[PATCH 2/2] drm/amdgpu: fix pm notifier handling

[Mario Limonciello]

On 5/5/2025 10:30 AM, Alex Deucher wrote:
> Set the s3/s0ix and s4 flags in the pm notifier so that we can skip
> the resource evictions properly in pm prepare based on whether
> we are suspending or hibernating.  Drop the eviction as processes
> are not frozen at this time, we we can end up getting stuck trying
> to evict VRAM while applications continue to submit work which
> causes the buffers to get pulled back into VRAM.
> 
> v2: Move suspend flags out of pm notifier (Mario)
> 
> Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4178
> Fixes: 2965e6355dcd ("drm/amd: Add Suspend/Hibernate notification callback support")
> Cc: Mario Limonciello <mario.limonciello at amd.com>
> Signed-off-by: Alex Deucher <alexander.deucher at amd.com>

Until we can get movement on the RFC for the extra eviction step this is 
fine for now.  I hope we can get that soon though.

For both:
Reviewed-by: Mario Limonciello <mario.limonciello at amd.com>

> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 18 +++++-------------
>   drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c    | 10 +---------
>   2 files changed, 6 insertions(+), 22 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index 368004b716ba8..5d25db16cb4b8 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -4979,28 +4979,20 @@ static int amdgpu_device_evict_resources(struct amdgpu_device *adev)
>    * @data: data
>    *
>    * This function is called when the system is about to suspend or hibernate.
> - * It is used to evict resources from the device before the system goes to
> - * sleep while there is still access to swap.
> + * It is used to set the appropriate flags so that eviction can be optimized
> + * in the pm prepare callback.
>    */
>   static int amdgpu_device_pm_notifier(struct notifier_block *nb, unsigned long mode,
>   				     void *data)
>   {
>   	struct amdgpu_device *adev = container_of(nb, struct amdgpu_device, pm_nb);
> -	int r;
>   
>   	switch (mode) {
>   	case PM_HIBERNATION_PREPARE:
>   		adev->in_s4 = true;
> -		fallthrough;
> -	case PM_SUSPEND_PREPARE:
> -		r = amdgpu_device_evict_resources(adev);
> -		/*
> -		 * This is considered non-fatal at this time because
> -		 * amdgpu_device_prepare() will also fatally evict resources.
> -		 * See https://gitlab.freedesktop.org/drm/amd/-/issues/3781
> -		 */
> -		if (r)
> -			drm_warn(adev_to_drm(adev), "Failed to evict resources, freeze active processes if problems occur: %d\n", r);
> +		break;
> +	case PM_POST_HIBERNATION:
> +		adev->in_s4 = false;
>   		break;
>   	}
>   
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> index 8baeaa5d5b119..e0ca9405719e8 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
> @@ -2643,13 +2643,8 @@ static int amdgpu_pmops_freeze(struct device *dev)
>   static int amdgpu_pmops_thaw(struct device *dev)
>   {
>   	struct drm_device *drm_dev = dev_get_drvdata(dev);
> -	struct amdgpu_device *adev = drm_to_adev(drm_dev);
> -	int r;
> -
> -	r = amdgpu_device_resume(drm_dev, true);
> -	adev->in_s4 = false;
>   
> -	return r;
> +	return amdgpu_device_resume(drm_dev, true);
>   }
>   
>   static int amdgpu_pmops_poweroff(struct device *dev)
> @@ -2662,9 +2657,6 @@ static int amdgpu_pmops_poweroff(struct device *dev)
>   static int amdgpu_pmops_restore(struct device *dev)
>   {
>   	struct drm_device *drm_dev = dev_get_drvdata(dev);
> -	struct amdgpu_device *adev = drm_to_adev(drm_dev);
> -
> -	adev->in_s4 = false;
>   
>   	return amdgpu_device_resume(drm_dev, true);
>   }