Date: Mon, 13 May 2024 08:41:27 -0700
From: "Paul E. McKenney" <paulmck@...nel.org>
To: Boqun Feng <boqun.feng@...il.com>
Cc: linux-arch@...r.kernel.org, linux-kernel@...r.kernel.org,
	elver@...gle.com, akpm@...ux-foundation.org, tglx@...utronix.de,
	peterz@...radead.org, dianders@...omium.org, pmladek@...e.com,
	arnd@...db.de, torvalds@...ux-foundation.org, kernel-team@...a.com
Subject: Re: [PATCH v2 cmpxchg 09/13] lib: Add one-byte emulation function

On Mon, May 13, 2024 at 07:44:00AM -0700, Boqun Feng wrote:
> On Wed, May 01, 2024 at 04:01:26PM -0700, Paul E. McKenney wrote:
> > Architectures are required to provide four-byte cmpxchg() and 64-bit
> > architectures are additionally required to provide eight-byte cmpxchg().
> > However, there are cases where one-byte cmpxchg() would be extremely
> > useful.  Therefore, provide cmpxchg_emu_u8() that emulates one-byte
> > cmpxchg() in terms of four-byte cmpxchg().
> > 
> > Note that this emulations is fully ordered, and can (for example) cause
> > one-byte cmpxchg_relaxed() to incur the overhead of full ordering.
> > If this causes problems for a given architecture, that architecture is
> > free to provide its own lighter-weight primitives.
> > 
> > [ paulmck: Apply Marco Elver feedback. ]
> > [ paulmck: Apply kernel test robot feedback. ]
> > [ paulmck: Drop two-byte support per Arnd Bergmann feedback. ]
> > 
> > Link: https://lore.kernel.org/all/0733eb10-5e7a-4450-9b8a-527b97c842ff@paulmck-laptop/
> > 
> > Signed-off-by: Paul E. McKenney <paulmck@...nel.org>
> > Acked-by: Marco Elver <elver@...gle.com>
> > Cc: Andrew Morton <akpm@...ux-foundation.org>
> > Cc: Thomas Gleixner <tglx@...utronix.de>
> > Cc: "Peter Zijlstra (Intel)" <peterz@...radead.org>
> > Cc: Douglas Anderson <dianders@...omium.org>
> > Cc: Petr Mladek <pmladek@...e.com>
> > Cc: Arnd Bergmann <arnd@...db.de>
> > Cc: <linux-arch@...r.kernel.org>
> > ---
> >  arch/Kconfig                |  3 +++
> >  include/linux/cmpxchg-emu.h | 15 +++++++++++++
> >  lib/Makefile                |  1 +
> >  lib/cmpxchg-emu.c           | 45 +++++++++++++++++++++++++++++++++++++
> >  4 files changed, 64 insertions(+)
> >  create mode 100644 include/linux/cmpxchg-emu.h
> >  create mode 100644 lib/cmpxchg-emu.c
> > 
> > diff --git a/arch/Kconfig b/arch/Kconfig
> > index 9f066785bb71d..284663392eef8 100644
> > --- a/arch/Kconfig
> > +++ b/arch/Kconfig
> > @@ -1609,4 +1609,7 @@ config CC_HAS_SANE_FUNCTION_ALIGNMENT
> >  	# strict alignment always, even with -falign-functions.
> >  	def_bool CC_HAS_MIN_FUNCTION_ALIGNMENT || CC_IS_CLANG
> >  
> > +config ARCH_NEED_CMPXCHG_1_EMU
> > +	bool
> > +
> >  endmenu
> > diff --git a/include/linux/cmpxchg-emu.h b/include/linux/cmpxchg-emu.h
> > new file mode 100644
> > index 0000000000000..998deec67740a
> > --- /dev/null
> > +++ b/include/linux/cmpxchg-emu.h
> > @@ -0,0 +1,15 @@
> > +/* SPDX-License-Identifier: GPL-2.0+ */
> > +/*
> > + * Emulated 1-byte and 2-byte cmpxchg operations for architectures
> > + * lacking direct support for these sizes.  These are implemented in terms
> > + * of 4-byte cmpxchg operations.
> > + *
> > + * Copyright (C) 2024 Paul E. McKenney.
> > + */
> > +
> > +#ifndef __LINUX_CMPXCHG_EMU_H
> > +#define __LINUX_CMPXCHG_EMU_H
> > +
> > +uintptr_t cmpxchg_emu_u8(volatile u8 *p, uintptr_t old, uintptr_t new);
> > +
> > +#endif /* __LINUX_CMPXCHG_EMU_H */
> > diff --git a/lib/Makefile b/lib/Makefile
> > index ffc6b2341b45a..cc3d52fdb477d 100644
> > --- a/lib/Makefile
> > +++ b/lib/Makefile
> > @@ -236,6 +236,7 @@ obj-$(CONFIG_FUNCTION_ERROR_INJECTION) += error-inject.o
> >  lib-$(CONFIG_GENERIC_BUG) += bug.o
> >  
> >  obj-$(CONFIG_HAVE_ARCH_TRACEHOOK) += syscall.o
> > +obj-$(CONFIG_ARCH_NEED_CMPXCHG_1_EMU) += cmpxchg-emu.o
> >  
> >  obj-$(CONFIG_DYNAMIC_DEBUG_CORE) += dynamic_debug.o
> >  #ensure exported functions have prototypes
> > diff --git a/lib/cmpxchg-emu.c b/lib/cmpxchg-emu.c
> > new file mode 100644
> > index 0000000000000..27f6f97cb60dd
> > --- /dev/null
> > +++ b/lib/cmpxchg-emu.c
> > @@ -0,0 +1,45 @@
> > +// SPDX-License-Identifier: GPL-2.0+
> > +/*
> > + * Emulated 1-byte cmpxchg operation for architectures lacking direct
> > + * support for this size.  This is implemented in terms of 4-byte cmpxchg
> > + * operations.
> > + *
> > + * Copyright (C) 2024 Paul E. McKenney.
> > + */
> > +
> > +#include <linux/types.h>
> > +#include <linux/export.h>
> > +#include <linux/instrumented.h>
> > +#include <linux/atomic.h>
> > +#include <linux/panic.h>
> > +#include <linux/bug.h>
> > +#include <asm-generic/rwonce.h>
> > +#include <linux/cmpxchg-emu.h>
> > +
> > +union u8_32 {
> > +	u8 b[4];
> > +	u32 w;
> > +};
> > +
> > +/* Emulate one-byte cmpxchg() in terms of 4-byte cmpxchg. */
> > +uintptr_t cmpxchg_emu_u8(volatile u8 *p, uintptr_t old, uintptr_t new)
> > +{
> > +	u32 *p32 = (u32 *)(((uintptr_t)p) & ~0x3);
> > +	int i = ((uintptr_t)p) & 0x3;
> > +	union u8_32 old32;
> > +	union u8_32 new32;
> > +	u32 ret;
> > +
> > +	ret = READ_ONCE(*p32);
> > +	do {
> > +		old32.w = ret;
> > +		if (old32.b[i] != old)
> > +			return old32.b[i];
> > +		new32.w = old32.w;
> > +		new32.b[i] = new;
> > +		instrument_atomic_read_write(p, 1);
> > +		ret = data_race(cmpxchg(p32, old32.w, new32.w)); // Overridden above.
> 
> Just out of curiosity, why is this `data_race` needed? cmpxchg is atomic
> so there should be no chance for a data race?

That is what I thought, too.  ;-)

The problem is that the cmpxchg() covers 32 bits, and so without that
data_race(), KCSAN would complain about data races with perfectly
legitimate concurrent accesses to the other three bytes.

The instrument_atomic_read_write(p, 1) beforehand tells KCSAN to complain
about concurrent accesses, but only to that one byte.

							Thanx, Paul

> Regards,
> Boqun
> 
> > +	} while (ret != old32.w);
> > +	return old;
> > +}
> > +EXPORT_SYMBOL_GPL(cmpxchg_emu_u8);
> > -- 
> > 2.40.1
> > 

Powered by blists - more mailing lists