On 11/07/2018 03:04 PM, Hans Verkuil wrote:
> The use of flush_schedule_work() made no sense and caused a syzkaller error.
> Replace with the correct cancel_delayed_work_sync().
> 
> Signed-off-by: Hans Verkuil <hans.verkuil>

Mistyped that SoB, this should of course be:

Signed-off-by: Hans Verkuil <hans.verk...@cisco.com>

Regards,

        Hans

> Reported-by: syzbot+69780d144754b8071...@syzkaller.appspotmail.com
> ---
> diff --git a/drivers/media/platform/vim2m.c b/drivers/media/platform/vim2m.c
> index d82db738f174..f938a2c54314 100644
> --- a/drivers/media/platform/vim2m.c
> +++ b/drivers/media/platform/vim2m.c
> @@ -805,10 +805,11 @@ static int vim2m_start_streaming(struct vb2_queue *q, 
> unsigned count)
>  static void vim2m_stop_streaming(struct vb2_queue *q)
>  {
>       struct vim2m_ctx *ctx = vb2_get_drv_priv(q);
> +     struct vim2m_dev *dev = ctx->dev;
>       struct vb2_v4l2_buffer *vbuf;
>       unsigned long flags;
> 
> -     flush_scheduled_work();
> +     cancel_delayed_work_sync(&dev->work_run);
>       for (;;) {
>               if (V4L2_TYPE_IS_OUTPUT(q->type))
>                       vbuf = v4l2_m2m_src_buf_remove(ctx->fh.m2m_ctx);
> 

Reply via email to