On Fri, 12 Nov 2021 12:15:28 +0100 Gerd Hoffmann <kra...@redhat.com> wrote:
> On Thu, Nov 11, 2021 at 10:39:59AM -0500, Michael S. Tsirkin wrote: > > On Thu, Nov 11, 2021 at 01:09:05PM +0100, Gerd Hoffmann wrote: > > > Hi, > > > > > > > When the acpihp driver is used the linux kernel will just call the aml > > > > methods and I suspect the pci device will stay invisible then because > > > > nobody flips the slot power control bit (with native-hotplug=on, for > > > > native-hotplug=off this isn't a problem of course). > > > > > > Hmm, on a quick smoke test with both patch series (mine + igors) applied > > > everything seems to work fine on a quick glance. Dunno why. Maybe the > > > pcieport driver turns on slot power even in case pciehp is not active. > > Digged deeper. Updating power status is handled by the plug() callback, > which is never called in case acpi hotplug is active. The guest seems > to never touch slot power control either, so it's working fine. Still > feels a bit fragile though. > > > Well power and hotplug capabilities are mostly unrelated, right? > > At least they are separate slot capabilities. The linux pciehp driver > checks whenever the power control is present before using it, so having > PwrCtrl- HotPlug+ seems to be a valid combination. > > We even have an option for that: pcie-root-port.power_controller_present > > So flipping that to off in case apci hotplug is active should make sure > we never run into trouble with pci devices being powered off. > > Igor? Can you add that to your patch series? Sorry, saw it too late. I'll test idea with my set of guests to see if there are any adverse effects. > > I feel switching to native so late would be inappropriate, looks more > > like a feature than a bugfix. Given that - we need Igor's patches. > > Given that - would you say I should apply yours? > > I think when setting power_controller_present=off for acpi hotplug it is > safe to merge both mine and igor's. > > take care, > Gerd >
