Skip to contentSkip to conversation

drm/i915: Use seqlock in engine stats

CI Tvrtko Ursulin 3 participants 4 messages Apr 26, 2018

Subsystems (2)
lore
Share & tools
lore

Tvrtko Ursulin

author maintainer
#0 lore raw

From: Tvrtko Ursulin <[email protected]>

We can convert engine stats from a spinlock to seqlock to ensure interrupt processing is never even a tiny bit delayed by parallel readers.

There is a smidgen bit more cost on the write lock side, and an extremely unlikely chance that readers will have to retry a few times in face of heavy interrupt load. But it should be extremely unlikely given how lightweight read side section is compared to the interrupt processing side, and also compared to the rest of the code paths which can lead into it. Furthermore, writer is the ones doing the real, latency sensitive work, while readers are only informative.

---
 drivers/gpu/drm/i915/intel_engine_cs.c  | 19 ++++++++++---------
 drivers/gpu/drm/i915/intel_ringbuffer.h | 11 ++++++-----
 2 files changed, 16 insertions(+), 14 deletions(-)
drivers/gpu/drm/i915/intel_engine_cs.c +10 -9
index 58be7fac5b8c..ac009f10c948 100644
@@ -306,7 +306,7 @@ intel_engine_setup(struct drm_i915_private *dev_priv,
 	/* Nothing to do here, execute in order of dependencies */
 	engine->schedule = NULL;
 
-	spin_lock_init(&engine->stats.lock);
+	seqlock_init(&engine->stats.lock);
 
 	ATOMIC_INIT_NOTIFIER_HEAD(&engine->context_status_notifier);
 
@@ -1481,7 +1481,7 @@ int intel_enable_engine_stats(struct intel_engine_cs *engine)
 		return -ENODEV;
 
 	tasklet_disable(&execlists->tasklet);
-	spin_lock_irqsave(&engine->stats.lock, flags);
+	write_seqlock_irqsave(&engine->stats.lock, flags);
 
 	if (unlikely(engine->stats.enabled == ~0)) {
 		err = -EBUSY;
@@ -1505,7 +1505,7 @@ int intel_enable_engine_stats(struct intel_engine_cs *engine)
 	}
 
 unlock:
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	write_sequnlock_irqrestore(&engine->stats.lock, flags);
 	tasklet_enable(&execlists->tasklet);
 
 	return err;
@@ -1534,12 +1534,13 @@ static ktime_t __intel_engine_get_busy_time(struct intel_engine_cs *engine)
  */
 ktime_t intel_engine_get_busy_time(struct intel_engine_cs *engine)
 {
+	unsigned int seq;
 	ktime_t total;
-	unsigned long flags;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
-	total = __intel_engine_get_busy_time(engine);
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	do {
+		seq = read_seqbegin(&engine->stats.lock);
+		total = __intel_engine_get_busy_time(engine);
+	} while (read_seqretry(&engine->stats.lock, seq));
 
 	return total;
 }
@@ -1557,13 +1558,13 @@ void intel_disable_engine_stats(struct intel_engine_cs *engine)
 	if (!intel_engine_supports_stats(engine))
 		return;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
+	write_seqlock_irqsave(&engine->stats.lock, flags);
 	WARN_ON_ONCE(engine->stats.enabled == 0);
 	if (--engine->stats.enabled == 0) {
 		engine->stats.total = __intel_engine_get_busy_time(engine);
 		engine->stats.active = 0;
 	}
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	write_sequnlock_irqrestore(&engine->stats.lock, flags);
 }
 
 #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
drivers/gpu/drm/i915/intel_ringbuffer.h +6 -5
index c5e27905b0e1..24af3f1088ba 100644
@@ -3,6 +3,7 @@
 #define _INTEL_RINGBUFFER_H_
 
 #include <linux/hashtable.h>
+#include <linux/seqlock.h>
 
 #include "i915_gem_batch_pool.h"
 #include "i915_gem_timeline.h"
@@ -595,7 +596,7 @@ struct intel_engine_cs {
 		/**
 		 * @lock: Lock protecting the below fields.
 		 */
-		spinlock_t lock;
+		seqlock_t lock;
 		/**
 		 * @enabled: Reference count indicating number of listeners.
 		 */
@@ -1064,7 +1065,7 @@ static inline void intel_engine_context_in(struct intel_engine_cs *engine)
 	if (READ_ONCE(engine->stats.enabled) == 0)
 		return;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
+	write_seqlock_irqsave(&engine->stats.lock, flags);
 
 	if (engine->stats.enabled > 0) {
 		if (engine->stats.active++ == 0)
@@ -1072,7 +1073,7 @@ static inline void intel_engine_context_in(struct intel_engine_cs *engine)
 		GEM_BUG_ON(engine->stats.active == 0);
 	}
 
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	write_sequnlock_irqrestore(&engine->stats.lock, flags);
 }
 
 static inline void intel_engine_context_out(struct intel_engine_cs *engine)
@@ -1082,7 +1083,7 @@ static inline void intel_engine_context_out(struct intel_engine_cs *engine)
 	if (READ_ONCE(engine->stats.enabled) == 0)
 		return;
 
-	spin_lock_irqsave(&engine->stats.lock, flags);
+	write_seqlock_irqsave(&engine->stats.lock, flags);
 
 	if (engine->stats.enabled > 0) {
 		ktime_t last;
@@ -1109,7 +1110,7 @@ static inline void intel_engine_context_out(struct intel_engine_cs *engine)
 		}
 	}
 
-	spin_unlock_irqrestore(&engine->stats.lock, flags);
+	write_sequnlock_irqrestore(&engine->stats.lock, flags);
 }
 
 int intel_enable_engine_stats(struct intel_engine_cs *engine);
Signature · 6 lines
2.14.1

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

✓ Fi.CI.BAT: success for drm/i915: Use seqlock in engine stats (rev3)

== Series Details ==

== Series Details ==

Series: drm/i915: Use seqlock in engine stats (rev3)
URL : https://patchwork.freedesktop.org/series/38347/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4100 -> Patchwork_8806 =

== Summary - SUCCESS ==

No regressions found.

External URL: https://patchwork.freedesktop.org/api/1.0/series/38347/revisions/3/mbox/

== Known issues ==

Here are the changes found in Patchwork_8806 that come from known issues:

=== IGT changes ===

    ==== Possible fixes ====

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

    igt@kms_flip@basic-flip-vs-dpms:
      fi-bxt-dsi:         INCOMPLETE (fdo#103927) -> PASS

fdo#103713 https://bugs.freedesktop.org/show_bug.cgi?id=103713
fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927

== Participating hosts (37 -> 35) ==

Missing (2): fi-ilk-m540 fi-skl-6700hq

== Build changes ==

* Linux: CI_DRM_4100 -> Patchwork_8806

CI_DRM_4100: 05cc701ce9d22b01f2f4afa9fddd521b931ed163 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4449: 0350f0e7f6a0e07281445fc3082aa70419f4aac7 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_8806: 9ce2e374546c8bd06ae6e97f8b5b7b6cdf8e7f0e @ git://anongit.freedesktop.org/gfx-ci/linux
piglit_4449: ad8992d3fb27fd604b9ab15e7963c42421ced85c @ git://anongit.freedesktop.org/piglit

== Linux commits ==

9ce2e374546c drm/i915: Use seqlock in engine stats

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_8806/issues.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#3 lore raw

✓ Fi.CI.IGT: success for drm/i915: Use seqlock in engine stats (rev3)

In reply to Tvrtko Ursulin
== Series Details ==

== Series Details ==

Series: drm/i915: Use seqlock in engine stats (rev3)
URL : https://patchwork.freedesktop.org/series/38347/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4100_full -> Patchwork_8806_full =

== Summary - WARNING ==

Minor unknown changes coming with Patchwork_8806_full need to be verified manually.

If you think the reported changes have nothing to do with the changes introduced in Patchwork_8806_full, 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/38347/revisions/3/mbox/

== Possible new issues ==

Here are the unknown changes that may have been introduced in Patchwork_8806_full:

=== IGT changes ===

    ==== Warnings ====

    igt@gem_exec_schedule@deep-render:
      shard-kbl:          SKIP -> PASS

== Known issues ==

Here are the changes found in Patchwork_8806_full that come from known issues:

=== IGT changes ===

    ==== Issues hit ====

    igt@kms_cursor_legacy@2x-flip-vs-cursor-atomic:
      shard-hsw:          PASS -> FAIL (fdo#104873)

    igt@kms_flip@2x-plain-flip-fb-recreate:
      shard-hsw:          PASS -> FAIL (fdo#100368)

    igt@kms_flip@modeset-vs-vblank-race:
      shard-glk:          PASS -> FAIL (fdo#103060) +1

    igt@kms_frontbuffer_tracking@fbc-tilingchange:
      shard-apl:          PASS -> FAIL (fdo#103167)

    igt@kms_sysfs_edid_timing:
      shard-apl:          PASS -> WARN (fdo#100047)

    
    ==== Possible fixes ====

    igt@kms_cursor_legacy@2x-long-flip-vs-cursor-legacy:
      shard-hsw:          FAIL (fdo#104873) -> PASS

    igt@kms_flip@absolute-wf_vblank-interruptible:
      shard-glk:          FAIL (fdo#106087) -> PASS

    igt@kms_flip@dpms-vs-vblank-race:
      shard-glk:          FAIL (fdo#103060) -> PASS

    igt@kms_flip@dpms-vs-vblank-race-interruptible:
      shard-apl:          FAIL (fdo#103060) -> PASS

    igt@kms_flip@flip-vs-wf_vblank-interruptible:
      shard-hsw:          FAIL (fdo#103928) -> PASS
      shard-glk:          FAIL (fdo#100368) -> PASS +1
      shard-apl:          FAIL (fdo#105312) -> PASS

    igt@kms_flip@plain-flip-fb-recreate:
      shard-hsw:          FAIL (fdo#100368) -> PASS

    igt@kms_frontbuffer_tracking@fbc-1p-primscrn-cur-indfb-draw-render:
      shard-apl:          FAIL (fdo#103167) -> PASS

fdo#100047 https://bugs.freedesktop.org/show_bug.cgi?id=100047
fdo#100368 https://bugs.freedesktop.org/show_bug.cgi?id=100368
fdo#103060 https://bugs.freedesktop.org/show_bug.cgi?id=103060
fdo#103167 https://bugs.freedesktop.org/show_bug.cgi?id=103167
fdo#103928 https://bugs.freedesktop.org/show_bug.cgi?id=103928
fdo#104873 https://bugs.freedesktop.org/show_bug.cgi?id=104873
fdo#105312 https://bugs.freedesktop.org/show_bug.cgi?id=105312
fdo#106087 https://bugs.freedesktop.org/show_bug.cgi?id=106087

== Participating hosts (6 -> 5) ==

Missing (1): shard-glkb

== Build changes ==

* Linux: CI_DRM_4100 -> Patchwork_8806

CI_DRM_4100: 05cc701ce9d22b01f2f4afa9fddd521b931ed163 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_4449: 0350f0e7f6a0e07281445fc3082aa70419f4aac7 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
Patchwork_8806: 9ce2e374546c8bd06ae6e97f8b5b7b6cdf8e7f0e @ git://anongit.freedesktop.org/gfx-ci/linux
piglit_4449: ad8992d3fb27fd604b9ab15e7963c42421ced85c @ git://anongit.freedesktop.org/piglit

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_8806/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#2 lore raw

Re: ✓ Fi.CI.BAT: success for drm/i915: Use seqlock in engine stats (rev3)

In reply to Patchwork
Quoted 50 lines · Patchwork

== Series Details ==

Series: drm/i915: Use seqlock in engine stats (rev3)
URL : https://patchwork.freedesktop.org/series/38347/
State : success

== Summary ==

= CI Bug Log - changes from CI_DRM_4100 -> Patchwork_8806 =

== Summary - SUCCESS ==

No regressions found.

   External URL: https://patchwork.freedesktop.org/api/1.0/series/38347/revisions/3/mbox/

== Known issues ==

   Here are the changes found in Patchwork_8806 that come from known issues:

   === IGT changes ===

     ==== Possible fixes ====

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

     igt@kms_flip@basic-flip-vs-dpms:
       fi-bxt-dsi:         INCOMPLETE (fdo#103927) -> PASS

     
   fdo#103713 https://bugs.freedesktop.org/show_bug.cgi?id=103713
   fdo#103927 https://bugs.freedesktop.org/show_bug.cgi?id=103927

== Participating hosts (37 -> 35) ==

   Missing    (2): fi-ilk-m540 fi-skl-6700hq

== Build changes ==

* Linux: CI_DRM_4100 -> Patchwork_8806

   CI_DRM_4100: 05cc701ce9d22b01f2f4afa9fddd521b931ed163 @ git://anongit.freedesktop.org/gfx-ci/linux
   IGT_4449: 0350f0e7f6a0e07281445fc3082aa70419f4aac7 @ git://anongit.freedesktop.org/xorg/app/intel-gpu-tools
   Patchwork_8806: 9ce2e374546c8bd06ae6e97f8b5b7b6cdf8e7f0e @ git://anongit.freedesktop.org/gfx-ci/linux
   piglit_4449: ad8992d3fb27fd604b9ab15e7963c42421ced85c @ git://anongit.freedesktop.org/piglit
Quoted 3 lines

== Linux commits ==

9ce2e374546c drm/i915: Use seqlock in engine stats

Pushed, thanks for the suggestion and review!

Regards,

Tvrtko

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