Hi Dave and Jean,
did some investigation into the "irda-usb doesn't work with ohci-hcd"
issue. Symptoms are on one hand the outgoing irda-packets get transmitted
(but corrupted) by the OHCI-HC and on the other hand the urb's are staying
queued without ever getting unlinked (but callback called).
Finally, it turned out the decision when to append a length=0 usb-packet
is partly done just the wrong way.
The patch below does
* append length=0 packet in the correct situation only,
i.e. basically when (length%maxpacketsize)==0
* make USB_ZERO_PACKET only effective on OUT transaction, because
it's useless for IN and uhci is restricted to out as well.
Testing with 2.5.24 and irda-usb demonstrates things are working again.
Martin
--------------------------
diff -ur linux-2.5.24/drivers/usb/host/ohci-hcd.c
v2.5.24-md/drivers/usb/host/ohci-hcd.c
--- linux-2.5.24/drivers/usb/host/ohci-hcd.c Sun Jul 7 18:45:12 2002
+++ v2.5.24-md/drivers/usb/host/ohci-hcd.c Tue Jul 9 16:18:11 2002
@@ -164,11 +164,13 @@
/* ... and maybe a zero length packet to wrap it up */
if (size == 0)
size++;
- else if ((urb->transfer_flags & USB_ZERO_PACKET) != 0
- && (urb->transfer_buffer_length
- % usb_maxpacket (urb->dev, pipe,
- usb_pipeout (pipe))) != 0)
- size++;
+ else if ((urb->transfer_flags & USB_ZERO_PACKET) != 0) {
+ size_t maxp;
+
+ maxp = usb_maxpacket(urb->dev,pipe,usb_pipeout(pipe));
+ if (usb_pipeout(pipe) &&
+(urb->transfer_buffer_length%maxp) == 0)
+ size++;
+ }
break;
case PIPE_ISOCHRONOUS: /* number of packets from URB */
size = urb->number_of_packets;
diff -ur linux-2.5.24/drivers/usb/host/ohci-q.c v2.5.24-md/drivers/usb/host/ohci-q.c
--- linux-2.5.24/drivers/usb/host/ohci-q.c Fri Jun 21 23:17:21 2002
+++ v2.5.24-md/drivers/usb/host/ohci-q.c Tue Jul 9 16:19:39 2002
@@ -640,7 +640,7 @@
data, data_len, urb, cnt);
cnt++;
if ((urb->transfer_flags & USB_ZERO_PACKET)
- && cnt < urb_priv->length) {
+ && usb_pipeout(urb->pipe) && cnt < urb_priv->length) {
td_fill (ohci, info | (cnt? TD_T_TOGGLE:toggle),
0, 0, urb, cnt);
cnt++;
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Stuff, things, and much much more.
http://thinkgeek.com/sf
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
