Re: [PATCH 06/13] mm/memory: Add dax_insert_pfn

[Alistair Popple <apopple@nvidia.com>]

David Hildenbrand <david@redhat.com> writes:

> On 27.06.24 02:54, Alistair Popple wrote:
>> Currently to map a DAX page the DAX driver calls vmf_insert_pfn. This
>> creates a special devmap PTE entry for the pfn but does not take a
>> reference on the underlying struct page for the mapping. This is
>> because DAX page refcounts are treated specially, as indicated by the
>> presence of a devmap entry.
>> To allow DAX page refcounts to be managed the same as normal page
>> refcounts introduce dax_insert_pfn. This will take a reference on the
>> underlying page much the same as vmf_insert_page, except it also
>> permits upgrading an existing mapping to be writable if
>> requested/possible.
>
> We have this comparably nasty vmf_insert_mixed() that FS dax abused to
> insert into !VM_MIXED VMAs. Is that abuse now stopping and are there
> maybe ways to get rid of vmf_insert_mixed()?

It's not something I've looked at but quite possibly - there are a
couple of other users of vmf_insert_mixed() that would need to be
removed though. I just added this as dax_insert_pfn() because as an API
it is really specific to the DAX use case. For example a more general
API would likely pass a page/folio.