]>
git.ipfire.org Git - thirdparty/zstd.git/log
Pavel P [Tue, 21 Jan 2025 00:27:09 +0000 (02:27 +0200)]
Use _M_X64 only without mixing with _M_AMD64
Yann Collet [Sun, 19 Jan 2025 15:37:57 +0000 (07:37 -0800)]
Merge pull request #4251 from facebook/x32_bmi2
BMI2 detection in 32-bit mode
Yann Collet [Sun, 19 Jan 2025 08:08:57 +0000 (00:08 -0800)]
no longer limit automated BMI2 detection to x64
this was previously no triggered in x86 32-bit mode,
due to a limitation in `bitstream.h`, that was fixed in #4248.
Now, `bmi2` will be automatically detected and triggered
at compilation time, if the corresponding instruction set is enabled,
even in 32-bit mode.
Also: updated library documentation, to feature STATIC_BMI2 build variable
Yann Collet [Sun, 19 Jan 2025 07:54:51 +0000 (23:54 -0800)]
Merge pull request #4250 from facebook/x86_32_avx2
added a CI test for x86 32-bit + avx2 combination
Yann Collet [Sun, 19 Jan 2025 06:58:03 +0000 (22:58 -0800)]
also add `-mbmi2` to the compilation test
Yann Collet [Sun, 19 Jan 2025 06:48:49 +0000 (22:48 -0800)]
minor: cosmetic, indentation
Yann Collet [Sun, 19 Jan 2025 06:40:18 +0000 (22:40 -0800)]
added -DSTATIC_BMI2=1 for the -mavx2 test
Yann Collet [Sun, 19 Jan 2025 04:43:30 +0000 (20:43 -0800)]
fixed minor static function declaration issue
in AVX2 mode only
Yann Collet [Sun, 19 Jan 2025 04:38:48 +0000 (20:38 -0800)]
added avx2 (x64) compilation test
Yann Collet [Sun, 19 Jan 2025 04:19:55 +0000 (20:19 -0800)]
added a CI test for x86 32-bit + avx2 combination
which is expected to be quite rare, but nonetheless possible.
This test is initially expected to fail, before integration of #4248 fix
Yann Collet [Sun, 19 Jan 2025 06:48:24 +0000 (22:48 -0800)]
Merge pull request #4248 from pps83/dev-bzhi32
Use _bzhi_u32 for 32-bit builds when building with STATIC_BMI2
Yann Collet [Sun, 19 Jan 2025 04:03:16 +0000 (20:03 -0800)]
Merge pull request #4246 from pps83/dev-asmx64-win
[asm] Enable x86_64 asm for windows builds
Yann Collet [Sun, 19 Jan 2025 02:56:46 +0000 (18:56 -0800)]
Merge pull request #4249 from facebook/zds_nonnull
minor: assert that state is not null
Pavel P [Sun, 19 Jan 2025 00:37:35 +0000 (02:37 +0200)]
update sizeof check
Pavel P [Sun, 19 Jan 2025 00:36:00 +0000 (02:36 +0200)]
remove unrelated changes
Pavel P [Sun, 19 Jan 2025 00:34:23 +0000 (02:34 +0200)]
change to BitContainerType
Pavel P [Sat, 18 Jan 2025 21:37:50 +0000 (23:37 +0200)]
handle 32bit size_t when building for x64
Pavel P [Sat, 18 Jan 2025 21:27:19 +0000 (23:27 +0200)]
handle 32bit size_t when building for x64
Yann Collet [Sat, 18 Jan 2025 21:08:04 +0000 (13:08 -0800)]
minor: assert that state is not null
replaces #4016
Yann Collet [Sat, 18 Jan 2025 21:06:07 +0000 (13:06 -0800)]
Merge pull request #4247 from facebook/test_endMsg
Added a human-readable final success message at end of tests
Yann Collet [Sat, 18 Jan 2025 20:22:52 +0000 (12:22 -0800)]
CI: replaced shortest target by check
Pavel P [Sat, 18 Jan 2025 19:32:30 +0000 (21:32 +0200)]
Use _bzhi_u32 for 32-bit builds when building with STATIC_BMI2
`_bzhi_u64` is available only for 64-bit builds, while `BIT_getLowerBits` expects `nbBits` to be less than `BIT_MASK_SIZE` (`BIT_MASK_SIZE` is 32)
Pavel P [Sat, 18 Jan 2025 03:01:43 +0000 (05:01 +0200)]
[asm] Enable x86_64 asm for windows builds
Yann Collet [Sat, 18 Jan 2025 03:30:55 +0000 (19:30 -0800)]
Added a final success message at end of tests
so that a human reading the test log can determine everything was fine without consulting the shell error code.
Also: made `make check` slightly shorter by moving one longer test to `make test`
Yann Collet [Thu, 16 Jan 2025 18:33:18 +0000 (10:33 -0800)]
Merge pull request #4242 from lalakii/dev
add prerelease option
Yann Collet [Thu, 16 Jan 2025 18:33:04 +0000 (10:33 -0800)]
Merge pull request #4235 from facebook/bench_nbThreads
benchmark mode specifies nb of threads with -v
Yann Collet [Thu, 16 Jan 2025 01:56:41 +0000 (17:56 -0800)]
Merge pull request #4238 from szsam/patch-1
fix out-of-bounds array index access
Yann Collet [Thu, 16 Jan 2025 01:56:05 +0000 (17:56 -0800)]
Merge pull request #4232 from facebook/convertSequences_SSE
Improve speed of ZSTD_compressSequencesAndLiterals() using AVX2
Yann Collet [Tue, 14 Jan 2025 23:57:05 +0000 (15:57 -0800)]
restore full equation
do not solve the equation, even though some members cancel each other,
this is done for clarity,
we'll let the compiler do the resolution at compile time.
Yann Collet [Tue, 14 Jan 2025 23:54:10 +0000 (15:54 -0800)]
Alignment instruction ZSTD_ALIGNED() in common/compiler.h
Yann Collet [Tue, 14 Jan 2025 22:56:10 +0000 (14:56 -0800)]
changed code compilation test to employ ZSTD_ARCH_X86_AVX2
Yann Collet [Tue, 14 Jan 2025 22:54:02 +0000 (14:54 -0800)]
removed unused branch
Yann Collet [Tue, 14 Jan 2025 22:44:02 +0000 (14:44 -0800)]
create new compilation macro ZSTD_ARCH_X86_AVX2
Yann Collet [Wed, 8 Jan 2025 22:25:03 +0000 (14:25 -0800)]
minor code doc update
Yann Collet [Wed, 8 Jan 2025 18:51:57 +0000 (10:51 -0800)]
added compilation-time checks to ensure AVX2 code is valid
since it depends on a specific definition of ZSTD_Sequence structure.
Yann Collet [Wed, 8 Jan 2025 07:59:01 +0000 (23:59 -0800)]
removed unused variable
Yann Collet [Wed, 8 Jan 2025 07:45:15 +0000 (23:45 -0800)]
fix minor conversion warning
Yann Collet [Wed, 8 Jan 2025 07:40:49 +0000 (23:40 -0800)]
no need for specialized variant
the branch is not in the hot loop
Yann Collet [Wed, 8 Jan 2025 07:34:19 +0000 (23:34 -0800)]
removed erroneous #includes
that were automatically added by the editor without notification
Yann Collet [Wed, 8 Jan 2025 07:32:05 +0000 (23:32 -0800)]
AVX2 version of ZSTD_get1BlockSummary()
Yann Collet [Wed, 8 Jan 2025 03:34:06 +0000 (19:34 -0800)]
minor +10% speed improvement for scalar ZSTD_get1BlockSummary()
Yann Collet [Wed, 8 Jan 2025 03:29:06 +0000 (19:29 -0800)]
added benchmark for get1BlockSummary()
Yann Collet [Wed, 8 Jan 2025 01:04:12 +0000 (17:04 -0800)]
generalize validation function
Yann Collet [Wed, 8 Jan 2025 00:42:36 +0000 (16:42 -0800)]
control long length within AVX2 implementation
Yann Collet [Tue, 7 Jan 2025 23:51:38 +0000 (15:51 -0800)]
fullbench can run a verification function
compressSequencesAndLiterals: fixed long lengths in scalar mode
Yann Collet [Sun, 29 Dec 2024 10:13:57 +0000 (02:13 -0800)]
initial implementation (incomplete)
needs to take care of long lengths > 65535
Yann Collet [Thu, 16 Jan 2025 01:10:26 +0000 (17:10 -0800)]
Merge pull request #4243 from facebook/fix_ci
fix ci tests
Yann Collet [Thu, 16 Jan 2025 00:31:00 +0000 (16:31 -0800)]
fixed zlib test
new versions of the library generate a different error code
Yann Collet [Thu, 16 Jan 2025 00:05:42 +0000 (16:05 -0800)]
zlib-wrapper: separate test components to better understand what is failing
Yann Collet [Wed, 15 Jan 2025 23:52:27 +0000 (15:52 -0800)]
update installed lib64gcc version
Yann Collet [Wed, 15 Jan 2025 23:39:20 +0000 (15:39 -0800)]
short-test-1: separate tests, giving each a name
to better observe where the problem might be
Yann Collet [Wed, 15 Jan 2025 23:28:23 +0000 (15:28 -0800)]
fixed cmake-build-and-test-check
missing the liblzma library
Yann Collet [Wed, 15 Jan 2025 23:24:27 +0000 (15:24 -0800)]
fix make-external-compressors "only-lzma" test
missing the liblzma-dev package
Yann Collet [Wed, 15 Jan 2025 23:15:27 +0000 (15:15 -0800)]
fix meson-linux build test
missing the liblzma package
Yann Collet [Wed, 15 Jan 2025 22:49:45 +0000 (14:49 -0800)]
fix clang-pgo test
Yann Collet [Tue, 14 Jan 2025 22:33:25 +0000 (14:33 -0800)]
Merge pull request #4234 from MessyHack/dev
[MSVC] Wrong define checked when building for 64bit with Visual Studio
Yann Collet [Tue, 14 Jan 2025 19:12:46 +0000 (11:12 -0800)]
Merge pull request #4237 from facebook/typo_s
fixed an obvious macro name typo
lalakii [Tue, 14 Jan 2025 18:16:42 +0000 (02:16 +0800)]
add prerelease option
Mingjie Shen [Tue, 14 Jan 2025 01:52:06 +0000 (20:52 -0500)]
return error if block delimiter is not found
Sean Bartell [Thu, 20 Jun 2024 20:52:43 +0000 (15:52 -0500)]
[contrib/pzstd] Prevent hangs when there are errors
When two threads are using a WorkQueue and the reader thread exits due
to an error, it must call WorkQueue::finish() to wake up the writer
thread. Otherwise, if the queue is full and the writer thread is waiting
for a free slot, it could hang forever.
This can happen in pratice when decompressing a large, corrupted file
that does not contain pzstd skippable frames.
Mingjie Shen [Mon, 13 Jan 2025 21:39:34 +0000 (16:39 -0500)]
fix out-of-bounds array index access
Yann Collet [Mon, 13 Jan 2025 07:08:26 +0000 (23:08 -0800)]
fixed macro name typo
Yann Collet [Sun, 12 Jan 2025 09:32:40 +0000 (01:32 -0800)]
bench: nb threads specified with -v
MessyHack [Sat, 11 Jan 2025 06:38:56 +0000 (22:38 -0800)]
should check defined(_M_X64) not defined(_M_X86) when building with MSVC.
_M_X86 is only defined under MSVC 32Bit
_M_X64 is only defined under MSVC 64Bit
Victor Zhang [Tue, 7 Jan 2025 18:06:08 +0000 (10:06 -0800)]
Merge pull request #4218 from facebook/externC
Move #includes out of `extern "C"` blocks
Yann Collet [Tue, 7 Jan 2025 17:04:57 +0000 (09:04 -0800)]
Merge pull request #4231 from sunpoet/dev
Update MANDIR for FreeBSD
Victor Zhang [Mon, 6 Jan 2025 17:26:57 +0000 (09:26 -0800)]
Merge pull request #4230 from facebook/variedOutput
Do not vary row matchfinder selection based on availability of SSE2/Neon
Po-Chuan Hsieh [Sun, 5 Jan 2025 14:26:08 +0000 (22:26 +0800)]
Update MANDIR for FreeBSD
share/man became a valid path for manpage since Jan 2020. And we
converted the whole ports tree to share/man around last March.
Victor Zhang [Fri, 3 Jan 2025 17:35:18 +0000 (09:35 -0800)]
Do not vary row matchfinder selection based on availability of SSE2/Neon
Move towards a stronger guarantee of reproducibility by removing this small difference for machines without SSE2/Neon.
The SIMD behavior is now the default for all platforms.
Nick Terrell [Fri, 20 Dec 2024 22:32:28 +0000 (17:32 -0500)]
[opt] Fix too short of match getting generated
The optimal parser with LDM enabled using minMatch > 3 could generate a match
length of 3 when minMatch >= 4. This is not allowed.
1. Fix the bug
2. Add validation logic to `ZSTD_buildSeqStore()` in debug mode for all block
compressors that checks we never generate too short a match. This way we don't
rely on the `generate_sequences` fuzzer to find this issue.
Credit to OSS-Fuzz
Victor Zhang [Fri, 3 Jan 2025 06:42:00 +0000 (22:42 -0800)]
Merge pull request #4229 from facebook/noFseek
Support for libc variants without fseeko/ftello
Victor Zhang [Thu, 2 Jan 2025 23:10:40 +0000 (15:10 -0800)]
chore: indentation alignment
Victor Zhang [Thu, 2 Jan 2025 23:05:58 +0000 (15:05 -0800)]
PR feedback
Victor Zhang [Thu, 2 Jan 2025 22:17:24 +0000 (14:17 -0800)]
Oops
Victor Zhang [Thu, 2 Jan 2025 22:02:10 +0000 (14:02 -0800)]
Support for libc variants without fseeko/ftello
Some older Android libc implementations don't support `fseeko` or `ftello`.
This commit adds a new compile-time macro `LIBC_NO_FSEEKO` as well as a usage in CMake for old Android APIs.
Yann Collet [Thu, 2 Jan 2025 18:57:49 +0000 (10:57 -0800)]
Merge pull request #4228 from facebook/skippableMagicVariant
ZSTD_getFrameHeader() for skippable frames
Yann Collet [Sun, 29 Dec 2024 22:25:33 +0000 (14:25 -0800)]
update type names
naming convention: Type names should start with a Capital letter (after the prefix)
Yann Collet [Sun, 29 Dec 2024 21:27:35 +0000 (13:27 -0800)]
Merge pull request #4227 from facebook/doc_skippableFrame
clarify documentation for Frame-level methods invoked on a skippable frame
Yann Collet [Sun, 29 Dec 2024 20:26:04 +0000 (12:26 -0800)]
improve ZSTD_getFrameHeader on skippable frames
now reports:
- the header size
- the magic variant (within @dictID field)
Yann Collet [Sun, 29 Dec 2024 10:35:53 +0000 (02:35 -0800)]
clarify doc on Frame-level methods when invoked on a skippable frame
following discussion at #4226
Yann Collet [Thu, 26 Dec 2024 19:41:44 +0000 (11:41 -0800)]
Merge pull request #4217 from facebook/ZSTD_compressSequencesAndLiterals
ZSTD_compressSequencesAndLiterals
Yann Collet [Tue, 24 Dec 2024 05:15:50 +0000 (21:15 -0800)]
fixed another invalid scenario
compressSequencesAndLiterals() doesn't support sequence validation
Yann Collet [Tue, 24 Dec 2024 03:43:17 +0000 (19:43 -0800)]
add one valid test case
ZSTD_compressSequencesAndLiterals() may return a specific error code
when data to compress is non-compressible.
Yann Collet [Tue, 24 Dec 2024 02:42:51 +0000 (18:42 -0800)]
added fuzzer test for compressSequencesAndLiterals()
piggy-backing onto existing compressSequences() fuzzer test
Nick Terrell [Fri, 20 Dec 2024 17:40:32 +0000 (09:40 -0800)]
[cmake] Fix -z noexecstack portability
Summary:
Issue reported by @ryandesign and @MarcusCalhoun-Lopez.
CMake doesn't support spaces in flags. This caused older versions of gcc to
ignore the unknown flag "-z noexecstack" on MacOS since it was interpreted as a
single flag, not two separate flags. Then, during compilation it was treated as
"-z" "noexecstack", which was correctly forwarded to the linker. But the MacOS
linker does not support `-z noexecstack` so compilation failed.
The fix is to use `-Wl,-z,noexecstack`. This is never misinterpreted by a
compiler. However, not all compilers support this syntax to forward flags to the
linker. To fix this issue, we check if all the relevant `noexecstack` flags have
been successfully set, and if they haven't we disable assembly.
See also PR#4056 and PR#4061. I decided to go a different route because this is
simpler. It might not successfully set these flags on some compilers, but in that
case it also disables assembly, so they aren't required.
Test Plan:
```
mkdir build-cmake
cmake ../build/cmake/CMakeLists.txt
make -j
```
See that the linker flag is successfully detected & that assembly is enabled.
Run the same commands on MacOS which doesn't support `-Wl,-z,noexecstack` and see
that everything compiles and that `LD_FLAG_WL_Z_NOEXECSTACK` and
`ZSTD_HAS_NOEXECSTACK` are both false.
Yann Collet [Fri, 20 Dec 2024 17:03:26 +0000 (09:03 -0800)]
restore invocation of ZSTD_entropyCompressSeqStore()
in the ZSTD_compressSequences() pipeline
Yann Collet [Fri, 20 Dec 2024 00:11:46 +0000 (16:11 -0800)]
minor: use MEM_writeLE24()
so that an empty frame needs only 3 bytes of dstCapacity.
Yann Collet [Thu, 19 Dec 2024 23:55:11 +0000 (15:55 -0800)]
added parameter litCapacity
to ZSTD_compressSequencesAndLiterals()
to enforce the litCapacity >= litSize+8 condition.
Yann Collet [Thu, 19 Dec 2024 23:39:57 +0000 (15:39 -0800)]
add a check, to return an error if Sequence validation is enabled
since ZSTD_compressSequencesAndLiterals() doesn't support it.
Yann Collet [Thu, 19 Dec 2024 23:00:12 +0000 (15:00 -0800)]
fixed minor error in preparation of one fullbench scenario
Yann Collet [Thu, 19 Dec 2024 22:41:33 +0000 (14:41 -0800)]
ensure that srcSize is controlled
Yann Collet [Thu, 19 Dec 2024 17:45:28 +0000 (09:45 -0800)]
add dedicated error code for special case
ZSTD_compressSequencesAndLiterals() cannot produce an uncompressed block
Yann Collet [Thu, 19 Dec 2024 15:38:14 +0000 (07:38 -0800)]
added a test for ZSTD_compressSequencesAndLiterals
checks that srcSize is present in the frame header
and bounds the window size.
Yann Collet [Thu, 19 Dec 2024 15:26:38 +0000 (07:26 -0800)]
ZSTD_compressSequencesAndLiterals requires srcSize as parameter
this makes it possible to adjust windowSize to its tightest.
Yann Collet [Wed, 18 Dec 2024 06:27:09 +0000 (22:27 -0800)]
fixed minor error in one benchmark scenario
Yann Collet [Wed, 18 Dec 2024 06:08:43 +0000 (22:08 -0800)]
fixed minor conversion warning
Yann Collet [Wed, 18 Dec 2024 05:33:26 +0000 (21:33 -0800)]
improved speed of the Sequences converter
Yann Collet [Wed, 18 Dec 2024 01:25:57 +0000 (17:25 -0800)]
added benchmark for ZSTD_convertBlockSequences_wBlockDelim()
Yann Collet [Tue, 17 Dec 2024 22:24:29 +0000 (14:24 -0800)]
change name to ZSTD_convertSequences*()
Yann Collet [Tue, 17 Dec 2024 05:48:10 +0000 (21:48 -0800)]
attempt to silence Visual Studio warning about fopen()