diff mbox

[RFC] PCI / PM: Avoid resuming PCI devices during system suspend

Message ID 1654338.pQ8g9cpnmS@vostro.rjw.lan
State Not Applicable
Headers show

Commit Message

Rafael J. Wysocki Jan. 12, 2015, 10:02 a.m. UTC 
On Thursday, January 08, 2015 10:51:06 AM Alan Stern wrote:
> On Thu, 8 Jan 2015, Rafael J. Wysocki wrote:
> 
> > From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > 
> > Commit f25c0ae2b4c4 (ACPI / PM: Avoid resuming devices in ACPI PM
> > domain during system suspend) modified the ACPI PM domain's system
> > suspend callbacks to allow devices attached to it to be left in the
> > runtime-suspended state during system suspend so as to optimize
> > the suspend process.
> > 
> > This was based on the general mechanism introduced by commit
> > aae4518b3124 (PM / sleep: Mechanism to avoid resuming runtime-suspended
> > devices unnecessarily).
> > 
> > Extend that approach to PCI devices by modifying the PCI bus type's
> > ->prepare callback to return 1 for devices that are runtime-suspended
> > when it is being executed and that are in a suitable power state and
> > need not be resumed going forward.
> 
> Does this correctly handle PCI devices that aren't included in the ACPI
> tables?  For example, add-on PCI cards?

Well, it would if it took the case when a device was configured for remote
wakeup at run time but was not supposed to wakeup the system from sleep into
account.

That needs to be checked in pci_dev_keep_suspended() rather than in the platform
callback to cover all devices.

Updated patch follows.

Rafael


---
From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Subject: PCI / PM: Avoid resuming PCI devices during system suspend

Commit f25c0ae2b4c4 (ACPI / PM: Avoid resuming devices in ACPI PM
domain during system suspend) modified the ACPI PM domain's system
suspend callbacks to allow devices attached to it to be left in the
runtime-suspended state during system suspend so as to optimize
the suspend process.

This was based on the general mechanism introduced by commit
aae4518b3124 (PM / sleep: Mechanism to avoid resuming runtime-suspended
devices unnecessarily).

Extend that approach to PCI devices by modifying the PCI bus type's
->prepare callback to return 1 for devices that are runtime-suspended
when it is being executed and that are in a suitable power state and
need not be resumed going forward.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 drivers/pci/pci-acpi.c   |   12 ++++++++++++
 drivers/pci/pci-driver.c |   11 ++++++-----
 drivers/pci/pci.c        |   26 ++++++++++++++++++++++++++
 drivers/pci/pci.h        |    6 ++++++
 4 files changed, 50 insertions(+), 5 deletions(-)


--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

Alan Stern Jan. 12, 2015, 4:39 p.m. UTC | #1 
On Mon, 12 Jan 2015, Rafael J. Wysocki wrote:

> On Thursday, January 08, 2015 10:51:06 AM Alan Stern wrote:
> > On Thu, 8 Jan 2015, Rafael J. Wysocki wrote:
> > 
> > > From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > > 
> > > Commit f25c0ae2b4c4 (ACPI / PM: Avoid resuming devices in ACPI PM
> > > domain during system suspend) modified the ACPI PM domain's system
> > > suspend callbacks to allow devices attached to it to be left in the
> > > runtime-suspended state during system suspend so as to optimize
> > > the suspend process.
> > > 
> > > This was based on the general mechanism introduced by commit
> > > aae4518b3124 (PM / sleep: Mechanism to avoid resuming runtime-suspended
> > > devices unnecessarily).
> > > 
> > > Extend that approach to PCI devices by modifying the PCI bus type's
> > > ->prepare callback to return 1 for devices that are runtime-suspended
> > > when it is being executed and that are in a suitable power state and
> > > need not be resumed going forward.
> > 
> > Does this correctly handle PCI devices that aren't included in the ACPI
> > tables?  For example, add-on PCI cards?
> 
> Well, it would if it took the case when a device was configured for remote
> wakeup at run time but was not supposed to wakeup the system from sleep into
> account.
> 
> That needs to be checked in pci_dev_keep_suspended() rather than in the platform
> callback to cover all devices.
> 
> Updated patch follows.
> 
> Rafael

This looks better, thank you.

Alan Stern


--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Rafael J. Wysocki Jan. 15, 2015, 9:45 a.m. UTC | #2 
On Monday, January 12, 2015 11:39:18 AM Alan Stern wrote:
> On Mon, 12 Jan 2015, Rafael J. Wysocki wrote:
> 
> > On Thursday, January 08, 2015 10:51:06 AM Alan Stern wrote:
> > > On Thu, 8 Jan 2015, Rafael J. Wysocki wrote:
> > > 
> > > > From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > > > 
> > > > Commit f25c0ae2b4c4 (ACPI / PM: Avoid resuming devices in ACPI PM
> > > > domain during system suspend) modified the ACPI PM domain's system
> > > > suspend callbacks to allow devices attached to it to be left in the
> > > > runtime-suspended state during system suspend so as to optimize
> > > > the suspend process.
> > > > 
> > > > This was based on the general mechanism introduced by commit
> > > > aae4518b3124 (PM / sleep: Mechanism to avoid resuming runtime-suspended
> > > > devices unnecessarily).
> > > > 
> > > > Extend that approach to PCI devices by modifying the PCI bus type's
> > > > ->prepare callback to return 1 for devices that are runtime-suspended
> > > > when it is being executed and that are in a suitable power state and
> > > > need not be resumed going forward.
> > > 
> > > Does this correctly handle PCI devices that aren't included in the ACPI
> > > tables?  For example, add-on PCI cards?
> > 
> > Well, it would if it took the case when a device was configured for remote
> > wakeup at run time but was not supposed to wakeup the system from sleep into
> > account.
> > 
> > That needs to be checked in pci_dev_keep_suspended() rather than in the platform
> > callback to cover all devices.
> > 
> > Updated patch follows.
> > 
> > Rafael
> 
> This looks better, thank you.

Well, it still has the problem that the current ACPI wakeup setting may not be
suitable depending on the PM_QOS_FLAG_REMOTE_WAKEUP value which I overlooked
before.

I need to restore the check against adev->wakeup.prepare_count in
acpi_pci_need_resume() to cover that I think.

Will post a v2 with that fixed.
diff mbox

Patch

Index: linux-pm/drivers/pci/pci.c
===================================================================
--- linux-pm.orig/drivers/pci/pci.c
+++ linux-pm/drivers/pci/pci.c
@@ -521,6 +521,11 @@  static inline int platform_pci_run_wake(
 			pci_platform_pm->run_wake(dev, enable) : -ENODEV;
 }
 
+static inline bool platform_pci_need_resume(struct pci_dev *dev)
+{
+	return pci_platform_pm ? pci_platform_pm->need_resume(dev) : false;
+}
+
 /**
  * pci_raw_set_power_state - Use PCI PM registers to set the power state of
  *                           given PCI device
@@ -1999,6 +2004,27 @@  bool pci_dev_run_wake(struct pci_dev *de
 }
 EXPORT_SYMBOL_GPL(pci_dev_run_wake);
 
+/**
+ * pci_dev_keep_suspended - Check if the device can stay in the suspended state.
+ * @pci_dev: Device to check.
+ *
+ * Return 'true' if the device is runtime-suspended, it doesn't have to be
+ * reconfigured due to wakeup settings difference between system and runtime
+ * suspend and the current power state of it is suitable for the upcoming
+ * (system) transition.
+ */
+bool pci_dev_keep_suspended(struct pci_dev *pci_dev)
+{
+	struct device *dev = &pci_dev->dev;
+
+	if (!pm_runtime_suspended(dev)
+	    || (device_can_wakeup(dev) && !device_may_wakeup(dev))
+	    || platform_pci_need_resume(pci_dev))
+		return false;
+
+	return pci_target_state(dev) == dev->current_state;
+}
+
 void pci_config_pm_runtime_get(struct pci_dev *pdev)
 {
 	struct device *dev = &pdev->dev;
Index: linux-pm/drivers/pci/pci.h
===================================================================
--- linux-pm.orig/drivers/pci/pci.h
+++ linux-pm/drivers/pci/pci.h
@@ -50,6 +50,10 @@  int pci_probe_reset_function(struct pci_
  *		for given device (the device's wake-up capability has to be
  *		enabled by @sleep_wake for this feature to work)
  *
+ * @need_resume: returns 'true' if the given device (which is currently
+ *		suspended) needs to be resumed to be configured for system
+ *		wakeup.
+ *
  * If given platform is generally capable of power managing PCI devices, all of
  * these callbacks are mandatory.
  */
@@ -59,6 +63,7 @@  struct pci_platform_pm_ops {
 	pci_power_t (*choose_state)(struct pci_dev *dev);
 	int (*sleep_wake)(struct pci_dev *dev, bool enable);
 	int (*run_wake)(struct pci_dev *dev, bool enable);
+	bool (*need_resume)(struct pci_dev *dev);
 };
 
 int pci_set_platform_pm(struct pci_platform_pm_ops *ops);
@@ -67,6 +72,7 @@  void pci_power_up(struct pci_dev *dev);
 void pci_disable_enabled_device(struct pci_dev *dev);
 int pci_finish_runtime_suspend(struct pci_dev *dev);
 int __pci_pme_wakeup(struct pci_dev *dev, void *ign);
+bool pci_dev_keep_suspended(struct pci_dev *dev);
 void pci_config_pm_runtime_get(struct pci_dev *dev);
 void pci_config_pm_runtime_put(struct pci_dev *dev);
 void pci_pm_init(struct pci_dev *dev);
Index: linux-pm/drivers/pci/pci-acpi.c
===================================================================
--- linux-pm.orig/drivers/pci/pci-acpi.c
+++ linux-pm/drivers/pci/pci-acpi.c
@@ -501,12 +501,24 @@  static int acpi_pci_run_wake(struct pci_
 	return 0;
 }
 
+static bool acpi_pci_need_resume(struct pci_dev *dev)
+{
+	struct acpi_device *adev = ACPI_COMPANION(&dev->dev);
+
+	if (!adev || !acpi_device_power_manageable(adev)
+	    || acpi_target_system_state() == ACPI_STATE_S0)
+		return false;
+
+	return !!adev->power.flags.dsw_present;
+}
+
 static struct pci_platform_pm_ops acpi_pci_platform_pm = {
 	.is_manageable = acpi_pci_power_manageable,
 	.set_state = acpi_pci_set_power_state,
 	.choose_state = acpi_pci_choose_state,
 	.sleep_wake = acpi_pci_sleep_wake,
 	.run_wake = acpi_pci_run_wake,
+	.need_resume = acpi_pci_need_resume,
 };
 
 void acpi_pci_add_bus(struct pci_bus *bus)
Index: linux-pm/drivers/pci/pci-driver.c
===================================================================
--- linux-pm.orig/drivers/pci/pci-driver.c
+++ linux-pm/drivers/pci/pci-driver.c
@@ -653,7 +653,6 @@  static bool pci_has_legacy_pm_support(st
 static int pci_pm_prepare(struct device *dev)
 {
 	struct device_driver *drv = dev->driver;
-	int error = 0;
 
 	/*
 	 * Devices having power.ignore_children set may still be necessary for
@@ -662,10 +661,12 @@  static int pci_pm_prepare(struct device
 	if (dev->power.ignore_children)
 		pm_runtime_resume(dev);
 
-	if (drv && drv->pm && drv->pm->prepare)
-		error = drv->pm->prepare(dev);
-
-	return error;
+	if (drv && drv->pm && drv->pm->prepare) {
+		int error = drv->pm->prepare(dev);
+		if (error)
+			return error;
+	}
+	return pci_dev_keep_suspended(to_pci_dev(dev));
 }