X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lnet%2Flnet%2Fapi-ni.c;h=8f5dc249214f1b6fd88ccdcf87d92d9c30471e78;hb=24b1bba70c43ca64318b54aa11d7ba584ee2b4c0;hp=5ecf560cf32ee5d604a12d48b584cb6542542cfe;hpb=a3488d2f33e3ba4ae1d9c127132a6b75b5bb0112;p=fs%2Flustre-release.git diff --git a/lnet/lnet/api-ni.c b/lnet/lnet/api-ni.c index 5ecf560..8f5dc24 100644 --- a/lnet/lnet/api-ni.c +++ b/lnet/lnet/api-ni.c @@ -15,11 +15,7 @@ * * You should have received a copy of the GNU General Public License * version 2 along with this program; If not, see - * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf - * - * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, - * CA 95054 USA or visit www.sun.com if you need additional information or - * have any questions. + * http://www.gnu.org/licenses/gpl-2.0.html * * GPL HEADER END */ @@ -27,7 +23,7 @@ * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2011, 2013, Intel Corporation. + * Copyright (c) 2011, 2016, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -35,54 +31,60 @@ */ #define DEBUG_SUBSYSTEM S_LNET -#include -#include -#ifdef __KERNEL__ #include -#endif +#include + +#include -#ifdef __KERNEL__ #define D_LNI D_CONSOLE -#else -#define D_LNI D_CONFIG -#endif -lnet_t the_lnet; /* THE state of the network */ +lnet_t the_lnet; /* THE state of the network */ EXPORT_SYMBOL(the_lnet); -#ifdef __KERNEL__ - static char *ip2nets = ""; -CFS_MODULE_PARM(ip2nets, "s", charp, 0444, - "LNET network <- IP table"); +module_param(ip2nets, charp, 0444); +MODULE_PARM_DESC(ip2nets, "LNET network <- IP table"); static char *networks = ""; -CFS_MODULE_PARM(networks, "s", charp, 0444, - "local networks"); +module_param(networks, charp, 0444); +MODULE_PARM_DESC(networks, "local networks"); static char *routes = ""; -CFS_MODULE_PARM(routes, "s", charp, 0444, - "routes to non-local networks"); +module_param(routes, charp, 0444); +MODULE_PARM_DESC(routes, "routes to non-local networks"); static int rnet_htable_size = LNET_REMOTE_NETS_HASH_DEFAULT; -CFS_MODULE_PARM(rnet_htable_size, "i", int, 0444, - "size of remote network hash table"); +module_param(rnet_htable_size, int, 0444); +MODULE_PARM_DESC(rnet_htable_size, "size of remote network hash table"); + +static int use_tcp_bonding = false; +module_param(use_tcp_bonding, int, 0444); +MODULE_PARM_DESC(use_tcp_bonding, + "Set to 1 to use socklnd bonding. 0 to use Multi-Rail"); -static void lnet_ping_target_fini(void); -static int lnet_ping(lnet_process_id_t id, int timeout_ms, - lnet_process_id_t *ids, int n_ids); +/* + * This sequence number keeps track of how many times DLC was used to + * update the configuration. It is incremented on any DLC update and + * checked when sending a message to determine if there is a need to + * re-run the selection algorithm to handle configuration change. + * Look at lnet_select_pathway() for more details on its usage. + */ +static atomic_t lnet_dlc_seq_no = ATOMIC_INIT(0); + +static int lnet_ping(lnet_process_id_t id, signed long timeout, + lnet_process_id_t __user *ids, int n_ids); static char * lnet_get_routes(void) { - return routes; + return routes; } static char * lnet_get_networks(void) { char *nets; - int rc; + int rc; if (*networks != 0 && *ip2nets != 0) { LCONSOLE_ERROR_MSG(0x101, "Please specify EITHER 'networks' or " @@ -106,6 +108,7 @@ lnet_init_locks(void) { spin_lock_init(&the_lnet.ln_eq_wait_lock); init_waitqueue_head(&the_lnet.ln_eq_waitq); + init_waitqueue_head(&the_lnet.ln_rc_waitq); mutex_init(&the_lnet.ln_lnd_mutex); mutex_init(&the_lnet.ln_api_mutex); } @@ -115,91 +118,45 @@ lnet_fini_locks(void) { } -#else - -static char * -lnet_get_routes(void) -{ - char *str = getenv("LNET_ROUTES"); - - return (str == NULL) ? "" : str; -} - -static char * -lnet_get_networks (void) -{ - static char default_networks[256]; - char *networks = getenv("LNET_NETWORKS"); - char *str; - char *sep; - int len; - int nob; - struct list_head *tmp; - - if (networks != NULL) - return networks; - - /* In userland, the default 'networks=' is the list of known net types */ - len = sizeof(default_networks); - str = default_networks; - *str = 0; - sep = ""; +struct kmem_cache *lnet_mes_cachep; /* MEs kmem_cache */ +struct kmem_cache *lnet_small_mds_cachep; /* <= LNET_SMALL_MD_SIZE bytes + * MDs kmem_cache */ - list_for_each(tmp, &the_lnet.ln_lnds) { - lnd_t *lnd = list_entry(tmp, lnd_t, lnd_list); - - nob = snprintf(str, len, "%s%s", sep, - libcfs_lnd2str(lnd->lnd_type)); - if (nob >= len) { - /* overflowed the string; leave it where it was */ - *str = 0; - break; - } - len -= nob; - str += nob; - sep = ","; - } - - return default_networks; -} +static int +lnet_descriptor_setup(void) +{ + /* create specific kmem_cache for MEs and small MDs (i.e., originally + * allocated in kmem_cache). + */ + lnet_mes_cachep = kmem_cache_create("lnet_MEs", sizeof(lnet_me_t), + 0, 0, NULL); + if (!lnet_mes_cachep) + return -ENOMEM; -# ifndef HAVE_LIBPTHREAD + lnet_small_mds_cachep = kmem_cache_create("lnet_small_MDs", + LNET_SMALL_MD_SIZE, 0, 0, + NULL); + if (!lnet_small_mds_cachep) + return -ENOMEM; -static void lnet_init_locks(void) -{ - the_lnet.ln_eq_wait_lock = 0; - the_lnet.ln_lnd_mutex = 0; - the_lnet.ln_api_mutex = 0; + return 0; } -static void lnet_fini_locks(void) +static void +lnet_descriptor_cleanup(void) { - LASSERT(the_lnet.ln_api_mutex == 0); - LASSERT(the_lnet.ln_lnd_mutex == 0); - LASSERT(the_lnet.ln_eq_wait_lock == 0); -} -# else - -static void lnet_init_locks(void) -{ - pthread_cond_init(&the_lnet.ln_eq_cond, NULL); - pthread_mutex_init(&the_lnet.ln_eq_wait_lock, NULL); - pthread_mutex_init(&the_lnet.ln_lnd_mutex, NULL); - pthread_mutex_init(&the_lnet.ln_api_mutex, NULL); -} + if (lnet_small_mds_cachep) { + kmem_cache_destroy(lnet_small_mds_cachep); + lnet_small_mds_cachep = NULL; + } -static void lnet_fini_locks(void) -{ - pthread_mutex_destroy(&the_lnet.ln_api_mutex); - pthread_mutex_destroy(&the_lnet.ln_lnd_mutex); - pthread_mutex_destroy(&the_lnet.ln_eq_wait_lock); - pthread_cond_destroy(&the_lnet.ln_eq_cond); + if (lnet_mes_cachep) { + kmem_cache_destroy(lnet_mes_cachep); + lnet_mes_cachep = NULL; + } } -# endif -#endif - static int lnet_create_remote_nets_table(void) { @@ -275,99 +232,98 @@ lnet_create_locks(void) static void lnet_assert_wire_constants(void) { - /* Wire protocol assertions generated by 'wirecheck' - * running on Linux robert.bartonsoftware.com 2.6.8-1.521 - * #1 Mon Aug 16 09:01:18 EDT 2004 i686 athlon i386 GNU/Linux - * with gcc version 3.3.3 20040412 (Red Hat Linux 3.3.3-7) */ - - /* Constants... */ - CLASSERT (LNET_PROTO_TCP_MAGIC == 0xeebc0ded); - CLASSERT (LNET_PROTO_TCP_VERSION_MAJOR == 1); - CLASSERT (LNET_PROTO_TCP_VERSION_MINOR == 0); - CLASSERT (LNET_MSG_ACK == 0); - CLASSERT (LNET_MSG_PUT == 1); - CLASSERT (LNET_MSG_GET == 2); - CLASSERT (LNET_MSG_REPLY == 3); - CLASSERT (LNET_MSG_HELLO == 4); - - /* Checks for struct ptl_handle_wire_t */ - CLASSERT ((int)sizeof(lnet_handle_wire_t) == 16); - CLASSERT ((int)offsetof(lnet_handle_wire_t, wh_interface_cookie) == 0); - CLASSERT ((int)sizeof(((lnet_handle_wire_t *)0)->wh_interface_cookie) == 8); - CLASSERT ((int)offsetof(lnet_handle_wire_t, wh_object_cookie) == 8); - CLASSERT ((int)sizeof(((lnet_handle_wire_t *)0)->wh_object_cookie) == 8); - - /* Checks for struct lnet_magicversion_t */ - CLASSERT ((int)sizeof(lnet_magicversion_t) == 8); - CLASSERT ((int)offsetof(lnet_magicversion_t, magic) == 0); - CLASSERT ((int)sizeof(((lnet_magicversion_t *)0)->magic) == 4); - CLASSERT ((int)offsetof(lnet_magicversion_t, version_major) == 4); - CLASSERT ((int)sizeof(((lnet_magicversion_t *)0)->version_major) == 2); - CLASSERT ((int)offsetof(lnet_magicversion_t, version_minor) == 6); - CLASSERT ((int)sizeof(((lnet_magicversion_t *)0)->version_minor) == 2); - - /* Checks for struct lnet_hdr_t */ - CLASSERT ((int)sizeof(lnet_hdr_t) == 72); - CLASSERT ((int)offsetof(lnet_hdr_t, dest_nid) == 0); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->dest_nid) == 8); - CLASSERT ((int)offsetof(lnet_hdr_t, src_nid) == 8); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->src_nid) == 8); - CLASSERT ((int)offsetof(lnet_hdr_t, dest_pid) == 16); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->dest_pid) == 4); - CLASSERT ((int)offsetof(lnet_hdr_t, src_pid) == 20); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->src_pid) == 4); - CLASSERT ((int)offsetof(lnet_hdr_t, type) == 24); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->type) == 4); - CLASSERT ((int)offsetof(lnet_hdr_t, payload_length) == 28); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->payload_length) == 4); - CLASSERT ((int)offsetof(lnet_hdr_t, msg) == 32); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg) == 40); - - /* Ack */ - CLASSERT ((int)offsetof(lnet_hdr_t, msg.ack.dst_wmd) == 32); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.ack.dst_wmd) == 16); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.ack.match_bits) == 48); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.ack.match_bits) == 8); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.ack.mlength) == 56); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.ack.mlength) == 4); - - /* Put */ - CLASSERT ((int)offsetof(lnet_hdr_t, msg.put.ack_wmd) == 32); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.put.ack_wmd) == 16); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.put.match_bits) == 48); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.put.match_bits) == 8); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.put.hdr_data) == 56); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.put.hdr_data) == 8); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.put.ptl_index) == 64); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.put.ptl_index) == 4); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.put.offset) == 68); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.put.offset) == 4); - - /* Get */ - CLASSERT ((int)offsetof(lnet_hdr_t, msg.get.return_wmd) == 32); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.get.return_wmd) == 16); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.get.match_bits) == 48); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.get.match_bits) == 8); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.get.ptl_index) == 56); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.get.ptl_index) == 4); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.get.src_offset) == 60); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.get.src_offset) == 4); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.get.sink_length) == 64); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.get.sink_length) == 4); - - /* Reply */ - CLASSERT ((int)offsetof(lnet_hdr_t, msg.reply.dst_wmd) == 32); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.reply.dst_wmd) == 16); - - /* Hello */ - CLASSERT ((int)offsetof(lnet_hdr_t, msg.hello.incarnation) == 32); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.hello.incarnation) == 8); - CLASSERT ((int)offsetof(lnet_hdr_t, msg.hello.type) == 40); - CLASSERT ((int)sizeof(((lnet_hdr_t *)0)->msg.hello.type) == 4); -} - -static lnd_t * -lnet_find_lnd_by_type (int type) + /* Wire protocol assertions generated by 'wirecheck' + * running on Linux robert.bartonsoftware.com 2.6.8-1.521 + * #1 Mon Aug 16 09:01:18 EDT 2004 i686 athlon i386 GNU/Linux + * with gcc version 3.3.3 20040412 (Red Hat Linux 3.3.3-7) */ + + /* Constants... */ + CLASSERT(LNET_PROTO_TCP_MAGIC == 0xeebc0ded); + CLASSERT(LNET_PROTO_TCP_VERSION_MAJOR == 1); + CLASSERT(LNET_PROTO_TCP_VERSION_MINOR == 0); + CLASSERT(LNET_MSG_ACK == 0); + CLASSERT(LNET_MSG_PUT == 1); + CLASSERT(LNET_MSG_GET == 2); + CLASSERT(LNET_MSG_REPLY == 3); + CLASSERT(LNET_MSG_HELLO == 4); + + /* Checks for struct lnet_handle_wire */ + CLASSERT((int)sizeof(struct lnet_handle_wire) == 16); + CLASSERT((int)offsetof(struct lnet_handle_wire, wh_interface_cookie) == 0); + CLASSERT((int)sizeof(((struct lnet_handle_wire *)0)->wh_interface_cookie) == 8); + CLASSERT((int)offsetof(struct lnet_handle_wire, wh_object_cookie) == 8); + CLASSERT((int)sizeof(((struct lnet_handle_wire *)0)->wh_object_cookie) == 8); + + /* Checks for struct lnet_magicversion_t */ + CLASSERT((int)sizeof(lnet_magicversion_t) == 8); + CLASSERT((int)offsetof(lnet_magicversion_t, magic) == 0); + CLASSERT((int)sizeof(((lnet_magicversion_t *)0)->magic) == 4); + CLASSERT((int)offsetof(lnet_magicversion_t, version_major) == 4); + CLASSERT((int)sizeof(((lnet_magicversion_t *)0)->version_major) == 2); + CLASSERT((int)offsetof(lnet_magicversion_t, version_minor) == 6); + CLASSERT((int)sizeof(((lnet_magicversion_t *)0)->version_minor) == 2); + + /* Checks for struct lnet_hdr_t */ + CLASSERT((int)sizeof(lnet_hdr_t) == 72); + CLASSERT((int)offsetof(lnet_hdr_t, dest_nid) == 0); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->dest_nid) == 8); + CLASSERT((int)offsetof(lnet_hdr_t, src_nid) == 8); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->src_nid) == 8); + CLASSERT((int)offsetof(lnet_hdr_t, dest_pid) == 16); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->dest_pid) == 4); + CLASSERT((int)offsetof(lnet_hdr_t, src_pid) == 20); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->src_pid) == 4); + CLASSERT((int)offsetof(lnet_hdr_t, type) == 24); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->type) == 4); + CLASSERT((int)offsetof(lnet_hdr_t, payload_length) == 28); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->payload_length) == 4); + CLASSERT((int)offsetof(lnet_hdr_t, msg) == 32); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg) == 40); + + /* Ack */ + CLASSERT((int)offsetof(lnet_hdr_t, msg.ack.dst_wmd) == 32); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.ack.dst_wmd) == 16); + CLASSERT((int)offsetof(lnet_hdr_t, msg.ack.match_bits) == 48); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.ack.match_bits) == 8); + CLASSERT((int)offsetof(lnet_hdr_t, msg.ack.mlength) == 56); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.ack.mlength) == 4); + + /* Put */ + CLASSERT((int)offsetof(lnet_hdr_t, msg.put.ack_wmd) == 32); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.put.ack_wmd) == 16); + CLASSERT((int)offsetof(lnet_hdr_t, msg.put.match_bits) == 48); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.put.match_bits) == 8); + CLASSERT((int)offsetof(lnet_hdr_t, msg.put.hdr_data) == 56); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.put.hdr_data) == 8); + CLASSERT((int)offsetof(lnet_hdr_t, msg.put.ptl_index) == 64); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.put.ptl_index) == 4); + CLASSERT((int)offsetof(lnet_hdr_t, msg.put.offset) == 68); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.put.offset) == 4); + + /* Get */ + CLASSERT((int)offsetof(lnet_hdr_t, msg.get.return_wmd) == 32); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.get.return_wmd) == 16); + CLASSERT((int)offsetof(lnet_hdr_t, msg.get.match_bits) == 48); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.get.match_bits) == 8); + CLASSERT((int)offsetof(lnet_hdr_t, msg.get.ptl_index) == 56); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.get.ptl_index) == 4); + CLASSERT((int)offsetof(lnet_hdr_t, msg.get.src_offset) == 60); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.get.src_offset) == 4); + CLASSERT((int)offsetof(lnet_hdr_t, msg.get.sink_length) == 64); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.get.sink_length) == 4); + + /* Reply */ + CLASSERT((int)offsetof(lnet_hdr_t, msg.reply.dst_wmd) == 32); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.reply.dst_wmd) == 16); + + /* Hello */ + CLASSERT((int)offsetof(lnet_hdr_t, msg.hello.incarnation) == 32); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.hello.incarnation) == 8); + CLASSERT((int)offsetof(lnet_hdr_t, msg.hello.type) == 40); + CLASSERT((int)sizeof(((lnet_hdr_t *)0)->msg.hello.type) == 4); +} + +static lnd_t *lnet_find_lnd_by_type(__u32 type) { lnd_t *lnd; struct list_head *tmp; @@ -376,7 +332,7 @@ lnet_find_lnd_by_type (int type) list_for_each(tmp, &the_lnet.ln_lnds) { lnd = list_entry(tmp, lnd_t, lnd_list); - if ((int)lnd->lnd_type == type) + if (lnd->lnd_type == type) return lnd; } return NULL; @@ -385,9 +341,8 @@ lnet_find_lnd_by_type (int type) void lnet_register_lnd (lnd_t *lnd) { - LNET_MUTEX_LOCK(&the_lnet.ln_lnd_mutex); + mutex_lock(&the_lnet.ln_lnd_mutex); - LASSERT(the_lnet.ln_init); LASSERT(libcfs_isknown_lnd(lnd->lnd_type)); LASSERT(lnet_find_lnd_by_type(lnd->lnd_type) == NULL); @@ -396,23 +351,22 @@ lnet_register_lnd (lnd_t *lnd) CDEBUG(D_NET, "%s LND registered\n", libcfs_lnd2str(lnd->lnd_type)); - LNET_MUTEX_UNLOCK(&the_lnet.ln_lnd_mutex); + mutex_unlock(&the_lnet.ln_lnd_mutex); } EXPORT_SYMBOL(lnet_register_lnd); void lnet_unregister_lnd (lnd_t *lnd) { - LNET_MUTEX_LOCK(&the_lnet.ln_lnd_mutex); + mutex_lock(&the_lnet.ln_lnd_mutex); - LASSERT(the_lnet.ln_init); LASSERT(lnet_find_lnd_by_type(lnd->lnd_type) == lnd); LASSERT(lnd->lnd_refcount == 0); list_del(&lnd->lnd_list); CDEBUG(D_NET, "%s LND unregistered\n", libcfs_lnd2str(lnd->lnd_type)); - LNET_MUTEX_UNLOCK(&the_lnet.ln_lnd_mutex); + mutex_unlock(&the_lnet.ln_lnd_mutex); } EXPORT_SYMBOL(lnet_unregister_lnd); @@ -457,76 +411,6 @@ lnet_counters_reset(void) lnet_net_unlock(LNET_LOCK_EX); } -EXPORT_SYMBOL(lnet_counters_reset); - -#ifdef LNET_USE_LIB_FREELIST - -int -lnet_freelist_init(lnet_freelist_t *fl, int n, int size) -{ - char *space; - - LASSERT (n > 0); - - size += offsetof (lnet_freeobj_t, fo_contents); - - LIBCFS_ALLOC(space, n * size); - if (space == NULL) - return (-ENOMEM); - - INIT_LIST_HEAD(&fl->fl_list); - fl->fl_objs = space; - fl->fl_nobjs = n; - fl->fl_objsize = size; - - do { - list_add((struct list_head *)space, &fl->fl_list); - space += size; - } while (--n != 0); - - return 0; -} - -void -lnet_freelist_fini(lnet_freelist_t *fl) -{ - struct list_head *el; - int count; - - if (fl->fl_nobjs == 0) - return; - - count = 0; - for (el = fl->fl_list.next; el != &fl->fl_list; el = el->next) - count++; - - LASSERT (count == fl->fl_nobjs); - - LIBCFS_FREE(fl->fl_objs, fl->fl_nobjs * fl->fl_objsize); - memset (fl, 0, sizeof (*fl)); -} - -#endif /* LNET_USE_LIB_FREELIST */ - -static __u64 lnet_create_interface_cookie(void) -{ - /* NB the interface cookie in wire handles guards against delayed - * replies and ACKs appearing valid after reboot. Initialisation time, - * even if it's only implemented to millisecond resolution is probably - * easily good enough. */ - struct timeval tv; - __u64 cookie; -#ifndef __KERNEL__ - int rc = gettimeofday (&tv, NULL); - LASSERT (rc == 0); -#else - do_gettimeofday(&tv); -#endif - cookie = tv.tv_sec; - cookie *= 1000000; - cookie += tv.tv_usec; - return cookie; -} static char * lnet_res_type2str(int type) @@ -575,9 +459,6 @@ lnet_res_container_cleanup(struct lnet_res_container *rec) count, lnet_res_type2str(rec->rec_type)); } -#ifdef LNET_USE_LIB_FREELIST - lnet_freelist_fini(&rec->rec_freelist); -#endif if (rec->rec_lh_hash != NULL) { LIBCFS_FREE(rec->rec_lh_hash, LNET_LH_HASH_SIZE * sizeof(rec->rec_lh_hash[0])); @@ -588,8 +469,7 @@ lnet_res_container_cleanup(struct lnet_res_container *rec) } static int -lnet_res_container_setup(struct lnet_res_container *rec, - int cpt, int type, int objnum, int objsz) +lnet_res_container_setup(struct lnet_res_container *rec, int cpt, int type) { int rc = 0; int i; @@ -599,12 +479,6 @@ lnet_res_container_setup(struct lnet_res_container *rec, rec->rec_type = type; INIT_LIST_HEAD(&rec->rec_active); -#ifdef LNET_USE_LIB_FREELIST - memset(&rec->rec_freelist, 0, sizeof(rec->rec_freelist)); - rc = lnet_freelist_init(&rec->rec_freelist, objnum, objsz); - if (rc != 0) - goto out; -#endif rec->rec_lh_cookie = (cpt << LNET_COOKIE_TYPE_BITS) | type; /* Arbitrary choice of hash table size */ @@ -640,7 +514,7 @@ lnet_res_containers_destroy(struct lnet_res_container **recs) } static struct lnet_res_container ** -lnet_res_containers_create(int type, int objnum, int objsz) +lnet_res_containers_create(int type) { struct lnet_res_container **recs; struct lnet_res_container *rec; @@ -655,7 +529,7 @@ lnet_res_containers_create(int type, int objnum, int objsz) } cfs_percpt_for_each(rec, i, recs) { - rc = lnet_res_container_setup(rec, i, type, objnum, objsz); + rc = lnet_res_container_setup(rec, i, type); if (rc != 0) { lnet_res_containers_destroy(recs); return NULL; @@ -702,19 +576,6 @@ lnet_res_lh_initialize(struct lnet_res_container *rec, lnet_libhandle_t *lh) list_add(&lh->lh_hash_chain, &rec->rec_lh_hash[hash]); } -#ifndef __KERNEL__ -/** - * Reserved API - do not use. - * Temporary workaround to allow uOSS and test programs force server - * mode in userspace. See comments near ln_server_mode_flag in - * lnet/lib-types.h */ - -void -lnet_server_mode() { - the_lnet.ln_server_mode_flag = 1; -} -#endif - static int lnet_unprepare(void); static int @@ -729,40 +590,33 @@ lnet_prepare(lnet_pid_t requested_pid) return -ENETDOWN; } - LASSERT (the_lnet.ln_refcount == 0); - - the_lnet.ln_routing = 0; - -#ifdef __KERNEL__ - LASSERT ((requested_pid & LNET_PID_USERFLAG) == 0); - the_lnet.ln_pid = requested_pid; -#else - if (the_lnet.ln_server_mode_flag) {/* server case (uOSS) */ - LASSERT ((requested_pid & LNET_PID_USERFLAG) == 0); - - if (current_uid() != 0) /* Only root can run user-space server */ - return -EPERM; - the_lnet.ln_pid = requested_pid; + LASSERT(the_lnet.ln_refcount == 0); - } else {/* client case (liblustre) */ + the_lnet.ln_routing = 0; - /* My PID must be unique on this node and flag I'm userspace */ - the_lnet.ln_pid = getpid() | LNET_PID_USERFLAG; - } -#endif + LASSERT((requested_pid & LNET_PID_USERFLAG) == 0); + the_lnet.ln_pid = requested_pid; INIT_LIST_HEAD(&the_lnet.ln_test_peers); - INIT_LIST_HEAD(&the_lnet.ln_nis); - INIT_LIST_HEAD(&the_lnet.ln_nis_cpt); - INIT_LIST_HEAD(&the_lnet.ln_nis_zombie); + INIT_LIST_HEAD(&the_lnet.ln_peers); + INIT_LIST_HEAD(&the_lnet.ln_nets); INIT_LIST_HEAD(&the_lnet.ln_routers); INIT_LIST_HEAD(&the_lnet.ln_drop_rules); + INIT_LIST_HEAD(&the_lnet.ln_delay_rules); + + rc = lnet_descriptor_setup(); + if (rc != 0) + goto failed; rc = lnet_create_remote_nets_table(); if (rc != 0) goto failed; - the_lnet.ln_interface_cookie = lnet_create_interface_cookie(); + /* + * NB the interface cookie in wire handles guards against delayed + * replies and ACKs appearing valid after reboot. + */ + the_lnet.ln_interface_cookie = ktime_get_real_ns(); the_lnet.ln_counters = cfs_percpt_alloc(lnet_cpt_table(), sizeof(lnet_counters_t)); @@ -781,22 +635,23 @@ lnet_prepare(lnet_pid_t requested_pid) goto failed; rc = lnet_res_container_setup(&the_lnet.ln_eq_container, 0, - LNET_COOKIE_TYPE_EQ, LNET_FL_MAX_EQS, - sizeof(lnet_eq_t)); + LNET_COOKIE_TYPE_EQ); if (rc != 0) goto failed; - recs = lnet_res_containers_create(LNET_COOKIE_TYPE_ME, LNET_FL_MAX_MES, - sizeof(lnet_me_t)); - if (recs == NULL) + recs = lnet_res_containers_create(LNET_COOKIE_TYPE_ME); + if (recs == NULL) { + rc = -ENOMEM; goto failed; + } the_lnet.ln_me_containers = recs; - recs = lnet_res_containers_create(LNET_COOKIE_TYPE_MD, LNET_FL_MAX_MDS, - sizeof(lnet_libmd_t)); - if (recs == NULL) + recs = lnet_res_containers_create(LNET_COOKIE_TYPE_MD); + if (recs == NULL) { + rc = -ENOMEM; goto failed; + } the_lnet.ln_md_containers = recs; @@ -825,9 +680,7 @@ lnet_unprepare (void) LASSERT(the_lnet.ln_refcount == 0); LASSERT(list_empty(&the_lnet.ln_test_peers)); - LASSERT(list_empty(&the_lnet.ln_nis)); - LASSERT(list_empty(&the_lnet.ln_nis_cpt)); - LASSERT(list_empty(&the_lnet.ln_nis_zombie)); + LASSERT(list_empty(&the_lnet.ln_nets)); lnet_portals_destroy(); @@ -852,23 +705,23 @@ lnet_unprepare (void) the_lnet.ln_counters = NULL; } lnet_destroy_remote_nets_table(); + lnet_descriptor_cleanup(); return 0; } lnet_ni_t * -lnet_net2ni_locked(__u32 net, int cpt) +lnet_net2ni_locked(__u32 net_id, int cpt) { - struct list_head *tmp; - lnet_ni_t *ni; + struct lnet_ni *ni; + struct lnet_net *net; LASSERT(cpt != LNET_LOCK_EX); - list_for_each(tmp, &the_lnet.ln_nis) { - ni = list_entry(tmp, lnet_ni_t, ni_list); - - if (LNET_NIDNET(ni->ni_nid) == net) { - lnet_ni_addref_locked(ni, cpt); + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + if (net->net_id == net_id) { + ni = list_entry(net->net_ni_list.next, struct lnet_ni, + ni_netlist); return ni; } } @@ -889,7 +742,20 @@ lnet_net2ni(__u32 net) } EXPORT_SYMBOL(lnet_net2ni); -static unsigned int +struct lnet_net * +lnet_get_net_locked(__u32 net_id) +{ + struct lnet_net *net; + + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + if (net->net_id == net_id) + return net; + } + + return NULL; +} + +unsigned int lnet_nid_cpt_hash(lnet_nid_t nid, unsigned int number) { __u64 key = nid; @@ -909,31 +775,41 @@ lnet_nid_cpt_hash(lnet_nid_t nid, unsigned int number) } int -lnet_cpt_of_nid_locked(lnet_nid_t nid) +lnet_cpt_of_nid_locked(lnet_nid_t nid, struct lnet_ni *ni) { - struct lnet_ni *ni; + struct lnet_net *net; /* must called with hold of lnet_net_lock */ if (LNET_CPT_NUMBER == 1) return 0; /* the only one */ - /* take lnet_net_lock(any) would be OK */ - if (!list_empty(&the_lnet.ln_nis_cpt)) { - list_for_each_entry(ni, &the_lnet.ln_nis_cpt, ni_cptlist) { - if (LNET_NIDNET(ni->ni_nid) != LNET_NIDNET(nid)) - continue; + /* + * If NI is provided then use the CPT identified in the NI cpt + * list if one exists. If one doesn't exist, then that NI is + * associated with all CPTs and it follows that the net it belongs + * to is implicitly associated with all CPTs, so just hash the nid + * and return that. + */ + if (ni != NULL) { + if (ni->ni_cpts != NULL) + return ni->ni_cpts[lnet_nid_cpt_hash(nid, + ni->ni_ncpts)]; + else + return lnet_nid_cpt_hash(nid, LNET_CPT_NUMBER); + } - LASSERT(ni->ni_cpts != NULL); - return ni->ni_cpts[lnet_nid_cpt_hash - (nid, ni->ni_ncpts)]; - } + /* no NI provided so look at the net */ + net = lnet_get_net_locked(LNET_NIDNET(nid)); + + if (net != NULL && net->net_cpts != NULL) { + return net->net_cpts[lnet_nid_cpt_hash(nid, net->net_ncpts)]; } return lnet_nid_cpt_hash(nid, LNET_CPT_NUMBER); } int -lnet_cpt_of_nid(lnet_nid_t nid) +lnet_cpt_of_nid(lnet_nid_t nid, struct lnet_ni *ni) { int cpt; int cpt2; @@ -941,11 +817,10 @@ lnet_cpt_of_nid(lnet_nid_t nid) if (LNET_CPT_NUMBER == 1) return 0; /* the only one */ - if (list_empty(&the_lnet.ln_nis_cpt)) - return lnet_nid_cpt_hash(nid, LNET_CPT_NUMBER); - cpt = lnet_net_lock_current(); - cpt2 = lnet_cpt_of_nid_locked(nid); + + cpt2 = lnet_cpt_of_nid_locked(nid, ni); + lnet_net_unlock(cpt); return cpt2; @@ -953,42 +828,66 @@ lnet_cpt_of_nid(lnet_nid_t nid) EXPORT_SYMBOL(lnet_cpt_of_nid); int -lnet_islocalnet(__u32 net) +lnet_islocalnet(__u32 net_id) { - struct lnet_ni *ni; + struct lnet_net *net; int cpt; + bool local; cpt = lnet_net_lock_current(); - ni = lnet_net2ni_locked(net, cpt); - if (ni != NULL) - lnet_ni_decref_locked(ni, cpt); + net = lnet_get_net_locked(net_id); + + local = net != NULL; lnet_net_unlock(cpt); - return ni != NULL; + return local; +} + +bool +lnet_is_ni_healthy_locked(struct lnet_ni *ni) +{ + if (ni->ni_state == LNET_NI_STATE_ACTIVE || + ni->ni_state == LNET_NI_STATE_DEGRADED) + return true; + + return false; } lnet_ni_t * lnet_nid2ni_locked(lnet_nid_t nid, int cpt) { + struct lnet_net *net; struct lnet_ni *ni; - struct list_head *tmp; LASSERT(cpt != LNET_LOCK_EX); - list_for_each(tmp, &the_lnet.ln_nis) { - ni = list_entry(tmp, lnet_ni_t, ni_list); - - if (ni->ni_nid == nid) { - lnet_ni_addref_locked(ni, cpt); - return ni; + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) { + if (ni->ni_nid == nid) + return ni; } } return NULL; } +lnet_ni_t * +lnet_nid2ni_addref(lnet_nid_t nid) +{ + lnet_ni_t *ni; + + lnet_net_lock(0); + ni = lnet_nid2ni_locked(nid, 0); + if (ni) + lnet_ni_addref_locked(ni, 0); + lnet_net_unlock(0); + + return ni; +} +EXPORT_SYMBOL(lnet_nid2ni_addref); + int lnet_islocalnid(lnet_nid_t nid) { @@ -997,44 +896,39 @@ lnet_islocalnid(lnet_nid_t nid) cpt = lnet_net_lock_current(); ni = lnet_nid2ni_locked(nid, cpt); - if (ni != NULL) - lnet_ni_decref_locked(ni, cpt); lnet_net_unlock(cpt); return ni != NULL; } int -lnet_count_acceptor_nis (void) +lnet_count_acceptor_nets(void) { /* Return the # of NIs that need the acceptor. */ int count = 0; -#if defined(__KERNEL__) || defined(HAVE_LIBPTHREAD) - struct list_head *tmp; - struct lnet_ni *ni; + struct lnet_net *net; int cpt; cpt = lnet_net_lock_current(); - list_for_each(tmp, &the_lnet.ln_nis) { - ni = list_entry(tmp, lnet_ni_t, ni_list); - - if (ni->ni_lnd->lnd_accept != NULL) + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + /* all socklnd type networks should have the acceptor + * thread started */ + if (net->net_lnd->lnd_accept != NULL) count++; } lnet_net_unlock(cpt); -#endif /* defined(__KERNEL__) || defined(HAVE_LIBPTHREAD) */ return count; } -static lnet_ping_info_t * +static struct lnet_ping_info * lnet_ping_info_create(int num_ni) { - lnet_ping_info_t *ping_info; + struct lnet_ping_info *ping_info; unsigned int infosz; - infosz = offsetof(lnet_ping_info_t, pi_ni[num_ni]); + infosz = offsetof(struct lnet_ping_info, pi_ni[num_ni]); LIBCFS_ALLOC(ping_info, infosz); if (ping_info == NULL) { CERROR("Can't allocate ping info[%d]\n", num_ni); @@ -1050,15 +944,30 @@ lnet_ping_info_create(int num_ni) } static inline int +lnet_get_net_ni_count_locked(struct lnet_net *net) +{ + struct lnet_ni *ni; + int count = 0; + + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) + count++; + + return count; +} + +static inline int lnet_get_ni_count(void) { - struct lnet_ni *ni; - int count = 0; + struct lnet_ni *ni; + struct lnet_net *net; + int count = 0; lnet_net_lock(0); - list_for_each_entry(ni, &the_lnet.ln_nis, ni_list) - count++; + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) + count++; + } lnet_net_unlock(0); @@ -1066,24 +975,27 @@ lnet_get_ni_count(void) } static inline void -lnet_ping_info_free(lnet_ping_info_t *pinfo) +lnet_ping_info_free(struct lnet_ping_info *pinfo) { LIBCFS_FREE(pinfo, - offsetof(lnet_ping_info_t, + offsetof(struct lnet_ping_info, pi_ni[pinfo->pi_nnis])); } static void lnet_ping_info_destroy(void) { + struct lnet_net *net; struct lnet_ni *ni; lnet_net_lock(LNET_LOCK_EX); - list_for_each_entry(ni, &the_lnet.ln_nis, ni_list) { - lnet_ni_lock(ni); - ni->ni_status = NULL; - lnet_ni_unlock(ni); + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) { + lnet_ni_lock(ni); + ni->ni_status = NULL; + lnet_ni_unlock(ni); + } } lnet_ping_info_free(the_lnet.ln_ping_info); @@ -1095,19 +1007,19 @@ lnet_ping_info_destroy(void) static void lnet_ping_event_handler(lnet_event_t *event) { - lnet_ping_info_t *pinfo = event->md.user_ptr; + struct lnet_ping_info *pinfo = event->md.user_ptr; if (event->unlinked) pinfo->pi_features = LNET_PING_FEAT_INVAL; } static int -lnet_ping_info_setup(lnet_ping_info_t **ppinfo, lnet_handle_md_t *md_handle, +lnet_ping_info_setup(struct lnet_ping_info **ppinfo, lnet_handle_md_t *md_handle, int ni_count, bool set_eq) { lnet_handle_me_t me_handle; lnet_process_id_t id = {LNET_NID_ANY, LNET_PID_ANY}; - lnet_md_t md = {0}; + lnet_md_t md = {NULL}; int rc, rc2; if (set_eq) { @@ -1136,7 +1048,7 @@ lnet_ping_info_setup(lnet_ping_info_t **ppinfo, lnet_handle_md_t *md_handle, /* initialize md content */ md.start = *ppinfo; - md.length = offsetof(lnet_ping_info_t, + md.length = offsetof(struct lnet_ping_info, pi_ni[(*ppinfo)->pi_nnis]); md.threshold = LNET_MD_THRESH_INF; md.max_size = 0; @@ -1167,7 +1079,7 @@ failed_0: } static void -lnet_ping_md_unlink(lnet_ping_info_t *pinfo, lnet_handle_md_t *md_handle) +lnet_ping_md_unlink(struct lnet_ping_info *pinfo, lnet_handle_md_t *md_handle) { sigset_t blocked = cfs_block_allsigs(); @@ -1177,41 +1089,47 @@ lnet_ping_md_unlink(lnet_ping_info_t *pinfo, lnet_handle_md_t *md_handle) /* NB md could be busy; this just starts the unlink */ while (pinfo->pi_features != LNET_PING_FEAT_INVAL) { CDEBUG(D_NET, "Still waiting for ping MD to unlink\n"); - cfs_pause(cfs_time_seconds(1)); + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(cfs_time_seconds(1)); } cfs_restore_sigs(blocked); } static void -lnet_ping_info_install_locked(lnet_ping_info_t *ping_info) +lnet_ping_info_install_locked(struct lnet_ping_info *ping_info) { int i; - lnet_ni_t *ni; - lnet_ni_status_t *ns; + struct lnet_ni *ni; + struct lnet_net *net; + struct lnet_ni_status *ns; i = 0; - list_for_each_entry(ni, &the_lnet.ln_nis, ni_list) { - LASSERT(i < ping_info->pi_nnis); + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) { + LASSERT(i < ping_info->pi_nnis); - ns = &ping_info->pi_ni[i]; + ns = &ping_info->pi_ni[i]; - ns->ns_nid = ni->ni_nid; + ns->ns_nid = ni->ni_nid; - lnet_ni_lock(ni); - ns->ns_status = (ni->ni_status != NULL) ? - ni->ni_status->ns_status : LNET_NI_STATUS_UP; - ni->ni_status = ns; - lnet_ni_unlock(ni); + lnet_ni_lock(ni); + ns->ns_status = (ni->ni_status != NULL) ? + ni->ni_status->ns_status : + LNET_NI_STATUS_UP; + ni->ni_status = ns; + lnet_ni_unlock(ni); + + i++; + } - i++; } } static void -lnet_ping_target_update(lnet_ping_info_t *pinfo, lnet_handle_md_t md_handle) +lnet_ping_target_update(struct lnet_ping_info *pinfo, lnet_handle_md_t md_handle) { - lnet_ping_info_t *old_pinfo = NULL; + struct lnet_ping_info *old_pinfo = NULL; lnet_handle_md_t old_md; /* switch the NIs to point to the new ping info created */ @@ -1240,7 +1158,7 @@ lnet_ping_target_update(lnet_ping_info_t *pinfo, lnet_handle_md_t md_handle) static void lnet_ping_target_fini(void) { - int rc; + int rc; lnet_ping_md_unlink(the_lnet.ln_ping_info, &the_lnet.ln_ping_target_md); @@ -1259,41 +1177,61 @@ lnet_ni_tq_credits(lnet_ni_t *ni) LASSERT(ni->ni_ncpts >= 1); if (ni->ni_ncpts == 1) - return ni->ni_maxtxcredits; + return ni->ni_net->net_tunables.lct_max_tx_credits; - credits = ni->ni_maxtxcredits / ni->ni_ncpts; - credits = max(credits, 8 * ni->ni_peertxcredits); - credits = min(credits, ni->ni_maxtxcredits); + credits = ni->ni_net->net_tunables.lct_max_tx_credits / ni->ni_ncpts; + credits = max(credits, 8 * ni->ni_net->net_tunables.lct_peer_tx_credits); + credits = min(credits, ni->ni_net->net_tunables.lct_max_tx_credits); return credits; } static void -lnet_clear_zombies_nis_locked(void) +lnet_ni_unlink_locked(lnet_ni_t *ni) +{ + if (!list_empty(&ni->ni_cptlist)) { + list_del_init(&ni->ni_cptlist); + lnet_ni_decref_locked(ni, 0); + } + + /* move it to zombie list and nobody can find it anymore */ + LASSERT(!list_empty(&ni->ni_netlist)); + list_move(&ni->ni_netlist, &ni->ni_net->net_ni_zombie); + lnet_ni_decref_locked(ni, 0); +} + +static void +lnet_clear_zombies_nis_locked(struct lnet_net *net) { int i; int islo; lnet_ni_t *ni; + struct list_head *zombie_list = &net->net_ni_zombie; - /* Now wait for the NI's I just nuked to show up on ln_zombie_nis - * and shut them down in guaranteed thread context */ + /* + * Now wait for the NIs I just nuked to show up on the zombie + * list and shut them down in guaranteed thread context + */ i = 2; - while (!list_empty(&the_lnet.ln_nis_zombie)) { + while (!list_empty(zombie_list)) { int *ref; int j; - ni = list_entry(the_lnet.ln_nis_zombie.next, - lnet_ni_t, ni_list); - list_del_init(&ni->ni_list); + ni = list_entry(zombie_list->next, + lnet_ni_t, ni_netlist); + list_del_init(&ni->ni_netlist); + /* the ni should be in deleting state. If it's not it's + * a bug */ + LASSERT(ni->ni_state == LNET_NI_STATE_DELETING); cfs_percpt_for_each(ref, j, ni->ni_refs) { if (*ref == 0) continue; /* still busy, add it back to zombie list */ - list_add(&ni->ni_list, &the_lnet.ln_nis_zombie); + list_add(&ni->ni_netlist, zombie_list); break; } - if (!list_empty(&ni->ni_list)) { + if (!list_empty(&ni->ni_netlist)) { lnet_net_unlock(LNET_LOCK_EX); ++i; if ((i & (-i)) == i) { @@ -1301,21 +1239,18 @@ lnet_clear_zombies_nis_locked(void) "Waiting for zombie LNI %s\n", libcfs_nid2str(ni->ni_nid)); } - cfs_pause(cfs_time_seconds(1)); + set_current_state(TASK_UNINTERRUPTIBLE); + schedule_timeout(cfs_time_seconds(1)); lnet_net_lock(LNET_LOCK_EX); continue; } - ni->ni_lnd->lnd_refcount--; lnet_net_unlock(LNET_LOCK_EX); - islo = ni->ni_lnd->lnd_type == LOLND; + islo = ni->ni_net->net_lnd->lnd_type == LOLND; LASSERT(!in_interrupt()); - (ni->ni_lnd->lnd_shutdown)(ni); - - /* can't deref lnd anymore now; it might have unregistered - * itself... */ + (net->net_lnd->lnd_shutdown)(ni); if (!islo) CDEBUG(D_LNI, "Removed LNI %s\n", @@ -1327,335 +1262,409 @@ lnet_clear_zombies_nis_locked(void) } } +/* shutdown down the NI and release refcount */ static void -lnet_shutdown_lndnis(void) +lnet_shutdown_lndni(struct lnet_ni *ni) { - int i; - lnet_ni_t *ni; + int i; + struct lnet_net *net = ni->ni_net; + + lnet_net_lock(LNET_LOCK_EX); + ni->ni_state = LNET_NI_STATE_DELETING; + lnet_ni_unlink_locked(ni); + lnet_net_unlock(LNET_LOCK_EX); + + /* clear messages for this NI on the lazy portal */ + for (i = 0; i < the_lnet.ln_nportals; i++) + lnet_clear_lazy_portal(ni, i, "Shutting down NI"); + + /* Do peer table cleanup for this ni */ + lnet_peer_tables_cleanup(ni); + + lnet_net_lock(LNET_LOCK_EX); + lnet_clear_zombies_nis_locked(net); + lnet_net_unlock(LNET_LOCK_EX); +} + +static void +lnet_shutdown_lndnet(struct lnet_net *net) +{ + struct lnet_ni *ni; + + lnet_net_lock(LNET_LOCK_EX); + + net->net_state = LNET_NET_STATE_DELETING; + + list_del_init(&net->net_list); + + while (!list_empty(&net->net_ni_list)) { + ni = list_entry(net->net_ni_list.next, + lnet_ni_t, ni_netlist); + lnet_net_unlock(LNET_LOCK_EX); + lnet_shutdown_lndni(ni); + lnet_net_lock(LNET_LOCK_EX); + } + + /* + * decrement ref count on lnd only when the entire network goes + * away + */ + net->net_lnd->lnd_refcount--; + + lnet_net_unlock(LNET_LOCK_EX); + + lnet_net_free(net); +} + +static void +lnet_shutdown_lndnets(void) +{ + struct lnet_net *net; /* NB called holding the global mutex */ /* All quiet on the API front */ LASSERT(!the_lnet.ln_shutdown); LASSERT(the_lnet.ln_refcount == 0); - LASSERT(list_empty(&the_lnet.ln_nis_zombie)); lnet_net_lock(LNET_LOCK_EX); the_lnet.ln_shutdown = 1; /* flag shutdown */ - /* Unlink NIs from the global table */ - while (!list_empty(&the_lnet.ln_nis)) { - ni = list_entry(the_lnet.ln_nis.next, - lnet_ni_t, ni_list); - /* move it to zombie list and nobody can find it anymore */ - list_move(&ni->ni_list, &the_lnet.ln_nis_zombie); - lnet_ni_decref_locked(ni, 0); /* drop ln_nis' ref */ - - if (!list_empty(&ni->ni_cptlist)) { - list_del_init(&ni->ni_cptlist); - lnet_ni_decref_locked(ni, 0); - } - } - - /* Drop the cached eqwait NI. */ - if (the_lnet.ln_eq_waitni != NULL) { - lnet_ni_decref_locked(the_lnet.ln_eq_waitni, 0); - the_lnet.ln_eq_waitni = NULL; + while (!list_empty(&the_lnet.ln_nets)) { + /* + * move the nets to the zombie list to avoid them being + * picked up for new work. LONET is also included in the + * Nets that will be moved to the zombie list + */ + net = list_entry(the_lnet.ln_nets.next, + struct lnet_net, net_list); + list_move(&net->net_list, &the_lnet.ln_net_zombie); } - /* Drop the cached loopback NI. */ + /* Drop the cached loopback Net. */ if (the_lnet.ln_loni != NULL) { lnet_ni_decref_locked(the_lnet.ln_loni, 0); the_lnet.ln_loni = NULL; } - lnet_net_unlock(LNET_LOCK_EX); - /* Clear lazy portals and drop delayed messages which hold refs - * on their lnet_msg_t::msg_rxpeer */ - for (i = 0; i < the_lnet.ln_nportals; i++) - LNetClearLazyPortal(i); - - /* Clear the peer table and wait for all peers to go (they hold refs on - * their NIs) */ - lnet_peer_tables_cleanup(NULL); + /* iterate through the net zombie list and delete each net */ + while (!list_empty(&the_lnet.ln_net_zombie)) { + net = list_entry(the_lnet.ln_net_zombie.next, + struct lnet_net, net_list); + lnet_shutdown_lndnet(net); + } lnet_net_lock(LNET_LOCK_EX); - - lnet_clear_zombies_nis_locked(); the_lnet.ln_shutdown = 0; lnet_net_unlock(LNET_LOCK_EX); } -int -lnet_shutdown_lndni(__u32 net) +static int +lnet_startup_lndni(struct lnet_ni *ni, struct lnet_lnd_tunables *tun) { - lnet_ping_info_t *pinfo; - lnet_handle_md_t md_handle; - lnet_ni_t *found_ni = NULL; - int ni_count; - int rc; + int rc = -EINVAL; + struct lnet_tx_queue *tq; + int i; + struct lnet_net *net = ni->ni_net; - if (LNET_NETTYP(net) == LOLND) - return -EINVAL; + mutex_lock(&the_lnet.ln_lnd_mutex); - ni_count = lnet_get_ni_count(); + if (tun) { + memcpy(&ni->ni_lnd_tunables, tun, sizeof(*tun)); + ni->ni_lnd_tunables_set = true; + } - /* create and link a new ping info, before removing the old one */ - rc = lnet_ping_info_setup(&pinfo, &md_handle, ni_count - 1, false); - if (rc != 0) - return rc; + rc = (net->net_lnd->lnd_startup)(ni); - /* proceed with shutting down the NI */ - lnet_net_lock(LNET_LOCK_EX); + mutex_unlock(&the_lnet.ln_lnd_mutex); - found_ni = lnet_net2ni_locked(net, 0); - if (found_ni == NULL) { + if (rc != 0) { + LCONSOLE_ERROR_MSG(0x105, "Error %d starting up LNI %s\n", + rc, libcfs_lnd2str(net->net_lnd->lnd_type)); + lnet_net_lock(LNET_LOCK_EX); + net->net_lnd->lnd_refcount--; lnet_net_unlock(LNET_LOCK_EX); - lnet_ping_md_unlink(pinfo, &md_handle); - lnet_ping_info_free(pinfo); - return -EINVAL; + goto failed0; } - /* decrement the reference counter on found_ni which was - * incremented when we called lnet_net2ni_locked() */ - lnet_ni_decref_locked(found_ni, 0); + ni->ni_state = LNET_NI_STATE_ACTIVE; - /* Move ni to zombie list so nobody can find it anymore */ - list_move(&found_ni->ni_list, &the_lnet.ln_nis_zombie); - - /* Drop the lock reference for the ln_nis ref. */ - lnet_ni_decref_locked(found_ni, 0); - - if (!list_empty(&found_ni->ni_cptlist)) { - list_del_init(&found_ni->ni_cptlist); - lnet_ni_decref_locked(found_ni, 0); + /* We keep a reference on the loopback net through the loopback NI */ + if (net->net_lnd->lnd_type == LOLND) { + lnet_ni_addref(ni); + LASSERT(the_lnet.ln_loni == NULL); + the_lnet.ln_loni = ni; + ni->ni_net->net_tunables.lct_peer_tx_credits = 0; + ni->ni_net->net_tunables.lct_peer_rtr_credits = 0; + ni->ni_net->net_tunables.lct_max_tx_credits = 0; + ni->ni_net->net_tunables.lct_peer_timeout = 0; + return 0; } - lnet_net_unlock(LNET_LOCK_EX); - - /* Do peer table cleanup for this ni */ - lnet_peer_tables_cleanup(found_ni); + if (ni->ni_net->net_tunables.lct_peer_tx_credits == 0 || + ni->ni_net->net_tunables.lct_max_tx_credits == 0) { + LCONSOLE_ERROR_MSG(0x107, "LNI %s has no %scredits\n", + libcfs_lnd2str(net->net_lnd->lnd_type), + ni->ni_net->net_tunables.lct_peer_tx_credits == 0 ? + "" : "per-peer "); + /* shutdown the NI since if we get here then it must've already + * been started + */ + lnet_shutdown_lndni(ni); + return -EINVAL; + } - lnet_net_lock(LNET_LOCK_EX); - lnet_clear_zombies_nis_locked(); - lnet_net_unlock(LNET_LOCK_EX); + cfs_percpt_for_each(tq, i, ni->ni_tx_queues) { + tq->tq_credits_min = + tq->tq_credits_max = + tq->tq_credits = lnet_ni_tq_credits(ni); + } - lnet_ping_target_update(pinfo, md_handle); + CDEBUG(D_LNI, "Added LNI %s [%d/%d/%d/%d]\n", + libcfs_nid2str(ni->ni_nid), + ni->ni_net->net_tunables.lct_peer_tx_credits, + lnet_ni_tq_credits(ni) * LNET_CPT_NUMBER, + ni->ni_net->net_tunables.lct_peer_rtr_credits, + ni->ni_net->net_tunables.lct_peer_timeout); return 0; +failed0: + lnet_ni_free(ni); + return rc; } static int -lnet_startup_lndnis(struct list_head *nilist, __s32 peer_timeout, - __s32 peer_cr, __s32 peer_buf_cr, __s32 credits, - int *ni_count) +lnet_startup_lndnet(struct lnet_net *net, struct lnet_lnd_tunables *tun) { - int rc = 0; struct lnet_ni *ni; - int lnd_type; + struct lnet_net *net_l = NULL; + struct list_head local_ni_list; + int rc; + int ni_count = 0; + __u32 lnd_type; lnd_t *lnd; - struct lnet_tx_queue *tq; - int i; + int peer_timeout = + net->net_tunables.lct_peer_timeout; + int maxtxcredits = + net->net_tunables.lct_max_tx_credits; + int peerrtrcredits = + net->net_tunables.lct_peer_rtr_credits; + + INIT_LIST_HEAD(&local_ni_list); + + /* + * make sure that this net is unique. If it isn't then + * we are adding interfaces to an already existing network, and + * 'net' is just a convenient way to pass in the list. + * if it is unique we need to find the LND and load it if + * necessary. + */ + if (lnet_net_unique(net->net_id, &the_lnet.ln_nets, &net_l)) { + lnd_type = LNET_NETTYP(net->net_id); + + LASSERT(libcfs_isknown_lnd(lnd_type)); + + if (lnd_type == CIBLND || lnd_type == OPENIBLND || + lnd_type == IIBLND || lnd_type == VIBLND) { + CERROR("LND %s obsoleted\n", libcfs_lnd2str(lnd_type)); + rc = -EINVAL; + goto failed0; + } - while (!list_empty(nilist)) { - ni = list_entry(nilist->next, lnet_ni_t, ni_list); - lnd_type = LNET_NETTYP(LNET_NIDNET(ni->ni_nid)); + mutex_lock(&the_lnet.ln_lnd_mutex); + lnd = lnet_find_lnd_by_type(lnd_type); - if (!libcfs_isknown_lnd(lnd_type)) - goto failed; - - if (lnd_type == CIBLND || - lnd_type == OPENIBLND || - lnd_type == IIBLND || - lnd_type == VIBLND) { - CERROR("LND %s obsoleted\n", - libcfs_lnd2str(lnd_type)); - goto failed; - } - - /* Make sure this new NI is unique. */ - lnet_net_lock(LNET_LOCK_EX); - if (!lnet_net_unique(LNET_NIDNET(ni->ni_nid), - &the_lnet.ln_nis)) { - if (lnd_type == LOLND) { - lnet_net_unlock(LNET_LOCK_EX); - list_del(&ni->ni_list); - lnet_ni_free(ni); - continue; - } - - CERROR("Net %s is not unique\n", - libcfs_net2str(LNET_NIDNET(ni->ni_nid))); - lnet_net_unlock(LNET_LOCK_EX); - goto failed; - } - lnet_net_unlock(LNET_LOCK_EX); - - LNET_MUTEX_LOCK(&the_lnet.ln_lnd_mutex); - lnd = lnet_find_lnd_by_type(lnd_type); - -#ifdef __KERNEL__ - if (lnd == NULL) { - LNET_MUTEX_UNLOCK(&the_lnet.ln_lnd_mutex); - rc = request_module("%s", - libcfs_lnd2modname(lnd_type)); - LNET_MUTEX_LOCK(&the_lnet.ln_lnd_mutex); + if (lnd == NULL) { + mutex_unlock(&the_lnet.ln_lnd_mutex); + rc = request_module("%s", libcfs_lnd2modname(lnd_type)); + mutex_lock(&the_lnet.ln_lnd_mutex); lnd = lnet_find_lnd_by_type(lnd_type); if (lnd == NULL) { - LNET_MUTEX_UNLOCK(&the_lnet.ln_lnd_mutex); + mutex_unlock(&the_lnet.ln_lnd_mutex); CERROR("Can't load LND %s, module %s, rc=%d\n", - libcfs_lnd2str(lnd_type), - libcfs_lnd2modname(lnd_type), rc); + libcfs_lnd2str(lnd_type), + libcfs_lnd2modname(lnd_type), rc); #ifndef HAVE_MODULE_LOADING_SUPPORT LCONSOLE_ERROR_MSG(0x104, "Your kernel must be " - "compiled with kernel module " - "loading support."); + "compiled with kernel module " + "loading support."); #endif - goto failed; + rc = -EINVAL; + goto failed0; } } -#else - if (lnd == NULL) { - LNET_MUTEX_UNLOCK(&the_lnet.ln_lnd_mutex); - CERROR("LND %s not supported\n", - libcfs_lnd2str(lnd_type)); - goto failed; - } -#endif lnet_net_lock(LNET_LOCK_EX); lnd->lnd_refcount++; lnet_net_unlock(LNET_LOCK_EX); - ni->ni_lnd = lnd; - - rc = (lnd->lnd_startup)(ni); + net->net_lnd = lnd; + + mutex_unlock(&the_lnet.ln_lnd_mutex); + + net_l = net; + } + + /* + * net_l: if the network being added is unique then net_l + * will point to that network + * if the network being added is not unique then + * net_l points to the existing network. + * + * When we enter the loop below, we'll pick NIs off he + * network beign added and start them up, then add them to + * a local ni list. Once we've successfully started all + * the NIs then we join the local NI list (of started up + * networks) with the net_l->net_ni_list, which should + * point to the correct network to add the new ni list to + * + * If any of the new NIs fail to start up, then we want to + * iterate through the local ni list, which should include + * any NIs which were successfully started up, and shut + * them down. + * + * After than we want to delete the network being added, + * to avoid a memory leak. + */ + + /* + * When a network uses TCP bonding then all its interfaces + * must be specified when the network is first defined: the + * TCP bonding code doesn't allow for interfaces to be added + * or removed. + */ + if (net_l != net && net_l != NULL && use_tcp_bonding && + LNET_NETTYP(net_l->net_id) == SOCKLND) { + rc = -EINVAL; + goto failed0; + } - LNET_MUTEX_UNLOCK(&the_lnet.ln_lnd_mutex); + while (!list_empty(&net->net_ni_added)) { + ni = list_entry(net->net_ni_added.next, struct lnet_ni, + ni_netlist); + list_del_init(&ni->ni_netlist); - if (rc != 0) { - LCONSOLE_ERROR_MSG(0x105, "Error %d starting up LNI %s" - "\n", - rc, libcfs_lnd2str(lnd->lnd_type)); - lnet_net_lock(LNET_LOCK_EX); - lnd->lnd_refcount--; - lnet_net_unlock(LNET_LOCK_EX); - goto failed; + /* make sure that the the NI we're about to start + * up is actually unique. if it's not fail. */ + if (!lnet_ni_unique_net(&net_l->net_ni_list, + ni->ni_interfaces[0])) { + rc = -EINVAL; + goto failed1; } - /* If given some LND tunable parameters, parse those now to - * override the values in the NI structure. */ - if (peer_buf_cr >= 0) - ni->ni_peerrtrcredits = peer_buf_cr; - if (peer_timeout >= 0) - ni->ni_peertimeout = peer_timeout; + /* adjust the pointer the parent network, just in case it + * the net is a duplicate */ + ni->ni_net = net_l; + + rc = lnet_startup_lndni(ni, tun); + + LASSERT(ni->ni_net->net_tunables.lct_peer_timeout <= 0 || + ni->ni_net->net_lnd->lnd_query != NULL); + + if (rc < 0) + goto failed1; + + lnet_ni_addref(ni); + list_add_tail(&ni->ni_netlist, &local_ni_list); + + ni_count++; + } + + lnet_net_lock(LNET_LOCK_EX); + list_splice_tail(&local_ni_list, &net_l->net_ni_list); + lnet_incr_dlc_seq(); + lnet_net_unlock(LNET_LOCK_EX); + + /* if the network is not unique then we don't want to keep + * it around after we're done. Free it. Otherwise add that + * net to the global the_lnet.ln_nets */ + if (net_l != net && net_l != NULL) { /* - * TODO - * Note: For now, don't allow the user to change - * peertxcredits as this number is used in the - * IB LND to control queue depth. - * if (peer_cr != -1) - * ni->ni_peertxcredits = peer_cr; + * TODO - note. currently the tunables can not be updated + * once added */ - if (credits >= 0) - ni->ni_maxtxcredits = credits; + lnet_net_free(net); + } else { + net->net_state = LNET_NET_STATE_ACTIVE; + /* + * restore tunables after it has been overwitten by the + * lnd + */ + if (peer_timeout != -1) + net->net_tunables.lct_peer_timeout = peer_timeout; + if (maxtxcredits != -1) + net->net_tunables.lct_max_tx_credits = maxtxcredits; + if (peerrtrcredits != -1) + net->net_tunables.lct_peer_rtr_credits = peerrtrcredits; - LASSERT(ni->ni_peertimeout <= 0 || lnd->lnd_query != NULL); + lnet_net_lock(LNET_LOCK_EX); + list_add_tail(&net->net_list, &the_lnet.ln_nets); + lnet_net_unlock(LNET_LOCK_EX); + } - list_del(&ni->ni_list); + return ni_count; - lnet_net_lock(LNET_LOCK_EX); - /* refcount for ln_nis */ - lnet_ni_addref_locked(ni, 0); - list_add_tail(&ni->ni_list, &the_lnet.ln_nis); - if (ni->ni_cpts != NULL) { - list_add_tail(&ni->ni_cptlist, - &the_lnet.ln_nis_cpt); - lnet_ni_addref_locked(ni, 0); - } +failed1: + /* + * shutdown the new NIs that are being started up + * free the NET being started + */ + while (!list_empty(&local_ni_list)) { + ni = list_entry(local_ni_list.next, struct lnet_ni, + ni_netlist); - lnet_net_unlock(LNET_LOCK_EX); + lnet_shutdown_lndni(ni); + } - /* increment the ni_count here to account for the LOLND as - * well. If we increment past this point then the number - * of count will be missing the LOLND, and then ping and - * will not report the LOLND - */ - if (ni_count != NULL) - (*ni_count)++; +failed0: + lnet_net_free(net); - if (lnd->lnd_type == LOLND) { - lnet_ni_addref(ni); - LASSERT(the_lnet.ln_loni == NULL); - the_lnet.ln_loni = ni; - continue; - } + return rc; +} -#ifndef __KERNEL__ - if (lnd->lnd_wait != NULL) { - if (the_lnet.ln_eq_waitni == NULL) { - lnet_ni_addref(ni); - the_lnet.ln_eq_waitni = ni; - } - } else { -# ifndef HAVE_LIBPTHREAD - LCONSOLE_ERROR_MSG(0x106, "LND %s not supported in a " - "single-threaded runtime\n", - libcfs_lnd2str(lnd_type)); - goto failed; -# endif - } -#endif - if (ni->ni_peertxcredits == 0 || - ni->ni_maxtxcredits == 0) { - LCONSOLE_ERROR_MSG(0x107, "LNI %s has no %scredits\n", - libcfs_lnd2str(lnd->lnd_type), - ni->ni_peertxcredits == 0 ? - "" : "per-peer "); - goto failed; - } +static int +lnet_startup_lndnets(struct list_head *netlist) +{ + struct lnet_net *net; + int rc; + int ni_count = 0; - cfs_percpt_for_each(tq, i, ni->ni_tx_queues) { - tq->tq_credits_min = - tq->tq_credits_max = - tq->tq_credits = lnet_ni_tq_credits(ni); - } + while (!list_empty(netlist)) { + net = list_entry(netlist->next, struct lnet_net, net_list); + list_del_init(&net->net_list); - CDEBUG(D_LNI, "Added LNI %s [%d/%d/%d/%d]\n", - libcfs_nid2str(ni->ni_nid), ni->ni_peertxcredits, - lnet_ni_tq_credits(ni) * LNET_CPT_NUMBER, - ni->ni_peerrtrcredits, ni->ni_peertimeout); + rc = lnet_startup_lndnet(net, NULL); + + if (rc < 0) + goto failed; + + ni_count += rc; } - return 0; + return ni_count; failed: - while (!list_empty(nilist)) { - ni = list_entry(nilist->next, lnet_ni_t, ni_list); - list_del(&ni->ni_list); - lnet_ni_free(ni); - } - return -EINVAL; + lnet_shutdown_lndnets(); + + return rc; } /** * Initialize LNet library. * - * Only userspace program needs to call this function - it's automatically - * called in the kernel at module loading time. Caller has to call LNetFini() - * after a call to LNetInit(), if and only if the latter returned 0. It must - * be called exactly once. + * Automatically called at module loading time. Caller has to call + * lnet_lib_exit() after a call to lnet_lib_init(), if and only if the + * latter returned 0. It must be called exactly once. * - * \return 0 on success, and -ve on failures. + * \retval 0 on success + * \retval -ve on failures. */ -int -LNetInit(void) +int lnet_lib_init(void) { - int rc; + int rc; lnet_assert_wire_constants(); - LASSERT(!the_lnet.ln_init); memset(&the_lnet, 0, sizeof(the_lnet)); @@ -1669,7 +1678,7 @@ LNetInit(void) CERROR("Can't have %d CPTs for LNet (max allowed is %d), " "please change setting of CPT-table and retry\n", the_lnet.ln_cpt_number, LNET_CPT_MAX); - return -1; + return -E2BIG; } while ((1 << the_lnet.ln_cpt_bits) < the_lnet.ln_cpt_number) @@ -1678,17 +1687,16 @@ LNetInit(void) rc = lnet_create_locks(); if (rc != 0) { CERROR("Can't create LNet global locks: %d\n", rc); - return -1; + return rc; } the_lnet.ln_refcount = 0; - the_lnet.ln_init = 1; LNetInvalidateHandle(&the_lnet.ln_rc_eqh); INIT_LIST_HEAD(&the_lnet.ln_lnds); + INIT_LIST_HEAD(&the_lnet.ln_net_zombie); INIT_LIST_HEAD(&the_lnet.ln_rcd_zombie); INIT_LIST_HEAD(&the_lnet.ln_rcd_deathrow); -#ifdef __KERNEL__ /* The hash table size is the number of bits it takes to express the set * ln_num_routes, minus 1 (better to under estimate than over so we * don't waste memory). */ @@ -1702,48 +1710,29 @@ LNetInit(void) /* All LNDs apart from the LOLND are in separate modules. They * register themselves when their module loads, and unregister * themselves when their module is unloaded. */ -#else - the_lnet.ln_remote_nets_hbits = 8; - - /* Register LNDs - * NB the order here determines default 'networks=' order */ -# ifdef HAVE_LIBPTHREAD - LNET_REGISTER_ULND(the_tcplnd); -# endif -#endif lnet_register_lnd(&the_lolnd); return 0; } -EXPORT_SYMBOL(LNetInit); /** * Finalize LNet library. * - * Only userspace program needs to call this function. It can be called - * at most once. - * - * \pre LNetInit() called with success. + * \pre lnet_lib_init() called with success. * \pre All LNet users called LNetNIFini() for matching LNetNIInit() calls. */ -void -LNetFini(void) +void lnet_lib_exit(void) { - LASSERT(the_lnet.ln_init); LASSERT(the_lnet.ln_refcount == 0); while (!list_empty(&the_lnet.ln_lnds)) lnet_unregister_lnd(list_entry(the_lnet.ln_lnds.next, lnd_t, lnd_list)); lnet_destroy_locks(); - - the_lnet.ln_init = 0; } -EXPORT_SYMBOL(LNetFini); /** * Set LNet PID and start LNet interfaces, routing, and forwarding. * - * Userspace program should call this after a successful call to LNetInit(). * Users must call this function at least once before any other functions. * For each successful call there must be a corresponding call to * LNetNIFini(). For subsequent calls to LNetNIInit(), \a requested_pid is @@ -1761,103 +1750,120 @@ LNetNIInit(lnet_pid_t requested_pid) { int im_a_router = 0; int rc; - int ni_count = 0; - int lnd_type; - struct lnet_ni *ni; - lnet_ping_info_t *pinfo; + int ni_count; + struct lnet_ping_info *pinfo; lnet_handle_md_t md_handle; struct list_head net_head; + struct lnet_net *net; INIT_LIST_HEAD(&net_head); - LNET_MUTEX_LOCK(&the_lnet.ln_api_mutex); + mutex_lock(&the_lnet.ln_api_mutex); - LASSERT(the_lnet.ln_init); CDEBUG(D_OTHER, "refs %d\n", the_lnet.ln_refcount); if (the_lnet.ln_refcount > 0) { rc = the_lnet.ln_refcount++; - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + mutex_unlock(&the_lnet.ln_api_mutex); return rc; } rc = lnet_prepare(requested_pid); - if (rc != 0) - goto failed0; + if (rc != 0) { + mutex_unlock(&the_lnet.ln_api_mutex); + return rc; + } - rc = lnet_parse_networks(&net_head, - !the_lnet.ln_nis_from_mod_params ? - lnet_get_networks() : ""); - if (rc < 0) - goto failed1; + /* create a network for Loopback network */ + net = lnet_net_alloc(LNET_MKNET(LOLND, 0), &net_head); + if (net == NULL) { + rc = -ENOMEM; + goto err_empty_list; + } - rc = lnet_startup_lndnis(&net_head, -1, -1, -1, -1, &ni_count); - if (rc != 0) - goto failed1; + /* Add in the loopback NI */ + if (lnet_ni_alloc(net, NULL, NULL) == NULL) { + rc = -ENOMEM; + goto err_empty_list; + } + + /* If LNet is being initialized via DLC it is possible + * that the user requests not to load module parameters (ones which + * are supported by DLC) on initialization. Therefore, make sure not + * to load networks, routes and forwarding from module parameters + * in this case. On cleanup in case of failure only clean up + * routes if it has been loaded */ + if (!the_lnet.ln_nis_from_mod_params) { + rc = lnet_parse_networks(&net_head, lnet_get_networks(), + use_tcp_bonding); + if (rc < 0) + goto err_empty_list; + } - if (the_lnet.ln_eq_waitni != NULL && ni_count > 1) { - lnd_type = the_lnet.ln_eq_waitni->ni_lnd->lnd_type; - LCONSOLE_ERROR_MSG(0x109, "LND %s can only run single-network" - "\n", - libcfs_lnd2str(lnd_type)); - goto failed2; + ni_count = lnet_startup_lndnets(&net_head); + if (ni_count < 0) { + rc = ni_count; + goto err_empty_list; } - rc = lnet_parse_routes(lnet_get_routes(), &im_a_router); - if (rc != 0) - goto failed2; + if (!the_lnet.ln_nis_from_mod_params) { + rc = lnet_parse_routes(lnet_get_routes(), &im_a_router); + if (rc != 0) + goto err_shutdown_lndnis; - rc = lnet_check_routes(); - if (rc != 0) - goto failed2; + rc = lnet_check_routes(); + if (rc != 0) + goto err_destroy_routes; - rc = lnet_rtrpools_alloc(im_a_router); - if (rc != 0) - goto failed2; + rc = lnet_rtrpools_alloc(im_a_router); + if (rc != 0) + goto err_destroy_routes; + } rc = lnet_acceptor_start(); if (rc != 0) - goto failed2; + goto err_destroy_routes; + the_lnet.ln_refcount = 1; /* Now I may use my own API functions... */ rc = lnet_ping_info_setup(&pinfo, &md_handle, ni_count, true); if (rc != 0) - goto failed3; + goto err_acceptor_stop; lnet_ping_target_update(pinfo, md_handle); rc = lnet_router_checker_start(); if (rc != 0) - goto failed4; + goto err_stop_ping; lnet_fault_init(); lnet_proc_init(); - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + mutex_unlock(&the_lnet.ln_api_mutex); return 0; -failed4: +err_stop_ping: + lnet_ping_target_fini(); +err_acceptor_stop: the_lnet.ln_refcount = 0; - lnet_ping_md_unlink(pinfo, &md_handle); - lnet_ping_info_free(pinfo); -failed3: lnet_acceptor_stop(); - rc = LNetEQFree(the_lnet.ln_ping_target_eq); - LASSERT(rc == 0); -failed2: - lnet_destroy_routes(); - lnet_shutdown_lndnis(); -failed1: +err_destroy_routes: + if (!the_lnet.ln_nis_from_mod_params) + lnet_destroy_routes(); +err_shutdown_lndnis: + lnet_shutdown_lndnets(); +err_empty_list: lnet_unprepare(); -failed0: LASSERT(rc < 0); - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + mutex_unlock(&the_lnet.ln_api_mutex); while (!list_empty(&net_head)) { - ni = list_entry(net_head.next, struct lnet_ni, ni_list); - list_del_init(&ni->ni_list); - lnet_ni_free(ni); + struct lnet_net *net; + + net = list_entry(net_head.next, struct lnet_net, net_list); + list_del_init(&net->net_list); + lnet_net_free(net); } return rc; } @@ -1875,33 +1881,32 @@ EXPORT_SYMBOL(LNetNIInit); int LNetNIFini() { - LNET_MUTEX_LOCK(&the_lnet.ln_api_mutex); + mutex_lock(&the_lnet.ln_api_mutex); - LASSERT (the_lnet.ln_init); - LASSERT (the_lnet.ln_refcount > 0); + LASSERT(the_lnet.ln_refcount > 0); - if (the_lnet.ln_refcount != 1) { - the_lnet.ln_refcount--; - } else { + if (the_lnet.ln_refcount != 1) { + the_lnet.ln_refcount--; + } else { LASSERT(!the_lnet.ln_niinit_self); lnet_fault_fini(); - lnet_proc_fini(); - lnet_router_checker_stop(); - lnet_ping_target_fini(); + lnet_proc_fini(); + lnet_router_checker_stop(); + lnet_ping_target_fini(); - /* Teardown fns that use my own API functions BEFORE here */ - the_lnet.ln_refcount = 0; + /* Teardown fns that use my own API functions BEFORE here */ + the_lnet.ln_refcount = 0; - lnet_acceptor_stop(); - lnet_destroy_routes(); - lnet_shutdown_lndnis(); - lnet_unprepare(); - } + lnet_acceptor_stop(); + lnet_destroy_routes(); + lnet_shutdown_lndnets(); + lnet_unprepare(); + } - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); - return 0; + mutex_unlock(&the_lnet.ln_api_mutex); + return 0; } EXPORT_SYMBOL(LNetNIFini); @@ -1919,73 +1924,155 @@ EXPORT_SYMBOL(LNetNIFini); * \param[out] net_config Network configuration */ static void -lnet_fill_ni_info(struct lnet_ni *ni, __u32 *cpt_count, __u64 *nid, - int *peer_timeout, int *peer_tx_credits, - int *peer_rtr_credits, int *max_tx_credits, - struct lnet_ioctl_net_config *net_config) +lnet_fill_ni_info(struct lnet_ni *ni, struct lnet_ioctl_config_data *config) { + struct lnet_ioctl_net_config *net_config; + struct lnet_ioctl_config_lnd_tunables *lnd_cfg = NULL; + size_t min_size, tunable_size = 0; int i; - if (ni == NULL) + if (!ni || !config) return; - if (net_config == NULL) + net_config = (struct lnet_ioctl_net_config *) config->cfg_bulk; + if (!net_config) return; - CLASSERT(ARRAY_SIZE(ni->ni_interfaces) == - ARRAY_SIZE(net_config->ni_interfaces)); + BUILD_BUG_ON(ARRAY_SIZE(ni->ni_interfaces) != + ARRAY_SIZE(net_config->ni_interfaces)); - if (ni->ni_interfaces[0] != NULL) { - for (i = 0; i < ARRAY_SIZE(ni->ni_interfaces); i++) { - if (ni->ni_interfaces[i] != NULL) { - strncpy(net_config->ni_interfaces[i], - ni->ni_interfaces[i], - sizeof(net_config->ni_interfaces[i])); - } - } + for (i = 0; i < ARRAY_SIZE(ni->ni_interfaces); i++) { + if (!ni->ni_interfaces[i]) + break; + + strncpy(net_config->ni_interfaces[i], + ni->ni_interfaces[i], + sizeof(net_config->ni_interfaces[i])); } - *nid = ni->ni_nid; - *peer_timeout = ni->ni_peertimeout; - *peer_tx_credits = ni->ni_peertxcredits; - *peer_rtr_credits = ni->ni_peerrtrcredits; - *max_tx_credits = ni->ni_maxtxcredits; + config->cfg_nid = ni->ni_nid; + config->cfg_config_u.cfg_net.net_peer_timeout = + ni->ni_net->net_tunables.lct_peer_timeout; + config->cfg_config_u.cfg_net.net_max_tx_credits = + ni->ni_net->net_tunables.lct_max_tx_credits; + config->cfg_config_u.cfg_net.net_peer_tx_credits = + ni->ni_net->net_tunables.lct_peer_tx_credits; + config->cfg_config_u.cfg_net.net_peer_rtr_credits = + ni->ni_net->net_tunables.lct_peer_rtr_credits; net_config->ni_status = ni->ni_status->ns_status; - for (i = 0; - ni->ni_cpts != NULL && i < ni->ni_ncpts && - i < LNET_MAX_SHOW_NUM_CPT; - i++) - net_config->ni_cpts[i] = ni->ni_cpts[i]; + if (ni->ni_cpts) { + int num_cpts = min(ni->ni_ncpts, LNET_MAX_SHOW_NUM_CPT); - *cpt_count = ni->ni_ncpts; + for (i = 0; i < num_cpts; i++) + net_config->ni_cpts[i] = ni->ni_cpts[i]; + + config->cfg_ncpts = num_cpts; + } + + /* + * See if user land tools sent in a newer and larger version + * of struct lnet_tunables than what the kernel uses. + */ + min_size = sizeof(*config) + sizeof(*net_config); + + if (config->cfg_hdr.ioc_len > min_size) + tunable_size = config->cfg_hdr.ioc_len - min_size; + + /* Don't copy too much data to user space */ + min_size = min(tunable_size, sizeof(ni->ni_lnd_tunables)); + lnd_cfg = (struct lnet_ioctl_config_lnd_tunables *)net_config->cfg_bulk; + + if (lnd_cfg && min_size) { + memcpy(&lnd_cfg->lt_tun, &ni->ni_lnd_tunables, min_size); + config->cfg_config_u.cfg_net.net_interface_count = 1; + + /* Tell user land that kernel side has less data */ + if (tunable_size > sizeof(ni->ni_lnd_tunables)) { + min_size = tunable_size - sizeof(ni->ni_lnd_tunables); + config->cfg_hdr.ioc_len -= min_size; + } + } } -int -lnet_get_net_config(int idx, __u32 *cpt_count, __u64 *nid, int *peer_timeout, - int *peer_tx_credits, int *peer_rtr_credits, - int *max_tx_credits, - struct lnet_ioctl_net_config *net_config) +struct lnet_ni * +lnet_get_ni_idx_locked(int idx) +{ + struct lnet_ni *ni; + struct lnet_net *net; + + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) { + if (idx-- == 0) + return ni; + } + } + + return NULL; +} + +struct lnet_ni * +lnet_get_next_ni_locked(struct lnet_net *mynet, struct lnet_ni *prev) { struct lnet_ni *ni; - struct list_head *tmp; - int cpt; - int rc = -ENOENT; + struct lnet_net *net = mynet; + + if (prev == NULL) { + if (net == NULL) + net = list_entry(the_lnet.ln_nets.next, struct lnet_net, + net_list); + ni = list_entry(net->net_ni_list.next, struct lnet_ni, + ni_netlist); + + return ni; + } + + if (prev->ni_netlist.next == &prev->ni_net->net_ni_list) { + /* if you reached the end of the ni list and the net is + * specified, then there are no more nis in that net */ + if (net != NULL) + return NULL; + + /* we reached the end of this net ni list. move to the + * next net */ + if (prev->ni_net->net_list.next == &the_lnet.ln_nets) + /* no more nets and no more NIs. */ + return NULL; + + /* get the next net */ + net = list_entry(prev->ni_net->net_list.next, struct lnet_net, + net_list); + /* get the ni on it */ + ni = list_entry(net->net_ni_list.next, struct lnet_ni, + ni_netlist); + + return ni; + } + + /* there are more nis left */ + ni = list_entry(prev->ni_netlist.next, struct lnet_ni, ni_netlist); + + return ni; +} + +int +lnet_get_net_config(struct lnet_ioctl_config_data *config) +{ + struct lnet_ni *ni; + int cpt; + int rc = -ENOENT; + int idx = config->cfg_count; cpt = lnet_net_lock_current(); - list_for_each(tmp, &the_lnet.ln_nis) { - ni = list_entry(tmp, lnet_ni_t, ni_list); - if (idx-- == 0) { - rc = 0; - lnet_ni_lock(ni); - lnet_fill_ni_info(ni, cpt_count, nid, peer_timeout, - peer_tx_credits, peer_rtr_credits, - max_tx_credits, net_config); - lnet_ni_unlock(ni); - break; - } + ni = lnet_get_ni_idx_locked(idx); + + if (ni != NULL) { + rc = 0; + lnet_ni_lock(ni); + lnet_fill_ni_info(ni, config); + lnet_ni_unlock(ni); } lnet_net_unlock(cpt); @@ -1993,42 +2080,117 @@ lnet_get_net_config(int idx, __u32 *cpt_count, __u64 *nid, int *peer_timeout, } int -lnet_dyn_add_ni(lnet_pid_t requested_pid, char *nets, - __s32 peer_timeout, __s32 peer_cr, __s32 peer_buf_cr, - __s32 credits) +lnet_dyn_add_ni(lnet_pid_t requested_pid, struct lnet_ioctl_config_data *conf) { - lnet_ping_info_t *pinfo; + char *nets = conf->cfg_config_u.cfg_net.net_intf; + struct lnet_ping_info *pinfo; lnet_handle_md_t md_handle; - struct lnet_ni *ni; + struct lnet_net *net; struct list_head net_head; int rc; + lnet_remotenet_t *rnet; + int net_ni_count; + int num_acceptor_nets; + __u32 net_type; + struct lnet_ioctl_config_lnd_tunables *lnd_tunables = NULL; INIT_LIST_HEAD(&net_head); - /* Create a ni structure for the network string */ - rc = lnet_parse_networks(&net_head, nets); - if (rc < 0) - return rc; + if (conf && conf->cfg_hdr.ioc_len > sizeof(*conf)) + lnd_tunables = (struct lnet_ioctl_config_lnd_tunables *)conf->cfg_bulk; + + /* Create a net/ni structures for the network string */ + rc = lnet_parse_networks(&net_head, nets, use_tcp_bonding); + if (rc <= 0) + return rc == 0 ? -EINVAL : rc; - LNET_MUTEX_LOCK(&the_lnet.ln_api_mutex); + mutex_lock(&the_lnet.ln_api_mutex); if (rc > 1) { - rc = -EINVAL; /* only add one interface per call */ + rc = -EINVAL; /* only add one network per call */ goto failed0; } - rc = lnet_ping_info_setup(&pinfo, &md_handle, 1 + lnet_get_ni_count(), + net = list_entry(net_head.next, struct lnet_net, net_list); + + lnet_net_lock(LNET_LOCK_EX); + rnet = lnet_find_rnet_locked(net->net_id); + lnet_net_unlock(LNET_LOCK_EX); + /* make sure that the net added doesn't invalidate the current + * configuration LNet is keeping */ + if (rnet != NULL) { + CERROR("Adding net %s will invalidate routing configuration\n", + nets); + rc = -EUSERS; + goto failed0; + } + + /* + * make sure you calculate the correct number of slots in the ping + * info. Since the ping info is a flattened list of all the NIs, + * we should allocate enough slots to accomodate the number of NIs + * which will be added. + * + * We can use lnet_get_net_ni_count_locked() since the net is not + * on a public list yet, so locking is not a problem + */ + net_ni_count = lnet_get_net_ni_count_locked(net); + + rc = lnet_ping_info_setup(&pinfo, &md_handle, + net_ni_count + lnet_get_ni_count(), false); if (rc != 0) goto failed0; - rc = lnet_startup_lndnis(&net_head, peer_timeout, peer_cr, - peer_buf_cr, credits, NULL); - if (rc != 0) + list_del_init(&net->net_list); + + if (lnd_tunables) + memcpy(&net->net_tunables, + &lnd_tunables->lt_cmn, sizeof(lnd_tunables->lt_cmn)); + + /* + * before starting this network get a count of the current TCP + * networks which require the acceptor thread running. If that + * count is == 0 before we start up this network, then we'd want to + * start up the acceptor thread after starting up this network + */ + num_acceptor_nets = lnet_count_acceptor_nets(); + + /* + * lnd_startup_lndnet() can deallocate 'net' even if it it returns + * success, because we endded up adding interfaces to an existing + * network. So grab the net_type now + */ + net_type = LNET_NETTYP(net->net_id); + + rc = lnet_startup_lndnet(net, + (lnd_tunables) ? &lnd_tunables->lt_tun : NULL); + if (rc < 0) goto failed1; + /* + * Start the acceptor thread if this is the first network + * being added that requires the thread. + */ + if (net_type == SOCKLND && num_acceptor_nets == 0) + { + rc = lnet_acceptor_start(); + if (rc < 0) { + /* shutdown the net that we just started */ + CERROR("Failed to start up acceptor thread\n"); + /* + * Note that if we needed to start the acceptor + * thread, then 'net' must have been the first TCP + * network, therefore was unique, and therefore + * wasn't deallocated by lnet_startup_lndnet() + */ + lnet_shutdown_lndnet(net); + goto failed1; + } + } + lnet_ping_target_update(pinfo, md_handle); - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + mutex_unlock(&the_lnet.ln_api_mutex); return 0; @@ -2036,52 +2198,86 @@ failed1: lnet_ping_md_unlink(pinfo, &md_handle); lnet_ping_info_free(pinfo); failed0: - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + mutex_unlock(&the_lnet.ln_api_mutex); while (!list_empty(&net_head)) { - ni = list_entry(net_head.next, struct lnet_ni, ni_list); - list_del_init(&ni->ni_list); - lnet_ni_free(ni); + net = list_entry(net_head.next, struct lnet_net, net_list); + list_del_init(&net->net_list); + lnet_net_free(net); } return rc; } int -lnet_dyn_del_ni(__u32 net) +lnet_dyn_del_ni(__u32 net_id) { - int rc; + struct lnet_net *net; + struct lnet_ping_info *pinfo; + lnet_handle_md_t md_handle; + int rc; + int net_ni_count; + + /* don't allow userspace to shutdown the LOLND */ + if (LNET_NETTYP(net_id) == LOLND) + return -EINVAL; + + mutex_lock(&the_lnet.ln_api_mutex); + + lnet_net_lock(0); - LNET_MUTEX_LOCK(&the_lnet.ln_api_mutex); - rc = lnet_shutdown_lndni(net); - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + net = lnet_get_net_locked(net_id); + if (net == NULL) { + rc = -EINVAL; + goto out; + } + + net_ni_count = lnet_get_net_ni_count_locked(net); + + lnet_net_unlock(0); + + /* create and link a new ping info, before removing the old one */ + rc = lnet_ping_info_setup(&pinfo, &md_handle, + lnet_get_ni_count() - net_ni_count, false); + if (rc != 0) + goto out; + + lnet_shutdown_lndnet(net); + + if (lnet_count_acceptor_nets() == 0) + lnet_acceptor_stop(); + + lnet_ping_target_update(pinfo, md_handle); + +out: + mutex_unlock(&the_lnet.ln_api_mutex); return rc; } +void lnet_incr_dlc_seq(void) +{ + atomic_inc(&lnet_dlc_seq_no); +} + +__u32 lnet_get_dlc_seq_locked(void) +{ + return atomic_read(&lnet_dlc_seq_no); +} + /** - * This is an ugly hack to export IOC_LIBCFS_DEBUG_PEER and - * IOC_LIBCFS_PORTALS_COMPATIBILITY commands to users, by tweaking the LNet - * internal ioctl handler. + * LNet ioctl handler. * - * IOC_LIBCFS_PORTALS_COMPATIBILITY is now deprecated, don't use it. - * - * \param cmd IOC_LIBCFS_DEBUG_PEER to print debugging data about a peer. - * The data will be printed to system console. Don't use it excessively. - * \param arg A pointer to lnet_process_id_t, process ID of the peer. - * - * \return Always return 0 when called by users directly (i.e., not via ioctl). */ int LNetCtl(unsigned int cmd, void *arg) { struct libcfs_ioctl_data *data = arg; struct lnet_ioctl_config_data *config; - lnet_process_id_t id = {0}; - lnet_ni_t *ni; - int rc; + lnet_process_id_t id = {0}; + lnet_ni_t *ni; + int rc; - CLASSERT(LIBCFS_IOC_DATA_MAX >= sizeof(struct lnet_ioctl_net_config) + - sizeof(struct lnet_ioctl_config_data)); - LASSERT(the_lnet.ln_init); + BUILD_BUG_ON(sizeof(struct lnet_ioctl_net_config) + + sizeof(struct lnet_ioctl_config_data) > LIBCFS_IOC_DATA_MAX); switch (cmd) { case IOC_LIBCFS_GET_NI: @@ -2094,24 +2290,42 @@ LNetCtl(unsigned int cmd, void *arg) case IOC_LIBCFS_ADD_ROUTE: config = arg; - LNET_MUTEX_LOCK(&the_lnet.ln_api_mutex); + + if (config->cfg_hdr.ioc_len < sizeof(*config)) + return -EINVAL; + + mutex_lock(&the_lnet.ln_api_mutex); rc = lnet_add_route(config->cfg_net, config->cfg_config_u.cfg_route.rtr_hop, config->cfg_nid, config->cfg_config_u.cfg_route. rtr_priority); - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); - return (rc != 0) ? rc : lnet_check_routes(); + if (rc == 0) { + rc = lnet_check_routes(); + if (rc != 0) + lnet_del_route(config->cfg_net, + config->cfg_nid); + } + mutex_unlock(&the_lnet.ln_api_mutex); + return rc; case IOC_LIBCFS_DEL_ROUTE: config = arg; - LNET_MUTEX_LOCK(&the_lnet.ln_api_mutex); + + if (config->cfg_hdr.ioc_len < sizeof(*config)) + return -EINVAL; + + mutex_lock(&the_lnet.ln_api_mutex); rc = lnet_del_route(config->cfg_net, config->cfg_nid); - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + mutex_unlock(&the_lnet.ln_api_mutex); return rc; case IOC_LIBCFS_GET_ROUTE: config = arg; + + if (config->cfg_hdr.ioc_len < sizeof(*config)) + return -EINVAL; + return lnet_get_route(config->cfg_count, &config->cfg_net, &config->cfg_config_u.cfg_route.rtr_hop, @@ -2121,70 +2335,78 @@ LNetCtl(unsigned int cmd, void *arg) rtr_priority); case IOC_LIBCFS_GET_NET: { - struct lnet_ioctl_net_config *net_config; + size_t total = sizeof(*config) + + sizeof(struct lnet_ioctl_net_config); config = arg; - net_config = (struct lnet_ioctl_net_config *) - config->cfg_bulk; - if (config == NULL || net_config == NULL) - return -1; - - return lnet_get_net_config(config->cfg_count, - &config->cfg_ncpts, - &config->cfg_nid, - &config->cfg_config_u. - cfg_net.net_peer_timeout, - &config->cfg_config_u.cfg_net. - net_peer_tx_credits, - &config->cfg_config_u.cfg_net. - net_peer_rtr_credits, - &config->cfg_config_u.cfg_net. - net_max_tx_credits, - net_config); + + if (config->cfg_hdr.ioc_len < total) + return -EINVAL; + + return lnet_get_net_config(config); } case IOC_LIBCFS_GET_LNET_STATS: { struct lnet_ioctl_lnet_stats *lnet_stats = arg; + if (lnet_stats->st_hdr.ioc_len < sizeof(*lnet_stats)) + return -EINVAL; + lnet_counters_get(&lnet_stats->st_cntrs); return 0; } -#if defined(__KERNEL__) && defined(LNET_ROUTER) case IOC_LIBCFS_CONFIG_RTR: config = arg; - LNET_MUTEX_LOCK(&the_lnet.ln_api_mutex); + + if (config->cfg_hdr.ioc_len < sizeof(*config)) + return -EINVAL; + + mutex_lock(&the_lnet.ln_api_mutex); if (config->cfg_config_u.cfg_buffers.buf_enable) { rc = lnet_rtrpools_enable(); - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + mutex_unlock(&the_lnet.ln_api_mutex); return rc; } lnet_rtrpools_disable(); - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + mutex_unlock(&the_lnet.ln_api_mutex); return 0; case IOC_LIBCFS_ADD_BUF: config = arg; - LNET_MUTEX_LOCK(&the_lnet.ln_api_mutex); + + if (config->cfg_hdr.ioc_len < sizeof(*config)) + return -EINVAL; + + mutex_lock(&the_lnet.ln_api_mutex); rc = lnet_rtrpools_adjust(config->cfg_config_u.cfg_buffers. buf_tiny, config->cfg_config_u.cfg_buffers. buf_small, config->cfg_config_u.cfg_buffers. buf_large); - LNET_MUTEX_UNLOCK(&the_lnet.ln_api_mutex); + mutex_unlock(&the_lnet.ln_api_mutex); return rc; -#endif case IOC_LIBCFS_GET_BUF: { struct lnet_ioctl_pool_cfg *pool_cfg; + size_t total = sizeof(*config) + sizeof(*pool_cfg); + config = arg; + + if (config->cfg_hdr.ioc_len < total) + return -EINVAL; + pool_cfg = (struct lnet_ioctl_pool_cfg *)config->cfg_bulk; return lnet_get_rtr_pool_cfg(config->cfg_count, pool_cfg); } case IOC_LIBCFS_GET_PEER_INFO: { struct lnet_ioctl_peer *peer_info = arg; + + if (peer_info->pr_hdr.ioc_len < sizeof(*peer_info)) + return -EINVAL; + return lnet_get_peer_info( peer_info->pr_count, &peer_info->pr_nid, @@ -2198,15 +2420,15 @@ LNetCtl(unsigned int cmd, void *arg) &peer_info->pr_lnd_u.pr_peer_credits.cr_peer_tx_qnob); } - case IOC_LIBCFS_NOTIFY_ROUTER: - return lnet_notify(NULL, data->ioc_nid, data->ioc_flags, - cfs_time_current() - - cfs_time_seconds(cfs_time_current_sec() - - (time_t)data->ioc_u64[0])); + case IOC_LIBCFS_NOTIFY_ROUTER: { + unsigned long jiffies_passed; - case IOC_LIBCFS_PORTALS_COMPATIBILITY: - /* This can be removed once lustre stops calling it */ - return 0; + jiffies_passed = ktime_get_real_seconds() - data->ioc_u64[0]; + jiffies_passed = cfs_time_seconds(jiffies_passed); + + return lnet_notify(NULL, data->ioc_nid, data->ioc_flags, + jiffies - jiffies_passed); + } case IOC_LIBCFS_LNET_DIST: rc = LNetDist(data->ioc_nid, &data->ioc_nid, &data->ioc_u32[1]); @@ -2225,57 +2447,51 @@ LNetCtl(unsigned int cmd, void *arg) case IOC_LIBCFS_LNET_FAULT: return lnet_fault_ctl(data->ioc_flags, data); - case IOC_LIBCFS_PING: + case IOC_LIBCFS_PING: { + signed long timeout; + id.nid = data->ioc_nid; id.pid = data->ioc_u32[0]; - rc = lnet_ping(id, data->ioc_u32[1], /* timeout */ - (lnet_process_id_t __user *)data->ioc_pbuf1, - data->ioc_plen1/sizeof(lnet_process_id_t)); + + /* Don't block longer than 2 minutes */ + if (data->ioc_u32[1] > 120 * MSEC_PER_SEC) + return -EINVAL; + + /* If timestamp is negative then disable timeout */ + if ((s32)data->ioc_u32[1] < 0) + timeout = MAX_SCHEDULE_TIMEOUT; + else + timeout = msecs_to_jiffies(data->ioc_u32[1]); + + rc = lnet_ping(id, timeout, data->ioc_pbuf1, + data->ioc_plen1 / sizeof(lnet_process_id_t)); if (rc < 0) return rc; data->ioc_count = rc; return 0; - - case IOC_LIBCFS_DEBUG_PEER: { - /* CAVEAT EMPTOR: this one designed for calling directly; not - * via an ioctl */ - id = *((lnet_process_id_t *) arg); - - lnet_debug_peer(id.nid); - - ni = lnet_net2ni(LNET_NIDNET(id.nid)); - if (ni == NULL) { - CDEBUG(D_WARNING, "No NI for %s\n", libcfs_id2str(id)); - } else { - if (ni->ni_lnd->lnd_ctl == NULL) { - CDEBUG(D_WARNING, "No ctl for %s\n", - libcfs_id2str(id)); - } else { - (void)ni->ni_lnd->lnd_ctl(ni, cmd, arg); - } - - lnet_ni_decref(ni); - } - return 0; } - default: ni = lnet_net2ni(data->ioc_net); if (ni == NULL) return -EINVAL; - if (ni->ni_lnd->lnd_ctl == NULL) + if (ni->ni_net->net_lnd->lnd_ctl == NULL) rc = -EINVAL; else - rc = ni->ni_lnd->lnd_ctl(ni, cmd, arg); + rc = ni->ni_net->net_lnd->lnd_ctl(ni, cmd, arg); - lnet_ni_decref(ni); return rc; } /* not reached */ } EXPORT_SYMBOL(LNetCtl); +void LNetDebugPeer(lnet_process_id_t id) +{ + lnet_debug_peer(id.nid); +} +EXPORT_SYMBOL(LNetDebugPeer); + /** * Retrieve the lnet_process_id_t ID of LNet interface at \a index. Note that * all interfaces share a same PID, as requested by LNetNIInit(). @@ -2291,25 +2507,24 @@ int LNetGetId(unsigned int index, lnet_process_id_t *id) { struct lnet_ni *ni; - struct list_head *tmp; + struct lnet_net *net; int cpt; int rc = -ENOENT; - LASSERT(the_lnet.ln_init); LASSERT(the_lnet.ln_refcount > 0); cpt = lnet_net_lock_current(); - list_for_each(tmp, &the_lnet.ln_nis) { - if (index-- != 0) - continue; - - ni = list_entry(tmp, lnet_ni_t, ni_list); + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) { + if (index-- != 0) + continue; - id->nid = ni->ni_nid; - id->pid = the_lnet.ln_pid; - rc = 0; - break; + id->nid = ni->ni_nid; + id->pid = the_lnet.ln_pid; + rc = 0; + break; + } } lnet_net_unlock(cpt); @@ -2324,37 +2539,34 @@ EXPORT_SYMBOL(LNetGetId); void LNetSnprintHandle(char *str, int len, lnet_handle_any_t h) { - snprintf(str, len, LPX64, h.cookie); + snprintf(str, len, "%#llx", h.cookie); } EXPORT_SYMBOL(LNetSnprintHandle); -static int -lnet_ping(lnet_process_id_t id, int timeout_ms, lnet_process_id_t __user *ids, - int n_ids) +static int lnet_ping(lnet_process_id_t id, signed long timeout, + lnet_process_id_t __user *ids, int n_ids) { lnet_handle_eq_t eqh; lnet_handle_md_t mdh; - lnet_event_t event; - lnet_md_t md = {0}; - int which; - int unlinked = 0; - int replied = 0; - const int a_long_time = 60000; /* mS */ - int infosz; - lnet_ping_info_t *info; + lnet_event_t event; + lnet_md_t md = { NULL }; + int which; + int unlinked = 0; + int replied = 0; + const signed long a_long_time = msecs_to_jiffies(60 * MSEC_PER_SEC); + int infosz; + struct lnet_ping_info *info; lnet_process_id_t tmpid; - int i; - int nob; - int rc; - int rc2; - sigset_t blocked; - - infosz = offsetof(lnet_ping_info_t, pi_ni[n_ids]); - - if (n_ids <= 0 || - id.nid == LNET_NID_ANY || - timeout_ms > 500000 || /* arbitrary limit! */ - n_ids > 20) /* arbitrary limit! */ + int i; + int nob; + int rc; + int rc2; + sigset_t blocked; + + infosz = offsetof(struct lnet_ping_info, pi_ni[n_ids]); + + /* n_ids limit is arbitrary */ + if (n_ids <= 0 || n_ids > 20 || id.nid == LNET_NID_ANY) return -EINVAL; if (id.pid == LNET_PID_ANY) @@ -2398,7 +2610,7 @@ lnet_ping(lnet_process_id_t id, int timeout_ms, lnet_process_id_t __user *ids, /* NB must wait for the UNLINK event below... */ unlinked = 1; - timeout_ms = a_long_time; + timeout = a_long_time; } do { @@ -2406,7 +2618,7 @@ lnet_ping(lnet_process_id_t id, int timeout_ms, lnet_process_id_t __user *ids, if (unlinked) blocked = cfs_block_allsigs(); - rc2 = LNetEQPoll(&eqh, 1, timeout_ms, &event, &which); + rc2 = LNetEQPoll(&eqh, 1, timeout, &event, &which); if (unlinked) cfs_restore_sigs(blocked); @@ -2416,7 +2628,7 @@ lnet_ping(lnet_process_id_t id, int timeout_ms, lnet_process_id_t __user *ids, (rc2 <= 0) ? -1 : event.status, (rc2 > 0 && event.unlinked) ? " unlinked" : ""); - LASSERT(rc2 != -EOVERFLOW); /* can't miss anything */ + LASSERT(rc2 != -EOVERFLOW); /* can't miss anything */ if (rc2 <= 0 || event.status != 0) { /* timeout or error */ @@ -2430,7 +2642,7 @@ lnet_ping(lnet_process_id_t id, int timeout_ms, lnet_process_id_t __user *ids, LNetMDUnlink(mdh); /* No assertion (racing with network) */ unlinked = 1; - timeout_ms = a_long_time; + timeout = a_long_time; } else if (rc2 == 0) { /* timed out waiting for unlink */ CWARN("ping %s: late network completion\n", @@ -2454,7 +2666,7 @@ lnet_ping(lnet_process_id_t id, int timeout_ms, lnet_process_id_t __user *ids, nob = rc; LASSERT(nob >= 0 && nob <= infosz); - rc = -EPROTO; /* if I can't parse... */ + rc = -EPROTO; /* if I can't parse... */ if (nob < 8) { /* can't check magic/version */ @@ -2477,23 +2689,24 @@ lnet_ping(lnet_process_id_t id, int timeout_ms, lnet_process_id_t __user *ids, goto out_1; } - if (nob < offsetof(lnet_ping_info_t, pi_ni[0])) { + if (nob < offsetof(struct lnet_ping_info, pi_ni[0])) { CERROR("%s: Short reply %d(%d min)\n", libcfs_id2str(id), - nob, (int)offsetof(lnet_ping_info_t, pi_ni[0])); + nob, (int)offsetof(struct lnet_ping_info, pi_ni[0])); goto out_1; } if (info->pi_nnis < n_ids) n_ids = info->pi_nnis; - if (nob < offsetof(lnet_ping_info_t, pi_ni[n_ids])) { + if (nob < offsetof(struct lnet_ping_info, pi_ni[n_ids])) { CERROR("%s: Short reply %d(%d expected)\n", libcfs_id2str(id), - nob, (int)offsetof(lnet_ping_info_t, pi_ni[n_ids])); + nob, (int)offsetof(struct lnet_ping_info, pi_ni[n_ids])); goto out_1; } - rc = -EFAULT; /* If I SEGV... */ + rc = -EFAULT; /* If I SEGV... */ + memset(&tmpid, 0, sizeof(tmpid)); for (i = 0; i < n_ids; i++) { tmpid.pid = info->pi_pid; tmpid.nid = info->pi_ni[i].ns_nid;