[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1605491118.ek4lw8ppgq.astroid@bobo.none>
Date: Mon, 16 Nov 2020 11:48:58 +1000
From: Nicholas Piggin <npiggin@...il.com>
To: Christophe Leroy <christophe.leroy@...roup.eu>,
Peter Zijlstra <peterz@...radead.org>
Cc: Alexey Kardashevskiy <aik@...abs.ru>,
Arnd Bergmann <arnd@...db.de>,
Boqun Feng <boqun.feng@...il.com>,
Christophe Leroy <christophe.leroy@....fr>,
linux-arch@...r.kernel.org, linux-kernel@...r.kernel.org,
linuxppc-dev@...ts.ozlabs.org,
Michael Ellerman <mpe@...erman.id.au>,
Will Deacon <will@...nel.org>
Subject: Re: [PATCH 1/3] asm-generic/atomic64: Add support for ARCH_ATOMIC
Excerpts from Peter Zijlstra's message of November 11, 2020 11:44 pm:
> On Wed, Nov 11, 2020 at 02:39:01PM +0100, Christophe Leroy wrote:
>> Hello,
>>
>> Le 11/11/2020 à 12:07, Nicholas Piggin a écrit :
>> > This passes atomic64 selftest on ppc32 on qemu (uniprocessor only)
>> > both before and after powerpc is converted to use ARCH_ATOMIC.
>>
>> Can you explain what this change does and why it is needed ?
>
> That certainly should've been in the Changelog. This enables atomic
> instrumentation, see asm-generic/atomic-instrumented.h. IOW, it makes
> atomic ops visible to K*SAN.
>
Right. This specific patch doesn't actually "do" anything except
allow generic atomic64 to be used with ARCH_ATOMIC. It does that
by re-naming some things to avoid name collisions and also providing
the arch_ prefix that ARCH_ATOMIC expects.
I don't know what should be in the changelog. I suppose the latter,
the former is discoverable by looking at ARCH_ATOMIC code and
patches but I guess it's polite and doesn't hurt to include the
former as well.
I'll send an update before long.
Thanks,
Nick
Powered by blists - more mailing lists