On 12/05/2017 10:27 AM, Jakub Jelinek wrote: > The most important change I've done in the testsuite was pointer-subtract-2.c > used -fsanitize=address,pointer-subtract, but the function was actually > doing pointer comparison. Guess that needs to be propagated upstream at > some point. Another thing is that in pointer-compare-1.c where you test > p - 1, p and p, p - 1 on the global variables, we need to ensure there is > some other array before it, otherwise we run into the issue that there is no > red zone before the first global (and when optimizing, global objects seems > to be sorted by decreasing size).
Hi. I've just done review request for that: https://reviews.llvm.org/D41481 Apart from that I enhanced detect_invalid_pointer_pairs run-time option that can control whether a pointer comparison (or subtraction) with nullptr is reported or not: https://reviews.llvm.org/D41479 Martin