*
* You should have received a copy of the GNU General Public License
* version 2 along with this program; If not, see
- * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
- *
- * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
- * CA 95054 USA or visit www.sun.com if you need additional information or
- * have any questions.
+ * http://www.gnu.org/licenses/gpl-2.0.html
*
* GPL HEADER END
*/
* Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
- * Copyright (c) 2012, Intel Corporation.
+ * Copyright (c) 2012, 2017, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
#ifndef __LNET_LIB_TYPES_H__
#define __LNET_LIB_TYPES_H__
-#if defined(__linux__)
-#include <lnet/linux/lib-types.h>
-#elif defined(__APPLE__)
-#include <lnet/darwin/lib-types.h>
-#elif defined(__WINNT__)
-#include <lnet/winnt/lib-types.h>
-#include <libcfs/libcfs_pack.h>
-#include <libcfs/libcfs_unpack.h>
-#else
-#error Unsupported Operating System
+#ifndef __KERNEL__
+# error This include is only for kernel use.
#endif
-#include <libcfs/libcfs.h>
-#include <libcfs/list.h>
-#include <lnet/types.h>
-
-#define WIRE_ATTR __attribute__((packed))
-
-/* Packed version of lnet_process_id_t to transfer via network */
-typedef struct {
- lnet_nid_t nid;
- lnet_pid_t pid; /* node id / process id */
-} WIRE_ATTR lnet_process_id_packed_t;
-
-/* The wire handle's interface cookie only matches one network interface in
- * one epoch (i.e. new cookie when the interface restarts or the node
- * reboots). The object cookie only matches one object on that interface
- * during that object's lifetime (i.e. no cookie re-use). */
-typedef struct {
- __u64 wh_interface_cookie;
- __u64 wh_object_cookie;
-} WIRE_ATTR lnet_handle_wire_t;
-
-typedef enum {
- LNET_MSG_ACK = 0,
- LNET_MSG_PUT,
- LNET_MSG_GET,
- LNET_MSG_REPLY,
- LNET_MSG_HELLO,
-} lnet_msg_type_t;
-
-/* The variant fields of the portals message header are aligned on an 8
- * byte boundary in the message header. Note that all types used in these
- * wire structs MUST be fixed size and the smaller types are placed at the
- * end. */
-typedef struct lnet_ack {
- lnet_handle_wire_t dst_wmd;
- __u64 match_bits;
- __u32 mlength;
-} WIRE_ATTR lnet_ack_t;
-
-typedef struct lnet_put {
- lnet_handle_wire_t ack_wmd;
- __u64 match_bits;
- __u64 hdr_data;
- __u32 ptl_index;
- __u32 offset;
-} WIRE_ATTR lnet_put_t;
-
-typedef struct lnet_get {
- lnet_handle_wire_t return_wmd;
- __u64 match_bits;
- __u32 ptl_index;
- __u32 src_offset;
- __u32 sink_length;
-} WIRE_ATTR lnet_get_t;
-
-typedef struct lnet_reply {
- lnet_handle_wire_t dst_wmd;
-} WIRE_ATTR lnet_reply_t;
-
-typedef struct lnet_hello {
- __u64 incarnation;
- __u32 type;
-} WIRE_ATTR lnet_hello_t;
-
-typedef struct {
- lnet_nid_t dest_nid;
- lnet_nid_t src_nid;
- lnet_pid_t dest_pid;
- lnet_pid_t src_pid;
- __u32 type; /* lnet_msg_type_t */
- __u32 payload_length; /* payload data to follow */
- /*<------__u64 aligned------->*/
- union {
- lnet_ack_t ack;
- lnet_put_t put;
- lnet_get_t get;
- lnet_reply_t reply;
- lnet_hello_t hello;
- } msg;
-} WIRE_ATTR lnet_hdr_t;
-
-/* A HELLO message contains a magic number and protocol version
- * code in the header's dest_nid, the peer's NID in the src_nid, and
- * LNET_MSG_HELLO in the type field. All other common fields are zero
- * (including payload_size; i.e. no payload).
- * This is for use by byte-stream LNDs (e.g. TCP/IP) to check the peer is
- * running the same protocol and to find out its NID. These LNDs should
- * exchange HELLO messages when a connection is first established. Individual
- * LNDs can put whatever else they fancy in lnet_hdr_t::msg.
+#include <linux/kthread.h>
+#include <linux/uio.h>
+#include <linux/semaphore.h>
+#include <linux/types.h>
+
+#include <uapi/linux/lnet/lnet-dlc.h>
+#include <uapi/linux/lnet/lnetctl.h>
+
+/* Max payload size */
+#define LNET_MAX_PAYLOAD LNET_MTU
+
+#define LNET_MAX_IOV (LNET_MAX_PAYLOAD >> PAGE_SHIFT)
+
+/*
+ * This is the maximum health value.
+ * All local and peer NIs created have their health default to this value.
*/
-typedef struct {
- __u32 magic; /* LNET_PROTO_TCP_MAGIC */
- __u16 version_major; /* increment on incompatible change */
- __u16 version_minor; /* increment on compatible change */
-} WIRE_ATTR lnet_magicversion_t;
-
-/* PROTO MAGIC for LNDs */
-#define LNET_PROTO_IB_MAGIC 0x0be91b91
-#define LNET_PROTO_RA_MAGIC 0x0be91b92
-#define LNET_PROTO_QSW_MAGIC 0x0be91b93
-#define LNET_PROTO_GNI_MAGIC 0xb00fbabe /* ask Kim */
-#define LNET_PROTO_TCP_MAGIC 0xeebc0ded
-#define LNET_PROTO_PTL_MAGIC 0x50746C4E /* 'PtlN' unique magic */
-#define LNET_PROTO_MX_MAGIC 0x4d583130 /* 'MX10'! */
-#define LNET_PROTO_ACCEPTOR_MAGIC 0xacce7100
-#define LNET_PROTO_PING_MAGIC 0x70696E67 /* 'ping' */
-
-/* Placeholder for a future "unified" protocol across all LNDs */
-/* Current LNDs that receive a request with this magic will respond with a
- * "stub" reply using their current protocol */
-#define LNET_PROTO_MAGIC 0x45726963 /* ! */
-
-
-#define LNET_PROTO_TCP_VERSION_MAJOR 1
-#define LNET_PROTO_TCP_VERSION_MINOR 0
-
-/* Acceptor connection request */
-typedef struct {
- __u32 acr_magic; /* PTL_ACCEPTOR_PROTO_MAGIC */
- __u32 acr_version; /* protocol version */
- __u64 acr_nid; /* target NID */
-} WIRE_ATTR lnet_acceptor_connreq_t;
-
-#define LNET_PROTO_ACCEPTOR_VERSION 1
+#define LNET_MAX_HEALTH_VALUE 1000
/* forward refs */
struct lnet_libmd;
-typedef struct lnet_msg {
- cfs_list_t msg_activelist;
- cfs_list_t msg_list; /* Q for credits/MD */
+enum lnet_msg_hstatus {
+ LNET_MSG_STATUS_OK = 0,
+ LNET_MSG_STATUS_LOCAL_INTERRUPT,
+ LNET_MSG_STATUS_LOCAL_DROPPED,
+ LNET_MSG_STATUS_LOCAL_ABORTED,
+ LNET_MSG_STATUS_LOCAL_NO_ROUTE,
+ LNET_MSG_STATUS_LOCAL_ERROR,
+ LNET_MSG_STATUS_LOCAL_TIMEOUT,
+ LNET_MSG_STATUS_REMOTE_ERROR,
+ LNET_MSG_STATUS_REMOTE_DROPPED,
+ LNET_MSG_STATUS_REMOTE_TIMEOUT,
+ LNET_MSG_STATUS_NETWORK_TIMEOUT,
+ LNET_MSG_STATUS_END,
+};
+
+struct lnet_rsp_tracker {
+ /* chain on the waiting list */
+ struct list_head rspt_on_list;
+ /* cpt to lock */
+ int rspt_cpt;
+ /* nid of next hop */
+ lnet_nid_t rspt_next_hop_nid;
+ /* deadline of the REPLY/ACK */
+ ktime_t rspt_deadline;
+ /* parent MD */
+ struct lnet_handle_md rspt_mdh;
+};
+
+struct lnet_msg {
+ struct list_head msg_activelist;
+ struct list_head msg_list; /* Q for credits/MD */
- lnet_process_id_t msg_target;
+ struct lnet_process_id msg_target;
+ /* Primary NID of the source. */
+ lnet_nid_t msg_initiator;
/* where is it from, it's only for building event */
lnet_nid_t msg_from;
__u32 msg_type;
- /* commited for sending */
+ /*
+ * hold parameters in case message is with held due
+ * to discovery
+ */
+ lnet_nid_t msg_src_nid_param;
+ lnet_nid_t msg_rtr_nid_param;
+
+ /*
+ * Deadline for the message after which it will be finalized if it
+ * has not completed.
+ */
+ ktime_t msg_deadline;
+
+ /* The message health status. */
+ enum lnet_msg_hstatus msg_health_status;
+ /* This is a recovery message */
+ bool msg_recovery;
+ /* 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 */
+ bool msg_no_resend;
+
+ /* committed for sending */
unsigned int msg_tx_committed:1;
/* CPT # this message committed for sending */
unsigned int msg_tx_cpt:15;
- /* commited for receiving */
+ /* committed for receiving */
unsigned int msg_rx_committed:1;
/* CPT # this message committed for receiving */
unsigned int msg_rx_cpt:15;
/* ready for pending on RX delay list */
unsigned int msg_rx_ready_delay:1;
- unsigned int msg_vmflush:1; /* VM trying to free memory */
- unsigned int msg_target_is_router:1; /* sending to a router */
- unsigned int msg_routing:1; /* being forwarded */
- unsigned int msg_ack:1; /* ack on finalize (PUT) */
- unsigned int msg_sending:1; /* outgoing message */
- unsigned int msg_receiving:1; /* being received */
- unsigned int msg_txcredit:1; /* taken an NI send credit */
- unsigned int msg_peertxcredit:1; /* taken a peer send credit */
- unsigned int msg_rtrcredit:1; /* taken a globel router credit */
- unsigned int msg_peerrtrcredit:1; /* taken a peer router credit */
- unsigned int msg_onactivelist:1; /* on the activelist */
-
- struct lnet_peer *msg_txpeer; /* peer I'm sending to */
- struct lnet_peer *msg_rxpeer; /* peer I received from */
-
- void *msg_private;
- struct lnet_libmd *msg_md;
-
- unsigned int msg_len;
- unsigned int msg_wanted;
- unsigned int msg_offset;
- unsigned int msg_niov;
- struct iovec *msg_iov;
- lnet_kiov_t *msg_kiov;
-
- lnet_event_t msg_ev;
- lnet_hdr_t msg_hdr;
-} lnet_msg_t;
-
-
-typedef struct lnet_libhandle {
- cfs_list_t lh_hash_chain;
- __u64 lh_cookie;
-} lnet_libhandle_t;
+ unsigned int msg_vmflush:1; /* VM trying to free memory */
+ unsigned int msg_target_is_router:1; /* sending to a router */
+ unsigned int msg_routing:1; /* being forwarded */
+ unsigned int msg_ack:1; /* ack on finalize (PUT) */
+ unsigned int msg_sending:1; /* outgoing message */
+ unsigned int msg_receiving:1; /* being received */
+ unsigned int msg_txcredit:1; /* taken an NI send credit */
+ unsigned int msg_peertxcredit:1; /* taken a peer send credit */
+ unsigned int msg_rtrcredit:1; /* taken a globel router credit */
+ unsigned int msg_peerrtrcredit:1; /* taken a peer router credit */
+ unsigned int msg_onactivelist:1; /* on the activelist */
+ unsigned int msg_rdma_get:1;
+
+ struct lnet_peer_ni *msg_txpeer; /* peer I'm sending to */
+ struct lnet_peer_ni *msg_rxpeer; /* peer I received from */
+
+ void *msg_private;
+ struct lnet_libmd *msg_md;
+ /* the NI the message was sent or received over */
+ struct lnet_ni *msg_txni;
+ struct lnet_ni *msg_rxni;
+
+ unsigned int msg_len;
+ unsigned int msg_wanted;
+ unsigned int msg_offset;
+ unsigned int msg_niov;
+ struct kvec *msg_iov;
+ lnet_kiov_t *msg_kiov;
+
+ struct lnet_event msg_ev;
+ struct lnet_hdr msg_hdr;
+};
+
+struct lnet_libhandle {
+ struct list_head lh_hash_chain;
+ __u64 lh_cookie;
+};
#define lh_entry(ptr, type, member) \
- ((type *)((char *)(ptr)-(char *)(&((type *)0)->member)))
+ ((type *)((char *)(ptr)-(char *)(&((type *)0)->member)))
-typedef struct lnet_eq {
- cfs_list_t eq_list;
- lnet_libhandle_t eq_lh;
- lnet_seq_t eq_enq_seq;
- lnet_seq_t eq_deq_seq;
+struct lnet_eq {
+ struct list_head eq_list;
+ struct lnet_libhandle eq_lh;
+ unsigned long eq_enq_seq;
+ unsigned long eq_deq_seq;
unsigned int eq_size;
lnet_eq_handler_t eq_callback;
- lnet_event_t *eq_events;
+ struct lnet_event *eq_events;
int **eq_refs; /* percpt refcount for EQ */
-} lnet_eq_t;
-
-typedef struct lnet_me {
- cfs_list_t me_list;
- lnet_libhandle_t me_lh;
- lnet_process_id_t me_match_id;
- unsigned int me_portal;
- unsigned int me_pos; /* hash offset in mt_hash */
- __u64 me_match_bits;
- __u64 me_ignore_bits;
- lnet_unlink_t me_unlink;
- struct lnet_libmd *me_md;
-} lnet_me_t;
-
-typedef struct lnet_libmd {
- cfs_list_t md_list;
- lnet_libhandle_t md_lh;
- lnet_me_t *md_me;
- char *md_start;
- unsigned int md_offset;
- unsigned int md_length;
- unsigned int md_max_size;
- int md_threshold;
- int md_refcount;
- unsigned int md_options;
- unsigned int md_flags;
- void *md_user_ptr;
- lnet_eq_t *md_eq;
- unsigned int md_niov; /* # frags */
- union {
- struct iovec iov[LNET_MAX_IOV];
- lnet_kiov_t kiov[LNET_MAX_IOV];
- } md_iov;
-} lnet_libmd_t;
-
-#define LNET_MD_FLAG_ZOMBIE (1 << 0)
-#define LNET_MD_FLAG_AUTO_UNLINK (1 << 1)
-
-#ifdef LNET_USE_LIB_FREELIST
-typedef struct
-{
- void *fl_objs; /* single contiguous array of objects */
- int fl_nobjs; /* the number of them */
- int fl_objsize; /* the size (including overhead) of each of them */
- cfs_list_t fl_list; /* where they are enqueued */
-} lnet_freelist_t;
-
-typedef struct
-{
- cfs_list_t fo_list; /* enqueue on fl_list */
- void *fo_contents; /* aligned contents */
-} lnet_freeobj_t;
-#endif
+};
+
+struct lnet_me {
+ struct list_head me_list;
+ struct lnet_libhandle me_lh;
+ struct lnet_process_id me_match_id;
+ unsigned int me_portal;
+ unsigned int me_pos; /* hash offset in mt_hash */
+ __u64 me_match_bits;
+ __u64 me_ignore_bits;
+ enum lnet_unlink me_unlink;
+ struct lnet_libmd *me_md;
+};
+
+struct lnet_libmd {
+ struct list_head md_list;
+ struct lnet_libhandle md_lh;
+ struct lnet_me *md_me;
+ char *md_start;
+ unsigned int md_offset;
+ unsigned int md_length;
+ unsigned int md_max_size;
+ int md_threshold;
+ int md_refcount;
+ unsigned int md_options;
+ unsigned int md_flags;
+ unsigned int md_niov; /* # frags at end of struct */
+ void *md_user_ptr;
+ struct lnet_rsp_tracker *md_rspt_ptr;
+ struct lnet_eq *md_eq;
+ struct lnet_handle_md md_bulk_handle;
+ union {
+ struct kvec iov[LNET_MAX_IOV];
+ lnet_kiov_t kiov[LNET_MAX_IOV];
+ } md_iov;
+};
+
+#define LNET_MD_FLAG_ZOMBIE (1 << 0)
+#define LNET_MD_FLAG_AUTO_UNLINK (1 << 1)
+#define LNET_MD_FLAG_ABORTED (1 << 2)
-typedef struct {
- /* info about peers we are trying to fail */
- cfs_list_t tp_list; /* ln_test_peers */
- lnet_nid_t tp_nid; /* matching nid */
- unsigned int tp_threshold; /* # failures to simulate */
-} lnet_test_peer_t;
+struct lnet_test_peer {
+ /* info about peers we are trying to fail */
+ struct list_head tp_list; /* ln_test_peers */
+ lnet_nid_t tp_nid; /* matching nid */
+ unsigned int tp_threshold; /* # failures to simulate */
+};
#define LNET_COOKIE_TYPE_MD 1
#define LNET_COOKIE_TYPE_ME 2
#define LNET_COOKIE_TYPE_BITS 2
#define LNET_COOKIE_MASK ((1ULL << LNET_COOKIE_TYPE_BITS) - 1ULL)
-struct lnet_ni; /* forward ref */
-
-typedef struct lnet_lnd
-{
- /* fields managed by portals */
- cfs_list_t lnd_list; /* stash in the LND table */
- int lnd_refcount; /* # active instances */
-
- /* fields initialised by the LND */
- unsigned int lnd_type;
-
- int (*lnd_startup) (struct lnet_ni *ni);
- void (*lnd_shutdown) (struct lnet_ni *ni);
- int (*lnd_ctl)(struct lnet_ni *ni, unsigned int cmd, void *arg);
-
- /* In data movement APIs below, payload buffers are described as a set
- * of 'niov' fragments which are...
- * EITHER
- * in virtual memory (struct iovec *iov != NULL)
- * OR
- * in pages (kernel only: plt_kiov_t *kiov != NULL).
- * The LND may NOT overwrite these fragment descriptors.
- * An 'offset' and may specify a byte offset within the set of
- * fragments to start from
- */
-
- /* Start sending a preformatted message. 'private' is NULL for PUT and
- * GET messages; otherwise this is a response to an incoming message
- * and 'private' is the 'private' passed to lnet_parse(). Return
- * non-zero for immediate failure, otherwise complete later with
- * lnet_finalize() */
- int (*lnd_send)(struct lnet_ni *ni, void *private, lnet_msg_t *msg);
-
- /* Start receiving 'mlen' bytes of payload data, skipping the following
- * 'rlen' - 'mlen' bytes. 'private' is the 'private' passed to
- * lnet_parse(). Return non-zero for immedaite failure, otherwise
- * complete later with lnet_finalize(). This also gives back a receive
- * credit if the LND does flow control. */
- int (*lnd_recv)(struct lnet_ni *ni, void *private, lnet_msg_t *msg,
- int delayed, unsigned int niov,
- struct iovec *iov, lnet_kiov_t *kiov,
- unsigned int offset, unsigned int mlen, unsigned int rlen);
-
- /* lnet_parse() has had to delay processing of this message
- * (e.g. waiting for a forwarding buffer or send credits). Give the
- * LND a chance to free urgently needed resources. If called, return 0
- * for success and do NOT give back a receive credit; that has to wait
- * until lnd_recv() gets called. On failure return < 0 and
- * release resources; lnd_recv() will not be called. */
- int (*lnd_eager_recv)(struct lnet_ni *ni, void *private, lnet_msg_t *msg,
- void **new_privatep);
-
- /* notification of peer health */
- void (*lnd_notify)(struct lnet_ni *ni, lnet_nid_t peer, int alive);
-
- /* query of peer aliveness */
- void (*lnd_query)(struct lnet_ni *ni, lnet_nid_t peer, cfs_time_t *when);
-
-#if defined(__KERNEL__) || defined(HAVE_LIBPTHREAD)
- /* accept a new connection */
- int (*lnd_accept)(struct lnet_ni *ni, cfs_socket_t *sock);
-#endif
-
-#ifndef __KERNEL__
- /* wait for something to happen */
- void (*lnd_wait)(struct lnet_ni *ni, int milliseconds);
-
- /* ensure non-RDMA messages can be received outside liblustre */
- int (*lnd_setasync)(struct lnet_ni *ni, lnet_process_id_t id, int nasync);
-#endif
-} lnd_t;
-
-#define LNET_NI_STATUS_UP 0x15aac0de
-#define LNET_NI_STATUS_DOWN 0xdeadface
-#define LNET_NI_STATUS_INVALID 0x00000000
-typedef struct {
- lnet_nid_t ns_nid;
- __u32 ns_status;
- __u32 ns_unused;
-} WIRE_ATTR lnet_ni_status_t;
+struct lnet_ni; /* forward ref */
+struct socket;
+
+struct lnet_lnd {
+ /* fields managed by portals */
+ struct list_head lnd_list; /* stash in the LND table */
+ int lnd_refcount; /* # active instances */
+
+ /* fields initialized by the LND */
+ __u32 lnd_type;
+
+ int (*lnd_startup)(struct lnet_ni *ni);
+ void (*lnd_shutdown)(struct lnet_ni *ni);
+ int (*lnd_ctl)(struct lnet_ni *ni, unsigned int cmd, void *arg);
+
+ /* In data movement APIs below, payload buffers are described as a set
+ * of 'niov' fragments which are...
+ * EITHER
+ * in virtual memory (struct kvec *iov != NULL)
+ * OR
+ * in pages (kernel only: plt_kiov_t *kiov != NULL).
+ * The LND may NOT overwrite these fragment descriptors.
+ * An 'offset' and may specify a byte offset within the set of
+ * fragments to start from
+ */
+
+ /* Start sending a preformatted message. 'private' is NULL for PUT and
+ * GET messages; otherwise this is a response to an incoming message
+ * and 'private' is the 'private' passed to lnet_parse(). Return
+ * non-zero for immediate failure, otherwise complete later with
+ * lnet_finalize() */
+ int (*lnd_send)(struct lnet_ni *ni, void *private,
+ struct lnet_msg *msg);
+
+ /* Start receiving 'mlen' bytes of payload data, skipping the following
+ * 'rlen' - 'mlen' bytes. 'private' is the 'private' passed to
+ * lnet_parse(). Return non-zero for immedaite failure, otherwise
+ * complete later with lnet_finalize(). This also gives back a receive
+ * credit if the LND does flow control. */
+ int (*lnd_recv)(struct lnet_ni *ni, void *private, struct lnet_msg *msg,
+ int delayed, unsigned int niov,
+ struct kvec *iov, lnet_kiov_t *kiov,
+ unsigned int offset, unsigned int mlen, unsigned int rlen);
+
+ /* lnet_parse() has had to delay processing of this message
+ * (e.g. waiting for a forwarding buffer or send credits). Give the
+ * LND a chance to free urgently needed resources. If called, return 0
+ * for success and do NOT give back a receive credit; that has to wait
+ * until lnd_recv() gets called. On failure return < 0 and
+ * release resources; lnd_recv() will not be called. */
+ int (*lnd_eager_recv)(struct lnet_ni *ni, void *private,
+ struct lnet_msg *msg, void **new_privatep);
+
+ /* notification of peer down */
+ void (*lnd_notify_peer_down)(lnet_nid_t peer);
+
+ /* query of peer aliveness */
+ void (*lnd_query)(struct lnet_ni *ni, lnet_nid_t peer, time64_t *when);
+
+ /* accept a new connection */
+ int (*lnd_accept)(struct lnet_ni *ni, struct socket *sock);
+};
struct lnet_tx_queue {
int tq_credits; /* # tx credits free */
int tq_credits_min; /* lowest it's been */
int tq_credits_max; /* total # tx credits */
- cfs_list_t tq_delayed; /* delayed TXs */
+ struct list_head tq_delayed; /* delayed TXs */
};
-#define LNET_MAX_INTERFACES 16
+enum lnet_net_state {
+ /* set when net block is allocated */
+ LNET_NET_STATE_INIT = 0,
+ /* set when NIs in net are started successfully */
+ LNET_NET_STATE_ACTIVE,
+ /* set if all NIs in net are in FAILED state */
+ LNET_NET_STATE_INACTIVE,
+ /* set when shutting down a NET */
+ LNET_NET_STATE_DELETING
+};
+
+enum lnet_ni_state {
+ /* initial state when NI is created */
+ LNET_NI_STATE_INIT = 0,
+ /* set when NI is brought up */
+ LNET_NI_STATE_ACTIVE,
+ /* set when NI is being shutdown */
+ LNET_NI_STATE_DELETING,
+};
+
+#define LNET_NI_RECOVERY_PENDING BIT(0)
+#define LNET_NI_RECOVERY_FAILED BIT(1)
+
+enum lnet_stats_type {
+ LNET_STATS_TYPE_SEND = 0,
+ LNET_STATS_TYPE_RECV,
+ LNET_STATS_TYPE_DROP
+};
+
+struct lnet_comm_count {
+ atomic_t co_get_count;
+ atomic_t co_put_count;
+ atomic_t co_reply_count;
+ atomic_t co_ack_count;
+ atomic_t co_hello_count;
+};
+
+struct lnet_element_stats {
+ struct lnet_comm_count el_send_stats;
+ struct lnet_comm_count el_recv_stats;
+ struct lnet_comm_count el_drop_stats;
+};
+
+struct lnet_health_local_stats {
+ atomic_t hlt_local_interrupt;
+ atomic_t hlt_local_dropped;
+ atomic_t hlt_local_aborted;
+ atomic_t hlt_local_no_route;
+ atomic_t hlt_local_timeout;
+ atomic_t hlt_local_error;
+};
+
+struct lnet_health_remote_stats {
+ atomic_t hlt_remote_dropped;
+ atomic_t hlt_remote_timeout;
+ atomic_t hlt_remote_error;
+ atomic_t hlt_network_timeout;
+};
+
+struct lnet_net {
+ /* chain on the ln_nets */
+ struct list_head net_list;
+
+ /* net ID, which is composed of
+ * (net_type << 16) | net_num.
+ * net_type can be one of the enumerated types defined in
+ * lnet/include/lnet/nidstr.h */
+ __u32 net_id;
+
+ /* priority of the network */
+ __u32 net_prio;
+
+ /* total number of CPTs in the array */
+ __u32 net_ncpts;
+
+ /* cumulative CPTs of all NIs in this net */
+ __u32 *net_cpts;
+
+ /* network tunables */
+ struct lnet_ioctl_config_lnd_cmn_tunables net_tunables;
+
+ /*
+ * boolean to indicate that the tunables have been set and
+ * shouldn't be reset
+ */
+ bool net_tunables_set;
+
+ /* procedural interface */
+ struct lnet_lnd *net_lnd;
+
+ /* list of NIs on this net */
+ struct list_head net_ni_list;
+
+ /* list of NIs being added, but not started yet */
+ struct list_head net_ni_added;
+
+ /* dying LND instances */
+ struct list_head net_ni_zombie;
+
+ /* network state */
+ enum lnet_net_state net_state;
+};
+
+struct lnet_ni {
+ /* chain on the lnet_net structure */
+ struct list_head ni_netlist;
+
+ /* chain on the recovery queue */
+ struct list_head ni_recovery;
+
+ /* MD handle for recovery ping */
+ struct lnet_handle_md ni_ping_mdh;
-typedef struct lnet_ni {
-#ifdef __KERNEL__
spinlock_t ni_lock;
-#else
-# ifndef HAVE_LIBPTHREAD
- int ni_lock;
-# else
- pthread_mutex_t ni_lock;
-# endif
-#endif
- cfs_list_t ni_list; /* chain on ln_nis */
- cfs_list_t ni_cptlist; /* chain on ln_nis_cpt */
- int ni_maxtxcredits; /* # tx credits */
- /* # per-peer send credits */
- int ni_peertxcredits;
- /* # per-peer router buffer credits */
- int ni_peerrtrcredits;
- /* seconds to consider peer dead */
- int ni_peertimeout;
- int ni_ncpts; /* number of CPTs */
- __u32 *ni_cpts; /* bond NI on some CPTs */
- lnet_nid_t ni_nid; /* interface's NID */
- void *ni_data; /* instance-specific data */
- lnd_t *ni_lnd; /* procedural interface */
- struct lnet_tx_queue **ni_tx_queues; /* percpt TX queues */
- int **ni_refs; /* percpt reference count */
- long ni_last_alive; /* when I was last alive */
- lnet_ni_status_t *ni_status; /* my health status */
- /* equivalent interfaces to use */
- char *ni_interfaces[LNET_MAX_INTERFACES];
-} lnet_ni_t;
+
+ /* number of CPTs */
+ int ni_ncpts;
+
+ /* bond NI on some CPTs */
+ __u32 *ni_cpts;
+
+ /* interface's NID */
+ lnet_nid_t ni_nid;
+
+ /* instance-specific data */
+ void *ni_data;
+
+ /* per ni credits */
+ atomic_t ni_tx_credits;
+
+ /* percpt TX queues */
+ struct lnet_tx_queue **ni_tx_queues;
+
+ /* percpt reference count */
+ int **ni_refs;
+
+ /* when I was last alive */
+ time64_t ni_last_alive;
+
+ /* pointer to parent network */
+ struct lnet_net *ni_net;
+
+ /* my health status */
+ struct lnet_ni_status *ni_status;
+
+ /* NI FSM. Protected by lnet_ni_lock() */
+ enum lnet_ni_state ni_state;
+
+ /* Recovery state. Protected by lnet_ni_lock() */
+ __u32 ni_recovery_state;
+
+ /* per NI LND tunables */
+ struct lnet_lnd_tunables ni_lnd_tunables;
+
+ /* lnd tunables set explicitly */
+ bool ni_lnd_tunables_set;
+
+ /* NI statistics */
+ struct lnet_element_stats ni_stats;
+ struct lnet_health_local_stats ni_hstats;
+
+ /* physical device CPT */
+ int ni_dev_cpt;
+
+ /* sequence number used to round robin over nis within a net */
+ __u32 ni_seq;
+
+ /*
+ * health value
+ * initialized to LNET_MAX_HEALTH_VALUE
+ * Value is decremented every time we fail to send a message over
+ * this NI because of a NI specific failure.
+ * Value is incremented if we successfully send a message.
+ */
+ atomic_t ni_healthv;
+
+ /*
+ * Set to 1 by the LND when it receives an event telling it the device
+ * has gone into a fatal state. Set to 0 when the LND receives an
+ * even telling it the device is back online.
+ */
+ atomic_t ni_fatal_error_on;
+
+ /*
+ * equivalent interfaces to use
+ * This is an array because socklnd bonding can still be configured
+ */
+ char *ni_interfaces[LNET_INTERFACES_NUM];
+ struct net *ni_net_ns; /* original net namespace */
+};
#define LNET_PROTO_PING_MATCHBITS 0x8000000000000000LL
-/* NB: value of these features equal to LNET_PROTO_PING_VERSION_x
- * of old LNet, so there shouldn't be any compatibility issue */
-#define LNET_PING_FEAT_INVAL (0) /* no feature */
-#define LNET_PING_FEAT_BASE (1 << 0) /* just a ping */
-#define LNET_PING_FEAT_NI_STATUS (1 << 1) /* return NI status */
-
-#define LNET_PING_FEAT_MASK (LNET_PING_FEAT_BASE | \
- LNET_PING_FEAT_NI_STATUS)
-
-typedef struct {
- __u32 pi_magic;
- __u32 pi_features;
- lnet_pid_t pi_pid;
- __u32 pi_nnis;
- lnet_ni_status_t pi_ni[0];
-} WIRE_ATTR lnet_ping_info_t;
-
-/* router checker data, per router */
-#define LNET_MAX_RTR_NIS 16
-#define LNET_PINGINFO_SIZE offsetof(lnet_ping_info_t, pi_ni[LNET_MAX_RTR_NIS])
-typedef struct {
- /* chain on the_lnet.ln_zombie_rcd or ln_deathrow_rcd */
- cfs_list_t rcd_list;
- lnet_handle_md_t rcd_mdh; /* ping buffer MD */
- struct lnet_peer *rcd_gateway; /* reference to gateway */
- lnet_ping_info_t *rcd_pinginfo; /* ping buffer */
-} lnet_rc_data_t;
-
-typedef struct lnet_peer {
- cfs_list_t lp_hashlist; /* chain on peer hash */
- cfs_list_t lp_txq; /* messages blocking for tx credits */
- cfs_list_t lp_rtrq; /* messages blocking for router credits */
- cfs_list_t lp_rtr_list; /* chain on router list */
- int lp_txcredits; /* # tx credits available */
- int lp_mintxcredits; /* low water mark */
- int lp_rtrcredits; /* # router credits */
- int lp_minrtrcredits; /* low water mark */
- unsigned int lp_alive:1; /* alive/dead? */
- unsigned int lp_notify:1; /* notification outstanding? */
- unsigned int lp_notifylnd:1; /* outstanding notification for LND? */
- unsigned int lp_notifying:1; /* some thread is handling notification */
- unsigned int lp_ping_notsent; /* SEND event outstanding from ping */
- int lp_alive_count; /* # times router went dead<->alive */
- long lp_txqnob; /* bytes queued for sending */
- cfs_time_t lp_timestamp; /* time of last aliveness news */
- cfs_time_t lp_ping_timestamp; /* time of last ping attempt */
- cfs_time_t lp_ping_deadline; /* != 0 if ping reply expected */
- cfs_time_t lp_last_alive; /* when I was last alive */
- cfs_time_t lp_last_query; /* when lp_ni was queried last time */
- lnet_ni_t *lp_ni; /* interface peer is on */
- lnet_nid_t lp_nid; /* peer's NID */
- int lp_refcount; /* # refs */
- int lp_cpt; /* CPT this peer attached on */
+/*
+ * Descriptor of a ping info buffer: keep a separate indicator of the
+ * size and a reference count. The type is used both as a source and
+ * sink of data, so we need to keep some information outside of the
+ * area that may be overwritten by network data.
+ */
+struct lnet_ping_buffer {
+ int pb_nnis;
+ atomic_t pb_refcnt;
+ struct lnet_ping_info pb_info;
+};
+
+#define LNET_PING_BUFFER_SIZE(NNIDS) \
+ offsetof(struct lnet_ping_buffer, pb_info.pi_ni[NNIDS])
+#define LNET_PING_BUFFER_LONI(PBUF) ((PBUF)->pb_info.pi_ni[0].ns_nid)
+#define LNET_PING_BUFFER_SEQNO(PBUF) ((PBUF)->pb_info.pi_ni[0].ns_status)
+
+#define LNET_PING_INFO_TO_BUFFER(PINFO) \
+ container_of((PINFO), struct lnet_ping_buffer, pb_info)
+
+struct lnet_peer_ni {
+ /* chain on lpn_peer_nis */
+ struct list_head lpni_peer_nis;
+ /* chain on remote peer list */
+ struct list_head lpni_on_remote_peer_ni_list;
+ /* chain on recovery queue */
+ struct list_head lpni_recovery;
+ /* chain on peer hash */
+ struct list_head lpni_hashlist;
+ /* messages blocking for tx credits */
+ struct list_head lpni_txq;
+ /* pointer to peer net I'm part of */
+ struct lnet_peer_net *lpni_peer_net;
+ /* statistics kept on each peer NI */
+ struct lnet_element_stats lpni_stats;
+ struct lnet_health_remote_stats lpni_hstats;
+ /* spin lock protecting credits and lpni_txq */
+ spinlock_t lpni_lock;
+ /* # tx credits available */
+ int lpni_txcredits;
+ /* low water mark */
+ int lpni_mintxcredits;
+ /*
+ * Each peer_ni in a gateway maintains its own credits. This
+ * allows more traffic to gateways that have multiple interfaces.
+ */
+ /* # router credits */
+ int lpni_rtrcredits;
+ /* low water mark */
+ int lpni_minrtrcredits;
+ /* bytes queued for sending */
+ long lpni_txqnob;
+ /* network peer is on */
+ struct lnet_net *lpni_net;
+ /* peer's NID */
+ lnet_nid_t lpni_nid;
+ /* # refs */
+ atomic_t lpni_refcount;
+ /* health value for the peer */
+ atomic_t lpni_healthv;
+ /* recovery ping mdh */
+ struct lnet_handle_md lpni_recovery_ping_mdh;
+ /* CPT this peer attached on */
+ int lpni_cpt;
+ /* state flags -- protected by lpni_lock */
+ unsigned lpni_state;
+ /* status of the peer NI as reported by the peer */
+ __u32 lpni_ns_status;
+ /* sequence number used to round robin over peer nis within a net */
+ __u32 lpni_seq;
+ /* sequence number used to round robin over gateways */
+ __u32 lpni_gw_seq;
+ /* returned RC ping features. Protected with lpni_lock */
+ unsigned int lpni_ping_feats;
+ /* preferred local nids: if only one, use lpni_pref.nid */
+ union lpni_pref {
+ lnet_nid_t nid;
+ lnet_nid_t *nids;
+ } lpni_pref;
+ /* number of preferred NIDs in lnpi_pref_nids */
+ __u32 lpni_pref_nnids;
+};
+
+/* Preferred path added due to traffic on non-MR peer_ni */
+#define LNET_PEER_NI_NON_MR_PREF (1 << 0)
+/* peer is being recovered. */
+#define LNET_PEER_NI_RECOVERY_PENDING (1 << 1)
+/* recovery ping failed */
+#define LNET_PEER_NI_RECOVERY_FAILED (1 << 2)
+/* peer is being deleted */
+#define LNET_PEER_NI_DELETING (1 << 3)
+
+struct lnet_peer {
+ /* chain on pt_peer_list */
+ struct list_head lp_peer_list;
+
+ /* list of peer nets */
+ struct list_head lp_peer_nets;
+
+ /* list of messages pending discovery*/
+ struct list_head lp_dc_pendq;
+
+ /* chain on router list */
+ struct list_head lp_rtr_list;
+
+ /* primary NID of the peer */
+ lnet_nid_t lp_primary_nid;
+
+ /* CPT of peer_table */
+ int lp_cpt;
+
+ /* number of NIDs on this peer */
+ int lp_nnis;
+
/* # refs from lnet_route_t::lr_gateway */
int lp_rtr_refcount;
- /* returned RC ping features */
- unsigned int lp_ping_feats;
- cfs_list_t lp_routes; /* routers on this peer */
- lnet_rc_data_t *lp_rcd; /* router checker state */
-} lnet_peer_t;
+ /* messages blocking for router credits */
+ struct list_head lp_rtrq;
+
+ /* routes on this peer */
+ struct list_head lp_routes;
+
+ /* time of last router check attempt */
+ time64_t lp_rtrcheck_timestamp;
+
+ /* reference count */
+ atomic_t lp_refcount;
+
+ /* lock protecting peer state flags and lpni_rtrq */
+ spinlock_t lp_lock;
+
+ /* peer state flags */
+ unsigned lp_state;
+
+ /* buffer for data pushed by peer */
+ struct lnet_ping_buffer *lp_data;
+
+ /* MD handle for ping in progress */
+ struct lnet_handle_md lp_ping_mdh;
+
+ /* MD handle for push in progress */
+ struct lnet_handle_md lp_push_mdh;
+
+ /* number of NIDs for sizing push data */
+ int lp_data_nnis;
+
+ /* NI config sequence number of peer */
+ __u32 lp_peer_seqno;
+
+ /* Local NI config sequence number acked by peer */
+ __u32 lp_node_seqno;
+
+ /* Local NI config sequence number sent to peer */
+ __u32 lp_node_seqno_sent;
+
+ /* Ping error encountered during discovery. */
+ int lp_ping_error;
+
+ /* Push error encountered during discovery. */
+ int lp_push_error;
+
+ /* Error encountered during discovery. */
+ int lp_dc_error;
+
+ /* time it was put on the ln_dc_working queue */
+ time64_t lp_last_queued;
+
+ /* link on discovery-related lists */
+ struct list_head lp_dc_list;
+
+ /* tasks waiting on discovery of this peer */
+ wait_queue_head_t lp_dc_waitq;
+};
+
+/*
+ * The status flags in lp_state. Their semantics have chosen so that
+ * lp_state can be zero-initialized.
+ *
+ * A peer is marked MULTI_RAIL in two cases: it was configured using DLC
+ * as multi-rail aware, or the LNET_PING_FEAT_MULTI_RAIL bit was set.
+ *
+ * A peer is marked NO_DISCOVERY if the LNET_PING_FEAT_DISCOVERY bit was
+ * NOT set when the peer was pinged by discovery.
+ *
+ * A peer is marked ROUTER if it indicates so in the feature bit.
+ */
+#define LNET_PEER_MULTI_RAIL (1 << 0) /* Multi-rail aware */
+#define LNET_PEER_NO_DISCOVERY (1 << 1) /* Peer disabled discovery */
+#define LNET_PEER_ROUTER_ENABLED (1 << 2) /* router feature enabled */
+
+/*
+ * A peer is marked CONFIGURED if it was configured by DLC.
+ *
+ * In addition, a peer is marked DISCOVERED if it has fully passed
+ * through Peer Discovery.
+ *
+ * When Peer Discovery is disabled, the discovery thread will mark
+ * peers REDISCOVER to indicate that they should be re-examined if
+ * discovery is (re)enabled on the node.
+ *
+ * A peer that was created as the result of inbound traffic will not
+ * be marked at all.
+ */
+#define LNET_PEER_CONFIGURED (1 << 3) /* Configured via DLC */
+#define LNET_PEER_DISCOVERED (1 << 4) /* Peer was discovered */
+#define LNET_PEER_REDISCOVER (1 << 5) /* Discovery was disabled */
+/*
+ * A peer is marked DISCOVERING when discovery is in progress.
+ * The other flags below correspond to stages of discovery.
+ */
+#define LNET_PEER_DISCOVERING (1 << 6) /* Discovering */
+#define LNET_PEER_DATA_PRESENT (1 << 7) /* Remote peer data present */
+#define LNET_PEER_NIDS_UPTODATE (1 << 8) /* Remote peer info uptodate */
+#define LNET_PEER_PING_SENT (1 << 9) /* Waiting for REPLY to Ping */
+#define LNET_PEER_PUSH_SENT (1 << 10) /* Waiting for ACK of Push */
+#define LNET_PEER_PING_FAILED (1 << 11) /* Ping send failure */
+#define LNET_PEER_PUSH_FAILED (1 << 12) /* Push send failure */
+/*
+ * A ping can be forced as a way to fix up state, or as a manual
+ * intervention by an admin.
+ * A push can be forced in circumstances that would normally not
+ * allow for one to happen.
+ */
+#define LNET_PEER_FORCE_PING (1 << 13) /* Forced Ping */
+#define LNET_PEER_FORCE_PUSH (1 << 14) /* Forced Push */
+
+struct lnet_peer_net {
+ /* chain on lp_peer_nets */
+ struct list_head lpn_peer_nets;
+
+ /* list of peer_nis on this network */
+ struct list_head lpn_peer_nis;
+
+ /* pointer to the peer I'm part of */
+ struct lnet_peer *lpn_peer;
+
+ /* Net ID */
+ __u32 lpn_net_id;
+
+ /* reference count */
+ atomic_t lpn_refcount;
+};
/* peer hash size */
-#define LNET_PEER_HASH_BITS 9
-#define LNET_PEER_HASH_SIZE (1 << LNET_PEER_HASH_BITS)
+#define LNET_PEER_HASH_BITS 9
+#define LNET_PEER_HASH_SIZE (1 << LNET_PEER_HASH_BITS)
-/* peer hash table */
+/*
+ * peer hash table - one per CPT
+ *
+ * protected by lnet_net_lock/EX for update
+ * pt_version
+ * pt_number
+ * pt_hash[...]
+ * pt_peer_list
+ * pt_peers
+ * protected by pt_zombie_lock:
+ * pt_zombie_list
+ * pt_zombies
+ *
+ * pt_zombie lock nests inside lnet_net_lock
+ */
struct lnet_peer_table {
int pt_version; /* /proc validity stamp */
- int pt_number; /* # peers extant */
- cfs_list_t pt_deathrow; /* zombie peers */
- cfs_list_t *pt_hash; /* NID->peer hash */
+ int pt_number; /* # peers_ni extant */
+ struct list_head *pt_hash; /* NID->peer hash */
+ struct list_head pt_peer_list; /* peers */
+ int pt_peers; /* # peers */
+ struct list_head pt_zombie_list; /* zombie peer_ni */
+ int pt_zombies; /* # zombie peers_ni */
+ spinlock_t pt_zombie_lock; /* protect list and count */
};
/* peer aliveness is enabled only on routers for peers in a network where the
- * lnet_ni_t::ni_peertimeout has been set to a positive value */
+ * struct lnet_ni::ni_peertimeout has been set to a positive value
+ */
#define lnet_peer_aliveness_enabled(lp) (the_lnet.ln_routing != 0 && \
- (lp)->lp_ni->ni_peertimeout > 0)
+ ((lp)->lpni_net) && \
+ (lp)->lpni_net->net_tunables.lct_peer_timeout > 0)
-typedef struct {
- cfs_list_t lr_list; /* chain on net */
- cfs_list_t lr_gwlist; /* chain on gateway */
- lnet_peer_t *lr_gateway; /* router node */
+struct lnet_route {
+ struct list_head lr_list; /* chain on net */
+ struct list_head lr_gwlist; /* chain on gateway */
+ struct lnet_peer *lr_gateway; /* router node */
__u32 lr_net; /* remote network number */
+ __u32 lr_lnet; /* local network number */
int lr_seq; /* sequence for round-robin */
- unsigned int lr_downis; /* number of down NIs */
- unsigned int lr_hops; /* how far I am */
-} lnet_route_t;
+ __u32 lr_hops; /* how far I am */
+ unsigned int lr_priority; /* route priority */
+};
#define LNET_REMOTE_NETS_HASH_DEFAULT (1U << 7)
#define LNET_REMOTE_NETS_HASH_MAX (1U << 16)
#define LNET_REMOTE_NETS_HASH_SIZE (1 << the_lnet.ln_remote_nets_hbits)
-typedef struct {
- cfs_list_t lrn_list; /* chain on ln_remote_nets_hash */
- cfs_list_t lrn_routes; /* routes to me */
- __u32 lrn_net; /* my net number */
-} lnet_remotenet_t;
-
-typedef struct {
- cfs_list_t rbp_bufs; /* my free buffer pool */
- cfs_list_t rbp_msgs; /* messages blocking for a buffer */
- int rbp_npages; /* # pages in each buffer */
- int rbp_nbuffers; /* # buffers */
- int rbp_credits; /* # free buffers / blocked messages */
- int rbp_mincredits; /* low water mark */
-} lnet_rtrbufpool_t;
-
-typedef struct {
- cfs_list_t rb_list; /* chain on rbp_bufs */
- lnet_rtrbufpool_t *rb_pool; /* owning pool */
- lnet_kiov_t rb_kiov[0]; /* the buffer space */
-} lnet_rtrbuf_t;
-
-typedef struct {
- __u32 msgs_alloc;
- __u32 msgs_max;
- __u32 errors;
- __u32 send_count;
- __u32 recv_count;
- __u32 route_count;
- __u32 drop_count;
- __u64 send_length;
- __u64 recv_length;
- __u64 route_length;
- __u64 drop_length;
-} WIRE_ATTR lnet_counters_t;
-
-#define LNET_PEER_HASHSIZE 503 /* prime! */
-
-#define LNET_NRBPOOLS 3 /* # different router buffer pools */
-
-enum {
+struct lnet_remotenet {
+ /* chain on ln_remote_nets_hash */
+ struct list_head lrn_list;
+ /* routes to me */
+ struct list_head lrn_routes;
+ /* my net number */
+ __u32 lrn_net;
+};
+
+/** lnet message has credit and can be submitted to lnd for send/receive */
+#define LNET_CREDIT_OK 0
+/** lnet message is waiting for credit */
+#define LNET_CREDIT_WAIT 1
+/** lnet message is waiting for discovery */
+#define LNET_DC_WAIT 2
+
+struct lnet_rtrbufpool {
+ /* my free buffer pool */
+ struct list_head rbp_bufs;
+ /* messages blocking for a buffer */
+ struct list_head rbp_msgs;
+ /* # pages in each buffer */
+ int rbp_npages;
+ /* requested number of buffers */
+ int rbp_req_nbuffers;
+ /* # buffers actually allocated */
+ int rbp_nbuffers;
+ /* # free buffers / blocked messages */
+ int rbp_credits;
+ /* low water mark */
+ int rbp_mincredits;
+};
+
+struct lnet_rtrbuf {
+ struct list_head rb_list; /* chain on rbp_bufs */
+ struct lnet_rtrbufpool *rb_pool; /* owning pool */
+ lnet_kiov_t rb_kiov[0]; /* the buffer space */
+};
+
+#define LNET_PEER_HASHSIZE 503 /* prime! */
+
+enum lnet_match_flags {
/* Didn't match anything */
LNET_MATCHMD_NONE = (1 << 0),
/* Matched OK */
/* Must be discarded */
LNET_MATCHMD_DROP = (1 << 2),
/* match and buffer is exhausted */
- LNET_MATCHMD_EXHAUSTED = (1 << 3),
+ LNET_MATCHMD_EXHAUSTED = (1 << 3),
/* match or drop */
- LNET_MATCHMD_FINISH = (LNET_MATCHMD_OK | LNET_MATCHMD_DROP),
+ LNET_MATCHMD_FINISH = (LNET_MATCHMD_OK | LNET_MATCHMD_DROP),
};
-/* Options for lnet_portal_t::ptl_options */
-#define LNET_PTL_LAZY (1 << 0)
-#define LNET_PTL_MATCH_UNIQUE (1 << 1) /* unique match, for RDMA */
-#define LNET_PTL_MATCH_WILDCARD (1 << 2) /* wildcard match, request portal */
+/* Options for struct lnet_portal::ptl_options */
+#define LNET_PTL_LAZY (1 << 0)
+#define LNET_PTL_MATCH_UNIQUE (1 << 1) /* unique match, for RDMA */
+#define LNET_PTL_MATCH_WILDCARD (1 << 2) /* wildcard match, request portal */
/* parameter for matching operations (GET, PUT) */
struct lnet_match_info {
__u64 mi_mbits;
- lnet_process_id_t mi_id;
+ struct lnet_process_id mi_id;
+ unsigned int mi_cpt;
unsigned int mi_opc;
unsigned int mi_portal;
unsigned int mi_rlength;
struct lnet_match_table {
/* reserved for upcoming patches, CPU partition ID */
unsigned int mt_cpt;
- unsigned int mt_portal; /* portal index */
+ unsigned int mt_portal; /* portal index */
/* match table is set as "enabled" if there's non-exhausted MD
- * attached on mt_mhash, it's only valide for wildcard portal */
+ * attached on mt_mhash, it's only valid for wildcard portal */
unsigned int mt_enabled;
/* bitmap to flag whether MEs on mt_hash are exhausted or not */
__u64 mt_exhausted[LNET_MT_EXHAUSTED_BMAP];
- cfs_list_t *mt_mhash; /* matching hash */
+ struct list_head *mt_mhash; /* matching hash */
};
/* these are only useful for wildcard portal */
/* dispatch routed PUT message by hashing source NID for wildcard portals */
#define LNET_PTL_ROTOR_HASH_RT 3
-typedef struct lnet_portal {
-#ifdef __KERNEL__
+struct lnet_portal {
spinlock_t ptl_lock;
-#else
-# ifndef HAVE_LIBPTHREAD
- int ptl_lock;
-# else
- pthread_mutex_t ptl_lock;
-# endif
-#endif
unsigned int ptl_index; /* portal ID, reserved */
/* flags on this portal: lazy, unique... */
unsigned int ptl_options;
- /* list of messags which are stealing buffer */
- cfs_list_t ptl_msg_stealing;
+ /* list of messages which are stealing buffer */
+ struct list_head ptl_msg_stealing;
/* messages blocking for MD */
- cfs_list_t ptl_msg_delayed;
+ struct list_head ptl_msg_delayed;
/* Match table for each CPT */
struct lnet_match_table **ptl_mtables;
/* spread rotor of incoming "PUT" */
- int ptl_rotor;
+ unsigned int ptl_rotor;
/* # active entries for this portal */
- int ptl_mt_nmaps;
+ int ptl_mt_nmaps;
/* array of active entries' cpu-partition-id */
- int ptl_mt_maps[0];
-} lnet_portal_t;
+ int ptl_mt_maps[0];
+};
#define LNET_LH_HASH_BITS 12
#define LNET_LH_HASH_SIZE (1ULL << LNET_LH_HASH_BITS)
struct lnet_res_container {
unsigned int rec_type; /* container type */
__u64 rec_lh_cookie; /* cookie generator */
- cfs_list_t rec_active; /* active resource list */
- cfs_list_t *rec_lh_hash; /* handle hash */
-#ifdef LNET_USE_LIB_FREELIST
- lnet_freelist_t rec_freelist; /* freelist for resources */
-#endif
+ struct list_head rec_active; /* active resource list */
+ struct list_head *rec_lh_hash; /* handle hash */
};
/* message container */
/* max # threads finalizing */
int msc_nfinalizers;
/* msgs waiting to complete finalizing */
- cfs_list_t msc_finalizing;
- cfs_list_t msc_active; /* active message list */
+ struct list_head msc_finalizing;
+ struct list_head msc_active; /* active message list */
/* threads doing finalization */
void **msc_finalizers;
-#ifdef LNET_USE_LIB_FREELIST
- lnet_freelist_t msc_freelist; /* freelist for messages */
-#endif
};
+/* Peer Discovery states */
+#define LNET_DC_STATE_SHUTDOWN 0 /* not started */
+#define LNET_DC_STATE_RUNNING 1 /* started up OK */
+#define LNET_DC_STATE_STOPPING 2 /* telling thread to stop */
+
/* Router Checker states */
-#define LNET_RC_STATE_SHUTDOWN 0 /* not started */
-#define LNET_RC_STATE_RUNNING 1 /* started up OK */
-#define LNET_RC_STATE_STOPPING 2 /* telling thread to stop */
+#define LNET_MT_STATE_SHUTDOWN 0 /* not started */
+#define LNET_MT_STATE_RUNNING 1 /* started up OK */
+#define LNET_MT_STATE_STOPPING 2 /* telling thread to stop */
-typedef struct
-{
+/* LNet states */
+#define LNET_STATE_SHUTDOWN 0 /* not started */
+#define LNET_STATE_RUNNING 1 /* started up OK */
+#define LNET_STATE_STOPPING 2 /* telling thread to stop */
+
+struct lnet {
/* CPU partition table of LNet */
struct cfs_cpt_table *ln_cpt_table;
/* number of CPTs in ln_cpt_table */
/* # portals */
int ln_nportals;
/* the vector of portals */
- lnet_portal_t **ln_portals;
+ struct lnet_portal **ln_portals;
/* percpt ME containers */
struct lnet_res_container **ln_me_containers;
/* percpt MD container */
/* Event Queue container */
struct lnet_res_container ln_eq_container;
-#ifdef __KERNEL__
- cfs_waitq_t ln_eq_waitq;
+ wait_queue_head_t ln_eq_waitq;
spinlock_t ln_eq_wait_lock;
-#else
-# ifndef HAVE_LIBPTHREAD
- int ln_eq_wait_lock;
-# else
- pthread_cond_t ln_eq_cond;
- pthread_mutex_t ln_eq_wait_lock;
-# endif
-#endif
+
unsigned int ln_remote_nets_hbits;
/* protect NI, peer table, credits, routers, rtrbuf... */
struct cfs_percpt_lock *ln_net_lock;
/* percpt message containers for active/finalizing/freed message */
struct lnet_msg_container **ln_msg_containers;
- lnet_counters_t **ln_counters;
+ struct lnet_counters **ln_counters;
struct lnet_peer_table **ln_peer_tables;
+ /* list of peer nis not on a local network */
+ struct list_head ln_remote_peer_ni_list;
/* failure simulation */
- cfs_list_t ln_test_peers;
-
- cfs_list_t ln_nis; /* LND instances */
- /* NIs bond on specific CPT(s) */
- cfs_list_t ln_nis_cpt;
- /* dying LND instances */
- cfs_list_t ln_nis_zombie;
- lnet_ni_t *ln_loni; /* the loopback NI */
- /* NI to wait for events in */
- lnet_ni_t *ln_eq_waitni;
+ struct list_head ln_test_peers;
+ struct list_head ln_drop_rules;
+ struct list_head ln_delay_rules;
+ /* LND instances */
+ struct list_head ln_nets;
+ /* the loopback NI */
+ struct lnet_ni *ln_loni;
+ /* network zombie list */
+ struct list_head ln_net_zombie;
+ /* resend messages list */
+ struct list_head ln_msg_resend;
+ /* spin lock to protect the msg resend list */
+ spinlock_t ln_msg_resend_lock;
/* remote networks with routes to them */
- cfs_list_t *ln_remote_nets_hash;
+ struct list_head *ln_remote_nets_hash;
/* validity stamp */
__u64 ln_remote_nets_version;
/* list of all known routers */
- cfs_list_t ln_routers;
+ struct list_head ln_routers;
/* validity stamp */
__u64 ln_routers_version;
/* percpt router buffer pools */
- lnet_rtrbufpool_t **ln_rtrpools;
-
- lnet_handle_md_t ln_ping_target_md;
- lnet_handle_eq_t ln_ping_target_eq;
- lnet_ping_info_t *ln_ping_info;
-
- /* router checker startup/shutdown state */
- int ln_rc_state;
- /* router checker's event queue */
- lnet_handle_eq_t ln_rc_eqh;
- /* rcd still pending on net */
- cfs_list_t ln_rcd_deathrow;
- /* rcd ready for free */
- cfs_list_t ln_rcd_zombie;
-#ifdef __KERNEL__
+ struct lnet_rtrbufpool **ln_rtrpools;
+
+ /*
+ * Ping target / Push source
+ *
+ * The ping target and push source share a single buffer. The
+ * ln_ping_target is protected against concurrent updates by
+ * ln_api_mutex.
+ */
+ struct lnet_handle_md ln_ping_target_md;
+ struct lnet_handle_eq ln_ping_target_eq;
+ struct lnet_ping_buffer *ln_ping_target;
+ atomic_t ln_ping_target_seqno;
+
+ /*
+ * Push Target
+ *
+ * ln_push_nnis contains the desired size of the push target.
+ * The lnet_net_lock is used to handle update races. The old
+ * buffer may linger a while after it has been unlinked, in
+ * which case the event handler cleans up.
+ */
+ struct lnet_handle_eq ln_push_target_eq;
+ struct lnet_handle_md ln_push_target_md;
+ struct lnet_ping_buffer *ln_push_target;
+ int ln_push_target_nnis;
+
+ /* discovery event queue handle */
+ struct lnet_handle_eq ln_dc_eqh;
+ /* discovery requests */
+ struct list_head ln_dc_request;
+ /* discovery working list */
+ struct list_head ln_dc_working;
+ /* discovery expired list */
+ struct list_head ln_dc_expired;
+ /* discovery thread wait queue */
+ wait_queue_head_t ln_dc_waitq;
+ /* discovery startup/shutdown state */
+ int ln_dc_state;
+
+ /* monitor thread startup/shutdown state */
+ int ln_mt_state;
/* serialise startup/shutdown */
- struct semaphore ln_rc_signal;
+ struct semaphore ln_mt_signal;
struct mutex ln_api_mutex;
struct mutex ln_lnd_mutex;
-#else
-# ifndef HAVE_LIBPTHREAD
- int ln_api_mutex;
- int ln_lnd_mutex;
-# else
- pthread_mutex_t ln_api_mutex;
- pthread_mutex_t ln_lnd_mutex;
-# endif
-#endif
- int ln_init; /* LNetInit() called? */
/* Have I called LNetNIInit myself? */
int ln_niinit_self;
/* LNetNIInit/LNetNIFini counter */
int ln_refcount;
- /* shutdown in progress */
- int ln_shutdown;
+ /* SHUTDOWN/RUNNING/STOPPING */
+ int ln_state;
int ln_routing; /* am I a router? */
lnet_pid_t ln_pid; /* requested pid */
/* uniquely identifies this ni in this epoch */
__u64 ln_interface_cookie;
/* registered LNDs */
- cfs_list_t ln_lnds;
+ struct list_head ln_lnds;
- /* space for network names */
- char *ln_network_tokens;
- int ln_network_tokens_nob;
/* test protocol compatibility flags */
int ln_testprotocompat;
-#ifndef __KERNEL__
- /* Temporary workaround to allow uOSS and test programs force
- * server mode in userspace. The only place where we use it is
- * lnet_prepare(). The only way to turn this flag on is to
- * call lnet_server_mode() */
- int ln_server_mode_flag;
-#endif
-} lnet_t;
+ /* 0 - load the NIs from the mod params
+ * 1 - do not load the NIs from the mod params
+ * Reverse logic to ensure that other calls to LNetNIInit
+ * need no change
+ */
+ bool ln_nis_from_mod_params;
+
+ /*
+ * waitq for the monitor thread. The monitor thread takes care of
+ * checking routes, timedout messages and resending messages.
+ */
+ wait_queue_head_t ln_mt_waitq;
+
+ /* per-cpt resend queues */
+ struct list_head **ln_mt_resendqs;
+ /* local NIs to recover */
+ struct list_head ln_mt_localNIRecovq;
+ /* local NIs to recover */
+ struct list_head ln_mt_peerNIRecovq;
+ /*
+ * An array of queues for GET/PUT waiting for REPLY/ACK respectively.
+ * There are CPT number of queues. Since response trackers will be
+ * added on the fast path we can't afford to grab the exclusive
+ * net lock to protect these queues. The CPT will be calculated
+ * based on the mdh cookie.
+ */
+ struct list_head **ln_mt_rstq;
+ /* recovery eq handler */
+ struct lnet_handle_eq ln_mt_eqh;
+
+};
#endif