On Tue, Apr 29, 2025 at 02:22:11PM +0200, Christoph Hellwig wrote: > So I guess the property variant didn't work out based on the replies > from Darrick last round? Correct. Calling setxattr on the property wouldn't call xfs_set_max_atomic_write_opt to change the operational limit, and I don't want to start up the whole magic xattrs discussion again by making xfs_xattr_set do extra things if name=="trusted.xfs:max_atomic_write". (Though iirc you were the most opposed to magic xattrs, and my opposition is 45% yuck factor and 55% "hch will nak it anyway" ;) ) --D
