Re: [PATCH] mm/page_alloc: Clarify some migratetype fallback code

[Vlastimil Babka]

On 2/18/25 21:38, Johannes Weiner wrote:
> It still makes sense to have the two passes, though, right? One pass
> where we try to steal a whole block starting from the biggest buddies;
> and then one pass where we try to steal an individual page starting
> from the smallest buddies.

Yes.

> Something like this, completely untested draft:

Looks good to me!