Instrumented binary crashes with SIGABRT

[metalnem]

Instrumented binary crashes with SIGABRT

[nulldatamap]

Turns out that the original problem I was trying to solve was caused by calling instrumented code outside Fuzzer.Run.

Sorry for wasting your time.

[Metalnem]

You should actually have two projects: one that you will instrument with sharpfuzz (class library where you would put the Repro function from your example), and the other one that will reference the instrumented assembly (console application that will run Fuzzer.Run function). The console app will be the target when running afl-fuzz.

[nulldatamap]

Yes, for my project that's how I did it and ran into this crash.
While reducing this case I found that I get the same crash even when it's just one project.
So splitting it up (while more correct) still causes this crash for me.

[Metalnem]

Single project will always crash. Two projects is the only way that fuzzing will work. In your case, two projects probably crashed because of some other reason. One possible reason is that your initial case was starting with byte 97: if your initial test cases are crashing the program, fuzzer will stop (test cases must always run without exceptions).

I've created two separate projects to try to reproduce your bug. They are in the attached archive, and you can run them by extracting the archive and executing the following script:

cd Runner
dotnet build
sharpfuzz bin/Debug/netcoreapp2.2/Target.dll
afl-fuzz -i testcases/ -o findings dotnet bin/Debug/netcoreapp2.2/Runner.dll

And everything should work as expected.

One additional thing to keep in mind is that you should always consume the whole input stream (I know that this was just a toy example, but I just want to make sure that you are aware of that).