Re: matroxfb problems with dualhead G400

Re: matroxfb problems with dualhead G400

[Petr Vandrovec]

On  3 Sep 01 at 0:25, Benjamin Gilbert wrote:
> I've been trying to set up dualhead text console with matroxfb and a
> 16MB G400, with mixed success.  What I think I'm seeing is that the
> scrollback from the primary head, when it gets long enough (i.e., cat
> /boot/System.map), intrudes on the secondary head.  If the secondary
> head is set to the same resolution and bit depth as the primary, I

> Kernel is vanilla 2.4.9.  Let me know what other information you need
> from me.

You must boot your kernel with 'video=scrollback:0'. Otherwise your
kernel die sooner or later... JJ's scrollback code does not cope with
more than one visible console, so you must disable it if you have more
than one display in the box.
                                            Best regards,
                                                Petr Vandrovec
                                                vandrove@vc.cvut.cz
                                                

Re:Re: matroxfb problems with dualhead G400

[Ghozlane Toumi]

Monday, September 03, 2001 4:51 PM, You wwrote:
> On  3 Sep 01 at 0:25, Benjamin Gilbert wrote:
> > [ software scroll back kills on dualhead ]
>
> You must boot your kernel with 'video=scrollback:0'. Otherwise your
> kernel die sooner or later... JJ's scrollback code does not cope with
> more than one visible console, so you must disable it if you have more
> than one display in the box.
>                                                 Petr Vandrovec

could we somehow detect in register_framebuffer that whe're going
multihead and disable software scrolling ?

I didn't look at the code so i don't know if it's feasible , just a
suggestion ..

ghoz

Re: matroxfb problems with dualhead G400

[Benjamin Gilbert]

On Mon, Sep 03, 2001 at 02:51:44PM +0000, Petr Vandrovec (VANDROVE@vc.cvut.cz) wrote:
> You must boot your kernel with 'video=scrollback:0'. Otherwise your
> kernel die sooner or later... JJ's scrollback code does not cope with
> more than one visible console, so you must disable it if you have more
> than one display in the box.

That doesn't fix it.  Shift-PgUp no longer works (of course), and
it's not immediately oopsing on me, but I still have the same type of
intrusion onto the secondary display.

--Benjamin Gilbert

Re:Re: matroxfb problems with dualhead G400

[Petr Vandrovec]

On  3 Sep 01 at 15:15, Ghozlane Toumi wrote:
> > You must boot your kernel with 'video=scrollback:0'. Otherwise your
> > kernel die sooner or later... JJ's scrollback code does not cope with
> > more than one visible console, so you must disable it if you have more
> > than one display in the box.

> could we somehow detect in register_framebuffer that whe're going
> multihead and disable software scrolling ?
> 
> I didn't look at the code so i don't know if it's feasible , just a
> suggestion ..

Unfortunately once scrollback gets enabled (and allocates its scrollback
buffers, fills couple of pointers here and there), it is not trivial to 
disable it again. If someone feels for doing that, do that. But I think
that changing default scrollback value from 32768 to 0 is better - as
you'll still have scrollback if your hardware can do it.
                                            Best regards,
                                                Petr Vandrovec
                                                vandrove@vc.cvut.cz
                                                
                                    
                                            

Re:Re: matroxfb problems with dualhead G400

[James Simmons]

> > could we somehow detect in register_framebuffer that whe're going
> > multihead and disable software scrolling ?
> > 
> > I didn't look at the code so i don't know if it's feasible , just a
> > suggestion ..
> 
> Unfortunately once scrollback gets enabled (and allocates its scrollback
> buffers, fills couple of pointers here and there), it is not trivial to 
> disable it again. If someone feels for doing that, do that. But I think
> that changing default scrollback value from 32768 to 0 is better - as
> you'll still have scrollback if your hardware can do it.

I agree for 2.4 kernels the solution Petr stated is the best. It is not
easy to properly implement soft scrollback. Sometime ago I worked on a new
scrollback implementation that was more in the upper console layers so
other systems could take advanatge of it. It is not a easy problem to
solve. In the next following weeks I should start working on this again.