Currently, dictionaries are taken into account only when `libfuzzer` is
used as a fuzzing engine (and also apparently `none` but I'm not sure what it is).
This patch makes it possible to make use of dictionaries with other fuzzing
engines too. I didn't touch the code handling options passed to libFuzzer
so as not to break anything :-)
* envoy: use Bazel to build corpora.
Rather than scraping this out of the source tree, build each corpus
under Bazel. The advantage is that we can now support synthesized
corpora, e.g. automated generation from unit tests.
Signed-off-by: Harvey Tuch <htuch@google.com>
* Review feedback.
Signed-off-by: Harvey Tuch <htuch@google.com>
* bzip2: Bug fixes; added citations
* bzip2: Bump bzip2 version to 1.0.6
* bzip2: Variable blockSize100k, workFactor, and small
* bzip2: Remove hardcoded buffer sizes; fix nZ (compressor) to point to real buffer size
* bzip: Remove ubsan from project.yaml beacuse of start-up crash google/oss-fuzz#1887
* lzo: Add lzo out-of-source
* lzo: Change maintainer of Docker image
* lzo: Add decompress target
* lzo: Update build script
* lzo: Fix build script
* lzo: Bail out if size==0 in decompress test
* Adding a seed file to lzo_decompress_target_seed
This commit adds a minimal lzo seed as a seed for the lzo_decompress_target. Still results in a heap-buffer-overflow at the moment.
* lzo: Switch to safer decompressor (HT @viniul) that does not crash and remove assertion
* lzo: Remove buggy addition (from OvS) to this PR
* Add more decompression targets to decompress_target
This commit adds more decompression targets to decompress_target.c. The target function is chosen based on the first byte of the data given by libfuzzer.
* Make decomp func ptr static, fix minor bug in func ptr init, and fix include paths in build script
Since Wireshark v2.9.0rc0-2266-g7e88bb5e53, building fuzzing binaries is
better integrated in the Wireshark build system. While that change was
supposed to be backwards compatible, unfortunately the resulting fuzzer
binaries are dynamically linked due to how cmake caches library paths.
The very first cmake invocation must be invoked with -DOSS_FUZZ=ON.
Changes:
- Use ninja for better control of parallel linker jobs.
- Installing files (make install) is no longer needed.
- Avoid setting a long list of cmake options (BUILD_xxx=OFF).
- Remove fortify comment, it does not belong here.
- Remove lemon workaround, it is fixed since v2.9.0rc0-2273-gbb2caa2a9f.
Was looking through this script to see what needs to be done for tensorflow and found this typo. Small fix, just to improve readability of script (or being too pedantic)
* Limit size of inputs for gnupg import fuzz target
And adds the targets out of the diff
* Removes option limiting size for gnupg target import
As it is done in the target itself
* Removes max length options for fuzz targets in gnupg
* [harfbuzz] Remove UBSan exemptions
We are clean on all sanitizers now.
* [harfbuz] Seed corpus from more fonts
* [harfbuzz] Share seed corpus amongst the two sanitizers
Is this really what those archives are for? Or just for archival?
* [harfbuzz] Remove hb-subset-get-codepoints-fuzzer
Folded into hb-subset-fuzzer upstream.
Our new fuzzer requires CPython, and I'm getting extremely
confusing (and implausible-looking based on reading of CPython source)
msan issues, so let's disable msan for this project for now.
webp_enc_dec: Limit alpha_quality to 99 when method is 6
The target fuzz_webp_enc_dec with asan crashes (timeout)
during encoding with heavy alpha compression.
Clamp alpha compression parameters for images bigger than 16*16.
Bug report 10838
Max Moroz