-/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
+/*
+ * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
+ *
+ * Copyright (c) 2011, 2017, Intel Corporation.
*
* Author: Zach Brown <zab@zabbo.net>
* Author: Peter J. Braam <braam@clusterfs.com>
*
*/
+#ifndef _SOCKLND_SOCKLND_H_
+#define _SOCKLND_SOCKLND_H_
+
#define DEBUG_PORTAL_ALLOC
-#ifndef EXPORT_SYMTAB
-# define EXPORT_SYMTAB
-#endif
+#define DEBUG_SUBSYSTEM S_LND
-#include <linux/config.h>
-#include <linux/module.h>
+#include <linux/crc32.h>
+#include <linux/errno.h>
+#include <linux/if.h>
+#include <linux/init.h>
#include <linux/kernel.h>
+#include <linux/kthread.h>
+#include <linux/kmod.h>
+#include <linux/list.h>
#include <linux/mm.h>
-#include <linux/string.h>
+#include <linux/module.h>
+#include <linux/pagemap.h>
#include <linux/stat.h>
-#include <linux/errno.h>
-#include <linux/smp_lock.h>
+#include <linux/string.h>
+#include <linux/syscalls.h>
+#include <linux/sysctl.h>
+#include <linux/uio.h>
#include <linux/unistd.h>
+#include <linux/hashtable.h>
#include <net/sock.h>
#include <net/tcp.h>
-#include <linux/uio.h>
-
-#include <asm/system.h>
-#include <asm/uaccess.h>
-#include <asm/irq.h>
-#include <linux/init.h>
-#include <linux/fs.h>
-#include <linux/file.h>
-#include <linux/stat.h>
-#include <linux/list.h>
-#include <linux/kmod.h>
-#include <linux/sysctl.h>
-#include <asm/uaccess.h>
-#include <asm/segment.h>
-#include <asm/div64.h>
-
-#define DEBUG_SUBSYSTEM S_NAL
-
-#include <linux/kp30.h>
-#include <linux/portals_compat25.h>
-#include <linux/kpr.h>
-#include <portals/p30.h>
-#include <portals/lib-p30.h>
-#include <portals/nal.h>
-#include <portals/socknal.h>
-
-#define SOCKNAL_N_AUTOCONNECTD 4 /* # socknal autoconnect daemons */
-
-#define SOCKNAL_MIN_RECONNECT_INTERVAL HZ /* first failed connection retry... */
-#define SOCKNAL_MAX_RECONNECT_INTERVAL (60*HZ) /* ...exponentially increasing to this */
-
-/* default vals for runtime tunables */
-#define SOCKNAL_IO_TIMEOUT 50 /* default comms timeout (seconds) */
-#define SOCKNAL_EAGER_ACK 0 /* default eager ack (boolean) */
-#define SOCKNAL_TYPED_CONNS 1 /* unidirectional large, bidirectional small? */
-#define SOCKNAL_ZC_MIN_FRAG (2<<10) /* default smallest zerocopy fragment */
-#define SOCKNAL_MIN_BULK (1<<10) /* smallest "large" message */
-#define SOCKNAL_BUFFER_SIZE (8<<20) /* default socket buffer size */
-#define SOCKNAL_NAGLE 0 /* enable/disable NAGLE? */
-#define SOCKNAL_IRQ_AFFINITY 1 /* enable/disable IRQ affinity? */
-#define SOCKNAL_KEEPALIVE_IDLE 30 /* # seconds idle before 1st probe */
-#define SOCKNAL_KEEPALIVE_COUNT 10 /* # unanswered probes to determine peer death */
-#define SOCKNAL_KEEPALIVE_INTVL 2 /* seconds between probes */
-
-#define SOCKNAL_PEER_HASH_SIZE 101 /* # peer lists */
-
-#define SOCKNAL_SMALL_FWD_NMSGS 128 /* # small messages I can be forwarding at any time */
-#define SOCKNAL_LARGE_FWD_NMSGS 64 /* # large messages I can be forwarding at any time */
-
-#define SOCKNAL_SMALL_FWD_PAGES 1 /* # pages in a small message fwd buffer */
-
-#define SOCKNAL_LARGE_FWD_PAGES (PAGE_ALIGN(PTL_MTU) >> PAGE_SHIFT)
- /* # pages in a large message fwd buffer */
-
-#define SOCKNAL_RESCHED 100 /* # scheduler loops before reschedule */
-#define SOCKNAL_ENOMEM_RETRY 1 /* jiffies between retries */
-
-#define SOCKNAL_MAX_INTERFACES 16 /* Largest number of interfaces we bind */
-
-#define SOCKNAL_ROUND_ROBIN 0 /* round robin / load balance */
-
-#define SOCKNAL_TX_LOW_WATER(sk) (((sk)->sk_sndbuf*8)/10)
-
-#define SOCKNAL_SINGLE_FRAG_TX 0 /* disable multi-fragment sends */
-#define SOCKNAL_SINGLE_FRAG_RX 0 /* disable multi-fragment receives */
-#define SOCKNAL_RISK_KMAP_DEADLOCK 1 /* risk kmap deadlock on multi-frag I/O
- * (backs off to single-frag if disabled) */
-
-#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,5,72))
-# define sk_allocation allocation
-# define sk_data_ready data_ready
-# define sk_write_space write_space
-# define sk_user_data user_data
-# define sk_prot prot
-# define sk_sndbuf sndbuf
-# define sk_socket socket
-#endif
+#include <lnet/lib-lnet.h>
+#include <lnet/socklnd.h>
-#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0))
-# define sk_wmem_queued wmem_queued
-# define sk_err err
+#ifndef NETIF_F_CSUM_MASK
+# define NETIF_F_CSUM_MASK NETIF_F_ALL_CSUM
#endif
-typedef struct /* pool of forwarding buffers */
-{
- spinlock_t fmp_lock; /* serialise */
- struct list_head fmp_idle_fmbs; /* free buffers */
- struct list_head fmp_blocked_conns; /* connections waiting for a buffer */
- int fmp_nactive_fmbs; /* # buffers in use */
- int fmp_buff_pages; /* # pages per buffer */
-} ksock_fmb_pool_t;
+/* assume one thread for each connection type */
+#define SOCKNAL_NSCHEDS 3
+#define SOCKNAL_NSCHEDS_HIGH (SOCKNAL_NSCHEDS << 1)
+#define SOCKNAL_PEER_HASH_BITS 7 /* log2 of # peer_ni lists */
+#define SOCKNAL_RESCHED 100 /* # scheduler loops before reschedule */
+#define SOCKNAL_INSANITY_RECONN 5000 /* connd is trying on reconn infinitely */
+#define SOCKNAL_ENOMEM_RETRY 1 /* seconds between retries */
-typedef struct /* per scheduler state */
-{
- spinlock_t kss_lock; /* serialise */
- struct list_head kss_rx_conns; /* conn waiting to be read */
- struct list_head kss_tx_conns; /* conn waiting to be written */
-#if SOCKNAL_ZC
- struct list_head kss_zctxdone_list; /* completed ZC transmits */
-#endif
- wait_queue_head_t kss_waitq; /* where scheduler sleeps */
- int kss_nconns; /* # connections assigned to this scheduler */
-} ksock_sched_t;
+#define SOCKNAL_SINGLE_FRAG_TX 0 /* disable multi-fragment sends */
+#define SOCKNAL_SINGLE_FRAG_RX 0 /* disable multi-fragment receives */
-typedef struct
-{
- int ksni_valid:1; /* been set yet? */
- int ksni_bound:1; /* bound to a cpu yet? */
- int ksni_sched:6; /* which scheduler (assumes < 64) */
-} ksock_irqinfo_t;
+#define SOCKNAL_VERSION_DEBUG 0 /* enable protocol version debugging */
-typedef struct
-{
- __u32 ksni_ipaddr; /* interface's IP address */
- __u32 ksni_netmask; /* interface's network mask */
- int ksni_nroutes; /* # routes using (active) */
- int ksni_npeers; /* # peers using (passive) */
-} ksock_interface_t;
-
-typedef struct
-{
- int ksnd_io_timeout; /* "stuck" socket timeout (seconds) */
- int ksnd_eager_ack; /* make TCP ack eagerly? */
- int ksnd_typed_conns; /* drive sockets by type? */
- int ksnd_min_bulk; /* smallest "large" message */
- int ksnd_buffer_size; /* socket buffer size */
- int ksnd_nagle; /* enable NAGLE? */
- int ksnd_irq_affinity; /* enable IRQ affinity? */
- int ksnd_keepalive_idle; /* # idle secs before 1st probe */
- int ksnd_keepalive_count; /* # probes */
- int ksnd_keepalive_intvl; /* time between probes */
-#if SOCKNAL_ZC
- unsigned int ksnd_zc_min_frag; /* minimum zero copy frag size */
+/* risk kmap deadlock on multi-frag I/O (backs off to single-frag if disabled).
+ * no risk if we're not running on a CONFIG_HIGHMEM platform. */
+#ifdef CONFIG_HIGHMEM
+# define SOCKNAL_RISK_KMAP_DEADLOCK 0
+#else
+# define SOCKNAL_RISK_KMAP_DEADLOCK 1
#endif
- struct ctl_table_header *ksnd_sysctl; /* sysctl interface */
-} ksock_tunables_t;
-
-typedef struct
-{
- int ksnd_init; /* initialisation state */
- __u64 ksnd_incarnation; /* my epoch */
-
- rwlock_t ksnd_global_lock; /* stabilize peer/conn ops */
- struct list_head *ksnd_peers; /* hash table of all my known peers */
- int ksnd_peer_hash_size; /* size of ksnd_peers */
- int ksnd_nthreads; /* # live threads */
- int ksnd_shuttingdown; /* tell threads to exit */
- int ksnd_nschedulers; /* # schedulers */
- ksock_sched_t *ksnd_schedulers; /* their state */
-
- atomic_t ksnd_npeers; /* total # peers extant */
- atomic_t ksnd_nclosing_conns; /* # closed conns extant */
-
- kpr_router_t ksnd_router; /* THE router */
-
- ksock_fmb_pool_t ksnd_small_fmp; /* small message forwarding buffers */
- ksock_fmb_pool_t ksnd_large_fmp; /* large message forwarding buffers */
-
- atomic_t ksnd_nactive_ltxs; /* #active ltxs */
-
- struct list_head ksnd_deathrow_conns; /* conns to be closed */
- struct list_head ksnd_zombie_conns; /* conns to be freed */
- struct list_head ksnd_enomem_conns; /* conns to be retried */
- wait_queue_head_t ksnd_reaper_waitq; /* reaper sleeps here */
- unsigned long ksnd_reaper_waketime; /* when reaper will wake */
- spinlock_t ksnd_reaper_lock; /* serialise */
-
- int ksnd_enomem_tx; /* test ENOMEM sender */
- int ksnd_stall_tx; /* test sluggish sender */
- int ksnd_stall_rx; /* test sluggish receiver */
-
- struct list_head ksnd_autoconnectd_routes; /* routes waiting to be connected */
- wait_queue_head_t ksnd_autoconnectd_waitq; /* autoconnectds sleep here */
- spinlock_t ksnd_autoconnectd_lock; /* serialise */
-
- ksock_irqinfo_t ksnd_irqinfo[NR_IRQS];/* irq->scheduler lookup */
-
- int ksnd_ninterfaces;
- ksock_interface_t ksnd_interfaces[SOCKNAL_MAX_INTERFACES]; /* published interfaces */
-} ksock_nal_data_t;
+/* per scheduler state */
+struct ksock_sched {
+ /* serialise */
+ spinlock_t kss_lock;
+ /* conn waiting to be written */
+ struct list_head kss_rx_conns;
+ struct list_head kss_tx_conns;
+ /* zombie noop tx list */
+ struct list_head kss_zombie_noop_txs;
+ /* where scheduler sleeps */
+ wait_queue_head_t kss_waitq;
+ /* # connections assigned to this scheduler */
+ int kss_nconns;
+ /* max allowed threads */
+ int kss_nthreads_max;
+ /* number of threads */
+ int kss_nthreads;
+ /* CPT id */
+ int kss_cpt;
+};
+
+#define KSOCK_CPT_SHIFT 16
+#define KSOCK_THREAD_ID(cpt, sid) (((cpt) << KSOCK_CPT_SHIFT) | (sid))
+#define KSOCK_THREAD_CPT(id) ((id) >> KSOCK_CPT_SHIFT)
+#define KSOCK_THREAD_SID(id) ((id) & ((1UL << KSOCK_CPT_SHIFT) - 1))
+
+struct ksock_interface { /* in-use interface */
+ int ksni_index; /* Linux interface index */
+ __u32 ksni_ipaddr; /* interface's IP address */
+ __u32 ksni_netmask; /* interface's network mask */
+ int ksni_nroutes; /* # routes using (active) */
+ int ksni_npeers; /* # peers using (passive) */
+ char ksni_name[IFNAMSIZ]; /* interface name */
+};
+
+struct ksock_tunables {
+ /* "stuck" socket timeout (seconds) */
+ int *ksnd_timeout;
+ /* # scheduler threads in each pool while starting */
+ int *ksnd_nscheds;
+ int *ksnd_nconnds; /* # connection daemons */
+ int *ksnd_nconnds_max; /* max # connection daemons */
+ int *ksnd_min_reconnectms; /* first connection retry after (ms)... */
+ int *ksnd_max_reconnectms; /* ...exponentially increasing to this */
+ int *ksnd_eager_ack; /* make TCP ack eagerly? */
+ int *ksnd_typed_conns; /* drive sockets by type? */
+ int *ksnd_min_bulk; /* smallest "large" message */
+ int *ksnd_tx_buffer_size; /* socket tx buffer size */
+ int *ksnd_rx_buffer_size; /* socket rx buffer size */
+ int *ksnd_nagle; /* enable NAGLE? */
+ int *ksnd_round_robin; /* round robin for multiple interfaces */
+ int *ksnd_keepalive; /* # secs for sending keepalive NOOP */
+ int *ksnd_keepalive_idle; /* # idle secs before 1st probe */
+ int *ksnd_keepalive_count; /* # probes */
+ int *ksnd_keepalive_intvl; /* time between probes */
+ int *ksnd_credits; /* # concurrent sends */
+ int *ksnd_peertxcredits; /* # concurrent sends to 1 peer_ni */
+ int *ksnd_peerrtrcredits; /* # per-peer_ni router buffer credits */
+ int *ksnd_peertimeout; /* seconds to consider peer_ni dead */
+ int *ksnd_enable_csum; /* enable check sum */
+ int *ksnd_inject_csum_error; /* set non-zero to inject checksum error */
+ int *ksnd_nonblk_zcack; /* always send zc-ack on non-blocking connection */
+ unsigned int *ksnd_zc_min_payload; /* minimum zero copy payload size */
+ int *ksnd_zc_recv; /* enable ZC receive (for Chelsio TOE) */
+ int *ksnd_zc_recv_min_nfrags; /* minimum # of fragments to enable ZC receive */
+ int *ksnd_irq_affinity; /* enable IRQ affinity? */
+#ifdef SOCKNAL_BACKOFF
+ int *ksnd_backoff_init; /* initial TCP backoff */
+ int *ksnd_backoff_max; /* maximum TCP backoff */
+#endif
+#if SOCKNAL_VERSION_DEBUG
+ int *ksnd_protocol; /* protocol version */
+#endif
+};
+
+struct ksock_net {
+ __u64 ksnn_incarnation; /* my epoch */
+ struct list_head ksnn_list; /* chain on global list */
+ atomic_t ksnn_npeers; /* # peers */
+ int ksnn_ninterfaces; /* IP interfaces */
+ struct ksock_interface ksnn_interfaces[LNET_INTERFACES_NUM];
+};
+/* When the ksock_net is shut down, this (negative) bias is added to
+ * ksnn_npeers, which prevents new peers from being added.
+ */
+#define SOCKNAL_SHUTDOWN_BIAS (INT_MIN+1)
+
+/** connd timeout */
+#define SOCKNAL_CONND_TIMEOUT 120
+/** reserved thread for accepting & creating new connd */
+#define SOCKNAL_CONND_RESV 1
+
+struct ksock_nal_data {
+ int ksnd_init; /* initialisation state */
+ int ksnd_nnets; /* # networks set up */
+ struct list_head ksnd_nets; /* list of nets */
+ /* stabilize peer_ni/conn ops */
+ rwlock_t ksnd_global_lock;
+ /* hash table of all my known peers */
+ DECLARE_HASHTABLE(ksnd_peers, SOCKNAL_PEER_HASH_BITS);
+
+ int ksnd_nthreads; /* # live threads */
+ int ksnd_shuttingdown; /* tell threads to exit */
+ /* schedulers information */
+ struct ksock_sched **ksnd_schedulers;
+
+ atomic_t ksnd_nactive_txs; /* #active txs */
+
+ /* conns to close: reaper_lock*/
+ struct list_head ksnd_deathrow_conns;
+ /* conns to free: reaper_lock */
+ struct list_head ksnd_zombie_conns;
+ /* conns to retry: reaper_lock*/
+ struct list_head ksnd_enomem_conns;
+ /* reaper sleeps here */
+ wait_queue_head_t ksnd_reaper_waitq;
+ /* when reaper will wake */
+ time64_t ksnd_reaper_waketime;
+ /* serialise */
+ spinlock_t ksnd_reaper_lock;
+
+ int ksnd_enomem_tx; /* test ENOMEM sender */
+ int ksnd_stall_tx; /* test sluggish sender */
+ int ksnd_stall_rx; /* test sluggish receiver */
+
+ /* incoming connection requests */
+ struct list_head ksnd_connd_connreqs;
+ /* routes waiting to be connected */
+ struct list_head ksnd_connd_routes;
+ /* connds sleep here */
+ wait_queue_head_t ksnd_connd_waitq;
+ /* # connds connecting */
+ int ksnd_connd_connecting;
+ /** time stamp of the last failed connecting attempt */
+ time64_t ksnd_connd_failed_stamp;
+ /** # starting connd */
+ unsigned ksnd_connd_starting;
+ /** time stamp of the last starting connd */
+ time64_t ksnd_connd_starting_stamp;
+ /** # running connd */
+ unsigned ksnd_connd_running;
+ /* serialise */
+ spinlock_t ksnd_connd_lock;
+
+ /* list head for freed noop tx */
+ struct list_head ksnd_idle_noop_txs;
+ /* serialise, g_lock unsafe */
+ spinlock_t ksnd_tx_lock;
+};
#define SOCKNAL_INIT_NOTHING 0
#define SOCKNAL_INIT_DATA 1
-#define SOCKNAL_INIT_LIB 2
-#define SOCKNAL_INIT_ALL 3
+#define SOCKNAL_INIT_ALL 2
/* A packet just assembled for transmission is represented by 1 or more
- * struct iovec fragments (the first frag contains the portals header),
- * followed by 0 or more ptl_kiov_t fragments.
+ * struct kvec fragments (the first frag contains the portals header),
+ * followed by 0 or more struct bio_vec fragments.
*
- * On the receive side, initially 1 struct iovec fragment is posted for
+ * On the receive side, initially 1 struct kvec fragment is posted for
* receive (the header). Once the header has been received, the payload is
- * received into either struct iovec or ptl_kiov_t fragments, depending on
+ * received into either struct kvec or struct bio_vec fragments, depending on
* what the header matched or whether the message needs forwarding. */
-struct ksock_conn; /* forward ref */
-struct ksock_peer; /* forward ref */
-struct ksock_route; /* forward ref */
+struct ksock_conn; /* forward ref */
+struct ksock_route; /* forward ref */
+struct ksock_proto; /* forward ref */
+
+struct ksock_tx { /* transmit packet */
+ struct list_head tx_list; /* queue on conn for transmission etc */
+ struct list_head tx_zc_list; /* queue on peer_ni for ZC request */
+ atomic_t tx_refcount; /* tx reference count */
+ int tx_nob; /* # packet bytes */
+ int tx_resid; /* residual bytes */
+ int tx_niov; /* # packet kvec frags */
+ struct kvec *tx_iov; /* packet kvec frags */
+ int tx_nkiov; /* # packet page frags */
+ unsigned short tx_zc_aborted; /* aborted ZC request */
+ unsigned short tx_zc_capable:1; /* payload is large enough for ZC */
+ unsigned short tx_zc_checked:1; /* Have I checked if I should ZC? */
+ unsigned short tx_nonblk:1; /* it's a non-blocking ACK */
+ struct bio_vec *tx_kiov; /* packet page frags */
+ struct ksock_conn *tx_conn; /* owning conn */
+ struct lnet_msg *tx_lnetmsg; /* lnet message for lnet_finalize() */
+ time64_t tx_deadline; /* when (in secs) tx times out */
+ struct ksock_msg tx_msg; /* socklnd message buffer */
+ int tx_desc_size; /* size of this descriptor */
+ enum lnet_msg_hstatus tx_hstatus; /* health status of tx */
+ union {
+ struct {
+ struct kvec iov; /* virt hdr */
+ struct bio_vec kiov[0]; /* paged payload */
+ } paged;
+ struct {
+ struct kvec iov[1]; /* virt hdr + payload */
+ } virt;
+ } tx_frags;
+};
+
+#define KSOCK_NOOP_TX_SIZE ((int)offsetof(struct ksock_tx, tx_frags.paged.kiov[0]))
+
+/* network zero copy callback descriptor embedded in struct ksock_tx */
-typedef struct /* transmit packet */
-{
- struct list_head tx_list; /* queue on conn for transmission etc */
- char tx_isfwd; /* forwarding / sourced here */
- int tx_nob; /* # packet bytes */
- int tx_resid; /* residual bytes */
- int tx_niov; /* # packet iovec frags */
- struct iovec *tx_iov; /* packet iovec frags */
- int tx_nkiov; /* # packet page frags */
- ptl_kiov_t *tx_kiov; /* packet page frags */
- struct ksock_conn *tx_conn; /* owning conn */
- ptl_hdr_t *tx_hdr; /* packet header (for debug only) */
-#if SOCKNAL_ZC
- zccd_t tx_zccd; /* zero copy callback descriptor */
+/* space for the rx frag descriptors; we either read a single contiguous
+ * header, or up to LNET_MAX_IOV frags of payload of either type. */
+union ksock_rxiovspace {
+ struct kvec iov[LNET_MAX_IOV];
+ struct bio_vec kiov[LNET_MAX_IOV];
+};
+
+#define SOCKNAL_RX_KSM_HEADER 1 /* reading ksock message header */
+#define SOCKNAL_RX_LNET_HEADER 2 /* reading lnet message header */
+#define SOCKNAL_RX_PARSE 3 /* Calling lnet_parse() */
+#define SOCKNAL_RX_PARSE_WAIT 4 /* waiting to be told to read the body */
+#define SOCKNAL_RX_LNET_PAYLOAD 5 /* reading lnet payload (to deliver here) */
+#define SOCKNAL_RX_SLOP 6 /* skipping body */
+
+struct ksock_conn {
+ struct ksock_peer_ni *ksnc_peer; /* owning peer_ni */
+ struct ksock_route *ksnc_route; /* owning route */
+ struct list_head ksnc_list; /* stash on peer_ni's conn list */
+ struct socket *ksnc_sock; /* actual socket */
+ void *ksnc_saved_data_ready; /* socket's original data_ready() callback */
+ void *ksnc_saved_write_space; /* socket's original write_space() callback */
+ atomic_t ksnc_conn_refcount; /* conn refcount */
+ atomic_t ksnc_sock_refcount; /* sock refcount */
+ struct ksock_sched *ksnc_scheduler; /* who schedules this connection */
+ __u32 ksnc_myipaddr; /* my IP */
+ __u32 ksnc_ipaddr; /* peer_ni's IP */
+ int ksnc_port; /* peer_ni's port */
+ signed int ksnc_type:3; /* type of connection,
+ * should be signed value */
+ unsigned int ksnc_closing:1; /* being shut down */
+ unsigned int ksnc_flip:1; /* flip or not, only for V2.x */
+ unsigned int ksnc_zc_capable:1; /* enable to ZC */
+ const struct ksock_proto *ksnc_proto; /* protocol for the connection */
+
+ /* READER */
+
+ /* where I enq waiting input or a forwarding descriptor */
+ struct list_head ksnc_rx_list;
+ time64_t ksnc_rx_deadline; /* when (in seconds) receive times out */
+ __u8 ksnc_rx_started; /* started receiving a message */
+ __u8 ksnc_rx_ready; /* data ready to read */
+ __u8 ksnc_rx_scheduled;/* being progressed */
+ __u8 ksnc_rx_state; /* what is being read */
+ int ksnc_rx_nob_left; /* # bytes to next hdr/body */
+ int ksnc_rx_nob_wanted; /* bytes actually wanted */
+ int ksnc_rx_niov; /* # kvec frags */
+ struct kvec *ksnc_rx_iov; /* the kvec frags */
+ int ksnc_rx_nkiov; /* # page frags */
+ struct bio_vec *ksnc_rx_kiov; /* the page frags */
+ union ksock_rxiovspace ksnc_rx_iov_space;/* space for frag descriptors */
+ __u32 ksnc_rx_csum; /* partial checksum for incoming
+ * data */
+ struct lnet_msg *ksnc_lnet_msg; /* rx lnet_finalize arg*/
+ struct ksock_msg ksnc_msg; /* incoming message buffer:
+ * V2.x message takes the
+ * whole struct
+ * V1.x message is a bare
+ * struct lnet_hdr, it's stored
+ * in ksnc_msg.ksm_u.lnetmsg
+ */
+ /* -- WRITER -- */
+ /* where I enq waiting for output space */
+ struct list_head ksnc_tx_list;
+ /* packets waiting to be sent */
+ struct list_head ksnc_tx_queue;
+ /* next TX that can carry a LNet message or ZC-ACK */
+ struct ksock_tx *ksnc_tx_carrier;
+ /* when (in seconds) tx times out */
+ time64_t ksnc_tx_deadline;
+ /* send buffer marker */
+ int ksnc_tx_bufnob;
+ /* # bytes queued */
+ atomic_t ksnc_tx_nob;
+ /* write space */
+ int ksnc_tx_ready;
+ /* being progressed */
+ int ksnc_tx_scheduled;
+ /* time stamp of the last posted TX */
+ time64_t ksnc_tx_last_post;
+};
+
+struct ksock_route {
+ struct list_head ksnr_list; /* chain on peer_ni route list*/
+ struct list_head ksnr_connd_list;/* chain on ksnr_connd_routes */
+ struct ksock_peer_ni *ksnr_peer; /* owning peer_ni */
+ atomic_t ksnr_refcount; /* # users */
+ time64_t ksnr_timeout; /* when (in secs) reconnection
+ * can happen next
+ */
+ time64_t ksnr_retry_interval;/* secs between retries */
+ int ksnr_myiface; /* interface index */
+ __u32 ksnr_ipaddr; /* IP address to connect to */
+ int ksnr_port; /* port to connect to */
+ unsigned int ksnr_scheduled:1;/* scheduled for attention */
+ unsigned int ksnr_connecting:1;/* connection in progress */
+ unsigned int ksnr_connected:4;/* connections by type */
+ unsigned int ksnr_deleted:1; /* been removed from peer_ni? */
+ unsigned int ksnr_share_count;/* created explicitly? */
+ int ksnr_conn_count;/* # conns for this route */
+};
+
+#define SOCKNAL_KEEPALIVE_PING 1 /* cookie for keepalive ping */
+
+struct ksock_peer_ni {
+ struct hlist_node ksnp_list; /* stash on global peer_ni list */
+ time64_t ksnp_last_alive;/* when (in seconds) I was last alive */
+ struct lnet_process_id ksnp_id; /* who's on the other end(s) */
+ atomic_t ksnp_refcount; /* # users */
+ int ksnp_closing; /* being closed */
+ int ksnp_accepting; /* # passive connections pending */
+ int ksnp_error; /* errno on closing last conn */
+ __u64 ksnp_zc_next_cookie;/* ZC completion cookie */
+ __u64 ksnp_incarnation; /* latest known peer_ni incarnation */
+ const struct ksock_proto *ksnp_proto; /* latest known protocol */
+ struct list_head ksnp_conns; /* all active connections */
+ struct list_head ksnp_routes; /* routes */
+ struct list_head ksnp_tx_queue; /* waiting packets */
+ spinlock_t ksnp_lock; /* serialize, g_lock unsafe */
+ /* zero copy requests wait for ACK */
+ struct list_head ksnp_zc_req_list;
+ time64_t ksnp_send_keepalive; /* time to send keepalive */
+ struct lnet_ni *ksnp_ni; /* which network */
+ int ksnp_n_passive_ips; /* # of... */
+ __u32 ksnp_passive_ips[LNET_INTERFACES_NUM]; /* preferred local interfaces */
+};
+
+struct ksock_connreq {
+ /* stash on ksnd_connd_connreqs */
+ struct list_head ksncr_list;
+ /* chosen NI */
+ struct lnet_ni *ksncr_ni;
+ /* accepted socket */
+ struct socket *ksncr_sock;
+};
+
+extern struct ksock_nal_data ksocknal_data;
+extern struct ksock_tunables ksocknal_tunables;
+
+#define SOCKNAL_MATCH_NO 0 /* TX can't match type of connection */
+#define SOCKNAL_MATCH_YES 1 /* TX matches type of connection */
+#define SOCKNAL_MATCH_MAY 2 /* TX can be sent on the connection, but not preferred */
+
+struct ksock_proto {
+ int pro_version; /* version number of protocol */
+ int (*pro_send_hello)(struct ksock_conn *, struct ksock_hello_msg *); /* handshake function */
+ int (*pro_recv_hello)(struct ksock_conn *, struct ksock_hello_msg *, int);/* handshake function */
+ void (*pro_pack)(struct ksock_tx *); /* message pack */
+ void (*pro_unpack)(struct ksock_msg *); /* message unpack */
+ struct ksock_tx *(*pro_queue_tx_msg)(struct ksock_conn *, struct ksock_tx *); /* queue tx on the connection */
+ int (*pro_queue_tx_zcack)(struct ksock_conn *, struct ksock_tx *, __u64); /* queue ZC ack on the connection */
+ int (*pro_handle_zcreq)(struct ksock_conn *, __u64, int); /* handle ZC request */
+ int (*pro_handle_zcack)(struct ksock_conn *, __u64, __u64); /* handle ZC ACK */
+ int (*pro_match_tx)(struct ksock_conn *, struct ksock_tx *, int); /* msg type matches the connection type:
+ * return value:
+ * return MATCH_NO : no
+ * return MATCH_YES : matching type
+ * return MATCH_MAY : can be backup */
+};
+
+extern const struct ksock_proto ksocknal_protocol_v1x;
+extern const struct ksock_proto ksocknal_protocol_v2x;
+extern const struct ksock_proto ksocknal_protocol_v3x;
+
+#define KSOCK_PROTO_V1_MAJOR LNET_PROTO_TCP_VERSION_MAJOR
+#define KSOCK_PROTO_V1_MINOR LNET_PROTO_TCP_VERSION_MINOR
+#define KSOCK_PROTO_V1 KSOCK_PROTO_V1_MAJOR
+
+#ifndef CPU_MASK_NONE
+#define CPU_MASK_NONE 0UL
#endif
-} ksock_tx_t;
-typedef struct /* forwarded packet */
+static inline __u32 ksocknal_csum(__u32 crc, unsigned char const *p, size_t len)
{
- ksock_tx_t ftx_tx; /* send info */
- struct iovec ftx_iov; /* hdr iovec */
-} ksock_ftx_t;
+#if 1
+ return crc32_le(crc, p, len);
+#else
+ while (len-- > 0)
+ crc = ((crc + 0x100) & ~0xff) | ((crc + *p++) & 0xff) ;
-#define KSOCK_ZCCD_2_TX(ptr) list_entry (ptr, ksock_tx_t, tx_zccd)
-/* network zero copy callback descriptor embedded in ksock_tx_t */
+ return crc;
+#endif
+}
-typedef struct /* locally transmitted packet */
+static inline int
+ksocknal_route_mask(void)
{
- ksock_tx_t ltx_tx; /* send info */
- void *ltx_private; /* lib_finalize() callback arg */
- void *ltx_cookie; /* lib_finalize() callback arg */
- ptl_hdr_t ltx_hdr; /* buffer for packet header */
- int ltx_desc_size; /* bytes allocated for this desc */
- struct iovec ltx_iov[1]; /* iov for hdr + payload */
- ptl_kiov_t ltx_kiov[0]; /* kiov for payload */
-} ksock_ltx_t;
-
-#define KSOCK_TX_2_KPR_FWD_DESC(ptr) list_entry ((kprfd_scratch_t *)ptr, kpr_fwd_desc_t, kprfd_scratch)
-/* forwarded packets (router->socknal) embedded in kpr_fwd_desc_t::kprfd_scratch */
-
-#define KSOCK_TX_2_KSOCK_LTX(ptr) list_entry (ptr, ksock_ltx_t, ltx_tx)
-/* local packets (lib->socknal) embedded in ksock_ltx_t::ltx_tx */
-
-/* NB list_entry() is used here as convenient macro for calculating a
- * pointer to a struct from the address of a member. */
-
-typedef struct /* Kernel portals Socket Forwarding message buffer */
-{ /* (socknal->router) */
- struct list_head fmb_list; /* queue idle */
- kpr_fwd_desc_t fmb_fwd; /* router's descriptor */
- ksock_fmb_pool_t *fmb_pool; /* owning pool */
- struct ksock_peer *fmb_peer; /* peer received from */
- ptl_hdr_t fmb_hdr; /* message header */
- ptl_kiov_t fmb_kiov[0]; /* payload frags */
-} ksock_fmb_t;
+ if (!*ksocknal_tunables.ksnd_typed_conns)
+ return BIT(SOCKLND_CONN_ANY);
-/* space for the rx frag descriptors; we either read a single contiguous
- * header, or up to PTL_MD_MAX_IOV frags of payload of either type. */
-typedef union {
- struct iovec iov[PTL_MD_MAX_IOV];
- ptl_kiov_t kiov[PTL_MD_MAX_IOV];
-} ksock_rxiovspace_t;
-
-#define SOCKNAL_RX_HEADER 1 /* reading header */
-#define SOCKNAL_RX_BODY 2 /* reading body (to deliver here) */
-#define SOCKNAL_RX_BODY_FWD 3 /* reading body (to forward) */
-#define SOCKNAL_RX_SLOP 4 /* skipping body */
-#define SOCKNAL_RX_GET_FMB 5 /* scheduled for forwarding */
-#define SOCKNAL_RX_FMB_SLEEP 6 /* blocked waiting for a fwd desc */
-
-typedef struct ksock_conn
-{
- struct ksock_peer *ksnc_peer; /* owning peer */
- struct ksock_route *ksnc_route; /* owning route */
- struct list_head ksnc_list; /* stash on peer's conn list */
- struct socket *ksnc_sock; /* actual socket */
- void *ksnc_saved_data_ready; /* socket's original data_ready() callback */
- void *ksnc_saved_write_space; /* socket's original write_space() callback */
- atomic_t ksnc_refcount; /* # users */
- ksock_sched_t *ksnc_scheduler; /* who schedules this connection */
- __u32 ksnc_myipaddr; /* my IP */
- __u32 ksnc_ipaddr; /* peer's IP */
- int ksnc_port; /* peer's port */
- int ksnc_closing; /* being shut down */
- int ksnc_type; /* type of connection */
- __u64 ksnc_incarnation; /* peer's incarnation */
-
- /* reader */
- struct list_head ksnc_rx_list; /* where I enq waiting input or a forwarding descriptor */
- unsigned long ksnc_rx_deadline; /* when (in jiffies) receive times out */
- int ksnc_rx_started; /* started receiving a message */
- int ksnc_rx_ready; /* data ready to read */
- int ksnc_rx_scheduled; /* being progressed */
- int ksnc_rx_state; /* what is being read */
- int ksnc_rx_nob_left; /* # bytes to next hdr/body */
- int ksnc_rx_nob_wanted; /* bytes actually wanted */
- int ksnc_rx_niov; /* # iovec frags */
- struct iovec *ksnc_rx_iov; /* the iovec frags */
- int ksnc_rx_nkiov; /* # page frags */
- ptl_kiov_t *ksnc_rx_kiov; /* the page frags */
- ksock_rxiovspace_t ksnc_rx_iov_space; /* space for frag descriptors */
- void *ksnc_cookie; /* rx lib_finalize passthru arg */
- ptl_hdr_t ksnc_hdr; /* where I read headers into */
-
- /* WRITER */
- struct list_head ksnc_tx_list; /* where I enq waiting for output space */
- struct list_head ksnc_tx_queue; /* packets waiting to be sent */
- unsigned long ksnc_tx_deadline; /* when (in jiffies) tx times out */
- int ksnc_tx_bufnob; /* send buffer marker */
- atomic_t ksnc_tx_nob; /* # bytes queued */
- int ksnc_tx_ready; /* write space */
- int ksnc_tx_scheduled; /* being progressed */
-
-#if !SOCKNAL_SINGLE_FRAG_RX
- struct iovec ksnc_rx_scratch_iov[PTL_MD_MAX_IOV];
-#endif
-#if !SOCKNAL_SINGLE_FRAG_TX
- struct iovec ksnc_tx_scratch_iov[PTL_MD_MAX_IOV];
-#endif
-} ksock_conn_t;
-
-#define KSNR_TYPED_ROUTES ((1 << SOCKNAL_CONN_CONTROL) | \
- (1 << SOCKNAL_CONN_BULK_IN) | \
- (1 << SOCKNAL_CONN_BULK_OUT))
+ return (BIT(SOCKLND_CONN_CONTROL) |
+ BIT(SOCKLND_CONN_BULK_IN) |
+ BIT(SOCKLND_CONN_BULK_OUT));
+}
-typedef struct ksock_route
-{
- struct list_head ksnr_list; /* chain on peer route list */
- struct list_head ksnr_connect_list; /* chain on autoconnect list */
- struct ksock_peer *ksnr_peer; /* owning peer */
- atomic_t ksnr_refcount; /* # users */
- unsigned long ksnr_timeout; /* when (in jiffies) reconnection can happen next */
- unsigned int ksnr_retry_interval; /* how long between retries */
- __u32 ksnr_myipaddr; /* my IP */
- __u32 ksnr_ipaddr; /* IP address to connect to */
- int ksnr_port; /* port to connect to */
- unsigned int ksnr_connecting:4; /* autoconnects in progress by type */
- unsigned int ksnr_connected:4; /* connections established by type */
- unsigned int ksnr_deleted:1; /* been removed from peer? */
- unsigned int ksnr_share_count; /* created explicitly? */
- int ksnr_conn_count; /* # conns established by this route */
-} ksock_route_t;
-
-typedef struct ksock_peer
-{
- struct list_head ksnp_list; /* stash on global peer list */
- ptl_nid_t ksnp_nid; /* who's on the other end(s) */
- atomic_t ksnp_refcount; /* # users */
- int ksnp_sharecount; /* lconf usage counter */
- int ksnp_closing; /* being closed */
- int ksnp_error; /* errno on closing last conn */
- struct list_head ksnp_conns; /* all active connections */
- struct list_head ksnp_routes; /* routes */
- struct list_head ksnp_tx_queue; /* waiting packets */
- unsigned long ksnp_last_alive; /* when (in jiffies) I was last alive */
- int ksnp_n_passive_ips; /* # of... */
- __u32 ksnp_passive_ips[SOCKNAL_MAX_INTERFACES]; /* preferred local interfaces */
-} ksock_peer_t;
-
-
-extern lib_nal_t ksocknal_lib;
-extern ksock_nal_data_t ksocknal_data;
-extern ksock_tunables_t ksocknal_tunables;
-
-static inline struct list_head *
-ksocknal_nid2peerlist (ptl_nid_t nid)
+static inline void
+ksocknal_conn_addref(struct ksock_conn *conn)
{
- unsigned int hash = ((unsigned int)nid) % ksocknal_data.ksnd_peer_hash_size;
+ LASSERT(atomic_read(&conn->ksnc_conn_refcount) > 0);
+ atomic_inc(&conn->ksnc_conn_refcount);
+}
+
+extern void ksocknal_queue_zombie_conn(struct ksock_conn *conn);
+extern void ksocknal_finalize_zcreq(struct ksock_conn *conn);
- return (&ksocknal_data.ksnd_peers [hash]);
+static inline void
+ksocknal_conn_decref(struct ksock_conn *conn)
+{
+ LASSERT(atomic_read(&conn->ksnc_conn_refcount) > 0);
+ if (atomic_dec_and_test(&conn->ksnc_conn_refcount))
+ ksocknal_queue_zombie_conn(conn);
}
static inline int
-ksocknal_getconnsock (ksock_conn_t *conn)
+ksocknal_connsock_addref(struct ksock_conn *conn)
{
- int rc = -ESHUTDOWN;
+ int rc = -ESHUTDOWN;
- read_lock (&ksocknal_data.ksnd_global_lock);
- if (!conn->ksnc_closing) {
- rc = 0;
- get_file (conn->ksnc_sock->file);
- }
- read_unlock (&ksocknal_data.ksnd_global_lock);
+ read_lock(&ksocknal_data.ksnd_global_lock);
+ if (!conn->ksnc_closing) {
+ LASSERT(atomic_read(&conn->ksnc_sock_refcount) > 0);
+ atomic_inc(&conn->ksnc_sock_refcount);
+ rc = 0;
+ }
+ read_unlock(&ksocknal_data.ksnd_global_lock);
- return (rc);
+ return (rc);
}
static inline void
-ksocknal_putconnsock (ksock_conn_t *conn)
+ksocknal_connsock_decref(struct ksock_conn *conn)
{
- fput (conn->ksnc_sock->file);
+ LASSERT(atomic_read(&conn->ksnc_sock_refcount) > 0);
+ if (atomic_dec_and_test(&conn->ksnc_sock_refcount)) {
+ LASSERT (conn->ksnc_closing);
+ sock_release(conn->ksnc_sock);
+ conn->ksnc_sock = NULL;
+ ksocknal_finalize_zcreq(conn);
+ }
}
-#ifndef CONFIG_SMP
-static inline
-int ksocknal_nsched(void)
-{
- return 1;
-}
-#else
-#include <linux/lustre_version.h>
-# if !(defined(CONFIG_X86) && (LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,21))) || defined(CONFIG_X86_64) || (LUSTRE_KERNEL_VERSION < 39) || ((LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)) && !defined(CONFIG_X86_HT))
-static inline int
-ksocknal_nsched(void)
+static inline void
+ksocknal_tx_addref(struct ksock_tx *tx)
{
- return num_online_cpus();
+ LASSERT(atomic_read(&tx->tx_refcount) > 0);
+ atomic_inc(&tx->tx_refcount);
}
-static inline int
-ksocknal_sched2cpu(int i)
+extern void ksocknal_tx_prep(struct ksock_conn *, struct ksock_tx *tx);
+extern void ksocknal_tx_done(struct lnet_ni *ni, struct ksock_tx *tx, int error);
+
+static inline void
+ksocknal_tx_decref(struct ksock_tx *tx)
{
- return i;
+ LASSERT(atomic_read(&tx->tx_refcount) > 0);
+ if (atomic_dec_and_test(&tx->tx_refcount))
+ ksocknal_tx_done(NULL, tx, 0);
}
-static inline int
-ksocknal_irqsched2cpu(int i)
+static inline void
+ksocknal_route_addref(struct ksock_route *route)
{
- return i;
+ LASSERT(atomic_read(&route->ksnr_refcount) > 0);
+ atomic_inc(&route->ksnr_refcount);
}
-# else
-static inline int
-ksocknal_nsched(void)
-{
- if (smp_num_siblings == 1)
- return (num_online_cpus());
- /* We need to know if this assumption is crap */
- LASSERT (smp_num_siblings == 2);
- return (num_online_cpus()/2);
-}
+extern void ksocknal_destroy_route(struct ksock_route *route);
-static inline int
-ksocknal_sched2cpu(int i)
+static inline void
+ksocknal_route_decref(struct ksock_route *route)
{
- if (smp_num_siblings == 1)
- return i;
+ LASSERT(atomic_read(&route->ksnr_refcount) > 0);
+ if (atomic_dec_and_test(&route->ksnr_refcount))
+ ksocknal_destroy_route (route);
+}
- return (i * 2);
+static inline void
+ksocknal_peer_addref(struct ksock_peer_ni *peer_ni)
+{
+ LASSERT(atomic_read(&peer_ni->ksnp_refcount) > 0);
+ atomic_inc(&peer_ni->ksnp_refcount);
}
-static inline int
-ksocknal_irqsched2cpu(int i)
+extern void ksocknal_destroy_peer(struct ksock_peer_ni *peer_ni);
+
+static inline void
+ksocknal_peer_decref(struct ksock_peer_ni *peer_ni)
{
- return (ksocknal_sched2cpu(i) + 1);
+ LASSERT (atomic_read (&peer_ni->ksnp_refcount) > 0);
+ if (atomic_dec_and_test(&peer_ni->ksnp_refcount))
+ ksocknal_destroy_peer(peer_ni);
}
-# endif
-#endif
-extern void ksocknal_put_route (ksock_route_t *route);
-extern void ksocknal_put_peer (ksock_peer_t *peer);
-extern ksock_peer_t *ksocknal_find_peer_locked (ptl_nid_t nid);
-extern ksock_peer_t *ksocknal_get_peer (ptl_nid_t nid);
-extern int ksocknal_del_route (ptl_nid_t nid, __u32 ipaddr,
- int single, int keep_conn);
-extern int ksocknal_create_conn (ksock_route_t *route,
- struct socket *sock, int type);
-extern void ksocknal_close_conn_locked (ksock_conn_t *conn, int why);
-extern void ksocknal_terminate_conn (ksock_conn_t *conn);
-extern void ksocknal_destroy_conn (ksock_conn_t *conn);
-extern void ksocknal_put_conn (ksock_conn_t *conn);
-extern int ksocknal_close_stale_conns_locked (ksock_peer_t *peer, __u64 incarnation);
-extern int ksocknal_close_conn_and_siblings (ksock_conn_t *conn, int why);
-extern int ksocknal_close_matching_conns (ptl_nid_t nid, __u32 ipaddr);
-
-extern void ksocknal_queue_tx_locked (ksock_tx_t *tx, ksock_conn_t *conn);
-extern void ksocknal_tx_done (ksock_tx_t *tx, int asynch);
-extern void ksocknal_fwd_packet (void *arg, kpr_fwd_desc_t *fwd);
-extern void ksocknal_fmb_callback (void *arg, int error);
-extern void ksocknal_notify (void *arg, ptl_nid_t gw_nid, int alive);
-extern int ksocknal_thread_start (int (*fn)(void *arg), void *arg);
-extern int ksocknal_new_packet (ksock_conn_t *conn, int skip);
-extern int ksocknal_scheduler (void *arg);
-extern void ksocknal_data_ready(struct sock *sk, int n);
-extern void ksocknal_write_space(struct sock *sk);
-extern int ksocknal_autoconnectd (void *arg);
-extern int ksocknal_reaper (void *arg);
-extern int ksocknal_get_conn_tunables (ksock_conn_t *conn, int *txmem,
- int *rxmem, int *nagle);
-extern int ksocknal_setup_sock (struct socket *sock);
-extern int ksocknal_send_hello (ksock_conn_t *conn, __u32 *ipaddrs, int nipaddrs);
-extern int ksocknal_recv_hello (ksock_conn_t *conn,
- ptl_nid_t *nid, __u64 *incarnation, __u32 *ipaddrs);
+int ksocknal_startup(struct lnet_ni *ni);
+void ksocknal_shutdown(struct lnet_ni *ni);
+int ksocknal_ctl(struct lnet_ni *ni, unsigned int cmd, void *arg);
+int ksocknal_send(struct lnet_ni *ni, void *private, struct lnet_msg *lntmsg);
+int ksocknal_recv(struct lnet_ni *ni, void *private, struct lnet_msg *lntmsg,
+ int delayed, unsigned int niov,
+ struct bio_vec *kiov,
+ unsigned int offset, unsigned int mlen, unsigned int rlen);
+int ksocknal_accept(struct lnet_ni *ni, struct socket *sock);
+
+int ksocknal_add_peer(struct lnet_ni *ni, struct lnet_process_id id, __u32 ip,
+ int port);
+struct ksock_peer_ni *ksocknal_find_peer_locked(struct lnet_ni *ni,
+ struct lnet_process_id id);
+struct ksock_peer_ni *ksocknal_find_peer(struct lnet_ni *ni,
+ struct lnet_process_id id);
+extern void ksocknal_peer_failed(struct ksock_peer_ni *peer_ni);
+extern int ksocknal_create_conn(struct lnet_ni *ni, struct ksock_route *route,
+ struct socket *sock, int type);
+extern void ksocknal_close_conn_locked(struct ksock_conn *conn, int why);
+extern void ksocknal_terminate_conn(struct ksock_conn *conn);
+extern void ksocknal_destroy_conn(struct ksock_conn *conn);
+extern int ksocknal_close_peer_conns_locked(struct ksock_peer_ni *peer_ni,
+ __u32 ipaddr, int why);
+extern int ksocknal_close_conn_and_siblings(struct ksock_conn *conn, int why);
+int ksocknal_close_matching_conns(struct lnet_process_id id, __u32 ipaddr);
+extern struct ksock_conn *ksocknal_find_conn_locked(struct ksock_peer_ni *peer_ni,
+ struct ksock_tx *tx, int nonblk);
+
+extern int ksocknal_launch_packet(struct lnet_ni *ni, struct ksock_tx *tx,
+ struct lnet_process_id id);
+extern struct ksock_tx *ksocknal_alloc_tx(int type, int size);
+extern void ksocknal_free_tx(struct ksock_tx *tx);
+extern struct ksock_tx *ksocknal_alloc_tx_noop(__u64 cookie, int nonblk);
+extern void ksocknal_next_tx_carrier(struct ksock_conn *conn);
+extern void ksocknal_queue_tx_locked(struct ksock_tx *tx, struct ksock_conn *conn);
+extern void ksocknal_txlist_done(struct lnet_ni *ni, struct list_head *txlist,
+ int error);
+extern void ksocknal_notify(lnet_nid_t gw_nid);
+extern void ksocknal_query(struct lnet_ni *ni, lnet_nid_t nid, time64_t *when);
+extern int ksocknal_thread_start(int (*fn)(void *arg), void *arg, char *name);
+extern void ksocknal_thread_fini(void);
+extern void ksocknal_launch_all_connections_locked(struct ksock_peer_ni *peer_ni);
+extern struct ksock_route *ksocknal_find_connectable_route_locked(struct ksock_peer_ni *peer_ni);
+extern struct ksock_route *ksocknal_find_connecting_route_locked(struct ksock_peer_ni *peer_ni);
+extern int ksocknal_new_packet(struct ksock_conn *conn, int skip);
+extern int ksocknal_scheduler(void *arg);
+extern int ksocknal_connd(void *arg);
+extern int ksocknal_reaper(void *arg);
+int ksocknal_send_hello(struct lnet_ni *ni, struct ksock_conn *conn,
+ lnet_nid_t peer_nid, struct ksock_hello_msg *hello);
+int ksocknal_recv_hello(struct lnet_ni *ni, struct ksock_conn *conn,
+ struct ksock_hello_msg *hello,
+ struct lnet_process_id *id,
+ __u64 *incarnation);
+extern void ksocknal_read_callback(struct ksock_conn *conn);
+extern void ksocknal_write_callback(struct ksock_conn *conn);
+
+extern int ksocknal_lib_zc_capable(struct ksock_conn *conn);
+extern void ksocknal_lib_save_callback(struct socket *sock, struct ksock_conn *conn);
+extern void ksocknal_lib_set_callback(struct socket *sock, struct ksock_conn *conn);
+extern void ksocknal_lib_reset_callback(struct socket *sock,
+ struct ksock_conn *conn);
+extern void ksocknal_lib_push_conn(struct ksock_conn *conn);
+extern int ksocknal_lib_get_conn_addrs(struct ksock_conn *conn);
+extern int ksocknal_lib_setup_sock(struct socket *so);
+extern int ksocknal_lib_send_iov(struct ksock_conn *conn, struct ksock_tx *tx,
+ struct kvec *scratch_iov);
+extern int ksocknal_lib_send_kiov(struct ksock_conn *conn, struct ksock_tx *tx,
+ struct kvec *scratch_iov);
+extern void ksocknal_lib_eager_ack(struct ksock_conn *conn);
+extern int ksocknal_lib_recv_iov(struct ksock_conn *conn,
+ struct kvec *scratchiov);
+extern int ksocknal_lib_recv_kiov(struct ksock_conn *conn, struct page **pages,
+ struct kvec *scratchiov);
+extern int ksocknal_lib_get_conn_tunables(struct ksock_conn *conn, int *txmem,
+ int *rxmem, int *nagle);
+
+extern int ksocknal_tunables_init(void);
+
+extern void ksocknal_lib_csum_tx(struct ksock_tx *tx);
+
+extern int ksocknal_lib_memory_pressure(struct ksock_conn *conn);
+extern int ksocknal_lib_bind_thread_to_cpu(int id);
+
+#endif /* _SOCKLND_SOCKLND_H_ */