On Fri, 5 Jun 2009, Tor Håkon Gjerde wrote:
> Hello.
>
> I am building a high speed usb audio device. It uses asynch isochronous
> mode for playback but the info from the feedback endpoint seems to get
> lost somewhere.
>
> Relevant ouput from usbmon (2 lines):
> f694c1c0 1057412404 S Zi:1:022:3 -115:64:3920 1 -18:0:4 4 <
> f694c5c0 1057420403 C Zi:1:022:3 0:64:3984:0 1 0:0:4 0
>
> I would assume the last line should look somewhat like this:
> f694c5c0 1057420403 C Zi:1:022:3 0:64:3984:0 1 0:0:4 4 = 00000600
>
> The feedback is all correct when I force the device to full speed.
What this means is that for Iso transfers, ehci-hcd does not report
anything in urb->actual_length. You have to look at the actual_length
fields of the individual iso_packet_descriptors.
This is a bug in ehci-hcd. The patch below ought to fix it.
Alan Stern
Index: usb-2.6/drivers/usb/host/ehci-sched.c
===================================================================
--- usb-2.6.orig/drivers/usb/host/ehci-sched.c
+++ usb-2.6/drivers/usb/host/ehci-sched.c
@@ -1618,10 +1618,12 @@ itd_complete (
/* HC need not update length with this error */
if (!(t & EHCI_ISOC_BABBLE))
- desc->actual_length = EHCI_ITD_LENGTH (t);
+ urb->actual_length += desc->actual_length =
+ EHCI_ITD_LENGTH(t);
} else if (likely ((t & EHCI_ISOC_ACTIVE) == 0)) {
desc->status = 0;
- desc->actual_length = EHCI_ITD_LENGTH (t);
+ urb->actual_length += desc->actual_length =
+ EHCI_ITD_LENGTH(t);
} else {
/* URB was too late */
desc->status = -EXDEV;
@@ -2012,7 +2014,8 @@ sitd_complete (
desc->status = -EPROTO;
} else {
desc->status = 0;
- desc->actual_length = desc->length - SITD_LENGTH (t);
+ urb->actual_length += desc->actual_length =
+ desc->length - SITD_LENGTH (t);
}
stream->depth -= stream->interval << 3;
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
