Re: [patch 0/3] [-rt] Fixes the timeout-bug in the rtmutex/PI-futex.

[Esben Nielsen]

On Wed, 26 Jul 2006, Ingo Molnar wrote:

> * Ingo Molnar <mingo@xxxxxxx> wrote:
>
> > and i also had to do the fixes below to get it to build.
>
> then it crashed with the assert below. I'll skip this one for now. I've
> attached the patches (cleaned up for whitespaces) plus the
> build-fixpatch.
>
> 	Ingo
>
> Brought up 2 CPUs
> BUG at kernel/rtmutex.c:773!
> ------------[ cut here ]------------
> kernel BUG at kernel/rtmutex.c:773!
> invalid opcode: 0000 [#1]
> PREEMPT SMP
> Modules linked in:
> CPU:    0
> EIP:    0060:[<c03e407c>]    Not tainted VLI
> EFLAGS: 00010246   (2.6.17-rt7 #14)
> EIP is at rt_lock_slowlock+0x21e/0x231
> eax: 00000020   ebx: c31d7000   ecx: c0477be4   edx: c31d97f0
> esi: 00000000   edi: c31d7d34   ebp: c31d7cdc   esp: c31d7c70
> ds: 007b   es: 007b   ss: 0068   preempt: 00000001

It is a very simple bug: The waiter was made garbage by the 
debug_rt_mutex_free_waiter() before it was used in the BUG_ON() line.
My UP machine boots now with CONFIG_DEBUG_RT_MUTEXEX on in -rt8. Fix 
is below. Apply on top of the patches you just send me with no further 
changes. I hope I don't have white space damage in this small patch!!

Esben

 kernel/rtmutex.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: linux-2.6.17-rt8/kernel/rtmutex.c
===================================================================
--- linux-2.6.17-rt8.orig/kernel/rtmutex.c
+++ linux-2.6.17-rt8/kernel/rtmutex.c
@@ -768,9 +768,9 @@ rt_lock_slowlock(struct rt_mutex *lock _
 	if (adjust_prio_final)
 		rt_mutex_adjust_prio_final(current, waiter.old_sched_lifo);

-	debug_rt_mutex_free_waiter(&waiter);
 	BUG_ON(current->boosting_prio != MAX_PRIO);
 	BUG_ON(waiter.old_sched_lifo != get_sched_lifo());
+	debug_rt_mutex_free_waiter(&waiter);
 }

 /*

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/