Re: IBM GXT6000P on parisc linux

On Sun September 30 2007 15:22, Berthold Gunreben wrote:
> Am Sonntag, 30. September 2007 21:35 schrieb Grant Grundler:
> > On Fri, Sep 21, 2007 at 07:49:21PM +0200, Berthold Gunreben wrote:
> > > Hi,
> > >
> > > I got a used HP parisc workstation, and found a GXT6000P video card in
> > > there. The card was detected and initialized by the firmware of this
> > > machine (a C3750).
> >
> > wow..that's odd. I didn't see any web page which shows HP ever
> > supported GXT6000P with HPUX. I suspect someone sold you the wrong card.
> > Can you explain in more detail what you mean with "detected and initialized
> > by the firmware"?
> >
> > I suspect magic BIOS code isn't getting run. (It _might_ run under
> > a C8000 but I don't expect it to work).
> 
> Well, what can I say. The workstation boots up and displays the usual firmware 
> messages like press Tab to choose resolution (only 640x480 seems to be 
> supported at that stage) and I can also enter commands at this console. The 
> card does not seem to be wrong in this respect.
>

Most likely, all the card hardware required to do 640x480 was a power-on
reset signal.

That rate is very non-hp - it is a rare parisc system monitor that can 
display at that scan.
 
> After booting linux, the messages of the kernel are displayed until the 
> console switches:
> 
> console handover: boot [ttyB0] -> real [tty0]
> 
> Then it takes some more seconds, until the display goes dark and I was not yet 
> successful in bringing it to life again.
>

Sounds like the card never got setup for anything that wasn't hardware defaulted
by the power-on signal.

If I recall correctly, that means (from non-parisc, vga mode land) that the video
interrupt generation is also disabled on the card.
 
> > You can determine if it's necessary by looking at the driver source code.
> > Try to find where it registers the IRQ handler with request_irq().
> > Sounds like either it doesn't or it can't (but should print something
> > to dmesg buffer in that case).
> 
> The driver has no occurence of irq, and it only handles framebuffer (normally, 
> this card also should have 3D capabilities. It looks like it just does not 
> need IRQs.
> 

Perhaps the driver was written to use _only_ the power-on defaults 
with a lot more work required to do anything else.  
Like the other 90% of the driver.

Mike

> It looks like the card is switched off by using 
> pdc_cons.flags &= ~CON_PRINTBUFFER
> 
> At least if I comment this line, booting continues with messages until some 
> point where the machine just hangs. 
> 
> I am sorry, that I did not post more information in the first place, but I 
> don't have much experience with graphics on parisc-linux ... 
> 
> Thanks for your efforts anyway.
> 
> Berthold
> _______________________________________________
> parisc-linux mailing list
> [email protected]
> http://lists.parisc-linux.org/mailman/listinfo/parisc-linux
> 
>