Re: [PATCH] userfaultfd: mark uffd_wp regardless of VM_WRITE flag

[Peter Xu <peterx@redhat.com>]

Hi, Andrew,

On Wed, Mar 16, 2022 at 03:05:53PM -0700, Andrew Morton wrote:
> 
> As I understand it, this patch (below) is still good to merge upstream,
> although Peter hasn't acked it (please).

Thanks for asking.  I didn't ack because I saw that it's queued a long time
ago into -mm, and also it's in -next for a long time too (my new uffd-wp
patchset is rebased to this one already).

I normally assume that means you read and ack that patch already, so if I
didn't see anything obviously wrong I'll just keep silent. Please let me
know if that's not the expected behavior..

So here it is..

Acked-by: Peter Xu <peterx@redhat.com>

> 
> And a whole bunch of followup patches are being thought about, but have
> yet to eventuate.

Is there a pointer/subject?

> 
> Do we feel that this patch warrants the cc:stable?  I'm suspecting
> "no", as it isn't clear that the use-case is really legitimate at this
> time?

Right. Uffd-wp+mprotect usage is IMHO not a major use case for uffd-wp per
my knowledge. At least I didn't even expect both work together, not until
Nadav started working on some of the problems..

Said that, for this specific case it should be literally only changing the
behavior of anonymous UFFD-WP && !WRITE case but nothing else (please
correct me otherwise..), then IMHO it's pretty safe to copy stable too
especially for the cleanly applicable branches.

Thanks,

-- 
Peter Xu