ddidderr/zstd-rs
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
10,057 Commits 17 Branches 77 Tags
v1.5.4
Commit Graph
2214 Commits
Author SHA1 Message Date
Elliot GorokhovskyandGitHub ff42ed1582 Rename "External Matchfinder" to "Block-Level Sequence Producer" (#3484) 
* change "external matchfinder" to "external sequence producer"

* migrate contrib/ to new naming convention

* fix contrib build

* fix error message

* update debug strings

* fix def of invalid sequences in zstd.h

* nit

* update CHANGELOG

* fix .gitignore
 2023-02-09 17:01:17 -05:00
Elliot Gorokhovsky 3fe5f1fbb9 assert externalRepSearch != ZSTD_ps_auto 2023-02-01 18:24:46 -08:00
Elliot Gorokhovsky 7f8189ca57 add ZSTD_c_fastExternalSequenceParsing cctxParam 2023-02-01 09:09:53 -08:00
Elliot GorokhovskyandGitHub 64052ef57d Guard against invalid sequences from external matchfinders (#3465) 2023-01-31 13:55:48 -05:00
daniellerozenblitandGitHub 00176638e3 Merge pull request #3460 from daniellerozenblit/fix-long-offsets-resolution-pointer 
fix long offset resolution
 2023-01-30 14:02:51 -05:00
daniellerozenblitGitHubNick Terrell
 		2bde9fbf85 Update lib/compress/zstd_compress.c 
Co-authored-by: Nick Terrell <nickrterrell@gmail.com>
 2023-01-27 16:58:53 -05:00
Nick TerrellandNick Terrell 423a74986f [fse] Delete unused functions 
Delete all unused FSE functions, now that we are no longer syncing
to/from upstream.

This avoids confusion about Zstd's stack usage like in Issue #3453.
It also removes dead code, which is always a plus.
 2023-01-27 13:15:07 -08:00
Danielle Rozenblit 9e4c66b9e9 record long offsets in ZSTD_symbolEncodingTypeStats_t + add test case 2023-01-27 12:04:29 -08:00
Danielle Rozenblit 814f4bfb99 fix long offset resolution 2023-01-27 08:21:47 -08:00
daniellerozenblitandGitHub f3255bfeff Merge pull request #3447 from daniellerozenblit/fuzz-sequence-compression 
Fuzz large offsets through sequence compression api
 2023-01-25 09:27:34 -05:00
Yonatan Komornik 1d636b4ba0 Bug fix redzones by unpoisoning only the intended buffer and not the followup redzone. 2023-01-24 12:54:43 -08:00
Danielle Rozenblit 7d600c628a fix bound check for ZSTD_copySequencesToSeqStoreNoBlockDelim() 2023-01-24 06:40:40 -08:00
daniellerozenblitandGitHub 9116000be6 Merge pull request #3439 from daniellerozenblit/sequence-validation-bug-fix 
Fix sequence validation and seqStore bounds check
 2023-01-23 13:50:37 -05:00
Danielle Rozenblit 815d1d4eda update external sequence error to fit error naming scheme 2023-01-23 09:58:34 -08:00
Danielle Rozenblit 1b65727e74 fix nits and add new error code for invalid external sequences 2023-01-23 07:59:02 -08:00
Nick TerrellandNick Terrell b4467c1061 Fix bufferless API with attached dictionary 
Fixes #3102.
 2023-01-20 16:15:16 -08:00
Nick TerrellandNick Terrell 329169189c Replace Huffman boolean args with flags bit set 2023-01-20 14:12:53 -08:00
Nick TerrellandNick Terrell 0cc1b0cb22 Delete unused Huffman functions 
Remove all Huffman functions that aren't used by zstd.
 2023-01-20 14:12:53 -08:00
Yann ColletandGitHub 6742f20a7f Merge pull request #3435 from facebook/c89build 
added c89 build test to CI
 2023-01-20 14:07:12 -08:00
Nick TerrellandNick Terrell 666944fbe6 Cap hashLog & chainLog to ensure that we only use 32 bits of hash 
* Cap shortCache chainLog to 24
* Cap row match finder hashLog so that rowLog <= 24
* Add unit tests to expose all cases. The row match finder unit tests
  are only run in 64-bit mode, because they allocate ~1GB.

Fixes #3336
 2023-01-20 14:05:26 -08:00
Danielle Rozenblit aa385ece13 fix sequence validation and bounds check in ZSTD_copySequencesToSeqStore() 2023-01-20 10:32:35 -08:00
Yann Collet ea684c335a added c89 build test to CI 2023-01-19 14:59:30 -08:00
Elliot GorokhovskyandGitHub bce0382c82 Bugfixes for the External Matchfinder API (#3433) 
* external matchfinder bugfixes + tests

* small doc fix
 2023-01-19 10:41:24 -05:00
daniellerozenblitandGitHub dc1c6cc5df Merge pull request #3418 from daniellerozenblit/fuzz-max-block-size 
Fuzz on maxBlockSize
 2023-01-19 08:18:04 -05:00
Danielle Rozenblit 8353a4b095 fix maxBlockSize resolution + add test cases 2023-01-17 12:24:18 -08:00
Yann Collet ac45e078a5 add explanation about new test 
as requested by @terrelln
 2023-01-12 15:49:01 -08:00
Yann Collet 796699c0bc fix root cause of #3416 
A minor change in 5434de0 changed a `<=` into a `<`,
and as an indirect consequence allowed compression attempt of literals when there are only 6 literals to compress
(previous limit was effectively 7 literals).

This is not in itself a problem, as the threshold is merely an heuristic,
but it emerged a bug that has always been there, and was just never triggered so far due to the previous limit.
This bug would make the literal compressor believes that all literals are the same symbol,
but for the exact case where nbLiterals==6, plus a pretty wild combination of other limit conditions,
this outcome could be false, resulting in data corruption.

Replaced the blind heuristic by an actual test for all limit cases,
so that even if the threshold is changed again in the future,
the detection of RLE mode will remain reliable.
 2023-01-12 15:41:08 -08:00
Danielle Rozenblit 06b096db47 additional tests and documentation updates + allow maxBlockSize to be set to 0 (goes to default) 2023-01-12 13:41:50 -08:00
Danielle Rozenblit 53eb5a758c add simple test for maxBlockSize expected functionality 2023-01-12 08:55:39 -08:00
Danielle Rozenblit 1fffcfe01d update minimum threshold for max block size 2023-01-11 11:09:57 -08:00
Danielle Rozenblit fe08137d9a resolve max block value in cctx and use when calculating the max block size 2023-01-09 07:53:53 -08:00
Yann Collet 71dbe8f9d4 minor: fix conversion warnings 2023-01-04 20:00:04 -08:00
daniellerozenblitandGitHub d913417f72 Merge branch 'dev' into fuzz-max-block-size 2023-01-04 16:34:07 -05:00
Danielle Rozenblit 908e812733 initial commit 2023-01-04 13:01:54 -08:00
Yann Collet ebba9ff425 update regression results 2023-01-03 14:04:23 -08:00
Yann Collet 5434de01e2 improve compression ratio of small alphabets 
fix #3328

In situations where the alphabet size is very small,
the evaluation of literal costs from the Optimal Parser is initially incorrect.
It takes some time to converge, during which compression is less efficient.
This is especially important for small files,
because there will not be enough data to converge,
so most of the parsing is selected based on incorrect metrics.

After this patch, the scenario ##3328 gets fixed,
delivering the expected 29 bytes compressed size (smallest known compressed size).
 2023-01-03 12:22:37 -08:00
daniellerozenblitandGitHub 1c818e3a0a Merge pull request #3302 from daniellerozenblit/optimal-huff-depth-speed 
Optimal huff depth speed improvements
 2023-01-03 12:51:51 -05:00
Danielle Rozenblit df714ddb0f implement suggestions 2023-01-03 07:20:21 -08:00
Yann Collet d07e72bb13 fixed incorrect assert 
commented Fweight instead
 2022-12-28 17:23:40 -08:00
Yann Collet 4a1a79a512 just add some comments to zstd_opt for improved clarity 2022-12-28 16:24:12 -08:00
Yann ColletandGitHub 481a2e1010 Merge pull request #3403 from facebook/setCParams 
ZSTD_CCtx_setCParams
 2022-12-28 14:07:13 -08:00
Elliot GorokhovskyandGitHub 2a402626dd External matchfinder API (#3333) 
* First building commit with sample matchfinder

* Set up ZSTD_externalMatchCtx struct

* move seqBuffer to ZSTD_Sequence*

* support non-contiguous dictionary

* clean up parens

* add clearExternalMatchfinder, handle allocation errors

* Add useExternalMatchfinder cParam

* validate useExternalMatchfinder cParam

* Disable LDM + external matchfinder

* Check for static CCtx

* Validate mState and mStateDestructor

* Improve LDM check to cover both branches

* Error API with optional fallback

* handle RLE properly for external matchfinder

* nit

* Move to a CDict-like model for resource ownership

* Add hidden useExternalMatchfinder bool to CCtx_params_s

* Eliminate malloc, move to cwksp allocation

* Handle CCtx reset properly

* Ensure seqStore has enough space for external sequences

* fix capitalization

* Add DEBUGLOG statements

* Add compressionLevel param to matchfinder API

* fix c99 issues and add a param combination error code

* nits

* Test external matchfinder API

* C90 compat for simpleExternalMatchFinder

* Fix some @nocommits and an ASAN bug

* nit

* nit

* nits

* forward declare copySequencesToSeqStore functions in zstd_compress_internal.h

* nit

* nit

* nits

* Update copyright headers

* Fix CMake zstreamtest build

* Fix copyright headers (again)

* typo

* Add externalMatchfinder demo program to make contrib

* Reduce memory consumption for small blockSize

* ZSTD_postProcessExternalMatchFinderResult nits

* test sum(matchlen) + sum(litlen) == srcSize in debug builds

* refExternalMatchFinder -> registerExternalMatchFinder

* C90 nit

* zstreamtest nits

* contrib nits

* contrib nits

* allow block splitter + external matchfinder, refactor

* add windowSize param

* add contrib/externalMatchfinder/README.md

* docs

* go back to old RLE heuristic because of the first block issue

* fix initializer element is not a constant expression

* ref contrib from zstd.h

* extremely pedantic compiler warning fix, meson fix, typo fix

* Additional docs on API limitations

* minor nits

* Refactor maxNbSeq calculation into a helper function

* Fix copyright
 2022-12-28 16:45:14 -05:00
Yann Collet b17743e41b Signal parameter change during MT compression 2022-12-28 13:14:58 -08:00
Yann Collet 89342d1e07 New xp library symbol : ZSTD_CCtx_setCParams() 
Inspired by #3395,
offer a new capability to set all parameters defined in a ZSTD_compressionParameters structure
with a single symbol invocation
to improve user code brevity.
 2022-12-27 23:49:22 -08:00
Yann ColletandGitHub 089b2797e3 Merge pull request #3398 from facebook/fix3316 
spec update : require minimum nb of literals for 4-streams mode
 2022-12-22 16:57:05 -08:00
Yann Collet 6a9c525903 spec update : require minimum nb of literals for 4-streams mode 
Reported by @shulib :
the specification for 4-streams mode
doesn't work when the amount of literals to compress is 5 bytes.
Extending it, it also doesn't work for sizes 1 or 2.

This patch updates the specification and the implementation
to require a minimum of 6 literals to trigger or accept the 4-streams mode.

The impact is expected to be a no-op :
the 4-streams mode is never triggered for such small quantity of literals anyway,
since it would be wasteful (it costs ~7.3 bytes more than single-stream mode).
An informal lower limit is set at ~256 bytes,
so the technical minimum is very far from this limit.

This is just meant for completeness of the specification.
 2022-12-22 16:14:34 -08:00
Yann Collet ea2895cef4 Support decompression of compressed blocks of size ZSTD_BLOCKSIZE_MAX exactly 2022-12-22 12:40:27 -08:00
Nick TerrellandNick Terrell 40a7188130 Fix make clangbuild & add CI 
Fix the errors for:
* `-Wdocumentation`
* `-Wconversion` except `-Wsign-conversion`
 2022-12-21 17:31:04 -08:00
Danielle Rozenblit c26f348dc8 fix CI errors 2022-12-20 12:43:46 -08:00
Danielle Rozenblit 482689b995 huf log speed optimization: unidirectional scan of logs + break when regressing 2022-12-20 12:27:38 -08:00