Re: [RFC PATCH 02/13] arm64: preempt: Treat should_resched() as unlikely

[Jinjie Ruan <ruanjinjie@huawei.com>]

在 2026/7/28 20:38, Mark Rutland 写道:
> The arm64 implementation of should_resched() doesn't treat the need to
> resched as unlikely, while all other implementations do, e.g.
> 
> * asm-generic since:
>   bdb43806589096ac ("sched: Extract the basic add/sub preempt_count modifiers")
> 
> * x86 since:
>   c2daa3bed53a8117 ("sched, x86: Provide a per-cpu preempt_count implementation")
> 
> * s390 since:
>   c360192bf4a8dc72 ("s390/preempt: move preempt_count to the lowcore")
> 
> Do the same for arm64.
> 
> Signed-off-by: Mark Rutland <mark.rutland@arm.com>
> Cc: Ada Couprie Diaz <ada.coupriediaz@arm.com>
> Cc: Ard Biesheuvel <ardb@kernel.org>
> Cc: Catalin Marinas <catalin.marinas@arm.com>
> Cc: Jinjie Ruan <ruanjinjie@huawei.com>
> Cc: Marc Zyngier <maz@kernel.org>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Vladimir Murzin <vladimir.murzin@arm.com>
> Cc: Will Deacon <will@kernel.org>
> Cc: Yang Shi <yang@os.amperecomputing.com>
> ---
>  arch/arm64/include/asm/preempt.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/arm64/include/asm/preempt.h b/arch/arm64/include/asm/preempt.h
> index ca2ad1a8db095..3ef24ab8f3399 100644
> --- a/arch/arm64/include/asm/preempt.h
> +++ b/arch/arm64/include/asm/preempt.h
> @@ -58,7 +58,7 @@ static inline void __preempt_count_sub(int val)
>  static inline bool should_resched(int preempt_offset)
>  {
>  	u64 pc = READ_ONCE(current_thread_info()->preempt_count);
> -	return pc == preempt_offset;
> +	return unlikely(pc == preempt_offset);

It is rare that preempt_count is 0 or a specific value. Therefore, it is
reasonable to use unlikely.

Reviewed-by: Jinjie Ruan <ruanjinjie@huawei.com>

>  }
>  
>  static inline bool __preempt_count_dec_and_test(void)