Re: [PATCH] cmpxchg: allow const-qualified old value in cmpxchg()

From: David Laight <david.laight.linux@gmail.com>
To: Hangbin Liu <liuhangbin@gmail.com>
Cc: Richard Henderson <richard.henderson@linaro.org>,
	Matt Turner <mattst88@gmail.com>,
	Magnus Lindholm <linmag7@gmail.com>,
	Vineet Gupta <vgupta@kernel.org>, Brian Cain <bcain@kernel.org>,
	"James E.J. Bottomley" <James.Bottomley@HansenPartnership.com>,
	Helge Deller <deller@gmx.de>,
	Madhavan Srinivasan <maddy@linux.ibm.com>,
	Michael Ellerman <mpe@ellerman.id.au>,
	Nicholas Piggin <npiggin@gmail.com>,
	"Christophe Leroy (CS GROUP)" <chleroy@kernel.org>,
	Paul Walmsley <pjw@kernel.org>,
	Palmer Dabbelt <palmer@dabbelt.com>,
	Albert Ou <aou@eecs.berkeley.edu>,
	Alexandre Ghiti <alex@ghiti.fr>,
	Yoshinori Sato <ysato@users.sourceforge.jp>,
	Rich Felker <dalias@libc.org>,
	John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>,
	"David S. Miller" <davem@davemloft.net>,
	Andreas Larsson <andreas@gaisler.com>,
	Thomas Gleixner <tglx@kernel.org>, Ingo Molnar <mingo@redhat.com>,
	Borislav Petkov <bp@alien8.de>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	x86@kernel.org, "H. Peter Anvin" <hpa@zytor.com>,
	Chris Zankel <chris@zankel.net>,
	Max Filippov <jcmvbkbc@gmail.com>,
	linux-alpha@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-snps-arc@lists.infradead.org,
	linux-hexagon@vger.kernel.org, linux-parisc@vger.kernel.org,
	linuxppc-dev@lists.ozlabs.org, linux-riscv@lists.infradead.org,
	linux-sh@vger.kernel.org, sparclinux@vger.kernel.org,
	Jakub Kicinski <kuba@kernel.org>
Subject: Re: [PATCH] cmpxchg: allow const-qualified old value in cmpxchg()
Date: Thu, 2 Apr 2026 09:57:47 +0100	[thread overview]
Message-ID: <20260402095747.6efbd456@pumpkin> (raw)
In-Reply-To: <20260402-cmpxchg-v1-1-4561e2c05d2c@gmail.com>

On Thu, 02 Apr 2026 14:56:01 +0800
Hangbin Liu <liuhangbin@gmail.com> wrote:

> The old value passed to cmpxchg() is semantically read-only: it is
> only loaded into a register as a comparand and is never written back.
> However, the macro currently assigns it implicitly to a local variable
> of type __typeof__(*(ptr)), which triggers -Werror=discarded-qualifiers
> when old is a const-qualified pointer and ptr points to a non-const type.
> 
> To avoid this, let's add an explicit cast to __typeof__(*(ptr)) for the
> old local variable in the cmpxchg macros. This explicit cast suppresses
> the -Wdiscarded-qualifiers diagnostic.
> 
> The new value is intentionally left without a cast: new will be stored
> into *ptr, so silently accepting a const-qualified new would allow
> callers to store a pointer-to-const into a non-const location without
> any compiler warning.
> 
> Suggested-by: Jakub Kicinski <kuba@kernel.org>
> Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
...
> 
> diff --git a/arch/alpha/include/asm/cmpxchg.h b/arch/alpha/include/asm/cmpxchg.h
> index ae1b96479d0c..b4b8dac759c4 100644
> --- a/arch/alpha/include/asm/cmpxchg.h
> +++ b/arch/alpha/include/asm/cmpxchg.h
> @@ -234,7 +234,7 @@ ____cmpxchg(volatile void *ptr, unsigned long old, unsigned long new,
>  
>  #define arch_cmpxchg_local(ptr, o, n)					\
>  ({									\
> -	__typeof__(*(ptr)) _o_ = (o);					\
> +	__typeof__(*(ptr)) _o_ = (__typeof__(*(ptr)))(o);		\
>  	__typeof__(*(ptr)) _n_ = (n);					\
>  	(__typeof__(*(ptr))) ____cmpxchg((ptr), (unsigned long)_o_,	\
>  					  (unsigned long)_n_,		\

That looks like it loses the check that 'o' has the same type as '*ptr'.
Maybe this works?
	auto _o_ = 1 ? (o) : *(ptr);

  David