[PATCH 0/2] Improvements to system power consumption at S5

[PATCH 0/2] Improvements to system power consumption at S5

[Mario Limonciello]

OEM systems that ship with Linux preloaded need to go through energy
certifications that match regulatory bodies in the regions that they
will ship.

If any of those certifications don't pass then OEMs might not be able
to ship systems preloaded in applicable regions.

Multiple models of systems are reported to fail in Linux but pass
for Windows on the exact same hardware.

By looking at a breakdown of power consumption across devices the
issue is that some devices aren't turned off when user puts the
system into S5.

This series modifies the PCI driver and PCIe port shutdown
codepaths to ensure that devices aren't needlessly woken up and
that the hierarchy for all ports is put into D3cold.

With this series power consumption at S5 drops on some affected
systems to ranges that should be acceptable to ship preloaded.

Cc: mpearson-lenovo@squebb.ca

Mario Limonciello (2):
  PCI: Avoid runtime resuming devices if system is shutting down
  PCI/portdrv: Place PCIe port hierarchy into D3cold at shutdown

 drivers/pci/pci-driver.c   |  4 +++-
 drivers/pci/pcie/portdrv.c | 11 ++++++++---
 2 files changed, 11 insertions(+), 4 deletions(-)

-- 
2.34.1

[PATCH 1/2] PCI: Avoid runtime resuming devices if system is shutting down

[Mario Limonciello]

When the system is going to S5 resuming devices is unnecessary at
best and may cause problems for power consumption at S5 at worst.

Check the target `system_state` in the pci driver shutdown() callback
and skip the runtime resume step if the system is being powered off.

Cc: mpearson-lenovo@squebb.ca
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
---
 drivers/pci/pci-driver.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
index 51ec9e7e784f..5a6f7a786952 100644
--- a/drivers/pci/pci-driver.c
+++ b/drivers/pci/pci-driver.c
@@ -507,7 +507,9 @@ static void pci_device_shutdown(struct device *dev)
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct pci_driver *drv = pci_dev->driver;
 
-	pm_runtime_resume(dev);
+	/* If system will go to S5, don't wake up PCI devices for shutdown */
+	if (system_state != SYSTEM_POWER_OFF)
+		pm_runtime_resume(dev);
 
 	if (drv && drv->shutdown)
 		drv->shutdown(pci_dev);
-- 
2.34.1

Re: [PATCH 1/2] PCI: Avoid runtime resuming devices if system is shutting down

[Rafael J. Wysocki]

On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
<mario.limonciello@amd.com> wrote:
>
> When the system is going to S5 resuming devices is unnecessary at
> best and may cause problems for power consumption at S5 at worst.
>
> Check the target `system_state` in the pci driver shutdown() callback
> and skip the runtime resume step if the system is being powered off.
>
> Cc: mpearson-lenovo@squebb.ca
> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> ---
>  drivers/pci/pci-driver.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> index 51ec9e7e784f..5a6f7a786952 100644
> --- a/drivers/pci/pci-driver.c
> +++ b/drivers/pci/pci-driver.c
> @@ -507,7 +507,9 @@ static void pci_device_shutdown(struct device *dev)
>         struct pci_dev *pci_dev = to_pci_dev(dev);
>         struct pci_driver *drv = pci_dev->driver;
>
> -       pm_runtime_resume(dev);
> +       /* If system will go to S5, don't wake up PCI devices for shutdown */
> +       if (system_state != SYSTEM_POWER_OFF)
> +               pm_runtime_resume(dev);

What's the guarantee that the driver's shutdown callback will not
access MMIO regions of a runtime-suspended device?

>         if (drv && drv->shutdown)
>                 drv->shutdown(pci_dev);
> --

Re: [PATCH 1/2] PCI: Avoid runtime resuming devices if system is shutting down

[Mario Limonciello]

On 12/13/2023 12:31, Rafael J. Wysocki wrote:
> On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
> <mario.limonciello@amd.com> wrote:
>>
>> When the system is going to S5 resuming devices is unnecessary at
>> best and may cause problems for power consumption at S5 at worst.
>>
>> Check the target `system_state` in the pci driver shutdown() callback
>> and skip the runtime resume step if the system is being powered off.
>>
>> Cc: mpearson-lenovo@squebb.ca
>> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
>> ---
>>   drivers/pci/pci-driver.c | 4 +++-
>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
>> index 51ec9e7e784f..5a6f7a786952 100644
>> --- a/drivers/pci/pci-driver.c
>> +++ b/drivers/pci/pci-driver.c
>> @@ -507,7 +507,9 @@ static void pci_device_shutdown(struct device *dev)
>>          struct pci_dev *pci_dev = to_pci_dev(dev);
>>          struct pci_driver *drv = pci_dev->driver;
>>
>> -       pm_runtime_resume(dev);
>> +       /* If system will go to S5, don't wake up PCI devices for shutdown */
>> +       if (system_state != SYSTEM_POWER_OFF)
>> +               pm_runtime_resume(dev);
> 
> What's the guarantee that the driver's shutdown callback will not
> access MMIO regions of a runtime-suspended device?

This is something that may need to be audited.

I have an expectation that drivers will runtime resume a device before 
they try to access it.

> 
>>          if (drv && drv->shutdown)
>>                  drv->shutdown(pci_dev);
>> --

Re: [PATCH 1/2] PCI: Avoid runtime resuming devices if system is shutting down

[Rafael J. Wysocki]

On Wed, Dec 13, 2023 at 7:33 PM Mario Limonciello
<mario.limonciello@amd.com> wrote:
>
> On 12/13/2023 12:31, Rafael J. Wysocki wrote:
> > On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
> > <mario.limonciello@amd.com> wrote:
> >>
> >> When the system is going to S5 resuming devices is unnecessary at
> >> best and may cause problems for power consumption at S5 at worst.
> >>
> >> Check the target `system_state` in the pci driver shutdown() callback
> >> and skip the runtime resume step if the system is being powered off.
> >>
> >> Cc: mpearson-lenovo@squebb.ca
> >> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> >> ---
> >>   drivers/pci/pci-driver.c | 4 +++-
> >>   1 file changed, 3 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> >> index 51ec9e7e784f..5a6f7a786952 100644
> >> --- a/drivers/pci/pci-driver.c
> >> +++ b/drivers/pci/pci-driver.c
> >> @@ -507,7 +507,9 @@ static void pci_device_shutdown(struct device *dev)
> >>          struct pci_dev *pci_dev = to_pci_dev(dev);
> >>          struct pci_driver *drv = pci_dev->driver;
> >>
> >> -       pm_runtime_resume(dev);
> >> +       /* If system will go to S5, don't wake up PCI devices for shutdown */
> >> +       if (system_state != SYSTEM_POWER_OFF)
> >> +               pm_runtime_resume(dev);
> >
> > What's the guarantee that the driver's shutdown callback will not
> > access MMIO regions of a runtime-suspended device?
>
> This is something that may need to be audited.
>
> I have an expectation that drivers will runtime resume a device before
> they try to access it.

But they don't need to do it, because the core does it right now,
which is my point.

[PATCH 2/2] PCI/portdrv: Place PCIe port hierarchy into D3cold at shutdown

[Mario Limonciello]

When a system is being powered off it's important that PCIe ports
have been put into D3cold as there is no other software to turn
off the devices at S5.

If PCIe ports are left in D0 then any GPIOs toggled by the ACPI
power resources may be left enabled and devices may consume excess
power.

Cc: mpearson-lenovo@squebb.ca
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
---
 drivers/pci/pcie/portdrv.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/drivers/pci/pcie/portdrv.c b/drivers/pci/pcie/portdrv.c
index 14a4b89a3b83..08238680c481 100644
--- a/drivers/pci/pcie/portdrv.c
+++ b/drivers/pci/pcie/portdrv.c
@@ -734,9 +734,14 @@ static void pcie_portdrv_remove(struct pci_dev *dev)
 static void pcie_portdrv_shutdown(struct pci_dev *dev)
 {
 	if (pci_bridge_d3_possible(dev)) {
-		pm_runtime_forbid(&dev->dev);
-		pm_runtime_get_noresume(&dev->dev);
-		pm_runtime_dont_use_autosuspend(&dev->dev);
+		/* whole hierarchy goes into a low power state for S5 */
+		if (system_state == SYSTEM_POWER_OFF) {
+			pci_set_power_state(dev, PCI_D3cold);
+		} else {
+			pm_runtime_forbid(&dev->dev);
+			pm_runtime_get_noresume(&dev->dev);
+			pm_runtime_dont_use_autosuspend(&dev->dev);
+		}
 	}
 
 	pcie_port_device_remove(dev);
-- 
2.34.1

Re: [PATCH 2/2] PCI/portdrv: Place PCIe port hierarchy into D3cold at shutdown

[Rafael J. Wysocki]

On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
<mario.limonciello@amd.com> wrote:
>
> When a system is being powered off it's important that PCIe ports
> have been put into D3cold as there is no other software to turn
> off the devices at S5.
>
> If PCIe ports are left in D0 then any GPIOs toggled by the ACPI
> power resources may be left enabled and devices may consume excess
> power.

Isn't that a platform firmware issue?

It is the responsibility of the platform firmware to properly put the
platform into S5, including power removal from devices that are not
armed for power-on.

> Cc: mpearson-lenovo@squebb.ca
> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> ---
>  drivers/pci/pcie/portdrv.c | 11 ++++++++---
>  1 file changed, 8 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/pci/pcie/portdrv.c b/drivers/pci/pcie/portdrv.c
> index 14a4b89a3b83..08238680c481 100644
> --- a/drivers/pci/pcie/portdrv.c
> +++ b/drivers/pci/pcie/portdrv.c
> @@ -734,9 +734,14 @@ static void pcie_portdrv_remove(struct pci_dev *dev)
>  static void pcie_portdrv_shutdown(struct pci_dev *dev)
>  {
>         if (pci_bridge_d3_possible(dev)) {
> -               pm_runtime_forbid(&dev->dev);
> -               pm_runtime_get_noresume(&dev->dev);
> -               pm_runtime_dont_use_autosuspend(&dev->dev);
> +               /* whole hierarchy goes into a low power state for S5 */
> +               if (system_state == SYSTEM_POWER_OFF) {
> +                       pci_set_power_state(dev, PCI_D3cold);
> +               } else {
> +                       pm_runtime_forbid(&dev->dev);
> +                       pm_runtime_get_noresume(&dev->dev);
> +                       pm_runtime_dont_use_autosuspend(&dev->dev);
> +               }
>         }

Wouldn't it be better to remove power from the port after running the
code below?

>         pcie_port_device_remove(dev);
> --

Re: [PATCH 2/2] PCI/portdrv: Place PCIe port hierarchy into D3cold at shutdown

[Mario Limonciello]

On 12/13/2023 12:38, Rafael J. Wysocki wrote:
> On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
> <mario.limonciello@amd.com> wrote:
>>
>> When a system is being powered off it's important that PCIe ports
>> have been put into D3cold as there is no other software to turn
>> off the devices at S5.
>>
>> If PCIe ports are left in D0 then any GPIOs toggled by the ACPI
>> power resources may be left enabled and devices may consume excess
>> power.
> 
> Isn't that a platform firmware issue?
> 
> It is the responsibility of the platform firmware to properly put the
> platform into S5, including power removal from devices that are not
> armed for power-on.

The specific issues that triggered this series were tied to the PCIe 
ports for dGPUs.  There is a GPIO that is toggled by _ON or _OFF.

Windows calls _OFF as part of S5..

> 
>> Cc: mpearson-lenovo@squebb.ca
>> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
>> ---
>>   drivers/pci/pcie/portdrv.c | 11 ++++++++---
>>   1 file changed, 8 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/pci/pcie/portdrv.c b/drivers/pci/pcie/portdrv.c
>> index 14a4b89a3b83..08238680c481 100644
>> --- a/drivers/pci/pcie/portdrv.c
>> +++ b/drivers/pci/pcie/portdrv.c
>> @@ -734,9 +734,14 @@ static void pcie_portdrv_remove(struct pci_dev *dev)
>>   static void pcie_portdrv_shutdown(struct pci_dev *dev)
>>   {
>>          if (pci_bridge_d3_possible(dev)) {
>> -               pm_runtime_forbid(&dev->dev);
>> -               pm_runtime_get_noresume(&dev->dev);
>> -               pm_runtime_dont_use_autosuspend(&dev->dev);
>> +               /* whole hierarchy goes into a low power state for S5 */
>> +               if (system_state == SYSTEM_POWER_OFF) {
>> +                       pci_set_power_state(dev, PCI_D3cold);
>> +               } else {
>> +                       pm_runtime_forbid(&dev->dev);
>> +                       pm_runtime_get_noresume(&dev->dev);
>> +                       pm_runtime_dont_use_autosuspend(&dev->dev);
>> +               }
>>          }
> 
> Wouldn't it be better to remove power from the port after running the
> code below?
> 

Yes; I think you're right.  I'll do some more testing with this.

>>          pcie_port_device_remove(dev);
>> --

Re: [PATCH 2/2] PCI/portdrv: Place PCIe port hierarchy into D3cold at shutdown

[Rafael J. Wysocki]

On Wed, Dec 13, 2023 at 7:42 PM Mario Limonciello
<mario.limonciello@amd.com> wrote:
>
> On 12/13/2023 12:38, Rafael J. Wysocki wrote:
> > On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
> > <mario.limonciello@amd.com> wrote:
> >>
> >> When a system is being powered off it's important that PCIe ports
> >> have been put into D3cold as there is no other software to turn
> >> off the devices at S5.
> >>
> >> If PCIe ports are left in D0 then any GPIOs toggled by the ACPI
> >> power resources may be left enabled and devices may consume excess
> >> power.
> >
> > Isn't that a platform firmware issue?
> >
> > It is the responsibility of the platform firmware to properly put the
> > platform into S5, including power removal from devices that are not
> > armed for power-on.
>
> The specific issues that triggered this series were tied to the PCIe
> ports for dGPUs.  There is a GPIO that is toggled by _ON or _OFF.
>
> Windows calls _OFF as part of S5..

I see.

> >
> >> Cc: mpearson-lenovo@squebb.ca
> >> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> >> ---
> >>   drivers/pci/pcie/portdrv.c | 11 ++++++++---
> >>   1 file changed, 8 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/pci/pcie/portdrv.c b/drivers/pci/pcie/portdrv.c
> >> index 14a4b89a3b83..08238680c481 100644
> >> --- a/drivers/pci/pcie/portdrv.c
> >> +++ b/drivers/pci/pcie/portdrv.c
> >> @@ -734,9 +734,14 @@ static void pcie_portdrv_remove(struct pci_dev *dev)
> >>   static void pcie_portdrv_shutdown(struct pci_dev *dev)
> >>   {
> >>          if (pci_bridge_d3_possible(dev)) {
> >> -               pm_runtime_forbid(&dev->dev);
> >> -               pm_runtime_get_noresume(&dev->dev);
> >> -               pm_runtime_dont_use_autosuspend(&dev->dev);
> >> +               /* whole hierarchy goes into a low power state for S5 */
> >> +               if (system_state == SYSTEM_POWER_OFF) {
> >> +                       pci_set_power_state(dev, PCI_D3cold);
> >> +               } else {
> >> +                       pm_runtime_forbid(&dev->dev);
> >> +                       pm_runtime_get_noresume(&dev->dev);
> >> +                       pm_runtime_dont_use_autosuspend(&dev->dev);
> >> +               }
> >>          }
> >
> > Wouldn't it be better to remove power from the port after running the
> > code below?
> >
>
> Yes; I think you're right.  I'll do some more testing with this.
>
> >>          pcie_port_device_remove(dev);
> >> --

IIRC, to do this all properly, you'd need to rework the shutdown path
to look like the hibernation power-off one.  Or even use the latter
for shutdown?

There was no reason to do that till now, so it has not been done, but
it looks like you have one.

Re: [PATCH 2/2] PCI/portdrv: Place PCIe port hierarchy into D3cold at shutdown

[Kai-Heng Feng]

Hi Mario and Rafael,

On Thu, Dec 14, 2023 at 2:46 AM Rafael J. Wysocki <rafael@kernel.org> wrote:
>
> On Wed, Dec 13, 2023 at 7:42 PM Mario Limonciello
> <mario.limonciello@amd.com> wrote:
> >
> > On 12/13/2023 12:38, Rafael J. Wysocki wrote:
> > > On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
> > > <mario.limonciello@amd.com> wrote:
> > >>
> > >> When a system is being powered off it's important that PCIe ports
> > >> have been put into D3cold as there is no other software to turn
> > >> off the devices at S5.
> > >>
> > >> If PCIe ports are left in D0 then any GPIOs toggled by the ACPI
> > >> power resources may be left enabled and devices may consume excess
> > >> power.
> > >
> > > Isn't that a platform firmware issue?
> > >
> > > It is the responsibility of the platform firmware to properly put the
> > > platform into S5, including power removal from devices that are not
> > > armed for power-on.
> >
> > The specific issues that triggered this series were tied to the PCIe
> > ports for dGPUs.  There is a GPIO that is toggled by _ON or _OFF.
> >
> > Windows calls _OFF as part of S5..
>
> I see.
>
> > >
> > >> Cc: mpearson-lenovo@squebb.ca
> > >> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> > >> ---
> > >>   drivers/pci/pcie/portdrv.c | 11 ++++++++---
> > >>   1 file changed, 8 insertions(+), 3 deletions(-)
> > >>
> > >> diff --git a/drivers/pci/pcie/portdrv.c b/drivers/pci/pcie/portdrv.c
> > >> index 14a4b89a3b83..08238680c481 100644
> > >> --- a/drivers/pci/pcie/portdrv.c
> > >> +++ b/drivers/pci/pcie/portdrv.c
> > >> @@ -734,9 +734,14 @@ static void pcie_portdrv_remove(struct pci_dev *dev)
> > >>   static void pcie_portdrv_shutdown(struct pci_dev *dev)
> > >>   {
> > >>          if (pci_bridge_d3_possible(dev)) {
> > >> -               pm_runtime_forbid(&dev->dev);
> > >> -               pm_runtime_get_noresume(&dev->dev);
> > >> -               pm_runtime_dont_use_autosuspend(&dev->dev);
> > >> +               /* whole hierarchy goes into a low power state for S5 */
> > >> +               if (system_state == SYSTEM_POWER_OFF) {
> > >> +                       pci_set_power_state(dev, PCI_D3cold);
> > >> +               } else {
> > >> +                       pm_runtime_forbid(&dev->dev);
> > >> +                       pm_runtime_get_noresume(&dev->dev);
> > >> +                       pm_runtime_dont_use_autosuspend(&dev->dev);
> > >> +               }
> > >>          }
> > >
> > > Wouldn't it be better to remove power from the port after running the
> > > code below?
> > >
> >
> > Yes; I think you're right.  I'll do some more testing with this.
> >
> > >>          pcie_port_device_remove(dev);
> > >> --
>
> IIRC, to do this all properly, you'd need to rework the shutdown path
> to look like the hibernation power-off one.  Or even use the latter
> for shutdown?
>
> There was no reason to do that till now, so it has not been done, but
> it looks like you have one.
>

I am working on exactly same thing but with a different approach.
Because this is needed for more than just PCI devices.
I haven't written a proper commit message yet, but the implementation
is quite simple:

diff --git a/drivers/acpi/device_pm.c b/drivers/acpi/device_pm.c
index f007116a8427..b90c6cf6faf4 100644
--- a/drivers/acpi/device_pm.c
+++ b/drivers/acpi/device_pm.c
@@ -967,15 +967,17 @@ EXPORT_SYMBOL_GPL(acpi_pm_set_device_wakeup);
  * @adev: ACPI device node corresponding to @dev.
  * @system_state: System state to choose the device state for.
  */
-static int acpi_dev_pm_low_power(struct device *dev, struct acpi_device *adev,
-                 u32 system_state)
+static int acpi_dev_pm_low_power(struct acpi_device *adev, void* data)
 {
     int ret, state;
+    u32 *system_state = data;

     if (!acpi_device_power_manageable(adev))
         return 0;

-    ret = acpi_dev_pm_get_state(dev, adev, system_state, NULL, &state);
+    acpi_dev_for_each_child(adev, acpi_dev_pm_low_power, data);
+
+    ret = acpi_dev_pm_get_state(&adev->dev, adev, *system_state, NULL, &state);
     return ret ? ret : acpi_device_set_power(adev, state);
 }

@@ -1016,7 +1018,7 @@ int acpi_dev_suspend(struct device *dev, bool wakeup)
         wakeup = false;
     }

-    error = acpi_dev_pm_low_power(dev, adev, target_state);
+    error = acpi_dev_pm_low_power(adev, &target_state);
     if (error && wakeup)
         acpi_device_wakeup_disable(adev);

@@ -1386,6 +1388,7 @@ static struct dev_pm_domain acpi_general_pm_domain = {
 static void acpi_dev_pm_detach(struct device *dev, bool power_off)
 {
     struct acpi_device *adev = ACPI_COMPANION(dev);
+    u32 state = ACPI_STATE_S0;

     if (adev && dev->pm_domain == &acpi_general_pm_domain) {
         dev_pm_domain_set(dev, NULL);
@@ -1400,7 +1403,7 @@ static void acpi_dev_pm_detach(struct device
*dev, bool power_off)
             dev_pm_qos_hide_latency_limit(dev);
             dev_pm_qos_hide_flags(dev);
             acpi_device_wakeup_disable(adev);
-            acpi_dev_pm_low_power(dev, adev, ACPI_STATE_S0);
+            acpi_dev_pm_low_power(adev, &state);
         }
     }
 }
@@ -1514,4 +1517,16 @@ bool acpi_dev_state_d0(struct device *dev)
 }
 EXPORT_SYMBOL_GPL(acpi_dev_state_d0);

+void acpi_dev_shutdown(struct device *dev)
+{
+    struct acpi_device *adev = ACPI_COMPANION(dev);
+    u32 state = ACPI_STATE_S5;
+
+    if (!adev)
+        return;
+
+    acpi_device_wakeup_disable(adev);
+    acpi_dev_pm_low_power(adev, &state);
+}
+EXPORT_SYMBOL_GPL(acpi_dev_shutdown);
 #endif /* CONFIG_PM */
diff --git a/drivers/base/core.c b/drivers/base/core.c
index 6ceaf50f5a67..7e7c99eade63 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -45,6 +45,15 @@ static void __fw_devlink_link_to_consumers(struct
device *dev);
 static bool fw_devlink_drv_reg_done;
 static bool fw_devlink_best_effort;

+#ifdef CONFIG_ACPI
+static inline void fw_dev_shutdown(struct device *dev)
+{
+    acpi_dev_shutdown(dev);
+}
+#else
+static inline void fw_dev_shutdown(struct device *dev) {  }
+#endif
+
 /**
  * __fwnode_link_add - Create a link between two fwnode_handles.
  * @con: Consumer end of the link.
@@ -4780,6 +4789,8 @@ void device_shutdown(void)
             dev->driver->shutdown(dev);
         }

+        fw_dev_shutdown(dev);
+
         device_unlock(dev);
         if (parent)
             device_unlock(parent);
diff --git a/include/linux/acpi.h b/include/linux/acpi.h
index 641dc4843987..374f9eb75c22 100644
--- a/include/linux/acpi.h
+++ b/include/linux/acpi.h
@@ -1130,6 +1130,7 @@ int acpi_subsys_runtime_resume(struct device *dev);
 int acpi_dev_pm_attach(struct device *dev, bool power_on);
 bool acpi_storage_d3(struct device *dev);
 bool acpi_dev_state_d0(struct device *dev);
+void acpi_dev_shutdown(struct device *dev);
 #else
 static inline int acpi_subsys_runtime_suspend(struct device *dev) { return 0; }
 static inline int acpi_subsys_runtime_resume(struct device *dev) { return 0; }
@@ -1145,6 +1146,7 @@ static inline bool acpi_dev_state_d0(struct device *dev)
 {
     return true;
 }
+static inline void acpi_dev_shutdown(struct device *dev) { }
 #endif

 #if defined(CONFIG_ACPI) && defined(CONFIG_PM_SLEEP)

Re: [PATCH 2/2] PCI/portdrv: Place PCIe port hierarchy into D3cold at shutdown

[Rafael J. Wysocki]

On Thu, Dec 14, 2023 at 4:46 AM Kai-Heng Feng
<kai.heng.feng@canonical.com> wrote:
>
> Hi Mario and Rafael,
>
> On Thu, Dec 14, 2023 at 2:46 AM Rafael J. Wysocki <rafael@kernel.org> wrote:
> >
> > On Wed, Dec 13, 2023 at 7:42 PM Mario Limonciello
> > <mario.limonciello@amd.com> wrote:
> > >
> > > On 12/13/2023 12:38, Rafael J. Wysocki wrote:
> > > > On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
> > > > <mario.limonciello@amd.com> wrote:
> > > >>
> > > >> When a system is being powered off it's important that PCIe ports
> > > >> have been put into D3cold as there is no other software to turn
> > > >> off the devices at S5.
> > > >>
> > > >> If PCIe ports are left in D0 then any GPIOs toggled by the ACPI
> > > >> power resources may be left enabled and devices may consume excess
> > > >> power.
> > > >
> > > > Isn't that a platform firmware issue?
> > > >
> > > > It is the responsibility of the platform firmware to properly put the
> > > > platform into S5, including power removal from devices that are not
> > > > armed for power-on.
> > >
> > > The specific issues that triggered this series were tied to the PCIe
> > > ports for dGPUs.  There is a GPIO that is toggled by _ON or _OFF.
> > >
> > > Windows calls _OFF as part of S5..
> >
> > I see.
> >
> > > >
> > > >> Cc: mpearson-lenovo@squebb.ca
> > > >> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> > > >> ---
> > > >>   drivers/pci/pcie/portdrv.c | 11 ++++++++---
> > > >>   1 file changed, 8 insertions(+), 3 deletions(-)
> > > >>
> > > >> diff --git a/drivers/pci/pcie/portdrv.c b/drivers/pci/pcie/portdrv.c
> > > >> index 14a4b89a3b83..08238680c481 100644
> > > >> --- a/drivers/pci/pcie/portdrv.c
> > > >> +++ b/drivers/pci/pcie/portdrv.c
> > > >> @@ -734,9 +734,14 @@ static void pcie_portdrv_remove(struct pci_dev *dev)
> > > >>   static void pcie_portdrv_shutdown(struct pci_dev *dev)
> > > >>   {
> > > >>          if (pci_bridge_d3_possible(dev)) {
> > > >> -               pm_runtime_forbid(&dev->dev);
> > > >> -               pm_runtime_get_noresume(&dev->dev);
> > > >> -               pm_runtime_dont_use_autosuspend(&dev->dev);
> > > >> +               /* whole hierarchy goes into a low power state for S5 */
> > > >> +               if (system_state == SYSTEM_POWER_OFF) {
> > > >> +                       pci_set_power_state(dev, PCI_D3cold);
> > > >> +               } else {
> > > >> +                       pm_runtime_forbid(&dev->dev);
> > > >> +                       pm_runtime_get_noresume(&dev->dev);
> > > >> +                       pm_runtime_dont_use_autosuspend(&dev->dev);
> > > >> +               }
> > > >>          }
> > > >
> > > > Wouldn't it be better to remove power from the port after running the
> > > > code below?
> > > >
> > >
> > > Yes; I think you're right.  I'll do some more testing with this.
> > >
> > > >>          pcie_port_device_remove(dev);
> > > >> --
> >
> > IIRC, to do this all properly, you'd need to rework the shutdown path
> > to look like the hibernation power-off one.  Or even use the latter
> > for shutdown?
> >
> > There was no reason to do that till now, so it has not been done, but
> > it looks like you have one.
> >
>
> I am working on exactly same thing but with a different approach.
> Because this is needed for more than just PCI devices.
> I haven't written a proper commit message yet, but the implementation
> is quite simple:

As I said, doing this properly requires something like the hibernation
power-off transition to be carried out for S5.

I think that the existing hibernation power-off code can be used as-is
for this purpose even.

> diff --git a/drivers/acpi/device_pm.c b/drivers/acpi/device_pm.c
> index f007116a8427..b90c6cf6faf4 100644
> --- a/drivers/acpi/device_pm.c
> +++ b/drivers/acpi/device_pm.c
> @@ -967,15 +967,17 @@ EXPORT_SYMBOL_GPL(acpi_pm_set_device_wakeup);
>   * @adev: ACPI device node corresponding to @dev.
>   * @system_state: System state to choose the device state for.
>   */
> -static int acpi_dev_pm_low_power(struct device *dev, struct acpi_device *adev,
> -                 u32 system_state)
> +static int acpi_dev_pm_low_power(struct acpi_device *adev, void* data)
>  {
>      int ret, state;
> +    u32 *system_state = data;
>
>      if (!acpi_device_power_manageable(adev))
>          return 0;
>
> -    ret = acpi_dev_pm_get_state(dev, adev, system_state, NULL, &state);
> +    acpi_dev_for_each_child(adev, acpi_dev_pm_low_power, data);
> +
> +    ret = acpi_dev_pm_get_state(&adev->dev, adev, *system_state, NULL, &state);
>      return ret ? ret : acpi_device_set_power(adev, state);
>  }
>
> @@ -1016,7 +1018,7 @@ int acpi_dev_suspend(struct device *dev, bool wakeup)
>          wakeup = false;
>      }
>
> -    error = acpi_dev_pm_low_power(dev, adev, target_state);
> +    error = acpi_dev_pm_low_power(adev, &target_state);
>      if (error && wakeup)
>          acpi_device_wakeup_disable(adev);
>
> @@ -1386,6 +1388,7 @@ static struct dev_pm_domain acpi_general_pm_domain = {
>  static void acpi_dev_pm_detach(struct device *dev, bool power_off)
>  {
>      struct acpi_device *adev = ACPI_COMPANION(dev);
> +    u32 state = ACPI_STATE_S0;
>
>      if (adev && dev->pm_domain == &acpi_general_pm_domain) {
>          dev_pm_domain_set(dev, NULL);
> @@ -1400,7 +1403,7 @@ static void acpi_dev_pm_detach(struct device
> *dev, bool power_off)
>              dev_pm_qos_hide_latency_limit(dev);
>              dev_pm_qos_hide_flags(dev);
>              acpi_device_wakeup_disable(adev);
> -            acpi_dev_pm_low_power(dev, adev, ACPI_STATE_S0);
> +            acpi_dev_pm_low_power(adev, &state);
>          }
>      }
>  }
> @@ -1514,4 +1517,16 @@ bool acpi_dev_state_d0(struct device *dev)
>  }
>  EXPORT_SYMBOL_GPL(acpi_dev_state_d0);
>
> +void acpi_dev_shutdown(struct device *dev)
> +{
> +    struct acpi_device *adev = ACPI_COMPANION(dev);
> +    u32 state = ACPI_STATE_S5;
> +
> +    if (!adev)
> +        return;
> +
> +    acpi_device_wakeup_disable(adev);
> +    acpi_dev_pm_low_power(adev, &state);
> +}
> +EXPORT_SYMBOL_GPL(acpi_dev_shutdown);
>  #endif /* CONFIG_PM */
> diff --git a/drivers/base/core.c b/drivers/base/core.c
> index 6ceaf50f5a67..7e7c99eade63 100644
> --- a/drivers/base/core.c
> +++ b/drivers/base/core.c
> @@ -45,6 +45,15 @@ static void __fw_devlink_link_to_consumers(struct
> device *dev);
>  static bool fw_devlink_drv_reg_done;
>  static bool fw_devlink_best_effort;
>
> +#ifdef CONFIG_ACPI
> +static inline void fw_dev_shutdown(struct device *dev)
> +{
> +    acpi_dev_shutdown(dev);
> +}
> +#else
> +static inline void fw_dev_shutdown(struct device *dev) {  }
> +#endif
> +
>  /**
>   * __fwnode_link_add - Create a link between two fwnode_handles.
>   * @con: Consumer end of the link.
> @@ -4780,6 +4789,8 @@ void device_shutdown(void)
>              dev->driver->shutdown(dev);
>          }
>
> +        fw_dev_shutdown(dev);
> +
>          device_unlock(dev);
>          if (parent)
>              device_unlock(parent);
> diff --git a/include/linux/acpi.h b/include/linux/acpi.h
> index 641dc4843987..374f9eb75c22 100644
> --- a/include/linux/acpi.h
> +++ b/include/linux/acpi.h
> @@ -1130,6 +1130,7 @@ int acpi_subsys_runtime_resume(struct device *dev);
>  int acpi_dev_pm_attach(struct device *dev, bool power_on);
>  bool acpi_storage_d3(struct device *dev);
>  bool acpi_dev_state_d0(struct device *dev);
> +void acpi_dev_shutdown(struct device *dev);
>  #else
>  static inline int acpi_subsys_runtime_suspend(struct device *dev) { return 0; }
>  static inline int acpi_subsys_runtime_resume(struct device *dev) { return 0; }
> @@ -1145,6 +1146,7 @@ static inline bool acpi_dev_state_d0(struct device *dev)
>  {
>      return true;
>  }
> +static inline void acpi_dev_shutdown(struct device *dev) { }
>  #endif
>
>  #if defined(CONFIG_ACPI) && defined(CONFIG_PM_SLEEP)
>

Re: [PATCH 2/2] PCI/portdrv: Place PCIe port hierarchy into D3cold at shutdown

[Mario Limonciello]

On 12/14/2023 03:00, Rafael J. Wysocki wrote:
> On Thu, Dec 14, 2023 at 4:46 AM Kai-Heng Feng
> <kai.heng.feng@canonical.com> wrote:
>>
>> Hi Mario and Rafael,
>>
>> On Thu, Dec 14, 2023 at 2:46 AM Rafael J. Wysocki <rafael@kernel.org> wrote:
>>>
>>> On Wed, Dec 13, 2023 at 7:42 PM Mario Limonciello
>>> <mario.limonciello@amd.com> wrote:
>>>>
>>>> On 12/13/2023 12:38, Rafael J. Wysocki wrote:
>>>>> On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
>>>>> <mario.limonciello@amd.com> wrote:
>>>>>>
>>>>>> When a system is being powered off it's important that PCIe ports
>>>>>> have been put into D3cold as there is no other software to turn
>>>>>> off the devices at S5.
>>>>>>
>>>>>> If PCIe ports are left in D0 then any GPIOs toggled by the ACPI
>>>>>> power resources may be left enabled and devices may consume excess
>>>>>> power.
>>>>>
>>>>> Isn't that a platform firmware issue?
>>>>>
>>>>> It is the responsibility of the platform firmware to properly put the
>>>>> platform into S5, including power removal from devices that are not
>>>>> armed for power-on.
>>>>
>>>> The specific issues that triggered this series were tied to the PCIe
>>>> ports for dGPUs.  There is a GPIO that is toggled by _ON or _OFF.
>>>>
>>>> Windows calls _OFF as part of S5..
>>>
>>> I see.
>>>
>>>>>
>>>>>> Cc: mpearson-lenovo@squebb.ca
>>>>>> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
>>>>>> ---
>>>>>>    drivers/pci/pcie/portdrv.c | 11 ++++++++---
>>>>>>    1 file changed, 8 insertions(+), 3 deletions(-)
>>>>>>
>>>>>> diff --git a/drivers/pci/pcie/portdrv.c b/drivers/pci/pcie/portdrv.c
>>>>>> index 14a4b89a3b83..08238680c481 100644
>>>>>> --- a/drivers/pci/pcie/portdrv.c
>>>>>> +++ b/drivers/pci/pcie/portdrv.c
>>>>>> @@ -734,9 +734,14 @@ static void pcie_portdrv_remove(struct pci_dev *dev)
>>>>>>    static void pcie_portdrv_shutdown(struct pci_dev *dev)
>>>>>>    {
>>>>>>           if (pci_bridge_d3_possible(dev)) {
>>>>>> -               pm_runtime_forbid(&dev->dev);
>>>>>> -               pm_runtime_get_noresume(&dev->dev);
>>>>>> -               pm_runtime_dont_use_autosuspend(&dev->dev);
>>>>>> +               /* whole hierarchy goes into a low power state for S5 */
>>>>>> +               if (system_state == SYSTEM_POWER_OFF) {
>>>>>> +                       pci_set_power_state(dev, PCI_D3cold);
>>>>>> +               } else {
>>>>>> +                       pm_runtime_forbid(&dev->dev);
>>>>>> +                       pm_runtime_get_noresume(&dev->dev);
>>>>>> +                       pm_runtime_dont_use_autosuspend(&dev->dev);
>>>>>> +               }
>>>>>>           }
>>>>>
>>>>> Wouldn't it be better to remove power from the port after running the
>>>>> code below?
>>>>>
>>>>
>>>> Yes; I think you're right.  I'll do some more testing with this.
>>>>
>>>>>>           pcie_port_device_remove(dev);
>>>>>> --
>>>
>>> IIRC, to do this all properly, you'd need to rework the shutdown path
>>> to look like the hibernation power-off one.  Or even use the latter
>>> for shutdown?
>>>
>>> There was no reason to do that till now, so it has not been done, but
>>> it looks like you have one.
>>>
>>
>> I am working on exactly same thing but with a different approach.
>> Because this is needed for more than just PCI devices.
>> I haven't written a proper commit message yet, but the implementation
>> is quite simple:
> 
> As I said, doing this properly requires something like the hibernation
> power-off transition to be carried out for S5.
> 
> I think that the existing hibernation power-off code can be used as-is
> for this purpose even.
> 

I feel Rafael is right here that unifying the hibernation and shutdown 
paths is the right direction.  Our team just double checked the 
"unpatched" Linux S4 measurements on a system that otherwise had 
problems with S5 and they show the same decreases in power my patch 
series showed.

KH,

I'm going to be OOO for a while with the holidays around the corner and 
some personal time.  If you end up working on some patches to unify the 
S4/S5 codepaths CC me on them and I'll look when I'm back from my leave.

Thanks,

Re: [PATCH 2/2] PCI/portdrv: Place PCIe port hierarchy into D3cold at shutdown

[Kai-Heng Feng]

On Fri, Dec 15, 2023 at 12:44 AM Mario Limonciello
<mario.limonciello@amd.com> wrote:
>
> On 12/14/2023 03:00, Rafael J. Wysocki wrote:
> > On Thu, Dec 14, 2023 at 4:46 AM Kai-Heng Feng
> > <kai.heng.feng@canonical.com> wrote:
> >>
> >> Hi Mario and Rafael,
> >>
> >> On Thu, Dec 14, 2023 at 2:46 AM Rafael J. Wysocki <rafael@kernel.org> wrote:
> >>>
> >>> On Wed, Dec 13, 2023 at 7:42 PM Mario Limonciello
> >>> <mario.limonciello@amd.com> wrote:
> >>>>
> >>>> On 12/13/2023 12:38, Rafael J. Wysocki wrote:
> >>>>> On Wed, Dec 13, 2023 at 7:27 PM Mario Limonciello
> >>>>> <mario.limonciello@amd.com> wrote:
> >>>>>>
> >>>>>> When a system is being powered off it's important that PCIe ports
> >>>>>> have been put into D3cold as there is no other software to turn
> >>>>>> off the devices at S5.
> >>>>>>
> >>>>>> If PCIe ports are left in D0 then any GPIOs toggled by the ACPI
> >>>>>> power resources may be left enabled and devices may consume excess
> >>>>>> power.
> >>>>>
> >>>>> Isn't that a platform firmware issue?
> >>>>>
> >>>>> It is the responsibility of the platform firmware to properly put the
> >>>>> platform into S5, including power removal from devices that are not
> >>>>> armed for power-on.
> >>>>
> >>>> The specific issues that triggered this series were tied to the PCIe
> >>>> ports for dGPUs.  There is a GPIO that is toggled by _ON or _OFF.
> >>>>
> >>>> Windows calls _OFF as part of S5..
> >>>
> >>> I see.
> >>>
> >>>>>
> >>>>>> Cc: mpearson-lenovo@squebb.ca
> >>>>>> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> >>>>>> ---
> >>>>>>    drivers/pci/pcie/portdrv.c | 11 ++++++++---
> >>>>>>    1 file changed, 8 insertions(+), 3 deletions(-)
> >>>>>>
> >>>>>> diff --git a/drivers/pci/pcie/portdrv.c b/drivers/pci/pcie/portdrv.c
> >>>>>> index 14a4b89a3b83..08238680c481 100644
> >>>>>> --- a/drivers/pci/pcie/portdrv.c
> >>>>>> +++ b/drivers/pci/pcie/portdrv.c
> >>>>>> @@ -734,9 +734,14 @@ static void pcie_portdrv_remove(struct pci_dev *dev)
> >>>>>>    static void pcie_portdrv_shutdown(struct pci_dev *dev)
> >>>>>>    {
> >>>>>>           if (pci_bridge_d3_possible(dev)) {
> >>>>>> -               pm_runtime_forbid(&dev->dev);
> >>>>>> -               pm_runtime_get_noresume(&dev->dev);
> >>>>>> -               pm_runtime_dont_use_autosuspend(&dev->dev);
> >>>>>> +               /* whole hierarchy goes into a low power state for S5 */
> >>>>>> +               if (system_state == SYSTEM_POWER_OFF) {
> >>>>>> +                       pci_set_power_state(dev, PCI_D3cold);
> >>>>>> +               } else {
> >>>>>> +                       pm_runtime_forbid(&dev->dev);
> >>>>>> +                       pm_runtime_get_noresume(&dev->dev);
> >>>>>> +                       pm_runtime_dont_use_autosuspend(&dev->dev);
> >>>>>> +               }
> >>>>>>           }
> >>>>>
> >>>>> Wouldn't it be better to remove power from the port after running the
> >>>>> code below?
> >>>>>
> >>>>
> >>>> Yes; I think you're right.  I'll do some more testing with this.
> >>>>
> >>>>>>           pcie_port_device_remove(dev);
> >>>>>> --
> >>>
> >>> IIRC, to do this all properly, you'd need to rework the shutdown path
> >>> to look like the hibernation power-off one.  Or even use the latter
> >>> for shutdown?
> >>>
> >>> There was no reason to do that till now, so it has not been done, but
> >>> it looks like you have one.
> >>>
> >>
> >> I am working on exactly same thing but with a different approach.
> >> Because this is needed for more than just PCI devices.
> >> I haven't written a proper commit message yet, but the implementation
> >> is quite simple:
> >
> > As I said, doing this properly requires something like the hibernation
> > power-off transition to be carried out for S5.
> >
> > I think that the existing hibernation power-off code can be used as-is
> > for this purpose even.
> >
>
> I feel Rafael is right here that unifying the hibernation and shutdown
> paths is the right direction.  Our team just double checked the
> "unpatched" Linux S4 measurements on a system that otherwise had
> problems with S5 and they show the same decreases in power my patch
> series showed.

I agree this is the right approach because Windows is using S4 for its
"Fast Startup" feature.

Is there any historical reason that .power_off and .shutdown are separated?
And, should we take a step further and also unify the .remove
callback? I recall there was a discussion on consolidating .shutdown
and .remove but somehow it didn't take off.

>
> KH,
>
> I'm going to be OOO for a while with the holidays around the corner and
> some personal time.  If you end up working on some patches to unify the
> S4/S5 codepaths CC me on them and I'll look when I'm back from my leave.

Sure thing. This can take a while.

Kai-Heng

>
> Thanks,
>