linux-kernel - Re: [PATCH v2 01/12] locking/atomic: Introduce inc/dec calls for FETCH-OP flavors

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 24 Jun 2016 20:48:57 +0200
From:	Peter Zijlstra <peterz@...radead.org>
To:	Davidlohr Bueso <dave@...olabs.net>
Cc:	mingo@...nel.org, davem@...emloft.net, cw00.choi@...sung.com,
	dougthompson@...ssion.com, bp@...en8.de, mchehab@....samsung.com,
	gregkh@...uxfoundation.org, pfg@....com, jikos@...nel.org,
	hans.verkuil@...co.com, awalls@...metrocast.net,
	dledford@...hat.com, sean.hefty@...el.com, kys@...rosoft.com,
	heiko.carstens@...ibm.com, James.Bottomley@...senPartnership.com,
	sumit.semwal@...aro.org, schwidefsky@...ibm.com,
	linux-kernel@...r.kernel.org, Davidlohr Bueso <dbueso@...e.de>
Subject: Re: [PATCH v2 01/12] locking/atomic: Introduce inc/dec calls for
 FETCH-OP flavors

On Fri, Jun 24, 2016 at 09:34:30AM -0700, Davidlohr Bueso wrote:
> On Thu, 23 Jun 2016, Peter Zijlstra wrote:
> 
> >-
> >/* atomic_fetch_inc_relaxed */
> >#ifndef atomic_fetch_inc_relaxed
> >+
> >+#ifndef atomic_fetch_inc
> 
> #define atomic_fetch_inc(v)  (atomic_fetch_add(1, v))

Ah yes.

> >+#define atomic_fetch_inc_relaxed(v)	atomic_fetch_add_relaxed(1, (v))
> >+#define atomic_fetch_inc_acquire(v)	atomic_fetch_add_acquire(1, (v))
> >+#define atomic_fetch_inc_release(v)	atomic_fetch_add_release(1, (v))
> 
> I was under the impression that if the archs don't define their own calls,
> then we always default to fully ordered. Which is why I based all this ifdefery
> on what is currently done with the other atomic_fetch_$ops.

Right, but this allows an arch to only define atomic_fetch_add_relaxed()
and have the various forms of fetch_inc() instantiated as well.

With your earlier version they would all fall back to the fully
sequential variant of fetch_add().