In case of ISOC transfer, if TX queue is full then we start
dropping latest frame, instead we should drop oldest frame &
add latest frame to TX queue.
Signed-off-by: Rupesh Gujare <rupesh.guj...@atmel.com>
---
drivers/staging/ozwpan/ozpd.c | 13 ++++++++++++-
1 file changed, 12 insertions(+), 1 deletion(-)
diff --git a/drivers/staging/ozwpan/ozpd.c b/drivers/staging/ozwpan/ozpd.c
index 86876bd..17fd7b2 100644
--- a/drivers/staging/ozwpan/ozpd.c
+++ b/drivers/staging/ozwpan/ozpd.c
@@ -844,9 +844,20 @@ int oz_send_isoc_unit(struct oz_pd *pd, u8 ep_num, const
u8 *data, int len)
struct oz_tx_frame *isoc_unit = NULL;
int nb = pd->nb_queued_isoc_frames;
if (nb >= pd->isoc_latency) {
+ struct list_head *e;
+ struct oz_tx_frame *f;
oz_dbg(TX_FRAMES, "Dropping ISOC Unit nb= %d\n",
nb);
- goto out;
+ spin_lock(&pd->tx_frame_lock);
+ list_for_each(e, &pd->tx_queue) {
+ f = container_of(e, struct oz_tx_frame,
+ link);
+ if (f->skb != NULL) {
+ oz_tx_isoc_free(pd, f);
+ break;
+ }
+ }
+ spin_unlock(&pd->tx_frame_lock);
}
isoc_unit = oz_tx_frame_alloc(pd);
if (isoc_unit == NULL)
--
1.7.9.5
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
