rtmutex assert failure (was [Patch] restore the RCU callback...)

[Bill Huey (hui) <billh@gnuppy.monkey.org>]

On Mon, Aug 21, 2006 at 05:21:35PM -0700, Robert Crocombe wrote:
> On 8/18/06, hui Bill Huey <billh@gnuppy.monkey.org> wrote:
> >Patch attached:
> 
> The problem still appears to happen, but now I get no trace at all,
> just a single line reported to the machine's console (and not on the
> serial console):
> 
> pdflush/314[CPU#2]: BUG in debug_rt_mutex_unlock at 
> kernel/rt_mutex_debug.c:471
> 
> i.e., a standard statement, except at pdflush instead of swapper this time.

Thanks for testing. I appreciate this and I'm more than just plain willing
to help out. Any thing goes to get this patch solid. :)

I'll come up with a patch tomorrow to try and get a clean stack trace. I've
made some change to the bug dump output to make it more preemption aware,
but, as you can, some tweeking is needed.

However, I don't think this is going to be a simple atomic violation problem
that I've seen numerous times before. That's what the previous patch tried to
address. It's seriously pointing to a problem with the rtmutex and/or how it's
used, maybe a corner case it's missing.

I'll start to look at it either tonight or tomorrow and see what's going on.
It's going to take me a bit so be patient. I'll definitely get back to you
on this. I've got a number of weeks to blow exclusively on -rt development,
so I'll be around for a little while longer, hopefully longer.

bill