[PATCH RFC iproute-next 0/5] cm_id, cq, mr, and pd resource tracking

[PATCH RFC iproute-next 0/5] cm_id, cq, mr, and pd resource tracking

[Steve Wise]

Hello,

This series enhances the iproute2 rdma tool to include dumping of
connection manager id (cm_id), completion queue (cq), memory region (mr),
and protection domain (pd) rdma resources.  It is the user-space part of
the resource tracking series under review now on the linux-rdma list [1].

Thanks,

Steve.

[1]https://www.spinics.net/lists/linux-rdma/msg60912.html

Steve Wise (5):
  rdma: update rdma_netlink.h
  rdma: Add CM_ID resource tracking information
  rdma: Add CQ resource tracking information
  rdma: Add MR resource tracking information
  rdma: Add PD resource tracking information

 include/json_writer.h            |   2 +
 include/uapi/rdma/rdma_netlink.h |  62 +++-
 lib/json_writer.c                |  11 +
 rdma/rdma.h                      |   1 +
 rdma/res.c                       | 657 ++++++++++++++++++++++++++++++++++++++-
 rdma/utils.c                     |  24 ++
 6 files changed, 750 insertions(+), 7 deletions(-)

-- 
1.8.3.1

[PATCH RFC iproute-next 1/5] rdma: update rdma_netlink.h

[Steve Wise]

From: Steve Wise <swise@opengridcomputing.com>

Pull in the latest rdma_netlink.h which has support for
the rdma nldev resource tracking objects being added
with this patch series.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
---
 include/uapi/rdma/rdma_netlink.h | 62 ++++++++++++++++++++++++++++++++++++++--
 1 file changed, 59 insertions(+), 3 deletions(-)

diff --git a/include/uapi/rdma/rdma_netlink.h b/include/uapi/rdma/rdma_netlink.h
index dbac3b8..7052a7c 100644
--- a/include/uapi/rdma/rdma_netlink.h
+++ b/include/uapi/rdma/rdma_netlink.h
@@ -1,6 +1,6 @@
 /* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
-#ifndef _RDMA_NETLINK_H
-#define _RDMA_NETLINK_H
+#ifndef _UAPI_RDMA_NETLINK_H
+#define _UAPI_RDMA_NETLINK_H
 
 #include <linux/types.h>
 
@@ -238,6 +238,14 @@ enum rdma_nldev_command {
 
 	RDMA_NLDEV_CMD_RES_QP_GET, /* can dump */
 
+	RDMA_NLDEV_CMD_RES_CM_ID_GET, /* can dump */
+
+	RDMA_NLDEV_CMD_RES_CQ_GET, /* can dump */
+
+	RDMA_NLDEV_CMD_RES_MR_GET, /* can dump */
+
+	RDMA_NLDEV_CMD_RES_PD_GET, /* can dump */
+
 	RDMA_NLDEV_NUM_OPS
 };
 
@@ -350,6 +358,54 @@ enum rdma_nldev_attr {
 	 */
 	RDMA_NLDEV_ATTR_RES_KERN_NAME,		/* string */
 
+	RDMA_NLDEV_ATTR_RES_CM_ID,		/* nested table */
+	RDMA_NLDEV_ATTR_RES_CM_ID_ENTRY,	/* nested table */
+	/*
+	 * rdma_cm_id port space.
+	 */
+	RDMA_NLDEV_ATTR_RES_PS,			/* u32 */
+	/*
+	 * Source and destination IP address and port attributes.
+	 */
+	RDMA_NLDEV_ATTR_RES_IPV4_SADDR,		/* u8[4] */
+	RDMA_NLDEV_ATTR_RES_IPV4_DADDR,		/* u8[4] */
+	RDMA_NLDEV_ATTR_RES_IPV6_SADDR,		/* u8[16] */
+	RDMA_NLDEV_ATTR_RES_IPV6_DADDR,		/* u8[16] */
+	RDMA_NLDEV_ATTR_RES_IP_SPORT,		/* u16 */
+	RDMA_NLDEV_ATTR_RES_IP_DPORT,		/* u16 */
+	/*
+	 * ARPHRD_INFINIBAND, ARPHRD_ETHER, ...
+	 */
+	RDMA_NLDEV_ATTR_RES_DEV_TYPE,		/* u8 */
+	/*
+	 * enum enum rdma_transport_type (IB, IWARP, ...)
+	 */
+	RDMA_NLDEV_ATTR_RES_TRANSPORT_TYPE,	/* u8 */
+	/*
+	 * enum rdma_network_type (IB, IPv4, IPv6,...)
+	 */
+	RDMA_NLDEV_ATTR_RES_NETWORK_TYPE,	/* u8 */
+
+	RDMA_NLDEV_ATTR_RES_CQ,			/* nested table */
+	RDMA_NLDEV_ATTR_RES_CQ_ENTRY,		/* nested table */
+	RDMA_NLDEV_ATTR_RES_CQE,		/* u32 */
+	RDMA_NLDEV_ATTR_RES_USECNT,		/* u64 */
+	RDMA_NLDEV_ATTR_RES_POLL_CTX,		/* u8 */
+
+	RDMA_NLDEV_ATTR_RES_MR,			/* nested table */
+	RDMA_NLDEV_ATTR_RES_MR_ENTRY,		/* nested table */
+	RDMA_NLDEV_ATTR_RES_RKEY,		/* u32 */
+	RDMA_NLDEV_ATTR_RES_LKEY,		/* u32 */
+	RDMA_NLDEV_ATTR_RES_IOVA,		/* u64 */
+	RDMA_NLDEV_ATTR_RES_MRLEN,		/* u64 */
+	RDMA_NLDEV_ATTR_RES_PGSIZE,		/* u32 */
+
+	RDMA_NLDEV_ATTR_RES_PD,			/* nested table */
+	RDMA_NLDEV_ATTR_RES_PD_ENTRY,		/* nested table */
+	RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY,	/* u32 */
+	RDMA_NLDEV_ATTR_RES_PD_FLAGS,		/* u32 */
+	RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY,	/* u32 */
+
 	RDMA_NLDEV_ATTR_MAX
 };
-#endif /* _RDMA_NETLINK_H */
+#endif /* _UAPI_RDMA_NETLINK_H */
-- 
1.8.3.1

[PATCH RFC iproute-next 5/5] rdma: Add PD resource tracking information

[Steve Wise]

Sample output:

# rdma resource show pd
link cxgb4_0/- local_dma_lkey 0x0 usecnt 4 flags 0x0 pid 30503 comm rping
link cxgb4_0/- local_dma_lkey 0x0 usecnt 0 flags 0x0 pid 30488 comm rping
link cxgb4_0/- local_dma_lkey 0x0 usecnt 5 flags 0x0 pid 30498 comm rping
link cxgb4_0/- local_dma_lkey 0x0 usecnt 0 flags 0x0 pid 30501 comm rping
link cxgb4_0/- local_dma_lkey 0x0 usecnt 0 flags 0x0 pid 0 comm [ib_srp]
link mlx4_0/- local_dma_lkey 0x0 usecnt 4 flags 0x0 pid 30494 comm rping
link mlx4_0/- local_dma_lkey 0x0 usecnt 0 flags 0x0 pid 30488 comm rping
link mlx4_0/- local_dma_lkey 0x0 usecnt 5 flags 0x0 pid 30489 comm rping
link mlx4_0/- local_dma_lkey 0x0 usecnt 0 flags 0x0 pid 30492 comm rping
link mlx4_0/- local_dma_lkey 0x8000 usecnt 0 flags 0x0 pid 0 comm [ib_srpt]
link mlx4_0/- local_dma_lkey 0x8000 usecnt 0 flags 0x0 pid 0 comm [ib_srp]
link mlx4_0/- local_dma_lkey 0x8000 usecnt 1 flags 0x0 pid 0 comm [ib_core]

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
---
 rdma/res.c | 96 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 96 insertions(+)

diff --git a/rdma/res.c b/rdma/res.c
index 2b67d25..5e6bc42 100644
--- a/rdma/res.c
+++ b/rdma/res.c
@@ -932,6 +932,93 @@ static int res_mr_parse_cb(const struct nlmsghdr *nlh, void *data)
 	return MNL_CB_OK;
 }
 
+static void print_flags(struct rd *rd, uint32_t val)
+{
+	if (rd->json_output)
+		jsonw_xint_field(rd->jw, "flags", val);
+	else
+		pr_out("flags 0x%x ", val);
+}
+
+static int res_pd_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;
+	uint32_t 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_PD])
+		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_PD];
+
+	mnl_attr_for_each_nested(nla_entry, nla_table) {
+		struct nlattr *nla_line[RDMA_NLDEV_ATTR_MAX] = {};
+		uint32_t local_dma_lkey, unsafe_global_rkey = 0;
+		uint32_t flags, pid = 0;
+		char *comm = NULL;
+		uint64_t usecnt;
+		int err;
+
+		err = mnl_attr_parse_nested(nla_entry, rd_attr_cb, nla_line);
+		if (err != MNL_CB_OK)
+			return MNL_CB_ERROR;
+
+		if (!nla_line[RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY] ||
+		    !nla_line[RDMA_NLDEV_ATTR_RES_USECNT] ||
+		    !nla_line[RDMA_NLDEV_ATTR_RES_PD_FLAGS] ||
+		    (!nla_line[RDMA_NLDEV_ATTR_RES_PID] &&
+		     !nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME])) {
+			return MNL_CB_ERROR;
+		}
+
+		local_dma_lkey = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY]);
+		usecnt = mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_USECNT]);
+		flags = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PD_FLAGS]);
+		if (nla_line[RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY])
+			unsafe_global_rkey = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY]);
+
+		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
+			pid = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
+			comm = get_task_name(pid);
+		}
+
+		if (rd_check_is_filtered(rd, "pid", pid))
+			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]);
+
+		if (rd->json_output)
+			jsonw_start_array(rd->jw);
+
+		print_link(rd, idx, name, 0, nla_line);
+		print_key(rd, "local_dma_lkey", local_dma_lkey);
+		print_usecnt(rd, usecnt);
+		print_flags(rd, flags);
+		if (nla_line[RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY])
+			print_key(rd, "unsafe_global_rkey", unsafe_global_rkey);
+		print_pid(rd, pid);
+		print_comm(rd, comm, nla_line);
+
+		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
@@ -1005,6 +1092,14 @@ filters mr_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
 
 RES_FUNC(res_mr,	RDMA_NLDEV_CMD_RES_MR_GET, mr_valid_filters, true);
 
+static const struct
+filters pd_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
+						   .is_number = false },
+						   { .name = "pid",
+						   .is_number = true }};
+
+RES_FUNC(res_pd,	RDMA_NLDEV_CMD_RES_PD_GET, pd_valid_filters, true);
+
 static int res_show(struct rd *rd)
 {
 	const struct rd_cmd cmds[] = {
@@ -1013,6 +1108,7 @@ static int res_show(struct rd *rd)
 		{ "cm_id",	res_cm_id	},
 		{ "cq",		res_cq		},
 		{ "mr",		res_mr		},
+		{ "pd",		res_pd		},
 		{ 0 }
 	};
 
-- 
1.8.3.1

Re: [PATCH RFC iproute-next 5/5] rdma: Add PD resource tracking information

[Leon Romanovsky]

[-- Attachment #1: Type: text/plain, Size: 306 bytes --]

On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> Sample output:
>
> # rdma resource show pd
> link cxgb4_0/- local_dma_lkey 0x0 usecnt 4 flags 0x0 pid 30503 comm rping

One more thing, flags need to be pre-parsed and accessible with "-d" command,
as we did with dev,link capabilities.

Thanks

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

RE: [PATCH RFC iproute-next 5/5] rdma: Add PD resource tracking information

[Steve Wise]

> 
> On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> > Sample output:
> >
> > # rdma resource show pd
> > link cxgb4_0/- local_dma_lkey 0x0 usecnt 4 flags 0x0 pid 30503 comm
rping
> 
> One more thing, flags need to be pre-parsed and accessible with "-d"
> command,
> as we did with dev,link capabilities.

Will do.  

Thanks for reviewing this series!

Steve.

RE: [PATCH RFC iproute-next 5/5] rdma: Add PD resource tracking information

[Steve Wise]

> > On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> > > Sample output:
> > >
> > > # rdma resource show pd
> > > link cxgb4_0/- local_dma_lkey 0x0 usecnt 4 flags 0x0 pid 30503 comm
> rping
> >
> > One more thing, flags need to be pre-parsed and accessible with "-d"
> > command,
> > as we did with dev,link capabilities.
> 
> Will do.
> 
> Thanks for reviewing this series!
> 

Turns out the only flags field was in the pd, and pd->flags only has one
flag, currently, indicating that the global_dma_rkey is in use.  Since the
kernel side only sends up the global_dma_rkey if that flag is set, I just
dropped the PD_FLAGS attribute in both patch series.  If pd flags grow, we
can add it, but I don't expect that.


Steve.

[PATCH RFC iproute-next 2/5] rdma: Add CM_ID resource tracking information

[Steve Wise]

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  show cm_id
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 0
link cxgb4_0/2 lqpn 1048 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IWARP pid 30503 comm rping src-addr 172.16.2.1 src-port 7174 dst-addr 172.16.2.1 dst-port 38246
link cxgb4_0/2 lqpn 1040 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IWARP pid 30498 comm rping src-addr 172.16.2.1 src-port 38246 dst-addr 172.16.2.1 dst-port 7174
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 0
link mlx4_0/1 lqpn 539 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IB pid 30494 comm rping src-addr 172.16.99.1 src-port 7174 dst-addr 172.16.99.1 dst-port 43670
link mlx4_0/1 lqpn 538 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IB pid 30492 comm rping src-addr 172.16.99.1 src-port 43670 dst-addr 172.16.99.1 dst-port 7174

# rdma resource show cm_id dst-port 7174
link cxgb4_0/2 lqpn 1040 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IWARP pid 30498 comm rping src-addr 172.16.2.1 src-port 38246 dst-addr 172.16.2.1 dst-port 7174
link mlx4_0/1 lqpn 538 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IB pid 30492 comm rping src-addr 172.16.99.1 src-port 43670 dst-addr 172.16.99.1 dst-port 7174

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
---
 rdma/rdma.h  |   1 +
 rdma/res.c   | 312 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
 rdma/utils.c |  12 +++
 3 files changed, 321 insertions(+), 4 deletions(-)

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-NAME FILTER-VALUE]\n");
+	pr_out("          resource show cm_id link [DEV/PORT]\n");
+	pr_out("          resource show cm_id link [DEV/PORT] [FILTER-NAME FILTER-VALUE]\n");
 	return 0;
 }
 
@@ -431,6 +433,278 @@ static int res_qp_parse_cb(const struct nlmsghdr *nlh, void *data)
 	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",
+						      "ADDR_RESOLVED", "ROUTE_QUERY", "ROUTE_RESOLVED",
+						      "CONNECT", "DISCONNECT",
+						      "ADDR_BOUND", "LISTEN", "DEVICE_REMOVAL", "DESTROYING" };
+
+	if (idx < ARRAY_SIZE(cm_id_states_str))
+		return cm_id_states_str[idx];
+	return "UNKNOWN";
+}
+
+enum rdma_port_space {
+	RDMA_PS_SDP   = 0x0001,
+	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";
+	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" };
+
+	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));
+		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));
+		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));
+		return;
+	}
+	pr_out("transport-type %s ", cm_id_transport_type_to_str(transport_type));
+}
+
+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)
+{
+	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);
+		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;
+		}
+
+		if (nla_line[RDMA_NLDEV_ATTR_PORT_INDEX])
+			port = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_PORT_INDEX]);
+
+		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]);
+		if (rd_check_is_filtered(rd, "lqpn", lqpn))
+			continue;
+
+		ps = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PS]);
+		if (rd_check_is_string_filtered(rd, "ps", cm_id_ps_to_str(ps)))
+			continue;
+
+		type = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_TYPE]);
+		if (rd_check_is_string_filtered(rd, "qp-type", qp_types_to_str(type)))
+			continue;
+
+		state = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_STATE]);
+		if (rd_check_is_string_filtered(rd, "state", cm_id_state_to_str(state)))
+			continue;
+
+		dev_type = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_DEV_TYPE]);
+		if (rd_check_is_string_filtered(rd, "dev-type", cm_id_dev_type_to_str(dev_type)))
+			continue;
+
+		transport_type = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_TRANSPORT_TYPE]);
+		if (rd_check_is_string_filtered(rd, "transport-type", cm_id_transport_type_to_str(transport_type)))
+			continue;
+
+		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
+			pid = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
+			comm = get_task_name(pid);
+		}
+		if (rd_check_is_filtered(rd, "pid", pid))
+			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]);
+		}
+
+		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]);
+			if (!inet_ntop(AF_INET, src_addr, src_addr_str, INET6_ADDRSTRLEN))
+				return -EINVAL;
+			dst_addr = mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV4_DADDR]);
+			if (!inet_ntop(AF_INET, dst_addr, dst_addr_str, INET6_ADDRSTRLEN))
+				return -EINVAL;
+		} else {
+			if (!nla_line[RDMA_NLDEV_ATTR_RES_IPV6_SADDR] ||
+			    !nla_line[RDMA_NLDEV_ATTR_RES_IPV6_DADDR])
+				return -EINVAL;
+			src_addr = mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]);
+			if (!inet_ntop(AF_INET6, src_addr, src_addr_str, INET6_ADDRSTRLEN))
+				return -EINVAL;
+			dst_addr = mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV6_DADDR]);
+			if (!inet_ntop(AF_INET6, dst_addr, dst_addr_str, INET6_ADDRSTRLEN))
+				return -EINVAL;
+		}
+		if (rd_check_is_string_filtered(rd, "src-addr", src_addr_str))
+			continue;
+		if (rd_check_is_string_filtered(rd, "dst-addr", dst_addr_str))
+			continue;
+
+		src_port = mnl_attr_get_u16(nla_line[RDMA_NLDEV_ATTR_RES_IP_SPORT]);
+		dst_port = mnl_attr_get_u16(nla_line[RDMA_NLDEV_ATTR_RES_IP_DPORT]);
+		if (rd_check_is_filtered(rd, "src-port", src_port))
+			continue;
+		if (rd_check_is_filtered(rd, "dst-port", dst_port))
+			continue;
+
+		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);
+
+		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",
 						   .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 },
@@ -455,11 +729,41 @@ filters qp_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
 
 RES_FUNC(res_qp,	RDMA_NLDEV_CMD_RES_QP_GET, qp_valid_filters, false);
 
+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",
+						   .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,
+	 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_type nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
 	[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,
+	[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,
+	[RDMA_NLDEV_ATTR_RES_IPV4_DADDR]	= MNL_TYPE_UNSPEC,
+	[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]	= MNL_TYPE_UNSPEC,
+	[RDMA_NLDEV_ATTR_RES_IPV6_DADDR]	= MNL_TYPE_UNSPEC,
+	[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

Re: [PATCH RFC iproute-next 2/5] rdma: Add CM_ID resource tracking information

[Leon Romanovsky]

[-- Attachment #1: Type: text/plain, Size: 15532 bytes --]

On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> 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  show cm_id
> 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 0
> link cxgb4_0/2 lqpn 1048 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IWARP pid 30503 comm rping src-addr 172.16.2.1 src-port 7174 dst-addr 172.16.2.1 dst-port 38246
> link cxgb4_0/2 lqpn 1040 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IWARP pid 30498 comm rping src-addr 172.16.2.1 src-port 38246 dst-addr 172.16.2.1 dst-port 7174
> 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 0
> link mlx4_0/1 lqpn 539 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IB pid 30494 comm rping src-addr 172.16.99.1 src-port 7174 dst-addr 172.16.99.1 dst-port 43670
> link mlx4_0/1 lqpn 538 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IB pid 30492 comm rping src-addr 172.16.99.1 src-port 43670 dst-addr 172.16.99.1 dst-port 7174
>
> # rdma resource show cm_id dst-port 7174
> link cxgb4_0/2 lqpn 1040 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IWARP pid 30498 comm rping src-addr 172.16.2.1 src-port 38246 dst-addr 172.16.2.1 dst-port 7174
> link mlx4_0/1 lqpn 538 qp-type RC state CONNECT ps TCP dev-type ETH transport-type IB pid 30492 comm rping src-addr 172.16.99.1 src-port 43670 dst-addr 172.16.99.1 dst-port 7174
>
> Signed-off-by: Steve Wise <swise@opengridcomputing.com>
> ---
>  rdma/rdma.h  |   1 +
>  rdma/res.c   | 312 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
>  rdma/utils.c |  12 +++
>  3 files changed, 321 insertions(+), 4 deletions(-)

Thanks, for doing it.

>
> 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-NAME FILTER-VALUE]\n");
> +	pr_out("          resource show cm_id link [DEV/PORT]\n");
> +	pr_out("          resource show cm_id link [DEV/PORT] [FILTER-NAME FILTER-VALUE]\n");
>  	return 0;
>  }
>
> @@ -431,6 +433,278 @@ static int res_qp_parse_cb(const struct nlmsghdr *nlh, void *data)
>  	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",
> +						      "ADDR_RESOLVED", "ROUTE_QUERY", "ROUTE_RESOLVED",
> +						      "CONNECT", "DISCONNECT",
> +						      "ADDR_BOUND", "LISTEN", "DEVICE_REMOVAL", "DESTROYING" };
> +
> +	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.

> +	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

> +	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" };
> +

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?

> +	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));
> +		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));
> +		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));
> +		return;
> +	}
> +	pr_out("transport-type %s ", cm_id_transport_type_to_str(transport_type));
> +}
> +
> +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)
> +{
> +	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);
> +		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?

> +		}
> +
> +		if (nla_line[RDMA_NLDEV_ATTR_PORT_INDEX])
> +			port = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_PORT_INDEX]);
> +
> +		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]);
> +		if (rd_check_is_filtered(rd, "lqpn", lqpn))
> +			continue;
> +
> +		ps = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PS]);
> +		if (rd_check_is_string_filtered(rd, "ps", cm_id_ps_to_str(ps)))
> +			continue;
> +
> +		type = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_TYPE]);
> +		if (rd_check_is_string_filtered(rd, "qp-type", qp_types_to_str(type)))
> +			continue;
> +
> +		state = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_STATE]);
> +		if (rd_check_is_string_filtered(rd, "state", cm_id_state_to_str(state)))
> +			continue;
> +
> +		dev_type = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_DEV_TYPE]);
> +		if (rd_check_is_string_filtered(rd, "dev-type", cm_id_dev_type_to_str(dev_type)))
> +			continue;
> +
> +		transport_type = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_TRANSPORT_TYPE]);
> +		if (rd_check_is_string_filtered(rd, "transport-type", cm_id_transport_type_to_str(transport_type)))
> +			continue;
> +
> +		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
> +			pid = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
> +			comm = get_task_name(pid);
> +		}
> +		if (rd_check_is_filtered(rd, "pid", pid))

free(comm) here

> +			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]);
> +		}
> +
> +		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]);
> +			if (!inet_ntop(AF_INET, src_addr, src_addr_str, INET6_ADDRSTRLEN))
> +				return -EINVAL;
> +			dst_addr = mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV4_DADDR]);
> +			if (!inet_ntop(AF_INET, dst_addr, dst_addr_str, INET6_ADDRSTRLEN))
> +				return -EINVAL;
> +		} else {
> +			if (!nla_line[RDMA_NLDEV_ATTR_RES_IPV6_SADDR] ||
> +			    !nla_line[RDMA_NLDEV_ATTR_RES_IPV6_DADDR])
> +				return -EINVAL;
> +			src_addr = mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]);
> +			if (!inet_ntop(AF_INET6, src_addr, src_addr_str, INET6_ADDRSTRLEN))
> +				return -EINVAL;
> +			dst_addr = mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV6_DADDR]);
> +			if (!inet_ntop(AF_INET6, dst_addr, dst_addr_str, INET6_ADDRSTRLEN))
> +				return -EINVAL;
> +		}
> +		if (rd_check_is_string_filtered(rd, "src-addr", src_addr_str))
> +			continue;
> +		if (rd_check_is_string_filtered(rd, "dst-addr", dst_addr_str))
> +			continue;
> +
> +		src_port = mnl_attr_get_u16(nla_line[RDMA_NLDEV_ATTR_RES_IP_SPORT]);
> +		dst_port = mnl_attr_get_u16(nla_line[RDMA_NLDEV_ATTR_RES_IP_DPORT]);
> +		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.

> +		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?

> +
> +		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",
>  						   .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?

> @@ -455,11 +729,41 @@ filters qp_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
>
>  RES_FUNC(res_qp,	RDMA_NLDEV_CMD_RES_QP_GET, qp_valid_filters, false);
>
> +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",
> +						   .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,
> +	 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_type nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
>  	[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,
> +	[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,
> +	[RDMA_NLDEV_ATTR_RES_IPV4_DADDR]	= MNL_TYPE_UNSPEC,
> +	[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]	= MNL_TYPE_UNSPEC,
> +	[RDMA_NLDEV_ATTR_RES_IPV6_DADDR]	= MNL_TYPE_UNSPEC,
> +	[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
>

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

RE: [PATCH RFC iproute-next 2/5] rdma: Add CM_ID resource tracking information

[Parav Pandit]

> -----Original Message-----
> From: linux-rdma-owner@vger.kernel.org [mailto:linux-rdma-
> owner@vger.kernel.org] On Behalf Of Leon Romanovsky
> Sent: Tuesday, February 20, 2018 6:58 AM
> To: Steve Wise <swise@opengridcomputing.com>
> Cc: dsahern@gmail.com; stephen@networkplumber.org;
> netdev@vger.kernel.org; linux-rdma@vger.kernel.org
> Subject: Re: [PATCH RFC iproute-next 2/5] rdma: Add CM_ID resource tracking
> information
> 
> On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> > 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  show
> > cm_id 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 0 link cxgb4_0/2 lqpn 1048
> > qp-type RC state CONNECT ps TCP dev-type ETH transport-type IWARP pid
> > 30503 comm rping src-addr 172.16.2.1 src-port 7174 dst-addr 172.16.2.1
> > dst-port 38246 link cxgb4_0/2 lqpn 1040 qp-type RC state CONNECT ps
> > TCP dev-type ETH transport-type IWARP pid 30498 comm rping src-addr
> > 172.16.2.1 src-port 38246 dst-addr 172.16.2.1 dst-port 7174 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 0 link mlx4_0/1 lqpn 539 qp-type RC state
> > CONNECT ps TCP dev-type ETH transport-type IB pid 30494 comm rping
> > src-addr 172.16.99.1 src-port 7174 dst-addr 172.16.99.1 dst-port 43670
> > link mlx4_0/1 lqpn 538 qp-type RC state CONNECT ps TCP dev-type ETH
> > transport-type IB pid 30492 comm rping src-addr 172.16.99.1 src-port
> > 43670 dst-addr 172.16.99.1 dst-port 7174
> >
> > # rdma resource show cm_id dst-port 7174 link cxgb4_0/2 lqpn 1040
> > qp-type RC state CONNECT ps TCP dev-type ETH transport-type IWARP pid
> > 30498 comm rping src-addr 172.16.2.1 src-port 38246 dst-addr
> > 172.16.2.1 dst-port 7174 link mlx4_0/1 lqpn 538 qp-type RC state
> > CONNECT ps TCP dev-type ETH transport-type IB pid 30492 comm rping
> > src-addr 172.16.99.1 src-port 43670 dst-addr 172.16.99.1 dst-port 7174
> >
> > Signed-off-by: Steve Wise <swise@opengridcomputing.com>
> > ---
> >  rdma/rdma.h  |   1 +
> >  rdma/res.c   | 312
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
> >  rdma/utils.c |  12 +++
> >  3 files changed, 321 insertions(+), 4 deletions(-)
> 
> Thanks, for doing it.
> 
> >
> > 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-NAME FILTER-
> VALUE]\n");
> > +	pr_out("          resource show cm_id link [DEV/PORT]\n");
> > +	pr_out("          resource show cm_id link [DEV/PORT] [FILTER-NAME
> FILTER-VALUE]\n");
> >  	return 0;
> >  }
> >
> > @@ -431,6 +433,278 @@ static int res_qp_parse_cb(const struct nlmsghdr
> *nlh, void *data)
> >  	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",
> > +						      "ADDR_RESOLVED",
> "ROUTE_QUERY", "ROUTE_RESOLVED",
> > +						      "CONNECT",
> "DISCONNECT",
> > +						      "ADDR_BOUND", "LISTEN",
> "DEVICE_REMOVAL", "DESTROYING" };
> > +
> > +	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.
> 
> > +	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
> 
> > +	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 "---";
> > +	}
> > +}
Since link information is given as {device,port}, not sure this is needed on how ARP is done.
I think we should rely to device,port information and skip this in NL.

RE: [PATCH RFC iproute-next 2/5] rdma: Add CM_ID resource tracking information

[Steve Wise]

> 
> On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> > 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  show
> cm_id
> > 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 0
> > link cxgb4_0/2 lqpn 1048 qp-type RC state CONNECT ps TCP dev-type ETH
> transport-type IWARP pid 30503 comm rping src-addr 172.16.2.1 src-port
> 7174 dst-addr 172.16.2.1 dst-port 38246
> > link cxgb4_0/2 lqpn 1040 qp-type RC state CONNECT ps TCP dev-type ETH
> transport-type IWARP pid 30498 comm rping src-addr 172.16.2.1 src-port
> 38246 dst-addr 172.16.2.1 dst-port 7174
> > 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 0
> > link mlx4_0/1 lqpn 539 qp-type RC state CONNECT ps TCP dev-type ETH
> transport-type IB pid 30494 comm rping src-addr 172.16.99.1 src-port 7174
> dst-addr 172.16.99.1 dst-port 43670
> > link mlx4_0/1 lqpn 538 qp-type RC state CONNECT ps TCP dev-type ETH
> transport-type IB pid 30492 comm rping src-addr 172.16.99.1 src-port 43670
> dst-addr 172.16.99.1 dst-port 7174
> >
> > # rdma resource show cm_id dst-port 7174
> > link cxgb4_0/2 lqpn 1040 qp-type RC state CONNECT ps TCP dev-type ETH
> transport-type IWARP pid 30498 comm rping src-addr 172.16.2.1 src-port
> 38246 dst-addr 172.16.2.1 dst-port 7174
> > link mlx4_0/1 lqpn 538 qp-type RC state CONNECT ps TCP dev-type ETH
> transport-type IB pid 30492 comm rping src-addr 172.16.99.1 src-port 43670
> dst-addr 172.16.99.1 dst-port 7174
> >
> > Signed-off-by: Steve Wise <swise@opengridcomputing.com>
> > ---
> >  rdma/rdma.h  |   1 +
> >  rdma/res.c   | 312
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
> >  rdma/utils.c |  12 +++
> >  3 files changed, 321 insertions(+), 4 deletions(-)
> 
> Thanks, for doing it.
> 
> >
> > 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-NAME
> FILTER-VALUE]\n");
> > +	pr_out("          resource show cm_id link [DEV/PORT]\n");
> > +	pr_out("          resource show cm_id link [DEV/PORT] [FILTER-NAME
> FILTER-VALUE]\n");
> >  	return 0;
> >  }
> >
> > @@ -431,6 +433,278 @@ static int res_qp_parse_cb(const struct nlmsghdr
> *nlh, void *data)
> >  	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",
> > +						      "ADDR_RESOLVED",
> "ROUTE_QUERY", "ROUTE_RESOLVED",
> > +						      "CONNECT",
> "DISCONNECT",
> > +						      "ADDR_BOUND",
> "LISTEN", "DEVICE_REMOVAL", "DESTROYING" };
> > +
> > +	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.
> 
> > +	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>.

> 
> > +	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" };
> > +
> 
> 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.

> > +	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));
> > +		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));
> > +		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));
> > +		return;
> > +	}
> > +	pr_out("transport-type %s ",
> cm_id_transport_type_to_str(transport_type));
> > +}
> > +
> > +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)
> > +{
> > +	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);
> > +		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).

> 
> > +		}
> > +
> > +		if (nla_line[RDMA_NLDEV_ATTR_PORT_INDEX])
> > +			port =
> mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_PORT_INDEX]);
> > +
> > +		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]);
> > +		if (rd_check_is_filtered(rd, "lqpn", lqpn))
> > +			continue;
> > +
> > +		ps =
> mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PS]);
> > +		if (rd_check_is_string_filtered(rd, "ps",
cm_id_ps_to_str(ps)))
> > +			continue;
> > +
> > +		type =
> mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_TYPE]);
> > +		if (rd_check_is_string_filtered(rd, "qp-type",
> qp_types_to_str(type)))
> > +			continue;
> > +
> > +		state =
> mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_STATE]);
> > +		if (rd_check_is_string_filtered(rd, "state",
> cm_id_state_to_str(state)))
> > +			continue;
> > +
> > +		dev_type =
> mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_DEV_TYPE]);
> > +		if (rd_check_is_string_filtered(rd, "dev-type",
> cm_id_dev_type_to_str(dev_type)))
> > +			continue;
> > +
> > +		transport_type =
> mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_TRANSPORT_TYPE]);
> > +		if (rd_check_is_string_filtered(rd, "transport-type",
> cm_id_transport_type_to_str(transport_type)))
> > +			continue;
> > +
> > +		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
> > +			pid =
> mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
> > +			comm = get_task_name(pid);
> > +		}
> > +		if (rd_check_is_filtered(rd, "pid", pid))
> 
> free(comm) here
> 

Oops!  Thanks!

> > +			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]);
> > +		}
> > +
> > +		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]);
> > +			if (!inet_ntop(AF_INET, src_addr, src_addr_str,
> INET6_ADDRSTRLEN))
> > +				return -EINVAL;
> > +			dst_addr =
> mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV4_DADDR]);
> > +			if (!inet_ntop(AF_INET, dst_addr, dst_addr_str,
> INET6_ADDRSTRLEN))
> > +				return -EINVAL;
> > +		} else {
> > +			if (!nla_line[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]
> ||
> > +			    !nla_line[RDMA_NLDEV_ATTR_RES_IPV6_DADDR])
> > +				return -EINVAL;
> > +			src_addr =
> mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]);
> > +			if (!inet_ntop(AF_INET6, src_addr, src_addr_str,
> INET6_ADDRSTRLEN))
> > +				return -EINVAL;
> > +			dst_addr =
> mnl_attr_get_payload(nla_line[RDMA_NLDEV_ATTR_RES_IPV6_DADDR]);
> > +			if (!inet_ntop(AF_INET6, dst_addr, dst_addr_str,
> INET6_ADDRSTRLEN))
> > +				return -EINVAL;
> > +		}
> > +		if (rd_check_is_string_filtered(rd, "src-addr",
src_addr_str))
> > +			continue;
> > +		if (rd_check_is_string_filtered(rd, "dst-addr",
dst_addr_str))
> > +			continue;
> > +
> > +		src_port =
> mnl_attr_get_u16(nla_line[RDMA_NLDEV_ATTR_RES_IP_SPORT]);
> > +		dst_port =
> mnl_attr_get_u16(nla_line[RDMA_NLDEV_ATTR_RES_IP_DPORT]);
> > +		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.

> 
> > +		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.


> > +
> > +		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",
> >  						   .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.

> 
> > @@ -455,11 +729,41 @@ filters
> qp_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
> >
> >  RES_FUNC(res_qp,	RDMA_NLDEV_CMD_RES_QP_GET, qp_valid_filters,
> false);
> >
> > +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",
> > +						   .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,
> > +	 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_type
> nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
> >  	[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,
> > +	[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,
> > +	[RDMA_NLDEV_ATTR_RES_IPV4_DADDR]	=
> MNL_TYPE_UNSPEC,
> > +	[RDMA_NLDEV_ATTR_RES_IPV6_SADDR]	=
> MNL_TYPE_UNSPEC,
> > +	[RDMA_NLDEV_ATTR_RES_IPV6_DADDR]	=
> MNL_TYPE_UNSPEC,
> > +	[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
> >

[PATCH RFC iproute-next 4/5] rdma: Add MR resource tracking information

[Steve Wise]

Sample output:

# rdma resource show mr
link cxgb4_0/- rkey 0x5e0e lkey 0x5e0e iova 0x7f9b60001c80 mrlen 64 pgsize 4096 pid 30503 comm rping
link cxgb4_0/- rkey 0x510d lkey 0x510d iova 0x17c7ee0 mrlen 16 pgsize 4096 pid 30503 comm rping
link cxgb4_0/- rkey 0x4a0c lkey 0x4a0c iova 0x17c7e38 mrlen 16 pgsize 4096 pid 30503 comm rping
link cxgb4_0/- rkey 0x480b lkey 0x480b iova 0x224b3f0 mrlen 64 pgsize 4096 pid 30498 comm rping
link cxgb4_0/- rkey 0x460a lkey 0x460a iova 0x224b350 mrlen 64 pgsize 4096 pid 30498 comm rping
link cxgb4_0/- rkey 0x4509 lkey 0x4509 iova 0x2245890 mrlen 16 pgsize 4096 pid 30498 comm rping
link cxgb4_0/- rkey 0x4208 lkey 0x4208 iova 0x22457e8 mrlen 16 pgsize 4096 pid 30498 comm rping
link mlx4_0/- rkey 0x38010700 lkey 0x38010700 iova 0x7f9b5c002f90 mrlen 64 pgsize 4096 pid 30494 comm rping
link mlx4_0/- rkey 0x38010600 lkey 0x38010600 iova 0x17c6c80 mrlen 16 pgsize 4096 pid 30494 comm rping
link mlx4_0/- rkey 0x38010500 lkey 0x38010500 iova 0x17c6bd8 mrlen 16 pgsize 4096 pid 30494 comm rping
link mlx4_0/- rkey 0x38010400 lkey 0x38010400 iova 0x1b68430 mrlen 64 pgsize 4096 pid 30489 comm rping
link mlx4_0/- rkey 0x38010300 lkey 0x38010300 iova 0x1b683a0 mrlen 64 pgsize 4096 pid 30489 comm rping
link mlx4_0/- rkey 0x38010200 lkey 0x38010200 iova 0x1b62890 mrlen 16 pgsize 4096 pid 30489 comm rping
link mlx4_0/- rkey 0x38010100 lkey 0x38010100 iova 0x1b627e8 mrlen 16 pgsize 4096 pid 30489 comm rping

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
---
 include/json_writer.h |   2 +
 lib/json_writer.c     |  11 +++++
 rdma/res.c            | 126 ++++++++++++++++++++++++++++++++++++++++++++++++++
 rdma/utils.c          |   7 +++
 4 files changed, 146 insertions(+)

diff --git a/include/json_writer.h b/include/json_writer.h
index 1516aaf..34f2ccc 100644
--- a/include/json_writer.h
+++ b/include/json_writer.h
@@ -39,6 +39,7 @@ void jsonw_bool(json_writer_t *self, bool value);
 void jsonw_float(json_writer_t *self, double number);
 void jsonw_float_fmt(json_writer_t *self, const char *fmt, double num);
 void jsonw_uint(json_writer_t *self, uint64_t number);
+void jsonw_xint(json_writer_t *self, uint64_t number);
 void jsonw_hu(json_writer_t *self, unsigned short number);
 void jsonw_int(json_writer_t *self, int64_t number);
 void jsonw_null(json_writer_t *self);
@@ -49,6 +50,7 @@ void jsonw_string_field(json_writer_t *self, const char *prop, const char *val);
 void jsonw_bool_field(json_writer_t *self, const char *prop, bool value);
 void jsonw_float_field(json_writer_t *self, const char *prop, double num);
 void jsonw_uint_field(json_writer_t *self, const char *prop, uint64_t num);
+void jsonw_xint_field(json_writer_t *self, const char *prop, uint64_t num);
 void jsonw_hu_field(json_writer_t *self, const char *prop, unsigned short num);
 void jsonw_int_field(json_writer_t *self, const char *prop, int64_t num);
 void jsonw_null_field(json_writer_t *self, const char *prop);
diff --git a/lib/json_writer.c b/lib/json_writer.c
index f3eeaf7..6d73a1b 100644
--- a/lib/json_writer.c
+++ b/lib/json_writer.c
@@ -224,6 +224,11 @@ void jsonw_uint(json_writer_t *self, uint64_t num)
 	jsonw_printf(self, "%"PRIu64, num);
 }
 
+void jsonw_xint(json_writer_t *self, uint64_t num)
+{
+	jsonw_printf(self, "%"PRIx64, num);
+}
+
 void jsonw_lluint(json_writer_t *self, unsigned long long int num)
 {
 	jsonw_printf(self, "%llu", num);
@@ -268,6 +273,12 @@ void jsonw_uint_field(json_writer_t *self, const char *prop, uint64_t num)
 	jsonw_uint(self, num);
 }
 
+void jsonw_xint_field(json_writer_t *self, const char *prop, uint64_t num)
+{
+	jsonw_name(self, prop);
+	jsonw_xint(self, num);
+}
+
 void jsonw_hu_field(json_writer_t *self, const char *prop, unsigned short num)
 {
 	jsonw_name(self, prop);
diff --git a/rdma/res.c b/rdma/res.c
index 27c1efd..2b67d25 100644
--- a/rdma/res.c
+++ b/rdma/res.c
@@ -819,6 +819,119 @@ static int res_cq_parse_cb(const struct nlmsghdr *nlh, void *data)
 	return MNL_CB_OK;
 }
 
+static void print_key(struct rd *rd, const char *name, uint32_t val)
+{
+	if (rd->json_output)
+		jsonw_xint_field(rd->jw, name, val);
+	else
+		pr_out("%s 0x%x ", name, val);
+}
+
+static void print_iova(struct rd *rd, uint64_t val)
+{
+	if (rd->json_output)
+		jsonw_xint_field(rd->jw, "iova", val);
+	else
+		pr_out("iova 0x%" PRIx64 " ", val);
+}
+
+static void print_mrlen(struct rd *rd, uint64_t val)
+{
+	if (rd->json_output)
+		jsonw_uint_field(rd->jw, "mrlen", val);
+	else
+		pr_out("mrlen %" PRIu64 " ", val);
+}
+
+static void print_pgsize(struct rd *rd, uint32_t val)
+{
+	if (rd->json_output)
+		jsonw_uint_field(rd->jw, "pgsize", val);
+	else
+		pr_out("pgsize %u ", val);
+}
+
+static int res_mr_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;
+	uint32_t 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_MR])
+		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_MR];
+
+	mnl_attr_for_each_nested(nla_entry, nla_table) {
+		struct nlattr *nla_line[RDMA_NLDEV_ATTR_MAX] = {};
+		uint32_t rkey, lkey, pgsize;
+		uint64_t iova, mrlen;
+		char *comm = NULL;
+		uint32_t pid = 0;
+		int err;
+
+		err = mnl_attr_parse_nested(nla_entry, rd_attr_cb, nla_line);
+		if (err != MNL_CB_OK)
+			return MNL_CB_ERROR;
+
+		if (!nla_line[RDMA_NLDEV_ATTR_RES_RKEY] ||
+		    !nla_line[RDMA_NLDEV_ATTR_RES_LKEY] ||
+		    !nla_line[RDMA_NLDEV_ATTR_RES_IOVA] ||
+		    !nla_line[RDMA_NLDEV_ATTR_RES_MRLEN] ||
+		    !nla_line[RDMA_NLDEV_ATTR_RES_PGSIZE] ||
+		    (!nla_line[RDMA_NLDEV_ATTR_RES_PID] &&
+		     !nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME])) {
+			return MNL_CB_ERROR;
+		}
+
+		rkey = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_RKEY]);
+		lkey = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_LKEY]);
+		iova = mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_IOVA]);
+		mrlen = mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_MRLEN]);
+		pgsize = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PGSIZE]);
+
+		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
+			pid = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
+			comm = get_task_name(pid);
+		}
+
+		if (rd_check_is_filtered(rd, "pid", pid))
+			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]);
+
+		if (rd->json_output)
+			jsonw_start_array(rd->jw);
+
+		print_link(rd, idx, name, 0, nla_line);
+		print_key(rd, "rkey", rkey);
+		print_key(rd, "lkey", lkey);
+		print_iova(rd, iova);
+		print_mrlen(rd, mrlen);
+		print_pgsize(rd, pgsize);
+		print_pid(rd, pid);
+		print_comm(rd, comm, nla_line);
+
+		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
@@ -880,6 +993,18 @@ filters cq_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
 
 RES_FUNC(res_cq,	RDMA_NLDEV_CMD_RES_CQ_GET, cq_valid_filters, true);
 
+static const struct
+filters mr_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
+						   .is_number = false },
+						   { .name = "rkey",
+						   .is_number = true },
+						   { .name = "lkey",
+						   .is_number = true },
+						   { .name = "pid",
+						   .is_number = true }};
+
+RES_FUNC(res_mr,	RDMA_NLDEV_CMD_RES_MR_GET, mr_valid_filters, true);
+
 static int res_show(struct rd *rd)
 {
 	const struct rd_cmd cmds[] = {
@@ -887,6 +1012,7 @@ static int res_show(struct rd *rd)
 		{ "qp",		res_qp		},
 		{ "cm_id",	res_cm_id	},
 		{ "cq",		res_cq		},
+		{ "mr",		res_mr		},
 		{ 0 }
 	};
 
diff --git a/rdma/utils.c b/rdma/utils.c
index 11b34fe..34b195a 100644
--- a/rdma/utils.c
+++ b/rdma/utils.c
@@ -392,6 +392,13 @@ static const enum mnl_attr_data_type nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
 	[RDMA_NLDEV_ATTR_RES_CQE] = MNL_TYPE_U32,
 	[RDMA_NLDEV_ATTR_RES_USECNT] = MNL_TYPE_U64,
 	[RDMA_NLDEV_ATTR_RES_POLL_CTX] = MNL_TYPE_U8,
+	[RDMA_NLDEV_ATTR_RES_MR] = MNL_TYPE_NESTED,
+	[RDMA_NLDEV_ATTR_RES_MR_ENTRY] = MNL_TYPE_NESTED,
+	[RDMA_NLDEV_ATTR_RES_RKEY] = MNL_TYPE_U32,
+	[RDMA_NLDEV_ATTR_RES_LKEY] = MNL_TYPE_U32,
+	[RDMA_NLDEV_ATTR_RES_IOVA] = MNL_TYPE_U64,
+	[RDMA_NLDEV_ATTR_RES_MRLEN] = MNL_TYPE_U64,
+	[RDMA_NLDEV_ATTR_RES_PGSIZE] = MNL_TYPE_U32,
 };
 
 int rd_attr_cb(const struct nlattr *attr, void *data)
-- 
1.8.3.1

Re: [PATCH RFC iproute-next 4/5] rdma: Add MR resource tracking information

[Leon Romanovsky]

[-- Attachment #1: Type: text/plain, Size: 9316 bytes --]

On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> Sample output:
>
> # rdma resource show mr
> link cxgb4_0/- rkey 0x5e0e lkey 0x5e0e iova 0x7f9b60001c80 mrlen 64 pgsize 4096 pid 30503 comm rping
> link cxgb4_0/- rkey 0x510d lkey 0x510d iova 0x17c7ee0 mrlen 16 pgsize 4096 pid 30503 comm rping
> link cxgb4_0/- rkey 0x4a0c lkey 0x4a0c iova 0x17c7e38 mrlen 16 pgsize 4096 pid 30503 comm rping
> link cxgb4_0/- rkey 0x480b lkey 0x480b iova 0x224b3f0 mrlen 64 pgsize 4096 pid 30498 comm rping
> link cxgb4_0/- rkey 0x460a lkey 0x460a iova 0x224b350 mrlen 64 pgsize 4096 pid 30498 comm rping
> link cxgb4_0/- rkey 0x4509 lkey 0x4509 iova 0x2245890 mrlen 16 pgsize 4096 pid 30498 comm rping
> link cxgb4_0/- rkey 0x4208 lkey 0x4208 iova 0x22457e8 mrlen 16 pgsize 4096 pid 30498 comm rping
> link mlx4_0/- rkey 0x38010700 lkey 0x38010700 iova 0x7f9b5c002f90 mrlen 64 pgsize 4096 pid 30494 comm rping
> link mlx4_0/- rkey 0x38010600 lkey 0x38010600 iova 0x17c6c80 mrlen 16 pgsize 4096 pid 30494 comm rping
> link mlx4_0/- rkey 0x38010500 lkey 0x38010500 iova 0x17c6bd8 mrlen 16 pgsize 4096 pid 30494 comm rping
> link mlx4_0/- rkey 0x38010400 lkey 0x38010400 iova 0x1b68430 mrlen 64 pgsize 4096 pid 30489 comm rping
> link mlx4_0/- rkey 0x38010300 lkey 0x38010300 iova 0x1b683a0 mrlen 64 pgsize 4096 pid 30489 comm rping
> link mlx4_0/- rkey 0x38010200 lkey 0x38010200 iova 0x1b62890 mrlen 16 pgsize 4096 pid 30489 comm rping
> link mlx4_0/- rkey 0x38010100 lkey 0x38010100 iova 0x1b627e8 mrlen 16 pgsize 4096 pid 30489 comm rping
>
> Signed-off-by: Steve Wise <swise@opengridcomputing.com>
> ---
>  include/json_writer.h |   2 +
>  lib/json_writer.c     |  11 +++++
>  rdma/res.c            | 126 ++++++++++++++++++++++++++++++++++++++++++++++++++
>  rdma/utils.c          |   7 +++
>  4 files changed, 146 insertions(+)
>
> diff --git a/include/json_writer.h b/include/json_writer.h
> index 1516aaf..34f2ccc 100644
> --- a/include/json_writer.h
> +++ b/include/json_writer.h
> @@ -39,6 +39,7 @@ void jsonw_bool(json_writer_t *self, bool value);
>  void jsonw_float(json_writer_t *self, double number);
>  void jsonw_float_fmt(json_writer_t *self, const char *fmt, double num);
>  void jsonw_uint(json_writer_t *self, uint64_t number);
> +void jsonw_xint(json_writer_t *self, uint64_t number);
>  void jsonw_hu(json_writer_t *self, unsigned short number);
>  void jsonw_int(json_writer_t *self, int64_t number);
>  void jsonw_null(json_writer_t *self);
> @@ -49,6 +50,7 @@ void jsonw_string_field(json_writer_t *self, const char *prop, const char *val);
>  void jsonw_bool_field(json_writer_t *self, const char *prop, bool value);
>  void jsonw_float_field(json_writer_t *self, const char *prop, double num);
>  void jsonw_uint_field(json_writer_t *self, const char *prop, uint64_t num);
> +void jsonw_xint_field(json_writer_t *self, const char *prop, uint64_t num);
>  void jsonw_hu_field(json_writer_t *self, const char *prop, unsigned short num);
>  void jsonw_int_field(json_writer_t *self, const char *prop, int64_t num);
>  void jsonw_null_field(json_writer_t *self, const char *prop);
> diff --git a/lib/json_writer.c b/lib/json_writer.c
> index f3eeaf7..6d73a1b 100644
> --- a/lib/json_writer.c
> +++ b/lib/json_writer.c
> @@ -224,6 +224,11 @@ void jsonw_uint(json_writer_t *self, uint64_t num)
>  	jsonw_printf(self, "%"PRIu64, num);
>  }
>
> +void jsonw_xint(json_writer_t *self, uint64_t num)
> +{
> +	jsonw_printf(self, "%"PRIx64, num);
> +}
> +
>  void jsonw_lluint(json_writer_t *self, unsigned long long int num)
>  {
>  	jsonw_printf(self, "%llu", num);
> @@ -268,6 +273,12 @@ void jsonw_uint_field(json_writer_t *self, const char *prop, uint64_t num)
>  	jsonw_uint(self, num);
>  }
>
> +void jsonw_xint_field(json_writer_t *self, const char *prop, uint64_t num)
> +{
> +	jsonw_name(self, prop);
> +	jsonw_xint(self, num);
> +}
> +
>  void jsonw_hu_field(json_writer_t *self, const char *prop, unsigned short num)
>  {
>  	jsonw_name(self, prop);
> diff --git a/rdma/res.c b/rdma/res.c
> index 27c1efd..2b67d25 100644
> --- a/rdma/res.c
> +++ b/rdma/res.c
> @@ -819,6 +819,119 @@ static int res_cq_parse_cb(const struct nlmsghdr *nlh, void *data)
>  	return MNL_CB_OK;
>  }
>
> +static void print_key(struct rd *rd, const char *name, uint32_t val)
> +{
> +	if (rd->json_output)
> +		jsonw_xint_field(rd->jw, name, val);
> +	else
> +		pr_out("%s 0x%x ", name, val);
> +}
> +
> +static void print_iova(struct rd *rd, uint64_t val)
> +{
> +	if (rd->json_output)
> +		jsonw_xint_field(rd->jw, "iova", val);
> +	else
> +		pr_out("iova 0x%" PRIx64 " ", val);
> +}
> +
> +static void print_mrlen(struct rd *rd, uint64_t val)
> +{
> +	if (rd->json_output)
> +		jsonw_uint_field(rd->jw, "mrlen", val);
> +	else
> +		pr_out("mrlen %" PRIu64 " ", val);
> +}
> +
> +static void print_pgsize(struct rd *rd, uint32_t val)
> +{
> +	if (rd->json_output)
> +		jsonw_uint_field(rd->jw, "pgsize", val);
> +	else
> +		pr_out("pgsize %u ", val);
> +}
> +
> +static int res_mr_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;
> +	uint32_t 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_MR])
> +		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_MR];
> +
> +	mnl_attr_for_each_nested(nla_entry, nla_table) {
> +		struct nlattr *nla_line[RDMA_NLDEV_ATTR_MAX] = {};
> +		uint32_t rkey, lkey, pgsize;
> +		uint64_t iova, mrlen;
> +		char *comm = NULL;
> +		uint32_t pid = 0;
> +		int err;
> +
> +		err = mnl_attr_parse_nested(nla_entry, rd_attr_cb, nla_line);
> +		if (err != MNL_CB_OK)
> +			return MNL_CB_ERROR;
> +
> +		if (!nla_line[RDMA_NLDEV_ATTR_RES_RKEY] ||
> +		    !nla_line[RDMA_NLDEV_ATTR_RES_LKEY] ||
> +		    !nla_line[RDMA_NLDEV_ATTR_RES_IOVA] ||
> +		    !nla_line[RDMA_NLDEV_ATTR_RES_MRLEN] ||
> +		    !nla_line[RDMA_NLDEV_ATTR_RES_PGSIZE] ||
> +		    (!nla_line[RDMA_NLDEV_ATTR_RES_PID] &&
> +		     !nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME])) {
> +			return MNL_CB_ERROR;
> +		}

I'm not sure that all this ATTR are needed.

> +
> +		rkey = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_RKEY]);
> +		lkey = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_LKEY]);
> +		iova = mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_IOVA]);
> +		mrlen = mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_MRLEN]);
> +		pgsize = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PGSIZE]);
> +
> +		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
> +			pid = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
> +			comm = get_task_name(pid);
> +		}
> +
> +		if (rd_check_is_filtered(rd, "pid", pid))

free(comm);

> +			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]);
> +
> +		if (rd->json_output)
> +			jsonw_start_array(rd->jw);
> +
> +		print_link(rd, idx, name, 0, nla_line);
> +		print_key(rd, "rkey", rkey);
> +		print_key(rd, "lkey", lkey);
> +		print_iova(rd, iova);
> +		print_mrlen(rd, mrlen);
> +		print_pgsize(rd, pgsize);
> +		print_pid(rd, pid);
> +		print_comm(rd, comm, nla_line);
> +
> +		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
> @@ -880,6 +993,18 @@ filters cq_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
>
>  RES_FUNC(res_cq,	RDMA_NLDEV_CMD_RES_CQ_GET, cq_valid_filters, true);
>
> +static const struct
> +filters mr_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
> +						   .is_number = false },
> +						   { .name = "rkey",
> +						   .is_number = true },
> +						   { .name = "lkey",
> +						   .is_number = true },
> +						   { .name = "pid",
> +						   .is_number = true }};
> +
> +RES_FUNC(res_mr,	RDMA_NLDEV_CMD_RES_MR_GET, mr_valid_filters, true);
> +
>  static int res_show(struct rd *rd)
>  {
>  	const struct rd_cmd cmds[] = {
> @@ -887,6 +1012,7 @@ static int res_show(struct rd *rd)
>  		{ "qp",		res_qp		},
>  		{ "cm_id",	res_cm_id	},
>  		{ "cq",		res_cq		},
> +		{ "mr",		res_mr		},
>  		{ 0 }
>  	};
>
> diff --git a/rdma/utils.c b/rdma/utils.c
> index 11b34fe..34b195a 100644
> --- a/rdma/utils.c
> +++ b/rdma/utils.c
> @@ -392,6 +392,13 @@ static const enum mnl_attr_data_type nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
>  	[RDMA_NLDEV_ATTR_RES_CQE] = MNL_TYPE_U32,
>  	[RDMA_NLDEV_ATTR_RES_USECNT] = MNL_TYPE_U64,
>  	[RDMA_NLDEV_ATTR_RES_POLL_CTX] = MNL_TYPE_U8,
> +	[RDMA_NLDEV_ATTR_RES_MR] = MNL_TYPE_NESTED,
> +	[RDMA_NLDEV_ATTR_RES_MR_ENTRY] = MNL_TYPE_NESTED,
> +	[RDMA_NLDEV_ATTR_RES_RKEY] = MNL_TYPE_U32,
> +	[RDMA_NLDEV_ATTR_RES_LKEY] = MNL_TYPE_U32,
> +	[RDMA_NLDEV_ATTR_RES_IOVA] = MNL_TYPE_U64,
> +	[RDMA_NLDEV_ATTR_RES_MRLEN] = MNL_TYPE_U64,
> +	[RDMA_NLDEV_ATTR_RES_PGSIZE] = MNL_TYPE_U32,
>  };
>
>  int rd_attr_cb(const struct nlattr *attr, void *data)
> --
> 1.8.3.1
>

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

RE: [PATCH RFC iproute-next 4/5] rdma: Add MR resource tracking information

[Steve Wise]

> -----Original Message-----
> From: Leon Romanovsky [mailto:leon@kernel.org]
> Sent: Tuesday, February 20, 2018 8:12 AM
> To: Steve Wise <swise@opengridcomputing.com>
> Cc: dsahern@gmail.com; stephen@networkplumber.org;
> netdev@vger.kernel.org; linux-rdma@vger.kernel.org
> Subject: Re: [PATCH RFC iproute-next 4/5] rdma: Add MR resource tracking
> information
> 
> On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> > Sample output:
> >
> > # rdma resource show mr
> > link cxgb4_0/- rkey 0x5e0e lkey 0x5e0e iova 0x7f9b60001c80 mrlen 64
> pgsize 4096 pid 30503 comm rping
> > link cxgb4_0/- rkey 0x510d lkey 0x510d iova 0x17c7ee0 mrlen 16 pgsize
> 4096 pid 30503 comm rping
> > link cxgb4_0/- rkey 0x4a0c lkey 0x4a0c iova 0x17c7e38 mrlen 16 pgsize
4096
> pid 30503 comm rping
> > link cxgb4_0/- rkey 0x480b lkey 0x480b iova 0x224b3f0 mrlen 64 pgsize
> 4096 pid 30498 comm rping
> > link cxgb4_0/- rkey 0x460a lkey 0x460a iova 0x224b350 mrlen 64 pgsize
> 4096 pid 30498 comm rping
> > link cxgb4_0/- rkey 0x4509 lkey 0x4509 iova 0x2245890 mrlen 16 pgsize
> 4096 pid 30498 comm rping
> > link cxgb4_0/- rkey 0x4208 lkey 0x4208 iova 0x22457e8 mrlen 16 pgsize
> 4096 pid 30498 comm rping
> > link mlx4_0/- rkey 0x38010700 lkey 0x38010700 iova 0x7f9b5c002f90 mrlen
> 64 pgsize 4096 pid 30494 comm rping
> > link mlx4_0/- rkey 0x38010600 lkey 0x38010600 iova 0x17c6c80 mrlen 16
> pgsize 4096 pid 30494 comm rping
> > link mlx4_0/- rkey 0x38010500 lkey 0x38010500 iova 0x17c6bd8 mrlen 16
> pgsize 4096 pid 30494 comm rping
> > link mlx4_0/- rkey 0x38010400 lkey 0x38010400 iova 0x1b68430 mrlen 64
> pgsize 4096 pid 30489 comm rping
> > link mlx4_0/- rkey 0x38010300 lkey 0x38010300 iova 0x1b683a0 mrlen 64
> pgsize 4096 pid 30489 comm rping
> > link mlx4_0/- rkey 0x38010200 lkey 0x38010200 iova 0x1b62890 mrlen 16
> pgsize 4096 pid 30489 comm rping
> > link mlx4_0/- rkey 0x38010100 lkey 0x38010100 iova 0x1b627e8 mrlen 16
> pgsize 4096 pid 30489 comm rping
> >
> > Signed-off-by: Steve Wise <swise@opengridcomputing.com>
> > ---
> >  include/json_writer.h |   2 +
> >  lib/json_writer.c     |  11 +++++
> >  rdma/res.c            | 126
> ++++++++++++++++++++++++++++++++++++++++++++++++++
> >  rdma/utils.c          |   7 +++
> >  4 files changed, 146 insertions(+)
> >
> > diff --git a/include/json_writer.h b/include/json_writer.h
> > index 1516aaf..34f2ccc 100644
> > --- a/include/json_writer.h
> > +++ b/include/json_writer.h
> > @@ -39,6 +39,7 @@ void jsonw_bool(json_writer_t *self, bool value);
> >  void jsonw_float(json_writer_t *self, double number);
> >  void jsonw_float_fmt(json_writer_t *self, const char *fmt, double num);
> >  void jsonw_uint(json_writer_t *self, uint64_t number);
> > +void jsonw_xint(json_writer_t *self, uint64_t number);
> >  void jsonw_hu(json_writer_t *self, unsigned short number);
> >  void jsonw_int(json_writer_t *self, int64_t number);
> >  void jsonw_null(json_writer_t *self);
> > @@ -49,6 +50,7 @@ void jsonw_string_field(json_writer_t *self, const
char
> *prop, const char *val);
> >  void jsonw_bool_field(json_writer_t *self, const char *prop, bool
value);
> >  void jsonw_float_field(json_writer_t *self, const char *prop, double
num);
> >  void jsonw_uint_field(json_writer_t *self, const char *prop, uint64_t
> num);
> > +void jsonw_xint_field(json_writer_t *self, const char *prop, uint64_t
> num);
> >  void jsonw_hu_field(json_writer_t *self, const char *prop, unsigned
short
> num);
> >  void jsonw_int_field(json_writer_t *self, const char *prop, int64_t
num);
> >  void jsonw_null_field(json_writer_t *self, const char *prop);
> > diff --git a/lib/json_writer.c b/lib/json_writer.c
> > index f3eeaf7..6d73a1b 100644
> > --- a/lib/json_writer.c
> > +++ b/lib/json_writer.c
> > @@ -224,6 +224,11 @@ void jsonw_uint(json_writer_t *self, uint64_t
> num)
> >  	jsonw_printf(self, "%"PRIu64, num);
> >  }
> >
> > +void jsonw_xint(json_writer_t *self, uint64_t num)
> > +{
> > +	jsonw_printf(self, "%"PRIx64, num);
> > +}
> > +
> >  void jsonw_lluint(json_writer_t *self, unsigned long long int num)
> >  {
> >  	jsonw_printf(self, "%llu", num);
> > @@ -268,6 +273,12 @@ void jsonw_uint_field(json_writer_t *self, const
> char *prop, uint64_t num)
> >  	jsonw_uint(self, num);
> >  }
> >
> > +void jsonw_xint_field(json_writer_t *self, const char *prop, uint64_t
num)
> > +{
> > +	jsonw_name(self, prop);
> > +	jsonw_xint(self, num);
> > +}
> > +
> >  void jsonw_hu_field(json_writer_t *self, const char *prop, unsigned
short
> num)
> >  {
> >  	jsonw_name(self, prop);
> > diff --git a/rdma/res.c b/rdma/res.c
> > index 27c1efd..2b67d25 100644
> > --- a/rdma/res.c
> > +++ b/rdma/res.c
> > @@ -819,6 +819,119 @@ static int res_cq_parse_cb(const struct nlmsghdr
> *nlh, void *data)
> >  	return MNL_CB_OK;
> >  }
> >
> > +static void print_key(struct rd *rd, const char *name, uint32_t val)
> > +{
> > +	if (rd->json_output)
> > +		jsonw_xint_field(rd->jw, name, val);
> > +	else
> > +		pr_out("%s 0x%x ", name, val);
> > +}
> > +
> > +static void print_iova(struct rd *rd, uint64_t val)
> > +{
> > +	if (rd->json_output)
> > +		jsonw_xint_field(rd->jw, "iova", val);
> > +	else
> > +		pr_out("iova 0x%" PRIx64 " ", val);
> > +}
> > +
> > +static void print_mrlen(struct rd *rd, uint64_t val)
> > +{
> > +	if (rd->json_output)
> > +		jsonw_uint_field(rd->jw, "mrlen", val);
> > +	else
> > +		pr_out("mrlen %" PRIu64 " ", val);
> > +}
> > +
> > +static void print_pgsize(struct rd *rd, uint32_t val)
> > +{
> > +	if (rd->json_output)
> > +		jsonw_uint_field(rd->jw, "pgsize", val);
> > +	else
> > +		pr_out("pgsize %u ", val);
> > +}
> > +
> > +static int res_mr_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;
> > +	uint32_t 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_MR])
> > +		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_MR];
> > +
> > +	mnl_attr_for_each_nested(nla_entry, nla_table) {
> > +		struct nlattr *nla_line[RDMA_NLDEV_ATTR_MAX] = {};
> > +		uint32_t rkey, lkey, pgsize;
> > +		uint64_t iova, mrlen;
> > +		char *comm = NULL;
> > +		uint32_t pid = 0;
> > +		int err;
> > +
> > +		err = mnl_attr_parse_nested(nla_entry, rd_attr_cb,
nla_line);
> > +		if (err != MNL_CB_OK)
> > +			return MNL_CB_ERROR;
> > +
> > +		if (!nla_line[RDMA_NLDEV_ATTR_RES_RKEY] ||
> > +		    !nla_line[RDMA_NLDEV_ATTR_RES_LKEY] ||
> > +		    !nla_line[RDMA_NLDEV_ATTR_RES_IOVA] ||
> > +		    !nla_line[RDMA_NLDEV_ATTR_RES_MRLEN] ||
> > +		    !nla_line[RDMA_NLDEV_ATTR_RES_PGSIZE] ||
> > +		    (!nla_line[RDMA_NLDEV_ATTR_RES_PID] &&
> > +		     !nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME])) {
> > +			return MNL_CB_ERROR;
> > +		}
> 
> I'm not sure that all this ATTR are needed.
> 

Yea you're right.  I'll revisit all these and decide which minimal ones are
really required.

> > +
> > +		rkey =
> mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_RKEY]);
> > +		lkey =
> mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_LKEY]);
> > +		iova =
> mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_IOVA]);
> > +		mrlen =
> mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_MRLEN]);
> > +		pgsize =
> mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PGSIZE]);
> > +
> > +		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
> > +			pid =
> mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
> > +			comm = get_task_name(pid);
> > +		}
> > +
> > +		if (rd_check_is_filtered(rd, "pid", pid))
> 
> free(comm);
>

yup
 
> > +			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]);
> > +
> > +		if (rd->json_output)
> > +			jsonw_start_array(rd->jw);
> > +
> > +		print_link(rd, idx, name, 0, nla_line);
> > +		print_key(rd, "rkey", rkey);
> > +		print_key(rd, "lkey", lkey);
> > +		print_iova(rd, iova);
> > +		print_mrlen(rd, mrlen);
> > +		print_pgsize(rd, pgsize);
> > +		print_pid(rd, pid);
> > +		print_comm(rd, comm, nla_line);
> > +
> > +		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
> > @@ -880,6 +993,18 @@ filters
> cq_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
> >
> >  RES_FUNC(res_cq,	RDMA_NLDEV_CMD_RES_CQ_GET, cq_valid_filters,
> true);
> >
> > +static const struct
> > +filters mr_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
> > +						   .is_number = false },
> > +						   { .name = "rkey",
> > +						   .is_number = true },
> > +						   { .name = "lkey",
> > +						   .is_number = true },
> > +						   { .name = "pid",
> > +						   .is_number = true }};
> > +
> > +RES_FUNC(res_mr,	RDMA_NLDEV_CMD_RES_MR_GET, mr_valid_filters,
> true);
> > +
> >  static int res_show(struct rd *rd)
> >  {
> >  	const struct rd_cmd cmds[] = {
> > @@ -887,6 +1012,7 @@ static int res_show(struct rd *rd)
> >  		{ "qp",		res_qp		},
> >  		{ "cm_id",	res_cm_id	},
> >  		{ "cq",		res_cq		},
> > +		{ "mr",		res_mr		},
> >  		{ 0 }
> >  	};
> >
> > diff --git a/rdma/utils.c b/rdma/utils.c
> > index 11b34fe..34b195a 100644
> > --- a/rdma/utils.c
> > +++ b/rdma/utils.c
> > @@ -392,6 +392,13 @@ static const enum mnl_attr_data_type
> nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
> >  	[RDMA_NLDEV_ATTR_RES_CQE] = MNL_TYPE_U32,
> >  	[RDMA_NLDEV_ATTR_RES_USECNT] = MNL_TYPE_U64,
> >  	[RDMA_NLDEV_ATTR_RES_POLL_CTX] = MNL_TYPE_U8,
> > +	[RDMA_NLDEV_ATTR_RES_MR] = MNL_TYPE_NESTED,
> > +	[RDMA_NLDEV_ATTR_RES_MR_ENTRY] = MNL_TYPE_NESTED,
> > +	[RDMA_NLDEV_ATTR_RES_RKEY] = MNL_TYPE_U32,
> > +	[RDMA_NLDEV_ATTR_RES_LKEY] = MNL_TYPE_U32,
> > +	[RDMA_NLDEV_ATTR_RES_IOVA] = MNL_TYPE_U64,
> > +	[RDMA_NLDEV_ATTR_RES_MRLEN] = MNL_TYPE_U64,
> > +	[RDMA_NLDEV_ATTR_RES_PGSIZE] = MNL_TYPE_U32,
> >  };
> >
> >  int rd_attr_cb(const struct nlattr *attr, void *data)
> > --
> > 1.8.3.1
> >

[PATCH RFC iproute-next 3/5] rdma: Add CQ resource tracking information

[Steve Wise]

Sample output:

# rdma resource show cq
link cxgb4_0/- cqe 46 usecnt 2 pid 30503 comm rping
link cxgb4_0/- cqe 46 usecnt 2 pid 30498 comm rping
link mlx4_0/- cqe 63 usecnt 2 pid 30494 comm rping
link mlx4_0/- cqe 63 usecnt 2 pid 30489 comm rping
link mlx4_0/- cqe 1023 usecnt 2 poll_ctx WORKQUEUE pid 0 comm [ib_core]

# rdma resource show cq pid 30489
link mlx4_0/- cqe 63 usecnt 2 pid 30489 comm rping

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
---
 rdma/res.c   | 123 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 rdma/utils.c |   5 +++
 2 files changed, 128 insertions(+)

diff --git a/rdma/res.c b/rdma/res.c
index beae7dc..27c1efd 100644
--- a/rdma/res.c
+++ b/rdma/res.c
@@ -21,6 +21,8 @@ static int res_help(struct rd *rd)
 	pr_out("          resource show qp link [DEV/PORT] [FILTER-NAME FILTER-VALUE]\n");
 	pr_out("          resource show cm_id link [DEV/PORT]\n");
 	pr_out("          resource show cm_id link [DEV/PORT] [FILTER-NAME FILTER-VALUE]\n");
+	pr_out("          resource show cq link [DEV/PORT]\n");
+	pr_out("          resource show cq link [DEV/PORT] [FILTER-NAME FILTER-VALUE]\n");
 	return 0;
 }
 
@@ -705,6 +707,118 @@ static int res_cm_id_parse_cb(const struct nlmsghdr *nlh, void *data)
 	return MNL_CB_OK;
 }
 
+static void print_cqe(struct rd *rd, uint32_t val)
+{
+	if (rd->json_output)
+		jsonw_uint_field(rd->jw, "cqe", val);
+	else
+		pr_out("cqe %u ", val);
+}
+
+static void print_usecnt(struct rd *rd, uint64_t val)
+{
+	if (rd->json_output)
+		jsonw_uint_field(rd->jw, "usecnt", val);
+	else
+		pr_out("usecnt %" PRIu64 " ", val);
+}
+
+static const char *poll_ctx_to_str(uint8_t idx)
+{
+	static const char * const cm_id_states_str[] = { "DIRECT", "SOFTIRQ",
+						      "WORKQUEUE"};
+
+	if (idx < ARRAY_SIZE(cm_id_states_str))
+		return cm_id_states_str[idx];
+	return "UNKNOWN";
+}
+
+static void print_poll_ctx(struct rd *rd, uint8_t poll_ctx)
+{
+	if (rd->json_output) {
+		jsonw_string_field(rd->jw, "poll_ctx", poll_ctx_to_str(poll_ctx));
+		return;
+	}
+	pr_out("poll_ctx %s ", poll_ctx_to_str(poll_ctx));
+}
+
+static int res_cq_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;
+	uint32_t 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_CQ])
+		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_CQ];
+
+	mnl_attr_for_each_nested(nla_entry, nla_table) {
+		struct nlattr *nla_line[RDMA_NLDEV_ATTR_MAX] = {};
+		char *comm = NULL;
+		uint32_t pid = 0;
+		uint8_t poll_ctx = 0;
+		uint64_t usecnt;
+		uint32_t cqe;
+		int err;
+
+		err = mnl_attr_parse_nested(nla_entry, rd_attr_cb, nla_line);
+		if (err != MNL_CB_OK)
+			return MNL_CB_ERROR;
+
+		if (!nla_line[RDMA_NLDEV_ATTR_RES_CQE] ||
+		    !nla_line[RDMA_NLDEV_ATTR_RES_USECNT] ||
+		    (!nla_line[RDMA_NLDEV_ATTR_RES_PID] &&
+		     !nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME])) {
+			return MNL_CB_ERROR;
+		}
+
+		cqe = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_CQE]);
+		usecnt = mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_USECNT]);
+		if (nla_line[RDMA_NLDEV_ATTR_RES_POLL_CTX])
+			poll_ctx = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_POLL_CTX]);
+
+		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
+			pid = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
+			comm = get_task_name(pid);
+		}
+
+		if (rd_check_is_filtered(rd, "pid", pid))
+			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]);
+
+		if (rd->json_output)
+			jsonw_start_array(rd->jw);
+
+		print_link(rd, idx, name, 0, nla_line);
+		print_cqe(rd, cqe);
+		print_usecnt(rd, usecnt);
+		if (nla_line[RDMA_NLDEV_ATTR_RES_POLL_CTX])
+			print_poll_ctx(rd, poll_ctx);
+		print_pid(rd, pid);
+		print_comm(rd, comm, nla_line);
+
+		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
@@ -758,12 +872,21 @@ filters cm_id_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
 RES_FUNC(res_cm_id,	RDMA_NLDEV_CMD_RES_CM_ID_GET, cm_id_valid_filters,
 	 false);
 
+static const struct
+filters cq_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
+						   .is_number = false },
+						   { .name = "pid",
+						   .is_number = true }};
+
+RES_FUNC(res_cq,	RDMA_NLDEV_CMD_RES_CQ_GET, cq_valid_filters, true);
+
 static int res_show(struct rd *rd)
 {
 	const struct rd_cmd cmds[] = {
 		{ NULL,		res_no_args	},
 		{ "qp",		res_qp		},
 		{ "cm_id",	res_cm_id	},
+		{ "cq",		res_cq		},
 		{ 0 }
 	};
 
diff --git a/rdma/utils.c b/rdma/utils.c
index 906ca73..11b34fe 100644
--- a/rdma/utils.c
+++ b/rdma/utils.c
@@ -387,6 +387,11 @@ static const enum mnl_attr_data_type nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
 	[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,
+	[RDMA_NLDEV_ATTR_RES_CQ] = MNL_TYPE_NESTED,
+	[RDMA_NLDEV_ATTR_RES_CQ_ENTRY] = MNL_TYPE_NESTED,
+	[RDMA_NLDEV_ATTR_RES_CQE] = MNL_TYPE_U32,
+	[RDMA_NLDEV_ATTR_RES_USECNT] = MNL_TYPE_U64,
+	[RDMA_NLDEV_ATTR_RES_POLL_CTX] = MNL_TYPE_U8,
 };
 
 int rd_attr_cb(const struct nlattr *attr, void *data)
-- 
1.8.3.1

Re: [PATCH RFC iproute-next 3/5] rdma: Add CQ resource tracking information

[Leon Romanovsky]

[-- Attachment #1: Type: text/plain, Size: 6477 bytes --]

On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> Sample output:
>
> # rdma resource show cq
> link cxgb4_0/- cqe 46 usecnt 2 pid 30503 comm rping
> link cxgb4_0/- cqe 46 usecnt 2 pid 30498 comm rping
> link mlx4_0/- cqe 63 usecnt 2 pid 30494 comm rping
> link mlx4_0/- cqe 63 usecnt 2 pid 30489 comm rping
> link mlx4_0/- cqe 1023 usecnt 2 poll_ctx WORKQUEUE pid 0 comm [ib_core]
>
> # rdma resource show cq pid 30489
> link mlx4_0/- cqe 63 usecnt 2 pid 30489 comm rping
>
> Signed-off-by: Steve Wise <swise@opengridcomputing.com>
> ---
>  rdma/res.c   | 123 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  rdma/utils.c |   5 +++
>  2 files changed, 128 insertions(+)
>
> diff --git a/rdma/res.c b/rdma/res.c
> index beae7dc..27c1efd 100644
> --- a/rdma/res.c
> +++ b/rdma/res.c
> @@ -21,6 +21,8 @@ static int res_help(struct rd *rd)
>  	pr_out("          resource show qp link [DEV/PORT] [FILTER-NAME FILTER-VALUE]\n");
>  	pr_out("          resource show cm_id link [DEV/PORT]\n");
>  	pr_out("          resource show cm_id link [DEV/PORT] [FILTER-NAME FILTER-VALUE]\n");
> +	pr_out("          resource show cq link [DEV/PORT]\n");
> +	pr_out("          resource show cq link [DEV/PORT] [FILTER-NAME FILTER-VALUE]\n");
>  	return 0;
>  }
>
> @@ -705,6 +707,118 @@ static int res_cm_id_parse_cb(const struct nlmsghdr *nlh, void *data)
>  	return MNL_CB_OK;
>  }
>
> +static void print_cqe(struct rd *rd, uint32_t val)
> +{
> +	if (rd->json_output)
> +		jsonw_uint_field(rd->jw, "cqe", val);
> +	else
> +		pr_out("cqe %u ", val);
> +}
> +
> +static void print_usecnt(struct rd *rd, uint64_t val)
> +{
> +	if (rd->json_output)
> +		jsonw_uint_field(rd->jw, "usecnt", val);
> +	else
> +		pr_out("usecnt %" PRIu64 " ", val);

Interesting, how many users are actually know what the "usecnt" actually means?
Will it be more clear to call it "users" instead of "usecnt"?

> +}
> +
> +static const char *poll_ctx_to_str(uint8_t idx)
> +{
> +	static const char * const cm_id_states_str[] = { "DIRECT", "SOFTIRQ",
> +						      "WORKQUEUE"};
> +
> +	if (idx < ARRAY_SIZE(cm_id_states_str))
> +		return cm_id_states_str[idx];
> +	return "UNKNOWN";
> +}
> +
> +static void print_poll_ctx(struct rd *rd, uint8_t poll_ctx)
> +{
> +	if (rd->json_output) {
> +		jsonw_string_field(rd->jw, "poll_ctx", poll_ctx_to_str(poll_ctx));
> +		return;
> +	}
> +	pr_out("poll_ctx %s ", poll_ctx_to_str(poll_ctx));
> +}
> +
> +static int res_cq_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;
> +	uint32_t 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_CQ])
> +		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_CQ];
> +
> +	mnl_attr_for_each_nested(nla_entry, nla_table) {
> +		struct nlattr *nla_line[RDMA_NLDEV_ATTR_MAX] = {};
> +		char *comm = NULL;
> +		uint32_t pid = 0;
> +		uint8_t poll_ctx = 0;
> +		uint64_t usecnt;
> +		uint32_t cqe;
> +		int err;
> +
> +		err = mnl_attr_parse_nested(nla_entry, rd_attr_cb, nla_line);
> +		if (err != MNL_CB_OK)
> +			return MNL_CB_ERROR;
> +
> +		if (!nla_line[RDMA_NLDEV_ATTR_RES_CQE] ||
> +		    !nla_line[RDMA_NLDEV_ATTR_RES_USECNT] ||

I'm not sure that we will have USECNT in the future, let's not put
requirement for RDMA_NLDEV_ATTR_RES_USECNT here.

> +		    (!nla_line[RDMA_NLDEV_ATTR_RES_PID] &&
> +		     !nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME])) {
> +			return MNL_CB_ERROR;
> +		}
> +
> +		cqe = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_CQE]);
> +		usecnt = mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_USECNT]);
> +		if (nla_line[RDMA_NLDEV_ATTR_RES_POLL_CTX])
> +			poll_ctx = mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_POLL_CTX]);
> +
> +		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
> +			pid = mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
> +			comm = get_task_name(pid);
> +		}
> +
> +		if (rd_check_is_filtered(rd, "pid", pid))

free(comm);

> +			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]);
> +
> +		if (rd->json_output)
> +			jsonw_start_array(rd->jw);
> +
> +		print_link(rd, idx, name, 0, nla_line);
> +		print_cqe(rd, cqe);
> +		print_usecnt(rd, usecnt);
> +		if (nla_line[RDMA_NLDEV_ATTR_RES_POLL_CTX])
> +			print_poll_ctx(rd, poll_ctx);
> +		print_pid(rd, pid);
> +		print_comm(rd, comm, nla_line);
> +
> +		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
> @@ -758,12 +872,21 @@ filters cm_id_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
>  RES_FUNC(res_cm_id,	RDMA_NLDEV_CMD_RES_CM_ID_GET, cm_id_valid_filters,
>  	 false);
>
> +static const struct
> +filters cq_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
> +						   .is_number = false },
> +						   { .name = "pid",
> +						   .is_number = true }};

Can you please add filter of usecnt too? It will give us easy view on
"over crowded" CQs.

> +
> +RES_FUNC(res_cq,	RDMA_NLDEV_CMD_RES_CQ_GET, cq_valid_filters, true);
> +
>  static int res_show(struct rd *rd)
>  {
>  	const struct rd_cmd cmds[] = {
>  		{ NULL,		res_no_args	},
>  		{ "qp",		res_qp		},
>  		{ "cm_id",	res_cm_id	},
> +		{ "cq",		res_cq		},
>  		{ 0 }
>  	};
>
> diff --git a/rdma/utils.c b/rdma/utils.c
> index 906ca73..11b34fe 100644
> --- a/rdma/utils.c
> +++ b/rdma/utils.c
> @@ -387,6 +387,11 @@ static const enum mnl_attr_data_type nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
>  	[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,
> +	[RDMA_NLDEV_ATTR_RES_CQ] = MNL_TYPE_NESTED,
> +	[RDMA_NLDEV_ATTR_RES_CQ_ENTRY] = MNL_TYPE_NESTED,
> +	[RDMA_NLDEV_ATTR_RES_CQE] = MNL_TYPE_U32,
> +	[RDMA_NLDEV_ATTR_RES_USECNT] = MNL_TYPE_U64,
> +	[RDMA_NLDEV_ATTR_RES_POLL_CTX] = MNL_TYPE_U8,
>  };
>
>  int rd_attr_cb(const struct nlattr *attr, void *data)
> --
> 1.8.3.1
>

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

RE: [PATCH RFC iproute-next 3/5] rdma: Add CQ resource tracking information

[Steve Wise]

> 
> On Wed, Feb 14, 2018 at 01:07:01PM -0800, Steve Wise wrote:
> > Sample output:
> >
> > # rdma resource show cq
> > link cxgb4_0/- cqe 46 usecnt 2 pid 30503 comm rping
> > link cxgb4_0/- cqe 46 usecnt 2 pid 30498 comm rping
> > link mlx4_0/- cqe 63 usecnt 2 pid 30494 comm rping
> > link mlx4_0/- cqe 63 usecnt 2 pid 30489 comm rping
> > link mlx4_0/- cqe 1023 usecnt 2 poll_ctx WORKQUEUE pid 0 comm [ib_core]
> >
> > # rdma resource show cq pid 30489
> > link mlx4_0/- cqe 63 usecnt 2 pid 30489 comm rping
> >
> > Signed-off-by: Steve Wise <swise@opengridcomputing.com>
> > ---
> >  rdma/res.c   | 123
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  rdma/utils.c |   5 +++
> >  2 files changed, 128 insertions(+)
> >
> > diff --git a/rdma/res.c b/rdma/res.c
> > index beae7dc..27c1efd 100644
> > --- a/rdma/res.c
> > +++ b/rdma/res.c
> > @@ -21,6 +21,8 @@ static int res_help(struct rd *rd)
> >  	pr_out("          resource show qp link [DEV/PORT] [FILTER-NAME
> FILTER-VALUE]\n");
> >  	pr_out("          resource show cm_id link [DEV/PORT]\n");
> >  	pr_out("          resource show cm_id link [DEV/PORT] [FILTER-NAME
> FILTER-VALUE]\n");
> > +	pr_out("          resource show cq link [DEV/PORT]\n");
> > +	pr_out("          resource show cq link [DEV/PORT] [FILTER-NAME
> FILTER-VALUE]\n");
> >  	return 0;
> >  }
> >
> > @@ -705,6 +707,118 @@ static int res_cm_id_parse_cb(const struct
> nlmsghdr *nlh, void *data)
> >  	return MNL_CB_OK;
> >  }
> >
> > +static void print_cqe(struct rd *rd, uint32_t val)
> > +{
> > +	if (rd->json_output)
> > +		jsonw_uint_field(rd->jw, "cqe", val);
> > +	else
> > +		pr_out("cqe %u ", val);
> > +}
> > +
> > +static void print_usecnt(struct rd *rd, uint64_t val)
> > +{
> > +	if (rd->json_output)
> > +		jsonw_uint_field(rd->jw, "usecnt", val);
> > +	else
> > +		pr_out("usecnt %" PRIu64 " ", val);
> 
> Interesting, how many users are actually know what the "usecnt" actually
> means?
> Will it be more clear to call it "users" instead of "usecnt"?
> 

Users is good.

> > +}
> > +
> > +static const char *poll_ctx_to_str(uint8_t idx)
> > +{
> > +	static const char * const cm_id_states_str[] = { "DIRECT", "SOFTIRQ",
> > +						      "WORKQUEUE"};
> > +
> > +	if (idx < ARRAY_SIZE(cm_id_states_str))
> > +		return cm_id_states_str[idx];
> > +	return "UNKNOWN";
> > +}
> > +
> > +static void print_poll_ctx(struct rd *rd, uint8_t poll_ctx)
> > +{
> > +	if (rd->json_output) {
> > +		jsonw_string_field(rd->jw, "poll_ctx",
> poll_ctx_to_str(poll_ctx));
> > +		return;
> > +	}
> > +	pr_out("poll_ctx %s ", poll_ctx_to_str(poll_ctx));
> > +}
> > +
> > +static int res_cq_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;
> > +	uint32_t 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_CQ])
> > +		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_CQ];
> > +
> > +	mnl_attr_for_each_nested(nla_entry, nla_table) {
> > +		struct nlattr *nla_line[RDMA_NLDEV_ATTR_MAX] = {};
> > +		char *comm = NULL;
> > +		uint32_t pid = 0;
> > +		uint8_t poll_ctx = 0;
> > +		uint64_t usecnt;
> > +		uint32_t cqe;
> > +		int err;
> > +
> > +		err = mnl_attr_parse_nested(nla_entry, rd_attr_cb, nla_line);
> > +		if (err != MNL_CB_OK)
> > +			return MNL_CB_ERROR;
> > +
> > +		if (!nla_line[RDMA_NLDEV_ATTR_RES_CQE] ||
> > +		    !nla_line[RDMA_NLDEV_ATTR_RES_USECNT] ||
> 
> I'm not sure that we will have USECNT in the future, let's not put
> requirement for RDMA_NLDEV_ATTR_RES_USECNT here.

Agreed.

> 
> > +		    (!nla_line[RDMA_NLDEV_ATTR_RES_PID] &&
> > +		     !nla_line[RDMA_NLDEV_ATTR_RES_KERN_NAME])) {
> > +			return MNL_CB_ERROR;
> > +		}
> > +
> > +		cqe =
> mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_CQE]);
> > +		usecnt =
> mnl_attr_get_u64(nla_line[RDMA_NLDEV_ATTR_RES_USECNT]);
> > +		if (nla_line[RDMA_NLDEV_ATTR_RES_POLL_CTX])
> > +			poll_ctx =
> mnl_attr_get_u8(nla_line[RDMA_NLDEV_ATTR_RES_POLL_CTX]);
> > +
> > +		if (nla_line[RDMA_NLDEV_ATTR_RES_PID]) {
> > +			pid =
> mnl_attr_get_u32(nla_line[RDMA_NLDEV_ATTR_RES_PID]);
> > +			comm = get_task_name(pid);
> > +		}
> > +
> > +		if (rd_check_is_filtered(rd, "pid", pid))
> 
> free(comm);
> 

😊

> > +			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]);
> > +
> > +		if (rd->json_output)
> > +			jsonw_start_array(rd->jw);
> > +
> > +		print_link(rd, idx, name, 0, nla_line);
> > +		print_cqe(rd, cqe);
> > +		print_usecnt(rd, usecnt);
> > +		if (nla_line[RDMA_NLDEV_ATTR_RES_POLL_CTX])
> > +			print_poll_ctx(rd, poll_ctx);
> > +		print_pid(rd, pid);
> > +		print_comm(rd, comm, nla_line);
> > +
> > +		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
> > @@ -758,12 +872,21 @@ filters
> cm_id_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
> >  RES_FUNC(res_cm_id,	RDMA_NLDEV_CMD_RES_CM_ID_GET,
> cm_id_valid_filters,
> >  	 false);
> >
> > +static const struct
> > +filters cq_valid_filters[MAX_NUMBER_OF_FILTERS] = {{ .name = "link",
> > +						   .is_number = false },
> > +						   { .name = "pid",
> > +						   .is_number = true }};
> 
> Can you please add filter of usecnt too? It will give us easy view on
> "over crowded" CQs.

Good idea.

> 
> > +
> > +RES_FUNC(res_cq,	RDMA_NLDEV_CMD_RES_CQ_GET, cq_valid_filters,
> true);
> > +
> >  static int res_show(struct rd *rd)
> >  {
> >  	const struct rd_cmd cmds[] = {
> >  		{ NULL,		res_no_args	},
> >  		{ "qp",		res_qp		},
> >  		{ "cm_id",	res_cm_id	},
> > +		{ "cq",		res_cq		},
> >  		{ 0 }
> >  	};
> >
> > diff --git a/rdma/utils.c b/rdma/utils.c
> > index 906ca73..11b34fe 100644
> > --- a/rdma/utils.c
> > +++ b/rdma/utils.c
> > @@ -387,6 +387,11 @@ static const enum mnl_attr_data_type
> nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
> >  	[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,
> > +	[RDMA_NLDEV_ATTR_RES_CQ] = MNL_TYPE_NESTED,
> > +	[RDMA_NLDEV_ATTR_RES_CQ_ENTRY] = MNL_TYPE_NESTED,
> > +	[RDMA_NLDEV_ATTR_RES_CQE] = MNL_TYPE_U32,
> > +	[RDMA_NLDEV_ATTR_RES_USECNT] = MNL_TYPE_U64,
> > +	[RDMA_NLDEV_ATTR_RES_POLL_CTX] = MNL_TYPE_U8,
> >  };
> >
> >  int rd_attr_cb(const struct nlattr *attr, void *data)
> > --
> > 1.8.3.1
> >