Re: [patch 2.6] remove PCI_BRIDGE_CTL_VGA handling from setup-bus.c

[Ivan Kokshaysky <ink@jurassic.park.msu.ru>]

On Thu, Jul 14, 2005 at 02:53:27PM +0100, Russell King wrote:
> What happens when there is no firmware?

It shouldn't be a problem. These days we have a lot of arch hooks
in the PCI layer. I'd probably start with the following:

static void __init
pcibios_enable_p2p_vga_fwd(struct pci_dev *dev)
{
	u16 class = dev->class >> 8;
	struct pci_bus *b;

	if (class == PCI_CLASS_DISPLAY_VGA ||
	    class == PCI_CLASS_NOT_DEFINED_VGA) {
		for (b = dev->bus; b->parent; b = b->parent) {
			b->bridge_ctl |= PCI_BRIDGE_CTL_VGA;
		}
	}
}

DECLARE_PCI_FIXUP_HEADER(PCI_ANY_ID, PCI_ANY_ID, pcibios_enable_p2p_vga_fwd);

> I'm sure this code would not have been added had there not been a reason
> for it.  Do we know why it was added?

It was quite a few years ago, so I don't recall all details...
I think some old alphas cannot cope with VGA behind the p2p bridges.
But I also recall that this code was problematic on alpha with multiple
VGA cards.
Note that setup-bus code in 2.4 does more reasonable thing - it
enables VGA port forwarding *only* for the first (by PCI enumeration)
VGA device, and doesn't enable this for others (if any). Unlike 2.6 :-(

Ivan.