X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lnet%2Flnet%2Fapi-ni.c;h=f86a31f3fc4323e62874e7e0996eaf0431e5d862;hp=53988ec2bab0137ccf376f5314c4565f0184f652;hb=47cc77462343533b4d706836e7e087f7a1844318;hpb=b34e754c1a0b0d03ceac8279aad2fd03c5be616e diff --git a/lnet/lnet/api-ni.c b/lnet/lnet/api-ni.c index 53988ec..f86a31f 100644 --- a/lnet/lnet/api-ni.c +++ b/lnet/lnet/api-ni.c @@ -27,7 +27,6 @@ */ /* * This file is part of Lustre, http://www.lustre.org/ - * Lustre is a trademark of Sun Microsystems, Inc. */ #define DEBUG_SUBSYSTEM S_LNET @@ -37,7 +36,10 @@ #include #include #include - +#ifdef HAVE_SCHED_HEADERS +#include +#endif +#include #include #define D_LNI D_CONSOLE @@ -68,10 +70,10 @@ static int rnet_htable_size = LNET_REMOTE_NETS_HASH_DEFAULT; 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; +static int use_tcp_bonding; 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"); + "use_tcp_bonding parameter has been removed"); unsigned int lnet_numa_range = 0; module_param(lnet_numa_range, uint, 0444); @@ -119,7 +121,12 @@ module_param_call(lnet_recovery_interval, recovery_interval_set, param_get_int, &lnet_recovery_interval, S_IRUGO|S_IWUSR); #endif MODULE_PARM_DESC(lnet_recovery_interval, - "Interval to recover unhealthy interfaces in seconds"); + "DEPRECATED - Interval to recover unhealthy interfaces in seconds"); + +unsigned int lnet_recovery_limit; +module_param(lnet_recovery_limit, uint, 0644); +MODULE_PARM_DESC(lnet_recovery_limit, + "How long to attempt recovery of unhealthy peer interfaces in seconds. Set to 0 to allow indefinite recovery"); static int lnet_interfaces_max = LNET_INTERFACES_MAX_DEFAULT; static int intf_max_set(const char *val, cfs_kernel_param_arg_t *kp); @@ -179,10 +186,8 @@ module_param_call(lnet_drop_asym_route, drop_asym_route_set, param_get_int, MODULE_PARM_DESC(lnet_drop_asym_route, "Set to 1 to drop asymmetrical route messages."); -#define LNET_TRANSACTION_TIMEOUT_NO_HEALTH_DEFAULT 50 -#define LNET_TRANSACTION_TIMEOUT_HEALTH_DEFAULT 10 - -unsigned lnet_transaction_timeout = LNET_TRANSACTION_TIMEOUT_HEALTH_DEFAULT; +#define LNET_TRANSACTION_TIMEOUT_DEFAULT 50 +unsigned int lnet_transaction_timeout = LNET_TRANSACTION_TIMEOUT_DEFAULT; static int transaction_to_set(const char *val, cfs_kernel_param_arg_t *kp); #ifdef HAVE_KERNEL_PARAM_OPS static struct kernel_param_ops param_ops_transaction_timeout = { @@ -200,8 +205,8 @@ module_param_call(lnet_transaction_timeout, transaction_to_set, param_get_int, MODULE_PARM_DESC(lnet_transaction_timeout, "Maximum number of seconds to wait for a peer response."); -#define LNET_RETRY_COUNT_HEALTH_DEFAULT 3 -unsigned lnet_retry_count = LNET_RETRY_COUNT_HEALTH_DEFAULT; +#define LNET_RETRY_COUNT_DEFAULT 2 +unsigned int lnet_retry_count = LNET_RETRY_COUNT_DEFAULT; static int retry_count_set(const char *val, cfs_kernel_param_arg_t *kp); #ifdef HAVE_KERNEL_PARAM_OPS static struct kernel_param_ops param_ops_retry_count = { @@ -219,8 +224,33 @@ module_param_call(lnet_retry_count, retry_count_set, param_get_int, MODULE_PARM_DESC(lnet_retry_count, "Maximum number of times to retry transmitting a message"); +unsigned int lnet_response_tracking = 3; +static int response_tracking_set(const char *val, cfs_kernel_param_arg_t *kp); + +#ifdef HAVE_KERNEL_PARAM_OPS +static struct kernel_param_ops param_ops_response_tracking = { + .set = response_tracking_set, + .get = param_get_int, +}; + +#define param_check_response_tracking(name, p) \ + __param_check(name, p, int) +module_param(lnet_response_tracking, response_tracking, 0644); +#else +module_param_call(lnet_response_tracking, response_tracking_set, param_get_int, + &lnet_response_tracking, 0644); +#endif +MODULE_PARM_DESC(lnet_response_tracking, + "(0|1|2|3) LNet Internal Only|GET Reply only|PUT ACK only|Full Tracking (default)"); -unsigned lnet_lnd_timeout = LNET_LND_DEFAULT_TIMEOUT; +#define LNET_LND_TIMEOUT_DEFAULT ((LNET_TRANSACTION_TIMEOUT_DEFAULT - 1) / \ + (LNET_RETRY_COUNT_DEFAULT + 1)) +unsigned int lnet_lnd_timeout = LNET_LND_TIMEOUT_DEFAULT; +static void lnet_set_lnd_timeout(void) +{ + lnet_lnd_timeout = (lnet_transaction_timeout - 1) / + (lnet_retry_count + 1); +} /* * This sequence number keeps track of how many times DLC was used to @@ -263,21 +293,9 @@ sensitivity_set(const char *val, cfs_kernel_param_arg_t *kp) return -EINVAL; } - /* - * if we're turning on health then use the health timeout - * defaults. - */ - if (*sensitivity == 0 && value != 0) { - lnet_transaction_timeout = LNET_TRANSACTION_TIMEOUT_HEALTH_DEFAULT; - lnet_retry_count = LNET_RETRY_COUNT_HEALTH_DEFAULT; - /* - * if we're turning off health then use the no health timeout - * default. - */ - } else if (*sensitivity != 0 && value == 0) { - lnet_transaction_timeout = - LNET_TRANSACTION_TIMEOUT_NO_HEALTH_DEFAULT; + if (*sensitivity != 0 && value == 0 && lnet_retry_count != 0) { lnet_retry_count = 0; + lnet_set_lnd_timeout(); } *sensitivity = value; @@ -290,30 +308,7 @@ sensitivity_set(const char *val, cfs_kernel_param_arg_t *kp) static int recovery_interval_set(const char *val, cfs_kernel_param_arg_t *kp) { - int rc; - unsigned *interval = (unsigned *)kp->arg; - unsigned long value; - - rc = kstrtoul(val, 0, &value); - if (rc) { - CERROR("Invalid module parameter value for 'lnet_recovery_interval'\n"); - return rc; - } - - if (value < 1) { - CERROR("lnet_recovery_interval must be at least 1 second\n"); - return -EINVAL; - } - - /* - * The purpose of locking the api_mutex here is to ensure that - * the correct value ends up stored properly. - */ - mutex_lock(&the_lnet.ln_api_mutex); - - *interval = value; - - mutex_unlock(&the_lnet.ln_api_mutex); + CWARN("'lnet_recovery_interval' has been deprecated\n"); return 0; } @@ -322,7 +317,7 @@ static int discovery_set(const char *val, cfs_kernel_param_arg_t *kp) { int rc; - unsigned *discovery = (unsigned *)kp->arg; + unsigned *discovery_off = (unsigned *)kp->arg; unsigned long value; struct lnet_ping_buffer *pbuf; @@ -340,14 +335,20 @@ discovery_set(const char *val, cfs_kernel_param_arg_t *kp) */ mutex_lock(&the_lnet.ln_api_mutex); - if (value == *discovery) { + if (value == *discovery_off) { mutex_unlock(&the_lnet.ln_api_mutex); return 0; } - *discovery = value; - + /* + * We still want to set the discovery value even when LNet is not + * running. This is the case when LNet is being loaded and we want + * the module parameters to take effect. Otherwise if we're + * changing the value dynamically, we want to set it after + * updating the peers + */ if (the_lnet.ln_state != LNET_STATE_RUNNING) { + *discovery_off = value; mutex_unlock(&the_lnet.ln_api_mutex); return 0; } @@ -361,7 +362,10 @@ discovery_set(const char *val, cfs_kernel_param_arg_t *kp) pbuf->pb_info.pi_features |= LNET_PING_FEAT_DISCOVERY; lnet_net_unlock(LNET_LOCK_EX); - lnet_push_update_to_peers(1); + /* only send a push when we're turning off discovery */ + if (*discovery_off <= 0 && value > 0) + lnet_push_update_to_peers(1); + *discovery_off = value; mutex_unlock(&the_lnet.ln_api_mutex); @@ -419,7 +423,7 @@ transaction_to_set(const char *val, cfs_kernel_param_arg_t *kp) */ mutex_lock(&the_lnet.ln_api_mutex); - if (value < lnet_retry_count || value == 0) { + if (value <= lnet_retry_count || value == 0) { mutex_unlock(&the_lnet.ln_api_mutex); CERROR("Invalid value for lnet_transaction_timeout (%lu). " "Has to be greater than lnet_retry_count (%u)\n", @@ -433,10 +437,10 @@ transaction_to_set(const char *val, cfs_kernel_param_arg_t *kp) } *transaction_to = value; - if (lnet_retry_count == 0) - lnet_lnd_timeout = value; - else - lnet_lnd_timeout = value / lnet_retry_count; + /* Update the lnet_lnd_timeout now that we've modified the + * transaction timeout + */ + lnet_set_lnd_timeout(); mutex_unlock(&the_lnet.ln_api_mutex); @@ -462,9 +466,9 @@ retry_count_set(const char *val, cfs_kernel_param_arg_t *kp) */ mutex_lock(&the_lnet.ln_api_mutex); - if (lnet_health_sensitivity == 0) { + if (lnet_health_sensitivity == 0 && value > 0) { mutex_unlock(&the_lnet.ln_api_mutex); - CERROR("Can not set retry_count when health feature is turned off\n"); + CERROR("Can not set lnet_retry_count when health feature is turned off\n"); return -EINVAL; } @@ -478,10 +482,10 @@ retry_count_set(const char *val, cfs_kernel_param_arg_t *kp) *retry_count = value; - if (value == 0) - lnet_lnd_timeout = lnet_transaction_timeout; - else - lnet_lnd_timeout = lnet_transaction_timeout / value; + /* Update the lnet_lnd_timeout now that we've modified the + * retry count + */ + lnet_set_lnd_timeout(); mutex_unlock(&the_lnet.ln_api_mutex); @@ -510,17 +514,40 @@ intf_max_set(const char *val, cfs_kernel_param_arg_t *kp) return 0; } -static char * +static int +response_tracking_set(const char *val, cfs_kernel_param_arg_t *kp) +{ + int rc; + unsigned long new_value; + + rc = kstrtoul(val, 0, &new_value); + if (rc) { + CERROR("Invalid value for 'lnet_response_tracking'\n"); + return -EINVAL; + } + + if (new_value < 0 || new_value > 3) { + CWARN("Invalid value (%lu) for 'lnet_response_tracking'\n", + new_value); + return -EINVAL; + } + + lnet_response_tracking = new_value; + + return 0; +} + +static const char * lnet_get_routes(void) { return routes; } -static char * +static const char * lnet_get_networks(void) { - char *nets; - int rc; + const char *nets; + int rc; if (*networks != 0 && *ip2nets != 0) { LCONSOLE_ERROR_MSG(0x101, "Please specify EITHER 'networks' or " @@ -544,22 +571,19 @@ lnet_init_locks(void) { spin_lock_init(&the_lnet.ln_eq_wait_lock); spin_lock_init(&the_lnet.ln_msg_resend_lock); - init_waitqueue_head(&the_lnet.ln_eq_waitq); - init_waitqueue_head(&the_lnet.ln_mt_waitq); + init_completion(&the_lnet.ln_mt_wait_complete); mutex_init(&the_lnet.ln_lnd_mutex); } -static void -lnet_fini_locks(void) -{ -} - struct kmem_cache *lnet_mes_cachep; /* MEs kmem_cache */ struct kmem_cache *lnet_small_mds_cachep; /* <= LNET_SMALL_MD_SIZE bytes * MDs kmem_cache */ +struct kmem_cache *lnet_udsp_cachep; /* udsp cache */ +struct kmem_cache *lnet_rspt_cachep; /* response tracker cache */ +struct kmem_cache *lnet_msg_cachep; static int -lnet_descriptor_setup(void) +lnet_slab_setup(void) { /* create specific kmem_cache for MEs and small MDs (i.e., originally * allocated in kmem_cache). @@ -575,12 +599,42 @@ lnet_descriptor_setup(void) if (!lnet_small_mds_cachep) return -ENOMEM; + lnet_udsp_cachep = kmem_cache_create("lnet_udsp", + sizeof(struct lnet_udsp), + 0, 0, NULL); + if (!lnet_udsp_cachep) + return -ENOMEM; + + lnet_rspt_cachep = kmem_cache_create("lnet_rspt", sizeof(struct lnet_rsp_tracker), + 0, 0, NULL); + if (!lnet_rspt_cachep) + return -ENOMEM; + + lnet_msg_cachep = kmem_cache_create("lnet_msg", sizeof(struct lnet_msg), + 0, 0, NULL); + if (!lnet_msg_cachep) + return -ENOMEM; + return 0; } static void -lnet_descriptor_cleanup(void) +lnet_slab_cleanup(void) { + if (lnet_msg_cachep) { + kmem_cache_destroy(lnet_msg_cachep); + lnet_msg_cachep = NULL; + } + + if (lnet_rspt_cachep) { + kmem_cache_destroy(lnet_rspt_cachep); + lnet_rspt_cachep = NULL; + } + + if (lnet_udsp_cachep) { + kmem_cache_destroy(lnet_udsp_cachep); + lnet_udsp_cachep = NULL; + } if (lnet_small_mds_cachep) { kmem_cache_destroy(lnet_small_mds_cachep); @@ -601,7 +655,7 @@ lnet_create_remote_nets_table(void) LASSERT(the_lnet.ln_remote_nets_hash == NULL); LASSERT(the_lnet.ln_remote_nets_hbits > 0); - LIBCFS_ALLOC(hash, LNET_REMOTE_NETS_HASH_SIZE * sizeof(*hash)); + CFS_ALLOC_PTR_ARRAY(hash, LNET_REMOTE_NETS_HASH_SIZE); if (hash == NULL) { CERROR("Failed to create remote nets hash table\n"); return -ENOMEM; @@ -624,9 +678,8 @@ lnet_destroy_remote_nets_table(void) for (i = 0; i < LNET_REMOTE_NETS_HASH_SIZE; i++) LASSERT(list_empty(&the_lnet.ln_remote_nets_hash[i])); - LIBCFS_FREE(the_lnet.ln_remote_nets_hash, - LNET_REMOTE_NETS_HASH_SIZE * - sizeof(the_lnet.ln_remote_nets_hash[0])); + CFS_FREE_PTR_ARRAY(the_lnet.ln_remote_nets_hash, + LNET_REMOTE_NETS_HASH_SIZE); the_lnet.ln_remote_nets_hash = NULL; } @@ -642,8 +695,6 @@ lnet_destroy_locks(void) cfs_percpt_lock_free(the_lnet.ln_net_lock); the_lnet.ln_net_lock = NULL; } - - lnet_fini_locks(); } static int @@ -671,144 +722,213 @@ 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) */ + * 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); + BUILD_BUG_ON(LNET_PROTO_TCP_MAGIC != 0xeebc0ded); + BUILD_BUG_ON(LNET_PROTO_TCP_VERSION_MAJOR != 1); + BUILD_BUG_ON(LNET_PROTO_TCP_VERSION_MINOR != 0); + BUILD_BUG_ON(LNET_MSG_ACK != 0); + BUILD_BUG_ON(LNET_MSG_PUT != 1); + BUILD_BUG_ON(LNET_MSG_GET != 2); + BUILD_BUG_ON(LNET_MSG_REPLY != 3); + BUILD_BUG_ON(LNET_MSG_HELLO != 4); + + BUILD_BUG_ON((int)sizeof(lnet_nid_t) != 8); + BUILD_BUG_ON((int)sizeof(lnet_pid_t) != 4); + + /* Checks for struct lnet_nid */ + BUILD_BUG_ON((int)sizeof(struct lnet_nid) != 20); + BUILD_BUG_ON((int)offsetof(struct lnet_nid, nid_size) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_nid *)0)->nid_size) != 1); + BUILD_BUG_ON((int)offsetof(struct lnet_nid, nid_type) != 1); + BUILD_BUG_ON((int)sizeof(((struct lnet_nid *)0)->nid_type) != 1); + BUILD_BUG_ON((int)offsetof(struct lnet_nid, nid_num) != 2); + BUILD_BUG_ON((int)sizeof(((struct lnet_nid *)0)->nid_num) != 2); + BUILD_BUG_ON((int)offsetof(struct lnet_nid, nid_addr) != 4); + BUILD_BUG_ON((int)sizeof(((struct lnet_nid *)0)->nid_addr) != 16); + + /* Checks for struct lnet_process_id_packed */ + BUILD_BUG_ON((int)sizeof(struct lnet_process_id_packed) != 12); + BUILD_BUG_ON((int)offsetof(struct lnet_process_id_packed, nid) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_process_id_packed *)0)->nid) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_process_id_packed, pid) != 8); + BUILD_BUG_ON((int)sizeof(((struct lnet_process_id_packed *)0)->pid) != 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); + BUILD_BUG_ON((int)sizeof(struct lnet_handle_wire) != 16); + BUILD_BUG_ON((int)offsetof(struct lnet_handle_wire, + wh_interface_cookie) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_handle_wire *)0)->wh_interface_cookie) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_handle_wire, + wh_object_cookie) != 8); + BUILD_BUG_ON((int)sizeof(((struct lnet_handle_wire *)0)->wh_object_cookie) != 8); /* Checks for struct struct lnet_magicversion */ - CLASSERT((int)sizeof(struct lnet_magicversion) == 8); - CLASSERT((int)offsetof(struct lnet_magicversion, magic) == 0); - CLASSERT((int)sizeof(((struct lnet_magicversion *)0)->magic) == 4); - CLASSERT((int)offsetof(struct lnet_magicversion, version_major) == 4); - CLASSERT((int)sizeof(((struct lnet_magicversion *)0)->version_major) == 2); - CLASSERT((int)offsetof(struct lnet_magicversion, version_minor) == 6); - CLASSERT((int)sizeof(((struct lnet_magicversion *)0)->version_minor) == 2); + BUILD_BUG_ON((int)sizeof(struct lnet_magicversion) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_magicversion, magic) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_magicversion *)0)->magic) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_magicversion, version_major) != 4); + BUILD_BUG_ON((int)sizeof(((struct lnet_magicversion *)0)->version_major) != 2); + BUILD_BUG_ON((int)offsetof(struct lnet_magicversion, + version_minor) != 6); + BUILD_BUG_ON((int)sizeof(((struct lnet_magicversion *)0)->version_minor) != 2); /* Checks for struct struct lnet_hdr */ - CLASSERT((int)sizeof(struct lnet_hdr) == 72); - CLASSERT((int)offsetof(struct lnet_hdr, dest_nid) == 0); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->dest_nid) == 8); - CLASSERT((int)offsetof(struct lnet_hdr, src_nid) == 8); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->src_nid) == 8); - CLASSERT((int)offsetof(struct lnet_hdr, dest_pid) == 16); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->dest_pid) == 4); - CLASSERT((int)offsetof(struct lnet_hdr, src_pid) == 20); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->src_pid) == 4); - CLASSERT((int)offsetof(struct lnet_hdr, type) == 24); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->type) == 4); - CLASSERT((int)offsetof(struct lnet_hdr, payload_length) == 28); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->payload_length) == 4); - CLASSERT((int)offsetof(struct lnet_hdr, msg) == 32); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg) == 40); + BUILD_BUG_ON((int)sizeof(struct lnet_hdr) != 72); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, dest_nid) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->dest_nid) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, src_nid) != 8); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->src_nid) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, dest_pid) != 16); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->dest_pid) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, src_pid) != 20); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->src_pid) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, type) != 24); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->type) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, payload_length) != 28); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->payload_length) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg) != 32); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg) != 40); /* Ack */ - CLASSERT((int)offsetof(struct lnet_hdr, msg.ack.dst_wmd) == 32); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.ack.dst_wmd) == 16); - CLASSERT((int)offsetof(struct lnet_hdr, msg.ack.match_bits) == 48); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.ack.match_bits) == 8); - CLASSERT((int)offsetof(struct lnet_hdr, msg.ack.mlength) == 56); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.ack.mlength) == 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.ack.dst_wmd) != 32); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.ack.dst_wmd) != 16); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.ack.match_bits) != 48); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.ack.match_bits) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.ack.mlength) != 56); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.ack.mlength) != 4); /* Put */ - CLASSERT((int)offsetof(struct lnet_hdr, msg.put.ack_wmd) == 32); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.put.ack_wmd) == 16); - CLASSERT((int)offsetof(struct lnet_hdr, msg.put.match_bits) == 48); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.put.match_bits) == 8); - CLASSERT((int)offsetof(struct lnet_hdr, msg.put.hdr_data) == 56); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.put.hdr_data) == 8); - CLASSERT((int)offsetof(struct lnet_hdr, msg.put.ptl_index) == 64); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.put.ptl_index) == 4); - CLASSERT((int)offsetof(struct lnet_hdr, msg.put.offset) == 68); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.put.offset) == 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.put.ack_wmd) != 32); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.put.ack_wmd) != 16); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.put.match_bits) != 48); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.put.match_bits) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.put.hdr_data) != 56); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.put.hdr_data) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.put.ptl_index) != 64); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.put.ptl_index) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.put.offset) != 68); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.put.offset) != 4); /* Get */ - CLASSERT((int)offsetof(struct lnet_hdr, msg.get.return_wmd) == 32); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.get.return_wmd) == 16); - CLASSERT((int)offsetof(struct lnet_hdr, msg.get.match_bits) == 48); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.get.match_bits) == 8); - CLASSERT((int)offsetof(struct lnet_hdr, msg.get.ptl_index) == 56); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.get.ptl_index) == 4); - CLASSERT((int)offsetof(struct lnet_hdr, msg.get.src_offset) == 60); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.get.src_offset) == 4); - CLASSERT((int)offsetof(struct lnet_hdr, msg.get.sink_length) == 64); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.get.sink_length) == 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.get.return_wmd) != 32); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.get.return_wmd) != 16); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.get.match_bits) != 48); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.get.match_bits) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.get.ptl_index) != 56); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.get.ptl_index) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.get.src_offset) != 60); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.get.src_offset) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.get.sink_length) != 64); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.get.sink_length) != 4); /* Reply */ - CLASSERT((int)offsetof(struct lnet_hdr, msg.reply.dst_wmd) == 32); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.reply.dst_wmd) == 16); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.reply.dst_wmd) != 32); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.reply.dst_wmd) != 16); /* Hello */ - CLASSERT((int)offsetof(struct lnet_hdr, msg.hello.incarnation) == 32); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.hello.incarnation) == 8); - CLASSERT((int)offsetof(struct lnet_hdr, msg.hello.type) == 40); - CLASSERT((int)sizeof(((struct lnet_hdr *)0)->msg.hello.type) == 4); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.hello.incarnation) != 32); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.hello.incarnation) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_hdr, msg.hello.type) != 40); + BUILD_BUG_ON((int)sizeof(((struct lnet_hdr *)0)->msg.hello.type) != 4); /* Checks for struct lnet_ni_status and related constants */ - CLASSERT(LNET_NI_STATUS_INVALID == 0x00000000); - CLASSERT(LNET_NI_STATUS_UP == 0x15aac0de); - CLASSERT(LNET_NI_STATUS_DOWN == 0xdeadface); + BUILD_BUG_ON(LNET_NI_STATUS_INVALID != 0x00000000); + BUILD_BUG_ON(LNET_NI_STATUS_UP != 0x15aac0de); + BUILD_BUG_ON(LNET_NI_STATUS_DOWN != 0xdeadface); /* Checks for struct lnet_ni_status */ - CLASSERT((int)sizeof(struct lnet_ni_status) == 16); - CLASSERT((int)offsetof(struct lnet_ni_status, ns_nid) == 0); - CLASSERT((int)sizeof(((struct lnet_ni_status *)0)->ns_nid) == 8); - CLASSERT((int)offsetof(struct lnet_ni_status, ns_status) == 8); - CLASSERT((int)sizeof(((struct lnet_ni_status *)0)->ns_status) == 4); - CLASSERT((int)offsetof(struct lnet_ni_status, ns_unused) == 12); - CLASSERT((int)sizeof(((struct lnet_ni_status *)0)->ns_unused) == 4); + BUILD_BUG_ON((int)sizeof(struct lnet_ni_status) != 16); + BUILD_BUG_ON((int)offsetof(struct lnet_ni_status, ns_nid) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_ni_status *)0)->ns_nid) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_ni_status, ns_status) != 8); + BUILD_BUG_ON((int)sizeof(((struct lnet_ni_status *)0)->ns_status) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_ni_status, ns_unused) != 12); + BUILD_BUG_ON((int)sizeof(((struct lnet_ni_status *)0)->ns_unused) != 4); /* Checks for struct lnet_ping_info and related constants */ - CLASSERT(LNET_PROTO_PING_MAGIC == 0x70696E67); - CLASSERT(LNET_PING_FEAT_INVAL == 0); - CLASSERT(LNET_PING_FEAT_BASE == 1); - CLASSERT(LNET_PING_FEAT_NI_STATUS == 2); - CLASSERT(LNET_PING_FEAT_RTE_DISABLED == 4); - CLASSERT(LNET_PING_FEAT_MULTI_RAIL == 8); - CLASSERT(LNET_PING_FEAT_DISCOVERY == 16); - CLASSERT(LNET_PING_FEAT_BITS == 31); + BUILD_BUG_ON(LNET_PROTO_PING_MAGIC != 0x70696E67); + BUILD_BUG_ON(LNET_PING_FEAT_INVAL != 0); + BUILD_BUG_ON(LNET_PING_FEAT_BASE != 1); + BUILD_BUG_ON(LNET_PING_FEAT_NI_STATUS != 2); + BUILD_BUG_ON(LNET_PING_FEAT_RTE_DISABLED != 4); + BUILD_BUG_ON(LNET_PING_FEAT_MULTI_RAIL != 8); + BUILD_BUG_ON(LNET_PING_FEAT_DISCOVERY != 16); + BUILD_BUG_ON(LNET_PING_FEAT_BITS != 31); /* Checks for struct lnet_ping_info */ - CLASSERT((int)sizeof(struct lnet_ping_info) == 16); - CLASSERT((int)offsetof(struct lnet_ping_info, pi_magic) == 0); - CLASSERT((int)sizeof(((struct lnet_ping_info *)0)->pi_magic) == 4); - CLASSERT((int)offsetof(struct lnet_ping_info, pi_features) == 4); - CLASSERT((int)sizeof(((struct lnet_ping_info *)0)->pi_features) == 4); - CLASSERT((int)offsetof(struct lnet_ping_info, pi_pid) == 8); - CLASSERT((int)sizeof(((struct lnet_ping_info *)0)->pi_pid) == 4); - CLASSERT((int)offsetof(struct lnet_ping_info, pi_nnis) == 12); - CLASSERT((int)sizeof(((struct lnet_ping_info *)0)->pi_nnis) == 4); - CLASSERT((int)offsetof(struct lnet_ping_info, pi_ni) == 16); - CLASSERT((int)sizeof(((struct lnet_ping_info *)0)->pi_ni) == 0); + BUILD_BUG_ON((int)sizeof(struct lnet_ping_info) != 16); + BUILD_BUG_ON((int)offsetof(struct lnet_ping_info, pi_magic) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_ping_info *)0)->pi_magic) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_ping_info, pi_features) != 4); + BUILD_BUG_ON((int)sizeof(((struct lnet_ping_info *)0)->pi_features) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_ping_info, pi_pid) != 8); + BUILD_BUG_ON((int)sizeof(((struct lnet_ping_info *)0)->pi_pid) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_ping_info, pi_nnis) != 12); + BUILD_BUG_ON((int)sizeof(((struct lnet_ping_info *)0)->pi_nnis) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_ping_info, pi_ni) != 16); + BUILD_BUG_ON((int)sizeof(((struct lnet_ping_info *)0)->pi_ni) != 0); + + /* Acceptor connection request */ + BUILD_BUG_ON(LNET_PROTO_ACCEPTOR_VERSION != 1); + + /* Checks for struct lnet_acceptor_connreq */ + BUILD_BUG_ON((int)sizeof(struct lnet_acceptor_connreq) != 16); + BUILD_BUG_ON((int)offsetof(struct lnet_acceptor_connreq, acr_magic) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_acceptor_connreq *)0)->acr_magic) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_acceptor_connreq, acr_version) != 4); + BUILD_BUG_ON((int)sizeof(((struct lnet_acceptor_connreq *)0)->acr_version) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_acceptor_connreq, acr_nid) != 8); + BUILD_BUG_ON((int)sizeof(((struct lnet_acceptor_connreq *)0)->acr_nid) != 8); + + /* Checks for struct lnet_acceptor_connreq_v2 */ + BUILD_BUG_ON((int)sizeof(struct lnet_acceptor_connreq_v2) != 28); + BUILD_BUG_ON((int)offsetof(struct lnet_acceptor_connreq_v2, acr_magic) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_acceptor_connreq_v2 *)0)->acr_magic) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_acceptor_connreq_v2, acr_version) != 4); + BUILD_BUG_ON((int)sizeof(((struct lnet_acceptor_connreq_v2 *)0)->acr_version) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_acceptor_connreq_v2, acr_nid) != 8); + BUILD_BUG_ON((int)sizeof(((struct lnet_acceptor_connreq_v2 *)0)->acr_nid) != 20); + + /* Checks for struct lnet_counters_common */ + BUILD_BUG_ON((int)sizeof(struct lnet_counters_common) != 60); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_msgs_alloc) != 0); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_msgs_alloc) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_msgs_max) != 4); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_msgs_max) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_errors) != 8); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_errors) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_send_count) != 12); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_send_count) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_recv_count) != 16); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_recv_count) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_route_count) != 20); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_route_count) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_drop_count) != 24); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_drop_count) != 4); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_send_length) != 28); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_send_length) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_recv_length) != 36); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_recv_length) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_route_length) != 44); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_route_length) != 8); + BUILD_BUG_ON((int)offsetof(struct lnet_counters_common, lcc_drop_length) != 52); + BUILD_BUG_ON((int)sizeof(((struct lnet_counters_common *)0)->lcc_drop_length) != 8); } -static struct lnet_lnd *lnet_find_lnd_by_type(__u32 type) +static const struct lnet_lnd *lnet_find_lnd_by_type(__u32 type) { - struct lnet_lnd *lnd; - struct list_head *tmp; + const struct lnet_lnd *lnd; /* holding lnd mutex */ - list_for_each(tmp, &the_lnet.ln_lnds) { - lnd = list_entry(tmp, struct lnet_lnd, lnd_list); + if (type >= NUM_LNDS) + return NULL; + lnd = the_lnet.ln_lnds[type]; + LASSERT(!lnd || lnd->lnd_type == type); - if (lnd->lnd_type == type) - return lnd; - } - return NULL; + return lnd; } unsigned int @@ -819,15 +939,14 @@ lnet_get_lnd_timeout(void) EXPORT_SYMBOL(lnet_get_lnd_timeout); void -lnet_register_lnd(struct lnet_lnd *lnd) +lnet_register_lnd(const struct lnet_lnd *lnd) { mutex_lock(&the_lnet.ln_lnd_mutex); LASSERT(libcfs_isknown_lnd(lnd->lnd_type)); LASSERT(lnet_find_lnd_by_type(lnd->lnd_type) == NULL); - list_add_tail(&lnd->lnd_list, &the_lnet.ln_lnds); - lnd->lnd_refcount = 0; + the_lnet.ln_lnds[lnd->lnd_type] = lnd; CDEBUG(D_NET, "%s LND registered\n", libcfs_lnd2str(lnd->lnd_type)); @@ -836,30 +955,30 @@ lnet_register_lnd(struct lnet_lnd *lnd) EXPORT_SYMBOL(lnet_register_lnd); void -lnet_unregister_lnd(struct lnet_lnd *lnd) +lnet_unregister_lnd(const struct lnet_lnd *lnd) { mutex_lock(&the_lnet.ln_lnd_mutex); LASSERT(lnet_find_lnd_by_type(lnd->lnd_type) == lnd); - LASSERT(lnd->lnd_refcount == 0); - list_del(&lnd->lnd_list); + the_lnet.ln_lnds[lnd->lnd_type] = NULL; CDEBUG(D_NET, "%s LND unregistered\n", libcfs_lnd2str(lnd->lnd_type)); mutex_unlock(&the_lnet.ln_lnd_mutex); } EXPORT_SYMBOL(lnet_unregister_lnd); -void -lnet_counters_get_common(struct lnet_counters_common *common) +static void +lnet_counters_get_common_locked(struct lnet_counters_common *common) { struct lnet_counters *ctr; int i; + /* FIXME !!! Their is no assert_lnet_net_locked() to ensure this + * actually called under the protection of the lnet_net_lock. + */ memset(common, 0, sizeof(*common)); - lnet_net_lock(LNET_LOCK_EX); - cfs_percpt_for_each(ctr, i, the_lnet.ln_counters) { common->lcc_msgs_max += ctr->lct_common.lcc_msgs_max; common->lcc_msgs_alloc += ctr->lct_common.lcc_msgs_alloc; @@ -873,23 +992,33 @@ lnet_counters_get_common(struct lnet_counters_common *common) common->lcc_route_length += ctr->lct_common.lcc_route_length; common->lcc_drop_length += ctr->lct_common.lcc_drop_length; } +} + +void +lnet_counters_get_common(struct lnet_counters_common *common) +{ + lnet_net_lock(LNET_LOCK_EX); + lnet_counters_get_common_locked(common); lnet_net_unlock(LNET_LOCK_EX); } EXPORT_SYMBOL(lnet_counters_get_common); -void +int lnet_counters_get(struct lnet_counters *counters) { struct lnet_counters *ctr; struct lnet_counters_health *health = &counters->lct_health; - int i; + int i, rc = 0; memset(counters, 0, sizeof(*counters)); - lnet_counters_get_common(&counters->lct_common); - lnet_net_lock(LNET_LOCK_EX); + if (the_lnet.ln_state != LNET_STATE_RUNNING) + GOTO(out_unlock, rc = -ENODEV); + + lnet_counters_get_common_locked(&counters->lct_common); + cfs_percpt_for_each(ctr, i, the_lnet.ln_counters) { health->lch_rst_alloc += ctr->lct_health.lch_rst_alloc; health->lch_resend_count += ctr->lct_health.lch_resend_count; @@ -916,7 +1045,9 @@ lnet_counters_get(struct lnet_counters *counters) health->lch_network_timeout_count += ctr->lct_health.lch_network_timeout_count; } +out_unlock: lnet_net_unlock(LNET_LOCK_EX); + return rc; } EXPORT_SYMBOL(lnet_counters_get); @@ -928,9 +1059,12 @@ lnet_counters_reset(void) lnet_net_lock(LNET_LOCK_EX); + if (the_lnet.ln_state != LNET_STATE_RUNNING) + goto avoid_reset; + cfs_percpt_for_each(counters, i, the_lnet.ln_counters) memset(counters, 0, sizeof(struct lnet_counters)); - +avoid_reset: lnet_net_unlock(LNET_LOCK_EX); } @@ -961,10 +1095,7 @@ lnet_res_container_cleanup(struct lnet_res_container *rec) struct list_head *e = rec->rec_active.next; list_del_init(e); - if (rec->rec_type == LNET_COOKIE_TYPE_EQ) { - lnet_eq_free(list_entry(e, struct lnet_eq, eq_list)); - - } else if (rec->rec_type == LNET_COOKIE_TYPE_MD) { + if (rec->rec_type == LNET_COOKIE_TYPE_MD) { lnet_md_free(list_entry(e, struct lnet_libmd, md_list)); } else { /* NB: Active MEs should be attached on portals */ @@ -982,8 +1113,7 @@ lnet_res_container_cleanup(struct lnet_res_container *rec) } if (rec->rec_lh_hash != NULL) { - LIBCFS_FREE(rec->rec_lh_hash, - LNET_LH_HASH_SIZE * sizeof(rec->rec_lh_hash[0])); + CFS_FREE_PTR_ARRAY(rec->rec_lh_hash, LNET_LH_HASH_SIZE); rec->rec_lh_hash = NULL; } @@ -1099,6 +1229,26 @@ lnet_res_lh_initialize(struct lnet_res_container *rec, list_add(&lh->lh_hash_chain, &rec->rec_lh_hash[hash]); } +struct list_head ** +lnet_create_array_of_queues(void) +{ + struct list_head **qs; + struct list_head *q; + int i; + + qs = cfs_percpt_alloc(lnet_cpt_table(), + sizeof(struct list_head)); + if (!qs) { + CERROR("Failed to allocate queues\n"); + return NULL; + } + + cfs_percpt_for_each(q, i, qs) + INIT_LIST_HEAD(q); + + return qs; +} + static int lnet_unprepare(void); static int @@ -1131,10 +1281,12 @@ lnet_prepare(lnet_pid_t requested_pid) INIT_LIST_HEAD(&the_lnet.ln_dc_expired); INIT_LIST_HEAD(&the_lnet.ln_mt_localNIRecovq); INIT_LIST_HEAD(&the_lnet.ln_mt_peerNIRecovq); + INIT_LIST_HEAD(&the_lnet.ln_udsp_list); init_waitqueue_head(&the_lnet.ln_dc_waitq); - LNetInvalidateEQHandle(&the_lnet.ln_mt_eqh); + the_lnet.ln_mt_handler = NULL; + init_completion(&the_lnet.ln_started); - rc = lnet_descriptor_setup(); + rc = lnet_slab_setup(); if (rc != 0) goto failed; @@ -1169,14 +1321,6 @@ lnet_prepare(lnet_pid_t requested_pid) if (rc != 0) goto failed; - 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); if (recs == NULL) { rc = -ENOMEM; @@ -1191,6 +1335,12 @@ lnet_prepare(lnet_pid_t requested_pid) goto failed; } + the_lnet.ln_mt_zombie_rstqs = lnet_create_array_of_queues(); + if (!the_lnet.ln_mt_zombie_rstqs) { + rc = -ENOMEM; + goto failed; + } + return 0; failed: @@ -1201,8 +1351,6 @@ lnet_prepare(lnet_pid_t requested_pid) static int lnet_unprepare (void) { - int rc; - /* NB no LNET_LOCK since this is the last reference. All LND instances * have shut down already, so it is safe to unlink and free all * descriptors, even those that appear committed to a network op (eg MD @@ -1214,12 +1362,14 @@ lnet_unprepare (void) LASSERT(list_empty(&the_lnet.ln_test_peers)); LASSERT(list_empty(&the_lnet.ln_nets)); - if (!LNetEQHandleIsInvalid(the_lnet.ln_mt_eqh)) { - rc = LNetEQFree(the_lnet.ln_mt_eqh); - LNetInvalidateEQHandle(&the_lnet.ln_mt_eqh); - LASSERT(rc == 0); + if (the_lnet.ln_mt_zombie_rstqs) { + lnet_clean_zombie_rstqs(); + the_lnet.ln_mt_zombie_rstqs = NULL; } + lnet_assert_handler_unused(the_lnet.ln_mt_handler); + the_lnet.ln_mt_handler = NULL; + lnet_portals_destroy(); if (the_lnet.ln_md_containers != NULL) { @@ -1227,11 +1377,6 @@ lnet_unprepare (void) the_lnet.ln_md_containers = NULL; } - if (the_lnet.ln_me_containers != NULL) { - lnet_res_containers_destroy(the_lnet.ln_me_containers); - the_lnet.ln_me_containers = NULL; - } - lnet_res_container_cleanup(&the_lnet.ln_eq_container); lnet_msg_containers_destroy(); @@ -1243,7 +1388,8 @@ lnet_unprepare (void) the_lnet.ln_counters = NULL; } lnet_destroy_remote_nets_table(); - lnet_descriptor_cleanup(); + lnet_udsp_destroy(true); + lnet_slab_cleanup(); return 0; } @@ -1295,8 +1441,83 @@ lnet_get_net_locked(__u32 net_id) return NULL; } -unsigned int -lnet_nid_cpt_hash(lnet_nid_t nid, unsigned int number) +void +lnet_net_clr_pref_rtrs(struct lnet_net *net) +{ + struct list_head zombies; + struct lnet_nid_list *ne; + struct lnet_nid_list *tmp; + + INIT_LIST_HEAD(&zombies); + + lnet_net_lock(LNET_LOCK_EX); + list_splice_init(&net->net_rtr_pref_nids, &zombies); + lnet_net_unlock(LNET_LOCK_EX); + + list_for_each_entry_safe(ne, tmp, &zombies, nl_list) { + list_del_init(&ne->nl_list); + LIBCFS_FREE(ne, sizeof(*ne)); + } +} + +int +lnet_net_add_pref_rtr(struct lnet_net *net, + struct lnet_nid *gw_nid) +__must_hold(&the_lnet.ln_api_mutex) +{ + struct lnet_nid_list *ne; + + /* This function is called with api_mutex held. When the api_mutex + * is held the list can not be modified, as it is only modified as + * a result of applying a UDSP and that happens under api_mutex + * lock. + */ + list_for_each_entry(ne, &net->net_rtr_pref_nids, nl_list) { + if (nid_same(&ne->nl_nid, gw_nid)) + return -EEXIST; + } + + LIBCFS_ALLOC(ne, sizeof(*ne)); + if (!ne) + return -ENOMEM; + + ne->nl_nid = *gw_nid; + + /* Lock the cpt to protect against addition and checks in the + * selection algorithm + */ + lnet_net_lock(LNET_LOCK_EX); + list_add(&ne->nl_list, &net->net_rtr_pref_nids); + lnet_net_unlock(LNET_LOCK_EX); + + return 0; +} + +bool +lnet_net_is_pref_rtr_locked(struct lnet_net *net, struct lnet_nid *rtr_nid) +{ + struct lnet_nid_list *ne; + + CDEBUG(D_NET, "%s: rtr pref empty: %d\n", + libcfs_net2str(net->net_id), + list_empty(&net->net_rtr_pref_nids)); + + if (list_empty(&net->net_rtr_pref_nids)) + return false; + + list_for_each_entry(ne, &net->net_rtr_pref_nids, nl_list) { + CDEBUG(D_NET, "Comparing pref %s with gw %s\n", + libcfs_nidstr(&ne->nl_nid), + libcfs_nidstr(rtr_nid)); + if (nid_same(rtr_nid, &ne->nl_nid)) + return true; + } + + return false; +} + +static unsigned int +lnet_nid4_cpt_hash(lnet_nid_t nid, unsigned int number) { __u64 key = nid; unsigned int val; @@ -1314,8 +1535,31 @@ lnet_nid_cpt_hash(lnet_nid_t nid, unsigned int number) return (unsigned int)(key + val + (val >> 1)) % number; } +unsigned int +lnet_nid_cpt_hash(struct lnet_nid *nid, unsigned int number) +{ + unsigned int val; + u32 h = 0; + int i; + + LASSERT(number >= 1 && number <= LNET_CPT_NUMBER); + + if (number == 1) + return 0; + + if (nid_is_nid4(nid)) + return lnet_nid4_cpt_hash(lnet_nid_to_nid4(nid), number); + + for (i = 0; i < 4; i++) + h = hash_32(nid->nid_addr[i]^h, 32); + val = hash_32(LNET_NID_NET(nid) ^ h, LNET_CPT_BITS); + if (val < number) + return val; + return (unsigned int)(h + val + (val >> 1)) % number; +} + int -lnet_cpt_of_nid_locked(lnet_nid_t nid, struct lnet_ni *ni) +lnet_cpt_of_nid_locked(struct lnet_nid *nid, struct lnet_ni *ni) { struct lnet_net *net; @@ -1339,7 +1583,7 @@ lnet_cpt_of_nid_locked(lnet_nid_t nid, struct lnet_ni *ni) } /* no NI provided so look at the net */ - net = lnet_get_net_locked(LNET_NIDNET(nid)); + net = lnet_get_net_locked(LNET_NID_NET(nid)); if (net != NULL && net->net_cpts != NULL) { return net->net_cpts[lnet_nid_cpt_hash(nid, net->net_ncpts)]; @@ -1349,7 +1593,7 @@ lnet_cpt_of_nid_locked(lnet_nid_t nid, struct lnet_ni *ni) } int -lnet_cpt_of_nid(lnet_nid_t nid, struct lnet_ni *ni) +lnet_nid2cpt(struct lnet_nid *nid, struct lnet_ni *ni) { int cpt; int cpt2; @@ -1365,37 +1609,60 @@ lnet_cpt_of_nid(lnet_nid_t nid, struct lnet_ni *ni) return cpt2; } +EXPORT_SYMBOL(lnet_nid2cpt); + +int +lnet_cpt_of_nid(lnet_nid_t nid4, struct lnet_ni *ni) +{ + struct lnet_nid nid; + + if (LNET_CPT_NUMBER == 1) + return 0; /* the only one */ + + lnet_nid4_to_nid(nid4, &nid); + return lnet_nid2cpt(&nid, ni); +} EXPORT_SYMBOL(lnet_cpt_of_nid); int -lnet_islocalnet(__u32 net_id) +lnet_islocalnet_locked(__u32 net_id) { struct lnet_net *net; - int cpt; - bool local; - - cpt = lnet_net_lock_current(); + bool local; net = lnet_get_net_locked(net_id); local = net != NULL; + return local; +} + +int +lnet_islocalnet(__u32 net_id) +{ + int cpt; + bool local; + + cpt = lnet_net_lock_current(); + + local = lnet_islocalnet_locked(net_id); + lnet_net_unlock(cpt); return local; } struct lnet_ni * -lnet_nid2ni_locked(lnet_nid_t nid, int cpt) +lnet_nid_to_ni_locked(struct lnet_nid *nid, int cpt) { struct lnet_net *net; - struct lnet_ni *ni; + struct lnet_ni *ni; LASSERT(cpt != LNET_LOCK_EX); 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) + if (nid_same(&ni->ni_nid, nid)) return ni; } } @@ -1403,13 +1670,25 @@ lnet_nid2ni_locked(lnet_nid_t nid, int cpt) return NULL; } +struct lnet_ni * +lnet_nid2ni_locked(lnet_nid_t nid4, int cpt) +{ + struct lnet_nid nid; + + lnet_nid4_to_nid(nid4, &nid); + return lnet_nid_to_ni_locked(&nid, cpt); +} + struct lnet_ni * -lnet_nid2ni_addref(lnet_nid_t nid) +lnet_nid2ni_addref(lnet_nid_t nid4) { struct lnet_ni *ni; + struct lnet_nid nid; + + lnet_nid4_to_nid(nid4, &nid); lnet_net_lock(0); - ni = lnet_nid2ni_locked(nid, 0); + ni = lnet_nid_to_ni_locked(&nid, 0); if (ni) lnet_ni_addref_locked(ni, 0); lnet_net_unlock(0); @@ -1418,8 +1697,23 @@ lnet_nid2ni_addref(lnet_nid_t nid) } EXPORT_SYMBOL(lnet_nid2ni_addref); +struct lnet_ni * +lnet_nid_to_ni_addref(struct lnet_nid *nid) +{ + struct lnet_ni *ni; + + lnet_net_lock(0); + ni = lnet_nid_to_ni_locked(nid, 0); + if (ni) + lnet_ni_addref_locked(ni, 0); + lnet_net_unlock(0); + + return ni; +} +EXPORT_SYMBOL(lnet_nid_to_ni_addref); + int -lnet_islocalnid(lnet_nid_t nid) +lnet_islocalnid4(lnet_nid_t nid) { struct lnet_ni *ni; int cpt; @@ -1432,6 +1726,19 @@ lnet_islocalnid(lnet_nid_t nid) } int +lnet_islocalnid(struct lnet_nid *nid) +{ + struct lnet_ni *ni; + int cpt; + + cpt = lnet_net_lock_current(); + ni = lnet_nid_to_ni_locked(nid, cpt); + lnet_net_unlock(cpt); + + return ni != NULL; +} + +int lnet_count_acceptor_nets(void) { /* Return the # of NIs that need the acceptor. */ @@ -1460,6 +1767,7 @@ lnet_ping_buffer_alloc(int nnis, gfp_t gfp) LIBCFS_ALLOC_GFP(pbuf, LNET_PING_BUFFER_SIZE(nnis), gfp); if (pbuf) { pbuf->pb_nnis = nnis; + pbuf->pb_needs_post = false; atomic_set(&pbuf->pb_refcnt, 1); } @@ -1469,7 +1777,7 @@ lnet_ping_buffer_alloc(int nnis, gfp_t gfp) void lnet_ping_buffer_free(struct lnet_ping_buffer *pbuf) { - LASSERT(lnet_ping_buffer_numref(pbuf) == 0); + LASSERT(atomic_read(&pbuf->pb_refcnt) == 0); LIBCFS_FREE(pbuf, LNET_PING_BUFFER_SIZE(pbuf->pb_nnis)); } @@ -1555,7 +1863,6 @@ lnet_swap_pinginfo(struct lnet_ping_buffer *pbuf) __swab64s(&stat->ns_nid); __swab32s(&stat->ns_status); } - return; } int @@ -1570,7 +1877,7 @@ lnet_ping_info_validate(struct lnet_ping_info *pinfo) /* Loopback is guaranteed to be present */ if (pinfo->pi_nnis < 1 || pinfo->pi_nnis > lnet_interfaces_max) return -ERANGE; - if (LNET_NETTYP(LNET_NIDNET(LNET_PING_INFO_LONI(pinfo))) != LOLND) + if (LNET_PING_INFO_LONI(pinfo) != LNET_NID_LO_0) return -EPROTO; return 0; } @@ -1600,7 +1907,7 @@ lnet_ping_target_destroy(void) static void lnet_ping_target_event_handler(struct lnet_event *event) { - struct lnet_ping_buffer *pbuf = event->md.user_ptr; + struct lnet_ping_buffer *pbuf = event->md_user_ptr; if (event->unlinked) lnet_ping_buffer_decref(pbuf); @@ -1615,18 +1922,13 @@ lnet_ping_target_setup(struct lnet_ping_buffer **ppbuf, .nid = LNET_NID_ANY, .pid = LNET_PID_ANY }; - struct lnet_handle_me me_handle; + struct lnet_me *me; struct lnet_md md = { NULL }; - int rc, rc2; + int rc; - if (set_eq) { - rc = LNetEQAlloc(0, lnet_ping_target_event_handler, - &the_lnet.ln_ping_target_eq); - if (rc != 0) { - CERROR("Can't allocate ping buffer EQ: %d\n", rc); - return rc; - } - } + if (set_eq) + the_lnet.ln_ping_target_handler = + lnet_ping_target_event_handler; *ppbuf = lnet_ping_target_create(ni_count); if (*ppbuf == NULL) { @@ -1635,11 +1937,11 @@ lnet_ping_target_setup(struct lnet_ping_buffer **ppbuf, } /* Ping target ME/MD */ - rc = LNetMEAttach(LNET_RESERVED_PORTAL, id, + me = LNetMEAttach(LNET_RESERVED_PORTAL, id, LNET_PROTO_PING_MATCHBITS, 0, - LNET_UNLINK, LNET_INS_AFTER, - &me_handle); - if (rc != 0) { + LNET_UNLINK, LNET_INS_AFTER); + if (IS_ERR(me)) { + rc = PTR_ERR(me); CERROR("Can't create ping target ME: %d\n", rc); goto fail_decref_ping_buffer; } @@ -1651,30 +1953,23 @@ lnet_ping_target_setup(struct lnet_ping_buffer **ppbuf, md.max_size = 0; md.options = LNET_MD_OP_GET | LNET_MD_TRUNCATE | LNET_MD_MANAGE_REMOTE; - md.eq_handle = the_lnet.ln_ping_target_eq; + md.handler = the_lnet.ln_ping_target_handler; md.user_ptr = *ppbuf; - rc = LNetMDAttach(me_handle, md, LNET_RETAIN, ping_mdh); + rc = LNetMDAttach(me, &md, LNET_RETAIN, ping_mdh); if (rc != 0) { CERROR("Can't attach ping target MD: %d\n", rc); - goto fail_unlink_ping_me; + goto fail_decref_ping_buffer; } lnet_ping_buffer_addref(*ppbuf); return 0; -fail_unlink_ping_me: - rc2 = LNetMEUnlink(me_handle); - LASSERT(rc2 == 0); fail_decref_ping_buffer: - LASSERT(lnet_ping_buffer_numref(*ppbuf) == 1); + LASSERT(atomic_read(&(*ppbuf)->pb_refcnt) == 1); lnet_ping_buffer_decref(*ppbuf); *ppbuf = NULL; fail_free_eq: - if (set_eq) { - rc2 = LNetEQFree(the_lnet.ln_ping_target_eq); - LASSERT(rc2 == 0); - } return rc; } @@ -1682,19 +1977,13 @@ static void lnet_ping_md_unlink(struct lnet_ping_buffer *pbuf, struct lnet_handle_md *ping_mdh) { - sigset_t blocked = cfs_block_allsigs(); - LNetMDUnlink(*ping_mdh); LNetInvalidateMDHandle(ping_mdh); /* NB the MD could be busy; this just starts the unlink */ - while (lnet_ping_buffer_numref(pbuf) > 1) { - CDEBUG(D_NET, "Still waiting for ping data MD to unlink\n"); - set_current_state(TASK_UNINTERRUPTIBLE); - schedule_timeout(cfs_time_seconds(1)); - } - - cfs_restore_sigs(blocked); + wait_var_event_warning(&pbuf->pb_refcnt, + atomic_read(&pbuf->pb_refcnt) <= 1, + "Still waiting for ping data MD to unlink\n"); } static void @@ -1713,12 +2002,12 @@ lnet_ping_target_install_locked(struct lnet_ping_buffer *pbuf) ns = &pbuf->pb_info.pi_ni[i]; - ns->ns_nid = ni->ni_nid; + if (!nid_is_nid4(&ni->ni_nid)) + continue; + ns->ns_nid = lnet_nid_to_nid4(&ni->ni_nid); lnet_ni_lock(ni); - ns->ns_status = (ni->ni_status != NULL) ? - ni->ni_status->ns_status : - LNET_NI_STATUS_UP; + ns->ns_status = lnet_ni_get_status_locked(ni); ni->ni_status = ns; lnet_ni_unlock(ni); @@ -1781,66 +2070,46 @@ lnet_ping_target_update(struct lnet_ping_buffer *pbuf, static void lnet_ping_target_fini(void) { - int rc; - lnet_ping_md_unlink(the_lnet.ln_ping_target, &the_lnet.ln_ping_target_md); - rc = LNetEQFree(the_lnet.ln_ping_target_eq); - LASSERT(rc == 0); - + lnet_assert_handler_unused(the_lnet.ln_ping_target_handler); lnet_ping_target_destroy(); } /* Resize the push target. */ int lnet_push_target_resize(void) { - struct lnet_process_id id = { LNET_NID_ANY, LNET_PID_ANY }; - struct lnet_md md = { NULL }; - struct lnet_handle_me meh; struct lnet_handle_md mdh; struct lnet_handle_md old_mdh; struct lnet_ping_buffer *pbuf; struct lnet_ping_buffer *old_pbuf; - int nnis = the_lnet.ln_push_target_nnis; + int nnis; int rc; +again: + nnis = the_lnet.ln_push_target_nnis; if (nnis <= 0) { - rc = -EINVAL; - goto fail_return; + CDEBUG(D_NET, "Invalid nnis %d\n", nnis); + return -EINVAL; } -again: + + /* NB: lnet_ping_buffer_alloc() sets pbuf refcount to 1. That ref is + * dropped when we need to resize again (see "old_pbuf" below) or when + * LNet is shutdown (see lnet_push_target_fini()) + */ pbuf = lnet_ping_buffer_alloc(nnis, GFP_NOFS); if (!pbuf) { - rc = -ENOMEM; - goto fail_return; - } - - rc = LNetMEAttach(LNET_RESERVED_PORTAL, id, - LNET_PROTO_PING_MATCHBITS, 0, - LNET_UNLINK, LNET_INS_AFTER, - &meh); - if (rc) { - CERROR("Can't create push target ME: %d\n", rc); - goto fail_decref_pbuf; + CDEBUG(D_NET, "Can't allocate pbuf for nnis %d\n", nnis); + return -ENOMEM; } - /* initialize md content */ - md.start = &pbuf->pb_info; - md.length = LNET_PING_INFO_SIZE(nnis); - md.threshold = LNET_MD_THRESH_INF; - md.max_size = 0; - md.options = LNET_MD_OP_PUT | LNET_MD_TRUNCATE | - LNET_MD_MANAGE_REMOTE; - md.user_ptr = pbuf; - md.eq_handle = the_lnet.ln_push_target_eq; - - rc = LNetMDAttach(meh, md, LNET_RETAIN, &mdh); + rc = lnet_push_target_post(pbuf, &mdh); if (rc) { - CERROR("Can't attach push MD: %d\n", rc); - goto fail_unlink_meh; + CDEBUG(D_NET, "Failed to post push target: %d\n", rc); + lnet_ping_buffer_decref(pbuf); + return rc; } - lnet_ping_buffer_addref(pbuf); lnet_net_lock(LNET_LOCK_EX); old_pbuf = the_lnet.ln_push_target; @@ -1851,34 +2120,81 @@ again: if (old_pbuf) { LNetMDUnlink(old_mdh); + /* Drop ref set by lnet_ping_buffer_alloc() */ lnet_ping_buffer_decref(old_pbuf); } + /* Received another push or reply that requires a larger buffer */ if (nnis < the_lnet.ln_push_target_nnis) goto again; CDEBUG(D_NET, "nnis %d success\n", nnis); - return 0; +} -fail_unlink_meh: - LNetMEUnlink(meh); -fail_decref_pbuf: - lnet_ping_buffer_decref(pbuf); -fail_return: - CDEBUG(D_NET, "nnis %d error %d\n", nnis, rc); - return rc; +int lnet_push_target_post(struct lnet_ping_buffer *pbuf, + struct lnet_handle_md *mdhp) +{ + struct lnet_process_id id = { LNET_NID_ANY, LNET_PID_ANY }; + struct lnet_md md = { NULL }; + struct lnet_me *me; + int rc; + + me = LNetMEAttach(LNET_RESERVED_PORTAL, id, + LNET_PROTO_PING_MATCHBITS, 0, + LNET_UNLINK, LNET_INS_AFTER); + if (IS_ERR(me)) { + rc = PTR_ERR(me); + CERROR("Can't create push target ME: %d\n", rc); + return rc; + } + + pbuf->pb_needs_post = false; + + /* This reference is dropped by lnet_push_target_event_handler() */ + lnet_ping_buffer_addref(pbuf); + + /* initialize md content */ + md.start = &pbuf->pb_info; + md.length = LNET_PING_INFO_SIZE(pbuf->pb_nnis); + md.threshold = 1; + md.max_size = 0; + md.options = LNET_MD_OP_PUT | LNET_MD_TRUNCATE; + md.user_ptr = pbuf; + md.handler = the_lnet.ln_push_target_handler; + + rc = LNetMDAttach(me, &md, LNET_UNLINK, mdhp); + if (rc) { + CERROR("Can't attach push MD: %d\n", rc); + lnet_ping_buffer_decref(pbuf); + pbuf->pb_needs_post = true; + return rc; + } + + CDEBUG(D_NET, "posted push target %p\n", pbuf); + + return 0; } static void lnet_push_target_event_handler(struct lnet_event *ev) { - struct lnet_ping_buffer *pbuf = ev->md.user_ptr; + struct lnet_ping_buffer *pbuf = ev->md_user_ptr; + + CDEBUG(D_NET, "type %d status %d unlinked %d\n", ev->type, ev->status, + ev->unlinked); if (pbuf->pb_info.pi_magic == __swab32(LNET_PROTO_PING_MAGIC)) lnet_swap_pinginfo(pbuf); + if (ev->type == LNET_EVENT_UNLINK) { + /* Drop ref added by lnet_push_target_post() */ + lnet_ping_buffer_decref(pbuf); + return; + } + lnet_peer_push_event(ev); if (ev->unlinked) + /* Drop ref added by lnet_push_target_post */ lnet_ping_buffer_decref(pbuf); } @@ -1890,12 +2206,11 @@ static int lnet_push_target_init(void) if (the_lnet.ln_push_target) return -EALREADY; - rc = LNetEQAlloc(0, lnet_push_target_event_handler, - &the_lnet.ln_push_target_eq); - if (rc) { - CERROR("Can't allocated push target EQ: %d\n", rc); - return rc; - } + the_lnet.ln_push_target_handler = + lnet_push_target_event_handler; + + rc = LNetSetLazyPortal(LNET_RESERVED_PORTAL); + LASSERT(rc == 0); /* Start at the required minimum, we'll enlarge if required. */ the_lnet.ln_push_target_nnis = LNET_INTERFACES_MIN; @@ -1903,8 +2218,8 @@ static int lnet_push_target_init(void) rc = lnet_push_target_resize(); if (rc) { - LNetEQFree(the_lnet.ln_push_target_eq); - LNetInvalidateEQHandle(&the_lnet.ln_push_target_eq); + LNetClearLazyPortal(LNET_RESERVED_PORTAL); + the_lnet.ln_push_target_handler = NULL; } return rc; @@ -1921,18 +2236,18 @@ static void lnet_push_target_fini(void) LNetInvalidateMDHandle(&the_lnet.ln_push_target_md); /* Wait for the unlink to complete. */ - while (lnet_ping_buffer_numref(the_lnet.ln_push_target) > 1) { - CDEBUG(D_NET, "Still waiting for ping data MD to unlink\n"); - set_current_state(TASK_UNINTERRUPTIBLE); - schedule_timeout(cfs_time_seconds(1)); - } + wait_var_event_warning(&the_lnet.ln_push_target->pb_refcnt, + atomic_read(&the_lnet.ln_push_target->pb_refcnt) <= 1, + "Still waiting for ping data MD to unlink\n"); + /* Drop ref set by lnet_ping_buffer_alloc() */ lnet_ping_buffer_decref(the_lnet.ln_push_target); the_lnet.ln_push_target = NULL; the_lnet.ln_push_target_nnis = 0; - LNetEQFree(the_lnet.ln_push_target_eq); - LNetInvalidateEQHandle(&the_lnet.ln_push_target_eq); + LNetClearLazyPortal(LNET_RESERVED_PORTAL); + lnet_assert_handler_unused(the_lnet.ln_push_target_handler); + the_lnet.ln_push_target_handler = NULL; } static int @@ -1993,15 +2308,22 @@ lnet_clear_zombies_nis_locked(struct lnet_net *net) } if (!list_empty(&ni->ni_netlist)) { + /* Unlock mutex while waiting to allow other + * threads to read the LNet state and fall through + * to avoid deadlock + */ lnet_net_unlock(LNET_LOCK_EX); + mutex_unlock(&the_lnet.ln_api_mutex); + ++i; if ((i & (-i)) == i) { CDEBUG(D_WARNING, "Waiting for zombie LNI %s\n", - libcfs_nid2str(ni->ni_nid)); + libcfs_nidstr(&ni->ni_nid)); } - set_current_state(TASK_UNINTERRUPTIBLE); - schedule_timeout(cfs_time_seconds(1)); + schedule_timeout_uninterruptible(cfs_time_seconds(1)); + + mutex_lock(&the_lnet.ln_api_mutex); lnet_net_lock(LNET_LOCK_EX); continue; } @@ -2011,11 +2333,18 @@ lnet_clear_zombies_nis_locked(struct lnet_net *net) islo = ni->ni_net->net_lnd->lnd_type == LOLND; LASSERT(!in_interrupt()); + /* Holding the mutex makes it safe for lnd_shutdown + * to call module_put(). Module unload cannot finish + * until lnet_unregister_lnd() completes, and that + * requires the mutex. + */ + mutex_lock(&the_lnet.ln_lnd_mutex); (net->net_lnd->lnd_shutdown)(ni); + mutex_unlock(&the_lnet.ln_lnd_mutex); if (!islo) CDEBUG(D_LNI, "Removed LNI %s\n", - libcfs_nid2str(ni->ni_nid)); + libcfs_nidstr(&ni->ni_nid)); lnet_ni_free(ni); i = 2; @@ -2054,8 +2383,6 @@ lnet_shutdown_lndnet(struct lnet_net *net) 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)) { @@ -2071,15 +2398,6 @@ lnet_shutdown_lndnet(struct lnet_net *net) /* Do peer table cleanup for this net */ lnet_peer_tables_cleanup(net); - 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); } @@ -2087,11 +2405,9 @@ static void lnet_shutdown_lndnets(void) { struct lnet_net *net; - struct list_head resend; + LIST_HEAD(resend); struct lnet_msg *msg, *tmp; - INIT_LIST_HEAD(&resend); - /* NB called holding the global mutex */ /* All quiet on the API front */ @@ -2101,16 +2417,12 @@ lnet_shutdown_lndnets(void) lnet_net_lock(LNET_LOCK_EX); the_lnet.ln_state = LNET_STATE_STOPPING; - 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); - } + /* + * 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 + */ + list_splice_init(&the_lnet.ln_nets, &the_lnet.ln_net_zombie); /* Drop the cached loopback Net. */ if (the_lnet.ln_loni != NULL) { @@ -2163,9 +2475,6 @@ lnet_startup_lndni(struct lnet_ni *ni, struct lnet_lnd_tunables *tun) 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); goto failed0; } @@ -2209,7 +2518,7 @@ lnet_startup_lndni(struct lnet_ni *ni, struct lnet_lnd_tunables *tun) atomic_set(&ni->ni_healthv, LNET_MAX_HEALTH_VALUE); CDEBUG(D_LNI, "Added LNI %s [%d/%d/%d/%d]\n", - libcfs_nid2str(ni->ni_nid), + libcfs_nidstr(&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, @@ -2225,21 +2534,19 @@ static int lnet_startup_lndnet(struct lnet_net *net, struct lnet_lnd_tunables *tun) { struct lnet_ni *ni; - struct lnet_net *net_l = NULL; - struct list_head local_ni_list; - int rc; - int ni_count = 0; - __u32 lnd_type; - struct lnet_lnd *lnd; - int peer_timeout = + struct lnet_net *net_l = NULL; + LIST_HEAD(local_ni_list); + int rc; + int ni_count = 0; + __u32 lnd_type; + const struct lnet_lnd *lnd; + int peer_timeout = net->net_tunables.lct_peer_timeout; - int maxtxcredits = + int maxtxcredits = net->net_tunables.lct_max_tx_credits; - int peerrtrcredits = + 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 @@ -2274,10 +2581,6 @@ lnet_startup_lndnet(struct lnet_net *net, struct lnet_lnd_tunables *tun) } } - lnet_net_lock(LNET_LOCK_EX); - lnd->lnd_refcount++; - lnet_net_unlock(LNET_LOCK_EX); - net->net_lnd = lnd; mutex_unlock(&the_lnet.ln_lnd_mutex); @@ -2306,19 +2609,6 @@ lnet_startup_lndnet(struct lnet_net *net, struct lnet_lnd_tunables *tun) * 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; - } - while (!list_empty(&net->net_ni_added)) { ni = list_entry(net->net_ni_added.next, struct lnet_ni, ni_netlist); @@ -2327,8 +2617,8 @@ lnet_startup_lndnet(struct lnet_net *net, struct lnet_lnd_tunables *tun) /* 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; + ni->ni_interface)) { + rc = -EEXIST; goto failed1; } @@ -2338,9 +2628,6 @@ lnet_startup_lndnet(struct lnet_net *net, struct lnet_lnd_tunables *tun) 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; @@ -2365,7 +2652,6 @@ lnet_startup_lndnet(struct lnet_net *net, struct lnet_lnd_tunables *tun) */ lnet_net_free(net); } else { - net->net_state = LNET_NET_STATE_ACTIVE; /* * restore tunables after it has been overwitten by the * lnd @@ -2437,6 +2723,85 @@ failed: return rc; } +static int lnet_genl_parse_list(struct sk_buff *msg, + const struct ln_key_list *data[], u16 idx) +{ + const struct ln_key_list *list = data[idx]; + const struct ln_key_props *props; + struct nlattr *node; + u16 count; + + if (!list) + return 0; + + if (!list->lkl_maxattr) + return -ERANGE; + + props = list->lkl_list; + if (!props) + return -EINVAL; + + node = nla_nest_start(msg, LN_SCALAR_ATTR_LIST); + if (!node) + return -ENOBUFS; + + for (count = 1; count <= list->lkl_maxattr; count++) { + struct nlattr *key = nla_nest_start(msg, count); + + if (count == 1) + nla_put_u16(msg, LN_SCALAR_ATTR_LIST_SIZE, + list->lkl_maxattr); + + nla_put_u16(msg, LN_SCALAR_ATTR_INDEX, count); + if (props[count].lkp_values) + nla_put_string(msg, LN_SCALAR_ATTR_VALUE, + props[count].lkp_values); + if (props[count].lkp_key_format) + nla_put_u16(msg, LN_SCALAR_ATTR_KEY_FORMAT, + props[count].lkp_key_format); + nla_put_u16(msg, LN_SCALAR_ATTR_NLA_TYPE, + props[count].lkp_data_type); + if (props[count].lkp_data_type == NLA_NESTED) { + int rc; + + rc = lnet_genl_parse_list(msg, data, ++idx); + if (rc < 0) + return rc; + } + + nla_nest_end(msg, key); + } + + nla_nest_end(msg, node); + return 0; +} + +int lnet_genl_send_scalar_list(struct sk_buff *msg, u32 portid, u32 seq, + const struct genl_family *family, int flags, + u8 cmd, const struct ln_key_list *data[]) +{ + int rc = 0; + void *hdr; + + if (!data[0]) + return -EINVAL; + + hdr = genlmsg_put(msg, portid, seq, family, flags, cmd); + if (!hdr) + GOTO(canceled, rc = -EMSGSIZE); + + rc = lnet_genl_parse_list(msg, data, 0); + if (rc < 0) + GOTO(canceled, rc); + + genlmsg_end(msg, hdr); +canceled: + if (rc < 0) + genlmsg_cancel(msg, hdr); + return rc; +} +EXPORT_SYMBOL(lnet_genl_send_scalar_list); + /** * Initialize LNet library. * @@ -2454,8 +2819,8 @@ int lnet_lib_init(void) lnet_assert_wire_constants(); /* refer to global cfs_cpt_table for now */ - the_lnet.ln_cpt_table = cfs_cpt_table; - the_lnet.ln_cpt_number = cfs_cpt_number(cfs_cpt_table); + the_lnet.ln_cpt_table = cfs_cpt_tab; + the_lnet.ln_cpt_number = cfs_cpt_number(cfs_cpt_tab); LASSERT(the_lnet.ln_cpt_number > 0); if (the_lnet.ln_cpt_number > LNET_CPT_MAX) { @@ -2476,7 +2841,6 @@ int lnet_lib_init(void) } the_lnet.ln_refcount = 0; - INIT_LIST_HEAD(&the_lnet.ln_lnds); INIT_LIST_HEAD(&the_lnet.ln_net_zombie); INIT_LIST_HEAD(&the_lnet.ln_msg_resend); @@ -2502,14 +2866,18 @@ int lnet_lib_init(void) * * \pre lnet_lib_init() called with success. * \pre All LNet users called LNetNIFini() for matching LNetNIInit() calls. + * + * As this happens at module-unload, all lnds must already be unloaded, + * so they must already be unregistered. */ void lnet_lib_exit(void) { - LASSERT(the_lnet.ln_refcount == 0); + int i; - while (!list_empty(&the_lnet.ln_lnds)) - lnet_unregister_lnd(list_entry(the_lnet.ln_lnds.next, - struct lnet_lnd, lnd_list)); + LASSERT(the_lnet.ln_refcount == 0); + lnet_unregister_lnd(&the_lolnd); + for (i = 0; i < NUM_LNDS; i++) + LASSERT(!the_lnet.ln_lnds[i]); lnet_destroy_locks(); } @@ -2536,11 +2904,9 @@ LNetNIInit(lnet_pid_t requested_pid) int ni_count; struct lnet_ping_buffer *pbuf; struct lnet_handle_md ping_mdh; - struct list_head net_head; + LIST_HEAD(net_head); struct lnet_net *net; - INIT_LIST_HEAD(&net_head); - mutex_lock(&the_lnet.ln_api_mutex); CDEBUG(D_OTHER, "refs %d\n", the_lnet.ln_refcount); @@ -2570,6 +2936,9 @@ LNetNIInit(lnet_pid_t requested_pid) goto err_empty_list; } + if (use_tcp_bonding) + CWARN("use_tcp_bonding has been removed. Use Multi-Rail and Dynamic Discovery instead, see LU-13641\n"); + /* 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 @@ -2577,8 +2946,7 @@ LNetNIInit(lnet_pid_t requested_pid) * 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); + rc = lnet_parse_networks(&net_head, lnet_get_networks()); if (rc < 0) goto err_empty_list; } @@ -2612,35 +2980,36 @@ LNetNIInit(lnet_pid_t requested_pid) lnet_ping_target_update(pbuf, ping_mdh); - rc = LNetEQAlloc(0, lnet_mt_event_handler, &the_lnet.ln_mt_eqh); - if (rc != 0) { - CERROR("Can't allocate monitor thread EQ: %d\n", rc); - goto err_stop_ping; - } - - rc = lnet_monitor_thr_start(); - if (rc != 0) - goto err_stop_ping; + the_lnet.ln_mt_handler = lnet_mt_event_handler; rc = lnet_push_target_init(); if (rc != 0) - goto err_stop_monitor_thr; + goto err_stop_ping; rc = lnet_peer_discovery_start(); if (rc != 0) goto err_destroy_push_target; + rc = lnet_monitor_thr_start(); + if (rc != 0) + goto err_stop_discovery_thr; + lnet_fault_init(); lnet_router_debugfs_init(); mutex_unlock(&the_lnet.ln_api_mutex); + complete_all(&the_lnet.ln_started); + + /* wait for all routers to start */ + lnet_wait_router_start(); + return 0; +err_stop_discovery_thr: + lnet_peer_discovery_stop(); err_destroy_push_target: lnet_push_target_fini(); -err_stop_monitor_thr: - lnet_monitor_thr_stop(); err_stop_ping: lnet_ping_target_fini(); err_acceptor_stop: @@ -2676,7 +3045,7 @@ EXPORT_SYMBOL(LNetNIInit); * \return always 0 for current implementation. */ int -LNetNIFini() +LNetNIFini(void) { mutex_lock(&the_lnet.ln_api_mutex); @@ -2690,9 +3059,9 @@ LNetNIFini() lnet_fault_fini(); lnet_router_debugfs_fini(); + lnet_monitor_thr_stop(); lnet_peer_discovery_stop(); lnet_push_target_fini(); - lnet_monitor_thr_stop(); lnet_ping_target_fini(); /* Teardown fns that use my own API functions BEFORE here */ @@ -2726,25 +3095,17 @@ lnet_fill_ni_info(struct lnet_ni *ni, struct lnet_ioctl_config_ni *cfg_ni, size_t min_size = 0; int i; - if (!ni || !cfg_ni || !tun) + if (!ni || !cfg_ni || !tun || !nid_is_nid4(&ni->ni_nid)) return; - if (ni->ni_interfaces[0] != NULL) { - for (i = 0; i < ARRAY_SIZE(ni->ni_interfaces); i++) { - if (ni->ni_interfaces[i] != NULL) { - strncpy(cfg_ni->lic_ni_intf[i], - ni->ni_interfaces[i], - sizeof(cfg_ni->lic_ni_intf[i])); - } - } + if (ni->ni_interface != NULL) { + strncpy(cfg_ni->lic_ni_intf, + ni->ni_interface, + sizeof(cfg_ni->lic_ni_intf)); } - cfg_ni->lic_nid = ni->ni_nid; - if (LNET_NETTYP(LNET_NIDNET(ni->ni_nid)) == LOLND) - cfg_ni->lic_status = LNET_NI_STATUS_UP; - else - cfg_ni->lic_status = ni->ni_status->ns_status; - cfg_ni->lic_tcp_bonding = use_tcp_bonding; + cfg_ni->lic_nid = lnet_nid_to_nid4(&ni->ni_nid); + cfg_ni->lic_status = lnet_ni_get_status_locked(ni); cfg_ni->lic_dev_cpt = ni->ni_dev_cpt; memcpy(&tun->lt_cmn, &ni->ni_net->net_tunables, sizeof(tun->lt_cmn)); @@ -2803,26 +3164,21 @@ lnet_fill_ni_info_legacy(struct lnet_ni *ni, size_t min_size, tunable_size = 0; int i; - if (!ni || !config) + if (!ni || !config || !nid_is_nid4(&ni->ni_nid)) return; net_config = (struct lnet_ioctl_net_config *) config->cfg_bulk; if (!net_config) return; - BUILD_BUG_ON(ARRAY_SIZE(ni->ni_interfaces) != - ARRAY_SIZE(net_config->ni_interfaces)); - - for (i = 0; i < ARRAY_SIZE(ni->ni_interfaces); i++) { - if (!ni->ni_interfaces[i]) - break; + if (!ni->ni_interface) + return; - strncpy(net_config->ni_interfaces[i], - ni->ni_interfaces[i], - sizeof(net_config->ni_interfaces[i])); - } + strncpy(net_config->ni_interface, + ni->ni_interface, + sizeof(net_config->ni_interface)); - config->cfg_nid = ni->ni_nid; + config->cfg_nid = lnet_nid_to_nid4(&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 = @@ -2832,10 +3188,7 @@ lnet_fill_ni_info_legacy(struct lnet_ni *ni, config->cfg_config_u.cfg_net.net_peer_rtr_credits = ni->ni_net->net_tunables.lct_peer_rtr_credits; - if (LNET_NETTYP(LNET_NIDNET(ni->ni_nid)) == LOLND) - net_config->ni_status = LNET_NI_STATUS_UP; - else - net_config->ni_status = ni->ni_status->ns_status; + net_config->ni_status = lnet_ni_get_status_locked(ni); if (ni->ni_cpts) { int num_cpts = min(ni->ni_ncpts, LNET_MAX_SHOW_NUM_CPT); @@ -2887,6 +3240,22 @@ lnet_get_ni_idx_locked(int idx) return NULL; } +int lnet_get_net_healthv_locked(struct lnet_net *net) +{ + struct lnet_ni *ni; + int best_healthv = 0; + int healthv, ni_fatal; + + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) { + healthv = atomic_read(&ni->ni_healthv); + ni_fatal = atomic_read(&ni->ni_fatal_error_on); + if (!ni_fatal && healthv > best_healthv) + best_healthv = healthv; + } + + return best_healthv; +} + struct lnet_ni * lnet_get_next_ni_locked(struct lnet_net *mynet, struct lnet_ni *prev) { @@ -3020,13 +3389,13 @@ int lnet_get_ni_stats(struct lnet_ioctl_element_msg_stats *msg_stats) static int lnet_add_net_common(struct lnet_net *net, struct lnet_ioctl_config_lnd_tunables *tun) { - __u32 net_id; + struct lnet_handle_md ping_mdh; struct lnet_ping_buffer *pbuf; - struct lnet_handle_md ping_mdh; - int rc; struct lnet_remotenet *rnet; - int net_ni_count; - int num_acceptor_nets; + struct lnet_ni *ni; + int net_ni_count; + __u32 net_id; + int rc; lnet_net_lock(LNET_LOCK_EX); rnet = lnet_find_rnet_locked(net->net_id); @@ -3067,14 +3436,6 @@ static int lnet_add_net_common(struct lnet_net *net, else memset(&net->net_tunables, -1, sizeof(net->net_tunables)); - /* - * 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(); - net_id = net->net_id; rc = lnet_startup_lndnet(net, @@ -3084,15 +3445,30 @@ static int lnet_add_net_common(struct lnet_net *net, lnet_net_lock(LNET_LOCK_EX); net = lnet_get_net_locked(net_id); - lnet_net_unlock(LNET_LOCK_EX); - LASSERT(net); + /* apply the UDSPs */ + rc = lnet_udsp_apply_policies_on_net(net); + if (rc) + CERROR("Failed to apply UDSPs on local net %s\n", + libcfs_net2str(net->net_id)); + + /* At this point we lost track of which NI was just added, so we + * just re-apply the policies on all of the NIs on this net + */ + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) { + rc = lnet_udsp_apply_policies_on_ni(ni); + if (rc) + CERROR("Failed to apply UDSPs on ni %s\n", + libcfs_nidstr(&ni->ni_nid)); + } + lnet_net_unlock(LNET_LOCK_EX); + /* * Start the acceptor thread if this is the first network * being added that requires the thread. */ - if (net->net_lnd->lnd_accept && num_acceptor_nets == 0) { + if (net->net_lnd->lnd_accept) { rc = lnet_acceptor_start(); if (rc < 0) { /* shutdown the net that we just started */ @@ -3116,24 +3492,37 @@ failed: return rc; } +static void +lnet_set_tune_defaults(struct lnet_ioctl_config_lnd_tunables *tun) +{ + if (tun) { + if (!tun->lt_cmn.lct_peer_timeout) + tun->lt_cmn.lct_peer_timeout = DEFAULT_PEER_TIMEOUT; + if (!tun->lt_cmn.lct_peer_tx_credits) + tun->lt_cmn.lct_peer_tx_credits = DEFAULT_PEER_CREDITS; + if (!tun->lt_cmn.lct_max_tx_credits) + tun->lt_cmn.lct_max_tx_credits = DEFAULT_CREDITS; + } +} + static int lnet_handle_legacy_ip2nets(char *ip2nets, struct lnet_ioctl_config_lnd_tunables *tun) { struct lnet_net *net; - char *nets; + const char *nets; int rc; - struct list_head net_head; - - INIT_LIST_HEAD(&net_head); + LIST_HEAD(net_head); rc = lnet_parse_ip2nets(&nets, ip2nets); if (rc < 0) return rc; - rc = lnet_parse_networks(&net_head, nets, use_tcp_bonding); + rc = lnet_parse_networks(&net_head, nets); if (rc < 0) return rc; + lnet_set_tune_defaults(tun); + mutex_lock(&the_lnet.ln_api_mutex); while (!list_empty(&net_head)) { net = list_entry(net_head.next, struct lnet_net, net_list); @@ -3191,10 +3580,12 @@ int lnet_dyn_add_ni(struct lnet_ioctl_config_ni *conf) } ni = lnet_ni_alloc_w_cpt_array(net, conf->lic_cpts, conf->lic_ncpts, - conf->lic_ni_intf[0]); + conf->lic_ni_intf); if (!ni) return -ENOMEM; + lnet_set_tune_defaults(tun); + mutex_lock(&the_lnet.ln_api_mutex); rc = lnet_add_net_common(net, tun); @@ -3247,8 +3638,7 @@ int lnet_dyn_del_ni(struct lnet_ioctl_config_ni *conf) lnet_shutdown_lndnet(net); - if (lnet_count_acceptor_nets() == 0) - lnet_acceptor_stop(); + lnet_acceptor_stop(); lnet_ping_target_update(pbuf, ping_mdh); @@ -3275,8 +3665,7 @@ int lnet_dyn_del_ni(struct lnet_ioctl_config_ni *conf) lnet_shutdown_lndni(ni); - if (lnet_count_acceptor_nets() == 0) - lnet_acceptor_stop(); + lnet_acceptor_stop(); lnet_ping_target_update(pbuf, ping_mdh); @@ -3303,16 +3692,14 @@ unlock_api_mutex: int lnet_dyn_add_net(struct lnet_ioctl_config_data *conf) { - struct lnet_net *net; - struct list_head net_head; - int rc; + struct lnet_net *net; + LIST_HEAD(net_head); + int rc; struct lnet_ioctl_config_lnd_tunables tun; - char *nets = conf->cfg_config_u.cfg_net.net_intf; - - INIT_LIST_HEAD(&net_head); + const char *nets = conf->cfg_config_u.cfg_net.net_intf; /* Create a net/ni structures for the network string */ - rc = lnet_parse_networks(&net_head, nets, use_tcp_bonding); + rc = lnet_parse_networks(&net_head, nets); if (rc <= 0) return rc == 0 ? -EINVAL : rc; @@ -3331,13 +3718,16 @@ lnet_dyn_add_net(struct lnet_ioctl_config_data *conf) memset(&tun, 0, sizeof(tun)); tun.lt_cmn.lct_peer_timeout = - conf->cfg_config_u.cfg_net.net_peer_timeout; + (!conf->cfg_config_u.cfg_net.net_peer_timeout) ? DEFAULT_PEER_TIMEOUT : + conf->cfg_config_u.cfg_net.net_peer_timeout; tun.lt_cmn.lct_peer_tx_credits = - conf->cfg_config_u.cfg_net.net_peer_tx_credits; + (!conf->cfg_config_u.cfg_net.net_peer_tx_credits) ? DEFAULT_PEER_CREDITS : + conf->cfg_config_u.cfg_net.net_peer_tx_credits; tun.lt_cmn.lct_peer_rtr_credits = conf->cfg_config_u.cfg_net.net_peer_rtr_credits; tun.lt_cmn.lct_max_tx_credits = - conf->cfg_config_u.cfg_net.net_max_tx_credits; + (!conf->cfg_config_u.cfg_net.net_max_tx_credits) ? DEFAULT_CREDITS : + conf->cfg_config_u.cfg_net.net_max_tx_credits; rc = lnet_add_net_common(net, &tun); @@ -3388,8 +3778,7 @@ lnet_dyn_del_net(__u32 net_id) lnet_shutdown_lndnet(net); - if (lnet_count_acceptor_nets() == 0) - lnet_acceptor_stop(); + lnet_acceptor_stop(); lnet_ping_target_update(pbuf, ping_mdh); @@ -3418,12 +3807,13 @@ lnet_ni_set_healthv(lnet_nid_t nid, int value, bool all) lnet_net_lock(LNET_LOCK_EX); 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 || all) { + if (all || (nid_is_nid4(&ni->ni_nid) && + lnet_nid_to_nid4(&ni->ni_nid) == nid)) { atomic_set(&ni->ni_healthv, value); if (list_empty(&ni->ni_recovery) && value < LNET_MAX_HEALTH_VALUE) { CERROR("manually adding local NI %s to recovery\n", - libcfs_nid2str(ni->ni_nid)); + libcfs_nidstr(&ni->ni_nid)); list_add_tail(&ni->ni_recovery, &the_lnet.ln_mt_localNIRecovq); lnet_ni_addref_locked(ni, 0); @@ -3438,6 +3828,30 @@ lnet_ni_set_healthv(lnet_nid_t nid, int value, bool all) lnet_net_unlock(LNET_LOCK_EX); } +static void +lnet_ni_set_conns_per_peer(lnet_nid_t nid, int value, bool all) +{ + struct lnet_net *net; + struct lnet_ni *ni; + + lnet_net_lock(LNET_LOCK_EX); + list_for_each_entry(net, &the_lnet.ln_nets, net_list) { + list_for_each_entry(ni, &net->net_ni_list, ni_netlist) { + if (lnet_nid_to_nid4(&ni->ni_nid) != nid && !all) + continue; + if (LNET_NETTYP(net->net_id) == SOCKLND) + ni->ni_lnd_tunables.lnd_tun_u.lnd_sock.lnd_conns_per_peer = value; + else if (LNET_NETTYP(net->net_id) == O2IBLND) + ni->ni_lnd_tunables.lnd_tun_u.lnd_o2ib.lnd_conns_per_peer = value; + if (!all) { + lnet_net_unlock(LNET_LOCK_EX); + return; + } + } + } + lnet_net_unlock(LNET_LOCK_EX); +} + static int lnet_get_local_ni_hstats(struct lnet_ioctl_local_ni_hstats *stats) { @@ -3459,7 +3873,10 @@ lnet_get_local_ni_hstats(struct lnet_ioctl_local_ni_hstats *stats) stats->hlni_local_no_route = atomic_read(&ni->ni_hstats.hlt_local_no_route); stats->hlni_local_timeout = atomic_read(&ni->ni_hstats.hlt_local_timeout); stats->hlni_local_error = atomic_read(&ni->ni_hstats.hlt_local_error); + stats->hlni_fatal_error = atomic_read(&ni->ni_fatal_error_on); stats->hlni_health_value = atomic_read(&ni->ni_healthv); + stats->hlni_ping_count = ni->ni_ping_count; + stats->hlni_next_ping = ni->ni_next_ping; unlock: lnet_net_unlock(cpt); @@ -3475,7 +3892,9 @@ lnet_get_local_ni_recovery_list(struct lnet_ioctl_recovery_list *list) lnet_net_lock(LNET_LOCK_EX); list_for_each_entry(ni, &the_lnet.ln_mt_localNIRecovq, ni_recovery) { - list->rlst_nid_array[i] = ni->ni_nid; + if (!nid_is_nid4(&ni->ni_nid)) + continue; + list->rlst_nid_array[i] = lnet_nid_to_nid4(&ni->ni_nid); i++; if (i >= LNET_MAX_SHOW_NUM_NID) break; @@ -3494,7 +3913,7 @@ lnet_get_peer_ni_recovery_list(struct lnet_ioctl_recovery_list *list) lnet_net_lock(LNET_LOCK_EX); list_for_each_entry(lpni, &the_lnet.ln_mt_peerNIRecovq, lpni_recovery) { - list->rlst_nid_array[i] = lpni->lpni_nid; + list->rlst_nid_array[i] = lnet_nid_to_nid4(&lpni->lpni_nid); i++; if (i >= LNET_MAX_SHOW_NUM_NID) break; @@ -3516,6 +3935,7 @@ LNetCtl(unsigned int cmd, void *arg) struct lnet_ioctl_config_data *config; struct lnet_process_id id = {0}; struct lnet_ni *ni; + struct lnet_nid nid; int rc; BUILD_BUG_ON(sizeof(struct lnet_ioctl_net_config) + @@ -3530,20 +3950,29 @@ LNetCtl(unsigned int cmd, void *arg) case IOC_LIBCFS_FAIL_NID: return lnet_fail_nid(data->ioc_nid, data->ioc_count); - case IOC_LIBCFS_ADD_ROUTE: + case IOC_LIBCFS_ADD_ROUTE: { + /* default router sensitivity to 1 */ + unsigned int sensitivity = 1; config = arg; if (config->cfg_hdr.ioc_len < sizeof(*config)) return -EINVAL; + if (config->cfg_config_u.cfg_route.rtr_sensitivity) { + sensitivity = + config->cfg_config_u.cfg_route.rtr_sensitivity; + } + + lnet_nid4_to_nid(config->cfg_nid, &nid); 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, + &nid, config->cfg_config_u.cfg_route. - rtr_priority); + rtr_priority, sensitivity); mutex_unlock(&the_lnet.ln_api_mutex); return rc; + } case IOC_LIBCFS_DEL_ROUTE: config = arg; @@ -3569,7 +3998,9 @@ LNetCtl(unsigned int cmd, void *arg) &config->cfg_nid, &config->cfg_config_u.cfg_route.rtr_flags, &config->cfg_config_u.cfg_route. - rtr_priority); + rtr_priority, + &config->cfg_config_u.cfg_route. + rtr_sensitivity); mutex_unlock(&the_lnet.ln_api_mutex); return rc; @@ -3635,7 +4066,15 @@ LNetCtl(unsigned int cmd, void *arg) return -EINVAL; mutex_lock(&the_lnet.ln_api_mutex); - lnet_counters_get(&lnet_stats->st_cntrs); + rc = lnet_counters_get(&lnet_stats->st_cntrs); + mutex_unlock(&the_lnet.ln_api_mutex); + return rc; + } + + case IOC_LIBCFS_RESET_LNET_STATS: + { + mutex_lock(&the_lnet.ln_api_mutex); + lnet_counters_reset(); mutex_unlock(&the_lnet.ln_api_mutex); return 0; } @@ -3745,7 +4184,7 @@ LNetCtl(unsigned int cmd, void *arg) mutex_lock(&the_lnet.ln_api_mutex); rc = lnet_add_peer_ni(cfg->prcfg_prim_nid, cfg->prcfg_cfg_nid, - cfg->prcfg_mr); + cfg->prcfg_mr, false); mutex_unlock(&the_lnet.ln_api_mutex); return rc; } @@ -3835,6 +4274,25 @@ LNetCtl(unsigned int cmd, void *arg) return 0; } + case IOC_LIBCFS_SET_CONNS_PER_PEER: { + struct lnet_ioctl_reset_conns_per_peer_cfg *cfg = arg; + int value; + + if (cfg->rcpp_hdr.ioc_len < sizeof(*cfg)) + return -EINVAL; + if (cfg->rcpp_value < 0) + value = 1; + else + value = cfg->rcpp_value; + CDEBUG(D_NET, + "Setting conns_per_peer to %d for %s. all = %d\n", + value, libcfs_nid2str(cfg->rcpp_nid), cfg->rcpp_all); + mutex_lock(&the_lnet.ln_api_mutex); + lnet_ni_set_conns_per_peer(cfg->rcpp_nid, value, cfg->rcpp_all); + mutex_unlock(&the_lnet.ln_api_mutex); + return 0; + } + case IOC_LIBCFS_NOTIFY_ROUTER: { time64_t deadline = ktime_get_real_seconds() - data->ioc_u64[0]; @@ -3856,9 +4314,7 @@ LNetCtl(unsigned int cmd, void *arg) return 0; case IOC_LIBCFS_TESTPROTOCOMPAT: - lnet_net_lock(LNET_LOCK_EX); the_lnet.ln_testprotocompat = data->ioc_flags; - lnet_net_unlock(LNET_LOCK_EX); return 0; case IOC_LIBCFS_LNET_FAULT: @@ -3873,9 +4329,9 @@ LNetCtl(unsigned int cmd, void *arg) /* If timeout is negative then set default of 3 minutes */ if (((s32)data->ioc_u32[1] <= 0) || data->ioc_u32[1] > (DEFAULT_PEER_TIMEOUT * MSEC_PER_SEC)) - timeout = msecs_to_jiffies(DEFAULT_PEER_TIMEOUT * MSEC_PER_SEC); + timeout = cfs_time_seconds(DEFAULT_PEER_TIMEOUT); else - timeout = msecs_to_jiffies(data->ioc_u32[1]); + timeout = nsecs_to_jiffies(data->ioc_u32[1] * NSEC_PER_MSEC); rc = lnet_ping(id, timeout, data->ioc_pbuf1, data->ioc_plen1 / sizeof(struct lnet_process_id)); @@ -3895,9 +4351,9 @@ LNetCtl(unsigned int cmd, void *arg) /* If timeout is negative then set default of 3 minutes */ if (((s32)ping->op_param) <= 0 || ping->op_param > (DEFAULT_PEER_TIMEOUT * MSEC_PER_SEC)) - timeout = msecs_to_jiffies(DEFAULT_PEER_TIMEOUT * MSEC_PER_SEC); + timeout = cfs_time_seconds(DEFAULT_PEER_TIMEOUT); else - timeout = msecs_to_jiffies(ping->op_param); + timeout = nsecs_to_jiffies(ping->op_param * NSEC_PER_MSEC); rc = lnet_ping(ping->ping_id, timeout, ping->ping_buf, @@ -3908,7 +4364,8 @@ LNetCtl(unsigned int cmd, void *arg) mutex_lock(&the_lnet.ln_api_mutex); lp = lnet_find_peer(ping->ping_id.nid); if (lp) { - ping->ping_id.nid = lp->lp_primary_nid; + ping->ping_id.nid = + lnet_nid_to_nid4(&lp->lp_primary_nid); ping->mr_info = lnet_peer_is_multi_rail(lp); lnet_peer_decref_locked(lp); } @@ -3931,7 +4388,8 @@ LNetCtl(unsigned int cmd, void *arg) mutex_lock(&the_lnet.ln_api_mutex); lp = lnet_find_peer(discover->ping_id.nid); if (lp) { - discover->ping_id.nid = lp->lp_primary_nid; + discover->ping_id.nid = + lnet_nid_to_nid4(&lp->lp_primary_nid); discover->mr_info = lnet_peer_is_multi_rail(lp); lnet_peer_decref_locked(lp); } @@ -3941,6 +4399,106 @@ LNetCtl(unsigned int cmd, void *arg) return 0; } + case IOC_LIBCFS_ADD_UDSP: { + struct lnet_ioctl_udsp *ioc_udsp = arg; + __u32 bulk_size = ioc_udsp->iou_hdr.ioc_len; + + mutex_lock(&the_lnet.ln_api_mutex); + rc = lnet_udsp_demarshal_add(arg, bulk_size); + if (!rc) { + rc = lnet_udsp_apply_policies(NULL, false); + CDEBUG(D_NET, "policy application returned %d\n", rc); + rc = 0; + } + mutex_unlock(&the_lnet.ln_api_mutex); + + return rc; + } + + case IOC_LIBCFS_DEL_UDSP: { + struct lnet_ioctl_udsp *ioc_udsp = arg; + int idx = ioc_udsp->iou_idx; + + if (ioc_udsp->iou_hdr.ioc_len < sizeof(*ioc_udsp)) + return -EINVAL; + + mutex_lock(&the_lnet.ln_api_mutex); + rc = lnet_udsp_del_policy(idx); + if (!rc) { + rc = lnet_udsp_apply_policies(NULL, false); + CDEBUG(D_NET, "policy re-application returned %d\n", + rc); + rc = 0; + } + mutex_unlock(&the_lnet.ln_api_mutex); + + return rc; + } + + case IOC_LIBCFS_GET_UDSP_SIZE: { + struct lnet_ioctl_udsp *ioc_udsp = arg; + struct lnet_udsp *udsp; + + if (ioc_udsp->iou_hdr.ioc_len < sizeof(*ioc_udsp)) + return -EINVAL; + + rc = 0; + + mutex_lock(&the_lnet.ln_api_mutex); + udsp = lnet_udsp_get_policy(ioc_udsp->iou_idx); + if (!udsp) { + rc = -ENOENT; + } else { + /* coming in iou_idx will hold the idx of the udsp + * to get the size of. going out the iou_idx will + * hold the size of the UDSP found at the passed + * in index. + */ + ioc_udsp->iou_idx = lnet_get_udsp_size(udsp); + if (ioc_udsp->iou_idx < 0) + rc = -EINVAL; + } + mutex_unlock(&the_lnet.ln_api_mutex); + + return rc; + } + + case IOC_LIBCFS_GET_UDSP: { + struct lnet_ioctl_udsp *ioc_udsp = arg; + struct lnet_udsp *udsp; + + if (ioc_udsp->iou_hdr.ioc_len < sizeof(*ioc_udsp)) + return -EINVAL; + + rc = 0; + + mutex_lock(&the_lnet.ln_api_mutex); + udsp = lnet_udsp_get_policy(ioc_udsp->iou_idx); + if (!udsp) + rc = -ENOENT; + else + rc = lnet_udsp_marshal(udsp, ioc_udsp); + mutex_unlock(&the_lnet.ln_api_mutex); + + return rc; + } + + case IOC_LIBCFS_GET_CONST_UDSP_INFO: { + struct lnet_ioctl_construct_udsp_info *info = arg; + + if (info->cud_hdr.ioc_len < sizeof(*info)) + return -EINVAL; + + CDEBUG(D_NET, "GET_UDSP_INFO for %s\n", + libcfs_nid2str(info->cud_nid)); + + mutex_lock(&the_lnet.ln_api_mutex); + lnet_udsp_get_construct_info(info); + mutex_unlock(&the_lnet.ln_api_mutex); + + return 0; + } + default: ni = lnet_net2ni_addref(data->ioc_net); if (ni == NULL) @@ -3981,7 +4539,7 @@ bool LNetIsPeerLocal(lnet_nid_t nid) cpt = lnet_net_lock_current(); 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) { + if (lnet_nid_to_nid4(&ni->ni_nid) == nid) { lnet_net_unlock(cpt); return true; } @@ -4018,10 +4576,13 @@ LNetGetId(unsigned int index, struct lnet_process_id *id) list_for_each_entry(net, &the_lnet.ln_nets, net_list) { list_for_each_entry(ni, &net->net_ni_list, ni_netlist) { + if (!nid_is_nid4(&ni->ni_nid)) + /* FIXME this needs to be handled */ + continue; if (index-- != 0) continue; - id->nid = ni->ni_nid; + id->nid = lnet_nid_to_nid4(&ni->ni_nid); id->pid = the_lnet.ln_pid; rc = 0; break; @@ -4033,24 +4594,44 @@ LNetGetId(unsigned int index, struct lnet_process_id *id) } EXPORT_SYMBOL(LNetGetId); +struct ping_data { + int rc; + int replied; + struct lnet_handle_md mdh; + struct completion completion; +}; + +static void +lnet_ping_event_handler(struct lnet_event *event) +{ + struct ping_data *pd = event->md_user_ptr; + + CDEBUG(D_NET, "ping event (%d %d)%s\n", + event->type, event->status, + event->unlinked ? " unlinked" : ""); + + if (event->status) { + if (!pd->rc) + pd->rc = event->status; + } else if (event->type == LNET_EVENT_REPLY) { + pd->replied = 1; + pd->rc = event->mlength; + } + if (event->unlinked) + complete(&pd->completion); +} + static int lnet_ping(struct lnet_process_id id, signed long timeout, struct lnet_process_id __user *ids, int n_ids) { - struct lnet_handle_eq eqh; - struct lnet_handle_md mdh; - struct lnet_event event; struct lnet_md md = { NULL }; - int which; - int unlinked = 0; - int replied = 0; - const signed long a_long_time = msecs_to_jiffies(60 * MSEC_PER_SEC); + struct ping_data pd = { 0 }; struct lnet_ping_buffer *pbuf; struct lnet_process_id tmpid; int i; int nob; int rc; int rc2; - sigset_t blocked; /* n_ids limit is arbitrary */ if (n_ids <= 0 || id.nid == LNET_NID_ANY) @@ -4070,92 +4651,46 @@ static int lnet_ping(struct lnet_process_id id, signed long timeout, if (!pbuf) return -ENOMEM; - /* NB 2 events max (including any unlink event) */ - rc = LNetEQAlloc(2, LNET_EQ_HANDLER_NONE, &eqh); - if (rc != 0) { - CERROR("Can't allocate EQ: %d\n", rc); - goto fail_ping_buffer_decref; - } - /* initialize md content */ md.start = &pbuf->pb_info; md.length = LNET_PING_INFO_SIZE(n_ids); md.threshold = 2; /* GET/REPLY */ md.max_size = 0; md.options = LNET_MD_TRUNCATE; - md.user_ptr = NULL; - md.eq_handle = eqh; + md.user_ptr = &pd; + md.handler = lnet_ping_event_handler; - rc = LNetMDBind(md, LNET_UNLINK, &mdh); + init_completion(&pd.completion); + + rc = LNetMDBind(&md, LNET_UNLINK, &pd.mdh); if (rc != 0) { CERROR("Can't bind MD: %d\n", rc); - goto fail_free_eq; + goto fail_ping_buffer_decref; } - rc = LNetGet(LNET_NID_ANY, mdh, id, + rc = LNetGet(LNET_NID_ANY, pd.mdh, id, LNET_RESERVED_PORTAL, LNET_PROTO_PING_MATCHBITS, 0, false); if (rc != 0) { /* Don't CERROR; this could be deliberate! */ - rc2 = LNetMDUnlink(mdh); + rc2 = LNetMDUnlink(pd.mdh); LASSERT(rc2 == 0); /* NB must wait for the UNLINK event below... */ - unlinked = 1; - timeout = a_long_time; - } - - do { - /* MUST block for unlink to complete */ - if (unlinked) - blocked = cfs_block_allsigs(); - - rc2 = LNetEQPoll(&eqh, 1, timeout, &event, &which); - - if (unlinked) - cfs_restore_sigs(blocked); - - CDEBUG(D_NET, "poll %d(%d %d)%s\n", rc2, - (rc2 <= 0) ? -1 : event.type, - (rc2 <= 0) ? -1 : event.status, - (rc2 > 0 && event.unlinked) ? " unlinked" : ""); - - LASSERT(rc2 != -EOVERFLOW); /* can't miss anything */ - - if (rc2 <= 0 || event.status != 0) { - /* timeout or error */ - if (!replied && rc == 0) - rc = (rc2 < 0) ? rc2 : - (rc2 == 0) ? -ETIMEDOUT : - event.status; - - if (!unlinked) { - /* Ensure completion in finite time... */ - LNetMDUnlink(mdh); - /* No assertion (racing with network) */ - unlinked = 1; - timeout = a_long_time; - } else if (rc2 == 0) { - /* timed out waiting for unlink */ - CWARN("ping %s: late network completion\n", - libcfs_id2str(id)); - } - } else if (event.type == LNET_EVENT_REPLY) { - replied = 1; - rc = event.mlength; - } - } while (rc2 <= 0 || !event.unlinked); + } - if (!replied) { - if (rc >= 0) - CWARN("%s: Unexpected rc >= 0 but no reply!\n", - libcfs_id2str(id)); + if (wait_for_completion_timeout(&pd.completion, timeout) == 0) { + /* Ensure completion in finite time... */ + LNetMDUnlink(pd.mdh); + wait_for_completion(&pd.completion); + } + if (!pd.replied) { rc = -EIO; - goto fail_free_eq; + goto fail_ping_buffer_decref; } - nob = rc; + nob = pd.rc; LASSERT(nob >= 0 && nob <= LNET_PING_INFO_SIZE(n_ids)); rc = -EPROTO; /* if I can't parse... */ @@ -4163,7 +4698,7 @@ static int lnet_ping(struct lnet_process_id id, signed long timeout, if (nob < 8) { CERROR("%s: ping info too short %d\n", libcfs_id2str(id), nob); - goto fail_free_eq; + goto fail_ping_buffer_decref; } if (pbuf->pb_info.pi_magic == __swab32(LNET_PROTO_PING_MAGIC)) { @@ -4171,20 +4706,20 @@ static int lnet_ping(struct lnet_process_id id, signed long timeout, } else if (pbuf->pb_info.pi_magic != LNET_PROTO_PING_MAGIC) { CERROR("%s: Unexpected magic %08x\n", libcfs_id2str(id), pbuf->pb_info.pi_magic); - goto fail_free_eq; + goto fail_ping_buffer_decref; } if ((pbuf->pb_info.pi_features & LNET_PING_FEAT_NI_STATUS) == 0) { CERROR("%s: ping w/o NI status: 0x%x\n", libcfs_id2str(id), pbuf->pb_info.pi_features); - goto fail_free_eq; + goto fail_ping_buffer_decref; } if (nob < LNET_PING_INFO_SIZE(0)) { CERROR("%s: Short reply %d(%d min)\n", libcfs_id2str(id), nob, (int)LNET_PING_INFO_SIZE(0)); - goto fail_free_eq; + goto fail_ping_buffer_decref; } if (pbuf->pb_info.pi_nnis < n_ids) @@ -4194,7 +4729,7 @@ static int lnet_ping(struct lnet_process_id id, signed long timeout, CERROR("%s: Short reply %d(%d expected)\n", libcfs_id2str(id), nob, (int)LNET_PING_INFO_SIZE(n_ids)); - goto fail_free_eq; + goto fail_ping_buffer_decref; } rc = -EFAULT; /* if I segv in copy_to_user()... */ @@ -4204,16 +4739,10 @@ static int lnet_ping(struct lnet_process_id id, signed long timeout, tmpid.pid = pbuf->pb_info.pi_pid; tmpid.nid = pbuf->pb_info.pi_ni[i].ns_nid; if (copy_to_user(&ids[i], &tmpid, sizeof(tmpid))) - goto fail_free_eq; + goto fail_ping_buffer_decref; } rc = pbuf->pb_info.pi_nnis; - fail_free_eq: - rc2 = LNetEQFree(eqh); - if (rc2 != 0) - CERROR("rc2 %d\n", rc2); - LASSERT(rc2 == 0); - fail_ping_buffer_decref: lnet_ping_buffer_decref(pbuf); return rc; @@ -4230,8 +4759,6 @@ lnet_discover(struct lnet_process_id id, __u32 force, int cpt; int i; int rc; - int max_intf = lnet_interfaces_max; - size_t buf_size; if (n_ids <= 0 || id.nid == LNET_NID_ANY) @@ -4241,15 +4768,13 @@ lnet_discover(struct lnet_process_id id, __u32 force, id.pid = LNET_PID_LUSTRE; /* - * if the user buffer has more space than the max_intf - * then only fill it up to max_intf + * If the user buffer has more space than the lnet_interfaces_max, + * then only fill it up to lnet_interfaces_max. */ - if (n_ids > max_intf) - n_ids = max_intf; - - buf_size = n_ids * sizeof(*buf); + if (n_ids > lnet_interfaces_max) + n_ids = lnet_interfaces_max; - LIBCFS_ALLOC(buf, buf_size); + CFS_ALLOC_PTR_ARRAY(buf, n_ids); if (!buf) return -ENOMEM; @@ -4275,34 +4800,37 @@ lnet_discover(struct lnet_process_id id, __u32 force, if (rc) goto out_decref; - /* Peer may have changed. */ + /* The lpni (or lp) for this NID may have changed and our ref is + * the only thing keeping the old one around. Release the ref + * and lookup the lpni again + */ + lnet_peer_ni_decref_locked(lpni); + lpni = lnet_find_peer_ni_locked(id.nid); + if (!lpni) { + rc = -ENOENT; + goto out; + } lp = lpni->lpni_peer_net->lpn_peer; - if (lp->lp_nnis < n_ids) - n_ids = lp->lp_nnis; i = 0; p = NULL; while ((p = lnet_get_next_peer_ni_locked(lp, NULL, p)) != NULL) { buf[i].pid = id.pid; - buf[i].nid = p->lpni_nid; + buf[i].nid = lnet_nid_to_nid4(&p->lpni_nid); if (++i >= n_ids) break; } + rc = i; - lnet_net_unlock(cpt); - - rc = -EFAULT; - if (copy_to_user(ids, buf, n_ids * sizeof(*buf))) - goto out_relock; - rc = n_ids; -out_relock: - lnet_net_lock(cpt); out_decref: lnet_peer_ni_decref_locked(lpni); out: lnet_net_unlock(cpt); - LIBCFS_FREE(buf, buf_size); + if (rc >= 0) + if (copy_to_user(ids, buf, rc * sizeof(*buf))) + rc = -EFAULT; + CFS_FREE_PTR_ARRAY(buf, n_ids); return rc; }