When we backup ring contents to reemit after a queue reset,
we don't backup ring contents from the bad context. When
we signal the fences, we should set an error on those
fences as well.
Fixes: 77cc0da39c7c ("drm/amdgpu: track ring state associated with a fence")
Signed-off-by: Alex Deucher <alexander.deuc...@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c | 34 ++++++++++++++++-------
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c | 4 +--
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h | 3 +-
3 files changed, 28 insertions(+), 13 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
index fd8cca241da62..1b689a4226291 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@@ -752,17 +752,31 @@ void amdgpu_fence_driver_force_completion(struct
amdgpu_ring *ring)
* which data needs to be saved out of the queue's ring buffer.
*/
-/**
- * amdgpu_fence_driver_guilty_force_completion - force signal of specified
sequence
- *
- * @fence: fence of the ring to signal
- *
- */
-void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *fence)
+void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_ring *ring,
+ struct amdgpu_fence
*guilty_fence)
{
- dma_fence_set_error(&fence->base, -ETIME);
- amdgpu_fence_write(fence->ring, fence->seq);
- amdgpu_fence_process(fence->ring);
+ struct dma_fence *unprocessed;
+ struct dma_fence __rcu **ptr;
+ struct amdgpu_fence *fence;
+ u64 i, seqno;
+
+ seqno = amdgpu_fence_read(ring);
+
+ for (i = seqno + 1; i <= ring->fence_drv.sync_seq; ++i) {
+ ptr = &ring->fence_drv.fences[i &
ring->fence_drv.num_fences_mask];
+ rcu_read_lock();
+ unprocessed = rcu_dereference(*ptr);
+
+ if (unprocessed && !dma_fence_is_signaled(unprocessed)) {
+ fence = container_of(unprocessed, struct amdgpu_fence,
base);
+
+ if (fence->context == guilty_fence->context)
+ dma_fence_set_error(&fence->base, -ETIME);
+ }
+ rcu_read_unlock();
+ }
+ amdgpu_fence_write(ring, guilty_fence->seq);
+ amdgpu_fence_process(ring);
}
void amdgpu_fence_save_wptr(struct dma_fence *fence)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
index 6379bb25bf5ce..725d6437fe8e3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@@ -812,9 +812,9 @@ int amdgpu_ring_reset_helper_end(struct amdgpu_ring *ring,
if (r)
return r;
- /* signal the fence of the bad job */
+ /* signal the fences of the bad job */
if (guilty_fence)
- amdgpu_fence_driver_guilty_force_completion(guilty_fence);
+ amdgpu_fence_driver_guilty_force_completion(ring, guilty_fence);
/* Re-emit the non-guilty commands */
if (ring->ring_backup_entries_to_copy) {
amdgpu_ring_alloc_reemit(ring,
ring->ring_backup_entries_to_copy);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
index 7670f5d82b9e4..7a362ce8435fc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
@@ -155,7 +155,8 @@ extern const struct drm_sched_backend_ops amdgpu_sched_ops;
void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring);
void amdgpu_fence_driver_set_error(struct amdgpu_ring *ring, int error);
void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring);
-void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *fence);
+void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_ring *ring,
+ struct amdgpu_fence *fence);
void amdgpu_fence_save_wptr(struct dma_fence *fence);
int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring);
--
2.51.0
