RFC: [MIPS] BCM1480HT set mips_io_port_base

RFC: [MIPS] BCM1480HT set mips_io_port_base

[Bjorn Helgaas]

I don't see anywhere that BCM1480HT sets mips_io_port_base (but maybe
I missed it).  We *do* set bcm1480ht_controller.io_map_base, so pci_iomap()
should work, but without mips_io_port_base, I don't think inb() et al.
will work.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>

diff --git a/arch/mips/pci/pci-bcm1480ht.c b/arch/mips/pci/pci-bcm1480ht.c
index 0fd0222..021e959 100644
--- a/arch/mips/pci/pci-bcm1480ht.c
+++ b/arch/mips/pci/pci-bcm1480ht.c
@@ -205,6 +205,7 @@ static int __init bcm1480ht_pcibios_init(void)
 			4 * 1024 * 1024);
 	bcm1480ht_controller.io_map_base = (unsigned long)
 		ioremap(A_BCM1480_PHYS_HT_IO_MATCH_BYTES, 65536);
+	set_io_port_base(bcm1480ht_controller.io_map_base);
 
 	register_pci_controller(&bcm1480ht_controller);
 

Re: RFC: [MIPS] BCM1480HT set mips_io_port_base

[Thomas Bogendoerfer]

On Wed, Feb 24, 2010 at 01:47:07PM -0700, Bjorn Helgaas wrote:
> I don't see anywhere that BCM1480HT sets mips_io_port_base (but maybe
> I missed it).  We *do* set bcm1480ht_controller.io_map_base, so pci_iomap()
> should work, but without mips_io_port_base, I don't think inb() et al.
> will work.

there are two hoses on the BCM1480 one is a via the PCI host controller
the other one via the hypertransport. set_io_port_base() is an old hack
to get inX/outX drivers working. Drivers using in/out will only work for
devices on the PCI host controller and will be broken for devies on the
HT side. Your change will make the devices on the HT side working, but
break the one on the PCI bus. Either way isn't too nice, but at least
nobody had a problem with the current behaviour. So it might be better
to not apply your patch.

Thomas.

-- 
Crap can work. Given enough thrust pigs will fly, but it's not necessary a
good idea.                                                [ RFC1925, 2.3 ]