Back again with something that is stumping me or I'm having a major brain fart. Either way here is what the deal is. I wanted to create a zune like window using two methods. The first would be using Dwm and I have successfully accomplished this. However, I'm running into one problem...resizing the window. Since the borders are gone I know that I have to handle the WM_NCHITTEST message but I'm kind of stuck here...somewhat. Here is the example of my code so far:
_WM_NCHITTEST:
assume edi:ptr POINT
assume esi:ptr RECT
push [ebp+14h]
pop edi
push 1
push edi
push [ebp+08h]
push NULL
call MapWindowPoints
lea esi, rc
push esi
push [ebp+08h]
call GetClientRect
ret HTNOWHERE
Now this code will actually cause the program to freeze any time the mouse enters into the window and the same applies if I remove ret HTNOWHERE to jmp _DEFER. My question is how do I properly handle this message. The examples that I have seen about resizing borderless forms use this same method but return a result of HTTOPLEFT, HTTOP, etc depending on calculations from MapWindowPoints and GetClientRect. But any time I try to handle this message and the mouse enters the window it freezes and force closes.
I have attached the code but please excuse the mess. :biggrin:
The argument of the RET[N] instruction are the numbers of byte that must be cleaned up (16 for the WndProc) and not a return value. EAX is used for that purpose. Also, you may want to turn of the prologue and epilogue for the WndProc, because the stack frame is currently created two times.
Quote from: qWord on January 23, 2013, 11:03:46 PM
The argument of the RET[N] instruction are the numbers of byte that must be cleaned up (16 for the WndProc) and not a return value. EAX is used for that purpose.
Indeed a source of confusion. Look at the proggie below with Olly to understand what the three variants do.
include \masm32\include\masm32rt.inc
.code
AppName db "Masm32:", 0
aproc proc arg1, arg2
return 12h
ret 34h
retn 56h
aproc endp
start: invoke aproc, chr$("argA"), chr$("argB")
exit
end start
Assembling with ML 6.15 there is no automatic stack frame for WndProc because there are no parameters. Do the newer versions of ML do this differently?
i hope not :biggrin:
but - why doesn't WndProc have 4 parameters ?
at any rate, you could force a prologue and epilogue by creating a dummy local variable
Quote from: MichaelW on January 24, 2013, 03:29:56 PM
Assembling with ML 6.15 there is no automatic stack frame for WndProc because there are no parameters. Do the newer versions of ML do this differently?
no, i was my mistake.
Thanks for the help guys but I finally found the mistake. For some reason (why I didn't catch it) I used the following:
lea esi, rc
instead of
lea esi, offset rc
once I changed this it worked perfectly.
you should note that the code you post breaks with calling convention and corrupt the stack (WndProc). That it runs, has something to do how different windows version handles such failures ( e.g. the 64 Bit versions ignore any exception in the WndProc).
Quote from: jasonabullard on January 25, 2013, 01:00:50 AM
Thanks for the help guys but I finally found the mistake. For some reason (why I didn't catch it) I used the following:
lea esi, rc
instead of
lea esi, offset rc
once I changed this it worked perfectly.
Lucky you :t
However, it helps occasionally to look at code through Olly's eyes:
.data
rc RECT <12, 34, 56, 78>
.code
lea esi, rc
lea esi, offset rc
mov esi, offset rc
00401025 ³. 8D35 00804000 lea esi, [rc]
0040102B ³. 8D35 00804000 lea esi, [rc]
00401031 ³. BE 00804000 mov esi, offset rc ;)
Hello ,
Were did you get uxtheme.inc and uxtheme.lib?
or can anybody login here and get it? http://www.asmcommunity.net/board/index.php?topic=16091.0
Thank you
I found one version