Xref: news-dnh.mv.net comp.os.msdos.djgpp:405
Newsgroups: comp.os.msdos.djgpp
Path: news-dnh.mv.net!mv!news.sprintlink.net!howland.reston.ans.net!news.starnet.net!wupost!news.utdallas.edu!corpgate!bcarh189.bnr.ca!nott!cunews!freenet.carleton.ca!FreeNet.Carleton.CA!ao950
From: ao950 AT FreeNet DOT Carleton DOT CA (Paul Derbyshire)
Subject: Screen Scrolling Snafu (heeeeeeeelp)
Sender: ao950 AT freenet2 DOT carleton DOT ca (Paul Derbyshire)
Reply-To: ao950 AT FreeNet DOT Carleton DOT CA (Paul Derbyshire)
Organization: The National Capital FreeNet
Date: Fri, 16 Jun 1995 00:54:13 GMT
Lines: 115
To: djgpp AT sun DOT soe DOT clarkson DOT edu
Dj-Gateway: from newsgroup comp.os.msdos.djgpp



Yeah, it's me again. I must have an older, bug-ridden version or something...

I have source code for a very rapid, direct-memory-copying screen
scrolling routine. (As direct as it can be in protected mode..>!)
It works fine when the scroll involves copying the memory to a higher
address, but not when copying to a lower one. When copying to a lower one
it almost works, but the top one fifth or so of the screen behaves very
strangely.
The scroll uses a buffer; 64000 bytes are copied from screen into the
buffer at (buffer+1000+scroll) where scroll is dx+320*dy (320x200). Then
64000 bytes are copied from buffer+1000 to screen. If scroll is negative
the problem occurs. I have checked and rechecked all the pointer
operations and math, and I have concluded that the problem lies with the
compiler and how it handles the assignment operation that does the memory
copying.

The source code follows for a short program that scrolls a field of random
pixels and slowly changes the scroll, as though a camera were flying in a
circle over a landscape of random dots.

#include <graphics.h>
#include <dos.h>


#define SETMODE 0
#define VIDEO_INT 0x10
#define G320x200x256 19
#define TEXT_COLOR 3
#define PUTPIXEL(x,y,c) *(video+(x)+(y)*320)=(c)


struct the_screen {
  unsigned char pixel[320][240];
  unsigned char overflows[2000];
};


void hsetmode (char mode);
void scroll (int scrollstep);


struct the_screen *screen = (struct the_screen *)0xD0000000;
char *video = (char *)0xD0000000;
struct the_screen vbuffer;
struct the_screen *vbuf = &vbuffer;
char *vb2 = (char *)&vbuffer+1000;


void main (void) {
  int x,y;
  int x1,y1;
  int x2,y2;
  int i,j;
  int scrs;
  int dx[16]={-6,-5,-4,-3, 0, 3, 4, 5, 6, 5, 4, 3, 0,-3,-4,-5};
  int dy[16]={ 0,-3,-4,-5,-6,-5,-4,-3, 0, 3, 4, 5, 6, 5, 4, 3};
  unsigned char color;
  hsetmode (G320x200x256);
  for (x=0; x<320; x++) {
    for (y=0; y<200; y++) {
      color=rand()/128;
      PUTPIXEL(x,y,color);
    }
  }
  for (i=0; i<16; i++) {
    scrs=dy[i]*320+dx[i];
    y1=(dy[i]<0) ? 200+dy[i] : 0;
    y2=(dy[i]<0) ? 200 : dy[i];
    x1=(dx[i]<0) ? 320+dx[i] : 0;
    x2=(dx[i]<0) ? 320 : dx[i];
    for (j=0; j<40; j++) {
      scroll (scrs);
      for (x=x1; x<x2; x++) {
        for (y=0; y<200; y++) {
          color=rand()/128;
          PUTPIXEL(x,y,color);
        }
      }
      for (y=y1; y<y2; y++) {
        for (x=0; x<320; x++) {
          color=rand()/128;
          PUTPIXEL(x,y,color);
        }
      }
    }
  }
  for (j=0; j<10000; j++);
  hsetmode (TEXT_COLOR);
}

void hsetmode (char mode) {
  union REGS regs;
  regs.h.ah=SETMODE;
  regs.h.al=mode;
  int86 (VIDEO_INT, &regs, &regs);
}
  
void scroll (int scrollstep) {
  *(struct the_screen *)(vb2+scrollstep)=*screen;
  *screen=*(struct the_screen *)vb2;
}


If anyone can help with this inexplicable problem, could they e-mail the
solution?
Thanks in advance,
	PGD

--
    .*.  "Clouds are not spheres, mountains are not cones, coastlines are not
 -()  <  circles, and bark is not smooth, nor does lightning travel in a
    `*'  straight line."    ,------------------------------------------------
         -- B. Mandelbrot  |  Paul Derbyshire (PGD) ao950 AT freenet DOT carleton DOT ca