X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lnet%2Fklnds%2Fsocklnd%2Fsocklnd.h;h=abd733b95cf4902e39951184d1658a5bdd47a3c1;hb=617ad3af720a3827a30f52ec5faa917f6c902a8b;hp=761c5c5dd8903ef84524c2f9dd07cb4292b9b1f1;hpb=93090d9b8250cf95d33c41809d558148cebde67b;p=fs%2Flustre-release.git diff --git a/lnet/klnds/socklnd/socklnd.h b/lnet/klnds/socklnd/socklnd.h index 761c5c5..abd733b 100644 --- a/lnet/klnds/socklnd/socklnd.h +++ b/lnet/klnds/socklnd/socklnd.h @@ -1,7 +1,7 @@ /* * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved. * - * Copyright (c) 2011, 2016, Intel Corporation. + * Copyright (c) 2011, 2017, Intel Corporation. * * Author: Zach Brown * Author: Peter J. Braam @@ -41,28 +41,20 @@ #include #include #include +#include #include #include #include #include #include #include +#include #include #include -#include -#include #include #include -#ifdef HAVE_TCP_SENDPAGE_USE_SOCKET -# define cfs_tcp_sendpage(sk, page, offset, size, flags) \ - tcp_sendpage((sk)->sk_socket, page, offset, size, flags) -#else /* !HAVE_TCP_SENDPAGE_USE_SOCKET */ -# define cfs_tcp_sendpage(sk, page, offset, size, flags) \ - tcp_sendpage(sk, page, offset, size, flags) -#endif /* HAVE_TCP_SENDPAGE_USE_SOCKET */ - #ifndef NETIF_F_CSUM_MASK # define NETIF_F_CSUM_MASK NETIF_F_ALL_CSUM #endif @@ -71,15 +63,15 @@ #define SOCKNAL_NSCHEDS 3 #define SOCKNAL_NSCHEDS_HIGH (SOCKNAL_NSCHEDS << 1) -#define SOCKNAL_PEER_HASH_SIZE 101 /* # 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 */ +#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 */ -#define SOCKNAL_SINGLE_FRAG_TX 0 /* disable multi-fragment sends */ -#define SOCKNAL_SINGLE_FRAG_RX 0 /* disable multi-fragment receives */ +#define SOCKNAL_SINGLE_FRAG_TX 0 /* disable multi-fragment sends */ +#define SOCKNAL_SINGLE_FRAG_RX 0 /* disable multi-fragment receives */ -#define SOCKNAL_VERSION_DEBUG 0 /* enable protocol version debugging */ +#define SOCKNAL_VERSION_DEBUG 0 /* enable protocol version debugging */ /* 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. */ @@ -89,32 +81,25 @@ # define SOCKNAL_RISK_KMAP_DEADLOCK 1 #endif -struct ksock_sched_info; - -struct ksock_sched { /* per scheduler state */ - spinlock_t kss_lock; /* serialise */ - struct list_head kss_rx_conns; /* conn waiting to be read */ +/* per scheduler state */ +struct ksock_sched { + /* serialise */ + spinlock_t kss_lock; /* conn waiting to be written */ - struct list_head kss_tx_conns; + struct list_head kss_rx_conns; + struct list_head kss_tx_conns; /* zombie noop tx list */ - struct list_head kss_zombie_noop_txs; - wait_queue_head_t kss_waitq; /* where scheduler sleeps */ + struct list_head kss_zombie_noop_txs; + /* where scheduler sleeps */ + wait_queue_head_t kss_waitq; /* # connections assigned to this scheduler */ - int kss_nconns; - struct ksock_sched_info *kss_info; /* owner of it */ -#if !SOCKNAL_SINGLE_FRAG_RX - struct page *kss_rx_scratch_pgs[LNET_MAX_IOV]; -#endif -#if !SOCKNAL_SINGLE_FRAG_TX || !SOCKNAL_SINGLE_FRAG_RX - struct kvec kss_scratch_iov[LNET_MAX_IOV]; -#endif -}; - -struct ksock_sched_info { - int ksi_nthreads_max; /* max allowed threads */ - int ksi_nthreads; /* number of threads */ - int ksi_cpt; /* CPT id */ - struct ksock_sched *ksi_scheds; /* array of schedulers */ + 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 @@ -123,6 +108,7 @@ struct ksock_sched_info { #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) */ @@ -160,9 +146,7 @@ struct ksock_tunables { 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 */ -#ifdef CPU_AFFINITY int *ksnd_irq_affinity; /* enable IRQ affinity? */ -#endif #ifdef SOCKNAL_BACKOFF int *ksnd_backoff_init; /* initial TCP backoff */ int *ksnd_backoff_max; /* maximum TCP backoff */ @@ -174,13 +158,15 @@ struct ksock_tunables { struct ksock_net { __u64 ksnn_incarnation; /* my epoch */ - spinlock_t ksnn_lock; /* serialise */ struct list_head ksnn_list; /* chain on global list */ - int ksnn_npeers; /* # peers */ - int ksnn_shutdown; /* shutting down? */ + atomic_t ksnn_npeers; /* # peers */ int ksnn_ninterfaces; /* IP interfaces */ - struct ksock_interface ksnn_interfaces[LNET_NUM_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 @@ -194,13 +180,12 @@ struct ksock_nal_data { /* stabilize peer_ni/conn ops */ rwlock_t ksnd_global_lock; /* hash table of all my known peers */ - struct list_head *ksnd_peers; - int ksnd_peer_hash_size; /* size of ksnd_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_info **ksnd_sched_info; + struct ksock_sched **ksnd_schedulers; atomic_t ksnd_nactive_txs; /* #active txs */ @@ -250,19 +235,17 @@ struct ksock_nal_data { #define SOCKNAL_INIT_DATA 1 #define SOCKNAL_INIT_ALL 2 -/* A packet just assembled for transmission is represented by 1 or more - * struct kvec fragments (the first frag contains the portals header), - * followed by 0 or more lnet_kiov_t fragments. +/* A packet just assembled for transmission is represented by 1 + * struct iovec fragment - the portals header - followed by 0 + * or more struct bio_vec fragments. * * 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 kvec or lnet_kiov_t 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_proto; /* forward ref */ + * received into struct bio_vec fragments. + */ +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 */ @@ -271,38 +254,29 @@ struct ksock_tx { /* transmit packet */ 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 */ - lnet_kiov_t *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 */ - union { - struct { - struct kvec iov; /* virt hdr */ - lnet_kiov_t kiov[0]; /* paged payload */ - } paged; - struct { - struct kvec iov[1]; /* virt hdr + payload */ - } virt; - } tx_frags; + 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 */ + struct kvec tx_hdr; /* virt hdr */ + struct bio_vec tx_payload[0]; /* paged payload */ }; -#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 */ +#define KSOCK_NOOP_TX_SIZE ((int)offsetof(struct ksock_tx, tx_payload[0])) /* 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]; - lnet_kiov_t kiov[LNET_MAX_IOV]; + struct bio_vec kiov[LNET_MAX_IOV]; }; #define SOCKNAL_RX_KSM_HEADER 1 /* reading ksock message header */ @@ -330,7 +304,7 @@ struct ksock_conn { 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 */ - struct ksock_proto *ksnc_proto; /* protocol for the connection */ + const struct ksock_proto *ksnc_proto; /* protocol for the connection */ /* READER */ @@ -345,11 +319,12 @@ struct ksock_conn { 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 */ - lnet_kiov_t *ksnc_rx_kiov; /* the page 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 */ - void *ksnc_cookie; /* rx lnet_finalize passthru arg */ + __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 @@ -379,47 +354,48 @@ struct ksock_conn { }; 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; /* 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_scheduled:1; /* scheduled for attention */ - unsigned int ksnr_connecting:1;/* connection establishment in progress */ - unsigned int ksnr_connected:4; /* connections established by type */ - unsigned int ksnr_deleted:1; /* been removed from peer_ni? */ - unsigned int ksnr_share_count; /* created explicitly? */ - int ksnr_conn_count; /* # conns established by this 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 list_head ksnp_list; /* stash on global peer_ni list */ + 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_sharecount; /* lconf usage counter */ - 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 */ - struct ksock_proto *ksnp_proto; /* latest known peer_ni protocol */ + 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 */ + 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_NUM_INTERFACES]; /* preferred local interfaces */ + 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 { @@ -455,9 +431,9 @@ struct ksock_proto { * return MATCH_MAY : can be backup */ }; -extern struct ksock_proto ksocknal_protocol_v1x; -extern struct ksock_proto ksocknal_protocol_v2x; -extern struct ksock_proto ksocknal_protocol_v3x; +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 @@ -482,20 +458,12 @@ static inline __u32 ksocknal_csum(__u32 crc, unsigned char const *p, size_t len) static inline int ksocknal_route_mask(void) { - if (!*ksocknal_tunables.ksnd_typed_conns) - return (1 << SOCKLND_CONN_ANY); - - return ((1 << SOCKLND_CONN_CONTROL) | - (1 << SOCKLND_CONN_BULK_IN) | - (1 << SOCKLND_CONN_BULK_OUT)); -} - -static inline struct list_head * -ksocknal_nid2peerlist (lnet_nid_t nid) -{ - unsigned int hash = ((unsigned int)nid) % ksocknal_data.ksnd_peer_hash_size; + if (!*ksocknal_tunables.ksnd_typed_conns) + return BIT(SOCKLND_CONN_ANY); - return (&ksocknal_data.ksnd_peers [hash]); + return (BIT(SOCKLND_CONN_CONTROL) | + BIT(SOCKLND_CONN_BULK_IN) | + BIT(SOCKLND_CONN_BULK_OUT)); } static inline void @@ -596,13 +564,20 @@ ksocknal_peer_decref(struct ksock_peer_ni *peer_ni) ksocknal_destroy_peer(peer_ni); } +static inline int ksocknal_timeout(void) +{ + return *ksocknal_tunables.ksnd_timeout ? + *ksocknal_tunables.ksnd_timeout : + lnet_get_lnd_timeout(); +} + 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 kvec *iov, lnet_kiov_t *kiov, + struct bio_vec *kiov, unsigned int offset, unsigned int mlen, unsigned int rlen); int ksocknal_accept(struct lnet_ni *ni, struct socket *sock); @@ -634,8 +609,8 @@ 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(struct lnet_ni *ni, lnet_nid_t gw_nid, int alive); -extern void ksocknal_query(struct lnet_ni *ni, lnet_nid_t nid, cfs_time_t *when); +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); @@ -662,11 +637,15 @@ extern void ksocknal_lib_reset_callback(struct socket *sock, 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); -extern int ksocknal_lib_send_kiov(struct ksock_conn *conn, struct ksock_tx *tx); +extern int ksocknal_lib_send_hdr(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); -extern int ksocknal_lib_recv_kiov(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);