xfs: remove xfs_inobp_check()
authorDave Chinner <dchinner@redhat.com>
Mon, 29 Jun 2020 21:49:20 +0000 (14:49 -0700)
committerDarrick J. Wong <darrick.wong@oracle.com>
Tue, 7 Jul 2020 14:15:09 +0000 (07:15 -0700)
commite2705b0304778916db87831217ec642e34d9d9fa
tree6e1045844533a0755a1a4a08b499d1cbfe367a8b
parenta69a1dc2842e4548efca956c86e0816f2662ccb7
xfs: remove xfs_inobp_check()

This debug code is called on every xfs_iflush() call, which then
checks every inode in the buffer for non-zero unlinked list field.
Hence it checks every inode in the cluster buffer every time a
single inode on that cluster it flushed. This is resulting in:

-   38.91%     5.33%  [kernel]  [k] xfs_iflush
   - 17.70% xfs_iflush
      - 9.93% xfs_inobp_check
           4.36% xfs_buf_offset

10% of the CPU time spent flushing inodes is repeatedly checking
unlinked fields in the buffer. We don't need to do this.

The other place we call xfs_inobp_check() is
xfs_iunlink_update_dinode(), and this is after we've done this
assert for the agino we are about to write into that inode:

ASSERT(xfs_verify_agino_or_null(mp, agno, next_agino));

which means we've already checked that the agino we are about to
write is not 0 on debug kernels. The inode buffer verifiers do
everything else we need, so let's just remove this debug code.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
fs/xfs/libxfs/xfs_inode_buf.c
fs/xfs/libxfs/xfs_inode_buf.h
fs/xfs/xfs_inode.c