Re: 答复: 答复: [外部邮件] Re: [PATCH] mm/mempool: use static key for boot-time debug enablement

[Vlastimil Babka (SUSE)]

On 5/28/26 14:59, Usama Arif wrote:
> 
> 
> On 28/05/2026 11:50, Li,Rongqing(ACG CCN) wrote:
>> 
>> 
>>>
>>> On 28/05/2026 04:00, Li,Rongqing(ACG CCN) wrote:
>>>
>>> I think 32 bit ARM doesnt?
>> 
>> You are right, 32-bit ARM doesn't. 
>> 
>> However, the correct architectural approach should be fixing the boot sequence 
>> inside arch/arm/ to match arm64 , powerpc and m68k, rather than compromising core MM 
>> code with temporary boilerplate variables.
>> 
>> I prefer to keep the mempool implementation clean. If ARM32 triggers the 
>> warning, the proper remedy is a follow-up patch to align its setup_arch() 
>> ordering.
>> 
>> What do you think?
>> 
> 
> I think it would be a prerequisite rather than a follow up patch inorder to not
> break 32 bit arm. I will let ARM and slab maintainers decide on this. 

Yeah we shouldn't just knowingly break it.
But I kind of doubt the argument that crucial mempool allocations happen
that early in the boot and we could miss errors in them? They are mostly
storage drivers, AFAICS. So I don't think we need the debugging to be
enabled early.