[PATCH v7 0/3] erofs-utils: introduce xattr name bloom filter

[Jingbo Xu]

changes since v6:
- patch 1: polish license disclaimer; tweak included headers (Gao Xiang)
- patch 2: drop unused `EROFS_XATTR_NAME_LEN_MAX`; polish commit message
  (Gao Xiang)
- patch 3: add warning when failed to calculate hashbit; tweak code of
  assigning `header->h_name_filter` (Gao Xiang)

changes since v5:
- patch 1: update distribution license
- patch 3: rebase to the latest -dev branch

changes since v4:
- patch 3: the feature is refactored and implemented in a more cohesive
  way, that is, the hashbit is calculated in erofs_export_xattr_ibody().
  For xattr_item in the long xattr name prefix style, reconstruct the
  full xattr name first, then derive the corresponding predefined short
  name prefix and finally calculate the hashbit.

changes since v3:
- patch 3: "-Exattr-name-filter" option rather than "--xattr-filter"
  option is newly introduced to enable this feature (Gao Xiang)

changes since v2:
- patch 2: introduce xattr_filter_reserved in on-disk superblock;
  remove EROFS_XATTR_FILTER_MASK
- patch 3: xattr_filter_reserved is always initialized to 0 by default


changes since RFC:
- the number of hash functions is 1, and now it's implemented as:
    xxh32(name, strlen(name), EROFS_XATTR_FILTER_SEED + index),
  where the constant magic number EROFS_XATTR_FILTER_SEED [*] is used to
  give a better spread for the mapping. (Alexander Larsson)
- fix the value of EROFS_FEATURE_COMPAT_XATTR_FILTER; rename
  EROFS_XATTR_BLOOM_* to EROFS_XATTR_FILTER_* (Gao Xiang)


RFC: https://lore.kernel.org/all/20230621083939.128961-1-jefflexu@linux.alibaba.com/
v2: https://lore.kernel.org/all/20230705071017.104130-1-jefflexu@linux.alibaba.com/
v3: https://lore.kernel.org/all/20230712121331.99671-1-jefflexu@linux.alibaba.com/
v4: https://lore.kernel.org/all/20230714025330.42950-1-jefflexu@linux.alibaba.com/
v5: https://lore.kernel.org/all/20230722042414.126630-1-jefflexu@linux.alibaba.com/
v6: https://lore.kernel.org/all/20230829124127.36719-1-jefflexu@linux.alibaba.com/

The xattr bloom filter feature is used to boost the negative xattr
lookup.

Refer to the kernel patch set [*] for more details.

[*] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=fd73a4395d47



Jingbo Xu (3):
  erofs-utils: add xxh32 library
  erofs-utils: update on-disk format for xattr name filter
  erofs-utils: mkfs: enable xattr name filter

 include/erofs/config.h   |   1 +
 include/erofs/internal.h |   1 +
 include/erofs/xxhash.h   |  27 +++++++++
 include/erofs_fs.h       |  10 +++-
 lib/Makefile.am          |   3 +-
 lib/xattr.c              |  65 ++++++++++++++++++++
 lib/xxhash.c             | 126 +++++++++++++++++++++++++++++++++++++++
 mkfs/main.c              |   7 +++
 8 files changed, 237 insertions(+), 3 deletions(-)
 create mode 100644 include/erofs/xxhash.h
 create mode 100644 lib/xxhash.c

-- 
2.19.1.6.gb485710b