Intel-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
* [Intel-gfx] [PATCH] drm/i915/reset: Handle reset timeouts under unrelated kernel hangs
@ 2022-06-28 19:17 Ashutosh Dixit
  2022-06-28 22:58 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for " Patchwork
                   ` (2 more replies)
  0 siblings, 3 replies; 5+ messages in thread
From: Ashutosh Dixit @ 2022-06-28 19:17 UTC (permalink / raw)
  To: intel-gfx; +Cc: Chris Wilson, dri-devel, Chris Wilson

From: Chris Wilson <chris@chris-wilson.co.uk>

When resuming after hibernate sometimes we see hangs in unrelated kernel
subsystems. These hangs often result in the following i915 trace:

i915 0000:00:02.0: [drm] \
	*ERROR* intel_gt_reset_global timed out, cancelling all in-flight rendering.

implying our reset task has been starved by the hanging kernel subsystem,
causing us to inappropiately declare the system as wedged beyond recovery.

The trace would be caused by our synchronize_srcu_expedited() taking more
than the allowed 5s due to the unrelated kernel hang. But we neither need
to perform that synchronisation inside the reset watchdog, nor do we need
such a short timeout before declaring the device as unrecoverable.

Bug: https://gitlab.freedesktop.org/drm/intel/-/issues/3575
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
---
 drivers/gpu/drm/i915/gt/intel_reset.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/intel_reset.c b/drivers/gpu/drm/i915/gt/intel_reset.c
index a5338c3fde7a0..e72744f6faedc 100644
--- a/drivers/gpu/drm/i915/gt/intel_reset.c
+++ b/drivers/gpu/drm/i915/gt/intel_reset.c
@@ -1259,12 +1259,9 @@ static void intel_gt_reset_global(struct intel_gt *gt,
 	kobject_uevent_env(kobj, KOBJ_CHANGE, reset_event);
 
 	/* Use a watchdog to ensure that our reset completes */
-	intel_wedge_on_timeout(&w, gt, 5 * HZ) {
+	intel_wedge_on_timeout(&w, gt, 60 * HZ) {
 		intel_display_prepare_reset(gt->i915);
 
-		/* Flush everyone using a resource about to be clobbered */
-		synchronize_srcu_expedited(&gt->reset.backoff_srcu);
-
 		intel_gt_reset(gt, engine_mask, reason);
 
 		intel_display_finish_reset(gt->i915);
@@ -1373,6 +1370,9 @@ void intel_gt_handle_error(struct intel_gt *gt,
 		}
 	}
 
+	/* Flush everyone using a resource about to be clobbered */
+	synchronize_srcu_expedited(&gt->reset.backoff_srcu);
+
 	intel_gt_reset_global(gt, engine_mask, msg);
 
 	if (!intel_uc_uses_guc_submission(&gt->uc)) {
-- 
2.36.1


^ permalink raw reply related	[flat|nested] 5+ messages in thread
* [Intel-gfx] [PATCH] drm/i915/reset: Handle reset timeouts under unrelated kernel hangs
@ 2022-06-30  4:39 Ashutosh Dixit
  0 siblings, 0 replies; 5+ messages in thread
From: Ashutosh Dixit @ 2022-06-30  4:39 UTC (permalink / raw)
  To: intel-gfx; +Cc: Chris Wilson, dri-devel

From: Chris Wilson <chris@chris-wilson.co.uk>

When resuming after hibernate sometimes we see hangs in unrelated kernel
subsystems. These hangs often result in the following i915 trace:

i915 0000:00:02.0: [drm] *ERROR* \
	intel_gt_reset_global timed out, cancelling all in-flight rendering

implying our reset task has been starved by the hanging kernel subsystem,
causing us to inappropiately declare the system as wedged beyond recovery.

The trace would be caused by our synchronize_srcu_expedited() taking more
than the allowed 5s due to the unrelated kernel hang. But we neither need
to perform that synchronisation inside the reset watchdog, nor do we need
such a short timeout before declaring the device as unrecoverable.

v2: Restore watchdog timeout to the previous 5 seconds (Ashutosh)

Bug: https://gitlab.freedesktop.org/drm/intel/-/issues/3575
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
---
 drivers/gpu/drm/i915/gt/intel_reset.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/intel_reset.c b/drivers/gpu/drm/i915/gt/intel_reset.c
index a5338c3fde7a..1cbe65a5b0fd 100644
--- a/drivers/gpu/drm/i915/gt/intel_reset.c
+++ b/drivers/gpu/drm/i915/gt/intel_reset.c
@@ -1262,9 +1262,6 @@ static void intel_gt_reset_global(struct intel_gt *gt,
 	intel_wedge_on_timeout(&w, gt, 5 * HZ) {
 		intel_display_prepare_reset(gt->i915);
 
-		/* Flush everyone using a resource about to be clobbered */
-		synchronize_srcu_expedited(&gt->reset.backoff_srcu);
-
 		intel_gt_reset(gt, engine_mask, reason);
 
 		intel_display_finish_reset(gt->i915);
@@ -1373,6 +1370,9 @@ void intel_gt_handle_error(struct intel_gt *gt,
 		}
 	}
 
+	/* Flush everyone using a resource about to be clobbered */
+	synchronize_srcu_expedited(&gt->reset.backoff_srcu);
+
 	intel_gt_reset_global(gt, engine_mask, msg);
 
 	if (!intel_uc_uses_guc_submission(&gt->uc)) {
-- 
2.34.1


^ permalink raw reply related	[flat|nested] 5+ messages in thread
end of thread, other threads:[~2022-06-30  4:40 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2022-06-28 19:17 [Intel-gfx] [PATCH] drm/i915/reset: Handle reset timeouts under unrelated kernel hangs Ashutosh Dixit
2022-06-28 22:58 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for " Patchwork
2022-06-28 23:21 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork
2022-06-29  5:35 ` [Intel-gfx] [PATCH] " Dixit, Ashutosh
  -- strict thread matches above, loose matches on Subject: below --
2022-06-30  4:39 Ashutosh Dixit

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox