diff options
Diffstat (limited to 'drivers/gpu/drm/i915/intel_ringbuffer.c')
| -rw-r--r-- | drivers/gpu/drm/i915/intel_ringbuffer.c | 2402 |
1 files changed, 1607 insertions, 795 deletions
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index 1ab842c6032..279488addf3 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -27,49 +27,78 @@ * */ -#include "drmP.h" -#include "drm.h" +#include <drm/drmP.h> #include "i915_drv.h" -#include "i915_drm.h" +#include <drm/i915_drm.h> #include "i915_trace.h" #include "intel_drv.h" -/* - * 965+ support PIPE_CONTROL commands, which provide finer grained control - * over cache flushing. +/* Early gen2 devices have a cacheline of just 32 bytes, using 64 is overkill, + * but keeps the logic simple. Indeed, the whole purpose of this macro is just + * to give some inclination as to some of the magic values used in the various + * workarounds! */ -struct pipe_control { - struct drm_i915_gem_object *obj; - volatile u32 *cpu_page; - u32 gtt_offset; -}; +#define CACHELINE_BYTES 64 -static inline int ring_space(struct intel_ring_buffer *ring) +static inline int __ring_space(int head, int tail, int size) { - int space = (ring->head & HEAD_ADDR) - (ring->tail + 8); + int space = head - (tail + I915_RING_FREE_SPACE); if (space < 0) - space += ring->size; + space += size; return space; } -static u32 i915_gem_get_seqno(struct drm_device *dev) +static inline int ring_space(struct intel_engine_cs *ring) { - drm_i915_private_t *dev_priv = dev->dev_private; - u32 seqno; + struct intel_ringbuffer *ringbuf = ring->buffer; + return __ring_space(ringbuf->head & HEAD_ADDR, ringbuf->tail, ringbuf->size); +} - seqno = dev_priv->next_seqno; +static bool intel_ring_stopped(struct intel_engine_cs *ring) +{ + struct drm_i915_private *dev_priv = ring->dev->dev_private; + return dev_priv->gpu_error.stop_rings & intel_ring_flag(ring); +} - /* reserve 0 for non-seqno */ - if (++dev_priv->next_seqno == 0) - dev_priv->next_seqno = 1; +void __intel_ring_advance(struct intel_engine_cs *ring) +{ + struct intel_ringbuffer *ringbuf = ring->buffer; + ringbuf->tail &= ringbuf->size - 1; + if (intel_ring_stopped(ring)) + return; + ring->write_tail(ring, ringbuf->tail); +} + +static int +gen2_render_ring_flush(struct intel_engine_cs *ring, + u32 invalidate_domains, + u32 flush_domains) +{ + u32 cmd; + int ret; + + cmd = MI_FLUSH; + if (((invalidate_domains|flush_domains) & I915_GEM_DOMAIN_RENDER) == 0) + cmd |= MI_NO_WRITE_FLUSH; + + if (invalidate_domains & I915_GEM_DOMAIN_SAMPLER) + cmd |= MI_READ_FLUSH; + + ret = intel_ring_begin(ring, 2); + if (ret) + return ret; + + intel_ring_emit(ring, cmd); + intel_ring_emit(ring, MI_NOOP); + intel_ring_advance(ring); - return seqno; + return 0; } static int -render_ring_flush(struct intel_ring_buffer *ring, - u32 invalidate_domains, - u32 flush_domains) +gen4_render_ring_flush(struct intel_engine_cs *ring, + u32 invalidate_domains, + u32 flush_domains) { struct drm_device *dev = ring->dev; u32 cmd; @@ -104,17 +133,8 @@ render_ring_flush(struct intel_ring_buffer *ring, */ cmd = MI_FLUSH | MI_NO_WRITE_FLUSH; - if ((invalidate_domains|flush_domains) & - I915_GEM_DOMAIN_RENDER) + if ((invalidate_domains|flush_domains) & I915_GEM_DOMAIN_RENDER) cmd &= ~MI_NO_WRITE_FLUSH; - if (INTEL_INFO(dev)->gen < 4) { - /* - * On the 965, the sampler cache always gets flushed - * and this bit is reserved. - */ - if (invalidate_domains & I915_GEM_DOMAIN_SAMPLER) - cmd |= MI_READ_FLUSH; - } if (invalidate_domains & I915_GEM_DOMAIN_INSTRUCTION) cmd |= MI_EXE_FLUSH; @@ -171,10 +191,9 @@ render_ring_flush(struct intel_ring_buffer *ring, * really our business. That leaves only stall at scoreboard. */ static int -intel_emit_post_sync_nonzero_flush(struct intel_ring_buffer *ring) +intel_emit_post_sync_nonzero_flush(struct intel_engine_cs *ring) { - struct pipe_control *pc = ring->private; - u32 scratch_addr = pc->gtt_offset + 128; + u32 scratch_addr = ring->scratch.gtt_offset + 2 * CACHELINE_BYTES; int ret; @@ -207,77 +226,271 @@ intel_emit_post_sync_nonzero_flush(struct intel_ring_buffer *ring) } static int -gen6_render_ring_flush(struct intel_ring_buffer *ring, +gen6_render_ring_flush(struct intel_engine_cs *ring, u32 invalidate_domains, u32 flush_domains) { u32 flags = 0; - struct pipe_control *pc = ring->private; - u32 scratch_addr = pc->gtt_offset + 128; + u32 scratch_addr = ring->scratch.gtt_offset + 2 * CACHELINE_BYTES; int ret; /* Force SNB workarounds for PIPE_CONTROL flushes */ - intel_emit_post_sync_nonzero_flush(ring); + ret = intel_emit_post_sync_nonzero_flush(ring); + if (ret) + return ret; /* Just flush everything. Experiments have shown that reducing the * number of bits based on the write domains has little performance * impact. */ - flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH; - flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH; - flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE; + if (flush_domains) { + flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH; + flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH; + /* + * Ensure that any following seqno writes only happen + * when the render cache is indeed flushed. + */ + flags |= PIPE_CONTROL_CS_STALL; + } + if (invalidate_domains) { + flags |= PIPE_CONTROL_TLB_INVALIDATE; + flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE; + /* + * TLB invalidate requires a post-sync write. + */ + flags |= PIPE_CONTROL_QW_WRITE | PIPE_CONTROL_CS_STALL; + } - ret = intel_ring_begin(ring, 6); + ret = intel_ring_begin(ring, 4); if (ret) return ret; - intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(5)); + intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(4)); intel_ring_emit(ring, flags); intel_ring_emit(ring, scratch_addr | PIPE_CONTROL_GLOBAL_GTT); - intel_ring_emit(ring, 0); /* lower dword */ - intel_ring_emit(ring, 0); /* uppwer dword */ - intel_ring_emit(ring, MI_NOOP); + intel_ring_emit(ring, 0); + intel_ring_advance(ring); + + return 0; +} + +static int +gen7_render_ring_cs_stall_wa(struct intel_engine_cs *ring) +{ + int ret; + + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + + intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(4)); + intel_ring_emit(ring, PIPE_CONTROL_CS_STALL | + PIPE_CONTROL_STALL_AT_SCOREBOARD); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, 0); + intel_ring_advance(ring); + + return 0; +} + +static int gen7_ring_fbc_flush(struct intel_engine_cs *ring, u32 value) +{ + int ret; + + if (!ring->fbc_dirty) + return 0; + + ret = intel_ring_begin(ring, 6); + if (ret) + return ret; + /* WaFbcNukeOn3DBlt:ivb/hsw */ + intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1)); + intel_ring_emit(ring, MSG_FBC_REND_STATE); + intel_ring_emit(ring, value); + intel_ring_emit(ring, MI_STORE_REGISTER_MEM(1) | MI_SRM_LRM_GLOBAL_GTT); + intel_ring_emit(ring, MSG_FBC_REND_STATE); + intel_ring_emit(ring, ring->scratch.gtt_offset + 256); + intel_ring_advance(ring); + + ring->fbc_dirty = false; + return 0; +} + +static int +gen7_render_ring_flush(struct intel_engine_cs *ring, + u32 invalidate_domains, u32 flush_domains) +{ + u32 flags = 0; + u32 scratch_addr = ring->scratch.gtt_offset + 2 * CACHELINE_BYTES; + int ret; + + /* + * Ensure that any following seqno writes only happen when the render + * cache is indeed flushed. + * + * Workaround: 4th PIPE_CONTROL command (except the ones with only + * read-cache invalidate bits set) must have the CS_STALL bit set. We + * don't try to be clever and just set it unconditionally. + */ + flags |= PIPE_CONTROL_CS_STALL; + + /* Just flush everything. Experiments have shown that reducing the + * number of bits based on the write domains has little performance + * impact. + */ + if (flush_domains) { + flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH; + flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH; + } + if (invalidate_domains) { + flags |= PIPE_CONTROL_TLB_INVALIDATE; + flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE; + /* + * TLB invalidate requires a post-sync write. + */ + flags |= PIPE_CONTROL_QW_WRITE; + flags |= PIPE_CONTROL_GLOBAL_GTT_IVB; + + /* Workaround: we must issue a pipe_control with CS-stall bit + * set before a pipe_control command that has the state cache + * invalidate bit set. */ + gen7_render_ring_cs_stall_wa(ring); + } + + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + + intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(4)); + intel_ring_emit(ring, flags); + intel_ring_emit(ring, scratch_addr); + intel_ring_emit(ring, 0); intel_ring_advance(ring); + if (!invalidate_domains && flush_domains) + return gen7_ring_fbc_flush(ring, FBC_REND_NUKE); + return 0; } -static void ring_write_tail(struct intel_ring_buffer *ring, +static int +gen8_render_ring_flush(struct intel_engine_cs *ring, + u32 invalidate_domains, u32 flush_domains) +{ + u32 flags = 0; + u32 scratch_addr = ring->scratch.gtt_offset + 2 * CACHELINE_BYTES; + int ret; + + flags |= PIPE_CONTROL_CS_STALL; + + if (flush_domains) { + flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH; + flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH; + } + if (invalidate_domains) { + flags |= PIPE_CONTROL_TLB_INVALIDATE; + flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE; + flags |= PIPE_CONTROL_QW_WRITE; + flags |= PIPE_CONTROL_GLOBAL_GTT_IVB; + } + + ret = intel_ring_begin(ring, 6); + if (ret) + return ret; + + intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(6)); + intel_ring_emit(ring, flags); + intel_ring_emit(ring, scratch_addr); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, 0); + intel_ring_advance(ring); + + return 0; + +} + +static void ring_write_tail(struct intel_engine_cs *ring, u32 value) { - drm_i915_private_t *dev_priv = ring->dev->dev_private; + struct drm_i915_private *dev_priv = ring->dev->dev_private; I915_WRITE_TAIL(ring, value); } -u32 intel_ring_get_active_head(struct intel_ring_buffer *ring) +u64 intel_ring_get_active_head(struct intel_engine_cs *ring) +{ + struct drm_i915_private *dev_priv = ring->dev->dev_private; + u64 acthd; + + if (INTEL_INFO(ring->dev)->gen >= 8) + acthd = I915_READ64_2x32(RING_ACTHD(ring->mmio_base), + RING_ACTHD_UDW(ring->mmio_base)); + else if (INTEL_INFO(ring->dev)->gen >= 4) + acthd = I915_READ(RING_ACTHD(ring->mmio_base)); + else + acthd = I915_READ(ACTHD); + + return acthd; +} + +static void ring_setup_phys_status_page(struct intel_engine_cs *ring) { - drm_i915_private_t *dev_priv = ring->dev->dev_private; - u32 acthd_reg = INTEL_INFO(ring->dev)->gen >= 4 ? - RING_ACTHD(ring->mmio_base) : ACTHD; + struct drm_i915_private *dev_priv = ring->dev->dev_private; + u32 addr; - return I915_READ(acthd_reg); + addr = dev_priv->status_page_dmah->busaddr; + if (INTEL_INFO(ring->dev)->gen >= 4) + addr |= (dev_priv->status_page_dmah->busaddr >> 28) & 0xf0; + I915_WRITE(HWS_PGA, addr); } -static int init_ring_common(struct intel_ring_buffer *ring) +static bool stop_ring(struct intel_engine_cs *ring) { - drm_i915_private_t *dev_priv = ring->dev->dev_private; - struct drm_i915_gem_object *obj = ring->obj; - u32 head; + struct drm_i915_private *dev_priv = to_i915(ring->dev); + + if (!IS_GEN2(ring->dev)) { + I915_WRITE_MODE(ring, _MASKED_BIT_ENABLE(STOP_RING)); + if (wait_for_atomic((I915_READ_MODE(ring) & MODE_IDLE) != 0, 1000)) { + DRM_ERROR("%s :timed out trying to stop ring\n", ring->name); + return false; + } + } - /* Stop the ring if it's running. */ I915_WRITE_CTL(ring, 0); I915_WRITE_HEAD(ring, 0); ring->write_tail(ring, 0); - /* Initialize the ring. */ - I915_WRITE_START(ring, obj->gtt_offset); - head = I915_READ_HEAD(ring) & HEAD_ADDR; + if (!IS_GEN2(ring->dev)) { + (void)I915_READ_CTL(ring); + I915_WRITE_MODE(ring, _MASKED_BIT_DISABLE(STOP_RING)); + } + + return (I915_READ_HEAD(ring) & HEAD_ADDR) == 0; +} + +static int init_ring_common(struct intel_engine_cs *ring) +{ + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_ringbuffer *ringbuf = ring->buffer; + struct drm_i915_gem_object *obj = ringbuf->obj; + int ret = 0; - /* G45 ring initialization fails to reset head to zero */ - if (head != 0) { + gen6_gt_force_wake_get(dev_priv, FORCEWAKE_ALL); + + if (!stop_ring(ring)) { + /* G45 ring initialization often fails to reset head to zero */ DRM_DEBUG_KMS("%s head not reset to zero " "ctl %08x head %08x tail %08x start %08x\n", ring->name, @@ -286,9 +499,7 @@ static int init_ring_common(struct intel_ring_buffer *ring) I915_READ_TAIL(ring), I915_READ_START(ring)); - I915_WRITE_HEAD(ring, 0); - - if (I915_READ_HEAD(ring) & HEAD_ADDR) { + if (!stop_ring(ring)) { DRM_ERROR("failed to set %s head to zero " "ctl %08x head %08x tail %08x start %08x\n", ring->name, @@ -296,117 +507,128 @@ static int init_ring_common(struct intel_ring_buffer *ring) I915_READ_HEAD(ring), I915_READ_TAIL(ring), I915_READ_START(ring)); + ret = -EIO; + goto out; } } + if (I915_NEED_GFX_HWS(dev)) + intel_ring_setup_status_page(ring); + else + ring_setup_phys_status_page(ring); + + /* Initialize the ring. This must happen _after_ we've cleared the ring + * registers with the above sequence (the readback of the HEAD registers + * also enforces ordering), otherwise the hw might lose the new ring + * register values. */ + I915_WRITE_START(ring, i915_gem_obj_ggtt_offset(obj)); I915_WRITE_CTL(ring, - ((ring->size - PAGE_SIZE) & RING_NR_PAGES) - | RING_REPORT_64K | RING_VALID); + ((ringbuf->size - PAGE_SIZE) & RING_NR_PAGES) + | RING_VALID); /* If the head is still not zero, the ring is dead */ - if ((I915_READ_CTL(ring) & RING_VALID) == 0 || - I915_READ_START(ring) != obj->gtt_offset || - (I915_READ_HEAD(ring) & HEAD_ADDR) != 0) { + if (wait_for((I915_READ_CTL(ring) & RING_VALID) != 0 && + I915_READ_START(ring) == i915_gem_obj_ggtt_offset(obj) && + (I915_READ_HEAD(ring) & HEAD_ADDR) == 0, 50)) { DRM_ERROR("%s initialization failed " - "ctl %08x head %08x tail %08x start %08x\n", - ring->name, - I915_READ_CTL(ring), - I915_READ_HEAD(ring), - I915_READ_TAIL(ring), - I915_READ_START(ring)); - return -EIO; + "ctl %08x (valid? %d) head %08x tail %08x start %08x [expected %08lx]\n", + ring->name, + I915_READ_CTL(ring), I915_READ_CTL(ring) & RING_VALID, + I915_READ_HEAD(ring), I915_READ_TAIL(ring), + I915_READ_START(ring), (unsigned long)i915_gem_obj_ggtt_offset(obj)); + ret = -EIO; + goto out; } if (!drm_core_check_feature(ring->dev, DRIVER_MODESET)) i915_kernel_lost_context(ring->dev); else { - ring->head = I915_READ_HEAD(ring); - ring->tail = I915_READ_TAIL(ring) & TAIL_ADDR; - ring->space = ring_space(ring); + ringbuf->head = I915_READ_HEAD(ring); + ringbuf->tail = I915_READ_TAIL(ring) & TAIL_ADDR; + ringbuf->space = ring_space(ring); + ringbuf->last_retired_head = -1; } - return 0; + memset(&ring->hangcheck, 0, sizeof(ring->hangcheck)); + +out: + gen6_gt_force_wake_put(dev_priv, FORCEWAKE_ALL); + + return ret; } static int -init_pipe_control(struct intel_ring_buffer *ring) +init_pipe_control(struct intel_engine_cs *ring) { - struct pipe_control *pc; - struct drm_i915_gem_object *obj; int ret; - if (ring->private) + if (ring->scratch.obj) return 0; - pc = kmalloc(sizeof(*pc), GFP_KERNEL); - if (!pc) - return -ENOMEM; - - obj = i915_gem_alloc_object(ring->dev, 4096); - if (obj == NULL) { + ring->scratch.obj = i915_gem_alloc_object(ring->dev, 4096); + if (ring->scratch.obj == NULL) { DRM_ERROR("Failed to allocate seqno page\n"); ret = -ENOMEM; goto err; } - i915_gem_object_set_cache_level(obj, I915_CACHE_LLC); + ret = i915_gem_object_set_cache_level(ring->scratch.obj, I915_CACHE_LLC); + if (ret) + goto err_unref; - ret = i915_gem_object_pin(obj, 4096, true); + ret = i915_gem_obj_ggtt_pin(ring->scratch.obj, 4096, 0); if (ret) goto err_unref; - pc->gtt_offset = obj->gtt_offset; - pc->cpu_page = kmap(obj->pages[0]); - if (pc->cpu_page == NULL) + ring->scratch.gtt_offset = i915_gem_obj_ggtt_offset(ring->scratch.obj); + ring->scratch.cpu_page = kmap(sg_page(ring->scratch.obj->pages->sgl)); + if (ring->scratch.cpu_page == NULL) { + ret = -ENOMEM; goto err_unpin; + } - pc->obj = obj; - ring->private = pc; + DRM_DEBUG_DRIVER("%s pipe control offset: 0x%08x\n", + ring->name, ring->scratch.gtt_offset); return 0; err_unpin: - i915_gem_object_unpin(obj); + i915_gem_object_ggtt_unpin(ring->scratch.obj); err_unref: - drm_gem_object_unreference(&obj->base); + drm_gem_object_unreference(&ring->scratch.obj->base); err: - kfree(pc); return ret; } -static void -cleanup_pipe_control(struct intel_ring_buffer *ring) -{ - struct pipe_control *pc = ring->private; - struct drm_i915_gem_object *obj; - - if (!ring->private) - return; - - obj = pc->obj; - kunmap(obj->pages[0]); - i915_gem_object_unpin(obj); - drm_gem_object_unreference(&obj->base); - - kfree(pc); - ring->private = NULL; -} - -static int init_render_ring(struct intel_ring_buffer *ring) +static int init_render_ring(struct intel_engine_cs *ring) { struct drm_device *dev = ring->dev; struct drm_i915_private *dev_priv = dev->dev_private; int ret = init_ring_common(ring); - if (INTEL_INFO(dev)->gen > 3) { - int mode = VS_TIMER_DISPATCH << 16 | VS_TIMER_DISPATCH; - 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)); - } + /* WaTimedSingleVertexDispatch:cl,bw,ctg,elk,ilk,snb */ + if (INTEL_INFO(dev)->gen >= 4 && INTEL_INFO(dev)->gen < 7) + I915_WRITE(MI_MODE, _MASKED_BIT_ENABLE(VS_TIMER_DISPATCH)); + + /* We need to disable the AsyncFlip performance optimisations in order + * to use MI_WAIT_FOR_EVENT within the CS. It should already be + * programmed to '1' on all products. + * + * WaDisableAsyncFlipPerfMode:snb,ivb,hsw,vlv,bdw,chv + */ + if (INTEL_INFO(dev)->gen >= 6) + I915_WRITE(MI_MODE, _MASKED_BIT_ENABLE(ASYNC_FLIP_PERF_DISABLE)); + + /* Required for the hardware to program scanline values for waiting */ + /* WaEnableFlushTlbInvalidationMode:snb */ + if (INTEL_INFO(dev)->gen == 6) + I915_WRITE(GFX_MODE, + _MASKED_BIT_ENABLE(GFX_TLB_INVALIDATE_EXPLICIT)); + + /* WaBCSVCSTlbInvalidationMode:ivb,vlv,hsw */ + if (IS_GEN7(dev)) + I915_WRITE(GFX_MODE_GEN7, + _MASKED_BIT_ENABLE(GFX_TLB_INVALIDATE_EXPLICIT) | + _MASKED_BIT_ENABLE(GFX_REPLAY_MODE)); if (INTEL_INFO(dev)->gen >= 5) { ret = init_pipe_control(ring); @@ -414,33 +636,81 @@ static int init_render_ring(struct intel_ring_buffer *ring) return ret; } - if (INTEL_INFO(dev)->gen >= 6) { - I915_WRITE(INSTPM, - INSTPM_FORCE_ORDERING << 16 | INSTPM_FORCE_ORDERING); + if (IS_GEN6(dev)) { + /* From the Sandybridge PRM, volume 1 part 3, page 24: + * "If this bit is set, STCunit will have LRA as replacement + * policy. [...] This bit must be reset. LRA replacement + * policy is not supported." + */ + I915_WRITE(CACHE_MODE_0, + _MASKED_BIT_DISABLE(CM0_STC_EVICT_DISABLE_LRA_SNB)); } + if (INTEL_INFO(dev)->gen >= 6) + I915_WRITE(INSTPM, _MASKED_BIT_ENABLE(INSTPM_FORCE_ORDERING)); + + if (HAS_L3_DPF(dev)) + I915_WRITE_IMR(ring, ~GT_PARITY_ERROR(dev)); + return ret; } -static void render_ring_cleanup(struct intel_ring_buffer *ring) +static void render_ring_cleanup(struct intel_engine_cs *ring) { - if (!ring->private) + struct drm_device *dev = ring->dev; + + if (ring->scratch.obj == NULL) return; - cleanup_pipe_control(ring); + if (INTEL_INFO(dev)->gen >= 5) { + kunmap(sg_page(ring->scratch.obj->pages->sgl)); + i915_gem_object_ggtt_unpin(ring->scratch.obj); + } + + drm_gem_object_unreference(&ring->scratch.obj->base); + ring->scratch.obj = NULL; } -static void -update_mboxes(struct intel_ring_buffer *ring, - u32 seqno, - u32 mmio_offset) +static int gen6_signal(struct intel_engine_cs *signaller, + unsigned int num_dwords) { - intel_ring_emit(ring, MI_SEMAPHORE_MBOX | - MI_SEMAPHORE_GLOBAL_GTT | - MI_SEMAPHORE_REGISTER | - MI_SEMAPHORE_UPDATE); - intel_ring_emit(ring, seqno); - intel_ring_emit(ring, mmio_offset); + struct drm_device *dev = signaller->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_engine_cs *useless; + int i, ret; + + /* NB: In order to be able to do semaphore MBOX updates for varying + * number of rings, it's easiest if we round up each individual update + * to a multiple of 2 (since ring updates must always be a multiple of + * 2) even though the actual update only requires 3 dwords. + */ +#define MBOX_UPDATE_DWORDS 4 + if (i915_semaphore_is_enabled(dev)) + num_dwords += ((I915_NUM_RINGS-1) * MBOX_UPDATE_DWORDS); + else + return intel_ring_begin(signaller, num_dwords); + + ret = intel_ring_begin(signaller, num_dwords); + if (ret) + return ret; +#undef MBOX_UPDATE_DWORDS + + for_each_ring(useless, dev_priv, i) { + u32 mbox_reg = signaller->semaphore.mbox.signal[i]; + if (mbox_reg != GEN6_NOSYNC) { + intel_ring_emit(signaller, MI_LOAD_REGISTER_IMM(1)); + intel_ring_emit(signaller, mbox_reg); + intel_ring_emit(signaller, signaller->outstanding_lazy_seqno); + intel_ring_emit(signaller, MI_NOOP); + } else { + intel_ring_emit(signaller, MI_NOOP); + intel_ring_emit(signaller, MI_NOOP); + intel_ring_emit(signaller, MI_NOOP); + intel_ring_emit(signaller, MI_NOOP); + } + } + + return 0; } /** @@ -453,33 +723,30 @@ update_mboxes(struct intel_ring_buffer *ring, * This acts like a signal in the canonical semaphore. */ static int -gen6_add_request(struct intel_ring_buffer *ring, - u32 *seqno) +gen6_add_request(struct intel_engine_cs *ring) { - u32 mbox1_reg; - u32 mbox2_reg; int ret; - ret = intel_ring_begin(ring, 10); + ret = ring->semaphore.signal(ring, 4); if (ret) return ret; - mbox1_reg = ring->signal_mbox[0]; - mbox2_reg = ring->signal_mbox[1]; - - *seqno = i915_gem_get_seqno(ring->dev); - - update_mboxes(ring, *seqno, mbox1_reg); - update_mboxes(ring, *seqno, mbox2_reg); intel_ring_emit(ring, MI_STORE_DWORD_INDEX); intel_ring_emit(ring, I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT); - intel_ring_emit(ring, *seqno); + intel_ring_emit(ring, ring->outstanding_lazy_seqno); intel_ring_emit(ring, MI_USER_INTERRUPT); - intel_ring_advance(ring); + __intel_ring_advance(ring); return 0; } +static inline bool i915_gem_has_seqno_wrapped(struct drm_device *dev, + u32 seqno) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + return dev_priv->last_seqno < seqno; +} + /** * intel_ring_sync - sync the waiter to the signaller on seqno * @@ -488,70 +755,45 @@ gen6_add_request(struct intel_ring_buffer *ring, * @seqno - seqno which the waiter will block on */ static int -intel_ring_sync(struct intel_ring_buffer *waiter, - struct intel_ring_buffer *signaller, - int ring, - u32 seqno) +gen6_ring_sync(struct intel_engine_cs *waiter, + struct intel_engine_cs *signaller, + u32 seqno) { - int ret; u32 dw1 = MI_SEMAPHORE_MBOX | MI_SEMAPHORE_COMPARE | MI_SEMAPHORE_REGISTER; + u32 wait_mbox = signaller->semaphore.mbox.wait[waiter->id]; + int ret; + + /* Throughout all of the GEM code, seqno passed implies our current + * seqno is >= the last seqno executed. However for hardware the + * comparison is strictly greater than. + */ + seqno -= 1; + + WARN_ON(wait_mbox == MI_SEMAPHORE_SYNC_INVALID); ret = intel_ring_begin(waiter, 4); if (ret) return ret; - intel_ring_emit(waiter, dw1 | signaller->semaphore_register[ring]); - intel_ring_emit(waiter, seqno); - intel_ring_emit(waiter, 0); - intel_ring_emit(waiter, MI_NOOP); + /* If seqno wrap happened, omit the wait with no-ops */ + if (likely(!i915_gem_has_seqno_wrapped(waiter->dev, seqno))) { + intel_ring_emit(waiter, dw1 | wait_mbox); + intel_ring_emit(waiter, seqno); + intel_ring_emit(waiter, 0); + intel_ring_emit(waiter, MI_NOOP); + } else { + intel_ring_emit(waiter, MI_NOOP); + intel_ring_emit(waiter, MI_NOOP); + intel_ring_emit(waiter, MI_NOOP); + intel_ring_emit(waiter, MI_NOOP); + } intel_ring_advance(waiter); return 0; } -/* VCS->RCS (RVSYNC) or BCS->RCS (RBSYNC) */ -int -render_ring_sync_to(struct intel_ring_buffer *waiter, - struct intel_ring_buffer *signaller, - u32 seqno) -{ - WARN_ON(signaller->semaphore_register[RCS] == MI_SEMAPHORE_SYNC_INVALID); - return intel_ring_sync(waiter, - signaller, - RCS, - seqno); -} - -/* RCS->VCS (VRSYNC) or BCS->VCS (VBSYNC) */ -int -gen6_bsd_ring_sync_to(struct intel_ring_buffer *waiter, - struct intel_ring_buffer *signaller, - u32 seqno) -{ - WARN_ON(signaller->semaphore_register[VCS] == MI_SEMAPHORE_SYNC_INVALID); - return intel_ring_sync(waiter, - signaller, - VCS, - seqno); -} - -/* RCS->BCS (BRSYNC) or VCS->BCS (BVSYNC) */ -int -gen6_blt_ring_sync_to(struct intel_ring_buffer *waiter, - struct intel_ring_buffer *signaller, - u32 seqno) -{ - WARN_ON(signaller->semaphore_register[BCS] == MI_SEMAPHORE_SYNC_INVALID); - return intel_ring_sync(waiter, - signaller, - BCS, - seqno); -} - - - #define PIPE_CONTROL_FLUSH(ring__, addr__) \ do { \ intel_ring_emit(ring__, GFX_OP_PIPE_CONTROL(4) | PIPE_CONTROL_QW_WRITE | \ @@ -562,13 +804,9 @@ do { \ } while (0) static int -pc_render_add_request(struct intel_ring_buffer *ring, - u32 *result) +pc_render_add_request(struct intel_engine_cs *ring) { - struct drm_device *dev = ring->dev; - u32 seqno = i915_gem_get_seqno(dev); - struct pipe_control *pc = ring->private; - u32 scratch_addr = pc->gtt_offset + 128; + u32 scratch_addr = ring->scratch.gtt_offset + 2 * CACHELINE_BYTES; int ret; /* For Ironlake, MI_USER_INTERRUPT was deprecated and apparently @@ -586,157 +824,180 @@ pc_render_add_request(struct intel_ring_buffer *ring, intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(4) | PIPE_CONTROL_QW_WRITE | PIPE_CONTROL_WRITE_FLUSH | PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE); - intel_ring_emit(ring, pc->gtt_offset | PIPE_CONTROL_GLOBAL_GTT); - intel_ring_emit(ring, seqno); + intel_ring_emit(ring, ring->scratch.gtt_offset | PIPE_CONTROL_GLOBAL_GTT); + intel_ring_emit(ring, ring->outstanding_lazy_seqno); intel_ring_emit(ring, 0); PIPE_CONTROL_FLUSH(ring, scratch_addr); - scratch_addr += 128; /* write to separate cachelines */ + scratch_addr += 2 * CACHELINE_BYTES; /* write to separate cachelines */ PIPE_CONTROL_FLUSH(ring, scratch_addr); - scratch_addr += 128; + scratch_addr += 2 * CACHELINE_BYTES; PIPE_CONTROL_FLUSH(ring, scratch_addr); - scratch_addr += 128; + scratch_addr += 2 * CACHELINE_BYTES; PIPE_CONTROL_FLUSH(ring, scratch_addr); - scratch_addr += 128; + scratch_addr += 2 * CACHELINE_BYTES; PIPE_CONTROL_FLUSH(ring, scratch_addr); - scratch_addr += 128; + scratch_addr += 2 * CACHELINE_BYTES; PIPE_CONTROL_FLUSH(ring, scratch_addr); + intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(4) | PIPE_CONTROL_QW_WRITE | PIPE_CONTROL_WRITE_FLUSH | PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE | PIPE_CONTROL_NOTIFY); - intel_ring_emit(ring, pc->gtt_offset | PIPE_CONTROL_GLOBAL_GTT); - intel_ring_emit(ring, seqno); + intel_ring_emit(ring, ring->scratch.gtt_offset | PIPE_CONTROL_GLOBAL_GTT); + intel_ring_emit(ring, ring->outstanding_lazy_seqno); intel_ring_emit(ring, 0); - intel_ring_advance(ring); - - *result = seqno; - return 0; -} - -static int -render_ring_add_request(struct intel_ring_buffer *ring, - u32 *result) -{ - struct drm_device *dev = ring->dev; - u32 seqno = i915_gem_get_seqno(dev); - int ret; - - ret = intel_ring_begin(ring, 4); - if (ret) - return ret; - - intel_ring_emit(ring, MI_STORE_DWORD_INDEX); - intel_ring_emit(ring, I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT); - intel_ring_emit(ring, seqno); - intel_ring_emit(ring, MI_USER_INTERRUPT); - intel_ring_advance(ring); + __intel_ring_advance(ring); - *result = seqno; return 0; } static u32 -gen6_ring_get_seqno(struct intel_ring_buffer *ring) +gen6_ring_get_seqno(struct intel_engine_cs *ring, bool lazy_coherency) { - struct drm_device *dev = ring->dev; - /* Workaround to force correct ordering between irq and seqno writes on * ivb (and maybe also on snb) by reading from a CS register (like * ACTHD) before reading the status page. */ - if (IS_GEN7(dev)) - intel_ring_get_active_head(ring); + if (!lazy_coherency) { + struct drm_i915_private *dev_priv = ring->dev->dev_private; + POSTING_READ(RING_ACTHD(ring->mmio_base)); + } + return intel_read_status_page(ring, I915_GEM_HWS_INDEX); } static u32 -ring_get_seqno(struct intel_ring_buffer *ring) +ring_get_seqno(struct intel_engine_cs *ring, bool lazy_coherency) { return intel_read_status_page(ring, I915_GEM_HWS_INDEX); } +static void +ring_set_seqno(struct intel_engine_cs *ring, u32 seqno) +{ + intel_write_status_page(ring, I915_GEM_HWS_INDEX, seqno); +} + static u32 -pc_render_get_seqno(struct intel_ring_buffer *ring) +pc_render_get_seqno(struct intel_engine_cs *ring, bool lazy_coherency) { - struct pipe_control *pc = ring->private; - return pc->cpu_page[0]; + return ring->scratch.cpu_page[0]; } static void -ironlake_enable_irq(drm_i915_private_t *dev_priv, u32 mask) +pc_render_set_seqno(struct intel_engine_cs *ring, u32 seqno) { - dev_priv->gt_irq_mask &= ~mask; - I915_WRITE(GTIMR, dev_priv->gt_irq_mask); - POSTING_READ(GTIMR); + ring->scratch.cpu_page[0] = seqno; } -static void -ironlake_disable_irq(drm_i915_private_t *dev_priv, u32 mask) +static bool +gen5_ring_get_irq(struct intel_engine_cs *ring) { - dev_priv->gt_irq_mask |= mask; - I915_WRITE(GTIMR, dev_priv->gt_irq_mask); - POSTING_READ(GTIMR); + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; + + if (!dev->irq_enabled) + return false; + + spin_lock_irqsave(&dev_priv->irq_lock, flags); + if (ring->irq_refcount++ == 0) + ilk_enable_gt_irq(dev_priv, ring->irq_enable_mask); + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); + + return true; } static void -i915_enable_irq(drm_i915_private_t *dev_priv, u32 mask) +gen5_ring_put_irq(struct intel_engine_cs *ring) { - dev_priv->irq_mask &= ~mask; - I915_WRITE(IMR, dev_priv->irq_mask); - POSTING_READ(IMR); + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->irq_lock, flags); + if (--ring->irq_refcount == 0) + ilk_disable_gt_irq(dev_priv, ring->irq_enable_mask); + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); +} + +static bool +i9xx_ring_get_irq(struct intel_engine_cs *ring) +{ + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; + + if (!dev->irq_enabled) + return false; + + spin_lock_irqsave(&dev_priv->irq_lock, flags); + if (ring->irq_refcount++ == 0) { + dev_priv->irq_mask &= ~ring->irq_enable_mask; + I915_WRITE(IMR, dev_priv->irq_mask); + POSTING_READ(IMR); + } + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); + + return true; } static void -i915_disable_irq(drm_i915_private_t *dev_priv, u32 mask) +i9xx_ring_put_irq(struct intel_engine_cs *ring) { - dev_priv->irq_mask |= mask; - I915_WRITE(IMR, dev_priv->irq_mask); - POSTING_READ(IMR); + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; + + spin_lock_irqsave(&dev_priv->irq_lock, flags); + if (--ring->irq_refcount == 0) { + dev_priv->irq_mask |= ring->irq_enable_mask; + I915_WRITE(IMR, dev_priv->irq_mask); + POSTING_READ(IMR); + } + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); } static bool -render_ring_get_irq(struct intel_ring_buffer *ring) +i8xx_ring_get_irq(struct intel_engine_cs *ring) { struct drm_device *dev = ring->dev; - drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; if (!dev->irq_enabled) return false; - spin_lock(&ring->irq_lock); + spin_lock_irqsave(&dev_priv->irq_lock, flags); if (ring->irq_refcount++ == 0) { - if (HAS_PCH_SPLIT(dev)) - ironlake_enable_irq(dev_priv, - GT_PIPE_NOTIFY | GT_USER_INTERRUPT); - else - i915_enable_irq(dev_priv, I915_USER_INTERRUPT); + dev_priv->irq_mask &= ~ring->irq_enable_mask; + I915_WRITE16(IMR, dev_priv->irq_mask); + POSTING_READ16(IMR); } - spin_unlock(&ring->irq_lock); + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); return true; } static void -render_ring_put_irq(struct intel_ring_buffer *ring) +i8xx_ring_put_irq(struct intel_engine_cs *ring) { struct drm_device *dev = ring->dev; - drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; - spin_lock(&ring->irq_lock); + spin_lock_irqsave(&dev_priv->irq_lock, flags); if (--ring->irq_refcount == 0) { - if (HAS_PCH_SPLIT(dev)) - ironlake_disable_irq(dev_priv, - GT_USER_INTERRUPT | - GT_PIPE_NOTIFY); - else - i915_disable_irq(dev_priv, I915_USER_INTERRUPT); + dev_priv->irq_mask |= ring->irq_enable_mask; + I915_WRITE16(IMR, dev_priv->irq_mask); + POSTING_READ16(IMR); } - spin_unlock(&ring->irq_lock); + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); } -void intel_ring_setup_status_page(struct intel_ring_buffer *ring) +void intel_ring_setup_status_page(struct intel_engine_cs *ring) { struct drm_device *dev = ring->dev; - drm_i915_private_t *dev_priv = ring->dev->dev_private; + struct drm_i915_private *dev_priv = ring->dev->dev_private; u32 mmio = 0; /* The ring status page addresses are no longer next to the rest of @@ -744,28 +1005,59 @@ void intel_ring_setup_status_page(struct intel_ring_buffer *ring) */ if (IS_GEN7(dev)) { switch (ring->id) { - case RING_RENDER: + case RCS: mmio = RENDER_HWS_PGA_GEN7; break; - case RING_BLT: + case BCS: mmio = BLT_HWS_PGA_GEN7; break; - case RING_BSD: + /* + * VCS2 actually doesn't exist on Gen7. Only shut up + * gcc switch check warning + */ + case VCS2: + case VCS: mmio = BSD_HWS_PGA_GEN7; break; + case VECS: + mmio = VEBOX_HWS_PGA_GEN7; + break; } } else if (IS_GEN6(ring->dev)) { mmio = RING_HWS_PGA_GEN6(ring->mmio_base); } else { + /* XXX: gen8 returns to sanity */ mmio = RING_HWS_PGA(ring->mmio_base); } I915_WRITE(mmio, (u32)ring->status_page.gfx_addr); POSTING_READ(mmio); + + /* + * Flush the TLB for this page + * + * FIXME: These two bits have disappeared on gen8, so a question + * arises: do we still need this and if so how should we go about + * invalidating the TLB? + */ + if (INTEL_INFO(dev)->gen >= 6 && INTEL_INFO(dev)->gen < 8) { + u32 reg = RING_INSTPM(ring->mmio_base); + + /* ring should be idle before issuing a sync flush*/ + WARN_ON((I915_READ_MODE(ring) & MODE_IDLE) == 0); + + I915_WRITE(reg, + _MASKED_BIT_ENABLE(INSTPM_TLB_INVALIDATE | + INSTPM_SYNC_FLUSH)); + if (wait_for((I915_READ(reg) & INSTPM_SYNC_FLUSH) == 0, + 1000)) + DRM_ERROR("%s: wait for SyncFlush to complete for TLB invalidation timed out\n", + ring->name); + } } static int -bsd_ring_flush(struct intel_ring_buffer *ring, +bsd_ring_flush(struct intel_engine_cs *ring, u32 invalidate_domains, u32 flush_domains) { @@ -782,110 +1074,154 @@ bsd_ring_flush(struct intel_ring_buffer *ring, } static int -ring_add_request(struct intel_ring_buffer *ring, - u32 *result) +i9xx_add_request(struct intel_engine_cs *ring) { - u32 seqno; int ret; ret = intel_ring_begin(ring, 4); if (ret) return ret; - seqno = i915_gem_get_seqno(ring->dev); - intel_ring_emit(ring, MI_STORE_DWORD_INDEX); intel_ring_emit(ring, I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT); - intel_ring_emit(ring, seqno); + intel_ring_emit(ring, ring->outstanding_lazy_seqno); intel_ring_emit(ring, MI_USER_INTERRUPT); - intel_ring_advance(ring); + __intel_ring_advance(ring); - *result = seqno; return 0; } static bool -gen6_ring_get_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag) +gen6_ring_get_irq(struct intel_engine_cs *ring) { struct drm_device *dev = ring->dev; - drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; if (!dev->irq_enabled) return false; - /* It looks like we need to prevent the gt from suspending while waiting - * for an notifiy irq, otherwise irqs seem to get lost on at least the - * blt/bsd rings on ivb. */ - if (IS_GEN7(dev)) - gen6_gt_force_wake_get(dev_priv); - - spin_lock(&ring->irq_lock); + spin_lock_irqsave(&dev_priv->irq_lock, flags); if (ring->irq_refcount++ == 0) { - ring->irq_mask &= ~rflag; - I915_WRITE_IMR(ring, ring->irq_mask); - ironlake_enable_irq(dev_priv, gflag); + if (HAS_L3_DPF(dev) && ring->id == RCS) + I915_WRITE_IMR(ring, + ~(ring->irq_enable_mask | + GT_PARITY_ERROR(dev))); + else + I915_WRITE_IMR(ring, ~ring->irq_enable_mask); + ilk_enable_gt_irq(dev_priv, ring->irq_enable_mask); } - spin_unlock(&ring->irq_lock); + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); return true; } static void -gen6_ring_put_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag) +gen6_ring_put_irq(struct intel_engine_cs *ring) { struct drm_device *dev = ring->dev; - drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; - spin_lock(&ring->irq_lock); + spin_lock_irqsave(&dev_priv->irq_lock, flags); if (--ring->irq_refcount == 0) { - ring->irq_mask |= rflag; - I915_WRITE_IMR(ring, ring->irq_mask); - ironlake_disable_irq(dev_priv, gflag); + if (HAS_L3_DPF(dev) && ring->id == RCS) + I915_WRITE_IMR(ring, ~GT_PARITY_ERROR(dev)); + else + I915_WRITE_IMR(ring, ~0); + ilk_disable_gt_irq(dev_priv, ring->irq_enable_mask); } - spin_unlock(&ring->irq_lock); + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); +} - if (IS_GEN7(dev)) - gen6_gt_force_wake_put(dev_priv); +static bool +hsw_vebox_get_irq(struct intel_engine_cs *ring) +{ + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; + + if (!dev->irq_enabled) + return false; + + spin_lock_irqsave(&dev_priv->irq_lock, flags); + if (ring->irq_refcount++ == 0) { + I915_WRITE_IMR(ring, ~ring->irq_enable_mask); + snb_enable_pm_irq(dev_priv, ring->irq_enable_mask); + } + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); + + return true; +} + +static void +hsw_vebox_put_irq(struct intel_engine_cs *ring) +{ + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; + + if (!dev->irq_enabled) + return; + + spin_lock_irqsave(&dev_priv->irq_lock, flags); + if (--ring->irq_refcount == 0) { + I915_WRITE_IMR(ring, ~0); + snb_disable_pm_irq(dev_priv, ring->irq_enable_mask); + } + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); } static bool -bsd_ring_get_irq(struct intel_ring_buffer *ring) +gen8_ring_get_irq(struct intel_engine_cs *ring) { struct drm_device *dev = ring->dev; - drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; if (!dev->irq_enabled) return false; - spin_lock(&ring->irq_lock); + spin_lock_irqsave(&dev_priv->irq_lock, flags); if (ring->irq_refcount++ == 0) { - if (IS_G4X(dev)) - i915_enable_irq(dev_priv, I915_BSD_USER_INTERRUPT); - else - ironlake_enable_irq(dev_priv, GT_BSD_USER_INTERRUPT); + if (HAS_L3_DPF(dev) && ring->id == RCS) { + I915_WRITE_IMR(ring, + ~(ring->irq_enable_mask | + GT_RENDER_L3_PARITY_ERROR_INTERRUPT)); + } else { + I915_WRITE_IMR(ring, ~ring->irq_enable_mask); + } + POSTING_READ(RING_IMR(ring->mmio_base)); } - spin_unlock(&ring->irq_lock); + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); return true; } + static void -bsd_ring_put_irq(struct intel_ring_buffer *ring) +gen8_ring_put_irq(struct intel_engine_cs *ring) { struct drm_device *dev = ring->dev; - drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_private *dev_priv = dev->dev_private; + unsigned long flags; - spin_lock(&ring->irq_lock); + spin_lock_irqsave(&dev_priv->irq_lock, flags); if (--ring->irq_refcount == 0) { - if (IS_G4X(dev)) - i915_disable_irq(dev_priv, I915_BSD_USER_INTERRUPT); - else - ironlake_disable_irq(dev_priv, GT_BSD_USER_INTERRUPT); + if (HAS_L3_DPF(dev) && ring->id == RCS) { + I915_WRITE_IMR(ring, + ~GT_RENDER_L3_PARITY_ERROR_INTERRUPT); + } else { + I915_WRITE_IMR(ring, ~0); + } + POSTING_READ(RING_IMR(ring->mmio_base)); } - spin_unlock(&ring->irq_lock); + spin_unlock_irqrestore(&dev_priv->irq_lock, flags); } static int -ring_dispatch_execbuffer(struct intel_ring_buffer *ring, u32 offset, u32 length) +i965_dispatch_execbuffer(struct intel_engine_cs *ring, + u64 offset, u32 length, + unsigned flags) { int ret; @@ -894,385 +1230,572 @@ ring_dispatch_execbuffer(struct intel_ring_buffer *ring, u32 offset, u32 length) return ret; intel_ring_emit(ring, - MI_BATCH_BUFFER_START | (2 << 6) | - MI_BATCH_NON_SECURE_I965); + MI_BATCH_BUFFER_START | + MI_BATCH_GTT | + (flags & I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE_I965)); intel_ring_emit(ring, offset); intel_ring_advance(ring); return 0; } +/* Just userspace ABI convention to limit the wa batch bo to a resonable size */ +#define I830_BATCH_LIMIT (256*1024) static int -render_ring_dispatch_execbuffer(struct intel_ring_buffer *ring, - u32 offset, u32 len) +i830_dispatch_execbuffer(struct intel_engine_cs *ring, + u64 offset, u32 len, + unsigned flags) { - struct drm_device *dev = ring->dev; int ret; - if (IS_I830(dev) || IS_845G(dev)) { + if (flags & I915_DISPATCH_PINNED) { ret = intel_ring_begin(ring, 4); if (ret) return ret; intel_ring_emit(ring, MI_BATCH_BUFFER); - intel_ring_emit(ring, offset | MI_BATCH_NON_SECURE); + intel_ring_emit(ring, offset | (flags & I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE)); intel_ring_emit(ring, offset + len - 8); - intel_ring_emit(ring, 0); + intel_ring_emit(ring, MI_NOOP); + intel_ring_advance(ring); } else { - ret = intel_ring_begin(ring, 2); + u32 cs_offset = ring->scratch.gtt_offset; + + if (len > I830_BATCH_LIMIT) + return -ENOSPC; + + ret = intel_ring_begin(ring, 9+3); if (ret) return ret; + /* Blit the batch (which has now all relocs applied) to the stable batch + * scratch bo area (so that the CS never stumbles over its tlb + * invalidation bug) ... */ + intel_ring_emit(ring, XY_SRC_COPY_BLT_CMD | + XY_SRC_COPY_BLT_WRITE_ALPHA | + XY_SRC_COPY_BLT_WRITE_RGB); + intel_ring_emit(ring, BLT_DEPTH_32 | BLT_ROP_GXCOPY | 4096); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, (DIV_ROUND_UP(len, 4096) << 16) | 1024); + intel_ring_emit(ring, cs_offset); + intel_ring_emit(ring, 0); + intel_ring_emit(ring, 4096); + intel_ring_emit(ring, offset); + intel_ring_emit(ring, MI_FLUSH); - if (INTEL_INFO(dev)->gen >= 4) { - intel_ring_emit(ring, - MI_BATCH_BUFFER_START | (2 << 6) | - MI_BATCH_NON_SECURE_I965); - intel_ring_emit(ring, offset); - } else { - intel_ring_emit(ring, - MI_BATCH_BUFFER_START | (2 << 6)); - intel_ring_emit(ring, offset | MI_BATCH_NON_SECURE); - } + /* ... and execute it. */ + intel_ring_emit(ring, MI_BATCH_BUFFER); + intel_ring_emit(ring, cs_offset | (flags & I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE)); + intel_ring_emit(ring, cs_offset + len - 8); + intel_ring_advance(ring); } + + return 0; +} + +static int +i915_dispatch_execbuffer(struct intel_engine_cs *ring, + u64 offset, u32 len, + unsigned flags) +{ + int ret; + + ret = intel_ring_begin(ring, 2); + if (ret) + return ret; + + intel_ring_emit(ring, MI_BATCH_BUFFER_START | MI_BATCH_GTT); + intel_ring_emit(ring, offset | (flags & I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE)); intel_ring_advance(ring); return 0; } -static void cleanup_status_page(struct intel_ring_buffer *ring) +static void cleanup_status_page(struct intel_engine_cs *ring) { - drm_i915_private_t *dev_priv = ring->dev->dev_private; struct drm_i915_gem_object *obj; obj = ring->status_page.obj; if (obj == NULL) return; - kunmap(obj->pages[0]); - i915_gem_object_unpin(obj); + kunmap(sg_page(obj->pages->sgl)); + i915_gem_object_ggtt_unpin(obj); drm_gem_object_unreference(&obj->base); ring->status_page.obj = NULL; +} + +static int init_status_page(struct intel_engine_cs *ring) +{ + struct drm_i915_gem_object *obj; + + if ((obj = ring->status_page.obj) == NULL) { + int ret; + + obj = i915_gem_alloc_object(ring->dev, 4096); + if (obj == NULL) { + DRM_ERROR("Failed to allocate status page\n"); + return -ENOMEM; + } + + ret = i915_gem_object_set_cache_level(obj, I915_CACHE_LLC); + if (ret) + goto err_unref; + + ret = i915_gem_obj_ggtt_pin(obj, 4096, 0); + if (ret) { +err_unref: + drm_gem_object_unreference(&obj->base); + return ret; + } + + ring->status_page.obj = obj; + } + + ring->status_page.gfx_addr = i915_gem_obj_ggtt_offset(obj); + ring->status_page.page_addr = kmap(sg_page(obj->pages->sgl)); + memset(ring->status_page.page_addr, 0, PAGE_SIZE); + + DRM_DEBUG_DRIVER("%s hws offset: 0x%08x\n", + ring->name, ring->status_page.gfx_addr); + + return 0; +} - memset(&dev_priv->hws_map, 0, sizeof(dev_priv->hws_map)); +static int init_phys_status_page(struct intel_engine_cs *ring) +{ + struct drm_i915_private *dev_priv = ring->dev->dev_private; + + if (!dev_priv->status_page_dmah) { + dev_priv->status_page_dmah = + drm_pci_alloc(ring->dev, PAGE_SIZE, PAGE_SIZE); + if (!dev_priv->status_page_dmah) + return -ENOMEM; + } + + ring->status_page.page_addr = dev_priv->status_page_dmah->vaddr; + memset(ring->status_page.page_addr, 0, PAGE_SIZE); + + return 0; } -static int init_status_page(struct intel_ring_buffer *ring) +static int allocate_ring_buffer(struct intel_engine_cs *ring) { struct drm_device *dev = ring->dev; - drm_i915_private_t *dev_priv = dev->dev_private; + struct drm_i915_private *dev_priv = to_i915(dev); + struct intel_ringbuffer *ringbuf = ring->buffer; struct drm_i915_gem_object *obj; int ret; - obj = i915_gem_alloc_object(dev, 4096); - if (obj == NULL) { - DRM_ERROR("Failed to allocate status page\n"); - ret = -ENOMEM; - goto err; - } + if (intel_ring_initialized(ring)) + return 0; - i915_gem_object_set_cache_level(obj, I915_CACHE_LLC); + obj = NULL; + if (!HAS_LLC(dev)) + obj = i915_gem_object_create_stolen(dev, ringbuf->size); + if (obj == NULL) + obj = i915_gem_alloc_object(dev, ringbuf->size); + if (obj == NULL) + return -ENOMEM; - ret = i915_gem_object_pin(obj, 4096, true); - if (ret != 0) { + ret = i915_gem_obj_ggtt_pin(obj, PAGE_SIZE, PIN_MAPPABLE); + if (ret) goto err_unref; - } - ring->status_page.gfx_addr = obj->gtt_offset; - ring->status_page.page_addr = kmap(obj->pages[0]); - if (ring->status_page.page_addr == NULL) { - memset(&dev_priv->hws_map, 0, sizeof(dev_priv->hws_map)); + ret = i915_gem_object_set_to_gtt_domain(obj, true); + if (ret) goto err_unpin; - } - ring->status_page.obj = obj; - memset(ring->status_page.page_addr, 0, PAGE_SIZE); - intel_ring_setup_status_page(ring); - DRM_DEBUG_DRIVER("%s hws offset: 0x%08x\n", - ring->name, ring->status_page.gfx_addr); + ringbuf->virtual_start = + ioremap_wc(dev_priv->gtt.mappable_base + i915_gem_obj_ggtt_offset(obj), + ringbuf->size); + if (ringbuf->virtual_start == NULL) { + ret = -EINVAL; + goto err_unpin; + } + ringbuf->obj = obj; return 0; err_unpin: - i915_gem_object_unpin(obj); + i915_gem_object_ggtt_unpin(obj); err_unref: drm_gem_object_unreference(&obj->base); -err: return ret; } -int intel_init_ring_buffer(struct drm_device *dev, - struct intel_ring_buffer *ring) +static int intel_init_ring_buffer(struct drm_device *dev, + struct intel_engine_cs *ring) { - struct drm_i915_gem_object *obj; + struct intel_ringbuffer *ringbuf = ring->buffer; int ret; + if (ringbuf == NULL) { + ringbuf = kzalloc(sizeof(*ringbuf), GFP_KERNEL); + if (!ringbuf) + return -ENOMEM; + ring->buffer = ringbuf; + } + ring->dev = dev; INIT_LIST_HEAD(&ring->active_list); INIT_LIST_HEAD(&ring->request_list); - INIT_LIST_HEAD(&ring->gpu_write_list); + ringbuf->size = 32 * PAGE_SIZE; + memset(ring->semaphore.sync_seqno, 0, sizeof(ring->semaphore.sync_seqno)); init_waitqueue_head(&ring->irq_queue); - spin_lock_init(&ring->irq_lock); - ring->irq_mask = ~0; if (I915_NEED_GFX_HWS(dev)) { ret = init_status_page(ring); if (ret) - return ret; + goto error; + } else { + BUG_ON(ring->id != RCS); + ret = init_phys_status_page(ring); + if (ret) + goto error; } - obj = i915_gem_alloc_object(dev, ring->size); - if (obj == NULL) { - DRM_ERROR("Failed to allocate ringbuffer\n"); - ret = -ENOMEM; - goto err_hws; + ret = allocate_ring_buffer(ring); + if (ret) { + DRM_ERROR("Failed to allocate ringbuffer %s: %d\n", ring->name, ret); + goto error; } - ring->obj = obj; + /* Workaround an erratum on the i830 which causes a hang if + * the TAIL pointer points to within the last 2 cachelines + * of the buffer. + */ + ringbuf->effective_size = ringbuf->size; + if (IS_I830(dev) || IS_845G(dev)) + ringbuf->effective_size -= 2 * CACHELINE_BYTES; - ret = i915_gem_object_pin(obj, PAGE_SIZE, true); + ret = i915_cmd_parser_init_ring(ring); if (ret) - goto err_unref; - - ring->map.size = ring->size; - ring->map.offset = dev->agp->base + obj->gtt_offset; - ring->map.type = 0; - ring->map.flags = 0; - ring->map.mtrr = 0; + goto error; - drm_core_ioremap_wc(&ring->map, dev); - if (ring->map.handle == NULL) { - DRM_ERROR("Failed to map ringbuffer.\n"); - ret = -EINVAL; - goto err_unpin; - } - - ring->virtual_start = ring->map.handle; ret = ring->init(ring); if (ret) - goto err_unmap; - - /* Workaround an erratum on the i830 which causes a hang if - * the TAIL pointer points to within the last 2 cachelines - * of the buffer. - */ - ring->effective_size = ring->size; - if (IS_I830(ring->dev)) - ring->effective_size -= 128; + goto error; return 0; -err_unmap: - drm_core_ioremapfree(&ring->map, dev); -err_unpin: - i915_gem_object_unpin(obj); -err_unref: - drm_gem_object_unreference(&obj->base); - ring->obj = NULL; -err_hws: - cleanup_status_page(ring); +error: + kfree(ringbuf); + ring->buffer = NULL; return ret; } -void intel_cleanup_ring_buffer(struct intel_ring_buffer *ring) +void intel_cleanup_ring_buffer(struct intel_engine_cs *ring) { - struct drm_i915_private *dev_priv; - int ret; + struct drm_i915_private *dev_priv = to_i915(ring->dev); + struct intel_ringbuffer *ringbuf = ring->buffer; - if (ring->obj == NULL) + if (!intel_ring_initialized(ring)) return; - /* Disable the ring buffer. The ring must be idle at this point */ - dev_priv = ring->dev->dev_private; - ret = intel_wait_ring_idle(ring); - if (ret) - DRM_ERROR("failed to quiesce %s whilst cleaning up: %d\n", - ring->name, ret); - - I915_WRITE_CTL(ring, 0); + intel_stop_ring_buffer(ring); + WARN_ON(!IS_GEN2(ring->dev) && (I915_READ_MODE(ring) & MODE_IDLE) == 0); - drm_core_ioremapfree(&ring->map, ring->dev); + iounmap(ringbuf->virtual_start); - i915_gem_object_unpin(ring->obj); - drm_gem_object_unreference(&ring->obj->base); - ring->obj = NULL; + i915_gem_object_ggtt_unpin(ringbuf->obj); + drm_gem_object_unreference(&ringbuf->obj->base); + ringbuf->obj = NULL; + ring->preallocated_lazy_request = NULL; + ring->outstanding_lazy_seqno = 0; if (ring->cleanup) ring->cleanup(ring); cleanup_status_page(ring); + + i915_cmd_parser_fini_ring(ring); + + kfree(ringbuf); + ring->buffer = NULL; } -static int intel_wrap_ring_buffer(struct intel_ring_buffer *ring) +static int intel_ring_wait_request(struct intel_engine_cs *ring, int n) { - unsigned int *virt; - int rem = ring->size - ring->tail; + struct intel_ringbuffer *ringbuf = ring->buffer; + struct drm_i915_gem_request *request; + u32 seqno = 0; + int ret; - if (ring->space < rem) { - int ret = intel_wait_ring_buffer(ring, rem); - if (ret) - return ret; + if (ringbuf->last_retired_head != -1) { + ringbuf->head = ringbuf->last_retired_head; + ringbuf->last_retired_head = -1; + + ringbuf->space = ring_space(ring); + if (ringbuf->space >= n) + return 0; } - virt = (unsigned int *)(ring->virtual_start + ring->tail); - rem /= 8; - while (rem--) { - *virt++ = MI_NOOP; - *virt++ = MI_NOOP; + list_for_each_entry(request, &ring->request_list, list) { + if (__ring_space(request->tail, ringbuf->tail, ringbuf->size) >= n) { + seqno = request->seqno; + break; + } } - ring->tail = 0; - ring->space = ring_space(ring); + if (seqno == 0) + return -ENOSPC; + + ret = i915_wait_seqno(ring, seqno); + if (ret) + return ret; + i915_gem_retire_requests_ring(ring); + ringbuf->head = ringbuf->last_retired_head; + ringbuf->last_retired_head = -1; + + ringbuf->space = ring_space(ring); return 0; } -int intel_wait_ring_buffer(struct intel_ring_buffer *ring, int n) +static int ring_wait_for_space(struct intel_engine_cs *ring, int n) { struct drm_device *dev = ring->dev; struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_ringbuffer *ringbuf = ring->buffer; unsigned long end; - u32 head; + int ret; - /* If the reported head position has wrapped or hasn't advanced, - * fallback to the slow and accurate path. - */ - head = intel_read_status_page(ring, 4); - if (head > ring->head) { - ring->head = head; - ring->space = ring_space(ring); - if (ring->space >= n) - return 0; - } + ret = intel_ring_wait_request(ring, n); + if (ret != -ENOSPC) + return ret; - trace_i915_ring_wait_begin(ring); - if (drm_core_check_feature(dev, DRIVER_GEM)) - /* With GEM the hangcheck timer should kick us out of the loop, - * leaving it early runs the risk of corrupting GEM state (due - * to running on almost untested codepaths). But on resume - * timers don't work yet, so prevent a complete hang in that - * case by choosing an insanely large timeout. */ - end = jiffies + 60 * HZ; - else - end = jiffies + 3 * HZ; + /* force the tail write in case we have been skipping them */ + __intel_ring_advance(ring); + + /* With GEM the hangcheck timer should kick us out of the loop, + * leaving it early runs the risk of corrupting GEM state (due + * to running on almost untested codepaths). But on resume + * timers don't work yet, so prevent a complete hang in that + * case by choosing an insanely large timeout. */ + end = jiffies + 60 * HZ; + trace_i915_ring_wait_begin(ring); do { - ring->head = I915_READ_HEAD(ring); - ring->space = ring_space(ring); - if (ring->space >= n) { - trace_i915_ring_wait_end(ring); - return 0; + ringbuf->head = I915_READ_HEAD(ring); + ringbuf->space = ring_space(ring); + if (ringbuf->space >= n) { + ret = 0; + break; } - if (dev->primary->master) { + if (!drm_core_check_feature(dev, DRIVER_MODESET) && + dev->primary->master) { struct drm_i915_master_private *master_priv = dev->primary->master->driver_priv; if (master_priv->sarea_priv) master_priv->sarea_priv->perf_boxes |= I915_BOX_WAIT; } msleep(1); - if (atomic_read(&dev_priv->mm.wedged)) - return -EAGAIN; - } while (!time_after(jiffies, end)); + + if (dev_priv->mm.interruptible && signal_pending(current)) { + ret = -ERESTARTSYS; + break; + } + + ret = i915_gem_check_wedge(&dev_priv->gpu_error, + dev_priv->mm.interruptible); + if (ret) + break; + + if (time_after(jiffies, end)) { + ret = -EBUSY; + break; + } + } while (1); trace_i915_ring_wait_end(ring); - return -EBUSY; + return ret; } -int intel_ring_begin(struct intel_ring_buffer *ring, - int num_dwords) +static int intel_wrap_ring_buffer(struct intel_engine_cs *ring) { - struct drm_i915_private *dev_priv = ring->dev->dev_private; - int n = 4*num_dwords; + uint32_t __iomem *virt; + struct intel_ringbuffer *ringbuf = ring->buffer; + int rem = ringbuf->size - ringbuf->tail; + + if (ringbuf->space < rem) { + int ret = ring_wait_for_space(ring, rem); + if (ret) + return ret; + } + + virt = ringbuf->virtual_start + ringbuf->tail; + rem /= 4; + while (rem--) + iowrite32(MI_NOOP, virt++); + + ringbuf->tail = 0; + ringbuf->space = ring_space(ring); + + return 0; +} + +int intel_ring_idle(struct intel_engine_cs *ring) +{ + u32 seqno; int ret; - if (unlikely(atomic_read(&dev_priv->mm.wedged))) - return -EIO; + /* We need to add any requests required to flush the objects and ring */ + if (ring->outstanding_lazy_seqno) { + ret = i915_add_request(ring, NULL); + if (ret) + return ret; + } + + /* Wait upon the last request to be completed */ + if (list_empty(&ring->request_list)) + return 0; + + seqno = list_entry(ring->request_list.prev, + struct drm_i915_gem_request, + list)->seqno; + + return i915_wait_seqno(ring, seqno); +} + +static int +intel_ring_alloc_seqno(struct intel_engine_cs *ring) +{ + if (ring->outstanding_lazy_seqno) + return 0; + + if (ring->preallocated_lazy_request == NULL) { + struct drm_i915_gem_request *request; - if (unlikely(ring->tail + n > ring->effective_size)) { + request = kmalloc(sizeof(*request), GFP_KERNEL); + if (request == NULL) + return -ENOMEM; + + ring->preallocated_lazy_request = request; + } + + return i915_gem_get_seqno(ring->dev, &ring->outstanding_lazy_seqno); +} + +static int __intel_ring_prepare(struct intel_engine_cs *ring, + int bytes) +{ + struct intel_ringbuffer *ringbuf = ring->buffer; + int ret; + + if (unlikely(ringbuf->tail + bytes > ringbuf->effective_size)) { ret = intel_wrap_ring_buffer(ring); if (unlikely(ret)) return ret; } - if (unlikely(ring->space < n)) { - ret = intel_wait_ring_buffer(ring, n); + if (unlikely(ringbuf->space < bytes)) { + ret = ring_wait_for_space(ring, bytes); if (unlikely(ret)) return ret; } - ring->space -= n; return 0; } -void intel_ring_advance(struct intel_ring_buffer *ring) -{ - ring->tail &= ring->size - 1; - ring->write_tail(ring, ring->tail); -} - -static const struct intel_ring_buffer render_ring = { - .name = "render ring", - .id = RING_RENDER, - .mmio_base = RENDER_RING_BASE, - .size = 32 * PAGE_SIZE, - .init = init_render_ring, - .write_tail = ring_write_tail, - .flush = render_ring_flush, - .add_request = render_ring_add_request, - .get_seqno = ring_get_seqno, - .irq_get = render_ring_get_irq, - .irq_put = render_ring_put_irq, - .dispatch_execbuffer = render_ring_dispatch_execbuffer, - .cleanup = render_ring_cleanup, - .sync_to = render_ring_sync_to, - .semaphore_register = {MI_SEMAPHORE_SYNC_INVALID, - MI_SEMAPHORE_SYNC_RV, - MI_SEMAPHORE_SYNC_RB}, - .signal_mbox = {GEN6_VRSYNC, GEN6_BRSYNC}, -}; - -/* ring buffer for bit-stream decoder */ - -static const struct intel_ring_buffer bsd_ring = { - .name = "bsd ring", - .id = RING_BSD, - .mmio_base = BSD_RING_BASE, - .size = 32 * PAGE_SIZE, - .init = init_ring_common, - .write_tail = ring_write_tail, - .flush = bsd_ring_flush, - .add_request = ring_add_request, - .get_seqno = ring_get_seqno, - .irq_get = bsd_ring_get_irq, - .irq_put = bsd_ring_put_irq, - .dispatch_execbuffer = ring_dispatch_execbuffer, -}; - - -static void gen6_bsd_ring_write_tail(struct intel_ring_buffer *ring, +int intel_ring_begin(struct intel_engine_cs *ring, + int num_dwords) +{ + struct drm_i915_private *dev_priv = ring->dev->dev_private; + int ret; + + ret = i915_gem_check_wedge(&dev_priv->gpu_error, + dev_priv->mm.interruptible); + if (ret) + return ret; + + ret = __intel_ring_prepare(ring, num_dwords * sizeof(uint32_t)); + if (ret) + return ret; + + /* Preallocate the olr before touching the ring */ + ret = intel_ring_alloc_seqno(ring); + if (ret) + return ret; + + ring->buffer->space -= num_dwords * sizeof(uint32_t); + return 0; +} + +/* Align the ring tail to a cacheline boundary */ +int intel_ring_cacheline_align(struct intel_engine_cs *ring) +{ + int num_dwords = (ring->buffer->tail & (CACHELINE_BYTES - 1)) / sizeof(uint32_t); + int ret; + + if (num_dwords == 0) + return 0; + + num_dwords = CACHELINE_BYTES / sizeof(uint32_t) - num_dwords; + ret = intel_ring_begin(ring, num_dwords); + if (ret) + return ret; + + while (num_dwords--) + intel_ring_emit(ring, MI_NOOP); + + intel_ring_advance(ring); + + return 0; +} + +void intel_ring_init_seqno(struct intel_engine_cs *ring, u32 seqno) +{ + struct drm_i915_private *dev_priv = ring->dev->dev_private; + + BUG_ON(ring->outstanding_lazy_seqno); + + if (INTEL_INFO(ring->dev)->gen >= 6) { + I915_WRITE(RING_SYNC_0(ring->mmio_base), 0); + I915_WRITE(RING_SYNC_1(ring->mmio_base), 0); + if (HAS_VEBOX(ring->dev)) + I915_WRITE(RING_SYNC_2(ring->mmio_base), 0); + } + + ring->set_seqno(ring, seqno); + ring->hangcheck.seqno = seqno; +} + +static void gen6_bsd_ring_write_tail(struct intel_engine_cs *ring, u32 value) { - drm_i915_private_t *dev_priv = ring->dev->dev_private; + struct drm_i915_private *dev_priv = ring->dev->dev_private; /* Every tail move must follow the sequence below */ + + /* Disable notification that the ring is IDLE. The GT + * will then assume that it is busy and bring it out of rc6. + */ I915_WRITE(GEN6_BSD_SLEEP_PSMI_CONTROL, - GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_MODIFY_MASK | - GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_DISABLE); - I915_WRITE(GEN6_BSD_RNCID, 0x0); + _MASKED_BIT_ENABLE(GEN6_BSD_SLEEP_MSG_DISABLE)); + + /* Clear the context id. Here be magic! */ + I915_WRITE64(GEN6_BSD_RNCID, 0x0); + /* Wait for the ring not to be idle, i.e. for it to wake up. */ if (wait_for((I915_READ(GEN6_BSD_SLEEP_PSMI_CONTROL) & - GEN6_BSD_SLEEP_PSMI_CONTROL_IDLE_INDICATOR) == 0, - 50)) - DRM_ERROR("timed out waiting for IDLE Indicator\n"); + GEN6_BSD_SLEEP_INDICATOR) == 0, + 50)) + DRM_ERROR("timed out waiting for the BSD ring to wake up\n"); + /* Now that the ring is fully powered up, update the tail */ I915_WRITE_TAIL(ring, value); + POSTING_READ(RING_TAIL(ring->mmio_base)); + + /* Let the ring send IDLE messages to the GT again, + * and so let it sleep to conserve power when idle. + */ I915_WRITE(GEN6_BSD_SLEEP_PSMI_CONTROL, - GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_MODIFY_MASK | - GEN6_BSD_SLEEP_PSMI_CONTROL_RC_ILDL_MESSAGE_ENABLE); + _MASKED_BIT_DISABLE(GEN6_BSD_SLEEP_MSG_DISABLE)); } -static int gen6_ring_flush(struct intel_ring_buffer *ring, - u32 invalidate, u32 flush) +static int gen6_bsd_ring_flush(struct intel_engine_cs *ring, + u32 invalidate, u32 flush) { uint32_t cmd; int ret; @@ -1282,19 +1805,58 @@ static int gen6_ring_flush(struct intel_ring_buffer *ring, return ret; cmd = MI_FLUSH_DW; + if (INTEL_INFO(ring->dev)->gen >= 8) + cmd += 1; + /* + * Bspec vol 1c.5 - video engine command streamer: + * "If ENABLED, all TLBs will be invalidated once the flush + * operation is complete. This bit is only valid when the + * Post-Sync Operation field is a value of 1h or 3h." + */ if (invalidate & I915_GEM_GPU_DOMAINS) - cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD; + cmd |= MI_INVALIDATE_TLB | MI_INVALIDATE_BSD | + MI_FLUSH_DW_STORE_INDEX | MI_FLUSH_DW_OP_STOREDW; intel_ring_emit(ring, cmd); - intel_ring_emit(ring, 0); - intel_ring_emit(ring, 0); + intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT); + if (INTEL_INFO(ring->dev)->gen >= 8) { + intel_ring_emit(ring, 0); /* upper addr */ + intel_ring_emit(ring, 0); /* value */ + } else { + intel_ring_emit(ring, 0); + intel_ring_emit(ring, MI_NOOP); + } + intel_ring_advance(ring); + return 0; +} + +static int +gen8_ring_dispatch_execbuffer(struct intel_engine_cs *ring, + u64 offset, u32 len, + unsigned flags) +{ + struct drm_i915_private *dev_priv = ring->dev->dev_private; + bool ppgtt = dev_priv->mm.aliasing_ppgtt != NULL && + !(flags & I915_DISPATCH_SECURE); + int ret; + + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + + /* FIXME(BDW): Address space and security selectors. */ + intel_ring_emit(ring, MI_BATCH_BUFFER_START_GEN8 | (ppgtt<<8)); + intel_ring_emit(ring, lower_32_bits(offset)); + intel_ring_emit(ring, upper_32_bits(offset)); intel_ring_emit(ring, MI_NOOP); intel_ring_advance(ring); + return 0; } static int -gen6_ring_dispatch_execbuffer(struct intel_ring_buffer *ring, - u32 offset, u32 len) +hsw_ring_dispatch_execbuffer(struct intel_engine_cs *ring, + u64 offset, u32 len, + unsigned flags) { int ret; @@ -1302,7 +1864,9 @@ gen6_ring_dispatch_execbuffer(struct intel_ring_buffer *ring, if (ret) return ret; - intel_ring_emit(ring, MI_BATCH_BUFFER_START | MI_BATCH_NON_SECURE_I965); + intel_ring_emit(ring, + MI_BATCH_BUFFER_START | MI_BATCH_PPGTT_HSW | + (flags & I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE_HSW)); /* bit0-7 is the length on GEN6+ */ intel_ring_emit(ring, offset); intel_ring_advance(ring); @@ -1310,284 +1874,532 @@ gen6_ring_dispatch_execbuffer(struct intel_ring_buffer *ring, return 0; } -static bool -gen6_render_ring_get_irq(struct intel_ring_buffer *ring) +static int +gen6_ring_dispatch_execbuffer(struct intel_engine_cs *ring, + u64 offset, u32 len, + unsigned flags) { - return gen6_ring_get_irq(ring, - GT_USER_INTERRUPT, - GEN6_RENDER_USER_INTERRUPT); -} + int ret; -static void -gen6_render_ring_put_irq(struct intel_ring_buffer *ring) -{ - return gen6_ring_put_irq(ring, - GT_USER_INTERRUPT, - GEN6_RENDER_USER_INTERRUPT); -} + ret = intel_ring_begin(ring, 2); + if (ret) + return ret; -static bool -gen6_bsd_ring_get_irq(struct intel_ring_buffer *ring) -{ - return gen6_ring_get_irq(ring, - GT_GEN6_BSD_USER_INTERRUPT, - GEN6_BSD_USER_INTERRUPT); -} + intel_ring_emit(ring, + MI_BATCH_BUFFER_START | + (flags & I915_DISPATCH_SECURE ? 0 : MI_BATCH_NON_SECURE_I965)); + /* bit0-7 is the length on GEN6+ */ + intel_ring_emit(ring, offset); + intel_ring_advance(ring); -static void -gen6_bsd_ring_put_irq(struct intel_ring_buffer *ring) -{ - return gen6_ring_put_irq(ring, - GT_GEN6_BSD_USER_INTERRUPT, - GEN6_BSD_USER_INTERRUPT); -} - -/* ring buffer for Video Codec for Gen6+ */ -static const struct intel_ring_buffer gen6_bsd_ring = { - .name = "gen6 bsd ring", - .id = RING_BSD, - .mmio_base = GEN6_BSD_RING_BASE, - .size = 32 * PAGE_SIZE, - .init = init_ring_common, - .write_tail = gen6_bsd_ring_write_tail, - .flush = gen6_ring_flush, - .add_request = gen6_add_request, - .get_seqno = gen6_ring_get_seqno, - .irq_get = gen6_bsd_ring_get_irq, - .irq_put = gen6_bsd_ring_put_irq, - .dispatch_execbuffer = gen6_ring_dispatch_execbuffer, - .sync_to = gen6_bsd_ring_sync_to, - .semaphore_register = {MI_SEMAPHORE_SYNC_VR, - MI_SEMAPHORE_SYNC_INVALID, - MI_SEMAPHORE_SYNC_VB}, - .signal_mbox = {GEN6_RVSYNC, GEN6_BVSYNC}, -}; + return 0; +} /* Blitter support (SandyBridge+) */ -static bool -blt_ring_get_irq(struct intel_ring_buffer *ring) +static int gen6_ring_flush(struct intel_engine_cs *ring, + u32 invalidate, u32 flush) { - return gen6_ring_get_irq(ring, - GT_BLT_USER_INTERRUPT, - GEN6_BLITTER_USER_INTERRUPT); -} + struct drm_device *dev = ring->dev; + uint32_t cmd; + int ret; -static void -blt_ring_put_irq(struct intel_ring_buffer *ring) -{ - gen6_ring_put_irq(ring, - GT_BLT_USER_INTERRUPT, - GEN6_BLITTER_USER_INTERRUPT); -} + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + cmd = MI_FLUSH_DW; + if (INTEL_INFO(ring->dev)->gen >= 8) + cmd += 1; + /* + * Bspec vol 1c.3 - blitter engine command streamer: + * "If ENABLED, all TLBs will be invalidated once the flush + * operation is complete. This bit is only valid when the + * Post-Sync Operation field is a value of 1h or 3h." + */ + if (invalidate & I915_GEM_DOMAIN_RENDER) + cmd |= MI_INVALIDATE_TLB | MI_FLUSH_DW_STORE_INDEX | + MI_FLUSH_DW_OP_STOREDW; + intel_ring_emit(ring, cmd); + intel_ring_emit(ring, I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT); + if (INTEL_INFO(ring->dev)->gen >= 8) { + intel_ring_emit(ring, 0); /* upper addr */ + intel_ring_emit(ring, 0); /* value */ + } else { + intel_ring_emit(ring, 0); + intel_ring_emit(ring, MI_NOOP); + } + intel_ring_advance(ring); -/* Workaround for some stepping of SNB, - * each time when BLT engine ring tail moved, - * the first command in the ring to be parsed - * should be MI_BATCH_BUFFER_START - */ -#define NEED_BLT_WORKAROUND(dev) \ - (IS_GEN6(dev) && (dev->pdev->revision < 8)) + if (IS_GEN7(dev) && !invalidate && flush) + return gen7_ring_fbc_flush(ring, FBC_REND_CACHE_CLEAN); -static inline struct drm_i915_gem_object * -to_blt_workaround(struct intel_ring_buffer *ring) -{ - return ring->private; + return 0; } -static int blt_ring_init(struct intel_ring_buffer *ring) +int intel_init_render_ring_buffer(struct drm_device *dev) { - if (NEED_BLT_WORKAROUND(ring->dev)) { + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_engine_cs *ring = &dev_priv->ring[RCS]; + + ring->name = "render ring"; + ring->id = RCS; + ring->mmio_base = RENDER_RING_BASE; + + if (INTEL_INFO(dev)->gen >= 6) { + ring->add_request = gen6_add_request; + ring->flush = gen7_render_ring_flush; + if (INTEL_INFO(dev)->gen == 6) + ring->flush = gen6_render_ring_flush; + if (INTEL_INFO(dev)->gen >= 8) { + ring->flush = gen8_render_ring_flush; + ring->irq_get = gen8_ring_get_irq; + ring->irq_put = gen8_ring_put_irq; + } else { + ring->irq_get = gen6_ring_get_irq; + ring->irq_put = gen6_ring_put_irq; + } + ring->irq_enable_mask = GT_RENDER_USER_INTERRUPT; + ring->get_seqno = gen6_ring_get_seqno; + ring->set_seqno = ring_set_seqno; + ring->semaphore.sync_to = gen6_ring_sync; + ring->semaphore.signal = gen6_signal; + /* + * The current semaphore is only applied on pre-gen8 platform. + * And there is no VCS2 ring on the pre-gen8 platform. So the + * semaphore between RCS and VCS2 is initialized as INVALID. + * Gen8 will initialize the sema between VCS2 and RCS later. + */ + ring->semaphore.mbox.wait[RCS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.wait[VCS] = MI_SEMAPHORE_SYNC_RV; + ring->semaphore.mbox.wait[BCS] = MI_SEMAPHORE_SYNC_RB; + ring->semaphore.mbox.wait[VECS] = MI_SEMAPHORE_SYNC_RVE; + ring->semaphore.mbox.wait[VCS2] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.signal[RCS] = GEN6_NOSYNC; + ring->semaphore.mbox.signal[VCS] = GEN6_VRSYNC; + ring->semaphore.mbox.signal[BCS] = GEN6_BRSYNC; + ring->semaphore.mbox.signal[VECS] = GEN6_VERSYNC; + ring->semaphore.mbox.signal[VCS2] = GEN6_NOSYNC; + } else if (IS_GEN5(dev)) { + ring->add_request = pc_render_add_request; + ring->flush = gen4_render_ring_flush; + ring->get_seqno = pc_render_get_seqno; + ring->set_seqno = pc_render_set_seqno; + ring->irq_get = gen5_ring_get_irq; + ring->irq_put = gen5_ring_put_irq; + ring->irq_enable_mask = GT_RENDER_USER_INTERRUPT | + GT_RENDER_PIPECTL_NOTIFY_INTERRUPT; + } else { + ring->add_request = i9xx_add_request; + if (INTEL_INFO(dev)->gen < 4) + ring->flush = gen2_render_ring_flush; + else + ring->flush = gen4_render_ring_flush; + ring->get_seqno = ring_get_seqno; + ring->set_seqno = ring_set_seqno; + if (IS_GEN2(dev)) { + ring->irq_get = i8xx_ring_get_irq; + ring->irq_put = i8xx_ring_put_irq; + } else { + ring->irq_get = i9xx_ring_get_irq; + ring->irq_put = i9xx_ring_put_irq; + } + ring->irq_enable_mask = I915_USER_INTERRUPT; + } + ring->write_tail = ring_write_tail; + if (IS_HASWELL(dev)) + ring->dispatch_execbuffer = hsw_ring_dispatch_execbuffer; + else if (IS_GEN8(dev)) + ring->dispatch_execbuffer = gen8_ring_dispatch_execbuffer; + else if (INTEL_INFO(dev)->gen >= 6) + ring->dispatch_execbuffer = gen6_ring_dispatch_execbuffer; + else if (INTEL_INFO(dev)->gen >= 4) + ring->dispatch_execbuffer = i965_dispatch_execbuffer; + else if (IS_I830(dev) || IS_845G(dev)) + ring->dispatch_execbuffer = i830_dispatch_execbuffer; + else + ring->dispatch_execbuffer = i915_dispatch_execbuffer; + ring->init = init_render_ring; + ring->cleanup = render_ring_cleanup; + + /* Workaround batchbuffer to combat CS tlb bug. */ + if (HAS_BROKEN_CS_TLB(dev)) { struct drm_i915_gem_object *obj; - u32 *ptr; int ret; - obj = i915_gem_alloc_object(ring->dev, 4096); - if (obj == NULL) + obj = i915_gem_alloc_object(dev, I830_BATCH_LIMIT); + if (obj == NULL) { + DRM_ERROR("Failed to allocate batch bo\n"); return -ENOMEM; - - ret = i915_gem_object_pin(obj, 4096, true); - if (ret) { - drm_gem_object_unreference(&obj->base); - return ret; } - ptr = kmap(obj->pages[0]); - *ptr++ = MI_BATCH_BUFFER_END; - *ptr++ = MI_NOOP; - kunmap(obj->pages[0]); - - ret = i915_gem_object_set_to_gtt_domain(obj, false); - if (ret) { - i915_gem_object_unpin(obj); + ret = i915_gem_obj_ggtt_pin(obj, 0, 0); + if (ret != 0) { drm_gem_object_unreference(&obj->base); + DRM_ERROR("Failed to ping batch bo\n"); return ret; } - ring->private = obj; + ring->scratch.obj = obj; + ring->scratch.gtt_offset = i915_gem_obj_ggtt_offset(obj); } - return init_ring_common(ring); + return intel_init_ring_buffer(dev, ring); } -static int blt_ring_begin(struct intel_ring_buffer *ring, - int num_dwords) +int intel_render_ring_init_dri(struct drm_device *dev, u64 start, u32 size) { - if (ring->private) { - int ret = intel_ring_begin(ring, num_dwords+2); - if (ret) - return ret; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_engine_cs *ring = &dev_priv->ring[RCS]; + struct intel_ringbuffer *ringbuf = ring->buffer; + int ret; - intel_ring_emit(ring, MI_BATCH_BUFFER_START); - intel_ring_emit(ring, to_blt_workaround(ring)->gtt_offset); + if (ringbuf == NULL) { + ringbuf = kzalloc(sizeof(*ringbuf), GFP_KERNEL); + if (!ringbuf) + return -ENOMEM; + ring->buffer = ringbuf; + } - return 0; - } else - return intel_ring_begin(ring, 4); -} + ring->name = "render ring"; + ring->id = RCS; + ring->mmio_base = RENDER_RING_BASE; -static int blt_ring_flush(struct intel_ring_buffer *ring, - u32 invalidate, u32 flush) -{ - uint32_t cmd; - int ret; + if (INTEL_INFO(dev)->gen >= 6) { + /* non-kms not supported on gen6+ */ + ret = -ENODEV; + goto err_ringbuf; + } - ret = blt_ring_begin(ring, 4); - if (ret) - return ret; + /* Note: gem is not supported on gen5/ilk without kms (the corresponding + * gem_init ioctl returns with -ENODEV). Hence we do not need to set up + * the special gen5 functions. */ + ring->add_request = i9xx_add_request; + if (INTEL_INFO(dev)->gen < 4) + ring->flush = gen2_render_ring_flush; + else + ring->flush = gen4_render_ring_flush; + ring->get_seqno = ring_get_seqno; + ring->set_seqno = ring_set_seqno; + if (IS_GEN2(dev)) { + ring->irq_get = i8xx_ring_get_irq; + ring->irq_put = i8xx_ring_put_irq; + } else { + ring->irq_get = i9xx_ring_get_irq; + ring->irq_put = i9xx_ring_put_irq; + } + ring->irq_enable_mask = I915_USER_INTERRUPT; + ring->write_tail = ring_write_tail; + if (INTEL_INFO(dev)->gen >= 4) + ring->dispatch_execbuffer = i965_dispatch_execbuffer; + else if (IS_I830(dev) || IS_845G(dev)) + ring->dispatch_execbuffer = i830_dispatch_execbuffer; + else + ring->dispatch_execbuffer = i915_dispatch_execbuffer; + ring->init = init_render_ring; + ring->cleanup = render_ring_cleanup; + + ring->dev = dev; + INIT_LIST_HEAD(&ring->active_list); + INIT_LIST_HEAD(&ring->request_list); + + ringbuf->size = size; + ringbuf->effective_size = ringbuf->size; + if (IS_I830(ring->dev) || IS_845G(ring->dev)) + ringbuf->effective_size -= 2 * CACHELINE_BYTES; + + ringbuf->virtual_start = ioremap_wc(start, size); + if (ringbuf->virtual_start == NULL) { + DRM_ERROR("can not ioremap virtual address for" + " ring buffer\n"); + ret = -ENOMEM; + goto err_ringbuf; + } + + if (!I915_NEED_GFX_HWS(dev)) { + ret = init_phys_status_page(ring); + if (ret) + goto err_vstart; + } - cmd = MI_FLUSH_DW; - if (invalidate & I915_GEM_DOMAIN_RENDER) - cmd |= MI_INVALIDATE_TLB; - intel_ring_emit(ring, cmd); - intel_ring_emit(ring, 0); - intel_ring_emit(ring, 0); - intel_ring_emit(ring, MI_NOOP); - intel_ring_advance(ring); return 0; + +err_vstart: + iounmap(ringbuf->virtual_start); +err_ringbuf: + kfree(ringbuf); + ring->buffer = NULL; + return ret; } -static void blt_ring_cleanup(struct intel_ring_buffer *ring) +int intel_init_bsd_ring_buffer(struct drm_device *dev) { - if (!ring->private) - return; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_engine_cs *ring = &dev_priv->ring[VCS]; - i915_gem_object_unpin(ring->private); - drm_gem_object_unreference(ring->private); - ring->private = NULL; -} - -static const struct intel_ring_buffer gen6_blt_ring = { - .name = "blt ring", - .id = RING_BLT, - .mmio_base = BLT_RING_BASE, - .size = 32 * PAGE_SIZE, - .init = blt_ring_init, - .write_tail = ring_write_tail, - .flush = blt_ring_flush, - .add_request = gen6_add_request, - .get_seqno = gen6_ring_get_seqno, - .irq_get = blt_ring_get_irq, - .irq_put = blt_ring_put_irq, - .dispatch_execbuffer = gen6_ring_dispatch_execbuffer, - .cleanup = blt_ring_cleanup, - .sync_to = gen6_blt_ring_sync_to, - .semaphore_register = {MI_SEMAPHORE_SYNC_BR, - MI_SEMAPHORE_SYNC_BV, - MI_SEMAPHORE_SYNC_INVALID}, - .signal_mbox = {GEN6_RBSYNC, GEN6_VBSYNC}, -}; + ring->name = "bsd ring"; + ring->id = VCS; -int intel_init_render_ring_buffer(struct drm_device *dev) -{ - drm_i915_private_t *dev_priv = dev->dev_private; - struct intel_ring_buffer *ring = &dev_priv->ring[RCS]; - - *ring = render_ring; + ring->write_tail = ring_write_tail; if (INTEL_INFO(dev)->gen >= 6) { + ring->mmio_base = GEN6_BSD_RING_BASE; + /* gen6 bsd needs a special wa for tail updates */ + if (IS_GEN6(dev)) + ring->write_tail = gen6_bsd_ring_write_tail; + ring->flush = gen6_bsd_ring_flush; ring->add_request = gen6_add_request; - ring->flush = gen6_render_ring_flush; - ring->irq_get = gen6_render_ring_get_irq; - ring->irq_put = gen6_render_ring_put_irq; ring->get_seqno = gen6_ring_get_seqno; - } else if (IS_GEN5(dev)) { - ring->add_request = pc_render_add_request; - ring->get_seqno = pc_render_get_seqno; + ring->set_seqno = ring_set_seqno; + if (INTEL_INFO(dev)->gen >= 8) { + ring->irq_enable_mask = + GT_RENDER_USER_INTERRUPT << GEN8_VCS1_IRQ_SHIFT; + ring->irq_get = gen8_ring_get_irq; + ring->irq_put = gen8_ring_put_irq; + ring->dispatch_execbuffer = + gen8_ring_dispatch_execbuffer; + } else { + ring->irq_enable_mask = GT_BSD_USER_INTERRUPT; + ring->irq_get = gen6_ring_get_irq; + ring->irq_put = gen6_ring_put_irq; + ring->dispatch_execbuffer = + gen6_ring_dispatch_execbuffer; + } + ring->semaphore.sync_to = gen6_ring_sync; + ring->semaphore.signal = gen6_signal; + /* + * The current semaphore is only applied on pre-gen8 platform. + * And there is no VCS2 ring on the pre-gen8 platform. So the + * semaphore between VCS and VCS2 is initialized as INVALID. + * Gen8 will initialize the sema between VCS2 and VCS later. + */ + ring->semaphore.mbox.wait[RCS] = MI_SEMAPHORE_SYNC_VR; + ring->semaphore.mbox.wait[VCS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.wait[BCS] = MI_SEMAPHORE_SYNC_VB; + ring->semaphore.mbox.wait[VECS] = MI_SEMAPHORE_SYNC_VVE; + ring->semaphore.mbox.wait[VCS2] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.signal[RCS] = GEN6_RVSYNC; + ring->semaphore.mbox.signal[VCS] = GEN6_NOSYNC; + ring->semaphore.mbox.signal[BCS] = GEN6_BVSYNC; + ring->semaphore.mbox.signal[VECS] = GEN6_VEVSYNC; + ring->semaphore.mbox.signal[VCS2] = GEN6_NOSYNC; + } else { + ring->mmio_base = BSD_RING_BASE; + ring->flush = bsd_ring_flush; + ring->add_request = i9xx_add_request; + ring->get_seqno = ring_get_seqno; + ring->set_seqno = ring_set_seqno; + if (IS_GEN5(dev)) { + ring->irq_enable_mask = ILK_BSD_USER_INTERRUPT; + ring->irq_get = gen5_ring_get_irq; + ring->irq_put = gen5_ring_put_irq; + } else { + ring->irq_enable_mask = I915_BSD_USER_INTERRUPT; + ring->irq_get = i9xx_ring_get_irq; + ring->irq_put = i9xx_ring_put_irq; + } + ring->dispatch_execbuffer = i965_dispatch_execbuffer; } + ring->init = init_ring_common; - if (!I915_NEED_GFX_HWS(dev)) { - ring->status_page.page_addr = dev_priv->status_page_dmah->vaddr; - memset(ring->status_page.page_addr, 0, PAGE_SIZE); + return intel_init_ring_buffer(dev, ring); +} + +/** + * Initialize the second BSD ring for Broadwell GT3. + * It is noted that this only exists on Broadwell GT3. + */ +int intel_init_bsd2_ring_buffer(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_engine_cs *ring = &dev_priv->ring[VCS2]; + + if ((INTEL_INFO(dev)->gen != 8)) { + DRM_ERROR("No dual-BSD ring on non-BDW machine\n"); + return -EINVAL; } + ring->name = "bds2_ring"; + ring->id = VCS2; + + ring->write_tail = ring_write_tail; + ring->mmio_base = GEN8_BSD2_RING_BASE; + ring->flush = gen6_bsd_ring_flush; + ring->add_request = gen6_add_request; + ring->get_seqno = gen6_ring_get_seqno; + ring->set_seqno = ring_set_seqno; + ring->irq_enable_mask = + GT_RENDER_USER_INTERRUPT << GEN8_VCS2_IRQ_SHIFT; + ring->irq_get = gen8_ring_get_irq; + ring->irq_put = gen8_ring_put_irq; + ring->dispatch_execbuffer = + gen8_ring_dispatch_execbuffer; + ring->semaphore.sync_to = gen6_ring_sync; + ring->semaphore.signal = gen6_signal; + /* + * The current semaphore is only applied on the pre-gen8. And there + * is no bsd2 ring on the pre-gen8. So now the semaphore_register + * between VCS2 and other ring is initialized as invalid. + * Gen8 will initialize the sema between VCS2 and other ring later. + */ + ring->semaphore.mbox.wait[RCS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.wait[VCS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.wait[BCS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.wait[VECS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.wait[VCS2] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.signal[RCS] = GEN6_NOSYNC; + ring->semaphore.mbox.signal[VCS] = GEN6_NOSYNC; + ring->semaphore.mbox.signal[BCS] = GEN6_NOSYNC; + ring->semaphore.mbox.signal[VECS] = GEN6_NOSYNC; + ring->semaphore.mbox.signal[VCS2] = GEN6_NOSYNC; + + ring->init = init_ring_common; + return intel_init_ring_buffer(dev, ring); } -int intel_render_ring_init_dri(struct drm_device *dev, u64 start, u32 size) +int intel_init_blt_ring_buffer(struct drm_device *dev) { - drm_i915_private_t *dev_priv = dev->dev_private; - struct intel_ring_buffer *ring = &dev_priv->ring[RCS]; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_engine_cs *ring = &dev_priv->ring[BCS]; + + ring->name = "blitter ring"; + ring->id = BCS; + + ring->mmio_base = BLT_RING_BASE; + ring->write_tail = ring_write_tail; + ring->flush = gen6_ring_flush; + ring->add_request = gen6_add_request; + ring->get_seqno = gen6_ring_get_seqno; + ring->set_seqno = ring_set_seqno; + if (INTEL_INFO(dev)->gen >= 8) { + ring->irq_enable_mask = + GT_RENDER_USER_INTERRUPT << GEN8_BCS_IRQ_SHIFT; + ring->irq_get = gen8_ring_get_irq; + ring->irq_put = gen8_ring_put_irq; + ring->dispatch_execbuffer = gen8_ring_dispatch_execbuffer; + } else { + ring->irq_enable_mask = GT_BLT_USER_INTERRUPT; + ring->irq_get = gen6_ring_get_irq; + ring->irq_put = gen6_ring_put_irq; + ring->dispatch_execbuffer = gen6_ring_dispatch_execbuffer; + } + ring->semaphore.sync_to = gen6_ring_sync; + ring->semaphore.signal = gen6_signal; + /* + * The current semaphore is only applied on pre-gen8 platform. And + * there is no VCS2 ring on the pre-gen8 platform. So the semaphore + * between BCS and VCS2 is initialized as INVALID. + * Gen8 will initialize the sema between BCS and VCS2 later. + */ + ring->semaphore.mbox.wait[RCS] = MI_SEMAPHORE_SYNC_BR; + ring->semaphore.mbox.wait[VCS] = MI_SEMAPHORE_SYNC_BV; + ring->semaphore.mbox.wait[BCS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.wait[VECS] = MI_SEMAPHORE_SYNC_BVE; + ring->semaphore.mbox.wait[VCS2] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.signal[RCS] = GEN6_RBSYNC; + ring->semaphore.mbox.signal[VCS] = GEN6_VBSYNC; + ring->semaphore.mbox.signal[BCS] = GEN6_NOSYNC; + ring->semaphore.mbox.signal[VECS] = GEN6_VEBSYNC; + ring->semaphore.mbox.signal[VCS2] = GEN6_NOSYNC; + ring->init = init_ring_common; - *ring = render_ring; - if (INTEL_INFO(dev)->gen >= 6) { - ring->add_request = gen6_add_request; - ring->irq_get = gen6_render_ring_get_irq; - ring->irq_put = gen6_render_ring_put_irq; - } else if (IS_GEN5(dev)) { - ring->add_request = pc_render_add_request; - ring->get_seqno = pc_render_get_seqno; + return intel_init_ring_buffer(dev, ring); +} + +int intel_init_vebox_ring_buffer(struct drm_device *dev) +{ + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_engine_cs *ring = &dev_priv->ring[VECS]; + + ring->name = "video enhancement ring"; + ring->id = VECS; + + ring->mmio_base = VEBOX_RING_BASE; + ring->write_tail = ring_write_tail; + ring->flush = gen6_ring_flush; + ring->add_request = gen6_add_request; + ring->get_seqno = gen6_ring_get_seqno; + ring->set_seqno = ring_set_seqno; + + if (INTEL_INFO(dev)->gen >= 8) { + ring->irq_enable_mask = + GT_RENDER_USER_INTERRUPT << GEN8_VECS_IRQ_SHIFT; + ring->irq_get = gen8_ring_get_irq; + ring->irq_put = gen8_ring_put_irq; + ring->dispatch_execbuffer = gen8_ring_dispatch_execbuffer; + } else { + ring->irq_enable_mask = PM_VEBOX_USER_INTERRUPT; + ring->irq_get = hsw_vebox_get_irq; + ring->irq_put = hsw_vebox_put_irq; + ring->dispatch_execbuffer = gen6_ring_dispatch_execbuffer; } + ring->semaphore.sync_to = gen6_ring_sync; + ring->semaphore.signal = gen6_signal; + ring->semaphore.mbox.wait[RCS] = MI_SEMAPHORE_SYNC_VER; + ring->semaphore.mbox.wait[VCS] = MI_SEMAPHORE_SYNC_VEV; + ring->semaphore.mbox.wait[BCS] = MI_SEMAPHORE_SYNC_VEB; + ring->semaphore.mbox.wait[VECS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.wait[VCS2] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore.mbox.signal[RCS] = GEN6_RVESYNC; + ring->semaphore.mbox.signal[VCS] = GEN6_VVESYNC; + ring->semaphore.mbox.signal[BCS] = GEN6_BVESYNC; + ring->semaphore.mbox.signal[VECS] = GEN6_NOSYNC; + ring->semaphore.mbox.signal[VCS2] = GEN6_NOSYNC; + ring->init = init_ring_common; - if (!I915_NEED_GFX_HWS(dev)) - ring->status_page.page_addr = dev_priv->status_page_dmah->vaddr; + return intel_init_ring_buffer(dev, ring); +} - ring->dev = dev; - INIT_LIST_HEAD(&ring->active_list); - INIT_LIST_HEAD(&ring->request_list); - INIT_LIST_HEAD(&ring->gpu_write_list); +int +intel_ring_flush_all_caches(struct intel_engine_cs *ring) +{ + int ret; - ring->size = size; - ring->effective_size = ring->size; - if (IS_I830(ring->dev)) - ring->effective_size -= 128; + if (!ring->gpu_caches_dirty) + return 0; - ring->map.offset = start; - ring->map.size = size; - ring->map.type = 0; - ring->map.flags = 0; - ring->map.mtrr = 0; + ret = ring->flush(ring, 0, I915_GEM_GPU_DOMAINS); + if (ret) + return ret; - drm_core_ioremap_wc(&ring->map, dev); - if (ring->map.handle == NULL) { - DRM_ERROR("can not ioremap virtual address for" - " ring buffer\n"); - return -ENOMEM; - } + trace_i915_gem_ring_flush(ring, 0, I915_GEM_GPU_DOMAINS); - ring->virtual_start = (void __force __iomem *)ring->map.handle; + ring->gpu_caches_dirty = false; return 0; } -int intel_init_bsd_ring_buffer(struct drm_device *dev) +int +intel_ring_invalidate_all_caches(struct intel_engine_cs *ring) { - drm_i915_private_t *dev_priv = dev->dev_private; - struct intel_ring_buffer *ring = &dev_priv->ring[VCS]; + uint32_t flush_domains; + int ret; - if (IS_GEN6(dev) || IS_GEN7(dev)) - *ring = gen6_bsd_ring; - else - *ring = bsd_ring; + flush_domains = 0; + if (ring->gpu_caches_dirty) + flush_domains = I915_GEM_GPU_DOMAINS; - return intel_init_ring_buffer(dev, ring); + ret = ring->flush(ring, I915_GEM_GPU_DOMAINS, flush_domains); + if (ret) + return ret; + + trace_i915_gem_ring_flush(ring, I915_GEM_GPU_DOMAINS, flush_domains); + + ring->gpu_caches_dirty = false; + return 0; } -int intel_init_blt_ring_buffer(struct drm_device *dev) +void +intel_stop_ring_buffer(struct intel_engine_cs *ring) { - drm_i915_private_t *dev_priv = dev->dev_private; - struct intel_ring_buffer *ring = &dev_priv->ring[BCS]; + int ret; - *ring = gen6_blt_ring; + if (!intel_ring_initialized(ring)) + return; - return intel_init_ring_buffer(dev, ring); + ret = intel_ring_idle(ring); + if (ret && !i915_reset_in_progress(&to_i915(ring->dev)->gpu_error)) + DRM_ERROR("failed to quiesce %s whilst cleaning up: %d\n", + ring->name, ret); + + stop_ring(ring); } |