This was replaced with detect_repeated_char_length(), but the
implementation of the old function was never removed.
Change-Id: I55485cec324dce01033b73f24474f1aca2a31bd3
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Remove declarations from huff_codes.h (preserving doc comments where
applicable) and mark functions that are only called within huff_codes.c
as static.
Change-Id: Idc0113d4eca9e97347def86a502073ef7126114b
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
heapify() is only used within proc_heap_base.c.
Change-Id: I68cc11c2a82fa7f6a989a1838c0e744c0c23feb3
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
compress_icf_map_g() is only used within igzip_icf_body.c.
Change-Id: Id488d6721c60c1909c922a5e0bd162b1542e71ca
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
detect_repeated_char_length() is only used within igzip.c.
Change-Id: I77ee5422e2cb58d81b9705ebcfad68f3a7017d6b
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
valid_lit_huff_table() and valid_dist_huff_table() were declared in an
internal header, but they were never defined or called anywhere.
Change-Id: I59ddf35f161276fca6d6b58081cf640bbea41252
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
This isn't referenced anywhere and isn't part of the public API.
Change-Id: I1e4809c8cc4ac64310fa151425e710abb7351079
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
These globals are only used within huff_codes.c, so they don't need to
be globally visible.
Change-Id: I1e118b3a95cfb7d21bf33c66559362483e460d58
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
This patch makes the project compile and run (tests and
performance tests as well) with CPUs which are not
supporting SSE instructions.
Signed-off-by: Jean-Yves Vet <jyvet@ddn.com>
Currently can't test because yasm doesn't support avx512 but should
get the syntax correct for when it does.
Change-Id: I672b47b83b96861d8b9bfb0af02e726a1949aca0
Signed-off-by: Greg Tucker <greg.b.tucker@intel.com>
Fix a possible 1 byte overflow by creating a combined write_bits and flush.
Change-Id: I2d2455e9e32a820522ff1d89d016db72a82baed9
Signed-off-by: Roy Oursler <roy.j.oursler@intel.com>
This fix prevents possible histogram overflow in compression.
Change-Id: Ie5f25d1bace7f443f432678fcfbd9050ac65113f
Signed-off-by: Roy Oursler <roy.j.oursler@intel.com>
Move current level 2 compression to be level 3 and create a level 2
compression which is the same algorithm as level 1 but with a larger
hashtable.
Change-Id: I398a6017c2ff218dded24e229fb51c23ea69f314
Signed-off-by: Roy Oursler <roy.j.oursler@intel.com>
Moved the afl fuzz test and added llvm fuzz tests including inflate
and round trip compress and inflate. Currently only works with clang,
std makefile and libFuzzer installed. Need to add checking and
support later when libfuzzer is more tightly integrated into the
compiler.
Change-Id: I2db9ad2335d6c5ed846886703b58225f67bcc935
Signed-off-by: Greg Tucker <greg.b.tucker@intel.com>
Modify igzip_rand_test to test that data after next_in can be modified without
producing incorrect compressed data.
Change-Id: Ic6b62f269c6783407fbec7222ae73f24d3735717
Signed-off-by: Roy Oursler <roy.j.oursler@intel.com>
isal_create_hufftables_subset failed to generate length symbols, but should
generate those symbols as a histogram does not guarantee finding all lengths
found in compression
Change-Id: I880210fe1b1078de8617cab0ecb93c9810b9c9de
Signed-off-by: Roy Oursler <roy.j.oursler@intel.com>
For machines with sse4.2 but not avx we can choose a faster version of
crc32_gzip. For highly compressible files using gzip format (rfc 1952) this can
be a significant advantage.
Change-Id: Ib699a68999290eed4b83f9bbf54dc6c304a5c445
Signed-off-by: Greg Tucker <greg.b.tucker@intel.com>
Remove igzip_sync_flush perf because the perf test does not give any new
information, and the data used does is not a representative work load.
Change-Id: I7b68f8b7c6da0944ace5a2a9e31db378135689ff
Signed-off-by: Roy Oursler <roy.j.oursler@intel.com>
Modify total_out to be total data decompressed into user supplied buffers,
rather than to user supplied and internal buffers.
Signed-off-by: Roy Oursler <roy.j.oursler@intel.com>
