linux-ext4 - Re: [PATCH 1/4] fs: send uevents for filesystem mount events

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20251224-imitieren-flugtauglich-dcef25c57c8d@brauner>
Date: Wed, 24 Dec 2025 13:47:25 +0100
From: Christian Brauner <brauner@...nel.org>
To: "Darrick J. Wong" <djwong@...nel.org>
Cc: linux-ext4@...r.kernel.org, linux-xfs@...r.kernel.org, hch@....de, 
	linux-fsdevel@...r.kernel.org
Subject: Re: [PATCH 1/4] fs: send uevents for filesystem mount events

On Wed, Dec 17, 2025 at 06:04:29PM -0800, Darrick J. Wong wrote:
> From: Darrick J. Wong <djwong@...nel.org>
> 
> Add the ability to send uevents whenever a filesystem mounts, unmounts,
> or goes down.  This will enable XFS to start daemons whenever a
> filesystem is first mounted.
> 
> Regrettably, we can't wire this directly into get_tree_bdev_flags or
> generic_shutdown_super because not all filesystems set up a kobject
> representation in sysfs, and the VFS has no idea if a filesystem
> actually does that.
> 
> Signed-off-by: "Darrick J. Wong" <djwong@...nel.org>
> ---

I have issues with uevents as a mechanism for this. Uevents are tied to
network namespaces and they are not really namespaced appropriately. Any
filesystem that hooks into this mechanism will spew uevents into the
initial network namespace unconditionally. Any container mountable
filesystem that wants to use this interface will spam the host with
this event though the even is completely useless without appropriate
meta information about the relevant mount namespaces and further
parameters. This is a design dead end going forward imho. So please
let's not do this.

Instead ties this to fanotify which is the right interface for this.
My suggestion would be to tie this to mount namespaces as that's the
appropriate object. Fanotify already supports listening for general
mount/umount events on mount namespaces. So extend it to send filesystem
creation/destruction events so that a caller may listen on the initial
mount namespace - where xfs fses can be mounted - you could even make it
filterable per filesystem type right away.