Skip to contentSkip to conversation

Add basic PM support for Nvidia and ATI AGP bridges

PATCH In review Matthew Garrett 3 participants 3 messages Nov 24, 2005 – Nov 25, 2005 Agpgart driver

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Add basic PM support for Nvidia and ATI AGP bridges

I retrieved these from the swsusp2 patchset, but they seem to be independently useful. As a result, I'm not sure who the original author is - however, they seem to be pretty obvious.

## All lines beginning with `## DP:' are a description of the patch. ## DP: Description: Add suspend/resume support to ATI and Nvidia AGP bridges ## DP: Patch author: Unknown
## DP: Upstream status: Not submitted

. $(dirname $0)/DPATCH

@DPATCH@

drivers/char/agp/ati-agp.c +31 -0
diff -ruNp 210-agp-resume-support.patch-old/drivers/char/agp/ati-agp.c 210-agp-resume-support.patch-new/drivers/char/agp/ati-agp.c
@@ -507,6 +507,33 @@ static void __devexit agp_ati_remove(str
 	agp_put_bridge(bridge);
 }
 
+#ifdef CONFIG_PM
+
+static int agp_ati_suspend(struct pci_dev *pdev, pm_message_t state)
+{
+	pci_save_state (pdev);
+	pci_set_power_state (pdev, 3);
+
+	return 0;
+}
+
+static int agp_ati_resume(struct pci_dev *pdev)
+{
+	struct agp_bridge_data *bridge = pci_get_drvdata(pdev);
+
+	/* set power state 0 and restore PCI space */
+	pci_set_power_state (pdev, 0);
+	pci_restore_state(pdev);
+
+	/* reconfigure AGP hardware again */
+	if (bridge->driver == &ati_generic_bridge)
+		return ati_configure();
+
+	return 0;
+}
+
+#endif /* CONFIG_PM */
+
 static struct pci_device_id agp_ati_pci_table[] = {
 	{
 	.class		= (PCI_CLASS_BRIDGE_HOST << 8),
@@ -526,6 +553,10 @@ static struct pci_driver agp_ati_pci_dri
 	.id_table	= agp_ati_pci_table,
 	.probe		= agp_ati_probe,
 	.remove		= agp_ati_remove,
+#ifdef CONFIG_PM
+	.suspend	= agp_ati_suspend,
+	.resume		= agp_ati_resume,
+#endif
 };
 
 static int __init agp_ati_init(void)
drivers/char/agp/nvidia-agp.c +29 -0
diff -ruNp 210-agp-resume-support.patch-old/drivers/char/agp/nvidia-agp.c 210-agp-resume-support.patch-new/drivers/char/agp/nvidia-agp.c
@@ -397,11 +397,40 @@ static struct pci_device_id agp_nvidia_p
 
 MODULE_DEVICE_TABLE(pci, agp_nvidia_pci_table);
 
+#ifdef CONFIG_PM
+static int agp_nvidia_suspend(struct pci_dev *pdev, pm_message_t state)
+{
+	pci_save_state (pdev);
+	pci_set_power_state (pdev, 3);
+
+	return 0;
+}
+
+static int agp_nvidia_resume(struct pci_dev *pdev)
+{
+	struct agp_bridge_data *bridge = pci_get_drvdata(pdev);
+
+	/* set power state 0 and restore PCI space */
+	pci_set_power_state (pdev, 0);
+	pci_restore_state(pdev);
+
+	/* reconfigure AGP hardware again */
+	if (bridge->driver == &nvidia_driver)
+		nvidia_configure();
+
+	return 0;
+}
+#endif
+
 static struct pci_driver agp_nvidia_pci_driver = {
 	.name		= "agpgart-nvidia",
 	.id_table	= agp_nvidia_pci_table,
 	.probe		= agp_nvidia_probe,
 	.remove		= agp_nvidia_remove,
+#ifdef CONFIG_PM
+	.suspend	= agp_nvidia_suspend,
+	.resume		= agp_nvidia_resume,
+#endif
 };
 
 static int __init agp_nvidia_init(void)
Signature · 1 line
Matthew Garrett | [email protected]
#1 lore raw
Quoted 15 lines · Matthew Garrett

I retrieved these from the swsusp2 patchset, but they seem to be independently useful. As a result, I'm not sure who the original author is - however, they seem to be pretty obvious.

@DPATCH@
diff -ruNp 210-agp-resume-support.patch-old/drivers/char/agp/ati-agp.c 210-agp-resume-support.patch-new/drivers/char/agp/ati-agp.c ---
210-agp-resume-support.patch-old/drivers/char/agp/ati-agp.c 2005-06-20
11:46:51.000000000 +1000 +++
210-agp-resume-support.patch-new/drivers/char/agp/ati-agp.c 2005-07-04
23:14:18.000000000 +1000 @@ -507,6 +507,33 @@ static void __devexit agp_ati_remove(str

	agp_put_bridge(bridge);
}

drivers/char/agp/ati-agp.c static void __devexit agp_ati_remove(str · line 515

Quoted 6 lines
drivers/char/agp/ati-agp.c +6 -0
+#ifdef CONFIG_PM
+
+static int agp_ati_suspend(struct pci_dev *pdev, pm_message_t state)
+{
+	pci_save_state (pdev);
+	pci_set_power_state (pdev, 3);

Please remove the spaces after the function names.

Quoted 10 lines

diff -ruNp 210-agp-resume-support.patch-old/drivers/char/agp/nvidia-agp.c 210-agp-resume-support.patch-new/drivers/char/agp/nvidia-agp.c ---
210-agp-resume-support.patch-old/drivers/char/agp/nvidia-agp.c 2005-06-20
11:46:51.000000000 +1000 +++
210-agp-resume-support.patch-new/drivers/char/agp/nvidia-agp.c 2005-07-04
23:14:18.000000000 +1000 @@ -397,11 +397,40 @@ static struct pci_device_id agp_nvidia_p

MODULE_DEVICE_TABLE(pci, agp_nvidia_pci_table);

drivers/char/agp/nvidia-agp.c static struct pci_device_id agp_nvidia_p · line 404

Quoted 5 lines
drivers/char/agp/nvidia-agp.c +5 -0
+#ifdef CONFIG_PM
+static int agp_nvidia_suspend(struct pci_dev *pdev, pm_message_t state)
+{
+	pci_save_state (pdev);
+	pci_set_power_state (pdev, 3);

Same here.

Eike

#2 lore raw
In reply to Matthew Garrett

Hi!

Quoted 12 lines

I retrieved these from the swsusp2 patchset, but they seem to be independently useful. As a result, I'm not sure who the original author is - however, they seem to be pretty obvious.

## All lines beginning with `## DP:' are a description of the patch. ## DP: Description: Add suspend/resume support to ATI and Nvidia AGP bridges ## DP: Patch author: Unknown
## DP: Upstream status: Not submitted

. $(dirname $0)/DPATCH

drivers/char/agp/ati-agp.c static void __devexit agp_ati_remove(str · line 515

Quoted 9 lines
drivers/char/agp/ati-agp.c +6 -0
@@ -507,6 +507,33 @@ static void __devexit agp_ati_remove(str
 	agp_put_bridge(bridge);
 }
 
+#ifdef CONFIG_PM
+
+static int agp_ati_suspend(struct pci_dev *pdev, pm_message_t state)
+{
+	pci_save_state (pdev);
+	pci_set_power_state (pdev, 3);

PCI_D3hot, please.

drivers/char/agp/ati-agp.c static void __devexit agp_ati_remove(str · line 525

Quoted 6 lines
drivers/char/agp/ati-agp.c +6 -0
+static int agp_ati_resume(struct pci_dev *pdev)
+{
+	struct agp_bridge_data *bridge = pci_get_drvdata(pdev);
+
+	/* set power state 0 and restore PCI space */
+	pci_set_power_state (pdev, 0);

PCI_D0... ....and same in other two functions. Otherwise it looks ok.

Pavel

Signature · 1 line
Thanks, Sharp!