Re: [PATCH v2 1/2] PCI/PM: refactor pci_pm_suspend_noirq()

Re: [PATCH v2 1/2] PCI/PM: refactor pci_pm_suspend_noirq()

[David E. Box]

Hi Rajvi,

On Thu, 2022-03-17 at 16:31 -0700, Rajvi Jingar wrote:
> The state of the device is saved during pci_pm_suspend_noirq(), if it
> has not already been saved, regardless of the skip_bus_pm flag value. So
> skip_bus_pm check is removed before saving the device state.
> 
> v2: add comments to the changes

...

> 
> Signed-off-by: Rajvi Jingar <rajvi.jingar@intel.com>
> Suggested-by: David E. Box <david.e.box@linux.intel.com>
> ---

Patch changelogs aren't kept in the commit message. Place them here after the
"---" line. In this location, it won't affect applying the patch.

David

>  drivers/pci/pci-driver.c | 18 ++++++------------
>  1 file changed, 6 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> index 588588cfda48..ffe76f238d7e 100644
> --- a/drivers/pci/pci-driver.c
> +++ b/drivers/pci/pci-driver.c
> @@ -834,20 +834,14 @@ static int pci_pm_suspend_noirq(struct device *dev)
>  		}
>  	}
>  
> -	if (pci_dev->skip_bus_pm) {
> +	if (!pci_dev->state_saved) {
> +		pci_save_state(pci_dev);
>  		/*
> -		 * Either the device is a bridge with a child in D0 below it, or
> -		 * the function is running for the second time in a row without
> -		 * going through full resume, which is possible only during
> -		 * suspend-to-idle in a spurious wakeup case.  The device should
> -		 * be in D0 at this point, but if it is a bridge, it may be
> -		 * necessary to save its state.
> +		 * If the device is a bridge with a child in D0 below it, it
> needs to
> +		 * stay in D0, so check skip_bus_pm to avoid putting it into a
> +		 * low-power state in that case.
>  		 */
> -		if (!pci_dev->state_saved)
> -			pci_save_state(pci_dev);
> -	} else if (!pci_dev->state_saved) {
> -		pci_save_state(pci_dev);
> -		if (pci_power_manageable(pci_dev))
> +		if (!pci_dev->skip_bus_pm && pci_power_manageable(pci_dev))
>  			pci_prepare_to_sleep(pci_dev);
>  	}
>  

Re: [PATCH v2 1/2] PCI/PM: refactor pci_pm_suspend_noirq()

[Rafael J. Wysocki]

On 3/18/2022 12:31 AM, Rajvi Jingar wrote:
> The state of the device is saved during pci_pm_suspend_noirq(), if it
> has not already been saved, regardless of the skip_bus_pm flag value. So
> skip_bus_pm check is removed before saving the device state.
>
> v2: add comments to the changes

As mentioned by Dave, this shouldn't be here.


> Signed-off-by: Rajvi Jingar <rajvi.jingar@intel.com>
> Suggested-by: David E. Box <david.e.box@linux.intel.com>
> ---
>   drivers/pci/pci-driver.c | 18 ++++++------------
>   1 file changed, 6 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> index 588588cfda48..ffe76f238d7e 100644
> --- a/drivers/pci/pci-driver.c
> +++ b/drivers/pci/pci-driver.c
> @@ -834,20 +834,14 @@ static int pci_pm_suspend_noirq(struct device *dev)
>   		}
>   	}
>   
> -	if (pci_dev->skip_bus_pm) {
> +	if (!pci_dev->state_saved) {
> +		pci_save_state(pci_dev);
>   		/*
> -		 * Either the device is a bridge with a child in D0 below it, or
> -		 * the function is running for the second time in a row without
> -		 * going through full resume, which is possible only during
> -		 * suspend-to-idle in a spurious wakeup case.  The device should
> -		 * be in D0 at this point, but if it is a bridge, it may be
> -		 * necessary to save its state.
> +		 * If the device is a bridge with a child in D0 below it, it needs to
> +		 * stay in D0, so check skip_bus_pm to avoid putting it into a
> +		 * low-power state in that case.
>   		 */
> -		if (!pci_dev->state_saved)
> -			pci_save_state(pci_dev);
> -	} else if (!pci_dev->state_saved) {
> -		pci_save_state(pci_dev);
> -		if (pci_power_manageable(pci_dev))
> +		if (!pci_dev->skip_bus_pm && pci_power_manageable(pci_dev))
>   			pci_prepare_to_sleep(pci_dev);
>   	}
>   

[PATCH v2 1/2] PCI/PM: refactor pci_pm_suspend_noirq()

[Rajvi Jingar]

The state of the device is saved during pci_pm_suspend_noirq(), if it
has not already been saved, regardless of the skip_bus_pm flag value. So
skip_bus_pm check is removed before saving the device state.

Signed-off-by: Rajvi Jingar <rajvi.jingar@linux.intel.com>
Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 v1->v2: no change
---
 drivers/pci/pci-driver.c | 18 ++++++------------
 1 file changed, 6 insertions(+), 12 deletions(-)

diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
index 49238ddd39ee..1f64de3e5280 100644
--- a/drivers/pci/pci-driver.c
+++ b/drivers/pci/pci-driver.c
@@ -867,20 +867,14 @@ static int pci_pm_suspend_noirq(struct device *dev)
 		}
 	}
 
-	if (pci_dev->skip_bus_pm) {
+	if (!pci_dev->state_saved) {
+		pci_save_state(pci_dev);
 		/*
-		 * Either the device is a bridge with a child in D0 below it, or
-		 * the function is running for the second time in a row without
-		 * going through full resume, which is possible only during
-		 * suspend-to-idle in a spurious wakeup case.  The device should
-		 * be in D0 at this point, but if it is a bridge, it may be
-		 * necessary to save its state.
+		 * If the device is a bridge with a child in D0 below it, it needs to
+		 * stay in D0, so check skip_bus_pm to avoid putting it into a
+		 * low-power state in that case.
 		 */
-		if (!pci_dev->state_saved)
-			pci_save_state(pci_dev);
-	} else if (!pci_dev->state_saved) {
-		pci_save_state(pci_dev);
-		if (pci_power_manageable(pci_dev))
+		if (!pci_dev->skip_bus_pm && pci_power_manageable(pci_dev))
 			pci_prepare_to_sleep(pci_dev);
 	}
 
-- 
2.25.1

[PATCH v2 2/2] PCI/PTM: fix to maintain pci_dev->ptm_enabled

[Rajvi Jingar]

pci_dev->ptm_enabled needs to be maintained to reflect the current PTM
state of the device. In pci_ptm_disable(), clear ptm_enabled from
'struct pci_dev' on disabling PTM state for the device.
In pci_restore_ptm_state(), set dev->ptm_enabled based on the restored
PTM state of the device.

Also, perform ptm_enabled check in pci_ptm_disable() to avoid config
space access in case if PTM is already disabled for the device.

Signed-off-by: Rajvi Jingar <rajvi.jingar@linux.intel.com>
---
 v1->v2:
   - add ptm_enabled check in pci_ptm_disable().
   - set the dev->ptm_enabled value in pci_restore_ptm_state().
---
 drivers/pci/pcie/ptm.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/pci/pcie/ptm.c b/drivers/pci/pcie/ptm.c
index 368a254e3124..8dc7d2285a0d 100644
--- a/drivers/pci/pcie/ptm.c
+++ b/drivers/pci/pcie/ptm.c
@@ -34,6 +34,9 @@ void pci_disable_ptm(struct pci_dev *dev)
 	int ptm;
 	u16 ctrl;
 
+	if (!dev->ptm_enabled)
+		return;
+
 	if (!pci_is_pcie(dev))
 		return;
 
@@ -44,6 +47,7 @@ void pci_disable_ptm(struct pci_dev *dev)
 	pci_read_config_word(dev, ptm + PCI_PTM_CTRL, &ctrl);
 	ctrl &= ~(PCI_PTM_CTRL_ENABLE | PCI_PTM_CTRL_ROOT);
 	pci_write_config_word(dev, ptm + PCI_PTM_CTRL, ctrl);
+	dev->ptm_enabled = 0;
 }
 
 void pci_save_ptm_state(struct pci_dev *dev)
@@ -83,6 +87,7 @@ void pci_restore_ptm_state(struct pci_dev *dev)
 
 	cap = (u16 *)&save_state->cap.data[0];
 	pci_write_config_word(dev, ptm + PCI_PTM_CTRL, *cap);
+	dev->ptm_enabled = !!(*cap & PCI_PTM_CTRL_ENABLE);
 }
 
 void pci_ptm_init(struct pci_dev *dev)
-- 
2.25.1

Re: [PATCH v2 2/2] PCI/PTM: fix to maintain pci_dev->ptm_enabled

[Rafael J. Wysocki]

On Fri, Jul 15, 2022 at 1:16 AM Rajvi Jingar
<rajvi.jingar@linux.intel.com> wrote:
>
> pci_dev->ptm_enabled needs to be maintained to reflect the current PTM
> state of the device. In pci_ptm_disable(), clear ptm_enabled from
> 'struct pci_dev' on disabling PTM state for the device.
> In pci_restore_ptm_state(), set dev->ptm_enabled based on the restored
> PTM state of the device.
>
> Also, perform ptm_enabled check in pci_ptm_disable() to avoid config
> space access in case if PTM is already disabled for the device.
>
> Signed-off-by: Rajvi Jingar <rajvi.jingar@linux.intel.com>
> ---
>  v1->v2:
>    - add ptm_enabled check in pci_ptm_disable().
>    - set the dev->ptm_enabled value in pci_restore_ptm_state().
> ---
>  drivers/pci/pcie/ptm.c | 5 +++++
>  1 file changed, 5 insertions(+)
>
> diff --git a/drivers/pci/pcie/ptm.c b/drivers/pci/pcie/ptm.c
> index 368a254e3124..8dc7d2285a0d 100644
> --- a/drivers/pci/pcie/ptm.c
> +++ b/drivers/pci/pcie/ptm.c
> @@ -34,6 +34,9 @@ void pci_disable_ptm(struct pci_dev *dev)
>         int ptm;
>         u16 ctrl;
>
> +       if (!dev->ptm_enabled)
> +               return;
> +
>         if (!pci_is_pcie(dev))
>                 return;

After the check you've added this doesn't need to be checked any more,
because ptm_enabled will not be set if pci_is_pcie(dev) is not true
AFAICS.

Otherwise the patch looks good to me, so with the above addressed,
please feel free to add

Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

to it.

Thanks!

>
> @@ -44,6 +47,7 @@ void pci_disable_ptm(struct pci_dev *dev)
>         pci_read_config_word(dev, ptm + PCI_PTM_CTRL, &ctrl);
>         ctrl &= ~(PCI_PTM_CTRL_ENABLE | PCI_PTM_CTRL_ROOT);
>         pci_write_config_word(dev, ptm + PCI_PTM_CTRL, ctrl);
> +       dev->ptm_enabled = 0;
>  }
>
>  void pci_save_ptm_state(struct pci_dev *dev)
> @@ -83,6 +87,7 @@ void pci_restore_ptm_state(struct pci_dev *dev)
>
>         cap = (u16 *)&save_state->cap.data[0];
>         pci_write_config_word(dev, ptm + PCI_PTM_CTRL, *cap);
> +       dev->ptm_enabled = !!(*cap & PCI_PTM_CTRL_ENABLE);
>  }
>
>  void pci_ptm_init(struct pci_dev *dev)
> --
> 2.25.1
>