leandrof/linux
1
0
Fork 0
forked from Minki/linux
Code Issues Pull Requests Packages Projects Releases Wiki Activity

media: videobuf2-core: sanity checks for requests and qbuf

Browse Source 
The combination of supports_requests == 1 and min_buffers_needed > 0
is not allowed, WARN on that and return an error.

Also check that if vb2_core_qbuf() is called from req_queue, that it
doesn't return an error, unless it is -EIO.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Cc: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Reviewed-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
This commit is contained in:
Hans Verkuil 2021-07-29 15:36:27 +02:00 committed by Mauro Carvalho Chehab
parent b72dd0f390
commit 9015fcc256
1 changed files with 22 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
drivers/media/common/videobuf2/videobuf2-core.c
View File

@ -1421,9 +1421,19 @@ int vb2_core_qbuf(struct vb2_queue *q, unsigned int index, void *pb,
static void vb2_req_queue(struct media_request_object *obj)
{
struct vb2_buffer *vb = container_of(obj, struct vb2_buffer, req_obj);
int err;
mutex_lock(vb->vb2_queue->lock);
vb2_core_qbuf(vb->vb2_queue, vb->index, NULL, NULL);
/*
* There is no method to propagate an error from vb2_core_qbuf(),
* so if this returns a non-0 value, then WARN.
*
* The only exception is -EIO which is returned if q->error is
* set. We just ignore that, and expect this will be caught the
* next time vb2_req_prepare() is called.
*/
err = vb2_core_qbuf(vb->vb2_queue, vb->index, NULL, NULL);
WARN_ON_ONCE(err && err != -EIO);
mutex_unlock(vb->vb2_queue->lock);
}
@ -2342,6 +2352,17 @@ int vb2_core_queue_init(struct vb2_queue *q)
if (WARN_ON(q->requires_requests && !q->supports_requests))
return -EINVAL;
/*
* This combination is not allowed since a non-zero value of
* q->min_buffers_needed can cause vb2_core_qbuf() to fail if
* it has to call start_streaming(), and the Request API expects
* that queueing a request (and thus queueing a buffer contained
* in that request) will always succeed. There is no method of
* propagating an error back to userspace.
*/
if (WARN_ON(q->supports_requests && q->min_buffers_needed))
return -EINVAL;
INIT_LIST_HEAD(&q->queued_list);
INIT_LIST_HEAD(&q->done_list);
spin_lock_init(&q->done_lock);