Re: [PATCH v2 2/4] PCI: Move PCIe ports to D3 during suspend

[Lukas Wunner <lukas@wunner.de>]

Hi Mika,

On Fri, Apr 08, 2016 at 01:36:28PM +0100, Mika Westerberg wrote:
> --- a/drivers/pci/pci-driver.c
> +++ b/drivers/pci/pci-driver.c
> @@ -744,6 +744,19 @@ static int pci_pm_suspend(struct device *dev)
>  	return 0;
>  }
>  
> +/*
> + * Check if given device can go to low power state. Currently we allow
> + * normal PCI devices and PCI bridges if their bridge_d3 is set.
> + */
> +static bool pci_can_suspend(struct pci_dev *pdev)
> +{
> +	if (!pci_has_subordinate(pdev))
> +		return true;
> +	else if (pdev->bridge_d3)
> +		return true;
> +	return false;
> +}
> +
>  static int pci_pm_suspend_noirq(struct device *dev)
>  {
>  	struct pci_dev *pci_dev = to_pci_dev(dev);
> @@ -777,7 +790,7 @@ static int pci_pm_suspend_noirq(struct device *dev)
>  
>  	if (!pci_dev->state_saved) {
>  		pci_save_state(pci_dev);
> -		if (!pci_has_subordinate(pci_dev))
> +		if (pci_can_suspend(pci_dev))
>  			pci_prepare_to_sleep(pci_dev);
>  	}

pci_can_suspend() is only used by this single function. It may be
worth to consider folding it into pci_pm_suspend_noirq(), i.e. simply

		if (!pci_has_subordinate(pci_dev) || pdev->bridge_d3)

together with the "Currently we allow..." comment above.

Best regards,

Lukas