drm/amdgpu: hold reference to fences in amdgpu_sa_bo_new (v2)

An arbitrary amount of time can pass between spin_unlock and
fence_wait_any_timeout, so we need to ensure that nobody frees the
fences from under us.

A stress test (rapidly starting and killing hundreds of glxgears
instances) ran into a deadlock in fence_wait_any_timeout after
about an hour, and this race condition appears to be a plausible
cause.

v2: agd: rebase on upstream

Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org

1 files changed, 4 insertions, 1 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
index 8b88edb0434b..ca72a2e487b9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
@@ -354,12 +354,15 @@ int amdgpu_sa_bo_new(struct amdgpu_sa_manager *sa_manager,
 
                 for (i = 0, count = 0; i < AMDGPU_MAX_RINGS; ++i)
                         if (fences[i])
-                                fences[count++] = fences[i];
+                                fences[count++] = fence_get(fences[i]);
 
                 if (count) {
                         spin_unlock(&sa_manager->wq.lock);
                         t = fence_wait_any_timeout(fences, count, false,
                                                    MAX_SCHEDULE_TIMEOUT);
+                        for (i = 0; i < count; ++i)
+                                fence_put(fences[i]);
+
                         r = (t > 0) ? 0 : t;
                         spin_lock(&sa_manager->wq.lock);
                 } else {