Re: [PATCH] PCI: Power on bridges before scanning new devices

[Mika Westerberg <mika.westerberg@linux.intel.com>]

On Sat, May 28, 2016 at 02:21:11PM +0200, Rafael J. Wysocki wrote:
> On Thursday, May 26, 2016 11:16:43 AM Mika Westerberg wrote:
> > On Wed, May 25, 2016 at 10:45:48PM +0200, Rafael J. Wysocki wrote:
> > > On Wednesday, May 25, 2016 04:19:48 PM Mika Westerberg wrote:
> > > > On Wed, May 25, 2016 at 12:13:09AM +0300, Mika Westerberg wrote:
> > > > > On Tue, May 24, 2016 at 03:53:23PM +0300, Mika Westerberg wrote:
> > > > > > > I dropped "ACPI / hotplug / PCI: Runtime resume bridge before rescan"
> > > > > > > on the assumption that "PCI: Power on bridges before scanning new
> > > > > > > devices" is sufficient to cover both the ACPI and the generic PCi
> > > > > > > rescan cases, but I'd like some reassurance about that.
> > > > > > 
> > > > > > I agree with your reasoning that the patch should not be needed anymore.
> > > > > > However, I have the machine which needed that patch at home so I'm not
> > > > > > able to test it now. I'll do that later today when I get back home.
> > > > > 
> > > > > I tried now on my Lenovo Yoga 900 laptop and unfortunately "PCI: Power
> > > > > on bridges before scanning new devices" seems not to be enough. This
> > > > > machine has SD-card reader connected to one PCIe port and once I unload
> > > > > the sdhci-pci driver and enable runtime PM for the device, next system
> > > > > suspend/resume cycle loses the SD-card reader PCI device.
> > > > > 
> > > > > I will investigate more tomorrow -- it is getting late here.
> > > > 
> > > > I think I found reason for the issue.
> > > > 
> > > > When the laptop resumes it will send ACPI BUS_CHECK event for the two
> > > > PCIe root ports. This ends up in acpiphp_check_bridge() where it goes
> > > > through all slots in that bridge checking if the devices are still
> > > > present. This happens before we call pci_scan_bridge() for the bridge
> > > > itself.
> > > > 
> > > > Since the bridge is in D3 config space of the device behind it is not
> > > > available and we determine that the device is not there anymore.
> > > > 
> > > > It looks like we either need that ACPI hotplug patch or alternatively we
> > > > could add pm_runtime_get/put() in acpiphp_check_bridge().
> > > 
> > > Have you tried the latter?
> > 
> > Indeed I tried and it worked fine. I can make formal patch doing that
> > which then replaces the current ACPI hotplug patch, if that is the
> > preferred way.
> 
> It is somewhat cleaner, so I'd prefer it.

Okay, I'm going to submit an updated version of that patch then.