On Mon, 2010-03-15 at 09:37 +0100, Sander Eikelenboom wrote:
> Hello Sarah,
>
> Attached is the syslog, tar'ed and bzip2,ed.
> (it grows rather fast with debug on :-) )
>
> Done the same as the previous time,
>
> - boot
> - use vgrabbj to grab a jpeg from the video device a few times.
>
>From the syslog, the situation I can tell is that isoc TDs are
successfully inserted into transfer ring, but the xHC does not process
them correctly: all isoc TDs are returned with short packet, data is not
transferred completely. I wonder why it acts like this. Please apply the
attached patch and send the syslog.
I verified the patchset with USB1.1 speaker and it works fine. It's a
OUT device while webcam is a IN device. I'll try webcam and see if it's
a code issue.
--
Thanks,
Andiry
diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
index 25f0928..db99567 100644
--- a/drivers/usb/host/xhci-ring.c
+++ b/drivers/usb/host/xhci-ring.c
@@ -1360,6 +1360,7 @@ static int handle_tx_event(struct xhci_hcd *xhci,
urb = td->urb;
urb_priv = urb->hcpriv;
idx = urb_priv->td_cnt;
+ status = 0;
/* handle completion code */
switch (trb_comp_code) {
@@ -1368,6 +1369,13 @@ static int handle_tx_event(struct xhci_hcd *xhci,
xhci_dbg(xhci, "Successful isoc "
"transfer!\n");
break;
+ case COMP_SHORT_TX:
+ xhci_dbg(xhci, "short transfer on isoc ep\n");
+ if (td->urb->transfer_flags & URB_SHORT_NOT_OK)
+ urb->iso_frame_desc[idx].status = -EREMOTEIO;
+ else
+ urb->iso_frame_desc[idx].status = 0;
+ break;
case COMP_UNDERRUN:
urb->iso_frame_desc[idx].status = -EREMOTEIO;
skip_td = 1;
@@ -2381,7 +2389,7 @@ static int xhci_queue_isoc_tx(struct xhci_hcd *xhci, gfp_t mem_flags,
if (!in_interrupt())
dev_dbg(&urb->dev->dev, "ep %#x - urb len = %#x (%d),"
- " addr = %#llx, num_trbs = %d\n",
+ " addr = %#llx, num_tds = %d\n",
urb->ep->desc.bEndpointAddress,
urb->transfer_buffer_length,
urb->transfer_buffer_length,
