On Sun, 8 Jan 2012 14:23:29 -0800 Linus Torvalds wrote: > On Sun, Jan 8, 2012 at 2:13 PM, Rogério Brito wrote: > > > > It doesn't boot with a vanilla kernel. It only boots when I pass the options > > above *and* compile the kernel with the following patch applied: > > > > ,----[ do_not_size_subtractive_decoding_transparent_pci_to_pci_bridges.patch ] > > | diff --git a/drivers/pci/setup-bus.c b/drivers/pci/setup-bus.c > > | index 86b69f85..84543f5 100644 > > | --- a/drivers/pci/setup-bus.c > > | +++ b/drivers/pci/setup-bus.c > > | @@ -849,6 +849,10 @@ void __ref __pci_bus_size_bridges(struct pci_bus *bus, > > |               break; > > | > > |       case PCI_CLASS_BRIDGE_PCI: > > | +             /* don't size subtractive decoding (transparent) > > | +              * PCI-to-PCI bridges */ > > | +             if (bus->self->transparent) > > | +                     break; > > |               pci_bridge_check_ranges(bus); > > |               if (bus->self->is_hotplug_bridge) { > > |                       additional_io_size  = pci_hotplug_io_size; > > `---- > > Ahh. I'd forgotten about that particular PCI patch. > > That is definitely the right thing to do, and commit 8fa5913d54f3 > ("PCI: remove transparent bridge sizing") did exactly that, but then > we reverted it in commit 12c22d6ef299 because it caused some odd > problems for some people. > > I think we should try to re-do that "avoid sizing transparent bridges" > commit, because it really should make it much easier to do PCI > allocations under some very common situations (there's a *lot* of > common intel transparent PCI bridges). > > Jesse, Yinghai, Ram - should we try to just re-do that commit in this > merge window, and see how that goes? We can always revert it again if > it causes problems.. I don't remember what problems we hit, but if Ram and Yinghai are willing to take a look at them we should go ahead and try again. Thanks, -- Jesse Barnes, Intel Open Source Technology Center