[PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Chris Wilson]

If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.

References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
---
 drivers/gpu/drm/i915/i915_gem.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 8b52cb768a67..a4031fab57b0 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3312,8 +3312,8 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 			intel_engine_dump(engine, &p, "%s\n", engine->name);
 	}
 
-	set_bit(I915_WEDGED, &i915->gpu_error.flags);
-	smp_mb__after_atomic();
+	if (test_and_set_bit(I915_WEDGED, &i915->gpu_error.flags))
+		goto out;
 
 	/*
 	 * First, stop submission to hw, but do not yet complete requests by
@@ -3372,6 +3372,7 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 		i915_gem_reset_finish_engine(engine);
 	}
 
+out:
 	GEM_TRACE("end\n");
 
 	wake_up_all(&i915->gpu_error.reset_queue);
-- 
2.18.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

✓ Fi.CI.BAT: success for drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Patchwork]

== Series Details ==

Series: drm/i915: Skip repeated calls to i915_gem_set_wedged()
URL   : https://patchwork.freedesktop.org/series/47067/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4528 -> Patchwork_9747 =

== Summary - SUCCESS ==

  No regressions found.

  External URL: https://patchwork.freedesktop.org/api/1.0/series/47067/revisions/1/mbox/

== Known issues ==

  Here are the changes found in Patchwork_9747 that come from known issues:

  === IGT changes ===

    ==== Issues hit ====

    igt@gem_ctx_create@basic-files:
      fi-glk-j4005:       PASS -> DMESG-WARN (fdo#105719)

    igt@kms_pipe_crc_basic@hang-read-crc-pipe-c:
      fi-skl-guc:         PASS -> FAIL (fdo#103191)

    
    ==== Possible fixes ====

    igt@drv_module_reload@basic-no-display:
      fi-glk-j4005:       DMESG-WARN (fdo#106725) -> PASS

    igt@kms_pipe_crc_basic@suspend-read-crc-pipe-c:
      fi-ivb-3520m:       FAIL (fdo#103375) -> PASS

    
  fdo#103191 https://bugs.freedesktop.org/show_bug.cgi?id=103191
  fdo#103375 https://bugs.freedesktop.org/show_bug.cgi?id=103375
  fdo#105719 https://bugs.freedesktop.org/show_bug.cgi?id=105719
  fdo#106725 https://bugs.freedesktop.org/show_bug.cgi?id=106725


== Participating hosts (48 -> 43) ==

  Additional (1): fi-kbl-8809g 
  Missing    (6): fi-ilk-m540 fi-hsw-4200u fi-byt-squawks fi-bsw-cyan fi-ctg-p8600 fi-pnv-d510 


== Build changes ==

    * Linux: CI_DRM_4528 -> Patchwork_9747

  CI_DRM_4528: 9e2763473a5f24fa8848e5508cae60c3b5af3cf3 @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_4571: 65fccc149b85968cdce4737266b056059c1510f3 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
  Patchwork_9747: 8c8dac2e80e2deaa195e17732c979a8949325c01 @ git://anongit.freedesktop.org/gfx-ci/linux


== Linux commits ==

8c8dac2e80e2 drm/i915: Skip repeated calls to i915_gem_set_wedged()

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_9747/issues.html
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

✓ Fi.CI.IGT: success for drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Patchwork]

== Series Details ==

Series: drm/i915: Skip repeated calls to i915_gem_set_wedged()
URL   : https://patchwork.freedesktop.org/series/47067/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4528_full -> Patchwork_9747_full =

== Summary - SUCCESS ==

  No regressions found.

  

== Known issues ==

  Here are the changes found in Patchwork_9747_full that come from known issues:

  === IGT changes ===

    ==== Possible fixes ====

    igt@kms_flip@flip-vs-expired-vblank-interruptible:
      shard-glk:          FAIL (fdo#105363) -> PASS

    
  fdo#105363 https://bugs.freedesktop.org/show_bug.cgi?id=105363


== Participating hosts (5 -> 5) ==

  No changes in participating hosts


== Build changes ==

    * Linux: CI_DRM_4528 -> Patchwork_9747

  CI_DRM_4528: 9e2763473a5f24fa8848e5508cae60c3b5af3cf3 @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_4571: 65fccc149b85968cdce4737266b056059c1510f3 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
  Patchwork_9747: 8c8dac2e80e2deaa195e17732c979a8949325c01 @ git://anongit.freedesktop.org/gfx-ci/linux
  piglit_4509: fdc5a4ca11124ab8413c7988896eec4c97336694 @ git://anongit.freedesktop.org/piglit

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_9747/shards.html
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Chris Wilson]

If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.

v2: Make sure the double set-wedged is synchronous, a parallel call
should not return before the driver is indeed wedged.

References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
---
 drivers/gpu/drm/i915/i915_gem.c       | 26 ++++++++++++++++++++++----
 drivers/gpu/drm/i915/i915_gpu_error.h |  3 ++-
 2 files changed, 24 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 8b52cb768a67..99c91c7ad46c 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3298,12 +3298,24 @@ static void nop_complete_submit_request(struct i915_request *request)
 	spin_unlock_irqrestore(&request->engine->timeline.lock, flags);
 }
 
+static void wait_for_wedged(struct drm_i915_private *i915)
+{
+	struct i915_gpu_error *error = &i915->gpu_error;
+	DEFINE_WAIT_BIT(wq_entry, &error->flags, I915_WEDGED);
+
+	__wait_on_bit(&error->reset_queue,
+		      &wq_entry, bit_wait, TASK_UNINTERRUPTIBLE);
+}
+
 void i915_gem_set_wedged(struct drm_i915_private *i915)
 {
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 
-	GEM_TRACE("start\n");
+	if (test_and_set_bit(I915_WEDGE_IN_PROGRESS, &i915->gpu_error.flags)) {
+		wait_for_wedged(i915);
+		return;
+	}
 
 	if (GEM_SHOW_DEBUG()) {
 		struct drm_printer p = drm_debug_printer(__func__);
@@ -3312,8 +3324,7 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 			intel_engine_dump(engine, &p, "%s\n", engine->name);
 	}
 
-	set_bit(I915_WEDGED, &i915->gpu_error.flags);
-	smp_mb__after_atomic();
+	GEM_TRACE("start\n");
 
 	/*
 	 * First, stop submission to hw, but do not yet complete requests by
@@ -3372,6 +3383,9 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 		i915_gem_reset_finish_engine(engine);
 	}
 
+	smp_mb__before_atomic();
+	set_bit(I915_WEDGED, &i915->gpu_error.flags);
+
 	GEM_TRACE("end\n");
 
 	wake_up_all(&i915->gpu_error.reset_queue);
@@ -3379,10 +3393,14 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 
 bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct i915_timeline *tl;
 
 	lockdep_assert_held(&i915->drm.struct_mutex);
-	if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
+
+	while (test_and_clear_bit(I915_WEDGE_IN_PROGRESS, &error->flags))
+		wait_for_wedged(i915);
+	if (!test_bit(I915_WEDGE_IN_PROGRESS, &error->flags))
 		return true;
 
 	GEM_TRACE("start\n");
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.h b/drivers/gpu/drm/i915/i915_gpu_error.h
index f893a4e8b783..1a78a8f330f2 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.h
+++ b/drivers/gpu/drm/i915/i915_gpu_error.h
@@ -267,8 +267,9 @@ struct i915_gpu_error {
 #define I915_RESET_BACKOFF	0
 #define I915_RESET_HANDOFF	1
 #define I915_RESET_MODESET	2
+#define I915_RESET_ENGINE	3
 #define I915_WEDGED		(BITS_PER_LONG - 1)
-#define I915_RESET_ENGINE	(I915_WEDGED - I915_NUM_ENGINES)
+#define I915_WEDGE_IN_PROGRESS	(I915_WEDGED - 1)
 
 	/** Number of times an engine has been reset */
 	u32 reset_engine_count[I915_NUM_ENGINES];
-- 
2.18.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

✗ Fi.CI.CHECKPATCH: warning for drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev2)

[Patchwork]

== Series Details ==

Series: drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev2)
URL   : https://patchwork.freedesktop.org/series/47067/
State : warning

== Summary ==

$ dim checkpatch origin/drm-tip
9b348bfd231b drm/i915: Skip repeated calls to i915_gem_set_wedged()
-:58: WARNING:MEMORY_BARRIER: memory barrier without comment
#58: FILE: drivers/gpu/drm/i915/i915_gem.c:3386:
+	smp_mb__before_atomic();

total: 0 errors, 1 warnings, 0 checks, 68 lines checked

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

✓ Fi.CI.BAT: success for drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev2)

[Patchwork]

== Series Details ==

Series: drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev2)
URL   : https://patchwork.freedesktop.org/series/47067/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4529 -> Patchwork_9749 =

== Summary - WARNING ==

  Minor unknown changes coming with Patchwork_9749 need to be verified
  manually.
  
  If you think the reported changes have nothing to do with the changes
  introduced in Patchwork_9749, please notify your bug team to allow them
  to document this new failure mode, which will reduce false positives in CI.

  External URL: https://patchwork.freedesktop.org/api/1.0/series/47067/revisions/2/mbox/

== Possible new issues ==

  Here are the unknown changes that may have been introduced in Patchwork_9749:

  === IGT changes ===

    ==== Warnings ====

    igt@drv_selftest@live_execlists:
      fi-kbl-7500u:       SKIP -> PASS +1

    
== Known issues ==

  Here are the changes found in Patchwork_9749 that come from known issues:

  === IGT changes ===

    ==== Issues hit ====

    igt@drv_selftest@mock_sanitycheck:
      fi-glk-j4005:       PASS -> DMESG-WARN (fdo#107344)

    igt@gem_exec_fence@nb-await-default:
      fi-glk-j4005:       PASS -> DMESG-WARN (fdo#105719)

    igt@kms_flip@basic-flip-vs-modeset:
      fi-glk-j4005:       PASS -> DMESG-WARN (fdo#106000)

    
    ==== Possible fixes ====

    igt@drv_selftest@live_hangcheck:
      fi-kbl-7500u:       DMESG-FAIL (fdo#106947) -> PASS
      fi-skl-guc:         DMESG-FAIL (fdo#107174) -> PASS

    igt@drv_selftest@live_workarounds:
      fi-skl-6700hq:      DMESG-FAIL (fdo#107292) -> PASS

    igt@gem_exec_gttfill@basic:
      fi-glk-j4005:       DMESG-WARN (fdo#106097) -> PASS

    
  fdo#105719 https://bugs.freedesktop.org/show_bug.cgi?id=105719
  fdo#106000 https://bugs.freedesktop.org/show_bug.cgi?id=106000
  fdo#106097 https://bugs.freedesktop.org/show_bug.cgi?id=106097
  fdo#106947 https://bugs.freedesktop.org/show_bug.cgi?id=106947
  fdo#107174 https://bugs.freedesktop.org/show_bug.cgi?id=107174
  fdo#107292 https://bugs.freedesktop.org/show_bug.cgi?id=107292
  fdo#107344 https://bugs.freedesktop.org/show_bug.cgi?id=107344


== Participating hosts (50 -> 43) ==

  Missing    (7): fi-ilk-m540 fi-hsw-4200u fi-byt-squawks fi-bsw-cyan fi-ctg-p8600 fi-skl-caroline fi-byt-clapper 


== Build changes ==

    * Linux: CI_DRM_4529 -> Patchwork_9749

  CI_DRM_4529: f44f476c6bb735cb4067af0c4417c49d105bf419 @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_4571: 65fccc149b85968cdce4737266b056059c1510f3 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
  Patchwork_9749: 9b348bfd231b38751c4d528a89d795745484c760 @ git://anongit.freedesktop.org/gfx-ci/linux


== Linux commits ==

9b348bfd231b drm/i915: Skip repeated calls to i915_gem_set_wedged()

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_9749/issues.html
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Chris Wilson]

If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.

v2: Make sure the double set-wedged is synchronous, a parallel call
should not return before the driver is indeed wedged.

References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
---
 drivers/gpu/drm/i915/i915_gem.c       | 31 ++++++++++++++++++++++-----
 drivers/gpu/drm/i915/i915_gpu_error.h |  3 ++-
 2 files changed, 28 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 8b52cb768a67..832c65734c47 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3298,12 +3298,26 @@ static void nop_complete_submit_request(struct i915_request *request)
 	spin_unlock_irqrestore(&request->engine->timeline.lock, flags);
 }
 
+static void wait_for_wedged(struct i915_gpu_error *error)
+{
+	DEFINE_WAIT_BIT(wq_entry, &error->flags, I915_WEDGED);
+
+	__wait_on_bit(&error->reset_queue,
+		      &wq_entry, bit_wait, TASK_UNINTERRUPTIBLE);
+}
+
 void i915_gem_set_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 
-	GEM_TRACE("start\n");
+	if (test_and_set_bit(I915_WEDGE_IN_PROGRESS, &error->flags)) {
+		wait_for_wedged(error);
+		return;
+	}
+	if (test_bit(I915_WEDGED, &error->flags))
+		return;
 
 	if (GEM_SHOW_DEBUG()) {
 		struct drm_printer p = drm_debug_printer(__func__);
@@ -3312,8 +3326,7 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 			intel_engine_dump(engine, &p, "%s\n", engine->name);
 	}
 
-	set_bit(I915_WEDGED, &i915->gpu_error.flags);
-	smp_mb__after_atomic();
+	GEM_TRACE("start\n");
 
 	/*
 	 * First, stop submission to hw, but do not yet complete requests by
@@ -3372,17 +3385,25 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 		i915_gem_reset_finish_engine(engine);
 	}
 
+	smp_mb__before_atomic();
+	set_bit(I915_WEDGED, &error->flags);
+	clear_bit(I915_WEDGE_IN_PROGRESS, &error->flags);
+
 	GEM_TRACE("end\n");
 
-	wake_up_all(&i915->gpu_error.reset_queue);
+	wake_up_all(&error->reset_queue);
 }
 
 bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct i915_timeline *tl;
 
 	lockdep_assert_held(&i915->drm.struct_mutex);
-	if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
+
+	if (test_bit(I915_WEDGE_IN_PROGRESS, &error->flags))
+		wait_for_wedged(error);
+	if (!test_bit(I915_WEDGED, &error->flags))
 		return true;
 
 	GEM_TRACE("start\n");
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.h b/drivers/gpu/drm/i915/i915_gpu_error.h
index f893a4e8b783..1a78a8f330f2 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.h
+++ b/drivers/gpu/drm/i915/i915_gpu_error.h
@@ -267,8 +267,9 @@ struct i915_gpu_error {
 #define I915_RESET_BACKOFF	0
 #define I915_RESET_HANDOFF	1
 #define I915_RESET_MODESET	2
+#define I915_RESET_ENGINE	3
 #define I915_WEDGED		(BITS_PER_LONG - 1)
-#define I915_RESET_ENGINE	(I915_WEDGED - I915_NUM_ENGINES)
+#define I915_WEDGE_IN_PROGRESS	(I915_WEDGED - 1)
 
 	/** Number of times an engine has been reset */
 	u32 reset_engine_count[I915_NUM_ENGINES];
-- 
2.18.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

✓ Fi.CI.IGT: success for drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev2)

[Patchwork]

== Series Details ==

Series: drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev2)
URL   : https://patchwork.freedesktop.org/series/47067/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4529_full -> Patchwork_9749_full =

== Summary - WARNING ==

  Minor unknown changes coming with Patchwork_9749_full need to be verified
  manually.
  
  If you think the reported changes have nothing to do with the changes
  introduced in Patchwork_9749_full, please notify your bug team to allow them
  to document this new failure mode, which will reduce false positives in CI.

  

== Possible new issues ==

  Here are the unknown changes that may have been introduced in Patchwork_9749_full:

  === IGT changes ===

    ==== Warnings ====

    igt@gem_mocs_settings@mocs-rc6-blt:
      shard-kbl:          PASS -> SKIP +1

    igt@gem_mocs_settings@mocs-rc6-vebox:
      shard-kbl:          SKIP -> PASS +2

    
== Known issues ==

  Here are the changes found in Patchwork_9749_full that come from known issues:

  === IGT changes ===

    ==== Issues hit ====

    igt@gem_eio@execbuf:
      shard-glk:          PASS -> INCOMPLETE (k.org#198133, fdo#103359) +15

    igt@gem_eio@in-flight-contexts-1us:
      shard-kbl:          PASS -> INCOMPLETE (fdo#103665) +14

    igt@gem_eio@in-flight-contexts-immediate:
      shard-hsw:          PASS -> INCOMPLETE (fdo#103540) +14

    igt@gem_eio@in-flight-immediate:
      shard-apl:          PASS -> INCOMPLETE (fdo#103927) +15

    igt@gem_eio@in-flight-suspend:
      shard-kbl:          PASS -> INCOMPLETE (fdo#106702, fdo#103665)

    igt@gem_eio@throttle:
      shard-snb:          PASS -> INCOMPLETE (fdo#105411) +16

    
  fdo#103359 https://bugs.freedesktop.org/show_bug.cgi?id=103359
  fdo#103540 https://bugs.freedesktop.org/show_bug.cgi?id=103540
  fdo#103665 https://bugs.freedesktop.org/show_bug.cgi?id=103665
  fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927
  fdo#105411 https://bugs.freedesktop.org/show_bug.cgi?id=105411
  fdo#106702 https://bugs.freedesktop.org/show_bug.cgi?id=106702
  k.org#198133 https://bugzilla.kernel.org/show_bug.cgi?id=198133


== Participating hosts (5 -> 5) ==

  No changes in participating hosts


== Build changes ==

    * Linux: CI_DRM_4529 -> Patchwork_9749

  CI_DRM_4529: f44f476c6bb735cb4067af0c4417c49d105bf419 @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_4571: 65fccc149b85968cdce4737266b056059c1510f3 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
  Patchwork_9749: 9b348bfd231b38751c4d528a89d795745484c760 @ git://anongit.freedesktop.org/gfx-ci/linux
  piglit_4509: fdc5a4ca11124ab8413c7988896eec4c97336694 @ git://anongit.freedesktop.org/piglit

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_9749/shards.html
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

✗ Fi.CI.CHECKPATCH: warning for drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev3)

[Patchwork]

== Series Details ==

Series: drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev3)
URL   : https://patchwork.freedesktop.org/series/47067/
State : warning

== Summary ==

$ dim checkpatch origin/drm-tip
fbfeaf105d25 drm/i915: Skip repeated calls to i915_gem_set_wedged()
-:60: WARNING:MEMORY_BARRIER: memory barrier without comment
#60: FILE: drivers/gpu/drm/i915/i915_gem.c:3388:
+	smp_mb__before_atomic();

total: 0 errors, 1 warnings, 0 checks, 73 lines checked

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH v4] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Chris Wilson]

If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.

v2: Make sure the double set-wedged is synchronous, a parallel call
should not return before the driver is indeed wedged.

References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
---
 drivers/gpu/drm/i915/i915_gem.c       | 32 ++++++++++++++++++++++-----
 drivers/gpu/drm/i915/i915_gpu_error.h |  3 ++-
 2 files changed, 29 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 8b52cb768a67..912be7356984 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3298,12 +3298,27 @@ static void nop_complete_submit_request(struct i915_request *request)
 	spin_unlock_irqrestore(&request->engine->timeline.lock, flags);
 }
 
+static void wait_for_wedged(struct i915_gpu_error *error)
+{
+	DEFINE_WAIT_BIT(wq_entry, &error->flags, I915_WEDGED);
+
+	__wait_on_bit(&error->reset_queue,
+		      &wq_entry, bit_wait, TASK_UNINTERRUPTIBLE);
+}
+
 void i915_gem_set_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 
-	GEM_TRACE("start\n");
+	if (test_bit(I915_WEDGED, &error->flags))
+		return;
+
+	if (test_and_set_bit(I915_WEDGE_IN_PROGRESS, &error->flags)) {
+		wait_for_wedged(error);
+		return;
+	}
 
 	if (GEM_SHOW_DEBUG()) {
 		struct drm_printer p = drm_debug_printer(__func__);
@@ -3312,8 +3327,7 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 			intel_engine_dump(engine, &p, "%s\n", engine->name);
 	}
 
-	set_bit(I915_WEDGED, &i915->gpu_error.flags);
-	smp_mb__after_atomic();
+	GEM_TRACE("start\n");
 
 	/*
 	 * First, stop submission to hw, but do not yet complete requests by
@@ -3372,17 +3386,25 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 		i915_gem_reset_finish_engine(engine);
 	}
 
+	smp_mb__before_atomic();
+	set_bit(I915_WEDGED, &error->flags);
+	clear_bit(I915_WEDGE_IN_PROGRESS, &error->flags);
+
 	GEM_TRACE("end\n");
 
-	wake_up_all(&i915->gpu_error.reset_queue);
+	wake_up_all(&error->reset_queue);
 }
 
 bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct i915_timeline *tl;
 
 	lockdep_assert_held(&i915->drm.struct_mutex);
-	if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
+
+	if (test_bit(I915_WEDGE_IN_PROGRESS, &error->flags))
+		wait_for_wedged(error);
+	if (!test_bit(I915_WEDGED, &error->flags))
 		return true;
 
 	GEM_TRACE("start\n");
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.h b/drivers/gpu/drm/i915/i915_gpu_error.h
index f893a4e8b783..1a78a8f330f2 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.h
+++ b/drivers/gpu/drm/i915/i915_gpu_error.h
@@ -267,8 +267,9 @@ struct i915_gpu_error {
 #define I915_RESET_BACKOFF	0
 #define I915_RESET_HANDOFF	1
 #define I915_RESET_MODESET	2
+#define I915_RESET_ENGINE	3
 #define I915_WEDGED		(BITS_PER_LONG - 1)
-#define I915_RESET_ENGINE	(I915_WEDGED - I915_NUM_ENGINES)
+#define I915_WEDGE_IN_PROGRESS	(I915_WEDGED - 1)
 
 	/** Number of times an engine has been reset */
 	u32 reset_engine_count[I915_NUM_ENGINES];
-- 
2.18.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

✓ Fi.CI.BAT: success for drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev3)

[Patchwork]

== Series Details ==

Series: drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev3)
URL   : https://patchwork.freedesktop.org/series/47067/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4530 -> Patchwork_9751 =

== Summary - SUCCESS ==

  No regressions found.

  External URL: https://patchwork.freedesktop.org/api/1.0/series/47067/revisions/3/mbox/

== Known issues ==

  Here are the changes found in Patchwork_9751 that come from known issues:

  === IGT changes ===

    ==== Issues hit ====

    igt@drv_selftest@live_workarounds:
      {fi-cfl-8109u}:     PASS -> DMESG-FAIL (fdo#107292)

    igt@gem_exec_suspend@basic-s4-devices:
      fi-kbl-7500u:       PASS -> DMESG-WARN (fdo#107139, fdo#105128)

    igt@kms_flip@basic-flip-vs-wf_vblank:
      fi-glk-j4005:       PASS -> FAIL (fdo#100368)

    igt@kms_pipe_crc_basic@read-crc-pipe-a-frame-sequence:
      fi-glk-j4005:       PASS -> FAIL (fdo#103481)

    
    ==== Possible fixes ====

    igt@drv_selftest@live_workarounds:
      fi-skl-6700hq:      DMESG-FAIL (fdo#107292) -> PASS

    igt@gem_exec_suspend@basic-s3:
      fi-glk-j4005:       DMESG-WARN (fdo#106097) -> PASS

    igt@kms_chamelium@dp-edid-read:
      fi-kbl-7500u:       FAIL (fdo#103841) -> PASS

    igt@kms_flip@basic-flip-vs-wf_vblank:
      fi-glk-dsi:         FAIL (fdo#100368) -> PASS

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

  fdo#100368 https://bugs.freedesktop.org/show_bug.cgi?id=100368
  fdo#103481 https://bugs.freedesktop.org/show_bug.cgi?id=103481
  fdo#103841 https://bugs.freedesktop.org/show_bug.cgi?id=103841
  fdo#105128 https://bugs.freedesktop.org/show_bug.cgi?id=105128
  fdo#106097 https://bugs.freedesktop.org/show_bug.cgi?id=106097
  fdo#107139 https://bugs.freedesktop.org/show_bug.cgi?id=107139
  fdo#107292 https://bugs.freedesktop.org/show_bug.cgi?id=107292


== Participating hosts (50 -> 43) ==

  Missing    (7): fi-ilk-m540 fi-hsw-4200u fi-byt-squawks fi-bsw-cyan fi-ctg-p8600 fi-skl-caroline fi-byt-clapper 


== Build changes ==

    * Linux: CI_DRM_4530 -> Patchwork_9751

  CI_DRM_4530: d27cc4a37a5cc1ef14a3aafdcb6682e5f6a85d09 @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_4571: 65fccc149b85968cdce4737266b056059c1510f3 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
  Patchwork_9751: fbfeaf105d2509bfd9333dfe8779b143b0dbd14a @ git://anongit.freedesktop.org/gfx-ci/linux


== Linux commits ==

fbfeaf105d25 drm/i915: Skip repeated calls to i915_gem_set_wedged()

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_9751/issues.html
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

✗ Fi.CI.CHECKPATCH: warning for drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev4)

[Patchwork]

== Series Details ==

Series: drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev4)
URL   : https://patchwork.freedesktop.org/series/47067/
State : warning

== Summary ==

$ dim checkpatch origin/drm-tip
a85ab6631205 drm/i915: Skip repeated calls to i915_gem_set_wedged()
-:61: WARNING:MEMORY_BARRIER: memory barrier without comment
#61: FILE: drivers/gpu/drm/i915/i915_gem.c:3389:
+	smp_mb__before_atomic();

total: 0 errors, 1 warnings, 0 checks, 74 lines checked

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

✗ Fi.CI.BAT: failure for drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev4)

[Patchwork]

== Series Details ==

Series: drm/i915: Skip repeated calls to i915_gem_set_wedged() (rev4)
URL   : https://patchwork.freedesktop.org/series/47067/
State : failure

== Summary ==

= CI Bug Log - changes from CI_DRM_4530 -> Patchwork_9752 =

== Summary - FAILURE ==

  Serious unknown changes coming with Patchwork_9752 absolutely need to be
  verified manually.
  
  If you think the reported changes have nothing to do with the changes
  introduced in Patchwork_9752, please notify your bug team to allow them
  to document this new failure mode, which will reduce false positives in CI.

  External URL: https://patchwork.freedesktop.org/api/1.0/series/47067/revisions/4/mbox/

== Possible new issues ==

  Here are the unknown changes that may have been introduced in Patchwork_9752:

  === IGT changes ===

    ==== Possible regressions ====

    igt@gem_cpu_reloc@basic:
      fi-byt-n2820:       PASS -> FAIL

    
== Known issues ==

  Here are the changes found in Patchwork_9752 that come from known issues:

  === IGT changes ===

    ==== Issues hit ====

    igt@drv_selftest@live_hangcheck:
      fi-kbl-7560u:       PASS -> DMESG-FAIL (fdo#106947, fdo#106560)

    igt@kms_flip@basic-flip-vs-modeset:
      fi-glk-j4005:       PASS -> DMESG-WARN (fdo#106000)

    igt@kms_flip@basic-flip-vs-wf_vblank:
      fi-glk-j4005:       PASS -> FAIL (fdo#100368)

    igt@kms_pipe_crc_basic@suspend-read-crc-pipe-b:
      fi-snb-2520m:       NOTRUN -> INCOMPLETE (fdo#103713)

    igt@kms_pipe_crc_basic@suspend-read-crc-pipe-c:
      fi-bxt-dsi:         PASS -> INCOMPLETE (fdo#103927)

    
    ==== Possible fixes ====

    igt@debugfs_test@read_all_entries:
      fi-snb-2520m:       INCOMPLETE (fdo#103713) -> PASS

    igt@drv_selftest@live_workarounds:
      fi-skl-6700hq:      DMESG-FAIL (fdo#107292) -> PASS

    igt@gem_exec_suspend@basic-s3:
      fi-glk-j4005:       DMESG-WARN (fdo#106097) -> PASS

    igt@kms_chamelium@dp-edid-read:
      fi-kbl-7500u:       FAIL (fdo#103841) -> PASS

    igt@kms_flip@basic-flip-vs-wf_vblank:
      fi-glk-dsi:         FAIL (fdo#100368) -> PASS

    
  fdo#100368 https://bugs.freedesktop.org/show_bug.cgi?id=100368
  fdo#103713 https://bugs.freedesktop.org/show_bug.cgi?id=103713
  fdo#103841 https://bugs.freedesktop.org/show_bug.cgi?id=103841
  fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927
  fdo#106000 https://bugs.freedesktop.org/show_bug.cgi?id=106000
  fdo#106097 https://bugs.freedesktop.org/show_bug.cgi?id=106097
  fdo#106560 https://bugs.freedesktop.org/show_bug.cgi?id=106560
  fdo#106947 https://bugs.freedesktop.org/show_bug.cgi?id=106947
  fdo#107292 https://bugs.freedesktop.org/show_bug.cgi?id=107292


== Participating hosts (50 -> 43) ==

  Missing    (7): fi-ilk-m540 fi-hsw-4200u fi-byt-squawks fi-bsw-cyan fi-ctg-p8600 fi-skl-caroline fi-byt-clapper 


== Build changes ==

    * Linux: CI_DRM_4530 -> Patchwork_9752

  CI_DRM_4530: d27cc4a37a5cc1ef14a3aafdcb6682e5f6a85d09 @ git://anongit.freedesktop.org/gfx-ci/linux
  IGT_4571: 65fccc149b85968cdce4737266b056059c1510f3 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
  Patchwork_9752: a85ab663120589128f229d9d90a434d726b37fdf @ git://anongit.freedesktop.org/gfx-ci/linux


== Linux commits ==

a85ab6631205 drm/i915: Skip repeated calls to i915_gem_set_wedged()

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_9752/issues.html
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Rodrigo Vivi]

On Mon, Jul 23, 2018 at 03:53:35PM +0100, Chris Wilson wrote:
> If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.
> 
> References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

> ---
>  drivers/gpu/drm/i915/i915_gem.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index 8b52cb768a67..a4031fab57b0 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -3312,8 +3312,8 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
>  			intel_engine_dump(engine, &p, "%s\n", engine->name);
>  	}
>  
> -	set_bit(I915_WEDGED, &i915->gpu_error.flags);
> -	smp_mb__after_atomic();
> +	if (test_and_set_bit(I915_WEDGED, &i915->gpu_error.flags))
> +		goto out;
>  
>  	/*
>  	 * First, stop submission to hw, but do not yet complete requests by
> @@ -3372,6 +3372,7 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
>  		i915_gem_reset_finish_engine(engine);
>  	}
>  
> +out:
>  	GEM_TRACE("end\n");
>  
>  	wake_up_all(&i915->gpu_error.reset_queue);
> -- 
> 2.18.0
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Chris Wilson]

Quoting Rodrigo Vivi (2018-07-23 23:23:32)
> On Mon, Jul 23, 2018 at 03:53:35PM +0100, Chris Wilson wrote:
> > If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.
> > 
> > References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
> > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> 
> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

Ta, I'll come back to v2+ when we have a test case that demonstrates the
danger in the race between wedging and unwedging. (Ever descending the
scale of probability.)
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Chris Wilson]

If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.

v2: Make sure the double set-wedged is synchronous, a parallel call
should not return before the driver is indeed wedged.
v3: Use a real mutex (and fingers crossed that lockdep is ok!)
v4: Fix incorrect conversion to return false if unwedged.

References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
---
 drivers/gpu/drm/i915/i915_gem.c               | 34 +++++++++++++------
 drivers/gpu/drm/i915/i915_gpu_error.h         |  4 ++-
 .../gpu/drm/i915/selftests/mock_gem_device.c  |  1 +
 3 files changed, 28 insertions(+), 11 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 0453eb42a1a3..98a3e2601ab4 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3300,10 +3300,15 @@ static void nop_complete_submit_request(struct i915_request *request)
 
 void i915_gem_set_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 
-	GEM_TRACE("start\n");
+	mutex_lock(&error->wedge_mutex);
+	if (test_bit(I915_WEDGED, &error->flags)) {
+		mutex_unlock(&error->wedge_mutex);
+		return;
+	}
 
 	if (GEM_SHOW_DEBUG()) {
 		struct drm_printer p = drm_debug_printer(__func__);
@@ -3312,8 +3317,7 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 			intel_engine_dump(engine, &p, "%s\n", engine->name);
 	}
 
-	if (test_and_set_bit(I915_WEDGED, &i915->gpu_error.flags))
-		goto out;
+	GEM_TRACE("start\n");
 
 	/*
 	 * First, stop submission to hw, but do not yet complete requests by
@@ -3373,20 +3377,28 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 		i915_gem_reset_finish_engine(engine);
 	}
 
-out:
+	smp_mb__before_atomic();
+	set_bit(I915_WEDGED, &error->flags);
+
 	GEM_TRACE("end\n");
+	mutex_unlock(&error->wedge_mutex);
 
-	wake_up_all(&i915->gpu_error.reset_queue);
+	wake_up_all(&error->reset_queue);
 }
 
 bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct i915_timeline *tl;
+	bool ret = false;
 
 	lockdep_assert_held(&i915->drm.struct_mutex);
-	if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
+
+	if (!test_bit(I915_WEDGED, &error->flags))
 		return true;
 
+	mutex_lock(&error->wedge_mutex);
+
 	GEM_TRACE("start\n");
 
 	/*
@@ -3420,7 +3432,7 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 		 */
 		if (dma_fence_default_wait(&rq->fence, true,
 					   MAX_SCHEDULE_TIMEOUT) < 0)
-			return false;
+			goto unlock;
 	}
 	i915_retire_requests(i915);
 	GEM_BUG_ON(i915->gt.active_requests);
@@ -3439,10 +3451,11 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 
 	GEM_TRACE("end\n");
 
-	smp_mb__before_atomic(); /* complete takeover before enabling execbuf */
-	clear_bit(I915_WEDGED, &i915->gpu_error.flags);
+	ret = true;
+unlock:
+	mutex_unlock(&i915->gpu_error.wedge_mutex);
 
-	return true;
+	return ret;
 }
 
 static void
@@ -5764,6 +5777,7 @@ int i915_gem_init_early(struct drm_i915_private *dev_priv)
 			  i915_gem_idle_work_handler);
 	init_waitqueue_head(&dev_priv->gpu_error.wait_queue);
 	init_waitqueue_head(&dev_priv->gpu_error.reset_queue);
+	mutex_init(&dev_priv->gpu_error.wedge_mutex);
 
 	atomic_set(&dev_priv->mm.bsd_engine_dispatch_index, 0);
 
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.h b/drivers/gpu/drm/i915/i915_gpu_error.h
index f893a4e8b783..49950b1231cf 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.h
+++ b/drivers/gpu/drm/i915/i915_gpu_error.h
@@ -267,8 +267,8 @@ struct i915_gpu_error {
 #define I915_RESET_BACKOFF	0
 #define I915_RESET_HANDOFF	1
 #define I915_RESET_MODESET	2
+#define I915_RESET_ENGINE	3
 #define I915_WEDGED		(BITS_PER_LONG - 1)
-#define I915_RESET_ENGINE	(I915_WEDGED - I915_NUM_ENGINES)
 
 	/** Number of times an engine has been reset */
 	u32 reset_engine_count[I915_NUM_ENGINES];
@@ -279,6 +279,8 @@ struct i915_gpu_error {
 	/** Reason for the current *global* reset */
 	const char *reason;
 
+	struct mutex wedge_mutex; /* serialises wedging/unwedging */
+
 	/**
 	 * Waitqueue to signal when a hang is detected. Used to for waiters
 	 * to release the struct_mutex for the reset to procede.
diff --git a/drivers/gpu/drm/i915/selftests/mock_gem_device.c b/drivers/gpu/drm/i915/selftests/mock_gem_device.c
index 43ed8b28aeaa..6beb1f47e988 100644
--- a/drivers/gpu/drm/i915/selftests/mock_gem_device.c
+++ b/drivers/gpu/drm/i915/selftests/mock_gem_device.c
@@ -186,6 +186,7 @@ struct drm_i915_private *mock_gem_device(void)
 
 	init_waitqueue_head(&i915->gpu_error.wait_queue);
 	init_waitqueue_head(&i915->gpu_error.reset_queue);
+	mutex_init(&i915->gpu_error.wedge_mutex);
 
 	i915->wq = alloc_ordered_workqueue("mock", 0);
 	if (!i915->wq)
-- 
2.18.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Chris Wilson]

Quoting Chris Wilson (2018-08-15 10:25:37)
> If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.
> 
> v2: Make sure the double set-wedged is synchronous, a parallel call
> should not return before the driver is indeed wedged.
> 
> References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>

Fwiw, I actually a landed a different version of this patch earlier.

The only advantage (esp. v3) is that we serialise the wedging more
carefully.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Chris Wilson]

If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.

v2: Make sure the double set-wedged is synchronous, a parallel call
should not return before the driver is indeed wedged.
v3: Use a real mutex (and fingers crossed that lockdep is ok!)

References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
---
 drivers/gpu/drm/i915/i915_gem.c               | 36 +++++++++++++------
 drivers/gpu/drm/i915/i915_gpu_error.h         |  4 ++-
 .../gpu/drm/i915/selftests/mock_gem_device.c  |  1 +
 3 files changed, 29 insertions(+), 12 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 0453eb42a1a3..d9e00705536d 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3300,10 +3300,15 @@ static void nop_complete_submit_request(struct i915_request *request)
 
 void i915_gem_set_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 
-	GEM_TRACE("start\n");
+	mutex_lock(&error->wedge_mutex);
+	if (test_bit(I915_WEDGED, &error->flags)) {
+		mutex_unlock(&error->wedge_mutex);
+		return;
+	}
 
 	if (GEM_SHOW_DEBUG()) {
 		struct drm_printer p = drm_debug_printer(__func__);
@@ -3312,8 +3317,7 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 			intel_engine_dump(engine, &p, "%s\n", engine->name);
 	}
 
-	if (test_and_set_bit(I915_WEDGED, &i915->gpu_error.flags))
-		goto out;
+	GEM_TRACE("start\n");
 
 	/*
 	 * First, stop submission to hw, but do not yet complete requests by
@@ -3373,19 +3377,27 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 		i915_gem_reset_finish_engine(engine);
 	}
 
-out:
+	smp_mb__before_atomic();
+	set_bit(I915_WEDGED, &error->flags);
+
 	GEM_TRACE("end\n");
+	mutex_unlock(&error->wedge_mutex);
 
-	wake_up_all(&i915->gpu_error.reset_queue);
+	wake_up_all(&error->reset_queue);
 }
 
 bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct i915_timeline *tl;
+	bool ret = false;
 
 	lockdep_assert_held(&i915->drm.struct_mutex);
-	if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
-		return true;
+
+	if (!test_bit(I915_WEDGED, &error->flags))
+		return false;
+
+	mutex_lock(&error->wedge_mutex);
 
 	GEM_TRACE("start\n");
 
@@ -3420,7 +3432,7 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 		 */
 		if (dma_fence_default_wait(&rq->fence, true,
 					   MAX_SCHEDULE_TIMEOUT) < 0)
-			return false;
+			goto unlock;
 	}
 	i915_retire_requests(i915);
 	GEM_BUG_ON(i915->gt.active_requests);
@@ -3439,10 +3451,11 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 
 	GEM_TRACE("end\n");
 
-	smp_mb__before_atomic(); /* complete takeover before enabling execbuf */
-	clear_bit(I915_WEDGED, &i915->gpu_error.flags);
+	ret = true;
+unlock:
+	mutex_unlock(&i915->gpu_error.wedge_mutex);
 
-	return true;
+	return ret;
 }
 
 static void
@@ -5764,6 +5777,7 @@ int i915_gem_init_early(struct drm_i915_private *dev_priv)
 			  i915_gem_idle_work_handler);
 	init_waitqueue_head(&dev_priv->gpu_error.wait_queue);
 	init_waitqueue_head(&dev_priv->gpu_error.reset_queue);
+	mutex_init(&dev_priv->gpu_error.wedge_mutex);
 
 	atomic_set(&dev_priv->mm.bsd_engine_dispatch_index, 0);
 
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.h b/drivers/gpu/drm/i915/i915_gpu_error.h
index f893a4e8b783..49950b1231cf 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.h
+++ b/drivers/gpu/drm/i915/i915_gpu_error.h
@@ -267,8 +267,8 @@ struct i915_gpu_error {
 #define I915_RESET_BACKOFF	0
 #define I915_RESET_HANDOFF	1
 #define I915_RESET_MODESET	2
+#define I915_RESET_ENGINE	3
 #define I915_WEDGED		(BITS_PER_LONG - 1)
-#define I915_RESET_ENGINE	(I915_WEDGED - I915_NUM_ENGINES)
 
 	/** Number of times an engine has been reset */
 	u32 reset_engine_count[I915_NUM_ENGINES];
@@ -279,6 +279,8 @@ struct i915_gpu_error {
 	/** Reason for the current *global* reset */
 	const char *reason;
 
+	struct mutex wedge_mutex; /* serialises wedging/unwedging */
+
 	/**
 	 * Waitqueue to signal when a hang is detected. Used to for waiters
 	 * to release the struct_mutex for the reset to procede.
diff --git a/drivers/gpu/drm/i915/selftests/mock_gem_device.c b/drivers/gpu/drm/i915/selftests/mock_gem_device.c
index 43ed8b28aeaa..6beb1f47e988 100644
--- a/drivers/gpu/drm/i915/selftests/mock_gem_device.c
+++ b/drivers/gpu/drm/i915/selftests/mock_gem_device.c
@@ -186,6 +186,7 @@ struct drm_i915_private *mock_gem_device(void)
 
 	init_waitqueue_head(&i915->gpu_error.wait_queue);
 	init_waitqueue_head(&i915->gpu_error.reset_queue);
+	mutex_init(&i915->gpu_error.wedge_mutex);
 
 	i915->wq = alloc_ordered_workqueue("mock", 0);
 	if (!i915->wq)
-- 
2.18.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH] drm/i915: Skip repeated calls to i915_gem_set_wedged()

[Chris Wilson]

If we already wedged, i915_gem_set_wedged() becomes a complicated no-op.

v2: Make sure the double set-wedged is synchronous, a parallel call
should not return before the driver is indeed wedged.

References: https://bugs.freedesktop.org/show_bug.cgi?id=107343
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
---
 drivers/gpu/drm/i915/i915_gem.c       | 33 ++++++++++++++++++++++-----
 drivers/gpu/drm/i915/i915_gpu_error.h |  3 ++-
 2 files changed, 29 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 0453eb42a1a3..4778d324529e 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3298,12 +3298,27 @@ static void nop_complete_submit_request(struct i915_request *request)
 	spin_unlock_irqrestore(&request->engine->timeline.lock, flags);
 }
 
+static void wait_for_wedged(struct i915_gpu_error *error)
+{
+	DEFINE_WAIT_BIT(wq_entry, &error->flags, I915_WEDGED);
+
+	__wait_on_bit(&error->reset_queue,
+		      &wq_entry, bit_wait, TASK_UNINTERRUPTIBLE);
+}
+
 void i915_gem_set_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 
-	GEM_TRACE("start\n");
+	if (test_bit(I915_WEDGED, &error->flags))
+		return;
+
+	if (test_and_set_bit(I915_WEDGE_IN_PROGRESS, &error->flags)) {
+		wait_for_wedged(error);
+		return;
+	}
 
 	if (GEM_SHOW_DEBUG()) {
 		struct drm_printer p = drm_debug_printer(__func__);
@@ -3312,8 +3327,7 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 			intel_engine_dump(engine, &p, "%s\n", engine->name);
 	}
 
-	if (test_and_set_bit(I915_WEDGED, &i915->gpu_error.flags))
-		goto out;
+	GEM_TRACE("start\n");
 
 	/*
 	 * First, stop submission to hw, but do not yet complete requests by
@@ -3373,18 +3387,25 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
 		i915_gem_reset_finish_engine(engine);
 	}
 
-out:
+	smp_mb__before_atomic();
+	set_bit(I915_WEDGED, &error->flags);
+	clear_bit(I915_WEDGE_IN_PROGRESS, &error->flags);
+
 	GEM_TRACE("end\n");
 
-	wake_up_all(&i915->gpu_error.reset_queue);
+	wake_up_all(&error->reset_queue);
 }
 
 bool i915_gem_unset_wedged(struct drm_i915_private *i915)
 {
+	struct i915_gpu_error *error = &i915->gpu_error;
 	struct i915_timeline *tl;
 
 	lockdep_assert_held(&i915->drm.struct_mutex);
-	if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
+
+	if (test_bit(I915_WEDGE_IN_PROGRESS, &error->flags))
+		wait_for_wedged(error);
+	if (!test_bit(I915_WEDGED, &error->flags))
 		return true;
 
 	GEM_TRACE("start\n");
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.h b/drivers/gpu/drm/i915/i915_gpu_error.h
index f893a4e8b783..1a78a8f330f2 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.h
+++ b/drivers/gpu/drm/i915/i915_gpu_error.h
@@ -267,8 +267,9 @@ struct i915_gpu_error {
 #define I915_RESET_BACKOFF	0
 #define I915_RESET_HANDOFF	1
 #define I915_RESET_MODESET	2
+#define I915_RESET_ENGINE	3
 #define I915_WEDGED		(BITS_PER_LONG - 1)
-#define I915_RESET_ENGINE	(I915_WEDGED - I915_NUM_ENGINES)
+#define I915_WEDGE_IN_PROGRESS	(I915_WEDGED - 1)
 
 	/** Number of times an engine has been reset */
 	u32 reset_engine_count[I915_NUM_ENGINES];
-- 
2.18.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx