Re: [PATCH v6 1/8] x86/runtime-const: Introduce runtime_const_mask_32()

Hello,

On 7/28/2026 11:07 AM, [email protected] wrote:
>> diff --git a/arch/x86/include/asm/runtime-const.h b/arch/x86/include/asm/runtime-const.h
>> index 4cd94fdcb45e2..b13f7036c1c9b 100644
>> --- a/arch/x86/include/asm/runtime-const.h
>> +++ b/arch/x86/include/asm/runtime-const.h
> [ ... ]
>> @@ -41,6 +41,15 @@
>>  		:"+r" (__ret));					\
>>  	__ret; })
>>  
>> +#define runtime_const_mask_32(val, sym) ({			\
>> +	typeof(0u+(val)) __ret = (val);				\
>> +	asm_inline("and $0x12345678, %k0\n1:\n"				\
>> +		   ".pushsection runtime_mask_" #sym ",\"a\"\n\t"\
>> +		   ".long 1b - 4 - .\n"				\
>> +		   ".popsection"				\
>> +		   : "+r" (__ret));				\
>> +	__ret; })
>> +
> 
> [Severity: Medium]
> Will this cause a build regression on non-x86 architectures when the generic
> code starts using runtime_const_mask_32()?

That only happens at Patch 8 and by Patch 7, we have the generic
fallback. Everything is aligned before the first user is introduced.

> 
> The commit message mentions this will be used by the futex hash computation in
> a subsequent commit. However, there doesn't appear to be a generic fallback
> for this macro in include/asm-generic/runtime-const.h.
> 
> Without a generic fallback, it seems architectures that don't support or
> haven't yet implemented runtime constants would fail to compile once the
> generic futex code uses this macro.
> 
> Should a fallback like the following be introduced in
> include/asm-generic/runtime-const.h alongside this x86 implementation to
> maintain cross-architecture compatibility and strict bisectability?
> 
> #define runtime_const_mask_32(val, sym) ((val) & (sym))

Done in Patch 7. Bisectibility is preserved by having the plumbing
in place (in some order) before first user is introduced in Patch 8.

-- 
Thanks and Regards,
Prateek