On Fri, Jan 24, 2025 at 04:49:28PM +0100, Miklos Szeredi wrote: > On Fri, 24 Jan 2025 at 16:38, Christian Brauner <brauner@xxxxxxxxxx> wrote: > > > > On Thu, Jan 23, 2025 at 08:41:07PM +0100, Miklos Szeredi wrote: > > > Notify when mount flags, propagation or idmap changes. > > > > > > Just like attach and detach, no details are given in the notification, only > > > the mount ID. > > > > > > Signed-off-by: Miklos Szeredi <mszeredi@xxxxxxxxxx> > > > --- > > > > I think this is a good next step but I would first go with the minimal > > functionality of notifying about mount topology changes for v6.15. > > I can totally relate to that. I added the fourth patch more as a > "let's see if this can also fit into the current framework". > > > Btw, if we notify in do_remount() on the mount that triggered > > superblock reconfiguration then we also need to trigger in > > vfs_cmd_reconfigure() aka fsconfig(FSCONFIG_CMD_RECONFIGURE) but the > > mount that was used to change superblock options is only available in > > fspick() currently. That would need to be handled. > > No, if we'd want to watch changes on super blocks, then we'd need to > iterate all the mounts of the superblock and notify each. Ah, I remember that old remount had unclear semantics where mount specific and superblock specific options are interleaved. So we would need to notify from do_remount() on mount specific changes. Right, then this change is correct and I agree about the superblock part.
