Is there some way to force inlining of memcpy? On my target (SPU), for a certain type T T x[10], y; memcpy (&y, x+3, sizeof(T)); is inlined, but int a = rand() % 10; memcpy(&y, x+a, sizeof(T)); is not. I've tried several optimization flags, do I overlook something causing this behaviour? I would patch gcc to achieve inlining if necessary. Thank you Thomas
