Skip to contentSkip to conversation

NET/r8169: Rework suspend and resume

PATCH Applied Rafael J. Wysocki 3 participants 7 messages Apr 5, 2009 – Apr 13, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 NET/r8169: Rework suspend and resume

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

The recent changes of the PCI PM core allow us to simplify the suspend and resume handling in a number of device drivers, since they don't need to carry out the general PCI PM operations, such as changing the power state of the device, during suspend and resume any more.

Simplify the suspend and resume callbacks of r8169 using the observation that the PCI PM core can take care of some operations carried out by the driver.

Additionally, make the shutdown callback of r8169 only put the device into a low power state if the system is going to be powered off (kexec is known to have problems with network adapters that are put into low power states on shutdown).

This patch has been tested on MSI Wind U100.

---
 drivers/net/r8169.c |   61 +++++++++++++++++++++++++++++++++-------------------
 1 file changed, 39 insertions(+), 22 deletions(-)
drivers/net/r8169.c +39 -22
Index: linux-2.6/drivers/net/r8169.c
===================================================================
@@ -3791,16 +3791,13 @@ static struct net_device_stats *rtl8169_
 	return &dev->stats;
 }
 
-#ifdef CONFIG_PM
-
-static int rtl8169_suspend(struct pci_dev *pdev, pm_message_t state)
+static void rtl8169_net_suspend(struct net_device *dev)
 {
-	struct net_device *dev = pci_get_drvdata(pdev);
 	struct rtl8169_private *tp = netdev_priv(dev);
 	void __iomem *ioaddr = tp->mmio_addr;
 
 	if (!netif_running(dev))
-		goto out_pci_suspend;
+		return;
 
 	netif_device_detach(dev);
 	netif_stop_queue(dev);
@@ -3812,24 +3809,25 @@ static int rtl8169_suspend(struct pci_de
 	rtl8169_rx_missed(dev, ioaddr);
 
 	spin_unlock_irq(&tp->lock);
+}
 
-out_pci_suspend:
-	pci_save_state(pdev);
-	pci_enable_wake(pdev, pci_choose_state(pdev, state),
-		(tp->features & RTL_FEATURE_WOL) ? 1 : 0);
-	pci_set_power_state(pdev, pci_choose_state(pdev, state));
+#ifdef CONFIG_PM
+
+static int rtl8169_suspend(struct device *device)
+{
+	struct pci_dev *pdev = to_pci_dev(device);
+	struct net_device *dev = pci_get_drvdata(pdev);
+
+	rtl8169_net_suspend(dev);
 
 	return 0;
 }
 
-static int rtl8169_resume(struct pci_dev *pdev)
+static int rtl8169_resume(struct device *device)
 {
+	struct pci_dev *pdev = to_pci_dev(device);
 	struct net_device *dev = pci_get_drvdata(pdev);
 
-	pci_set_power_state(pdev, PCI_D0);
-	pci_restore_state(pdev);
-	pci_enable_wake(pdev, PCI_D0, 0);
-
 	if (!netif_running(dev))
 		goto out;
 
@@ -3840,23 +3838,42 @@ out:
 	return 0;
 }
 
+static struct dev_pm_ops rtl8169_pm_ops = {
+	.suspend = rtl8169_suspend,
+	.resume = rtl8169_resume,
+	.freeze = rtl8169_suspend,
+	.thaw = rtl8169_resume,
+	.poweroff = rtl8169_suspend,
+	.restore = rtl8169_resume,
+};
+
+#define RTL8169_PM_OPS	(&rtl8169_pm_ops)
+
+#else /* !CONFIG_PM */
+
+#define RTL8169_PM_OPS	NULL
+
+#endif /* !CONFIG_PM */
+
 static void rtl_shutdown(struct pci_dev *pdev)
 {
-	rtl8169_suspend(pdev, PMSG_SUSPEND);
-}
+	struct net_device *dev = pci_get_drvdata(pdev);
 
-#endif /* CONFIG_PM */
+	rtl8169_net_suspend(dev);
+
+	if (system_state == SYSTEM_POWER_OFF) {
+		pci_wake_from_d3(pdev, true);
+		pci_set_power_state(pdev, PCI_D3hot);
+	}
+}
 
 static struct pci_driver rtl8169_pci_driver = {
 	.name		= MODULENAME,
 	.id_table	= rtl8169_pci_tbl,
 	.probe		= rtl8169_init_one,
 	.remove		= __devexit_p(rtl8169_remove_one),
-#ifdef CONFIG_PM
-	.suspend	= rtl8169_suspend,
-	.resume		= rtl8169_resume,
 	.shutdown	= rtl_shutdown,
-#endif
+	.driver.pm	= RTL8169_PM_OPS,
 };
 
 static int __init rtl8169_init_module(void)
#1 lore raw
Quoted 12 lines · On Sun, 05 April 2009 "Rafael J. Wysocki" <[email protected]> wrote:

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

The recent changes of the PCI PM core allow us to simplify the suspend and resume handling in a number of device drivers, since they don't need to carry out the general PCI PM operations, such as changing the power state of the device, during suspend and resume any more.

Simplify the suspend and resume callbacks of r8169 using the observation that the PCI PM core can take care of some operations carried out by the driver.

Quoted 6 lines

Additionally, make the shutdown callback of r8169 only put the device into a low power state if the system is going to be powered off (kexec is known to have problems with network adapters that are put into low power states on shutdown).

This patch has been tested on MSI Wind U100.

Tested against 2.6.29.1 on my Commell LE-365 and suspend (S3) + WoL still works as expected.

02:08.0 Ethernet controller [0200]: Realtek Semiconductor Co., Ltd.

                     RTL-8169 Gigabit Ethernet [10ec:8169] (rev 10)

Bruno

#2 lore raw
Quoted 18 lines · Bruno Prémont
Quoted 17 lines · On Sun, 05 April 2009 "Rafael J. Wysocki" <[email protected]> wrote:

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

The recent changes of the PCI PM core allow us to simplify the suspend and resume handling in a number of device drivers, since they don't need to carry out the general PCI PM operations, such as changing the power state of the device, during suspend and resume any more.

Simplify the suspend and resume callbacks of r8169 using the observation that the PCI PM core can take care of some operations carried out by the driver.

Additionally, make the shutdown callback of r8169 only put the device into a low power state if the system is going to be powered off (kexec is known to have problems with network adapters that are put into low power states on shutdown).

Quoted 7 lines
Quoted 1 line

This patch has been tested on MSI Wind U100.

Tested against 2.6.29.1 on my Commell LE-365 and suspend (S3) + WoL still works as expected.

02:08.0 Ethernet controller [0200]: Realtek Semiconductor Co., Ltd.

                     RTL-8169 Gigabit Ethernet [10ec:8169] (rev 10)

Thanks for testing!

Best,
Rafael

#3 lore raw
In reply to Bruno Prémont
Quoted 18 lines · Bruno Prémont
Quoted 17 lines · On Sun, 05 April 2009 "Rafael J. Wysocki" <[email protected]> wrote:

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

The recent changes of the PCI PM core allow us to simplify the suspend and resume handling in a number of device drivers, since they don't need to carry out the general PCI PM operations, such as changing the power state of the device, during suspend and resume any more.

Simplify the suspend and resume callbacks of r8169 using the observation that the PCI PM core can take care of some operations carried out by the driver.

Additionally, make the shutdown callback of r8169 only put the device into a low power state if the system is going to be powered off (kexec is known to have problems with network adapters that are put into low power states on shutdown).

Quoted 7 lines
Quoted 1 line

This patch has been tested on MSI Wind U100.

Tested against 2.6.29.1 on my Commell LE-365 and suspend (S3) + WoL still works as expected.

02:08.0 Ethernet controller [0200]: Realtek Semiconductor Co., Ltd.

                     RTL-8169 Gigabit Ethernet [10ec:8169] (rev 10)

Thanks for testing!

Best,
Rafael

maintainer Applied
#4 lore raw

From: "Rafael J. Wysocki" <[email protected]>
Date: Sun, 5 Apr 2009 23:59:31 +0200

Quoted 21 lines · Bruno Prémont
Quoted 20 lines · On Sun, 05 April 2009 "Rafael J. Wysocki" <[email protected]> wrote:
Quoted 18 lines

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

The recent changes of the PCI PM core allow us to simplify the suspend and resume handling in a number of device drivers, since they don't need to carry out the general PCI PM operations, such as changing the power state of the device, during suspend and resume any more.

Simplify the suspend and resume callbacks of r8169 using the observation that the PCI PM core can take care of some operations carried out by the driver.

Additionally, make the shutdown callback of r8169 only put the device into a low power state if the system is going to be powered off (kexec is known to have problems with network adapters that are put into low power states on shutdown).

This patch has been tested on MSI Wind U100.

Quoted 7 lines
Quoted 5 lines

Tested against 2.6.29.1 on my Commell LE-365 and suspend (S3) + WoL still works as expected.

02:08.0 Ethernet controller [0200]: Realtek Semiconductor Co., Ltd.

                     RTL-8169 Gigabit Ethernet [10ec:8169] (rev 10)

Thanks for testing!

I've applied this patch to net-next-2.6, thanks everyone.

maintainer Applied
#5 lore raw
In reply to Rafael J. Wysocki

From: "Rafael J. Wysocki" <[email protected]>
Date: Sun, 5 Apr 2009 23:59:31 +0200

Quoted 21 lines · Bruno Prémont
Quoted 20 lines · On Sun, 05 April 2009 "Rafael J. Wysocki" <[email protected]> wrote:
Quoted 18 lines

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

The recent changes of the PCI PM core allow us to simplify the suspend and resume handling in a number of device drivers, since they don't need to carry out the general PCI PM operations, such as changing the power state of the device, during suspend and resume any more.

Simplify the suspend and resume callbacks of r8169 using the observation that the PCI PM core can take care of some operations carried out by the driver.

Additionally, make the shutdown callback of r8169 only put the device into a low power state if the system is going to be powered off (kexec is known to have problems with network adapters that are put into low power states on shutdown).

This patch has been tested on MSI Wind U100.

Quoted 7 lines
Quoted 5 lines

Tested against 2.6.29.1 on my Commell LE-365 and suspend (S3) + WoL still works as expected.

02:08.0 Ethernet controller [0200]: Realtek Semiconductor Co., Ltd.

                     RTL-8169 Gigabit Ethernet [10ec:8169] (rev 10)

Thanks for testing!

I've applied this patch to net-next-2.6, thanks everyone.

#6 lore raw
In reply to Rafael J. Wysocki
Quoted 12 lines · On Sun, 05 April 2009 "Rafael J. Wysocki" <[email protected]> wrote:

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

The recent changes of the PCI PM core allow us to simplify the suspend and resume handling in a number of device drivers, since they don't need to carry out the general PCI PM operations, such as changing the power state of the device, during suspend and resume any more.

Simplify the suspend and resume callbacks of r8169 using the observation that the PCI PM core can take care of some operations carried out by the driver.

Quoted 6 lines

Additionally, make the shutdown callback of r8169 only put the device into a low power state if the system is going to be powered off (kexec is known to have problems with network adapters that are put into low power states on shutdown).

This patch has been tested on MSI Wind U100.

Tested against 2.6.29.1 on my Commell LE-365 and suspend (S3) + WoL still works as expected.

02:08.0 Ethernet controller [0200]: Realtek Semiconductor Co., Ltd.

                     RTL-8169 Gigabit Ethernet [10ec:8169] (rev 10)

Bruno