On Fri, 20 Jul 2018 18:51:33 -0500, Scott Cheloha wrote: > The block sizes sum to 1 when cast and malloc(3) breaks us off a byte. > read(2) then sees that it would be writing outside the valid address > space and fails.
OK millert@ for the fix. > Maybe pedantic, but should we also be checking for addition overflow? > OpenBSD is fine -- those additions will not overflow, as we check prior to > this point that all block sizes are <= SSIZE_MAX. But the standard doesn't > guarantee that SSIZE_MAX * 2 < SIZE_MAX. Are such considerations left to > the porter? I think this is safe on any 2s-complement machine unless sizeof(ssize_t) < sizeof(size_t). I wouldn't worry about it. - todd
