[PATCH] SB600 for the Nemo board has non-zero devices on non-root bus

On 04 December 2017 at 12:40PM, Darren Stevens wrote:

Hello Bjorn

Firstly sorry for not being able to join in this discussion, I have been
moving house and only got my X1000 set up again yesterday..

On 30/11/2017, Bjorn Helgaas wrote:

quoted

I *think* something like the patch below should make this work if you
use the "pci=pcie_scan_all" parameter.  We have some x86 DMI quirks
that set PCI_SCAN_ALL_PCIE_DEVS automatically.  I don't know how to do
something similar on powerpc, but maybe you do?

Actually the root ports on the Nemo's PA6T processor don't respond to the
SB600 unless we turn on a special 'relax pci-e' bit in one of its control
registers. We use a small out of tree init routine to do this, and there
would be the ideal place to put a call to
pci_set_flag(PCI_SCAN_ALL_PCIE_DEVS).

This patch fixes the last major hurdle to getting the X1000 fully supported in
the linux kernel, so thanks very much for that.

Regards
Darren

On 15 December 2017 at 09:25PM, Bjorn Helgaas wrote:

On Fri, Dec 15, 2017 at 09:04:51AM +0100, Christian Zigotzky wrote:

quoted

On 09 December 2017 at 7:03PM, Christian Zigotzky wrote:

quoted

On 08 December 2017 at 12:59PM, Michael Ellerman wrote:

quoted

quoted

Darren's idea of doing it at the same time you tweak the SB600 "relax
pci-e" bit is ideal because then the two pieces are obviously
connected and it wouldn't affect any other systems at all.

Yes that would be ideal. That patch is currently out-of-tree I gather,
but I guess everyone who's using these machines must have that patch
anyway.

Darren what does that code look like? Can we get it upstream and close
the loop on this?

cheers

Hi Michael,

Please find attached the code.

Thanks,
Christian

Hi All,

I haven't received any response yet. Is this the correct patch you
are looking for?

This is a powerpc patch that doesn't affect the PCI core, so I would
say this is Michael's bailiwick.

I guess you're only looking for a hint about whether this is the right
approach, because it's obviously fully baked yet (no changelog,
signed-off-by, etc, not a "safe for all powerpc" run-time solution,
not in Linux indentation style, etc).

It looks like the "pasemi,1682m-iob" DT property is required and
possibly sufficient to identify this system at run-time.

My advice is to finish that stuff up, post it to the powerpc
maintainers and the linuxppc-dev@lists.ozlabs.org list, and go from
there.

Darren,

Where is this small out of tree init routine in our patch? I haven't found it yet. Please post this routine here. Please find attached our latest Nemo patch.

Thanks,
Christian