drm/amdgpu: always emit the job vm fence
Subsystems (2)
Alex Deucher
maintainerPATCH 1/1 drm/amdgpu: always emit the job vm fence
We need the fence to reemit the gds switch or spm update after a queue reset.
--- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +4 -6
@@ -853,12 +853,10 @@ void amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, job->oa_size); } -if (vm_flush_needed || pasid_mapping_needed || cleaner_shader_needed) { -amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); - fence = &job->hw_vm_fence->base; -/* get a ref for the job */ -dma_fence_get (fence); - } +amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); + fence = &job->hw_vm_fence->base; +/* get a ref for the job */ +dma_fence_get (fence);if (vm_flush_needed) {mutex_lock (&id_mgr->lock);
Signature · 1 line
2.54.0
Hi Alex,
Quoted 7 lines · On 2026. június 3., szerda 21:45:35 közép-európai nyári idő Alex Deucher wrote:
We need the fence to reemit the gds switch or spm update after a queue reset.
Fixes a17ef941212b ("drm/amdgpu: rework ring reset backup and reemit v9")Signed-off-by Alex Deucher <[email protected]>
The patch makes good sense to me. Thanks!
Quoted 10 lines
--- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.cindex2f3470208829e..7e0e2281719b1100644
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c void amdgpu_vm_flush(struct amdgpu_ring *ring, struct · line 856
Quoted 4 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0
-1@@ -853,12 +853,10 @@ void amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, job->oa_size); } - if (vm_flush_needed || pasid_mapping_needed ||
cleaner_shader_needed) {
Quoted 12 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +4
-5-amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); - fence = &job->hw_vm_fence->base; -/* get a ref for the job */ -dma_fence_get (fence); - } +amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); + fence = &job->hw_vm_fence->base; +/* get a ref for the job */ +dma_fence_get (fence);if (vm_flush_needed) {mutex_lock (&id_mgr->lock);
Quoted 7 lines · Alex Deucher
We need the fence to reemit the gds switch or spm update after a queue reset.
Fixes a17ef941212b ("drm/amdgpu: rework ring reset backup and reemit v9")Signed-off-by Alex Deucher <[email protected]>
That was avoided because it means another entry in the EOP ring buffer which can be bad for performance.
But correctness is obviously more important, just to keep in mind when we suddenly see 1% fps decrease and don't know where it's coming from.
Quoted 25 lines
--- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +4
-6@@ -853,12 +853,10 @@ void amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, job->oa_size); } -if (vm_flush_needed || pasid_mapping_needed || cleaner_shader_needed) { -amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); - fence = &job->hw_vm_fence->base; -/* get a ref for the job */ -dma_fence_get (fence); - } +amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); + fence = &job->hw_vm_fence->base; +/* get a ref for the job */ +dma_fence_get (fence);if (vm_flush_needed) {mutex_lock (&id_mgr->lock);
Quoted 9 lines · On 2026. június 5., péntek 11:24:45 közép-európai nyári idő Christian König…
Quoted 7 lines · Alex Deucher
We need the fence to reemit the gds switch or spm update after a queue reset.
Fixes a17ef941212b ("drm/amdgpu: rework ring reset backup and reemit v9")Signed-off-by Alex Deucher <[email protected]>
Quoted 7 lines
That was avoided because it means another entry in the EOP ring buffer which can be bad for performance.
But correctness is obviously more important, just to keep in mind when we suddenly see 1% fps decrease and don't know where it's coming from.
Reviewed-by Christian König <[email protected]>
Thanks Christian, that's a valid point.
Do you think we should worry about an actual perf impact here?
If we want to avoid adding an extra fence, then an alternative solution could be to include the emitted commands in the ib_wptr of the job's own fence when a VM fence was not emitted.
What do you guys think?
Thanks,
Timur
Quoted 15 lines
Quoted 15 lines
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.cindex2f3470208829e..7e0e2281719b1100644drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0
-0@@ -853,12 +853,10 @@ void amdgpu_vm_flush(struct amdgpu_ring *ring,struct
amdgpu_job*job,>
job->oa_size);
Quoted 3 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
Quoted 3 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0
-1} - if (vm_flush_needed || pasid_mapping_needed ||
cleaner_shader_needed) {
Quoted 13 lines
Quoted 13 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +4
-5-amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); - fence = &job->hw_vm_fence->base; -/* get a ref for the job */ -dma_fence_get (fence); - } +amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); + fence = &job->hw_vm_fence->base; +/* get a ref for the job */ +dma_fence_get (fence);if (vm_flush_needed) {mutex_lock (&id_mgr->lock);
Quoted 20 lines · Timur Kristóf
Quoted 9 lines · On 2026. június 5., péntek 11:24:45 közép-európai nyári idő Christian König…
Quoted 7 lines · Alex Deucher
We need the fence to reemit the gds switch or spm update after a queue reset.
Fixes a17ef941212b ("drm/amdgpu: rework ring reset backup and reemit v9")Signed-off-by Alex Deucher <[email protected]>Quoted 7 lines
That was avoided because it means another entry in the EOP ring buffer which can be bad for performance.
But correctness is obviously more important, just to keep in mind when we suddenly see 1% fps decrease and don't know where it's coming from.
Reviewed-by Christian König <[email protected]>
Quoted 8 lines
Thanks Christian, that's a valid point.
Do you think we should worry about an actual perf impact here?If we want to avoid adding an extra fence, then an alternative solution could be to include the emitted commands in the
ib_wptrof the job's own fence when a VM fence was not emitted.What do you guys think?
@Christian Koenig Is there a reason we need to emit gds and spm before the vm fence if the vm fence is required? I've sent out a series to retain that behavior, but if it doesn't matter, we could simplify things more and move gds and spm into the ib fence.
Alex
Quoted 40 lines
Thanks,
TimurQuoted 15 lines
Quoted 15 lines
---
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-)(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.cindex2f3470208829e..7e0e2281719b1100644drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0
-0@@ -853,12 +853,10 @@ void amdgpu_vm_flush(struct amdgpu_ring *ring,struct
amdgpu_job*job,>job->oa_size);Quoted 3 lines
Quoted 3 lines
}- if (
vm_flush_needed||pasid_mapping_needed||
cleaner_shader_needed) {Quoted 13 lines
Quoted 13 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +4
-5-amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); - fence = &job->hw_vm_fence->base; -/* get a ref for the job */ -dma_fence_get (fence); - } +amdgpu_fence_emit (ring, job->hw_vm_fence,0 ); + fence = &job->hw_vm_fence->base; +/* get a ref for the job */ +dma_fence_get (fence);if (vm_flush_needed) {mutex_lock (&id_mgr->lock);