<html> <head> <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1"> <style type="text/css" style="display:none;"></style> </head> <body dir="ltr"> <div id="divtagdefaultwrapper" style="font-size:12pt;color:#000000;background-color:#FFFFFF;font-family:Calibri,Arial,Helvetica,sans-serif;"> Just to add to that I had no more GPU hangs with those patches on my CZ system. Tom <div style="color: rgb(0, 0, 0);"> <div> <hr tabindex="-1" style="display:inline-block; width:98%"> <div id="x_divRplyFwdMsg" dir="ltr">From: amd-gfx <amd-gfx-bounces@lists.freedesktop.org> on behalf of zhoucm1 <david1.zhou@amd.com> Sent: Tuesday, July 12, 2016 21:52 To: Christian König; amd-gfx@lists.freedesktop.org Subject: Re: [PATCH 1/3] drm/amdgpu: check flush fence context instead of same ring v2 <div> </div> </div> </div> <div class="PlainText"> On 2016年07月12日 20:05, Christian König wrote: > From: Christian König <christian.koenig@amd.com> > > Otherwise we can run into the following situation: > > 1. Process A grabs ID 1 for ring 0. > 2. Process B grabs ID 1 for ring 0. > 3. Process A grabs ID 1 for ring 1. > 4. Process A tries to reuse ID1 for ring 0 but things he doesn't need to flush. > > v2: check the context of the flush fence instead of messing with the owner field. > > Signed-off-by: Christian König <christian.koenig@amd.com> The set is Reviewed-by: Chunming Zhou <david1.zhou@amd.com> > --- > drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 9 ++++++--- > 1 file changed, 6 insertions(+), 3 deletions(-) > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c > index f8615a4..57337d4 100644 > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c > @@ -185,6 +185,7 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, > struct amdgpu_job *job) > { > struct amdgpu_device *adev = ring->adev; > + uint64_t fence_context = adev->fence_context + ring->idx; > struct fence *updates = sync->last_vm_update; > struct amdgpu_vm_id *id, *idle; > struct fence **fences; > @@ -244,7 +245,6 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, > i = ring->idx; > do { > struct fence *flushed; > - bool same_ring = ring->idx == i; > > id = vm->ids[i++]; > if (i == AMDGPU_MAX_RINGS) > @@ -262,8 +262,11 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring, > if (job->vm_pd_addr != id->pd_gpu_addr) > continue; > > - if (!same_ring && > - (!id->last_flush || !fence_is_signaled(id->last_flush))) > + if (!id->last_flush) > + continue; > + > + if (id->last_flush->context != fence_context && > + !fence_is_signaled(id->last_flush)) > continue; > > flushed = id->flushed_updates; _______________________________________________ amd-gfx mailing list amd-gfx@lists.freedesktop.org <a href="https://lists.freedesktop.org/mailman/listinfo/amd-gfx" id="LPlnk415996">https://lists.freedesktop.org/mailman/listinfo/amd-gfx</a> </div> </div> </div> </body> </html>