[PATCH v12 0/4] Allow USB devices to remain runtime-suspended when sleeping

[PATCH v12 0/4] Allow USB devices to remain runtime-suspended when sleeping

[Tomeu Vizoso]

Hi,

this is v12 of an attempt to make it easier for devices to remain in
runtime PM when the system goes to sleep, mainly to reduce the time
spent resuming devices.

For this, we interpret the absence of all PM callback implementations as
it being safe to do direct_complete, so their ancestors aren't prevented
from remaining runtime-suspended.

Additionally, the prepare() callback of USB devices will return 1 if
runtime PM is enabled and the current wakeup settings are correct.

With these changes, a uvcvideo device (for example) stays in runtime
suspend when the system goes to sleep and is left in that state when the
system resumes, not delaying it unnecessarily.

Thanks,

Tomeu

Changes in v12:
- Include linux/pm_domain.h in vga_switcheroo.c for dev_pm_domain_set()

Changes in v11:
- Move calls to dev_pm_domain_set() out from &dev->power.lock as that
  isn't needed for dev->pm_domain.

Changes in v10:
- Remove superfluous call to pm_runtime_enabled() as suggested by Alan

Changes in v9:
- Add docs noting the need for the device lock to be held before calling
  device_is_bound()
- Add docs noting the need for the device lock to be held before calling
  dev_pm_domain_set()
- Move to CONFIG_PM_SLEEP as suggested by Rafael and Ulf.
- Rename from device_check_pm_callbacks to device_pm_check_callbacks to
  follow with the naming convention of existing API.
- Re-add calling to device_pm_check_callbacks from device registration
  and when updating the PM domain of a device.

Changes in v8:
- Add device_is_bound()
- Add dev_pm_domain_set() and update code to use it
- Move no_pm_callbacks field into CONFIG_PM_SLEEP
- Call device_check_pm_callbacks only after a device is bound or unbound

Changes in v7:
- Reduce indentation by adding a label in device_prepare()

Changes in v6:
- Add stub for !CONFIG_PM.
- Move implementation of device_check_pm_callbacks to power/main.c as it
  doesn't belong to CONFIG_PM_SLEEP.
- Take dev->power.lock before modifying flag.

Changes in v5:
- Check for all dev_pm_ops instances associated to a device, updating a
  no_pm_callbacks flag at the times when that could change.

Tomeu Vizoso (4):
  device core: add device_is_bound()
  PM / Domains: add setter for dev.pm_domain
  PM / sleep: Go direct_complete if driver has no callbacks
  USB / PM: Allow USB devices to remain runtime-suspended when sleeping

 arch/arm/mach-omap2/omap_device.c |  7 ++++---
 drivers/acpi/acpi_lpss.c          |  5 ++++-
 drivers/acpi/device_pm.c          |  5 +++--
 drivers/base/dd.c                 | 21 +++++++++++++++++++--
 drivers/base/power/clock_ops.c    |  5 +++--
 drivers/base/power/common.c       | 24 ++++++++++++++++++++++++
 drivers/base/power/domain.c       |  8 ++++++--
 drivers/base/power/main.c         | 35 +++++++++++++++++++++++++++++++++++
 drivers/base/power/power.h        |  3 +++
 drivers/gpu/vga/vga_switcheroo.c  | 11 ++++++-----
 drivers/misc/mei/pci-me.c         |  5 +++--
 drivers/misc/mei/pci-txe.c        |  5 +++--
 drivers/usb/core/port.c           |  6 ++++++
 drivers/usb/core/usb.c            |  8 +++++++-
 include/linux/device.h            |  2 ++
 include/linux/pm.h                |  1 +
 include/linux/pm_domain.h         |  3 +++
 17 files changed, 132 insertions(+), 22 deletions(-)

-- 
2.5.0

[PATCH v12 1/4] device core: add device_is_bound()

[Tomeu Vizoso]

Adds a function that tells whether a device is already bound to a
driver.

This is needed to warn when there is an attempt to change the PM domain
of a device that has finished probing already. The reason why we want to
enforce that is because in the general case that can cause problems and
also that we can simplify code quite a bit if we can always assume that.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---

Changes in v9:
- Add docs noting the need for the device lock to be held before calling
  device_is_bound()

Changes in v8:
- Add device_is_bound()

 drivers/base/dd.c      | 18 ++++++++++++++++--
 include/linux/device.h |  2 ++
 2 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/drivers/base/dd.c b/drivers/base/dd.c
index 7399be790b5d..13a0d66e5782 100644
--- a/drivers/base/dd.c
+++ b/drivers/base/dd.c
@@ -223,9 +223,23 @@ static int deferred_probe_initcall(void)
 }
 late_initcall(deferred_probe_initcall);
 
+/**
+ * device_is_bound() - Check if device is bound to a driver
+ * @dev: device to check
+ *
+ * Returns true if passed device has already finished probing successfully
+ * against a driver.
+ *
+ * This function must be called with the device lock held.
+ */
+bool device_is_bound(struct device *dev)
+{
+	return klist_node_attached(&dev->p->knode_driver);
+}
+
 static void driver_bound(struct device *dev)
 {
-	if (klist_node_attached(&dev->p->knode_driver)) {
+	if (device_is_bound(dev)) {
 		printk(KERN_WARNING "%s: device %s already bound\n",
 			__func__, kobject_name(&dev->kobj));
 		return;
@@ -601,7 +615,7 @@ static int __device_attach(struct device *dev, bool allow_async)
 
 	device_lock(dev);
 	if (dev->driver) {
-		if (klist_node_attached(&dev->p->knode_driver)) {
+		if (device_is_bound(dev)) {
 			ret = 1;
 			goto out_unlock;
 		}
diff --git a/include/linux/device.h b/include/linux/device.h
index f627ba20a46c..6d6f1fec092f 100644
--- a/include/linux/device.h
+++ b/include/linux/device.h
@@ -1044,6 +1044,8 @@ extern int __must_check driver_attach(struct device_driver *drv);
 extern void device_initial_probe(struct device *dev);
 extern int __must_check device_reprobe(struct device *dev);
 
+extern bool device_is_bound(struct device *dev);
+
 /*
  * Easy functions for dynamically creating devices on the fly
  */
-- 
2.5.0

[PATCH v12 2/4] PM / Domains: add setter for dev.pm_domain

[Tomeu Vizoso]

Adds a function that sets the pointer to dev_pm_domain in struct device
and that warns if the device has already finished probing. The reason
why we want to enforce that is because in the general case that can
cause problems and also that we can simplify code quite a bit if we can
always assume that.

This patch also changes all current code that directly sets the
dev.pm_domain pointer.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
---

Changes in v12:
- Include linux/pm_domain.h in vga_switcheroo.c for dev_pm_domain_set()

Changes in v11:
- Move calls to dev_pm_domain_set() out from &dev->power.lock as that
  isn't needed for dev->pm_domain.

Changes in v9:
- Add docs noting the need for the device lock to be held before calling
  dev_pm_domain_set()

Changes in v8:
- Add dev_pm_domain_set() and update code to use it

 arch/arm/mach-omap2/omap_device.c |  7 ++++---
 drivers/acpi/acpi_lpss.c          |  5 ++++-
 drivers/acpi/device_pm.c          |  5 +++--
 drivers/base/power/clock_ops.c    |  5 +++--
 drivers/base/power/common.c       | 21 +++++++++++++++++++++
 drivers/base/power/domain.c       |  6 ++++--
 drivers/gpu/vga/vga_switcheroo.c  | 11 ++++++-----
 drivers/misc/mei/pci-me.c         |  5 +++--
 drivers/misc/mei/pci-txe.c        |  5 +++--
 include/linux/pm_domain.h         |  3 +++
 10 files changed, 54 insertions(+), 19 deletions(-)

diff --git a/arch/arm/mach-omap2/omap_device.c b/arch/arm/mach-omap2/omap_device.c
index 3750ed14f8c5..0437537751bc 100644
--- a/arch/arm/mach-omap2/omap_device.c
+++ b/arch/arm/mach-omap2/omap_device.c
@@ -32,6 +32,7 @@
 #include <linux/io.h>
 #include <linux/clk.h>
 #include <linux/clkdev.h>
+#include <linux/pm_domain.h>
 #include <linux/pm_runtime.h>
 #include <linux/of.h>
 #include <linux/notifier.h>
@@ -168,7 +169,7 @@ static int omap_device_build_from_dt(struct platform_device *pdev)
 			r->name = dev_name(&pdev->dev);
 	}
 
-	pdev->dev.pm_domain = &omap_device_pm_domain;
+	dev_pm_domain_set(&pdev->dev, &omap_device_pm_domain);
 
 	if (device_active) {
 		omap_device_enable(pdev);
@@ -180,7 +181,7 @@ odbfd_exit1:
 odbfd_exit:
 	/* if data/we are at fault.. load up a fail handler */
 	if (ret)
-		pdev->dev.pm_domain = &omap_device_fail_pm_domain;
+		dev_pm_domain_set(&pdev->dev, &omap_device_fail_pm_domain);
 
 	return ret;
 }
@@ -701,7 +702,7 @@ int omap_device_register(struct platform_device *pdev)
 {
 	pr_debug("omap_device: %s: registering\n", pdev->name);
 
-	pdev->dev.pm_domain = &omap_device_pm_domain;
+	dev_pm_domain_set(&pdev->dev, &omap_device_pm_domain);
 	return platform_device_add(pdev);
 }
 
diff --git a/drivers/acpi/acpi_lpss.c b/drivers/acpi/acpi_lpss.c
index 047281a6ae11..c570b1d9f094 100644
--- a/drivers/acpi/acpi_lpss.c
+++ b/drivers/acpi/acpi_lpss.c
@@ -18,6 +18,7 @@
 #include <linux/mutex.h>
 #include <linux/platform_device.h>
 #include <linux/platform_data/clk-lpss.h>
+#include <linux/pm_domain.h>
 #include <linux/pm_runtime.h>
 #include <linux/delay.h>
 
@@ -875,13 +876,14 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
 
 	switch (action) {
 	case BUS_NOTIFY_BIND_DRIVER:
-		pdev->dev.pm_domain = &acpi_lpss_pm_domain;
+		dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
 		break;
 	case BUS_NOTIFY_DRIVER_NOT_BOUND:
 	case BUS_NOTIFY_UNBOUND_DRIVER:
 		pdev->dev.pm_domain = NULL;
 		break;
 	case BUS_NOTIFY_ADD_DEVICE:
+		dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
 		if (pdata->dev_desc->flags & LPSS_LTR)
 			return sysfs_create_group(&pdev->dev.kobj,
 						  &lpss_attr_group);
@@ -889,6 +891,7 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
 	case BUS_NOTIFY_DEL_DEVICE:
 		if (pdata->dev_desc->flags & LPSS_LTR)
 			sysfs_remove_group(&pdev->dev.kobj, &lpss_attr_group);
+		dev_pm_domain_set(&pdev->dev, NULL);
 		break;
 	default:
 		break;
diff --git a/drivers/acpi/device_pm.c b/drivers/acpi/device_pm.c
index 08a02cdc737c..cd2c3d6d40e0 100644
--- a/drivers/acpi/device_pm.c
+++ b/drivers/acpi/device_pm.c
@@ -22,6 +22,7 @@
 #include <linux/export.h>
 #include <linux/mutex.h>
 #include <linux/pm_qos.h>
+#include <linux/pm_domain.h>
 #include <linux/pm_runtime.h>
 
 #include "internal.h"
@@ -1059,7 +1060,7 @@ static void acpi_dev_pm_detach(struct device *dev, bool power_off)
 	struct acpi_device *adev = ACPI_COMPANION(dev);
 
 	if (adev && dev->pm_domain == &acpi_general_pm_domain) {
-		dev->pm_domain = NULL;
+		dev_pm_domain_set(dev, NULL);
 		acpi_remove_pm_notifier(adev);
 		if (power_off) {
 			/*
@@ -1111,7 +1112,7 @@ int acpi_dev_pm_attach(struct device *dev, bool power_on)
 		return -EBUSY;
 
 	acpi_add_pm_notifier(adev, dev, acpi_pm_notify_work_func);
-	dev->pm_domain = &acpi_general_pm_domain;
+	dev_pm_domain_set(dev, &acpi_general_pm_domain);
 	if (power_on) {
 		acpi_dev_pm_full_power(adev);
 		acpi_device_wakeup(adev, ACPI_STATE_S0, false);
diff --git a/drivers/base/power/clock_ops.c b/drivers/base/power/clock_ops.c
index 60ee5591ee8f..b36d01e0589d 100644
--- a/drivers/base/power/clock_ops.c
+++ b/drivers/base/power/clock_ops.c
@@ -15,6 +15,7 @@
 #include <linux/clkdev.h>
 #include <linux/slab.h>
 #include <linux/err.h>
+#include <linux/pm_domain.h>
 #include <linux/pm_runtime.h>
 
 #ifdef CONFIG_PM_CLK
@@ -348,7 +349,7 @@ static int pm_clk_notify(struct notifier_block *nb,
 		if (error)
 			break;
 
-		dev->pm_domain = clknb->pm_domain;
+		dev_pm_domain_set(dev, clknb->pm_domain);
 		if (clknb->con_ids[0]) {
 			for (con_id = clknb->con_ids; *con_id; con_id++)
 				pm_clk_add(dev, *con_id);
@@ -361,7 +362,7 @@ static int pm_clk_notify(struct notifier_block *nb,
 		if (dev->pm_domain != clknb->pm_domain)
 			break;
 
-		dev->pm_domain = NULL;
+		dev_pm_domain_set(dev, NULL);
 		pm_clk_destroy(dev);
 		break;
 	}
diff --git a/drivers/base/power/common.c b/drivers/base/power/common.c
index f48e33385b3e..02812bcabcac 100644
--- a/drivers/base/power/common.c
+++ b/drivers/base/power/common.c
@@ -128,3 +128,24 @@ void dev_pm_domain_detach(struct device *dev, bool power_off)
 		dev->pm_domain->detach(dev, power_off);
 }
 EXPORT_SYMBOL_GPL(dev_pm_domain_detach);
+
+/**
+ * dev_pm_domain_set - Set PM domain of a device.
+ * @dev: Device whose PM domain is to be set.
+ * @pd: PM domain to be set, or NULL.
+ *
+ * Sets the PM domain the device belongs to. The PM domain of a device needs
+ * to be set before its probe finishes (it's bound to a driver).
+ *
+ * This function must be called with the device lock held.
+ */
+void dev_pm_domain_set(struct device *dev, struct dev_pm_domain *pd)
+{
+	if (dev->pm_domain == pd)
+		return;
+
+	WARN(device_is_bound(dev),
+	     "PM domains can only be changed for unbound devices\n");
+	dev->pm_domain = pd;
+}
+EXPORT_SYMBOL_GPL(dev_pm_domain_set);
diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
index b80379012840..e8b526faec1f 100644
--- a/drivers/base/power/domain.c
+++ b/drivers/base/power/domain.c
@@ -1188,10 +1188,11 @@ static struct generic_pm_domain_data *genpd_alloc_dev_data(struct device *dev,
 	}
 
 	dev->power.subsys_data->domain_data = &gpd_data->base;
-	dev->pm_domain = &genpd->domain;
 
 	spin_unlock_irq(&dev->power.lock);
 
+	dev_pm_domain_set(dev, &genpd->domain);
+
 	return gpd_data;
 
  err_free:
@@ -1205,9 +1206,10 @@ static struct generic_pm_domain_data *genpd_alloc_dev_data(struct device *dev,
 static void genpd_free_dev_data(struct device *dev,
 				struct generic_pm_domain_data *gpd_data)
 {
+	dev_pm_domain_set(dev, NULL);
+
 	spin_lock_irq(&dev->power.lock);
 
-	dev->pm_domain = NULL;
 	dev->power.subsys_data->domain_data = NULL;
 
 	spin_unlock_irq(&dev->power.lock);
diff --git a/drivers/gpu/vga/vga_switcheroo.c b/drivers/gpu/vga/vga_switcheroo.c
index d64d9058bce5..665ab9fd0e01 100644
--- a/drivers/gpu/vga/vga_switcheroo.c
+++ b/drivers/gpu/vga/vga_switcheroo.c
@@ -36,6 +36,7 @@
 #include <linux/fs.h>
 #include <linux/module.h>
 #include <linux/pci.h>
+#include <linux/pm_domain.h>
 #include <linux/pm_runtime.h>
 #include <linux/seq_file.h>
 #include <linux/uaccess.h>
@@ -918,17 +919,17 @@ int vga_switcheroo_init_domain_pm_ops(struct device *dev,
 		domain->ops.runtime_suspend = vga_switcheroo_runtime_suspend;
 		domain->ops.runtime_resume = vga_switcheroo_runtime_resume;
 
-		dev->pm_domain = domain;
+		dev_pm_domain_set(dev, domain);
 		return 0;
 	}
-	dev->pm_domain = NULL;
+	dev_pm_domain_set(dev, NULL);
 	return -EINVAL;
 }
 EXPORT_SYMBOL(vga_switcheroo_init_domain_pm_ops);
 
 void vga_switcheroo_fini_domain_pm_ops(struct device *dev)
 {
-	dev->pm_domain = NULL;
+	dev_pm_domain_set(dev, NULL);
 }
 EXPORT_SYMBOL(vga_switcheroo_fini_domain_pm_ops);
 
@@ -989,10 +990,10 @@ vga_switcheroo_init_domain_pm_optimus_hdmi_audio(struct device *dev,
 		domain->ops.runtime_resume =
 			vga_switcheroo_runtime_resume_hdmi_audio;
 
-		dev->pm_domain = domain;
+		dev_pm_domain_set(dev, domain);
 		return 0;
 	}
-	dev->pm_domain = NULL;
+	dev_pm_domain_set(dev, NULL);
 	return -EINVAL;
 }
 EXPORT_SYMBOL(vga_switcheroo_init_domain_pm_optimus_hdmi_audio);
diff --git a/drivers/misc/mei/pci-me.c b/drivers/misc/mei/pci-me.c
index 27678d8154e0..75fc9c688df8 100644
--- a/drivers/misc/mei/pci-me.c
+++ b/drivers/misc/mei/pci-me.c
@@ -31,6 +31,7 @@
 #include <linux/jiffies.h>
 #include <linux/interrupt.h>
 
+#include <linux/pm_domain.h>
 #include <linux/pm_runtime.h>
 
 #include <linux/mei.h>
@@ -436,7 +437,7 @@ static inline void mei_me_set_pm_domain(struct mei_device *dev)
 		dev->pg_domain.ops.runtime_resume = mei_me_pm_runtime_resume;
 		dev->pg_domain.ops.runtime_idle = mei_me_pm_runtime_idle;
 
-		pdev->dev.pm_domain = &dev->pg_domain;
+		dev_pm_domain_set(&pdev->dev, &dev->pg_domain);
 	}
 }
 
@@ -448,7 +449,7 @@ static inline void mei_me_set_pm_domain(struct mei_device *dev)
 static inline void mei_me_unset_pm_domain(struct mei_device *dev)
 {
 	/* stop using pm callbacks if any */
-	dev->dev->pm_domain = NULL;
+	dev_pm_domain_set(dev->dev, NULL);
 }
 
 static const struct dev_pm_ops mei_me_pm_ops = {
diff --git a/drivers/misc/mei/pci-txe.c b/drivers/misc/mei/pci-txe.c
index 0882c0201907..71f8a7475717 100644
--- a/drivers/misc/mei/pci-txe.c
+++ b/drivers/misc/mei/pci-txe.c
@@ -27,6 +27,7 @@
 #include <linux/jiffies.h>
 #include <linux/interrupt.h>
 #include <linux/workqueue.h>
+#include <linux/pm_domain.h>
 #include <linux/pm_runtime.h>
 
 #include <linux/mei.h>
@@ -388,7 +389,7 @@ static inline void mei_txe_set_pm_domain(struct mei_device *dev)
 		dev->pg_domain.ops.runtime_resume = mei_txe_pm_runtime_resume;
 		dev->pg_domain.ops.runtime_idle = mei_txe_pm_runtime_idle;
 
-		pdev->dev.pm_domain = &dev->pg_domain;
+		dev_pm_domain_set(&pdev->dev, &dev->pg_domain);
 	}
 }
 
@@ -400,7 +401,7 @@ static inline void mei_txe_set_pm_domain(struct mei_device *dev)
 static inline void mei_txe_unset_pm_domain(struct mei_device *dev)
 {
 	/* stop using pm callbacks if any */
-	dev->dev->pm_domain = NULL;
+	dev_pm_domain_set(dev->dev, NULL);
 }
 
 static const struct dev_pm_ops mei_txe_pm_ops = {
diff --git a/include/linux/pm_domain.h b/include/linux/pm_domain.h
index ba4ced38efae..db21d3995f7e 100644
--- a/include/linux/pm_domain.h
+++ b/include/linux/pm_domain.h
@@ -240,12 +240,15 @@ static inline int of_genpd_add_provider_onecell(struct device_node *np,
 #ifdef CONFIG_PM
 extern int dev_pm_domain_attach(struct device *dev, bool power_on);
 extern void dev_pm_domain_detach(struct device *dev, bool power_off);
+extern void dev_pm_domain_set(struct device *dev, struct dev_pm_domain *pd);
 #else
 static inline int dev_pm_domain_attach(struct device *dev, bool power_on)
 {
 	return -ENODEV;
 }
 static inline void dev_pm_domain_detach(struct device *dev, bool power_off) {}
+static inline void dev_pm_domain_set(struct device *dev,
+				     struct dev_pm_domain *pd) {}
 #endif
 
 #endif /* _LINUX_PM_DOMAIN_H */
-- 
2.5.0

[PATCH v12 3/4] PM / sleep: Go direct_complete if driver has no callbacks

[Tomeu Vizoso]

If a suitable prepare callback cannot be found for a given device and
its driver has no PM callbacks at all, assume that it can go direct to
complete when the system goes to sleep.

The reason for this is that there's lots of devices in a system that do
no PM at all and there's no reason for them to prevent their ancestors
to do direct_complete if they can support it.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
---

Changes in v9:
- Move to CONFIG_PM_SLEEP as suggested by Rafael and Ulf.
- Rename from device_check_pm_callbacks to device_pm_check_callbacks to
  follow with the naming convention of existing API.
- Re-add calling to device_pm_check_callbacks from device registration
  and when updating the PM domain of a device.

Changes in v8:
- Move no_pm_callbacks field into CONFIG_PM_SLEEP
- Call device_check_pm_callbacks only after a device is bound or unbound

Changes in v7:
- Reduce indentation by adding a label in device_prepare()

Changes in v6:
- Add stub for !CONFIG_PM.
- Move implementation of device_check_pm_callbacks to power/main.c as it
  doesn't belong to CONFIG_PM_SLEEP.
- Take dev->power.lock before modifying flag.

Changes in v5:
- Check for all dev_pm_ops instances associated to a device, updating a
  no_pm_callbacks flag at the times when that could change.

 drivers/base/dd.c           |  3 +++
 drivers/base/power/common.c |  3 +++
 drivers/base/power/domain.c |  2 ++
 drivers/base/power/main.c   | 35 +++++++++++++++++++++++++++++++++++
 drivers/base/power/power.h  |  3 +++
 include/linux/pm.h          |  1 +
 6 files changed, 47 insertions(+)

diff --git a/drivers/base/dd.c b/drivers/base/dd.c
index 13a0d66e5782..049942176b00 100644
--- a/drivers/base/dd.c
+++ b/drivers/base/dd.c
@@ -250,6 +250,8 @@ static void driver_bound(struct device *dev)
 
 	klist_add_tail(&dev->p->knode_driver, &dev->driver->p->klist_devices);
 
+	device_pm_check_callbacks(dev);
+
 	/*
 	 * Make sure the device is no longer in one of the deferred lists and
 	 * kick off retrying all pending devices
@@ -766,6 +768,7 @@ static void __device_release_driver(struct device *dev)
 		pm_runtime_reinit(dev);
 
 		klist_remove(&dev->p->knode_driver);
+		device_pm_check_callbacks(dev);
 		if (dev->bus)
 			blocking_notifier_call_chain(&dev->bus->p->bus_notifier,
 						     BUS_NOTIFY_UNBOUND_DRIVER,
diff --git a/drivers/base/power/common.c b/drivers/base/power/common.c
index 02812bcabcac..93ed14cc2252 100644
--- a/drivers/base/power/common.c
+++ b/drivers/base/power/common.c
@@ -14,6 +14,8 @@
 #include <linux/acpi.h>
 #include <linux/pm_domain.h>
 
+#include "power.h"
+
 /**
  * dev_pm_get_subsys_data - Create or refcount power.subsys_data for device.
  * @dev: Device to handle.
@@ -147,5 +149,6 @@ void dev_pm_domain_set(struct device *dev, struct dev_pm_domain *pd)
 	WARN(device_is_bound(dev),
 	     "PM domains can only be changed for unbound devices\n");
 	dev->pm_domain = pd;
+	device_pm_check_callbacks(dev);
 }
 EXPORT_SYMBOL_GPL(dev_pm_domain_set);
diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
index e8b526faec1f..6ac9a7f33b64 100644
--- a/drivers/base/power/domain.c
+++ b/drivers/base/power/domain.c
@@ -20,6 +20,8 @@
 #include <linux/suspend.h>
 #include <linux/export.h>
 
+#include "power.h"
+
 #define GENPD_RETRY_MAX_MS	250		/* Approximate */
 
 #define GENPD_DEV_CALLBACK(genpd, type, callback, dev)		\
diff --git a/drivers/base/power/main.c b/drivers/base/power/main.c
index 9d626ac08d9c..6e7c3ccea24b 100644
--- a/drivers/base/power/main.c
+++ b/drivers/base/power/main.c
@@ -125,6 +125,7 @@ void device_pm_add(struct device *dev)
 {
 	pr_debug("PM: Adding info for %s:%s\n",
 		 dev->bus ? dev->bus->name : "No Bus", dev_name(dev));
+	device_pm_check_callbacks(dev);
 	mutex_lock(&dpm_list_mtx);
 	if (dev->parent && dev->parent->power.is_prepared)
 		dev_warn(dev, "parent %s should not be sleeping\n",
@@ -147,6 +148,7 @@ void device_pm_remove(struct device *dev)
 	mutex_unlock(&dpm_list_mtx);
 	device_wakeup_disable(dev);
 	pm_runtime_remove(dev);
+	device_pm_check_callbacks(dev);
 }
 
 /**
@@ -1572,6 +1574,11 @@ static int device_prepare(struct device *dev, pm_message_t state)
 
 	dev->power.wakeup_path = device_may_wakeup(dev);
 
+	if (dev->power.no_pm_callbacks) {
+		ret = 1;	/* Let device go direct_complete */
+		goto unlock;
+	}
+
 	if (dev->pm_domain) {
 		info = "preparing power domain ";
 		callback = dev->pm_domain->ops.prepare;
@@ -1594,6 +1601,7 @@ static int device_prepare(struct device *dev, pm_message_t state)
 	if (callback)
 		ret = callback(dev);
 
+unlock:
 	device_unlock(dev);
 
 	if (ret < 0) {
@@ -1736,3 +1744,30 @@ void dpm_for_each_dev(void *data, void (*fn)(struct device *, void *))
 	device_pm_unlock();
 }
 EXPORT_SYMBOL_GPL(dpm_for_each_dev);
+
+static bool pm_ops_is_empty(const struct dev_pm_ops *ops)
+{
+	if (!ops)
+		return true;
+
+	return !ops->prepare &&
+	       !ops->suspend &&
+	       !ops->suspend_late &&
+	       !ops->suspend_noirq &&
+	       !ops->resume_noirq &&
+	       !ops->resume_early &&
+	       !ops->resume &&
+	       !ops->complete;
+}
+
+void device_pm_check_callbacks(struct device *dev)
+{
+	spin_lock_irq(&dev->power.lock);
+	dev->power.no_pm_callbacks =
+		(!dev->bus || pm_ops_is_empty(dev->bus->pm)) &&
+		(!dev->class || pm_ops_is_empty(dev->class->pm)) &&
+		(!dev->type || pm_ops_is_empty(dev->type->pm)) &&
+		(!dev->pm_domain || pm_ops_is_empty(&dev->pm_domain->ops)) &&
+		(!dev->driver || pm_ops_is_empty(dev->driver->pm));
+	spin_unlock_irq(&dev->power.lock);
+}
diff --git a/drivers/base/power/power.h b/drivers/base/power/power.h
index 8b06193d4a5e..50e30e7b059d 100644
--- a/drivers/base/power/power.h
+++ b/drivers/base/power/power.h
@@ -125,6 +125,7 @@ extern void device_pm_remove(struct device *);
 extern void device_pm_move_before(struct device *, struct device *);
 extern void device_pm_move_after(struct device *, struct device *);
 extern void device_pm_move_last(struct device *);
+extern void device_pm_check_callbacks(struct device *dev);
 
 #else /* !CONFIG_PM_SLEEP */
 
@@ -143,6 +144,8 @@ static inline void device_pm_move_after(struct device *deva,
 					struct device *devb) {}
 static inline void device_pm_move_last(struct device *dev) {}
 
+static inline void device_pm_check_callbacks(struct device *dev) {}
+
 #endif /* !CONFIG_PM_SLEEP */
 
 static inline void device_pm_init(struct device *dev)
diff --git a/include/linux/pm.h b/include/linux/pm.h
index 528be6787796..6a5d654f4447 100644
--- a/include/linux/pm.h
+++ b/include/linux/pm.h
@@ -573,6 +573,7 @@ struct dev_pm_info {
 	struct wakeup_source	*wakeup;
 	bool			wakeup_path:1;
 	bool			syscore:1;
+	bool			no_pm_callbacks:1;	/* Owned by the PM core */
 #else
 	unsigned int		should_wakeup:1;
 #endif
-- 
2.5.0

[PATCH v12 4/4] USB / PM: Allow USB devices to remain runtime-suspended when sleeping

[Tomeu Vizoso]

Have dev_pm_ops.prepare return 1 for USB devices and ports so that USB
devices can remain runtime-suspended when the system goes to a sleep
state, if their wakeup state is correct and they have runtime PM enabled.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
---

Changes in v10:
- Remove superfluous call to pm_runtime_enabled() as suggested by Alan

 drivers/usb/core/port.c | 6 ++++++
 drivers/usb/core/usb.c  | 8 +++++++-
 2 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/drivers/usb/core/port.c b/drivers/usb/core/port.c
index 460c855be0d0..14718a9ffcfb 100644
--- a/drivers/usb/core/port.c
+++ b/drivers/usb/core/port.c
@@ -249,12 +249,18 @@ static int usb_port_runtime_suspend(struct device *dev)
 
 	return retval;
 }
+
+static int usb_port_prepare(struct device *dev)
+{
+	return 1;
+}
 #endif
 
 static const struct dev_pm_ops usb_port_pm_ops = {
 #ifdef CONFIG_PM
 	.runtime_suspend =	usb_port_runtime_suspend,
 	.runtime_resume =	usb_port_runtime_resume,
+	.prepare =		usb_port_prepare,
 #endif
 };
 
diff --git a/drivers/usb/core/usb.c b/drivers/usb/core/usb.c
index 77e4c9bc0ab1..ebb29caa3fe4 100644
--- a/drivers/usb/core/usb.c
+++ b/drivers/usb/core/usb.c
@@ -311,7 +311,13 @@ static int usb_dev_uevent(struct device *dev, struct kobj_uevent_env *env)
 
 static int usb_dev_prepare(struct device *dev)
 {
-	return 0;		/* Implement eventually? */
+	struct usb_device *udev = to_usb_device(dev);
+
+	/* Return 0 if the current wakeup setting is wrong, otherwise 1 */
+	if (udev->do_remote_wakeup != device_may_wakeup(dev))
+		return 0;
+
+	return 1;
 }
 
 static void usb_dev_complete(struct device *dev)
-- 
2.5.0

Re: [PATCH v12 0/4] Allow USB devices to remain runtime-suspended when sleeping

[Rafael J. Wysocki]

On Thursday, January 07, 2016 04:46:11 PM Tomeu Vizoso wrote:
> Hi,
> 
> this is v12 of an attempt to make it easier for devices to remain in
> runtime PM when the system goes to sleep, mainly to reduce the time
> spent resuming devices.
> 
> For this, we interpret the absence of all PM callback implementations as
> it being safe to do direct_complete, so their ancestors aren't prevented
> from remaining runtime-suspended.
> 
> Additionally, the prepare() callback of USB devices will return 1 if
> runtime PM is enabled and the current wakeup settings are correct.
> 
> With these changes, a uvcvideo device (for example) stays in runtime
> suspend when the system goes to sleep and is left in that state when the
> system resumes, not delaying it unnecessarily.
> 
> Thanks,
> 
> Tomeu
> 
> Changes in v12:
> - Include linux/pm_domain.h in vga_switcheroo.c for dev_pm_domain_set()
> 
> Changes in v11:
> - Move calls to dev_pm_domain_set() out from &dev->power.lock as that
>   isn't needed for dev->pm_domain.
> 
> Changes in v10:
> - Remove superfluous call to pm_runtime_enabled() as suggested by Alan
> 
> Changes in v9:
> - Add docs noting the need for the device lock to be held before calling
>   device_is_bound()
> - Add docs noting the need for the device lock to be held before calling
>   dev_pm_domain_set()
> - Move to CONFIG_PM_SLEEP as suggested by Rafael and Ulf.
> - Rename from device_check_pm_callbacks to device_pm_check_callbacks to
>   follow with the naming convention of existing API.
> - Re-add calling to device_pm_check_callbacks from device registration
>   and when updating the PM domain of a device.
> 
> Changes in v8:
> - Add device_is_bound()
> - Add dev_pm_domain_set() and update code to use it
> - Move no_pm_callbacks field into CONFIG_PM_SLEEP
> - Call device_check_pm_callbacks only after a device is bound or unbound
> 
> Changes in v7:
> - Reduce indentation by adding a label in device_prepare()
> 
> Changes in v6:
> - Add stub for !CONFIG_PM.
> - Move implementation of device_check_pm_callbacks to power/main.c as it
>   doesn't belong to CONFIG_PM_SLEEP.
> - Take dev->power.lock before modifying flag.
> 
> Changes in v5:
> - Check for all dev_pm_ops instances associated to a device, updating a
>   no_pm_callbacks flag at the times when that could change.
> 
> Tomeu Vizoso (4):
>   device core: add device_is_bound()
>   PM / Domains: add setter for dev.pm_domain
>   PM / sleep: Go direct_complete if driver has no callbacks
>   USB / PM: Allow USB devices to remain runtime-suspended when sleeping
> 
>  arch/arm/mach-omap2/omap_device.c |  7 ++++---
>  drivers/acpi/acpi_lpss.c          |  5 ++++-
>  drivers/acpi/device_pm.c          |  5 +++--
>  drivers/base/dd.c                 | 21 +++++++++++++++++++--
>  drivers/base/power/clock_ops.c    |  5 +++--
>  drivers/base/power/common.c       | 24 ++++++++++++++++++++++++
>  drivers/base/power/domain.c       |  8 ++++++--
>  drivers/base/power/main.c         | 35 +++++++++++++++++++++++++++++++++++
>  drivers/base/power/power.h        |  3 +++
>  drivers/gpu/vga/vga_switcheroo.c  | 11 ++++++-----
>  drivers/misc/mei/pci-me.c         |  5 +++--
>  drivers/misc/mei/pci-txe.c        |  5 +++--
>  drivers/usb/core/port.c           |  6 ++++++
>  drivers/usb/core/usb.c            |  8 +++++++-
>  include/linux/device.h            |  2 ++
>  include/linux/pm.h                |  1 +
>  include/linux/pm_domain.h         |  3 +++
>  17 files changed, 132 insertions(+), 22 deletions(-)

Series applied, thanks!


-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.

Re: [PATCH v12 2/4] PM / Domains: add setter for dev.pm_domain

[Andy Shevchenko]

On Thu, Jan 7, 2016 at 5:46 PM, Tomeu Vizoso <tomeu.vizoso@collabora.com> wrote:
> Adds a function that sets the pointer to dev_pm_domain in struct device
> and that warns if the device has already finished probing. The reason
> why we want to enforce that is because in the general case that can
> cause problems and also that we can simplify code quite a bit if we can
> always assume that.
>
> This patch also changes all current code that directly sets the
> dev.pm_domain pointer.


> --- a/drivers/acpi/acpi_lpss.c
> +++ b/drivers/acpi/acpi_lpss.c

> @@ -875,13 +876,14 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
>
>         switch (action) {
>         case BUS_NOTIFY_BIND_DRIVER:
> -               pdev->dev.pm_domain = &acpi_lpss_pm_domain;
> +               dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
>                 break;
>         case BUS_NOTIFY_DRIVER_NOT_BOUND:
>         case BUS_NOTIFY_UNBOUND_DRIVER:
>                 pdev->dev.pm_domain = NULL;

Missed?

>                 break;
>         case BUS_NOTIFY_ADD_DEVICE:
> +               dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
>                 if (pdata->dev_desc->flags & LPSS_LTR)
>                         return sysfs_create_group(&pdev->dev.kobj,
>                                                   &lpss_attr_group);
> @@ -889,6 +891,7 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
>         case BUS_NOTIFY_DEL_DEVICE:
>                 if (pdata->dev_desc->flags & LPSS_LTR)
>                         sysfs_remove_group(&pdev->dev.kobj, &lpss_attr_group);
> +               dev_pm_domain_set(&pdev->dev, NULL);
>                 break;
>         default:
>                 break;

This looks wrong. I didn't test yet, but I have concerns here. Why did
you add those calls?

Have you been aware about the commit b5f88dd1d6ef ("Revert "ACPI /
LPSS: allow to use specific PM domain during ->probe()") ?

Sorry for being too late with this.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH v12 2/4] PM / Domains: add setter for dev.pm_domain

[Andy Shevchenko]

On Fri, Jan 29, 2016 at 5:38 PM, Andy Shevchenko
<andy.shevchenko@gmail.com> wrote:
> On Thu, Jan 7, 2016 at 5:46 PM, Tomeu Vizoso <tomeu.vizoso@collabora.com> wrote:
>> Adds a function that sets the pointer to dev_pm_domain in struct device
>> and that warns if the device has already finished probing. The reason
>> why we want to enforce that is because in the general case that can
>> cause problems and also that we can simplify code quite a bit if we can
>> always assume that.
>>
>> This patch also changes all current code that directly sets the
>> dev.pm_domain pointer.
>
>
>> --- a/drivers/acpi/acpi_lpss.c
>> +++ b/drivers/acpi/acpi_lpss.c
>
>> @@ -875,13 +876,14 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
>>
>>         switch (action) {
>>         case BUS_NOTIFY_BIND_DRIVER:
>> -               pdev->dev.pm_domain = &acpi_lpss_pm_domain;
>> +               dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
>>                 break;
>>         case BUS_NOTIFY_DRIVER_NOT_BOUND:
>>         case BUS_NOTIFY_UNBOUND_DRIVER:
>>                 pdev->dev.pm_domain = NULL;
>
> Missed?
>
>>                 break;
>>         case BUS_NOTIFY_ADD_DEVICE:
>> +               dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
>>                 if (pdata->dev_desc->flags & LPSS_LTR)
>>                         return sysfs_create_group(&pdev->dev.kobj,
>>                                                   &lpss_attr_group);
>> @@ -889,6 +891,7 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
>>         case BUS_NOTIFY_DEL_DEVICE:
>>                 if (pdata->dev_desc->flags & LPSS_LTR)
>>                         sysfs_remove_group(&pdev->dev.kobj, &lpss_attr_group);
>> +               dev_pm_domain_set(&pdev->dev, NULL);
>>                 break;
>>         default:
>>                 break;
>
> This looks wrong. I didn't test yet, but I have concerns here. Why did
> you add those calls?


Okay, for first glance it seems working on Intel Braswell. But can
yoyu point me out what was going on before this patch?

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH v12 2/4] PM / Domains: add setter for dev.pm_domain

[Andy Shevchenko]

On Fri, Jan 29, 2016 at 5:51 PM, Andy Shevchenko
<andy.shevchenko@gmail.com> wrote:
> On Fri, Jan 29, 2016 at 5:38 PM, Andy Shevchenko
> <andy.shevchenko@gmail.com> wrote:
>> On Thu, Jan 7, 2016 at 5:46 PM, Tomeu Vizoso <tomeu.vizoso@collabora.com> wrote:
>>> Adds a function that sets the pointer to dev_pm_domain in struct device
>>> and that warns if the device has already finished probing. The reason
>>> why we want to enforce that is because in the general case that can
>>> cause problems and also that we can simplify code quite a bit if we can
>>> always assume that.
>>>
>>> This patch also changes all current code that directly sets the
>>> dev.pm_domain pointer.
>>
>>
>>> --- a/drivers/acpi/acpi_lpss.c
>>> +++ b/drivers/acpi/acpi_lpss.c
>>
>>> @@ -875,13 +876,14 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
>>>
>>>         switch (action) {
>>>         case BUS_NOTIFY_BIND_DRIVER:
>>> -               pdev->dev.pm_domain = &acpi_lpss_pm_domain;
>>> +               dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
>>>                 break;
>>>         case BUS_NOTIFY_DRIVER_NOT_BOUND:
>>>         case BUS_NOTIFY_UNBOUND_DRIVER:
>>>                 pdev->dev.pm_domain = NULL;
>>
>> Missed?
>>
>>>                 break;
>>>         case BUS_NOTIFY_ADD_DEVICE:
>>> +               dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
>>>                 if (pdata->dev_desc->flags & LPSS_LTR)
>>>                         return sysfs_create_group(&pdev->dev.kobj,
>>>                                                   &lpss_attr_group);
>>> @@ -889,6 +891,7 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
>>>         case BUS_NOTIFY_DEL_DEVICE:
>>>                 if (pdata->dev_desc->flags & LPSS_LTR)
>>>                         sysfs_remove_group(&pdev->dev.kobj, &lpss_attr_group);
>>> +               dev_pm_domain_set(&pdev->dev, NULL);
>>>                 break;
>>>         default:
>>>                 break;
>>
>> This looks wrong. I didn't test yet, but I have concerns here. Why did
>> you add those calls?
>
>
> Okay, for first glance it seems working on Intel Braswell.

One more test unveils this one

# modprobe -r sdhci-acpi
[ 1289.909441] ------------[ cut here ]------------
[ 1289.918205] WARNING: CPU: 1 PID: 4374 at
/home/andy/prj/linux-otc/drivers/base/power/common.c:150
dev_pm_domain_set+0x51/0x60()
[ 1289.934681] PM domains can only be changed for unbound devices
[ 1289.944843] Modules linked in: sdhci_acpi(-) sdhci mmc_core
led_class [last unloaded: dw_dmac_core]
[ 1289.958802] CPU: 1 PID: 4374 Comm: modprobe Not tainted 4.5.0-rc1+ #3
[ 1289.969736]  ffffffff81c38330 ffff88007bb53d18 ffffffff8133162f
ffff88007bb53d60
[ 1289.981844]  ffff88007bb53d50 ffffffff8105cd12 ffff88017a007410
0000000000000000
[ 1289.993996]  0000000000000001 0000000000000080 0000000000000000
ffff88007bb53db0
[ 1290.006123] Call Trace:
[ 1290.012600]  [<ffffffff8133162f>] dump_stack+0x44/0x55
[ 1290.022052]  [<ffffffff8105cd12>] warn_slowpath_common+0x82/0xc0
[ 1290.032462]  [<ffffffff8105cd9c>] warn_slowpath_fmt+0x4c/0x50
[ 1290.042589]  [<ffffffff814867c1>] dev_pm_domain_set+0x51/0x60
[ 1290.052695]  [<ffffffff813a62a9>] acpi_dev_pm_detach+0x3f/0x84
[ 1290.062910]  [<ffffffff81486747>] dev_pm_domain_detach+0x27/0x30
[ 1290.073294]  [<ffffffff81480188>] platform_drv_remove+0x38/0x40
[ 1290.083585]  [<ffffffff8147da51>] __device_release_driver+0xa1/0x160
[ 1290.094335]  [<ffffffff8147e5a6>] driver_detach+0xa6/0xb0
[ 1290.104009]  [<ffffffff8147d435>] bus_remove_driver+0x55/0xd0
[ 1290.114053]  [<ffffffff8147eccc>] driver_unregister+0x2c/0x50
[ 1290.124076]  [<ffffffff81480232>] platform_driver_unregister+0x12/0x20
[ 1290.134957]  [<ffffffffa003681c>] sdhci_acpi_driver_exit+0x10/0x12
[sdhci_acpi]
[ 1290.146730]  [<ffffffff810ce0e9>] SyS_delete_module+0x199/0x240
[ 1290.156907]  [<ffffffff8100219b>] ? exit_to_usermode_loop+0x8b/0x90
[ 1290.167478]  [<ffffffff81002b83>] do_syscall_32_irqs_off+0x53/0xa0
[ 1290.177949]  [<ffffffff818b7f9a>] entry_INT80_compat+0x2a/0x40
[ 1290.187986] ---[ end trace a8351c0506ee239b ]---
[ 1290.213855] mmc0: card 0001 removed

Which function is guilty?

> But can
> yoyu point me out what was going on before this patch?

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH v12 2/4] PM / Domains: add setter for dev.pm_domain

[Rafael J. Wysocki]

On Friday, January 29, 2016 06:35:49 PM Andy Shevchenko wrote:
> On Fri, Jan 29, 2016 at 5:51 PM, Andy Shevchenko
> <andy.shevchenko@gmail.com> wrote:
> > On Fri, Jan 29, 2016 at 5:38 PM, Andy Shevchenko
> > <andy.shevchenko@gmail.com> wrote:
> >> On Thu, Jan 7, 2016 at 5:46 PM, Tomeu Vizoso <tomeu.vizoso@collabora.com> wrote:
> >>> Adds a function that sets the pointer to dev_pm_domain in struct device
> >>> and that warns if the device has already finished probing. The reason
> >>> why we want to enforce that is because in the general case that can
> >>> cause problems and also that we can simplify code quite a bit if we can
> >>> always assume that.
> >>>
> >>> This patch also changes all current code that directly sets the
> >>> dev.pm_domain pointer.
> >>
> >>
> >>> --- a/drivers/acpi/acpi_lpss.c
> >>> +++ b/drivers/acpi/acpi_lpss.c
> >>
> >>> @@ -875,13 +876,14 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
> >>>
> >>>         switch (action) {
> >>>         case BUS_NOTIFY_BIND_DRIVER:
> >>> -               pdev->dev.pm_domain = &acpi_lpss_pm_domain;
> >>> +               dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
> >>>                 break;
> >>>         case BUS_NOTIFY_DRIVER_NOT_BOUND:
> >>>         case BUS_NOTIFY_UNBOUND_DRIVER:
> >>>                 pdev->dev.pm_domain = NULL;
> >>
> >> Missed?
> >>
> >>>                 break;
> >>>         case BUS_NOTIFY_ADD_DEVICE:
> >>> +               dev_pm_domain_set(&pdev->dev, &acpi_lpss_pm_domain);
> >>>                 if (pdata->dev_desc->flags & LPSS_LTR)
> >>>                         return sysfs_create_group(&pdev->dev.kobj,
> >>>                                                   &lpss_attr_group);
> >>> @@ -889,6 +891,7 @@ static int acpi_lpss_platform_notify(struct notifier_block *nb,
> >>>         case BUS_NOTIFY_DEL_DEVICE:
> >>>                 if (pdata->dev_desc->flags & LPSS_LTR)
> >>>                         sysfs_remove_group(&pdev->dev.kobj, &lpss_attr_group);
> >>> +               dev_pm_domain_set(&pdev->dev, NULL);
> >>>                 break;
> >>>         default:
> >>>                 break;
> >>
> >> This looks wrong. I didn't test yet, but I have concerns here. Why did
> >> you add those calls?
> >
> >
> > Okay, for first glance it seems working on Intel Braswell.
> 
> One more test unveils this one
> 
> # modprobe -r sdhci-acpi
> [ 1289.909441] ------------[ cut here ]------------
> [ 1289.918205] WARNING: CPU: 1 PID: 4374 at
> /home/andy/prj/linux-otc/drivers/base/power/common.c:150
> dev_pm_domain_set+0x51/0x60()
> [ 1289.934681] PM domains can only be changed for unbound devices
> [ 1289.944843] Modules linked in: sdhci_acpi(-) sdhci mmc_core
> led_class [last unloaded: dw_dmac_core]
> [ 1289.958802] CPU: 1 PID: 4374 Comm: modprobe Not tainted 4.5.0-rc1+ #3
> [ 1289.969736]  ffffffff81c38330 ffff88007bb53d18 ffffffff8133162f
> ffff88007bb53d60
> [ 1289.981844]  ffff88007bb53d50 ffffffff8105cd12 ffff88017a007410
> 0000000000000000
> [ 1289.993996]  0000000000000001 0000000000000080 0000000000000000
> ffff88007bb53db0
> [ 1290.006123] Call Trace:
> [ 1290.012600]  [<ffffffff8133162f>] dump_stack+0x44/0x55
> [ 1290.022052]  [<ffffffff8105cd12>] warn_slowpath_common+0x82/0xc0
> [ 1290.032462]  [<ffffffff8105cd9c>] warn_slowpath_fmt+0x4c/0x50
> [ 1290.042589]  [<ffffffff814867c1>] dev_pm_domain_set+0x51/0x60
> [ 1290.052695]  [<ffffffff813a62a9>] acpi_dev_pm_detach+0x3f/0x84
> [ 1290.062910]  [<ffffffff81486747>] dev_pm_domain_detach+0x27/0x30
> [ 1290.073294]  [<ffffffff81480188>] platform_drv_remove+0x38/0x40
> [ 1290.083585]  [<ffffffff8147da51>] __device_release_driver+0xa1/0x160
> [ 1290.094335]  [<ffffffff8147e5a6>] driver_detach+0xa6/0xb0
> [ 1290.104009]  [<ffffffff8147d435>] bus_remove_driver+0x55/0xd0
> [ 1290.114053]  [<ffffffff8147eccc>] driver_unregister+0x2c/0x50
> [ 1290.124076]  [<ffffffff81480232>] platform_driver_unregister+0x12/0x20
> [ 1290.134957]  [<ffffffffa003681c>] sdhci_acpi_driver_exit+0x10/0x12
> [sdhci_acpi]
> [ 1290.146730]  [<ffffffff810ce0e9>] SyS_delete_module+0x199/0x240
> [ 1290.156907]  [<ffffffff8100219b>] ? exit_to_usermode_loop+0x8b/0x90
> [ 1290.167478]  [<ffffffff81002b83>] do_syscall_32_irqs_off+0x53/0xa0
> [ 1290.177949]  [<ffffffff818b7f9a>] entry_INT80_compat+0x2a/0x40
> [ 1290.187986] ---[ end trace a8351c0506ee239b ]---
> [ 1290.213855] mmc0: card 0001 removed
> 
> Which function is guilty?

acpi_dev_pm_detach() calls dev_pm_domain_set() to clear the pm_domain pointer.

One might argue that this is happening too early.

Thanks,
Rafael

Re: [PATCH v12 2/4] PM / Domains: add setter for dev.pm_domain

[Valdis.Kletnieks]

On Fri, 29 Jan 2016 22:35:18 +0100, "Rafael J. Wysocki" said:
> > One more test unveils this one
> > 
> > # modprobe -r sdhci-acpi
> > [ 1289.909441] ------------[ cut here ]------------
> > [ 1289.918205] WARNING: CPU: 1 PID: 4374 at
> > /home/andy/prj/linux-otc/drivers/base/power/common.c:150
> > dev_pm_domain_set+0x51/0x60()
> > [ 1289.934681] PM domains can only be changed for unbound devices
> > [ 1289.944843] Modules linked in: sdhci_acpi(-) sdhci mmc_core
> > led_class [last unloaded: dw_dmac_core]
> > [ 1289.958802] CPU: 1 PID: 4374 Comm: modprobe Not tainted 4.5.0-rc1+ #3
> > [ 1289.969736]  ffffffff81c38330 ffff88007bb53d18 ffffffff8133162f
> > ffff88007bb53d60
> > [ 1289.981844]  ffff88007bb53d50 ffffffff8105cd12 ffff88017a007410
> > 0000000000000000
> > [ 1289.993996]  0000000000000001 0000000000000080 0000000000000000
> > ffff88007bb53db0
> > [ 1290.006123] Call Trace:
> > [ 1290.012600]  [<ffffffff8133162f>] dump_stack+0x44/0x55
> > [ 1290.022052]  [<ffffffff8105cd12>] warn_slowpath_common+0x82/0xc0
> > [ 1290.032462]  [<ffffffff8105cd9c>] warn_slowpath_fmt+0x4c/0x50
> > [ 1290.042589]  [<ffffffff814867c1>] dev_pm_domain_set+0x51/0x60

My Dell Latitude laptopi on next-20160201 is throwing a similar error
at shutdown, except the traceback continues:

mei_me_remove+0xbd/0xc0
pci_device_shutdown+0x32/0x50
device_shutdown+0xe2/0x210

(sorry, that's all I have, transcribing from one frame of video shot while
the system was shutting down....)

Re: [PATCH v12 2/4] PM / Domains: add setter for dev.pm_domain

[Rafael J. Wysocki]

On Mon, Feb 8, 2016 at 10:44 PM,  <Valdis.Kletnieks@vt.edu> wrote:
> On Fri, 29 Jan 2016 22:35:18 +0100, "Rafael J. Wysocki" said:
>> > One more test unveils this one
>> >
>> > # modprobe -r sdhci-acpi
>> > [ 1289.909441] ------------[ cut here ]------------
>> > [ 1289.918205] WARNING: CPU: 1 PID: 4374 at
>> > /home/andy/prj/linux-otc/drivers/base/power/common.c:150
>> > dev_pm_domain_set+0x51/0x60()
>> > [ 1289.934681] PM domains can only be changed for unbound devices
>> > [ 1289.944843] Modules linked in: sdhci_acpi(-) sdhci mmc_core
>> > led_class [last unloaded: dw_dmac_core]
>> > [ 1289.958802] CPU: 1 PID: 4374 Comm: modprobe Not tainted 4.5.0-rc1+ #3
>> > [ 1289.969736]  ffffffff81c38330 ffff88007bb53d18 ffffffff8133162f
>> > ffff88007bb53d60
>> > [ 1289.981844]  ffff88007bb53d50 ffffffff8105cd12 ffff88017a007410
>> > 0000000000000000
>> > [ 1289.993996]  0000000000000001 0000000000000080 0000000000000000
>> > ffff88007bb53db0
>> > [ 1290.006123] Call Trace:
>> > [ 1290.012600]  [<ffffffff8133162f>] dump_stack+0x44/0x55
>> > [ 1290.022052]  [<ffffffff8105cd12>] warn_slowpath_common+0x82/0xc0
>> > [ 1290.032462]  [<ffffffff8105cd9c>] warn_slowpath_fmt+0x4c/0x50
>> > [ 1290.042589]  [<ffffffff814867c1>] dev_pm_domain_set+0x51/0x60
>
> My Dell Latitude laptopi on next-20160201 is throwing a similar error
> at shutdown, except the traceback continues:
>
> mei_me_remove+0xbd/0xc0
> pci_device_shutdown+0x32/0x50
> device_shutdown+0xe2/0x210
>
> (sorry, that's all I have, transcribing from one frame of video shot while
> the system was shutting down....)

This should be fixed in 4.5-rc3, so please retest.

Thanks,
Rafael

Re: [PATCH v12 2/4] PM / Domains: add setter for dev.pm_domain

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 613 bytes --]

On Mon, 08 Feb 2016 22:50:39 +0100, "Rafael J. Wysocki" said:
> On Mon, Feb 8, 2016 at 10:44 PM,  <Valdis.Kletnieks@vt.edu> wrote:

> > My Dell Latitude laptopi on next-20160201 is throwing a similar error
> > at shutdown, except the traceback continues:
> >
> > mei_me_remove+0xbd/0xc0
> > pci_device_shutdown+0x32/0x50
> > device_shutdown+0xe2/0x210
> >
> > (sorry, that's all I have, transcribing from one frame of video shot while
> > the system was shutting down....)
>
> This should be fixed in 4.5-rc3, so please retest.

Confirming that next-20160201 has the problem, but it is fixed as of next-20160208.

[-- Attachment #2: Type: application/pgp-signature, Size: 848 bytes --]