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

[David Howells]

Lennart Poettering <mzxreary@xxxxxxxxxxx> wrote:

> systemd cares about all mount points in PID1's mount namespace.
> 
> The fact that mount tables can grow large is why we want something
> better than constantly reparsing the whole /proc/self/mountinfo. But
> filtering subsets of that is something we don't really care about.

With the notifications stuff I've done, you can do, for example:

	pipe2(pipefd, O_NOTIFICATION_PIPE);
	ioctl(pipefd[0], IOC_WATCH_QUEUE_SET_SIZE, 256);
	watch_mount(AT_FDCWD, "/", 0, pipefd[0], 0x02);

And that will catch all mount object changes in the subtree rooted at the
given path, in this case "/".

If you want to limit it to just the notifications on that mount, you would
need to install a filter:

	struct watch_notification_filter filter = {
		.nr_filters	= 1,
		.filters = {
			[0]	= {
				.type		= WATCH_TYPE_MOUNT_NOTIFY,
				.subtype_filter[0]= UINT_MAX,
				.info_mask	= NOTIFY_MOUNT_IS_RECURSIVE,
				.info_filter	= 0,
			},
		},
	};
	ioctl(fd, IOC_WATCH_QUEUE_SET_FILTER, &filter);

Note that this doesn't monitor for superblock changes and events.  They must
be watched individually with something like:

	watch_sb(AT_FDCWD, "/afs", AT_NO_AUTOMOUNT, pipefd[0], 0x27);

David