[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <5d65491caf6249c8b72c7a6ced95614c@AcuMS.aculab.com>
Date: Mon, 13 Jun 2022 21:29:46 +0000
From: David Laight <David.Laight@...LAB.COM>
To: "'Luck, Tony'" <tony.luck@...el.com>,
"Lobakin, Alexandr" <alexandr.lobakin@...el.com>,
Marco Elver <elver@...gle.com>
CC: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
Arnd Bergmann <arnd@...db.de>,
Yury Norov <yury.norov@...il.com>,
Mark Rutland <mark.rutland@....com>,
Matt Turner <mattst88@...il.com>,
Brian Cain <bcain@...cinc.com>,
Geert Uytterhoeven <geert@...ux-m68k.org>,
"Yoshinori Sato" <ysato@...rs.sourceforge.jp>,
Rich Felker <dalias@...c.org>,
"David S. Miller" <davem@...emloft.net>,
Kees Cook <keescook@...omium.org>,
"Peter Zijlstra (Intel)" <peterz@...radead.org>,
Borislav Petkov <bp@...e.de>,
"Greg Kroah-Hartman" <gregkh@...uxfoundation.org>,
"linux-alpha@...r.kernel.org" <linux-alpha@...r.kernel.org>,
"linux-hexagon@...r.kernel.org" <linux-hexagon@...r.kernel.org>,
"linux-ia64@...r.kernel.org" <linux-ia64@...r.kernel.org>,
"linux-m68k@...ts.linux-m68k.org" <linux-m68k@...ts.linux-m68k.org>,
"linux-sh@...r.kernel.org" <linux-sh@...r.kernel.org>,
"sparclinux@...r.kernel.org" <sparclinux@...r.kernel.org>,
"linux-arch@...r.kernel.org" <linux-arch@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH v2 2/6] bitops: always define asm-generic non-atomic
bitops
From: Luck, Tony
> Sent: 13 June 2022 17:27
>
> >> It's listed in Documentation/atomic_bitops.txt.
> >
> > Oh, so my memory was actually correct that I saw it in the docs
> > somewhere.
> > WDYT, should I mention this here in the code (block comment) as well
> > that it's atomic and must not lose `volatile` as Andy suggested or
> > it's sufficient to have it in the docs (+ it's not underscored)?
>
> I think a comment that the "volatile" is required to prevent re-ordering
> is enough.
>
> But maybe others are sufficiently clear on the meaning? I once wasted
> time looking for the non-atomic __test_bit() version (to use in some code
> that was already protected by a spin lock, so didn't need the overhead
> of an "atomic" version) before realizing there wasn't a non-atomic one.
Does it make any sense for 'test bit' to be atomic?
I'm not even sure is needs any ordering constraints either.
The result is always stale - the value can be changed by
another cpu at any time.
The set/clear atomic bit-ops require a RMW bus cycle - which has
to be locked (or similar) to avoid corruption.
The atomic 'test and set' (etc) are RMW and return a valid state.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)
Powered by blists - more mailing lists