On 09/23/2014 08:41 PM, Bjorn Helgaas wrote:

On Tue, Sep 23, 2014 at 7:18 PM, Liviu Dudau [off-list ref] wrote:

quoted

On Tue, Sep 23, 2014 at 08:01:12PM +0100, Liviu Dudau wrote:

quoted

If the firmware has not assigned all the bus resources and
we are not just probing the PCIe busses, it makes sense to
assign the unassigned resources in pci_scan_root_bus().

Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Jason Gunthorpe <redacted>
Cc: Rob Herring <robh+dt@kernel.org>
Signed-off-by: Liviu Dudau <Liviu.Dudau@arm.com>
---
  drivers/pci/probe.c | 3 +++
  1 file changed, 3 insertions(+)

diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
index ef891d2..508cf61 100644
--- a/drivers/pci/probe.c
+++ b/drivers/pci/probe.c

@@ -1953,6 +1953,9 @@ struct pci_bus *pci_scan_root_bus(struct device *parent, int bus,
       if (!found)
               pci_bus_update_busn_res_end(b, max);

+     if (!pci_has_flag(PCI_PROBE_ONLY))
+             pci_assign_unassigned_bus_resources(b);
+
       pci_bus_add_devices(b);
       return b;
  }
--

2.1.0

Bjorn,

If you are OK with this patch, can you let me know how do you feel about
making pci_scan_root_bus() set up the bus->msi pointer as well?

I'm not opposed to it, but I have the same question as for setting up
the domain: how does pci_scan_root_bus() learn what to assign to
bus->msi?  It currently only gets a "void *sysdata" so there's no
obvious place to put it there.  You could add a pcibios interface to
retrieve it, I suppose, but I'm starting to get uncomfortable with
adding more of those because we have such a mess of them already.

Bjorn

[Suravee] Liviu and I had a talk during Linaro Connect, and we came up
with a new binding for the pcie controller called "msi-parent" which is
supposed to contain phandle to the corresponded msi-controller.

quoted

Side note: I don't know exactly how to do it now, but setting bus->msi is
needed by platforms that are based on my patches and want to use MSI,
otherwise they have to open code pci_scan_root_bus() to set it. I haven't
made any attempts to do it in this series as the GICv2m patches are not
ready yet, but I can see the need arriving soon.

[Suravee] I just sent out an RFC which implements the described binding 
here.

https://lkml.org/lkml/2014/9/28/149

Any feedback/comments are welcome :)

Thanks,

Suravee