I knew in the common calling convention of integer argument it’s rdi rsi ... , but what about the doubles? As the argument in the test is actually a struct, am I supposed to use rdi as the address of the struct? But I’ve seen from other’s solutions they are just using rsp+8 rsp+16 rsp+24…
You’d use rdi if the struct was passed as a pointer. In this case, however, it is passed by value and thus would be pushed on the stack. In System V AMD64 ABI, ignoring floating point arguments, the first integer/pointer arguments are passed to the corresponding 6 registers (rdi, rsi, rdx, rcx, r8, r9) and the rest are pushed on the stack in reverse order.
Also, it might be useful to see what GCC outputs in different cases.