Merge tag 'amd-drm-fixes-6.17-2025-09-18' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.17-2025-09-18:

amdgpu:
- GC 11.0.1/4 cleaner shader support
- DC irq fix
- OD fix

amdkfd:
- S0ix fix

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://lore.kernel.org/r/20250918191428.2553105-1-alexander.deucher@amd.com

Author: airlied

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c

@@ -250,16 +250,24 @@ void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
 
 void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool suspend_proc)
 {
-	if (adev->kfd.dev)
-		kgd2kfd_suspend(adev->kfd.dev, suspend_proc);
+	if (adev->kfd.dev) {
+		if (adev->in_s0ix)
+			kgd2kfd_stop_sched_all_nodes(adev->kfd.dev);
+		else
+			kgd2kfd_suspend(adev->kfd.dev, suspend_proc);
+	}
 }
 
 int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool resume_proc)
 {
 	int r = 0;
 
-	if (adev->kfd.dev)
-		r = kgd2kfd_resume(adev->kfd.dev, resume_proc);
+	if (adev->kfd.dev) {
+		if (adev->in_s0ix)
+			r = kgd2kfd_start_sched_all_nodes(adev->kfd.dev);
+		else
+			r = kgd2kfd_resume(adev->kfd.dev, resume_proc);
+	}
 
 	return r;
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h

@@ -426,7 +426,9 @@ void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint64_t throttle_bitmask);
 int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd);
 void kgd2kfd_unlock_kfd(struct kfd_dev *kfd);
 int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id);
+int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd);
 int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id);
+int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd);
 bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id);
 bool kgd2kfd_vmfault_fast_path(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry,
 			       bool retry_fault);
@@ -516,11 +518,21 @@ static inline int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id)
 	return 0;
 }
 
+static inline int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd)
+{
+	return 0;
+}
+
 static inline int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id)
 {
 	return 0;
 }
 
+static inline int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd)
+{
+	return 0;
+}
+
 static inline bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id)
 {
 	return false;

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -5136,7 +5136,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients)
 	adev->in_suspend = true;
 
 	if (amdgpu_sriov_vf(adev)) {
-		if (!adev->in_s0ix && !adev->in_runpm)
+		if (!adev->in_runpm)
 			amdgpu_amdkfd_suspend_process(adev);
 		amdgpu_virt_fini_data_exchange(adev);
 		r = amdgpu_virt_request_full_gpu(adev, false);
@@ -5156,10 +5156,8 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients)
 
 	amdgpu_device_ip_suspend_phase1(adev);
 
-	if (!adev->in_s0ix) {
-		amdgpu_amdkfd_suspend(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
-		amdgpu_userq_suspend(adev);
-	}
+	amdgpu_amdkfd_suspend(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
+	amdgpu_userq_suspend(adev);
 
 	r = amdgpu_device_evict_resources(adev);
 	if (r)
@@ -5254,15 +5252,13 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients)
 		goto exit;
 	}
 
-	if (!adev->in_s0ix) {
-		r = amdgpu_amdkfd_resume(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
-		if (r)
-			goto exit;
+	r = amdgpu_amdkfd_resume(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);
+	if (r)
+		goto exit;
 
-		r = amdgpu_userq_resume(adev);
-		if (r)
-			goto exit;
-	}
+	r = amdgpu_userq_resume(adev);
+	if (r)
+		goto exit;
 
 	r = amdgpu_device_ip_late_init(adev);
 	if (r)
@@ -5275,7 +5271,7 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients)
 		amdgpu_virt_init_data_exchange(adev);
 		amdgpu_virt_release_full_gpu(adev, true);
 
-		if (!adev->in_s0ix && !r && !adev->in_runpm)
+		if (!r && !adev->in_runpm)
 			r = amdgpu_amdkfd_resume_process(adev);
 	}
 

drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c

@@ -1654,6 +1654,21 @@ static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block)
 			}
 		}
 		break;
+	case IP_VERSION(11, 0, 1):
+	case IP_VERSION(11, 0, 4):
+		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;
+		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);
+		if (adev->gfx.pfp_fw_version >= 102 &&
+		    adev->gfx.mec_fw_version >= 66 &&
+		    adev->mes.fw_version[0] >= 128) {
+			adev->gfx.enable_cleaner_shader = true;
+			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);
+			if (r) {
+				adev->gfx.enable_cleaner_shader = false;
+				dev_err(adev->dev, "Failed to initialize cleaner shader\n");
+			}
+		}
+		break;
 	case IP_VERSION(11, 5, 0):
 	case IP_VERSION(11, 5, 1):
 		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;

drivers/gpu/drm/amd/amdkfd/kfd_device.c

@@ -1550,6 +1550,25 @@ int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id)
 	return ret;
 }
 
+int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd)
+{
+	struct kfd_node *node;
+	int i, r;
+
+	if (!kfd->init_complete)
+		return 0;
+
+	for (i = 0; i < kfd->num_nodes; i++) {
+		node = kfd->nodes[i];
+		r = node->dqm->ops.unhalt(node->dqm);
+		if (r) {
+			dev_err(kfd_device, "Error in starting scheduler\n");
+			return r;
+		}
+	}
+	return 0;
+}
+
 int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id)
 {
 	struct kfd_node *node;
@@ -1567,6 +1586,23 @@ int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id)
 	return node->dqm->ops.halt(node->dqm);
 }
 
+int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd)
+{
+	struct kfd_node *node;
+	int i, r;
+
+	if (!kfd->init_complete)
+		return 0;
+
+	for (i = 0; i < kfd->num_nodes; i++) {
+		node = kfd->nodes[i];
+		r = node->dqm->ops.halt(node->dqm);
+		if (r)
+			return r;
+	}
+	return 0;
+}
+
 bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id)
 {
 	struct kfd_node *node;

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c

@@ -8717,7 +8717,16 @@ static int amdgpu_dm_encoder_init(struct drm_device *dev,
 static void manage_dm_interrupts(struct amdgpu_device *adev,
 				 struct amdgpu_crtc *acrtc,
 				 struct dm_crtc_state *acrtc_state)
-{
+{	/*
+	 * We cannot be sure that the frontend index maps to the same
+	 * backend index - some even map to more than one.
+	 * So we have to go through the CRTC to find the right IRQ.
+	 */
+	int irq_type = amdgpu_display_crtc_idx_to_irq_type(
+			adev,
+			acrtc->crtc_id);
+	struct drm_device *dev = adev_to_drm(adev);
+
 	struct drm_vblank_crtc_config config = {0};
 	struct dc_crtc_timing *timing;
 	int offdelay;
@@ -8770,7 +8779,35 @@ static void manage_dm_interrupts(struct amdgpu_device *adev,
 
 		drm_crtc_vblank_on_config(&acrtc->base,
 					  &config);
+		/* Allow RX6xxx, RX7700, RX7800 GPUs to call amdgpu_irq_get.*/
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(3, 0, 0):
+		case IP_VERSION(3, 0, 2):
+		case IP_VERSION(3, 0, 3):
+		case IP_VERSION(3, 2, 0):
+			if (amdgpu_irq_get(adev, &adev->pageflip_irq, irq_type))
+				drm_err(dev, "DM_IRQ: Cannot get pageflip irq!\n");
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+			if (amdgpu_irq_get(adev, &adev->vline0_irq, irq_type))
+				drm_err(dev, "DM_IRQ: Cannot get vline0 irq!\n");
+#endif
+		}
+
 	} else {
+		/* Allow RX6xxx, RX7700, RX7800 GPUs to call amdgpu_irq_put.*/
+		switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) {
+		case IP_VERSION(3, 0, 0):
+		case IP_VERSION(3, 0, 2):
+		case IP_VERSION(3, 0, 3):
+		case IP_VERSION(3, 2, 0):
+#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY)
+			if (amdgpu_irq_put(adev, &adev->vline0_irq, irq_type))
+				drm_err(dev, "DM_IRQ: Cannot put vline0 irq!\n");
+#endif
+			if (amdgpu_irq_put(adev, &adev->pageflip_irq, irq_type))
+				drm_err(dev, "DM_IRQ: Cannot put pageflip irq!\n");
+		}
+
 		drm_crtc_vblank_off(&acrtc->base);
 	}
 }

drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c

@@ -2236,7 +2236,7 @@ static int smu_resume(struct amdgpu_ip_block *ip_block)
 			return ret;
 	}
 
-	if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_MANUAL) {
+	if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_MANUAL && smu->od_enabled) {
 		ret = smu_od_edit_dpm_table(smu, PP_OD_COMMIT_DPM_TABLE, NULL, 0);
 		if (ret)
 			return ret;