Created attachment 6332 [details]
dmesg-2.6.14-rc4-gregkh-patches

complete dmesg-output

What type of device are you removing?

The log shows that it was a USB printer.

Frieder, can you attach the contents of /proc/bus/usb/devices with the printer
plugged in?

first scenario:
the printer is on
the oops occurs during gentoo

Created attachment 6334 [details]
proc-bus-usb-devices

in reply to  Comment #3

Created attachment 6335 [details]
proc-bus-usb-devices_new

the other does not contain the required informations

Aha -- multiple altsettings!  Just as I thought.

Greg, this is the problem I warned about earlier.  Your endpoint-attribute sysfs
code doesn't work when altsettings get changed.

On Thu, Oct 20, 2005 at 07:44:44AM -0700, bugme-daemon@kernel-bugs.osdl.org wrote:
> Aha -- multiple altsettings!  Just as I thought.
> 
> Greg, this is the problem I warned about earlier.  Your endpoint-attribute
> sysfs
> code doesn't work when altsettings get changed.

Ok, then we need some usb core changes in order to handle this properly,
right?

Created attachment 6357 [details]
Fix Greg's endpoint attribute stuff

Greg: Somehow I _knew_ you were going to say that!  :-)

Frieder, see if this patch helps.  I wrote it for 2.6.14-rc4 plus
gregkh-all-2.6.14-rc4.patch.  My copy of gregkh-all is a little old, so let me
know if you have any trouble applying this patch.

Greg: Eww!  Don't you believe in wrapping lines before 80 columns or checking
for error returns from kzalloc etc.?

Alan your patch applied fine and seems to fix the oops
thx