PCRE2 is a complicated target which is prone to timeouts. This timeout-proneness is caused, in part, by the input representing instructions rather than flat data. Additionally, the state space of the program is significantly larger than the code coverage suggests. These two issues together make this a very hard target to fuzz, which has led to interesting problems in OSS-Fuzz and in local testing.
I think this target is interesting from a benchmarking perspective as it introduces complications which are not present in other targets in Fuzzbench and would like to add it to the standard benchmark suite. I have used the standard OSS-Fuzz integration script.
PCRE2 is a complicated target which is prone to timeouts. This timeout-proneness is caused, in part, by the input representing instructions rather than flat data. Additionally, the state space of the program is significantly larger than the code coverage suggests. These two issues together make this a very hard target to fuzz, which has led to interesting problems in OSS-Fuzz and in local testing.
I think this target is interesting from a benchmarking perspective as it introduces complications which are not present in other targets in Fuzzbench and would like to add it to the standard benchmark suite. I have used the standard OSS-Fuzz integration script.
CC @PhilipHazel @alexdowad for input.