Re: [PATCH 1/3] Increase lockdep limits: MAX_STACK_TRACE_ENTRIES

[Joao Correia <joaomiguelcorreia@gmail.com>]

On Thu, Jul 9, 2009 at 6:07 PM, Peter Zijlstra<a.p.zijlstra@chello.nl> wrote:
> On Thu, 2009-07-09 at 12:10 -0400, Dave Jones wrote:
>> On Thu, Jul 09, 2009 at 10:02:36AM +0200, Peter Zijlstra wrote:
>>  > On Thu, 2009-07-09 at 00:39 -0400, Dave Jones wrote:
>>  > > On Wed, Jul 08, 2009 at 08:36:04PM +0200, Peter Zijlstra wrote:
>>  > >  > Looking at a lockdep dump hch gave me I can see that that is certainly
>>  > >  > possible, I see tons of very deep callchains.
>>  > >  >
>>  > >  > /me wonders if we're getting significantly deeper..
>>  > >
>>  > > Looking at /proc/lockdep, I'm curious..
>>  > > Take a look at http://davej.fedorapeople.org/lockdep
>>  > > scroll down to c12c0924
>>  > >
>>  > > What's up with all those old_style_spin_init's ?
>>  >
>>  > What kernel are you running?
>>
>> ..31rc2
>>
>>  > Does your lib/dma_debug.c:dma_debug_init()
>>  > have spin_lock_init() in that HASH_SIZE loop?
>>
>> it's doing it by hand..
>>
>>  717         for (i = 0; i < HASH_SIZE; ++i) {
>>  718                 INIT_LIST_HEAD(&dma_entry_hash[i].list);
>>  719                 dma_entry_hash[i].lock = SPIN_LOCK_UNLOCKED;
>>  720         }
>
> Hmm, that's the problem, it should read:
>
>        for (i = 0; i < HASH_SIZE; ++i) {
>                INIT_LIST_HEAD(&dma_entry_hash[i].list);
>                spin_lock_init(&dma_entry_hash[i].lock);
>        }
>
> and does in -tip, so maybe Ingo took that patch, but I thought Joerg
> would push that Linus wards. Joerg?
>
>

Indeed, changing to spin_lock_init keeps the values at a much
healthier value, and the warnings do not trigger anymore. Thanks for
looking into this.

Disregard my patch submission please.

Joao Correia