Skip to contentSkip to conversation

lrc lifecycle cleanups

PATCH In review Nick Hoath 4 participants 13 messages Oct 20, 2015

Subsystems (2)
lore
Share & tools
lore

Nick Hoath

author
#0 lore raw

cover lrc lifecycle cleanups

These changes are a result of the requests made in VIZ-4277. Make the lrc path more like the legacy submission path. Attach the CPU mappings to vma (un)bind, so that the shrinker also cleans those up.
Pin the CPU mappings while context is busy (pending requests), so that the mappings aren't released/made continuously as this is an expensive process.

V2: Removed unecessary changes in the lrc retiring code path
Removed unecessary map/unmap

Nick Hoath (6):
drm/i195: Rename gt_irq_handler variable
drm/i915: Break out common code from gen8_gt_irq_handler drm/i915: Unify execlist and legacy request life-cycles drm/i915: Improve dynamic management/eviction of lrc backing objects
drm/i915: Add the CPU mapping of the hw context to the pinned
items.
drm/i915: Only update ringbuf address when necessary

 drivers/gpu/drm/i915/i915_debugfs.c     |  14 ++--
 drivers/gpu/drm/i915/i915_drv.h         |  10 ++-
 drivers/gpu/drm/i915/i915_gem.c         |  26 +++----
 drivers/gpu/drm/i915/i915_gem_gtt.c     |   8 +++
 drivers/gpu/drm/i915/i915_irq.c         |  67 ++++++++---------
 drivers/gpu/drm/i915/intel_lrc.c        | 123 +++++++++++++++++---------------
 drivers/gpu/drm/i915/intel_lrc.h        |   2 +-
 drivers/gpu/drm/i915/intel_ringbuffer.c |  71 +++++++++---------
 drivers/gpu/drm/i915/intel_ringbuffer.h |   4 --
 9 files changed, 173 insertions(+), 152 deletions(-)
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#1 lore raw

PATCH 1/6 drm/i195: Rename gt_irq_handler variable

Renamed tmp variable to the more descriptive iir. (Daniel Vetter/ Thomas Daniel)

Issue: VIZ-4277

Cc Daniel Vetter <[email protected]>
Cc David Gordon <[email protected]>
Cc Thomas Daniel <[email protected]>
---
 drivers/gpu/drm/i915/i915_irq.c | 46 ++++++++++++++++++++---------------------
 1 file changed, 23 insertions(+), 23 deletions(-)
drivers/gpu/drm/i915/i915_irq.c +23 -23
index d68328f..fbf9153 100644
@@ -1296,64 +1296,64 @@ static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 	irqreturn_t ret = IRQ_NONE;
 
 	if (master_ctl & (GEN8_GT_RCS_IRQ | GEN8_GT_BCS_IRQ)) {
-		u32 tmp = I915_READ_FW(GEN8_GT_IIR(0));
-		if (tmp) {
-			I915_WRITE_FW(GEN8_GT_IIR(0), tmp);
+		u32 iir = I915_READ_FW(GEN8_GT_IIR(0));
+		if (iir) {
+			I915_WRITE_FW(GEN8_GT_IIR(0), iir);
 			ret = IRQ_HANDLED;
 
-			if (tmp & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_RCS_IRQ_SHIFT))
+			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_RCS_IRQ_SHIFT))
 				intel_lrc_irq_handler(&dev_priv->ring[RCS]);
-			if (tmp & (GT_RENDER_USER_INTERRUPT << GEN8_RCS_IRQ_SHIFT))
+			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_RCS_IRQ_SHIFT))
 				notify_ring(&dev_priv->ring[RCS]);
 
-			if (tmp & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_BCS_IRQ_SHIFT))
+			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_BCS_IRQ_SHIFT))
 				intel_lrc_irq_handler(&dev_priv->ring[BCS]);
-			if (tmp & (GT_RENDER_USER_INTERRUPT << GEN8_BCS_IRQ_SHIFT))
+			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_BCS_IRQ_SHIFT))
 				notify_ring(&dev_priv->ring[BCS]);
 		} else
 			DRM_ERROR("The master control interrupt lied (GT0)!\n");
 	}
 
 	if (master_ctl & (GEN8_GT_VCS1_IRQ | GEN8_GT_VCS2_IRQ)) {
-		u32 tmp = I915_READ_FW(GEN8_GT_IIR(1));
-		if (tmp) {
-			I915_WRITE_FW(GEN8_GT_IIR(1), tmp);
+		u32 iir = I915_READ_FW(GEN8_GT_IIR(1));
+		if (iir) {
+			I915_WRITE_FW(GEN8_GT_IIR(1), iir);
 			ret = IRQ_HANDLED;
 
-			if (tmp & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VCS1_IRQ_SHIFT))
+			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VCS1_IRQ_SHIFT))
 				intel_lrc_irq_handler(&dev_priv->ring[VCS]);
-			if (tmp & (GT_RENDER_USER_INTERRUPT << GEN8_VCS1_IRQ_SHIFT))
+			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_VCS1_IRQ_SHIFT))
 				notify_ring(&dev_priv->ring[VCS]);
 
-			if (tmp & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VCS2_IRQ_SHIFT))
+			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VCS2_IRQ_SHIFT))
 				intel_lrc_irq_handler(&dev_priv->ring[VCS2]);
-			if (tmp & (GT_RENDER_USER_INTERRUPT << GEN8_VCS2_IRQ_SHIFT))
+			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_VCS2_IRQ_SHIFT))
 				notify_ring(&dev_priv->ring[VCS2]);
 		} else
 			DRM_ERROR("The master control interrupt lied (GT1)!\n");
 	}
 
 	if (master_ctl & GEN8_GT_VECS_IRQ) {
-		u32 tmp = I915_READ_FW(GEN8_GT_IIR(3));
-		if (tmp) {
-			I915_WRITE_FW(GEN8_GT_IIR(3), tmp);
+		u32 iir = I915_READ_FW(GEN8_GT_IIR(3));
+		if (iir) {
+			I915_WRITE_FW(GEN8_GT_IIR(3), iir);
 			ret = IRQ_HANDLED;
 
-			if (tmp & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VECS_IRQ_SHIFT))
+			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VECS_IRQ_SHIFT))
 				intel_lrc_irq_handler(&dev_priv->ring[VECS]);
-			if (tmp & (GT_RENDER_USER_INTERRUPT << GEN8_VECS_IRQ_SHIFT))
+			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_VECS_IRQ_SHIFT))
 				notify_ring(&dev_priv->ring[VECS]);
 		} else
 			DRM_ERROR("The master control interrupt lied (GT3)!\n");
 	}
 
 	if (master_ctl & GEN8_GT_PM_IRQ) {
-		u32 tmp = I915_READ_FW(GEN8_GT_IIR(2));
-		if (tmp & dev_priv->pm_rps_events) {
+		u32 iir = I915_READ_FW(GEN8_GT_IIR(2));
+		if (iir & dev_priv->pm_rps_events) {
 			I915_WRITE_FW(GEN8_GT_IIR(2),
-				      tmp & dev_priv->pm_rps_events);
+				      iir & dev_priv->pm_rps_events);
 			ret = IRQ_HANDLED;
-			gen6_rps_irq_handler(dev_priv, tmp);
+			gen6_rps_irq_handler(dev_priv, iir);
 		} else
 			DRM_ERROR("The master control interrupt lied (PM)!\n");
 	}
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#2 lore raw

PATCH 2/6 drm/i915: Break out common code from gen8_gt_irq_handler

In reply to Nick Hoath

Break out common code from gen8_gt_irq_handler and put it in to an always inlined function. gcc optimises out the shift at compile time. (Thomas Daniel/Daniel Vetter/Chris Wilson)

Issue: VIZ-4277

Cc Thomas Daniel <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Chris Wilson <[email protected]>
---
 drivers/gpu/drm/i915/i915_irq.c | 40 ++++++++++++++++++++--------------------
 1 file changed, 20 insertions(+), 20 deletions(-)
drivers/gpu/drm/i915/i915_irq.c +20 -20
index fbf9153..7837f5e 100644
@@ -1290,6 +1290,16 @@ static void snb_gt_irq_handler(struct drm_device *dev,
 		ivybridge_parity_error_irq_handler(dev, gt_iir);
 }
 
+static __always_inline void
+		gen8_cs_irq_handler(struct intel_engine_cs *ring, u32 iir,
+				    int test_shift)
+{
+	if (iir & (GT_RENDER_USER_INTERRUPT << test_shift))
+		notify_ring(ring);
+	if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << test_shift))
+		intel_lrc_irq_handler(ring);
+}
+
 static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 				       u32 master_ctl)
 {
@@ -1301,15 +1311,11 @@ static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 			I915_WRITE_FW(GEN8_GT_IIR(0), iir);
 			ret = IRQ_HANDLED;
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_RCS_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[RCS]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_RCS_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[RCS]);
+			gen8_cs_irq_handler(&dev_priv->ring[RCS],
+					iir, GEN8_RCS_IRQ_SHIFT);
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_BCS_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[BCS]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_BCS_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[BCS]);
+			gen8_cs_irq_handler(&dev_priv->ring[BCS],
+					iir, GEN8_BCS_IRQ_SHIFT);
 		} else
 			DRM_ERROR("The master control interrupt lied (GT0)!\n");
 	}
@@ -1320,15 +1326,11 @@ static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 			I915_WRITE_FW(GEN8_GT_IIR(1), iir);
 			ret = IRQ_HANDLED;
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VCS1_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[VCS]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_VCS1_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[VCS]);
+			gen8_cs_irq_handler(&dev_priv->ring[VCS],
+					iir, GEN8_VCS1_IRQ_SHIFT);
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VCS2_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[VCS2]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_VCS2_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[VCS2]);
+			gen8_cs_irq_handler(&dev_priv->ring[VCS2],
+					iir, GEN8_VCS2_IRQ_SHIFT);
 		} else
 			DRM_ERROR("The master control interrupt lied (GT1)!\n");
 	}
@@ -1339,10 +1341,8 @@ static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 			I915_WRITE_FW(GEN8_GT_IIR(3), iir);
 			ret = IRQ_HANDLED;
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VECS_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[VECS]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_VECS_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[VECS]);
+			gen8_cs_irq_handler(&dev_priv->ring[VECS],
+					iir, GEN8_VECS_IRQ_SHIFT);
 		} else
 			DRM_ERROR("The master control interrupt lied (GT3)!\n");
 	}
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw

Re: [PATCH 2/6] drm/i915: Break out common code from gen8_gt_irq_handler

Quoted 9 lines · Nick Hoath

Break out common code from gen8_gt_irq_handler and put it in to an always inlined function. gcc optimises out the shift at compile time. (Thomas Daniel/Daniel Vetter/Chris Wilson)

Issue: VIZ-4277

Cc Thomas Daniel <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Chris Wilson <[email protected]>

Merged the first two patches to dinq, thanks.
-Daniel

Quoted 81 lines
---
 drivers/gpu/drm/i915/i915_irq.c | 40 ++++++++++++++++++++--------------------
 1 file changed, 20 insertions(+), 20 deletions(-)
drivers/gpu/drm/i915/i915_irq.c +20 -20
index fbf9153..7837f5e 100644
@@ -1290,6 +1290,16 @@ static void snb_gt_irq_handler(struct drm_device *dev,
 		ivybridge_parity_error_irq_handler(dev, gt_iir);
 }
 
+static __always_inline void
+		gen8_cs_irq_handler(struct intel_engine_cs *ring, u32 iir,
+				    int test_shift)
+{
+	if (iir & (GT_RENDER_USER_INTERRUPT << test_shift))
+		notify_ring(ring);
+	if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << test_shift))
+		intel_lrc_irq_handler(ring);
+}
+
 static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 				       u32 master_ctl)
 {
@@ -1301,15 +1311,11 @@ static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 			I915_WRITE_FW(GEN8_GT_IIR(0), iir);
 			ret = IRQ_HANDLED;
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_RCS_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[RCS]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_RCS_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[RCS]);
+			gen8_cs_irq_handler(&dev_priv->ring[RCS],
+					iir, GEN8_RCS_IRQ_SHIFT);
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_BCS_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[BCS]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_BCS_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[BCS]);
+			gen8_cs_irq_handler(&dev_priv->ring[BCS],
+					iir, GEN8_BCS_IRQ_SHIFT);
 		} else
 			DRM_ERROR("The master control interrupt lied (GT0)!\n");
 	}
@@ -1320,15 +1326,11 @@ static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 			I915_WRITE_FW(GEN8_GT_IIR(1), iir);
 			ret = IRQ_HANDLED;
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VCS1_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[VCS]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_VCS1_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[VCS]);
+			gen8_cs_irq_handler(&dev_priv->ring[VCS],
+					iir, GEN8_VCS1_IRQ_SHIFT);
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VCS2_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[VCS2]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_VCS2_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[VCS2]);
+			gen8_cs_irq_handler(&dev_priv->ring[VCS2],
+					iir, GEN8_VCS2_IRQ_SHIFT);
 		} else
 			DRM_ERROR("The master control interrupt lied (GT1)!\n");
 	}
@@ -1339,10 +1341,8 @@ static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
 			I915_WRITE_FW(GEN8_GT_IIR(3), iir);
 			ret = IRQ_HANDLED;
 
-			if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << GEN8_VECS_IRQ_SHIFT))
-				intel_lrc_irq_handler(&dev_priv->ring[VECS]);
-			if (iir & (GT_RENDER_USER_INTERRUPT << GEN8_VECS_IRQ_SHIFT))
-				notify_ring(&dev_priv->ring[VECS]);
+			gen8_cs_irq_handler(&dev_priv->ring[VECS],
+					iir, GEN8_VECS_IRQ_SHIFT);
 		} else
 			DRM_ERROR("The master control interrupt lied (GT3)!\n");
 	}
Signature · 1 line
1.9.1
Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#4 lore raw

Re: [PATCH 2/6] drm/i915: Break out common code from gen8_gt_irq_handler

Quoted 26 lines · Daniel Vetter
Quoted 9 lines · Nick Hoath

Break out common code from gen8_gt_irq_handler and put it in to an always inlined function. gcc optimises out the shift at compile time. (Thomas Daniel/Daniel Vetter/Chris Wilson)

Issue: VIZ-4277

Cc Thomas Daniel <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Chris Wilson <[email protected]>

Merged the first two patches to dinq, thanks.
-Daniel

Quoted 12 lines
---
 drivers/gpu/drm/i915/i915_irq.c | 40 ++++++++++++++++++++--------------------
 1 file changed, 20 insertions(+), 20 deletions(-)
drivers/gpu/drm/i915/i915_irq.c +0 -0
index fbf9153..7837f5e 100644
@@ -1290,6 +1290,16 @@ static void snb_gt_irq_handler(struct drm_device *dev,
 		ivybridge_parity_error_irq_handler(dev, gt_iir);
 }
 
Quoted 3 lines

drivers/gpu/drm/i915/i915_irq.c static void snb_gt_irq_handler(struct drm_device *dev, · line 1295

Quoted 3 lines
drivers/gpu/drm/i915/i915_irq.c +3 -0
+static __always_inline void
+		gen8_cs_irq_handler(struct intel_engine_cs *ring, u32 iir,
+				    int test_shift)

Pray tell you fixed the formatting first?
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#5 lore raw

PATCH 3/6 drm/i915: Unify execlist and legacy request life-cycles

In reply to Nick Hoath

There is a desire to simplify the i915 driver by reducing the number of different code paths introduced by the LRC / execlists support. As the execlists request is now part of the gem request it is possible and desirable to unify the request life-cycles for execlist and legacy requests.

A request is considered retireable if its seqno passed (i.e. the request has completed) and either it was never submitted to the ELSP or its context completed. This ensures that context save is carried out before the last request for a context is considered retireable. request_complete() now checks the elsp_submitted count when deciding if a request is complete. Requests that were not waiting for a context
switch interrupt (either as a result of being merged into a following request or by being a legacy request) will be considered retireable as soon as their seqno has passed.

Removed the extra request reference held for the execlist request.

Removed intel_execlists_retire_requests() and all references to intel_engine_cs.execlist_retired_req_list.

Changed gen8_cs_irq_handler() so that notify_ring() is called when contexts complete as well as when a user interrupt occurs so that notification happens when a request is complete and context save has finished.

v2: Rebase over the read-read optimisation changes

v3: Reworked IRQ handler after removing IRQ handler cleanup patch

v4: Fixed various pin leaks

v5: Removed ctx_complete flag & associated changes. Removed extraneous
request pin of context.
(Chris Wilson/Daniel Vetter)

Issue: VIZ-4277

---
 drivers/gpu/drm/i915/i915_drv.h         |  2 +-
 drivers/gpu/drm/i915/i915_gem.c         | 23 ++++++++---------
 drivers/gpu/drm/i915/i915_irq.c         |  7 ++---
 drivers/gpu/drm/i915/intel_lrc.c        | 45 ++++-----------------------------
 drivers/gpu/drm/i915/intel_lrc.h        |  2 +-
 drivers/gpu/drm/i915/intel_ringbuffer.h |  1 -
 6 files changed, 21 insertions(+), 59 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +1 -1
index 8afda45..ae08e57 100644
@@ -2914,7 +2914,7 @@ static inline bool i915_gem_request_completed(struct drm_i915_gem_request *req,
 
 	seqno = req->ring->get_seqno(req->ring, lazy_coherency);
 
-	return i915_seqno_passed(seqno, req->seqno);
+	return i915_seqno_passed(seqno, req->seqno) && !req->elsp_submitted;
 }
 
 int __must_check i915_gem_get_seqno(struct drm_device *dev, u32 *seqno);
drivers/gpu/drm/i915/i915_gem.c +10 -13
index e57061a..290a1ac 100644
@@ -2848,12 +2848,16 @@ i915_gem_retire_requests_ring(struct intel_engine_cs *ring)
 
 		if (!list_empty(&obj->last_read_req[ring->id]->list))
 			break;
+		if (!i915_gem_request_completed(obj->last_read_req[ring->id],
+				true))
+			break;
 
 		i915_gem_object_retire__read(obj, ring->id);
 	}
 
 	if (unlikely(ring->trace_irq_req &&
-		     i915_gem_request_completed(ring->trace_irq_req, true))) {
+		     i915_gem_request_completed(ring->trace_irq_req,
+				true))) {
 		ring->irq_put(ring);
 		i915_gem_request_assign(&ring->trace_irq_req, NULL);
 	}
@@ -2872,15 +2876,6 @@ i915_gem_retire_requests(struct drm_device *dev)
 	for_each_ring(ring, dev_priv, i) {
 		i915_gem_retire_requests_ring(ring);
 		idle &= list_empty(&ring->request_list);
-		if (i915.enable_execlists) {
-			unsigned long flags;
-
-			spin_lock_irqsave(&ring->execlist_lock, flags);
-			idle &= list_empty(&ring->execlist_queue);
-			spin_unlock_irqrestore(&ring->execlist_lock, flags);
-
-			intel_execlists_retire_requests(ring);
-		}
 	}
 
 	if (idle)
@@ -2956,12 +2951,14 @@ i915_gem_object_flush_active(struct drm_i915_gem_object *obj)
 		if (req == NULL)
 			continue;
 
-		if (list_empty(&req->list))
-			goto retire;
+		if (list_empty(&req->list)) {
+			if (i915_gem_request_completed(req, true))
+				i915_gem_object_retire__read(obj, i);
+			continue;
+		}
 
 		if (i915_gem_request_completed(req, true)) {
 			__i915_gem_request_retire__upto(req);
-retire:
 			i915_gem_object_retire__read(obj, i);
 		}
 	}
drivers/gpu/drm/i915/i915_irq.c +4 -3
index 7837f5e..86a6662 100644
@@ -1294,10 +1294,11 @@ static __always_inline void
 		gen8_cs_irq_handler(struct intel_engine_cs *ring, u32 iir,
 				    int test_shift)
 {
-	if (iir & (GT_RENDER_USER_INTERRUPT << test_shift))
-		notify_ring(ring);
+	bool need_notify = false;
 	if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << test_shift))
-		intel_lrc_irq_handler(ring);
+		need_notify = intel_lrc_irq_handler(ring);
+	if ((iir & (GT_RENDER_USER_INTERRUPT << test_shift)) || need_notify)
+		notify_ring(ring);
 }
 
 static irqreturn_t gen8_gt_irq_handler(struct drm_i915_private *dev_priv,
drivers/gpu/drm/i915/intel_lrc.c +5 -40

97 lines not on this page · raw message

drivers/gpu/drm/i915/intel_lrc.h +1 -1
index 4e60d54..e6a4900 100644
@@ -95,7 +95,7 @@ int intel_execlists_submission(struct i915_execbuffer_params *params,
 			       struct list_head *vmas);
 u32 intel_execlists_ctx_id(struct drm_i915_gem_object *ctx_obj);
 
-void intel_lrc_irq_handler(struct intel_engine_cs *ring);
+bool intel_lrc_irq_handler(struct intel_engine_cs *ring);
 void intel_execlists_retire_requests(struct intel_engine_cs *ring);
 
 #endif /* _INTEL_LRC_H_ */
drivers/gpu/drm/i915/intel_ringbuffer.h +0 -1
index 49fa41d..d99b167 100644
@@ -264,7 +264,6 @@ struct  intel_engine_cs {
 	/* Execlists */
 	spinlock_t execlist_lock;
 	struct list_head execlist_queue;
-	struct list_head execlist_retired_req_list;
 	u8 next_context_status_buffer;
 	u32             irq_keep_mask; /* bitmask for interrupts that should not be masked */
 	int		(*emit_request)(struct drm_i915_gem_request *request);
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#6 lore raw

Re: [PATCH 3/6] drm/i915: Unify execlist and legacy request life-cycles

Quoted 68 lines · Nick Hoath

There is a desire to simplify the i915 driver by reducing the number of different code paths introduced by the LRC / execlists support. As the execlists request is now part of the gem request it is possible and desirable to unify the request life-cycles for execlist and legacy requests.

A request is considered retireable if its seqno passed (i.e. the request has completed) and either it was never submitted to the ELSP or its context completed. This ensures that context save is carried out before the last request for a context is considered retireable. request_complete() now checks the elsp_submitted count when deciding if a request is complete. Requests that were not waiting for a context
switch interrupt (either as a result of being merged into a following request or by being a legacy request) will be considered retireable as soon as their seqno has passed.

Removed the extra request reference held for the execlist request.

Removed intel_execlists_retire_requests() and all references to intel_engine_cs.execlist_retired_req_list.

Changed gen8_cs_irq_handler() so that notify_ring() is called when contexts complete as well as when a user interrupt occurs so that notification happens when a request is complete and context save has finished.

v2: Rebase over the read-read optimisation changes

v3: Reworked IRQ handler after removing IRQ handler cleanup patch

v4: Fixed various pin leaks

v5: Removed ctx_complete flag & associated changes. Removed extraneous
request pin of context.
(Chris Wilson/Daniel Vetter)

Issue: VIZ-4277

---
 drivers/gpu/drm/i915/i915_drv.h         |  2 +-
 drivers/gpu/drm/i915/i915_gem.c         | 23 ++++++++---------
 drivers/gpu/drm/i915/i915_irq.c         |  7 ++---
 drivers/gpu/drm/i915/intel_lrc.c        | 45 ++++-----------------------------
 drivers/gpu/drm/i915/intel_lrc.h        |  2 +-
 drivers/gpu/drm/i915/intel_ringbuffer.h |  1 -
 6 files changed, 21 insertions(+), 59 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +1 -1
index 8afda45..ae08e57 100644
@@ -2914,7 +2914,7 @@ static inline bool i915_gem_request_completed(struct drm_i915_gem_request *req,
 
 	seqno = req->ring->get_seqno(req->ring, lazy_coherency);
 
-	return i915_seqno_passed(seqno, req->seqno);
+	return i915_seqno_passed(seqno, req->seqno) && !req->elsp_submitted;
 }
 
 int __must_check i915_gem_get_seqno(struct drm_device *dev, u32 *seqno);
drivers/gpu/drm/i915/i915_gem.c +0 -0
index e57061a..290a1ac 100644

drivers/gpu/drm/i915/i915_gem.c i915_gem_retire_requests_ring(struct intel_engine_cs *ring) · line 2853

Quoted 6 lines
drivers/gpu/drm/i915/i915_gem.c +3 -0
@@ -2848,12 +2848,16 @@ i915_gem_retire_requests_ring(struct intel_engine_cs *ring)
 
 		if (!list_empty(&obj->last_read_req[ring->id]->list))
 			break;
+		if (!i915_gem_request_completed(obj->last_read_req[ring->id],
+				true))
+			break;

What???? No. request_completed is and should only ever be has the seqno passed.
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#7 lore raw

PATCH 4/6 drm/i915: Improve dynamic management/eviction of lrc backing objects

In reply to Nick Hoath

Shovel all context related objects through the active queue and obj management.

- Added callback in vma_(un)bind to add CPU (un)mapping at same time if desired
- Inserted LRC hw context & ringbuf to vma active list

Issue: VIZ-4277

---
 drivers/gpu/drm/i915/i915_drv.h         |  4 ++
 drivers/gpu/drm/i915/i915_gem.c         |  3 ++
 drivers/gpu/drm/i915/i915_gem_gtt.c     |  8 ++++
 drivers/gpu/drm/i915/intel_lrc.c        | 28 +++++++++++--
 drivers/gpu/drm/i915/intel_ringbuffer.c | 71 ++++++++++++++++++---------------
 drivers/gpu/drm/i915/intel_ringbuffer.h |  3 --
 6 files changed, 79 insertions(+), 38 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +4 -0
index ae08e57..0dd4ace 100644
@@ -2161,6 +2161,10 @@ struct drm_i915_gem_object {
 			struct work_struct *work;
 		} userptr;
 	};
+
+	/** Support for automatic CPU side mapping of object */
+	int (*mmap)(struct drm_i915_gem_object *obj, bool unmap);
+	void *mappable;
 };
 #define to_intel_bo(x) container_of(x, struct drm_i915_gem_object, base)
 
drivers/gpu/drm/i915/i915_gem.c +3 -0
index 290a1ac..8bd318a 100644
@@ -3222,6 +3222,9 @@ static int __i915_vma_unbind(struct i915_vma *vma, bool wait)
 	if (vma->pin_count)
 		return -EBUSY;
 
+	if (obj->mmap)
+		obj->mmap(obj, true);
+
 	BUG_ON(obj->pages == NULL);
 
 	if (wait) {
drivers/gpu/drm/i915/i915_gem_gtt.c +8 -0
index 43f35d1..2812757 100644
@@ -3517,6 +3517,14 @@ int i915_vma_bind(struct i915_vma *vma, enum i915_cache_level cache_level,
 
 	vma->bound |= bind_flags;
 
+	if (vma->obj->mmap) {
+		ret = vma->obj->mmap(vma->obj, false);
+		if (ret) {
+			i915_vma_unbind(vma);
+			return ret;
+		}
+	}
+
 	return 0;
 }
 
drivers/gpu/drm/i915/intel_lrc.c +25 -3
index 8428ebd..069950e 100644
@@ -719,6 +719,18 @@ intel_logical_ring_advance_and_submit(struct drm_i915_gem_request *request)
 
 	intel_logical_ring_advance(request->ringbuf);
 
+	/* Push the hw context on to the active list */
+	i915_vma_move_to_active(
+			i915_gem_obj_to_ggtt(
+				request->ctx->engine[ring->id].state),
+			request);
+
+	/* Push the ringbuf on to the active list */
+	i915_vma_move_to_active(
+			i915_gem_obj_to_ggtt(
+			request->ctx->engine[ring->id].ringbuf->obj),
+			request);
+
 	request->tail = request->ringbuf->tail;
 
 	if (intel_ring_stopped(ring))
@@ -987,10 +999,15 @@ static int intel_lr_context_do_pin(struct intel_engine_cs *ring,
 	if (ret)
 		return ret;
 
-	ret = intel_pin_and_map_ringbuffer_obj(ring->dev, ringbuf);
+	ret = i915_gem_obj_ggtt_pin(ringbuf->obj, PAGE_SIZE,
+			PIN_MAPPABLE);
 	if (ret)
 		goto unpin_ctx_obj;
 
+	ret = i915_gem_object_set_to_gtt_domain(ringbuf->obj, true);
+	if (ret)
+		goto unpin_rb_obj;
+
 	ctx_obj->dirty = true;
 
 	/* Invalidate GuC TLB. */
@@ -999,6 +1016,8 @@ static int intel_lr_context_do_pin(struct intel_engine_cs *ring,
 
 	return ret;
 
+unpin_rb_obj:
+	i915_gem_object_ggtt_unpin(ringbuf->obj);
 unpin_ctx_obj:
 	i915_gem_object_ggtt_unpin(ctx_obj);
 
@@ -1033,7 +1052,7 @@ void intel_lr_context_unpin(struct drm_i915_gem_request *rq)
 	if (ctx_obj) {
 		WARN_ON(!mutex_is_locked(&ring->dev->struct_mutex));
 		if (--rq->ctx->engine[ring->id].pin_count == 0) {
-			intel_unpin_ringbuffer_obj(ringbuf);
+			i915_gem_object_ggtt_unpin(ringbuf->obj);
 			i915_gem_object_ggtt_unpin(ctx_obj);
 		}
 	}
@@ -2351,7 +2370,7 @@ void intel_lr_context_free(struct intel_context *ctx)
 			struct intel_engine_cs *ring = ringbuf->ring;
 
 			if (ctx == ring->default_context) {
-				intel_unpin_ringbuffer_obj(ringbuf);
+				i915_gem_object_ggtt_unpin(ringbuf->obj);
 				i915_gem_object_ggtt_unpin(ctx_obj);
 			}
 			WARN_ON(ctx->engine[ring->id].pin_count);
@@ -2518,5 +2537,8 @@ void intel_lr_context_reset(struct drm_device *dev,
 
 		ringbuf->head = 0;
 		ringbuf->tail = 0;
+
+		i915_gem_object_ggtt_unpin(
+				ctx->engine[ring->id].state);
 	}
 }
drivers/gpu/drm/i915/intel_ringbuffer.c +39 -32

109 lines not on this page · raw message

drivers/gpu/drm/i915/intel_ringbuffer.h +0 -3
index d99b167..8daaf99 100644
@@ -421,9 +421,6 @@ intel_write_status_page(struct intel_engine_cs *ring,
 
 struct intel_ringbuffer *
 intel_engine_create_ringbuffer(struct intel_engine_cs *engine, int size);
-int intel_pin_and_map_ringbuffer_obj(struct drm_device *dev,
-				     struct intel_ringbuffer *ringbuf);
-void intel_unpin_ringbuffer_obj(struct intel_ringbuffer *ringbuf);
 void intel_ringbuffer_free(struct intel_ringbuffer *ring);
 
 void intel_stop_ring_buffer(struct intel_engine_cs *ring);
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#8 lore raw

Re: [PATCH 4/6] drm/i915: Improve dynamic management/eviction of lrc backing objects

Quoted 38 lines · Nick Hoath

Shovel all context related objects through the active queue and obj management.

- Added callback in vma_(un)bind to add CPU (un)mapping at same time if desired
- Inserted LRC hw context & ringbuf to vma active list

Issue: VIZ-4277

---
 drivers/gpu/drm/i915/i915_drv.h         |  4 ++
 drivers/gpu/drm/i915/i915_gem.c         |  3 ++
 drivers/gpu/drm/i915/i915_gem_gtt.c     |  8 ++++
 drivers/gpu/drm/i915/intel_lrc.c        | 28 +++++++++++--
 drivers/gpu/drm/i915/intel_ringbuffer.c | 71 ++++++++++++++++++---------------
 drivers/gpu/drm/i915/intel_ringbuffer.h |  3 --
 6 files changed, 79 insertions(+), 38 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +4 -0
index ae08e57..0dd4ace 100644
@@ -2161,6 +2161,10 @@ struct drm_i915_gem_object {
 			struct work_struct *work;
 		} userptr;
 	};
+
+	/** Support for automatic CPU side mapping of object */
+	int (*mmap)(struct drm_i915_gem_object *obj, bool unmap);
+	void *mappable;
 };
 #define to_intel_bo(x) container_of(x, struct drm_i915_gem_object, base)
 
drivers/gpu/drm/i915/i915_gem.c +0 -0
index 290a1ac..8bd318a 100644

drivers/gpu/drm/i915/i915_gem.c static int __i915_vma_unbind(struct i915_vma *vma, bool wait) · line 3226

Quoted 5 lines
drivers/gpu/drm/i915/i915_gem.c +2 -0
@@ -3222,6 +3222,9 @@ static int __i915_vma_unbind(struct i915_vma *vma, bool wait)
 	if (vma->pin_count)
 		return -EBUSY;
 
+	if (obj->mmap)
+		obj->mmap(obj, true);

This is not a per-obj callback.
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#9 lore raw

Re: [PATCH 4/6] drm/i915: Improve dynamic management/eviction of lrc backing objects

In reply to Nick Hoath
reproduce:

Hi Nick,

[auto build test WARNING on next-20151020 -- if it's inappropriate base, please suggest rules for selecting the more suitable base]

url:    https://github.com/0day-ci/linux/commits/Nick-Hoath/lrc-lifecycle-cleanups/20151020-173347
config: i386-randconfig-x009-10191220 (attached as .config)
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386 

All warnings (new ones prefixed by >>):

drivers/gpu/drm/i915/i915_gem_gtt.c: In function 'i915_vma_bind':

>> drivers/gpu/drm/i915/i915_gem_gtt.c:3523:4: warning: ignoring return value of 'i915_vma_unbind', declared with attribute warn_unused_result [-Wunused-result]
       i915_vma_unbind(vma);
       ^

vim +/i915_vma_unbind +3523 drivers/gpu/drm/i915/i915_gem_gtt.c

3507 vma->node.start,
3508 vma->node.size);
3509 vma->pin_count--;
3510 if (ret)
3511 return ret;
3512 }
3513
3514 ret = vma->vm->bind_vma(vma, cache_level, bind_flags);
3515 if (ret)
3516 return ret;
3517
3518 vma->bound |= bind_flags;
3519
3520 if (vma->obj->mmap) {
3521 ret = vma->obj->mmap(vma->obj, false);
3522 if (ret) {

> 3523				i915_vma_unbind(vma);

3524 return ret;
3525 }
3526 }
3527
3528 return 0;
3529 }
3530
3531 /**

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#10 lore raw

PATCH 5/6 drm/i915: Add the CPU mapping of the hw context to the pinned items.

In reply to Nick Hoath

Pin the hw ctx mapping so that it is not mapped/unmapped per bb when doing GuC submission.

v2: Removed interim development extra mapping. (Daniel Vetter)

Issue: VIZ-4277

Cc David Gordon <[email protected]>
Cc Daniel Vetter <[email protected]>
---
 drivers/gpu/drm/i915/i915_debugfs.c | 14 ++++-------
 drivers/gpu/drm/i915/i915_drv.h     |  4 +++-
 drivers/gpu/drm/i915/intel_lrc.c    | 46 ++++++++++++++++++++++++++-----------
 3 files changed, 40 insertions(+), 24 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +5 -9
index a3b22bd..f0a172e 100644
@@ -1970,10 +1970,9 @@ static int i915_context_status(struct seq_file *m, void *unused)
 
 static void i915_dump_lrc_obj(struct seq_file *m,
 			      struct intel_engine_cs *ring,
-			      struct drm_i915_gem_object *ctx_obj)
+			      struct drm_i915_gem_object *ctx_obj,
+			      uint32_t *reg_state)
 {
-	struct page *page;
-	uint32_t *reg_state;
 	int j;
 	unsigned long ggtt_offset = 0;
 
@@ -1996,17 +1995,13 @@ static void i915_dump_lrc_obj(struct seq_file *m,
 		return;
 	}
 
-	page = i915_gem_object_get_page(ctx_obj, LRC_STATE_PN);
-	if (!WARN_ON(page == NULL)) {
-		reg_state = kmap_atomic(page);
-
+	if (!WARN_ON(reg_state == NULL)) {
 		for (j = 0; j < 0x600 / sizeof(u32) / 4; j += 4) {
 			seq_printf(m, "\t[0x%08lx] 0x%08x 0x%08x 0x%08x 0x%08x\n",
 				   ggtt_offset + 4096 + (j * 4),
 				   reg_state[j], reg_state[j + 1],
 				   reg_state[j + 2], reg_state[j + 3]);
 		}
-		kunmap_atomic(reg_state);
 	}
 
 	seq_putc(m, '\n');
@@ -2034,7 +2029,8 @@ static int i915_dump_lrc(struct seq_file *m, void *unused)
 		for_each_ring(ring, dev_priv, i) {
 			if (ring->default_context != ctx)
 				i915_dump_lrc_obj(m, ring,
-						  ctx->engine[i].state);
+						  ctx->engine[i].state,
+						  ctx->engine[i].reg_state);
 		}
 	}
 
drivers/gpu/drm/i915/i915_drv.h +3 -1
index 0dd4ace..dc69d67 100644
@@ -881,8 +881,10 @@ struct intel_context {
 	} legacy_hw_ctx;
 
 	/* Execlists */
-	struct {
+	struct intel_context_engine {
 		struct drm_i915_gem_object *state;
+		uint32_t *reg_state;
+		struct page *page;
 		struct intel_ringbuffer *ringbuf;
 		int pin_count;
 	} engine[I915_NUM_RINGS];
drivers/gpu/drm/i915/intel_lrc.c +32 -14

103 lines not on this page · raw message

Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#11 lore raw

Re: [PATCH 5/6] drm/i915: Add the CPU mapping of the hw context to the pinned items.

Quoted 4 lines · Nick Hoath

Pin the hw ctx mapping so that it is not mapped/unmapped per bb when doing GuC submission.

v2: Removed interim development extra mapping. (Daniel Vetter)

You've missed the point here. I've shown how to do kmapping once neatly, what you need the unbind callback for is to amoritize the cost of the ioremap.
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
author
#12 lore raw

PATCH 6/6 drm/i915: Only update ringbuf address when necessary

In reply to Nick Hoath

We now only need to update the address of the ringbuf object in the hw context when it is pinned, and the hw context is first CPU mapped

Issue: VIZ-4277

---
 drivers/gpu/drm/i915/intel_lrc.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/intel_lrc.c +3 -1
index a35efcd..2e529a4 100644
@@ -368,7 +368,6 @@ static int execlists_update_context(struct drm_i915_gem_request *rq)
 	WARN_ON(!i915_gem_obj_is_pinned(rb_obj));
 
 	reg_state[CTX_RING_TAIL+1] = rq->tail;
-	reg_state[CTX_RING_BUFFER_START+1] = i915_gem_obj_ggtt_offset(rb_obj);
 
 	if (ppgtt && !USES_FULL_48BIT_PPGTT(ppgtt->base.dev)) {
 		/* True 32b PPGTT with dynamic page allocation: update PDP
@@ -1027,6 +1026,9 @@ static int intel_lr_context_do_pin(
 	if (ret)
 		goto unpin_rb_obj;
 
+	ctx->engine[ring->id].reg_state[CTX_RING_BUFFER_START+1] =
+			i915_gem_obj_ggtt_offset(ringbuf->obj);
+
 	ctx_obj->dirty = true;
 
 	/* Invalidate GuC TLB. */
Signature · 6 lines
1.9.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx