x64汇编调用printf崩溃

原来的代码在Linux上运行正常

BITS 64

extern exit
extern printf

global main

SECTION .data
fmt0    db '%hd', 0Ah, 0h

SECTION .text
main:
label_1:    enter 6, 0
label_2:    xchg rbp, r8

label_9:    push r8

label_10:   mov rsi, 1
label_11:   mov rdi, fmt0
label_12:   xor rax, rax
label_13:   call printf

label_14:   pop r8

label_18:   xchg rbp, r8
label_19:   leave
label_20:   call exit

后来把代码移植到Windows上，把传参改成Windows的调用约定

label_10:   mov rdx, 1
label_11:   mov rcx, fmt0

nasm汇编，MinGW-GCC链接到64位msvcrt.dll。

运行时崩溃，gdb调试发现调用printf前后栈中内容不一致，r8没有恢复。

---

找到一个可能的原因
Windows下用register传参依旧需要分配栈空间。并且需要16字节对齐。
The x64 Application Binary Interface (ABI) is a 4 register fast-call calling convention, with stack-backing for those registers.