Subject: Re: trap in ipw(4)
To: Luke Mewburn <lukem@NetBSD.org>
From: john heasley <heas@shrubbery.net>
List: tech-net
Date: 12/16/2004 11:53:07
Thu, Dec 16, 2004 at 04:18:31PM +1100, Luke Mewburn:
> On Wed, Dec 15, 2004 at 03:25:38PM -0800, john heasley wrote:
>   | I've hit a trap in the ipw driver several times on an ibm t41p w/ a
>   | 2100 3B mini-pci.  I can almost reliably trigger it by booting, loggi=
ng
>   | in from a remote, and running dmesg.  It always occurs at the same pl=
ace
>   | each time;  IIRC, from the two occasions where it wasnt stuck in X, it
>   | complained about the value of %ebx, which appears to be m (the mbuf *=
).
>   |=20
>   | It seems as if the values of ifp and m (in ipw_start) get swapped
>   | somehow.  I looked through the code and do not see any place where th=
ese
>   | could get swapped.
>   |=20
>   | The firmware is the most recent with -current of ~12/4.  It is not do=
ing
>   | WEP.
>   |=20
>   | Has anyone seen this before?  Clues on how to debug?
>=20
> I've had hangs on my system that I've suspected were ipw(4) related;
> I couldn't get to the console because I was in X at the time.
>=20
> Please submit a PR about this.

Done, kern/28683 with a hack.

Pavel Cahyna pointed out that gdb's trace was probably bogus. :)