X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lnet%2Flnet%2Fapi-ni.c;h=d39e6b932380a86404a8bce2c26e36c6d1fe3bf6;hp=1b3aa1a436fbaa9a76c6a3501b59983817dd83aa;hb=5c883ea2748ae9e430a9cd863a9b630b2a74440a;hpb=2be10428ac22426c5868b699b6c0b80c040465dc

diff --git a/lnet/lnet/api-ni.c b/lnet/lnet/api-ni.c
index 1b3aa1a4..d39e6b9 100644
--- a/lnet/lnet/api-ni.c
+++ b/lnet/lnet/api-ni.c
@@ -80,18 +80,18 @@ MODULE_PARM_DESC(lnet_numa_range,
 
 /*
  * lnet_health_sensitivity determines by how much we decrement the health
- * value on sending error. The value defaults to 0, which means health
- * checking is turned off by default.
+ * value on sending error. The value defaults to 100, which means health
+ * interface health is decremented by 100 points every failure.
  */
-unsigned int lnet_health_sensitivity = 0;
+unsigned int lnet_health_sensitivity = 100;
 static int sensitivity_set(const char *val, cfs_kernel_param_arg_t *kp);
+#ifdef HAVE_KERNEL_PARAM_OPS
 static struct kernel_param_ops param_ops_health_sensitivity = {
 	.set = sensitivity_set,
 	.get = param_get_int,
 };
 #define param_check_health_sensitivity(name, p) \
 		__param_check(name, p, int)
-#ifdef HAVE_KERNEL_PARAM_OPS
 module_param(lnet_health_sensitivity, health_sensitivity, S_IRUGO|S_IWUSR);
 #else
 module_param_call(lnet_health_sensitivity, sensitivity_set, param_get_int,
@@ -106,13 +106,13 @@ MODULE_PARM_DESC(lnet_health_sensitivity,
  */
 unsigned int lnet_recovery_interval = 1;
 static int recovery_interval_set(const char *val, cfs_kernel_param_arg_t *kp);
+#ifdef HAVE_KERNEL_PARAM_OPS
 static struct kernel_param_ops param_ops_recovery_interval = {
 	.set = recovery_interval_set,
 	.get = param_get_int,
 };
 #define param_check_recovery_interval(name, p) \
 		__param_check(name, p, int)
-#ifdef HAVE_KERNEL_PARAM_OPS
 module_param(lnet_recovery_interval, recovery_interval, S_IRUGO|S_IWUSR);
 #else
 module_param_call(lnet_recovery_interval, recovery_interval_set, param_get_int,
@@ -160,8 +160,31 @@ module_param_call(lnet_peer_discovery_disabled, discovery_set, param_get_int,
 MODULE_PARM_DESC(lnet_peer_discovery_disabled,
 		"Set to 1 to disable peer discovery on this node.");
 
-unsigned lnet_transaction_timeout = 5;
+unsigned int lnet_drop_asym_route;
+static int drop_asym_route_set(const char *val, cfs_kernel_param_arg_t *kp);
+
+static struct kernel_param_ops param_ops_drop_asym_route = {
+	.set = drop_asym_route_set,
+	.get = param_get_int,
+};
+
+#define param_check_drop_asym_route(name, p)	\
+	__param_check(name, p, int)
+#ifdef HAVE_KERNEL_PARAM_OPS
+module_param(lnet_drop_asym_route, drop_asym_route, 0644);
+#else
+module_param_call(lnet_drop_asym_route, drop_asym_route_set, param_get_int,
+		  &param_ops_drop_asym_route, 0644);
+#endif
+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 50
+
+unsigned lnet_transaction_timeout = LNET_TRANSACTION_TIMEOUT_HEALTH_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 = {
 	.set = transaction_to_set,
 	.get = param_get_int,
@@ -169,17 +192,18 @@ static struct kernel_param_ops param_ops_transaction_timeout = {
 
 #define param_check_transaction_timeout(name, p) \
 		__param_check(name, p, int)
-#ifdef HAVE_KERNEL_PARAM_OPS
 module_param(lnet_transaction_timeout, transaction_timeout, S_IRUGO|S_IWUSR);
 #else
 module_param_call(lnet_transaction_timeout, transaction_to_set, param_get_int,
 		  &lnet_transaction_timeout, S_IRUGO|S_IWUSR);
 #endif
-MODULE_PARM_DESC(lnet_peer_discovery_disabled,
-		"Set to 1 to disable peer discovery on this node.");
+MODULE_PARM_DESC(lnet_transaction_timeout,
+		"Maximum number of seconds to wait for a peer response.");
 
-unsigned lnet_retry_count = 0;
+#define LNET_RETRY_COUNT_HEALTH_DEFAULT 2
+unsigned lnet_retry_count = LNET_RETRY_COUNT_HEALTH_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 = {
 	.set = retry_count_set,
 	.get = param_get_int,
@@ -187,7 +211,6 @@ static struct kernel_param_ops param_ops_retry_count = {
 
 #define param_check_retry_count(name, p) \
 		__param_check(name, p, int)
-#ifdef HAVE_KERNEL_PARAM_OPS
 module_param(lnet_retry_count, retry_count, S_IRUGO|S_IWUSR);
 #else
 module_param_call(lnet_retry_count, retry_count_set, param_get_int,
@@ -198,6 +221,7 @@ MODULE_PARM_DESC(lnet_retry_count,
 
 
 unsigned lnet_lnd_timeout = LNET_LND_DEFAULT_TIMEOUT;
+unsigned int lnet_current_net_count;
 
 /*
  * This sequence number keeps track of how many times DLC was used to
@@ -233,11 +257,6 @@ sensitivity_set(const char *val, cfs_kernel_param_arg_t *kp)
 	 */
 	mutex_lock(&the_lnet.ln_api_mutex);
 
-	if (the_lnet.ln_state != LNET_STATE_RUNNING) {
-		mutex_unlock(&the_lnet.ln_api_mutex);
-		return 0;
-	}
-
 	if (value > LNET_MAX_HEALTH_VALUE) {
 		mutex_unlock(&the_lnet.ln_api_mutex);
 		CERROR("Invalid health value. Maximum: %d value = %lu\n",
@@ -245,6 +264,23 @@ 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;
+		lnet_retry_count = 0;
+	}
+
 	*sensitivity = value;
 
 	mutex_unlock(&the_lnet.ln_api_mutex);
@@ -276,11 +312,6 @@ recovery_interval_set(const char *val, cfs_kernel_param_arg_t *kp)
 	 */
 	mutex_lock(&the_lnet.ln_api_mutex);
 
-	if (the_lnet.ln_state != LNET_STATE_RUNNING) {
-		mutex_unlock(&the_lnet.ln_api_mutex);
-		return 0;
-	}
-
 	*interval = value;
 
 	mutex_unlock(&the_lnet.ln_api_mutex);
@@ -339,15 +370,16 @@ discovery_set(const char *val, cfs_kernel_param_arg_t *kp)
 }
 
 static int
-transaction_to_set(const char *val, cfs_kernel_param_arg_t *kp)
+drop_asym_route_set(const char *val, cfs_kernel_param_arg_t *kp)
 {
 	int rc;
-	unsigned *transaction_to = (unsigned *)kp->arg;
+	unsigned int *drop_asym_route = (unsigned int *)kp->arg;
 	unsigned long value;
 
 	rc = kstrtoul(val, 0, &value);
 	if (rc) {
-		CERROR("Invalid module parameter value for 'lnet_transaction_timeout'\n");
+		CERROR("Invalid module parameter value for "
+		       "'lnet_drop_asym_route'\n");
 		return rc;
 	}
 
@@ -357,11 +389,37 @@ transaction_to_set(const char *val, cfs_kernel_param_arg_t *kp)
 	 */
 	mutex_lock(&the_lnet.ln_api_mutex);
 
-	if (the_lnet.ln_state != LNET_STATE_RUNNING) {
+	if (value == *drop_asym_route) {
 		mutex_unlock(&the_lnet.ln_api_mutex);
 		return 0;
 	}
 
+	*drop_asym_route = value;
+
+	mutex_unlock(&the_lnet.ln_api_mutex);
+
+	return 0;
+}
+
+static int
+transaction_to_set(const char *val, cfs_kernel_param_arg_t *kp)
+{
+	int rc;
+	unsigned *transaction_to = (unsigned *)kp->arg;
+	unsigned long value;
+
+	rc = kstrtoul(val, 0, &value);
+	if (rc) {
+		CERROR("Invalid module parameter value for 'lnet_transaction_timeout'\n");
+		return rc;
+	}
+
+	/*
+	 * 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);
+
 	if (value < lnet_retry_count || value == 0) {
 		mutex_unlock(&the_lnet.ln_api_mutex);
 		CERROR("Invalid value for lnet_transaction_timeout (%lu). "
@@ -405,9 +463,10 @@ retry_count_set(const char *val, cfs_kernel_param_arg_t *kp)
 	 */
 	mutex_lock(&the_lnet.ln_api_mutex);
 
-	if (the_lnet.ln_state != LNET_STATE_RUNNING) {
+	if (lnet_health_sensitivity == 0) {
 		mutex_unlock(&the_lnet.ln_api_mutex);
-		return 0;
+		CERROR("Can not set retry_count when health feature is turned off\n");
+		return -EINVAL;
 	}
 
 	if (value > lnet_transaction_timeout) {
@@ -418,11 +477,6 @@ retry_count_set(const char *val, cfs_kernel_param_arg_t *kp)
 		return -EINVAL;
 	}
 
-	if (value == *retry_count) {
-		mutex_unlock(&the_lnet.ln_api_mutex);
-		return 0;
-	}
-
 	*retry_count = value;
 
 	if (value == 0)
@@ -492,21 +546,18 @@ 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_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 <size-xxx> kmem_cache).
@@ -522,12 +573,32 @@ lnet_descriptor_setup(void)
 	if (!lnet_small_mds_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_small_mds_cachep) {
 		kmem_cache_destroy(lnet_small_mds_cachep);
@@ -589,8 +660,6 @@ lnet_destroy_locks(void)
 		cfs_percpt_lock_free(the_lnet.ln_net_lock);
 		the_lnet.ln_net_lock = NULL;
 	}
-
-	lnet_fini_locks();
 }
 
 static int
@@ -621,141 +690,142 @@ static void lnet_assert_wire_constants(void)
 	 * 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);
 
 	/* 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);
 }
 
-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
@@ -766,15 +836,14 @@ 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));
 
@@ -783,14 +852,13 @@ 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);
@@ -1046,6 +1114,26 @@ 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
@@ -1079,8 +1167,10 @@ lnet_prepare(lnet_pid_t requested_pid)
 	INIT_LIST_HEAD(&the_lnet.ln_mt_localNIRecovq);
 	INIT_LIST_HEAD(&the_lnet.ln_mt_peerNIRecovq);
 	init_waitqueue_head(&the_lnet.ln_dc_waitq);
+	LNetInvalidateEQHandle(&the_lnet.ln_mt_eqh);
+	init_completion(&the_lnet.ln_started);
 
-	rc = lnet_descriptor_setup();
+	rc = lnet_slab_setup();
 	if (rc != 0)
 		goto failed;
 
@@ -1115,14 +1205,6 @@ 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;
@@ -1137,6 +1219,12 @@ 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:
@@ -1147,6 +1235,8 @@ 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
@@ -1158,6 +1248,17 @@ lnet_unprepare (void)
 	LASSERT(list_empty(&the_lnet.ln_test_peers));
 	LASSERT(list_empty(&the_lnet.ln_nets));
 
+	if (the_lnet.ln_mt_zombie_rstqs) {
+		lnet_clean_zombie_rstqs();
+		the_lnet.ln_mt_zombie_rstqs = NULL;
+	}
+
+	if (!LNetEQHandleIsInvalid(the_lnet.ln_mt_eqh)) {
+		rc = LNetEQFree(the_lnet.ln_mt_eqh);
+		LNetInvalidateEQHandle(&the_lnet.ln_mt_eqh);
+		LASSERT(rc == 0);
+	}
+
 	lnet_portals_destroy();
 
 	if (the_lnet.ln_md_containers != NULL) {
@@ -1165,11 +1266,6 @@ 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();
@@ -1181,7 +1277,7 @@ lnet_unprepare (void)
 		the_lnet.ln_counters = NULL;
 	}
 	lnet_destroy_remote_nets_table();
-	lnet_descriptor_cleanup();
+	lnet_slab_cleanup();
 
 	return 0;
 }
@@ -1306,18 +1402,28 @@ lnet_cpt_of_nid(lnet_nid_t nid, struct lnet_ni *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;
@@ -1407,7 +1513,7 @@ 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));
 }
 
@@ -1475,6 +1581,44 @@ lnet_get_ni_count(void)
 }
 
 int
+lnet_get_net_count(void)
+{
+	struct lnet_net *net;
+	int count = 0;
+
+	lnet_net_lock(0);
+
+	list_for_each_entry(net, &the_lnet.ln_nets, net_list) {
+		count++;
+	}
+
+	lnet_net_unlock(0);
+
+	return count;
+}
+
+void
+lnet_swap_pinginfo(struct lnet_ping_buffer *pbuf)
+{
+	struct lnet_ni_status *stat;
+	int nnis;
+	int i;
+
+	__swab32s(&pbuf->pb_info.pi_magic);
+	__swab32s(&pbuf->pb_info.pi_features);
+	__swab32s(&pbuf->pb_info.pi_pid);
+	__swab32s(&pbuf->pb_info.pi_nnis);
+	nnis = pbuf->pb_info.pi_nnis;
+	if (nnis > pbuf->pb_nnis)
+		nnis = pbuf->pb_nnis;
+	for (i = 0; i < nnis; i++) {
+		stat = &pbuf->pb_info.pi_ni[i];
+		__swab64s(&stat->ns_nid);
+		__swab32s(&stat->ns_status);
+	}
+}
+
+int
 lnet_ping_info_validate(struct lnet_ping_info *pinfo)
 {
 	if (!pinfo)
@@ -1531,7 +1675,7 @@ 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;
 
@@ -1551,11 +1695,11 @@ 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;
 	}
@@ -1570,7 +1714,7 @@ lnet_ping_target_setup(struct lnet_ping_buffer **ppbuf,
 	md.eq_handle = the_lnet.ln_ping_target_eq;
 	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;
@@ -1580,10 +1724,9 @@ lnet_ping_target_setup(struct lnet_ping_buffer **ppbuf,
 	return 0;
 
 fail_unlink_ping_me:
-	rc2 = LNetMEUnlink(me_handle);
-	LASSERT(rc2 == 0);
+	LNetMEUnlink(me);
 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:
@@ -1604,10 +1747,9 @@ lnet_ping_md_unlink(struct lnet_ping_buffer *pbuf,
 	LNetInvalidateMDHandle(ping_mdh);
 
 	/* NB the MD could be busy; this just starts the unlink */
-	while (lnet_ping_buffer_numref(pbuf) > 1) {
+	while (atomic_read(&pbuf->pb_refcnt) > 1) {
 		CDEBUG(D_NET, "Still waiting for ping data MD to unlink\n");
-		set_current_state(TASK_UNINTERRUPTIBLE);
-		schedule_timeout(cfs_time_seconds(1));
+		schedule_timeout_uninterruptible(cfs_time_seconds(1));
 	}
 
 	cfs_restore_sigs(blocked);
@@ -1713,7 +1855,7 @@ 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_me *me;
 	struct lnet_handle_md mdh;
 	struct lnet_handle_md old_mdh;
 	struct lnet_ping_buffer *pbuf;
@@ -1732,11 +1874,12 @@ again:
 		goto fail_return;
 	}
 
-	rc = LNetMEAttach(LNET_RESERVED_PORTAL, id,
+	me = LNetMEAttach(LNET_RESERVED_PORTAL, id,
 			  LNET_PROTO_PING_MATCHBITS, 0,
-			  LNET_UNLINK, LNET_INS_AFTER,
-			  &meh);
-	if (rc) {
+			  LNET_UNLINK, LNET_INS_AFTER);
+
+	if (IS_ERR(me)) {
+		rc = PTR_ERR(me);
 		CERROR("Can't create push target ME: %d\n", rc);
 		goto fail_decref_pbuf;
 	}
@@ -1751,10 +1894,10 @@ again:
 	md.user_ptr  = pbuf;
 	md.eq_handle = the_lnet.ln_push_target_eq;
 
-	rc = LNetMDAttach(meh, md, LNET_RETAIN, &mdh);
+	rc = LNetMDAttach(me, md, LNET_RETAIN, &mdh);
 	if (rc) {
 		CERROR("Can't attach push MD: %d\n", rc);
-		goto fail_unlink_meh;
+		goto fail_unlink_me;
 	}
 	lnet_ping_buffer_addref(pbuf);
 
@@ -1777,8 +1920,8 @@ again:
 
 	return 0;
 
-fail_unlink_meh:
-	LNetMEUnlink(meh);
+fail_unlink_me:
+	LNetMEUnlink(me);
 fail_decref_pbuf:
 	lnet_ping_buffer_decref(pbuf);
 fail_return:
@@ -1837,10 +1980,9 @@ 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) {
+	while (atomic_read(&the_lnet.ln_push_target->pb_refcnt) > 1) {
 		CDEBUG(D_NET, "Still waiting for ping data MD to unlink\n");
-		set_current_state(TASK_UNINTERRUPTIBLE);
-		schedule_timeout(cfs_time_seconds(1));
+		schedule_timeout_uninterruptible(cfs_time_seconds(1));
 	}
 
 	lnet_ping_buffer_decref(the_lnet.ln_push_target);
@@ -1916,8 +2058,7 @@ lnet_clear_zombies_nis_locked(struct lnet_net *net)
 				       "Waiting for zombie LNI %s\n",
 				       libcfs_nid2str(ni->ni_nid));
 			}
-			set_current_state(TASK_UNINTERRUPTIBLE);
-			schedule_timeout(cfs_time_seconds(1));
+			schedule_timeout_uninterruptible(cfs_time_seconds(1));
 			lnet_net_lock(LNET_LOCK_EX);
 			continue;
 		}
@@ -1927,7 +2068,14 @@ 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",
@@ -1970,8 +2118,6 @@ 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)) {
@@ -1987,15 +2133,6 @@ 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);
 }
 
@@ -2003,11 +2140,9 @@ 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 */
@@ -2079,9 +2214,6 @@ 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;
 	}
 
@@ -2141,21 +2273,19 @@ 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
@@ -2190,10 +2320,6 @@ 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);
@@ -2244,7 +2370,7 @@ lnet_startup_lndnet(struct lnet_net *net, struct lnet_lnd_tunables *tun)
 		 * 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;
+			rc = -EEXIST;
 			goto failed1;
 		}
 
@@ -2254,9 +2380,6 @@ 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;
 
@@ -2281,7 +2404,6 @@ 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
@@ -2298,6 +2420,9 @@ lnet_startup_lndnet(struct lnet_net *net, struct lnet_lnd_tunables *tun)
 		lnet_net_unlock(LNET_LOCK_EX);
 	}
 
+	/* update net count */
+	lnet_current_net_count = lnet_get_net_count();
+
 	return ni_count;
 
 failed1:
@@ -2392,12 +2517,8 @@ int lnet_lib_init(void)
 	}
 
 	the_lnet.ln_refcount = 0;
-	LNetInvalidateEQHandle(&the_lnet.ln_rc_eqh);
-	INIT_LIST_HEAD(&the_lnet.ln_lnds);
 	INIT_LIST_HEAD(&the_lnet.ln_net_zombie);
-	INIT_LIST_HEAD(&the_lnet.ln_rcd_zombie);
 	INIT_LIST_HEAD(&the_lnet.ln_msg_resend);
-	INIT_LIST_HEAD(&the_lnet.ln_rcd_deathrow);
 
 	/* The hash table size is the number of bits it takes to express the set
 	 * ln_num_routes, minus 1 (better to under estimate than over so we
@@ -2421,14 +2542,18 @@ 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();
 }
 
@@ -2455,11 +2580,9 @@ 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);
@@ -2513,10 +2636,6 @@ LNetNIInit(lnet_pid_t requested_pid)
 		if (rc != 0)
 			goto err_shutdown_lndnis;
 
-		rc = lnet_check_routes();
-		if (rc != 0)
-			goto err_destroy_routes;
-
 		rc = lnet_rtrpools_alloc(im_a_router);
 		if (rc != 0)
 			goto err_destroy_routes;
@@ -2535,29 +2654,40 @@ LNetNIInit(lnet_pid_t requested_pid)
 
 	lnet_ping_target_update(pbuf, ping_mdh);
 
-	rc = lnet_monitor_thr_start();
-	if (rc != 0)
+	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_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:
@@ -2593,7 +2723,7 @@ EXPORT_SYMBOL(LNetNIInit);
  * \return always 0 for current implementation.
  */
 int
-LNetNIFini()
+LNetNIFini(void)
 {
 	mutex_lock(&the_lnet.ln_api_mutex);
 
@@ -2606,10 +2736,10 @@ LNetNIFini()
 
 		lnet_fault_fini();
 
-		lnet_router_debugfs_init();
+		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 */
@@ -3039,9 +3169,7 @@ static int lnet_handle_legacy_ip2nets(char *ip2nets,
 	struct lnet_net *net;
 	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)
@@ -3220,14 +3348,12 @@ 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);
-
 	/* Create a net/ni structures for the network string */
 	rc = lnet_parse_networks(&net_head, nets, use_tcp_bonding);
 	if (rc <= 0)
@@ -3447,26 +3573,28 @@ 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;
+		}
+
 		mutex_lock(&the_lnet.ln_api_mutex);
 		rc = lnet_add_route(config->cfg_net,
 				    config->cfg_config_u.cfg_route.rtr_hop,
 				    config->cfg_nid,
 				    config->cfg_config_u.cfg_route.
-					rtr_priority);
-		if (rc == 0) {
-			rc = lnet_check_routes();
-			if (rc != 0)
-				lnet_del_route(config->cfg_net,
-					       config->cfg_nid);
-		}
+					rtr_priority, sensitivity);
 		mutex_unlock(&the_lnet.ln_api_mutex);
 		return rc;
+	}
 
 	case IOC_LIBCFS_DEL_ROUTE:
 		config = arg;
@@ -3492,7 +3620,9 @@ 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;
 
@@ -3766,7 +3896,7 @@ LNetCtl(unsigned int cmd, void *arg)
 		 * that deadline to the wall clock.
 		 */
 		deadline += ktime_get_seconds();
-		return lnet_notify(NULL, data->ioc_nid, data->ioc_flags,
+		return lnet_notify(NULL, data->ioc_nid, data->ioc_flags, false,
 				   deadline);
 	}
 
@@ -3779,9 +3909,7 @@ 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:
@@ -3796,9 +3924,9 @@ 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));
@@ -3818,9 +3946,9 @@ 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,
@@ -3966,7 +4094,7 @@ static int lnet_ping(struct lnet_process_id id, signed long timeout,
 	int which;
 	int unlinked = 0;
 	int replied = 0;
-	const signed long a_long_time = msecs_to_jiffies(60 * MSEC_PER_SEC);
+	const signed long a_long_time = cfs_time_seconds(60);
 	struct lnet_ping_buffer *pbuf;
 	struct lnet_process_id tmpid;
 	int i;