steakhal wrote: > In the state dump I see that `stdout` seems to be NULL (last line in > "constraints"). This explains why the `StateNull` becomes NULL, because call > to `assumeNoAliasingWithStdStreams` was called already. I think the better > solution is to check NULL-ness of the std stream variable > `assumeNoAliasingWithStdStreams` and do not assume if it is NULL. There is > not a case when `fopen` returns non-null for sure, but at least not in the > current situation, so the current fix is not as good. We could add an > `assert` to check if both `StateNull` and `StateNotNull` are non-zero.
Exactly. I didn't want to rush too much, but I can share that my current idea is to call `assumeNoAliasingWithStdStreams` only on the success path. https://github.com/llvm/llvm-project/pull/100901 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits