Re: [PATCH v2 1/4] fs: add owner of freeze/thaw

[Christian Brauner]

On Thu, Apr 03, 2025 at 04:56:57PM +0200, Jan Kara wrote:
> On Wed 02-04-25 16:07:31, Christian Brauner wrote:
> > For some kernel subsystems it is paramount that they are guaranteed that
> > they are the owner of the freeze to avoid any risk of deadlocks. This is
> > the case for the power subsystem. Enable it to recognize whether it did
> > actually freeze the filesystem.
> > 
> > If userspace has 10 filesystems and suspend/hibernate manges to freeze 5
> > and then fails on the 6th for whatever odd reason (current or future)
> > then power needs to undo the freeze of the first 5 filesystems. It can't
> > just walk the list again because while it's unlikely that a new
> > filesystem got added in the meantime it still cannot tell which
> > filesystems the power subsystem actually managed to get a freeze
> > reference count on that needs to be dropped during thaw.
> > 
> > There's various ways out of this ugliness. For example, record the
> > filesystems the power subsystem managed to freeze on a temporary list in
> > the callbacks and then walk that list backwards during thaw to undo the
> > freezing or make sure that the power subsystem just actually exclusively
> > freezes things it can freeze and marking such filesystems as being owned
> > by power for the duration of the suspend or resume cycle. I opted for
> > the latter as that seemed the clean thing to do even if it means more
> > code changes.
> > 
> > Signed-off-by: Christian Brauner <brauner@xxxxxxxxxx>
> 
> I have realized a slight catch with this approach that if hibernation races
> with filesystem freezing (e.g. DM reconfiguration), then hibernation need
> not freeze a filesystem because it's already frozen but userspace may thaw
> the filesystem before hibernation actually happens (relatively harmless).
> If the race happens the other way around, DM reconfiguration may
> unexpectedly fail with EBUSY (rather unexpected). So somehow tracking which
> fs was frozen by suspend while properly nesting with other freeze users may
> be actually a better approach (maybe just a sb flag even though it's
> somewhat hacky?).

The approach that I originally had was to add FREEZE_POWER which adds a
simple boolean into the sb_writers instead of a holder and then this
simply nests with the rest. I'll try to post that diff tomorrow.