Date: Tue, 19 Aug 1997 02:47:05 +0300 (EET DST)
From: Samuli Takala <tax AT cc DOT hut DOT fi>
To: Jesse Legg <snowcrash AT ecr DOT net>
cc: djgpp AT delorie DOT com
Subject: Re: inline asm putpixel routine using table
In-Reply-To: <33F8B04B.AEC7D9AF@ecr.net>
Message-ID: <Pine.OSF.3.96.970819023856.6408A-100000@alpha.hut.fi>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk

Hi Jesse,

On Mon, 18 Aug 1997, Jesse Legg wrote:
>     Hi, I've been working with writing my own graphics routines in DJGPP
> 
> using mostly inline asm. My only references are a couple of (old) Pascal
[snipped and rearranged a bit]

> Also, I initialize the 'int screen_y[199]' array like this:
> void calc_screen_y(int width)
> {
>  int i;
>  for (i =0; i < 200; i++)
>      screen_y[i] = i*width;
> }

That should be int screen_y[200], otherwise with y=199 you'll overrun the
array.

> void _setpixel(int x, int y, char col) {
>  __asm__ __volatile__ ("
>    movw %3, %%es
>    addl %0, %%edi
>    movl %1, %%ebx
>    addl %%ebx, %%ebx

Problem here is that we are dealing with a 32-bit compiler so the size of
'int' is 32-bits. So to get the right offset to the screen_y array, you need
to multiply y by 4, not by 2:
     shll $2, %%ebx

>    addl %%ds:_screen_y(%%ebx), %%edi
>    movb %%al, %%es:(%%edi)"
>    : /* no outputs */
>    : "g" (x), "g" (y), "a" (col), "g" (_dos_ds), "D" (0xA0000)
>    : "eax", "ebx", "edi");  }


Hope this helps,
Samppa

Samuli Takala    Samuli DOT Takala AT hut DOT fi     finger -l tax AT hut DOT fi for long version