On Tue, Sep 23, 2014 at 03:06:35PM +0100, Russell King - ARM Linux wrote:
> On Tue, Sep 23, 2014 at 07:56:01AM -0600, Bjorn Helgaas wrote:
> > This doesn't require any coordination with the PCI core, so I was just
> > leaving this up to the arch. But I guess I can at least give you my
> > opinion :)
>
> However, PCI core people have more knowledge of the issues here than I do.
>
> > > diff --git a/arch/arm/kernel/bios32.c b/arch/arm/kernel/bios32.c
> > > index 17a26c1..03c56ba 100644
> > > --- a/arch/arm/kernel/bios32.c
> > > +++ b/arch/arm/kernel/bios32.c
> > > @@ -290,6 +290,7 @@ void pcibios_fixup_bus(struct pci_bus *bus)
> > > {
> > > struct pci_dev *dev;
> > > u16 features = PCI_COMMAND_SERR | PCI_COMMAND_PARITY |
> > > PCI_COMMAND_FAST_BACK;
> > > + bool has_pcie_dev = false;
> > >
> > > /*
> > > * Walk the devices on this bus, working out what we can
> > > @@ -298,6 +299,8 @@ void pcibios_fixup_bus(struct pci_bus *bus)
> > > list_for_each_entry(dev, &bus->devices, bus_list) {
> > > u16 status;
> > >
> > > + if (!has_pcie_dev)
> > > + has_pcie_dev = pci_is_pcie(dev);
> > > pci_read_config_word(dev, PCI_STATUS, &status);
> > >
> > > /*
> > > @@ -354,9 +357,11 @@ void pcibios_fixup_bus(struct pci_bus *bus)
> > >
> > > /*
> > > * Report what we did for this bus
> > > + * (only if the bus doesn't have any PCIe devices)
> > > */
> > > - printk(KERN_INFO "PCI: bus%d: Fast back to back transfers
> > > %sabled\n",
> > > - bus->number, (features & PCI_COMMAND_FAST_BACK) ? "en" :
> > > "dis");
> > > + if (!has_pcie_dev)
> > > + pr_info("PCI: bus%d: Fast back to back transfers
> > > %sabled\n",
> > > + bus->number, (features & PCI_COMMAND_FAST_BACK) ?
> > > "en" : "dis");
> >
> > My first choice would be to just drop the printk altogether.
>
> It can be useful information.
>
> > If we want to keep the printk, it should be enough to test "bus->self
> > && !pci_is_pcie(bus->self)" to see whether Fast Back-to-Back can be
> > enabled.
>
> This is exactly the kind of issue that needs to be picked up by core
> PCI people.I agree. Wouldn't it be more useful to move this into the core? Disabling fast back-to-back transfers hardly seems like an ARM- specific "fixup" to me. Thierry
pgphI_uhs2GhX.pgp
Description: PGP signature
