BTRFS extended attributes mounted on a non-extended-attributes compiled kernel

BTRFS extended attributes mounted on a non-extended-attributes compiled kernel

[Martin]

What happens if...

I have a btrfs that has utilised posix ACLs / extended attributes and I
then subsequently mount that onto a system that does not have the kernel
modules compiled for those features?


Crash and burn?

Or are the extra filesystem features benignly ignored until remounted on
the original system with all the kernel modules?


Thanks,
Martin

Re: BTRFS extended attributes mounted on a non-extended-attributes compiled kernel

[Hugo Mills]

[-- Attachment #1: Type: text/plain, Size: 1361 bytes --]

On Wed, Dec 11, 2013 at 05:51:06PM +0000, Martin wrote:
> What happens if...
> 
> I have a btrfs that has utilised posix ACLs / extended attributes and I
> then subsequently mount that onto a system that does not have the kernel
> modules compiled for those features?
> 
> 
> Crash and burn?
> 
> Or are the extra filesystem features benignly ignored until remounted on
> the original system with all the kernel modules?

   Thinking about it, it's probably going to be OK. btrfs itself
doesn't have any way of turning off EA support, so you'll always have
the EAs managed correctly. The ACL support (which is implemented
through EAs, if I remember correctly) can be turned off, so the
meaning of the ACL EAs will be ignored, but the EA content should
still be there for when you move to an ACL-enabled system again. Note
that this gives you a "convenient" way of bypassing POSIX ACLs, by
switching to a kernel that doesn't enforce them.

   I've not actually tried this, so I'm willing to be proved wrong,
but I'll be surprised if that's the case. :)

   Hugo.

-- 
=== Hugo Mills: hugo@... carfax.org.uk | darksatanic.net | lug.org.uk ===
  PGP key: 65E74AC0 from wwwkeys.eu.pgp.net or http://www.carfax.org.uk
    --- Jazz is the sort of music where no-one plays anything the ---    
                             same way once.                              

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

Re: BTRFS extended attributes mounted on a non-extended-attributes compiled kernel

[Chris Mason]

Quoting Hugo Mills (2013-12-11 14:01:04)
> On Wed, Dec 11, 2013 at 05:51:06PM +0000, Martin wrote:
> > What happens if...
> > 
> > I have a btrfs that has utilised posix ACLs / extended attributes and I
> > then subsequently mount that onto a system that does not have the kernel
> > modules compiled for those features?
> > 
> > 
> > Crash and burn?
> > 
> > Or are the extra filesystem features benignly ignored until remounted on
> > the original system with all the kernel modules?
> 
>    Thinking about it, it's probably going to be OK. btrfs itself
> doesn't have any way of turning off EA support, so you'll always have
> the EAs managed correctly. The ACL support (which is implemented
> through EAs, if I remember correctly) can be turned off, so the
> meaning of the ACL EAs will be ignored, but the EA content should
> still be there for when you move to an ACL-enabled system again. Note
> that this gives you a "convenient" way of bypassing POSIX ACLs, by
> switching to a kernel that doesn't enforce them.
> 
>    I've not actually tried this, so I'm willing to be proved wrong,
> but I'll be surprised if that's the case. :)

I do expect it to silently work.  If you have directories that inherit
acls etc, you might not get fully consistent results if you try to
change anything on the non-xattr/acl kernel.

-chris