Re: [patch] Reorganize the cpufreq cpu hotplug locking to not be totally bizare

[Arjan van de Ven <arjan@linux.intel.com>]

On Wed, 2006-07-26 at 13:42 -0700, Srivatsa Vaddagiri wrote:
> On Wed, Jul 26, 2006 at 09:42:34PM +0200, Arjan van de Ven wrote:
> > As a quick hack I made non-lock_cpu_hotplug()'ing versions of the 3 key
> > workqueue functions (patch below). It works, it's correct, it's just so
> > ugly that I'm almost too ashamed to post it. I haven't found a better
> > solution yet though... time to take a step back I suppose.
> 
> My worry is that such special cases might be needed in more places as we
> discover further or as code evolves. Fundamentally looks like the locked and 
> unlocked paths of the kernel cannot be separated so well because of interaction 
> between subsystems. /me thinks rwsem seems to be a sane thing to go after.

rwsems unfortunately help you zilch; an rwsem is just a mutex with a
performance tweak, but from the deadlock perspective it's really a
mutex.

I'm really starting to feel that the hotplug lock would have been better
of being a refcount (with a waitqueue for zero) than a lock. While
"refcount+waitqueue" sort of IS a lock, the semantics make more sense
imo...

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