Re: [PATCH 3/4] s390/percpu: Drop unused output register from arch_this_cpu_to_op()

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] s390/percpu: unpredictable LAN/LAO behavior from R1==R3
--

commit 657f51d5e5b9873bdbdaca586024d213577edd7b
Author: Heiko Carstens <hca@linux.ibm.com>

s390/percpu: Drop unused output register from arch_this_cpu_to_op()

The lan/lang/lao/laog instructions return the old memory content, but
arch_this_cpu_to_op() does not need it. This commit changes the val__
constraint to use it as an input/output register instead of a dedicated
old__ register, aiming to reduce register pressure.

> diff --git a/arch/s390/include/asm/percpu.h b/arch/s390/include/asm/percpu.h
> index fd0a3a9951ce7..7d87e0a262c37 100644
> --- a/arch/s390/include/asm/percpu.h
> +++ b/arch/s390/include/asm/percpu.h
> @@ -257,18 +257,18 @@ do {									\
>  	typedef typeof(pcp) pcp_op_T__; 				\
>  	union register_pair rp__;					\
>  	pcp_op_T__ val__ = (val);					\
> -	pcp_op_T__ old__, *ptr__;					\
> +	pcp_op_T__ *ptr__;						\
>  									\
>  	ptr__ = PERCPU_PTR(&(pcp));					\
>  	asm_inline volatile(						\
>  		__PCPU_BEGIN("%[lcreg]","%[lcoff]","%[ptr__]",		\
>  			     "%[pair__]","%N[pair__]")			\
> -		op "    %[old__],%[val__],0(%N[pair__])\n"		\
> +		op "    %[val__],%[val__],0(%N[pair__])\n"		\

[Severity: High]
Does this inline assembly string force the compiler to emit the atomic
instruction using the identical register for both the r1 and r3 operands?

According to the z/Architecture Principles of Operation, if r1 and r3
designate the same register in laa, lan, and lao instructions, the results
are unpredictable.

By hardcoding `%[val__],%[val__]`, the exact same register is used for both r1
(the output destination for the old value) and r3 (the input addend/mask). Does
this emit unpredictable assembly for invocations of this_cpu_and_* or
this_cpu_or_*?

Also, arch_this_cpu_add() was modified similarly just one commit earlier in
this series ("s390/percpu: Drop unused output register from
arch_this_cpu_add()"). Does that macro have the same unpredictability
regression for laa?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261012114249.2898572-1-hca@linux.ibm.com?part=3