[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAHVXubiLJWtAit9T6OYx00qHu2QOVNqYRZZiOZHtmDBrDoW5Ew@mail.gmail.com>
Date: Mon, 15 Jul 2024 13:48:57 +0200
From: Alexandre Ghiti <alexghiti@...osinc.com>
To: Andrea Parri <parri.andrea@...il.com>
Cc: Alexandre Ghiti <alex@...ti.fr>, Jonathan Corbet <corbet@....net>,
Paul Walmsley <paul.walmsley@...ive.com>, Palmer Dabbelt <palmer@...belt.com>,
Albert Ou <aou@...s.berkeley.edu>, Nathan Chancellor <nathan@...nel.org>,
Peter Zijlstra <peterz@...radead.org>, Ingo Molnar <mingo@...hat.com>, Will Deacon <will@...nel.org>,
Waiman Long <longman@...hat.com>, Boqun Feng <boqun.feng@...il.com>, Arnd Bergmann <arnd@...db.de>,
Leonardo Bras <leobras@...hat.com>, Guo Ren <guoren@...nel.org>, linux-doc@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-riscv@...ts.infradead.org,
linux-arch@...r.kernel.org
Subject: Re: [PATCH v2 01/10] riscv: Implement cmpxchg32/64() using Zacas
Hi Andrea,
On Wed, Jul 10, 2024 at 1:47 AM Andrea Parri <parri.andrea@...il.com> wrote:
>
> > > Is this second IS_ENABLED(CONFIG_RISCV_ISA_ZACAS) check actually needed?
> > > (just wondering - no real objection)
> >
> > To me yes, otherwise a toolchain without zacas support would fail to
> > assemble the amocas instruction.
>
> To elaborate on my question: Such a toolchain may be able to recognize
> that the block of code following the zacas: label (and comprising the
> amocas instruction) can't be reached/executed if the first IS_ENABLED()
> evaluates to false (due to the goto end; statement), and consequently it
> may compile out the entire block/instruction no matter the presence or
> not of the second IS_ENABLE() check. IOW, such a toolchain/compiler may
> not actually have to assemble the amocas instruction under such config.
> In fact, this is how the current gcc trunk (which doesn't support zacas)
> seems to behave. And this very same optimization/code removal seems to
> be performed by clang when CONFIG_RISCV_ISA_ZACAS=n. IAC, I'd agree it
> is good to be explicit in the sources and keep both of these checks.
Indeed, clang works fine without the second IS_ENABLED(). I'll remove
it then as the code is complex enough.
Thanks,
Alex
>
>
> > > Why the semicolon?
> >
> > That fixes a clang warning reported by Nathan here:
> > https://lore.kernel.org/linux-riscv/20240528193110.GA2196855@thelio-3990X/
>
> I see. Thanks for the pointer.
>
>
> > > This is because the compiler doesn't realize __ret is actually
> > > initialized, right? IAC, seems a bit unexpected to initialize
> > > with (old) (which indicates SUCCESS of the CMPXCHG operation);
> > > how about using (new) for the initialization of __ret instead?
> > > would (new) still work for you?
> >
> > But amocas rd register must contain the expected old value in order to
> > actually work right?
>
> Agreed. Thanks for the clarification.
>
> Andrea
Powered by blists - more mailing lists