Yann Collet [Wed, 17 Apr 2019 22:33:37 +0000 (15:33 -0700)]
re-enable FORCE_INLINE
was disabled for tests
Yann Collet [Wed, 17 Apr 2019 22:01:53 +0000 (15:01 -0700)]
fix out-of-bound read within LZ4_decompress_fast()
and deprecate LZ4_decompress_fast(),
with deprecation warnings enabled by default.
Note that, as a consequence of the fix,
LZ4_decompress_fast is now __slower__ than LZ4_decompress_safe().
That's because, since it doesn't know the input buffer size,
it must progress more cautiously into the input buffer
to ensure to out-of-bound read.
Yann Collet [Wed, 17 Apr 2019 20:31:24 +0000 (13:31 -0700)]
fuzzer: added tests to detect LZ4_decompress_fast() out of bound read
Yann Collet [Wed, 17 Apr 2019 17:43:38 +0000 (10:43 -0700)]
Merge pull request #674 from nrgiii/dev
Fix
Norm Green [Wed, 17 Apr 2019 16:35:18 +0000 (09:35 -0700)]
Merge branch 'fuzzerAix' into dev
Norm Green [Wed, 17 Apr 2019 16:20:09 +0000 (09:20 -0700)]
Fix AIX errors/warnings
Yann Collet [Wed, 17 Apr 2019 03:46:04 +0000 (20:46 -0700)]
fixed version number for lz4hc
Yann Collet [Tue, 16 Apr 2019 18:52:54 +0000 (11:52 -0700)]
minor benchmark update
was using wrong memory timing
Yann Collet [Tue, 16 Apr 2019 18:26:03 +0000 (11:26 -0700)]
ensure consistent definition and usage of FREEMEM
as suggested by @sloutsky in #671
Yann Collet [Tue, 16 Apr 2019 18:20:31 +0000 (11:20 -0700)]
minor news update
for msys2
Yann Collet [Tue, 16 Apr 2019 00:30:31 +0000 (17:30 -0700)]
updated NEWS for v1.9.0
Yann Collet [Mon, 15 Apr 2019 22:32:03 +0000 (15:32 -0700)]
Merge pull request #669 from lz4/fixframe
Fix lz4frame inefficiency
Yann Collet [Mon, 15 Apr 2019 21:13:10 +0000 (14:13 -0700)]
simplified output_directive
Yann Collet [Mon, 15 Apr 2019 18:22:25 +0000 (11:22 -0700)]
fix comma for pedantic
Yann Collet [Mon, 15 Apr 2019 18:18:33 +0000 (11:18 -0700)]
Merge branch 'dev' into fixframe
Yann Collet [Mon, 15 Apr 2019 18:17:35 +0000 (11:17 -0700)]
added a fake initialization to please an analyzer
which is unable to understand that the variable is necessarily initialized
in spite of an assert just before.
Yann Collet [Mon, 15 Apr 2019 18:12:54 +0000 (11:12 -0700)]
fix fuzzer tests on dirty context
context is no longer dirty after a failed compressed block.
Actually, all indexes are fine.
It remains compatible with continued streaming, and reset*_fast().
Yann Collet [Mon, 15 Apr 2019 18:09:56 +0000 (11:09 -0700)]
unified limitedOutput_directive
between lz4.c and lz4hc.c .
was left in a strange state after the "amalgamation" patch.
Now only 3 directives remain,
same name across both implementations,
single definition place.
Might allow some light simplification due to reduced nb of states possible.
Yann Collet [Mon, 15 Apr 2019 17:46:48 +0000 (10:46 -0700)]
Merge branch 'dev' into fixframe
Yann Collet [Mon, 15 Apr 2019 17:44:46 +0000 (10:44 -0700)]
added more margin for canary in frametest
Yann Collet [Mon, 15 Apr 2019 17:33:40 +0000 (10:33 -0700)]
decompress*_fast() function do not generate deprecation warnings
they are classified as deprecated in the API documentation (lz4.h)
but do not yet trigger a warning,
to give time to existing applications to move away.
Also, the _fast() variants are still ~5% faster than the _safe() ones
after Dave's patch.
Yann Collet [Mon, 15 Apr 2019 17:28:36 +0000 (10:28 -0700)]
fixed lz4frame with linked blocks
when one block was not compressible,
it would tag the context as `dirty`,
resulting in compression automatically bailing out of all future blocks,
making the rest of the frame uncompressible.
Yann Collet [Mon, 15 Apr 2019 17:14:36 +0000 (10:14 -0700)]
frametest: removed uninitialized warning
was a false positive, but better remove it anyway
Yann Collet [Mon, 15 Apr 2019 09:11:59 +0000 (02:11 -0700)]
assert() the failed test
ensure canary remains within buffer limits
Yann Collet [Mon, 15 Apr 2019 08:54:40 +0000 (01:54 -0700)]
fixed frametest
One test could write a canary value out of bound
in exceptional conditions involving multiple flushes
triggered by -s3421 -t462948.
Yann Collet [Sat, 13 Apr 2019 22:55:33 +0000 (15:55 -0700)]
Merge branch 'dev' of github.com:Cyan4973/lz4 into dev
Yann Collet [Sat, 13 Apr 2019 22:55:06 +0000 (15:55 -0700)]
fixed incorrect assertion condition
output can use the full length of output buffer
Yann Collet [Sat, 13 Apr 2019 21:15:48 +0000 (14:15 -0700)]
Merge pull request #668 from lz4/visual
Fix some Visual conversion warnings
Yann Collet [Sat, 13 Apr 2019 01:00:43 +0000 (18:00 -0700)]
updated benchmark table
Yann Collet [Sat, 13 Apr 2019 00:40:23 +0000 (17:40 -0700)]
and even more visual conversion warnings
Yann Collet [Sat, 13 Apr 2019 00:29:44 +0000 (17:29 -0700)]
Merge pull request #667 from lz4/pipeMultiple
Fix combination -m -c
Yann Collet [Sat, 13 Apr 2019 00:06:53 +0000 (17:06 -0700)]
some more Visual conversion warning
Yann Collet [Sat, 13 Apr 2019 00:03:28 +0000 (17:03 -0700)]
fixed minor Visual conversion warnings
Yann Collet [Fri, 12 Apr 2019 23:56:22 +0000 (16:56 -0700)]
fix minor visual warning
yet some overly cautious overflow risk flag,
while it's actually impossible, due to previous test just one line above.
Changing the cast position, just to be please the thing.
Yann Collet [Fri, 12 Apr 2019 23:49:01 +0000 (16:49 -0700)]
fixed minor Visual warnings
since Visual 2017,
worries about potential overflow, which are actually impossible.
Replaced (c * a) by (c ? a : 0).
Will likely replaced a * by a cmov.
Probably harmless for performance.
Yann Collet [Fri, 12 Apr 2019 22:55:38 +0000 (15:55 -0700)]
added tests for -d -m -c
seems to work properly
Yann Collet [Fri, 12 Apr 2019 22:17:48 +0000 (15:17 -0700)]
fixed lz4 -m -c
can compress multiple files into stdout
Yann Collet [Fri, 12 Apr 2019 21:28:44 +0000 (14:28 -0700)]
Merge pull request #666 from lz4/blockIndep
added command -BI for "Independent Blocks"
Yann Collet [Fri, 12 Apr 2019 20:27:11 +0000 (13:27 -0700)]
added command -BI for "Independent Blocks"
This is the reverse of `-BD`, and the current default.
This command can be useful to reverse a previous `-BD` command.
It may in the future be more important
if `lz4` switches to generating dependent blocks by default.
Yann Collet [Fri, 12 Apr 2019 20:15:05 +0000 (13:15 -0700)]
updated man page
with --favor-decSpeed
Yann Collet [Fri, 12 Apr 2019 18:27:44 +0000 (11:27 -0700)]
updated doc to underline difference between block and frame
as this is a very frequent source of confusion for new users.
Yann Collet [Thu, 11 Apr 2019 22:43:40 +0000 (15:43 -0700)]
Merge pull request #665 from lz4/dict
Dictionary compression
Yann Collet [Thu, 11 Apr 2019 22:42:54 +0000 (15:42 -0700)]
Merge pull request #664 from lz4/maxdist
introduce LZ4_DISTANCE_MAX build macro
Yann Collet [Thu, 11 Apr 2019 22:19:48 +0000 (15:19 -0700)]
improved documentation for LZ4 dictionary compression
Yann Collet [Thu, 11 Apr 2019 21:15:33 +0000 (14:15 -0700)]
introduce LZ4_DISTANCE_MAX build macro
make it possible to generate LZ4-compressed block
with a controlled maximum offset (necessarily <= 65535).
This could be useful for compatibility with decoders
using a very limited memory budget (<64 KB).
Answer #154
Yann Collet [Thu, 11 Apr 2019 20:55:35 +0000 (13:55 -0700)]
updated block format with more details on MF_LIMIT
aka distance from end of block for the last match.
Yann Collet [Wed, 10 Apr 2019 04:30:14 +0000 (21:30 -0700)]
Merge pull request #663 from lz4/headerSize
made LZ4F_getHeaderSize() public
Yann Collet [Wed, 10 Apr 2019 01:23:32 +0000 (18:23 -0700)]
added versions in comments
Yann Collet [Wed, 10 Apr 2019 01:10:02 +0000 (18:10 -0700)]
made LZ4F_getHeaderSize() public
Yann Collet [Tue, 9 Apr 2019 23:26:49 +0000 (16:26 -0700)]
Merge pull request #659 from lz4/resetFast
LZ4_resetStream*_fast()
Yann Collet [Tue, 9 Apr 2019 22:37:59 +0000 (15:37 -0700)]
fixed loadDictHC
by making a full initialization
instead of a fast reset.
Yann Collet [Tue, 9 Apr 2019 21:00:30 +0000 (14:00 -0700)]
re-enable LZ4_resetStreamHC()
towards deprecation, but still available and fully supported
Yann Collet [Tue, 9 Apr 2019 20:55:42 +0000 (13:55 -0700)]
modified LZ4_initStreamHC() to look like LZ4_initStream()
it is now a pure initializer, for statically allocated states.
It can initialize any memory area, and because of this, requires size.
Yann Collet [Mon, 8 Apr 2019 23:51:22 +0000 (16:51 -0700)]
check some more initialization result
ensure it's not NULL.
Yann Collet [Mon, 8 Apr 2019 23:47:21 +0000 (16:47 -0700)]
removed LZ4_stream_t alignment test on Visual
it fails on x86 32-bit mode :
Visual reports an alignment of 8-bytes (even with alignof())
but actually only align LZ4_stream_t on 4 bytes.
The alignment check then fails, resulting in missed initialization.
Yann Collet [Mon, 8 Apr 2019 21:21:09 +0000 (14:21 -0700)]
checkFrame: removed %zu
not liked by mingw
Yann Collet [Mon, 8 Apr 2019 21:10:59 +0000 (14:10 -0700)]
fullbench: assert lz4_stream_t initialization
Yann Collet [Mon, 8 Apr 2019 21:01:32 +0000 (14:01 -0700)]
fixed cleaning tmp directory
Yann Collet [Mon, 8 Apr 2019 20:49:26 +0000 (13:49 -0700)]
slightly faster huge file test
level down 5->4
size down 6G->5G
Yann Collet [Mon, 8 Apr 2019 20:45:01 +0000 (13:45 -0700)]
added make list target to tests
Yann Collet [Mon, 8 Apr 2019 19:49:54 +0000 (12:49 -0700)]
LZ4_initStream() checks alignment restriction
updated associated documentation
Yann Collet [Fri, 5 Apr 2019 22:35:19 +0000 (15:35 -0700)]
added comment on initStream + _extState_
as suggested by @felixhandte
Yann Collet [Fri, 5 Apr 2019 19:54:13 +0000 (12:54 -0700)]
created LZ4_initStream()
- promoted LZ4_resetStream_fast() to stable
- moved LZ4_resetStream() into deprecate, but without triggering a compiler warning
- update all sources to no longer rely on LZ4_resetStream()
note : LZ4_initStream() proposal is slightly different :
it's able to initialize any buffer, provided that it's large enough.
To this end, it accepts a void*, and returns an LZ4_stream_t*.
Yann Collet [Fri, 5 Apr 2019 18:55:34 +0000 (11:55 -0700)]
fixed _GNU_SOURCE duplicate
Yann Collet [Fri, 5 Apr 2019 18:47:06 +0000 (11:47 -0700)]
fuzzer: fixed strict c99 with mmap+MAP_ANONYMOUS
Yann Collet [Fri, 5 Apr 2019 17:41:16 +0000 (10:41 -0700)]
fixed strict iso C90
Yann Collet [Fri, 5 Apr 2019 00:03:05 +0000 (17:03 -0700)]
created LZ4_initStreamHC()
- promoted LZ4_resetStreamHC_fast() to stable
- moved LZ4_resetStreamHC() to deprecated (but do not generate a warning yet)
- Updated doc, to highlight difference between init and reset
- switched all invocations of LZ4_resetStreamHC() onto LZ4_initStreamHC()
- misc: ensure `make all` also builds /tests
Yann Collet [Thu, 4 Apr 2019 20:42:58 +0000 (13:42 -0700)]
Merge pull request #658 from lz4/_fast
Deprecated LZ4_decompres_fast*() functions
Yann Collet [Thu, 4 Apr 2019 19:47:36 +0000 (12:47 -0700)]
make `_fast*()` decoder generate a deprecation warning
updated modification
Yann Collet [Thu, 4 Apr 2019 19:24:46 +0000 (12:24 -0700)]
moved LZ4_decompress_fast*() into deprecated section
Yann Collet [Thu, 4 Apr 2019 00:15:06 +0000 (17:15 -0700)]
Merge pull request #657 from lz4/destSize
moved _destSize() into "stable API" status
Yann Collet [Wed, 3 Apr 2019 23:28:42 +0000 (16:28 -0700)]
fixed an old bug in LZ4F_flush()
which remained undetected so far,
as it requires a fairly large number of conditions to be triggered,
starting with enabling Block checksum, which is disabled by default,
and which usage is known to be extremely rare.
Yann Collet [Wed, 3 Apr 2019 21:27:21 +0000 (14:27 -0700)]
fixed doc
and bumped version number fo v1.9.0
Yann Collet [Wed, 3 Apr 2019 21:18:00 +0000 (14:18 -0700)]
moved _destSize() into "stable API" status
as requested in #642
Yann Collet [Wed, 3 Apr 2019 17:03:26 +0000 (10:03 -0700)]
Merge pull request #656 from lz4/armtest
LZ4_FAST_DEC_LOOP macros
Yann Collet [Wed, 3 Apr 2019 15:59:29 +0000 (08:59 -0700)]
minor comments and reformatting
Yann Collet [Wed, 3 Apr 2019 00:16:43 +0000 (17:16 -0700)]
fixed minor conversion warnings
Yann Collet [Tue, 2 Apr 2019 23:22:11 +0000 (16:22 -0700)]
created LZ4_FAST_DEC_LOOP build macro
Yann Collet [Tue, 2 Apr 2019 23:06:37 +0000 (16:06 -0700)]
fixed a few minor conversion warnings
Yann Collet [Sun, 3 Mar 2019 21:40:45 +0000 (13:40 -0800)]
Merge pull request #652 from vtorri/dev
Allow installation of lz4 for Windows 10 with MSYS2
Vincent Torri [Sun, 3 Mar 2019 21:06:38 +0000 (22:06 +0100)]
Allow installation of lz4 for Windows 10 with MSYS2
Yann Collet [Tue, 12 Feb 2019 00:58:53 +0000 (16:58 -0800)]
Merge pull request #645 from djwatson/optimize_decompress_generic
Optimize decompress generic
Dave Watson [Fri, 8 Feb 2019 21:57:43 +0000 (13:57 -0800)]
decompress_generic: Limit fastpath to x86
New fastpath currently shows a regression on qualcomm
arm chips. Restrict it to x86 for now
Dave Watson [Mon, 28 Jan 2019 17:36:25 +0000 (09:36 -0800)]
decompress_generic: Add fastpath for small offsets
For small offsets of size 1, 2, 4 and 8, we can set a single uint64_t,
and then use it to do a memset() variation. In particular, this makes
the somewhat-common RLE (offset 1) about 2-4x faster than the previous
implementation - we avoid not only the load blocked by store, but also
avoid the loads entirely.
Dave Watson [Sat, 26 Jan 2019 00:19:05 +0000 (16:19 -0800)]
decompress_generic: Unroll loops a bit more
Generally we want our wildcopy loops to look like the
memcpy loops from our libc, but without the final byte copy checks.
We can unroll a bit to make long copies even faster.
The only catch is that this affects the value of FASTLOOP_SAFE_DISTANCE.
Dave Watson [Sat, 26 Jan 2019 00:06:55 +0000 (16:06 -0800)]
decompress_generic: remove msan write
This store is also causing load-blocked-by-store issues, remove it.
The msan warning will have to be fixed another way if it is still an issue.
Dave Watson [Sat, 26 Jan 2019 00:01:09 +0000 (16:01 -0800)]
decompress_generic: re-add fastpath
This is the remaineder of the original 'shortcut'. If true, we can avoid
the loop in LZ4_wildCopy, and directly copy instead.
Dave Watson [Fri, 25 Jan 2019 22:47:15 +0000 (14:47 -0800)]
decompress_generic: drop partial copy check in fast loop
We've already checked that we are more than FASTLOOP_SAFE_DISTANCE
away from the end, so this branch can never be true, we will have
already jumped to the second decode loop.
Dave Watson [Fri, 25 Jan 2019 22:40:27 +0000 (14:40 -0800)]
decompress_generic: Optimize literal copies
Use LZ4_wildCopy16 for variable-length literals. For literal counts that
fit in the flag byte, copy directly. We can also omit oend checks for
roughly the same reason as the previous shortcut: We check once that both
match length and literal length fit in FASTLOOP_SAFE_DISTANCE, including
wildcopy distance.
Dave Watson [Thu, 24 Jan 2019 22:17:24 +0000 (14:17 -0800)]
decompress_generic: optimize match copy
Add an LZ4_wildCopy16, that will wildcopy, potentially smashing up
to 16 bytes, and use it for match copy. On x64, this avoids many
blocked loads due to store forwarding, similar to issue #411.
Dave Watson [Thu, 24 Jan 2019 21:42:25 +0000 (13:42 -0800)]
decompress_generic: Add a loop fastpath
Copy the main loop, and change checks such that op is always less
than oend-SAFE_DISTANCE. Currently these are added for the literal
copy length check, and for the match copy length check.
Otherwise the first loop is exactly the same as the second. Follow on
diffs will optimize the first copy loop based on this new requirement.
I also tried instead making a separate inlineable function for the copy
loop (similar to existing partialDecode flags, etc), but I think the
changes might be significant enough to warrent doubling the code, instead
pulling out common functionality to separate functions.
This is the basic transformation that will allow several following optimisations.
Dave Watson [Thu, 24 Jan 2019 21:07:57 +0000 (13:07 -0800)]
decompress_generic: Refactor variable length fields
Make a helper function to read variable lengths for literals and
match length.
Yann Collet [Thu, 7 Feb 2019 18:38:09 +0000 (10:38 -0800)]
Merge pull request #648 from aregm/fix-VS2017-solution
Build fixed by removing unavailable project
Areg Melik-Adamyan [Thu, 7 Feb 2019 04:29:31 +0000 (22:29 -0600)]
Build fixed by removing unavailable project
Yann Collet [Mon, 4 Feb 2019 22:50:28 +0000 (14:50 -0800)]
Merge pull request #646 from jbms/fix-clang-optimize-attribute-ppc64le
Eliminate optimize attribute warning with clang on PPC64LE
Jeremy Maitin-Shepard [Mon, 4 Feb 2019 20:22:56 +0000 (12:22 -0800)]
Eliminate optimize attribute warning with clang on PPC64LE
Yann Collet [Wed, 23 Jan 2019 17:03:13 +0000 (09:03 -0800)]
Merge pull request #644 from lzutao/meson-msvc-export
meson: Add -DLZ4_DLL_EXPORT=1 to build dynamic lib on Windows
Lzu Tao [Wed, 23 Jan 2019 08:40:26 +0000 (15:40 +0700)]
meson: Add -DLZ4_DLL_EXPORT=1 to build dynamic lib on Windows
Thanks @nacho for pointing it out.
Yann Collet [Fri, 11 Jan 2019 20:11:45 +0000 (12:11 -0800)]
Merge pull request #638 from lzutao/travis
Travis: Clean up .travis.yml
Yann Collet [Fri, 11 Jan 2019 20:11:35 +0000 (12:11 -0800)]
Merge pull request #639 from lzutao/meson
meson: Small improvements