On Wed, Jul 31, 2024 at 08:41:58AM -0700, Christoph Hellwig wrote: > On Tue, Jul 30, 2024 at 03:37:25PM -0700, Darrick J. Wong wrote: > > On Tue, Jul 30, 2024 at 02:41:41PM -0700, Christoph Hellwig wrote: > > > Is this really a xfs_spaceman thingy? The code itself looks fine, > > > but the association with space management is kinda weird. > > > > I dunno. It could just as easily go into xfs_io I suppose; the tiny > > advantage of putting it in spaceman is that spaceman grabs the > > fsgeometry structure on file open so we don't have to do that again. > > For the online changes xfs_io seems ok, and for the offline ones xfs_db > seems like a perfevt fit anyway. If fsprops can be managed both online and offline, then xfs_admin is probably the right user facing interface to document. i.e. We already have xfs_admin vectoring between xfs_io when the filesystem is online and xfs_db when the filesystem is offline to do things like change UUIDs and labels. This would make setting fsprops somewhat simpler for admins and scripts as they then only have to learn/code one mechanism instead of two... -Dave. -- Dave Chinner david@xxxxxxxxxxxxx
