lkml.org 
[lkml]   [2008]   [Aug]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 0 of 9] x86/smp function calls: convert x86 tlb flushes to use function calls [POST 2]
Ingo Molnar wrote:
> * Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>
>
>> I think this might be a spurious "holding multiple locks in the same
>> class" bug. All the queue locks are presumably in the same class, and
>> ipi_call_lock_irq() wants to hold them all to lock out any IPIs.
>> Spurious because this is the only place which holds more than one
>> queue lock, and it always locks 0->N.
>>
>> I guess the fix is to use an outer lock and use spin_lock_nested()
>> (now that it exists). Something along these lines?
>>
>
> this is not a good idea:
>
>
>> +/* Hold queues_lock when taking more than one queue[].lock at once */
>> +static DEFINE_SPINLOCK(queues_lock);
>>
>
> because it adds an artificial extra spinlock for no good reason and
> weakens the lock dependency checking as well.
>
> Just add a lock class descriptor to each call_function_queue lock, so
> that lockdep can see that it's truly all in the correct order.
>
> I.e. dont turn lockdep off artificially.

Are you sure? I thought this is exactly the case where
spin_lock_nest_lock() is supposed to be used? Admittedly it's very
simple, but the extra lock does two things: 1) it guarantees that the
queue locks can be taken in any order, and 2) it's a single lock on
which we can do spin_lock_irq(), rather than doing it in the loop for
each individual lock (which I think was bogus).

I don't see how it weakens lockdep in any way. Does it hide any
potential lock misuse?

J



\
 
 \ /
  Last update: 2008-08-19 17:01    [W:0.074 / U:33.080 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site