Skip to contentSkip to conversation

PCI PM: More refinements of suspend-resume framework

PATCH In review Rafael J. Wysocki 5 participants 35 messages Feb 1, 2009 – Feb 10, 2009 PCI

lore
Share & tools
lore
#1 lore raw

PATCH 1/3 PCI PM: Do not disable and enable bridges during suspend-resume

In reply to Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

It is a mistake to disable and enable PCI bridges and PCI Express ports during suspend-resume (at least at the time when it is currently done). Disabling them may lead to problems with accessing devices behind them and they should be automatically enabled when their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

---
 drivers/pci/pci-driver.c |   10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/pci/pci-driver.c +6 -4
Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }
 
 static void pci_pm_default_suspend_generic(struct pci_dev *pci_dev)
 {
-	/* If device is enabled at this point, disable it */
-	pci_disable_enabled_device(pci_dev);
+	/* If a non-bridge device is enabled at this point, disable it */
+	if (!pci_is_bridge(pci_dev))
+		pci_disable_enabled_device(pci_dev);
 	/*
 	 * Save state with interrupts enabled, because in principle the bus the
 	 * device is on may be put into a low power state after this code runs.
#15 lore raw

PATCH 2/3 PCI PM: Let the core be more careful with respect to drivers using new framework

In reply to Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attepmt to do anything else to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices without drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

For all devices the core will disable wake-up during resume.

---
 drivers/pci/pci-driver.c |  138 ++++++++++++++++++++++++++++-------------------
 1 file changed, 85 insertions(+), 53 deletions(-)
drivers/pci/pci-driver.c +85 -53
Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================
@@ -370,7 +370,6 @@ static int pci_legacy_suspend(struct dev
 	}
 
 	pci_save_state(pci_dev);
-	pci_dev->state_saved = true;
 	/*
 	 * This is for compatibility with existing code with legacy PM support.
 	 */
@@ -424,39 +423,22 @@ static void pci_pm_default_resume_noirq(
 	pci_fixup_device(pci_fixup_resume_early, pci_dev);
 }
 
-static int pci_pm_default_resume(struct pci_dev *pci_dev)
+static void pci_pm_default_resume(struct pci_dev *pci_dev)
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (pci_is_bridge(pci_dev))
-		return 0;
-
-	pci_enable_wake(pci_dev, PCI_D0, false);
-	return pci_pm_reenable_device(pci_dev);
+	if (!pci_is_bridge(pci_dev))
+		pci_enable_wake(pci_dev, PCI_D0, false);
 }
 
-static void pci_pm_default_suspend_generic(struct pci_dev *pci_dev)
+static void pci_pm_default_suspend(struct pci_dev *pci_dev)
 {
-	/* If a non-bridge device is enabled at this point, disable it */
+	/* Disable non-bridge devices without PM support */
 	if (!pci_is_bridge(pci_dev))
 		pci_disable_enabled_device(pci_dev);
-	/*
-	 * Save state with interrupts enabled, because in principle the bus the
-	 * device is on may be put into a low power state after this code runs.
-	 */
 	pci_save_state(pci_dev);
 }
 
-static void pci_pm_default_suspend(struct pci_dev *pci_dev, bool prepare)
-{
-	pci_pm_default_suspend_generic(pci_dev);
-
-	if (prepare && !pci_is_bridge(pci_dev))
-		pci_prepare_to_sleep(pci_dev);
-
-	pci_fixup_device(pci_fixup_suspend, pci_dev);
-}
-
 static bool pci_has_legacy_pm_support(struct pci_dev *pci_dev)
 {
 	struct pci_driver *drv = pci_dev->driver;
@@ -500,20 +482,40 @@ static int pci_pm_suspend(struct device 
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
-	int error = 0;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_SUSPEND);
 
-	if (pm && pm->suspend) {
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		goto Fixup;
+	}
+
+	pci_dev->state_saved = false;
+
+	if (pm->suspend) {
+		int error;
+
 		error = pm->suspend(dev);
 		suspend_report_result(pm->suspend, error);
+		if (error)
+			return error;
+		if (!pci_dev->state_saved)
+			WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: State of device not saved by %pF\n",
+				pm->suspend);
 	}
 
-	if (!error)
-		pci_pm_default_suspend(pci_dev, !!pm);
+	if (!pci_dev->state_saved) {
+		pci_save_state(pci_dev);
+		if (!pci_is_bridge(pci_dev))
+			pci_prepare_to_sleep(pci_dev);
+	}
 
-	return error;
+ Fixup:
+	pci_fixup_device(pci_fixup_suspend, pci_dev);
+
+	return 0;
 }
 
 static int pci_pm_suspend_noirq(struct device *dev)
@@ -556,7 +558,7 @@ static int pci_pm_resume_noirq(struct de
 static int pci_pm_resume(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	/*
@@ -569,12 +571,16 @@ static int pci_pm_resume(struct device *
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	error = pci_pm_default_resume(pci_dev);
+	pci_pm_default_resume(pci_dev);
 
-	if (!error && drv && drv->pm && drv->pm->resume)
-		error = drv->pm->resume(dev);
+	if (pm) {
+		if (pm->resume)
+			error = pm->resume(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
-	return error;
+	return 0;
 }
 
 #else /* !CONFIG_SUSPEND */
@@ -591,21 +597,31 @@ static int pci_pm_resume(struct device *
 static int pci_pm_freeze(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
-	int error = 0;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_FREEZE);
 
-	if (drv && drv->pm && drv->pm->freeze) {
-		error = drv->pm->freeze(dev);
-		suspend_report_result(drv->pm->freeze, error);
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		return 0;
 	}
 
-	if (!error)
-		pci_pm_default_suspend_generic(pci_dev);
+	pci_dev->state_saved = false;
 
-	return error;
+	if (pm->freeze) {
+		int error;
+
+		error = pm->freeze(dev);
+		suspend_report_result(pm->freeze, error);
+		if (error)
+			return error;
+	}
+
+	if (!pci_dev->state_saved)
+		pci_save_state(pci_dev);
+
+	return 0;
 }
 
 static int pci_pm_freeze_noirq(struct device *dev)
@@ -648,16 +664,18 @@ static int pci_pm_thaw_noirq(struct devi
 static int pci_pm_thaw(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	pci_pm_reenable_device(pci_dev);
-
-	if (drv && drv->pm && drv->pm->thaw)
-		error =  drv->pm->thaw(dev);
+	if (pm) {
+		if (pm->thaw)
+			error = pm->thaw(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
 	return error;
 }
@@ -671,13 +689,23 @@ static int pci_pm_poweroff(struct device
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_HIBERNATE);
 
-	if (pm && pm->poweroff) {
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		goto Fixup;
+	}
+
+	pci_dev->state_saved = false;
+
+	if (pm->poweroff) {
 		error = pm->poweroff(dev);
 		suspend_report_result(pm->poweroff, error);
 	}
 
-	if (!error)
-		pci_pm_default_suspend(pci_dev, !!pm);
+	if (!pci_dev->state_saved && !pci_is_bridge(pci_dev))
+		pci_prepare_to_sleep(pci_dev);
+
+ Fixup:
+	pci_fixup_device(pci_fixup_suspend, pci_dev);
 
 	return error;
 }
@@ -718,7 +746,7 @@ static int pci_pm_restore_noirq(struct d
 static int pci_pm_restore(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	/*
@@ -731,10 +759,14 @@ static int pci_pm_restore(struct device 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	error = pci_pm_default_resume(pci_dev);
+	pci_pm_default_resume(pci_dev);
 
-	if (!error && drv && drv->pm && drv->pm->restore)
-		error = drv->pm->restore(dev);
+	if (pm) {
+		if (pm->restore)
+			error = pm->restore(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
 	return error;
 }
#23 lore raw

PATCH 3/3 PCI PM: Make warning in pci_legacy_suspend more useful

In reply to Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

The warning in pci_legacy_suspend() would be much more useful if it printed the name of the function that did the wrong thing. Make it do so.

---
 drivers/pci/pci-driver.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/pci/pci-driver.c +3 -1
Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================
@@ -365,7 +365,9 @@ static int pci_legacy_suspend(struct dev
 		if (pci_dev->state_saved)
 			goto Fixup;
 
-		if (WARN_ON_ONCE(pci_dev->current_state != PCI_D0))
+		if (WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: Device state not saved by %pF\n",
+				drv->suspend))
 			goto Fixup;
 	}
 
#0 lore raw

cover PCI PM: More refinements of suspend-resume framework

Hi,

After the recent discussion with Linus, I have the following three fixes of the PCI PM framework.

The first patch fixes the bug that bridges (and PCIe ports) are disabled during suspend, althouth they shouldn't.

The second one makes the PCI PM core handle devices more carefully (details in the changelog). [Note to Linus: devices are still put into low power states with interrupts on after this patch. Moving that to the late suspend phase will be the next step.]

The last patch makes the warning in pci_legacy_suspend() more useful.

Thanks,
Rafael

#24 lore raw

PATCH 3/3 PCI PM: Make warning in pci_legacy_suspend more useful

From: Rafael J. Wysocki <[email protected]>

The warning in pci_legacy_suspend() would be much more useful if it printed the name of the function that did the wrong thing. Make it do so.

---
 drivers/pci/pci-driver.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/pci/pci-driver.c +3 -1
Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================
@@ -365,7 +365,9 @@ static int pci_legacy_suspend(struct dev
 		if (pci_dev->state_saved)
 			goto Fixup;
 
-		if (WARN_ON_ONCE(pci_dev->current_state != PCI_D0))
+		if (WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: Device state not saved by %pF\n",
+				drv->suspend))
 			goto Fixup;
 	}
 
#2 lore raw

PATCH 1/3 PCI PM: Do not disable and enable bridges during suspend-resume

In reply to Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

It is a mistake to disable and enable PCI bridges and PCI Express ports during suspend-resume (at least at the time when it is currently done). Disabling them may lead to problems with accessing devices behind them and they should be automatically enabled when their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

---
 drivers/pci/pci-driver.c |   10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/pci/pci-driver.c +6 -4
Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }
 
 static void pci_pm_default_suspend_generic(struct pci_dev *pci_dev)
 {
-	/* If device is enabled at this point, disable it */
-	pci_disable_enabled_device(pci_dev);
+	/* If a non-bridge device is enabled at this point, disable it */
+	if (!pci_is_bridge(pci_dev))
+		pci_disable_enabled_device(pci_dev);
 	/*
 	 * Save state with interrupts enabled, because in principle the bus the
 	 * device is on may be put into a low power state after this code runs.
#16 lore raw

PATCH 2/3 PCI PM: Let the core be more careful with respect to drivers using new framework

In reply to Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attepmt to do anything else to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices without drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

For all devices the core will disable wake-up during resume.

---
 drivers/pci/pci-driver.c |  138 ++++++++++++++++++++++++++++-------------------
 1 file changed, 85 insertions(+), 53 deletions(-)
drivers/pci/pci-driver.c +85 -53
Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================
@@ -370,7 +370,6 @@ static int pci_legacy_suspend(struct dev
 	}
 
 	pci_save_state(pci_dev);
-	pci_dev->state_saved = true;
 	/*
 	 * This is for compatibility with existing code with legacy PM support.
 	 */
@@ -424,39 +423,22 @@ static void pci_pm_default_resume_noirq(
 	pci_fixup_device(pci_fixup_resume_early, pci_dev);
 }
 
-static int pci_pm_default_resume(struct pci_dev *pci_dev)
+static void pci_pm_default_resume(struct pci_dev *pci_dev)
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (pci_is_bridge(pci_dev))
-		return 0;
-
-	pci_enable_wake(pci_dev, PCI_D0, false);
-	return pci_pm_reenable_device(pci_dev);
+	if (!pci_is_bridge(pci_dev))
+		pci_enable_wake(pci_dev, PCI_D0, false);
 }
 
-static void pci_pm_default_suspend_generic(struct pci_dev *pci_dev)
+static void pci_pm_default_suspend(struct pci_dev *pci_dev)
 {
-	/* If a non-bridge device is enabled at this point, disable it */
+	/* Disable non-bridge devices without PM support */
 	if (!pci_is_bridge(pci_dev))
 		pci_disable_enabled_device(pci_dev);
-	/*
-	 * Save state with interrupts enabled, because in principle the bus the
-	 * device is on may be put into a low power state after this code runs.
-	 */
 	pci_save_state(pci_dev);
 }
 
-static void pci_pm_default_suspend(struct pci_dev *pci_dev, bool prepare)
-{
-	pci_pm_default_suspend_generic(pci_dev);
-
-	if (prepare && !pci_is_bridge(pci_dev))
-		pci_prepare_to_sleep(pci_dev);
-
-	pci_fixup_device(pci_fixup_suspend, pci_dev);
-}
-
 static bool pci_has_legacy_pm_support(struct pci_dev *pci_dev)
 {
 	struct pci_driver *drv = pci_dev->driver;
@@ -500,20 +482,40 @@ static int pci_pm_suspend(struct device 
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
-	int error = 0;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_SUSPEND);
 
-	if (pm && pm->suspend) {
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		goto Fixup;
+	}
+
+	pci_dev->state_saved = false;
+
+	if (pm->suspend) {
+		int error;
+
 		error = pm->suspend(dev);
 		suspend_report_result(pm->suspend, error);
+		if (error)
+			return error;
+		if (!pci_dev->state_saved)
+			WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: State of device not saved by %pF\n",
+				pm->suspend);
 	}
 
-	if (!error)
-		pci_pm_default_suspend(pci_dev, !!pm);
+	if (!pci_dev->state_saved) {
+		pci_save_state(pci_dev);
+		if (!pci_is_bridge(pci_dev))
+			pci_prepare_to_sleep(pci_dev);
+	}
 
-	return error;
+ Fixup:
+	pci_fixup_device(pci_fixup_suspend, pci_dev);
+
+	return 0;
 }
 
 static int pci_pm_suspend_noirq(struct device *dev)
@@ -556,7 +558,7 @@ static int pci_pm_resume_noirq(struct de
 static int pci_pm_resume(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	/*
@@ -569,12 +571,16 @@ static int pci_pm_resume(struct device *
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	error = pci_pm_default_resume(pci_dev);
+	pci_pm_default_resume(pci_dev);
 
-	if (!error && drv && drv->pm && drv->pm->resume)
-		error = drv->pm->resume(dev);
+	if (pm) {
+		if (pm->resume)
+			error = pm->resume(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
-	return error;
+	return 0;
 }
 
 #else /* !CONFIG_SUSPEND */
@@ -591,21 +597,31 @@ static int pci_pm_resume(struct device *
 static int pci_pm_freeze(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
-	int error = 0;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_FREEZE);
 
-	if (drv && drv->pm && drv->pm->freeze) {
-		error = drv->pm->freeze(dev);
-		suspend_report_result(drv->pm->freeze, error);
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		return 0;
 	}
 
-	if (!error)
-		pci_pm_default_suspend_generic(pci_dev);
+	pci_dev->state_saved = false;
 
-	return error;
+	if (pm->freeze) {
+		int error;
+
+		error = pm->freeze(dev);
+		suspend_report_result(pm->freeze, error);
+		if (error)
+			return error;
+	}
+
+	if (!pci_dev->state_saved)
+		pci_save_state(pci_dev);
+
+	return 0;
 }
 
 static int pci_pm_freeze_noirq(struct device *dev)
@@ -648,16 +664,18 @@ static int pci_pm_thaw_noirq(struct devi
 static int pci_pm_thaw(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	pci_pm_reenable_device(pci_dev);
-
-	if (drv && drv->pm && drv->pm->thaw)
-		error =  drv->pm->thaw(dev);
+	if (pm) {
+		if (pm->thaw)
+			error = pm->thaw(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
 	return error;
 }
@@ -671,13 +689,23 @@ static int pci_pm_poweroff(struct device
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_HIBERNATE);
 
-	if (pm && pm->poweroff) {
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		goto Fixup;
+	}
+
+	pci_dev->state_saved = false;
+
+	if (pm->poweroff) {
 		error = pm->poweroff(dev);
 		suspend_report_result(pm->poweroff, error);
 	}
 
-	if (!error)
-		pci_pm_default_suspend(pci_dev, !!pm);
+	if (!pci_dev->state_saved && !pci_is_bridge(pci_dev))
+		pci_prepare_to_sleep(pci_dev);
+
+ Fixup:
+	pci_fixup_device(pci_fixup_suspend, pci_dev);
 
 	return error;
 }
@@ -718,7 +746,7 @@ static int pci_pm_restore_noirq(struct d
 static int pci_pm_restore(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	/*
@@ -731,10 +759,14 @@ static int pci_pm_restore(struct device 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	error = pci_pm_default_resume(pci_dev);
+	pci_pm_default_resume(pci_dev);
 
-	if (!error && drv && drv->pm && drv->pm->restore)
-		error = drv->pm->restore(dev);
+	if (pm) {
+		if (pm->restore)
+			error = pm->restore(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
 	return error;
 }
#17 lore raw

Re: [linux-pm] [PATCH 2/3] PCI PM: Let the core be more careful with respect to drivers using new framework

Hi Rafael.

Just a couple of typos in the comment:

Quoted 18 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

Quoted 3 lines

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attepmt to do anything else

s/attepmt/attempt/

Quoted 8 lines

to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices without drivers the PCI PM core will only save

s/without/with/

Quoted 11 lines

their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

Quoted 3 lines

For all devices the core will disable wake-up during resume.

I'm not pretending to understand the gory details of the code itself, and so don't want you to add a Reviewed-by or such like for me, thanks.

Regards,

Nigel

#22 lore raw

Re: [PATCH 2/3] PCI PM: Let the core be more careful with respect to drivers using new framework

In reply to Rafael J. Wysocki

Hi Rafael.

Just a couple of typos in the comment:

Quoted 18 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

Quoted 3 lines

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attepmt to do anything else

s/attepmt/attempt/

Quoted 8 lines

to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices without drivers the PCI PM core will only save

s/without/with/

Quoted 11 lines

their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

Quoted 3 lines

For all devices the core will disable wake-up during resume.

I'm not pretending to understand the gory details of the code itself, and so don't want you to add a Reviewed-by or such like for me, thanks.

Regards,

Nigel

#25 lore raw

Re: [linux-pm] [PATCH 3/3] PCI PM: Make warning in pci_legacy_suspend more useful

In reply to Rafael J. Wysocki

Hi again.

Quoted 16 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

The warning in pci_legacy_suspend() would be much more useful if it printed the name of the function that did the wrong thing. Make it do so.

---
 drivers/pci/pci-driver.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================

drivers/pci/pci-driver.c static int pci_legacy_suspend(struct dev · line 372

Quoted 9 lines
drivers/pci/pci-driver.c +3 -1
@@ -365,7 +365,9 @@ static int pci_legacy_suspend(struct dev
 		if (pci_dev->state_saved)
 			goto Fixup;
 
-		if (WARN_ON_ONCE(pci_dev->current_state != PCI_D0))
+		if (WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: Device state not saved by %pF\n",
+				drv->suspend))
 			goto Fixup;
 	}

Am I right in thinking that WARN_ONCE will only warn about the first driver that has the problem? If so, wouldn't it be better to make it warn about all drivers that have the problem, but only once per device?

Regards,

Nigel

#28 lore raw

Re: [PATCH 3/3] PCI PM: Make warning in pci_legacy_suspend more useful

In reply to Rafael J. Wysocki

Hi again.

Quoted 16 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

The warning in pci_legacy_suspend() would be much more useful if it printed the name of the function that did the wrong thing. Make it do so.

---
 drivers/pci/pci-driver.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================

drivers/pci/pci-driver.c static int pci_legacy_suspend(struct dev · line 372

Quoted 9 lines
drivers/pci/pci-driver.c +3 -1
@@ -365,7 +365,9 @@ static int pci_legacy_suspend(struct dev
 		if (pci_dev->state_saved)
 			goto Fixup;
 
-		if (WARN_ON_ONCE(pci_dev->current_state != PCI_D0))
+		if (WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: Device state not saved by %pF\n",
+				drv->suspend))
 			goto Fixup;
 	}

Am I right in thinking that WARN_ONCE will only warn about the first driver that has the problem? If so, wouldn't it be better to make it warn about all drivers that have the problem, but only once per device?

Regards,

Nigel

#26 lore raw

Re: [PATCH 3/3] PCI PM: Make warning in pci_legacy_suspend more useful

In reply to Nigel Cunningham
Quoted 22 lines · Nigel Cunningham

Hi again.

Quoted 19 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

The warning in pci_legacy_suspend() would be much more useful if it printed the name of the function that did the wrong thing. Make it do so.

---
 drivers/pci/pci-driver.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================

drivers/pci/pci-driver.c +0 -0
@@ -365,7 +365,9 @@ static int pci_legacy_suspend(struct dev
 		if (pci_dev->state_saved)
 			goto Fixup;
 
Quoted 10 lines

drivers/pci/pci-driver.c static int pci_legacy_suspend(struct dev · line 372

Quoted 6 lines
drivers/pci/pci-driver.c +3 -1
-		if (WARN_ON_ONCE(pci_dev->current_state != PCI_D0))
+		if (WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: Device state not saved by %pF\n",
+				drv->suspend))
 			goto Fixup;
 	}

Am I right in thinking that WARN_ONCE will only warn about the first driver that has the problem? If so, wouldn't it be better to make it warn about all drivers that have the problem, but only once per device?

That would be better, but I don't think we are going to see many of these anyway. :-)

Thanks,
Rafael

#27 lore raw

Re: [linux-pm] [PATCH 3/3] PCI PM: Make warning in pci_legacy_suspend more useful

In reply to Nigel Cunningham
Quoted 22 lines · Nigel Cunningham

Hi again.

Quoted 19 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

The warning in pci_legacy_suspend() would be much more useful if it printed the name of the function that did the wrong thing. Make it do so.

---
 drivers/pci/pci-driver.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================

drivers/pci/pci-driver.c +0 -0
@@ -365,7 +365,9 @@ static int pci_legacy_suspend(struct dev
 		if (pci_dev->state_saved)
 			goto Fixup;
 
Quoted 10 lines

drivers/pci/pci-driver.c static int pci_legacy_suspend(struct dev · line 372

Quoted 6 lines
drivers/pci/pci-driver.c +3 -1
-		if (WARN_ON_ONCE(pci_dev->current_state != PCI_D0))
+		if (WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: Device state not saved by %pF\n",
+				drv->suspend))
 			goto Fixup;
 	}

Am I right in thinking that WARN_ONCE will only warn about the first driver that has the problem? If so, wouldn't it be better to make it warn about all drivers that have the problem, but only once per device?

That would be better, but I don't think we are going to see many of these anyway. :-)

Thanks,
Rafael

#18 lore raw

Re: [PATCH 2/3] PCI PM: Let the core be more careful with respect to drivers using new framework

In reply to Nigel Cunningham
Quoted 23 lines · Nigel Cunningham

Hi Rafael.

Just a couple of typos in the comment:

Quoted 18 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

Quoted 5 lines
Quoted 3 lines

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attepmt to do anything else

s/attepmt/attempt/

OK

Quoted 10 lines
Quoted 8 lines

to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices without drivers the PCI PM core will only save

s/without/with/

Ouch, thanks.

Quoted 11 lines
Quoted 11 lines

their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

Quoted 6 lines
Quoted 3 lines

For all devices the core will disable wake-up during resume.

I'm not pretending to understand the gory details of the code itself, and so don't want you to add a Reviewed-by or such like for me, thanks.

Thanks for the fixes, I'll resend the patch with updated changelog.

Best,
Rafael

#19 lore raw

Re: [linux-pm] [PATCH 2/3] PCI PM: Let the core be more careful with respect to drivers using new framework

In reply to Nigel Cunningham
Quoted 23 lines · Nigel Cunningham

Hi Rafael.

Just a couple of typos in the comment:

Quoted 18 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

Quoted 5 lines
Quoted 3 lines

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attepmt to do anything else

s/attepmt/attempt/

OK

Quoted 10 lines
Quoted 8 lines

to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices without drivers the PCI PM core will only save

s/without/with/

Ouch, thanks.

Quoted 11 lines
Quoted 11 lines

their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

Quoted 6 lines
Quoted 3 lines

For all devices the core will disable wake-up during resume.

I'm not pretending to understand the gory details of the code itself, and so don't want you to add a Reviewed-by or such like for me, thanks.

Thanks for the fixes, I'll resend the patch with updated changelog.

Best,
Rafael

#20 lore raw

Re: [PATCH 2/3] PCI PM: Let the core be more careful with respect to drivers using new framework

Quoted 56 lines · Rafael J. Wysocki
Quoted 23 lines · Nigel Cunningham

Hi Rafael.

Just a couple of typos in the comment:

Quoted 18 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

Quoted 5 lines
Quoted 3 lines

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attepmt to do anything else

s/attepmt/attempt/

OK

Quoted 10 lines
Quoted 8 lines

to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices without drivers the PCI PM core will only save

s/without/with/

Ouch, thanks.

Quoted 11 lines
Quoted 11 lines

their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

For all devices the core will disable wake-up during resume.

I'm not pretending to understand the gory details of the code itself, and so don't want you to add a Reviewed-by or such like for me, thanks.

Thanks for the fixes, I'll resend the patch with updated changelog.

Appended.

Thanks,
Rafael

---
From: Rafael J. Wysocki <[email protected]>
Subject: PCI PM: Let the core be more careful with respect to drivers using new framework

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attempt to do anything else to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

For all devices the core will disable wake-up during resume.

---
 drivers/pci/pci-driver.c |  138 ++++++++++++++++++++++++++++-------------------
 1 file changed, 85 insertions(+), 53 deletions(-)
drivers/pci/pci-driver.c +85 -53
Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================
@@ -370,7 +370,6 @@ static int pci_legacy_suspend(struct dev
 	}
 
 	pci_save_state(pci_dev);
-	pci_dev->state_saved = true;
 	/*
 	 * This is for compatibility with existing code with legacy PM support.
 	 */
@@ -424,39 +423,22 @@ static void pci_pm_default_resume_noirq(
 	pci_fixup_device(pci_fixup_resume_early, pci_dev);
 }
 
-static int pci_pm_default_resume(struct pci_dev *pci_dev)
+static void pci_pm_default_resume(struct pci_dev *pci_dev)
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (pci_is_bridge(pci_dev))
-		return 0;
-
-	pci_enable_wake(pci_dev, PCI_D0, false);
-	return pci_pm_reenable_device(pci_dev);
+	if (!pci_is_bridge(pci_dev))
+		pci_enable_wake(pci_dev, PCI_D0, false);
 }
 
-static void pci_pm_default_suspend_generic(struct pci_dev *pci_dev)
+static void pci_pm_default_suspend(struct pci_dev *pci_dev)
 {
-	/* If a non-bridge device is enabled at this point, disable it */
+	/* Disable non-bridge devices without PM support */
 	if (!pci_is_bridge(pci_dev))
 		pci_disable_enabled_device(pci_dev);
-	/*
-	 * Save state with interrupts enabled, because in principle the bus the
-	 * device is on may be put into a low power state after this code runs.
-	 */
 	pci_save_state(pci_dev);
 }
 
-static void pci_pm_default_suspend(struct pci_dev *pci_dev, bool prepare)
-{
-	pci_pm_default_suspend_generic(pci_dev);
-
-	if (prepare && !pci_is_bridge(pci_dev))
-		pci_prepare_to_sleep(pci_dev);
-
-	pci_fixup_device(pci_fixup_suspend, pci_dev);
-}
-
 static bool pci_has_legacy_pm_support(struct pci_dev *pci_dev)
 {
 	struct pci_driver *drv = pci_dev->driver;
@@ -500,20 +482,40 @@ static int pci_pm_suspend(struct device 
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
-	int error = 0;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_SUSPEND);
 
-	if (pm && pm->suspend) {
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		goto Fixup;
+	}
+
+	pci_dev->state_saved = false;
+
+	if (pm->suspend) {
+		int error;
+
 		error = pm->suspend(dev);
 		suspend_report_result(pm->suspend, error);
+		if (error)
+			return error;
+		if (!pci_dev->state_saved)
+			WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: State of device not saved by %pF\n",
+				pm->suspend);
 	}
 
-	if (!error)
-		pci_pm_default_suspend(pci_dev, !!pm);
+	if (!pci_dev->state_saved) {
+		pci_save_state(pci_dev);
+		if (!pci_is_bridge(pci_dev))
+			pci_prepare_to_sleep(pci_dev);
+	}
 
-	return error;
+ Fixup:
+	pci_fixup_device(pci_fixup_suspend, pci_dev);
+
+	return 0;
 }
 
 static int pci_pm_suspend_noirq(struct device *dev)
@@ -556,7 +558,7 @@ static int pci_pm_resume_noirq(struct de
 static int pci_pm_resume(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	/*
@@ -569,12 +571,16 @@ static int pci_pm_resume(struct device *
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	error = pci_pm_default_resume(pci_dev);
+	pci_pm_default_resume(pci_dev);
 
-	if (!error && drv && drv->pm && drv->pm->resume)
-		error = drv->pm->resume(dev);
+	if (pm) {
+		if (pm->resume)
+			error = pm->resume(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
-	return error;
+	return 0;
 }
 
 #else /* !CONFIG_SUSPEND */
@@ -591,21 +597,31 @@ static int pci_pm_resume(struct device *
 static int pci_pm_freeze(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
-	int error = 0;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_FREEZE);
 
-	if (drv && drv->pm && drv->pm->freeze) {
-		error = drv->pm->freeze(dev);
-		suspend_report_result(drv->pm->freeze, error);
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		return 0;
 	}
 
-	if (!error)
-		pci_pm_default_suspend_generic(pci_dev);
+	pci_dev->state_saved = false;
 
-	return error;
+	if (pm->freeze) {
+		int error;
+
+		error = pm->freeze(dev);
+		suspend_report_result(pm->freeze, error);
+		if (error)
+			return error;
+	}
+
+	if (!pci_dev->state_saved)
+		pci_save_state(pci_dev);
+
+	return 0;
 }
 
 static int pci_pm_freeze_noirq(struct device *dev)
@@ -648,16 +664,18 @@ static int pci_pm_thaw_noirq(struct devi
 static int pci_pm_thaw(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	pci_pm_reenable_device(pci_dev);
-
-	if (drv && drv->pm && drv->pm->thaw)
-		error =  drv->pm->thaw(dev);
+	if (pm) {
+		if (pm->thaw)
+			error = pm->thaw(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
 	return error;
 }
@@ -671,13 +689,23 @@ static int pci_pm_poweroff(struct device
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_HIBERNATE);
 
-	if (pm && pm->poweroff) {
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		goto Fixup;
+	}
+
+	pci_dev->state_saved = false;
+
+	if (pm->poweroff) {
 		error = pm->poweroff(dev);
 		suspend_report_result(pm->poweroff, error);
 	}
 
-	if (!error)
-		pci_pm_default_suspend(pci_dev, !!pm);
+	if (!pci_dev->state_saved && !pci_is_bridge(pci_dev))
+		pci_prepare_to_sleep(pci_dev);
+
+ Fixup:
+	pci_fixup_device(pci_fixup_suspend, pci_dev);
 
 	return error;
 }
@@ -718,7 +746,7 @@ static int pci_pm_restore_noirq(struct d
 static int pci_pm_restore(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	/*
@@ -731,10 +759,14 @@ static int pci_pm_restore(struct device 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	error = pci_pm_default_resume(pci_dev);
+	pci_pm_default_resume(pci_dev);
 
-	if (!error && drv && drv->pm && drv->pm->restore)
-		error = drv->pm->restore(dev);
+	if (pm) {
+		if (pm->restore)
+			error = pm->restore(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
 	return error;
 }
#21 lore raw

Re: [linux-pm] [PATCH 2/3] PCI PM: Let the core be more careful with respect to drivers using new framework

In reply to Rafael J. Wysocki
Quoted 56 lines · Rafael J. Wysocki
Quoted 23 lines · Nigel Cunningham

Hi Rafael.

Just a couple of typos in the comment:

Quoted 18 lines · Rafael J. Wysocki

From: Rafael J. Wysocki <[email protected]>

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

Quoted 5 lines
Quoted 3 lines

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attepmt to do anything else

s/attepmt/attempt/

OK

Quoted 10 lines
Quoted 8 lines

to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices without drivers the PCI PM core will only save

s/without/with/

Ouch, thanks.

Quoted 11 lines
Quoted 11 lines

their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

For all devices the core will disable wake-up during resume.

I'm not pretending to understand the gory details of the code itself, and so don't want you to add a Reviewed-by or such like for me, thanks.

Thanks for the fixes, I'll resend the patch with updated changelog.

Appended.

Thanks,
Rafael

---
From: Rafael J. Wysocki <[email protected]>
Subject: PCI PM: Let the core be more careful with respect to drivers using new framework

Currently, the PM core always attempts to manage devices with drivers that use the new PM framework. In particular, it attempts to disable the devices (which is unnecessary), to save their state (which may be undesirable if the driver has done that already) and to put them into low power states (again, this may be undesirable if the driver has already put the device into a low power state). That need not be the right thing to do, so make the core be more careful in this respect.

Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.

For bridge devices without drivers the PCI PM core will save their state on suspend and restore it (early) during resume, after putting them into D0 if necessary. It will not attempt to do anything else to these devices.

For non-bridge devices without drivers the PCI PM core will disable them and save their state on suspend. During resume, it will put them into D0, if necessary, restore their state (early) and reenable them.

For bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Still, the core will restore their state (early) during resume, after putting them into D0, if necessary.

For non-bridge devices with drivers the PCI PM core will only save their state on suspend if the driver hasn't done that already. Also, if the state of the device hasn't been saved by the driver, the core will attempt to put the device into a low power state. During resume the core will restore the state of the device (early), after putting it into D0, if necessary.

For all devices the core will disable wake-up during resume.

---
 drivers/pci/pci-driver.c |  138 ++++++++++++++++++++++++++++-------------------
 1 file changed, 85 insertions(+), 53 deletions(-)
drivers/pci/pci-driver.c +85 -53
Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================
@@ -370,7 +370,6 @@ static int pci_legacy_suspend(struct dev
 	}
 
 	pci_save_state(pci_dev);
-	pci_dev->state_saved = true;
 	/*
 	 * This is for compatibility with existing code with legacy PM support.
 	 */
@@ -424,39 +423,22 @@ static void pci_pm_default_resume_noirq(
 	pci_fixup_device(pci_fixup_resume_early, pci_dev);
 }
 
-static int pci_pm_default_resume(struct pci_dev *pci_dev)
+static void pci_pm_default_resume(struct pci_dev *pci_dev)
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (pci_is_bridge(pci_dev))
-		return 0;
-
-	pci_enable_wake(pci_dev, PCI_D0, false);
-	return pci_pm_reenable_device(pci_dev);
+	if (!pci_is_bridge(pci_dev))
+		pci_enable_wake(pci_dev, PCI_D0, false);
 }
 
-static void pci_pm_default_suspend_generic(struct pci_dev *pci_dev)
+static void pci_pm_default_suspend(struct pci_dev *pci_dev)
 {
-	/* If a non-bridge device is enabled at this point, disable it */
+	/* Disable non-bridge devices without PM support */
 	if (!pci_is_bridge(pci_dev))
 		pci_disable_enabled_device(pci_dev);
-	/*
-	 * Save state with interrupts enabled, because in principle the bus the
-	 * device is on may be put into a low power state after this code runs.
-	 */
 	pci_save_state(pci_dev);
 }
 
-static void pci_pm_default_suspend(struct pci_dev *pci_dev, bool prepare)
-{
-	pci_pm_default_suspend_generic(pci_dev);
-
-	if (prepare && !pci_is_bridge(pci_dev))
-		pci_prepare_to_sleep(pci_dev);
-
-	pci_fixup_device(pci_fixup_suspend, pci_dev);
-}
-
 static bool pci_has_legacy_pm_support(struct pci_dev *pci_dev)
 {
 	struct pci_driver *drv = pci_dev->driver;
@@ -500,20 +482,40 @@ static int pci_pm_suspend(struct device 
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
-	int error = 0;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_SUSPEND);
 
-	if (pm && pm->suspend) {
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		goto Fixup;
+	}
+
+	pci_dev->state_saved = false;
+
+	if (pm->suspend) {
+		int error;
+
 		error = pm->suspend(dev);
 		suspend_report_result(pm->suspend, error);
+		if (error)
+			return error;
+		if (!pci_dev->state_saved)
+			WARN_ONCE(pci_dev->current_state != PCI_D0,
+				"PCI PM: State of device not saved by %pF\n",
+				pm->suspend);
 	}
 
-	if (!error)
-		pci_pm_default_suspend(pci_dev, !!pm);
+	if (!pci_dev->state_saved) {
+		pci_save_state(pci_dev);
+		if (!pci_is_bridge(pci_dev))
+			pci_prepare_to_sleep(pci_dev);
+	}
 
-	return error;
+ Fixup:
+	pci_fixup_device(pci_fixup_suspend, pci_dev);
+
+	return 0;
 }
 
 static int pci_pm_suspend_noirq(struct device *dev)
@@ -556,7 +558,7 @@ static int pci_pm_resume_noirq(struct de
 static int pci_pm_resume(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	/*
@@ -569,12 +571,16 @@ static int pci_pm_resume(struct device *
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	error = pci_pm_default_resume(pci_dev);
+	pci_pm_default_resume(pci_dev);
 
-	if (!error && drv && drv->pm && drv->pm->resume)
-		error = drv->pm->resume(dev);
+	if (pm) {
+		if (pm->resume)
+			error = pm->resume(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
-	return error;
+	return 0;
 }
 
 #else /* !CONFIG_SUSPEND */
@@ -591,21 +597,31 @@ static int pci_pm_resume(struct device *
 static int pci_pm_freeze(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
-	int error = 0;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_FREEZE);
 
-	if (drv && drv->pm && drv->pm->freeze) {
-		error = drv->pm->freeze(dev);
-		suspend_report_result(drv->pm->freeze, error);
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		return 0;
 	}
 
-	if (!error)
-		pci_pm_default_suspend_generic(pci_dev);
+	pci_dev->state_saved = false;
 
-	return error;
+	if (pm->freeze) {
+		int error;
+
+		error = pm->freeze(dev);
+		suspend_report_result(pm->freeze, error);
+		if (error)
+			return error;
+	}
+
+	if (!pci_dev->state_saved)
+		pci_save_state(pci_dev);
+
+	return 0;
 }
 
 static int pci_pm_freeze_noirq(struct device *dev)
@@ -648,16 +664,18 @@ static int pci_pm_thaw_noirq(struct devi
 static int pci_pm_thaw(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	pci_pm_reenable_device(pci_dev);
-
-	if (drv && drv->pm && drv->pm->thaw)
-		error =  drv->pm->thaw(dev);
+	if (pm) {
+		if (pm->thaw)
+			error = pm->thaw(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
 	return error;
 }
@@ -671,13 +689,23 @@ static int pci_pm_poweroff(struct device
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_suspend(dev, PMSG_HIBERNATE);
 
-	if (pm && pm->poweroff) {
+	if (!pm) {
+		pci_pm_default_suspend(pci_dev);
+		goto Fixup;
+	}
+
+	pci_dev->state_saved = false;
+
+	if (pm->poweroff) {
 		error = pm->poweroff(dev);
 		suspend_report_result(pm->poweroff, error);
 	}
 
-	if (!error)
-		pci_pm_default_suspend(pci_dev, !!pm);
+	if (!pci_dev->state_saved && !pci_is_bridge(pci_dev))
+		pci_prepare_to_sleep(pci_dev);
+
+ Fixup:
+	pci_fixup_device(pci_fixup_suspend, pci_dev);
 
 	return error;
 }
@@ -718,7 +746,7 @@ static int pci_pm_restore_noirq(struct d
 static int pci_pm_restore(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
-	struct device_driver *drv = dev->driver;
+	struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
 	int error = 0;
 
 	/*
@@ -731,10 +759,14 @@ static int pci_pm_restore(struct device 
 	if (pci_has_legacy_pm_support(pci_dev))
 		return pci_legacy_resume(dev);
 
-	error = pci_pm_default_resume(pci_dev);
+	pci_pm_default_resume(pci_dev);
 
-	if (!error && drv && drv->pm && drv->pm->restore)
-		error = drv->pm->restore(dev);
+	if (pm) {
+		if (pm->restore)
+			error = pm->restore(dev);
+	} else {
+		pci_pm_reenable_device(pci_dev);
+	}
 
 	return error;
 }
#29 lore raw
In reply to Rafael J. Wysocki
Quoted 8 lines · Rafael J. Wysocki

Hi,

After the recent discussion with Linus, I have the following three fixes of the PCI PM framework.

The first patch fixes the bug that bridges (and PCIe ports) are disabled during suspend, althouth they shouldn't.

Quoted 6 lines

The second one makes the PCI PM core handle devices more carefully (details in the changelog). [Note to Linus: devices are still put into low power states with interrupts on after this patch. Moving that to the late suspend phase will be the next step.]

The last patch makes the warning in pci_legacy_suspend() more useful.

Linus, do you want these as part of the next pull request? If so, I'll check them out, pull them in, and let them run the linux-next build gauntlet for a day before sending them. Otherwise I'll just send a pull request today with the 8 or so fixes I have queued now.

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center
#30 lore raw
In reply to Rafael J. Wysocki
Quoted 8 lines · Rafael J. Wysocki

Hi,

After the recent discussion with Linus, I have the following three fixes of the PCI PM framework.

The first patch fixes the bug that bridges (and PCIe ports) are disabled during suspend, althouth they shouldn't.

Quoted 6 lines

The second one makes the PCI PM core handle devices more carefully (details in the changelog). [Note to Linus: devices are still put into low power states with interrupts on after this patch. Moving that to the late suspend phase will be the next step.]

The last patch makes the warning in pci_legacy_suspend() more useful.

Linus, do you want these as part of the next pull request? If so, I'll check them out, pull them in, and let them run the linux-next build gauntlet for a day before sending them. Otherwise I'll just send a pull request today with the 8 or so fixes I have queued now.

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center
#31 lore raw
Quoted 7 lines · Jesse Barnes
Quoted 2 lines

The last patch makes the warning in pci_legacy_suspend() more useful.

Linus, do you want these as part of the next pull request? If so, I'll check them out, pull them in, and let them run the linux-next build gauntlet for a day before sending them. Otherwise I'll just send a pull request today with the 8 or so fixes I have queued now.

I want them eventually, but I don't know if we got confirmation that this particular series actually fixed the PCI-E bridge issue? The one that Parag reported.. It would be good to have confirmation. Maybe we do already and I missed that email?

Linus

#32 lore raw
In reply to Jesse Barnes
Quoted 7 lines · Jesse Barnes
Quoted 2 lines

The last patch makes the warning in pci_legacy_suspend() more useful.

Linus, do you want these as part of the next pull request? If so, I'll check them out, pull them in, and let them run the linux-next build gauntlet for a day before sending them. Otherwise I'll just send a pull request today with the 8 or so fixes I have queued now.

I want them eventually, but I don't know if we got confirmation that this particular series actually fixed the PCI-E bridge issue? The one that Parag reported.. It would be good to have confirmation. Maybe we do already and I missed that email?

Linus

#33 lore raw
Quoted 10 lines · Linus Torvalds
Quoted 7 lines · Jesse Barnes
Quoted 2 lines

The last patch makes the warning in pci_legacy_suspend() more useful.

Linus, do you want these as part of the next pull request? If so, I'll check them out, pull them in, and let them run the linux-next build gauntlet for a day before sending them. Otherwise I'll just send a pull request today with the 8 or so fixes I have queued now.

Quoted 4 lines

I want them eventually, but I don't know if we got confirmation that this particular series actually fixed the PCI-E bridge issue? The one that Parag reported.. It would be good to have confirmation. Maybe we do already and I missed that email?

An additional patch against the PCIe port driver will be necessary to fix this issue, but I'm waiting for feedback from Parag about the patches already posted to find out what's the minimal fix.

Thanks,
Rafael

#34 lore raw
In reply to Linus Torvalds
Quoted 10 lines · Linus Torvalds
Quoted 7 lines · Jesse Barnes
Quoted 2 lines

The last patch makes the warning in pci_legacy_suspend() more useful.

Linus, do you want these as part of the next pull request? If so, I'll check them out, pull them in, and let them run the linux-next build gauntlet for a day before sending them. Otherwise I'll just send a pull request today with the 8 or so fixes I have queued now.

Quoted 4 lines

I want them eventually, but I don't know if we got confirmation that this particular series actually fixed the PCI-E bridge issue? The one that Parag reported.. It would be good to have confirmation. Maybe we do already and I missed that email?

An additional patch against the PCIe port driver will be necessary to fix this issue, but I'm waiting for feedback from Parag about the patches already posted to find out what's the minimal fix.

Thanks,
Rafael

#3 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

In reply to Rafael J. Wysocki

Hi!

Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 6 lines

Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================

drivers/pci/pci-driver.c +0 -0
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 10 lines
drivers/pci/pci-driver.c +3 -2
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#4 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

In reply to Rafael J. Wysocki

Hi!

Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 6 lines

Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================

drivers/pci/pci-driver.c +0 -0
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 10 lines
drivers/pci/pci-driver.c +3 -2
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#5 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

Quoted 20 lines · Pavel Machek

Hi!

Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 13 lines

Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================

drivers/pci/pci-driver.c +2 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
Quoted 6 lines

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 3 lines
drivers/pci/pci-driver.c +1 -0
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Can you explain what you mean, please?

Thanks,
Rafael

#6 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

In reply to Pavel Machek
Quoted 20 lines · Pavel Machek

Hi!

Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 13 lines

Index: linux-2.6/drivers/pci/pci-driver.c
===================================================================

drivers/pci/pci-driver.c +2 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
Quoted 6 lines

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 3 lines
drivers/pci/pci-driver.c +1 -0
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Can you explain what you mean, please?

Thanks,
Rafael

#7 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

Hi!

Quoted 12 lines
Quoted 12 lines
Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 9 lines
drivers/pci/pci-driver.c +2 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
Quoted 8 lines
Quoted 6 lines

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 3 lines
drivers/pci/pci-driver.c +1 -0
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Can you explain what you mean, please?

It looks to me like the patch does not match the changelog.

Changelog says "enable bridge during resume", but code does return 0 if it seems bridge.
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#8 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

In reply to Rafael J. Wysocki

Hi!

Quoted 12 lines
Quoted 12 lines
Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 9 lines
drivers/pci/pci-driver.c +2 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
Quoted 8 lines
Quoted 6 lines

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 3 lines
drivers/pci/pci-driver.c +1 -0
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Can you explain what you mean, please?

It looks to me like the patch does not match the changelog.

Changelog says "enable bridge during resume", but code does return 0 if it seems bridge.
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#9 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

Quoted 21 lines · Pavel Machek

Hi!

Quoted 19 lines
Quoted 18 lines
Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 12 lines
drivers/pci/pci-driver.c +3 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Quoted 6 lines
Quoted 1 line

Can you explain what you mean, please?

It looks to me like the patch does not match the changelog.

Changelog says "enable bridge during resume", but code does return 0 if it seems bridge.

Ah, I see. The changelog was supposed to mean that the bridges would not be disabled during suspend and also would not be enabled during resume (which shouldn't be necessary, since they would be automatically enabled as a result of restoring their configuration registers).

Thanks,
Rafael

Quoted 1 line

Pavel

Signature · 3 lines
Everyone knows that debugging is twice as hard as writing a program
in the first place.  So if you're as clever as you can be when you write it,
how will you ever debug it? --- Brian Kernighan
#10 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

In reply to Pavel Machek
Quoted 21 lines · Pavel Machek

Hi!

Quoted 19 lines
Quoted 18 lines
Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 12 lines
drivers/pci/pci-driver.c +3 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Quoted 6 lines
Quoted 1 line

Can you explain what you mean, please?

It looks to me like the patch does not match the changelog.

Changelog says "enable bridge during resume", but code does return 0 if it seems bridge.

Ah, I see. The changelog was supposed to mean that the bridges would not be disabled during suspend and also would not be enabled during resume (which shouldn't be necessary, since they would be automatically enabled as a result of restoring their configuration registers).

Thanks,
Rafael

Quoted 1 line

Pavel

Signature · 3 lines
Everyone knows that debugging is twice as hard as writing a program
in the first place.  So if you're as clever as you can be when you write it,
how will you ever debug it? --- Brian Kernighan
#11 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

Quoted 26 lines · Pavel Machek
Quoted 25 lines

Hi!

Quoted 12 lines
Quoted 12 lines
Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 9 lines
drivers/pci/pci-driver.c +2 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
Quoted 8 lines
Quoted 6 lines

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 3 lines
drivers/pci/pci-driver.c +1 -0
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Can you explain what you mean, please?

It looks to me like the patch does not match the changelog.

Quoted 7 lines
Quoted 2 lines

Changelog says "enable bridge during resume", but code does return 0 if it seems bridge.

Ah, I see. The changelog was supposed to mean that the bridges would not be disabled during suspend and also would not be enabled during resume (which shouldn't be necessary, since they would be automatically enabled as a result of restoring their configuration registers).

You can add my acked-by: when you resubmit with fixed changelog... :-).
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#12 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

In reply to Rafael J. Wysocki
Quoted 26 lines · Pavel Machek
Quoted 25 lines

Hi!

Quoted 12 lines
Quoted 12 lines
Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 9 lines
drivers/pci/pci-driver.c +2 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
Quoted 8 lines
Quoted 6 lines

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 3 lines
drivers/pci/pci-driver.c +1 -0
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Can you explain what you mean, please?

It looks to me like the patch does not match the changelog.

Quoted 7 lines
Quoted 2 lines

Changelog says "enable bridge during resume", but code does return 0 if it seems bridge.

Ah, I see. The changelog was supposed to mean that the bridges would not be disabled during suspend and also would not be enabled during resume (which shouldn't be necessary, since they would be automatically enabled as a result of restoring their configuration registers).

You can add my acked-by: when you resubmit with fixed changelog... :-).
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#13 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

Quoted 29 lines · Pavel Machek
Quoted 29 lines · Pavel Machek
Quoted 27 lines

Hi!

Quoted 12 lines
Quoted 12 lines
Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 9 lines
drivers/pci/pci-driver.c +2 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
Quoted 8 lines
Quoted 6 lines

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 3 lines
drivers/pci/pci-driver.c +1 -0
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Can you explain what you mean, please?

It looks to me like the patch does not match the changelog.

Changelog says "enable bridge during resume", but code does return 0 if it seems bridge.

Quoted 7 lines
Quoted 4 lines

Ah, I see. The changelog was supposed to mean that the bridges would not be disabled during suspend and also would not be enabled during resume (which shouldn't be necessary, since they would be automatically enabled as a result of restoring their configuration registers).

You can add my acked-by: when you resubmit with fixed changelog... :-).

It's merged, so that wouldn't be of any use.

Thanks,
Rafael

#14 lore raw

Re: [PATCH 1/3] PCI PM: Do not disable and enable bridges during suspend-resume

In reply to Pavel Machek
Quoted 29 lines · Pavel Machek
Quoted 29 lines · Pavel Machek
Quoted 27 lines

Hi!

Quoted 12 lines
Quoted 12 lines
Quoted 2 lines

their standard config spaces are restored. Fix this by not attempting to disable bridges during suspend and enable them during resume.

...

Quoted 9 lines
drivers/pci/pci-driver.c +2 -2
@@ -428,16 +428,18 @@ static int pci_pm_default_resume(struct 
 {
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
-	if (!pci_is_bridge(pci_dev))
-		pci_enable_wake(pci_dev, PCI_D0, false);
+	if (pci_is_bridge(pci_dev))
+		return 0;
 
Quoted 8 lines
Quoted 6 lines

drivers/pci/pci-driver.c static int pci_pm_default_resume(struct · line 436

Quoted 3 lines
drivers/pci/pci-driver.c +1 -0
+	pci_enable_wake(pci_dev, PCI_D0, false);
 	return pci_pm_reenable_device(pci_dev);
 }

Are you sure? This goes from doing reenable_device to not doing it for bridges, seemingly contradicting changelog?

Can you explain what you mean, please?

It looks to me like the patch does not match the changelog.

Changelog says "enable bridge during resume", but code does return 0 if it seems bridge.

Quoted 7 lines
Quoted 4 lines

Ah, I see. The changelog was supposed to mean that the bridges would not be disabled during suspend and also would not be enabled during resume (which shouldn't be necessary, since they would be automatically enabled as a result of restoring their configuration registers).

You can add my acked-by: when you resubmit with fixed changelog... :-).

It's merged, so that wouldn't be of any use.

Thanks,
Rafael