Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20150517223327.GR17573@brightrain.aerifal.cx>
Date: Sun, 17 May 2015 18:33:27 -0400
From: Rich Felker <dalias@...c.org>
To: musl@...ts.openwall.com
Subject: Re: Deduplicating atomics written in terms of CAS

On Mon, May 18, 2015 at 12:23:07AM +0200, Jens Gustedt wrote:
> Am Sonntag, den 17.05.2015, 13:59 -0400 schrieb Rich Felker:
> > > Ah sorry, I probably went too fast. My last paragraph would be for all
> > > atomic operations, so in particular 32 bit. A macro "a_load" would
> > > make intentions clearer and would perhaps allow to implement an
> > > optional compile time check to see if we use any object consistently
> > > as atomic or not.
> > 
> > The reason I'm mildly against this is that all current reads of
> > atomics, except via the return value of a_cas or a_fetch_add, are
> > relaxed-order. We don't care if we see a stale value; if staleness
> > could be a problem, the caller takes care of that in an efficient way.
> > Having a_load that's relaxed-order whereas all the existing atomics
> > are seq_cst order would be an inconsistent API design.
> 
> I still wasn't clear enough, sorry. My idea was not that such a
> function or macro should change anything on the binary code that is
> produced, at least for production builds. I just thought to
> encapsulate all atomic accesses into a type and functions that allow
> to have a compile check.

I understand that. But if it were called a_load, its semantics (no
synchronization/relaxed order) would be inconsistent with all other
a_* atomics which are seq_cst. That's what I don't like.

Rich

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.