Re: [BUG] Crda oopses the system

[Maxim Levitsky <maximlevitsky@gmail.com>]

On Wed, 2009-05-13 at 14:28 -0700, Luis R. Rodriguez wrote:
> On Tue, May 12, 2009 at 3:07 PM, Maxim Levitsky <maximlevitsky@gmail.com> wrote:
> > On Tue, 2009-05-12 at 15:00 -0700, Luis R. Rodriguez wrote:
> >> On Tue, May 12, 2009 at 10:34 AM, Maxim Levitsky
> >> <maximlevitsky@gmail.com> wrote:
> >> > On Tue, 2009-05-12 at 10:24 -0700, Luis R. Rodriguez wrote:
> >> >> On Tue, May 12, 2009 at 8:05 AM, Maxim Levitsky <maximlevitsky@gmail.com> wrote:
> >> >> > On Sun, 2009-05-10 at 12:36 -0700, Luis R. Rodriguez wrote:
> >> >> >> On Sun, May 10, 2009 at 12:34 PM, Maxim Levitsky
> >> >> >> <maximlevitsky@gmail.com> wrote:
> >> >> >> > On Sun, 2009-05-10 at 11:33 -0700, Luis R. Rodriguez wrote:
> >> >> >> >> On Sun, May 10, 2009 at 5:20 AM, Maxim Levitsky <maximlevitsky@gmail.com> wrote:
> >> >> >> >> > On Wed, 2009-05-06 at 16:34 -0400, Pavel Roskin wrote:
> >> >> >> >> >> gdb /lib/modules/`uname -r`/kernel/net/wireless/cfg80211.ko
> >> >> >> >> >
> >> >> >> >> > I have recompiled the kernel with debugging info.
> >> >> >> >> >
> >> >> >> >> > This is new backtrace:
> >> >> >
> >> >> > Ok, here is what I found:
> >> >> >
> >> >> > The real BUG_ON that fires up is at
> >> >> >
> >> >> > wireless/reg.c: __set_regdom: BUG_ON(!country_ie_regdomain)
> >> >> >
> >> >> > Confirmed by replacing it with printk.
> >> >> >
> >> >> >
> >> >> > I attach a beacon sample that triggers this:
> >> >>
> >> >> Thanks for that. Was this from wireless-testing? If so as of what SHA1
> >> >> sum? Can you provide steps to reproduce? What AP? What card?
> >> > wireless-testing, b2382a4aeff07a481ccf860e4f716b48b52e3781
> >>
> >> That HEAD is buggy, its missing a fix for the minstrel/pid fix.
> >> Although since you are running iwl3945 it explains why you haven't
> >> seen that oops.
> >>
> >> Will look into this -- thanks. How easy can you reproduce BTW? Does it
> >> happen immediately upon assoc? Does it happen with other APs at the
> >> university?
> > Happens always when I try to associate.
> 
> Great so it will be easy to confirm a fix. Can you please try the 4
> patches I just posted? I haven't finished reviewing your case but so
> far from the pcap review (which was really helpful) I don't see
> anything funky from the beacon except that the FCS is bad. But that
> shouldn't matter.
This is a fault of iwlwifi monitor mode. it always reports 0 there.

> 
> Anyway please try the patches I posted and I'll finish reviewing this
> in the meantime.
I don't see anything yet.

> 
> > Other APs, I don't see many other APs there, (you mean other essids,
> > right?)
> 
> I meant other APs at the University on the same network.
This happens with all (I obviously couldn't test them all...) AP here.
(more correctly to say there were no ap that didn't trigger that bug)
I now converted that BUG_ON to printk, so I can  now use crda, and still
see if this bug exists.


> 
> > Most of them are encrypted anyway.
> 
> OK this should not matter.
> 
>   Luis


Thanks a lot.
	Best regards,
		Maxim Levitsky