On 8/8/26 08:42, Sidong Yang wrote:
On 8/1/26 16:46, Pavel Begunkov wrote:+ if (dma_resv_wait_timeout(dmabuf->resv, DMA_RESV_USAGE_KERNEL, + true, 0) < 0) { + dma_resv_unlock(dmabuf->resv); + goto retry; + }Shouldn't this be <= 0? With a zero timeout dma_resv_wait_timeout() returns 0 for a still pending fence; only -ERESTARTSYS is negative.
You're right, and this chunk is gone in v6 -- Pavel Begunkov
