On 9/23/19 6:52 PM, Darrick J. Wong wrote: > From: Darrick J. Wong <darrick.wong@xxxxxxxxxx> > > Revert this commit, as it caused periodic regressions in xfs/173 w/ > 1k blocks[1]. > > [1] https://lore.kernel.org/lkml/20190919014602.GN15734@shao2-debian/ > > Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx> Came across this randomly ... since there's no explanation in the changelog as to /why/ this broke things, I think it's due to: static inline bool xfs_alloc_is_userdata(int datatype) { return (datatype & ~XFS_ALLOC_NOBUSY) != 0; } which would previously have returned true if (datatype & XFS_ALLOC_USERDATA), but without the any flag set returned false. So failing to set XFS_ALLOC_USERDATA indirectly changed the result of this test. So, you're welcome, future code archaeologists. :) -Eric
