On 07/04/15 10:51 -0400, Hans-Peter Nilsson wrote:
I was more thinking of something like:#include <atomic> #include <iostream> using std::cout; using std::endl; struct SoSo { double d; int x alignas(sizeof(int)); }; SoSo s __attribute__((__aligned__(16))); int main(void) { cout << "alignof(s): " << alignof(s) << endl; cout << "alignof(s.d): " << alignof(s.d) << endl; cout << "alignof(s.x): " << alignof(s.x) << endl; } in which I fear s.x would get an alignof the same as the s.d or s, now or after a while, i.e. higher than specified. (I get for cris-elf at revision 221891: alignof(s): 16 alignof(s.d): 1 alignof(s.x): 4 which is kind-of-expected except I thought s.d would get the s alignment so that just leaves it open whether that could possibly change.)
The docs are clear that alignof(s.x) is not related to its position in struct SoSo: https://gcc.gnu.org/onlinedocs/gcc/Alignment.html I'm not going to worry about that behaviour changing.
