On 8/14/24 09:33, Yafang Shao wrote:
> On Wed, Aug 14, 2024 at 1:27 PM Christoph Hellwig <hch@xxxxxxxxxxxxx> wrote:
>>
>> > There are already memalloc_noreclaim_{save,restore} which imply __GFP_MEMALLOC:
>> >
>> > memalloc_noreclaim_save - Marks implicit __GFP_MEMALLOC scope.
>>
>> .. and those are horrible misnamed :(
Yes I agree, sorry about that.
> What about renaming it to memalloc_memalloc_save ?
While it looks weird, it could be indeed better than the current name. It's
not obvious, so it should force the user to read the description.
memalloc_noreclaim_save() might look too obviously "this disables reclaim"
but it's misleading as that's not the full story of PF_MEMALLOC.
>>
>> If we can't even keep our APIs consistently name, who is supposed
>> to understand all this?
>>
>
