Re: 2.6.18-rc5-mm1 + all hotfixes -- BUG: MAX_STACK_TRACE_ENTRIES too low!

[Miles Lane]

On 9/3/06, Miles Lane <miles.lane@xxxxxxxxx> wrote:
> On 9/3/06, Ingo Molnar <mingo@xxxxxxx> wrote:
> >
> > * Miles Lane <miles.lane@xxxxxxxxx> wrote:
> >
> > > Sorry Andrew.  I don't see clues here to help me target the report to
> > > a maintainer. I hope this helps.
> > >
> > > BUG: MAX_STACK_TRACE_ENTRIES too low!
> > > turning off the locking correctness validator.
> >
> > Miles, could you try the patch below? (Andrew: if this solves Miles'
> > problem then i think this is v2.6.18 material too. [The other
> > possibility would be some permanent stack-trace entries leak, in which
> > case the patch will not help. If that happens then we'll have to debug
> > this some more.])
> >
> >         Ingo
> >
> > ---------------->
> > From: Ingo Molnar <mingo@xxxxxxx>
> > Subject: lockdep: double the number of stack-trace entries
> >
> > Miles Lane reported the "BUG: MAX_STACK_TRACE_ENTRIES too low!" message,
> > which means that during normal use his system produced enough lockdep
> > events so that the 128-thousand entries stack-trace array got exhausted.
> > Double the size of the array.
> >
> > Signed-off-by: Ingo Molnar <mingo@xxxxxxx>
> > ---
> >  kernel/lockdep_internals.h |    2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > Index: linux/kernel/lockdep_internals.h
> > ===================================================================
> > --- linux.orig/kernel/lockdep_internals.h
> > +++ linux/kernel/lockdep_internals.h
> > @@ -27,7 +27,7 @@
> >   * Stack-trace: tightly packed array of stack backtrace
> >   * addresses. Protected by the hash_lock.
> >   */
> > -#define MAX_STACK_TRACE_ENTRIES        131072UL
> > +#define MAX_STACK_TRACE_ENTRIES        262144UL
> >
> >  extern struct list_head all_lock_classes;
> >
> >
>
> Ingo, there seemed to be a difference between the file you editted and
>  the one in Andrew's tree.  I remade you patch so it applies cleanly.
> I'll test and let you know.  One word of caution, I only hit the
> problem once and I'm not sure how to trigger the condition.  I'll do
> my best.
>
> Thanks,
>         Miles
>
> --- kernel/lockdep_internals.h~ 2006-09-03 09:59:29.000000000 -0700
> +++ kernel/lockdep_internals.h  2006-09-03 10:00:55.000000000 -0700
> @@ -27,7 +27,7 @@
>   * Stack-trace: tightly packed array of stack backtrace
>   * addresses. Protected by the hash_lock.
>   */
> -#define MAX_STACK_TRACE_ENTRIES        131072UL
> +#define MAX_STACK_TRACE_ENTRIES        262144UL
>
>  extern struct list_head all_lock_classes;

By the way, after making this change "make all install modules
modules_install" didn't seem to notice that the file had been
modified.  I backed up .config, ran "make mrproper", etc.  Is this a
build dependency checker bug?

Thanks,
       Miles

--
VGER BF report: H 4.35163e-06
-
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/