linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Bhavik Sachdev <b.sachdev1904@gmail.com>
To: Alexander Viro <viro@zeniv.linux.org.uk>,
	Christian Brauner <brauner@kernel.org>
Cc: linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
	Aleksa Sarai <cyphar@cyphar.com>,
	Bhavik Sachdev <b.sachdev1904@gmail.com>,
	Pavel Tikhomirov <ptikhomirov@virtuozzo.com>,
	Jan Kara <jack@suse.cz>, John Garry <john.g.garry@oracle.com>,
	Arnaldo Carvalho de Melo <acme@redhat.com>,
	"Darrick J . Wong" <djwong@kernel.org>,
	Namhyung Kim <namhyung@kernel.org>,
	Ingo Molnar <mingo@kernel.org>, Andrei Vagin <avagin@gmail.com>,
	Alexander Mikhalitsyn <alexander@mihalicyn.com>
Subject: [PATCH v2 0/1] statmount: mountinfo for "unmounted" mounts
Date: Sat, 11 Oct 2025 18:16:10 +0530	[thread overview]
Message-ID: <20251011124753.1820802-1-b.sachdev1904@gmail.com> (raw)

By "unmounted" mounts we mean mounts that have been unmounted using
umount2(mnt, MNT_DETACH) but we still have file descriptors to files on
that mount. We want to add the ability to handle such mounts in CRIU
(Checkpoint/Restore in Userspace).

Currently, we have no way to get mount info for these mounts as they do
not appear in /proc/<pid>/mountinfo and statmount does not work on them.

We solve this problem by introducing a file descriptor parameter to
statmount, along with a STATMOUNT_FD flag. Even if this file descriptor
is on a "unmounted" mount we are still able to get mountinfo for the
mount. We report the mountpoint of the mount to be "[detached]" and
mnt_ns_id to be 0.

v1 of this patchset, took a different approach and introduced a new
umount_mnt_ns, to which "unmounted" mounts would be moved to (instead of
their namespace being NULL) thus allowing them to be still available via
statmount:
https://lore.kernel.org/linux-fsdevel/20251002125422.203598-1-b.sachdev1904@gmail.com/

That approach complicated namespace locking and modified performance
sensitive code.
See: https://lore.kernel.org/linux-fsdevel/7e4d9eb5-6dde-4c59-8ee3-358233f082d0@virtuozzo.com/

Christian also talked about a separate approach of tying the _lifetime_
of the mount namespace to the lifetime of the unmounted mounts through
the passive reference count by moving them to a separate rb_root
`unmounted` in the namespace instance.

This approach has a few problems, some of them are:
1. It further extends the scope of the namespace semaphore.
2. Weird to be able to statmount() via mount id if the mount namespace
is dead.

For a more complete description, see:
https://lore.kernel.org/linux-fsdevel/20251006-erlesen-anlagen-9af59899a969@brauner/

Aleska Sarai also pointed out that this fd based approach is similiar to
the fstatfs(2) which returns information about a mounted filesystem when
given a fd open on that filesystem.
https://lore.kernel.org/linux-fsdevel/2025-10-07-lavish-refried-navy-journey-EqHk9K@cyphar.com/

We use this patchset with CRIU to support checkpoint/restore of
"unmounted" mounts in this pull request:
https://github.com/checkpoint-restore/criu/pull/2754.

All these patches are also available in this branch on github:
https://github.com/bsach64/linux/tree/statmount-fd-v2

Bhavik Sachdev (1):
  statmount: accept fd as a parameter

 fs/namespace.c             | 80 ++++++++++++++++++++++++++++----------
 include/uapi/linux/mount.h |  8 ++++
 2 files changed, 67 insertions(+), 21 deletions(-)

-- 
2.51.0


             reply	other threads:[~2025-10-11 12:48 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-10-11 12:46 Bhavik Sachdev [this message]
2025-10-11 12:46 ` [PATCH v2 1/1] statmount: accept fd as a parameter Bhavik Sachdev

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20251011124753.1820802-1-b.sachdev1904@gmail.com \
    --to=b.sachdev1904@gmail.com \
    --cc=acme@redhat.com \
    --cc=alexander@mihalicyn.com \
    --cc=avagin@gmail.com \
    --cc=brauner@kernel.org \
    --cc=cyphar@cyphar.com \
    --cc=djwong@kernel.org \
    --cc=jack@suse.cz \
    --cc=john.g.garry@oracle.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@kernel.org \
    --cc=namhyung@kernel.org \
    --cc=ptikhomirov@virtuozzo.com \
    --cc=viro@zeniv.linux.org.uk \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).