Re: [PATCH -tip v3 09/11] data_race: Avoid nested statement expression

From: Arnd Bergmann <arnd@arndb.de>
To: Marco Elver <elver@google.com>
Cc: Will Deacon <will@kernel.org>,
	Nick Desaulniers <ndesaulniers@google.com>,
	"Paul E. McKenney" <paulmck@kernel.org>,
	Dmitry Vyukov <dvyukov@google.com>,
	Alexander Potapenko <glider@google.com>,
	Andrey Konovalov <andreyknvl@google.com>,
	kasan-dev <kasan-dev@googlegroups.com>,
	LKML <linux-kernel@vger.kernel.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Ingo Molnar <mingo@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	clang-built-linux <clang-built-linux@googlegroups.com>,
	Borislav Petkov <bp@alien8.de>,
	Stephen Rothwell <sfr@canb.auug.org.au>
Subject: Re: [PATCH -tip v3 09/11] data_race: Avoid nested statement expression
Date: Wed, 27 May 2020 11:26:51 +0200	[thread overview]
Message-ID: <CAK8P3a1BH5nXDK2VS7jWc_u2B1kztr4u9JMXhWF9-iZdrsb-7Q@mail.gmail.com> (raw)
In-Reply-To: <CANpmjNO2A39XRQ9OstwKGKpZ6wQ4ebVcBNfH_ZhCTi8RG6WqYw@mail.gmail.com>

On Wed, May 27, 2020 at 9:44 AM 'Marco Elver' via Clang Built Linux
<clang-built-linux@googlegroups.com> wrote:
> On Wed, 27 May 2020 at 09:22, Will Deacon <will@kernel.org> wrote:
> >
> > Nice! FWIW, I'm planning to have Alpha override __READ_ONCE_SCALAR()
> > eventually, so that smp_read_barrier_depends() can disappear forever. I
> > just bit off more than I can chew for 5.8 :(
> >
> > However, '__unqual_scalar_typeof()' is still useful for
> > load-acquire/store-release on arm64, so we still need a better solution to
> > the build-time regression imo. I'm not fond of picking random C11 features
> > to accomplish that, but I also don't have any better ideas...
>
> We already use _Static_assert in the kernel, so it's not the first use
> of a C11 feature.
>
> > Is there any mileage in the clever trick from Rasmus?
> >
> > https://lore.kernel.org/r/6cbc8ae1-8eb1-a5a0-a584-2081fca1c4aa@rasmusvillemoes.dk
>
> Apparently that one only works with GCC 7 or newer, and is only
> properly defined behaviour since C11. It also relies on multiple
> _Pragma. I'd probably take the arguably much cleaner _Generic solution
> over that. ;-)

I'd have to try, but I suspect we could force gcc-4.9 or higher to
accept it by always passing --std=gnu11 instead of --std=gnu89,
but that still wouldn't help us with gcc-4.8, and it's definitely not
something we could consider changing for v5.8.

However, if we find a solution that is nicer and faster but does
requires C11 or some other features from a newer compiler,
I think making it version dependent is a good idea and lets us
drop the worse code eventually.

> I think given that Peter and Arnd already did some testing, and it
> works as intended, if you don't mind, I'll send a patch for the
> _Generic version. At least that'll give us a more optimized
> __unqual_scalar_typeof(). Any further optimizations to READ_ONCE()
> like you mentioned then become a little less urgent.

Right. I think there is still room for optimization around here, but
for v5.8 I'm happy enough with Marco's__unqual_scalar_typeof()
change. Stephen Rothwell is probably the one who's most affected
by compile speed, so it would be good to get an Ack/Nak from him
on whether this brings speed and memory usage back to normal
for him as well.

      Arnd