Re: [PATCH 1/2] PCI: apple: Integrate pwrctrl API

[Marc Zyngier <maz@kernel.org>]

On Tue, 07 Jul 2026 13:04:21 +0100,
Yureka Lilian <yureka@cyberchaos.dev> wrote:
> 
> On 7/7/26 11:25, Marc Zyngier wrote:
> > On Mon, 06 Jul 2026 23:38:27 +0100,
> > Yureka Lilian <yureka@cyberchaos.dev> wrote:
> >> Integrate the PCI pwrctrl framework into the Apple PCIe host driver to
> >> provide standardized power management for PCI devices.
> >> 
> >> Notably, this allows enabling powering on the WiFi, SD card reader on
> >> various Macs by means of the pwrctrl framework before probing the ports.
> >> 
> >> Previously, a custom solution for powering on the WiFi and SD card
> >> reader was proposed[1], but we can now use the new pci-pwrctrl-generic
> >> driver for this purpose.
> >> 
> >> Link[1]: https://lore.kernel.org/lkml/20220502093832.32778-4-marcan@marcan.st/
> >> 
> > nit: this paragraph and the accompanying link don't belong in the
> > commit message and should be moved below the --- mark or even better,
> > to the cover letter.
> ack, will leave it out of the commit message of the individual commit in v2
> >> Signed-off-by: Yureka Lilian <yureka@cyberchaos.dev>
> >> ---
> >>   drivers/pci/controller/Kconfig      |  1 +
> >>   drivers/pci/controller/pcie-apple.c | 16 ++++++++++++++++
> >>   2 files changed, 17 insertions(+)
> >> 
> >> diff --git a/drivers/pci/controller/Kconfig b/drivers/pci/controller/Kconfig
> >> index 2247709ef6d6..af64630d28fa 100644
> >> --- a/drivers/pci/controller/Kconfig
> >> +++ b/drivers/pci/controller/Kconfig
> >> @@ -46,6 +46,7 @@ config PCIE_APPLE
> >>   	depends on OF
> >>   	depends on PCI_MSI
> >>   	select PCI_HOST_COMMON
> >> +	select PCI_PWRCTRL_GENERIC
> >>   	select IRQ_MSI_LIB
> >>   	help
> >>   	  Say Y here if you want to enable PCIe controller support on Apple
> >> diff --git a/drivers/pci/controller/pcie-apple.c b/drivers/pci/controller/pcie-apple.c
> >> index c2cffc0659f4..db038a9d4831 100644
> >> --- a/drivers/pci/controller/pcie-apple.c
> >> +++ b/drivers/pci/controller/pcie-apple.c
> >> @@ -30,6 +30,7 @@
> >>   #include <linux/msi.h>
> >>   #include <linux/of_irq.h>
> >>   #include <linux/pci-ecam.h>
> >> +#include <linux/pci-pwrctrl.h>
> >>     #include "pci-host-common.h"
> >>   @@ -825,6 +826,21 @@ static int apple_pcie_init(struct
> >> pci_config_window *cfg)
> >>   	if (WARN_ON(!pcie))
> >>   		return -ENOENT;
> >>   +	ret = pci_pwrctrl_create_devices(pcie->dev);
> >> +	if (ret) {
> >> +		dev_err(pcie->dev, "Failed to create pwrctrl devices: %pe\n", ret);
> >> +		return ret;
> >> +	}
> >> +
> >> +	ret = pci_pwrctrl_power_on_devices(pcie->dev);
> >> +	if (ret) {
> >> +		if (ret != -EPROBE_DEFER) {
> >> +			dev_err(pcie->dev, "Failed to power on devices: %pe\n", ret);
> >> +			pci_pwrctrl_destroy_devices(pcie->dev);
> >> +		}
> >> +		return ret;
> >> +	}
> >> +
> > Why is this done globally while the whole driver works on a per-port
> > basis, and that the proposed DT updates are also per port?
> 
> pci_pwrctrl_power_on_devices takes a struct device as parameter, but
> pcie-apple does not allocate device structs for the individual ports.
> This could be changed of course. But since pci_pwrctrl_* operate on
> the subnodes recursively, it works just fine this way.

Works fine is one thing. Being consistent with the way the rest of the
driver works is another. pci_pwrctrl_create_device() and
pci_pwrctrl_power_on_device() appear to do exactly what would be
required for a single port. They just needs to be exported made
global/exported.

> Additionally, we would like to be sure all the endpoints can be powered
> before we start initializing the individual ports. Otherwise we could
> end up in a situation where some ports are initialized but others are
> not when we realize some driver needed to power on the endpoints for
> port n is not yet bound.

How can a driver be bound to a device connected to a port if we
haven't been through the initial link-up dance, which is what
apple_pcie_setup_port() does? The power supplies are per-port for a
reason.

> There is no appropriate pci_pwrctrl API for checking for the
> availability without changing the state, so this is an additional
> reason to do it early.

Availability of what? To be clear, I'm not suggesting delaying
switching of the power. I'm merely suggesting managing it on a per
port basis.

Thanks,

	M.

-- 
Without deviation from the norm, progress is not possible.