Hm, good point about the bitfields. The paper I cite [1] actually talks specifically about bitfields as their precise semantics in the presence of "poison"-styl

Hm, good point about the bitfields. The paper I cite [1] actually talks specifically about bitfields as their precise semantics in the presence of "poison"-style uninitialized memory is not entirely clear yet. [1]: http://www.cs.utah.edu/~regehr/papers/undef-pldi17.pdf http://www.cs.utah.edu/~regehr/papers/undef-pldi17.pdf