[PATCH] drm/i915: Rollback seqno when request creation fails

[PATCH] drm/i915: Rollback seqno when request creation fails

[Matthew Brost]

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

[Intel-gfx] [PATCH] drm/i915: Rollback seqno when request creation fails

[Matthew Brost]

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

[Intel-gfx] ✗ Fi.CI.BUILD: failure for drm/i915: Rollback seqno when request creation fails

[Patchwork]

== Series Details ==

Series: drm/i915: Rollback seqno when request creation fails
URL   : https://patchwork.freedesktop.org/series/97562/
State : failure

== Summary ==

CALL    scripts/checksyscalls.sh
  CALL    scripts/atomic/check-atomics.sh
  DESCEND objtool
  CHK     include/generated/compile.h
  CC [M]  drivers/gpu/drm/i915/gt/intel_timeline.o
drivers/gpu/drm/i915/gt/intel_timeline.c: In function ‘intel_timeline_rollback_seqno’:
drivers/gpu/drm/i915/gt/intel_timeline.c:306:2: error: implicit declaration of function ‘timeline_rollback’; did you mean ‘timeline_advance’? [-Werror=implicit-function-declaration]
  timeline_rollback(tl);
  ^~~~~~~~~~~~~~~~~
  timeline_advance
cc1: all warnings being treated as errors
scripts/Makefile.build:287: recipe for target 'drivers/gpu/drm/i915/gt/intel_timeline.o' failed
make[4]: *** [drivers/gpu/drm/i915/gt/intel_timeline.o] Error 1
scripts/Makefile.build:549: recipe for target 'drivers/gpu/drm/i915' failed
make[3]: *** [drivers/gpu/drm/i915] Error 2
scripts/Makefile.build:549: recipe for target 'drivers/gpu/drm' failed
make[2]: *** [drivers/gpu/drm] Error 2
scripts/Makefile.build:549: recipe for target 'drivers/gpu' failed
make[1]: *** [drivers/gpu] Error 2
Makefile:1846: recipe for target 'drivers' failed
make: *** [drivers] Error 2

Re: [PATCH] drm/i915: Rollback seqno when request creation fails

[kernel test robot]

Hi Matthew,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on drm-intel/for-linux-next]
[also build test ERROR on drm-tip/drm-tip drm-exynos/exynos-drm-next drm/drm-next tegra-drm/drm/tegra/for-next v5.16-rc3 next-20211203]
[cannot apply to airlied/drm-next]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Matthew-Brost/drm-i915-Rollback-seqno-when-request-creation-fails/20211204-020638
base:   git://anongit.freedesktop.org/drm-intel for-linux-next
config: i386-randconfig-m021-20211203 (https://download.01.org/0day-ci/archive/20211204/202112040802.QqhQIE5v-lkp@intel.com/config)
compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
reproduce (this is a W=1 build):
        # https://github.com/0day-ci/linux/commit/af590220a3160c7a680487eac25eb2dc24baf42d
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Matthew-Brost/drm-i915-Rollback-seqno-when-request-creation-fails/20211204-020638
        git checkout af590220a3160c7a680487eac25eb2dc24baf42d
        # save the config file to linux build tree
        mkdir build_dir
        make W=1 O=build_dir ARCH=i386 SHELL=/bin/bash drivers/gpu/drm/i915/

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   drivers/gpu/drm/i915/gt/intel_timeline.c: In function 'intel_timeline_rollback_seqno':
>> drivers/gpu/drm/i915/gt/intel_timeline.c:306:2: error: implicit declaration of function 'timeline_rollback' [-Werror=implicit-function-declaration]
     306 |  timeline_rollback(tl);
         |  ^~~~~~~~~~~~~~~~~
   cc1: some warnings being treated as errors


vim +/timeline_rollback +306 drivers/gpu/drm/i915/gt/intel_timeline.c

   303	
   304	void intel_timeline_rollback_seqno(struct intel_timeline *tl)
   305	{
 > 306		timeline_rollback(tl);
   307	}
   308	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

Re: [Intel-gfx] [PATCH] drm/i915: Rollback seqno when request creation fails

[kernel test robot]

Hi Matthew,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on drm-intel/for-linux-next]
[also build test ERROR on drm-tip/drm-tip drm-exynos/exynos-drm-next drm/drm-next tegra-drm/drm/tegra/for-next v5.16-rc3 next-20211203]
[cannot apply to airlied/drm-next]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Matthew-Brost/drm-i915-Rollback-seqno-when-request-creation-fails/20211204-020638
base:   git://anongit.freedesktop.org/drm-intel for-linux-next
config: i386-randconfig-m021-20211203 (https://download.01.org/0day-ci/archive/20211204/202112040802.QqhQIE5v-lkp@intel.com/config)
compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
reproduce (this is a W=1 build):
        # https://github.com/0day-ci/linux/commit/af590220a3160c7a680487eac25eb2dc24baf42d
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Matthew-Brost/drm-i915-Rollback-seqno-when-request-creation-fails/20211204-020638
        git checkout af590220a3160c7a680487eac25eb2dc24baf42d
        # save the config file to linux build tree
        mkdir build_dir
        make W=1 O=build_dir ARCH=i386 SHELL=/bin/bash drivers/gpu/drm/i915/

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   drivers/gpu/drm/i915/gt/intel_timeline.c: In function 'intel_timeline_rollback_seqno':
>> drivers/gpu/drm/i915/gt/intel_timeline.c:306:2: error: implicit declaration of function 'timeline_rollback' [-Werror=implicit-function-declaration]
     306 |  timeline_rollback(tl);
         |  ^~~~~~~~~~~~~~~~~
   cc1: some warnings being treated as errors


vim +/timeline_rollback +306 drivers/gpu/drm/i915/gt/intel_timeline.c

   303	
   304	void intel_timeline_rollback_seqno(struct intel_timeline *tl)
   305	{
 > 306		timeline_rollback(tl);
   307	}
   308	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

Re: [PATCH] drm/i915: Rollback seqno when request creation fails

[kernel test robot]

[-- Attachment #1: Type: text/plain, Size: 2244 bytes --]

Hi Matthew,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on drm-intel/for-linux-next]
[also build test ERROR on drm-tip/drm-tip drm-exynos/exynos-drm-next drm/drm-next tegra-drm/drm/tegra/for-next v5.16-rc3 next-20211203]
[cannot apply to airlied/drm-next]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Matthew-Brost/drm-i915-Rollback-seqno-when-request-creation-fails/20211204-020638
base:   git://anongit.freedesktop.org/drm-intel for-linux-next
config: i386-randconfig-m021-20211203 (https://download.01.org/0day-ci/archive/20211204/202112040802.QqhQIE5v-lkp(a)intel.com/config)
compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
reproduce (this is a W=1 build):
        # https://github.com/0day-ci/linux/commit/af590220a3160c7a680487eac25eb2dc24baf42d
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Matthew-Brost/drm-i915-Rollback-seqno-when-request-creation-fails/20211204-020638
        git checkout af590220a3160c7a680487eac25eb2dc24baf42d
        # save the config file to linux build tree
        mkdir build_dir
        make W=1 O=build_dir ARCH=i386 SHELL=/bin/bash drivers/gpu/drm/i915/

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   drivers/gpu/drm/i915/gt/intel_timeline.c: In function 'intel_timeline_rollback_seqno':
>> drivers/gpu/drm/i915/gt/intel_timeline.c:306:2: error: implicit declaration of function 'timeline_rollback' [-Werror=implicit-function-declaration]
     306 |  timeline_rollback(tl);
         |  ^~~~~~~~~~~~~~~~~
   cc1: some warnings being treated as errors


vim +/timeline_rollback +306 drivers/gpu/drm/i915/gt/intel_timeline.c

   303	
   304	void intel_timeline_rollback_seqno(struct intel_timeline *tl)
   305	{
 > 306		timeline_rollback(tl);
   307	}
   308	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org

[PATCH] drm/i915: Rollback seqno when request creation fails

[Matthew Brost]

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.

v2:
 (CI)
  - Fix build error

Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 drivers/gpu/drm/i915/gt/intel_timeline.c | 8 ++++++++
 drivers/gpu/drm/i915/gt/intel_timeline.h | 1 +
 drivers/gpu/drm/i915/i915_request.c      | 1 +
 3 files changed, 10 insertions(+)

diff --git a/drivers/gpu/drm/i915/gt/intel_timeline.c b/drivers/gpu/drm/i915/gt/intel_timeline.c
index 438bbc7b8147..3785e5605549 100644
--- a/drivers/gpu/drm/i915/gt/intel_timeline.c
+++ b/drivers/gpu/drm/i915/gt/intel_timeline.c
@@ -301,6 +301,14 @@ 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)
+{
+	GEM_BUG_ON(!atomic_read(&tl->pin_count));
+	GEM_BUG_ON(tl->seqno & tl->has_initial_breadcrumb);
+
+	tl->seqno -= 1 + tl->has_initial_breadcrumb;
+}
+
 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