On Fri, 10 Jan 2003 22:15:57 +0100 (CET)
Thomas Runge <coto@core.de> wrote:

> When it comes to read the data from the isochronous endpoint,
> on -release nothing happens, the programm just hangs
> and waits forever. This is not nice, but better than what
> happens on -current.
> 
> On -current I get a panic dropping me right into ddb:
> 
> uvm_fault(0xc696b98c, 0x0, 0, 1) -> e
> 
> ohci_softintr(c050c000,c6bbbb38,c013a7cc,c050c000,0) at ohci_softintr+0x290
> softintr_dispatch(1,c02f0010,30,10,10) at softintr_dispatch+0xb1
> Xsoftnet() at Xsoftnet+0x47
> --- interrupt ---
> <snip>


This panic is caused by an isochronous transfer via ugen(4)
on ohci(4). It seems to be related to ugen rather than ohci,
but isn't understood yet.

For the present, the following patch will fix this problem.

Index: ohci.c
===================================================================
RCS file: /cvsroot/src/sys/dev/usb/ohci.c,v
retrieving revision 1.135
diff -u -r1.135 ohci.c
--- ohci.c	2002/12/10 14:07:37	1.135
+++ ohci.c	2002/12/11 17:52:23
@@ -3374,8 +3374,9 @@
 {
 
 	DPRINTFN(1,("ohci_device_isoc_done: xfer=%p\n", xfer));
-
+#if 0
 	xfer->hcpriv = NULL;
+#endif
 }
 
 usbd_status


---
SOMEYA Yoshihiko <someya@netbsd.org>