Re: [PATCH] fs: add the FIGETFROZEN ioctl call

[Eric Sandeen <sandeen@sandeen.net>]

On 4/18/16 1:20 PM, Florian Margaine wrote:
> On Mon, 2016-04-18 at 11:20 -0400, Eric Sandeen wrote:
>> > 
>> > On 4/14/16 10:17 PM, Dave Chinner wrote:
>>> > > On Thu, Apr 14, 2016 at 09:57:07AM +0200, Florian Margaine wrote:
>>>> > > > This lets userland get the filesystem freezing status, aka
>>>> > > > whether the
>>>> > > > filesystem is frozen or not. This is so that an application can
>>>> > > > know if
>>>> > > > it should freeze the filesystem or if it isn't necessary when
>>>> > > > taking a
>>>> > > > snapshot.
>>> > > 
>>> > > freezing nests, so there is no reason for avoiding a freeze when
>>> > > doing a snapshot. 
>> > 
>> > Sadly, no:
>> > 
>> > # xfs_freeze -f /mnt/test
>> > # xfs_freeze -f /mnt/test
>> > xfs_freeze: cannot freeze filesystem at /mnt/test: Device or resource
>> > busy
>> > 
>> > It used to, but it was broken^Wchanged quite some time ago.
> I guess I can provide a patch for this. Silently making it a no-op if
> the FS is already frozen in ioctl_fsfreeze() should be good enough?

That doesn't work, because you can go:

Process A: Freeze
Process A: Start snapshotting
Process B: Freeze; already frozen, no-op
Process B: Start snapshotting
Process A: Snapshot done, unfreeze
Process B: Now B is snapshotting an unfrozen filesystem

Freeze needs to be nested so that two freeze calls require two thaw
calls to make the filesystem active again, so that any given process
calling freeze can be reasonably sure that it will stay frozen until
it calls unfreeze at some later point.

-Eric