From:	Richard Nichols <tiberius AT mailhost DOT net>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Getting to the bottom of this
Date:	Sat, 10 Jan 1998 12:03:04 +1000
Organization:	None
Lines:	71
Message-ID:	<34B6D6D8.6B48@mailhost.net>
Reply-To:	tiberius AT mailhost DOT net
NNTP-Posting-Host:	138.77.56.32
Mime-Version:	1.0
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp

Hmm..  after my numerous posts on bugs and stuff I've decided that
it must be NASM that's screwing my library (or my use of NASM).
My library is a hulking beast like allegro (only about 45% of that
functionality yet though) which is making debugging difficult. I
would wager that I have numberous bugs which are causing weird errors.

Here is the way I am using nasm:

-I only preserve ds when I use it. I often clobber es and all offset
 regs
-I store ebp and mov ebp, esp to reference the function's args if
I've messed around on the stack I mov esp, ebp at the end before
popping ebp
-does using ebp with debug info on cause any worries?
-I have MIDAS sound system running in the background. Is this a
problem?

Here is a sample procedure to write to the LFB. I've gotten myself
a selector to the LFB. I know this function causes problems so
what's wrong???

;
ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ
; gfxSendToLFB : Sends the data pointed to to the lfb
;
; void gfxSendToLFB(UCHAR *data, int length);
;
ÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ
;_gfxSendToLFB:
;        push ebp
;        mov ebp, esp

;        push dword [_vidLFB_Seg]        ; es = LFB segment
;        pop es                          ;
;        cld                             ;
;        xor edi, edi                    ; edi = 0
;        mov esi, [ebp +  8]             ; esi = data
;        mov edx, [ebp + 12]             ; edx = count
;        mov ecx, edx                    ; ecx = count/4
;        shr ecx, 2                      ;
;        rep movsd                       ; copy data
;        mov ecx, edx                    ; ecx = count mod 4
;        and ecx, 3                      ;
;        rep movsb                       ; copy left over data
;
;        mov esp, ebp
;        pop ebp
;        ret

And another thing, I'm using NASM 0.97 and when I do [SECTION .text
ALIGN=4]
I get an error: standard coff format does not support section alignment
the examples from djppasm.doc all have this error. Has something changed
in a subsequent version of NASM?

hmm.. I know I've got a lot of problems there but I'm really starting to
get
annoyed with this...

-Rich

-- 
  .------------------------------+--------------------------------.
  | Richard Nichols              | Inspire Media (Demos/Games)    |
  | mailto:tiberius AT mailhost DOT net | mailto:InspireMedia AT post1 DOT com  |
  | http://tiberius.home.ml.org/ | http://inspire.home.ml.org/    |
  +------------------------------+--------------------------------+
  |               Organizer, coder for Inspire Media              |
  `---------------------------------------------------------------'
        "The life of the coder is that of the hermit, except
           the hermit can't make shadebobs." -- Leviathan

- Raw text -