IRQ-Tables once again

[Eric W. Biederman]

Stefan Reinauer <stepan at suse.de> writes:

> * Li-Ta Lo <ollie at lanl.gov> [040122 17:22]:
> > > good. things seem to work now. Except that the secondary bus of the 
> > > 8151 gets set to 0, where it seems that in our case this should be 2
> > > (the AGP device is visible on bus 2) Don't think this matters though
>  
> > Did you just dump the IRQ table from normal BIOS and put it in
> > LinuxBIOS ?
> 
> > That is the same problem we found here. The normal BIOS tries to make
> > the secondary bus of 8151 as bus 0. It also tries to hide the bridge
> > from IRQ and/or MP table.
> 
> My assumption that the secondary bus is 0 on the 8151 comes from reading
> the SECONDARY_BUS register while creating the mptable. 

Hmm.  That sounds like a bug.
 
> This happens before the irq table is used for the first time.
> 
> Some  pirq tables I've looked at showed the complete device tree of 
> the machine, including the 2 cpus and all the bridges. We could probably
> generate a lot of this information automatically already.
> For getting the slots described right we need an additional slot
> command/tag in the config mechanism. 
> Then we could just walk over all north/southbridges and generate both
> an irq table and an mptable from the device tree.

That is one of my informal requirements for a stable freebios 2.0 release.
To be able to generate the mptable and irq tables from the device tree.

Eric