Re: [PATCH] arch,locking: Ciao arch_mutex_cpu_relax()

[Davidlohr Bueso <davidlohr@hp.com>]

On Mon, 2014-06-23 at 08:58 +0200, Peter Zijlstra wrote:
> On Fri, Jun 20, 2014 at 11:21:13AM -0700, Davidlohr Bueso wrote:
> > From: Davidlohr Bueso <davidlohr@hp.com>
> > 
> > The arch_mutex_cpu_relax() function, introduced by 34b133f, is
> > hacky and ugly. It was added a few years ago to address the fact
> > that common cpu_relax() calls include yielding on s390, and thus
> > impact the optimistic spinning functionality of mutexes. Nowadays
> > we use this function well beyond mutexes: rwsem, qrwlock, mcs and
> > lockref. Since the macro that defines the call is in the mutex header,
> > any users must include mutex.h and the naming is misleading as well.
> > 
> > This patch (i) renames the call to arch_cpu_relax (for lack of a better
> > name), and (ii) defines it in each arch's asm/processor.h local header,
> > just like for regular cpu_relax() functions. On all archs, except s390,
> > arch_cpu_relax is simply cpu_relax, and thus we can take it out of
> > mutex.h. While this can seem redundant or weird, I believe it is a
> > good choice as it allows us to move out arch specific logic from generic
> > locking primitives and enables future(?) archs to transparently define
> > it, similarly to System Z.
> > 
> > Please note that these changes are only tested on x86-64.
> 
> While I like the general idea; does anyone have a better name for this?
> So in particular, the difference is that on s390:
> 
>  cpu_relax()                - yields the vcpu
>  arch_{,mutex_}cpu_relax()  - will actually spin-wait

iirc Heiko had suggested cpu_relax_simple() in the past. I don't think
it's any better or worse than arch_cpu_relax(). For s390
cpu_relax_noyield() would perhaps be suitable, but not very descriptive
for the rest of the archs. I'm really lacking creativity for this name.

Thanks,
Davidlohr