On Thursday 27 February 2014 12:36:27 Jason Gunthorpe wrote:

On Thu, Feb 27, 2014 at 07:12:59PM +0000, Liviu Dudau wrote:

quoted

The outstanding issue is how to fix pci_address_to_pio() as it will not
for for range->cpu_addr > IO_SPACE_LIMIT (16MB in my case).

The default actually looks fine to me, it is the correct behavior for
systems that actually have a dedicated IO space (like x86) where the
'CPU' value for IO is the exact value used in the IO accessor
instructions. In this case the IO_SPACE_LIMIT test is appropriate.

Right.

It also looks correct for architectures that use the CPU MMIO address
as the IO address directly (where IO_SPACE_LIMIT would be 4G)

Are you aware of any that still do? I thought we had stopped doing
that.

Architectures that use the virtual IO window technique will always
require a custom pci_address_to_pio implementation.

Hmm, at the moment we only call it from of_address_to_resource(),
which in turn does not get called on PCI devices, and does not
call pci_address_to_pio for 'simple' platform devices. The only
case I can think of where it actually matters is when we have
ISA devices in DT that use an I/O port address in the reg property,
and that case hopefully won't happen on ARM32 or ARM64.

BTW, something that occured to me after reading the patches:

For ARM64 you might want to think about doing away with the fixed
virtual IO window like we see in ARM32. Just use the CPU MMIO address
directly within the kernel, and implement a ioport_map to setup the MM
on demand.

I think the legacy reasons for having all those layers of translation
are probably not applicable to ARM64, and it is much simpler without
the extra translation step....

Arnd, what do you think?

Either I don't like it or I misunderstand you ;-)

Most PCI drivers normally don't call ioport_map or pci_iomap, so
we can't just do it there. If you are thinking of calling ioport_map
for every PCI device that has an I/O BAR and storing the virtual
address in the pci_dev resource, I don't see what that gains us
in terms of complexity, and it will also break /dev/port.

	Arnd