Messages in this thread | | | Date | Mon, 12 Sep 2022 10:46:47 +0200 | Subject | Re: [PATCH v3 4/6] drm/amdgpu: Enable signaling on fence. | From | Christian König <> |
| |
Am 09.09.22 um 19:08 schrieb Arvind Yadav: > Here's enabling software signaling on fence because > amdgpu_ctx_add_fence() is checking the status of fence > and emits warning. > > Signed-off-by: Arvind Yadav <Arvind.Yadav@amd.com> > --- > > Changes in v1, v2: This new patch was not part of previous series. > > --- > drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c > index afe22f83d4a6..21221d705588 100644 > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c > @@ -730,6 +730,8 @@ uint64_t amdgpu_ctx_add_fence(struct amdgpu_ctx *ctx, > > dma_fence_get(fence); > > + dma_fence_enable_sw_signaling(fence); > +
That looks like a step into the right direction, but still isn't correct.
The code using this interface should call amdgpu_ctx_wait_prev_fence() before calling amdgpu_ctx_add_fence(). And amdgpu_ctx_wait_prev_fence() in turn calls dma_fence_wait() which should also enables the signaling.
So when we need this here something is still very wrong on the logic :)
Thanks, Christian.
> spin_lock(&ctx->ring_lock); > centity->fences[idx] = fence; > centity->sequence++;
| |