diff options
Diffstat (limited to 'drivers/gpu')
180 files changed, 17472 insertions, 6334 deletions
diff --git a/drivers/gpu/drm/drm_bufs.c b/drivers/gpu/drm/drm_bufs.c index 3e257a50bf5..61e1ef90d4e 100644 --- a/drivers/gpu/drm/drm_bufs.c +++ b/drivers/gpu/drm/drm_bufs.c @@ -46,10 +46,11 @@ static struct drm_map_list *drm_find_matching_map(struct drm_device *dev, list_for_each_entry(entry, &dev->maplist, head) { /* * Because the kernel-userspace ABI is fixed at a 32-bit offset - * while PCI resources may live above that, we ignore the map - * offset for maps of type _DRM_FRAMEBUFFER or _DRM_REGISTERS. - * It is assumed that each driver will have only one resource of - * each type. + * while PCI resources may live above that, we only compare the + * lower 32 bits of the map offset for maps of type + * _DRM_FRAMEBUFFER or _DRM_REGISTERS. + * It is assumed that if a driver have more than one resource + * of each type, the lower 32 bits are different. */ if (!entry->map || map->type != entry->map->type || @@ -59,9 +60,12 @@ static struct drm_map_list *drm_find_matching_map(struct drm_device *dev, case _DRM_SHM: if (map->flags != _DRM_CONTAINS_LOCK) break; + return entry; case _DRM_REGISTERS: case _DRM_FRAME_BUFFER: - return entry; + if ((entry->map->offset & 0xffffffff) == + (map->offset & 0xffffffff)) + return entry; default: /* Make gcc happy */ ; } @@ -183,9 +187,6 @@ static int drm_addmap_core(struct drm_device * dev, resource_size_t offset, return -EINVAL; } #endif -#ifdef __alpha__ - map->offset += dev->hose->mem_space->start; -#endif /* Some drivers preinitialize some maps, without the X Server * needing to be aware of it. Therefore, we just return success * when the server tries to create a duplicate map. diff --git a/drivers/gpu/drm/drm_crtc.c b/drivers/gpu/drm/drm_crtc.c index 872747c5a54..82db1850666 100644 --- a/drivers/gpu/drm/drm_crtc.c +++ b/drivers/gpu/drm/drm_crtc.c @@ -886,9 +886,6 @@ int drm_mode_group_init(struct drm_device *dev, struct drm_mode_group *group) total_objects += dev->mode_config.num_connector; total_objects += dev->mode_config.num_encoder; - if (total_objects == 0) - return -EINVAL; - group->id_list = kzalloc(total_objects * sizeof(uint32_t), GFP_KERNEL); if (!group->id_list) return -ENOMEM; @@ -1113,7 +1110,7 @@ int drm_mode_getresources(struct drm_device *dev, void *data, if (card_res->count_fbs >= fb_count) { copied = 0; fb_id = (uint32_t __user *)(unsigned long)card_res->fb_id_ptr; - list_for_each_entry(fb, &file_priv->fbs, head) { + list_for_each_entry(fb, &file_priv->fbs, filp_head) { if (put_user(fb->base.id, fb_id + copied)) { ret = -EFAULT; goto out; diff --git a/drivers/gpu/drm/drm_crtc_helper.c b/drivers/gpu/drm/drm_crtc_helper.c index 92369655dca..f88a9b2c977 100644 --- a/drivers/gpu/drm/drm_crtc_helper.c +++ b/drivers/gpu/drm/drm_crtc_helper.c @@ -560,6 +560,11 @@ int drm_crtc_helper_set_config(struct drm_mode_set *set) mode_changed = true; } else if (set->fb == NULL) { mode_changed = true; + } else if (set->fb->depth != set->crtc->fb->depth) { + mode_changed = true; + } else if (set->fb->bits_per_pixel != + set->crtc->fb->bits_per_pixel) { + mode_changed = true; } else fb_changed = true; } diff --git a/drivers/gpu/drm/drm_debugfs.c b/drivers/gpu/drm/drm_debugfs.c index 9d8c892d07c..9d2668a5087 100644 --- a/drivers/gpu/drm/drm_debugfs.c +++ b/drivers/gpu/drm/drm_debugfs.c @@ -90,7 +90,6 @@ int drm_debugfs_create_files(struct drm_info_list *files, int count, struct drm_device *dev = minor->dev; struct dentry *ent; struct drm_info_node *tmp; - char name[64]; int i, ret; for (i = 0; i < count; i++) { @@ -108,6 +107,9 @@ int drm_debugfs_create_files(struct drm_info_list *files, int count, ent = debugfs_create_file(files[i].name, S_IFREG | S_IRUGO, root, tmp, &drm_debugfs_fops); if (!ent) { + char name[64]; + strncpy(name, root->d_name.name, + min(root->d_name.len, 64U)); DRM_ERROR("Cannot create /sys/kernel/debug/dri/%s/%s\n", name, files[i].name); kfree(tmp); diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c index adc9358c9be..7425e5c9bd7 100644 --- a/drivers/gpu/drm/drm_edid.c +++ b/drivers/gpu/drm/drm_edid.c @@ -127,6 +127,23 @@ static const u8 edid_header[] = { 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x00 }; + /* + * Sanity check the header of the base EDID block. Return 8 if the header + * is perfect, down to 0 if it's totally wrong. + */ +int drm_edid_header_is_valid(const u8 *raw_edid) +{ + int i, score = 0; + + for (i = 0; i < sizeof(edid_header); i++) + if (raw_edid[i] == edid_header[i]) + score++; + + return score; +} +EXPORT_SYMBOL(drm_edid_header_is_valid); + + /* * Sanity check the EDID block (base or extension). Return 0 if the block * doesn't check out, or 1 if it's valid. @@ -139,12 +156,7 @@ drm_edid_block_valid(u8 *raw_edid) struct edid *edid = (struct edid *)raw_edid; if (raw_edid[0] == 0x00) { - int score = 0; - - for (i = 0; i < sizeof(edid_header); i++) - if (raw_edid[i] == edid_header[i]) - score++; - + int score = drm_edid_header_is_valid(raw_edid); if (score == 8) ; else if (score >= 6) { DRM_DEBUG("Fixing EDID header, your hardware may be failing\n"); @@ -184,9 +196,9 @@ drm_edid_block_valid(u8 *raw_edid) bad: if (raw_edid) { - DRM_ERROR("Raw EDID:\n"); - print_hex_dump_bytes(KERN_ERR, DUMP_PREFIX_NONE, raw_edid, EDID_LENGTH); - printk("\n"); + printk(KERN_ERR "Raw EDID:\n"); + print_hex_dump(KERN_ERR, " \t", DUMP_PREFIX_NONE, 16, 1, + raw_edid, EDID_LENGTH, false); } return 0; } @@ -258,6 +270,17 @@ drm_do_probe_ddc_edid(struct i2c_adapter *adapter, unsigned char *buf, return ret == 2 ? 0 : -1; } +static bool drm_edid_is_zero(u8 *in_edid, int length) +{ + int i; + u32 *raw_edid = (u32 *)in_edid; + + for (i = 0; i < length / 4; i++) + if (*(raw_edid + i) != 0) + return false; + return true; +} + static u8 * drm_do_get_edid(struct drm_connector *connector, struct i2c_adapter *adapter) { @@ -273,6 +296,10 @@ drm_do_get_edid(struct drm_connector *connector, struct i2c_adapter *adapter) goto out; if (drm_edid_block_valid(block)) break; + if (i == 0 && drm_edid_is_zero(block, EDID_LENGTH)) { + connector->null_edid_counter++; + goto carp; + } } if (i == 4) goto carp; @@ -1413,6 +1440,73 @@ end: EXPORT_SYMBOL(drm_detect_monitor_audio); /** + * drm_add_display_info - pull display info out if present + * @edid: EDID data + * @info: display info (attached to connector) + * + * Grab any available display info and stuff it into the drm_display_info + * structure that's part of the connector. Useful for tracking bpp and + * color spaces. + */ +static void drm_add_display_info(struct edid *edid, + struct drm_display_info *info) +{ + u8 *edid_ext; + + info->width_mm = edid->width_cm * 10; + info->height_mm = edid->height_cm * 10; + + /* driver figures it out in this case */ + info->bpc = 0; + info->color_formats = 0; + + /* Only defined for 1.4 with digital displays */ + if (edid->revision < 4) + return; + + if (!(edid->input & DRM_EDID_INPUT_DIGITAL)) + return; + + switch (edid->input & DRM_EDID_DIGITAL_DEPTH_MASK) { + case DRM_EDID_DIGITAL_DEPTH_6: + info->bpc = 6; + break; + case DRM_EDID_DIGITAL_DEPTH_8: + info->bpc = 8; + break; + case DRM_EDID_DIGITAL_DEPTH_10: + info->bpc = 10; + break; + case DRM_EDID_DIGITAL_DEPTH_12: + info->bpc = 12; + break; + case DRM_EDID_DIGITAL_DEPTH_14: + info->bpc = 14; + break; + case DRM_EDID_DIGITAL_DEPTH_16: + info->bpc = 16; + break; + case DRM_EDID_DIGITAL_DEPTH_UNDEF: + default: + info->bpc = 0; + break; + } + + info->color_formats = DRM_COLOR_FORMAT_RGB444; + if (info->color_formats & DRM_EDID_FEATURE_RGB_YCRCB444) + info->color_formats = DRM_COLOR_FORMAT_YCRCB444; + if (info->color_formats & DRM_EDID_FEATURE_RGB_YCRCB422) + info->color_formats = DRM_COLOR_FORMAT_YCRCB422; + + /* Get data from CEA blocks if present */ + edid_ext = drm_find_cea_extension(edid); + if (!edid_ext) + return; + + info->cea_rev = edid_ext[1]; +} + +/** * drm_add_edid_modes - add modes from EDID data, if available * @connector: connector we're probing * @edid: edid data @@ -1460,8 +1554,7 @@ int drm_add_edid_modes(struct drm_connector *connector, struct edid *edid) if (quirks & (EDID_QUIRK_PREFER_LARGE_60 | EDID_QUIRK_PREFER_LARGE_75)) edid_fixup_preferred(connector, quirks); - connector->display_info.width_mm = edid->width_cm * 10; - connector->display_info.height_mm = edid->height_cm * 10; + drm_add_display_info(edid, &connector->display_info); return num_modes; } diff --git a/drivers/gpu/drm/drm_fb_helper.c b/drivers/gpu/drm/drm_fb_helper.c index 11d7a72c22d..802b61ac313 100644 --- a/drivers/gpu/drm/drm_fb_helper.c +++ b/drivers/gpu/drm/drm_fb_helper.c @@ -70,174 +70,50 @@ fail: } EXPORT_SYMBOL(drm_fb_helper_single_add_all_connectors); -/** - * drm_fb_helper_connector_parse_command_line - parse command line for connector - * @connector - connector to parse line for - * @mode_option - per connector mode option - * - * This parses the connector specific then generic command lines for - * modes and options to configure the connector. - * - * This uses the same parameters as the fb modedb.c, except for extra - * <xres>x<yres>[M][R][-<bpp>][@<refresh>][i][m][eDd] - * - * enable/enable Digital/disable bit at the end - */ -static bool drm_fb_helper_connector_parse_command_line(struct drm_fb_helper_connector *fb_helper_conn, - const char *mode_option) -{ - const char *name; - unsigned int namelen; - int res_specified = 0, bpp_specified = 0, refresh_specified = 0; - unsigned int xres = 0, yres = 0, bpp = 32, refresh = 0; - int yres_specified = 0, cvt = 0, rb = 0, interlace = 0, margins = 0; - int i; - enum drm_connector_force force = DRM_FORCE_UNSPECIFIED; - struct drm_fb_helper_cmdline_mode *cmdline_mode; - struct drm_connector *connector; - - if (!fb_helper_conn) - return false; - connector = fb_helper_conn->connector; - - cmdline_mode = &fb_helper_conn->cmdline_mode; - if (!mode_option) - mode_option = fb_mode_option; - - if (!mode_option) { - cmdline_mode->specified = false; - return false; - } - - name = mode_option; - namelen = strlen(name); - for (i = namelen-1; i >= 0; i--) { - switch (name[i]) { - case '@': - namelen = i; - if (!refresh_specified && !bpp_specified && - !yres_specified) { - refresh = simple_strtol(&name[i+1], NULL, 10); - refresh_specified = 1; - if (cvt || rb) - cvt = 0; - } else - goto done; - break; - case '-': - namelen = i; - if (!bpp_specified && !yres_specified) { - bpp = simple_strtol(&name[i+1], NULL, 10); - bpp_specified = 1; - if (cvt || rb) - cvt = 0; - } else - goto done; - break; - case 'x': - if (!yres_specified) { - yres = simple_strtol(&name[i+1], NULL, 10); - yres_specified = 1; - } else - goto done; - case '0' ... '9': - break; - case 'M': - if (!yres_specified) - cvt = 1; - break; - case 'R': - if (cvt) - rb = 1; - break; - case 'm': - if (!cvt) - margins = 1; - break; - case 'i': - if (!cvt) - interlace = 1; - break; - case 'e': - force = DRM_FORCE_ON; - break; - case 'D': - if ((connector->connector_type != DRM_MODE_CONNECTOR_DVII) && - (connector->connector_type != DRM_MODE_CONNECTOR_HDMIB)) - force = DRM_FORCE_ON; - else - force = DRM_FORCE_ON_DIGITAL; - break; - case 'd': - force = DRM_FORCE_OFF; - break; - default: - goto done; - } - } - if (i < 0 && yres_specified) { - xres = simple_strtol(name, NULL, 10); - res_specified = 1; - } -done: - - DRM_DEBUG_KMS("cmdline mode for connector %s %dx%d@%dHz%s%s%s\n", - drm_get_connector_name(connector), xres, yres, - (refresh) ? refresh : 60, (rb) ? " reduced blanking" : - "", (margins) ? " with margins" : "", (interlace) ? - " interlaced" : ""); - - if (force) { - const char *s; - switch (force) { - case DRM_FORCE_OFF: s = "OFF"; break; - case DRM_FORCE_ON_DIGITAL: s = "ON - dig"; break; - default: - case DRM_FORCE_ON: s = "ON"; break; - } - - DRM_INFO("forcing %s connector %s\n", - drm_get_connector_name(connector), s); - connector->force = force; - } - - if (res_specified) { - cmdline_mode->specified = true; - cmdline_mode->xres = xres; - cmdline_mode->yres = yres; - } - - if (refresh_specified) { - cmdline_mode->refresh_specified = true; - cmdline_mode->refresh = refresh; - } - - if (bpp_specified) { - cmdline_mode->bpp_specified = true; - cmdline_mode->bpp = bpp; - } - cmdline_mode->rb = rb ? true : false; - cmdline_mode->cvt = cvt ? true : false; - cmdline_mode->interlace = interlace ? true : false; - - return true; -} - static int drm_fb_helper_parse_command_line(struct drm_fb_helper *fb_helper) { struct drm_fb_helper_connector *fb_helper_conn; int i; for (i = 0; i < fb_helper->connector_count; i++) { + struct drm_cmdline_mode *mode; + struct drm_connector *connector; char *option = NULL; fb_helper_conn = fb_helper->connector_info[i]; + connector = fb_helper_conn->connector; + mode = &fb_helper_conn->cmdline_mode; /* do something on return - turn off connector maybe */ - if (fb_get_options(drm_get_connector_name(fb_helper_conn->connector), &option)) + if (fb_get_options(drm_get_connector_name(connector), &option)) continue; - drm_fb_helper_connector_parse_command_line(fb_helper_conn, option); + if (drm_mode_parse_command_line_for_connector(option, + connector, + mode)) { + if (mode->force) { + const char *s; + switch (mode->force) { + case DRM_FORCE_OFF: s = "OFF"; break; + case DRM_FORCE_ON_DIGITAL: s = "ON - dig"; break; + default: + case DRM_FORCE_ON: s = "ON"; break; + } + + DRM_INFO("forcing %s connector %s\n", + drm_get_connector_name(connector), s); + connector->force = mode->force; + } + + DRM_DEBUG_KMS("cmdline mode for connector %s %dx%d@%dHz%s%s%s\n", + drm_get_connector_name(connector), + mode->xres, mode->yres, + mode->refresh_specified ? mode->refresh : 60, + mode->rb ? " reduced blanking" : "", + mode->margins ? " with margins" : "", + mode->interlace ? " interlaced" : ""); + } + } return 0; } @@ -901,7 +777,7 @@ int drm_fb_helper_single_fb_probe(struct drm_fb_helper *fb_helper, /* first up get a count of crtcs now in use and new min/maxes width/heights */ for (i = 0; i < fb_helper->connector_count; i++) { struct drm_fb_helper_connector *fb_helper_conn = fb_helper->connector_info[i]; - struct drm_fb_helper_cmdline_mode *cmdline_mode; + struct drm_cmdline_mode *cmdline_mode; cmdline_mode = &fb_helper_conn->cmdline_mode; @@ -1123,7 +999,7 @@ static struct drm_display_mode *drm_has_preferred_mode(struct drm_fb_helper_conn static bool drm_has_cmdline_mode(struct drm_fb_helper_connector *fb_connector) { - struct drm_fb_helper_cmdline_mode *cmdline_mode; + struct drm_cmdline_mode *cmdline_mode; cmdline_mode = &fb_connector->cmdline_mode; return cmdline_mode->specified; } @@ -1131,7 +1007,7 @@ static bool drm_has_cmdline_mode(struct drm_fb_helper_connector *fb_connector) static struct drm_display_mode *drm_pick_cmdline_mode(struct drm_fb_helper_connector *fb_helper_conn, int width, int height) { - struct drm_fb_helper_cmdline_mode *cmdline_mode; + struct drm_cmdline_mode *cmdline_mode; struct drm_display_mode *mode = NULL; cmdline_mode = &fb_helper_conn->cmdline_mode; @@ -1163,19 +1039,8 @@ static struct drm_display_mode *drm_pick_cmdline_mode(struct drm_fb_helper_conne } create_mode: - if (cmdline_mode->cvt) - mode = drm_cvt_mode(fb_helper_conn->connector->dev, - cmdline_mode->xres, cmdline_mode->yres, - cmdline_mode->refresh_specified ? cmdline_mode->refresh : 60, - cmdline_mode->rb, cmdline_mode->interlace, - cmdline_mode->margins); - else - mode = drm_gtf_mode(fb_helper_conn->connector->dev, - cmdline_mode->xres, cmdline_mode->yres, - cmdline_mode->refresh_specified ? cmdline_mode->refresh : 60, - cmdline_mode->interlace, - cmdline_mode->margins); - drm_mode_set_crtcinfo(mode, CRTC_INTERLACE_HALVE_V); + mode = drm_mode_create_from_cmdline_mode(fb_helper_conn->connector->dev, + cmdline_mode); list_add(&mode->head, &fb_helper_conn->connector->modes); return mode; } @@ -1516,17 +1381,33 @@ bool drm_fb_helper_initial_config(struct drm_fb_helper *fb_helper, int bpp_sel) } EXPORT_SYMBOL(drm_fb_helper_initial_config); -bool drm_fb_helper_hotplug_event(struct drm_fb_helper *fb_helper) +/** + * drm_fb_helper_hotplug_event - respond to a hotplug notification by + * probing all the outputs attached to the fb. + * @fb_helper: the drm_fb_helper + * + * LOCKING: + * Called at runtime, must take mode config lock. + * + * Scan the connectors attached to the fb_helper and try to put together a + * setup after *notification of a change in output configuration. + * + * RETURNS: + * 0 on success and a non-zero error code otherwise. + */ +int drm_fb_helper_hotplug_event(struct drm_fb_helper *fb_helper) { + struct drm_device *dev = fb_helper->dev; int count = 0; u32 max_width, max_height, bpp_sel; bool bound = false, crtcs_bound = false; struct drm_crtc *crtc; if (!fb_helper->fb) - return false; + return 0; - list_for_each_entry(crtc, &fb_helper->dev->mode_config.crtc_list, head) { + mutex_lock(&dev->mode_config.mutex); + list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { if (crtc->fb) crtcs_bound = true; if (crtc->fb == fb_helper->fb) @@ -1535,7 +1416,8 @@ bool drm_fb_helper_hotplug_event(struct drm_fb_helper *fb_helper) if (!bound && crtcs_bound) { fb_helper->delayed_hotplug = true; - return false; + mutex_unlock(&dev->mode_config.mutex); + return 0; } DRM_DEBUG_KMS("\n"); @@ -1546,6 +1428,7 @@ bool drm_fb_helper_hotplug_event(struct drm_fb_helper *fb_helper) count = drm_fb_helper_probe_connector_modes(fb_helper, max_width, max_height); drm_setup_crtcs(fb_helper); + mutex_unlock(&dev->mode_config.mutex); return drm_fb_helper_single_fb_probe(fb_helper, bpp_sel); } diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c index 74e4ff57801..186d62eb063 100644 --- a/drivers/gpu/drm/drm_gem.c +++ b/drivers/gpu/drm/drm_gem.c @@ -34,6 +34,7 @@ #include <linux/module.h> #include <linux/mman.h> #include <linux/pagemap.h> +#include <linux/shmem_fs.h> #include "drmP.h" /** @file drm_gem.c @@ -128,7 +129,7 @@ drm_gem_destroy(struct drm_device *dev) } /** - * Initialize an already allocate GEM object of the specified size with + * Initialize an already allocated GEM object of the specified size with * shmfs backing store. */ int drm_gem_object_init(struct drm_device *dev, @@ -150,6 +151,27 @@ int drm_gem_object_init(struct drm_device *dev, EXPORT_SYMBOL(drm_gem_object_init); /** + * Initialize an already allocated GEM object of the specified size with + * no GEM provided backing store. Instead the caller is responsible for + * backing the object and handling it. + */ +int drm_gem_private_object_init(struct drm_device *dev, + struct drm_gem_object *obj, size_t size) +{ + BUG_ON((size & (PAGE_SIZE - 1)) != 0); + + obj->dev = dev; + obj->filp = NULL; + + kref_init(&obj->refcount); + atomic_set(&obj->handle_count, 0); + obj->size = size; + + return 0; +} +EXPORT_SYMBOL(drm_gem_private_object_init); + +/** * Allocate a GEM object of the specified size with shmfs backing store */ struct drm_gem_object * @@ -210,6 +232,8 @@ drm_gem_handle_delete(struct drm_file *filp, u32 handle) idr_remove(&filp->object_idr, handle); spin_unlock(&filp->table_lock); + if (dev->driver->gem_close_object) + dev->driver->gem_close_object(obj, filp); drm_gem_object_handle_unreference_unlocked(obj); return 0; @@ -226,7 +250,8 @@ drm_gem_handle_create(struct drm_file *file_priv, struct drm_gem_object *obj, u32 *handlep) { - int ret; + struct drm_device *dev = obj->dev; + int ret; /* * Get the user-visible handle using idr. @@ -247,6 +272,15 @@ again: return ret; drm_gem_object_handle_reference(obj); + + if (dev->driver->gem_open_object) { + ret = dev->driver->gem_open_object(obj, file_priv); + if (ret) { + drm_gem_handle_delete(file_priv, *handlep); + return ret; + } + } + return 0; } EXPORT_SYMBOL(drm_gem_handle_create); @@ -401,7 +435,12 @@ drm_gem_open(struct drm_device *dev, struct drm_file *file_private) static int drm_gem_object_release_handle(int id, void *ptr, void *data) { + struct drm_file *file_priv = data; struct drm_gem_object *obj = ptr; + struct drm_device *dev = obj->dev; + + if (dev->driver->gem_close_object) + dev->driver->gem_close_object(obj, file_priv); drm_gem_object_handle_unreference_unlocked(obj); @@ -417,7 +456,7 @@ void drm_gem_release(struct drm_device *dev, struct drm_file *file_private) { idr_for_each(&file_private->object_idr, - &drm_gem_object_release_handle, NULL); + &drm_gem_object_release_handle, file_private); idr_remove_all(&file_private->object_idr); idr_destroy(&file_private->object_idr); @@ -426,7 +465,8 @@ drm_gem_release(struct drm_device *dev, struct drm_file *file_private) void drm_gem_object_release(struct drm_gem_object *obj) { - fput(obj->filp); + if (obj->filp) + fput(obj->filp); } EXPORT_SYMBOL(drm_gem_object_release); diff --git a/drivers/gpu/drm/drm_ioc32.c b/drivers/gpu/drm/drm_ioc32.c index d61d185cf04..4a058c7af6c 100644 --- a/drivers/gpu/drm/drm_ioc32.c +++ b/drivers/gpu/drm/drm_ioc32.c @@ -28,6 +28,7 @@ * IN THE SOFTWARE. */ #include <linux/compat.h> +#include <linux/ratelimit.h> #include "drmP.h" #include "drm_core.h" @@ -253,10 +254,10 @@ static int compat_drm_addmap(struct file *file, unsigned int cmd, return -EFAULT; m32.handle = (unsigned long)handle; - if (m32.handle != (unsigned long)handle && printk_ratelimit()) - printk(KERN_ERR "compat_drm_addmap truncated handle" - " %p for type %d offset %x\n", - handle, m32.type, m32.offset); + if (m32.handle != (unsigned long)handle) + printk_ratelimited(KERN_ERR "compat_drm_addmap truncated handle" + " %p for type %d offset %x\n", + handle, m32.type, m32.offset); if (copy_to_user(argp, &m32, sizeof(m32))) return -EFAULT; diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c index a1f12cb043d..3830e9e478c 100644 --- a/drivers/gpu/drm/drm_irq.c +++ b/drivers/gpu/drm/drm_irq.c @@ -291,11 +291,14 @@ static void drm_irq_vgaarb_nokms(void *cookie, bool state) if (!dev->irq_enabled) return; - if (state) - dev->driver->irq_uninstall(dev); - else { - dev->driver->irq_preinstall(dev); - dev->driver->irq_postinstall(dev); + if (state) { + if (dev->driver->irq_uninstall) + dev->driver->irq_uninstall(dev); + } else { + if (dev->driver->irq_preinstall) + dev->driver->irq_preinstall(dev); + if (dev->driver->irq_postinstall) + dev->driver->irq_postinstall(dev); } } @@ -338,7 +341,8 @@ int drm_irq_install(struct drm_device *dev) DRM_DEBUG("irq=%d\n", drm_dev_to_irq(dev)); /* Before installing handler */ - dev->driver->irq_preinstall(dev); + if (dev->driver->irq_preinstall) + dev->driver->irq_preinstall(dev); /* Install handler */ if (drm_core_check_feature(dev, DRIVER_IRQ_SHARED)) @@ -363,11 +367,16 @@ int drm_irq_install(struct drm_device *dev) vga_client_register(dev->pdev, (void *)dev, drm_irq_vgaarb_nokms, NULL); /* After installing handler */ - ret = dev->driver->irq_postinstall(dev); + if (dev->driver->irq_postinstall) + ret = dev->driver->irq_postinstall(dev); + if (ret < 0) { mutex_lock(&dev->struct_mutex); dev->irq_enabled = 0; mutex_unlock(&dev->struct_mutex); + if (!drm_core_check_feature(dev, DRIVER_MODESET)) + vga_client_register(dev->pdev, NULL, NULL, NULL); + free_irq(drm_dev_to_irq(dev), dev); } return ret; @@ -413,7 +422,8 @@ int drm_irq_uninstall(struct drm_device *dev) if (!drm_core_check_feature(dev, DRIVER_MODESET)) vga_client_register(dev->pdev, NULL, NULL, NULL); - dev->driver->irq_uninstall(dev); + if (dev->driver->irq_uninstall) + dev->driver->irq_uninstall(dev); free_irq(drm_dev_to_irq(dev), dev); @@ -684,10 +694,11 @@ int drm_calc_vbltimestamp_from_scanoutpos(struct drm_device *dev, int crtc, */ *vblank_time = ns_to_timeval(timeval_to_ns(&raw_time) - delta_ns); - DRM_DEBUG("crtc %d : v %d p(%d,%d)@ %d.%d -> %d.%d [e %d us, %d rep]\n", - crtc, (int) vbl_status, hpos, vpos, raw_time.tv_sec, - raw_time.tv_usec, vblank_time->tv_sec, vblank_time->tv_usec, - (int) duration_ns/1000, i); + DRM_DEBUG("crtc %d : v %d p(%d,%d)@ %ld.%ld -> %ld.%ld [e %d us, %d rep]\n", + crtc, (int)vbl_status, hpos, vpos, + (long)raw_time.tv_sec, (long)raw_time.tv_usec, + (long)vblank_time->tv_sec, (long)vblank_time->tv_usec, + (int)duration_ns/1000, i); vbl_status = DRM_VBLANKTIME_SCANOUTPOS_METHOD; if (invbl) diff --git a/drivers/gpu/drm/drm_modes.c b/drivers/gpu/drm/drm_modes.c index 25bf87390f5..ad74fb4dc54 100644 --- a/drivers/gpu/drm/drm_modes.c +++ b/drivers/gpu/drm/drm_modes.c @@ -974,3 +974,192 @@ void drm_mode_connector_list_update(struct drm_connector *connector) } } EXPORT_SYMBOL(drm_mode_connector_list_update); + +/** + * drm_mode_parse_command_line_for_connector - parse command line for connector + * @mode_option - per connector mode option + * @connector - connector to parse line for + * + * This parses the connector specific then generic command lines for + * modes and options to configure the connector. + * + * This uses the same parameters as the fb modedb.c, except for extra + * <xres>x<yres>[M][R][-<bpp>][@<refresh>][i][m][eDd] + * + * enable/enable Digital/disable bit at the end + */ +bool drm_mode_parse_command_line_for_connector(const char *mode_option, + struct drm_connector *connector, + struct drm_cmdline_mode *mode) +{ + const char *name; + unsigned int namelen; + bool res_specified = false, bpp_specified = false, refresh_specified = false; + unsigned int xres = 0, yres = 0, bpp = 32, refresh = 0; + bool yres_specified = false, cvt = false, rb = false; + bool interlace = false, margins = false, was_digit = false; + int i; + enum drm_connector_force force = DRM_FORCE_UNSPECIFIED; + +#ifdef CONFIG_FB + if (!mode_option) + mode_option = fb_mode_option; +#endif + + if (!mode_option) { + mode->specified = false; + return false; + } + + name = mode_option; + namelen = strlen(name); + for (i = namelen-1; i >= 0; i--) { + switch (name[i]) { + case '@': + if (!refresh_specified && !bpp_specified && + !yres_specified && !cvt && !rb && was_digit) { + refresh = simple_strtol(&name[i+1], NULL, 10); + refresh_specified = true; + was_digit = false; + } else + goto done; + break; + case '-': + if (!bpp_specified && !yres_specified && !cvt && + !rb && was_digit) { + bpp = simple_strtol(&name[i+1], NULL, 10); + bpp_specified = true; + was_digit = false; + } else + goto done; + break; + case 'x': + if (!yres_specified && was_digit) { + yres = simple_strtol(&name[i+1], NULL, 10); + yres_specified = true; + was_digit = false; + } else + goto done; + case '0' ... '9': + was_digit = true; + break; + case 'M': + if (yres_specified || cvt || was_digit) + goto done; + cvt = true; + break; + case 'R': + if (yres_specified || cvt || rb || was_digit) + goto done; + rb = true; + break; + case 'm': + if (cvt || yres_specified || was_digit) + goto done; + margins = true; + break; + case 'i': + if (cvt || yres_specified || was_digit) + goto done; + interlace = true; + break; + case 'e': + if (yres_specified || bpp_specified || refresh_specified || + was_digit || (force != DRM_FORCE_UNSPECIFIED)) + goto done; + + force = DRM_FORCE_ON; + break; + case 'D': + if (yres_specified || bpp_specified || refresh_specified || + was_digit || (force != DRM_FORCE_UNSPECIFIED)) + goto done; + + if ((connector->connector_type != DRM_MODE_CONNECTOR_DVII) && + (connector->connector_type != DRM_MODE_CONNECTOR_HDMIB)) + force = DRM_FORCE_ON; + else + force = DRM_FORCE_ON_DIGITAL; + break; + case 'd': + if (yres_specified || bpp_specified || refresh_specified || + was_digit || (force != DRM_FORCE_UNSPECIFIED)) + goto done; + + force = DRM_FORCE_OFF; + break; + default: + goto done; + } + } + + if (i < 0 && yres_specified) { + char *ch; + xres = simple_strtol(name, &ch, 10); + if ((ch != NULL) && (*ch == 'x')) + res_specified = true; + else + i = ch - name; + } else if (!yres_specified && was_digit) { + /* catch mode that begins with digits but has no 'x' */ + i = 0; + } +done: + if (i >= 0) { + printk(KERN_WARNING + "parse error at position %i in video mode '%s'\n", + i, name); + mode->specified = false; + return false; + } + + if (res_specified) { + mode->specified = true; + mode->xres = xres; + mode->yres = yres; + } + + if (refresh_specified) { + mode->refresh_specified = true; + mode->refresh = refresh; + } + + if (bpp_specified) { + mode->bpp_specified = true; + mode->bpp = bpp; + } + mode->rb = rb; + mode->cvt = cvt; + mode->interlace = interlace; + mode->margins = margins; + mode->force = force; + + return true; +} +EXPORT_SYMBOL(drm_mode_parse_command_line_for_connector); + +struct drm_display_mode * +drm_mode_create_from_cmdline_mode(struct drm_device *dev, + struct drm_cmdline_mode *cmd) +{ + struct drm_display_mode *mode; + + if (cmd->cvt) + mode = drm_cvt_mode(dev, + cmd->xres, cmd->yres, + cmd->refresh_specified ? cmd->refresh : 60, + cmd->rb, cmd->interlace, + cmd->margins); + else + mode = drm_gtf_mode(dev, + cmd->xres, cmd->yres, + cmd->refresh_specified ? cmd->refresh : 60, + cmd->interlace, + cmd->margins); + if (!mode) + return NULL; + + drm_mode_set_crtcinfo(mode, CRTC_INTERLACE_HALVE_V); + return mode; +} +EXPORT_SYMBOL(drm_mode_create_from_cmdline_mode); diff --git a/drivers/gpu/drm/drm_pci.c b/drivers/gpu/drm/drm_pci.c index e1aee4f6a7c..b6a19cb07ca 100644 --- a/drivers/gpu/drm/drm_pci.c +++ b/drivers/gpu/drm/drm_pci.c @@ -251,7 +251,7 @@ err: } -int drm_pci_irq_by_busid(struct drm_device *dev, struct drm_irq_busid *p) +static int drm_pci_irq_by_busid(struct drm_device *dev, struct drm_irq_busid *p) { if ((p->busnum >> 8) != drm_get_pci_domain(dev) || (p->busnum & 0xff) != dev->pdev->bus->number || @@ -292,6 +292,7 @@ static struct drm_bus drm_pci_bus = { .get_name = drm_pci_get_name, .set_busid = drm_pci_set_busid, .set_unique = drm_pci_set_unique, + .irq_by_busid = drm_pci_irq_by_busid, .agp_init = drm_pci_agp_init, }; diff --git a/drivers/gpu/drm/drm_platform.c b/drivers/gpu/drm/drm_platform.c index 7223f06d8e5..2a8b6265ad3 100644 --- a/drivers/gpu/drm/drm_platform.c +++ b/drivers/gpu/drm/drm_platform.c @@ -123,14 +123,15 @@ static int drm_platform_set_busid(struct drm_device *dev, struct drm_master *mas { int len, ret; - master->unique_len = 10 + strlen(dev->platformdev->name); + master->unique_len = 13 + strlen(dev->platformdev->name); + master->unique_size = master->unique_len; master->unique = kmalloc(master->unique_len + 1, GFP_KERNEL); if (master->unique == NULL) return -ENOMEM; len = snprintf(master->unique, master->unique_len, - "platform:%s", dev->platformdev->name); + "platform:%s:%02d", dev->platformdev->name, dev->platformdev->id); if (len > master->unique_len) { DRM_ERROR("Unique buffer overflowed\n"); diff --git a/drivers/gpu/drm/drm_scatter.c b/drivers/gpu/drm/drm_scatter.c index d15e09b0ae0..7525e0311e5 100644 --- a/drivers/gpu/drm/drm_scatter.c +++ b/drivers/gpu/drm/drm_scatter.c @@ -83,30 +83,26 @@ int drm_sg_alloc(struct drm_device *dev, struct drm_scatter_gather * request) if (dev->sg) return -EINVAL; - entry = kmalloc(sizeof(*entry), GFP_KERNEL); + entry = kzalloc(sizeof(*entry), GFP_KERNEL); if (!entry) return -ENOMEM; - memset(entry, 0, sizeof(*entry)); pages = (request->size + PAGE_SIZE - 1) / PAGE_SIZE; DRM_DEBUG("size=%ld pages=%ld\n", request->size, pages); entry->pages = pages; - entry->pagelist = kmalloc(pages * sizeof(*entry->pagelist), GFP_KERNEL); + entry->pagelist = kcalloc(pages, sizeof(*entry->pagelist), GFP_KERNEL); if (!entry->pagelist) { kfree(entry); return -ENOMEM; } - memset(entry->pagelist, 0, pages * sizeof(*entry->pagelist)); - - entry->busaddr = kmalloc(pages * sizeof(*entry->busaddr), GFP_KERNEL); + entry->busaddr = kcalloc(pages, sizeof(*entry->busaddr), GFP_KERNEL); if (!entry->busaddr) { kfree(entry->pagelist); kfree(entry); return -ENOMEM; } - memset((void *)entry->busaddr, 0, pages * sizeof(*entry->busaddr)); entry->virtual = drm_vmalloc_dma(pages << PAGE_SHIFT); if (!entry->virtual) { diff --git a/drivers/gpu/drm/drm_stub.c b/drivers/gpu/drm/drm_stub.c index 001273d57f2..6d7b083c5b7 100644 --- a/drivers/gpu/drm/drm_stub.c +++ b/drivers/gpu/drm/drm_stub.c @@ -62,6 +62,26 @@ struct idr drm_minors_idr; struct class *drm_class; struct proc_dir_entry *drm_proc_root; struct dentry *drm_debugfs_root; + +int drm_err(const char *func, const char *format, ...) +{ + struct va_format vaf; + va_list args; + int r; + + va_start(args, format); + + vaf.fmt = format; + vaf.va = &args; + + r = printk(KERN_ERR "[" DRM_NAME ":%s] *ERROR* %pV", func, &vaf); + + va_end(args); + + return r; +} +EXPORT_SYMBOL(drm_err); + void drm_ut_debug_printk(unsigned int request_level, const char *prefix, const char *function_name, @@ -78,6 +98,7 @@ void drm_ut_debug_printk(unsigned int request_level, } } EXPORT_SYMBOL(drm_ut_debug_printk); + static int drm_minor_get_id(struct drm_device *dev, int type) { int new_id; diff --git a/drivers/gpu/drm/drm_vm.c b/drivers/gpu/drm/drm_vm.c index 2c3fcbdfd8f..5db96d45fc7 100644 --- a/drivers/gpu/drm/drm_vm.c +++ b/drivers/gpu/drm/drm_vm.c @@ -526,7 +526,7 @@ static int drm_mmap_dma(struct file *filp, struct vm_area_struct *vma) static resource_size_t drm_core_get_reg_ofs(struct drm_device *dev) { #ifdef __alpha__ - return dev->hose->dense_mem_base - dev->hose->mem_space->start; + return dev->hose->dense_mem_base; #else return 0; #endif diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c index 87c8e29465e..3c395a59da3 100644 --- a/drivers/gpu/drm/i915/i915_debugfs.c +++ b/drivers/gpu/drm/i915/i915_debugfs.c @@ -106,11 +106,12 @@ static const char *get_tiling_flag(struct drm_i915_gem_object *obj) } } -static const char *agp_type_str(int type) +static const char *cache_level_str(int type) { switch (type) { - case 0: return " uncached"; - case 1: return " snooped"; + case I915_CACHE_NONE: return " uncached"; + case I915_CACHE_LLC: return " snooped (LLC)"; + case I915_CACHE_LLC_MLC: return " snooped (LLC+MLC)"; default: return ""; } } @@ -127,7 +128,7 @@ describe_obj(struct seq_file *m, struct drm_i915_gem_object *obj) obj->base.write_domain, obj->last_rendering_seqno, obj->last_fenced_seqno, - agp_type_str(obj->agp_type == AGP_USER_CACHED_MEMORY), + cache_level_str(obj->cache_level), obj->dirty ? " dirty" : "", obj->madv == I915_MADV_DONTNEED ? " purgeable" : ""); if (obj->base.name) @@ -498,7 +499,7 @@ static int i915_interrupt_info(struct seq_file *m, void *data) seq_printf(m, "Interrupts received: %d\n", atomic_read(&dev_priv->irq_received)); for (i = 0; i < I915_NUM_RINGS; i++) { - if (IS_GEN6(dev)) { + if (IS_GEN6(dev) || IS_GEN7(dev)) { seq_printf(m, "Graphics Interrupt mask (%s): %08x\n", dev_priv->ring[i].name, I915_READ_IMR(&dev_priv->ring[i])); @@ -714,7 +715,7 @@ static void print_error_buffers(struct seq_file *m, dirty_flag(err->dirty), purgeable_flag(err->purgeable), ring_str(err->ring), - agp_type_str(err->agp_type)); + cache_level_str(err->cache_level)); if (err->name) seq_printf(m, " (name: %d)", err->name); @@ -775,7 +776,7 @@ static int i915_error_state(struct seq_file *m, void *unused) seq_printf(m, " INSTPM: 0x%08x\n", error->instpm); seq_printf(m, " seqno: 0x%08x\n", error->seqno); - for (i = 0; i < 16; i++) + for (i = 0; i < dev_priv->num_fence_regs; i++) seq_printf(m, " fence[%d] = %08llx\n", i, error->fence[i]); if (error->active_bo) @@ -852,6 +853,7 @@ static int i915_cur_delayinfo(struct seq_file *m, void *unused) struct drm_info_node *node = (struct drm_info_node *) m->private; struct drm_device *dev = node->minor->dev; drm_i915_private_t *dev_priv = dev->dev_private; + int ret; if (IS_GEN5(dev)) { u16 rgvswctl = I915_READ16(MEMSWCTL); @@ -863,7 +865,7 @@ static int i915_cur_delayinfo(struct seq_file *m, void *unused) MEMSTAT_VID_SHIFT); seq_printf(m, "Current P-state: %d\n", (rgvstat & MEMSTAT_PSTATE_MASK) >> MEMSTAT_PSTATE_SHIFT); - } else if (IS_GEN6(dev)) { + } else if (IS_GEN6(dev) || IS_GEN7(dev)) { u32 gt_perf_status = I915_READ(GEN6_GT_PERF_STATUS); u32 rp_state_limits = I915_READ(GEN6_RP_STATE_LIMITS); u32 rp_state_cap = I915_READ(GEN6_RP_STATE_CAP); @@ -873,7 +875,11 @@ static int i915_cur_delayinfo(struct seq_file *m, void *unused) int max_freq; /* RPSTAT1 is in the GT power well */ - __gen6_gt_force_wake_get(dev_priv); + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + gen6_gt_force_wake_get(dev_priv); rpstat = I915_READ(GEN6_RPSTAT1); rpupei = I915_READ(GEN6_RP_CUR_UP_EI); @@ -883,6 +889,9 @@ static int i915_cur_delayinfo(struct seq_file *m, void *unused) rpcurdown = I915_READ(GEN6_RP_CUR_DOWN); rpprevdown = I915_READ(GEN6_RP_PREV_DOWN); + gen6_gt_force_wake_put(dev_priv); + mutex_unlock(&dev->struct_mutex); + seq_printf(m, "GT_PERF_STATUS: 0x%08x\n", gt_perf_status); seq_printf(m, "RPSTAT1: 0x%08x\n", rpstat); seq_printf(m, "Render p-state ratio: %d\n", @@ -917,8 +926,6 @@ static int i915_cur_delayinfo(struct seq_file *m, void *unused) max_freq = rp_state_cap & 0xff; seq_printf(m, "Max non-overclocked (RP0) frequency: %dMHz\n", max_freq * 50); - - __gen6_gt_force_wake_put(dev_priv); } else { seq_printf(m, "no P-state info available\n"); } @@ -1058,6 +1065,9 @@ static int i915_fbc_status(struct seq_file *m, void *unused) case FBC_MULTIPLE_PIPES: seq_printf(m, "multiple pipes are enabled"); break; + case FBC_MODULE_PARAM: + seq_printf(m, "disabled per module param (default off)"); + break; default: seq_printf(m, "unknown reason"); } @@ -1113,6 +1123,44 @@ static int i915_emon_status(struct seq_file *m, void *unused) return 0; } +static int i915_ring_freq_table(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + int gpu_freq, ia_freq; + + if (!(IS_GEN6(dev) || IS_GEN7(dev))) { + seq_printf(m, "unsupported on this chipset\n"); + return 0; + } + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + + seq_printf(m, "GPU freq (MHz)\tEffective CPU freq (MHz)\n"); + + for (gpu_freq = dev_priv->min_delay; gpu_freq <= dev_priv->max_delay; + gpu_freq++) { + I915_WRITE(GEN6_PCODE_DATA, gpu_freq); + I915_WRITE(GEN6_PCODE_MAILBOX, GEN6_PCODE_READY | + GEN6_PCODE_READ_MIN_FREQ_TABLE); + if (wait_for((I915_READ(GEN6_PCODE_MAILBOX) & + GEN6_PCODE_READY) == 0, 10)) { + DRM_ERROR("pcode read of freq table timed out\n"); + continue; + } + ia_freq = I915_READ(GEN6_PCODE_DATA); + seq_printf(m, "%d\t\t%d\n", gpu_freq * 50, ia_freq * 100); + } + + mutex_unlock(&dev->struct_mutex); + + return 0; +} + static int i915_gfxec(struct seq_file *m, void *unused) { struct drm_info_node *node = (struct drm_info_node *) m->private; @@ -1186,6 +1234,46 @@ static int i915_gem_framebuffer_info(struct seq_file *m, void *data) return 0; } +static int i915_context_status(struct seq_file *m, void *unused) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + drm_i915_private_t *dev_priv = dev->dev_private; + int ret; + + ret = mutex_lock_interruptible(&dev->mode_config.mutex); + if (ret) + return ret; + + if (dev_priv->pwrctx) { + seq_printf(m, "power context "); + describe_obj(m, dev_priv->pwrctx); + seq_printf(m, "\n"); + } + + if (dev_priv->renderctx) { + seq_printf(m, "render context "); + describe_obj(m, dev_priv->renderctx); + seq_printf(m, "\n"); + } + + mutex_unlock(&dev->mode_config.mutex); + + return 0; +} + +static int i915_gen6_forcewake_count_info(struct seq_file *m, void *data) +{ + struct drm_info_node *node = (struct drm_info_node *) m->private; + struct drm_device *dev = node->minor->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + seq_printf(m, "forcewake count = %d\n", + atomic_read(&dev_priv->forcewake_count)); + + return 0; +} + static int i915_wedged_open(struct inode *inode, struct file *filp) @@ -1250,6 +1338,155 @@ static const struct file_operations i915_wedged_fops = { .llseek = default_llseek, }; +static int +i915_max_freq_open(struct inode *inode, + struct file *filp) +{ + filp->private_data = inode->i_private; + return 0; +} + +static ssize_t +i915_max_freq_read(struct file *filp, + char __user *ubuf, + size_t max, + loff_t *ppos) +{ + struct drm_device *dev = filp->private_data; + drm_i915_private_t *dev_priv = dev->dev_private; + char buf[80]; + int len; + + len = snprintf(buf, sizeof (buf), + "max freq: %d\n", dev_priv->max_delay * 50); + + if (len > sizeof (buf)) + len = sizeof (buf); + + return simple_read_from_buffer(ubuf, max, ppos, buf, len); +} + +static ssize_t +i915_max_freq_write(struct file *filp, + const char __user *ubuf, + size_t cnt, + loff_t *ppos) +{ + struct drm_device *dev = filp->private_data; + struct drm_i915_private *dev_priv = dev->dev_private; + char buf[20]; + int val = 1; + + if (cnt > 0) { + if (cnt > sizeof (buf) - 1) + return -EINVAL; + + if (copy_from_user(buf, ubuf, cnt)) + return -EFAULT; + buf[cnt] = 0; + + val = simple_strtoul(buf, NULL, 0); + } + + DRM_DEBUG_DRIVER("Manually setting max freq to %d\n", val); + + /* + * Turbo will still be enabled, but won't go above the set value. + */ + dev_priv->max_delay = val / 50; + + gen6_set_rps(dev, val / 50); + + return cnt; +} + +static const struct file_operations i915_max_freq_fops = { + .owner = THIS_MODULE, + .open = i915_max_freq_open, + .read = i915_max_freq_read, + .write = i915_max_freq_write, + .llseek = default_llseek, +}; + +static int +i915_cache_sharing_open(struct inode *inode, + struct file *filp) +{ + filp->private_data = inode->i_private; + return 0; +} + +static ssize_t +i915_cache_sharing_read(struct file *filp, + char __user *ubuf, + size_t max, + loff_t *ppos) +{ + struct drm_device *dev = filp->private_data; + drm_i915_private_t *dev_priv = dev->dev_private; + char buf[80]; + u32 snpcr; + int len; + + mutex_lock(&dev_priv->dev->struct_mutex); + snpcr = I915_READ(GEN6_MBCUNIT_SNPCR); + mutex_unlock(&dev_priv->dev->struct_mutex); + + len = snprintf(buf, sizeof (buf), + "%d\n", (snpcr & GEN6_MBC_SNPCR_MASK) >> + GEN6_MBC_SNPCR_SHIFT); + + if (len > sizeof (buf)) + len = sizeof (buf); + + return simple_read_from_buffer(ubuf, max, ppos, buf, len); +} + +static ssize_t +i915_cache_sharing_write(struct file *filp, + const char __user *ubuf, + size_t cnt, + loff_t *ppos) +{ + struct drm_device *dev = filp->private_data; + struct drm_i915_private *dev_priv = dev->dev_private; + char buf[20]; + u32 snpcr; + int val = 1; + + if (cnt > 0) { + if (cnt > sizeof (buf) - 1) + return -EINVAL; + + if (copy_from_user(buf, ubuf, cnt)) + return -EFAULT; + buf[cnt] = 0; + + val = simple_strtoul(buf, NULL, 0); + } + + if (val < 0 || val > 3) + return -EINVAL; + + DRM_DEBUG_DRIVER("Manually setting uncore sharing to %d\n", val); + + /* Update the cache sharing policy here as well */ + snpcr = I915_READ(GEN6_MBCUNIT_SNPCR); + snpcr &= ~GEN6_MBC_SNPCR_MASK; + snpcr |= (val << GEN6_MBC_SNPCR_SHIFT); + I915_WRITE(GEN6_MBCUNIT_SNPCR, snpcr); + + return cnt; +} + +static const struct file_operations i915_cache_sharing_fops = { + .owner = THIS_MODULE, + .open = i915_cache_sharing_open, + .read = i915_cache_sharing_read, + .write = i915_cache_sharing_write, + .llseek = default_llseek, +}; + /* As the drm_debugfs_init() routines are called before dev->dev_private is * allocated we need to hook into the minor for release. */ static int @@ -1288,6 +1525,97 @@ static int i915_wedged_create(struct dentry *root, struct drm_minor *minor) return drm_add_fake_info_node(minor, ent, &i915_wedged_fops); } +static int i915_forcewake_open(struct inode *inode, struct file *file) +{ + struct drm_device *dev = inode->i_private; + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + if (!IS_GEN6(dev)) + return 0; + + ret = mutex_lock_interruptible(&dev->struct_mutex); + if (ret) + return ret; + gen6_gt_force_wake_get(dev_priv); + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +int i915_forcewake_release(struct inode *inode, struct file *file) +{ + struct drm_device *dev = inode->i_private; + struct drm_i915_private *dev_priv = dev->dev_private; + + if (!IS_GEN6(dev)) + return 0; + + /* + * It's bad that we can potentially hang userspace if struct_mutex gets + * forever stuck. However, if we cannot acquire this lock it means that + * almost certainly the driver has hung, is not unload-able. Therefore + * hanging here is probably a minor inconvenience not to be seen my + * almost every user. + */ + mutex_lock(&dev->struct_mutex); + gen6_gt_force_wake_put(dev_priv); + mutex_unlock(&dev->struct_mutex); + + return 0; +} + +static const struct file_operations i915_forcewake_fops = { + .owner = THIS_MODULE, + .open = i915_forcewake_open, + .release = i915_forcewake_release, +}; + +static int i915_forcewake_create(struct dentry *root, struct drm_minor *minor) +{ + struct drm_device *dev = minor->dev; + struct dentry *ent; + + ent = debugfs_create_file("i915_forcewake_user", + S_IRUSR, + root, dev, + &i915_forcewake_fops); + if (IS_ERR(ent)) + return PTR_ERR(ent); + + return drm_add_fake_info_node(minor, ent, &i915_forcewake_fops); +} + +static int i915_max_freq_create(struct dentry *root, struct drm_minor *minor) +{ + struct drm_device *dev = minor->dev; + struct dentry *ent; + + ent = debugfs_create_file("i915_max_freq", + S_IRUGO | S_IWUSR, + root, dev, + &i915_max_freq_fops); + if (IS_ERR(ent)) + return PTR_ERR(ent); + + return drm_add_fake_info_node(minor, ent, &i915_max_freq_fops); +} + +static int i915_cache_sharing_create(struct dentry *root, struct drm_minor *minor) +{ + struct drm_device *dev = minor->dev; + struct dentry *ent; + + ent = debugfs_create_file("i915_cache_sharing", + S_IRUGO | S_IWUSR, + root, dev, + &i915_cache_sharing_fops); + if (IS_ERR(ent)) + return PTR_ERR(ent); + + return drm_add_fake_info_node(minor, ent, &i915_cache_sharing_fops); +} + static struct drm_info_list i915_debugfs_list[] = { {"i915_capabilities", i915_capabilities, 0}, {"i915_gem_objects", i915_gem_object_info, 0}, @@ -1319,11 +1647,14 @@ static struct drm_info_list i915_debugfs_list[] = { {"i915_inttoext_table", i915_inttoext_table, 0}, {"i915_drpc_info", i915_drpc_info, 0}, {"i915_emon_status", i915_emon_status, 0}, + {"i915_ring_freq_table", i915_ring_freq_table, 0}, {"i915_gfxec", i915_gfxec, 0}, {"i915_fbc_status", i915_fbc_status, 0}, {"i915_sr_status", i915_sr_status, 0}, {"i915_opregion", i915_opregion, 0}, {"i915_gem_framebuffer", i915_gem_framebuffer_info, 0}, + {"i915_context_status", i915_context_status, 0}, + {"i915_gen6_forcewake_count", i915_gen6_forcewake_count_info, 0}, }; #define I915_DEBUGFS_ENTRIES ARRAY_SIZE(i915_debugfs_list) @@ -1335,6 +1666,16 @@ int i915_debugfs_init(struct drm_minor *minor) if (ret) return ret; + ret = i915_forcewake_create(minor->debugfs_root, minor); + if (ret) + return ret; + ret = i915_max_freq_create(minor->debugfs_root, minor); + if (ret) + return ret; + ret = i915_cache_sharing_create(minor->debugfs_root, minor); + if (ret) + return ret; + return drm_debugfs_create_files(i915_debugfs_list, I915_DEBUGFS_ENTRIES, minor->debugfs_root, minor); @@ -1344,8 +1685,14 @@ void i915_debugfs_cleanup(struct drm_minor *minor) { drm_debugfs_remove_files(i915_debugfs_list, I915_DEBUGFS_ENTRIES, minor); + drm_debugfs_remove_files((struct drm_info_list *) &i915_forcewake_fops, + 1, minor); drm_debugfs_remove_files((struct drm_info_list *) &i915_wedged_fops, 1, minor); + drm_debugfs_remove_files((struct drm_info_list *) &i915_max_freq_fops, + 1, minor); + drm_debugfs_remove_files((struct drm_info_list *) &i915_cache_sharing_fops, + 1, minor); } #endif /* CONFIG_DEBUG_FS */ diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c index 12876f2795d..8a3942c4f09 100644 --- a/drivers/gpu/drm/i915/i915_dma.c +++ b/drivers/gpu/drm/i915/i915_dma.c @@ -61,7 +61,6 @@ static void i915_write_hws_pga(struct drm_device *dev) static int i915_init_phys_hws(struct drm_device *dev) { drm_i915_private_t *dev_priv = dev->dev_private; - struct intel_ring_buffer *ring = LP_RING(dev_priv); /* Program Hardware Status Page */ dev_priv->status_page_dmah = @@ -71,10 +70,9 @@ static int i915_init_phys_hws(struct drm_device *dev) DRM_ERROR("Can not allocate hardware status page\n"); return -ENOMEM; } - ring->status_page.page_addr = - (void __force __iomem *)dev_priv->status_page_dmah->vaddr; - memset_io(ring->status_page.page_addr, 0, PAGE_SIZE); + memset_io((void __force __iomem *)dev_priv->status_page_dmah->vaddr, + 0, PAGE_SIZE); i915_write_hws_pga(dev); @@ -571,7 +569,7 @@ static int i915_quiescent(struct drm_device *dev) struct intel_ring_buffer *ring = LP_RING(dev->dev_private); i915_kernel_lost_context(dev); - return intel_wait_ring_buffer(ring, ring->size - 8); + return intel_wait_ring_idle(ring); } static int i915_flush_ioctl(struct drm_device *dev, void *data, @@ -1073,6 +1071,9 @@ static void i915_setup_compression(struct drm_device *dev, int size) unsigned long cfb_base; unsigned long ll_base = 0; + /* Just in case the BIOS is doing something questionable. */ + intel_disable_fbc(dev); + compressed_fb = drm_mm_search_free(&dev_priv->mm.stolen, size, 4096, 0); if (compressed_fb) compressed_fb = drm_mm_get_block(compressed_fb, size, 4096); @@ -1099,7 +1100,6 @@ static void i915_setup_compression(struct drm_device *dev, int size) dev_priv->cfb_size = size; - intel_disable_fbc(dev); dev_priv->compressed_fb = compressed_fb; if (HAS_PCH_SPLIT(dev)) I915_WRITE(ILK_DPFC_CB_BASE, compressed_fb->start); @@ -1176,11 +1176,11 @@ static bool i915_switcheroo_can_switch(struct pci_dev *pdev) return can_switch; } -static int i915_load_modeset_init(struct drm_device *dev) +static int i915_load_gem_init(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; unsigned long prealloc_size, gtt_size, mappable_size; - int ret = 0; + int ret; prealloc_size = dev_priv->mm.gtt->stolen_size; gtt_size = dev_priv->mm.gtt->gtt_total_entries << PAGE_SHIFT; @@ -1204,7 +1204,7 @@ static int i915_load_modeset_init(struct drm_device *dev) ret = i915_gem_init_ringbuffer(dev); mutex_unlock(&dev->struct_mutex); if (ret) - goto out; + return ret; /* Try to set up FBC with a reasonable compressed buffer size */ if (I915_HAS_FBC(dev) && i915_powersave) { @@ -1222,6 +1222,13 @@ static int i915_load_modeset_init(struct drm_device *dev) /* Allow hardware batchbuffers unless told otherwise. */ dev_priv->allow_batchbuffer = 1; + return 0; +} + +static int i915_load_modeset_init(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; ret = intel_parse_bios(dev); if (ret) @@ -1236,7 +1243,7 @@ static int i915_load_modeset_init(struct drm_device *dev) */ ret = vga_client_register(dev->pdev, dev, NULL, i915_vga_set_decode); if (ret && ret != -ENODEV) - goto cleanup_ringbuffer; + goto out; intel_register_dsm_handler(); @@ -1253,10 +1260,16 @@ static int i915_load_modeset_init(struct drm_device *dev) intel_modeset_init(dev); - ret = drm_irq_install(dev); + ret = i915_load_gem_init(dev); if (ret) goto cleanup_vga_switcheroo; + intel_modeset_gem_init(dev); + + ret = drm_irq_install(dev); + if (ret) + goto cleanup_gem; + /* Always safe in the mode setting case. */ /* FIXME: do pre/post-mode set stuff in core KMS code */ dev->vblank_disable_allowed = 1; @@ -1274,14 +1287,14 @@ static int i915_load_modeset_init(struct drm_device *dev) cleanup_irq: drm_irq_uninstall(dev); +cleanup_gem: + mutex_lock(&dev->struct_mutex); + i915_gem_cleanup_ringbuffer(dev); + mutex_unlock(&dev->struct_mutex); cleanup_vga_switcheroo: vga_switcheroo_unregister_client(dev->pdev); cleanup_vga_client: vga_client_register(dev->pdev, NULL, NULL, NULL); -cleanup_ringbuffer: - mutex_lock(&dev->struct_mutex); - i915_gem_cleanup_ringbuffer(dev); - mutex_unlock(&dev->struct_mutex); out: return ret; } @@ -1930,7 +1943,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) if (!dev_priv->mm.gtt) { DRM_ERROR("Failed to initialize GTT\n"); ret = -ENODEV; - goto out_iomapfree; + goto out_rmmap; } agp_size = dev_priv->mm.gtt->gtt_mappable_entries << PAGE_SHIFT; @@ -1974,18 +1987,13 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) if (dev_priv->wq == NULL) { DRM_ERROR("Failed to create our workqueue.\n"); ret = -ENOMEM; - goto out_iomapfree; + goto out_mtrrfree; } /* enable GEM by default */ dev_priv->has_gem = 1; - dev->driver->get_vblank_counter = i915_get_vblank_counter; - dev->max_vblank_count = 0xffffff; /* only 24 bits of frame count */ - if (IS_G4X(dev) || IS_GEN5(dev) || IS_GEN6(dev)) { - dev->max_vblank_count = 0xffffffff; /* full 32 bit counter */ - dev->driver->get_vblank_counter = gm45_get_vblank_counter; - } + intel_irq_init(dev); /* Try to make sure MCHBAR is enabled before poking at it */ intel_setup_mchbar(dev); @@ -2025,6 +2033,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) spin_lock_init(&dev_priv->irq_lock); spin_lock_init(&dev_priv->error_lock); + spin_lock_init(&dev_priv->rps_lock); if (IS_MOBILE(dev) || !IS_GEN2(dev)) dev_priv->num_pipe = 2; @@ -2065,13 +2074,21 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) return 0; out_gem_unload: + if (dev_priv->mm.inactive_shrinker.shrink) + unregister_shrinker(&dev_priv->mm.inactive_shrinker); + if (dev->pdev->msi_enabled) pci_disable_msi(dev->pdev); intel_teardown_gmbus(dev); intel_teardown_mchbar(dev); destroy_workqueue(dev_priv->wq); -out_iomapfree: +out_mtrrfree: + if (dev_priv->mm.gtt_mtrr >= 0) { + mtrr_del(dev_priv->mm.gtt_mtrr, dev->agp->base, + dev->agp->agp_info.aper_size * 1024 * 1024); + dev_priv->mm.gtt_mtrr = -1; + } io_mapping_free(dev_priv->mm.gtt_mapping); out_rmmap: pci_iounmap(dev->pdev, dev_priv->regs); @@ -2144,9 +2161,8 @@ int i915_driver_unload(struct drm_device *dev) /* Flush any outstanding unpin_work. */ flush_workqueue(dev_priv->wq); - i915_gem_free_all_phys_object(dev); - mutex_lock(&dev->struct_mutex); + i915_gem_free_all_phys_object(dev); i915_gem_cleanup_ringbuffer(dev); mutex_unlock(&dev->struct_mutex); if (I915_HAS_FBC(dev) && i915_powersave) diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index c34a8dd31d0..ce045a8cf82 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -37,35 +37,70 @@ #include <linux/console.h> #include "drm_crtc_helper.h" -static int i915_modeset = -1; +static int i915_modeset __read_mostly = -1; module_param_named(modeset, i915_modeset, int, 0400); +MODULE_PARM_DESC(modeset, + "Use kernel modesetting [KMS] (0=DRM_I915_KMS from .config, " + "1=on, -1=force vga console preference [default])"); -unsigned int i915_fbpercrtc = 0; +unsigned int i915_fbpercrtc __always_unused = 0; module_param_named(fbpercrtc, i915_fbpercrtc, int, 0400); -int i915_panel_ignore_lid = 0; +int i915_panel_ignore_lid __read_mostly = 0; module_param_named(panel_ignore_lid, i915_panel_ignore_lid, int, 0600); +MODULE_PARM_DESC(panel_ignore_lid, + "Override lid status (0=autodetect [default], 1=lid open, " + "-1=lid closed)"); -unsigned int i915_powersave = 1; +unsigned int i915_powersave __read_mostly = 1; module_param_named(powersave, i915_powersave, int, 0600); +MODULE_PARM_DESC(powersave, + "Enable powersavings, fbc, downclocking, etc. (default: true)"); -unsigned int i915_semaphores = 1; +unsigned int i915_semaphores __read_mostly = 0; module_param_named(semaphores, i915_semaphores, int, 0600); +MODULE_PARM_DESC(semaphores, + "Use semaphores for inter-ring sync (default: false)"); -unsigned int i915_enable_rc6 = 0; +unsigned int i915_enable_rc6 __read_mostly = 0; module_param_named(i915_enable_rc6, i915_enable_rc6, int, 0600); +MODULE_PARM_DESC(i915_enable_rc6, + "Enable power-saving render C-state 6 (default: true)"); -unsigned int i915_lvds_downclock = 0; +unsigned int i915_enable_fbc __read_mostly = 1; +module_param_named(i915_enable_fbc, i915_enable_fbc, int, 0600); +MODULE_PARM_DESC(i915_enable_fbc, + "Enable frame buffer compression for power savings " + "(default: false)"); + +unsigned int i915_lvds_downclock __read_mostly = 0; module_param_named(lvds_downclock, i915_lvds_downclock, int, 0400); +MODULE_PARM_DESC(lvds_downclock, + "Use panel (LVDS/eDP) downclocking for power savings " + "(default: false)"); -unsigned int i915_panel_use_ssc = 1; +unsigned int i915_panel_use_ssc __read_mostly = 1; module_param_named(lvds_use_ssc, i915_panel_use_ssc, int, 0600); +MODULE_PARM_DESC(lvds_use_ssc, + "Use Spread Spectrum Clock with panels [LVDS/eDP] " + "(default: true)"); -int i915_vbt_sdvo_panel_type = -1; +int i915_vbt_sdvo_panel_type __read_mostly = -1; module_param_named(vbt_sdvo_panel_type, i915_vbt_sdvo_panel_type, int, 0600); +MODULE_PARM_DESC(vbt_sdvo_panel_type, + "Override selection of SDVO panel mode in the VBT " + "(default: auto)"); -static bool i915_try_reset = true; +static bool i915_try_reset __read_mostly = true; module_param_named(reset, i915_try_reset, bool, 0600); +MODULE_PARM_DESC(reset, "Attempt GPU resets (default: true)"); + +bool i915_enable_hangcheck __read_mostly = true; +module_param_named(enable_hangcheck, i915_enable_hangcheck, bool, 0644); +MODULE_PARM_DESC(enable_hangcheck, + "Periodically check GPU activity for detecting hangs. " + "WARNING: Disabling this can cause system wide hangs. " + "(default: true)"); static struct drm_driver driver; extern int intel_agp_enabled; @@ -169,7 +204,7 @@ static const struct intel_device_info intel_ironlake_d_info = { static const struct intel_device_info intel_ironlake_m_info = { .gen = 5, .is_mobile = 1, .need_gfx_hws = 1, .has_hotplug = 1, - .has_fbc = 0, /* disabled due to buggy hardware */ + .has_fbc = 1, .has_bsd_ring = 1, }; @@ -188,6 +223,21 @@ static const struct intel_device_info intel_sandybridge_m_info = { .has_blt_ring = 1, }; +static const struct intel_device_info intel_ivybridge_d_info = { + .is_ivybridge = 1, .gen = 7, + .need_gfx_hws = 1, .has_hotplug = 1, + .has_bsd_ring = 1, + .has_blt_ring = 1, +}; + +static const struct intel_device_info intel_ivybridge_m_info = { + .is_ivybridge = 1, .gen = 7, .is_mobile = 1, + .need_gfx_hws = 1, .has_hotplug = 1, + .has_fbc = 0, /* FBC is not enabled on Ivybridge mobile yet */ + .has_bsd_ring = 1, + .has_blt_ring = 1, +}; + static const struct pci_device_id pciidlist[] = { /* aka */ INTEL_VGA_DEVICE(0x3577, &intel_i830_info), /* I830_M */ INTEL_VGA_DEVICE(0x2562, &intel_845g_info), /* 845_G */ @@ -227,6 +277,11 @@ static const struct pci_device_id pciidlist[] = { /* aka */ INTEL_VGA_DEVICE(0x0116, &intel_sandybridge_m_info), INTEL_VGA_DEVICE(0x0126, &intel_sandybridge_m_info), INTEL_VGA_DEVICE(0x010A, &intel_sandybridge_d_info), + INTEL_VGA_DEVICE(0x0156, &intel_ivybridge_m_info), /* GT1 mobile */ + INTEL_VGA_DEVICE(0x0166, &intel_ivybridge_m_info), /* GT2 mobile */ + INTEL_VGA_DEVICE(0x0152, &intel_ivybridge_d_info), /* GT1 desktop */ + INTEL_VGA_DEVICE(0x0162, &intel_ivybridge_d_info), /* GT2 desktop */ + INTEL_VGA_DEVICE(0x015a, &intel_ivybridge_d_info), /* GT1 server */ {0, 0, 0} }; @@ -235,7 +290,9 @@ MODULE_DEVICE_TABLE(pci, pciidlist); #endif #define INTEL_PCH_DEVICE_ID_MASK 0xff00 +#define INTEL_PCH_IBX_DEVICE_ID_TYPE 0x3b00 #define INTEL_PCH_CPT_DEVICE_ID_TYPE 0x1c00 +#define INTEL_PCH_PPT_DEVICE_ID_TYPE 0x1e00 void intel_detect_pch (struct drm_device *dev) { @@ -254,16 +311,23 @@ void intel_detect_pch (struct drm_device *dev) int id; id = pch->device & INTEL_PCH_DEVICE_ID_MASK; - if (id == INTEL_PCH_CPT_DEVICE_ID_TYPE) { + if (id == INTEL_PCH_IBX_DEVICE_ID_TYPE) { + dev_priv->pch_type = PCH_IBX; + DRM_DEBUG_KMS("Found Ibex Peak PCH\n"); + } else if (id == INTEL_PCH_CPT_DEVICE_ID_TYPE) { dev_priv->pch_type = PCH_CPT; DRM_DEBUG_KMS("Found CougarPoint PCH\n"); + } else if (id == INTEL_PCH_PPT_DEVICE_ID_TYPE) { + /* PantherPoint is CPT compatible */ + dev_priv->pch_type = PCH_CPT; + DRM_DEBUG_KMS("Found PatherPoint PCH\n"); } } pci_dev_put(pch); } } -void __gen6_gt_force_wake_get(struct drm_i915_private *dev_priv) +static void __gen6_gt_force_wake_get(struct drm_i915_private *dev_priv) { int count; @@ -279,20 +343,51 @@ void __gen6_gt_force_wake_get(struct drm_i915_private *dev_priv) udelay(10); } -void __gen6_gt_force_wake_put(struct drm_i915_private *dev_priv) +/* + * Generally this is called implicitly by the register read function. However, + * if some sequence requires the GT to not power down then this function should + * be called at the beginning of the sequence followed by a call to + * gen6_gt_force_wake_put() at the end of the sequence. + */ +void gen6_gt_force_wake_get(struct drm_i915_private *dev_priv) +{ + WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex)); + + /* Forcewake is atomic in case we get in here without the lock */ + if (atomic_add_return(1, &dev_priv->forcewake_count) == 1) + __gen6_gt_force_wake_get(dev_priv); +} + +static void __gen6_gt_force_wake_put(struct drm_i915_private *dev_priv) { I915_WRITE_NOTRACE(FORCEWAKE, 0); POSTING_READ(FORCEWAKE); } +/* + * see gen6_gt_force_wake_get() + */ +void gen6_gt_force_wake_put(struct drm_i915_private *dev_priv) +{ + WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex)); + + if (atomic_dec_and_test(&dev_priv->forcewake_count)) + __gen6_gt_force_wake_put(dev_priv); +} + void __gen6_gt_wait_for_fifo(struct drm_i915_private *dev_priv) { - int loop = 500; - u32 fifo = I915_READ_NOTRACE(GT_FIFO_FREE_ENTRIES); - while (fifo < 20 && loop--) { - udelay(10); - fifo = I915_READ_NOTRACE(GT_FIFO_FREE_ENTRIES); + if (dev_priv->gt_fifo_count < GT_FIFO_NUM_RESERVED_ENTRIES ) { + int loop = 500; + u32 fifo = I915_READ_NOTRACE(GT_FIFO_FREE_ENTRIES); + while (fifo <= GT_FIFO_NUM_RESERVED_ENTRIES && loop--) { + udelay(10); + fifo = I915_READ_NOTRACE(GT_FIFO_FREE_ENTRIES); + } + WARN_ON(loop < 0 && fifo <= GT_FIFO_NUM_RESERVED_ENTRIES); + dev_priv->gt_fifo_count = fifo; } + dev_priv->gt_fifo_count--; } static int i915_drm_freeze(struct drm_device *dev) @@ -519,8 +614,12 @@ int i915_reset(struct drm_device *dev, u8 flags) if (get_seconds() - dev_priv->last_gpu_reset < 5) { DRM_ERROR("GPU hanging too fast, declaring wedged!\n"); } else switch (INTEL_INFO(dev)->gen) { + case 7: case 6: ret = gen6_do_reset(dev, flags); + /* If reset with a user forcewake, try to restore */ + if (atomic_read(&dev_priv->forcewake_count)) + __gen6_gt_force_wake_get(dev_priv); break; case 5: ret = ironlake_do_reset(dev, flags); @@ -704,14 +803,6 @@ static struct drm_driver driver = { .resume = i915_resume, .device_is_agp = i915_driver_device_is_agp, - .enable_vblank = i915_enable_vblank, - .disable_vblank = i915_disable_vblank, - .get_vblank_timestamp = i915_get_vblank_timestamp, - .get_scanout_position = i915_get_crtc_scanoutpos, - .irq_preinstall = i915_driver_irq_preinstall, - .irq_postinstall = i915_driver_irq_postinstall, - .irq_uninstall = i915_driver_irq_uninstall, - .irq_handler = i915_driver_irq_handler, .reclaim_buffers = drm_core_reclaim_buffers, .master_create = i915_master_create, .master_destroy = i915_master_destroy, diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index 1c1b27c97e5..7916bd97d5c 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -36,6 +36,7 @@ #include <linux/io-mapping.h> #include <linux/i2c.h> #include <drm/intel-gtt.h> +#include <linux/backlight.h> /* General customization: */ @@ -188,7 +189,7 @@ struct drm_i915_error_state { u32 dirty:1; u32 purgeable:1; u32 ring:4; - u32 agp_type:1; + u32 cache_level:2; } *active_bo, *pinned_bo; u32 active_bo_count, pinned_bo_count; struct intel_overlay_error_state *overlay; @@ -203,12 +204,24 @@ struct drm_i915_display_funcs { int (*get_display_clock_speed)(struct drm_device *dev); int (*get_fifo_size)(struct drm_device *dev, int plane); void (*update_wm)(struct drm_device *dev); + int (*crtc_mode_set)(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, + struct drm_framebuffer *old_fb); + void (*fdi_link_train)(struct drm_crtc *crtc); + void (*init_clock_gating)(struct drm_device *dev); + void (*init_pch_clock_gating)(struct drm_device *dev); + int (*queue_flip)(struct drm_device *dev, struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj); + int (*update_plane)(struct drm_crtc *crtc, struct drm_framebuffer *fb, + int x, int y); /* clock updates for mode set */ /* cursor updates */ /* render clock increase/decrease */ /* display clock increase/decrease */ /* pll clock increase/decrease */ - /* clock gating init */ }; struct intel_device_info { @@ -223,6 +236,7 @@ struct intel_device_info { u8 is_pineview : 1; u8 is_broadwater : 1; u8 is_crestline : 1; + u8 is_ivybridge : 1; u8 has_fbc : 1; u8 has_pipe_cxsr : 1; u8 has_hotplug : 1; @@ -242,6 +256,7 @@ enum no_fbc_reason { FBC_BAD_PLANE, /* fbc not supported on plane */ FBC_NOT_TILED, /* buffer not tiled */ FBC_MULTIPLE_PIPES, /* more than one pipe active */ + FBC_MODULE_PARAM, }; enum intel_pch { @@ -250,8 +265,10 @@ enum intel_pch { }; #define QUIRK_PIPEA_FORCE (1<<0) +#define QUIRK_LVDS_SSC_DISABLE (1<<1) struct intel_fbdev; +struct intel_fbc_work; typedef struct drm_i915_private { struct drm_device *dev; @@ -262,6 +279,7 @@ typedef struct drm_i915_private { int relative_constants_mode; void __iomem *regs; + u32 gt_fifo_count; struct intel_gmbus { struct i2c_adapter adapter; @@ -316,11 +334,10 @@ typedef struct drm_i915_private { uint32_t last_instdone1; unsigned long cfb_size; - unsigned long cfb_pitch; - unsigned long cfb_offset; - int cfb_fence; - int cfb_plane; + unsigned int cfb_fb; + enum plane cfb_plane; int cfb_y; + struct intel_fbc_work *fbc_work; struct intel_opregion opregion; @@ -528,6 +545,7 @@ typedef struct drm_i915_private { u32 savePIPEB_LINK_M1; u32 savePIPEB_LINK_N1; u32 saveMCHBAR_RENDER_STANDBY; + u32 savePCH_PORT_HOTPLUG; struct { /** Bridge to intel-gtt-ko */ @@ -673,9 +691,14 @@ typedef struct drm_i915_private { int child_dev_num; struct child_device_config *child_dev; struct drm_connector *int_lvds_connector; + struct drm_connector *int_edp_connector; bool mchbar_need_disable; + struct work_struct rps_work; + spinlock_t rps_lock; + u32 pm_iir; + u8 cur_delay; u8 min_delay; u8 max_delay; @@ -702,9 +725,20 @@ typedef struct drm_i915_private { /* list of fbdev register on this device */ struct intel_fbdev *fbdev; + struct backlight_device *backlight; + struct drm_property *broadcast_rgb_property; + struct drm_property *force_audio_property; + + atomic_t forcewake_count; } drm_i915_private_t; +enum i915_cache_level { + I915_CACHE_NONE, + I915_CACHE_LLC, + I915_CACHE_LLC_MLC, /* gen6+ */ +}; + struct drm_i915_gem_object { struct drm_gem_object base; @@ -791,6 +825,8 @@ struct drm_i915_gem_object { unsigned int pending_fenced_gpu_access:1; unsigned int fenced_gpu_access:1; + unsigned int cache_level:2; + struct page **pages; /** @@ -827,8 +863,6 @@ struct drm_i915_gem_object { /** Record of address bit 17 of each page at last unbind. */ unsigned long *bit_17; - /** AGP mapping type (AGP_USER_MEMORY or AGP_USER_CACHED_MEMORY */ - uint32_t agp_type; /** * If present, while GEM_DOMAIN_CPU is in the read domain this array @@ -888,13 +922,6 @@ struct drm_i915_file_private { } mm; }; -enum intel_chip_family { - CHIP_I8XX = 0x01, - CHIP_I9XX = 0x02, - CHIP_I915 = 0x04, - CHIP_I965 = 0x08, -}; - #define INTEL_INFO(dev) (((struct drm_i915_private *) (dev)->dev_private)->info) #define IS_I830(dev) ((dev)->pci_device == 0x3577) @@ -915,13 +942,21 @@ enum intel_chip_family { #define IS_G33(dev) (INTEL_INFO(dev)->is_g33) #define IS_IRONLAKE_D(dev) ((dev)->pci_device == 0x0042) #define IS_IRONLAKE_M(dev) ((dev)->pci_device == 0x0046) +#define IS_IVYBRIDGE(dev) (INTEL_INFO(dev)->is_ivybridge) #define IS_MOBILE(dev) (INTEL_INFO(dev)->is_mobile) +/* + * The genX designation typically refers to the render engine, so render + * capability related checks should use IS_GEN, while display and other checks + * have their own (e.g. HAS_PCH_SPLIT for ILK+ display, IS_foo for particular + * chips, etc.). + */ #define IS_GEN2(dev) (INTEL_INFO(dev)->gen == 2) #define IS_GEN3(dev) (INTEL_INFO(dev)->gen == 3) #define IS_GEN4(dev) (INTEL_INFO(dev)->gen == 4) #define IS_GEN5(dev) (INTEL_INFO(dev)->gen == 5) #define IS_GEN6(dev) (INTEL_INFO(dev)->gen == 6) +#define IS_GEN7(dev) (INTEL_INFO(dev)->gen == 7) #define HAS_BSD(dev) (INTEL_INFO(dev)->has_bsd_ring) #define HAS_BLT(dev) (INTEL_INFO(dev)->has_blt_ring) @@ -948,8 +983,8 @@ enum intel_chip_family { #define HAS_PIPE_CXSR(dev) (INTEL_INFO(dev)->has_pipe_cxsr) #define I915_HAS_FBC(dev) (INTEL_INFO(dev)->has_fbc) -#define HAS_PCH_SPLIT(dev) (IS_GEN5(dev) || IS_GEN6(dev)) -#define HAS_PIPE_CONTROL(dev) (IS_GEN5(dev) || IS_GEN6(dev)) +#define HAS_PCH_SPLIT(dev) (IS_GEN5(dev) || IS_GEN6(dev) || IS_IVYBRIDGE(dev)) +#define HAS_PIPE_CONTROL(dev) (INTEL_INFO(dev)->gen >= 5) #define INTEL_PCH_TYPE(dev) (((struct drm_i915_private *)(dev)->dev_private)->pch_type) #define HAS_PCH_CPT(dev) (INTEL_PCH_TYPE(dev) == PCH_CPT) @@ -959,19 +994,19 @@ enum intel_chip_family { extern struct drm_ioctl_desc i915_ioctls[]; extern int i915_max_ioctl; -extern unsigned int i915_fbpercrtc; -extern int i915_panel_ignore_lid; -extern unsigned int i915_powersave; -extern unsigned int i915_semaphores; -extern unsigned int i915_lvds_downclock; -extern unsigned int i915_panel_use_ssc; -extern int i915_vbt_sdvo_panel_type; -extern unsigned int i915_enable_rc6; +extern unsigned int i915_fbpercrtc __always_unused; +extern int i915_panel_ignore_lid __read_mostly; +extern unsigned int i915_powersave __read_mostly; +extern unsigned int i915_semaphores __read_mostly; +extern unsigned int i915_lvds_downclock __read_mostly; +extern unsigned int i915_panel_use_ssc __read_mostly; +extern int i915_vbt_sdvo_panel_type __read_mostly; +extern unsigned int i915_enable_rc6 __read_mostly; +extern unsigned int i915_enable_fbc __read_mostly; +extern bool i915_enable_hangcheck __read_mostly; extern int i915_suspend(struct drm_device *dev, pm_message_t state); extern int i915_resume(struct drm_device *dev); -extern void i915_save_display(struct drm_device *dev); -extern void i915_restore_display(struct drm_device *dev); extern int i915_master_create(struct drm_device *dev, struct drm_master *master); extern void i915_master_destroy(struct drm_device *dev, struct drm_master *master); @@ -1006,18 +1041,12 @@ extern int i915_irq_emit(struct drm_device *dev, void *data, extern int i915_irq_wait(struct drm_device *dev, void *data, struct drm_file *file_priv); -extern irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS); -extern void i915_driver_irq_preinstall(struct drm_device * dev); -extern int i915_driver_irq_postinstall(struct drm_device *dev); -extern void i915_driver_irq_uninstall(struct drm_device * dev); +extern void intel_irq_init(struct drm_device *dev); + extern int i915_vblank_pipe_set(struct drm_device *dev, void *data, struct drm_file *file_priv); extern int i915_vblank_pipe_get(struct drm_device *dev, void *data, struct drm_file *file_priv); -extern int i915_enable_vblank(struct drm_device *dev, int crtc); -extern void i915_disable_vblank(struct drm_device *dev, int crtc); -extern u32 i915_get_vblank_counter(struct drm_device *dev, int crtc); -extern u32 gm45_get_vblank_counter(struct drm_device *dev, int crtc); extern int i915_vblank_swap(struct drm_device *dev, void *data, struct drm_file *file_priv); @@ -1028,13 +1057,6 @@ void i915_disable_pipestat(drm_i915_private_t *dev_priv, int pipe, u32 mask); void intel_enable_asle (struct drm_device *dev); -int i915_get_vblank_timestamp(struct drm_device *dev, int crtc, - int *max_error, - struct timeval *vblank_time, - unsigned flags); - -int i915_get_crtc_scanoutpos(struct drm_device *dev, int pipe, - int *vpos, int *hpos); #ifdef CONFIG_DEBUG_FS extern void i915_destroy_error_state(struct drm_device *dev); @@ -1151,7 +1173,7 @@ void i915_gem_clflush_object(struct drm_i915_gem_object *obj); int __must_check i915_gem_object_set_domain(struct drm_i915_gem_object *obj, uint32_t read_domains, uint32_t write_domain); -int __must_check i915_gem_object_flush_gpu(struct drm_i915_gem_object *obj); +int __must_check i915_gem_object_finish_gpu(struct drm_i915_gem_object *obj); int __must_check i915_gem_init_ringbuffer(struct drm_device *dev); void i915_gem_cleanup_ringbuffer(struct drm_device *dev); void i915_gem_do_init(struct drm_device *dev, @@ -1170,7 +1192,8 @@ int __must_check i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj, bool write); int __must_check -i915_gem_object_set_to_display_plane(struct drm_i915_gem_object *obj, +i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj, + u32 alignment, struct intel_ring_buffer *pipelined); int i915_gem_attach_phys_object(struct drm_device *dev, struct drm_i915_gem_object *obj, @@ -1182,11 +1205,18 @@ void i915_gem_free_all_phys_object(struct drm_device *dev); void i915_gem_release(struct drm_device *dev, struct drm_file *file); uint32_t -i915_gem_get_unfenced_gtt_alignment(struct drm_i915_gem_object *obj); +i915_gem_get_unfenced_gtt_alignment(struct drm_device *dev, + uint32_t size, + int tiling_mode); + +int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj, + enum i915_cache_level cache_level); /* i915_gem_gtt.c */ void i915_gem_restore_gtt_mappings(struct drm_device *dev); int __must_check i915_gem_gtt_bind_object(struct drm_i915_gem_object *obj); +void i915_gem_gtt_rebind_object(struct drm_i915_gem_object *obj, + enum i915_cache_level cache_level); void i915_gem_gtt_unbind_object(struct drm_i915_gem_object *obj); /* i915_gem_evict.c */ @@ -1265,14 +1295,11 @@ static inline void intel_unregister_dsm_handler(void) { return; } /* modesetting */ extern void intel_modeset_init(struct drm_device *dev); +extern void intel_modeset_gem_init(struct drm_device *dev); extern void intel_modeset_cleanup(struct drm_device *dev); extern int intel_modeset_vga_set_state(struct drm_device *dev, bool state); -extern void i8xx_disable_fbc(struct drm_device *dev); -extern void g4x_disable_fbc(struct drm_device *dev); -extern void ironlake_disable_fbc(struct drm_device *dev); -extern void intel_disable_fbc(struct drm_device *dev); -extern void intel_enable_fbc(struct drm_crtc *crtc, unsigned long interval); extern bool intel_fbc_enabled(struct drm_device *dev); +extern void intel_disable_fbc(struct drm_device *dev); extern bool ironlake_set_drps(struct drm_device *dev, u8 val); extern void ironlake_enable_rc6(struct drm_device *dev); extern void gen6_set_rps(struct drm_device *dev, u8 val); @@ -1312,13 +1339,34 @@ extern void intel_display_print_error_state(struct seq_file *m, LOCK_TEST_WITH_RETURN(dev, file); \ } while (0) +/* On SNB platform, before reading ring registers forcewake bit + * must be set to prevent GT core from power down and stale values being + * returned. + */ +void gen6_gt_force_wake_get(struct drm_i915_private *dev_priv); +void gen6_gt_force_wake_put(struct drm_i915_private *dev_priv); +void __gen6_gt_wait_for_fifo(struct drm_i915_private *dev_priv); + +/* We give fast paths for the really cool registers */ +#define NEEDS_FORCE_WAKE(dev_priv, reg) \ + (((dev_priv)->info->gen >= 6) && \ + ((reg) < 0x40000) && \ + ((reg) != FORCEWAKE)) #define __i915_read(x, y) \ static inline u##x i915_read##x(struct drm_i915_private *dev_priv, u32 reg) { \ - u##x val = read##y(dev_priv->regs + reg); \ + u##x val = 0; \ + if (NEEDS_FORCE_WAKE((dev_priv), (reg))) { \ + gen6_gt_force_wake_get(dev_priv); \ + val = read##y(dev_priv->regs + reg); \ + gen6_gt_force_wake_put(dev_priv); \ + } else { \ + val = read##y(dev_priv->regs + reg); \ + } \ trace_i915_reg_rw(false, reg, val, sizeof(val)); \ return val; \ } + __i915_read(8, b) __i915_read(16, w) __i915_read(32, l) @@ -1328,6 +1376,9 @@ __i915_read(64, q) #define __i915_write(x, y) \ static inline void i915_write##x(struct drm_i915_private *dev_priv, u32 reg, u##x val) { \ trace_i915_reg_rw(true, reg, val, sizeof(val)); \ + if (NEEDS_FORCE_WAKE((dev_priv), (reg))) { \ + __gen6_gt_wait_for_fifo(dev_priv); \ + } \ write##y(val, dev_priv->regs + reg); \ } __i915_write(8, b) @@ -1356,33 +1407,4 @@ __i915_write(64, q) #define POSTING_READ16(reg) (void)I915_READ16_NOTRACE(reg) -/* On SNB platform, before reading ring registers forcewake bit - * must be set to prevent GT core from power down and stale values being - * returned. - */ -void __gen6_gt_force_wake_get(struct drm_i915_private *dev_priv); -void __gen6_gt_force_wake_put(struct drm_i915_private *dev_priv); -void __gen6_gt_wait_for_fifo(struct drm_i915_private *dev_priv); - -static inline u32 i915_gt_read(struct drm_i915_private *dev_priv, u32 reg) -{ - u32 val; - - if (dev_priv->info->gen >= 6) { - __gen6_gt_force_wake_get(dev_priv); - val = I915_READ(reg); - __gen6_gt_force_wake_put(dev_priv); - } else - val = I915_READ(reg); - - return val; -} - -static inline void i915_gt_write(struct drm_i915_private *dev_priv, - u32 reg, u32 val) -{ - if (dev_priv->info->gen >= 6) - __gen6_gt_wait_for_fifo(dev_priv); - I915_WRITE(reg, val); -} #endif diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 7ce3f353af3..a546a71fb06 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -31,6 +31,7 @@ #include "i915_drv.h" #include "i915_trace.h" #include "intel_drv.h" +#include <linux/shmem_fs.h> #include <linux/slab.h> #include <linux/swap.h> #include <linux/pci.h> @@ -56,9 +57,7 @@ static int i915_gem_phys_pwrite(struct drm_device *dev, static void i915_gem_free_object_tail(struct drm_i915_gem_object *obj); static int i915_gem_inactive_shrink(struct shrinker *shrinker, - int nr_to_scan, - gfp_t gfp_mask); - + struct shrink_control *sc); /* some bookkeeping */ static void i915_gem_info_add_obj(struct drm_i915_private *dev_priv, @@ -356,13 +355,12 @@ i915_gem_shmem_pread_fast(struct drm_device *dev, * page_offset = offset within page * page_length = bytes to copy for this page */ - page_offset = offset & (PAGE_SIZE-1); + page_offset = offset_in_page(offset); page_length = remain; if ((page_offset + remain) > PAGE_SIZE) page_length = PAGE_SIZE - page_offset; - page = read_cache_page_gfp(mapping, offset >> PAGE_SHIFT, - GFP_HIGHUSER | __GFP_RECLAIMABLE); + page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT); if (IS_ERR(page)) return PTR_ERR(page); @@ -455,9 +453,9 @@ i915_gem_shmem_pread_slow(struct drm_device *dev, * data_page_offset = offset with data_page_index page. * page_length = bytes to copy for this page */ - shmem_page_offset = offset & ~PAGE_MASK; + shmem_page_offset = offset_in_page(offset); data_page_index = data_ptr / PAGE_SIZE - first_data_page; - data_page_offset = data_ptr & ~PAGE_MASK; + data_page_offset = offset_in_page(data_ptr); page_length = remain; if ((shmem_page_offset + page_length) > PAGE_SIZE) @@ -465,10 +463,11 @@ i915_gem_shmem_pread_slow(struct drm_device *dev, if ((data_page_offset + page_length) > PAGE_SIZE) page_length = PAGE_SIZE - data_page_offset; - page = read_cache_page_gfp(mapping, offset >> PAGE_SHIFT, - GFP_HIGHUSER | __GFP_RECLAIMABLE); - if (IS_ERR(page)) - return PTR_ERR(page); + page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT); + if (IS_ERR(page)) { + ret = PTR_ERR(page); + goto out; + } if (do_bit17_swizzling) { slow_shmem_bit17_copy(page, @@ -640,8 +639,8 @@ i915_gem_gtt_pwrite_fast(struct drm_device *dev, * page_offset = offset within page * page_length = bytes to copy for this page */ - page_base = (offset & ~(PAGE_SIZE-1)); - page_offset = offset & (PAGE_SIZE-1); + page_base = offset & PAGE_MASK; + page_offset = offset_in_page(offset); page_length = remain; if ((page_offset + remain) > PAGE_SIZE) page_length = PAGE_SIZE - page_offset; @@ -652,7 +651,6 @@ i915_gem_gtt_pwrite_fast(struct drm_device *dev, */ if (fast_user_write(dev_priv->mm.gtt_mapping, page_base, page_offset, user_data, page_length)) - return -EFAULT; remain -= page_length; @@ -732,9 +730,9 @@ i915_gem_gtt_pwrite_slow(struct drm_device *dev, * page_length = bytes to copy for this page */ gtt_page_base = offset & PAGE_MASK; - gtt_page_offset = offset & ~PAGE_MASK; + gtt_page_offset = offset_in_page(offset); data_page_index = data_ptr / PAGE_SIZE - first_data_page; - data_page_offset = data_ptr & ~PAGE_MASK; + data_page_offset = offset_in_page(data_ptr); page_length = remain; if ((gtt_page_offset + page_length) > PAGE_SIZE) @@ -793,13 +791,12 @@ i915_gem_shmem_pwrite_fast(struct drm_device *dev, * page_offset = offset within page * page_length = bytes to copy for this page */ - page_offset = offset & (PAGE_SIZE-1); + page_offset = offset_in_page(offset); page_length = remain; if ((page_offset + remain) > PAGE_SIZE) page_length = PAGE_SIZE - page_offset; - page = read_cache_page_gfp(mapping, offset >> PAGE_SHIFT, - GFP_HIGHUSER | __GFP_RECLAIMABLE); + page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT); if (IS_ERR(page)) return PTR_ERR(page); @@ -898,9 +895,9 @@ i915_gem_shmem_pwrite_slow(struct drm_device *dev, * data_page_offset = offset with data_page_index page. * page_length = bytes to copy for this page */ - shmem_page_offset = offset & ~PAGE_MASK; + shmem_page_offset = offset_in_page(offset); data_page_index = data_ptr / PAGE_SIZE - first_data_page; - data_page_offset = data_ptr & ~PAGE_MASK; + data_page_offset = offset_in_page(data_ptr); page_length = remain; if ((shmem_page_offset + page_length) > PAGE_SIZE) @@ -908,8 +905,7 @@ i915_gem_shmem_pwrite_slow(struct drm_device *dev, if ((data_page_offset + page_length) > PAGE_SIZE) page_length = PAGE_SIZE - data_page_offset; - page = read_cache_page_gfp(mapping, offset >> PAGE_SHIFT, - GFP_HIGHUSER | __GFP_RECLAIMABLE); + page = shmem_read_mapping_page(mapping, offset >> PAGE_SHIFT); if (IS_ERR(page)) { ret = PTR_ERR(page); goto out; @@ -1220,11 +1216,11 @@ int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf) ret = i915_gem_object_bind_to_gtt(obj, 0, true); if (ret) goto unlock; - } - ret = i915_gem_object_set_to_gtt_domain(obj, write); - if (ret) - goto unlock; + ret = i915_gem_object_set_to_gtt_domain(obj, write); + if (ret) + goto unlock; + } if (obj->tiling_mode == I915_TILING_NONE) ret = i915_gem_object_put_fence(obj); @@ -1378,25 +1374,24 @@ i915_gem_free_mmap_offset(struct drm_i915_gem_object *obj) } static uint32_t -i915_gem_get_gtt_size(struct drm_i915_gem_object *obj) +i915_gem_get_gtt_size(struct drm_device *dev, uint32_t size, int tiling_mode) { - struct drm_device *dev = obj->base.dev; - uint32_t size; + uint32_t gtt_size; if (INTEL_INFO(dev)->gen >= 4 || - obj->tiling_mode == I915_TILING_NONE) - return obj->base.size; + tiling_mode == I915_TILING_NONE) + return size; /* Previous chips need a power-of-two fence region when tiling */ if (INTEL_INFO(dev)->gen == 3) - size = 1024*1024; + gtt_size = 1024*1024; else - size = 512*1024; + gtt_size = 512*1024; - while (size < obj->base.size) - size <<= 1; + while (gtt_size < size) + gtt_size <<= 1; - return size; + return gtt_size; } /** @@ -1407,58 +1402,52 @@ i915_gem_get_gtt_size(struct drm_i915_gem_object *obj) * potential fence register mapping. */ static uint32_t -i915_gem_get_gtt_alignment(struct drm_i915_gem_object *obj) +i915_gem_get_gtt_alignment(struct drm_device *dev, + uint32_t size, + int tiling_mode) { - struct drm_device *dev = obj->base.dev; - /* * Minimum alignment is 4k (GTT page size), but might be greater * if a fence register is needed for the object. */ if (INTEL_INFO(dev)->gen >= 4 || - obj->tiling_mode == I915_TILING_NONE) + tiling_mode == I915_TILING_NONE) return 4096; /* * Previous chips need to be aligned to the size of the smallest * fence register that can contain the object. */ - return i915_gem_get_gtt_size(obj); + return i915_gem_get_gtt_size(dev, size, tiling_mode); } /** * i915_gem_get_unfenced_gtt_alignment - return required GTT alignment for an * unfenced object - * @obj: object to check + * @dev: the device + * @size: size of the object + * @tiling_mode: tiling mode of the object * * Return the required GTT alignment for an object, only taking into account * unfenced tiled surface requirements. */ uint32_t -i915_gem_get_unfenced_gtt_alignment(struct drm_i915_gem_object *obj) +i915_gem_get_unfenced_gtt_alignment(struct drm_device *dev, + uint32_t size, + int tiling_mode) { - struct drm_device *dev = obj->base.dev; - int tile_height; - /* * Minimum alignment is 4k (GTT page size) for sane hw. */ if (INTEL_INFO(dev)->gen >= 4 || IS_G33(dev) || - obj->tiling_mode == I915_TILING_NONE) + tiling_mode == I915_TILING_NONE) return 4096; - /* - * Older chips need unfenced tiled buffers to be aligned to the left - * edge of an even tile row (where tile rows are counted as if the bo is - * placed in a fenced gtt region). + /* Previous hardware however needs to be aligned to a power-of-two + * tile height. The simplest method for determining this is to reuse + * the power-of-tile object size. */ - if (IS_GEN2(dev) || - (obj->tiling_mode == I915_TILING_Y && HAS_128_BYTE_Y_TILING(dev))) - tile_height = 32; - else - tile_height = 8; - - return tile_height * obj->stride * 2; + return i915_gem_get_gtt_size(dev, size, tiling_mode); } int @@ -1558,12 +1547,10 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj, inode = obj->base.filp->f_path.dentry->d_inode; mapping = inode->i_mapping; + gfpmask |= mapping_gfp_mask(mapping); + for (i = 0; i < page_count; i++) { - page = read_cache_page_gfp(mapping, i, - GFP_HIGHUSER | - __GFP_COLD | - __GFP_RECLAIMABLE | - gfpmask); + page = shmem_read_mapping_page_gfp(mapping, i, gfpmask); if (IS_ERR(page)) goto err_pages; @@ -1701,13 +1688,10 @@ i915_gem_object_truncate(struct drm_i915_gem_object *obj) /* Our goal here is to return as much of the memory as * is possible back to the system as we are called from OOM. * To do this we must instruct the shmfs to drop all of its - * backing pages, *now*. Here we mirror the actions taken - * when by shmem_delete_inode() to release the backing store. + * backing pages, *now*. */ inode = obj->base.filp->f_path.dentry->d_inode; - truncate_inode_pages(inode->i_mapping, 0); - if (inode->i_op->truncate_range) - inode->i_op->truncate_range(inode, 0, (loff_t)-1); + shmem_truncate_range(inode, 0, (loff_t)-1); obj->madv = __I915_MADV_PURGED; } @@ -1779,8 +1763,11 @@ i915_add_request(struct intel_ring_buffer *ring, ring->outstanding_lazy_request = false; if (!dev_priv->mm.suspended) { - mod_timer(&dev_priv->hangcheck_timer, - jiffies + msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)); + if (i915_enable_hangcheck) { + mod_timer(&dev_priv->hangcheck_timer, + jiffies + + msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)); + } if (was_empty) queue_delayed_work(dev_priv->wq, &dev_priv->mm.retire_work, HZ); @@ -2080,8 +2067,8 @@ i915_wait_request(struct intel_ring_buffer *ring, if (!ier) { DRM_ERROR("something (likely vbetool) disabled " "interrupts, re-enabling\n"); - i915_driver_irq_preinstall(ring->dev); - i915_driver_irq_postinstall(ring->dev); + ring->dev->driver->irq_preinstall(ring->dev); + ring->dev->driver->irq_postinstall(ring->dev); } trace_i915_gem_request_wait_begin(ring, seqno); @@ -2151,6 +2138,30 @@ i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj) return 0; } +static void i915_gem_object_finish_gtt(struct drm_i915_gem_object *obj) +{ + u32 old_write_domain, old_read_domains; + + /* Act a barrier for all accesses through the GTT */ + mb(); + + /* Force a pagefault for domain tracking on next user access */ + i915_gem_release_mmap(obj); + + if ((obj->base.read_domains & I915_GEM_DOMAIN_GTT) == 0) + return; + + old_read_domains = obj->base.read_domains; + old_write_domain = obj->base.write_domain; + + obj->base.read_domains &= ~I915_GEM_DOMAIN_GTT; + obj->base.write_domain &= ~I915_GEM_DOMAIN_GTT; + + trace_i915_gem_object_change_domain(obj, + old_read_domains, + old_write_domain); +} + /** * Unbinds an object from the GTT aperture. */ @@ -2167,23 +2178,28 @@ i915_gem_object_unbind(struct drm_i915_gem_object *obj) return -EINVAL; } - /* blow away mappings if mapped through GTT */ - i915_gem_release_mmap(obj); - - /* Move the object to the CPU domain to ensure that - * any possible CPU writes while it's not in the GTT - * are flushed when we go to remap it. This will - * also ensure that all pending GPU writes are finished - * before we unbind. - */ - ret = i915_gem_object_set_to_cpu_domain(obj, 1); + ret = i915_gem_object_finish_gpu(obj); if (ret == -ERESTARTSYS) return ret; /* Continue on if we fail due to EIO, the GPU is hung so we * should be safe and we need to cleanup or else we might * cause memory corruption through use-after-free. */ + + i915_gem_object_finish_gtt(obj); + + /* Move the object to the CPU domain to ensure that + * any possible CPU writes while it's not in the GTT + * are flushed when we go to remap it. + */ + if (ret == 0) + ret = i915_gem_object_set_to_cpu_domain(obj, 1); + if (ret == -ERESTARTSYS) + return ret; if (ret) { + /* In the event of a disaster, abandon all caches and + * hope for the best. + */ i915_gem_clflush_object(obj); obj->base.read_domains = obj->base.write_domain = I915_GEM_DOMAIN_CPU; } @@ -2673,6 +2689,7 @@ i915_gem_object_get_fence(struct drm_i915_gem_object *obj, update: obj->tiling_changed = false; switch (INTEL_INFO(dev)->gen) { + case 7: case 6: ret = sandybridge_write_fence_reg(obj, pipelined); break; @@ -2706,6 +2723,7 @@ i915_gem_clear_fence_reg(struct drm_device *dev, uint32_t fence_reg = reg - dev_priv->fence_regs; switch (INTEL_INFO(dev)->gen) { + case 7: case 6: I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 + fence_reg*8, 0); break; @@ -2750,9 +2768,16 @@ i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj, return -EINVAL; } - fence_size = i915_gem_get_gtt_size(obj); - fence_alignment = i915_gem_get_gtt_alignment(obj); - unfenced_alignment = i915_gem_get_unfenced_gtt_alignment(obj); + fence_size = i915_gem_get_gtt_size(dev, + obj->base.size, + obj->tiling_mode); + fence_alignment = i915_gem_get_gtt_alignment(dev, + obj->base.size, + obj->tiling_mode); + unfenced_alignment = + i915_gem_get_unfenced_gtt_alignment(dev, + obj->base.size, + obj->tiling_mode); if (alignment == 0) alignment = map_and_fenceable ? fence_alignment : @@ -2878,6 +2903,17 @@ i915_gem_clflush_object(struct drm_i915_gem_object *obj) if (obj->pages == NULL) return; + /* If the GPU is snooping the contents of the CPU cache, + * we do not need to manually clear the CPU cache lines. However, + * the caches are only snooped when the render cache is + * flushed/invalidated. As we always have to emit invalidations + * and flushes when moving into and out of the RENDER domain, correct + * snooping behaviour occurs naturally as the result of our domain + * tracking. + */ + if (obj->cache_level != I915_CACHE_NONE) + return; + trace_i915_gem_object_clflush(obj); drm_clflush_pages(obj->pages, obj->base.size / PAGE_SIZE); @@ -2913,8 +2949,6 @@ i915_gem_object_flush_gtt_write_domain(struct drm_i915_gem_object *obj) */ wmb(); - i915_gem_release_mmap(obj); - old_write_domain = obj->base.write_domain; obj->base.write_domain = 0; @@ -2994,51 +3028,139 @@ i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj, bool write) return 0; } +int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj, + enum i915_cache_level cache_level) +{ + int ret; + + if (obj->cache_level == cache_level) + return 0; + + if (obj->pin_count) { + DRM_DEBUG("can not change the cache level of pinned objects\n"); + return -EBUSY; + } + + if (obj->gtt_space) { + ret = i915_gem_object_finish_gpu(obj); + if (ret) + return ret; + + i915_gem_object_finish_gtt(obj); + + /* Before SandyBridge, you could not use tiling or fence + * registers with snooped memory, so relinquish any fences + * currently pointing to our region in the aperture. + */ + if (INTEL_INFO(obj->base.dev)->gen < 6) { + ret = i915_gem_object_put_fence(obj); + if (ret) + return ret; + } + + i915_gem_gtt_rebind_object(obj, cache_level); + } + + if (cache_level == I915_CACHE_NONE) { + u32 old_read_domains, old_write_domain; + + /* If we're coming from LLC cached, then we haven't + * actually been tracking whether the data is in the + * CPU cache or not, since we only allow one bit set + * in obj->write_domain and have been skipping the clflushes. + * Just set it to the CPU cache for now. + */ + WARN_ON(obj->base.write_domain & ~I915_GEM_DOMAIN_CPU); + WARN_ON(obj->base.read_domains & ~I915_GEM_DOMAIN_CPU); + + old_read_domains = obj->base.read_domains; + old_write_domain = obj->base.write_domain; + + obj->base.read_domains = I915_GEM_DOMAIN_CPU; + obj->base.write_domain = I915_GEM_DOMAIN_CPU; + + trace_i915_gem_object_change_domain(obj, + old_read_domains, + old_write_domain); + } + + obj->cache_level = cache_level; + return 0; +} + /* - * Prepare buffer for display plane. Use uninterruptible for possible flush - * wait, as in modesetting process we're not supposed to be interrupted. + * Prepare buffer for display plane (scanout, cursors, etc). + * Can be called from an uninterruptible phase (modesetting) and allows + * any flushes to be pipelined (for pageflips). + * + * For the display plane, we want to be in the GTT but out of any write + * domains. So in many ways this looks like set_to_gtt_domain() apart from the + * ability to pipeline the waits, pinning and any additional subtleties + * that may differentiate the display plane from ordinary buffers. */ int -i915_gem_object_set_to_display_plane(struct drm_i915_gem_object *obj, +i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj, + u32 alignment, struct intel_ring_buffer *pipelined) { - uint32_t old_read_domains; + u32 old_read_domains, old_write_domain; int ret; - /* Not valid to be called on unbound objects. */ - if (obj->gtt_space == NULL) - return -EINVAL; - ret = i915_gem_object_flush_gpu_write_domain(obj); if (ret) return ret; - - /* Currently, we are always called from an non-interruptible context. */ if (pipelined != obj->ring) { ret = i915_gem_object_wait_rendering(obj); - if (ret) + if (ret == -ERESTARTSYS) return ret; } + /* The display engine is not coherent with the LLC cache on gen6. As + * a result, we make sure that the pinning that is about to occur is + * done with uncached PTEs. This is lowest common denominator for all + * chipsets. + * + * However for gen6+, we could do better by using the GFDT bit instead + * of uncaching, which would allow us to flush all the LLC-cached data + * with that bit in the PTE to main memory with just one PIPE_CONTROL. + */ + ret = i915_gem_object_set_cache_level(obj, I915_CACHE_NONE); + if (ret) + return ret; + + /* As the user may map the buffer once pinned in the display plane + * (e.g. libkms for the bootup splash), we have to ensure that we + * always use map_and_fenceable for all scanout buffers. + */ + ret = i915_gem_object_pin(obj, alignment, true); + if (ret) + return ret; + i915_gem_object_flush_cpu_write_domain(obj); + old_write_domain = obj->base.write_domain; old_read_domains = obj->base.read_domains; + + /* It should now be out of any other write domains, and we can update + * the domain values for our changes. + */ + BUG_ON((obj->base.write_domain & ~I915_GEM_DOMAIN_GTT) != 0); obj->base.read_domains |= I915_GEM_DOMAIN_GTT; trace_i915_gem_object_change_domain(obj, old_read_domains, - obj->base.write_domain); + old_write_domain); return 0; } int -i915_gem_object_flush_gpu(struct drm_i915_gem_object *obj) +i915_gem_object_finish_gpu(struct drm_i915_gem_object *obj) { int ret; - if (!obj->active) + if ((obj->base.read_domains & I915_GEM_GPU_DOMAINS) == 0) return 0; if (obj->base.write_domain & I915_GEM_GPU_DOMAINS) { @@ -3047,6 +3169,9 @@ i915_gem_object_flush_gpu(struct drm_i915_gem_object *obj) return ret; } + /* Ensure that we invalidate the GPU's caches and TLBs. */ + obj->base.read_domains &= ~I915_GEM_GPU_DOMAINS; + return i915_gem_object_wait_rendering(obj); } @@ -3554,6 +3679,7 @@ struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev, { struct drm_i915_private *dev_priv = dev->dev_private; struct drm_i915_gem_object *obj; + struct address_space *mapping; obj = kzalloc(sizeof(*obj), GFP_KERNEL); if (obj == NULL) @@ -3564,12 +3690,31 @@ struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev, return NULL; } + mapping = obj->base.filp->f_path.dentry->d_inode->i_mapping; + mapping_set_gfp_mask(mapping, GFP_HIGHUSER | __GFP_RECLAIMABLE); + i915_gem_info_add_obj(dev_priv, size); obj->base.write_domain = I915_GEM_DOMAIN_CPU; obj->base.read_domains = I915_GEM_DOMAIN_CPU; - obj->agp_type = AGP_USER_MEMORY; + if (IS_GEN6(dev)) { + /* On Gen6, we can have the GPU use the LLC (the CPU + * cache) for about a 10% performance improvement + * compared to uncached. Graphics requests other than + * display scanout are coherent with the CPU in + * accessing this cache. This means in this mode we + * don't need to clflush on the CPU side, and on the + * GPU side we only need to flush internal caches to + * get data visible to the CPU. + * + * However, we maintain the display planes as UC, and so + * need to rebind when first used as such. + */ + obj->cache_level = I915_CACHE_LLC; + } else + obj->cache_level = I915_CACHE_NONE; + obj->base.driver_private = NULL; obj->fence_reg = I915_FENCE_REG_NONE; INIT_LIST_HEAD(&obj->mm_list); @@ -3845,25 +3990,10 @@ i915_gem_load(struct drm_device *dev) dev_priv->num_fence_regs = 8; /* Initialize fence registers to zero */ - switch (INTEL_INFO(dev)->gen) { - case 6: - for (i = 0; i < 16; i++) - I915_WRITE64(FENCE_REG_SANDYBRIDGE_0 + (i * 8), 0); - break; - case 5: - case 4: - for (i = 0; i < 16; i++) - I915_WRITE64(FENCE_REG_965_0 + (i * 8), 0); - break; - case 3: - if (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) - for (i = 0; i < 8; i++) - I915_WRITE(FENCE_REG_945_8 + (i * 4), 0); - case 2: - for (i = 0; i < 8; i++) - I915_WRITE(FENCE_REG_830_0 + (i * 4), 0); - break; + for (i = 0; i < dev_priv->num_fence_regs; i++) { + i915_gem_clear_fence_reg(dev, &dev_priv->fence_regs[i]); } + i915_gem_detect_bit_6_swizzle(dev); init_waitqueue_head(&dev_priv->pending_flip_queue); @@ -3954,8 +4084,7 @@ void i915_gem_detach_phys_object(struct drm_device *dev, page_count = obj->base.size / PAGE_SIZE; for (i = 0; i < page_count; i++) { - struct page *page = read_cache_page_gfp(mapping, i, - GFP_HIGHUSER | __GFP_RECLAIMABLE); + struct page *page = shmem_read_mapping_page(mapping, i); if (!IS_ERR(page)) { char *dst = kmap_atomic(page); memcpy(dst, vaddr + i*PAGE_SIZE, PAGE_SIZE); @@ -4016,8 +4145,7 @@ i915_gem_attach_phys_object(struct drm_device *dev, struct page *page; char *dst, *src; - page = read_cache_page_gfp(mapping, i, - GFP_HIGHUSER | __GFP_RECLAIMABLE); + page = shmem_read_mapping_page(mapping, i); if (IS_ERR(page)) return PTR_ERR(page); @@ -4094,9 +4222,7 @@ i915_gpu_is_active(struct drm_device *dev) } static int -i915_gem_inactive_shrink(struct shrinker *shrinker, - int nr_to_scan, - gfp_t gfp_mask) +i915_gem_inactive_shrink(struct shrinker *shrinker, struct shrink_control *sc) { struct drm_i915_private *dev_priv = container_of(shrinker, @@ -4104,6 +4230,7 @@ i915_gem_inactive_shrink(struct shrinker *shrinker, mm.inactive_shrinker); struct drm_device *dev = dev_priv->dev; struct drm_i915_gem_object *obj, *next; + int nr_to_scan = sc->nr_to_scan; int cnt; if (!mutex_trylock(&dev->struct_mutex)) diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c index 20a4cc5b818..4934cf84c32 100644 --- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c @@ -187,10 +187,6 @@ i915_gem_object_set_to_gpu_domain(struct drm_i915_gem_object *obj, if ((flush_domains | invalidate_domains) & I915_GEM_DOMAIN_CPU) i915_gem_clflush_object(obj); - /* blow away mappings if mapped through GTT */ - if ((flush_domains | invalidate_domains) & I915_GEM_DOMAIN_GTT) - i915_gem_release_mmap(obj); - if (obj->base.pending_write_domain) cd->flips |= atomic_read(&obj->pending_flip); diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c index b0abdc64aa9..7a709cd8d54 100644 --- a/drivers/gpu/drm/i915/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/i915_gem_gtt.c @@ -29,6 +29,26 @@ #include "i915_trace.h" #include "intel_drv.h" +/* XXX kill agp_type! */ +static unsigned int cache_level_to_agp_type(struct drm_device *dev, + enum i915_cache_level cache_level) +{ + switch (cache_level) { + case I915_CACHE_LLC_MLC: + if (INTEL_INFO(dev)->gen >= 6) + return AGP_USER_CACHED_MEMORY_LLC_MLC; + /* Older chipsets do not have this extra level of CPU + * cacheing, so fallthrough and request the PTE simply + * as cached. + */ + case I915_CACHE_LLC: + return AGP_USER_CACHED_MEMORY; + default: + case I915_CACHE_NONE: + return AGP_USER_MEMORY; + } +} + void i915_gem_restore_gtt_mappings(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -40,21 +60,7 @@ void i915_gem_restore_gtt_mappings(struct drm_device *dev) list_for_each_entry(obj, &dev_priv->mm.gtt_list, gtt_list) { i915_gem_clflush_object(obj); - - if (dev_priv->mm.gtt->needs_dmar) { - BUG_ON(!obj->sg_list); - - intel_gtt_insert_sg_entries(obj->sg_list, - obj->num_sg, - obj->gtt_space->start - >> PAGE_SHIFT, - obj->agp_type); - } else - intel_gtt_insert_pages(obj->gtt_space->start - >> PAGE_SHIFT, - obj->base.size >> PAGE_SHIFT, - obj->pages, - obj->agp_type); + i915_gem_gtt_rebind_object(obj, obj->cache_level); } intel_gtt_chipset_flush(); @@ -64,6 +70,7 @@ int i915_gem_gtt_bind_object(struct drm_i915_gem_object *obj) { struct drm_device *dev = obj->base.dev; struct drm_i915_private *dev_priv = dev->dev_private; + unsigned int agp_type = cache_level_to_agp_type(dev, obj->cache_level); int ret; if (dev_priv->mm.gtt->needs_dmar) { @@ -77,16 +84,37 @@ int i915_gem_gtt_bind_object(struct drm_i915_gem_object *obj) intel_gtt_insert_sg_entries(obj->sg_list, obj->num_sg, obj->gtt_space->start >> PAGE_SHIFT, - obj->agp_type); + agp_type); } else intel_gtt_insert_pages(obj->gtt_space->start >> PAGE_SHIFT, obj->base.size >> PAGE_SHIFT, obj->pages, - obj->agp_type); + agp_type); return 0; } +void i915_gem_gtt_rebind_object(struct drm_i915_gem_object *obj, + enum i915_cache_level cache_level) +{ + struct drm_device *dev = obj->base.dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned int agp_type = cache_level_to_agp_type(dev, cache_level); + + if (dev_priv->mm.gtt->needs_dmar) { + BUG_ON(!obj->sg_list); + + intel_gtt_insert_sg_entries(obj->sg_list, + obj->num_sg, + obj->gtt_space->start >> PAGE_SHIFT, + agp_type); + } else + intel_gtt_insert_pages(obj->gtt_space->start >> PAGE_SHIFT, + obj->base.size >> PAGE_SHIFT, + obj->pages, + agp_type); +} + void i915_gem_gtt_unbind_object(struct drm_i915_gem_object *obj) { intel_gtt_clear_range(obj->gtt_space->start >> PAGE_SHIFT, diff --git a/drivers/gpu/drm/i915/i915_gem_tiling.c b/drivers/gpu/drm/i915/i915_gem_tiling.c index 281ad3d6115..99c4faa59d8 100644 --- a/drivers/gpu/drm/i915/i915_gem_tiling.c +++ b/drivers/gpu/drm/i915/i915_gem_tiling.c @@ -92,7 +92,7 @@ i915_gem_detect_bit_6_swizzle(struct drm_device *dev) uint32_t swizzle_x = I915_BIT_6_SWIZZLE_UNKNOWN; uint32_t swizzle_y = I915_BIT_6_SWIZZLE_UNKNOWN; - if (IS_GEN5(dev) || IS_GEN6(dev)) { + if (INTEL_INFO(dev)->gen >= 5) { /* On Ironlake whatever DRAM config, GPU always do * same swizzling setup. */ @@ -348,7 +348,9 @@ i915_gem_set_tiling(struct drm_device *dev, void *data, /* Rebind if we need a change of alignment */ if (!obj->map_and_fenceable) { u32 unfenced_alignment = - i915_gem_get_unfenced_gtt_alignment(obj); + i915_gem_get_unfenced_gtt_alignment(dev, + obj->base.size, + args->tiling_mode); if (obj->gtt_offset & (unfenced_alignment - 1)) ret = i915_gem_object_unbind(obj); } diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c index 188b497e507..9cbb0cd8f46 100644 --- a/drivers/gpu/drm/i915/i915_irq.c +++ b/drivers/gpu/drm/i915/i915_irq.c @@ -152,7 +152,7 @@ i915_pipe_enabled(struct drm_device *dev, int pipe) /* Called from drm generic code, passed a 'crtc', which * we use as a pipe index */ -u32 i915_get_vblank_counter(struct drm_device *dev, int pipe) +static u32 i915_get_vblank_counter(struct drm_device *dev, int pipe) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; unsigned long high_frame; @@ -184,7 +184,7 @@ u32 i915_get_vblank_counter(struct drm_device *dev, int pipe) return (high1 << 8) | low; } -u32 gm45_get_vblank_counter(struct drm_device *dev, int pipe) +static u32 gm45_get_vblank_counter(struct drm_device *dev, int pipe) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; int reg = PIPE_FRMCOUNT_GM45(pipe); @@ -198,7 +198,7 @@ u32 gm45_get_vblank_counter(struct drm_device *dev, int pipe) return I915_READ(reg); } -int i915_get_crtc_scanoutpos(struct drm_device *dev, int pipe, +static int i915_get_crtc_scanoutpos(struct drm_device *dev, int pipe, int *vpos, int *hpos) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; @@ -264,7 +264,7 @@ int i915_get_crtc_scanoutpos(struct drm_device *dev, int pipe, return ret; } -int i915_get_vblank_timestamp(struct drm_device *dev, int pipe, +static int i915_get_vblank_timestamp(struct drm_device *dev, int pipe, int *max_error, struct timeval *vblank_time, unsigned flags) @@ -306,12 +306,15 @@ static void i915_hotplug_work_func(struct work_struct *work) struct drm_mode_config *mode_config = &dev->mode_config; struct intel_encoder *encoder; + mutex_lock(&mode_config->mutex); DRM_DEBUG_KMS("running encoder hotplug functions\n"); list_for_each_entry(encoder, &mode_config->encoder_list, base.head) if (encoder->hot_plug) encoder->hot_plug(encoder); + mutex_unlock(&mode_config->mutex); + /* Just fire off a uevent and let userspace tell us what to do */ drm_helper_hpd_irq_event(dev); } @@ -361,28 +364,38 @@ static void notify_ring(struct drm_device *dev, ring->irq_seqno = seqno; wake_up_all(&ring->irq_queue); - - dev_priv->hangcheck_count = 0; - mod_timer(&dev_priv->hangcheck_timer, - jiffies + msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)); + if (i915_enable_hangcheck) { + dev_priv->hangcheck_count = 0; + mod_timer(&dev_priv->hangcheck_timer, + jiffies + + msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)); + } } -static void gen6_pm_irq_handler(struct drm_device *dev) +static void gen6_pm_rps_work(struct work_struct *work) { - drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + drm_i915_private_t *dev_priv = container_of(work, drm_i915_private_t, + rps_work); u8 new_delay = dev_priv->cur_delay; - u32 pm_iir; + u32 pm_iir, pm_imr; + + spin_lock_irq(&dev_priv->rps_lock); + pm_iir = dev_priv->pm_iir; + dev_priv->pm_iir = 0; + pm_imr = I915_READ(GEN6_PMIMR); + spin_unlock_irq(&dev_priv->rps_lock); - pm_iir = I915_READ(GEN6_PMIIR); if (!pm_iir) return; + mutex_lock(&dev_priv->dev->struct_mutex); if (pm_iir & GEN6_PM_RP_UP_THRESHOLD) { if (dev_priv->cur_delay != dev_priv->max_delay) new_delay = dev_priv->cur_delay + 1; if (new_delay > dev_priv->max_delay) new_delay = dev_priv->max_delay; } else if (pm_iir & (GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_DOWN_TIMEOUT)) { + gen6_gt_force_wake_get(dev_priv); if (dev_priv->cur_delay != dev_priv->min_delay) new_delay = dev_priv->cur_delay - 1; if (new_delay < dev_priv->min_delay) { @@ -396,13 +409,19 @@ static void gen6_pm_irq_handler(struct drm_device *dev) I915_WRITE(GEN6_RP_INTERRUPT_LIMITS, I915_READ(GEN6_RP_INTERRUPT_LIMITS) & ~0x3f0000); } - + gen6_gt_force_wake_put(dev_priv); } - gen6_set_rps(dev, new_delay); + gen6_set_rps(dev_priv->dev, new_delay); dev_priv->cur_delay = new_delay; - I915_WRITE(GEN6_PMIIR, pm_iir); + /* + * rps_lock not held here because clearing is non-destructive. There is + * an *extremely* unlikely race with gen6_rps_enable() that is prevented + * by holding struct_mutex for the duration of the write. + */ + I915_WRITE(GEN6_PMIMR, pm_imr & ~pm_iir); + mutex_unlock(&dev_priv->dev->struct_mutex); } static void pch_irq_handler(struct drm_device *dev) @@ -448,8 +467,97 @@ static void pch_irq_handler(struct drm_device *dev) DRM_DEBUG_DRIVER("PCH transcoder A underrun interrupt\n"); } -static irqreturn_t ironlake_irq_handler(struct drm_device *dev) +static irqreturn_t ivybridge_irq_handler(DRM_IRQ_ARGS) { + struct drm_device *dev = (struct drm_device *) arg; + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + int ret = IRQ_NONE; + u32 de_iir, gt_iir, de_ier, pch_iir, pm_iir; + struct drm_i915_master_private *master_priv; + + atomic_inc(&dev_priv->irq_received); + + /* disable master interrupt before clearing iir */ + de_ier = I915_READ(DEIER); + I915_WRITE(DEIER, de_ier & ~DE_MASTER_IRQ_CONTROL); + POSTING_READ(DEIER); + + de_iir = I915_READ(DEIIR); + gt_iir = I915_READ(GTIIR); + pch_iir = I915_READ(SDEIIR); + pm_iir = I915_READ(GEN6_PMIIR); + + if (de_iir == 0 && gt_iir == 0 && pch_iir == 0 && pm_iir == 0) + goto done; + + ret = IRQ_HANDLED; + + if (dev->primary->master) { + master_priv = dev->primary->master->driver_priv; + if (master_priv->sarea_priv) + master_priv->sarea_priv->last_dispatch = + READ_BREADCRUMB(dev_priv); + } + + if (gt_iir & (GT_USER_INTERRUPT | GT_PIPE_NOTIFY)) + notify_ring(dev, &dev_priv->ring[RCS]); + if (gt_iir & GT_GEN6_BSD_USER_INTERRUPT) + notify_ring(dev, &dev_priv->ring[VCS]); + if (gt_iir & GT_BLT_USER_INTERRUPT) + notify_ring(dev, &dev_priv->ring[BCS]); + + if (de_iir & DE_GSE_IVB) + intel_opregion_gse_intr(dev); + + if (de_iir & DE_PLANEA_FLIP_DONE_IVB) { + intel_prepare_page_flip(dev, 0); + intel_finish_page_flip_plane(dev, 0); + } + + if (de_iir & DE_PLANEB_FLIP_DONE_IVB) { + intel_prepare_page_flip(dev, 1); + intel_finish_page_flip_plane(dev, 1); + } + + if (de_iir & DE_PIPEA_VBLANK_IVB) + drm_handle_vblank(dev, 0); + + if (de_iir & DE_PIPEB_VBLANK_IVB) + drm_handle_vblank(dev, 1); + + /* check event from PCH */ + if (de_iir & DE_PCH_EVENT_IVB) { + if (pch_iir & SDE_HOTPLUG_MASK_CPT) + queue_work(dev_priv->wq, &dev_priv->hotplug_work); + pch_irq_handler(dev); + } + + if (pm_iir & GEN6_PM_DEFERRED_EVENTS) { + unsigned long flags; + spin_lock_irqsave(&dev_priv->rps_lock, flags); + WARN(dev_priv->pm_iir & pm_iir, "Missed a PM interrupt\n"); + I915_WRITE(GEN6_PMIMR, pm_iir); + dev_priv->pm_iir |= pm_iir; + spin_unlock_irqrestore(&dev_priv->rps_lock, flags); + queue_work(dev_priv->wq, &dev_priv->rps_work); + } + + /* should clear PCH hotplug event before clear CPU irq */ + I915_WRITE(SDEIIR, pch_iir); + I915_WRITE(GTIIR, gt_iir); + I915_WRITE(DEIIR, de_iir); + I915_WRITE(GEN6_PMIIR, pm_iir); + +done: + I915_WRITE(DEIER, de_ier); + POSTING_READ(DEIER); + + return ret; +} + +static irqreturn_t ironlake_irq_handler(DRM_IRQ_ARGS) +{ + struct drm_device *dev = (struct drm_device *) arg; drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; int ret = IRQ_NONE; u32 de_iir, gt_iir, de_ier, pch_iir, pm_iir; @@ -457,6 +565,8 @@ static irqreturn_t ironlake_irq_handler(struct drm_device *dev) struct drm_i915_master_private *master_priv; u32 bsd_usr_interrupt = GT_BSD_USER_INTERRUPT; + atomic_inc(&dev_priv->irq_received); + if (IS_GEN6(dev)) bsd_usr_interrupt = GT_GEN6_BSD_USER_INTERRUPT; @@ -526,13 +636,30 @@ static irqreturn_t ironlake_irq_handler(struct drm_device *dev) i915_handle_rps_change(dev); } - if (IS_GEN6(dev)) - gen6_pm_irq_handler(dev); + if (IS_GEN6(dev) && pm_iir & GEN6_PM_DEFERRED_EVENTS) { + /* + * IIR bits should never already be set because IMR should + * prevent an interrupt from being shown in IIR. The warning + * displays a case where we've unsafely cleared + * dev_priv->pm_iir. Although missing an interrupt of the same + * type is not a problem, it displays a problem in the logic. + * + * The mask bit in IMR is cleared by rps_work. + */ + unsigned long flags; + spin_lock_irqsave(&dev_priv->rps_lock, flags); + WARN(dev_priv->pm_iir & pm_iir, "Missed a PM interrupt\n"); + I915_WRITE(GEN6_PMIMR, pm_iir); + dev_priv->pm_iir |= pm_iir; + spin_unlock_irqrestore(&dev_priv->rps_lock, flags); + queue_work(dev_priv->wq, &dev_priv->rps_work); + } /* should clear PCH hotplug event before clear CPU irq */ I915_WRITE(SDEIIR, pch_iir); I915_WRITE(GTIIR, gt_iir); I915_WRITE(DEIIR, de_iir); + I915_WRITE(GEN6_PMIIR, pm_iir); done: I915_WRITE(DEIER, de_ier); @@ -676,7 +803,7 @@ static u32 capture_bo_list(struct drm_i915_error_buffer *err, err->dirty = obj->dirty; err->purgeable = obj->madv != I915_MADV_WILLNEED; err->ring = obj->ring ? obj->ring->id : 0; - err->agp_type = obj->agp_type == AGP_USER_CACHED_MEMORY; + err->cache_level = obj->cache_level; if (++i == count) break; @@ -1087,7 +1214,7 @@ static void i915_pageflip_stall_check(struct drm_device *dev, int pipe) } } -irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS) +static irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS) { struct drm_device *dev = (struct drm_device *) arg; drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; @@ -1103,9 +1230,6 @@ irqreturn_t i915_driver_irq_handler(DRM_IRQ_ARGS) atomic_inc(&dev_priv->irq_received); - if (HAS_PCH_SPLIT(dev)) - return ironlake_irq_handler(dev); - iir = I915_READ(IIR); if (INTEL_INFO(dev)->gen >= 4) @@ -1335,7 +1459,7 @@ int i915_irq_wait(struct drm_device *dev, void *data, /* Called from drm generic code, passed 'crtc' which * we use as a pipe index */ -int i915_enable_vblank(struct drm_device *dev, int pipe) +static int i915_enable_vblank(struct drm_device *dev, int pipe) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; unsigned long irqflags; @@ -1344,10 +1468,7 @@ int i915_enable_vblank(struct drm_device *dev, int pipe) return -EINVAL; spin_lock_irqsave(&dev_priv->irq_lock, irqflags); - if (HAS_PCH_SPLIT(dev)) - ironlake_enable_display_irq(dev_priv, (pipe == 0) ? - DE_PIPEA_VBLANK: DE_PIPEB_VBLANK); - else if (INTEL_INFO(dev)->gen >= 4) + if (INTEL_INFO(dev)->gen >= 4) i915_enable_pipestat(dev_priv, pipe, PIPE_START_VBLANK_INTERRUPT_ENABLE); else @@ -1362,10 +1483,42 @@ int i915_enable_vblank(struct drm_device *dev, int pipe) return 0; } +static int ironlake_enable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + if (!i915_pipe_enabled(dev, pipe)) + return -EINVAL; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + ironlake_enable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK: DE_PIPEB_VBLANK); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); + + return 0; +} + +static int ivybridge_enable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + if (!i915_pipe_enabled(dev, pipe)) + return -EINVAL; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + ironlake_enable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK_IVB : DE_PIPEB_VBLANK_IVB); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); + + return 0; +} + /* Called from drm generic code, passed 'crtc' which * we use as a pipe index */ -void i915_disable_vblank(struct drm_device *dev, int pipe) +static void i915_disable_vblank(struct drm_device *dev, int pipe) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; unsigned long irqflags; @@ -1375,13 +1528,31 @@ void i915_disable_vblank(struct drm_device *dev, int pipe) I915_WRITE(INSTPM, INSTPM_AGPBUSY_DIS << 16 | INSTPM_AGPBUSY_DIS); - if (HAS_PCH_SPLIT(dev)) - ironlake_disable_display_irq(dev_priv, (pipe == 0) ? - DE_PIPEA_VBLANK: DE_PIPEB_VBLANK); - else - i915_disable_pipestat(dev_priv, pipe, - PIPE_VBLANK_INTERRUPT_ENABLE | - PIPE_START_VBLANK_INTERRUPT_ENABLE); + i915_disable_pipestat(dev_priv, pipe, + PIPE_VBLANK_INTERRUPT_ENABLE | + PIPE_START_VBLANK_INTERRUPT_ENABLE); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); +} + +static void ironlake_disable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + ironlake_disable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK: DE_PIPEB_VBLANK); + spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); +} + +static void ivybridge_disable_vblank(struct drm_device *dev, int pipe) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + unsigned long irqflags; + + spin_lock_irqsave(&dev_priv->irq_lock, irqflags); + ironlake_disable_display_irq(dev_priv, (pipe == 0) ? + DE_PIPEA_VBLANK_IVB : DE_PIPEB_VBLANK_IVB); spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); } @@ -1498,6 +1669,9 @@ void i915_hangcheck_elapsed(unsigned long data) uint32_t acthd, instdone, instdone1; bool err = false; + if (!i915_enable_hangcheck) + return; + /* If all work is done then ACTHD clearly hasn't advanced. */ if (i915_hangcheck_ring_idle(&dev_priv->ring[RCS], &err) && i915_hangcheck_ring_idle(&dev_priv->ring[VCS], &err) && @@ -1566,7 +1740,25 @@ static void ironlake_irq_preinstall(struct drm_device *dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + atomic_set(&dev_priv->irq_received, 0); + + INIT_WORK(&dev_priv->hotplug_work, i915_hotplug_work_func); + INIT_WORK(&dev_priv->error_work, i915_error_work_func); + if (IS_GEN6(dev) || IS_IVYBRIDGE(dev)) + INIT_WORK(&dev_priv->rps_work, gen6_pm_rps_work); + I915_WRITE(HWSTAM, 0xeffe); + if (IS_GEN6(dev) || IS_GEN7(dev)) { + /* Workaround stalls observed on Sandy Bridge GPUs by + * making the blitter command streamer generate a + * write to the Hardware Status Page for + * MI_USER_INTERRUPT. This appears to serialize the + * previous seqno write out before the interrupt + * happens. + */ + I915_WRITE(GEN6_BLITTER_HWSTAM, ~GEN6_BLITTER_USER_INTERRUPT); + I915_WRITE(GEN6_BSD_HWSTAM, ~GEN6_BSD_USER_INTERRUPT); + } /* XXX hotplug from PCH */ @@ -1594,6 +1786,13 @@ static int ironlake_irq_postinstall(struct drm_device *dev) u32 render_irqs; u32 hotplug_mask; + DRM_INIT_WAITQUEUE(&dev_priv->ring[RCS].irq_queue); + if (HAS_BSD(dev)) + DRM_INIT_WAITQUEUE(&dev_priv->ring[VCS].irq_queue); + if (HAS_BLT(dev)) + DRM_INIT_WAITQUEUE(&dev_priv->ring[BCS].irq_queue); + + dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B; dev_priv->irq_mask = ~display_mask; /* should always can generate irq */ @@ -1650,7 +1849,57 @@ static int ironlake_irq_postinstall(struct drm_device *dev) return 0; } -void i915_driver_irq_preinstall(struct drm_device * dev) +static int ivybridge_irq_postinstall(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + /* enable kind of interrupts always enabled */ + u32 display_mask = DE_MASTER_IRQ_CONTROL | DE_GSE_IVB | + DE_PCH_EVENT_IVB | DE_PLANEA_FLIP_DONE_IVB | + DE_PLANEB_FLIP_DONE_IVB; + u32 render_irqs; + u32 hotplug_mask; + + DRM_INIT_WAITQUEUE(&dev_priv->ring[RCS].irq_queue); + if (HAS_BSD(dev)) + DRM_INIT_WAITQUEUE(&dev_priv->ring[VCS].irq_queue); + if (HAS_BLT(dev)) + DRM_INIT_WAITQUEUE(&dev_priv->ring[BCS].irq_queue); + + dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B; + dev_priv->irq_mask = ~display_mask; + + /* should always can generate irq */ + I915_WRITE(DEIIR, I915_READ(DEIIR)); + I915_WRITE(DEIMR, dev_priv->irq_mask); + I915_WRITE(DEIER, display_mask | DE_PIPEA_VBLANK_IVB | + DE_PIPEB_VBLANK_IVB); + POSTING_READ(DEIER); + + dev_priv->gt_irq_mask = ~0; + + I915_WRITE(GTIIR, I915_READ(GTIIR)); + I915_WRITE(GTIMR, dev_priv->gt_irq_mask); + + render_irqs = GT_USER_INTERRUPT | GT_GEN6_BSD_USER_INTERRUPT | + GT_BLT_USER_INTERRUPT; + I915_WRITE(GTIER, render_irqs); + POSTING_READ(GTIER); + + hotplug_mask = (SDE_CRT_HOTPLUG_CPT | + SDE_PORTB_HOTPLUG_CPT | + SDE_PORTC_HOTPLUG_CPT | + SDE_PORTD_HOTPLUG_CPT); + dev_priv->pch_irq_mask = ~hotplug_mask; + + I915_WRITE(SDEIIR, I915_READ(SDEIIR)); + I915_WRITE(SDEIMR, dev_priv->pch_irq_mask); + I915_WRITE(SDEIER, hotplug_mask); + POSTING_READ(SDEIER); + + return 0; +} + +static void i915_driver_irq_preinstall(struct drm_device * dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; int pipe; @@ -1660,11 +1909,6 @@ void i915_driver_irq_preinstall(struct drm_device * dev) INIT_WORK(&dev_priv->hotplug_work, i915_hotplug_work_func); INIT_WORK(&dev_priv->error_work, i915_error_work_func); - if (HAS_PCH_SPLIT(dev)) { - ironlake_irq_preinstall(dev); - return; - } - if (I915_HAS_HOTPLUG(dev)) { I915_WRITE(PORT_HOTPLUG_EN, 0); I915_WRITE(PORT_HOTPLUG_STAT, I915_READ(PORT_HOTPLUG_STAT)); @@ -1682,23 +1926,14 @@ void i915_driver_irq_preinstall(struct drm_device * dev) * Must be called after intel_modeset_init or hotplug interrupts won't be * enabled correctly. */ -int i915_driver_irq_postinstall(struct drm_device *dev) +static int i915_driver_irq_postinstall(struct drm_device *dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; u32 enable_mask = I915_INTERRUPT_ENABLE_FIX | I915_INTERRUPT_ENABLE_VAR; u32 error_mask; - DRM_INIT_WAITQUEUE(&dev_priv->ring[RCS].irq_queue); - if (HAS_BSD(dev)) - DRM_INIT_WAITQUEUE(&dev_priv->ring[VCS].irq_queue); - if (HAS_BLT(dev)) - DRM_INIT_WAITQUEUE(&dev_priv->ring[BCS].irq_queue); - dev_priv->vblank_pipe = DRM_I915_VBLANK_PIPE_A | DRM_I915_VBLANK_PIPE_B; - if (HAS_PCH_SPLIT(dev)) - return ironlake_irq_postinstall(dev); - /* Unmask the interrupts that we always want on. */ dev_priv->irq_mask = ~I915_INTERRUPT_ENABLE_FIX; @@ -1770,6 +2005,12 @@ int i915_driver_irq_postinstall(struct drm_device *dev) static void ironlake_irq_uninstall(struct drm_device *dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; + + if (!dev_priv) + return; + + dev_priv->vblank_pipe = 0; + I915_WRITE(HWSTAM, 0xffffffff); I915_WRITE(DEIMR, 0xffffffff); @@ -1781,7 +2022,7 @@ static void ironlake_irq_uninstall(struct drm_device *dev) I915_WRITE(GTIIR, I915_READ(GTIIR)); } -void i915_driver_irq_uninstall(struct drm_device * dev) +static void i915_driver_irq_uninstall(struct drm_device * dev) { drm_i915_private_t *dev_priv = (drm_i915_private_t *) dev->dev_private; int pipe; @@ -1791,11 +2032,6 @@ void i915_driver_irq_uninstall(struct drm_device * dev) dev_priv->vblank_pipe = 0; - if (HAS_PCH_SPLIT(dev)) { - ironlake_irq_uninstall(dev); - return; - } - if (I915_HAS_HOTPLUG(dev)) { I915_WRITE(PORT_HOTPLUG_EN, 0); I915_WRITE(PORT_HOTPLUG_STAT, I915_READ(PORT_HOTPLUG_STAT)); @@ -1812,3 +2048,43 @@ void i915_driver_irq_uninstall(struct drm_device * dev) I915_READ(PIPESTAT(pipe)) & 0x8000ffff); I915_WRITE(IIR, I915_READ(IIR)); } + +void intel_irq_init(struct drm_device *dev) +{ + dev->driver->get_vblank_counter = i915_get_vblank_counter; + dev->max_vblank_count = 0xffffff; /* only 24 bits of frame count */ + if (IS_G4X(dev) || IS_GEN5(dev) || IS_GEN6(dev) || IS_IVYBRIDGE(dev)) { + dev->max_vblank_count = 0xffffffff; /* full 32 bit counter */ + dev->driver->get_vblank_counter = gm45_get_vblank_counter; + } + + if (drm_core_check_feature(dev, DRIVER_MODESET)) + dev->driver->get_vblank_timestamp = i915_get_vblank_timestamp; + else + dev->driver->get_vblank_timestamp = NULL; + dev->driver->get_scanout_position = i915_get_crtc_scanoutpos; + + if (IS_IVYBRIDGE(dev)) { + /* Share pre & uninstall handlers with ILK/SNB */ + dev->driver->irq_handler = ivybridge_irq_handler; + dev->driver->irq_preinstall = ironlake_irq_preinstall; + dev->driver->irq_postinstall = ivybridge_irq_postinstall; + dev->driver->irq_uninstall = ironlake_irq_uninstall; + dev->driver->enable_vblank = ivybridge_enable_vblank; + dev->driver->disable_vblank = ivybridge_disable_vblank; + } else if (HAS_PCH_SPLIT(dev)) { + dev->driver->irq_handler = ironlake_irq_handler; + dev->driver->irq_preinstall = ironlake_irq_preinstall; + dev->driver->irq_postinstall = ironlake_irq_postinstall; + dev->driver->irq_uninstall = ironlake_irq_uninstall; + dev->driver->enable_vblank = ironlake_enable_vblank; + dev->driver->disable_vblank = ironlake_disable_vblank; + } else { + dev->driver->irq_preinstall = i915_driver_irq_preinstall; + dev->driver->irq_postinstall = i915_driver_irq_postinstall; + dev->driver->irq_uninstall = i915_driver_irq_uninstall; + dev->driver->irq_handler = i915_driver_irq_handler; + dev->driver->enable_vblank = i915_enable_vblank; + dev->driver->disable_vblank = i915_disable_vblank; + } +} diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h index f39ac3a0fa9..542453f7498 100644 --- a/drivers/gpu/drm/i915/i915_reg.h +++ b/drivers/gpu/drm/i915/i915_reg.h @@ -78,6 +78,14 @@ #define GRDOM_RENDER (1<<2) #define GRDOM_MEDIA (3<<2) +#define GEN6_MBCUNIT_SNPCR 0x900c /* for LLC config */ +#define GEN6_MBC_SNPCR_SHIFT 21 +#define GEN6_MBC_SNPCR_MASK (3<<21) +#define GEN6_MBC_SNPCR_MAX (0<<21) +#define GEN6_MBC_SNPCR_MED (1<<21) +#define GEN6_MBC_SNPCR_LOW (2<<21) +#define GEN6_MBC_SNPCR_MIN (3<<21) /* only 1/16th of the cache is shared */ + #define GEN6_GDRST 0x941c #define GEN6_GRDOM_FULL (1 << 0) #define GEN6_GRDOM_RENDER (1 << 1) @@ -291,6 +299,9 @@ #define RING_MAX_IDLE(base) ((base)+0x54) #define RING_HWS_PGA(base) ((base)+0x80) #define RING_HWS_PGA_GEN6(base) ((base)+0x2080) +#define RENDER_HWS_PGA_GEN7 (0x04080) +#define BSD_HWS_PGA_GEN7 (0x04180) +#define BLT_HWS_PGA_GEN7 (0x04280) #define RING_ACTHD(base) ((base)+0x74) #define RING_NOPID(base) ((base)+0x94) #define RING_IMR(base) ((base)+0xa8) @@ -364,6 +375,7 @@ # define MI_FLUSH_ENABLE (1 << 11) #define GFX_MODE 0x02520 +#define GFX_MODE_GEN7 0x0229c #define GFX_RUN_LIST_ENABLE (1<<15) #define GFX_TLB_INVALIDATE_ALWAYS (1<<13) #define GFX_SURFACE_FAULT_ENABLE (1<<12) @@ -371,6 +383,9 @@ #define GFX_PSMI_GRANULARITY (1<<10) #define GFX_PPGTT_ENABLE (1<<9) +#define GFX_MODE_ENABLE(bit) (((bit) << 16) | (bit)) +#define GFX_MODE_DISABLE(bit) (((bit) << 16) | (0)) + #define SCPD0 0x0209c /* 915+ only */ #define IER 0x020a0 #define IIR 0x020a4 @@ -528,6 +543,7 @@ #define GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_ENABLE 0 #define GEN6_BSD_SLEEP_PSMI_CONTROL_IDLE_INDICATOR (1 << 3) +#define GEN6_BSD_HWSTAM 0x12098 #define GEN6_BSD_IMR 0x120a8 #define GEN6_BSD_USER_INTERRUPT (1 << 12) @@ -575,6 +591,7 @@ #define DPFC_CTL_PLANEA (0<<30) #define DPFC_CTL_PLANEB (1<<30) #define DPFC_CTL_FENCE_EN (1<<29) +#define DPFC_CTL_PERSISTENT_MODE (1<<25) #define DPFC_SR_EN (1<<10) #define DPFC_CTL_LIMIT_1X (0<<6) #define DPFC_CTL_LIMIT_2X (1<<6) @@ -1305,6 +1322,7 @@ #define ADPA_PIPE_SELECT_MASK (1<<30) #define ADPA_PIPE_A_SELECT 0 #define ADPA_PIPE_B_SELECT (1<<30) +#define ADPA_PIPE_SELECT(pipe) ((pipe) << 30) #define ADPA_USE_VGA_HVPOLARITY (1<<15) #define ADPA_SETS_HVPOLARITY 0 #define ADPA_VSYNC_CNTL_DISABLE (1<<11) @@ -1447,6 +1465,7 @@ /* Selects pipe B for LVDS data. Must be set on pre-965. */ #define LVDS_PIPEB_SELECT (1 << 30) #define LVDS_PIPE_MASK (1 << 30) +#define LVDS_PIPE(pipe) ((pipe) << 30) /* LVDS dithering flag on 965/g4x platform */ #define LVDS_ENABLE_DITHER (1 << 25) /* LVDS sync polarity flags. Set to invert (i.e. negative) */ @@ -1486,9 +1505,6 @@ #define LVDS_B0B3_POWER_DOWN (0 << 2) #define LVDS_B0B3_POWER_UP (3 << 2) -#define LVDS_PIPE_ENABLED(V, P) \ - (((V) & (LVDS_PIPE_MASK | LVDS_PORT_EN)) == ((P) << 30 | LVDS_PORT_EN)) - /* Video Data Island Packet control */ #define VIDEO_DIP_DATA 0x61178 #define VIDEO_DIP_CTL 0x61170 @@ -1501,6 +1517,7 @@ #define VIDEO_DIP_SELECT_AVI (0 << 19) #define VIDEO_DIP_SELECT_VENDOR (1 << 19) #define VIDEO_DIP_SELECT_SPD (3 << 19) +#define VIDEO_DIP_SELECT_MASK (3 << 19) #define VIDEO_DIP_FREQ_ONCE (0 << 16) #define VIDEO_DIP_FREQ_VSYNC (1 << 16) #define VIDEO_DIP_FREQ_2VSYNC (2 << 16) @@ -2079,9 +2096,6 @@ #define DP_PIPEB_SELECT (1 << 30) #define DP_PIPE_MASK (1 << 30) -#define DP_PIPE_ENABLED(V, P) \ - (((V) & (DP_PIPE_MASK | DP_PORT_EN)) == ((P) << 30 | DP_PORT_EN)) - /* Link training mode - select a suitable mode for each stage */ #define DP_LINK_TRAIN_PAT_1 (0 << 28) #define DP_LINK_TRAIN_PAT_2 (1 << 28) @@ -2778,6 +2792,19 @@ #define DE_PIPEA_VSYNC (1 << 3) #define DE_PIPEA_FIFO_UNDERRUN (1 << 0) +/* More Ivybridge lolz */ +#define DE_ERR_DEBUG_IVB (1<<30) +#define DE_GSE_IVB (1<<29) +#define DE_PCH_EVENT_IVB (1<<28) +#define DE_DP_A_HOTPLUG_IVB (1<<27) +#define DE_AUX_CHANNEL_A_IVB (1<<26) +#define DE_SPRITEB_FLIP_DONE_IVB (1<<9) +#define DE_SPRITEA_FLIP_DONE_IVB (1<<4) +#define DE_PLANEB_FLIP_DONE_IVB (1<<8) +#define DE_PLANEA_FLIP_DONE_IVB (1<<3) +#define DE_PIPEB_VBLANK_IVB (1<<5) +#define DE_PIPEA_VBLANK_IVB (1<<0) + #define DEISR 0x44000 #define DEIMR 0x44004 #define DEIIR 0x44008 @@ -2809,6 +2836,7 @@ #define ILK_eDP_A_DISABLE (1<<24) #define ILK_DESKTOP (1<<23) #define ILK_DSPCLK_GATE 0x42020 +#define IVB_VRHUNIT_CLK_GATE (1<<28) #define ILK_DPARB_CLK_GATE (1<<5) #define ILK_DPFD_CLK_GATE (1<<7) @@ -3005,6 +3033,20 @@ #define _TRANSA_DP_LINK_M2 0xe0048 #define _TRANSA_DP_LINK_N2 0xe004c +/* Per-transcoder DIP controls */ + +#define _VIDEO_DIP_CTL_A 0xe0200 +#define _VIDEO_DIP_DATA_A 0xe0208 +#define _VIDEO_DIP_GCP_A 0xe0210 + +#define _VIDEO_DIP_CTL_B 0xe1200 +#define _VIDEO_DIP_DATA_B 0xe1208 +#define _VIDEO_DIP_GCP_B 0xe1210 + +#define TVIDEO_DIP_CTL(pipe) _PIPE(pipe, _VIDEO_DIP_CTL_A, _VIDEO_DIP_CTL_B) +#define TVIDEO_DIP_DATA(pipe) _PIPE(pipe, _VIDEO_DIP_DATA_A, _VIDEO_DIP_DATA_B) +#define TVIDEO_DIP_GCP(pipe) _PIPE(pipe, _VIDEO_DIP_GCP_A, _VIDEO_DIP_GCP_B) + #define _TRANS_HTOTAL_B 0xe1000 #define _TRANS_HBLANK_B 0xe1004 #define _TRANS_HSYNC_B 0xe1008 @@ -3057,6 +3099,19 @@ #define TRANS_6BPC (2<<5) #define TRANS_12BPC (3<<5) +#define _TRANSA_CHICKEN2 0xf0064 +#define _TRANSB_CHICKEN2 0xf1064 +#define TRANS_CHICKEN2(pipe) _PIPE(pipe, _TRANSA_CHICKEN2, _TRANSB_CHICKEN2) +#define TRANS_AUTOTRAIN_GEN_STALL_DIS (1<<31) + +#define SOUTH_CHICKEN1 0xc2000 +#define FDIA_PHASE_SYNC_SHIFT_OVR 19 +#define FDIA_PHASE_SYNC_SHIFT_EN 18 +#define FDI_PHASE_SYNC_OVR(pipe) (1<<(FDIA_PHASE_SYNC_SHIFT_OVR - ((pipe) * 2))) +#define FDI_PHASE_SYNC_EN(pipe) (1<<(FDIA_PHASE_SYNC_SHIFT_EN - ((pipe) * 2))) +#define SOUTH_CHICKEN2 0xc2004 +#define DPLS_EDP_PPS_FIX_DIS (1<<0) + #define _FDI_RXA_CHICKEN 0xc200c #define _FDI_RXB_CHICKEN 0xc2010 #define FDI_RX_PHASE_SYNC_POINTER_OVR (1<<1) @@ -3104,7 +3159,15 @@ #define FDI_TX_ENHANCE_FRAME_ENABLE (1<<18) /* Ironlake: hardwired to 1 */ #define FDI_TX_PLL_ENABLE (1<<14) + +/* Ivybridge has different bits for lolz */ +#define FDI_LINK_TRAIN_PATTERN_1_IVB (0<<8) +#define FDI_LINK_TRAIN_PATTERN_2_IVB (1<<8) +#define FDI_LINK_TRAIN_PATTERN_IDLE_IVB (2<<8) +#define FDI_LINK_TRAIN_NONE_IVB (3<<8) + /* both Tx and Rx */ +#define FDI_LINK_TRAIN_AUTO (1<<10) #define FDI_SCRAMBLING_ENABLE (0<<7) #define FDI_SCRAMBLING_DISABLE (1<<7) @@ -3114,6 +3177,8 @@ #define FDI_RX_CTL(pipe) _PIPE(pipe, _FDI_RXA_CTL, _FDI_RXB_CTL) #define FDI_RX_ENABLE (1<<31) /* train, dp width same as FDI_TX */ +#define FDI_FS_ERRC_ENABLE (1<<27) +#define FDI_FE_ERRC_ENABLE (1<<26) #define FDI_DP_PORT_WIDTH_X8 (7<<19) #define FDI_8BPC (0<<16) #define FDI_10BPC (1<<16) @@ -3194,14 +3259,12 @@ #define ADPA_CRT_HOTPLUG_VOLREF_475MV (1<<17) #define ADPA_CRT_HOTPLUG_FORCE_TRIGGER (1<<16) -#define ADPA_PIPE_ENABLED(V, P) \ - (((V) & (ADPA_TRANS_SELECT_MASK | ADPA_DAC_ENABLE)) == ((P) << 30 | ADPA_DAC_ENABLE)) - /* or SDVOB */ #define HDMIB 0xe1140 #define PORT_ENABLE (1 << 31) #define TRANSCODER_A (0) #define TRANSCODER_B (1 << 30) +#define TRANSCODER(pipe) ((pipe) << 30) #define TRANSCODER_MASK (1 << 30) #define COLOR_FORMAT_8bpc (0) #define COLOR_FORMAT_12bpc (3 << 26) @@ -3218,9 +3281,6 @@ #define HSYNC_ACTIVE_HIGH (1 << 3) #define PORT_DETECTED (1 << 2) -#define HDMI_PIPE_ENABLED(V, P) \ - (((V) & (TRANSCODER_MASK | PORT_ENABLE)) == ((P) << 30 | PORT_ENABLE)) - /* PCH SDVOB multiplex with HDMIB */ #define PCH_SDVOB HDMIB @@ -3287,6 +3347,7 @@ #define PORT_TRANS_B_SEL_CPT (1<<29) #define PORT_TRANS_C_SEL_CPT (2<<29) #define PORT_TRANS_SEL_MASK (3<<29) +#define PORT_TRANS_SEL_CPT(pipe) ((pipe) << 29) #define TRANS_DP_CTL_A 0xe0300 #define TRANS_DP_CTL_B 0xe1300 @@ -3329,6 +3390,7 @@ #define FORCEWAKE_ACK 0x130090 #define GT_FIFO_FREE_ENTRIES 0x120008 +#define GT_FIFO_NUM_RESERVED_ENTRIES 20 #define GEN6_RPNSWREQ 0xA008 #define GEN6_TURBO_DISABLE (1<<31) @@ -3386,7 +3448,7 @@ #define GEN6_PMINTRMSK 0xA168 #define GEN6_PMISR 0x44020 -#define GEN6_PMIMR 0x44024 +#define GEN6_PMIMR 0x44024 /* rps_lock */ #define GEN6_PMIIR 0x44028 #define GEN6_PMIER 0x4402C #define GEN6_PM_MBOX_EVENT (1<<25) @@ -3396,11 +3458,16 @@ #define GEN6_PM_RP_DOWN_THRESHOLD (1<<4) #define GEN6_PM_RP_UP_EI_EXPIRED (1<<2) #define GEN6_PM_RP_DOWN_EI_EXPIRED (1<<1) +#define GEN6_PM_DEFERRED_EVENTS (GEN6_PM_RP_UP_THRESHOLD | \ + GEN6_PM_RP_DOWN_THRESHOLD | \ + GEN6_PM_RP_DOWN_TIMEOUT) #define GEN6_PCODE_MAILBOX 0x138124 #define GEN6_PCODE_READY (1<<31) #define GEN6_READ_OC_PARAMS 0xc -#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x9 +#define GEN6_PCODE_WRITE_MIN_FREQ_TABLE 0x8 +#define GEN6_PCODE_READ_MIN_FREQ_TABLE 0x9 #define GEN6_PCODE_DATA 0x138128 +#define GEN6_PCODE_FREQ_IA_RATIO_SHIFT 8 #endif /* _I915_REG_H_ */ diff --git a/drivers/gpu/drm/i915/i915_suspend.c b/drivers/gpu/drm/i915/i915_suspend.c index da474153a0a..f10742359ec 100644 --- a/drivers/gpu/drm/i915/i915_suspend.c +++ b/drivers/gpu/drm/i915/i915_suspend.c @@ -597,7 +597,7 @@ static void i915_restore_modeset_reg(struct drm_device *dev) return; } -void i915_save_display(struct drm_device *dev) +static void i915_save_display(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -689,7 +689,7 @@ void i915_save_display(struct drm_device *dev) i915_save_vga(dev); } -void i915_restore_display(struct drm_device *dev) +static void i915_restore_display(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -760,15 +760,13 @@ void i915_restore_display(struct drm_device *dev) /* FIXME: restore TV & SDVO state */ /* only restore FBC info on the platform that supports FBC*/ + intel_disable_fbc(dev); if (I915_HAS_FBC(dev)) { if (HAS_PCH_SPLIT(dev)) { - ironlake_disable_fbc(dev); I915_WRITE(ILK_DPFC_CB_BASE, dev_priv->saveDPFC_CB_BASE); } else if (IS_GM45(dev)) { - g4x_disable_fbc(dev); I915_WRITE(DPFC_CB_BASE, dev_priv->saveDPFC_CB_BASE); } else { - i8xx_disable_fbc(dev); I915_WRITE(FBC_CFB_BASE, dev_priv->saveFBC_CFB_BASE); I915_WRITE(FBC_LL_BASE, dev_priv->saveFBC_LL_BASE); I915_WRITE(FBC_CONTROL2, dev_priv->saveFBC_CONTROL2); @@ -780,6 +778,7 @@ void i915_restore_display(struct drm_device *dev) I915_WRITE(CPU_VGACNTRL, dev_priv->saveVGACNTRL); else I915_WRITE(VGACNTRL, dev_priv->saveVGACNTRL); + I915_WRITE(VGA0, dev_priv->saveVGA0); I915_WRITE(VGA1, dev_priv->saveVGA1); I915_WRITE(VGA_PD, dev_priv->saveVGA_PD); @@ -796,6 +795,8 @@ int i915_save_state(struct drm_device *dev) pci_read_config_byte(dev->pdev, LBB, &dev_priv->saveLBB); + mutex_lock(&dev->struct_mutex); + /* Hardware status page */ dev_priv->saveHWS = I915_READ(HWS_PGA); @@ -811,6 +812,7 @@ int i915_save_state(struct drm_device *dev) dev_priv->saveFDI_RXB_IMR = I915_READ(_FDI_RXB_IMR); dev_priv->saveMCHBAR_RENDER_STANDBY = I915_READ(RSTDBYCTL); + dev_priv->savePCH_PORT_HOTPLUG = I915_READ(PCH_PORT_HOTPLUG); } else { dev_priv->saveIER = I915_READ(IER); dev_priv->saveIMR = I915_READ(IMR); @@ -835,6 +837,8 @@ int i915_save_state(struct drm_device *dev) for (i = 0; i < 3; i++) dev_priv->saveSWF2[i] = I915_READ(SWF30 + (i << 2)); + mutex_unlock(&dev->struct_mutex); + return 0; } @@ -845,6 +849,8 @@ int i915_restore_state(struct drm_device *dev) pci_write_config_byte(dev->pdev, LBB, dev_priv->saveLBB); + mutex_lock(&dev->struct_mutex); + /* Hardware status page */ I915_WRITE(HWS_PGA, dev_priv->saveHWS); @@ -858,21 +864,27 @@ int i915_restore_state(struct drm_device *dev) I915_WRITE(GTIMR, dev_priv->saveGTIMR); I915_WRITE(_FDI_RXA_IMR, dev_priv->saveFDI_RXA_IMR); I915_WRITE(_FDI_RXB_IMR, dev_priv->saveFDI_RXB_IMR); + I915_WRITE(PCH_PORT_HOTPLUG, dev_priv->savePCH_PORT_HOTPLUG); } else { I915_WRITE(IER, dev_priv->saveIER); I915_WRITE(IMR, dev_priv->saveIMR); } + mutex_unlock(&dev->struct_mutex); - /* Clock gating state */ - intel_enable_clock_gating(dev); + if (drm_core_check_feature(dev, DRIVER_MODESET)) + intel_init_clock_gating(dev); if (IS_IRONLAKE_M(dev)) { ironlake_enable_drps(dev); intel_init_emon(dev); } - if (IS_GEN6(dev)) + if (IS_GEN6(dev)) { gen6_enable_rps(dev_priv); + gen6_update_ring_freq(dev_priv); + } + + mutex_lock(&dev->struct_mutex); /* Cache mode state */ I915_WRITE (CACHE_MODE_0, dev_priv->saveCACHE_MODE_0 | 0xffff0000); @@ -887,6 +899,8 @@ int i915_restore_state(struct drm_device *dev) for (i = 0; i < 3; i++) I915_WRITE(SWF30 + (i << 2), dev_priv->saveSWF2[i]); + mutex_unlock(&dev->struct_mutex); + intel_i2c_reset(dev); return 0; diff --git a/drivers/gpu/drm/i915/intel_bios.c b/drivers/gpu/drm/i915/intel_bios.c index fb5b4d426ae..61abef8a811 100644 --- a/drivers/gpu/drm/i915/intel_bios.c +++ b/drivers/gpu/drm/i915/intel_bios.c @@ -74,7 +74,7 @@ get_blocksize(void *p) static void fill_detail_timing_data(struct drm_display_mode *panel_fixed_mode, - struct lvds_dvo_timing *dvo_timing) + const struct lvds_dvo_timing *dvo_timing) { panel_fixed_mode->hdisplay = (dvo_timing->hactive_hi << 8) | dvo_timing->hactive_lo; @@ -115,20 +115,75 @@ fill_detail_timing_data(struct drm_display_mode *panel_fixed_mode, drm_mode_set_name(panel_fixed_mode); } +static bool +lvds_dvo_timing_equal_size(const struct lvds_dvo_timing *a, + const struct lvds_dvo_timing *b) +{ + if (a->hactive_hi != b->hactive_hi || + a->hactive_lo != b->hactive_lo) + return false; + + if (a->hsync_off_hi != b->hsync_off_hi || + a->hsync_off_lo != b->hsync_off_lo) + return false; + + if (a->hsync_pulse_width != b->hsync_pulse_width) + return false; + + if (a->hblank_hi != b->hblank_hi || + a->hblank_lo != b->hblank_lo) + return false; + + if (a->vactive_hi != b->vactive_hi || + a->vactive_lo != b->vactive_lo) + return false; + + if (a->vsync_off != b->vsync_off) + return false; + + if (a->vsync_pulse_width != b->vsync_pulse_width) + return false; + + if (a->vblank_hi != b->vblank_hi || + a->vblank_lo != b->vblank_lo) + return false; + + return true; +} + +static const struct lvds_dvo_timing * +get_lvds_dvo_timing(const struct bdb_lvds_lfp_data *lvds_lfp_data, + const struct bdb_lvds_lfp_data_ptrs *lvds_lfp_data_ptrs, + int index) +{ + /* + * the size of fp_timing varies on the different platform. + * So calculate the DVO timing relative offset in LVDS data + * entry to get the DVO timing entry + */ + + int lfp_data_size = + lvds_lfp_data_ptrs->ptr[1].dvo_timing_offset - + lvds_lfp_data_ptrs->ptr[0].dvo_timing_offset; + int dvo_timing_offset = + lvds_lfp_data_ptrs->ptr[0].dvo_timing_offset - + lvds_lfp_data_ptrs->ptr[0].fp_timing_offset; + char *entry = (char *)lvds_lfp_data->data + lfp_data_size * index; + + return (struct lvds_dvo_timing *)(entry + dvo_timing_offset); +} + /* Try to find integrated panel data */ static void parse_lfp_panel_data(struct drm_i915_private *dev_priv, struct bdb_header *bdb) { - struct bdb_lvds_options *lvds_options; - struct bdb_lvds_lfp_data *lvds_lfp_data; - struct bdb_lvds_lfp_data_ptrs *lvds_lfp_data_ptrs; - struct bdb_lvds_lfp_data_entry *entry; - struct lvds_dvo_timing *dvo_timing; + const struct bdb_lvds_options *lvds_options; + const struct bdb_lvds_lfp_data *lvds_lfp_data; + const struct bdb_lvds_lfp_data_ptrs *lvds_lfp_data_ptrs; + const struct lvds_dvo_timing *panel_dvo_timing; struct drm_display_mode *panel_fixed_mode; - int lfp_data_size, dvo_timing_offset; - int i, temp_downclock; - struct drm_display_mode *temp_mode; + int i, downclock; lvds_options = find_section(bdb, BDB_LVDS_OPTIONS); if (!lvds_options) @@ -150,75 +205,44 @@ parse_lfp_panel_data(struct drm_i915_private *dev_priv, dev_priv->lvds_vbt = 1; - lfp_data_size = lvds_lfp_data_ptrs->ptr[1].dvo_timing_offset - - lvds_lfp_data_ptrs->ptr[0].dvo_timing_offset; - entry = (struct bdb_lvds_lfp_data_entry *) - ((uint8_t *)lvds_lfp_data->data + (lfp_data_size * - lvds_options->panel_type)); - dvo_timing_offset = lvds_lfp_data_ptrs->ptr[0].dvo_timing_offset - - lvds_lfp_data_ptrs->ptr[0].fp_timing_offset; - - /* - * the size of fp_timing varies on the different platform. - * So calculate the DVO timing relative offset in LVDS data - * entry to get the DVO timing entry - */ - dvo_timing = (struct lvds_dvo_timing *) - ((unsigned char *)entry + dvo_timing_offset); + panel_dvo_timing = get_lvds_dvo_timing(lvds_lfp_data, + lvds_lfp_data_ptrs, + lvds_options->panel_type); panel_fixed_mode = kzalloc(sizeof(*panel_fixed_mode), GFP_KERNEL); if (!panel_fixed_mode) return; - fill_detail_timing_data(panel_fixed_mode, dvo_timing); + fill_detail_timing_data(panel_fixed_mode, panel_dvo_timing); dev_priv->lfp_lvds_vbt_mode = panel_fixed_mode; DRM_DEBUG_KMS("Found panel mode in BIOS VBT tables:\n"); drm_mode_debug_printmodeline(panel_fixed_mode); - temp_mode = kzalloc(sizeof(*temp_mode), GFP_KERNEL); - temp_downclock = panel_fixed_mode->clock; /* - * enumerate the LVDS panel timing info entry in VBT to check whether - * the LVDS downclock is found. + * Iterate over the LVDS panel timing info to find the lowest clock + * for the native resolution. */ + downclock = panel_dvo_timing->clock; for (i = 0; i < 16; i++) { - entry = (struct bdb_lvds_lfp_data_entry *) - ((uint8_t *)lvds_lfp_data->data + (lfp_data_size * i)); - dvo_timing = (struct lvds_dvo_timing *) - ((unsigned char *)entry + dvo_timing_offset); - - fill_detail_timing_data(temp_mode, dvo_timing); - - if (temp_mode->hdisplay == panel_fixed_mode->hdisplay && - temp_mode->hsync_start == panel_fixed_mode->hsync_start && - temp_mode->hsync_end == panel_fixed_mode->hsync_end && - temp_mode->htotal == panel_fixed_mode->htotal && - temp_mode->vdisplay == panel_fixed_mode->vdisplay && - temp_mode->vsync_start == panel_fixed_mode->vsync_start && - temp_mode->vsync_end == panel_fixed_mode->vsync_end && - temp_mode->vtotal == panel_fixed_mode->vtotal && - temp_mode->clock < temp_downclock) { - /* - * downclock is already found. But we expect - * to find the lower downclock. - */ - temp_downclock = temp_mode->clock; - } - /* clear it to zero */ - memset(temp_mode, 0, sizeof(*temp_mode)); + const struct lvds_dvo_timing *dvo_timing; + + dvo_timing = get_lvds_dvo_timing(lvds_lfp_data, + lvds_lfp_data_ptrs, + i); + if (lvds_dvo_timing_equal_size(dvo_timing, panel_dvo_timing) && + dvo_timing->clock < downclock) + downclock = dvo_timing->clock; } - kfree(temp_mode); - if (temp_downclock < panel_fixed_mode->clock && - i915_lvds_downclock) { + + if (downclock < panel_dvo_timing->clock && i915_lvds_downclock) { dev_priv->lvds_downclock_avail = 1; - dev_priv->lvds_downclock = temp_downclock; - DRM_DEBUG_KMS("LVDS downclock is found in VBT. ", - "Normal Clock %dKHz, downclock %dKHz\n", - temp_downclock, panel_fixed_mode->clock); + dev_priv->lvds_downclock = downclock * 10; + DRM_DEBUG_KMS("LVDS downclock is found in VBT. " + "Normal Clock %dKHz, downclock %dKHz\n", + panel_fixed_mode->clock, 10*downclock); } - return; } /* Try to find sdvo panel data */ diff --git a/drivers/gpu/drm/i915/intel_crt.c b/drivers/gpu/drm/i915/intel_crt.c index d03fc05b39c..0979d887788 100644 --- a/drivers/gpu/drm/i915/intel_crt.c +++ b/drivers/gpu/drm/i915/intel_crt.c @@ -288,6 +288,8 @@ static bool intel_crt_detect_ddc(struct drm_connector *connector) * This may be a DVI-I connector with a shared DDC * link between analog and digital outputs, so we * have to check the EDID input spec of the attached device. + * + * On the other hand, what should we do if it is a broken EDID? */ if (edid != NULL) { is_digital = edid->input & DRM_EDID_INPUT_DIGITAL; @@ -298,6 +300,8 @@ static bool intel_crt_detect_ddc(struct drm_connector *connector) if (!is_digital) { DRM_DEBUG_KMS("CRT detected via DDC:0x50 [EDID]\n"); return true; + } else { + DRM_DEBUG_KMS("CRT not detected via DDC:0x50 [EDID reports a digital panel]\n"); } } @@ -305,13 +309,11 @@ static bool intel_crt_detect_ddc(struct drm_connector *connector) } static enum drm_connector_status -intel_crt_load_detect(struct drm_crtc *crtc, struct intel_crt *crt) +intel_crt_load_detect(struct intel_crt *crt) { - struct drm_encoder *encoder = &crt->base.base; - struct drm_device *dev = encoder->dev; + struct drm_device *dev = crt->base.base.dev; struct drm_i915_private *dev_priv = dev->dev_private; - struct intel_crtc *intel_crtc = to_intel_crtc(crtc); - uint32_t pipe = intel_crtc->pipe; + uint32_t pipe = to_intel_crtc(crt->base.base.crtc)->pipe; uint32_t save_bclrpat; uint32_t save_vtotal; uint32_t vtotal, vactive; @@ -432,7 +434,6 @@ intel_crt_detect(struct drm_connector *connector, bool force) struct drm_device *dev = connector->dev; struct intel_crt *crt = intel_attached_crt(connector); struct drm_crtc *crtc; - int dpms_mode; enum drm_connector_status status; if (I915_HAS_HOTPLUG(dev)) { @@ -454,17 +455,18 @@ intel_crt_detect(struct drm_connector *connector, bool force) /* for pre-945g platforms use load detect */ crtc = crt->base.base.crtc; if (crtc && crtc->enabled) { - status = intel_crt_load_detect(crtc, crt); + status = intel_crt_load_detect(crt); } else { - crtc = intel_get_load_detect_pipe(&crt->base, connector, - NULL, &dpms_mode); - if (crtc) { + struct intel_load_detect_pipe tmp; + + if (intel_get_load_detect_pipe(&crt->base, connector, NULL, + &tmp)) { if (intel_crt_detect_ddc(connector)) status = connector_status_connected; else - status = intel_crt_load_detect(crtc, crt); - intel_release_load_detect_pipe(&crt->base, - connector, dpms_mode); + status = intel_crt_load_detect(crt); + intel_release_load_detect_pipe(&crt->base, connector, + &tmp); } else status = connector_status_unknown; } diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index 373c2a005ec..56a8554d903 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -24,6 +24,7 @@ * Eric Anholt <eric@anholt.net> */ +#include <linux/cpufreq.h> #include <linux/module.h> #include <linux/input.h> #include <linux/i2c.h> @@ -76,255 +77,6 @@ struct intel_limit { int, int, intel_clock_t *); }; -#define I8XX_DOT_MIN 25000 -#define I8XX_DOT_MAX 350000 -#define I8XX_VCO_MIN 930000 -#define I8XX_VCO_MAX 1400000 -#define I8XX_N_MIN 3 -#define I8XX_N_MAX 16 -#define I8XX_M_MIN 96 -#define I8XX_M_MAX 140 -#define I8XX_M1_MIN 18 -#define I8XX_M1_MAX 26 -#define I8XX_M2_MIN 6 -#define I8XX_M2_MAX 16 -#define I8XX_P_MIN 4 -#define I8XX_P_MAX 128 -#define I8XX_P1_MIN 2 -#define I8XX_P1_MAX 33 -#define I8XX_P1_LVDS_MIN 1 -#define I8XX_P1_LVDS_MAX 6 -#define I8XX_P2_SLOW 4 -#define I8XX_P2_FAST 2 -#define I8XX_P2_LVDS_SLOW 14 -#define I8XX_P2_LVDS_FAST 7 -#define I8XX_P2_SLOW_LIMIT 165000 - -#define I9XX_DOT_MIN 20000 -#define I9XX_DOT_MAX 400000 -#define I9XX_VCO_MIN 1400000 -#define I9XX_VCO_MAX 2800000 -#define PINEVIEW_VCO_MIN 1700000 -#define PINEVIEW_VCO_MAX 3500000 -#define I9XX_N_MIN 1 -#define I9XX_N_MAX 6 -/* Pineview's Ncounter is a ring counter */ -#define PINEVIEW_N_MIN 3 -#define PINEVIEW_N_MAX 6 -#define I9XX_M_MIN 70 -#define I9XX_M_MAX 120 -#define PINEVIEW_M_MIN 2 -#define PINEVIEW_M_MAX 256 -#define I9XX_M1_MIN 10 -#define I9XX_M1_MAX 22 -#define I9XX_M2_MIN 5 -#define I9XX_M2_MAX 9 -/* Pineview M1 is reserved, and must be 0 */ -#define PINEVIEW_M1_MIN 0 -#define PINEVIEW_M1_MAX 0 -#define PINEVIEW_M2_MIN 0 -#define PINEVIEW_M2_MAX 254 -#define I9XX_P_SDVO_DAC_MIN 5 -#define I9XX_P_SDVO_DAC_MAX 80 -#define I9XX_P_LVDS_MIN 7 -#define I9XX_P_LVDS_MAX 98 -#define PINEVIEW_P_LVDS_MIN 7 -#define PINEVIEW_P_LVDS_MAX 112 -#define I9XX_P1_MIN 1 -#define I9XX_P1_MAX 8 -#define I9XX_P2_SDVO_DAC_SLOW 10 -#define I9XX_P2_SDVO_DAC_FAST 5 -#define I9XX_P2_SDVO_DAC_SLOW_LIMIT 200000 -#define I9XX_P2_LVDS_SLOW 14 -#define I9XX_P2_LVDS_FAST 7 -#define I9XX_P2_LVDS_SLOW_LIMIT 112000 - -/*The parameter is for SDVO on G4x platform*/ -#define G4X_DOT_SDVO_MIN 25000 -#define G4X_DOT_SDVO_MAX 270000 -#define G4X_VCO_MIN 1750000 -#define G4X_VCO_MAX 3500000 -#define G4X_N_SDVO_MIN 1 -#define G4X_N_SDVO_MAX 4 -#define G4X_M_SDVO_MIN 104 -#define G4X_M_SDVO_MAX 138 -#define G4X_M1_SDVO_MIN 17 -#define G4X_M1_SDVO_MAX 23 -#define G4X_M2_SDVO_MIN 5 -#define G4X_M2_SDVO_MAX 11 -#define G4X_P_SDVO_MIN 10 -#define G4X_P_SDVO_MAX 30 -#define G4X_P1_SDVO_MIN 1 -#define G4X_P1_SDVO_MAX 3 -#define G4X_P2_SDVO_SLOW 10 -#define G4X_P2_SDVO_FAST 10 -#define G4X_P2_SDVO_LIMIT 270000 - -/*The parameter is for HDMI_DAC on G4x platform*/ -#define G4X_DOT_HDMI_DAC_MIN 22000 -#define G4X_DOT_HDMI_DAC_MAX 400000 -#define G4X_N_HDMI_DAC_MIN 1 -#define G4X_N_HDMI_DAC_MAX 4 -#define G4X_M_HDMI_DAC_MIN 104 -#define G4X_M_HDMI_DAC_MAX 138 -#define G4X_M1_HDMI_DAC_MIN 16 -#define G4X_M1_HDMI_DAC_MAX 23 -#define G4X_M2_HDMI_DAC_MIN 5 -#define G4X_M2_HDMI_DAC_MAX 11 -#define G4X_P_HDMI_DAC_MIN 5 -#define G4X_P_HDMI_DAC_MAX 80 -#define G4X_P1_HDMI_DAC_MIN 1 -#define G4X_P1_HDMI_DAC_MAX 8 -#define G4X_P2_HDMI_DAC_SLOW 10 -#define G4X_P2_HDMI_DAC_FAST 5 -#define G4X_P2_HDMI_DAC_LIMIT 165000 - -/*The parameter is for SINGLE_CHANNEL_LVDS on G4x platform*/ -#define G4X_DOT_SINGLE_CHANNEL_LVDS_MIN 20000 -#define G4X_DOT_SINGLE_CHANNEL_LVDS_MAX 115000 -#define G4X_N_SINGLE_CHANNEL_LVDS_MIN 1 -#define G4X_N_SINGLE_CHANNEL_LVDS_MAX 3 -#define G4X_M_SINGLE_CHANNEL_LVDS_MIN 104 -#define G4X_M_SINGLE_CHANNEL_LVDS_MAX 138 -#define G4X_M1_SINGLE_CHANNEL_LVDS_MIN 17 -#define G4X_M1_SINGLE_CHANNEL_LVDS_MAX 23 -#define G4X_M2_SINGLE_CHANNEL_LVDS_MIN 5 -#define G4X_M2_SINGLE_CHANNEL_LVDS_MAX 11 -#define G4X_P_SINGLE_CHANNEL_LVDS_MIN 28 -#define G4X_P_SINGLE_CHANNEL_LVDS_MAX 112 -#define G4X_P1_SINGLE_CHANNEL_LVDS_MIN 2 -#define G4X_P1_SINGLE_CHANNEL_LVDS_MAX 8 -#define G4X_P2_SINGLE_CHANNEL_LVDS_SLOW 14 -#define G4X_P2_SINGLE_CHANNEL_LVDS_FAST 14 -#define G4X_P2_SINGLE_CHANNEL_LVDS_LIMIT 0 - -/*The parameter is for DUAL_CHANNEL_LVDS on G4x platform*/ -#define G4X_DOT_DUAL_CHANNEL_LVDS_MIN 80000 -#define G4X_DOT_DUAL_CHANNEL_LVDS_MAX 224000 -#define G4X_N_DUAL_CHANNEL_LVDS_MIN 1 -#define G4X_N_DUAL_CHANNEL_LVDS_MAX 3 -#define G4X_M_DUAL_CHANNEL_LVDS_MIN 104 -#define G4X_M_DUAL_CHANNEL_LVDS_MAX 138 -#define G4X_M1_DUAL_CHANNEL_LVDS_MIN 17 -#define G4X_M1_DUAL_CHANNEL_LVDS_MAX 23 -#define G4X_M2_DUAL_CHANNEL_LVDS_MIN 5 -#define G4X_M2_DUAL_CHANNEL_LVDS_MAX 11 -#define G4X_P_DUAL_CHANNEL_LVDS_MIN 14 -#define G4X_P_DUAL_CHANNEL_LVDS_MAX 42 -#define G4X_P1_DUAL_CHANNEL_LVDS_MIN 2 -#define G4X_P1_DUAL_CHANNEL_LVDS_MAX 6 -#define G4X_P2_DUAL_CHANNEL_LVDS_SLOW 7 -#define G4X_P2_DUAL_CHANNEL_LVDS_FAST 7 -#define G4X_P2_DUAL_CHANNEL_LVDS_LIMIT 0 - -/*The parameter is for DISPLAY PORT on G4x platform*/ -#define G4X_DOT_DISPLAY_PORT_MIN 161670 -#define G4X_DOT_DISPLAY_PORT_MAX 227000 -#define G4X_N_DISPLAY_PORT_MIN 1 -#define G4X_N_DISPLAY_PORT_MAX 2 -#define G4X_M_DISPLAY_PORT_MIN 97 -#define G4X_M_DISPLAY_PORT_MAX 108 -#define G4X_M1_DISPLAY_PORT_MIN 0x10 -#define G4X_M1_DISPLAY_PORT_MAX 0x12 -#define G4X_M2_DISPLAY_PORT_MIN 0x05 -#define G4X_M2_DISPLAY_PORT_MAX 0x06 -#define G4X_P_DISPLAY_PORT_MIN 10 -#define G4X_P_DISPLAY_PORT_MAX 20 -#define G4X_P1_DISPLAY_PORT_MIN 1 -#define G4X_P1_DISPLAY_PORT_MAX 2 -#define G4X_P2_DISPLAY_PORT_SLOW 10 -#define G4X_P2_DISPLAY_PORT_FAST 10 -#define G4X_P2_DISPLAY_PORT_LIMIT 0 - -/* Ironlake / Sandybridge */ -/* as we calculate clock using (register_value + 2) for - N/M1/M2, so here the range value for them is (actual_value-2). - */ -#define IRONLAKE_DOT_MIN 25000 -#define IRONLAKE_DOT_MAX 350000 -#define IRONLAKE_VCO_MIN 1760000 -#define IRONLAKE_VCO_MAX 3510000 -#define IRONLAKE_M1_MIN 12 -#define IRONLAKE_M1_MAX 22 -#define IRONLAKE_M2_MIN 5 -#define IRONLAKE_M2_MAX 9 -#define IRONLAKE_P2_DOT_LIMIT 225000 /* 225Mhz */ - -/* We have parameter ranges for different type of outputs. */ - -/* DAC & HDMI Refclk 120Mhz */ -#define IRONLAKE_DAC_N_MIN 1 -#define IRONLAKE_DAC_N_MAX 5 -#define IRONLAKE_DAC_M_MIN 79 -#define IRONLAKE_DAC_M_MAX 127 -#define IRONLAKE_DAC_P_MIN 5 -#define IRONLAKE_DAC_P_MAX 80 -#define IRONLAKE_DAC_P1_MIN 1 -#define IRONLAKE_DAC_P1_MAX 8 -#define IRONLAKE_DAC_P2_SLOW 10 -#define IRONLAKE_DAC_P2_FAST 5 - -/* LVDS single-channel 120Mhz refclk */ -#define IRONLAKE_LVDS_S_N_MIN 1 -#define IRONLAKE_LVDS_S_N_MAX 3 -#define IRONLAKE_LVDS_S_M_MIN 79 -#define IRONLAKE_LVDS_S_M_MAX 118 -#define IRONLAKE_LVDS_S_P_MIN 28 -#define IRONLAKE_LVDS_S_P_MAX 112 -#define IRONLAKE_LVDS_S_P1_MIN 2 -#define IRONLAKE_LVDS_S_P1_MAX 8 -#define IRONLAKE_LVDS_S_P2_SLOW 14 -#define IRONLAKE_LVDS_S_P2_FAST 14 - -/* LVDS dual-channel 120Mhz refclk */ -#define IRONLAKE_LVDS_D_N_MIN 1 -#define IRONLAKE_LVDS_D_N_MAX 3 -#define IRONLAKE_LVDS_D_M_MIN 79 -#define IRONLAKE_LVDS_D_M_MAX 127 -#define IRONLAKE_LVDS_D_P_MIN 14 -#define IRONLAKE_LVDS_D_P_MAX 56 -#define IRONLAKE_LVDS_D_P1_MIN 2 -#define IRONLAKE_LVDS_D_P1_MAX 8 -#define IRONLAKE_LVDS_D_P2_SLOW 7 -#define IRONLAKE_LVDS_D_P2_FAST 7 - -/* LVDS single-channel 100Mhz refclk */ -#define IRONLAKE_LVDS_S_SSC_N_MIN 1 -#define IRONLAKE_LVDS_S_SSC_N_MAX 2 -#define IRONLAKE_LVDS_S_SSC_M_MIN 79 -#define IRONLAKE_LVDS_S_SSC_M_MAX 126 -#define IRONLAKE_LVDS_S_SSC_P_MIN 28 -#define IRONLAKE_LVDS_S_SSC_P_MAX 112 -#define IRONLAKE_LVDS_S_SSC_P1_MIN 2 -#define IRONLAKE_LVDS_S_SSC_P1_MAX 8 -#define IRONLAKE_LVDS_S_SSC_P2_SLOW 14 -#define IRONLAKE_LVDS_S_SSC_P2_FAST 14 - -/* LVDS dual-channel 100Mhz refclk */ -#define IRONLAKE_LVDS_D_SSC_N_MIN 1 -#define IRONLAKE_LVDS_D_SSC_N_MAX 3 -#define IRONLAKE_LVDS_D_SSC_M_MIN 79 -#define IRONLAKE_LVDS_D_SSC_M_MAX 126 -#define IRONLAKE_LVDS_D_SSC_P_MIN 14 -#define IRONLAKE_LVDS_D_SSC_P_MAX 42 -#define IRONLAKE_LVDS_D_SSC_P1_MIN 2 -#define IRONLAKE_LVDS_D_SSC_P1_MAX 6 -#define IRONLAKE_LVDS_D_SSC_P2_SLOW 7 -#define IRONLAKE_LVDS_D_SSC_P2_FAST 7 - -/* DisplayPort */ -#define IRONLAKE_DP_N_MIN 1 -#define IRONLAKE_DP_N_MAX 2 -#define IRONLAKE_DP_M_MIN 81 -#define IRONLAKE_DP_M_MAX 90 -#define IRONLAKE_DP_P_MIN 10 -#define IRONLAKE_DP_P_MAX 20 -#define IRONLAKE_DP_P2_FAST 10 -#define IRONLAKE_DP_P2_SLOW 10 -#define IRONLAKE_DP_P2_LIMIT 0 -#define IRONLAKE_DP_P1_MIN 1 -#define IRONLAKE_DP_P1_MAX 2 - /* FDI */ #define IRONLAKE_FDI_FREQ 2700000 /* in kHz for mode->clock */ @@ -353,292 +105,253 @@ intel_fdi_link_freq(struct drm_device *dev) } static const intel_limit_t intel_limits_i8xx_dvo = { - .dot = { .min = I8XX_DOT_MIN, .max = I8XX_DOT_MAX }, - .vco = { .min = I8XX_VCO_MIN, .max = I8XX_VCO_MAX }, - .n = { .min = I8XX_N_MIN, .max = I8XX_N_MAX }, - .m = { .min = I8XX_M_MIN, .max = I8XX_M_MAX }, - .m1 = { .min = I8XX_M1_MIN, .max = I8XX_M1_MAX }, - .m2 = { .min = I8XX_M2_MIN, .max = I8XX_M2_MAX }, - .p = { .min = I8XX_P_MIN, .max = I8XX_P_MAX }, - .p1 = { .min = I8XX_P1_MIN, .max = I8XX_P1_MAX }, - .p2 = { .dot_limit = I8XX_P2_SLOW_LIMIT, - .p2_slow = I8XX_P2_SLOW, .p2_fast = I8XX_P2_FAST }, + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 930000, .max = 1400000 }, + .n = { .min = 3, .max = 16 }, + .m = { .min = 96, .max = 140 }, + .m1 = { .min = 18, .max = 26 }, + .m2 = { .min = 6, .max = 16 }, + .p = { .min = 4, .max = 128 }, + .p1 = { .min = 2, .max = 33 }, + .p2 = { .dot_limit = 165000, + .p2_slow = 4, .p2_fast = 2 }, .find_pll = intel_find_best_PLL, }; static const intel_limit_t intel_limits_i8xx_lvds = { - .dot = { .min = I8XX_DOT_MIN, .max = I8XX_DOT_MAX }, - .vco = { .min = I8XX_VCO_MIN, .max = I8XX_VCO_MAX }, - .n = { .min = I8XX_N_MIN, .max = I8XX_N_MAX }, - .m = { .min = I8XX_M_MIN, .max = I8XX_M_MAX }, - .m1 = { .min = I8XX_M1_MIN, .max = I8XX_M1_MAX }, - .m2 = { .min = I8XX_M2_MIN, .max = I8XX_M2_MAX }, - .p = { .min = I8XX_P_MIN, .max = I8XX_P_MAX }, - .p1 = { .min = I8XX_P1_LVDS_MIN, .max = I8XX_P1_LVDS_MAX }, - .p2 = { .dot_limit = I8XX_P2_SLOW_LIMIT, - .p2_slow = I8XX_P2_LVDS_SLOW, .p2_fast = I8XX_P2_LVDS_FAST }, + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 930000, .max = 1400000 }, + .n = { .min = 3, .max = 16 }, + .m = { .min = 96, .max = 140 }, + .m1 = { .min = 18, .max = 26 }, + .m2 = { .min = 6, .max = 16 }, + .p = { .min = 4, .max = 128 }, + .p1 = { .min = 1, .max = 6 }, + .p2 = { .dot_limit = 165000, + .p2_slow = 14, .p2_fast = 7 }, .find_pll = intel_find_best_PLL, }; - + static const intel_limit_t intel_limits_i9xx_sdvo = { - .dot = { .min = I9XX_DOT_MIN, .max = I9XX_DOT_MAX }, - .vco = { .min = I9XX_VCO_MIN, .max = I9XX_VCO_MAX }, - .n = { .min = I9XX_N_MIN, .max = I9XX_N_MAX }, - .m = { .min = I9XX_M_MIN, .max = I9XX_M_MAX }, - .m1 = { .min = I9XX_M1_MIN, .max = I9XX_M1_MAX }, - .m2 = { .min = I9XX_M2_MIN, .max = I9XX_M2_MAX }, - .p = { .min = I9XX_P_SDVO_DAC_MIN, .max = I9XX_P_SDVO_DAC_MAX }, - .p1 = { .min = I9XX_P1_MIN, .max = I9XX_P1_MAX }, - .p2 = { .dot_limit = I9XX_P2_SDVO_DAC_SLOW_LIMIT, - .p2_slow = I9XX_P2_SDVO_DAC_SLOW, .p2_fast = I9XX_P2_SDVO_DAC_FAST }, + .dot = { .min = 20000, .max = 400000 }, + .vco = { .min = 1400000, .max = 2800000 }, + .n = { .min = 1, .max = 6 }, + .m = { .min = 70, .max = 120 }, + .m1 = { .min = 10, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 5, .max = 80 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 200000, + .p2_slow = 10, .p2_fast = 5 }, .find_pll = intel_find_best_PLL, }; static const intel_limit_t intel_limits_i9xx_lvds = { - .dot = { .min = I9XX_DOT_MIN, .max = I9XX_DOT_MAX }, - .vco = { .min = I9XX_VCO_MIN, .max = I9XX_VCO_MAX }, - .n = { .min = I9XX_N_MIN, .max = I9XX_N_MAX }, - .m = { .min = I9XX_M_MIN, .max = I9XX_M_MAX }, - .m1 = { .min = I9XX_M1_MIN, .max = I9XX_M1_MAX }, - .m2 = { .min = I9XX_M2_MIN, .max = I9XX_M2_MAX }, - .p = { .min = I9XX_P_LVDS_MIN, .max = I9XX_P_LVDS_MAX }, - .p1 = { .min = I9XX_P1_MIN, .max = I9XX_P1_MAX }, - /* The single-channel range is 25-112Mhz, and dual-channel - * is 80-224Mhz. Prefer single channel as much as possible. - */ - .p2 = { .dot_limit = I9XX_P2_LVDS_SLOW_LIMIT, - .p2_slow = I9XX_P2_LVDS_SLOW, .p2_fast = I9XX_P2_LVDS_FAST }, + .dot = { .min = 20000, .max = 400000 }, + .vco = { .min = 1400000, .max = 2800000 }, + .n = { .min = 1, .max = 6 }, + .m = { .min = 70, .max = 120 }, + .m1 = { .min = 10, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 7, .max = 98 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 112000, + .p2_slow = 14, .p2_fast = 7 }, .find_pll = intel_find_best_PLL, }; - /* below parameter and function is for G4X Chipset Family*/ + static const intel_limit_t intel_limits_g4x_sdvo = { - .dot = { .min = G4X_DOT_SDVO_MIN, .max = G4X_DOT_SDVO_MAX }, - .vco = { .min = G4X_VCO_MIN, .max = G4X_VCO_MAX}, - .n = { .min = G4X_N_SDVO_MIN, .max = G4X_N_SDVO_MAX }, - .m = { .min = G4X_M_SDVO_MIN, .max = G4X_M_SDVO_MAX }, - .m1 = { .min = G4X_M1_SDVO_MIN, .max = G4X_M1_SDVO_MAX }, - .m2 = { .min = G4X_M2_SDVO_MIN, .max = G4X_M2_SDVO_MAX }, - .p = { .min = G4X_P_SDVO_MIN, .max = G4X_P_SDVO_MAX }, - .p1 = { .min = G4X_P1_SDVO_MIN, .max = G4X_P1_SDVO_MAX}, - .p2 = { .dot_limit = G4X_P2_SDVO_LIMIT, - .p2_slow = G4X_P2_SDVO_SLOW, - .p2_fast = G4X_P2_SDVO_FAST + .dot = { .min = 25000, .max = 270000 }, + .vco = { .min = 1750000, .max = 3500000}, + .n = { .min = 1, .max = 4 }, + .m = { .min = 104, .max = 138 }, + .m1 = { .min = 17, .max = 23 }, + .m2 = { .min = 5, .max = 11 }, + .p = { .min = 10, .max = 30 }, + .p1 = { .min = 1, .max = 3}, + .p2 = { .dot_limit = 270000, + .p2_slow = 10, + .p2_fast = 10 }, .find_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_g4x_hdmi = { - .dot = { .min = G4X_DOT_HDMI_DAC_MIN, .max = G4X_DOT_HDMI_DAC_MAX }, - .vco = { .min = G4X_VCO_MIN, .max = G4X_VCO_MAX}, - .n = { .min = G4X_N_HDMI_DAC_MIN, .max = G4X_N_HDMI_DAC_MAX }, - .m = { .min = G4X_M_HDMI_DAC_MIN, .max = G4X_M_HDMI_DAC_MAX }, - .m1 = { .min = G4X_M1_HDMI_DAC_MIN, .max = G4X_M1_HDMI_DAC_MAX }, - .m2 = { .min = G4X_M2_HDMI_DAC_MIN, .max = G4X_M2_HDMI_DAC_MAX }, - .p = { .min = G4X_P_HDMI_DAC_MIN, .max = G4X_P_HDMI_DAC_MAX }, - .p1 = { .min = G4X_P1_HDMI_DAC_MIN, .max = G4X_P1_HDMI_DAC_MAX}, - .p2 = { .dot_limit = G4X_P2_HDMI_DAC_LIMIT, - .p2_slow = G4X_P2_HDMI_DAC_SLOW, - .p2_fast = G4X_P2_HDMI_DAC_FAST - }, + .dot = { .min = 22000, .max = 400000 }, + .vco = { .min = 1750000, .max = 3500000}, + .n = { .min = 1, .max = 4 }, + .m = { .min = 104, .max = 138 }, + .m1 = { .min = 16, .max = 23 }, + .m2 = { .min = 5, .max = 11 }, + .p = { .min = 5, .max = 80 }, + .p1 = { .min = 1, .max = 8}, + .p2 = { .dot_limit = 165000, + .p2_slow = 10, .p2_fast = 5 }, .find_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_g4x_single_channel_lvds = { - .dot = { .min = G4X_DOT_SINGLE_CHANNEL_LVDS_MIN, - .max = G4X_DOT_SINGLE_CHANNEL_LVDS_MAX }, - .vco = { .min = G4X_VCO_MIN, - .max = G4X_VCO_MAX }, - .n = { .min = G4X_N_SINGLE_CHANNEL_LVDS_MIN, - .max = G4X_N_SINGLE_CHANNEL_LVDS_MAX }, - .m = { .min = G4X_M_SINGLE_CHANNEL_LVDS_MIN, - .max = G4X_M_SINGLE_CHANNEL_LVDS_MAX }, - .m1 = { .min = G4X_M1_SINGLE_CHANNEL_LVDS_MIN, - .max = G4X_M1_SINGLE_CHANNEL_LVDS_MAX }, - .m2 = { .min = G4X_M2_SINGLE_CHANNEL_LVDS_MIN, - .max = G4X_M2_SINGLE_CHANNEL_LVDS_MAX }, - .p = { .min = G4X_P_SINGLE_CHANNEL_LVDS_MIN, - .max = G4X_P_SINGLE_CHANNEL_LVDS_MAX }, - .p1 = { .min = G4X_P1_SINGLE_CHANNEL_LVDS_MIN, - .max = G4X_P1_SINGLE_CHANNEL_LVDS_MAX }, - .p2 = { .dot_limit = G4X_P2_SINGLE_CHANNEL_LVDS_LIMIT, - .p2_slow = G4X_P2_SINGLE_CHANNEL_LVDS_SLOW, - .p2_fast = G4X_P2_SINGLE_CHANNEL_LVDS_FAST + .dot = { .min = 20000, .max = 115000 }, + .vco = { .min = 1750000, .max = 3500000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 104, .max = 138 }, + .m1 = { .min = 17, .max = 23 }, + .m2 = { .min = 5, .max = 11 }, + .p = { .min = 28, .max = 112 }, + .p1 = { .min = 2, .max = 8 }, + .p2 = { .dot_limit = 0, + .p2_slow = 14, .p2_fast = 14 }, .find_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_g4x_dual_channel_lvds = { - .dot = { .min = G4X_DOT_DUAL_CHANNEL_LVDS_MIN, - .max = G4X_DOT_DUAL_CHANNEL_LVDS_MAX }, - .vco = { .min = G4X_VCO_MIN, - .max = G4X_VCO_MAX }, - .n = { .min = G4X_N_DUAL_CHANNEL_LVDS_MIN, - .max = G4X_N_DUAL_CHANNEL_LVDS_MAX }, - .m = { .min = G4X_M_DUAL_CHANNEL_LVDS_MIN, - .max = G4X_M_DUAL_CHANNEL_LVDS_MAX }, - .m1 = { .min = G4X_M1_DUAL_CHANNEL_LVDS_MIN, - .max = G4X_M1_DUAL_CHANNEL_LVDS_MAX }, - .m2 = { .min = G4X_M2_DUAL_CHANNEL_LVDS_MIN, - .max = G4X_M2_DUAL_CHANNEL_LVDS_MAX }, - .p = { .min = G4X_P_DUAL_CHANNEL_LVDS_MIN, - .max = G4X_P_DUAL_CHANNEL_LVDS_MAX }, - .p1 = { .min = G4X_P1_DUAL_CHANNEL_LVDS_MIN, - .max = G4X_P1_DUAL_CHANNEL_LVDS_MAX }, - .p2 = { .dot_limit = G4X_P2_DUAL_CHANNEL_LVDS_LIMIT, - .p2_slow = G4X_P2_DUAL_CHANNEL_LVDS_SLOW, - .p2_fast = G4X_P2_DUAL_CHANNEL_LVDS_FAST + .dot = { .min = 80000, .max = 224000 }, + .vco = { .min = 1750000, .max = 3500000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 104, .max = 138 }, + .m1 = { .min = 17, .max = 23 }, + .m2 = { .min = 5, .max = 11 }, + .p = { .min = 14, .max = 42 }, + .p1 = { .min = 2, .max = 6 }, + .p2 = { .dot_limit = 0, + .p2_slow = 7, .p2_fast = 7 }, .find_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_g4x_display_port = { - .dot = { .min = G4X_DOT_DISPLAY_PORT_MIN, - .max = G4X_DOT_DISPLAY_PORT_MAX }, - .vco = { .min = G4X_VCO_MIN, - .max = G4X_VCO_MAX}, - .n = { .min = G4X_N_DISPLAY_PORT_MIN, - .max = G4X_N_DISPLAY_PORT_MAX }, - .m = { .min = G4X_M_DISPLAY_PORT_MIN, - .max = G4X_M_DISPLAY_PORT_MAX }, - .m1 = { .min = G4X_M1_DISPLAY_PORT_MIN, - .max = G4X_M1_DISPLAY_PORT_MAX }, - .m2 = { .min = G4X_M2_DISPLAY_PORT_MIN, - .max = G4X_M2_DISPLAY_PORT_MAX }, - .p = { .min = G4X_P_DISPLAY_PORT_MIN, - .max = G4X_P_DISPLAY_PORT_MAX }, - .p1 = { .min = G4X_P1_DISPLAY_PORT_MIN, - .max = G4X_P1_DISPLAY_PORT_MAX}, - .p2 = { .dot_limit = G4X_P2_DISPLAY_PORT_LIMIT, - .p2_slow = G4X_P2_DISPLAY_PORT_SLOW, - .p2_fast = G4X_P2_DISPLAY_PORT_FAST }, + .dot = { .min = 161670, .max = 227000 }, + .vco = { .min = 1750000, .max = 3500000}, + .n = { .min = 1, .max = 2 }, + .m = { .min = 97, .max = 108 }, + .m1 = { .min = 0x10, .max = 0x12 }, + .m2 = { .min = 0x05, .max = 0x06 }, + .p = { .min = 10, .max = 20 }, + .p1 = { .min = 1, .max = 2}, + .p2 = { .dot_limit = 0, + .p2_slow = 10, .p2_fast = 10 }, .find_pll = intel_find_pll_g4x_dp, }; static const intel_limit_t intel_limits_pineview_sdvo = { - .dot = { .min = I9XX_DOT_MIN, .max = I9XX_DOT_MAX}, - .vco = { .min = PINEVIEW_VCO_MIN, .max = PINEVIEW_VCO_MAX }, - .n = { .min = PINEVIEW_N_MIN, .max = PINEVIEW_N_MAX }, - .m = { .min = PINEVIEW_M_MIN, .max = PINEVIEW_M_MAX }, - .m1 = { .min = PINEVIEW_M1_MIN, .max = PINEVIEW_M1_MAX }, - .m2 = { .min = PINEVIEW_M2_MIN, .max = PINEVIEW_M2_MAX }, - .p = { .min = I9XX_P_SDVO_DAC_MIN, .max = I9XX_P_SDVO_DAC_MAX }, - .p1 = { .min = I9XX_P1_MIN, .max = I9XX_P1_MAX }, - .p2 = { .dot_limit = I9XX_P2_SDVO_DAC_SLOW_LIMIT, - .p2_slow = I9XX_P2_SDVO_DAC_SLOW, .p2_fast = I9XX_P2_SDVO_DAC_FAST }, + .dot = { .min = 20000, .max = 400000}, + .vco = { .min = 1700000, .max = 3500000 }, + /* Pineview's Ncounter is a ring counter */ + .n = { .min = 3, .max = 6 }, + .m = { .min = 2, .max = 256 }, + /* Pineview only has one combined m divider, which we treat as m2. */ + .m1 = { .min = 0, .max = 0 }, + .m2 = { .min = 0, .max = 254 }, + .p = { .min = 5, .max = 80 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 200000, + .p2_slow = 10, .p2_fast = 5 }, .find_pll = intel_find_best_PLL, }; static const intel_limit_t intel_limits_pineview_lvds = { - .dot = { .min = I9XX_DOT_MIN, .max = I9XX_DOT_MAX }, - .vco = { .min = PINEVIEW_VCO_MIN, .max = PINEVIEW_VCO_MAX }, - .n = { .min = PINEVIEW_N_MIN, .max = PINEVIEW_N_MAX }, - .m = { .min = PINEVIEW_M_MIN, .max = PINEVIEW_M_MAX }, - .m1 = { .min = PINEVIEW_M1_MIN, .max = PINEVIEW_M1_MAX }, - .m2 = { .min = PINEVIEW_M2_MIN, .max = PINEVIEW_M2_MAX }, - .p = { .min = PINEVIEW_P_LVDS_MIN, .max = PINEVIEW_P_LVDS_MAX }, - .p1 = { .min = I9XX_P1_MIN, .max = I9XX_P1_MAX }, - /* Pineview only supports single-channel mode. */ - .p2 = { .dot_limit = I9XX_P2_LVDS_SLOW_LIMIT, - .p2_slow = I9XX_P2_LVDS_SLOW, .p2_fast = I9XX_P2_LVDS_SLOW }, + .dot = { .min = 20000, .max = 400000 }, + .vco = { .min = 1700000, .max = 3500000 }, + .n = { .min = 3, .max = 6 }, + .m = { .min = 2, .max = 256 }, + .m1 = { .min = 0, .max = 0 }, + .m2 = { .min = 0, .max = 254 }, + .p = { .min = 7, .max = 112 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 112000, + .p2_slow = 14, .p2_fast = 14 }, .find_pll = intel_find_best_PLL, }; +/* Ironlake / Sandybridge + * + * We calculate clock using (register_value + 2) for N/M1/M2, so here + * the range value for them is (actual_value - 2). + */ static const intel_limit_t intel_limits_ironlake_dac = { - .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, - .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, - .n = { .min = IRONLAKE_DAC_N_MIN, .max = IRONLAKE_DAC_N_MAX }, - .m = { .min = IRONLAKE_DAC_M_MIN, .max = IRONLAKE_DAC_M_MAX }, - .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, - .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, - .p = { .min = IRONLAKE_DAC_P_MIN, .max = IRONLAKE_DAC_P_MAX }, - .p1 = { .min = IRONLAKE_DAC_P1_MIN, .max = IRONLAKE_DAC_P1_MAX }, - .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, - .p2_slow = IRONLAKE_DAC_P2_SLOW, - .p2_fast = IRONLAKE_DAC_P2_FAST }, + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 5 }, + .m = { .min = 79, .max = 127 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 5, .max = 80 }, + .p1 = { .min = 1, .max = 8 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 10, .p2_fast = 5 }, .find_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_ironlake_single_lvds = { - .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, - .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, - .n = { .min = IRONLAKE_LVDS_S_N_MIN, .max = IRONLAKE_LVDS_S_N_MAX }, - .m = { .min = IRONLAKE_LVDS_S_M_MIN, .max = IRONLAKE_LVDS_S_M_MAX }, - .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, - .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, - .p = { .min = IRONLAKE_LVDS_S_P_MIN, .max = IRONLAKE_LVDS_S_P_MAX }, - .p1 = { .min = IRONLAKE_LVDS_S_P1_MIN, .max = IRONLAKE_LVDS_S_P1_MAX }, - .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, - .p2_slow = IRONLAKE_LVDS_S_P2_SLOW, - .p2_fast = IRONLAKE_LVDS_S_P2_FAST }, + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 79, .max = 118 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 28, .max = 112 }, + .p1 = { .min = 2, .max = 8 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 14, .p2_fast = 14 }, .find_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_ironlake_dual_lvds = { - .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, - .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, - .n = { .min = IRONLAKE_LVDS_D_N_MIN, .max = IRONLAKE_LVDS_D_N_MAX }, - .m = { .min = IRONLAKE_LVDS_D_M_MIN, .max = IRONLAKE_LVDS_D_M_MAX }, - .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, - .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, - .p = { .min = IRONLAKE_LVDS_D_P_MIN, .max = IRONLAKE_LVDS_D_P_MAX }, - .p1 = { .min = IRONLAKE_LVDS_D_P1_MIN, .max = IRONLAKE_LVDS_D_P1_MAX }, - .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, - .p2_slow = IRONLAKE_LVDS_D_P2_SLOW, - .p2_fast = IRONLAKE_LVDS_D_P2_FAST }, + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 79, .max = 127 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 14, .max = 56 }, + .p1 = { .min = 2, .max = 8 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 7, .p2_fast = 7 }, .find_pll = intel_g4x_find_best_PLL, }; +/* LVDS 100mhz refclk limits. */ static const intel_limit_t intel_limits_ironlake_single_lvds_100m = { - .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, - .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, - .n = { .min = IRONLAKE_LVDS_S_SSC_N_MIN, .max = IRONLAKE_LVDS_S_SSC_N_MAX }, - .m = { .min = IRONLAKE_LVDS_S_SSC_M_MIN, .max = IRONLAKE_LVDS_S_SSC_M_MAX }, - .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, - .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, - .p = { .min = IRONLAKE_LVDS_S_SSC_P_MIN, .max = IRONLAKE_LVDS_S_SSC_P_MAX }, - .p1 = { .min = IRONLAKE_LVDS_S_SSC_P1_MIN,.max = IRONLAKE_LVDS_S_SSC_P1_MAX }, - .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, - .p2_slow = IRONLAKE_LVDS_S_SSC_P2_SLOW, - .p2_fast = IRONLAKE_LVDS_S_SSC_P2_FAST }, + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 2 }, + .m = { .min = 79, .max = 126 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 28, .max = 112 }, + .p1 = { .min = 2,.max = 8 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 14, .p2_fast = 14 }, .find_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_ironlake_dual_lvds_100m = { - .dot = { .min = IRONLAKE_DOT_MIN, .max = IRONLAKE_DOT_MAX }, - .vco = { .min = IRONLAKE_VCO_MIN, .max = IRONLAKE_VCO_MAX }, - .n = { .min = IRONLAKE_LVDS_D_SSC_N_MIN, .max = IRONLAKE_LVDS_D_SSC_N_MAX }, - .m = { .min = IRONLAKE_LVDS_D_SSC_M_MIN, .max = IRONLAKE_LVDS_D_SSC_M_MAX }, - .m1 = { .min = IRONLAKE_M1_MIN, .max = IRONLAKE_M1_MAX }, - .m2 = { .min = IRONLAKE_M2_MIN, .max = IRONLAKE_M2_MAX }, - .p = { .min = IRONLAKE_LVDS_D_SSC_P_MIN, .max = IRONLAKE_LVDS_D_SSC_P_MAX }, - .p1 = { .min = IRONLAKE_LVDS_D_SSC_P1_MIN,.max = IRONLAKE_LVDS_D_SSC_P1_MAX }, - .p2 = { .dot_limit = IRONLAKE_P2_DOT_LIMIT, - .p2_slow = IRONLAKE_LVDS_D_SSC_P2_SLOW, - .p2_fast = IRONLAKE_LVDS_D_SSC_P2_FAST }, + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000 }, + .n = { .min = 1, .max = 3 }, + .m = { .min = 79, .max = 126 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 14, .max = 42 }, + .p1 = { .min = 2,.max = 6 }, + .p2 = { .dot_limit = 225000, + .p2_slow = 7, .p2_fast = 7 }, .find_pll = intel_g4x_find_best_PLL, }; static const intel_limit_t intel_limits_ironlake_display_port = { - .dot = { .min = IRONLAKE_DOT_MIN, - .max = IRONLAKE_DOT_MAX }, - .vco = { .min = IRONLAKE_VCO_MIN, - .max = IRONLAKE_VCO_MAX}, - .n = { .min = IRONLAKE_DP_N_MIN, - .max = IRONLAKE_DP_N_MAX }, - .m = { .min = IRONLAKE_DP_M_MIN, - .max = IRONLAKE_DP_M_MAX }, - .m1 = { .min = IRONLAKE_M1_MIN, - .max = IRONLAKE_M1_MAX }, - .m2 = { .min = IRONLAKE_M2_MIN, - .max = IRONLAKE_M2_MAX }, - .p = { .min = IRONLAKE_DP_P_MIN, - .max = IRONLAKE_DP_P_MAX }, - .p1 = { .min = IRONLAKE_DP_P1_MIN, - .max = IRONLAKE_DP_P1_MAX}, - .p2 = { .dot_limit = IRONLAKE_DP_P2_LIMIT, - .p2_slow = IRONLAKE_DP_P2_SLOW, - .p2_fast = IRONLAKE_DP_P2_FAST }, + .dot = { .min = 25000, .max = 350000 }, + .vco = { .min = 1760000, .max = 3510000}, + .n = { .min = 1, .max = 2 }, + .m = { .min = 81, .max = 90 }, + .m1 = { .min = 12, .max = 22 }, + .m2 = { .min = 5, .max = 9 }, + .p = { .min = 10, .max = 20 }, + .p1 = { .min = 1, .max = 2}, + .p2 = { .dot_limit = 0, + .p2_slow = 10, .p2_fast = 10 }, .find_pll = intel_find_pll_ironlake_dp, }; @@ -1165,7 +878,7 @@ static void assert_panel_unlocked(struct drm_i915_private *dev_priv, int pp_reg, lvds_reg; u32 val; enum pipe panel_pipe = PIPE_A; - bool locked = locked; + bool locked = true; if (HAS_PCH_SPLIT(dev_priv->dev)) { pp_reg = PCH_PP_CONTROL; @@ -1267,11 +980,76 @@ static void assert_transcoder_disabled(struct drm_i915_private *dev_priv, pipe_name(pipe)); } +static bool dp_pipe_enabled(struct drm_i915_private *dev_priv, + enum pipe pipe, u32 port_sel, u32 val) +{ + if ((val & DP_PORT_EN) == 0) + return false; + + if (HAS_PCH_CPT(dev_priv->dev)) { + u32 trans_dp_ctl_reg = TRANS_DP_CTL(pipe); + u32 trans_dp_ctl = I915_READ(trans_dp_ctl_reg); + if ((trans_dp_ctl & TRANS_DP_PORT_SEL_MASK) != port_sel) + return false; + } else { + if ((val & DP_PIPE_MASK) != (pipe << 30)) + return false; + } + return true; +} + +static bool hdmi_pipe_enabled(struct drm_i915_private *dev_priv, + enum pipe pipe, u32 val) +{ + if ((val & PORT_ENABLE) == 0) + return false; + + if (HAS_PCH_CPT(dev_priv->dev)) { + if ((val & PORT_TRANS_SEL_MASK) != PORT_TRANS_SEL_CPT(pipe)) + return false; + } else { + if ((val & TRANSCODER_MASK) != TRANSCODER(pipe)) + return false; + } + return true; +} + +static bool lvds_pipe_enabled(struct drm_i915_private *dev_priv, + enum pipe pipe, u32 val) +{ + if ((val & LVDS_PORT_EN) == 0) + return false; + + if (HAS_PCH_CPT(dev_priv->dev)) { + if ((val & PORT_TRANS_SEL_MASK) != PORT_TRANS_SEL_CPT(pipe)) + return false; + } else { + if ((val & LVDS_PIPE_MASK) != LVDS_PIPE(pipe)) + return false; + } + return true; +} + +static bool adpa_pipe_enabled(struct drm_i915_private *dev_priv, + enum pipe pipe, u32 val) +{ + if ((val & ADPA_DAC_ENABLE) == 0) + return false; + if (HAS_PCH_CPT(dev_priv->dev)) { + if ((val & PORT_TRANS_SEL_MASK) != PORT_TRANS_SEL_CPT(pipe)) + return false; + } else { + if ((val & ADPA_PIPE_SELECT_MASK) != ADPA_PIPE_SELECT(pipe)) + return false; + } + return true; +} + static void assert_pch_dp_disabled(struct drm_i915_private *dev_priv, - enum pipe pipe, int reg) + enum pipe pipe, int reg, u32 port_sel) { u32 val = I915_READ(reg); - WARN(DP_PIPE_ENABLED(val, pipe), + WARN(dp_pipe_enabled(dev_priv, pipe, port_sel, val), "PCH DP (0x%08x) enabled on transcoder %c, should be disabled\n", reg, pipe_name(pipe)); } @@ -1280,7 +1058,7 @@ static void assert_pch_hdmi_disabled(struct drm_i915_private *dev_priv, enum pipe pipe, int reg) { u32 val = I915_READ(reg); - WARN(HDMI_PIPE_ENABLED(val, pipe), + WARN(hdmi_pipe_enabled(dev_priv, val, pipe), "PCH DP (0x%08x) enabled on transcoder %c, should be disabled\n", reg, pipe_name(pipe)); } @@ -1291,19 +1069,19 @@ static void assert_pch_ports_disabled(struct drm_i915_private *dev_priv, int reg; u32 val; - assert_pch_dp_disabled(dev_priv, pipe, PCH_DP_B); - assert_pch_dp_disabled(dev_priv, pipe, PCH_DP_C); - assert_pch_dp_disabled(dev_priv, pipe, PCH_DP_D); + assert_pch_dp_disabled(dev_priv, pipe, PCH_DP_B, TRANS_DP_PORT_SEL_B); + assert_pch_dp_disabled(dev_priv, pipe, PCH_DP_C, TRANS_DP_PORT_SEL_C); + assert_pch_dp_disabled(dev_priv, pipe, PCH_DP_D, TRANS_DP_PORT_SEL_D); reg = PCH_ADPA; val = I915_READ(reg); - WARN(ADPA_PIPE_ENABLED(val, pipe), + WARN(adpa_pipe_enabled(dev_priv, val, pipe), "PCH VGA enabled on transcoder %c, should be disabled\n", pipe_name(pipe)); reg = PCH_LVDS; val = I915_READ(reg); - WARN(LVDS_PIPE_ENABLED(val, pipe), + WARN(lvds_pipe_enabled(dev_priv, val, pipe), "PCH LVDS enabled on transcoder %c, should be disabled\n", pipe_name(pipe)); @@ -1445,12 +1223,15 @@ static void intel_enable_transcoder(struct drm_i915_private *dev_priv, reg = TRANSCONF(pipe); val = I915_READ(reg); - /* - * make the BPC in transcoder be consistent with - * that in pipeconf reg. - */ - val &= ~PIPE_BPC_MASK; - val |= I915_READ(PIPECONF(pipe)) & PIPE_BPC_MASK; + + if (HAS_PCH_IBX(dev_priv->dev)) { + /* + * make the BPC in transcoder be consistent with + * that in pipeconf reg. + */ + val &= ~PIPE_BPC_MASK; + val |= I915_READ(PIPECONF(pipe)) & PIPE_BPC_MASK; + } I915_WRITE(reg, val | TRANS_ENABLE); if (wait_for(I915_READ(reg) & TRANS_STATE_ENABLE, 100)) DRM_ERROR("failed to enable transcoder %d\n", pipe); @@ -1560,6 +1341,17 @@ static void intel_disable_pipe(struct drm_i915_private *dev_priv, intel_wait_for_pipe_off(dev_priv->dev, pipe); } +/* + * Plane regs are double buffered, going from enabled->disabled needs a + * trigger in order to latch. The display address reg provides this. + */ +static void intel_flush_display_plane(struct drm_i915_private *dev_priv, + enum plane plane) +{ + I915_WRITE(DSPADDR(plane), I915_READ(DSPADDR(plane))); + I915_WRITE(DSPSURF(plane), I915_READ(DSPSURF(plane))); +} + /** * intel_enable_plane - enable a display plane on a given pipe * @dev_priv: i915 private structure @@ -1583,20 +1375,10 @@ static void intel_enable_plane(struct drm_i915_private *dev_priv, return; I915_WRITE(reg, val | DISPLAY_PLANE_ENABLE); + intel_flush_display_plane(dev_priv, plane); intel_wait_for_vblank(dev_priv->dev, pipe); } -/* - * Plane regs are double buffered, going from enabled->disabled needs a - * trigger in order to latch. The display address reg provides this. - */ -static void intel_flush_display_plane(struct drm_i915_private *dev_priv, - enum plane plane) -{ - u32 reg = DSPADDR(plane); - I915_WRITE(reg, I915_READ(reg)); -} - /** * intel_disable_plane - disable a display plane * @dev_priv: i915 private structure @@ -1622,19 +1404,24 @@ static void intel_disable_plane(struct drm_i915_private *dev_priv, } static void disable_pch_dp(struct drm_i915_private *dev_priv, - enum pipe pipe, int reg) + enum pipe pipe, int reg, u32 port_sel) { u32 val = I915_READ(reg); - if (DP_PIPE_ENABLED(val, pipe)) + if (dp_pipe_enabled(dev_priv, pipe, port_sel, val)) { + DRM_DEBUG_KMS("Disabling pch dp %x on pipe %d\n", reg, pipe); I915_WRITE(reg, val & ~DP_PORT_EN); + } } static void disable_pch_hdmi(struct drm_i915_private *dev_priv, enum pipe pipe, int reg) { u32 val = I915_READ(reg); - if (HDMI_PIPE_ENABLED(val, pipe)) + if (hdmi_pipe_enabled(dev_priv, val, pipe)) { + DRM_DEBUG_KMS("Disabling pch HDMI %x on pipe %d\n", + reg, pipe); I915_WRITE(reg, val & ~PORT_ENABLE); + } } /* Disable any ports connected to this transcoder */ @@ -1646,18 +1433,19 @@ static void intel_disable_pch_ports(struct drm_i915_private *dev_priv, val = I915_READ(PCH_PP_CONTROL); I915_WRITE(PCH_PP_CONTROL, val | PANEL_UNLOCK_REGS); - disable_pch_dp(dev_priv, pipe, PCH_DP_B); - disable_pch_dp(dev_priv, pipe, PCH_DP_C); - disable_pch_dp(dev_priv, pipe, PCH_DP_D); + disable_pch_dp(dev_priv, pipe, PCH_DP_B, TRANS_DP_PORT_SEL_B); + disable_pch_dp(dev_priv, pipe, PCH_DP_C, TRANS_DP_PORT_SEL_C); + disable_pch_dp(dev_priv, pipe, PCH_DP_D, TRANS_DP_PORT_SEL_D); reg = PCH_ADPA; val = I915_READ(reg); - if (ADPA_PIPE_ENABLED(val, pipe)) + if (adpa_pipe_enabled(dev_priv, val, pipe)) I915_WRITE(reg, val & ~ADPA_DAC_ENABLE); reg = PCH_LVDS; val = I915_READ(reg); - if (LVDS_PIPE_ENABLED(val, pipe)) { + if (lvds_pipe_enabled(dev_priv, val, pipe)) { + DRM_DEBUG_KMS("disable lvds on pipe %d val 0x%08x\n", pipe, val); I915_WRITE(reg, val & ~LVDS_PORT_EN); POSTING_READ(reg); udelay(100); @@ -1668,6 +1456,28 @@ static void intel_disable_pch_ports(struct drm_i915_private *dev_priv, disable_pch_hdmi(dev_priv, pipe, HDMID); } +static void i8xx_disable_fbc(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 fbc_ctl; + + /* Disable compression */ + fbc_ctl = I915_READ(FBC_CONTROL); + if ((fbc_ctl & FBC_CTL_EN) == 0) + return; + + fbc_ctl &= ~FBC_CTL_EN; + I915_WRITE(FBC_CONTROL, fbc_ctl); + + /* Wait for compressing bit to clear */ + if (wait_for((I915_READ(FBC_STATUS) & FBC_STAT_COMPRESSING) == 0, 10)) { + DRM_DEBUG_KMS("FBC idle timed out\n"); + return; + } + + DRM_DEBUG_KMS("disabled FBC\n"); +} + static void i8xx_enable_fbc(struct drm_crtc *crtc, unsigned long interval) { struct drm_device *dev = crtc->dev; @@ -1676,36 +1486,25 @@ static void i8xx_enable_fbc(struct drm_crtc *crtc, unsigned long interval) struct intel_framebuffer *intel_fb = to_intel_framebuffer(fb); struct drm_i915_gem_object *obj = intel_fb->obj; struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int cfb_pitch; int plane, i; u32 fbc_ctl, fbc_ctl2; - if (fb->pitch == dev_priv->cfb_pitch && - obj->fence_reg == dev_priv->cfb_fence && - intel_crtc->plane == dev_priv->cfb_plane && - I915_READ(FBC_CONTROL) & FBC_CTL_EN) - return; - - i8xx_disable_fbc(dev); - - dev_priv->cfb_pitch = dev_priv->cfb_size / FBC_LL_SIZE; - - if (fb->pitch < dev_priv->cfb_pitch) - dev_priv->cfb_pitch = fb->pitch; + cfb_pitch = dev_priv->cfb_size / FBC_LL_SIZE; + if (fb->pitch < cfb_pitch) + cfb_pitch = fb->pitch; /* FBC_CTL wants 64B units */ - dev_priv->cfb_pitch = (dev_priv->cfb_pitch / 64) - 1; - dev_priv->cfb_fence = obj->fence_reg; - dev_priv->cfb_plane = intel_crtc->plane; - plane = dev_priv->cfb_plane == 0 ? FBC_CTL_PLANEA : FBC_CTL_PLANEB; + cfb_pitch = (cfb_pitch / 64) - 1; + plane = intel_crtc->plane == 0 ? FBC_CTL_PLANEA : FBC_CTL_PLANEB; /* Clear old tags */ for (i = 0; i < (FBC_LL_SIZE / 32) + 1; i++) I915_WRITE(FBC_TAG + (i * 4), 0); /* Set it up... */ - fbc_ctl2 = FBC_CTL_FENCE_DBL | FBC_CTL_IDLE_IMM | plane; - if (obj->tiling_mode != I915_TILING_NONE) - fbc_ctl2 |= FBC_CTL_CPU_FENCE; + fbc_ctl2 = FBC_CTL_FENCE_DBL | FBC_CTL_IDLE_IMM | FBC_CTL_CPU_FENCE; + fbc_ctl2 |= plane; I915_WRITE(FBC_CONTROL2, fbc_ctl2); I915_WRITE(FBC_FENCE_OFF, crtc->y); @@ -1713,36 +1512,13 @@ static void i8xx_enable_fbc(struct drm_crtc *crtc, unsigned long interval) fbc_ctl = FBC_CTL_EN | FBC_CTL_PERIODIC; if (IS_I945GM(dev)) fbc_ctl |= FBC_CTL_C3_IDLE; /* 945 needs special SR handling */ - fbc_ctl |= (dev_priv->cfb_pitch & 0xff) << FBC_CTL_STRIDE_SHIFT; + fbc_ctl |= (cfb_pitch & 0xff) << FBC_CTL_STRIDE_SHIFT; fbc_ctl |= (interval & 0x2fff) << FBC_CTL_INTERVAL_SHIFT; - if (obj->tiling_mode != I915_TILING_NONE) - fbc_ctl |= dev_priv->cfb_fence; - I915_WRITE(FBC_CONTROL, fbc_ctl); - - DRM_DEBUG_KMS("enabled FBC, pitch %ld, yoff %d, plane %d, ", - dev_priv->cfb_pitch, crtc->y, dev_priv->cfb_plane); -} - -void i8xx_disable_fbc(struct drm_device *dev) -{ - struct drm_i915_private *dev_priv = dev->dev_private; - u32 fbc_ctl; - - /* Disable compression */ - fbc_ctl = I915_READ(FBC_CONTROL); - if ((fbc_ctl & FBC_CTL_EN) == 0) - return; - - fbc_ctl &= ~FBC_CTL_EN; + fbc_ctl |= obj->fence_reg; I915_WRITE(FBC_CONTROL, fbc_ctl); - /* Wait for compressing bit to clear */ - if (wait_for((I915_READ(FBC_STATUS) & FBC_STAT_COMPRESSING) == 0, 10)) { - DRM_DEBUG_KMS("FBC idle timed out\n"); - return; - } - - DRM_DEBUG_KMS("disabled FBC\n"); + DRM_DEBUG_KMS("enabled FBC, pitch %d, yoff %d, plane %d, ", + cfb_pitch, crtc->y, intel_crtc->plane); } static bool i8xx_fbc_enabled(struct drm_device *dev) @@ -1764,30 +1540,9 @@ static void g4x_enable_fbc(struct drm_crtc *crtc, unsigned long interval) unsigned long stall_watermark = 200; u32 dpfc_ctl; - dpfc_ctl = I915_READ(DPFC_CONTROL); - if (dpfc_ctl & DPFC_CTL_EN) { - if (dev_priv->cfb_pitch == dev_priv->cfb_pitch / 64 - 1 && - dev_priv->cfb_fence == obj->fence_reg && - dev_priv->cfb_plane == intel_crtc->plane && - dev_priv->cfb_y == crtc->y) - return; - - I915_WRITE(DPFC_CONTROL, dpfc_ctl & ~DPFC_CTL_EN); - intel_wait_for_vblank(dev, intel_crtc->pipe); - } - - dev_priv->cfb_pitch = (dev_priv->cfb_pitch / 64) - 1; - dev_priv->cfb_fence = obj->fence_reg; - dev_priv->cfb_plane = intel_crtc->plane; - dev_priv->cfb_y = crtc->y; - dpfc_ctl = plane | DPFC_SR_EN | DPFC_CTL_LIMIT_1X; - if (obj->tiling_mode != I915_TILING_NONE) { - dpfc_ctl |= DPFC_CTL_FENCE_EN | dev_priv->cfb_fence; - I915_WRITE(DPFC_CHICKEN, DPFC_HT_MODIFY); - } else { - I915_WRITE(DPFC_CHICKEN, ~DPFC_HT_MODIFY); - } + dpfc_ctl |= DPFC_CTL_FENCE_EN | obj->fence_reg; + I915_WRITE(DPFC_CHICKEN, DPFC_HT_MODIFY); I915_WRITE(DPFC_RECOMP_CTL, DPFC_RECOMP_STALL_EN | (stall_watermark << DPFC_RECOMP_STALL_WM_SHIFT) | @@ -1800,7 +1555,7 @@ static void g4x_enable_fbc(struct drm_crtc *crtc, unsigned long interval) DRM_DEBUG_KMS("enabled fbc on plane %d\n", intel_crtc->plane); } -void g4x_disable_fbc(struct drm_device *dev) +static void g4x_disable_fbc(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; u32 dpfc_ctl; @@ -1828,7 +1583,7 @@ static void sandybridge_blit_fbc_update(struct drm_device *dev) u32 blt_ecoskpd; /* Make sure blitter notifies FBC of writes */ - __gen6_gt_force_wake_get(dev_priv); + gen6_gt_force_wake_get(dev_priv); blt_ecoskpd = I915_READ(GEN6_BLITTER_ECOSKPD); blt_ecoskpd |= GEN6_BLITTER_FBC_NOTIFY << GEN6_BLITTER_LOCK_SHIFT; @@ -1839,7 +1594,7 @@ static void sandybridge_blit_fbc_update(struct drm_device *dev) GEN6_BLITTER_LOCK_SHIFT); I915_WRITE(GEN6_BLITTER_ECOSKPD, blt_ecoskpd); POSTING_READ(GEN6_BLITTER_ECOSKPD); - __gen6_gt_force_wake_put(dev_priv); + gen6_gt_force_wake_put(dev_priv); } static void ironlake_enable_fbc(struct drm_crtc *crtc, unsigned long interval) @@ -1855,32 +1610,12 @@ static void ironlake_enable_fbc(struct drm_crtc *crtc, unsigned long interval) u32 dpfc_ctl; dpfc_ctl = I915_READ(ILK_DPFC_CONTROL); - if (dpfc_ctl & DPFC_CTL_EN) { - if (dev_priv->cfb_pitch == dev_priv->cfb_pitch / 64 - 1 && - dev_priv->cfb_fence == obj->fence_reg && - dev_priv->cfb_plane == intel_crtc->plane && - dev_priv->cfb_offset == obj->gtt_offset && - dev_priv->cfb_y == crtc->y) - return; - - I915_WRITE(ILK_DPFC_CONTROL, dpfc_ctl & ~DPFC_CTL_EN); - intel_wait_for_vblank(dev, intel_crtc->pipe); - } - - dev_priv->cfb_pitch = (dev_priv->cfb_pitch / 64) - 1; - dev_priv->cfb_fence = obj->fence_reg; - dev_priv->cfb_plane = intel_crtc->plane; - dev_priv->cfb_offset = obj->gtt_offset; - dev_priv->cfb_y = crtc->y; - dpfc_ctl &= DPFC_RESERVED; dpfc_ctl |= (plane | DPFC_CTL_LIMIT_1X); - if (obj->tiling_mode != I915_TILING_NONE) { - dpfc_ctl |= (DPFC_CTL_FENCE_EN | dev_priv->cfb_fence); - I915_WRITE(ILK_DPFC_CHICKEN, DPFC_HT_MODIFY); - } else { - I915_WRITE(ILK_DPFC_CHICKEN, ~DPFC_HT_MODIFY); - } + /* Set persistent mode for front-buffer rendering, ala X. */ + dpfc_ctl |= DPFC_CTL_PERSISTENT_MODE; + dpfc_ctl |= (DPFC_CTL_FENCE_EN | obj->fence_reg); + I915_WRITE(ILK_DPFC_CHICKEN, DPFC_HT_MODIFY); I915_WRITE(ILK_DPFC_RECOMP_CTL, DPFC_RECOMP_STALL_EN | (stall_watermark << DPFC_RECOMP_STALL_WM_SHIFT) | @@ -1892,7 +1627,7 @@ static void ironlake_enable_fbc(struct drm_crtc *crtc, unsigned long interval) if (IS_GEN6(dev)) { I915_WRITE(SNB_DPFC_CTL_SA, - SNB_CPU_FENCE_ENABLE | dev_priv->cfb_fence); + SNB_CPU_FENCE_ENABLE | obj->fence_reg); I915_WRITE(DPFC_CPU_FENCE_OFFSET, crtc->y); sandybridge_blit_fbc_update(dev); } @@ -1900,7 +1635,7 @@ static void ironlake_enable_fbc(struct drm_crtc *crtc, unsigned long interval) DRM_DEBUG_KMS("enabled fbc on plane %d\n", intel_crtc->plane); } -void ironlake_disable_fbc(struct drm_device *dev) +static void ironlake_disable_fbc(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; u32 dpfc_ctl; @@ -1932,24 +1667,109 @@ bool intel_fbc_enabled(struct drm_device *dev) return dev_priv->display.fbc_enabled(dev); } -void intel_enable_fbc(struct drm_crtc *crtc, unsigned long interval) +static void intel_fbc_work_fn(struct work_struct *__work) +{ + struct intel_fbc_work *work = + container_of(to_delayed_work(__work), + struct intel_fbc_work, work); + struct drm_device *dev = work->crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + + mutex_lock(&dev->struct_mutex); + if (work == dev_priv->fbc_work) { + /* Double check that we haven't switched fb without cancelling + * the prior work. + */ + if (work->crtc->fb == work->fb) { + dev_priv->display.enable_fbc(work->crtc, + work->interval); + + dev_priv->cfb_plane = to_intel_crtc(work->crtc)->plane; + dev_priv->cfb_fb = work->crtc->fb->base.id; + dev_priv->cfb_y = work->crtc->y; + } + + dev_priv->fbc_work = NULL; + } + mutex_unlock(&dev->struct_mutex); + + kfree(work); +} + +static void intel_cancel_fbc_work(struct drm_i915_private *dev_priv) { - struct drm_i915_private *dev_priv = crtc->dev->dev_private; + if (dev_priv->fbc_work == NULL) + return; + + DRM_DEBUG_KMS("cancelling pending FBC enable\n"); + + /* Synchronisation is provided by struct_mutex and checking of + * dev_priv->fbc_work, so we can perform the cancellation + * entirely asynchronously. + */ + if (cancel_delayed_work(&dev_priv->fbc_work->work)) + /* tasklet was killed before being run, clean up */ + kfree(dev_priv->fbc_work); + + /* Mark the work as no longer wanted so that if it does + * wake-up (because the work was already running and waiting + * for our mutex), it will discover that is no longer + * necessary to run. + */ + dev_priv->fbc_work = NULL; +} + +static void intel_enable_fbc(struct drm_crtc *crtc, unsigned long interval) +{ + struct intel_fbc_work *work; + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; if (!dev_priv->display.enable_fbc) return; - dev_priv->display.enable_fbc(crtc, interval); + intel_cancel_fbc_work(dev_priv); + + work = kzalloc(sizeof *work, GFP_KERNEL); + if (work == NULL) { + dev_priv->display.enable_fbc(crtc, interval); + return; + } + + work->crtc = crtc; + work->fb = crtc->fb; + work->interval = interval; + INIT_DELAYED_WORK(&work->work, intel_fbc_work_fn); + + dev_priv->fbc_work = work; + + DRM_DEBUG_KMS("scheduling delayed FBC enable\n"); + + /* Delay the actual enabling to let pageflipping cease and the + * display to settle before starting the compression. Note that + * this delay also serves a second purpose: it allows for a + * vblank to pass after disabling the FBC before we attempt + * to modify the control registers. + * + * A more complicated solution would involve tracking vblanks + * following the termination of the page-flipping sequence + * and indeed performing the enable as a co-routine and not + * waiting synchronously upon the vblank. + */ + schedule_delayed_work(&work->work, msecs_to_jiffies(50)); } void intel_disable_fbc(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; + intel_cancel_fbc_work(dev_priv); + if (!dev_priv->display.disable_fbc) return; dev_priv->display.disable_fbc(dev); + dev_priv->cfb_plane = -1; } /** @@ -2019,6 +1839,11 @@ static void intel_update_fbc(struct drm_device *dev) intel_fb = to_intel_framebuffer(fb); obj = intel_fb->obj; + if (!i915_enable_fbc) { + DRM_DEBUG_KMS("fbc disabled per module param (default off)\n"); + dev_priv->no_fbc_reason = FBC_MODULE_PARAM; + goto out_disable; + } if (intel_fb->obj->base.size > dev_priv->cfb_size) { DRM_DEBUG_KMS("framebuffer too large, disabling " "compression\n"); @@ -2043,8 +1868,13 @@ static void intel_update_fbc(struct drm_device *dev) dev_priv->no_fbc_reason = FBC_BAD_PLANE; goto out_disable; } - if (obj->tiling_mode != I915_TILING_X) { - DRM_DEBUG_KMS("framebuffer not tiled, disabling compression\n"); + + /* The use of a CPU fence is mandatory in order to detect writes + * by the CPU to the scanout and trigger updates to the FBC. + */ + if (obj->tiling_mode != I915_TILING_X || + obj->fence_reg == I915_FENCE_REG_NONE) { + DRM_DEBUG_KMS("framebuffer not tiled or fenced, disabling compression\n"); dev_priv->no_fbc_reason = FBC_NOT_TILED; goto out_disable; } @@ -2053,6 +1883,44 @@ static void intel_update_fbc(struct drm_device *dev) if (in_dbg_master()) goto out_disable; + /* If the scanout has not changed, don't modify the FBC settings. + * Note that we make the fundamental assumption that the fb->obj + * cannot be unpinned (and have its GTT offset and fence revoked) + * without first being decoupled from the scanout and FBC disabled. + */ + if (dev_priv->cfb_plane == intel_crtc->plane && + dev_priv->cfb_fb == fb->base.id && + dev_priv->cfb_y == crtc->y) + return; + + if (intel_fbc_enabled(dev)) { + /* We update FBC along two paths, after changing fb/crtc + * configuration (modeswitching) and after page-flipping + * finishes. For the latter, we know that not only did + * we disable the FBC at the start of the page-flip + * sequence, but also more than one vblank has passed. + * + * For the former case of modeswitching, it is possible + * to switch between two FBC valid configurations + * instantaneously so we do need to disable the FBC + * before we can modify its control registers. We also + * have to wait for the next vblank for that to take + * effect. However, since we delay enabling FBC we can + * assume that a vblank has passed since disabling and + * that we can safely alter the registers in the deferred + * callback. + * + * In the scenario that we go from a valid to invalid + * and then back to valid FBC configuration we have + * no strict enforcement that a vblank occurred since + * disabling the FBC. However, along all current pipe + * disabling paths we do need to wait for a vblank at + * some point. And we wait before enabling FBC anyway. + */ + DRM_DEBUG_KMS("disabling active FBC for update\n"); + intel_disable_fbc(dev); + } + intel_enable_fbc(crtc, 500); return; @@ -2095,14 +1963,10 @@ intel_pin_and_fence_fb_obj(struct drm_device *dev, } dev_priv->mm.interruptible = false; - ret = i915_gem_object_pin(obj, alignment, true); + ret = i915_gem_object_pin_to_display_plane(obj, alignment, pipelined); if (ret) goto err_interruptible; - ret = i915_gem_object_set_to_display_plane(obj, pipelined); - if (ret) - goto err_unpin; - /* Install a fence for tiled scan-out. Pre-i965 always needs a * fence, whereas 965+ only requires a fence if using * framebuffer compression. For simplicity, we always install @@ -2124,10 +1988,8 @@ err_interruptible: return ret; } -/* Assume fb object is pinned & idle & fenced and just update base pointers */ -static int -intel_pipe_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, - int x, int y, enum mode_set_atomic state) +static int i9xx_update_plane(struct drm_crtc *crtc, struct drm_framebuffer *fb, + int x, int y) { struct drm_device *dev = crtc->dev; struct drm_i915_private *dev_priv = dev->dev_private; @@ -2170,7 +2032,7 @@ intel_pipe_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, dspcntr |= DISPPLANE_32BPP_NO_ALPHA; break; default: - DRM_ERROR("Unknown color depth\n"); + DRM_ERROR("Unknown color depth %d\n", fb->bits_per_pixel); return -EINVAL; } if (INTEL_INFO(dev)->gen >= 4) { @@ -2180,10 +2042,6 @@ intel_pipe_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, dspcntr &= ~DISPPLANE_TILED; } - if (HAS_PCH_SPLIT(dev)) - /* must disable */ - dspcntr |= DISPPLANE_TRICKLE_FEED_DISABLE; - I915_WRITE(reg, dspcntr); Start = obj->gtt_offset; @@ -2200,6 +2058,99 @@ intel_pipe_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, I915_WRITE(DSPADDR(plane), Start + Offset); POSTING_READ(reg); + return 0; +} + +static int ironlake_update_plane(struct drm_crtc *crtc, + struct drm_framebuffer *fb, int x, int y) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct intel_framebuffer *intel_fb; + struct drm_i915_gem_object *obj; + int plane = intel_crtc->plane; + unsigned long Start, Offset; + u32 dspcntr; + u32 reg; + + switch (plane) { + case 0: + case 1: + break; + default: + DRM_ERROR("Can't update plane %d in SAREA\n", plane); + return -EINVAL; + } + + intel_fb = to_intel_framebuffer(fb); + obj = intel_fb->obj; + + reg = DSPCNTR(plane); + dspcntr = I915_READ(reg); + /* Mask out pixel format bits in case we change it */ + dspcntr &= ~DISPPLANE_PIXFORMAT_MASK; + switch (fb->bits_per_pixel) { + case 8: + dspcntr |= DISPPLANE_8BPP; + break; + case 16: + if (fb->depth != 16) + return -EINVAL; + + dspcntr |= DISPPLANE_16BPP; + break; + case 24: + case 32: + if (fb->depth == 24) + dspcntr |= DISPPLANE_32BPP_NO_ALPHA; + else if (fb->depth == 30) + dspcntr |= DISPPLANE_32BPP_30BIT_NO_ALPHA; + else + return -EINVAL; + break; + default: + DRM_ERROR("Unknown color depth %d\n", fb->bits_per_pixel); + return -EINVAL; + } + + if (obj->tiling_mode != I915_TILING_NONE) + dspcntr |= DISPPLANE_TILED; + else + dspcntr &= ~DISPPLANE_TILED; + + /* must disable */ + dspcntr |= DISPPLANE_TRICKLE_FEED_DISABLE; + + I915_WRITE(reg, dspcntr); + + Start = obj->gtt_offset; + Offset = y * fb->pitch + x * (fb->bits_per_pixel / 8); + + DRM_DEBUG_KMS("Writing base %08lX %08lX %d %d %d\n", + Start, Offset, x, y, fb->pitch); + I915_WRITE(DSPSTRIDE(plane), fb->pitch); + I915_WRITE(DSPSURF(plane), Start); + I915_WRITE(DSPTILEOFF(plane), (y << 16) | x); + I915_WRITE(DSPADDR(plane), Offset); + POSTING_READ(reg); + + return 0; +} + +/* Assume fb object is pinned & idle & fenced and just update base pointers */ +static int +intel_pipe_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, + int x, int y, enum mode_set_atomic state) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + int ret; + + ret = dev_priv->display.update_plane(crtc, fb, x, y); + if (ret) + return ret; + intel_update_fbc(dev); intel_increase_pllclock(crtc); @@ -2217,7 +2168,7 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, /* no fb bound */ if (!crtc->fb) { - DRM_DEBUG_KMS("No FB bound\n"); + DRM_ERROR("No FB bound\n"); return 0; } @@ -2226,6 +2177,7 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, case 1: break; default: + DRM_ERROR("no plane for crtc\n"); return -EINVAL; } @@ -2235,6 +2187,7 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, NULL); if (ret != 0) { mutex_unlock(&dev->struct_mutex); + DRM_ERROR("pin & fence failed\n"); return ret; } @@ -2254,7 +2207,7 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, * This should only fail upon a hung GPU, in which case we * can safely continue. */ - ret = i915_gem_object_flush_gpu(obj); + ret = i915_gem_object_finish_gpu(obj); (void) ret; } @@ -2263,6 +2216,7 @@ intel_pipe_set_base(struct drm_crtc *crtc, int x, int y, if (ret) { i915_gem_object_unpin(to_intel_framebuffer(crtc->fb)->obj); mutex_unlock(&dev->struct_mutex); + DRM_ERROR("failed to update base address\n"); return ret; } @@ -2339,8 +2293,13 @@ static void intel_fdi_normal_train(struct drm_crtc *crtc) /* enable normal train */ reg = FDI_TX_CTL(pipe); temp = I915_READ(reg); - temp &= ~FDI_LINK_TRAIN_NONE; - temp |= FDI_LINK_TRAIN_NONE | FDI_TX_ENHANCE_FRAME_ENABLE; + if (IS_IVYBRIDGE(dev)) { + temp &= ~FDI_LINK_TRAIN_NONE_IVB; + temp |= FDI_LINK_TRAIN_NONE_IVB | FDI_TX_ENHANCE_FRAME_ENABLE; + } else { + temp &= ~FDI_LINK_TRAIN_NONE; + temp |= FDI_LINK_TRAIN_NONE | FDI_TX_ENHANCE_FRAME_ENABLE; + } I915_WRITE(reg, temp); reg = FDI_RX_CTL(pipe); @@ -2357,6 +2316,23 @@ static void intel_fdi_normal_train(struct drm_crtc *crtc) /* wait one idle pattern time */ POSTING_READ(reg); udelay(1000); + + /* IVB wants error correction enabled */ + if (IS_IVYBRIDGE(dev)) + I915_WRITE(reg, I915_READ(reg) | FDI_FS_ERRC_ENABLE | + FDI_FE_ERRC_ENABLE); +} + +static void cpt_phase_pointer_enable(struct drm_device *dev, int pipe) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 flags = I915_READ(SOUTH_CHICKEN1); + + flags |= FDI_PHASE_SYNC_OVR(pipe); + I915_WRITE(SOUTH_CHICKEN1, flags); /* once to unlock... */ + flags |= FDI_PHASE_SYNC_EN(pipe); + I915_WRITE(SOUTH_CHICKEN1, flags); /* then again to enable */ + POSTING_READ(SOUTH_CHICKEN1); } /* The FDI link training functions for ILK/Ibexpeak. */ @@ -2509,6 +2485,9 @@ static void gen6_fdi_link_train(struct drm_crtc *crtc) POSTING_READ(reg); udelay(150); + if (HAS_PCH_CPT(dev)) + cpt_phase_pointer_enable(dev, pipe); + for (i = 0; i < 4; i++ ) { reg = FDI_TX_CTL(pipe); temp = I915_READ(reg); @@ -2584,7 +2563,119 @@ static void gen6_fdi_link_train(struct drm_crtc *crtc) DRM_DEBUG_KMS("FDI train done.\n"); } -static void ironlake_fdi_enable(struct drm_crtc *crtc) +/* Manual link training for Ivy Bridge A0 parts */ +static void ivb_manual_fdi_link_train(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + u32 reg, temp, i; + + /* Train 1: umask FDI RX Interrupt symbol_lock and bit_lock bit + for train result */ + reg = FDI_RX_IMR(pipe); + temp = I915_READ(reg); + temp &= ~FDI_RX_SYMBOL_LOCK; + temp &= ~FDI_RX_BIT_LOCK; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(150); + + /* enable CPU FDI TX and PCH FDI RX */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~(7 << 19); + temp |= (intel_crtc->fdi_lanes - 1) << 19; + temp &= ~(FDI_LINK_TRAIN_AUTO | FDI_LINK_TRAIN_NONE_IVB); + temp |= FDI_LINK_TRAIN_PATTERN_1_IVB; + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= FDI_LINK_TRAIN_400MV_0DB_SNB_B; + I915_WRITE(reg, temp | FDI_TX_ENABLE); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_AUTO; + temp &= ~FDI_LINK_TRAIN_PATTERN_MASK_CPT; + temp |= FDI_LINK_TRAIN_PATTERN_1_CPT; + I915_WRITE(reg, temp | FDI_RX_ENABLE); + + POSTING_READ(reg); + udelay(150); + + if (HAS_PCH_CPT(dev)) + cpt_phase_pointer_enable(dev, pipe); + + for (i = 0; i < 4; i++ ) { + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= snb_b_fdi_train_param[i]; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(500); + + reg = FDI_RX_IIR(pipe); + temp = I915_READ(reg); + DRM_DEBUG_KMS("FDI_RX_IIR 0x%x\n", temp); + + if (temp & FDI_RX_BIT_LOCK || + (I915_READ(reg) & FDI_RX_BIT_LOCK)) { + I915_WRITE(reg, temp | FDI_RX_BIT_LOCK); + DRM_DEBUG_KMS("FDI train 1 done.\n"); + break; + } + } + if (i == 4) + DRM_ERROR("FDI train 1 fail!\n"); + + /* Train 2 */ + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_NONE_IVB; + temp |= FDI_LINK_TRAIN_PATTERN_2_IVB; + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= FDI_LINK_TRAIN_400MV_0DB_SNB_B; + I915_WRITE(reg, temp); + + reg = FDI_RX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_PATTERN_MASK_CPT; + temp |= FDI_LINK_TRAIN_PATTERN_2_CPT; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(150); + + for (i = 0; i < 4; i++ ) { + reg = FDI_TX_CTL(pipe); + temp = I915_READ(reg); + temp &= ~FDI_LINK_TRAIN_VOL_EMP_MASK; + temp |= snb_b_fdi_train_param[i]; + I915_WRITE(reg, temp); + + POSTING_READ(reg); + udelay(500); + + reg = FDI_RX_IIR(pipe); + temp = I915_READ(reg); + DRM_DEBUG_KMS("FDI_RX_IIR 0x%x\n", temp); + + if (temp & FDI_RX_SYMBOL_LOCK) { + I915_WRITE(reg, temp | FDI_RX_SYMBOL_LOCK); + DRM_DEBUG_KMS("FDI train 2 done.\n"); + break; + } + } + if (i == 4) + DRM_ERROR("FDI train 2 fail!\n"); + + DRM_DEBUG_KMS("FDI train done.\n"); +} + +static void ironlake_fdi_pll_enable(struct drm_crtc *crtc) { struct drm_device *dev = crtc->dev; struct drm_i915_private *dev_priv = dev->dev_private; @@ -2625,6 +2716,17 @@ static void ironlake_fdi_enable(struct drm_crtc *crtc) } } +static void cpt_phase_pointer_disable(struct drm_device *dev, int pipe) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 flags = I915_READ(SOUTH_CHICKEN1); + + flags &= ~(FDI_PHASE_SYNC_EN(pipe)); + I915_WRITE(SOUTH_CHICKEN1, flags); /* once to disable... */ + flags &= ~(FDI_PHASE_SYNC_OVR(pipe)); + I915_WRITE(SOUTH_CHICKEN1, flags); /* then again to lock */ + POSTING_READ(SOUTH_CHICKEN1); +} static void ironlake_fdi_disable(struct drm_crtc *crtc) { struct drm_device *dev = crtc->dev; @@ -2654,6 +2756,8 @@ static void ironlake_fdi_disable(struct drm_crtc *crtc) I915_WRITE(FDI_RX_CHICKEN(pipe), I915_READ(FDI_RX_CHICKEN(pipe) & ~FDI_RX_PHASE_SYNC_POINTER_EN)); + } else if (HAS_PCH_CPT(dev)) { + cpt_phase_pointer_disable(dev, pipe); } /* still set train pattern 1 */ @@ -2757,10 +2861,7 @@ static void ironlake_pch_enable(struct drm_crtc *crtc) u32 reg, temp; /* For PCH output, training FDI link */ - if (IS_GEN6(dev)) - gen6_fdi_link_train(crtc); - else - ironlake_fdi_link_train(crtc); + dev_priv->display.fdi_link_train(crtc); intel_enable_pch_pll(dev_priv, pipe); @@ -2789,6 +2890,7 @@ static void ironlake_pch_enable(struct drm_crtc *crtc) /* For PCH DP, enable TRANS_DP_CTL */ if (HAS_PCH_CPT(dev) && intel_pipe_has_type(crtc, INTEL_OUTPUT_DISPLAYPORT)) { + u32 bpc = (I915_READ(PIPECONF(pipe)) & PIPE_BPC_MASK) >> 5; reg = TRANS_DP_CTL(pipe); temp = I915_READ(reg); temp &= ~(TRANS_DP_PORT_SEL_MASK | @@ -2796,7 +2898,7 @@ static void ironlake_pch_enable(struct drm_crtc *crtc) TRANS_DP_BPC_MASK); temp |= (TRANS_DP_OUTPUT_ENABLE | TRANS_DP_ENH_FRAMING); - temp |= TRANS_DP_8BPC; + temp |= bpc << 9; /* same format but at 11:9 */ if (crtc->mode.flags & DRM_MODE_FLAG_PHSYNC) temp |= TRANS_DP_HSYNC_ACTIVE_HIGH; @@ -2850,7 +2952,7 @@ static void ironlake_crtc_enable(struct drm_crtc *crtc) is_pch_port = intel_crtc_driving_pch(crtc); if (is_pch_port) - ironlake_fdi_enable(crtc); + ironlake_fdi_pll_enable(crtc); else ironlake_fdi_disable(crtc); @@ -2866,14 +2968,22 @@ static void ironlake_crtc_enable(struct drm_crtc *crtc) I915_WRITE(PF_WIN_SZ(pipe), dev_priv->pch_pf_size); } + /* + * On ILK+ LUT must be loaded before the pipe is running but with + * clocks enabled + */ + intel_crtc_load_lut(crtc); + intel_enable_pipe(dev_priv, pipe, is_pch_port); intel_enable_plane(dev_priv, plane, pipe); if (is_pch_port) ironlake_pch_enable(crtc); - intel_crtc_load_lut(crtc); + mutex_lock(&dev->struct_mutex); intel_update_fbc(dev); + mutex_unlock(&dev->struct_mutex); + intel_crtc_update_cursor(crtc, true); } @@ -2895,9 +3005,8 @@ static void ironlake_crtc_disable(struct drm_crtc *crtc) intel_disable_plane(dev_priv, plane, pipe); - if (dev_priv->cfb_plane == plane && - dev_priv->display.disable_fbc) - dev_priv->display.disable_fbc(dev); + if (dev_priv->cfb_plane == plane) + intel_disable_fbc(dev); intel_disable_pipe(dev_priv, pipe); @@ -2969,8 +3078,11 @@ static void ironlake_crtc_disable(struct drm_crtc *crtc) intel_crtc->active = false; intel_update_watermarks(dev); + + mutex_lock(&dev->struct_mutex); intel_update_fbc(dev); intel_clear_scanline_wait(dev); + mutex_unlock(&dev->struct_mutex); } static void ironlake_crtc_dpms(struct drm_crtc *crtc, int mode) @@ -3058,9 +3170,8 @@ static void i9xx_crtc_disable(struct drm_crtc *crtc) intel_crtc_dpms_overlay(intel_crtc, false); intel_crtc_update_cursor(crtc, false); - if (dev_priv->cfb_plane == plane && - dev_priv->display.disable_fbc) - dev_priv->display.disable_fbc(dev); + if (dev_priv->cfb_plane == plane) + intel_disable_fbc(dev); intel_disable_plane(dev_priv, plane, pipe); intel_disable_pipe(dev_priv, pipe); @@ -3497,11 +3608,11 @@ static unsigned long intel_calculate_wm(unsigned long clock_in_khz, 1000; entries_required = DIV_ROUND_UP(entries_required, wm->cacheline_size); - DRM_DEBUG_KMS("FIFO entries required for mode: %d\n", entries_required); + DRM_DEBUG_KMS("FIFO entries required for mode: %ld\n", entries_required); wm_size = fifo_size - (entries_required + wm->guard_size); - DRM_DEBUG_KMS("FIFO watermark level: %d\n", wm_size); + DRM_DEBUG_KMS("FIFO watermark level: %ld\n", wm_size); /* Don't promote wm_size to unsigned... */ if (wm_size > (long)wm->max_wm) @@ -3823,13 +3934,13 @@ static bool g4x_check_srwm(struct drm_device *dev, display_wm, cursor_wm); if (display_wm > display->max_wm) { - DRM_DEBUG_KMS("display watermark is too large(%d), disabling\n", + DRM_DEBUG_KMS("display watermark is too large(%d/%ld), disabling\n", display_wm, display->max_wm); return false; } if (cursor_wm > cursor->max_wm) { - DRM_DEBUG_KMS("cursor watermark is too large(%d), disabling\n", + DRM_DEBUG_KMS("cursor watermark is too large(%d/%ld), disabling\n", cursor_wm, cursor->max_wm); return false; } @@ -4143,54 +4254,6 @@ static void i830_update_wm(struct drm_device *dev) #define ILK_LP0_PLANE_LATENCY 700 #define ILK_LP0_CURSOR_LATENCY 1300 -static bool ironlake_compute_wm0(struct drm_device *dev, - int pipe, - const struct intel_watermark_params *display, - int display_latency_ns, - const struct intel_watermark_params *cursor, - int cursor_latency_ns, - int *plane_wm, - int *cursor_wm) -{ - struct drm_crtc *crtc; - int htotal, hdisplay, clock, pixel_size; - int line_time_us, line_count; - int entries, tlb_miss; - - crtc = intel_get_crtc_for_pipe(dev, pipe); - if (crtc->fb == NULL || !crtc->enabled) - return false; - - htotal = crtc->mode.htotal; - hdisplay = crtc->mode.hdisplay; - clock = crtc->mode.clock; - pixel_size = crtc->fb->bits_per_pixel / 8; - - /* Use the small buffer method to calculate plane watermark */ - entries = ((clock * pixel_size / 1000) * display_latency_ns) / 1000; - tlb_miss = display->fifo_size*display->cacheline_size - hdisplay * 8; - if (tlb_miss > 0) - entries += tlb_miss; - entries = DIV_ROUND_UP(entries, display->cacheline_size); - *plane_wm = entries + display->guard_size; - if (*plane_wm > (int)display->max_wm) - *plane_wm = display->max_wm; - - /* Use the large buffer method to calculate cursor watermark */ - line_time_us = ((htotal * 1000) / clock); - line_count = (cursor_latency_ns / line_time_us + 1000) / 1000; - entries = line_count * 64 * pixel_size; - tlb_miss = cursor->fifo_size*cursor->cacheline_size - hdisplay * 8; - if (tlb_miss > 0) - entries += tlb_miss; - entries = DIV_ROUND_UP(entries, cursor->cacheline_size); - *cursor_wm = entries + cursor->guard_size; - if (*cursor_wm > (int)cursor->max_wm) - *cursor_wm = (int)cursor->max_wm; - - return true; -} - /* * Check the wm result. * @@ -4299,12 +4362,12 @@ static void ironlake_update_wm(struct drm_device *dev) unsigned int enabled; enabled = 0; - if (ironlake_compute_wm0(dev, 0, - &ironlake_display_wm_info, - ILK_LP0_PLANE_LATENCY, - &ironlake_cursor_wm_info, - ILK_LP0_CURSOR_LATENCY, - &plane_wm, &cursor_wm)) { + if (g4x_compute_wm0(dev, 0, + &ironlake_display_wm_info, + ILK_LP0_PLANE_LATENCY, + &ironlake_cursor_wm_info, + ILK_LP0_CURSOR_LATENCY, + &plane_wm, &cursor_wm)) { I915_WRITE(WM0_PIPEA_ILK, (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm); DRM_DEBUG_KMS("FIFO watermarks For pipe A -" @@ -4313,12 +4376,12 @@ static void ironlake_update_wm(struct drm_device *dev) enabled |= 1; } - if (ironlake_compute_wm0(dev, 1, - &ironlake_display_wm_info, - ILK_LP0_PLANE_LATENCY, - &ironlake_cursor_wm_info, - ILK_LP0_CURSOR_LATENCY, - &plane_wm, &cursor_wm)) { + if (g4x_compute_wm0(dev, 1, + &ironlake_display_wm_info, + ILK_LP0_PLANE_LATENCY, + &ironlake_cursor_wm_info, + ILK_LP0_CURSOR_LATENCY, + &plane_wm, &cursor_wm)) { I915_WRITE(WM0_PIPEB_ILK, (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm); DRM_DEBUG_KMS("FIFO watermarks For pipe B -" @@ -4383,10 +4446,10 @@ static void sandybridge_update_wm(struct drm_device *dev) unsigned int enabled; enabled = 0; - if (ironlake_compute_wm0(dev, 0, - &sandybridge_display_wm_info, latency, - &sandybridge_cursor_wm_info, latency, - &plane_wm, &cursor_wm)) { + if (g4x_compute_wm0(dev, 0, + &sandybridge_display_wm_info, latency, + &sandybridge_cursor_wm_info, latency, + &plane_wm, &cursor_wm)) { I915_WRITE(WM0_PIPEA_ILK, (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm); DRM_DEBUG_KMS("FIFO watermarks For pipe A -" @@ -4395,10 +4458,10 @@ static void sandybridge_update_wm(struct drm_device *dev) enabled |= 1; } - if (ironlake_compute_wm0(dev, 1, - &sandybridge_display_wm_info, latency, - &sandybridge_cursor_wm_info, latency, - &plane_wm, &cursor_wm)) { + if (g4x_compute_wm0(dev, 1, + &sandybridge_display_wm_info, latency, + &sandybridge_cursor_wm_info, latency, + &plane_wm, &cursor_wm)) { I915_WRITE(WM0_PIPEB_ILK, (plane_wm << WM0_PIPE_PLANE_SHIFT) | cursor_wm); DRM_DEBUG_KMS("FIFO watermarks For pipe B -" @@ -4513,37 +4576,161 @@ static void intel_update_watermarks(struct drm_device *dev) static inline bool intel_panel_use_ssc(struct drm_i915_private *dev_priv) { - return dev_priv->lvds_use_ssc && i915_panel_use_ssc; + return dev_priv->lvds_use_ssc && i915_panel_use_ssc + && !(dev_priv->quirks & QUIRK_LVDS_SSC_DISABLE); } -static int intel_crtc_mode_set(struct drm_crtc *crtc, - struct drm_display_mode *mode, - struct drm_display_mode *adjusted_mode, - int x, int y, - struct drm_framebuffer *old_fb) +/** + * intel_choose_pipe_bpp_dither - figure out what color depth the pipe should send + * @crtc: CRTC structure + * + * A pipe may be connected to one or more outputs. Based on the depth of the + * attached framebuffer, choose a good color depth to use on the pipe. + * + * If possible, match the pipe depth to the fb depth. In some cases, this + * isn't ideal, because the connected output supports a lesser or restricted + * set of depths. Resolve that here: + * LVDS typically supports only 6bpc, so clamp down in that case + * HDMI supports only 8bpc or 12bpc, so clamp to 8bpc with dither for 10bpc + * Displays may support a restricted set as well, check EDID and clamp as + * appropriate. + * + * RETURNS: + * Dithering requirement (i.e. false if display bpc and pipe bpc match, + * true if they don't match). + */ +static bool intel_choose_pipe_bpp_dither(struct drm_crtc *crtc, + unsigned int *pipe_bpp) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_encoder *encoder; + struct drm_connector *connector; + unsigned int display_bpc = UINT_MAX, bpc; + + /* Walk the encoders & connectors on this crtc, get min bpc */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct intel_encoder *intel_encoder = to_intel_encoder(encoder); + + if (encoder->crtc != crtc) + continue; + + if (intel_encoder->type == INTEL_OUTPUT_LVDS) { + unsigned int lvds_bpc; + + if ((I915_READ(PCH_LVDS) & LVDS_A3_POWER_MASK) == + LVDS_A3_POWER_UP) + lvds_bpc = 8; + else + lvds_bpc = 6; + + if (lvds_bpc < display_bpc) { + DRM_DEBUG_DRIVER("clamping display bpc (was %d) to LVDS (%d)\n", display_bpc, lvds_bpc); + display_bpc = lvds_bpc; + } + continue; + } + + if (intel_encoder->type == INTEL_OUTPUT_EDP) { + /* Use VBT settings if we have an eDP panel */ + unsigned int edp_bpc = dev_priv->edp.bpp / 3; + + if (edp_bpc < display_bpc) { + DRM_DEBUG_DRIVER("clamping display bpc (was %d) to eDP (%d)\n", display_bpc, edp_bpc); + display_bpc = edp_bpc; + } + continue; + } + + /* Not one of the known troublemakers, check the EDID */ + list_for_each_entry(connector, &dev->mode_config.connector_list, + head) { + if (connector->encoder != encoder) + continue; + + /* Don't use an invalid EDID bpc value */ + if (connector->display_info.bpc && + connector->display_info.bpc < display_bpc) { + DRM_DEBUG_DRIVER("clamping display bpc (was %d) to EDID reported max of %d\n", display_bpc, connector->display_info.bpc); + display_bpc = connector->display_info.bpc; + } + } + + /* + * HDMI is either 12 or 8, so if the display lets 10bpc sneak + * through, clamp it down. (Note: >12bpc will be caught below.) + */ + if (intel_encoder->type == INTEL_OUTPUT_HDMI) { + if (display_bpc > 8 && display_bpc < 12) { + DRM_DEBUG_DRIVER("forcing bpc to 12 for HDMI\n"); + display_bpc = 12; + } else { + DRM_DEBUG_DRIVER("forcing bpc to 8 for HDMI\n"); + display_bpc = 8; + } + } + } + + /* + * We could just drive the pipe at the highest bpc all the time and + * enable dithering as needed, but that costs bandwidth. So choose + * the minimum value that expresses the full color range of the fb but + * also stays within the max display bpc discovered above. + */ + + switch (crtc->fb->depth) { + case 8: + bpc = 8; /* since we go through a colormap */ + break; + case 15: + case 16: + bpc = 6; /* min is 18bpp */ + break; + case 24: + bpc = min((unsigned int)8, display_bpc); + break; + case 30: + bpc = min((unsigned int)10, display_bpc); + break; + case 48: + bpc = min((unsigned int)12, display_bpc); + break; + default: + DRM_DEBUG("unsupported depth, assuming 24 bits\n"); + bpc = min((unsigned int)8, display_bpc); + break; + } + + DRM_DEBUG_DRIVER("setting pipe bpc to %d (max display bpc %d)\n", + bpc, display_bpc); + + *pipe_bpp = bpc * 3; + + return display_bpc != bpc; +} + +static int i9xx_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, + struct drm_framebuffer *old_fb) { struct drm_device *dev = crtc->dev; struct drm_i915_private *dev_priv = dev->dev_private; struct intel_crtc *intel_crtc = to_intel_crtc(crtc); int pipe = intel_crtc->pipe; int plane = intel_crtc->plane; - u32 fp_reg, dpll_reg; int refclk, num_connectors = 0; intel_clock_t clock, reduced_clock; u32 dpll, fp = 0, fp2 = 0, dspcntr, pipeconf; bool ok, has_reduced_clock = false, is_sdvo = false, is_dvo = false; bool is_crt = false, is_lvds = false, is_tv = false, is_dp = false; - struct intel_encoder *has_edp_encoder = NULL; struct drm_mode_config *mode_config = &dev->mode_config; struct intel_encoder *encoder; const intel_limit_t *limit; int ret; - struct fdi_m_n m_n = {0}; - u32 reg, temp; + u32 temp; u32 lvds_sync = 0; - int target_clock; - - drm_vblank_pre_modeset(dev, pipe); list_for_each_entry(encoder, &mode_config->encoder_list, base.head) { if (encoder->base.crtc != crtc) @@ -4571,9 +4758,6 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, case INTEL_OUTPUT_DISPLAYPORT: is_dp = true; break; - case INTEL_OUTPUT_EDP: - has_edp_encoder = encoder; - break; } num_connectors++; @@ -4585,9 +4769,6 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, refclk / 1000); } else if (!IS_GEN2(dev)) { refclk = 96000; - if (HAS_PCH_SPLIT(dev) && - (!has_edp_encoder || intel_encoder_is_pch_edp(&has_edp_encoder->base))) - refclk = 120000; /* 120Mhz refclk */ } else { refclk = 48000; } @@ -4601,7 +4782,6 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, ok = limit->find_pll(limit, crtc, adjusted_mode->clock, refclk, &clock); if (!ok) { DRM_ERROR("Couldn't find PLL settings for mode!\n"); - drm_vblank_post_modeset(dev, pipe); return -EINVAL; } @@ -4645,143 +4825,6 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, } } - /* FDI link */ - if (HAS_PCH_SPLIT(dev)) { - int pixel_multiplier = intel_mode_get_pixel_multiplier(adjusted_mode); - int lane = 0, link_bw, bpp; - /* CPU eDP doesn't require FDI link, so just set DP M/N - according to current link config */ - if (has_edp_encoder && !intel_encoder_is_pch_edp(&has_edp_encoder->base)) { - target_clock = mode->clock; - intel_edp_link_config(has_edp_encoder, - &lane, &link_bw); - } else { - /* [e]DP over FDI requires target mode clock - instead of link clock */ - if (is_dp || intel_encoder_is_pch_edp(&has_edp_encoder->base)) - target_clock = mode->clock; - else - target_clock = adjusted_mode->clock; - - /* FDI is a binary signal running at ~2.7GHz, encoding - * each output octet as 10 bits. The actual frequency - * is stored as a divider into a 100MHz clock, and the - * mode pixel clock is stored in units of 1KHz. - * Hence the bw of each lane in terms of the mode signal - * is: - */ - link_bw = intel_fdi_link_freq(dev) * MHz(100)/KHz(1)/10; - } - - /* determine panel color depth */ - temp = I915_READ(PIPECONF(pipe)); - temp &= ~PIPE_BPC_MASK; - if (is_lvds) { - /* the BPC will be 6 if it is 18-bit LVDS panel */ - if ((I915_READ(PCH_LVDS) & LVDS_A3_POWER_MASK) == LVDS_A3_POWER_UP) - temp |= PIPE_8BPC; - else - temp |= PIPE_6BPC; - } else if (has_edp_encoder) { - switch (dev_priv->edp.bpp/3) { - case 8: - temp |= PIPE_8BPC; - break; - case 10: - temp |= PIPE_10BPC; - break; - case 6: - temp |= PIPE_6BPC; - break; - case 12: - temp |= PIPE_12BPC; - break; - } - } else - temp |= PIPE_8BPC; - I915_WRITE(PIPECONF(pipe), temp); - - switch (temp & PIPE_BPC_MASK) { - case PIPE_8BPC: - bpp = 24; - break; - case PIPE_10BPC: - bpp = 30; - break; - case PIPE_6BPC: - bpp = 18; - break; - case PIPE_12BPC: - bpp = 36; - break; - default: - DRM_ERROR("unknown pipe bpc value\n"); - bpp = 24; - } - - if (!lane) { - /* - * Account for spread spectrum to avoid - * oversubscribing the link. Max center spread - * is 2.5%; use 5% for safety's sake. - */ - u32 bps = target_clock * bpp * 21 / 20; - lane = bps / (link_bw * 8) + 1; - } - - intel_crtc->fdi_lanes = lane; - - if (pixel_multiplier > 1) - link_bw *= pixel_multiplier; - ironlake_compute_m_n(bpp, lane, target_clock, link_bw, &m_n); - } - - /* Ironlake: try to setup display ref clock before DPLL - * enabling. This is only under driver's control after - * PCH B stepping, previous chipset stepping should be - * ignoring this setting. - */ - if (HAS_PCH_SPLIT(dev)) { - temp = I915_READ(PCH_DREF_CONTROL); - /* Always enable nonspread source */ - temp &= ~DREF_NONSPREAD_SOURCE_MASK; - temp |= DREF_NONSPREAD_SOURCE_ENABLE; - temp &= ~DREF_SSC_SOURCE_MASK; - temp |= DREF_SSC_SOURCE_ENABLE; - I915_WRITE(PCH_DREF_CONTROL, temp); - - POSTING_READ(PCH_DREF_CONTROL); - udelay(200); - - if (has_edp_encoder) { - if (intel_panel_use_ssc(dev_priv)) { - temp |= DREF_SSC1_ENABLE; - I915_WRITE(PCH_DREF_CONTROL, temp); - - POSTING_READ(PCH_DREF_CONTROL); - udelay(200); - } - temp &= ~DREF_CPU_SOURCE_OUTPUT_MASK; - - /* Enable CPU source on CPU attached eDP */ - if (!intel_encoder_is_pch_edp(&has_edp_encoder->base)) { - if (intel_panel_use_ssc(dev_priv)) - temp |= DREF_CPU_SOURCE_OUTPUT_DOWNSPREAD; - else - temp |= DREF_CPU_SOURCE_OUTPUT_NONSPREAD; - } else { - /* Enable SSC on PCH eDP if needed */ - if (intel_panel_use_ssc(dev_priv)) { - DRM_ERROR("enabling SSC on PCH\n"); - temp |= DREF_SUPERSPREAD_SOURCE_ENABLE; - } - } - I915_WRITE(PCH_DREF_CONTROL, temp); - POSTING_READ(PCH_DREF_CONTROL); - udelay(200); - } - } - if (IS_PINEVIEW(dev)) { fp = (1 << clock.n) << 16 | clock.m1 << 8 | clock.m2; if (has_reduced_clock) @@ -4794,25 +4837,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, reduced_clock.m2; } - /* Enable autotuning of the PLL clock (if permissible) */ - if (HAS_PCH_SPLIT(dev)) { - int factor = 21; - - if (is_lvds) { - if ((intel_panel_use_ssc(dev_priv) && - dev_priv->lvds_ssc_freq == 100) || - (I915_READ(PCH_LVDS) & LVDS_CLKB_POWER_MASK) == LVDS_CLKB_POWER_UP) - factor = 25; - } else if (is_sdvo && is_tv) - factor = 20; - - if (clock.m1 < factor * clock.n) - fp |= FP_CB_TUNE; - } - - dpll = 0; - if (!HAS_PCH_SPLIT(dev)) - dpll = DPLL_VGA_MODE_DIS; + dpll = DPLL_VGA_MODE_DIS; if (!IS_GEN2(dev)) { if (is_lvds) @@ -4824,12 +4849,10 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, if (pixel_multiplier > 1) { if (IS_I945G(dev) || IS_I945GM(dev) || IS_G33(dev)) dpll |= (pixel_multiplier - 1) << SDVO_MULTIPLIER_SHIFT_HIRES; - else if (HAS_PCH_SPLIT(dev)) - dpll |= (pixel_multiplier - 1) << PLL_REF_SDVO_HDMI_MULTIPLIER_SHIFT; } dpll |= DPLL_DVO_HIGH_SPEED; } - if (is_dp || intel_encoder_is_pch_edp(&has_edp_encoder->base)) + if (is_dp) dpll |= DPLL_DVO_HIGH_SPEED; /* compute bitmask from p1 value */ @@ -4837,9 +4860,6 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA01_P1_POST_DIV_SHIFT_PINEVIEW; else { dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA01_P1_POST_DIV_SHIFT; - /* also FPA1 */ - if (HAS_PCH_SPLIT(dev)) - dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA1_P1_POST_DIV_SHIFT; if (IS_G4X(dev) && has_reduced_clock) dpll |= (1 << (reduced_clock.p1 - 1)) << DPLL_FPA1_P1_POST_DIV_SHIFT; } @@ -4857,7 +4877,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, dpll |= DPLLB_LVDS_P2_CLOCK_DIV_14; break; } - if (INTEL_INFO(dev)->gen >= 4 && !HAS_PCH_SPLIT(dev)) + if (INTEL_INFO(dev)->gen >= 4) dpll |= (6 << PLL_LOAD_PULSE_PHASE_SHIFT); } else { if (is_lvds) { @@ -4891,12 +4911,10 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, /* Ironlake's plane is forced to pipe, bit 24 is to enable color space conversion */ - if (!HAS_PCH_SPLIT(dev)) { - if (pipe == 0) - dspcntr &= ~DISPPLANE_SEL_PIPE_MASK; - else - dspcntr |= DISPPLANE_SEL_PIPE_B; - } + if (pipe == 0) + dspcntr &= ~DISPPLANE_SEL_PIPE_MASK; + else + dspcntr |= DISPPLANE_SEL_PIPE_B; if (pipe == 0 && INTEL_INFO(dev)->gen < 4) { /* Enable pixel doubling when the dot clock is > 90% of the (display) @@ -4912,27 +4930,525 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, pipeconf &= ~PIPECONF_DOUBLE_WIDE; } - if (!HAS_PCH_SPLIT(dev)) - dpll |= DPLL_VCO_ENABLE; + dpll |= DPLL_VCO_ENABLE; DRM_DEBUG_KMS("Mode for pipe %c:\n", pipe == 0 ? 'A' : 'B'); drm_mode_debug_printmodeline(mode); - /* assign to Ironlake registers */ - if (HAS_PCH_SPLIT(dev)) { - fp_reg = PCH_FP0(pipe); - dpll_reg = PCH_DPLL(pipe); + I915_WRITE(FP0(pipe), fp); + I915_WRITE(DPLL(pipe), dpll & ~DPLL_VCO_ENABLE); + + POSTING_READ(DPLL(pipe)); + udelay(150); + + /* The LVDS pin pair needs to be on before the DPLLs are enabled. + * This is an exception to the general rule that mode_set doesn't turn + * things on. + */ + if (is_lvds) { + temp = I915_READ(LVDS); + temp |= LVDS_PORT_EN | LVDS_A0A2_CLKA_POWER_UP; + if (pipe == 1) { + temp |= LVDS_PIPEB_SELECT; + } else { + temp &= ~LVDS_PIPEB_SELECT; + } + /* set the corresponsding LVDS_BORDER bit */ + temp |= dev_priv->lvds_border_bits; + /* Set the B0-B3 data pairs corresponding to whether we're going to + * set the DPLLs for dual-channel mode or not. + */ + if (clock.p2 == 7) + temp |= LVDS_B0B3_POWER_UP | LVDS_CLKB_POWER_UP; + else + temp &= ~(LVDS_B0B3_POWER_UP | LVDS_CLKB_POWER_UP); + + /* It would be nice to set 24 vs 18-bit mode (LVDS_A3_POWER_UP) + * appropriately here, but we need to look more thoroughly into how + * panels behave in the two modes. + */ + /* set the dithering flag on LVDS as needed */ + if (INTEL_INFO(dev)->gen >= 4) { + if (dev_priv->lvds_dither) + temp |= LVDS_ENABLE_DITHER; + else + temp &= ~LVDS_ENABLE_DITHER; + } + if (adjusted_mode->flags & DRM_MODE_FLAG_NHSYNC) + lvds_sync |= LVDS_HSYNC_POLARITY; + if (adjusted_mode->flags & DRM_MODE_FLAG_NVSYNC) + lvds_sync |= LVDS_VSYNC_POLARITY; + if ((temp & (LVDS_HSYNC_POLARITY | LVDS_VSYNC_POLARITY)) + != lvds_sync) { + char flags[2] = "-+"; + DRM_INFO("Changing LVDS panel from " + "(%chsync, %cvsync) to (%chsync, %cvsync)\n", + flags[!(temp & LVDS_HSYNC_POLARITY)], + flags[!(temp & LVDS_VSYNC_POLARITY)], + flags[!(lvds_sync & LVDS_HSYNC_POLARITY)], + flags[!(lvds_sync & LVDS_VSYNC_POLARITY)]); + temp &= ~(LVDS_HSYNC_POLARITY | LVDS_VSYNC_POLARITY); + temp |= lvds_sync; + } + I915_WRITE(LVDS, temp); + } + + if (is_dp) { + intel_dp_set_m_n(crtc, mode, adjusted_mode); + } + + I915_WRITE(DPLL(pipe), dpll); + + /* Wait for the clocks to stabilize. */ + POSTING_READ(DPLL(pipe)); + udelay(150); + + if (INTEL_INFO(dev)->gen >= 4) { + temp = 0; + if (is_sdvo) { + temp = intel_mode_get_pixel_multiplier(adjusted_mode); + if (temp > 1) + temp = (temp - 1) << DPLL_MD_UDI_MULTIPLIER_SHIFT; + else + temp = 0; + } + I915_WRITE(DPLL_MD(pipe), temp); + } else { + /* The pixel multiplier can only be updated once the + * DPLL is enabled and the clocks are stable. + * + * So write it again. + */ + I915_WRITE(DPLL(pipe), dpll); + } + + intel_crtc->lowfreq_avail = false; + if (is_lvds && has_reduced_clock && i915_powersave) { + I915_WRITE(FP1(pipe), fp2); + intel_crtc->lowfreq_avail = true; + if (HAS_PIPE_CXSR(dev)) { + DRM_DEBUG_KMS("enabling CxSR downclocking\n"); + pipeconf |= PIPECONF_CXSR_DOWNCLOCK; + } + } else { + I915_WRITE(FP1(pipe), fp); + if (HAS_PIPE_CXSR(dev)) { + DRM_DEBUG_KMS("disabling CxSR downclocking\n"); + pipeconf &= ~PIPECONF_CXSR_DOWNCLOCK; + } + } + + if (adjusted_mode->flags & DRM_MODE_FLAG_INTERLACE) { + pipeconf |= PIPECONF_INTERLACE_W_FIELD_INDICATION; + /* the chip adds 2 halflines automatically */ + adjusted_mode->crtc_vdisplay -= 1; + adjusted_mode->crtc_vtotal -= 1; + adjusted_mode->crtc_vblank_start -= 1; + adjusted_mode->crtc_vblank_end -= 1; + adjusted_mode->crtc_vsync_end -= 1; + adjusted_mode->crtc_vsync_start -= 1; + } else + pipeconf &= ~PIPECONF_INTERLACE_W_FIELD_INDICATION; /* progressive */ + + I915_WRITE(HTOTAL(pipe), + (adjusted_mode->crtc_hdisplay - 1) | + ((adjusted_mode->crtc_htotal - 1) << 16)); + I915_WRITE(HBLANK(pipe), + (adjusted_mode->crtc_hblank_start - 1) | + ((adjusted_mode->crtc_hblank_end - 1) << 16)); + I915_WRITE(HSYNC(pipe), + (adjusted_mode->crtc_hsync_start - 1) | + ((adjusted_mode->crtc_hsync_end - 1) << 16)); + + I915_WRITE(VTOTAL(pipe), + (adjusted_mode->crtc_vdisplay - 1) | + ((adjusted_mode->crtc_vtotal - 1) << 16)); + I915_WRITE(VBLANK(pipe), + (adjusted_mode->crtc_vblank_start - 1) | + ((adjusted_mode->crtc_vblank_end - 1) << 16)); + I915_WRITE(VSYNC(pipe), + (adjusted_mode->crtc_vsync_start - 1) | + ((adjusted_mode->crtc_vsync_end - 1) << 16)); + + /* pipesrc and dspsize control the size that is scaled from, + * which should always be the user's requested size. + */ + I915_WRITE(DSPSIZE(plane), + ((mode->vdisplay - 1) << 16) | + (mode->hdisplay - 1)); + I915_WRITE(DSPPOS(plane), 0); + I915_WRITE(PIPESRC(pipe), + ((mode->hdisplay - 1) << 16) | (mode->vdisplay - 1)); + + I915_WRITE(PIPECONF(pipe), pipeconf); + POSTING_READ(PIPECONF(pipe)); + intel_enable_pipe(dev_priv, pipe, false); + + intel_wait_for_vblank(dev, pipe); + + I915_WRITE(DSPCNTR(plane), dspcntr); + POSTING_READ(DSPCNTR(plane)); + intel_enable_plane(dev_priv, plane, pipe); + + ret = intel_pipe_set_base(crtc, x, y, old_fb); + + intel_update_watermarks(dev); + + return ret; +} + +static void ironlake_update_pch_refclk(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_mode_config *mode_config = &dev->mode_config; + struct drm_crtc *crtc; + struct intel_encoder *encoder; + struct intel_encoder *has_edp_encoder = NULL; + u32 temp; + bool has_lvds = false; + + /* We need to take the global config into account */ + list_for_each_entry(crtc, &mode_config->crtc_list, head) { + if (!crtc->enabled) + continue; + + list_for_each_entry(encoder, &mode_config->encoder_list, + base.head) { + if (encoder->base.crtc != crtc) + continue; + + switch (encoder->type) { + case INTEL_OUTPUT_LVDS: + has_lvds = true; + case INTEL_OUTPUT_EDP: + has_edp_encoder = encoder; + break; + } + } + } + + /* Ironlake: try to setup display ref clock before DPLL + * enabling. This is only under driver's control after + * PCH B stepping, previous chipset stepping should be + * ignoring this setting. + */ + temp = I915_READ(PCH_DREF_CONTROL); + /* Always enable nonspread source */ + temp &= ~DREF_NONSPREAD_SOURCE_MASK; + temp |= DREF_NONSPREAD_SOURCE_ENABLE; + temp &= ~DREF_SSC_SOURCE_MASK; + temp |= DREF_SSC_SOURCE_ENABLE; + I915_WRITE(PCH_DREF_CONTROL, temp); + + POSTING_READ(PCH_DREF_CONTROL); + udelay(200); + + if (has_edp_encoder) { + if (intel_panel_use_ssc(dev_priv)) { + temp |= DREF_SSC1_ENABLE; + I915_WRITE(PCH_DREF_CONTROL, temp); + + POSTING_READ(PCH_DREF_CONTROL); + udelay(200); + } + temp &= ~DREF_CPU_SOURCE_OUTPUT_MASK; + + /* Enable CPU source on CPU attached eDP */ + if (!intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + if (intel_panel_use_ssc(dev_priv)) + temp |= DREF_CPU_SOURCE_OUTPUT_DOWNSPREAD; + else + temp |= DREF_CPU_SOURCE_OUTPUT_NONSPREAD; + } else { + /* Enable SSC on PCH eDP if needed */ + if (intel_panel_use_ssc(dev_priv)) { + DRM_ERROR("enabling SSC on PCH\n"); + temp |= DREF_SUPERSPREAD_SOURCE_ENABLE; + } + } + I915_WRITE(PCH_DREF_CONTROL, temp); + POSTING_READ(PCH_DREF_CONTROL); + udelay(200); + } +} + +static int ironlake_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, + struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int plane = intel_crtc->plane; + int refclk, num_connectors = 0; + intel_clock_t clock, reduced_clock; + u32 dpll, fp = 0, fp2 = 0, dspcntr, pipeconf; + bool ok, has_reduced_clock = false, is_sdvo = false; + bool is_crt = false, is_lvds = false, is_tv = false, is_dp = false; + struct intel_encoder *has_edp_encoder = NULL; + struct drm_mode_config *mode_config = &dev->mode_config; + struct intel_encoder *encoder; + const intel_limit_t *limit; + int ret; + struct fdi_m_n m_n = {0}; + u32 temp; + u32 lvds_sync = 0; + int target_clock, pixel_multiplier, lane, link_bw, factor; + unsigned int pipe_bpp; + bool dither; + + list_for_each_entry(encoder, &mode_config->encoder_list, base.head) { + if (encoder->base.crtc != crtc) + continue; + + switch (encoder->type) { + case INTEL_OUTPUT_LVDS: + is_lvds = true; + break; + case INTEL_OUTPUT_SDVO: + case INTEL_OUTPUT_HDMI: + is_sdvo = true; + if (encoder->needs_tv_clock) + is_tv = true; + break; + case INTEL_OUTPUT_TVOUT: + is_tv = true; + break; + case INTEL_OUTPUT_ANALOG: + is_crt = true; + break; + case INTEL_OUTPUT_DISPLAYPORT: + is_dp = true; + break; + case INTEL_OUTPUT_EDP: + has_edp_encoder = encoder; + break; + } + + num_connectors++; + } + + if (is_lvds && intel_panel_use_ssc(dev_priv) && num_connectors < 2) { + refclk = dev_priv->lvds_ssc_freq * 1000; + DRM_DEBUG_KMS("using SSC reference clock of %d MHz\n", + refclk / 1000); + } else { + refclk = 96000; + if (!has_edp_encoder || + intel_encoder_is_pch_edp(&has_edp_encoder->base)) + refclk = 120000; /* 120Mhz refclk */ + } + + /* + * Returns a set of divisors for the desired target clock with the given + * refclk, or FALSE. The returned values represent the clock equation: + * reflck * (5 * (m1 + 2) + (m2 + 2)) / (n + 2) / p1 / p2. + */ + limit = intel_limit(crtc, refclk); + ok = limit->find_pll(limit, crtc, adjusted_mode->clock, refclk, &clock); + if (!ok) { + DRM_ERROR("Couldn't find PLL settings for mode!\n"); + return -EINVAL; + } + + /* Ensure that the cursor is valid for the new mode before changing... */ + intel_crtc_update_cursor(crtc, true); + + if (is_lvds && dev_priv->lvds_downclock_avail) { + has_reduced_clock = limit->find_pll(limit, crtc, + dev_priv->lvds_downclock, + refclk, + &reduced_clock); + if (has_reduced_clock && (clock.p != reduced_clock.p)) { + /* + * If the different P is found, it means that we can't + * switch the display clock by using the FP0/FP1. + * In such case we will disable the LVDS downclock + * feature. + */ + DRM_DEBUG_KMS("Different P is found for " + "LVDS clock/downclock\n"); + has_reduced_clock = 0; + } + } + /* SDVO TV has fixed PLL values depend on its clock range, + this mirrors vbios setting. */ + if (is_sdvo && is_tv) { + if (adjusted_mode->clock >= 100000 + && adjusted_mode->clock < 140500) { + clock.p1 = 2; + clock.p2 = 10; + clock.n = 3; + clock.m1 = 16; + clock.m2 = 8; + } else if (adjusted_mode->clock >= 140500 + && adjusted_mode->clock <= 200000) { + clock.p1 = 1; + clock.p2 = 10; + clock.n = 6; + clock.m1 = 12; + clock.m2 = 8; + } + } + + /* FDI link */ + pixel_multiplier = intel_mode_get_pixel_multiplier(adjusted_mode); + lane = 0; + /* CPU eDP doesn't require FDI link, so just set DP M/N + according to current link config */ + if (has_edp_encoder && + !intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + target_clock = mode->clock; + intel_edp_link_config(has_edp_encoder, + &lane, &link_bw); } else { - fp_reg = FP0(pipe); - dpll_reg = DPLL(pipe); + /* [e]DP over FDI requires target mode clock + instead of link clock */ + if (is_dp || intel_encoder_is_pch_edp(&has_edp_encoder->base)) + target_clock = mode->clock; + else + target_clock = adjusted_mode->clock; + + /* FDI is a binary signal running at ~2.7GHz, encoding + * each output octet as 10 bits. The actual frequency + * is stored as a divider into a 100MHz clock, and the + * mode pixel clock is stored in units of 1KHz. + * Hence the bw of each lane in terms of the mode signal + * is: + */ + link_bw = intel_fdi_link_freq(dev) * MHz(100)/KHz(1)/10; } + /* determine panel color depth */ + temp = I915_READ(PIPECONF(pipe)); + temp &= ~PIPE_BPC_MASK; + dither = intel_choose_pipe_bpp_dither(crtc, &pipe_bpp); + switch (pipe_bpp) { + case 18: + temp |= PIPE_6BPC; + break; + case 24: + temp |= PIPE_8BPC; + break; + case 30: + temp |= PIPE_10BPC; + break; + case 36: + temp |= PIPE_12BPC; + break; + default: + WARN(1, "intel_choose_pipe_bpp returned invalid value %d\n", + pipe_bpp); + temp |= PIPE_8BPC; + pipe_bpp = 24; + break; + } + + intel_crtc->bpp = pipe_bpp; + I915_WRITE(PIPECONF(pipe), temp); + + if (!lane) { + /* + * Account for spread spectrum to avoid + * oversubscribing the link. Max center spread + * is 2.5%; use 5% for safety's sake. + */ + u32 bps = target_clock * intel_crtc->bpp * 21 / 20; + lane = bps / (link_bw * 8) + 1; + } + + intel_crtc->fdi_lanes = lane; + + if (pixel_multiplier > 1) + link_bw *= pixel_multiplier; + ironlake_compute_m_n(intel_crtc->bpp, lane, target_clock, link_bw, + &m_n); + + ironlake_update_pch_refclk(dev); + + fp = clock.n << 16 | clock.m1 << 8 | clock.m2; + if (has_reduced_clock) + fp2 = reduced_clock.n << 16 | reduced_clock.m1 << 8 | + reduced_clock.m2; + + /* Enable autotuning of the PLL clock (if permissible) */ + factor = 21; + if (is_lvds) { + if ((intel_panel_use_ssc(dev_priv) && + dev_priv->lvds_ssc_freq == 100) || + (I915_READ(PCH_LVDS) & LVDS_CLKB_POWER_MASK) == LVDS_CLKB_POWER_UP) + factor = 25; + } else if (is_sdvo && is_tv) + factor = 20; + + if (clock.m < factor * clock.n) + fp |= FP_CB_TUNE; + + dpll = 0; + + if (is_lvds) + dpll |= DPLLB_MODE_LVDS; + else + dpll |= DPLLB_MODE_DAC_SERIAL; + if (is_sdvo) { + int pixel_multiplier = intel_mode_get_pixel_multiplier(adjusted_mode); + if (pixel_multiplier > 1) { + dpll |= (pixel_multiplier - 1) << PLL_REF_SDVO_HDMI_MULTIPLIER_SHIFT; + } + dpll |= DPLL_DVO_HIGH_SPEED; + } + if (is_dp || intel_encoder_is_pch_edp(&has_edp_encoder->base)) + dpll |= DPLL_DVO_HIGH_SPEED; + + /* compute bitmask from p1 value */ + dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA01_P1_POST_DIV_SHIFT; + /* also FPA1 */ + dpll |= (1 << (clock.p1 - 1)) << DPLL_FPA1_P1_POST_DIV_SHIFT; + + switch (clock.p2) { + case 5: + dpll |= DPLL_DAC_SERIAL_P2_CLOCK_DIV_5; + break; + case 7: + dpll |= DPLLB_LVDS_P2_CLOCK_DIV_7; + break; + case 10: + dpll |= DPLL_DAC_SERIAL_P2_CLOCK_DIV_10; + break; + case 14: + dpll |= DPLLB_LVDS_P2_CLOCK_DIV_14; + break; + } + + if (is_sdvo && is_tv) + dpll |= PLL_REF_INPUT_TVCLKINBC; + else if (is_tv) + /* XXX: just matching BIOS for now */ + /* dpll |= PLL_REF_INPUT_TVCLKINBC; */ + dpll |= 3; + else if (is_lvds && intel_panel_use_ssc(dev_priv) && num_connectors < 2) + dpll |= PLLB_REF_INPUT_SPREADSPECTRUMIN; + else + dpll |= PLL_REF_INPUT_DREFCLK; + + /* setup pipeconf */ + pipeconf = I915_READ(PIPECONF(pipe)); + + /* Set up the display plane register */ + dspcntr = DISPPLANE_GAMMA_ENABLE; + + DRM_DEBUG_KMS("Mode for pipe %c:\n", pipe == 0 ? 'A' : 'B'); + drm_mode_debug_printmodeline(mode); + /* PCH eDP needs FDI, but CPU eDP does not */ if (!has_edp_encoder || intel_encoder_is_pch_edp(&has_edp_encoder->base)) { - I915_WRITE(fp_reg, fp); - I915_WRITE(dpll_reg, dpll & ~DPLL_VCO_ENABLE); + I915_WRITE(PCH_FP0(pipe), fp); + I915_WRITE(PCH_DPLL(pipe), dpll & ~DPLL_VCO_ENABLE); - POSTING_READ(dpll_reg); + POSTING_READ(PCH_DPLL(pipe)); udelay(150); } @@ -4964,11 +5480,7 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, * things on. */ if (is_lvds) { - reg = LVDS; - if (HAS_PCH_SPLIT(dev)) - reg = PCH_LVDS; - - temp = I915_READ(reg); + temp = I915_READ(PCH_LVDS); temp |= LVDS_PORT_EN | LVDS_A0A2_CLKA_POWER_UP; if (pipe == 1) { if (HAS_PCH_CPT(dev)) @@ -4995,13 +5507,6 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, * appropriately here, but we need to look more thoroughly into how * panels behave in the two modes. */ - /* set the dithering flag on non-PCH LVDS as needed */ - if (INTEL_INFO(dev)->gen >= 4 && !HAS_PCH_SPLIT(dev)) { - if (dev_priv->lvds_dither) - temp |= LVDS_ENABLE_DITHER; - else - temp &= ~LVDS_ENABLE_DITHER; - } if (adjusted_mode->flags & DRM_MODE_FLAG_NHSYNC) lvds_sync |= LVDS_HSYNC_POLARITY; if (adjusted_mode->flags & DRM_MODE_FLAG_NVSYNC) @@ -5018,22 +5523,18 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, temp &= ~(LVDS_HSYNC_POLARITY | LVDS_VSYNC_POLARITY); temp |= lvds_sync; } - I915_WRITE(reg, temp); + I915_WRITE(PCH_LVDS, temp); } - /* set the dithering flag and clear for anything other than a panel. */ - if (HAS_PCH_SPLIT(dev)) { - pipeconf &= ~PIPECONF_DITHER_EN; - pipeconf &= ~PIPECONF_DITHER_TYPE_MASK; - if (dev_priv->lvds_dither && (is_lvds || has_edp_encoder)) { - pipeconf |= PIPECONF_DITHER_EN; - pipeconf |= PIPECONF_DITHER_TYPE_ST1; - } + pipeconf &= ~PIPECONF_DITHER_EN; + pipeconf &= ~PIPECONF_DITHER_TYPE_MASK; + if ((is_lvds && dev_priv->lvds_dither) || dither) { + pipeconf |= PIPECONF_DITHER_EN; + pipeconf |= PIPECONF_DITHER_TYPE_ST1; } - if (is_dp || intel_encoder_is_pch_edp(&has_edp_encoder->base)) { intel_dp_set_m_n(crtc, mode, adjusted_mode); - } else if (HAS_PCH_SPLIT(dev)) { + } else { /* For non-DP output, clear any trans DP clock recovery setting.*/ I915_WRITE(TRANSDATA_M1(pipe), 0); I915_WRITE(TRANSDATA_N1(pipe), 0); @@ -5041,43 +5542,32 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, I915_WRITE(TRANSDPLINK_N1(pipe), 0); } - if (!has_edp_encoder || intel_encoder_is_pch_edp(&has_edp_encoder->base)) { - I915_WRITE(dpll_reg, dpll); + if (!has_edp_encoder || + intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + I915_WRITE(PCH_DPLL(pipe), dpll); /* Wait for the clocks to stabilize. */ - POSTING_READ(dpll_reg); + POSTING_READ(PCH_DPLL(pipe)); udelay(150); - if (INTEL_INFO(dev)->gen >= 4 && !HAS_PCH_SPLIT(dev)) { - temp = 0; - if (is_sdvo) { - temp = intel_mode_get_pixel_multiplier(adjusted_mode); - if (temp > 1) - temp = (temp - 1) << DPLL_MD_UDI_MULTIPLIER_SHIFT; - else - temp = 0; - } - I915_WRITE(DPLL_MD(pipe), temp); - } else { - /* The pixel multiplier can only be updated once the - * DPLL is enabled and the clocks are stable. - * - * So write it again. - */ - I915_WRITE(dpll_reg, dpll); - } + /* The pixel multiplier can only be updated once the + * DPLL is enabled and the clocks are stable. + * + * So write it again. + */ + I915_WRITE(PCH_DPLL(pipe), dpll); } intel_crtc->lowfreq_avail = false; if (is_lvds && has_reduced_clock && i915_powersave) { - I915_WRITE(fp_reg + 4, fp2); + I915_WRITE(PCH_FP1(pipe), fp2); intel_crtc->lowfreq_avail = true; if (HAS_PIPE_CXSR(dev)) { DRM_DEBUG_KMS("enabling CxSR downclocking\n"); pipeconf |= PIPECONF_CXSR_DOWNCLOCK; } } else { - I915_WRITE(fp_reg + 4, fp); + I915_WRITE(PCH_FP1(pipe), fp); if (HAS_PIPE_CXSR(dev)) { DRM_DEBUG_KMS("disabling CxSR downclocking\n"); pipeconf &= ~PIPECONF_CXSR_DOWNCLOCK; @@ -5116,33 +5606,24 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, (adjusted_mode->crtc_vsync_start - 1) | ((adjusted_mode->crtc_vsync_end - 1) << 16)); - /* pipesrc and dspsize control the size that is scaled from, - * which should always be the user's requested size. + /* pipesrc controls the size that is scaled from, which should + * always be the user's requested size. */ - if (!HAS_PCH_SPLIT(dev)) { - I915_WRITE(DSPSIZE(plane), - ((mode->vdisplay - 1) << 16) | - (mode->hdisplay - 1)); - I915_WRITE(DSPPOS(plane), 0); - } I915_WRITE(PIPESRC(pipe), ((mode->hdisplay - 1) << 16) | (mode->vdisplay - 1)); - if (HAS_PCH_SPLIT(dev)) { - I915_WRITE(PIPE_DATA_M1(pipe), TU_SIZE(m_n.tu) | m_n.gmch_m); - I915_WRITE(PIPE_DATA_N1(pipe), m_n.gmch_n); - I915_WRITE(PIPE_LINK_M1(pipe), m_n.link_m); - I915_WRITE(PIPE_LINK_N1(pipe), m_n.link_n); + I915_WRITE(PIPE_DATA_M1(pipe), TU_SIZE(m_n.tu) | m_n.gmch_m); + I915_WRITE(PIPE_DATA_N1(pipe), m_n.gmch_n); + I915_WRITE(PIPE_LINK_M1(pipe), m_n.link_m); + I915_WRITE(PIPE_LINK_N1(pipe), m_n.link_n); - if (has_edp_encoder && !intel_encoder_is_pch_edp(&has_edp_encoder->base)) { - ironlake_set_pll_edp(crtc, adjusted_mode->clock); - } + if (has_edp_encoder && + !intel_encoder_is_pch_edp(&has_edp_encoder->base)) { + ironlake_set_pll_edp(crtc, adjusted_mode->clock); } I915_WRITE(PIPECONF(pipe), pipeconf); POSTING_READ(PIPECONF(pipe)); - if (!HAS_PCH_SPLIT(dev)) - intel_enable_pipe(dev_priv, pipe, false); intel_wait_for_vblank(dev, pipe); @@ -5159,8 +5640,30 @@ static int intel_crtc_mode_set(struct drm_crtc *crtc, intel_update_watermarks(dev); + return ret; +} + +static int intel_crtc_mode_set(struct drm_crtc *crtc, + struct drm_display_mode *mode, + struct drm_display_mode *adjusted_mode, + int x, int y, + struct drm_framebuffer *old_fb) +{ + struct drm_device *dev = crtc->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int pipe = intel_crtc->pipe; + int ret; + + drm_vblank_pre_modeset(dev, pipe); + + ret = dev_priv->display.crtc_mode_set(crtc, mode, adjusted_mode, + x, y, old_fb); + drm_vblank_post_modeset(dev, pipe); + intel_crtc->dpms_mode = DRM_MODE_DPMS_ON; + return ret; } @@ -5350,21 +5853,15 @@ static int intel_crtc_cursor_set(struct drm_crtc *crtc, goto fail_locked; } - ret = i915_gem_object_pin(obj, PAGE_SIZE, true); - if (ret) { - DRM_ERROR("failed to pin cursor bo\n"); - goto fail_locked; - } - - ret = i915_gem_object_set_to_gtt_domain(obj, 0); + ret = i915_gem_object_pin_to_display_plane(obj, 0, NULL); if (ret) { DRM_ERROR("failed to move cursor bo into the GTT\n"); - goto fail_unpin; + goto fail_locked; } ret = i915_gem_object_put_fence(obj); if (ret) { - DRM_ERROR("failed to move cursor bo into the GTT\n"); + DRM_ERROR("failed to release fence for cursor"); goto fail_unpin; } @@ -5481,43 +5978,140 @@ static struct drm_display_mode load_detect_mode = { 704, 832, 0, 480, 489, 491, 520, 0, DRM_MODE_FLAG_NHSYNC | DRM_MODE_FLAG_NVSYNC), }; -struct drm_crtc *intel_get_load_detect_pipe(struct intel_encoder *intel_encoder, - struct drm_connector *connector, - struct drm_display_mode *mode, - int *dpms_mode) +static struct drm_framebuffer * +intel_framebuffer_create(struct drm_device *dev, + struct drm_mode_fb_cmd *mode_cmd, + struct drm_i915_gem_object *obj) +{ + struct intel_framebuffer *intel_fb; + int ret; + + intel_fb = kzalloc(sizeof(*intel_fb), GFP_KERNEL); + if (!intel_fb) { + drm_gem_object_unreference_unlocked(&obj->base); + return ERR_PTR(-ENOMEM); + } + + ret = intel_framebuffer_init(dev, intel_fb, mode_cmd, obj); + if (ret) { + drm_gem_object_unreference_unlocked(&obj->base); + kfree(intel_fb); + return ERR_PTR(ret); + } + + return &intel_fb->base; +} + +static u32 +intel_framebuffer_pitch_for_width(int width, int bpp) +{ + u32 pitch = DIV_ROUND_UP(width * bpp, 8); + return ALIGN(pitch, 64); +} + +static u32 +intel_framebuffer_size_for_mode(struct drm_display_mode *mode, int bpp) +{ + u32 pitch = intel_framebuffer_pitch_for_width(mode->hdisplay, bpp); + return ALIGN(pitch * mode->vdisplay, PAGE_SIZE); +} + +static struct drm_framebuffer * +intel_framebuffer_create_for_mode(struct drm_device *dev, + struct drm_display_mode *mode, + int depth, int bpp) +{ + struct drm_i915_gem_object *obj; + struct drm_mode_fb_cmd mode_cmd; + + obj = i915_gem_alloc_object(dev, + intel_framebuffer_size_for_mode(mode, bpp)); + if (obj == NULL) + return ERR_PTR(-ENOMEM); + + mode_cmd.width = mode->hdisplay; + mode_cmd.height = mode->vdisplay; + mode_cmd.depth = depth; + mode_cmd.bpp = bpp; + mode_cmd.pitch = intel_framebuffer_pitch_for_width(mode_cmd.width, bpp); + + return intel_framebuffer_create(dev, &mode_cmd, obj); +} + +static struct drm_framebuffer * +mode_fits_in_fbdev(struct drm_device *dev, + struct drm_display_mode *mode) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_i915_gem_object *obj; + struct drm_framebuffer *fb; + + if (dev_priv->fbdev == NULL) + return NULL; + + obj = dev_priv->fbdev->ifb.obj; + if (obj == NULL) + return NULL; + + fb = &dev_priv->fbdev->ifb.base; + if (fb->pitch < intel_framebuffer_pitch_for_width(mode->hdisplay, + fb->bits_per_pixel)) + return NULL; + + if (obj->base.size < mode->vdisplay * fb->pitch) + return NULL; + + return fb; +} + +bool intel_get_load_detect_pipe(struct intel_encoder *intel_encoder, + struct drm_connector *connector, + struct drm_display_mode *mode, + struct intel_load_detect_pipe *old) { struct intel_crtc *intel_crtc; struct drm_crtc *possible_crtc; - struct drm_crtc *supported_crtc =NULL; struct drm_encoder *encoder = &intel_encoder->base; struct drm_crtc *crtc = NULL; struct drm_device *dev = encoder->dev; - struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private; - struct drm_crtc_helper_funcs *crtc_funcs; + struct drm_framebuffer *old_fb; int i = -1; + DRM_DEBUG_KMS("[CONNECTOR:%d:%s], [ENCODER:%d:%s]\n", + connector->base.id, drm_get_connector_name(connector), + encoder->base.id, drm_get_encoder_name(encoder)); + /* * Algorithm gets a little messy: + * * - if the connector already has an assigned crtc, use it (but make * sure it's on first) + * * - try to find the first unused crtc that can drive this connector, * and use that if we find one - * - if there are no unused crtcs available, try to use the first - * one we found that supports the connector */ /* See if we already have a CRTC for this connector */ if (encoder->crtc) { crtc = encoder->crtc; - /* Make sure the crtc and connector are running */ + intel_crtc = to_intel_crtc(crtc); - *dpms_mode = intel_crtc->dpms_mode; + old->dpms_mode = intel_crtc->dpms_mode; + old->load_detect_temp = false; + + /* Make sure the crtc and connector are running */ if (intel_crtc->dpms_mode != DRM_MODE_DPMS_ON) { + struct drm_encoder_helper_funcs *encoder_funcs; + struct drm_crtc_helper_funcs *crtc_funcs; + crtc_funcs = crtc->helper_private; crtc_funcs->dpms(crtc, DRM_MODE_DPMS_ON); + + encoder_funcs = encoder->helper_private; encoder_funcs->dpms(encoder, DRM_MODE_DPMS_ON); } - return crtc; + + return true; } /* Find an unused one (if possible) */ @@ -5529,46 +6123,66 @@ struct drm_crtc *intel_get_load_detect_pipe(struct intel_encoder *intel_encoder, crtc = possible_crtc; break; } - if (!supported_crtc) - supported_crtc = possible_crtc; } /* * If we didn't find an unused CRTC, don't use any. */ if (!crtc) { - return NULL; + DRM_DEBUG_KMS("no pipe available for load-detect\n"); + return false; } encoder->crtc = crtc; connector->encoder = encoder; - intel_encoder->load_detect_temp = true; intel_crtc = to_intel_crtc(crtc); - *dpms_mode = intel_crtc->dpms_mode; + old->dpms_mode = intel_crtc->dpms_mode; + old->load_detect_temp = true; + old->release_fb = NULL; - if (!crtc->enabled) { - if (!mode) - mode = &load_detect_mode; - drm_crtc_helper_set_mode(crtc, mode, 0, 0, crtc->fb); - } else { - if (intel_crtc->dpms_mode != DRM_MODE_DPMS_ON) { - crtc_funcs = crtc->helper_private; - crtc_funcs->dpms(crtc, DRM_MODE_DPMS_ON); - } + if (!mode) + mode = &load_detect_mode; + + old_fb = crtc->fb; + + /* We need a framebuffer large enough to accommodate all accesses + * that the plane may generate whilst we perform load detection. + * We can not rely on the fbcon either being present (we get called + * during its initialisation to detect all boot displays, or it may + * not even exist) or that it is large enough to satisfy the + * requested mode. + */ + crtc->fb = mode_fits_in_fbdev(dev, mode); + if (crtc->fb == NULL) { + DRM_DEBUG_KMS("creating tmp fb for load-detection\n"); + crtc->fb = intel_framebuffer_create_for_mode(dev, mode, 24, 32); + old->release_fb = crtc->fb; + } else + DRM_DEBUG_KMS("reusing fbdev for load-detection framebuffer\n"); + if (IS_ERR(crtc->fb)) { + DRM_DEBUG_KMS("failed to allocate framebuffer for load-detection\n"); + crtc->fb = old_fb; + return false; + } - /* Add this connector to the crtc */ - encoder_funcs->mode_set(encoder, &crtc->mode, &crtc->mode); - encoder_funcs->commit(encoder); + if (!drm_crtc_helper_set_mode(crtc, mode, 0, 0, old_fb)) { + DRM_DEBUG_KMS("failed to set mode on load-detect pipe\n"); + if (old->release_fb) + old->release_fb->funcs->destroy(old->release_fb); + crtc->fb = old_fb; + return false; } + /* let the connector get through one full cycle before testing */ intel_wait_for_vblank(dev, intel_crtc->pipe); - return crtc; + return true; } void intel_release_load_detect_pipe(struct intel_encoder *intel_encoder, - struct drm_connector *connector, int dpms_mode) + struct drm_connector *connector, + struct intel_load_detect_pipe *old) { struct drm_encoder *encoder = &intel_encoder->base; struct drm_device *dev = encoder->dev; @@ -5576,19 +6190,24 @@ void intel_release_load_detect_pipe(struct intel_encoder *intel_encoder, struct drm_encoder_helper_funcs *encoder_funcs = encoder->helper_private; struct drm_crtc_helper_funcs *crtc_funcs = crtc->helper_private; - if (intel_encoder->load_detect_temp) { - encoder->crtc = NULL; + DRM_DEBUG_KMS("[CONNECTOR:%d:%s], [ENCODER:%d:%s]\n", + connector->base.id, drm_get_connector_name(connector), + encoder->base.id, drm_get_encoder_name(encoder)); + + if (old->load_detect_temp) { connector->encoder = NULL; - intel_encoder->load_detect_temp = false; - crtc->enabled = drm_helper_crtc_in_use(crtc); drm_helper_disable_unused_functions(dev); + + if (old->release_fb) + old->release_fb->funcs->destroy(old->release_fb); + + return; } /* Switch crtc and encoder back off if necessary */ - if (crtc->enabled && dpms_mode != DRM_MODE_DPMS_ON) { - if (encoder->crtc == crtc) - encoder_funcs->dpms(encoder, dpms_mode); - crtc_funcs->dpms(crtc, dpms_mode); + if (old->dpms_mode != DRM_MODE_DPMS_ON) { + encoder_funcs->dpms(encoder, old->dpms_mode); + crtc_funcs->dpms(crtc, old->dpms_mode); } } @@ -5945,6 +6564,7 @@ static void intel_unpin_work_fn(struct work_struct *__work) drm_gem_object_unreference(&work->pending_flip_obj->base); drm_gem_object_unreference(&work->old_fb_obj->base); + intel_update_fbc(work->dev); mutex_unlock(&work->dev->struct_mutex); kfree(work); } @@ -6055,6 +6675,197 @@ void intel_prepare_page_flip(struct drm_device *dev, int plane) spin_unlock_irqrestore(&dev->event_lock, flags); } +static int intel_gen2_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + unsigned long offset; + u32 flip_mask; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, LP_RING(dev_priv)); + if (ret) + goto out; + + /* Offset into the new buffer for cases of shared fbs between CRTCs */ + offset = crtc->y * fb->pitch + crtc->x * fb->bits_per_pixel/8; + + ret = BEGIN_LP_RING(6); + if (ret) + goto out; + + /* Can't queue multiple flips, so wait for the previous + * one to finish before executing the next. + */ + if (intel_crtc->plane) + flip_mask = MI_WAIT_FOR_PLANE_B_FLIP; + else + flip_mask = MI_WAIT_FOR_PLANE_A_FLIP; + OUT_RING(MI_WAIT_FOR_EVENT | flip_mask); + OUT_RING(MI_NOOP); + OUT_RING(MI_DISPLAY_FLIP | + MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); + OUT_RING(fb->pitch); + OUT_RING(obj->gtt_offset + offset); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); +out: + return ret; +} + +static int intel_gen3_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + unsigned long offset; + u32 flip_mask; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, LP_RING(dev_priv)); + if (ret) + goto out; + + /* Offset into the new buffer for cases of shared fbs between CRTCs */ + offset = crtc->y * fb->pitch + crtc->x * fb->bits_per_pixel/8; + + ret = BEGIN_LP_RING(6); + if (ret) + goto out; + + if (intel_crtc->plane) + flip_mask = MI_WAIT_FOR_PLANE_B_FLIP; + else + flip_mask = MI_WAIT_FOR_PLANE_A_FLIP; + OUT_RING(MI_WAIT_FOR_EVENT | flip_mask); + OUT_RING(MI_NOOP); + OUT_RING(MI_DISPLAY_FLIP_I915 | + MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); + OUT_RING(fb->pitch); + OUT_RING(obj->gtt_offset + offset); + OUT_RING(MI_NOOP); + + ADVANCE_LP_RING(); +out: + return ret; +} + +static int intel_gen4_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + uint32_t pf, pipesrc; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, LP_RING(dev_priv)); + if (ret) + goto out; + + ret = BEGIN_LP_RING(4); + if (ret) + goto out; + + /* i965+ uses the linear or tiled offsets from the + * Display Registers (which do not change across a page-flip) + * so we need only reprogram the base address. + */ + OUT_RING(MI_DISPLAY_FLIP | + MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); + OUT_RING(fb->pitch); + OUT_RING(obj->gtt_offset | obj->tiling_mode); + + /* XXX Enabling the panel-fitter across page-flip is so far + * untested on non-native modes, so ignore it for now. + * pf = I915_READ(pipe == 0 ? PFA_CTL_1 : PFB_CTL_1) & PF_ENABLE; + */ + pf = 0; + pipesrc = I915_READ(PIPESRC(intel_crtc->pipe)) & 0x0fff0fff; + OUT_RING(pf | pipesrc); + ADVANCE_LP_RING(); +out: + return ret; +} + +static int intel_gen6_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + uint32_t pf, pipesrc; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, LP_RING(dev_priv)); + if (ret) + goto out; + + ret = BEGIN_LP_RING(4); + if (ret) + goto out; + + OUT_RING(MI_DISPLAY_FLIP | + MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); + OUT_RING(fb->pitch | obj->tiling_mode); + OUT_RING(obj->gtt_offset); + + pf = I915_READ(PF_CTL(intel_crtc->pipe)) & PF_ENABLE; + pipesrc = I915_READ(PIPESRC(intel_crtc->pipe)) & 0x0fff0fff; + OUT_RING(pf | pipesrc); + ADVANCE_LP_RING(); +out: + return ret; +} + +/* + * On gen7 we currently use the blit ring because (in early silicon at least) + * the render ring doesn't give us interrpts for page flip completion, which + * means clients will hang after the first flip is queued. Fortunately the + * blit ring generates interrupts properly, so use it instead. + */ +static int intel_gen7_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + struct intel_ring_buffer *ring = &dev_priv->ring[BCS]; + int ret; + + ret = intel_pin_and_fence_fb_obj(dev, obj, ring); + if (ret) + goto out; + + ret = intel_ring_begin(ring, 4); + if (ret) + goto out; + + intel_ring_emit(ring, MI_DISPLAY_FLIP_I915 | (intel_crtc->plane << 19)); + intel_ring_emit(ring, (fb->pitch | obj->tiling_mode)); + intel_ring_emit(ring, (obj->gtt_offset)); + intel_ring_emit(ring, (MI_NOOP)); + intel_ring_advance(ring); +out: + return ret; +} + +static int intel_default_queue_flip(struct drm_device *dev, + struct drm_crtc *crtc, + struct drm_framebuffer *fb, + struct drm_i915_gem_object *obj) +{ + return -ENODEV; +} + static int intel_crtc_page_flip(struct drm_crtc *crtc, struct drm_framebuffer *fb, struct drm_pending_vblank_event *event) @@ -6065,9 +6876,7 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc, struct drm_i915_gem_object *obj; struct intel_crtc *intel_crtc = to_intel_crtc(crtc); struct intel_unpin_work *work; - unsigned long flags, offset; - int pipe = intel_crtc->pipe; - u32 pf, pipesrc; + unsigned long flags; int ret; work = kzalloc(sizeof *work, GFP_KERNEL); @@ -6096,9 +6905,6 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc, obj = intel_fb->obj; mutex_lock(&dev->struct_mutex); - ret = intel_pin_and_fence_fb_obj(dev, obj, LP_RING(dev_priv)); - if (ret) - goto cleanup_work; /* Reference the objects for the scheduled work. */ drm_gem_object_reference(&work->old_fb_obj->base); @@ -6110,101 +6916,31 @@ static int intel_crtc_page_flip(struct drm_crtc *crtc, if (ret) goto cleanup_objs; - if (IS_GEN3(dev) || IS_GEN2(dev)) { - u32 flip_mask; - - /* Can't queue multiple flips, so wait for the previous - * one to finish before executing the next. - */ - ret = BEGIN_LP_RING(2); - if (ret) - goto cleanup_objs; - - if (intel_crtc->plane) - flip_mask = MI_WAIT_FOR_PLANE_B_FLIP; - else - flip_mask = MI_WAIT_FOR_PLANE_A_FLIP; - OUT_RING(MI_WAIT_FOR_EVENT | flip_mask); - OUT_RING(MI_NOOP); - ADVANCE_LP_RING(); - } - work->pending_flip_obj = obj; work->enable_stall_check = true; - /* Offset into the new buffer for cases of shared fbs between CRTCs */ - offset = crtc->y * fb->pitch + crtc->x * fb->bits_per_pixel/8; - - ret = BEGIN_LP_RING(4); - if (ret) - goto cleanup_objs; - /* Block clients from rendering to the new back buffer until * the flip occurs and the object is no longer visible. */ atomic_add(1 << intel_crtc->plane, &work->old_fb_obj->pending_flip); - switch (INTEL_INFO(dev)->gen) { - case 2: - OUT_RING(MI_DISPLAY_FLIP | - MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); - OUT_RING(fb->pitch); - OUT_RING(obj->gtt_offset + offset); - OUT_RING(MI_NOOP); - break; - - case 3: - OUT_RING(MI_DISPLAY_FLIP_I915 | - MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); - OUT_RING(fb->pitch); - OUT_RING(obj->gtt_offset + offset); - OUT_RING(MI_NOOP); - break; - - case 4: - case 5: - /* i965+ uses the linear or tiled offsets from the - * Display Registers (which do not change across a page-flip) - * so we need only reprogram the base address. - */ - OUT_RING(MI_DISPLAY_FLIP | - MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); - OUT_RING(fb->pitch); - OUT_RING(obj->gtt_offset | obj->tiling_mode); - - /* XXX Enabling the panel-fitter across page-flip is so far - * untested on non-native modes, so ignore it for now. - * pf = I915_READ(pipe == 0 ? PFA_CTL_1 : PFB_CTL_1) & PF_ENABLE; - */ - pf = 0; - pipesrc = I915_READ(PIPESRC(pipe)) & 0x0fff0fff; - OUT_RING(pf | pipesrc); - break; - - case 6: - OUT_RING(MI_DISPLAY_FLIP | - MI_DISPLAY_FLIP_PLANE(intel_crtc->plane)); - OUT_RING(fb->pitch | obj->tiling_mode); - OUT_RING(obj->gtt_offset); - - pf = I915_READ(PF_CTL(pipe)) & PF_ENABLE; - pipesrc = I915_READ(PIPESRC(pipe)) & 0x0fff0fff; - OUT_RING(pf | pipesrc); - break; - } - ADVANCE_LP_RING(); + ret = dev_priv->display.queue_flip(dev, crtc, fb, obj); + if (ret) + goto cleanup_pending; + intel_disable_fbc(dev); mutex_unlock(&dev->struct_mutex); trace_i915_flip_request(intel_crtc->plane, obj); return 0; +cleanup_pending: + atomic_sub(1 << intel_crtc->plane, &work->old_fb_obj->pending_flip); cleanup_objs: drm_gem_object_unreference(&work->old_fb_obj->base); drm_gem_object_unreference(&obj->base); -cleanup_work: mutex_unlock(&dev->struct_mutex); spin_lock_irqsave(&dev->event_lock, flags); @@ -6322,6 +7058,7 @@ static void intel_crtc_init(struct drm_device *dev, int pipe) intel_crtc_reset(&intel_crtc->base); intel_crtc->active = true; /* force the pipe off on setup_init_config */ + intel_crtc->bpp = 24; /* default for pre-Ironlake */ if (HAS_PCH_SPLIT(dev)) { intel_helper_funcs.prepare = ironlake_crtc_prepare; @@ -6501,7 +7238,8 @@ static void intel_setup_outputs(struct drm_device *dev) intel_encoder_clones(dev, encoder->clone_mask); } - intel_panel_setup_backlight(dev); + /* disable all the possible outputs/crtcs before entering KMS mode */ + drm_helper_disable_unused_functions(dev); } static void intel_user_framebuffer_destroy(struct drm_framebuffer *fb) @@ -6545,6 +7283,11 @@ int intel_framebuffer_init(struct drm_device *dev, switch (mode_cmd->bpp) { case 8: case 16: + /* Only pre-ILK can handle 5:5:5 */ + if (mode_cmd->depth == 15 && !HAS_PCH_SPLIT(dev)) + return -EINVAL; + break; + case 24: case 32: break; @@ -6569,27 +7312,12 @@ intel_user_framebuffer_create(struct drm_device *dev, struct drm_mode_fb_cmd *mode_cmd) { struct drm_i915_gem_object *obj; - struct intel_framebuffer *intel_fb; - int ret; obj = to_intel_bo(drm_gem_object_lookup(dev, filp, mode_cmd->handle)); if (&obj->base == NULL) return ERR_PTR(-ENOENT); - intel_fb = kzalloc(sizeof(*intel_fb), GFP_KERNEL); - if (!intel_fb) { - drm_gem_object_unreference_unlocked(&obj->base); - return ERR_PTR(-ENOMEM); - } - - ret = intel_framebuffer_init(dev, intel_fb, mode_cmd, obj); - if (ret) { - drm_gem_object_unreference_unlocked(&obj->base); - kfree(intel_fb); - return ERR_PTR(ret); - } - - return &intel_fb->base; + return intel_framebuffer_create(dev, mode_cmd, obj); } static const struct drm_mode_config_funcs intel_mode_funcs = { @@ -6603,13 +7331,14 @@ intel_alloc_context_page(struct drm_device *dev) struct drm_i915_gem_object *ctx; int ret; + WARN_ON(!mutex_is_locked(&dev->struct_mutex)); + ctx = i915_gem_alloc_object(dev, 4096); if (!ctx) { DRM_DEBUG("failed to alloc power context, RC6 disabled\n"); return NULL; } - mutex_lock(&dev->struct_mutex); ret = i915_gem_object_pin(ctx, 4096, true); if (ret) { DRM_ERROR("failed to pin power context: %d\n", ret); @@ -6621,7 +7350,6 @@ intel_alloc_context_page(struct drm_device *dev) DRM_ERROR("failed to set-domain on power context: %d\n", ret); goto err_unpin; } - mutex_unlock(&dev->struct_mutex); return ctx; @@ -6756,6 +7484,11 @@ void gen6_disable_rps(struct drm_device *dev) I915_WRITE(GEN6_RPNSWREQ, 1 << 31); I915_WRITE(GEN6_PMINTRMSK, 0xffffffff); I915_WRITE(GEN6_PMIER, 0); + + spin_lock_irq(&dev_priv->rps_lock); + dev_priv->pm_iir = 0; + spin_unlock_irq(&dev_priv->rps_lock); + I915_WRITE(GEN6_PMIIR, I915_READ(GEN6_PMIIR)); } @@ -6849,7 +7582,7 @@ void gen6_enable_rps(struct drm_i915_private *dev_priv) { u32 rp_state_cap = I915_READ(GEN6_RP_STATE_CAP); u32 gt_perf_status = I915_READ(GEN6_GT_PERF_STATUS); - u32 pcu_mbox; + u32 pcu_mbox, rc6_mask = 0; int cur_freq, min_freq, max_freq; int i; @@ -6860,7 +7593,8 @@ void gen6_enable_rps(struct drm_i915_private *dev_priv) * userspace... */ I915_WRITE(GEN6_RC_STATE, 0); - __gen6_gt_force_wake_get(dev_priv); + mutex_lock(&dev_priv->dev->struct_mutex); + gen6_gt_force_wake_get(dev_priv); /* disable the counters and set deterministic thresholds */ I915_WRITE(GEN6_RC_CONTROL, 0); @@ -6880,9 +7614,12 @@ void gen6_enable_rps(struct drm_i915_private *dev_priv) I915_WRITE(GEN6_RC6p_THRESHOLD, 100000); I915_WRITE(GEN6_RC6pp_THRESHOLD, 64000); /* unused */ + if (i915_enable_rc6) + rc6_mask = GEN6_RC_CTL_RC6p_ENABLE | + GEN6_RC_CTL_RC6_ENABLE; + I915_WRITE(GEN6_RC_CONTROL, - GEN6_RC_CTL_RC6p_ENABLE | - GEN6_RC_CTL_RC6_ENABLE | + rc6_mask | GEN6_RC_CTL_EI_MODE(1) | GEN6_RC_CTL_HW_ENABLE); @@ -6954,170 +7691,300 @@ void gen6_enable_rps(struct drm_i915_private *dev_priv) GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_DOWN_EI_EXPIRED); + spin_lock_irq(&dev_priv->rps_lock); + WARN_ON(dev_priv->pm_iir != 0); I915_WRITE(GEN6_PMIMR, 0); + spin_unlock_irq(&dev_priv->rps_lock); /* enable all PM interrupts */ I915_WRITE(GEN6_PMINTRMSK, 0); - __gen6_gt_force_wake_put(dev_priv); + gen6_gt_force_wake_put(dev_priv); + mutex_unlock(&dev_priv->dev->struct_mutex); } -void intel_enable_clock_gating(struct drm_device *dev) +void gen6_update_ring_freq(struct drm_i915_private *dev_priv) { - struct drm_i915_private *dev_priv = dev->dev_private; - int pipe; + int min_freq = 15; + int gpu_freq, ia_freq, max_ia_freq; + int scaling_factor = 180; + max_ia_freq = cpufreq_quick_get_max(0); /* - * Disable clock gating reported to work incorrectly according to the - * specs, but enable as much else as we can. + * Default to measured freq if none found, PCU will ensure we don't go + * over */ - if (HAS_PCH_SPLIT(dev)) { - uint32_t dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE; + if (!max_ia_freq) + max_ia_freq = tsc_khz; - if (IS_GEN5(dev)) { - /* Required for FBC */ - dspclk_gate |= DPFCUNIT_CLOCK_GATE_DISABLE | - DPFCRUNIT_CLOCK_GATE_DISABLE | - DPFDUNIT_CLOCK_GATE_DISABLE; - /* Required for CxSR */ - dspclk_gate |= DPARBUNIT_CLOCK_GATE_DISABLE; - - I915_WRITE(PCH_3DCGDIS0, - MARIUNIT_CLOCK_GATE_DISABLE | - SVSMUNIT_CLOCK_GATE_DISABLE); - I915_WRITE(PCH_3DCGDIS1, - VFMUNIT_CLOCK_GATE_DISABLE); - } + /* Convert from kHz to MHz */ + max_ia_freq /= 1000; - I915_WRITE(PCH_DSPCLK_GATE_D, dspclk_gate); + mutex_lock(&dev_priv->dev->struct_mutex); - /* - * On Ibex Peak and Cougar Point, we need to disable clock - * gating for the panel power sequencer or it will fail to - * start up when no ports are active. - */ - I915_WRITE(SOUTH_DSPCLK_GATE_D, PCH_DPLSUNIT_CLOCK_GATE_DISABLE); + /* + * For each potential GPU frequency, load a ring frequency we'd like + * to use for memory access. We do this by specifying the IA frequency + * the PCU should use as a reference to determine the ring frequency. + */ + for (gpu_freq = dev_priv->max_delay; gpu_freq >= dev_priv->min_delay; + gpu_freq--) { + int diff = dev_priv->max_delay - gpu_freq; /* - * According to the spec the following bits should be set in - * order to enable memory self-refresh - * The bit 22/21 of 0x42004 - * The bit 5 of 0x42020 - * The bit 15 of 0x45000 + * For GPU frequencies less than 750MHz, just use the lowest + * ring freq. */ - if (IS_GEN5(dev)) { - I915_WRITE(ILK_DISPLAY_CHICKEN2, - (I915_READ(ILK_DISPLAY_CHICKEN2) | - ILK_DPARB_GATE | ILK_VSDPFD_FULL)); - I915_WRITE(ILK_DSPCLK_GATE, - (I915_READ(ILK_DSPCLK_GATE) | - ILK_DPARB_CLK_GATE)); - I915_WRITE(DISP_ARB_CTL, - (I915_READ(DISP_ARB_CTL) | - DISP_FBC_WM_DIS)); - I915_WRITE(WM3_LP_ILK, 0); - I915_WRITE(WM2_LP_ILK, 0); - I915_WRITE(WM1_LP_ILK, 0); - } - /* - * Based on the document from hardware guys the following bits - * should be set unconditionally in order to enable FBC. - * The bit 22 of 0x42000 - * The bit 22 of 0x42004 - * The bit 7,8,9 of 0x42020. - */ - if (IS_IRONLAKE_M(dev)) { - I915_WRITE(ILK_DISPLAY_CHICKEN1, - I915_READ(ILK_DISPLAY_CHICKEN1) | - ILK_FBCQ_DIS); - I915_WRITE(ILK_DISPLAY_CHICKEN2, - I915_READ(ILK_DISPLAY_CHICKEN2) | - ILK_DPARB_GATE); - I915_WRITE(ILK_DSPCLK_GATE, - I915_READ(ILK_DSPCLK_GATE) | - ILK_DPFC_DIS1 | - ILK_DPFC_DIS2 | - ILK_CLK_FBC); + if (gpu_freq < min_freq) + ia_freq = 800; + else + ia_freq = max_ia_freq - ((diff * scaling_factor) / 2); + ia_freq = DIV_ROUND_CLOSEST(ia_freq, 100); + + I915_WRITE(GEN6_PCODE_DATA, + (ia_freq << GEN6_PCODE_FREQ_IA_RATIO_SHIFT) | + gpu_freq); + I915_WRITE(GEN6_PCODE_MAILBOX, GEN6_PCODE_READY | + GEN6_PCODE_WRITE_MIN_FREQ_TABLE); + if (wait_for((I915_READ(GEN6_PCODE_MAILBOX) & + GEN6_PCODE_READY) == 0, 10)) { + DRM_ERROR("pcode write of freq table timed out\n"); + continue; } + } + + mutex_unlock(&dev_priv->dev->struct_mutex); +} + +static void ironlake_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE; + + /* Required for FBC */ + dspclk_gate |= DPFCUNIT_CLOCK_GATE_DISABLE | + DPFCRUNIT_CLOCK_GATE_DISABLE | + DPFDUNIT_CLOCK_GATE_DISABLE; + /* Required for CxSR */ + dspclk_gate |= DPARBUNIT_CLOCK_GATE_DISABLE; + + I915_WRITE(PCH_3DCGDIS0, + MARIUNIT_CLOCK_GATE_DISABLE | + SVSMUNIT_CLOCK_GATE_DISABLE); + I915_WRITE(PCH_3DCGDIS1, + VFMUNIT_CLOCK_GATE_DISABLE); + + I915_WRITE(PCH_DSPCLK_GATE_D, dspclk_gate); + + /* + * According to the spec the following bits should be set in + * order to enable memory self-refresh + * The bit 22/21 of 0x42004 + * The bit 5 of 0x42020 + * The bit 15 of 0x45000 + */ + I915_WRITE(ILK_DISPLAY_CHICKEN2, + (I915_READ(ILK_DISPLAY_CHICKEN2) | + ILK_DPARB_GATE | ILK_VSDPFD_FULL)); + I915_WRITE(ILK_DSPCLK_GATE, + (I915_READ(ILK_DSPCLK_GATE) | + ILK_DPARB_CLK_GATE)); + I915_WRITE(DISP_ARB_CTL, + (I915_READ(DISP_ARB_CTL) | + DISP_FBC_WM_DIS)); + I915_WRITE(WM3_LP_ILK, 0); + I915_WRITE(WM2_LP_ILK, 0); + I915_WRITE(WM1_LP_ILK, 0); + /* + * Based on the document from hardware guys the following bits + * should be set unconditionally in order to enable FBC. + * The bit 22 of 0x42000 + * The bit 22 of 0x42004 + * The bit 7,8,9 of 0x42020. + */ + if (IS_IRONLAKE_M(dev)) { + I915_WRITE(ILK_DISPLAY_CHICKEN1, + I915_READ(ILK_DISPLAY_CHICKEN1) | + ILK_FBCQ_DIS); I915_WRITE(ILK_DISPLAY_CHICKEN2, I915_READ(ILK_DISPLAY_CHICKEN2) | - ILK_ELPIN_409_SELECT); + ILK_DPARB_GATE); + I915_WRITE(ILK_DSPCLK_GATE, + I915_READ(ILK_DSPCLK_GATE) | + ILK_DPFC_DIS1 | + ILK_DPFC_DIS2 | + ILK_CLK_FBC); + } - if (IS_GEN5(dev)) { - I915_WRITE(_3D_CHICKEN2, - _3D_CHICKEN2_WM_READ_PIPELINED << 16 | - _3D_CHICKEN2_WM_READ_PIPELINED); - } + I915_WRITE(ILK_DISPLAY_CHICKEN2, + I915_READ(ILK_DISPLAY_CHICKEN2) | + ILK_ELPIN_409_SELECT); + I915_WRITE(_3D_CHICKEN2, + _3D_CHICKEN2_WM_READ_PIPELINED << 16 | + _3D_CHICKEN2_WM_READ_PIPELINED); +} - if (IS_GEN6(dev)) { - I915_WRITE(WM3_LP_ILK, 0); - I915_WRITE(WM2_LP_ILK, 0); - I915_WRITE(WM1_LP_ILK, 0); +static void gen6_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int pipe; + uint32_t dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE; - /* - * According to the spec the following bits should be - * set in order to enable memory self-refresh and fbc: - * The bit21 and bit22 of 0x42000 - * The bit21 and bit22 of 0x42004 - * The bit5 and bit7 of 0x42020 - * The bit14 of 0x70180 - * The bit14 of 0x71180 - */ - I915_WRITE(ILK_DISPLAY_CHICKEN1, - I915_READ(ILK_DISPLAY_CHICKEN1) | - ILK_FBCQ_DIS | ILK_PABSTRETCH_DIS); - I915_WRITE(ILK_DISPLAY_CHICKEN2, - I915_READ(ILK_DISPLAY_CHICKEN2) | - ILK_DPARB_GATE | ILK_VSDPFD_FULL); - I915_WRITE(ILK_DSPCLK_GATE, - I915_READ(ILK_DSPCLK_GATE) | - ILK_DPARB_CLK_GATE | - ILK_DPFD_CLK_GATE); - - for_each_pipe(pipe) - I915_WRITE(DSPCNTR(pipe), - I915_READ(DSPCNTR(pipe)) | - DISPPLANE_TRICKLE_FEED_DISABLE); - } - } else if (IS_G4X(dev)) { - uint32_t dspclk_gate; - I915_WRITE(RENCLK_GATE_D1, 0); - I915_WRITE(RENCLK_GATE_D2, VF_UNIT_CLOCK_GATE_DISABLE | - GS_UNIT_CLOCK_GATE_DISABLE | - CL_UNIT_CLOCK_GATE_DISABLE); - I915_WRITE(RAMCLK_GATE_D, 0); - dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE | - OVRUNIT_CLOCK_GATE_DISABLE | - OVCUNIT_CLOCK_GATE_DISABLE; - if (IS_GM45(dev)) - dspclk_gate |= DSSUNIT_CLOCK_GATE_DISABLE; - I915_WRITE(DSPCLK_GATE_D, dspclk_gate); - } else if (IS_CRESTLINE(dev)) { - I915_WRITE(RENCLK_GATE_D1, I965_RCC_CLOCK_GATE_DISABLE); - I915_WRITE(RENCLK_GATE_D2, 0); - I915_WRITE(DSPCLK_GATE_D, 0); - I915_WRITE(RAMCLK_GATE_D, 0); - I915_WRITE16(DEUC, 0); - } else if (IS_BROADWATER(dev)) { - I915_WRITE(RENCLK_GATE_D1, I965_RCZ_CLOCK_GATE_DISABLE | - I965_RCC_CLOCK_GATE_DISABLE | - I965_RCPB_CLOCK_GATE_DISABLE | - I965_ISC_CLOCK_GATE_DISABLE | - I965_FBC_CLOCK_GATE_DISABLE); - I915_WRITE(RENCLK_GATE_D2, 0); - } else if (IS_GEN3(dev)) { - u32 dstate = I915_READ(D_STATE); + I915_WRITE(PCH_DSPCLK_GATE_D, dspclk_gate); + + I915_WRITE(ILK_DISPLAY_CHICKEN2, + I915_READ(ILK_DISPLAY_CHICKEN2) | + ILK_ELPIN_409_SELECT); + + I915_WRITE(WM3_LP_ILK, 0); + I915_WRITE(WM2_LP_ILK, 0); + I915_WRITE(WM1_LP_ILK, 0); + + /* + * According to the spec the following bits should be + * set in order to enable memory self-refresh and fbc: + * The bit21 and bit22 of 0x42000 + * The bit21 and bit22 of 0x42004 + * The bit5 and bit7 of 0x42020 + * The bit14 of 0x70180 + * The bit14 of 0x71180 + */ + I915_WRITE(ILK_DISPLAY_CHICKEN1, + I915_READ(ILK_DISPLAY_CHICKEN1) | + ILK_FBCQ_DIS | ILK_PABSTRETCH_DIS); + I915_WRITE(ILK_DISPLAY_CHICKEN2, + I915_READ(ILK_DISPLAY_CHICKEN2) | + ILK_DPARB_GATE | ILK_VSDPFD_FULL); + I915_WRITE(ILK_DSPCLK_GATE, + I915_READ(ILK_DSPCLK_GATE) | + ILK_DPARB_CLK_GATE | + ILK_DPFD_CLK_GATE); + + for_each_pipe(pipe) { + I915_WRITE(DSPCNTR(pipe), + I915_READ(DSPCNTR(pipe)) | + DISPPLANE_TRICKLE_FEED_DISABLE); + intel_flush_display_plane(dev_priv, pipe); + } +} + +static void ivybridge_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int pipe; + uint32_t dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE; + + I915_WRITE(PCH_DSPCLK_GATE_D, dspclk_gate); - dstate |= DSTATE_PLL_D3_OFF | DSTATE_GFX_CLOCK_GATING | - DSTATE_DOT_CLOCK_GATING; - I915_WRITE(D_STATE, dstate); - } else if (IS_I85X(dev) || IS_I865G(dev)) { - I915_WRITE(RENCLK_GATE_D1, SV_CLOCK_GATE_DISABLE); - } else if (IS_I830(dev)) { - I915_WRITE(DSPCLK_GATE_D, OVRUNIT_CLOCK_GATE_DISABLE); + I915_WRITE(WM3_LP_ILK, 0); + I915_WRITE(WM2_LP_ILK, 0); + I915_WRITE(WM1_LP_ILK, 0); + + I915_WRITE(ILK_DSPCLK_GATE, IVB_VRHUNIT_CLK_GATE); + + for_each_pipe(pipe) { + I915_WRITE(DSPCNTR(pipe), + I915_READ(DSPCNTR(pipe)) | + DISPPLANE_TRICKLE_FEED_DISABLE); + intel_flush_display_plane(dev_priv, pipe); } } +static void g4x_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + uint32_t dspclk_gate; + + I915_WRITE(RENCLK_GATE_D1, 0); + I915_WRITE(RENCLK_GATE_D2, VF_UNIT_CLOCK_GATE_DISABLE | + GS_UNIT_CLOCK_GATE_DISABLE | + CL_UNIT_CLOCK_GATE_DISABLE); + I915_WRITE(RAMCLK_GATE_D, 0); + dspclk_gate = VRHUNIT_CLOCK_GATE_DISABLE | + OVRUNIT_CLOCK_GATE_DISABLE | + OVCUNIT_CLOCK_GATE_DISABLE; + if (IS_GM45(dev)) + dspclk_gate |= DSSUNIT_CLOCK_GATE_DISABLE; + I915_WRITE(DSPCLK_GATE_D, dspclk_gate); +} + +static void crestline_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(RENCLK_GATE_D1, I965_RCC_CLOCK_GATE_DISABLE); + I915_WRITE(RENCLK_GATE_D2, 0); + I915_WRITE(DSPCLK_GATE_D, 0); + I915_WRITE(RAMCLK_GATE_D, 0); + I915_WRITE16(DEUC, 0); +} + +static void broadwater_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(RENCLK_GATE_D1, I965_RCZ_CLOCK_GATE_DISABLE | + I965_RCC_CLOCK_GATE_DISABLE | + I965_RCPB_CLOCK_GATE_DISABLE | + I965_ISC_CLOCK_GATE_DISABLE | + I965_FBC_CLOCK_GATE_DISABLE); + I915_WRITE(RENCLK_GATE_D2, 0); +} + +static void gen3_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + u32 dstate = I915_READ(D_STATE); + + dstate |= DSTATE_PLL_D3_OFF | DSTATE_GFX_CLOCK_GATING | + DSTATE_DOT_CLOCK_GATING; + I915_WRITE(D_STATE, dstate); +} + +static void i85x_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(RENCLK_GATE_D1, SV_CLOCK_GATE_DISABLE); +} + +static void i830_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + I915_WRITE(DSPCLK_GATE_D, OVRUNIT_CLOCK_GATE_DISABLE); +} + +static void ibx_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + /* + * On Ibex Peak and Cougar Point, we need to disable clock + * gating for the panel power sequencer or it will fail to + * start up when no ports are active. + */ + I915_WRITE(SOUTH_DSPCLK_GATE_D, PCH_DPLSUNIT_CLOCK_GATE_DISABLE); +} + +static void cpt_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + int pipe; + + /* + * On Ibex Peak and Cougar Point, we need to disable clock + * gating for the panel power sequencer or it will fail to + * start up when no ports are active. + */ + I915_WRITE(SOUTH_DSPCLK_GATE_D, PCH_DPLSUNIT_CLOCK_GATE_DISABLE); + I915_WRITE(SOUTH_CHICKEN2, I915_READ(SOUTH_CHICKEN2) | + DPLS_EDP_PPS_FIX_DIS); + /* Without this, mode sets may fail silently on FDI */ + for_each_pipe(pipe) + I915_WRITE(TRANS_CHICKEN2(pipe), TRANS_AUTOTRAIN_GEN_STALL_DIS); +} + static void ironlake_teardown_rc6(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -7185,9 +8052,12 @@ void ironlake_enable_rc6(struct drm_device *dev) if (!i915_enable_rc6) return; + mutex_lock(&dev->struct_mutex); ret = ironlake_setup_rc6(dev); - if (ret) + if (ret) { + mutex_unlock(&dev->struct_mutex); return; + } /* * GPU can automatically power down the render unit if given a page @@ -7196,6 +8066,7 @@ void ironlake_enable_rc6(struct drm_device *dev) ret = BEGIN_LP_RING(6); if (ret) { ironlake_teardown_rc6(dev); + mutex_unlock(&dev->struct_mutex); return; } @@ -7211,10 +8082,33 @@ void ironlake_enable_rc6(struct drm_device *dev) OUT_RING(MI_FLUSH); ADVANCE_LP_RING(); + /* + * Wait for the command parser to advance past MI_SET_CONTEXT. The HW + * does an implicit flush, combined with MI_FLUSH above, it should be + * safe to assume that renderctx is valid + */ + ret = intel_wait_ring_idle(LP_RING(dev_priv)); + if (ret) { + DRM_ERROR("failed to enable ironlake power power savings\n"); + ironlake_teardown_rc6(dev); + mutex_unlock(&dev->struct_mutex); + return; + } + I915_WRITE(PWRCTXA, dev_priv->pwrctx->gtt_offset | PWRCTX_EN); I915_WRITE(RSTDBYCTL, I915_READ(RSTDBYCTL) & ~RCX_SW_EXIT); + mutex_unlock(&dev->struct_mutex); } +void intel_init_clock_gating(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + + dev_priv->display.init_clock_gating(dev); + + if (dev_priv->display.init_pch_clock_gating) + dev_priv->display.init_pch_clock_gating(dev); +} /* Set up chip specific display functions */ static void intel_init_display(struct drm_device *dev) @@ -7222,10 +8116,15 @@ static void intel_init_display(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; /* We always want a DPMS function */ - if (HAS_PCH_SPLIT(dev)) + if (HAS_PCH_SPLIT(dev)) { dev_priv->display.dpms = ironlake_crtc_dpms; - else + dev_priv->display.crtc_mode_set = ironlake_crtc_mode_set; + dev_priv->display.update_plane = ironlake_update_plane; + } else { dev_priv->display.dpms = i9xx_crtc_dpms; + dev_priv->display.crtc_mode_set = i9xx_crtc_mode_set; + dev_priv->display.update_plane = i9xx_update_plane; + } if (I915_HAS_FBC(dev)) { if (HAS_PCH_SPLIT(dev)) { @@ -7269,6 +8168,11 @@ static void intel_init_display(struct drm_device *dev) /* For FIFO watermark updates */ if (HAS_PCH_SPLIT(dev)) { + if (HAS_PCH_IBX(dev)) + dev_priv->display.init_pch_clock_gating = ibx_init_clock_gating; + else if (HAS_PCH_CPT(dev)) + dev_priv->display.init_pch_clock_gating = cpt_init_clock_gating; + if (IS_GEN5(dev)) { if (I915_READ(MLTR_ILK) & ILK_SRLT_MASK) dev_priv->display.update_wm = ironlake_update_wm; @@ -7277,6 +8181,8 @@ static void intel_init_display(struct drm_device *dev) "Disable CxSR\n"); dev_priv->display.update_wm = NULL; } + dev_priv->display.fdi_link_train = ironlake_fdi_link_train; + dev_priv->display.init_clock_gating = ironlake_init_clock_gating; } else if (IS_GEN6(dev)) { if (SNB_READ_WM0_LATENCY()) { dev_priv->display.update_wm = sandybridge_update_wm; @@ -7285,6 +8191,20 @@ static void intel_init_display(struct drm_device *dev) "Disable CxSR\n"); dev_priv->display.update_wm = NULL; } + dev_priv->display.fdi_link_train = gen6_fdi_link_train; + dev_priv->display.init_clock_gating = gen6_init_clock_gating; + } else if (IS_IVYBRIDGE(dev)) { + /* FIXME: detect B0+ stepping and use auto training */ + dev_priv->display.fdi_link_train = ivb_manual_fdi_link_train; + if (SNB_READ_WM0_LATENCY()) { + dev_priv->display.update_wm = sandybridge_update_wm; + } else { + DRM_DEBUG_KMS("Failed to read display plane latency. " + "Disable CxSR\n"); + dev_priv->display.update_wm = NULL; + } + dev_priv->display.init_clock_gating = ivybridge_init_clock_gating; + } else dev_priv->display.update_wm = NULL; } else if (IS_PINEVIEW(dev)) { @@ -7302,23 +8222,61 @@ static void intel_init_display(struct drm_device *dev) dev_priv->display.update_wm = NULL; } else dev_priv->display.update_wm = pineview_update_wm; - } else if (IS_G4X(dev)) + dev_priv->display.init_clock_gating = gen3_init_clock_gating; + } else if (IS_G4X(dev)) { dev_priv->display.update_wm = g4x_update_wm; - else if (IS_GEN4(dev)) + dev_priv->display.init_clock_gating = g4x_init_clock_gating; + } else if (IS_GEN4(dev)) { dev_priv->display.update_wm = i965_update_wm; - else if (IS_GEN3(dev)) { + if (IS_CRESTLINE(dev)) + dev_priv->display.init_clock_gating = crestline_init_clock_gating; + else if (IS_BROADWATER(dev)) + dev_priv->display.init_clock_gating = broadwater_init_clock_gating; + } else if (IS_GEN3(dev)) { dev_priv->display.update_wm = i9xx_update_wm; dev_priv->display.get_fifo_size = i9xx_get_fifo_size; + dev_priv->display.init_clock_gating = gen3_init_clock_gating; + } else if (IS_I865G(dev)) { + dev_priv->display.update_wm = i830_update_wm; + dev_priv->display.init_clock_gating = i85x_init_clock_gating; + dev_priv->display.get_fifo_size = i830_get_fifo_size; } else if (IS_I85X(dev)) { dev_priv->display.update_wm = i9xx_update_wm; dev_priv->display.get_fifo_size = i85x_get_fifo_size; + dev_priv->display.init_clock_gating = i85x_init_clock_gating; } else { dev_priv->display.update_wm = i830_update_wm; + dev_priv->display.init_clock_gating = i830_init_clock_gating; if (IS_845G(dev)) dev_priv->display.get_fifo_size = i845_get_fifo_size; else dev_priv->display.get_fifo_size = i830_get_fifo_size; } + + /* Default just returns -ENODEV to indicate unsupported */ + dev_priv->display.queue_flip = intel_default_queue_flip; + + switch (INTEL_INFO(dev)->gen) { + case 2: + dev_priv->display.queue_flip = intel_gen2_queue_flip; + break; + + case 3: + dev_priv->display.queue_flip = intel_gen3_queue_flip; + break; + + case 4: + case 5: + dev_priv->display.queue_flip = intel_gen4_queue_flip; + break; + + case 6: + dev_priv->display.queue_flip = intel_gen6_queue_flip; + break; + case 7: + dev_priv->display.queue_flip = intel_gen7_queue_flip; + break; + } } /* @@ -7334,6 +8292,15 @@ static void quirk_pipea_force (struct drm_device *dev) DRM_DEBUG_DRIVER("applying pipe a force quirk\n"); } +/* + * Some machines (Lenovo U160) do not work with SSC on LVDS for some reason + */ +static void quirk_ssc_force_disable(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + dev_priv->quirks |= QUIRK_LVDS_SSC_DISABLE; +} + struct intel_quirk { int device; int subsystem_vendor; @@ -7362,6 +8329,12 @@ struct intel_quirk intel_quirks[] = { /* 855 & before need to leave pipe A & dpll A up */ { 0x3582, PCI_ANY_ID, PCI_ANY_ID, quirk_pipea_force }, { 0x2562, PCI_ANY_ID, PCI_ANY_ID, quirk_pipea_force }, + + /* Lenovo U160 cannot use SSC on LVDS */ + { 0x0046, 0x17aa, 0x3920, quirk_ssc_force_disable }, + + /* Sony Vaio Y cannot use SSC on LVDS */ + { 0x0046, 0x104d, 0x9076, quirk_ssc_force_disable }, }; static void intel_init_quirks(struct drm_device *dev) @@ -7439,27 +8412,31 @@ void intel_modeset_init(struct drm_device *dev) intel_crtc_init(dev, i); } - intel_setup_outputs(dev); - - intel_enable_clock_gating(dev); - /* Just disable it once at startup */ i915_disable_vga(dev); + intel_setup_outputs(dev); + + intel_init_clock_gating(dev); if (IS_IRONLAKE_M(dev)) { ironlake_enable_drps(dev); intel_init_emon(dev); } - if (IS_GEN6(dev)) + if (IS_GEN6(dev) || IS_GEN7(dev)) { gen6_enable_rps(dev_priv); - - if (IS_IRONLAKE_M(dev)) - ironlake_enable_rc6(dev); + gen6_update_ring_freq(dev_priv); + } INIT_WORK(&dev_priv->idle_work, intel_idle_update); setup_timer(&dev_priv->idle_timer, intel_gpu_idle_timer, (unsigned long)dev); +} + +void intel_modeset_gem_init(struct drm_device *dev) +{ + if (IS_IRONLAKE_M(dev)) + ironlake_enable_rc6(dev); intel_setup_overlay(dev); } @@ -7485,12 +8462,11 @@ void intel_modeset_cleanup(struct drm_device *dev) intel_increase_pllclock(crtc); } - if (dev_priv->display.disable_fbc) - dev_priv->display.disable_fbc(dev); + intel_disable_fbc(dev); if (IS_IRONLAKE_M(dev)) ironlake_disable_drps(dev); - if (IS_GEN6(dev)) + if (IS_GEN6(dev) || IS_GEN7(dev)) gen6_disable_rps(dev); if (IS_IRONLAKE_M(dev)) @@ -7503,6 +8479,9 @@ void intel_modeset_cleanup(struct drm_device *dev) drm_irq_uninstall(dev); cancel_work_sync(&dev_priv->hotplug_work); + /* flush any delayed tasks or pending work */ + flush_scheduled_work(); + /* Shut off idle work before the crtcs get freed. */ list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { intel_crtc = to_intel_crtc(crtc); diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c index a4d80314e7f..44fef5e1c49 100644 --- a/drivers/gpu/drm/i915/intel_dp.c +++ b/drivers/gpu/drm/i915/intel_dp.c @@ -53,14 +53,12 @@ struct intel_dp { int dpms_mode; uint8_t link_bw; uint8_t lane_count; - uint8_t dpcd[4]; + uint8_t dpcd[8]; struct i2c_adapter adapter; struct i2c_algo_dp_aux_data algo; bool is_pch_edp; uint8_t train_set[4]; uint8_t link_status[DP_LINK_STATUS_SIZE]; - - struct drm_property *force_audio_property; }; /** @@ -140,8 +138,8 @@ intel_dp_max_lane_count(struct intel_dp *intel_dp) { int max_lane_count = 4; - if (intel_dp->dpcd[0] >= 0x11) { - max_lane_count = intel_dp->dpcd[2] & 0x1f; + if (intel_dp->dpcd[DP_DPCD_REV] >= 0x11) { + max_lane_count = intel_dp->dpcd[DP_MAX_LANE_COUNT] & 0x1f; switch (max_lane_count) { case 1: case 2: case 4: break; @@ -155,7 +153,7 @@ intel_dp_max_lane_count(struct intel_dp *intel_dp) static int intel_dp_max_link_bw(struct intel_dp *intel_dp) { - int max_link_bw = intel_dp->dpcd[1]; + int max_link_bw = intel_dp->dpcd[DP_MAX_LINK_RATE]; switch (max_link_bw) { case DP_LINK_BW_1_62: @@ -181,12 +179,14 @@ intel_dp_link_clock(uint8_t link_bw) static int intel_dp_link_required(struct drm_device *dev, struct intel_dp *intel_dp, int pixel_clock) { - struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc = intel_dp->base.base.crtc; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int bpp = 24; - if (is_edp(intel_dp)) - return (pixel_clock * dev_priv->edp.bpp + 7) / 8; - else - return pixel_clock * 3; + if (intel_crtc) + bpp = intel_crtc->bpp; + + return (pixel_clock * bpp + 7) / 8; } static int @@ -317,9 +317,17 @@ intel_dp_aux_ch(struct intel_dp *intel_dp, else precharge = 5; - if (I915_READ(ch_ctl) & DP_AUX_CH_CTL_SEND_BUSY) { - DRM_ERROR("dp_aux_ch not started status 0x%08x\n", - I915_READ(ch_ctl)); + /* Try to wait for any previous AUX channel activity */ + for (try = 0; try < 3; try++) { + status = I915_READ(ch_ctl); + if ((status & DP_AUX_CH_CTL_SEND_BUSY) == 0) + break; + msleep(1); + } + + if (try == 3) { + WARN(1, "dp_aux_ch not started status 0x%08x\n", + I915_READ(ch_ctl)); return -EBUSY; } @@ -684,7 +692,7 @@ intel_dp_set_m_n(struct drm_crtc *crtc, struct drm_display_mode *mode, struct drm_encoder *encoder; struct drm_i915_private *dev_priv = dev->dev_private; struct intel_crtc *intel_crtc = to_intel_crtc(crtc); - int lane_count = 4, bpp = 24; + int lane_count = 4; struct intel_dp_m_n m_n; int pipe = intel_crtc->pipe; @@ -703,7 +711,6 @@ intel_dp_set_m_n(struct drm_crtc *crtc, struct drm_display_mode *mode, break; } else if (is_edp(intel_dp)) { lane_count = dev_priv->edp.lanes; - bpp = dev_priv->edp.bpp; break; } } @@ -713,7 +720,7 @@ intel_dp_set_m_n(struct drm_crtc *crtc, struct drm_display_mode *mode, * the number of bytes_per_pixel post-LUT, which we always * set up for 8-bits of R/G/B, or 3 bytes total. */ - intel_dp_compute_m_n(bpp, lane_count, + intel_dp_compute_m_n(intel_crtc->bpp, lane_count, mode->clock, adjusted_mode->clock, &m_n); if (HAS_PCH_SPLIT(dev)) { @@ -772,11 +779,13 @@ intel_dp_mode_set(struct drm_encoder *encoder, struct drm_display_mode *mode, memset(intel_dp->link_configuration, 0, DP_LINK_CONFIGURATION_SIZE); intel_dp->link_configuration[0] = intel_dp->link_bw; intel_dp->link_configuration[1] = intel_dp->lane_count; + intel_dp->link_configuration[8] = DP_SET_ANSI_8B10B; /* * Check for DPCD version > 1.1 and enhanced framing support */ - if (intel_dp->dpcd[0] >= 0x11 && (intel_dp->dpcd[2] & DP_ENHANCED_FRAME_CAP)) { + if (intel_dp->dpcd[DP_DPCD_REV] >= 0x11 && + (intel_dp->dpcd[DP_MAX_LANE_COUNT] & DP_ENHANCED_FRAME_CAP)) { intel_dp->link_configuration[1] |= DP_LANE_COUNT_ENHANCED_FRAME_EN; intel_dp->DP |= DP_ENHANCED_FRAMING; } @@ -944,11 +953,44 @@ static void ironlake_edp_pll_off(struct drm_encoder *encoder) udelay(200); } +/* If the sink supports it, try to set the power state appropriately */ +static void intel_dp_sink_dpms(struct intel_dp *intel_dp, int mode) +{ + int ret, i; + + /* Should have a valid DPCD by this point */ + if (intel_dp->dpcd[DP_DPCD_REV] < 0x11) + return; + + if (mode != DRM_MODE_DPMS_ON) { + ret = intel_dp_aux_native_write_1(intel_dp, DP_SET_POWER, + DP_SET_POWER_D3); + if (ret != 1) + DRM_DEBUG_DRIVER("failed to write sink power state\n"); + } else { + /* + * When turning on, we need to retry for 1ms to give the sink + * time to wake up. + */ + for (i = 0; i < 3; i++) { + ret = intel_dp_aux_native_write_1(intel_dp, + DP_SET_POWER, + DP_SET_POWER_D0); + if (ret == 1) + break; + msleep(1); + } + } +} + static void intel_dp_prepare(struct drm_encoder *encoder) { struct intel_dp *intel_dp = enc_to_intel_dp(encoder); struct drm_device *dev = encoder->dev; + /* Wake up the sink first */ + intel_dp_sink_dpms(intel_dp, DRM_MODE_DPMS_ON); + if (is_edp(intel_dp)) { ironlake_edp_backlight_off(dev); ironlake_edp_panel_off(dev); @@ -979,6 +1021,8 @@ static void intel_dp_commit(struct drm_encoder *encoder) if (is_edp(intel_dp)) ironlake_edp_backlight_on(dev); + + intel_dp->dpms_mode = DRM_MODE_DPMS_ON; } static void @@ -992,6 +1036,7 @@ intel_dp_dpms(struct drm_encoder *encoder, int mode) if (mode != DRM_MODE_DPMS_ON) { if (is_edp(intel_dp)) ironlake_edp_backlight_off(dev); + intel_dp_sink_dpms(intel_dp, mode); intel_dp_link_down(intel_dp); if (is_edp(intel_dp)) ironlake_edp_panel_off(dev); @@ -1000,6 +1045,7 @@ intel_dp_dpms(struct drm_encoder *encoder, int mode) } else { if (is_edp(intel_dp)) ironlake_edp_panel_vdd_on(intel_dp); + intel_dp_sink_dpms(intel_dp, mode); if (!(dp_reg & DP_PORT_EN)) { intel_dp_start_link_train(intel_dp); if (is_edp(intel_dp)) { @@ -1015,20 +1061,41 @@ intel_dp_dpms(struct drm_encoder *encoder, int mode) } /* + * Native read with retry for link status and receiver capability reads for + * cases where the sink may still be asleep. + */ +static bool +intel_dp_aux_native_read_retry(struct intel_dp *intel_dp, uint16_t address, + uint8_t *recv, int recv_bytes) +{ + int ret, i; + + /* + * Sinks are *supposed* to come up within 1ms from an off state, + * but we're also supposed to retry 3 times per the spec. + */ + for (i = 0; i < 3; i++) { + ret = intel_dp_aux_native_read(intel_dp, address, recv, + recv_bytes); + if (ret == recv_bytes) + return true; + msleep(1); + } + + return false; +} + +/* * Fetch AUX CH registers 0x202 - 0x207 which contain * link status information */ static bool intel_dp_get_link_status(struct intel_dp *intel_dp) { - int ret; - - ret = intel_dp_aux_native_read(intel_dp, - DP_LANE0_1_STATUS, - intel_dp->link_status, DP_LINK_STATUS_SIZE); - if (ret != DP_LINK_STATUS_SIZE) - return false; - return true; + return intel_dp_aux_native_read_retry(intel_dp, + DP_LANE0_1_STATUS, + intel_dp->link_status, + DP_LINK_STATUS_SIZE); } static uint8_t @@ -1280,10 +1347,16 @@ intel_dp_start_link_train(struct intel_dp *intel_dp) u32 reg; uint32_t DP = intel_dp->DP; - /* Enable output, wait for it to become active */ - I915_WRITE(intel_dp->output_reg, intel_dp->DP); - POSTING_READ(intel_dp->output_reg); - intel_wait_for_vblank(dev, intel_crtc->pipe); + /* + * On CPT we have to enable the port in training pattern 1, which + * will happen below in intel_dp_set_link_train. Otherwise, enable + * the port and wait for it to become active. + */ + if (!HAS_PCH_CPT(dev)) { + I915_WRITE(intel_dp->output_reg, intel_dp->DP); + POSTING_READ(intel_dp->output_reg); + intel_wait_for_vblank(dev, intel_crtc->pipe); + } /* Write the link configuration data */ intel_dp_aux_native_write(intel_dp, DP_LINK_BW_SET, @@ -1316,7 +1389,8 @@ intel_dp_start_link_train(struct intel_dp *intel_dp) reg = DP | DP_LINK_TRAIN_PAT_1; if (!intel_dp_set_link_train(intel_dp, reg, - DP_TRAINING_PATTERN_1)) + DP_TRAINING_PATTERN_1 | + DP_LINK_SCRAMBLING_DISABLE)) break; /* Set training pattern 1 */ @@ -1391,7 +1465,8 @@ intel_dp_complete_link_train(struct intel_dp *intel_dp) /* channel eq pattern */ if (!intel_dp_set_link_train(intel_dp, reg, - DP_TRAINING_PATTERN_2)) + DP_TRAINING_PATTERN_2 | + DP_LINK_SCRAMBLING_DISABLE)) break; udelay(400); @@ -1505,6 +1580,18 @@ intel_dp_link_down(struct intel_dp *intel_dp) POSTING_READ(intel_dp->output_reg); } +static bool +intel_dp_get_dpcd(struct intel_dp *intel_dp) +{ + if (intel_dp_aux_native_read_retry(intel_dp, 0x000, intel_dp->dpcd, + sizeof (intel_dp->dpcd)) && + (intel_dp->dpcd[DP_DPCD_REV] != 0)) { + return true; + } + + return false; +} + /* * According to DP spec * 5.1.2: @@ -1517,21 +1604,41 @@ intel_dp_link_down(struct intel_dp *intel_dp) static void intel_dp_check_link_status(struct intel_dp *intel_dp) { + if (intel_dp->dpms_mode != DRM_MODE_DPMS_ON) + return; + if (!intel_dp->base.base.crtc) return; + /* Try to read receiver status if the link appears to be up */ if (!intel_dp_get_link_status(intel_dp)) { intel_dp_link_down(intel_dp); return; } + /* Now read the DPCD to see if it's actually running */ + if (!intel_dp_get_dpcd(intel_dp)) { + intel_dp_link_down(intel_dp); + return; + } + if (!intel_channel_eq_ok(intel_dp)) { + DRM_DEBUG_KMS("%s: channel EQ not ok, retraining\n", + drm_get_encoder_name(&intel_dp->base.base)); intel_dp_start_link_train(intel_dp); intel_dp_complete_link_train(intel_dp); } } static enum drm_connector_status +intel_dp_detect_dpcd(struct intel_dp *intel_dp) +{ + if (intel_dp_get_dpcd(intel_dp)) + return connector_status_connected; + return connector_status_disconnected; +} + +static enum drm_connector_status ironlake_dp_detect(struct intel_dp *intel_dp) { enum drm_connector_status status; @@ -1544,17 +1651,7 @@ ironlake_dp_detect(struct intel_dp *intel_dp) return status; } - status = connector_status_disconnected; - if (intel_dp_aux_native_read(intel_dp, - 0x000, intel_dp->dpcd, - sizeof (intel_dp->dpcd)) - == sizeof(intel_dp->dpcd)) { - if (intel_dp->dpcd[0] != 0) - status = connector_status_connected; - } - DRM_DEBUG_KMS("DPCD: %hx%hx%hx%hx\n", intel_dp->dpcd[0], - intel_dp->dpcd[1], intel_dp->dpcd[2], intel_dp->dpcd[3]); - return status; + return intel_dp_detect_dpcd(intel_dp); } static enum drm_connector_status @@ -1562,7 +1659,6 @@ g4x_dp_detect(struct intel_dp *intel_dp) { struct drm_device *dev = intel_dp->base.base.dev; struct drm_i915_private *dev_priv = dev->dev_private; - enum drm_connector_status status; uint32_t temp, bit; switch (intel_dp->output_reg) { @@ -1584,15 +1680,7 @@ g4x_dp_detect(struct intel_dp *intel_dp) if ((temp & bit) == 0) return connector_status_disconnected; - status = connector_status_disconnected; - if (intel_dp_aux_native_read(intel_dp, 0x000, intel_dp->dpcd, - sizeof (intel_dp->dpcd)) == sizeof (intel_dp->dpcd)) - { - if (intel_dp->dpcd[0] != 0) - status = connector_status_connected; - } - - return status; + return intel_dp_detect_dpcd(intel_dp); } /** @@ -1615,6 +1703,12 @@ intel_dp_detect(struct drm_connector *connector, bool force) status = ironlake_dp_detect(intel_dp); else status = g4x_dp_detect(intel_dp); + + DRM_DEBUG_KMS("DPCD: %02hx%02hx%02hx%02hx%02hx%02hx%02hx%02hx\n", + intel_dp->dpcd[0], intel_dp->dpcd[1], intel_dp->dpcd[2], + intel_dp->dpcd[3], intel_dp->dpcd[4], intel_dp->dpcd[5], + intel_dp->dpcd[6], intel_dp->dpcd[7]); + if (status != connector_status_connected) return status; @@ -1702,7 +1796,7 @@ intel_dp_set_property(struct drm_connector *connector, if (ret) return ret; - if (property == intel_dp->force_audio_property) { + if (property == dev_priv->force_audio_property) { int i = val; bool has_audio; @@ -1747,6 +1841,11 @@ done: static void intel_dp_destroy (struct drm_connector *connector) { + struct drm_device *dev = connector->dev; + + if (intel_dpd_is_edp(dev)) + intel_panel_destroy_backlight(dev); + drm_sysfs_connector_remove(connector); drm_connector_cleanup(connector); kfree(connector); @@ -1792,8 +1891,7 @@ intel_dp_hot_plug(struct intel_encoder *intel_encoder) { struct intel_dp *intel_dp = container_of(intel_encoder, struct intel_dp, base); - if (intel_dp->dpms_mode == DRM_MODE_DPMS_ON) - intel_dp_check_link_status(intel_dp); + intel_dp_check_link_status(intel_dp); } /* Return which DP Port should be selected for Transcoder DP control */ @@ -1841,16 +1939,7 @@ bool intel_dpd_is_edp(struct drm_device *dev) static void intel_dp_add_properties(struct intel_dp *intel_dp, struct drm_connector *connector) { - struct drm_device *dev = connector->dev; - - intel_dp->force_audio_property = - drm_property_create(dev, DRM_MODE_PROP_RANGE, "force_audio", 2); - if (intel_dp->force_audio_property) { - intel_dp->force_audio_property->values[0] = -1; - intel_dp->force_audio_property->values[1] = 1; - drm_connector_attach_property(connector, intel_dp->force_audio_property, 0); - } - + intel_attach_force_audio_property(connector); intel_attach_broadcast_rgb_property(connector); } @@ -1947,7 +2036,7 @@ intel_dp_init(struct drm_device *dev, int output_reg) /* Cache some DPCD data in the eDP case */ if (is_edp(intel_dp)) { - int ret; + bool ret; u32 pp_on, pp_div; pp_on = I915_READ(PCH_PP_ON_DELAYS); @@ -1960,13 +2049,12 @@ intel_dp_init(struct drm_device *dev, int output_reg) dev_priv->panel_t12 *= 100; /* t12 in 100ms units */ ironlake_edp_panel_vdd_on(intel_dp); - ret = intel_dp_aux_native_read(intel_dp, DP_DPCD_REV, - intel_dp->dpcd, - sizeof(intel_dp->dpcd)); + ret = intel_dp_get_dpcd(intel_dp); ironlake_edp_panel_vdd_off(intel_dp); - if (ret == sizeof(intel_dp->dpcd)) { - if (intel_dp->dpcd[0] >= 0x11) - dev_priv->no_aux_handshake = intel_dp->dpcd[3] & + if (ret) { + if (intel_dp->dpcd[DP_DPCD_REV] >= 0x11) + dev_priv->no_aux_handshake = + intel_dp->dpcd[DP_MAX_DOWNSPREAD] & DP_NO_AUX_HANDSHAKE_LINK_TRAINING; } else { /* if this fails, presume the device is a ghost */ @@ -1989,6 +2077,8 @@ intel_dp_init(struct drm_device *dev, int output_reg) DRM_MODE_TYPE_PREFERRED; } } + dev_priv->int_edp_connector = connector; + intel_panel_setup_backlight(dev); } intel_dp_add_properties(intel_dp, connector); diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h index 1d20712d527..0b2ee9d3998 100644 --- a/drivers/gpu/drm/i915/intel_drv.h +++ b/drivers/gpu/drm/i915/intel_drv.h @@ -140,7 +140,6 @@ struct intel_fbdev { struct intel_encoder { struct drm_encoder base; int type; - bool load_detect_temp; bool needs_tv_clock; void (*hot_plug)(struct intel_encoder *); int crtc_mask; @@ -171,6 +170,7 @@ struct intel_crtc { int16_t cursor_x, cursor_y; int16_t cursor_width, cursor_height; bool cursor_visible; + unsigned int bpp; }; #define to_intel_crtc(x) container_of(x, struct intel_crtc, base) @@ -178,10 +178,28 @@ struct intel_crtc { #define to_intel_encoder(x) container_of(x, struct intel_encoder, base) #define to_intel_framebuffer(x) container_of(x, struct intel_framebuffer, base) +#define DIP_HEADER_SIZE 5 + #define DIP_TYPE_AVI 0x82 #define DIP_VERSION_AVI 0x2 #define DIP_LEN_AVI 13 +#define DIP_TYPE_SPD 0x3 +#define DIP_VERSION_SPD 0x1 +#define DIP_LEN_SPD 25 +#define DIP_SPD_UNKNOWN 0 +#define DIP_SPD_DSTB 0x1 +#define DIP_SPD_DVDP 0x2 +#define DIP_SPD_DVHS 0x3 +#define DIP_SPD_HDDVR 0x4 +#define DIP_SPD_DVC 0x5 +#define DIP_SPD_DSC 0x6 +#define DIP_SPD_VCD 0x7 +#define DIP_SPD_GAME 0x8 +#define DIP_SPD_PC 0x9 +#define DIP_SPD_BD 0xa +#define DIP_SPD_SCD 0xb + struct dip_infoframe { uint8_t type; /* HB0 */ uint8_t ver; /* HB1 */ @@ -206,6 +224,11 @@ struct dip_infoframe { uint16_t left_bar_end; uint16_t right_bar_start; } avi; + struct { + uint8_t vn[8]; + uint8_t pd[16]; + uint8_t sdi; + } spd; uint8_t payload[27]; } __attribute__ ((packed)) body; } __attribute__((packed)); @@ -234,9 +257,17 @@ struct intel_unpin_work { bool enable_stall_check; }; +struct intel_fbc_work { + struct delayed_work work; + struct drm_crtc *crtc; + struct drm_framebuffer *fb; + int interval; +}; + int intel_ddc_get_modes(struct drm_connector *c, struct i2c_adapter *adapter); extern bool intel_ddc_probe(struct intel_encoder *intel_encoder, int ddc_bus); +extern void intel_attach_force_audio_property(struct drm_connector *connector); extern void intel_attach_broadcast_rgb_property(struct drm_connector *connector); extern void intel_crt_init(struct drm_device *dev); @@ -266,9 +297,10 @@ extern void intel_pch_panel_fitting(struct drm_device *dev, extern u32 intel_panel_get_max_backlight(struct drm_device *dev); extern u32 intel_panel_get_backlight(struct drm_device *dev); extern void intel_panel_set_backlight(struct drm_device *dev, u32 level); -extern void intel_panel_setup_backlight(struct drm_device *dev); +extern int intel_panel_setup_backlight(struct drm_device *dev); extern void intel_panel_enable_backlight(struct drm_device *dev); extern void intel_panel_disable_backlight(struct drm_device *dev); +extern void intel_panel_destroy_backlight(struct drm_device *dev); extern enum drm_connector_status intel_panel_detect(struct drm_device *dev); extern void intel_crtc_load_lut(struct drm_crtc *crtc); @@ -291,13 +323,19 @@ int intel_get_pipe_from_crtc_id(struct drm_device *dev, void *data, struct drm_file *file_priv); extern void intel_wait_for_vblank(struct drm_device *dev, int pipe); extern void intel_wait_for_pipe_off(struct drm_device *dev, int pipe); -extern struct drm_crtc *intel_get_load_detect_pipe(struct intel_encoder *intel_encoder, - struct drm_connector *connector, - struct drm_display_mode *mode, - int *dpms_mode); + +struct intel_load_detect_pipe { + struct drm_framebuffer *release_fb; + bool load_detect_temp; + int dpms_mode; +}; +extern bool intel_get_load_detect_pipe(struct intel_encoder *intel_encoder, + struct drm_connector *connector, + struct drm_display_mode *mode, + struct intel_load_detect_pipe *old); extern void intel_release_load_detect_pipe(struct intel_encoder *intel_encoder, struct drm_connector *connector, - int dpms_mode); + struct intel_load_detect_pipe *old); extern struct drm_connector* intel_sdvo_find(struct drm_device *dev, int sdvoB); extern int intel_sdvo_supports_hotplug(struct drm_connector *connector); @@ -311,6 +349,7 @@ extern void intel_enable_clock_gating(struct drm_device *dev); extern void ironlake_enable_drps(struct drm_device *dev); extern void ironlake_disable_drps(struct drm_device *dev); extern void gen6_enable_rps(struct drm_i915_private *dev_priv); +extern void gen6_update_ring_freq(struct drm_i915_private *dev_priv); extern void gen6_disable_rps(struct drm_device *dev); extern void intel_init_emon(struct drm_device *dev); @@ -339,4 +378,6 @@ extern int intel_overlay_attrs(struct drm_device *dev, void *data, extern void intel_fb_output_poll_changed(struct drm_device *dev); extern void intel_fb_restore_mode(struct drm_device *dev); + +extern void intel_init_clock_gating(struct drm_device *dev); #endif /* __INTEL_DRV_H__ */ diff --git a/drivers/gpu/drm/i915/intel_hdmi.c b/drivers/gpu/drm/i915/intel_hdmi.c index f289b864297..226ba830f38 100644 --- a/drivers/gpu/drm/i915/intel_hdmi.c +++ b/drivers/gpu/drm/i915/intel_hdmi.c @@ -45,7 +45,8 @@ struct intel_hdmi { bool has_hdmi_sink; bool has_audio; int force_audio; - struct drm_property *force_audio_property; + void (*write_infoframe)(struct drm_encoder *encoder, + struct dip_infoframe *frame); }; static struct intel_hdmi *enc_to_intel_hdmi(struct drm_encoder *encoder) @@ -59,37 +60,70 @@ static struct intel_hdmi *intel_attached_hdmi(struct drm_connector *connector) struct intel_hdmi, base); } -void intel_dip_infoframe_csum(struct dip_infoframe *avi_if) +void intel_dip_infoframe_csum(struct dip_infoframe *frame) { - uint8_t *data = (uint8_t *)avi_if; + uint8_t *data = (uint8_t *)frame; uint8_t sum = 0; unsigned i; - avi_if->checksum = 0; - avi_if->ecc = 0; + frame->checksum = 0; + frame->ecc = 0; - for (i = 0; i < sizeof(*avi_if); i++) + /* Header isn't part of the checksum */ + for (i = 5; i < frame->len; i++) sum += data[i]; - avi_if->checksum = 0x100 - sum; + frame->checksum = 0x100 - sum; } -static void intel_hdmi_set_avi_infoframe(struct drm_encoder *encoder) +static u32 intel_infoframe_index(struct dip_infoframe *frame) { - struct dip_infoframe avi_if = { - .type = DIP_TYPE_AVI, - .ver = DIP_VERSION_AVI, - .len = DIP_LEN_AVI, - }; - uint32_t *data = (uint32_t *)&avi_if; + u32 flags = 0; + + switch (frame->type) { + case DIP_TYPE_AVI: + flags |= VIDEO_DIP_SELECT_AVI; + break; + case DIP_TYPE_SPD: + flags |= VIDEO_DIP_SELECT_SPD; + break; + default: + DRM_DEBUG_DRIVER("unknown info frame type %d\n", frame->type); + break; + } + + return flags; +} + +static u32 intel_infoframe_flags(struct dip_infoframe *frame) +{ + u32 flags = 0; + + switch (frame->type) { + case DIP_TYPE_AVI: + flags |= VIDEO_DIP_ENABLE_AVI | VIDEO_DIP_FREQ_VSYNC; + break; + case DIP_TYPE_SPD: + flags |= VIDEO_DIP_ENABLE_SPD | VIDEO_DIP_FREQ_2VSYNC; + break; + default: + DRM_DEBUG_DRIVER("unknown info frame type %d\n", frame->type); + break; + } + + return flags; +} + +static void i9xx_write_infoframe(struct drm_encoder *encoder, + struct dip_infoframe *frame) +{ + uint32_t *data = (uint32_t *)frame; struct drm_device *dev = encoder->dev; struct drm_i915_private *dev_priv = dev->dev_private; struct intel_hdmi *intel_hdmi = enc_to_intel_hdmi(encoder); - u32 port; - unsigned i; + u32 port, flags, val = I915_READ(VIDEO_DIP_CTL); + unsigned i, len = DIP_HEADER_SIZE + frame->len; - if (!intel_hdmi->has_hdmi_sink) - return; /* XXX first guess at handling video port, is this corrent? */ if (intel_hdmi->sdvox_reg == SDVOB) @@ -99,18 +133,87 @@ static void intel_hdmi_set_avi_infoframe(struct drm_encoder *encoder) else return; - I915_WRITE(VIDEO_DIP_CTL, VIDEO_DIP_ENABLE | port | - VIDEO_DIP_SELECT_AVI | VIDEO_DIP_FREQ_VSYNC); + flags = intel_infoframe_index(frame); + + val &= ~VIDEO_DIP_SELECT_MASK; - intel_dip_infoframe_csum(&avi_if); - for (i = 0; i < sizeof(avi_if); i += 4) { + I915_WRITE(VIDEO_DIP_CTL, val | port | flags); + + for (i = 0; i < len; i += 4) { I915_WRITE(VIDEO_DIP_DATA, *data); data++; } - I915_WRITE(VIDEO_DIP_CTL, VIDEO_DIP_ENABLE | port | - VIDEO_DIP_SELECT_AVI | VIDEO_DIP_FREQ_VSYNC | - VIDEO_DIP_ENABLE_AVI); + flags |= intel_infoframe_flags(frame); + + I915_WRITE(VIDEO_DIP_CTL, VIDEO_DIP_ENABLE | val | port | flags); +} + +static void ironlake_write_infoframe(struct drm_encoder *encoder, + struct dip_infoframe *frame) +{ + uint32_t *data = (uint32_t *)frame; + struct drm_device *dev = encoder->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_crtc *crtc = encoder->crtc; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); + int reg = TVIDEO_DIP_CTL(intel_crtc->pipe); + unsigned i, len = DIP_HEADER_SIZE + frame->len; + u32 flags, val = I915_READ(reg); + + intel_wait_for_vblank(dev, intel_crtc->pipe); + + flags = intel_infoframe_index(frame); + + val &= ~VIDEO_DIP_SELECT_MASK; + + I915_WRITE(reg, val | flags); + + for (i = 0; i < len; i += 4) { + I915_WRITE(TVIDEO_DIP_DATA(intel_crtc->pipe), *data); + data++; + } + + flags |= intel_infoframe_flags(frame); + + I915_WRITE(reg, VIDEO_DIP_ENABLE | val | flags); +} +static void intel_set_infoframe(struct drm_encoder *encoder, + struct dip_infoframe *frame) +{ + struct intel_hdmi *intel_hdmi = enc_to_intel_hdmi(encoder); + + if (!intel_hdmi->has_hdmi_sink) + return; + + intel_dip_infoframe_csum(frame); + intel_hdmi->write_infoframe(encoder, frame); +} + +static void intel_hdmi_set_avi_infoframe(struct drm_encoder *encoder) +{ + struct dip_infoframe avi_if = { + .type = DIP_TYPE_AVI, + .ver = DIP_VERSION_AVI, + .len = DIP_LEN_AVI, + }; + + intel_set_infoframe(encoder, &avi_if); +} + +static void intel_hdmi_set_spd_infoframe(struct drm_encoder *encoder) +{ + struct dip_infoframe spd_if; + + memset(&spd_if, 0, sizeof(spd_if)); + spd_if.type = DIP_TYPE_SPD; + spd_if.ver = DIP_VERSION_SPD; + spd_if.len = DIP_LEN_SPD; + strcpy(spd_if.body.spd.vn, "Intel"); + strcpy(spd_if.body.spd.pd, "Integrated gfx"); + spd_if.body.spd.sdi = DIP_SPD_PC; + + intel_set_infoframe(encoder, &spd_if); } static void intel_hdmi_mode_set(struct drm_encoder *encoder, @@ -125,12 +228,18 @@ static void intel_hdmi_mode_set(struct drm_encoder *encoder, u32 sdvox; sdvox = SDVO_ENCODING_HDMI | SDVO_BORDER_ENABLE; - sdvox |= intel_hdmi->color_range; + if (!HAS_PCH_SPLIT(dev)) + sdvox |= intel_hdmi->color_range; if (adjusted_mode->flags & DRM_MODE_FLAG_PVSYNC) sdvox |= SDVO_VSYNC_ACTIVE_HIGH; if (adjusted_mode->flags & DRM_MODE_FLAG_PHSYNC) sdvox |= SDVO_HSYNC_ACTIVE_HIGH; + if (intel_crtc->bpp > 24) + sdvox |= COLOR_FORMAT_12bpc; + else + sdvox |= COLOR_FORMAT_8bpc; + /* Required on CPT */ if (intel_hdmi->has_hdmi_sink && HAS_PCH_CPT(dev)) sdvox |= HDMI_MODE_SELECT; @@ -151,6 +260,7 @@ static void intel_hdmi_mode_set(struct drm_encoder *encoder, POSTING_READ(intel_hdmi->sdvox_reg); intel_hdmi_set_avi_infoframe(encoder); + intel_hdmi_set_spd_infoframe(encoder); } static void intel_hdmi_dpms(struct drm_encoder *encoder, int mode) @@ -194,7 +304,7 @@ static int intel_hdmi_mode_valid(struct drm_connector *connector, if (mode->clock > 165000) return MODE_CLOCK_HIGH; if (mode->clock < 20000) - return MODE_CLOCK_HIGH; + return MODE_CLOCK_LOW; if (mode->flags & DRM_MODE_FLAG_DBLSCAN) return MODE_NO_DBLESCAN; @@ -287,7 +397,7 @@ intel_hdmi_set_property(struct drm_connector *connector, if (ret) return ret; - if (property == intel_hdmi->force_audio_property) { + if (property == dev_priv->force_audio_property) { int i = val; bool has_audio; @@ -365,16 +475,7 @@ static const struct drm_encoder_funcs intel_hdmi_enc_funcs = { static void intel_hdmi_add_properties(struct intel_hdmi *intel_hdmi, struct drm_connector *connector) { - struct drm_device *dev = connector->dev; - - intel_hdmi->force_audio_property = - drm_property_create(dev, DRM_MODE_PROP_RANGE, "force_audio", 2); - if (intel_hdmi->force_audio_property) { - intel_hdmi->force_audio_property->values[0] = -1; - intel_hdmi->force_audio_property->values[1] = 1; - drm_connector_attach_property(connector, intel_hdmi->force_audio_property, 0); - } - + intel_attach_force_audio_property(connector); intel_attach_broadcast_rgb_property(connector); } @@ -437,6 +538,11 @@ void intel_hdmi_init(struct drm_device *dev, int sdvox_reg) intel_hdmi->sdvox_reg = sdvox_reg; + if (!HAS_PCH_SPLIT(dev)) + intel_hdmi->write_infoframe = i9xx_write_infoframe; + else + intel_hdmi->write_infoframe = ironlake_write_infoframe; + drm_encoder_helper_add(&intel_encoder->base, &intel_hdmi_helper_funcs); intel_hdmi_add_properties(intel_hdmi, connector); diff --git a/drivers/gpu/drm/i915/intel_i2c.c b/drivers/gpu/drm/i915/intel_i2c.c index d3b903bce7c..d98cee60b60 100644 --- a/drivers/gpu/drm/i915/intel_i2c.c +++ b/drivers/gpu/drm/i915/intel_i2c.c @@ -401,8 +401,7 @@ int intel_setup_gmbus(struct drm_device *dev) bus->reg0 = i | GMBUS_RATE_100KHZ; /* XXX force bit banging until GMBUS is fully debugged */ - if (IS_GEN2(dev)) - bus->force_bit = intel_gpio_create(dev_priv, i); + bus->force_bit = intel_gpio_create(dev_priv, i); } intel_i2c_reset(dev_priv->dev); diff --git a/drivers/gpu/drm/i915/intel_lvds.c b/drivers/gpu/drm/i915/intel_lvds.c index 67cb076d271..31da77f5c05 100644 --- a/drivers/gpu/drm/i915/intel_lvds.c +++ b/drivers/gpu/drm/i915/intel_lvds.c @@ -72,14 +72,16 @@ static void intel_lvds_enable(struct intel_lvds *intel_lvds) { struct drm_device *dev = intel_lvds->base.base.dev; struct drm_i915_private *dev_priv = dev->dev_private; - u32 ctl_reg, lvds_reg; + u32 ctl_reg, lvds_reg, stat_reg; if (HAS_PCH_SPLIT(dev)) { ctl_reg = PCH_PP_CONTROL; lvds_reg = PCH_LVDS; + stat_reg = PCH_PP_STATUS; } else { ctl_reg = PP_CONTROL; lvds_reg = LVDS; + stat_reg = PP_STATUS; } I915_WRITE(lvds_reg, I915_READ(lvds_reg) | LVDS_PORT_EN); @@ -94,17 +96,16 @@ static void intel_lvds_enable(struct intel_lvds *intel_lvds) DRM_DEBUG_KMS("applying panel-fitter: %x, %x\n", intel_lvds->pfit_control, intel_lvds->pfit_pgm_ratios); - if (wait_for((I915_READ(PP_STATUS) & PP_ON) == 0, 1000)) { - DRM_ERROR("timed out waiting for panel to power off\n"); - } else { - I915_WRITE(PFIT_PGM_RATIOS, intel_lvds->pfit_pgm_ratios); - I915_WRITE(PFIT_CONTROL, intel_lvds->pfit_control); - intel_lvds->pfit_dirty = false; - } + + I915_WRITE(PFIT_PGM_RATIOS, intel_lvds->pfit_pgm_ratios); + I915_WRITE(PFIT_CONTROL, intel_lvds->pfit_control); + intel_lvds->pfit_dirty = false; } I915_WRITE(ctl_reg, I915_READ(ctl_reg) | POWER_TARGET_ON); POSTING_READ(lvds_reg); + if (wait_for((I915_READ(stat_reg) & PP_ON) != 0, 1000)) + DRM_ERROR("timed out waiting for panel to power on\n"); intel_panel_enable_backlight(dev); } @@ -113,24 +114,25 @@ static void intel_lvds_disable(struct intel_lvds *intel_lvds) { struct drm_device *dev = intel_lvds->base.base.dev; struct drm_i915_private *dev_priv = dev->dev_private; - u32 ctl_reg, lvds_reg; + u32 ctl_reg, lvds_reg, stat_reg; if (HAS_PCH_SPLIT(dev)) { ctl_reg = PCH_PP_CONTROL; lvds_reg = PCH_LVDS; + stat_reg = PCH_PP_STATUS; } else { ctl_reg = PP_CONTROL; lvds_reg = LVDS; + stat_reg = PP_STATUS; } intel_panel_disable_backlight(dev); I915_WRITE(ctl_reg, I915_READ(ctl_reg) & ~POWER_TARGET_ON); + if (wait_for((I915_READ(stat_reg) & PP_ON) == 0, 1000)) + DRM_ERROR("timed out waiting for panel to power off\n"); if (intel_lvds->pfit_control) { - if (wait_for((I915_READ(PP_STATUS) & PP_ON) == 0, 1000)) - DRM_ERROR("timed out waiting for panel to power off\n"); - I915_WRITE(PFIT_CONTROL, 0); intel_lvds->pfit_dirty = true; } @@ -398,53 +400,21 @@ out: static void intel_lvds_prepare(struct drm_encoder *encoder) { - struct drm_device *dev = encoder->dev; - struct drm_i915_private *dev_priv = dev->dev_private; struct intel_lvds *intel_lvds = to_intel_lvds(encoder); - /* We try to do the minimum that is necessary in order to unlock - * the registers for mode setting. - * - * On Ironlake, this is quite simple as we just set the unlock key - * and ignore all subtleties. (This may cause some issues...) - * + /* * Prior to Ironlake, we must disable the pipe if we want to adjust * the panel fitter. However at all other times we can just reset * the registers regardless. */ - - if (HAS_PCH_SPLIT(dev)) { - I915_WRITE(PCH_PP_CONTROL, - I915_READ(PCH_PP_CONTROL) | PANEL_UNLOCK_REGS); - } else if (intel_lvds->pfit_dirty) { - I915_WRITE(PP_CONTROL, - (I915_READ(PP_CONTROL) | PANEL_UNLOCK_REGS) - & ~POWER_TARGET_ON); - } else { - I915_WRITE(PP_CONTROL, - I915_READ(PP_CONTROL) | PANEL_UNLOCK_REGS); - } + if (!HAS_PCH_SPLIT(encoder->dev) && intel_lvds->pfit_dirty) + intel_lvds_disable(intel_lvds); } static void intel_lvds_commit(struct drm_encoder *encoder) { - struct drm_device *dev = encoder->dev; - struct drm_i915_private *dev_priv = dev->dev_private; struct intel_lvds *intel_lvds = to_intel_lvds(encoder); - /* Undo any unlocking done in prepare to prevent accidental - * adjustment of the registers. - */ - if (HAS_PCH_SPLIT(dev)) { - u32 val = I915_READ(PCH_PP_CONTROL); - if ((val & PANEL_UNLOCK_REGS) == PANEL_UNLOCK_REGS) - I915_WRITE(PCH_PP_CONTROL, val & 0x3); - } else { - u32 val = I915_READ(PP_CONTROL); - if ((val & PANEL_UNLOCK_REGS) == PANEL_UNLOCK_REGS) - I915_WRITE(PP_CONTROL, val & 0x3); - } - /* Always do a full power on as we do not know what state * we were left in. */ @@ -582,6 +552,8 @@ static void intel_lvds_destroy(struct drm_connector *connector) struct drm_device *dev = connector->dev; struct drm_i915_private *dev_priv = dev->dev_private; + intel_panel_destroy_backlight(dev); + if (dev_priv->lid_notifier.notifier_call) acpi_lid_notifier_unregister(&dev_priv->lid_notifier); drm_sysfs_connector_remove(connector); @@ -690,6 +662,14 @@ static const struct dmi_system_id intel_no_lvds[] = { }, { .callback = intel_no_lvds_dmi_callback, + .ident = "Dell OptiPlex FX170", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex FX170"), + }, + }, + { + .callback = intel_no_lvds_dmi_callback, .ident = "AOpen Mini PC", .matches = { DMI_MATCH(DMI_SYS_VENDOR, "AOpen"), @@ -727,6 +707,14 @@ static const struct dmi_system_id intel_no_lvds[] = { DMI_MATCH(DMI_PRODUCT_NAME, "U800"), }, }, + { + .callback = intel_no_lvds_dmi_callback, + .ident = "Asus EeeBox PC EB1007", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK Computer INC."), + DMI_MATCH(DMI_PRODUCT_NAME, "EB1007"), + }, + }, { } /* terminating entry */ }; @@ -1024,6 +1012,19 @@ out: pwm = I915_READ(BLC_PWM_PCH_CTL1); pwm |= PWM_PCH_ENABLE; I915_WRITE(BLC_PWM_PCH_CTL1, pwm); + /* + * Unlock registers and just + * leave them unlocked + */ + I915_WRITE(PCH_PP_CONTROL, + I915_READ(PCH_PP_CONTROL) | PANEL_UNLOCK_REGS); + } else { + /* + * Unlock registers and just + * leave them unlocked + */ + I915_WRITE(PP_CONTROL, + I915_READ(PP_CONTROL) | PANEL_UNLOCK_REGS); } dev_priv->lid_notifier.notifier_call = intel_lid_notify; if (acpi_lid_notifier_register(&dev_priv->lid_notifier)) { @@ -1033,6 +1034,9 @@ out: /* keep the LVDS connector */ dev_priv->int_lvds_connector = connector; drm_sysfs_connector_add(connector); + + intel_panel_setup_backlight(dev); + return true; failed: diff --git a/drivers/gpu/drm/i915/intel_modes.c b/drivers/gpu/drm/i915/intel_modes.c index 9034dd8f33c..3b26a3ba02d 100644 --- a/drivers/gpu/drm/i915/intel_modes.c +++ b/drivers/gpu/drm/i915/intel_modes.c @@ -81,6 +81,36 @@ int intel_ddc_get_modes(struct drm_connector *connector, return ret; } +static const char *force_audio_names[] = { + "off", + "auto", + "on", +}; + +void +intel_attach_force_audio_property(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct drm_property *prop; + int i; + + prop = dev_priv->force_audio_property; + if (prop == NULL) { + prop = drm_property_create(dev, DRM_MODE_PROP_ENUM, + "audio", + ARRAY_SIZE(force_audio_names)); + if (prop == NULL) + return; + + for (i = 0; i < ARRAY_SIZE(force_audio_names); i++) + drm_property_add_enum(prop, i, i-1, force_audio_names[i]); + + dev_priv->force_audio_property = prop; + } + drm_connector_attach_property(connector, prop, 0); +} + static const char *broadcast_rgb_names[] = { "Full", "Limited 16:235", diff --git a/drivers/gpu/drm/i915/intel_opregion.c b/drivers/gpu/drm/i915/intel_opregion.c index d2c71042290..b8e8158bb16 100644 --- a/drivers/gpu/drm/i915/intel_opregion.c +++ b/drivers/gpu/drm/i915/intel_opregion.c @@ -227,7 +227,6 @@ void intel_opregion_asle_intr(struct drm_device *dev) asle->aslc = asle_stat; } -/* Only present on Ironlake+ */ void intel_opregion_gse_intr(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -297,19 +296,26 @@ static int intel_opregion_video_event(struct notifier_block *nb, /* The only video events relevant to opregion are 0x80. These indicate either a docking event, lid switch or display switch request. In Linux, these are handled by the dock, button and video drivers. - We might want to fix the video driver to be opregion-aware in - future, but right now we just indicate to the firmware that the - request has been handled */ + */ struct opregion_acpi *acpi; + struct acpi_bus_event *event = data; + int ret = NOTIFY_OK; + + if (strcmp(event->device_class, ACPI_VIDEO_CLASS) != 0) + return NOTIFY_DONE; if (!system_opregion) return NOTIFY_DONE; acpi = system_opregion->acpi; + + if (event->type == 0x80 && !(acpi->cevt & 0x1)) + ret = NOTIFY_BAD; + acpi->csts = 0; - return NOTIFY_OK; + return ret; } static struct notifier_block intel_opregion_notifier = { diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c index a670c006982..d3603808682 100644 --- a/drivers/gpu/drm/i915/intel_overlay.c +++ b/drivers/gpu/drm/i915/intel_overlay.c @@ -773,14 +773,10 @@ static int intel_overlay_do_put_image(struct intel_overlay *overlay, if (ret != 0) return ret; - ret = i915_gem_object_pin(new_bo, PAGE_SIZE, true); + ret = i915_gem_object_pin_to_display_plane(new_bo, 0, NULL); if (ret != 0) return ret; - ret = i915_gem_object_set_to_gtt_domain(new_bo, 0); - if (ret != 0) - goto out_unpin; - ret = i915_gem_object_put_fence(new_bo); if (ret) goto out_unpin; @@ -1409,6 +1405,11 @@ void intel_setup_overlay(struct drm_device *dev) overlay = kzalloc(sizeof(struct intel_overlay), GFP_KERNEL); if (!overlay) return; + + mutex_lock(&dev->struct_mutex); + if (WARN_ON(dev_priv->overlay)) + goto out_free; + overlay->dev = dev; reg_bo = i915_gem_alloc_object(dev, PAGE_SIZE); @@ -1448,7 +1449,7 @@ void intel_setup_overlay(struct drm_device *dev) regs = intel_overlay_map_regs(overlay); if (!regs) - goto out_free_bo; + goto out_unpin_bo; memset(regs, 0, sizeof(struct overlay_registers)); update_polyphase_filter(regs); @@ -1457,14 +1458,17 @@ void intel_setup_overlay(struct drm_device *dev) intel_overlay_unmap_regs(overlay, regs); dev_priv->overlay = overlay; + mutex_unlock(&dev->struct_mutex); DRM_INFO("initialized overlay support\n"); return; out_unpin_bo: - i915_gem_object_unpin(reg_bo); + if (!OVERLAY_NEEDS_PHYSICAL(dev)) + i915_gem_object_unpin(reg_bo); out_free_bo: drm_gem_object_unreference(®_bo->base); out_free: + mutex_unlock(&dev->struct_mutex); kfree(overlay); return; } diff --git a/drivers/gpu/drm/i915/intel_panel.c b/drivers/gpu/drm/i915/intel_panel.c index a06ff07a4d3..a9e0c7bcd31 100644 --- a/drivers/gpu/drm/i915/intel_panel.c +++ b/drivers/gpu/drm/i915/intel_panel.c @@ -83,11 +83,15 @@ intel_pch_panel_fitting(struct drm_device *dev, u32 scaled_height = mode->hdisplay * adjusted_mode->vdisplay; if (scaled_width > scaled_height) { /* pillar */ width = scaled_height / mode->vdisplay; + if (width & 1) + width++; x = (adjusted_mode->hdisplay - width + 1) / 2; y = 0; height = adjusted_mode->vdisplay; } else if (scaled_width < scaled_height) { /* letter */ height = scaled_width / mode->hdisplay; + if (height & 1) + height++; y = (adjusted_mode->vdisplay - height + 1) / 2; x = 0; width = adjusted_mode->hdisplay; @@ -273,7 +277,7 @@ void intel_panel_enable_backlight(struct drm_device *dev) dev_priv->backlight_enabled = true; } -void intel_panel_setup_backlight(struct drm_device *dev) +static void intel_panel_init_backlight(struct drm_device *dev) { struct drm_i915_private *dev_priv = dev->dev_private; @@ -305,3 +309,73 @@ intel_panel_detect(struct drm_device *dev) return connector_status_unknown; } + +#ifdef CONFIG_BACKLIGHT_CLASS_DEVICE +static int intel_panel_update_status(struct backlight_device *bd) +{ + struct drm_device *dev = bl_get_data(bd); + intel_panel_set_backlight(dev, bd->props.brightness); + return 0; +} + +static int intel_panel_get_brightness(struct backlight_device *bd) +{ + struct drm_device *dev = bl_get_data(bd); + return intel_panel_get_backlight(dev); +} + +static const struct backlight_ops intel_panel_bl_ops = { + .update_status = intel_panel_update_status, + .get_brightness = intel_panel_get_brightness, +}; + +int intel_panel_setup_backlight(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct backlight_properties props; + struct drm_connector *connector; + + intel_panel_init_backlight(dev); + + if (dev_priv->int_lvds_connector) + connector = dev_priv->int_lvds_connector; + else if (dev_priv->int_edp_connector) + connector = dev_priv->int_edp_connector; + else + return -ENODEV; + + props.type = BACKLIGHT_RAW; + props.max_brightness = intel_panel_get_max_backlight(dev); + dev_priv->backlight = + backlight_device_register("intel_backlight", + &connector->kdev, dev, + &intel_panel_bl_ops, &props); + + if (IS_ERR(dev_priv->backlight)) { + DRM_ERROR("Failed to register backlight: %ld\n", + PTR_ERR(dev_priv->backlight)); + dev_priv->backlight = NULL; + return -ENODEV; + } + dev_priv->backlight->props.brightness = intel_panel_get_backlight(dev); + return 0; +} + +void intel_panel_destroy_backlight(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + if (dev_priv->backlight) + backlight_device_unregister(dev_priv->backlight); +} +#else +int intel_panel_setup_backlight(struct drm_device *dev) +{ + intel_panel_init_backlight(dev); + return 0; +} + +void intel_panel_destroy_backlight(struct drm_device *dev) +{ + return; +} +#endif diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index e9e6f71418a..c30626ea9f9 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -236,7 +236,8 @@ init_pipe_control(struct intel_ring_buffer *ring) ret = -ENOMEM; goto err; } - obj->agp_type = AGP_USER_CACHED_MEMORY; + + i915_gem_object_set_cache_level(obj, I915_CACHE_LLC); ret = i915_gem_object_pin(obj, 4096, true); if (ret) @@ -286,9 +287,13 @@ static int init_render_ring(struct intel_ring_buffer *ring) if (INTEL_INFO(dev)->gen > 3) { int mode = VS_TIMER_DISPATCH << 16 | VS_TIMER_DISPATCH; - if (IS_GEN6(dev)) + if (IS_GEN6(dev) || IS_GEN7(dev)) mode |= MI_FLUSH_ENABLE << 16 | MI_FLUSH_ENABLE; I915_WRITE(MI_MODE, mode); + if (IS_GEN7(dev)) + I915_WRITE(GFX_MODE_GEN7, + GFX_MODE_DISABLE(GFX_TLB_INVALIDATE_ALWAYS) | + GFX_MODE_ENABLE(GFX_REPLAY_MODE)); } if (INTEL_INFO(dev)->gen >= 6) { @@ -551,10 +556,31 @@ render_ring_put_irq(struct intel_ring_buffer *ring) void intel_ring_setup_status_page(struct intel_ring_buffer *ring) { + struct drm_device *dev = ring->dev; drm_i915_private_t *dev_priv = ring->dev->dev_private; - u32 mmio = IS_GEN6(ring->dev) ? - RING_HWS_PGA_GEN6(ring->mmio_base) : - RING_HWS_PGA(ring->mmio_base); + u32 mmio = 0; + + /* The ring status page addresses are no longer next to the rest of + * the ring registers as of gen7. + */ + if (IS_GEN7(dev)) { + switch (ring->id) { + case RING_RENDER: + mmio = RENDER_HWS_PGA_GEN7; + break; + case RING_BLT: + mmio = BLT_HWS_PGA_GEN7; + break; + case RING_BSD: + mmio = BSD_HWS_PGA_GEN7; + break; + } + } else if (IS_GEN6(ring->dev)) { + mmio = RING_HWS_PGA_GEN6(ring->mmio_base); + } else { + mmio = RING_HWS_PGA(ring->mmio_base); + } + I915_WRITE(mmio, (u32)ring->status_page.gfx_addr); POSTING_READ(mmio); } @@ -600,7 +626,7 @@ ring_add_request(struct intel_ring_buffer *ring, } static bool -ring_get_irq(struct intel_ring_buffer *ring, u32 flag) +gen6_ring_get_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag) { struct drm_device *dev = ring->dev; drm_i915_private_t *dev_priv = dev->dev_private; @@ -609,71 +635,67 @@ ring_get_irq(struct intel_ring_buffer *ring, u32 flag) return false; spin_lock(&ring->irq_lock); - if (ring->irq_refcount++ == 0) - ironlake_enable_irq(dev_priv, flag); + if (ring->irq_refcount++ == 0) { + ring->irq_mask &= ~rflag; + I915_WRITE_IMR(ring, ring->irq_mask); + ironlake_enable_irq(dev_priv, gflag); + } spin_unlock(&ring->irq_lock); return true; } static void -ring_put_irq(struct intel_ring_buffer *ring, u32 flag) +gen6_ring_put_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag) { struct drm_device *dev = ring->dev; drm_i915_private_t *dev_priv = dev->dev_private; spin_lock(&ring->irq_lock); - if (--ring->irq_refcount == 0) - ironlake_disable_irq(dev_priv, flag); + if (--ring->irq_refcount == 0) { + ring->irq_mask |= rflag; + I915_WRITE_IMR(ring, ring->irq_mask); + ironlake_disable_irq(dev_priv, gflag); + } spin_unlock(&ring->irq_lock); } static bool -gen6_ring_get_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag) +bsd_ring_get_irq(struct intel_ring_buffer *ring) { struct drm_device *dev = ring->dev; drm_i915_private_t *dev_priv = dev->dev_private; if (!dev->irq_enabled) - return false; + return false; spin_lock(&ring->irq_lock); if (ring->irq_refcount++ == 0) { - ring->irq_mask &= ~rflag; - I915_WRITE_IMR(ring, ring->irq_mask); - ironlake_enable_irq(dev_priv, gflag); + if (IS_G4X(dev)) + i915_enable_irq(dev_priv, I915_BSD_USER_INTERRUPT); + else + ironlake_enable_irq(dev_priv, GT_BSD_USER_INTERRUPT); } spin_unlock(&ring->irq_lock); return true; } - static void -gen6_ring_put_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag) +bsd_ring_put_irq(struct intel_ring_buffer *ring) { struct drm_device *dev = ring->dev; drm_i915_private_t *dev_priv = dev->dev_private; spin_lock(&ring->irq_lock); if (--ring->irq_refcount == 0) { - ring->irq_mask |= rflag; - I915_WRITE_IMR(ring, ring->irq_mask); - ironlake_disable_irq(dev_priv, gflag); + if (IS_G4X(dev)) + i915_disable_irq(dev_priv, I915_BSD_USER_INTERRUPT); + else + ironlake_disable_irq(dev_priv, GT_BSD_USER_INTERRUPT); } spin_unlock(&ring->irq_lock); } -static bool -bsd_ring_get_irq(struct intel_ring_buffer *ring) -{ - return ring_get_irq(ring, GT_BSD_USER_INTERRUPT); -} -static void -bsd_ring_put_irq(struct intel_ring_buffer *ring) -{ - ring_put_irq(ring, GT_BSD_USER_INTERRUPT); -} - static int ring_dispatch_execbuffer(struct intel_ring_buffer *ring, u32 offset, u32 length) { @@ -759,7 +781,8 @@ static int init_status_page(struct intel_ring_buffer *ring) ret = -ENOMEM; goto err; } - obj->agp_type = AGP_USER_CACHED_MEMORY; + + i915_gem_object_set_cache_level(obj, I915_CACHE_LLC); ret = i915_gem_object_pin(obj, 4096, true); if (ret != 0) { @@ -800,6 +823,7 @@ int intel_init_ring_buffer(struct drm_device *dev, INIT_LIST_HEAD(&ring->request_list); INIT_LIST_HEAD(&ring->gpu_write_list); + init_waitqueue_head(&ring->irq_queue); spin_lock_init(&ring->irq_lock); ring->irq_mask = ~0; @@ -872,7 +896,7 @@ void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring) /* Disable the ring buffer. The ring must be idle at this point */ dev_priv = ring->dev->dev_private; - ret = intel_wait_ring_buffer(ring, ring->size - 8); + ret = intel_wait_ring_idle(ring); if (ret) DRM_ERROR("failed to quiesce %s whilst cleaning up: %d\n", ring->name, ret); @@ -1301,6 +1325,9 @@ int intel_render_ring_init_dri(struct drm_device *dev, u64 start, u32 size) ring->get_seqno = pc_render_get_seqno; } + if (!I915_NEED_GFX_HWS(dev)) + ring->status_page.page_addr = dev_priv->status_page_dmah->vaddr; + ring->dev = dev; INIT_LIST_HEAD(&ring->active_list); INIT_LIST_HEAD(&ring->request_list); @@ -1333,7 +1360,7 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev) drm_i915_private_t *dev_priv = dev->dev_private; struct intel_ring_buffer *ring = &dev_priv->ring[VCS]; - if (IS_GEN6(dev)) + if (IS_GEN6(dev) || IS_GEN7(dev)) *ring = gen6_bsd_ring; else *ring = bsd_ring; diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h index f23cc5f037a..39ac2b634ae 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.h +++ b/drivers/gpu/drm/i915/intel_ringbuffer.h @@ -14,27 +14,24 @@ struct intel_hw_status_page { struct drm_i915_gem_object *obj; }; -#define I915_RING_READ(reg) i915_gt_read(dev_priv, reg) -#define I915_RING_WRITE(reg, val) i915_gt_write(dev_priv, reg, val) +#define I915_READ_TAIL(ring) I915_READ(RING_TAIL((ring)->mmio_base)) +#define I915_WRITE_TAIL(ring, val) I915_WRITE(RING_TAIL((ring)->mmio_base), val) -#define I915_READ_TAIL(ring) I915_RING_READ(RING_TAIL((ring)->mmio_base)) -#define I915_WRITE_TAIL(ring, val) I915_RING_WRITE(RING_TAIL((ring)->mmio_base), val) +#define I915_READ_START(ring) I915_READ(RING_START((ring)->mmio_base)) +#define I915_WRITE_START(ring, val) I915_WRITE(RING_START((ring)->mmio_base), val) -#define I915_READ_START(ring) I915_RING_READ(RING_START((ring)->mmio_base)) -#define I915_WRITE_START(ring, val) I915_RING_WRITE(RING_START((ring)->mmio_base), val) +#define I915_READ_HEAD(ring) I915_READ(RING_HEAD((ring)->mmio_base)) +#define I915_WRITE_HEAD(ring, val) I915_WRITE(RING_HEAD((ring)->mmio_base), val) -#define I915_READ_HEAD(ring) I915_RING_READ(RING_HEAD((ring)->mmio_base)) -#define I915_WRITE_HEAD(ring, val) I915_RING_WRITE(RING_HEAD((ring)->mmio_base), val) +#define I915_READ_CTL(ring) I915_READ(RING_CTL((ring)->mmio_base)) +#define I915_WRITE_CTL(ring, val) I915_WRITE(RING_CTL((ring)->mmio_base), val) -#define I915_READ_CTL(ring) I915_RING_READ(RING_CTL((ring)->mmio_base)) -#define I915_WRITE_CTL(ring, val) I915_RING_WRITE(RING_CTL((ring)->mmio_base), val) +#define I915_READ_IMR(ring) I915_READ(RING_IMR((ring)->mmio_base)) +#define I915_WRITE_IMR(ring, val) I915_WRITE(RING_IMR((ring)->mmio_base), val) -#define I915_READ_IMR(ring) I915_RING_READ(RING_IMR((ring)->mmio_base)) -#define I915_WRITE_IMR(ring, val) I915_RING_WRITE(RING_IMR((ring)->mmio_base), val) - -#define I915_READ_NOPID(ring) I915_RING_READ(RING_NOPID((ring)->mmio_base)) -#define I915_READ_SYNC_0(ring) I915_RING_READ(RING_SYNC_0((ring)->mmio_base)) -#define I915_READ_SYNC_1(ring) I915_RING_READ(RING_SYNC_1((ring)->mmio_base)) +#define I915_READ_NOPID(ring) I915_READ(RING_NOPID((ring)->mmio_base)) +#define I915_READ_SYNC_0(ring) I915_READ(RING_SYNC_0((ring)->mmio_base)) +#define I915_READ_SYNC_1(ring) I915_READ(RING_SYNC_1((ring)->mmio_base)) struct intel_ring_buffer { const char *name; @@ -164,7 +161,13 @@ intel_read_status_page(struct intel_ring_buffer *ring, #define I915_BREADCRUMB_INDEX 0x21 void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring); + int __must_check intel_wait_ring_buffer(struct intel_ring_buffer *ring, int n); +static inline int intel_wait_ring_idle(struct intel_ring_buffer *ring) +{ + return intel_wait_ring_buffer(ring, ring->size - 8); +} + int __must_check intel_ring_begin(struct intel_ring_buffer *ring, int n); static inline void intel_ring_emit(struct intel_ring_buffer *ring, diff --git a/drivers/gpu/drm/i915/intel_sdvo.c b/drivers/gpu/drm/i915/intel_sdvo.c index 4324f33212d..30fe554d893 100644 --- a/drivers/gpu/drm/i915/intel_sdvo.c +++ b/drivers/gpu/drm/i915/intel_sdvo.c @@ -148,8 +148,6 @@ struct intel_sdvo_connector { int format_supported_num; struct drm_property *tv_format; - struct drm_property *force_audio_property; - /* add the property for the SDVO-TV */ struct drm_property *left; struct drm_property *right; @@ -1712,7 +1710,7 @@ intel_sdvo_set_property(struct drm_connector *connector, if (ret) return ret; - if (property == intel_sdvo_connector->force_audio_property) { + if (property == dev_priv->force_audio_property) { int i = val; bool has_audio; @@ -2037,15 +2035,7 @@ intel_sdvo_add_hdmi_properties(struct intel_sdvo_connector *connector) { struct drm_device *dev = connector->base.base.dev; - connector->force_audio_property = - drm_property_create(dev, DRM_MODE_PROP_RANGE, "force_audio", 2); - if (connector->force_audio_property) { - connector->force_audio_property->values[0] = -1; - connector->force_audio_property->values[1] = 1; - drm_connector_attach_property(&connector->base.base, - connector->force_audio_property, 0); - } - + intel_attach_force_audio_property(&connector->base.base); if (INTEL_INFO(dev)->gen >= 4 && IS_MOBILE(dev)) intel_attach_broadcast_rgb_property(&connector->base.base); } @@ -2544,21 +2534,19 @@ bool intel_sdvo_init(struct drm_device *dev, int sdvo_reg) if (!intel_sdvo) return false; + intel_sdvo->sdvo_reg = sdvo_reg; + intel_sdvo->slave_addr = intel_sdvo_get_slave_addr(dev, sdvo_reg) >> 1; + intel_sdvo_select_i2c_bus(dev_priv, intel_sdvo, sdvo_reg); if (!intel_sdvo_init_ddc_proxy(intel_sdvo, dev)) { kfree(intel_sdvo); return false; } - intel_sdvo->sdvo_reg = sdvo_reg; - + /* encoder type will be decided later */ intel_encoder = &intel_sdvo->base; intel_encoder->type = INTEL_OUTPUT_SDVO; - /* encoder type will be decided later */ drm_encoder_init(dev, &intel_encoder->base, &intel_sdvo_enc_funcs, 0); - intel_sdvo->slave_addr = intel_sdvo_get_slave_addr(dev, sdvo_reg) >> 1; - intel_sdvo_select_i2c_bus(dev_priv, intel_sdvo, sdvo_reg); - /* Read the regs to test if we can talk to the device */ for (i = 0; i < 0x40; i++) { u8 byte; diff --git a/drivers/gpu/drm/i915/intel_tv.c b/drivers/gpu/drm/i915/intel_tv.c index 6b22c1dcc01..210d570fd51 100644 --- a/drivers/gpu/drm/i915/intel_tv.c +++ b/drivers/gpu/drm/i915/intel_tv.c @@ -1236,6 +1236,8 @@ intel_tv_detect_type (struct intel_tv *intel_tv, struct drm_connector *connector) { struct drm_encoder *encoder = &intel_tv->base.base; + struct drm_crtc *crtc = encoder->crtc; + struct intel_crtc *intel_crtc = to_intel_crtc(crtc); struct drm_device *dev = encoder->dev; struct drm_i915_private *dev_priv = dev->dev_private; unsigned long irqflags; @@ -1258,6 +1260,10 @@ intel_tv_detect_type (struct intel_tv *intel_tv, /* Poll for TV detection */ tv_ctl &= ~(TV_ENC_ENABLE | TV_TEST_MODE_MASK); tv_ctl |= TV_TEST_MODE_MONITOR_DETECT; + if (intel_crtc->pipe == 1) + tv_ctl |= TV_ENC_PIPEB_SELECT; + else + tv_ctl &= ~TV_ENC_PIPEB_SELECT; tv_dac &= ~(TVDAC_SENSE_MASK | DAC_A_MASK | DAC_B_MASK | DAC_C_MASK); tv_dac |= (TVDAC_STATE_CHG_EN | @@ -1277,26 +1283,26 @@ intel_tv_detect_type (struct intel_tv *intel_tv, to_intel_crtc(intel_tv->base.base.crtc)->pipe); type = -1; - if (wait_for((tv_dac = I915_READ(TV_DAC)) & TVDAC_STATE_CHG, 20) == 0) { - DRM_DEBUG_KMS("TV detected: %x, %x\n", tv_ctl, tv_dac); - /* - * A B C - * 0 1 1 Composite - * 1 0 X svideo - * 0 0 0 Component - */ - if ((tv_dac & TVDAC_SENSE_MASK) == (TVDAC_B_SENSE | TVDAC_C_SENSE)) { - DRM_DEBUG_KMS("Detected Composite TV connection\n"); - type = DRM_MODE_CONNECTOR_Composite; - } else if ((tv_dac & (TVDAC_A_SENSE|TVDAC_B_SENSE)) == TVDAC_A_SENSE) { - DRM_DEBUG_KMS("Detected S-Video TV connection\n"); - type = DRM_MODE_CONNECTOR_SVIDEO; - } else if ((tv_dac & TVDAC_SENSE_MASK) == 0) { - DRM_DEBUG_KMS("Detected Component TV connection\n"); - type = DRM_MODE_CONNECTOR_Component; - } else { - DRM_DEBUG_KMS("Unrecognised TV connection\n"); - } + tv_dac = I915_READ(TV_DAC); + DRM_DEBUG_KMS("TV detected: %x, %x\n", tv_ctl, tv_dac); + /* + * A B C + * 0 1 1 Composite + * 1 0 X svideo + * 0 0 0 Component + */ + if ((tv_dac & TVDAC_SENSE_MASK) == (TVDAC_B_SENSE | TVDAC_C_SENSE)) { + DRM_DEBUG_KMS("Detected Composite TV connection\n"); + type = DRM_MODE_CONNECTOR_Composite; + } else if ((tv_dac & (TVDAC_A_SENSE|TVDAC_B_SENSE)) == TVDAC_A_SENSE) { + DRM_DEBUG_KMS("Detected S-Video TV connection\n"); + type = DRM_MODE_CONNECTOR_SVIDEO; + } else if ((tv_dac & TVDAC_SENSE_MASK) == 0) { + DRM_DEBUG_KMS("Detected Component TV connection\n"); + type = DRM_MODE_CONNECTOR_Component; + } else { + DRM_DEBUG_KMS("Unrecognised TV connection\n"); + type = -1; } I915_WRITE(TV_DAC, save_tv_dac & ~TVDAC_STATE_CHG_EN); @@ -1361,15 +1367,14 @@ intel_tv_detect(struct drm_connector *connector, bool force) if (intel_tv->base.base.crtc && intel_tv->base.base.crtc->enabled) { type = intel_tv_detect_type(intel_tv, connector); } else if (force) { - struct drm_crtc *crtc; - int dpms_mode; + struct intel_load_detect_pipe tmp; - crtc = intel_get_load_detect_pipe(&intel_tv->base, connector, - &mode, &dpms_mode); - if (crtc) { + if (intel_get_load_detect_pipe(&intel_tv->base, connector, + &mode, &tmp)) { type = intel_tv_detect_type(intel_tv, connector); - intel_release_load_detect_pipe(&intel_tv->base, connector, - dpms_mode); + intel_release_load_detect_pipe(&intel_tv->base, + connector, + &tmp); } else return connector_status_unknown; } else diff --git a/drivers/gpu/drm/mga/mga_drv.h b/drivers/gpu/drm/mga/mga_drv.h index 1084fa4d261..54558a01969 100644 --- a/drivers/gpu/drm/mga/mga_drv.h +++ b/drivers/gpu/drm/mga/mga_drv.h @@ -195,29 +195,10 @@ extern long mga_compat_ioctl(struct file *filp, unsigned int cmd, #define mga_flush_write_combine() DRM_WRITEMEMORYBARRIER() -#if defined(__linux__) && defined(__alpha__) -#define MGA_BASE(reg) ((unsigned long)(dev_priv->mmio->handle)) -#define MGA_ADDR(reg) (MGA_BASE(reg) + reg) - -#define MGA_DEREF(reg) (*(volatile u32 *)MGA_ADDR(reg)) -#define MGA_DEREF8(reg) (*(volatile u8 *)MGA_ADDR(reg)) - -#define MGA_READ(reg) (_MGA_READ((u32 *)MGA_ADDR(reg))) -#define MGA_READ8(reg) (_MGA_READ((u8 *)MGA_ADDR(reg))) -#define MGA_WRITE(reg, val) do { DRM_WRITEMEMORYBARRIER(); MGA_DEREF(reg) = val; } while (0) -#define MGA_WRITE8(reg, val) do { DRM_WRITEMEMORYBARRIER(); MGA_DEREF8(reg) = val; } while (0) - -static inline u32 _MGA_READ(u32 *addr) -{ - DRM_MEMORYBARRIER(); - return *(volatile u32 *)addr; -} -#else #define MGA_READ8(reg) DRM_READ8(dev_priv->mmio, (reg)) #define MGA_READ(reg) DRM_READ32(dev_priv->mmio, (reg)) #define MGA_WRITE8(reg, val) DRM_WRITE8(dev_priv->mmio, (reg), (val)) #define MGA_WRITE(reg, val) DRM_WRITE32(dev_priv->mmio, (reg), (val)) -#endif #define DWGREG0 0x1c00 #define DWGREG0_END 0x1dff diff --git a/drivers/gpu/drm/nouveau/Kconfig b/drivers/gpu/drm/nouveau/Kconfig index de70959b9ed..ca1639918f5 100644 --- a/drivers/gpu/drm/nouveau/Kconfig +++ b/drivers/gpu/drm/nouveau/Kconfig @@ -11,6 +11,8 @@ config DRM_NOUVEAU select FRAMEBUFFER_CONSOLE if !EXPERT select FB_BACKLIGHT if DRM_NOUVEAU_BACKLIGHT select ACPI_VIDEO if ACPI && X86 && BACKLIGHT_CLASS_DEVICE && VIDEO_OUTPUT_CONTROL && INPUT + select ACPI_WMI if ACPI + select MXM_WMI if ACPI help Choose this option for open-source nVidia support. diff --git a/drivers/gpu/drm/nouveau/Makefile b/drivers/gpu/drm/nouveau/Makefile index e12c97fd8db..0583677e458 100644 --- a/drivers/gpu/drm/nouveau/Makefile +++ b/drivers/gpu/drm/nouveau/Makefile @@ -20,6 +20,8 @@ nouveau-y := nouveau_drv.o nouveau_state.o nouveau_channel.o nouveau_mem.o \ nv40_graph.o nv50_graph.o nvc0_graph.o \ nv40_grctx.o nv50_grctx.o nvc0_grctx.o \ nv84_crypt.o \ + nva3_copy.o nvc0_copy.o \ + nv40_mpeg.o nv50_mpeg.o \ nv04_instmem.o nv50_instmem.o nvc0_instmem.o \ nv50_evo.o nv50_crtc.o nv50_dac.o nv50_sor.o \ nv50_cursor.o nv50_display.o \ diff --git a/drivers/gpu/drm/nouveau/nouveau_acpi.c b/drivers/gpu/drm/nouveau/nouveau_acpi.c index a54238058dc..525744d593c 100644 --- a/drivers/gpu/drm/nouveau/nouveau_acpi.c +++ b/drivers/gpu/drm/nouveau/nouveau_acpi.c @@ -4,6 +4,8 @@ #include <acpi/acpi_drivers.h> #include <acpi/acpi_bus.h> #include <acpi/video.h> +#include <acpi/acpi.h> +#include <linux/mxm-wmi.h> #include "drmP.h" #include "drm.h" @@ -35,15 +37,71 @@ static struct nouveau_dsm_priv { bool dsm_detected; + bool optimus_detected; acpi_handle dhandle; acpi_handle rom_handle; } nouveau_dsm_priv; +#define NOUVEAU_DSM_HAS_MUX 0x1 +#define NOUVEAU_DSM_HAS_OPT 0x2 + static const char nouveau_dsm_muid[] = { 0xA0, 0xA0, 0x95, 0x9D, 0x60, 0x00, 0x48, 0x4D, 0xB3, 0x4D, 0x7E, 0x5F, 0xEA, 0x12, 0x9F, 0xD4, }; +static const char nouveau_op_dsm_muid[] = { + 0xF8, 0xD8, 0x86, 0xA4, 0xDA, 0x0B, 0x1B, 0x47, + 0xA7, 0x2B, 0x60, 0x42, 0xA6, 0xB5, 0xBE, 0xE0, +}; + +static int nouveau_optimus_dsm(acpi_handle handle, int func, int arg, uint32_t *result) +{ + struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL }; + struct acpi_object_list input; + union acpi_object params[4]; + union acpi_object *obj; + int err; + + input.count = 4; + input.pointer = params; + params[0].type = ACPI_TYPE_BUFFER; + params[0].buffer.length = sizeof(nouveau_op_dsm_muid); + params[0].buffer.pointer = (char *)nouveau_op_dsm_muid; + params[1].type = ACPI_TYPE_INTEGER; + params[1].integer.value = 0x00000100; + params[2].type = ACPI_TYPE_INTEGER; + params[2].integer.value = func; + params[3].type = ACPI_TYPE_BUFFER; + params[3].buffer.length = 0; + + err = acpi_evaluate_object(handle, "_DSM", &input, &output); + if (err) { + printk(KERN_INFO "failed to evaluate _DSM: %d\n", err); + return err; + } + + obj = (union acpi_object *)output.pointer; + + if (obj->type == ACPI_TYPE_INTEGER) + if (obj->integer.value == 0x80000002) { + return -ENODEV; + } + + if (obj->type == ACPI_TYPE_BUFFER) { + if (obj->buffer.length == 4 && result) { + *result = 0; + *result |= obj->buffer.pointer[0]; + *result |= (obj->buffer.pointer[1] << 8); + *result |= (obj->buffer.pointer[2] << 16); + *result |= (obj->buffer.pointer[3] << 24); + } + } + + kfree(output.pointer); + return 0; +} + static int nouveau_dsm(acpi_handle handle, int func, int arg, uint32_t *result) { struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL }; @@ -92,6 +150,8 @@ static int nouveau_dsm(acpi_handle handle, int func, int arg, uint32_t *result) static int nouveau_dsm_switch_mux(acpi_handle handle, int mux_id) { + mxm_wmi_call_mxmx(mux_id == NOUVEAU_DSM_LED_STAMINA ? MXM_MXDS_ADAPTER_IGD : MXM_MXDS_ADAPTER_0); + mxm_wmi_call_mxds(mux_id == NOUVEAU_DSM_LED_STAMINA ? MXM_MXDS_ADAPTER_IGD : MXM_MXDS_ADAPTER_0); return nouveau_dsm(handle, NOUVEAU_DSM_LED, mux_id, NULL); } @@ -148,11 +208,11 @@ static struct vga_switcheroo_handler nouveau_dsm_handler = { .get_client_id = nouveau_dsm_get_client_id, }; -static bool nouveau_dsm_pci_probe(struct pci_dev *pdev) +static int nouveau_dsm_pci_probe(struct pci_dev *pdev) { acpi_handle dhandle, nvidia_handle; acpi_status status; - int ret; + int ret, retval = 0; uint32_t result; dhandle = DEVICE_ACPI_HANDLE(&pdev->dev); @@ -166,11 +226,17 @@ static bool nouveau_dsm_pci_probe(struct pci_dev *pdev) ret = nouveau_dsm(dhandle, NOUVEAU_DSM_SUPPORTED, NOUVEAU_DSM_SUPPORTED_FUNCTIONS, &result); - if (ret < 0) - return false; + if (ret == 0) + retval |= NOUVEAU_DSM_HAS_MUX; - nouveau_dsm_priv.dhandle = dhandle; - return true; + ret = nouveau_optimus_dsm(dhandle, 0, 0, &result); + if (ret == 0) + retval |= NOUVEAU_DSM_HAS_OPT; + + if (retval) + nouveau_dsm_priv.dhandle = dhandle; + + return retval; } static bool nouveau_dsm_detect(void) @@ -179,22 +245,41 @@ static bool nouveau_dsm_detect(void) struct acpi_buffer buffer = {sizeof(acpi_method_name), acpi_method_name}; struct pci_dev *pdev = NULL; int has_dsm = 0; + int has_optimus; int vga_count = 0; + bool guid_valid; + int retval; + bool ret = false; + + /* lookup the MXM GUID */ + guid_valid = mxm_wmi_supported(); + if (guid_valid) + printk("MXM: GUID detected in BIOS\n"); + + /* now do DSM detection */ while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) { vga_count++; - has_dsm |= (nouveau_dsm_pci_probe(pdev) == true); + retval = nouveau_dsm_pci_probe(pdev); + if (retval & NOUVEAU_DSM_HAS_MUX) + has_dsm |= 1; + if (retval & NOUVEAU_DSM_HAS_OPT) + has_optimus = 1; } - if (vga_count == 2 && has_dsm) { + if (vga_count == 2 && has_dsm && guid_valid) { acpi_get_name(nouveau_dsm_priv.dhandle, ACPI_FULL_PATHNAME, &buffer); printk(KERN_INFO "VGA switcheroo: detected DSM switching method %s handle\n", acpi_method_name); nouveau_dsm_priv.dsm_detected = true; - return true; + ret = true; } - return false; + + if (has_optimus == 1) + nouveau_dsm_priv.optimus_detected = true; + + return ret; } void nouveau_register_dsm_handler(void) @@ -247,7 +332,7 @@ bool nouveau_acpi_rom_supported(struct pci_dev *pdev) acpi_status status; acpi_handle dhandle, rom_handle; - if (!nouveau_dsm_priv.dsm_detected) + if (!nouveau_dsm_priv.dsm_detected && !nouveau_dsm_priv.optimus_detected) return false; dhandle = DEVICE_ACPI_HANDLE(&pdev->dev); diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.c b/drivers/gpu/drm/nouveau/nouveau_bios.c index 90aef64b76f..b311faba34f 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bios.c +++ b/drivers/gpu/drm/nouveau/nouveau_bios.c @@ -135,13 +135,14 @@ static void load_vbios_pramin(struct drm_device *dev, uint8_t *data) int i; if (dev_priv->card_type >= NV_50) { - uint32_t vbios_vram = (nv_rd32(dev, 0x619f04) & ~0xff) << 8; - - if (!vbios_vram) - vbios_vram = (nv_rd32(dev, 0x1700) << 16) + 0xf0000; + u64 addr = (u64)(nv_rd32(dev, 0x619f04) & 0xffffff00) << 8; + if (!addr) { + addr = (u64)nv_rd32(dev, 0x1700) << 16; + addr += 0xf0000; + } old_bar0_pramin = nv_rd32(dev, 0x1700); - nv_wr32(dev, 0x1700, vbios_vram >> 16); + nv_wr32(dev, 0x1700, addr >> 16); } /* bail if no rom signature */ @@ -5049,11 +5050,7 @@ int get_pll_limits(struct drm_device *dev, uint32_t limit_match, struct pll_lims pll_lim->vco1.max_n = record[11]; pll_lim->min_p = record[12]; pll_lim->max_p = record[13]; - /* where did this go to?? */ - if ((entry[0] & 0xf0) == 0x80) - pll_lim->refclk = 27000; - else - pll_lim->refclk = 100000; + pll_lim->refclk = ROM16(entry[9]) * 1000; } /* @@ -5190,7 +5187,7 @@ static int parse_bit_A_tbl_entry(struct drm_device *dev, struct nvbios *bios, st load_table_ptr = ROM16(bios->data[bitentry->offset]); if (load_table_ptr == 0x0) { - NV_ERROR(dev, "Pointer to BIT loadval table invalid\n"); + NV_DEBUG(dev, "Pointer to BIT loadval table invalid\n"); return -EINVAL; } @@ -5969,6 +5966,12 @@ apply_dcb_connector_quirks(struct nvbios *bios, int idx) if (cte->type == DCB_CONNECTOR_HDMI_1) cte->type = DCB_CONNECTOR_DVI_I; } + + /* Gigabyte GV-NX86T512H */ + if (nv_match_device(dev, 0x0402, 0x1458, 0x3455)) { + if (cte->type == DCB_CONNECTOR_HDMI_1) + cte->type = DCB_CONNECTOR_DVI_I; + } } static const u8 hpd_gpio[16] = { @@ -6035,6 +6038,7 @@ parse_dcb_connector_table(struct nvbios *bios) case DCB_CONNECTOR_DVI_I: case DCB_CONNECTOR_DVI_D: case DCB_CONNECTOR_LVDS: + case DCB_CONNECTOR_LVDS_SPWG: case DCB_CONNECTOR_DP: case DCB_CONNECTOR_eDP: case DCB_CONNECTOR_HDMI_0: @@ -6380,6 +6384,37 @@ apply_dcb_encoder_quirks(struct drm_device *dev, int idx, u32 *conn, u32 *conf) } } + /* Some other twisted XFX board (rhbz#694914) + * + * The DVI/VGA encoder combo that's supposed to represent the + * DVI-I connector actually point at two different ones, and + * the HDMI connector ends up paired with the VGA instead. + * + * Connector table is missing anything for VGA at all, pointing it + * an invalid conntab entry 2 so we figure it out ourself. + */ + if (nv_match_device(dev, 0x0615, 0x1682, 0x2605)) { + if (idx == 0) { + *conn = 0x02002300; /* VGA, connector 2 */ + *conf = 0x00000028; + } else + if (idx == 1) { + *conn = 0x01010312; /* DVI, connector 0 */ + *conf = 0x00020030; + } else + if (idx == 2) { + *conn = 0x04020310; /* VGA, connector 0 */ + *conf = 0x00000028; + } else + if (idx == 3) { + *conn = 0x02021322; /* HDMI, connector 1 */ + *conf = 0x00020010; + } else { + *conn = 0x0000000e; /* EOL */ + *conf = 0x00000000; + } + } + return true; } diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.h b/drivers/gpu/drm/nouveau/nouveau_bios.h index 8a54fa7edf5..050c314119d 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bios.h +++ b/drivers/gpu/drm/nouveau/nouveau_bios.h @@ -82,6 +82,7 @@ enum dcb_connector_type { DCB_CONNECTOR_DVI_I = 0x30, DCB_CONNECTOR_DVI_D = 0x31, DCB_CONNECTOR_LVDS = 0x40, + DCB_CONNECTOR_LVDS_SPWG = 0x41, DCB_CONNECTOR_DP = 0x46, DCB_CONNECTOR_eDP = 0x47, DCB_CONNECTOR_HDMI_0 = 0x60, diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 2ad49cbf7c8..890d50e4d68 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -49,16 +49,12 @@ nouveau_bo_del_ttm(struct ttm_buffer_object *bo) DRM_ERROR("bo %p still attached to GEM object\n", bo); nv10_mem_put_tile_region(dev, nvbo->tile, NULL); - if (nvbo->vma.node) { - nouveau_vm_unmap(&nvbo->vma); - nouveau_vm_put(&nvbo->vma); - } kfree(nvbo); } static void nouveau_bo_fixup_align(struct nouveau_bo *nvbo, u32 flags, - int *align, int *size, int *page_shift) + int *align, int *size) { struct drm_nouveau_private *dev_priv = nouveau_bdev(nvbo->bo.bdev); @@ -82,67 +78,51 @@ nouveau_bo_fixup_align(struct nouveau_bo *nvbo, u32 flags, } } } else { - if (likely(dev_priv->chan_vm)) { - if (!(flags & TTM_PL_FLAG_TT) && *size > 256 * 1024) - *page_shift = dev_priv->chan_vm->lpg_shift; - else - *page_shift = dev_priv->chan_vm->spg_shift; - } else { - *page_shift = 12; - } - - *size = roundup(*size, (1 << *page_shift)); - *align = max((1 << *page_shift), *align); + *size = roundup(*size, (1 << nvbo->page_shift)); + *align = max((1 << nvbo->page_shift), *align); } *size = roundup(*size, PAGE_SIZE); } int -nouveau_bo_new(struct drm_device *dev, struct nouveau_channel *chan, - int size, int align, uint32_t flags, uint32_t tile_mode, - uint32_t tile_flags, struct nouveau_bo **pnvbo) +nouveau_bo_new(struct drm_device *dev, int size, int align, + uint32_t flags, uint32_t tile_mode, uint32_t tile_flags, + struct nouveau_bo **pnvbo) { struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_bo *nvbo; - int ret = 0, page_shift = 0; + int ret; nvbo = kzalloc(sizeof(struct nouveau_bo), GFP_KERNEL); if (!nvbo) return -ENOMEM; INIT_LIST_HEAD(&nvbo->head); INIT_LIST_HEAD(&nvbo->entry); + INIT_LIST_HEAD(&nvbo->vma_list); nvbo->tile_mode = tile_mode; nvbo->tile_flags = tile_flags; nvbo->bo.bdev = &dev_priv->ttm.bdev; - nouveau_bo_fixup_align(nvbo, flags, &align, &size, &page_shift); - align >>= PAGE_SHIFT; - - if (dev_priv->chan_vm) { - ret = nouveau_vm_get(dev_priv->chan_vm, size, page_shift, - NV_MEM_ACCESS_RW, &nvbo->vma); - if (ret) { - kfree(nvbo); - return ret; - } + nvbo->page_shift = 12; + if (dev_priv->bar1_vm) { + if (!(flags & TTM_PL_FLAG_TT) && size > 256 * 1024) + nvbo->page_shift = dev_priv->bar1_vm->lpg_shift; } + nouveau_bo_fixup_align(nvbo, flags, &align, &size); nvbo->bo.mem.num_pages = size >> PAGE_SHIFT; nouveau_bo_placement_set(nvbo, flags, 0); - nvbo->channel = chan; ret = ttm_bo_init(&dev_priv->ttm.bdev, &nvbo->bo, size, - ttm_bo_type_device, &nvbo->placement, align, 0, - false, NULL, size, nouveau_bo_del_ttm); + ttm_bo_type_device, &nvbo->placement, + align >> PAGE_SHIFT, 0, false, NULL, size, + nouveau_bo_del_ttm); if (ret) { /* ttm will call nouveau_bo_del_ttm if it fails.. */ return ret; } - nvbo->channel = NULL; - if (nvbo->vma.node) - nvbo->bo.offset = nvbo->vma.offset; *pnvbo = nvbo; return 0; } @@ -312,8 +292,6 @@ nouveau_bo_validate(struct nouveau_bo *nvbo, bool interruptible, if (ret) return ret; - if (nvbo->vma.node) - nvbo->bo.offset = nvbo->vma.offset; return 0; } @@ -440,7 +418,6 @@ nouveau_bo_init_mem_type(struct ttm_bo_device *bdev, uint32_t type, TTM_MEMTYPE_FLAG_CMA; man->available_caching = TTM_PL_MASK_CACHING; man->default_caching = TTM_PL_FLAG_CACHED; - man->gpu_offset = dev_priv->gart_info.aper_base; break; default: NV_ERROR(dev, "Unknown GART type: %d\n", @@ -501,19 +478,12 @@ static int nvc0_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, struct ttm_mem_reg *old_mem, struct ttm_mem_reg *new_mem) { - struct nouveau_mem *old_node = old_mem->mm_node; - struct nouveau_mem *new_node = new_mem->mm_node; - struct nouveau_bo *nvbo = nouveau_bo(bo); + struct nouveau_mem *node = old_mem->mm_node; + u64 src_offset = node->vma[0].offset; + u64 dst_offset = node->vma[1].offset; u32 page_count = new_mem->num_pages; - u64 src_offset, dst_offset; int ret; - src_offset = old_node->tmp_vma.offset; - if (new_node->tmp_vma.node) - dst_offset = new_node->tmp_vma.offset; - else - dst_offset = nvbo->vma.offset; - page_count = new_mem->num_pages; while (page_count) { int line_count = (page_count > 2047) ? 2047 : page_count; @@ -547,19 +517,13 @@ static int nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, struct ttm_mem_reg *old_mem, struct ttm_mem_reg *new_mem) { - struct nouveau_mem *old_node = old_mem->mm_node; - struct nouveau_mem *new_node = new_mem->mm_node; + struct nouveau_mem *node = old_mem->mm_node; struct nouveau_bo *nvbo = nouveau_bo(bo); u64 length = (new_mem->num_pages << PAGE_SHIFT); - u64 src_offset, dst_offset; + u64 src_offset = node->vma[0].offset; + u64 dst_offset = node->vma[1].offset; int ret; - src_offset = old_node->tmp_vma.offset; - if (new_node->tmp_vma.node) - dst_offset = new_node->tmp_vma.offset; - else - dst_offset = nvbo->vma.offset; - while (length) { u32 amount, stride, height; @@ -695,6 +659,27 @@ nv04_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, } static int +nouveau_vma_getmap(struct nouveau_channel *chan, struct nouveau_bo *nvbo, + struct ttm_mem_reg *mem, struct nouveau_vma *vma) +{ + struct nouveau_mem *node = mem->mm_node; + int ret; + + ret = nouveau_vm_get(chan->vm, mem->num_pages << PAGE_SHIFT, + node->page_shift, NV_MEM_ACCESS_RO, vma); + if (ret) + return ret; + + if (mem->mem_type == TTM_PL_VRAM) + nouveau_vm_map(vma, node); + else + nouveau_vm_map_sg(vma, 0, mem->num_pages << PAGE_SHIFT, + node, node->pages); + + return 0; +} + +static int nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, bool intr, bool no_wait_reserve, bool no_wait_gpu, struct ttm_mem_reg *new_mem) @@ -711,31 +696,20 @@ nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, bool intr, mutex_lock_nested(&chan->mutex, NOUVEAU_KCHANNEL_MUTEX); } - /* create temporary vma for old memory, this will get cleaned - * up after ttm destroys the ttm_mem_reg + /* create temporary vmas for the transfer and attach them to the + * old nouveau_mem node, these will get cleaned up after ttm has + * destroyed the ttm_mem_reg */ if (dev_priv->card_type >= NV_50) { struct nouveau_mem *node = old_mem->mm_node; - if (!node->tmp_vma.node) { - u32 page_shift = nvbo->vma.node->type; - if (old_mem->mem_type == TTM_PL_TT) - page_shift = nvbo->vma.vm->spg_shift; - - ret = nouveau_vm_get(chan->vm, - old_mem->num_pages << PAGE_SHIFT, - page_shift, NV_MEM_ACCESS_RO, - &node->tmp_vma); - if (ret) - goto out; - } - if (old_mem->mem_type == TTM_PL_VRAM) - nouveau_vm_map(&node->tmp_vma, node); - else { - nouveau_vm_map_sg(&node->tmp_vma, 0, - old_mem->num_pages << PAGE_SHIFT, - node, node->pages); - } + ret = nouveau_vma_getmap(chan, nvbo, old_mem, &node->vma[0]); + if (ret) + goto out; + + ret = nouveau_vma_getmap(chan, nvbo, new_mem, &node->vma[1]); + if (ret) + goto out; } if (dev_priv->card_type < NV_50) @@ -762,7 +736,6 @@ nouveau_bo_move_flipd(struct ttm_buffer_object *bo, bool evict, bool intr, bool no_wait_reserve, bool no_wait_gpu, struct ttm_mem_reg *new_mem) { - struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); u32 placement_memtype = TTM_PL_FLAG_TT | TTM_PL_MASK_CACHING; struct ttm_placement placement; struct ttm_mem_reg tmp_mem; @@ -782,23 +755,7 @@ nouveau_bo_move_flipd(struct ttm_buffer_object *bo, bool evict, bool intr, if (ret) goto out; - if (dev_priv->card_type >= NV_50) { - struct nouveau_bo *nvbo = nouveau_bo(bo); - struct nouveau_mem *node = tmp_mem.mm_node; - struct nouveau_vma *vma = &nvbo->vma; - if (vma->node->type != vma->vm->spg_shift) - vma = &node->tmp_vma; - nouveau_vm_map_sg(vma, 0, tmp_mem.num_pages << PAGE_SHIFT, - node, node->pages); - } - ret = nouveau_bo_move_m2mf(bo, true, intr, no_wait_reserve, no_wait_gpu, &tmp_mem); - - if (dev_priv->card_type >= NV_50) { - struct nouveau_bo *nvbo = nouveau_bo(bo); - nouveau_vm_unmap(&nvbo->vma); - } - if (ret) goto out; @@ -844,30 +801,22 @@ out: static void nouveau_bo_move_ntfy(struct ttm_buffer_object *bo, struct ttm_mem_reg *new_mem) { - struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); struct nouveau_mem *node = new_mem->mm_node; struct nouveau_bo *nvbo = nouveau_bo(bo); - struct nouveau_vma *vma = &nvbo->vma; - struct nouveau_vm *vm = vma->vm; - - if (dev_priv->card_type < NV_50) - return; - - switch (new_mem->mem_type) { - case TTM_PL_VRAM: - nouveau_vm_map(vma, node); - break; - case TTM_PL_TT: - if (vma->node->type != vm->spg_shift) { + struct nouveau_vma *vma; + + list_for_each_entry(vma, &nvbo->vma_list, head) { + if (new_mem->mem_type == TTM_PL_VRAM) { + nouveau_vm_map(vma, new_mem->mm_node); + } else + if (new_mem->mem_type == TTM_PL_TT && + nvbo->page_shift == vma->vm->spg_shift) { + nouveau_vm_map_sg(vma, 0, new_mem-> + num_pages << PAGE_SHIFT, + node, node->pages); + } else { nouveau_vm_unmap(vma); - vma = &node->tmp_vma; } - nouveau_vm_map_sg(vma, 0, new_mem->num_pages << PAGE_SHIFT, - node, node->pages); - break; - default: - nouveau_vm_unmap(&nvbo->vma); - break; } } @@ -1113,3 +1062,54 @@ struct ttm_bo_driver nouveau_bo_driver = { .io_mem_free = &nouveau_ttm_io_mem_free, }; +struct nouveau_vma * +nouveau_bo_vma_find(struct nouveau_bo *nvbo, struct nouveau_vm *vm) +{ + struct nouveau_vma *vma; + list_for_each_entry(vma, &nvbo->vma_list, head) { + if (vma->vm == vm) + return vma; + } + + return NULL; +} + +int +nouveau_bo_vma_add(struct nouveau_bo *nvbo, struct nouveau_vm *vm, + struct nouveau_vma *vma) +{ + const u32 size = nvbo->bo.mem.num_pages << PAGE_SHIFT; + struct nouveau_mem *node = nvbo->bo.mem.mm_node; + int ret; + + ret = nouveau_vm_get(vm, size, nvbo->page_shift, + NV_MEM_ACCESS_RW, vma); + if (ret) + return ret; + + if (nvbo->bo.mem.mem_type == TTM_PL_VRAM) + nouveau_vm_map(vma, nvbo->bo.mem.mm_node); + else + if (nvbo->bo.mem.mem_type == TTM_PL_TT) + nouveau_vm_map_sg(vma, 0, size, node, node->pages); + + list_add_tail(&vma->head, &nvbo->vma_list); + vma->refcount = 1; + return 0; +} + +void +nouveau_bo_vma_del(struct nouveau_bo *nvbo, struct nouveau_vma *vma) +{ + if (vma->node) { + if (nvbo->bo.mem.mem_type != TTM_PL_SYSTEM) { + spin_lock(&nvbo->bo.bdev->fence_lock); + ttm_bo_wait(&nvbo->bo, false, false, false); + spin_unlock(&nvbo->bo.bdev->fence_lock); + nouveau_vm_unmap(vma); + } + + nouveau_vm_put(vma); + list_del(&vma->head); + } +} diff --git a/drivers/gpu/drm/nouveau/nouveau_channel.c b/drivers/gpu/drm/nouveau/nouveau_channel.c index 4cea35c57d1..b0d753f45bb 100644 --- a/drivers/gpu/drm/nouveau/nouveau_channel.c +++ b/drivers/gpu/drm/nouveau/nouveau_channel.c @@ -27,40 +27,63 @@ #include "nouveau_drv.h" #include "nouveau_drm.h" #include "nouveau_dma.h" +#include "nouveau_ramht.h" static int -nouveau_channel_pushbuf_ctxdma_init(struct nouveau_channel *chan) +nouveau_channel_pushbuf_init(struct nouveau_channel *chan) { + u32 mem = nouveau_vram_pushbuf ? TTM_PL_FLAG_VRAM : TTM_PL_FLAG_TT; struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_bo *pb = chan->pushbuf_bo; - struct nouveau_gpuobj *pushbuf = NULL; - int ret = 0; + int ret; + + /* allocate buffer object */ + ret = nouveau_bo_new(dev, 65536, 0, mem, 0, 0, &chan->pushbuf_bo); + if (ret) + goto out; + + ret = nouveau_bo_pin(chan->pushbuf_bo, mem); + if (ret) + goto out; + + ret = nouveau_bo_map(chan->pushbuf_bo); + if (ret) + goto out; + /* create DMA object covering the entire memtype where the push + * buffer resides, userspace can submit its own push buffers from + * anywhere within the same memtype. + */ + chan->pushbuf_base = chan->pushbuf_bo->bo.offset; if (dev_priv->card_type >= NV_50) { + ret = nouveau_bo_vma_add(chan->pushbuf_bo, chan->vm, + &chan->pushbuf_vma); + if (ret) + goto out; + if (dev_priv->card_type < NV_C0) { ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, 0, (1ULL << 40), NV_MEM_ACCESS_RO, NV_MEM_TARGET_VM, - &pushbuf); + &chan->pushbuf); } - chan->pushbuf_base = pb->bo.offset; + chan->pushbuf_base = chan->pushbuf_vma.offset; } else - if (pb->bo.mem.mem_type == TTM_PL_TT) { + if (chan->pushbuf_bo->bo.mem.mem_type == TTM_PL_TT) { ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, 0, dev_priv->gart_info.aper_size, NV_MEM_ACCESS_RO, - NV_MEM_TARGET_GART, &pushbuf); - chan->pushbuf_base = pb->bo.mem.start << PAGE_SHIFT; + NV_MEM_TARGET_GART, + &chan->pushbuf); } else if (dev_priv->card_type != NV_04) { ret = nouveau_gpuobj_dma_new(chan, NV_CLASS_DMA_IN_MEMORY, 0, dev_priv->fb_available_size, NV_MEM_ACCESS_RO, - NV_MEM_TARGET_VRAM, &pushbuf); - chan->pushbuf_base = pb->bo.mem.start << PAGE_SHIFT; + NV_MEM_TARGET_VRAM, + &chan->pushbuf); } else { /* NV04 cmdbuf hack, from original ddx.. not sure of it's * exact reason for existing :) PCI access to cmdbuf in @@ -70,47 +93,22 @@ nouveau_channel_pushbuf_ctxdma_init(struct nouveau_channel *chan) pci_resource_start(dev->pdev, 1), dev_priv->fb_available_size, NV_MEM_ACCESS_RO, - NV_MEM_TARGET_PCI, &pushbuf); - chan->pushbuf_base = pb->bo.mem.start << PAGE_SHIFT; + NV_MEM_TARGET_PCI, + &chan->pushbuf); } - nouveau_gpuobj_ref(pushbuf, &chan->pushbuf); - nouveau_gpuobj_ref(NULL, &pushbuf); - return ret; -} - -static struct nouveau_bo * -nouveau_channel_user_pushbuf_alloc(struct drm_device *dev) -{ - struct nouveau_bo *pushbuf = NULL; - int location, ret; - - if (nouveau_vram_pushbuf) - location = TTM_PL_FLAG_VRAM; - else - location = TTM_PL_FLAG_TT; - - ret = nouveau_bo_new(dev, NULL, 65536, 0, location, 0, 0x0000, &pushbuf); - if (ret) { - NV_ERROR(dev, "error allocating DMA push buffer: %d\n", ret); - return NULL; - } - - ret = nouveau_bo_pin(pushbuf, location); +out: if (ret) { - NV_ERROR(dev, "error pinning DMA push buffer: %d\n", ret); - nouveau_bo_ref(NULL, &pushbuf); - return NULL; - } - - ret = nouveau_bo_map(pushbuf); - if (ret) { - nouveau_bo_unpin(pushbuf); - nouveau_bo_ref(NULL, &pushbuf); - return NULL; + NV_ERROR(dev, "error initialising pushbuf: %d\n", ret); + nouveau_bo_vma_del(chan->pushbuf_bo, &chan->pushbuf_vma); + nouveau_gpuobj_ref(NULL, &chan->pushbuf); + if (chan->pushbuf_bo) { + nouveau_bo_unmap(chan->pushbuf_bo); + nouveau_bo_ref(NULL, &chan->pushbuf_bo); + } } - return pushbuf; + return 0; } /* allocates and initializes a fifo for user space consumption */ @@ -121,6 +119,7 @@ nouveau_channel_alloc(struct drm_device *dev, struct nouveau_channel **chan_ret, { struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_fpriv *fpriv = nouveau_fpriv(file_priv); struct nouveau_channel *chan; unsigned long flags; int ret; @@ -160,19 +159,14 @@ nouveau_channel_alloc(struct drm_device *dev, struct nouveau_channel **chan_ret, INIT_LIST_HEAD(&chan->nvsw.flip); INIT_LIST_HEAD(&chan->fence.pending); - /* Allocate DMA push buffer */ - chan->pushbuf_bo = nouveau_channel_user_pushbuf_alloc(dev); - if (!chan->pushbuf_bo) { - ret = -ENOMEM; - NV_ERROR(dev, "pushbuf %d\n", ret); + /* setup channel's memory and vm */ + ret = nouveau_gpuobj_channel_init(chan, vram_handle, gart_handle); + if (ret) { + NV_ERROR(dev, "gpuobj %d\n", ret); nouveau_channel_put(&chan); return ret; } - nouveau_dma_pre_init(chan); - chan->user_put = 0x40; - chan->user_get = 0x44; - /* Allocate space for per-channel fixed notifier memory */ ret = nouveau_notifier_init_channel(chan); if (ret) { @@ -181,21 +175,17 @@ nouveau_channel_alloc(struct drm_device *dev, struct nouveau_channel **chan_ret, return ret; } - /* Setup channel's default objects */ - ret = nouveau_gpuobj_channel_init(chan, vram_handle, gart_handle); + /* Allocate DMA push buffer */ + ret = nouveau_channel_pushbuf_init(chan); if (ret) { - NV_ERROR(dev, "gpuobj %d\n", ret); + NV_ERROR(dev, "pushbuf %d\n", ret); nouveau_channel_put(&chan); return ret; } - /* Create a dma object for the push buffer */ - ret = nouveau_channel_pushbuf_ctxdma_init(chan); - if (ret) { - NV_ERROR(dev, "pbctxdma %d\n", ret); - nouveau_channel_put(&chan); - return ret; - } + nouveau_dma_pre_init(chan); + chan->user_put = 0x40; + chan->user_get = 0x44; /* disable the fifo caches */ pfifo->reassign(dev, false); @@ -220,6 +210,11 @@ nouveau_channel_alloc(struct drm_device *dev, struct nouveau_channel **chan_ret, nouveau_debugfs_channel_init(chan); NV_DEBUG(dev, "channel %d initialised\n", chan->id); + if (fpriv) { + spin_lock(&fpriv->lock); + list_add(&chan->list, &fpriv->channels); + spin_unlock(&fpriv->lock); + } *chan_ret = chan; return 0; } @@ -236,29 +231,23 @@ nouveau_channel_get_unlocked(struct nouveau_channel *ref) } struct nouveau_channel * -nouveau_channel_get(struct drm_device *dev, struct drm_file *file_priv, int id) +nouveau_channel_get(struct drm_file *file_priv, int id) { - struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fpriv *fpriv = nouveau_fpriv(file_priv); struct nouveau_channel *chan; - unsigned long flags; - - if (unlikely(id < 0 || id >= NOUVEAU_MAX_CHANNEL_NR)) - return ERR_PTR(-EINVAL); - - spin_lock_irqsave(&dev_priv->channels.lock, flags); - chan = nouveau_channel_get_unlocked(dev_priv->channels.ptr[id]); - spin_unlock_irqrestore(&dev_priv->channels.lock, flags); - - if (unlikely(!chan)) - return ERR_PTR(-EINVAL); - if (unlikely(file_priv && chan->file_priv != file_priv)) { - nouveau_channel_put_unlocked(&chan); - return ERR_PTR(-EINVAL); + spin_lock(&fpriv->lock); + list_for_each_entry(chan, &fpriv->channels, list) { + if (chan->id == id) { + chan = nouveau_channel_get_unlocked(chan); + spin_unlock(&fpriv->lock); + mutex_lock(&chan->mutex); + return chan; + } } + spin_unlock(&fpriv->lock); - mutex_lock(&chan->mutex); - return chan; + return ERR_PTR(-EINVAL); } void @@ -268,9 +257,8 @@ nouveau_channel_put_unlocked(struct nouveau_channel **pchan) struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct nouveau_crypt_engine *pcrypt = &dev_priv->engine.crypt; unsigned long flags; + int i; /* decrement the refcount, and we're done if there's still refs */ if (likely(!atomic_dec_and_test(&chan->users))) { @@ -294,19 +282,12 @@ nouveau_channel_put_unlocked(struct nouveau_channel **pchan) /* boot it off the hardware */ pfifo->reassign(dev, false); - /* We want to give pgraph a chance to idle and get rid of all - * potential errors. We need to do this without the context - * switch lock held, otherwise the irq handler is unable to - * process them. - */ - if (pgraph->channel(dev) == chan) - nouveau_wait_for_idle(dev); - /* destroy the engine specific contexts */ pfifo->destroy_context(chan); - pgraph->destroy_context(chan); - if (pcrypt->destroy_context) - pcrypt->destroy_context(chan); + for (i = 0; i < NVOBJ_ENGINE_NR; i++) { + if (chan->engctx[i]) + dev_priv->eng[i]->context_del(chan, i); + } pfifo->reassign(dev, true); @@ -320,12 +301,14 @@ nouveau_channel_put_unlocked(struct nouveau_channel **pchan) /* destroy any resources the channel owned */ nouveau_gpuobj_ref(NULL, &chan->pushbuf); if (chan->pushbuf_bo) { + nouveau_bo_vma_del(chan->pushbuf_bo, &chan->pushbuf_vma); nouveau_bo_unmap(chan->pushbuf_bo); nouveau_bo_unpin(chan->pushbuf_bo); nouveau_bo_ref(NULL, &chan->pushbuf_bo); } - nouveau_gpuobj_channel_takedown(chan); + nouveau_ramht_ref(NULL, &chan->ramht, chan); nouveau_notifier_takedown_channel(chan); + nouveau_gpuobj_channel_takedown(chan); nouveau_channel_ref(NULL, pchan); } @@ -391,10 +374,11 @@ nouveau_channel_cleanup(struct drm_device *dev, struct drm_file *file_priv) NV_DEBUG(dev, "clearing FIFO enables from file_priv\n"); for (i = 0; i < engine->fifo.channels; i++) { - chan = nouveau_channel_get(dev, file_priv, i); + chan = nouveau_channel_get(file_priv, i); if (IS_ERR(chan)) continue; + list_del(&chan->list); atomic_dec(&chan->users); nouveau_channel_put(&chan); } @@ -414,7 +398,7 @@ nouveau_ioctl_fifo_alloc(struct drm_device *dev, void *data, struct nouveau_channel *chan; int ret; - if (dev_priv->engine.graph.accel_blocked) + if (!dev_priv->eng[NVOBJ_ENGINE_GR]) return -ENODEV; if (init->fb_ctxdma_handle == ~0 || init->tt_ctxdma_handle == ~0) @@ -467,10 +451,11 @@ nouveau_ioctl_fifo_free(struct drm_device *dev, void *data, struct drm_nouveau_channel_free *req = data; struct nouveau_channel *chan; - chan = nouveau_channel_get(dev, file_priv, req->channel); + chan = nouveau_channel_get(file_priv, req->channel); if (IS_ERR(chan)) return PTR_ERR(chan); + list_del(&chan->list); atomic_dec(&chan->users); nouveau_channel_put(&chan); return 0; diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c index 7ae151109a6..939d4df0777 100644 --- a/drivers/gpu/drm/nouveau/nouveau_connector.c +++ b/drivers/gpu/drm/nouveau/nouveau_connector.c @@ -40,7 +40,7 @@ static void nouveau_connector_hotplug(void *, int); static struct nouveau_encoder * -find_encoder_by_type(struct drm_connector *connector, int type) +find_encoder(struct drm_connector *connector, int type) { struct drm_device *dev = connector->dev; struct nouveau_encoder *nv_encoder; @@ -170,8 +170,8 @@ nouveau_connector_of_detect(struct drm_connector *connector) struct device_node *cn, *dn = pci_device_to_OF_node(dev->pdev); if (!dn || - !((nv_encoder = find_encoder_by_type(connector, OUTPUT_TMDS)) || - (nv_encoder = find_encoder_by_type(connector, OUTPUT_ANALOG)))) + !((nv_encoder = find_encoder(connector, OUTPUT_TMDS)) || + (nv_encoder = find_encoder(connector, OUTPUT_ANALOG)))) return NULL; for_each_child_of_node(dn, cn) { @@ -233,6 +233,7 @@ nouveau_connector_detect(struct drm_connector *connector, bool force) struct drm_device *dev = connector->dev; struct nouveau_connector *nv_connector = nouveau_connector(connector); struct nouveau_encoder *nv_encoder = NULL; + struct nouveau_encoder *nv_partner; struct nouveau_i2c_chan *i2c; int type; @@ -266,19 +267,22 @@ nouveau_connector_detect(struct drm_connector *connector, bool force) * same i2c channel so the value returned from ddc_detect * isn't necessarily correct. */ - if (nv_connector->dcb->type == DCB_CONNECTOR_DVI_I) { + nv_partner = NULL; + if (nv_encoder->dcb->type == OUTPUT_TMDS) + nv_partner = find_encoder(connector, OUTPUT_ANALOG); + if (nv_encoder->dcb->type == OUTPUT_ANALOG) + nv_partner = find_encoder(connector, OUTPUT_TMDS); + + if (nv_partner && ((nv_encoder->dcb->type == OUTPUT_ANALOG && + nv_partner->dcb->type == OUTPUT_TMDS) || + (nv_encoder->dcb->type == OUTPUT_TMDS && + nv_partner->dcb->type == OUTPUT_ANALOG))) { if (nv_connector->edid->input & DRM_EDID_INPUT_DIGITAL) type = OUTPUT_TMDS; else type = OUTPUT_ANALOG; - nv_encoder = find_encoder_by_type(connector, type); - if (!nv_encoder) { - NV_ERROR(dev, "Detected %d encoder on %s, " - "but no object!\n", type, - drm_get_connector_name(connector)); - return connector_status_disconnected; - } + nv_encoder = find_encoder(connector, type); } nouveau_connector_set_encoder(connector, nv_encoder); @@ -292,9 +296,9 @@ nouveau_connector_detect(struct drm_connector *connector, bool force) } detect_analog: - nv_encoder = find_encoder_by_type(connector, OUTPUT_ANALOG); + nv_encoder = find_encoder(connector, OUTPUT_ANALOG); if (!nv_encoder && !nouveau_tv_disable) - nv_encoder = find_encoder_by_type(connector, OUTPUT_TV); + nv_encoder = find_encoder(connector, OUTPUT_TV); if (nv_encoder && force) { struct drm_encoder *encoder = to_drm_encoder(nv_encoder); struct drm_encoder_helper_funcs *helper = @@ -327,7 +331,7 @@ nouveau_connector_detect_lvds(struct drm_connector *connector, bool force) nv_connector->edid = NULL; } - nv_encoder = find_encoder_by_type(connector, OUTPUT_LVDS); + nv_encoder = find_encoder(connector, OUTPUT_LVDS); if (!nv_encoder) return connector_status_disconnected; @@ -405,7 +409,7 @@ nouveau_connector_force(struct drm_connector *connector) } else type = OUTPUT_ANY; - nv_encoder = find_encoder_by_type(connector, type); + nv_encoder = find_encoder(connector, type); if (!nv_encoder) { NV_ERROR(connector->dev, "can't find encoder to force %s on!\n", drm_get_connector_name(connector)); @@ -442,7 +446,7 @@ nouveau_connector_set_property(struct drm_connector *connector, } /* LVDS always needs gpu scaling */ - if (nv_connector->dcb->type == DCB_CONNECTOR_LVDS && + if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS && value == DRM_MODE_SCALE_NONE) return -EINVAL; @@ -650,6 +654,7 @@ nouveau_connector_get_modes(struct drm_connector *connector) ret = get_slave_funcs(encoder)->get_modes(encoder, connector); if (nv_connector->dcb->type == DCB_CONNECTOR_LVDS || + nv_connector->dcb->type == DCB_CONNECTOR_LVDS_SPWG || nv_connector->dcb->type == DCB_CONNECTOR_eDP) ret += nouveau_connector_scaler_modes_add(connector); @@ -810,6 +815,7 @@ nouveau_connector_create(struct drm_device *dev, int index) type = DRM_MODE_CONNECTOR_HDMIA; break; case DCB_CONNECTOR_LVDS: + case DCB_CONNECTOR_LVDS_SPWG: type = DRM_MODE_CONNECTOR_LVDS; funcs = &nouveau_connector_funcs_lvds; break; @@ -838,7 +844,7 @@ nouveau_connector_create(struct drm_device *dev, int index) drm_connector_helper_add(connector, &nouveau_connector_helper_funcs); /* Check if we need dithering enabled */ - if (dcb->type == DCB_CONNECTOR_LVDS) { + if (connector->connector_type == DRM_MODE_CONNECTOR_LVDS) { bool dummy, is_24bit = false; ret = nouveau_bios_parse_lvds_table(dev, 0, &dummy, &is_24bit); @@ -883,7 +889,7 @@ nouveau_connector_create(struct drm_device *dev, int index) nv_connector->use_dithering ? DRM_MODE_DITHERING_ON : DRM_MODE_DITHERING_OFF); - if (dcb->type != DCB_CONNECTOR_LVDS) { + if (connector->connector_type != DRM_MODE_CONNECTOR_LVDS) { if (dev_priv->card_type >= NV_50) connector->polled = DRM_CONNECTOR_POLL_HPD; else diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c b/drivers/gpu/drm/nouveau/nouveau_display.c index 764c15d537b..eb514ea2937 100644 --- a/drivers/gpu/drm/nouveau/nouveau_display.c +++ b/drivers/gpu/drm/nouveau/nouveau_display.c @@ -276,7 +276,7 @@ nouveau_crtc_page_flip(struct drm_crtc *crtc, struct drm_framebuffer *fb, struct nouveau_fence *fence; int ret; - if (dev_priv->engine.graph.accel_blocked) + if (!dev_priv->channel) return -ENODEV; s = kzalloc(sizeof(*s), GFP_KERNEL); diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.c b/drivers/gpu/drm/nouveau/nouveau_dma.c index 568caedd721..00bc6eaad55 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dma.c +++ b/drivers/gpu/drm/nouveau/nouveau_dma.c @@ -167,8 +167,13 @@ nv50_dma_push(struct nouveau_channel *chan, struct nouveau_bo *bo, int delta, int length) { struct nouveau_bo *pb = chan->pushbuf_bo; - uint64_t offset = bo->bo.offset + delta; + struct nouveau_vma *vma; int ip = (chan->dma.ib_put * 2) + chan->dma.ib_base; + u64 offset; + + vma = nouveau_bo_vma_find(bo, chan->vm); + BUG_ON(!vma); + offset = vma->offset + delta; BUG_ON(chan->dma.ib_free < 1); nouveau_bo_wr32(pb, ip++, lower_32_bits(offset)); diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.c b/drivers/gpu/drm/nouveau/nouveau_drv.c index 155ebdcbf06..b30ddd8d2e2 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drv.c +++ b/drivers/gpu/drm/nouveau/nouveau_drv.c @@ -73,7 +73,7 @@ int nouveau_ignorelid = 0; module_param_named(ignorelid, nouveau_ignorelid, int, 0400); MODULE_PARM_DESC(noaccel, "Disable all acceleration"); -int nouveau_noaccel = 0; +int nouveau_noaccel = -1; module_param_named(noaccel, nouveau_noaccel, int, 0400); MODULE_PARM_DESC(nofbaccel, "Disable fbcon acceleration"); @@ -119,6 +119,10 @@ MODULE_PARM_DESC(msi, "Enable MSI (default: off)\n"); int nouveau_msi; module_param_named(msi, nouveau_msi, int, 0400); +MODULE_PARM_DESC(ctxfw, "Use external HUB/GPC ucode (fermi)\n"); +int nouveau_ctxfw; +module_param_named(ctxfw, nouveau_ctxfw, int, 0400); + int nouveau_fbpercrtc; #if 0 module_param_named(fbpercrtc, nouveau_fbpercrtc, int, 0400); @@ -162,11 +166,10 @@ nouveau_pci_suspend(struct pci_dev *pdev, pm_message_t pm_state) struct drm_device *dev = pci_get_drvdata(pdev); struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; struct nouveau_channel *chan; struct drm_crtc *crtc; - int ret, i; + int ret, i, e; if (pm_state.event == PM_EVENT_PRETHAW) return 0; @@ -206,12 +209,20 @@ nouveau_pci_suspend(struct pci_dev *pdev, pm_message_t pm_state) nouveau_channel_idle(chan); } - pgraph->fifo_access(dev, false); - nouveau_wait_for_idle(dev); pfifo->reassign(dev, false); pfifo->disable(dev); pfifo->unload_context(dev); - pgraph->unload_context(dev); + + for (e = NVOBJ_ENGINE_NR - 1; e >= 0; e--) { + if (!dev_priv->eng[e]) + continue; + + ret = dev_priv->eng[e]->fini(dev, e, true); + if (ret) { + NV_ERROR(dev, "... engine %d failed: %d\n", i, ret); + goto out_abort; + } + } ret = pinstmem->suspend(dev); if (ret) { @@ -242,9 +253,12 @@ nouveau_pci_suspend(struct pci_dev *pdev, pm_message_t pm_state) out_abort: NV_INFO(dev, "Re-enabling acceleration..\n"); + for (e = e + 1; e < NVOBJ_ENGINE_NR; e++) { + if (dev_priv->eng[e]) + dev_priv->eng[e]->init(dev, e); + } pfifo->enable(dev); pfifo->reassign(dev, true); - pgraph->fifo_access(dev, true); return ret; } @@ -299,8 +313,10 @@ nouveau_pci_resume(struct pci_dev *pdev) engine->mc.init(dev); engine->timer.init(dev); engine->fb.init(dev); - engine->graph.init(dev); - engine->crypt.init(dev); + for (i = 0; i < NVOBJ_ENGINE_NR; i++) { + if (dev_priv->eng[i]) + dev_priv->eng[i]->init(dev, i); + } engine->fifo.init(dev); nouveau_irq_postinstall(dev); @@ -345,7 +361,7 @@ nouveau_pci_resume(struct pci_dev *pdev) list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) { struct nouveau_crtc *nv_crtc = nouveau_crtc(crtc); - u32 offset = nv_crtc->cursor.nvbo->bo.mem.start << PAGE_SHIFT; + u32 offset = nv_crtc->cursor.nvbo->bo.offset; nv_crtc->cursor.set_offset(nv_crtc, offset); nv_crtc->cursor.set_pos(nv_crtc, nv_crtc->cursor_saved_x, @@ -380,7 +396,9 @@ static struct drm_driver driver = { .firstopen = nouveau_firstopen, .lastclose = nouveau_lastclose, .unload = nouveau_unload, + .open = nouveau_open, .preclose = nouveau_preclose, + .postclose = nouveau_postclose, #if defined(CONFIG_DRM_NOUVEAU_DEBUG) .debugfs_init = nouveau_debugfs_init, .debugfs_cleanup = nouveau_debugfs_takedown, @@ -411,6 +429,8 @@ static struct drm_driver driver = { .gem_init_object = nouveau_gem_object_new, .gem_free_object = nouveau_gem_object_del, + .gem_open_object = nouveau_gem_object_open, + .gem_close_object = nouveau_gem_object_close, .name = DRIVER_NAME, .desc = DRIVER_DESC, diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h index a76514a209b..d7d51deb34b 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drv.h +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h @@ -46,9 +46,17 @@ #include "ttm/ttm_module.h" struct nouveau_fpriv { - struct ttm_object_file *tfile; + spinlock_t lock; + struct list_head channels; + struct nouveau_vm *vm; }; +static inline struct nouveau_fpriv * +nouveau_fpriv(struct drm_file *file_priv) +{ + return file_priv ? file_priv->driver_priv : NULL; +} + #define DRM_FILE_PAGE_OFFSET (0x100000000ULL >> PAGE_SHIFT) #include "nouveau_drm.h" @@ -69,7 +77,7 @@ struct nouveau_mem { struct drm_device *dev; struct nouveau_vma bar_vma; - struct nouveau_vma tmp_vma; + struct nouveau_vma vma[2]; u8 page_shift; struct drm_mm_node *tag; @@ -107,7 +115,8 @@ struct nouveau_bo { struct nouveau_channel *channel; - struct nouveau_vma vma; + struct list_head vma_list; + unsigned page_shift; uint32_t tile_mode; uint32_t tile_flags; @@ -150,13 +159,12 @@ enum nouveau_flags { #define NVOBJ_ENGINE_SW 0 #define NVOBJ_ENGINE_GR 1 -#define NVOBJ_ENGINE_PPP 2 -#define NVOBJ_ENGINE_COPY 3 -#define NVOBJ_ENGINE_VP 4 -#define NVOBJ_ENGINE_CRYPT 5 -#define NVOBJ_ENGINE_BSP 6 -#define NVOBJ_ENGINE_DISPLAY 0xcafe0001 -#define NVOBJ_ENGINE_INT 0xdeadbeef +#define NVOBJ_ENGINE_CRYPT 2 +#define NVOBJ_ENGINE_COPY0 3 +#define NVOBJ_ENGINE_COPY1 4 +#define NVOBJ_ENGINE_MPEG 5 +#define NVOBJ_ENGINE_DISPLAY 15 +#define NVOBJ_ENGINE_NR 16 #define NVOBJ_FLAG_DONT_MAP (1 << 0) #define NVOBJ_FLAG_ZERO_ALLOC (1 << 1) @@ -177,9 +185,10 @@ struct nouveau_gpuobj { uint32_t flags; u32 size; - u32 pinst; - u32 cinst; - u64 vinst; + u32 pinst; /* PRAMIN BAR offset */ + u32 cinst; /* Channel offset */ + u64 vinst; /* VRAM address */ + u64 linst; /* VM address */ uint32_t engine; uint32_t class; @@ -202,6 +211,7 @@ enum nouveau_channel_mutex_class { struct nouveau_channel { struct drm_device *dev; + struct list_head list; int id; /* references to the channel data structure */ @@ -229,15 +239,18 @@ struct nouveau_channel { uint32_t sequence; uint32_t sequence_ack; atomic_t last_sequence_irq; + struct nouveau_vma vma; } fence; /* DMA push buffer */ struct nouveau_gpuobj *pushbuf; struct nouveau_bo *pushbuf_bo; + struct nouveau_vma pushbuf_vma; uint32_t pushbuf_base; /* Notifier memory */ struct nouveau_bo *notifier_bo; + struct nouveau_vma notifier_vma; struct drm_mm notifier_heap; /* PFIFO context */ @@ -245,11 +258,8 @@ struct nouveau_channel { struct nouveau_gpuobj *cache; void *fifo_priv; - /* PGRAPH context */ - /* XXX may be merge 2 pointers as private data ??? */ - struct nouveau_gpuobj *ramin_grctx; - struct nouveau_gpuobj *crypt_ctx; - void *pgraph_ctx; + /* Execution engine contexts */ + void *engctx[NVOBJ_ENGINE_NR]; /* NV50 VM */ struct nouveau_vm *vm; @@ -282,6 +292,7 @@ struct nouveau_channel { uint32_t sw_subchannel[8]; + struct nouveau_vma dispc_vma[2]; struct { struct nouveau_gpuobj *vblsem; uint32_t vblsem_head; @@ -298,6 +309,18 @@ struct nouveau_channel { } debugfs; }; +struct nouveau_exec_engine { + void (*destroy)(struct drm_device *, int engine); + int (*init)(struct drm_device *, int engine); + int (*fini)(struct drm_device *, int engine, bool suspend); + int (*context_new)(struct nouveau_channel *, int engine); + void (*context_del)(struct nouveau_channel *, int engine); + int (*object_new)(struct nouveau_channel *, int engine, + u32 handle, u16 class); + void (*set_tile_region)(struct drm_device *dev, int i); + void (*tlb_flush)(struct drm_device *, int engine); +}; + struct nouveau_instmem_engine { void *priv; @@ -306,7 +329,8 @@ struct nouveau_instmem_engine { int (*suspend)(struct drm_device *dev); void (*resume)(struct drm_device *dev); - int (*get)(struct nouveau_gpuobj *, u32 size, u32 align); + int (*get)(struct nouveau_gpuobj *, struct nouveau_channel *, + u32 size, u32 align); void (*put)(struct nouveau_gpuobj *); int (*map)(struct nouveau_gpuobj *); void (*unmap)(struct nouveau_gpuobj *); @@ -364,30 +388,6 @@ struct nouveau_fifo_engine { void (*tlb_flush)(struct drm_device *dev); }; -struct nouveau_pgraph_engine { - bool accel_blocked; - bool registered; - int grctx_size; - void *priv; - - /* NV2x/NV3x context table (0x400780) */ - struct nouveau_gpuobj *ctx_table; - - int (*init)(struct drm_device *); - void (*takedown)(struct drm_device *); - - void (*fifo_access)(struct drm_device *, bool); - - struct nouveau_channel *(*channel)(struct drm_device *); - int (*create_context)(struct nouveau_channel *); - void (*destroy_context)(struct nouveau_channel *); - int (*load_context)(struct nouveau_channel *); - int (*unload_context)(struct drm_device *); - void (*tlb_flush)(struct drm_device *dev); - - void (*set_tile_region)(struct drm_device *dev, int i); -}; - struct nouveau_display_engine { void *priv; int (*early_init)(struct drm_device *); @@ -426,6 +426,19 @@ struct nouveau_pm_voltage { int nr_level; }; +struct nouveau_pm_memtiming { + int id; + u32 reg_100220; + u32 reg_100224; + u32 reg_100228; + u32 reg_10022c; + u32 reg_100230; + u32 reg_100234; + u32 reg_100238; + u32 reg_10023c; + u32 reg_100240; +}; + #define NOUVEAU_PM_MAX_LEVEL 8 struct nouveau_pm_level { struct device_attribute dev_attr; @@ -436,19 +449,21 @@ struct nouveau_pm_level { u32 memory; u32 shader; u32 unk05; + u32 unk0a; u8 voltage; u8 fanspeed; u16 memscript; + struct nouveau_pm_memtiming *timing; }; struct nouveau_pm_temp_sensor_constants { u16 offset_constant; s16 offset_mult; - u16 offset_div; - u16 slope_mult; - u16 slope_div; + s16 offset_div; + s16 slope_mult; + s16 slope_div; }; struct nouveau_pm_threshold_temp { @@ -457,17 +472,6 @@ struct nouveau_pm_threshold_temp { s16 fan_boost; }; -struct nouveau_pm_memtiming { - u32 reg_100220; - u32 reg_100224; - u32 reg_100228; - u32 reg_10022c; - u32 reg_100230; - u32 reg_100234; - u32 reg_100238; - u32 reg_10023c; -}; - struct nouveau_pm_memtimings { bool supported; struct nouveau_pm_memtiming *timing; @@ -499,18 +503,11 @@ struct nouveau_pm_engine { int (*temp_get)(struct drm_device *); }; -struct nouveau_crypt_engine { - bool registered; - - int (*init)(struct drm_device *); - void (*takedown)(struct drm_device *); - int (*create_context)(struct nouveau_channel *); - void (*destroy_context)(struct nouveau_channel *); - void (*tlb_flush)(struct drm_device *dev); -}; - struct nouveau_vram_engine { + struct nouveau_mm *mm; + int (*init)(struct drm_device *); + void (*takedown)(struct drm_device *dev); int (*get)(struct drm_device *, u64, u32 align, u32 size_nc, u32 type, struct nouveau_mem **); void (*put)(struct drm_device *, struct nouveau_mem **); @@ -523,12 +520,10 @@ struct nouveau_engine { struct nouveau_mc_engine mc; struct nouveau_timer_engine timer; struct nouveau_fb_engine fb; - struct nouveau_pgraph_engine graph; struct nouveau_fifo_engine fifo; struct nouveau_display_engine display; struct nouveau_gpio_engine gpio; struct nouveau_pm_engine pm; - struct nouveau_crypt_engine crypt; struct nouveau_vram_engine vram; }; @@ -632,11 +627,13 @@ enum nouveau_card_type { struct drm_nouveau_private { struct drm_device *dev; + bool noaccel; /* the card type, takes NV_* as values */ enum nouveau_card_type card_type; /* exact chipset, derived from NV_PMC_BOOT_0 */ int chipset; + int stepping; int flags; void __iomem *mmio; @@ -647,6 +644,7 @@ struct drm_nouveau_private { u32 ramin_base; bool ramin_available; struct drm_mm ramin_heap; + struct nouveau_exec_engine *eng[NVOBJ_ENGINE_NR]; struct list_head gpuobj_list; struct list_head classes; @@ -722,7 +720,6 @@ struct drm_nouveau_private { /* VRAM/fb configuration */ uint64_t vram_size; uint64_t vram_sys_base; - u32 vram_rblock_size; uint64_t fb_phys; uint64_t fb_available_size; @@ -745,10 +742,6 @@ struct drm_nouveau_private { uint32_t crtc_owner; uint32_t dac_users[4]; - struct nouveau_suspend_resume { - uint32_t *ramin_copy; - } susres; - struct backlight_device *backlight; struct { @@ -757,8 +750,6 @@ struct drm_nouveau_private { struct nouveau_fbdev *nfbdev; struct apertures_struct *apertures; - - bool powered_down; }; static inline struct drm_nouveau_private * @@ -812,12 +803,15 @@ extern int nouveau_override_conntype; extern char *nouveau_perflvl; extern int nouveau_perflvl_wr; extern int nouveau_msi; +extern int nouveau_ctxfw; extern int nouveau_pci_suspend(struct pci_dev *pdev, pm_message_t pm_state); extern int nouveau_pci_resume(struct pci_dev *pdev); /* nouveau_state.c */ +extern int nouveau_open(struct drm_device *, struct drm_file *); extern void nouveau_preclose(struct drm_device *dev, struct drm_file *); +extern void nouveau_postclose(struct drm_device *, struct drm_file *); extern int nouveau_load(struct drm_device *, unsigned long flags); extern int nouveau_firstopen(struct drm_device *); extern void nouveau_lastclose(struct drm_device *); @@ -875,7 +869,7 @@ extern int nouveau_channel_alloc(struct drm_device *dev, extern struct nouveau_channel * nouveau_channel_get_unlocked(struct nouveau_channel *); extern struct nouveau_channel * -nouveau_channel_get(struct drm_device *, struct drm_file *, int id); +nouveau_channel_get(struct drm_file *, int id); extern void nouveau_channel_put_unlocked(struct nouveau_channel **); extern void nouveau_channel_put(struct nouveau_channel **); extern void nouveau_channel_ref(struct nouveau_channel *chan, @@ -883,17 +877,27 @@ extern void nouveau_channel_ref(struct nouveau_channel *chan, extern void nouveau_channel_idle(struct nouveau_channel *chan); /* nouveau_object.c */ -#define NVOBJ_CLASS(d,c,e) do { \ +#define NVOBJ_ENGINE_ADD(d, e, p) do { \ + struct drm_nouveau_private *dev_priv = (d)->dev_private; \ + dev_priv->eng[NVOBJ_ENGINE_##e] = (p); \ +} while (0) + +#define NVOBJ_ENGINE_DEL(d, e) do { \ + struct drm_nouveau_private *dev_priv = (d)->dev_private; \ + dev_priv->eng[NVOBJ_ENGINE_##e] = NULL; \ +} while (0) + +#define NVOBJ_CLASS(d, c, e) do { \ int ret = nouveau_gpuobj_class_new((d), (c), NVOBJ_ENGINE_##e); \ if (ret) \ return ret; \ -} while(0) +} while (0) -#define NVOBJ_MTHD(d,c,m,e) do { \ +#define NVOBJ_MTHD(d, c, m, e) do { \ int ret = nouveau_gpuobj_mthd_new((d), (c), (m), (e)); \ if (ret) \ return ret; \ -} while(0) +} while (0) extern int nouveau_gpuobj_early_init(struct drm_device *); extern int nouveau_gpuobj_init(struct drm_device *); @@ -903,7 +907,7 @@ extern void nouveau_gpuobj_resume(struct drm_device *dev); extern int nouveau_gpuobj_class_new(struct drm_device *, u32 class, u32 eng); extern int nouveau_gpuobj_mthd_new(struct drm_device *, u32 class, u32 mthd, int (*exec)(struct nouveau_channel *, - u32 class, u32 mthd, u32 data)); + u32 class, u32 mthd, u32 data)); extern int nouveau_gpuobj_mthd_call(struct nouveau_channel *, u32, u32, u32); extern int nouveau_gpuobj_mthd_call2(struct drm_device *, int, u32, u32, u32); extern int nouveau_gpuobj_channel_init(struct nouveau_channel *, @@ -1137,88 +1141,57 @@ extern int nvc0_fifo_load_context(struct nouveau_channel *); extern int nvc0_fifo_unload_context(struct drm_device *); /* nv04_graph.c */ -extern int nv04_graph_init(struct drm_device *); -extern void nv04_graph_takedown(struct drm_device *); -extern void nv04_graph_fifo_access(struct drm_device *, bool); -extern struct nouveau_channel *nv04_graph_channel(struct drm_device *); -extern int nv04_graph_create_context(struct nouveau_channel *); -extern void nv04_graph_destroy_context(struct nouveau_channel *); -extern int nv04_graph_load_context(struct nouveau_channel *); -extern int nv04_graph_unload_context(struct drm_device *); +extern int nv04_graph_create(struct drm_device *); +extern int nv04_graph_object_new(struct nouveau_channel *, int, u32, u16); extern int nv04_graph_mthd_page_flip(struct nouveau_channel *chan, u32 class, u32 mthd, u32 data); extern struct nouveau_bitfield nv04_graph_nsource[]; /* nv10_graph.c */ -extern int nv10_graph_init(struct drm_device *); -extern void nv10_graph_takedown(struct drm_device *); +extern int nv10_graph_create(struct drm_device *); extern struct nouveau_channel *nv10_graph_channel(struct drm_device *); -extern int nv10_graph_create_context(struct nouveau_channel *); -extern void nv10_graph_destroy_context(struct nouveau_channel *); -extern int nv10_graph_load_context(struct nouveau_channel *); -extern int nv10_graph_unload_context(struct drm_device *); -extern void nv10_graph_set_tile_region(struct drm_device *dev, int i); extern struct nouveau_bitfield nv10_graph_intr[]; extern struct nouveau_bitfield nv10_graph_nstatus[]; /* nv20_graph.c */ -extern int nv20_graph_create_context(struct nouveau_channel *); -extern void nv20_graph_destroy_context(struct nouveau_channel *); -extern int nv20_graph_load_context(struct nouveau_channel *); -extern int nv20_graph_unload_context(struct drm_device *); -extern int nv20_graph_init(struct drm_device *); -extern void nv20_graph_takedown(struct drm_device *); -extern int nv30_graph_init(struct drm_device *); -extern void nv20_graph_set_tile_region(struct drm_device *dev, int i); +extern int nv20_graph_create(struct drm_device *); /* nv40_graph.c */ -extern int nv40_graph_init(struct drm_device *); -extern void nv40_graph_takedown(struct drm_device *); -extern struct nouveau_channel *nv40_graph_channel(struct drm_device *); -extern int nv40_graph_create_context(struct nouveau_channel *); -extern void nv40_graph_destroy_context(struct nouveau_channel *); -extern int nv40_graph_load_context(struct nouveau_channel *); -extern int nv40_graph_unload_context(struct drm_device *); +extern int nv40_graph_create(struct drm_device *); extern void nv40_grctx_init(struct nouveau_grctx *); -extern void nv40_graph_set_tile_region(struct drm_device *dev, int i); /* nv50_graph.c */ -extern int nv50_graph_init(struct drm_device *); -extern void nv50_graph_takedown(struct drm_device *); -extern void nv50_graph_fifo_access(struct drm_device *, bool); -extern struct nouveau_channel *nv50_graph_channel(struct drm_device *); -extern int nv50_graph_create_context(struct nouveau_channel *); -extern void nv50_graph_destroy_context(struct nouveau_channel *); -extern int nv50_graph_load_context(struct nouveau_channel *); -extern int nv50_graph_unload_context(struct drm_device *); +extern int nv50_graph_create(struct drm_device *); extern int nv50_grctx_init(struct nouveau_grctx *); -extern void nv50_graph_tlb_flush(struct drm_device *dev); -extern void nv84_graph_tlb_flush(struct drm_device *dev); extern struct nouveau_enum nv50_data_error_names[]; +extern int nv50_graph_isr_chid(struct drm_device *dev, u64 inst); /* nvc0_graph.c */ -extern int nvc0_graph_init(struct drm_device *); -extern void nvc0_graph_takedown(struct drm_device *); -extern void nvc0_graph_fifo_access(struct drm_device *, bool); -extern struct nouveau_channel *nvc0_graph_channel(struct drm_device *); -extern int nvc0_graph_create_context(struct nouveau_channel *); -extern void nvc0_graph_destroy_context(struct nouveau_channel *); -extern int nvc0_graph_load_context(struct nouveau_channel *); -extern int nvc0_graph_unload_context(struct drm_device *); +extern int nvc0_graph_create(struct drm_device *); +extern int nvc0_graph_isr_chid(struct drm_device *dev, u64 inst); /* nv84_crypt.c */ -extern int nv84_crypt_init(struct drm_device *dev); -extern void nv84_crypt_fini(struct drm_device *dev); -extern int nv84_crypt_create_context(struct nouveau_channel *); -extern void nv84_crypt_destroy_context(struct nouveau_channel *); -extern void nv84_crypt_tlb_flush(struct drm_device *dev); +extern int nv84_crypt_create(struct drm_device *); + +/* nva3_copy.c */ +extern int nva3_copy_create(struct drm_device *dev); + +/* nvc0_copy.c */ +extern int nvc0_copy_create(struct drm_device *dev, int engine); + +/* nv40_mpeg.c */ +extern int nv40_mpeg_create(struct drm_device *dev); + +/* nv50_mpeg.c */ +extern int nv50_mpeg_create(struct drm_device *dev); /* nv04_instmem.c */ extern int nv04_instmem_init(struct drm_device *); extern void nv04_instmem_takedown(struct drm_device *); extern int nv04_instmem_suspend(struct drm_device *); extern void nv04_instmem_resume(struct drm_device *); -extern int nv04_instmem_get(struct nouveau_gpuobj *, u32 size, u32 align); +extern int nv04_instmem_get(struct nouveau_gpuobj *, struct nouveau_channel *, + u32 size, u32 align); extern void nv04_instmem_put(struct nouveau_gpuobj *); extern int nv04_instmem_map(struct nouveau_gpuobj *); extern void nv04_instmem_unmap(struct nouveau_gpuobj *); @@ -1229,7 +1202,8 @@ extern int nv50_instmem_init(struct drm_device *); extern void nv50_instmem_takedown(struct drm_device *); extern int nv50_instmem_suspend(struct drm_device *); extern void nv50_instmem_resume(struct drm_device *); -extern int nv50_instmem_get(struct nouveau_gpuobj *, u32 size, u32 align); +extern int nv50_instmem_get(struct nouveau_gpuobj *, struct nouveau_channel *, + u32 size, u32 align); extern void nv50_instmem_put(struct nouveau_gpuobj *); extern int nv50_instmem_map(struct nouveau_gpuobj *); extern void nv50_instmem_unmap(struct nouveau_gpuobj *); @@ -1296,10 +1270,9 @@ extern int nv04_crtc_create(struct drm_device *, int index); /* nouveau_bo.c */ extern struct ttm_bo_driver nouveau_bo_driver; -extern int nouveau_bo_new(struct drm_device *, struct nouveau_channel *, - int size, int align, uint32_t flags, - uint32_t tile_mode, uint32_t tile_flags, - struct nouveau_bo **); +extern int nouveau_bo_new(struct drm_device *, int size, int align, + uint32_t flags, uint32_t tile_mode, + uint32_t tile_flags, struct nouveau_bo **); extern int nouveau_bo_pin(struct nouveau_bo *, uint32_t flags); extern int nouveau_bo_unpin(struct nouveau_bo *); extern int nouveau_bo_map(struct nouveau_bo *); @@ -1314,6 +1287,12 @@ extern void nouveau_bo_fence(struct nouveau_bo *, struct nouveau_fence *); extern int nouveau_bo_validate(struct nouveau_bo *, bool interruptible, bool no_wait_reserve, bool no_wait_gpu); +extern struct nouveau_vma * +nouveau_bo_vma_find(struct nouveau_bo *, struct nouveau_vm *); +extern int nouveau_bo_vma_add(struct nouveau_bo *, struct nouveau_vm *, + struct nouveau_vma *); +extern void nouveau_bo_vma_del(struct nouveau_bo *, struct nouveau_vma *); + /* nouveau_fence.c */ struct nouveau_fence; extern int nouveau_fence_init(struct drm_device *); @@ -1359,12 +1338,14 @@ static inline struct nouveau_fence *nouveau_fence_ref(struct nouveau_fence *obj) } /* nouveau_gem.c */ -extern int nouveau_gem_new(struct drm_device *, struct nouveau_channel *, - int size, int align, uint32_t domain, - uint32_t tile_mode, uint32_t tile_flags, - struct nouveau_bo **); +extern int nouveau_gem_new(struct drm_device *, int size, int align, + uint32_t domain, uint32_t tile_mode, + uint32_t tile_flags, struct nouveau_bo **); extern int nouveau_gem_object_new(struct drm_gem_object *); extern void nouveau_gem_object_del(struct drm_gem_object *); +extern int nouveau_gem_object_open(struct drm_gem_object *, struct drm_file *); +extern void nouveau_gem_object_close(struct drm_gem_object *, + struct drm_file *); extern int nouveau_gem_ioctl_new(struct drm_device *, void *, struct drm_file *); extern int nouveau_gem_ioctl_pushbuf(struct drm_device *, void *, @@ -1402,8 +1383,8 @@ bool nv50_gpio_irq_enable(struct drm_device *, enum dcb_gpio_tag, bool on); /* nv50_calc. */ int nv50_calc_pll(struct drm_device *, struct pll_lims *, int clk, int *N1, int *M1, int *N2, int *M2, int *P); -int nv50_calc_pll2(struct drm_device *, struct pll_lims *, - int clk, int *N, int *fN, int *M, int *P); +int nva3_calc_pll(struct drm_device *, struct pll_lims *, + int clk, int *N, int *fN, int *M, int *P); #ifndef ioread32_native #ifdef __BIG_ENDIAN @@ -1579,6 +1560,13 @@ nv_match_device(struct drm_device *dev, unsigned device, dev->pdev->subsystem_device == sub_device; } +static inline void * +nv_engine(struct drm_device *dev, int engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + return (void *)dev_priv->eng[engine]; +} + /* returns 1 if device is one of the nv4x using the 0x4497 object class, * helpful to determine a number of other hardware features */ diff --git a/drivers/gpu/drm/nouveau/nouveau_fb.h b/drivers/gpu/drm/nouveau/nouveau_fb.h index a3a88ad00f8..95c843e684b 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fb.h +++ b/drivers/gpu/drm/nouveau/nouveau_fb.h @@ -30,6 +30,7 @@ struct nouveau_framebuffer { struct drm_framebuffer base; struct nouveau_bo *nvbo; + struct nouveau_vma vma; u32 r_dma; u32 r_format; u32 r_pitch; diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.c b/drivers/gpu/drm/nouveau/nouveau_fbcon.c index 39aee6d4daf..14a8627efe4 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fbcon.c +++ b/drivers/gpu/drm/nouveau/nouveau_fbcon.c @@ -279,6 +279,7 @@ nouveau_fbcon_create(struct nouveau_fbdev *nfbdev, struct fb_info *info; struct drm_framebuffer *fb; struct nouveau_framebuffer *nouveau_fb; + struct nouveau_channel *chan; struct nouveau_bo *nvbo; struct drm_mode_fb_cmd mode_cmd; struct pci_dev *pdev = dev->pdev; @@ -296,8 +297,8 @@ nouveau_fbcon_create(struct nouveau_fbdev *nfbdev, size = mode_cmd.pitch * mode_cmd.height; size = roundup(size, PAGE_SIZE); - ret = nouveau_gem_new(dev, dev_priv->channel, size, 0, - NOUVEAU_GEM_DOMAIN_VRAM, 0, 0x0000, &nvbo); + ret = nouveau_gem_new(dev, size, 0, NOUVEAU_GEM_DOMAIN_VRAM, + 0, 0x0000, &nvbo); if (ret) { NV_ERROR(dev, "failed to allocate framebuffer\n"); goto out; @@ -318,6 +319,15 @@ nouveau_fbcon_create(struct nouveau_fbdev *nfbdev, goto out; } + chan = nouveau_nofbaccel ? NULL : dev_priv->channel; + if (chan && dev_priv->card_type >= NV_50) { + ret = nouveau_bo_vma_add(nvbo, chan->vm, &nfbdev->nouveau_fb.vma); + if (ret) { + NV_ERROR(dev, "failed to map fb into chan: %d\n", ret); + chan = NULL; + } + } + mutex_lock(&dev->struct_mutex); info = framebuffer_alloc(0, device); @@ -448,6 +458,7 @@ nouveau_fbcon_destroy(struct drm_device *dev, struct nouveau_fbdev *nfbdev) if (nouveau_fb->nvbo) { nouveau_bo_unmap(nouveau_fb->nvbo); + nouveau_bo_vma_del(nouveau_fb->nvbo, &nouveau_fb->vma); drm_gem_object_unreference_unlocked(nouveau_fb->nvbo->gem); nouveau_fb->nvbo = NULL; } diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.c b/drivers/gpu/drm/nouveau/nouveau_fence.c index 4b9f4493c9f..8d02d875376 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fence.c +++ b/drivers/gpu/drm/nouveau/nouveau_fence.c @@ -336,34 +336,32 @@ semaphore_acquire(struct nouveau_channel *chan, struct nouveau_semaphore *sema) { struct drm_nouveau_private *dev_priv = chan->dev->dev_private; struct nouveau_fence *fence = NULL; + u64 offset = chan->fence.vma.offset + sema->mem->start; int ret; if (dev_priv->chipset < 0x84) { - ret = RING_SPACE(chan, 3); + ret = RING_SPACE(chan, 4); if (ret) return ret; - BEGIN_RING(chan, NvSubSw, NV_SW_SEMAPHORE_OFFSET, 2); - OUT_RING (chan, sema->mem->start); + BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 3); + OUT_RING (chan, NvSema); + OUT_RING (chan, offset); OUT_RING (chan, 1); } else if (dev_priv->chipset < 0xc0) { - struct nouveau_vma *vma = &dev_priv->fence.bo->vma; - u64 offset = vma->offset + sema->mem->start; - - ret = RING_SPACE(chan, 5); + ret = RING_SPACE(chan, 7); if (ret) return ret; + BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 1); + OUT_RING (chan, chan->vram_handle); BEGIN_RING(chan, NvSubSw, 0x0010, 4); OUT_RING (chan, upper_32_bits(offset)); OUT_RING (chan, lower_32_bits(offset)); OUT_RING (chan, 1); OUT_RING (chan, 1); /* ACQUIRE_EQ */ } else { - struct nouveau_vma *vma = &dev_priv->fence.bo->vma; - u64 offset = vma->offset + sema->mem->start; - ret = RING_SPACE(chan, 5); if (ret) return ret; @@ -391,35 +389,33 @@ semaphore_release(struct nouveau_channel *chan, struct nouveau_semaphore *sema) { struct drm_nouveau_private *dev_priv = chan->dev->dev_private; struct nouveau_fence *fence = NULL; + u64 offset = chan->fence.vma.offset + sema->mem->start; int ret; if (dev_priv->chipset < 0x84) { - ret = RING_SPACE(chan, 4); + ret = RING_SPACE(chan, 5); if (ret) return ret; - BEGIN_RING(chan, NvSubSw, NV_SW_SEMAPHORE_OFFSET, 1); - OUT_RING (chan, sema->mem->start); + BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 2); + OUT_RING (chan, NvSema); + OUT_RING (chan, offset); BEGIN_RING(chan, NvSubSw, NV_SW_SEMAPHORE_RELEASE, 1); OUT_RING (chan, 1); } else if (dev_priv->chipset < 0xc0) { - struct nouveau_vma *vma = &dev_priv->fence.bo->vma; - u64 offset = vma->offset + sema->mem->start; - - ret = RING_SPACE(chan, 5); + ret = RING_SPACE(chan, 7); if (ret) return ret; + BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 1); + OUT_RING (chan, chan->vram_handle); BEGIN_RING(chan, NvSubSw, 0x0010, 4); OUT_RING (chan, upper_32_bits(offset)); OUT_RING (chan, lower_32_bits(offset)); OUT_RING (chan, 1); OUT_RING (chan, 2); /* RELEASE */ } else { - struct nouveau_vma *vma = &dev_priv->fence.bo->vma; - u64 offset = vma->offset + sema->mem->start; - ret = RING_SPACE(chan, 5); if (ret) return ret; @@ -504,22 +500,22 @@ nouveau_fence_channel_init(struct nouveau_channel *chan) struct nouveau_gpuobj *obj = NULL; int ret; - if (dev_priv->card_type >= NV_C0) - goto out_initialised; + if (dev_priv->card_type < NV_C0) { + /* Create an NV_SW object for various sync purposes */ + ret = nouveau_gpuobj_gr_new(chan, NvSw, NV_SW); + if (ret) + return ret; - /* Create an NV_SW object for various sync purposes */ - ret = nouveau_gpuobj_gr_new(chan, NvSw, NV_SW); - if (ret) - return ret; + ret = RING_SPACE(chan, 2); + if (ret) + return ret; - /* we leave subchannel empty for nvc0 */ - ret = RING_SPACE(chan, 2); - if (ret) - return ret; - BEGIN_RING(chan, NvSubSw, 0, 1); - OUT_RING(chan, NvSw); + BEGIN_RING(chan, NvSubSw, 0, 1); + OUT_RING (chan, NvSw); + FIRE_RING (chan); + } - /* Create a DMA object for the shared cross-channel sync area. */ + /* Setup area of memory shared between all channels for x-chan sync */ if (USE_SEMA(dev) && dev_priv->chipset < 0x84) { struct ttm_mem_reg *mem = &dev_priv->fence.bo->bo.mem; @@ -534,23 +530,14 @@ nouveau_fence_channel_init(struct nouveau_channel *chan) nouveau_gpuobj_ref(NULL, &obj); if (ret) return ret; - - ret = RING_SPACE(chan, 2); - if (ret) - return ret; - BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 1); - OUT_RING(chan, NvSema); } else { - ret = RING_SPACE(chan, 2); + /* map fence bo into channel's vm */ + ret = nouveau_bo_vma_add(dev_priv->fence.bo, chan->vm, + &chan->fence.vma); if (ret) return ret; - BEGIN_RING(chan, NvSubSw, NV_SW_DMA_SEMAPHORE, 1); - OUT_RING (chan, chan->vram_handle); /* whole VM */ } - FIRE_RING(chan); - -out_initialised: INIT_LIST_HEAD(&chan->fence.pending); spin_lock_init(&chan->fence.lock); atomic_set(&chan->fence.last_sequence_irq, 0); @@ -560,10 +547,10 @@ out_initialised: void nouveau_fence_channel_fini(struct nouveau_channel *chan) { + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; struct nouveau_fence *tmp, *fence; spin_lock(&chan->fence.lock); - list_for_each_entry_safe(fence, tmp, &chan->fence.pending, entry) { fence->signalled = true; list_del(&fence->entry); @@ -573,8 +560,9 @@ nouveau_fence_channel_fini(struct nouveau_channel *chan) kref_put(&fence->refcount, nouveau_fence_del); } - spin_unlock(&chan->fence.lock); + + nouveau_bo_vma_del(dev_priv->fence.bo, &chan->fence.vma); } int @@ -586,7 +574,7 @@ nouveau_fence_init(struct drm_device *dev) /* Create a shared VRAM heap for cross-channel sync. */ if (USE_SEMA(dev)) { - ret = nouveau_bo_new(dev, NULL, size, 0, TTM_PL_FLAG_VRAM, + ret = nouveau_bo_new(dev, size, 0, TTM_PL_FLAG_VRAM, 0, 0, &dev_priv->fence.bo); if (ret) return ret; diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c index b52e4601824..5f0bc57fdaa 100644 --- a/drivers/gpu/drm/nouveau/nouveau_gem.c +++ b/drivers/gpu/drm/nouveau/nouveau_gem.c @@ -60,9 +60,71 @@ nouveau_gem_object_del(struct drm_gem_object *gem) } int -nouveau_gem_new(struct drm_device *dev, struct nouveau_channel *chan, - int size, int align, uint32_t domain, uint32_t tile_mode, - uint32_t tile_flags, struct nouveau_bo **pnvbo) +nouveau_gem_object_open(struct drm_gem_object *gem, struct drm_file *file_priv) +{ + struct nouveau_fpriv *fpriv = nouveau_fpriv(file_priv); + struct nouveau_bo *nvbo = nouveau_gem_object(gem); + struct nouveau_vma *vma; + int ret; + + if (!fpriv->vm) + return 0; + + ret = ttm_bo_reserve(&nvbo->bo, false, false, false, 0); + if (ret) + return ret; + + vma = nouveau_bo_vma_find(nvbo, fpriv->vm); + if (!vma) { + vma = kzalloc(sizeof(*vma), GFP_KERNEL); + if (!vma) { + ret = -ENOMEM; + goto out; + } + + ret = nouveau_bo_vma_add(nvbo, fpriv->vm, vma); + if (ret) { + kfree(vma); + goto out; + } + } else { + vma->refcount++; + } + +out: + ttm_bo_unreserve(&nvbo->bo); + return ret; +} + +void +nouveau_gem_object_close(struct drm_gem_object *gem, struct drm_file *file_priv) +{ + struct nouveau_fpriv *fpriv = nouveau_fpriv(file_priv); + struct nouveau_bo *nvbo = nouveau_gem_object(gem); + struct nouveau_vma *vma; + int ret; + + if (!fpriv->vm) + return; + + ret = ttm_bo_reserve(&nvbo->bo, false, false, false, 0); + if (ret) + return; + + vma = nouveau_bo_vma_find(nvbo, fpriv->vm); + if (vma) { + if (--vma->refcount == 0) { + nouveau_bo_vma_del(nvbo, vma); + kfree(vma); + } + } + ttm_bo_unreserve(&nvbo->bo); +} + +int +nouveau_gem_new(struct drm_device *dev, int size, int align, uint32_t domain, + uint32_t tile_mode, uint32_t tile_flags, + struct nouveau_bo **pnvbo) { struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_bo *nvbo; @@ -76,7 +138,7 @@ nouveau_gem_new(struct drm_device *dev, struct nouveau_channel *chan, if (!flags || domain & NOUVEAU_GEM_DOMAIN_CPU) flags |= TTM_PL_FLAG_SYSTEM; - ret = nouveau_bo_new(dev, chan, size, align, flags, tile_mode, + ret = nouveau_bo_new(dev, size, align, flags, tile_mode, tile_flags, pnvbo); if (ret) return ret; @@ -103,17 +165,28 @@ nouveau_gem_new(struct drm_device *dev, struct nouveau_channel *chan, } static int -nouveau_gem_info(struct drm_gem_object *gem, struct drm_nouveau_gem_info *rep) +nouveau_gem_info(struct drm_file *file_priv, struct drm_gem_object *gem, + struct drm_nouveau_gem_info *rep) { + struct nouveau_fpriv *fpriv = nouveau_fpriv(file_priv); struct nouveau_bo *nvbo = nouveau_gem_object(gem); + struct nouveau_vma *vma; if (nvbo->bo.mem.mem_type == TTM_PL_TT) rep->domain = NOUVEAU_GEM_DOMAIN_GART; else rep->domain = NOUVEAU_GEM_DOMAIN_VRAM; - rep->size = nvbo->bo.mem.num_pages << PAGE_SHIFT; rep->offset = nvbo->bo.offset; + if (fpriv->vm) { + vma = nouveau_bo_vma_find(nvbo, fpriv->vm); + if (!vma) + return -EINVAL; + + rep->offset = vma->offset; + } + + rep->size = nvbo->bo.mem.num_pages << PAGE_SHIFT; rep->map_handle = nvbo->bo.addr_space_offset; rep->tile_mode = nvbo->tile_mode; rep->tile_flags = nvbo->tile_flags; @@ -127,7 +200,6 @@ nouveau_gem_ioctl_new(struct drm_device *dev, void *data, struct drm_nouveau_private *dev_priv = dev->dev_private; struct drm_nouveau_gem_new *req = data; struct nouveau_bo *nvbo = NULL; - struct nouveau_channel *chan = NULL; int ret = 0; if (unlikely(dev_priv->ttm.bdev.dev_mapping == NULL)) @@ -138,28 +210,21 @@ nouveau_gem_ioctl_new(struct drm_device *dev, void *data, return -EINVAL; } - if (req->channel_hint) { - chan = nouveau_channel_get(dev, file_priv, req->channel_hint); - if (IS_ERR(chan)) - return PTR_ERR(chan); - } - - ret = nouveau_gem_new(dev, chan, req->info.size, req->align, + ret = nouveau_gem_new(dev, req->info.size, req->align, req->info.domain, req->info.tile_mode, req->info.tile_flags, &nvbo); - if (chan) - nouveau_channel_put(&chan); if (ret) return ret; - ret = nouveau_gem_info(nvbo->gem, &req->info); - if (ret) - goto out; - ret = drm_gem_handle_create(file_priv, nvbo->gem, &req->info.handle); + if (ret == 0) { + ret = nouveau_gem_info(file_priv, nvbo->gem, &req->info); + if (ret) + drm_gem_handle_delete(file_priv, req->info.handle); + } + /* drop reference from allocate - handle holds it now */ drm_gem_object_unreference_unlocked(nvbo->gem); -out: return ret; } @@ -318,6 +383,7 @@ static int validate_list(struct nouveau_channel *chan, struct list_head *list, struct drm_nouveau_gem_pushbuf_bo *pbbo, uint64_t user_pbbo_ptr) { + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; struct drm_nouveau_gem_pushbuf_bo __user *upbbo = (void __force __user *)(uintptr_t)user_pbbo_ptr; struct drm_device *dev = chan->dev; @@ -356,24 +422,26 @@ validate_list(struct nouveau_channel *chan, struct list_head *list, return ret; } - if (nvbo->bo.offset == b->presumed.offset && - ((nvbo->bo.mem.mem_type == TTM_PL_VRAM && - b->presumed.domain & NOUVEAU_GEM_DOMAIN_VRAM) || - (nvbo->bo.mem.mem_type == TTM_PL_TT && - b->presumed.domain & NOUVEAU_GEM_DOMAIN_GART))) - continue; + if (dev_priv->card_type < NV_50) { + if (nvbo->bo.offset == b->presumed.offset && + ((nvbo->bo.mem.mem_type == TTM_PL_VRAM && + b->presumed.domain & NOUVEAU_GEM_DOMAIN_VRAM) || + (nvbo->bo.mem.mem_type == TTM_PL_TT && + b->presumed.domain & NOUVEAU_GEM_DOMAIN_GART))) + continue; - if (nvbo->bo.mem.mem_type == TTM_PL_TT) - b->presumed.domain = NOUVEAU_GEM_DOMAIN_GART; - else - b->presumed.domain = NOUVEAU_GEM_DOMAIN_VRAM; - b->presumed.offset = nvbo->bo.offset; - b->presumed.valid = 0; - relocs++; - - if (DRM_COPY_TO_USER(&upbbo[nvbo->pbbo_index].presumed, - &b->presumed, sizeof(b->presumed))) - return -EFAULT; + if (nvbo->bo.mem.mem_type == TTM_PL_TT) + b->presumed.domain = NOUVEAU_GEM_DOMAIN_GART; + else + b->presumed.domain = NOUVEAU_GEM_DOMAIN_VRAM; + b->presumed.offset = nvbo->bo.offset; + b->presumed.valid = 0; + relocs++; + + if (DRM_COPY_TO_USER(&upbbo[nvbo->pbbo_index].presumed, + &b->presumed, sizeof(b->presumed))) + return -EFAULT; + } } return relocs; @@ -548,7 +616,7 @@ nouveau_gem_ioctl_pushbuf(struct drm_device *dev, void *data, struct nouveau_fence *fence = NULL; int i, j, ret = 0, do_reloc = 0; - chan = nouveau_channel_get(dev, file_priv, req->channel); + chan = nouveau_channel_get(file_priv, req->channel); if (IS_ERR(chan)) return PTR_ERR(chan); @@ -782,7 +850,7 @@ nouveau_gem_ioctl_info(struct drm_device *dev, void *data, if (!gem) return -ENOENT; - ret = nouveau_gem_info(gem, req); + ret = nouveau_gem_info(file_priv, gem, req); drm_gem_object_unreference_unlocked(gem); return ret; } diff --git a/drivers/gpu/drm/nouveau/nouveau_grctx.h b/drivers/gpu/drm/nouveau/nouveau_grctx.h index 4a8ad1307fa..86c2e374e93 100644 --- a/drivers/gpu/drm/nouveau/nouveau_grctx.h +++ b/drivers/gpu/drm/nouveau/nouveau_grctx.h @@ -87,10 +87,10 @@ _cp_bra(struct nouveau_grctx *ctx, u32 mod, int flag, int state, int name) cp_out(ctx, CP_BRA | (mod << 18) | ip | flag | (state ? 0 : CP_BRA_IF_CLEAR)); } -#define cp_bra(c,f,s,n) _cp_bra((c), 0, CP_FLAG_##f, CP_FLAG_##f##_##s, n) +#define cp_bra(c, f, s, n) _cp_bra((c), 0, CP_FLAG_##f, CP_FLAG_##f##_##s, n) #ifdef CP_BRA_MOD -#define cp_cal(c,f,s,n) _cp_bra((c), 1, CP_FLAG_##f, CP_FLAG_##f##_##s, n) -#define cp_ret(c,f,s) _cp_bra((c), 2, CP_FLAG_##f, CP_FLAG_##f##_##s, 0) +#define cp_cal(c, f, s, n) _cp_bra((c), 1, CP_FLAG_##f, CP_FLAG_##f##_##s, n) +#define cp_ret(c, f, s) _cp_bra((c), 2, CP_FLAG_##f, CP_FLAG_##f##_##s, 0) #endif static inline void @@ -98,14 +98,14 @@ _cp_wait(struct nouveau_grctx *ctx, int flag, int state) { cp_out(ctx, CP_WAIT | flag | (state ? CP_WAIT_SET : 0)); } -#define cp_wait(c,f,s) _cp_wait((c), CP_FLAG_##f, CP_FLAG_##f##_##s) +#define cp_wait(c, f, s) _cp_wait((c), CP_FLAG_##f, CP_FLAG_##f##_##s) static inline void _cp_set(struct nouveau_grctx *ctx, int flag, int state) { cp_out(ctx, CP_SET | flag | (state ? CP_SET_1 : 0)); } -#define cp_set(c,f,s) _cp_set((c), CP_FLAG_##f, CP_FLAG_##f##_##s) +#define cp_set(c, f, s) _cp_set((c), CP_FLAG_##f, CP_FLAG_##f##_##s) static inline void cp_pos(struct nouveau_grctx *ctx, int offset) diff --git a/drivers/gpu/drm/nouveau/nouveau_hw.c b/drivers/gpu/drm/nouveau/nouveau_hw.c index 053edf9d2f6..ba896e54b79 100644 --- a/drivers/gpu/drm/nouveau/nouveau_hw.c +++ b/drivers/gpu/drm/nouveau/nouveau_hw.c @@ -900,6 +900,7 @@ nv_save_state_ext(struct drm_device *dev, int head, } /* NV11 and NV20 don't have this, they stop at 0x52. */ if (nv_gf4_disp_arch(dev)) { + rd_cio_state(dev, head, regp, NV_CIO_CRE_42); rd_cio_state(dev, head, regp, NV_CIO_CRE_53); rd_cio_state(dev, head, regp, NV_CIO_CRE_54); @@ -1003,6 +1004,7 @@ nv_load_state_ext(struct drm_device *dev, int head, nouveau_wait_eq(dev, 650000000, NV_PRMCIO_INP0__COLOR, 0x8, 0x0); } + wr_cio_state(dev, head, regp, NV_CIO_CRE_42); wr_cio_state(dev, head, regp, NV_CIO_CRE_53); wr_cio_state(dev, head, regp, NV_CIO_CRE_54); diff --git a/drivers/gpu/drm/nouveau/nouveau_irq.c b/drivers/gpu/drm/nouveau/nouveau_irq.c index 2ba7265bc96..868c7fd7485 100644 --- a/drivers/gpu/drm/nouveau/nouveau_irq.c +++ b/drivers/gpu/drm/nouveau/nouveau_irq.c @@ -79,7 +79,7 @@ nouveau_irq_handler(DRM_IRQ_ARGS) int i; stat = nv_rd32(dev, NV03_PMC_INTR_0); - if (!stat) + if (stat == 0 || stat == ~0) return IRQ_NONE; spin_lock_irqsave(&dev_priv->context_switch_lock, flags); diff --git a/drivers/gpu/drm/nouveau/nouveau_mem.c b/drivers/gpu/drm/nouveau/nouveau_mem.c index c3e953b0899..f9ae2fc3d6f 100644 --- a/drivers/gpu/drm/nouveau/nouveau_mem.c +++ b/drivers/gpu/drm/nouveau/nouveau_mem.c @@ -51,8 +51,7 @@ nv10_mem_update_tile_region(struct drm_device *dev, struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - int i = tile - dev_priv->tile.reg; + int i = tile - dev_priv->tile.reg, j; unsigned long save; nouveau_fence_unref(&tile->fence); @@ -70,7 +69,10 @@ nv10_mem_update_tile_region(struct drm_device *dev, nouveau_wait_for_idle(dev); pfb->set_tile_region(dev, i); - pgraph->set_tile_region(dev, i); + for (j = 0; j < NVOBJ_ENGINE_NR; j++) { + if (dev_priv->eng[j] && dev_priv->eng[j]->set_tile_region) + dev_priv->eng[j]->set_tile_region(dev, i); + } pfifo->cache_pull(dev, true); pfifo->reassign(dev, true); @@ -395,7 +397,7 @@ nouveau_mem_vram_init(struct drm_device *dev) if (pci_dma_supported(dev->pdev, DMA_BIT_MASK(40))) dma_bits = 40; } else - if (drm_pci_device_is_pcie(dev) && + if (0 && pci_is_pcie(dev->pdev) && dev_priv->chipset > 0x40 && dev_priv->chipset != 0x45) { if (pci_dma_supported(dev->pdev, DMA_BIT_MASK(39))) @@ -421,38 +423,6 @@ nouveau_mem_vram_init(struct drm_device *dev) return ret; } - /* reserve space at end of VRAM for PRAMIN */ - if (dev_priv->card_type >= NV_50) { - dev_priv->ramin_rsvd_vram = 1 * 1024 * 1024; - } else - if (dev_priv->card_type >= NV_40) { - u32 vs = hweight8((nv_rd32(dev, 0x001540) & 0x0000ff00) >> 8); - u32 rsvd; - - /* estimate grctx size, the magics come from nv40_grctx.c */ - if (dev_priv->chipset == 0x40) rsvd = 0x6aa0 * vs; - else if (dev_priv->chipset < 0x43) rsvd = 0x4f00 * vs; - else if (nv44_graph_class(dev)) rsvd = 0x4980 * vs; - else rsvd = 0x4a40 * vs; - rsvd += 16 * 1024; - rsvd *= dev_priv->engine.fifo.channels; - - /* pciegart table */ - if (drm_pci_device_is_pcie(dev)) - rsvd += 512 * 1024; - - /* object storage */ - rsvd += 512 * 1024; - - dev_priv->ramin_rsvd_vram = round_up(rsvd, 4096); - } else { - dev_priv->ramin_rsvd_vram = 512 * 1024; - } - - ret = dev_priv->engine.vram.init(dev); - if (ret) - return ret; - NV_INFO(dev, "Detected %dMiB VRAM\n", (int)(dev_priv->vram_size >> 20)); if (dev_priv->vram_sys_base) { NV_INFO(dev, "Stolen system memory at: 0x%010llx\n", @@ -477,7 +447,7 @@ nouveau_mem_vram_init(struct drm_device *dev) } if (dev_priv->card_type < NV_50) { - ret = nouveau_bo_new(dev, NULL, 256*1024, 0, TTM_PL_FLAG_VRAM, + ret = nouveau_bo_new(dev, 256*1024, 0, TTM_PL_FLAG_VRAM, 0, 0, &dev_priv->vga_ram); if (ret == 0) ret = nouveau_bo_pin(dev_priv->vga_ram, @@ -595,10 +565,10 @@ nouveau_mem_timing_init(struct drm_device *dev) if (!memtimings->timing) return; - /* Get "some number" from the timing reg for NV_40 + /* Get "some number" from the timing reg for NV_40 and NV_50 * Used in calculations later */ - if(dev_priv->card_type == NV_40) { - magic_number = (nv_rd32(dev,0x100228) & 0x0f000000) >> 24; + if (dev_priv->card_type >= NV_40 && dev_priv->chipset < 0x98) { + magic_number = (nv_rd32(dev, 0x100228) & 0x0f000000) >> 24; } entry = mem + mem[1]; @@ -641,51 +611,68 @@ nouveau_mem_timing_init(struct drm_device *dev) /* XXX: I don't trust the -1's and +1's... they must come * from somewhere! */ timing->reg_100224 = (tUNK_0 + tUNK_19 + 1 + magic_number) << 24 | - tUNK_18 << 16 | + max(tUNK_18, (u8) 1) << 16 | (tUNK_1 + tUNK_19 + 1 + magic_number) << 8; - if(dev_priv->chipset == 0xa8) { + if (dev_priv->chipset == 0xa8) { timing->reg_100224 |= (tUNK_2 - 1); } else { timing->reg_100224 |= (tUNK_2 + 2 - magic_number); } timing->reg_100228 = (tUNK_12 << 16 | tUNK_11 << 8 | tUNK_10); - if(dev_priv->chipset >= 0xa3 && dev_priv->chipset < 0xaa) { + if (dev_priv->chipset >= 0xa3 && dev_priv->chipset < 0xaa) timing->reg_100228 |= (tUNK_19 - 1) << 24; - } + else + timing->reg_100228 |= magic_number << 24; - if(dev_priv->card_type == NV_40) { + if (dev_priv->card_type == NV_40) { /* NV40: don't know what the rest of the regs are.. * And don't need to know either */ - timing->reg_100228 |= 0x20200000 | magic_number << 24; - } else if(dev_priv->card_type >= NV_50) { - /* XXX: reg_10022c */ - timing->reg_10022c = tUNK_2 - 1; + timing->reg_100228 |= 0x20200000; + } else if (dev_priv->card_type >= NV_50) { + if (dev_priv->chipset < 0x98 || + (dev_priv->chipset == 0x98 && + dev_priv->stepping <= 0xa1)) { + timing->reg_10022c = (0x14 + tUNK_2) << 24 | + 0x16 << 16 | + (tUNK_2 - 1) << 8 | + (tUNK_2 - 1); + } else { + /* XXX: reg_10022c for recentish cards */ + timing->reg_10022c = tUNK_2 - 1; + } timing->reg_100230 = (tUNK_20 << 24 | tUNK_21 << 16 | tUNK_13 << 8 | tUNK_13); timing->reg_100234 = (tRAS << 24 | tRC); - timing->reg_100234 += max(tUNK_10,tUNK_11) << 16; + timing->reg_100234 += max(tUNK_10, tUNK_11) << 16; - if(dev_priv->chipset < 0xa3) { + if (dev_priv->chipset < 0x98 || + (dev_priv->chipset == 0x98 && + dev_priv->stepping <= 0xa1)) { timing->reg_100234 |= (tUNK_2 + 2) << 8; } else { /* XXX: +6? */ timing->reg_100234 |= (tUNK_19 + 6) << 8; } - /* XXX; reg_100238, reg_10023c - * reg_100238: 0x00?????? - * reg_10023c: 0x!!??0202 for NV50+ cards (empirical evidence) */ + /* XXX; reg_100238 + * reg_100238: 0x00?????? */ timing->reg_10023c = 0x202; - if(dev_priv->chipset < 0xa3) { + if (dev_priv->chipset < 0x98 || + (dev_priv->chipset == 0x98 && + dev_priv->stepping <= 0xa1)) { timing->reg_10023c |= 0x4000000 | (tUNK_2 - 1) << 16; } else { - /* currently unknown + /* XXX: reg_10023c + * currently unknown * 10023c seen as 06xxxxxx, 0bxxxxxx or 0fxxxxxx */ } + + /* XXX: reg_100240? */ } + timing->id = i; NV_DEBUG(dev, "Entry %d: 220: %08x %08x %08x %08x\n", i, timing->reg_100220, timing->reg_100224, @@ -693,10 +680,11 @@ nouveau_mem_timing_init(struct drm_device *dev) NV_DEBUG(dev, " 230: %08x %08x %08x %08x\n", timing->reg_100230, timing->reg_100234, timing->reg_100238, timing->reg_10023c); + NV_DEBUG(dev, " 240: %08x\n", timing->reg_100240); } memtimings->nr_timing = entries; - memtimings->supported = true; + memtimings->supported = (dev_priv->chipset <= 0x98); } void @@ -709,37 +697,31 @@ nouveau_mem_timing_fini(struct drm_device *dev) } static int -nouveau_vram_manager_init(struct ttm_mem_type_manager *man, unsigned long p_size) +nouveau_vram_manager_init(struct ttm_mem_type_manager *man, unsigned long psize) { - struct drm_nouveau_private *dev_priv = nouveau_bdev(man->bdev); - struct nouveau_mm *mm; - u64 size, block, rsvd; - int ret; - - rsvd = (256 * 1024); /* vga memory */ - size = (p_size << PAGE_SHIFT) - rsvd; - block = dev_priv->vram_rblock_size; - - ret = nouveau_mm_init(&mm, rsvd >> 12, size >> 12, block >> 12); - if (ret) - return ret; - - man->priv = mm; + /* nothing to do */ return 0; } static int nouveau_vram_manager_fini(struct ttm_mem_type_manager *man) { - struct nouveau_mm *mm = man->priv; - int ret; + /* nothing to do */ + return 0; +} - ret = nouveau_mm_fini(&mm); - if (ret) - return ret; +static inline void +nouveau_mem_node_cleanup(struct nouveau_mem *node) +{ + if (node->vma[0].node) { + nouveau_vm_unmap(&node->vma[0]); + nouveau_vm_put(&node->vma[0]); + } - man->priv = NULL; - return 0; + if (node->vma[1].node) { + nouveau_vm_unmap(&node->vma[1]); + nouveau_vm_put(&node->vma[1]); + } } static void @@ -748,14 +730,9 @@ nouveau_vram_manager_del(struct ttm_mem_type_manager *man, { struct drm_nouveau_private *dev_priv = nouveau_bdev(man->bdev); struct nouveau_vram_engine *vram = &dev_priv->engine.vram; - struct nouveau_mem *node = mem->mm_node; struct drm_device *dev = dev_priv->dev; - if (node->tmp_vma.node) { - nouveau_vm_unmap(&node->tmp_vma); - nouveau_vm_put(&node->tmp_vma); - } - + nouveau_mem_node_cleanup(mem->mm_node); vram->put(dev, (struct nouveau_mem **)&mem->mm_node); } @@ -774,7 +751,7 @@ nouveau_vram_manager_new(struct ttm_mem_type_manager *man, int ret; if (nvbo->tile_flags & NOUVEAU_GEM_TILE_NONCONTIG) - size_nc = 1 << nvbo->vma.node->type; + size_nc = 1 << nvbo->page_shift; ret = vram->get(dev, mem->num_pages << PAGE_SHIFT, mem->page_alignment << PAGE_SHIFT, size_nc, @@ -784,9 +761,7 @@ nouveau_vram_manager_new(struct ttm_mem_type_manager *man, return (ret == -ENOSPC) ? 0 : ret; } - node->page_shift = 12; - if (nvbo->vma.node) - node->page_shift = nvbo->vma.node->type; + node->page_shift = nvbo->page_shift; mem->mm_node = node; mem->start = node->offset >> PAGE_SHIFT; @@ -842,12 +817,8 @@ static void nouveau_gart_manager_del(struct ttm_mem_type_manager *man, struct ttm_mem_reg *mem) { - struct nouveau_mem *node = mem->mm_node; - - if (node->tmp_vma.node) { - nouveau_vm_unmap(&node->tmp_vma); - nouveau_vm_put(&node->tmp_vma); - } + nouveau_mem_node_cleanup(mem->mm_node); + kfree(mem->mm_node); mem->mm_node = NULL; } @@ -858,11 +829,7 @@ nouveau_gart_manager_new(struct ttm_mem_type_manager *man, struct ttm_mem_reg *mem) { struct drm_nouveau_private *dev_priv = nouveau_bdev(bo->bdev); - struct nouveau_bo *nvbo = nouveau_bo(bo); - struct nouveau_vma *vma = &nvbo->vma; - struct nouveau_vm *vm = vma->vm; struct nouveau_mem *node; - int ret; if (unlikely((mem->num_pages << PAGE_SHIFT) >= dev_priv->gart_info.aper_size)) @@ -871,24 +838,8 @@ nouveau_gart_manager_new(struct ttm_mem_type_manager *man, node = kzalloc(sizeof(*node), GFP_KERNEL); if (!node) return -ENOMEM; + node->page_shift = 12; - /* This node must be for evicting large-paged VRAM - * to system memory. Due to a nv50 limitation of - * not being able to mix large/small pages within - * the same PDE, we need to create a temporary - * small-paged VMA for the eviction. - */ - if (vma->node->type != vm->spg_shift) { - ret = nouveau_vm_get(vm, (u64)vma->node->length << 12, - vm->spg_shift, NV_MEM_ACCESS_RW, - &node->tmp_vma); - if (ret) { - kfree(node); - return ret; - } - } - - node->page_shift = nvbo->vma.node->type; mem->mm_node = node; mem->start = 0; return 0; diff --git a/drivers/gpu/drm/nouveau/nouveau_mm.c b/drivers/gpu/drm/nouveau/nouveau_mm.c index 7609756b6fa..1640dec3b82 100644 --- a/drivers/gpu/drm/nouveau/nouveau_mm.c +++ b/drivers/gpu/drm/nouveau/nouveau_mm.c @@ -158,11 +158,18 @@ int nouveau_mm_fini(struct nouveau_mm **prmm) { struct nouveau_mm *rmm = *prmm; - struct nouveau_mm_node *heap = + struct nouveau_mm_node *node, *heap = list_first_entry(&rmm->nodes, struct nouveau_mm_node, nl_entry); - if (!list_is_singular(&rmm->nodes)) + if (!list_is_singular(&rmm->nodes)) { + printk(KERN_ERR "nouveau_mm not empty at destroy time!\n"); + list_for_each_entry(node, &rmm->nodes, nl_entry) { + printk(KERN_ERR "0x%02x: 0x%08x 0x%08x\n", + node->type, node->offset, node->length); + } + WARN_ON(1); return -EBUSY; + } kfree(heap); kfree(rmm); diff --git a/drivers/gpu/drm/nouveau/nouveau_mm.h b/drivers/gpu/drm/nouveau/nouveau_mm.h index 1f7483aae9a..b9c016d2155 100644 --- a/drivers/gpu/drm/nouveau/nouveau_mm.h +++ b/drivers/gpu/drm/nouveau/nouveau_mm.h @@ -52,6 +52,7 @@ int nouveau_mm_get(struct nouveau_mm *, int type, u32 size, u32 size_nc, void nouveau_mm_put(struct nouveau_mm *, struct nouveau_mm_node *); int nv50_vram_init(struct drm_device *); +void nv50_vram_fini(struct drm_device *); int nv50_vram_new(struct drm_device *, u64 size, u32 align, u32 size_nc, u32 memtype, struct nouveau_mem **); void nv50_vram_del(struct drm_device *, struct nouveau_mem **); diff --git a/drivers/gpu/drm/nouveau/nouveau_notifier.c b/drivers/gpu/drm/nouveau/nouveau_notifier.c index 5b39718ae1f..6abdbe6530a 100644 --- a/drivers/gpu/drm/nouveau/nouveau_notifier.c +++ b/drivers/gpu/drm/nouveau/nouveau_notifier.c @@ -34,6 +34,7 @@ int nouveau_notifier_init_channel(struct nouveau_channel *chan) { struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_bo *ntfy = NULL; uint32_t flags, ttmpl; int ret; @@ -46,7 +47,7 @@ nouveau_notifier_init_channel(struct nouveau_channel *chan) ttmpl = TTM_PL_FLAG_TT; } - ret = nouveau_gem_new(dev, NULL, PAGE_SIZE, 0, flags, 0, 0, &ntfy); + ret = nouveau_gem_new(dev, PAGE_SIZE, 0, flags, 0, 0, &ntfy); if (ret) return ret; @@ -58,14 +59,22 @@ nouveau_notifier_init_channel(struct nouveau_channel *chan) if (ret) goto out_err; + if (dev_priv->card_type >= NV_50) { + ret = nouveau_bo_vma_add(ntfy, chan->vm, &chan->notifier_vma); + if (ret) + goto out_err; + } + ret = drm_mm_init(&chan->notifier_heap, 0, ntfy->bo.mem.size); if (ret) goto out_err; chan->notifier_bo = ntfy; out_err: - if (ret) + if (ret) { + nouveau_bo_vma_del(ntfy, &chan->notifier_vma); drm_gem_object_unreference_unlocked(ntfy->gem); + } return ret; } @@ -78,6 +87,7 @@ nouveau_notifier_takedown_channel(struct nouveau_channel *chan) if (!chan->notifier_bo) return; + nouveau_bo_vma_del(chan->notifier_bo, &chan->notifier_vma); nouveau_bo_unmap(chan->notifier_bo); mutex_lock(&dev->struct_mutex); nouveau_bo_unpin(chan->notifier_bo); @@ -122,10 +132,10 @@ nouveau_notifier_alloc(struct nouveau_channel *chan, uint32_t handle, target = NV_MEM_TARGET_VRAM; else target = NV_MEM_TARGET_GART; - offset = chan->notifier_bo->bo.mem.start << PAGE_SHIFT; + offset = chan->notifier_bo->bo.offset; } else { target = NV_MEM_TARGET_VM; - offset = chan->notifier_bo->vma.offset; + offset = chan->notifier_vma.offset; } offset += mem->start; @@ -183,7 +193,7 @@ nouveau_ioctl_notifier_alloc(struct drm_device *dev, void *data, if (unlikely(dev_priv->card_type >= NV_C0)) return -EINVAL; - chan = nouveau_channel_get(dev, file_priv, na->channel); + chan = nouveau_channel_get(file_priv, na->channel); if (IS_ERR(chan)) return PTR_ERR(chan); diff --git a/drivers/gpu/drm/nouveau/nouveau_object.c b/drivers/gpu/drm/nouveau/nouveau_object.c index 67a16e01ffa..159b7c437d3 100644 --- a/drivers/gpu/drm/nouveau/nouveau_object.c +++ b/drivers/gpu/drm/nouveau/nouveau_object.c @@ -125,7 +125,7 @@ nouveau_gpuobj_mthd_call2(struct drm_device *dev, int chid, int ret = -EINVAL; spin_lock_irqsave(&dev_priv->channels.lock, flags); - if (chid > 0 && chid < dev_priv->engine.fifo.channels) + if (chid >= 0 && chid < dev_priv->engine.fifo.channels) chan = dev_priv->channels.ptr[chid]; if (chan) ret = nouveau_gpuobj_mthd_call(chan, class, mthd, data); @@ -191,7 +191,7 @@ nouveau_gpuobj_new(struct drm_device *dev, struct nouveau_channel *chan, list_add_tail(&gpuobj->list, &dev_priv->gpuobj_list); spin_unlock(&dev_priv->ramin_lock); - if (chan) { + if (!(flags & NVOBJ_FLAG_VM) && chan) { ramin = drm_mm_search_free(&chan->ramin_heap, size, align, 0); if (ramin) ramin = drm_mm_get_block(ramin, size, align); @@ -208,7 +208,7 @@ nouveau_gpuobj_new(struct drm_device *dev, struct nouveau_channel *chan, gpuobj->vinst = ramin->start + chan->ramin->vinst; gpuobj->node = ramin; } else { - ret = instmem->get(gpuobj, size, align); + ret = instmem->get(gpuobj, chan, size, align); if (ret) { nouveau_gpuobj_ref(NULL, &gpuobj); return ret; @@ -361,20 +361,6 @@ nouveau_gpuobj_new_fake(struct drm_device *dev, u32 pinst, u64 vinst, return 0; } - -static uint32_t -nouveau_gpuobj_class_instmem_size(struct drm_device *dev, int class) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - - /*XXX: dodgy hack for now */ - if (dev_priv->card_type >= NV_50) - return 24; - if (dev_priv->card_type >= NV_40) - return 32; - return 16; -} - /* DMA objects are used to reference a piece of memory in the framebuffer, PCI or AGP address space. Each object is 16 bytes big @@ -606,11 +592,11 @@ nouveau_gpuobj_dma_new(struct nouveau_channel *chan, int class, u64 base, set to 0? */ static int -nouveau_gpuobj_sw_new(struct nouveau_channel *chan, int class, - struct nouveau_gpuobj **gpuobj_ret) +nouveau_gpuobj_sw_new(struct nouveau_channel *chan, u32 handle, u16 class) { struct drm_nouveau_private *dev_priv = chan->dev->dev_private; struct nouveau_gpuobj *gpuobj; + int ret; gpuobj = kzalloc(sizeof(*gpuobj), GFP_KERNEL); if (!gpuobj) @@ -624,8 +610,10 @@ nouveau_gpuobj_sw_new(struct nouveau_channel *chan, int class, spin_lock(&dev_priv->ramin_lock); list_add_tail(&gpuobj->list, &dev_priv->gpuobj_list); spin_unlock(&dev_priv->ramin_lock); - *gpuobj_ret = gpuobj; - return 0; + + ret = nouveau_ramht_insert(chan, handle, gpuobj); + nouveau_gpuobj_ref(NULL, &gpuobj); + return ret; } int @@ -634,101 +622,30 @@ nouveau_gpuobj_gr_new(struct nouveau_channel *chan, u32 handle, int class) struct drm_nouveau_private *dev_priv = chan->dev->dev_private; struct drm_device *dev = chan->dev; struct nouveau_gpuobj_class *oc; - struct nouveau_gpuobj *gpuobj; int ret; NV_DEBUG(dev, "ch%d class=0x%04x\n", chan->id, class); list_for_each_entry(oc, &dev_priv->classes, head) { - if (oc->id == class) - goto found; - } + struct nouveau_exec_engine *eng = dev_priv->eng[oc->engine]; - NV_ERROR(dev, "illegal object class: 0x%x\n", class); - return -EINVAL; - -found: - switch (oc->engine) { - case NVOBJ_ENGINE_SW: - if (dev_priv->card_type < NV_C0) { - ret = nouveau_gpuobj_sw_new(chan, class, &gpuobj); - if (ret) - return ret; - goto insert; - } - break; - case NVOBJ_ENGINE_GR: - if ((dev_priv->card_type >= NV_20 && !chan->ramin_grctx) || - (dev_priv->card_type < NV_20 && !chan->pgraph_ctx)) { - struct nouveau_pgraph_engine *pgraph = - &dev_priv->engine.graph; + if (oc->id != class) + continue; - ret = pgraph->create_context(chan); - if (ret) - return ret; - } - break; - case NVOBJ_ENGINE_CRYPT: - if (!chan->crypt_ctx) { - struct nouveau_crypt_engine *pcrypt = - &dev_priv->engine.crypt; + if (oc->engine == NVOBJ_ENGINE_SW) + return nouveau_gpuobj_sw_new(chan, handle, class); - ret = pcrypt->create_context(chan); + if (!chan->engctx[oc->engine]) { + ret = eng->context_new(chan, oc->engine); if (ret) return ret; } - break; - } - - /* we're done if this is fermi */ - if (dev_priv->card_type >= NV_C0) - return 0; - - ret = nouveau_gpuobj_new(dev, chan, - nouveau_gpuobj_class_instmem_size(dev, class), - 16, - NVOBJ_FLAG_ZERO_ALLOC | NVOBJ_FLAG_ZERO_FREE, - &gpuobj); - if (ret) { - NV_ERROR(dev, "error creating gpuobj: %d\n", ret); - return ret; - } - if (dev_priv->card_type >= NV_50) { - nv_wo32(gpuobj, 0, class); - nv_wo32(gpuobj, 20, 0x00010000); - } else { - switch (class) { - case NV_CLASS_NULL: - nv_wo32(gpuobj, 0, 0x00001030); - nv_wo32(gpuobj, 4, 0xFFFFFFFF); - break; - default: - if (dev_priv->card_type >= NV_40) { - nv_wo32(gpuobj, 0, class); -#ifdef __BIG_ENDIAN - nv_wo32(gpuobj, 8, 0x01000000); -#endif - } else { -#ifdef __BIG_ENDIAN - nv_wo32(gpuobj, 0, class | 0x00080000); -#else - nv_wo32(gpuobj, 0, class); -#endif - } - } + return eng->object_new(chan, oc->engine, handle, class); } - dev_priv->engine.instmem.flush(dev); - - gpuobj->engine = oc->engine; - gpuobj->class = oc->id; -insert: - ret = nouveau_ramht_insert(chan, handle, gpuobj); - if (ret) - NV_ERROR(dev, "error adding gpuobj to RAMHT: %d\n", ret); - nouveau_gpuobj_ref(NULL, &gpuobj); - return ret; + NV_ERROR(dev, "illegal object class: 0x%x\n", class); + return -EINVAL; } static int @@ -746,9 +663,6 @@ nouveau_gpuobj_channel_init_pramin(struct nouveau_channel *chan) size = 0x2000; base = 0; - /* PGRAPH context */ - size += dev_priv->engine.graph.grctx_size; - if (dev_priv->card_type == NV_50) { /* Various fixed table thingos */ size += 0x1400; /* mostly unknown stuff */ @@ -776,35 +690,64 @@ nouveau_gpuobj_channel_init_pramin(struct nouveau_channel *chan) return 0; } +static int +nvc0_gpuobj_channel_init(struct nouveau_channel *chan, struct nouveau_vm *vm) +{ + struct drm_device *dev = chan->dev; + struct nouveau_gpuobj *pgd = NULL; + struct nouveau_vm_pgd *vpgd; + int ret, i; + + ret = nouveau_gpuobj_new(dev, NULL, 4096, 0x1000, 0, &chan->ramin); + if (ret) + return ret; + + /* create page directory for this vm if none currently exists, + * will be destroyed automagically when last reference to the + * vm is removed + */ + if (list_empty(&vm->pgd_list)) { + ret = nouveau_gpuobj_new(dev, NULL, 65536, 0x1000, 0, &pgd); + if (ret) + return ret; + } + nouveau_vm_ref(vm, &chan->vm, pgd); + nouveau_gpuobj_ref(NULL, &pgd); + + /* point channel at vm's page directory */ + vpgd = list_first_entry(&vm->pgd_list, struct nouveau_vm_pgd, head); + nv_wo32(chan->ramin, 0x0200, lower_32_bits(vpgd->obj->vinst)); + nv_wo32(chan->ramin, 0x0204, upper_32_bits(vpgd->obj->vinst)); + nv_wo32(chan->ramin, 0x0208, 0xffffffff); + nv_wo32(chan->ramin, 0x020c, 0x000000ff); + + /* map display semaphore buffers into channel's vm */ + for (i = 0; i < 2; i++) { + struct nv50_display_crtc *dispc = &nv50_display(dev)->crtc[i]; + + ret = nouveau_bo_vma_add(dispc->sem.bo, chan->vm, + &chan->dispc_vma[i]); + if (ret) + return ret; + } + + return 0; +} + int nouveau_gpuobj_channel_init(struct nouveau_channel *chan, uint32_t vram_h, uint32_t tt_h) { struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fpriv *fpriv = nouveau_fpriv(chan->file_priv); + struct nouveau_vm *vm = fpriv ? fpriv->vm : dev_priv->chan_vm; struct nouveau_gpuobj *vram = NULL, *tt = NULL; int ret, i; NV_DEBUG(dev, "ch%d vram=0x%08x tt=0x%08x\n", chan->id, vram_h, tt_h); - - if (dev_priv->card_type == NV_C0) { - struct nouveau_vm *vm = dev_priv->chan_vm; - struct nouveau_vm_pgd *vpgd; - - ret = nouveau_gpuobj_new(dev, NULL, 4096, 0x1000, 0, - &chan->ramin); - if (ret) - return ret; - - nouveau_vm_ref(vm, &chan->vm, NULL); - - vpgd = list_first_entry(&vm->pgd_list, struct nouveau_vm_pgd, head); - nv_wo32(chan->ramin, 0x0200, lower_32_bits(vpgd->obj->vinst)); - nv_wo32(chan->ramin, 0x0204, upper_32_bits(vpgd->obj->vinst)); - nv_wo32(chan->ramin, 0x0208, 0xffffffff); - nv_wo32(chan->ramin, 0x020c, 0x000000ff); - return 0; - } + if (dev_priv->card_type == NV_C0) + return nvc0_gpuobj_channel_init(chan, vm); /* Allocate a chunk of memory for per-channel object storage */ ret = nouveau_gpuobj_channel_init_pramin(chan); @@ -817,7 +760,7 @@ nouveau_gpuobj_channel_init(struct nouveau_channel *chan, * - Allocate per-channel page-directory * - Link with shared channel VM */ - if (dev_priv->chan_vm) { + if (vm) { u32 pgd_offs = (dev_priv->chipset == 0x50) ? 0x1400 : 0x0200; u64 vm_vinst = chan->ramin->vinst + pgd_offs; u32 vm_pinst = chan->ramin->pinst; @@ -830,7 +773,7 @@ nouveau_gpuobj_channel_init(struct nouveau_channel *chan, if (ret) return ret; - nouveau_vm_ref(dev_priv->chan_vm, &chan->vm, chan->vm_pd); + nouveau_vm_ref(vm, &chan->vm, chan->vm_pd); } /* RAMHT */ @@ -854,7 +797,7 @@ nouveau_gpuobj_channel_init(struct nouveau_channel *chan, struct nouveau_gpuobj *sem = NULL; struct nv50_display_crtc *dispc = &nv50_display(dev)->crtc[i]; - u64 offset = dispc->sem.bo->bo.mem.start << PAGE_SHIFT; + u64 offset = dispc->sem.bo->bo.offset; ret = nouveau_gpuobj_dma_new(chan, 0x3d, offset, 0xfff, NV_MEM_ACCESS_RW, @@ -927,13 +870,22 @@ void nouveau_gpuobj_channel_takedown(struct nouveau_channel *chan) { struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + int i; NV_DEBUG(dev, "ch%d\n", chan->id); - nouveau_ramht_ref(NULL, &chan->ramht, chan); + if (dev_priv->card_type >= NV_50) { + struct nv50_display *disp = nv50_display(dev); - nouveau_vm_ref(NULL, &chan->vm, chan->vm_pd); - nouveau_gpuobj_ref(NULL, &chan->vm_pd); + for (i = 0; i < 2; i++) { + struct nv50_display_crtc *dispc = &disp->crtc[i]; + nouveau_bo_vma_del(dispc->sem.bo, &chan->dispc_vma[i]); + } + + nouveau_vm_ref(NULL, &chan->vm, chan->vm_pd); + nouveau_gpuobj_ref(NULL, &chan->vm_pd); + } if (drm_mm_initialized(&chan->ramin_heap)) drm_mm_takedown(&chan->ramin_heap); @@ -995,7 +947,7 @@ int nouveau_ioctl_grobj_alloc(struct drm_device *dev, void *data, if (init->handle == ~0) return -EINVAL; - chan = nouveau_channel_get(dev, file_priv, init->channel); + chan = nouveau_channel_get(file_priv, init->channel); if (IS_ERR(chan)) return PTR_ERR(chan); @@ -1022,7 +974,7 @@ int nouveau_ioctl_gpuobj_free(struct drm_device *dev, void *data, struct nouveau_channel *chan; int ret; - chan = nouveau_channel_get(dev, file_priv, objfree->channel); + chan = nouveau_channel_get(file_priv, objfree->channel); if (IS_ERR(chan)) return PTR_ERR(chan); diff --git a/drivers/gpu/drm/nouveau/nouveau_perf.c b/drivers/gpu/drm/nouveau/nouveau_perf.c index 670e3cb697e..ef9dec0e6f8 100644 --- a/drivers/gpu/drm/nouveau/nouveau_perf.c +++ b/drivers/gpu/drm/nouveau/nouveau_perf.c @@ -72,6 +72,68 @@ legacy_perf_init(struct drm_device *dev) pm->nr_perflvl = 1; } +static struct nouveau_pm_memtiming * +nouveau_perf_timing(struct drm_device *dev, struct bit_entry *P, + u16 memclk, u8 *entry, u8 recordlen, u8 entries) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_pm_engine *pm = &dev_priv->engine.pm; + struct nvbios *bios = &dev_priv->vbios; + u8 ramcfg; + int i; + + /* perf v2 has a separate "timing map" table, we have to match + * the target memory clock to a specific entry, *then* use + * ramcfg to select the correct subentry + */ + if (P->version == 2) { + u8 *tmap = ROMPTR(bios, P->data[4]); + if (!tmap) { + NV_DEBUG(dev, "no timing map pointer\n"); + return NULL; + } + + if (tmap[0] != 0x10) { + NV_WARN(dev, "timing map 0x%02x unknown\n", tmap[0]); + return NULL; + } + + entry = tmap + tmap[1]; + recordlen = tmap[2] + (tmap[4] * tmap[3]); + for (i = 0; i < tmap[5]; i++, entry += recordlen) { + if (memclk >= ROM16(entry[0]) && + memclk <= ROM16(entry[2])) + break; + } + + if (i == tmap[5]) { + NV_WARN(dev, "no match in timing map table\n"); + return NULL; + } + + entry += tmap[2]; + recordlen = tmap[3]; + entries = tmap[4]; + } + + ramcfg = (nv_rd32(dev, NV_PEXTDEV_BOOT_0) & 0x0000003c) >> 2; + if (bios->ram_restrict_tbl_ptr) + ramcfg = bios->data[bios->ram_restrict_tbl_ptr + ramcfg]; + + if (ramcfg >= entries) { + NV_WARN(dev, "ramcfg strap out of bounds!\n"); + return NULL; + } + + entry += ramcfg * recordlen; + if (entry[1] >= pm->memtimings.nr_timing) { + NV_WARN(dev, "timingset %d does not exist\n", entry[1]); + return NULL; + } + + return &pm->memtimings.timing[entry[1]]; +} + void nouveau_perf_init(struct drm_device *dev) { @@ -120,10 +182,17 @@ nouveau_perf_init(struct drm_device *dev) entries = perf[2]; } + if (entries > NOUVEAU_PM_MAX_LEVEL) { + NV_DEBUG(dev, "perf table has too many entries - buggy vbios?\n"); + entries = NOUVEAU_PM_MAX_LEVEL; + } + entry = perf + headerlen; for (i = 0; i < entries; i++) { struct nouveau_pm_level *perflvl = &pm->perflvl[pm->nr_perflvl]; + perflvl->timing = NULL; + if (entry[0] == 0xff) { entry += recordlen; continue; @@ -174,9 +243,21 @@ nouveau_perf_init(struct drm_device *dev) #define subent(n) entry[perf[2] + ((n) * perf[3])] perflvl->fanspeed = 0; /*XXX*/ perflvl->voltage = entry[2]; - perflvl->core = (ROM16(subent(0)) & 0xfff) * 1000; - perflvl->shader = (ROM16(subent(1)) & 0xfff) * 1000; - perflvl->memory = (ROM16(subent(2)) & 0xfff) * 1000; + if (dev_priv->card_type == NV_50) { + perflvl->core = ROM16(subent(0)) & 0xfff; + perflvl->shader = ROM16(subent(1)) & 0xfff; + perflvl->memory = ROM16(subent(2)) & 0xfff; + } else { + perflvl->shader = ROM16(subent(3)) & 0xfff; + perflvl->core = perflvl->shader / 2; + perflvl->unk0a = ROM16(subent(4)) & 0xfff; + perflvl->memory = ROM16(subent(5)) & 0xfff; + } + + perflvl->core *= 1000; + perflvl->shader *= 1000; + perflvl->memory *= 1000; + perflvl->unk0a *= 1000; break; } @@ -190,6 +271,16 @@ nouveau_perf_init(struct drm_device *dev) } } + /* get the corresponding memory timings */ + if (version > 0x15) { + /* last 3 args are for < 0x40, ignored for >= 0x40 */ + perflvl->timing = + nouveau_perf_timing(dev, &P, + perflvl->memory / 1000, + entry + perf[3], + perf[5], perf[4]); + } + snprintf(perflvl->name, sizeof(perflvl->name), "performance_level_%d", i); perflvl->id = i; diff --git a/drivers/gpu/drm/nouveau/nouveau_pm.c b/drivers/gpu/drm/nouveau/nouveau_pm.c index 4399e2f34db..da8d994d5e8 100644 --- a/drivers/gpu/drm/nouveau/nouveau_pm.c +++ b/drivers/gpu/drm/nouveau/nouveau_pm.c @@ -156,7 +156,7 @@ nouveau_pm_perflvl_get(struct drm_device *dev, struct nouveau_pm_level *perflvl) static void nouveau_pm_perflvl_info(struct nouveau_pm_level *perflvl, char *ptr, int len) { - char c[16], s[16], v[16], f[16]; + char c[16], s[16], v[16], f[16], t[16]; c[0] = '\0'; if (perflvl->core) @@ -174,8 +174,12 @@ nouveau_pm_perflvl_info(struct nouveau_pm_level *perflvl, char *ptr, int len) if (perflvl->fanspeed) snprintf(f, sizeof(f), " fanspeed %d%%", perflvl->fanspeed); - snprintf(ptr, len, "memory %dMHz%s%s%s%s\n", perflvl->memory / 1000, - c, s, v, f); + t[0] = '\0'; + if (perflvl->timing) + snprintf(t, sizeof(t), " timing %d", perflvl->timing->id); + + snprintf(ptr, len, "memory %dMHz%s%s%s%s%s\n", perflvl->memory / 1000, + c, s, v, f, t); } static ssize_t @@ -449,7 +453,7 @@ nouveau_hwmon_fini(struct drm_device *dev) #endif } -#ifdef CONFIG_ACPI +#if defined(CONFIG_ACPI) && defined(CONFIG_POWER_SUPPLY) static int nouveau_pm_acpi_event(struct notifier_block *nb, unsigned long val, void *data) { @@ -476,10 +480,10 @@ nouveau_pm_init(struct drm_device *dev) char info[256]; int ret, i; + nouveau_mem_timing_init(dev); nouveau_volt_init(dev); nouveau_perf_init(dev); nouveau_temp_init(dev); - nouveau_mem_timing_init(dev); NV_INFO(dev, "%d available performance level(s)\n", pm->nr_perflvl); for (i = 0; i < pm->nr_perflvl; i++) { @@ -490,6 +494,7 @@ nouveau_pm_init(struct drm_device *dev) /* determine current ("boot") performance level */ ret = nouveau_pm_perflvl_get(dev, &pm->boot); if (ret == 0) { + strncpy(pm->boot.name, "boot", 4); pm->cur = &pm->boot; nouveau_pm_perflvl_info(&pm->boot, info, sizeof(info)); @@ -507,7 +512,7 @@ nouveau_pm_init(struct drm_device *dev) nouveau_sysfs_init(dev); nouveau_hwmon_init(dev); -#ifdef CONFIG_ACPI +#if defined(CONFIG_ACPI) && defined(CONFIG_POWER_SUPPLY) pm->acpi_nb.notifier_call = nouveau_pm_acpi_event; register_acpi_notifier(&pm->acpi_nb); #endif @@ -524,12 +529,12 @@ nouveau_pm_fini(struct drm_device *dev) if (pm->cur != &pm->boot) nouveau_pm_perflvl_set(dev, &pm->boot); - nouveau_mem_timing_fini(dev); nouveau_temp_fini(dev); nouveau_perf_fini(dev); nouveau_volt_fini(dev); + nouveau_mem_timing_fini(dev); -#ifdef CONFIG_ACPI +#if defined(CONFIG_ACPI) && defined(CONFIG_POWER_SUPPLY) unregister_acpi_notifier(&pm->acpi_nb); #endif nouveau_hwmon_fini(dev); diff --git a/drivers/gpu/drm/nouveau/nouveau_reg.h b/drivers/gpu/drm/nouveau/nouveau_reg.h index 04e8fb79526..f18cdfc3400 100644 --- a/drivers/gpu/drm/nouveau/nouveau_reg.h +++ b/drivers/gpu/drm/nouveau/nouveau_reg.h @@ -639,9 +639,9 @@ # define NV50_PCONNECTOR_I2C_PORT_4 0x0000e240 # define NV50_PCONNECTOR_I2C_PORT_5 0x0000e258 -#define NV50_AUXCH_DATA_OUT(i,n) ((n) * 4 + (i) * 0x50 + 0x0000e4c0) +#define NV50_AUXCH_DATA_OUT(i, n) ((n) * 4 + (i) * 0x50 + 0x0000e4c0) #define NV50_AUXCH_DATA_OUT__SIZE 4 -#define NV50_AUXCH_DATA_IN(i,n) ((n) * 4 + (i) * 0x50 + 0x0000e4d0) +#define NV50_AUXCH_DATA_IN(i, n) ((n) * 4 + (i) * 0x50 + 0x0000e4d0) #define NV50_AUXCH_DATA_IN__SIZE 4 #define NV50_AUXCH_ADDR(i) ((i) * 0x50 + 0x0000e4e0) #define NV50_AUXCH_CTRL(i) ((i) * 0x50 + 0x0000e4e4) @@ -829,7 +829,7 @@ #define NV50_PDISPLAY_SOR_BACKLIGHT 0x0061c084 #define NV50_PDISPLAY_SOR_BACKLIGHT_ENABLE 0x80000000 #define NV50_PDISPLAY_SOR_BACKLIGHT_LEVEL 0x00000fff -#define NV50_SOR_DP_CTRL(i,l) (0x0061c10c + (i) * 0x800 + (l) * 0x80) +#define NV50_SOR_DP_CTRL(i, l) (0x0061c10c + (i) * 0x800 + (l) * 0x80) #define NV50_SOR_DP_CTRL_ENABLED 0x00000001 #define NV50_SOR_DP_CTRL_ENHANCED_FRAME_ENABLED 0x00004000 #define NV50_SOR_DP_CTRL_LANE_MASK 0x001f0000 @@ -841,10 +841,10 @@ #define NV50_SOR_DP_CTRL_TRAINING_PATTERN_DISABLED 0x00000000 #define NV50_SOR_DP_CTRL_TRAINING_PATTERN_1 0x01000000 #define NV50_SOR_DP_CTRL_TRAINING_PATTERN_2 0x02000000 -#define NV50_SOR_DP_UNK118(i,l) (0x0061c118 + (i) * 0x800 + (l) * 0x80) -#define NV50_SOR_DP_UNK120(i,l) (0x0061c120 + (i) * 0x800 + (l) * 0x80) -#define NV50_SOR_DP_UNK128(i,l) (0x0061c128 + (i) * 0x800 + (l) * 0x80) -#define NV50_SOR_DP_UNK130(i,l) (0x0061c130 + (i) * 0x800 + (l) * 0x80) +#define NV50_SOR_DP_UNK118(i, l) (0x0061c118 + (i) * 0x800 + (l) * 0x80) +#define NV50_SOR_DP_UNK120(i, l) (0x0061c120 + (i) * 0x800 + (l) * 0x80) +#define NV50_SOR_DP_UNK128(i, l) (0x0061c128 + (i) * 0x800 + (l) * 0x80) +#define NV50_SOR_DP_UNK130(i, l) (0x0061c130 + (i) * 0x800 + (l) * 0x80) #define NV50_PDISPLAY_USER(i) ((i) * 0x1000 + 0x00640000) #define NV50_PDISPLAY_USER_PUT(i) ((i) * 0x1000 + 0x00640000) diff --git a/drivers/gpu/drm/nouveau/nouveau_sgdma.c b/drivers/gpu/drm/nouveau/nouveau_sgdma.c index 4bce801bc58..c444cadbf84 100644 --- a/drivers/gpu/drm/nouveau/nouveau_sgdma.c +++ b/drivers/gpu/drm/nouveau/nouveau_sgdma.c @@ -42,7 +42,8 @@ nouveau_sgdma_populate(struct ttm_backend *be, unsigned long num_pages, nvbe->nr_pages = 0; while (num_pages--) { - if (dma_addrs[nvbe->nr_pages] != DMA_ERROR_CODE) { + /* this code path isn't called and is incorrect anyways */ + if (0) { /*dma_addrs[nvbe->nr_pages] != DMA_ERROR_CODE)*/ nvbe->pages[nvbe->nr_pages] = dma_addrs[nvbe->nr_pages]; nvbe->ttm_alloced[nvbe->nr_pages] = true; @@ -428,7 +429,7 @@ nouveau_sgdma_init(struct drm_device *dev) u32 aper_size, align; int ret; - if (dev_priv->card_type >= NV_40 && drm_pci_device_is_pcie(dev)) + if (dev_priv->card_type >= NV_40 && pci_is_pcie(dev->pdev)) aper_size = 512 * 1024 * 1024; else aper_size = 64 * 1024 * 1024; @@ -457,7 +458,7 @@ nouveau_sgdma_init(struct drm_device *dev) dev_priv->gart_info.type = NOUVEAU_GART_HW; dev_priv->gart_info.func = &nv50_sgdma_backend; } else - if (drm_pci_device_is_pcie(dev) && + if (0 && pci_is_pcie(dev->pdev) && dev_priv->chipset > 0x40 && dev_priv->chipset != 0x45) { if (nv44_graph_class(dev)) { dev_priv->gart_info.func = &nv44_sgdma_backend; diff --git a/drivers/gpu/drm/nouveau/nouveau_state.c b/drivers/gpu/drm/nouveau/nouveau_state.c index 915fbce8959..10656e430b4 100644 --- a/drivers/gpu/drm/nouveau/nouveau_state.c +++ b/drivers/gpu/drm/nouveau/nouveau_state.c @@ -65,14 +65,6 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->timer.takedown = nv04_timer_takedown; engine->fb.init = nv04_fb_init; engine->fb.takedown = nv04_fb_takedown; - engine->graph.init = nv04_graph_init; - engine->graph.takedown = nv04_graph_takedown; - engine->graph.fifo_access = nv04_graph_fifo_access; - engine->graph.channel = nv04_graph_channel; - engine->graph.create_context = nv04_graph_create_context; - engine->graph.destroy_context = nv04_graph_destroy_context; - engine->graph.load_context = nv04_graph_load_context; - engine->graph.unload_context = nv04_graph_unload_context; engine->fifo.channels = 16; engine->fifo.init = nv04_fifo_init; engine->fifo.takedown = nv04_fifo_fini; @@ -98,9 +90,8 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->pm.clock_get = nv04_pm_clock_get; engine->pm.clock_pre = nv04_pm_clock_pre; engine->pm.clock_set = nv04_pm_clock_set; - engine->crypt.init = nouveau_stub_init; - engine->crypt.takedown = nouveau_stub_takedown; engine->vram.init = nouveau_mem_detect; + engine->vram.takedown = nouveau_stub_takedown; engine->vram.flags_valid = nouveau_mem_flags_valid; break; case 0x10: @@ -123,15 +114,6 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->fb.init_tile_region = nv10_fb_init_tile_region; engine->fb.set_tile_region = nv10_fb_set_tile_region; engine->fb.free_tile_region = nv10_fb_free_tile_region; - engine->graph.init = nv10_graph_init; - engine->graph.takedown = nv10_graph_takedown; - engine->graph.channel = nv10_graph_channel; - engine->graph.create_context = nv10_graph_create_context; - engine->graph.destroy_context = nv10_graph_destroy_context; - engine->graph.fifo_access = nv04_graph_fifo_access; - engine->graph.load_context = nv10_graph_load_context; - engine->graph.unload_context = nv10_graph_unload_context; - engine->graph.set_tile_region = nv10_graph_set_tile_region; engine->fifo.channels = 32; engine->fifo.init = nv10_fifo_init; engine->fifo.takedown = nv04_fifo_fini; @@ -157,9 +139,8 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->pm.clock_get = nv04_pm_clock_get; engine->pm.clock_pre = nv04_pm_clock_pre; engine->pm.clock_set = nv04_pm_clock_set; - engine->crypt.init = nouveau_stub_init; - engine->crypt.takedown = nouveau_stub_takedown; engine->vram.init = nouveau_mem_detect; + engine->vram.takedown = nouveau_stub_takedown; engine->vram.flags_valid = nouveau_mem_flags_valid; break; case 0x20: @@ -182,15 +163,6 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->fb.init_tile_region = nv10_fb_init_tile_region; engine->fb.set_tile_region = nv10_fb_set_tile_region; engine->fb.free_tile_region = nv10_fb_free_tile_region; - engine->graph.init = nv20_graph_init; - engine->graph.takedown = nv20_graph_takedown; - engine->graph.channel = nv10_graph_channel; - engine->graph.create_context = nv20_graph_create_context; - engine->graph.destroy_context = nv20_graph_destroy_context; - engine->graph.fifo_access = nv04_graph_fifo_access; - engine->graph.load_context = nv20_graph_load_context; - engine->graph.unload_context = nv20_graph_unload_context; - engine->graph.set_tile_region = nv20_graph_set_tile_region; engine->fifo.channels = 32; engine->fifo.init = nv10_fifo_init; engine->fifo.takedown = nv04_fifo_fini; @@ -216,9 +188,8 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->pm.clock_get = nv04_pm_clock_get; engine->pm.clock_pre = nv04_pm_clock_pre; engine->pm.clock_set = nv04_pm_clock_set; - engine->crypt.init = nouveau_stub_init; - engine->crypt.takedown = nouveau_stub_takedown; engine->vram.init = nouveau_mem_detect; + engine->vram.takedown = nouveau_stub_takedown; engine->vram.flags_valid = nouveau_mem_flags_valid; break; case 0x30: @@ -241,15 +212,6 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->fb.init_tile_region = nv30_fb_init_tile_region; engine->fb.set_tile_region = nv10_fb_set_tile_region; engine->fb.free_tile_region = nv30_fb_free_tile_region; - engine->graph.init = nv30_graph_init; - engine->graph.takedown = nv20_graph_takedown; - engine->graph.fifo_access = nv04_graph_fifo_access; - engine->graph.channel = nv10_graph_channel; - engine->graph.create_context = nv20_graph_create_context; - engine->graph.destroy_context = nv20_graph_destroy_context; - engine->graph.load_context = nv20_graph_load_context; - engine->graph.unload_context = nv20_graph_unload_context; - engine->graph.set_tile_region = nv20_graph_set_tile_region; engine->fifo.channels = 32; engine->fifo.init = nv10_fifo_init; engine->fifo.takedown = nv04_fifo_fini; @@ -277,9 +239,8 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->pm.clock_set = nv04_pm_clock_set; engine->pm.voltage_get = nouveau_voltage_gpio_get; engine->pm.voltage_set = nouveau_voltage_gpio_set; - engine->crypt.init = nouveau_stub_init; - engine->crypt.takedown = nouveau_stub_takedown; engine->vram.init = nouveau_mem_detect; + engine->vram.takedown = nouveau_stub_takedown; engine->vram.flags_valid = nouveau_mem_flags_valid; break; case 0x40: @@ -303,15 +264,6 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->fb.init_tile_region = nv30_fb_init_tile_region; engine->fb.set_tile_region = nv40_fb_set_tile_region; engine->fb.free_tile_region = nv30_fb_free_tile_region; - engine->graph.init = nv40_graph_init; - engine->graph.takedown = nv40_graph_takedown; - engine->graph.fifo_access = nv04_graph_fifo_access; - engine->graph.channel = nv40_graph_channel; - engine->graph.create_context = nv40_graph_create_context; - engine->graph.destroy_context = nv40_graph_destroy_context; - engine->graph.load_context = nv40_graph_load_context; - engine->graph.unload_context = nv40_graph_unload_context; - engine->graph.set_tile_region = nv40_graph_set_tile_region; engine->fifo.channels = 32; engine->fifo.init = nv40_fifo_init; engine->fifo.takedown = nv04_fifo_fini; @@ -340,9 +292,8 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->pm.voltage_get = nouveau_voltage_gpio_get; engine->pm.voltage_set = nouveau_voltage_gpio_set; engine->pm.temp_get = nv40_temp_get; - engine->crypt.init = nouveau_stub_init; - engine->crypt.takedown = nouveau_stub_takedown; engine->vram.init = nouveau_mem_detect; + engine->vram.takedown = nouveau_stub_takedown; engine->vram.flags_valid = nouveau_mem_flags_valid; break; case 0x50: @@ -368,19 +319,6 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->timer.takedown = nv04_timer_takedown; engine->fb.init = nv50_fb_init; engine->fb.takedown = nv50_fb_takedown; - engine->graph.init = nv50_graph_init; - engine->graph.takedown = nv50_graph_takedown; - engine->graph.fifo_access = nv50_graph_fifo_access; - engine->graph.channel = nv50_graph_channel; - engine->graph.create_context = nv50_graph_create_context; - engine->graph.destroy_context = nv50_graph_destroy_context; - engine->graph.load_context = nv50_graph_load_context; - engine->graph.unload_context = nv50_graph_unload_context; - if (dev_priv->chipset == 0x50 || - dev_priv->chipset == 0xac) - engine->graph.tlb_flush = nv50_graph_tlb_flush; - else - engine->graph.tlb_flush = nv84_graph_tlb_flush; engine->fifo.channels = 128; engine->fifo.init = nv50_fifo_init; engine->fifo.takedown = nv50_fifo_takedown; @@ -432,25 +370,8 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->pm.temp_get = nv84_temp_get; else engine->pm.temp_get = nv40_temp_get; - switch (dev_priv->chipset) { - case 0x84: - case 0x86: - case 0x92: - case 0x94: - case 0x96: - case 0xa0: - engine->crypt.init = nv84_crypt_init; - engine->crypt.takedown = nv84_crypt_fini; - engine->crypt.create_context = nv84_crypt_create_context; - engine->crypt.destroy_context = nv84_crypt_destroy_context; - engine->crypt.tlb_flush = nv84_crypt_tlb_flush; - break; - default: - engine->crypt.init = nouveau_stub_init; - engine->crypt.takedown = nouveau_stub_takedown; - break; - } engine->vram.init = nv50_vram_init; + engine->vram.takedown = nv50_vram_fini; engine->vram.get = nv50_vram_new; engine->vram.put = nv50_vram_del; engine->vram.flags_valid = nv50_vram_flags_valid; @@ -472,14 +393,6 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->timer.takedown = nv04_timer_takedown; engine->fb.init = nvc0_fb_init; engine->fb.takedown = nvc0_fb_takedown; - engine->graph.init = nvc0_graph_init; - engine->graph.takedown = nvc0_graph_takedown; - engine->graph.fifo_access = nvc0_graph_fifo_access; - engine->graph.channel = nvc0_graph_channel; - engine->graph.create_context = nvc0_graph_create_context; - engine->graph.destroy_context = nvc0_graph_destroy_context; - engine->graph.load_context = nvc0_graph_load_context; - engine->graph.unload_context = nvc0_graph_unload_context; engine->fifo.channels = 128; engine->fifo.init = nvc0_fifo_init; engine->fifo.takedown = nvc0_fifo_takedown; @@ -503,12 +416,12 @@ static int nouveau_init_engine_ptrs(struct drm_device *dev) engine->gpio.irq_register = nv50_gpio_irq_register; engine->gpio.irq_unregister = nv50_gpio_irq_unregister; engine->gpio.irq_enable = nv50_gpio_irq_enable; - engine->crypt.init = nouveau_stub_init; - engine->crypt.takedown = nouveau_stub_takedown; engine->vram.init = nvc0_vram_init; + engine->vram.takedown = nv50_vram_fini; engine->vram.get = nvc0_vram_new; engine->vram.put = nv50_vram_del; engine->vram.flags_valid = nvc0_vram_flags_valid; + engine->pm.temp_get = nv84_temp_get; break; default: NV_ERROR(dev, "NV%02x unsupported\n", dev_priv->chipset); @@ -542,8 +455,8 @@ nouveau_card_init_channel(struct drm_device *dev) struct drm_nouveau_private *dev_priv = dev->dev_private; int ret; - ret = nouveau_channel_alloc(dev, &dev_priv->channel, - (struct drm_file *)-2, NvDmaFB, NvDmaTT); + ret = nouveau_channel_alloc(dev, &dev_priv->channel, NULL, + NvDmaFB, NvDmaTT); if (ret) return ret; @@ -593,7 +506,7 @@ nouveau_card_init(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_engine *engine; - int ret; + int ret, e = 0; vga_client_register(dev->pdev, dev, NULL, nouveau_vga_set_decode); vga_switcheroo_register_client(dev->pdev, nouveau_switcheroo_set_state, @@ -622,7 +535,7 @@ nouveau_card_init(struct drm_device *dev) nouveau_pm_init(dev); - ret = nouveau_mem_vram_init(dev); + ret = engine->vram.init(dev); if (ret) goto out_bios; @@ -634,10 +547,14 @@ nouveau_card_init(struct drm_device *dev) if (ret) goto out_gpuobj; - ret = nouveau_mem_gart_init(dev); + ret = nouveau_mem_vram_init(dev); if (ret) goto out_instmem; + ret = nouveau_mem_gart_init(dev); + if (ret) + goto out_ttmvram; + /* PMC */ ret = engine->mc.init(dev); if (ret) @@ -658,23 +575,80 @@ nouveau_card_init(struct drm_device *dev) if (ret) goto out_timer; - if (nouveau_noaccel) - engine->graph.accel_blocked = true; - else { - /* PGRAPH */ - ret = engine->graph.init(dev); - if (ret) - goto out_fb; + if (!dev_priv->noaccel) { + switch (dev_priv->card_type) { + case NV_04: + nv04_graph_create(dev); + break; + case NV_10: + nv10_graph_create(dev); + break; + case NV_20: + case NV_30: + nv20_graph_create(dev); + break; + case NV_40: + nv40_graph_create(dev); + break; + case NV_50: + nv50_graph_create(dev); + break; + case NV_C0: + nvc0_graph_create(dev); + break; + default: + break; + } - /* PCRYPT */ - ret = engine->crypt.init(dev); - if (ret) - goto out_graph; + switch (dev_priv->chipset) { + case 0x84: + case 0x86: + case 0x92: + case 0x94: + case 0x96: + case 0xa0: + nv84_crypt_create(dev); + break; + } + + switch (dev_priv->card_type) { + case NV_50: + switch (dev_priv->chipset) { + case 0xa3: + case 0xa5: + case 0xa8: + case 0xaf: + nva3_copy_create(dev); + break; + } + break; + case NV_C0: + nvc0_copy_create(dev, 0); + nvc0_copy_create(dev, 1); + break; + default: + break; + } + + if (dev_priv->card_type == NV_40) + nv40_mpeg_create(dev); + else + if (dev_priv->card_type == NV_50 && + (dev_priv->chipset < 0x98 || dev_priv->chipset == 0xa0)) + nv50_mpeg_create(dev); + + for (e = 0; e < NVOBJ_ENGINE_NR; e++) { + if (dev_priv->eng[e]) { + ret = dev_priv->eng[e]->init(dev, e); + if (ret) + goto out_engine; + } + } /* PFIFO */ ret = engine->fifo.init(dev); if (ret) - goto out_crypt; + goto out_engine; } ret = engine->display.create(dev); @@ -691,7 +665,7 @@ nouveau_card_init(struct drm_device *dev) /* what about PVIDEO/PCRTC/PRAMDAC etc? */ - if (!engine->graph.accel_blocked) { + if (dev_priv->eng[NVOBJ_ENGINE_GR]) { ret = nouveau_fence_init(dev); if (ret) goto out_irq; @@ -713,15 +687,18 @@ out_vblank: drm_vblank_cleanup(dev); engine->display.destroy(dev); out_fifo: - if (!nouveau_noaccel) + if (!dev_priv->noaccel) engine->fifo.takedown(dev); -out_crypt: - if (!nouveau_noaccel) - engine->crypt.takedown(dev); -out_graph: - if (!nouveau_noaccel) - engine->graph.takedown(dev); -out_fb: +out_engine: + if (!dev_priv->noaccel) { + for (e = e - 1; e >= 0; e--) { + if (!dev_priv->eng[e]) + continue; + dev_priv->eng[e]->fini(dev, e, false); + dev_priv->eng[e]->destroy(dev,e ); + } + } + engine->fb.takedown(dev); out_timer: engine->timer.takedown(dev); @@ -731,12 +708,14 @@ out_mc: engine->mc.takedown(dev); out_gart: nouveau_mem_gart_fini(dev); +out_ttmvram: + nouveau_mem_vram_fini(dev); out_instmem: engine->instmem.takedown(dev); out_gpuobj: nouveau_gpuobj_takedown(dev); out_vram: - nouveau_mem_vram_fini(dev); + engine->vram.takedown(dev); out_bios: nouveau_pm_fini(dev); nouveau_bios_takedown(dev); @@ -751,16 +730,26 @@ static void nouveau_card_takedown(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_engine *engine = &dev_priv->engine; + int e; - if (!engine->graph.accel_blocked) { - nouveau_fence_fini(dev); + drm_kms_helper_poll_fini(dev); + nouveau_fbcon_fini(dev); + + if (dev_priv->channel) { nouveau_channel_put_unlocked(&dev_priv->channel); + nouveau_fence_fini(dev); } - if (!nouveau_noaccel) { + engine->display.destroy(dev); + + if (!dev_priv->noaccel) { engine->fifo.takedown(dev); - engine->crypt.takedown(dev); - engine->graph.takedown(dev); + for (e = NVOBJ_ENGINE_NR - 1; e >= 0; e--) { + if (dev_priv->eng[e]) { + dev_priv->eng[e]->fini(dev, e, false); + dev_priv->eng[e]->destroy(dev,e ); + } + } } engine->fb.takedown(dev); engine->timer.takedown(dev); @@ -778,10 +767,11 @@ static void nouveau_card_takedown(struct drm_device *dev) ttm_bo_clean_mm(&dev_priv->ttm.bdev, TTM_PL_TT); mutex_unlock(&dev->struct_mutex); nouveau_mem_gart_fini(dev); + nouveau_mem_vram_fini(dev); engine->instmem.takedown(dev); nouveau_gpuobj_takedown(dev); - nouveau_mem_vram_fini(dev); + engine->vram.takedown(dev); nouveau_irq_fini(dev); drm_vblank_cleanup(dev); @@ -792,6 +782,41 @@ static void nouveau_card_takedown(struct drm_device *dev) vga_client_register(dev->pdev, NULL, NULL, NULL); } +int +nouveau_open(struct drm_device *dev, struct drm_file *file_priv) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fpriv *fpriv; + int ret; + + fpriv = kzalloc(sizeof(*fpriv), GFP_KERNEL); + if (unlikely(!fpriv)) + return -ENOMEM; + + spin_lock_init(&fpriv->lock); + INIT_LIST_HEAD(&fpriv->channels); + + if (dev_priv->card_type == NV_50) { + ret = nouveau_vm_new(dev, 0, (1ULL << 40), 0x0020000000ULL, + &fpriv->vm); + if (ret) { + kfree(fpriv); + return ret; + } + } else + if (dev_priv->card_type >= NV_C0) { + ret = nouveau_vm_new(dev, 0, (1ULL << 40), 0x0008000000ULL, + &fpriv->vm); + if (ret) { + kfree(fpriv); + return ret; + } + } + + file_priv->driver_priv = fpriv; + return 0; +} + /* here a client dies, release the stuff that was allocated for its * file_priv */ void nouveau_preclose(struct drm_device *dev, struct drm_file *file_priv) @@ -799,6 +824,14 @@ void nouveau_preclose(struct drm_device *dev, struct drm_file *file_priv) nouveau_channel_cleanup(dev, file_priv); } +void +nouveau_postclose(struct drm_device *dev, struct drm_file *file_priv) +{ + struct nouveau_fpriv *fpriv = nouveau_fpriv(file_priv); + nouveau_vm_ref(NULL, &fpriv->vm, NULL); + kfree(fpriv); +} + /* first module load, setup the mmio/fb mapping */ /* KMS: we need mmio at load time, not when the first drm client opens. */ int nouveau_firstopen(struct drm_device *dev) @@ -866,7 +899,7 @@ static int nouveau_remove_conflicting_drivers(struct drm_device *dev) #ifdef CONFIG_X86 primary = dev->pdev->resource[PCI_ROM_RESOURCE].flags & IORESOURCE_ROM_SHADOW; #endif - + remove_conflicting_framebuffers(dev_priv->apertures, "nouveaufb", primary); return 0; } @@ -910,19 +943,21 @@ int nouveau_load(struct drm_device *dev, unsigned long flags) #ifdef __BIG_ENDIAN /* Put the card in BE mode if it's not */ - if (nv_rd32(dev, NV03_PMC_BOOT_1)) - nv_wr32(dev, NV03_PMC_BOOT_1, 0x00000001); + if (nv_rd32(dev, NV03_PMC_BOOT_1) != 0x01000001) + nv_wr32(dev, NV03_PMC_BOOT_1, 0x01000001); DRM_MEMORYBARRIER(); #endif /* Time to determine the card architecture */ reg0 = nv_rd32(dev, NV03_PMC_BOOT_0); + dev_priv->stepping = 0; /* XXX: add stepping for pre-NV10? */ /* We're dealing with >=NV10 */ if ((reg0 & 0x0f000000) > 0) { /* Bit 27-20 contain the architecture in hex */ dev_priv->chipset = (reg0 & 0xff00000) >> 20; + dev_priv->stepping = (reg0 & 0xff); /* NV04 or NV05 */ } else if ((reg0 & 0xff00fff0) == 0x20004000) { if (reg0 & 0x00f00000) @@ -961,6 +996,25 @@ int nouveau_load(struct drm_device *dev, unsigned long flags) NV_INFO(dev, "Detected an NV%2x generation card (0x%08x)\n", dev_priv->card_type, reg0); + /* Determine whether we'll attempt acceleration or not, some + * cards are disabled by default here due to them being known + * non-functional, or never been tested due to lack of hw. + */ + dev_priv->noaccel = !!nouveau_noaccel; + if (nouveau_noaccel == -1) { + switch (dev_priv->chipset) { + case 0xc1: /* known broken */ + case 0xc8: /* never tested */ + NV_INFO(dev, "acceleration disabled by default, pass " + "noaccel=0 to force enable\n"); + dev_priv->noaccel = true; + break; + default: + dev_priv->noaccel = false; + break; + } + } + ret = nouveau_remove_conflicting_drivers(dev); if (ret) goto err_mmio; @@ -1025,11 +1079,7 @@ void nouveau_lastclose(struct drm_device *dev) int nouveau_unload(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_engine *engine = &dev_priv->engine; - drm_kms_helper_poll_fini(dev); - nouveau_fbcon_fini(dev); - engine->display.destroy(dev); nouveau_card_takedown(dev); iounmap(dev_priv->mmio); @@ -1059,7 +1109,7 @@ int nouveau_ioctl_getparam(struct drm_device *dev, void *data, case NOUVEAU_GETPARAM_BUS_TYPE: if (drm_pci_device_is_agp(dev)) getparam->value = NV_AGP; - else if (drm_pci_device_is_pcie(dev)) + else if (pci_is_pcie(dev->pdev)) getparam->value = NV_PCIE; else getparam->value = NV_PCI; diff --git a/drivers/gpu/drm/nouveau/nouveau_temp.c b/drivers/gpu/drm/nouveau/nouveau_temp.c index 649b0413b09..081ca7b03e8 100644 --- a/drivers/gpu/drm/nouveau/nouveau_temp.c +++ b/drivers/gpu/drm/nouveau/nouveau_temp.c @@ -43,7 +43,7 @@ nouveau_temp_vbios_parse(struct drm_device *dev, u8 *temp) /* Set the default sensor's contants */ sensor->offset_constant = 0; - sensor->offset_mult = 1; + sensor->offset_mult = 0; sensor->offset_div = 1; sensor->slope_mult = 1; sensor->slope_div = 1; @@ -99,6 +99,13 @@ nouveau_temp_vbios_parse(struct drm_device *dev, u8 *temp) sensor->slope_mult = 431; sensor->slope_div = 10000; break; + + case 0x67: + sensor->offset_mult = -26149; + sensor->offset_div = 100; + sensor->slope_mult = 484; + sensor->slope_div = 10000; + break; } } @@ -109,7 +116,7 @@ nouveau_temp_vbios_parse(struct drm_device *dev, u8 *temp) /* Read the entries from the table */ for (i = 0; i < entries; i++) { - u16 value = ROM16(temp[1]); + s16 value = ROM16(temp[1]); switch (temp[0]) { case 0x01: @@ -160,8 +167,8 @@ nv40_sensor_setup(struct drm_device *dev) struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_pm_engine *pm = &dev_priv->engine.pm; struct nouveau_pm_temp_sensor_constants *sensor = &pm->sensor_constants; - u32 offset = sensor->offset_mult / sensor->offset_div; - u32 sensor_calibration; + s32 offset = sensor->offset_mult / sensor->offset_div; + s32 sensor_calibration; /* set up the sensors */ sensor_calibration = 120 - offset - sensor->offset_constant; diff --git a/drivers/gpu/drm/nouveau/nouveau_vm.c b/drivers/gpu/drm/nouveau/nouveau_vm.c index 0059e6f58a8..244fd38fdb8 100644 --- a/drivers/gpu/drm/nouveau/nouveau_vm.c +++ b/drivers/gpu/drm/nouveau/nouveau_vm.c @@ -58,6 +58,7 @@ nouveau_vm_map_at(struct nouveau_vma *vma, u64 delta, struct nouveau_mem *node) num -= len; pte += len; if (unlikely(end >= max)) { + phys += len << (bits + 12); pde++; pte = 0; } @@ -368,23 +369,26 @@ nouveau_vm_link(struct nouveau_vm *vm, struct nouveau_gpuobj *pgd) } static void -nouveau_vm_unlink(struct nouveau_vm *vm, struct nouveau_gpuobj *pgd) +nouveau_vm_unlink(struct nouveau_vm *vm, struct nouveau_gpuobj *mpgd) { struct nouveau_vm_pgd *vpgd, *tmp; + struct nouveau_gpuobj *pgd = NULL; - if (!pgd) + if (!mpgd) return; mutex_lock(&vm->mm->mutex); list_for_each_entry_safe(vpgd, tmp, &vm->pgd_list, head) { - if (vpgd->obj != pgd) - continue; - - list_del(&vpgd->head); - nouveau_gpuobj_ref(NULL, &vpgd->obj); - kfree(vpgd); + if (vpgd->obj == mpgd) { + pgd = vpgd->obj; + list_del(&vpgd->head); + kfree(vpgd); + break; + } } mutex_unlock(&vm->mm->mutex); + + nouveau_gpuobj_ref(NULL, &pgd); } static void @@ -395,8 +399,8 @@ nouveau_vm_del(struct nouveau_vm *vm) list_for_each_entry_safe(vpgd, tmp, &vm->pgd_list, head) { nouveau_vm_unlink(vm, vpgd->obj); } - WARN_ON(nouveau_mm_fini(&vm->mm) != 0); + nouveau_mm_fini(&vm->mm); kfree(vm->pgt); kfree(vm); } diff --git a/drivers/gpu/drm/nouveau/nouveau_vm.h b/drivers/gpu/drm/nouveau/nouveau_vm.h index 2e06b55cfdc..579ca8cc223 100644 --- a/drivers/gpu/drm/nouveau/nouveau_vm.h +++ b/drivers/gpu/drm/nouveau/nouveau_vm.h @@ -41,6 +41,8 @@ struct nouveau_vm_pgd { }; struct nouveau_vma { + struct list_head head; + int refcount; struct nouveau_vm *vm; struct nouveau_mm_node *node; u64 offset; @@ -53,8 +55,7 @@ struct nouveau_vm { int refcount; struct list_head pgd_list; - atomic_t pgraph_refs; - atomic_t pcrypt_refs; + atomic_t engref[16]; struct nouveau_vm_pgt *pgt; u32 fpde; diff --git a/drivers/gpu/drm/nouveau/nouveau_volt.c b/drivers/gpu/drm/nouveau/nouveau_volt.c index 04fdc00a67d..75e872741d9 100644 --- a/drivers/gpu/drm/nouveau/nouveau_volt.c +++ b/drivers/gpu/drm/nouveau/nouveau_volt.c @@ -159,8 +159,16 @@ nouveau_volt_init(struct drm_device *dev) headerlen = volt[1]; recordlen = volt[2]; entries = volt[3]; - vidshift = hweight8(volt[5]); vidmask = volt[4]; + /* no longer certain what volt[5] is, if it's related to + * the vid shift then it's definitely not a function of + * how many bits are set. + * + * after looking at a number of nva3+ vbios images, they + * all seem likely to have a static shift of 2.. lets + * go with that for now until proven otherwise. + */ + vidshift = 2; break; default: NV_WARN(dev, "voltage table 0x%02x unknown\n", volt[0]); diff --git a/drivers/gpu/drm/nouveau/nv04_crtc.c b/drivers/gpu/drm/nouveau/nv04_crtc.c index 748b9d9c294..118261d4927 100644 --- a/drivers/gpu/drm/nouveau/nv04_crtc.c +++ b/drivers/gpu/drm/nouveau/nv04_crtc.c @@ -376,7 +376,10 @@ nv_crtc_mode_set_vga(struct drm_crtc *crtc, struct drm_display_mode *mode) */ /* framebuffer can be larger than crtc scanout area. */ - regp->CRTC[NV_CIO_CRE_RPC0_INDEX] = XLATE(fb->pitch / 8, 8, NV_CIO_CRE_RPC0_OFFSET_10_8); + regp->CRTC[NV_CIO_CRE_RPC0_INDEX] = + XLATE(fb->pitch / 8, 8, NV_CIO_CRE_RPC0_OFFSET_10_8); + regp->CRTC[NV_CIO_CRE_42] = + XLATE(fb->pitch / 8, 11, NV_CIO_CRE_42_OFFSET_11); regp->CRTC[NV_CIO_CRE_RPC1_INDEX] = mode->crtc_hdisplay < 1280 ? MASK(NV_CIO_CRE_RPC1_LARGE) : 0x00; regp->CRTC[NV_CIO_CRE_LSR_INDEX] = XLATE(horizBlankEnd, 6, NV_CIO_CRE_LSR_HBE_6) | @@ -790,8 +793,7 @@ nv04_crtc_do_mode_set_base(struct drm_crtc *crtc, if (atomic) { drm_fb = passed_fb; fb = nouveau_framebuffer(passed_fb); - } - else { + } else { /* If not atomic, we can go ahead and pin, and unpin the * old fb we were passed. */ @@ -825,8 +827,11 @@ nv04_crtc_do_mode_set_base(struct drm_crtc *crtc, regp->CRTC[NV_CIO_CR_OFFSET_INDEX] = drm_fb->pitch >> 3; regp->CRTC[NV_CIO_CRE_RPC0_INDEX] = XLATE(drm_fb->pitch >> 3, 8, NV_CIO_CRE_RPC0_OFFSET_10_8); + regp->CRTC[NV_CIO_CRE_42] = + XLATE(drm_fb->pitch / 8, 11, NV_CIO_CRE_42_OFFSET_11); crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_RPC0_INDEX); crtc_wr_cio_state(crtc, regp, NV_CIO_CR_OFFSET_INDEX); + crtc_wr_cio_state(crtc, regp, NV_CIO_CRE_42); /* Update the framebuffer location. */ regp->fb_start = nv_crtc->fb.offset & ~3; @@ -944,14 +949,14 @@ nv04_crtc_cursor_set(struct drm_crtc *crtc, struct drm_file *file_priv, struct drm_gem_object *gem; int ret = 0; - if (width != 64 || height != 64) - return -EINVAL; - if (!buffer_handle) { nv_crtc->cursor.hide(nv_crtc, true); return 0; } + if (width != 64 || height != 64) + return -EINVAL; + gem = drm_gem_object_lookup(dev, file_priv, buffer_handle); if (!gem) return -ENOENT; @@ -1030,7 +1035,7 @@ nv04_crtc_create(struct drm_device *dev, int crtc_num) drm_crtc_helper_add(&nv_crtc->base, &nv04_crtc_helper_funcs); drm_mode_crtc_set_gamma_size(&nv_crtc->base, 256); - ret = nouveau_bo_new(dev, NULL, 64*64*4, 0x100, TTM_PL_FLAG_VRAM, + ret = nouveau_bo_new(dev, 64*64*4, 0x100, TTM_PL_FLAG_VRAM, 0, 0x0000, &nv_crtc->cursor.nvbo); if (!ret) { ret = nouveau_bo_pin(nv_crtc->cursor.nvbo, TTM_PL_FLAG_VRAM); diff --git a/drivers/gpu/drm/nouveau/nv04_graph.c b/drivers/gpu/drm/nouveau/nv04_graph.c index af75015068d..dbdea8ed392 100644 --- a/drivers/gpu/drm/nouveau/nv04_graph.c +++ b/drivers/gpu/drm/nouveau/nv04_graph.c @@ -28,9 +28,11 @@ #include "nouveau_drv.h" #include "nouveau_hw.h" #include "nouveau_util.h" +#include "nouveau_ramht.h" -static int nv04_graph_register(struct drm_device *dev); -static void nv04_graph_isr(struct drm_device *dev); +struct nv04_graph_engine { + struct nouveau_exec_engine base; +}; static uint32_t nv04_graph_ctx_regs[] = { 0x0040053c, @@ -350,7 +352,7 @@ struct graph_state { uint32_t nv04[ARRAY_SIZE(nv04_graph_ctx_regs)]; }; -struct nouveau_channel * +static struct nouveau_channel * nv04_graph_channel(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; @@ -365,26 +367,6 @@ nv04_graph_channel(struct drm_device *dev) return dev_priv->channels.ptr[chid]; } -static void -nv04_graph_context_switch(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct nouveau_channel *chan = NULL; - int chid; - - nouveau_wait_for_idle(dev); - - /* If previous context is valid, we need to save it */ - pgraph->unload_context(dev); - - /* Load context for next channel */ - chid = dev_priv->engine.fifo.channel_id(dev); - chan = dev_priv->channels.ptr[chid]; - if (chan) - nv04_graph_load_context(chan); -} - static uint32_t *ctx_reg(struct graph_state *ctx, uint32_t reg) { int i; @@ -397,48 +379,11 @@ static uint32_t *ctx_reg(struct graph_state *ctx, uint32_t reg) return NULL; } -int nv04_graph_create_context(struct nouveau_channel *chan) -{ - struct graph_state *pgraph_ctx; - NV_DEBUG(chan->dev, "nv04_graph_context_create %d\n", chan->id); - - chan->pgraph_ctx = pgraph_ctx = kzalloc(sizeof(*pgraph_ctx), - GFP_KERNEL); - if (pgraph_ctx == NULL) - return -ENOMEM; - - *ctx_reg(pgraph_ctx, NV04_PGRAPH_DEBUG_3) = 0xfad4ff31; - - return 0; -} - -void nv04_graph_destroy_context(struct nouveau_channel *chan) -{ - struct drm_device *dev = chan->dev; - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct graph_state *pgraph_ctx = chan->pgraph_ctx; - unsigned long flags; - - spin_lock_irqsave(&dev_priv->context_switch_lock, flags); - pgraph->fifo_access(dev, false); - - /* Unload the context if it's the currently active one */ - if (pgraph->channel(dev) == chan) - pgraph->unload_context(dev); - - /* Free the context resources */ - kfree(pgraph_ctx); - chan->pgraph_ctx = NULL; - - pgraph->fifo_access(dev, true); - spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); -} - -int nv04_graph_load_context(struct nouveau_channel *chan) +static int +nv04_graph_load_context(struct nouveau_channel *chan) { + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; struct drm_device *dev = chan->dev; - struct graph_state *pgraph_ctx = chan->pgraph_ctx; uint32_t tmp; int i; @@ -456,20 +401,19 @@ int nv04_graph_load_context(struct nouveau_channel *chan) return 0; } -int +static int nv04_graph_unload_context(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; struct nouveau_channel *chan = NULL; struct graph_state *ctx; uint32_t tmp; int i; - chan = pgraph->channel(dev); + chan = nv04_graph_channel(dev); if (!chan) return 0; - ctx = chan->pgraph_ctx; + ctx = chan->engctx[NVOBJ_ENGINE_GR]; for (i = 0; i < ARRAY_SIZE(nv04_graph_ctx_regs); i++) ctx->nv04[i] = nv_rd32(dev, nv04_graph_ctx_regs[i]); @@ -481,23 +425,85 @@ nv04_graph_unload_context(struct drm_device *dev) return 0; } -int nv04_graph_init(struct drm_device *dev) +static int +nv04_graph_context_new(struct nouveau_channel *chan, int engine) { + struct graph_state *pgraph_ctx; + NV_DEBUG(chan->dev, "nv04_graph_context_create %d\n", chan->id); + + pgraph_ctx = kzalloc(sizeof(*pgraph_ctx), GFP_KERNEL); + if (pgraph_ctx == NULL) + return -ENOMEM; + + *ctx_reg(pgraph_ctx, NV04_PGRAPH_DEBUG_3) = 0xfad4ff31; + + chan->engctx[engine] = pgraph_ctx; + return 0; +} + +static void +nv04_graph_context_del(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; - uint32_t tmp; + struct graph_state *pgraph_ctx = chan->engctx[engine]; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000000); + + /* Unload the context if it's the currently active one */ + if (nv04_graph_channel(dev) == chan) + nv04_graph_unload_context(dev); + + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + /* Free the context resources */ + kfree(pgraph_ctx); + chan->engctx[engine] = NULL; +} + +int +nv04_graph_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct drm_device *dev = chan->dev; + struct nouveau_gpuobj *obj = NULL; int ret; + ret = nouveau_gpuobj_new(dev, chan, 16, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 1; + obj->class = class; + +#ifdef __BIG_ENDIAN + nv_wo32(obj, 0x00, 0x00080000 | class); +#else + nv_wo32(obj, 0x00, class); +#endif + nv_wo32(obj, 0x04, 0x00000000); + nv_wo32(obj, 0x08, 0x00000000); + nv_wo32(obj, 0x0c, 0x00000000); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static int +nv04_graph_init(struct drm_device *dev, int engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t tmp; + nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PGRAPH); nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | NV_PMC_ENABLE_PGRAPH); - ret = nv04_graph_register(dev); - if (ret) - return ret; - /* Enable PGRAPH interrupts */ - nouveau_irq_register(dev, 12, nv04_graph_isr); nv_wr32(dev, NV03_PGRAPH_INTR, 0xFFFFFFFF); nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0xFFFFFFFF); @@ -507,7 +513,7 @@ int nv04_graph_init(struct drm_device *dev) nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0x001FFFFF);*/ nv_wr32(dev, NV04_PGRAPH_DEBUG_0, 0x1231c000); /*1231C000 blob, 001 haiku*/ - //*V_WRITE(NV04_PGRAPH_DEBUG_1, 0xf2d91100);*/ + /*V_WRITE(NV04_PGRAPH_DEBUG_1, 0xf2d91100);*/ nv_wr32(dev, NV04_PGRAPH_DEBUG_1, 0x72111100); /*0x72111100 blob , 01 haiku*/ /*nv_wr32(dev, NV04_PGRAPH_DEBUG_2, 0x11d5f870);*/ @@ -531,21 +537,17 @@ int nv04_graph_init(struct drm_device *dev) return 0; } -void nv04_graph_takedown(struct drm_device *dev) +static int +nv04_graph_fini(struct drm_device *dev, int engine, bool suspend) { + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000000); + if (!nv_wait(dev, NV04_PGRAPH_STATUS, ~0, 0) && suspend) { + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000001); + return -EBUSY; + } + nv04_graph_unload_context(dev); nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0x00000000); - nouveau_irq_unregister(dev, 12); -} - -void -nv04_graph_fifo_access(struct drm_device *dev, bool enabled) -{ - if (enabled) - nv_wr32(dev, NV04_PGRAPH_FIFO, - nv_rd32(dev, NV04_PGRAPH_FIFO) | 1); - else - nv_wr32(dev, NV04_PGRAPH_FIFO, - nv_rd32(dev, NV04_PGRAPH_FIFO) & ~1); + return 0; } static int @@ -969,13 +971,138 @@ nv04_graph_mthd_bind_chroma(struct nouveau_channel *chan, return 1; } -static int -nv04_graph_register(struct drm_device *dev) +static struct nouveau_bitfield nv04_graph_intr[] = { + { NV_PGRAPH_INTR_NOTIFY, "NOTIFY" }, + {} +}; + +static struct nouveau_bitfield nv04_graph_nstatus[] = { + { NV04_PGRAPH_NSTATUS_STATE_IN_USE, "STATE_IN_USE" }, + { NV04_PGRAPH_NSTATUS_INVALID_STATE, "INVALID_STATE" }, + { NV04_PGRAPH_NSTATUS_BAD_ARGUMENT, "BAD_ARGUMENT" }, + { NV04_PGRAPH_NSTATUS_PROTECTION_FAULT, "PROTECTION_FAULT" }, + {} +}; + +struct nouveau_bitfield nv04_graph_nsource[] = { + { NV03_PGRAPH_NSOURCE_NOTIFICATION, "NOTIFICATION" }, + { NV03_PGRAPH_NSOURCE_DATA_ERROR, "DATA_ERROR" }, + { NV03_PGRAPH_NSOURCE_PROTECTION_ERROR, "PROTECTION_ERROR" }, + { NV03_PGRAPH_NSOURCE_RANGE_EXCEPTION, "RANGE_EXCEPTION" }, + { NV03_PGRAPH_NSOURCE_LIMIT_COLOR, "LIMIT_COLOR" }, + { NV03_PGRAPH_NSOURCE_LIMIT_ZETA, "LIMIT_ZETA" }, + { NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD, "ILLEGAL_MTHD" }, + { NV03_PGRAPH_NSOURCE_DMA_R_PROTECTION, "DMA_R_PROTECTION" }, + { NV03_PGRAPH_NSOURCE_DMA_W_PROTECTION, "DMA_W_PROTECTION" }, + { NV03_PGRAPH_NSOURCE_FORMAT_EXCEPTION, "FORMAT_EXCEPTION" }, + { NV03_PGRAPH_NSOURCE_PATCH_EXCEPTION, "PATCH_EXCEPTION" }, + { NV03_PGRAPH_NSOURCE_STATE_INVALID, "STATE_INVALID" }, + { NV03_PGRAPH_NSOURCE_DOUBLE_NOTIFY, "DOUBLE_NOTIFY" }, + { NV03_PGRAPH_NSOURCE_NOTIFY_IN_USE, "NOTIFY_IN_USE" }, + { NV03_PGRAPH_NSOURCE_METHOD_CNT, "METHOD_CNT" }, + { NV03_PGRAPH_NSOURCE_BFR_NOTIFICATION, "BFR_NOTIFICATION" }, + { NV03_PGRAPH_NSOURCE_DMA_VTX_PROTECTION, "DMA_VTX_PROTECTION" }, + { NV03_PGRAPH_NSOURCE_DMA_WIDTH_A, "DMA_WIDTH_A" }, + { NV03_PGRAPH_NSOURCE_DMA_WIDTH_B, "DMA_WIDTH_B" }, + {} +}; + +static void +nv04_graph_context_switch(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_channel *chan = NULL; + int chid; - if (dev_priv->engine.graph.registered) - return 0; + nouveau_wait_for_idle(dev); + + /* If previous context is valid, we need to save it */ + nv04_graph_unload_context(dev); + + /* Load context for next channel */ + chid = dev_priv->engine.fifo.channel_id(dev); + chan = dev_priv->channels.ptr[chid]; + if (chan) + nv04_graph_load_context(chan); +} + +static void +nv04_graph_isr(struct drm_device *dev) +{ + u32 stat; + + while ((stat = nv_rd32(dev, NV03_PGRAPH_INTR))) { + u32 nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE); + u32 nstatus = nv_rd32(dev, NV03_PGRAPH_NSTATUS); + u32 addr = nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR); + u32 chid = (addr & 0x0f000000) >> 24; + u32 subc = (addr & 0x0000e000) >> 13; + u32 mthd = (addr & 0x00001ffc); + u32 data = nv_rd32(dev, NV04_PGRAPH_TRAPPED_DATA); + u32 class = nv_rd32(dev, 0x400180 + subc * 4) & 0xff; + u32 show = stat; + + if (stat & NV_PGRAPH_INTR_NOTIFY) { + if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) { + if (!nouveau_gpuobj_mthd_call2(dev, chid, class, mthd, data)) + show &= ~NV_PGRAPH_INTR_NOTIFY; + } + } + + if (stat & NV_PGRAPH_INTR_CONTEXT_SWITCH) { + nv_wr32(dev, NV03_PGRAPH_INTR, NV_PGRAPH_INTR_CONTEXT_SWITCH); + stat &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; + show &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; + nv04_graph_context_switch(dev); + } + + nv_wr32(dev, NV03_PGRAPH_INTR, stat); + nv_wr32(dev, NV04_PGRAPH_FIFO, 0x00000001); + + if (show && nouveau_ratelimit()) { + NV_INFO(dev, "PGRAPH -"); + nouveau_bitfield_print(nv04_graph_intr, show); + printk(" nsource:"); + nouveau_bitfield_print(nv04_graph_nsource, nsource); + printk(" nstatus:"); + nouveau_bitfield_print(nv04_graph_nstatus, nstatus); + printk("\n"); + NV_INFO(dev, "PGRAPH - ch %d/%d class 0x%04x " + "mthd 0x%04x data 0x%08x\n", + chid, subc, class, mthd, data); + } + } +} + +static void +nv04_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv04_graph_engine *pgraph = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 12); + + NVOBJ_ENGINE_DEL(dev, GR); + kfree(pgraph); +} + +int +nv04_graph_create(struct drm_device *dev) +{ + struct nv04_graph_engine *pgraph; + + pgraph = kzalloc(sizeof(*pgraph), GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + pgraph->base.destroy = nv04_graph_destroy; + pgraph->base.init = nv04_graph_init; + pgraph->base.fini = nv04_graph_fini; + pgraph->base.context_new = nv04_graph_context_new; + pgraph->base.context_del = nv04_graph_context_del; + pgraph->base.object_new = nv04_graph_object_new; + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + nouveau_irq_register(dev, 12, nv04_graph_isr); /* dvd subpicture */ NVOBJ_CLASS(dev, 0x0038, GR); @@ -1222,93 +1349,5 @@ nv04_graph_register(struct drm_device *dev) NVOBJ_CLASS(dev, 0x506e, SW); NVOBJ_MTHD (dev, 0x506e, 0x0150, nv04_graph_mthd_set_ref); NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); - - dev_priv->engine.graph.registered = true; return 0; -}; - -static struct nouveau_bitfield nv04_graph_intr[] = { - { NV_PGRAPH_INTR_NOTIFY, "NOTIFY" }, - {} -}; - -static struct nouveau_bitfield nv04_graph_nstatus[] = -{ - { NV04_PGRAPH_NSTATUS_STATE_IN_USE, "STATE_IN_USE" }, - { NV04_PGRAPH_NSTATUS_INVALID_STATE, "INVALID_STATE" }, - { NV04_PGRAPH_NSTATUS_BAD_ARGUMENT, "BAD_ARGUMENT" }, - { NV04_PGRAPH_NSTATUS_PROTECTION_FAULT, "PROTECTION_FAULT" }, - {} -}; - -struct nouveau_bitfield nv04_graph_nsource[] = -{ - { NV03_PGRAPH_NSOURCE_NOTIFICATION, "NOTIFICATION" }, - { NV03_PGRAPH_NSOURCE_DATA_ERROR, "DATA_ERROR" }, - { NV03_PGRAPH_NSOURCE_PROTECTION_ERROR, "PROTECTION_ERROR" }, - { NV03_PGRAPH_NSOURCE_RANGE_EXCEPTION, "RANGE_EXCEPTION" }, - { NV03_PGRAPH_NSOURCE_LIMIT_COLOR, "LIMIT_COLOR" }, - { NV03_PGRAPH_NSOURCE_LIMIT_ZETA, "LIMIT_ZETA" }, - { NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD, "ILLEGAL_MTHD" }, - { NV03_PGRAPH_NSOURCE_DMA_R_PROTECTION, "DMA_R_PROTECTION" }, - { NV03_PGRAPH_NSOURCE_DMA_W_PROTECTION, "DMA_W_PROTECTION" }, - { NV03_PGRAPH_NSOURCE_FORMAT_EXCEPTION, "FORMAT_EXCEPTION" }, - { NV03_PGRAPH_NSOURCE_PATCH_EXCEPTION, "PATCH_EXCEPTION" }, - { NV03_PGRAPH_NSOURCE_STATE_INVALID, "STATE_INVALID" }, - { NV03_PGRAPH_NSOURCE_DOUBLE_NOTIFY, "DOUBLE_NOTIFY" }, - { NV03_PGRAPH_NSOURCE_NOTIFY_IN_USE, "NOTIFY_IN_USE" }, - { NV03_PGRAPH_NSOURCE_METHOD_CNT, "METHOD_CNT" }, - { NV03_PGRAPH_NSOURCE_BFR_NOTIFICATION, "BFR_NOTIFICATION" }, - { NV03_PGRAPH_NSOURCE_DMA_VTX_PROTECTION, "DMA_VTX_PROTECTION" }, - { NV03_PGRAPH_NSOURCE_DMA_WIDTH_A, "DMA_WIDTH_A" }, - { NV03_PGRAPH_NSOURCE_DMA_WIDTH_B, "DMA_WIDTH_B" }, - {} -}; - -static void -nv04_graph_isr(struct drm_device *dev) -{ - u32 stat; - - while ((stat = nv_rd32(dev, NV03_PGRAPH_INTR))) { - u32 nsource = nv_rd32(dev, NV03_PGRAPH_NSOURCE); - u32 nstatus = nv_rd32(dev, NV03_PGRAPH_NSTATUS); - u32 addr = nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR); - u32 chid = (addr & 0x0f000000) >> 24; - u32 subc = (addr & 0x0000e000) >> 13; - u32 mthd = (addr & 0x00001ffc); - u32 data = nv_rd32(dev, NV04_PGRAPH_TRAPPED_DATA); - u32 class = nv_rd32(dev, 0x400180 + subc * 4) & 0xff; - u32 show = stat; - - if (stat & NV_PGRAPH_INTR_NOTIFY) { - if (nsource & NV03_PGRAPH_NSOURCE_ILLEGAL_MTHD) { - if (!nouveau_gpuobj_mthd_call2(dev, chid, class, mthd, data)) - show &= ~NV_PGRAPH_INTR_NOTIFY; - } - } - - if (stat & NV_PGRAPH_INTR_CONTEXT_SWITCH) { - nv_wr32(dev, NV03_PGRAPH_INTR, NV_PGRAPH_INTR_CONTEXT_SWITCH); - stat &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; - show &= ~NV_PGRAPH_INTR_CONTEXT_SWITCH; - nv04_graph_context_switch(dev); - } - - nv_wr32(dev, NV03_PGRAPH_INTR, stat); - nv_wr32(dev, NV04_PGRAPH_FIFO, 0x00000001); - - if (show && nouveau_ratelimit()) { - NV_INFO(dev, "PGRAPH -"); - nouveau_bitfield_print(nv04_graph_intr, show); - printk(" nsource:"); - nouveau_bitfield_print(nv04_graph_nsource, nsource); - printk(" nstatus:"); - nouveau_bitfield_print(nv04_graph_nstatus, nstatus); - printk("\n"); - NV_INFO(dev, "PGRAPH - ch %d/%d class 0x%04x " - "mthd 0x%04x data 0x%08x\n", - chid, subc, class, mthd, data); - } - } } diff --git a/drivers/gpu/drm/nouveau/nv04_instmem.c b/drivers/gpu/drm/nouveau/nv04_instmem.c index b8e3edb5c06..c1248e0740a 100644 --- a/drivers/gpu/drm/nouveau/nv04_instmem.c +++ b/drivers/gpu/drm/nouveau/nv04_instmem.c @@ -28,6 +28,31 @@ int nv04_instmem_init(struct drm_device *dev) /* RAMIN always available */ dev_priv->ramin_available = true; + /* Reserve space at end of VRAM for PRAMIN */ + if (dev_priv->card_type >= NV_40) { + u32 vs = hweight8((nv_rd32(dev, 0x001540) & 0x0000ff00) >> 8); + u32 rsvd; + + /* estimate grctx size, the magics come from nv40_grctx.c */ + if (dev_priv->chipset == 0x40) rsvd = 0x6aa0 * vs; + else if (dev_priv->chipset < 0x43) rsvd = 0x4f00 * vs; + else if (nv44_graph_class(dev)) rsvd = 0x4980 * vs; + else rsvd = 0x4a40 * vs; + rsvd += 16 * 1024; + rsvd *= dev_priv->engine.fifo.channels; + + /* pciegart table */ + if (pci_is_pcie(dev->pdev)) + rsvd += 512 * 1024; + + /* object storage */ + rsvd += 512 * 1024; + + dev_priv->ramin_rsvd_vram = round_up(rsvd, 4096); + } else { + dev_priv->ramin_rsvd_vram = 512 * 1024; + } + /* Setup shared RAMHT */ ret = nouveau_gpuobj_new_fake(dev, 0x10000, ~0, 4096, NVOBJ_FLAG_ZERO_ALLOC, &ramht); @@ -95,6 +120,9 @@ nv04_instmem_takedown(struct drm_device *dev) nouveau_ramht_ref(NULL, &dev_priv->ramht, NULL); nouveau_gpuobj_ref(NULL, &dev_priv->ramro); nouveau_gpuobj_ref(NULL, &dev_priv->ramfc); + + if (drm_mm_initialized(&dev_priv->ramin_heap)) + drm_mm_takedown(&dev_priv->ramin_heap); } int @@ -109,7 +137,8 @@ nv04_instmem_resume(struct drm_device *dev) } int -nv04_instmem_get(struct nouveau_gpuobj *gpuobj, u32 size, u32 align) +nv04_instmem_get(struct nouveau_gpuobj *gpuobj, struct nouveau_channel *chan, + u32 size, u32 align) { struct drm_nouveau_private *dev_priv = gpuobj->dev->dev_private; struct drm_mm_node *ramin = NULL; diff --git a/drivers/gpu/drm/nouveau/nv10_graph.c b/drivers/gpu/drm/nouveau/nv10_graph.c index 8c92edb7bbc..7255e4a4d3f 100644 --- a/drivers/gpu/drm/nouveau/nv10_graph.c +++ b/drivers/gpu/drm/nouveau/nv10_graph.c @@ -28,10 +28,9 @@ #include "nouveau_drv.h" #include "nouveau_util.h" -static int nv10_graph_register(struct drm_device *); -static void nv10_graph_isr(struct drm_device *); - -#define NV10_FIFO_NUMBER 32 +struct nv10_graph_engine { + struct nouveau_exec_engine base; +}; struct pipe_state { uint32_t pipe_0x0000[0x040/4]; @@ -414,9 +413,9 @@ struct graph_state { static void nv10_graph_save_pipe(struct nouveau_channel *chan) { - struct drm_device *dev = chan->dev; - struct graph_state *pgraph_ctx = chan->pgraph_ctx; + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; struct pipe_state *pipe = &pgraph_ctx->pipe_state; + struct drm_device *dev = chan->dev; PIPE_SAVE(dev, pipe->pipe_0x4400, 0x4400); PIPE_SAVE(dev, pipe->pipe_0x0200, 0x0200); @@ -432,9 +431,9 @@ static void nv10_graph_save_pipe(struct nouveau_channel *chan) static void nv10_graph_load_pipe(struct nouveau_channel *chan) { - struct drm_device *dev = chan->dev; - struct graph_state *pgraph_ctx = chan->pgraph_ctx; + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; struct pipe_state *pipe = &pgraph_ctx->pipe_state; + struct drm_device *dev = chan->dev; uint32_t xfmode0, xfmode1; int i; @@ -482,9 +481,9 @@ static void nv10_graph_load_pipe(struct nouveau_channel *chan) static void nv10_graph_create_pipe(struct nouveau_channel *chan) { - struct drm_device *dev = chan->dev; - struct graph_state *pgraph_ctx = chan->pgraph_ctx; + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; struct pipe_state *fifo_pipe_state = &pgraph_ctx->pipe_state; + struct drm_device *dev = chan->dev; uint32_t *fifo_pipe_state_addr; int i; #define PIPE_INIT(addr) \ @@ -661,8 +660,6 @@ static void nv10_graph_load_dma_vtxbuf(struct nouveau_channel *chan, uint32_t inst) { struct drm_device *dev = chan->dev; - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; uint32_t st2, st2_dl, st2_dh, fifo_ptr, fifo[0x60/4]; uint32_t ctx_user, ctx_switch[5]; int i, subchan = -1; @@ -711,8 +708,8 @@ static void nv10_graph_load_dma_vtxbuf(struct nouveau_channel *chan, 0x2c000000 | chan->id << 20 | subchan << 16 | 0x18c); nv_wr32(dev, NV10_PGRAPH_FFINTFC_ST2_DL, inst); nv_mask(dev, NV10_PGRAPH_CTX_CONTROL, 0, 0x10000); - pgraph->fifo_access(dev, true); - pgraph->fifo_access(dev, false); + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000001); + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000000); /* Restore the FIFO state */ for (i = 0; i < ARRAY_SIZE(fifo); i++) @@ -729,11 +726,12 @@ static void nv10_graph_load_dma_vtxbuf(struct nouveau_channel *chan, nv_wr32(dev, NV10_PGRAPH_CTX_USER, ctx_user); } -int nv10_graph_load_context(struct nouveau_channel *chan) +static int +nv10_graph_load_context(struct nouveau_channel *chan) { struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; - struct graph_state *pgraph_ctx = chan->pgraph_ctx; + struct graph_state *pgraph_ctx = chan->engctx[NVOBJ_ENGINE_GR]; uint32_t tmp; int i; @@ -757,21 +755,20 @@ int nv10_graph_load_context(struct nouveau_channel *chan) return 0; } -int +static int nv10_graph_unload_context(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; struct nouveau_channel *chan; struct graph_state *ctx; uint32_t tmp; int i; - chan = pgraph->channel(dev); + chan = nv10_graph_channel(dev); if (!chan) return 0; - ctx = chan->pgraph_ctx; + ctx = chan->engctx[NVOBJ_ENGINE_GR]; for (i = 0; i < ARRAY_SIZE(nv10_graph_ctx_regs); i++) ctx->nv10[i] = nv_rd32(dev, nv10_graph_ctx_regs[i]); @@ -805,7 +802,7 @@ nv10_graph_context_switch(struct drm_device *dev) /* Load context for next channel */ chid = (nv_rd32(dev, NV04_PGRAPH_TRAPPED_ADDR) >> 20) & 0x1f; chan = dev_priv->channels.ptr[chid]; - if (chan && chan->pgraph_ctx) + if (chan && chan->engctx[NVOBJ_ENGINE_GR]) nv10_graph_load_context(chan); } @@ -836,7 +833,8 @@ nv10_graph_channel(struct drm_device *dev) return dev_priv->channels.ptr[chid]; } -int nv10_graph_create_context(struct nouveau_channel *chan) +static int +nv10_graph_context_new(struct nouveau_channel *chan, int engine) { struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; @@ -844,11 +842,10 @@ int nv10_graph_create_context(struct nouveau_channel *chan) NV_DEBUG(dev, "nv10_graph_context_create %d\n", chan->id); - chan->pgraph_ctx = pgraph_ctx = kzalloc(sizeof(*pgraph_ctx), - GFP_KERNEL); + pgraph_ctx = kzalloc(sizeof(*pgraph_ctx), GFP_KERNEL); if (pgraph_ctx == NULL) return -ENOMEM; - + chan->engctx[engine] = pgraph_ctx; NV_WRITE_CTX(0x00400e88, 0x08000000); NV_WRITE_CTX(0x00400e9c, 0x4b7fffff); @@ -873,30 +870,30 @@ int nv10_graph_create_context(struct nouveau_channel *chan) return 0; } -void nv10_graph_destroy_context(struct nouveau_channel *chan) +static void +nv10_graph_context_del(struct nouveau_channel *chan, int engine) { struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct graph_state *pgraph_ctx = chan->pgraph_ctx; + struct graph_state *pgraph_ctx = chan->engctx[engine]; unsigned long flags; spin_lock_irqsave(&dev_priv->context_switch_lock, flags); - pgraph->fifo_access(dev, false); + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000000); /* Unload the context if it's the currently active one */ - if (pgraph->channel(dev) == chan) - pgraph->unload_context(dev); + if (nv10_graph_channel(dev) == chan) + nv10_graph_unload_context(dev); + + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); /* Free the context resources */ + chan->engctx[engine] = NULL; kfree(pgraph_ctx); - chan->pgraph_ctx = NULL; - - pgraph->fifo_access(dev, true); - spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); } -void +static void nv10_graph_set_tile_region(struct drm_device *dev, int i) { struct drm_nouveau_private *dev_priv = dev->dev_private; @@ -907,22 +904,18 @@ nv10_graph_set_tile_region(struct drm_device *dev, int i) nv_wr32(dev, NV10_PGRAPH_TILE(i), tile->addr); } -int nv10_graph_init(struct drm_device *dev) +static int +nv10_graph_init(struct drm_device *dev, int engine) { struct drm_nouveau_private *dev_priv = dev->dev_private; - uint32_t tmp; - int ret, i; + u32 tmp; + int i; nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PGRAPH); nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | NV_PMC_ENABLE_PGRAPH); - ret = nv10_graph_register(dev); - if (ret) - return ret; - - nouveau_irq_register(dev, 12, nv10_graph_isr); nv_wr32(dev, NV03_PGRAPH_INTR , 0xFFFFFFFF); nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0xFFFFFFFF); @@ -963,18 +956,25 @@ int nv10_graph_init(struct drm_device *dev) return 0; } -void nv10_graph_takedown(struct drm_device *dev) +static int +nv10_graph_fini(struct drm_device *dev, int engine, bool suspend) { + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000000); + if (!nv_wait(dev, NV04_PGRAPH_STATUS, ~0, 0) && suspend) { + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000001); + return -EBUSY; + } + nv10_graph_unload_context(dev); nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0x00000000); - nouveau_irq_unregister(dev, 12); + return 0; } static int nv17_graph_mthd_lma_window(struct nouveau_channel *chan, u32 class, u32 mthd, u32 data) { + struct graph_state *ctx = chan->engctx[NVOBJ_ENGINE_GR]; struct drm_device *dev = chan->dev; - struct graph_state *ctx = chan->pgraph_ctx; struct pipe_state *pipe = &ctx->pipe_state; uint32_t pipe_0x0040[1], pipe_0x64c0[8], pipe_0x6a80[3], pipe_0x6ab0[3]; uint32_t xfmode0, xfmode1; @@ -1061,64 +1061,13 @@ nv17_graph_mthd_lma_enable(struct nouveau_channel *chan, return 0; } -static int -nv10_graph_register(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - - if (dev_priv->engine.graph.registered) - return 0; - - NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ - NVOBJ_CLASS(dev, 0x0030, GR); /* null */ - NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ - NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ - NVOBJ_CLASS(dev, 0x005f, GR); /* imageblit */ - NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ - NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ - NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ - NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ - NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ - NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ - NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ - NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ - NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ - NVOBJ_CLASS(dev, 0x0052, GR); /* swzsurf */ - NVOBJ_CLASS(dev, 0x0093, GR); /* surf3d */ - NVOBJ_CLASS(dev, 0x0094, GR); /* tex_tri */ - NVOBJ_CLASS(dev, 0x0095, GR); /* multitex_tri */ - - /* celcius */ - if (dev_priv->chipset <= 0x10) { - NVOBJ_CLASS(dev, 0x0056, GR); - } else - if (dev_priv->chipset < 0x17 || dev_priv->chipset == 0x1a) { - NVOBJ_CLASS(dev, 0x0096, GR); - } else { - NVOBJ_CLASS(dev, 0x0099, GR); - NVOBJ_MTHD (dev, 0x0099, 0x1638, nv17_graph_mthd_lma_window); - NVOBJ_MTHD (dev, 0x0099, 0x163c, nv17_graph_mthd_lma_window); - NVOBJ_MTHD (dev, 0x0099, 0x1640, nv17_graph_mthd_lma_window); - NVOBJ_MTHD (dev, 0x0099, 0x1644, nv17_graph_mthd_lma_window); - NVOBJ_MTHD (dev, 0x0099, 0x1658, nv17_graph_mthd_lma_enable); - } - - /* nvsw */ - NVOBJ_CLASS(dev, 0x506e, SW); - NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); - - dev_priv->engine.graph.registered = true; - return 0; -} - struct nouveau_bitfield nv10_graph_intr[] = { { NV_PGRAPH_INTR_NOTIFY, "NOTIFY" }, { NV_PGRAPH_INTR_ERROR, "ERROR" }, {} }; -struct nouveau_bitfield nv10_graph_nstatus[] = -{ +struct nouveau_bitfield nv10_graph_nstatus[] = { { NV10_PGRAPH_NSTATUS_STATE_IN_USE, "STATE_IN_USE" }, { NV10_PGRAPH_NSTATUS_INVALID_STATE, "INVALID_STATE" }, { NV10_PGRAPH_NSTATUS_BAD_ARGUMENT, "BAD_ARGUMENT" }, @@ -1173,3 +1122,73 @@ nv10_graph_isr(struct drm_device *dev) } } } + +static void +nv10_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv10_graph_engine *pgraph = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 12); + kfree(pgraph); +} + +int +nv10_graph_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv10_graph_engine *pgraph; + + pgraph = kzalloc(sizeof(*pgraph), GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + pgraph->base.destroy = nv10_graph_destroy; + pgraph->base.init = nv10_graph_init; + pgraph->base.fini = nv10_graph_fini; + pgraph->base.context_new = nv10_graph_context_new; + pgraph->base.context_del = nv10_graph_context_del; + pgraph->base.object_new = nv04_graph_object_new; + pgraph->base.set_tile_region = nv10_graph_set_tile_region; + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + nouveau_irq_register(dev, 12, nv10_graph_isr); + + /* nvsw */ + NVOBJ_CLASS(dev, 0x506e, SW); + NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); + + NVOBJ_CLASS(dev, 0x0030, GR); /* null */ + NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ + NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ + NVOBJ_CLASS(dev, 0x005f, GR); /* imageblit */ + NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ + NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ + NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ + NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ + NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ + NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ + NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ + NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ + NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ + NVOBJ_CLASS(dev, 0x0052, GR); /* swzsurf */ + NVOBJ_CLASS(dev, 0x0093, GR); /* surf3d */ + NVOBJ_CLASS(dev, 0x0094, GR); /* tex_tri */ + NVOBJ_CLASS(dev, 0x0095, GR); /* multitex_tri */ + + /* celcius */ + if (dev_priv->chipset <= 0x10) { + NVOBJ_CLASS(dev, 0x0056, GR); + } else + if (dev_priv->chipset < 0x17 || dev_priv->chipset == 0x1a) { + NVOBJ_CLASS(dev, 0x0096, GR); + } else { + NVOBJ_CLASS(dev, 0x0099, GR); + NVOBJ_MTHD (dev, 0x0099, 0x1638, nv17_graph_mthd_lma_window); + NVOBJ_MTHD (dev, 0x0099, 0x163c, nv17_graph_mthd_lma_window); + NVOBJ_MTHD (dev, 0x0099, 0x1640, nv17_graph_mthd_lma_window); + NVOBJ_MTHD (dev, 0x0099, 0x1644, nv17_graph_mthd_lma_window); + NVOBJ_MTHD (dev, 0x0099, 0x1658, nv17_graph_mthd_lma_enable); + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv20_graph.c b/drivers/gpu/drm/nouveau/nv20_graph.c index 8464b76798d..183e37512ef 100644 --- a/drivers/gpu/drm/nouveau/nv20_graph.c +++ b/drivers/gpu/drm/nouveau/nv20_graph.c @@ -24,6 +24,14 @@ * */ +struct nv20_graph_engine { + struct nouveau_exec_engine base; + struct nouveau_gpuobj *ctxtab; + void (*grctx_init)(struct nouveau_gpuobj *); + u32 grctx_size; + u32 grctx_user; +}; + #define NV20_GRCTX_SIZE (3580*4) #define NV25_GRCTX_SIZE (3529*4) #define NV2A_GRCTX_SIZE (3500*4) @@ -32,12 +40,54 @@ #define NV34_GRCTX_SIZE (18140) #define NV35_36_GRCTX_SIZE (22396) -static int nv20_graph_register(struct drm_device *); -static int nv30_graph_register(struct drm_device *); -static void nv20_graph_isr(struct drm_device *); +int +nv20_graph_unload_context(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_channel *chan; + struct nouveau_gpuobj *grctx; + u32 tmp; + + chan = nv10_graph_channel(dev); + if (!chan) + return 0; + grctx = chan->engctx[NVOBJ_ENGINE_GR]; + + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, grctx->pinst >> 4); + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_XFER, + NV20_PGRAPH_CHANNEL_CTX_XFER_SAVE); + + nouveau_wait_for_idle(dev); + + nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10000000); + tmp = nv_rd32(dev, NV10_PGRAPH_CTX_USER) & 0x00ffffff; + tmp |= (pfifo->channels - 1) << 24; + nv_wr32(dev, NV10_PGRAPH_CTX_USER, tmp); + return 0; +} + +static void +nv20_graph_rdi(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + int i, writecount = 32; + uint32_t rdi_index = 0x2c80000; + + if (dev_priv->chipset == 0x20) { + rdi_index = 0x3d0000; + writecount = 15; + } + + nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, rdi_index); + for (i = 0; i < writecount; i++) + nv_wr32(dev, NV10_PGRAPH_RDI_DATA, 0); + + nouveau_wait_for_idle(dev); +} static void -nv20_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) +nv20_graph_context_init(struct nouveau_gpuobj *ctx) { int i; @@ -87,7 +137,7 @@ nv20_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) } static void -nv25_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) +nv25_graph_context_init(struct nouveau_gpuobj *ctx) { int i; @@ -146,7 +196,7 @@ nv25_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) } static void -nv2a_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) +nv2a_graph_context_init(struct nouveau_gpuobj *ctx) { int i; @@ -196,7 +246,7 @@ nv2a_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) } static void -nv30_31_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) +nv30_31_graph_context_init(struct nouveau_gpuobj *ctx) { int i; @@ -254,7 +304,7 @@ nv30_31_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) } static void -nv34_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) +nv34_graph_context_init(struct nouveau_gpuobj *ctx) { int i; @@ -312,7 +362,7 @@ nv34_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) } static void -nv35_36_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) +nv35_36_graph_context_init(struct nouveau_gpuobj *ctx) { int i; @@ -370,148 +420,57 @@ nv35_36_graph_context_init(struct drm_device *dev, struct nouveau_gpuobj *ctx) } int -nv20_graph_create_context(struct nouveau_channel *chan) +nv20_graph_context_new(struct nouveau_channel *chan, int engine) { + struct nv20_graph_engine *pgraph = nv_engine(chan->dev, engine); + struct nouveau_gpuobj *grctx = NULL; struct drm_device *dev = chan->dev; - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - void (*ctx_init)(struct drm_device *, struct nouveau_gpuobj *); - unsigned int idoffs = 0x28; int ret; - switch (dev_priv->chipset) { - case 0x20: - ctx_init = nv20_graph_context_init; - idoffs = 0; - break; - case 0x25: - case 0x28: - ctx_init = nv25_graph_context_init; - break; - case 0x2a: - ctx_init = nv2a_graph_context_init; - idoffs = 0; - break; - case 0x30: - case 0x31: - ctx_init = nv30_31_graph_context_init; - break; - case 0x34: - ctx_init = nv34_graph_context_init; - break; - case 0x35: - case 0x36: - ctx_init = nv35_36_graph_context_init; - break; - default: - BUG_ON(1); - } - - ret = nouveau_gpuobj_new(dev, chan, pgraph->grctx_size, 16, - NVOBJ_FLAG_ZERO_ALLOC, &chan->ramin_grctx); + ret = nouveau_gpuobj_new(dev, NULL, pgraph->grctx_size, 16, + NVOBJ_FLAG_ZERO_ALLOC, &grctx); if (ret) return ret; /* Initialise default context values */ - ctx_init(dev, chan->ramin_grctx); + pgraph->grctx_init(grctx); /* nv20: nv_wo32(dev, chan->ramin_grctx->gpuobj, 10, chan->id<<24); */ - nv_wo32(chan->ramin_grctx, idoffs, - (chan->id << 24) | 0x1); /* CTX_USER */ + /* CTX_USER */ + nv_wo32(grctx, pgraph->grctx_user, (chan->id << 24) | 0x1); - nv_wo32(pgraph->ctx_table, chan->id * 4, chan->ramin_grctx->pinst >> 4); + nv_wo32(pgraph->ctxtab, chan->id * 4, grctx->pinst >> 4); + chan->engctx[engine] = grctx; return 0; } void -nv20_graph_destroy_context(struct nouveau_channel *chan) +nv20_graph_context_del(struct nouveau_channel *chan, int engine) { + struct nv20_graph_engine *pgraph = nv_engine(chan->dev, engine); + struct nouveau_gpuobj *grctx = chan->engctx[engine]; struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; unsigned long flags; spin_lock_irqsave(&dev_priv->context_switch_lock, flags); - pgraph->fifo_access(dev, false); + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000000); /* Unload the context if it's the currently active one */ - if (pgraph->channel(dev) == chan) - pgraph->unload_context(dev); + if (nv10_graph_channel(dev) == chan) + nv20_graph_unload_context(dev); - pgraph->fifo_access(dev, true); + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000001); spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); /* Free the context resources */ - nv_wo32(pgraph->ctx_table, chan->id * 4, 0); - nouveau_gpuobj_ref(NULL, &chan->ramin_grctx); -} - -int -nv20_graph_load_context(struct nouveau_channel *chan) -{ - struct drm_device *dev = chan->dev; - uint32_t inst; - - if (!chan->ramin_grctx) - return -EINVAL; - inst = chan->ramin_grctx->pinst >> 4; - - nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, inst); - nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_XFER, - NV20_PGRAPH_CHANNEL_CTX_XFER_LOAD); - nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10010100); - - nouveau_wait_for_idle(dev); - return 0; -} + nv_wo32(pgraph->ctxtab, chan->id * 4, 0); -int -nv20_graph_unload_context(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; - struct nouveau_channel *chan; - uint32_t inst, tmp; - - chan = pgraph->channel(dev); - if (!chan) - return 0; - inst = chan->ramin_grctx->pinst >> 4; - - nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, inst); - nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_XFER, - NV20_PGRAPH_CHANNEL_CTX_XFER_SAVE); - - nouveau_wait_for_idle(dev); - - nv_wr32(dev, NV10_PGRAPH_CTX_CONTROL, 0x10000000); - tmp = nv_rd32(dev, NV10_PGRAPH_CTX_USER) & 0x00ffffff; - tmp |= (pfifo->channels - 1) << 24; - nv_wr32(dev, NV10_PGRAPH_CTX_USER, tmp); - return 0; + nouveau_gpuobj_ref(NULL, &grctx); + chan->engctx[engine] = NULL; } static void -nv20_graph_rdi(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - int i, writecount = 32; - uint32_t rdi_index = 0x2c80000; - - if (dev_priv->chipset == 0x20) { - rdi_index = 0x3d0000; - writecount = 15; - } - - nv_wr32(dev, NV10_PGRAPH_RDI_INDEX, rdi_index); - for (i = 0; i < writecount; i++) - nv_wr32(dev, NV10_PGRAPH_RDI_DATA, 0); - - nouveau_wait_for_idle(dev); -} - -void nv20_graph_set_tile_region(struct drm_device *dev, int i) { struct drm_nouveau_private *dev_priv = dev->dev_private; @@ -536,56 +495,22 @@ nv20_graph_set_tile_region(struct drm_device *dev, int i) } int -nv20_graph_init(struct drm_device *dev) +nv20_graph_init(struct drm_device *dev, int engine) { + struct nv20_graph_engine *pgraph = nv_engine(dev, engine); struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; uint32_t tmp, vramsz; - int ret, i; - - switch (dev_priv->chipset) { - case 0x20: - pgraph->grctx_size = NV20_GRCTX_SIZE; - break; - case 0x25: - case 0x28: - pgraph->grctx_size = NV25_GRCTX_SIZE; - break; - case 0x2a: - pgraph->grctx_size = NV2A_GRCTX_SIZE; - break; - default: - NV_ERROR(dev, "unknown chipset, disabling acceleration\n"); - pgraph->accel_blocked = true; - return 0; - } + int i; nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PGRAPH); nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | NV_PMC_ENABLE_PGRAPH); - if (!pgraph->ctx_table) { - /* Create Context Pointer Table */ - ret = nouveau_gpuobj_new(dev, NULL, 32 * 4, 16, - NVOBJ_FLAG_ZERO_ALLOC, - &pgraph->ctx_table); - if (ret) - return ret; - } - - nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_TABLE, - pgraph->ctx_table->pinst >> 4); + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_TABLE, pgraph->ctxtab->pinst >> 4); nv20_graph_rdi(dev); - ret = nv20_graph_register(dev); - if (ret) { - nouveau_gpuobj_ref(NULL, &pgraph->ctx_table); - return ret; - } - - nouveau_irq_register(dev, 12, nv20_graph_isr); nv_wr32(dev, NV03_PGRAPH_INTR , 0xFFFFFFFF); nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0xFFFFFFFF); @@ -657,67 +582,20 @@ nv20_graph_init(struct drm_device *dev) return 0; } -void -nv20_graph_takedown(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - - nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0x00000000); - nouveau_irq_unregister(dev, 12); - - nouveau_gpuobj_ref(NULL, &pgraph->ctx_table); -} - int -nv30_graph_init(struct drm_device *dev) +nv30_graph_init(struct drm_device *dev, int engine) { + struct nv20_graph_engine *pgraph = nv_engine(dev, engine); struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - int ret, i; - - switch (dev_priv->chipset) { - case 0x30: - case 0x31: - pgraph->grctx_size = NV30_31_GRCTX_SIZE; - break; - case 0x34: - pgraph->grctx_size = NV34_GRCTX_SIZE; - break; - case 0x35: - case 0x36: - pgraph->grctx_size = NV35_36_GRCTX_SIZE; - break; - default: - NV_ERROR(dev, "unknown chipset, disabling acceleration\n"); - pgraph->accel_blocked = true; - return 0; - } + int i; nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PGRAPH); nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) | NV_PMC_ENABLE_PGRAPH); - if (!pgraph->ctx_table) { - /* Create Context Pointer Table */ - ret = nouveau_gpuobj_new(dev, NULL, 32 * 4, 16, - NVOBJ_FLAG_ZERO_ALLOC, - &pgraph->ctx_table); - if (ret) - return ret; - } - - ret = nv30_graph_register(dev); - if (ret) { - nouveau_gpuobj_ref(NULL, &pgraph->ctx_table); - return ret; - } - - nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_TABLE, - pgraph->ctx_table->pinst >> 4); + nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_TABLE, pgraph->ctxtab->pinst >> 4); - nouveau_irq_register(dev, 12, nv20_graph_isr); nv_wr32(dev, NV03_PGRAPH_INTR , 0xFFFFFFFF); nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0xFFFFFFFF); @@ -775,85 +653,16 @@ nv30_graph_init(struct drm_device *dev) return 0; } -static int -nv20_graph_register(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - - if (dev_priv->engine.graph.registered) - return 0; - - NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ - NVOBJ_CLASS(dev, 0x0030, GR); /* null */ - NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ - NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ - NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ - NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ - NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ - NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ - NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ - NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ - NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ - NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ - NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ - NVOBJ_CLASS(dev, 0x009e, GR); /* swzsurf */ - NVOBJ_CLASS(dev, 0x0096, GR); /* celcius */ - - /* kelvin */ - if (dev_priv->chipset < 0x25) - NVOBJ_CLASS(dev, 0x0097, GR); - else - NVOBJ_CLASS(dev, 0x0597, GR); - - /* nvsw */ - NVOBJ_CLASS(dev, 0x506e, SW); - NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); - - dev_priv->engine.graph.registered = true; - return 0; -} - -static int -nv30_graph_register(struct drm_device *dev) +int +nv20_graph_fini(struct drm_device *dev, int engine, bool suspend) { - struct drm_nouveau_private *dev_priv = dev->dev_private; - - if (dev_priv->engine.graph.registered) - return 0; - - NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ - NVOBJ_CLASS(dev, 0x0030, GR); /* null */ - NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ - NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ - NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ - NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ - NVOBJ_CLASS(dev, 0x038a, GR); /* ifc (nv30) */ - NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ - NVOBJ_CLASS(dev, 0x0389, GR); /* sifm (nv30) */ - NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ - NVOBJ_CLASS(dev, 0x0362, GR); /* surf2d (nv30) */ - NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ - NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ - NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ - NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ - NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ - NVOBJ_CLASS(dev, 0x039e, GR); /* swzsurf */ - - /* rankine */ - if (0x00000003 & (1 << (dev_priv->chipset & 0x0f))) - NVOBJ_CLASS(dev, 0x0397, GR); - else - if (0x00000010 & (1 << (dev_priv->chipset & 0x0f))) - NVOBJ_CLASS(dev, 0x0697, GR); - else - if (0x000001e0 & (1 << (dev_priv->chipset & 0x0f))) - NVOBJ_CLASS(dev, 0x0497, GR); - - /* nvsw */ - NVOBJ_CLASS(dev, 0x506e, SW); - NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); - - dev_priv->engine.graph.registered = true; + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000000); + if (!nv_wait(dev, NV04_PGRAPH_STATUS, ~0, 0) && suspend) { + nv_mask(dev, NV04_PGRAPH_FIFO, 0x00000001, 0x00000001); + return -EBUSY; + } + nv20_graph_unload_context(dev); + nv_wr32(dev, NV03_PGRAPH_INTR_EN, 0x00000000); return 0; } @@ -897,3 +706,137 @@ nv20_graph_isr(struct drm_device *dev) } } } + +static void +nv20_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv20_graph_engine *pgraph = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 12); + nouveau_gpuobj_ref(NULL, &pgraph->ctxtab); + + NVOBJ_ENGINE_DEL(dev, GR); + kfree(pgraph); +} + +int +nv20_graph_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv20_graph_engine *pgraph; + int ret; + + pgraph = kzalloc(sizeof(*pgraph), GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + pgraph->base.destroy = nv20_graph_destroy; + pgraph->base.fini = nv20_graph_fini; + pgraph->base.context_new = nv20_graph_context_new; + pgraph->base.context_del = nv20_graph_context_del; + pgraph->base.object_new = nv04_graph_object_new; + pgraph->base.set_tile_region = nv20_graph_set_tile_region; + + pgraph->grctx_user = 0x0028; + if (dev_priv->card_type == NV_20) { + pgraph->base.init = nv20_graph_init; + switch (dev_priv->chipset) { + case 0x20: + pgraph->grctx_init = nv20_graph_context_init; + pgraph->grctx_size = NV20_GRCTX_SIZE; + pgraph->grctx_user = 0x0000; + break; + case 0x25: + case 0x28: + pgraph->grctx_init = nv25_graph_context_init; + pgraph->grctx_size = NV25_GRCTX_SIZE; + break; + case 0x2a: + pgraph->grctx_init = nv2a_graph_context_init; + pgraph->grctx_size = NV2A_GRCTX_SIZE; + pgraph->grctx_user = 0x0000; + break; + default: + NV_ERROR(dev, "PGRAPH: unknown chipset\n"); + kfree(pgraph); + return 0; + } + } else { + pgraph->base.init = nv30_graph_init; + switch (dev_priv->chipset) { + case 0x30: + case 0x31: + pgraph->grctx_init = nv30_31_graph_context_init; + pgraph->grctx_size = NV30_31_GRCTX_SIZE; + break; + case 0x34: + pgraph->grctx_init = nv34_graph_context_init; + pgraph->grctx_size = NV34_GRCTX_SIZE; + break; + case 0x35: + case 0x36: + pgraph->grctx_init = nv35_36_graph_context_init; + pgraph->grctx_size = NV35_36_GRCTX_SIZE; + break; + default: + NV_ERROR(dev, "PGRAPH: unknown chipset\n"); + kfree(pgraph); + return 0; + } + } + + /* Create Context Pointer Table */ + ret = nouveau_gpuobj_new(dev, NULL, 32 * 4, 16, NVOBJ_FLAG_ZERO_ALLOC, + &pgraph->ctxtab); + if (ret) { + kfree(pgraph); + return ret; + } + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + nouveau_irq_register(dev, 12, nv20_graph_isr); + + /* nvsw */ + NVOBJ_CLASS(dev, 0x506e, SW); + NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); + + NVOBJ_CLASS(dev, 0x0030, GR); /* null */ + NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ + NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ + NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ + NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ + NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ + NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ + NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ + NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ + NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ + NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ + NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ + if (dev_priv->card_type == NV_20) { + NVOBJ_CLASS(dev, 0x009e, GR); /* swzsurf */ + NVOBJ_CLASS(dev, 0x0096, GR); /* celcius */ + + /* kelvin */ + if (dev_priv->chipset < 0x25) + NVOBJ_CLASS(dev, 0x0097, GR); + else + NVOBJ_CLASS(dev, 0x0597, GR); + } else { + NVOBJ_CLASS(dev, 0x038a, GR); /* ifc (nv30) */ + NVOBJ_CLASS(dev, 0x0389, GR); /* sifm (nv30) */ + NVOBJ_CLASS(dev, 0x0362, GR); /* surf2d (nv30) */ + NVOBJ_CLASS(dev, 0x039e, GR); /* swzsurf */ + + /* rankine */ + if (0x00000003 & (1 << (dev_priv->chipset & 0x0f))) + NVOBJ_CLASS(dev, 0x0397, GR); + else + if (0x00000010 & (1 << (dev_priv->chipset & 0x0f))) + NVOBJ_CLASS(dev, 0x0697, GR); + else + if (0x000001e0 & (1 << (dev_priv->chipset & 0x0f))) + NVOBJ_CLASS(dev, 0x0497, GR); + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv40_fifo.c b/drivers/gpu/drm/nouveau/nv40_fifo.c index 49b9a35a9cd..68cb2d991c8 100644 --- a/drivers/gpu/drm/nouveau/nv40_fifo.c +++ b/drivers/gpu/drm/nouveau/nv40_fifo.c @@ -115,6 +115,7 @@ nv40_fifo_do_load_context(struct drm_device *dev, int chid) nv_wr32(dev, 0x32e8, nv_ri32(dev, fc + 68)); nv_wr32(dev, 0x2088, nv_ri32(dev, fc + 76)); nv_wr32(dev, 0x3300, nv_ri32(dev, fc + 80)); + nv_wr32(dev, 0x330c, nv_ri32(dev, fc + 84)); nv_wr32(dev, NV03_PFIFO_CACHE1_GET, 0); nv_wr32(dev, NV03_PFIFO_CACHE1_PUT, 0); @@ -186,6 +187,7 @@ nv40_fifo_unload_context(struct drm_device *dev) tmp |= (nv_rd32(dev, NV04_PFIFO_CACHE1_PUT) << 16); nv_wi32(dev, fc + 72, tmp); #endif + nv_wi32(dev, fc + 84, nv_rd32(dev, 0x330c)); nv40_fifo_do_load_context(dev, pfifo->channels - 1); nv_wr32(dev, NV03_PFIFO_CACHE1_PUSH1, diff --git a/drivers/gpu/drm/nouveau/nv40_graph.c b/drivers/gpu/drm/nouveau/nv40_graph.c index fceb44c0ec7..ba14a93d8af 100644 --- a/drivers/gpu/drm/nouveau/nv40_graph.c +++ b/drivers/gpu/drm/nouveau/nv40_graph.c @@ -28,183 +28,105 @@ #include "drm.h" #include "nouveau_drv.h" #include "nouveau_grctx.h" +#include "nouveau_ramht.h" -static int nv40_graph_register(struct drm_device *); -static void nv40_graph_isr(struct drm_device *); +struct nv40_graph_engine { + struct nouveau_exec_engine base; + u32 grctx_size; +}; -struct nouveau_channel * -nv40_graph_channel(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - uint32_t inst; - int i; - - inst = nv_rd32(dev, NV40_PGRAPH_CTXCTL_CUR); - if (!(inst & NV40_PGRAPH_CTXCTL_CUR_LOADED)) - return NULL; - inst = (inst & NV40_PGRAPH_CTXCTL_CUR_INSTANCE) << 4; - - for (i = 0; i < dev_priv->engine.fifo.channels; i++) { - struct nouveau_channel *chan = dev_priv->channels.ptr[i]; - - if (chan && chan->ramin_grctx && - chan->ramin_grctx->pinst == inst) - return chan; - } - - return NULL; -} - -int -nv40_graph_create_context(struct nouveau_channel *chan) +static int +nv40_graph_context_new(struct nouveau_channel *chan, int engine) { + struct nv40_graph_engine *pgraph = nv_engine(chan->dev, engine); struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; + struct nouveau_gpuobj *grctx = NULL; struct nouveau_grctx ctx = {}; unsigned long flags; int ret; - ret = nouveau_gpuobj_new(dev, chan, pgraph->grctx_size, 16, - NVOBJ_FLAG_ZERO_ALLOC, &chan->ramin_grctx); + ret = nouveau_gpuobj_new(dev, NULL, pgraph->grctx_size, 16, + NVOBJ_FLAG_ZERO_ALLOC, &grctx); if (ret) return ret; /* Initialise default context values */ ctx.dev = chan->dev; ctx.mode = NOUVEAU_GRCTX_VALS; - ctx.data = chan->ramin_grctx; + ctx.data = grctx; nv40_grctx_init(&ctx); - nv_wo32(chan->ramin_grctx, 0, chan->ramin_grctx->pinst); + nv_wo32(grctx, 0, grctx->vinst); /* init grctx pointer in ramfc, and on PFIFO if channel is * already active there */ spin_lock_irqsave(&dev_priv->context_switch_lock, flags); - nv_wo32(chan->ramfc, 0x38, chan->ramin_grctx->pinst >> 4); + nv_wo32(chan->ramfc, 0x38, grctx->vinst >> 4); nv_mask(dev, 0x002500, 0x00000001, 0x00000000); if ((nv_rd32(dev, 0x003204) & 0x0000001f) == chan->id) - nv_wr32(dev, 0x0032e0, chan->ramin_grctx->pinst >> 4); + nv_wr32(dev, 0x0032e0, grctx->vinst >> 4); nv_mask(dev, 0x002500, 0x00000001, 0x00000001); spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + chan->engctx[engine] = grctx; return 0; } -void -nv40_graph_destroy_context(struct nouveau_channel *chan) +static void +nv40_graph_context_del(struct nouveau_channel *chan, int engine) { + struct nouveau_gpuobj *grctx = chan->engctx[engine]; struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; + u32 inst = 0x01000000 | (grctx->pinst >> 4); unsigned long flags; spin_lock_irqsave(&dev_priv->context_switch_lock, flags); - pgraph->fifo_access(dev, false); - - /* Unload the context if it's the currently active one */ - if (pgraph->channel(dev) == chan) - pgraph->unload_context(dev); - - pgraph->fifo_access(dev, true); + nv_mask(dev, 0x400720, 0x00000000, 0x00000001); + if (nv_rd32(dev, 0x40032c) == inst) + nv_mask(dev, 0x40032c, 0x01000000, 0x00000000); + if (nv_rd32(dev, 0x400330) == inst) + nv_mask(dev, 0x400330, 0x01000000, 0x00000000); + nv_mask(dev, 0x400720, 0x00000001, 0x00000001); spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); /* Free the context resources */ - nouveau_gpuobj_ref(NULL, &chan->ramin_grctx); -} - -static int -nv40_graph_transfer_context(struct drm_device *dev, uint32_t inst, int save) -{ - uint32_t old_cp, tv = 1000, tmp; - int i; - - old_cp = nv_rd32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER); - nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, inst); - - tmp = nv_rd32(dev, NV40_PGRAPH_CTXCTL_0310); - tmp |= save ? NV40_PGRAPH_CTXCTL_0310_XFER_SAVE : - NV40_PGRAPH_CTXCTL_0310_XFER_LOAD; - nv_wr32(dev, NV40_PGRAPH_CTXCTL_0310, tmp); - - tmp = nv_rd32(dev, NV40_PGRAPH_CTXCTL_0304); - tmp |= NV40_PGRAPH_CTXCTL_0304_XFER_CTX; - nv_wr32(dev, NV40_PGRAPH_CTXCTL_0304, tmp); - - nouveau_wait_for_idle(dev); - - for (i = 0; i < tv; i++) { - if (nv_rd32(dev, NV40_PGRAPH_CTXCTL_030C) == 0) - break; - } - - nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, old_cp); - - if (i == tv) { - uint32_t ucstat = nv_rd32(dev, NV40_PGRAPH_CTXCTL_UCODE_STAT); - NV_ERROR(dev, "Failed: Instance=0x%08x Save=%d\n", inst, save); - NV_ERROR(dev, "IP: 0x%02x, Opcode: 0x%08x\n", - ucstat >> NV40_PGRAPH_CTXCTL_UCODE_STAT_IP_SHIFT, - ucstat & NV40_PGRAPH_CTXCTL_UCODE_STAT_OP_MASK); - NV_ERROR(dev, "0x40030C = 0x%08x\n", - nv_rd32(dev, NV40_PGRAPH_CTXCTL_030C)); - return -EBUSY; - } - - return 0; + nouveau_gpuobj_ref(NULL, &grctx); + chan->engctx[engine] = NULL; } -/* Restore the context for a specific channel into PGRAPH */ int -nv40_graph_load_context(struct nouveau_channel *chan) +nv40_graph_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) { struct drm_device *dev = chan->dev; - uint32_t inst; + struct nouveau_gpuobj *obj = NULL; int ret; - if (!chan->ramin_grctx) - return -EINVAL; - inst = chan->ramin_grctx->pinst >> 4; - - ret = nv40_graph_transfer_context(dev, inst, 0); + ret = nouveau_gpuobj_new(dev, chan, 20, 16, NVOBJ_FLAG_ZERO_FREE, &obj); if (ret) return ret; - - /* 0x40032C, no idea of it's exact function. Could simply be a - * record of the currently active PGRAPH context. It's currently - * unknown as to what bit 24 does. The nv ddx has it set, so we will - * set it here too. - */ - nv_wr32(dev, NV20_PGRAPH_CHANNEL_CTX_POINTER, inst); - nv_wr32(dev, NV40_PGRAPH_CTXCTL_CUR, - (inst & NV40_PGRAPH_CTXCTL_CUR_INSTANCE) | - NV40_PGRAPH_CTXCTL_CUR_LOADED); - /* 0x32E0 records the instance address of the active FIFO's PGRAPH - * context. If at any time this doesn't match 0x40032C, you will - * receive PGRAPH_INTR_CONTEXT_SWITCH - */ - nv_wr32(dev, NV40_PFIFO_GRCTX_INSTANCE, inst); - return 0; -} - -int -nv40_graph_unload_context(struct drm_device *dev) -{ - uint32_t inst; - int ret; - - inst = nv_rd32(dev, NV40_PGRAPH_CTXCTL_CUR); - if (!(inst & NV40_PGRAPH_CTXCTL_CUR_LOADED)) - return 0; - inst &= NV40_PGRAPH_CTXCTL_CUR_INSTANCE; - - ret = nv40_graph_transfer_context(dev, inst, 1); - - nv_wr32(dev, NV40_PGRAPH_CTXCTL_CUR, inst); + obj->engine = 1; + obj->class = class; + + nv_wo32(obj, 0x00, class); + nv_wo32(obj, 0x04, 0x00000000); +#ifndef __BIG_ENDIAN + nv_wo32(obj, 0x08, 0x00000000); +#else + nv_wo32(obj, 0x08, 0x01000000); +#endif + nv_wo32(obj, 0x0c, 0x00000000); + nv_wo32(obj, 0x10, 0x00000000); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); return ret; } -void +static void nv40_graph_set_tile_region(struct drm_device *dev, int i) { struct drm_nouveau_private *dev_priv = dev->dev_private; @@ -257,14 +179,14 @@ nv40_graph_set_tile_region(struct drm_device *dev, int i) * C51 0x4e */ int -nv40_graph_init(struct drm_device *dev) +nv40_graph_init(struct drm_device *dev, int engine) { - struct drm_nouveau_private *dev_priv = - (struct drm_nouveau_private *)dev->dev_private; + struct nv40_graph_engine *pgraph = nv_engine(dev, engine); + struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; struct nouveau_grctx ctx = {}; uint32_t vramsz, *cp; - int ret, i, j; + int i, j; nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & ~NV_PMC_ENABLE_PGRAPH); @@ -280,7 +202,7 @@ nv40_graph_init(struct drm_device *dev) ctx.data = cp; ctx.ctxprog_max = 256; nv40_grctx_init(&ctx); - dev_priv->engine.graph.grctx_size = ctx.ctxvals_pos * 4; + pgraph->grctx_size = ctx.ctxvals_pos * 4; nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_INDEX, 0); for (i = 0; i < ctx.ctxprog_len; i++) @@ -288,14 +210,9 @@ nv40_graph_init(struct drm_device *dev) kfree(cp); - ret = nv40_graph_register(dev); - if (ret) - return ret; - /* No context present currently */ nv_wr32(dev, NV40_PGRAPH_CTXCTL_CUR, 0x00000000); - nouveau_irq_register(dev, 12, nv40_graph_isr); nv_wr32(dev, NV03_PGRAPH_INTR , 0xFFFFFFFF); nv_wr32(dev, NV40_PGRAPH_INTR_EN, 0xFFFFFFFF); @@ -428,47 +345,21 @@ nv40_graph_init(struct drm_device *dev) return 0; } -void nv40_graph_takedown(struct drm_device *dev) -{ - nouveau_irq_unregister(dev, 12); -} - static int -nv40_graph_register(struct drm_device *dev) +nv40_graph_fini(struct drm_device *dev, int engine, bool suspend) { - struct drm_nouveau_private *dev_priv = dev->dev_private; - - if (dev_priv->engine.graph.registered) - return 0; - - NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ - NVOBJ_CLASS(dev, 0x0030, GR); /* null */ - NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ - NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ - NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ - NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ - NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ - NVOBJ_CLASS(dev, 0x3089, GR); /* sifm (nv40) */ - NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ - NVOBJ_CLASS(dev, 0x3062, GR); /* surf2d (nv40) */ - NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ - NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ - NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ - NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ - NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ - NVOBJ_CLASS(dev, 0x309e, GR); /* swzsurf */ - - /* curie */ - if (nv44_graph_class(dev)) - NVOBJ_CLASS(dev, 0x4497, GR); - else - NVOBJ_CLASS(dev, 0x4097, GR); - - /* nvsw */ - NVOBJ_CLASS(dev, 0x506e, SW); - NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); - - dev_priv->engine.graph.registered = true; + u32 inst = nv_rd32(dev, 0x40032c); + if (inst & 0x01000000) { + nv_wr32(dev, 0x400720, 0x00000000); + nv_wr32(dev, 0x400784, inst); + nv_mask(dev, 0x400310, 0x00000020, 0x00000020); + nv_mask(dev, 0x400304, 0x00000001, 0x00000001); + if (!nv_wait(dev, 0x400300, 0x00000001, 0x00000000)) { + u32 insn = nv_rd32(dev, 0x400308); + NV_ERROR(dev, "PGRAPH: ctxprog timeout 0x%08x\n", insn); + } + nv_mask(dev, 0x40032c, 0x01000000, 0x00000000); + } return 0; } @@ -476,17 +367,17 @@ static int nv40_graph_isr_chid(struct drm_device *dev, u32 inst) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_channel *chan; + struct nouveau_gpuobj *grctx; unsigned long flags; int i; spin_lock_irqsave(&dev_priv->channels.lock, flags); for (i = 0; i < dev_priv->engine.fifo.channels; i++) { - chan = dev_priv->channels.ptr[i]; - if (!chan || !chan->ramin_grctx) + if (!dev_priv->channels.ptr[i]) continue; + grctx = dev_priv->channels.ptr[i]->engctx[NVOBJ_ENGINE_GR]; - if (inst == chan->ramin_grctx->pinst) + if (grctx && grctx->pinst == inst) break; } spin_unlock_irqrestore(&dev_priv->channels.lock, flags); @@ -537,3 +428,63 @@ nv40_graph_isr(struct drm_device *dev) } } } + +static void +nv40_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv40_graph_engine *pgraph = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 12); + + NVOBJ_ENGINE_DEL(dev, GR); + kfree(pgraph); +} + +int +nv40_graph_create(struct drm_device *dev) +{ + struct nv40_graph_engine *pgraph; + + pgraph = kzalloc(sizeof(*pgraph), GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + pgraph->base.destroy = nv40_graph_destroy; + pgraph->base.init = nv40_graph_init; + pgraph->base.fini = nv40_graph_fini; + pgraph->base.context_new = nv40_graph_context_new; + pgraph->base.context_del = nv40_graph_context_del; + pgraph->base.object_new = nv40_graph_object_new; + pgraph->base.set_tile_region = nv40_graph_set_tile_region; + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + nouveau_irq_register(dev, 12, nv40_graph_isr); + + NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ + NVOBJ_CLASS(dev, 0x0030, GR); /* null */ + NVOBJ_CLASS(dev, 0x0039, GR); /* m2mf */ + NVOBJ_CLASS(dev, 0x004a, GR); /* gdirect */ + NVOBJ_CLASS(dev, 0x009f, GR); /* imageblit (nv12) */ + NVOBJ_CLASS(dev, 0x008a, GR); /* ifc */ + NVOBJ_CLASS(dev, 0x0089, GR); /* sifm */ + NVOBJ_CLASS(dev, 0x3089, GR); /* sifm (nv40) */ + NVOBJ_CLASS(dev, 0x0062, GR); /* surf2d */ + NVOBJ_CLASS(dev, 0x3062, GR); /* surf2d (nv40) */ + NVOBJ_CLASS(dev, 0x0043, GR); /* rop */ + NVOBJ_CLASS(dev, 0x0012, GR); /* beta1 */ + NVOBJ_CLASS(dev, 0x0072, GR); /* beta4 */ + NVOBJ_CLASS(dev, 0x0019, GR); /* cliprect */ + NVOBJ_CLASS(dev, 0x0044, GR); /* pattern */ + NVOBJ_CLASS(dev, 0x309e, GR); /* swzsurf */ + + /* curie */ + if (nv44_graph_class(dev)) + NVOBJ_CLASS(dev, 0x4497, GR); + else + NVOBJ_CLASS(dev, 0x4097, GR); + + /* nvsw */ + NVOBJ_CLASS(dev, 0x506e, SW); + NVOBJ_MTHD (dev, 0x506e, 0x0500, nv04_graph_mthd_page_flip); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv40_mpeg.c b/drivers/gpu/drm/nouveau/nv40_mpeg.c new file mode 100644 index 00000000000..ad03a0e1fc7 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv40_mpeg.c @@ -0,0 +1,311 @@ +/* + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" + +struct nv40_mpeg_engine { + struct nouveau_exec_engine base; +}; + +static int +nv40_mpeg_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ctx = NULL; + unsigned long flags; + int ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + ret = nouveau_gpuobj_new(dev, NULL, 264 * 4, 16, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &ctx); + if (ret) + return ret; + + nv_wo32(ctx, 0x78, 0x02001ec1); + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv_mask(dev, 0x002500, 0x00000001, 0x00000000); + if ((nv_rd32(dev, 0x003204) & 0x1f) == chan->id) + nv_wr32(dev, 0x00330c, ctx->pinst >> 4); + nv_wo32(chan->ramfc, 0x54, ctx->pinst >> 4); + nv_mask(dev, 0x002500, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + chan->engctx[engine] = ctx; + return 0; +} + +static void +nv40_mpeg_context_del(struct nouveau_channel *chan, int engine) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + unsigned long flags; + u32 inst = 0x80000000 | (ctx->pinst >> 4); + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000000); + if (nv_rd32(dev, 0x00b318) == inst) + nv_mask(dev, 0x00b318, 0x80000000, 0x00000000); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + nouveau_gpuobj_ref(NULL, &ctx); + chan->engctx[engine] = NULL; +} + +static int +nv40_mpeg_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct drm_device *dev = chan->dev; + struct nouveau_gpuobj *obj = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 20, 16, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 2; + obj->class = class; + + nv_wo32(obj, 0x00, class); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static int +nv40_mpeg_init(struct drm_device *dev, int engine) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv40_mpeg_engine *pmpeg = nv_engine(dev, engine); + int i; + + /* VPE init */ + nv_mask(dev, 0x000200, 0x00000002, 0x00000000); + nv_mask(dev, 0x000200, 0x00000002, 0x00000002); + nv_wr32(dev, 0x00b0e0, 0x00000020); /* nvidia: rd 0x01, wr 0x20 */ + nv_wr32(dev, 0x00b0e8, 0x00000020); /* nvidia: rd 0x01, wr 0x20 */ + + for (i = 0; i < dev_priv->engine.fb.num_tiles; i++) + pmpeg->base.set_tile_region(dev, i); + + /* PMPEG init */ + nv_wr32(dev, 0x00b32c, 0x00000000); + nv_wr32(dev, 0x00b314, 0x00000100); + nv_wr32(dev, 0x00b220, 0x00000044); + nv_wr32(dev, 0x00b300, 0x02001ec1); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000001); + + nv_wr32(dev, 0x00b100, 0xffffffff); + nv_wr32(dev, 0x00b140, 0xffffffff); + + if (!nv_wait(dev, 0x00b200, 0x00000001, 0x00000000)) { + NV_ERROR(dev, "PMPEG init: 0x%08x\n", nv_rd32(dev, 0x00b200)); + return -EBUSY; + } + + return 0; +} + +static int +nv40_mpeg_fini(struct drm_device *dev, int engine, bool suspend) +{ + /*XXX: context save? */ + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000000); + nv_wr32(dev, 0x00b140, 0x00000000); + return 0; +} + +static int +nv40_mpeg_mthd_dma(struct nouveau_channel *chan, u32 class, u32 mthd, u32 data) +{ + struct drm_device *dev = chan->dev; + u32 inst = data << 4; + u32 dma0 = nv_ri32(dev, inst + 0); + u32 dma1 = nv_ri32(dev, inst + 4); + u32 dma2 = nv_ri32(dev, inst + 8); + u32 base = (dma2 & 0xfffff000) | (dma0 >> 20); + u32 size = dma1 + 1; + + /* only allow linear DMA objects */ + if (!(dma0 & 0x00002000)) + return -EINVAL; + + if (mthd == 0x0190) { + /* DMA_CMD */ + nv_mask(dev, 0x00b300, 0x00030000, (dma0 & 0x00030000)); + nv_wr32(dev, 0x00b334, base); + nv_wr32(dev, 0x00b324, size); + } else + if (mthd == 0x01a0) { + /* DMA_DATA */ + nv_mask(dev, 0x00b300, 0x000c0000, (dma0 & 0x00030000) << 2); + nv_wr32(dev, 0x00b360, base); + nv_wr32(dev, 0x00b364, size); + } else { + /* DMA_IMAGE, VRAM only */ + if (dma0 & 0x000c0000) + return -EINVAL; + + nv_wr32(dev, 0x00b370, base); + nv_wr32(dev, 0x00b374, size); + } + + return 0; +} + +static int +nv40_mpeg_isr_chid(struct drm_device *dev, u32 inst) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ctx; + unsigned long flags; + int i; + + spin_lock_irqsave(&dev_priv->channels.lock, flags); + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + if (!dev_priv->channels.ptr[i]) + continue; + + ctx = dev_priv->channels.ptr[i]->engctx[NVOBJ_ENGINE_MPEG]; + if (ctx && ctx->pinst == inst) + break; + } + spin_unlock_irqrestore(&dev_priv->channels.lock, flags); + return i; +} + +static void +nv40_vpe_set_tile_region(struct drm_device *dev, int i) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_tile_reg *tile = &dev_priv->tile.reg[i]; + + nv_wr32(dev, 0x00b008 + (i * 0x10), tile->pitch); + nv_wr32(dev, 0x00b004 + (i * 0x10), tile->limit); + nv_wr32(dev, 0x00b000 + (i * 0x10), tile->addr); +} + +static void +nv40_mpeg_isr(struct drm_device *dev) +{ + u32 inst = (nv_rd32(dev, 0x00b318) & 0x000fffff) << 4; + u32 chid = nv40_mpeg_isr_chid(dev, inst); + u32 stat = nv_rd32(dev, 0x00b100); + u32 type = nv_rd32(dev, 0x00b230); + u32 mthd = nv_rd32(dev, 0x00b234); + u32 data = nv_rd32(dev, 0x00b238); + u32 show = stat; + + if (stat & 0x01000000) { + /* happens on initial binding of the object */ + if (type == 0x00000020 && mthd == 0x0000) { + nv_mask(dev, 0x00b308, 0x00000000, 0x00000000); + show &= ~0x01000000; + } + + if (type == 0x00000010) { + if (!nouveau_gpuobj_mthd_call2(dev, chid, 0x3174, mthd, data)) + show &= ~0x01000000; + } + } + + nv_wr32(dev, 0x00b100, stat); + nv_wr32(dev, 0x00b230, 0x00000001); + + if (show && nouveau_ratelimit()) { + NV_INFO(dev, "PMPEG: Ch %d [0x%08x] 0x%08x 0x%08x 0x%08x 0x%08x\n", + chid, inst, stat, type, mthd, data); + } +} + +static void +nv40_vpe_isr(struct drm_device *dev) +{ + if (nv_rd32(dev, 0x00b100)) + nv40_mpeg_isr(dev); + + if (nv_rd32(dev, 0x00b800)) { + u32 stat = nv_rd32(dev, 0x00b800); + NV_INFO(dev, "PMSRCH: 0x%08x\n", stat); + nv_wr32(dev, 0xb800, stat); + } +} + +static void +nv40_mpeg_destroy(struct drm_device *dev, int engine) +{ + struct nv40_mpeg_engine *pmpeg = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 0); + + NVOBJ_ENGINE_DEL(dev, MPEG); + kfree(pmpeg); +} + +int +nv40_mpeg_create(struct drm_device *dev) +{ + struct nv40_mpeg_engine *pmpeg; + + pmpeg = kzalloc(sizeof(*pmpeg), GFP_KERNEL); + if (!pmpeg) + return -ENOMEM; + + pmpeg->base.destroy = nv40_mpeg_destroy; + pmpeg->base.init = nv40_mpeg_init; + pmpeg->base.fini = nv40_mpeg_fini; + pmpeg->base.context_new = nv40_mpeg_context_new; + pmpeg->base.context_del = nv40_mpeg_context_del; + pmpeg->base.object_new = nv40_mpeg_object_new; + + /* ISR vector, PMC_ENABLE bit, and TILE regs are shared between + * all VPE engines, for this driver's purposes the PMPEG engine + * will be treated as the "master" and handle the global VPE + * bits too + */ + pmpeg->base.set_tile_region = nv40_vpe_set_tile_region; + nouveau_irq_register(dev, 0, nv40_vpe_isr); + + NVOBJ_ENGINE_ADD(dev, MPEG, &pmpeg->base); + NVOBJ_CLASS(dev, 0x3174, MPEG); + NVOBJ_MTHD (dev, 0x3174, 0x0190, nv40_mpeg_mthd_dma); + NVOBJ_MTHD (dev, 0x3174, 0x01a0, nv40_mpeg_mthd_dma); + NVOBJ_MTHD (dev, 0x3174, 0x01b0, nv40_mpeg_mthd_dma); + +#if 0 + NVOBJ_ENGINE_ADD(dev, ME, &pme->base); + NVOBJ_CLASS(dev, 0x4075, ME); +#endif + return 0; + +} diff --git a/drivers/gpu/drm/nouveau/nv50_calc.c b/drivers/gpu/drm/nouveau/nv50_calc.c index de81151648f..8cf63a8b30c 100644 --- a/drivers/gpu/drm/nouveau/nv50_calc.c +++ b/drivers/gpu/drm/nouveau/nv50_calc.c @@ -23,7 +23,6 @@ */ #include "drmP.h" -#include "drm_fixed.h" #include "nouveau_drv.h" #include "nouveau_hw.h" @@ -47,45 +46,52 @@ nv50_calc_pll(struct drm_device *dev, struct pll_lims *pll, int clk, } int -nv50_calc_pll2(struct drm_device *dev, struct pll_lims *pll, int clk, - int *N, int *fN, int *M, int *P) +nva3_calc_pll(struct drm_device *dev, struct pll_lims *pll, int clk, + int *pN, int *pfN, int *pM, int *P) { - fixed20_12 fb_div, a, b; - u32 refclk = pll->refclk / 10; - u32 max_vco_freq = pll->vco1.maxfreq / 10; - u32 max_vco_inputfreq = pll->vco1.max_inputfreq / 10; - clk /= 10; + u32 best_err = ~0, err; + int M, lM, hM, N, fN; - *P = max_vco_freq / clk; + *P = pll->vco1.maxfreq / clk; if (*P > pll->max_p) *P = pll->max_p; if (*P < pll->min_p) *P = pll->min_p; - /* *M = floor((refclk + max_vco_inputfreq) / max_vco_inputfreq); */ - a.full = dfixed_const(refclk + max_vco_inputfreq); - b.full = dfixed_const(max_vco_inputfreq); - a.full = dfixed_div(a, b); - a.full = dfixed_floor(a); - *M = dfixed_trunc(a); + lM = (pll->refclk + pll->vco1.max_inputfreq) / pll->vco1.max_inputfreq; + lM = max(lM, (int)pll->vco1.min_m); + hM = (pll->refclk + pll->vco1.min_inputfreq) / pll->vco1.min_inputfreq; + hM = min(hM, (int)pll->vco1.max_m); - /* fb_div = (vco * *M) / refclk; */ - fb_div.full = dfixed_const(clk * *P); - fb_div.full = dfixed_mul(fb_div, a); - a.full = dfixed_const(refclk); - fb_div.full = dfixed_div(fb_div, a); + for (M = lM; M <= hM; M++) { + u32 tmp = clk * *P * M; + N = tmp / pll->refclk; + fN = tmp % pll->refclk; + if (!pfN && fN >= pll->refclk / 2) + N++; - /* *N = floor(fb_div); */ - a.full = dfixed_floor(fb_div); - *N = dfixed_trunc(fb_div); + if (N < pll->vco1.min_n) + continue; + if (N > pll->vco1.max_n) + break; - /* *fN = (fmod(fb_div, 1.0) * 8192) - 4096; */ - b.full = dfixed_const(8192); - a.full = dfixed_mul(a, b); - fb_div.full = dfixed_mul(fb_div, b); - fb_div.full = fb_div.full - a.full; - *fN = dfixed_trunc(fb_div) - 4096; - *fN &= 0xffff; + err = abs(clk - (pll->refclk * N / M / *P)); + if (err < best_err) { + best_err = err; + *pN = N; + *pM = M; + } - return clk; + if (pfN) { + *pfN = (((fN << 13) / pll->refclk) - 4096) & 0xffff; + return clk; + } + } + + if (unlikely(best_err == ~0)) { + NV_ERROR(dev, "unable to find matching pll values\n"); + return -EINVAL; + } + + return pll->refclk * *pN / *pM / *P; } diff --git a/drivers/gpu/drm/nouveau/nv50_crtc.c b/drivers/gpu/drm/nouveau/nv50_crtc.c index a19ccaa025b..46ad59ea218 100644 --- a/drivers/gpu/drm/nouveau/nv50_crtc.c +++ b/drivers/gpu/drm/nouveau/nv50_crtc.c @@ -104,7 +104,7 @@ nv50_crtc_blank(struct nouveau_crtc *nv_crtc, bool blanked) OUT_RING(evo, nv_crtc->lut.depth == 8 ? NV50_EVO_CRTC_CLUT_MODE_OFF : NV50_EVO_CRTC_CLUT_MODE_ON); - OUT_RING(evo, (nv_crtc->lut.nvbo->bo.mem.start << PAGE_SHIFT) >> 8); + OUT_RING(evo, nv_crtc->lut.nvbo->bo.offset >> 8); if (dev_priv->chipset != 0x50) { BEGIN_RING(evo, 0, NV84_EVO_CRTC(index, CLUT_DMA), 1); OUT_RING(evo, NvEvoVRAM); @@ -286,7 +286,7 @@ nv50_crtc_set_clock(struct drm_device *dev, int head, int pclk) nv_wr32(dev, pll.reg + 8, reg2 | (P << 28) | (M2 << 16) | N2); } else if (dev_priv->chipset < NV_C0) { - ret = nv50_calc_pll2(dev, &pll, pclk, &N1, &N2, &M1, &P); + ret = nva3_calc_pll(dev, &pll, pclk, &N1, &N2, &M1, &P); if (ret <= 0) return 0; @@ -298,7 +298,7 @@ nv50_crtc_set_clock(struct drm_device *dev, int head, int pclk) nv_wr32(dev, pll.reg + 4, reg1 | (P << 16) | (M1 << 8) | N1); nv_wr32(dev, pll.reg + 8, N2); } else { - ret = nv50_calc_pll2(dev, &pll, pclk, &N1, &N2, &M1, &P); + ret = nva3_calc_pll(dev, &pll, pclk, &N1, &N2, &M1, &P); if (ret <= 0) return 0; @@ -349,14 +349,14 @@ nv50_crtc_cursor_set(struct drm_crtc *crtc, struct drm_file *file_priv, struct drm_gem_object *gem; int ret = 0, i; - if (width != 64 || height != 64) - return -EINVAL; - if (!buffer_handle) { nv_crtc->cursor.hide(nv_crtc, true); return 0; } + if (width != 64 || height != 64) + return -EINVAL; + gem = drm_gem_object_lookup(dev, file_priv, buffer_handle); if (!gem) return -ENOENT; @@ -372,7 +372,7 @@ nv50_crtc_cursor_set(struct drm_crtc *crtc, struct drm_file *file_priv, nouveau_bo_unmap(cursor); - nv_crtc->cursor.set_offset(nv_crtc, nv_crtc->cursor.nvbo->bo.mem.start << PAGE_SHIFT); + nv_crtc->cursor.set_offset(nv_crtc, nv_crtc->cursor.nvbo->bo.offset); nv_crtc->cursor.show(nv_crtc, true); out: @@ -532,8 +532,7 @@ nv50_crtc_do_mode_set_base(struct drm_crtc *crtc, if (atomic) { drm_fb = passed_fb; fb = nouveau_framebuffer(passed_fb); - } - else { + } else { /* If not atomic, we can go ahead and pin, and unpin the * old fb we were passed. */ @@ -547,7 +546,7 @@ nv50_crtc_do_mode_set_base(struct drm_crtc *crtc, } } - nv_crtc->fb.offset = fb->nvbo->bo.mem.start << PAGE_SHIFT; + nv_crtc->fb.offset = fb->nvbo->bo.offset; nv_crtc->fb.tile_flags = nouveau_bo_tile_layout(fb->nvbo); nv_crtc->fb.cpp = drm_fb->bits_per_pixel / 8; if (!nv_crtc->fb.blanked && dev_priv->chipset != 0x50) { @@ -748,7 +747,7 @@ nv50_crtc_create(struct drm_device *dev, int index) } nv_crtc->lut.depth = 0; - ret = nouveau_bo_new(dev, NULL, 4096, 0x100, TTM_PL_FLAG_VRAM, + ret = nouveau_bo_new(dev, 4096, 0x100, TTM_PL_FLAG_VRAM, 0, 0x0000, &nv_crtc->lut.nvbo); if (!ret) { ret = nouveau_bo_pin(nv_crtc->lut.nvbo, TTM_PL_FLAG_VRAM); @@ -774,7 +773,7 @@ nv50_crtc_create(struct drm_device *dev, int index) drm_crtc_helper_add(&nv_crtc->base, &nv50_crtc_helper_funcs); drm_mode_crtc_set_gamma_size(&nv_crtc->base, 256); - ret = nouveau_bo_new(dev, NULL, 64*64*4, 0x100, TTM_PL_FLAG_VRAM, + ret = nouveau_bo_new(dev, 64*64*4, 0x100, TTM_PL_FLAG_VRAM, 0, 0x0000, &nv_crtc->cursor.nvbo); if (!ret) { ret = nouveau_bo_pin(nv_crtc->cursor.nvbo, TTM_PL_FLAG_VRAM); diff --git a/drivers/gpu/drm/nouveau/nv50_display.c b/drivers/gpu/drm/nouveau/nv50_display.c index 75a376cc342..db1a5f4b711 100644 --- a/drivers/gpu/drm/nouveau/nv50_display.c +++ b/drivers/gpu/drm/nouveau/nv50_display.c @@ -409,14 +409,12 @@ nv50_display_flip_next(struct drm_crtc *crtc, struct drm_framebuffer *fb, struct nouveau_channel *evo = dispc->sync; int ret; - ret = RING_SPACE(evo, 24); + ret = RING_SPACE(evo, chan ? 25 : 27); if (unlikely(ret)) return ret; /* synchronise with the rendering channel, if necessary */ if (likely(chan)) { - u64 offset = dispc->sem.bo->vma.offset + dispc->sem.offset; - ret = RING_SPACE(chan, 10); if (ret) { WIND_RING(evo); @@ -438,6 +436,8 @@ nv50_display_flip_next(struct drm_crtc *crtc, struct drm_framebuffer *fb, else OUT_RING (chan, chan->vram_handle); } else { + u64 offset = chan->dispc_vma[nv_crtc->index].offset; + offset += dispc->sem.offset; BEGIN_NVC0(chan, 2, NvSubM2MF, 0x0010, 4); OUT_RING (chan, upper_32_bits(offset)); OUT_RING (chan, lower_32_bits(offset)); @@ -458,8 +458,19 @@ nv50_display_flip_next(struct drm_crtc *crtc, struct drm_framebuffer *fb, /* queue the flip on the crtc's "display sync" channel */ BEGIN_RING(evo, 0, 0x0100, 1); OUT_RING (evo, 0xfffe0000); - BEGIN_RING(evo, 0, 0x0084, 5); - OUT_RING (evo, chan ? 0x00000100 : 0x00000010); + if (chan) { + BEGIN_RING(evo, 0, 0x0084, 1); + OUT_RING (evo, 0x00000100); + } else { + BEGIN_RING(evo, 0, 0x0084, 1); + OUT_RING (evo, 0x00000010); + /* allows gamma somehow, PDISP will bitch at you if + * you don't wait for vblank before changing this.. + */ + BEGIN_RING(evo, 0, 0x00e0, 1); + OUT_RING (evo, 0x40000000); + } + BEGIN_RING(evo, 0, 0x0088, 4); OUT_RING (evo, dispc->sem.offset); OUT_RING (evo, 0xf00d0000 | dispc->sem.value); OUT_RING (evo, 0x74b1e000); @@ -473,7 +484,7 @@ nv50_display_flip_next(struct drm_crtc *crtc, struct drm_framebuffer *fb, OUT_RING (evo, 0x00000000); OUT_RING (evo, 0x00000000); BEGIN_RING(evo, 0, 0x0800, 5); - OUT_RING (evo, (nv_fb->nvbo->bo.mem.start << PAGE_SHIFT) >> 8); + OUT_RING (evo, nv_fb->nvbo->bo.offset >> 8); OUT_RING (evo, 0); OUT_RING (evo, (fb->height << 16) | fb->width); OUT_RING (evo, nv_fb->r_pitch); @@ -517,13 +528,25 @@ nv50_display_script_select(struct drm_device *dev, struct dcb_entry *dcb, if (bios->fp.if_is_24bit) script |= 0x0200; } else { + /* determine number of lvds links */ + if (nv_connector && nv_connector->edid && + nv_connector->dcb->type == DCB_CONNECTOR_LVDS_SPWG) { + /* http://www.spwg.org */ + if (((u8 *)nv_connector->edid)[121] == 2) + script |= 0x0100; + } else if (pxclk >= bios->fp.duallink_transition_clk) { script |= 0x0100; + } + + /* determine panel depth */ + if (script & 0x0100) { if (bios->fp.strapless_is_24bit & 2) script |= 0x0200; - } else - if (bios->fp.strapless_is_24bit & 1) - script |= 0x0200; + } else { + if (bios->fp.strapless_is_24bit & 1) + script |= 0x0200; + } if (nv_connector && nv_connector->edid && (nv_connector->edid->revision >= 4) && diff --git a/drivers/gpu/drm/nouveau/nv50_evo.c b/drivers/gpu/drm/nouveau/nv50_evo.c index c8e83c1a4de..c99d9751880 100644 --- a/drivers/gpu/drm/nouveau/nv50_evo.c +++ b/drivers/gpu/drm/nouveau/nv50_evo.c @@ -38,6 +38,7 @@ nv50_evo_channel_del(struct nouveau_channel **pevo) return; *pevo = NULL; + nouveau_ramht_ref(NULL, &evo->ramht, evo); nouveau_gpuobj_channel_takedown(evo); nouveau_bo_unmap(evo->pushbuf_bo); nouveau_bo_ref(NULL, &evo->pushbuf_bo); @@ -116,7 +117,7 @@ nv50_evo_channel_new(struct drm_device *dev, int chid, evo->user_get = 4; evo->user_put = 0; - ret = nouveau_bo_new(dev, NULL, 4096, 0, TTM_PL_FLAG_VRAM, 0, 0, + ret = nouveau_bo_new(dev, 4096, 0, TTM_PL_FLAG_VRAM, 0, 0, &evo->pushbuf_bo); if (ret == 0) ret = nouveau_bo_pin(evo->pushbuf_bo, TTM_PL_FLAG_VRAM); @@ -153,7 +154,7 @@ nv50_evo_channel_init(struct nouveau_channel *evo) { struct drm_device *dev = evo->dev; int id = evo->id, ret, i; - u64 pushbuf = evo->pushbuf_bo->bo.mem.start << PAGE_SHIFT; + u64 pushbuf = evo->pushbuf_bo->bo.offset; u32 tmp; tmp = nv_rd32(dev, NV50_PDISPLAY_EVO_CTRL(id)); @@ -331,16 +332,15 @@ nv50_evo_create(struct drm_device *dev) if (ret) goto err; - ret = nouveau_bo_new(dev, NULL, 4096, 0x1000, TTM_PL_FLAG_VRAM, + ret = nouveau_bo_new(dev, 4096, 0x1000, TTM_PL_FLAG_VRAM, 0, 0x0000, &dispc->sem.bo); if (!ret) { - offset = dispc->sem.bo->bo.mem.start << PAGE_SHIFT; - ret = nouveau_bo_pin(dispc->sem.bo, TTM_PL_FLAG_VRAM); if (!ret) ret = nouveau_bo_map(dispc->sem.bo); if (ret) nouveau_bo_ref(NULL, &dispc->sem.bo); + offset = dispc->sem.bo->bo.offset; } if (ret) diff --git a/drivers/gpu/drm/nouveau/nv50_fbcon.c b/drivers/gpu/drm/nouveau/nv50_fbcon.c index 791ded1c5c6..dc75a720652 100644 --- a/drivers/gpu/drm/nouveau/nv50_fbcon.c +++ b/drivers/gpu/drm/nouveau/nv50_fbcon.c @@ -159,7 +159,7 @@ nv50_fbcon_accel_init(struct fb_info *info) struct drm_device *dev = nfbdev->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_channel *chan = dev_priv->channel; - struct nouveau_bo *nvbo = nfbdev->nouveau_fb.nvbo; + struct nouveau_framebuffer *fb = &nfbdev->nouveau_fb; int ret, format; switch (info->var.bits_per_pixel) { @@ -247,8 +247,8 @@ nv50_fbcon_accel_init(struct fb_info *info) OUT_RING(chan, info->fix.line_length); OUT_RING(chan, info->var.xres_virtual); OUT_RING(chan, info->var.yres_virtual); - OUT_RING(chan, upper_32_bits(nvbo->vma.offset)); - OUT_RING(chan, lower_32_bits(nvbo->vma.offset)); + OUT_RING(chan, upper_32_bits(fb->vma.offset)); + OUT_RING(chan, lower_32_bits(fb->vma.offset)); BEGIN_RING(chan, NvSub2D, 0x0230, 2); OUT_RING(chan, format); OUT_RING(chan, 1); @@ -256,8 +256,8 @@ nv50_fbcon_accel_init(struct fb_info *info) OUT_RING(chan, info->fix.line_length); OUT_RING(chan, info->var.xres_virtual); OUT_RING(chan, info->var.yres_virtual); - OUT_RING(chan, upper_32_bits(nvbo->vma.offset)); - OUT_RING(chan, lower_32_bits(nvbo->vma.offset)); + OUT_RING(chan, upper_32_bits(fb->vma.offset)); + OUT_RING(chan, lower_32_bits(fb->vma.offset)); return 0; } diff --git a/drivers/gpu/drm/nouveau/nv50_graph.c b/drivers/gpu/drm/nouveau/nv50_graph.c index b02a5b1e7d3..d43c46caa76 100644 --- a/drivers/gpu/drm/nouveau/nv50_graph.c +++ b/drivers/gpu/drm/nouveau/nv50_graph.c @@ -33,14 +33,97 @@ #include "nouveau_vm.h" #include "nv50_evo.h" -static int nv50_graph_register(struct drm_device *); -static void nv50_graph_isr(struct drm_device *); +struct nv50_graph_engine { + struct nouveau_exec_engine base; + u32 ctxprog[512]; + u32 ctxprog_size; + u32 grctx_size; +}; + +static void +nv50_graph_fifo_access(struct drm_device *dev, bool enabled) +{ + const uint32_t mask = 0x00010001; + + if (enabled) + nv_wr32(dev, 0x400500, nv_rd32(dev, 0x400500) | mask); + else + nv_wr32(dev, 0x400500, nv_rd32(dev, 0x400500) & ~mask); +} + +static struct nouveau_channel * +nv50_graph_channel(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + uint32_t inst; + int i; + + /* Be sure we're not in the middle of a context switch or bad things + * will happen, such as unloading the wrong pgraph context. + */ + if (!nv_wait(dev, 0x400300, 0x00000001, 0x00000000)) + NV_ERROR(dev, "Ctxprog is still running\n"); + + inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR); + if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED)) + return NULL; + inst = (inst & NV50_PGRAPH_CTXCTL_CUR_INSTANCE) << 12; + + for (i = 0; i < dev_priv->engine.fifo.channels; i++) { + struct nouveau_channel *chan = dev_priv->channels.ptr[i]; + + if (chan && chan->ramin && chan->ramin->vinst == inst) + return chan; + } + + return NULL; +} + +static int +nv50_graph_do_load_context(struct drm_device *dev, uint32_t inst) +{ + uint32_t fifo = nv_rd32(dev, 0x400500); + + nv_wr32(dev, 0x400500, fifo & ~1); + nv_wr32(dev, 0x400784, inst); + nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x40); + nv_wr32(dev, 0x400320, nv_rd32(dev, 0x400320) | 0x11); + nv_wr32(dev, 0x400040, 0xffffffff); + (void)nv_rd32(dev, 0x400040); + nv_wr32(dev, 0x400040, 0x00000000); + nv_wr32(dev, 0x400304, nv_rd32(dev, 0x400304) | 1); + + if (nouveau_wait_for_idle(dev)) + nv_wr32(dev, 0x40032c, inst | (1<<31)); + nv_wr32(dev, 0x400500, fifo); + + return 0; +} + +static int +nv50_graph_unload_context(struct drm_device *dev) +{ + uint32_t inst; + + inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR); + if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED)) + return 0; + inst &= NV50_PGRAPH_CTXCTL_CUR_INSTANCE; + + nouveau_wait_for_idle(dev); + nv_wr32(dev, 0x400784, inst); + nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x20); + nv_wr32(dev, 0x400304, nv_rd32(dev, 0x400304) | 0x01); + nouveau_wait_for_idle(dev); + + nv_wr32(dev, NV50_PGRAPH_CTXCTL_CUR, inst); + return 0; +} static void nv50_graph_init_reset(struct drm_device *dev) { uint32_t pmc_e = NV_PMC_ENABLE_PGRAPH | (1 << 21); - NV_DEBUG(dev, "\n"); nv_wr32(dev, NV03_PMC_ENABLE, nv_rd32(dev, NV03_PMC_ENABLE) & ~pmc_e); @@ -52,7 +135,6 @@ nv50_graph_init_intr(struct drm_device *dev) { NV_DEBUG(dev, "\n"); - nouveau_irq_register(dev, 12, nv50_graph_isr); nv_wr32(dev, NV03_PGRAPH_INTR, 0xffffffff); nv_wr32(dev, 0x400138, 0xffffffff); nv_wr32(dev, NV40_PGRAPH_INTR_EN, 0xffffffff); @@ -135,34 +217,14 @@ nv50_graph_init_zcull(struct drm_device *dev) static int nv50_graph_init_ctxctl(struct drm_device *dev) { - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_grctx ctx = {}; - uint32_t *cp; + struct nv50_graph_engine *pgraph = nv_engine(dev, NVOBJ_ENGINE_GR); int i; NV_DEBUG(dev, "\n"); - cp = kmalloc(512 * 4, GFP_KERNEL); - if (!cp) { - NV_ERROR(dev, "failed to allocate ctxprog\n"); - dev_priv->engine.graph.accel_blocked = true; - return 0; - } - - ctx.dev = dev; - ctx.mode = NOUVEAU_GRCTX_PROG; - ctx.data = cp; - ctx.ctxprog_max = 512; - if (!nv50_grctx_init(&ctx)) { - dev_priv->engine.graph.grctx_size = ctx.ctxvals_pos * 4; - - nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_INDEX, 0); - for (i = 0; i < ctx.ctxprog_len; i++) - nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_DATA, cp[i]); - } else { - dev_priv->engine.graph.accel_blocked = true; - } - kfree(cp); + nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_INDEX, 0); + for (i = 0; i < pgraph->ctxprog_size; i++) + nv_wr32(dev, NV40_PGRAPH_CTXCTL_UCODE_DATA, pgraph->ctxprog[i]); nv_wr32(dev, 0x40008c, 0x00000004); /* HW_CTX_SWITCH_ENABLED */ nv_wr32(dev, 0x400320, 4); @@ -171,8 +233,8 @@ nv50_graph_init_ctxctl(struct drm_device *dev) return 0; } -int -nv50_graph_init(struct drm_device *dev) +static int +nv50_graph_init(struct drm_device *dev, int engine) { int ret; @@ -186,105 +248,70 @@ nv50_graph_init(struct drm_device *dev) if (ret) return ret; - ret = nv50_graph_register(dev); - if (ret) - return ret; nv50_graph_init_intr(dev); return 0; } -void -nv50_graph_takedown(struct drm_device *dev) -{ - NV_DEBUG(dev, "\n"); - nv_wr32(dev, 0x40013c, 0x00000000); - nouveau_irq_unregister(dev, 12); -} - -void -nv50_graph_fifo_access(struct drm_device *dev, bool enabled) -{ - const uint32_t mask = 0x00010001; - - if (enabled) - nv_wr32(dev, 0x400500, nv_rd32(dev, 0x400500) | mask); - else - nv_wr32(dev, 0x400500, nv_rd32(dev, 0x400500) & ~mask); -} - -struct nouveau_channel * -nv50_graph_channel(struct drm_device *dev) +static int +nv50_graph_fini(struct drm_device *dev, int engine, bool suspend) { - struct drm_nouveau_private *dev_priv = dev->dev_private; - uint32_t inst; - int i; - - /* Be sure we're not in the middle of a context switch or bad things - * will happen, such as unloading the wrong pgraph context. - */ - if (!nv_wait(dev, 0x400300, 0x00000001, 0x00000000)) - NV_ERROR(dev, "Ctxprog is still running\n"); - - inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR); - if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED)) - return NULL; - inst = (inst & NV50_PGRAPH_CTXCTL_CUR_INSTANCE) << 12; - - for (i = 0; i < dev_priv->engine.fifo.channels; i++) { - struct nouveau_channel *chan = dev_priv->channels.ptr[i]; - - if (chan && chan->ramin && chan->ramin->vinst == inst) - return chan; + nv_mask(dev, 0x400500, 0x00010001, 0x00000000); + if (!nv_wait(dev, 0x400700, ~0, 0) && suspend) { + nv_mask(dev, 0x400500, 0x00010001, 0x00010001); + return -EBUSY; } - - return NULL; + nv50_graph_unload_context(dev); + nv_wr32(dev, 0x40013c, 0x00000000); + return 0; } -int -nv50_graph_create_context(struct nouveau_channel *chan) +static int +nv50_graph_context_new(struct nouveau_channel *chan, int engine) { struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_gpuobj *ramin = chan->ramin; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; + struct nouveau_gpuobj *grctx = NULL; + struct nv50_graph_engine *pgraph = nv_engine(dev, engine); struct nouveau_grctx ctx = {}; int hdr, ret; NV_DEBUG(dev, "ch%d\n", chan->id); - ret = nouveau_gpuobj_new(dev, chan, pgraph->grctx_size, 0, + ret = nouveau_gpuobj_new(dev, NULL, pgraph->grctx_size, 0, NVOBJ_FLAG_ZERO_ALLOC | - NVOBJ_FLAG_ZERO_FREE, &chan->ramin_grctx); + NVOBJ_FLAG_ZERO_FREE, &grctx); if (ret) return ret; hdr = (dev_priv->chipset == 0x50) ? 0x200 : 0x20; nv_wo32(ramin, hdr + 0x00, 0x00190002); - nv_wo32(ramin, hdr + 0x04, chan->ramin_grctx->vinst + - pgraph->grctx_size - 1); - nv_wo32(ramin, hdr + 0x08, chan->ramin_grctx->vinst); + nv_wo32(ramin, hdr + 0x04, grctx->vinst + grctx->size - 1); + nv_wo32(ramin, hdr + 0x08, grctx->vinst); nv_wo32(ramin, hdr + 0x0c, 0); nv_wo32(ramin, hdr + 0x10, 0); nv_wo32(ramin, hdr + 0x14, 0x00010000); ctx.dev = chan->dev; ctx.mode = NOUVEAU_GRCTX_VALS; - ctx.data = chan->ramin_grctx; + ctx.data = grctx; nv50_grctx_init(&ctx); - nv_wo32(chan->ramin_grctx, 0x00000, chan->ramin->vinst >> 12); + nv_wo32(grctx, 0x00000, chan->ramin->vinst >> 12); dev_priv->engine.instmem.flush(dev); - atomic_inc(&chan->vm->pgraph_refs); + + atomic_inc(&chan->vm->engref[NVOBJ_ENGINE_GR]); + chan->engctx[NVOBJ_ENGINE_GR] = grctx; return 0; } -void -nv50_graph_destroy_context(struct nouveau_channel *chan) +static void +nv50_graph_context_del(struct nouveau_channel *chan, int engine) { + struct nouveau_gpuobj *grctx = chan->engctx[engine]; struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; int i, hdr = (dev_priv->chipset == 0x50) ? 0x200 : 0x20; unsigned long flags; @@ -296,72 +323,49 @@ nv50_graph_destroy_context(struct nouveau_channel *chan) spin_lock_irqsave(&dev_priv->context_switch_lock, flags); pfifo->reassign(dev, false); - pgraph->fifo_access(dev, false); + nv50_graph_fifo_access(dev, false); - if (pgraph->channel(dev) == chan) - pgraph->unload_context(dev); + if (nv50_graph_channel(dev) == chan) + nv50_graph_unload_context(dev); for (i = hdr; i < hdr + 24; i += 4) nv_wo32(chan->ramin, i, 0); dev_priv->engine.instmem.flush(dev); - pgraph->fifo_access(dev, true); + nv50_graph_fifo_access(dev, true); pfifo->reassign(dev, true); spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); - nouveau_gpuobj_ref(NULL, &chan->ramin_grctx); + nouveau_gpuobj_ref(NULL, &grctx); - atomic_dec(&chan->vm->pgraph_refs); + atomic_dec(&chan->vm->engref[engine]); + chan->engctx[engine] = NULL; } static int -nv50_graph_do_load_context(struct drm_device *dev, uint32_t inst) -{ - uint32_t fifo = nv_rd32(dev, 0x400500); - - nv_wr32(dev, 0x400500, fifo & ~1); - nv_wr32(dev, 0x400784, inst); - nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x40); - nv_wr32(dev, 0x400320, nv_rd32(dev, 0x400320) | 0x11); - nv_wr32(dev, 0x400040, 0xffffffff); - (void)nv_rd32(dev, 0x400040); - nv_wr32(dev, 0x400040, 0x00000000); - nv_wr32(dev, 0x400304, nv_rd32(dev, 0x400304) | 1); - - if (nouveau_wait_for_idle(dev)) - nv_wr32(dev, 0x40032c, inst | (1<<31)); - nv_wr32(dev, 0x400500, fifo); - - return 0; -} - -int -nv50_graph_load_context(struct nouveau_channel *chan) -{ - uint32_t inst = chan->ramin->vinst >> 12; - - NV_DEBUG(chan->dev, "ch%d\n", chan->id); - return nv50_graph_do_load_context(chan->dev, inst); -} - -int -nv50_graph_unload_context(struct drm_device *dev) +nv50_graph_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) { - uint32_t inst; + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *obj = NULL; + int ret; - inst = nv_rd32(dev, NV50_PGRAPH_CTXCTL_CUR); - if (!(inst & NV50_PGRAPH_CTXCTL_CUR_LOADED)) - return 0; - inst &= NV50_PGRAPH_CTXCTL_CUR_INSTANCE; + ret = nouveau_gpuobj_new(dev, chan, 16, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 1; + obj->class = class; - nouveau_wait_for_idle(dev); - nv_wr32(dev, 0x400784, inst); - nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) | 0x20); - nv_wr32(dev, 0x400304, nv_rd32(dev, 0x400304) | 0x01); - nouveau_wait_for_idle(dev); + nv_wo32(obj, 0x00, class); + nv_wo32(obj, 0x04, 0x00000000); + nv_wo32(obj, 0x08, 0x00000000); + nv_wo32(obj, 0x0c, 0x00000000); + dev_priv->engine.instmem.flush(dev); - nv_wr32(dev, NV50_PGRAPH_CTXCTL_CUR, inst); - return 0; + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; } static void @@ -442,68 +446,15 @@ nv50_graph_nvsw_mthd_page_flip(struct nouveau_channel *chan, return 0; } -static int -nv50_graph_register(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - - if (dev_priv->engine.graph.registered) - return 0; - - NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ - NVOBJ_MTHD (dev, 0x506e, 0x018c, nv50_graph_nvsw_dma_vblsem); - NVOBJ_MTHD (dev, 0x506e, 0x0400, nv50_graph_nvsw_vblsem_offset); - NVOBJ_MTHD (dev, 0x506e, 0x0404, nv50_graph_nvsw_vblsem_release_val); - NVOBJ_MTHD (dev, 0x506e, 0x0408, nv50_graph_nvsw_vblsem_release); - NVOBJ_MTHD (dev, 0x506e, 0x0500, nv50_graph_nvsw_mthd_page_flip); - - NVOBJ_CLASS(dev, 0x0030, GR); /* null */ - NVOBJ_CLASS(dev, 0x5039, GR); /* m2mf */ - NVOBJ_CLASS(dev, 0x502d, GR); /* 2d */ - - /* tesla */ - if (dev_priv->chipset == 0x50) - NVOBJ_CLASS(dev, 0x5097, GR); /* tesla (nv50) */ - else - if (dev_priv->chipset < 0xa0) - NVOBJ_CLASS(dev, 0x8297, GR); /* tesla (nv8x/nv9x) */ - else { - switch (dev_priv->chipset) { - case 0xa0: - case 0xaa: - case 0xac: - NVOBJ_CLASS(dev, 0x8397, GR); - break; - case 0xa3: - case 0xa5: - case 0xa8: - NVOBJ_CLASS(dev, 0x8597, GR); - break; - case 0xaf: - NVOBJ_CLASS(dev, 0x8697, GR); - break; - } - } - - /* compute */ - NVOBJ_CLASS(dev, 0x50c0, GR); - if (dev_priv->chipset > 0xa0 && - dev_priv->chipset != 0xaa && - dev_priv->chipset != 0xac) - NVOBJ_CLASS(dev, 0x85c0, GR); - dev_priv->engine.graph.registered = true; - return 0; -} - -void -nv50_graph_tlb_flush(struct drm_device *dev) +static void +nv50_graph_tlb_flush(struct drm_device *dev, int engine) { nv50_vm_flush_engine(dev, 0); } -void -nv84_graph_tlb_flush(struct drm_device *dev) +static void +nv84_graph_tlb_flush(struct drm_device *dev, int engine) { struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_timer_engine *ptimer = &dev_priv->engine.timer; @@ -548,8 +499,7 @@ nv84_graph_tlb_flush(struct drm_device *dev) spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); } -static struct nouveau_enum nv50_mp_exec_error_names[] = -{ +static struct nouveau_enum nv50_mp_exec_error_names[] = { { 3, "STACK_UNDERFLOW", NULL }, { 4, "QUADON_ACTIVE", NULL }, { 8, "TIMEOUT", NULL }, @@ -663,7 +613,7 @@ nv50_pgraph_mp_trap(struct drm_device *dev, int tpid, int display) nv_rd32(dev, addr + 0x20); pc = nv_rd32(dev, addr + 0x24); oplow = nv_rd32(dev, addr + 0x70); - ophigh= nv_rd32(dev, addr + 0x74); + ophigh = nv_rd32(dev, addr + 0x74); NV_INFO(dev, "PGRAPH_TRAP_MP_EXEC - " "TP %d MP %d: ", tpid, i); nouveau_enum_print(nv50_mp_exec_error_names, status); @@ -991,7 +941,7 @@ nv50_pgraph_trap_handler(struct drm_device *dev, u32 display, u64 inst, u32 chid return 1; } -static int +int nv50_graph_isr_chid(struct drm_device *dev, u64 inst) { struct drm_nouveau_private *dev_priv = dev->dev_private; @@ -1073,3 +1023,101 @@ nv50_graph_isr(struct drm_device *dev) if (nv_rd32(dev, 0x400824) & (1 << 31)) nv_wr32(dev, 0x400824, nv_rd32(dev, 0x400824) & ~(1 << 31)); } + +static void +nv50_graph_destroy(struct drm_device *dev, int engine) +{ + struct nv50_graph_engine *pgraph = nv_engine(dev, engine); + + NVOBJ_ENGINE_DEL(dev, GR); + + nouveau_irq_unregister(dev, 12); + kfree(pgraph); +} + +int +nv50_graph_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_graph_engine *pgraph; + struct nouveau_grctx ctx = {}; + int ret; + + pgraph = kzalloc(sizeof(*pgraph),GFP_KERNEL); + if (!pgraph) + return -ENOMEM; + + ctx.dev = dev; + ctx.mode = NOUVEAU_GRCTX_PROG; + ctx.data = pgraph->ctxprog; + ctx.ctxprog_max = ARRAY_SIZE(pgraph->ctxprog); + + ret = nv50_grctx_init(&ctx); + if (ret) { + NV_ERROR(dev, "PGRAPH: ctxprog build failed\n"); + kfree(pgraph); + return 0; + } + + pgraph->grctx_size = ctx.ctxvals_pos * 4; + pgraph->ctxprog_size = ctx.ctxprog_len; + + pgraph->base.destroy = nv50_graph_destroy; + pgraph->base.init = nv50_graph_init; + pgraph->base.fini = nv50_graph_fini; + pgraph->base.context_new = nv50_graph_context_new; + pgraph->base.context_del = nv50_graph_context_del; + pgraph->base.object_new = nv50_graph_object_new; + if (dev_priv->chipset == 0x50 || dev_priv->chipset == 0xac) + pgraph->base.tlb_flush = nv50_graph_tlb_flush; + else + pgraph->base.tlb_flush = nv84_graph_tlb_flush; + + nouveau_irq_register(dev, 12, nv50_graph_isr); + + /* NVSW really doesn't live here... */ + NVOBJ_CLASS(dev, 0x506e, SW); /* nvsw */ + NVOBJ_MTHD (dev, 0x506e, 0x018c, nv50_graph_nvsw_dma_vblsem); + NVOBJ_MTHD (dev, 0x506e, 0x0400, nv50_graph_nvsw_vblsem_offset); + NVOBJ_MTHD (dev, 0x506e, 0x0404, nv50_graph_nvsw_vblsem_release_val); + NVOBJ_MTHD (dev, 0x506e, 0x0408, nv50_graph_nvsw_vblsem_release); + NVOBJ_MTHD (dev, 0x506e, 0x0500, nv50_graph_nvsw_mthd_page_flip); + + NVOBJ_ENGINE_ADD(dev, GR, &pgraph->base); + NVOBJ_CLASS(dev, 0x0030, GR); /* null */ + NVOBJ_CLASS(dev, 0x5039, GR); /* m2mf */ + NVOBJ_CLASS(dev, 0x502d, GR); /* 2d */ + + /* tesla */ + if (dev_priv->chipset == 0x50) + NVOBJ_CLASS(dev, 0x5097, GR); /* tesla (nv50) */ + else + if (dev_priv->chipset < 0xa0) + NVOBJ_CLASS(dev, 0x8297, GR); /* tesla (nv8x/nv9x) */ + else { + switch (dev_priv->chipset) { + case 0xa0: + case 0xaa: + case 0xac: + NVOBJ_CLASS(dev, 0x8397, GR); + break; + case 0xa3: + case 0xa5: + case 0xa8: + NVOBJ_CLASS(dev, 0x8597, GR); + break; + case 0xaf: + NVOBJ_CLASS(dev, 0x8697, GR); + break; + } + } + + /* compute */ + NVOBJ_CLASS(dev, 0x50c0, GR); + if (dev_priv->chipset > 0xa0 && + dev_priv->chipset != 0xaa && + dev_priv->chipset != 0xac) + NVOBJ_CLASS(dev, 0x85c0, GR); + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nv50_grctx.c b/drivers/gpu/drm/nouveau/nv50_grctx.c index 336aab2a24a..de9abff12b9 100644 --- a/drivers/gpu/drm/nouveau/nv50_grctx.c +++ b/drivers/gpu/drm/nouveau/nv50_grctx.c @@ -747,7 +747,7 @@ nv50_graph_construct_mmio(struct nouveau_grctx *ctx) gr_def(ctx, offset + 0x64, 0x0000001f); gr_def(ctx, offset + 0x68, 0x0000000f); gr_def(ctx, offset + 0x6c, 0x0000000f); - } else if(dev_priv->chipset < 0xa0) { + } else if (dev_priv->chipset < 0xa0) { cp_ctx(ctx, offset + 0x50, 1); cp_ctx(ctx, offset + 0x70, 1); } else { @@ -924,7 +924,7 @@ nv50_graph_construct_mmio_ddata(struct nouveau_grctx *ctx) dd_emit(ctx, 1, 0); /* 0000007f MULTISAMPLE_SAMPLES_LOG2 */ } else { dd_emit(ctx, 1, 0); /* 0000000f MULTISAMPLE_SAMPLES_LOG2 */ - } + } dd_emit(ctx, 1, 0xc); /* 000000ff SEMANTIC_COLOR.BFC0_ID */ if (dev_priv->chipset != 0x50) dd_emit(ctx, 1, 0); /* 00000001 SEMANTIC_COLOR.CLMP_EN */ @@ -1803,9 +1803,7 @@ nv50_graph_construct_gene_unk24xx(struct nouveau_grctx *ctx) xf_emit(ctx, 1, 0); /* 1ff */ xf_emit(ctx, 8, 0); /* 0? */ xf_emit(ctx, 9, 0); /* ffffffff, 7ff */ - } - else - { + } else { xf_emit(ctx, 0xc, 0); /* RO */ /* SEEK */ xf_emit(ctx, 0xe10, 0); /* 190 * 9: 8*ffffffff, 7ff */ @@ -2836,7 +2834,7 @@ nv50_graph_construct_xfer_tprop(struct nouveau_grctx *ctx) xf_emit(ctx, 1, 1); /* 00000001 DST_LINEAR */ if (IS_NVA3F(dev_priv->chipset)) xf_emit(ctx, 1, 1); /* 0000001f tesla UNK169C */ - if(dev_priv->chipset == 0x50) + if (dev_priv->chipset == 0x50) xf_emit(ctx, 1, 0); /* ff */ else xf_emit(ctx, 3, 0); /* 1, 7, 3ff */ diff --git a/drivers/gpu/drm/nouveau/nv50_instmem.c b/drivers/gpu/drm/nouveau/nv50_instmem.c index 4f95a1e5822..a7c12c94a5a 100644 --- a/drivers/gpu/drm/nouveau/nv50_instmem.c +++ b/drivers/gpu/drm/nouveau/nv50_instmem.c @@ -305,9 +305,9 @@ struct nv50_gpuobj_node { u32 align; }; - int -nv50_instmem_get(struct nouveau_gpuobj *gpuobj, u32 size, u32 align) +nv50_instmem_get(struct nouveau_gpuobj *gpuobj, struct nouveau_channel *chan, + u32 size, u32 align) { struct drm_device *dev = gpuobj->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; @@ -336,7 +336,7 @@ nv50_instmem_get(struct nouveau_gpuobj *gpuobj, u32 size, u32 align) if (!(gpuobj->flags & NVOBJ_FLAG_VM_USER)) flags |= NV_MEM_ACCESS_SYS; - ret = nouveau_vm_get(dev_priv->chan_vm, size, 12, flags, + ret = nouveau_vm_get(chan->vm, size, 12, flags, &node->chan_vma); if (ret) { vram->put(dev, &node->vram); @@ -345,7 +345,7 @@ nv50_instmem_get(struct nouveau_gpuobj *gpuobj, u32 size, u32 align) } nouveau_vm_map(&node->chan_vma, node->vram); - gpuobj->vinst = node->chan_vma.offset; + gpuobj->linst = node->chan_vma.offset; } gpuobj->size = size; diff --git a/drivers/gpu/drm/nouveau/nv50_mpeg.c b/drivers/gpu/drm/nouveau/nv50_mpeg.c new file mode 100644 index 00000000000..b57a2d180ad --- /dev/null +++ b/drivers/gpu/drm/nouveau/nv50_mpeg.c @@ -0,0 +1,256 @@ +/* + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_ramht.h" + +struct nv50_mpeg_engine { + struct nouveau_exec_engine base; +}; + +static inline u32 +CTX_PTR(struct drm_device *dev, u32 offset) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset == 0x50) + offset += 0x0260; + else + offset += 0x0060; + + return offset; +} + +static int +nv50_mpeg_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramin = chan->ramin; + struct nouveau_gpuobj *ctx = NULL; + int ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + ret = nouveau_gpuobj_new(dev, chan, 128 * 4, 0, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &ctx); + if (ret) + return ret; + + nv_wo32(ramin, CTX_PTR(dev, 0x00), 0x80190002); + nv_wo32(ramin, CTX_PTR(dev, 0x04), ctx->vinst + ctx->size - 1); + nv_wo32(ramin, CTX_PTR(dev, 0x08), ctx->vinst); + nv_wo32(ramin, CTX_PTR(dev, 0x0c), 0); + nv_wo32(ramin, CTX_PTR(dev, 0x10), 0); + nv_wo32(ramin, CTX_PTR(dev, 0x14), 0x00010000); + + nv_wo32(ctx, 0x70, 0x00801ec1); + nv_wo32(ctx, 0x7c, 0x0000037c); + dev_priv->engine.instmem.flush(dev); + + chan->engctx[engine] = ctx; + return 0; +} + +static void +nv50_mpeg_context_del(struct nouveau_channel *chan, int engine) +{ + struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + unsigned long flags; + u32 inst, i; + + if (!chan->ramin) + return; + + inst = chan->ramin->vinst >> 12; + inst |= 0x80000000; + + spin_lock_irqsave(&dev_priv->context_switch_lock, flags); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000000); + if (nv_rd32(dev, 0x00b318) == inst) + nv_mask(dev, 0x00b318, 0x80000000, 0x00000000); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000001); + spin_unlock_irqrestore(&dev_priv->context_switch_lock, flags); + + for (i = 0x00; i <= 0x14; i += 4) + nv_wo32(chan->ramin, CTX_PTR(dev, i), 0x00000000); + nouveau_gpuobj_ref(NULL, &ctx); + chan->engctx[engine] = NULL; +} + +static int +nv50_mpeg_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *obj = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 16, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 2; + obj->class = class; + + nv_wo32(obj, 0x00, class); + nv_wo32(obj, 0x04, 0x00000000); + nv_wo32(obj, 0x08, 0x00000000); + nv_wo32(obj, 0x0c, 0x00000000); + dev_priv->engine.instmem.flush(dev); + + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; +} + +static void +nv50_mpeg_tlb_flush(struct drm_device *dev, int engine) +{ + nv50_vm_flush_engine(dev, 0x08); +} + +static int +nv50_mpeg_init(struct drm_device *dev, int engine) +{ + nv_wr32(dev, 0x00b32c, 0x00000000); + nv_wr32(dev, 0x00b314, 0x00000100); + nv_wr32(dev, 0x00b0e0, 0x0000001a); + + nv_wr32(dev, 0x00b220, 0x00000044); + nv_wr32(dev, 0x00b300, 0x00801ec1); + nv_wr32(dev, 0x00b390, 0x00000000); + nv_wr32(dev, 0x00b394, 0x00000000); + nv_wr32(dev, 0x00b398, 0x00000000); + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000001); + + nv_wr32(dev, 0x00b100, 0xffffffff); + nv_wr32(dev, 0x00b140, 0xffffffff); + + if (!nv_wait(dev, 0x00b200, 0x00000001, 0x00000000)) { + NV_ERROR(dev, "PMPEG init: 0x%08x\n", nv_rd32(dev, 0x00b200)); + return -EBUSY; + } + + return 0; +} + +static int +nv50_mpeg_fini(struct drm_device *dev, int engine, bool suspend) +{ + /*XXX: context save for s/r */ + nv_mask(dev, 0x00b32c, 0x00000001, 0x00000000); + nv_wr32(dev, 0x00b140, 0x00000000); + return 0; +} + +static void +nv50_mpeg_isr(struct drm_device *dev) +{ + u32 stat = nv_rd32(dev, 0x00b100); + u32 type = nv_rd32(dev, 0x00b230); + u32 mthd = nv_rd32(dev, 0x00b234); + u32 data = nv_rd32(dev, 0x00b238); + u32 show = stat; + + if (stat & 0x01000000) { + /* happens on initial binding of the object */ + if (type == 0x00000020 && mthd == 0x0000) { + nv_wr32(dev, 0x00b308, 0x00000100); + show &= ~0x01000000; + } + } + + if (show && nouveau_ratelimit()) { + NV_INFO(dev, "PMPEG - 0x%08x 0x%08x 0x%08x 0x%08x\n", + stat, type, mthd, data); + } + + nv_wr32(dev, 0x00b100, stat); + nv_wr32(dev, 0x00b230, 0x00000001); + nv50_fb_vm_trap(dev, 1); +} + +static void +nv50_vpe_isr(struct drm_device *dev) +{ + if (nv_rd32(dev, 0x00b100)) + nv50_mpeg_isr(dev); + + if (nv_rd32(dev, 0x00b800)) { + u32 stat = nv_rd32(dev, 0x00b800); + NV_INFO(dev, "PMSRCH: 0x%08x\n", stat); + nv_wr32(dev, 0xb800, stat); + } +} + +static void +nv50_mpeg_destroy(struct drm_device *dev, int engine) +{ + struct nv50_mpeg_engine *pmpeg = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 0); + + NVOBJ_ENGINE_DEL(dev, MPEG); + kfree(pmpeg); +} + +int +nv50_mpeg_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nv50_mpeg_engine *pmpeg; + + pmpeg = kzalloc(sizeof(*pmpeg), GFP_KERNEL); + if (!pmpeg) + return -ENOMEM; + + pmpeg->base.destroy = nv50_mpeg_destroy; + pmpeg->base.init = nv50_mpeg_init; + pmpeg->base.fini = nv50_mpeg_fini; + pmpeg->base.context_new = nv50_mpeg_context_new; + pmpeg->base.context_del = nv50_mpeg_context_del; + pmpeg->base.object_new = nv50_mpeg_object_new; + pmpeg->base.tlb_flush = nv50_mpeg_tlb_flush; + + if (dev_priv->chipset == 0x50) { + nouveau_irq_register(dev, 0, nv50_vpe_isr); + NVOBJ_ENGINE_ADD(dev, MPEG, &pmpeg->base); + NVOBJ_CLASS(dev, 0x3174, MPEG); +#if 0 + NVOBJ_ENGINE_ADD(dev, ME, &pme->base); + NVOBJ_CLASS(dev, 0x4075, ME); +#endif + } else { + nouveau_irq_register(dev, 0, nv50_mpeg_isr); + NVOBJ_ENGINE_ADD(dev, MPEG, &pmpeg->base); + NVOBJ_CLASS(dev, 0x8274, MPEG); + } + + return 0; + +} diff --git a/drivers/gpu/drm/nouveau/nv50_pm.c b/drivers/gpu/drm/nouveau/nv50_pm.c index 7dbb305d7e6..8a2810011bd 100644 --- a/drivers/gpu/drm/nouveau/nv50_pm.c +++ b/drivers/gpu/drm/nouveau/nv50_pm.c @@ -47,6 +47,21 @@ nv50_pm_clock_get(struct drm_device *dev, u32 id) reg0 = nv_rd32(dev, pll.reg + 0); reg1 = nv_rd32(dev, pll.reg + 4); + + if ((reg0 & 0x80000000) == 0) { + if (id == PLL_SHADER) { + NV_DEBUG(dev, "Shader PLL is disabled. " + "Shader clock is twice the core\n"); + ret = nv50_pm_clock_get(dev, PLL_CORE); + if (ret > 0) + return ret << 1; + } else if (id == PLL_MEMORY) { + NV_DEBUG(dev, "Memory PLL is disabled. " + "Memory clock is equal to the ref_clk\n"); + return pll.refclk; + } + } + P = (reg0 & 0x00070000) >> 16; N = (reg1 & 0x0000ff00) >> 8; M = (reg1 & 0x000000ff); diff --git a/drivers/gpu/drm/nouveau/nv50_sor.c b/drivers/gpu/drm/nouveau/nv50_sor.c index c25c5938642..ffe8b483b7b 100644 --- a/drivers/gpu/drm/nouveau/nv50_sor.c +++ b/drivers/gpu/drm/nouveau/nv50_sor.c @@ -318,6 +318,8 @@ nv50_sor_create(struct drm_connector *connector, struct dcb_entry *entry) uint32_t tmp; tmp = nv_rd32(dev, 0x61c700 + (or * 0x800)); + if (!tmp) + tmp = nv_rd32(dev, 0x610798 + (or * 8)); switch ((tmp & 0x00000f00) >> 8) { case 8: diff --git a/drivers/gpu/drm/nouveau/nv50_vm.c b/drivers/gpu/drm/nouveau/nv50_vm.c index 6c269449074..40b84f22d81 100644 --- a/drivers/gpu/drm/nouveau/nv50_vm.c +++ b/drivers/gpu/drm/nouveau/nv50_vm.c @@ -151,23 +151,21 @@ nv50_vm_flush(struct nouveau_vm *vm) struct drm_nouveau_private *dev_priv = vm->dev->dev_private; struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct nouveau_crypt_engine *pcrypt = &dev_priv->engine.crypt; + int i; pinstmem->flush(vm->dev); /* BAR */ - if (vm != dev_priv->chan_vm) { + if (vm == dev_priv->bar1_vm || vm == dev_priv->bar3_vm) { nv50_vm_flush_engine(vm->dev, 6); return; } pfifo->tlb_flush(vm->dev); - - if (atomic_read(&vm->pgraph_refs)) - pgraph->tlb_flush(vm->dev); - if (atomic_read(&vm->pcrypt_refs)) - pcrypt->tlb_flush(vm->dev); + for (i = 0; i < NVOBJ_ENGINE_NR; i++) { + if (atomic_read(&vm->engref[i])) + dev_priv->eng[i]->tlb_flush(vm->dev, i); + } } void diff --git a/drivers/gpu/drm/nouveau/nv50_vram.c b/drivers/gpu/drm/nouveau/nv50_vram.c index ffbc3d8cf5b..af32daecd1e 100644 --- a/drivers/gpu/drm/nouveau/nv50_vram.c +++ b/drivers/gpu/drm/nouveau/nv50_vram.c @@ -51,9 +51,7 @@ void nv50_vram_del(struct drm_device *dev, struct nouveau_mem **pmem) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct ttm_bo_device *bdev = &dev_priv->ttm.bdev; - struct ttm_mem_type_manager *man = &bdev->man[TTM_PL_VRAM]; - struct nouveau_mm *mm = man->priv; + struct nouveau_mm *mm = dev_priv->engine.vram.mm; struct nouveau_mm_node *this; struct nouveau_mem *mem; @@ -84,9 +82,7 @@ nv50_vram_new(struct drm_device *dev, u64 size, u32 align, u32 size_nc, u32 memtype, struct nouveau_mem **pmem) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct ttm_bo_device *bdev = &dev_priv->ttm.bdev; - struct ttm_mem_type_manager *man = &bdev->man[TTM_PL_VRAM]; - struct nouveau_mm *mm = man->priv; + struct nouveau_mm *mm = dev_priv->engine.vram.mm; struct nouveau_mm_node *r; struct nouveau_mem *mem; int comp = (memtype & 0x300) >> 8; @@ -190,22 +186,35 @@ int nv50_vram_init(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_vram_engine *vram = &dev_priv->engine.vram; + const u32 rsvd_head = ( 256 * 1024) >> 12; /* vga memory */ + const u32 rsvd_tail = (1024 * 1024) >> 12; /* vbios etc */ + u32 rblock, length; dev_priv->vram_size = nv_rd32(dev, 0x10020c); dev_priv->vram_size |= (dev_priv->vram_size & 0xff) << 32; dev_priv->vram_size &= 0xffffffff00ULL; - switch (dev_priv->chipset) { - case 0xaa: - case 0xac: - case 0xaf: + /* IGPs, no funky reordering happens here, they don't have VRAM */ + if (dev_priv->chipset == 0xaa || + dev_priv->chipset == 0xac || + dev_priv->chipset == 0xaf) { dev_priv->vram_sys_base = (u64)nv_rd32(dev, 0x100e10) << 12; - dev_priv->vram_rblock_size = 4096; - break; - default: - dev_priv->vram_rblock_size = nv50_vram_rblock(dev); - break; + rblock = 4096 >> 12; + } else { + rblock = nv50_vram_rblock(dev) >> 12; } - return 0; + length = (dev_priv->vram_size >> 12) - rsvd_head - rsvd_tail; + + return nouveau_mm_init(&vram->mm, rsvd_head, length, rblock); +} + +void +nv50_vram_fini(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_vram_engine *vram = &dev_priv->engine.vram; + + nouveau_mm_fini(&vram->mm); } diff --git a/drivers/gpu/drm/nouveau/nv84_crypt.c b/drivers/gpu/drm/nouveau/nv84_crypt.c index fabc7fd30b1..edece9c616e 100644 --- a/drivers/gpu/drm/nouveau/nv84_crypt.c +++ b/drivers/gpu/drm/nouveau/nv84_crypt.c @@ -26,46 +26,48 @@ #include "nouveau_drv.h" #include "nouveau_util.h" #include "nouveau_vm.h" +#include "nouveau_ramht.h" -static void nv84_crypt_isr(struct drm_device *); +struct nv84_crypt_engine { + struct nouveau_exec_engine base; +}; -int -nv84_crypt_create_context(struct nouveau_channel *chan) +static int +nv84_crypt_context_new(struct nouveau_channel *chan, int engine) { struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_gpuobj *ramin = chan->ramin; + struct nouveau_gpuobj *ctx; int ret; NV_DEBUG(dev, "ch%d\n", chan->id); - ret = nouveau_gpuobj_new(dev, chan, 256, 0, - NVOBJ_FLAG_ZERO_ALLOC | NVOBJ_FLAG_ZERO_FREE, - &chan->crypt_ctx); + ret = nouveau_gpuobj_new(dev, chan, 256, 0, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &ctx); if (ret) return ret; nv_wo32(ramin, 0xa0, 0x00190000); - nv_wo32(ramin, 0xa4, chan->crypt_ctx->vinst + 0xff); - nv_wo32(ramin, 0xa8, chan->crypt_ctx->vinst); + nv_wo32(ramin, 0xa4, ctx->vinst + ctx->size - 1); + nv_wo32(ramin, 0xa8, ctx->vinst); nv_wo32(ramin, 0xac, 0); nv_wo32(ramin, 0xb0, 0); nv_wo32(ramin, 0xb4, 0); - dev_priv->engine.instmem.flush(dev); - atomic_inc(&chan->vm->pcrypt_refs); + + atomic_inc(&chan->vm->engref[engine]); + chan->engctx[engine] = ctx; return 0; } -void -nv84_crypt_destroy_context(struct nouveau_channel *chan) +static void +nv84_crypt_context_del(struct nouveau_channel *chan, int engine) { + struct nouveau_gpuobj *ctx = chan->engctx[engine]; struct drm_device *dev = chan->dev; u32 inst; - if (!chan->crypt_ctx) - return; - inst = (chan->ramin->vinst >> 12); inst |= 0x80000000; @@ -80,43 +82,39 @@ nv84_crypt_destroy_context(struct nouveau_channel *chan) nv_mask(dev, 0x10218c, 0x80000000, 0x00000000); nv_wr32(dev, 0x10200c, 0x00000010); - nouveau_gpuobj_ref(NULL, &chan->crypt_ctx); - atomic_dec(&chan->vm->pcrypt_refs); -} + nouveau_gpuobj_ref(NULL, &ctx); -void -nv84_crypt_tlb_flush(struct drm_device *dev) -{ - nv50_vm_flush_engine(dev, 0x0a); + atomic_dec(&chan->vm->engref[engine]); + chan->engctx[engine] = NULL; } -int -nv84_crypt_init(struct drm_device *dev) +static int +nv84_crypt_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) { + struct drm_device *dev = chan->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_crypt_engine *pcrypt = &dev_priv->engine.crypt; - - if (!pcrypt->registered) { - NVOBJ_CLASS(dev, 0x74c1, CRYPT); - pcrypt->registered = true; - } + struct nouveau_gpuobj *obj = NULL; + int ret; - nv_mask(dev, 0x000200, 0x00004000, 0x00000000); - nv_mask(dev, 0x000200, 0x00004000, 0x00004000); + ret = nouveau_gpuobj_new(dev, chan, 16, 16, NVOBJ_FLAG_ZERO_FREE, &obj); + if (ret) + return ret; + obj->engine = 5; + obj->class = class; - nouveau_irq_register(dev, 14, nv84_crypt_isr); - nv_wr32(dev, 0x102130, 0xffffffff); - nv_wr32(dev, 0x102140, 0xffffffbf); + nv_wo32(obj, 0x00, class); + dev_priv->engine.instmem.flush(dev); - nv_wr32(dev, 0x10200c, 0x00000010); - return 0; + ret = nouveau_ramht_insert(chan, handle, obj); + nouveau_gpuobj_ref(NULL, &obj); + return ret; } -void -nv84_crypt_fini(struct drm_device *dev) +static void +nv84_crypt_tlb_flush(struct drm_device *dev, int engine) { - nv_wr32(dev, 0x102140, 0x00000000); - nouveau_irq_unregister(dev, 14); + nv50_vm_flush_engine(dev, 0x0a); } static void @@ -138,3 +136,58 @@ nv84_crypt_isr(struct drm_device *dev) nv50_fb_vm_trap(dev, show); } + +static int +nv84_crypt_fini(struct drm_device *dev, int engine, bool suspend) +{ + nv_wr32(dev, 0x102140, 0x00000000); + return 0; +} + +static int +nv84_crypt_init(struct drm_device *dev, int engine) +{ + nv_mask(dev, 0x000200, 0x00004000, 0x00000000); + nv_mask(dev, 0x000200, 0x00004000, 0x00004000); + + nv_wr32(dev, 0x102130, 0xffffffff); + nv_wr32(dev, 0x102140, 0xffffffbf); + + nv_wr32(dev, 0x10200c, 0x00000010); + return 0; +} + +static void +nv84_crypt_destroy(struct drm_device *dev, int engine) +{ + struct nv84_crypt_engine *pcrypt = nv_engine(dev, engine); + + NVOBJ_ENGINE_DEL(dev, CRYPT); + + nouveau_irq_unregister(dev, 14); + kfree(pcrypt); +} + +int +nv84_crypt_create(struct drm_device *dev) +{ + struct nv84_crypt_engine *pcrypt; + + pcrypt = kzalloc(sizeof(*pcrypt), GFP_KERNEL); + if (!pcrypt) + return -ENOMEM; + + pcrypt->base.destroy = nv84_crypt_destroy; + pcrypt->base.init = nv84_crypt_init; + pcrypt->base.fini = nv84_crypt_fini; + pcrypt->base.context_new = nv84_crypt_context_new; + pcrypt->base.context_del = nv84_crypt_context_del; + pcrypt->base.object_new = nv84_crypt_object_new; + pcrypt->base.tlb_flush = nv84_crypt_tlb_flush; + + nouveau_irq_register(dev, 14, nv84_crypt_isr); + + NVOBJ_ENGINE_ADD(dev, CRYPT, &pcrypt->base); + NVOBJ_CLASS (dev, 0x74c1, CRYPT); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nva3_copy.c b/drivers/gpu/drm/nouveau/nva3_copy.c new file mode 100644 index 00000000000..8f356d58e40 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nva3_copy.c @@ -0,0 +1,226 @@ +/* + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include <linux/firmware.h> +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_util.h" +#include "nouveau_vm.h" +#include "nouveau_ramht.h" +#include "nva3_copy.fuc.h" + +struct nva3_copy_engine { + struct nouveau_exec_engine base; +}; + +static int +nva3_copy_context_new(struct nouveau_channel *chan, int engine) +{ + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramin = chan->ramin; + struct nouveau_gpuobj *ctx = NULL; + int ret; + + NV_DEBUG(dev, "ch%d\n", chan->id); + + ret = nouveau_gpuobj_new(dev, chan, 256, 0, NVOBJ_FLAG_ZERO_ALLOC | + NVOBJ_FLAG_ZERO_FREE, &ctx); + if (ret) + return ret; + + nv_wo32(ramin, 0xc0, 0x00190000); + nv_wo32(ramin, 0xc4, ctx->vinst + ctx->size - 1); + nv_wo32(ramin, 0xc8, ctx->vinst); + nv_wo32(ramin, 0xcc, 0x00000000); + nv_wo32(ramin, 0xd0, 0x00000000); + nv_wo32(ramin, 0xd4, 0x00000000); + dev_priv->engine.instmem.flush(dev); + + atomic_inc(&chan->vm->engref[engine]); + chan->engctx[engine] = ctx; + return 0; +} + +static int +nva3_copy_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + + /* fuc engine doesn't need an object, our ramht code does.. */ + ctx->engine = 3; + ctx->class = class; + return nouveau_ramht_insert(chan, handle, ctx); +} + +static void +nva3_copy_context_del(struct nouveau_channel *chan, int engine) +{ + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + u32 inst; + + inst = (chan->ramin->vinst >> 12); + inst |= 0x40000000; + + /* disable fifo access */ + nv_wr32(dev, 0x104048, 0x00000000); + /* mark channel as unloaded if it's currently active */ + if (nv_rd32(dev, 0x104050) == inst) + nv_mask(dev, 0x104050, 0x40000000, 0x00000000); + /* mark next channel as invalid if it's about to be loaded */ + if (nv_rd32(dev, 0x104054) == inst) + nv_mask(dev, 0x104054, 0x40000000, 0x00000000); + /* restore fifo access */ + nv_wr32(dev, 0x104048, 0x00000003); + + for (inst = 0xc0; inst <= 0xd4; inst += 4) + nv_wo32(chan->ramin, inst, 0x00000000); + + nouveau_gpuobj_ref(NULL, &ctx); + + atomic_dec(&chan->vm->engref[engine]); + chan->engctx[engine] = ctx; +} + +static void +nva3_copy_tlb_flush(struct drm_device *dev, int engine) +{ + nv50_vm_flush_engine(dev, 0x0d); +} + +static int +nva3_copy_init(struct drm_device *dev, int engine) +{ + int i; + + nv_mask(dev, 0x000200, 0x00002000, 0x00000000); + nv_mask(dev, 0x000200, 0x00002000, 0x00002000); + nv_wr32(dev, 0x104014, 0xffffffff); /* disable all interrupts */ + + /* upload ucode */ + nv_wr32(dev, 0x1041c0, 0x01000000); + for (i = 0; i < sizeof(nva3_pcopy_data) / 4; i++) + nv_wr32(dev, 0x1041c4, nva3_pcopy_data[i]); + + nv_wr32(dev, 0x104180, 0x01000000); + for (i = 0; i < sizeof(nva3_pcopy_code) / 4; i++) { + if ((i & 0x3f) == 0) + nv_wr32(dev, 0x104188, i >> 6); + nv_wr32(dev, 0x104184, nva3_pcopy_code[i]); + } + + /* start it running */ + nv_wr32(dev, 0x10410c, 0x00000000); + nv_wr32(dev, 0x104104, 0x00000000); /* ENTRY */ + nv_wr32(dev, 0x104100, 0x00000002); /* TRIGGER */ + return 0; +} + +static int +nva3_copy_fini(struct drm_device *dev, int engine, bool suspend) +{ + nv_mask(dev, 0x104048, 0x00000003, 0x00000000); + + /* trigger fuc context unload */ + nv_wait(dev, 0x104008, 0x0000000c, 0x00000000); + nv_mask(dev, 0x104054, 0x40000000, 0x00000000); + nv_wr32(dev, 0x104000, 0x00000008); + nv_wait(dev, 0x104008, 0x00000008, 0x00000000); + + nv_wr32(dev, 0x104014, 0xffffffff); + return 0; +} + +static struct nouveau_enum nva3_copy_isr_error_name[] = { + { 0x0001, "ILLEGAL_MTHD" }, + { 0x0002, "INVALID_ENUM" }, + { 0x0003, "INVALID_BITFIELD" }, + {} +}; + +static void +nva3_copy_isr(struct drm_device *dev) +{ + u32 dispatch = nv_rd32(dev, 0x10401c); + u32 stat = nv_rd32(dev, 0x104008) & dispatch & ~(dispatch >> 16); + u32 inst = nv_rd32(dev, 0x104050) & 0x3fffffff; + u32 ssta = nv_rd32(dev, 0x104040) & 0x0000ffff; + u32 addr = nv_rd32(dev, 0x104040) >> 16; + u32 mthd = (addr & 0x07ff) << 2; + u32 subc = (addr & 0x3800) >> 11; + u32 data = nv_rd32(dev, 0x104044); + int chid = nv50_graph_isr_chid(dev, inst); + + if (stat & 0x00000040) { + NV_INFO(dev, "PCOPY: DISPATCH_ERROR ["); + nouveau_enum_print(nva3_copy_isr_error_name, ssta); + printk("] ch %d [0x%08x] subc %d mthd 0x%04x data 0x%08x\n", + chid, inst, subc, mthd, data); + nv_wr32(dev, 0x104004, 0x00000040); + stat &= ~0x00000040; + } + + if (stat) { + NV_INFO(dev, "PCOPY: unhandled intr 0x%08x\n", stat); + nv_wr32(dev, 0x104004, stat); + } + nv50_fb_vm_trap(dev, 1); +} + +static void +nva3_copy_destroy(struct drm_device *dev, int engine) +{ + struct nva3_copy_engine *pcopy = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, 22); + + NVOBJ_ENGINE_DEL(dev, COPY0); + kfree(pcopy); +} + +int +nva3_copy_create(struct drm_device *dev) +{ + struct nva3_copy_engine *pcopy; + + pcopy = kzalloc(sizeof(*pcopy), GFP_KERNEL); + if (!pcopy) + return -ENOMEM; + + pcopy->base.destroy = nva3_copy_destroy; + pcopy->base.init = nva3_copy_init; + pcopy->base.fini = nva3_copy_fini; + pcopy->base.context_new = nva3_copy_context_new; + pcopy->base.context_del = nva3_copy_context_del; + pcopy->base.object_new = nva3_copy_object_new; + pcopy->base.tlb_flush = nva3_copy_tlb_flush; + + nouveau_irq_register(dev, 22, nva3_copy_isr); + + NVOBJ_ENGINE_ADD(dev, COPY0, &pcopy->base); + NVOBJ_CLASS(dev, 0x85b5, COPY0); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nva3_copy.fuc b/drivers/gpu/drm/nouveau/nva3_copy.fuc new file mode 100644 index 00000000000..eaf35f8321e --- /dev/null +++ b/drivers/gpu/drm/nouveau/nva3_copy.fuc @@ -0,0 +1,870 @@ +/* fuc microcode for copy engine on nva3- chipsets + * + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +/* To build for nva3:nvc0 + * m4 -DNVA3 nva3_copy.fuc | envyas -a -w -m fuc -V nva3 -o nva3_copy.fuc.h + * + * To build for nvc0- + * m4 -DNVC0 nva3_copy.fuc | envyas -a -w -m fuc -V nva3 -o nvc0_copy.fuc.h + */ + +ifdef(`NVA3', +.section nva3_pcopy_data, +.section nvc0_pcopy_data +) + +ctx_object: .b32 0 +ifdef(`NVA3', +ctx_dma: +ctx_dma_query: .b32 0 +ctx_dma_src: .b32 0 +ctx_dma_dst: .b32 0 +,) +.equ ctx_dma_count 3 +ctx_query_address_high: .b32 0 +ctx_query_address_low: .b32 0 +ctx_query_counter: .b32 0 +ctx_src_address_high: .b32 0 +ctx_src_address_low: .b32 0 +ctx_src_pitch: .b32 0 +ctx_src_tile_mode: .b32 0 +ctx_src_xsize: .b32 0 +ctx_src_ysize: .b32 0 +ctx_src_zsize: .b32 0 +ctx_src_zoff: .b32 0 +ctx_src_xoff: .b32 0 +ctx_src_yoff: .b32 0 +ctx_src_cpp: .b32 0 +ctx_dst_address_high: .b32 0 +ctx_dst_address_low: .b32 0 +ctx_dst_pitch: .b32 0 +ctx_dst_tile_mode: .b32 0 +ctx_dst_xsize: .b32 0 +ctx_dst_ysize: .b32 0 +ctx_dst_zsize: .b32 0 +ctx_dst_zoff: .b32 0 +ctx_dst_xoff: .b32 0 +ctx_dst_yoff: .b32 0 +ctx_dst_cpp: .b32 0 +ctx_format: .b32 0 +ctx_swz_const0: .b32 0 +ctx_swz_const1: .b32 0 +ctx_xcnt: .b32 0 +ctx_ycnt: .b32 0 +.align 256 + +dispatch_table: +// mthd 0x0000, NAME +.b16 0x000 1 +.b32 ctx_object ~0xffffffff +// mthd 0x0100, NOP +.b16 0x040 1 +.b32 0x00010000 + cmd_nop ~0xffffffff +// mthd 0x0140, PM_TRIGGER +.b16 0x050 1 +.b32 0x00010000 + cmd_pm_trigger ~0xffffffff +ifdef(`NVA3', ` +// mthd 0x0180-0x018c, DMA_ +.b16 0x060 ctx_dma_count +dispatch_dma: +.b32 0x00010000 + cmd_dma ~0xffffffff +.b32 0x00010000 + cmd_dma ~0xffffffff +.b32 0x00010000 + cmd_dma ~0xffffffff +',) +// mthd 0x0200-0x0218, SRC_TILE +.b16 0x80 7 +.b32 ctx_src_tile_mode ~0x00000fff +.b32 ctx_src_xsize ~0x0007ffff +.b32 ctx_src_ysize ~0x00001fff +.b32 ctx_src_zsize ~0x000007ff +.b32 ctx_src_zoff ~0x00000fff +.b32 ctx_src_xoff ~0x0007ffff +.b32 ctx_src_yoff ~0x00001fff +// mthd 0x0220-0x0238, DST_TILE +.b16 0x88 7 +.b32 ctx_dst_tile_mode ~0x00000fff +.b32 ctx_dst_xsize ~0x0007ffff +.b32 ctx_dst_ysize ~0x00001fff +.b32 ctx_dst_zsize ~0x000007ff +.b32 ctx_dst_zoff ~0x00000fff +.b32 ctx_dst_xoff ~0x0007ffff +.b32 ctx_dst_yoff ~0x00001fff +// mthd 0x0300-0x0304, EXEC, WRCACHE_FLUSH +.b16 0xc0 2 +.b32 0x00010000 + cmd_exec ~0xffffffff +.b32 0x00010000 + cmd_wrcache_flush ~0xffffffff +// mthd 0x030c-0x0340, various stuff +.b16 0xc3 14 +.b32 ctx_src_address_high ~0x000000ff +.b32 ctx_src_address_low ~0xfffffff0 +.b32 ctx_dst_address_high ~0x000000ff +.b32 ctx_dst_address_low ~0xfffffff0 +.b32 ctx_src_pitch ~0x0007ffff +.b32 ctx_dst_pitch ~0x0007ffff +.b32 ctx_xcnt ~0x0000ffff +.b32 ctx_ycnt ~0x00001fff +.b32 ctx_format ~0x0333ffff +.b32 ctx_swz_const0 ~0xffffffff +.b32 ctx_swz_const1 ~0xffffffff +.b32 ctx_query_address_high ~0x000000ff +.b32 ctx_query_address_low ~0xffffffff +.b32 ctx_query_counter ~0xffffffff +.b16 0x800 0 + +ifdef(`NVA3', +.section nva3_pcopy_code, +.section nvc0_pcopy_code +) + +main: + clear b32 $r0 + mov $sp $r0 + + // setup i0 handler and route fifo and ctxswitch to it + mov $r1 ih + mov $iv0 $r1 + mov $r1 0x400 + movw $r2 0xfff3 + sethi $r2 0 + iowr I[$r2 + 0x300] $r2 + + // enable interrupts + or $r2 0xc + iowr I[$r1] $r2 + bset $flags ie0 + + // enable fifo access and context switching + mov $r1 0x1200 + mov $r2 3 + iowr I[$r1] $r2 + + // sleep forever, waking for interrupts + bset $flags $p0 + spin: + sleep $p0 + bra spin + +// i0 handler +ih: + iord $r1 I[$r0 + 0x200] + + and $r2 $r1 0x00000008 + bra e ih_no_chsw + call chsw + ih_no_chsw: + and $r2 $r1 0x00000004 + bra e ih_no_cmd + call dispatch + + ih_no_cmd: + and $r1 $r1 0x0000000c + iowr I[$r0 + 0x100] $r1 + iret + +// $p1 direction (0 = unload, 1 = load) +// $r3 channel +swctx: + mov $r4 0x7700 + mov $xtargets $r4 +ifdef(`NVA3', ` + // target 7 hardcoded to ctx dma object + mov $xdbase $r0 +', ` // NVC0 + // read SCRATCH3 to decide if we are PCOPY0 or PCOPY1 + mov $r4 0x2100 + iord $r4 I[$r4 + 0] + and $r4 1 + shl b32 $r4 4 + add b32 $r4 0x30 + + // channel is in vram + mov $r15 0x61c + shl b32 $r15 6 + mov $r5 0x114 + iowrs I[$r15] $r5 + + // read 16-byte PCOPYn info, containing context pointer, from channel + shl b32 $r5 $r3 4 + add b32 $r5 2 + mov $xdbase $r5 + mov $r5 $sp + // get a chunk of stack space, aligned to 256 byte boundary + sub b32 $r5 0x100 + mov $r6 0xff + not b32 $r6 + and $r5 $r6 + sethi $r5 0x00020000 + xdld $r4 $r5 + xdwait + sethi $r5 0 + + // set context pointer, from within channel VM + mov $r14 0 + iowrs I[$r15] $r14 + ld b32 $r4 D[$r5 + 0] + shr b32 $r4 8 + ld b32 $r6 D[$r5 + 4] + shl b32 $r6 24 + or $r4 $r6 + mov $xdbase $r4 +') + // 256-byte context, at start of data segment + mov b32 $r4 $r0 + sethi $r4 0x60000 + + // swap! + bra $p1 swctx_load + xdst $r0 $r4 + bra swctx_done + swctx_load: + xdld $r0 $r4 + swctx_done: + xdwait + ret + +chsw: + // read current channel + mov $r2 0x1400 + iord $r3 I[$r2] + + // if it's active, unload it and return + xbit $r15 $r3 0x1e + bra e chsw_no_unload + bclr $flags $p1 + call swctx + bclr $r3 0x1e + iowr I[$r2] $r3 + mov $r4 1 + iowr I[$r2 + 0x200] $r4 + ret + + // read next channel + chsw_no_unload: + iord $r3 I[$r2 + 0x100] + + // is there a channel waiting to be loaded? + xbit $r13 $r3 0x1e + bra e chsw_finish_load + bset $flags $p1 + call swctx +ifdef(`NVA3', + // load dma objects back into TARGET regs + mov $r5 ctx_dma + mov $r6 ctx_dma_count + chsw_load_ctx_dma: + ld b32 $r7 D[$r5 + $r6 * 4] + add b32 $r8 $r6 0x180 + shl b32 $r8 8 + iowr I[$r8] $r7 + sub b32 $r6 1 + bra nc chsw_load_ctx_dma +,) + + chsw_finish_load: + mov $r3 2 + iowr I[$r2 + 0x200] $r3 + ret + +dispatch: + // read incoming fifo command + mov $r3 0x1900 + iord $r2 I[$r3 + 0x100] + iord $r3 I[$r3 + 0x000] + and $r4 $r2 0x7ff + // $r2 will be used to store exception data + shl b32 $r2 0x10 + + // lookup method in the dispatch table, ILLEGAL_MTHD if not found + mov $r5 dispatch_table + clear b32 $r6 + clear b32 $r7 + dispatch_loop: + ld b16 $r6 D[$r5 + 0] + ld b16 $r7 D[$r5 + 2] + add b32 $r5 4 + cmpu b32 $r4 $r6 + bra c dispatch_illegal_mthd + add b32 $r7 $r6 + cmpu b32 $r4 $r7 + bra c dispatch_valid_mthd + sub b32 $r7 $r6 + shl b32 $r7 3 + add b32 $r5 $r7 + bra dispatch_loop + + // ensure no bits set in reserved fields, INVALID_BITFIELD + dispatch_valid_mthd: + sub b32 $r4 $r6 + shl b32 $r4 3 + add b32 $r4 $r5 + ld b32 $r5 D[$r4 + 4] + and $r5 $r3 + cmpu b32 $r5 0 + bra ne dispatch_invalid_bitfield + + // depending on dispatch flags: execute method, or save data as state + ld b16 $r5 D[$r4 + 0] + ld b16 $r6 D[$r4 + 2] + cmpu b32 $r6 0 + bra ne dispatch_cmd + st b32 D[$r5] $r3 + bra dispatch_done + dispatch_cmd: + bclr $flags $p1 + call $r5 + bra $p1 dispatch_error + bra dispatch_done + + dispatch_invalid_bitfield: + or $r2 2 + dispatch_illegal_mthd: + or $r2 1 + + // store exception data in SCRATCH0/SCRATCH1, signal hostirq + dispatch_error: + mov $r4 0x1000 + iowr I[$r4 + 0x000] $r2 + iowr I[$r4 + 0x100] $r3 + mov $r2 0x40 + iowr I[$r0] $r2 + hostirq_wait: + iord $r2 I[$r0 + 0x200] + and $r2 0x40 + cmpu b32 $r2 0 + bra ne hostirq_wait + + dispatch_done: + mov $r2 0x1d00 + mov $r3 1 + iowr I[$r2] $r3 + ret + +// No-operation +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_nop: + ret + +// PM_TRIGGER +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_pm_trigger: + mov $r2 0x2200 + clear b32 $r3 + sethi $r3 0x20000 + iowr I[$r2] $r3 + ret + +ifdef(`NVA3', +// SET_DMA_* method handler +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_dma: + sub b32 $r4 dispatch_dma + shr b32 $r4 1 + bset $r3 0x1e + st b32 D[$r4 + ctx_dma] $r3 + add b32 $r4 0x600 + shl b32 $r4 6 + iowr I[$r4] $r3 + ret +,) + +// Calculates the hw swizzle mask and adjusts the surface's xcnt to match +// +cmd_exec_set_format: + // zero out a chunk of the stack to store the swizzle into + add $sp -0x10 + st b32 D[$sp + 0x00] $r0 + st b32 D[$sp + 0x04] $r0 + st b32 D[$sp + 0x08] $r0 + st b32 D[$sp + 0x0c] $r0 + + // extract cpp, src_ncomp and dst_ncomp from FORMAT + ld b32 $r4 D[$r0 + ctx_format] + extr $r5 $r4 16:17 + add b32 $r5 1 + extr $r6 $r4 20:21 + add b32 $r6 1 + extr $r7 $r4 24:25 + add b32 $r7 1 + + // convert FORMAT swizzle mask to hw swizzle mask + bclr $flags $p2 + clear b32 $r8 + clear b32 $r9 + ncomp_loop: + and $r10 $r4 0xf + shr b32 $r4 4 + clear b32 $r11 + bpc_loop: + cmpu b8 $r10 4 + bra nc cmp_c0 + mulu $r12 $r10 $r5 + add b32 $r12 $r11 + bset $flags $p2 + bra bpc_next + cmp_c0: + bra ne cmp_c1 + mov $r12 0x10 + add b32 $r12 $r11 + bra bpc_next + cmp_c1: + cmpu b8 $r10 6 + bra nc cmp_zero + mov $r12 0x14 + add b32 $r12 $r11 + bra bpc_next + cmp_zero: + mov $r12 0x80 + bpc_next: + st b8 D[$sp + $r8] $r12 + add b32 $r8 1 + add b32 $r11 1 + cmpu b32 $r11 $r5 + bra c bpc_loop + add b32 $r9 1 + cmpu b32 $r9 $r7 + bra c ncomp_loop + + // SRC_XCNT = (xcnt * src_cpp), or 0 if no src ref in swz (hw will hang) + mulu $r6 $r5 + st b32 D[$r0 + ctx_src_cpp] $r6 + ld b32 $r8 D[$r0 + ctx_xcnt] + mulu $r6 $r8 + bra $p2 dst_xcnt + clear b32 $r6 + + dst_xcnt: + mulu $r7 $r5 + st b32 D[$r0 + ctx_dst_cpp] $r7 + mulu $r7 $r8 + + mov $r5 0x810 + shl b32 $r5 6 + iowr I[$r5 + 0x000] $r6 + iowr I[$r5 + 0x100] $r7 + add b32 $r5 0x800 + ld b32 $r6 D[$r0 + ctx_dst_cpp] + sub b32 $r6 1 + shl b32 $r6 8 + ld b32 $r7 D[$r0 + ctx_src_cpp] + sub b32 $r7 1 + or $r6 $r7 + iowr I[$r5 + 0x000] $r6 + add b32 $r5 0x100 + ld b32 $r6 D[$sp + 0x00] + iowr I[$r5 + 0x000] $r6 + ld b32 $r6 D[$sp + 0x04] + iowr I[$r5 + 0x100] $r6 + ld b32 $r6 D[$sp + 0x08] + iowr I[$r5 + 0x200] $r6 + ld b32 $r6 D[$sp + 0x0c] + iowr I[$r5 + 0x300] $r6 + add b32 $r5 0x400 + ld b32 $r6 D[$r0 + ctx_swz_const0] + iowr I[$r5 + 0x000] $r6 + ld b32 $r6 D[$r0 + ctx_swz_const1] + iowr I[$r5 + 0x100] $r6 + add $sp 0x10 + ret + +// Setup to handle a tiled surface +// +// Calculates a number of parameters the hardware requires in order +// to correctly handle tiling. +// +// Offset calculation is performed as follows (Tp/Th/Td from TILE_MODE): +// nTx = round_up(w * cpp, 1 << Tp) >> Tp +// nTy = round_up(h, 1 << Th) >> Th +// Txo = (x * cpp) & ((1 << Tp) - 1) +// Tx = (x * cpp) >> Tp +// Tyo = y & ((1 << Th) - 1) +// Ty = y >> Th +// Tzo = z & ((1 << Td) - 1) +// Tz = z >> Td +// +// off = (Tzo << Tp << Th) + (Tyo << Tp) + Txo +// off += ((Tz * nTy * nTx)) + (Ty * nTx) + Tx) << Td << Th << Tp; +// +// Inputs: +// $r4: hw command (0x104800) +// $r5: ctx offset adjustment for src/dst selection +// $p2: set if dst surface +// +cmd_exec_set_surface_tiled: + // translate TILE_MODE into Tp, Th, Td shift values + ld b32 $r7 D[$r5 + ctx_src_tile_mode] + extr $r9 $r7 8:11 + extr $r8 $r7 4:7 +ifdef(`NVA3', + add b32 $r8 2 +, + add b32 $r8 3 +) + extr $r7 $r7 0:3 + cmp b32 $r7 0xe + bra ne xtile64 + mov $r7 4 + bra xtileok + xtile64: + xbit $r7 $flags $p2 + add b32 $r7 17 + bset $r4 $r7 + mov $r7 6 + xtileok: + + // Op = (x * cpp) & ((1 << Tp) - 1) + // Tx = (x * cpp) >> Tp + ld b32 $r10 D[$r5 + ctx_src_xoff] + ld b32 $r11 D[$r5 + ctx_src_cpp] + mulu $r10 $r11 + mov $r11 1 + shl b32 $r11 $r7 + sub b32 $r11 1 + and $r12 $r10 $r11 + shr b32 $r10 $r7 + + // Tyo = y & ((1 << Th) - 1) + // Ty = y >> Th + ld b32 $r13 D[$r5 + ctx_src_yoff] + mov $r14 1 + shl b32 $r14 $r8 + sub b32 $r14 1 + and $r11 $r13 $r14 + shr b32 $r13 $r8 + + // YTILE = ((1 << Th) << 12) | ((1 << Th) - Tyo) + add b32 $r14 1 + shl b32 $r15 $r14 12 + sub b32 $r14 $r11 + or $r15 $r14 + xbit $r6 $flags $p2 + add b32 $r6 0x208 + shl b32 $r6 8 + iowr I[$r6 + 0x000] $r15 + + // Op += Tyo << Tp + shl b32 $r11 $r7 + add b32 $r12 $r11 + + // nTx = ((w * cpp) + ((1 << Tp) - 1) >> Tp) + ld b32 $r15 D[$r5 + ctx_src_xsize] + ld b32 $r11 D[$r5 + ctx_src_cpp] + mulu $r15 $r11 + mov $r11 1 + shl b32 $r11 $r7 + sub b32 $r11 1 + add b32 $r15 $r11 + shr b32 $r15 $r7 + push $r15 + + // nTy = (h + ((1 << Th) - 1)) >> Th + ld b32 $r15 D[$r5 + ctx_src_ysize] + mov $r11 1 + shl b32 $r11 $r8 + sub b32 $r11 1 + add b32 $r15 $r11 + shr b32 $r15 $r8 + push $r15 + + // Tys = Tp + Th + // CFG_YZ_TILE_SIZE = ((1 << Th) >> 2) << Td + add b32 $r7 $r8 + sub b32 $r8 2 + mov $r11 1 + shl b32 $r11 $r8 + shl b32 $r11 $r9 + + // Tzo = z & ((1 << Td) - 1) + // Tz = z >> Td + // Op += Tzo << Tys + // Ts = Tys + Td + ld b32 $r8 D[$r5 + ctx_src_zoff] + mov $r14 1 + shl b32 $r14 $r9 + sub b32 $r14 1 + and $r15 $r8 $r14 + shl b32 $r15 $r7 + add b32 $r12 $r15 + add b32 $r7 $r9 + shr b32 $r8 $r9 + + // Ot = ((Tz * nTy * nTx) + (Ty * nTx) + Tx) << Ts + pop $r15 + pop $r9 + mulu $r13 $r9 + add b32 $r10 $r13 + mulu $r8 $r9 + mulu $r8 $r15 + add b32 $r10 $r8 + shl b32 $r10 $r7 + + // PITCH = (nTx - 1) << Ts + sub b32 $r9 1 + shl b32 $r9 $r7 + iowr I[$r6 + 0x200] $r9 + + // SRC_ADDRESS_LOW = (Ot + Op) & 0xffffffff + // CFG_ADDRESS_HIGH |= ((Ot + Op) >> 32) << 16 + ld b32 $r7 D[$r5 + ctx_src_address_low] + ld b32 $r8 D[$r5 + ctx_src_address_high] + add b32 $r10 $r12 + add b32 $r7 $r10 + adc b32 $r8 0 + shl b32 $r8 16 + or $r8 $r11 + sub b32 $r6 0x600 + iowr I[$r6 + 0x000] $r7 + add b32 $r6 0x400 + iowr I[$r6 + 0x000] $r8 + ret + +// Setup to handle a linear surface +// +// Nothing to see here.. Sets ADDRESS and PITCH, pretty non-exciting +// +cmd_exec_set_surface_linear: + xbit $r6 $flags $p2 + add b32 $r6 0x202 + shl b32 $r6 8 + ld b32 $r7 D[$r5 + ctx_src_address_low] + iowr I[$r6 + 0x000] $r7 + add b32 $r6 0x400 + ld b32 $r7 D[$r5 + ctx_src_address_high] + shl b32 $r7 16 + iowr I[$r6 + 0x000] $r7 + add b32 $r6 0x400 + ld b32 $r7 D[$r5 + ctx_src_pitch] + iowr I[$r6 + 0x000] $r7 + ret + +// wait for regs to be available for use +cmd_exec_wait: + push $r0 + push $r1 + mov $r0 0x800 + shl b32 $r0 6 + loop: + iord $r1 I[$r0] + and $r1 1 + bra ne loop + pop $r1 + pop $r0 + ret + +cmd_exec_query: + // if QUERY_SHORT not set, write out { -, 0, TIME_LO, TIME_HI } + xbit $r4 $r3 13 + bra ne query_counter + call cmd_exec_wait + mov $r4 0x80c + shl b32 $r4 6 + ld b32 $r5 D[$r0 + ctx_query_address_low] + add b32 $r5 4 + iowr I[$r4 + 0x000] $r5 + iowr I[$r4 + 0x100] $r0 + mov $r5 0xc + iowr I[$r4 + 0x200] $r5 + add b32 $r4 0x400 + ld b32 $r5 D[$r0 + ctx_query_address_high] + shl b32 $r5 16 + iowr I[$r4 + 0x000] $r5 + add b32 $r4 0x500 + mov $r5 0x00000b00 + sethi $r5 0x00010000 + iowr I[$r4 + 0x000] $r5 + mov $r5 0x00004040 + shl b32 $r5 1 + sethi $r5 0x80800000 + iowr I[$r4 + 0x100] $r5 + mov $r5 0x00001110 + sethi $r5 0x13120000 + iowr I[$r4 + 0x200] $r5 + mov $r5 0x00001514 + sethi $r5 0x17160000 + iowr I[$r4 + 0x300] $r5 + mov $r5 0x00002601 + sethi $r5 0x00010000 + mov $r4 0x800 + shl b32 $r4 6 + iowr I[$r4 + 0x000] $r5 + + // write COUNTER + query_counter: + call cmd_exec_wait + mov $r4 0x80c + shl b32 $r4 6 + ld b32 $r5 D[$r0 + ctx_query_address_low] + iowr I[$r4 + 0x000] $r5 + iowr I[$r4 + 0x100] $r0 + mov $r5 0x4 + iowr I[$r4 + 0x200] $r5 + add b32 $r4 0x400 + ld b32 $r5 D[$r0 + ctx_query_address_high] + shl b32 $r5 16 + iowr I[$r4 + 0x000] $r5 + add b32 $r4 0x500 + mov $r5 0x00000300 + iowr I[$r4 + 0x000] $r5 + mov $r5 0x00001110 + sethi $r5 0x13120000 + iowr I[$r4 + 0x100] $r5 + ld b32 $r5 D[$r0 + ctx_query_counter] + add b32 $r4 0x500 + iowr I[$r4 + 0x000] $r5 + mov $r5 0x00002601 + sethi $r5 0x00010000 + mov $r4 0x800 + shl b32 $r4 6 + iowr I[$r4 + 0x000] $r5 + ret + +// Execute a copy operation +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// 000002000 QUERY_SHORT +// 000001000 QUERY +// 000000100 DST_LINEAR +// 000000010 SRC_LINEAR +// 000000001 FORMAT +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_exec: + call cmd_exec_wait + + // if format requested, call function to calculate it, otherwise + // fill in cpp/xcnt for both surfaces as if (cpp == 1) + xbit $r15 $r3 0 + bra e cmd_exec_no_format + call cmd_exec_set_format + mov $r4 0x200 + bra cmd_exec_init_src_surface + cmd_exec_no_format: + mov $r6 0x810 + shl b32 $r6 6 + mov $r7 1 + st b32 D[$r0 + ctx_src_cpp] $r7 + st b32 D[$r0 + ctx_dst_cpp] $r7 + ld b32 $r7 D[$r0 + ctx_xcnt] + iowr I[$r6 + 0x000] $r7 + iowr I[$r6 + 0x100] $r7 + clear b32 $r4 + + cmd_exec_init_src_surface: + bclr $flags $p2 + clear b32 $r5 + xbit $r15 $r3 4 + bra e src_tiled + call cmd_exec_set_surface_linear + bra cmd_exec_init_dst_surface + src_tiled: + call cmd_exec_set_surface_tiled + bset $r4 7 + + cmd_exec_init_dst_surface: + bset $flags $p2 + mov $r5 ctx_dst_address_high - ctx_src_address_high + xbit $r15 $r3 8 + bra e dst_tiled + call cmd_exec_set_surface_linear + bra cmd_exec_kick + dst_tiled: + call cmd_exec_set_surface_tiled + bset $r4 8 + + cmd_exec_kick: + mov $r5 0x800 + shl b32 $r5 6 + ld b32 $r6 D[$r0 + ctx_ycnt] + iowr I[$r5 + 0x100] $r6 + mov $r6 0x0041 + // SRC_TARGET = 1, DST_TARGET = 2 + sethi $r6 0x44000000 + or $r4 $r6 + iowr I[$r5] $r4 + + // if requested, queue up a QUERY write after the copy has completed + xbit $r15 $r3 12 + bra e cmd_exec_done + call cmd_exec_query + + cmd_exec_done: + ret + +// Flush write cache +// +// Inputs: +// $r1: irqh state +// $r2: hostirq state +// $r3: data +// $r4: dispatch table entry +// Outputs: +// $r1: irqh state +// $p1: set on error +// $r2: hostirq state +// $r3: data +cmd_wrcache_flush: + mov $r2 0x2200 + clear b32 $r3 + sethi $r3 0x10000 + iowr I[$r2] $r3 + ret + +.align 0x100 diff --git a/drivers/gpu/drm/nouveau/nva3_copy.fuc.h b/drivers/gpu/drm/nouveau/nva3_copy.fuc.h new file mode 100644 index 00000000000..2731de22ebe --- /dev/null +++ b/drivers/gpu/drm/nouveau/nva3_copy.fuc.h @@ -0,0 +1,534 @@ +uint32_t nva3_pcopy_data[] = { + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00010000, + 0x00000000, + 0x00000000, + 0x00010040, + 0x00010160, + 0x00000000, + 0x00010050, + 0x00010162, + 0x00000000, + 0x00030060, + 0x00010170, + 0x00000000, + 0x00010170, + 0x00000000, + 0x00010170, + 0x00000000, + 0x00070080, + 0x00000028, + 0xfffff000, + 0x0000002c, + 0xfff80000, + 0x00000030, + 0xffffe000, + 0x00000034, + 0xfffff800, + 0x00000038, + 0xfffff000, + 0x0000003c, + 0xfff80000, + 0x00000040, + 0xffffe000, + 0x00070088, + 0x00000054, + 0xfffff000, + 0x00000058, + 0xfff80000, + 0x0000005c, + 0xffffe000, + 0x00000060, + 0xfffff800, + 0x00000064, + 0xfffff000, + 0x00000068, + 0xfff80000, + 0x0000006c, + 0xffffe000, + 0x000200c0, + 0x00010492, + 0x00000000, + 0x0001051b, + 0x00000000, + 0x000e00c3, + 0x0000001c, + 0xffffff00, + 0x00000020, + 0x0000000f, + 0x00000048, + 0xffffff00, + 0x0000004c, + 0x0000000f, + 0x00000024, + 0xfff80000, + 0x00000050, + 0xfff80000, + 0x00000080, + 0xffff0000, + 0x00000084, + 0xffffe000, + 0x00000074, + 0xfccc0000, + 0x00000078, + 0x00000000, + 0x0000007c, + 0x00000000, + 0x00000010, + 0xffffff00, + 0x00000014, + 0x00000000, + 0x00000018, + 0x00000000, + 0x00000800, +}; + +uint32_t nva3_pcopy_code[] = { + 0x04fe04bd, + 0x3517f000, + 0xf10010fe, + 0xf1040017, + 0xf0fff327, + 0x22d00023, + 0x0c25f0c0, + 0xf40012d0, + 0x17f11031, + 0x27f01200, + 0x0012d003, + 0xf40031f4, + 0x0ef40028, + 0x8001cffd, + 0xf40812c4, + 0x21f4060b, + 0x0412c472, + 0xf4060bf4, + 0x11c4c321, + 0x4001d00c, + 0x47f101f8, + 0x4bfe7700, + 0x0007fe00, + 0xf00204b9, + 0x01f40643, + 0x0604fa09, + 0xfa060ef4, + 0x03f80504, + 0x27f100f8, + 0x23cf1400, + 0x1e3fc800, + 0xf4170bf4, + 0x21f40132, + 0x1e3af052, + 0xf00023d0, + 0x24d00147, + 0xcf00f880, + 0x3dc84023, + 0x220bf41e, + 0xf40131f4, + 0x57f05221, + 0x0367f004, + 0xa07856bc, + 0xb6018068, + 0x87d00884, + 0x0162b600, + 0xf0f018f4, + 0x23d00237, + 0xf100f880, + 0xcf190037, + 0x33cf4032, + 0xff24e400, + 0x1024b607, + 0x010057f1, + 0x74bd64bd, + 0x58005658, + 0x50b60157, + 0x0446b804, + 0xbb4d08f4, + 0x47b80076, + 0x0f08f404, + 0xb60276bb, + 0x57bb0374, + 0xdf0ef400, + 0xb60246bb, + 0x45bb0344, + 0x01459800, + 0xb00453fd, + 0x1bf40054, + 0x00455820, + 0xb0014658, + 0x1bf40064, + 0x00538009, + 0xf4300ef4, + 0x55f90132, + 0xf40c01f4, + 0x25f0250e, + 0x0125f002, + 0x100047f1, + 0xd00042d0, + 0x27f04043, + 0x0002d040, + 0xf08002cf, + 0x24b04024, + 0xf71bf400, + 0x1d0027f1, + 0xd00137f0, + 0x00f80023, + 0x27f100f8, + 0x34bd2200, + 0xd00233f0, + 0x00f80023, + 0x012842b7, + 0xf00145b6, + 0x43801e39, + 0x0040b701, + 0x0644b606, + 0xf80043d0, + 0xf030f400, + 0xb00001b0, + 0x01b00101, + 0x0301b002, + 0xc71d0498, + 0x50b63045, + 0x3446c701, + 0xc70160b6, + 0x70b63847, + 0x0232f401, + 0x94bd84bd, + 0xb60f4ac4, + 0xb4bd0445, + 0xf404a430, + 0xa5ff0f18, + 0x00cbbbc0, + 0xf40231f4, + 0x1bf4220e, + 0x10c7f00c, + 0xf400cbbb, + 0xa430160e, + 0x0c18f406, + 0xbb14c7f0, + 0x0ef400cb, + 0x80c7f107, + 0x01c83800, + 0xb60180b6, + 0xb5b801b0, + 0xc308f404, + 0xb80190b6, + 0x08f40497, + 0x0065fdb2, + 0x98110680, + 0x68fd2008, + 0x0502f400, + 0x75fd64bd, + 0x1c078000, + 0xf10078fd, + 0xb6081057, + 0x56d00654, + 0x4057d000, + 0x080050b7, + 0xb61c0698, + 0x64b60162, + 0x11079808, + 0xfd0172b6, + 0x56d00567, + 0x0050b700, + 0x0060b401, + 0xb40056d0, + 0x56d00160, + 0x0260b440, + 0xb48056d0, + 0x56d00360, + 0x0050b7c0, + 0x1e069804, + 0x980056d0, + 0x56d01f06, + 0x1030f440, + 0x579800f8, + 0x6879c70a, + 0xb66478c7, + 0x77c70280, + 0x0e76b060, + 0xf0091bf4, + 0x0ef40477, + 0x027cf00f, + 0xfd1170b6, + 0x77f00947, + 0x0f5a9806, + 0xfd115b98, + 0xb7f000ab, + 0x04b7bb01, + 0xff01b2b6, + 0xa7bbc4ab, + 0x105d9805, + 0xbb01e7f0, + 0xe2b604e8, + 0xb4deff01, + 0xb605d8bb, + 0xef9401e0, + 0x02ebbb0c, + 0xf005fefd, + 0x60b7026c, + 0x64b60208, + 0x006fd008, + 0xbb04b7bb, + 0x5f9800cb, + 0x115b980b, + 0xf000fbfd, + 0xb7bb01b7, + 0x01b2b604, + 0xbb00fbbb, + 0xf0f905f7, + 0xf00c5f98, + 0xb8bb01b7, + 0x01b2b604, + 0xbb00fbbb, + 0xf0f905f8, + 0xb60078bb, + 0xb7f00282, + 0x04b8bb01, + 0x9804b9bb, + 0xe7f00e58, + 0x04e9bb01, + 0xff01e2b6, + 0xf7bbf48e, + 0x00cfbb04, + 0xbb0079bb, + 0xf0fc0589, + 0xd9fd90fc, + 0x00adbb00, + 0xfd0089fd, + 0xa8bb008f, + 0x04a7bb00, + 0xbb0192b6, + 0x69d00497, + 0x08579880, + 0xbb075898, + 0x7abb00ac, + 0x0081b600, + 0xfd1084b6, + 0x62b7058b, + 0x67d00600, + 0x0060b700, + 0x0068d004, + 0x6cf000f8, + 0x0260b702, + 0x0864b602, + 0xd0085798, + 0x60b70067, + 0x57980400, + 0x1074b607, + 0xb70067d0, + 0x98040060, + 0x67d00957, + 0xf900f800, + 0xf110f900, + 0xb6080007, + 0x01cf0604, + 0x0114f000, + 0xfcfa1bf4, + 0xf800fc10, + 0x0d34c800, + 0xf5701bf4, + 0xf103ab21, + 0xb6080c47, + 0x05980644, + 0x0450b605, + 0xd00045d0, + 0x57f04040, + 0x8045d00c, + 0x040040b7, + 0xb6040598, + 0x45d01054, + 0x0040b700, + 0x0057f105, + 0x0153f00b, + 0xf10045d0, + 0xb6404057, + 0x53f10154, + 0x45d08080, + 0x1057f140, + 0x1253f111, + 0x8045d013, + 0x151457f1, + 0x171653f1, + 0xf1c045d0, + 0xf0260157, + 0x47f10153, + 0x44b60800, + 0x0045d006, + 0x03ab21f5, + 0x080c47f1, + 0x980644b6, + 0x45d00505, + 0x4040d000, + 0xd00457f0, + 0x40b78045, + 0x05980400, + 0x1054b604, + 0xb70045d0, + 0xf1050040, + 0xd0030057, + 0x57f10045, + 0x53f11110, + 0x45d01312, + 0x06059840, + 0x050040b7, + 0xf10045d0, + 0xf0260157, + 0x47f10153, + 0x44b60800, + 0x0045d006, + 0x21f500f8, + 0x3fc803ab, + 0x0e0bf400, + 0x018921f5, + 0x020047f1, + 0xf11e0ef4, + 0xb6081067, + 0x77f00664, + 0x11078001, + 0x981c0780, + 0x67d02007, + 0x4067d000, + 0x32f444bd, + 0xc854bd02, + 0x0bf4043f, + 0x8221f50a, + 0x0a0ef403, + 0x027621f5, + 0xf40749f0, + 0x57f00231, + 0x083fc82c, + 0xf50a0bf4, + 0xf4038221, + 0x21f50a0e, + 0x49f00276, + 0x0057f108, + 0x0654b608, + 0xd0210698, + 0x67f04056, + 0x0063f141, + 0x0546fd44, + 0xc80054d0, + 0x0bf40c3f, + 0xc521f507, + 0xf100f803, + 0xbd220027, + 0x0133f034, + 0xf80023d0, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, +}; diff --git a/drivers/gpu/drm/nouveau/nva3_pm.c b/drivers/gpu/drm/nouveau/nva3_pm.c index dbbafed3640..e4b2b9e934b 100644 --- a/drivers/gpu/drm/nouveau/nva3_pm.c +++ b/drivers/gpu/drm/nouveau/nva3_pm.c @@ -27,32 +27,74 @@ #include "nouveau_bios.h" #include "nouveau_pm.h" -/*XXX: boards using limits 0x40 need fixing, the register layout - * is correct here, but, there's some other funny magic - * that modifies things, so it's not likely we'll set/read - * the correct timings yet.. working on it... +/* This is actually a lot more complex than it appears here, but hopefully + * this should be able to deal with what the VBIOS leaves for us.. + * + * If not, well, I'll jump off that bridge when I come to it. */ struct nva3_pm_state { - struct pll_lims pll; - int N, M, P; + enum pll_types type; + u32 src0; + u32 src1; + u32 ctrl; + u32 coef; + u32 old_pnm; + u32 new_pnm; + u32 new_div; }; +static int +nva3_pm_pll_offset(u32 id) +{ + static const u32 pll_map[] = { + 0x00, PLL_CORE, + 0x01, PLL_SHADER, + 0x02, PLL_MEMORY, + 0x00, 0x00 + }; + const u32 *map = pll_map; + + while (map[1]) { + if (id == map[1]) + return map[0]; + map += 2; + } + + return -ENOENT; +} + int nva3_pm_clock_get(struct drm_device *dev, u32 id) { + u32 src0, src1, ctrl, coef; struct pll_lims pll; - int P, N, M, ret; - u32 reg; + int ret, off; + int P, N, M; ret = get_pll_limits(dev, id, &pll); if (ret) return ret; - reg = nv_rd32(dev, pll.reg + 4); - P = (reg & 0x003f0000) >> 16; - N = (reg & 0x0000ff00) >> 8; - M = (reg & 0x000000ff); + off = nva3_pm_pll_offset(id); + if (off < 0) + return off; + + src0 = nv_rd32(dev, 0x4120 + (off * 4)); + src1 = nv_rd32(dev, 0x4160 + (off * 4)); + ctrl = nv_rd32(dev, pll.reg + 0); + coef = nv_rd32(dev, pll.reg + 4); + NV_DEBUG(dev, "PLL %02x: 0x%08x 0x%08x 0x%08x 0x%08x\n", + id, src0, src1, ctrl, coef); + + if (ctrl & 0x00000008) { + u32 div = ((src1 & 0x003c0000) >> 18) + 1; + return (pll.refclk * 2) / div; + } + + P = (coef & 0x003f0000) >> 16; + N = (coef & 0x0000ff00) >> 8; + M = (coef & 0x000000ff); return pll.refclk * N / M / P; } @@ -60,36 +102,103 @@ void * nva3_pm_clock_pre(struct drm_device *dev, struct nouveau_pm_level *perflvl, u32 id, int khz) { - struct nva3_pm_state *state; - int dummy, ret; + struct nva3_pm_state *pll; + struct pll_lims limits; + int N, M, P, diff; + int ret, off; + + ret = get_pll_limits(dev, id, &limits); + if (ret < 0) + return (ret == -ENOENT) ? NULL : ERR_PTR(ret); + + off = nva3_pm_pll_offset(id); + if (id < 0) + return ERR_PTR(-EINVAL); - state = kzalloc(sizeof(*state), GFP_KERNEL); - if (!state) + + pll = kzalloc(sizeof(*pll), GFP_KERNEL); + if (!pll) return ERR_PTR(-ENOMEM); + pll->type = id; + pll->src0 = 0x004120 + (off * 4); + pll->src1 = 0x004160 + (off * 4); + pll->ctrl = limits.reg + 0; + pll->coef = limits.reg + 4; - ret = get_pll_limits(dev, id, &state->pll); - if (ret < 0) { - kfree(state); - return (ret == -ENOENT) ? NULL : ERR_PTR(ret); + /* If target clock is within [-2, 3) MHz of a divisor, we'll + * use that instead of calculating MNP values + */ + pll->new_div = min((limits.refclk * 2) / (khz - 2999), 16); + if (pll->new_div) { + diff = khz - ((limits.refclk * 2) / pll->new_div); + if (diff < -2000 || diff >= 3000) + pll->new_div = 0; } - ret = nv50_calc_pll2(dev, &state->pll, khz, &state->N, &dummy, - &state->M, &state->P); - if (ret < 0) { - kfree(state); - return ERR_PTR(ret); + if (!pll->new_div) { + ret = nva3_calc_pll(dev, &limits, khz, &N, NULL, &M, &P); + if (ret < 0) + return ERR_PTR(ret); + + pll->new_pnm = (P << 16) | (N << 8) | M; + pll->new_div = 2 - 1; + } else { + pll->new_pnm = 0; + pll->new_div--; } - return state; + if ((nv_rd32(dev, pll->src1) & 0x00000101) != 0x00000101) + pll->old_pnm = nv_rd32(dev, pll->coef); + return pll; } void nva3_pm_clock_set(struct drm_device *dev, void *pre_state) { - struct nva3_pm_state *state = pre_state; - u32 reg = state->pll.reg; + struct nva3_pm_state *pll = pre_state; + u32 ctrl = 0; + + /* For the memory clock, NVIDIA will build a "script" describing + * the reclocking process and ask PDAEMON to execute it. + */ + if (pll->type == PLL_MEMORY) { + nv_wr32(dev, 0x100210, 0); + nv_wr32(dev, 0x1002dc, 1); + nv_wr32(dev, 0x004018, 0x00001000); + ctrl = 0x18000100; + } + + if (pll->old_pnm || !pll->new_pnm) { + nv_mask(dev, pll->src1, 0x003c0101, 0x00000101 | + (pll->new_div << 18)); + nv_wr32(dev, pll->ctrl, 0x0001001d | ctrl); + nv_mask(dev, pll->ctrl, 0x00000001, 0x00000000); + } + + if (pll->new_pnm) { + nv_mask(dev, pll->src0, 0x00000101, 0x00000101); + nv_wr32(dev, pll->coef, pll->new_pnm); + nv_wr32(dev, pll->ctrl, 0x0001001d | ctrl); + nv_mask(dev, pll->ctrl, 0x00000010, 0x00000000); + nv_mask(dev, pll->ctrl, 0x00020010, 0x00020010); + nv_wr32(dev, pll->ctrl, 0x00010015 | ctrl); + nv_mask(dev, pll->src1, 0x00000100, 0x00000000); + nv_mask(dev, pll->src1, 0x00000001, 0x00000000); + if (pll->type == PLL_MEMORY) + nv_wr32(dev, 0x4018, 0x10005000); + } else { + nv_mask(dev, pll->ctrl, 0x00000001, 0x00000000); + nv_mask(dev, pll->src0, 0x00000100, 0x00000000); + nv_mask(dev, pll->src0, 0x00000001, 0x00000000); + if (pll->type == PLL_MEMORY) + nv_wr32(dev, 0x4018, 0x1000d000); + } + + if (pll->type == PLL_MEMORY) { + nv_wr32(dev, 0x1002dc, 0); + nv_wr32(dev, 0x100210, 0x80000000); + } - nv_wr32(dev, reg + 4, (state->P << 16) | (state->N << 8) | state->M); - kfree(state); + kfree(pll); } diff --git a/drivers/gpu/drm/nouveau/nvc0_copy.c b/drivers/gpu/drm/nouveau/nvc0_copy.c new file mode 100644 index 00000000000..dddf006f6d8 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_copy.c @@ -0,0 +1,243 @@ +/* + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +#include <linux/firmware.h> +#include "drmP.h" +#include "nouveau_drv.h" +#include "nouveau_util.h" +#include "nouveau_vm.h" +#include "nouveau_ramht.h" +#include "nvc0_copy.fuc.h" + +struct nvc0_copy_engine { + struct nouveau_exec_engine base; + u32 irq; + u32 pmc; + u32 fuc; + u32 ctx; +}; + +static int +nvc0_copy_context_new(struct nouveau_channel *chan, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(chan->dev, engine); + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_gpuobj *ramin = chan->ramin; + struct nouveau_gpuobj *ctx = NULL; + int ret; + + ret = nouveau_gpuobj_new(dev, chan, 256, 256, + NVOBJ_FLAG_VM | NVOBJ_FLAG_VM_USER | + NVOBJ_FLAG_ZERO_ALLOC, &ctx); + if (ret) + return ret; + + nv_wo32(ramin, pcopy->ctx + 0, lower_32_bits(ctx->linst)); + nv_wo32(ramin, pcopy->ctx + 4, upper_32_bits(ctx->linst)); + dev_priv->engine.instmem.flush(dev); + + chan->engctx[engine] = ctx; + return 0; +} + +static int +nvc0_copy_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) +{ + return 0; +} + +static void +nvc0_copy_context_del(struct nouveau_channel *chan, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(chan->dev, engine); + struct nouveau_gpuobj *ctx = chan->engctx[engine]; + struct drm_device *dev = chan->dev; + u32 inst; + + inst = (chan->ramin->vinst >> 12); + inst |= 0x40000000; + + /* disable fifo access */ + nv_wr32(dev, pcopy->fuc + 0x048, 0x00000000); + /* mark channel as unloaded if it's currently active */ + if (nv_rd32(dev, pcopy->fuc + 0x050) == inst) + nv_mask(dev, pcopy->fuc + 0x050, 0x40000000, 0x00000000); + /* mark next channel as invalid if it's about to be loaded */ + if (nv_rd32(dev, pcopy->fuc + 0x054) == inst) + nv_mask(dev, pcopy->fuc + 0x054, 0x40000000, 0x00000000); + /* restore fifo access */ + nv_wr32(dev, pcopy->fuc + 0x048, 0x00000003); + + nv_wo32(chan->ramin, pcopy->ctx + 0, 0x00000000); + nv_wo32(chan->ramin, pcopy->ctx + 4, 0x00000000); + nouveau_gpuobj_ref(NULL, &ctx); + + chan->engctx[engine] = ctx; +} + +static int +nvc0_copy_init(struct drm_device *dev, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(dev, engine); + int i; + + nv_mask(dev, 0x000200, pcopy->pmc, 0x00000000); + nv_mask(dev, 0x000200, pcopy->pmc, pcopy->pmc); + nv_wr32(dev, pcopy->fuc + 0x014, 0xffffffff); + + nv_wr32(dev, pcopy->fuc + 0x1c0, 0x01000000); + for (i = 0; i < sizeof(nvc0_pcopy_data) / 4; i++) + nv_wr32(dev, pcopy->fuc + 0x1c4, nvc0_pcopy_data[i]); + + nv_wr32(dev, pcopy->fuc + 0x180, 0x01000000); + for (i = 0; i < sizeof(nvc0_pcopy_code) / 4; i++) { + if ((i & 0x3f) == 0) + nv_wr32(dev, pcopy->fuc + 0x188, i >> 6); + nv_wr32(dev, pcopy->fuc + 0x184, nvc0_pcopy_code[i]); + } + + nv_wr32(dev, pcopy->fuc + 0x084, engine - NVOBJ_ENGINE_COPY0); + nv_wr32(dev, pcopy->fuc + 0x10c, 0x00000000); + nv_wr32(dev, pcopy->fuc + 0x104, 0x00000000); /* ENTRY */ + nv_wr32(dev, pcopy->fuc + 0x100, 0x00000002); /* TRIGGER */ + return 0; +} + +static int +nvc0_copy_fini(struct drm_device *dev, int engine, bool suspend) +{ + struct nvc0_copy_engine *pcopy = nv_engine(dev, engine); + + nv_mask(dev, pcopy->fuc + 0x048, 0x00000003, 0x00000000); + + /* trigger fuc context unload */ + nv_wait(dev, pcopy->fuc + 0x008, 0x0000000c, 0x00000000); + nv_mask(dev, pcopy->fuc + 0x054, 0x40000000, 0x00000000); + nv_wr32(dev, pcopy->fuc + 0x000, 0x00000008); + nv_wait(dev, pcopy->fuc + 0x008, 0x00000008, 0x00000000); + + nv_wr32(dev, pcopy->fuc + 0x014, 0xffffffff); + return 0; +} + +static struct nouveau_enum nvc0_copy_isr_error_name[] = { + { 0x0001, "ILLEGAL_MTHD" }, + { 0x0002, "INVALID_ENUM" }, + { 0x0003, "INVALID_BITFIELD" }, + {} +}; + +static void +nvc0_copy_isr(struct drm_device *dev, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(dev, engine); + u32 disp = nv_rd32(dev, pcopy->fuc + 0x01c); + u32 stat = nv_rd32(dev, pcopy->fuc + 0x008) & disp & ~(disp >> 16); + u64 inst = (u64)(nv_rd32(dev, pcopy->fuc + 0x050) & 0x0fffffff) << 12; + u32 chid = nvc0_graph_isr_chid(dev, inst); + u32 ssta = nv_rd32(dev, pcopy->fuc + 0x040) & 0x0000ffff; + u32 addr = nv_rd32(dev, pcopy->fuc + 0x040) >> 16; + u32 mthd = (addr & 0x07ff) << 2; + u32 subc = (addr & 0x3800) >> 11; + u32 data = nv_rd32(dev, pcopy->fuc + 0x044); + + if (stat & 0x00000040) { + NV_INFO(dev, "PCOPY: DISPATCH_ERROR ["); + nouveau_enum_print(nvc0_copy_isr_error_name, ssta); + printk("] ch %d [0x%010llx] subc %d mthd 0x%04x data 0x%08x\n", + chid, inst, subc, mthd, data); + nv_wr32(dev, pcopy->fuc + 0x004, 0x00000040); + stat &= ~0x00000040; + } + + if (stat) { + NV_INFO(dev, "PCOPY: unhandled intr 0x%08x\n", stat); + nv_wr32(dev, pcopy->fuc + 0x004, stat); + } +} + +static void +nvc0_copy_isr_0(struct drm_device *dev) +{ + nvc0_copy_isr(dev, NVOBJ_ENGINE_COPY0); +} + +static void +nvc0_copy_isr_1(struct drm_device *dev) +{ + nvc0_copy_isr(dev, NVOBJ_ENGINE_COPY1); +} + +static void +nvc0_copy_destroy(struct drm_device *dev, int engine) +{ + struct nvc0_copy_engine *pcopy = nv_engine(dev, engine); + + nouveau_irq_unregister(dev, pcopy->irq); + + if (engine == NVOBJ_ENGINE_COPY0) + NVOBJ_ENGINE_DEL(dev, COPY0); + else + NVOBJ_ENGINE_DEL(dev, COPY1); + kfree(pcopy); +} + +int +nvc0_copy_create(struct drm_device *dev, int engine) +{ + struct nvc0_copy_engine *pcopy; + + pcopy = kzalloc(sizeof(*pcopy), GFP_KERNEL); + if (!pcopy) + return -ENOMEM; + + pcopy->base.destroy = nvc0_copy_destroy; + pcopy->base.init = nvc0_copy_init; + pcopy->base.fini = nvc0_copy_fini; + pcopy->base.context_new = nvc0_copy_context_new; + pcopy->base.context_del = nvc0_copy_context_del; + pcopy->base.object_new = nvc0_copy_object_new; + + if (engine == 0) { + pcopy->irq = 5; + pcopy->pmc = 0x00000040; + pcopy->fuc = 0x104000; + pcopy->ctx = 0x0230; + nouveau_irq_register(dev, pcopy->irq, nvc0_copy_isr_0); + NVOBJ_ENGINE_ADD(dev, COPY0, &pcopy->base); + NVOBJ_CLASS(dev, 0x90b5, COPY0); + } else { + pcopy->irq = 6; + pcopy->pmc = 0x00000080; + pcopy->fuc = 0x105000; + pcopy->ctx = 0x0240; + nouveau_irq_register(dev, pcopy->irq, nvc0_copy_isr_1); + NVOBJ_ENGINE_ADD(dev, COPY1, &pcopy->base); + NVOBJ_CLASS(dev, 0x90b8, COPY1); + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nvc0_copy.fuc.h b/drivers/gpu/drm/nouveau/nvc0_copy.fuc.h new file mode 100644 index 00000000000..419903880e9 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_copy.fuc.h @@ -0,0 +1,527 @@ +uint32_t nvc0_pcopy_data[] = { + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00010000, + 0x00000000, + 0x00000000, + 0x00010040, + 0x0001019f, + 0x00000000, + 0x00010050, + 0x000101a1, + 0x00000000, + 0x00070080, + 0x0000001c, + 0xfffff000, + 0x00000020, + 0xfff80000, + 0x00000024, + 0xffffe000, + 0x00000028, + 0xfffff800, + 0x0000002c, + 0xfffff000, + 0x00000030, + 0xfff80000, + 0x00000034, + 0xffffe000, + 0x00070088, + 0x00000048, + 0xfffff000, + 0x0000004c, + 0xfff80000, + 0x00000050, + 0xffffe000, + 0x00000054, + 0xfffff800, + 0x00000058, + 0xfffff000, + 0x0000005c, + 0xfff80000, + 0x00000060, + 0xffffe000, + 0x000200c0, + 0x000104b8, + 0x00000000, + 0x00010541, + 0x00000000, + 0x000e00c3, + 0x00000010, + 0xffffff00, + 0x00000014, + 0x0000000f, + 0x0000003c, + 0xffffff00, + 0x00000040, + 0x0000000f, + 0x00000018, + 0xfff80000, + 0x00000044, + 0xfff80000, + 0x00000074, + 0xffff0000, + 0x00000078, + 0xffffe000, + 0x00000068, + 0xfccc0000, + 0x0000006c, + 0x00000000, + 0x00000070, + 0x00000000, + 0x00000004, + 0xffffff00, + 0x00000008, + 0x00000000, + 0x0000000c, + 0x00000000, + 0x00000800, +}; + +uint32_t nvc0_pcopy_code[] = { + 0x04fe04bd, + 0x3517f000, + 0xf10010fe, + 0xf1040017, + 0xf0fff327, + 0x22d00023, + 0x0c25f0c0, + 0xf40012d0, + 0x17f11031, + 0x27f01200, + 0x0012d003, + 0xf40031f4, + 0x0ef40028, + 0x8001cffd, + 0xf40812c4, + 0x21f4060b, + 0x0412c4ca, + 0xf5070bf4, + 0xc4010221, + 0x01d00c11, + 0xf101f840, + 0xfe770047, + 0x47f1004b, + 0x44cf2100, + 0x0144f000, + 0xb60444b6, + 0xf7f13040, + 0xf4b6061c, + 0x1457f106, + 0x00f5d101, + 0xb6043594, + 0x57fe0250, + 0x0145fe00, + 0x010052b7, + 0x00ff67f1, + 0x56fd60bd, + 0x0253f004, + 0xf80545fa, + 0x0053f003, + 0xd100e7f0, + 0x549800fe, + 0x0845b600, + 0xb6015698, + 0x46fd1864, + 0x0047fe05, + 0xf00204b9, + 0x01f40643, + 0x0604fa09, + 0xfa060ef4, + 0x03f80504, + 0x27f100f8, + 0x23cf1400, + 0x1e3fc800, + 0xf4170bf4, + 0x21f40132, + 0x1e3af053, + 0xf00023d0, + 0x24d00147, + 0xcf00f880, + 0x3dc84023, + 0x090bf41e, + 0xf40131f4, + 0x37f05321, + 0x8023d002, + 0x37f100f8, + 0x32cf1900, + 0x0033cf40, + 0x07ff24e4, + 0xf11024b6, + 0xbd010057, + 0x5874bd64, + 0x57580056, + 0x0450b601, + 0xf40446b8, + 0x76bb4d08, + 0x0447b800, + 0xbb0f08f4, + 0x74b60276, + 0x0057bb03, + 0xbbdf0ef4, + 0x44b60246, + 0x0045bb03, + 0xfd014598, + 0x54b00453, + 0x201bf400, + 0x58004558, + 0x64b00146, + 0x091bf400, + 0xf4005380, + 0x32f4300e, + 0xf455f901, + 0x0ef40c01, + 0x0225f025, + 0xf10125f0, + 0xd0100047, + 0x43d00042, + 0x4027f040, + 0xcf0002d0, + 0x24f08002, + 0x0024b040, + 0xf1f71bf4, + 0xf01d0027, + 0x23d00137, + 0xf800f800, + 0x0027f100, + 0xf034bd22, + 0x23d00233, + 0xf400f800, + 0x01b0f030, + 0x0101b000, + 0xb00201b0, + 0x04980301, + 0x3045c71a, + 0xc70150b6, + 0x60b63446, + 0x3847c701, + 0xf40170b6, + 0x84bd0232, + 0x4ac494bd, + 0x0445b60f, + 0xa430b4bd, + 0x0f18f404, + 0xbbc0a5ff, + 0x31f400cb, + 0x220ef402, + 0xf00c1bf4, + 0xcbbb10c7, + 0x160ef400, + 0xf406a430, + 0xc7f00c18, + 0x00cbbb14, + 0xf1070ef4, + 0x380080c7, + 0x80b601c8, + 0x01b0b601, + 0xf404b5b8, + 0x90b6c308, + 0x0497b801, + 0xfdb208f4, + 0x06800065, + 0x1d08980e, + 0xf40068fd, + 0x64bd0502, + 0x800075fd, + 0x78fd1907, + 0x1057f100, + 0x0654b608, + 0xd00056d0, + 0x50b74057, + 0x06980800, + 0x0162b619, + 0x980864b6, + 0x72b60e07, + 0x0567fd01, + 0xb70056d0, + 0xb4010050, + 0x56d00060, + 0x0160b400, + 0xb44056d0, + 0x56d00260, + 0x0360b480, + 0xb7c056d0, + 0x98040050, + 0x56d01b06, + 0x1c069800, + 0xf44056d0, + 0x00f81030, + 0xc7075798, + 0x78c76879, + 0x0380b664, + 0xb06077c7, + 0x1bf40e76, + 0x0477f009, + 0xf00f0ef4, + 0x70b6027c, + 0x0947fd11, + 0x980677f0, + 0x5b980c5a, + 0x00abfd0e, + 0xbb01b7f0, + 0xb2b604b7, + 0xc4abff01, + 0x9805a7bb, + 0xe7f00d5d, + 0x04e8bb01, + 0xff01e2b6, + 0xd8bbb4de, + 0x01e0b605, + 0xbb0cef94, + 0xfefd02eb, + 0x026cf005, + 0x020860b7, + 0xd00864b6, + 0xb7bb006f, + 0x00cbbb04, + 0x98085f98, + 0xfbfd0e5b, + 0x01b7f000, + 0xb604b7bb, + 0xfbbb01b2, + 0x05f7bb00, + 0x5f98f0f9, + 0x01b7f009, + 0xb604b8bb, + 0xfbbb01b2, + 0x05f8bb00, + 0x78bbf0f9, + 0x0282b600, + 0xbb01b7f0, + 0xb9bb04b8, + 0x0b589804, + 0xbb01e7f0, + 0xe2b604e9, + 0xf48eff01, + 0xbb04f7bb, + 0x79bb00cf, + 0x0589bb00, + 0x90fcf0fc, + 0xbb00d9fd, + 0x89fd00ad, + 0x008ffd00, + 0xbb00a8bb, + 0x92b604a7, + 0x0497bb01, + 0x988069d0, + 0x58980557, + 0x00acbb04, + 0xb6007abb, + 0x84b60081, + 0x058bfd10, + 0x060062b7, + 0xb70067d0, + 0xd0040060, + 0x00f80068, + 0xb7026cf0, + 0xb6020260, + 0x57980864, + 0x0067d005, + 0x040060b7, + 0xb6045798, + 0x67d01074, + 0x0060b700, + 0x06579804, + 0xf80067d0, + 0xf900f900, + 0x0007f110, + 0x0604b608, + 0xf00001cf, + 0x1bf40114, + 0xfc10fcfa, + 0xc800f800, + 0x1bf40d34, + 0xd121f570, + 0x0c47f103, + 0x0644b608, + 0xb6020598, + 0x45d00450, + 0x4040d000, + 0xd00c57f0, + 0x40b78045, + 0x05980400, + 0x1054b601, + 0xb70045d0, + 0xf1050040, + 0xf00b0057, + 0x45d00153, + 0x4057f100, + 0x0154b640, + 0x808053f1, + 0xf14045d0, + 0xf1111057, + 0xd0131253, + 0x57f18045, + 0x53f11514, + 0x45d01716, + 0x0157f1c0, + 0x0153f026, + 0x080047f1, + 0xd00644b6, + 0x21f50045, + 0x47f103d1, + 0x44b6080c, + 0x02059806, + 0xd00045d0, + 0x57f04040, + 0x8045d004, + 0x040040b7, + 0xb6010598, + 0x45d01054, + 0x0040b700, + 0x0057f105, + 0x0045d003, + 0x111057f1, + 0x131253f1, + 0x984045d0, + 0x40b70305, + 0x45d00500, + 0x0157f100, + 0x0153f026, + 0x080047f1, + 0xd00644b6, + 0x00f80045, + 0x03d121f5, + 0xf4003fc8, + 0x21f50e0b, + 0x47f101af, + 0x0ef40200, + 0x1067f11e, + 0x0664b608, + 0x800177f0, + 0x07800e07, + 0x1d079819, + 0xd00067d0, + 0x44bd4067, + 0xbd0232f4, + 0x043fc854, + 0xf50a0bf4, + 0xf403a821, + 0x21f50a0e, + 0x49f0029c, + 0x0231f407, + 0xc82c57f0, + 0x0bf4083f, + 0xa821f50a, + 0x0a0ef403, + 0x029c21f5, + 0xf10849f0, + 0xb6080057, + 0x06980654, + 0x4056d01e, + 0xf14167f0, + 0xfd440063, + 0x54d00546, + 0x0c3fc800, + 0xf5070bf4, + 0xf803eb21, + 0x0027f100, + 0xf034bd22, + 0x23d00133, + 0x0000f800, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, +}; diff --git a/drivers/gpu/drm/nouveau/nvc0_fb.c b/drivers/gpu/drm/nouveau/nvc0_fb.c index 26a996025dd..08e6b118f02 100644 --- a/drivers/gpu/drm/nouveau/nvc0_fb.c +++ b/drivers/gpu/drm/nouveau/nvc0_fb.c @@ -1,5 +1,5 @@ /* - * Copyright 2010 Red Hat Inc. + * Copyright 2011 Red Hat Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -23,16 +23,80 @@ */ #include "drmP.h" - +#include "drm.h" #include "nouveau_drv.h" +#include "nouveau_drm.h" + +struct nvc0_fb_priv { + struct page *r100c10_page; + dma_addr_t r100c10; +}; + +static void +nvc0_fb_destroy(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + struct nvc0_fb_priv *priv = pfb->priv; + + if (priv->r100c10_page) { + pci_unmap_page(dev->pdev, priv->r100c10, PAGE_SIZE, + PCI_DMA_BIDIRECTIONAL); + __free_page(priv->r100c10_page); + } + + kfree(priv); + pfb->priv = NULL; +} + +static int +nvc0_fb_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_fb_engine *pfb = &dev_priv->engine.fb; + struct nvc0_fb_priv *priv; + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + pfb->priv = priv; + + priv->r100c10_page = alloc_page(GFP_KERNEL | __GFP_ZERO); + if (!priv->r100c10_page) { + nvc0_fb_destroy(dev); + return -ENOMEM; + } + + priv->r100c10 = pci_map_page(dev->pdev, priv->r100c10_page, 0, + PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + if (pci_dma_mapping_error(dev->pdev, priv->r100c10)) { + nvc0_fb_destroy(dev); + return -EFAULT; + } + + return 0; +} int nvc0_fb_init(struct drm_device *dev) { + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvc0_fb_priv *priv; + int ret; + + if (!dev_priv->engine.fb.priv) { + ret = nvc0_fb_create(dev); + if (ret) + return ret; + } + priv = dev_priv->engine.fb.priv; + + nv_wr32(dev, 0x100c10, priv->r100c10 >> 8); return 0; } void nvc0_fb_takedown(struct drm_device *dev) { + nvc0_fb_destroy(dev); } diff --git a/drivers/gpu/drm/nouveau/nvc0_fbcon.c b/drivers/gpu/drm/nouveau/nvc0_fbcon.c index fa5d4c23438..a495e48197c 100644 --- a/drivers/gpu/drm/nouveau/nvc0_fbcon.c +++ b/drivers/gpu/drm/nouveau/nvc0_fbcon.c @@ -159,7 +159,7 @@ nvc0_fbcon_accel_init(struct fb_info *info) struct drm_device *dev = nfbdev->dev; struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_channel *chan = dev_priv->channel; - struct nouveau_bo *nvbo = nfbdev->nouveau_fb.nvbo; + struct nouveau_framebuffer *fb = &nfbdev->nouveau_fb; int ret, format; ret = nouveau_gpuobj_gr_new(chan, 0x902d, 0x902d); @@ -203,8 +203,8 @@ nvc0_fbcon_accel_init(struct fb_info *info) BEGIN_NVC0(chan, 2, NvSub2D, 0x0000, 1); OUT_RING (chan, 0x0000902d); BEGIN_NVC0(chan, 2, NvSub2D, 0x0104, 2); - OUT_RING (chan, upper_32_bits(chan->notifier_bo->bo.offset)); - OUT_RING (chan, lower_32_bits(chan->notifier_bo->bo.offset)); + OUT_RING (chan, upper_32_bits(chan->notifier_vma.offset)); + OUT_RING (chan, lower_32_bits(chan->notifier_vma.offset)); BEGIN_NVC0(chan, 2, NvSub2D, 0x0290, 1); OUT_RING (chan, 0); BEGIN_NVC0(chan, 2, NvSub2D, 0x0888, 1); @@ -249,8 +249,8 @@ nvc0_fbcon_accel_init(struct fb_info *info) OUT_RING (chan, info->fix.line_length); OUT_RING (chan, info->var.xres_virtual); OUT_RING (chan, info->var.yres_virtual); - OUT_RING (chan, upper_32_bits(nvbo->vma.offset)); - OUT_RING (chan, lower_32_bits(nvbo->vma.offset)); + OUT_RING (chan, upper_32_bits(fb->vma.offset)); + OUT_RING (chan, lower_32_bits(fb->vma.offset)); BEGIN_NVC0(chan, 2, NvSub2D, 0x0230, 10); OUT_RING (chan, format); OUT_RING (chan, 1); @@ -260,8 +260,8 @@ nvc0_fbcon_accel_init(struct fb_info *info) OUT_RING (chan, info->fix.line_length); OUT_RING (chan, info->var.xres_virtual); OUT_RING (chan, info->var.yres_virtual); - OUT_RING (chan, upper_32_bits(nvbo->vma.offset)); - OUT_RING (chan, lower_32_bits(nvbo->vma.offset)); + OUT_RING (chan, upper_32_bits(fb->vma.offset)); + OUT_RING (chan, lower_32_bits(fb->vma.offset)); FIRE_RING (chan); return 0; diff --git a/drivers/gpu/drm/nouveau/nvc0_fifo.c b/drivers/gpu/drm/nouveau/nvc0_fifo.c index 2886f2726a9..6f9f341c3e8 100644 --- a/drivers/gpu/drm/nouveau/nvc0_fifo.c +++ b/drivers/gpu/drm/nouveau/nvc0_fifo.c @@ -37,7 +37,7 @@ struct nvc0_fifo_priv { }; struct nvc0_fifo_chan { - struct nouveau_bo *user; + struct nouveau_gpuobj *user; struct nouveau_gpuobj *ramfc; }; @@ -106,7 +106,7 @@ nvc0_fifo_create_context(struct nouveau_channel *chan) struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; struct nvc0_fifo_priv *priv = pfifo->priv; struct nvc0_fifo_chan *fifoch; - u64 ib_virt, user_vinst; + u64 ib_virt = chan->pushbuf_base + chan->dma.ib_base * 4; int ret; chan->fifo_priv = kzalloc(sizeof(*fifoch), GFP_KERNEL); @@ -115,28 +115,13 @@ nvc0_fifo_create_context(struct nouveau_channel *chan) fifoch = chan->fifo_priv; /* allocate vram for control regs, map into polling area */ - ret = nouveau_bo_new(dev, NULL, 0x1000, 0, TTM_PL_FLAG_VRAM, - 0, 0, &fifoch->user); + ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 0x1000, + NVOBJ_FLAG_ZERO_ALLOC, &fifoch->user); if (ret) goto error; - ret = nouveau_bo_pin(fifoch->user, TTM_PL_FLAG_VRAM); - if (ret) { - nouveau_bo_ref(NULL, &fifoch->user); - goto error; - } - - user_vinst = fifoch->user->bo.mem.start << PAGE_SHIFT; - - ret = nouveau_bo_map(fifoch->user); - if (ret) { - nouveau_bo_unpin(fifoch->user); - nouveau_bo_ref(NULL, &fifoch->user); - goto error; - } - nouveau_vm_map_at(&priv->user_vma, chan->id * 0x1000, - fifoch->user->bo.mem.mm_node); + *(struct nouveau_mem **)fifoch->user->node); chan->user = ioremap_wc(pci_resource_start(dev->pdev, 1) + priv->user_vma.offset + (chan->id * 0x1000), @@ -146,20 +131,6 @@ nvc0_fifo_create_context(struct nouveau_channel *chan) goto error; } - ib_virt = chan->pushbuf_base + chan->dma.ib_base * 4; - - /* zero channel regs */ - nouveau_bo_wr32(fifoch->user, 0x0040/4, 0); - nouveau_bo_wr32(fifoch->user, 0x0044/4, 0); - nouveau_bo_wr32(fifoch->user, 0x0048/4, 0); - nouveau_bo_wr32(fifoch->user, 0x004c/4, 0); - nouveau_bo_wr32(fifoch->user, 0x0050/4, 0); - nouveau_bo_wr32(fifoch->user, 0x0058/4, 0); - nouveau_bo_wr32(fifoch->user, 0x005c/4, 0); - nouveau_bo_wr32(fifoch->user, 0x0060/4, 0); - nouveau_bo_wr32(fifoch->user, 0x0088/4, 0); - nouveau_bo_wr32(fifoch->user, 0x008c/4, 0); - /* ramfc */ ret = nouveau_gpuobj_new_fake(dev, chan->ramin->pinst, chan->ramin->vinst, 0x100, @@ -167,8 +138,8 @@ nvc0_fifo_create_context(struct nouveau_channel *chan) if (ret) goto error; - nv_wo32(fifoch->ramfc, 0x08, lower_32_bits(user_vinst)); - nv_wo32(fifoch->ramfc, 0x0c, upper_32_bits(user_vinst)); + nv_wo32(fifoch->ramfc, 0x08, lower_32_bits(fifoch->user->vinst)); + nv_wo32(fifoch->ramfc, 0x0c, upper_32_bits(fifoch->user->vinst)); nv_wo32(fifoch->ramfc, 0x10, 0x0000face); nv_wo32(fifoch->ramfc, 0x30, 0xfffff902); nv_wo32(fifoch->ramfc, 0x48, lower_32_bits(ib_virt)); @@ -223,11 +194,7 @@ nvc0_fifo_destroy_context(struct nouveau_channel *chan) return; nouveau_gpuobj_ref(NULL, &fifoch->ramfc); - if (fifoch->user) { - nouveau_bo_unmap(fifoch->user); - nouveau_bo_unpin(fifoch->user); - nouveau_bo_ref(NULL, &fifoch->user); - } + nouveau_gpuobj_ref(NULL, &fifoch->user); kfree(fifoch); } @@ -240,6 +207,21 @@ nvc0_fifo_load_context(struct nouveau_channel *chan) int nvc0_fifo_unload_context(struct drm_device *dev) { + int i; + + for (i = 0; i < 128; i++) { + if (!(nv_rd32(dev, 0x003004 + (i * 8)) & 1)) + continue; + + nv_mask(dev, 0x003004 + (i * 8), 0x00000001, 0x00000000); + nv_wr32(dev, 0x002634, i); + if (!nv_wait(dev, 0x002634, 0xffffffff, i)) { + NV_INFO(dev, "PFIFO: kick ch %d failed: 0x%08x\n", + i, nv_rd32(dev, 0x002634)); + return -EBUSY; + } + } + return 0; } @@ -309,6 +291,7 @@ nvc0_fifo_init(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_fifo_engine *pfifo = &dev_priv->engine.fifo; + struct nouveau_channel *chan; struct nvc0_fifo_priv *priv; int ret, i; @@ -351,23 +334,74 @@ nvc0_fifo_init(struct drm_device *dev) nv_wr32(dev, 0x002a00, 0xffffffff); /* clears PFIFO.INTR bit 30 */ nv_wr32(dev, 0x002100, 0xffffffff); nv_wr32(dev, 0x002140, 0xbfffffff); + + /* restore PFIFO context table */ + for (i = 0; i < 128; i++) { + chan = dev_priv->channels.ptr[i]; + if (!chan || !chan->fifo_priv) + continue; + + nv_wr32(dev, 0x003000 + (i * 8), 0xc0000000 | + (chan->ramin->vinst >> 12)); + nv_wr32(dev, 0x003004 + (i * 8), 0x001f0001); + } + nvc0_fifo_playlist_update(dev); + return 0; } struct nouveau_enum nvc0_fifo_fault_unit[] = { - { 0, "PGRAPH" }, - { 3, "PEEPHOLE" }, - { 4, "BAR1" }, - { 5, "BAR3" }, - { 7, "PFIFO" }, + { 0x00, "PGRAPH" }, + { 0x03, "PEEPHOLE" }, + { 0x04, "BAR1" }, + { 0x05, "BAR3" }, + { 0x07, "PFIFO" }, + { 0x10, "PBSP" }, + { 0x11, "PPPP" }, + { 0x13, "PCOUNTER" }, + { 0x14, "PVP" }, + { 0x15, "PCOPY0" }, + { 0x16, "PCOPY1" }, + { 0x17, "PDAEMON" }, {} }; struct nouveau_enum nvc0_fifo_fault_reason[] = { - { 0, "PT_NOT_PRESENT" }, - { 1, "PT_TOO_SHORT" }, - { 2, "PAGE_NOT_PRESENT" }, - { 3, "VM_LIMIT_EXCEEDED" }, + { 0x00, "PT_NOT_PRESENT" }, + { 0x01, "PT_TOO_SHORT" }, + { 0x02, "PAGE_NOT_PRESENT" }, + { 0x03, "VM_LIMIT_EXCEEDED" }, + { 0x04, "NO_CHANNEL" }, + { 0x05, "PAGE_SYSTEM_ONLY" }, + { 0x06, "PAGE_READ_ONLY" }, + { 0x0a, "COMPRESSED_SYSRAM" }, + { 0x0c, "INVALID_STORAGE_TYPE" }, + {} +}; + +struct nouveau_enum nvc0_fifo_fault_hubclient[] = { + { 0x01, "PCOPY0" }, + { 0x02, "PCOPY1" }, + { 0x04, "DISPATCH" }, + { 0x05, "CTXCTL" }, + { 0x06, "PFIFO" }, + { 0x07, "BAR_READ" }, + { 0x08, "BAR_WRITE" }, + { 0x0b, "PVP" }, + { 0x0c, "PPPP" }, + { 0x0d, "PBSP" }, + { 0x11, "PCOUNTER" }, + { 0x12, "PDAEMON" }, + { 0x14, "CCACHE" }, + { 0x15, "CCACHE_POST" }, + {} +}; + +struct nouveau_enum nvc0_fifo_fault_gpcclient[] = { + { 0x01, "TEX" }, + { 0x0c, "ESETUP" }, + { 0x0e, "CTXCTL" }, + { 0x0f, "PROP" }, {} }; @@ -385,12 +419,20 @@ nvc0_fifo_isr_vm_fault(struct drm_device *dev, int unit) u32 valo = nv_rd32(dev, 0x2804 + (unit * 0x10)); u32 vahi = nv_rd32(dev, 0x2808 + (unit * 0x10)); u32 stat = nv_rd32(dev, 0x280c + (unit * 0x10)); + u32 client = (stat & 0x00001f00) >> 8; NV_INFO(dev, "PFIFO: %s fault at 0x%010llx [", (stat & 0x00000080) ? "write" : "read", (u64)vahi << 32 | valo); nouveau_enum_print(nvc0_fifo_fault_reason, stat & 0x0000000f); printk("] from "); nouveau_enum_print(nvc0_fifo_fault_unit, unit); + if (stat & 0x00000040) { + printk("/"); + nouveau_enum_print(nvc0_fifo_fault_hubclient, client); + } else { + printk("/GPC%d/", (stat & 0x1f000000) >> 24); + nouveau_enum_print(nvc0_fifo_fault_gpcclient, client); + } printk(" on channel 0x%010llx\n", (u64)inst << 12); } diff --git a/drivers/gpu/drm/nouveau/nvc0_graph.c b/drivers/gpu/drm/nouveau/nvc0_graph.c index 3de9b721d8d..5b2f6f42046 100644 --- a/drivers/gpu/drm/nouveau/nvc0_graph.c +++ b/drivers/gpu/drm/nouveau/nvc0_graph.c @@ -28,29 +28,69 @@ #include "nouveau_drv.h" #include "nouveau_mm.h" + #include "nvc0_graph.h" +#include "nvc0_grhub.fuc.h" +#include "nvc0_grgpc.fuc.h" -static void nvc0_graph_isr(struct drm_device *); -static void nvc0_runk140_isr(struct drm_device *); -static int nvc0_graph_unload_context_to(struct drm_device *dev, u64 chan); +static void +nvc0_graph_ctxctl_debug_unit(struct drm_device *dev, u32 base) +{ + NV_INFO(dev, "PGRAPH: %06x - done 0x%08x\n", base, + nv_rd32(dev, base + 0x400)); + NV_INFO(dev, "PGRAPH: %06x - stat 0x%08x 0x%08x 0x%08x 0x%08x\n", base, + nv_rd32(dev, base + 0x800), nv_rd32(dev, base + 0x804), + nv_rd32(dev, base + 0x808), nv_rd32(dev, base + 0x80c)); + NV_INFO(dev, "PGRAPH: %06x - stat 0x%08x 0x%08x 0x%08x 0x%08x\n", base, + nv_rd32(dev, base + 0x810), nv_rd32(dev, base + 0x814), + nv_rd32(dev, base + 0x818), nv_rd32(dev, base + 0x81c)); +} -void -nvc0_graph_fifo_access(struct drm_device *dev, bool enabled) +static void +nvc0_graph_ctxctl_debug(struct drm_device *dev) { + u32 gpcnr = nv_rd32(dev, 0x409604) & 0xffff; + u32 gpc; + + nvc0_graph_ctxctl_debug_unit(dev, 0x409000); + for (gpc = 0; gpc < gpcnr; gpc++) + nvc0_graph_ctxctl_debug_unit(dev, 0x502000 + (gpc * 0x8000)); } -struct nouveau_channel * -nvc0_graph_channel(struct drm_device *dev) +static int +nvc0_graph_load_context(struct nouveau_channel *chan) { - return NULL; + struct drm_device *dev = chan->dev; + + nv_wr32(dev, 0x409840, 0x00000030); + nv_wr32(dev, 0x409500, 0x80000000 | chan->ramin->vinst >> 12); + nv_wr32(dev, 0x409504, 0x00000003); + if (!nv_wait(dev, 0x409800, 0x00000010, 0x00000010)) + NV_ERROR(dev, "PGRAPH: load_ctx timeout\n"); + + return 0; +} + +static int +nvc0_graph_unload_context_to(struct drm_device *dev, u64 chan) +{ + nv_wr32(dev, 0x409840, 0x00000003); + nv_wr32(dev, 0x409500, 0x80000000 | chan >> 12); + nv_wr32(dev, 0x409504, 0x00000009); + if (!nv_wait(dev, 0x409800, 0x00000001, 0x00000000)) { + NV_ERROR(dev, "PGRAPH: unload_ctx timeout\n"); + return -EBUSY; + } + + return 0; } static int nvc0_graph_construct_context(struct nouveau_channel *chan) { struct drm_nouveau_private *dev_priv = chan->dev->dev_private; - struct nvc0_graph_priv *priv = dev_priv->engine.graph.priv; - struct nvc0_graph_chan *grch = chan->pgraph_ctx; + struct nvc0_graph_priv *priv = nv_engine(chan->dev, NVOBJ_ENGINE_GR); + struct nvc0_graph_chan *grch = chan->engctx[NVOBJ_ENGINE_GR]; struct drm_device *dev = chan->dev; int ret, i; u32 *ctx; @@ -59,24 +99,44 @@ nvc0_graph_construct_context(struct nouveau_channel *chan) if (!ctx) return -ENOMEM; - nvc0_graph_load_context(chan); - - nv_wo32(grch->grctx, 0x1c, 1); - nv_wo32(grch->grctx, 0x20, 0); - nv_wo32(grch->grctx, 0x28, 0); - nv_wo32(grch->grctx, 0x2c, 0); - dev_priv->engine.instmem.flush(dev); - - ret = nvc0_grctx_generate(chan); - if (ret) { - kfree(ctx); - return ret; + if (!nouveau_ctxfw) { + nv_wr32(dev, 0x409840, 0x80000000); + nv_wr32(dev, 0x409500, 0x80000000 | chan->ramin->vinst >> 12); + nv_wr32(dev, 0x409504, 0x00000001); + if (!nv_wait(dev, 0x409800, 0x80000000, 0x80000000)) { + NV_ERROR(dev, "PGRAPH: HUB_SET_CHAN timeout\n"); + nvc0_graph_ctxctl_debug(dev); + ret = -EBUSY; + goto err; + } + } else { + nvc0_graph_load_context(chan); + + nv_wo32(grch->grctx, 0x1c, 1); + nv_wo32(grch->grctx, 0x20, 0); + nv_wo32(grch->grctx, 0x28, 0); + nv_wo32(grch->grctx, 0x2c, 0); + dev_priv->engine.instmem.flush(dev); } - ret = nvc0_graph_unload_context_to(dev, chan->ramin->vinst); - if (ret) { - kfree(ctx); - return ret; + ret = nvc0_grctx_generate(chan); + if (ret) + goto err; + + if (!nouveau_ctxfw) { + nv_wr32(dev, 0x409840, 0x80000000); + nv_wr32(dev, 0x409500, 0x80000000 | chan->ramin->vinst >> 12); + nv_wr32(dev, 0x409504, 0x00000002); + if (!nv_wait(dev, 0x409800, 0x80000000, 0x80000000)) { + NV_ERROR(dev, "PGRAPH: HUB_CTX_SAVE timeout\n"); + nvc0_graph_ctxctl_debug(dev); + ret = -EBUSY; + goto err; + } + } else { + ret = nvc0_graph_unload_context_to(dev, chan->ramin->vinst); + if (ret) + goto err; } for (i = 0; i < priv->grctx_size; i += 4) @@ -84,62 +144,65 @@ nvc0_graph_construct_context(struct nouveau_channel *chan) priv->grctx_vals = ctx; return 0; + +err: + kfree(ctx); + return ret; } static int nvc0_graph_create_context_mmio_list(struct nouveau_channel *chan) { - struct drm_nouveau_private *dev_priv = chan->dev->dev_private; - struct nvc0_graph_priv *priv = dev_priv->engine.graph.priv; - struct nvc0_graph_chan *grch = chan->pgraph_ctx; + struct nvc0_graph_priv *priv = nv_engine(chan->dev, NVOBJ_ENGINE_GR); + struct nvc0_graph_chan *grch = chan->engctx[NVOBJ_ENGINE_GR]; struct drm_device *dev = chan->dev; int i = 0, gpc, tp, ret; u32 magic; - ret = nouveau_gpuobj_new(dev, NULL, 0x2000, 256, NVOBJ_FLAG_VM, + ret = nouveau_gpuobj_new(dev, chan, 0x2000, 256, NVOBJ_FLAG_VM, &grch->unk408004); if (ret) return ret; - ret = nouveau_gpuobj_new(dev, NULL, 0x8000, 256, NVOBJ_FLAG_VM, + ret = nouveau_gpuobj_new(dev, chan, 0x8000, 256, NVOBJ_FLAG_VM, &grch->unk40800c); if (ret) return ret; - ret = nouveau_gpuobj_new(dev, NULL, 384 * 1024, 4096, + ret = nouveau_gpuobj_new(dev, chan, 384 * 1024, 4096, NVOBJ_FLAG_VM | NVOBJ_FLAG_VM_USER, &grch->unk418810); if (ret) return ret; - ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 0, NVOBJ_FLAG_VM, + ret = nouveau_gpuobj_new(dev, chan, 0x1000, 0, NVOBJ_FLAG_VM, &grch->mmio); if (ret) return ret; nv_wo32(grch->mmio, i++ * 4, 0x00408004); - nv_wo32(grch->mmio, i++ * 4, grch->unk408004->vinst >> 8); + nv_wo32(grch->mmio, i++ * 4, grch->unk408004->linst >> 8); nv_wo32(grch->mmio, i++ * 4, 0x00408008); nv_wo32(grch->mmio, i++ * 4, 0x80000018); nv_wo32(grch->mmio, i++ * 4, 0x0040800c); - nv_wo32(grch->mmio, i++ * 4, grch->unk40800c->vinst >> 8); + nv_wo32(grch->mmio, i++ * 4, grch->unk40800c->linst >> 8); nv_wo32(grch->mmio, i++ * 4, 0x00408010); nv_wo32(grch->mmio, i++ * 4, 0x80000000); nv_wo32(grch->mmio, i++ * 4, 0x00418810); - nv_wo32(grch->mmio, i++ * 4, 0x80000000 | grch->unk418810->vinst >> 12); + nv_wo32(grch->mmio, i++ * 4, 0x80000000 | grch->unk418810->linst >> 12); nv_wo32(grch->mmio, i++ * 4, 0x00419848); - nv_wo32(grch->mmio, i++ * 4, 0x10000000 | grch->unk418810->vinst >> 12); + nv_wo32(grch->mmio, i++ * 4, 0x10000000 | grch->unk418810->linst >> 12); nv_wo32(grch->mmio, i++ * 4, 0x00419004); - nv_wo32(grch->mmio, i++ * 4, grch->unk40800c->vinst >> 8); + nv_wo32(grch->mmio, i++ * 4, grch->unk40800c->linst >> 8); nv_wo32(grch->mmio, i++ * 4, 0x00419008); nv_wo32(grch->mmio, i++ * 4, 0x00000000); nv_wo32(grch->mmio, i++ * 4, 0x00418808); - nv_wo32(grch->mmio, i++ * 4, grch->unk408004->vinst >> 8); + nv_wo32(grch->mmio, i++ * 4, grch->unk408004->linst >> 8); nv_wo32(grch->mmio, i++ * 4, 0x0041880c); nv_wo32(grch->mmio, i++ * 4, 0x80000018); @@ -147,7 +210,7 @@ nvc0_graph_create_context_mmio_list(struct nouveau_channel *chan) nv_wo32(grch->mmio, i++ * 4, 0x00405830); nv_wo32(grch->mmio, i++ * 4, magic); for (gpc = 0; gpc < priv->gpc_nr; gpc++) { - for (tp = 0; tp < priv->tp_nr[gpc]; tp++, magic += 0x02fc) { + for (tp = 0; tp < priv->tp_nr[gpc]; tp++, magic += 0x0324) { u32 reg = 0x504520 + (gpc * 0x8000) + (tp * 0x0800); nv_wo32(grch->mmio, i++ * 4, reg); nv_wo32(grch->mmio, i++ * 4, magic); @@ -158,37 +221,35 @@ nvc0_graph_create_context_mmio_list(struct nouveau_channel *chan) return 0; } -int -nvc0_graph_create_context(struct nouveau_channel *chan) +static int +nvc0_graph_context_new(struct nouveau_channel *chan, int engine) { - struct drm_nouveau_private *dev_priv = chan->dev->dev_private; + struct drm_device *dev = chan->dev; + struct drm_nouveau_private *dev_priv = dev->dev_private; struct nouveau_instmem_engine *pinstmem = &dev_priv->engine.instmem; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct nvc0_graph_priv *priv = pgraph->priv; + struct nvc0_graph_priv *priv = nv_engine(dev, engine); struct nvc0_graph_chan *grch; - struct drm_device *dev = chan->dev; struct nouveau_gpuobj *grctx; int ret, i; - chan->pgraph_ctx = kzalloc(sizeof(*grch), GFP_KERNEL); - if (!chan->pgraph_ctx) + grch = kzalloc(sizeof(*grch), GFP_KERNEL); + if (!grch) return -ENOMEM; - grch = chan->pgraph_ctx; + chan->engctx[NVOBJ_ENGINE_GR] = grch; - ret = nouveau_gpuobj_new(dev, NULL, priv->grctx_size, 256, + ret = nouveau_gpuobj_new(dev, chan, priv->grctx_size, 256, NVOBJ_FLAG_VM | NVOBJ_FLAG_ZERO_ALLOC, &grch->grctx); if (ret) goto error; - chan->ramin_grctx = grch->grctx; grctx = grch->grctx; ret = nvc0_graph_create_context_mmio_list(chan); if (ret) goto error; - nv_wo32(chan->ramin, 0x0210, lower_32_bits(grctx->vinst) | 4); - nv_wo32(chan->ramin, 0x0214, upper_32_bits(grctx->vinst)); + nv_wo32(chan->ramin, 0x0210, lower_32_bits(grctx->linst) | 4); + nv_wo32(chan->ramin, 0x0214, upper_32_bits(grctx->linst)); pinstmem->flush(dev); if (!priv->grctx_vals) { @@ -200,104 +261,54 @@ nvc0_graph_create_context(struct nouveau_channel *chan) for (i = 0; i < priv->grctx_size; i += 4) nv_wo32(grctx, i, priv->grctx_vals[i / 4]); - nv_wo32(grctx, 0xf4, 0); - nv_wo32(grctx, 0xf8, 0); - nv_wo32(grctx, 0x10, grch->mmio_nr); - nv_wo32(grctx, 0x14, lower_32_bits(grch->mmio->vinst)); - nv_wo32(grctx, 0x18, upper_32_bits(grch->mmio->vinst)); - nv_wo32(grctx, 0x1c, 1); - nv_wo32(grctx, 0x20, 0); - nv_wo32(grctx, 0x28, 0); - nv_wo32(grctx, 0x2c, 0); + if (!nouveau_ctxfw) { + nv_wo32(grctx, 0x00, grch->mmio_nr); + nv_wo32(grctx, 0x04, grch->mmio->linst >> 8); + } else { + nv_wo32(grctx, 0xf4, 0); + nv_wo32(grctx, 0xf8, 0); + nv_wo32(grctx, 0x10, grch->mmio_nr); + nv_wo32(grctx, 0x14, lower_32_bits(grch->mmio->linst)); + nv_wo32(grctx, 0x18, upper_32_bits(grch->mmio->linst)); + nv_wo32(grctx, 0x1c, 1); + nv_wo32(grctx, 0x20, 0); + nv_wo32(grctx, 0x28, 0); + nv_wo32(grctx, 0x2c, 0); + } pinstmem->flush(dev); return 0; error: - pgraph->destroy_context(chan); + priv->base.context_del(chan, engine); return ret; } -void -nvc0_graph_destroy_context(struct nouveau_channel *chan) +static void +nvc0_graph_context_del(struct nouveau_channel *chan, int engine) { - struct nvc0_graph_chan *grch; - - grch = chan->pgraph_ctx; - chan->pgraph_ctx = NULL; - if (!grch) - return; + struct nvc0_graph_chan *grch = chan->engctx[engine]; nouveau_gpuobj_ref(NULL, &grch->mmio); nouveau_gpuobj_ref(NULL, &grch->unk418810); nouveau_gpuobj_ref(NULL, &grch->unk40800c); nouveau_gpuobj_ref(NULL, &grch->unk408004); nouveau_gpuobj_ref(NULL, &grch->grctx); - chan->ramin_grctx = NULL; + chan->engctx[engine] = NULL; } -int -nvc0_graph_load_context(struct nouveau_channel *chan) +static int +nvc0_graph_object_new(struct nouveau_channel *chan, int engine, + u32 handle, u16 class) { - struct drm_device *dev = chan->dev; - - nv_wr32(dev, 0x409840, 0x00000030); - nv_wr32(dev, 0x409500, 0x80000000 | chan->ramin->vinst >> 12); - nv_wr32(dev, 0x409504, 0x00000003); - if (!nv_wait(dev, 0x409800, 0x00000010, 0x00000010)) - NV_ERROR(dev, "PGRAPH: load_ctx timeout\n"); - return 0; } static int -nvc0_graph_unload_context_to(struct drm_device *dev, u64 chan) +nvc0_graph_fini(struct drm_device *dev, int engine, bool suspend) { - nv_wr32(dev, 0x409840, 0x00000003); - nv_wr32(dev, 0x409500, 0x80000000 | chan >> 12); - nv_wr32(dev, 0x409504, 0x00000009); - if (!nv_wait(dev, 0x409800, 0x00000001, 0x00000000)) { - NV_ERROR(dev, "PGRAPH: unload_ctx timeout\n"); - return -EBUSY; - } - return 0; } -int -nvc0_graph_unload_context(struct drm_device *dev) -{ - u64 inst = (u64)(nv_rd32(dev, 0x409b00) & 0x0fffffff) << 12; - return nvc0_graph_unload_context_to(dev, inst); -} - -static void -nvc0_graph_destroy(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct nvc0_graph_priv *priv; - - priv = pgraph->priv; - if (!priv) - return; - - nouveau_irq_unregister(dev, 12); - nouveau_irq_unregister(dev, 25); - - nouveau_gpuobj_ref(NULL, &priv->unk4188b8); - nouveau_gpuobj_ref(NULL, &priv->unk4188b4); - - if (priv->grctx_vals) - kfree(priv->grctx_vals); - kfree(priv); -} - -void -nvc0_graph_takedown(struct drm_device *dev) -{ - nvc0_graph_destroy(dev); -} - static int nvc0_graph_mthd_page_flip(struct nouveau_channel *chan, u32 class, u32 mthd, u32 data) @@ -306,119 +317,10 @@ nvc0_graph_mthd_page_flip(struct nouveau_channel *chan, return 0; } -static int -nvc0_graph_create(struct drm_device *dev) -{ - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct nvc0_graph_priv *priv; - int ret, gpc, i; - - priv = kzalloc(sizeof(*priv), GFP_KERNEL); - if (!priv) - return -ENOMEM; - pgraph->priv = priv; - - ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 256, 0, &priv->unk4188b4); - if (ret) - goto error; - - ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 256, 0, &priv->unk4188b8); - if (ret) - goto error; - - for (i = 0; i < 0x1000; i += 4) { - nv_wo32(priv->unk4188b4, i, 0x00000010); - nv_wo32(priv->unk4188b8, i, 0x00000010); - } - - priv->gpc_nr = nv_rd32(dev, 0x409604) & 0x0000001f; - priv->rop_nr = (nv_rd32(dev, 0x409604) & 0x001f0000) >> 16; - for (gpc = 0; gpc < priv->gpc_nr; gpc++) { - priv->tp_nr[gpc] = nv_rd32(dev, GPC_UNIT(gpc, 0x2608)); - priv->tp_total += priv->tp_nr[gpc]; - } - - /*XXX: these need figuring out... */ - switch (dev_priv->chipset) { - case 0xc0: - if (priv->tp_total == 11) { /* 465, 3/4/4/0, 4 */ - priv->magic_not_rop_nr = 0x07; - /* filled values up to tp_total, the rest 0 */ - priv->magicgpc980[0] = 0x22111000; - priv->magicgpc980[1] = 0x00000233; - priv->magicgpc980[2] = 0x00000000; - priv->magicgpc980[3] = 0x00000000; - priv->magicgpc918 = 0x000ba2e9; - } else - if (priv->tp_total == 14) { /* 470, 3/3/4/4, 5 */ - priv->magic_not_rop_nr = 0x05; - priv->magicgpc980[0] = 0x11110000; - priv->magicgpc980[1] = 0x00233222; - priv->magicgpc980[2] = 0x00000000; - priv->magicgpc980[3] = 0x00000000; - priv->magicgpc918 = 0x00092493; - } else - if (priv->tp_total == 15) { /* 480, 3/4/4/4, 6 */ - priv->magic_not_rop_nr = 0x06; - priv->magicgpc980[0] = 0x11110000; - priv->magicgpc980[1] = 0x03332222; - priv->magicgpc980[2] = 0x00000000; - priv->magicgpc980[3] = 0x00000000; - priv->magicgpc918 = 0x00088889; - } - break; - case 0xc3: /* 450, 4/0/0/0, 2 */ - priv->magic_not_rop_nr = 0x03; - priv->magicgpc980[0] = 0x00003210; - priv->magicgpc980[1] = 0x00000000; - priv->magicgpc980[2] = 0x00000000; - priv->magicgpc980[3] = 0x00000000; - priv->magicgpc918 = 0x00200000; - break; - case 0xc4: /* 460, 3/4/0/0, 4 */ - priv->magic_not_rop_nr = 0x01; - priv->magicgpc980[0] = 0x02321100; - priv->magicgpc980[1] = 0x00000000; - priv->magicgpc980[2] = 0x00000000; - priv->magicgpc980[3] = 0x00000000; - priv->magicgpc918 = 0x00124925; - break; - } - - if (!priv->magic_not_rop_nr) { - NV_ERROR(dev, "PGRAPH: unknown config: %d/%d/%d/%d, %d\n", - priv->tp_nr[0], priv->tp_nr[1], priv->tp_nr[2], - priv->tp_nr[3], priv->rop_nr); - /* use 0xc3's values... */ - priv->magic_not_rop_nr = 0x03; - priv->magicgpc980[0] = 0x00003210; - priv->magicgpc980[1] = 0x00000000; - priv->magicgpc980[2] = 0x00000000; - priv->magicgpc980[3] = 0x00000000; - priv->magicgpc918 = 0x00200000; - } - - nouveau_irq_register(dev, 12, nvc0_graph_isr); - nouveau_irq_register(dev, 25, nvc0_runk140_isr); - NVOBJ_CLASS(dev, 0x902d, GR); /* 2D */ - NVOBJ_CLASS(dev, 0x9039, GR); /* M2MF */ - NVOBJ_MTHD (dev, 0x9039, 0x0500, nvc0_graph_mthd_page_flip); - NVOBJ_CLASS(dev, 0x9097, GR); /* 3D */ - NVOBJ_CLASS(dev, 0x90c0, GR); /* COMPUTE */ - return 0; - -error: - nvc0_graph_destroy(dev); - return ret; -} - static void nvc0_graph_init_obj418880(struct drm_device *dev) { - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; - struct nvc0_graph_priv *priv = pgraph->priv; + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); int i; nv_wr32(dev, GPC_BCAST(0x0880), 0x00000000); @@ -449,44 +351,45 @@ nvc0_graph_init_regs(struct drm_device *dev) static void nvc0_graph_init_gpc_0(struct drm_device *dev) { - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvc0_graph_priv *priv = dev_priv->engine.graph.priv; - int gpc; - - // TP ROP UNKVAL(magic_not_rop_nr) - // 450: 4/0/0/0 2 3 - // 460: 3/4/0/0 4 1 - // 465: 3/4/4/0 4 7 - // 470: 3/3/4/4 5 5 - // 480: 3/4/4/4 6 6 - - // magicgpc918 - // 450: 00200000 00000000001000000000000000000000 - // 460: 00124925 00000000000100100100100100100101 - // 465: 000ba2e9 00000000000010111010001011101001 - // 470: 00092493 00000000000010010010010010010011 - // 480: 00088889 00000000000010001000100010001001 - - /* filled values up to tp_total, remainder 0 */ - // 450: 00003210 00000000 00000000 00000000 - // 460: 02321100 00000000 00000000 00000000 - // 465: 22111000 00000233 00000000 00000000 - // 470: 11110000 00233222 00000000 00000000 - // 480: 11110000 03332222 00000000 00000000 - - nv_wr32(dev, GPC_BCAST(0x0980), priv->magicgpc980[0]); - nv_wr32(dev, GPC_BCAST(0x0984), priv->magicgpc980[1]); - nv_wr32(dev, GPC_BCAST(0x0988), priv->magicgpc980[2]); - nv_wr32(dev, GPC_BCAST(0x098c), priv->magicgpc980[3]); + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); + const u32 magicgpc918 = DIV_ROUND_UP(0x00800000, priv->tp_total); + u32 data[TP_MAX / 8]; + u8 tpnr[GPC_MAX]; + int i, gpc, tpc; + + /* + * TP ROP UNKVAL(magic_not_rop_nr) + * 450: 4/0/0/0 2 3 + * 460: 3/4/0/0 4 1 + * 465: 3/4/4/0 4 7 + * 470: 3/3/4/4 5 5 + * 480: 3/4/4/4 6 6 + */ + + memset(data, 0x00, sizeof(data)); + memcpy(tpnr, priv->tp_nr, sizeof(priv->tp_nr)); + for (i = 0, gpc = -1; i < priv->tp_total; i++) { + do { + gpc = (gpc + 1) % priv->gpc_nr; + } while (!tpnr[gpc]); + tpc = priv->tp_nr[gpc] - tpnr[gpc]--; + + data[i / 8] |= tpc << ((i % 8) * 4); + } + + nv_wr32(dev, GPC_BCAST(0x0980), data[0]); + nv_wr32(dev, GPC_BCAST(0x0984), data[1]); + nv_wr32(dev, GPC_BCAST(0x0988), data[2]); + nv_wr32(dev, GPC_BCAST(0x098c), data[3]); for (gpc = 0; gpc < priv->gpc_nr; gpc++) { nv_wr32(dev, GPC_UNIT(gpc, 0x0914), priv->magic_not_rop_nr << 8 | priv->tp_nr[gpc]); nv_wr32(dev, GPC_UNIT(gpc, 0x0910), 0x00040000 | priv->tp_total); - nv_wr32(dev, GPC_UNIT(gpc, 0x0918), priv->magicgpc918); + nv_wr32(dev, GPC_UNIT(gpc, 0x0918), magicgpc918); } - nv_wr32(dev, GPC_BCAST(0x1bd4), priv->magicgpc918); + nv_wr32(dev, GPC_BCAST(0x1bd4), magicgpc918); nv_wr32(dev, GPC_BCAST(0x08ac), priv->rop_nr); } @@ -509,8 +412,7 @@ nvc0_graph_init_units(struct drm_device *dev) static void nvc0_graph_init_gpc_1(struct drm_device *dev) { - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvc0_graph_priv *priv = dev_priv->engine.graph.priv; + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); int gpc, tp; for (gpc = 0; gpc < priv->gpc_nr; gpc++) { @@ -535,8 +437,7 @@ nvc0_graph_init_gpc_1(struct drm_device *dev) static void nvc0_graph_init_rop(struct drm_device *dev) { - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvc0_graph_priv *priv = dev_priv->engine.graph.priv; + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); int rop; for (rop = 0; rop < priv->rop_nr; rop++) { @@ -547,62 +448,79 @@ nvc0_graph_init_rop(struct drm_device *dev) } } -static int -nvc0_fuc_load_fw(struct drm_device *dev, u32 fuc_base, - const char *code_fw, const char *data_fw) +static void +nvc0_graph_init_fuc(struct drm_device *dev, u32 fuc_base, + struct nvc0_graph_fuc *code, struct nvc0_graph_fuc *data) { - const struct firmware *fw; - char name[32]; - int ret, i; - - snprintf(name, sizeof(name), "nouveau/%s", data_fw); - ret = request_firmware(&fw, name, &dev->pdev->dev); - if (ret) { - NV_ERROR(dev, "failed to load %s\n", data_fw); - return ret; - } + int i; nv_wr32(dev, fuc_base + 0x01c0, 0x01000000); - for (i = 0; i < fw->size / 4; i++) - nv_wr32(dev, fuc_base + 0x01c4, ((u32 *)fw->data)[i]); - release_firmware(fw); - - snprintf(name, sizeof(name), "nouveau/%s", code_fw); - ret = request_firmware(&fw, name, &dev->pdev->dev); - if (ret) { - NV_ERROR(dev, "failed to load %s\n", code_fw); - return ret; - } + for (i = 0; i < data->size / 4; i++) + nv_wr32(dev, fuc_base + 0x01c4, data->data[i]); nv_wr32(dev, fuc_base + 0x0180, 0x01000000); - for (i = 0; i < fw->size / 4; i++) { + for (i = 0; i < code->size / 4; i++) { if ((i & 0x3f) == 0) nv_wr32(dev, fuc_base + 0x0188, i >> 6); - nv_wr32(dev, fuc_base + 0x0184, ((u32 *)fw->data)[i]); + nv_wr32(dev, fuc_base + 0x0184, code->data[i]); } - release_firmware(fw); - - return 0; } static int nvc0_graph_init_ctxctl(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nvc0_graph_priv *priv = dev_priv->engine.graph.priv; + struct nvc0_graph_priv *priv = nv_engine(dev, NVOBJ_ENGINE_GR); u32 r000260; - int ret; + int i; + + if (!nouveau_ctxfw) { + /* load HUB microcode */ + r000260 = nv_mask(dev, 0x000260, 0x00000001, 0x00000000); + nv_wr32(dev, 0x4091c0, 0x01000000); + for (i = 0; i < sizeof(nvc0_grhub_data) / 4; i++) + nv_wr32(dev, 0x4091c4, nvc0_grhub_data[i]); + + nv_wr32(dev, 0x409180, 0x01000000); + for (i = 0; i < sizeof(nvc0_grhub_code) / 4; i++) { + if ((i & 0x3f) == 0) + nv_wr32(dev, 0x409188, i >> 6); + nv_wr32(dev, 0x409184, nvc0_grhub_code[i]); + } + + /* load GPC microcode */ + nv_wr32(dev, 0x41a1c0, 0x01000000); + for (i = 0; i < sizeof(nvc0_grgpc_data) / 4; i++) + nv_wr32(dev, 0x41a1c4, nvc0_grgpc_data[i]); + + nv_wr32(dev, 0x41a180, 0x01000000); + for (i = 0; i < sizeof(nvc0_grgpc_code) / 4; i++) { + if ((i & 0x3f) == 0) + nv_wr32(dev, 0x41a188, i >> 6); + nv_wr32(dev, 0x41a184, nvc0_grgpc_code[i]); + } + nv_wr32(dev, 0x000260, r000260); + + /* start HUB ucode running, it'll init the GPCs */ + nv_wr32(dev, 0x409800, dev_priv->chipset); + nv_wr32(dev, 0x40910c, 0x00000000); + nv_wr32(dev, 0x409100, 0x00000002); + if (!nv_wait(dev, 0x409800, 0x80000000, 0x80000000)) { + NV_ERROR(dev, "PGRAPH: HUB_INIT timed out\n"); + nvc0_graph_ctxctl_debug(dev); + return -EBUSY; + } + + priv->grctx_size = nv_rd32(dev, 0x409804); + return 0; + } /* load fuc microcode */ r000260 = nv_mask(dev, 0x000260, 0x00000001, 0x00000000); - ret = nvc0_fuc_load_fw(dev, 0x409000, "fuc409c", "fuc409d"); - if (ret == 0) - ret = nvc0_fuc_load_fw(dev, 0x41a000, "fuc41ac", "fuc41ad"); + nvc0_graph_init_fuc(dev, 0x409000, &priv->fuc409c, &priv->fuc409d); + nvc0_graph_init_fuc(dev, 0x41a000, &priv->fuc41ac, &priv->fuc41ad); nv_wr32(dev, 0x000260, r000260); - if (ret) - return ret; - /* start both of them running */ nv_wr32(dev, 0x409840, 0xffffffff); nv_wr32(dev, 0x41a10c, 0x00000000); @@ -644,41 +562,19 @@ nvc0_graph_init_ctxctl(struct drm_device *dev) return 0; } -int -nvc0_graph_init(struct drm_device *dev) +static int +nvc0_graph_init(struct drm_device *dev, int engine) { - struct drm_nouveau_private *dev_priv = dev->dev_private; - struct nouveau_pgraph_engine *pgraph = &dev_priv->engine.graph; int ret; - dev_priv->engine.graph.accel_blocked = true; - - switch (dev_priv->chipset) { - case 0xc0: - case 0xc3: - case 0xc4: - break; - default: - NV_ERROR(dev, "PGRAPH: unsupported chipset, please report!\n"); - if (nouveau_noaccel != 0) - return 0; - break; - } - nv_mask(dev, 0x000200, 0x18001000, 0x00000000); nv_mask(dev, 0x000200, 0x18001000, 0x18001000); - if (!pgraph->priv) { - ret = nvc0_graph_create(dev); - if (ret) - return ret; - } - nvc0_graph_init_obj418880(dev); nvc0_graph_init_regs(dev); - //nvc0_graph_init_unitplemented_magics(dev); + /*nvc0_graph_init_unitplemented_magics(dev);*/ nvc0_graph_init_gpc_0(dev); - //nvc0_graph_init_unitplemented_c242(dev); + /*nvc0_graph_init_unitplemented_c242(dev);*/ nv_wr32(dev, 0x400500, 0x00010001); nv_wr32(dev, 0x400100, 0xffffffff); @@ -697,12 +593,13 @@ nvc0_graph_init(struct drm_device *dev) nv_wr32(dev, 0x400054, 0x34ce3464); ret = nvc0_graph_init_ctxctl(dev); - if (ret == 0) - dev_priv->engine.graph.accel_blocked = false; + if (ret) + return ret; + return 0; } -static int +int nvc0_graph_isr_chid(struct drm_device *dev, u64 inst) { struct drm_nouveau_private *dev_priv = dev->dev_private; @@ -724,6 +621,22 @@ nvc0_graph_isr_chid(struct drm_device *dev, u64 inst) } static void +nvc0_graph_ctxctl_isr(struct drm_device *dev) +{ + u32 ustat = nv_rd32(dev, 0x409c18); + + if (ustat & 0x00000001) + NV_INFO(dev, "PGRAPH: CTXCTRL ucode error\n"); + if (ustat & 0x00080000) + NV_INFO(dev, "PGRAPH: CTXCTRL watchdog timeout\n"); + if (ustat & ~0x00080001) + NV_INFO(dev, "PGRAPH: CTXCTRL 0x%08x\n", ustat); + + nvc0_graph_ctxctl_debug(dev); + nv_wr32(dev, 0x409c20, ustat); +} + +static void nvc0_graph_isr(struct drm_device *dev) { u64 inst = (u64)(nv_rd32(dev, 0x409b00) & 0x0fffffff) << 12; @@ -774,11 +687,7 @@ nvc0_graph_isr(struct drm_device *dev) } if (stat & 0x00080000) { - u32 ustat = nv_rd32(dev, 0x409c18); - - NV_INFO(dev, "PGRAPH: CTXCTRL ustat 0x%08x\n", ustat); - - nv_wr32(dev, 0x409c20, ustat); + nvc0_graph_ctxctl_isr(dev); nv_wr32(dev, 0x400100, 0x00080000); stat &= ~0x00080000; } @@ -802,7 +711,198 @@ nvc0_runk140_isr(struct drm_device *dev) u32 st0 = nv_mask(dev, reg + 0x1020, 0, 0); u32 st1 = nv_mask(dev, reg + 0x1420, 0, 0); - NV_INFO(dev, "PRUNK140: %d 0x%08x 0x%08x\n", unit, st0, st1); + NV_DEBUG(dev, "PRUNK140: %d 0x%08x 0x%08x\n", unit, st0, st1); units &= ~(1 << unit); } } + +static int +nvc0_graph_create_fw(struct drm_device *dev, const char *fwname, + struct nvc0_graph_fuc *fuc) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + const struct firmware *fw; + char f[32]; + int ret; + + snprintf(f, sizeof(f), "nouveau/nv%02x_%s", dev_priv->chipset, fwname); + ret = request_firmware(&fw, f, &dev->pdev->dev); + if (ret) { + snprintf(f, sizeof(f), "nouveau/%s", fwname); + ret = request_firmware(&fw, f, &dev->pdev->dev); + if (ret) { + NV_ERROR(dev, "failed to load %s\n", fwname); + return ret; + } + } + + fuc->size = fw->size; + fuc->data = kmemdup(fw->data, fuc->size, GFP_KERNEL); + release_firmware(fw); + return (fuc->data != NULL) ? 0 : -ENOMEM; +} + +static void +nvc0_graph_destroy_fw(struct nvc0_graph_fuc *fuc) +{ + if (fuc->data) { + kfree(fuc->data); + fuc->data = NULL; + } +} + +static void +nvc0_graph_destroy(struct drm_device *dev, int engine) +{ + struct nvc0_graph_priv *priv = nv_engine(dev, engine); + + if (nouveau_ctxfw) { + nvc0_graph_destroy_fw(&priv->fuc409c); + nvc0_graph_destroy_fw(&priv->fuc409d); + nvc0_graph_destroy_fw(&priv->fuc41ac); + nvc0_graph_destroy_fw(&priv->fuc41ad); + } + + nouveau_irq_unregister(dev, 12); + nouveau_irq_unregister(dev, 25); + + nouveau_gpuobj_ref(NULL, &priv->unk4188b8); + nouveau_gpuobj_ref(NULL, &priv->unk4188b4); + + if (priv->grctx_vals) + kfree(priv->grctx_vals); + + NVOBJ_ENGINE_DEL(dev, GR); + kfree(priv); +} + +int +nvc0_graph_create(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nvc0_graph_priv *priv; + int ret, gpc, i; + u32 fermi; + + fermi = nvc0_graph_class(dev); + if (!fermi) { + NV_ERROR(dev, "PGRAPH: unsupported chipset, please report!\n"); + return 0; + } + + priv = kzalloc(sizeof(*priv), GFP_KERNEL); + if (!priv) + return -ENOMEM; + + priv->base.destroy = nvc0_graph_destroy; + priv->base.init = nvc0_graph_init; + priv->base.fini = nvc0_graph_fini; + priv->base.context_new = nvc0_graph_context_new; + priv->base.context_del = nvc0_graph_context_del; + priv->base.object_new = nvc0_graph_object_new; + + NVOBJ_ENGINE_ADD(dev, GR, &priv->base); + nouveau_irq_register(dev, 12, nvc0_graph_isr); + nouveau_irq_register(dev, 25, nvc0_runk140_isr); + + if (nouveau_ctxfw) { + NV_INFO(dev, "PGRAPH: using external firmware\n"); + if (nvc0_graph_create_fw(dev, "fuc409c", &priv->fuc409c) || + nvc0_graph_create_fw(dev, "fuc409d", &priv->fuc409d) || + nvc0_graph_create_fw(dev, "fuc41ac", &priv->fuc41ac) || + nvc0_graph_create_fw(dev, "fuc41ad", &priv->fuc41ad)) { + ret = 0; + goto error; + } + } + + ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 256, 0, &priv->unk4188b4); + if (ret) + goto error; + + ret = nouveau_gpuobj_new(dev, NULL, 0x1000, 256, 0, &priv->unk4188b8); + if (ret) + goto error; + + for (i = 0; i < 0x1000; i += 4) { + nv_wo32(priv->unk4188b4, i, 0x00000010); + nv_wo32(priv->unk4188b8, i, 0x00000010); + } + + priv->gpc_nr = nv_rd32(dev, 0x409604) & 0x0000001f; + priv->rop_nr = (nv_rd32(dev, 0x409604) & 0x001f0000) >> 16; + for (gpc = 0; gpc < priv->gpc_nr; gpc++) { + priv->tp_nr[gpc] = nv_rd32(dev, GPC_UNIT(gpc, 0x2608)); + priv->tp_total += priv->tp_nr[gpc]; + } + + /*XXX: these need figuring out... */ + switch (dev_priv->chipset) { + case 0xc0: + if (priv->tp_total == 11) { /* 465, 3/4/4/0, 4 */ + priv->magic_not_rop_nr = 0x07; + } else + if (priv->tp_total == 14) { /* 470, 3/3/4/4, 5 */ + priv->magic_not_rop_nr = 0x05; + } else + if (priv->tp_total == 15) { /* 480, 3/4/4/4, 6 */ + priv->magic_not_rop_nr = 0x06; + } + break; + case 0xc3: /* 450, 4/0/0/0, 2 */ + priv->magic_not_rop_nr = 0x03; + break; + case 0xc4: /* 460, 3/4/0/0, 4 */ + priv->magic_not_rop_nr = 0x01; + break; + case 0xc1: /* 2/0/0/0, 1 */ + priv->magic_not_rop_nr = 0x01; + break; + case 0xc8: /* 4/4/3/4, 5 */ + priv->magic_not_rop_nr = 0x06; + break; + case 0xce: /* 4/4/0/0, 4 */ + priv->magic_not_rop_nr = 0x03; + break; + } + + if (!priv->magic_not_rop_nr) { + NV_ERROR(dev, "PGRAPH: unknown config: %d/%d/%d/%d, %d\n", + priv->tp_nr[0], priv->tp_nr[1], priv->tp_nr[2], + priv->tp_nr[3], priv->rop_nr); + /* use 0xc3's values... */ + priv->magic_not_rop_nr = 0x03; + } + + NVOBJ_CLASS(dev, 0x902d, GR); /* 2D */ + NVOBJ_CLASS(dev, 0x9039, GR); /* M2MF */ + NVOBJ_MTHD (dev, 0x9039, 0x0500, nvc0_graph_mthd_page_flip); + NVOBJ_CLASS(dev, 0x9097, GR); /* 3D */ + if (fermi >= 0x9197) + NVOBJ_CLASS(dev, 0x9197, GR); /* 3D (NVC1-) */ + if (fermi >= 0x9297) + NVOBJ_CLASS(dev, 0x9297, GR); /* 3D (NVC8-) */ + NVOBJ_CLASS(dev, 0x90c0, GR); /* COMPUTE */ + return 0; + +error: + nvc0_graph_destroy(dev, NVOBJ_ENGINE_GR); + return ret; +} + +MODULE_FIRMWARE("nouveau/nvc0_fuc409c"); +MODULE_FIRMWARE("nouveau/nvc0_fuc409d"); +MODULE_FIRMWARE("nouveau/nvc0_fuc41ac"); +MODULE_FIRMWARE("nouveau/nvc0_fuc41ad"); +MODULE_FIRMWARE("nouveau/nvc3_fuc409c"); +MODULE_FIRMWARE("nouveau/nvc3_fuc409d"); +MODULE_FIRMWARE("nouveau/nvc3_fuc41ac"); +MODULE_FIRMWARE("nouveau/nvc3_fuc41ad"); +MODULE_FIRMWARE("nouveau/nvc4_fuc409c"); +MODULE_FIRMWARE("nouveau/nvc4_fuc409d"); +MODULE_FIRMWARE("nouveau/nvc4_fuc41ac"); +MODULE_FIRMWARE("nouveau/nvc4_fuc41ad"); +MODULE_FIRMWARE("nouveau/fuc409c"); +MODULE_FIRMWARE("nouveau/fuc409d"); +MODULE_FIRMWARE("nouveau/fuc41ac"); +MODULE_FIRMWARE("nouveau/fuc41ad"); diff --git a/drivers/gpu/drm/nouveau/nvc0_graph.fuc b/drivers/gpu/drm/nouveau/nvc0_graph.fuc new file mode 100644 index 00000000000..2a4b6dc8f9d --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_graph.fuc @@ -0,0 +1,400 @@ +/* fuc microcode util functions for nvc0 PGRAPH + * + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +define(`mmctx_data', `.b32 eval((($2 - 1) << 26) | $1)') +define(`queue_init', `.skip eval((2 * 4) + ((8 * 4) * 2))') + +ifdef(`include_code', ` +// Error codes +define(`E_BAD_COMMAND', 0x01) +define(`E_CMD_OVERFLOW', 0x02) + +// Util macros to help with debugging ucode hangs etc +define(`T_WAIT', 0) +define(`T_MMCTX', 1) +define(`T_STRWAIT', 2) +define(`T_STRINIT', 3) +define(`T_AUTO', 4) +define(`T_CHAN', 5) +define(`T_LOAD', 6) +define(`T_SAVE', 7) +define(`T_LCHAN', 8) +define(`T_LCTXH', 9) + +define(`trace_set', ` + mov $r8 0x83c + shl b32 $r8 6 + clear b32 $r9 + bset $r9 $1 + iowr I[$r8 + 0x000] $r9 // CC_SCRATCH[7] +') + +define(`trace_clr', ` + mov $r8 0x85c + shl b32 $r8 6 + clear b32 $r9 + bset $r9 $1 + iowr I[$r8 + 0x000] $r9 // CC_SCRATCH[7] +') + +// queue_put - add request to queue +// +// In : $r13 queue pointer +// $r14 command +// $r15 data +// +queue_put: + // make sure we have space.. + ld b32 $r8 D[$r13 + 0x0] // GET + ld b32 $r9 D[$r13 + 0x4] // PUT + xor $r8 8 + cmpu b32 $r8 $r9 + bra ne queue_put_next + mov $r15 E_CMD_OVERFLOW + call error + ret + + // store cmd/data on queue + queue_put_next: + and $r8 $r9 7 + shl b32 $r8 3 + add b32 $r8 $r13 + add b32 $r8 8 + st b32 D[$r8 + 0x0] $r14 + st b32 D[$r8 + 0x4] $r15 + + // update PUT + add b32 $r9 1 + and $r9 0xf + st b32 D[$r13 + 0x4] $r9 + ret + +// queue_get - fetch request from queue +// +// In : $r13 queue pointer +// +// Out: $p1 clear on success (data available) +// $r14 command +// $r15 data +// +queue_get: + bset $flags $p1 + ld b32 $r8 D[$r13 + 0x0] // GET + ld b32 $r9 D[$r13 + 0x4] // PUT + cmpu b32 $r8 $r9 + bra e queue_get_done + // fetch first cmd/data pair + and $r9 $r8 7 + shl b32 $r9 3 + add b32 $r9 $r13 + add b32 $r9 8 + ld b32 $r14 D[$r9 + 0x0] + ld b32 $r15 D[$r9 + 0x4] + + // update GET + add b32 $r8 1 + and $r8 0xf + st b32 D[$r13 + 0x0] $r8 + bclr $flags $p1 +queue_get_done: + ret + +// nv_rd32 - read 32-bit value from nv register +// +// In : $r14 register +// Out: $r15 value +// +nv_rd32: + mov $r11 0x728 + shl b32 $r11 6 + mov b32 $r12 $r14 + bset $r12 31 // MMIO_CTRL_PENDING + iowr I[$r11 + 0x000] $r12 // MMIO_CTRL + nv_rd32_wait: + iord $r12 I[$r11 + 0x000] + xbit $r12 $r12 31 + bra ne nv_rd32_wait + mov $r10 6 // DONE_MMIO_RD + call wait_doneo + iord $r15 I[$r11 + 0x100] // MMIO_RDVAL + ret + +// nv_wr32 - write 32-bit value to nv register +// +// In : $r14 register +// $r15 value +// +nv_wr32: + mov $r11 0x728 + shl b32 $r11 6 + iowr I[$r11 + 0x200] $r15 // MMIO_WRVAL + mov b32 $r12 $r14 + bset $r12 31 // MMIO_CTRL_PENDING + bset $r12 30 // MMIO_CTRL_WRITE + iowr I[$r11 + 0x000] $r12 // MMIO_CTRL + nv_wr32_wait: + iord $r12 I[$r11 + 0x000] + xbit $r12 $r12 31 + bra ne nv_wr32_wait + ret + +// (re)set watchdog timer +// +// In : $r15 timeout +// +watchdog_reset: + mov $r8 0x430 + shl b32 $r8 6 + bset $r15 31 + iowr I[$r8 + 0x000] $r15 + ret + +// clear watchdog timer +watchdog_clear: + mov $r8 0x430 + shl b32 $r8 6 + iowr I[$r8 + 0x000] $r0 + ret + +// wait_done{z,o} - wait on FUC_DONE bit to become clear/set +// +// In : $r10 bit to wait on +// +define(`wait_done', ` +$1: + trace_set(T_WAIT); + mov $r8 0x818 + shl b32 $r8 6 + iowr I[$r8 + 0x000] $r10 // CC_SCRATCH[6] = wait bit + wait_done_$1: + mov $r8 0x400 + shl b32 $r8 6 + iord $r8 I[$r8 + 0x000] // DONE + xbit $r8 $r8 $r10 + bra $2 wait_done_$1 + trace_clr(T_WAIT) + ret +') +wait_done(wait_donez, ne) +wait_done(wait_doneo, e) + +// mmctx_size - determine size of a mmio list transfer +// +// In : $r14 mmio list head +// $r15 mmio list tail +// Out: $r15 transfer size (in bytes) +// +mmctx_size: + clear b32 $r9 + nv_mmctx_size_loop: + ld b32 $r8 D[$r14] + shr b32 $r8 26 + add b32 $r8 1 + shl b32 $r8 2 + add b32 $r9 $r8 + add b32 $r14 4 + cmpu b32 $r14 $r15 + bra ne nv_mmctx_size_loop + mov b32 $r15 $r9 + ret + +// mmctx_xfer - execute a list of mmio transfers +// +// In : $r10 flags +// bit 0: direction (0 = save, 1 = load) +// bit 1: set if first transfer +// bit 2: set if last transfer +// $r11 base +// $r12 mmio list head +// $r13 mmio list tail +// $r14 multi_stride +// $r15 multi_mask +// +mmctx_xfer: + trace_set(T_MMCTX) + mov $r8 0x710 + shl b32 $r8 6 + clear b32 $r9 + or $r11 $r11 + bra e mmctx_base_disabled + iowr I[$r8 + 0x000] $r11 // MMCTX_BASE + bset $r9 0 // BASE_EN + mmctx_base_disabled: + or $r14 $r14 + bra e mmctx_multi_disabled + iowr I[$r8 + 0x200] $r14 // MMCTX_MULTI_STRIDE + iowr I[$r8 + 0x300] $r15 // MMCTX_MULTI_MASK + bset $r9 1 // MULTI_EN + mmctx_multi_disabled: + add b32 $r8 0x100 + + xbit $r11 $r10 0 + shl b32 $r11 16 // DIR + bset $r11 12 // QLIMIT = 0x10 + xbit $r14 $r10 1 + shl b32 $r14 17 + or $r11 $r14 // START_TRIGGER + iowr I[$r8 + 0x000] $r11 // MMCTX_CTRL + + // loop over the mmio list, and send requests to the hw + mmctx_exec_loop: + // wait for space in mmctx queue + mmctx_wait_free: + iord $r14 I[$r8 + 0x000] // MMCTX_CTRL + and $r14 0x1f + bra e mmctx_wait_free + + // queue up an entry + ld b32 $r14 D[$r12] + or $r14 $r9 + iowr I[$r8 + 0x300] $r14 + add b32 $r12 4 + cmpu b32 $r12 $r13 + bra ne mmctx_exec_loop + + xbit $r11 $r10 2 + bra ne mmctx_stop + // wait for queue to empty + mmctx_fini_wait: + iord $r11 I[$r8 + 0x000] // MMCTX_CTRL + and $r11 0x1f + cmpu b32 $r11 0x10 + bra ne mmctx_fini_wait + mov $r10 2 // DONE_MMCTX + call wait_donez + bra mmctx_done + mmctx_stop: + xbit $r11 $r10 0 + shl b32 $r11 16 // DIR + bset $r11 12 // QLIMIT = 0x10 + bset $r11 18 // STOP_TRIGGER + iowr I[$r8 + 0x000] $r11 // MMCTX_CTRL + mmctx_stop_wait: + // wait for STOP_TRIGGER to clear + iord $r11 I[$r8 + 0x000] // MMCTX_CTRL + xbit $r11 $r11 18 + bra ne mmctx_stop_wait + mmctx_done: + trace_clr(T_MMCTX) + ret + +// Wait for DONE_STRAND +// +strand_wait: + push $r10 + mov $r10 2 + call wait_donez + pop $r10 + ret + +// unknown - call before issuing strand commands +// +strand_pre: + mov $r8 0x4afc + sethi $r8 0x20000 + mov $r9 0xc + iowr I[$r8] $r9 + call strand_wait + ret + +// unknown - call after issuing strand commands +// +strand_post: + mov $r8 0x4afc + sethi $r8 0x20000 + mov $r9 0xd + iowr I[$r8] $r9 + call strand_wait + ret + +// Selects strand set?! +// +// In: $r14 id +// +strand_set: + mov $r10 0x4ffc + sethi $r10 0x20000 + sub b32 $r11 $r10 0x500 + mov $r12 0xf + iowr I[$r10 + 0x000] $r12 // 0x93c = 0xf + mov $r12 0xb + iowr I[$r11 + 0x000] $r12 // 0x928 = 0xb + call strand_wait + iowr I[$r10 + 0x000] $r14 // 0x93c = <id> + mov $r12 0xa + iowr I[$r11 + 0x000] $r12 // 0x928 = 0xa + call strand_wait + ret + +// Initialise strand context data +// +// In : $r15 context base +// Out: $r15 context size (in bytes) +// +// Strandset(?) 3 hardcoded currently +// +strand_ctx_init: + trace_set(T_STRINIT) + call strand_pre + mov $r14 3 + call strand_set + mov $r10 0x46fc + sethi $r10 0x20000 + add b32 $r11 $r10 0x400 + iowr I[$r10 + 0x100] $r0 // STRAND_FIRST_GENE = 0 + mov $r12 1 + iowr I[$r11 + 0x000] $r12 // STRAND_CMD = LATCH_FIRST_GENE + call strand_wait + sub b32 $r12 $r0 1 + iowr I[$r10 + 0x000] $r12 // STRAND_GENE_CNT = 0xffffffff + mov $r12 2 + iowr I[$r11 + 0x000] $r12 // STRAND_CMD = LATCH_GENE_CNT + call strand_wait + call strand_post + + // read the size of each strand, poke the context offset of + // each into STRAND_{SAVE,LOAD}_SWBASE now, no need to worry + // about it later then. + mov $r8 0x880 + shl b32 $r8 6 + iord $r9 I[$r8 + 0x000] // STRANDS + add b32 $r8 0x2200 + shr b32 $r14 $r15 8 + ctx_init_strand_loop: + iowr I[$r8 + 0x000] $r14 // STRAND_SAVE_SWBASE + iowr I[$r8 + 0x100] $r14 // STRAND_LOAD_SWBASE + iord $r10 I[$r8 + 0x200] // STRAND_SIZE + shr b32 $r10 6 + add b32 $r10 1 + add b32 $r14 $r10 + add b32 $r8 4 + sub b32 $r9 1 + bra ne ctx_init_strand_loop + + shl b32 $r14 8 + sub b32 $r15 $r14 $r15 + trace_clr(T_STRINIT) + ret +') diff --git a/drivers/gpu/drm/nouveau/nvc0_graph.h b/drivers/gpu/drm/nouveau/nvc0_graph.h index 40e26f9c56c..55689e99728 100644 --- a/drivers/gpu/drm/nouveau/nvc0_graph.h +++ b/drivers/gpu/drm/nouveau/nvc0_graph.h @@ -28,13 +28,25 @@ #define GPC_MAX 4 #define TP_MAX 32 -#define ROP_BCAST(r) (0x408800 + (r)) -#define ROP_UNIT(u,r) (0x410000 + (u) * 0x400 + (r)) -#define GPC_BCAST(r) (0x418000 + (r)) -#define GPC_UNIT(t,r) (0x500000 + (t) * 0x8000 + (r)) -#define TP_UNIT(t,m,r) (0x504000 + (t) * 0x8000 + (m) * 0x800 + (r)) +#define ROP_BCAST(r) (0x408800 + (r)) +#define ROP_UNIT(u, r) (0x410000 + (u) * 0x400 + (r)) +#define GPC_BCAST(r) (0x418000 + (r)) +#define GPC_UNIT(t, r) (0x500000 + (t) * 0x8000 + (r)) +#define TP_UNIT(t, m, r) (0x504000 + (t) * 0x8000 + (m) * 0x800 + (r)) + +struct nvc0_graph_fuc { + u32 *data; + u32 size; +}; struct nvc0_graph_priv { + struct nouveau_exec_engine base; + + struct nvc0_graph_fuc fuc409c; + struct nvc0_graph_fuc fuc409d; + struct nvc0_graph_fuc fuc41ac; + struct nvc0_graph_fuc fuc41ad; + u8 gpc_nr; u8 rop_nr; u8 tp_nr[GPC_MAX]; @@ -45,20 +57,39 @@ struct nvc0_graph_priv { struct nouveau_gpuobj *unk4188b4; struct nouveau_gpuobj *unk4188b8; - u8 magic_not_rop_nr; - u32 magicgpc980[4]; - u32 magicgpc918; + u8 magic_not_rop_nr; }; struct nvc0_graph_chan { struct nouveau_gpuobj *grctx; - struct nouveau_gpuobj *unk408004; // 0x418810 too - struct nouveau_gpuobj *unk40800c; // 0x419004 too - struct nouveau_gpuobj *unk418810; // 0x419848 too + struct nouveau_gpuobj *unk408004; /* 0x418810 too */ + struct nouveau_gpuobj *unk40800c; /* 0x419004 too */ + struct nouveau_gpuobj *unk418810; /* 0x419848 too */ struct nouveau_gpuobj *mmio; int mmio_nr; }; int nvc0_grctx_generate(struct nouveau_channel *); +/* nvc0_graph.c uses this also to determine supported chipsets */ +static inline u32 +nvc0_graph_class(struct drm_device *dev) +{ + struct drm_nouveau_private *dev_priv = dev->dev_private; + + switch (dev_priv->chipset) { + case 0xc0: + case 0xc3: + case 0xc4: + case 0xce: /* guess, mmio trace shows only 0x9097 state */ + return 0x9097; + case 0xc1: + return 0x9197; + case 0xc8: + return 0x9297; + default: + return 0; + } +} + #endif diff --git a/drivers/gpu/drm/nouveau/nvc0_grctx.c b/drivers/gpu/drm/nouveau/nvc0_grctx.c index f880ff776db..31018eaf527 100644 --- a/drivers/gpu/drm/nouveau/nvc0_grctx.c +++ b/drivers/gpu/drm/nouveau/nvc0_grctx.c @@ -45,6 +45,9 @@ nv_mthd(struct drm_device *dev, u32 class, u32 mthd, u32 data) static void nvc0_grctx_generate_9097(struct drm_device *dev) { + u32 fermi = nvc0_graph_class(dev); + u32 mthd; + nv_mthd(dev, 0x9097, 0x0800, 0x00000000); nv_mthd(dev, 0x9097, 0x0840, 0x00000000); nv_mthd(dev, 0x9097, 0x0880, 0x00000000); @@ -824,134 +827,10 @@ nvc0_grctx_generate_9097(struct drm_device *dev) nv_mthd(dev, 0x9097, 0x1eb8, 0x00000001); nv_mthd(dev, 0x9097, 0x1ed8, 0x00000001); nv_mthd(dev, 0x9097, 0x1ef8, 0x00000001); - nv_mthd(dev, 0x9097, 0x3400, 0x00000000); - nv_mthd(dev, 0x9097, 0x3404, 0x00000000); - nv_mthd(dev, 0x9097, 0x3408, 0x00000000); - nv_mthd(dev, 0x9097, 0x340c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3410, 0x00000000); - nv_mthd(dev, 0x9097, 0x3414, 0x00000000); - nv_mthd(dev, 0x9097, 0x3418, 0x00000000); - nv_mthd(dev, 0x9097, 0x341c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3420, 0x00000000); - nv_mthd(dev, 0x9097, 0x3424, 0x00000000); - nv_mthd(dev, 0x9097, 0x3428, 0x00000000); - nv_mthd(dev, 0x9097, 0x342c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3430, 0x00000000); - nv_mthd(dev, 0x9097, 0x3434, 0x00000000); - nv_mthd(dev, 0x9097, 0x3438, 0x00000000); - nv_mthd(dev, 0x9097, 0x343c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3440, 0x00000000); - nv_mthd(dev, 0x9097, 0x3444, 0x00000000); - nv_mthd(dev, 0x9097, 0x3448, 0x00000000); - nv_mthd(dev, 0x9097, 0x344c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3450, 0x00000000); - nv_mthd(dev, 0x9097, 0x3454, 0x00000000); - nv_mthd(dev, 0x9097, 0x3458, 0x00000000); - nv_mthd(dev, 0x9097, 0x345c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3460, 0x00000000); - nv_mthd(dev, 0x9097, 0x3464, 0x00000000); - nv_mthd(dev, 0x9097, 0x3468, 0x00000000); - nv_mthd(dev, 0x9097, 0x346c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3470, 0x00000000); - nv_mthd(dev, 0x9097, 0x3474, 0x00000000); - nv_mthd(dev, 0x9097, 0x3478, 0x00000000); - nv_mthd(dev, 0x9097, 0x347c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3480, 0x00000000); - nv_mthd(dev, 0x9097, 0x3484, 0x00000000); - nv_mthd(dev, 0x9097, 0x3488, 0x00000000); - nv_mthd(dev, 0x9097, 0x348c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3490, 0x00000000); - nv_mthd(dev, 0x9097, 0x3494, 0x00000000); - nv_mthd(dev, 0x9097, 0x3498, 0x00000000); - nv_mthd(dev, 0x9097, 0x349c, 0x00000000); - nv_mthd(dev, 0x9097, 0x34a0, 0x00000000); - nv_mthd(dev, 0x9097, 0x34a4, 0x00000000); - nv_mthd(dev, 0x9097, 0x34a8, 0x00000000); - nv_mthd(dev, 0x9097, 0x34ac, 0x00000000); - nv_mthd(dev, 0x9097, 0x34b0, 0x00000000); - nv_mthd(dev, 0x9097, 0x34b4, 0x00000000); - nv_mthd(dev, 0x9097, 0x34b8, 0x00000000); - nv_mthd(dev, 0x9097, 0x34bc, 0x00000000); - nv_mthd(dev, 0x9097, 0x34c0, 0x00000000); - nv_mthd(dev, 0x9097, 0x34c4, 0x00000000); - nv_mthd(dev, 0x9097, 0x34c8, 0x00000000); - nv_mthd(dev, 0x9097, 0x34cc, 0x00000000); - nv_mthd(dev, 0x9097, 0x34d0, 0x00000000); - nv_mthd(dev, 0x9097, 0x34d4, 0x00000000); - nv_mthd(dev, 0x9097, 0x34d8, 0x00000000); - nv_mthd(dev, 0x9097, 0x34dc, 0x00000000); - nv_mthd(dev, 0x9097, 0x34e0, 0x00000000); - nv_mthd(dev, 0x9097, 0x34e4, 0x00000000); - nv_mthd(dev, 0x9097, 0x34e8, 0x00000000); - nv_mthd(dev, 0x9097, 0x34ec, 0x00000000); - nv_mthd(dev, 0x9097, 0x34f0, 0x00000000); - nv_mthd(dev, 0x9097, 0x34f4, 0x00000000); - nv_mthd(dev, 0x9097, 0x34f8, 0x00000000); - nv_mthd(dev, 0x9097, 0x34fc, 0x00000000); - nv_mthd(dev, 0x9097, 0x3500, 0x00000000); - nv_mthd(dev, 0x9097, 0x3504, 0x00000000); - nv_mthd(dev, 0x9097, 0x3508, 0x00000000); - nv_mthd(dev, 0x9097, 0x350c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3510, 0x00000000); - nv_mthd(dev, 0x9097, 0x3514, 0x00000000); - nv_mthd(dev, 0x9097, 0x3518, 0x00000000); - nv_mthd(dev, 0x9097, 0x351c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3520, 0x00000000); - nv_mthd(dev, 0x9097, 0x3524, 0x00000000); - nv_mthd(dev, 0x9097, 0x3528, 0x00000000); - nv_mthd(dev, 0x9097, 0x352c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3530, 0x00000000); - nv_mthd(dev, 0x9097, 0x3534, 0x00000000); - nv_mthd(dev, 0x9097, 0x3538, 0x00000000); - nv_mthd(dev, 0x9097, 0x353c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3540, 0x00000000); - nv_mthd(dev, 0x9097, 0x3544, 0x00000000); - nv_mthd(dev, 0x9097, 0x3548, 0x00000000); - nv_mthd(dev, 0x9097, 0x354c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3550, 0x00000000); - nv_mthd(dev, 0x9097, 0x3554, 0x00000000); - nv_mthd(dev, 0x9097, 0x3558, 0x00000000); - nv_mthd(dev, 0x9097, 0x355c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3560, 0x00000000); - nv_mthd(dev, 0x9097, 0x3564, 0x00000000); - nv_mthd(dev, 0x9097, 0x3568, 0x00000000); - nv_mthd(dev, 0x9097, 0x356c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3570, 0x00000000); - nv_mthd(dev, 0x9097, 0x3574, 0x00000000); - nv_mthd(dev, 0x9097, 0x3578, 0x00000000); - nv_mthd(dev, 0x9097, 0x357c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3580, 0x00000000); - nv_mthd(dev, 0x9097, 0x3584, 0x00000000); - nv_mthd(dev, 0x9097, 0x3588, 0x00000000); - nv_mthd(dev, 0x9097, 0x358c, 0x00000000); - nv_mthd(dev, 0x9097, 0x3590, 0x00000000); - nv_mthd(dev, 0x9097, 0x3594, 0x00000000); - nv_mthd(dev, 0x9097, 0x3598, 0x00000000); - nv_mthd(dev, 0x9097, 0x359c, 0x00000000); - nv_mthd(dev, 0x9097, 0x35a0, 0x00000000); - nv_mthd(dev, 0x9097, 0x35a4, 0x00000000); - nv_mthd(dev, 0x9097, 0x35a8, 0x00000000); - nv_mthd(dev, 0x9097, 0x35ac, 0x00000000); - nv_mthd(dev, 0x9097, 0x35b0, 0x00000000); - nv_mthd(dev, 0x9097, 0x35b4, 0x00000000); - nv_mthd(dev, 0x9097, 0x35b8, 0x00000000); - nv_mthd(dev, 0x9097, 0x35bc, 0x00000000); - nv_mthd(dev, 0x9097, 0x35c0, 0x00000000); - nv_mthd(dev, 0x9097, 0x35c4, 0x00000000); - nv_mthd(dev, 0x9097, 0x35c8, 0x00000000); - nv_mthd(dev, 0x9097, 0x35cc, 0x00000000); - nv_mthd(dev, 0x9097, 0x35d0, 0x00000000); - nv_mthd(dev, 0x9097, 0x35d4, 0x00000000); - nv_mthd(dev, 0x9097, 0x35d8, 0x00000000); - nv_mthd(dev, 0x9097, 0x35dc, 0x00000000); - nv_mthd(dev, 0x9097, 0x35e0, 0x00000000); - nv_mthd(dev, 0x9097, 0x35e4, 0x00000000); - nv_mthd(dev, 0x9097, 0x35e8, 0x00000000); - nv_mthd(dev, 0x9097, 0x35ec, 0x00000000); - nv_mthd(dev, 0x9097, 0x35f0, 0x00000000); - nv_mthd(dev, 0x9097, 0x35f4, 0x00000000); - nv_mthd(dev, 0x9097, 0x35f8, 0x00000000); - nv_mthd(dev, 0x9097, 0x35fc, 0x00000000); + if (fermi == 0x9097) { + for (mthd = 0x3400; mthd <= 0x35fc; mthd += 4) + nv_mthd(dev, 0x9097, mthd, 0x00000000); + } nv_mthd(dev, 0x9097, 0x030c, 0x00000001); nv_mthd(dev, 0x9097, 0x1944, 0x00000000); nv_mthd(dev, 0x9097, 0x1514, 0x00000000); @@ -1321,6 +1200,37 @@ nvc0_grctx_generate_9097(struct drm_device *dev) } static void +nvc0_grctx_generate_9197(struct drm_device *dev) +{ + u32 fermi = nvc0_graph_class(dev); + u32 mthd; + + if (fermi == 0x9197) { + for (mthd = 0x3400; mthd <= 0x35fc; mthd += 4) + nv_mthd(dev, 0x9197, mthd, 0x00000000); + } + nv_mthd(dev, 0x9197, 0x02e4, 0x0000b001); +} + +static void +nvc0_grctx_generate_9297(struct drm_device *dev) +{ + u32 fermi = nvc0_graph_class(dev); + u32 mthd; + + if (fermi == 0x9297) { + for (mthd = 0x3400; mthd <= 0x35fc; mthd += 4) + nv_mthd(dev, 0x9297, mthd, 0x00000000); + } + nv_mthd(dev, 0x9297, 0x036c, 0x00000000); + nv_mthd(dev, 0x9297, 0x0370, 0x00000000); + nv_mthd(dev, 0x9297, 0x07a4, 0x00000000); + nv_mthd(dev, 0x9297, 0x07a8, 0x00000000); + nv_mthd(dev, 0x9297, 0x0374, 0x00000000); + nv_mthd(dev, 0x9297, 0x0378, 0x00000020); +} + +static void nvc0_grctx_generate_902d(struct drm_device *dev) { nv_mthd(dev, 0x902d, 0x0200, 0x000000cf); @@ -1559,8 +1469,15 @@ nvc0_grctx_generate_unk47xx(struct drm_device *dev) static void nvc0_grctx_generate_shaders(struct drm_device *dev) { - nv_wr32(dev, 0x405800, 0x078000bf); - nv_wr32(dev, 0x405830, 0x02180000); + struct drm_nouveau_private *dev_priv = dev->dev_private; + + if (dev_priv->chipset != 0xc1) { + nv_wr32(dev, 0x405800, 0x078000bf); + nv_wr32(dev, 0x405830, 0x02180000); + } else { + nv_wr32(dev, 0x405800, 0x0f8000bf); + nv_wr32(dev, 0x405830, 0x02180218); + } nv_wr32(dev, 0x405834, 0x00000000); nv_wr32(dev, 0x405838, 0x00000000); nv_wr32(dev, 0x405854, 0x00000000); @@ -1586,10 +1503,16 @@ nvc0_grctx_generate_unk60xx(struct drm_device *dev) static void nvc0_grctx_generate_unk64xx(struct drm_device *dev) { + struct drm_nouveau_private *dev_priv = dev->dev_private; + nv_wr32(dev, 0x4064a8, 0x00000000); nv_wr32(dev, 0x4064ac, 0x00003fff); nv_wr32(dev, 0x4064b4, 0x00000000); nv_wr32(dev, 0x4064b8, 0x00000000); + if (dev_priv->chipset == 0xc1) { + nv_wr32(dev, 0x4064c0, 0x80140078); + nv_wr32(dev, 0x4064c4, 0x0086ffff); + } } static void @@ -1622,21 +1545,14 @@ static void nvc0_grctx_generate_rop(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; + int chipset = dev_priv->chipset; - // ROPC_BROADCAST + /* ROPC_BROADCAST */ nv_wr32(dev, 0x408800, 0x02802a3c); nv_wr32(dev, 0x408804, 0x00000040); - nv_wr32(dev, 0x408808, 0x0003e00d); - switch (dev_priv->chipset) { - case 0xc0: - nv_wr32(dev, 0x408900, 0x0080b801); - break; - case 0xc3: - case 0xc4: - nv_wr32(dev, 0x408900, 0x3080b801); - break; - } - nv_wr32(dev, 0x408904, 0x02000001); + nv_wr32(dev, 0x408808, chipset != 0xc1 ? 0x0003e00d : 0x1003e005); + nv_wr32(dev, 0x408900, 0x3080b801); + nv_wr32(dev, 0x408904, chipset != 0xc1 ? 0x02000001 : 0x62000001); nv_wr32(dev, 0x408908, 0x00c80929); nv_wr32(dev, 0x40890c, 0x00000000); nv_wr32(dev, 0x408980, 0x0000011d); @@ -1645,9 +1561,11 @@ nvc0_grctx_generate_rop(struct drm_device *dev) static void nvc0_grctx_generate_gpc(struct drm_device *dev) { + struct drm_nouveau_private *dev_priv = dev->dev_private; + int chipset = dev_priv->chipset; int i; - // GPC_BROADCAST + /* GPC_BROADCAST */ nv_wr32(dev, 0x418380, 0x00000016); nv_wr32(dev, 0x418400, 0x38004e00); nv_wr32(dev, 0x418404, 0x71e0ffff); @@ -1676,7 +1594,7 @@ nvc0_grctx_generate_gpc(struct drm_device *dev) nv_wr32(dev, 0x41880c, 0x00000000); nv_wr32(dev, 0x418810, 0x00000000); nv_wr32(dev, 0x418828, 0x00008442); - nv_wr32(dev, 0x418830, 0x00000001); + nv_wr32(dev, 0x418830, chipset != 0xc1 ? 0x00000001 : 0x10000001); nv_wr32(dev, 0x4188d8, 0x00000008); nv_wr32(dev, 0x4188e0, 0x01000000); nv_wr32(dev, 0x4188e8, 0x00000000); @@ -1684,7 +1602,7 @@ nvc0_grctx_generate_gpc(struct drm_device *dev) nv_wr32(dev, 0x4188f0, 0x00000000); nv_wr32(dev, 0x4188f4, 0x00000000); nv_wr32(dev, 0x4188f8, 0x00000000); - nv_wr32(dev, 0x4188fc, 0x00100000); + nv_wr32(dev, 0x4188fc, chipset != 0xc1 ? 0x00100000 : 0x00100018); nv_wr32(dev, 0x41891c, 0x00ff00ff); nv_wr32(dev, 0x418924, 0x00000000); nv_wr32(dev, 0x418928, 0x00ffff00); @@ -1715,6 +1633,8 @@ nvc0_grctx_generate_gpc(struct drm_device *dev) nv_wr32(dev, 0x418c24, 0x00000000); nv_wr32(dev, 0x418c28, 0x00000000); nv_wr32(dev, 0x418c2c, 0x00000000); + if (chipset == 0xc1) + nv_wr32(dev, 0x418c6c, 0x00000001); nv_wr32(dev, 0x418c80, 0x20200004); nv_wr32(dev, 0x418c8c, 0x00000001); nv_wr32(dev, 0x419000, 0x00000780); @@ -1727,10 +1647,13 @@ static void nvc0_grctx_generate_tp(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; + int chipset = dev_priv->chipset; - // GPC_BROADCAST.TP_BROADCAST + /* GPC_BROADCAST.TP_BROADCAST */ + nv_wr32(dev, 0x419818, 0x00000000); + nv_wr32(dev, 0x41983c, 0x00038bc7); nv_wr32(dev, 0x419848, 0x00000000); - nv_wr32(dev, 0x419864, 0x0000012a); + nv_wr32(dev, 0x419864, chipset != 0xc1 ? 0x0000012a : 0x00000129); nv_wr32(dev, 0x419888, 0x00000000); nv_wr32(dev, 0x419a00, 0x000001f0); nv_wr32(dev, 0x419a04, 0x00000001); @@ -1740,8 +1663,8 @@ nvc0_grctx_generate_tp(struct drm_device *dev) nv_wr32(dev, 0x419a14, 0x00000200); nv_wr32(dev, 0x419a1c, 0x00000000); nv_wr32(dev, 0x419a20, 0x00000800); - if (dev_priv->chipset != 0xc0) - nv_wr32(dev, 0x00419ac4, 0x0007f440); // 0xc3 + if (chipset != 0xc0 && chipset != 0xc8) + nv_wr32(dev, 0x00419ac4, 0x0007f440); nv_wr32(dev, 0x419b00, 0x0a418820); nv_wr32(dev, 0x419b04, 0x062080e6); nv_wr32(dev, 0x419b08, 0x020398a4); @@ -1749,17 +1672,19 @@ nvc0_grctx_generate_tp(struct drm_device *dev) nv_wr32(dev, 0x419b10, 0x0a418820); nv_wr32(dev, 0x419b14, 0x000000e6); nv_wr32(dev, 0x419bd0, 0x00900103); - nv_wr32(dev, 0x419be0, 0x00000001); + nv_wr32(dev, 0x419be0, chipset != 0xc1 ? 0x00000001 : 0x00400001); nv_wr32(dev, 0x419be4, 0x00000000); nv_wr32(dev, 0x419c00, 0x00000002); nv_wr32(dev, 0x419c04, 0x00000006); nv_wr32(dev, 0x419c08, 0x00000002); nv_wr32(dev, 0x419c20, 0x00000000); - nv_wr32(dev, 0x419cbc, 0x28137606); + nv_wr32(dev, 0x419cb0, 0x00060048); //XXX: 0xce 0x00020048 nv_wr32(dev, 0x419ce8, 0x00000000); nv_wr32(dev, 0x419cf4, 0x00000183); - nv_wr32(dev, 0x419d20, 0x02180000); + nv_wr32(dev, 0x419d20, chipset != 0xc1 ? 0x02180000 : 0x12180000); nv_wr32(dev, 0x419d24, 0x00001fff); + if (chipset == 0xc1) + nv_wr32(dev, 0x419d44, 0x02180218); nv_wr32(dev, 0x419e04, 0x00000000); nv_wr32(dev, 0x419e08, 0x00000000); nv_wr32(dev, 0x419e0c, 0x00000000); @@ -1785,11 +1710,11 @@ nvc0_grctx_generate_tp(struct drm_device *dev) nv_wr32(dev, 0x419e8c, 0x00000000); nv_wr32(dev, 0x419e90, 0x00000000); nv_wr32(dev, 0x419e98, 0x00000000); - if (dev_priv->chipset != 0xc0) + if (chipset != 0xc0 && chipset != 0xc8) nv_wr32(dev, 0x419ee0, 0x00011110); nv_wr32(dev, 0x419f50, 0x00000000); nv_wr32(dev, 0x419f54, 0x00000000); - if (dev_priv->chipset != 0xc0) + if (chipset != 0xc0 && chipset != 0xc8) nv_wr32(dev, 0x419f58, 0x00000000); } @@ -1797,10 +1722,11 @@ int nvc0_grctx_generate(struct nouveau_channel *chan) { struct drm_nouveau_private *dev_priv = chan->dev->dev_private; - struct nvc0_graph_priv *priv = dev_priv->engine.graph.priv; - struct nvc0_graph_chan *grch = chan->pgraph_ctx; + struct nvc0_graph_priv *priv = nv_engine(chan->dev, NVOBJ_ENGINE_GR); + struct nvc0_graph_chan *grch = chan->engctx[NVOBJ_ENGINE_GR]; struct drm_device *dev = chan->dev; int i, gpc, tp, id; + u32 fermi = nvc0_graph_class(dev); u32 r000260, tmp; r000260 = nv_rd32(dev, 0x000260); @@ -1857,10 +1783,11 @@ nvc0_grctx_generate(struct nouveau_channel *chan) nv_wr32(dev, 0x40587c, 0x00000000); if (1) { - const u8 chipset_tp_max[] = { 16, 0, 0, 4, 8 }; + const u8 chipset_tp_max[] = { 16, 4, 0, 4, 8, 0, 0, 0, + 16, 0, 0, 0, 0, 0, 8, 0 }; u8 max = chipset_tp_max[dev_priv->chipset & 0x0f]; u8 tpnr[GPC_MAX]; - u8 data[32]; + u8 data[TP_MAX]; memcpy(tpnr, priv->tp_nr, sizeof(priv->tp_nr)); memset(data, 0x1f, sizeof(data)); @@ -1912,13 +1839,13 @@ nvc0_grctx_generate(struct nouveau_channel *chan) for (i = 1; i < 7; i++) data2[1] |= ((1 << (i + 5)) % ntpcv) << ((i - 1) * 5); - // GPC_BROADCAST + /* GPC_BROADCAST */ nv_wr32(dev, 0x418bb8, (priv->tp_total << 8) | priv->magic_not_rop_nr); for (i = 0; i < 6; i++) nv_wr32(dev, 0x418b08 + (i * 4), data[i]); - // GPC_BROADCAST.TP_BROADCAST + /* GPC_BROADCAST.TP_BROADCAST */ nv_wr32(dev, 0x419bd0, (priv->tp_total << 8) | priv->magic_not_rop_nr | data2[0]); @@ -1926,7 +1853,7 @@ nvc0_grctx_generate(struct nouveau_channel *chan) for (i = 0; i < 6; i++) nv_wr32(dev, 0x419b00 + (i * 4), data[i]); - // UNK78xx + /* UNK78xx */ nv_wr32(dev, 0x4078bc, (priv->tp_total << 8) | priv->magic_not_rop_nr); for (i = 0; i < 6; i++) @@ -1944,7 +1871,7 @@ nvc0_grctx_generate(struct nouveau_channel *chan) gpc = -1; for (i = 0, gpc = -1; i < 32; i++) { int ltp = i * (priv->tp_total - 1) / 32; - + do { gpc = (gpc + 1) % priv->gpc_nr; } while (!tpnr[gpc]); @@ -2633,6 +2560,8 @@ nvc0_grctx_generate(struct nouveau_channel *chan) nv_icmd(dev, 0x0000053f, 0xffff0000); nv_icmd(dev, 0x00000585, 0x0000003f); nv_icmd(dev, 0x00000576, 0x00000003); + if (dev_priv->chipset == 0xc1) + nv_icmd(dev, 0x0000057b, 0x00000059); nv_icmd(dev, 0x00000586, 0x00000040); nv_icmd(dev, 0x00000582, 0x00000080); nv_icmd(dev, 0x00000583, 0x00000080); @@ -2865,6 +2794,10 @@ nvc0_grctx_generate(struct nouveau_channel *chan) nv_wr32(dev, 0x404154, 0x00000400); nvc0_grctx_generate_9097(dev); + if (fermi >= 0x9197) + nvc0_grctx_generate_9197(dev); + if (fermi >= 0x9297) + nvc0_grctx_generate_9297(dev); nvc0_grctx_generate_902d(dev); nvc0_grctx_generate_9039(dev); nvc0_grctx_generate_90c0(dev); diff --git a/drivers/gpu/drm/nouveau/nvc0_grgpc.fuc b/drivers/gpu/drm/nouveau/nvc0_grgpc.fuc new file mode 100644 index 00000000000..0ec2add72a7 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_grgpc.fuc @@ -0,0 +1,474 @@ +/* fuc microcode for nvc0 PGRAPH/GPC + * + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +/* To build: + * m4 nvc0_grgpc.fuc | envyas -a -w -m fuc -V nva3 -o nvc0_grgpc.fuc.h + */ + +/* TODO + * - bracket certain functions with scratch writes, useful for debugging + * - watchdog timer around ctx operations + */ + +.section nvc0_grgpc_data +include(`nvc0_graph.fuc') +gpc_id: .b32 0 +gpc_mmio_list_head: .b32 0 +gpc_mmio_list_tail: .b32 0 + +tpc_count: .b32 0 +tpc_mask: .b32 0 +tpc_mmio_list_head: .b32 0 +tpc_mmio_list_tail: .b32 0 + +cmd_queue: queue_init + +// chipset descriptions +chipsets: +.b8 0xc0 0 0 0 +.b16 nvc0_gpc_mmio_head +.b16 nvc0_gpc_mmio_tail +.b16 nvc0_tpc_mmio_head +.b16 nvc0_tpc_mmio_tail +.b8 0xc1 0 0 0 +.b16 nvc0_gpc_mmio_head +.b16 nvc1_gpc_mmio_tail +.b16 nvc0_tpc_mmio_head +.b16 nvc1_tpc_mmio_tail +.b8 0xc3 0 0 0 +.b16 nvc0_gpc_mmio_head +.b16 nvc0_gpc_mmio_tail +.b16 nvc0_tpc_mmio_head +.b16 nvc3_tpc_mmio_tail +.b8 0xc4 0 0 0 +.b16 nvc0_gpc_mmio_head +.b16 nvc0_gpc_mmio_tail +.b16 nvc0_tpc_mmio_head +.b16 nvc3_tpc_mmio_tail +.b8 0xc8 0 0 0 +.b16 nvc0_gpc_mmio_head +.b16 nvc0_gpc_mmio_tail +.b16 nvc0_tpc_mmio_head +.b16 nvc0_tpc_mmio_tail +.b8 0xce 0 0 0 +.b16 nvc0_gpc_mmio_head +.b16 nvc0_gpc_mmio_tail +.b16 nvc0_tpc_mmio_head +.b16 nvc3_tpc_mmio_tail +.b8 0 0 0 0 + +// GPC mmio lists +nvc0_gpc_mmio_head: +mmctx_data(0x000380, 1) +mmctx_data(0x000400, 6) +mmctx_data(0x000450, 9) +mmctx_data(0x000600, 1) +mmctx_data(0x000684, 1) +mmctx_data(0x000700, 5) +mmctx_data(0x000800, 1) +mmctx_data(0x000808, 3) +mmctx_data(0x000828, 1) +mmctx_data(0x000830, 1) +mmctx_data(0x0008d8, 1) +mmctx_data(0x0008e0, 1) +mmctx_data(0x0008e8, 6) +mmctx_data(0x00091c, 1) +mmctx_data(0x000924, 3) +mmctx_data(0x000b00, 1) +mmctx_data(0x000b08, 6) +mmctx_data(0x000bb8, 1) +mmctx_data(0x000c08, 1) +mmctx_data(0x000c10, 8) +mmctx_data(0x000c80, 1) +mmctx_data(0x000c8c, 1) +mmctx_data(0x001000, 3) +mmctx_data(0x001014, 1) +nvc0_gpc_mmio_tail: +mmctx_data(0x000c6c, 1); +nvc1_gpc_mmio_tail: + +// TPC mmio lists +nvc0_tpc_mmio_head: +mmctx_data(0x000018, 1) +mmctx_data(0x00003c, 1) +mmctx_data(0x000048, 1) +mmctx_data(0x000064, 1) +mmctx_data(0x000088, 1) +mmctx_data(0x000200, 6) +mmctx_data(0x00021c, 2) +mmctx_data(0x000300, 6) +mmctx_data(0x0003d0, 1) +mmctx_data(0x0003e0, 2) +mmctx_data(0x000400, 3) +mmctx_data(0x000420, 1) +mmctx_data(0x0004b0, 1) +mmctx_data(0x0004e8, 1) +mmctx_data(0x0004f4, 1) +mmctx_data(0x000520, 2) +mmctx_data(0x000604, 4) +mmctx_data(0x000644, 20) +mmctx_data(0x000698, 1) +mmctx_data(0x000750, 2) +nvc0_tpc_mmio_tail: +mmctx_data(0x000758, 1) +mmctx_data(0x0002c4, 1) +mmctx_data(0x0004bc, 1) +mmctx_data(0x0006e0, 1) +nvc3_tpc_mmio_tail: +mmctx_data(0x000544, 1) +nvc1_tpc_mmio_tail: + + +.section nvc0_grgpc_code +bra init +define(`include_code') +include(`nvc0_graph.fuc') + +// reports an exception to the host +// +// In: $r15 error code (see nvc0_graph.fuc) +// +error: + push $r14 + mov $r14 -0x67ec // 0x9814 + sethi $r14 0x400000 + call nv_wr32 // HUB_CTXCTL_CC_SCRATCH[5] = error code + add b32 $r14 0x41c + mov $r15 1 + call nv_wr32 // HUB_CTXCTL_INTR_UP_SET + pop $r14 + ret + +// GPC fuc initialisation, executed by triggering ucode start, will +// fall through to main loop after completion. +// +// Input: +// CC_SCRATCH[0]: chipset (PMC_BOOT_0 read returns 0x0bad0bad... sigh) +// CC_SCRATCH[1]: context base +// +// Output: +// CC_SCRATCH[0]: +// 31:31: set to signal completion +// CC_SCRATCH[1]: +// 31:0: GPC context size +// +init: + clear b32 $r0 + mov $sp $r0 + + // enable fifo access + mov $r1 0x1200 + mov $r2 2 + iowr I[$r1 + 0x000] $r2 // FIFO_ENABLE + + // setup i0 handler, and route all interrupts to it + mov $r1 ih + mov $iv0 $r1 + mov $r1 0x400 + iowr I[$r1 + 0x300] $r0 // INTR_DISPATCH + + // enable fifo interrupt + mov $r2 4 + iowr I[$r1 + 0x000] $r2 // INTR_EN_SET + + // enable interrupts + bset $flags ie0 + + // figure out which GPC we are, and how many TPCs we have + mov $r1 0x608 + shl b32 $r1 6 + iord $r2 I[$r1 + 0x000] // UNITS + mov $r3 1 + and $r2 0x1f + shl b32 $r3 $r2 + sub b32 $r3 1 + st b32 D[$r0 + tpc_count] $r2 + st b32 D[$r0 + tpc_mask] $r3 + add b32 $r1 0x400 + iord $r2 I[$r1 + 0x000] // MYINDEX + st b32 D[$r0 + gpc_id] $r2 + + // find context data for this chipset + mov $r2 0x800 + shl b32 $r2 6 + iord $r2 I[$r2 + 0x000] // CC_SCRATCH[0] + mov $r1 chipsets - 12 + init_find_chipset: + add b32 $r1 12 + ld b32 $r3 D[$r1 + 0x00] + cmpu b32 $r3 $r2 + bra e init_context + cmpu b32 $r3 0 + bra ne init_find_chipset + // unknown chipset + ret + + // initialise context base, and size tracking + init_context: + mov $r2 0x800 + shl b32 $r2 6 + iord $r2 I[$r2 + 0x100] // CC_SCRATCH[1], initial base + clear b32 $r3 // track GPC context size here + + // set mmctx base addresses now so we don't have to do it later, + // they don't currently ever change + mov $r4 0x700 + shl b32 $r4 6 + shr b32 $r5 $r2 8 + iowr I[$r4 + 0x000] $r5 // MMCTX_SAVE_SWBASE + iowr I[$r4 + 0x100] $r5 // MMCTX_LOAD_SWBASE + + // calculate GPC mmio context size, store the chipset-specific + // mmio list pointers somewhere we can get at them later without + // re-parsing the chipset list + clear b32 $r14 + clear b32 $r15 + ld b16 $r14 D[$r1 + 4] + ld b16 $r15 D[$r1 + 6] + st b16 D[$r0 + gpc_mmio_list_head] $r14 + st b16 D[$r0 + gpc_mmio_list_tail] $r15 + call mmctx_size + add b32 $r2 $r15 + add b32 $r3 $r15 + + // calculate per-TPC mmio context size, store the list pointers + ld b16 $r14 D[$r1 + 8] + ld b16 $r15 D[$r1 + 10] + st b16 D[$r0 + tpc_mmio_list_head] $r14 + st b16 D[$r0 + tpc_mmio_list_tail] $r15 + call mmctx_size + ld b32 $r14 D[$r0 + tpc_count] + mulu $r14 $r15 + add b32 $r2 $r14 + add b32 $r3 $r14 + + // round up base/size to 256 byte boundary (for strand SWBASE) + add b32 $r4 0x1300 + shr b32 $r3 2 + iowr I[$r4 + 0x000] $r3 // MMCTX_LOAD_COUNT, wtf for?!? + shr b32 $r2 8 + shr b32 $r3 6 + add b32 $r2 1 + add b32 $r3 1 + shl b32 $r2 8 + shl b32 $r3 8 + + // calculate size of strand context data + mov b32 $r15 $r2 + call strand_ctx_init + add b32 $r3 $r15 + + // save context size, and tell HUB we're done + mov $r1 0x800 + shl b32 $r1 6 + iowr I[$r1 + 0x100] $r3 // CC_SCRATCH[1] = context size + add b32 $r1 0x800 + clear b32 $r2 + bset $r2 31 + iowr I[$r1 + 0x000] $r2 // CC_SCRATCH[0] |= 0x80000000 + +// Main program loop, very simple, sleeps until woken up by the interrupt +// handler, pulls a command from the queue and executes its handler +// +main: + bset $flags $p0 + sleep $p0 + mov $r13 cmd_queue + call queue_get + bra $p1 main + + // 0x0000-0x0003 are all context transfers + cmpu b32 $r14 0x04 + bra nc main_not_ctx_xfer + // fetch $flags and mask off $p1/$p2 + mov $r1 $flags + mov $r2 0x0006 + not b32 $r2 + and $r1 $r2 + // set $p1/$p2 according to transfer type + shl b32 $r14 1 + or $r1 $r14 + mov $flags $r1 + // transfer context data + call ctx_xfer + bra main + + main_not_ctx_xfer: + shl b32 $r15 $r14 16 + or $r15 E_BAD_COMMAND + call error + bra main + +// interrupt handler +ih: + push $r8 + mov $r8 $flags + push $r8 + push $r9 + push $r10 + push $r11 + push $r13 + push $r14 + push $r15 + + // incoming fifo command? + iord $r10 I[$r0 + 0x200] // INTR + and $r11 $r10 0x00000004 + bra e ih_no_fifo + // queue incoming fifo command for later processing + mov $r11 0x1900 + mov $r13 cmd_queue + iord $r14 I[$r11 + 0x100] // FIFO_CMD + iord $r15 I[$r11 + 0x000] // FIFO_DATA + call queue_put + add b32 $r11 0x400 + mov $r14 1 + iowr I[$r11 + 0x000] $r14 // FIFO_ACK + + // ack, and wake up main() + ih_no_fifo: + iowr I[$r0 + 0x100] $r10 // INTR_ACK + + pop $r15 + pop $r14 + pop $r13 + pop $r11 + pop $r10 + pop $r9 + pop $r8 + mov $flags $r8 + pop $r8 + bclr $flags $p0 + iret + +// Set this GPC's bit in HUB_BAR, used to signal completion of various +// activities to the HUB fuc +// +hub_barrier_done: + mov $r15 1 + ld b32 $r14 D[$r0 + gpc_id] + shl b32 $r15 $r14 + mov $r14 -0x6be8 // 0x409418 - HUB_BAR_SET + sethi $r14 0x400000 + call nv_wr32 + ret + +// Disables various things, waits a bit, and re-enables them.. +// +// Not sure how exactly this helps, perhaps "ENABLE" is not such a +// good description for the bits we turn off? Anyways, without this, +// funny things happen. +// +ctx_redswitch: + mov $r14 0x614 + shl b32 $r14 6 + mov $r15 0x020 + iowr I[$r14] $r15 // GPC_RED_SWITCH = POWER + mov $r15 8 + ctx_redswitch_delay: + sub b32 $r15 1 + bra ne ctx_redswitch_delay + mov $r15 0xa20 + iowr I[$r14] $r15 // GPC_RED_SWITCH = UNK11, ENABLE, POWER + ret + +// Transfer GPC context data between GPU and storage area +// +// In: $r15 context base address +// $p1 clear on save, set on load +// $p2 set if opposite direction done/will be done, so: +// on save it means: "a load will follow this save" +// on load it means: "a save preceeded this load" +// +ctx_xfer: + // set context base address + mov $r1 0xa04 + shl b32 $r1 6 + iowr I[$r1 + 0x000] $r15// MEM_BASE + bra not $p1 ctx_xfer_not_load + call ctx_redswitch + ctx_xfer_not_load: + + // strands + mov $r1 0x4afc + sethi $r1 0x20000 + mov $r2 0xc + iowr I[$r1] $r2 // STRAND_CMD(0x3f) = 0x0c + call strand_wait + mov $r2 0x47fc + sethi $r2 0x20000 + iowr I[$r2] $r0 // STRAND_FIRST_GENE(0x3f) = 0x00 + xbit $r2 $flags $p1 + add b32 $r2 3 + iowr I[$r1] $r2 // STRAND_CMD(0x3f) = 0x03/0x04 (SAVE/LOAD) + + // mmio context + xbit $r10 $flags $p1 // direction + or $r10 2 // first + mov $r11 0x0000 + sethi $r11 0x500000 + ld b32 $r12 D[$r0 + gpc_id] + shl b32 $r12 15 + add b32 $r11 $r12 // base = NV_PGRAPH_GPCn + ld b32 $r12 D[$r0 + gpc_mmio_list_head] + ld b32 $r13 D[$r0 + gpc_mmio_list_tail] + mov $r14 0 // not multi + call mmctx_xfer + + // per-TPC mmio context + xbit $r10 $flags $p1 // direction + or $r10 4 // last + mov $r11 0x4000 + sethi $r11 0x500000 // base = NV_PGRAPH_GPC0_TPC0 + ld b32 $r12 D[$r0 + gpc_id] + shl b32 $r12 15 + add b32 $r11 $r12 // base = NV_PGRAPH_GPCn_TPC0 + ld b32 $r12 D[$r0 + tpc_mmio_list_head] + ld b32 $r13 D[$r0 + tpc_mmio_list_tail] + ld b32 $r15 D[$r0 + tpc_mask] + mov $r14 0x800 // stride = 0x800 + call mmctx_xfer + + // wait for strands to finish + call strand_wait + + // if load, or a save without a load following, do some + // unknown stuff that's done after finishing a block of + // strand commands + bra $p1 ctx_xfer_post + bra not $p2 ctx_xfer_done + ctx_xfer_post: + mov $r1 0x4afc + sethi $r1 0x20000 + mov $r2 0xd + iowr I[$r1] $r2 // STRAND_CMD(0x3f) = 0x0d + call strand_wait + + // mark completion in HUB's barrier + ctx_xfer_done: + call hub_barrier_done + ret + +.align 256 diff --git a/drivers/gpu/drm/nouveau/nvc0_grgpc.fuc.h b/drivers/gpu/drm/nouveau/nvc0_grgpc.fuc.h new file mode 100644 index 00000000000..1896c898f5b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_grgpc.fuc.h @@ -0,0 +1,483 @@ +uint32_t nvc0_grgpc_data[] = { + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x000000c0, + 0x011000b0, + 0x01640114, + 0x000000c1, + 0x011400b0, + 0x01780114, + 0x000000c3, + 0x011000b0, + 0x01740114, + 0x000000c4, + 0x011000b0, + 0x01740114, + 0x000000c8, + 0x011000b0, + 0x01640114, + 0x000000ce, + 0x011000b0, + 0x01740114, + 0x00000000, + 0x00000380, + 0x14000400, + 0x20000450, + 0x00000600, + 0x00000684, + 0x10000700, + 0x00000800, + 0x08000808, + 0x00000828, + 0x00000830, + 0x000008d8, + 0x000008e0, + 0x140008e8, + 0x0000091c, + 0x08000924, + 0x00000b00, + 0x14000b08, + 0x00000bb8, + 0x00000c08, + 0x1c000c10, + 0x00000c80, + 0x00000c8c, + 0x08001000, + 0x00001014, + 0x00000c6c, + 0x00000018, + 0x0000003c, + 0x00000048, + 0x00000064, + 0x00000088, + 0x14000200, + 0x0400021c, + 0x14000300, + 0x000003d0, + 0x040003e0, + 0x08000400, + 0x00000420, + 0x000004b0, + 0x000004e8, + 0x000004f4, + 0x04000520, + 0x0c000604, + 0x4c000644, + 0x00000698, + 0x04000750, + 0x00000758, + 0x000002c4, + 0x000004bc, + 0x000006e0, + 0x00000544, +}; + +uint32_t nvc0_grgpc_code[] = { + 0x03060ef5, + 0x9800d898, + 0x86f001d9, + 0x0489b808, + 0xf00c1bf4, + 0x21f502f7, + 0x00f802ec, + 0xb60798c4, + 0x8dbb0384, + 0x0880b600, + 0x80008e80, + 0x90b6018f, + 0x0f94f001, + 0xf801d980, + 0x0131f400, + 0x9800d898, + 0x89b801d9, + 0x210bf404, + 0xb60789c4, + 0x9dbb0394, + 0x0890b600, + 0x98009e98, + 0x80b6019f, + 0x0f84f001, + 0xf400d880, + 0x00f80132, + 0x0728b7f1, + 0xb906b4b6, + 0xc9f002ec, + 0x00bcd01f, + 0xc800bccf, + 0x1bf41fcc, + 0x06a7f0fa, + 0x010321f5, + 0xf840bfcf, + 0x28b7f100, + 0x06b4b607, + 0xb980bfd0, + 0xc9f002ec, + 0x1ec9f01f, + 0xcf00bcd0, + 0xccc800bc, + 0xfa1bf41f, + 0x87f100f8, + 0x84b60430, + 0x1ff9f006, + 0xf8008fd0, + 0x3087f100, + 0x0684b604, + 0xf80080d0, + 0x3c87f100, + 0x0684b608, + 0x99f094bd, + 0x0089d000, + 0x081887f1, + 0xd00684b6, + 0x87f1008a, + 0x84b60400, + 0x0088cf06, + 0xf4888aff, + 0x87f1f31b, + 0x84b6085c, + 0xf094bd06, + 0x89d00099, + 0xf100f800, + 0xb6083c87, + 0x94bd0684, + 0xd00099f0, + 0x87f10089, + 0x84b60818, + 0x008ad006, + 0x040087f1, + 0xcf0684b6, + 0x8aff0088, + 0xf30bf488, + 0x085c87f1, + 0xbd0684b6, + 0x0099f094, + 0xf80089d0, + 0x9894bd00, + 0x85b600e8, + 0x0180b61a, + 0xbb0284b6, + 0xe0b60098, + 0x04efb804, + 0xb9eb1bf4, + 0x00f8029f, + 0x083c87f1, + 0xbd0684b6, + 0x0199f094, + 0xf10089d0, + 0xb6071087, + 0x94bd0684, + 0xf405bbfd, + 0x8bd0090b, + 0x0099f000, + 0xf405eefd, + 0x8ed00c0b, + 0xc08fd080, + 0xb70199f0, + 0xc8010080, + 0xb4b600ab, + 0x0cb9f010, + 0xb601aec8, + 0xbefd11e4, + 0x008bd005, + 0xf0008ecf, + 0x0bf41fe4, + 0x00ce98fa, + 0xd005e9fd, + 0xc0b6c08e, + 0x04cdb804, + 0xc8e81bf4, + 0x1bf402ab, + 0x008bcf18, + 0xb01fb4f0, + 0x1bf410b4, + 0x02a7f0f7, + 0xf4c921f4, + 0xabc81b0e, + 0x10b4b600, + 0xf00cb9f0, + 0x8bd012b9, + 0x008bcf00, + 0xf412bbc8, + 0x87f1fa1b, + 0x84b6085c, + 0xf094bd06, + 0x89d00199, + 0xf900f800, + 0x02a7f0a0, + 0xfcc921f4, + 0xf100f8a0, + 0xf04afc87, + 0x97f00283, + 0x0089d00c, + 0x020721f5, + 0x87f100f8, + 0x83f04afc, + 0x0d97f002, + 0xf50089d0, + 0xf8020721, + 0xfca7f100, + 0x02a3f04f, + 0x0500aba2, + 0xd00fc7f0, + 0xc7f000ac, + 0x00bcd00b, + 0x020721f5, + 0xf000aed0, + 0xbcd00ac7, + 0x0721f500, + 0xf100f802, + 0xb6083c87, + 0x94bd0684, + 0xd00399f0, + 0x21f50089, + 0xe7f00213, + 0x3921f503, + 0xfca7f102, + 0x02a3f046, + 0x0400aba0, + 0xf040a0d0, + 0xbcd001c7, + 0x0721f500, + 0x010c9202, + 0xf000acd0, + 0xbcd002c7, + 0x0721f500, + 0x2621f502, + 0x8087f102, + 0x0684b608, + 0xb70089cf, + 0x95220080, + 0x8ed008fe, + 0x408ed000, + 0xb6808acf, + 0xa0b606a5, + 0x00eabb01, + 0xb60480b6, + 0x1bf40192, + 0x08e4b6e8, + 0xf1f2efbc, + 0xb6085c87, + 0x94bd0684, + 0xd00399f0, + 0x00f80089, + 0xe7f1e0f9, + 0xe3f09814, + 0x8d21f440, + 0x041ce0b7, + 0xf401f7f0, + 0xe0fc8d21, + 0x04bd00f8, + 0xf10004fe, + 0xf0120017, + 0x12d00227, + 0x3e17f100, + 0x0010fe04, + 0x040017f1, + 0xf0c010d0, + 0x12d00427, + 0x1031f400, + 0x060817f1, + 0xcf0614b6, + 0x37f00012, + 0x1f24f001, + 0xb60432bb, + 0x02800132, + 0x04038003, + 0x040010b7, + 0x800012cf, + 0x27f10002, + 0x24b60800, + 0x0022cf06, + 0xb65817f0, + 0x13980c10, + 0x0432b800, + 0xb00b0bf4, + 0x1bf40034, + 0xf100f8f1, + 0xb6080027, + 0x22cf0624, + 0xf134bd40, + 0xb6070047, + 0x25950644, + 0x0045d008, + 0xbd4045d0, + 0x58f4bde4, + 0x1f58021e, + 0x020e4003, + 0xf5040f40, + 0xbb013d21, + 0x3fbb002f, + 0x041e5800, + 0x40051f58, + 0x0f400a0e, + 0x3d21f50c, + 0x030e9801, + 0xbb00effd, + 0x3ebb002e, + 0x0040b700, + 0x0235b613, + 0xb60043d0, + 0x35b60825, + 0x0120b606, + 0xb60130b6, + 0x34b60824, + 0x022fb908, + 0x026321f5, + 0xf1003fbb, + 0xb6080017, + 0x13d00614, + 0x0010b740, + 0xf024bd08, + 0x12d01f29, + 0x0031f400, + 0xf00028f4, + 0x21f41cd7, + 0xf401f439, + 0xf404e4b0, + 0x81fe1e18, + 0x0627f001, + 0x12fd20bd, + 0x01e4b604, + 0xfe051efd, + 0x21f50018, + 0x0ef404c3, + 0x10ef94d3, + 0xf501f5f0, + 0xf402ec21, + 0x80f9c60e, + 0xf90188fe, + 0xf990f980, + 0xf9b0f9a0, + 0xf9e0f9d0, + 0x800acff0, + 0xf404abc4, + 0xb7f11d0b, + 0xd7f01900, + 0x40becf1c, + 0xf400bfcf, + 0xb0b70421, + 0xe7f00400, + 0x00bed001, + 0xfc400ad0, + 0xfce0fcf0, + 0xfcb0fcd0, + 0xfc90fca0, + 0x0088fe80, + 0x32f480fc, + 0xf001f800, + 0x0e9801f7, + 0x04febb00, + 0x9418e7f1, + 0xf440e3f0, + 0x00f88d21, + 0x0614e7f1, + 0xf006e4b6, + 0xefd020f7, + 0x08f7f000, + 0xf401f2b6, + 0xf7f1fd1b, + 0xefd00a20, + 0xf100f800, + 0xb60a0417, + 0x1fd00614, + 0x0711f400, + 0x04a421f5, + 0x4afc17f1, + 0xf00213f0, + 0x12d00c27, + 0x0721f500, + 0xfc27f102, + 0x0223f047, + 0xf00020d0, + 0x20b6012c, + 0x0012d003, + 0xf001acf0, + 0xb7f002a5, + 0x50b3f000, + 0xb6000c98, + 0xbcbb0fc4, + 0x010c9800, + 0xf0020d98, + 0x21f500e7, + 0xacf0015c, + 0x04a5f001, + 0x4000b7f1, + 0x9850b3f0, + 0xc4b6000c, + 0x00bcbb0f, + 0x98050c98, + 0x0f98060d, + 0x00e7f104, + 0x5c21f508, + 0x0721f501, + 0x0601f402, + 0xf11412f4, + 0xf04afc17, + 0x27f00213, + 0x0012d00d, + 0x020721f5, + 0x048f21f5, + 0x000000f8, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, +}; diff --git a/drivers/gpu/drm/nouveau/nvc0_grhub.fuc b/drivers/gpu/drm/nouveau/nvc0_grhub.fuc new file mode 100644 index 00000000000..a1a599124cf --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_grhub.fuc @@ -0,0 +1,808 @@ +/* fuc microcode for nvc0 PGRAPH/HUB + * + * Copyright 2011 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Ben Skeggs + */ + +/* To build: + * m4 nvc0_grhub.fuc | envyas -a -w -m fuc -V nva3 -o nvc0_grhub.fuc.h + */ + +.section nvc0_grhub_data +include(`nvc0_graph.fuc') +gpc_count: .b32 0 +rop_count: .b32 0 +cmd_queue: queue_init +hub_mmio_list_head: .b32 0 +hub_mmio_list_tail: .b32 0 + +ctx_current: .b32 0 + +chipsets: +.b8 0xc0 0 0 0 +.b16 nvc0_hub_mmio_head +.b16 nvc0_hub_mmio_tail +.b8 0xc1 0 0 0 +.b16 nvc0_hub_mmio_head +.b16 nvc1_hub_mmio_tail +.b8 0xc3 0 0 0 +.b16 nvc0_hub_mmio_head +.b16 nvc0_hub_mmio_tail +.b8 0xc4 0 0 0 +.b16 nvc0_hub_mmio_head +.b16 nvc0_hub_mmio_tail +.b8 0xc8 0 0 0 +.b16 nvc0_hub_mmio_head +.b16 nvc0_hub_mmio_tail +.b8 0xce 0 0 0 +.b16 nvc0_hub_mmio_head +.b16 nvc0_hub_mmio_tail +.b8 0 0 0 0 + +nvc0_hub_mmio_head: +mmctx_data(0x17e91c, 2) +mmctx_data(0x400204, 2) +mmctx_data(0x404004, 11) +mmctx_data(0x404044, 1) +mmctx_data(0x404094, 14) +mmctx_data(0x4040d0, 7) +mmctx_data(0x4040f8, 1) +mmctx_data(0x404130, 3) +mmctx_data(0x404150, 3) +mmctx_data(0x404164, 2) +mmctx_data(0x404174, 3) +mmctx_data(0x404200, 8) +mmctx_data(0x404404, 14) +mmctx_data(0x404460, 4) +mmctx_data(0x404480, 1) +mmctx_data(0x404498, 1) +mmctx_data(0x404604, 4) +mmctx_data(0x404618, 32) +mmctx_data(0x404698, 21) +mmctx_data(0x4046f0, 2) +mmctx_data(0x404700, 22) +mmctx_data(0x405800, 1) +mmctx_data(0x405830, 3) +mmctx_data(0x405854, 1) +mmctx_data(0x405870, 4) +mmctx_data(0x405a00, 2) +mmctx_data(0x405a18, 1) +mmctx_data(0x406020, 1) +mmctx_data(0x406028, 4) +mmctx_data(0x4064a8, 2) +mmctx_data(0x4064b4, 2) +mmctx_data(0x407804, 1) +mmctx_data(0x40780c, 6) +mmctx_data(0x4078bc, 1) +mmctx_data(0x408000, 7) +mmctx_data(0x408064, 1) +mmctx_data(0x408800, 3) +mmctx_data(0x408900, 4) +mmctx_data(0x408980, 1) +nvc0_hub_mmio_tail: +mmctx_data(0x4064c0, 2) +nvc1_hub_mmio_tail: + +.align 256 +chan_data: +chan_mmio_count: .b32 0 +chan_mmio_address: .b32 0 + +.align 256 +xfer_data: .b32 0 + +.section nvc0_grhub_code +bra init +define(`include_code') +include(`nvc0_graph.fuc') + +// reports an exception to the host +// +// In: $r15 error code (see nvc0_graph.fuc) +// +error: + push $r14 + mov $r14 0x814 + shl b32 $r14 6 + iowr I[$r14 + 0x000] $r15 // CC_SCRATCH[5] = error code + mov $r14 0xc1c + shl b32 $r14 6 + mov $r15 1 + iowr I[$r14 + 0x000] $r15 // INTR_UP_SET + pop $r14 + ret + +// HUB fuc initialisation, executed by triggering ucode start, will +// fall through to main loop after completion. +// +// Input: +// CC_SCRATCH[0]: chipset (PMC_BOOT_0 read returns 0x0bad0bad... sigh) +// +// Output: +// CC_SCRATCH[0]: +// 31:31: set to signal completion +// CC_SCRATCH[1]: +// 31:0: total PGRAPH context size +// +init: + clear b32 $r0 + mov $sp $r0 + mov $xdbase $r0 + + // enable fifo access + mov $r1 0x1200 + mov $r2 2 + iowr I[$r1 + 0x000] $r2 // FIFO_ENABLE + + // setup i0 handler, and route all interrupts to it + mov $r1 ih + mov $iv0 $r1 + mov $r1 0x400 + iowr I[$r1 + 0x300] $r0 // INTR_DISPATCH + + // route HUB_CHANNEL_SWITCH to fuc interrupt 8 + mov $r3 0x404 + shl b32 $r3 6 + mov $r2 0x2003 // { HUB_CHANNEL_SWITCH, ZERO } -> intr 8 + iowr I[$r3 + 0x000] $r2 + + // not sure what these are, route them because NVIDIA does, and + // the IRQ handler will signal the host if we ever get one.. we + // may find out if/why we need to handle these if so.. + // + mov $r2 0x2004 + iowr I[$r3 + 0x004] $r2 // { 0x04, ZERO } -> intr 9 + mov $r2 0x200b + iowr I[$r3 + 0x008] $r2 // { 0x0b, ZERO } -> intr 10 + mov $r2 0x200c + iowr I[$r3 + 0x01c] $r2 // { 0x0c, ZERO } -> intr 15 + + // enable all INTR_UP interrupts + mov $r2 0xc24 + shl b32 $r2 6 + not b32 $r3 $r0 + iowr I[$r2] $r3 + + // enable fifo, ctxsw, 9, 10, 15 interrupts + mov $r2 -0x78fc // 0x8704 + sethi $r2 0 + iowr I[$r1 + 0x000] $r2 // INTR_EN_SET + + // fifo level triggered, rest edge + sub b32 $r1 0x100 + mov $r2 4 + iowr I[$r1] $r2 + + // enable interrupts + bset $flags ie0 + + // fetch enabled GPC/ROP counts + mov $r14 -0x69fc // 0x409604 + sethi $r14 0x400000 + call nv_rd32 + extr $r1 $r15 16:20 + st b32 D[$r0 + rop_count] $r1 + and $r15 0x1f + st b32 D[$r0 + gpc_count] $r15 + + // set BAR_REQMASK to GPC mask + mov $r1 1 + shl b32 $r1 $r15 + sub b32 $r1 1 + mov $r2 0x40c + shl b32 $r2 6 + iowr I[$r2 + 0x000] $r1 + iowr I[$r2 + 0x100] $r1 + + // find context data for this chipset + mov $r2 0x800 + shl b32 $r2 6 + iord $r2 I[$r2 + 0x000] // CC_SCRATCH[0] + mov $r15 chipsets - 8 + init_find_chipset: + add b32 $r15 8 + ld b32 $r3 D[$r15 + 0x00] + cmpu b32 $r3 $r2 + bra e init_context + cmpu b32 $r3 0 + bra ne init_find_chipset + // unknown chipset + ret + + // context size calculation, reserve first 256 bytes for use by fuc + init_context: + mov $r1 256 + + // calculate size of mmio context data + ld b16 $r14 D[$r15 + 4] + ld b16 $r15 D[$r15 + 6] + sethi $r14 0 + st b32 D[$r0 + hub_mmio_list_head] $r14 + st b32 D[$r0 + hub_mmio_list_tail] $r15 + call mmctx_size + + // set mmctx base addresses now so we don't have to do it later, + // they don't (currently) ever change + mov $r3 0x700 + shl b32 $r3 6 + shr b32 $r4 $r1 8 + iowr I[$r3 + 0x000] $r4 // MMCTX_SAVE_SWBASE + iowr I[$r3 + 0x100] $r4 // MMCTX_LOAD_SWBASE + add b32 $r3 0x1300 + add b32 $r1 $r15 + shr b32 $r15 2 + iowr I[$r3 + 0x000] $r15 // MMCTX_LOAD_COUNT, wtf for?!? + + // strands, base offset needs to be aligned to 256 bytes + shr b32 $r1 8 + add b32 $r1 1 + shl b32 $r1 8 + mov b32 $r15 $r1 + call strand_ctx_init + add b32 $r1 $r15 + + // initialise each GPC in sequence by passing in the offset of its + // context data in GPCn_CC_SCRATCH[1], and starting its FUC (which + // has previously been uploaded by the host) running. + // + // the GPC fuc init sequence will set GPCn_CC_SCRATCH[0] bit 31 + // when it has completed, and return the size of its context data + // in GPCn_CC_SCRATCH[1] + // + ld b32 $r3 D[$r0 + gpc_count] + mov $r4 0x2000 + sethi $r4 0x500000 + init_gpc: + // setup, and start GPC ucode running + add b32 $r14 $r4 0x804 + mov b32 $r15 $r1 + call nv_wr32 // CC_SCRATCH[1] = ctx offset + add b32 $r14 $r4 0x800 + mov b32 $r15 $r2 + call nv_wr32 // CC_SCRATCH[0] = chipset + add b32 $r14 $r4 0x10c + clear b32 $r15 + call nv_wr32 + add b32 $r14 $r4 0x104 + call nv_wr32 // ENTRY + add b32 $r14 $r4 0x100 + mov $r15 2 // CTRL_START_TRIGGER + call nv_wr32 // CTRL + + // wait for it to complete, and adjust context size + add b32 $r14 $r4 0x800 + init_gpc_wait: + call nv_rd32 + xbit $r15 $r15 31 + bra e init_gpc_wait + add b32 $r14 $r4 0x804 + call nv_rd32 + add b32 $r1 $r15 + + // next! + add b32 $r4 0x8000 + sub b32 $r3 1 + bra ne init_gpc + + // save context size, and tell host we're ready + mov $r2 0x800 + shl b32 $r2 6 + iowr I[$r2 + 0x100] $r1 // CC_SCRATCH[1] = context size + add b32 $r2 0x800 + clear b32 $r1 + bset $r1 31 + iowr I[$r2 + 0x000] $r1 // CC_SCRATCH[0] |= 0x80000000 + +// Main program loop, very simple, sleeps until woken up by the interrupt +// handler, pulls a command from the queue and executes its handler +// +main: + // sleep until we have something to do + bset $flags $p0 + sleep $p0 + mov $r13 cmd_queue + call queue_get + bra $p1 main + + // context switch, requested by GPU? + cmpu b32 $r14 0x4001 + bra ne main_not_ctx_switch + trace_set(T_AUTO) + mov $r1 0xb00 + shl b32 $r1 6 + iord $r2 I[$r1 + 0x100] // CHAN_NEXT + iord $r1 I[$r1 + 0x000] // CHAN_CUR + + xbit $r3 $r1 31 + bra e chsw_no_prev + xbit $r3 $r2 31 + bra e chsw_prev_no_next + push $r2 + mov b32 $r2 $r1 + trace_set(T_SAVE) + bclr $flags $p1 + bset $flags $p2 + call ctx_xfer + trace_clr(T_SAVE); + pop $r2 + trace_set(T_LOAD); + bset $flags $p1 + call ctx_xfer + trace_clr(T_LOAD); + bra chsw_done + chsw_prev_no_next: + push $r2 + mov b32 $r2 $r1 + bclr $flags $p1 + bclr $flags $p2 + call ctx_xfer + pop $r2 + mov $r1 0xb00 + shl b32 $r1 6 + iowr I[$r1] $r2 + bra chsw_done + chsw_no_prev: + xbit $r3 $r2 31 + bra e chsw_done + bset $flags $p1 + bclr $flags $p2 + call ctx_xfer + + // ack the context switch request + chsw_done: + mov $r1 0xb0c + shl b32 $r1 6 + mov $r2 1 + iowr I[$r1 + 0x000] $r2 // 0x409b0c + trace_clr(T_AUTO) + bra main + + // request to set current channel? (*not* a context switch) + main_not_ctx_switch: + cmpu b32 $r14 0x0001 + bra ne main_not_ctx_chan + mov b32 $r2 $r15 + call ctx_chan + bra main_done + + // request to store current channel context? + main_not_ctx_chan: + cmpu b32 $r14 0x0002 + bra ne main_not_ctx_save + trace_set(T_SAVE) + bclr $flags $p1 + bclr $flags $p2 + call ctx_xfer + trace_clr(T_SAVE) + bra main_done + + main_not_ctx_save: + shl b32 $r15 $r14 16 + or $r15 E_BAD_COMMAND + call error + bra main + + main_done: + mov $r1 0x820 + shl b32 $r1 6 + clear b32 $r2 + bset $r2 31 + iowr I[$r1 + 0x000] $r2 // CC_SCRATCH[0] |= 0x80000000 + bra main + +// interrupt handler +ih: + push $r8 + mov $r8 $flags + push $r8 + push $r9 + push $r10 + push $r11 + push $r13 + push $r14 + push $r15 + + // incoming fifo command? + iord $r10 I[$r0 + 0x200] // INTR + and $r11 $r10 0x00000004 + bra e ih_no_fifo + // queue incoming fifo command for later processing + mov $r11 0x1900 + mov $r13 cmd_queue + iord $r14 I[$r11 + 0x100] // FIFO_CMD + iord $r15 I[$r11 + 0x000] // FIFO_DATA + call queue_put + add b32 $r11 0x400 + mov $r14 1 + iowr I[$r11 + 0x000] $r14 // FIFO_ACK + + // context switch request? + ih_no_fifo: + and $r11 $r10 0x00000100 + bra e ih_no_ctxsw + // enqueue a context switch for later processing + mov $r13 cmd_queue + mov $r14 0x4001 + call queue_put + + // anything we didn't handle, bring it to the host's attention + ih_no_ctxsw: + mov $r11 0x104 + not b32 $r11 + and $r11 $r10 $r11 + bra e ih_no_other + mov $r10 0xc1c + shl b32 $r10 6 + iowr I[$r10] $r11 // INTR_UP_SET + + // ack, and wake up main() + ih_no_other: + iowr I[$r0 + 0x100] $r10 // INTR_ACK + + pop $r15 + pop $r14 + pop $r13 + pop $r11 + pop $r10 + pop $r9 + pop $r8 + mov $flags $r8 + pop $r8 + bclr $flags $p0 + iret + +// Not real sure, but, MEM_CMD 7 will hang forever if this isn't done +ctx_4160s: + mov $r14 0x4160 + sethi $r14 0x400000 + mov $r15 1 + call nv_wr32 + ctx_4160s_wait: + call nv_rd32 + xbit $r15 $r15 4 + bra e ctx_4160s_wait + ret + +// Without clearing again at end of xfer, some things cause PGRAPH +// to hang with STATUS=0x00000007 until it's cleared.. fbcon can +// still function with it set however... +ctx_4160c: + mov $r14 0x4160 + sethi $r14 0x400000 + clear b32 $r15 + call nv_wr32 + ret + +// Again, not real sure +// +// In: $r15 value to set 0x404170 to +// +ctx_4170s: + mov $r14 0x4170 + sethi $r14 0x400000 + or $r15 0x10 + call nv_wr32 + ret + +// Waits for a ctx_4170s() call to complete +// +ctx_4170w: + mov $r14 0x4170 + sethi $r14 0x400000 + call nv_rd32 + and $r15 0x10 + bra ne ctx_4170w + ret + +// Disables various things, waits a bit, and re-enables them.. +// +// Not sure how exactly this helps, perhaps "ENABLE" is not such a +// good description for the bits we turn off? Anyways, without this, +// funny things happen. +// +ctx_redswitch: + mov $r14 0x614 + shl b32 $r14 6 + mov $r15 0x270 + iowr I[$r14] $r15 // HUB_RED_SWITCH = ENABLE_GPC, POWER_ALL + mov $r15 8 + ctx_redswitch_delay: + sub b32 $r15 1 + bra ne ctx_redswitch_delay + mov $r15 0x770 + iowr I[$r14] $r15 // HUB_RED_SWITCH = ENABLE_ALL, POWER_ALL + ret + +// Not a clue what this is for, except that unless the value is 0x10, the +// strand context is saved (and presumably restored) incorrectly.. +// +// In: $r15 value to set to (0x00/0x10 are used) +// +ctx_86c: + mov $r14 0x86c + shl b32 $r14 6 + iowr I[$r14] $r15 // HUB(0x86c) = val + mov $r14 -0x75ec + sethi $r14 0x400000 + call nv_wr32 // ROP(0xa14) = val + mov $r14 -0x5794 + sethi $r14 0x410000 + call nv_wr32 // GPC(0x86c) = val + ret + +// ctx_load - load's a channel's ctxctl data, and selects its vm +// +// In: $r2 channel address +// +ctx_load: + trace_set(T_CHAN) + + // switch to channel, somewhat magic in parts.. + mov $r10 12 // DONE_UNK12 + call wait_donez + mov $r1 0xa24 + shl b32 $r1 6 + iowr I[$r1 + 0x000] $r0 // 0x409a24 + mov $r3 0xb00 + shl b32 $r3 6 + iowr I[$r3 + 0x100] $r2 // CHAN_NEXT + mov $r1 0xa0c + shl b32 $r1 6 + mov $r4 7 + iowr I[$r1 + 0x000] $r2 // MEM_CHAN + iowr I[$r1 + 0x100] $r4 // MEM_CMD + ctx_chan_wait_0: + iord $r4 I[$r1 + 0x100] + and $r4 0x1f + bra ne ctx_chan_wait_0 + iowr I[$r3 + 0x000] $r2 // CHAN_CUR + + // load channel header, fetch PGRAPH context pointer + mov $xtargets $r0 + bclr $r2 31 + shl b32 $r2 4 + add b32 $r2 2 + + trace_set(T_LCHAN) + mov $r1 0xa04 + shl b32 $r1 6 + iowr I[$r1 + 0x000] $r2 // MEM_BASE + mov $r1 0xa20 + shl b32 $r1 6 + mov $r2 0x0002 + sethi $r2 0x80000000 + iowr I[$r1 + 0x000] $r2 // MEM_TARGET = vram + mov $r1 0x10 // chan + 0x0210 + mov $r2 xfer_data + sethi $r2 0x00020000 // 16 bytes + xdld $r1 $r2 + xdwait + trace_clr(T_LCHAN) + + // update current context + ld b32 $r1 D[$r0 + xfer_data + 4] + shl b32 $r1 24 + ld b32 $r2 D[$r0 + xfer_data + 0] + shr b32 $r2 8 + or $r1 $r2 + st b32 D[$r0 + ctx_current] $r1 + + // set transfer base to start of context, and fetch context header + trace_set(T_LCTXH) + mov $r2 0xa04 + shl b32 $r2 6 + iowr I[$r2 + 0x000] $r1 // MEM_BASE + mov $r2 1 + mov $r1 0xa20 + shl b32 $r1 6 + iowr I[$r1 + 0x000] $r2 // MEM_TARGET = vm + mov $r1 chan_data + sethi $r1 0x00060000 // 256 bytes + xdld $r0 $r1 + xdwait + trace_clr(T_LCTXH) + + trace_clr(T_CHAN) + ret + +// ctx_chan - handler for HUB_SET_CHAN command, will set a channel as +// the active channel for ctxctl, but not actually transfer +// any context data. intended for use only during initial +// context construction. +// +// In: $r2 channel address +// +ctx_chan: + call ctx_4160s + call ctx_load + mov $r10 12 // DONE_UNK12 + call wait_donez + mov $r1 0xa10 + shl b32 $r1 6 + mov $r2 5 + iowr I[$r1 + 0x000] $r2 // MEM_CMD = 5 (???) + ctx_chan_wait: + iord $r2 I[$r1 + 0x000] + or $r2 $r2 + bra ne ctx_chan_wait + call ctx_4160c + ret + +// Execute per-context state overrides list +// +// Only executed on the first load of a channel. Might want to look into +// removing this and having the host directly modify the channel's context +// to change this state... The nouveau DRM already builds this list as +// it's definitely needed for NVIDIA's, so we may as well use it for now +// +// Input: $r1 mmio list length +// +ctx_mmio_exec: + // set transfer base to be the mmio list + ld b32 $r3 D[$r0 + chan_mmio_address] + mov $r2 0xa04 + shl b32 $r2 6 + iowr I[$r2 + 0x000] $r3 // MEM_BASE + + clear b32 $r3 + ctx_mmio_loop: + // fetch next 256 bytes of mmio list if necessary + and $r4 $r3 0xff + bra ne ctx_mmio_pull + mov $r5 xfer_data + sethi $r5 0x00060000 // 256 bytes + xdld $r3 $r5 + xdwait + + // execute a single list entry + ctx_mmio_pull: + ld b32 $r14 D[$r4 + xfer_data + 0x00] + ld b32 $r15 D[$r4 + xfer_data + 0x04] + call nv_wr32 + + // next! + add b32 $r3 8 + sub b32 $r1 1 + bra ne ctx_mmio_loop + + // set transfer base back to the current context + ctx_mmio_done: + ld b32 $r3 D[$r0 + ctx_current] + iowr I[$r2 + 0x000] $r3 // MEM_BASE + + // disable the mmio list now, we don't need/want to execute it again + st b32 D[$r0 + chan_mmio_count] $r0 + mov $r1 chan_data + sethi $r1 0x00060000 // 256 bytes + xdst $r0 $r1 + xdwait + ret + +// Transfer HUB context data between GPU and storage area +// +// In: $r2 channel address +// $p1 clear on save, set on load +// $p2 set if opposite direction done/will be done, so: +// on save it means: "a load will follow this save" +// on load it means: "a save preceeded this load" +// +ctx_xfer: + bra not $p1 ctx_xfer_pre + bra $p2 ctx_xfer_pre_load + ctx_xfer_pre: + mov $r15 0x10 + call ctx_86c + call ctx_4160s + bra not $p1 ctx_xfer_exec + + ctx_xfer_pre_load: + mov $r15 2 + call ctx_4170s + call ctx_4170w + call ctx_redswitch + clear b32 $r15 + call ctx_4170s + call ctx_load + + // fetch context pointer, and initiate xfer on all GPCs + ctx_xfer_exec: + ld b32 $r1 D[$r0 + ctx_current] + mov $r2 0x414 + shl b32 $r2 6 + iowr I[$r2 + 0x000] $r0 // BAR_STATUS = reset + mov $r14 -0x5b00 + sethi $r14 0x410000 + mov b32 $r15 $r1 + call nv_wr32 // GPC_BCAST_WRCMD_DATA = ctx pointer + add b32 $r14 4 + xbit $r15 $flags $p1 + xbit $r2 $flags $p2 + shl b32 $r2 1 + or $r15 $r2 + call nv_wr32 // GPC_BCAST_WRCMD_CMD = GPC_XFER(type) + + // strands + mov $r1 0x4afc + sethi $r1 0x20000 + mov $r2 0xc + iowr I[$r1] $r2 // STRAND_CMD(0x3f) = 0x0c + call strand_wait + mov $r2 0x47fc + sethi $r2 0x20000 + iowr I[$r2] $r0 // STRAND_FIRST_GENE(0x3f) = 0x00 + xbit $r2 $flags $p1 + add b32 $r2 3 + iowr I[$r1] $r2 // STRAND_CMD(0x3f) = 0x03/0x04 (SAVE/LOAD) + + // mmio context + xbit $r10 $flags $p1 // direction + or $r10 6 // first, last + mov $r11 0 // base = 0 + ld b32 $r12 D[$r0 + hub_mmio_list_head] + ld b32 $r13 D[$r0 + hub_mmio_list_tail] + mov $r14 0 // not multi + call mmctx_xfer + + // wait for GPCs to all complete + mov $r10 8 // DONE_BAR + call wait_doneo + + // wait for strand xfer to complete + call strand_wait + + // post-op + bra $p1 ctx_xfer_post + mov $r10 12 // DONE_UNK12 + call wait_donez + mov $r1 0xa10 + shl b32 $r1 6 + mov $r2 5 + iowr I[$r1] $r2 // MEM_CMD + ctx_xfer_post_save_wait: + iord $r2 I[$r1] + or $r2 $r2 + bra ne ctx_xfer_post_save_wait + + bra $p2 ctx_xfer_done + ctx_xfer_post: + mov $r15 2 + call ctx_4170s + clear b32 $r15 + call ctx_86c + call strand_post + call ctx_4170w + clear b32 $r15 + call ctx_4170s + + bra not $p1 ctx_xfer_no_post_mmio + ld b32 $r1 D[$r0 + chan_mmio_count] + or $r1 $r1 + bra e ctx_xfer_no_post_mmio + call ctx_mmio_exec + + ctx_xfer_no_post_mmio: + call ctx_4160c + + ctx_xfer_done: + ret + +.align 256 diff --git a/drivers/gpu/drm/nouveau/nvc0_grhub.fuc.h b/drivers/gpu/drm/nouveau/nvc0_grhub.fuc.h new file mode 100644 index 00000000000..b3b541b6d04 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvc0_grhub.fuc.h @@ -0,0 +1,838 @@ +uint32_t nvc0_grhub_data[] = { + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x000000c0, + 0x012c0090, + 0x000000c1, + 0x01300090, + 0x000000c3, + 0x012c0090, + 0x000000c4, + 0x012c0090, + 0x000000c8, + 0x012c0090, + 0x000000ce, + 0x012c0090, + 0x00000000, + 0x0417e91c, + 0x04400204, + 0x28404004, + 0x00404044, + 0x34404094, + 0x184040d0, + 0x004040f8, + 0x08404130, + 0x08404150, + 0x04404164, + 0x08404174, + 0x1c404200, + 0x34404404, + 0x0c404460, + 0x00404480, + 0x00404498, + 0x0c404604, + 0x7c404618, + 0x50404698, + 0x044046f0, + 0x54404700, + 0x00405800, + 0x08405830, + 0x00405854, + 0x0c405870, + 0x04405a00, + 0x00405a18, + 0x00406020, + 0x0c406028, + 0x044064a8, + 0x044064b4, + 0x00407804, + 0x1440780c, + 0x004078bc, + 0x18408000, + 0x00408064, + 0x08408800, + 0x0c408900, + 0x00408980, + 0x044064c0, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, +}; + +uint32_t nvc0_grhub_code[] = { + 0x03090ef5, + 0x9800d898, + 0x86f001d9, + 0x0489b808, + 0xf00c1bf4, + 0x21f502f7, + 0x00f802ec, + 0xb60798c4, + 0x8dbb0384, + 0x0880b600, + 0x80008e80, + 0x90b6018f, + 0x0f94f001, + 0xf801d980, + 0x0131f400, + 0x9800d898, + 0x89b801d9, + 0x210bf404, + 0xb60789c4, + 0x9dbb0394, + 0x0890b600, + 0x98009e98, + 0x80b6019f, + 0x0f84f001, + 0xf400d880, + 0x00f80132, + 0x0728b7f1, + 0xb906b4b6, + 0xc9f002ec, + 0x00bcd01f, + 0xc800bccf, + 0x1bf41fcc, + 0x06a7f0fa, + 0x010321f5, + 0xf840bfcf, + 0x28b7f100, + 0x06b4b607, + 0xb980bfd0, + 0xc9f002ec, + 0x1ec9f01f, + 0xcf00bcd0, + 0xccc800bc, + 0xfa1bf41f, + 0x87f100f8, + 0x84b60430, + 0x1ff9f006, + 0xf8008fd0, + 0x3087f100, + 0x0684b604, + 0xf80080d0, + 0x3c87f100, + 0x0684b608, + 0x99f094bd, + 0x0089d000, + 0x081887f1, + 0xd00684b6, + 0x87f1008a, + 0x84b60400, + 0x0088cf06, + 0xf4888aff, + 0x87f1f31b, + 0x84b6085c, + 0xf094bd06, + 0x89d00099, + 0xf100f800, + 0xb6083c87, + 0x94bd0684, + 0xd00099f0, + 0x87f10089, + 0x84b60818, + 0x008ad006, + 0x040087f1, + 0xcf0684b6, + 0x8aff0088, + 0xf30bf488, + 0x085c87f1, + 0xbd0684b6, + 0x0099f094, + 0xf80089d0, + 0x9894bd00, + 0x85b600e8, + 0x0180b61a, + 0xbb0284b6, + 0xe0b60098, + 0x04efb804, + 0xb9eb1bf4, + 0x00f8029f, + 0x083c87f1, + 0xbd0684b6, + 0x0199f094, + 0xf10089d0, + 0xb6071087, + 0x94bd0684, + 0xf405bbfd, + 0x8bd0090b, + 0x0099f000, + 0xf405eefd, + 0x8ed00c0b, + 0xc08fd080, + 0xb70199f0, + 0xc8010080, + 0xb4b600ab, + 0x0cb9f010, + 0xb601aec8, + 0xbefd11e4, + 0x008bd005, + 0xf0008ecf, + 0x0bf41fe4, + 0x00ce98fa, + 0xd005e9fd, + 0xc0b6c08e, + 0x04cdb804, + 0xc8e81bf4, + 0x1bf402ab, + 0x008bcf18, + 0xb01fb4f0, + 0x1bf410b4, + 0x02a7f0f7, + 0xf4c921f4, + 0xabc81b0e, + 0x10b4b600, + 0xf00cb9f0, + 0x8bd012b9, + 0x008bcf00, + 0xf412bbc8, + 0x87f1fa1b, + 0x84b6085c, + 0xf094bd06, + 0x89d00199, + 0xf900f800, + 0x02a7f0a0, + 0xfcc921f4, + 0xf100f8a0, + 0xf04afc87, + 0x97f00283, + 0x0089d00c, + 0x020721f5, + 0x87f100f8, + 0x83f04afc, + 0x0d97f002, + 0xf50089d0, + 0xf8020721, + 0xfca7f100, + 0x02a3f04f, + 0x0500aba2, + 0xd00fc7f0, + 0xc7f000ac, + 0x00bcd00b, + 0x020721f5, + 0xf000aed0, + 0xbcd00ac7, + 0x0721f500, + 0xf100f802, + 0xb6083c87, + 0x94bd0684, + 0xd00399f0, + 0x21f50089, + 0xe7f00213, + 0x3921f503, + 0xfca7f102, + 0x02a3f046, + 0x0400aba0, + 0xf040a0d0, + 0xbcd001c7, + 0x0721f500, + 0x010c9202, + 0xf000acd0, + 0xbcd002c7, + 0x0721f500, + 0x2621f502, + 0x8087f102, + 0x0684b608, + 0xb70089cf, + 0x95220080, + 0x8ed008fe, + 0x408ed000, + 0xb6808acf, + 0xa0b606a5, + 0x00eabb01, + 0xb60480b6, + 0x1bf40192, + 0x08e4b6e8, + 0xf1f2efbc, + 0xb6085c87, + 0x94bd0684, + 0xd00399f0, + 0x00f80089, + 0xe7f1e0f9, + 0xe4b60814, + 0x00efd006, + 0x0c1ce7f1, + 0xf006e4b6, + 0xefd001f7, + 0xf8e0fc00, + 0xfe04bd00, + 0x07fe0004, + 0x0017f100, + 0x0227f012, + 0xf10012d0, + 0xfe05b917, + 0x17f10010, + 0x10d00400, + 0x0437f1c0, + 0x0634b604, + 0x200327f1, + 0xf10032d0, + 0xd0200427, + 0x27f10132, + 0x32d0200b, + 0x0c27f102, + 0x0732d020, + 0x0c2427f1, + 0xb90624b6, + 0x23d00003, + 0x0427f100, + 0x0023f087, + 0xb70012d0, + 0xf0010012, + 0x12d00427, + 0x1031f400, + 0x9604e7f1, + 0xf440e3f0, + 0xf1c76821, + 0x01018090, + 0x801ff4f0, + 0x17f0000f, + 0x041fbb01, + 0xf10112b6, + 0xb6040c27, + 0x21d00624, + 0x4021d000, + 0x080027f1, + 0xcf0624b6, + 0xf7f00022, + 0x08f0b654, + 0xb800f398, + 0x0bf40432, + 0x0034b00b, + 0xf8f11bf4, + 0x0017f100, + 0x02fe5801, + 0xf003ff58, + 0x0e8000e3, + 0x150f8014, + 0x013d21f5, + 0x070037f1, + 0x950634b6, + 0x34d00814, + 0x4034d000, + 0x130030b7, + 0xb6001fbb, + 0x3fd002f5, + 0x0815b600, + 0xb60110b6, + 0x1fb90814, + 0x6321f502, + 0x001fbb02, + 0xf1000398, + 0xf0200047, + 0x4ea05043, + 0x1fb90804, + 0x8d21f402, + 0x08004ea0, + 0xf4022fb9, + 0x4ea08d21, + 0xf4bd010c, + 0xa08d21f4, + 0xf401044e, + 0x4ea08d21, + 0xf7f00100, + 0x8d21f402, + 0x08004ea0, + 0xc86821f4, + 0x0bf41fff, + 0x044ea0fa, + 0x6821f408, + 0xb7001fbb, + 0xb6800040, + 0x1bf40132, + 0x0027f1b4, + 0x0624b608, + 0xb74021d0, + 0xbd080020, + 0x1f19f014, + 0xf40021d0, + 0x28f40031, + 0x08d7f000, + 0xf43921f4, + 0xe4b1f401, + 0x1bf54001, + 0x87f100d1, + 0x84b6083c, + 0xf094bd06, + 0x89d00499, + 0x0017f100, + 0x0614b60b, + 0xcf4012cf, + 0x13c80011, + 0x7e0bf41f, + 0xf41f23c8, + 0x20f95a0b, + 0xf10212b9, + 0xb6083c87, + 0x94bd0684, + 0xd00799f0, + 0x32f40089, + 0x0231f401, + 0x082921f5, + 0x085c87f1, + 0xbd0684b6, + 0x0799f094, + 0xfc0089d0, + 0x3c87f120, + 0x0684b608, + 0x99f094bd, + 0x0089d006, + 0xf50131f4, + 0xf1082921, + 0xb6085c87, + 0x94bd0684, + 0xd00699f0, + 0x0ef40089, + 0xb920f931, + 0x32f40212, + 0x0232f401, + 0x082921f5, + 0x17f120fc, + 0x14b60b00, + 0x0012d006, + 0xc8130ef4, + 0x0bf41f23, + 0x0131f40d, + 0xf50232f4, + 0xf1082921, + 0xb60b0c17, + 0x27f00614, + 0x0012d001, + 0x085c87f1, + 0xbd0684b6, + 0x0499f094, + 0xf50089d0, + 0xb0ff200e, + 0x1bf401e4, + 0x02f2b90d, + 0x07b521f5, + 0xb0420ef4, + 0x1bf402e4, + 0x3c87f12e, + 0x0684b608, + 0x99f094bd, + 0x0089d007, + 0xf40132f4, + 0x21f50232, + 0x87f10829, + 0x84b6085c, + 0xf094bd06, + 0x89d00799, + 0x110ef400, + 0xf010ef94, + 0x21f501f5, + 0x0ef502ec, + 0x17f1fed1, + 0x14b60820, + 0xf024bd06, + 0x12d01f29, + 0xbe0ef500, + 0xfe80f9fe, + 0x80f90188, + 0xa0f990f9, + 0xd0f9b0f9, + 0xf0f9e0f9, + 0xc4800acf, + 0x0bf404ab, + 0x00b7f11d, + 0x08d7f019, + 0xcf40becf, + 0x21f400bf, + 0x00b0b704, + 0x01e7f004, + 0xe400bed0, + 0xf40100ab, + 0xd7f00d0b, + 0x01e7f108, + 0x0421f440, + 0x0104b7f1, + 0xabffb0bd, + 0x0d0bf4b4, + 0x0c1ca7f1, + 0xd006a4b6, + 0x0ad000ab, + 0xfcf0fc40, + 0xfcd0fce0, + 0xfca0fcb0, + 0xfe80fc90, + 0x80fc0088, + 0xf80032f4, + 0x60e7f101, + 0x40e3f041, + 0xf401f7f0, + 0x21f48d21, + 0x04ffc868, + 0xf8fa0bf4, + 0x60e7f100, + 0x40e3f041, + 0x21f4f4bd, + 0xf100f88d, + 0xf04170e7, + 0xf5f040e3, + 0x8d21f410, + 0xe7f100f8, + 0xe3f04170, + 0x6821f440, + 0xf410f4f0, + 0x00f8f31b, + 0x0614e7f1, + 0xf106e4b6, + 0xd00270f7, + 0xf7f000ef, + 0x01f2b608, + 0xf1fd1bf4, + 0xd00770f7, + 0x00f800ef, + 0x086ce7f1, + 0xd006e4b6, + 0xe7f100ef, + 0xe3f08a14, + 0x8d21f440, + 0xa86ce7f1, + 0xf441e3f0, + 0x00f88d21, + 0x083c87f1, + 0xbd0684b6, + 0x0599f094, + 0xf00089d0, + 0x21f40ca7, + 0x2417f1c9, + 0x0614b60a, + 0xf10010d0, + 0xb60b0037, + 0x32d00634, + 0x0c17f140, + 0x0614b60a, + 0xd00747f0, + 0x14d00012, + 0x4014cf40, + 0xf41f44f0, + 0x32d0fa1b, + 0x000bfe00, + 0xb61f2af0, + 0x20b60424, + 0x3c87f102, + 0x0684b608, + 0x99f094bd, + 0x0089d008, + 0x0a0417f1, + 0xd00614b6, + 0x17f10012, + 0x14b60a20, + 0x0227f006, + 0x800023f1, + 0xf00012d0, + 0x27f11017, + 0x23f00300, + 0x0512fa02, + 0x87f103f8, + 0x84b6085c, + 0xf094bd06, + 0x89d00899, + 0xc1019800, + 0x981814b6, + 0x25b6c002, + 0x0512fd08, + 0xf1160180, + 0xb6083c87, + 0x94bd0684, + 0xd00999f0, + 0x27f10089, + 0x24b60a04, + 0x0021d006, + 0xf10127f0, + 0xb60a2017, + 0x12d00614, + 0x0017f100, + 0x0613f002, + 0xf80501fa, + 0x5c87f103, + 0x0684b608, + 0x99f094bd, + 0x0089d009, + 0x085c87f1, + 0xbd0684b6, + 0x0599f094, + 0xf80089d0, + 0x3121f500, + 0xb821f506, + 0x0ca7f006, + 0xf1c921f4, + 0xb60a1017, + 0x27f00614, + 0x0012d005, + 0xfd0012cf, + 0x1bf40522, + 0x4921f5fa, + 0x9800f806, + 0x27f18103, + 0x24b60a04, + 0x0023d006, + 0x34c434bd, + 0x0f1bf4ff, + 0x030057f1, + 0xfa0653f0, + 0x03f80535, + 0x98c04e98, + 0x21f4c14f, + 0x0830b68d, + 0xf40112b6, + 0x0398df1b, + 0x0023d016, + 0xf1800080, + 0xf0020017, + 0x01fa0613, + 0xf803f806, + 0x0611f400, + 0xf01102f4, + 0x21f510f7, + 0x21f50698, + 0x11f40631, + 0x02f7f01c, + 0x065721f5, + 0x066621f5, + 0x067821f5, + 0x21f5f4bd, + 0x21f50657, + 0x019806b8, + 0x1427f116, + 0x0624b604, + 0xf10020d0, + 0xf0a500e7, + 0x1fb941e3, + 0x8d21f402, + 0xf004e0b6, + 0x2cf001fc, + 0x0124b602, + 0xf405f2fd, + 0x17f18d21, + 0x13f04afc, + 0x0c27f002, + 0xf50012d0, + 0xf1020721, + 0xf047fc27, + 0x20d00223, + 0x012cf000, + 0xd00320b6, + 0xacf00012, + 0x06a5f001, + 0x9800b7f0, + 0x0d98140c, + 0x00e7f015, + 0x015c21f5, + 0xf508a7f0, + 0xf5010321, + 0xf4020721, + 0xa7f02201, + 0xc921f40c, + 0x0a1017f1, + 0xf00614b6, + 0x12d00527, + 0x0012cf00, + 0xf40522fd, + 0x02f4fa1b, + 0x02f7f032, + 0x065721f5, + 0x21f5f4bd, + 0x21f50698, + 0x21f50226, + 0xf4bd0666, + 0x065721f5, + 0x981011f4, + 0x11fd8001, + 0x070bf405, + 0x07df21f5, + 0x064921f5, + 0x000000f8, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, +}; diff --git a/drivers/gpu/drm/nouveau/nvc0_instmem.c b/drivers/gpu/drm/nouveau/nvc0_instmem.c index 82357d2df1f..b701c439c92 100644 --- a/drivers/gpu/drm/nouveau/nvc0_instmem.c +++ b/drivers/gpu/drm/nouveau/nvc0_instmem.c @@ -32,7 +32,6 @@ struct nvc0_instmem_priv { struct nouveau_channel *bar1; struct nouveau_gpuobj *bar3_pgd; struct nouveau_channel *bar3; - struct nouveau_gpuobj *chan_pgd; }; int @@ -181,17 +180,11 @@ nvc0_instmem_init(struct drm_device *dev) goto error; /* channel vm */ - ret = nouveau_vm_new(dev, 0, (1ULL << 40), 0x0008000000ULL, &vm); + ret = nouveau_vm_new(dev, 0, (1ULL << 40), 0x0008000000ULL, + &dev_priv->chan_vm); if (ret) goto error; - ret = nouveau_gpuobj_new(dev, NULL, 0x8000, 4096, 0, &priv->chan_pgd); - if (ret) - goto error; - - nouveau_vm_ref(vm, &dev_priv->chan_vm, priv->chan_pgd); - nouveau_vm_ref(NULL, &vm, NULL); - nvc0_instmem_resume(dev); return 0; error: @@ -211,8 +204,7 @@ nvc0_instmem_takedown(struct drm_device *dev) nv_wr32(dev, 0x1704, 0x00000000); nv_wr32(dev, 0x1714, 0x00000000); - nouveau_vm_ref(NULL, &dev_priv->chan_vm, priv->chan_pgd); - nouveau_gpuobj_ref(NULL, &priv->chan_pgd); + nouveau_vm_ref(NULL, &dev_priv->chan_vm, NULL); nvc0_channel_del(&priv->bar1); nouveau_vm_ref(NULL, &dev_priv->bar1_vm, priv->bar1_pgd); diff --git a/drivers/gpu/drm/nouveau/nvc0_vm.c b/drivers/gpu/drm/nouveau/nvc0_vm.c index a179e6c55af..9e352944a35 100644 --- a/drivers/gpu/drm/nouveau/nvc0_vm.c +++ b/drivers/gpu/drm/nouveau/nvc0_vm.c @@ -105,7 +105,11 @@ nvc0_vm_flush(struct nouveau_vm *vm) struct drm_device *dev = vm->dev; struct nouveau_vm_pgd *vpgd; unsigned long flags; - u32 engine = (dev_priv->chan_vm == vm) ? 1 : 5; + u32 engine; + + engine = 1; + if (vm == dev_priv->bar1_vm || vm == dev_priv->bar3_vm) + engine |= 4; pinstmem->flush(vm->dev); diff --git a/drivers/gpu/drm/nouveau/nvc0_vram.c b/drivers/gpu/drm/nouveau/nvc0_vram.c index 67c6ec6f34e..e45a24d84e9 100644 --- a/drivers/gpu/drm/nouveau/nvc0_vram.c +++ b/drivers/gpu/drm/nouveau/nvc0_vram.c @@ -61,9 +61,7 @@ nvc0_vram_new(struct drm_device *dev, u64 size, u32 align, u32 ncmin, u32 type, struct nouveau_mem **pmem) { struct drm_nouveau_private *dev_priv = dev->dev_private; - struct ttm_bo_device *bdev = &dev_priv->ttm.bdev; - struct ttm_mem_type_manager *man = &bdev->man[TTM_PL_VRAM]; - struct nouveau_mm *mm = man->priv; + struct nouveau_mm *mm = dev_priv->engine.vram.mm; struct nouveau_mm_node *r; struct nouveau_mem *mem; int ret; @@ -105,9 +103,15 @@ int nvc0_vram_init(struct drm_device *dev) { struct drm_nouveau_private *dev_priv = dev->dev_private; + struct nouveau_vram_engine *vram = &dev_priv->engine.vram; + const u32 rsvd_head = ( 256 * 1024) >> 12; /* vga memory */ + const u32 rsvd_tail = (1024 * 1024) >> 12; /* vbios etc */ + u32 length; dev_priv->vram_size = nv_rd32(dev, 0x10f20c) << 20; dev_priv->vram_size *= nv_rd32(dev, 0x121c74); - dev_priv->vram_rblock_size = 4096; - return 0; + + length = (dev_priv->vram_size >> 12) - rsvd_head - rsvd_tail; + + return nouveau_mm_init(&vram->mm, rsvd_head, length, 1); } diff --git a/drivers/gpu/drm/nouveau/nvreg.h b/drivers/gpu/drm/nouveau/nvreg.h index fe0f253089a..bbfb1a68fb1 100644 --- a/drivers/gpu/drm/nouveau/nvreg.h +++ b/drivers/gpu/drm/nouveau/nvreg.h @@ -277,6 +277,8 @@ # define NV_CIO_CRE_EBR_VDE_11 2:2 # define NV_CIO_CRE_EBR_VRS_11 4:4 # define NV_CIO_CRE_EBR_VBS_11 6:6 +# define NV_CIO_CRE_42 0x42 +# define NV_CIO_CRE_42_OFFSET_11 6:6 # define NV_CIO_CRE_43 0x43 # define NV_CIO_CRE_44 0x44 /* head control */ # define NV_CIO_CRE_CSB 0x45 /* colour saturation boost */ diff --git a/drivers/gpu/drm/radeon/Kconfig b/drivers/gpu/drm/radeon/Kconfig index 9746fee59f5..ea92bbe3ed3 100644 --- a/drivers/gpu/drm/radeon/Kconfig +++ b/drivers/gpu/drm/radeon/Kconfig @@ -28,11 +28,4 @@ config DRM_RADEON_KMS The kernel will also perform security check on command stream provided by the user, we want to catch and forbid any illegal use of the GPU such as DMA into random system memory or into memory - not owned by the process supplying the command stream. This part - of the code is still incomplete and this why we propose that patch - as a staging driver addition, future security might forbid current - experimental userspace to run. - - This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX - (radeon up to X1950). Works is underway to provide support for R6XX, - R7XX and newer hardware (radeon from HD2XXX to HD4XXX). + not owned by the process supplying the command stream. diff --git a/drivers/gpu/drm/radeon/Makefile b/drivers/gpu/drm/radeon/Makefile index 3896ef81110..9f363e0c4b6 100644 --- a/drivers/gpu/drm/radeon/Makefile +++ b/drivers/gpu/drm/radeon/Makefile @@ -5,6 +5,7 @@ ccflags-y := -Iinclude/drm hostprogs-y := mkregtable +clean-files := rn50_reg_safe.h r100_reg_safe.h r200_reg_safe.h rv515_reg_safe.h r300_reg_safe.h r420_reg_safe.h rs600_reg_safe.h r600_reg_safe.h evergreen_reg_safe.h cayman_reg_safe.h quiet_cmd_mkregtable = MKREGTABLE $@ cmd_mkregtable = $(obj)/mkregtable $< > $@ diff --git a/drivers/gpu/drm/radeon/atom.c b/drivers/gpu/drm/radeon/atom.c index 7bd74568909..e88c64417a8 100644 --- a/drivers/gpu/drm/radeon/atom.c +++ b/drivers/gpu/drm/radeon/atom.c @@ -652,12 +652,12 @@ static void atom_op_compare(atom_exec_context *ctx, int *ptr, int arg) static void atom_op_delay(atom_exec_context *ctx, int *ptr, int arg) { - uint8_t count = U8((*ptr)++); + unsigned count = U8((*ptr)++); SDEBUG(" count: %d\n", count); if (arg == ATOM_UNIT_MICROSEC) udelay(count); else - schedule_timeout_uninterruptible(msecs_to_jiffies(count)); + msleep(count); } static void atom_op_div(atom_exec_context *ctx, int *ptr, int arg) @@ -1245,6 +1245,9 @@ struct atom_context *atom_parse(struct card_info *card, void *bios) char name[512]; int i; + if (!ctx) + return NULL; + ctx->card = card; ctx->bios = bios; diff --git a/drivers/gpu/drm/radeon/atombios.h b/drivers/gpu/drm/radeon/atombios.h index 7fd88497b93..1b50ad8919d 100644 --- a/drivers/gpu/drm/radeon/atombios.h +++ b/drivers/gpu/drm/radeon/atombios.h @@ -726,6 +726,7 @@ typedef struct _DIG_ENCODER_CONTROL_PARAMETERS_V2 #define ATOM_ENCODER_CMD_DP_VIDEO_ON 0x0d #define ATOM_ENCODER_CMD_QUERY_DP_LINK_TRAINING_STATUS 0x0e #define ATOM_ENCODER_CMD_SETUP 0x0f +#define ATOM_ENCODER_CMD_SETUP_PANEL_MODE 0x10 // ucStatus #define ATOM_ENCODER_STATUS_LINK_TRAINING_COMPLETE 0x10 @@ -765,13 +766,19 @@ typedef struct _DIG_ENCODER_CONTROL_PARAMETERS_V3 USHORT usPixelClock; // in 10KHz; for bios convenient ATOM_DIG_ENCODER_CONFIG_V3 acConfig; UCHAR ucAction; - UCHAR ucEncoderMode; + union { + UCHAR ucEncoderMode; // =0: DP encoder // =1: LVDS encoder // =2: DVI encoder // =3: HDMI encoder // =4: SDVO encoder // =5: DP audio + UCHAR ucPanelMode; // only valid when ucAction == ATOM_ENCODER_CMD_SETUP_PANEL_MODE + // =0: external DP + // =1: internal DP2 + // =0x11: internal DP1 for NutMeg/Travis DP translator + }; UCHAR ucLaneNum; // how many lanes to enable UCHAR ucBitPerColor; // only valid for DP mode when ucAction = ATOM_ENCODER_CMD_SETUP UCHAR ucReserved; @@ -816,13 +823,19 @@ typedef struct _DIG_ENCODER_CONTROL_PARAMETERS_V4 UCHAR ucConfig; }; UCHAR ucAction; - UCHAR ucEncoderMode; + union { + UCHAR ucEncoderMode; // =0: DP encoder // =1: LVDS encoder // =2: DVI encoder // =3: HDMI encoder // =4: SDVO encoder // =5: DP audio + UCHAR ucPanelMode; // only valid when ucAction == ATOM_ENCODER_CMD_SETUP_PANEL_MODE + // =0: external DP + // =1: internal DP2 + // =0x11: internal DP1 for NutMeg/Travis DP translator + }; UCHAR ucLaneNum; // how many lanes to enable UCHAR ucBitPerColor; // only valid for DP mode when ucAction = ATOM_ENCODER_CMD_SETUP UCHAR ucHPD_ID; // HPD ID (1-6). =0 means to skip HDP programming. New comparing to previous version @@ -836,6 +849,11 @@ typedef struct _DIG_ENCODER_CONTROL_PARAMETERS_V4 #define PANEL_12BIT_PER_COLOR 0x04 #define PANEL_16BIT_PER_COLOR 0x05 +//define ucPanelMode +#define DP_PANEL_MODE_EXTERNAL_DP_MODE 0x00 +#define DP_PANEL_MODE_INTERNAL_DP2_MODE 0x01 +#define DP_PANEL_MODE_INTERNAL_DP1_MODE 0x11 + /****************************************************************************/ // Structures used by UNIPHYTransmitterControlTable // LVTMATransmitterControlTable @@ -1182,6 +1200,7 @@ typedef struct _EXTERNAL_ENCODER_CONTROL_PARAMETERS_V3 #define EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING_OFF 0x10 #define EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING 0x11 #define EXTERNAL_ENCODER_ACTION_V3_DACLOAD_DETECTION 0x12 +#define EXTERNAL_ENCODER_ACTION_V3_DDC_SETUP 0x14 // ucConfig #define EXTERNAL_ENCODER_CONFIG_V3_DPLINKRATE_MASK 0x03 diff --git a/drivers/gpu/drm/radeon/atombios_crtc.c b/drivers/gpu/drm/radeon/atombios_crtc.c index 529a3a70473..c742944d380 100644 --- a/drivers/gpu/drm/radeon/atombios_crtc.c +++ b/drivers/gpu/drm/radeon/atombios_crtc.c @@ -420,7 +420,7 @@ static void atombios_crtc_program_ss(struct drm_crtc *crtc, if (ASIC_IS_DCE5(rdev)) { args.v3.usSpreadSpectrumAmountFrac = cpu_to_le16(0); - args.v3.ucSpreadSpectrumType = ss->type; + args.v3.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; switch (pll_id) { case ATOM_PPLL1: args.v3.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V3_P1PLL; @@ -440,10 +440,12 @@ static void atombios_crtc_program_ss(struct drm_crtc *crtc, case ATOM_PPLL_INVALID: return; } - args.v2.ucEnable = enable; + args.v3.ucEnable = enable; + if ((ss->percentage == 0) || (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v3.ucEnable = ATOM_DISABLE; } else if (ASIC_IS_DCE4(rdev)) { args.v2.usSpreadSpectrumPercentage = cpu_to_le16(ss->percentage); - args.v2.ucSpreadSpectrumType = ss->type; + args.v2.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; switch (pll_id) { case ATOM_PPLL1: args.v2.ucSpreadSpectrumType |= ATOM_PPLL_SS_TYPE_V2_P1PLL; @@ -464,32 +466,36 @@ static void atombios_crtc_program_ss(struct drm_crtc *crtc, return; } args.v2.ucEnable = enable; + if ((ss->percentage == 0) || (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v2.ucEnable = ATOM_DISABLE; } else if (ASIC_IS_DCE3(rdev)) { args.v1.usSpreadSpectrumPercentage = cpu_to_le16(ss->percentage); - args.v1.ucSpreadSpectrumType = ss->type; + args.v1.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; args.v1.ucSpreadSpectrumStep = ss->step; args.v1.ucSpreadSpectrumDelay = ss->delay; args.v1.ucSpreadSpectrumRange = ss->range; args.v1.ucPpll = pll_id; args.v1.ucEnable = enable; } else if (ASIC_IS_AVIVO(rdev)) { - if (enable == ATOM_DISABLE) { + if ((enable == ATOM_DISABLE) || (ss->percentage == 0) || + (ss->type & ATOM_EXTERNAL_SS_MASK)) { atombios_disable_ss(crtc); return; } args.lvds_ss_2.usSpreadSpectrumPercentage = cpu_to_le16(ss->percentage); - args.lvds_ss_2.ucSpreadSpectrumType = ss->type; + args.lvds_ss_2.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; args.lvds_ss_2.ucSpreadSpectrumStep = ss->step; args.lvds_ss_2.ucSpreadSpectrumDelay = ss->delay; args.lvds_ss_2.ucSpreadSpectrumRange = ss->range; args.lvds_ss_2.ucEnable = enable; } else { - if (enable == ATOM_DISABLE) { + if ((enable == ATOM_DISABLE) || (ss->percentage == 0) || + (ss->type & ATOM_EXTERNAL_SS_MASK)) { atombios_disable_ss(crtc); return; } args.lvds_ss.usSpreadSpectrumPercentage = cpu_to_le16(ss->percentage); - args.lvds_ss.ucSpreadSpectrumType = ss->type; + args.lvds_ss.ucSpreadSpectrumType = ss->type & ATOM_SS_CENTRE_SPREAD_MODE_MASK; args.lvds_ss.ucSpreadSpectrumStepSize_Delay = (ss->step & 3) << 2; args.lvds_ss.ucSpreadSpectrumStepSize_Delay |= (ss->delay & 7) << 4; args.lvds_ss.ucEnable = enable; @@ -512,6 +518,7 @@ static u32 atombios_adjust_pll(struct drm_crtc *crtc, struct radeon_device *rdev = dev->dev_private; struct drm_encoder *encoder = NULL; struct radeon_encoder *radeon_encoder = NULL; + struct drm_connector *connector = NULL; u32 adjusted_clock = mode->clock; int encoder_mode = 0; u32 dp_clock = mode->clock; @@ -546,9 +553,12 @@ static u32 atombios_adjust_pll(struct drm_crtc *crtc, list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { if (encoder->crtc == crtc) { radeon_encoder = to_radeon_encoder(encoder); + connector = radeon_get_connector_for_encoder(encoder); + if (connector) + bpc = connector->display_info.bpc; encoder_mode = atombios_get_encoder_mode(encoder); - if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT | ATOM_DEVICE_DFP_SUPPORT)) { - struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); + if ((radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT | ATOM_DEVICE_DFP_SUPPORT)) || + radeon_encoder_is_dp_bridge(encoder)) { if (connector) { struct radeon_connector *radeon_connector = to_radeon_connector(connector); struct radeon_connector_atom_dig *dig_connector = @@ -612,7 +622,7 @@ static u32 atombios_adjust_pll(struct drm_crtc *crtc, args.v1.usPixelClock = cpu_to_le16(mode->clock / 10); args.v1.ucTransmitterID = radeon_encoder->encoder_id; args.v1.ucEncodeMode = encoder_mode; - if (ss_enabled) + if (ss_enabled && ss->percentage) args.v1.ucConfig |= ADJUST_DISPLAY_CONFIG_SS_ENABLE; @@ -625,10 +635,11 @@ static u32 atombios_adjust_pll(struct drm_crtc *crtc, args.v3.sInput.ucTransmitterID = radeon_encoder->encoder_id; args.v3.sInput.ucEncodeMode = encoder_mode; args.v3.sInput.ucDispPllConfig = 0; - if (ss_enabled) + if (ss_enabled && ss->percentage) args.v3.sInput.ucDispPllConfig |= DISPPLL_CONFIG_SS_ENABLE; - if (radeon_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) { + if (radeon_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT) || + radeon_encoder_is_dp_bridge(encoder)) { struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; if (encoder_mode == ATOM_ENCODER_MODE_DP) { args.v3.sInput.ucDispPllConfig |= @@ -660,6 +671,13 @@ static u32 atombios_adjust_pll(struct drm_crtc *crtc, DISPPLL_CONFIG_DUAL_LINK; } } + if (radeon_encoder_is_dp_bridge(encoder)) { + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + struct radeon_encoder *ext_radeon_encoder = to_radeon_encoder(ext_encoder); + args.v3.sInput.ucExtTransmitterID = ext_radeon_encoder->encoder_id; + } else + args.v3.sInput.ucExtTransmitterID = 0; + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); adjusted_clock = le32_to_cpu(args.v3.sOutput.ulDispPllFreq) * 10; @@ -746,7 +764,7 @@ static void atombios_crtc_set_dcpll(struct drm_crtc *crtc, } static void atombios_crtc_program_pll(struct drm_crtc *crtc, - int crtc_id, + u32 crtc_id, int pll_id, u32 encoder_mode, u32 encoder_id, @@ -754,7 +772,10 @@ static void atombios_crtc_program_pll(struct drm_crtc *crtc, u32 ref_div, u32 fb_div, u32 frac_fb_div, - u32 post_div) + u32 post_div, + int bpc, + bool ss_enabled, + struct radeon_atom_ss *ss) { struct drm_device *dev = crtc->dev; struct radeon_device *rdev = dev->dev_private; @@ -801,6 +822,8 @@ static void atombios_crtc_program_pll(struct drm_crtc *crtc, args.v3.ucPostDiv = post_div; args.v3.ucPpll = pll_id; args.v3.ucMiscInfo = (pll_id << 2); + if (ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v3.ucMiscInfo |= PIXEL_CLOCK_MISC_REF_DIV_SRC; args.v3.ucTransmitterId = encoder_id; args.v3.ucEncoderMode = encoder_mode; break; @@ -812,18 +835,45 @@ static void atombios_crtc_program_pll(struct drm_crtc *crtc, args.v5.ulFbDivDecFrac = cpu_to_le32(frac_fb_div * 100000); args.v5.ucPostDiv = post_div; args.v5.ucMiscInfo = 0; /* HDMI depth, etc. */ + if (ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_REF_DIV_SRC; + switch (bpc) { + case 8: + default: + args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_24BPP; + break; + case 10: + args.v5.ucMiscInfo |= PIXEL_CLOCK_V5_MISC_HDMI_30BPP; + break; + } args.v5.ucTransmitterID = encoder_id; args.v5.ucEncoderMode = encoder_mode; args.v5.ucPpll = pll_id; break; case 6: - args.v6.ulCrtcPclkFreq.ucCRTC = crtc_id; - args.v6.ulCrtcPclkFreq.ulPixelClock = cpu_to_le32(clock / 10); + args.v6.ulDispEngClkFreq = cpu_to_le32(crtc_id << 24 | clock / 10); args.v6.ucRefDiv = ref_div; args.v6.usFbDiv = cpu_to_le16(fb_div); args.v6.ulFbDivDecFrac = cpu_to_le32(frac_fb_div * 100000); args.v6.ucPostDiv = post_div; args.v6.ucMiscInfo = 0; /* HDMI depth, etc. */ + if (ss_enabled && (ss->type & ATOM_EXTERNAL_SS_MASK)) + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_REF_DIV_SRC; + switch (bpc) { + case 8: + default: + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_24BPP; + break; + case 10: + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_30BPP; + break; + case 12: + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_36BPP; + break; + case 16: + args.v6.ucMiscInfo |= PIXEL_CLOCK_V6_MISC_HDMI_48BPP; + break; + } args.v6.ucTransmitterID = encoder_id; args.v6.ucEncoderMode = encoder_mode; args.v6.ucPpll = pll_id; @@ -855,6 +905,7 @@ static void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode int encoder_mode = 0; struct radeon_atom_ss ss; bool ss_enabled = false; + int bpc = 8; list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { if (encoder->crtc == crtc) { @@ -891,41 +942,30 @@ static void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; int dp_clock; + bpc = connector->display_info.bpc; switch (encoder_mode) { case ATOM_ENCODER_MODE_DP: /* DP/eDP */ dp_clock = dig_connector->dp_clock / 10; - if (radeon_encoder->active_device & (ATOM_DEVICE_LCD_SUPPORT)) { - if (ASIC_IS_DCE4(rdev)) - ss_enabled = - radeon_atombios_get_asic_ss_info(rdev, &ss, - dig->lcd_ss_id, - dp_clock); - else + if (ASIC_IS_DCE4(rdev)) + ss_enabled = + radeon_atombios_get_asic_ss_info(rdev, &ss, + ASIC_INTERNAL_SS_ON_DP, + dp_clock); + else { + if (dp_clock == 16200) { ss_enabled = radeon_atombios_get_ppll_ss_info(rdev, &ss, - dig->lcd_ss_id); - } else { - if (ASIC_IS_DCE4(rdev)) - ss_enabled = - radeon_atombios_get_asic_ss_info(rdev, &ss, - ASIC_INTERNAL_SS_ON_DP, - dp_clock); - else { - if (dp_clock == 16200) { - ss_enabled = - radeon_atombios_get_ppll_ss_info(rdev, &ss, - ATOM_DP_SS_ID2); - if (!ss_enabled) - ss_enabled = - radeon_atombios_get_ppll_ss_info(rdev, &ss, - ATOM_DP_SS_ID1); - } else + ATOM_DP_SS_ID2); + if (!ss_enabled) ss_enabled = radeon_atombios_get_ppll_ss_info(rdev, &ss, ATOM_DP_SS_ID1); - } + } else + ss_enabled = + radeon_atombios_get_ppll_ss_info(rdev, &ss, + ATOM_DP_SS_ID1); } break; case ATOM_ENCODER_MODE_LVDS: @@ -974,7 +1014,7 @@ static void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode atombios_crtc_program_pll(crtc, radeon_crtc->crtc_id, radeon_crtc->pll_id, encoder_mode, radeon_encoder->encoder_id, mode->clock, - ref_div, fb_div, frac_fb_div, post_div); + ref_div, fb_div, frac_fb_div, post_div, bpc, ss_enabled, &ss); if (ss_enabled) { /* calculate ss amount and step size */ @@ -982,7 +1022,7 @@ static void atombios_crtc_set_pll(struct drm_crtc *crtc, struct drm_display_mode u32 step_size; u32 amount = (((fb_div * 10) + frac_fb_div) * ss.percentage) / 10000; ss.amount = (amount / 10) & ATOM_PPLL_SS_AMOUNT_V2_FBDIV_MASK; - ss.amount |= ((amount - (ss.amount * 10)) << ATOM_PPLL_SS_AMOUNT_V2_NFRAC_SHIFT) & + ss.amount |= ((amount - (amount / 10)) << ATOM_PPLL_SS_AMOUNT_V2_NFRAC_SHIFT) & ATOM_PPLL_SS_AMOUNT_V2_NFRAC_MASK; if (ss.type & ATOM_PPLL_SS_TYPE_V2_CENTRE_SPREAD) step_size = (4 * amount * ref_div * (ss.rate * 2048)) / @@ -1011,7 +1051,7 @@ static int dce4_crtc_do_set_base(struct drm_crtc *crtc, uint64_t fb_location; uint32_t fb_format, fb_pitch_pixels, tiling_flags; u32 fb_swap = EVERGREEN_GRPH_ENDIAN_SWAP(EVERGREEN_GRPH_ENDIAN_NONE); - u32 tmp; + u32 tmp, viewport_w, viewport_h; int r; /* no fb bound */ @@ -1137,8 +1177,10 @@ static int dce4_crtc_do_set_base(struct drm_crtc *crtc, y &= ~1; WREG32(EVERGREEN_VIEWPORT_START + radeon_crtc->crtc_offset, (x << 16) | y); + viewport_w = crtc->mode.hdisplay; + viewport_h = (crtc->mode.vdisplay + 1) & ~1; WREG32(EVERGREEN_VIEWPORT_SIZE + radeon_crtc->crtc_offset, - (crtc->mode.hdisplay << 16) | crtc->mode.vdisplay); + (viewport_w << 16) | viewport_h); /* pageflip setup */ /* make sure flip is at vb rather than hb */ @@ -1179,7 +1221,7 @@ static int avivo_crtc_do_set_base(struct drm_crtc *crtc, uint64_t fb_location; uint32_t fb_format, fb_pitch_pixels, tiling_flags; u32 fb_swap = R600_D1GRPH_SWAP_ENDIAN_NONE; - u32 tmp; + u32 tmp, viewport_w, viewport_h; int r; /* no fb bound */ @@ -1304,8 +1346,10 @@ static int avivo_crtc_do_set_base(struct drm_crtc *crtc, y &= ~1; WREG32(AVIVO_D1MODE_VIEWPORT_START + radeon_crtc->crtc_offset, (x << 16) | y); + viewport_w = crtc->mode.hdisplay; + viewport_h = (crtc->mode.vdisplay + 1) & ~1; WREG32(AVIVO_D1MODE_VIEWPORT_SIZE + radeon_crtc->crtc_offset, - (crtc->mode.hdisplay << 16) | crtc->mode.vdisplay); + (viewport_w << 16) | viewport_h); /* pageflip setup */ /* make sure flip is at vb rather than hb */ @@ -1395,11 +1439,19 @@ static int radeon_atom_pick_pll(struct drm_crtc *crtc) uint32_t pll_in_use = 0; if (ASIC_IS_DCE4(rdev)) { - /* if crtc is driving DP and we have an ext clock, use that */ list_for_each_entry(test_encoder, &dev->mode_config.encoder_list, head) { if (test_encoder->crtc && (test_encoder->crtc == crtc)) { + /* in DP mode, the DP ref clock can come from PPLL, DCPLL, or ext clock, + * depending on the asic: + * DCE4: PPLL or ext clock + * DCE5: DCPLL or ext clock + * + * Setting ATOM_PPLL_INVALID will cause SetPixelClock to skip + * PPLL/DCPLL programming and only program the DP DTO for the + * crtc virtual pixel clock. + */ if (atombios_get_encoder_mode(test_encoder) == ATOM_ENCODER_MODE_DP) { - if (rdev->clock.dp_extclk) + if (ASIC_IS_DCE5(rdev) || rdev->clock.dp_extclk) return ATOM_PPLL_INVALID; } } @@ -1515,6 +1567,8 @@ static void atombios_crtc_commit(struct drm_crtc *crtc) static void atombios_crtc_disable(struct drm_crtc *crtc) { struct radeon_crtc *radeon_crtc = to_radeon_crtc(crtc); + struct radeon_atom_ss ss; + atombios_crtc_dpms(crtc, DRM_MODE_DPMS_OFF); switch (radeon_crtc->pll_id) { @@ -1522,7 +1576,7 @@ static void atombios_crtc_disable(struct drm_crtc *crtc) case ATOM_PPLL2: /* disable the ppll */ atombios_crtc_program_pll(crtc, radeon_crtc->crtc_id, radeon_crtc->pll_id, - 0, 0, ATOM_DISABLE, 0, 0, 0, 0); + 0, 0, ATOM_DISABLE, 0, 0, 0, 0, 0, false, &ss); break; default: break; diff --git a/drivers/gpu/drm/radeon/atombios_dp.c b/drivers/gpu/drm/radeon/atombios_dp.c index 695de9a3850..7ad43c6b1db 100644 --- a/drivers/gpu/drm/radeon/atombios_dp.c +++ b/drivers/gpu/drm/radeon/atombios_dp.c @@ -43,158 +43,242 @@ static char *pre_emph_names[] = { "0dB", "3.5dB", "6dB", "9.5dB" }; -static const int dp_clocks[] = { - 54000, /* 1 lane, 1.62 Ghz */ - 90000, /* 1 lane, 2.70 Ghz */ - 108000, /* 2 lane, 1.62 Ghz */ - 180000, /* 2 lane, 2.70 Ghz */ - 216000, /* 4 lane, 1.62 Ghz */ - 360000, /* 4 lane, 2.70 Ghz */ +/***** radeon AUX functions *****/ +union aux_channel_transaction { + PROCESS_AUX_CHANNEL_TRANSACTION_PS_ALLOCATION v1; + PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS_V2 v2; }; -static const int num_dp_clocks = sizeof(dp_clocks) / sizeof(int); +static int radeon_process_aux_ch(struct radeon_i2c_chan *chan, + u8 *send, int send_bytes, + u8 *recv, int recv_size, + u8 delay, u8 *ack) +{ + struct drm_device *dev = chan->dev; + struct radeon_device *rdev = dev->dev_private; + union aux_channel_transaction args; + int index = GetIndexIntoMasterTable(COMMAND, ProcessAuxChannelTransaction); + unsigned char *base; + int recv_bytes; + + memset(&args, 0, sizeof(args)); -/* common helper functions */ -static int dp_lanes_for_mode_clock(u8 dpcd[DP_DPCD_SIZE], int mode_clock) + base = (unsigned char *)rdev->mode_info.atom_context->scratch; + + memcpy(base, send, send_bytes); + + args.v1.lpAuxRequest = 0; + args.v1.lpDataOut = 16; + args.v1.ucDataOutLen = 0; + args.v1.ucChannelID = chan->rec.i2c_id; + args.v1.ucDelay = delay / 10; + if (ASIC_IS_DCE4(rdev)) + args.v2.ucHPD_ID = chan->rec.hpd; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + + *ack = args.v1.ucReplyStatus; + + /* timeout */ + if (args.v1.ucReplyStatus == 1) { + DRM_DEBUG_KMS("dp_aux_ch timeout\n"); + return -ETIMEDOUT; + } + + /* flags not zero */ + if (args.v1.ucReplyStatus == 2) { + DRM_DEBUG_KMS("dp_aux_ch flags not zero\n"); + return -EBUSY; + } + + /* error */ + if (args.v1.ucReplyStatus == 3) { + DRM_DEBUG_KMS("dp_aux_ch error\n"); + return -EIO; + } + + recv_bytes = args.v1.ucDataOutLen; + if (recv_bytes > recv_size) + recv_bytes = recv_size; + + if (recv && recv_size) + memcpy(recv, base + 16, recv_bytes); + + return recv_bytes; +} + +static int radeon_dp_aux_native_write(struct radeon_connector *radeon_connector, + u16 address, u8 *send, u8 send_bytes, u8 delay) { - int i; - u8 max_link_bw; - u8 max_lane_count; + struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; + int ret; + u8 msg[20]; + int msg_bytes = send_bytes + 4; + u8 ack; - if (!dpcd) - return 0; + if (send_bytes > 16) + return -1; - max_link_bw = dpcd[DP_MAX_LINK_RATE]; - max_lane_count = dpcd[DP_MAX_LANE_COUNT] & DP_MAX_LANE_COUNT_MASK; + msg[0] = address; + msg[1] = address >> 8; + msg[2] = AUX_NATIVE_WRITE << 4; + msg[3] = (msg_bytes << 4) | (send_bytes - 1); + memcpy(&msg[4], send, send_bytes); - switch (max_link_bw) { - case DP_LINK_BW_1_62: - default: - for (i = 0; i < num_dp_clocks; i++) { - if (i % 2) - continue; - switch (max_lane_count) { - case 1: - if (i > 1) - return 0; - break; - case 2: - if (i > 3) - return 0; - break; - case 4: - default: - break; - } - if (dp_clocks[i] > mode_clock) { - if (i < 2) - return 1; - else if (i < 4) - return 2; - else - return 4; - } - } - break; - case DP_LINK_BW_2_7: - for (i = 0; i < num_dp_clocks; i++) { - switch (max_lane_count) { - case 1: - if (i > 1) - return 0; - break; - case 2: - if (i > 3) - return 0; - break; - case 4: - default: - break; - } - if (dp_clocks[i] > mode_clock) { - if (i < 2) - return 1; - else if (i < 4) - return 2; - else - return 4; - } - } - break; + while (1) { + ret = radeon_process_aux_ch(dig_connector->dp_i2c_bus, + msg, msg_bytes, NULL, 0, delay, &ack); + if (ret < 0) + return ret; + if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_ACK) + break; + else if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_DEFER) + udelay(400); + else + return -EIO; } - return 0; + return send_bytes; } -static int dp_link_clock_for_mode_clock(u8 dpcd[DP_DPCD_SIZE], int mode_clock) +static int radeon_dp_aux_native_read(struct radeon_connector *radeon_connector, + u16 address, u8 *recv, int recv_bytes, u8 delay) { - int i; - u8 max_link_bw; - u8 max_lane_count; + struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; + u8 msg[4]; + int msg_bytes = 4; + u8 ack; + int ret; - if (!dpcd) - return 0; + msg[0] = address; + msg[1] = address >> 8; + msg[2] = AUX_NATIVE_READ << 4; + msg[3] = (msg_bytes << 4) | (recv_bytes - 1); + + while (1) { + ret = radeon_process_aux_ch(dig_connector->dp_i2c_bus, + msg, msg_bytes, recv, recv_bytes, delay, &ack); + if (ret == 0) + return -EPROTO; + if (ret < 0) + return ret; + if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_ACK) + return ret; + else if ((ack & AUX_NATIVE_REPLY_MASK) == AUX_NATIVE_REPLY_DEFER) + udelay(400); + else + return -EIO; + } +} - max_link_bw = dpcd[DP_MAX_LINK_RATE]; - max_lane_count = dpcd[DP_MAX_LANE_COUNT] & DP_MAX_LANE_COUNT_MASK; +static void radeon_write_dpcd_reg(struct radeon_connector *radeon_connector, + u16 reg, u8 val) +{ + radeon_dp_aux_native_write(radeon_connector, reg, &val, 1, 0); +} - switch (max_link_bw) { - case DP_LINK_BW_1_62: +static u8 radeon_read_dpcd_reg(struct radeon_connector *radeon_connector, + u16 reg) +{ + u8 val = 0; + + radeon_dp_aux_native_read(radeon_connector, reg, &val, 1, 0); + + return val; +} + +int radeon_dp_i2c_aux_ch(struct i2c_adapter *adapter, int mode, + u8 write_byte, u8 *read_byte) +{ + struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; + struct radeon_i2c_chan *auxch = (struct radeon_i2c_chan *)adapter; + u16 address = algo_data->address; + u8 msg[5]; + u8 reply[2]; + unsigned retry; + int msg_bytes; + int reply_bytes = 1; + int ret; + u8 ack; + + /* Set up the command byte */ + if (mode & MODE_I2C_READ) + msg[2] = AUX_I2C_READ << 4; + else + msg[2] = AUX_I2C_WRITE << 4; + + if (!(mode & MODE_I2C_STOP)) + msg[2] |= AUX_I2C_MOT << 4; + + msg[0] = address; + msg[1] = address >> 8; + + switch (mode) { + case MODE_I2C_WRITE: + msg_bytes = 5; + msg[3] = msg_bytes << 4; + msg[4] = write_byte; + break; + case MODE_I2C_READ: + msg_bytes = 4; + msg[3] = msg_bytes << 4; + break; default: - for (i = 0; i < num_dp_clocks; i++) { - if (i % 2) - continue; - switch (max_lane_count) { - case 1: - if (i > 1) - return 0; - break; - case 2: - if (i > 3) - return 0; - break; - case 4: - default: - break; - } - if (dp_clocks[i] > mode_clock) - return 162000; - } + msg_bytes = 4; + msg[3] = 3 << 4; break; - case DP_LINK_BW_2_7: - for (i = 0; i < num_dp_clocks; i++) { - switch (max_lane_count) { - case 1: - if (i > 1) - return 0; - break; - case 2: - if (i > 3) - return 0; - break; - case 4: - default: - break; - } - if (dp_clocks[i] > mode_clock) - return (i % 2) ? 270000 : 162000; - } } - return 0; -} + for (retry = 0; retry < 4; retry++) { + ret = radeon_process_aux_ch(auxch, + msg, msg_bytes, reply, reply_bytes, 0, &ack); + if (ret < 0) { + DRM_DEBUG_KMS("aux_ch failed %d\n", ret); + return ret; + } -int dp_mode_valid(u8 dpcd[DP_DPCD_SIZE], int mode_clock) -{ - int lanes = dp_lanes_for_mode_clock(dpcd, mode_clock); - int dp_clock = dp_link_clock_for_mode_clock(dpcd, mode_clock); + switch (ack & AUX_NATIVE_REPLY_MASK) { + case AUX_NATIVE_REPLY_ACK: + /* I2C-over-AUX Reply field is only valid + * when paired with AUX ACK. + */ + break; + case AUX_NATIVE_REPLY_NACK: + DRM_DEBUG_KMS("aux_ch native nack\n"); + return -EREMOTEIO; + case AUX_NATIVE_REPLY_DEFER: + DRM_DEBUG_KMS("aux_ch native defer\n"); + udelay(400); + continue; + default: + DRM_ERROR("aux_ch invalid native reply 0x%02x\n", ack); + return -EREMOTEIO; + } - if ((lanes == 0) || (dp_clock == 0)) - return MODE_CLOCK_HIGH; + switch (ack & AUX_I2C_REPLY_MASK) { + case AUX_I2C_REPLY_ACK: + if (mode == MODE_I2C_READ) + *read_byte = reply[0]; + return ret; + case AUX_I2C_REPLY_NACK: + DRM_DEBUG_KMS("aux_i2c nack\n"); + return -EREMOTEIO; + case AUX_I2C_REPLY_DEFER: + DRM_DEBUG_KMS("aux_i2c defer\n"); + udelay(400); + break; + default: + DRM_ERROR("aux_i2c invalid reply 0x%02x\n", ack); + return -EREMOTEIO; + } + } - return MODE_OK; + DRM_ERROR("aux i2c too many retries, giving up\n"); + return -EREMOTEIO; } +/***** general DP utility functions *****/ + static u8 dp_link_status(u8 link_status[DP_LINK_STATUS_SIZE], int r) { return link_status[r - DP_LANE0_1_STATUS]; @@ -242,7 +326,7 @@ static bool dp_channel_eq_ok(u8 link_status[DP_LINK_STATUS_SIZE], return true; } -static u8 dp_get_adjust_request_voltage(uint8_t link_status[DP_LINK_STATUS_SIZE], +static u8 dp_get_adjust_request_voltage(u8 link_status[DP_LINK_STATUS_SIZE], int lane) { @@ -255,7 +339,7 @@ static u8 dp_get_adjust_request_voltage(uint8_t link_status[DP_LINK_STATUS_SIZE] return ((l >> s) & 0x3) << DP_TRAIN_VOLTAGE_SWING_SHIFT; } -static u8 dp_get_adjust_request_pre_emphasis(uint8_t link_status[DP_LINK_STATUS_SIZE], +static u8 dp_get_adjust_request_pre_emphasis(u8 link_status[DP_LINK_STATUS_SIZE], int lane) { int i = DP_ADJUST_REQUEST_LANE0_1 + (lane >> 1); @@ -267,22 +351,8 @@ static u8 dp_get_adjust_request_pre_emphasis(uint8_t link_status[DP_LINK_STATUS_ return ((l >> s) & 0x3) << DP_TRAIN_PRE_EMPHASIS_SHIFT; } -/* XXX fix me -- chip specific */ #define DP_VOLTAGE_MAX DP_TRAIN_VOLTAGE_SWING_1200 -static u8 dp_pre_emphasis_max(u8 voltage_swing) -{ - switch (voltage_swing & DP_TRAIN_VOLTAGE_SWING_MASK) { - case DP_TRAIN_VOLTAGE_SWING_400: - return DP_TRAIN_PRE_EMPHASIS_6; - case DP_TRAIN_VOLTAGE_SWING_600: - return DP_TRAIN_PRE_EMPHASIS_6; - case DP_TRAIN_VOLTAGE_SWING_800: - return DP_TRAIN_PRE_EMPHASIS_3_5; - case DP_TRAIN_VOLTAGE_SWING_1200: - default: - return DP_TRAIN_PRE_EMPHASIS_0; - } -} +#define DP_PRE_EMPHASIS_MAX DP_TRAIN_PRE_EMPHASIS_9_5 static void dp_get_adjust_train(u8 link_status[DP_LINK_STATUS_SIZE], int lane_count, @@ -308,10 +378,10 @@ static void dp_get_adjust_train(u8 link_status[DP_LINK_STATUS_SIZE], } if (v >= DP_VOLTAGE_MAX) - v = DP_VOLTAGE_MAX | DP_TRAIN_MAX_SWING_REACHED; + v |= DP_TRAIN_MAX_SWING_REACHED; - if (p >= dp_pre_emphasis_max(v)) - p = dp_pre_emphasis_max(v) | DP_TRAIN_MAX_PRE_EMPHASIS_REACHED; + if (p >= DP_PRE_EMPHASIS_MAX) + p |= DP_TRAIN_MAX_PRE_EMPHASIS_REACHED; DRM_DEBUG_KMS("using signal parameters: voltage %s pre_emph %s\n", voltage_names[(v & DP_TRAIN_VOLTAGE_SWING_MASK) >> DP_TRAIN_VOLTAGE_SWING_SHIFT], @@ -321,110 +391,109 @@ static void dp_get_adjust_train(u8 link_status[DP_LINK_STATUS_SIZE], train_set[lane] = v | p; } -union aux_channel_transaction { - PROCESS_AUX_CHANNEL_TRANSACTION_PS_ALLOCATION v1; - PROCESS_AUX_CHANNEL_TRANSACTION_PARAMETERS_V2 v2; -}; - -/* radeon aux chan functions */ -bool radeon_process_aux_ch(struct radeon_i2c_chan *chan, u8 *req_bytes, - int num_bytes, u8 *read_byte, - u8 read_buf_len, u8 delay) +/* convert bits per color to bits per pixel */ +/* get bpc from the EDID */ +static int convert_bpc_to_bpp(int bpc) { - struct drm_device *dev = chan->dev; - struct radeon_device *rdev = dev->dev_private; - union aux_channel_transaction args; - int index = GetIndexIntoMasterTable(COMMAND, ProcessAuxChannelTransaction); - unsigned char *base; - int retry_count = 0; - - memset(&args, 0, sizeof(args)); - - base = (unsigned char *)rdev->mode_info.atom_context->scratch; - -retry: - memcpy(base, req_bytes, num_bytes); - - args.v1.lpAuxRequest = 0; - args.v1.lpDataOut = 16; - args.v1.ucDataOutLen = 0; - args.v1.ucChannelID = chan->rec.i2c_id; - args.v1.ucDelay = delay / 10; - if (ASIC_IS_DCE4(rdev)) - args.v2.ucHPD_ID = chan->rec.hpd; + if (bpc == 0) + return 24; + else + return bpc * 3; +} - atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); +/* get the max pix clock supported by the link rate and lane num */ +static int dp_get_max_dp_pix_clock(int link_rate, + int lane_num, + int bpp) +{ + return (link_rate * lane_num * 8) / bpp; +} - if (args.v1.ucReplyStatus && !args.v1.ucDataOutLen) { - if (args.v1.ucReplyStatus == 0x20 && retry_count++ < 10) - goto retry; - DRM_DEBUG_KMS("failed to get auxch %02x%02x %02x %02x 0x%02x %02x after %d retries\n", - req_bytes[1], req_bytes[0], req_bytes[2], req_bytes[3], - chan->rec.i2c_id, args.v1.ucReplyStatus, retry_count); - return false; +static int dp_get_max_link_rate(u8 dpcd[DP_DPCD_SIZE]) +{ + switch (dpcd[DP_MAX_LINK_RATE]) { + case DP_LINK_BW_1_62: + default: + return 162000; + case DP_LINK_BW_2_7: + return 270000; + case DP_LINK_BW_5_4: + return 540000; } +} - if (args.v1.ucDataOutLen && read_byte && read_buf_len) { - if (read_buf_len < args.v1.ucDataOutLen) { - DRM_ERROR("Buffer to small for return answer %d %d\n", - read_buf_len, args.v1.ucDataOutLen); - return false; - } - { - int len = min(read_buf_len, args.v1.ucDataOutLen); - memcpy(read_byte, base + 16, len); - } - } - return true; +static u8 dp_get_max_lane_number(u8 dpcd[DP_DPCD_SIZE]) +{ + return dpcd[DP_MAX_LANE_COUNT] & DP_MAX_LANE_COUNT_MASK; } -bool radeon_dp_aux_native_write(struct radeon_connector *radeon_connector, uint16_t address, - uint8_t send_bytes, uint8_t *send) +static u8 dp_get_dp_link_rate_coded(int link_rate) { - struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; - u8 msg[20]; - u8 msg_len, dp_msg_len; - bool ret; + switch (link_rate) { + case 162000: + default: + return DP_LINK_BW_1_62; + case 270000: + return DP_LINK_BW_2_7; + case 540000: + return DP_LINK_BW_5_4; + } +} - dp_msg_len = 4; - msg[0] = address; - msg[1] = address >> 8; - msg[2] = AUX_NATIVE_WRITE << 4; - dp_msg_len += send_bytes; - msg[3] = (dp_msg_len << 4) | (send_bytes - 1); +/***** radeon specific DP functions *****/ - if (send_bytes > 16) - return false; +/* First get the min lane# when low rate is used according to pixel clock + * (prefer low rate), second check max lane# supported by DP panel, + * if the max lane# < low rate lane# then use max lane# instead. + */ +static int radeon_dp_get_dp_lane_number(struct drm_connector *connector, + u8 dpcd[DP_DPCD_SIZE], + int pix_clock) +{ + int bpp = convert_bpc_to_bpp(connector->display_info.bpc); + int max_link_rate = dp_get_max_link_rate(dpcd); + int max_lane_num = dp_get_max_lane_number(dpcd); + int lane_num; + int max_dp_pix_clock; + + for (lane_num = 1; lane_num < max_lane_num; lane_num <<= 1) { + max_dp_pix_clock = dp_get_max_dp_pix_clock(max_link_rate, lane_num, bpp); + if (pix_clock <= max_dp_pix_clock) + break; + } - memcpy(&msg[4], send, send_bytes); - msg_len = 4 + send_bytes; - ret = radeon_process_aux_ch(dig_connector->dp_i2c_bus, msg, msg_len, NULL, 0, 0); - return ret; + return lane_num; } -bool radeon_dp_aux_native_read(struct radeon_connector *radeon_connector, uint16_t address, - uint8_t delay, uint8_t expected_bytes, - uint8_t *read_p) +static int radeon_dp_get_dp_link_clock(struct drm_connector *connector, + u8 dpcd[DP_DPCD_SIZE], + int pix_clock) { - struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; - u8 msg[20]; - u8 msg_len, dp_msg_len; - bool ret = false; - msg_len = 4; - dp_msg_len = 4; - msg[0] = address; - msg[1] = address >> 8; - msg[2] = AUX_NATIVE_READ << 4; - msg[3] = (dp_msg_len) << 4; - msg[3] |= expected_bytes - 1; + int bpp = convert_bpc_to_bpp(connector->display_info.bpc); + int lane_num, max_pix_clock; + + if (radeon_connector_encoder_is_dp_bridge(connector)) + return 270000; + + lane_num = radeon_dp_get_dp_lane_number(connector, dpcd, pix_clock); + max_pix_clock = dp_get_max_dp_pix_clock(162000, lane_num, bpp); + if (pix_clock <= max_pix_clock) + return 162000; + max_pix_clock = dp_get_max_dp_pix_clock(270000, lane_num, bpp); + if (pix_clock <= max_pix_clock) + return 270000; + if (radeon_connector_is_dp12_capable(connector)) { + max_pix_clock = dp_get_max_dp_pix_clock(540000, lane_num, bpp); + if (pix_clock <= max_pix_clock) + return 540000; + } - ret = radeon_process_aux_ch(dig_connector->dp_i2c_bus, msg, msg_len, read_p, expected_bytes, delay); - return ret; + return dp_get_max_link_rate(dpcd); } -/* radeon dp functions */ -static u8 radeon_dp_encoder_service(struct radeon_device *rdev, int action, int dp_clock, - uint8_t ucconfig, uint8_t lane_num) +static u8 radeon_dp_encoder_service(struct radeon_device *rdev, + int action, int dp_clock, + u8 ucconfig, u8 lane_num) { DP_ENCODER_SERVICE_PARAMETERS args; int index = GetIndexIntoMasterTable(COMMAND, DPEncoderService); @@ -454,60 +523,86 @@ bool radeon_dp_getdpcd(struct radeon_connector *radeon_connector) { struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; u8 msg[25]; - int ret; + int ret, i; - ret = radeon_dp_aux_native_read(radeon_connector, DP_DPCD_REV, 0, 8, msg); - if (ret) { + ret = radeon_dp_aux_native_read(radeon_connector, DP_DPCD_REV, msg, 8, 0); + if (ret > 0) { memcpy(dig_connector->dpcd, msg, 8); - { - int i; - DRM_DEBUG_KMS("DPCD: "); - for (i = 0; i < 8; i++) - DRM_DEBUG_KMS("%02x ", msg[i]); - DRM_DEBUG_KMS("\n"); - } + DRM_DEBUG_KMS("DPCD: "); + for (i = 0; i < 8; i++) + DRM_DEBUG_KMS("%02x ", msg[i]); + DRM_DEBUG_KMS("\n"); return true; } dig_connector->dpcd[0] = 0; return false; } +static void radeon_dp_set_panel_mode(struct drm_encoder *encoder, + struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + int panel_mode = DP_PANEL_MODE_EXTERNAL_DP_MODE; + + if (!ASIC_IS_DCE4(rdev)) + return; + + if (radeon_connector_encoder_is_dp_bridge(connector)) + panel_mode = DP_PANEL_MODE_INTERNAL_DP1_MODE; + + atombios_dig_encoder_setup(encoder, + ATOM_ENCODER_CMD_SETUP_PANEL_MODE, + panel_mode); +} + void radeon_dp_set_link_config(struct drm_connector *connector, struct drm_display_mode *mode) { - struct radeon_connector *radeon_connector; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); struct radeon_connector_atom_dig *dig_connector; - if ((connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort) && - (connector->connector_type != DRM_MODE_CONNECTOR_eDP)) - return; - - radeon_connector = to_radeon_connector(connector); if (!radeon_connector->con_priv) return; dig_connector = radeon_connector->con_priv; - dig_connector->dp_clock = - dp_link_clock_for_mode_clock(dig_connector->dpcd, mode->clock); - dig_connector->dp_lane_count = - dp_lanes_for_mode_clock(dig_connector->dpcd, mode->clock); + if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) || + (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) { + dig_connector->dp_clock = + radeon_dp_get_dp_link_clock(connector, dig_connector->dpcd, mode->clock); + dig_connector->dp_lane_count = + radeon_dp_get_dp_lane_number(connector, dig_connector->dpcd, mode->clock); + } } -int radeon_dp_mode_valid_helper(struct radeon_connector *radeon_connector, +int radeon_dp_mode_valid_helper(struct drm_connector *connector, struct drm_display_mode *mode) { - struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct radeon_connector_atom_dig *dig_connector; + int dp_clock; + + if (!radeon_connector->con_priv) + return MODE_CLOCK_HIGH; + dig_connector = radeon_connector->con_priv; + + dp_clock = + radeon_dp_get_dp_link_clock(connector, dig_connector->dpcd, mode->clock); + + if ((dp_clock == 540000) && + (!radeon_connector_is_dp12_capable(connector))) + return MODE_CLOCK_HIGH; - return dp_mode_valid(dig_connector->dpcd, mode->clock); + return MODE_OK; } -static bool atom_dp_get_link_status(struct radeon_connector *radeon_connector, - u8 link_status[DP_LINK_STATUS_SIZE]) +static bool radeon_dp_get_link_status(struct radeon_connector *radeon_connector, + u8 link_status[DP_LINK_STATUS_SIZE]) { int ret; - ret = radeon_dp_aux_native_read(radeon_connector, DP_LANE0_1_STATUS, 100, - DP_LINK_STATUS_SIZE, link_status); - if (!ret) { + ret = radeon_dp_aux_native_read(radeon_connector, DP_LANE0_1_STATUS, + link_status, DP_LINK_STATUS_SIZE, 100); + if (ret <= 0) { DRM_ERROR("displayport link status failed\n"); return false; } @@ -520,290 +615,333 @@ static bool atom_dp_get_link_status(struct radeon_connector *radeon_connector, bool radeon_dp_needs_link_train(struct radeon_connector *radeon_connector) { - struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; u8 link_status[DP_LINK_STATUS_SIZE]; + struct radeon_connector_atom_dig *dig = radeon_connector->con_priv; - if (!atom_dp_get_link_status(radeon_connector, link_status)) + if (!radeon_dp_get_link_status(radeon_connector, link_status)) return false; - if (dp_channel_eq_ok(link_status, dig_connector->dp_lane_count)) + if (dp_channel_eq_ok(link_status, dig->dp_lane_count)) return false; return true; } -static void dp_set_power(struct radeon_connector *radeon_connector, u8 power_state) -{ - struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; - - if (dig_connector->dpcd[0] >= 0x11) { - radeon_dp_aux_native_write(radeon_connector, DP_SET_POWER, 1, - &power_state); - } -} - -static void dp_set_downspread(struct radeon_connector *radeon_connector, u8 downspread) -{ - radeon_dp_aux_native_write(radeon_connector, DP_DOWNSPREAD_CTRL, 1, - &downspread); -} +struct radeon_dp_link_train_info { + struct radeon_device *rdev; + struct drm_encoder *encoder; + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + int enc_id; + int dp_clock; + int dp_lane_count; + int rd_interval; + bool tp3_supported; + u8 dpcd[8]; + u8 train_set[4]; + u8 link_status[DP_LINK_STATUS_SIZE]; + u8 tries; + bool use_dpencoder; +}; -static void dp_set_link_bw_lanes(struct radeon_connector *radeon_connector, - u8 link_configuration[DP_LINK_CONFIGURATION_SIZE]) +static void radeon_dp_update_vs_emph(struct radeon_dp_link_train_info *dp_info) { - radeon_dp_aux_native_write(radeon_connector, DP_LINK_BW_SET, 2, - link_configuration); + /* set the initial vs/emph on the source */ + atombios_dig_transmitter_setup(dp_info->encoder, + ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH, + 0, dp_info->train_set[0]); /* sets all lanes at once */ + + /* set the vs/emph on the sink */ + radeon_dp_aux_native_write(dp_info->radeon_connector, DP_TRAINING_LANE0_SET, + dp_info->train_set, dp_info->dp_lane_count, 0); } -static void dp_update_dpvs_emph(struct radeon_connector *radeon_connector, - struct drm_encoder *encoder, - u8 train_set[4]) +static void radeon_dp_set_tp(struct radeon_dp_link_train_info *dp_info, int tp) { - struct radeon_connector_atom_dig *dig_connector = radeon_connector->con_priv; - int i; + int rtp = 0; - for (i = 0; i < dig_connector->dp_lane_count; i++) - atombios_dig_transmitter_setup(encoder, - ATOM_TRANSMITTER_ACTION_SETUP_VSEMPH, - i, train_set[i]); + /* set training pattern on the source */ + if (ASIC_IS_DCE4(dp_info->rdev) || !dp_info->use_dpencoder) { + switch (tp) { + case DP_TRAINING_PATTERN_1: + rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN1; + break; + case DP_TRAINING_PATTERN_2: + rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN2; + break; + case DP_TRAINING_PATTERN_3: + rtp = ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN3; + break; + } + atombios_dig_encoder_setup(dp_info->encoder, rtp, 0); + } else { + switch (tp) { + case DP_TRAINING_PATTERN_1: + rtp = 0; + break; + case DP_TRAINING_PATTERN_2: + rtp = 1; + break; + } + radeon_dp_encoder_service(dp_info->rdev, ATOM_DP_ACTION_TRAINING_PATTERN_SEL, + dp_info->dp_clock, dp_info->enc_id, rtp); + } - radeon_dp_aux_native_write(radeon_connector, DP_TRAINING_LANE0_SET, - dig_connector->dp_lane_count, train_set); + /* enable training pattern on the sink */ + radeon_write_dpcd_reg(dp_info->radeon_connector, DP_TRAINING_PATTERN_SET, tp); } -static void dp_set_training(struct radeon_connector *radeon_connector, - u8 training) +static int radeon_dp_link_train_init(struct radeon_dp_link_train_info *dp_info) { - radeon_dp_aux_native_write(radeon_connector, DP_TRAINING_PATTERN_SET, - 1, &training); -} + u8 tmp; -void dp_link_train(struct drm_encoder *encoder, - struct drm_connector *connector) -{ - struct drm_device *dev = encoder->dev; - struct radeon_device *rdev = dev->dev_private; - struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); - struct radeon_encoder_atom_dig *dig; - struct radeon_connector *radeon_connector; - struct radeon_connector_atom_dig *dig_connector; - int enc_id = 0; - bool clock_recovery, channel_eq; - u8 link_status[DP_LINK_STATUS_SIZE]; - u8 link_configuration[DP_LINK_CONFIGURATION_SIZE]; - u8 tries, voltage; - u8 train_set[4]; - int i; + /* power up the sink */ + if (dp_info->dpcd[0] >= 0x11) + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_SET_POWER, DP_SET_POWER_D0); + + /* possibly enable downspread on the sink */ + if (dp_info->dpcd[3] & 0x1) + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_DOWNSPREAD_CTRL, DP_SPREAD_AMP_0_5); + else + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_DOWNSPREAD_CTRL, 0); - if ((connector->connector_type != DRM_MODE_CONNECTOR_DisplayPort) && - (connector->connector_type != DRM_MODE_CONNECTOR_eDP)) - return; + radeon_dp_set_panel_mode(dp_info->encoder, dp_info->connector); - if (!radeon_encoder->enc_priv) - return; - dig = radeon_encoder->enc_priv; + /* set the lane count on the sink */ + tmp = dp_info->dp_lane_count; + if (dp_info->dpcd[0] >= 0x11) + tmp |= DP_LANE_COUNT_ENHANCED_FRAME_EN; + radeon_write_dpcd_reg(dp_info->radeon_connector, DP_LANE_COUNT_SET, tmp); - radeon_connector = to_radeon_connector(connector); - if (!radeon_connector->con_priv) - return; - dig_connector = radeon_connector->con_priv; + /* set the link rate on the sink */ + tmp = dp_get_dp_link_rate_coded(dp_info->dp_clock); + radeon_write_dpcd_reg(dp_info->radeon_connector, DP_LINK_BW_SET, tmp); - if (dig->dig_encoder) - enc_id |= ATOM_DP_CONFIG_DIG2_ENCODER; + /* start training on the source */ + if (ASIC_IS_DCE4(dp_info->rdev) || !dp_info->use_dpencoder) + atombios_dig_encoder_setup(dp_info->encoder, + ATOM_ENCODER_CMD_DP_LINK_TRAINING_START, 0); else - enc_id |= ATOM_DP_CONFIG_DIG1_ENCODER; - if (dig->linkb) - enc_id |= ATOM_DP_CONFIG_LINK_B; - else - enc_id |= ATOM_DP_CONFIG_LINK_A; - - memset(link_configuration, 0, DP_LINK_CONFIGURATION_SIZE); - if (dig_connector->dp_clock == 270000) - link_configuration[0] = DP_LINK_BW_2_7; - else - link_configuration[0] = DP_LINK_BW_1_62; - link_configuration[1] = dig_connector->dp_lane_count; - if (dig_connector->dpcd[0] >= 0x11) - link_configuration[1] |= DP_LANE_COUNT_ENHANCED_FRAME_EN; + radeon_dp_encoder_service(dp_info->rdev, ATOM_DP_ACTION_TRAINING_START, + dp_info->dp_clock, dp_info->enc_id, 0); - /* power up the sink */ - dp_set_power(radeon_connector, DP_SET_POWER_D0); /* disable the training pattern on the sink */ - dp_set_training(radeon_connector, DP_TRAINING_PATTERN_DISABLE); - /* set link bw and lanes on the sink */ - dp_set_link_bw_lanes(radeon_connector, link_configuration); - /* disable downspread on the sink */ - dp_set_downspread(radeon_connector, 0); - if (ASIC_IS_DCE4(rdev)) { - /* start training on the source */ - atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_LINK_TRAINING_START); - /* set training pattern 1 on the source */ - atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN1); - } else { - /* start training on the source */ - radeon_dp_encoder_service(rdev, ATOM_DP_ACTION_TRAINING_START, - dig_connector->dp_clock, enc_id, 0); - /* set training pattern 1 on the source */ - radeon_dp_encoder_service(rdev, ATOM_DP_ACTION_TRAINING_PATTERN_SEL, - dig_connector->dp_clock, enc_id, 0); - } + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_TRAINING_PATTERN_SET, + DP_TRAINING_PATTERN_DISABLE); - /* set initial vs/emph */ - memset(train_set, 0, 4); + return 0; +} + +static int radeon_dp_link_train_finish(struct radeon_dp_link_train_info *dp_info) +{ udelay(400); - /* set training pattern 1 on the sink */ - dp_set_training(radeon_connector, DP_TRAINING_PATTERN_1); - dp_update_dpvs_emph(radeon_connector, encoder, train_set); + /* disable the training pattern on the sink */ + radeon_write_dpcd_reg(dp_info->radeon_connector, + DP_TRAINING_PATTERN_SET, + DP_TRAINING_PATTERN_DISABLE); + + /* disable the training pattern on the source */ + if (ASIC_IS_DCE4(dp_info->rdev) || !dp_info->use_dpencoder) + atombios_dig_encoder_setup(dp_info->encoder, + ATOM_ENCODER_CMD_DP_LINK_TRAINING_COMPLETE, 0); + else + radeon_dp_encoder_service(dp_info->rdev, ATOM_DP_ACTION_TRAINING_COMPLETE, + dp_info->dp_clock, dp_info->enc_id, 0); + + return 0; +} + +static int radeon_dp_link_train_cr(struct radeon_dp_link_train_info *dp_info) +{ + bool clock_recovery; + u8 voltage; + int i; + + radeon_dp_set_tp(dp_info, DP_TRAINING_PATTERN_1); + memset(dp_info->train_set, 0, 4); + radeon_dp_update_vs_emph(dp_info); + + udelay(400); /* clock recovery loop */ clock_recovery = false; - tries = 0; + dp_info->tries = 0; voltage = 0xff; - for (;;) { - udelay(100); - if (!atom_dp_get_link_status(radeon_connector, link_status)) + while (1) { + if (dp_info->rd_interval == 0) + udelay(100); + else + mdelay(dp_info->rd_interval * 4); + + if (!radeon_dp_get_link_status(dp_info->radeon_connector, dp_info->link_status)) break; - if (dp_clock_recovery_ok(link_status, dig_connector->dp_lane_count)) { + if (dp_clock_recovery_ok(dp_info->link_status, dp_info->dp_lane_count)) { clock_recovery = true; break; } - for (i = 0; i < dig_connector->dp_lane_count; i++) { - if ((train_set[i] & DP_TRAIN_MAX_SWING_REACHED) == 0) + for (i = 0; i < dp_info->dp_lane_count; i++) { + if ((dp_info->train_set[i] & DP_TRAIN_MAX_SWING_REACHED) == 0) break; } - if (i == dig_connector->dp_lane_count) { + if (i == dp_info->dp_lane_count) { DRM_ERROR("clock recovery reached max voltage\n"); break; } - if ((train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK) == voltage) { - ++tries; - if (tries == 5) { + if ((dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK) == voltage) { + ++dp_info->tries; + if (dp_info->tries == 5) { DRM_ERROR("clock recovery tried 5 times\n"); break; } } else - tries = 0; + dp_info->tries = 0; - voltage = train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK; + voltage = dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK; /* Compute new train_set as requested by sink */ - dp_get_adjust_train(link_status, dig_connector->dp_lane_count, train_set); - dp_update_dpvs_emph(radeon_connector, encoder, train_set); + dp_get_adjust_train(dp_info->link_status, dp_info->dp_lane_count, dp_info->train_set); + + radeon_dp_update_vs_emph(dp_info); } - if (!clock_recovery) + if (!clock_recovery) { DRM_ERROR("clock recovery failed\n"); - else + return -1; + } else { DRM_DEBUG_KMS("clock recovery at voltage %d pre-emphasis %d\n", - train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK, - (train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK) >> + dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK, + (dp_info->train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK) >> DP_TRAIN_PRE_EMPHASIS_SHIFT); + return 0; + } +} +static int radeon_dp_link_train_ce(struct radeon_dp_link_train_info *dp_info) +{ + bool channel_eq; - /* set training pattern 2 on the sink */ - dp_set_training(radeon_connector, DP_TRAINING_PATTERN_2); - /* set training pattern 2 on the source */ - if (ASIC_IS_DCE4(rdev)) - atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_LINK_TRAINING_PATTERN2); + if (dp_info->tp3_supported) + radeon_dp_set_tp(dp_info, DP_TRAINING_PATTERN_3); else - radeon_dp_encoder_service(rdev, ATOM_DP_ACTION_TRAINING_PATTERN_SEL, - dig_connector->dp_clock, enc_id, 1); + radeon_dp_set_tp(dp_info, DP_TRAINING_PATTERN_2); /* channel equalization loop */ - tries = 0; + dp_info->tries = 0; channel_eq = false; - for (;;) { - udelay(400); - if (!atom_dp_get_link_status(radeon_connector, link_status)) + while (1) { + if (dp_info->rd_interval == 0) + udelay(400); + else + mdelay(dp_info->rd_interval * 4); + + if (!radeon_dp_get_link_status(dp_info->radeon_connector, dp_info->link_status)) break; - if (dp_channel_eq_ok(link_status, dig_connector->dp_lane_count)) { + if (dp_channel_eq_ok(dp_info->link_status, dp_info->dp_lane_count)) { channel_eq = true; break; } /* Try 5 times */ - if (tries > 5) { + if (dp_info->tries > 5) { DRM_ERROR("channel eq failed: 5 tries\n"); break; } /* Compute new train_set as requested by sink */ - dp_get_adjust_train(link_status, dig_connector->dp_lane_count, train_set); - dp_update_dpvs_emph(radeon_connector, encoder, train_set); + dp_get_adjust_train(dp_info->link_status, dp_info->dp_lane_count, dp_info->train_set); - tries++; + radeon_dp_update_vs_emph(dp_info); + dp_info->tries++; } - if (!channel_eq) + if (!channel_eq) { DRM_ERROR("channel eq failed\n"); - else + return -1; + } else { DRM_DEBUG_KMS("channel eq at voltage %d pre-emphasis %d\n", - train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK, - (train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK) + dp_info->train_set[0] & DP_TRAIN_VOLTAGE_SWING_MASK, + (dp_info->train_set[0] & DP_TRAIN_PRE_EMPHASIS_MASK) >> DP_TRAIN_PRE_EMPHASIS_SHIFT); - - /* disable the training pattern on the sink */ - dp_set_training(radeon_connector, DP_TRAINING_PATTERN_DISABLE); - - /* disable the training pattern on the source */ - if (ASIC_IS_DCE4(rdev)) - atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_LINK_TRAINING_COMPLETE); - else - radeon_dp_encoder_service(rdev, ATOM_DP_ACTION_TRAINING_COMPLETE, - dig_connector->dp_clock, enc_id, 0); + return 0; + } } -int radeon_dp_i2c_aux_ch(struct i2c_adapter *adapter, int mode, - uint8_t write_byte, uint8_t *read_byte) +void radeon_dp_link_train(struct drm_encoder *encoder, + struct drm_connector *connector) { - struct i2c_algo_dp_aux_data *algo_data = adapter->algo_data; - struct radeon_i2c_chan *auxch = (struct radeon_i2c_chan *)adapter; - int ret = 0; - uint16_t address = algo_data->address; - uint8_t msg[5]; - uint8_t reply[2]; - int msg_len, dp_msg_len; - int reply_bytes; - - /* Set up the command byte */ - if (mode & MODE_I2C_READ) - msg[2] = AUX_I2C_READ << 4; - else - msg[2] = AUX_I2C_WRITE << 4; + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_encoder_atom_dig *dig; + struct radeon_connector *radeon_connector; + struct radeon_connector_atom_dig *dig_connector; + struct radeon_dp_link_train_info dp_info; + int index; + u8 tmp, frev, crev; - if (!(mode & MODE_I2C_STOP)) - msg[2] |= AUX_I2C_MOT << 4; + if (!radeon_encoder->enc_priv) + return; + dig = radeon_encoder->enc_priv; - msg[0] = address; - msg[1] = address >> 8; + radeon_connector = to_radeon_connector(connector); + if (!radeon_connector->con_priv) + return; + dig_connector = radeon_connector->con_priv; - reply_bytes = 1; + if ((dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_DISPLAYPORT) && + (dig_connector->dp_sink_type != CONNECTOR_OBJECT_ID_eDP)) + return; - msg_len = 4; - dp_msg_len = 3; - switch (mode) { - case MODE_I2C_WRITE: - msg[4] = write_byte; - msg_len++; - dp_msg_len += 2; - break; - case MODE_I2C_READ: - dp_msg_len += 1; - break; - default: - break; + /* DPEncoderService newer than 1.1 can't program properly the + * training pattern. When facing such version use the + * DIGXEncoderControl (X== 1 | 2) + */ + dp_info.use_dpencoder = true; + index = GetIndexIntoMasterTable(COMMAND, DPEncoderService); + if (atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) { + if (crev > 1) { + dp_info.use_dpencoder = false; + } } - msg[3] = (dp_msg_len) << 4; - ret = radeon_process_aux_ch(auxch, msg, msg_len, reply, reply_bytes, 0); + dp_info.enc_id = 0; + if (dig->dig_encoder) + dp_info.enc_id |= ATOM_DP_CONFIG_DIG2_ENCODER; + else + dp_info.enc_id |= ATOM_DP_CONFIG_DIG1_ENCODER; + if (dig->linkb) + dp_info.enc_id |= ATOM_DP_CONFIG_LINK_B; + else + dp_info.enc_id |= ATOM_DP_CONFIG_LINK_A; - if (ret) { - if (read_byte) - *read_byte = reply[0]; - return reply_bytes; - } - return -EREMOTEIO; + dp_info.rd_interval = radeon_read_dpcd_reg(radeon_connector, DP_TRAINING_AUX_RD_INTERVAL); + tmp = radeon_read_dpcd_reg(radeon_connector, DP_MAX_LANE_COUNT); + if (ASIC_IS_DCE5(rdev) && (tmp & DP_TPS3_SUPPORTED)) + dp_info.tp3_supported = true; + else + dp_info.tp3_supported = false; + + memcpy(dp_info.dpcd, dig_connector->dpcd, 8); + dp_info.rdev = rdev; + dp_info.encoder = encoder; + dp_info.connector = connector; + dp_info.radeon_connector = radeon_connector; + dp_info.dp_lane_count = dig_connector->dp_lane_count; + dp_info.dp_clock = dig_connector->dp_clock; + + if (radeon_dp_link_train_init(&dp_info)) + goto done; + if (radeon_dp_link_train_cr(&dp_info)) + goto done; + if (radeon_dp_link_train_ce(&dp_info)) + goto done; +done: + if (radeon_dp_link_train_finish(&dp_info)) + return; } - diff --git a/drivers/gpu/drm/radeon/cayman_blit_shaders.c b/drivers/gpu/drm/radeon/cayman_blit_shaders.c index e148ab04b80..7b4eeb7b4a8 100644 --- a/drivers/gpu/drm/radeon/cayman_blit_shaders.c +++ b/drivers/gpu/drm/radeon/cayman_blit_shaders.c @@ -39,17 +39,335 @@ const u32 cayman_default_state[] = { - /* XXX fill in additional blit state */ + 0xc0066900, + 0x00000000, + 0x00000060, /* DB_RENDER_CONTROL */ + 0x00000000, /* DB_COUNT_CONTROL */ + 0x00000000, /* DB_DEPTH_VIEW */ + 0x0000002a, /* DB_RENDER_OVERRIDE */ + 0x00000000, /* DB_RENDER_OVERRIDE2 */ + 0x00000000, /* DB_HTILE_DATA_BASE */ 0xc0026900, - 0x00000316, - 0x0000000e, /* VGT_VERTEX_REUSE_BLOCK_CNTL */ - 0x00000010, /* */ + 0x0000000a, + 0x00000000, /* DB_STENCIL_CLEAR */ + 0x00000000, /* DB_DEPTH_CLEAR */ + + 0xc0036900, + 0x0000000f, + 0x00000000, /* DB_DEPTH_INFO */ + 0x00000000, /* DB_Z_INFO */ + 0x00000000, /* DB_STENCIL_INFO */ + + 0xc0016900, + 0x00000080, + 0x00000000, /* PA_SC_WINDOW_OFFSET */ + + 0xc00d6900, + 0x00000083, + 0x0000ffff, /* PA_SC_CLIPRECT_RULE */ + 0x00000000, /* PA_SC_CLIPRECT_0_TL */ + 0x20002000, /* PA_SC_CLIPRECT_0_BR */ + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0x00000000, + 0x20002000, + 0xaaaaaaaa, /* PA_SC_EDGERULE */ + 0x00000000, /* PA_SU_HARDWARE_SCREEN_OFFSET */ + 0x0000000f, /* CB_TARGET_MASK */ + 0x0000000f, /* CB_SHADER_MASK */ + + 0xc0226900, + 0x00000094, + 0x80000000, /* PA_SC_VPORT_SCISSOR_0_TL */ + 0x20002000, /* PA_SC_VPORT_SCISSOR_0_BR */ + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x80000000, + 0x20002000, + 0x00000000, /* PA_SC_VPORT_ZMIN_0 */ + 0x3f800000, /* PA_SC_VPORT_ZMAX_0 */ + + 0xc0016900, + 0x000000d4, + 0x00000000, /* SX_MISC */ 0xc0026900, 0x000000d9, 0x00000000, /* CP_RINGID */ 0x00000000, /* CP_VMID */ + + 0xc0096900, + 0x00000100, + 0x00ffffff, /* VGT_MAX_VTX_INDX */ + 0x00000000, /* VGT_MIN_VTX_INDX */ + 0x00000000, /* VGT_INDX_OFFSET */ + 0x00000000, /* VGT_MULTI_PRIM_IB_RESET_INDX */ + 0x00000000, /* SX_ALPHA_TEST_CONTROL */ + 0x00000000, /* CB_BLEND_RED */ + 0x00000000, /* CB_BLEND_GREEN */ + 0x00000000, /* CB_BLEND_BLUE */ + 0x00000000, /* CB_BLEND_ALPHA */ + + 0xc0016900, + 0x00000187, + 0x00000100, /* SPI_VS_OUT_ID_0 */ + + 0xc0026900, + 0x00000191, + 0x00000100, /* SPI_PS_INPUT_CNTL_0 */ + 0x00000101, /* SPI_PS_INPUT_CNTL_1 */ + + 0xc0016900, + 0x000001b1, + 0x00000000, /* SPI_VS_OUT_CONFIG */ + + 0xc0106900, + 0x000001b3, + 0x20000001, /* SPI_PS_IN_CONTROL_0 */ + 0x00000000, /* SPI_PS_IN_CONTROL_1 */ + 0x00000000, /* SPI_INTERP_CONTROL_0 */ + 0x00000000, /* SPI_INPUT_Z */ + 0x00000000, /* SPI_FOG_CNTL */ + 0x00100000, /* SPI_BARYC_CNTL */ + 0x00000000, /* SPI_PS_IN_CONTROL_2 */ + 0x00000000, /* SPI_COMPUTE_INPUT_CNTL */ + 0x00000000, /* SPI_COMPUTE_NUM_THREAD_X */ + 0x00000000, /* SPI_COMPUTE_NUM_THREAD_Y */ + 0x00000000, /* SPI_COMPUTE_NUM_THREAD_Z */ + 0x00000000, /* SPI_GPR_MGMT */ + 0x00000000, /* SPI_LDS_MGMT */ + 0x00000000, /* SPI_STACK_MGMT */ + 0x00000000, /* SPI_WAVE_MGMT_1 */ + 0x00000000, /* SPI_WAVE_MGMT_2 */ + + 0xc0016900, + 0x000001e0, + 0x00000000, /* CB_BLEND0_CONTROL */ + + 0xc00e6900, + 0x00000200, + 0x00000000, /* DB_DEPTH_CONTROL */ + 0x00000000, /* DB_EQAA */ + 0x00cc0010, /* CB_COLOR_CONTROL */ + 0x00000210, /* DB_SHADER_CONTROL */ + 0x00010000, /* PA_CL_CLIP_CNTL */ + 0x00000004, /* PA_SU_SC_MODE_CNTL */ + 0x00000100, /* PA_CL_VTE_CNTL */ + 0x00000000, /* PA_CL_VS_OUT_CNTL */ + 0x00000000, /* PA_CL_NANINF_CNTL */ + 0x00000000, /* PA_SU_LINE_STIPPLE_CNTL */ + 0x00000000, /* PA_SU_LINE_STIPPLE_SCALE */ + 0x00000000, /* PA_SU_PRIM_FILTER_CNTL */ + 0x00000000, /* */ + 0x00000000, /* */ + + 0xc0026900, + 0x00000229, + 0x00000000, /* SQ_PGM_START_FS */ + 0x00000000, + + 0xc0016900, + 0x0000023b, + 0x00000000, /* SQ_LDS_ALLOC_PS */ + + 0xc0066900, + 0x00000240, + 0x00000000, /* SQ_ESGS_RING_ITEMSIZE */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0046900, + 0x00000247, + 0x00000000, /* SQ_GS_VERT_ITEMSIZE */ + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0116900, + 0x00000280, + 0x00000000, /* PA_SU_POINT_SIZE */ + 0x00000000, /* PA_SU_POINT_MINMAX */ + 0x00000008, /* PA_SU_LINE_CNTL */ + 0x00000000, /* PA_SC_LINE_STIPPLE */ + 0x00000000, /* VGT_OUTPUT_PATH_CNTL */ + 0x00000000, /* VGT_HOS_CNTL */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, /* VGT_GS_MODE */ + + 0xc0026900, + 0x00000292, + 0x00000000, /* PA_SC_MODE_CNTL_0 */ + 0x00000000, /* PA_SC_MODE_CNTL_1 */ + + 0xc0016900, + 0x000002a1, + 0x00000000, /* VGT_PRIMITIVEID_EN */ + + 0xc0016900, + 0x000002a5, + 0x00000000, /* VGT_MULTI_PRIM_IB_RESET_EN */ + + 0xc0026900, + 0x000002a8, + 0x00000000, /* VGT_INSTANCE_STEP_RATE_0 */ + 0x00000000, + + 0xc0026900, + 0x000002ad, + 0x00000000, /* VGT_REUSE_OFF */ + 0x00000000, + + 0xc0016900, + 0x000002d5, + 0x00000000, /* VGT_SHADER_STAGES_EN */ + + 0xc0016900, + 0x000002dc, + 0x0000aa00, /* DB_ALPHA_TO_MASK */ + + 0xc0066900, + 0x000002de, + 0x00000000, /* PA_SU_POLY_OFFSET_DB_FMT_CNTL */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + + 0xc0026900, + 0x000002e5, + 0x00000000, /* VGT_STRMOUT_CONFIG */ + 0x00000000, + + 0xc01b6900, + 0x000002f5, + 0x76543210, /* PA_SC_CENTROID_PRIORITY_0 */ + 0xfedcba98, /* PA_SC_CENTROID_PRIORITY_1 */ + 0x00000000, /* PA_SC_LINE_CNTL */ + 0x00000000, /* PA_SC_AA_CONFIG */ + 0x00000005, /* PA_SU_VTX_CNTL */ + 0x3f800000, /* PA_CL_GB_VERT_CLIP_ADJ */ + 0x3f800000, /* PA_CL_GB_VERT_DISC_ADJ */ + 0x3f800000, /* PA_CL_GB_HORZ_CLIP_ADJ */ + 0x3f800000, /* PA_CL_GB_HORZ_DISC_ADJ */ + 0x00000000, /* PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0 */ + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0x00000000, + 0xffffffff, /* PA_SC_AA_MASK_X0Y0_X1Y0 */ + 0xffffffff, + + 0xc0026900, + 0x00000316, + 0x0000000e, /* VGT_VERTEX_REUSE_BLOCK_CNTL */ + 0x00000010, /* */ +}; + +const u32 cayman_vs[] = +{ + 0x00000004, + 0x80400400, + 0x0000a03c, + 0x95000688, + 0x00004000, + 0x15000688, + 0x00000000, + 0x88000000, + 0x04000000, + 0x67961001, +#ifdef __BIG_ENDIAN + 0x00020000, +#else + 0x00000000, +#endif + 0x00000000, + 0x04000000, + 0x67961000, +#ifdef __BIG_ENDIAN + 0x00020008, +#else + 0x00000008, +#endif + 0x00000000, +}; + +const u32 cayman_ps[] = +{ + 0x00000004, + 0xa00c0000, + 0x00000008, + 0x80400000, + 0x00000000, + 0x95000688, + 0x00000000, + 0x88000000, + 0x00380400, + 0x00146b10, + 0x00380000, + 0x20146b10, + 0x00380400, + 0x40146b00, + 0x80380000, + 0x60146b00, + 0x00000010, + 0x000d1000, + 0xb0800000, + 0x00000000, }; +const u32 cayman_ps_size = ARRAY_SIZE(cayman_ps); +const u32 cayman_vs_size = ARRAY_SIZE(cayman_vs); const u32 cayman_default_size = ARRAY_SIZE(cayman_default_state); diff --git a/drivers/gpu/drm/radeon/cayman_blit_shaders.h b/drivers/gpu/drm/radeon/cayman_blit_shaders.h index 33b75e5d0fa..f5d0e9a6026 100644 --- a/drivers/gpu/drm/radeon/cayman_blit_shaders.h +++ b/drivers/gpu/drm/radeon/cayman_blit_shaders.h @@ -25,8 +25,11 @@ #ifndef CAYMAN_BLIT_SHADERS_H #define CAYMAN_BLIT_SHADERS_H +extern const u32 cayman_ps[]; +extern const u32 cayman_vs[]; extern const u32 cayman_default_state[]; +extern const u32 cayman_ps_size, cayman_vs_size; extern const u32 cayman_default_size; #endif diff --git a/drivers/gpu/drm/radeon/evergreen.c b/drivers/gpu/drm/radeon/evergreen.c index c20eac3379e..fb5fa089886 100644 --- a/drivers/gpu/drm/radeon/evergreen.c +++ b/drivers/gpu/drm/radeon/evergreen.c @@ -88,21 +88,40 @@ u32 evergreen_page_flip(struct radeon_device *rdev, int crtc_id, u64 crtc_base) /* get temperature in millidegrees */ int evergreen_get_temp(struct radeon_device *rdev) { - u32 temp = (RREG32(CG_MULT_THERMAL_STATUS) & ASIC_T_MASK) >> - ASIC_T_SHIFT; - u32 actual_temp = 0; - - if (temp & 0x400) - actual_temp = -256; - else if (temp & 0x200) - actual_temp = 255; - else if (temp & 0x100) { - actual_temp = temp & 0x1ff; - actual_temp |= ~0x1ff; - } else - actual_temp = temp & 0xff; + u32 temp, toffset; + int actual_temp = 0; + + if (rdev->family == CHIP_JUNIPER) { + toffset = (RREG32(CG_THERMAL_CTRL) & TOFFSET_MASK) >> + TOFFSET_SHIFT; + temp = (RREG32(CG_TS0_STATUS) & TS0_ADC_DOUT_MASK) >> + TS0_ADC_DOUT_SHIFT; + + if (toffset & 0x100) + actual_temp = temp / 2 - (0x200 - toffset); + else + actual_temp = temp / 2 + toffset; + + actual_temp = actual_temp * 1000; + + } else { + temp = (RREG32(CG_MULT_THERMAL_STATUS) & ASIC_T_MASK) >> + ASIC_T_SHIFT; - return (actual_temp * 1000) / 2; + if (temp & 0x400) + actual_temp = -256; + else if (temp & 0x200) + actual_temp = 255; + else if (temp & 0x100) { + actual_temp = temp & 0x1ff; + actual_temp |= ~0x1ff; + } else + actual_temp = temp & 0xff; + + actual_temp = (actual_temp * 1000) / 2; + } + + return actual_temp; } int sumo_get_temp(struct radeon_device *rdev) @@ -121,11 +140,17 @@ void evergreen_pm_misc(struct radeon_device *rdev) struct radeon_voltage *voltage = &ps->clock_info[req_cm_idx].voltage; if (voltage->type == VOLTAGE_SW) { + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage->voltage == 0xff01) + return; if (voltage->voltage && (voltage->voltage != rdev->pm.current_vddc)) { radeon_atom_set_voltage(rdev, voltage->voltage, SET_VOLTAGE_TYPE_ASIC_VDDC); rdev->pm.current_vddc = voltage->voltage; DRM_DEBUG("Setting: vddc: %d\n", voltage->voltage); } + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage->vddci == 0xff01) + return; if (voltage->vddci && (voltage->vddci != rdev->pm.current_vddci)) { radeon_atom_set_voltage(rdev, voltage->vddci, SET_VOLTAGE_TYPE_ASIC_VDDCI); rdev->pm.current_vddci = voltage->vddci; @@ -718,7 +743,7 @@ static void evergreen_program_watermarks(struct radeon_device *rdev, !evergreen_average_bandwidth_vs_available_bandwidth(&wm) || !evergreen_check_latency_hiding(&wm) || (rdev->disp_priority == 2)) { - DRM_INFO("force priority to high\n"); + DRM_DEBUG_KMS("force priority to high\n"); priority_a_cnt |= PRIORITY_ALWAYS_ON; priority_b_cnt |= PRIORITY_ALWAYS_ON; } @@ -960,17 +985,19 @@ void evergreen_mc_stop(struct radeon_device *rdev, struct evergreen_mc_save *sav { save->vga_control[0] = RREG32(D1VGA_CONTROL); save->vga_control[1] = RREG32(D2VGA_CONTROL); - save->vga_control[2] = RREG32(EVERGREEN_D3VGA_CONTROL); - save->vga_control[3] = RREG32(EVERGREEN_D4VGA_CONTROL); - save->vga_control[4] = RREG32(EVERGREEN_D5VGA_CONTROL); - save->vga_control[5] = RREG32(EVERGREEN_D6VGA_CONTROL); save->vga_render_control = RREG32(VGA_RENDER_CONTROL); save->vga_hdp_control = RREG32(VGA_HDP_CONTROL); save->crtc_control[0] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET); save->crtc_control[1] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { + save->vga_control[2] = RREG32(EVERGREEN_D3VGA_CONTROL); + save->vga_control[3] = RREG32(EVERGREEN_D4VGA_CONTROL); save->crtc_control[2] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET); save->crtc_control[3] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET); + } + if (rdev->num_crtc >= 6) { + save->vga_control[4] = RREG32(EVERGREEN_D5VGA_CONTROL); + save->vga_control[5] = RREG32(EVERGREEN_D6VGA_CONTROL); save->crtc_control[4] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET); save->crtc_control[5] = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET); } @@ -979,35 +1006,45 @@ void evergreen_mc_stop(struct radeon_device *rdev, struct evergreen_mc_save *sav WREG32(VGA_RENDER_CONTROL, 0); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC0_REGISTER_OFFSET, 1); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC1_REGISTER_OFFSET, 1); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC2_REGISTER_OFFSET, 1); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC3_REGISTER_OFFSET, 1); + } + if (rdev->num_crtc >= 6) { WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC4_REGISTER_OFFSET, 1); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC5_REGISTER_OFFSET, 1); } WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); } WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); } WREG32(D1VGA_CONTROL, 0); WREG32(D2VGA_CONTROL, 0); - WREG32(EVERGREEN_D3VGA_CONTROL, 0); - WREG32(EVERGREEN_D4VGA_CONTROL, 0); - WREG32(EVERGREEN_D5VGA_CONTROL, 0); - WREG32(EVERGREEN_D6VGA_CONTROL, 0); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_D3VGA_CONTROL, 0); + WREG32(EVERGREEN_D4VGA_CONTROL, 0); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_D5VGA_CONTROL, 0); + WREG32(EVERGREEN_D6VGA_CONTROL, 0); + } } void evergreen_mc_resume(struct radeon_device *rdev, struct evergreen_mc_save *save) @@ -1030,7 +1067,7 @@ void evergreen_mc_resume(struct radeon_device *rdev, struct evergreen_mc_save *s WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + EVERGREEN_CRTC1_REGISTER_OFFSET, (u32)rdev->mc.vram_start); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC2_REGISTER_OFFSET, upper_32_bits(rdev->mc.vram_start)); WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC2_REGISTER_OFFSET, @@ -1048,7 +1085,8 @@ void evergreen_mc_resume(struct radeon_device *rdev, struct evergreen_mc_save *s (u32)rdev->mc.vram_start); WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS + EVERGREEN_CRTC3_REGISTER_OFFSET, (u32)rdev->mc.vram_start); - + } + if (rdev->num_crtc >= 6) { WREG32(EVERGREEN_GRPH_PRIMARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC4_REGISTER_OFFSET, upper_32_bits(rdev->mc.vram_start)); WREG32(EVERGREEN_GRPH_SECONDARY_SURFACE_ADDRESS_HIGH + EVERGREEN_CRTC4_REGISTER_OFFSET, @@ -1076,31 +1114,41 @@ void evergreen_mc_resume(struct radeon_device *rdev, struct evergreen_mc_save *s /* Restore video state */ WREG32(D1VGA_CONTROL, save->vga_control[0]); WREG32(D2VGA_CONTROL, save->vga_control[1]); - WREG32(EVERGREEN_D3VGA_CONTROL, save->vga_control[2]); - WREG32(EVERGREEN_D4VGA_CONTROL, save->vga_control[3]); - WREG32(EVERGREEN_D5VGA_CONTROL, save->vga_control[4]); - WREG32(EVERGREEN_D6VGA_CONTROL, save->vga_control[5]); + if (rdev->num_crtc >= 4) { + WREG32(EVERGREEN_D3VGA_CONTROL, save->vga_control[2]); + WREG32(EVERGREEN_D4VGA_CONTROL, save->vga_control[3]); + } + if (rdev->num_crtc >= 6) { + WREG32(EVERGREEN_D5VGA_CONTROL, save->vga_control[4]); + WREG32(EVERGREEN_D6VGA_CONTROL, save->vga_control[5]); + } WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC0_REGISTER_OFFSET, 1); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC1_REGISTER_OFFSET, 1); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC2_REGISTER_OFFSET, 1); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC3_REGISTER_OFFSET, 1); + } + if (rdev->num_crtc >= 6) { WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC4_REGISTER_OFFSET, 1); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC5_REGISTER_OFFSET, 1); } WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET, save->crtc_control[0]); WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET, save->crtc_control[1]); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, save->crtc_control[2]); WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, save->crtc_control[3]); + } + if (rdev->num_crtc >= 6) { WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, save->crtc_control[4]); WREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, save->crtc_control[5]); } WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); WREG32(EVERGREEN_CRTC_UPDATE_LOCK + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); } @@ -1334,9 +1382,6 @@ int evergreen_cp_resume(struct radeon_device *rdev) /* set the wb address wether it's enabled or not */ WREG32(CP_RB_RPTR_ADDR, -#ifdef __BIG_ENDIAN - RB_RPTR_SWAP(2) | -#endif ((rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFFFFFFFC)); WREG32(CP_RB_RPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFF); WREG32(SCRATCH_ADDR, ((rdev->wb.gpu_addr + RADEON_WB_SCRATCH_OFFSET) >> 8) & 0xFFFFFFFF); @@ -1415,6 +1460,8 @@ static u32 evergreen_get_tile_pipe_to_backend_map(struct radeon_device *rdev, case CHIP_CEDAR: case CHIP_REDWOOD: case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: case CHIP_TURKS: case CHIP_CAICOS: force_no_swizzle = false; @@ -1544,6 +1591,8 @@ static void evergreen_program_channel_remap(struct radeon_device *rdev) case CHIP_REDWOOD: case CHIP_CEDAR: case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: case CHIP_TURKS: case CHIP_CAICOS: default: @@ -1578,7 +1627,7 @@ static void evergreen_gpu_init(struct radeon_device *rdev) u32 sq_stack_resource_mgmt_2; u32 sq_stack_resource_mgmt_3; u32 vgt_cache_invalidation; - u32 hdp_host_path_cntl; + u32 hdp_host_path_cntl, tmp; int i, j, num_shader_engines, ps_thread_count; switch (rdev->family) { @@ -1689,6 +1738,54 @@ static void evergreen_gpu_init(struct radeon_device *rdev) rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; break; + case CHIP_SUMO: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 2; + if (rdev->pdev->device == 0x9648) + rdev->config.evergreen.max_simds = 3; + else if ((rdev->pdev->device == 0x9647) || + (rdev->pdev->device == 0x964a)) + rdev->config.evergreen.max_simds = 4; + else + rdev->config.evergreen.max_simds = 5; + rdev->config.evergreen.max_backends = 2 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 248; + rdev->config.evergreen.max_gs_threads = 32; + rdev->config.evergreen.max_stack_entries = 256; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 256; + rdev->config.evergreen.sx_max_export_pos_size = 64; + rdev->config.evergreen.sx_max_export_smx_size = 192; + rdev->config.evergreen.max_hw_contexts = 8; + rdev->config.evergreen.sq_num_cf_insts = 2; + + rdev->config.evergreen.sc_prim_fifo_size = 0x40; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; + case CHIP_SUMO2: + rdev->config.evergreen.num_ses = 1; + rdev->config.evergreen.max_pipes = 4; + rdev->config.evergreen.max_tile_pipes = 4; + rdev->config.evergreen.max_simds = 2; + rdev->config.evergreen.max_backends = 1 * rdev->config.evergreen.num_ses; + rdev->config.evergreen.max_gprs = 256; + rdev->config.evergreen.max_threads = 248; + rdev->config.evergreen.max_gs_threads = 32; + rdev->config.evergreen.max_stack_entries = 512; + rdev->config.evergreen.sx_num_of_sets = 4; + rdev->config.evergreen.sx_max_export_size = 256; + rdev->config.evergreen.sx_max_export_pos_size = 64; + rdev->config.evergreen.sx_max_export_smx_size = 192; + rdev->config.evergreen.max_hw_contexts = 8; + rdev->config.evergreen.sq_num_cf_insts = 2; + + rdev->config.evergreen.sc_prim_fifo_size = 0x40; + rdev->config.evergreen.sc_hiz_tile_fifo_size = 0x30; + rdev->config.evergreen.sc_earlyz_tile_fifo_size = 0x130; + break; case CHIP_BARTS: rdev->config.evergreen.num_ses = 2; rdev->config.evergreen.max_pipes = 4; @@ -1780,7 +1877,10 @@ static void evergreen_gpu_init(struct radeon_device *rdev) mc_shared_chmap = RREG32(MC_SHARED_CHMAP); - mc_arb_ramcfg = RREG32(MC_ARB_RAMCFG); + if (rdev->flags & RADEON_IS_IGP) + mc_arb_ramcfg = RREG32(FUS_MC_ARB_RAMCFG); + else + mc_arb_ramcfg = RREG32(MC_ARB_RAMCFG); switch (rdev->config.evergreen.max_tile_pipes) { case 1: @@ -1897,7 +1997,7 @@ static void evergreen_gpu_init(struct radeon_device *rdev) gb_backend_map = 0x66442200; break; case CHIP_JUNIPER: - gb_backend_map = 0x00006420; + gb_backend_map = 0x00002200; break; default: gb_backend_map = @@ -1933,13 +2033,18 @@ static void evergreen_gpu_init(struct radeon_device *rdev) rdev->config.evergreen.tile_config |= (3 << 0); break; } - rdev->config.evergreen.tile_config |= - ((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) << 4; + /* num banks is 8 on all fusion asics. 0 = 4, 1 = 8, 2 = 16 */ + if (rdev->flags & RADEON_IS_IGP) + rdev->config.evergreen.tile_config |= 1 << 4; + else + rdev->config.evergreen.tile_config |= + ((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) << 4; rdev->config.evergreen.tile_config |= ((mc_arb_ramcfg & BURSTLENGTH_MASK) >> BURSTLENGTH_SHIFT) << 8; rdev->config.evergreen.tile_config |= ((gb_addr_config & 0x30000000) >> 28) << 12; + rdev->config.evergreen.backend_map = gb_backend_map; WREG32(GB_BACKEND_MAP, gb_backend_map); WREG32(GB_ADDR_CONFIG, gb_addr_config); WREG32(DMIF_ADDR_CONFIG, gb_addr_config); @@ -2032,6 +2137,8 @@ static void evergreen_gpu_init(struct radeon_device *rdev) switch (rdev->family) { case CHIP_CEDAR: case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: case CHIP_CAICOS: /* no vertex cache */ sq_config &= ~VC_ENABLE; @@ -2053,6 +2160,8 @@ static void evergreen_gpu_init(struct radeon_device *rdev) switch (rdev->family) { case CHIP_CEDAR: case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: ps_thread_count = 96; break; default: @@ -2092,6 +2201,8 @@ static void evergreen_gpu_init(struct radeon_device *rdev) switch (rdev->family) { case CHIP_CEDAR: case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: case CHIP_CAICOS: vgt_cache_invalidation = CACHE_INVALIDATION(TC_ONLY); break; @@ -2138,6 +2249,10 @@ static void evergreen_gpu_init(struct radeon_device *rdev) for (i = SQ_ALU_CONST_BUFFER_SIZE_HS_0; i < 0x29000; i += 4) WREG32(i, 0); + tmp = RREG32(HDP_MISC_CNTL); + tmp |= HDP_FLUSH_INVALIDATE_CACHE; + WREG32(HDP_MISC_CNTL, tmp); + hdp_host_path_cntl = RREG32(HDP_HOST_PATH_CNTL); WREG32(HDP_HOST_PATH_CNTL, hdp_host_path_cntl); @@ -2154,7 +2269,10 @@ int evergreen_mc_init(struct radeon_device *rdev) /* Get VRAM informations */ rdev->mc.vram_is_ddr = true; - tmp = RREG32(MC_ARB_RAMCFG); + if (rdev->flags & RADEON_IS_IGP) + tmp = RREG32(FUS_MC_ARB_RAMCFG); + else + tmp = RREG32(MC_ARB_RAMCFG); if (tmp & CHANSIZE_OVERRIDE) { chansize = 16; } else if (tmp & CHANSIZE_MASK) { @@ -2320,18 +2438,22 @@ void evergreen_disable_interrupt_state(struct radeon_device *rdev) WREG32(GRBM_INT_CNTL, 0); WREG32(INT_MASK + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); WREG32(INT_MASK + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(INT_MASK + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); WREG32(INT_MASK + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { WREG32(INT_MASK + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); WREG32(INT_MASK + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); } WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET, 0); WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET, 0); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, 0); WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, 0); + } + if (rdev->num_crtc >= 6) { WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, 0); WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, 0); } @@ -2450,19 +2572,25 @@ int evergreen_irq_set(struct radeon_device *rdev) WREG32(INT_MASK + EVERGREEN_CRTC0_REGISTER_OFFSET, crtc1); WREG32(INT_MASK + EVERGREEN_CRTC1_REGISTER_OFFSET, crtc2); - if (!(rdev->flags & RADEON_IS_IGP)) { + if (rdev->num_crtc >= 4) { WREG32(INT_MASK + EVERGREEN_CRTC2_REGISTER_OFFSET, crtc3); WREG32(INT_MASK + EVERGREEN_CRTC3_REGISTER_OFFSET, crtc4); + } + if (rdev->num_crtc >= 6) { WREG32(INT_MASK + EVERGREEN_CRTC4_REGISTER_OFFSET, crtc5); WREG32(INT_MASK + EVERGREEN_CRTC5_REGISTER_OFFSET, crtc6); } WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET, grph1); WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC1_REGISTER_OFFSET, grph2); - WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, grph3); - WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, grph4); - WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, grph5); - WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, grph6); + if (rdev->num_crtc >= 4) { + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC2_REGISTER_OFFSET, grph3); + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC3_REGISTER_OFFSET, grph4); + } + if (rdev->num_crtc >= 6) { + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC4_REGISTER_OFFSET, grph5); + WREG32(GRPH_INT_CONTROL + EVERGREEN_CRTC5_REGISTER_OFFSET, grph6); + } WREG32(DC_HPD1_INT_CONTROL, hpd1); WREG32(DC_HPD2_INT_CONTROL, hpd2); @@ -2486,53 +2614,57 @@ static inline void evergreen_irq_ack(struct radeon_device *rdev) rdev->irq.stat_regs.evergreen.disp_int_cont5 = RREG32(DISP_INTERRUPT_STATUS_CONTINUE5); rdev->irq.stat_regs.evergreen.d1grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC0_REGISTER_OFFSET); rdev->irq.stat_regs.evergreen.d2grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC1_REGISTER_OFFSET); - rdev->irq.stat_regs.evergreen.d3grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET); - rdev->irq.stat_regs.evergreen.d4grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET); - rdev->irq.stat_regs.evergreen.d5grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET); - rdev->irq.stat_regs.evergreen.d6grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET); + if (rdev->num_crtc >= 4) { + rdev->irq.stat_regs.evergreen.d3grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET); + rdev->irq.stat_regs.evergreen.d4grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET); + } + if (rdev->num_crtc >= 6) { + rdev->irq.stat_regs.evergreen.d5grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET); + rdev->irq.stat_regs.evergreen.d6grph_int = RREG32(GRPH_INT_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET); + } if (rdev->irq.stat_regs.evergreen.d1grph_int & GRPH_PFLIP_INT_OCCURRED) WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC0_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); if (rdev->irq.stat_regs.evergreen.d2grph_int & GRPH_PFLIP_INT_OCCURRED) WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC1_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); - if (rdev->irq.stat_regs.evergreen.d3grph_int & GRPH_PFLIP_INT_OCCURRED) - WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); - if (rdev->irq.stat_regs.evergreen.d4grph_int & GRPH_PFLIP_INT_OCCURRED) - WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); - if (rdev->irq.stat_regs.evergreen.d5grph_int & GRPH_PFLIP_INT_OCCURRED) - WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); - if (rdev->irq.stat_regs.evergreen.d6grph_int & GRPH_PFLIP_INT_OCCURRED) - WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); - if (rdev->irq.stat_regs.evergreen.disp_int & LB_D1_VBLANK_INTERRUPT) WREG32(VBLANK_STATUS + EVERGREEN_CRTC0_REGISTER_OFFSET, VBLANK_ACK); if (rdev->irq.stat_regs.evergreen.disp_int & LB_D1_VLINE_INTERRUPT) WREG32(VLINE_STATUS + EVERGREEN_CRTC0_REGISTER_OFFSET, VLINE_ACK); - if (rdev->irq.stat_regs.evergreen.disp_int_cont & LB_D2_VBLANK_INTERRUPT) WREG32(VBLANK_STATUS + EVERGREEN_CRTC1_REGISTER_OFFSET, VBLANK_ACK); if (rdev->irq.stat_regs.evergreen.disp_int_cont & LB_D2_VLINE_INTERRUPT) WREG32(VLINE_STATUS + EVERGREEN_CRTC1_REGISTER_OFFSET, VLINE_ACK); - if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & LB_D3_VBLANK_INTERRUPT) - WREG32(VBLANK_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET, VBLANK_ACK); - if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & LB_D3_VLINE_INTERRUPT) - WREG32(VLINE_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET, VLINE_ACK); - - if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & LB_D4_VBLANK_INTERRUPT) - WREG32(VBLANK_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET, VBLANK_ACK); - if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & LB_D4_VLINE_INTERRUPT) - WREG32(VLINE_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET, VLINE_ACK); - - if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & LB_D5_VBLANK_INTERRUPT) - WREG32(VBLANK_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET, VBLANK_ACK); - if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & LB_D5_VLINE_INTERRUPT) - WREG32(VLINE_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET, VLINE_ACK); - - if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & LB_D6_VBLANK_INTERRUPT) - WREG32(VBLANK_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET, VBLANK_ACK); - if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & LB_D6_VLINE_INTERRUPT) - WREG32(VLINE_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET, VLINE_ACK); + if (rdev->num_crtc >= 4) { + if (rdev->irq.stat_regs.evergreen.d3grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.d4grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & LB_D3_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont2 & LB_D3_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC2_REGISTER_OFFSET, VLINE_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & LB_D4_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont3 & LB_D4_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC3_REGISTER_OFFSET, VLINE_ACK); + } + + if (rdev->num_crtc >= 6) { + if (rdev->irq.stat_regs.evergreen.d5grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.d6grph_int & GRPH_PFLIP_INT_OCCURRED) + WREG32(GRPH_INT_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET, GRPH_PFLIP_INT_CLEAR); + if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & LB_D5_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont4 & LB_D5_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC4_REGISTER_OFFSET, VLINE_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & LB_D6_VBLANK_INTERRUPT) + WREG32(VBLANK_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET, VBLANK_ACK); + if (rdev->irq.stat_regs.evergreen.disp_int_cont5 & LB_D6_VLINE_INTERRUPT) + WREG32(VLINE_STATUS + EVERGREEN_CRTC5_REGISTER_OFFSET, VLINE_ACK); + } if (rdev->irq.stat_regs.evergreen.disp_int & DC_HPD1_INTERRUPT) { tmp = RREG32(DC_HPD1_INT_CONTROL); @@ -2607,29 +2739,29 @@ static inline u32 evergreen_get_ih_wptr(struct radeon_device *rdev) int evergreen_irq_process(struct radeon_device *rdev) { - u32 wptr = evergreen_get_ih_wptr(rdev); - u32 rptr = rdev->ih.rptr; + u32 wptr; + u32 rptr; u32 src_id, src_data; u32 ring_index; unsigned long flags; bool queue_hotplug = false; - DRM_DEBUG("r600_irq_process start: rptr %d, wptr %d\n", rptr, wptr); - if (!rdev->ih.enabled) + if (!rdev->ih.enabled || rdev->shutdown) return IRQ_NONE; - spin_lock_irqsave(&rdev->ih.lock, flags); + wptr = evergreen_get_ih_wptr(rdev); + rptr = rdev->ih.rptr; + DRM_DEBUG("r600_irq_process start: rptr %d, wptr %d\n", rptr, wptr); + spin_lock_irqsave(&rdev->ih.lock, flags); if (rptr == wptr) { spin_unlock_irqrestore(&rdev->ih.lock, flags); return IRQ_NONE; } - if (rdev->shutdown) { - spin_unlock_irqrestore(&rdev->ih.lock, flags); - return IRQ_NONE; - } - restart_ih: + /* Order reading of wptr vs. reading of IH ring data */ + rmb(); + /* display interrupts */ evergreen_irq_ack(rdev); @@ -2857,7 +2989,7 @@ restart_ih: radeon_fence_process(rdev); break; case 233: /* GUI IDLE */ - DRM_DEBUG("IH: CP EOP\n"); + DRM_DEBUG("IH: GUI idle\n"); rdev->pm.gui_idle = true; wake_up(&rdev->irq.idle_queue); break; @@ -3143,6 +3275,7 @@ void evergreen_fini(struct radeon_device *rdev) r700_cp_fini(rdev); r600_irq_fini(rdev); radeon_wb_fini(rdev); + radeon_ib_pool_fini(rdev); radeon_irq_kms_fini(rdev); evergreen_pcie_gart_fini(rdev); radeon_gem_fini(rdev); diff --git a/drivers/gpu/drm/radeon/evergreen_blit_kms.c b/drivers/gpu/drm/radeon/evergreen_blit_kms.c index ba06a69c6de..2eb251858e7 100644 --- a/drivers/gpu/drm/radeon/evergreen_blit_kms.c +++ b/drivers/gpu/drm/radeon/evergreen_blit_kms.c @@ -31,6 +31,7 @@ #include "evergreend.h" #include "evergreen_blit_shaders.h" +#include "cayman_blit_shaders.h" #define DI_PT_RECTLIST 0x11 #define DI_INDEX_SIZE_16_BIT 0x0 @@ -152,6 +153,8 @@ set_vtx_resource(struct radeon_device *rdev, u64 gpu_addr) if ((rdev->family == CHIP_CEDAR) || (rdev->family == CHIP_PALM) || + (rdev->family == CHIP_SUMO) || + (rdev->family == CHIP_SUMO2) || (rdev->family == CHIP_CAICOS)) cp_set_surface_sync(rdev, PACKET3_TC_ACTION_ENA, 48, gpu_addr); @@ -199,6 +202,16 @@ static void set_scissors(struct radeon_device *rdev, int x1, int y1, int x2, int y2) { + /* workaround some hw bugs */ + if (x2 == 0) + x1 = 1; + if (y2 == 0) + y1 = 1; + if (rdev->family == CHIP_CAYMAN) { + if ((x2 == 1) && (y2 == 1)) + x2 = 2; + } + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONTEXT_REG, 2)); radeon_ring_write(rdev, (PA_SC_SCREEN_SCISSOR_TL - PACKET3_SET_CONTEXT_REG_START) >> 2); radeon_ring_write(rdev, (x1 << 0) | (y1 << 16)); @@ -239,7 +252,7 @@ draw_auto(struct radeon_device *rdev) } -/* emits 36 */ +/* emits 39 */ static void set_default_state(struct radeon_device *rdev) { @@ -255,238 +268,289 @@ set_default_state(struct radeon_device *rdev) u64 gpu_addr; int dwords; - switch (rdev->family) { - case CHIP_CEDAR: - default: - num_ps_gprs = 93; - num_vs_gprs = 46; - num_temp_gprs = 4; - num_gs_gprs = 31; - num_es_gprs = 31; - num_hs_gprs = 23; - num_ls_gprs = 23; - num_ps_threads = 96; - num_vs_threads = 16; - num_gs_threads = 16; - num_es_threads = 16; - num_hs_threads = 16; - num_ls_threads = 16; - num_ps_stack_entries = 42; - num_vs_stack_entries = 42; - num_gs_stack_entries = 42; - num_es_stack_entries = 42; - num_hs_stack_entries = 42; - num_ls_stack_entries = 42; - break; - case CHIP_REDWOOD: - num_ps_gprs = 93; - num_vs_gprs = 46; - num_temp_gprs = 4; - num_gs_gprs = 31; - num_es_gprs = 31; - num_hs_gprs = 23; - num_ls_gprs = 23; - num_ps_threads = 128; - num_vs_threads = 20; - num_gs_threads = 20; - num_es_threads = 20; - num_hs_threads = 20; - num_ls_threads = 20; - num_ps_stack_entries = 42; - num_vs_stack_entries = 42; - num_gs_stack_entries = 42; - num_es_stack_entries = 42; - num_hs_stack_entries = 42; - num_ls_stack_entries = 42; - break; - case CHIP_JUNIPER: - num_ps_gprs = 93; - num_vs_gprs = 46; - num_temp_gprs = 4; - num_gs_gprs = 31; - num_es_gprs = 31; - num_hs_gprs = 23; - num_ls_gprs = 23; - num_ps_threads = 128; - num_vs_threads = 20; - num_gs_threads = 20; - num_es_threads = 20; - num_hs_threads = 20; - num_ls_threads = 20; - num_ps_stack_entries = 85; - num_vs_stack_entries = 85; - num_gs_stack_entries = 85; - num_es_stack_entries = 85; - num_hs_stack_entries = 85; - num_ls_stack_entries = 85; - break; - case CHIP_CYPRESS: - case CHIP_HEMLOCK: - num_ps_gprs = 93; - num_vs_gprs = 46; - num_temp_gprs = 4; - num_gs_gprs = 31; - num_es_gprs = 31; - num_hs_gprs = 23; - num_ls_gprs = 23; - num_ps_threads = 128; - num_vs_threads = 20; - num_gs_threads = 20; - num_es_threads = 20; - num_hs_threads = 20; - num_ls_threads = 20; - num_ps_stack_entries = 85; - num_vs_stack_entries = 85; - num_gs_stack_entries = 85; - num_es_stack_entries = 85; - num_hs_stack_entries = 85; - num_ls_stack_entries = 85; - break; - case CHIP_PALM: - num_ps_gprs = 93; - num_vs_gprs = 46; - num_temp_gprs = 4; - num_gs_gprs = 31; - num_es_gprs = 31; - num_hs_gprs = 23; - num_ls_gprs = 23; - num_ps_threads = 96; - num_vs_threads = 16; - num_gs_threads = 16; - num_es_threads = 16; - num_hs_threads = 16; - num_ls_threads = 16; - num_ps_stack_entries = 42; - num_vs_stack_entries = 42; - num_gs_stack_entries = 42; - num_es_stack_entries = 42; - num_hs_stack_entries = 42; - num_ls_stack_entries = 42; - break; - case CHIP_BARTS: - num_ps_gprs = 93; - num_vs_gprs = 46; - num_temp_gprs = 4; - num_gs_gprs = 31; - num_es_gprs = 31; - num_hs_gprs = 23; - num_ls_gprs = 23; - num_ps_threads = 128; - num_vs_threads = 20; - num_gs_threads = 20; - num_es_threads = 20; - num_hs_threads = 20; - num_ls_threads = 20; - num_ps_stack_entries = 85; - num_vs_stack_entries = 85; - num_gs_stack_entries = 85; - num_es_stack_entries = 85; - num_hs_stack_entries = 85; - num_ls_stack_entries = 85; - break; - case CHIP_TURKS: - num_ps_gprs = 93; - num_vs_gprs = 46; - num_temp_gprs = 4; - num_gs_gprs = 31; - num_es_gprs = 31; - num_hs_gprs = 23; - num_ls_gprs = 23; - num_ps_threads = 128; - num_vs_threads = 20; - num_gs_threads = 20; - num_es_threads = 20; - num_hs_threads = 20; - num_ls_threads = 20; - num_ps_stack_entries = 42; - num_vs_stack_entries = 42; - num_gs_stack_entries = 42; - num_es_stack_entries = 42; - num_hs_stack_entries = 42; - num_ls_stack_entries = 42; - break; - case CHIP_CAICOS: - num_ps_gprs = 93; - num_vs_gprs = 46; - num_temp_gprs = 4; - num_gs_gprs = 31; - num_es_gprs = 31; - num_hs_gprs = 23; - num_ls_gprs = 23; - num_ps_threads = 128; - num_vs_threads = 10; - num_gs_threads = 10; - num_es_threads = 10; - num_hs_threads = 10; - num_ls_threads = 10; - num_ps_stack_entries = 42; - num_vs_stack_entries = 42; - num_gs_stack_entries = 42; - num_es_stack_entries = 42; - num_hs_stack_entries = 42; - num_ls_stack_entries = 42; - break; - } - - if ((rdev->family == CHIP_CEDAR) || - (rdev->family == CHIP_PALM) || - (rdev->family == CHIP_CAICOS)) - sq_config = 0; - else - sq_config = VC_ENABLE; - - sq_config |= (EXPORT_SRC_C | - CS_PRIO(0) | - LS_PRIO(0) | - HS_PRIO(0) | - PS_PRIO(0) | - VS_PRIO(1) | - GS_PRIO(2) | - ES_PRIO(3)); - - sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(num_ps_gprs) | - NUM_VS_GPRS(num_vs_gprs) | - NUM_CLAUSE_TEMP_GPRS(num_temp_gprs)); - sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(num_gs_gprs) | - NUM_ES_GPRS(num_es_gprs)); - sq_gpr_resource_mgmt_3 = (NUM_HS_GPRS(num_hs_gprs) | - NUM_LS_GPRS(num_ls_gprs)); - sq_thread_resource_mgmt = (NUM_PS_THREADS(num_ps_threads) | - NUM_VS_THREADS(num_vs_threads) | - NUM_GS_THREADS(num_gs_threads) | - NUM_ES_THREADS(num_es_threads)); - sq_thread_resource_mgmt_2 = (NUM_HS_THREADS(num_hs_threads) | - NUM_LS_THREADS(num_ls_threads)); - sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(num_ps_stack_entries) | - NUM_VS_STACK_ENTRIES(num_vs_stack_entries)); - sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(num_gs_stack_entries) | - NUM_ES_STACK_ENTRIES(num_es_stack_entries)); - sq_stack_resource_mgmt_3 = (NUM_HS_STACK_ENTRIES(num_hs_stack_entries) | - NUM_LS_STACK_ENTRIES(num_ls_stack_entries)); - /* set clear context state */ radeon_ring_write(rdev, PACKET3(PACKET3_CLEAR_STATE, 0)); radeon_ring_write(rdev, 0); - /* disable dyn gprs */ - radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); - radeon_ring_write(rdev, (SQ_DYN_GPR_CNTL_PS_FLUSH_REQ - PACKET3_SET_CONFIG_REG_START) >> 2); - radeon_ring_write(rdev, 0); + if (rdev->family < CHIP_CAYMAN) { + switch (rdev->family) { + case CHIP_CEDAR: + default: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 96; + num_vs_threads = 16; + num_gs_threads = 16; + num_es_threads = 16; + num_hs_threads = 16; + num_ls_threads = 16; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_REDWOOD: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_JUNIPER: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 85; + num_vs_stack_entries = 85; + num_gs_stack_entries = 85; + num_es_stack_entries = 85; + num_hs_stack_entries = 85; + num_ls_stack_entries = 85; + break; + case CHIP_CYPRESS: + case CHIP_HEMLOCK: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 85; + num_vs_stack_entries = 85; + num_gs_stack_entries = 85; + num_es_stack_entries = 85; + num_hs_stack_entries = 85; + num_ls_stack_entries = 85; + break; + case CHIP_PALM: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 96; + num_vs_threads = 16; + num_gs_threads = 16; + num_es_threads = 16; + num_hs_threads = 16; + num_ls_threads = 16; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_SUMO: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 96; + num_vs_threads = 25; + num_gs_threads = 25; + num_es_threads = 25; + num_hs_threads = 25; + num_ls_threads = 25; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_SUMO2: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 96; + num_vs_threads = 25; + num_gs_threads = 25; + num_es_threads = 25; + num_hs_threads = 25; + num_ls_threads = 25; + num_ps_stack_entries = 85; + num_vs_stack_entries = 85; + num_gs_stack_entries = 85; + num_es_stack_entries = 85; + num_hs_stack_entries = 85; + num_ls_stack_entries = 85; + break; + case CHIP_BARTS: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 85; + num_vs_stack_entries = 85; + num_gs_stack_entries = 85; + num_es_stack_entries = 85; + num_hs_stack_entries = 85; + num_ls_stack_entries = 85; + break; + case CHIP_TURKS: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 20; + num_gs_threads = 20; + num_es_threads = 20; + num_hs_threads = 20; + num_ls_threads = 20; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + case CHIP_CAICOS: + num_ps_gprs = 93; + num_vs_gprs = 46; + num_temp_gprs = 4; + num_gs_gprs = 31; + num_es_gprs = 31; + num_hs_gprs = 23; + num_ls_gprs = 23; + num_ps_threads = 128; + num_vs_threads = 10; + num_gs_threads = 10; + num_es_threads = 10; + num_hs_threads = 10; + num_ls_threads = 10; + num_ps_stack_entries = 42; + num_vs_stack_entries = 42; + num_gs_stack_entries = 42; + num_es_stack_entries = 42; + num_hs_stack_entries = 42; + num_ls_stack_entries = 42; + break; + } - /* SQ config */ - radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 11)); - radeon_ring_write(rdev, (SQ_CONFIG - PACKET3_SET_CONFIG_REG_START) >> 2); - radeon_ring_write(rdev, sq_config); - radeon_ring_write(rdev, sq_gpr_resource_mgmt_1); - radeon_ring_write(rdev, sq_gpr_resource_mgmt_2); - radeon_ring_write(rdev, sq_gpr_resource_mgmt_3); - radeon_ring_write(rdev, 0); - radeon_ring_write(rdev, 0); - radeon_ring_write(rdev, sq_thread_resource_mgmt); - radeon_ring_write(rdev, sq_thread_resource_mgmt_2); - radeon_ring_write(rdev, sq_stack_resource_mgmt_1); - radeon_ring_write(rdev, sq_stack_resource_mgmt_2); - radeon_ring_write(rdev, sq_stack_resource_mgmt_3); + if ((rdev->family == CHIP_CEDAR) || + (rdev->family == CHIP_PALM) || + (rdev->family == CHIP_SUMO) || + (rdev->family == CHIP_SUMO2) || + (rdev->family == CHIP_CAICOS)) + sq_config = 0; + else + sq_config = VC_ENABLE; + + sq_config |= (EXPORT_SRC_C | + CS_PRIO(0) | + LS_PRIO(0) | + HS_PRIO(0) | + PS_PRIO(0) | + VS_PRIO(1) | + GS_PRIO(2) | + ES_PRIO(3)); + + sq_gpr_resource_mgmt_1 = (NUM_PS_GPRS(num_ps_gprs) | + NUM_VS_GPRS(num_vs_gprs) | + NUM_CLAUSE_TEMP_GPRS(num_temp_gprs)); + sq_gpr_resource_mgmt_2 = (NUM_GS_GPRS(num_gs_gprs) | + NUM_ES_GPRS(num_es_gprs)); + sq_gpr_resource_mgmt_3 = (NUM_HS_GPRS(num_hs_gprs) | + NUM_LS_GPRS(num_ls_gprs)); + sq_thread_resource_mgmt = (NUM_PS_THREADS(num_ps_threads) | + NUM_VS_THREADS(num_vs_threads) | + NUM_GS_THREADS(num_gs_threads) | + NUM_ES_THREADS(num_es_threads)); + sq_thread_resource_mgmt_2 = (NUM_HS_THREADS(num_hs_threads) | + NUM_LS_THREADS(num_ls_threads)); + sq_stack_resource_mgmt_1 = (NUM_PS_STACK_ENTRIES(num_ps_stack_entries) | + NUM_VS_STACK_ENTRIES(num_vs_stack_entries)); + sq_stack_resource_mgmt_2 = (NUM_GS_STACK_ENTRIES(num_gs_stack_entries) | + NUM_ES_STACK_ENTRIES(num_es_stack_entries)); + sq_stack_resource_mgmt_3 = (NUM_HS_STACK_ENTRIES(num_hs_stack_entries) | + NUM_LS_STACK_ENTRIES(num_ls_stack_entries)); + + /* disable dyn gprs */ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(rdev, (SQ_DYN_GPR_CNTL_PS_FLUSH_REQ - PACKET3_SET_CONFIG_REG_START) >> 2); + radeon_ring_write(rdev, 0); + + /* setup LDS */ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 1)); + radeon_ring_write(rdev, (SQ_LDS_RESOURCE_MGMT - PACKET3_SET_CONFIG_REG_START) >> 2); + radeon_ring_write(rdev, 0x10001000); + + /* SQ config */ + radeon_ring_write(rdev, PACKET3(PACKET3_SET_CONFIG_REG, 11)); + radeon_ring_write(rdev, (SQ_CONFIG - PACKET3_SET_CONFIG_REG_START) >> 2); + radeon_ring_write(rdev, sq_config); + radeon_ring_write(rdev, sq_gpr_resource_mgmt_1); + radeon_ring_write(rdev, sq_gpr_resource_mgmt_2); + radeon_ring_write(rdev, sq_gpr_resource_mgmt_3); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, 0); + radeon_ring_write(rdev, sq_thread_resource_mgmt); + radeon_ring_write(rdev, sq_thread_resource_mgmt_2); + radeon_ring_write(rdev, sq_stack_resource_mgmt_1); + radeon_ring_write(rdev, sq_stack_resource_mgmt_2); + radeon_ring_write(rdev, sq_stack_resource_mgmt_3); + } /* CONTEXT_CONTROL */ radeon_ring_write(rdev, 0xc0012800); @@ -560,7 +624,10 @@ int evergreen_blit_init(struct radeon_device *rdev) mutex_init(&rdev->r600_blit.mutex); rdev->r600_blit.state_offset = 0; - rdev->r600_blit.state_len = evergreen_default_size; + if (rdev->family < CHIP_CAYMAN) + rdev->r600_blit.state_len = evergreen_default_size; + else + rdev->r600_blit.state_len = cayman_default_size; dwords = rdev->r600_blit.state_len; while (dwords & 0xf) { @@ -572,11 +639,17 @@ int evergreen_blit_init(struct radeon_device *rdev) obj_size = ALIGN(obj_size, 256); rdev->r600_blit.vs_offset = obj_size; - obj_size += evergreen_vs_size * 4; + if (rdev->family < CHIP_CAYMAN) + obj_size += evergreen_vs_size * 4; + else + obj_size += cayman_vs_size * 4; obj_size = ALIGN(obj_size, 256); rdev->r600_blit.ps_offset = obj_size; - obj_size += evergreen_ps_size * 4; + if (rdev->family < CHIP_CAYMAN) + obj_size += evergreen_ps_size * 4; + else + obj_size += cayman_ps_size * 4; obj_size = ALIGN(obj_size, 256); r = radeon_bo_create(rdev, obj_size, PAGE_SIZE, true, RADEON_GEM_DOMAIN_VRAM, @@ -599,16 +672,29 @@ int evergreen_blit_init(struct radeon_device *rdev) return r; } - memcpy_toio(ptr + rdev->r600_blit.state_offset, - evergreen_default_state, rdev->r600_blit.state_len * 4); - - if (num_packet2s) - memcpy_toio(ptr + rdev->r600_blit.state_offset + (rdev->r600_blit.state_len * 4), - packet2s, num_packet2s * 4); - for (i = 0; i < evergreen_vs_size; i++) - *(u32 *)((unsigned long)ptr + rdev->r600_blit.vs_offset + i * 4) = cpu_to_le32(evergreen_vs[i]); - for (i = 0; i < evergreen_ps_size; i++) - *(u32 *)((unsigned long)ptr + rdev->r600_blit.ps_offset + i * 4) = cpu_to_le32(evergreen_ps[i]); + if (rdev->family < CHIP_CAYMAN) { + memcpy_toio(ptr + rdev->r600_blit.state_offset, + evergreen_default_state, rdev->r600_blit.state_len * 4); + + if (num_packet2s) + memcpy_toio(ptr + rdev->r600_blit.state_offset + (rdev->r600_blit.state_len * 4), + packet2s, num_packet2s * 4); + for (i = 0; i < evergreen_vs_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.vs_offset + i * 4) = cpu_to_le32(evergreen_vs[i]); + for (i = 0; i < evergreen_ps_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.ps_offset + i * 4) = cpu_to_le32(evergreen_ps[i]); + } else { + memcpy_toio(ptr + rdev->r600_blit.state_offset, + cayman_default_state, rdev->r600_blit.state_len * 4); + + if (num_packet2s) + memcpy_toio(ptr + rdev->r600_blit.state_offset + (rdev->r600_blit.state_len * 4), + packet2s, num_packet2s * 4); + for (i = 0; i < cayman_vs_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.vs_offset + i * 4) = cpu_to_le32(cayman_vs[i]); + for (i = 0; i < cayman_ps_size; i++) + *(u32 *)((unsigned long)ptr + rdev->r600_blit.ps_offset + i * 4) = cpu_to_le32(cayman_ps[i]); + } radeon_bo_kunmap(rdev->r600_blit.shader_obj); radeon_bo_unreserve(rdev->r600_blit.shader_obj); @@ -692,7 +778,7 @@ int evergreen_blit_prepare_copy(struct radeon_device *rdev, int size_bytes) /* calculate number of loops correctly */ ring_size = num_loops * dwords_per_loop; /* set default + shaders */ - ring_size += 52; /* shaders + def state */ + ring_size += 55; /* shaders + def state */ ring_size += 10; /* fence emit for VB IB */ ring_size += 5; /* done copy */ ring_size += 10; /* fence emit for done copy */ diff --git a/drivers/gpu/drm/radeon/evergreen_cs.c b/drivers/gpu/drm/radeon/evergreen_cs.c index 23d36417158..a134790903d 100644 --- a/drivers/gpu/drm/radeon/evergreen_cs.c +++ b/drivers/gpu/drm/radeon/evergreen_cs.c @@ -428,7 +428,7 @@ static inline int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u3 last_reg = ARRAY_SIZE(evergreen_reg_safe_bm); i = (reg >> 7); - if (i > last_reg) { + if (i >= last_reg) { dev_warn(p->dev, "forbidden register 0x%08x at %d\n", reg, idx); return -EINVAL; } @@ -856,7 +856,6 @@ static inline int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u3 case SQ_PGM_START_PS: case SQ_PGM_START_HS: case SQ_PGM_START_LS: - case GDS_ADDR_BASE: case SQ_CONST_MEM_BASE: case SQ_ALU_CONST_CACHE_GS_0: case SQ_ALU_CONST_CACHE_GS_1: @@ -946,6 +945,34 @@ static inline int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u3 } ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); break; + case SX_MEMORY_EXPORT_BASE: + if (p->rdev->family >= CHIP_CAYMAN) { + dev_warn(p->dev, "bad SET_CONFIG_REG " + "0x%04X\n", reg); + return -EINVAL; + } + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONFIG_REG " + "0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + break; + case CAYMAN_SX_SCATTER_EXPORT_BASE: + if (p->rdev->family < CHIP_CAYMAN) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONTEXT_REG " + "0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + break; default: dev_warn(p->dev, "forbidden register 0x%08x at %d\n", reg, idx); return -EINVAL; @@ -1153,6 +1180,34 @@ static int evergreen_packet3_check(struct radeon_cs_parser *p, return r; } break; + case PACKET3_DISPATCH_DIRECT: + if (pkt->count != 3) { + DRM_ERROR("bad DISPATCH_DIRECT\n"); + return -EINVAL; + } + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream %d\n", __func__, __LINE__, idx); + return r; + } + break; + case PACKET3_DISPATCH_INDIRECT: + if (pkt->count != 1) { + DRM_ERROR("bad DISPATCH_INDIRECT\n"); + return -EINVAL; + } + r = evergreen_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad DISPATCH_INDIRECT\n"); + return -EINVAL; + } + ib[idx+0] = idx_value + (u32)(reloc->lobj.gpu_offset & 0xffffffff); + r = evergreen_cs_track_check(p); + if (r) { + dev_warn(p->dev, "%s:%d invalid cmd stream\n", __func__, __LINE__); + return r; + } + break; case PACKET3_WAIT_REG_MEM: if (pkt->count != 5) { DRM_ERROR("bad WAIT_REG_MEM\n"); diff --git a/drivers/gpu/drm/radeon/evergreend.h b/drivers/gpu/drm/radeon/evergreend.h index 94533849927..7363d9dec90 100644 --- a/drivers/gpu/drm/radeon/evergreend.h +++ b/drivers/gpu/drm/radeon/evergreend.h @@ -64,6 +64,8 @@ #define GB_BACKEND_MAP 0x98FC #define DMIF_ADDR_CONFIG 0xBD4 #define HDP_ADDR_CONFIG 0x2F48 +#define HDP_MISC_CNTL 0x2F4C +#define HDP_FLUSH_INVALIDATE_CACHE (1 << 0) #define CC_SYS_RB_BACKEND_DISABLE 0x3F88 #define GC_USER_RB_BACKEND_DISABLE 0x9B7C @@ -166,10 +168,16 @@ #define SE_DB_BUSY (1 << 30) #define SE_CB_BUSY (1 << 31) /* evergreen */ +#define CG_THERMAL_CTRL 0x72c +#define TOFFSET_MASK 0x00003FE0 +#define TOFFSET_SHIFT 5 #define CG_MULT_THERMAL_STATUS 0x740 #define ASIC_T(x) ((x) << 16) -#define ASIC_T_MASK 0x7FF0000 +#define ASIC_T_MASK 0x07FF0000 #define ASIC_T_SHIFT 16 +#define CG_TS0_STATUS 0x760 +#define TS0_ADC_DOUT_MASK 0x000003FF +#define TS0_ADC_DOUT_SHIFT 0 /* APU */ #define CG_THERMAL_STATUS 0x678 @@ -200,6 +208,7 @@ #define BURSTLENGTH_SHIFT 9 #define BURSTLENGTH_MASK 0x00000200 #define CHANSIZE_OVERRIDE (1 << 11) +#define FUS_MC_ARB_RAMCFG 0x2768 #define MC_VM_AGP_TOP 0x2028 #define MC_VM_AGP_BOT 0x202C #define MC_VM_AGP_BASE 0x2030 @@ -342,6 +351,7 @@ #define COLOR_BUFFER_SIZE(x) ((x) << 0) #define POSITION_BUFFER_SIZE(x) ((x) << 8) #define SMX_BUFFER_SIZE(x) ((x) << 16) +#define SX_MEMORY_EXPORT_BASE 0x9010 #define SX_MISC 0x28350 #define CB_PERF_CTR0_SEL_0 0x9A20 @@ -457,7 +467,7 @@ #define IH_RB_WPTR_ADDR_LO 0x3e14 #define IH_CNTL 0x3e18 # define ENABLE_INTR (1 << 0) -# define IH_MC_SWAP(x) ((x) << 2) +# define IH_MC_SWAP(x) ((x) << 1) # define IH_MC_SWAP_NONE 0 # define IH_MC_SWAP_16BIT 1 # define IH_MC_SWAP_32BIT 2 @@ -538,7 +548,7 @@ # define LB_D5_VBLANK_INTERRUPT (1 << 3) # define DC_HPD5_INTERRUPT (1 << 17) # define DC_HPD5_RX_INTERRUPT (1 << 18) -#define DISP_INTERRUPT_STATUS_CONTINUE5 0x6050 +#define DISP_INTERRUPT_STATUS_CONTINUE5 0x6150 # define LB_D6_VLINE_INTERRUPT (1 << 2) # define LB_D6_VBLANK_INTERRUPT (1 << 3) # define DC_HPD6_INTERRUPT (1 << 17) @@ -1113,6 +1123,7 @@ #define CAYMAN_PA_SC_AA_CONFIG 0x28BE0 #define CAYMAN_MSAA_NUM_SAMPLES_SHIFT 0 #define CAYMAN_MSAA_NUM_SAMPLES_MASK 0x7 +#define CAYMAN_SX_SCATTER_EXPORT_BASE 0x28358 /* cayman packet3 addition */ #define CAYMAN_PACKET3_DEALLOC_STATE 0x14 diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c index 7aade20f63a..44c4750f451 100644 --- a/drivers/gpu/drm/radeon/ni.c +++ b/drivers/gpu/drm/radeon/ni.c @@ -417,7 +417,7 @@ static u32 cayman_get_tile_pipe_to_backend_map(struct radeon_device *rdev, num_shader_engines = 1; if (num_shader_engines > rdev->config.cayman.max_shader_engines) num_shader_engines = rdev->config.cayman.max_shader_engines; - if (num_backends_per_asic > num_shader_engines) + if (num_backends_per_asic < num_shader_engines) num_backends_per_asic = num_shader_engines; if (num_backends_per_asic > (rdev->config.cayman.max_backends_per_se * num_shader_engines)) num_backends_per_asic = rdev->config.cayman.max_backends_per_se * num_shader_engines; @@ -674,7 +674,7 @@ static void cayman_gpu_init(struct radeon_device *rdev) cc_rb_backend_disable = RREG32(CC_RB_BACKEND_DISABLE); cc_gc_shader_pipe_config = RREG32(CC_GC_SHADER_PIPE_CONFIG); - cgts_tcc_disable = RREG32(CGTS_TCC_DISABLE); + cgts_tcc_disable = 0xff000000; gc_user_rb_backend_disable = RREG32(GC_USER_RB_BACKEND_DISABLE); gc_user_shader_pipe_config = RREG32(GC_USER_SHADER_PIPE_CONFIG); cgts_user_tcc_disable = RREG32(CGTS_USER_TCC_DISABLE); @@ -829,10 +829,11 @@ static void cayman_gpu_init(struct radeon_device *rdev) rdev->config.cayman.tile_config |= ((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) << 4; rdev->config.cayman.tile_config |= - (gb_addr_config & PIPE_INTERLEAVE_SIZE_MASK) >> PIPE_INTERLEAVE_SIZE_SHIFT; + ((gb_addr_config & PIPE_INTERLEAVE_SIZE_MASK) >> PIPE_INTERLEAVE_SIZE_SHIFT) << 8; rdev->config.cayman.tile_config |= ((gb_addr_config & ROW_SIZE_MASK) >> ROW_SIZE_SHIFT) << 12; + rdev->config.cayman.backend_map = gb_backend_map; WREG32(GB_BACKEND_MAP, gb_backend_map); WREG32(GB_ADDR_CONFIG, gb_addr_config); WREG32(DMIF_ADDR_CONFIG, gb_addr_config); @@ -871,7 +872,7 @@ static void cayman_gpu_init(struct radeon_device *rdev) smx_dc_ctl0 = RREG32(SMX_DC_CTL0); smx_dc_ctl0 &= ~NUMBER_OF_SETS(0x1ff); - smx_dc_ctl0 |= NUMBER_OF_SETS(rdev->config.evergreen.sx_num_of_sets); + smx_dc_ctl0 |= NUMBER_OF_SETS(rdev->config.cayman.sx_num_of_sets); WREG32(SMX_DC_CTL0, smx_dc_ctl0); WREG32(SPI_CONFIG_CNTL_1, VTX_DONE_DELAY(4) | CRC_SIMD_ID_WADDR_DISABLE); @@ -887,20 +888,20 @@ static void cayman_gpu_init(struct radeon_device *rdev) WREG32(TA_CNTL_AUX, DISABLE_CUBE_ANISO); - WREG32(SX_EXPORT_BUFFER_SIZES, (COLOR_BUFFER_SIZE((rdev->config.evergreen.sx_max_export_size / 4) - 1) | - POSITION_BUFFER_SIZE((rdev->config.evergreen.sx_max_export_pos_size / 4) - 1) | - SMX_BUFFER_SIZE((rdev->config.evergreen.sx_max_export_smx_size / 4) - 1))); + WREG32(SX_EXPORT_BUFFER_SIZES, (COLOR_BUFFER_SIZE((rdev->config.cayman.sx_max_export_size / 4) - 1) | + POSITION_BUFFER_SIZE((rdev->config.cayman.sx_max_export_pos_size / 4) - 1) | + SMX_BUFFER_SIZE((rdev->config.cayman.sx_max_export_smx_size / 4) - 1))); - WREG32(PA_SC_FIFO_SIZE, (SC_PRIM_FIFO_SIZE(rdev->config.evergreen.sc_prim_fifo_size) | - SC_HIZ_TILE_FIFO_SIZE(rdev->config.evergreen.sc_hiz_tile_fifo_size) | - SC_EARLYZ_TILE_FIFO_SIZE(rdev->config.evergreen.sc_earlyz_tile_fifo_size))); + WREG32(PA_SC_FIFO_SIZE, (SC_PRIM_FIFO_SIZE(rdev->config.cayman.sc_prim_fifo_size) | + SC_HIZ_TILE_FIFO_SIZE(rdev->config.cayman.sc_hiz_tile_fifo_size) | + SC_EARLYZ_TILE_FIFO_SIZE(rdev->config.cayman.sc_earlyz_tile_fifo_size))); WREG32(VGT_NUM_INSTANCES, 1); WREG32(CP_PERFMON_CNTL, 0); - WREG32(SQ_MS_FIFO_SIZES, (CACHE_FIFO_SIZE(16 * rdev->config.evergreen.sq_num_cf_insts) | + WREG32(SQ_MS_FIFO_SIZES, (CACHE_FIFO_SIZE(16 * rdev->config.cayman.sq_num_cf_insts) | FETCH_FIFO_HIWATER(0x4) | DONE_FIFO_HIWATER(0xe0) | ALU_UPDATE_FIFO_HIWATER(0x8))); @@ -931,6 +932,10 @@ static void cayman_gpu_init(struct radeon_device *rdev) WREG32(CB_PERF_CTR3_SEL_0, 0); WREG32(CB_PERF_CTR3_SEL_1, 0); + tmp = RREG32(HDP_MISC_CNTL); + tmp |= HDP_FLUSH_INVALIDATE_CACHE; + WREG32(HDP_MISC_CNTL, tmp); + hdp_host_path_cntl = RREG32(HDP_HOST_PATH_CNTL); WREG32(HDP_HOST_PATH_CNTL, hdp_host_path_cntl); @@ -1383,14 +1388,12 @@ static int cayman_startup(struct radeon_device *rdev) return r; cayman_gpu_init(rdev); -#if 0 - r = cayman_blit_init(rdev); + r = evergreen_blit_init(rdev); if (r) { - cayman_blit_fini(rdev); + evergreen_blit_fini(rdev); rdev->asic->copy = NULL; dev_warn(rdev->dev, "failed blitter (%d) falling back to memcpy\n", r); } -#endif /* allocate wb buffer */ r = radeon_wb_init(rdev); @@ -1448,7 +1451,7 @@ int cayman_resume(struct radeon_device *rdev) int cayman_suspend(struct radeon_device *rdev) { - /* int r; */ + int r; /* FIXME: we should wait for ring to be empty */ cayman_cp_enable(rdev, false); @@ -1457,14 +1460,13 @@ int cayman_suspend(struct radeon_device *rdev) radeon_wb_disable(rdev); cayman_pcie_gart_disable(rdev); -#if 0 /* unpin shaders bo */ r = radeon_bo_reserve(rdev->r600_blit.shader_obj, false); if (likely(r == 0)) { radeon_bo_unpin(rdev->r600_blit.shader_obj); radeon_bo_unreserve(rdev->r600_blit.shader_obj); } -#endif + return 0; } @@ -1576,10 +1578,11 @@ int cayman_init(struct radeon_device *rdev) void cayman_fini(struct radeon_device *rdev) { - /* cayman_blit_fini(rdev); */ + evergreen_blit_fini(rdev); cayman_cp_fini(rdev); r600_irq_fini(rdev); radeon_wb_fini(rdev); + radeon_ib_pool_fini(rdev); radeon_irq_kms_fini(rdev); cayman_pcie_gart_fini(rdev); radeon_gem_fini(rdev); diff --git a/drivers/gpu/drm/radeon/nid.h b/drivers/gpu/drm/radeon/nid.h index 0f9a08b53fb..4672869cdb2 100644 --- a/drivers/gpu/drm/radeon/nid.h +++ b/drivers/gpu/drm/radeon/nid.h @@ -136,6 +136,8 @@ #define HDP_NONSURFACE_INFO 0x2C08 #define HDP_NONSURFACE_SIZE 0x2C0C #define HDP_ADDR_CONFIG 0x2F48 +#define HDP_MISC_CNTL 0x2F4C +#define HDP_FLUSH_INVALIDATE_CACHE (1 << 0) #define CC_SYS_RB_BACKEND_DISABLE 0x3F88 #define GC_USER_SYS_RB_BACKEND_DISABLE 0x3F8C @@ -318,7 +320,7 @@ #define CGTS_USER_TCC_DISABLE 0x914C #define TCC_DISABLE_MASK 0xFFFF0000 #define TCC_DISABLE_SHIFT 16 -#define CGTS_SM_CTRL_REG 0x915C +#define CGTS_SM_CTRL_REG 0x9150 #define OVERRIDE (1 << 21) #define TA_CNTL_AUX 0x9508 @@ -351,7 +353,7 @@ #define MULTI_GPU_TILE_SIZE_MASK 0x03000000 #define MULTI_GPU_TILE_SIZE_SHIFT 24 #define ROW_SIZE(x) ((x) << 28) -#define ROW_SIZE_MASK 0x30000007 +#define ROW_SIZE_MASK 0x30000000 #define ROW_SIZE_SHIFT 28 #define NUM_LOWER_PIPES(x) ((x) << 30) #define NUM_LOWER_PIPES_MASK 0x40000000 diff --git a/drivers/gpu/drm/radeon/r100_track.h b/drivers/gpu/drm/radeon/r100_track.h index 2fef9de7f36..686f9dc5d4b 100644 --- a/drivers/gpu/drm/radeon/r100_track.h +++ b/drivers/gpu/drm/radeon/r100_track.h @@ -63,7 +63,7 @@ struct r100_cs_track { unsigned num_arrays; unsigned max_indx; unsigned color_channel_mask; - struct r100_cs_track_array arrays[11]; + struct r100_cs_track_array arrays[16]; struct r100_cs_track_cb cb[R300_MAX_CB]; struct r100_cs_track_cb zb; struct r100_cs_track_cb aa; @@ -146,6 +146,12 @@ static inline int r100_packet3_load_vbpntr(struct radeon_cs_parser *p, ib = p->ib->ptr; track = (struct r100_cs_track *)p->track; c = radeon_get_ib_value(p, idx++) & 0x1F; + if (c > 16) { + DRM_ERROR("Only 16 vertex buffers are allowed %d\n", + pkt->opcode); + r100_cs_dump_packet(p, pkt); + return -EINVAL; + } track->num_arrays = c; for (i = 0; i < (c - 1); i+=2, idx+=3) { r = r100_cs_packet_next_reloc(p, &reloc); diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c index 6f27593901c..aa5571b73aa 100644 --- a/drivers/gpu/drm/radeon/r600.c +++ b/drivers/gpu/drm/radeon/r600.c @@ -87,6 +87,10 @@ MODULE_FIRMWARE("radeon/CYPRESS_rlc.bin"); MODULE_FIRMWARE("radeon/PALM_pfp.bin"); MODULE_FIRMWARE("radeon/PALM_me.bin"); MODULE_FIRMWARE("radeon/SUMO_rlc.bin"); +MODULE_FIRMWARE("radeon/SUMO_pfp.bin"); +MODULE_FIRMWARE("radeon/SUMO_me.bin"); +MODULE_FIRMWARE("radeon/SUMO2_pfp.bin"); +MODULE_FIRMWARE("radeon/SUMO2_me.bin"); int r600_debugfs_mc_info_init(struct radeon_device *rdev); @@ -586,6 +590,9 @@ void r600_pm_misc(struct radeon_device *rdev) struct radeon_voltage *voltage = &ps->clock_info[req_cm_idx].voltage; if ((voltage->type == VOLTAGE_SW) && voltage->voltage) { + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage->voltage == 0xff01) + return; if (voltage->voltage != rdev->pm.current_vddc) { radeon_atom_set_voltage(rdev, voltage->voltage, SET_VOLTAGE_TYPE_ASIC_VDDC); rdev->pm.current_vddc = voltage->voltage; @@ -1655,6 +1662,7 @@ void r600_gpu_init(struct radeon_device *rdev) R6XX_MAX_BACKENDS_MASK) >> 16)), (cc_rb_backend_disable >> 16)); rdev->config.r600.tile_config = tiling_config; + rdev->config.r600.backend_map = backend_map; tiling_config |= BACKEND_MAP(backend_map); WREG32(GB_TILING_CONFIG, tiling_config); WREG32(DCP_TILING_CONFIG, tiling_config & 0xffff); @@ -2024,6 +2032,14 @@ int r600_init_microcode(struct radeon_device *rdev) chip_name = "PALM"; rlc_chip_name = "SUMO"; break; + case CHIP_SUMO: + chip_name = "SUMO"; + rlc_chip_name = "SUMO"; + break; + case CHIP_SUMO2: + chip_name = "SUMO2"; + rlc_chip_name = "SUMO"; + break; default: BUG(); } @@ -2197,9 +2213,6 @@ int r600_cp_resume(struct radeon_device *rdev) /* set the wb address whether it's enabled or not */ WREG32(CP_RB_RPTR_ADDR, -#ifdef __BIG_ENDIAN - RB_RPTR_SWAP(2) | -#endif ((rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFFFFFFFC)); WREG32(CP_RB_RPTR_ADDR_HI, upper_32_bits(rdev->wb.gpu_addr + RADEON_WB_CP_RPTR_OFFSET) & 0xFF); WREG32(SCRATCH_ADDR, ((rdev->wb.gpu_addr + RADEON_WB_SCRATCH_OFFSET) >> 8) & 0xFFFFFFFF); @@ -2613,6 +2626,7 @@ void r600_fini(struct radeon_device *rdev) r600_cp_fini(rdev); r600_irq_fini(rdev); radeon_wb_fini(rdev); + radeon_ib_pool_fini(rdev); radeon_irq_kms_fini(rdev); r600_pcie_gart_fini(rdev); radeon_agp_fini(rdev); @@ -2978,10 +2992,6 @@ int r600_irq_init(struct radeon_device *rdev) /* RPTR_REARM only works if msi's are enabled */ if (rdev->msi_enabled) ih_cntl |= RPTR_REARM; - -#ifdef __BIG_ENDIAN - ih_cntl |= IH_MC_SWAP(IH_MC_SWAP_32BIT); -#endif WREG32(IH_CNTL, ih_cntl); /* force the active interrupt state to all disabled */ @@ -3282,29 +3292,35 @@ static inline u32 r600_get_ih_wptr(struct radeon_device *rdev) int r600_irq_process(struct radeon_device *rdev) { - u32 wptr = r600_get_ih_wptr(rdev); - u32 rptr = rdev->ih.rptr; + u32 wptr; + u32 rptr; u32 src_id, src_data; u32 ring_index; unsigned long flags; bool queue_hotplug = false; - DRM_DEBUG("r600_irq_process start: rptr %d, wptr %d\n", rptr, wptr); - if (!rdev->ih.enabled) + if (!rdev->ih.enabled || rdev->shutdown) return IRQ_NONE; + /* No MSIs, need a dummy read to flush PCI DMAs */ + if (!rdev->msi_enabled) + RREG32(IH_RB_WPTR); + + wptr = r600_get_ih_wptr(rdev); + rptr = rdev->ih.rptr; + DRM_DEBUG("r600_irq_process start: rptr %d, wptr %d\n", rptr, wptr); + spin_lock_irqsave(&rdev->ih.lock, flags); if (rptr == wptr) { spin_unlock_irqrestore(&rdev->ih.lock, flags); return IRQ_NONE; } - if (rdev->shutdown) { - spin_unlock_irqrestore(&rdev->ih.lock, flags); - return IRQ_NONE; - } restart_ih: + /* Order reading of wptr vs. reading of IH ring data */ + rmb(); + /* display interrupts */ r600_irq_ack(rdev); @@ -3432,7 +3448,7 @@ restart_ih: radeon_fence_process(rdev); break; case 233: /* GUI IDLE */ - DRM_DEBUG("IH: CP EOP\n"); + DRM_DEBUG("IH: GUI idle\n"); rdev->pm.gui_idle = true; wake_up(&rdev->irq.idle_queue); break; diff --git a/drivers/gpu/drm/radeon/r600_cp.c b/drivers/gpu/drm/radeon/r600_cp.c index c3ab959bdc7..45fd592f960 100644 --- a/drivers/gpu/drm/radeon/r600_cp.c +++ b/drivers/gpu/drm/radeon/r600_cp.c @@ -1802,8 +1802,8 @@ static void r600_cp_init_ring_buffer(struct drm_device *dev, /* Set ring buffer size */ #ifdef __BIG_ENDIAN RADEON_WRITE(R600_CP_RB_CNTL, - RADEON_BUF_SWAP_32BIT | - RADEON_RB_NO_UPDATE | + R600_BUF_SWAP_32BIT | + R600_RB_NO_UPDATE | (dev_priv->ring.rptr_update_l2qw << 8) | dev_priv->ring.size_l2qw); #else @@ -1820,15 +1820,15 @@ static void r600_cp_init_ring_buffer(struct drm_device *dev, #ifdef __BIG_ENDIAN RADEON_WRITE(R600_CP_RB_CNTL, - RADEON_BUF_SWAP_32BIT | - RADEON_RB_NO_UPDATE | - RADEON_RB_RPTR_WR_ENA | + R600_BUF_SWAP_32BIT | + R600_RB_NO_UPDATE | + R600_RB_RPTR_WR_ENA | (dev_priv->ring.rptr_update_l2qw << 8) | dev_priv->ring.size_l2qw); #else RADEON_WRITE(R600_CP_RB_CNTL, - RADEON_RB_NO_UPDATE | - RADEON_RB_RPTR_WR_ENA | + R600_RB_NO_UPDATE | + R600_RB_RPTR_WR_ENA | (dev_priv->ring.rptr_update_l2qw << 8) | dev_priv->ring.size_l2qw); #endif @@ -1851,13 +1851,8 @@ static void r600_cp_init_ring_buffer(struct drm_device *dev, - ((unsigned long) dev->sg->virtual) + dev_priv->gart_vm_start; } - RADEON_WRITE(R600_CP_RB_RPTR_ADDR, -#ifdef __BIG_ENDIAN - (2 << 0) | -#endif - (rptr_addr & 0xfffffffc)); - RADEON_WRITE(R600_CP_RB_RPTR_ADDR_HI, - upper_32_bits(rptr_addr)); + RADEON_WRITE(R600_CP_RB_RPTR_ADDR, (rptr_addr & 0xfffffffc)); + RADEON_WRITE(R600_CP_RB_RPTR_ADDR_HI, upper_32_bits(rptr_addr)); #ifdef __BIG_ENDIAN RADEON_WRITE(R600_CP_RB_CNTL, diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c index fd18be9871a..cf83aa05a68 100644 --- a/drivers/gpu/drm/radeon/r600_cs.c +++ b/drivers/gpu/drm/radeon/r600_cs.c @@ -71,20 +71,21 @@ struct r600_cs_track { u64 db_bo_mc; }; -#define FMT_8_BIT(fmt, vc) [fmt] = { 1, 1, 1, vc } -#define FMT_16_BIT(fmt, vc) [fmt] = { 1, 1, 2, vc } -#define FMT_24_BIT(fmt) [fmt] = { 1, 1, 3, 0 } -#define FMT_32_BIT(fmt, vc) [fmt] = { 1, 1, 4, vc } -#define FMT_48_BIT(fmt) [fmt] = { 1, 1, 6, 0 } -#define FMT_64_BIT(fmt, vc) [fmt] = { 1, 1, 8, vc } -#define FMT_96_BIT(fmt) [fmt] = { 1, 1, 12, 0 } -#define FMT_128_BIT(fmt, vc) [fmt] = { 1, 1, 16, vc } +#define FMT_8_BIT(fmt, vc) [fmt] = { 1, 1, 1, vc, CHIP_R600 } +#define FMT_16_BIT(fmt, vc) [fmt] = { 1, 1, 2, vc, CHIP_R600 } +#define FMT_24_BIT(fmt) [fmt] = { 1, 1, 3, 0, CHIP_R600 } +#define FMT_32_BIT(fmt, vc) [fmt] = { 1, 1, 4, vc, CHIP_R600 } +#define FMT_48_BIT(fmt) [fmt] = { 1, 1, 6, 0, CHIP_R600 } +#define FMT_64_BIT(fmt, vc) [fmt] = { 1, 1, 8, vc, CHIP_R600 } +#define FMT_96_BIT(fmt) [fmt] = { 1, 1, 12, 0, CHIP_R600 } +#define FMT_128_BIT(fmt, vc) [fmt] = { 1, 1, 16,vc, CHIP_R600 } struct gpu_formats { unsigned blockwidth; unsigned blockheight; unsigned blocksize; unsigned valid_color; + enum radeon_family min_family; }; static const struct gpu_formats color_formats_table[] = { @@ -154,7 +155,11 @@ static const struct gpu_formats color_formats_table[] = { [V_038004_FMT_BC3] = { 4, 4, 16, 0 }, [V_038004_FMT_BC4] = { 4, 4, 8, 0 }, [V_038004_FMT_BC5] = { 4, 4, 16, 0}, + [V_038004_FMT_BC6] = { 4, 4, 16, 0, CHIP_CEDAR}, /* Evergreen-only */ + [V_038004_FMT_BC7] = { 4, 4, 16, 0, CHIP_CEDAR}, /* Evergreen-only */ + /* The other Evergreen formats */ + [V_038004_FMT_32_AS_32_32_32_32] = { 1, 1, 4, 0, CHIP_CEDAR}, }; static inline bool fmt_is_valid_color(u32 format) @@ -168,11 +173,14 @@ static inline bool fmt_is_valid_color(u32 format) return false; } -static inline bool fmt_is_valid_texture(u32 format) +static inline bool fmt_is_valid_texture(u32 format, enum radeon_family family) { if (format >= ARRAY_SIZE(color_formats_table)) return false; + if (family < color_formats_table[format].min_family) + return false; + if (color_formats_table[format].blockwidth > 0) return true; @@ -907,12 +915,11 @@ static inline int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx { struct r600_cs_track *track = (struct r600_cs_track *)p->track; struct radeon_cs_reloc *reloc; - u32 last_reg = ARRAY_SIZE(r600_reg_safe_bm); u32 m, i, tmp, *ib; int r; i = (reg >> 7); - if (i > last_reg) { + if (i >= ARRAY_SIZE(r600_reg_safe_bm)) { dev_warn(p->dev, "forbidden register 0x%08x at %d\n", reg, idx); return -EINVAL; } @@ -1192,6 +1199,15 @@ static inline int r600_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx } ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); break; + case SX_MEMORY_EXPORT_BASE: + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + dev_warn(p->dev, "bad SET_CONFIG_REG " + "0x%04X\n", reg); + return -EINVAL; + } + ib[idx] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + break; default: dev_warn(p->dev, "forbidden register 0x%08x at %d\n", reg, idx); return -EINVAL; @@ -1325,7 +1341,7 @@ static inline int r600_check_texture_resource(struct radeon_cs_parser *p, u32 i return -EINVAL; } format = G_038004_DATA_FORMAT(word1); - if (!fmt_is_valid_texture(format)) { + if (!fmt_is_valid_texture(format, p->family)) { dev_warn(p->dev, "%s:%d texture invalid format %d\n", __func__, __LINE__, format); return -EINVAL; diff --git a/drivers/gpu/drm/radeon/r600d.h b/drivers/gpu/drm/radeon/r600d.h index b2b944bcd05..0245ae6c204 100644 --- a/drivers/gpu/drm/radeon/r600d.h +++ b/drivers/gpu/drm/radeon/r600d.h @@ -536,7 +536,7 @@ #define IH_RB_WPTR_ADDR_LO 0x3e14 #define IH_CNTL 0x3e18 # define ENABLE_INTR (1 << 0) -# define IH_MC_SWAP(x) ((x) << 2) +# define IH_MC_SWAP(x) ((x) << 1) # define IH_MC_SWAP_NONE 0 # define IH_MC_SWAP_16BIT 1 # define IH_MC_SWAP_32BIT 2 @@ -1309,6 +1309,9 @@ #define V_038004_FMT_BC3 0x00000033 #define V_038004_FMT_BC4 0x00000034 #define V_038004_FMT_BC5 0x00000035 +#define V_038004_FMT_BC6 0x00000036 +#define V_038004_FMT_BC7 0x00000037 +#define V_038004_FMT_32_AS_32_32_32_32 0x00000038 #define R_038010_SQ_TEX_RESOURCE_WORD4_0 0x038010 #define S_038010_FORMAT_COMP_X(x) (((x) & 0x3) << 0) #define G_038010_FORMAT_COMP_X(x) (((x) >> 0) & 0x3) diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index ba643b57605..32807baf55e 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -60,7 +60,7 @@ * are considered as fatal) */ -#include <asm/atomic.h> +#include <linux/atomic.h> #include <linux/wait.h> #include <linux/list.h> #include <linux/kref.h> @@ -165,6 +165,7 @@ struct radeon_clock { uint32_t default_sclk; uint32_t default_dispclk; uint32_t dp_extclk; + uint32_t max_pixel_clock; }; /* @@ -178,6 +179,7 @@ void radeon_pm_resume(struct radeon_device *rdev); void radeon_combios_get_power_modes(struct radeon_device *rdev); void radeon_atombios_get_power_modes(struct radeon_device *rdev); void radeon_atom_set_voltage(struct radeon_device *rdev, u16 voltage_level, u8 voltage_type); +int radeon_atom_get_max_vddc(struct radeon_device *rdev, u16 *voltage); void rs690_pm_info(struct radeon_device *rdev); extern int rv6xx_get_temp(struct radeon_device *rdev); extern int rv770_get_temp(struct radeon_device *rdev); @@ -1001,6 +1003,7 @@ struct r600_asic { unsigned tiling_npipes; unsigned tiling_group_size; unsigned tile_config; + unsigned backend_map; struct r100_gpu_lockup lockup; }; @@ -1026,6 +1029,7 @@ struct rv770_asic { unsigned tiling_npipes; unsigned tiling_group_size; unsigned tile_config; + unsigned backend_map; struct r100_gpu_lockup lockup; }; @@ -1052,6 +1056,7 @@ struct evergreen_asic { unsigned tiling_npipes; unsigned tiling_group_size; unsigned tile_config; + unsigned backend_map; struct r100_gpu_lockup lockup; }; @@ -1172,7 +1177,7 @@ struct radeon_device { /* Register mmio */ resource_size_t rmmio_base; resource_size_t rmmio_size; - void *rmmio; + void __iomem *rmmio; radeon_rreg_t mc_rreg; radeon_wreg_t mc_wreg; radeon_rreg_t pll_rreg; @@ -1249,20 +1254,20 @@ int radeon_gpu_wait_for_idle(struct radeon_device *rdev); static inline uint32_t r100_mm_rreg(struct radeon_device *rdev, uint32_t reg) { if (reg < rdev->rmmio_size) - return readl(((void __iomem *)rdev->rmmio) + reg); + return readl((rdev->rmmio) + reg); else { - writel(reg, ((void __iomem *)rdev->rmmio) + RADEON_MM_INDEX); - return readl(((void __iomem *)rdev->rmmio) + RADEON_MM_DATA); + writel(reg, (rdev->rmmio) + RADEON_MM_INDEX); + return readl((rdev->rmmio) + RADEON_MM_DATA); } } static inline void r100_mm_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) { if (reg < rdev->rmmio_size) - writel(v, ((void __iomem *)rdev->rmmio) + reg); + writel(v, (rdev->rmmio) + reg); else { - writel(reg, ((void __iomem *)rdev->rmmio) + RADEON_MM_INDEX); - writel(v, ((void __iomem *)rdev->rmmio) + RADEON_MM_DATA); + writel(reg, (rdev->rmmio) + RADEON_MM_INDEX); + writel(v, (rdev->rmmio) + RADEON_MM_DATA); } } @@ -1294,10 +1299,10 @@ static inline void r100_io_wreg(struct radeon_device *rdev, u32 reg, u32 v) /* * Registers read & write functions. */ -#define RREG8(reg) readb(((void __iomem *)rdev->rmmio) + (reg)) -#define WREG8(reg, v) writeb(v, ((void __iomem *)rdev->rmmio) + (reg)) -#define RREG16(reg) readw(((void __iomem *)rdev->rmmio) + (reg)) -#define WREG16(reg, v) writew(v, ((void __iomem *)rdev->rmmio) + (reg)) +#define RREG8(reg) readb((rdev->rmmio) + (reg)) +#define WREG8(reg, v) writeb(v, (rdev->rmmio) + (reg)) +#define RREG16(reg) readw((rdev->rmmio) + (reg)) +#define WREG16(reg, v) writew(v, (rdev->rmmio) + (reg)) #define RREG32(reg) r100_mm_rreg(rdev, (reg)) #define DREG32(reg) printk(KERN_INFO "REGISTER: " #reg " : 0x%08X\n", r100_mm_rreg(rdev, (reg))) #define WREG32(reg, v) r100_mm_wreg(rdev, (reg), (v)) diff --git a/drivers/gpu/drm/radeon/radeon_asic.c b/drivers/gpu/drm/radeon/radeon_asic.c index ca576191d05..df8218bb83a 100644 --- a/drivers/gpu/drm/radeon/radeon_asic.c +++ b/drivers/gpu/drm/radeon/radeon_asic.c @@ -625,7 +625,7 @@ static struct radeon_asic r600_asic = { .fence_ring_emit = &r600_fence_ring_emit, .cs_parse = &r600_cs_parse, .copy_blit = &r600_copy_blit, - .copy_dma = &r600_copy_blit, + .copy_dma = NULL, .copy = &r600_copy_blit, .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, @@ -672,7 +672,7 @@ static struct radeon_asic rs780_asic = { .fence_ring_emit = &r600_fence_ring_emit, .cs_parse = &r600_cs_parse, .copy_blit = &r600_copy_blit, - .copy_dma = &r600_copy_blit, + .copy_dma = NULL, .copy = &r600_copy_blit, .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, @@ -719,7 +719,7 @@ static struct radeon_asic rv770_asic = { .fence_ring_emit = &r600_fence_ring_emit, .cs_parse = &r600_cs_parse, .copy_blit = &r600_copy_blit, - .copy_dma = &r600_copy_blit, + .copy_dma = NULL, .copy = &r600_copy_blit, .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, @@ -766,7 +766,7 @@ static struct radeon_asic evergreen_asic = { .fence_ring_emit = &r600_fence_ring_emit, .cs_parse = &evergreen_cs_parse, .copy_blit = &evergreen_copy_blit, - .copy_dma = &evergreen_copy_blit, + .copy_dma = NULL, .copy = &evergreen_copy_blit, .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, @@ -782,6 +782,7 @@ static struct radeon_asic evergreen_asic = { .hpd_fini = &evergreen_hpd_fini, .hpd_sense = &evergreen_hpd_sense, .hpd_set_polarity = &evergreen_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, .gui_idle = &r600_gui_idle, .pm_misc = &evergreen_pm_misc, .pm_prepare = &evergreen_pm_prepare, @@ -812,7 +813,7 @@ static struct radeon_asic sumo_asic = { .fence_ring_emit = &r600_fence_ring_emit, .cs_parse = &evergreen_cs_parse, .copy_blit = &evergreen_copy_blit, - .copy_dma = &evergreen_copy_blit, + .copy_dma = NULL, .copy = &evergreen_copy_blit, .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, @@ -828,6 +829,7 @@ static struct radeon_asic sumo_asic = { .hpd_fini = &evergreen_hpd_fini, .hpd_sense = &evergreen_hpd_sense, .hpd_set_polarity = &evergreen_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, .gui_idle = &r600_gui_idle, .pm_misc = &evergreen_pm_misc, .pm_prepare = &evergreen_pm_prepare, @@ -858,7 +860,7 @@ static struct radeon_asic btc_asic = { .fence_ring_emit = &r600_fence_ring_emit, .cs_parse = &evergreen_cs_parse, .copy_blit = &evergreen_copy_blit, - .copy_dma = &evergreen_copy_blit, + .copy_dma = NULL, .copy = &evergreen_copy_blit, .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, @@ -874,6 +876,7 @@ static struct radeon_asic btc_asic = { .hpd_fini = &evergreen_hpd_fini, .hpd_sense = &evergreen_hpd_sense, .hpd_set_polarity = &evergreen_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, .gui_idle = &r600_gui_idle, .pm_misc = &evergreen_pm_misc, .pm_prepare = &evergreen_pm_prepare, @@ -903,9 +906,9 @@ static struct radeon_asic cayman_asic = { .get_vblank_counter = &evergreen_get_vblank_counter, .fence_ring_emit = &r600_fence_ring_emit, .cs_parse = &evergreen_cs_parse, - .copy_blit = NULL, + .copy_blit = &evergreen_copy_blit, .copy_dma = NULL, - .copy = NULL, + .copy = &evergreen_copy_blit, .get_engine_clock = &radeon_atom_get_engine_clock, .set_engine_clock = &radeon_atom_set_engine_clock, .get_memory_clock = &radeon_atom_get_memory_clock, @@ -920,6 +923,7 @@ static struct radeon_asic cayman_asic = { .hpd_fini = &evergreen_hpd_fini, .hpd_sense = &evergreen_hpd_sense, .hpd_set_polarity = &evergreen_hpd_set_polarity, + .ioctl_wait_idle = r600_ioctl_wait_idle, .gui_idle = &r600_gui_idle, .pm_misc = &evergreen_pm_misc, .pm_prepare = &evergreen_pm_prepare, @@ -934,6 +938,13 @@ static struct radeon_asic cayman_asic = { int radeon_asic_init(struct radeon_device *rdev) { radeon_register_accessor_init(rdev); + + /* set the number of crtcs */ + if (rdev->flags & RADEON_SINGLE_CRTC) + rdev->num_crtc = 1; + else + rdev->num_crtc = 2; + switch (rdev->family) { case CHIP_R100: case CHIP_RV100: @@ -1013,18 +1024,32 @@ int radeon_asic_init(struct radeon_device *rdev) case CHIP_JUNIPER: case CHIP_CYPRESS: case CHIP_HEMLOCK: + /* set num crtcs */ + if (rdev->family == CHIP_CEDAR) + rdev->num_crtc = 4; + else + rdev->num_crtc = 6; rdev->asic = &evergreen_asic; break; case CHIP_PALM: + case CHIP_SUMO: + case CHIP_SUMO2: rdev->asic = &sumo_asic; break; case CHIP_BARTS: case CHIP_TURKS: case CHIP_CAICOS: + /* set num crtcs */ + if (rdev->family == CHIP_CAICOS) + rdev->num_crtc = 4; + else + rdev->num_crtc = 6; rdev->asic = &btc_asic; break; case CHIP_CAYMAN: rdev->asic = &cayman_asic; + /* set num crtcs */ + rdev->num_crtc = 6; break; default: /* FIXME: not supported yet */ @@ -1036,18 +1061,6 @@ int radeon_asic_init(struct radeon_device *rdev) rdev->asic->set_memory_clock = NULL; } - /* set the number of crtcs */ - if (rdev->flags & RADEON_SINGLE_CRTC) - rdev->num_crtc = 1; - else { - if (ASIC_IS_DCE41(rdev)) - rdev->num_crtc = 2; - else if (ASIC_IS_DCE4(rdev)) - rdev->num_crtc = 6; - else - rdev->num_crtc = 2; - } - return 0; } diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c index dd881d035f0..bf2b61584cd 100644 --- a/drivers/gpu/drm/radeon/radeon_atombios.c +++ b/drivers/gpu/drm/radeon/radeon_atombios.c @@ -1246,6 +1246,10 @@ bool radeon_atom_get_clock_info(struct drm_device *dev) } *dcpll = *p1pll; + rdev->clock.max_pixel_clock = le16_to_cpu(firmware_info->info.usMaxPixelClock); + if (rdev->clock.max_pixel_clock == 0) + rdev->clock.max_pixel_clock = 40000; + return true; } @@ -1574,9 +1578,17 @@ struct radeon_encoder_atom_dig *radeon_atombios_get_lvds_info(struct ATOM_FAKE_EDID_PATCH_RECORD *fake_edid_record; ATOM_PANEL_RESOLUTION_PATCH_RECORD *panel_res_record; bool bad_record = false; - u8 *record = (u8 *)(mode_info->atom_context->bios + - data_offset + - le16_to_cpu(lvds_info->info.usModePatchTableOffset)); + u8 *record; + + if ((frev == 1) && (crev < 2)) + /* absolute */ + record = (u8 *)(mode_info->atom_context->bios + + le16_to_cpu(lvds_info->info.usModePatchTableOffset)); + else + /* relative */ + record = (u8 *)(mode_info->atom_context->bios + + data_offset + + le16_to_cpu(lvds_info->info.usModePatchTableOffset)); while (*record != ATOM_RECORD_END_TYPE) { switch (*record) { case LCD_MODE_PATCH_RECORD_MODE_TYPE: @@ -2308,6 +2320,14 @@ static bool radeon_atombios_parse_pplib_clock_info(struct radeon_device *rdev, le16_to_cpu(clock_info->r600.usVDDC); } + /* patch up vddc if necessary */ + if (rdev->pm.power_state[state_index].clock_info[mode_index].voltage.voltage == 0xff01) { + u16 vddc; + + if (radeon_atom_get_max_vddc(rdev, &vddc) == 0) + rdev->pm.power_state[state_index].clock_info[mode_index].voltage.voltage = vddc; + } + if (rdev->flags & RADEON_IS_IGP) { /* skip invalid modes */ if (rdev->pm.power_state[state_index].clock_info[mode_index].sclk == 0) @@ -2595,6 +2615,10 @@ void radeon_atom_set_voltage(struct radeon_device *rdev, u16 voltage_level, u8 v if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) return; + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage_level == 0xff01) + return; + switch (crev) { case 1: args.v1.ucVoltageType = voltage_type; @@ -2614,7 +2638,35 @@ void radeon_atom_set_voltage(struct radeon_device *rdev, u16 voltage_level, u8 v atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); } +int radeon_atom_get_max_vddc(struct radeon_device *rdev, + u16 *voltage) +{ + union set_voltage args; + int index = GetIndexIntoMasterTable(COMMAND, SetVoltage); + u8 frev, crev; + if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) + return -EINVAL; + + switch (crev) { + case 1: + return -EINVAL; + case 2: + args.v2.ucVoltageType = SET_VOLTAGE_GET_MAX_VOLTAGE; + args.v2.ucVoltageMode = 0; + args.v2.usVoltageLevel = 0; + + atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + + *voltage = le16_to_cpu(args.v2.usVoltageLevel); + break; + default: + DRM_ERROR("Unknown table version %d, %d\n", frev, crev); + return -EINVAL; + } + + return 0; +} void radeon_atom_initialize_bios_scratch_regs(struct drm_device *dev) { diff --git a/drivers/gpu/drm/radeon/radeon_bios.c b/drivers/gpu/drm/radeon/radeon_bios.c index 1aba85cad1a..229a20f10e2 100644 --- a/drivers/gpu/drm/radeon/radeon_bios.c +++ b/drivers/gpu/drm/radeon/radeon_bios.c @@ -104,7 +104,7 @@ static bool radeon_read_bios(struct radeon_device *rdev) static bool radeon_atrm_get_bios(struct radeon_device *rdev) { int ret; - int size = 64 * 1024; + int size = 256 * 1024; int i; if (!radeon_atrm_supported(rdev->pdev)) @@ -331,7 +331,7 @@ static bool avivo_read_disabled_bios(struct radeon_device *rdev) seprom_cntl1 = RREG32(RADEON_SEPROM_CNTL1); viph_control = RREG32(RADEON_VIPH_CONTROL); - bus_cntl = RREG32(RADEON_BUS_CNTL); + bus_cntl = RREG32(RV370_BUS_CNTL); d1vga_control = RREG32(AVIVO_D1VGA_CONTROL); d2vga_control = RREG32(AVIVO_D2VGA_CONTROL); vga_render_control = RREG32(AVIVO_VGA_RENDER_CONTROL); @@ -350,7 +350,7 @@ static bool avivo_read_disabled_bios(struct radeon_device *rdev) WREG32(RADEON_VIPH_CONTROL, (viph_control & ~RADEON_VIPH_EN)); /* enable the rom */ - WREG32(RADEON_BUS_CNTL, (bus_cntl & ~RADEON_BUS_BIOS_DIS_ROM)); + WREG32(RV370_BUS_CNTL, (bus_cntl & ~RV370_BUS_BIOS_DIS_ROM)); /* Disable VGA mode */ WREG32(AVIVO_D1VGA_CONTROL, @@ -367,7 +367,7 @@ static bool avivo_read_disabled_bios(struct radeon_device *rdev) /* restore regs */ WREG32(RADEON_SEPROM_CNTL1, seprom_cntl1); WREG32(RADEON_VIPH_CONTROL, viph_control); - WREG32(RADEON_BUS_CNTL, bus_cntl); + WREG32(RV370_BUS_CNTL, bus_cntl); WREG32(AVIVO_D1VGA_CONTROL, d1vga_control); WREG32(AVIVO_D2VGA_CONTROL, d2vga_control); WREG32(AVIVO_VGA_RENDER_CONTROL, vga_render_control); @@ -390,7 +390,10 @@ static bool legacy_read_disabled_bios(struct radeon_device *rdev) seprom_cntl1 = RREG32(RADEON_SEPROM_CNTL1); viph_control = RREG32(RADEON_VIPH_CONTROL); - bus_cntl = RREG32(RADEON_BUS_CNTL); + if (rdev->flags & RADEON_IS_PCIE) + bus_cntl = RREG32(RV370_BUS_CNTL); + else + bus_cntl = RREG32(RADEON_BUS_CNTL); crtc_gen_cntl = RREG32(RADEON_CRTC_GEN_CNTL); crtc2_gen_cntl = 0; crtc_ext_cntl = RREG32(RADEON_CRTC_EXT_CNTL); @@ -412,7 +415,10 @@ static bool legacy_read_disabled_bios(struct radeon_device *rdev) WREG32(RADEON_VIPH_CONTROL, (viph_control & ~RADEON_VIPH_EN)); /* enable the rom */ - WREG32(RADEON_BUS_CNTL, (bus_cntl & ~RADEON_BUS_BIOS_DIS_ROM)); + if (rdev->flags & RADEON_IS_PCIE) + WREG32(RV370_BUS_CNTL, (bus_cntl & ~RV370_BUS_BIOS_DIS_ROM)); + else + WREG32(RADEON_BUS_CNTL, (bus_cntl & ~RADEON_BUS_BIOS_DIS_ROM)); /* Turn off mem requests and CRTC for both controllers */ WREG32(RADEON_CRTC_GEN_CNTL, @@ -439,7 +445,10 @@ static bool legacy_read_disabled_bios(struct radeon_device *rdev) /* restore regs */ WREG32(RADEON_SEPROM_CNTL1, seprom_cntl1); WREG32(RADEON_VIPH_CONTROL, viph_control); - WREG32(RADEON_BUS_CNTL, bus_cntl); + if (rdev->flags & RADEON_IS_PCIE) + WREG32(RV370_BUS_CNTL, bus_cntl); + else + WREG32(RADEON_BUS_CNTL, bus_cntl); WREG32(RADEON_CRTC_GEN_CNTL, crtc_gen_cntl); if (!(rdev->flags & RADEON_SINGLE_CRTC)) { WREG32(RADEON_CRTC2_GEN_CNTL, crtc2_gen_cntl); diff --git a/drivers/gpu/drm/radeon/radeon_clocks.c b/drivers/gpu/drm/radeon/radeon_clocks.c index 5249af8931e..dcd0863e31a 100644 --- a/drivers/gpu/drm/radeon/radeon_clocks.c +++ b/drivers/gpu/drm/radeon/radeon_clocks.c @@ -96,7 +96,7 @@ uint32_t radeon_legacy_get_memory_clock(struct radeon_device *rdev) * Read XTAL (ref clock), SCLK and MCLK from Open Firmware device * tree. Hopefully, ATI OF driver is kind enough to fill these */ -static bool __devinit radeon_read_clocks_OF(struct drm_device *dev) +static bool radeon_read_clocks_OF(struct drm_device *dev) { struct radeon_device *rdev = dev->dev_private; struct device_node *dp = rdev->pdev->dev.of_node; @@ -117,7 +117,7 @@ static bool __devinit radeon_read_clocks_OF(struct drm_device *dev) p1pll->reference_div = RREG32_PLL(RADEON_PPLL_REF_DIV) & 0x3ff; if (p1pll->reference_div < 2) p1pll->reference_div = 12; - p2pll->reference_div = p1pll->reference_div; + p2pll->reference_div = p1pll->reference_div; /* These aren't in the device-tree */ if (rdev->family >= CHIP_R420) { @@ -139,6 +139,8 @@ static bool __devinit radeon_read_clocks_OF(struct drm_device *dev) p2pll->pll_out_min = 12500; p2pll->pll_out_max = 35000; } + /* not sure what the max should be in all cases */ + rdev->clock.max_pixel_clock = 35000; spll->reference_freq = mpll->reference_freq = p1pll->reference_freq; spll->reference_div = mpll->reference_div = @@ -151,7 +153,7 @@ static bool __devinit radeon_read_clocks_OF(struct drm_device *dev) else rdev->clock.default_sclk = radeon_legacy_get_engine_clock(rdev); - + val = of_get_property(dp, "ATY,MCLK", NULL); if (val && *val) rdev->clock.default_mclk = (*val) / 10; @@ -160,11 +162,11 @@ static bool __devinit radeon_read_clocks_OF(struct drm_device *dev) radeon_legacy_get_memory_clock(rdev); DRM_INFO("Using device-tree clock info\n"); - + return true; } #else -static bool __devinit radeon_read_clocks_OF(struct drm_device *dev) +static bool radeon_read_clocks_OF(struct drm_device *dev) { return false; } diff --git a/drivers/gpu/drm/radeon/radeon_combios.c b/drivers/gpu/drm/radeon/radeon_combios.c index 8caf546c8e9..e0138b674ac 100644 --- a/drivers/gpu/drm/radeon/radeon_combios.c +++ b/drivers/gpu/drm/radeon/radeon_combios.c @@ -505,12 +505,18 @@ static struct radeon_i2c_bus_rec combios_setup_i2c_bus(struct radeon_device *rde * DDC_VGA = RADEON_GPIO_VGA_DDC * DDC_LCD = RADEON_GPIOPAD_MASK * DDC_GPIO = RADEON_MDGPIO_MASK - * r1xx/r2xx + * r1xx * DDC_MONID = RADEON_GPIO_MONID * DDC_CRT2 = RADEON_GPIO_CRT2_DDC - * r3xx + * r200 * DDC_MONID = RADEON_GPIO_MONID * DDC_CRT2 = RADEON_GPIO_DVI_DDC + * r300/r350 + * DDC_MONID = RADEON_GPIO_DVI_DDC + * DDC_CRT2 = RADEON_GPIO_DVI_DDC + * rv2xx/rv3xx + * DDC_MONID = RADEON_GPIO_MONID + * DDC_CRT2 = RADEON_GPIO_MONID * rs3xx/rs4xx * DDC_MONID = RADEON_GPIOPAD_MASK * DDC_CRT2 = RADEON_GPIO_MONID @@ -537,17 +543,26 @@ static struct radeon_i2c_bus_rec combios_setup_i2c_bus(struct radeon_device *rde rdev->family == CHIP_RS400 || rdev->family == CHIP_RS480) ddc_line = RADEON_GPIOPAD_MASK; - else + else if (rdev->family == CHIP_R300 || + rdev->family == CHIP_R350) { + ddc_line = RADEON_GPIO_DVI_DDC; + ddc = DDC_DVI; + } else ddc_line = RADEON_GPIO_MONID; break; case DDC_CRT2: - if (rdev->family == CHIP_RS300 || - rdev->family == CHIP_RS400 || - rdev->family == CHIP_RS480) - ddc_line = RADEON_GPIO_MONID; - else if (rdev->family >= CHIP_R300) { + if (rdev->family == CHIP_R200 || + rdev->family == CHIP_R300 || + rdev->family == CHIP_R350) { ddc_line = RADEON_GPIO_DVI_DDC; ddc = DDC_DVI; + } else if (rdev->family == CHIP_RS300 || + rdev->family == CHIP_RS400 || + rdev->family == CHIP_RS480) + ddc_line = RADEON_GPIO_MONID; + else if (rdev->family >= CHIP_RV350) { + ddc_line = RADEON_GPIO_MONID; + ddc = DDC_MONID; } else ddc_line = RADEON_GPIO_CRT2_DDC; break; @@ -709,26 +724,42 @@ void radeon_combios_i2c_init(struct radeon_device *rdev) struct drm_device *dev = rdev->ddev; struct radeon_i2c_bus_rec i2c; + /* actual hw pads + * r1xx/rs2xx/rs3xx + * 0x60, 0x64, 0x68, 0x6c, gpiopads, mm + * r200 + * 0x60, 0x64, 0x68, mm + * r300/r350 + * 0x60, 0x64, mm + * rv2xx/rv3xx/rs4xx + * 0x60, 0x64, 0x68, gpiopads, mm + */ + /* 0x60 */ i2c = combios_setup_i2c_bus(rdev, DDC_DVI, 0, 0); rdev->i2c_bus[0] = radeon_i2c_create(dev, &i2c, "DVI_DDC"); - + /* 0x64 */ i2c = combios_setup_i2c_bus(rdev, DDC_VGA, 0, 0); rdev->i2c_bus[1] = radeon_i2c_create(dev, &i2c, "VGA_DDC"); + /* mm i2c */ i2c.valid = true; i2c.hw_capable = true; i2c.mm_i2c = true; i2c.i2c_id = 0xa0; rdev->i2c_bus[2] = radeon_i2c_create(dev, &i2c, "MM_I2C"); - if (rdev->family == CHIP_RS300 || - rdev->family == CHIP_RS400 || - rdev->family == CHIP_RS480) { + if (rdev->family == CHIP_R300 || + rdev->family == CHIP_R350) { + /* only 2 sw i2c pads */ + } else if (rdev->family == CHIP_RS300 || + rdev->family == CHIP_RS400 || + rdev->family == CHIP_RS480) { u16 offset; u8 id, blocks, clk, data; int i; + /* 0x68 */ i2c = combios_setup_i2c_bus(rdev, DDC_CRT2, 0, 0); rdev->i2c_bus[3] = radeon_i2c_create(dev, &i2c, "MONID"); @@ -740,6 +771,7 @@ void radeon_combios_i2c_init(struct radeon_device *rdev) if (id == 136) { clk = RBIOS8(offset + 3 + (i * 5) + 3); data = RBIOS8(offset + 3 + (i * 5) + 4); + /* gpiopad */ i2c = combios_setup_i2c_bus(rdev, DDC_MONID, (1 << clk), (1 << data)); rdev->i2c_bus[4] = radeon_i2c_create(dev, &i2c, "GPIOPAD_MASK"); @@ -747,14 +779,16 @@ void radeon_combios_i2c_init(struct radeon_device *rdev) } } } - - } else if (rdev->family >= CHIP_R300) { + } else if ((rdev->family == CHIP_R200) || + (rdev->family >= CHIP_R300)) { + /* 0x68 */ i2c = combios_setup_i2c_bus(rdev, DDC_MONID, 0, 0); rdev->i2c_bus[3] = radeon_i2c_create(dev, &i2c, "MONID"); } else { + /* 0x68 */ i2c = combios_setup_i2c_bus(rdev, DDC_MONID, 0, 0); rdev->i2c_bus[3] = radeon_i2c_create(dev, &i2c, "MONID"); - + /* 0x6c */ i2c = combios_setup_i2c_bus(rdev, DDC_CRT2, 0, 0); rdev->i2c_bus[4] = radeon_i2c_create(dev, &i2c, "CRT2_DDC"); } @@ -833,6 +867,11 @@ bool radeon_combios_get_clock_info(struct drm_device *dev) rdev->clock.default_sclk = sclk; rdev->clock.default_mclk = mclk; + if (RBIOS32(pll_info + 0x16)) + rdev->clock.max_pixel_clock = RBIOS32(pll_info + 0x16); + else + rdev->clock.max_pixel_clock = 35000; /* might need something asic specific */ + return true; } return false; @@ -1515,10 +1554,12 @@ bool radeon_get_legacy_connector_info_from_table(struct drm_device *dev) (rdev->pdev->subsystem_device == 0x4a48)) { /* Mac X800 */ rdev->mode_info.connector_table = CT_MAC_X800; - } else if ((rdev->pdev->device == 0x4150) && + } else if ((of_machine_is_compatible("PowerMac7,2") || + of_machine_is_compatible("PowerMac7,3")) && + (rdev->pdev->device == 0x4150) && (rdev->pdev->subsystem_vendor == 0x1002) && (rdev->pdev->subsystem_device == 0x4150)) { - /* Mac G5 9600 */ + /* Mac G5 tower 9600 */ rdev->mode_info.connector_table = CT_MAC_G5_9600; } else #endif /* CONFIG_PPC_PMAC */ @@ -2504,12 +2545,19 @@ bool radeon_get_legacy_connector_info_from_bios(struct drm_device *dev) return true; } +static const char *thermal_controller_names[] = { + "NONE", + "lm63", + "adm1032", +}; + void radeon_combios_get_power_modes(struct radeon_device *rdev) { struct drm_device *dev = rdev->ddev; u16 offset, misc, misc2 = 0; u8 rev, blocks, tmp; int state_index = 0; + struct radeon_i2c_bus_rec i2c_bus; rdev->pm.default_power_state_index = -1; @@ -2524,6 +2572,72 @@ void radeon_combios_get_power_modes(struct radeon_device *rdev) return; } + /* check for a thermal chip */ + offset = combios_get_table_offset(dev, COMBIOS_OVERDRIVE_INFO_TABLE); + if (offset) { + u8 thermal_controller = 0, gpio = 0, i2c_addr = 0, clk_bit = 0, data_bit = 0; + + rev = RBIOS8(offset); + + if (rev == 0) { + thermal_controller = RBIOS8(offset + 3); + gpio = RBIOS8(offset + 4) & 0x3f; + i2c_addr = RBIOS8(offset + 5); + } else if (rev == 1) { + thermal_controller = RBIOS8(offset + 4); + gpio = RBIOS8(offset + 5) & 0x3f; + i2c_addr = RBIOS8(offset + 6); + } else if (rev == 2) { + thermal_controller = RBIOS8(offset + 4); + gpio = RBIOS8(offset + 5) & 0x3f; + i2c_addr = RBIOS8(offset + 6); + clk_bit = RBIOS8(offset + 0xa); + data_bit = RBIOS8(offset + 0xb); + } + if ((thermal_controller > 0) && (thermal_controller < 3)) { + DRM_INFO("Possible %s thermal controller at 0x%02x\n", + thermal_controller_names[thermal_controller], + i2c_addr >> 1); + if (gpio == DDC_LCD) { + /* MM i2c */ + i2c_bus.valid = true; + i2c_bus.hw_capable = true; + i2c_bus.mm_i2c = true; + i2c_bus.i2c_id = 0xa0; + } else if (gpio == DDC_GPIO) + i2c_bus = combios_setup_i2c_bus(rdev, gpio, 1 << clk_bit, 1 << data_bit); + else + i2c_bus = combios_setup_i2c_bus(rdev, gpio, 0, 0); + rdev->pm.i2c_bus = radeon_i2c_lookup(rdev, &i2c_bus); + if (rdev->pm.i2c_bus) { + struct i2c_board_info info = { }; + const char *name = thermal_controller_names[thermal_controller]; + info.addr = i2c_addr >> 1; + strlcpy(info.type, name, sizeof(info.type)); + i2c_new_device(&rdev->pm.i2c_bus->adapter, &info); + } + } + } else { + /* boards with a thermal chip, but no overdrive table */ + + /* Asus 9600xt has an f75375 on the monid bus */ + if ((dev->pdev->device == 0x4152) && + (dev->pdev->subsystem_vendor == 0x1043) && + (dev->pdev->subsystem_device == 0xc002)) { + i2c_bus = combios_setup_i2c_bus(rdev, DDC_MONID, 0, 0); + rdev->pm.i2c_bus = radeon_i2c_lookup(rdev, &i2c_bus); + if (rdev->pm.i2c_bus) { + struct i2c_board_info info = { }; + const char *name = "f75375"; + info.addr = 0x28; + strlcpy(info.type, name, sizeof(info.type)); + i2c_new_device(&rdev->pm.i2c_bus->adapter, &info); + DRM_INFO("Possible %s thermal controller at 0x%02x\n", + name, info.addr); + } + } + } + if (rdev->flags & RADEON_IS_MOBILITY) { offset = combios_get_table_offset(dev, COMBIOS_POWERPLAY_INFO_TABLE); if (offset) { diff --git a/drivers/gpu/drm/radeon/radeon_connectors.c b/drivers/gpu/drm/radeon/radeon_connectors.c index 5f45fa12bb8..4f0c1ecac72 100644 --- a/drivers/gpu/drm/radeon/radeon_connectors.c +++ b/drivers/gpu/drm/radeon/radeon_connectors.c @@ -44,26 +44,37 @@ extern void radeon_legacy_backlight_init(struct radeon_encoder *radeon_encoder, struct drm_connector *drm_connector); +bool radeon_connector_encoder_is_dp_bridge(struct drm_connector *connector); + void radeon_connector_hotplug(struct drm_connector *connector) { struct drm_device *dev = connector->dev; struct radeon_device *rdev = dev->dev_private; struct radeon_connector *radeon_connector = to_radeon_connector(connector); - if (radeon_connector->hpd.hpd != RADEON_HPD_NONE) - radeon_hpd_set_polarity(rdev, radeon_connector->hpd.hpd); + /* bail if the connector does not have hpd pin, e.g., + * VGA, TV, etc. + */ + if (radeon_connector->hpd.hpd == RADEON_HPD_NONE) + return; - if ((connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort) || - (connector->connector_type == DRM_MODE_CONNECTOR_eDP)) { - if ((radeon_dp_getsinktype(radeon_connector) == CONNECTOR_OBJECT_ID_DISPLAYPORT) || - (radeon_dp_getsinktype(radeon_connector) == CONNECTOR_OBJECT_ID_eDP)) { - if (radeon_dp_needs_link_train(radeon_connector)) { - if (connector->encoder) - dp_link_train(connector->encoder, connector); - } - } - } + radeon_hpd_set_polarity(rdev, radeon_connector->hpd.hpd); + /* if the connector is already off, don't turn it back on */ + if (connector->dpms != DRM_MODE_DPMS_ON) + return; + + /* just deal with DP (not eDP) here. */ + if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort) { + int saved_dpms = connector->dpms; + + if (radeon_hpd_sense(rdev, radeon_connector->hpd.hpd) && + radeon_dp_needs_link_train(radeon_connector)) + drm_helper_connector_dpms(connector, DRM_MODE_DPMS_ON); + else + drm_helper_connector_dpms(connector, DRM_MODE_DPMS_OFF); + connector->dpms = saved_dpms; + } } static void radeon_property_change_mode(struct drm_encoder *encoder) @@ -421,16 +432,73 @@ int radeon_connector_set_property(struct drm_connector *connector, struct drm_pr return 0; } +/* + * Some integrated ATI Radeon chipset implementations (e. g. + * Asus M2A-VM HDMI) may indicate the availability of a DDC, + * even when there's no monitor connected. For these connectors + * following DDC probe extension will be applied: check also for the + * availability of EDID with at least a correct EDID header. Only then, + * DDC is assumed to be available. This prevents drm_get_edid() and + * drm_edid_block_valid() from periodically dumping data and kernel + * errors into the logs and onto the terminal. + */ +static bool radeon_connector_needs_extended_probe(struct radeon_device *dev, + uint32_t supported_device, + int connector_type) +{ + /* Asus M2A-VM HDMI board sends data to i2c bus even, + * if HDMI add-on card is not plugged in or HDMI is disabled in + * BIOS. Valid DDC can only be assumed, if also a valid EDID header + * can be retrieved via i2c bus during DDC probe */ + if ((dev->pdev->device == 0x791e) && + (dev->pdev->subsystem_vendor == 0x1043) && + (dev->pdev->subsystem_device == 0x826d)) { + if ((connector_type == DRM_MODE_CONNECTOR_HDMIA) && + (supported_device == ATOM_DEVICE_DFP2_SUPPORT)) + return true; + } + /* ECS A740GM-M with ATI RADEON 2100 sends data to i2c bus + * for a DVI connector that is not implemented */ + if ((dev->pdev->device == 0x796e) && + (dev->pdev->subsystem_vendor == 0x1019) && + (dev->pdev->subsystem_device == 0x2615)) { + if ((connector_type == DRM_MODE_CONNECTOR_DVID) && + (supported_device == ATOM_DEVICE_DFP2_SUPPORT)) + return true; + } + /* TOSHIBA Satellite L300D with ATI Mobility Radeon x1100 + * (RS690M) sends data to i2c bus for a HDMI connector that + * is not implemented */ + if ((dev->pdev->device == 0x791f) && + (dev->pdev->subsystem_vendor == 0x1179) && + (dev->pdev->subsystem_device == 0xff68)) { + if ((connector_type == DRM_MODE_CONNECTOR_HDMIA) && + (supported_device == ATOM_DEVICE_DFP2_SUPPORT)) + return true; + } + + /* Default: no EDID header probe required for DDC probing */ + return false; +} + static void radeon_fixup_lvds_native_mode(struct drm_encoder *encoder, struct drm_connector *connector) { struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + struct drm_display_mode *t, *mode; + + /* If the EDID preferred mode doesn't match the native mode, use it */ + list_for_each_entry_safe(mode, t, &connector->probed_modes, head) { + if (mode->type & DRM_MODE_TYPE_PREFERRED) { + if (mode->hdisplay != native_mode->hdisplay || + mode->vdisplay != native_mode->vdisplay) + memcpy(native_mode, mode, sizeof(*mode)); + } + } /* Try to get native mode details from EDID if necessary */ if (!native_mode->clock) { - struct drm_display_mode *t, *mode; - list_for_each_entry_safe(mode, t, &connector->probed_modes, head) { if (mode->hdisplay == native_mode->hdisplay && mode->vdisplay == native_mode->vdisplay) { @@ -441,6 +509,7 @@ static void radeon_fixup_lvds_native_mode(struct drm_encoder *encoder, } } } + if (!native_mode->clock) { DRM_DEBUG_KMS("No LVDS native mode details, disabling RMX\n"); radeon_encoder->rmx_type = RMX_OFF; @@ -625,8 +694,14 @@ static int radeon_vga_get_modes(struct drm_connector *connector) static int radeon_vga_mode_valid(struct drm_connector *connector, struct drm_display_mode *mode) { + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + /* XXX check mode bandwidth */ - /* XXX verify against max DAC output frequency */ + + if ((mode->clock / 10) > rdev->clock.max_pixel_clock) + return MODE_CLOCK_HIGH; + return MODE_OK; } @@ -646,7 +721,8 @@ radeon_vga_detect(struct drm_connector *connector, bool force) ret = connector_status_disconnected; if (radeon_connector->ddc_bus) - dret = radeon_ddc_probe(radeon_connector); + dret = radeon_ddc_probe(radeon_connector, + radeon_connector->requires_extended_probe); if (dret) { if (radeon_connector->edid) { kfree(radeon_connector->edid); @@ -818,7 +894,8 @@ radeon_dvi_detect(struct drm_connector *connector, bool force) bool dret = false; if (radeon_connector->ddc_bus) - dret = radeon_ddc_probe(radeon_connector); + dret = radeon_ddc_probe(radeon_connector, + radeon_connector->requires_extended_probe); if (dret) { if (radeon_connector->edid) { kfree(radeon_connector->edid); @@ -829,6 +906,13 @@ radeon_dvi_detect(struct drm_connector *connector, bool force) if (!radeon_connector->edid) { DRM_ERROR("%s: probed a monitor but no|invalid EDID\n", drm_get_connector_name(connector)); + /* rs690 seems to have a problem with connectors not existing and always + * return a block of 0's. If we see this just stop polling on this output */ + if ((rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740) && radeon_connector->base.null_edid_counter) { + ret = connector_status_disconnected; + DRM_ERROR("%s: detected RS690 floating bus bug, stopping ddc detect\n", drm_get_connector_name(connector)); + radeon_connector->ddc_bus = NULL; + } } else { radeon_connector->use_digital = !!(radeon_connector->edid->input & DRM_EDID_INPUT_DIGITAL); @@ -1014,6 +1098,11 @@ static int radeon_dvi_mode_valid(struct drm_connector *connector, } else return MODE_CLOCK_HIGH; } + + /* check against the max pixel clock */ + if ((mode->clock / 10) > rdev->clock.max_pixel_clock) + return MODE_CLOCK_HIGH; + return MODE_OK; } @@ -1051,36 +1140,153 @@ static int radeon_dp_get_modes(struct drm_connector *connector) { struct radeon_connector *radeon_connector = to_radeon_connector(connector); struct radeon_connector_atom_dig *radeon_dig_connector = radeon_connector->con_priv; + struct drm_encoder *encoder = radeon_best_single_encoder(connector); int ret; - if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) { + if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) || + (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) { + struct drm_display_mode *mode; + if (!radeon_dig_connector->edp_on) atombios_set_edp_panel_power(connector, ATOM_TRANSMITTER_ACTION_POWER_ON); - } - ret = radeon_ddc_get_modes(radeon_connector); - if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) { + ret = radeon_ddc_get_modes(radeon_connector); if (!radeon_dig_connector->edp_on) atombios_set_edp_panel_power(connector, ATOM_TRANSMITTER_ACTION_POWER_OFF); + + if (ret > 0) { + if (encoder) { + radeon_fixup_lvds_native_mode(encoder, connector); + /* add scaled modes */ + radeon_add_common_modes(encoder, connector); + } + return ret; + } + + encoder = radeon_best_single_encoder(connector); + if (!encoder) + return 0; + + /* we have no EDID modes */ + mode = radeon_fp_native_mode(encoder); + if (mode) { + ret = 1; + drm_mode_probed_add(connector, mode); + /* add the width/height from vbios tables if available */ + connector->display_info.width_mm = mode->width_mm; + connector->display_info.height_mm = mode->height_mm; + /* add scaled modes */ + radeon_add_common_modes(encoder, connector); + } + } else { + /* need to setup ddc on the bridge */ + if (radeon_connector_encoder_is_dp_bridge(connector)) { + if (encoder) + radeon_atom_ext_encoder_setup_ddc(encoder); + } + ret = radeon_ddc_get_modes(radeon_connector); } return ret; } +bool radeon_connector_encoder_is_dp_bridge(struct drm_connector *connector) +{ + struct drm_mode_object *obj; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + int i; + bool found = false; + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) + break; + + obj = drm_mode_object_find(connector->dev, connector->encoder_ids[i], DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + + encoder = obj_to_encoder(obj); + radeon_encoder = to_radeon_encoder(encoder); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_TRAVIS: + case ENCODER_OBJECT_ID_NUTMEG: + found = true; + break; + default: + break; + } + } + + return found; +} + +bool radeon_connector_encoder_is_hbr2(struct drm_connector *connector) +{ + struct drm_mode_object *obj; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + int i; + bool found = false; + + for (i = 0; i < DRM_CONNECTOR_MAX_ENCODER; i++) { + if (connector->encoder_ids[i] == 0) + break; + + obj = drm_mode_object_find(connector->dev, connector->encoder_ids[i], DRM_MODE_OBJECT_ENCODER); + if (!obj) + continue; + + encoder = obj_to_encoder(obj); + radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->caps & ATOM_ENCODER_CAP_RECORD_HBR2) + found = true; + } + + return found; +} + +bool radeon_connector_is_dp12_capable(struct drm_connector *connector) +{ + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; + + if (ASIC_IS_DCE5(rdev) && + (rdev->clock.dp_extclk >= 53900) && + radeon_connector_encoder_is_hbr2(connector)) { + return true; + } + + return false; +} + static enum drm_connector_status radeon_dp_detect(struct drm_connector *connector, bool force) { + struct drm_device *dev = connector->dev; + struct radeon_device *rdev = dev->dev_private; struct radeon_connector *radeon_connector = to_radeon_connector(connector); enum drm_connector_status ret = connector_status_disconnected; struct radeon_connector_atom_dig *radeon_dig_connector = radeon_connector->con_priv; + struct drm_encoder *encoder = radeon_best_single_encoder(connector); if (radeon_connector->edid) { kfree(radeon_connector->edid); radeon_connector->edid = NULL; } - if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) { + if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) || + (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) { + if (encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + /* check if panel is valid */ + if (native_mode->hdisplay >= 320 && native_mode->vdisplay >= 240) + ret = connector_status_connected; + } /* eDP is always DP */ radeon_dig_connector->dp_sink_type = CONNECTOR_OBJECT_ID_DISPLAYPORT; if (!radeon_dig_connector->edp_on) @@ -1092,13 +1298,35 @@ radeon_dp_detect(struct drm_connector *connector, bool force) atombios_set_edp_panel_power(connector, ATOM_TRANSMITTER_ACTION_POWER_OFF); } else { + /* need to setup ddc on the bridge */ + if (radeon_connector_encoder_is_dp_bridge(connector)) { + if (encoder) + radeon_atom_ext_encoder_setup_ddc(encoder); + } radeon_dig_connector->dp_sink_type = radeon_dp_getsinktype(radeon_connector); - if (radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) { - if (radeon_dp_getdpcd(radeon_connector)) - ret = connector_status_connected; + if (radeon_hpd_sense(rdev, radeon_connector->hpd.hpd)) { + ret = connector_status_connected; + if (radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) + radeon_dp_getdpcd(radeon_connector); } else { - if (radeon_ddc_probe(radeon_connector)) - ret = connector_status_connected; + if (radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) { + if (radeon_dp_getdpcd(radeon_connector)) + ret = connector_status_connected; + } else { + if (radeon_ddc_probe(radeon_connector, + radeon_connector->requires_extended_probe)) + ret = connector_status_connected; + } + } + + if ((ret == connector_status_disconnected) && + radeon_connector->dac_load_detect) { + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + struct drm_encoder_helper_funcs *encoder_funcs; + if (encoder) { + encoder_funcs = encoder->helper_private; + ret = encoder_funcs->detect(encoder, connector); + } } } @@ -1114,11 +1342,39 @@ static int radeon_dp_mode_valid(struct drm_connector *connector, /* XXX check mode bandwidth */ - if ((radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) || - (radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) - return radeon_dp_mode_valid_helper(radeon_connector, mode); - else + if ((connector->connector_type == DRM_MODE_CONNECTOR_eDP) || + (connector->connector_type == DRM_MODE_CONNECTOR_LVDS)) { + struct drm_encoder *encoder = radeon_best_single_encoder(connector); + + if ((mode->hdisplay < 320) || (mode->vdisplay < 240)) + return MODE_PANEL; + + if (encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_display_mode *native_mode = &radeon_encoder->native_mode; + + /* AVIVO hardware supports downscaling modes larger than the panel + * to the panel size, but I'm not sure this is desirable. + */ + if ((mode->hdisplay > native_mode->hdisplay) || + (mode->vdisplay > native_mode->vdisplay)) + return MODE_PANEL; + + /* if scaling is disabled, block non-native modes */ + if (radeon_encoder->rmx_type == RMX_OFF) { + if ((mode->hdisplay != native_mode->hdisplay) || + (mode->vdisplay != native_mode->vdisplay)) + return MODE_PANEL; + } + } return MODE_OK; + } else { + if ((radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) || + (radeon_dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) + return radeon_dp_mode_valid_helper(connector, mode); + else + return MODE_OK; + } } struct drm_connector_helper_funcs radeon_dp_connector_helper_funcs = { @@ -1151,8 +1407,11 @@ radeon_add_atom_connector(struct drm_device *dev, struct drm_connector *connector; struct radeon_connector *radeon_connector; struct radeon_connector_atom_dig *radeon_dig_connector; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; uint32_t subpixel_order = SubPixelNone; bool shared_ddc = false; + bool is_dp_bridge = false; if (connector_type == DRM_MODE_CONNECTOR_Unknown) return; @@ -1184,6 +1443,21 @@ radeon_add_atom_connector(struct drm_device *dev, } } + /* check if it's a dp bridge */ + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + radeon_encoder = to_radeon_encoder(encoder); + if (radeon_encoder->devices & supported_device) { + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_TRAVIS: + case ENCODER_OBJECT_ID_NUTMEG: + is_dp_bridge = true; + break; + default: + break; + } + } + } + radeon_connector = kzalloc(sizeof(struct radeon_connector), GFP_KERNEL); if (!radeon_connector) return; @@ -1195,67 +1469,52 @@ radeon_add_atom_connector(struct drm_device *dev, radeon_connector->shared_ddc = shared_ddc; radeon_connector->connector_object_id = connector_object_id; radeon_connector->hpd = *hpd; + radeon_connector->requires_extended_probe = + radeon_connector_needs_extended_probe(rdev, supported_device, + connector_type); radeon_connector->router = *router; if (router->ddc_valid || router->cd_valid) { radeon_connector->router_bus = radeon_i2c_lookup(rdev, &router->i2c_info); if (!radeon_connector->router_bus) DRM_ERROR("Failed to assign router i2c bus! Check dmesg for i2c errors.\n"); } - switch (connector_type) { - case DRM_MODE_CONNECTOR_VGA: - drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); - if (i2c_bus->valid) { - radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); - if (!radeon_connector->ddc_bus) - DRM_ERROR("VGA: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); - } - radeon_connector->dac_load_detect = true; - drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.load_detect_property, - 1); - /* no HPD on analog connectors */ - radeon_connector->hpd.hpd = RADEON_HPD_NONE; - connector->polled = DRM_CONNECTOR_POLL_CONNECT; - connector->interlace_allowed = true; - connector->doublescan_allowed = true; - break; - case DRM_MODE_CONNECTOR_DVIA: - drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); - if (i2c_bus->valid) { - radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); - if (!radeon_connector->ddc_bus) - DRM_ERROR("DVIA: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); - } - radeon_connector->dac_load_detect = true; - drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.load_detect_property, - 1); - /* no HPD on analog connectors */ - radeon_connector->hpd.hpd = RADEON_HPD_NONE; - connector->interlace_allowed = true; - connector->doublescan_allowed = true; - break; - case DRM_MODE_CONNECTOR_DVII: - case DRM_MODE_CONNECTOR_DVID: + + if (is_dp_bridge) { radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); if (!radeon_dig_connector) goto failed; radeon_dig_connector->igp_lane_info = igp_lane_info; radeon_connector->con_priv = radeon_dig_connector; - drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + drm_connector_init(dev, &radeon_connector->base, &radeon_dp_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dp_connector_helper_funcs); if (i2c_bus->valid) { + /* add DP i2c bus */ + if (connector_type == DRM_MODE_CONNECTOR_eDP) + radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "eDP-auxch"); + else + radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "DP-auxch"); + if (!radeon_dig_connector->dp_i2c_bus) + DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); if (!radeon_connector->ddc_bus) - DRM_ERROR("DVI: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); } - subpixel_order = SubPixelHorizontalRGB; - drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.coherent_mode_property, - 1); - if (ASIC_IS_AVIVO(rdev)) { + switch (connector_type) { + case DRM_MODE_CONNECTOR_VGA: + case DRM_MODE_CONNECTOR_DVIA: + default: + connector->interlace_allowed = true; + connector->doublescan_allowed = true; + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + break; + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_DVID: + case DRM_MODE_CONNECTOR_HDMIA: + case DRM_MODE_CONNECTOR_HDMIB: + case DRM_MODE_CONNECTOR_DisplayPort: drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.underscan_property, UNDERSCAN_OFF); @@ -1265,131 +1524,240 @@ radeon_add_atom_connector(struct drm_device *dev, drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.underscan_vborder_property, 0); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = true; + if (connector_type == DRM_MODE_CONNECTOR_HDMIB) + connector->doublescan_allowed = true; + else + connector->doublescan_allowed = false; + if (connector_type == DRM_MODE_CONNECTOR_DVII) { + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + } + break; + case DRM_MODE_CONNECTOR_LVDS: + case DRM_MODE_CONNECTOR_eDP: + drm_connector_attach_property(&radeon_connector->base, + dev->mode_config.scaling_mode_property, + DRM_MODE_SCALE_FULLSCREEN); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; } - if (connector_type == DRM_MODE_CONNECTOR_DVII) { + } else { + switch (connector_type) { + case DRM_MODE_CONNECTOR_VGA: + drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("VGA: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } radeon_connector->dac_load_detect = true; drm_connector_attach_property(&radeon_connector->base, rdev->mode_info.load_detect_property, 1); - } - connector->interlace_allowed = true; - if (connector_type == DRM_MODE_CONNECTOR_DVII) + /* no HPD on analog connectors */ + radeon_connector->hpd.hpd = RADEON_HPD_NONE; + connector->polled = DRM_CONNECTOR_POLL_CONNECT; + connector->interlace_allowed = true; connector->doublescan_allowed = true; - else - connector->doublescan_allowed = false; - break; - case DRM_MODE_CONNECTOR_HDMIA: - case DRM_MODE_CONNECTOR_HDMIB: - radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); - if (!radeon_dig_connector) - goto failed; - radeon_dig_connector->igp_lane_info = igp_lane_info; - radeon_connector->con_priv = radeon_dig_connector; - drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); - if (i2c_bus->valid) { - radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); - if (!radeon_connector->ddc_bus) - DRM_ERROR("HDMI: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); - } - drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.coherent_mode_property, - 1); - if (ASIC_IS_AVIVO(rdev)) { + break; + case DRM_MODE_CONNECTOR_DVIA: + drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_vga_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DVIA: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + radeon_connector->dac_load_detect = true; drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.underscan_property, - UNDERSCAN_OFF); + rdev->mode_info.load_detect_property, + 1); + /* no HPD on analog connectors */ + radeon_connector->hpd.hpd = RADEON_HPD_NONE; + connector->interlace_allowed = true; + connector->doublescan_allowed = true; + break; + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_DVID: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DVI: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + subpixel_order = SubPixelHorizontalRGB; drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.underscan_hborder_property, - 0); + rdev->mode_info.coherent_mode_property, + 1); + if (ASIC_IS_AVIVO(rdev)) { + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_property, + UNDERSCAN_OFF); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_hborder_property, + 0); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_vborder_property, + 0); + } + if (connector_type == DRM_MODE_CONNECTOR_DVII) { + radeon_connector->dac_load_detect = true; + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.load_detect_property, + 1); + } + connector->interlace_allowed = true; + if (connector_type == DRM_MODE_CONNECTOR_DVII) + connector->doublescan_allowed = true; + else + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_HDMIA: + case DRM_MODE_CONNECTOR_HDMIB: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_dvi_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dvi_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("HDMI: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.underscan_vborder_property, - 0); - } - subpixel_order = SubPixelHorizontalRGB; - connector->interlace_allowed = true; - if (connector_type == DRM_MODE_CONNECTOR_HDMIB) - connector->doublescan_allowed = true; - else - connector->doublescan_allowed = false; - break; - case DRM_MODE_CONNECTOR_DisplayPort: - case DRM_MODE_CONNECTOR_eDP: - radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); - if (!radeon_dig_connector) - goto failed; - radeon_dig_connector->igp_lane_info = igp_lane_info; - radeon_connector->con_priv = radeon_dig_connector; - drm_connector_init(dev, &radeon_connector->base, &radeon_dp_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_dp_connector_helper_funcs); - if (i2c_bus->valid) { - /* add DP i2c bus */ - if (connector_type == DRM_MODE_CONNECTOR_eDP) - radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "eDP-auxch"); + rdev->mode_info.coherent_mode_property, + 1); + if (ASIC_IS_AVIVO(rdev)) { + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_property, + UNDERSCAN_OFF); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_hborder_property, + 0); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_vborder_property, + 0); + } + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = true; + if (connector_type == DRM_MODE_CONNECTOR_HDMIB) + connector->doublescan_allowed = true; else + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_DisplayPort: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_dp_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dp_connector_helper_funcs); + if (i2c_bus->valid) { + /* add DP i2c bus */ radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "DP-auxch"); - if (!radeon_dig_connector->dp_i2c_bus) - DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); - radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); - if (!radeon_connector->ddc_bus) - DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); - } - subpixel_order = SubPixelHorizontalRGB; - drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.coherent_mode_property, - 1); - if (ASIC_IS_AVIVO(rdev)) { + if (!radeon_dig_connector->dp_i2c_bus) + DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + subpixel_order = SubPixelHorizontalRGB; drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.underscan_property, - UNDERSCAN_OFF); + rdev->mode_info.coherent_mode_property, + 1); + if (ASIC_IS_AVIVO(rdev)) { + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_property, + UNDERSCAN_OFF); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_hborder_property, + 0); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.underscan_vborder_property, + 0); + } + connector->interlace_allowed = true; + /* in theory with a DP to VGA converter... */ + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_eDP: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_dp_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_dp_connector_helper_funcs); + if (i2c_bus->valid) { + /* add DP i2c bus */ + radeon_dig_connector->dp_i2c_bus = radeon_i2c_create_dp(dev, i2c_bus, "eDP-auxch"); + if (!radeon_dig_connector->dp_i2c_bus) + DRM_ERROR("DP: Failed to assign dp ddc bus! Check dmesg for i2c errors.\n"); + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("DP: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.underscan_hborder_property, - 0); + dev->mode_config.scaling_mode_property, + DRM_MODE_SCALE_FULLSCREEN); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_SVIDEO: + case DRM_MODE_CONNECTOR_Composite: + case DRM_MODE_CONNECTOR_9PinDIN: + drm_connector_init(dev, &radeon_connector->base, &radeon_tv_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_tv_connector_helper_funcs); + radeon_connector->dac_load_detect = true; drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.underscan_vborder_property, - 0); - } - connector->interlace_allowed = true; - /* in theory with a DP to VGA converter... */ - connector->doublescan_allowed = false; - break; - case DRM_MODE_CONNECTOR_SVIDEO: - case DRM_MODE_CONNECTOR_Composite: - case DRM_MODE_CONNECTOR_9PinDIN: - drm_connector_init(dev, &radeon_connector->base, &radeon_tv_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_tv_connector_helper_funcs); - radeon_connector->dac_load_detect = true; - drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.load_detect_property, - 1); - drm_connector_attach_property(&radeon_connector->base, - rdev->mode_info.tv_std_property, - radeon_atombios_get_tv_info(rdev)); - /* no HPD on analog connectors */ - radeon_connector->hpd.hpd = RADEON_HPD_NONE; - connector->interlace_allowed = false; - connector->doublescan_allowed = false; - break; - case DRM_MODE_CONNECTOR_LVDS: - radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); - if (!radeon_dig_connector) - goto failed; - radeon_dig_connector->igp_lane_info = igp_lane_info; - radeon_connector->con_priv = radeon_dig_connector; - drm_connector_init(dev, &radeon_connector->base, &radeon_lvds_connector_funcs, connector_type); - drm_connector_helper_add(&radeon_connector->base, &radeon_lvds_connector_helper_funcs); - if (i2c_bus->valid) { - radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); - if (!radeon_connector->ddc_bus) - DRM_ERROR("LVDS: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + rdev->mode_info.load_detect_property, + 1); + drm_connector_attach_property(&radeon_connector->base, + rdev->mode_info.tv_std_property, + radeon_atombios_get_tv_info(rdev)); + /* no HPD on analog connectors */ + radeon_connector->hpd.hpd = RADEON_HPD_NONE; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; + case DRM_MODE_CONNECTOR_LVDS: + radeon_dig_connector = kzalloc(sizeof(struct radeon_connector_atom_dig), GFP_KERNEL); + if (!radeon_dig_connector) + goto failed; + radeon_dig_connector->igp_lane_info = igp_lane_info; + radeon_connector->con_priv = radeon_dig_connector; + drm_connector_init(dev, &radeon_connector->base, &radeon_lvds_connector_funcs, connector_type); + drm_connector_helper_add(&radeon_connector->base, &radeon_lvds_connector_helper_funcs); + if (i2c_bus->valid) { + radeon_connector->ddc_bus = radeon_i2c_lookup(rdev, i2c_bus); + if (!radeon_connector->ddc_bus) + DRM_ERROR("LVDS: Failed to assign ddc bus! Check dmesg for i2c errors.\n"); + } + drm_connector_attach_property(&radeon_connector->base, + dev->mode_config.scaling_mode_property, + DRM_MODE_SCALE_FULLSCREEN); + subpixel_order = SubPixelHorizontalRGB; + connector->interlace_allowed = false; + connector->doublescan_allowed = false; + break; } - drm_connector_attach_property(&radeon_connector->base, - dev->mode_config.scaling_mode_property, - DRM_MODE_SCALE_FULLSCREEN); - subpixel_order = SubPixelHorizontalRGB; - connector->interlace_allowed = false; - connector->doublescan_allowed = false; - break; } if (radeon_connector->hpd.hpd == RADEON_HPD_NONE) { @@ -1450,6 +1818,9 @@ radeon_add_legacy_connector(struct drm_device *dev, radeon_connector->devices = supported_device; radeon_connector->connector_object_id = connector_object_id; radeon_connector->hpd = *hpd; + radeon_connector->requires_extended_probe = + radeon_connector_needs_extended_probe(rdev, supported_device, + connector_type); switch (connector_type) { case DRM_MODE_CONNECTOR_VGA: drm_connector_init(dev, &radeon_connector->base, &radeon_vga_connector_funcs, connector_type); diff --git a/drivers/gpu/drm/radeon/radeon_cp.c b/drivers/gpu/drm/radeon/radeon_cp.c index 75867792a4e..045ec59478f 100644 --- a/drivers/gpu/drm/radeon/radeon_cp.c +++ b/drivers/gpu/drm/radeon/radeon_cp.c @@ -2115,7 +2115,7 @@ int radeon_driver_load(struct drm_device *dev, unsigned long flags) if (drm_pci_device_is_agp(dev)) dev_priv->flags |= RADEON_IS_AGP; - else if (drm_pci_device_is_pcie(dev)) + else if (pci_is_pcie(dev->pdev)) dev_priv->flags |= RADEON_IS_PCIE; else dev_priv->flags |= RADEON_IS_PCI; diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c index 8c191694187..fae00c0d75a 100644 --- a/drivers/gpu/drm/radeon/radeon_cs.c +++ b/drivers/gpu/drm/radeon/radeon_cs.c @@ -228,6 +228,7 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) parser.filp = filp; parser.rdev = rdev; parser.dev = rdev->dev; + parser.family = rdev->family; r = radeon_cs_parser_init(&parser, data); if (r) { DRM_ERROR("Failed to initialize parser !\n"); diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c index 890217e678d..b51e15725c6 100644 --- a/drivers/gpu/drm/radeon/radeon_device.c +++ b/drivers/gpu/drm/radeon/radeon_device.c @@ -32,6 +32,7 @@ #include <drm/radeon_drm.h> #include <linux/vgaarb.h> #include <linux/vga_switcheroo.h> +#include <linux/efi.h> #include "radeon_reg.h" #include "radeon.h" #include "atom.h" @@ -82,6 +83,8 @@ static const char radeon_family_name[][16] = { "CYPRESS", "HEMLOCK", "PALM", + "SUMO", + "SUMO2", "BARTS", "TURKS", "CAICOS", @@ -213,6 +216,8 @@ int radeon_wb_init(struct radeon_device *rdev) return r; } + /* clear wb memory */ + memset((char *)rdev->wb.wb, 0, RADEON_GPU_PAGE_SIZE); /* disable event_write fences */ rdev->wb.use_event = false; /* disabled via module param */ @@ -296,6 +301,8 @@ void radeon_vram_location(struct radeon_device *rdev, struct radeon_mc *mc, u64 mc->mc_vram_size = mc->aper_size; } mc->vram_end = mc->vram_start + mc->mc_vram_size - 1; + if (radeon_vram_limit && radeon_vram_limit < mc->real_vram_size) + mc->real_vram_size = radeon_vram_limit; dev_info(rdev->dev, "VRAM: %lluM 0x%016llX - 0x%016llX (%lluM used)\n", mc->mc_vram_size >> 20, mc->vram_start, mc->vram_end, mc->real_vram_size >> 20); @@ -344,6 +351,9 @@ bool radeon_card_posted(struct radeon_device *rdev) { uint32_t reg; + if (efi_enabled && rdev->pdev->subsystem_vendor == PCI_VENDOR_ID_APPLE) + return false; + /* first check CRTCs */ if (ASIC_IS_DCE41(rdev)) { reg = RREG32(EVERGREEN_CRTC_CONTROL + EVERGREEN_CRTC0_REGISTER_OFFSET) | @@ -700,8 +710,9 @@ int radeon_device_init(struct radeon_device *rdev, rdev->gpu_lockup = false; rdev->accel_working = false; - DRM_INFO("initializing kernel modesetting (%s 0x%04X:0x%04X).\n", - radeon_family_name[rdev->family], pdev->vendor, pdev->device); + DRM_INFO("initializing kernel modesetting (%s 0x%04X:0x%04X 0x%04X:0x%04X).\n", + radeon_family_name[rdev->family], pdev->vendor, pdev->device, + pdev->subsystem_vendor, pdev->subsystem_device); /* mutex initialization are all done here so we * can recall function without having locking issues */ @@ -752,6 +763,7 @@ int radeon_device_init(struct radeon_device *rdev, dma_bits = rdev->need_dma32 ? 32 : 40; r = pci_set_dma_mask(rdev->pdev, DMA_BIT_MASK(dma_bits)); if (r) { + rdev->need_dma32 = true; printk(KERN_WARNING "radeon: No suitable DMA available.\n"); } @@ -923,6 +935,9 @@ int radeon_resume_kms(struct drm_device *dev) radeon_fbdev_set_suspend(rdev, 0); console_unlock(); + /* init dig PHYs */ + if (rdev->is_atom_bios) + radeon_atom_encoder_init(rdev); /* reset hpd state */ radeon_hpd_init(rdev); /* blat the mode back in */ diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c index bdbab5c43bd..1a858944e4f 100644 --- a/drivers/gpu/drm/radeon/radeon_display.c +++ b/drivers/gpu/drm/radeon/radeon_display.c @@ -264,6 +264,8 @@ static void radeon_unpin_work_func(struct work_struct *__work) radeon_bo_unreserve(work->old_rbo); } else DRM_ERROR("failed to reserve buffer after flip\n"); + + drm_gem_object_unreference_unlocked(&work->old_rbo->gem_base); kfree(work); } @@ -280,7 +282,7 @@ void radeon_crtc_handle_flip(struct radeon_device *rdev, int crtc_id) spin_lock_irqsave(&rdev->ddev->event_lock, flags); work = radeon_crtc->unpin_work; if (work == NULL || - !radeon_fence_signaled(work->fence)) { + (work->fence && !radeon_fence_signaled(work->fence))) { spin_unlock_irqrestore(&rdev->ddev->event_lock, flags); return; } @@ -346,7 +348,6 @@ static int radeon_crtc_page_flip(struct drm_crtc *crtc, struct radeon_framebuffer *new_radeon_fb; struct drm_gem_object *obj; struct radeon_bo *rbo; - struct radeon_fence *fence; struct radeon_unpin_work *work; unsigned long flags; u32 tiling_flags, pitch_pixels; @@ -357,42 +358,35 @@ static int radeon_crtc_page_flip(struct drm_crtc *crtc, if (work == NULL) return -ENOMEM; - r = radeon_fence_create(rdev, &fence); - if (unlikely(r != 0)) { - kfree(work); - DRM_ERROR("flip queue: failed to create fence.\n"); - return -ENOMEM; - } work->event = event; work->rdev = rdev; work->crtc_id = radeon_crtc->crtc_id; - work->fence = radeon_fence_ref(fence); old_radeon_fb = to_radeon_framebuffer(crtc->fb); new_radeon_fb = to_radeon_framebuffer(fb); /* schedule unpin of the old buffer */ obj = old_radeon_fb->obj; + /* take a reference to the old object */ + drm_gem_object_reference(obj); rbo = gem_to_radeon_bo(obj); work->old_rbo = rbo; + obj = new_radeon_fb->obj; + rbo = gem_to_radeon_bo(obj); + if (rbo->tbo.sync_obj) + work->fence = radeon_fence_ref(rbo->tbo.sync_obj); INIT_WORK(&work->work, radeon_unpin_work_func); /* We borrow the event spin lock for protecting unpin_work */ spin_lock_irqsave(&dev->event_lock, flags); if (radeon_crtc->unpin_work) { - spin_unlock_irqrestore(&dev->event_lock, flags); - kfree(work); - radeon_fence_unref(&fence); - DRM_DEBUG_DRIVER("flip queue: crtc already busy\n"); - return -EBUSY; + r = -EBUSY; + goto unlock_free; } radeon_crtc->unpin_work = work; radeon_crtc->deferred_flip_completion = 0; spin_unlock_irqrestore(&dev->event_lock, flags); /* pin the new buffer */ - obj = new_radeon_fb->obj; - rbo = gem_to_radeon_bo(obj); - DRM_DEBUG_DRIVER("flip-ioctl() cur_fbo = %p, cur_bbo = %p\n", work->old_rbo, rbo); @@ -460,45 +454,28 @@ static int radeon_crtc_page_flip(struct drm_crtc *crtc, goto pflip_cleanup1; } - /* 32 ought to cover us */ - r = radeon_ring_lock(rdev, 32); - if (r) { - DRM_ERROR("failed to lock the ring before flip\n"); - goto pflip_cleanup2; - } - - /* emit the fence */ - radeon_fence_emit(rdev, fence); /* set the proper interrupt */ radeon_pre_page_flip(rdev, radeon_crtc->crtc_id); - /* fire the ring */ - radeon_ring_unlock_commit(rdev); return 0; -pflip_cleanup2: - drm_vblank_put(dev, radeon_crtc->crtc_id); - pflip_cleanup1: - r = radeon_bo_reserve(rbo, false); - if (unlikely(r != 0)) { + if (unlikely(radeon_bo_reserve(rbo, false) != 0)) { DRM_ERROR("failed to reserve new rbo in error path\n"); goto pflip_cleanup; } - r = radeon_bo_unpin(rbo); - if (unlikely(r != 0)) { - radeon_bo_unreserve(rbo); - r = -EINVAL; + if (unlikely(radeon_bo_unpin(rbo) != 0)) { DRM_ERROR("failed to unpin new rbo in error path\n"); - goto pflip_cleanup; } radeon_bo_unreserve(rbo); pflip_cleanup: spin_lock_irqsave(&dev->event_lock, flags); radeon_crtc->unpin_work = NULL; +unlock_free: + drm_gem_object_unreference_unlocked(old_radeon_fb->obj); spin_unlock_irqrestore(&dev->event_lock, flags); - radeon_fence_unref(&fence); + radeon_fence_unref(&work->fence); kfree(work); return r; @@ -774,8 +751,17 @@ static int radeon_ddc_dump(struct drm_connector *connector) if (!radeon_connector->ddc_bus) return -1; edid = drm_get_edid(connector, &radeon_connector->ddc_bus->adapter); + /* Log EDID retrieval status here. In particular with regard to + * connectors with requires_extended_probe flag set, that will prevent + * function radeon_dvi_detect() to fetch EDID on this connector, + * as long as there is no valid EDID header found */ if (edid) { + DRM_INFO("Radeon display connector %s: Found valid EDID", + drm_get_connector_name(connector)); kfree(edid); + } else { + DRM_INFO("Radeon display connector %s: No monitor connected or invalid EDID", + drm_get_connector_name(connector)); } return ret; } @@ -1087,8 +1073,9 @@ void radeon_compute_pll_legacy(struct radeon_pll *pll, *frac_fb_div_p = best_frac_feedback_div; *ref_div_p = best_ref_div; *post_div_p = best_post_div; - DRM_DEBUG_KMS("%d %d, pll dividers - fb: %d.%d ref: %d, post %d\n", - freq, best_freq / 1000, best_feedback_div, best_frac_feedback_div, + DRM_DEBUG_KMS("%lld %d, pll dividers - fb: %d.%d ref: %d, post %d\n", + (long long)freq, + best_freq / 1000, best_feedback_div, best_frac_feedback_div, best_ref_div, best_post_div); } @@ -1344,6 +1331,11 @@ int radeon_modeset_init(struct radeon_device *rdev) if (!ret) { return ret; } + + /* init dig PHYs */ + if (rdev->is_atom_bios) + radeon_atom_encoder_init(rdev); + /* initialize hpd */ radeon_hpd_init(rdev); diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index 63d2de8771d..e71d2ed7fa1 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -50,9 +50,11 @@ * 2.7.0 - fixups for r600 2D tiling support. (no external ABI change), add eg dyn gpr regs * 2.8.0 - pageflip support, r500 US_FORMAT regs. r500 ARGB2101010 colorbuf, r300->r500 CMASK, clock crystal query * 2.9.0 - r600 tiling (s3tc,rgtc) working, SET_PREDICATION packet 3 on r600 + eg, backend query + * 2.10.0 - fusion 2D tiling + * 2.11.0 - backend map, initial compute support for the CS checker */ #define KMS_DRIVER_MAJOR 2 -#define KMS_DRIVER_MINOR 9 +#define KMS_DRIVER_MINOR 11 #define KMS_DRIVER_PATCHLEVEL 0 int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags); int radeon_driver_unload_kms(struct drm_device *dev); @@ -112,7 +114,7 @@ int radeon_benchmarking = 0; int radeon_testing = 0; int radeon_connector_table = 0; int radeon_tv = 1; -int radeon_audio = 1; +int radeon_audio = 0; int radeon_disp_priority = 0; int radeon_hw_i2c = 0; int radeon_pcie_gen2 = 0; @@ -150,7 +152,7 @@ module_param_named(connector_table, radeon_connector_table, int, 0444); MODULE_PARM_DESC(tv, "TV enable (0 = disable)"); module_param_named(tv, radeon_tv, int, 0444); -MODULE_PARM_DESC(audio, "Audio enable (0 = disable)"); +MODULE_PARM_DESC(audio, "Audio enable (1 = enable)"); module_param_named(audio, radeon_audio, int, 0444); MODULE_PARM_DESC(disp_priority, "Display Priority (0 = auto, 1 = normal, 2 = high)"); diff --git a/drivers/gpu/drm/radeon/radeon_encoders.c b/drivers/gpu/drm/radeon/radeon_encoders.c index b4274883227..319d85d7e75 100644 --- a/drivers/gpu/drm/radeon/radeon_encoders.c +++ b/drivers/gpu/drm/radeon/radeon_encoders.c @@ -229,6 +229,22 @@ radeon_get_connector_for_encoder(struct drm_encoder *encoder) return NULL; } +static struct drm_connector * +radeon_get_connector_for_encoder_init(struct drm_encoder *encoder) +{ + struct drm_device *dev = encoder->dev; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_connector *connector; + struct radeon_connector *radeon_connector; + + list_for_each_entry(connector, &dev->mode_config.connector_list, head) { + radeon_connector = to_radeon_connector(connector); + if (radeon_encoder->devices & radeon_connector->devices) + return connector; + } + return NULL; +} + struct drm_encoder *radeon_atom_get_external_encoder(struct drm_encoder *encoder) { struct drm_device *dev = encoder->dev; @@ -250,6 +266,25 @@ struct drm_encoder *radeon_atom_get_external_encoder(struct drm_encoder *encoder return NULL; } +bool radeon_encoder_is_dp_bridge(struct drm_encoder *encoder) +{ + struct drm_encoder *other_encoder = radeon_atom_get_external_encoder(encoder); + + if (other_encoder) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(other_encoder); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_TRAVIS: + case ENCODER_OBJECT_ID_NUTMEG: + return true; + default: + return false; + } + } + + return false; +} + void radeon_panel_mode_fixup(struct drm_encoder *encoder, struct drm_display_mode *adjusted_mode) { @@ -332,7 +367,8 @@ static bool radeon_atom_mode_fixup(struct drm_encoder *encoder, } if (ASIC_IS_DCE3(rdev) && - (radeon_encoder->active_device & (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT))) { + ((radeon_encoder->active_device & (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) || + radeon_encoder_is_dp_bridge(encoder))) { struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); radeon_dp_set_link_config(connector, mode); } @@ -621,21 +657,20 @@ atombios_get_encoder_mode(struct drm_encoder *encoder) struct radeon_connector *radeon_connector; struct radeon_connector_atom_dig *dig_connector; + /* dp bridges are always DP */ + if (radeon_encoder_is_dp_bridge(encoder)) + return ATOM_ENCODER_MODE_DP; + + /* DVO is always DVO */ + if (radeon_encoder->encoder_id == ATOM_ENCODER_MODE_DVO) + return ATOM_ENCODER_MODE_DVO; + connector = radeon_get_connector_for_encoder(encoder); - if (!connector) { - switch (radeon_encoder->encoder_id) { - case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: - case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: - case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: - case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: - case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DVO1: - return ATOM_ENCODER_MODE_DVI; - case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC1: - case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_DAC2: - default: - return ATOM_ENCODER_MODE_CRT; - } - } + /* if we don't have an active device yet, just use one of + * the connectors tied to the encoder. + */ + if (!connector) + connector = radeon_get_connector_for_encoder_init(encoder); radeon_connector = to_radeon_connector(connector); switch (connector->connector_type) { @@ -668,7 +703,6 @@ atombios_get_encoder_mode(struct drm_encoder *encoder) return ATOM_ENCODER_MODE_LVDS; break; case DRM_MODE_CONNECTOR_DisplayPort: - case DRM_MODE_CONNECTOR_eDP: dig_connector = radeon_connector->con_priv; if ((dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_DISPLAYPORT) || (dig_connector->dp_sink_type == CONNECTOR_OBJECT_ID_eDP)) @@ -682,6 +716,8 @@ atombios_get_encoder_mode(struct drm_encoder *encoder) } else return ATOM_ENCODER_MODE_DVI; break; + case DRM_MODE_CONNECTOR_eDP: + return ATOM_ENCODER_MODE_DP; case DRM_MODE_CONNECTOR_DVIA: case DRM_MODE_CONNECTOR_VGA: return ATOM_ENCODER_MODE_CRT; @@ -747,7 +783,7 @@ union dig_encoder_control { }; void -atombios_dig_encoder_setup(struct drm_encoder *encoder, int action) +atombios_dig_encoder_setup(struct drm_encoder *encoder, int action, int panel_mode) { struct drm_device *dev = encoder->dev; struct radeon_device *rdev = dev->dev_private; @@ -760,6 +796,7 @@ atombios_dig_encoder_setup(struct drm_encoder *encoder, int action) int dp_clock = 0; int dp_lane_count = 0; int hpd_id = RADEON_HPD_NONE; + int bpc = 8; if (connector) { struct radeon_connector *radeon_connector = to_radeon_connector(connector); @@ -769,6 +806,7 @@ atombios_dig_encoder_setup(struct drm_encoder *encoder, int action) dp_clock = dig_connector->dp_clock; dp_lane_count = dig_connector->dp_lane_count; hpd_id = radeon_connector->hpd.hpd; + bpc = connector->display_info.bpc; } /* no dig encoder assigned */ @@ -791,7 +829,10 @@ atombios_dig_encoder_setup(struct drm_encoder *encoder, int action) args.v1.ucAction = action; args.v1.usPixelClock = cpu_to_le16(radeon_encoder->pixel_clock / 10); - args.v1.ucEncoderMode = atombios_get_encoder_mode(encoder); + if (action == ATOM_ENCODER_CMD_SETUP_PANEL_MODE) + args.v3.ucPanelMode = panel_mode; + else + args.v1.ucEncoderMode = atombios_get_encoder_mode(encoder); if ((args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP) || (args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP_MST)) @@ -810,7 +851,27 @@ atombios_dig_encoder_setup(struct drm_encoder *encoder, int action) args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V4_DPLINKRATE_5_40GHZ; } args.v4.acConfig.ucDigSel = dig->dig_encoder; - args.v4.ucBitPerColor = PANEL_8BIT_PER_COLOR; + switch (bpc) { + case 0: + args.v4.ucBitPerColor = PANEL_BPC_UNDEFINE; + break; + case 6: + args.v4.ucBitPerColor = PANEL_6BIT_PER_COLOR; + break; + case 8: + default: + args.v4.ucBitPerColor = PANEL_8BIT_PER_COLOR; + break; + case 10: + args.v4.ucBitPerColor = PANEL_10BIT_PER_COLOR; + break; + case 12: + args.v4.ucBitPerColor = PANEL_12BIT_PER_COLOR; + break; + case 16: + args.v4.ucBitPerColor = PANEL_16BIT_PER_COLOR; + break; + } if (hpd_id == RADEON_HPD_NONE) args.v4.ucHPD_ID = 0; else @@ -819,7 +880,27 @@ atombios_dig_encoder_setup(struct drm_encoder *encoder, int action) if ((args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP) && (dp_clock == 270000)) args.v1.ucConfig |= ATOM_ENCODER_CONFIG_V3_DPLINKRATE_2_70GHZ; args.v3.acConfig.ucDigSel = dig->dig_encoder; - args.v3.ucBitPerColor = PANEL_8BIT_PER_COLOR; + switch (bpc) { + case 0: + args.v3.ucBitPerColor = PANEL_BPC_UNDEFINE; + break; + case 6: + args.v3.ucBitPerColor = PANEL_6BIT_PER_COLOR; + break; + case 8: + default: + args.v3.ucBitPerColor = PANEL_8BIT_PER_COLOR; + break; + case 10: + args.v3.ucBitPerColor = PANEL_10BIT_PER_COLOR; + break; + case 12: + args.v3.ucBitPerColor = PANEL_12BIT_PER_COLOR; + break; + case 16: + args.v3.ucBitPerColor = PANEL_16BIT_PER_COLOR; + break; + } } else { if ((args.v1.ucEncoderMode == ATOM_ENCODER_MODE_DP) && (dp_clock == 270000)) args.v1.ucConfig |= ATOM_ENCODER_CONFIG_DPLINKRATE_2_70GHZ; @@ -859,7 +940,7 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t struct radeon_device *rdev = dev->dev_private; struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; - struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); + struct drm_connector *connector; union dig_transmitter_control args; int index = 0; uint8_t frev, crev; @@ -869,6 +950,16 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t int dp_lane_count = 0; int connector_object_id = 0; int igp_lane_info = 0; + int dig_encoder = dig->dig_encoder; + + if (action == ATOM_TRANSMITTER_ACTION_INIT) { + connector = radeon_get_connector_for_encoder_init(encoder); + /* just needed to avoid bailing in the encoder check. the encoder + * isn't used for init + */ + dig_encoder = 0; + } else + connector = radeon_get_connector_for_encoder(encoder); if (connector) { struct radeon_connector *radeon_connector = to_radeon_connector(connector); @@ -883,7 +974,7 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t } /* no dig encoder assigned */ - if (dig->dig_encoder == -1) + if (dig_encoder == -1) return; if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_DP) @@ -931,10 +1022,10 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t else args.v3.ucLaneNum = 4; - if (dig->linkb) { + if (dig->linkb) args.v3.acConfig.ucLinkSel = 1; + if (dig_encoder & 1) args.v3.acConfig.ucEncoderSel = 1; - } /* Select the PLL for the PHY * DP PHY should be clocked from external src if there is @@ -946,11 +1037,16 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t } if (ASIC_IS_DCE5(rdev)) { - if (is_dp && rdev->clock.dp_extclk) - args.v4.acConfig.ucRefClkSource = 3; /* external src */ - else + /* On DCE5 DCPLL usually generates the DP ref clock */ + if (is_dp) { + if (rdev->clock.dp_extclk) + args.v4.acConfig.ucRefClkSource = ENCODER_REFCLK_SRC_EXTCLK; + else + args.v4.acConfig.ucRefClkSource = ENCODER_REFCLK_SRC_DCPLL; + } else args.v4.acConfig.ucRefClkSource = pll_id; } else { + /* On DCE4, if there is an external clock, it generates the DP ref clock */ if (is_dp && rdev->clock.dp_extclk) args.v3.acConfig.ucRefClkSource = 2; /* external src */ else @@ -978,7 +1074,7 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t args.v3.acConfig.fDualLinkConnector = 1; } } else if (ASIC_IS_DCE32(rdev)) { - args.v2.acConfig.ucEncoderSel = dig->dig_encoder; + args.v2.acConfig.ucEncoderSel = dig_encoder; if (dig->linkb) args.v2.acConfig.ucLinkSel = 1; @@ -994,9 +1090,10 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t break; } - if (is_dp) + if (is_dp) { args.v2.acConfig.fCoherentMode = 1; - else if (radeon_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) { + args.v2.acConfig.fDPConnector = 1; + } else if (radeon_encoder->devices & (ATOM_DEVICE_DFP_SUPPORT)) { if (dig->coherent_mode) args.v2.acConfig.fCoherentMode = 1; if (radeon_encoder->pixel_clock > 165000) @@ -1005,7 +1102,7 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t } else { args.v1.ucConfig = ATOM_TRANSMITTER_CONFIG_CLKSRC_PPLL; - if (dig->dig_encoder) + if (dig_encoder) args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG2_ENCODER; else args.v1.ucConfig |= ATOM_TRANSMITTER_CONFIG_DIG1_ENCODER; @@ -1047,7 +1144,7 @@ atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); } -void +bool atombios_set_edp_panel_power(struct drm_connector *connector, int action) { struct radeon_connector *radeon_connector = to_radeon_connector(connector); @@ -1058,23 +1155,37 @@ atombios_set_edp_panel_power(struct drm_connector *connector, int action) uint8_t frev, crev; if (connector->connector_type != DRM_MODE_CONNECTOR_eDP) - return; + goto done; if (!ASIC_IS_DCE4(rdev)) - return; + goto done; if ((action != ATOM_TRANSMITTER_ACTION_POWER_ON) && (action != ATOM_TRANSMITTER_ACTION_POWER_OFF)) - return; + goto done; if (!atom_parse_cmd_header(rdev->mode_info.atom_context, index, &frev, &crev)) - return; + goto done; memset(&args, 0, sizeof(args)); args.v1.ucAction = action; atom_execute_table(rdev->mode_info.atom_context, index, (uint32_t *)&args); + + /* wait for the panel to power up */ + if (action == ATOM_TRANSMITTER_ACTION_POWER_ON) { + int i; + + for (i = 0; i < 300; i++) { + if (radeon_hpd_sense(rdev, radeon_connector->hpd.hpd)) + return true; + mdelay(1); + } + return false; + } +done: + return true; } union external_encoder_control { @@ -1092,13 +1203,19 @@ atombios_external_encoder_setup(struct drm_encoder *encoder, struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); struct radeon_encoder *ext_radeon_encoder = to_radeon_encoder(ext_encoder); union external_encoder_control args; - struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); + struct drm_connector *connector; int index = GetIndexIntoMasterTable(COMMAND, ExternalEncoderControl); u8 frev, crev; int dp_clock = 0; int dp_lane_count = 0; int connector_object_id = 0; u32 ext_enum = (ext_radeon_encoder->encoder_enum & ENUM_ID_MASK) >> ENUM_ID_SHIFT; + int bpc = 8; + + if (action == EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT) + connector = radeon_get_connector_for_encoder_init(encoder); + else + connector = radeon_get_connector_for_encoder(encoder); if (connector) { struct radeon_connector *radeon_connector = to_radeon_connector(connector); @@ -1109,6 +1226,7 @@ atombios_external_encoder_setup(struct drm_encoder *encoder, dp_lane_count = dig_connector->dp_lane_count; connector_object_id = (radeon_connector->connector_object_id & OBJECT_ID_MASK) >> OBJECT_ID_SHIFT; + bpc = connector->display_info.bpc; } memset(&args, 0, sizeof(args)); @@ -1166,7 +1284,27 @@ atombios_external_encoder_setup(struct drm_encoder *encoder, args.v3.sExtEncoder.ucConfig |= EXTERNAL_ENCODER_CONFIG_V3_ENCODER3; break; } - args.v3.sExtEncoder.ucBitPerColor = PANEL_8BIT_PER_COLOR; + switch (bpc) { + case 0: + args.v3.sExtEncoder.ucBitPerColor = PANEL_BPC_UNDEFINE; + break; + case 6: + args.v3.sExtEncoder.ucBitPerColor = PANEL_6BIT_PER_COLOR; + break; + case 8: + default: + args.v3.sExtEncoder.ucBitPerColor = PANEL_8BIT_PER_COLOR; + break; + case 10: + args.v3.sExtEncoder.ucBitPerColor = PANEL_10BIT_PER_COLOR; + break; + case 12: + args.v3.sExtEncoder.ucBitPerColor = PANEL_12BIT_PER_COLOR; + break; + case 16: + args.v3.sExtEncoder.ucBitPerColor = PANEL_16BIT_PER_COLOR; + break; + } break; default: DRM_ERROR("Unknown table version: %d, %d\n", frev, crev); @@ -1294,7 +1432,11 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) if (is_dig) { switch (mode) { case DRM_MODE_DPMS_ON: - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE_OUTPUT, 0, 0); + /* some early dce3.2 boards have a bug in their transmitter control table */ + if ((rdev->family == CHIP_RV710) || (rdev->family == CHIP_RV730)) + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); + else + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE_OUTPUT, 0, 0); if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_DP) { struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); @@ -1307,9 +1449,11 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) ATOM_TRANSMITTER_ACTION_POWER_ON); radeon_dig_connector->edp_on = true; } - dp_link_train(encoder, connector); if (ASIC_IS_DCE4(rdev)) - atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_ON); + atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_OFF, 0); + radeon_dp_link_train(encoder, connector); + if (ASIC_IS_DCE4(rdev)) + atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_ON, 0); } if (radeon_encoder->devices & (ATOM_DEVICE_LCD_SUPPORT)) atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_LCD_BLON, 0, 0); @@ -1322,7 +1466,7 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); if (ASIC_IS_DCE4(rdev)) - atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_OFF); + atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_DP_VIDEO_OFF, 0); if (connector && (connector->connector_type == DRM_MODE_CONNECTOR_eDP)) { struct radeon_connector *radeon_connector = to_radeon_connector(connector); @@ -1383,26 +1527,29 @@ radeon_atom_encoder_dpms(struct drm_encoder *encoder, int mode) } if (ext_encoder) { - int action; - switch (mode) { case DRM_MODE_DPMS_ON: default: - if (ASIC_IS_DCE41(rdev)) - action = EXTERNAL_ENCODER_ACTION_V3_ENABLE_OUTPUT; - else - action = ATOM_ENABLE; + if (ASIC_IS_DCE41(rdev)) { + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_ENABLE_OUTPUT); + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING_OFF); + } else + atombios_external_encoder_setup(encoder, ext_encoder, ATOM_ENABLE); break; case DRM_MODE_DPMS_STANDBY: case DRM_MODE_DPMS_SUSPEND: case DRM_MODE_DPMS_OFF: - if (ASIC_IS_DCE41(rdev)) - action = EXTERNAL_ENCODER_ACTION_V3_DISABLE_OUTPUT; - else - action = ATOM_DISABLE; + if (ASIC_IS_DCE41(rdev)) { + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_ENCODER_BLANKING); + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_DISABLE_OUTPUT); + } else + atombios_external_encoder_setup(encoder, ext_encoder, ATOM_DISABLE); break; } - atombios_external_encoder_setup(encoder, ext_encoder, action); } radeon_atombios_encoder_dpms_scratch_regs(encoder, (mode == DRM_MODE_DPMS_ON) ? true : false); @@ -1601,12 +1748,9 @@ static int radeon_atom_pick_dig_encoder(struct drm_encoder *encoder) /* DCE4/5 */ if (ASIC_IS_DCE4(rdev)) { dig = radeon_encoder->enc_priv; - if (ASIC_IS_DCE41(rdev)) { - if (dig->linkb) - return 1; - else - return 0; - } else { + if (ASIC_IS_DCE41(rdev)) + return radeon_crtc->crtc_id; + else { switch (radeon_encoder->encoder_id) { case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: if (dig->linkb) @@ -1662,6 +1806,34 @@ static int radeon_atom_pick_dig_encoder(struct drm_encoder *encoder) return 1; } +/* This only needs to be called once at startup */ +void +radeon_atom_encoder_init(struct radeon_device *rdev) +{ + struct drm_device *dev = rdev->ddev; + struct drm_encoder *encoder; + + list_for_each_entry(encoder, &dev->mode_config.encoder_list, head) { + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + + switch (radeon_encoder->encoder_id) { + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY1: + case ENCODER_OBJECT_ID_INTERNAL_UNIPHY2: + case ENCODER_OBJECT_ID_INTERNAL_KLDSCP_LVTMA: + atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_INIT, 0, 0); + break; + default: + break; + } + + if (ext_encoder && ASIC_IS_DCE41(rdev)) + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT); + } +} + static void radeon_atom_encoder_mode_set(struct drm_encoder *encoder, struct drm_display_mode *mode, @@ -1696,19 +1868,17 @@ radeon_atom_encoder_mode_set(struct drm_encoder *encoder, /* disable the transmitter */ atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); /* setup and enable the encoder */ - atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_SETUP); + atombios_dig_encoder_setup(encoder, ATOM_ENCODER_CMD_SETUP, 0); - /* init and enable the transmitter */ - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_INIT, 0, 0); + /* enable the transmitter */ atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); } else { /* disable the encoder and transmitter */ atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); - atombios_dig_encoder_setup(encoder, ATOM_DISABLE); + atombios_dig_encoder_setup(encoder, ATOM_DISABLE, 0); /* setup and enable the encoder and transmitter */ - atombios_dig_encoder_setup(encoder, ATOM_ENABLE); - atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_INIT, 0, 0); + atombios_dig_encoder_setup(encoder, ATOM_ENABLE, 0); atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_SETUP, 0, 0); atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_ENABLE, 0, 0); } @@ -1733,12 +1903,10 @@ radeon_atom_encoder_mode_set(struct drm_encoder *encoder, } if (ext_encoder) { - if (ASIC_IS_DCE41(rdev)) { - atombios_external_encoder_setup(encoder, ext_encoder, - EXTERNAL_ENCODER_ACTION_V3_ENCODER_INIT); + if (ASIC_IS_DCE41(rdev)) atombios_external_encoder_setup(encoder, ext_encoder, EXTERNAL_ENCODER_ACTION_V3_ENCODER_SETUP); - } else + else atombios_external_encoder_setup(encoder, ext_encoder, ATOM_ENABLE); } @@ -1840,13 +2008,73 @@ radeon_atom_dac_detect(struct drm_encoder *encoder, struct drm_connector *connec return connector_status_disconnected; } +static enum drm_connector_status +radeon_atom_dig_detect(struct drm_encoder *encoder, struct drm_connector *connector) +{ + struct drm_device *dev = encoder->dev; + struct radeon_device *rdev = dev->dev_private; + struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); + struct radeon_connector *radeon_connector = to_radeon_connector(connector); + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + u32 bios_0_scratch; + + if (!ASIC_IS_DCE4(rdev)) + return connector_status_unknown; + + if (!ext_encoder) + return connector_status_unknown; + + if ((radeon_connector->devices & ATOM_DEVICE_CRT_SUPPORT) == 0) + return connector_status_unknown; + + /* load detect on the dp bridge */ + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_DACLOAD_DETECTION); + + bios_0_scratch = RREG32(R600_BIOS_0_SCRATCH); + + DRM_DEBUG_KMS("Bios 0 scratch %x %08x\n", bios_0_scratch, radeon_encoder->devices); + if (radeon_connector->devices & ATOM_DEVICE_CRT1_SUPPORT) { + if (bios_0_scratch & ATOM_S0_CRT1_MASK) + return connector_status_connected; + } + if (radeon_connector->devices & ATOM_DEVICE_CRT2_SUPPORT) { + if (bios_0_scratch & ATOM_S0_CRT2_MASK) + return connector_status_connected; + } + if (radeon_connector->devices & ATOM_DEVICE_CV_SUPPORT) { + if (bios_0_scratch & (ATOM_S0_CV_MASK|ATOM_S0_CV_MASK_A)) + return connector_status_connected; + } + if (radeon_connector->devices & ATOM_DEVICE_TV1_SUPPORT) { + if (bios_0_scratch & (ATOM_S0_TV1_COMPOSITE | ATOM_S0_TV1_COMPOSITE_A)) + return connector_status_connected; /* CTV */ + else if (bios_0_scratch & (ATOM_S0_TV1_SVIDEO | ATOM_S0_TV1_SVIDEO_A)) + return connector_status_connected; /* STV */ + } + return connector_status_disconnected; +} + +void +radeon_atom_ext_encoder_setup_ddc(struct drm_encoder *encoder) +{ + struct drm_encoder *ext_encoder = radeon_atom_get_external_encoder(encoder); + + if (ext_encoder) + /* ddc_setup on the dp bridge */ + atombios_external_encoder_setup(encoder, ext_encoder, + EXTERNAL_ENCODER_ACTION_V3_DDC_SETUP); + +} + static void radeon_atom_encoder_prepare(struct drm_encoder *encoder) { struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); struct drm_connector *connector = radeon_get_connector_for_encoder(encoder); - if (radeon_encoder->active_device & - (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) { + if ((radeon_encoder->active_device & + (ATOM_DEVICE_DFP_SUPPORT | ATOM_DEVICE_LCD_SUPPORT)) || + radeon_encoder_is_dp_bridge(encoder)) { struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; if (dig) dig->dig_encoder = radeon_atom_pick_dig_encoder(encoder); @@ -1855,11 +2083,17 @@ static void radeon_atom_encoder_prepare(struct drm_encoder *encoder) radeon_atom_output_lock(encoder, true); radeon_atom_encoder_dpms(encoder, DRM_MODE_DPMS_OFF); - /* select the clock/data port if it uses a router */ if (connector) { struct radeon_connector *radeon_connector = to_radeon_connector(connector); + + /* select the clock/data port if it uses a router */ if (radeon_connector->router.cd_valid) radeon_router_select_cd_port(radeon_connector); + + /* turn eDP panel on for mode set */ + if (connector->connector_type == DRM_MODE_CONNECTOR_eDP) + atombios_set_edp_panel_power(connector, + ATOM_TRANSMITTER_ACTION_POWER_ON); } /* this is needed for the pll/ss setup to work correctly in some cases */ @@ -1914,7 +2148,7 @@ static void radeon_atom_encoder_disable(struct drm_encoder *encoder) else { /* disable the encoder and transmitter */ atombios_dig_transmitter_setup(encoder, ATOM_TRANSMITTER_ACTION_DISABLE, 0, 0); - atombios_dig_encoder_setup(encoder, ATOM_DISABLE); + atombios_dig_encoder_setup(encoder, ATOM_DISABLE, 0); } break; case ENCODER_OBJECT_ID_INTERNAL_DDI: @@ -1996,7 +2230,7 @@ static const struct drm_encoder_helper_funcs radeon_atom_dig_helper_funcs = { .mode_set = radeon_atom_encoder_mode_set, .commit = radeon_atom_encoder_commit, .disable = radeon_atom_encoder_disable, - /* no detect for TMDS/LVDS yet */ + .detect = radeon_atom_dig_detect, }; static const struct drm_encoder_helper_funcs radeon_atom_dac_helper_funcs = { @@ -2089,6 +2323,9 @@ radeon_add_atom_encoder(struct drm_device *dev, default: encoder->possible_crtcs = 0x3; break; + case 4: + encoder->possible_crtcs = 0xf; + break; case 6: encoder->possible_crtcs = 0x3f; break; @@ -2116,8 +2353,6 @@ radeon_add_atom_encoder(struct drm_device *dev, } else { drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_TMDS); radeon_encoder->enc_priv = radeon_atombios_set_dig_info(radeon_encoder); - if (ASIC_IS_AVIVO(rdev)) - radeon_encoder->underscan_type = UNDERSCAN_AUTO; } drm_encoder_helper_add(encoder, &radeon_atom_dig_helper_funcs); break; @@ -2150,8 +2385,6 @@ radeon_add_atom_encoder(struct drm_device *dev, } else { drm_encoder_init(dev, encoder, &radeon_atom_enc_funcs, DRM_MODE_ENCODER_TMDS); radeon_encoder->enc_priv = radeon_atombios_set_dig_info(radeon_encoder); - if (ASIC_IS_AVIVO(rdev)) - radeon_encoder->underscan_type = UNDERSCAN_AUTO; } drm_encoder_helper_add(encoder, &radeon_atom_dig_helper_funcs); break; diff --git a/drivers/gpu/drm/radeon/radeon_family.h b/drivers/gpu/drm/radeon/radeon_family.h index 6f1d9e563e7..ec2f1ea84f8 100644 --- a/drivers/gpu/drm/radeon/radeon_family.h +++ b/drivers/gpu/drm/radeon/radeon_family.h @@ -81,6 +81,8 @@ enum radeon_family { CHIP_CYPRESS, CHIP_HEMLOCK, CHIP_PALM, + CHIP_SUMO, + CHIP_SUMO2, CHIP_BARTS, CHIP_TURKS, CHIP_CAICOS, diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c index bbcd1dd7bac..7fd4e3e5ad5 100644 --- a/drivers/gpu/drm/radeon/radeon_fence.c +++ b/drivers/gpu/drm/radeon/radeon_fence.c @@ -29,7 +29,7 @@ * Dave Airlie */ #include <linux/seq_file.h> -#include <asm/atomic.h> +#include <linux/atomic.h> #include <linux/wait.h> #include <linux/list.h> #include <linux/kref.h> @@ -40,6 +40,35 @@ #include "radeon.h" #include "radeon_trace.h" +static void radeon_fence_write(struct radeon_device *rdev, u32 seq) +{ + if (rdev->wb.enabled) { + u32 scratch_index; + if (rdev->wb.use_event) + scratch_index = R600_WB_EVENT_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; + else + scratch_index = RADEON_WB_SCRATCH_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; + rdev->wb.wb[scratch_index/4] = cpu_to_le32(seq);; + } else + WREG32(rdev->fence_drv.scratch_reg, seq); +} + +static u32 radeon_fence_read(struct radeon_device *rdev) +{ + u32 seq; + + if (rdev->wb.enabled) { + u32 scratch_index; + if (rdev->wb.use_event) + scratch_index = R600_WB_EVENT_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; + else + scratch_index = RADEON_WB_SCRATCH_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; + seq = le32_to_cpu(rdev->wb.wb[scratch_index/4]); + } else + seq = RREG32(rdev->fence_drv.scratch_reg); + return seq; +} + int radeon_fence_emit(struct radeon_device *rdev, struct radeon_fence *fence) { unsigned long irq_flags; @@ -50,12 +79,12 @@ int radeon_fence_emit(struct radeon_device *rdev, struct radeon_fence *fence) return 0; } fence->seq = atomic_add_return(1, &rdev->fence_drv.seq); - if (!rdev->cp.ready) { + if (!rdev->cp.ready) /* FIXME: cp is not running assume everythings is done right * away */ - WREG32(rdev->fence_drv.scratch_reg, fence->seq); - } else + radeon_fence_write(rdev, fence->seq); + else radeon_fence_ring_emit(rdev, fence); trace_radeon_fence_emit(rdev->ddev, fence->seq); @@ -73,15 +102,7 @@ static bool radeon_fence_poll_locked(struct radeon_device *rdev) bool wake = false; unsigned long cjiffies; - if (rdev->wb.enabled) { - u32 scratch_index; - if (rdev->wb.use_event) - scratch_index = R600_WB_EVENT_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; - else - scratch_index = RADEON_WB_SCRATCH_OFFSET + rdev->fence_drv.scratch_reg - rdev->scratch.reg_base; - seq = le32_to_cpu(rdev->wb.wb[scratch_index/4]); - } else - seq = RREG32(rdev->fence_drv.scratch_reg); + seq = radeon_fence_read(rdev); if (seq != rdev->fence_drv.last_seq) { rdev->fence_drv.last_seq = seq; rdev->fence_drv.last_jiffies = jiffies; @@ -251,7 +272,7 @@ retry: r = radeon_gpu_reset(rdev); if (r) return r; - WREG32(rdev->fence_drv.scratch_reg, fence->seq); + radeon_fence_write(rdev, fence->seq); rdev->gpu_lockup = false; } timeout = RADEON_FENCE_JIFFIES_TIMEOUT; @@ -322,7 +343,7 @@ void radeon_fence_unref(struct radeon_fence **fence) *fence = NULL; if (tmp) { - kref_put(&tmp->kref, &radeon_fence_destroy); + kref_put(&tmp->kref, radeon_fence_destroy); } } @@ -351,7 +372,7 @@ int radeon_fence_driver_init(struct radeon_device *rdev) write_unlock_irqrestore(&rdev->fence_drv.lock, irq_flags); return r; } - WREG32(rdev->fence_drv.scratch_reg, 0); + radeon_fence_write(rdev, 0); atomic_set(&rdev->fence_drv.seq, 0); INIT_LIST_HEAD(&rdev->fence_drv.created); INIT_LIST_HEAD(&rdev->fence_drv.emited); @@ -391,7 +412,7 @@ static int radeon_debugfs_fence_info(struct seq_file *m, void *data) struct radeon_fence *fence; seq_printf(m, "Last signaled fence 0x%08X\n", - RREG32(rdev->fence_drv.scratch_reg)); + radeon_fence_read(rdev)); if (!list_empty(&rdev->fence_drv.emited)) { fence = list_entry(rdev->fence_drv.emited.prev, struct radeon_fence, list); diff --git a/drivers/gpu/drm/radeon/radeon_gart.c b/drivers/gpu/drm/radeon/radeon_gart.c index 8a955bbdb60..a533f52fd16 100644 --- a/drivers/gpu/drm/radeon/radeon_gart.c +++ b/drivers/gpu/drm/radeon/radeon_gart.c @@ -181,9 +181,9 @@ int radeon_gart_bind(struct radeon_device *rdev, unsigned offset, p = t / (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); for (i = 0; i < pages; i++, p++) { - /* On TTM path, we only use the DMA API if TTM_PAGE_FLAG_DMA32 - * is requested. */ - if (dma_addr[i] != DMA_ERROR_CODE) { + /* we reverted the patch using dma_addr in TTM for now but this + * code stops building on alpha so just comment it out for now */ + if (0) { /*dma_addr[i] != DMA_ERROR_CODE) */ rdev->gart.ttm_alloced[p] = true; rdev->gart.pages_addr[p] = dma_addr[i]; } else { diff --git a/drivers/gpu/drm/radeon/radeon_i2c.c b/drivers/gpu/drm/radeon/radeon_i2c.c index 983cbac75af..6c111c1fa3f 100644 --- a/drivers/gpu/drm/radeon/radeon_i2c.c +++ b/drivers/gpu/drm/radeon/radeon_i2c.c @@ -32,17 +32,17 @@ * radeon_ddc_probe * */ -bool radeon_ddc_probe(struct radeon_connector *radeon_connector) +bool radeon_ddc_probe(struct radeon_connector *radeon_connector, bool requires_extended_probe) { - u8 out_buf[] = { 0x0, 0x0}; - u8 buf[2]; + u8 out = 0x0; + u8 buf[8]; int ret; struct i2c_msg msgs[] = { { .addr = 0x50, .flags = 0, .len = 1, - .buf = out_buf, + .buf = &out, }, { .addr = 0x50, @@ -52,15 +52,31 @@ bool radeon_ddc_probe(struct radeon_connector *radeon_connector) } }; + /* Read 8 bytes from i2c for extended probe of EDID header */ + if (requires_extended_probe) + msgs[1].len = 8; + /* on hw with routers, select right port */ if (radeon_connector->router.ddc_valid) radeon_router_select_ddc_port(radeon_connector); ret = i2c_transfer(&radeon_connector->ddc_bus->adapter, msgs, 2); - if (ret == 2) - return true; - - return false; + if (ret != 2) + /* Couldn't find an accessible DDC on this connector */ + return false; + if (requires_extended_probe) { + /* Probe also for valid EDID header + * EDID header starts with: + * 0x00,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0x00. + * Only the first 6 bytes must be valid as + * drm_edid_block_valid() can fix the last 2 bytes */ + if (drm_edid_header_is_valid(buf) < 6) { + /* Couldn't find an accessible EDID on this + * connector */ + return false; + } + } + return true; } /* bit banging i2c */ @@ -888,6 +904,7 @@ struct radeon_i2c_chan *radeon_i2c_create(struct drm_device *dev, i2c->rec = *rec; i2c->adapter.owner = THIS_MODULE; + i2c->adapter.class = I2C_CLASS_DDC; i2c->dev = dev; i2c_set_adapdata(&i2c->adapter, i2c); if (rec->mm_i2c || @@ -947,6 +964,7 @@ struct radeon_i2c_chan *radeon_i2c_create_dp(struct drm_device *dev, i2c->rec = *rec; i2c->adapter.owner = THIS_MODULE; + i2c->adapter.class = I2C_CLASS_DDC; i2c->dev = dev; snprintf(i2c->adapter.name, sizeof(i2c->adapter.name), "Radeon aux bus %s", name); diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index bd58af65858..be2c1224e68 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -60,7 +60,7 @@ int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags) /* update BUS flag */ if (drm_pci_device_is_agp(dev)) { flags |= RADEON_IS_AGP; - } else if (drm_pci_device_is_pcie(dev)) { + } else if (pci_is_pcie(dev->pdev)) { flags |= RADEON_IS_PCIE; } else { flags |= RADEON_IS_PCI; @@ -237,6 +237,19 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) case RADEON_INFO_FUSION_GART_WORKING: value = 1; break; + case RADEON_INFO_BACKEND_MAP: + if (rdev->family >= CHIP_CAYMAN) + value = rdev->config.cayman.backend_map; + else if (rdev->family >= CHIP_CEDAR) + value = rdev->config.evergreen.backend_map; + else if (rdev->family >= CHIP_RV770) + value = rdev->config.rv770.backend_map; + else if (rdev->family >= CHIP_R600) + value = rdev->config.r600.backend_map; + else { + return -EINVAL; + } + break; default: DRM_DEBUG_KMS("Invalid request %d\n", info->request); return -EINVAL; diff --git a/drivers/gpu/drm/radeon/radeon_mem.c b/drivers/gpu/drm/radeon/radeon_mem.c index ed95155c4b1..988548efea9 100644 --- a/drivers/gpu/drm/radeon/radeon_mem.c +++ b/drivers/gpu/drm/radeon/radeon_mem.c @@ -139,7 +139,7 @@ static int init_heap(struct mem_block **heap, int start, int size) if (!blocks) return -ENOMEM; - *heap = kmalloc(sizeof(**heap), GFP_KERNEL); + *heap = kzalloc(sizeof(**heap), GFP_KERNEL); if (!*heap) { kfree(blocks); return -ENOMEM; @@ -150,7 +150,6 @@ static int init_heap(struct mem_block **heap, int start, int size) blocks->file_priv = NULL; blocks->next = blocks->prev = *heap; - memset(*heap, 0, sizeof(**heap)); (*heap)->file_priv = (struct drm_file *) - 1; (*heap)->next = (*heap)->prev = blocks; return 0; diff --git a/drivers/gpu/drm/radeon/radeon_mode.h b/drivers/gpu/drm/radeon/radeon_mode.h index 9c57538231d..68820f5f630 100644 --- a/drivers/gpu/drm/radeon/radeon_mode.h +++ b/drivers/gpu/drm/radeon/radeon_mode.h @@ -438,6 +438,9 @@ struct radeon_connector { struct radeon_i2c_chan *ddc_bus; /* some systems have an hdmi and vga port with a shared ddc line */ bool shared_ddc; + /* for some Radeon chip families we apply an additional EDID header + check as part of the DDC probe */ + bool requires_extended_probe; bool use_digital; /* we need to mind the EDID between detect and get modes due to analog/digital/tvencoder */ @@ -464,22 +467,30 @@ radeon_atombios_get_tv_info(struct radeon_device *rdev); extern struct drm_connector * radeon_get_connector_for_encoder(struct drm_encoder *encoder); +extern bool radeon_encoder_is_dp_bridge(struct drm_encoder *encoder); +extern bool radeon_connector_encoder_is_dp_bridge(struct drm_connector *connector); +extern bool radeon_connector_encoder_is_hbr2(struct drm_connector *connector); +extern bool radeon_connector_is_dp12_capable(struct drm_connector *connector); + extern void radeon_connector_hotplug(struct drm_connector *connector); -extern bool radeon_dp_needs_link_train(struct radeon_connector *radeon_connector); -extern int radeon_dp_mode_valid_helper(struct radeon_connector *radeon_connector, +extern int radeon_dp_mode_valid_helper(struct drm_connector *connector, struct drm_display_mode *mode); extern void radeon_dp_set_link_config(struct drm_connector *connector, struct drm_display_mode *mode); -extern void dp_link_train(struct drm_encoder *encoder, - struct drm_connector *connector); +extern void radeon_dp_link_train(struct drm_encoder *encoder, + struct drm_connector *connector); +extern bool radeon_dp_needs_link_train(struct radeon_connector *radeon_connector); extern u8 radeon_dp_getsinktype(struct radeon_connector *radeon_connector); extern bool radeon_dp_getdpcd(struct radeon_connector *radeon_connector); -extern void atombios_dig_encoder_setup(struct drm_encoder *encoder, int action); +extern void atombios_dig_encoder_setup(struct drm_encoder *encoder, int action, int panel_mode); +extern void radeon_atom_encoder_init(struct radeon_device *rdev); extern void atombios_dig_transmitter_setup(struct drm_encoder *encoder, int action, uint8_t lane_num, uint8_t lane_set); +extern void radeon_atom_ext_encoder_setup_ddc(struct drm_encoder *encoder); +extern struct drm_encoder *radeon_atom_get_external_encoder(struct drm_encoder *encoder); extern int radeon_dp_i2c_aux_ch(struct i2c_adapter *adapter, int mode, - uint8_t write_byte, uint8_t *read_byte); + u8 write_byte, u8 *read_byte); extern void radeon_i2c_init(struct radeon_device *rdev); extern void radeon_i2c_fini(struct radeon_device *rdev); @@ -507,7 +518,8 @@ extern void radeon_i2c_put_byte(struct radeon_i2c_chan *i2c, u8 val); extern void radeon_router_select_ddc_port(struct radeon_connector *radeon_connector); extern void radeon_router_select_cd_port(struct radeon_connector *radeon_connector); -extern bool radeon_ddc_probe(struct radeon_connector *radeon_connector); +extern bool radeon_ddc_probe(struct radeon_connector *radeon_connector, + bool requires_extended_probe); extern int radeon_ddc_get_modes(struct radeon_connector *radeon_connector); extern struct drm_encoder *radeon_best_encoder(struct drm_connector *connector); @@ -545,7 +557,7 @@ struct drm_encoder *radeon_encoder_legacy_tmds_ext_add(struct drm_device *dev, i extern void atombios_dvo_setup(struct drm_encoder *encoder, int action); extern void atombios_digital_setup(struct drm_encoder *encoder, int action); extern int atombios_get_encoder_mode(struct drm_encoder *encoder); -extern void atombios_set_edp_panel_power(struct drm_connector *connector, int action); +extern bool atombios_set_edp_panel_power(struct drm_connector *connector, int action); extern void radeon_encoder_set_active_device(struct drm_encoder *encoder); extern void radeon_crtc_load_lut(struct drm_crtc *crtc); diff --git a/drivers/gpu/drm/radeon/radeon_pm.c b/drivers/gpu/drm/radeon/radeon_pm.c index 86eda1ea94d..6fabe89fa6a 100644 --- a/drivers/gpu/drm/radeon/radeon_pm.c +++ b/drivers/gpu/drm/radeon/radeon_pm.c @@ -487,6 +487,7 @@ static int radeon_hwmon_init(struct radeon_device *rdev) case THERMAL_TYPE_RV6XX: case THERMAL_TYPE_RV770: case THERMAL_TYPE_EVERGREEN: + case THERMAL_TYPE_NI: case THERMAL_TYPE_SUMO: rdev->pm.int_hwmon_dev = hwmon_device_register(rdev->dev); if (IS_ERR(rdev->pm.int_hwmon_dev)) { @@ -593,6 +594,9 @@ int radeon_pm_init(struct radeon_device *rdev) if (rdev->pm.default_vddc) radeon_atom_set_voltage(rdev, rdev->pm.default_vddc, SET_VOLTAGE_TYPE_ASIC_VDDC); + if (rdev->pm.default_vddci) + radeon_atom_set_voltage(rdev, rdev->pm.default_vddci, + SET_VOLTAGE_TYPE_ASIC_VDDCI); if (rdev->pm.default_sclk) radeon_set_engine_clock(rdev, rdev->pm.default_sclk); if (rdev->pm.default_mclk) diff --git a/drivers/gpu/drm/radeon/radeon_reg.h b/drivers/gpu/drm/radeon/radeon_reg.h index ec93a75369e..b4ce8645570 100644 --- a/drivers/gpu/drm/radeon/radeon_reg.h +++ b/drivers/gpu/drm/radeon/radeon_reg.h @@ -300,6 +300,8 @@ # define RADEON_BUS_READ_BURST (1 << 30) #define RADEON_BUS_CNTL1 0x0034 # define RADEON_BUS_WAIT_ON_LOCK_EN (1 << 4) +#define RV370_BUS_CNTL 0x004c +# define RV370_BUS_BIOS_DIS_ROM (1 << 2) /* rv370/rv380, rv410, r423/r430/r480, r5xx */ #define RADEON_MSI_REARM_EN 0x0160 # define RV370_MSI_REARM_EN (1 << 0) @@ -3293,7 +3295,7 @@ # define RADEON_RB_BUFSZ_MASK (0x3f << 0) # define RADEON_RB_BLKSZ_SHIFT 8 # define RADEON_RB_BLKSZ_MASK (0x3f << 8) -# define RADEON_BUF_SWAP_32BIT (1 << 17) +# define RADEON_BUF_SWAP_32BIT (2 << 16) # define RADEON_MAX_FETCH_SHIFT 18 # define RADEON_MAX_FETCH_MASK (0x3 << 18) # define RADEON_RB_NO_UPDATE (1 << 27) diff --git a/drivers/gpu/drm/radeon/radeon_ring.c b/drivers/gpu/drm/radeon/radeon_ring.c index c6776e48fdd..08c0233db1b 100644 --- a/drivers/gpu/drm/radeon/radeon_ring.c +++ b/drivers/gpu/drm/radeon/radeon_ring.c @@ -194,7 +194,7 @@ int radeon_ib_pool_init(struct radeon_device *rdev) r = radeon_bo_kmap(rdev->ib_pool.robj, &ptr); radeon_bo_unreserve(rdev->ib_pool.robj); if (r) { - DRM_ERROR("radeon: failed to map ib poll (%d).\n", r); + DRM_ERROR("radeon: failed to map ib pool (%d).\n", r); return r; } for (i = 0; i < RADEON_IB_POOL_SIZE; i++) { diff --git a/drivers/gpu/drm/radeon/radeon_test.c b/drivers/gpu/drm/radeon/radeon_test.c index dee4a0c1b4b..602fa3541c4 100644 --- a/drivers/gpu/drm/radeon/radeon_test.c +++ b/drivers/gpu/drm/radeon/radeon_test.c @@ -40,10 +40,14 @@ void radeon_test_moves(struct radeon_device *rdev) size = 1024 * 1024; /* Number of tests = - * (Total GTT - IB pool - writeback page - ring buffer) / test size + * (Total GTT - IB pool - writeback page - ring buffers) / test size */ - n = ((u32)(rdev->mc.gtt_size - RADEON_IB_POOL_SIZE*64*1024 - RADEON_GPU_PAGE_SIZE - - rdev->cp.ring_size)) / size; + n = rdev->mc.gtt_size - RADEON_IB_POOL_SIZE*64*1024 - rdev->cp.ring_size; + if (rdev->wb.wb_obj) + n -= RADEON_GPU_PAGE_SIZE; + if (rdev->ih.ring_obj) + n -= rdev->ih.ring_size; + n /= size; gtt_obj = kzalloc(n * sizeof(*gtt_obj), GFP_KERNEL); if (!gtt_obj) { @@ -132,9 +136,15 @@ void radeon_test_moves(struct radeon_device *rdev) gtt_start++, vram_start++) { if (*vram_start != gtt_start) { DRM_ERROR("Incorrect GTT->VRAM copy %d: Got 0x%p, " - "expected 0x%p (GTT map 0x%p-0x%p)\n", - i, *vram_start, gtt_start, gtt_map, - gtt_end); + "expected 0x%p (GTT/VRAM offset " + "0x%16llx/0x%16llx)\n", + i, *vram_start, gtt_start, + (unsigned long long) + (gtt_addr - rdev->mc.gtt_start + + (void*)gtt_start - gtt_map), + (unsigned long long) + (vram_addr - rdev->mc.vram_start + + (void*)gtt_start - gtt_map)); radeon_bo_kunmap(vram_obj); goto out_cleanup; } @@ -175,9 +185,15 @@ void radeon_test_moves(struct radeon_device *rdev) gtt_start++, vram_start++) { if (*gtt_start != vram_start) { DRM_ERROR("Incorrect VRAM->GTT copy %d: Got 0x%p, " - "expected 0x%p (VRAM map 0x%p-0x%p)\n", - i, *gtt_start, vram_start, vram_map, - vram_end); + "expected 0x%p (VRAM/GTT offset " + "0x%16llx/0x%16llx)\n", + i, *gtt_start, vram_start, + (unsigned long long) + (vram_addr - rdev->mc.vram_start + + (void*)vram_start - vram_map), + (unsigned long long) + (gtt_addr - rdev->mc.gtt_start + + (void*)vram_start - vram_map)); radeon_bo_kunmap(gtt_obj[i]); goto out_cleanup; } diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index 60125ddba1e..9b86fb0e412 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -450,6 +450,29 @@ static int radeon_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_ return -EINVAL; mem->bus.base = rdev->mc.aper_base; mem->bus.is_iomem = true; +#ifdef __alpha__ + /* + * Alpha: use bus.addr to hold the ioremap() return, + * so we can modify bus.base below. + */ + if (mem->placement & TTM_PL_FLAG_WC) + mem->bus.addr = + ioremap_wc(mem->bus.base + mem->bus.offset, + mem->bus.size); + else + mem->bus.addr = + ioremap_nocache(mem->bus.base + mem->bus.offset, + mem->bus.size); + + /* + * Alpha: Use just the bus offset plus + * the hose/domain memory base for bus.base. + * It then can be used to build PTEs for VRAM + * access, as done in ttm_bo_vm_fault(). + */ + mem->bus.base = (mem->bus.base & 0x0ffffffffUL) + + rdev->ddev->hose->dense_mem_base; +#endif break; default: return -EINVAL; diff --git a/drivers/gpu/drm/radeon/reg_srcs/cayman b/drivers/gpu/drm/radeon/reg_srcs/cayman index 6334f8ac120..2316977eb92 100644 --- a/drivers/gpu/drm/radeon/reg_srcs/cayman +++ b/drivers/gpu/drm/radeon/reg_srcs/cayman @@ -33,6 +33,7 @@ cayman 0x9400 0x00008E48 SQ_EX_ALLOC_TABLE_SLOTS 0x00009100 SPI_CONFIG_CNTL 0x0000913C SPI_CONFIG_CNTL_1 +0x00009508 TA_CNTL_AUX 0x00009830 DB_DEBUG 0x00009834 DB_DEBUG2 0x00009838 DB_DEBUG3 @@ -207,6 +208,7 @@ cayman 0x9400 0x0002834C PA_SC_VPORT_ZMAX_15 0x00028350 SX_MISC 0x00028354 SX_SURFACE_SYNC +0x0002835C SX_SCATTER_EXPORT_SIZE 0x00028380 SQ_VTX_SEMANTIC_0 0x00028384 SQ_VTX_SEMANTIC_1 0x00028388 SQ_VTX_SEMANTIC_2 @@ -431,6 +433,7 @@ cayman 0x9400 0x00028700 SPI_STACK_MGMT 0x00028704 SPI_WAVE_MGMT_1 0x00028708 SPI_WAVE_MGMT_2 +0x00028720 GDS_ADDR_BASE 0x00028724 GDS_ADDR_SIZE 0x00028780 CB_BLEND0_CONTROL 0x00028784 CB_BLEND1_CONTROL diff --git a/drivers/gpu/drm/radeon/reg_srcs/evergreen b/drivers/gpu/drm/radeon/reg_srcs/evergreen index 7e1637176e0..161737a28c2 100644 --- a/drivers/gpu/drm/radeon/reg_srcs/evergreen +++ b/drivers/gpu/drm/radeon/reg_srcs/evergreen @@ -44,8 +44,10 @@ evergreen 0x9400 0x00008E28 SQ_STATIC_THREAD_MGMT_3 0x00008E2C SQ_LDS_RESOURCE_MGMT 0x00008E48 SQ_EX_ALLOC_TABLE_SLOTS +0x00009014 SX_MEMORY_EXPORT_SIZE 0x00009100 SPI_CONFIG_CNTL 0x0000913C SPI_CONFIG_CNTL_1 +0x00009508 TA_CNTL_AUX 0x00009700 VC_CNTL 0x00009714 VC_ENHANCE 0x00009830 DB_DEBUG @@ -441,7 +443,9 @@ evergreen 0x9400 0x000286EC SPI_COMPUTE_NUM_THREAD_X 0x000286F0 SPI_COMPUTE_NUM_THREAD_Y 0x000286F4 SPI_COMPUTE_NUM_THREAD_Z +0x00028720 GDS_ADDR_BASE 0x00028724 GDS_ADDR_SIZE +0x00028728 GDS_ORDERED_WAVE_PER_SE 0x00028780 CB_BLEND0_CONTROL 0x00028784 CB_BLEND1_CONTROL 0x00028788 CB_BLEND2_CONTROL diff --git a/drivers/gpu/drm/radeon/reg_srcs/r600 b/drivers/gpu/drm/radeon/reg_srcs/r600 index 92f1900dc7c..0380c5c15f8 100644 --- a/drivers/gpu/drm/radeon/reg_srcs/r600 +++ b/drivers/gpu/drm/radeon/reg_srcs/r600 @@ -429,6 +429,7 @@ r600 0x9400 0x00028438 SX_ALPHA_REF 0x00028410 SX_ALPHA_TEST_CONTROL 0x00028350 SX_MISC +0x00009014 SX_MEMORY_EXPORT_SIZE 0x00009604 TC_INVALIDATE 0x00009400 TD_FILTER4 0x00009404 TD_FILTER4_1 @@ -758,6 +759,5 @@ r600 0x9400 0x00009714 VC_ENHANCE 0x00009830 DB_DEBUG 0x00009838 DB_WATERMARKS -0x00028D28 DB_SRESULTS_COMPARE_STATE0 0x00028D44 DB_ALPHA_TO_MASK 0x00009700 VC_CNTL diff --git a/drivers/gpu/drm/radeon/rs600.c b/drivers/gpu/drm/radeon/rs600.c index 6e3b11e5abb..4b5d0e6974a 100644 --- a/drivers/gpu/drm/radeon/rs600.c +++ b/drivers/gpu/drm/radeon/rs600.c @@ -426,7 +426,7 @@ int rs600_gart_init(struct radeon_device *rdev) return radeon_gart_table_vram_alloc(rdev); } -int rs600_gart_enable(struct radeon_device *rdev) +static int rs600_gart_enable(struct radeon_device *rdev) { u32 tmp; int r, i; @@ -440,8 +440,8 @@ int rs600_gart_enable(struct radeon_device *rdev) return r; radeon_gart_restore(rdev); /* Enable bus master */ - tmp = RREG32(R_00004C_BUS_CNTL) & C_00004C_BUS_MASTER_DIS; - WREG32(R_00004C_BUS_CNTL, tmp); + tmp = RREG32(RADEON_BUS_CNTL) & ~RS600_BUS_MASTER_DIS; + WREG32(RADEON_BUS_CNTL, tmp); /* FIXME: setup default page */ WREG32_MC(R_000100_MC_PT0_CNTL, (S_000100_EFFECTIVE_L2_CACHE_SIZE(6) | @@ -530,7 +530,7 @@ int rs600_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) addr = addr & 0xFFFFFFFFFFFFF000ULL; addr |= R600_PTE_VALID | R600_PTE_SYSTEM | R600_PTE_SNOOPED; addr |= R600_PTE_READABLE | R600_PTE_WRITEABLE; - writeq(addr, ((void __iomem *)ptr) + (i * 8)); + writeq(addr, ptr + (i * 8)); return 0; } diff --git a/drivers/gpu/drm/radeon/rv770.c b/drivers/gpu/drm/radeon/rv770.c index ef8a5babe9f..4720d000d44 100644 --- a/drivers/gpu/drm/radeon/rv770.c +++ b/drivers/gpu/drm/radeon/rv770.c @@ -105,6 +105,9 @@ void rv770_pm_misc(struct radeon_device *rdev) struct radeon_voltage *voltage = &ps->clock_info[req_cm_idx].voltage; if ((voltage->type == VOLTAGE_SW) && voltage->voltage) { + /* 0xff01 is a flag rather then an actual voltage */ + if (voltage->voltage == 0xff01) + return; if (voltage->voltage != rdev->pm.current_vddc) { radeon_atom_set_voltage(rdev, voltage->voltage, SET_VOLTAGE_TYPE_ASIC_VDDC); rdev->pm.current_vddc = voltage->voltage; @@ -572,6 +575,12 @@ static void rv770_program_channel_remap(struct radeon_device *rdev) else tcp_chan_steer = 0x00fac688; + /* RV770 CE has special chremap setup */ + if (rdev->pdev->device == 0x944e) { + tcp_chan_steer = 0x00b08b08; + mc_shared_chremap = 0x00b08b08; + } + WREG32(TCP_CHAN_STEER, tcp_chan_steer); WREG32(MC_SHARED_CHREMAP, mc_shared_chremap); } @@ -769,6 +778,7 @@ static void rv770_gpu_init(struct radeon_device *rdev) (cc_rb_backend_disable >> 16)); rdev->config.rv770.tile_config = gb_tiling_config; + rdev->config.rv770.backend_map = backend_map; gb_tiling_config |= BACKEND_MAP(backend_map); WREG32(GB_TILING_CONFIG, gb_tiling_config); @@ -1359,6 +1369,7 @@ void rv770_fini(struct radeon_device *rdev) r700_cp_fini(rdev); r600_irq_fini(rdev); radeon_wb_fini(rdev); + radeon_ib_pool_fini(rdev); radeon_irq_kms_fini(rdev); rv770_pcie_gart_fini(rdev); rv770_vram_scratch_fini(rdev); diff --git a/drivers/gpu/drm/savage/savage_bci.c b/drivers/gpu/drm/savage/savage_bci.c index bf5f83ea14f..cb1ee4e0050 100644 --- a/drivers/gpu/drm/savage/savage_bci.c +++ b/drivers/gpu/drm/savage/savage_bci.c @@ -647,9 +647,6 @@ int savage_driver_firstopen(struct drm_device *dev) ret = drm_addmap(dev, aperture_base, SAVAGE_APERTURE_SIZE, _DRM_FRAME_BUFFER, _DRM_WRITE_COMBINING, &dev_priv->aperture); - if (ret) - return ret; - return ret; } diff --git a/drivers/gpu/drm/sis/sis_drv.h b/drivers/gpu/drm/sis/sis_drv.h index ef940bad63f..194303c177a 100644 --- a/drivers/gpu/drm/sis/sis_drv.h +++ b/drivers/gpu/drm/sis/sis_drv.h @@ -48,8 +48,8 @@ enum sis_family { #define SIS_BASE (dev_priv->mmio) -#define SIS_READ(reg) DRM_READ32(SIS_BASE, reg); -#define SIS_WRITE(reg, val) DRM_WRITE32(SIS_BASE, reg, val); +#define SIS_READ(reg) DRM_READ32(SIS_BASE, reg) +#define SIS_WRITE(reg, val) DRM_WRITE32(SIS_BASE, reg, val) typedef struct drm_sis_private { drm_local_map_t *mmio; diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 2e618b5ac46..a4d38d85909 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -37,7 +37,7 @@ #include <linux/mm.h> #include <linux/file.h> #include <linux/module.h> -#include <asm/atomic.h> +#include <linux/atomic.h> #define TTM_ASSERT_LOCKED(param) #define TTM_DEBUG(fmt, arg...) @@ -353,8 +353,10 @@ static int ttm_bo_add_ttm(struct ttm_buffer_object *bo, bool zero_alloc) ret = ttm_tt_set_user(bo->ttm, current, bo->buffer_start, bo->num_pages); - if (unlikely(ret != 0)) + if (unlikely(ret != 0)) { ttm_tt_destroy(bo->ttm); + bo->ttm = NULL; + } break; default: printk(KERN_ERR TTM_PFX "Illegal buffer object type\n"); @@ -390,10 +392,12 @@ static int ttm_bo_handle_move_mem(struct ttm_buffer_object *bo, * Create and bind a ttm if required. */ - if (!(new_man->flags & TTM_MEMTYPE_FLAG_FIXED) && (bo->ttm == NULL)) { - ret = ttm_bo_add_ttm(bo, false); - if (ret) - goto out_err; + if (!(new_man->flags & TTM_MEMTYPE_FLAG_FIXED)) { + if (bo->ttm == NULL) { + ret = ttm_bo_add_ttm(bo, false); + if (ret) + goto out_err; + } ret = ttm_tt_set_placement_caching(bo->ttm, mem->placement); if (ret) diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c index 77dbf408c0d..ae3c6f5dd2b 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_util.c +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c @@ -635,13 +635,13 @@ int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo, if (ret) return ret; - ttm_bo_free_old_node(bo); if ((man->flags & TTM_MEMTYPE_FLAG_FIXED) && (bo->ttm != NULL)) { ttm_tt_unbind(bo->ttm); ttm_tt_destroy(bo->ttm); bo->ttm = NULL; } + ttm_bo_free_old_node(bo); } else { /** * This should help pipeline ordinary buffer moves. diff --git a/drivers/gpu/drm/ttm/ttm_lock.c b/drivers/gpu/drm/ttm/ttm_lock.c index de41e55a944..075daf44bce 100644 --- a/drivers/gpu/drm/ttm/ttm_lock.c +++ b/drivers/gpu/drm/ttm/ttm_lock.c @@ -30,7 +30,7 @@ #include "ttm/ttm_lock.h" #include "ttm/ttm_module.h" -#include <asm/atomic.h> +#include <linux/atomic.h> #include <linux/errno.h> #include <linux/wait.h> #include <linux/sched.h> diff --git a/drivers/gpu/drm/ttm/ttm_object.c b/drivers/gpu/drm/ttm/ttm_object.c index 75e9d6f86ba..93577f2e295 100644 --- a/drivers/gpu/drm/ttm/ttm_object.c +++ b/drivers/gpu/drm/ttm/ttm_object.c @@ -55,7 +55,7 @@ #include <linux/spinlock.h> #include <linux/slab.h> #include <linux/module.h> -#include <asm/atomic.h> +#include <linux/atomic.h> struct ttm_object_file { struct ttm_object_device *tdev; @@ -206,7 +206,7 @@ void ttm_base_object_unref(struct ttm_base_object **p_base) */ write_lock(&tdev->object_lock); - (void)kref_put(&base->refcount, &ttm_release_base); + kref_put(&base->refcount, ttm_release_base); write_unlock(&tdev->object_lock); } EXPORT_SYMBOL(ttm_base_object_unref); diff --git a/drivers/gpu/drm/ttm/ttm_page_alloc.c b/drivers/gpu/drm/ttm/ttm_page_alloc.c index 9d9d92945f8..727e93daac3 100644 --- a/drivers/gpu/drm/ttm/ttm_page_alloc.c +++ b/drivers/gpu/drm/ttm/ttm_page_alloc.c @@ -40,7 +40,7 @@ #include <linux/slab.h> #include <linux/dma-mapping.h> -#include <asm/atomic.h> +#include <linux/atomic.h> #include "ttm/ttm_bo_driver.h" #include "ttm/ttm_page_alloc.h" @@ -355,7 +355,7 @@ restart: if (nr_free) goto restart; - /* Not allowed to fall tough or break because + /* Not allowed to fall through or break because * following context is inside spinlock while we are * outside here. */ @@ -395,12 +395,14 @@ static int ttm_pool_get_num_unused_pages(void) /** * Callback for mm to request pool to reduce number of page held. */ -static int ttm_pool_mm_shrink(struct shrinker *shrink, int shrink_pages, gfp_t gfp_mask) +static int ttm_pool_mm_shrink(struct shrinker *shrink, + struct shrink_control *sc) { static atomic_t start_pool = ATOMIC_INIT(0); unsigned i; unsigned pool_offset = atomic_add_return(1, &start_pool); struct ttm_page_pool *pool; + int shrink_pages = sc->nr_to_scan; pool_offset = pool_offset % NUM_POOLS; /* select start pool in round robin fashion */ @@ -554,7 +556,7 @@ out: } /** - * Fill the given pool if there isn't enough pages and requested number of + * Fill the given pool if there aren't enough pages and the requested number of * pages is small. */ static void ttm_page_pool_fill_locked(struct ttm_page_pool *pool, @@ -574,8 +576,8 @@ static void ttm_page_pool_fill_locked(struct ttm_page_pool *pool, pool->fill_lock = true; - /* If allocation request is small and there is not enough - * pages in pool we fill the pool first */ + /* If allocation request is small and there are not enough + * pages in a pool we fill the pool up first. */ if (count < _manager->options.small && count > pool->npages) { struct list_head new_pages; @@ -612,9 +614,9 @@ static void ttm_page_pool_fill_locked(struct ttm_page_pool *pool, } /** - * Cut count nubmer of pages from the pool and put them to return list + * Cut 'count' number of pages from the pool and put them on the return list. * - * @return count of pages still to allocate to fill the request. + * @return count of pages still required to fulfill the request. */ static unsigned ttm_page_pool_get_pages(struct ttm_page_pool *pool, struct list_head *pages, int ttm_flags, @@ -635,7 +637,7 @@ static unsigned ttm_page_pool_get_pages(struct ttm_page_pool *pool, goto out; } /* find the last pages to include for requested number of pages. Split - * pool to begin and halves to reduce search space. */ + * pool to begin and halve it to reduce search space. */ if (count <= pool->npages/2) { i = 0; list_for_each(p, &pool->list) { @@ -649,7 +651,7 @@ static unsigned ttm_page_pool_get_pages(struct ttm_page_pool *pool, break; } } - /* Cut count number of pages from pool */ + /* Cut 'count' number of pages from the pool */ list_cut_position(pages, &pool->list, p); pool->npages -= count; count = 0; diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 90e23e0bfad..58c271ebc0f 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -31,6 +31,7 @@ #include <linux/sched.h> #include <linux/highmem.h> #include <linux/pagemap.h> +#include <linux/shmem_fs.h> #include <linux/file.h> #include <linux/swap.h> #include <linux/slab.h> @@ -484,7 +485,7 @@ static int ttm_tt_swapin(struct ttm_tt *ttm) swap_space = swap_storage->f_path.dentry->d_inode->i_mapping; for (i = 0; i < ttm->num_pages; ++i) { - from_page = read_mapping_page(swap_space, i, NULL); + from_page = shmem_read_mapping_page(swap_space, i); if (IS_ERR(from_page)) { ret = PTR_ERR(from_page); goto out_err; @@ -557,7 +558,7 @@ int ttm_tt_swapout(struct ttm_tt *ttm, struct file *persistent_swap_storage) from_page = ttm->pages[i]; if (unlikely(from_page == NULL)) continue; - to_page = read_mapping_page(swap_space, i, NULL); + to_page = shmem_read_mapping_page(swap_space, i); if (unlikely(IS_ERR(to_page))) { ret = PTR_ERR(to_page); goto out_err; diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c b/drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c index f1a52f9e729..07ce02da78a 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c @@ -585,11 +585,10 @@ int vmw_overlay_init(struct vmw_private *dev_priv) return -ENOSYS; } - overlay = kmalloc(sizeof(*overlay), GFP_KERNEL); + overlay = kzalloc(sizeof(*overlay), GFP_KERNEL); if (!overlay) return -ENOMEM; - memset(overlay, 0, sizeof(*overlay)); mutex_init(&overlay->mutex); for (i = 0; i < VMW_MAX_NUM_STREAMS; i++) { overlay->stream[i].buf = NULL; diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c index 5408b1b7996..bfe1bcce7f8 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c @@ -612,11 +612,9 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data, srf->sizes[0].height == 64 && srf->format == SVGA3D_A8R8G8B8) { - srf->snooper.image = kmalloc(64 * 64 * 4, GFP_KERNEL); - /* clear the image */ - if (srf->snooper.image) { - memset(srf->snooper.image, 0x00, 64 * 64 * 4); - } else { + /* allocate image area and clear it */ + srf->snooper.image = kzalloc(64 * 64 * 4, GFP_KERNEL); + if (!srf->snooper.image) { DRM_ERROR("Failed to allocate cursor_image\n"); ret = -ENOMEM; goto out_err1; diff --git a/drivers/gpu/vga/vga_switcheroo.c b/drivers/gpu/vga/vga_switcheroo.c index e01cacba685..58434e804d9 100644 --- a/drivers/gpu/vga/vga_switcheroo.c +++ b/drivers/gpu/vga/vga_switcheroo.c @@ -215,13 +215,9 @@ static int vga_switchoff(struct vga_switcheroo_client *client) /* stage one happens before delay */ static int vga_switchto_stage1(struct vga_switcheroo_client *new_client) { - int ret; int i; struct vga_switcheroo_client *active = NULL; - if (new_client->active == true) - return 0; - for (i = 0; i < VGA_SWITCHEROO_MAX_CLIENTS; i++) { if (vgasr_priv.clients[i].active == true) { active = &vgasr_priv.clients[i]; @@ -231,11 +227,6 @@ static int vga_switchto_stage1(struct vga_switcheroo_client *new_client) if (!active) return 0; - /* power up the first device */ - ret = pci_enable_device(new_client->pdev); - if (ret) - return ret; - if (new_client->pwr_state == VGA_SWITCHEROO_OFF) vga_switchon(new_client); @@ -372,6 +363,9 @@ vga_switcheroo_debugfs_write(struct file *filp, const char __user *ubuf, goto out; } + if (client->active == true) + goto out; + /* okay we want a switch - test if devices are willing to switch */ can_switch = true; for (i = 0; i < VGA_SWITCHEROO_MAX_CLIENTS; i++) { diff --git a/drivers/gpu/vga/vgaarb.c b/drivers/gpu/vga/vgaarb.c index be8d4cb5861..c72f1c0b5e6 100644 --- a/drivers/gpu/vga/vgaarb.c +++ b/drivers/gpu/vga/vgaarb.c @@ -61,7 +61,7 @@ struct vga_device { unsigned int mem_lock_cnt; /* legacy MEM lock count */ unsigned int io_norm_cnt; /* normal IO count */ unsigned int mem_norm_cnt; /* normal MEM count */ - + bool bridge_has_one_vga; /* allow IRQ enable/disable hook */ void *cookie; void (*irq_set_state)(void *cookie, bool enable); @@ -165,6 +165,8 @@ static struct vga_device *__vga_tryget(struct vga_device *vgadev, unsigned int wants, legacy_wants, match; struct vga_device *conflict; unsigned int pci_bits; + u32 flags = 0; + /* Account for "normal" resources to lock. If we decode the legacy, * counterpart, we need to request it as well */ @@ -237,16 +239,23 @@ static struct vga_device *__vga_tryget(struct vga_device *vgadev, /* looks like he doesn't have a lock, we can steal * them from him */ - vga_irq_set_state(conflict, false); + flags = 0; pci_bits = 0; - if (lwants & (VGA_RSRC_LEGACY_MEM|VGA_RSRC_NORMAL_MEM)) - pci_bits |= PCI_COMMAND_MEMORY; - if (lwants & (VGA_RSRC_LEGACY_IO|VGA_RSRC_NORMAL_IO)) - pci_bits |= PCI_COMMAND_IO; - pci_set_vga_state(conflict->pdev, false, pci_bits, - change_bridge); + if (!conflict->bridge_has_one_vga) { + vga_irq_set_state(conflict, false); + flags |= PCI_VGA_STATE_CHANGE_DECODES; + if (lwants & (VGA_RSRC_LEGACY_MEM|VGA_RSRC_NORMAL_MEM)) + pci_bits |= PCI_COMMAND_MEMORY; + if (lwants & (VGA_RSRC_LEGACY_IO|VGA_RSRC_NORMAL_IO)) + pci_bits |= PCI_COMMAND_IO; + } + + if (change_bridge) + flags |= PCI_VGA_STATE_CHANGE_BRIDGE; + + pci_set_vga_state(conflict->pdev, false, pci_bits, flags); conflict->owns &= ~lwants; /* If he also owned non-legacy, that is no longer the case */ if (lwants & VGA_RSRC_LEGACY_MEM) @@ -261,14 +270,24 @@ enable_them: * also have in "decodes". We can lock resources we don't decode but * not own them. */ + flags = 0; pci_bits = 0; - if (wants & (VGA_RSRC_LEGACY_MEM|VGA_RSRC_NORMAL_MEM)) - pci_bits |= PCI_COMMAND_MEMORY; - if (wants & (VGA_RSRC_LEGACY_IO|VGA_RSRC_NORMAL_IO)) - pci_bits |= PCI_COMMAND_IO; - pci_set_vga_state(vgadev->pdev, true, pci_bits, !!(wants & VGA_RSRC_LEGACY_MASK)); - vga_irq_set_state(vgadev, true); + if (!vgadev->bridge_has_one_vga) { + flags |= PCI_VGA_STATE_CHANGE_DECODES; + if (wants & (VGA_RSRC_LEGACY_MEM|VGA_RSRC_NORMAL_MEM)) + pci_bits |= PCI_COMMAND_MEMORY; + if (wants & (VGA_RSRC_LEGACY_IO|VGA_RSRC_NORMAL_IO)) + pci_bits |= PCI_COMMAND_IO; + } + if (!!(wants & VGA_RSRC_LEGACY_MASK)) + flags |= PCI_VGA_STATE_CHANGE_BRIDGE; + + pci_set_vga_state(vgadev->pdev, true, pci_bits, flags); + + if (!vgadev->bridge_has_one_vga) { + vga_irq_set_state(vgadev, true); + } vgadev->owns |= (wants & vgadev->decodes); lock_them: vgadev->locks |= (rsrc & VGA_RSRC_LEGACY_MASK); @@ -421,6 +440,62 @@ bail: } EXPORT_SYMBOL(vga_put); +/* Rules for using a bridge to control a VGA descendant decoding: + if a bridge has only one VGA descendant then it can be used + to control the VGA routing for that device. + It should always use the bridge closest to the device to control it. + If a bridge has a direct VGA descendant, but also have a sub-bridge + VGA descendant then we cannot use that bridge to control the direct VGA descendant. + So for every device we register, we need to iterate all its parent bridges + so we can invalidate any devices using them properly. +*/ +static void vga_arbiter_check_bridge_sharing(struct vga_device *vgadev) +{ + struct vga_device *same_bridge_vgadev; + struct pci_bus *new_bus, *bus; + struct pci_dev *new_bridge, *bridge; + + vgadev->bridge_has_one_vga = true; + + if (list_empty(&vga_list)) + return; + + /* okay iterate the new devices bridge hierarachy */ + new_bus = vgadev->pdev->bus; + while (new_bus) { + new_bridge = new_bus->self; + + if (new_bridge) { + /* go through list of devices already registered */ + list_for_each_entry(same_bridge_vgadev, &vga_list, list) { + bus = same_bridge_vgadev->pdev->bus; + bridge = bus->self; + + /* see if the share a bridge with this device */ + if (new_bridge == bridge) { + /* if their direct parent bridge is the same + as any bridge of this device then it can't be used + for that device */ + same_bridge_vgadev->bridge_has_one_vga = false; + } + + /* now iterate the previous devices bridge hierarchy */ + /* if the new devices parent bridge is in the other devices + hierarchy then we can't use it to control this device */ + while (bus) { + bridge = bus->self; + if (bridge) { + if (bridge == vgadev->pdev->bus->self) + vgadev->bridge_has_one_vga = false; + } + bus = bus->parent; + } + } + } + new_bus = new_bus->parent; + } +} + /* * Currently, we assume that the "initial" setup of the system is * not sane, that is we come up with conflicting devices and let @@ -500,6 +575,8 @@ static bool vga_arbiter_add_pci_device(struct pci_dev *pdev) vga_default = pci_dev_get(pdev); #endif + vga_arbiter_check_bridge_sharing(vgadev); + /* Add to the list */ list_add(&vgadev->list, &vga_list); vga_count++; @@ -1094,10 +1171,9 @@ static int vga_arb_open(struct inode *inode, struct file *file) pr_debug("%s\n", __func__); - priv = kmalloc(sizeof(struct vga_arb_private), GFP_KERNEL); + priv = kzalloc(sizeof(*priv), GFP_KERNEL); if (priv == NULL) return -ENOMEM; - memset(priv, 0, sizeof(*priv)); spin_lock_init(&priv->lock); file->private_data = priv; @@ -1222,6 +1298,7 @@ static int __init vga_arb_device_init(void) { int rc; struct pci_dev *pdev; + struct vga_device *vgadev; rc = misc_register(&vga_arb_device); if (rc < 0) @@ -1238,6 +1315,13 @@ static int __init vga_arb_device_init(void) vga_arbiter_add_pci_device(pdev); pr_info("vgaarb: loaded\n"); + + list_for_each_entry(vgadev, &vga_list, list) { + if (vgadev->bridge_has_one_vga) + pr_info("vgaarb: bridge control possible %s\n", pci_name(vgadev->pdev)); + else + pr_info("vgaarb: no bridge control possible %s\n", pci_name(vgadev->pdev)); + } return rc; } subsys_initcall(vga_arb_device_init); |