MartinNowack

@operasfantom No - looks fantastic. There is only the undefined behaviour issue. I really like that the *san support is streamlined like this. Excellent job!!

Hi @EpureRares, It's hard to guess from your report. In case you test C++ applications, you should add `--libcxx` as part KLEE's invocation to link against the libc++ bitcode library....

@Laplace1814 I just saw you opened and closed this PR a couple of times (#1454, #1455, #1546). You don't need to do this, just commit changes to your branch: https://github.com/Laplace1814/klee/tree/my-pr...

@Laplace1814 I understand, it's a bit strange at the beginning - but you'll get used to it. And thanks a lot for your effort! For writing test cases, start with...

@Laplace1814 Great, now reduce the test case to its minimum: (I use godbolt for example: https://godbolt.org/z/3nTYqY6av) So, this is your original source code (reduce it as much as possible -...

@Laplace1814 Excellent. Now, add back your implementation to check that KLEE doesn't crash even with the intrinsic enabled.

@Laplace1814 No. Here is your original implementation https://github.com/klee/klee/pull/1457/commits/3246ca9e7a48fad844e6a38832bdcbd09e0fec1a But with all these changes, it's not there anymore. So you need to bring it back. At the end, the goal of...

@Laplace1814 Excellent, now let's get that code in shape: Please replace `unsigned mode` with an `enum class mode` and define it. (Here is a good example - it's a C++11...

What setup are you using (which version of KLEE/LLVM/STP)?

@gftea Does building the Docker image fail for you? In that case, please add the error message to this report.