nanogui: Thread: Playing with GrOpenClientFramebuffer

Hello,

I'm looking for porting some framebuffer app I've done to nanogui.

I've a PDA, with a 4 bits per pixel framebuffer.

my prg holds a false frame buffer.
I want to copy this false fb to the real fb at the right place:

void FlushFB()
{
  GR_WINDOW_FB_INFO fbinfo;
  GR_WINDOW_INFO winfo;
  char *d,*s;
  GrGetWindowFBInfo(mainwid,&fbinfo );
  GrGetWindowInfo(mainwid,&winfo );
  d=(char *)fbinfo.winpixels;
  s=fb;
  printf("x = %d y = %d pitch = %d \n",winfo.x,winfo.y,fbinfo.pitch);
}
The printf return the same line even if I move the window.
x = 3 y = 16 pitch = 320

All is false.. x,y and pitch, that should be 320/2 (4bpp)

here is the initialisation stuff:


void initfb()
{
  if (GrOpen() < 0) {
    fprintf(stderr, "cannot open graphics\n");
    exit(1);
  }
  fb2=(char *)GrOpenClientFramebuffer();

  if(!fb2)
    printf("cannot share framebuffer!!\n");

  mainwid = GrNewWindowEx(GR_WM_PROPS_APPFRAME |GR_WM_PROPS_APPWINDOW | 
GR_WM_PROPS_CAPTION |GR_WM_PROPS_CLOSEBOX,
			  (GR_CHAR *)"vt100em",
			  GR_ROOT_WINDOW_ID, 0, 0, 320, 220,
			 WHITE);

  wgc = GrNewGC();
  GrSetGCBackground(wgc, WHITE);
  GrSetGCForeground(wgc, BLACK);
  GrSetGCUseBackground(wgc, GR_TRUE);

  GrSelectEvents(mainwid, GR_EVENT_MASK_EXPOSURE |
		 GR_EVENT_MASK_BUTTON_DOWN |GR_EVENT_MASK_BUTTON_UP | 
GR_EVENT_MASK_KEY_DOWN|GR_EVENT_MASK_CLOSE_REQ);

  fb = (char *) malloc(rb * 240); // the buffer is volontary too big..

/*
  bm = (GR_BITMAP *) malloc((rb/2) * 240); // old idea, of course very slow
*/
  GrMapWindow(mainwid);

  GrReqShmCmds(655350);  // is it usefull??


/*
  framebuf_fd = open ("/dev/fb0", O_RDWR);
  fb = (char *)mmap(NULL, rb * 240, PROT_READ | PROT_WRITE,
                          MAP_SHARED, framebuf_fd, 0);
*/
}

when I look at  nanox/clientfb.c:
    159         fbinfo->pitch = fbinfo->xres * fbinfo->bytespp;
    164         /* winpixels only valid for non-portrait modes*/
    165         physoffset = info.y*fbinfo->pitch + info.x*fbinfo->bytespp;
    166         fbinfo->winpixels = physpixels? (physpixels + physoffset): 
NULL;

of course pitch is false, bytespp is theorically 0.5..

Is it a bug?

How should I do,then ?


Pierre

_________________________________________________________________
Téléchargez MSN Explorer gratuitement à l'adresse 
http://explorer.msn.fr/intl.asp.

> when I look at  nanox/clientfb.c:
>     159         fbinfo->pitch = fbinfo->xres * fbinfo->bytespp;
>     164         /* winpixels only valid for non-portrait modes*/
>     165         physoffset = info.y*fbinfo->pitch +
info.x*fbinfo->bytespp;
>     166         fbinfo->winpixels = physpixels? (physpixels + physoffset):
> NULL;
>
> of course pitch is false, bytespp is theorically 0.5..
>
> Is it a bug?

Yes, it looks like line 159 is written to only work with
8bpp or greater.  Try changing line 159 to something like

fpinfo->pitch = (fpinfo->xres+1)* 2;

and then things should work.  If this works, I'll change the
line 159 to something more portable, that calcs pitch properly
for all BPP ranges.

Regards,

Greg