| Message ID | 20220204100429.2049-4-christian.koenig@amd.com |
|---|---|
| State | New |
| Headers | show |
| Series | [1/6] dma-buf: consolidate dma_fence subclass checking | expand |
On Fri, 2022-02-04 at 11:04 +0100, Christian König wrote: > Chaining of dma_fence_chain objects is only allowed through the prev > fence and not through the contained fence. > > Warn about that when we create a dma_fence_chain. > > v2: fix comment style > > Signed-off-by: Christian König <christian.koenig@amd.com> > Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> > Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com> It looks like this blows up in generic drm code... https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_22201/shard-skl10/igt@syncobj_timeline@transfer-timeline-point.html /Thomas
Am 09.02.22 um 15:02 schrieb Thomas Hellström: > On Fri, 2022-02-04 at 11:04 +0100, Christian König wrote: >> Chaining of dma_fence_chain objects is only allowed through the prev >> fence and not through the contained fence. >> >> Warn about that when we create a dma_fence_chain. >> >> v2: fix comment style >> >> Signed-off-by: Christian König <christian.koenig@amd.com> >> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> >> Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com> > It looks like this blows up in generic drm code... > > https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_22201/shard-skl10/igt@syncobj_timeline@transfer-timeline-point.html Thanks for the notice. Going to take a look. I'm wondering why the last CI report I've got didn't showed that. Christian. > > /Thomas > >
diff --git a/drivers/dma-buf/dma-fence-chain.c b/drivers/dma-buf/dma-fence-chain.c index 1b4cb3e5cec9..084c6927b735 100644 --- a/drivers/dma-buf/dma-fence-chain.c +++ b/drivers/dma-buf/dma-fence-chain.c @@ -254,5 +254,14 @@ void dma_fence_chain_init(struct dma_fence_chain *chain, dma_fence_init(&chain->base, &dma_fence_chain_ops, &chain->lock, context, seqno); + + /* + * Chaining dma_fence_chain container together is only allowed through + * the prev fence and not through the contained fence. + * + * The correct way of handling this is to flatten out the fence + * structure into a dma_fence_array by the caller instead. + */ + WARN_ON(dma_fence_is_chain(fence)); } EXPORT_SYMBOL(dma_fence_chain_init);