Re: [tip:core/locking] locking: Check spinlock_t/rwlock_t argument type on non-SMP builds too

[Dave <kilroyd@googlemail.com>]

Ingo Molnar wrote:
> * tip-bot for David Kilroy <kilroyd@googlemail.com> wrote:
> 
>> Commit-ID:  02626aa5ecc03f94585164b97bedabe15302e3c3
>> Gitweb:     http://git.kernel.org/tip/02626aa5ecc03f94585164b97bedabe15302e3c3
>> Author:     David Kilroy <kilroyd@googlemail.com>
>> AuthorDate: Wed, 22 Jul 2009 19:11:35 +0100
>> Committer:  Ingo Molnar <mingo@elte.hu>
>> CommitDate: Sun, 2 Aug 2009 14:59:29 +0200
>>
>> locking: Check spinlock_t/rwlock_t argument type on non-SMP builds too
> 
> -tip testing found UP crashes and i bisected it down to:
> 
>  02626aa5ecc03f94585164b97bedabe15302e3c3 is first bad commit
>  commit 02626aa5ecc03f94585164b97bedabe15302e3c3
>  Author: David Kilroy <kilroyd@googlemail.com>
>  Date:   Wed Jul 22 19:11:35 2009 +0100
> 
>     locking: Check spinlock_t/rwlock_t argument type on non-SMP builds too
> 
> the crash looks like this:
> 
> [   55.257999] ip[3699]: segfault at 31108c6d80 ip 00000031108c6d80 sp 00007fff4aa183d8 error 14
> 
> sometimes it's just a spontaneous reboot with no log message.


Crap. Looking really carefully:

+static inline int _spin_trylock_bh(spinlock_t *lock)
+{ __LOCK(lock); return 1; }

Should have been:

+static inline int _spin_trylock_bh(spinlock_t *lock)
+{ __LOCK_BH(lock); return 1; }

I can't say if that's definitely causing your issue, but it's certainly
wrong. I'll go through it all again, and send a v3 of the patch.


Dave.