Received-SPF: pass (google.com: domain of linux-kernel+bounces-153086-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
Date: Mon, 22 Apr 2024 10:24:46 +0100
From: Mark Rutland <mark.rutland@arm.com>
To: George Guo <dongtai.guo@linux.dev>
Cc: peterz@infradead.org, jpoimboe@kernel.org, jbaron@akamai.com,
	rostedt@goodmis.org, ardb@kernel.org, catalin.marinas@arm.com,
	will@kernel.org, linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org, George Guo <guodongtai@kylinos.cn>
Subject: Re: [PATCH 1/1] arm64: optimize code duplication in
 arch_static_branch/_jump function
Message-ID: <ZiYs3itsmdOmuPq4@FVFF77S0Q05N>
References: <20240422063853.3568733-1-dongtai.guo@linux.dev>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20240422063853.3568733-1-dongtai.guo@linux.dev>

On Mon, Apr 22, 2024 at 02:38:53PM +0800, George Guo wrote:
> From: George Guo <guodongtai@kylinos.cn>
> 
> Extracted the jump table definition code from the arch_static_branch and
> arch_static_branch_jump functions into a macro JUMP_TABLE_ENTRY to reduce
> code duplication and improve readability.

The commit title says this is an optimization, but the commit message says this
is a cleanup (and this clearly is not an optimization).

This seems to be copying what x86 did in commit:

  e1aa35c4c4bc71e4 ("jump_label, x86: Factor out the __jump_table generation")

.. where the commit message is much clearer.

> 
> Signed-off-by: George Guo <guodongtai@kylinos.cn>
> ---
>  arch/arm64/include/asm/jump_label.h | 19 +++++++++----------
>  1 file changed, 9 insertions(+), 10 deletions(-)
> 
> diff --git a/arch/arm64/include/asm/jump_label.h b/arch/arm64/include/asm/jump_label.h
> index 6aafbb789991..69407b70821e 100644
> --- a/arch/arm64/include/asm/jump_label.h
> +++ b/arch/arm64/include/asm/jump_label.h
> @@ -15,16 +15,19 @@
>  
>  #define JUMP_LABEL_NOP_SIZE		AARCH64_INSN_SIZE
>  
> +#define JUMP_TABLE_ENTRY				\
> +	 ".pushsection	__jump_table, \"aw\"	\n\t"	\
> +	 ".align	3			\n\t"	\
> +	 ".long		1b - ., %l[l_yes] - .	\n\t"	\
> +	 ".quad		%c0 - .			\n\t"	\
> +	 ".popsection				\n\t"
> +
>  static __always_inline bool arch_static_branch(struct static_key * const key,
>  					       const bool branch)
>  {
>  	asm goto(
>  		"1:	nop					\n\t"
> -		 "	.pushsection	__jump_table, \"aw\"	\n\t"
> -		 "	.align		3			\n\t"
> -		 "	.long		1b - ., %l[l_yes] - .	\n\t"
> -		 "	.quad		%c0 - .			\n\t"
> -		 "	.popsection				\n\t"
> +		JUMP_TABLE_ENTRY
>  		 :  :  "i"(&((char *)key)[branch]) :  : l_yes);

If we really need to factor this out, I'd prefer that the JUMP_TABLE_ENTRY()
macro took the label and key as arguments, similar to what we do for
_ASM_EXTABLE_*().

Mark.

>  
>  	return false;
> @@ -37,11 +40,7 @@ static __always_inline bool arch_static_branch_jump(struct static_key * const ke
>  {
>  	asm goto(
>  		"1:	b		%l[l_yes]		\n\t"
> -		 "	.pushsection	__jump_table, \"aw\"	\n\t"
> -		 "	.align		3			\n\t"
> -		 "	.long		1b - ., %l[l_yes] - .	\n\t"
> -		 "	.quad		%c0 - .			\n\t"
> -		 "	.popsection				\n\t"
> +		JUMP_TABLE_ENTRY
>  		 :  :  "i"(&((char *)key)[branch]) :  : l_yes);
>  
>  	return false;
> -- 
> 2.34.1
> 
>