On Mon, Dec 09, 2013 at 04:12:24AM +0000, Hanjun Guo wrote:

On 2013-12-7 1:23, Arnd Bergmann wrote:

quoted

On Friday 06 December 2013, Tomasz Nowicki wrote:

quoted

On 05.12.2013 23:04, Arnd Bergmann wrote:

quoted

On Wednesday 04 December 2013, Hanjun Guo wrote:

quoted

On 2013?12?04? 00:41, Matthew Garrett wrote:

quoted

Given the number of #ifdefs you're adding, wouldn't it make more sense
to just add stub functions to include/linux/pci.h?

Thanks for the suggestion :)

I can add stub functions in include/linux/pci.h for raw_pci_read()/
raw_pci_write(), then can remove #ifdefs for acpi_os_read/write_pci_configuration().

Actually I wonder about the usefulness of this patch in either form: Since ACPI
on ARM64 is only for servers, I would very much expect them to always come with
PCI, either physical host bridges with attached devices, or logical PCI functions
used to describe the on-SoC I/O devices. Even in case of virtual machines, you'd
normally use PCI as the method to communicate data about the virtio channels.

Can you name a realistic use-case where you'd want ACPI but not PCI?

Yes you can describe SoC I/O devices using logical PCI functions only if 
they are on PCI, correct me if I am wrong. Also, devices can be placed 
only on IOMEM (like for ARM SoC) and it is hard to predict which way 
vendors chose. So way don't let it be configurable? ACPI spec says 
nothing like PCI is needed for ACPI, AFAIK.

You are right that today's ARM SoCs basically never use PCI to describe
internal devices (IIRC VIA VT8500 is an exception, but their PCI was
just a software fabrication).

However, when we're talking about ACPI on ARM64, that is nothing like classic
ARM SoCs: As Jon Masters mentioned, this is about new server hardware following
a (still secret, but hopefully not much longer) hardware specification that is
explicitly designed to allow interoperability between vendors, so they
must have put some thought into how to make the hardware discoverable. It
seems that they are modeling things after how it's done on x86, and the
only sensible way to have discoverable hardware there is PCI. This is
also what all x86 SoCs do.

I think the concern here is that ACPI is only for server platform or not.

Since ACPI has lots of content related to power management, I think ACPI
can be used for mobile devices and other platform too, not only for ARM
servers, and with this patch, we can support both requirement.

'Can be used' is one thing, will it really be used is another? I don't
think so, it was (well, is) difficult enough to make the transition to
FDT, I don't see how ACPI would solve the current issues.

I see ACPI as a server distro requirement and there are indeed benefits
in abstracting the hardware behind standard description, AML. Of course,
this would work even better with probe-able buses like PCIe and I'm
pretty sure this would be the case on high-end servers. But even if a
server distro like RHEL supports a SoC without PCIe, I would expect them
to only provide a single binary Image with CONFIG_PCI enabled.

This patch is small enough and allows ACPI build with !CONFIG_PCI for
the time being but longer term I would expect such SoCs without PCI to
be able to run on a kernel with CONFIG_PCI enabled.

-- 
Catalin