On 1/30/20 2:19 PM, Darrick J. Wong wrote: > On Thu, Jan 30, 2020 at 01:32:30PM -0600, Eric Sandeen wrote: >> On 1/23/20 6:17 PM, Darrick J. Wong wrote: >>> From: Darrick J. Wong <darrick.wong@xxxxxxxxxx> >>> >>> Make sure the root inode gets created where repair thinks it should be >>> created. >> >> Actual mkfs-time location calculation is still completely separate from >> the code in xfs_ialloc_calc_rootino though, right? Maybe there's nothing >> to do about that. > > Correct, because proto.c uses the regular inode allocation routines to > create the root inode, and mkfs doesn't have the ability to compute the > root inode and Make It So. > >> I mostly find myself wondering what a user will do next if this check fails. > > Complain. :) > > To be fair, if there was a mismatch prior to this patch, the user would > end up with a filesystem that formats fine, mounts ok, and explodes in > xfs_repair. Better we fail early than have repair shred the filesystem > after they've loaded up their production data and deleted the backups. Oh, for sure. But it's still: "haha something went wrong, HAND PLONK" I guess the savvy user would come to the list or file a bug. (Hm, you know, a routine to dump cfg-> out for debugging would be neat.) -Eric
