Re: offset x screen
On Thu, Aug 16, 2001 at 04:14:51PM +0200, Michel Dänzer wrote:
> Bruce McIntyre wrote:
> >
> > On Thu, Aug 16, 2001 at 07:00:05PM +1000, Bruce McIntyre wrote:
> > > On Thu, Aug 16, 2001 at 02:13:37AM +0200, Michel Dänzer wrote:
> > >
> > > > I have hopefully found the bug in the fbdev driver now, please test
> > > >
> > > > http://penguinppc.org/~daenzer/patches/fbdev-fboffset.diff
> > > >
> > > > or
> > > >
> > > > http://penguinppc.org/~daenzer/fbdev_drv.o
> > > >
> > > > which is the 4.1.0 fbdev driver built with the patch applied. If it
> > > > works, I'll submit it to Branden and upstream.
> > >
> > > I tried your patched fbdev_drv.o in /usr/X11R6/lib/modules/drivers.
[snip]
> > There is a slight change... the virtual screen does not scroll; instead it
> > wraps around underneath the 20 pixel margin and further... (hard to
> > describe.. I'll try a picture.
>
> That used to work before?
>
> > "ioctl FBIOPAN_DISPLAY: Invalid argument" was repeated many times in
> > the console.
>
> Sounds like your framebuffer device doesn't support panning in fact.
There used to be some panning vertically, not a lot and I'm afraid I cannot
remember at what resolutions. Panning still works well vertically.
[snip]
> > There is now also quite severe barrel distortion.
> > <- - - - - - - - - - - - - - 800 - - - - - - - - - - - - - - ->
> >
> > <---> Around 20 pixels
> >
> > +---------------------------------------------------------------+
> > | | | |
> > | | |
> >
> > | |<- screen starts |<- virtual screen ends.
> > and ends
> >
> > <---> this portion, as well as the overlap part of the rest of the
> > <-overlap-----------> screen are both displaced downwards 1
> > of virtual pixel.
> > screen.
>
> Thanks for trying to describe what you see, but I'm afraid I don't get the
> picture.
(Scratches his head) Ok, I'll do a series of pictures
Fig.2 ( \ is mouse )
(a)
+---------------------------------------------------------+
| \
| Here the mouse is at the top left corner. This
| is where the console window starts under my set-
up.
(b)
+---------------------------------------------------------+
|\
| This is where the mouse ends up if you scroll it
| to the right.... It is displaced down one pixel
| but can still move further right... see below
| Everything from this point on is down one pixel.
|
| \ The mouse gets to this horiz position before
overlap kicks in
(c)
+----------------------------------------------------------+
| \
| Here, I have continued moving the mouse to the right.
| Screen redraws go over the top of the cosole window
placed to the far left. You can move a window here
and if you move the mouse to the far left redraws
will, in turn, go over the top of it. This is the
overlap region of the previous figure.
At the moment I have tried depths of 8 and 16.
At a depth of 8, the starting position of (a) and the finishing of (b)
are further right.
It refuses to go into a normal 1024x768 mode. Instead it goes to a reso-
lution which resembles 800x600. 224 pixels are wrapped instead of pan-
ned horizontally and pans are normal for 168 pixels vertically. When the
modeline has only 800x600 and then 640x480, the overlap does not occur.
The mouse cannot go beyond those limits prescribed by (a) and (b)
> You are using XFree86 4.1.0, aren't you?
4.1.0-2
Reply to: