Age | Commit message (Collapse) | Author |
|
BCJ (branch/call/jump) filters in XZ will convert relative jumps into
absolute addresses to increase the redundancy. Since binary code has
plenty of relative jumps, this often improves compression ratio of .xz.
Preliminary test shows an improvement of 2% in the compression ratio
when compressing ELF files, which is already around 25% compression
ratio when using xz.
This patch enables the BCJ for x86, arm and thumb code filters (the
actual data being compressed, not the supported architectures where
the filter can run). This increases the size of the .a in ~12 KiB.
Bug: 27817327
TEST=Used xz-embedded to decompress a file compressed with `xz --x86 --lzma2`
Change-Id: Iadb9dd8174ba404a6dfa8e9bddecd878af4b2fcf
|
|
These files are part of the xz-embedded latest stable release but were
missing in this repo.
Bug: 27817327
TEST=mmma external/lzma
Change-Id: I949decdeb53c83ed6a14d1ce468e018c0f4533d7
|
|
The libxz_host module isn't actually being used.
Bug: 24619596
TEST=mma
Change-Id: I824b84ad97ce588ab913b79f42c0bc1d3febf265
|
|
Change-Id: If3a297bf3aa58df9ea18845a94787bb55384fe1b
|
|
Change-Id: I1168abf8ef4a488d8ab3cdc53f4a3052c78d6bb7
|
|
Change-Id: I8d6cda9748273826b6b6ba631c9e830cb385af5d
|