Merge drm/drm-next into drm-misc-next

Backmerging to get a late RC of v6.10 before moving into v6.11. Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
2026-04-18 06:33:43 -04:00 · 2024-07-29 09:35:54 +02:00
parent 0c3f3a83c7 d4ef5d2b7e
commit 0e8655b4e8
2694 changed files with 387715 additions and 28887 deletions
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -679,7 +679,7 @@ uint32_t amdgpu_device_xcc_rreg(struct amdgpu_device *adev,
 		    amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags,
 							 GC_HWIP, false,
 							 &rlcg_flag)) {
-			ret = amdgpu_virt_rlcg_reg_rw(adev, reg, 0, rlcg_flag, xcc_id);
+			ret = amdgpu_virt_rlcg_reg_rw(adev, reg, 0, rlcg_flag, GET_INST(GC, xcc_id));
 		} else if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
 		    amdgpu_sriov_runtime(adev) &&
 		    down_read_trylock(&adev->reset_domain->sem)) {
@@ -810,7 +810,7 @@ void amdgpu_device_xcc_wreg(struct amdgpu_device *adev,
 		    amdgpu_virt_get_rlcg_reg_access_flag(adev, acc_flags,
 							 GC_HWIP, true,
 							 &rlcg_flag)) {
-			amdgpu_virt_rlcg_reg_rw(adev, reg, v, rlcg_flag, xcc_id);
+			amdgpu_virt_rlcg_reg_rw(adev, reg, v, rlcg_flag, GET_INST(GC, xcc_id));
 		} else if (!(acc_flags & AMDGPU_REGS_NO_KIQ) &&
 		    amdgpu_sriov_runtime(adev) &&
 		    down_read_trylock(&adev->reset_domain->sem)) {
@@ -1308,6 +1308,7 @@ static int amdgpu_device_asic_init(struct amdgpu_device *adev)
 	amdgpu_asic_pre_asic_init(adev);

 	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
 	    amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0)) {
 		amdgpu_psp_wait_for_bootloader(adev);
 		ret = amdgpu_atomfirmware_asic_init(adev, true);
@@ -2349,7 +2350,6 @@ void amdgpu_device_set_sriov_virtual_display(struct amdgpu_device *adev)
 static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 {
 	const char *chip_name;
-	char fw_name[40];
 	int err;
 	const struct gpu_info_firmware_header_v1_0 *hdr;

@@ -2383,12 +2383,12 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 		break;
 	}

-	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_gpu_info.bin", chip_name);
-	err = amdgpu_ucode_request(adev, &adev->firmware.gpu_info_fw, fw_name);
+	err = amdgpu_ucode_request(adev, &adev->firmware.gpu_info_fw,
+				   "amdgpu/%s_gpu_info.bin", chip_name);
 	if (err) {
 		dev_err(adev->dev,
-			"Failed to get gpu_info firmware \"%s\"\n",
-			fw_name);
+			"Failed to get gpu_info firmware \"%s_gpu_info.bin\"\n",
+			chip_name);
 		goto out;
 	}

@@ -3142,7 +3142,8 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
 		return r;
 	}

-	amdgpu_ras_set_error_query_ready(adev, true);
+	if (!amdgpu_in_reset(adev))
+		amdgpu_ras_set_error_query_ready(adev, true);

 	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE);
 	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_GATE);
@@ -4048,6 +4049,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	mutex_init(&adev->grbm_idx_mutex);
 	mutex_init(&adev->mn_lock);
 	mutex_init(&adev->virt.vf_errors.lock);
+	mutex_init(&adev->virt.rlcg_reg_lock);
 	hash_init(adev->mn_hash);
 	mutex_init(&adev->psp.mutex);
 	mutex_init(&adev->notifier_lock);
@@ -5011,7 +5013,8 @@ static int amdgpu_device_recover_vram(struct amdgpu_device *adev)
 		shadow = vmbo->shadow;

 		/* No need to recover an evicted BO */
-		if (shadow->tbo.resource->mem_type != TTM_PL_TT ||
+		if (!shadow->tbo.resource ||
+		    shadow->tbo.resource->mem_type != TTM_PL_TT ||
 		    shadow->tbo.resource->start == AMDGPU_BO_INVALID_OFFSET ||
 		    shadow->parent->tbo.resource->mem_type != TTM_PL_VRAM)
 			continue;
@@ -5055,29 +5058,29 @@ static int amdgpu_device_recover_vram(struct amdgpu_device *adev)
 * amdgpu_device_reset_sriov - reset ASIC for SR-IOV vf
 *
 * @adev: amdgpu_device pointer
- * @from_hypervisor: request from hypervisor
+ * @reset_context: amdgpu reset context pointer
 *
 * do VF FLR and reinitialize Asic
 * return 0 means succeeded otherwise failed
 */
 static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
-				     bool from_hypervisor)
+				     struct amdgpu_reset_context *reset_context)
 {
 	int r;
 	struct amdgpu_hive_info *hive = NULL;
-	int retry_limit = 0;

-retry:
-	amdgpu_amdkfd_pre_reset(adev);
-
-	amdgpu_device_stop_pending_resets(adev);
-
-	if (from_hypervisor)
+	if (test_bit(AMDGPU_HOST_FLR, &reset_context->flags)) {
+		if (!amdgpu_ras_get_fed_status(adev))
+			amdgpu_virt_ready_to_reset(adev);
+		amdgpu_virt_wait_reset(adev);
+		clear_bit(AMDGPU_HOST_FLR, &reset_context->flags);
 		r = amdgpu_virt_request_full_gpu(adev, true);
-	else
+	} else {
 		r = amdgpu_virt_reset_gpu(adev);
+	}
 	if (r)
 		return r;
+
 	amdgpu_ras_set_fed(adev, false);
 	amdgpu_irq_gpu_reset_resume_helper(adev);

@@ -5087,7 +5090,7 @@ retry:
 	/* Resume IP prior to SMC */
 	r = amdgpu_device_ip_reinit_early_sriov(adev);
 	if (r)
-		goto error;
+		return r;

 	amdgpu_virt_init_data_exchange(adev);

@@ -5098,38 +5101,41 @@ retry:
 	/* now we are okay to resume SMC/CP/SDMA */
 	r = amdgpu_device_ip_reinit_late_sriov(adev);
 	if (r)
-		goto error;
+		return r;

 	hive = amdgpu_get_xgmi_hive(adev);
 	/* Update PSP FW topology after reset */
 	if (hive && adev->gmc.xgmi.num_physical_nodes > 1)
 		r = amdgpu_xgmi_update_topology(hive, adev);
-
 	if (hive)
 		amdgpu_put_xgmi_hive(hive);
+	if (r)
+		return r;

-	if (!r) {
-		r = amdgpu_ib_ring_tests(adev);
+	r = amdgpu_ib_ring_tests(adev);
+	if (r)
+		return r;

-		amdgpu_amdkfd_post_reset(adev);
-	}
-
-error:
-	if (!r && adev->virt.gim_feature & AMDGIM_FEATURE_GIM_FLR_VRAMLOST) {
+	if (adev->virt.gim_feature & AMDGIM_FEATURE_GIM_FLR_VRAMLOST) {
 		amdgpu_inc_vram_lost(adev);
 		r = amdgpu_device_recover_vram(adev);
 	}
+	if (r)
+		return r;
+
+	/* need to be called during full access so we can't do it later like
+	 * bare-metal does.
+	 */
+	amdgpu_amdkfd_post_reset(adev);
 	amdgpu_virt_release_full_gpu(adev, true);

-	if (AMDGPU_RETRY_SRIOV_RESET(r)) {
-		if (retry_limit < AMDGPU_MAX_RETRY_LIMIT) {
-			retry_limit++;
-			goto retry;
-		} else
-			DRM_ERROR("GPU reset retry is beyond the retry limit\n");
-	}
-
-	return r;
+	/* Aldebaran and gfx_11_0_3 support ras in SRIOV, so need resume ras during reset */
+	if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 2) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3))
+		amdgpu_ras_resume(adev);
+	return 0;
 }

 /**
@@ -5220,11 +5226,14 @@ int amdgpu_device_mode1_reset(struct amdgpu_device *adev)

 	dev_info(adev->dev, "GPU mode1 reset\n");

+	/* Cache the state before bus master disable. The saved config space
+	 * values are used in other cases like restore after mode-2 reset.
+	 */
+	amdgpu_device_cache_pci_state(adev->pdev);
+
 	/* disable BM */
 	pci_clear_master(adev->pdev);

-	amdgpu_device_cache_pci_state(adev->pdev);
-
 	if (amdgpu_dpm_is_mode1_reset_supported(adev)) {
 		dev_info(adev->dev, "GPU smu mode1 reset\n");
 		ret = amdgpu_dpm_mode1_reset(adev);
@@ -5371,11 +5380,13 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle,
 	if (!test_bit(AMDGPU_SKIP_COREDUMP, &reset_context->flags)) {
 		amdgpu_reset_reg_dumps(tmp_adev);

+		dev_info(tmp_adev->dev, "Dumping IP State\n");
 		/* Trigger ip dump before we reset the asic */
 		for (i = 0; i < tmp_adev->num_ip_blocks; i++)
 			if (tmp_adev->ip_blocks[i].version->funcs->dump_ip_state)
 				tmp_adev->ip_blocks[i].version->funcs
 				->dump_ip_state((void *)tmp_adev);
+		dev_info(tmp_adev->dev, "Dumping IP State Completed\n");
 	}

 	reset_context->reset_device_list = device_list_handle;
@@ -5688,6 +5699,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 	int i, r = 0;
 	bool need_emergency_restart = false;
 	bool audio_suspended = false;
+	int retry_limit = AMDGPU_MAX_RETRY_LIMIT;

 	/*
 	 * Special case: RAS triggered and full reset isn't supported
@@ -5722,7 +5734,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 	 * to put adev in the 1st position.
 	 */
 	INIT_LIST_HEAD(&device_list);
-	if (!amdgpu_sriov_vf(adev) && (adev->gmc.xgmi.num_physical_nodes > 1)) {
+	if (!amdgpu_sriov_vf(adev) && (adev->gmc.xgmi.num_physical_nodes > 1) && hive) {
 		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
 			list_add_tail(&tmp_adev->reset_list, &device_list);
 			if (adev->shutdown)
@@ -5769,8 +5781,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,

 		cancel_delayed_work_sync(&tmp_adev->delayed_init_work);

-		if (!amdgpu_sriov_vf(tmp_adev))
-			amdgpu_amdkfd_pre_reset(tmp_adev);
+		amdgpu_amdkfd_pre_reset(tmp_adev, reset_context);

 		/*
 		 * Mark these ASICs to be reseted as untracked first
@@ -5823,34 +5834,40 @@ retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 				  r, adev_to_drm(tmp_adev)->unique);
 			tmp_adev->asic_reset_res = r;
 		}
-
-		if (!amdgpu_sriov_vf(tmp_adev))
-			/*
-			* Drop all pending non scheduler resets. Scheduler resets
-			* were already dropped during drm_sched_stop
-			*/
-			amdgpu_device_stop_pending_resets(tmp_adev);
 	}

 	/* Actual ASIC resets if needed.*/
 	/* Host driver will handle XGMI hive reset for SRIOV */
 	if (amdgpu_sriov_vf(adev)) {
-		r = amdgpu_device_reset_sriov(adev, job ? false : true);
+		if (amdgpu_ras_get_fed_status(adev) || amdgpu_virt_rcvd_ras_interrupt(adev)) {
+			dev_dbg(adev->dev, "Detected RAS error, wait for FLR completion\n");
+			amdgpu_ras_set_fed(adev, true);
+			set_bit(AMDGPU_HOST_FLR, &reset_context->flags);
+		}
+
+		r = amdgpu_device_reset_sriov(adev, reset_context);
+		if (AMDGPU_RETRY_SRIOV_RESET(r) && (retry_limit--) > 0) {
+			amdgpu_virt_release_full_gpu(adev, true);
+			goto retry;
+		}
 		if (r)
 			adev->asic_reset_res = r;
-
-		/* Aldebaran and gfx_11_0_3 support ras in SRIOV, so need resume ras during reset */
-		if (amdgpu_ip_version(adev, GC_HWIP, 0) ==
-			    IP_VERSION(9, 4, 2) ||
-		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
-		    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(11, 0, 3))
-			amdgpu_ras_resume(adev);
 	} else {
 		r = amdgpu_do_asic_reset(device_list_handle, reset_context);
 		if (r && r == -EAGAIN)
 			goto retry;
 	}

+	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+		/*
+		 * Drop any pending non scheduler resets queued before reset is done.
+		 * Any reset scheduled after this point would be valid. Scheduler resets
+		 * were already dropped during drm_sched_stop and no new ones can come
+		 * in before drm_sched_start.
+		 */
+		amdgpu_device_stop_pending_resets(tmp_adev);
+	}
+
 skip_hw_reset:

 	/* Post ASIC reset for all devs .*/
@@ -5944,13 +5961,18 @@ static void amdgpu_device_partner_bandwidth(struct amdgpu_device *adev,
 	*speed = PCI_SPEED_UNKNOWN;
 	*width = PCIE_LNK_WIDTH_UNKNOWN;

-	while ((parent = pci_upstream_bridge(parent))) {
-		/* skip upstream/downstream switches internal to dGPU*/
-		if (parent->vendor == PCI_VENDOR_ID_ATI)
-			continue;
-		*speed = pcie_get_speed_cap(parent);
-		*width = pcie_get_width_cap(parent);
-		break;
+	if (amdgpu_device_pcie_dynamic_switching_supported(adev)) {
+		while ((parent = pci_upstream_bridge(parent))) {
+			/* skip upstream/downstream switches internal to dGPU*/
+			if (parent->vendor == PCI_VENDOR_ID_ATI)
+				continue;
+			*speed = pcie_get_speed_cap(parent);
+			*width = pcie_get_width_cap(parent);
+			break;
+		}
+	} else {
+		/* use the current speeds rather than max if switching is not supported */
+		pcie_bandwidth_available(adev->pdev, NULL, speed, width);
 	}
 }

@@ -6165,7 +6187,7 @@ int amdgpu_device_baco_exit(struct drm_device *dev)
 	    adev->nbio.funcs->enable_doorbell_interrupt)
 		adev->nbio.funcs->enable_doorbell_interrupt(adev, true);

-	if (amdgpu_passthrough(adev) &&
+	if (amdgpu_passthrough(adev) && adev->nbio.funcs &&
 	    adev->nbio.funcs->clear_doorbell_interrupt)
 		adev->nbio.funcs->clear_doorbell_interrupt(adev);

@@ -6265,19 +6287,11 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
 	struct amdgpu_reset_context reset_context;
 	u32 memsize;
 	struct list_head device_list;
-	struct amdgpu_hive_info *hive;
-	int hive_ras_recovery = 0;
-	struct amdgpu_ras *ras;

 	/* PCI error slot reset should be skipped During RAS recovery */
-	hive = amdgpu_get_xgmi_hive(adev);
-	if (hive) {
-		hive_ras_recovery = atomic_read(&hive->ras_recovery);
-		amdgpu_put_xgmi_hive(hive);
-	}
-	ras = amdgpu_ras_get_context(adev);
-	if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3)) &&
-		 ras && (atomic_read(&ras->in_recovery) || hive_ras_recovery))
+	if ((amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 3) ||
+	    amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4)) &&
+	    amdgpu_ras_in_recovery(adev))
 		return PCI_ERS_RESULT_RECOVERED;

 	DRM_INFO("PCI error: slot reset callback!!\n");
@@ -6519,6 +6533,22 @@ void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev,
 	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
 }

+/**
+ * amdgpu_device_get_gang - return a reference to the current gang
+ * @adev: amdgpu_device pointer
+ *
+ * Returns: A new reference to the current gang leader.
+ */
+struct dma_fence *amdgpu_device_get_gang(struct amdgpu_device *adev)
+{
+	struct dma_fence *fence;
+
+	rcu_read_lock();
+	fence = dma_fence_get_rcu_safe(&adev->gang_submit);
+	rcu_read_unlock();
+	return fence;
+}
+
 /**
 * amdgpu_device_switch_gang - switch to a new gang
 * @adev: amdgpu_device pointer
@@ -6535,10 +6565,7 @@ struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,

 	do {
 		dma_fence_put(old);
-		rcu_read_lock();
-		old = dma_fence_get_rcu_safe(&adev->gang_submit);
-		rcu_read_unlock();
-
+		old = amdgpu_device_get_gang(adev);
 		if (old == gang)
 			break;