diff options
author | Yann Collet <cyan@fb.com> | 2022-09-12 21:28:24 (GMT) |
---|---|---|
committer | Yann Collet <cyan@fb.com> | 2022-09-12 21:28:24 (GMT) |
commit | d91d16b48aa22f179f636123869ea4fbe1d8ccdb (patch) | |
tree | 7b31c42ff590424a071ef17afbfb124e4d45b494 /.github/workflows/README.md | |
parent | 9bed6b56e9beb1cf820f01e47e70a13442d33544 (diff) | |
download | lz4-d91d16b48aa22f179f636123869ea4fbe1d8ccdb.zip lz4-d91d16b48aa22f179f636123869ea4fbe1d8ccdb.tar.gz lz4-d91d16b48aa22f179f636123869ea4fbe1d8ccdb.tar.bz2 |
updated documentation : no more issue with 32-bit compilation on recent compilers
Diffstat (limited to '.github/workflows/README.md')
-rw-r--r-- | .github/workflows/README.md | 29 |
1 files changed, 8 insertions, 21 deletions
diff --git a/.github/workflows/README.md b/.github/workflows/README.md index 9ad2d99..2c9071a 100644 --- a/.github/workflows/README.md +++ b/.github/workflows/README.md @@ -6,31 +6,18 @@ This directory contains [GitHub Actions](https://github.com/features/actions) wo For now, `lz4-ubsan-*` uses the `-fsanitize-recover=pointer-overflow` flag: there are known cases of pointer overflow arithmetic within `lz4.c` fast compression. -These cases are not dangerous with current architecture, +These cases are not dangerous with known architecture, but they are not guaranteed to work by the C standard, which means that, in some future, some new architecture or some new compiler -may decide to do something funny that would break this behavior. -Hence it's proper to remove them. -This has been done in `lz4hc.c`. -However, the same attempt in `lz4.c` resulted in massive speed loss, -which is not acceptable to solve a "potential future" problem that does not exist anywhere today. -Therefore, a better work-around will have to be found. +may decide to do something funny that could break this behavior. +Hence, in anticipation, it's better to remove them. +This has been already achieved in `lz4hc.c`. +However, the same attempt in `lz4.c` resulted in massive speed losses, +which is not an acceptable cost for preemptively solving a "potential future" problem +not active anywhere today. +Therefore, a more acceptable work-around will have to be found first. -## C Compilers (`lz4-c-compilers`) - -- Our test doesn't use `gcc-4.5` due to installation issue of its package. (`apt-get install gcc-4.5` fails on GH-Actions VM) - -- Currently, the following 32bit executable tests fail with all versions of `clang`. - - `CC=clang-X CFLAGS='-O3' make V=1 -C tests clean test-lz4c32` - - `CC=clang-X CFLAGS='-O3 -mx32' make V=1 -C tests clean test-lz4c32` - - See [#991](https://github.com/lz4/lz4/issues/991) for details. - -- Currently, the following 32bit executable tests fail with `gcc-11` - - `CC=gcc-11 CFLAGS='-O3' make V=1 -C tests clean test-lz4c32` - - `CC=gcc-11 CFLAGS='-O3 -mx32' make V=1 -C tests clean test-lz4c32` - - See [#991](https://github.com/lz4/lz4/issues/991) for details. - ## cppcheck (`lz4-cppcheck`) |