Am 16.11.21 um 19:30 schrieb Amit Pundir:
On Tue, 16 Nov 2021 at 21:21, Rob Clark <robdclark@xxxxxxxxx> wrote:
From: Rob Clark <robdclark@xxxxxxxxxxxx>
drm_sched_job_add_dependency() could drop the last ref, so we need to do
the dma_fence_get() first.
It fixed the splats I saw on RB5 (sm8250 | A650). Thanks.
Tested-by: Amit Pundir <amit.pundir@xxxxxxxxxx>
I've added my rb, pushed this with the original fix to drm-misc-fixes
and cleaned up the obvious fallout between drm-misc-fixes and
drm-misc-next in drm-tip.
Thanks for the help and sorry for the noise,
Christian.
Cc: Christian König <christian.koenig@xxxxxxx>
Fixes: 9c2ba265352a drm/scheduler: ("use new iterator in drm_sched_job_add_implicit_dependencies v2")
Signed-off-by: Rob Clark <robdclark@xxxxxxxxxxxx>
---
Applies on top of "drm/scheduler: fix drm_sched_job_add_implicit_dependencies"
but I don't think that has a stable commit sha yet.
drivers/gpu/drm/scheduler/sched_main.c | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
index 94fe51b3caa2..f91fb31ab7a7 100644
--- a/drivers/gpu/drm/scheduler/sched_main.c
+++ b/drivers/gpu/drm/scheduler/sched_main.c
@@ -704,12 +704,13 @@ int drm_sched_job_add_implicit_dependencies(struct drm_sched_job *job,
int ret;
dma_resv_for_each_fence(&cursor, obj->resv, write, fence) {
- ret = drm_sched_job_add_dependency(job, fence);
- if (ret)
- return ret;
-
/* Make sure to grab an additional ref on the added fence */
dma_fence_get(fence);
+ ret = drm_sched_job_add_dependency(job, fence);
+ if (ret) {
+ dma_fence_put(fence);
+ return ret;
+ }
}
return 0;
}
--
2.33.1
