RE: [PATCH RFC iproute-next 2/5] rdma: Add CM_ID resource tracking information
flat view
From: Steve Wise <hidden>
Date: 2018-02-26 15:27:05
On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:quoted
Sample output: # rdma resource 2: cxgb4_0: pd 5 cq 2 qp 2 cm_id 3 mr 7 3: mlx4_0: pd 7 cq 3 qp 3 cm_id 3 mr 7 # rdma resource show cm_id [root@stevo1 iproute2]# /root/stevo/iproute2/rdma/rdma resource showcm_idquoted
link cxgb4_0/- lqpn 0 qp-type RC state LISTEN ps TCP dev-type ---
transport-
type IWARP pid 30485 comm rping src-addr 0.0.0.0 src-port 7174 dst-addr 0.0.0.0 dst-port 0quoted
link cxgb4_0/2 lqpn 1048 qp-type RC state CONNECT ps TCP dev-type ETHtransport-type IWARP pid 30503 comm rping src-addr 172.16.2.1 src-port 7174 dst-addr 172.16.2.1 dst-port 38246quoted
link cxgb4_0/2 lqpn 1040 qp-type RC state CONNECT ps TCP dev-type ETHtransport-type IWARP pid 30498 comm rping src-addr 172.16.2.1 src-port 38246 dst-addr 172.16.2.1 dst-port 7174quoted
link mlx4_0/- lqpn 0 qp-type RC state LISTEN ps TCP dev-type ---
transport-
type IB pid 30485 comm rping src-addr 0.0.0.0 src-port 7174 dst-addr
0.0.0.0
dst-port 0quoted
link mlx4_0/1 lqpn 539 qp-type RC state CONNECT ps TCP dev-type ETHtransport-type IB pid 30494 comm rping src-addr 172.16.99.1 src-port 7174 dst-addr 172.16.99.1 dst-port 43670quoted
link mlx4_0/1 lqpn 538 qp-type RC state CONNECT ps TCP dev-type ETHtransport-type IB pid 30492 comm rping src-addr 172.16.99.1 src-port 43670 dst-addr 172.16.99.1 dst-port 7174quoted
# rdma resource show cm_id dst-port 7174 link cxgb4_0/2 lqpn 1040 qp-type RC state CONNECT ps TCP dev-type ETHtransport-type IWARP pid 30498 comm rping src-addr 172.16.2.1 src-port 38246 dst-addr 172.16.2.1 dst-port 7174quoted
link mlx4_0/1 lqpn 538 qp-type RC state CONNECT ps TCP dev-type ETHtransport-type IB pid 30492 comm rping src-addr 172.16.99.1 src-port 43670 dst-addr 172.16.99.1 dst-port 7174quoted
Signed-off-by: Steve Wise <redacted> --- rdma/rdma.h | 1 + rdma/res.c | 312++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-quoted
rdma/utils.c | 12 +++ 3 files changed, 321 insertions(+), 4 deletions(-)Thanks, for doing it.quoted
diff --git a/rdma/rdma.h b/rdma/rdma.h index 5809f70..1ef0942 100644 --- a/rdma/rdma.h +++ b/rdma/rdma.h@@ -18,6 +18,7 @@ #include <libmnl/libmnl.h> #include <rdma/rdma_netlink.h> #include <time.h> +#include <net/if_arp.h> #include "list.h" #include "utils.h"diff --git a/rdma/res.c b/rdma/res.c index 2a63e71..beae7dc 100644 --- a/rdma/res.c +++ b/rdma/res.c@@ -16,9 +16,11 @@ static int res_help(struct rd *rd) { pr_out("Usage: %s resource\n", rd->filename); pr_out(" resource show [DEV]\n"); - pr_out(" resource show [qp]\n"); + pr_out(" resource show [qp|cm_id]\n"); pr_out(" resource show qp link [DEV/PORT]\n"); pr_out(" resource show qp link [DEV/PORT] [FILTER-NAMEFILTER-VALUE]\n");quoted
+ pr_out(" resource show cm_id link [DEV/PORT]\n"); + pr_out(" resource show cm_id link [DEV/PORT] [FILTER-NAMEFILTER-VALUE]\n");quoted
return 0; }@@ -431,6 +433,278 @@ static int res_qp_parse_cb(const struct nlmsghdr*nlh, void *data)quoted
return MNL_CB_OK; } +static void print_qp_type(struct rd *rd, uint32_t val) +{ + if (rd->json_output) + jsonw_string_field(rd->jw, "qp-type", + qp_types_to_str(val)); + else + pr_out("qp-type %s ", qp_types_to_str(val)); +} + +static const char *cm_id_state_to_str(uint8_t idx) +{ + static const char * const cm_id_states_str[] = { "IDLE","ADDR_QUERY",quoted
+ "ADDR_RESOLVED","ROUTE_QUERY", "ROUTE_RESOLVED",quoted
+ "CONNECT","DISCONNECT",quoted
+ "ADDR_BOUND","LISTEN", "DEVICE_REMOVAL", "DESTROYING" };quoted
+ + if (idx < ARRAY_SIZE(cm_id_states_str)) + return cm_id_states_str[idx]; + return "UNKNOWN"; +} + +enum rdma_port_space { + RDMA_PS_SDP = 0x0001,Do we still support this PS? It is not set in the kernel and Parav posted internal patch to remove it.quoted
+ RDMA_PS_IPOIB = 0x0002, + RDMA_PS_IB = 0x013F, + RDMA_PS_TCP = 0x0106, + RDMA_PS_UDP = 0x0111, +}; + +static const char *cm_id_ps_to_str(uint32_t ps) +{ + switch (ps) { + case RDMA_PS_SDP: + return "SDP";The same question
You're right. No more PS_SDP. Also, rdma tool should get the rdma_port_space enum from <include/rdma/rdma_cma.h>.
quoted
+ case RDMA_PS_IPOIB: + return "IPoIB"; + case RDMA_PS_IB: + return "IPoIB"; + case RDMA_PS_TCP: + return "TCP"; + case RDMA_PS_UDP: + return "UDP"; + default: + return "---"; + } +} + +static const char *cm_id_dev_type_to_str(uint8_t dev_type) +{ + switch (dev_type) { + case ARPHRD_INFINIBAND: + return "IB"; + case ARPHRD_ETHER: + return "ETH"; + default: + return "---"; + } +} + +static const char *cm_id_transport_type_to_str(uint8_t transport_type) +{ + static const char * const transport_type_str[] = { "IB", "IWARP","USNIC", "USNIC/UDP" };quoted
+I know that it is part of CM_ID, but wonder if node_type of device is not enough. The same question goes for device type, isn't it part of "rdma dev .." output?
I've gotten rid off dev_type, transport_type, and network_type based on feedback from the kernel patches. So I'll remove it from this series as well.
quoted
+ if (transport_type < ARRAY_SIZE(transport_type_str)) + return transport_type_str[transport_type]; + return "---"; +} + +static void print_cm_id_state(struct rd *rd, uint8_t state) +{ + if (rd->json_output) { + jsonw_string_field(rd->jw, "state",cm_id_state_to_str(state));quoted
+ return; + } + pr_out("state %s ", cm_id_state_to_str(state)); +} + +static void print_ps(struct rd *rd, uint32_t ps) +{ + if (rd->json_output) { + jsonw_string_field(rd->jw, "ps", cm_id_ps_to_str(ps)); + return; + } + pr_out("ps %s ", cm_id_ps_to_str(ps)); +} + +static void print_dev_type(struct rd *rd, uint8_t dev_type) +{ + if (rd->json_output) { + jsonw_string_field(rd->jw, "dev-type",cm_id_dev_type_to_str(dev_type));quoted
+ return; + } + pr_out("dev-type %s ", cm_id_dev_type_to_str(dev_type)); +} + +static void print_transport_type(struct rd *rd, uint8_t transport_type) +{ + if (rd->json_output) { + jsonw_string_field(rd->jw, "transport-type",cm_id_transport_type_to_str(transport_type));quoted
+ return; + } + pr_out("transport-type %s ",cm_id_transport_type_to_str(transport_type));quoted
+} + +static void print_ipaddr(struct rd *rd, const char *key, char *addrstr) +{ + if (rd->json_output) { + jsonw_string_field(rd->jw, key, addrstr); + return; + } + pr_out("%s %s ", key, addrstr); +} + +static void print_ipport(struct rd *rd, const char *key, uint16_t
ipport)
quoted
+{ + if (rd->json_output) { + jsonw_uint_field(rd->jw, key, ipport); + return; + } + pr_out("%s %u ", key, ipport); +} + +static int res_cm_id_parse_cb(const struct nlmsghdr *nlh, void *data) +{ + struct nlattr *tb[RDMA_NLDEV_ATTR_MAX] = {}; + struct nlattr *nla_table, *nla_entry; + struct rd *rd = data; + const char *name; + int idx; + + mnl_attr_parse(nlh, 0, rd_attr_cb, tb); + if (!tb[RDMA_NLDEV_ATTR_DEV_INDEX] || + !tb[RDMA_NLDEV_ATTR_DEV_NAME] || + !tb[RDMA_NLDEV_ATTR_RES_CM_ID]) + return MNL_CB_ERROR; + + name = mnl_attr_get_str(tb[RDMA_NLDEV_ATTR_DEV_NAME]); + idx = mnl_attr_get_u32(tb[RDMA_NLDEV_ATTR_DEV_INDEX]); + nla_table = tb[RDMA_NLDEV_ATTR_RES_CM_ID]; + mnl_attr_for_each_nested(nla_entry, nla_table) { + struct nlattr *nla_line[RDMA_NLDEV_ATTR_MAX] = {}; + uint8_t dev_type, transport_type; + char src_addr_str[INET6_ADDRSTRLEN]; + char dst_addr_str[INET6_ADDRSTRLEN]; + uint8_t *src_addr, *dst_addr; + uint16_t src_port, dst_port; + uint32_t port = 0, pid = 0; + uint8_t type, state; + uint32_t lqpn = 0, ps; + char *comm = NULL; + int err; + + err = mnl_attr_parse_nested(nla_entry, rd_attr_cb,
nla_line);
quoted
+ if (err != MNL_CB_OK) + return -EINVAL; + + if (!nla_line[RDMA_NLDEV_ATTR_RES_TYPE] || + !nla_line[RDMA_NLDEV_ATTR_RES_STATE] || + (!nla_line[RDMA_NLDEV_ATTR_RES_IPV4_SADDR] && + !nla_line[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]) || + (!nla_line[RDMA_NLDEV_ATTR_RES_IPV4_DADDR] && + !nla_line[RDMA_NLDEV_ATTR_RES_IPV6_DADDR]) || + !nla_line[RDMA_NLDEV_ATTR_RES_IP_SPORT] || + !nla_line[RDMA_NLDEV_ATTR_RES_IP_DPORT] || + (!nla_line[RDMA_NLDEV_ATTR_RES_PID] && + !nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME])) { + return MNL_CB_ERROR;It is unreadable, any chances to use intermediate variables with descriptive names?
This gets cleaned up because the kernel driver will now pass only the sockaddr_storage up for the addresses. It simplifies this code, so I think it'll be cleaner on the next version (I'll send it out today or tomorrow hopefully).
quoted
+ } + + if (nla_line[RDMA_NLDEV_ATTR_PORT_INDEX]) + port =mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_PORT_INDEX]);quoted
+ + if (port && port != rd->port_idx) + continue; + + if (nla_line[RDMA_NLDEV_ATTR_RES_LQPN]) + lqpn =mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_LQPN]);quoted
+ if (rd_check_is_filtered(rd, "lqpn", lqpn)) + continue; + + ps =mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PS]);quoted
+ if (rd_check_is_string_filtered(rd, "ps",
cm_id_ps_to_str(ps)))
quoted
+ continue; + + type =mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_TYPE]);quoted
+ if (rd_check_is_string_filtered(rd, "qp-type",qp_types_to_str(type)))quoted
+ continue; + + state =mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_STATE]);quoted
+ if (rd_check_is_string_filtered(rd, "state",cm_id_state_to_str(state)))quoted
+ continue; + + dev_type =mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_DEV_TYPE]);quoted
+ if (rd_check_is_string_filtered(rd, "dev-type",cm_id_dev_type_to_str(dev_type)))quoted
+ continue; + + transport_type =mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_TRANSPORT_TYPE]);quoted
+ if (rd_check_is_string_filtered(rd, "transport-type",cm_id_transport_type_to_str(transport_type)))quoted
+ continue; + + if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) { + pid =mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);quoted
+ comm = get_task_name(pid); + } + if (rd_check_is_filtered(rd, "pid", pid))free(comm) here
Oops! Thanks!
quoted
+ continue; + + if (nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME]) { + /* discard const from mnl_attr_get_str */ + comm = (char*)mnl_attr_get_str(nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME]);quoted
+ } + + if (nla_line[RDMA_NLDEV_ATTR_RES_IPV4_SADDR]) { + if (!nla_line[RDMA_NLDEV_ATTR_RES_IPV4_DADDR]) + return -EINVAL; + src_addr =mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV4_SADDR]);quoted
+ if (!inet_ntop(AF_INET, src_addr, src_addr_str,INET6_ADDRSTRLEN))quoted
+ return -EINVAL; + dst_addr =mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV4_DADDR]);quoted
+ if (!inet_ntop(AF_INET, dst_addr, dst_addr_str,INET6_ADDRSTRLEN))quoted
+ return -EINVAL; + } else { + if (!nla_line[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]||quoted
+ !nla_line[RDMA_NLDEV_ATTR_RES_IPV6_DADDR]) + return -EINVAL; + src_addr =mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]);quoted
+ if (!inet_ntop(AF_INET6, src_addr, src_addr_str,INET6_ADDRSTRLEN))quoted
+ return -EINVAL; + dst_addr =mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV6_DADDR]);quoted
+ if (!inet_ntop(AF_INET6, dst_addr, dst_addr_str,INET6_ADDRSTRLEN))quoted
+ return -EINVAL; + } + if (rd_check_is_string_filtered(rd, "src-addr",
src_addr_str))
quoted
+ continue; + if (rd_check_is_string_filtered(rd, "dst-addr",
dst_addr_str))
quoted
+ continue; + + src_port =mnl_attr_get_u16(nla_line[RDMA_NLDEV_ATTR_RES_IP_SPORT]);quoted
+ dst_port =mnl_attr_get_u16(nla_line[RDMA_NLDEV_ATTR_RES_IP_DPORT]);quoted
+ if (rd_check_is_filtered(rd, "src-port", src_port)) + continue; + if (rd_check_is_filtered(rd, "dst-port", dst_port)) + continue; +The same memory leaks as above, I put get_task_name() to the end of QP parsing code with purpose to avoid dealing with free() calls.
Sounds good. I'll look at your qp code.
quoted
+ if (rd->json_output) + jsonw_start_array(rd->jw); + + print_link(rd, idx, name, port, nla_line); + print_lqpn(rd, lqpn); + print_qp_type(rd, type); + print_cm_id_state(rd, state); + print_ps(rd, ps); + print_dev_type(rd, dev_type); + print_transport_type(rd, transport_type); + print_pid(rd, pid); + print_comm(rd, comm, nla_line); + print_ipaddr(rd, "src-addr", src_addr_str); + print_ipport(rd, "src-port", src_port); + print_ipaddr(rd, "dst-addr", dst_addr_str); + print_ipport(rd, "dst-port", dst_port);Does "ip tool" have standard Re presentation for addr<->port tupples? What about the following format src 1.1.1.1:1234 dst 2.2.2.2:6789?
That looks fine for me. I'm not sure if there's an 'ip' command that displays ports along with addresses. But let us go with add:port.
quoted
+ + if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) + free(comm); + + if (rd->json_output) + jsonw_end_array(rd->jw); + else + pr_out("\n"); + } + return MNL_CB_OK; +} + RES_FUNC(res_no_args, RDMA_NLDEV_CMD_RES_GET, NULL, true); static const struct@@ -438,9 +712,9 @@ filters qp_valid_filters[MAX_NUMBER_OF_FILTERS]= {{ .name = "link",quoted
.is_number = false }, { .name = "lqpn", .is_number = true }, - { .name = "rqpn", - .is_number = true }, - { .name = "pid", + { .name = "type", + .is_number = false }, + { .name = "cm_id_state", .is_number = true }, { .name = "sq-psn", .is_number = true },Why did you change qp_valid_filters?
Oops...I'll fix this.
quoted
@@ -455,11 +729,41 @@ filtersqp_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",quoted
RES_FUNC(res_qp, RDMA_NLDEV_CMD_RES_QP_GET, qp_valid_filters,false);quoted
+static const struct +filters cm_id_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link", + .is_number = false }, + { .name = "lqpn", + .is_number = true }, + { .name = "qp-type", + .is_number = false }, + { .name = "state", + .is_number = false }, + { .name = "ps", + .is_number = false }, + { .name = "dev-type", + .is_number = false }, + { .name =
"transport-type",
quoted
+ .is_number = false }, + { .name = "pid", + .is_number = true }, + { .name = "src-addr", + .is_number = false }, + { .name = "src-port", + .is_number = true }, + { .name = "dst-addr", + .is_number = false }, + { .name = "dst-port", + .is_number = true }}; + +RES_FUNC(res_cm_id, RDMA_NLDEV_CMD_RES_CM_ID_GET,cm_id_valid_filters,quoted
+ false); + static int res_show(struct rd *rd) { const struct rd_cmd cmds[] = { { NULL, res_no_args }, { "qp", res_qp }, + { "cm_id", res_cm_id }, { 0 } };diff --git a/rdma/utils.c b/rdma/utils.c index f946016..906ca73 100644 --- a/rdma/utils.c +++ b/rdma/utils.c@@ -375,6 +375,18 @@ static const enum mnl_attr_data_typenldev_policy[RDMA_NLDEV_ATTR_MAX] = {quoted
[RDMA_NLDEV_ATTR_RES_STATE] = MNL_TYPE_U8, [RDMA_NLDEV_ATTR_RES_PID] = MNL_TYPE_U32, [RDMA_NLDEV_ATTR_RES_KERN_NAME] =MNL_TYPE_NUL_STRING,quoted
+ [RDMA_NLDEV_ATTR_RES_CM_ID] = MNL_TYPE_NESTED, + [RDMA_NLDEV_ATTR_RES_CM_ID_ENTRY] = MNL_TYPE_NESTED, + [RDMA_NLDEV_ATTR_RES_PS] = MNL_TYPE_U32, + [RDMA_NLDEV_ATTR_RES_IPV4_SADDR] =MNL_TYPE_UNSPEC,quoted
+ [RDMA_NLDEV_ATTR_RES_IPV4_DADDR] =MNL_TYPE_UNSPEC,quoted
+ [RDMA_NLDEV_ATTR_RES_IPV6_SADDR] =MNL_TYPE_UNSPEC,quoted
+ [RDMA_NLDEV_ATTR_RES_IPV6_DADDR] =MNL_TYPE_UNSPEC,quoted
+ [RDMA_NLDEV_ATTR_RES_IP_SPORT] = MNL_TYPE_U16, + [RDMA_NLDEV_ATTR_RES_IP_DPORT] = MNL_TYPE_U16, + [RDMA_NLDEV_ATTR_RES_DEV_TYPE] = MNL_TYPE_U8, + [RDMA_NLDEV_ATTR_RES_TRANSPORT_TYPE] = MNL_TYPE_U8, + [RDMA_NLDEV_ATTR_RES_NETWORK_TYPE] = MNL_TYPE_U8, }; int rd_attr_cb(const struct nlattr *attr, void *data) -- 1.8.3.1