Re: [PATCH 1/3] acpi: add real mutex function calls

[Daniel Walker <dwalker@mvista.com>]

On Mon, 2008-07-21 at 21:20 +0200, Andi Kleen wrote:
> Daniel Walker wrote:
> > On Mon, 2008-07-21 at 09:51 +0800, Zhao Yakui wrote:
> >> On Sat, 2008-07-19 at 11:16 -0700, Daniel Walker wrote:
> >>> Instead of re-using semaphores for the mutex operation, I've
> >>> added usage of the kernel mutex for the os mutex implementation.
> >>>
> >> What is the advantage that the kernel mutex is used for the ACPI mutex
> >> implementation instead of using semaphore?
> >> And it seems that too much ACPICA source code is touched.
> > 
> > In general you would want to use a mutex whenever your using mutex-like
> > semantics. If I see a mutex used in code then I have a pretty good idea
> > the locking is sane..
> 
> With a mutex the locking can be totally broken and with a semaphore the 
> locking can be completely fine. They are both just tools which can be 
> used correctly and also incorrectly.

True ..

> The main advantage of mutexes is that they can be slightly more 
> efficient (doesn't matter for the ACPI case, the ACPI interpreter is not 
> performance critical) and that they are easier to check with lockdep. 
> But there are also other considerations like in ACPICA's case portability.

The ACPI interpreter actually specifically implements a mutex in the
AML. The interpreter wants a mutex, and not a semaphore. I'm not sure
why your bringing up portability issues ..

> My feeling is that adding own semaphore wrappers to ACPICA wouldn't
> be an improvement. Defering to Bob Moore for his opinion, since
> he maintains ACPICA.

I'm not trying to mod the locking that ACPI internally does (which I
think is mutex compatible) .. I'm just changing what the linux wrappers
are calling .. The additional lockdep calls are linux specific but they
aren't changing how the locking is already being done.

Daniel