linux-kernel - Re: Time to re-enable Runtime PM per default for PCI devcies?

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <5c618c30-b389-4545-e9c4-9fcec729b80d@gmail.com>
Date:   Thu, 31 Dec 2020 10:52:58 +0100
From:   Heiner Kallweit <hkallweit1@...il.com>
To:     Lukas Wunner <lukas@...ner.de>
Cc:     "Rafael J. Wysocki" <rafael@...nel.org>,
        Bjorn Helgaas <helgaas@...nel.org>,
        "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Bjorn Helgaas <bhelgaas@...gle.com>,
        Mika Westerberg <mika.westerberg@...ux.intel.com>,
        Kai Heng Feng <kai.heng.feng@...onical.com>,
        "linux-pci@...r.kernel.org" <linux-pci@...r.kernel.org>,
        Linux PM <linux-pm@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: Time to re-enable Runtime PM per default for PCI devcies?

On 31.12.2020 10:38, Heiner Kallweit wrote:
> On 31.12.2020 05:07, Lukas Wunner wrote:
>> On Wed, Dec 30, 2020 at 11:56:04PM +0100, Heiner Kallweit wrote:
>>> --- a/drivers/pci/pci.c
>>> +++ b/drivers/pci/pci.c
>>> @@ -3024,7 +3024,9 @@ void pci_pm_init(struct pci_dev *dev)
>>>  	u16 status;
>>>  	u16 pmc;
>>>  
>>> -	pm_runtime_forbid(&dev->dev);
>>> +	if (pci_acpi_forbid_runtime_pm())
>>> +		pm_runtime_forbid(&dev->dev);
>>> +
>>
>> Generic PCI code usually does not call ACPI-specific functions directly,
>> but rather through a pci_platform_pm_ops callback.
>>
>> FWIW, if platform_pci_power_manageable() returns true, it can probably
>> be assumed that allowing runtime PM by default is okay.  So as a first
>> step, you may want to call that instead of adding a new callback.
>>
> I don't think that's sufficient. Most likely all the broken old systems
> return true for platform_pci_power_manageable(). So yes, most likely
> we'd need a new callback if we want to have the platform ops abstraction.
> But it could be an optional callback, something like: forbid_runtime_pm
> The question is just: is it worth it?
> 
> By the way: pci_set_platform_pm() returns an error if a callback isn't
> set, but no existing caller bothers to check the return code.
> 
>> Thanks,
>>
>> Lukas
>>
> Heiner
> 

This would be the version with the abstraction.

---
 drivers/pci/pci-acpi.c | 6 ++++++
 drivers/pci/pci.c      | 9 ++++++++-
 drivers/pci/pci.h      | 5 ++++-
 3 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/drivers/pci/pci-acpi.c b/drivers/pci/pci-acpi.c
index 53502a751..b57a79af7 100644
--- a/drivers/pci/pci-acpi.c
+++ b/drivers/pci/pci-acpi.c
@@ -1108,6 +1108,11 @@ static bool acpi_pci_need_resume(struct pci_dev *dev)
 	return !!adev->power.flags.dsw_present;
 }
 
+static bool acpi_pci_forbid_runtime_pm(void)
+{
+	return acpi_gbl_FADT.header.revision < 6;
+}
+
 static const struct pci_platform_pm_ops acpi_pci_platform_pm = {
 	.bridge_d3 = acpi_pci_bridge_d3,
 	.is_manageable = acpi_pci_power_manageable,
@@ -1117,6 +1122,7 @@ static const struct pci_platform_pm_ops acpi_pci_platform_pm = {
 	.choose_state = acpi_pci_choose_state,
 	.set_wakeup = acpi_pci_wakeup,
 	.need_resume = acpi_pci_need_resume,
+	.forbid_runtime_pm = acpi_pci_forbid_runtime_pm,
 };
 
 void acpi_pci_add_bus(struct pci_bus *bus)
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index b9fecc25d..3af832b7f 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -982,6 +982,12 @@ static inline bool platform_pci_need_resume(struct pci_dev *dev)
 	return pci_platform_pm ? pci_platform_pm->need_resume(dev) : false;
 }
 
+static inline bool platform_pci_forbid_runtime_pm(void)
+{
+	return pci_platform_pm && pci_platform_pm->forbid_runtime_pm ?
+	       pci_platform_pm->forbid_runtime_pm() : false;
+}
+
 static inline bool platform_pci_bridge_d3(struct pci_dev *dev)
 {
 	if (pci_platform_pm && pci_platform_pm->bridge_d3)
@@ -3024,7 +3030,8 @@ void pci_pm_init(struct pci_dev *dev)
 	u16 status;
 	u16 pmc;
 
-	pm_runtime_forbid(&dev->dev);
+	if (platform_pci_forbid_runtime_pm())
+		pm_runtime_forbid(&dev->dev);
 	pm_runtime_set_active(&dev->dev);
 	pm_runtime_enable(&dev->dev);
 	device_enable_async_suspend(&dev->dev);
diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
index 5c5936509..d2d406ee4 100644
--- a/drivers/pci/pci.h
+++ b/drivers/pci/pci.h
@@ -71,8 +71,10 @@ int pci_bus_error_reset(struct pci_dev *dev);
  *		 suspended) needs to be resumed to be configured for system
  *		 wakeup.
  *
+ * @forbid_runtime_pm: returns true in case of known issues breaking runtime pm
+ *
  * If given platform is generally capable of power managing PCI devices, all of
- * these callbacks are mandatory.
+ * these callbacks are mandatory (except forbid_runtime_pm).
  */
 struct pci_platform_pm_ops {
 	bool (*bridge_d3)(struct pci_dev *dev);
@@ -83,6 +85,7 @@ struct pci_platform_pm_ops {
 	pci_power_t (*choose_state)(struct pci_dev *dev);
 	int (*set_wakeup)(struct pci_dev *dev, bool enable);
 	bool (*need_resume)(struct pci_dev *dev);
+	bool (*forbid_runtime_pm)(void);
 };
 
 int pci_set_platform_pm(const struct pci_platform_pm_ops *ops);
-- 
2.29.2