Re: [PATCH v2 1/2] drm/xe: Inspect all forcewake bits for runtime idleness

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The runtime PM autosuspend retry mechanism is broken, causing unconditional suspension even when forcewake bits are set.
--

commit 87cc16a66c9eb41f1b093edd9b83a3cac1ad67b5
Author: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
drm/xe: Inspect all forcewake bits for runtime idleness

This commit adds checks for all GT force wake bits in the idle check
for runtime suspend to avoid suspending while the GT is kept awake. 
It also adds a call to pm_runtime_autosuspend() inside the idle check
to ensure a PM subsystem retry at a later time.

> diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
> index 1e04e8ef2611f..278cfd390364c 100644
> --- a/drivers/gpu/drm/xe/xe_pci.c
> +++ b/drivers/gpu/drm/xe/xe_pci.c

[ ... ]

> @@ -1437,10 +1437,14 @@ static int xe_pci_runtime_idle(struct device *dev)
>  {
>  	struct pci_dev *pdev = to_pci_dev(dev);
>  	struct xe_device *xe = pdev_to_xe_device(pdev);
> +	int ret;
>  
>  	xe_pm_d3cold_allowed_toggle(xe);
>  
> -	return 0;
> +	ret = xe_pm_check_runtime_idle(xe);
> +	pm_runtime_autosuspend(dev);

[Severity: High]
Does this cause an immediate suspension rather than a delayed retry?

When xe_pm_check_runtime_idle() detects forcewake bits and returns -EBUSY,
pm_runtime_autosuspend() is called to queue a retry. However, because
pm_runtime_mark_last_busy() is not called to update the timestamp, the PM
core calculates that the autosuspend timer is already expired.

This causes rpm_suspend() to execute immediately. Since the
xe_pci_runtime_suspend() callback does not re-check idleness, the device
appears to unconditionally suspend while forcewake bits are still active.

> +
> +	return ret;
>  }
>  
>  static const struct dev_pm_ops xe_pm_ops = {

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260910212823.2358362-1-vinay.belgaumkar@intel.com?part=1