Commit Graph
813 Commits
Author SHA1 Message Date
Pavel P 0a183620a3 Reorder __BMI2__ check
+ if `__BMI2__` defined, then set STATIC_BMI2 for all compilers
 + use `defined(_MSC_VER) && defined(__AVX2__)` as fallback for ms compiler
2025-01-24 03:02:47 +02:00
Pavel P d486ccc9e9 Update comment for STATIC_BMI2 macro 2025-01-24 03:02:47 +02:00
Pavel P 1b15e888fc Move STATIC_BMI2 block as-is to portability_macros.h 2025-01-24 03:02:46 +02:00
Yann ColletandGitHub a7b59bcb7f Merge pull request #4257 from pps83/dev-x64test
Use _M_X64 only without mixing with _M_AMD64
2025-01-23 12:50:27 -08:00
Yann ColletandGitHub 55c0c5bdca Merge pull request #4258 from pps83/dev-ZSTD_ALIGNED
Implement ZSTD_ALIGNED for ms compiler
2025-01-22 15:09:35 -08:00
Pavel P a0872a8372 Implement ZSTD_ALIGNED for ms compiler 2025-01-21 02:33:25 +02:00
Pavel P 6c1d1cc600 Use _M_X64 only without mixing with _M_AMD64 2025-01-21 02:27:39 +02:00
Yann ColletandGitHub 48b186f76b Merge pull request #4253 from facebook/BitContainerType
minor: use BitContainerType when appropriate
2025-01-19 18:35:36 -08:00
Yann Collet 82346b92bb minor: generalize BitContainerType
technically equivalent to `size_t`,
but it's the proper type for underlying register representation.

This makes it possible to control register type, and therefore size, independently from `size_t`,
which can be useful on systems where `size_t` is 32-bit, while the architecture supports 64-bit registers.
2025-01-19 18:05:57 -08:00
Yann Collet 4bbf4a285d enable DYNAMIC_BMI2 by default on x86 (32-bit mode)
so far was only enabled for x64 (64-bit mode)
2025-01-19 08:11:59 -08:00
Yann Collet a556559841 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
2025-01-19 00:08:57 -08:00
Yann Collet 27d7940631 minor: cosmetic, indentation 2025-01-18 22:49:16 -08:00
Yann Collet 9efb09749b 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
2025-01-18 22:49:16 -08:00
Yann ColletandGitHub a469e7c083 Merge pull request #4248 from pps83/dev-bzhi32
Use _bzhi_u32 for 32-bit builds when building with STATIC_BMI2
2025-01-18 22:48:24 -08:00
Pavel P fcd684b9b4 update sizeof check 2025-01-19 02:37:35 +02:00
Pavel P d60c4d75e9 remove unrelated changes 2025-01-19 02:36:00 +02:00
Pavel P 462484d5dc change to BitContainerType 2025-01-19 02:34:41 +02:00
Pavel P 26e5fb3614 handle 32bit size_t when building for x64 2025-01-18 23:37:50 +02:00
Pavel P 936927a427 handle 32bit size_t when building for x64 2025-01-18 23:30:55 +02:00
Pavel P ee17f4c6d2 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)
2025-01-18 21:33:04 +02:00
Pavel P 46e17b805b [asm] Enable x86_64 asm for windows builds 2025-01-18 05:33:08 +02:00
Yann ColletandYann Collet 8bff69af86 Alignment instruction ZSTD_ALIGNED() in common/compiler.h 2025-01-15 17:11:27 -08:00
Yann Collet 6f8e6f3c97 create new compilation macro ZSTD_ARCH_X86_AVX2 2025-01-15 17:11:27 -08:00
MessyHack 42d704ad5e 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
2025-01-10 22:47:48 -08:00
Victor ZhangandGitHub a610550e2c Merge pull request #4218 from facebook/externC
Move #includes out of `extern "C"` blocks
2025-01-07 10:06:08 -08:00
Yann Collet a2ff6ea784 improve ZSTD_getFrameHeader on skippable frames
now reports:
- the header size
- the magic variant (within @dictID field)
2024-12-29 12:26:04 -08:00
Yann Collet b339efff2b add dedicated error code for special case
ZSTD_compressSequencesAndLiterals() cannot produce an uncompressed block
2024-12-20 10:37:00 -08:00
Yann ColletandYann Collet 0a5c0807af minor conversion warning fix 2024-12-20 10:36:59 -08:00
Yann Collet 477a01067f codemod: symbolEncodingType_e -> SymbolEncodingType_e 2024-12-20 10:36:56 -08:00
Yann Collet b4a40a845f move Sequences definition to zstd_compress_internal.h
they should not be in common/zstd_internal.h,
since these definitions are not shared beyond lib/compress/.
2024-12-20 10:36:55 -08:00
Victor Zhang 8f49db5a02 Revert "Remove unnecessary extern C declarations from xxhash.h"
This reverts commit 10b9d81909.
2024-12-19 17:54:41 -08:00
Victor Zhang 10b9d81909 Remove unnecessary extern C declarations from xxhash.h 2024-12-19 16:54:32 -08:00
Victor Zhang d0d5ce4c00 Remove extern C blocks from lib/* internal APIs (except xxhash.h) 2024-12-19 16:00:11 -08:00
Victor Zhang d51e6072a8 Test: remove extern C from some lib/common files 2024-12-19 14:59:02 -08:00
Victor Zhang a7bb6d6c49 Oopsie with xxhash.h [1/?] 2024-12-18 12:41:53 -08:00
Victor Zhang 07ffcc6b65 Separate xxhash includes from extern C blocks 2024-12-18 12:35:10 -08:00
Victor Zhang 63acf9a995 Oopsie with huf.h, debug.h 2024-12-18 09:56:50 -08:00
Victor Zhang 58a7f4b869 Oopsie with threading.h 2024-12-17 18:37:33 -08:00
Victor Zhang 5222dd87cf Oopsie with fse.h 2024-12-17 18:11:58 -08:00
Victor Zhang fc726da774 Move #includes out of extern "C" blocks
Do some include shuffling for `**.h` files within lib, programs, tests, and zlibWrapper.
`lib/legacy` and `lib/deprecated` are untouched.
`#include`s within `extern "C"` blocks in .cpp files are untouched.

todo: shuffling for `xxhash.h`
2024-12-17 17:55:07 -08:00
郑苏波 (Super Zheng)andNick Terrell 5e0a83ec25 Disallow 32-bit mode in clang section
Fix register %rbx is only available in 64-bit mode
2024-12-04 06:47:32 -08:00
Yann Collet 7bad787d8b made ZSTD_isPower2() an inline function 2024-10-23 11:50:57 -07:00
Yann ColletandYann Collet 4ce91cbf2b fixed workspace alignment on non 64-bit systems 2024-10-23 11:50:57 -07:00
Adenilson Cavalcanti 6dbd49bcd0 [riscv] Enable support for weak symbols
Both gcc and clang support weak symbols on RISC-V, therefore
let's enable it.

This should fix issue #4069.
2024-08-06 16:55:32 -07:00
Adenilson Cavalcanti c3c28c4d5a [zstd][android] Fix build with NDK r27
The NDK cross compiler declares the target as __linux (which is
not technically incorrect), which triggers the enablement of _GNU_SOURCE
in the newly added code that requires the presence of qsort_r() used
in the COVER dictionary code.

Even though the NDK uses llvm/libc, it doesn't declare qsort_r()
in the stdlib.h header.

The build fix is to only activate the _GNU_SOURCE macro if the OS is
*not* Android, as then we will fallback to the C90 compliant code.

This patch should solve the reported issue number #4103.
2024-07-29 17:13:58 -07:00
Adenilson Cavalcanti 345bcb5ff7 [zstd][dict] Ensure that dictionary training functions are fully reentrant
The two main functions used for dictionary training using the COVER
algorithm require initialization of a COVER_ctx_t where a call
to qsort() is performed.

The issue is that the standard C99 qsort() function doesn't offer
a way to pass an extra parameter for the comparison function callback
(e.g. a pointer to a context) and currently zstd relies on a *global*
static variable to hold a pointer to a context needed to perform
the sort operation.

If a zstd library user invokes either ZDICT_trainFromBuffer_cover or
ZDICT_optimizeTrainFromBuffer_cover from multiple threads, the
global context may be overwritten before/during the call/execution to qsort()
in the initialization of the COVER_ctx_t, thus yielding to crashes
and other bad things (Tm) as reported on issue #4045.

Enters qsort_r(): it was designed to address precisely this situation,
to quote from the documention [1]: "the comparison function does not need to
use global variables to pass through arbitrary arguments, and is therefore
reentrant and safe to use in threads."

It is available with small variations for multiple OSes (GNU, BSD[2],
Windows[3]), and the ISO C11 [4] standard features on annex B-21 qsort_s() as
part of the <stdlib.h>. Let's hope that compilers eventually catch up
with it.

For now, we have to handle the small variations in function parameters
for each platform.

The current fix solves the problem by allowing each executing thread
pass its own COVER_ctx_t instance to qsort_r(), removing the use of
a global pointer and allowing the code to be reentrant.

Unfortunately for *BSD, we cannot leverage qsort_r() given that its API
has changed on newer versions of FreeBSD (14.0) and the other BSD variants
(e.g. NetBSD, OpenBSD) don't implement it.

For such cases we provide a fallback that will work only requiring support
for compilers implementing support for C90.

[1] https://man7.org/linux/man-pages/man3/qsort_r.3.html
[2] https://man.freebsd.org/cgi/man.cgi?query=qsort_r
[3] https://learn.microsoft.com/en-us/cpp/c-runtime-library/reference/qsort-s?view=msvc-170
[4] https://www.open-std.org/jtc1/sc22/wg14/www/docs/n1548.pdf
2024-07-01 23:52:31 -07:00
Yann ColletandGitHub 3de0541aef Merge pull request #4079 from elasota/truncated-huff-state-error
Throw error if Huffman weight initial states are truncated
2024-06-30 16:17:03 -07:00
elasota 0938308ff6 Throw error if Huffman weight initial states are truncated 2024-06-20 17:46:16 -04:00
Dimitri Papadopoulos 44e83e9180 Fix typos not found by codespell 2024-06-20 20:16:25 +02:00
Joseph Chen 5fadd8e6b1 revert FSE_readNCount_body attribute 2024-05-15 10:47:50 +08:00