4.9-stable review patch. If anyone has any objections, please let me know.
------------------
From: Lucas Stach <l.stach@xxxxxxxxxxxxxx>
commit f3cd1b064f1179d9e6188c6d67297a2360880e10 upstream.
The fence allocation needs to be protected by the GPU mutex, otherwise
the fence seqnos of concurrent submits might not match the insertion order
of the jobs in the kernel ring. This breaks the assumption that jobs
complete with monotonically increasing fence seqnos.
Fixes: d9853490176c (drm/etnaviv: take GPU lock later in the submit process)
Signed-off-by: Lucas Stach <l.stach@xxxxxxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
drivers/gpu/drm/etnaviv/etnaviv_gpu.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
--- a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
@@ -1299,6 +1299,8 @@ int etnaviv_gpu_submit(struct etnaviv_gp
goto out_pm_put;
}
+ mutex_lock(&gpu->lock);
+
fence = etnaviv_gpu_fence_alloc(gpu);
if (!fence) {
event_free(gpu, event);
@@ -1306,8 +1308,6 @@ int etnaviv_gpu_submit(struct etnaviv_gp
goto out_pm_put;
}
- mutex_lock(&gpu->lock);
-
gpu->event[event].fence = fence;
submit->fence = fence->seqno;
gpu->active_fence = submit->fence;