On Fri, Dec 04, 2020 at 06:45:56PM +0300, Konstantin Komarov wrote: > This adds compression > > Signed-off-by: Konstantin Komarov <almaz.alexandrovich@xxxxxxxxxxxxxxxxxxxx> > --- > fs/ntfs3/lib/common_defs.h | 196 +++++++++++ > fs/ntfs3/lib/decompress_common.c | 314 +++++++++++++++++ > fs/ntfs3/lib/decompress_common.h | 558 +++++++++++++++++++++++++++++++ > fs/ntfs3/lib/lzx_common.c | 204 +++++++++++ > fs/ntfs3/lib/lzx_common.h | 31 ++ > fs/ntfs3/lib/lzx_constants.h | 113 +++++++ > fs/ntfs3/lib/lzx_decompress.c | 553 ++++++++++++++++++++++++++++++ > fs/ntfs3/lib/xpress_constants.h | 23 ++ > fs/ntfs3/lib/xpress_decompress.c | 165 +++++++++ > fs/ntfs3/lznt.c | 452 +++++++++++++++++++++++++ > 10 files changed, 2609 insertions(+) > create mode 100644 fs/ntfs3/lib/common_defs.h > create mode 100644 fs/ntfs3/lib/decompress_common.c > create mode 100644 fs/ntfs3/lib/decompress_common.h > create mode 100644 fs/ntfs3/lib/lzx_common.c > create mode 100644 fs/ntfs3/lib/lzx_common.h > create mode 100644 fs/ntfs3/lib/lzx_constants.h > create mode 100644 fs/ntfs3/lib/lzx_decompress.c > create mode 100644 fs/ntfs3/lib/xpress_constants.h > create mode 100644 fs/ntfs3/lib/xpress_decompress.c > create mode 100644 fs/ntfs3/lznt.c This really could use a much better commit message. Including mentioning where the LZX and XPRESS decompression code came from (https://github.com/ebiggers/ntfs-3g-system-compression). Also note you've marked the files as "SPDX-License-Identifier: GPL-2.0", but they really are "SPDX-License-Identifier: GPL-2.0-or-later". Also I still think you should consider using the simpler version from ntfs-3g-system-compression commit 3ddd227ee8e3, which I had originally intended to be included in NTFS-3G itself. That version was fewer lines of code and fewer files, as it was simplified for decompression-only. The latest version (the one you're using) is shared with a project that also implements compression (so that I can more easily maintain both projects), so it's more complex than needed for decompression-only support. But in the kernel context it may make sense to go with a simpler version. There are a few performance optimizations you'd miss by going with the older version, but they weren't too significant, and probably you don't need to squeeze out every bit of performance possible when reading XPRESS and LZX-compressed files in this context? - Eric
