Julius Pfrommer
fb4a98f655
JSON: Switch to another base64 implementation after oss-fuzz problems
|
5 years ago | |
|---|---|---|
| .. | ||
| fuzz_binary_message_corpus | 6 years ago | |
| fuzz_json | 5 years ago | |
| CMakeLists.txt | 5 years ago | |
| README.md | 5 years ago | |
| check_build.sh | 5 years ago | |
| corpus_generator.c | 5 years ago | |
| custom_memory_manager.c | 5 years ago | |
| custom_memory_manager.h | 5 years ago | |
| fuzz_binary_decode.cc | 5 years ago | |
| fuzz_binary_message.cc | 5 years ago | |
| fuzz_binary_message.options | 6 years ago | |
| fuzz_binary_message_header.dict | 6 years ago | |
| fuzz_json_decode.cc | 5 years ago | |
| fuzz_json_decode_encode.cc | 5 years ago | |
| fuzz_src_ua_util.cc | 5 years ago | |
| fuzz_src_ua_util.options | 6 years ago | |
| fuzz_src_ua_util_endpoints.dict | 6 years ago | |
| generate_corpus.sh | 6 years ago | |
| oss-fuzz-copy.sh | 7 years ago | |
| ua_debug_dump_pkgs_file.c | 5 years ago | |
README.md
I Can haz fuzz
open62541 is continuously tested with the awesome oss-fuzz project from Google: https://github.com/google/oss-fuzz
Currently tested is processing of binary messages and encoding/decoding of binary encoded data.
Reproduce locally
Reproduce build failure
You can just execute the script under tests/fuzz/check_build.sh, which does the following:
# clone oss-fuzz repo
cd oss-fuzz
# Change $OPEN62541_DIR to your local checkout of open62541
python infra/helper.py build_fuzzers --sanitizer address open62541 $OPEN62541_DIR && python infra/helper.py check_build --sanitizer address open62541
Reproduce issues
Download the testcase file and store it e.g. in your Download folder.
# clone oss-fuzz repo
cd oss-fuzz
# Change $OPEN62541_DIR to your local checkout of open62541
# And change $DOWNLOADS to your download location
python infra/helper.py build_fuzzers --sanitizer address open62541 $OPEN62541_DIR && python infra/helper.py reproduce open62541 fuzz_binary_decode $DOWNLOADS/clusterfuzz-testcase-minimized-fuzz_binary_decode-5686300273803264
Status
Update the corpus
To update the current corpus used for fuzzing you need to follow these steps. It will execute all the unit tests, dump the received data packages to a directory and then update and merge the corpus.
The script will create two directories:
open62541/build_fuzzandopen62541/build_corpus. Make sure that these directories are not existing or do not contain any important data.Run the generate script:
open62541/tests/fuzz/generate_corpus.sh
This script will build all the unit tests, dump the packages and then merge the current corpus with the new packages.
- If there is new coverage with the generated data there will be new files in the directory:
open62541/fuzz/fuzz_binary_message_corpus/generated
Commit the new files and then you can delete the build directories created in step 1.