drm/amd: Evict resources during PM ops prepare() callback

Linux PM core has a prepare() callback run before suspend.

If the system is under high memory pressure, the resources may need
to be evicted into swap instead.  If the storage backing for swap
is offlined during the suspend() step then such a call may fail.

So move this step into prepare() to move evict majority of
resources and update all non-pmops callers to call the same callback.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2362
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 6b89d92e22a47e6311e16a006d64908a6247d10b..17c4872a028774a48919bb5f320a0bce324c83e7 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -1408,6 +1408,7 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
 void amdgpu_driver_release_kms(struct drm_device *dev);
 
 int amdgpu_device_ip_suspend(struct amdgpu_device *adev);
+int amdgpu_device_prepare(struct drm_device *dev);
 int amdgpu_device_suspend(struct drm_device *dev, bool fbcon);
 int amdgpu_device_resume(struct drm_device *dev, bool fbcon);
 u32 amdgpu_get_vblank_counter_kms(struct drm_crtc *crtc);

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index d5b81a086e699ab365b9ffa0169c3c9cf3e015c5..0a7cc9107f072a5a4eaad6785d7a775c5dd4db8a 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -1760,6 +1760,7 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev,
        } else {
                pr_info("switched off\n");
                dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
+               amdgpu_device_prepare(dev);
                amdgpu_device_suspend(dev, true);
                amdgpu_device_cache_pci_state(pdev);
                /* Shut down the device */
@@ -4335,6 +4336,31 @@ static int amdgpu_device_evict_resources(struct amdgpu_device *adev)
 /*
  * Suspend & resume.
  */
+/**
+ * amdgpu_device_prepare - prepare for device suspend
+ *
+ * @dev: drm dev pointer
+ *
+ * Prepare to put the hw in the suspend state (all asics).
+ * Returns 0 for success or an error on failure.
+ * Called at driver suspend.
+ */
+int amdgpu_device_prepare(struct drm_device *dev)
+{
+       struct amdgpu_device *adev = drm_to_adev(dev);
+       int r;
+
+       if (dev->switch_power_state == DRM_SWITCH_POWER_OFF)
+               return 0;
+
+       /* Evict the majority of BOs before starting suspend sequence */
+       r = amdgpu_device_evict_resources(adev);
+       if (r)
+               return r;
+
+       return 0;
+}
+
 /**
  * amdgpu_device_suspend - initiate device suspend
  *
@@ -4355,11 +4381,6 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon)
 
        adev->in_suspend = true;
 
-       /* Evict the majority of BOs before grabbing the full access */
-       r = amdgpu_device_evict_resources(adev);
-       if (r)
-               return r;
-
        if (amdgpu_sriov_vf(adev)) {
                amdgpu_virt_fini_data_exchange(adev);
                r = amdgpu_virt_request_full_gpu(adev, false);

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 7fd3826c7b92b1e52348c26829ce78f590001a74..6cc6e3991410ed476e5dbf33c039f18705cb7430 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -2427,8 +2427,9 @@ static int amdgpu_pmops_prepare(struct device *dev)
        /* Return a positive number here so
         * DPM_FLAG_SMART_SUSPEND works properly
         */
-       if (amdgpu_device_supports_boco(drm_dev))
-               return pm_runtime_suspended(dev);
+       if (amdgpu_device_supports_boco(drm_dev) &&
+           pm_runtime_suspended(dev))
+               return 1;
 
        /* if we will not support s3 or s2i for the device
         *  then skip suspend
@@ -2437,7 +2438,7 @@ static int amdgpu_pmops_prepare(struct device *dev)
            !amdgpu_acpi_is_s3_active(adev))
                return 1;
 
-       return 0;
+       return amdgpu_device_prepare(drm_dev);
 }
 
 static void amdgpu_pmops_complete(struct device *dev)
@@ -2637,6 +2638,9 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev)
        if (amdgpu_device_supports_boco(drm_dev))
                adev->mp1_state = PP_MP1_STATE_UNLOAD;
 
+       ret = amdgpu_device_prepare(drm_dev);
+       if (ret)
+               return ret;
        ret = amdgpu_device_suspend(drm_dev, false);
        if (ret) {
                adev->in_runpm = false;