This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: Use atomics in libgomp mutex

From: Alan Modra <amodra at gmail dot com>
To: Jakub Jelinek <jakub at redhat dot com>
Cc: gcc-patches at gcc dot gnu dot org, Richard Henderson <rth at redhat dot com>
Date: Mon, 28 Nov 2011 20:05:36 +1030
Subject: Re: Use atomics in libgomp mutex
References: <20111125000328.GD5085@bubble.grove.modra.org> <20111125073839.GN27242@tyan-ft48-01.lab.bos.redhat.com> <20111127225720.GB7827@bubble.grove.modra.org> <20111128000709.GD7827@bubble.grove.modra.org> <20111128081502.GV27242@tyan-ft48-01.lab.bos.redhat.com>

On Mon, Nov 28, 2011 at 09:15:02AM +0100, Jakub Jelinek wrote:
> On Mon, Nov 28, 2011 at 10:37:09AM +1030, Alan Modra wrote:
> > 14 A flush region without a list is implied at the following locations:
> > 15 â During a barrier region.
> > 16 â At entry to and exit from parallel, critical, and ordered regions.
> > [snip]

> If it is just GOMP_critical_start, couldn't it use an extra barrier
> before resp. after it calls mutex_lock?

I believe gomp_ordered_sync needs a barrier, and I'm unsure about
GOMP_parallel_start but I think one is needed there too.  Yes, putting
the barriers where they are actually needed is the best solution, but
I'm far from an OpenMP expert so am unsure whether these are the only
places requiring a barrier.

-- 
Alan Modra
Australia Development Lab, IBM

References:
- Fix libgomp semaphores
  - From: Alan Modra
- Re: Fix libgomp semaphores
  - From: Jakub Jelinek
- Use atomics in libgomp mutex
  - From: Alan Modra
- Re: Use atomics in libgomp mutex
  - From: Alan Modra
- Re: Use atomics in libgomp mutex
  - From: Jakub Jelinek

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]