Re: [PATCH v3 02/15] userfaultfd: introduce struct mfill_state

[Mike Rapoport]

On Wed, Apr 01, 2026 at 10:44:49AM -0700, Andrew Morton wrote:
> On Wed, 1 Apr 2026 10:37:17 -0700 Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
> 
> > On Wed, 1 Apr 2026 10:36:03 +0300 Mike Rapoport <rppt@xxxxxxxxxx> wrote:
> > 
> > > Here's a fixup (it causes a conflict in patch 4 though).
> > > Andrew, I can send v4 if you prefer.
> > 
> > Looks pretty simple.  "userfaultfd: introduce mfill_get_vma() and
> > mfill_put_vma()" now leaves things like this:
> 
> err, actually, it gets messy.
> 
> Can we please have a v4?  I think there are a few other things pending?

Sent.
 
> I've noted
> https://lkml.kernel.org/r/202603311743.uZYPu1Gn-lkp@xxxxxxxxx

Alice fixed this on the Rust side and the fix is queued to char-misc-linus:

https://lore.kernel.org/all/CAH5fLgg63SYHQ34yqy7o2cA5f2Gh0aNTwTeQkzbWb+kV4Rr3Qw@xxxxxxxxxxxxxx

> https://lkml.kernel.org/r/acqrtd7E6UxrX6Ji@xxxxxxxxxxxxxxxxxxxx

We've been through this since Peter's first posting and I think there's a
consensus about necessity of vm_uffd_ops and the argument was about what
exact API they should have.
 
-- 
Sincerely yours,
Mike.