[0/7] Cleanup btrfs_item_* related helpers

Message

Josef Bacik Oct. 21, 2021, 6:58 p.m. UTC

Hello,

We've got a lot of odd patterns with our btrfs_item* helpers.  In a lot of
places we use teh btrfs_item_*_nr() helpers, in some places we get the
btrfs_item and then use the direct helpers on that.  Our btrfs_item_key()
helpers only take the slot.

Fix this up by using a leaf+slot combination everywhere, and only use the offset
calculations in the helpers.

This gives us a net deletion of about 40 lines, and also allows me to more
easily change the btrfs_item math when I change the size of btrfs_header in the
future without needing to update all the callsites to pass in an extent buffer.

Thanks,

Josef Bacik (7):
  btrfs: use btrfs_item_size_nr/btrfs_item_offset_nr everywhere
  btrfs: add btrfs_set_item_*_nr() helpers
  btrfs: make btrfs_file_extent_inline_item_len take a slot
  btrfs: introduce item_nr token variant helpers
  btrfs: drop the _nr from the item helpers
  btrfs: remove the btrfs_item_end() helper
  btrfs: rename btrfs_item_end_nr to btrfs_item_data_end

 fs/btrfs/backref.c                   |  16 ++-
 fs/btrfs/ctree.c                     | 148 +++++++++++----------------
 fs/btrfs/ctree.h                     |  55 ++++++----
 fs/btrfs/dev-replace.c               |   4 +-
 fs/btrfs/dir-item.c                  |  12 +--
 fs/btrfs/extent-tree.c               |  14 +--
 fs/btrfs/file-item.c                 |  24 ++---
 fs/btrfs/inode-item.c                |  14 ++-
 fs/btrfs/inode.c                     |   3 +-
 fs/btrfs/ioctl.c                     |   6 +-
 fs/btrfs/print-tree.c                |   8 +-
 fs/btrfs/props.c                     |   2 +-
 fs/btrfs/ref-verify.c                |   2 +-
 fs/btrfs/reflink.c                   |   2 +-
 fs/btrfs/relocation.c                |   2 +-
 fs/btrfs/root-tree.c                 |   4 +-
 fs/btrfs/scrub.c                     |   2 +-
 fs/btrfs/send.c                      |  18 ++--
 fs/btrfs/tests/extent-buffer-tests.c |  17 +--
 fs/btrfs/tree-checker.c              |  56 +++++-----
 fs/btrfs/tree-log.c                  |  34 +++---
 fs/btrfs/uuid-tree.c                 |  10 +-
 fs/btrfs/verity.c                    |   2 +-
 fs/btrfs/volumes.c                   |   6 +-
 fs/btrfs/xattr.c                     |   8 +-
 25 files changed, 215 insertions(+), 254 deletions(-)

Comments

David Sterba Nov. 8, 2021, 7:30 p.m. UTC | #1

On Thu, Oct 21, 2021 at 02:58:30PM -0400, Josef Bacik wrote:
> Hello,
> 
> We've got a lot of odd patterns with our btrfs_item* helpers.  In a lot of
> places we use teh btrfs_item_*_nr() helpers, in some places we get the
> btrfs_item and then use the direct helpers on that.  Our btrfs_item_key()
> helpers only take the slot.
> 
> Fix this up by using a leaf+slot combination everywhere, and only use the offset
> calculations in the helpers.
> 
> This gives us a net deletion of about 40 lines, and also allows me to more
> easily change the btrfs_item math when I change the size of btrfs_header in the
> future without needing to update all the callsites to pass in an extent buffer.
> 
> Thanks,
> 
> Josef Bacik (7):
>   btrfs: use btrfs_item_size_nr/btrfs_item_offset_nr everywhere
>   btrfs: add btrfs_set_item_*_nr() helpers
>   btrfs: make btrfs_file_extent_inline_item_len take a slot
>   btrfs: introduce item_nr token variant helpers
>   btrfs: drop the _nr from the item helpers
>   btrfs: remove the btrfs_item_end() helper
>   btrfs: rename btrfs_item_end_nr to btrfs_item_data_end

Added to misc-next, thanks. It's a big conflict surface but hopefully
all incorrect uses of the helpers in the old/new code should be caught
by a mismatching types of the parameters.