On Tue, Aug 06, 2024 at 08:53:01AM +0200, Lukas Wunner wrote:
> On Mon, Aug 05, 2024 at 07:05:55PM +0530, Manivannan Sadhasivam wrote:
> > On Fri, Aug 02, 2024 at 12:13:31PM +0200, Lukas Wunner wrote:
> > > The PCI core cannot put devices into D3cold without help from the
> > > platform. Checking whether D3cold is possible (or allowed or
> > > whatever) thus requires asking platform support code via
> > > platform_pci_power_manageable(), platform_pci_choose_state() etc.
> > >
> > > I think patch [3/4] is a little confusing because it creates
> > > infrastructure to decide whether D3cold is supported (allowed?)
> > > but we already have that in the platform_pci_*() functions.
> > > So I'm not sure if patch [3/4] adds value. I think generally
> > > speaking if D3hot isn't possible (allowed?), D3cold is assumed
> > > to not be possible either.
> >
> > Why? D3Hot is useful for runtime PM and if the platform doesn't want to do
> > runtime PM, it can always skip D3Hot (not ideal though).
>
> AFAICS we always program the device to go to D3hot and the platform
> then cuts power, thereby putting it into D3cold. So D3hot is never
> skipped. See __pci_set_power_state():
>
> if (state == PCI_D3cold) {
> /*
> * To put the device in D3cold, put it into D3hot in the native
> * way, then put it into D3cold using platform ops.
> */
> error = pci_set_low_power_state(dev, PCI_D3hot, locked);
>
> if (pci_platform_power_transition(dev, PCI_D3cold))
> return error;
>
This is applicable only to pci_set_power_state(), but AFAIK PCIe spec doesn't
mandate switching to D3Hot for entering D3Cold. So the PCIe host controller
drivers (especically non-ACPI platforms) may just send PME_Turn_Off followed by
removing the slot power (which again is not controlled by pci_set_power_state())
as there are no non-ACPI related hooks as of now.
- Mani
--
மணிவண்ணன் சதாசிவம்
