[RFC v3 03/24] fs: distinguish between user initiated freeze and kernel initiated freeze

[Luis Chamberlain]

On Wed, Jan 18, 2023 at 10:28:12AM +0100, Jan Kara wrote:
> On Tue 17-01-23 18:25:40, Darrick J. Wong wrote:
> > [add linux-xfs to cc on this one]
> > 
> > On Fri, Jan 13, 2023 at 04:33:48PM -0800, Luis Chamberlain wrote:
> > > Userspace can initiate a freeze call using ioctls. If the kernel decides
> > > to freeze a filesystem later it must be able to distinguish if userspace
> > > had initiated the freeze, so that it does not unfreeze it later
> > > automatically on resume.
> > 
> > Hm.  Zooming out a bit here, I want to think about how kernel freezes
> > should behave...
> > 
> > > Likewise if the kernel is initiating a freeze on its own it should *not*
> > > fail to freeze a filesystem if a user had already frozen it on our behalf.
> > 
> > ...because kernel freezes can absorb an existing userspace freeze.  Does
> > that mean that userspace should be prevented from undoing a kernel
> > freeze?  Even in that absorption case?
> > 
> > Also, should we permit multiple kernel freezes of the same fs at the
> > same time?  And if we do allow that, would they nest like freeze used to
> > do?
> > 
> > (My suggestions here are 'yes', 'yes', and '**** no'.)
> 
> Yeah, makes sense to me. So I think the mental model to make things safe
> is that there are two flags - frozen_by_user, frozen_by_kernel - and the
> superblock is kept frozen as long as either of these is set.

Makes sense to me.

  Luis