Commit Graph

4628 Commits

Author SHA1 Message Date
Yann Collet
96bb29aa14 Merge pull request #878 from terrelln/adjust
Don't `size -= 1` in ZSTD_adjustCParams()
2017-09-30 14:52:54 -07:00
Nick Terrell
c5d6dde502 Don't size -= 1 in ZSTD_adjustCParams()
The window size could end up too small if the source size is 2^n + 1.

Credit to OSS-Fuzz
2017-09-30 14:20:06 -07:00
Yann Collet
ee1ed78fcb fix proper naming on FSE_createCTable() arguments in fse.h 2017-09-30 11:08:50 -07:00
Yann Collet
5b10345b26 added ZSTD_COMPRESSBOUND() as a macro
ZSTD_compressBound() works fine, but is only useful for dynamic allocation.
For static allocation, only a macro can provide the amount during compilation time.
2017-09-29 23:17:41 -07:00
Yann Collet
e580dc6a4a Merge pull request #860 from felixhandte/zstd-lz4-support-tests
Add Default LZ4 Support When Available
2017-09-29 22:32:54 -07:00
Yann Collet
8afb151c9b cli: fixed wrong initialization in MT mode
It's not good to mix old and new API
ZSTD_resetCStream() doesn't just set pledgedSrcSize :
it also sets the CCtx for a single thread compression.

Problem is, when 2+ threads are defined in cctx->requestedParams,
ZSTD_compress_generic() will want to start MT compression,
since initialization is supposed to have already happened (thanks to ZSTD_resetCStream())
except that the underlying ZSTDMT_CCtx* object is not created,
resulting in a segfault.

This is an invalid construction
(correct one is to use ZSTD_CCtx_setPledgedSrcSize()).
I haven't found a nice way to mitigate this impact if someone makes the same mistake.

At some point, removing the old API to keep only the new API within fileio.c will limit these risks.
2017-09-29 22:14:37 -07:00
Yann Collet
657946a913 Merge pull request #875 from facebook/zstdcat
decode more data before triggering error
2017-09-29 20:12:25 -07:00
Yann Collet
fbd5ab7027 minor fix : no longer use fake srcSize during resource creation
srcSize is read and provided at each file, not at resource creation.
This used to be useful with older API, because it could not re-adapt parameters between sessions.

At some point, it will be better to remove the old code, and only keep the new_api.
It works fine by now.
2017-09-29 19:40:27 -07:00
Yann Collet
db1668a43b fix : srcSize written in frame header when multiple files compressed
This information used to be disabled when nbFiles>1.
It was badly initialized later in the code, resulting in an error.
2017-09-29 18:05:18 -07:00
Yann Collet
7c9669f272 Merge pull request #873 from facebook/shorterTests
Leaner tests
2017-09-29 17:26:46 -07:00
Yann Collet
1416bc0f07 erase existence of a buffer when it's sent out of the pool
In some complex scenario,
the buffer would be freed because it's too large,
another buffer would be allocated, but fail,
trigger an error,
and the general buffer pool would then be freed,
where the definition of the already freed buffer would be found
(beyond total index, but still), and freed again, resulting in double-free error.
2017-09-29 16:27:47 -07:00
Yann Collet
8afcc80e07 decode more data before triggering error
fixes #874 :
when a frame is not properly terminated by a "last block" signal,
zstd -d used to detect it immediately and error out.
This version will decode and flush the last block, and only then issue an error.
2017-09-29 15:54:09 -07:00
Yann Collet
e963800e27 zstdmt : fixed : buffer dst0 wasn't properly set to null after usage
now it's possible to unconditionnally invoke ZSTD_releaseAllJobRessources()
wether previous compression was completed correctly or not.
2017-09-28 23:01:31 -07:00
Yann Collet
754ae5cc0b removed ZSTDMT_waitForAllJobsCompleted() from ZSTDMT_freeCCtx()
as per @terrelln comment
2017-09-28 20:45:31 -07:00
Yann Collet
e0065cf660 make test : removed zstreamtest unit tests for variants
slightly reduced time to create dictionary
at beginning of unit tests
2017-09-28 18:34:38 -07:00
Yann Collet
47c6a95d07 zstreamtest : run unit tests only during "normal" session
not during --mt, --newapi and --opaque
this avoids running them 4x during `make test`
2017-09-28 18:27:22 -07:00
Yann Collet
86b4fe5b45 adjustCParams : restored previous behavior
unknowns srcSize presumed small if there is a dictionary (dictSize>0)
and presumed large otherwise.
2017-09-28 18:14:28 -07:00
W. Felix Handte
dc27c36495 Update documentation to reflect other format support 2017-09-28 19:43:12 -04:00
W. Felix Handte
d0519d4b0c Add CLI Program Name Detection for LZ4 2017-09-28 19:18:15 -04:00
W. Felix Handte
5705d9f25a Add basic tests for the lz4 integration 2017-09-28 19:16:43 -04:00
Yann Collet
b93598d6a4 zstdmt : reduced maximum nb of threads
to avoid memory address space issues on 32-bits systems
(see https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=876416#17)
2017-09-28 13:49:12 -07:00
Yann Collet
e4ec427720 Merge branch 'dev' into shorterTests
fixed conflicts
2017-09-28 12:19:28 -07:00
Yann Collet
bbef058ae6 zstreamtest --newapi : reduced maximum allocated memory 2017-09-28 11:48:45 -07:00
Yann Collet
8074261d00 zstdmt : move on when not enough memory for a new input buffer
just continue operations without input forward progress,
instead of an error that stops current compression session.
2017-09-28 11:46:19 -07:00
Yann Collet
2cd15dd9a4 fixed minor Visual conversion warning 2017-09-28 02:33:41 -07:00
Yann Collet
377abcc02c zstdmt : better behavior when freeing a context right after a memory allocation error
wait for all jobs to be completed, so that freeing can happen safely
2017-09-28 02:23:44 -07:00
Yann Collet
d6770f80af minor : rewrite unit tests using CHECK_Z macro 2017-09-28 02:14:48 -07:00
Yann Collet
9fe50ed623 fixed maximum windowLog for zstreamtest --newapi
for compatibility with low memory VM
2017-09-28 01:42:06 -07:00
Yann Collet
9b5b47ac93 ensure adjustCParams adjust hLog and cLog even without srcSize
It would previously exit when srcSize is unknown.
But in the case of custom parameters,
hLog and cLog can still be too large in comparison with windowLog.

Reduces maximum memory allocated during zstreamtest --newapi
2017-09-28 01:25:40 -07:00
Yann Collet
824f75ea7c Merge pull request #863 from facebook/newFormats
magicless frames (#591)
2017-09-28 00:32:16 -07:00
Yann Collet
fac8c70859 Merge pull request #872 from terrelln/fuzz-minor
[fuzz] Small changes for oss-fuzz integration
2017-09-27 19:12:18 -07:00
Nick Terrell
d9c1e9125f [fuzz] Small changes for oss-fuzz integration 2017-09-27 18:23:06 -07:00
Yann Collet
aa800c4793 reduced memory usage of zstreamtest --newapi
to run on memory-constrained VM
2017-09-27 18:00:15 -07:00
Yann Collet
bc32b40b98 reduced zstreamtest --mt memory load
adjust compression level, hence memory usage, depending on nb threads
in order to run correctly on memory-starved VM.
2017-09-27 17:27:38 -07:00
Yann Collet
54a827fff0 Merge branch 'dev' into newFormats
Fixed conflicts in zstdmt_compress.c
2017-09-27 16:39:40 -07:00
Yann Collet
e45a2aea9b Merge pull request #869 from terrelln/dev
[libzstd] pthread function prefixed with ZSTD_
2017-09-27 16:35:08 -07:00
Yann Collet
3182ea2e64 Merge pull request #866 from facebook/list
improved --list display
2017-09-27 16:34:29 -07:00
Yann Collet
e57b4114e9 Merge pull request #871 from terrelln/opt
[libzstd][opt] Simplify repcode logic
2017-09-27 16:34:15 -07:00
Yann Collet
02502191e5 separated --long tests between short and long tests
A fast subset of these tests is now played in short test mode
2017-09-27 15:48:06 -07:00
Yann Collet
f9de54acfb reduced memory requirements for --long tests
in new --long test section
2017-09-27 15:38:27 -07:00
Nick Terrell
b555b7ef41 [libzstd][opt] Simplify repcode logic 2017-09-27 15:30:12 -07:00
Yann Collet
bdd0f6f046 improved make clean in tests/fuzz 2017-09-27 15:20:08 -07:00
Yann Collet
60059df051 shorter make test to avoid time out on travis CI
Timed tests (fuzzer) are reduced
Long tests are shortened (less data generated)
2017-09-27 15:16:27 -07:00
Yann Collet
495b5269ea Merge pull request #870 from facebook/cpptest
Change c++ test to use CXX and CXXFLAGS environment variables
2017-09-27 13:58:12 -07:00
Yann Collet
ea1f50bf73 removed ZSTD_decompressBegin() from ZSTD_initDCtx_internal()
It does not feel "right" from a dependency perspective.
ZSTD_initDCtx_internal() is triggered once, on DCtx creation,
while ZSTD_decompressBegin() is invoked at the beginning of each new frame,
and is also a user-facing prototype.

Downside : a DCtx must be init before first usage !
This was always the intention by the way, and is documented as such.
This stage is automatically done within ZSTD_decompress() and variants,
and also within ZSTD_decompressStream().
Only ZSTD_decompressContinue() is impacted,
it must be preceded by a ZSTD_decompressBegin(), as detailed in doc.

A test has been fixed, to no longer rely on undocumented assumption that ZSTD_decompressBegin() is invoked during init.
2017-09-27 13:51:05 -07:00
Yann Collet
60ca44b545 switched name to cxxtest 2017-09-27 12:24:13 -07:00
Yann Collet
c994932788 fixed ZSTD_format_e value validation 2017-09-27 12:22:22 -07:00
Yann Collet
763f8b5e45 Change c++ test to use CXX and CXXFLAGS environment variables
Fix OS-X warning on compiling .c files with clang++
Also changed test name from gpptest to cpptest, since it's no longer g++ specific
2017-09-27 12:09:52 -07:00
Nick Terrell
6c41adfb28 [libzstd] pthread function prefixed with ZSTD_
* `sed -i 's/pthread_/ZSTD_pthread_/g' lib/{,common,compress,decompress,dictBuilder}/*.[hc]`
* Fix up `lib/common/threading.[hc]`
* `sed -i s/PTHREAD_MUTEX_LOCK/ZSTD_PTHREAD_MUTEX_LOCK/g lib/compress/zstdmt_compress.c`
2017-09-27 11:48:48 -07:00
Yann Collet
ecf1778e23 updated ZSTD_format_e value validation
also updated manual
2017-09-27 11:19:21 -07:00