1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
2 * vim:expandtab:shiftwidth=8:tabstop=8:
4 * Copyright (C) 2001, 2002 Cluster File Systems, Inc.
5 * Author: Zach Brown <zab@zabbo.net>
6 * Author: Peter J. Braam <braam@clusterfs.com>
7 * Author: Phil Schwan <phil@clusterfs.com>
8 * Author: Eric Barton <eric@bartonsoftware.com>
10 * This file is part of Portals, http://www.sf.net/projects/sandiaportals/
12 * Portals is free software; you can redistribute it and/or
13 * modify it under the terms of version 2 of the GNU General Public
14 * License as published by the Free Software Foundation.
16 * Portals is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 * GNU General Public License for more details.
21 * You should have received a copy of the GNU General Public License
22 * along with Portals; if not, write to the Free Software
23 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
29 * LIB functions follow
33 ksocknal_read(nal_cb_t *nal, void *private, void *dst_addr,
34 user_ptr src_addr, size_t len)
36 CDEBUG(D_NET, LPX64": reading %ld bytes from %p -> %p\n",
37 nal->ni.nid, (long)len, src_addr, dst_addr);
39 memcpy( dst_addr, src_addr, len );
44 ksocknal_write(nal_cb_t *nal, void *private, user_ptr dst_addr,
45 void *src_addr, size_t len)
47 CDEBUG(D_NET, LPX64": writing %ld bytes from %p -> %p\n",
48 nal->ni.nid, (long)len, src_addr, dst_addr);
50 memcpy( dst_addr, src_addr, len );
55 ksocknal_callback (nal_cb_t * nal, void *private, lib_eq_t *eq,
58 CDEBUG(D_NET, LPX64": callback eq %p ev %p\n",
61 if (eq->event_callback != NULL)
62 eq->event_callback(ev);
68 ksocknal_malloc(nal_cb_t *nal, size_t len)
72 PORTAL_ALLOC(buf, len);
81 ksocknal_free(nal_cb_t *nal, void *buf, size_t len)
83 PORTAL_FREE(buf, len);
87 ksocknal_printf(nal_cb_t *nal, const char *fmt, ...)
93 vsnprintf (msg, sizeof (msg), fmt, ap); /* sprint safely */
96 msg[sizeof (msg) - 1] = 0; /* ensure terminated */
98 CDEBUG (D_NET, "%s", msg);
102 ksocknal_cli(nal_cb_t *nal, unsigned long *flags)
104 ksock_nal_data_t *data = nal->nal_data;
106 spin_lock(&data->ksnd_nal_cb_lock);
110 ksocknal_sti(nal_cb_t *nal, unsigned long *flags)
112 ksock_nal_data_t *data;
113 data = nal->nal_data;
115 spin_unlock(&data->ksnd_nal_cb_lock);
119 ksocknal_dist(nal_cb_t *nal, ptl_nid_t nid, unsigned long *dist)
121 /* I would guess that if ksocknal_get_peer (nid) == NULL,
122 and we're not routing, then 'nid' is very distant :) */
123 if ( nal->ni.nid == nid ) {
133 ksocknal_get_ltx (int may_block)
136 ksock_ltx_t *ltx = NULL;
139 spin_lock_irqsave (&ksocknal_data.ksnd_idle_ltx_lock, flags);
141 if (!list_empty (&ksocknal_data.ksnd_idle_ltx_list)) {
142 ltx = list_entry(ksocknal_data.ksnd_idle_ltx_list.next,
143 ksock_ltx_t, ltx_tx.tx_list);
144 list_del (<x->ltx_tx.tx_list);
145 ksocknal_data.ksnd_active_ltxs++;
150 if (!list_empty(&ksocknal_data.ksnd_idle_nblk_ltx_list)) {
151 ltx = list_entry(ksocknal_data.ksnd_idle_nblk_ltx_list.next,
152 ksock_ltx_t, ltx_tx.tx_list);
153 list_del (<x->ltx_tx.tx_list);
154 ksocknal_data.ksnd_active_ltxs++;
159 spin_unlock_irqrestore(&ksocknal_data.ksnd_idle_ltx_lock,
162 wait_event (ksocknal_data.ksnd_idle_ltx_waitq,
163 !list_empty (&ksocknal_data.ksnd_idle_ltx_list));
166 spin_unlock_irqrestore (&ksocknal_data.ksnd_idle_ltx_lock, flags);
172 ksocknal_put_ltx (ksock_ltx_t *ltx)
176 spin_lock_irqsave (&ksocknal_data.ksnd_idle_ltx_lock, flags);
178 ksocknal_data.ksnd_active_ltxs--;
179 list_add_tail (<x->ltx_tx.tx_list, ltx->ltx_idle);
181 /* normal tx desc => wakeup anyone blocking for one */
182 if (ltx->ltx_idle == &ksocknal_data.ksnd_idle_ltx_list)
183 wake_up (&ksocknal_data.ksnd_idle_ltx_waitq);
185 spin_unlock_irqrestore (&ksocknal_data.ksnd_idle_ltx_lock, flags);
190 ksocknal_kvaddr_to_page (unsigned long vaddr)
194 if (vaddr >= VMALLOC_START &&
196 page = vmalloc_to_page ((void *)vaddr);
198 else if (vaddr >= PKMAP_BASE &&
199 vaddr < (PKMAP_BASE + LAST_PKMAP * PAGE_SIZE))
200 page = vmalloc_to_page ((void *)vaddr);
201 /* in 2.4 ^ just walks the page tables */
204 page = virt_to_page (vaddr);
215 ksocknal_send_iov (ksock_conn_t *conn, ksock_tx_t *tx)
217 struct socket *sock = conn->ksnc_sock;
218 struct iovec *iov = tx->tx_iov;
219 int fragsize = iov->iov_len;
220 unsigned long vaddr = (unsigned long)iov->iov_base;
221 int more = (tx->tx_niov > 1) ||
222 (tx->tx_nkiov > 0) ||
223 (!list_empty (&conn->ksnc_tx_queue));
225 int offset = vaddr & (PAGE_SIZE - 1);
226 int zcsize = MIN (fragsize, PAGE_SIZE - offset);
231 /* NB we can't trust socket ops to either consume our iovs
232 * or leave them alone, so we only send 1 frag at a time. */
233 LASSERT (fragsize <= tx->tx_resid);
234 LASSERT (tx->tx_niov > 0);
237 if (zcsize >= ksocknal_data.ksnd_zc_min_frag &&
238 (sock->sk->route_caps & NETIF_F_SG) &&
239 (sock->sk->route_caps & (NETIF_F_IP_CSUM | NETIF_F_NO_CSUM | NETIF_F_HW_CSUM)) &&
240 (page = ksocknal_kvaddr_to_page (vaddr)) != NULL) {
242 CDEBUG(D_NET, "vaddr %p, page %p->%p + offset %x for %d\n",
243 (void *)vaddr, page, page_address(page), offset, zcsize);
245 if (fragsize > zcsize) {
250 rc = tcp_sendpage_zccd(sock, page, offset, zcsize,
251 more ? (MSG_DONTWAIT | MSG_MORE) : MSG_DONTWAIT,
256 /* NB don't pass tx's iov; sendmsg may or may not update it */
257 struct iovec fragiov = { .iov_base = (void *)vaddr,
258 .iov_len = fragsize};
259 struct msghdr msg = {
266 .msg_flags = more ? (MSG_DONTWAIT | MSG_MORE) : MSG_DONTWAIT
268 mm_segment_t oldmm = get_fs();
271 rc = sock_sendmsg(sock, &msg, fragsize);
280 if (rc < iov->iov_len) {
281 /* didn't send whole iov entry... */
282 iov->iov_base = (void *)(vaddr + rc);
284 /* ...but did we send everything we tried to send? */
285 return ((rc == fragsize) ? 1 : -EAGAIN);
294 ksocknal_send_kiov (ksock_conn_t *conn, ksock_tx_t *tx)
296 struct socket *sock = conn->ksnc_sock;
297 ptl_kiov_t *kiov = tx->tx_kiov;
298 int fragsize = kiov->kiov_len;
299 struct page *page = kiov->kiov_page;
300 int offset = kiov->kiov_offset;
301 int more = (tx->tx_nkiov > 1) ||
302 (!list_empty (&conn->ksnc_tx_queue));
305 /* NB we can't trust socket ops to either consume our iovs
306 * or leave them alone, so we only send 1 frag at a time. */
307 LASSERT (fragsize <= tx->tx_resid);
308 LASSERT (offset + fragsize <= PAGE_SIZE);
309 LASSERT (tx->tx_niov == 0);
310 LASSERT (tx->tx_nkiov > 0);
313 if (fragsize >= ksocknal_data.ksnd_zc_min_frag &&
314 (sock->sk->route_caps & NETIF_F_SG) &&
315 (sock->sk->route_caps & (NETIF_F_IP_CSUM | NETIF_F_NO_CSUM | NETIF_F_HW_CSUM))) {
317 CDEBUG(D_NET, "page %p + offset %x for %d\n",
318 page, offset, fragsize);
320 rc = tcp_sendpage_zccd(sock, page, offset, fragsize,
321 more ? (MSG_DONTWAIT | MSG_MORE) : MSG_DONTWAIT,
326 char *addr = ((char *)kmap (page)) + offset;
327 struct iovec fragiov = {.iov_base = addr,
328 .iov_len = fragsize};
329 struct msghdr msg = {
336 .msg_flags = more ? (MSG_DONTWAIT | MSG_MORE) : MSG_DONTWAIT
338 mm_segment_t oldmm = get_fs();
341 rc = sock_sendmsg(sock, &msg, fragsize);
353 /* didn't send whole frag */
354 kiov->kiov_offset = offset + rc;
355 kiov->kiov_len = fragsize - rc;
359 /* everything went */
360 LASSERT (rc == fragsize);
367 ksocknal_sendmsg (ksock_conn_t *conn, ksock_tx_t *tx)
369 /* Return 0 on success, < 0 on error.
370 * caller checks tx_resid to determine progress/completion */
374 if (ksocknal_data.ksnd_stall_tx != 0) {
375 set_current_state (TASK_UNINTERRUPTIBLE);
376 schedule_timeout (ksocknal_data.ksnd_stall_tx * HZ);
379 rc = ksocknal_getconnsock (conn);
384 LASSERT (tx->tx_resid != 0);
386 if (conn->ksnc_closing) {
391 if (tx->tx_niov != 0)
392 rc = ksocknal_send_iov (conn, tx);
394 rc = ksocknal_send_kiov (conn, tx);
396 if (rc <= 0) { /* error or socket full? */
397 /* NB: rc == 0 and rc == -EAGAIN both mean try
398 * again later (linux stack returns -EAGAIN for
399 * this, but Adaptech TOE returns 0) */
405 /* Consider the connection alive since we managed to chuck
406 * more data into it. Really, we'd like to consider it
407 * alive only when the peer ACKs something, but
408 * write_space() only gets called back while SOCK_NOSPACE
409 * is set. Instead, we presume peer death has occurred if
410 * the socket doesn't drain within a timout */
411 conn->ksnc_tx_deadline = jiffies +
412 ksocknal_data.ksnd_io_timeout * HZ;
413 conn->ksnc_peer->ksnp_last_alive = jiffies;
415 if (tx->tx_resid == 0) { /* sent everything */
421 ksocknal_putconnsock (conn);
426 ksocknal_eager_ack (ksock_conn_t *conn)
429 mm_segment_t oldmm = get_fs();
430 struct socket *sock = conn->ksnc_sock;
432 /* Remind the socket to ACK eagerly. If I don't, the socket might
433 * think I'm about to send something it could piggy-back the ACK
434 * on, introducing delay in completing zero-copy sends in my
438 sock->ops->setsockopt (sock, SOL_TCP, TCP_QUICKACK,
439 (char *)&opt, sizeof (opt));
444 ksocknal_recv_iov (ksock_conn_t *conn)
446 struct iovec *iov = conn->ksnc_rx_iov;
447 int fragsize = iov->iov_len;
448 unsigned long vaddr = (unsigned long)iov->iov_base;
449 struct iovec fragiov = { .iov_base = (void *)vaddr,
450 .iov_len = fragsize};
451 struct msghdr msg = {
460 mm_segment_t oldmm = get_fs();
463 /* NB we can't trust socket ops to either consume our iovs
464 * or leave them alone, so we only receive 1 frag at a time. */
465 LASSERT (conn->ksnc_rx_niov > 0);
466 LASSERT (fragsize <= conn->ksnc_rx_nob_wanted);
469 rc = sock_recvmsg (conn->ksnc_sock, &msg, fragsize, MSG_DONTWAIT);
470 /* NB this is just a boolean............................^ */
476 /* received something... */
477 conn->ksnc_peer->ksnp_last_alive = jiffies;
478 conn->ksnc_rx_deadline = jiffies +
479 ksocknal_data.ksnd_io_timeout * HZ;
480 mb(); /* order with setting rx_started */
481 conn->ksnc_rx_started = 1;
483 conn->ksnc_rx_nob_wanted -= rc;
484 conn->ksnc_rx_nob_left -= rc;
487 iov->iov_base = (void *)(vaddr + rc);
488 iov->iov_len = fragsize - rc;
493 conn->ksnc_rx_niov--;
498 ksocknal_recv_kiov (ksock_conn_t *conn)
500 ptl_kiov_t *kiov = conn->ksnc_rx_kiov;
501 struct page *page = kiov->kiov_page;
502 int offset = kiov->kiov_offset;
503 int fragsize = kiov->kiov_len;
504 unsigned long vaddr = ((unsigned long)kmap (page)) + offset;
505 struct iovec fragiov = { .iov_base = (void *)vaddr,
506 .iov_len = fragsize};
507 struct msghdr msg = {
516 mm_segment_t oldmm = get_fs();
519 /* NB we can't trust socket ops to either consume our iovs
520 * or leave them alone, so we only receive 1 frag at a time. */
521 LASSERT (fragsize <= conn->ksnc_rx_nob_wanted);
522 LASSERT (conn->ksnc_rx_nkiov > 0);
523 LASSERT (offset + fragsize <= PAGE_SIZE);
526 rc = sock_recvmsg (conn->ksnc_sock, &msg, fragsize, MSG_DONTWAIT);
527 /* NB this is just a boolean............................^ */
535 /* received something... */
536 conn->ksnc_peer->ksnp_last_alive = jiffies;
537 conn->ksnc_rx_deadline = jiffies +
538 ksocknal_data.ksnd_io_timeout * HZ;
539 mb(); /* order with setting rx_started */
540 conn->ksnc_rx_started = 1;
542 conn->ksnc_rx_nob_wanted -= rc;
543 conn->ksnc_rx_nob_left -= rc;
546 kiov->kiov_offset = offset + rc;
547 kiov->kiov_len = fragsize - rc;
551 conn->ksnc_rx_kiov++;
552 conn->ksnc_rx_nkiov--;
557 ksocknal_recvmsg (ksock_conn_t *conn)
559 /* Return 1 on success, 0 on EOF, < 0 on error.
560 * Caller checks ksnc_rx_nob_wanted to determine
561 * progress/completion. */
565 if (ksocknal_data.ksnd_stall_rx != 0) {
566 set_current_state (TASK_UNINTERRUPTIBLE);
567 schedule_timeout (ksocknal_data.ksnd_stall_rx * HZ);
570 rc = ksocknal_getconnsock (conn);
575 if (conn->ksnc_closing) {
580 if (conn->ksnc_rx_niov != 0)
581 rc = ksocknal_recv_iov (conn);
583 rc = ksocknal_recv_kiov (conn);
586 /* error/EOF or partial receive */
589 } else if (rc == 0 && conn->ksnc_rx_started) {
590 /* EOF in the middle of a message */
596 /* Completed a fragment */
598 if (conn->ksnc_rx_nob_wanted == 0) {
599 /* Completed a message segment (header or payload) */
600 if ((ksocknal_data.ksnd_eager_ack & conn->ksnc_type) != 0 &&
601 (conn->ksnc_rx_state == SOCKNAL_RX_BODY ||
602 conn->ksnc_rx_state == SOCKNAL_RX_BODY_FWD)) {
603 /* Remind the socket to ack eagerly... */
604 ksocknal_eager_ack(conn);
611 ksocknal_putconnsock (conn);
617 ksocknal_zc_callback (zccd_t *zcd)
619 ksock_tx_t *tx = KSOCK_ZCCD_2_TX(zcd);
620 ksock_sched_t *sched = tx->tx_conn->ksnc_scheduler;
624 /* Schedule tx for cleanup (can't do it now due to lock conflicts) */
626 spin_lock_irqsave (&sched->kss_lock, flags);
628 list_add_tail (&tx->tx_list, &sched->kss_zctxdone_list);
629 wake_up (&sched->kss_waitq);
631 spin_unlock_irqrestore (&sched->kss_lock, flags);
637 ksocknal_tx_done (ksock_tx_t *tx, int asynch)
642 if (tx->tx_conn != NULL) {
643 /* This tx got queued on a conn; do the accounting... */
644 atomic_sub (tx->tx_nob, &tx->tx_conn->ksnc_tx_nob);
646 /* zero copy completion isn't always from
647 * process_transmit() so it needs to keep a ref on
650 ksocknal_put_conn (tx->tx_conn);
656 if (tx->tx_isfwd) { /* was a forwarded packet? */
657 kpr_fwd_done (&ksocknal_data.ksnd_router,
658 KSOCK_TX_2_KPR_FWD_DESC (tx), 0);
664 ltx = KSOCK_TX_2_KSOCK_LTX (tx);
666 lib_finalize (&ksocknal_lib, ltx->ltx_private, ltx->ltx_cookie);
668 ksocknal_put_ltx (ltx);
673 ksocknal_tx_launched (ksock_tx_t *tx)
676 if (atomic_read (&tx->tx_zccd.zccd_count) != 1) {
677 ksock_conn_t *conn = tx->tx_conn;
679 /* zccd skbufs are still in-flight. First take a ref on
680 * conn, so it hangs about for ksocknal_tx_done... */
681 atomic_inc (&conn->ksnc_refcount);
683 /* ...then drop the initial ref on zccd, so the zero copy
684 * callback can occur */
685 zccd_put (&tx->tx_zccd);
689 /* Any zero-copy-ness (if any) has completed; I can complete the
690 * transmit now, avoiding an extra schedule */
691 ksocknal_tx_done (tx, 0);
695 ksocknal_process_transmit (ksock_conn_t *conn, ksock_tx_t *tx)
699 rc = ksocknal_sendmsg (conn, tx);
701 CDEBUG (D_NET, "send(%d) %d\n", tx->tx_resid, rc);
702 LASSERT (rc != -EAGAIN);
706 if (tx->tx_resid != 0) {
707 /* didn't send everything */
711 ksocknal_tx_launched (tx);
715 if (!conn->ksnc_closing)
716 CERROR ("[%p] Error %d on write to "LPX64" ip %08x:%d\n",
717 conn, rc, conn->ksnc_peer->ksnp_nid,
718 conn->ksnc_ipaddr, conn->ksnc_port);
720 ksocknal_close_conn_and_siblings (conn, rc);
721 ksocknal_tx_launched (tx);
727 ksocknal_launch_autoconnect_locked (ksock_route_t *route)
731 /* called holding write lock on ksnd_global_lock */
733 LASSERT (!route->ksnr_deleted);
734 LASSERT ((route->ksnr_connected & (1 << SOCKNAL_CONN_ANY)) == 0);
735 LASSERT ((route->ksnr_connected & KSNR_TYPED_ROUTES) != KSNR_TYPED_ROUTES);
736 LASSERT (!route->ksnr_connecting);
738 if (ksocknal_data.ksnd_typed_conns)
739 route->ksnr_connecting =
740 KSNR_TYPED_ROUTES & ~route->ksnr_connected;
742 route->ksnr_connecting = (1 << SOCKNAL_CONN_ANY);
744 atomic_inc (&route->ksnr_refcount); /* extra ref for asynchd */
746 spin_lock_irqsave (&ksocknal_data.ksnd_autoconnectd_lock, flags);
748 list_add_tail (&route->ksnr_connect_list,
749 &ksocknal_data.ksnd_autoconnectd_routes);
750 wake_up (&ksocknal_data.ksnd_autoconnectd_waitq);
752 spin_unlock_irqrestore (&ksocknal_data.ksnd_autoconnectd_lock, flags);
756 ksocknal_find_target_peer_locked (ksock_tx_t *tx, ptl_nid_t nid)
758 ptl_nid_t target_nid;
760 ksock_peer_t *peer = ksocknal_find_peer_locked (nid);
766 CERROR ("Can't send packet to "LPX64
767 ": routed target is not a peer\n", nid);
771 rc = kpr_lookup (&ksocknal_data.ksnd_router, nid, tx->tx_nob,
774 CERROR ("Can't route to "LPX64": router error %d\n", nid, rc);
778 peer = ksocknal_find_peer_locked (target_nid);
782 CERROR ("Can't send packet to "LPX64": no peer entry\n", target_nid);
787 ksocknal_find_conn_locked (ksock_tx_t *tx, ksock_peer_t *peer)
789 struct list_head *tmp;
790 ksock_conn_t *typed = NULL;
792 ksock_conn_t *fallback = NULL;
795 /* Find the conn with the shortest tx queue */
796 list_for_each (tmp, &peer->ksnp_conns) {
797 ksock_conn_t *c = list_entry(tmp, ksock_conn_t, ksnc_list);
798 int nob = atomic_read(&c->ksnc_tx_nob);
800 LASSERT (!c->ksnc_closing);
802 if (fallback == NULL || nob < fnob) {
807 if (!ksocknal_data.ksnd_typed_conns)
810 switch (c->ksnc_type) {
813 case SOCKNAL_CONN_ANY:
815 case SOCKNAL_CONN_BULK_IN:
817 case SOCKNAL_CONN_BULK_OUT:
818 if (tx->tx_nob < ksocknal_data.ksnd_min_bulk)
821 case SOCKNAL_CONN_CONTROL:
822 if (tx->tx_nob >= ksocknal_data.ksnd_min_bulk)
827 if (typed == NULL || nob < tnob) {
833 /* prefer the typed selection */
834 return ((typed != NULL) ? typed : fallback);
838 ksocknal_queue_tx_locked (ksock_tx_t *tx, ksock_conn_t *conn)
841 ksock_sched_t *sched = conn->ksnc_scheduler;
843 /* called holding global lock (read or irq-write) */
845 CDEBUG (D_NET, "Sending to "LPX64" on port %d\n",
846 conn->ksnc_peer->ksnp_nid, conn->ksnc_port);
848 atomic_add (tx->tx_nob, &conn->ksnc_tx_nob);
849 tx->tx_resid = tx->tx_nob;
853 zccd_init (&tx->tx_zccd, ksocknal_zc_callback);
854 /* NB this sets 1 ref on zccd, so the callback can only occur after
855 * I've released this ref. */
858 spin_lock_irqsave (&sched->kss_lock, flags);
860 conn->ksnc_tx_deadline = jiffies +
861 ksocknal_data.ksnd_io_timeout * HZ;
862 mb(); /* order with list_add_tail */
864 list_add_tail (&tx->tx_list, &conn->ksnc_tx_queue);
866 if (conn->ksnc_tx_ready && /* able to send */
867 !conn->ksnc_tx_scheduled) { /* not scheduled to send */
868 /* +1 ref for scheduler */
869 atomic_inc (&conn->ksnc_refcount);
870 list_add_tail (&conn->ksnc_tx_list,
871 &sched->kss_tx_conns);
872 conn->ksnc_tx_scheduled = 1;
873 wake_up (&sched->kss_waitq);
876 spin_unlock_irqrestore (&sched->kss_lock, flags);
880 ksocknal_find_connectable_route_locked (ksock_peer_t *peer)
882 struct list_head *tmp;
883 ksock_route_t *route;
884 ksock_route_t *candidate = NULL;
888 list_for_each (tmp, &peer->ksnp_routes) {
889 route = list_entry (tmp, ksock_route_t, ksnr_list);
890 bits = route->ksnr_connected;
892 if ((bits & KSNR_TYPED_ROUTES) == KSNR_TYPED_ROUTES ||
893 (bits & (1 << SOCKNAL_CONN_ANY)) != 0 ||
894 route->ksnr_connecting != 0) {
895 /* All typed connections have been established, or
896 * an untyped connection has been established, or
897 * connections are currently being established */
902 /* too soon to retry this guy? */
903 if (!time_after_eq (jiffies, route->ksnr_timeout))
906 /* always do eager routes */
907 if (route->ksnr_eager)
910 if (candidate == NULL) {
911 /* If we don't find any other route that is fully
912 * connected or connecting, the first connectable
913 * route is returned. If it fails to connect, it
914 * will get placed at the end of the list */
919 return (found ? NULL : candidate);
923 ksocknal_find_connecting_route_locked (ksock_peer_t *peer)
925 struct list_head *tmp;
926 ksock_route_t *route;
928 list_for_each (tmp, &peer->ksnp_routes) {
929 route = list_entry (tmp, ksock_route_t, ksnr_list);
931 if (route->ksnr_connecting != 0)
939 ksocknal_launch_packet (ksock_tx_t *tx, ptl_nid_t nid)
944 ksock_route_t *route;
947 /* Ensure the frags we've been given EXACTLY match the number of
948 * bytes we want to send. Many TCP/IP stacks disregard any total
949 * size parameters passed to them and just look at the frags.
951 * We always expect at least 1 mapped fragment containing the
952 * complete portals header. */
953 LASSERT (lib_iov_nob (tx->tx_niov, tx->tx_iov) +
954 lib_kiov_nob (tx->tx_nkiov, tx->tx_kiov) == tx->tx_nob);
955 LASSERT (tx->tx_niov >= 1);
956 LASSERT (tx->tx_iov[0].iov_len >= sizeof (ptl_hdr_t));
958 CDEBUG (D_NET, "packet %p type %d, nob %d niov %d nkiov %d\n",
959 tx, ((ptl_hdr_t *)tx->tx_iov[0].iov_base)->type,
960 tx->tx_nob, tx->tx_niov, tx->tx_nkiov);
962 tx->tx_conn = NULL; /* only set when assigned a conn */
964 g_lock = &ksocknal_data.ksnd_global_lock;
967 peer = ksocknal_find_target_peer_locked (tx, nid);
969 read_unlock (g_lock);
970 return (-EHOSTUNREACH);
973 if (ksocknal_find_connectable_route_locked(peer) == NULL) {
974 conn = ksocknal_find_conn_locked (tx, peer);
976 /* I've got no autoconnect routes that need to be
977 * connecting and I do have an actual connection... */
978 ksocknal_queue_tx_locked (tx, conn);
979 read_unlock (g_lock);
984 /* Making one or more connections; I'll need a write lock... */
986 atomic_inc (&peer->ksnp_refcount); /* +1 ref for me while I unlock */
987 read_unlock (g_lock);
988 write_lock_irqsave (g_lock, flags);
990 if (peer->ksnp_closing) { /* peer deleted as I blocked! */
991 write_unlock_irqrestore (g_lock, flags);
992 ksocknal_put_peer (peer);
993 return (-EHOSTUNREACH);
995 ksocknal_put_peer (peer); /* drop ref I got above */
998 /* launch any/all autoconnections that need it */
999 route = ksocknal_find_connectable_route_locked (peer);
1003 ksocknal_launch_autoconnect_locked (route);
1006 conn = ksocknal_find_conn_locked (tx, peer);
1008 /* Connection exists; queue message on it */
1009 ksocknal_queue_tx_locked (tx, conn);
1010 write_unlock_irqrestore (g_lock, flags);
1014 route = ksocknal_find_connecting_route_locked (peer);
1015 if (route != NULL) {
1016 /* At least 1 connection is being established; queue the
1018 list_add_tail (&tx->tx_list, &peer->ksnp_tx_queue);
1019 write_unlock_irqrestore (g_lock, flags);
1023 write_unlock_irqrestore (g_lock, flags);
1024 return (-EHOSTUNREACH);
1028 ksocknal_setup_hdr (nal_cb_t *nal, void *private, lib_msg_t *cookie,
1029 ptl_hdr_t *hdr, int type)
1033 /* I may not block for a transmit descriptor if I might block the
1034 * receiver, or an interrupt handler. */
1035 ltx = ksocknal_get_ltx (!(type == PTL_MSG_ACK ||
1036 type == PTL_MSG_REPLY ||
1039 CERROR ("Can't allocate tx desc\n");
1043 /* Init local send packet (storage for hdr, finalize() args) */
1044 ltx->ltx_hdr = *hdr;
1045 ltx->ltx_private = private;
1046 ltx->ltx_cookie = cookie;
1048 /* Init common ltx_tx */
1049 ltx->ltx_tx.tx_isfwd = 0;
1050 ltx->ltx_tx.tx_nob = sizeof (*hdr);
1052 /* We always have 1 mapped frag for the header */
1053 ltx->ltx_tx.tx_niov = 1;
1054 ltx->ltx_tx.tx_iov = <x->ltx_iov_space.hdr;
1055 ltx->ltx_tx.tx_iov[0].iov_base = <x->ltx_hdr;
1056 ltx->ltx_tx.tx_iov[0].iov_len = sizeof (ltx->ltx_hdr);
1058 ltx->ltx_tx.tx_kiov = NULL;
1059 ltx->ltx_tx.tx_nkiov = 0;
1065 ksocknal_send (nal_cb_t *nal, void *private, lib_msg_t *cookie,
1066 ptl_hdr_t *hdr, int type, ptl_nid_t nid, ptl_pid_t pid,
1067 unsigned int payload_niov, struct iovec *payload_iov,
1073 /* NB 'private' is different depending on what we're sending.
1074 * Just ignore it until we can rely on it
1078 "sending "LPSZ" bytes in %d mapped frags to nid: "LPX64
1079 " pid %d\n", payload_len, payload_niov, nid, pid);
1081 ltx = ksocknal_setup_hdr (nal, private, cookie, hdr, type);
1085 /* append the payload_iovs to the one pointing at the header */
1086 LASSERT (ltx->ltx_tx.tx_niov == 1 && ltx->ltx_tx.tx_nkiov == 0);
1087 LASSERT (payload_niov <= PTL_MD_MAX_IOV);
1089 memcpy (ltx->ltx_tx.tx_iov + 1, payload_iov,
1090 payload_niov * sizeof (*payload_iov));
1091 ltx->ltx_tx.tx_niov = 1 + payload_niov;
1092 ltx->ltx_tx.tx_nob = sizeof (*hdr) + payload_len;
1094 rc = ksocknal_launch_packet (<x->ltx_tx, nid);
1098 ksocknal_put_ltx (ltx);
1103 ksocknal_send_pages (nal_cb_t *nal, void *private, lib_msg_t *cookie,
1104 ptl_hdr_t *hdr, int type, ptl_nid_t nid, ptl_pid_t pid,
1105 unsigned int payload_niov, ptl_kiov_t *payload_iov, size_t payload_len)
1110 /* NB 'private' is different depending on what we're sending.
1111 * Just ignore it until we can rely on it */
1114 "sending "LPSZ" bytes in %d mapped frags to nid: "LPX64" pid %d\n",
1115 payload_len, payload_niov, nid, pid);
1117 ltx = ksocknal_setup_hdr (nal, private, cookie, hdr, type);
1121 LASSERT (ltx->ltx_tx.tx_niov == 1 && ltx->ltx_tx.tx_nkiov == 0);
1122 LASSERT (payload_niov <= PTL_MD_MAX_IOV);
1124 ltx->ltx_tx.tx_kiov = ltx->ltx_iov_space.payload.kiov;
1125 memcpy (ltx->ltx_tx.tx_kiov, payload_iov,
1126 payload_niov * sizeof (*payload_iov));
1127 ltx->ltx_tx.tx_nkiov = payload_niov;
1128 ltx->ltx_tx.tx_nob = sizeof (*hdr) + payload_len;
1130 rc = ksocknal_launch_packet (<x->ltx_tx, nid);
1134 ksocknal_put_ltx (ltx);
1139 ksocknal_fwd_packet (void *arg, kpr_fwd_desc_t *fwd)
1141 ptl_nid_t nid = fwd->kprfd_gateway_nid;
1142 ksock_tx_t *tx = (ksock_tx_t *)&fwd->kprfd_scratch;
1145 CDEBUG (D_NET, "Forwarding [%p] -> "LPX64" ("LPX64"))\n", fwd,
1146 fwd->kprfd_gateway_nid, fwd->kprfd_target_nid);
1148 /* I'm the gateway; must be the last hop */
1149 if (nid == ksocknal_lib.ni.nid)
1150 nid = fwd->kprfd_target_nid;
1152 tx->tx_isfwd = 1; /* This is a forwarding packet */
1153 tx->tx_nob = fwd->kprfd_nob;
1154 tx->tx_niov = fwd->kprfd_niov;
1155 tx->tx_iov = fwd->kprfd_iov;
1158 tx->tx_hdr = (ptl_hdr_t *)fwd->kprfd_iov[0].iov_base;
1160 rc = ksocknal_launch_packet (tx, nid);
1162 kpr_fwd_done (&ksocknal_data.ksnd_router, fwd, rc);
1166 ksocknal_thread_start (int (*fn)(void *arg), void *arg)
1168 long pid = kernel_thread (fn, arg, 0);
1173 atomic_inc (&ksocknal_data.ksnd_nthreads);
1178 ksocknal_thread_fini (void)
1180 atomic_dec (&ksocknal_data.ksnd_nthreads);
1184 ksocknal_fmb_callback (void *arg, int error)
1186 ksock_fmb_t *fmb = (ksock_fmb_t *)arg;
1187 ksock_fmb_pool_t *fmp = fmb->fmb_pool;
1188 ptl_hdr_t *hdr = (ptl_hdr_t *) page_address(fmb->fmb_pages[0]);
1189 ksock_conn_t *conn = NULL;
1190 ksock_sched_t *sched;
1191 unsigned long flags;
1194 CERROR("Failed to route packet from "LPX64" to "LPX64": %d\n",
1195 NTOH__u64(hdr->src_nid), NTOH__u64(hdr->dest_nid),
1198 CDEBUG (D_NET, "routed packet from "LPX64" to "LPX64": OK\n",
1199 NTOH__u64 (hdr->src_nid), NTOH__u64 (hdr->dest_nid));
1201 /* drop peer ref taken on init */
1202 ksocknal_put_peer (fmb->fmb_peer);
1204 spin_lock_irqsave (&fmp->fmp_lock, flags);
1206 list_add (&fmb->fmb_list, &fmp->fmp_idle_fmbs);
1208 if (!list_empty (&fmp->fmp_blocked_conns)) {
1209 conn = list_entry (fmb->fmb_pool->fmp_blocked_conns.next,
1210 ksock_conn_t, ksnc_rx_list);
1211 list_del (&conn->ksnc_rx_list);
1214 spin_unlock_irqrestore (&fmp->fmp_lock, flags);
1219 CDEBUG (D_NET, "Scheduling conn %p\n", conn);
1220 LASSERT (conn->ksnc_rx_scheduled);
1221 LASSERT (conn->ksnc_rx_state == SOCKNAL_RX_FMB_SLEEP);
1223 conn->ksnc_rx_state = SOCKNAL_RX_GET_FMB;
1225 sched = conn->ksnc_scheduler;
1227 spin_lock_irqsave (&sched->kss_lock, flags);
1229 list_add_tail (&conn->ksnc_rx_list, &sched->kss_rx_conns);
1230 wake_up (&sched->kss_waitq);
1232 spin_unlock_irqrestore (&sched->kss_lock, flags);
1236 ksocknal_get_idle_fmb (ksock_conn_t *conn)
1238 int payload_nob = conn->ksnc_rx_nob_left;
1239 int packet_nob = sizeof (ptl_hdr_t) + payload_nob;
1240 unsigned long flags;
1241 ksock_fmb_pool_t *pool;
1244 LASSERT (conn->ksnc_rx_state == SOCKNAL_RX_GET_FMB);
1245 LASSERT (ksocknal_data.ksnd_fmbs != NULL);
1247 if (packet_nob <= SOCKNAL_SMALL_FWD_PAGES * PAGE_SIZE)
1248 pool = &ksocknal_data.ksnd_small_fmp;
1250 pool = &ksocknal_data.ksnd_large_fmp;
1252 spin_lock_irqsave (&pool->fmp_lock, flags);
1254 if (!list_empty (&pool->fmp_idle_fmbs)) {
1255 fmb = list_entry(pool->fmp_idle_fmbs.next,
1256 ksock_fmb_t, fmb_list);
1257 list_del (&fmb->fmb_list);
1258 spin_unlock_irqrestore (&pool->fmp_lock, flags);
1263 /* deschedule until fmb free */
1265 conn->ksnc_rx_state = SOCKNAL_RX_FMB_SLEEP;
1267 list_add_tail (&conn->ksnc_rx_list,
1268 &pool->fmp_blocked_conns);
1270 spin_unlock_irqrestore (&pool->fmp_lock, flags);
1275 ksocknal_init_fmb (ksock_conn_t *conn, ksock_fmb_t *fmb)
1277 int payload_nob = conn->ksnc_rx_nob_left;
1278 int packet_nob = sizeof (ptl_hdr_t) + payload_nob;
1279 ptl_nid_t dest_nid = NTOH__u64 (conn->ksnc_hdr.dest_nid);
1280 int niov; /* at least the header */
1283 LASSERT (conn->ksnc_rx_scheduled);
1284 LASSERT (conn->ksnc_rx_state == SOCKNAL_RX_GET_FMB);
1285 LASSERT (conn->ksnc_rx_nob_wanted == conn->ksnc_rx_nob_left);
1286 LASSERT (payload_nob >= 0);
1287 LASSERT (packet_nob <= fmb->fmb_npages * PAGE_SIZE);
1288 LASSERT (sizeof (ptl_hdr_t) < PAGE_SIZE);
1290 /* Got a forwarding buffer; copy the header we just read into the
1291 * forwarding buffer. If there's payload, start reading reading it
1292 * into the buffer, otherwise the forwarding buffer can be kicked
1295 * NB fmb->fmb_iov spans the WHOLE packet.
1296 * conn->ksnc_rx_iov spans just the payload.
1298 fmb->fmb_iov[0].iov_base = page_address (fmb->fmb_pages[0]);
1301 memcpy (fmb->fmb_iov[0].iov_base, &conn->ksnc_hdr, sizeof (ptl_hdr_t));
1303 /* Take a ref on the conn's peer to prevent module unload before
1304 * forwarding completes. NB we ref peer and not conn since because
1305 * all refs on conn after it has been closed must remove themselves
1307 fmb->fmb_peer = conn->ksnc_peer;
1308 atomic_inc (&conn->ksnc_peer->ksnp_refcount);
1310 if (payload_nob == 0) { /* got complete packet already */
1311 CDEBUG (D_NET, "%p "LPX64"->"LPX64" %d fwd_start (immediate)\n",
1312 conn, NTOH__u64 (conn->ksnc_hdr.src_nid),
1313 dest_nid, packet_nob);
1315 fmb->fmb_iov[0].iov_len = sizeof (ptl_hdr_t);
1317 kpr_fwd_init (&fmb->fmb_fwd, dest_nid,
1318 packet_nob, 1, fmb->fmb_iov,
1319 ksocknal_fmb_callback, fmb);
1321 /* forward it now */
1322 kpr_fwd_start (&ksocknal_data.ksnd_router, &fmb->fmb_fwd);
1324 ksocknal_new_packet (conn, 0); /* on to next packet */
1329 if (packet_nob <= PAGE_SIZE) { /* whole packet fits in first page */
1330 fmb->fmb_iov[0].iov_len = packet_nob;
1332 fmb->fmb_iov[0].iov_len = PAGE_SIZE;
1333 nob = packet_nob - PAGE_SIZE;
1336 LASSERT (niov < fmb->fmb_npages);
1337 fmb->fmb_iov[niov].iov_base =
1338 page_address (fmb->fmb_pages[niov]);
1339 fmb->fmb_iov[niov].iov_len = MIN (PAGE_SIZE, nob);
1345 kpr_fwd_init (&fmb->fmb_fwd, dest_nid,
1346 packet_nob, niov, fmb->fmb_iov,
1347 ksocknal_fmb_callback, fmb);
1349 conn->ksnc_cookie = fmb; /* stash fmb for later */
1350 conn->ksnc_rx_state = SOCKNAL_RX_BODY_FWD; /* read in the payload */
1352 /* payload is desc's iov-ed buffer, but skipping the hdr */
1353 LASSERT (niov <= sizeof (conn->ksnc_rx_iov_space) /
1354 sizeof (struct iovec));
1356 conn->ksnc_rx_iov = (struct iovec *)&conn->ksnc_rx_iov_space;
1357 conn->ksnc_rx_iov[0].iov_base =
1358 (void *)(((unsigned long)fmb->fmb_iov[0].iov_base) +
1359 sizeof (ptl_hdr_t));
1360 conn->ksnc_rx_iov[0].iov_len =
1361 fmb->fmb_iov[0].iov_len - sizeof (ptl_hdr_t);
1364 memcpy(&conn->ksnc_rx_iov[1], &fmb->fmb_iov[1],
1365 (niov - 1) * sizeof (struct iovec));
1367 conn->ksnc_rx_niov = niov;
1369 CDEBUG (D_NET, "%p "LPX64"->"LPX64" %d reading body\n", conn,
1370 NTOH__u64 (conn->ksnc_hdr.src_nid), dest_nid, payload_nob);
1375 ksocknal_fwd_parse (ksock_conn_t *conn)
1378 ptl_nid_t dest_nid = NTOH__u64 (conn->ksnc_hdr.dest_nid);
1379 ptl_nid_t src_nid = NTOH__u64 (conn->ksnc_hdr.src_nid);
1380 int body_len = NTOH__u32 (conn->ksnc_hdr.payload_length);
1381 char str[PTL_NALFMT_SIZE];
1383 CDEBUG (D_NET, "%p "LPX64"->"LPX64" %d parsing header\n", conn,
1384 src_nid, dest_nid, conn->ksnc_rx_nob_left);
1386 LASSERT (conn->ksnc_rx_state == SOCKNAL_RX_HEADER);
1387 LASSERT (conn->ksnc_rx_scheduled);
1389 if (body_len < 0) { /* length corrupt (overflow) */
1390 CERROR("dropping packet from "LPX64" (%s) for "LPX64" (%s): "
1391 "packet size %d illegal\n",
1392 src_nid, portals_nid2str(TCPNAL, src_nid, str),
1393 dest_nid, portals_nid2str(TCPNAL, dest_nid, str),
1396 ksocknal_new_packet (conn, 0); /* on to new packet */
1400 if (ksocknal_data.ksnd_fmbs == NULL) { /* not forwarding */
1401 CERROR("dropping packet from "LPX64" (%s) for "LPX64
1402 " (%s): not forwarding\n",
1403 src_nid, portals_nid2str(TCPNAL, src_nid, str),
1404 dest_nid, portals_nid2str(TCPNAL, dest_nid, str));
1405 /* on to new packet (skip this one's body) */
1406 ksocknal_new_packet (conn, body_len);
1410 if (body_len > PTL_MTU) { /* too big to forward */
1411 CERROR ("dropping packet from "LPX64" (%s) for "LPX64
1412 "(%s): packet size %d too big\n",
1413 src_nid, portals_nid2str(TCPNAL, src_nid, str),
1414 dest_nid, portals_nid2str(TCPNAL, dest_nid, str),
1416 /* on to new packet (skip this one's body) */
1417 ksocknal_new_packet (conn, body_len);
1421 /* should have gone direct */
1422 peer = ksocknal_get_peer (conn->ksnc_hdr.dest_nid);
1424 CERROR ("dropping packet from "LPX64" (%s) for "LPX64
1425 "(%s): target is a peer\n",
1426 src_nid, portals_nid2str(TCPNAL, src_nid, str),
1427 dest_nid, portals_nid2str(TCPNAL, dest_nid, str));
1428 ksocknal_put_peer (peer); /* drop ref from get above */
1430 /* on to next packet (skip this one's body) */
1431 ksocknal_new_packet (conn, body_len);
1435 conn->ksnc_rx_state = SOCKNAL_RX_GET_FMB; /* Getting FMB now */
1436 conn->ksnc_rx_nob_left = body_len; /* stash packet size */
1437 conn->ksnc_rx_nob_wanted = body_len; /* (no slop) */
1441 ksocknal_new_packet (ksock_conn_t *conn, int nob_to_skip)
1443 static char ksocknal_slop_buffer[4096];
1449 if (nob_to_skip == 0) { /* right at next packet boundary now */
1450 conn->ksnc_rx_started = 0;
1451 mb (); /* racing with timeout thread */
1453 conn->ksnc_rx_state = SOCKNAL_RX_HEADER;
1454 conn->ksnc_rx_nob_wanted = sizeof (ptl_hdr_t);
1455 conn->ksnc_rx_nob_left = sizeof (ptl_hdr_t);
1457 conn->ksnc_rx_iov = (struct iovec *)&conn->ksnc_rx_iov_space;
1458 conn->ksnc_rx_iov[0].iov_base = (char *)&conn->ksnc_hdr;
1459 conn->ksnc_rx_iov[0].iov_len = sizeof (ptl_hdr_t);
1460 conn->ksnc_rx_niov = 1;
1462 conn->ksnc_rx_kiov = NULL;
1463 conn->ksnc_rx_nkiov = 0;
1467 /* Set up to skip as much a possible now. If there's more left
1468 * (ran out of iov entries) we'll get called again */
1470 conn->ksnc_rx_state = SOCKNAL_RX_SLOP;
1471 conn->ksnc_rx_nob_left = nob_to_skip;
1472 conn->ksnc_rx_iov = (struct iovec *)&conn->ksnc_rx_iov_space;
1477 nob = MIN (nob_to_skip, sizeof (ksocknal_slop_buffer));
1479 conn->ksnc_rx_iov[niov].iov_base = ksocknal_slop_buffer;
1480 conn->ksnc_rx_iov[niov].iov_len = nob;
1485 } while (nob_to_skip != 0 && /* mustn't overflow conn's rx iov */
1486 niov < sizeof(conn->ksnc_rx_iov_space) / sizeof (struct iovec));
1488 conn->ksnc_rx_niov = niov;
1489 conn->ksnc_rx_kiov = NULL;
1490 conn->ksnc_rx_nkiov = 0;
1491 conn->ksnc_rx_nob_wanted = skipped;
1496 ksocknal_process_receive (ksock_conn_t *conn)
1501 LASSERT (atomic_read (&conn->ksnc_refcount) > 0);
1503 /* doesn't need a forwarding buffer */
1504 if (conn->ksnc_rx_state != SOCKNAL_RX_GET_FMB)
1508 fmb = ksocknal_get_idle_fmb (conn);
1510 /* conn descheduled waiting for idle fmb */
1514 if (ksocknal_init_fmb (conn, fmb)) {
1515 /* packet forwarded */
1520 /* NB: sched lock NOT held */
1521 LASSERT (conn->ksnc_rx_state == SOCKNAL_RX_HEADER ||
1522 conn->ksnc_rx_state == SOCKNAL_RX_BODY ||
1523 conn->ksnc_rx_state == SOCKNAL_RX_BODY_FWD ||
1524 conn->ksnc_rx_state == SOCKNAL_RX_SLOP);
1526 LASSERT (conn->ksnc_rx_nob_wanted > 0);
1528 rc = ksocknal_recvmsg(conn);
1532 CWARN ("[%p] EOF from "LPX64" ip %08x:%d\n",
1533 conn, conn->ksnc_peer->ksnp_nid,
1534 conn->ksnc_ipaddr, conn->ksnc_port);
1535 else if (!conn->ksnc_closing)
1536 CERROR ("[%p] Error %d on read from "LPX64" ip %08x:%d\n",
1537 conn, rc, conn->ksnc_peer->ksnp_nid,
1538 conn->ksnc_ipaddr, conn->ksnc_port);
1540 ksocknal_close_conn_and_siblings (conn, rc);
1541 return (rc == 0 ? -ESHUTDOWN : rc);
1544 if (conn->ksnc_rx_nob_wanted != 0) {
1549 switch (conn->ksnc_rx_state) {
1550 case SOCKNAL_RX_HEADER:
1551 if (conn->ksnc_hdr.type != HTON__u32(PTL_MSG_HELLO) &&
1552 NTOH__u64(conn->ksnc_hdr.dest_nid) != ksocknal_lib.ni.nid) {
1553 /* This packet isn't for me */
1554 ksocknal_fwd_parse (conn);
1555 switch (conn->ksnc_rx_state) {
1556 case SOCKNAL_RX_HEADER: /* skipped (zero payload) */
1557 return (0); /* => come back later */
1558 case SOCKNAL_RX_SLOP: /* skipping packet's body */
1559 goto try_read; /* => go read it */
1560 case SOCKNAL_RX_GET_FMB: /* forwarding */
1561 goto get_fmb; /* => go get a fwd msg buffer */
1568 /* sets wanted_len, iovs etc */
1569 lib_parse(&ksocknal_lib, &conn->ksnc_hdr, conn);
1571 if (conn->ksnc_rx_nob_wanted != 0) { /* need to get payload? */
1572 conn->ksnc_rx_state = SOCKNAL_RX_BODY;
1573 goto try_read; /* go read the payload */
1575 /* Fall through (completed packet for me) */
1577 case SOCKNAL_RX_BODY:
1578 /* payload all received */
1579 lib_finalize(&ksocknal_lib, NULL, conn->ksnc_cookie);
1582 case SOCKNAL_RX_SLOP:
1583 /* starting new packet? */
1584 if (ksocknal_new_packet (conn, conn->ksnc_rx_nob_left))
1585 return (0); /* come back later */
1586 goto try_read; /* try to finish reading slop now */
1588 case SOCKNAL_RX_BODY_FWD:
1589 /* payload all received */
1590 CDEBUG (D_NET, "%p "LPX64"->"LPX64" %d fwd_start (got body)\n",
1591 conn, NTOH__u64 (conn->ksnc_hdr.src_nid),
1592 NTOH__u64 (conn->ksnc_hdr.dest_nid),
1593 conn->ksnc_rx_nob_left);
1595 /* forward the packet. NB ksocknal_init_fmb() put fmb into
1596 * conn->ksnc_cookie */
1597 fmb = (ksock_fmb_t *)conn->ksnc_cookie;
1598 kpr_fwd_start (&ksocknal_data.ksnd_router, &fmb->fmb_fwd);
1600 /* no slop in forwarded packets */
1601 LASSERT (conn->ksnc_rx_nob_left == 0);
1603 ksocknal_new_packet (conn, 0); /* on to next packet */
1604 return (0); /* (later) */
1612 return (-EINVAL); /* keep gcc happy */
1616 ksocknal_recv (nal_cb_t *nal, void *private, lib_msg_t *msg,
1617 unsigned int niov, struct iovec *iov, size_t mlen, size_t rlen)
1619 ksock_conn_t *conn = (ksock_conn_t *)private;
1621 LASSERT (mlen <= rlen);
1622 LASSERT (niov <= PTL_MD_MAX_IOV);
1624 conn->ksnc_cookie = msg;
1625 conn->ksnc_rx_nob_wanted = mlen;
1626 conn->ksnc_rx_nob_left = rlen;
1628 conn->ksnc_rx_nkiov = 0;
1629 conn->ksnc_rx_kiov = NULL;
1630 conn->ksnc_rx_niov = niov;
1631 conn->ksnc_rx_iov = conn->ksnc_rx_iov_space.iov;
1632 memcpy (conn->ksnc_rx_iov, iov, niov * sizeof (*iov));
1635 lib_iov_nob (conn->ksnc_rx_niov, conn->ksnc_rx_iov) +
1636 lib_kiov_nob (conn->ksnc_rx_nkiov, conn->ksnc_rx_kiov));
1642 ksocknal_recv_pages (nal_cb_t *nal, void *private, lib_msg_t *msg,
1643 unsigned int niov, ptl_kiov_t *kiov, size_t mlen, size_t rlen)
1645 ksock_conn_t *conn = (ksock_conn_t *)private;
1647 LASSERT (mlen <= rlen);
1648 LASSERT (niov <= PTL_MD_MAX_IOV);
1650 conn->ksnc_cookie = msg;
1651 conn->ksnc_rx_nob_wanted = mlen;
1652 conn->ksnc_rx_nob_left = rlen;
1654 conn->ksnc_rx_niov = 0;
1655 conn->ksnc_rx_iov = NULL;
1656 conn->ksnc_rx_nkiov = niov;
1657 conn->ksnc_rx_kiov = conn->ksnc_rx_iov_space.kiov;
1658 memcpy (conn->ksnc_rx_kiov, kiov, niov * sizeof (*kiov));
1661 lib_iov_nob (conn->ksnc_rx_niov, conn->ksnc_rx_iov) +
1662 lib_kiov_nob (conn->ksnc_rx_nkiov, conn->ksnc_rx_kiov));
1667 int ksocknal_scheduler (void *arg)
1669 ksock_sched_t *sched = (ksock_sched_t *)arg;
1672 unsigned long flags;
1675 int id = sched - ksocknal_data.ksnd_schedulers;
1678 snprintf (name, sizeof (name),"ksocknald_%02d", id);
1679 kportal_daemonize (name);
1680 kportal_blockallsigs ();
1682 current->flags |= PF_MEMALLOC;
1684 #if (CONFIG_SMP && CPU_AFFINITY)
1685 if ((cpu_online_map & (1 << id)) != 0) {
1687 current->cpus_allowed = (1 << id);
1689 set_cpus_allowed (current, 1<<id);
1692 CERROR ("Can't set CPU affinity for %s\n", name);
1694 #endif /* CONFIG_SMP && CPU_AFFINITY */
1696 spin_lock_irqsave (&sched->kss_lock, flags);
1698 while (!ksocknal_data.ksnd_shuttingdown) {
1699 int did_something = 0;
1701 /* Ensure I progress everything semi-fairly */
1703 if (!list_empty (&sched->kss_rx_conns)) {
1704 conn = list_entry(sched->kss_rx_conns.next,
1705 ksock_conn_t, ksnc_rx_list);
1706 list_del(&conn->ksnc_rx_list);
1708 LASSERT(conn->ksnc_rx_scheduled);
1709 LASSERT(conn->ksnc_rx_ready);
1711 /* clear rx_ready in case receive isn't complete.
1712 * Do it BEFORE we call process_recv, since
1713 * data_ready can set it any time after we release
1715 conn->ksnc_rx_ready = 0;
1716 spin_unlock_irqrestore(&sched->kss_lock, flags);
1718 rc = ksocknal_process_receive(conn);
1720 spin_lock_irqsave(&sched->kss_lock, flags);
1722 /* I'm the only one that can clear this flag */
1723 LASSERT(conn->ksnc_rx_scheduled);
1725 /* Did process_receive get everything it wanted? */
1727 conn->ksnc_rx_ready = 1;
1729 if (conn->ksnc_rx_state == SOCKNAL_RX_FMB_SLEEP ||
1730 conn->ksnc_rx_state == SOCKNAL_RX_GET_FMB) {
1731 /* Conn blocked for a forwarding buffer.
1732 * It will get queued for my attention when
1733 * one becomes available (and it might just
1734 * already have been!). Meanwhile my ref
1735 * on it stays put. */
1736 } else if (conn->ksnc_rx_ready) {
1737 /* reschedule for rx */
1738 list_add_tail (&conn->ksnc_rx_list,
1739 &sched->kss_rx_conns);
1741 conn->ksnc_rx_scheduled = 0;
1743 ksocknal_put_conn(conn);
1749 if (!list_empty (&sched->kss_tx_conns)) {
1750 conn = list_entry(sched->kss_tx_conns.next,
1751 ksock_conn_t, ksnc_tx_list);
1752 list_del (&conn->ksnc_tx_list);
1754 LASSERT(conn->ksnc_tx_scheduled);
1755 LASSERT(conn->ksnc_tx_ready);
1756 LASSERT(!list_empty(&conn->ksnc_tx_queue));
1758 tx = list_entry(conn->ksnc_tx_queue.next,
1759 ksock_tx_t, tx_list);
1760 /* dequeue now so empty list => more to send */
1761 list_del(&tx->tx_list);
1763 /* Clear tx_ready in case send isn't complete. Do
1764 * it BEFORE we call process_transmit, since
1765 * write_space can set it any time after we release
1767 conn->ksnc_tx_ready = 0;
1768 spin_unlock_irqrestore (&sched->kss_lock, flags);
1770 rc = ksocknal_process_transmit(conn, tx);
1772 spin_lock_irqsave (&sched->kss_lock, flags);
1774 if (rc != -EAGAIN) {
1775 /* error or everything went: assume more can go */
1776 conn->ksnc_tx_ready = 1;
1778 /* back onto HEAD of tx_queue */
1779 list_add (&tx->tx_list, &conn->ksnc_tx_queue);
1782 if (conn->ksnc_tx_ready &&
1783 !list_empty (&conn->ksnc_tx_queue)) {
1784 /* reschedule for tx */
1785 list_add_tail (&conn->ksnc_tx_list,
1786 &sched->kss_tx_conns);
1788 conn->ksnc_tx_scheduled = 0;
1790 ksocknal_put_conn (conn);
1796 if (!list_empty (&sched->kss_zctxdone_list)) {
1798 list_entry(sched->kss_zctxdone_list.next,
1799 ksock_tx_t, tx_list);
1802 list_del (&tx->tx_list);
1803 spin_unlock_irqrestore (&sched->kss_lock, flags);
1805 ksocknal_tx_done (tx, 1);
1807 spin_lock_irqsave (&sched->kss_lock, flags);
1810 if (!did_something || /* nothing to do */
1811 ++nloops == SOCKNAL_RESCHED) { /* hogging CPU? */
1812 spin_unlock_irqrestore (&sched->kss_lock, flags);
1816 if (!did_something) { /* wait for something to do */
1818 rc = wait_event_interruptible (sched->kss_waitq,
1819 ksocknal_data.ksnd_shuttingdown ||
1820 !list_empty(&sched->kss_rx_conns) ||
1821 !list_empty(&sched->kss_tx_conns) ||
1822 !list_empty(&sched->kss_zctxdone_list));
1824 rc = wait_event_interruptible (sched->kss_waitq,
1825 ksocknal_data.ksnd_shuttingdown ||
1826 !list_empty(&sched->kss_rx_conns) ||
1827 !list_empty(&sched->kss_tx_conns));
1833 spin_lock_irqsave (&sched->kss_lock, flags);
1837 spin_unlock_irqrestore (&sched->kss_lock, flags);
1838 ksocknal_thread_fini ();
1843 ksocknal_data_ready (struct sock *sk, int n)
1845 unsigned long flags;
1847 ksock_sched_t *sched;
1850 /* interleave correctly with closing sockets... */
1851 read_lock (&ksocknal_data.ksnd_global_lock);
1853 conn = sk->sk_user_data;
1854 if (conn == NULL) { /* raced with ksocknal_close_sock */
1855 LASSERT (sk->sk_data_ready != &ksocknal_data_ready);
1856 sk->sk_data_ready (sk, n);
1858 sched = conn->ksnc_scheduler;
1860 spin_lock_irqsave (&sched->kss_lock, flags);
1862 conn->ksnc_rx_ready = 1;
1864 if (!conn->ksnc_rx_scheduled) { /* not being progressed */
1865 list_add_tail(&conn->ksnc_rx_list,
1866 &sched->kss_rx_conns);
1867 conn->ksnc_rx_scheduled = 1;
1868 /* extra ref for scheduler */
1869 atomic_inc (&conn->ksnc_refcount);
1871 wake_up (&sched->kss_waitq);
1874 spin_unlock_irqrestore (&sched->kss_lock, flags);
1877 read_unlock (&ksocknal_data.ksnd_global_lock);
1883 ksocknal_write_space (struct sock *sk)
1885 unsigned long flags;
1887 ksock_sched_t *sched;
1889 /* interleave correctly with closing sockets... */
1890 read_lock (&ksocknal_data.ksnd_global_lock);
1892 conn = sk->sk_user_data;
1894 CDEBUG(D_NET, "sk %p wspace %d low water %d conn %p%s%s%s\n",
1895 sk, tcp_wspace(sk), SOCKNAL_TX_LOW_WATER(sk), conn,
1896 (conn == NULL) ? "" : (conn->ksnc_tx_ready ?
1897 " ready" : " blocked"),
1898 (conn == NULL) ? "" : (conn->ksnc_tx_scheduled ?
1899 " scheduled" : " idle"),
1900 (conn == NULL) ? "" : (list_empty (&conn->ksnc_tx_queue) ?
1901 " empty" : " queued"));
1903 if (conn == NULL) { /* raced with ksocknal_close_sock */
1904 LASSERT (sk->sk_write_space != &ksocknal_write_space);
1905 sk->sk_write_space (sk);
1907 read_unlock (&ksocknal_data.ksnd_global_lock);
1911 if (tcp_wspace(sk) >= SOCKNAL_TX_LOW_WATER(sk)) { /* got enough space */
1912 clear_bit (SOCK_NOSPACE, &sk->sk_socket->flags);
1914 sched = conn->ksnc_scheduler;
1916 spin_lock_irqsave (&sched->kss_lock, flags);
1918 conn->ksnc_tx_ready = 1;
1920 if (!conn->ksnc_tx_scheduled && // not being progressed
1921 !list_empty(&conn->ksnc_tx_queue)){//packets to send
1922 list_add_tail (&conn->ksnc_tx_list,
1923 &sched->kss_tx_conns);
1924 conn->ksnc_tx_scheduled = 1;
1925 /* extra ref for scheduler */
1926 atomic_inc (&conn->ksnc_refcount);
1928 wake_up (&sched->kss_waitq);
1931 spin_unlock_irqrestore (&sched->kss_lock, flags);
1934 read_unlock (&ksocknal_data.ksnd_global_lock);
1938 ksocknal_sock_write (struct socket *sock, void *buffer, int nob)
1941 mm_segment_t oldmm = get_fs();
1944 struct iovec iov = {
1948 struct msghdr msg = {
1953 .msg_control = NULL,
1954 .msg_controllen = 0,
1959 rc = sock_sendmsg (sock, &msg, iov.iov_len);
1966 CERROR ("Unexpected zero rc\n");
1967 return (-ECONNABORTED);
1970 buffer = ((char *)buffer) + rc;
1978 ksocknal_sock_read (struct socket *sock, void *buffer, int nob)
1981 mm_segment_t oldmm = get_fs();
1984 struct iovec iov = {
1988 struct msghdr msg = {
1993 .msg_control = NULL,
1994 .msg_controllen = 0,
1999 rc = sock_recvmsg (sock, &msg, iov.iov_len, 0);
2006 return (-ECONNABORTED);
2008 buffer = ((char *)buffer) + rc;
2016 ksocknal_hello (struct socket *sock, ptl_nid_t *nid, int *type, __u64 *incarnation)
2020 ptl_magicversion_t *hmv = (ptl_magicversion_t *)&hdr.dest_nid;
2022 LASSERT (sizeof (*hmv) == sizeof (hdr.dest_nid));
2024 memset (&hdr, 0, sizeof (hdr));
2025 hmv->magic = __cpu_to_le32 (PORTALS_PROTO_MAGIC);
2026 hmv->version_major = __cpu_to_le32 (PORTALS_PROTO_VERSION_MAJOR);
2027 hmv->version_minor = __cpu_to_le32 (PORTALS_PROTO_VERSION_MINOR);
2029 hdr.src_nid = __cpu_to_le64 (ksocknal_lib.ni.nid);
2030 hdr.type = __cpu_to_le32 (PTL_MSG_HELLO);
2032 hdr.msg.hello.type = __cpu_to_le32 (*type);
2033 hdr.msg.hello.incarnation =
2034 __cpu_to_le64 (ksocknal_data.ksnd_incarnation);
2036 /* Assume sufficient socket buffering for this message */
2037 rc = ksocknal_sock_write (sock, &hdr, sizeof (hdr));
2039 CERROR ("Error %d sending HELLO to "LPX64"\n", rc, *nid);
2043 rc = ksocknal_sock_read (sock, hmv, sizeof (*hmv));
2045 CERROR ("Error %d reading HELLO from "LPX64"\n", rc, *nid);
2049 if (hmv->magic != __le32_to_cpu (PORTALS_PROTO_MAGIC)) {
2050 CERROR ("Bad magic %#08x (%#08x expected) from "LPX64"\n",
2051 __cpu_to_le32 (hmv->magic), PORTALS_PROTO_MAGIC, *nid);
2055 if (hmv->version_major != __cpu_to_le16 (PORTALS_PROTO_VERSION_MAJOR) ||
2056 hmv->version_minor != __cpu_to_le16 (PORTALS_PROTO_VERSION_MINOR)) {
2057 CERROR ("Incompatible protocol version %d.%d (%d.%d expected)"
2059 __le16_to_cpu (hmv->version_major),
2060 __le16_to_cpu (hmv->version_minor),
2061 PORTALS_PROTO_VERSION_MAJOR,
2062 PORTALS_PROTO_VERSION_MINOR,
2067 #if (PORTALS_PROTO_VERSION_MAJOR != 0)
2068 # error "This code only understands protocol version 0.x"
2070 /* version 0 sends magic/version as the dest_nid of a 'hello' header,
2071 * so read the rest of it in now... */
2073 rc = ksocknal_sock_read (sock, hmv + 1, sizeof (hdr) - sizeof (*hmv));
2075 CERROR ("Error %d reading rest of HELLO hdr from "LPX64"\n",
2080 /* ...and check we got what we expected */
2081 if (hdr.type != __cpu_to_le32 (PTL_MSG_HELLO) ||
2082 hdr.payload_length != __cpu_to_le32 (0)) {
2083 CERROR ("Expecting a HELLO hdr with 0 payload,"
2084 " but got type %d with %d payload from "LPX64"\n",
2085 __le32_to_cpu (hdr.type),
2086 __le32_to_cpu (hdr.payload_length), *nid);
2090 if (__le64_to_cpu(hdr.src_nid) == PTL_NID_ANY) {
2091 CERROR("Expecting a HELLO hdr with a NID, but got PTL_NID_ANY\n");
2095 if (*nid == PTL_NID_ANY) { /* don't know peer's nid yet */
2096 *nid = __le64_to_cpu(hdr.src_nid);
2097 } else if (*nid != __le64_to_cpu (hdr.src_nid)) {
2098 CERROR ("Connected to nid "LPX64", but expecting "LPX64"\n",
2099 __le64_to_cpu (hdr.src_nid), *nid);
2103 if (*type == SOCKNAL_CONN_NONE) {
2104 /* I've accepted this connection; peer determines type */
2105 *type = __le32_to_cpu(hdr.msg.hello.type);
2107 case SOCKNAL_CONN_ANY:
2108 case SOCKNAL_CONN_CONTROL:
2110 case SOCKNAL_CONN_BULK_IN:
2111 *type = SOCKNAL_CONN_BULK_OUT;
2113 case SOCKNAL_CONN_BULK_OUT:
2114 *type = SOCKNAL_CONN_BULK_IN;
2117 CERROR ("Unexpected type %d from "LPX64"\n", *type, *nid);
2120 } else if (__le32_to_cpu(hdr.msg.hello.type) != SOCKNAL_CONN_NONE) {
2121 CERROR ("Mismatched types: me %d "LPX64" %d\n",
2122 *type, *nid, __le32_to_cpu(hdr.msg.hello.type));
2126 *incarnation = __le64_to_cpu(hdr.msg.hello.incarnation);
2132 ksocknal_setup_sock (struct socket *sock)
2134 mm_segment_t oldmm = get_fs ();
2137 struct linger linger;
2139 sock->sk->allocation = GFP_NOFS;
2141 /* Ensure this socket aborts active sends immediately when we close
2145 linger.l_linger = 0;
2148 rc = sock_setsockopt (sock, SOL_SOCKET, SO_LINGER,
2149 (char *)&linger, sizeof (linger));
2152 CERROR ("Can't set SO_LINGER: %d\n", rc);
2158 rc = sock->ops->setsockopt (sock, SOL_TCP, TCP_LINGER2,
2159 (char *)&option, sizeof (option));
2162 CERROR ("Can't set SO_LINGER2: %d\n", rc);
2166 #if SOCKNAL_USE_KEEPALIVES
2167 /* Keepalives: If 3/4 of the timeout elapses, start probing every
2168 * second until the timeout elapses. */
2170 option = (ksocknal_data.ksnd_io_timeout * 3) / 4;
2172 rc = sock->ops->setsockopt (sock, SOL_TCP, TCP_KEEPIDLE,
2173 (char *)&option, sizeof (option));
2176 CERROR ("Can't set TCP_KEEPIDLE: %d\n", rc);
2182 rc = sock->ops->setsockopt (sock, SOL_TCP, TCP_KEEPINTVL,
2183 (char *)&option, sizeof (option));
2186 CERROR ("Can't set TCP_KEEPINTVL: %d\n", rc);
2190 option = ksocknal_data.ksnd_io_timeout / 4;
2192 rc = sock->ops->setsockopt (sock, SOL_TCP, TCP_KEEPCNT,
2193 (char *)&option, sizeof (option));
2196 CERROR ("Can't set TCP_KEEPINTVL: %d\n", rc);
2202 rc = sock_setsockopt (sock, SOL_SOCKET, SO_KEEPALIVE,
2203 (char *)&option, sizeof (option));
2206 CERROR ("Can't set SO_KEEPALIVE: %d\n", rc);
2214 ksocknal_connect_peer (ksock_route_t *route, int type)
2216 struct sockaddr_in peer_addr;
2217 mm_segment_t oldmm = get_fs();
2220 struct socket *sock;
2223 rc = sock_create (PF_INET, SOCK_STREAM, 0, &sock);
2225 CERROR ("Can't create autoconnect socket: %d\n", rc);
2229 /* Ugh; have to map_fd for compatibility with sockets passed in
2230 * from userspace. And we actually need the sock->file refcounting
2231 * that this gives you :) */
2233 fd = sock_map_fd (sock);
2235 sock_release (sock);
2236 CERROR ("sock_map_fd error %d\n", fd);
2240 /* NB the fd now owns the ref on sock->file */
2241 LASSERT (sock->file != NULL);
2242 LASSERT (file_count(sock->file) == 1);
2244 /* Set the socket timeouts, so our connection attempt completes in
2246 tv.tv_sec = ksocknal_data.ksnd_io_timeout;
2250 rc = sock_setsockopt (sock, SOL_SOCKET, SO_SNDTIMEO,
2251 (char *)&tv, sizeof (tv));
2254 CERROR ("Can't set send timeout %d: %d\n",
2255 ksocknal_data.ksnd_io_timeout, rc);
2260 rc = sock_setsockopt (sock, SOL_SOCKET, SO_RCVTIMEO,
2261 (char *)&tv, sizeof (tv));
2264 CERROR ("Can't set receive timeout %d: %d\n",
2265 ksocknal_data.ksnd_io_timeout, rc);
2269 if (route->ksnr_nonagel) {
2273 rc = sock->ops->setsockopt (sock, SOL_TCP, TCP_NODELAY,
2274 (char *)&option, sizeof (option));
2277 CERROR ("Can't disable nagel: %d\n", rc);
2282 if (route->ksnr_buffer_size != 0) {
2283 int option = route->ksnr_buffer_size;
2286 rc = sock_setsockopt (sock, SOL_SOCKET, SO_SNDBUF,
2287 (char *)&option, sizeof (option));
2290 CERROR ("Can't set send buffer %d: %d\n",
2291 route->ksnr_buffer_size, rc);
2296 rc = sock_setsockopt (sock, SOL_SOCKET, SO_RCVBUF,
2297 (char *)&option, sizeof (option));
2300 CERROR ("Can't set receive buffer %d: %d\n",
2301 route->ksnr_buffer_size, rc);
2306 memset (&peer_addr, 0, sizeof (peer_addr));
2307 peer_addr.sin_family = AF_INET;
2308 peer_addr.sin_port = htons (route->ksnr_port);
2309 peer_addr.sin_addr.s_addr = htonl (route->ksnr_ipaddr);
2311 rc = sock->ops->connect (sock, (struct sockaddr *)&peer_addr,
2312 sizeof (peer_addr), sock->file->f_flags);
2314 CERROR ("Error %d connecting to "LPX64"\n", rc,
2315 route->ksnr_peer->ksnp_nid);
2319 rc = ksocknal_create_conn (route, sock, route->ksnr_irq_affinity, type);
2321 /* Take an extra ref on sock->file to compensate for the
2322 * upcoming close which will lose fd's ref on it. */
2323 get_file (sock->file);
2332 ksocknal_autoconnect (ksock_route_t *route)
2334 LIST_HEAD (zombies);
2337 unsigned long flags;
2342 for (type = 0; type < SOCKNAL_CONN_NTYPES; type++)
2343 if ((route->ksnr_connecting & (1 << type)) != 0)
2345 LASSERT (type < SOCKNAL_CONN_NTYPES);
2347 rc = ksocknal_connect_peer (route, type);
2352 /* successfully autoconnected: create_conn did the
2353 * route/conn binding and scheduled any blocked packets */
2355 if (route->ksnr_connecting == 0) {
2356 /* No more connections required */
2361 /* Connection attempt failed */
2363 write_lock_irqsave (&ksocknal_data.ksnd_global_lock, flags);
2365 peer = route->ksnr_peer;
2366 route->ksnr_connecting = 0;
2368 /* This is a retry rather than a new connection */
2369 LASSERT (route->ksnr_retry_interval != 0);
2370 route->ksnr_timeout = jiffies + route->ksnr_retry_interval;
2371 route->ksnr_retry_interval = MIN (route->ksnr_retry_interval * 2,
2372 SOCKNAL_MAX_RECONNECT_INTERVAL);
2374 if (!list_empty (&peer->ksnp_tx_queue) &&
2375 ksocknal_find_connecting_route_locked (peer) == NULL) {
2376 LASSERT (list_empty (&peer->ksnp_conns));
2378 /* None of the connections that the blocked packets are
2379 * waiting for have been successful. Complete them now... */
2381 tx = list_entry (peer->ksnp_tx_queue.next,
2382 ksock_tx_t, tx_list);
2383 list_del (&tx->tx_list);
2384 list_add_tail (&tx->tx_list, &zombies);
2385 } while (!list_empty (&peer->ksnp_tx_queue));
2388 /* make this route least-favourite for re-selection */
2389 if (!route->ksnr_deleted) {
2390 list_del(&route->ksnr_list);
2391 list_add_tail(&route->ksnr_list, &peer->ksnp_routes);
2394 write_unlock_irqrestore (&ksocknal_data.ksnd_global_lock, flags);
2396 while (!list_empty (&zombies)) {
2397 tx = list_entry (zombies.next, ksock_tx_t, tx_list);
2399 CERROR ("Deleting packet type %d len %d ("LPX64"->"LPX64")\n",
2400 NTOH__u32 (tx->tx_hdr->type),
2401 NTOH__u32 (tx->tx_hdr->payload_length),
2402 NTOH__u64 (tx->tx_hdr->src_nid),
2403 NTOH__u64 (tx->tx_hdr->dest_nid));
2405 list_del (&tx->tx_list);
2407 ksocknal_tx_done (tx, 0);
2412 ksocknal_autoconnectd (void *arg)
2414 long id = (long)arg;
2416 unsigned long flags;
2417 ksock_route_t *route;
2420 snprintf (name, sizeof (name), "ksocknal_ad%02ld", id);
2421 kportal_daemonize (name);
2422 kportal_blockallsigs ();
2424 spin_lock_irqsave (&ksocknal_data.ksnd_autoconnectd_lock, flags);
2426 while (!ksocknal_data.ksnd_shuttingdown) {
2428 if (!list_empty (&ksocknal_data.ksnd_autoconnectd_routes)) {
2429 route = list_entry (ksocknal_data.ksnd_autoconnectd_routes.next,
2430 ksock_route_t, ksnr_connect_list);
2432 list_del (&route->ksnr_connect_list);
2433 spin_unlock_irqrestore (&ksocknal_data.ksnd_autoconnectd_lock, flags);
2435 ksocknal_autoconnect (route);
2436 ksocknal_put_route (route);
2438 spin_lock_irqsave (&ksocknal_data.ksnd_autoconnectd_lock, flags);
2442 spin_unlock_irqrestore (&ksocknal_data.ksnd_autoconnectd_lock, flags);
2444 rc = wait_event_interruptible (ksocknal_data.ksnd_autoconnectd_waitq,
2445 ksocknal_data.ksnd_shuttingdown ||
2446 !list_empty (&ksocknal_data.ksnd_autoconnectd_routes));
2448 spin_lock_irqsave (&ksocknal_data.ksnd_autoconnectd_lock, flags);
2451 spin_unlock_irqrestore (&ksocknal_data.ksnd_autoconnectd_lock, flags);
2453 ksocknal_thread_fini ();
2458 ksocknal_find_timed_out_conn (ksock_peer_t *peer)
2460 /* We're called with a shared lock on ksnd_global_lock */
2462 struct list_head *ctmp;
2463 ksock_sched_t *sched;
2465 list_for_each (ctmp, &peer->ksnp_conns) {
2466 conn = list_entry (ctmp, ksock_conn_t, ksnc_list);
2467 sched = conn->ksnc_scheduler;
2469 /* Don't need the {get,put}connsock dance to deref ksnc_sock... */
2470 LASSERT (!conn->ksnc_closing);
2472 if (conn->ksnc_rx_started &&
2473 time_after_eq (jiffies, conn->ksnc_rx_deadline)) {
2474 /* Timed out incomplete incoming message */
2475 atomic_inc (&conn->ksnc_refcount);
2476 CERROR ("Timed out RX from "LPX64" %p\n",
2477 peer->ksnp_nid, conn);
2481 if ((!list_empty (&conn->ksnc_tx_queue) ||
2482 conn->ksnc_sock->sk->sk_wmem_queued != 0) &&
2483 time_after_eq (jiffies, conn->ksnc_tx_deadline)) {
2484 /* Timed out messages queued for sending, or
2485 * messages buffered in the socket's send buffer */
2486 atomic_inc (&conn->ksnc_refcount);
2487 CERROR ("Timed out TX to "LPX64" %s%d %p\n",
2489 list_empty (&conn->ksnc_tx_queue) ? "" : "Q ",
2490 conn->ksnc_sock->sk->sk_wmem_queued, conn);
2499 ksocknal_check_peer_timeouts (int idx)
2501 struct list_head *peers = &ksocknal_data.ksnd_peers[idx];
2502 struct list_head *ptmp;
2507 /* NB. We expect to have a look at all the peers and not find any
2508 * connections to time out, so we just use a shared lock while we
2510 read_lock (&ksocknal_data.ksnd_global_lock);
2512 list_for_each (ptmp, peers) {
2513 peer = list_entry (ptmp, ksock_peer_t, ksnp_list);
2514 conn = ksocknal_find_timed_out_conn (peer);
2517 read_unlock (&ksocknal_data.ksnd_global_lock);
2519 CERROR ("Timeout out conn->"LPX64" ip %x:%d\n",
2520 peer->ksnp_nid, conn->ksnc_ipaddr,
2522 ksocknal_close_conn_and_siblings (conn, -ETIMEDOUT);
2524 /* NB we won't find this one again, but we can't
2525 * just proceed with the next peer, since we dropped
2526 * ksnd_global_lock and it might be dead already! */
2527 ksocknal_put_conn (conn);
2532 read_unlock (&ksocknal_data.ksnd_global_lock);
2536 ksocknal_reaper (void *arg)
2539 unsigned long flags;
2544 unsigned long deadline = jiffies;
2546 kportal_daemonize ("ksocknal_reaper");
2547 kportal_blockallsigs ();
2549 init_waitqueue_entry (&wait, current);
2551 spin_lock_irqsave (&ksocknal_data.ksnd_reaper_lock, flags);
2553 while (!ksocknal_data.ksnd_shuttingdown) {
2555 if (!list_empty (&ksocknal_data.ksnd_deathrow_conns)) {
2556 conn = list_entry (ksocknal_data.ksnd_deathrow_conns.next,
2557 ksock_conn_t, ksnc_list);
2558 list_del (&conn->ksnc_list);
2560 spin_unlock_irqrestore (&ksocknal_data.ksnd_reaper_lock, flags);
2562 ksocknal_terminate_conn (conn);
2563 ksocknal_put_conn (conn);
2565 spin_lock_irqsave (&ksocknal_data.ksnd_reaper_lock, flags);
2569 if (!list_empty (&ksocknal_data.ksnd_zombie_conns)) {
2570 conn = list_entry (ksocknal_data.ksnd_zombie_conns.next,
2571 ksock_conn_t, ksnc_list);
2572 list_del (&conn->ksnc_list);
2574 spin_unlock_irqrestore (&ksocknal_data.ksnd_reaper_lock, flags);
2576 ksocknal_destroy_conn (conn);
2578 spin_lock_irqsave (&ksocknal_data.ksnd_reaper_lock, flags);
2582 spin_unlock_irqrestore (&ksocknal_data.ksnd_reaper_lock, flags);
2584 /* careful with the jiffy wrap... */
2585 while ((timeout = ((int)deadline - (int)jiffies)) <= 0) {
2588 int chunk = ksocknal_data.ksnd_peer_hash_size;
2590 /* Time to check for timeouts on a few more peers: I do
2591 * checks every 'p' seconds on a proportion of the peer
2592 * table and I need to check every connection 'n' times
2593 * within a timeout interval, to ensure I detect a
2594 * timeout on any connection within (n+1)/n times the
2595 * timeout interval. */
2597 if (ksocknal_data.ksnd_io_timeout > n * p)
2598 chunk = (chunk * n * p) /
2599 ksocknal_data.ksnd_io_timeout;
2603 for (i = 0; i < chunk; i++) {
2604 ksocknal_check_peer_timeouts (peer_index);
2605 peer_index = (peer_index + 1) %
2606 ksocknal_data.ksnd_peer_hash_size;
2612 add_wait_queue (&ksocknal_data.ksnd_reaper_waitq, &wait);
2613 set_current_state (TASK_INTERRUPTIBLE);
2615 if (!ksocknal_data.ksnd_shuttingdown &&
2616 list_empty (&ksocknal_data.ksnd_deathrow_conns) &&
2617 list_empty (&ksocknal_data.ksnd_zombie_conns))
2618 schedule_timeout (timeout);
2620 set_current_state (TASK_RUNNING);
2621 remove_wait_queue (&ksocknal_data.ksnd_reaper_waitq, &wait);
2623 spin_lock_irqsave (&ksocknal_data.ksnd_reaper_lock, flags);
2626 spin_unlock_irqrestore (&ksocknal_data.ksnd_reaper_lock, flags);
2628 ksocknal_thread_fini ();
2632 nal_cb_t ksocknal_lib = {
2633 nal_data: &ksocknal_data, /* NAL private data */
2634 cb_send: ksocknal_send,
2635 cb_send_pages: ksocknal_send_pages,
2636 cb_recv: ksocknal_recv,
2637 cb_recv_pages: ksocknal_recv_pages,
2638 cb_read: ksocknal_read,
2639 cb_write: ksocknal_write,
2640 cb_callback: ksocknal_callback,
2641 cb_malloc: ksocknal_malloc,
2642 cb_free: ksocknal_free,
2643 cb_printf: ksocknal_printf,
2644 cb_cli: ksocknal_cli,
2645 cb_sti: ksocknal_sti,
2646 cb_dist: ksocknal_dist