Re: [PATCH] mm/huge_memory: bypass THP tuneables for huge pfnmap mappings

["Zi Yan" <ziy@nvidia.com>]

On Mon Sep 7, 2026 at 12:30 PM EDT, Lorenzo Stoakes (ARM) wrote:
> On Mon, Sep 07, 2026 at 06:19:30PM +0200, David Hildenbrand (Arm) wrote:
>> On 9/7/26 18:15, Lorenzo Stoakes (ARM) wrote:
>> > On Mon, Sep 07, 2026 at 06:08:31PM +0200, David Hildenbrand (Arm) wrote:
>> >>
>> >>>
>> >>> That's a good idea! Much better thanks.
>> >>>
>> >>>
>> >>> Yeah, it's an annoying one as kinda want to say 'file-backed VMA' but I think
>> >>> you're right vma_xxx_file() -> vma_file_xxx() is better.
>> >>>
>> >>> Andrew - could you fix that up in place?
>> >>>
>> >>> Obviously if there's further feedback I can respin as well.
>> >>
>> >> Just as a general note I really hate these in-place updates.
>> >>
>> >> For a reviewer perspective (like me stumbling over this) this is just a mess.
>> >>
>> >> Resending with b4 is as easy as it gets.

Probably worth putting it in the process documentation. Always resend if
a patch has any change.

>> >
>> > Ack, you are right to be honest, and I do empathise obviously.
>> >
>> > I will respin these in future!
>> >
>> > Let me know if you want me to respin this to make life a bit easier?
>>
>> Sorry for being grumpy. With 1500 unread mails, playing Hensel and searching for
>> breadcrumbs in my inbox is hard :D
>
> Tsk, that's not grumpy, you're speaking to the king of grumpiness :P
>
>>
>> (I did see some reply from Zi to an mm-unstable patch, but not sure if that was
>> applied in the end -- or whether Andrew fixed up any of that)
>>
>> I just saw my reply to my other mail, maybe a v2 would be best :)
>
> Sure no problem will respin.

The latest version from Andrew's tree is
https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git/commit/?h=mm-everything&id=e384abeb559d10d6505aec053ede9368d81d4c71

It has my renaming suggestion folded in.


-- 
Best Regards,
Yan, Zi