Attachment 137940 Details for Bug 75985 – [PATCH 2/3] PCI: Apply quirks on runtime resume despite being unbound

[patch] [PATCH 2/3] PCI: Apply quirks on runtime resume despite being unbound

0002-PCI-Apply-quirks-on-runtime-resume-despite-being-unb.patch (text/plain), 2.21 KB, created by Lukas Wunner on 2018-03-09 15:40:55 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Lukas Wunner

Created: 2018-03-09 15:40:55 UTC

Size: 2.21 KB

patch

obsolete

>From fcd287b1ced8c09dca250d4f7815c05195c1bd60 Mon Sep 17 00:00:00 2001
>From: Lukas Wunner <lukas@wunner.de>
>Date: Fri, 9 Mar 2018 13:53:25 +0100
>Subject: [PATCH 2/3] PCI: Apply quirks on runtime resume despite being unbound
>
>The integrated HDA controller on Nvidia GPUs is hidden by some laptop
>BIOSes on resume unless an external display is attached, rendering the
>HDA controller invisible and preventing its use if a display is attached
>later.  We avoid that by exposing it from a quirk, but that quirk is
>currently not applied to unbound GPUs.
>
>This leads to failure in the following scenario:
>
>* The HDA controller is bound and in runtime suspend.  The GPU is not
>  bound and the user manually allows runtime PM on it via sysfs.
>  => As a result, the root port above the GPU runtime suspends and cuts
>  power to the GPU.
>
>* The user loads the driver module for the GPU.
>  => The GPU is runtime resumed and bound to the driver.  On runtime
>  resume, the BIOS reinstates power to the GPU, detects that no HDMI/DP
>  display is attached and hides the HDA controller.  The HDA controller
>  thus becomes inaccessible to its driver.  The quirk to expose it is
>  not applied because the GPU was not yet bound on runtime resume.
>
>Fix by applying quirks of type "resume_early" on runtime resume even if
>the device is not bound.
>
>Cc: Bjorn Helgaas <bhelgaas@google.com>
>Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
>Signed-off-by: Lukas Wunner <lukas@wunner.de>
>---
> drivers/pci/pci-driver.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
>diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
>index 79afafbfd0b8..4fab6c832d77 100644
>--- a/drivers/pci/pci-driver.c
>+++ b/drivers/pci/pci-driver.c
>@@ -1271,6 +1271,7 @@ static int pci_pm_runtime_resume(struct device *dev)
> 	 * D3cold when the bridge above it runtime suspended.
> 	 */
> 	pci_restore_standard_config(pci_dev);
>+	pci_fixup_device(pci_fixup_resume_early, pci_dev);
> 
> 	if (!pci_dev->driver)
> 		return 0;
>@@ -1278,7 +1279,6 @@ static int pci_pm_runtime_resume(struct device *dev)
> 	if (!pm || !pm->runtime_resume)
> 		return -ENOSYS;
> 
>-	pci_fixup_device(pci_fixup_resume_early, pci_dev);
> 	pci_enable_wake(pci_dev, PCI_D0, false);
> 	pci_fixup_device(pci_fixup_resume, pci_dev);
> 
>-- 
>2.16.2
>

Actions: View | Splinter Review

Attachments on bug 75985: 95562 | 128640 | 136369 | 136416 | 136418 | 137764 | 137939 | 137940 | 137941 | 137989 | 144596 | 145614 | 145615 | 145616 | 145627 | 145640 | 145778 | 145830

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.