ping?
On Tue, Jun 19, 2012 at 11:12 PM, Dima Zavin dmitriyz@google.com wrote:
Tomasz,
I've encountered an issue with this patch when userspace does several stream_on/stream_off cycles. When the user tries to qbuf a buffer after doing stream_off, we trigger the "dmabuf already pinned" warning since we didn't unmap the buffer as dqbuf was never called.
The below patch adds calls to unmap in queue_cancel, but my feeling is that we probably should be calling detach too (i.e. put_dmabuf).
Thoughts?
--Dima
Subject: [PATCH] v4l: vb2: unmap dmabufs on STREAM_OFF event
Currently, if the user issues a STREAM_OFF request and then tries to re-enqueue buffers, it will trigger a warning in the vb2 allocators as the buffer would still be mapped from before STREAM_OFF was called. The current expectation is that buffers will be unmapped in dqbuf, but that will never be called on the mapped buffers after a STREAM_OFF event.
Cc: Sumit Semwal sumit.semwal@ti.com Cc: Tomasz Stanislawski t.stanislaws@samsung.com Signed-off-by: Dima Zavin dima@android.com
drivers/media/video/videobuf2-core.c | 22 ++++++++++++++++++++-- 1 files changed, 20 insertions(+), 2 deletions(-)
diff --git a/drivers/media/video/videobuf2-core.c b/drivers/media/video/videobuf2-core.c index b431dc6..e2a8f12 100644 --- a/drivers/media/video/videobuf2-core.c +++ b/drivers/media/video/videobuf2-core.c @@ -1592,8 +1592,26 @@ static void __vb2_queue_cancel(struct vb2_queue *q) /* * Reinitialize all buffers for next use. */
for (i = 0; i < q->num_buffers; ++i)q->bufs[i]->state = VB2_BUF_STATE_DEQUEUED;
for (i = 0; i < q->num_buffers; ++i) {struct vb2_buffer *vb = q->bufs[i];int plane;vb->state = VB2_BUF_STATE_DEQUEUED;if (q->memory != V4L2_MEMORY_DMABUF)continue;for (plane = 0; plane < vb->num_planes; ++plane) {struct vb2_plane *p = &vb->planes[plane];if (!p->mem_priv)continue;if (p->dbuf_mapped) {call_memop(q, unmap_dmabuf, p->mem_priv);p->dbuf_mapped = 0;}}}}
/**
1.7.7.3