Hello. On 07/08/2014 01:41 PM, Ian Molton wrote:
This patch fixes a race condition whereby a frame being captured may generate an interrupt between requesting capture to halt and freeing buffers.
This condition is exposed by the earlier patch that explicitly calls vb2_buffer_done() during stop streaming.
The solution is to wait for capture to finish prior to finalising these buffers.
Hm, my spell checker trips on "finalising"...
Signed-off-by: Ian Molton <ian.molton@xxxxxxxxxxxxxxx> Signed-off-by: William Towle <william.towle@xxxxxxxxxxxxxxx> --- drivers/media/platform/soc_camera/rcar_vin.c | 43 ++++++++++++++++++---------- 1 file changed, 28 insertions(+), 15 deletions(-)
diff --git a/drivers/media/platform/soc_camera/rcar_vin.c b/drivers/media/platform/soc_camera/rcar_vin.c index 06ce705..aeda4e2 100644 --- a/drivers/media/platform/soc_camera/rcar_vin.c +++ b/drivers/media/platform/soc_camera/rcar_vin.c
[...]
@@ -462,7 +485,6 @@ static void rcar_vin_videobuf_release(struct vb2_buffer *vb) struct rcar_vin_priv *priv = ici->priv; unsigned int i; int buf_in_use = 0; - spin_lock_irq(&priv->lock);
This seems like a random whitespace change. This empty should be present. [...]
@@ -517,12 +527,15 @@ static void rcar_vin_stop_streaming(struct vb2_queue *vq) spin_lock_irq(&priv->lock); + rcar_vin_wait_stop_streaming(priv); + for (i = 0; i < vq->num_buffers; ++i) if (vq->bufs[i]->state == VB2_BUF_STATE_ACTIVE) vb2_buffer_done(vq->bufs[i], VB2_BUF_STATE_ERROR); list_for_each_safe(buf_head, tmp, &priv->capture) list_del_init(buf_head); +
Also quite a random "drove-by" change.
spin_unlock_irq(&priv->lock); }
WBR, Sergei -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
