Re: [ath5k-devel] ath5k_tasklet_rx BUG_ON(bf->skb == NULL)

[Hugh Dickins <hugh@veritas.com>]

On Thu, 8 Jan 2009, Maxim Levitsky wrote:
> On Thu, 2009-01-08 at 13:49 +0000, Hugh Dickins wrote:
> > 
> > When running swapping load tests, wireless alive but not in active use,
> > I've now twice hit the BUG_ON(bf->skb == NULL) in ath5k_tasklet_rx().
> > 
> > So, that BUG_ON(bf->skb == NULL) appears to be unsafe under
> > memory pressure; but the fix wasn't obvious to me, so over
> > to you!
> > 
> > I'd be glad to try patches, of course, but it's not happening
> > often enough for me to report back success quickly - unless I
> > stumble on a quicker way to reproduce it, it'll need a week or
> > two to grow confident of a fix.
> 
> I use aspire one, and I don't see that, did this happen several times?

A couple of times, when running memory swapping loads: I doubt it
would be any problem when there's easily freeable memory around.

> 
> Could you try compat-wireless:
> 
> http://wireless.kernel.org/en/users/Download

Thanks, I've downloaded the tarball, but the difference between the
drivers/net/wireless/ath5k/base.c in there and in Linus's git is
very little: I'd be be very surprised if it fixes this bug.

As I said, it'll take a week or two to be confident that the issue
has been fixed, so I'd rather embark on that once a likely fix has
gone in.

(Of course, just removing the BUG_ON, and making sure there's
no oops on the NULL pointer, would fix my immediate issue:
but I doubt the right fix will be as simple as that.)

> 
> It has many fixes that made my wireless on that notebook more or less
> usable.

I find 2.6.28's wireless very usable on that machine: I bet you're
a wireless specialist who stresses the wireless side of things in
ways I wouldn't begin to imagine: whereas I just browse and download
and scp a little, but as a memory specialist try swapping loads
which perhaps that driver had not been subjected to before.

Hugh