Vesal Vojdani

This is workable, but it does somewhat bias the race detection category to multithreaded reachability checking... I recognize the value of witnesses, but it is fairly difficult for us to...

I think it may have been this paper: [Deterministic parallel fixpoint computation](https://dl.acm.org/doi/10.1145/3371082). The abstract looks impressive in terms of achieved parallelism, but I didn't check if it would work for...

So I don't really see any good reason to keeping the localized vars set... I think the motivation may have been trying to deal with cases where regions collapse and...

This is not a concern for running the test, but `pngtest.c` is [referenced](https://github.com/glennrp/libpng/blob/e755fb79ba945fea8a318dc343e73d22a39e2f4e/libpng-manual.txt#L60) multiple times in the manual as a demonstration of how to use the library. This usage pattern...

Hmm... I tried the Juliet suite examples again with enabled. The **above examples work**, but there seems to be some additional trickiness with the way Juliet suite wrappers create threads...

What is the status of this thing now? You have a student working on the remaining issues for this?

Yes, the SEGV issue that remains here is not a problem within libpng. The problem is that pngwriter has some undefined behavior that confuses ASAN, and this can be fixed...

Note: This has nothing to do with the SEGV "vulnerability", but it does fix the ASAN reports of a memory leak within `png2pnm`, which people keep reporting. The fix requires...