Due to the page limitation, we didn't include the raw data for hyperparameter study and all 10 rounds coverage/reach information in the paper. In this folder, we attached the raw coverage/reach data along with the script for p-value calculation.
In folder p-value, we include the raw coverage/reach/bug data in csv format,
and script p-val-calc.py allows calcuate p-value based on FF_AFL or FF_AFL++.
We use scipy 1.10.0 and as indicated by the maintainer
, different version of scipy
might approximate differently, therefore can results in different p-values.
Regarding the hyperparameter study, the raw data could be found in folder hyperparam,
we only briefly summarized the best vs worst results in the appendix.
In folder vuln, we attached a full list of bugs and CVEs FishFuzz found.
In folder artifact, we attached the dockerfile and the script to reproduce our evaluation results (todo).
In folder fuzzbench, we include an experimental configuration for fuzzbench, to build it locally, copy the FF_AFL to that folder.