RE: [Xen-devel] xen-unstable: Xen BUG at sched_credit.c:925

[Dan Magenheimer <dan.magenheimer@xxxxxxxxxx>]

(Note this is clean xen-unstable with no changes.)
Well after compiling with debug=y, the following BUG
is triggered during a different activity.

(XEN) Xen call trace:
(XEN)    [<ffff82c480117a73>] csched_schedule+0xa8/0x557
(XEN)    [<ffff82c48011c647>] schedule+0xd6/0x46a
(XEN)    [<ffff82c48011da20>] do_softirq+0x6a/0x77
(XEN)
(XEN)
(XEN) ****************************************
(XEN) Panic on CPU 1:
(XEN) Xen BUG at sched_credit.c:326
(XEN) ****************************************

Looks like some memory corruption or something... maybe
a free'd pointer that is used as if never freed.

While this could be a tmem bug, tmem is so filled with
asserts when debug=y is enabled that it is less likely
(though certainly still possible).  Very likely, tmem
is churning otherwise unused memory a lot more than
normal, surfacing a (non-tmem) latent bent pointer bug.

This of course will be a real bear to track down,
especially in code with which I am completely unfamiliar.
Any ideas?

Thanks,
Dan

> -----Original Message-----
> From: Keir Fraser [mailto:keir.fraser@xxxxxxxxxxxxx]
> Sent: Friday, November 13, 2009 10:16 AM
> To: Dan Magenheimer; Xen-Devel (E-mail)
> Subject: Re: [Xen-devel] xen-unstable: Xen BUG at sched_credit.c:925
> 
> 
> On 13/11/2009 17:04, "Dan Magenheimer" 
> <dan.magenheimer@xxxxxxxxxx> wrote:
> 
> > I'm seeing this BUG/crash when shutting down a domain.
> > I've only seen it when running a domain with tmem, though
> > I'm not clear on what the relationship would be.
> > 
> > I haven't tested tmem with xen-unstable for a couple of
> > months now and I see five patches to sched_credit.c
> > since Aug 25... should I just try backing those out,
> > or any other ideas on what might be going on?  The
> > code around the BUG is not very intuitive to me
> > and it looks like that chunk of code hasn't been
> > changed for years.
> 
> Worth a try, but I don't think any of those changesets are 
> likely to cause
> this bug either. It's not a BUG() I've seen or had reported triggering
> before.
> 
>  -- Keir
> 
> 
>

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel