Re: [PATCH RFC v3 01/19] mm: thread user_addr through page allocator for cache-friendly zeroing

[Michael S. Tsirkin]

On Fri, Apr 24, 2026 at 08:30:40AM -0400, Gregory Price wrote:
> On Fri, Apr 24, 2026 at 09:28:36AM +0200, David Hildenbrand (Arm) wrote:
> > On 4/23/26 16:57, Gregory Price wrote:
> > > On Thu, Apr 23, 2026 at 04:13:50PM +0200, David Hildenbrand (Arm) wrote:
> > >> On 4/23/26 15:42, Gregory Price wrote:
> > >>
> > >> Maybe we could forward the vma+addr here and call a vma_alloc_froze_folio() if
> > >> we have a VMA+addr to have a clean interface.
> > >>
> > >> But really, that hugetlb code is rather messy. I'd vote for leaving hugetlb
> > >> alone on a v1, and focusing on non-hugetlb first.
> > >>
> > > 
> > > If we're ok increasing the buddy surface this way, then I'd vote for
> > > only updating the exact interfaces that MST needs to update for his use
> > > case in a base set of patches, and then have each additional updated
> > > location (or logical set of locations) updated in follow-ups.
> > > 
> > > My initial go around with this - the patch was hard to read at best.
> > > 
> > > But I also think we should also seriously consider not increasing the
> > > surface of the buddy. 
> > 
> > Exactly, that's why I am saying that vma_alloc_folio() is the only external
> > interface people should be using with a user address. all other _noprof helpers
> > are supposed to be internal.
> > 
> > For hugetlb, we might need another interface for frozen folios later, which is
> > why I suggest to defer that.
> > 
> 
> Yeah I follow now - we're of the same mind on all this then.
> 
> ~Gregory


OK v4 incoming, thanks.