On 4/30/25 04:40, Prike Liang wrote:
> The dma_resv_add_fence() already refers to the added fence.
> So when attaching the evciton fence to the gem bo, it needn't
> refer to it anymore.
>
> Signed-off-by: Prike Liang <prike.li...@amd.com>
> Reviewed-by: Christian König <christian.koe...@amd.com>
This is a bug fix and as such should always come as first patch in a series.
Please make sure to commit this one to amd-staging-drm-next ASAP.
Regards,
Christian.
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c | 7 ++-----
> 1 file changed, 2 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c
> index 6c9b2b43a929..7a5f02ef45a7 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c
> @@ -189,7 +189,6 @@ void amdgpu_eviction_fence_destroy(struct
> amdgpu_eviction_fence_mgr *evf_mgr)
> int amdgpu_eviction_fence_attach(struct amdgpu_eviction_fence_mgr *evf_mgr,
> struct amdgpu_bo *bo)
> {
> - struct dma_fence *ef;
> struct amdgpu_eviction_fence *ev_fence;
> struct dma_resv *resv = bo->tbo.base.resv;
> int ret;
> @@ -205,10 +204,8 @@ int amdgpu_eviction_fence_attach(struct
> amdgpu_eviction_fence_mgr *evf_mgr,
>
> spin_lock(&evf_mgr->ev_fence_lock);
> ev_fence = evf_mgr->ev_fence;
> - if (ev_fence) {
> - ef = dma_fence_get(&ev_fence->base);
> - dma_resv_add_fence(resv, ef, DMA_RESV_USAGE_BOOKKEEP);
> - }
> + if (ev_fence)
> + dma_resv_add_fence(resv, &ev_fence->base,
> DMA_RESV_USAGE_BOOKKEEP);
> spin_unlock(&evf_mgr->ev_fence_lock);
>
> return 0;
