On Tue, Dec 13, 2016 at 2:36 AM, Martin Sebor <mse...@gmail.com> wrote: > The attached patch avoids infinite recursion when traversing phi > nodes in maybe_warn_alloc_args_overflow by using a bitmap to keep > track of those already visited and breaking out.
It looks somewhat excessive (the whole PHI node walk looks exponential in the number of alloca calls given a good enough testcase). It also looks like operand_signed_p really returns only a wild guess, neither conservatively true or false. Is that correct? Can you instead scrap the weird anti-range handling please? Thanks, Richard. > Thanks > Martin
