On Wed, Feb 26, 2020 at 09:19:53AM -0800, Eric Sandeen wrote: > On 2/4/20 4:46 PM, Darrick J. Wong wrote: > > From: Darrick J. Wong <darrick.wong@xxxxxxxxxx> > > > > xfs_repair has a very old check that evidently excuses the AG 0 inode > > btrees pointing to blocks that are already marked XR_E_INUSE_FS* (e.g. > > AG headers). mkfs never formats filesystems that way and it looks like > > an error, so purge the check. After this, we always complain if inodes > > overlap with AG headers because that should never happen. > > On a previous version, you and Brian had a fairly long conversation about > the warning this presents, and how it doesn't tell the user what to do > about it, and how the warning will persist, and may generate bug reports > or questions. > > It sounded like you had a plan to address that, which does not seem to be > present in this patch? So I'm not sure Brian's concerns have been resolved > yet. I'm confused about "the warning this presents" -- are you talking about this patch specifically, where we couldn't figure out the weird masking behavior that dated back to 2001 and the hysterical raisins? Or are you referring to Brian's criticism of earlier versions of this series that would whine about our root inode computation not leading to the root inode without actually telling the user what to do about it? If it's the second, then I the answer is that I added another patch ("xfs_repair: try to correct sb_unit value from secondaries") to try to recover a working sunit value from the backup superblocks, or try some power of two guesses to see if we find one that matches, and then reset the value to something that will make the computation work again. --D > > -Eric