All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] drm/i915: avoid concurrent writes to aux_inv
@ 2022-02-26  1:50 ` fei.yang
  0 siblings, 0 replies; 5+ messages in thread
From: fei.yang @ 2022-02-26  1:50 UTC (permalink / raw)
  To: intel-gfx; +Cc: Fei Yang, chris.p.wilson, dri-devel

From: Fei Yang <fei.yang@intel.com>

GPU hangs have been observed when multiple engines write to the same aux_inv
register at the same time. To avoid this each engine should only invalidate
its own auxiliary table. The function gen12_emit_flush_xcs() currently
invalidate the auxiliary table for all engines because the rq->engine is not
necessarily the engine eventually carrying out the request, and potentially
the engine could even be a virtual one (with engine->instance being -1).
With this patch, auxiliary table invalidation is done only for the engine
executing the request. And the mmio address for the aux_inv register is set
after the engine instance becomes certain.

Signed-off-by: Chris Wilson <chris.p.wilson@intel.com>
Signed-off-by: Fei Yang <fei.yang@intel.com>
---
 drivers/gpu/drm/i915/gt/gen8_engine_cs.c      | 38 +++----------------
 .../drm/i915/gt/intel_execlists_submission.c  | 38 +++++++++++++++++++
 drivers/gpu/drm/i915/i915_request.h           |  2 +
 3 files changed, 45 insertions(+), 33 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/gen8_engine_cs.c b/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
index 1f8cf4f790b2..9e39d04f46cc 100644
--- a/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
+++ b/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
@@ -165,30 +165,6 @@ static u32 preparser_disable(bool state)
 	return MI_ARB_CHECK | 1 << 8 | state;
 }
 
-static i915_reg_t aux_inv_reg(const struct intel_engine_cs *engine)
-{
-	static const i915_reg_t vd[] = {
-		GEN12_VD0_AUX_NV,
-		GEN12_VD1_AUX_NV,
-		GEN12_VD2_AUX_NV,
-		GEN12_VD3_AUX_NV,
-	};
-
-	static const i915_reg_t ve[] = {
-		GEN12_VE0_AUX_NV,
-		GEN12_VE1_AUX_NV,
-	};
-
-	if (engine->class == VIDEO_DECODE_CLASS)
-		return vd[engine->instance];
-
-	if (engine->class == VIDEO_ENHANCEMENT_CLASS)
-		return ve[engine->instance];
-
-	GEM_BUG_ON("unknown aux_inv reg\n");
-	return INVALID_MMIO_REG;
-}
-
 static u32 *gen12_emit_aux_table_inv(const i915_reg_t inv_reg, u32 *cs)
 {
 	*cs++ = MI_LOAD_REGISTER_IMM(1);
@@ -280,7 +256,7 @@ int gen12_emit_flush_xcs(struct i915_request *rq, u32 mode)
 	if (mode & EMIT_INVALIDATE)
 		aux_inv = rq->engine->mask & ~BIT(BCS0);
 	if (aux_inv)
-		cmd += 2 * hweight32(aux_inv) + 2;
+		cmd += 4;
 
 	cs = intel_ring_begin(rq, cmd);
 	if (IS_ERR(cs))
@@ -311,14 +287,10 @@ int gen12_emit_flush_xcs(struct i915_request *rq, u32 mode)
 	*cs++ = 0; /* value */
 
 	if (aux_inv) { /* hsdes: 1809175790 */
-		struct intel_engine_cs *engine;
-		unsigned int tmp;
-
-		*cs++ = MI_LOAD_REGISTER_IMM(hweight32(aux_inv));
-		for_each_engine_masked(engine, rq->engine->gt, aux_inv, tmp) {
-			*cs++ = i915_mmio_reg_offset(aux_inv_reg(engine));
-			*cs++ = AUX_INV;
-		}
+		*cs++ = MI_LOAD_REGISTER_IMM(1);
+		rq->vd_ve_aux_inv = cs;
+		*cs++ = 0; /* address to be set at submission to HW */
+		*cs++ = AUX_INV;
 		*cs++ = MI_NOOP;
 	}
 
diff --git a/drivers/gpu/drm/i915/gt/intel_execlists_submission.c b/drivers/gpu/drm/i915/gt/intel_execlists_submission.c
index 961d795220a3..f56726c861b7 100644
--- a/drivers/gpu/drm/i915/gt/intel_execlists_submission.c
+++ b/drivers/gpu/drm/i915/gt/intel_execlists_submission.c
@@ -1253,6 +1253,34 @@ static bool completed(const struct i915_request *rq)
 	return __i915_request_is_complete(rq);
 }
 
+static i915_reg_t aux_inv_reg(const struct intel_engine_cs *engine)
+{
+	static const i915_reg_t vd[] = {
+		GEN12_VD0_AUX_NV,
+		GEN12_VD1_AUX_NV,
+		GEN12_VD2_AUX_NV,
+		GEN12_VD3_AUX_NV,
+	};
+
+	static const i915_reg_t ve[] = {
+		GEN12_VE0_AUX_NV,
+		GEN12_VE1_AUX_NV,
+	};
+
+	if (engine->class == VIDEO_DECODE_CLASS) {
+		GEM_BUG_ON(engine->instance >= ARRAY_SIZE(vd));
+		return vd[engine->instance];
+	}
+
+	if (engine->class == VIDEO_ENHANCEMENT_CLASS) {
+		GEM_BUG_ON(engine->instance >= ARRAY_SIZE(ve));
+		return ve[engine->instance];
+	}
+
+	GEM_BUG_ON("unknown aux_inv reg\n");
+	return INVALID_MMIO_REG;
+}
+
 static void execlists_dequeue(struct intel_engine_cs *engine)
 {
 	struct intel_engine_execlists * const execlists = &engine->execlists;
@@ -1533,6 +1561,16 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
 			}
 
 			if (__i915_request_submit(rq)) {
+				/* hsdes: 1809175790 */
+				if (rq->vd_ve_aux_inv &&
+				    (engine->class == VIDEO_DECODE_CLASS ||
+				     engine->class == VIDEO_ENHANCEMENT_CLASS)) {
+					*rq->vd_ve_aux_inv =
+						i915_mmio_reg_offset(
+							aux_inv_reg(engine));
+					rq->vd_ve_aux_inv = NULL;
+					rq->execution_mask = engine->mask;
+				}
 				if (!merge) {
 					*port++ = i915_request_get(last);
 					last = NULL;
diff --git a/drivers/gpu/drm/i915/i915_request.h b/drivers/gpu/drm/i915/i915_request.h
index 28b1f9db5487..69de32e5e15d 100644
--- a/drivers/gpu/drm/i915/i915_request.h
+++ b/drivers/gpu/drm/i915/i915_request.h
@@ -350,6 +350,8 @@ struct i915_request {
 		struct list_head link;
 		unsigned long delay;
 	} mock;)
+
+	u32 *vd_ve_aux_inv;
 };
 
 #define I915_FENCE_GFP (GFP_KERNEL | __GFP_RETRY_MAYFAIL | __GFP_NOWARN)
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 5+ messages in thread

* [Intel-gfx] [PATCH] drm/i915: avoid concurrent writes to aux_inv
@ 2022-02-26  1:50 ` fei.yang
  0 siblings, 0 replies; 5+ messages in thread
From: fei.yang @ 2022-02-26  1:50 UTC (permalink / raw)
  To: intel-gfx; +Cc: chris.p.wilson, dri-devel

From: Fei Yang <fei.yang@intel.com>

GPU hangs have been observed when multiple engines write to the same aux_inv
register at the same time. To avoid this each engine should only invalidate
its own auxiliary table. The function gen12_emit_flush_xcs() currently
invalidate the auxiliary table for all engines because the rq->engine is not
necessarily the engine eventually carrying out the request, and potentially
the engine could even be a virtual one (with engine->instance being -1).
With this patch, auxiliary table invalidation is done only for the engine
executing the request. And the mmio address for the aux_inv register is set
after the engine instance becomes certain.

Signed-off-by: Chris Wilson <chris.p.wilson@intel.com>
Signed-off-by: Fei Yang <fei.yang@intel.com>
---
 drivers/gpu/drm/i915/gt/gen8_engine_cs.c      | 38 +++----------------
 .../drm/i915/gt/intel_execlists_submission.c  | 38 +++++++++++++++++++
 drivers/gpu/drm/i915/i915_request.h           |  2 +
 3 files changed, 45 insertions(+), 33 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/gen8_engine_cs.c b/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
index 1f8cf4f790b2..9e39d04f46cc 100644
--- a/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
+++ b/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
@@ -165,30 +165,6 @@ static u32 preparser_disable(bool state)
 	return MI_ARB_CHECK | 1 << 8 | state;
 }
 
-static i915_reg_t aux_inv_reg(const struct intel_engine_cs *engine)
-{
-	static const i915_reg_t vd[] = {
-		GEN12_VD0_AUX_NV,
-		GEN12_VD1_AUX_NV,
-		GEN12_VD2_AUX_NV,
-		GEN12_VD3_AUX_NV,
-	};
-
-	static const i915_reg_t ve[] = {
-		GEN12_VE0_AUX_NV,
-		GEN12_VE1_AUX_NV,
-	};
-
-	if (engine->class == VIDEO_DECODE_CLASS)
-		return vd[engine->instance];
-
-	if (engine->class == VIDEO_ENHANCEMENT_CLASS)
-		return ve[engine->instance];
-
-	GEM_BUG_ON("unknown aux_inv reg\n");
-	return INVALID_MMIO_REG;
-}
-
 static u32 *gen12_emit_aux_table_inv(const i915_reg_t inv_reg, u32 *cs)
 {
 	*cs++ = MI_LOAD_REGISTER_IMM(1);
@@ -280,7 +256,7 @@ int gen12_emit_flush_xcs(struct i915_request *rq, u32 mode)
 	if (mode & EMIT_INVALIDATE)
 		aux_inv = rq->engine->mask & ~BIT(BCS0);
 	if (aux_inv)
-		cmd += 2 * hweight32(aux_inv) + 2;
+		cmd += 4;
 
 	cs = intel_ring_begin(rq, cmd);
 	if (IS_ERR(cs))
@@ -311,14 +287,10 @@ int gen12_emit_flush_xcs(struct i915_request *rq, u32 mode)
 	*cs++ = 0; /* value */
 
 	if (aux_inv) { /* hsdes: 1809175790 */
-		struct intel_engine_cs *engine;
-		unsigned int tmp;
-
-		*cs++ = MI_LOAD_REGISTER_IMM(hweight32(aux_inv));
-		for_each_engine_masked(engine, rq->engine->gt, aux_inv, tmp) {
-			*cs++ = i915_mmio_reg_offset(aux_inv_reg(engine));
-			*cs++ = AUX_INV;
-		}
+		*cs++ = MI_LOAD_REGISTER_IMM(1);
+		rq->vd_ve_aux_inv = cs;
+		*cs++ = 0; /* address to be set at submission to HW */
+		*cs++ = AUX_INV;
 		*cs++ = MI_NOOP;
 	}
 
diff --git a/drivers/gpu/drm/i915/gt/intel_execlists_submission.c b/drivers/gpu/drm/i915/gt/intel_execlists_submission.c
index 961d795220a3..f56726c861b7 100644
--- a/drivers/gpu/drm/i915/gt/intel_execlists_submission.c
+++ b/drivers/gpu/drm/i915/gt/intel_execlists_submission.c
@@ -1253,6 +1253,34 @@ static bool completed(const struct i915_request *rq)
 	return __i915_request_is_complete(rq);
 }
 
+static i915_reg_t aux_inv_reg(const struct intel_engine_cs *engine)
+{
+	static const i915_reg_t vd[] = {
+		GEN12_VD0_AUX_NV,
+		GEN12_VD1_AUX_NV,
+		GEN12_VD2_AUX_NV,
+		GEN12_VD3_AUX_NV,
+	};
+
+	static const i915_reg_t ve[] = {
+		GEN12_VE0_AUX_NV,
+		GEN12_VE1_AUX_NV,
+	};
+
+	if (engine->class == VIDEO_DECODE_CLASS) {
+		GEM_BUG_ON(engine->instance >= ARRAY_SIZE(vd));
+		return vd[engine->instance];
+	}
+
+	if (engine->class == VIDEO_ENHANCEMENT_CLASS) {
+		GEM_BUG_ON(engine->instance >= ARRAY_SIZE(ve));
+		return ve[engine->instance];
+	}
+
+	GEM_BUG_ON("unknown aux_inv reg\n");
+	return INVALID_MMIO_REG;
+}
+
 static void execlists_dequeue(struct intel_engine_cs *engine)
 {
 	struct intel_engine_execlists * const execlists = &engine->execlists;
@@ -1533,6 +1561,16 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
 			}
 
 			if (__i915_request_submit(rq)) {
+				/* hsdes: 1809175790 */
+				if (rq->vd_ve_aux_inv &&
+				    (engine->class == VIDEO_DECODE_CLASS ||
+				     engine->class == VIDEO_ENHANCEMENT_CLASS)) {
+					*rq->vd_ve_aux_inv =
+						i915_mmio_reg_offset(
+							aux_inv_reg(engine));
+					rq->vd_ve_aux_inv = NULL;
+					rq->execution_mask = engine->mask;
+				}
 				if (!merge) {
 					*port++ = i915_request_get(last);
 					last = NULL;
diff --git a/drivers/gpu/drm/i915/i915_request.h b/drivers/gpu/drm/i915/i915_request.h
index 28b1f9db5487..69de32e5e15d 100644
--- a/drivers/gpu/drm/i915/i915_request.h
+++ b/drivers/gpu/drm/i915/i915_request.h
@@ -350,6 +350,8 @@ struct i915_request {
 		struct list_head link;
 		unsigned long delay;
 	} mock;)
+
+	u32 *vd_ve_aux_inv;
 };
 
 #define I915_FENCE_GFP (GFP_KERNEL | __GFP_RETRY_MAYFAIL | __GFP_NOWARN)
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 5+ messages in thread

* [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915: avoid concurrent writes to aux_inv
  2022-02-26  1:50 ` [Intel-gfx] " fei.yang
  (?)
@ 2022-02-26  3:56 ` Patchwork
  -1 siblings, 0 replies; 5+ messages in thread
From: Patchwork @ 2022-02-26  3:56 UTC (permalink / raw)
  To: fei.yang; +Cc: intel-gfx

== Series Details ==

Series: drm/i915: avoid concurrent writes to aux_inv
URL   : https://patchwork.freedesktop.org/series/100772/
State : warning

== Summary ==

$ dim checkpatch origin/drm-tip
2d74b0515d73 drm/i915: avoid concurrent writes to aux_inv
-:6: WARNING:COMMIT_LOG_LONG_LINE: Possible unwrapped commit description (prefer a maximum 75 chars per line)
#6: 
GPU hangs have been observed when multiple engines write to the same aux_inv

-:130: CHECK:OPEN_ENDED_LINE: Lines should not end with a '('
#130: FILE: drivers/gpu/drm/i915/gt/intel_execlists_submission.c:1569:
+						i915_mmio_reg_offset(

total: 0 errors, 1 warnings, 1 checks, 114 lines checked



^ permalink raw reply	[flat|nested] 5+ messages in thread

* [Intel-gfx] ✗ Fi.CI.SPARSE: warning for drm/i915: avoid concurrent writes to aux_inv
  2022-02-26  1:50 ` [Intel-gfx] " fei.yang
  (?)
  (?)
@ 2022-02-26  3:57 ` Patchwork
  -1 siblings, 0 replies; 5+ messages in thread
From: Patchwork @ 2022-02-26  3:57 UTC (permalink / raw)
  To: fei.yang; +Cc: intel-gfx

== Series Details ==

Series: drm/i915: avoid concurrent writes to aux_inv
URL   : https://patchwork.freedesktop.org/series/100772/
State : warning

== Summary ==

$ dim sparse --fast origin/drm-tip
Sparse version: v0.6.2
Fast mode used, each commit won't be checked separately.



^ permalink raw reply	[flat|nested] 5+ messages in thread

* [Intel-gfx] ✗ Fi.CI.BAT: failure for drm/i915: avoid concurrent writes to aux_inv
  2022-02-26  1:50 ` [Intel-gfx] " fei.yang
                   ` (2 preceding siblings ...)
  (?)
@ 2022-02-26  4:20 ` Patchwork
  -1 siblings, 0 replies; 5+ messages in thread
From: Patchwork @ 2022-02-26  4:20 UTC (permalink / raw)
  To: fei.yang; +Cc: intel-gfx

[-- Attachment #1: Type: text/plain, Size: 2678 bytes --]

== Series Details ==

Series: drm/i915: avoid concurrent writes to aux_inv
URL   : https://patchwork.freedesktop.org/series/100772/
State : failure

== Summary ==

CI Bug Log - changes from CI_DRM_11290 -> Patchwork_22423
====================================================

Summary
-------

  **FAILURE**

  Serious unknown changes coming with Patchwork_22423 absolutely need to be
  verified manually.
  
  If you think the reported changes have nothing to do with the changes
  introduced in Patchwork_22423, please notify your bug team to allow them
  to document this new failure mode, which will reduce false positives in CI.

  External URL: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_22423/index.html

Participating hosts (41 -> 17)
------------------------------

  ERROR: It appears as if the changes made in Patchwork_22423 prevented too many machines from booting.

  Missing    (24): fi-kbl-soraka fi-bdw-gvtdvm fi-apl-guc fi-skl-6600u fi-bxt-dsi fi-bsw-n3050 fi-glk-dsi fi-kbl-7500u fi-bsw-nick fi-skl-6700k2 fi-kbl-7567u fi-skl-guc fi-cfl-8700k fi-glk-j4005 fi-ehl-2 bat-jsl-2 bat-jsl-1 fi-jsl-1 fi-bsw-cyan fi-cfl-guc fi-kbl-x1275 fi-cfl-8109u fi-kbl-8809g fi-bsw-kefka 

Known issues
------------

  Here are the changes found in Patchwork_22423 that come from known issues:

### IGT changes ###

#### Possible fixes ####

  * igt@kms_busy@basic@flip:
    - {bat-adlp-6}:       [DMESG-WARN][1] ([i915#3576]) -> [PASS][2]
   [1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11290/bat-adlp-6/igt@kms_busy@basic@flip.html
   [2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_22423/bat-adlp-6/igt@kms_busy@basic@flip.html

  
  {name}: This element is suppressed. This means it is ignored when computing
          the status of the difference (SUCCESS, WARNING, or FAILURE).

  [fdo#109315]: https://bugs.freedesktop.org/show_bug.cgi?id=109315
  [i915#1759]: https://gitlab.freedesktop.org/drm/intel/issues/1759
  [i915#2575]: https://gitlab.freedesktop.org/drm/intel/issues/2575
  [i915#3576]: https://gitlab.freedesktop.org/drm/intel/issues/3576


Build changes
-------------

  * Linux: CI_DRM_11290 -> Patchwork_22423

  CI-20190529: 20190529
  CI_DRM_11290: e4658cb77436a0a406de83fef483b52f84e17208 @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_6359: 57049558c452272b27eeb099fac07e55a924bbf9 @ https://gitlab.freedesktop.org/drm/igt-gpu-tools.git
  Patchwork_22423: 2d74b0515d735ff44ad36fc6849a32a38893b649 @ git://anongit.freedesktop.org/gfx-ci/linux


== Linux commits ==

2d74b0515d73 drm/i915: avoid concurrent writes to aux_inv

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_22423/index.html

[-- Attachment #2: Type: text/html, Size: 3102 bytes --]

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2022-02-26  4:20 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-02-26  1:50 [PATCH] drm/i915: avoid concurrent writes to aux_inv fei.yang
2022-02-26  1:50 ` [Intel-gfx] " fei.yang
2022-02-26  3:56 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for " Patchwork
2022-02-26  3:57 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2022-02-26  4:20 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.