mount command now shows subvol and subvolid

mount command now shows subvol and subvolid

[Chris Murphy]

Does anyone know when this changed? Maybe it's a 4.2 thing... anyway
it's very much welcome!

/dev/sda2 on /home type btrfs
(rw,relatime,seclabel,space_cache,subvolid=258,subvol=/home)

/dev/sdc1 on /brick0 type btrfs
(rw,relatime,seclabel,space_cache,subvolid=5,subvol=/)

/dev/sde on /brick1 type btrfs
(rw,relatime,seclabel,space_cache,subvolid=5,subvol=/)




-- 
Chris Murphy

Re: mount command now shows subvol and subvolid

[Omar Sandoval]

On Sun, Aug 30, 2015 at 11:17:44AM -0600, Chris Murphy wrote:
> Does anyone know when this changed? Maybe it's a 4.2 thing... anyway
> it's very much welcome!
> 
> /dev/sda2 on /home type btrfs
> (rw,relatime,seclabel,space_cache,subvolid=258,subvol=/home)
> 
> /dev/sdc1 on /brick0 type btrfs
> (rw,relatime,seclabel,space_cache,subvolid=5,subvol=/)
> 
> /dev/sde on /brick1 type btrfs
> (rw,relatime,seclabel,space_cache,subvolid=5,subvol=/)
> 
> 
> 
> 
> -- 
> Chris Murphy

Yup, 4.2 :)

c8d3fe028f64 Btrfs: show subvol= and subvolid= in /proc/mounts
05dbe6837b60 Btrfs: unify subvol= and subvolid= mounting
bb289b7be62d Btrfs: fail on mismatched subvol and subvolid mount options
fa3306595071 Btrfs: clean up error handling in mount_subvol()
e6e4dbe894ef Btrfs: remove all subvol options before mounting top-level
773cd04ec191 Btrfs: lock superblock before remounting for rw subvol

-- 
Omar

Re: mount command now shows subvol and subvolid

[David Sterba]

On Sun, Aug 30, 2015 at 11:17:44AM -0600, Chris Murphy wrote:
> Does anyone know when this changed? Maybe it's a 4.2 thing... anyway
> it's very much welcome!

And another "side effect" is that /proc/self/mountinfo will report the
mounted subvolume, even if it was an implicit mount of non-toplevel
subvolume.

Re: mount command now shows subvol and subvolid

[Duncan]

David Sterba posted on Mon, 31 Aug 2015 15:45:17 +0200 as excerpted:

> On Sun, Aug 30, 2015 at 11:17:44AM -0600, Chris Murphy wrote:
>> Does anyone know when this changed? Maybe it's a 4.2 thing... anyway
>> it's very much welcome!
> 
> And another "side effect" is that /proc/self/mountinfo will report the
> mounted subvolume, even if it was an implicit mount of non-toplevel
> subvolume.

Another very nice side effect, likely accidental, but still very nice... 
is how subvol= ends up being reported for bind-mounts.

I don't use subvolumes here, but I do use multiple separate btrfs, so 
subvolid=5, subvol=/ for all of them, would be technically correct, if 
not particularly useful.  But while the subvolid=5 is indeed constant 
(and actually useful given the following), subvol= turns out to be rather 
more helpful than the / that would normally be expected.

I have a number of bind-mounts.  Formerly mount's output for these wasn't 
particularly helpful, since I ended up with a number of...

/dev/sda5 on <mountpoint> ...

... lines (/dev/sda5 being my rootfs), for instance.  The mountpoint is 
listed, but exactly what part of the filesystem is being bind-mounted 
there isn't (tho /proc/self/mountinfo does contain it).  While the 
filesystem component (/etc/bind, for instance) actually being bind-
mounted was generally what I was interested in, it wasn't actually 
displayed, and could only be indirectly derived from the mountpoint,
by looking up the mountpoint in fstab, for instance, or by using /proc/
self/mountinfo instead of the mount output or /proc/mounts.

With the new subvolume thing, despite the fact that as I said I don't use 
subvolumes so subvolume=/ would be technically correct, the subvol= now 
points at the particular component of the filesystem being bind-mounted 
at the given mountpoint.

So while I still have a bunch of /dev/sda5 on <mountpoint> lines, now the 
subvolume= bit actually tells me what's bind-mounted there, as in (sda5 
is /, sda4 is /var/log)...

/dev/sda5 on /mnt/cbind/etc/bind type btrfs (...,subvol=/etc/bind)
/dev/sda4 on /mnt/cbind/var/log/named type btrfs (...,subvol=/named)

Again, technically the subvol is actually / in both cases, since I don't 
use subvols.  And the subvolid is indeed 5, as expected.  But the 
actually reported subvol=/etc/bind is *so* much more helpful than the 
subvol=/ that I might have expected. =:^)

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

Re: mount command now shows subvol and subvolid

[Karel Zak]

On Tue, Sep 01, 2015 at 07:17:53AM +0000, Duncan wrote:
> David Sterba posted on Mon, 31 Aug 2015 15:45:17 +0200 as excerpted:
> 
> > On Sun, Aug 30, 2015 at 11:17:44AM -0600, Chris Murphy wrote:
> >> Does anyone know when this changed? Maybe it's a 4.2 thing... anyway
> >> it's very much welcome!
> > 
> > And another "side effect" is that /proc/self/mountinfo will report the
> > mounted subvolume, even if it was an implicit mount of non-toplevel
> > subvolume.
> 
> Another very nice side effect, likely accidental, but still very nice... 
> is how subvol= ends up being reported for bind-mounts.
> 
> I don't use subvolumes here, but I do use multiple separate btrfs, so 
> subvolid=5, subvol=/ for all of them, would be technically correct, if 
> not particularly useful.  But while the subvolid=5 is indeed constant 
> (and actually useful given the following), subvol= turns out to be rather 
> more helpful than the / that would normally be expected.
> 
> I have a number of bind-mounts.  Formerly mount's output for these wasn't 
> particularly helpful, since I ended up with a number of...
> 
> /dev/sda5 on <mountpoint> ...

 Try findmnt(8) rather than mount(8).

    Karel

-- 
 Karel Zak  <kzak@redhat.com>
 http://karelzak.blogspot.com