You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Aug 23, 2024. It is now read-only.
Hi @thdusdl1219, I have a question about the method of coverage calculation.
I tried to run your program, but the coverage rate is approximately a straight line.
First, the target program is JavaScriptCore. By adjusting the afl-llvm-pass.so.cc code, the number of inserted piles is recorded each time the pile is inserted, that is, the variable inst_blocks is summed. The result is that JavaScriptCore has 875,102 points in total. I use ((MAP_SIZE << 3)-count_bits(virgin_bits)) to record the position covered by the fuzzing process. Of course, the coverage of the initial seed will be recorded first. I run the program with a single thread.
After running the original seed, the number of locations covered by the record is 147,483, and the coverage rate is about 17%. After running for 10 hours, the number of locations covered was 148,864, and the coverage rate was still around 17%. There was no significant improvement as mentioned in the paper. Of course, it may be because the target program is different, but I don't think it should have such a big impact.
It may be that the calculation method is different, so I want to know how you calculate the coverage of the target program.
Thank you.
The text was updated successfully, but these errors were encountered:
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Hi @thdusdl1219, I have a question about the method of coverage calculation.
I tried to run your program, but the coverage rate is approximately a straight line.
First, the target program is JavaScriptCore. By adjusting the afl-llvm-pass.so.cc code, the number of inserted piles is recorded each time the pile is inserted, that is, the variable inst_blocks is summed. The result is that JavaScriptCore has 875,102 points in total. I use ((MAP_SIZE << 3)-count_bits(virgin_bits)) to record the position covered by the fuzzing process. Of course, the coverage of the initial seed will be recorded first. I run the program with a single thread.
After running the original seed, the number of locations covered by the record is 147,483, and the coverage rate is about 17%. After running for 10 hours, the number of locations covered was 148,864, and the coverage rate was still around 17%. There was no significant improvement as mentioned in the paper. Of course, it may be because the target program is different, but I don't think it should have such a big impact.
It may be that the calculation method is different, so I want to know how you calculate the coverage of the target program.
Thank you.
The text was updated successfully, but these errors were encountered: