On 3/1/20 12:55 PM, Dave Chinner wrote: > On Sun, Mar 01, 2020 at 09:50:03AM -0800, Eric Sandeen wrote: >> The changes to xfs_admin which allowed online label setting via >> ioctl had some unintended consequences in terms of changing command >> order and processing. It's going to be somewhat tricky to fix, so >> back it out for now. > > What are the symptoms and behaviour of these "unintended > consequences"? And why are they tricky to fix? Yeah, I should have probably said more in the commit log. https://bugzilla.kernel.org/show_bug.cgi?id=206429 was the first clue, "xfs_admin can't print both label and UUID for mounted filesystems" The main problem is that if /any/ options that trigger xfs_io get specified, they are the only ones that get run: # Try making the changes online, if supported if [ -n "$IO_OPTS" ] && mntpt="$(find_mntpt_for_arg "$1")" then eval xfs_io -x -p xfs_admin $IO_OPTS "$mntpt" test "$?" -eq 0 && exit 0 fi and the non-io / db opts don't get run at all. So sure, we could then move on to the db commands, but we actually built them all up along the way as well: l) DB_OPTS=$DB_OPTS" -r -c label" IO_OPTS=$IO_OPTS" -r -c label" ;; so we'd need to keep those separate, and not re-run them in db. And another thing that I struggled with was preserving the order; you'd kind of expect that if you specify commands in a certain order they'd be executed in that order, and that used to be true. Now it's not, even if we don't exit in the "if IO_OPTS" case above. So I experimented with building up an array of commands, invoking xfs_db or xfs_io one command at a time as needed for each, and ... it just got worse and worse, TBH. It's broken now, and so far a clean solution isn't evident, and I hate to leave it broken across another release. -Eric
