[PATCH iproute2-next 0/2] Print netdevice names and indexes for relevant RDMA devices

From: Leon Romanovsky <redacted>

Hi,

This is user space part of the kernel patch [1], which exposed net
device name and index through RDMA netlink interface.

First patch is actually updates rdma_netlink.h to the latest kernel
variant, while second patch implements the print.

[1] https://patchwork.kernel.org/patch/10307339/

Leon Romanovsky (2):
  rdma: Update RDMA header file
  rdma: Print net device name and index for RDMA device

 rdma/include/uapi/rdma/rdma_netlink.h | 42 +++++++++++++++++++++++++++++++++++
 rdma/link.c                           | 21 ++++++++++++++++++
 rdma/utils.c                          |  2 ++
 3 files changed, 65 insertions(+)

From: Leon Romanovsky <redacted>

Update RDMA netlink header file to kernel commit 29cf1351d450
("RDMA/nldev: provide detailed PD information")

Signed-off-by: Leon Romanovsky <redacted>
---
 rdma/include/uapi/rdma/rdma_netlink.h | 38 +++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)

diff --git a/rdma/include/uapi/rdma/rdma_netlink.h b/rdma/include/uapi/rdma/rdma_netlink.h
index dbac3b82..9446a721 100644
--- a/rdma/include/uapi/rdma/rdma_netlink.h
+++ b/rdma/include/uapi/rdma/rdma_netlink.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 +358,36 @@ 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 socket addresses
+	 */
+	RDMA_NLDEV_ATTR_RES_SRC_ADDR,		/* __kernel_sockaddr_storage */
+	RDMA_NLDEV_ATTR_RES_DST_ADDR,		/* __kernel_sockaddr_storage */
+
+	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_PD,			/* nested table */
+	RDMA_NLDEV_ATTR_RES_PD_ENTRY,		/* nested table */
+	RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY,	/* u32 */
+	RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY,	/* u32 */
+
 	RDMA_NLDEV_ATTR_MAX
 };
 #endif /* _RDMA_NETLINK_H */

From: Leon Romanovsky <redacted>

The RDMA devices are operated in RoCE and iWARP modes have net device
underneath. Present their names in regular output and their net index
in detailed mode.

[root@nps ~]# rdma link show mlx5_3/1
4/1: mlx5_3/1: state ACTIVE physical_state LINK_UP netdev ens7
[root@nps ~]# rdma link show mlx5_3/1 -d
4/1: mlx5_3/1: state ACTIVE physical_state LINK_UP netdev ens7 netdev_index 7
    caps: <CM, IP_BASED_GIDS>

Signed-off-by: Leon Romanovsky <redacted>
---
 rdma/include/uapi/rdma/rdma_netlink.h |  4 ++++
 rdma/link.c                           | 21 +++++++++++++++++++++
 rdma/utils.c                          |  2 ++
 3 files changed, 27 insertions(+)

diff --git a/rdma/include/uapi/rdma/rdma_netlink.h b/rdma/include/uapi/rdma/rdma_netlink.h
index 9446a721..45474f13 100644
--- a/rdma/include/uapi/rdma/rdma_netlink.h
+++ b/rdma/include/uapi/rdma/rdma_netlink.h

@@ -388,6 +388,10 @@ enum rdma_nldev_attr {
 	RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY,	/* u32 */
 	RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY,	/* u32 */

+	/* Netdev information for relevant protocols, like RoCE and iWARP */
+	RDMA_NLDEV_ATTR_NDEV_INDEX,		/* u32 */
+	RDMA_NLDEV_ATTR_NDEV_NAME,		/* string */
+
 	RDMA_NLDEV_ATTR_MAX
 };
 #endif /* _RDMA_NETLINK_H */

diff --git a/rdma/link.c b/rdma/link.c
index 66bcd50e..7e914c87 100644
--- a/rdma/link.c
+++ b/rdma/link.c

@@ -205,6 +205,26 @@ static void link_print_phys_state(struct rd *rd, struct nlattr **tb)
 		pr_out("physical_state %s ", phys_state_to_str(phys_state));
 }

+static void link_print_netdev(struct rd *rd, struct nlattr **tb)
+{
+	const char *netdev_name;
+	uint32_t idx;
+
+	if (!tb[RDMA_NLDEV_ATTR_NDEV_NAME] || !tb[RDMA_NLDEV_ATTR_NDEV_INDEX])
+		return;
+
+	netdev_name = mnl_attr_get_str(tb[RDMA_NLDEV_ATTR_NDEV_NAME]);
+	idx = mnl_attr_get_u32(tb[RDMA_NLDEV_ATTR_NDEV_INDEX]);
+	if (rd->json_output) {
+		jsonw_string_field(rd->jw, "netdev", netdev_name);
+		jsonw_uint_field(rd->jw, "netdev_index", idx);
+	} else {
+		pr_out("netdev %s ", netdev_name);
+		if (rd->show_details)
+			pr_out("netdev_index %u ", idx);
+	}
+}
+
 static int link_parse_cb(const struct nlmsghdr *nlh, void *data)
 {
 	struct nlattr *tb[RDMA_NLDEV_ATTR_MAX] = {};

@@ -241,6 +261,7 @@ static int link_parse_cb(const struct nlmsghdr *nlh, void *data)
 	link_print_lmc(rd, tb);
 	link_print_state(rd, tb);
 	link_print_phys_state(rd, tb);
+	link_print_netdev(rd, tb);
 	if (rd->show_details)
 		link_print_caps(rd, tb);

diff --git a/rdma/utils.c b/rdma/utils.c
index f9460162..4fed80ab 100644
--- a/rdma/utils.c
+++ b/rdma/utils.c

@@ -375,6 +375,8 @@ 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_NDEV_INDEX]		= MNL_TYPE_U32,
+	[RDMA_NLDEV_ATTR_NDEV_NAME]		= MNL_TYPE_NUL_STRING,
 };

 int rd_attr_cb(const struct nlattr *attr, void *data)

On 3/26/18 2:28 AM, Leon Romanovsky wrote:

Why wasn't the above included in patch 1 as part of the header file update?

On Mon, Mar 26, 2018 at 08:06:52AM -0600, David Ahern wrote:

On 3/26/18 2:28 AM, Leon Romanovsky wrote:

quoted

diff --git a/rdma/include/uapi/rdma/rdma_netlink.h b/rdma/include/uapi/rdma/rdma_netlink.h
index 9446a721..45474f13 100644
--- a/rdma/include/uapi/rdma/rdma_netlink.h
+++ b/rdma/include/uapi/rdma/rdma_netlink.h

@@ -388,6 +388,10 @@ enum rdma_nldev_attr {
 	RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY,	/* u32 */
 	RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY,	/* u32 */

+	/* Netdev information for relevant protocols, like RoCE and iWARP */
+	RDMA_NLDEV_ATTR_NDEV_INDEX,		/* u32 */
+	RDMA_NLDEV_ATTR_NDEV_NAME,		/* string */
+
 	RDMA_NLDEV_ATTR_MAX
 };
 #endif /* _RDMA_NETLINK_H */

Why wasn't the above included in patch 1 as part of the header file update?

Kernel part is not accepted yet, this is why I separated them.

Thanks

On 3/26/2018 3:28 AM, Leon Romanovsky wrote:

Why is the netdev_index under 'show_details'?

Reviewed-by: Steve Wise <redacted>

On 3/26/18 9:09 AM, Leon Romanovsky wrote:

On Mon, Mar 26, 2018 at 08:06:52AM -0600, David Ahern wrote:

quoted

On 3/26/18 2:28 AM, Leon Romanovsky wrote:

quoted

diff --git a/rdma/include/uapi/rdma/rdma_netlink.h b/rdma/include/uapi/rdma/rdma_netlink.h
index 9446a721..45474f13 100644
--- a/rdma/include/uapi/rdma/rdma_netlink.h
+++ b/rdma/include/uapi/rdma/rdma_netlink.h

@@ -388,6 +388,10 @@ enum rdma_nldev_attr {
 	RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY,	/* u32 */
 	RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY,	/* u32 */

+	/* Netdev information for relevant protocols, like RoCE and iWARP */
+	RDMA_NLDEV_ATTR_NDEV_INDEX,		/* u32 */
+	RDMA_NLDEV_ATTR_NDEV_NAME,		/* string */
+
 	RDMA_NLDEV_ATTR_MAX
 };
 #endif /* _RDMA_NETLINK_H */

Why wasn't the above included in patch 1 as part of the header file update?

Kernel part is not accepted yet, this is why I separated them.

Please don't mix patches in a set like this. The entire set should be
consistent -- either all of them are in the kernel or all of them are
waiting to be committed.

On Mon, Mar 26, 2018 at 09:44:32AM -0600, David Ahern wrote:

On 3/26/18 9:09 AM, Leon Romanovsky wrote:

quoted

On Mon, Mar 26, 2018 at 08:06:52AM -0600, David Ahern wrote:

quoted

On 3/26/18 2:28 AM, Leon Romanovsky wrote:

quoted

diff --git a/rdma/include/uapi/rdma/rdma_netlink.h b/rdma/include/uapi/rdma/rdma_netlink.h
index 9446a721..45474f13 100644
--- a/rdma/include/uapi/rdma/rdma_netlink.h
+++ b/rdma/include/uapi/rdma/rdma_netlink.h

@@ -388,6 +388,10 @@ enum rdma_nldev_attr {
 	RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY,	/* u32 */
 	RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY,	/* u32 */

+	/* Netdev information for relevant protocols, like RoCE and iWARP */
+	RDMA_NLDEV_ATTR_NDEV_INDEX,		/* u32 */
+	RDMA_NLDEV_ATTR_NDEV_NAME,		/* string */
+
 	RDMA_NLDEV_ATTR_MAX
 };
 #endif /* _RDMA_NETLINK_H */

Why wasn't the above included in patch 1 as part of the header file update?

Kernel part is not accepted yet, this is why I separated them.

Please don't mix patches in a set like this. The entire set should be
consistent -- either all of them are in the kernel or all of them are
waiting to be committed.

Ok, I'll do.

On Mon, Mar 26, 2018 at 10:17:12AM -0500, Steve Wise wrote:

On 3/26/2018 3:28 AM, Leon Romanovsky wrote:

quoted

From: Leon Romanovsky <redacted>

The RDMA devices are operated in RoCE and iWARP modes have net device
underneath. Present their names in regular output and their net index
in detailed mode.

[root@nps ~]# rdma link show mlx5_3/1
4/1: mlx5_3/1: state ACTIVE physical_state LINK_UP netdev ens7
[root@nps ~]# rdma link show mlx5_3/1 -d
4/1: mlx5_3/1: state ACTIVE physical_state LINK_UP netdev ens7 netdev_index 7
    caps: <CM, IP_BASED_GIDS>

Signed-off-by: Leon Romanovsky <redacted>
---
 rdma/include/uapi/rdma/rdma_netlink.h |  4 ++++
 rdma/link.c                           | 21 +++++++++++++++++++++
 rdma/utils.c                          |  2 ++
 3 files changed, 27 insertions(+)

diff --git a/rdma/include/uapi/rdma/rdma_netlink.h b/rdma/include/uapi/rdma/rdma_netlink.h
index 9446a721..45474f13 100644
--- a/rdma/include/uapi/rdma/rdma_netlink.h
+++ b/rdma/include/uapi/rdma/rdma_netlink.h

@@ -388,6 +388,10 @@ enum rdma_nldev_attr {
 	RDMA_NLDEV_ATTR_RES_LOCAL_DMA_LKEY,	/* u32 */
 	RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY,	/* u32 */

+	/* Netdev information for relevant protocols, like RoCE and iWARP */
+	RDMA_NLDEV_ATTR_NDEV_INDEX,		/* u32 */
+	RDMA_NLDEV_ATTR_NDEV_NAME,		/* string */
+
 	RDMA_NLDEV_ATTR_MAX
 };
 #endif /* _RDMA_NETLINK_H */

diff --git a/rdma/link.c b/rdma/link.c
index 66bcd50e..7e914c87 100644
--- a/rdma/link.c
+++ b/rdma/link.c

@@ -205,6 +205,26 @@ static void link_print_phys_state(struct rd *rd, struct nlattr **tb)
 		pr_out("physical_state %s ", phys_state_to_str(phys_state));
 }

+static void link_print_netdev(struct rd *rd, struct nlattr **tb)
+{
+	const char *netdev_name;
+	uint32_t idx;
+
+	if (!tb[RDMA_NLDEV_ATTR_NDEV_NAME] || !tb[RDMA_NLDEV_ATTR_NDEV_INDEX])
+		return;
+
+	netdev_name = mnl_attr_get_str(tb[RDMA_NLDEV_ATTR_NDEV_NAME]);
+	idx = mnl_attr_get_u32(tb[RDMA_NLDEV_ATTR_NDEV_INDEX]);
+	if (rd->json_output) {
+		jsonw_string_field(rd->jw, "netdev", netdev_name);
+		jsonw_uint_field(rd->jw, "netdev_index", idx);
+	} else {
+		pr_out("netdev %s ", netdev_name);
+		if (rd->show_details)
+			pr_out("netdev_index %u ", idx);
+	}
+}
+

Why is the netdev_index under 'show_details'?

Because I think that it is useless for day-to-day usage of average
rdmatool user and mostly will be used as pipe to other scripts, if JSON
doesn't good fit for any reasons. So I decided do not clutter "general
output".

Reviewed-by: Steve Wise <redacted>

Thanks

quoted

 static int link_parse_cb(const struct nlmsghdr *nlh, void *data)
 {
 	struct nlattr *tb[RDMA_NLDEV_ATTR_MAX] = {};

@@ -241,6 +261,7 @@ static int link_parse_cb(const struct nlmsghdr *nlh, void *data)
 	link_print_lmc(rd, tb);
 	link_print_state(rd, tb);
 	link_print_phys_state(rd, tb);
+	link_print_netdev(rd, tb);
 	if (rd->show_details)
 		link_print_caps(rd, tb);

diff --git a/rdma/utils.c b/rdma/utils.c
index f9460162..4fed80ab 100644
--- a/rdma/utils.c
+++ b/rdma/utils.c

@@ -375,6 +375,8 @@ 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_NDEV_INDEX]		= MNL_TYPE_U32,
+	[RDMA_NLDEV_ATTR_NDEV_NAME]		= MNL_TYPE_NUL_STRING,
 };

 int rd_attr_cb(const struct nlattr *attr, void *data)
--

2.14.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html