On 4/30/25 04:40, Prike Liang wrote: > Lock the eviction fence before trying to signal it. > > Signed-off-by: Prike Liang <prike.li...@amd.com> > --- > drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c > b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c > index 242bfb91c4f7..fed065892568 100644 > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c > @@ -108,7 +108,9 @@ amdgpu_eviction_fence_suspend_worker(struct work_struct > *work) > struct amdgpu_eviction_fence *ev_fence; > > mutex_lock(&uq_mgr->userq_mutex); > + spin_lock(&evf_mgr->ev_fence_lock); > ev_fence = evf_mgr->ev_fence; > + spin_unlock(&evf_mgr->ev_fence_lock);
That's a good catch, but won't work like this. You need to grab a reference to the fence while holding the lock, e.g. something like dma_fence_get(evf_mgr->ev_fence); Regards, Christian. > if (!ev_fence) > goto unlock; >
