On Mon, Nov 29, 2021 at 03:49:46PM +0000, Joao Martins wrote:
> Hmmm -- if by individual helpers moving to __dev_dax_{pte,pmd,pud}_fault()
> it would be slightly less straighforward. Unless you might mean to move
> to check_vma() (around the dax_alive() check) and that might actually
> remove the opencoding of dax_read_lock in dax_mmap() even.
>
> I would rather prefer that this cleanup around dax_read_{un,}lock is
> a separate patch separate to this series, unless you feel strongly that
> it needs to be part of this set.
Feel free to keep it as-is.
