xfs: fix incorrect decoding in xchk_btree_cur_fsbno
authorDarrick J. Wong <djwong@kernel.org>
Wed, 13 Oct 2021 22:11:02 +0000 (15:11 -0700)
committerDarrick J. Wong <djwong@kernel.org>
Tue, 19 Oct 2021 18:45:13 +0000 (11:45 -0700)
commit94a14cfd3b6e639c3fb90994ea24e8513f1b0cce
tree3bf86d8cbab1285deea2102162b94a6fe782bdc9
parent892a666fafa19ab04b5e948f6c92f98f1dafb489
xfs: fix incorrect decoding in xchk_btree_cur_fsbno

During review of subsequent patches, Dave and I noticed that this
function doesn't work quite right -- accessing cur->bc_ino depends on
the ROOT_IN_INODE flag, not LONG_PTRS.  Fix that and the parentheses
isssue.  While we're at it, remove the piece that accesses cur->bc_ag,
because block 0 of an AG is never part of a btree.

Note: This changes the btree scrubber tracepoints behavior -- if the
cursor has no buffer for a certain level, it will always report
NULLFSBLOCK.  It is assumed that anyone tracing the online fsck code
will also be tracing xchk_start/xchk_done or otherwise be aware of what
exactly is being scrubbed.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
fs/xfs/scrub/trace.c