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

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

[Rafael J. Wysocki]

On 2/24/2022 9:29 PM, 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.
>
> 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 | 15 ++-------------
>   1 file changed, 2 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> index 588588cfda48..ac3f7e1676a9 100644
> --- a/drivers/pci/pci-driver.c
> +++ b/drivers/pci/pci-driver.c
> @@ -834,20 +834,9 @@ static int pci_pm_suspend_noirq(struct device *dev)
>   		}
>   	}
>   
> -	if (pci_dev->skip_bus_pm) {
> -		/*
> -		 * 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.
> -		 */

The comment is outdated, but the role of skip_bus_pm needs to be 
explained.  It only applies to the bridge with a child in D0 case now, 
because the other one has gone away, so I would put a new comment before 
the reworked check below.


> -		if (!pci_dev->state_saved)
> -			pci_save_state(pci_dev);
> -	} else if (!pci_dev->state_saved) {
> +	if (!pci_dev->state_saved) {
>   		pci_save_state(pci_dev);
> -		if (pci_power_manageable(pci_dev))

For example:

/*

  * 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->skip_bus_pm && pci_power_manageable(pci_dev))
>   			pci_prepare_to_sleep(pci_dev);
>   	}
>   

Re: [PATCH 2/2] PCI/PM: Fix pci_pm_suspend_noirq() to disable PTM

[Rafael J. Wysocki]

On 2/24/2022 9:29 PM, Rajvi Jingar wrote:
> For the PCIe devices (like nvme) that do not go into D3 state still need to
> disable PTM on PCIe root ports to allow the port to enter a lower-power PM
> state and the SoC to reach a lower-power idle state as a whole. Move the
> pci_disable_ptm() out of pci_prepare_to_sleep() as this code path is not
> followed for devices that do not go into D3. This patch fixes the issue
> seen on Dell XPS 9300 with Ice Lake CPU and Dell Precision 5530 with Coffee
> Lake CPU platforms to get improved residency in low power idle states.
>
> Signed-off-by: Rajvi Jingar <rajvi.jingar@intel.com>
> Suggested-by: David E. Box <david.e.box@linux.intel.com>

I would add a Fixes tag pointing to the commit that introduced 
pci_disable_ptm().

Otherwise I agree with this change:

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


> ---
>   drivers/pci/pci-driver.c | 11 +++++++++++
>   drivers/pci/pci.c        | 10 ----------
>   2 files changed, 11 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> index ac3f7e1676a9..8be3f81afdf6 100644
> --- a/drivers/pci/pci-driver.c
> +++ b/drivers/pci/pci-driver.c
> @@ -836,6 +836,17 @@ static int pci_pm_suspend_noirq(struct device *dev)
>   
>   	if (!pci_dev->state_saved) {
>   		pci_save_state(pci_dev);
> +
> +		/*
> +		 * There are systems (for example, Intel mobile chips since Coffee
> +		 * Lake) where the power drawn while suspended can be significantly
> +		 * reduced by disabling PTM on PCIe root ports as this allows the
> +		 * port to enter a lower-power PM state and the SoC to reach a
> +		 * lower-power idle state as a whole.
> +		 */
> +		if (pci_pcie_type(pci_dev) == PCI_EXP_TYPE_ROOT_PORT)
> +			pci_disable_ptm(pci_dev);
> +
>   		if (!pci_dev->skip_bus_pm && pci_power_manageable(pci_dev))
>   			pci_prepare_to_sleep(pci_dev);
>   	}
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 9ecce435fb3f..f8768672c064 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -2660,16 +2660,6 @@ int pci_prepare_to_sleep(struct pci_dev *dev)
>   	if (target_state == PCI_POWER_ERROR)
>   		return -EIO;
>   
> -	/*
> -	 * There are systems (for example, Intel mobile chips since Coffee
> -	 * Lake) where the power drawn while suspended can be significantly
> -	 * reduced by disabling PTM on PCIe root ports as this allows the
> -	 * port to enter a lower-power PM state and the SoC to reach a
> -	 * lower-power idle state as a whole.
> -	 */
> -	if (pci_pcie_type(dev) == PCI_EXP_TYPE_ROOT_PORT)
> -		pci_disable_ptm(dev);
> -
>   	pci_enable_wake(dev, target_state, wakeup);
>   
>   	error = pci_set_power_state(dev, target_state);

[PATCH 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>
---
 This patch is a resend of https://lkml.org/lkml/2022/6/8/1462
---
 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