Minimize fuzz corpora with all fuzz engines used for fuzzing #167
Comments
This was referenced Aug 21, 2024
fanquake
added a commit
that referenced
this issue
Sep 4, 2024
56730ca delete_nonreduced_fuzz_inputs: also reduce with afl-cmin (dergoegge) 75c040f delete_nonreduced_fuzz_inputs: use llvm 18 (dergoegge) Pull request description: * Use llvm 18 * Minimize corpora with both afl++ and libfuzzer (#167) ACKs for top commit: maflcko: tested ACK 56730ca Tree-SHA512: d3b6b4a40b55ee012392f21fe68776b1c101b4820fac5b49db1d1cb292ac87e474acd96f35bb4de3dc6e821042726e067e5261d89ed30b79ccb014741360b3fd
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
At the moment, our script for minimizing the fuzz corpora only uses libFuzzer. Not minimizing with the same engines that we use for fuzzing will lead to coverage loss each time we minimize, because the engines differ in the way they collect coverage (e..g libFuzzer uses llvm's pc-guard but afl++ has it's own plugins).
(I need to find a source other than word of mouth on this but llvm's pc-guard apparently under reports coverage quite heavily due to missing support for instrumenting
iselectexpressions).We should extend the script to minimize with afl++ and honggfuzz as well.
The text was updated successfully, but these errors were encountered: