diff options
Diffstat (limited to 'drivers/gpu/drm/radeon/r100.c')
| -rw-r--r-- | drivers/gpu/drm/radeon/r100.c | 1725 |
1 files changed, 1330 insertions, 395 deletions
diff --git a/drivers/gpu/drm/radeon/r100.c b/drivers/gpu/drm/radeon/r100.c index 68e728e8be4..161094c07d9 100644 --- a/drivers/gpu/drm/radeon/r100.c +++ b/drivers/gpu/drm/radeon/r100.c @@ -29,23 +29,41 @@ #include "drmP.h" #include "drm.h" #include "radeon_drm.h" -#include "radeon_microcode.h" #include "radeon_reg.h" #include "radeon.h" +#include "r100d.h" +#include "rs100d.h" +#include "rv200d.h" +#include "rv250d.h" + +#include <linux/firmware.h> +#include <linux/platform_device.h> + +#include "r100_reg_safe.h" +#include "rn50_reg_safe.h" + +/* Firmware Names */ +#define FIRMWARE_R100 "radeon/R100_cp.bin" +#define FIRMWARE_R200 "radeon/R200_cp.bin" +#define FIRMWARE_R300 "radeon/R300_cp.bin" +#define FIRMWARE_R420 "radeon/R420_cp.bin" +#define FIRMWARE_RS690 "radeon/RS690_cp.bin" +#define FIRMWARE_RS600 "radeon/RS600_cp.bin" +#define FIRMWARE_R520 "radeon/R520_cp.bin" + +MODULE_FIRMWARE(FIRMWARE_R100); +MODULE_FIRMWARE(FIRMWARE_R200); +MODULE_FIRMWARE(FIRMWARE_R300); +MODULE_FIRMWARE(FIRMWARE_R420); +MODULE_FIRMWARE(FIRMWARE_RS690); +MODULE_FIRMWARE(FIRMWARE_RS600); +MODULE_FIRMWARE(FIRMWARE_R520); + +#include "r100_track.h" /* This files gather functions specifics to: * r100,rv100,rs100,rv200,rs200,r200,rv250,rs300,rv280 - * - * Some of these functions might be used by newer ASICs. */ -void r100_hdp_reset(struct radeon_device *rdev); -void r100_gpu_init(struct radeon_device *rdev); -int r100_gui_wait_for_idle(struct radeon_device *rdev); -int r100_mc_wait_for_idle(struct radeon_device *rdev); -void r100_gpu_wait_for_vsync(struct radeon_device *rdev); -void r100_gpu_wait_for_vsync2(struct radeon_device *rdev); -int r100_debugfs_mc_info_init(struct radeon_device *rdev); - /* * PCI GART @@ -58,23 +76,28 @@ void r100_pci_gart_tlb_flush(struct radeon_device *rdev) * could end up in wrong address. */ } -int r100_pci_gart_enable(struct radeon_device *rdev) +int r100_pci_gart_init(struct radeon_device *rdev) { - uint32_t tmp; int r; + if (rdev->gart.table.ram.ptr) { + WARN(1, "R100 PCI GART already initialized.\n"); + return 0; + } /* Initialize common gart structure */ r = radeon_gart_init(rdev); - if (r) { + if (r) return r; - } - if (rdev->gart.table.ram.ptr == NULL) { - rdev->gart.table_size = rdev->gart.num_gpu_pages * 4; - r = radeon_gart_table_ram_alloc(rdev); - if (r) { - return r; - } - } + rdev->gart.table_size = rdev->gart.num_gpu_pages * 4; + rdev->asic->gart_tlb_flush = &r100_pci_gart_tlb_flush; + rdev->asic->gart_set_page = &r100_pci_gart_set_page; + return radeon_gart_table_ram_alloc(rdev); +} + +int r100_pci_gart_enable(struct radeon_device *rdev) +{ + uint32_t tmp; + /* discard memory request outside of configured range */ tmp = RREG32(RADEON_AIC_CNTL) | RADEON_DIS_OUT_OF_PCI_GART_ACCESS; WREG32(RADEON_AIC_CNTL, tmp); @@ -114,148 +137,13 @@ int r100_pci_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) return 0; } -int r100_gart_enable(struct radeon_device *rdev) -{ - if (rdev->flags & RADEON_IS_AGP) { - r100_pci_gart_disable(rdev); - return 0; - } - return r100_pci_gart_enable(rdev); -} - - -/* - * MC - */ -void r100_mc_disable_clients(struct radeon_device *rdev) -{ - uint32_t ov0_scale_cntl, crtc_ext_cntl, crtc_gen_cntl, crtc2_gen_cntl; - - /* FIXME: is this function correct for rs100,rs200,rs300 ? */ - if (r100_gui_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait GUI idle while " - "programming pipes. Bad things might happen.\n"); - } - - /* stop display and memory access */ - ov0_scale_cntl = RREG32(RADEON_OV0_SCALE_CNTL); - WREG32(RADEON_OV0_SCALE_CNTL, ov0_scale_cntl & ~RADEON_SCALER_ENABLE); - crtc_ext_cntl = RREG32(RADEON_CRTC_EXT_CNTL); - WREG32(RADEON_CRTC_EXT_CNTL, crtc_ext_cntl | RADEON_CRTC_DISPLAY_DIS); - crtc_gen_cntl = RREG32(RADEON_CRTC_GEN_CNTL); - - r100_gpu_wait_for_vsync(rdev); - - WREG32(RADEON_CRTC_GEN_CNTL, - (crtc_gen_cntl & ~(RADEON_CRTC_CUR_EN | RADEON_CRTC_ICON_EN)) | - RADEON_CRTC_DISP_REQ_EN_B | RADEON_CRTC_EXT_DISP_EN); - - if (!(rdev->flags & RADEON_SINGLE_CRTC)) { - crtc2_gen_cntl = RREG32(RADEON_CRTC2_GEN_CNTL); - - r100_gpu_wait_for_vsync2(rdev); - WREG32(RADEON_CRTC2_GEN_CNTL, - (crtc2_gen_cntl & - ~(RADEON_CRTC2_CUR_EN | RADEON_CRTC2_ICON_EN)) | - RADEON_CRTC2_DISP_REQ_EN_B); - } - - udelay(500); -} - -void r100_mc_setup(struct radeon_device *rdev) -{ - uint32_t tmp; - int r; - - r = r100_debugfs_mc_info_init(rdev); - if (r) { - DRM_ERROR("Failed to register debugfs file for R100 MC !\n"); - } - /* Write VRAM size in case we are limiting it */ - WREG32(RADEON_CONFIG_MEMSIZE, rdev->mc.real_vram_size); - /* Novell bug 204882 for RN50/M6/M7 with 8/16/32MB VRAM, - * if the aperture is 64MB but we have 32MB VRAM - * we report only 32MB VRAM but we have to set MC_FB_LOCATION - * to 64MB, otherwise the gpu accidentially dies */ - tmp = rdev->mc.vram_location + rdev->mc.mc_vram_size - 1; - tmp = REG_SET(RADEON_MC_FB_TOP, tmp >> 16); - tmp |= REG_SET(RADEON_MC_FB_START, rdev->mc.vram_location >> 16); - WREG32(RADEON_MC_FB_LOCATION, tmp); - - /* Enable bus mastering */ - tmp = RREG32(RADEON_BUS_CNTL) & ~RADEON_BUS_MASTER_DIS; - WREG32(RADEON_BUS_CNTL, tmp); - - if (rdev->flags & RADEON_IS_AGP) { - tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - 1; - tmp = REG_SET(RADEON_MC_AGP_TOP, tmp >> 16); - tmp |= REG_SET(RADEON_MC_AGP_START, rdev->mc.gtt_location >> 16); - WREG32(RADEON_MC_AGP_LOCATION, tmp); - WREG32(RADEON_AGP_BASE, rdev->mc.agp_base); - } else { - WREG32(RADEON_MC_AGP_LOCATION, 0x0FFFFFFF); - WREG32(RADEON_AGP_BASE, 0); - } - - tmp = RREG32(RADEON_HOST_PATH_CNTL) & RADEON_HDP_APER_CNTL; - tmp |= (7 << 28); - WREG32(RADEON_HOST_PATH_CNTL, tmp | RADEON_HDP_SOFT_RESET | RADEON_HDP_READ_BUFFER_INVALIDATE); - (void)RREG32(RADEON_HOST_PATH_CNTL); - WREG32(RADEON_HOST_PATH_CNTL, tmp); - (void)RREG32(RADEON_HOST_PATH_CNTL); -} - -int r100_mc_init(struct radeon_device *rdev) -{ - int r; - - if (r100_debugfs_rbbm_init(rdev)) { - DRM_ERROR("Failed to register debugfs file for RBBM !\n"); - } - - r100_gpu_init(rdev); - /* Disable gart which also disable out of gart access */ - r100_pci_gart_disable(rdev); - - /* Setup GPU memory space */ - rdev->mc.gtt_location = 0xFFFFFFFFUL; - if (rdev->flags & RADEON_IS_AGP) { - r = radeon_agp_init(rdev); - if (r) { - printk(KERN_WARNING "[drm] Disabling AGP\n"); - rdev->flags &= ~RADEON_IS_AGP; - rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; - } else { - rdev->mc.gtt_location = rdev->mc.agp_base; - } - } - r = radeon_mc_setup(rdev); - if (r) { - return r; - } - - r100_mc_disable_clients(rdev); - if (r100_mc_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait MC idle while " - "programming pipes. Bad things might happen.\n"); - } - - r100_mc_setup(rdev); - return 0; -} - -void r100_mc_fini(struct radeon_device *rdev) +void r100_pci_gart_fini(struct radeon_device *rdev) { r100_pci_gart_disable(rdev); radeon_gart_table_ram_free(rdev); radeon_gart_fini(rdev); } - -/* - * Interrupts - */ int r100_irq_set(struct radeon_device *rdev) { uint32_t tmp = 0; @@ -273,6 +161,17 @@ int r100_irq_set(struct radeon_device *rdev) return 0; } +void r100_irq_disable(struct radeon_device *rdev) +{ + u32 tmp; + + WREG32(R_000040_GEN_INT_CNTL, 0); + /* Wait and acknowledge irq */ + mdelay(1); + tmp = RREG32(R_000044_GEN_INT_STATUS); + WREG32(R_000044_GEN_INT_STATUS, tmp); +} + static inline uint32_t r100_irq_ack(struct radeon_device *rdev) { uint32_t irqs = RREG32(RADEON_GEN_INT_STATUS); @@ -293,6 +192,9 @@ int r100_irq_process(struct radeon_device *rdev) if (!status) { return IRQ_NONE; } + if (rdev->shutdown) { + return IRQ_NONE; + } while (status) { /* SW interrupt */ if (status & RADEON_SW_INT_TEST) { @@ -318,10 +220,6 @@ u32 r100_get_vblank_counter(struct radeon_device *rdev, int crtc) return RREG32(RADEON_CRTC2_CRNT_FRAME); } - -/* - * Fence emission - */ void r100_fence_ring_emit(struct radeon_device *rdev, struct radeon_fence *fence) { @@ -337,10 +235,6 @@ void r100_fence_ring_emit(struct radeon_device *rdev, radeon_ring_write(rdev, RADEON_SW_INT_FIRE); } - -/* - * Writeback - */ int r100_wb_init(struct radeon_device *rdev) { int r; @@ -367,14 +261,21 @@ int r100_wb_init(struct radeon_device *rdev) return r; } } - WREG32(0x774, rdev->wb.gpu_addr); - WREG32(0x70C, rdev->wb.gpu_addr + 1024); - WREG32(0x770, 0xff); + WREG32(R_000774_SCRATCH_ADDR, rdev->wb.gpu_addr); + WREG32(R_00070C_CP_RB_RPTR_ADDR, + S_00070C_RB_RPTR_ADDR((rdev->wb.gpu_addr + 1024) >> 2)); + WREG32(R_000770_SCRATCH_UMSK, 0xff); return 0; } +void r100_wb_disable(struct radeon_device *rdev) +{ + WREG32(R_000770_SCRATCH_UMSK, 0); +} + void r100_wb_fini(struct radeon_device *rdev) { + r100_wb_disable(rdev); if (rdev->wb.wb_obj) { radeon_object_kunmap(rdev->wb.wb_obj); radeon_object_unpin(rdev->wb.wb_obj); @@ -457,10 +358,21 @@ int r100_copy_blit(struct radeon_device *rdev, return r; } +static int r100_cp_wait_for_idle(struct radeon_device *rdev) +{ + unsigned i; + u32 tmp; + + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(R_000E40_RBBM_STATUS); + if (!G_000E40_CP_CMDSTRM_BUSY(tmp)) { + return 0; + } + udelay(1); + } + return -1; +} -/* - * CP - */ void r100_ring_start(struct radeon_device *rdev) { int r; @@ -478,33 +390,33 @@ void r100_ring_start(struct radeon_device *rdev) radeon_ring_unlock_commit(rdev); } -static void r100_cp_load_microcode(struct radeon_device *rdev) + +/* Load the microcode for the CP */ +static int r100_cp_init_microcode(struct radeon_device *rdev) { - int i; + struct platform_device *pdev; + const char *fw_name = NULL; + int err; - if (r100_gui_wait_for_idle(rdev)) { - printk(KERN_WARNING "Failed to wait GUI idle while " - "programming pipes. Bad things might happen.\n"); - } + DRM_DEBUG("\n"); - WREG32(RADEON_CP_ME_RAM_ADDR, 0); + pdev = platform_device_register_simple("radeon_cp", 0, NULL, 0); + err = IS_ERR(pdev); + if (err) { + printk(KERN_ERR "radeon_cp: Failed to register firmware\n"); + return -EINVAL; + } if ((rdev->family == CHIP_R100) || (rdev->family == CHIP_RV100) || (rdev->family == CHIP_RV200) || (rdev->family == CHIP_RS100) || (rdev->family == CHIP_RS200)) { DRM_INFO("Loading R100 Microcode\n"); - for (i = 0; i < 256; i++) { - WREG32(RADEON_CP_ME_RAM_DATAH, R100_cp_microcode[i][1]); - WREG32(RADEON_CP_ME_RAM_DATAL, R100_cp_microcode[i][0]); - } + fw_name = FIRMWARE_R100; } else if ((rdev->family == CHIP_R200) || (rdev->family == CHIP_RV250) || (rdev->family == CHIP_RV280) || (rdev->family == CHIP_RS300)) { DRM_INFO("Loading R200 Microcode\n"); - for (i = 0; i < 256; i++) { - WREG32(RADEON_CP_ME_RAM_DATAH, R200_cp_microcode[i][1]); - WREG32(RADEON_CP_ME_RAM_DATAL, R200_cp_microcode[i][0]); - } + fw_name = FIRMWARE_R200; } else if ((rdev->family == CHIP_R300) || (rdev->family == CHIP_R350) || (rdev->family == CHIP_RV350) || @@ -512,31 +424,19 @@ static void r100_cp_load_microcode(struct radeon_device *rdev) (rdev->family == CHIP_RS400) || (rdev->family == CHIP_RS480)) { DRM_INFO("Loading R300 Microcode\n"); - for (i = 0; i < 256; i++) { - WREG32(RADEON_CP_ME_RAM_DATAH, R300_cp_microcode[i][1]); - WREG32(RADEON_CP_ME_RAM_DATAL, R300_cp_microcode[i][0]); - } + fw_name = FIRMWARE_R300; } else if ((rdev->family == CHIP_R420) || (rdev->family == CHIP_R423) || (rdev->family == CHIP_RV410)) { DRM_INFO("Loading R400 Microcode\n"); - for (i = 0; i < 256; i++) { - WREG32(RADEON_CP_ME_RAM_DATAH, R420_cp_microcode[i][1]); - WREG32(RADEON_CP_ME_RAM_DATAL, R420_cp_microcode[i][0]); - } + fw_name = FIRMWARE_R420; } else if ((rdev->family == CHIP_RS690) || (rdev->family == CHIP_RS740)) { DRM_INFO("Loading RS690/RS740 Microcode\n"); - for (i = 0; i < 256; i++) { - WREG32(RADEON_CP_ME_RAM_DATAH, RS690_cp_microcode[i][1]); - WREG32(RADEON_CP_ME_RAM_DATAL, RS690_cp_microcode[i][0]); - } + fw_name = FIRMWARE_RS690; } else if (rdev->family == CHIP_RS600) { DRM_INFO("Loading RS600 Microcode\n"); - for (i = 0; i < 256; i++) { - WREG32(RADEON_CP_ME_RAM_DATAH, RS600_cp_microcode[i][1]); - WREG32(RADEON_CP_ME_RAM_DATAL, RS600_cp_microcode[i][0]); - } + fw_name = FIRMWARE_RS600; } else if ((rdev->family == CHIP_RV515) || (rdev->family == CHIP_R520) || (rdev->family == CHIP_RV530) || @@ -544,9 +444,44 @@ static void r100_cp_load_microcode(struct radeon_device *rdev) (rdev->family == CHIP_RV560) || (rdev->family == CHIP_RV570)) { DRM_INFO("Loading R500 Microcode\n"); - for (i = 0; i < 256; i++) { - WREG32(RADEON_CP_ME_RAM_DATAH, R520_cp_microcode[i][1]); - WREG32(RADEON_CP_ME_RAM_DATAL, R520_cp_microcode[i][0]); + fw_name = FIRMWARE_R520; + } + + err = request_firmware(&rdev->me_fw, fw_name, &pdev->dev); + platform_device_unregister(pdev); + if (err) { + printk(KERN_ERR "radeon_cp: Failed to load firmware \"%s\"\n", + fw_name); + } else if (rdev->me_fw->size % 8) { + printk(KERN_ERR + "radeon_cp: Bogus length %zu in firmware \"%s\"\n", + rdev->me_fw->size, fw_name); + err = -EINVAL; + release_firmware(rdev->me_fw); + rdev->me_fw = NULL; + } + return err; +} + +static void r100_cp_load_microcode(struct radeon_device *rdev) +{ + const __be32 *fw_data; + int i, size; + + if (r100_gui_wait_for_idle(rdev)) { + printk(KERN_WARNING "Failed to wait GUI idle while " + "programming pipes. Bad things might happen.\n"); + } + + if (rdev->me_fw) { + size = rdev->me_fw->size / 4; + fw_data = (const __be32 *)&rdev->me_fw->data[0]; + WREG32(RADEON_CP_ME_RAM_ADDR, 0); + for (i = 0; i < size; i += 2) { + WREG32(RADEON_CP_ME_RAM_DATAH, + be32_to_cpup(&fw_data[i])); + WREG32(RADEON_CP_ME_RAM_DATAL, + be32_to_cpup(&fw_data[i + 1])); } } } @@ -585,6 +520,15 @@ int r100_cp_init(struct radeon_device *rdev, unsigned ring_size) } else { DRM_INFO("radeon: cp idle (0x%08X)\n", tmp); } + + if (!rdev->me_fw) { + r = r100_cp_init_microcode(rdev); + if (r) { + DRM_ERROR("Failed to load firmware!\n"); + return r; + } + } + /* Align ring size */ rb_bufsz = drm_order(ring_size / 8); ring_size = (1 << (rb_bufsz + 1)) * 4; @@ -658,9 +602,11 @@ int r100_cp_init(struct radeon_device *rdev, unsigned ring_size) void r100_cp_fini(struct radeon_device *rdev) { + if (r100_cp_wait_for_idle(rdev)) { + DRM_ERROR("Wait for CP idle timeout, shutting down CP.\n"); + } /* Disable ring */ - rdev->cp.ready = false; - WREG32(RADEON_CP_CSQ_CNTL, 0); + r100_cp_disable(rdev); radeon_ring_fini(rdev); DRM_INFO("radeon: cp finalized\n"); } @@ -710,6 +656,12 @@ int r100_cp_reset(struct radeon_device *rdev) return -1; } +void r100_cp_commit(struct radeon_device *rdev) +{ + WREG32(RADEON_CP_RB_WPTR, rdev->cp.wptr); + (void)RREG32(RADEON_CP_RB_WPTR); +} + /* * CS functions @@ -762,13 +714,11 @@ int r100_cs_parse_packet0(struct radeon_cs_parser *p, void r100_cs_dump_packet(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_chunk *ib_chunk; volatile uint32_t *ib; unsigned i; unsigned idx; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; idx = pkt->idx; for (i = 0; i <= (pkt->count + 1); i++, idx++) { DRM_INFO("ib[%d]=0x%08X\n", idx, ib[idx]); @@ -795,7 +745,7 @@ int r100_cs_packet_parse(struct radeon_cs_parser *p, idx, ib_chunk->length_dw); return -EINVAL; } - header = ib_chunk->kdata[idx]; + header = radeon_get_ib_value(p, idx); pkt->idx = idx; pkt->type = CP_PACKET_GET_TYPE(header); pkt->count = CP_PACKET_GET_COUNT(header); @@ -838,7 +788,6 @@ int r100_cs_packet_parse(struct radeon_cs_parser *p, */ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) { - struct radeon_cs_chunk *ib_chunk; struct drm_mode_object *obj; struct drm_crtc *crtc; struct radeon_crtc *radeon_crtc; @@ -846,8 +795,9 @@ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) int crtc_id; int r; uint32_t header, h_idx, reg; + volatile uint32_t *ib; - ib_chunk = &p->chunks[p->chunk_ib_idx]; + ib = p->ib->ptr; /* parse the wait until */ r = r100_cs_packet_parse(p, &waitreloc, p->idx); @@ -862,24 +812,24 @@ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) return r; } - if (ib_chunk->kdata[waitreloc.idx + 1] != RADEON_WAIT_CRTC_VLINE) { + if (radeon_get_ib_value(p, waitreloc.idx + 1) != RADEON_WAIT_CRTC_VLINE) { DRM_ERROR("vline wait had illegal wait until\n"); r = -EINVAL; return r; } /* jump over the NOP */ - r = r100_cs_packet_parse(p, &p3reloc, p->idx); + r = r100_cs_packet_parse(p, &p3reloc, p->idx + waitreloc.count + 2); if (r) return r; h_idx = p->idx - 2; - p->idx += waitreloc.count; - p->idx += p3reloc.count; + p->idx += waitreloc.count + 2; + p->idx += p3reloc.count + 2; - header = ib_chunk->kdata[h_idx]; - crtc_id = ib_chunk->kdata[h_idx + 5]; - reg = ib_chunk->kdata[h_idx] >> 2; + header = radeon_get_ib_value(p, h_idx); + crtc_id = radeon_get_ib_value(p, h_idx + 5); + reg = CP_PACKET0_GET_REG(header); mutex_lock(&p->rdev->ddev->mode_config.mutex); obj = drm_mode_object_find(p->rdev->ddev, crtc_id, DRM_MODE_OBJECT_CRTC); if (!obj) { @@ -893,16 +843,16 @@ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) if (!crtc->enabled) { /* if the CRTC isn't enabled - we need to nop out the wait until */ - ib_chunk->kdata[h_idx + 2] = PACKET2(0); - ib_chunk->kdata[h_idx + 3] = PACKET2(0); + ib[h_idx + 2] = PACKET2(0); + ib[h_idx + 3] = PACKET2(0); } else if (crtc_id == 1) { switch (reg) { case AVIVO_D1MODE_VLINE_START_END: - header &= R300_CP_PACKET0_REG_MASK; + header &= ~R300_CP_PACKET0_REG_MASK; header |= AVIVO_D2MODE_VLINE_START_END >> 2; break; case RADEON_CRTC_GUI_TRIG_VLINE: - header &= R300_CP_PACKET0_REG_MASK; + header &= ~R300_CP_PACKET0_REG_MASK; header |= RADEON_CRTC2_GUI_TRIG_VLINE >> 2; break; default: @@ -910,8 +860,8 @@ int r100_cs_packet_parse_vline(struct radeon_cs_parser *p) r = -EINVAL; goto out; } - ib_chunk->kdata[h_idx] = header; - ib_chunk->kdata[h_idx + 3] |= RADEON_ENG_DISPLAY_SELECT_CRTC1; + ib[h_idx] = header; + ib[h_idx + 3] |= RADEON_ENG_DISPLAY_SELECT_CRTC1; } out: mutex_unlock(&p->rdev->ddev->mode_config.mutex); @@ -932,7 +882,6 @@ out: int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, struct radeon_cs_reloc **cs_reloc) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_chunk *relocs_chunk; struct radeon_cs_packet p3reloc; unsigned idx; @@ -943,7 +892,6 @@ int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, return -EINVAL; } *cs_reloc = NULL; - ib_chunk = &p->chunks[p->chunk_ib_idx]; relocs_chunk = &p->chunks[p->chunk_relocs_idx]; r = r100_cs_packet_parse(p, &p3reloc, p->idx); if (r) { @@ -956,7 +904,7 @@ int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, r100_cs_dump_packet(p, &p3reloc); return -EINVAL; } - idx = ib_chunk->kdata[p3reloc.idx + 1]; + idx = radeon_get_ib_value(p, p3reloc.idx + 1); if (idx >= relocs_chunk->length_dw) { DRM_ERROR("Relocs at %d after relocations chunk end %d !\n", idx, relocs_chunk->length_dw); @@ -968,147 +916,357 @@ int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, return 0; } +static int r100_get_vtx_size(uint32_t vtx_fmt) +{ + int vtx_size; + vtx_size = 2; + /* ordered according to bits in spec */ + if (vtx_fmt & RADEON_SE_VTX_FMT_W0) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_FPCOLOR) + vtx_size += 3; + if (vtx_fmt & RADEON_SE_VTX_FMT_FPALPHA) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_PKCOLOR) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_FPSPEC) + vtx_size += 3; + if (vtx_fmt & RADEON_SE_VTX_FMT_FPFOG) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_PKSPEC) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_ST0) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_ST1) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_Q1) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_ST2) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_Q2) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_ST3) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_Q3) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_Q0) + vtx_size++; + /* blend weight */ + if (vtx_fmt & (0x7 << 15)) + vtx_size += (vtx_fmt >> 15) & 0x7; + if (vtx_fmt & RADEON_SE_VTX_FMT_N0) + vtx_size += 3; + if (vtx_fmt & RADEON_SE_VTX_FMT_XY1) + vtx_size += 2; + if (vtx_fmt & RADEON_SE_VTX_FMT_Z1) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_W1) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_N1) + vtx_size++; + if (vtx_fmt & RADEON_SE_VTX_FMT_Z) + vtx_size++; + return vtx_size; +} + static int r100_packet0_check(struct radeon_cs_parser *p, - struct radeon_cs_packet *pkt) + struct radeon_cs_packet *pkt, + unsigned idx, unsigned reg) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_reloc *reloc; + struct r100_cs_track *track; volatile uint32_t *ib; uint32_t tmp; - unsigned reg; - unsigned i; - unsigned idx; - bool onereg; int r; + int i, face; u32 tile_flags = 0; + u32 idx_value; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; - idx = pkt->idx + 1; - reg = pkt->reg; - onereg = false; - if (CP_PACKET0_GET_ONE_REG_WR(ib_chunk->kdata[pkt->idx])) { - onereg = true; - } - for (i = 0; i <= pkt->count; i++, idx++, reg += 4) { - switch (reg) { - case RADEON_CRTC_GUI_TRIG_VLINE: - r = r100_cs_packet_parse_vline(p); - if (r) { - DRM_ERROR("No reloc for ib[%d]=0x%04X\n", - idx, reg); - r100_cs_dump_packet(p, pkt); - return r; - } - break; + track = (struct r100_cs_track *)p->track; + + idx_value = radeon_get_ib_value(p, idx); + + switch (reg) { + case RADEON_CRTC_GUI_TRIG_VLINE: + r = r100_cs_packet_parse_vline(p); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + break; /* FIXME: only allow PACKET3 blit? easier to check for out of * range access */ - case RADEON_DST_PITCH_OFFSET: - case RADEON_SRC_PITCH_OFFSET: - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for ib[%d]=0x%04X\n", - idx, reg); - r100_cs_dump_packet(p, pkt); - return r; - } - tmp = ib_chunk->kdata[idx] & 0x003fffff; - tmp += (((u32)reloc->lobj.gpu_offset) >> 10); - - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) - tile_flags |= RADEON_DST_TILE_MACRO; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) { - if (reg == RADEON_SRC_PITCH_OFFSET) { - DRM_ERROR("Cannot src blit from microtiled surface\n"); - r100_cs_dump_packet(p, pkt); - return -EINVAL; - } - tile_flags |= RADEON_DST_TILE_MICRO; - } + case RADEON_DST_PITCH_OFFSET: + case RADEON_SRC_PITCH_OFFSET: + r = r100_reloc_pitch_offset(p, pkt, idx, reg); + if (r) + return r; + break; + case RADEON_RB3D_DEPTHOFFSET: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->zb.robj = reloc->robj; + track->zb.offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case RADEON_RB3D_COLOROFFSET: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->cb[0].robj = reloc->robj; + track->cb[0].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case RADEON_PP_TXOFFSET_0: + case RADEON_PP_TXOFFSET_1: + case RADEON_PP_TXOFFSET_2: + i = (reg - RADEON_PP_TXOFFSET_0) / 24; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[i].robj = reloc->robj; + break; + case RADEON_PP_CUBIC_OFFSET_T0_0: + case RADEON_PP_CUBIC_OFFSET_T0_1: + case RADEON_PP_CUBIC_OFFSET_T0_2: + case RADEON_PP_CUBIC_OFFSET_T0_3: + case RADEON_PP_CUBIC_OFFSET_T0_4: + i = (reg - RADEON_PP_CUBIC_OFFSET_T0_0) / 4; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->textures[0].cube_info[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[0].cube_info[i].robj = reloc->robj; + break; + case RADEON_PP_CUBIC_OFFSET_T1_0: + case RADEON_PP_CUBIC_OFFSET_T1_1: + case RADEON_PP_CUBIC_OFFSET_T1_2: + case RADEON_PP_CUBIC_OFFSET_T1_3: + case RADEON_PP_CUBIC_OFFSET_T1_4: + i = (reg - RADEON_PP_CUBIC_OFFSET_T1_0) / 4; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->textures[1].cube_info[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[1].cube_info[i].robj = reloc->robj; + break; + case RADEON_PP_CUBIC_OFFSET_T2_0: + case RADEON_PP_CUBIC_OFFSET_T2_1: + case RADEON_PP_CUBIC_OFFSET_T2_2: + case RADEON_PP_CUBIC_OFFSET_T2_3: + case RADEON_PP_CUBIC_OFFSET_T2_4: + i = (reg - RADEON_PP_CUBIC_OFFSET_T2_0) / 4; + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + track->textures[2].cube_info[i].offset = idx_value; + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + track->textures[2].cube_info[i].robj = reloc->robj; + break; + case RADEON_RE_WIDTH_HEIGHT: + track->maxy = ((idx_value >> 16) & 0x7FF); + break; + case RADEON_RB3D_COLORPITCH: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } - tmp |= tile_flags; - ib[idx] = (ib_chunk->kdata[idx] & 0x3fc00000) | tmp; - break; - case RADEON_RB3D_DEPTHOFFSET: - case RADEON_RB3D_COLOROFFSET: - case R300_RB3D_COLOROFFSET0: - case R300_ZB_DEPTHOFFSET: - case R200_PP_TXOFFSET_0: - case R200_PP_TXOFFSET_1: - case R200_PP_TXOFFSET_2: - case R200_PP_TXOFFSET_3: - case R200_PP_TXOFFSET_4: - case R200_PP_TXOFFSET_5: - case RADEON_PP_TXOFFSET_0: - case RADEON_PP_TXOFFSET_1: - case RADEON_PP_TXOFFSET_2: - case R300_TX_OFFSET_0: - case R300_TX_OFFSET_0+4: - case R300_TX_OFFSET_0+8: - case R300_TX_OFFSET_0+12: - case R300_TX_OFFSET_0+16: - case R300_TX_OFFSET_0+20: - case R300_TX_OFFSET_0+24: - case R300_TX_OFFSET_0+28: - case R300_TX_OFFSET_0+32: - case R300_TX_OFFSET_0+36: - case R300_TX_OFFSET_0+40: - case R300_TX_OFFSET_0+44: - case R300_TX_OFFSET_0+48: - case R300_TX_OFFSET_0+52: - case R300_TX_OFFSET_0+56: - case R300_TX_OFFSET_0+60: - /* rn50 has no 3D engine so fail on any 3d setup */ - if (ASIC_IS_RN50(p->rdev)) { - DRM_ERROR("attempt to use RN50 3D engine failed\n"); - return -EINVAL; - } - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for ib[%d]=0x%04X\n", - idx, reg); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); - break; - case R300_RB3D_COLORPITCH0: - case RADEON_RB3D_COLORPITCH: - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for ib[%d]=0x%04X\n", - idx, reg); - r100_cs_dump_packet(p, pkt); - return r; - } + if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) + tile_flags |= RADEON_COLOR_TILE_ENABLE; + if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) + tile_flags |= RADEON_COLOR_MICROTILE_ENABLE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MACRO) - tile_flags |= RADEON_COLOR_TILE_ENABLE; - if (reloc->lobj.tiling_flags & RADEON_TILING_MICRO) - tile_flags |= RADEON_COLOR_MICROTILE_ENABLE; + tmp = idx_value & ~(0x7 << 16); + tmp |= tile_flags; + ib[idx] = tmp; - tmp = ib_chunk->kdata[idx] & ~(0x7 << 16); - tmp |= tile_flags; - ib[idx] = tmp; + track->cb[0].pitch = idx_value & RADEON_COLORPITCH_MASK; + break; + case RADEON_RB3D_DEPTHPITCH: + track->zb.pitch = idx_value & RADEON_DEPTHPITCH_MASK; + break; + case RADEON_RB3D_CNTL: + switch ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f) { + case 7: + case 8: + case 9: + case 11: + case 12: + track->cb[0].cpp = 1; break; - case RADEON_RB3D_ZPASS_ADDR: - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for ib[%d]=0x%04X\n", - idx, reg); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + case 3: + case 4: + case 15: + track->cb[0].cpp = 2; + break; + case 6: + track->cb[0].cpp = 4; + break; + default: + DRM_ERROR("Invalid color buffer format (%d) !\n", + ((idx_value >> RADEON_RB3D_COLOR_FORMAT_SHIFT) & 0x1f)); + return -EINVAL; + } + track->z_enabled = !!(idx_value & RADEON_Z_ENABLE); + break; + case RADEON_RB3D_ZSTENCILCNTL: + switch (idx_value & 0xf) { + case 0: + track->zb.cpp = 2; + break; + case 2: + case 3: + case 4: + case 5: + case 9: + case 11: + track->zb.cpp = 4; break; default: - /* FIXME: we don't want to allow anyothers packet */ break; } - if (onereg) { - /* FIXME: forbid onereg write to register on relocate */ + break; + case RADEON_RB3D_ZPASS_ADDR: + r = r100_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("No reloc for ib[%d]=0x%04X\n", + idx, reg); + r100_cs_dump_packet(p, pkt); + return r; + } + ib[idx] = idx_value + ((u32)reloc->lobj.gpu_offset); + break; + case RADEON_PP_CNTL: + { + uint32_t temp = idx_value >> 4; + for (i = 0; i < track->num_texture; i++) + track->textures[i].enabled = !!(temp & (1 << i)); + } + break; + case RADEON_SE_VF_CNTL: + track->vap_vf_cntl = idx_value; + break; + case RADEON_SE_VTX_FMT: + track->vtx_size = r100_get_vtx_size(idx_value); + break; + case RADEON_PP_TEX_SIZE_0: + case RADEON_PP_TEX_SIZE_1: + case RADEON_PP_TEX_SIZE_2: + i = (reg - RADEON_PP_TEX_SIZE_0) / 8; + track->textures[i].width = (idx_value & RADEON_TEX_USIZE_MASK) + 1; + track->textures[i].height = ((idx_value & RADEON_TEX_VSIZE_MASK) >> RADEON_TEX_VSIZE_SHIFT) + 1; + break; + case RADEON_PP_TEX_PITCH_0: + case RADEON_PP_TEX_PITCH_1: + case RADEON_PP_TEX_PITCH_2: + i = (reg - RADEON_PP_TEX_PITCH_0) / 8; + track->textures[i].pitch = idx_value + 32; + break; + case RADEON_PP_TXFILTER_0: + case RADEON_PP_TXFILTER_1: + case RADEON_PP_TXFILTER_2: + i = (reg - RADEON_PP_TXFILTER_0) / 24; + track->textures[i].num_levels = ((idx_value & RADEON_MAX_MIP_LEVEL_MASK) + >> RADEON_MAX_MIP_LEVEL_SHIFT); + tmp = (idx_value >> 23) & 0x7; + if (tmp == 2 || tmp == 6) + track->textures[i].roundup_w = false; + tmp = (idx_value >> 27) & 0x7; + if (tmp == 2 || tmp == 6) + track->textures[i].roundup_h = false; + break; + case RADEON_PP_TXFORMAT_0: + case RADEON_PP_TXFORMAT_1: + case RADEON_PP_TXFORMAT_2: + i = (reg - RADEON_PP_TXFORMAT_0) / 24; + if (idx_value & RADEON_TXFORMAT_NON_POWER2) { + track->textures[i].use_pitch = 1; + } else { + track->textures[i].use_pitch = 0; + track->textures[i].width = 1 << ((idx_value >> RADEON_TXFORMAT_WIDTH_SHIFT) & RADEON_TXFORMAT_WIDTH_MASK); + track->textures[i].height = 1 << ((idx_value >> RADEON_TXFORMAT_HEIGHT_SHIFT) & RADEON_TXFORMAT_HEIGHT_MASK); + } + if (idx_value & RADEON_TXFORMAT_CUBIC_MAP_ENABLE) + track->textures[i].tex_coord_type = 2; + switch ((idx_value & RADEON_TXFORMAT_FORMAT_MASK)) { + case RADEON_TXFORMAT_I8: + case RADEON_TXFORMAT_RGB332: + case RADEON_TXFORMAT_Y8: + track->textures[i].cpp = 1; + break; + case RADEON_TXFORMAT_AI88: + case RADEON_TXFORMAT_ARGB1555: + case RADEON_TXFORMAT_RGB565: + case RADEON_TXFORMAT_ARGB4444: + case RADEON_TXFORMAT_VYUY422: + case RADEON_TXFORMAT_YVYU422: + case RADEON_TXFORMAT_DXT1: + case RADEON_TXFORMAT_SHADOW16: + case RADEON_TXFORMAT_LDUDV655: + case RADEON_TXFORMAT_DUDV88: + track->textures[i].cpp = 2; + break; + case RADEON_TXFORMAT_ARGB8888: + case RADEON_TXFORMAT_RGBA8888: + case RADEON_TXFORMAT_DXT23: + case RADEON_TXFORMAT_DXT45: + case RADEON_TXFORMAT_SHADOW32: + case RADEON_TXFORMAT_LDUDUV8888: + track->textures[i].cpp = 4; break; } + track->textures[i].cube_info[4].width = 1 << ((idx_value >> 16) & 0xf); + track->textures[i].cube_info[4].height = 1 << ((idx_value >> 20) & 0xf); + break; + case RADEON_PP_CUBIC_FACES_0: + case RADEON_PP_CUBIC_FACES_1: + case RADEON_PP_CUBIC_FACES_2: + tmp = idx_value; + i = (reg - RADEON_PP_CUBIC_FACES_0) / 4; + for (face = 0; face < 4; face++) { + track->textures[i].cube_info[face].width = 1 << ((tmp >> (face * 8)) & 0xf); + track->textures[i].cube_info[face].height = 1 << ((tmp >> ((face * 8) + 4)) & 0xf); + } + break; + default: + printk(KERN_ERR "Forbidden register 0x%04X in cs at %d\n", + reg, idx); + return -EINVAL; } return 0; } @@ -1117,15 +1275,14 @@ int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt, struct radeon_object *robj) { - struct radeon_cs_chunk *ib_chunk; unsigned idx; - - ib_chunk = &p->chunks[p->chunk_ib_idx]; + u32 value; idx = pkt->idx + 1; - if ((ib_chunk->kdata[idx+2] + 1) > radeon_object_size(robj)) { + value = radeon_get_ib_value(p, idx + 2); + if ((value + 1) > radeon_object_size(robj)) { DRM_ERROR("[drm] Buffer too small for PACKET3 INDX_BUFFER " "(need %u have %lu) !\n", - ib_chunk->kdata[idx+2] + 1, + value + 1, radeon_object_size(robj)); return -EINVAL; } @@ -1135,47 +1292,20 @@ int r100_cs_track_check_pkt3_indx_buffer(struct radeon_cs_parser *p, static int r100_packet3_check(struct radeon_cs_parser *p, struct radeon_cs_packet *pkt) { - struct radeon_cs_chunk *ib_chunk; struct radeon_cs_reloc *reloc; + struct r100_cs_track *track; unsigned idx; - unsigned i, c; volatile uint32_t *ib; int r; ib = p->ib->ptr; - ib_chunk = &p->chunks[p->chunk_ib_idx]; idx = pkt->idx + 1; + track = (struct r100_cs_track *)p->track; switch (pkt->opcode) { case PACKET3_3D_LOAD_VBPNTR: - c = ib_chunk->kdata[idx++]; - for (i = 0; i < (c - 1); i += 2, idx += 3) { - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for packet3 %d\n", - pkt->opcode); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for packet3 %d\n", - pkt->opcode); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx+2] = ib_chunk->kdata[idx+2] + ((u32)reloc->lobj.gpu_offset); - } - if (c & 1) { - r = r100_cs_packet_next_reloc(p, &reloc); - if (r) { - DRM_ERROR("No reloc for packet3 %d\n", - pkt->opcode); - r100_cs_dump_packet(p, pkt); - return r; - } - ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); - } + r = r100_packet3_load_vbpntr(p, pkt, idx); + if (r) + return r; break; case PACKET3_INDX_BUFFER: r = r100_cs_packet_next_reloc(p, &reloc); @@ -1184,14 +1314,13 @@ static int r100_packet3_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); + ib[idx+1] = radeon_get_ib_value(p, idx+1) + ((u32)reloc->lobj.gpu_offset); r = r100_cs_track_check_pkt3_indx_buffer(p, pkt, reloc->robj); if (r) { return r; } break; case 0x23: - /* FIXME: cleanup */ /* 3D_RNDR_GEN_INDX_PRIM on r100/r200 */ r = r100_cs_packet_next_reloc(p, &reloc); if (r) { @@ -1199,19 +1328,72 @@ static int r100_packet3_check(struct radeon_cs_parser *p, r100_cs_dump_packet(p, pkt); return r; } - ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); + ib[idx] = radeon_get_ib_value(p, idx) + ((u32)reloc->lobj.gpu_offset); + track->num_arrays = 1; + track->vtx_size = r100_get_vtx_size(radeon_get_ib_value(p, idx + 2)); + + track->arrays[0].robj = reloc->robj; + track->arrays[0].esize = track->vtx_size; + + track->max_indx = radeon_get_ib_value(p, idx+1); + + track->vap_vf_cntl = radeon_get_ib_value(p, idx+3); + track->immd_dwords = pkt->count - 1; + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; break; case PACKET3_3D_DRAW_IMMD: + if (((radeon_get_ib_value(p, idx + 1) >> 4) & 0x3) != 3) { + DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); + return -EINVAL; + } + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); + track->immd_dwords = pkt->count - 1; + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; /* triggers drawing using in-packet vertex data */ case PACKET3_3D_DRAW_IMMD_2: + if (((radeon_get_ib_value(p, idx) >> 4) & 0x3) != 3) { + DRM_ERROR("PRIM_WALK must be 3 for IMMD draw\n"); + return -EINVAL; + } + track->vap_vf_cntl = radeon_get_ib_value(p, idx); + track->immd_dwords = pkt->count; + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; /* triggers drawing using in-packet vertex data */ case PACKET3_3D_DRAW_VBUF_2: + track->vap_vf_cntl = radeon_get_ib_value(p, idx); + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; /* triggers drawing of vertex buffers setup elsewhere */ case PACKET3_3D_DRAW_INDX_2: + track->vap_vf_cntl = radeon_get_ib_value(p, idx); + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; /* triggers drawing using indices to vertex buffer */ case PACKET3_3D_DRAW_VBUF: + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; /* triggers drawing of vertex buffers setup elsewhere */ case PACKET3_3D_DRAW_INDX: + track->vap_vf_cntl = radeon_get_ib_value(p, idx + 1); + r = r100_cs_track_check(p->rdev, track); + if (r) + return r; + break; /* triggers drawing using indices to vertex buffer */ case PACKET3_NOP: break; @@ -1225,8 +1407,12 @@ static int r100_packet3_check(struct radeon_cs_parser *p, int r100_cs_parse(struct radeon_cs_parser *p) { struct radeon_cs_packet pkt; + struct r100_cs_track *track; int r; + track = kzalloc(sizeof(*track), GFP_KERNEL); + r100_cs_track_clear(p->rdev, track); + p->track = track; do { r = r100_cs_packet_parse(p, &pkt, p->idx); if (r) { @@ -1235,7 +1421,16 @@ int r100_cs_parse(struct radeon_cs_parser *p) p->idx += pkt.count + 2; switch (pkt.type) { case PACKET_TYPE0: - r = r100_packet0_check(p, &pkt); + if (p->rdev->family >= CHIP_R200) + r = r100_cs_parse_packet0(p, &pkt, + p->rdev->config.r100.reg_safe_bm, + p->rdev->config.r100.reg_safe_bm_size, + &r200_packet0_check); + else + r = r100_cs_parse_packet0(p, &pkt, + p->rdev->config.r100.reg_safe_bm, + p->rdev->config.r100.reg_safe_bm_size, + &r100_packet0_check); break; case PACKET_TYPE2: break; @@ -1568,6 +1763,20 @@ void r100_vram_init_sizes(struct radeon_device *rdev) rdev->mc.real_vram_size = rdev->mc.aper_size; } +void r100_vga_set_state(struct radeon_device *rdev, bool state) +{ + uint32_t temp; + + temp = RREG32(RADEON_CONFIG_CNTL); + if (state == false) { + temp &= ~(1<<8); + temp |= (1<<9); + } else { + temp &= ~(1<<9); + } + WREG32(RADEON_CONFIG_CNTL, temp); +} + void r100_vram_info(struct radeon_device *rdev) { r100_vram_get_type(rdev); @@ -1632,9 +1841,17 @@ void r100_pll_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) r100_pll_errata_after_data(rdev); } -int r100_init(struct radeon_device *rdev) +void r100_set_safe_registers(struct radeon_device *rdev) { - return 0; + if (ASIC_IS_RN50(rdev)) { + rdev->config.r100.reg_safe_bm = rn50_reg_safe_bm; + rdev->config.r100.reg_safe_bm_size = ARRAY_SIZE(rn50_reg_safe_bm); + } else if (rdev->family < CHIP_R200) { + rdev->config.r100.reg_safe_bm = r100_reg_safe_bm; + rdev->config.r100.reg_safe_bm_size = ARRAY_SIZE(r100_reg_safe_bm); + } else { + r200_set_safe_registers(rdev); + } } /* @@ -1839,6 +2056,11 @@ int r100_set_surface_reg(struct radeon_device *rdev, int reg, flags |= R300_SURF_TILE_MICRO; } + if (tiling_flags & RADEON_TILING_SWAP_16BIT) + flags |= RADEON_SURF_AP0_SWP_16BPP | RADEON_SURF_AP1_SWP_16BPP; + if (tiling_flags & RADEON_TILING_SWAP_32BIT) + flags |= RADEON_SURF_AP0_SWP_32BPP | RADEON_SURF_AP1_SWP_32BPP; + DRM_DEBUG("writing surface %d %d %x %x\n", reg, flags, offset, offset+obj_size-1); WREG32(RADEON_SURFACE0_INFO + surf_index, flags); WREG32(RADEON_SURFACE0_LOWER_BOUND + surf_index, offset); @@ -1927,9 +2149,11 @@ void r100_bandwidth_update(struct radeon_device *rdev) mode1 = &rdev->mode_info.crtcs[0]->base.mode; pixel_bytes1 = rdev->mode_info.crtcs[0]->base.fb->bits_per_pixel / 8; } - if (rdev->mode_info.crtcs[1]->base.enabled) { - mode2 = &rdev->mode_info.crtcs[1]->base.mode; - pixel_bytes2 = rdev->mode_info.crtcs[1]->base.fb->bits_per_pixel / 8; + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + if (rdev->mode_info.crtcs[1]->base.enabled) { + mode2 = &rdev->mode_info.crtcs[1]->base.mode; + pixel_bytes2 = rdev->mode_info.crtcs[1]->base.fb->bits_per_pixel / 8; + } } min_mem_eff.full = rfixed_const_8(0); @@ -2334,3 +2558,714 @@ void r100_bandwidth_update(struct radeon_device *rdev) (unsigned int)RREG32(RADEON_GRPH2_BUFFER_CNTL)); } } + +static inline void r100_cs_track_texture_print(struct r100_cs_track_texture *t) +{ + DRM_ERROR("pitch %d\n", t->pitch); + DRM_ERROR("width %d\n", t->width); + DRM_ERROR("height %d\n", t->height); + DRM_ERROR("num levels %d\n", t->num_levels); + DRM_ERROR("depth %d\n", t->txdepth); + DRM_ERROR("bpp %d\n", t->cpp); + DRM_ERROR("coordinate type %d\n", t->tex_coord_type); + DRM_ERROR("width round to power of 2 %d\n", t->roundup_w); + DRM_ERROR("height round to power of 2 %d\n", t->roundup_h); +} + +static int r100_cs_track_cube(struct radeon_device *rdev, + struct r100_cs_track *track, unsigned idx) +{ + unsigned face, w, h; + struct radeon_object *cube_robj; + unsigned long size; + + for (face = 0; face < 5; face++) { + cube_robj = track->textures[idx].cube_info[face].robj; + w = track->textures[idx].cube_info[face].width; + h = track->textures[idx].cube_info[face].height; + + size = w * h; + size *= track->textures[idx].cpp; + + size += track->textures[idx].cube_info[face].offset; + + if (size > radeon_object_size(cube_robj)) { + DRM_ERROR("Cube texture offset greater than object size %lu %lu\n", + size, radeon_object_size(cube_robj)); + r100_cs_track_texture_print(&track->textures[idx]); + return -1; + } + } + return 0; +} + +static int r100_cs_track_texture_check(struct radeon_device *rdev, + struct r100_cs_track *track) +{ + struct radeon_object *robj; + unsigned long size; + unsigned u, i, w, h; + int ret; + + for (u = 0; u < track->num_texture; u++) { + if (!track->textures[u].enabled) + continue; + robj = track->textures[u].robj; + if (robj == NULL) { + DRM_ERROR("No texture bound to unit %u\n", u); + return -EINVAL; + } + size = 0; + for (i = 0; i <= track->textures[u].num_levels; i++) { + if (track->textures[u].use_pitch) { + if (rdev->family < CHIP_R300) + w = (track->textures[u].pitch / track->textures[u].cpp) / (1 << i); + else + w = track->textures[u].pitch / (1 << i); + } else { + w = track->textures[u].width / (1 << i); + if (rdev->family >= CHIP_RV515) + w |= track->textures[u].width_11; + if (track->textures[u].roundup_w) + w = roundup_pow_of_two(w); + } + h = track->textures[u].height / (1 << i); + if (rdev->family >= CHIP_RV515) + h |= track->textures[u].height_11; + if (track->textures[u].roundup_h) + h = roundup_pow_of_two(h); + size += w * h; + } + size *= track->textures[u].cpp; + switch (track->textures[u].tex_coord_type) { + case 0: + break; + case 1: + size *= (1 << track->textures[u].txdepth); + break; + case 2: + if (track->separate_cube) { + ret = r100_cs_track_cube(rdev, track, u); + if (ret) + return ret; + } else + size *= 6; + break; + default: + DRM_ERROR("Invalid texture coordinate type %u for unit " + "%u\n", track->textures[u].tex_coord_type, u); + return -EINVAL; + } + if (size > radeon_object_size(robj)) { + DRM_ERROR("Texture of unit %u needs %lu bytes but is " + "%lu\n", u, size, radeon_object_size(robj)); + r100_cs_track_texture_print(&track->textures[u]); + return -EINVAL; + } + } + return 0; +} + +int r100_cs_track_check(struct radeon_device *rdev, struct r100_cs_track *track) +{ + unsigned i; + unsigned long size; + unsigned prim_walk; + unsigned nverts; + + for (i = 0; i < track->num_cb; i++) { + if (track->cb[i].robj == NULL) { + DRM_ERROR("[drm] No buffer for color buffer %d !\n", i); + return -EINVAL; + } + size = track->cb[i].pitch * track->cb[i].cpp * track->maxy; + size += track->cb[i].offset; + if (size > radeon_object_size(track->cb[i].robj)) { + DRM_ERROR("[drm] Buffer too small for color buffer %d " + "(need %lu have %lu) !\n", i, size, + radeon_object_size(track->cb[i].robj)); + DRM_ERROR("[drm] color buffer %d (%u %u %u %u)\n", + i, track->cb[i].pitch, track->cb[i].cpp, + track->cb[i].offset, track->maxy); + return -EINVAL; + } + } + if (track->z_enabled) { + if (track->zb.robj == NULL) { + DRM_ERROR("[drm] No buffer for z buffer !\n"); + return -EINVAL; + } + size = track->zb.pitch * track->zb.cpp * track->maxy; + size += track->zb.offset; + if (size > radeon_object_size(track->zb.robj)) { + DRM_ERROR("[drm] Buffer too small for z buffer " + "(need %lu have %lu) !\n", size, + radeon_object_size(track->zb.robj)); + DRM_ERROR("[drm] zbuffer (%u %u %u %u)\n", + track->zb.pitch, track->zb.cpp, + track->zb.offset, track->maxy); + return -EINVAL; + } + } + prim_walk = (track->vap_vf_cntl >> 4) & 0x3; + nverts = (track->vap_vf_cntl >> 16) & 0xFFFF; + switch (prim_walk) { + case 1: + for (i = 0; i < track->num_arrays; i++) { + size = track->arrays[i].esize * track->max_indx * 4; + if (track->arrays[i].robj == NULL) { + DRM_ERROR("(PW %u) Vertex array %u no buffer " + "bound\n", prim_walk, i); + return -EINVAL; + } + if (size > radeon_object_size(track->arrays[i].robj)) { + DRM_ERROR("(PW %u) Vertex array %u need %lu dwords " + "have %lu dwords\n", prim_walk, i, + size >> 2, + radeon_object_size(track->arrays[i].robj) >> 2); + DRM_ERROR("Max indices %u\n", track->max_indx); + return -EINVAL; + } + } + break; + case 2: + for (i = 0; i < track->num_arrays; i++) { + size = track->arrays[i].esize * (nverts - 1) * 4; + if (track->arrays[i].robj == NULL) { + DRM_ERROR("(PW %u) Vertex array %u no buffer " + "bound\n", prim_walk, i); + return -EINVAL; + } + if (size > radeon_object_size(track->arrays[i].robj)) { + DRM_ERROR("(PW %u) Vertex array %u need %lu dwords " + "have %lu dwords\n", prim_walk, i, size >> 2, + radeon_object_size(track->arrays[i].robj) >> 2); + return -EINVAL; + } + } + break; + case 3: + size = track->vtx_size * nverts; + if (size != track->immd_dwords) { + DRM_ERROR("IMMD draw %u dwors but needs %lu dwords\n", + track->immd_dwords, size); + DRM_ERROR("VAP_VF_CNTL.NUM_VERTICES %u, VTX_SIZE %u\n", + nverts, track->vtx_size); + return -EINVAL; + } + break; + default: + DRM_ERROR("[drm] Invalid primitive walk %d for VAP_VF_CNTL\n", + prim_walk); + return -EINVAL; + } + return r100_cs_track_texture_check(rdev, track); +} + +void r100_cs_track_clear(struct radeon_device *rdev, struct r100_cs_track *track) +{ + unsigned i, face; + + if (rdev->family < CHIP_R300) { + track->num_cb = 1; + if (rdev->family <= CHIP_RS200) + track->num_texture = 3; + else + track->num_texture = 6; + track->maxy = 2048; + track->separate_cube = 1; + } else { + track->num_cb = 4; + track->num_texture = 16; + track->maxy = 4096; + track->separate_cube = 0; + } + + for (i = 0; i < track->num_cb; i++) { + track->cb[i].robj = NULL; + track->cb[i].pitch = 8192; + track->cb[i].cpp = 16; + track->cb[i].offset = 0; + } + track->z_enabled = true; + track->zb.robj = NULL; + track->zb.pitch = 8192; + track->zb.cpp = 4; + track->zb.offset = 0; + track->vtx_size = 0x7F; + track->immd_dwords = 0xFFFFFFFFUL; + track->num_arrays = 11; + track->max_indx = 0x00FFFFFFUL; + for (i = 0; i < track->num_arrays; i++) { + track->arrays[i].robj = NULL; + track->arrays[i].esize = 0x7F; + } + for (i = 0; i < track->num_texture; i++) { + track->textures[i].pitch = 16536; + track->textures[i].width = 16536; + track->textures[i].height = 16536; + track->textures[i].width_11 = 1 << 11; + track->textures[i].height_11 = 1 << 11; + track->textures[i].num_levels = 12; + if (rdev->family <= CHIP_RS200) { + track->textures[i].tex_coord_type = 0; + track->textures[i].txdepth = 0; + } else { + track->textures[i].txdepth = 16; + track->textures[i].tex_coord_type = 1; + } + track->textures[i].cpp = 64; + track->textures[i].robj = NULL; + /* CS IB emission code makes sure texture unit are disabled */ + track->textures[i].enabled = false; + track->textures[i].roundup_w = true; + track->textures[i].roundup_h = true; + if (track->separate_cube) + for (face = 0; face < 5; face++) { + track->textures[i].cube_info[face].robj = NULL; + track->textures[i].cube_info[face].width = 16536; + track->textures[i].cube_info[face].height = 16536; + track->textures[i].cube_info[face].offset = 0; + } + } +} + +int r100_ring_test(struct radeon_device *rdev) +{ + uint32_t scratch; + uint32_t tmp = 0; + unsigned i; + int r; + + r = radeon_scratch_get(rdev, &scratch); + if (r) { + DRM_ERROR("radeon: cp failed to get scratch reg (%d).\n", r); + return r; + } + WREG32(scratch, 0xCAFEDEAD); + r = radeon_ring_lock(rdev, 2); + if (r) { + DRM_ERROR("radeon: cp failed to lock ring (%d).\n", r); + radeon_scratch_free(rdev, scratch); + return r; + } + radeon_ring_write(rdev, PACKET0(scratch, 0)); + radeon_ring_write(rdev, 0xDEADBEEF); + radeon_ring_unlock_commit(rdev); + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(scratch); + if (tmp == 0xDEADBEEF) { + break; + } + DRM_UDELAY(1); + } + if (i < rdev->usec_timeout) { + DRM_INFO("ring test succeeded in %d usecs\n", i); + } else { + DRM_ERROR("radeon: ring test failed (sracth(0x%04X)=0x%08X)\n", + scratch, tmp); + r = -EINVAL; + } + radeon_scratch_free(rdev, scratch); + return r; +} + +void r100_ring_ib_execute(struct radeon_device *rdev, struct radeon_ib *ib) +{ + radeon_ring_write(rdev, PACKET0(RADEON_CP_IB_BASE, 1)); + radeon_ring_write(rdev, ib->gpu_addr); + radeon_ring_write(rdev, ib->length_dw); +} + +int r100_ib_test(struct radeon_device *rdev) +{ + struct radeon_ib *ib; + uint32_t scratch; + uint32_t tmp = 0; + unsigned i; + int r; + + r = radeon_scratch_get(rdev, &scratch); + if (r) { + DRM_ERROR("radeon: failed to get scratch reg (%d).\n", r); + return r; + } + WREG32(scratch, 0xCAFEDEAD); + r = radeon_ib_get(rdev, &ib); + if (r) { + return r; + } + ib->ptr[0] = PACKET0(scratch, 0); + ib->ptr[1] = 0xDEADBEEF; + ib->ptr[2] = PACKET2(0); + ib->ptr[3] = PACKET2(0); + ib->ptr[4] = PACKET2(0); + ib->ptr[5] = PACKET2(0); + ib->ptr[6] = PACKET2(0); + ib->ptr[7] = PACKET2(0); + ib->length_dw = 8; + r = radeon_ib_schedule(rdev, ib); + if (r) { + radeon_scratch_free(rdev, scratch); + radeon_ib_free(rdev, &ib); + return r; + } + r = radeon_fence_wait(ib->fence, false); + if (r) { + return r; + } + for (i = 0; i < rdev->usec_timeout; i++) { + tmp = RREG32(scratch); + if (tmp == 0xDEADBEEF) { + break; + } + DRM_UDELAY(1); + } + if (i < rdev->usec_timeout) { + DRM_INFO("ib test succeeded in %u usecs\n", i); + } else { + DRM_ERROR("radeon: ib test failed (sracth(0x%04X)=0x%08X)\n", + scratch, tmp); + r = -EINVAL; + } + radeon_scratch_free(rdev, scratch); + radeon_ib_free(rdev, &ib); + return r; +} + +void r100_ib_fini(struct radeon_device *rdev) +{ + radeon_ib_pool_fini(rdev); +} + +int r100_ib_init(struct radeon_device *rdev) +{ + int r; + + r = radeon_ib_pool_init(rdev); + if (r) { + dev_err(rdev->dev, "failled initializing IB pool (%d).\n", r); + r100_ib_fini(rdev); + return r; + } + r = r100_ib_test(rdev); + if (r) { + dev_err(rdev->dev, "failled testing IB (%d).\n", r); + r100_ib_fini(rdev); + return r; + } + return 0; +} + +void r100_mc_stop(struct radeon_device *rdev, struct r100_mc_save *save) +{ + /* Shutdown CP we shouldn't need to do that but better be safe than + * sorry + */ + rdev->cp.ready = false; + WREG32(R_000740_CP_CSQ_CNTL, 0); + + /* Save few CRTC registers */ + save->GENMO_WT = RREG8(R_0003C2_GENMO_WT); + save->CRTC_EXT_CNTL = RREG32(R_000054_CRTC_EXT_CNTL); + save->CRTC_GEN_CNTL = RREG32(R_000050_CRTC_GEN_CNTL); + save->CUR_OFFSET = RREG32(R_000260_CUR_OFFSET); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + save->CRTC2_GEN_CNTL = RREG32(R_0003F8_CRTC2_GEN_CNTL); + save->CUR2_OFFSET = RREG32(R_000360_CUR2_OFFSET); + } + + /* Disable VGA aperture access */ + WREG8(R_0003C2_GENMO_WT, C_0003C2_VGA_RAM_EN & save->GENMO_WT); + /* Disable cursor, overlay, crtc */ + WREG32(R_000260_CUR_OFFSET, save->CUR_OFFSET | S_000260_CUR_LOCK(1)); + WREG32(R_000054_CRTC_EXT_CNTL, save->CRTC_EXT_CNTL | + S_000054_CRTC_DISPLAY_DIS(1)); + WREG32(R_000050_CRTC_GEN_CNTL, + (C_000050_CRTC_CUR_EN & save->CRTC_GEN_CNTL) | + S_000050_CRTC_DISP_REQ_EN_B(1)); + WREG32(R_000420_OV0_SCALE_CNTL, + C_000420_OV0_OVERLAY_EN & RREG32(R_000420_OV0_SCALE_CNTL)); + WREG32(R_000260_CUR_OFFSET, C_000260_CUR_LOCK & save->CUR_OFFSET); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + WREG32(R_000360_CUR2_OFFSET, save->CUR2_OFFSET | + S_000360_CUR2_LOCK(1)); + WREG32(R_0003F8_CRTC2_GEN_CNTL, + (C_0003F8_CRTC2_CUR_EN & save->CRTC2_GEN_CNTL) | + S_0003F8_CRTC2_DISPLAY_DIS(1) | + S_0003F8_CRTC2_DISP_REQ_EN_B(1)); + WREG32(R_000360_CUR2_OFFSET, + C_000360_CUR2_LOCK & save->CUR2_OFFSET); + } +} + +void r100_mc_resume(struct radeon_device *rdev, struct r100_mc_save *save) +{ + /* Update base address for crtc */ + WREG32(R_00023C_DISPLAY_BASE_ADDR, rdev->mc.vram_location); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + WREG32(R_00033C_CRTC2_DISPLAY_BASE_ADDR, + rdev->mc.vram_location); + } + /* Restore CRTC registers */ + WREG8(R_0003C2_GENMO_WT, save->GENMO_WT); + WREG32(R_000054_CRTC_EXT_CNTL, save->CRTC_EXT_CNTL); + WREG32(R_000050_CRTC_GEN_CNTL, save->CRTC_GEN_CNTL); + if (!(rdev->flags & RADEON_SINGLE_CRTC)) { + WREG32(R_0003F8_CRTC2_GEN_CNTL, save->CRTC2_GEN_CNTL); + } +} + +void r100_vga_render_disable(struct radeon_device *rdev) +{ + u32 tmp; + + tmp = RREG8(R_0003C2_GENMO_WT); + WREG8(R_0003C2_GENMO_WT, C_0003C2_VGA_RAM_EN & tmp); +} + +static void r100_debugfs(struct radeon_device *rdev) +{ + int r; + + r = r100_debugfs_mc_info_init(rdev); + if (r) + dev_warn(rdev->dev, "Failed to create r100_mc debugfs file.\n"); +} + +static void r100_mc_program(struct radeon_device *rdev) +{ + struct r100_mc_save save; + + /* Stops all mc clients */ + r100_mc_stop(rdev, &save); + if (rdev->flags & RADEON_IS_AGP) { + WREG32(R_00014C_MC_AGP_LOCATION, + S_00014C_MC_AGP_START(rdev->mc.gtt_start >> 16) | + S_00014C_MC_AGP_TOP(rdev->mc.gtt_end >> 16)); + WREG32(R_000170_AGP_BASE, lower_32_bits(rdev->mc.agp_base)); + if (rdev->family > CHIP_RV200) + WREG32(R_00015C_AGP_BASE_2, + upper_32_bits(rdev->mc.agp_base) & 0xff); + } else { + WREG32(R_00014C_MC_AGP_LOCATION, 0x0FFFFFFF); + WREG32(R_000170_AGP_BASE, 0); + if (rdev->family > CHIP_RV200) + WREG32(R_00015C_AGP_BASE_2, 0); + } + /* Wait for mc idle */ + if (r100_mc_wait_for_idle(rdev)) + dev_warn(rdev->dev, "Wait for MC idle timeout.\n"); + /* Program MC, should be a 32bits limited address space */ + WREG32(R_000148_MC_FB_LOCATION, + S_000148_MC_FB_START(rdev->mc.vram_start >> 16) | + S_000148_MC_FB_TOP(rdev->mc.vram_end >> 16)); + r100_mc_resume(rdev, &save); +} + +void r100_clock_startup(struct radeon_device *rdev) +{ + u32 tmp; + + if (radeon_dynclks != -1 && radeon_dynclks) + radeon_legacy_set_clock_gating(rdev, 1); + /* We need to force on some of the block */ + tmp = RREG32_PLL(R_00000D_SCLK_CNTL); + tmp |= S_00000D_FORCE_CP(1) | S_00000D_FORCE_VIP(1); + if ((rdev->family == CHIP_RV250) || (rdev->family == CHIP_RV280)) + tmp |= S_00000D_FORCE_DISP1(1) | S_00000D_FORCE_DISP2(1); + WREG32_PLL(R_00000D_SCLK_CNTL, tmp); +} + +static int r100_startup(struct radeon_device *rdev) +{ + int r; + + r100_mc_program(rdev); + /* Resume clock */ + r100_clock_startup(rdev); + /* Initialize GPU configuration (# pipes, ...) */ + r100_gpu_init(rdev); + /* Initialize GART (initialize after TTM so we can allocate + * memory through TTM but finalize after TTM) */ + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_enable(rdev); + if (r) + return r; + } + /* Enable IRQ */ + rdev->irq.sw_int = true; + r100_irq_set(rdev); + /* 1M ring buffer */ + r = r100_cp_init(rdev, 1024 * 1024); + if (r) { + dev_err(rdev->dev, "failled initializing CP (%d).\n", r); + return r; + } + r = r100_wb_init(rdev); + if (r) + dev_err(rdev->dev, "failled initializing WB (%d).\n", r); + r = r100_ib_init(rdev); + if (r) { + dev_err(rdev->dev, "failled initializing IB (%d).\n", r); + return r; + } + return 0; +} + +int r100_resume(struct radeon_device *rdev) +{ + /* Make sur GART are not working */ + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + /* Resume clock before doing reset */ + r100_clock_startup(rdev); + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* post */ + radeon_combios_asic_init(rdev->ddev); + /* Resume clock after posting */ + r100_clock_startup(rdev); + return r100_startup(rdev); +} + +int r100_suspend(struct radeon_device *rdev) +{ + r100_cp_disable(rdev); + r100_wb_disable(rdev); + r100_irq_disable(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_disable(rdev); + return 0; +} + +void r100_fini(struct radeon_device *rdev) +{ + r100_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + radeon_gem_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + radeon_fence_driver_fini(rdev); + radeon_object_fini(rdev); + radeon_atombios_fini(rdev); + kfree(rdev->bios); + rdev->bios = NULL; +} + +int r100_mc_init(struct radeon_device *rdev) +{ + int r; + u32 tmp; + + /* Setup GPU memory space */ + rdev->mc.vram_location = 0xFFFFFFFFUL; + rdev->mc.gtt_location = 0xFFFFFFFFUL; + if (rdev->flags & RADEON_IS_IGP) { + tmp = G_00015C_MC_FB_START(RREG32(R_00015C_NB_TOM)); + rdev->mc.vram_location = tmp << 16; + } + if (rdev->flags & RADEON_IS_AGP) { + r = radeon_agp_init(rdev); + if (r) { + printk(KERN_WARNING "[drm] Disabling AGP\n"); + rdev->flags &= ~RADEON_IS_AGP; + rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; + } else { + rdev->mc.gtt_location = rdev->mc.agp_base; + } + } + r = radeon_mc_setup(rdev); + if (r) + return r; + return 0; +} + +int r100_init(struct radeon_device *rdev) +{ + int r; + + /* Register debugfs file specific to this group of asics */ + r100_debugfs(rdev); + /* Disable VGA */ + r100_vga_render_disable(rdev); + /* Initialize scratch registers */ + radeon_scratch_init(rdev); + /* Initialize surface registers */ + radeon_surface_init(rdev); + /* TODO: disable VGA need to use VGA request */ + /* BIOS*/ + if (!radeon_get_bios(rdev)) { + if (ASIC_IS_AVIVO(rdev)) + return -EINVAL; + } + if (rdev->is_atom_bios) { + dev_err(rdev->dev, "Expecting combios for RS400/RS480 GPU\n"); + return -EINVAL; + } else { + r = radeon_combios_init(rdev); + if (r) + return r; + } + /* Reset gpu before posting otherwise ATOM will enter infinite loop */ + if (radeon_gpu_reset(rdev)) { + dev_warn(rdev->dev, + "GPU reset failed ! (0xE40=0x%08X, 0x7C0=0x%08X)\n", + RREG32(R_000E40_RBBM_STATUS), + RREG32(R_0007C0_CP_STAT)); + } + /* check if cards are posted or not */ + if (!radeon_card_posted(rdev) && rdev->bios) { + DRM_INFO("GPU not posted. posting now...\n"); + radeon_combios_asic_init(rdev->ddev); + } + /* Set asic errata */ + r100_errata(rdev); + /* Initialize clocks */ + radeon_get_clock_info(rdev->ddev); + /* Get vram informations */ + r100_vram_info(rdev); + /* Initialize memory controller (also test AGP) */ + r = r100_mc_init(rdev); + if (r) + return r; + /* Fence driver */ + r = radeon_fence_driver_init(rdev); + if (r) + return r; + r = radeon_irq_kms_init(rdev); + if (r) + return r; + /* Memory manager */ + r = radeon_object_init(rdev); + if (r) + return r; + if (rdev->flags & RADEON_IS_PCI) { + r = r100_pci_gart_init(rdev); + if (r) + return r; + } + r100_set_safe_registers(rdev); + rdev->accel_working = true; + r = r100_startup(rdev); + if (r) { + /* Somethings want wront with the accel init stop accel */ + dev_err(rdev->dev, "Disabling GPU acceleration\n"); + r100_suspend(rdev); + r100_cp_fini(rdev); + r100_wb_fini(rdev); + r100_ib_fini(rdev); + if (rdev->flags & RADEON_IS_PCI) + r100_pci_gart_fini(rdev); + radeon_irq_kms_fini(rdev); + rdev->accel_working = false; + } + return 0; +} |