From 1741dd4aa7552055659bf74ab829a21579407f10 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Wed, 4 Aug 2010 15:18:12 +0100 Subject: drm/i915: Unroll wrapping of the ringbuffer. The tail is quadword aligned, so we can add two MI_NOOP as a time. Signed-off-by: Chris Wilson Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/intel_ringbuffer.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_ringbuffer.c') diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index 26362f8495a..df8302a11eb 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -682,9 +682,11 @@ int intel_wrap_ring_buffer(struct drm_device *dev, } virt = (unsigned int *)(ring->virtual_start + ring->tail); - rem /= 4; - while (rem--) + rem /= 8; + while (rem--) { *virt++ = MI_NOOP; + *virt++ = MI_NOOP; + } ring->tail = 0; ring->space = ring->head - 8; -- cgit v1.2.3-70-g09d2 From d97ed3396399126cfca1e12e2b2e2d8bbc4924e5 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Wed, 4 Aug 2010 15:18:13 +0100 Subject: drm/i915: Move ringbuffer accounting to begin/advance. As we check that the ringbuffer will not wrap upon emission, we do not need to check that incrementing the tail wrapped every time. However, we do upon advancing just in case the tail is now pointing at the very end of the ring. Likewise we can account for the space used during emission in begin() and avoid decrementing it for every emit. Signed-off-by: Chris Wilson Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/intel_ringbuffer.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_ringbuffer.c') diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index df8302a11eb..7ab72af045e 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -731,6 +731,8 @@ void intel_ring_begin(struct drm_device *dev, intel_wrap_ring_buffer(dev, ring); if (unlikely(ring->space < n)) intel_wait_ring_buffer(dev, ring, n); + + ring->space -= n; } void intel_ring_emit(struct drm_device *dev, @@ -739,13 +741,12 @@ void intel_ring_emit(struct drm_device *dev, unsigned int *virt = ring->virtual_start + ring->tail; *virt = data; ring->tail += 4; - ring->tail &= ring->size - 1; - ring->space -= 4; } void intel_ring_advance(struct drm_device *dev, struct intel_ring_buffer *ring) { + ring->tail &= ring->size - 1; ring->advance_ring(dev, ring); } -- cgit v1.2.3-70-g09d2 From e898cd221db65273bfc102fa20e4e228e0b8c7e1 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Wed, 4 Aug 2010 15:18:14 +0100 Subject: drm/i915: Inline ringbuffer_emit() As the function has been reduced to a store plus increment, the body is now smaller than the call so inline it. Signed-off-by: Chris Wilson Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/intel_ringbuffer.c | 8 -------- drivers/gpu/drm/i915/intel_ringbuffer.h | 12 ++++++++++-- 2 files changed, 10 insertions(+), 10 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_ringbuffer.c') diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index 7ab72af045e..a5d664e0b17 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -735,14 +735,6 @@ void intel_ring_begin(struct drm_device *dev, ring->space -= n; } -void intel_ring_emit(struct drm_device *dev, - struct intel_ring_buffer *ring, unsigned int data) -{ - unsigned int *virt = ring->virtual_start + ring->tail; - *virt = data; - ring->tail += 4; -} - void intel_ring_advance(struct drm_device *dev, struct intel_ring_buffer *ring) { diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h index d5568d3766d..9b67eead187 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.h +++ b/drivers/gpu/drm/i915/intel_ringbuffer.h @@ -106,8 +106,16 @@ int intel_wrap_ring_buffer(struct drm_device *dev, struct intel_ring_buffer *ring); void intel_ring_begin(struct drm_device *dev, struct intel_ring_buffer *ring, int n); -void intel_ring_emit(struct drm_device *dev, - struct intel_ring_buffer *ring, u32 data); + +static inline void intel_ring_emit(struct drm_device *dev, + struct intel_ring_buffer *ring, + unsigned int data) +{ + unsigned int *virt = ring->virtual_start + ring->tail; + *virt = data; + ring->tail += 4; +} + void intel_fill_struct(struct drm_device *dev, struct intel_ring_buffer *ring, void *data, -- cgit v1.2.3-70-g09d2 From 6f392d548658a17600da7faaf8a5df25ee5f01f6 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Sat, 7 Aug 2010 11:01:22 +0100 Subject: drm/i915: Use a common seqno for all rings. This will be used by the eviction logic to maintain fairness between the rings. Signed-off-by: Chris Wilson Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/i915_drv.h | 3 +-- drivers/gpu/drm/i915/i915_gem.c | 2 ++ drivers/gpu/drm/i915/intel_ringbuffer.c | 46 +++++++++++++++++++-------------- drivers/gpu/drm/i915/intel_ringbuffer.h | 1 - 4 files changed, 29 insertions(+), 23 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_ringbuffer.c') diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index 151056501a5..def6ee0a352 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -244,6 +244,7 @@ typedef struct drm_i915_private { struct pci_dev *bridge_dev; struct intel_ring_buffer render_ring; struct intel_ring_buffer bsd_ring; + uint32_t next_seqno; drm_dma_handle_t *status_page_dmah; void *seqno_page; @@ -573,8 +574,6 @@ typedef struct drm_i915_private { */ struct delayed_work retire_work; - uint32_t next_gem_seqno; - /** * Waiting sequence number, if any */ diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index f150bfd2c85..45b998218d0 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -4714,6 +4714,8 @@ i915_gem_init_ringbuffer(struct drm_device *dev) goto cleanup_render_ring; } + dev_priv->next_seqno = 1; + return 0; cleanup_render_ring: diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index a5d664e0b17..3a024255722 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -33,18 +33,35 @@ #include "i915_drm.h" #include "i915_trace.h" +static u32 i915_gem_get_seqno(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + u32 seqno; + + seqno = dev_priv->next_seqno; + + /* reserve 0 for non-seqno */ + if (++dev_priv->next_seqno == 0) + dev_priv->next_seqno = 1; + + return seqno; +} + static void render_ring_flush(struct drm_device *dev, struct intel_ring_buffer *ring, u32 invalidate_domains, u32 flush_domains) { + drm_i915_private_t *dev_priv = dev->dev_private; + u32 cmd; + #if WATCH_EXEC DRM_INFO("%s: invalidate %08x flush %08x\n", __func__, invalidate_domains, flush_domains); #endif - u32 cmd; - trace_i915_gem_request_flush(dev, ring->next_seqno, + + trace_i915_gem_request_flush(dev, dev_priv->next_seqno, invalidate_domains, flush_domains); if ((invalidate_domains | flush_domains) & I915_GEM_GPU_DOMAINS) { @@ -233,9 +250,10 @@ render_ring_add_request(struct drm_device *dev, struct drm_file *file_priv, u32 flush_domains) { - u32 seqno; drm_i915_private_t *dev_priv = dev->dev_private; - seqno = intel_ring_get_seqno(dev, ring); + u32 seqno; + + seqno = i915_gem_get_seqno(dev); if (IS_GEN6(dev)) { BEGIN_LP_RING(6); @@ -405,7 +423,9 @@ bsd_ring_add_request(struct drm_device *dev, u32 flush_domains) { u32 seqno; - seqno = intel_ring_get_seqno(dev, ring); + + seqno = i915_gem_get_seqno(dev); + intel_ring_begin(dev, ring, 4); intel_ring_emit(dev, ring, MI_STORE_DWORD_INDEX); intel_ring_emit(dev, ring, @@ -479,7 +499,7 @@ render_ring_dispatch_gem_execbuffer(struct drm_device *dev, exec_start = (uint32_t) exec_offset + exec->batch_start_offset; exec_len = (uint32_t) exec->batch_len; - trace_i915_gem_request_submit(dev, dev_priv->mm.next_gem_seqno + 1); + trace_i915_gem_request_submit(dev, dev_priv->next_seqno + 1); count = nbox ? nbox : 1; @@ -757,18 +777,6 @@ void intel_fill_struct(struct drm_device *dev, intel_ring_advance(dev, ring); } -u32 intel_ring_get_seqno(struct drm_device *dev, - struct intel_ring_buffer *ring) -{ - u32 seqno; - seqno = ring->next_seqno; - - /* reserve 0 for non-seqno */ - if (++ring->next_seqno == 0) - ring->next_seqno = 1; - return seqno; -} - struct intel_ring_buffer render_ring = { .name = "render ring", .regs = { @@ -786,7 +794,6 @@ struct intel_ring_buffer render_ring = { .head = 0, .tail = 0, .space = 0, - .next_seqno = 1, .user_irq_refcount = 0, .irq_gem_seqno = 0, .waiting_gem_seqno = 0, @@ -825,7 +832,6 @@ struct intel_ring_buffer bsd_ring = { .head = 0, .tail = 0, .space = 0, - .next_seqno = 1, .user_irq_refcount = 0, .irq_gem_seqno = 0, .waiting_gem_seqno = 0, diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h index 9b67eead187..525e7d3edda 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.h +++ b/drivers/gpu/drm/i915/intel_ringbuffer.h @@ -26,7 +26,6 @@ struct intel_ring_buffer { unsigned int head; unsigned int tail; unsigned int space; - u32 next_seqno; struct intel_hw_status_page status_page; u32 irq_gem_seqno; /* last seq seem at irq time */ -- cgit v1.2.3-70-g09d2 From dd785e35cb3c430c2290d351e67715864f7e5db5 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Sat, 7 Aug 2010 11:01:34 +0100 Subject: drm/i915/ringbuffer: Set ring->gem_buffer = NULL on init unwind The cleanup path for early abort failed to nullify the gem_buffer. The likely consequence of this is zero, since a failure here should mean aborting the module load. Signed-off-by: Chris Wilson Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/intel_ringbuffer.c | 31 +++++++++++++++++-------------- 1 file changed, 17 insertions(+), 14 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_ringbuffer.c') diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index 3a024255722..7823b964817 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -608,9 +608,10 @@ err: int intel_init_ring_buffer(struct drm_device *dev, struct intel_ring_buffer *ring) { - int ret; struct drm_i915_gem_object *obj_priv; struct drm_gem_object *obj; + int ret; + ring->dev = dev; if (I915_NEED_GFX_HWS(dev)) { @@ -623,16 +624,14 @@ int intel_init_ring_buffer(struct drm_device *dev, if (obj == NULL) { DRM_ERROR("Failed to allocate ringbuffer\n"); ret = -ENOMEM; - goto cleanup; + goto err_hws; } ring->gem_object = obj; ret = i915_gem_object_pin(obj, ring->alignment); - if (ret != 0) { - drm_gem_object_unreference(obj); - goto cleanup; - } + if (ret) + goto err_unref; obj_priv = to_intel_bo(obj); ring->map.size = ring->size; @@ -644,18 +643,14 @@ int intel_init_ring_buffer(struct drm_device *dev, drm_core_ioremap_wc(&ring->map, dev); if (ring->map.handle == NULL) { DRM_ERROR("Failed to map ringbuffer.\n"); - i915_gem_object_unpin(obj); - drm_gem_object_unreference(obj); ret = -EINVAL; - goto cleanup; + goto err_unpin; } ring->virtual_start = ring->map.handle; ret = ring->init(dev, ring); - if (ret != 0) { - intel_cleanup_ring_buffer(dev, ring); - return ret; - } + if (ret) + goto err_unmap; if (!drm_core_check_feature(dev, DRIVER_MODESET)) i915_kernel_lost_context(dev); @@ -669,7 +664,15 @@ int intel_init_ring_buffer(struct drm_device *dev, INIT_LIST_HEAD(&ring->active_list); INIT_LIST_HEAD(&ring->request_list); return ret; -cleanup: + +err_unmap: + drm_core_ioremapfree(&ring->map, dev); +err_unpin: + i915_gem_object_unpin(obj); +err_unref: + drm_gem_object_unreference(obj); + ring->gem_object = NULL; +err_hws: cleanup_status_page(dev, ring); return ret; } -- cgit v1.2.3-70-g09d2 From 1cafd34731cd14e5a72edaf0f41717c8126cfce9 Mon Sep 17 00:00:00 2001 From: Zou Nan hai Date: Fri, 25 Jun 2010 13:40:24 +0800 Subject: drm/i915 invalidate indirect state pointers at end of ring exec This is required by the spec, and without this some 3D programs will hang after resume from RC6 we enable that. Signed-off-by: Zou Nan hai Signed-off-by: Eric Anholt --- drivers/gpu/drm/i915/i915_dma.c | 7 +++++++ drivers/gpu/drm/i915/i915_reg.h | 1 + drivers/gpu/drm/i915/intel_ringbuffer.c | 9 +++++++++ 3 files changed, 17 insertions(+) (limited to 'drivers/gpu/drm/i915/intel_ringbuffer.c') diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c index f19ffe87af3..44af317731b 100644 --- a/drivers/gpu/drm/i915/i915_dma.c +++ b/drivers/gpu/drm/i915/i915_dma.c @@ -499,6 +499,13 @@ static int i915_dispatch_batchbuffer(struct drm_device * dev, } } + + if (IS_G4X(dev) || IS_IRONLAKE(dev)) { + BEGIN_LP_RING(2); + OUT_RING(MI_FLUSH | MI_NO_WRITE_FLUSH | MI_INVALIDATE_ISP); + OUT_RING(MI_NOOP); + ADVANCE_LP_RING(); + } i915_emit_breadcrumb(dev); return 0; diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h index 97a35a42da2..21fd657663a 100644 --- a/drivers/gpu/drm/i915/i915_reg.h +++ b/drivers/gpu/drm/i915/i915_reg.h @@ -170,6 +170,7 @@ #define MI_NO_WRITE_FLUSH (1 << 2) #define MI_SCENE_COUNT (1 << 3) /* just increment scene count */ #define MI_END_SCENE (1 << 4) /* flush binner and incr scene count */ +#define MI_INVALIDATE_ISP (1 << 5) /* invalidate indirect state pointers */ #define MI_BATCH_BUFFER_END MI_INSTR(0x0a, 0) #define MI_REPORT_HEAD MI_INSTR(0x07, 0) #define MI_OVERLAY_FLIP MI_INSTR(0x11,0) diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index 7823b964817..51e9c9e718c 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -535,7 +535,16 @@ render_ring_dispatch_gem_execbuffer(struct drm_device *dev, intel_ring_advance(dev, ring); } + if (IS_G4X(dev) || IS_IRONLAKE(dev)) { + intel_ring_begin(dev, ring, 2); + intel_ring_emit(dev, ring, MI_FLUSH | + MI_NO_WRITE_FLUSH | + MI_INVALIDATE_ISP ); + intel_ring_emit(dev, ring, MI_NOOP); + intel_ring_advance(dev, ring); + } /* XXX breadcrumb */ + return 0; } -- cgit v1.2.3-70-g09d2 From a69ffdbfcba8eabf2ca9d384b578e6f28b339c61 Mon Sep 17 00:00:00 2001 From: Zhenyu Wang Date: Mon, 30 Aug 2010 16:12:42 +0800 Subject: drm/i915: Enable MI_FLUSH on Sandybridge MI_FLUSH is being deprecated, but still available on Sandybridge. Make sure it's enabled as userspace still uses MI_FLUSH. Signed-off-by: Zhenyu Wang Cc: stable@kernel.org Signed-off-by: Chris Wilson --- drivers/gpu/drm/i915/i915_reg.h | 1 + drivers/gpu/drm/i915/intel_ringbuffer.c | 8 ++++++-- 2 files changed, 7 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/i915/intel_ringbuffer.c') diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h index 67e3ec1a6af..d094e912922 100644 --- a/drivers/gpu/drm/i915/i915_reg.h +++ b/drivers/gpu/drm/i915/i915_reg.h @@ -319,6 +319,7 @@ #define MI_MODE 0x0209c # define VS_TIMER_DISPATCH (1 << 6) +# define MI_FLUSH_ENABLE (1 << 11) #define SCPD0 0x0209c /* 915+ only */ #define IER 0x020a0 diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index 51e9c9e718c..cb3508f78bc 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -220,9 +220,13 @@ static int init_render_ring(struct drm_device *dev, { drm_i915_private_t *dev_priv = dev->dev_private; int ret = init_ring_common(dev, ring); + int mode; + if (IS_I9XX(dev) && !IS_GEN3(dev)) { - I915_WRITE(MI_MODE, - (VS_TIMER_DISPATCH) << 16 | VS_TIMER_DISPATCH); + mode = VS_TIMER_DISPATCH << 16 | VS_TIMER_DISPATCH; + if (IS_GEN6(dev)) + mode |= MI_FLUSH_ENABLE << 16 | MI_FLUSH_ENABLE; + I915_WRITE(MI_MODE, mode); } return ret; } -- cgit v1.2.3-70-g09d2