Sorry it took a few days to get back to you...
On Sat, 13 Mar 2004, Matthias Andree wrote:
> > yet, can you post both the strace output and the dmesg output for the same
> > run?
>
> Not on the list, too large. I can show the URL of a bzip2ed copy of the
> logs on some FTP/HTTP server once I've had the leisure to reboot the
> machine several times, and I can mail them off-list.
I looked at your logs. Interestingly, the problems occur with control
transfers, not bulk transfers, but the strace only shows USBDEVFS_BULK
ioctls, not USBDEVFS_CONTROL. I can't imagine what that means or how it
happened. (Unless it's a bug in strace?)
The syslog includes some information about the failed control transfers.
They are Get_Descriptor requests, but exactly which descriptor isn't
shown. Below is a patch that will provide more complete information.
Try running it as before. The strace log may or may not help, it's hard
to say; you might as well include it too.
Ultimately your downfall was caused by the reply to one of those control
transfers. The program requested a 9-byte descriptor from the scanner and
the reply was longer than 9 bytes (that's the "Babble" near the end of
your syslog). Many VIA USB controllers have a bug that causes them to
stop working when they receive an unexpectedly long packet, and that
happened to you. When the controller dies it takes everything along with
it. (When I have a chance I'll change the UHCI driver so that doesn't
happen -- but for now it does.)
This patch will quite likely apply to a 2.4.recent kernel as well as to
2.6.4, although you will have to edit the third line to take out the
"/core" element in the pathname. It would be a good idea for you to try
running it under 2.4 too, so I can compare the resulting syslogs and see
what has changed.
Alan Stern
===== devio.c 1.87 vs edited =====
--- 1.87/drivers/usb/core/devio.c Mon Mar 1 06:40:11 2004
+++ edited/drivers/usb/core/devio.c Tue Mar 16 17:06:27 2004
@@ -549,6 +549,9 @@
if (!(tbuf = (unsigned char *)__get_free_page(GFP_KERNEL)))
return -ENOMEM;
tmo = (ctrl.timeout * HZ + 999) / 1000;
+ printk(KERN_INFO "usbfs: proc_control: dev %d rqt %x rq %x val %x ind %x len
%u\n",
+ dev->devnum, ctrl.bRequestType, ctrl.bRequest, ctrl.wValue,
ctrl.wIndex,
+ ctrl.wLength);
if (ctrl.bRequestType & 0x80) {
if (ctrl.wLength && !access_ok(VERIFY_WRITE, ctrl.data, ctrl.wLength))
{
free_page((unsigned long)tbuf);
-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
