调用CreateWindowExA后RIP损坏，疑栈分配错误求助

在Win64汇编开发中，调用CreateWindowExA函数后出现RIP寄存器损坏问题，通过x64dbg调试发现异常发生在NtUserCreateWindowEx调用之后，怀疑是栈分配操作存在错误。手动编写的汇编代码和使用Invoke宏的版本均出现相同错误，现将相关代码及编译配置整理如下：

手动编写的汇编代码 #

include win64a.inc

.code
WinMain proc
push rbp                        ; save old rbp
mov rbp, rsp                    ; set-up rbp
sub rsp, 20h                    ; allocate space for local vars
mov rcx, 0                      ; 1 param
mov rdx, IDC_ARROW              ; 2 param 
sub rsp, 20h                    ; allocate home-space
call LoadCursorA                
mov [rbp - 8h], rax             ; 1 local param, hCursor 
mov rcx, 0                      ; 1 param
mov rdx, IDI_APPLICATION        ; 2 param
call LoadIconA                  ; home-space allocated at 11 
mov [rbp - 10h], rax            ; 2 local param, hIcon
add rsp, 20h                    ; deallocate home-space
; Develope WNDCLASSEXA structure
push [rbp-10h]                  
pushaddr wndName
push 0
push COLOR_WINDOWFRAME
push [rbp-8h]
push [rbp-10h]
push IMAGE_BASE
push 0
pushaddr WndProc
push sizeof WNDCLASSEX

mov rcx, rsp                    ; 1 param - ptr to the structure
sub rsp, 20h                    ; allocate home-space
call RegisterClassExA
sub rsp, 40h                                ; Allocate stack for params
xor rcx, rcx                                ; dwExStyle
lea rdx, wndName                            ; lpClassName
lea r8, wndName                             ; lpWindowName
mov r9, WS_OVERLAPPEDWINDOW or WS_VISIBLE   ; dwStyle
mov qword ptr [rsp + 20h], 0                ; x
mov qword ptr [rsp + 28h], 0                ; y
mov qword ptr [rsp + 30h], 600              ; w
mov qword ptr [rsp + 38h], 800              ; h
mov qword ptr [rsp + 40h], 0                ; hWndParent
mov qword ptr [rsp + 48h], 0                ; hMenu
mov qword ptr [rsp + 50h], IMAGE_BASE       ; hInstance
mov qword ptr [rsp + 58h], 0                ; lpParam
call CreateWindowExA
add rsp, 40h                                ; Deallocate stack
mov [rbp-18h], rax                          ; 3 local param, hWnd
leave
WinMain endp

WndProc proc hWnd:HWND, uMsg:UINT, wParam:WPARAM, lParam:LPARAM
WndProc endp

.data
wndName db "test", 0
end

win64a.inc头文件内容 #

OPTION DOTNAME
OPTION PROLOGUE:rbpFramePrologue
OPTION EPILOGUE:rbpFrameEpilogue
include win64.inc
include temphls.inc
include ntdll.inc
includelib ntdll.lib
include user32.inc
includelib user32.lib
include gdi32.inc
includelib gdi32.lib
include kernel32.inc
includelib kernel32.lib
include shell32.inc
includelib shell32.lib
;---------------------------
pushaddr macro x
    db 68h
    dd x
endm
IMAGE_BASE  equ 400000h

使用Invoke宏的版本代码 #

include win64a.inc  

.code 

WinMain proc <13>
local hWnd:qword
local msg:MSG
local wnd:WNDCLASSEX
mov rdi, offset wndName                 ;WndName
invoke LoadCursorA, 0, IDC_APPSTARTING
mov wnd.hCursor, rax
invoke LoadIconA, 0, IDI_APPLICATION
mov wnd.hIcon, rax
mov wnd.hIconSm, rax
mov wnd.lpszClassName, rdi
mov wnd.hInstance, IMAGE_BASE
mov rax, offset Wndproc
mov wnd.lpfnWndProc, rax
mov wnd.cbSize, sizeof WNDCLASSEX
mov wnd.hbrBackground, COLOR_WINDOWFRAME
lea rax, wnd
invoke RegisterClassExA, rax
invoke CreateWindowExA, 0, rdi, rdi, WS_OVERLAPPEDWINDOW or WS_VISIBLE, 0, 0, 600, 800, 0, 0, IMAGE_BASE, 0
mov hWnd, rax
ret
WinMain endp

Wndproc proc <12,8,4,8,8> hWnd:HWND, uMsg:UINT, wParam:WPARAM, lParam:LPARAM
jmp NtdllDefWindowProc_
ret
Wndproc endp
.data
wndName db "test", 0
end

编译链接配置 #

ml64.exe编译命令 #

ml64.exe /I"%masm64_path%/include" /Zp16 /c /Cp %filename%.asm >>errors.txt

link.exe链接命令 #

link /LIBPATH:"%masm64_path%\lib" /LARGEADDRESSAWARE:NO /SUBSYSTEM:WINDOWS ^
 /STUB:%masm64_path%\bin\stubby.exe ^
 /entry:WinMain /fixed %filename%.obj

内容的提问来源于stack exchange，提问作者mantissa