We have this pattern all over the kernel & userspace:
if (xfs_sb_version_hascrc(&mp->m_sb))
xfs_btree_init_block(mp, bp, XFS_ABTB_CRC_MAGIC, 0, 1,
agno, XFS_BTREE_CRC_BLOCKS);
else
xfs_btree_init_block(mp, bp, XFS_ABTB_MAGIC, 0, 1,
agno, 0);
The last flag (XFS_BTREE_CRC_BLOCKS) can be gleaned from
the features on the superblock via mp just as the if/else does,
so no need to pass that in from the caller. That's patch 1.
Then the difference is simply CRC vs not-CRC magic. This can also
be determined inside the called function by passing in a btree number,
and looking up the proper magic in the xfs_magics array.
patch2 makes xfs_btree_magic() more generic, taking a
crc flag & btnum, instead of just a cursor.
patch3 then makes use of xfs_btree_magic in the block init routine
to determine the proper magic.
With those changes, the if/else can go away, and we can simply call:
xfs_btree_init_block(mp, bp, XFS_BTNUM_BNO, 0, 1, agno, 0);
and let the lower function sort out crc vs. no crc differences
by itself.
It's not a huge reduction in code, but it reduces a lot of the
boilerplate, particularly in growfs code. This may also make it
slightly easier to factor the growfs tree init code into a loop
(not sure about that, yet) and/or to move mkfs & growfs tree init
code into a libxfs/ call.
libxfs/xfs_bmap.c | 19 ++++---------------
libxfs/xfs_bmap_btree.c | 10 ++--------
libxfs/xfs_btree.c | 34 +++++++++++++++++++++-------------
libxfs/xfs_btree.h | 4 ++--
libxfs/xfs_types.h | 2 ++
xfs_fsops.c | 39 +++++++++------------------------------
6 files changed, 40 insertions(+), 68 deletions(-)
-Eric
--
To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
