Re: why MASM generated this code
From: Frank Kotler (fbkotler_at_comcast.net)
Date: 12/31/03
- Previous message: Randall Hyde: "Re: HLA v1.60 is now available"
- In reply to: Madhur: "why MASM generated this code"
- Next in thread: Aharon Lavie: "Re: why MASM generated this code"
- Reply: Aharon Lavie: "Re: why MASM generated this code"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 31 Dec 2003 02:26:56 GMT
Madhur wrote:
> Why this code is generated by the MASM.
> The procedure accepts 4 bytes of arguments but it pushes
> 6 bytes onto stack.
>
> anyproc ptoro arg:dword
>
> a1 db "sdfdf",0
>
> invoke anyproc,a1
I'm Nasmist, and I may be misunderstanding Masm's syntax, but wouldn't
you want "invoke anyproc, offset a1" to get a dword parameter? If I'm
correct, the invocation doesn't match the prototype (I assume you
spelled that correctly in the "real" code...). I would expect Masm to
complain about this, but I don't know what Masm does...
> * push 000h ;4 bytes pushed
Is it possible we're only pushing 2 bytes here? (it would be preceeded
by a "66h" size override prefix, assuming that this is 32-bit code) If
that were the case, it would serve to keep the stack 4-byte aligned...
> * mov al, a1
> * movzx ax, al
> * push ax ;2 bytes pushed
These three lines are more-or-less what I'd expect if you were asking
for a byte parameter... Try "offset" and see if it generates what you
expect.
This isn't intended as "Masm-bashing" (although it may sound the same),
but if you want complete control over the code that's emitted, don't
*use* the "compiler-like" features that your assembler may provide.
> 2nd question
> ========
> I have seen this code to be used in local variables.
> lea eax,[ebp-4]
>
> What is meaning of lea in the context of registers?
> Registers dont have addresses?
> Shouldn't lea be used only with program variables.
> What is the meaning of this statement.
That's pretty much where you *do* need lea. I see a lot of code that
uses "lea eax, myvar" ("lea eax, [myvar]" in Nasm syntax). This
essentially says, "calculate the address of the variable whose address
is myvar". This works, but is pointless - "mov eax, offset myvar" would
do the same thing, but shorter. Some versions of Masm actually changed
your "lea eax, myvar" to "mov eax, offset myvar", but I understand it
doesn't do that anymore - haven't tried it...
When you've got a local variable on the stack, "offset" won't work. You
could "mov eax, ebp"/"sub eax, 4" (or "add eax, -4" which is closer to
what lea actually does) - lea just does the calculation in a single
instruction. (it's not using the address "of" ebp - as you observe, ebp
doesn't have an address - but the address "in" ebp)
I hope that doesn't make things *more* confusing...
Best,
Frank
- Previous message: Randall Hyde: "Re: HLA v1.60 is now available"
- In reply to: Madhur: "why MASM generated this code"
- Next in thread: Aharon Lavie: "Re: why MASM generated this code"
- Reply: Aharon Lavie: "Re: why MASM generated this code"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|
