Merge branch 'hns3-RAS'

Guangbin Huang says:

====================
net: hns3: add RAS compatibility adaptation solution

This patchset adds RAS compatibility adaptation solution for new devices.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

drivers/net/ethernet/hisilicon/hns3/hnae3.h

@@ -91,6 +91,7 @@ enum HNAE3_DEV_CAP_BITS {
 	HNAE3_DEV_SUPPORT_STASH_B,
 	HNAE3_DEV_SUPPORT_UDP_TUNNEL_CSUM_B,
 	HNAE3_DEV_SUPPORT_PAUSE_B,
+	HNAE3_DEV_SUPPORT_RAS_IMP_B,
 	HNAE3_DEV_SUPPORT_RXD_ADV_LAYOUT_B,
 	HNAE3_DEV_SUPPORT_PORT_VLAN_BYPASS_B,
 	HNAE3_DEV_SUPPORT_VLAN_FLTR_MDF_B,
@@ -129,6 +130,9 @@ enum HNAE3_DEV_CAP_BITS {
 #define hnae3_dev_phy_imp_supported(hdev) \
 	test_bit(HNAE3_DEV_SUPPORT_PHY_IMP_B, (hdev)->ae_dev->caps)
 
+#define hnae3_dev_ras_imp_supported(hdev) \
+	test_bit(HNAE3_DEV_SUPPORT_RAS_IMP_B, (hdev)->ae_dev->caps)
+
 #define hnae3_dev_tqp_txrx_indep_supported(hdev) \
 	test_bit(HNAE3_DEV_SUPPORT_TQP_TXRX_INDEP_B, (hdev)->ae_dev->caps)
 

drivers/net/ethernet/hisilicon/hns3/hns3_debugfs.c

@@ -349,6 +349,9 @@ static struct hns3_dbg_cap_info hns3_dbg_cap[] = {
 	}, {
 		.name = "support imp-controlled PHY",
 		.cap_bit = HNAE3_DEV_SUPPORT_PHY_IMP_B,
+	}, {
+		.name = "support imp-controlled RAS",
+		.cap_bit = HNAE3_DEV_SUPPORT_RAS_IMP_B,
 	}, {
 		.name = "support rxd advanced layout",
 		.cap_bit = HNAE3_DEV_SUPPORT_RXD_ADV_LAYOUT_B,

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.c

@@ -178,7 +178,8 @@ static bool hclge_is_special_opcode(u16 opcode)
 			     HCLGE_QUERY_CLEAR_MPF_RAS_INT,
 			     HCLGE_QUERY_CLEAR_PF_RAS_INT,
 			     HCLGE_QUERY_CLEAR_ALL_MPF_MSIX_INT,
-			     HCLGE_QUERY_CLEAR_ALL_PF_MSIX_INT};
+			     HCLGE_QUERY_CLEAR_ALL_PF_MSIX_INT,
+			     HCLGE_QUERY_ALL_ERR_INFO};
 	int i;
 
 	for (i = 0; i < ARRAY_SIZE(spec_opcode); i++) {
@@ -386,6 +387,8 @@ static void hclge_parse_capability(struct hclge_dev *hdev,
 		set_bit(HNAE3_DEV_SUPPORT_PAUSE_B, ae_dev->caps);
 	if (hnae3_get_bit(caps, HCLGE_CAP_PHY_IMP_B))
 		set_bit(HNAE3_DEV_SUPPORT_PHY_IMP_B, ae_dev->caps);
+	if (hnae3_get_bit(caps, HCLGE_CAP_RAS_IMP_B))
+		set_bit(HNAE3_DEV_SUPPORT_RAS_IMP_B, ae_dev->caps);
 	if (hnae3_get_bit(caps, HCLGE_CAP_RXD_ADV_LAYOUT_B))
 		set_bit(HNAE3_DEV_SUPPORT_RXD_ADV_LAYOUT_B, ae_dev->caps);
 	if (hnae3_get_bit(caps, HCLGE_CAP_PORT_VLAN_BYPASS_B)) {

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h

@@ -293,6 +293,8 @@ enum hclge_opcode_type {
 	HCLGE_QUERY_MSIX_INT_STS_BD_NUM	= 0x1513,
 	HCLGE_QUERY_CLEAR_ALL_MPF_MSIX_INT	= 0x1514,
 	HCLGE_QUERY_CLEAR_ALL_PF_MSIX_INT	= 0x1515,
+	HCLGE_QUERY_ALL_ERR_BD_NUM		= 0x1516,
+	HCLGE_QUERY_ALL_ERR_INFO		= 0x1517,
 	HCLGE_CONFIG_ROCEE_RAS_INT_EN	= 0x1580,
 	HCLGE_QUERY_CLEAR_ROCEE_RAS_INT = 0x1581,
 	HCLGE_ROCEE_PF_RAS_INT_CMD	= 0x1584,
@@ -390,6 +392,7 @@ enum HCLGE_CAP_BITS {
 	HCLGE_CAP_HW_PAD_B,
 	HCLGE_CAP_STASH_B,
 	HCLGE_CAP_UDP_TUNNEL_CSUM_B,
+	HCLGE_CAP_RAS_IMP_B = 12,
 	HCLGE_CAP_FEC_B = 13,
 	HCLGE_CAP_PAUSE_B = 14,
 	HCLGE_CAP_RXD_ADV_LAYOUT_B = 15,

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_err.c

@@ -631,6 +631,134 @@ static const struct hclge_hw_error hclge_rocee_qmm_ovf_err_int[] = {
 	{ /* sentinel */ }
 };
 
+static const struct hclge_hw_module_id hclge_hw_module_id_st[] = {
+	{
+		.module_id = MODULE_NONE,
+		.msg = "MODULE_NONE"
+	}, {
+		.module_id = MODULE_BIOS_COMMON,
+		.msg = "MODULE_BIOS_COMMON"
+	}, {
+		.module_id = MODULE_GE,
+		.msg = "MODULE_GE"
+	}, {
+		.module_id = MODULE_IGU_EGU,
+		.msg = "MODULE_IGU_EGU"
+	}, {
+		.module_id = MODULE_LGE,
+		.msg = "MODULE_LGE"
+	}, {
+		.module_id = MODULE_NCSI,
+		.msg = "MODULE_NCSI"
+	}, {
+		.module_id = MODULE_PPP,
+		.msg = "MODULE_PPP"
+	}, {
+		.module_id = MODULE_QCN,
+		.msg = "MODULE_QCN"
+	}, {
+		.module_id = MODULE_RCB_RX,
+		.msg = "MODULE_RCB_RX"
+	}, {
+		.module_id = MODULE_RTC,
+		.msg = "MODULE_RTC"
+	}, {
+		.module_id = MODULE_SSU,
+		.msg = "MODULE_SSU"
+	}, {
+		.module_id = MODULE_TM,
+		.msg = "MODULE_TM"
+	}, {
+		.module_id = MODULE_RCB_TX,
+		.msg = "MODULE_RCB_TX"
+	}, {
+		.module_id = MODULE_TXDMA,
+		.msg = "MODULE_TXDMA"
+	}, {
+		.module_id = MODULE_MASTER,
+		.msg = "MODULE_MASTER"
+	}, {
+		.module_id = MODULE_ROCEE_TOP,
+		.msg = "MODULE_ROCEE_TOP"
+	}, {
+		.module_id = MODULE_ROCEE_TIMER,
+		.msg = "MODULE_ROCEE_TIMER"
+	}, {
+		.module_id = MODULE_ROCEE_MDB,
+		.msg = "MODULE_ROCEE_MDB"
+	}, {
+		.module_id = MODULE_ROCEE_TSP,
+		.msg = "MODULE_ROCEE_TSP"
+	}, {
+		.module_id = MODULE_ROCEE_TRP,
+		.msg = "MODULE_ROCEE_TRP"
+	}, {
+		.module_id = MODULE_ROCEE_SCC,
+		.msg = "MODULE_ROCEE_SCC"
+	}, {
+		.module_id = MODULE_ROCEE_CAEP,
+		.msg = "MODULE_ROCEE_CAEP"
+	}, {
+		.module_id = MODULE_ROCEE_GEN_AC,
+		.msg = "MODULE_ROCEE_GEN_AC"
+	}, {
+		.module_id = MODULE_ROCEE_QMM,
+		.msg = "MODULE_ROCEE_QMM"
+	}, {
+		.module_id = MODULE_ROCEE_LSAN,
+		.msg = "MODULE_ROCEE_LSAN"
+	}
+};
+
+static const struct hclge_hw_type_id hclge_hw_type_id_st[] = {
+	{
+		.type_id = NONE_ERROR,
+		.msg = "none_error"
+	}, {
+		.type_id = FIFO_ERROR,
+		.msg = "fifo_error"
+	}, {
+		.type_id = MEMORY_ERROR,
+		.msg = "memory_error"
+	}, {
+		.type_id = POISON_ERROR,
+		.msg = "poison_error"
+	}, {
+		.type_id = MSIX_ECC_ERROR,
+		.msg = "msix_ecc_error"
+	}, {
+		.type_id = TQP_INT_ECC_ERROR,
+		.msg = "tqp_int_ecc_error"
+	}, {
+		.type_id = PF_ABNORMAL_INT_ERROR,
+		.msg = "pf_abnormal_int_error"
+	}, {
+		.type_id = MPF_ABNORMAL_INT_ERROR,
+		.msg = "mpf_abnormal_int_error"
+	}, {
+		.type_id = COMMON_ERROR,
+		.msg = "common_error"
+	}, {
+		.type_id = PORT_ERROR,
+		.msg = "port_error"
+	}, {
+		.type_id = ETS_ERROR,
+		.msg = "ets_error"
+	}, {
+		.type_id = NCSI_ERROR,
+		.msg = "ncsi_error"
+	}, {
+		.type_id = GLB_ERROR,
+		.msg = "glb_error"
+	}, {
+		.type_id = ROCEE_NORMAL_ERR,
+		.msg = "rocee_normal_error"
+	}, {
+		.type_id = ROCEE_OVF_ERR,
+		.msg = "rocee_ovf_error"
+	}
+};
+
 static void hclge_log_error(struct device *dev, char *reg,
 			    const struct hclge_hw_error *err,
 			    u32 err_sts, unsigned long *reset_requests)
@@ -1611,11 +1739,27 @@ static const struct hclge_hw_blk hw_blk[] = {
 	{ /* sentinel */ }
 };
 
+static void hclge_config_all_msix_error(struct hclge_dev *hdev, bool enable)
+{
+	u32 reg_val;
+
+	reg_val = hclge_read_dev(&hdev->hw, HCLGE_PF_OTHER_INT_REG);
+
+	if (enable)
+		reg_val |= BIT(HCLGE_VECTOR0_ALL_MSIX_ERR_B);
+	else
+		reg_val &= ~BIT(HCLGE_VECTOR0_ALL_MSIX_ERR_B);
+
+	hclge_write_dev(&hdev->hw, HCLGE_PF_OTHER_INT_REG, reg_val);
+}
+
 int hclge_config_nic_hw_error(struct hclge_dev *hdev, bool state)
 {
 	const struct hclge_hw_blk *module = hw_blk;
 	int ret = 0;
 
+	hclge_config_all_msix_error(hdev, state);
+
 	while (module->name) {
 		if (module->config_err_int) {
 			ret = module->config_err_int(hdev, state);
@@ -1876,11 +2020,8 @@ static int hclge_handle_pf_msix_error(struct hclge_dev *hdev,
 static int hclge_handle_all_hw_msix_error(struct hclge_dev *hdev,
 					  unsigned long *reset_requests)
 {
-	struct hclge_mac_tnl_stats mac_tnl_stats;
-	struct device *dev = &hdev->pdev->dev;
 	u32 mpf_bd_num, pf_bd_num, bd_num;
 	struct hclge_desc *desc;
-	u32 status;
 	int ret;
 
 	/* query the number of bds for the MSIx int status */
@@ -1903,29 +2044,7 @@ static int hclge_handle_all_hw_msix_error(struct hclge_dev *hdev,
 	if (ret)
 		goto msi_error;
 
-	/* query and clear mac tnl interruptions */
-	hclge_cmd_setup_basic_desc(&desc[0], HCLGE_OPC_QUERY_MAC_TNL_INT,
-				   true);
-	ret = hclge_cmd_send(&hdev->hw, &desc[0], 1);
-	if (ret) {
-		dev_err(dev, "query mac tnl int cmd failed (%d)\n", ret);
-		goto msi_error;
-	}
-
-	status = le32_to_cpu(desc->data[0]);
-	if (status) {
-		/* When mac tnl interrupt occurs, we record current time and
-		 * register status here in a fifo, then clear the status. So
-		 * that if link status changes suddenly at some time, we can
-		 * query them by debugfs.
-		 */
-		mac_tnl_stats.time = local_clock();
-		mac_tnl_stats.status = status;
-		kfifo_put(&hdev->mac_tnl_log, mac_tnl_stats);
-		ret = hclge_clear_mac_tnl_int(hdev);
-		if (ret)
-			dev_err(dev, "clear mac tnl int failed (%d)\n", ret);
-	}
+	ret = hclge_handle_mac_tnl(hdev);
 
 msi_error:
 	kfree(desc);
@@ -1947,10 +2066,43 @@ int hclge_handle_hw_msix_error(struct hclge_dev *hdev,
 	return hclge_handle_all_hw_msix_error(hdev, reset_requests);
 }
 
-void hclge_handle_all_hns_hw_errors(struct hnae3_ae_dev *ae_dev)
+int hclge_handle_mac_tnl(struct hclge_dev *hdev)
 {
-#define HCLGE_DESC_NO_DATA_LEN 8
+	struct hclge_mac_tnl_stats mac_tnl_stats;
+	struct device *dev = &hdev->pdev->dev;
+	struct hclge_desc desc;
+	u32 status;
+	int ret;
+
+	/* query and clear mac tnl interruptions */
+	hclge_cmd_setup_basic_desc(&desc, HCLGE_OPC_QUERY_MAC_TNL_INT, true);
+	ret = hclge_cmd_send(&hdev->hw, &desc, 1);
+	if (ret) {
+		dev_err(dev, "failed to query mac tnl int, ret = %d.\n", ret);
+		return ret;
+	}
 
+	status = le32_to_cpu(desc.data[0]);
+	if (status) {
+		/* When mac tnl interrupt occurs, we record current time and
+		 * register status here in a fifo, then clear the status. So
+		 * that if link status changes suddenly at some time, we can
+		 * query them by debugfs.
+		 */
+		mac_tnl_stats.time = local_clock();
+		mac_tnl_stats.status = status;
+		kfifo_put(&hdev->mac_tnl_log, mac_tnl_stats);
+		ret = hclge_clear_mac_tnl_int(hdev);
+		if (ret)
+			dev_err(dev, "failed to clear mac tnl int, ret = %d.\n",
+				ret);
+	}
+
+	return ret;
+}
+
+void hclge_handle_all_hns_hw_errors(struct hnae3_ae_dev *ae_dev)
+{
 	struct hclge_dev *hdev = ae_dev->priv;
 	struct device *dev = &hdev->pdev->dev;
 	u32 mpf_bd_num, pf_bd_num, bd_num;
@@ -1999,3 +2151,205 @@ void hclge_handle_all_hns_hw_errors(struct hnae3_ae_dev *ae_dev)
 msi_error:
 	kfree(desc);
 }
+
+bool hclge_find_error_source(struct hclge_dev *hdev)
+{
+	u32 msix_src_flag, hw_err_src_flag;
+
+	msix_src_flag = hclge_read_dev(&hdev->hw, HCLGE_MISC_VECTOR_INT_STS) &
+			HCLGE_VECTOR0_REG_MSIX_MASK;
+
+	hw_err_src_flag = hclge_read_dev(&hdev->hw,
+					 HCLGE_RAS_PF_OTHER_INT_STS_REG) &
+			  HCLGE_RAS_REG_ERR_MASK;
+
+	return msix_src_flag || hw_err_src_flag;
+}
+
+void hclge_handle_occurred_error(struct hclge_dev *hdev)
+{
+	struct hnae3_ae_dev *ae_dev = pci_get_drvdata(hdev->pdev);
+
+	if (hclge_find_error_source(hdev))
+		hclge_handle_error_info_log(ae_dev);
+}
+
+static void
+hclge_handle_error_type_reg_log(struct device *dev,
+				struct hclge_mod_err_info *mod_info,
+				struct hclge_type_reg_err_info *type_reg_info)
+{
+#define HCLGE_ERR_TYPE_MASK 0x7F
+#define HCLGE_ERR_TYPE_IS_RAS_OFFSET 7
+
+	u8 mod_id, total_module, type_id, total_type, i, is_ras;
+	u8 index_module = MODULE_NONE;
+	u8 index_type = NONE_ERROR;
+
+	mod_id = mod_info->mod_id;
+	type_id = type_reg_info->type_id & HCLGE_ERR_TYPE_MASK;
+	is_ras = type_reg_info->type_id >> HCLGE_ERR_TYPE_IS_RAS_OFFSET;
+
+	total_module = ARRAY_SIZE(hclge_hw_module_id_st);
+	total_type = ARRAY_SIZE(hclge_hw_type_id_st);
+
+	for (i = 0; i < total_module; i++) {
+		if (mod_id == hclge_hw_module_id_st[i].module_id) {
+			index_module = i;
+			break;
+		}
+	}
+
+	for (i = 0; i < total_type; i++) {
+		if (type_id == hclge_hw_type_id_st[i].type_id) {
+			index_type = i;
+			break;
+		}
+	}
+
+	if (index_module != MODULE_NONE && index_type != NONE_ERROR)
+		dev_err(dev,
+			"found %s %s, is %s error.\n",
+			hclge_hw_module_id_st[index_module].msg,
+			hclge_hw_type_id_st[index_type].msg,
+			is_ras ? "ras" : "msix");
+	else
+		dev_err(dev,
+			"unknown module[%u] or type[%u].\n", mod_id, type_id);
+
+	dev_err(dev, "reg_value:\n");
+	for (i = 0; i < type_reg_info->reg_num; i++)
+		dev_err(dev, "0x%08x\n", type_reg_info->hclge_reg[i]);
+}
+
+static void hclge_handle_error_module_log(struct hnae3_ae_dev *ae_dev,
+					  const u32 *buf, u32 buf_size)
+{
+	struct hclge_type_reg_err_info *type_reg_info;
+	struct hclge_dev *hdev = ae_dev->priv;
+	struct device *dev = &hdev->pdev->dev;
+	struct hclge_mod_err_info *mod_info;
+	struct hclge_sum_err_info *sum_info;
+	u8 mod_num, err_num, i;
+	u32 offset = 0;
+
+	sum_info = (struct hclge_sum_err_info *)&buf[offset++];
+	if (sum_info->reset_type &&
+	    sum_info->reset_type != HNAE3_NONE_RESET)
+		set_bit(sum_info->reset_type, &ae_dev->hw_err_reset_req);
+	mod_num = sum_info->mod_num;
+
+	while (mod_num--) {
+		if (offset >= buf_size) {
+			dev_err(dev, "The offset(%u) exceeds buf's size(%u).\n",
+				offset, buf_size);
+			return;
+		}
+		mod_info = (struct hclge_mod_err_info *)&buf[offset++];
+		err_num = mod_info->err_num;
+
+		for (i = 0; i < err_num; i++) {
+			if (offset >= buf_size) {
+				dev_err(dev,
+					"The offset(%u) exceeds buf size(%u).\n",
+					offset, buf_size);
+				return;
+			}
+
+			type_reg_info = (struct hclge_type_reg_err_info *)
+					    &buf[offset++];
+			hclge_handle_error_type_reg_log(dev, mod_info,
+							type_reg_info);
+
+			offset += type_reg_info->reg_num;
+		}
+	}
+}
+
+static int hclge_query_all_err_bd_num(struct hclge_dev *hdev, u32 *bd_num)
+{
+	struct device *dev = &hdev->pdev->dev;
+	struct hclge_desc desc_bd;
+	int ret;
+
+	hclge_cmd_setup_basic_desc(&desc_bd, HCLGE_QUERY_ALL_ERR_BD_NUM, true);
+	ret = hclge_cmd_send(&hdev->hw, &desc_bd, 1);
+	if (ret) {
+		dev_err(dev, "failed to query error bd_num, ret = %d.\n", ret);
+		return ret;
+	}
+
+	*bd_num = le32_to_cpu(desc_bd.data[0]);
+	if (!(*bd_num)) {
+		dev_err(dev, "The value of bd_num is 0!\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int hclge_query_all_err_info(struct hclge_dev *hdev,
+				    struct hclge_desc *desc, u32 bd_num)
+{
+	struct device *dev = &hdev->pdev->dev;
+	int ret;
+
+	hclge_cmd_setup_basic_desc(desc, HCLGE_QUERY_ALL_ERR_INFO, true);
+	ret = hclge_cmd_send(&hdev->hw, desc, bd_num);
+	if (ret)
+		dev_err(dev, "failed to query error info, ret = %d.\n", ret);
+
+	return ret;
+}
+
+int hclge_handle_error_info_log(struct hnae3_ae_dev *ae_dev)
+{
+	u32 bd_num, desc_len, buf_len, buf_size, i;
+	struct hclge_dev *hdev = ae_dev->priv;
+	struct hclge_desc *desc;
+	__le32 *desc_data;
+	u32 *buf;
+	int ret;
+
+	ret = hclge_query_all_err_bd_num(hdev, &bd_num);
+	if (ret)
+		goto out;
+
+	desc_len = bd_num * sizeof(struct hclge_desc);
+	desc = kzalloc(desc_len, GFP_KERNEL);
+	if (!desc) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ret = hclge_query_all_err_info(hdev, desc, bd_num);
+	if (ret)
+		goto err_desc;
+
+	buf_len = bd_num * sizeof(struct hclge_desc) - HCLGE_DESC_NO_DATA_LEN;
+	buf_size = buf_len / sizeof(u32);
+
+	desc_data = kzalloc(buf_len, GFP_KERNEL);
+	if (!desc_data)
+		return -ENOMEM;
+
+	buf = kzalloc(buf_len, GFP_KERNEL);
+	if (!buf) {
+		ret = -ENOMEM;
+		goto err_buf_alloc;
+	}
+
+	memcpy(desc_data, &desc[0].data[0], buf_len);
+	for (i = 0; i < buf_size; i++)
+		buf[i] = le32_to_cpu(desc_data[i]);
+
+	hclge_handle_error_module_log(ae_dev, buf, buf_size);
+	kfree(buf);
+
+err_buf_alloc:
+	kfree(desc_data);
+err_desc:
+	kfree(desc);
+out:
+	return ret;
+}

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_err.h

@@ -15,6 +15,8 @@
 #define HCLGE_RAS_PF_OTHER_INT_STS_REG   0x20B00
 #define HCLGE_RAS_REG_NFE_MASK   0xFF00
 #define HCLGE_RAS_REG_ROCEE_ERR_MASK   0x3000000
+#define HCLGE_RAS_REG_ERR_MASK \
+	(HCLGE_RAS_REG_NFE_MASK | HCLGE_RAS_REG_ROCEE_ERR_MASK)
 
 #define HCLGE_VECTOR0_REG_MSIX_MASK   0x1FF00
 
@@ -107,6 +109,10 @@
 #define HCLGE_ROCEE_OVF_ERR_INT_MASK		0x10000
 #define HCLGE_ROCEE_OVF_ERR_TYPE_MASK		0x3F
 
+#define HCLGE_DESC_DATA_MAX			8
+#define HCLGE_REG_NUM_MAX			256
+#define HCLGE_DESC_NO_DATA_LEN			8
+
 enum hclge_err_int_type {
 	HCLGE_ERR_INT_MSIX = 0,
 	HCLGE_ERR_INT_RAS_CE = 1,
@@ -114,6 +120,56 @@ enum hclge_err_int_type {
 	HCLGE_ERR_INT_RAS_FE = 3,
 };
 
+enum hclge_mod_name_list {
+	MODULE_NONE		= 0,
+	MODULE_BIOS_COMMON	= 1,
+	MODULE_GE		= 2,
+	MODULE_IGU_EGU		= 3,
+	MODULE_LGE		= 4,
+	MODULE_NCSI		= 5,
+	MODULE_PPP		= 6,
+	MODULE_QCN		= 7,
+	MODULE_RCB_RX		= 8,
+	MODULE_RTC		= 9,
+	MODULE_SSU		= 10,
+	MODULE_TM		= 11,
+	MODULE_RCB_TX		= 12,
+	MODULE_TXDMA		= 13,
+	MODULE_MASTER		= 14,
+	/* add new MODULE NAME for NIC here in order */
+	MODULE_ROCEE_TOP	= 40,
+	MODULE_ROCEE_TIMER	= 41,
+	MODULE_ROCEE_MDB	= 42,
+	MODULE_ROCEE_TSP	= 43,
+	MODULE_ROCEE_TRP	= 44,
+	MODULE_ROCEE_SCC	= 45,
+	MODULE_ROCEE_CAEP	= 46,
+	MODULE_ROCEE_GEN_AC	= 47,
+	MODULE_ROCEE_QMM	= 48,
+	MODULE_ROCEE_LSAN	= 49,
+	/* add new MODULE NAME for RoCEE here in order */
+};
+
+enum hclge_err_type_list {
+	NONE_ERROR		= 0,
+	FIFO_ERROR		= 1,
+	MEMORY_ERROR		= 2,
+	POISON_ERROR		= 3,
+	MSIX_ECC_ERROR		= 4,
+	TQP_INT_ECC_ERROR	= 5,
+	PF_ABNORMAL_INT_ERROR	= 6,
+	MPF_ABNORMAL_INT_ERROR	= 7,
+	COMMON_ERROR		= 8,
+	PORT_ERROR		= 9,
+	ETS_ERROR		= 10,
+	NCSI_ERROR		= 11,
+	GLB_ERROR		= 12,
+	/* add new ERROR TYPE for NIC here in order */
+	ROCEE_NORMAL_ERR	= 40,
+	ROCEE_OVF_ERR		= 41,
+	/* add new ERROR TYPE for ROCEE here in order */
+};
+
 struct hclge_hw_blk {
 	u32 msk;
 	const char *name;
@@ -126,11 +182,44 @@ struct hclge_hw_error {
 	enum hnae3_reset_type reset_level;
 };
 
+struct hclge_hw_module_id {
+	enum hclge_mod_name_list module_id;
+	const char *msg;
+};
+
+struct hclge_hw_type_id {
+	enum hclge_err_type_list type_id;
+	const char *msg;
+};
+
+struct hclge_sum_err_info {
+	u8 reset_type;
+	u8 mod_num;
+	u8 rsv[2];
+};
+
+struct hclge_mod_err_info {
+	u8 mod_id;
+	u8 err_num;
+	u8 rsv[2];
+};
+
+struct hclge_type_reg_err_info {
+	u8 type_id;
+	u8 reg_num;
+	u8 rsv[2];
+	u32 hclge_reg[HCLGE_REG_NUM_MAX];
+};
+
 int hclge_config_mac_tnl_int(struct hclge_dev *hdev, bool en);
 int hclge_config_nic_hw_error(struct hclge_dev *hdev, bool state);
 int hclge_config_rocee_ras_interrupt(struct hclge_dev *hdev, bool en);
 void hclge_handle_all_hns_hw_errors(struct hnae3_ae_dev *ae_dev);
+bool hclge_find_error_source(struct hclge_dev *hdev);
+void hclge_handle_occurred_error(struct hclge_dev *hdev);
 pci_ers_result_t hclge_handle_hw_ras_error(struct hnae3_ae_dev *ae_dev);
 int hclge_handle_hw_msix_error(struct hclge_dev *hdev,
 			       unsigned long *reset_requests);
+int hclge_handle_error_info_log(struct hnae3_ae_dev *ae_dev);
+int hclge_handle_mac_tnl(struct hclge_dev *hdev);
 #endif

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c

@@ -3307,11 +3307,13 @@ static int hclge_set_vf_link_state(struct hnae3_handle *handle, int vf,
 
 static u32 hclge_check_event_cause(struct hclge_dev *hdev, u32 *clearval)
 {
-	u32 cmdq_src_reg, msix_src_reg;
+	u32 cmdq_src_reg, msix_src_reg, hw_err_src_reg;
 
 	/* fetch the events from their corresponding regs */
 	cmdq_src_reg = hclge_read_dev(&hdev->hw, HCLGE_VECTOR0_CMDQ_SRC_REG);
 	msix_src_reg = hclge_read_dev(&hdev->hw, HCLGE_MISC_VECTOR_INT_STS);
+	hw_err_src_reg = hclge_read_dev(&hdev->hw,
+					HCLGE_RAS_PF_OTHER_INT_STS_REG);
 
 	/* Assumption: If by any chance reset and mailbox events are reported
 	 * together then we will only process reset event in this go and will
@@ -3339,11 +3341,10 @@ static u32 hclge_check_event_cause(struct hclge_dev *hdev, u32 *clearval)
 		return HCLGE_VECTOR0_EVENT_RST;
 	}
 
-	/* check for vector0 msix event source */
-	if (msix_src_reg & HCLGE_VECTOR0_REG_MSIX_MASK) {
-		*clearval = msix_src_reg;
+	/* check for vector0 msix event and hardware error event source */
+	if (msix_src_reg & HCLGE_VECTOR0_REG_MSIX_MASK ||
+	    hw_err_src_reg & HCLGE_RAS_REG_ERR_MASK)
 		return HCLGE_VECTOR0_EVENT_ERR;
-	}
 
 	/* check for vector0 mailbox(=CMDQ RX) event source */
 	if (BIT(HCLGE_VECTOR0_RX_CMDQ_INT_B) & cmdq_src_reg) {
@@ -3354,9 +3355,8 @@ static u32 hclge_check_event_cause(struct hclge_dev *hdev, u32 *clearval)
 
 	/* print other vector0 event source */
 	dev_info(&hdev->pdev->dev,
-		 "CMDQ INT status:0x%x, other INT status:0x%x\n",
-		 cmdq_src_reg, msix_src_reg);
-	*clearval = msix_src_reg;
+		 "INT status: CMDQ(%#x) HW errors(%#x) other(%#x)\n",
+		 cmdq_src_reg, hw_err_src_reg, msix_src_reg);
 
 	return HCLGE_VECTOR0_EVENT_OTHER;
 }
@@ -3427,15 +3427,10 @@ static irqreturn_t hclge_misc_irq_handle(int irq, void *data)
 
 	hclge_clear_event_cause(hdev, event_cause, clearval);
 
-	/* Enable interrupt if it is not cause by reset. And when
-	 * clearval equal to 0, it means interrupt status may be
-	 * cleared by hardware before driver reads status register.
-	 * For this case, vector0 interrupt also should be enabled.
-	 */
-	if (!clearval ||
-	    event_cause == HCLGE_VECTOR0_EVENT_MBX) {
+	/* Enable interrupt if it is not caused by reset event or error event */
+	if (event_cause == HCLGE_VECTOR0_EVENT_MBX ||
+	    event_cause == HCLGE_VECTOR0_EVENT_OTHER)
 		hclge_enable_vector(&hdev->misc_vector, true);
-	}
 
 	return IRQ_HANDLED;
 }
@@ -4240,6 +4235,38 @@ static void hclge_reset_subtask(struct hclge_dev *hdev)
 	hdev->reset_type = HNAE3_NONE_RESET;
 }
 
+static void hclge_handle_err_reset_request(struct hclge_dev *hdev)
+{
+	struct hnae3_ae_dev *ae_dev = pci_get_drvdata(hdev->pdev);
+	enum hnae3_reset_type reset_type;
+
+	if (ae_dev->hw_err_reset_req) {
+		reset_type = hclge_get_reset_level(ae_dev,
+						   &ae_dev->hw_err_reset_req);
+		hclge_set_def_reset_request(ae_dev, reset_type);
+	}
+
+	if (hdev->default_reset_request && ae_dev->ops->reset_event)
+		ae_dev->ops->reset_event(hdev->pdev, NULL);
+
+	/* enable interrupt after error handling complete */
+	hclge_enable_vector(&hdev->misc_vector, true);
+}
+
+static void hclge_handle_err_recovery(struct hclge_dev *hdev)
+{
+	struct hnae3_ae_dev *ae_dev = pci_get_drvdata(hdev->pdev);
+
+	ae_dev->hw_err_reset_req = 0;
+
+	if (hclge_find_error_source(hdev)) {
+		hclge_handle_error_info_log(ae_dev);
+		hclge_handle_mac_tnl(hdev);
+	}
+
+	hclge_handle_err_reset_request(hdev);
+}
+
 static void hclge_misc_err_recovery(struct hclge_dev *hdev)
 {
 	struct hnae3_ae_dev *ae_dev = pci_get_drvdata(hdev->pdev);
@@ -4247,19 +4274,16 @@ static void hclge_misc_err_recovery(struct hclge_dev *hdev)
 	u32 msix_sts_reg;
 
 	msix_sts_reg = hclge_read_dev(&hdev->hw, HCLGE_MISC_VECTOR_INT_STS);
-
 	if (msix_sts_reg & HCLGE_VECTOR0_REG_MSIX_MASK) {
-		if (hclge_handle_hw_msix_error(hdev,
-					       &hdev->default_reset_request))
+		if (hclge_handle_hw_msix_error
+				(hdev, &hdev->default_reset_request))
 			dev_info(dev, "received msix interrupt 0x%x\n",
 				 msix_sts_reg);
-
-		if (hdev->default_reset_request)
-			if (ae_dev->ops->reset_event)
-				ae_dev->ops->reset_event(hdev->pdev, NULL);
 	}
 
-	hclge_enable_vector(&hdev->misc_vector, true);
+	hclge_handle_hw_ras_error(ae_dev);
+
+	hclge_handle_err_reset_request(hdev);
 }
 
 static void hclge_errhand_service_task(struct hclge_dev *hdev)
@@ -4267,7 +4291,10 @@ static void hclge_errhand_service_task(struct hclge_dev *hdev)
 	if (!test_and_clear_bit(HCLGE_STATE_ERR_SERVICE_SCHED, &hdev->state))
 		return;
 
-	hclge_misc_err_recovery(hdev);
+	if (hnae3_dev_ras_imp_supported(hdev))
+		hclge_handle_err_recovery(hdev);
+	else
+		hclge_misc_err_recovery(hdev);
 }
 
 static void hclge_reset_service_task(struct hclge_dev *hdev)
@@ -11524,7 +11551,10 @@ static int hclge_init_ae_dev(struct hnae3_ae_dev *ae_dev)
 	hclge_clear_resetting_state(hdev);
 
 	/* Log and clear the hw errors those already occurred */
-	hclge_handle_all_hns_hw_errors(ae_dev);
+	if (hnae3_dev_ras_imp_supported(hdev))
+		hclge_handle_occurred_error(hdev);
+	else
+		hclge_handle_all_hns_hw_errors(ae_dev);
 
 	/* request delayed reset for the error recovery because an immediate
 	 * global reset on a PF affecting pending initialization of other PFs
@@ -11877,7 +11907,10 @@ static int hclge_reset_ae_dev(struct hnae3_ae_dev *ae_dev)
 	}
 
 	/* Log and clear the hw errors those already occurred */
-	hclge_handle_all_hns_hw_errors(ae_dev);
+	if (hnae3_dev_ras_imp_supported(hdev))
+		hclge_handle_occurred_error(hdev);
+	else
+		hclge_handle_all_hns_hw_errors(ae_dev);
 
 	/* Re-enable the hw error interrupts because
 	 * the interrupts get disabled on global reset.

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h

@@ -190,6 +190,7 @@ enum HLCGE_PORT_TYPE {
 #define HCLGE_VECTOR0_IMP_RESET_INT_B	1
 #define HCLGE_VECTOR0_IMP_CMDQ_ERR_B	4U
 #define HCLGE_VECTOR0_IMP_RD_POISON_B	5U
+#define HCLGE_VECTOR0_ALL_MSIX_ERR_B	6U
 
 #define HCLGE_MAC_DEFAULT_FRAME \
 	(ETH_HLEN + ETH_FCS_LEN + 2 * VLAN_HLEN + ETH_DATA_LEN)