All of lore.kernel.org
 help / color / mirror / Atom feed
From: Matthew Brost <matthew.brost@intel.com>
To: <intel-gfx@lists.freedesktop.org>, <dri-devel@lists.freedesktop.org>
Cc: daniele.ceraolospurio@intel.com, john.c.harrison@intel.com
Subject: [PATCH] drm/i915: Rollback seqno when request creation fails
Date: Fri,  3 Dec 2021 09:59:10 -0800	[thread overview]
Message-ID: <20211203175910.28516-1-matthew.brost@intel.com> (raw)

gem_ctx_create.basic-files can slam on kernel contexts to the extent
where request creation fails because the ring is full. When this happens
seqno numbers are skipped which can result the below GEM_BUG_ON blowing
in gt/intel_engine_pm.c:__engine_unpark:

GEM_BUG_ON(ce->timeline->seqno !=
           READ_ONCE(*ce->timeline->hwsp_seqno));

Fixup request creation code to roll back seqno when request creation
fails.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 drivers/gpu/drm/i915/gt/intel_timeline.c | 5 +++++
 drivers/gpu/drm/i915/gt/intel_timeline.h | 1 +
 drivers/gpu/drm/i915/i915_request.c      | 1 +
 3 files changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/i915/gt/intel_timeline.c b/drivers/gpu/drm/i915/gt/intel_timeline.c
index 438bbc7b8147..64ea9a90c7a0 100644
--- a/drivers/gpu/drm/i915/gt/intel_timeline.c
+++ b/drivers/gpu/drm/i915/gt/intel_timeline.c
@@ -301,6 +301,11 @@ static u32 timeline_advance(struct intel_timeline *tl)
 	return tl->seqno += 1 + tl->has_initial_breadcrumb;
 }
 
+void intel_timeline_rollback_seqno(struct intel_timeline *tl)
+{
+	timeline_rollback(tl);
+}
+
 static noinline int
 __intel_timeline_get_seqno(struct intel_timeline *tl,
 			   u32 *seqno)
diff --git a/drivers/gpu/drm/i915/gt/intel_timeline.h b/drivers/gpu/drm/i915/gt/intel_timeline.h
index 57308c4d664a..a2f2e0ea186f 100644
--- a/drivers/gpu/drm/i915/gt/intel_timeline.h
+++ b/drivers/gpu/drm/i915/gt/intel_timeline.h
@@ -72,6 +72,7 @@ void intel_timeline_enter(struct intel_timeline *tl);
 int intel_timeline_get_seqno(struct intel_timeline *tl,
 			     struct i915_request *rq,
 			     u32 *seqno);
+void intel_timeline_rollback_seqno(struct intel_timeline *tl);
 void intel_timeline_exit(struct intel_timeline *tl);
 void intel_timeline_unpin(struct intel_timeline *tl);
 
diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c
index a72c8f0346a0..86f32ee082f7 100644
--- a/drivers/gpu/drm/i915/i915_request.c
+++ b/drivers/gpu/drm/i915/i915_request.c
@@ -966,6 +966,7 @@ __i915_request_create(struct intel_context *ce, gfp_t gfp)
 
 err_unwind:
 	ce->ring->emit = rq->head;
+	intel_timeline_rollback_seqno(tl);
 
 	/* Make sure we didn't add ourselves to external state before freeing */
 	GEM_BUG_ON(!list_empty(&rq->sched.signalers_list));
-- 
2.33.1


WARNING: multiple messages have this Message-ID (diff)
From: Matthew Brost <matthew.brost@intel.com>
To: <intel-gfx@lists.freedesktop.org>, <dri-devel@lists.freedesktop.org>
Subject: [Intel-gfx] [PATCH] drm/i915: Rollback seqno when request creation fails
Date: Fri,  3 Dec 2021 09:59:10 -0800	[thread overview]
Message-ID: <20211203175910.28516-1-matthew.brost@intel.com> (raw)

gem_ctx_create.basic-files can slam on kernel contexts to the extent
where request creation fails because the ring is full. When this happens
seqno numbers are skipped which can result the below GEM_BUG_ON blowing
in gt/intel_engine_pm.c:__engine_unpark:

GEM_BUG_ON(ce->timeline->seqno !=
           READ_ONCE(*ce->timeline->hwsp_seqno));

Fixup request creation code to roll back seqno when request creation
fails.

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 drivers/gpu/drm/i915/gt/intel_timeline.c | 5 +++++
 drivers/gpu/drm/i915/gt/intel_timeline.h | 1 +
 drivers/gpu/drm/i915/i915_request.c      | 1 +
 3 files changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/i915/gt/intel_timeline.c b/drivers/gpu/drm/i915/gt/intel_timeline.c
index 438bbc7b8147..64ea9a90c7a0 100644
--- a/drivers/gpu/drm/i915/gt/intel_timeline.c
+++ b/drivers/gpu/drm/i915/gt/intel_timeline.c
@@ -301,6 +301,11 @@ static u32 timeline_advance(struct intel_timeline *tl)
 	return tl->seqno += 1 + tl->has_initial_breadcrumb;
 }
 
+void intel_timeline_rollback_seqno(struct intel_timeline *tl)
+{
+	timeline_rollback(tl);
+}
+
 static noinline int
 __intel_timeline_get_seqno(struct intel_timeline *tl,
 			   u32 *seqno)
diff --git a/drivers/gpu/drm/i915/gt/intel_timeline.h b/drivers/gpu/drm/i915/gt/intel_timeline.h
index 57308c4d664a..a2f2e0ea186f 100644
--- a/drivers/gpu/drm/i915/gt/intel_timeline.h
+++ b/drivers/gpu/drm/i915/gt/intel_timeline.h
@@ -72,6 +72,7 @@ void intel_timeline_enter(struct intel_timeline *tl);
 int intel_timeline_get_seqno(struct intel_timeline *tl,
 			     struct i915_request *rq,
 			     u32 *seqno);
+void intel_timeline_rollback_seqno(struct intel_timeline *tl);
 void intel_timeline_exit(struct intel_timeline *tl);
 void intel_timeline_unpin(struct intel_timeline *tl);
 
diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c
index a72c8f0346a0..86f32ee082f7 100644
--- a/drivers/gpu/drm/i915/i915_request.c
+++ b/drivers/gpu/drm/i915/i915_request.c
@@ -966,6 +966,7 @@ __i915_request_create(struct intel_context *ce, gfp_t gfp)
 
 err_unwind:
 	ce->ring->emit = rq->head;
+	intel_timeline_rollback_seqno(tl);
 
 	/* Make sure we didn't add ourselves to external state before freeing */
 	GEM_BUG_ON(!list_empty(&rq->sched.signalers_list));
-- 
2.33.1


             reply	other threads:[~2021-12-03 18:05 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-03 17:59 Matthew Brost [this message]
2021-12-03 17:59 ` [Intel-gfx] [PATCH] drm/i915: Rollback seqno when request creation fails Matthew Brost
2021-12-03 22:14 ` [Intel-gfx] ✗ Fi.CI.BUILD: failure for " Patchwork
2021-12-04  0:50 ` [PATCH] " kernel test robot
2021-12-04  0:50   ` kernel test robot
2021-12-04  0:50   ` [Intel-gfx] " kernel test robot
2021-12-03 23:51 Matthew Brost

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20211203175910.28516-1-matthew.brost@intel.com \
    --to=matthew.brost@intel.com \
    --cc=daniele.ceraolospurio@intel.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=john.c.harrison@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
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.