From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id B1B7A4598C; Sat, 14 Sep 2024 10:21:06 +0200 (CEST) Received: from mails.dpdk.org (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id 74205402B1; Sat, 14 Sep 2024 10:21:05 +0200 (CEST) Received: from szxga02-in.huawei.com (szxga02-in.huawei.com [45.249.212.188]) by mails.dpdk.org (Postfix) with ESMTP id D62D34026A for ; Sat, 14 Sep 2024 10:21:02 +0200 (CEST) Received: from mail.maildlp.com (unknown [172.19.162.254]) by szxga02-in.huawei.com (SkyGuard) with ESMTP id 4X5PG434P8zQj7f; Sat, 14 Sep 2024 16:18:52 +0800 (CST) Received: from dggpeml500024.china.huawei.com (unknown [7.185.36.10]) by mail.maildlp.com (Postfix) with ESMTPS id 5281B18010A; Sat, 14 Sep 2024 16:20:57 +0800 (CST) Received: from [10.67.121.161] (10.67.121.161) by dggpeml500024.china.huawei.com (7.185.36.10) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.1.2507.39; Sat, 14 Sep 2024 16:20:57 +0800 Message-ID: Date: Sat, 14 Sep 2024 16:20:56 +0800 MIME-Version: 1.0 User-Agent: Mozilla Thunderbird Subject: Re: [PATCH v7 2/8] ethdev: add telemetry cmd for registers To: Jie Hai , , , , CC: References: <20231214015650.3738578-1-haijie1@huawei.com> <20240914071320.1982-1-haijie1@huawei.com> <20240914071320.1982-3-haijie1@huawei.com> Content-Language: en-US From: fengchengwen In-Reply-To: <20240914071320.1982-3-haijie1@huawei.com> Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit X-Originating-IP: [10.67.121.161] X-ClientProxiedBy: dggems701-chm.china.huawei.com (10.3.19.178) To dggpeml500024.china.huawei.com (7.185.36.10) X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org On 2024/9/14 15:13, Jie Hai wrote: > This patch adds a telemetry command for registers dump, > and supports obtaining the registers of a specified module. > > In one way, the number of registers that can be exported > is limited by the number of elements carried by dict and > container. In another way, the length of the string > exported by telemetry is limited by MAX_OUTPUT_LEN. > Therefore, when the number of registers to be exported > exceeds, some information will be lost. Warn on the former > case. > > An example usage is shown below: > --> /ethdev/regs,0,ring > { > "/ethdev/regs": { > "registers_length": 318, > "registers_width": 4, > "register_offset": "0x0", > "version": "0x1140011", > "group_0": { > "Q0_ring_rx_bd_num": "0x0", > "Q0_ring_rx_bd_len": "0x0", > ... > }, > "group_1": { > ... > }, > ... > } > > Signed-off-by: Jie Hai > --- > lib/ethdev/rte_ethdev_telemetry.c | 128 ++++++++++++++++++++++++++++++ > 1 file changed, 128 insertions(+) > > diff --git a/lib/ethdev/rte_ethdev_telemetry.c b/lib/ethdev/rte_ethdev_telemetry.c > index 6b873e7abe68..1d59c693883e 100644 > --- a/lib/ethdev/rte_ethdev_telemetry.c > +++ b/lib/ethdev/rte_ethdev_telemetry.c > @@ -1395,6 +1395,132 @@ eth_dev_handle_port_tm_node_caps(const char *cmd __rte_unused, > return ret; > } > > +static void > +eth_dev_add_reg_data(struct rte_tel_data *d, struct rte_dev_reg_info *reg_info, > + uint32_t idx) > +{ > + if (reg_info->width == sizeof(uint32_t)) > + rte_tel_data_add_dict_uint_hex(d, reg_info->names[idx].name, > + *((uint32_t *)reg_info->data + idx), 0); > + else > + rte_tel_data_add_dict_uint_hex(d, reg_info->names[idx].name, > + *((uint64_t *)reg_info->data + idx), 0); > +} > + > +static int > +eth_dev_store_regs(struct rte_tel_data *d, struct rte_dev_reg_info *reg_info) > +{ > + struct rte_tel_data *groups[RTE_TEL_MAX_DICT_ENTRIES]; > + char group_name[RTE_TEL_MAX_STRING_LEN] = {0}; > + struct rte_tel_data *group = NULL; > + uint32_t grp_num = 0; > + uint32_t i; > + int ret; > + > + rte_tel_data_start_dict(d); > + rte_tel_data_add_dict_uint(d, "register_length", reg_info->length); > + rte_tel_data_add_dict_uint(d, "register_width", reg_info->width); > + rte_tel_data_add_dict_uint_hex(d, "register_offset", reg_info->offset, 0); > + rte_tel_data_add_dict_uint_hex(d, "version", reg_info->version, 0); > + > + for (i = 0; i < reg_info->length; i++) { > + if (i % RTE_TEL_MAX_DICT_ENTRIES != 0) { > + eth_dev_add_reg_data(group, reg_info, i); > + continue; > + } > + > + group = rte_tel_data_alloc(); > + if (group == NULL) { > + ret = -ENOMEM; > + RTE_ETHDEV_LOG_LINE(WARNING, "No enough memory for group data"); > + goto out; > + } > + groups[grp_num++] = group; Although the probability is very small, the groups array may overflow if the length is very large. Suggest add judgement at the begin of this function: uint32_t max_cap = (RTE_TEL_MAX_DICT_ENTRIES - 4) * RTE_TEL_MAX_DICT_ENTRIES; // it will be 65000+ register if (reg_info->length > max_cap) { // just output error and exit } > + rte_tel_data_start_dict(group); > + eth_dev_add_reg_data(group, reg_info, i); > + } > + > + for (i = 0; i < grp_num; i++) { > + snprintf(group_name, RTE_TEL_MAX_STRING_LEN, "group_%u", i); > + ret = rte_tel_data_add_dict_container(d, group_name, groups[i], 0); > + if (ret == -ENOSPC) { > + RTE_ETHDEV_LOG_LINE(WARNING, > + "Reduce register number to be displayed from %u to %u due to limited capacity of telemetry", > + reg_info->length, i * RTE_TEL_MAX_DICT_ENTRIES); > + break; > + } It may lead to memory leak when break. If we add such judgement before, there is no need to add such judgement here. > + } > + return 0; > +out: > + for (i = 0; i < grp_num; i++) > + rte_tel_data_free(groups[i]); > + > + return ret; > +} > + > +static int > +eth_dev_get_port_regs(int port_id, struct rte_tel_data *d, char *filter) > +{ > + struct rte_dev_reg_info reg_info; > + int ret; > + > + memset(®_info, 0, sizeof(reg_info)); > + reg_info.filter = filter; > + > + ret = rte_eth_dev_get_reg_info_ext(port_id, ®_info); > + if (ret != 0) { > + RTE_ETHDEV_LOG_LINE(ERR, "Error getting device reg info: %d", ret); how about "Failed to get device reg info, ret = %d" > + return ret; > + } > + > + reg_info.data = calloc(reg_info.length, reg_info.width); > + if (reg_info.data == NULL) { > + RTE_ETHDEV_LOG_LINE(ERR, "Fail to allocate memory for reg_info.data"); > + return -ENOMEM; > + } > + > + reg_info.names = calloc(reg_info.length, sizeof(struct rte_eth_reg_name)); > + if (reg_info.names == NULL) { > + RTE_ETHDEV_LOG_LINE(ERR, "Fail to allocate memory for reg_info.names"); > + free(reg_info.data); > + return -ENOMEM; > + } > + > + ret = rte_eth_dev_get_reg_info_ext(port_id, ®_info); > + if (ret != 0) { > + RTE_ETHDEV_LOG_LINE(ERR, "Error getting regs from device: %d", ret); It seemed the ret is device id, how about "Failed to get device reg info, ret = %d" > + ret = -EINVAL; > + goto out; > + } > + > + ret = eth_dev_store_regs(d, ®_info); > +out: > + free(reg_info.data); > + free(reg_info.names); > + > + return ret; > +} > + > +static int > +eth_dev_handle_port_regs(const char *cmd __rte_unused, > + const char *params, > + struct rte_tel_data *d) > +{ > + char *filter, *end_param; > + uint16_t port_id; > + int ret; > + > + ret = eth_dev_parse_port_params(params, &port_id, &end_param, true); > + if (ret != 0) > + return ret; > + > + filter = strtok(end_param, ","); > + if (filter != NULL && strlen(filter) == 0) > + filter = NULL; > + > + return eth_dev_get_port_regs(port_id, d, filter); > +} > + > RTE_INIT(ethdev_init_telemetry) > { > rte_telemetry_register_cmd("/ethdev/list", eth_dev_handle_port_list, > @@ -1436,4 +1562,6 @@ RTE_INIT(ethdev_init_telemetry) > "Returns TM Level Capabilities info for a port. Parameters: int port_id, int level_id (see tm_capability for the max)"); > rte_telemetry_register_cmd("/ethdev/tm_node_capability", eth_dev_handle_port_tm_node_caps, > "Returns TM Node Capabilities info for a port. Parameters: int port_id, int node_id (see tm_capability for the max)"); > + rte_telemetry_register_cmd("/ethdev/regs", eth_dev_handle_port_regs, > + "Returns all or filtered registers info for a port. Parameters: int port_id, string module_name (Optional if show all)"); > }