Received-SPF: pass (google.com: domain of linux-kernel+bounces-192860-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Date: Tue, 28 May 2024 20:16:31 +0200
From: Andrea Parri <parri.andrea@gmail.com>
To: Alexandre Ghiti <alexghiti@rivosinc.com>
Cc: Jonathan Corbet <corbet@lwn.net>,
	Paul Walmsley <paul.walmsley@sifive.com>,
	Palmer Dabbelt <palmer@dabbelt.com>,
	Albert Ou <aou@eecs.berkeley.edu>,
	Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@redhat.com>, Will Deacon <will@kernel.org>,
	Waiman Long <longman@redhat.com>, Boqun Feng <boqun.feng@gmail.com>,
	Arnd Bergmann <arnd@arndb.de>, Leonardo Bras <leobras@redhat.com>,
	Guo Ren <guoren@kernel.org>, linux-doc@vger.kernel.org,
	linux-kernel@vger.kernel.org, linux-riscv@lists.infradead.org,
	linux-arch@vger.kernel.org
Subject: Re: [PATCH 1/7] riscv: Implement cmpxchg32/64() using Zacas
Message-ID: <ZlYff9x12FICHoP0@andrea>
References: <20240528151052.313031-1-alexghiti@rivosinc.com>
 <20240528151052.313031-2-alexghiti@rivosinc.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20240528151052.313031-2-alexghiti@rivosinc.com>

> +	asm goto(ALTERNATIVE("nop", "j %[zacas]", 0,			\
> +			     RISCV_ISA_EXT_ZACAS, 1)			\
> +			: : : : zacas);					\
> +									\
>  	__asm__ __volatile__ (						\
>  		prepend							\
>  		"0:	lr" lr_sfx " %0, %2\n"				\
>  		"	bne  %0, %z3, 1f\n"				\
> -		"	sc" sc_sfx " %1, %z4, %2\n"			\
> +		"	sc" sc_cas_sfx " %1, %z4, %2\n"			\
>  		"	bnez %1, 0b\n"					\
>  		append							\
>  		"1:\n"							\
>  		: "=&r" (r), "=&r" (__rc), "+A" (*(p))			\
>  		: "rJ" (co o), "rJ" (n)					\
>  		: "memory");						\
> +	goto end;							\
> +									\
> +zacas:									\
> +	__asm__ __volatile__ (						\
> +		prepend							\
> +		"	amocas" sc_cas_sfx " %0, %z2, %1\n"		\
> +		append							\
> +		: "+&r" (r), "+A" (*(p))				\
> +		: "rJ" (n)						\
> +		: "memory");						\

With this, a cmpxchg32() will result in something like

  amocas.w.rl     a5,a4,(s1)
  fence           rw,rw

(cf. my remarks in patch #4); this will/should provide enough sync,
but you might want to try the alternative and currently more common
mapping for "fully-ordered AMO sequences", aka

  amocas.w.aqrl   a5,a4,(s1)

Similarly for cmpxchg64 and other sizes.

  Andrea