On Tue, 29 Oct 2013, Richard Biener wrote: > LLVM covers addition, subtraction and multiply on signed and unsigned > int, long and long long types. Not sure why they offer anything for > unsigned - possibly for size_t arithmetic and security concerns with > malloc? For practicability and to be less error-prone I'd have done > the builtins in a type-generic way (like tgmath) as using the > wrong typed builtin can lead both to undetected overflow, unwanted > truncation of arguments and possibly memory overflow of 'result' > (if you ignore warnings about incompatible pointer types).
One difficulty of doing it as type-generic is that you might sometimes want to detect overflow on types smaller than int, which C is rather prone to promote to int. (Of course in such cases, you can also write naive code computing e.g. short_1 * short_2 as an int value then comparing with SHRT_MIN and SHRT_MAX, but it seems reasonable to want to use the same interface as for wider types.) In general, the type that arithmetic would happen on is not necessarily the same as the type you then store the result in, and it's the latter type you want to know if the arithmetic overflows. Maybe it could be type-generic on the result pointer only (i.e., say "compute the result as if the arguments are infinite-precision signed integers, store the result reduced modulo 2^N and return overflow status if that isn't the same as the infinite-precision result", where the first two arguments could have arbitrary integer types). -- Joseph S. Myers jos...@codesourcery.com
