This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Memory barriers vs lock/unlock
- From: Alexander Terekhov <alexander dot terekhov at gmail dot com>
- To: Peter Dimov <pdimov at mmltd dot net>
- Cc: Paolo Carlini <pcarlini at suse dot de>, libstdc++ <libstdc++ at gcc dot gnu dot org>, Jonathan Wakely <cow at compsoc dot man dot ac dot uk>
- Date: Wed, 9 Nov 2005 12:54:49 +0100
- Subject: Re: Memory barriers vs lock/unlock
- References: <4370AEC8.4020504@suse.de> <007801c5e499$be267dd0$6401a8c0@pdimov2>
On 11/8/05, Peter Dimov <pdimov@mmltd.net> wrote:
[...]
> When __exchange_and_add is implemented in terms of __sync_fetch_and_add,
> which seems to guarantee full ordering
I recently gathered that at least some of supposedly fully-fenced IA64
__sync_* intrinsics seem to have mf at the wrong place.
http://www.opengroup.org/austin/mailarchives/ag/msg08717.html
http://sourceware.org/cgi-bin/cvsweb.cgi/libc/nptl/sysdeps/ia64/pthread_spin_unlock.c?rev=1.3&content-type=text/x-cvsweb-markup&cvsroot=glibc
http://www.opengroup.org/austin/mailarchives/ag/msg08740.html
regards,
alexander.