MASM to NASM (or other)
in [ASM]
|
From: playstationdude on 7 Sep 2007 01:09 hey everyone I'm a CS student who's learning x86 assembly language right now. The book we're using, Assembly Language for Intel-Based Computers by Kip Irvine is written for MASM. Not owning a Windows computer, I cannot use MASM. The preface of the book said it is easy to convert the code from MASM to TASM or NASM, but hard to for the GNU assembler. My instructor said he didn't know how to do this, but he would be fine if I could figure out how to do it. Does anyone know anything about this? Thanks!
From: cr88192 on 7 Sep 2007 03:20 "playstationdude" <playstationdude(a)gmail.com> wrote in message news:1189141795.800646.158630(a)r34g2000hsd.googlegroups.com... > hey everyone > I'm a CS student who's learning x86 assembly language right now. The > book we're using, Assembly Language for Intel-Based Computers by Kip > Irvine is written for MASM. Not owning a Windows computer, I cannot > use MASM. The preface of the book said it is easy to convert the code > from MASM to TASM or NASM, but hard to for the GNU assembler. My > instructor said he didn't know how to do this, but he would be fine if > I could figure out how to do it. Does anyone know anything about > this? Thanks! > not too hard, but not trivial either. easier if you are using code that does not use a lot of masm-specific builtin cruft. maybe, find some examples of nasm code, and look at that. note that, beyond this, dos, windows, and linux, are very different OS's, so, depending on what you are doing, it may take a lot more modification than just converting between assemblers to make it work... here are a few major differences (sorry if my MASM examples have errors, as I haven't really use MASM in over a decade): NASM does not use 'PROC' or similar directives, you have to do all this yourself (noting where things are on the stack, ...). NASM does not infer memory references. in MASM, one would type: mov eax, foo but, in NASM, this is: mov eax, [foo] and, in MASM: mov eax, offset bar is, in NASM: mov eax, bar .... gas, yes, is a bit harder: mov eax, ecx becomes: movl %ecx, %eax .... or such...
From: Rod Pemberton on 7 Sep 2007 05:09 "playstationdude" <playstationdude(a)gmail.com> wrote in message news:1189141795.800646.158630(a)r34g2000hsd.googlegroups.com... > hey everyone > I'm a CS student who's learning x86 assembly language right now. The > book we're using, Assembly Language for Intel-Based Computers by Kip > Irvine is written for MASM. Not owning a Windows computer, I cannot > use MASM. The preface of the book said it is easy to convert the code > from MASM to TASM or NASM, but hard to for the GNU assembler. My > instructor said he didn't know how to do this, but he would be fine if > I could figure out how to do it. Does anyone know anything about > this? Thanks! > About a,b,c, or d? a) MASM (Intel syntax) to TASM ? b) MASM (Intel syntax) to NASM (NASM syntax) 1) drop the "t/d/q/word ptr" 2) add size specifiers "byte/word/dword" 3) put memory references in brackets [] 4) ? other... c) MASM (Intel syntax) vs. GAS (AT&T syntax) http://www.ibiblio.org/gferg/ldp/GCC-Inline-Assembly-HOWTO.html http://www.delorie.com/djgpp/doc/brennan/brennan_att_inline_djgpp.html d) all of the above There are some programs out there that do conversions between a few of the [plural of syntax: syntaxes, syntaces, ?]. Searching for you, I find: NOMYSO (perl MASM/TASM->NASM) http://www.devoresoftware.com/nomyso/ Intel2GAS (NASM->GAS) http://www.niksula.cs.hut.fi/~mtiihone/intel2gas/ A86CNVRT (MASM->A86) ftp://ftp.simtel.net/pub/simtelnet/msdos/asmutl/a86cnvrt.zip as386.sed (MASM->GAS) http://www.delorie.com/djgpp/faq/converting/asm2s-sed.html I seem to recall seeing some others, but didn't run across them just now. Rod Pemberton
From: Frank Kotler on 7 Sep 2007 05:42 playstationdude wrote: > hey everyone > I'm a CS student who's learning x86 assembly language right now. The > book we're using, Assembly Language for Intel-Based Computers by Kip > Irvine is written for MASM. Not owning a Windows computer, I cannot > use MASM. The preface of the book said it is easy to convert the code > from MASM to TASM or NASM, but hard to for the GNU assembler. My > instructor said he didn't know how to do this, but he would be fine if > I could figure out how to do it. Does anyone know anything about > this? Thanks! Well... I've converted a fair amount of code from Masm syntax to Nasm syntax - but for a "Windows computer"... although dos code, mostly. Porting code intended for dos/doze (apparently the book does both?) to... your header indicates Linux... that's a different matter! According to the blurb at Amazon, the 16-bit stuff will run under dosemu. (I haven't had much luck with graphics programs under dosemu) The 32-bit stuff just might run under WINE. Have you got WINE installed and running? I tried it the other day, and just got error messages. I can get the examples and libraries from Mr. Irvine's site... in .exe format. :( From what I can see, the code you'll be writing does stuff like: TITLE MASM Template (main.asm) ; Description: ; ; Revision date: INCLUDE Irvine32.inc ..data myMessage BYTE "MASM program example",0dh,0ah,0 ..code main PROC call Clrscr mov edx,OFFSET myMessage call WriteString exit main ENDP END main We could go through that line-by-line and discuss what would need to be changed to get it to assemble with Nasm (or Gas), but it wouldn't run. The code in the libraries you call is completely wrong. WriteString takes the offset (address) of the string to print in edx. Unless I'm mistaken, it'll eventually call WriteFile, with "stdout" as a handle (in Windows, we'll need to "get" that handle if we haven't got it)... parameters will be passed on the stack... WriteFile wants to know the length of the string, so we'll need to find that, and push it in the right place. WriteFile wants the address of a place to return how many bytes were actually written - probably make a temporary variable on the stack for that. For Linux, our WriteString function would be quite different. We still need the length of the string, we know the handle for stdout... We'll be using "sys_write", not WriteFile... returns bytes written in eax. A bit simpler, actually. Sys_write wants the address of the string in ecx, not edx - easily fixed: mov ecx, edx ; now we want the length in edx xor edx, edx getlen: cmp byte [ecx + edx], 0 jz gotlen inc edx jmp getlen gotlen: ; file handle to write to in ebx - stdout is 1 mov ebx, 1 ; system call number in eax - __NR_write is 4 mov eax, 4 ; call the OS int 80h ret Now if we put that code in a file ("decorated" a little bit), assembled it, and put the resulting file into a library (along with Clrscr and exit), the example - suitably "translated", assembled, and linked agaist that library - would work in Linux (barring errors). I don't expect you, as a beginner, to do all that. If there were only a few routines in the library, I'd offer to attempt to make a Linux version, but I expect the number of routines exceeds my ambition (quite small). I hate to tell you to run out and buy Windows. I hate to tell you to drop the course. I hate to tell you... you're between a rock and a hard place. If WINE will work with this stuff, that might be your salvation. Otherwise... not easy. (but perhaps doable if you're determined enough) Best, Frank
|
Pages: 1 Prev: HAY Herbert, wanna write a legal PE executable ? Next: Philips Chip P87LPC764 |