LU-14798 lnet: add LNet GPU Direct Support

[fs/lustre-release.git] / lnet / include / lnet / lib-types.h

diff --git a/lnet/include/lnet/lib-types.h b/lnet/include/lnet/lib-types.h
index 010cf32..ebe6065 100644 (file)
--- a/lnet/include/lnet/lib-types.h
+++ b/lnet/include/lnet/lib-types.h
@@ -27,7 +27,6 @@
  */
 /*
  * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
  *
  * lnet/include/lnet/lib-types.h
  *
@@ -46,7 +45,10 @@
 #include <linux/uio.h>
 #include <linux/semaphore.h>
 #include <linux/types.h>
+#include <linux/kref.h>
+#include <net/genetlink.h>
 
+#include <uapi/linux/lnet/lnet-nl.h>
 #include <uapi/linux/lnet/lnet-dlc.h>
 #include <uapi/linux/lnet/lnetctl.h>
 #include <uapi/linux/lnet/nidstr.h>
@@ -123,6 +125,8 @@ struct lnet_msg {
        enum lnet_msg_hstatus   msg_health_status;
        /* This is a recovery message */
        bool                    msg_recovery;
+       /* force an RDMA even if the message size is < 4K */
+       bool                    msg_rdma_force;
        /* the number of times a transmission has been retried */
        int                     msg_retry_count;
        /* flag to indicate that we do not want to resend this message */
@@ -307,6 +311,10 @@ struct lnet_lnd {
 
        /* accept a new connection */
        int (*lnd_accept)(struct lnet_ni *ni, struct socket *sock);
+
+       /* get dma_dev priority */
+       unsigned int (*lnd_get_dev_prio)(struct lnet_ni *ni,
+                                        unsigned int dev_idx);
 };
 
 struct lnet_tx_queue {
@@ -473,6 +481,13 @@ struct lnet_ni {
        /* Recovery state. Protected by lnet_ni_lock() */
        __u32                   ni_recovery_state;
 
+       /* When to send the next recovery ping */
+       time64_t                ni_next_ping;
+       /* How many pings sent during current recovery period did not receive
+        * a reply. NB: reset whenever _any_ message arrives on this NI
+        */
+       unsigned int            ni_ping_count;
+
        /* per NI LND tunables */
        struct lnet_lnd_tunables ni_lnd_tunables;
 
@@ -509,10 +524,9 @@ struct lnet_ni {
        __u32                   ni_sel_priority;
 
        /*
-        * equivalent interfaces to use
-        * This is an array because socklnd bonding can still be configured
+        * equivalent interface to use
         */
-       char                    *ni_interfaces[LNET_INTERFACES_NUM];
+       char                    *ni_interface;
        struct net              *ni_net_ns;     /* original net namespace */
 };
 
@@ -581,7 +595,7 @@ struct lnet_peer_ni {
        /* peer's NID */
        lnet_nid_t              lpni_nid;
        /* # refs */
-       atomic_t                lpni_refcount;
+       struct kref             lpni_kref;
        /* health value for the peer */
        atomic_t                lpni_healthv;
        /* recovery ping mdh */
@@ -646,6 +660,8 @@ struct lnet_peer {
 
        /* source NID to use during discovery */
        lnet_nid_t              lp_disc_src_nid;
+       /* destination NID to use during discovery */
+       lnet_nid_t              lp_disc_dst_nid;
 
        /* net to perform discovery on */
        __u32                   lp_disc_net_id;
@@ -804,8 +820,8 @@ struct lnet_peer_net {
        /* peer net health */
        int                     lpn_healthv;
 
-       /* time of last router net check attempt */
-       time64_t                lpn_rtrcheck_timestamp;
+       /* time of next router ping on this net */
+       time64_t                lpn_next_ping;
 
        /* selection sequence number */
        __u32                   lpn_seq;
@@ -862,7 +878,7 @@ struct lnet_route {
        int                     lr_seq;         /* sequence for round-robin */
        __u32                   lr_hops;        /* how far I am */
        unsigned int            lr_priority;    /* route priority */
-       bool                    lr_alive;       /* cached route aliveness */
+       atomic_t                lr_alive;       /* cached route aliveness */
        bool                    lr_single_hop;  /* this route is single-hop */
 };
 
@@ -1246,4 +1262,44 @@ struct lnet {
        struct list_head                ln_udsp_list;
 };
 
+static const struct nla_policy scalar_attr_policy[LN_SCALAR_CNT + 1] = {
+       [LN_SCALAR_ATTR_LIST]           = { .type = NLA_NESTED },
+       [LN_SCALAR_ATTR_LIST_SIZE]      = { .type = NLA_U16 },
+       [LN_SCALAR_ATTR_INDEX]          = { .type = NLA_U16 },
+       [LN_SCALAR_ATTR_NLA_TYPE]       = { .type = NLA_U16 },
+       [LN_SCALAR_ATTR_VALUE]          = { .type = NLA_STRING },
+       [LN_SCALAR_ATTR_KEY_FORMAT]     = { .type = NLA_U16 },
+};
+
+int lnet_genl_send_scalar_list(struct sk_buff *msg, u32 portid, u32 seq,
+                              const struct genl_family *family, int flags,
+                              u8 cmd, const struct ln_key_list *data[]);
+
+/* Special workaround for pre-4.19 kernels to send error messages
+ * from dumpit routines. Newer kernels will send message with
+ * NL_SET_ERR_MSG information by default if NETLINK_EXT_ACK is set.
+ */
+static inline int lnet_nl_send_error(struct sk_buff *msg, int portid, int seq,
+                                    int error)
+{
+#ifndef HAVE_NL_DUMP_WITH_EXT_ACK
+       struct nlmsghdr *nlh;
+
+       if (!error)
+               return 0;
+
+       nlh = nlmsg_put(msg, portid, seq, NLMSG_ERROR, sizeof(error), 0);
+       if (!nlh)
+               return -ENOMEM;
+#ifdef HAVE_NL_PARSE_WITH_EXT_ACK
+       netlink_ack(msg, nlh, error, NULL);
+#else
+       netlink_ack(msg, nlh, error);
+#endif
+       return nlmsg_len(nlh);
+#else
+       return error;
+#endif
+}
+
 #endif