From: Parav Pandit <parav@xxxxxxxxxxxx>
Add an interface via netlink command to query whether rdma devices are
shared among multiple net namespaces or not.
When using rdma tool, it can be queried as,
$rdma sys show netns
netns: shared
Signed-off-by: Parav Pandit <parav@xxxxxxxxxxxx>
Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx>
---
drivers/infiniband/core/core_priv.h | 2 ++
drivers/infiniband/core/device.c | 10 ++++----
drivers/infiniband/core/net_namespace.c | 15 ++++++++++++
drivers/infiniband/core/nldev.c | 32 +++++++++++++++++++++++++
include/uapi/rdma/rdma_netlink.h | 10 +++++++-
5 files changed, 64 insertions(+), 5 deletions(-)
diff --git a/drivers/infiniband/core/core_priv.h b/drivers/infiniband/core/core_priv.h
index 9ebce87c7961..1ba188efec44 100644
--- a/drivers/infiniband/core/core_priv.h
+++ b/drivers/infiniband/core/core_priv.h
@@ -59,6 +59,8 @@ extern struct list_head ib_device_list;
extern struct mutex ib_device_mutex;
extern struct rw_semaphore ib_lists_rwsem;
+extern ushort ib_devices_shared_netns;
+
int ib_device_register_sysfs(struct ib_device *device,
int (*port_callback)(struct ib_device *,
u8, struct kobject *));
diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 12877984b10c..68934bf14f76 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -676,11 +676,13 @@ int ib_register_device(struct ib_device *device, const char *name,
goto cg_cleanup;
}
- device->netns_shared = 1;
- /* Perform this under device mutex lock, so that it can synchronize
- * with _init_net() to avoid duplicate additions for a given device.
+ /* Perform compat device creation under device mutex lock, so that it
+ * can synchronize with _init_net() notifier to avoid duplicate
+ * additions for a given device.
*/
- rdma_compatdev_add(device);
+ device->netns_shared = ib_devices_shared_netns;
+ if (ib_devices_shared_netns)
+ rdma_compatdev_add(device);
device->reg_state = IB_DEV_REGISTERED;
diff --git a/drivers/infiniband/core/net_namespace.c b/drivers/infiniband/core/net_namespace.c
index 569049dd4216..57b1fe3d1d16 100644
--- a/drivers/infiniband/core/net_namespace.c
+++ b/drivers/infiniband/core/net_namespace.c
@@ -32,6 +32,17 @@ struct rdma_dev_net {
};
static unsigned int rdma_dev_net_id;
+/**
+ * ib_devices_shared_netns - rdma devices shared between multiple
+ * network namespaces or not.
+ * true = shared mode:
+ * All IB devices can be shared among multiple network namespace.
+ * This is default mode.
+ * false = single namespace mode:
+ * IB device resides in only one net namespace.
+ *
+ */
+ushort ib_devices_shared_netns = 1;
static void rdma_compatdev_release(struct device *dev)
{
@@ -180,6 +191,9 @@ static __net_init int rdma_dev_init_net(struct net *net)
* which also tries to add compat devices.
*/
mutex_lock(&ib_device_mutex);
+ if (!ib_devices_shared_netns)
+ goto done;
+
/* Hold ib_lists_rwsem read lock; thereby not assume that
* ib_device_mutex is always locked while accessing ib_device_list.
*/
@@ -187,6 +201,7 @@ static __net_init int rdma_dev_init_net(struct net *net)
list_for_each_entry(device, &ib_device_list, core_list)
rdma_compatdev_create(device, net);
up_read(&ib_lists_rwsem);
+done:
mutex_unlock(&ib_device_mutex);
return 0;
}
diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c
index 9591193cf285..881b620bed7a 100644
--- a/drivers/infiniband/core/nldev.c
+++ b/drivers/infiniband/core/nldev.c
@@ -107,6 +107,7 @@ static const struct nla_policy nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
[RDMA_NLDEV_ATTR_DRIVER_U32] = { .type = NLA_U32 },
[RDMA_NLDEV_ATTR_DRIVER_S64] = { .type = NLA_S64 },
[RDMA_NLDEV_ATTR_DRIVER_U64] = { .type = NLA_U64 },
+ [RDMA_NLDEV_SYS_ATTR_NETNS_MODE] = { .type = NLA_U8 },
};
static int put_driver_name_print_type(struct sk_buff *msg, const char *name,
@@ -1107,6 +1108,34 @@ static int nldev_res_get_pd_dumpit(struct sk_buff *skb,
return res_get_common_dumpit(skb, cb, RDMA_RESTRACK_PD);
}
+static int nldev_get_sys_get_dumpit(struct sk_buff *skb,
+ struct netlink_callback *cb)
+{
+ struct nlattr *tb[RDMA_NLDEV_ATTR_MAX];
+ struct nlmsghdr *nlh;
+ int err;
+
+ err = nlmsg_parse(cb->nlh, 0, tb, RDMA_NLDEV_ATTR_MAX - 1,
+ nldev_policy, NULL);
+ if (err)
+ return err;
+
+ nlh = nlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
+ RDMA_NL_GET_TYPE(RDMA_NL_NLDEV,
+ RDMA_NLDEV_CMD_SYS_GET),
+ 0, 0);
+
+ err = nla_put_u8(skb, RDMA_NLDEV_SYS_ATTR_NETNS_MODE,
+ (u8)ib_devices_shared_netns);
+ if (err) {
+ nlmsg_cancel(skb, nlh);
+ return err;
+ }
+
+ nlmsg_end(skb, nlh);
+ return skb->len;
+}
+
static const struct rdma_nl_cbs nldev_cb_table[RDMA_NLDEV_NUM_OPS] = {
[RDMA_NLDEV_CMD_GET] = {
.doit = nldev_get_doit,
@@ -1149,6 +1178,9 @@ static const struct rdma_nl_cbs nldev_cb_table[RDMA_NLDEV_NUM_OPS] = {
[RDMA_NLDEV_CMD_RES_PD_GET] = {
.dump = nldev_res_get_pd_dumpit,
},
+ [RDMA_NLDEV_CMD_SYS_GET] = {
+ .dump = nldev_get_sys_get_dumpit,
+ },
};
void __init nldev_init(void)
diff --git a/include/uapi/rdma/rdma_netlink.h b/include/uapi/rdma/rdma_netlink.h
index 2e18b77a817f..1cda30f392e7 100644
--- a/include/uapi/rdma/rdma_netlink.h
+++ b/include/uapi/rdma/rdma_netlink.h
@@ -229,7 +229,9 @@ enum rdma_nldev_command {
RDMA_NLDEV_CMD_GET, /* can dump */
RDMA_NLDEV_CMD_SET,
- /* 3 - 4 are free to use */
+ RDMA_NLDEV_CMD_SYS_GET, /* can dump */
+
+ /* 4 is free to use */
RDMA_NLDEV_CMD_PORT_GET = 5, /* can dump */
@@ -430,6 +432,12 @@ enum rdma_nldev_attr {
RDMA_NLDEV_ATTR_DRIVER_S64, /* s64 */
RDMA_NLDEV_ATTR_DRIVER_U64, /* u64 */
+ /*
+ * net namespace mode for rdma subsystem:
+ * either shared or exclusive among multiple net namespaces.
+ */
+ RDMA_NLDEV_SYS_ATTR_NETNS_MODE, /* u8 */
+
/*
* Always the end
*/
--
2.19.1
