On Fri, 26 May 2000, Petko Manolov wrote:
> > I still have problems with the default pegasus.c driver. The fix which was
> > applied is either wrong or triggers something wrong in the OHCI stack.
> The fix was suggested by Johannes Erdfelt and really fix the problem
> at least on Jason's machine and mine (both ohci-based). It seems to
> me that your machine has HW problem.
That would really suck ; I've got a make-kpkg running with perhaps a fix
(basically, using async unlink now that there's some OHCI support)
> > Now, basically whenever there's a timeout, the kernel panics (BUG() in
> > sched.c:640). This seems to be because usb_unlink_urb calls schedule(), at
> > least in the OHCI implementation.
> Actually it calls schedule_timeout(), which expect user context of
> course.
Yep. It doesn't however have a in_interrupt() guard, like JE's uhci stack.
I've just duplicated that guard, will see whether it helps.
> > FWIW, the machine is a Toshiba Satelite 230CDX, with Potato on it.
> I am curious about driver's behavior on some other machines
> (preferably uhci-based), can you test it?
Will try ASAP, on a 430TX K6 (Asus TX97XE mobo).
-- Cyrille
------------------------------------------------------------------------------
Grumpf.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
