Re: [PATCH v4 0/2] fix MADV_COLLAPSE issue if THP settings are disabled

[David Hildenbrand]

On 25.06.25 12:15, Lorenzo Stoakes wrote:
> On Wed, Jun 25, 2025 at 12:07:12PM +0200, David Hildenbrand wrote:
> > > Yes, I don't mean it will prevent madvise_collapse(), just as you said
> > > that it could be problematic (it's horrible to try to collapse 512MB).
> >
> > Well, assume you have a VM at that is 2 GiB and could use 4 THPs. It's
> > stupid, but there might be some selected use cases where it's not completely
> > stupid.
>
> I guess we limit the stupidity by MADV_COLLAPSE working on a PMD-aligned range
> so it'll be a no-op for an attempt to collapse 2 MiB (unless I'm misreading the
> code).

At least the doc says: "MADV_COLLAPSE will automatically clamp the 
provided range to be hugepage-aligned.", and skimming over the code that 
seems to be the case.

Which makes sense to me: never collapse outside the provided range.

So if someone would ever specify something smaller (e.g., 2MiB aligned 
range with 512MiB PMD size) we could decide to collapse a 2MiB THP there 
instead. Right now, it would be ignored.

Guess the rule is simply to not collapse outside of the requested range. 
Maybe that rule can be used to support collapsing of smaller THPs in the 
future as well. Maybe  (can of worms) :)

--
Cheers,

David / dhildenb