David LeBlanc and I (and a bunch of others) just had a little email exchange about some fascinating integer overflow vulnerabilities in gcc.
Long story made short: the code you add to detect integer overflows might actually be removed by the compiler because of assumptions made by the optimizer. I was going to write a post on the subject, but David did it for me :-) A frankly, no-one knows int-overflow science quite like LeBlanc.
I can't help but be reminded of another compiler optimization vulnerability we discovered a few years back. I wonder what else might be in store for us from the world of compiler optimizations?