Re: [PATCH][4/5] RapidIO support: ppc32

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




On Jun 8, 2005, at 12:52 AM, Matt Porter wrote:

On Tue, Jun 07, 2005 at 11:43:26PM -0500, Kumar Gala wrote:
Two questions:
1. how well does will all of this handle > 32-bit phys addr?

It does and it doesn't handle > 32-bit phys addr. It depends on which
configuration you are talking about.  If you are talking about I/O
32-bit, it's no problem.  If you are talking about handling DMA
32-bit,
then we need to handle a 64-bit DMA addr in the ppc32 implementations
and
also extend the arch messaging interface to let callers know when an
implementation can handle high DMA (system memory >4GB). This is all
pretty easy to handle once we need to support such a processor. So
far, nothing is available publicly. :)

Well 8548 is semi-public :)

For RIO MMIO purposes (which is functionality I'm working on now),
it has the similar issues that PCI memory space has on processors with
I/O above 4GB.  However, on RIO our resources hold a bus address since
a physical address doesn't make sense since address spaces our
per-device.
If we ever support a 66-bit address space device on 32-bit processor, we
might need a u64 resource.

I assume you mean 36-bit, what would one do with 66-bit addresses :)

2. can we make any of this a platform driver?

Hrm, so you would rather see RIO host bridges look like a driver
on another "bus"?  I have seen them as a component just like PCI
host bridges.  That is, they are instantiated by arch-code and
then initialized by a subsys initcall. This does mean that we
will be enumerating much later (during driver initcalls), but
it might be a better model if we ever see a rumored PCIE->RIO
bridge. Supporting that as a RIO master port would require driver
time init of the RIO fabric. There's some ordering issues that we'd
have to see about working out. None of this is needed right now,
though.

I would prefer if we could have the memory offsets and irq's not be
straight from the #define's

I think this and #2 are separate issues. We can pass the mpc85xx
rio init code some parameters to abstract things to different
parts. This is similar to how we init different SoC's PCI host
bridges with some common code on PPC32 (marvell, 85xx, etc).

I was just looking at doing this to support RIO on the 8548. At
the time I wrote this 85xx support there wasn't any info on the
8548 available, but it's an easy thing to extend.

Agreed, they are separate issues, I'm cool on waiting to see what happens with RIO <-> PCIE bridges in the future. For now if you can look at abstracting the offset, irq info that would be good (especially since 8548 does msg'g a bit differently).

- kumar

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[Index of Archives]     [Kernel Newbies]     [Netfilter]     [Bugtraq]     [Photo]     [Stuff]     [Gimp]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Video 4 Linux]     [Linux for the blind]     [Linux Resources]
  Powered by Linux