PCI bridge bases, IDE on the B&W G3

PCI bridge bases, IDE on the B&W G3

[Samuel Rydh]

Hi,

I recently installed an ACARD 6260 IDE card in my B&W G3.
While trying to get it working with the 2.4 kernel, I found
three problems:

- The B&W G3 has two PCI-buses. The second one is
accessed through a DEC,21154 controller. The expension
slots as well as the internal CMD646 IDE-controller is on
the second bus. The problem is that the bridge reports
zero in the both baseregisters (the new PCI resource
allocation scheme then fails miserably). This problem
is simply fixed by (arch/ppc/kernel/pci.c):

 void __init pcibios_fixup_bus(struct pci_bus *bus)
 {
+       pci_read_bridge_bases(bus);
+
        if ( ppc_md.pcibios_fixup_bus )
                ppc_md.pcibios_fixup_bus(bus);
 }

(i386 does it this way too).

- The pci card was never enabled. I solved this by calling
pci_enable_device(dev) by hand. But... shouldn't this really
be done automatically by the generic ide layer?

- If yaboot loads the kernel (or the yaboot config file?)
from the ACARD IDE-bus, then the card will not work (lost
interrupts). The same problem (lost interrupts) occurs if
one tries to boot from the built-in CMD646 controller.
Possibly, IDE devices are not left in a good state by the
firmware/yaboot on the B&W G3?

The ACARD card works great if the kernel is loaded from
a CMD-controlled device.


Cheers,

/Samuel



----------------------------------------------------------
 E-mail <samuel@ibrium.se>  WWW: <http://www.ibrium.se>
  Phone/fax: (home) +46 8 4418431, (work) +46 8 7908470
----------------------------------------------------------

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: PCI bridge bases, IDE on the B&W G3

[Tom Rini]

On Thu, Sep 21, 2000 at 09:44:47PM +0200, Samuel Rydh wrote:

> - The B&W G3 has two PCI-buses. The second one is
> accessed through a DEC,21154 controller. The expension
> slots as well as the internal CMD646 IDE-controller is on
> the second bus. The problem is that the bridge reports
> zero in the both baseregisters (the new PCI resource
> allocation scheme then fails miserably). This problem
> is simply fixed by (arch/ppc/kernel/pci.c):
>
>  void __init pcibios_fixup_bus(struct pci_bus *bus)
>  {
> +       pci_read_bridge_bases(bus);
> +
>         if ( ppc_md.pcibios_fixup_bus )
>                 ppc_md.pcibios_fixup_bus(bus);
>  }

Hmm, I wonder if this will fix all of thouse pci resource conflicts I see here.

> - The pci card was never enabled. I solved this by calling
> pci_enable_device(dev) by hand. But... shouldn't this really
> be done automatically by the generic ide layer?

Nope.  the driver for that card should be doing pci_enable_device(dev).

> - If yaboot loads the kernel (or the yaboot config file?)
> from the ACARD IDE-bus, then the card will not work (lost
> interrupts). The same problem (lost interrupts) occurs if
> one tries to boot from the built-in CMD646 controller.
> Possibly, IDE devices are not left in a good state by the
> firmware/yaboot on the B&W G3?

iirc, around 2.4.0-test5 or 6 there were some IDE updates and the only working
IDE driver for us right now is/was ide-pmac.  Ben knows more tho :)

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

2.4.0-test9 on B&W G3

[Kumar Gala]

I have also been seeing the same old 'lost interrupt' problem when trying
to boot a B&W G3 using the latest yaboot (0.8) on the latest pmac-devel
rsync kernel.  I was wondering if this was a know bug with the latest 2.4
tree kernel's.  I also booted the same kernel on a new G4/450 with no
problems.

Thanks

 - kumar gala


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: PCI bridge bases, IDE on the B&W G3

[Geert Uytterhoeven]

On Thu, 21 Sep 2000, Samuel Rydh wrote:
> - The B&W G3 has two PCI-buses. The second one is
> accessed through a DEC,21154 controller. The expension
> slots as well as the internal CMD646 IDE-controller is on
> the second bus. The problem is that the bridge reports
> zero in the both baseregisters (the new PCI resource
> allocation scheme then fails miserably). This problem
> is simply fixed by (arch/ppc/kernel/pci.c):
>
>  void __init pcibios_fixup_bus(struct pci_bus *bus)
>  {
> +       pci_read_bridge_bases(bus);
> +
>         if ( ppc_md.pcibios_fixup_bus )
>                 ppc_md.pcibios_fixup_bus(bus);
>  }
>
> (i386 does it this way too).

Aha! I think this will also solve the problems on the G4.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/