From 8693a824873a0a97be77c1d5f1e98777f06f7305 Mon Sep 17 00:00:00 2001 From: Damien Lespiau Date: Fri, 3 May 2013 18:48:11 +0100 Subject: drm/i915: Add references to some workaround we implement We did not mention the workaround name when implementing those. This should help us track what we already implement. Signed-off-by: Damien Lespiau Reviewed-by: Imre Deak Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 2 ++ 1 file changed, 2 insertions(+) (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 1d5d613eb6be..3d2c236e15ab 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -515,6 +515,8 @@ static int init_render_ring(struct intel_ring_buffer *ring) /* 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 */ if (INTEL_INFO(dev)->gen >= 6) I915_WRITE(MI_MODE, _MASKED_BIT_ENABLE(ASYNC_FLIP_PERF_DISABLE)); -- cgit v1.2.3 From 92cab7345131db7af18f630a799ce6b2e8e624c5 Mon Sep 17 00:00:00 2001 From: Mika Kuoppala Date: Fri, 24 May 2013 17:16:07 +0300 Subject: drm/i915: track ring progression using seqnos Instead of relying in acthd, track ring seqno progression to detect if ring has hung. v2: put hangcheck stuff inside struct (Chris Wilson) v3: initialize hangcheck.seqno (Ben Widawsky) Signed-off-by: Mika Kuoppala Reviewed-by: Ben Widawsky Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 1 + 1 file changed, 1 insertion(+) (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 3d2c236e15ab..5698faeee6f6 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -1502,6 +1502,7 @@ void intel_ring_init_seqno(struct intel_ring_buffer *ring, u32 seqno) } ring->set_seqno(ring, seqno); + ring->hangcheck.seqno = seqno; } void intel_ring_advance(struct intel_ring_buffer *ring) -- cgit v1.2.3 From 56b085a01befc7907d270e9acb349580015e7bad Mon Sep 17 00:00:00 2001 From: Wei Yongjun Date: Tue, 28 May 2013 17:51:44 +0800 Subject: drm/i915: fix error return code in init_pipe_control() Fix to return -ENOMEM in the kmap() error handling case instead of 0, as done elsewhere in this function. Signed-off-by: Wei Yongjun Signed-off-by: Daniel Vetter --- 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 5698faeee6f6..9b97cf66a5ae 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -464,9 +464,11 @@ init_pipe_control(struct intel_ring_buffer *ring) goto err_unref; pc->gtt_offset = obj->gtt_offset; - pc->cpu_page = kmap(sg_page(obj->pages->sgl)); - if (pc->cpu_page == NULL) + pc->cpu_page = kmap(sg_page(obj->pages->sgl)); + if (pc->cpu_page == NULL) { + ret = -ENOMEM; goto err_unpin; + } DRM_DEBUG_DRIVER("%s pipe control offset: 0x%08x\n", ring->name, pc->gtt_offset); -- cgit v1.2.3 From 5586181fce2b2e89a0e281d78ffbdfa103bb0dde Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:17 -0700 Subject: drm/i915: Comments for semaphore clarification Semaphores are tied very closely to the rings in the GPU. Trivial patch adds comments to the existing code so that when we add new rings we can include comments there as well. It also helps distinguish the ring to semaphore mailbox interactions by using the ringname in the semaphore data structures. This patch should have no functional impact. v2: The English parts (as opposed to register names) of the comments were reversed. (Damien) Signed-off-by: Ben Widawsky Reviewed-by: Damien Lespiau Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 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 9b97cf66a5ae..2d2a3622639c 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -1671,9 +1671,9 @@ int intel_init_render_ring_buffer(struct drm_device *dev) ring->get_seqno = gen6_ring_get_seqno; ring->set_seqno = ring_set_seqno; ring->sync_to = gen6_ring_sync; - ring->semaphore_register[0] = MI_SEMAPHORE_SYNC_INVALID; - ring->semaphore_register[1] = MI_SEMAPHORE_SYNC_RV; - ring->semaphore_register[2] = MI_SEMAPHORE_SYNC_RB; + ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_RV; + ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_RB; ring->signal_mbox[0] = GEN6_VRSYNC; ring->signal_mbox[1] = GEN6_BRSYNC; } else if (IS_GEN5(dev)) { @@ -1830,9 +1830,9 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev) ring->irq_put = gen6_ring_put_irq; ring->dispatch_execbuffer = gen6_ring_dispatch_execbuffer; ring->sync_to = gen6_ring_sync; - ring->semaphore_register[0] = MI_SEMAPHORE_SYNC_VR; - ring->semaphore_register[1] = MI_SEMAPHORE_SYNC_INVALID; - ring->semaphore_register[2] = MI_SEMAPHORE_SYNC_VB; + ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_VR; + ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_VB; ring->signal_mbox[0] = GEN6_RVSYNC; ring->signal_mbox[1] = GEN6_BVSYNC; } else { @@ -1876,9 +1876,9 @@ int intel_init_blt_ring_buffer(struct drm_device *dev) ring->irq_put = gen6_ring_put_irq; ring->dispatch_execbuffer = gen6_ring_dispatch_execbuffer; ring->sync_to = gen6_ring_sync; - ring->semaphore_register[0] = MI_SEMAPHORE_SYNC_BR; - ring->semaphore_register[1] = MI_SEMAPHORE_SYNC_BV; - ring->semaphore_register[2] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_BR; + ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_BV; + ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_INVALID; ring->signal_mbox[0] = GEN6_RBSYNC; ring->signal_mbox[1] = GEN6_VBSYNC; ring->init = init_ring_common; -- cgit v1.2.3 From ad776f8b09d66e0145479fdbde2a710e5892441f Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:18 -0700 Subject: drm/i915: Semaphore MBOX update generalization This replaces the existing MBOX update code with a more generalized calculation for emitting mbox updates. We also create a sentinel for doing the updates so we can more abstractly deal with the rings. When doing MBOX updates the code must be aware of the /other/ rings. Until now the platforms which supported semaphores had a fixed number of rings and so it made sense for the code to be very specialized (hardcoded). The patch does contain a functional change, but should have no behavioral changes. Signed-off-by: Ben Widawsky Reviewed-by: Damien Lespiau Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 43 ++++++++++++++++++++++----------- 1 file changed, 29 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 2d2a3622639c..5df179127e8b 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -582,9 +582,16 @@ static void update_mboxes(struct intel_ring_buffer *ring, u32 mmio_offset) { +/* 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 intel_ring_emit(ring, MI_LOAD_REGISTER_IMM(1)); intel_ring_emit(ring, mmio_offset); intel_ring_emit(ring, ring->outstanding_lazy_request); + intel_ring_emit(ring, MI_NOOP); } /** @@ -599,19 +606,24 @@ update_mboxes(struct intel_ring_buffer *ring, static int gen6_add_request(struct intel_ring_buffer *ring) { - u32 mbox1_reg; - u32 mbox2_reg; - int ret; + struct drm_device *dev = ring->dev; + struct drm_i915_private *dev_priv = dev->dev_private; + struct intel_ring_buffer *useless; + int i, ret; - ret = intel_ring_begin(ring, 10); + ret = intel_ring_begin(ring, ((I915_NUM_RINGS-1) * + MBOX_UPDATE_DWORDS) + + 4); if (ret) return ret; +#undef MBOX_UPDATE_DWORDS - mbox1_reg = ring->signal_mbox[0]; - mbox2_reg = ring->signal_mbox[1]; + for_each_ring(useless, dev_priv, i) { + u32 mbox_reg = ring->signal_mbox[i]; + if (mbox_reg != GEN6_NOSYNC) + update_mboxes(ring, mbox_reg); + } - update_mboxes(ring, mbox1_reg); - update_mboxes(ring, 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, ring->outstanding_lazy_request); @@ -1674,8 +1686,9 @@ int intel_init_render_ring_buffer(struct drm_device *dev) ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_INVALID; ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_RV; ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_RB; - ring->signal_mbox[0] = GEN6_VRSYNC; - ring->signal_mbox[1] = GEN6_BRSYNC; + ring->signal_mbox[RCS] = GEN6_NOSYNC; + ring->signal_mbox[VCS] = GEN6_VRSYNC; + ring->signal_mbox[BCS] = GEN6_BRSYNC; } else if (IS_GEN5(dev)) { ring->add_request = pc_render_add_request; ring->flush = gen4_render_ring_flush; @@ -1833,8 +1846,9 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev) ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_VR; ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_INVALID; ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_VB; - ring->signal_mbox[0] = GEN6_RVSYNC; - ring->signal_mbox[1] = GEN6_BVSYNC; + ring->signal_mbox[RCS] = GEN6_RVSYNC; + ring->signal_mbox[VCS] = GEN6_NOSYNC; + ring->signal_mbox[BCS] = GEN6_BVSYNC; } else { ring->mmio_base = BSD_RING_BASE; ring->flush = bsd_ring_flush; @@ -1879,8 +1893,9 @@ int intel_init_blt_ring_buffer(struct drm_device *dev) ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_BR; ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_BV; ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_INVALID; - ring->signal_mbox[0] = GEN6_RBSYNC; - ring->signal_mbox[1] = GEN6_VBSYNC; + ring->signal_mbox[RCS] = GEN6_RBSYNC; + ring->signal_mbox[VCS] = GEN6_VBSYNC; + ring->signal_mbox[BCS] = GEN6_NOSYNC; ring->init = init_ring_common; return intel_init_ring_buffer(dev, ring); -- cgit v1.2.3 From 4a3dd19d94c65323d71b2ffc7e63940f00acfb47 Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:19 -0700 Subject: drm/i915: Introduce VECS: the 4th ring The video enhancement command streamer is a new ring on HSW which does what it sounds like it does. This patch provides the most minimal inception of the ring. In order to support a new ring, we need to bump the number. The patch may look trivial to the untrained eye, but bumping the number of rings is a bit scary. As such the patch is not terribly useful by itself, but a pretty nice place to find issues during a bisection. Reviewed-by: Damien Lespiau Signed-off-by: Ben Widawsky Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 2 ++ 1 file changed, 2 insertions(+) (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 5df179127e8b..ead979a94105 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -915,6 +915,8 @@ void intel_ring_setup_status_page(struct intel_ring_buffer *ring) case VCS: mmio = BSD_HWS_PGA_GEN7; break; + case VECS: + BUG(); } } else if (IS_GEN6(ring->dev)) { mmio = RING_HWS_PGA_GEN6(ring->mmio_base); -- cgit v1.2.3 From 1950de14fd1b8ea27a411929156c7eccee2ad7c3 Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:20 -0700 Subject: drm/i915: Add VECS semaphore bits Like the other rings, the VECS supports semaphores. The semaphore stuff is a bit wonky so this patch on it's own should be nice for review. This patch should have no functional impact. v2: Fix the English parts of clarification (again, register names were right, text was reversed) (Damien) Restore the still valid invariant. (Damien) The bsd semaphore register should be MI_SEMAPHORE_SYNC_VVE (Damien) Signed-off-by: Ben Widawsky Reviewed-by: Damien Lespiau Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 6 ++++++ 1 file changed, 6 insertions(+) (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 ead979a94105..93ccd1e982e0 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -1688,9 +1688,11 @@ int intel_init_render_ring_buffer(struct drm_device *dev) ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_INVALID; ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_RV; ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_RB; + ring->semaphore_register[VECS] = MI_SEMAPHORE_SYNC_RVE; ring->signal_mbox[RCS] = GEN6_NOSYNC; ring->signal_mbox[VCS] = GEN6_VRSYNC; ring->signal_mbox[BCS] = GEN6_BRSYNC; + ring->signal_mbox[VECS] = GEN6_VERSYNC; } else if (IS_GEN5(dev)) { ring->add_request = pc_render_add_request; ring->flush = gen4_render_ring_flush; @@ -1848,9 +1850,11 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev) ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_VR; ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_INVALID; ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_VB; + ring->semaphore_register[VECS] = MI_SEMAPHORE_SYNC_VVE; ring->signal_mbox[RCS] = GEN6_RVSYNC; ring->signal_mbox[VCS] = GEN6_NOSYNC; ring->signal_mbox[BCS] = GEN6_BVSYNC; + ring->signal_mbox[VECS] = GEN6_VEVSYNC; } else { ring->mmio_base = BSD_RING_BASE; ring->flush = bsd_ring_flush; @@ -1895,9 +1899,11 @@ int intel_init_blt_ring_buffer(struct drm_device *dev) ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_BR; ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_BV; ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_INVALID; + ring->semaphore_register[VECS] = MI_SEMAPHORE_SYNC_BVE; ring->signal_mbox[RCS] = GEN6_RBSYNC; ring->signal_mbox[VCS] = GEN6_VBSYNC; ring->signal_mbox[BCS] = GEN6_NOSYNC; + ring->signal_mbox[VECS] = GEN6_VEBSYNC; ring->init = init_ring_common; return intel_init_ring_buffer(dev, ring); -- cgit v1.2.3 From ea251324cac6c1e0402db073e5193f33aedd94f3 Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:21 -0700 Subject: drm/i915: Rename ring flush functions Historically we considered the render ring to have special flush semantics and everything else to fall under a more general umbrella. Probably by coincidence more than anything we decided to make the bsd ring have the default *other* flush. As the new vebox ring exposes, the bsd ring is actually the weird one. Doing this allows us to call gen6_ring_flush for the vebox because calling blt_ring_flush would be weird... This patch should have no functional change. Reviewed-by: Damien Lespiau Signed-off-by: Ben Widawsky Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 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 93ccd1e982e0..3022e1579e58 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -1565,8 +1565,8 @@ static void gen6_bsd_ring_write_tail(struct intel_ring_buffer *ring, _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_ring_buffer *ring, + u32 invalidate, u32 flush) { uint32_t cmd; int ret; @@ -1637,8 +1637,8 @@ gen6_ring_dispatch_execbuffer(struct intel_ring_buffer *ring, /* Blitter support (SandyBridge+) */ -static int blt_ring_flush(struct intel_ring_buffer *ring, - u32 invalidate, u32 flush) +static int gen6_ring_flush(struct intel_ring_buffer *ring, + u32 invalidate, u32 flush) { uint32_t cmd; int ret; @@ -1838,7 +1838,7 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev) /* gen6 bsd needs a special wa for tail updates */ if (IS_GEN6(dev)) ring->write_tail = gen6_bsd_ring_write_tail; - ring->flush = gen6_ring_flush; + 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; @@ -1887,7 +1887,7 @@ int intel_init_blt_ring_buffer(struct drm_device *dev) ring->mmio_base = BLT_RING_BASE; ring->write_tail = ring_write_tail; - ring->flush = blt_ring_flush; + ring->flush = gen6_ring_flush; ring->add_request = gen6_add_request; ring->get_seqno = gen6_ring_get_seqno; ring->set_seqno = ring_set_seqno; -- cgit v1.2.3 From 9a8a2213a778509b724c8fda04be70528a1f7130 Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:23 -0700 Subject: drm/i915: Vebox ringbuffer init v2: Add set_seqno which didn't exist before rebase (Haihao) Signed-off-by: Ben Widawsky Reviewed-by: Damien Lespiau Signed-off-by: Xiang, Haihao Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 35 ++++++++++++++++++++++++++++++++- 1 file changed, 34 insertions(+), 1 deletion(-) (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 3022e1579e58..89dfc63677ad 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -916,7 +916,8 @@ void intel_ring_setup_status_page(struct intel_ring_buffer *ring) mmio = BSD_HWS_PGA_GEN7; break; case VECS: - BUG(); + mmio = VEBOX_HWS_PGA_GEN7; + break; } } else if (IS_GEN6(ring->dev)) { mmio = RING_HWS_PGA_GEN6(ring->mmio_base); @@ -1909,6 +1910,38 @@ int intel_init_blt_ring_buffer(struct drm_device *dev) return intel_init_ring_buffer(dev, ring); } +int intel_init_vebox_ring_buffer(struct drm_device *dev) +{ + drm_i915_private_t *dev_priv = dev->dev_private; + struct intel_ring_buffer *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; + ring->irq_enable_mask = 0; + ring->irq_get = NULL; + ring->irq_put = NULL; + ring->dispatch_execbuffer = gen6_ring_dispatch_execbuffer; + ring->sync_to = gen6_ring_sync; + ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_VER; + ring->semaphore_register[VCS] = MI_SEMAPHORE_SYNC_VEV; + ring->semaphore_register[BCS] = MI_SEMAPHORE_SYNC_VEB; + ring->semaphore_register[VECS] = MI_SEMAPHORE_SYNC_INVALID; + ring->signal_mbox[RCS] = GEN6_RVESYNC; + ring->signal_mbox[VCS] = GEN6_VVESYNC; + ring->signal_mbox[BCS] = GEN6_BVESYNC; + ring->signal_mbox[VECS] = GEN6_NOSYNC; + ring->init = init_ring_common; + + return intel_init_ring_buffer(dev, ring); +} + int intel_ring_flush_all_caches(struct intel_ring_buffer *ring) { -- cgit v1.2.3 From aeb0659338793746b8a4e482fa588ba1dd9ee559 Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:28 -0700 Subject: drm/i915: Convert irq_refounct to struct It's overkill on older gens, but it's useful for newer gens. Reviewed-by: Damien Lespiau Signed-off-by: Ben Widawsky Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 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 89dfc63677ad..c7a89bb051a0 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -795,7 +795,7 @@ gen5_ring_get_irq(struct intel_ring_buffer *ring) return false; spin_lock_irqsave(&dev_priv->irq_lock, flags); - if (ring->irq_refcount++ == 0) { + if (ring->irq_refcount.gt++ == 0) { dev_priv->gt_irq_mask &= ~ring->irq_enable_mask; I915_WRITE(GTIMR, dev_priv->gt_irq_mask); POSTING_READ(GTIMR); @@ -813,7 +813,7 @@ gen5_ring_put_irq(struct intel_ring_buffer *ring) unsigned long flags; spin_lock_irqsave(&dev_priv->irq_lock, flags); - if (--ring->irq_refcount == 0) { + if (--ring->irq_refcount.gt == 0) { dev_priv->gt_irq_mask |= ring->irq_enable_mask; I915_WRITE(GTIMR, dev_priv->gt_irq_mask); POSTING_READ(GTIMR); @@ -832,7 +832,7 @@ i9xx_ring_get_irq(struct intel_ring_buffer *ring) return false; spin_lock_irqsave(&dev_priv->irq_lock, flags); - if (ring->irq_refcount++ == 0) { + if (ring->irq_refcount.gt++ == 0) { dev_priv->irq_mask &= ~ring->irq_enable_mask; I915_WRITE(IMR, dev_priv->irq_mask); POSTING_READ(IMR); @@ -850,7 +850,7 @@ i9xx_ring_put_irq(struct intel_ring_buffer *ring) unsigned long flags; spin_lock_irqsave(&dev_priv->irq_lock, flags); - if (--ring->irq_refcount == 0) { + if (--ring->irq_refcount.gt == 0) { dev_priv->irq_mask |= ring->irq_enable_mask; I915_WRITE(IMR, dev_priv->irq_mask); POSTING_READ(IMR); @@ -869,7 +869,7 @@ i8xx_ring_get_irq(struct intel_ring_buffer *ring) return false; spin_lock_irqsave(&dev_priv->irq_lock, flags); - if (ring->irq_refcount++ == 0) { + if (ring->irq_refcount.gt++ == 0) { dev_priv->irq_mask &= ~ring->irq_enable_mask; I915_WRITE16(IMR, dev_priv->irq_mask); POSTING_READ16(IMR); @@ -887,7 +887,7 @@ i8xx_ring_put_irq(struct intel_ring_buffer *ring) unsigned long flags; spin_lock_irqsave(&dev_priv->irq_lock, flags); - if (--ring->irq_refcount == 0) { + if (--ring->irq_refcount.gt == 0) { dev_priv->irq_mask |= ring->irq_enable_mask; I915_WRITE16(IMR, dev_priv->irq_mask); POSTING_READ16(IMR); @@ -980,7 +980,7 @@ gen6_ring_get_irq(struct intel_ring_buffer *ring) gen6_gt_force_wake_get(dev_priv); spin_lock_irqsave(&dev_priv->irq_lock, flags); - if (ring->irq_refcount++ == 0) { + if (ring->irq_refcount.gt++ == 0) { if (HAS_L3_GPU_CACHE(dev) && ring->id == RCS) I915_WRITE_IMR(ring, ~(ring->irq_enable_mask | GEN6_RENDER_L3_PARITY_ERROR)); @@ -1003,7 +1003,7 @@ gen6_ring_put_irq(struct intel_ring_buffer *ring) unsigned long flags; spin_lock_irqsave(&dev_priv->irq_lock, flags); - if (--ring->irq_refcount == 0) { + if (--ring->irq_refcount.gt == 0) { if (HAS_L3_GPU_CACHE(dev) && ring->id == RCS) I915_WRITE_IMR(ring, ~GEN6_RENDER_L3_PARITY_ERROR); else -- cgit v1.2.3 From cc609d5da5c78c92a2e2565604b2603a0965b494 Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:29 -0700 Subject: drm/i915: consolidate interrupt naming scheme The motivation here is we're going to add some new interrupt definitions and handling outside of the GT interrupts which is all we've managed so far (with some RPS exceptions). By consolidating the names in the future we can make thing a bit cleaner as we don't need to define register names twice, and we can leverage pretty decent overlap in HW registers since ILK. To explain briefly what is in the comments: there are two sets of interrupt masking/enabling registers. At least so far, the definitions of the two sets overlap. The old code setup distinct names for interrupts in each set, ie. one for global, and one for ring. This made things confusing when using the wrong defines in the wrong places. rebase: Modified VLV bits v2: Renamed GT_RENDER_MASTER to GT_RENDER_CS_MASTER (Damien) Reviewed-by: Damien Lespiau Signed-off-by: Ben Widawsky Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 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 c7a89bb051a0..5ab8cc2cafe2 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -560,7 +560,7 @@ static int init_render_ring(struct intel_ring_buffer *ring) I915_WRITE(INSTPM, _MASKED_BIT_ENABLE(INSTPM_FORCE_ORDERING)); if (HAS_L3_GPU_CACHE(dev)) - I915_WRITE_IMR(ring, ~GEN6_RENDER_L3_PARITY_ERROR); + I915_WRITE_IMR(ring, ~GT_RENDER_L3_PARITY_ERROR_INTERRUPT); return ret; } @@ -982,8 +982,9 @@ gen6_ring_get_irq(struct intel_ring_buffer *ring) spin_lock_irqsave(&dev_priv->irq_lock, flags); if (ring->irq_refcount.gt++ == 0) { if (HAS_L3_GPU_CACHE(dev) && ring->id == RCS) - I915_WRITE_IMR(ring, ~(ring->irq_enable_mask | - GEN6_RENDER_L3_PARITY_ERROR)); + I915_WRITE_IMR(ring, + ~(ring->irq_enable_mask | + GT_RENDER_L3_PARITY_ERROR_INTERRUPT)); else I915_WRITE_IMR(ring, ~ring->irq_enable_mask); dev_priv->gt_irq_mask &= ~ring->irq_enable_mask; @@ -1005,7 +1006,8 @@ gen6_ring_put_irq(struct intel_ring_buffer *ring) spin_lock_irqsave(&dev_priv->irq_lock, flags); if (--ring->irq_refcount.gt == 0) { if (HAS_L3_GPU_CACHE(dev) && ring->id == RCS) - I915_WRITE_IMR(ring, ~GEN6_RENDER_L3_PARITY_ERROR); + I915_WRITE_IMR(ring, + ~GT_RENDER_L3_PARITY_ERROR_INTERRUPT); else I915_WRITE_IMR(ring, ~0); dev_priv->gt_irq_mask |= ring->irq_enable_mask; @@ -1682,7 +1684,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev) ring->flush = gen6_render_ring_flush; ring->irq_get = gen6_ring_get_irq; ring->irq_put = gen6_ring_put_irq; - ring->irq_enable_mask = GT_USER_INTERRUPT; + ring->irq_enable_mask = GT_RENDER_USER_INTERRUPT; ring->get_seqno = gen6_ring_get_seqno; ring->set_seqno = ring_set_seqno; ring->sync_to = gen6_ring_sync; @@ -1701,7 +1703,8 @@ int intel_init_render_ring_buffer(struct drm_device *dev) 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_USER_INTERRUPT | GT_PIPE_NOTIFY; + 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) @@ -1843,7 +1846,7 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev) ring->add_request = gen6_add_request; ring->get_seqno = gen6_ring_get_seqno; ring->set_seqno = ring_set_seqno; - ring->irq_enable_mask = GEN6_BSD_USER_INTERRUPT; + 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; @@ -1863,7 +1866,7 @@ int intel_init_bsd_ring_buffer(struct drm_device *dev) ring->get_seqno = ring_get_seqno; ring->set_seqno = ring_set_seqno; if (IS_GEN5(dev)) { - ring->irq_enable_mask = GT_BSD_USER_INTERRUPT; + ring->irq_enable_mask = ILK_BSD_USER_INTERRUPT; ring->irq_get = gen5_ring_get_irq; ring->irq_put = gen5_ring_put_irq; } else { @@ -1892,7 +1895,7 @@ int intel_init_blt_ring_buffer(struct drm_device *dev) ring->add_request = gen6_add_request; ring->get_seqno = gen6_ring_get_seqno; ring->set_seqno = ring_set_seqno; - ring->irq_enable_mask = GEN6_BLITTER_USER_INTERRUPT; + 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; -- cgit v1.2.3 From a19d2933cbc4c7b8e3d72e9fd2d48847c25bb41d Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:30 -0700 Subject: drm/i915: vebox interrupt get/put v2: Use the correct lock to protect PM interrupt regs, this was accidentally lost from earlier (Haihao) Fix return types (Ben) Reviewed-by: Damien Lespiau Signed-off-by: Ben Widawsky Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 46 +++++++++++++++++++++++++++++++-- 1 file changed, 44 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 5ab8cc2cafe2..8a6a0ee6d350 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -1019,6 +1019,48 @@ gen6_ring_put_irq(struct intel_ring_buffer *ring) gen6_gt_force_wake_put(dev_priv); } +static bool +hsw_vebox_get_irq(struct intel_ring_buffer *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->rps.lock, flags); + if (ring->irq_refcount.pm++ == 0) { + u32 pm_imr = I915_READ(GEN6_PMIMR); + I915_WRITE_IMR(ring, ~ring->irq_enable_mask); + I915_WRITE(GEN6_PMIMR, pm_imr & ~ring->irq_enable_mask); + POSTING_READ(GEN6_PMIMR); + } + spin_unlock_irqrestore(&dev_priv->rps.lock, flags); + + return true; +} + +static void +hsw_vebox_put_irq(struct intel_ring_buffer *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->rps.lock, flags); + if (--ring->irq_refcount.pm == 0) { + u32 pm_imr = I915_READ(GEN6_PMIMR); + I915_WRITE_IMR(ring, ~0); + I915_WRITE(GEN6_PMIMR, pm_imr | ring->irq_enable_mask); + POSTING_READ(GEN6_PMIMR); + } + spin_unlock_irqrestore(&dev_priv->rps.lock, flags); +} + static int i965_dispatch_execbuffer(struct intel_ring_buffer *ring, u32 offset, u32 length, @@ -1928,8 +1970,8 @@ int intel_init_vebox_ring_buffer(struct drm_device *dev) ring->get_seqno = gen6_ring_get_seqno; ring->set_seqno = ring_set_seqno; ring->irq_enable_mask = 0; - ring->irq_get = NULL; - ring->irq_put = NULL; + ring->irq_get = hsw_vebox_get_irq; + ring->irq_put = hsw_vebox_put_irq; ring->dispatch_execbuffer = gen6_ring_dispatch_execbuffer; ring->sync_to = gen6_ring_sync; ring->semaphore_register[RCS] = MI_SEMAPHORE_SYNC_VER; -- cgit v1.2.3 From 12638c57f31952127c734c26315e1348fa1334c2 Mon Sep 17 00:00:00 2001 From: Ben Widawsky Date: Tue, 28 May 2013 19:22:31 -0700 Subject: drm/i915: Enable vebox interrupts Similar to a patch originally written by: v2: Reversed the meanings of masked and enabled (Haihao) Made non-destructive writes in case enable/disabler rps runs first (Haihao) v3: Reword error message (Damien) Modify postinstall to do the right thing based on previous fixup. (Ben) CC: Xiang, Haihao Reviewed-by: Damien Lespiau Signed-off-by: Ben Widawsky Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (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 8a6a0ee6d350..0e72da6ad0fa 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -1969,7 +1969,8 @@ int intel_init_vebox_ring_buffer(struct drm_device *dev) ring->add_request = gen6_add_request; ring->get_seqno = gen6_ring_get_seqno; ring->set_seqno = ring_set_seqno; - ring->irq_enable_mask = 0; + ring->irq_enable_mask = PM_VEBOX_USER_INTERRUPT | + PM_VEBOX_CS_ERROR_INTERRUPT; ring->irq_get = hsw_vebox_get_irq; ring->irq_put = hsw_vebox_put_irq; ring->dispatch_execbuffer = gen6_ring_dispatch_execbuffer; -- cgit v1.2.3 From fd3da6c95b6d865446fa9b29df6edff4343e385a Mon Sep 17 00:00:00 2001 From: Rodrigo Vivi Date: Thu, 6 Jun 2013 16:58:16 -0300 Subject: drm/i915: WA: FBC Render Nuke. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit WaFbcNukeOn3DBlt for IVB, HSW. According BSPec: "Workaround: Do not enable Render Command Streamer tracking for FBC. Instead insert a LRI to address 0x50380 with data 0x00000004 after the PIPE_CONTROL that follows each render submission." v2: Chris noticed that flush_domains check was missing here and also suggested to do LRI only when fbc is enabled. To avoid do a I915_READ on every flush lets use the module parameter check. v3: Adding Wa name as Damien suggested. v4: Ville noticed VLV doesn't support fbc at all and comment came wrong from spec. v5: Ville noticed than on blt a Cache Clean LRI should be used instead the Nuke one. v6: Check for flush domain on blt (by Ville). Check for scanout dirty (by Chris). v7: Apply proper fbc_dirty implemented by Chris. v8: remove unused variables. Cc: Ville Syrjälä Cc: Chris Wilson Signed-off-by: Rodrigo Vivi Reviewed-by: Chris Wilson Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 29 +++++++++++++++++++++++++++++ 1 file changed, 29 insertions(+) (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 0e72da6ad0fa..1ef081cd1706 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -280,6 +280,27 @@ gen7_render_ring_cs_stall_wa(struct intel_ring_buffer *ring) return 0; } +static int gen7_ring_fbc_flush(struct intel_ring_buffer *ring, u32 value) +{ + int ret; + + if (!ring->fbc_dirty) + return 0; + + ret = intel_ring_begin(ring, 4); + if (ret) + return ret; + intel_ring_emit(ring, MI_NOOP); + /* 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_advance(ring); + + ring->fbc_dirty = false; + return 0; +} + static int gen7_render_ring_flush(struct intel_ring_buffer *ring, u32 invalidate_domains, u32 flush_domains) @@ -336,6 +357,9 @@ gen7_render_ring_flush(struct intel_ring_buffer *ring, intel_ring_emit(ring, 0); intel_ring_advance(ring); + if (flush_domains) + return gen7_ring_fbc_flush(ring, FBC_REND_NUKE); + return 0; } @@ -1685,6 +1709,7 @@ gen6_ring_dispatch_execbuffer(struct intel_ring_buffer *ring, static int gen6_ring_flush(struct intel_ring_buffer *ring, u32 invalidate, u32 flush) { + struct drm_device *dev = ring->dev; uint32_t cmd; int ret; @@ -1707,6 +1732,10 @@ static int gen6_ring_flush(struct intel_ring_buffer *ring, intel_ring_emit(ring, 0); intel_ring_emit(ring, MI_NOOP); intel_ring_advance(ring); + + if (IS_GEN7(dev) && flush) + return gen7_ring_fbc_flush(ring, FBC_REND_CACHE_CLEAN); + return 0; } -- cgit v1.2.3 From 50f018dff180942dc40e601de6e97145a4aaeaa9 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Mon, 10 Jun 2013 11:20:19 +0100 Subject: drm/i915: Initialize ring->hangcheck upon ring init When we reset and restart a ring, we also want to clear any existing hangcheck. Signed-off-by: Chris Wilson Reviewed-by: Mika Kuoppala Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 2 ++ 1 file changed, 2 insertions(+) (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 1ef081cd1706..a3cfa35b0573 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -453,6 +453,8 @@ static int init_ring_common(struct intel_ring_buffer *ring) ring->last_retired_head = -1; } + memset(&ring->hangcheck, 0, sizeof(ring->hangcheck)); + out: if (HAS_FORCE_WAKE(dev)) gen6_gt_force_wake_put(dev_priv); -- cgit v1.2.3 From 0025c0772de7451c2302fa628f038b213a0783bf Mon Sep 17 00:00:00 2001 From: Mika Kuoppala Date: Wed, 12 Jun 2013 12:35:30 +0300 Subject: drm/i915: change i915_add_request to macro Only execbuffer needed all the parameters on i915_add_request(). By putting __i915_add_request behind macro, all current callsites become cleaner. Following patch will introduce a new parameter for __i915_add_request. With this patch, only the relevant callsite will reflect the change making commit smaller and easier to understand. v2: _i915_add_request as function name (Chris Wilson) v3: change name __i915_add_request and fix ordering of params (Ben Widawsky) Signed-off-by: Mika Kuoppala Reviewed-by: Chris Wilson Acked-by: Ben Widawsky Signed-off-by: Daniel Vetter --- drivers/gpu/drm/i915/intel_ringbuffer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (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 a3cfa35b0573..e51ab552046c 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -1512,7 +1512,7 @@ int intel_ring_idle(struct intel_ring_buffer *ring) /* We need to add any requests required to flush the objects and ring */ if (ring->outstanding_lazy_request) { - ret = i915_add_request(ring, NULL, NULL); + ret = i915_add_request(ring, NULL); if (ret) return ret; } -- cgit v1.2.3