On 1 August 2016 at 09:06, Jeffrey Walton wrote: >>> $ cat test.cc >>> #include <x86intrin.h> >>> >>> #if _LP64 || __LP64__ >>> typedef unsigned long my_u64; >>> #else >>> typedef unsigned long long my_u64; >>> #endif >> >> The definition doesn't depend on ILP32 or LP64: >> >> ../gcc/config/i386/immintrin.h:_rdrand64_step (unsigned long long *__P) >> >> Looks like you want unsigned long long for Intel. If that doesn't work >> for Neon then use the preprocessor to check for that, not for LP64. > > Thanks. > > Forgive my ignorance... 'unsigned long' is 64 bits, which is what that > interface needs (if I am reading the docs correctly). The functions takes a pointer to an unsigned 64-bit type, not necessarily unsigned long. long is 32-bits on IA-32, 64 bits on x86_64. > It looks like > 'unsigned long long' is 64-bits too (other wise, GCC would not be > using it). long long is 64-bits on IA-32 and x86_64. That's probably why GCC uses unsigned long long. So you should not be using a different type for IA-32 and x86_64. > Is it safe to cast a pointer of type 'unsigned long' to a pointer of > 'unsigned long long' when using this particular function? No, they are distinct types. > Does UL and > ULL provide the same alignment guarantees?
