Re: Upcoming: Notifications, FS notifications and fsinfo()

[Linus Torvalds]

On Mon, Apr 6, 2020 at 11:48 AM Lennart Poettering <mzxreary@xxxxxxxxxxx> wrote:
>
> On Mo, 06.04.20 09:34, Linus Torvalds (torvalds@xxxxxxxxxxxxxxxxxxxx) wrote:
>
> > On Mon, Apr 6, 2020 at 2:17 AM Karel Zak <kzak@xxxxxxxxxx> wrote:
> > >
> > > On Fri, Apr 03, 2020 at 04:30:24PM -0400, J. Bruce Fields wrote:
> > > >
> > > > nfs-utils/support/misc/mountpoint.c:check_is_mountpoint() stats the file
> > > > and ".." and returns true if they have different st_dev or the same
> > > > st_ino.  Comparing mount ids sounds better.
> > >
> > > BTW, this traditional st_dev+st_ino way is not reliable for bind mounts.
> > > For mountpoint(1) we search the directory in /proc/self/mountinfo.
> >
> > These days you should probably use openat2() with RESOLVE_NO_XDEV.
>
> Note that opening a file is relatively "heavy" i.e. typically triggers
> autofs and stuff, and results in security checks (which can fail and
> such, and show up in audit).

For the use that Bruce outlined, openat2() with RESOLVE_NO_XDEV is
absolutely the right thing.

He already did the stat() of the file (and ".."), RESOLVE_NO_XDEV is
only an improvement. It's also a lot better than trying to parse
mountinfo.

Now, I don't disagree that a statx() flag to also indicate "that's a
top-level mount" might be a good idea, and may be the right answer for
other cases.

I'm just saying that considering what Bruce does now, RESOLVE_NO_XDEV
sounds like the nobrainer approach, and needs no new support outside
of what we already had for other reasons.

(And O_PATH _may_ or may not be part of what you want to do, it's an
independent separate issue, but automount behavior wrt a O_PATH lookup
is somewhat unclear - see Al's other emails on that subject)

             Linus