Re: [PATCH 1/2] PCI/PM: Revert "PCI/PM: Drop unused runtime PM support code for PCIe ports"

From: Rafael J. Wysocki
Date: Tue Jan 19 2016 - 17:09:47 EST

On Tuesday, January 19, 2016 03:52:34 PM Bjorn Helgaas wrote:
> [+cc Rafael, author of fe9a743a2601]
> On Fri, Jan 15, 2016 at 10:27:44PM +0800, Qipeng Zha wrote:
> > This reverts commit fe9a743a2601 ("PCI/PM: Drop unused runtime
> > PM support code for PCIe ports"). To support PCIe ports
> > runtime PM for Intel Broxton platform.
> This changelog is not as useful as it could be. It's true that this
> reverts fe9a743a2601, but what's important is that this adds support
> code for runtime PM of PCIe ports, so that should be mentioned first.
> The fe9a743a2601 changelog says "If we are to support runtime PM of
> PCIe ports, it will have to be done in a different way most likely
> anyway." I don't know what Rafael had in mind there, but we certainly
> need to ask him before adding it back the same way it was.


The previous approach had a few problems that showed up in testing and
that's why the code was first disabled and then removed from the kernel.

We can't re-enable it or those problems will show up again.

I have not looked at adding PCIe port runtime PM support in a cleaner
way yet, but it is on my radar.