| 7 Oct 2025 |
lovesegfault | woke up to Crashes : 370949 [unique: 78, blocklist: 0, verified: 67] | 14:40:05 |
lovesegfault | :D | 14:40:08 |
lovesegfault | it's weird, the honggfuzz session stalls at ~4% coverage, but found a lot of unique crashes. the afl++ session gets up to >16% coverage, but finds no crashes whatsoever 😵💫 | 16:18:59 |
lovesegfault | i wonder what's up | 16:19:07 |
lovesegfault | oh, afl++ found one! | 16:39:08 |
fzakaria | oh really ? | 18:56:34 |
fzakaria | are you running nix eval ? Or you hooked it up individually with a harness | 18:56:50 |
fzakaria | my laptop was going too slow | 18:57:37 |
lovesegfault | i hooked it up with a harness | 19:00:05 |
lovesegfault | i'll make a draft pr in a bit | 19:00:11 |
lovesegfault | https://github.com/NixOS/nix/pull/14175 | 19:27:22 |
fzakaria | that makes sense so it goes faster | 19:51:47 |
fzakaria | i just wasn't sure if i would "reset it" correctly | 19:51:53 |
fzakaria | did you give it a dictionary? | 19:53:11 |
lovesegfault | i did not | 19:54:37 |
fzakaria | I saw you could; the bash.dict in AFL++ is a good example. That PR is pretty slick; I don't understand yet the N fuzzers but must be a feature with harness i'm not familiar with | 19:56:27 |
lovesegfault | i took it from this: https://aflplus.plus/docs/fuzzing_in_depth/#c-using-multiple-cores | 19:58:01 |
fzakaria | afl-tmin is pretty cool too. | 19:58:09 |
lovesegfault | i'm trying out a dictionary, let's see | 19:58:19 |
fzakaria | (even minimizes without a crash to give you same coverage) | 19:58:25 |
fzakaria | the sweet of tools is insane | 19:58:57 |
fzakaria | like https://aflplus.plus/docs/technical_details/#13-the-afl-analyze-tool | 19:59:24 |
lovesegfault | i wonder if i need each secondary fuzzer to get passed -x as well | 20:00:15 |
fzakaria | here is my dumb (I wanted to learn it) post https://fzakaria.com/2025/10/07/fuzzing-for-fun-and-profit | 20:00:16 |
fzakaria | might be cool to look at queue and have AFL try to do nix build; see what interesting derivations it comes up with :P | 20:01:44 |