[PATCH RESEND] PCI: mvebu: Dynamically detect if the PEX link is up to enable hot plug

[Bjorn Helgaas]

On Mon, Oct 7, 2013 at 4:40 PM, Jason Gunthorpe
<jgunthorpe at obsidianresearch.com> wrote:
> On Mon, Oct 07, 2013 at 04:28:52PM -0600, Bjorn Helgaas wrote:
>> [+cc Seungwon]
>>
>> On Tue, Oct 1, 2013 at 11:58 AM, Jason Gunthorpe
>> <jgunthorpe at obsidianresearch.com> wrote:
>> > Otherwise hotplugging the PEX doesn't work at all since the driver
>> > detects the link state at probe time. Simply replacing the two tests
>> > of haslink with a register read is enough to fix it.
>> >
>> > Tested on kirkwood with repeated plug/unplug of the link partner.
>> >
>> > Signed-off-by: Jason Gunthorpe <jgunthorpe at obsidianresearch.com>
>>
>> Last I heard (Sep 26), you had some mvebu changes in your tree
>> already, Jason, so I was assuming you'd handle pci-mvebu.c changes, at
>> least for the v3.13 merge window.  But you marked this "resend"; does
>> that mean you're waiting for me to do something with it?
>
> Are you thinking of Jason C.? He is handling much of the mvebu stuff.

Oops, sorry, yep, indeed I meant Jason C.  Sorry about confusing you guys :)

> Hopefully Jason C. can clarify which route these patches should take
> to hit the 3.13 merge window.
>
> This was reposted at the request of Thomas, IIRC.
>
>>   PCI: mvebu - Support a bridge with no IO port window
>
> This one is broken, I have yet to fix it still.
>
>>   PCI: mvebu: add I/O access wrappers
>
> This isn't one of mine ..

That's Seungwon's.  I just wanted to collect all the mvebu patches
from the PCI patchwork.

Bjorn