[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20160115004638.GI13558@port70.net>
Date: Fri, 15 Jan 2016 01:46:38 +0100
From: Szabolcs Nagy <nsz@...t70.net>
To: musl@...ts.openwall.com
Subject: Re: atomic.h cleanup
* Rich Felker <dalias@...c.org> [2016-01-14 18:32:24 -0500]:
> On Thu, Jan 14, 2016 at 11:37:48PM +0100, Jens Gustedt wrote:
> > Am Donnerstag, den 14.01.2016, 17:12 -0500 schrieb Rich Felker:
> > > There are already multiple reasons we don't use the compiler's
> > > atomics, either directly or indirectly via stdatomic.h. They're not
> > > supported in some old/alternative compilers, they generate highly
> > > suboptimal code even on modern compilers for some important archs
> > > (e.g. ARM),
> >
> > I have seen some pretty good assembler when using the
> > __atomic... builtins, so I can't completely follow, here.
>
> It generates "dmb sy" all over the place instead of "dmb ish".
> Synchronizing with external bus devices is NOT something you want to
> happen in thread synchronization primitives.
>
fixed in gcc-6 :)
https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=224317
(not backported to earlier branches though)
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
