[PATCH] PCI PM: Read device power state from register after updating it (rev. 2)

[PATCH] PCI PM: Read device power state from register after updating it (rev. 2)

[Rafael J. Wysocki]

From: Rafael J. Wysocki <rjw@sisk.pl>

After attempting to change the power state of a PCI device
pci_raw_set_power_state() doesn't check if the value it wrote into
the device's PCI_PM_CTRL register has been stored in there, but
unconditionally modifies the device's current_state field to reflect
the change.  This may cause problems to happen if the power state of
the device hasn't been changed in fact, because it will make the PCI
PM core think that the device is in a power state it really is not
in.

To prevent such situations from happening modify
pci_raw_set_power_state() so that it reads the device's PCI_PM_CTRL
register after writing into it and uses the value read from the
register to update the device's current_state field.  Also make it
print a message saying that the device refused to change its power
state as requested (returning an error code in such cases would cause
suspend regressions to appear on some systems, where device drivers'
suspend routines return error codes if pci_set_power_state() fails).

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
---
 drivers/pci/pci.c |    7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/pci/pci.c
===================================================================
--- linux-2.6.orig/drivers/pci/pci.c
+++ linux-2.6/drivers/pci/pci.c
@@ -513,7 +513,12 @@ static int pci_raw_set_power_state(struc
 	else if (state == PCI_D2 || dev->current_state == PCI_D2)
 		udelay(PCI_PM_D2_DELAY);
 
-	dev->current_state = state;
+	pci_read_config_word(dev, dev->pm_cap + PCI_PM_CTRL, &pmcsr);
+	dev->current_state = (pmcsr & PCI_PM_CTRL_STATE_MASK);
+	/* Return error code if we have failed to change the state */
+	if (dev->current_state != state)
+		dev_info(&dev->dev, "Refused to change power state, "
+			"currently in D%d\n", dev->current_state);
 
 	/* According to section 5.4.1 of the "PCI BUS POWER MANAGEMENT
 	 * INTERFACE SPECIFICATION, REV. 1.2", a device transitioning

Re: [PATCH] PCI PM: Read device power state from register after updating it (rev. 2)

[Arjan van de Ven]

On Mon, 21 Sep 2009 23:37:01 +0200
"Rafael J. Wysocki" <rjw@sisk.pl> wrote:
>  
> -	dev->current_state = state;
> +	pci_read_config_word(dev, dev->pm_cap + PCI_PM_CTRL, &pmcsr);
> +	dev->current_state = (pmcsr & PCI_PM_CTRL_STATE_MASK);
> +	/* Return error code if we have failed to change the state */
> +	if (dev->current_state != state)
> +		dev_info(&dev->dev, "Refused to change power state, "
> +			"currently in D%d\n", dev->current_state);

I would suspect that you want this message only once...
to avoid it flooding logs and such

or at least ratelimiting it


-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

Re: [PATCH] PCI PM: Read device power state from register after updating it (rev. 2)

[Rafael J. Wysocki]

On Tuesday 22 September 2009, Arjan van de Ven wrote:
> On Mon, 21 Sep 2009 23:37:01 +0200
> "Rafael J. Wysocki" <rjw@sisk.pl> wrote:
> >  
> > -	dev->current_state = state;
> > +	pci_read_config_word(dev, dev->pm_cap + PCI_PM_CTRL, &pmcsr);
> > +	dev->current_state = (pmcsr & PCI_PM_CTRL_STATE_MASK);
> > +	/* Return error code if we have failed to change the state */
> > +	if (dev->current_state != state)
> > +		dev_info(&dev->dev, "Refused to change power state, "
> > +			"currently in D%d\n", dev->current_state);
> 
> I would suspect that you want this message only once...
> to avoid it flooding logs and such

Hmm?  I don't think we use pci_set_power_state() _that_ often.

> or at least ratelimiting it

That sounds like a good idea, though, will respin.

Re: [PATCH] PCI PM: Read device power state from register after updating it (rev. 2)

[Andreas Mohr]

Hi Rafael,

first you say "returning an error code in such cases would..."
and then the patch content has...?

Andreas Mohr

Re: [PATCH] PCI PM: Read device power state from register after updating it (rev. 2)

[Rafael J. Wysocki]

On Tuesday 22 September 2009, Andreas Mohr wrote:
> Hi Rafael,

Hi,

> first you say "returning an error code in such cases would..."
> and then the patch content has...?

I don't quite understand your comment, care to elaborate?

Rafael

Re: [PATCH] PCI PM: Read device power state from register after updating it (rev. 2)

[Rafael J. Wysocki]

On Tuesday 22 September 2009, kevin granade wrote:
> On Tue, Sep 22, 2009 at 2:50 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> 
> > On Tuesday 22 September 2009, Andreas Mohr wrote:
> > > Hi Rafael,
> >
> > Hi,
> >
> > > first you say "returning an error code in such cases would..."
> > > and then the patch content has...?
> >
> > I don't quite understand your comment, care to elaborate?
> >
> >
> I think this is what was being referenced:
> 
> +       /* Return error code if we have failed to change the state */
> +       if (dev->current_state != state)
> +               dev_info(&dev->dev, "Refused to change power state, "
> +                       "currently in D%d\n", dev->current_state);
> 
> The comment says, "Return error code", where it should probably say, "Log
> error".  Or possibly no comment at all.

Ah, thanks.

Rafael

[PATCH] PCI PM: Read device power state from register after updating it (rev. 3)

[Rafael J. Wysocki]

From: Rafael J. Wysocki <rjw@sisk.pl>

After attempting to change the power state of a PCI device
pci_raw_set_power_state() doesn't check if the value it wrote into
the device's PCI_PM_CTRL register has been stored in there, but
unconditionally modifies the device's current_state field to reflect
the change.  This may cause problems to happen if the power state of
the device hasn't been changed in fact, because it will make the PCI
PM core think that the device is in a power state it really is not
in.

To prevent such situations from happening modify
pci_raw_set_power_state() so that it reads the device's PCI_PM_CTRL
register after writing into it and uses the value read from the
register to update the device's current_state field.  Also make it
print a message saying that the device refused to change its power
state as requested (returning an error code in such cases would cause
suspend regressions to appear on some systems, where device drivers'
suspend routines return error codes if pci_set_power_state() fails).

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
---
Respin to fix issues pointed out by the reviewers.

Thanks,
Rafael
---
 drivers/pci/pci.c |    6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/pci/pci.c
===================================================================
--- linux-2.6.orig/drivers/pci/pci.c
+++ linux-2.6/drivers/pci/pci.c
@@ -513,7 +513,11 @@ static int pci_raw_set_power_state(struc
 	else if (state == PCI_D2 || dev->current_state == PCI_D2)
 		udelay(PCI_PM_D2_DELAY);
 
-	dev->current_state = state;
+	pci_read_config_word(dev, dev->pm_cap + PCI_PM_CTRL, &pmcsr);
+	dev->current_state = (pmcsr & PCI_PM_CTRL_STATE_MASK);
+	if (dev->current_state != state && printk_ratelimit())
+		dev_info(&dev->dev, "Refused to change power state, "
+			"currently in D%d\n", dev->current_state);
 
 	/* According to section 5.4.1 of the "PCI BUS POWER MANAGEMENT
 	 * INTERFACE SPECIFICATION, REV. 1.2", a device transitioning