libafl_libfuzzer optimization experiment - #1903
addisoncrump wants to merge 6 commits into
Conversation
402ce53 to
3060a52
Compare
|
having some linkage issues |
47223ec to
79256a9
Compare
|
Found an issue with one of our recent optimisations. Marked as draft until it's fixed up. |
297871f to
04a773e
Compare
|
Good to go! Hunted down all the bugs with the recent optimisations (that I'm aware of... 💀) libafl_libfuzzer is not affected by the issue identified in #1902 because we don't use libafl_cc, so this is good to go whenever y'all are ready to fire the experiment. |
|
Rebased on the libafl updates used in #1902 for sanity. |
|
Hi @addisoncrump, is this ready for experiments? If so, would you mind making a trivial modification to service/gcbrun_experiment.py? In addition, could you please write your experiment request in this format? For example, I reckon the command for this case is: Where FuzzBench will reuse existing results of |
|
We have a couple of more changes in the queue ready, so I'll send this in in a bit 🙂 |
|
this is ready with all the pr merged? @addisoncrump |
|
Yup, let me update the commit |
|
All of the relevant libafl changes landed, so I removed libfuzzer_best. For now, this is good to go ✔️ Let's compare:
Thanks! |
|
Well, that didn't work. Coming back to this in the morning. |
|
Superceded by other recent PRs. |
We made some heavy optimisations and new features that we'd like to test. Can an experiment be run for:
libafl (depends on Update LibAFL #1902; temporarily rebased)please skip while we remediatelibafl_ccThanks!