[ros-diffs] [greatlrd] 33783: revert my last change

[James Tabor]

Hi!
I checked this and it is all wrong,,,,, I reverted part of revision 19267:

Replacing the code with the original wine code that behaves like
"Windows code" but should be called inside EngCreateBitmap if (pvBits)
Width = DIB_GetDIBWidthBytes( lWidth, iFormat); .

As it should be:

INT FASTCALL DIB_GetDIBWidthBytes (INT width, INT depth)
{
//  return ((width * depth + 31) & ~31) >> 3;
  int words;

switch(depth)
  {
     case 1:  words = (width + 31) / 32; break;
     case 4:  words = (width + 7) / 8; break;
     case 8:  words = (width + 3) / 4; break;
     case 15:
     case 16: words = (width + 1) / 2; break;
     case 24: words = (width * 3 + 3)/4; break;

default:
        DPRINT1("(%d): Unsupported depth\n", depth );
    /* fall through */
    case 32:
        words = width;
  }
  return 4 * words;
}

[James Tabor]

OMG!

eng/surface.c:149:    NewBitmap = EngCreateBitmap(Size,
DIB_GetDIBWidthBytes(Size.cx, BitsPerFormat(Format)), Format, 0,
NULL);

Are you all aware that DIB_GetDIBWidthBytes is called again in
IntCreateBitmap? Follow the flow.

No wonder bitmaps is fubared!

Please someone fix this!
James

On Sat, May 31, 2008 at 1:41 PM, James Tabor [e-mail] wrote:
> Hi!
> I checked this and it is all wrong,,,,, I reverted part of revision 19267:
>
> Replacing the code with the original wine code that behaves like
> "Windows code" but should be called inside EngCreateBitmap if (pvBits)
> Width = DIB_GetDIBWidthBytes( lWidth, iFormat); .
>
> As it should be:
>
> INT FASTCALL DIB_GetDIBWidthBytes (INT width, INT depth)
> {
> //  return ((width * depth + 31) & ~31) >> 3;
>  int words;
>
> switch(depth)
>  {
>     case 1:  words = (width + 31) / 32; break;
>     case 4:  words = (width + 7) / 8; break;
>     case 8:  words = (width + 3) / 4; break;
>     case 15:
>     case 16: words = (width + 1) / 2; break;
>     case 24: words = (width * 3 + 3)/4; break;
> default:
>    DPRINT1("(%d): Unsupported depth\n", depth );
>
> /* fall through */
>    case 32:
>        words = width;
>  }
>  return 4 * words;
> }

[Aleksey Bragin]

And even more, James, there are live real cases during reactos start  
up process that DIB_GetDIBWidthBytes is called with negative width.

WBR,
Aleksey Bragin.

On Jun 1, 2008, at 1:31 AM, James Tabor wrote:
> OMG!
>
> eng/surface.c:149:    NewBitmap = EngCreateBitmap(Size,
> DIB_GetDIBWidthBytes(Size.cx, BitsPerFormat(Format)), Format, 0,
> NULL);
>
> Are you all aware that DIB_GetDIBWidthBytes is called again in
> IntCreateBitmap? Follow the flow.
>
> No wonder bitmaps is fubared!
>
> Please someone fix this!
> James
>
> On Sat, May 31, 2008 at 1:41 PM, James Tabor  
> [e-mail] wrote:
> > Hi!
> > I checked this and it is all wrong,,,,, I reverted part of  
> > revision 19267:
> >
> > Replacing the code with the original wine code that behaves like
> > "Windows code" but should be called inside EngCreateBitmap if  
> > (pvBits)
> > Width = DIB_GetDIBWidthBytes( lWidth, iFormat); .
> >
> > As it should be:
> >
> > INT FASTCALL DIB_GetDIBWidthBytes (INT width, INT depth)
> > {
> > //  return ((width * depth + 31) & ~31) >> 3;
> >  int words;
> >
> > switch(depth)
> >  {
> >     case 1:  words = (width + 31) / 32; break;
> >     case 4:  words = (width + 7) / 8; break;
> >     case 8:  words = (width + 3) / 4; break;
> >     case 15:
> >     case 16: words = (width + 1) / 2; break;
> >     case 24: words = (width * 3 + 3)/4; break;
> > default:
> >    DPRINT1("(%d): Unsupported depth\n", depth );
> >
> > /* fall through */
> >    case 32:
> >        words = width;
> >  }
> >  return 4 * words;
> > }
>
> Ros-dev mailing list
> [e-mail]
> http://www.reactos.org/mailman/listinfo/ros-dev