#include <lnet/lib-lnet.h>
#include <lnet/lib-dlc.h>
+static void
+lnet_peer_remove_from_remote_list(struct lnet_peer_ni *lpni)
+{
+ if (!list_empty(&lpni->lpni_on_remote_peer_ni_list)) {
+ list_del_init(&lpni->lpni_on_remote_peer_ni_list);
+ lnet_peer_ni_decref_locked(lpni);
+ }
+}
+
+void
+lnet_peer_net_added(struct lnet_net *net)
+{
+ struct lnet_peer_ni *lpni, *tmp;
+
+ list_for_each_entry_safe(lpni, tmp, &the_lnet.ln_remote_peer_ni_list,
+ lpni_on_remote_peer_ni_list) {
+
+ if (LNET_NIDNET(lpni->lpni_nid) == net->net_id) {
+ lpni->lpni_net = net;
+
+ spin_lock(&lpni->lpni_lock);
+ lpni->lpni_txcredits =
+ lpni->lpni_net->net_tunables.lct_peer_tx_credits;
+ lpni->lpni_mintxcredits = lpni->lpni_txcredits;
+ lpni->lpni_rtrcredits =
+ lnet_peer_buffer_credits(lpni->lpni_net);
+ lpni->lpni_minrtrcredits = lpni->lpni_rtrcredits;
+ spin_unlock(&lpni->lpni_lock);
+
+ lnet_peer_remove_from_remote_list(lpni);
+ }
+ }
+}
+
+static void
+lnet_peer_tables_destroy(void)
+{
+ struct lnet_peer_table *ptable;
+ struct list_head *hash;
+ int i;
+ int j;
+
+ if (!the_lnet.ln_peer_tables)
+ return;
+
+ cfs_percpt_for_each(ptable, i, the_lnet.ln_peer_tables) {
+ hash = ptable->pt_hash;
+ if (!hash) /* not intialized */
+ break;
+
+ LASSERT(list_empty(&ptable->pt_zombie_list));
+
+ ptable->pt_hash = NULL;
+ for (j = 0; j < LNET_PEER_HASH_SIZE; j++)
+ LASSERT(list_empty(&hash[j]));
+
+ LIBCFS_FREE(hash, LNET_PEER_HASH_SIZE * sizeof(*hash));
+ }
+
+ cfs_percpt_free(the_lnet.ln_peer_tables);
+ the_lnet.ln_peer_tables = NULL;
+}
+
int
lnet_peer_tables_create(void)
{
return -ENOMEM;
}
+ spin_lock_init(&ptable->pt_zombie_lock);
+ INIT_LIST_HEAD(&ptable->pt_zombie_list);
+
for (j = 0; j < LNET_PEER_HASH_SIZE; j++)
INIT_LIST_HEAD(&hash[j]);
ptable->pt_hash = hash; /* sign of initialization */
return 0;
}
-void
-lnet_peer_tables_destroy(void)
+static struct lnet_peer_ni *
+lnet_peer_ni_alloc(lnet_nid_t nid)
{
- struct lnet_peer_table *ptable;
- struct list_head *hash;
- int i;
- int j;
+ struct lnet_peer_ni *lpni;
+ struct lnet_net *net;
+ int cpt;
+
+ cpt = lnet_nid_cpt_hash(nid, LNET_CPT_NUMBER);
+
+ LIBCFS_CPT_ALLOC(lpni, lnet_cpt_table(), cpt, sizeof(*lpni));
+ if (!lpni)
+ return NULL;
+
+ INIT_LIST_HEAD(&lpni->lpni_txq);
+ INIT_LIST_HEAD(&lpni->lpni_rtrq);
+ INIT_LIST_HEAD(&lpni->lpni_routes);
+ INIT_LIST_HEAD(&lpni->lpni_hashlist);
+ INIT_LIST_HEAD(&lpni->lpni_on_peer_net_list);
+ INIT_LIST_HEAD(&lpni->lpni_on_remote_peer_ni_list);
+
+ spin_lock_init(&lpni->lpni_lock);
+
+ lpni->lpni_alive = !lnet_peers_start_down(); /* 1 bit!! */
+ lpni->lpni_last_alive = cfs_time_current(); /* assumes alive */
+ lpni->lpni_ping_feats = LNET_PING_FEAT_INVAL;
+ lpni->lpni_nid = nid;
+ lpni->lpni_cpt = cpt;
+ lnet_set_peer_ni_health_locked(lpni, true);
+
+ net = lnet_get_net_locked(LNET_NIDNET(nid));
+ lpni->lpni_net = net;
+ if (net) {
+ lpni->lpni_txcredits = net->net_tunables.lct_peer_tx_credits;
+ lpni->lpni_mintxcredits = lpni->lpni_txcredits;
+ lpni->lpni_rtrcredits = lnet_peer_buffer_credits(net);
+ lpni->lpni_minrtrcredits = lpni->lpni_rtrcredits;
+ } else {
+ /*
+ * This peer_ni is not on a local network, so we
+ * cannot add the credits here. In case the net is
+ * added later, add the peer_ni to the remote peer ni
+ * list so it can be easily found and revisited.
+ */
+ /* FIXME: per-net implementation instead? */
+ atomic_inc(&lpni->lpni_refcount);
+ list_add_tail(&lpni->lpni_on_remote_peer_ni_list,
+ &the_lnet.ln_remote_peer_ni_list);
+ }
+
+ /* TODO: update flags */
+
+ return lpni;
+}
+
+static struct lnet_peer_net *
+lnet_peer_net_alloc(__u32 net_id)
+{
+ struct lnet_peer_net *lpn;
- if (the_lnet.ln_peer_tables == NULL)
+ LIBCFS_CPT_ALLOC(lpn, lnet_cpt_table(), CFS_CPT_ANY, sizeof(*lpn));
+ if (!lpn)
+ return NULL;
+
+ INIT_LIST_HEAD(&lpn->lpn_on_peer_list);
+ INIT_LIST_HEAD(&lpn->lpn_peer_nis);
+ lpn->lpn_net_id = net_id;
+
+ return lpn;
+}
+
+static struct lnet_peer *
+lnet_peer_alloc(lnet_nid_t nid)
+{
+ struct lnet_peer *lp;
+
+ LIBCFS_CPT_ALLOC(lp, lnet_cpt_table(), CFS_CPT_ANY, sizeof(*lp));
+ if (!lp)
+ return NULL;
+
+ INIT_LIST_HEAD(&lp->lp_on_lnet_peer_list);
+ INIT_LIST_HEAD(&lp->lp_peer_nets);
+ lp->lp_primary_nid = nid;
+
+ /* TODO: update flags */
+
+ return lp;
+}
+
+
+static void
+lnet_try_destroy_peer_hierarchy_locked(struct lnet_peer_ni *lpni)
+{
+ struct lnet_peer_net *peer_net;
+ struct lnet_peer *peer;
+
+ /* TODO: could the below situation happen? accessing an already
+ * destroyed peer? */
+ if (lpni->lpni_peer_net == NULL ||
+ lpni->lpni_peer_net->lpn_peer == NULL)
return;
- cfs_percpt_for_each(ptable, i, the_lnet.ln_peer_tables) {
- hash = ptable->pt_hash;
- if (hash == NULL) /* not intialized */
- break;
+ peer_net = lpni->lpni_peer_net;
+ peer = lpni->lpni_peer_net->lpn_peer;
- ptable->pt_hash = NULL;
- for (j = 0; j < LNET_PEER_HASH_SIZE; j++)
- LASSERT(list_empty(&hash[j]));
+ list_del_init(&lpni->lpni_on_peer_net_list);
+ lpni->lpni_peer_net = NULL;
- LIBCFS_FREE(hash, LNET_PEER_HASH_SIZE * sizeof(*hash));
+ /* if peer_net is empty, then remove it from the peer */
+ if (list_empty(&peer_net->lpn_peer_nis)) {
+ list_del_init(&peer_net->lpn_on_peer_list);
+ peer_net->lpn_peer = NULL;
+ LIBCFS_FREE(peer_net, sizeof(*peer_net));
+
+ /* if the peer is empty then remove it from the
+ * the_lnet.ln_peers */
+ if (list_empty(&peer->lp_peer_nets)) {
+ list_del_init(&peer->lp_on_lnet_peer_list);
+ LIBCFS_FREE(peer, sizeof(*peer));
+ }
}
+}
- cfs_percpt_free(the_lnet.ln_peer_tables);
- the_lnet.ln_peer_tables = NULL;
+/* called with lnet_net_lock LNET_LOCK_EX held */
+static void
+lnet_peer_ni_del_locked(struct lnet_peer_ni *lpni)
+{
+ struct lnet_peer_table *ptable = NULL;
+
+ lnet_peer_remove_from_remote_list(lpni);
+
+ /* remove peer ni from the hash list. */
+ list_del_init(&lpni->lpni_hashlist);
+
+ /* decrement the ref count on the peer table */
+ ptable = the_lnet.ln_peer_tables[lpni->lpni_cpt];
+ LASSERT(atomic_read(&ptable->pt_number) > 0);
+ atomic_dec(&ptable->pt_number);
+
+ /*
+ * The peer_ni can no longer be found with a lookup. But there
+ * can be current users, so keep track of it on the zombie
+ * list until the reference count has gone to zero.
+ *
+ * The last reference may be lost in a place where the
+ * lnet_net_lock locks only a single cpt, and that cpt may not
+ * be lpni->lpni_cpt. So the zombie list of this peer_table
+ * has its own lock.
+ */
+ spin_lock(&ptable->pt_zombie_lock);
+ list_add(&lpni->lpni_hashlist, &ptable->pt_zombie_list);
+ ptable->pt_zombies++;
+ spin_unlock(&ptable->pt_zombie_lock);
+
+ /* no need to keep this peer on the hierarchy anymore */
+ lnet_try_destroy_peer_hierarchy_locked(lpni);
+
+ /* decrement reference on peer */
+ lnet_peer_ni_decref_locked(lpni);
+}
+
+void lnet_peer_uninit()
+{
+ struct lnet_peer_ni *lpni, *tmp;
+
+ lnet_net_lock(LNET_LOCK_EX);
+
+ /* remove all peer_nis from the remote peer and the hash list */
+ list_for_each_entry_safe(lpni, tmp, &the_lnet.ln_remote_peer_ni_list,
+ lpni_on_remote_peer_ni_list)
+ lnet_peer_ni_del_locked(lpni);
+
+ lnet_peer_tables_destroy();
+
+ lnet_net_unlock(LNET_LOCK_EX);
}
static void
-lnet_peer_table_cleanup_locked(lnet_ni_t *ni, struct lnet_peer_table *ptable)
+lnet_peer_del_locked(struct lnet_peer *peer)
+{
+ struct lnet_peer_ni *lpni = NULL, *lpni2;
+
+ lpni = lnet_get_next_peer_ni_locked(peer, NULL, lpni);
+ while (lpni != NULL) {
+ lpni2 = lnet_get_next_peer_ni_locked(peer, NULL, lpni);
+ lnet_peer_ni_del_locked(lpni);
+ lpni = lpni2;
+ }
+}
+
+static void
+lnet_peer_table_cleanup_locked(struct lnet_net *net,
+ struct lnet_peer_table *ptable)
{
int i;
- struct lnet_peer_ni *lp;
- struct lnet_peer_ni *tmp;
+ struct lnet_peer_ni *next;
+ struct lnet_peer_ni *lpni;
+ struct lnet_peer *peer;
for (i = 0; i < LNET_PEER_HASH_SIZE; i++) {
- list_for_each_entry_safe(lp, tmp, &ptable->pt_hash[i],
+ list_for_each_entry_safe(lpni, next, &ptable->pt_hash[i],
lpni_hashlist) {
- if (ni != NULL && ni->ni_net != lp->lpni_net)
+ if (net != NULL && net != lpni->lpni_net)
+ continue;
+
+ peer = lpni->lpni_peer_net->lpn_peer;
+ if (peer->lp_primary_nid != lpni->lpni_nid) {
+ lnet_peer_ni_del_locked(lpni);
continue;
- list_del_init(&lp->lpni_hashlist);
- /* Lose hash table's ref */
- ptable->pt_zombies++;
- lnet_peer_ni_decref_locked(lp);
+ }
+ /*
+ * Removing the primary NID implies removing
+ * the entire peer. Advance next beyond any
+ * peer_ni that belongs to the same peer.
+ */
+ list_for_each_entry_from(next, &ptable->pt_hash[i],
+ lpni_hashlist) {
+ if (next->lpni_peer_net->lpn_peer != peer)
+ break;
+ }
+ lnet_peer_del_locked(peer);
}
}
}
static void
-lnet_peer_table_finalize_wait_locked(struct lnet_peer_table *ptable,
- int cpt_locked)
+lnet_peer_ni_finalize_wait(struct lnet_peer_table *ptable)
{
- int i;
+ int i = 3;
- for (i = 3; ptable->pt_zombies != 0; i++) {
- lnet_net_unlock(cpt_locked);
+ spin_lock(&ptable->pt_zombie_lock);
+ while (ptable->pt_zombies) {
+ spin_unlock(&ptable->pt_zombie_lock);
if (IS_PO2(i)) {
CDEBUG(D_WARNING,
}
set_current_state(TASK_UNINTERRUPTIBLE);
schedule_timeout(cfs_time_seconds(1) >> 1);
- lnet_net_lock(cpt_locked);
+ spin_lock(&ptable->pt_zombie_lock);
}
+ spin_unlock(&ptable->pt_zombie_lock);
}
static void
-lnet_peer_table_del_rtrs_locked(lnet_ni_t *ni, struct lnet_peer_table *ptable,
- int cpt_locked)
+lnet_peer_table_del_rtrs_locked(struct lnet_net *net,
+ struct lnet_peer_table *ptable)
{
struct lnet_peer_ni *lp;
struct lnet_peer_ni *tmp;
for (i = 0; i < LNET_PEER_HASH_SIZE; i++) {
list_for_each_entry_safe(lp, tmp, &ptable->pt_hash[i],
lpni_hashlist) {
- if (ni->ni_net != lp->lpni_net)
+ if (net != lp->lpni_net)
continue;
if (lp->lpni_rtr_refcount == 0)
lpni_nid = lp->lpni_nid;
- lnet_net_unlock(cpt_locked);
+ lnet_net_unlock(LNET_LOCK_EX);
lnet_del_route(LNET_NIDNET(LNET_NID_ANY), lpni_nid);
- lnet_net_lock(cpt_locked);
+ lnet_net_lock(LNET_LOCK_EX);
}
}
}
void
-lnet_peer_tables_cleanup(lnet_ni_t *ni)
+lnet_peer_tables_cleanup(struct lnet_net *net)
{
int i;
struct lnet_peer_table *ptable;
- LASSERT(the_lnet.ln_shutdown || ni != NULL);
+ LASSERT(the_lnet.ln_shutdown || net != NULL);
/* If just deleting the peers for a NI, get rid of any routes these
* peers are gateways for. */
cfs_percpt_for_each(ptable, i, the_lnet.ln_peer_tables) {
lnet_net_lock(LNET_LOCK_EX);
- lnet_peer_table_del_rtrs_locked(ni, ptable, i);
+ lnet_peer_table_del_rtrs_locked(net, ptable);
lnet_net_unlock(LNET_LOCK_EX);
}
/* Start the cleanup process */
cfs_percpt_for_each(ptable, i, the_lnet.ln_peer_tables) {
lnet_net_lock(LNET_LOCK_EX);
- lnet_peer_table_cleanup_locked(ni, ptable);
+ lnet_peer_table_cleanup_locked(net, ptable);
lnet_net_unlock(LNET_LOCK_EX);
}
- /* Wait until all peers have been destroyed. */
- cfs_percpt_for_each(ptable, i, the_lnet.ln_peer_tables) {
- lnet_net_lock(LNET_LOCK_EX);
- lnet_peer_table_finalize_wait_locked(ptable, i);
- lnet_net_unlock(LNET_LOCK_EX);
- }
+ cfs_percpt_for_each(ptable, i, the_lnet.ln_peer_tables)
+ lnet_peer_ni_finalize_wait(ptable);
}
static struct lnet_peer_ni *
}
struct lnet_peer_ni *
-lnet_find_peer_ni_locked(lnet_nid_t nid, int cpt)
+lnet_find_peer_ni_locked(lnet_nid_t nid)
{
struct lnet_peer_ni *lpni;
struct lnet_peer_table *ptable;
+ int cpt;
+
+ cpt = lnet_nid_cpt_hash(nid, LNET_CPT_NUMBER);
ptable = the_lnet.ln_peer_tables[cpt];
lpni = lnet_get_peer_ni_locked(ptable, nid);
return lpni;
}
-static void
-lnet_try_destroy_peer_hierarchy_locked(struct lnet_peer_ni *lpni)
+struct lnet_peer *
+lnet_find_or_create_peer_locked(lnet_nid_t dst_nid, int cpt)
+{
+ struct lnet_peer_ni *lpni;
+ struct lnet_peer *lp;
+
+ lpni = lnet_find_peer_ni_locked(dst_nid);
+ if (!lpni) {
+ lpni = lnet_nid2peerni_locked(dst_nid, cpt);
+ if (IS_ERR(lpni))
+ return ERR_CAST(lpni);
+ }
+
+ lp = lpni->lpni_peer_net->lpn_peer;
+ lnet_peer_ni_decref_locked(lpni);
+
+ return lp;
+}
+
+struct lnet_peer_ni *
+lnet_get_peer_ni_idx_locked(int idx, struct lnet_peer_net **lpn,
+ struct lnet_peer **lp)
+{
+ struct lnet_peer_ni *lpni;
+
+ list_for_each_entry((*lp), &the_lnet.ln_peers, lp_on_lnet_peer_list) {
+ list_for_each_entry((*lpn), &((*lp)->lp_peer_nets), lpn_on_peer_list) {
+ list_for_each_entry(lpni, &((*lpn)->lpn_peer_nis),
+ lpni_on_peer_net_list)
+ if (idx-- == 0)
+ return lpni;
+ }
+ }
+
+ return NULL;
+}
+
+struct lnet_peer_ni *
+lnet_get_next_peer_ni_locked(struct lnet_peer *peer,
+ struct lnet_peer_net *peer_net,
+ struct lnet_peer_ni *prev)
+{
+ struct lnet_peer_ni *lpni;
+ struct lnet_peer_net *net = peer_net;
+
+ if (!prev) {
+ if (!net)
+ net = list_entry(peer->lp_peer_nets.next,
+ struct lnet_peer_net,
+ lpn_on_peer_list);
+ lpni = list_entry(net->lpn_peer_nis.next, struct lnet_peer_ni,
+ lpni_on_peer_net_list);
+
+ return lpni;
+ }
+
+ if (prev->lpni_on_peer_net_list.next ==
+ &prev->lpni_peer_net->lpn_peer_nis) {
+ /*
+ * if you reached the end of the peer ni list and the peer
+ * net is specified then there are no more peer nis in that
+ * net.
+ */
+ if (net)
+ return NULL;
+
+ /*
+ * we reached the end of this net ni list. move to the
+ * next net
+ */
+ if (prev->lpni_peer_net->lpn_on_peer_list.next ==
+ &peer->lp_peer_nets)
+ /* no more nets and no more NIs. */
+ return NULL;
+
+ /* get the next net */
+ net = list_entry(prev->lpni_peer_net->lpn_on_peer_list.next,
+ struct lnet_peer_net,
+ lpn_on_peer_list);
+ /* get the ni on it */
+ lpni = list_entry(net->lpn_peer_nis.next, struct lnet_peer_ni,
+ lpni_on_peer_net_list);
+
+ return lpni;
+ }
+
+ /* there are more nis left */
+ lpni = list_entry(prev->lpni_on_peer_net_list.next,
+ struct lnet_peer_ni, lpni_on_peer_net_list);
+
+ return lpni;
+}
+
+bool
+lnet_peer_is_ni_pref_locked(struct lnet_peer_ni *lpni, struct lnet_ni *ni)
+{
+ int i;
+
+ for (i = 0; i < lpni->lpni_pref_nnids; i++) {
+ if (lpni->lpni_pref_nids[i] == ni->ni_nid)
+ return true;
+ }
+ return false;
+}
+
+lnet_nid_t
+lnet_peer_primary_nid(lnet_nid_t nid)
+{
+ struct lnet_peer_ni *lpni;
+ lnet_nid_t primary_nid = nid;
+ int cpt;
+
+ cpt = lnet_net_lock_current();
+ lpni = lnet_find_peer_ni_locked(nid);
+ if (lpni) {
+ primary_nid = lpni->lpni_peer_net->lpn_peer->lp_primary_nid;
+ lnet_peer_ni_decref_locked(lpni);
+ }
+ lnet_net_unlock(cpt);
+
+ return primary_nid;
+}
+
+struct lnet_peer_net *
+lnet_peer_get_net_locked(struct lnet_peer *peer, __u32 net_id)
{
struct lnet_peer_net *peer_net;
+ list_for_each_entry(peer_net, &peer->lp_peer_nets, lpn_on_peer_list) {
+ if (peer_net->lpn_net_id == net_id)
+ return peer_net;
+ }
+ return NULL;
+}
+
+static int
+lnet_peer_setup_hierarchy(struct lnet_peer *lp, struct lnet_peer_ni *lpni,
+ lnet_nid_t nid)
+{
+ struct lnet_peer_net *lpn = NULL;
+ struct lnet_peer_table *ptable;
+ __u32 net_id = LNET_NIDNET(nid);
+
+ /*
+ * Create the peer_ni, peer_net, and peer if they don't exist
+ * yet.
+ */
+ if (lp) {
+ lpn = lnet_peer_get_net_locked(lp, net_id);
+ } else {
+ lp = lnet_peer_alloc(nid);
+ if (!lp)
+ goto out_enomem;
+ }
+
+ if (!lpn) {
+ lpn = lnet_peer_net_alloc(net_id);
+ if (!lpn)
+ goto out_maybe_free_lp;
+ }
+
+ if (!lpni) {
+ lpni = lnet_peer_ni_alloc(nid);
+ if (!lpni)
+ goto out_maybe_free_lpn;
+ }
+
+ /* Install the new peer_ni */
+ lnet_net_lock(LNET_LOCK_EX);
+ /* Add peer_ni to global peer table hash, if necessary. */
+ if (list_empty(&lpni->lpni_hashlist)) {
+ ptable = the_lnet.ln_peer_tables[lpni->lpni_cpt];
+ list_add_tail(&lpni->lpni_hashlist,
+ &ptable->pt_hash[lnet_nid2peerhash(nid)]);
+ ptable->pt_version++;
+ atomic_inc(&ptable->pt_number);
+ atomic_inc(&lpni->lpni_refcount);
+ }
+
+ /* Detach the peer_ni from an existing peer, if necessary. */
+ if (lpni->lpni_peer_net && lpni->lpni_peer_net->lpn_peer != lp)
+ lnet_try_destroy_peer_hierarchy_locked(lpni);
+
+ /* Add peer_ni to peer_net */
+ lpni->lpni_peer_net = lpn;
+ list_add_tail(&lpni->lpni_on_peer_net_list, &lpn->lpn_peer_nis);
+
+ /* Add peer_net to peer */
+ if (!lpn->lpn_peer) {
+ lpn->lpn_peer = lp;
+ list_add_tail(&lpn->lpn_on_peer_list, &lp->lp_peer_nets);
+ }
+
+ /* Add peer to global peer list */
+ if (list_empty(&lp->lp_on_lnet_peer_list))
+ list_add_tail(&lp->lp_on_lnet_peer_list, &the_lnet.ln_peers);
+ lnet_net_unlock(LNET_LOCK_EX);
+
+ return 0;
+
+out_maybe_free_lpn:
+ if (list_empty(&lpn->lpn_on_peer_list))
+ LIBCFS_FREE(lpn, sizeof(*lpn));
+out_maybe_free_lp:
+ if (list_empty(&lp->lp_on_lnet_peer_list))
+ LIBCFS_FREE(lp, sizeof(*lp));
+out_enomem:
+ return -ENOMEM;
+}
+
+static int
+lnet_add_prim_lpni(lnet_nid_t nid)
+{
+ int rc;
struct lnet_peer *peer;
+ struct lnet_peer_ni *lpni;
- /* TODO: could the below situation happen? accessing an already
- * destroyed peer? */
- if (lpni->lpni_peer_net == NULL ||
- lpni->lpni_peer_net->lpn_peer == NULL)
- return;
+ LASSERT(nid != LNET_NID_ANY);
+
+ /*
+ * lookup the NID and its peer
+ * if the peer doesn't exist, create it.
+ * if this is a non-MR peer then change its state to MR and exit.
+ * if this is an MR peer and it's a primary NI: NO-OP.
+ * if this is an MR peer and it's not a primary NI. Operation not
+ * allowed.
+ *
+ * The adding and deleting of peer nis is being serialized through
+ * the api_mutex. So we can look up peers with the mutex locked
+ * safely. Only when we need to change the ptable, do we need to
+ * exclusively lock the lnet_net_lock()
+ */
+ lpni = lnet_find_peer_ni_locked(nid);
+ if (!lpni) {
+ rc = lnet_peer_setup_hierarchy(NULL, NULL, nid);
+ if (rc != 0)
+ return rc;
+ lpni = lnet_find_peer_ni_locked(nid);
+ }
+
+ LASSERT(lpni);
+
+ lnet_peer_ni_decref_locked(lpni);
- peer_net = lpni->lpni_peer_net;
peer = lpni->lpni_peer_net->lpn_peer;
- list_del_init(&lpni->lpni_on_peer_net_list);
- lpni->lpni_peer_net = NULL;
+ /*
+ * If we found a lpni with the same nid as the NID we're trying to
+ * create, then we're trying to create an already existing lpni
+ * that belongs to a different peer
+ */
+ if (peer->lp_primary_nid != nid)
+ return -EEXIST;
- /* if peer_net is empty, then remove it from the peer */
- if (list_empty(&peer_net->lpn_peer_nis)) {
- list_del_init(&peer_net->lpn_on_peer_list);
- peer_net->lpn_peer = NULL;
- LIBCFS_FREE(peer_net, sizeof(*peer_net));
+ /*
+ * if we found an lpni that is not a multi-rail, which could occur
+ * if lpni is already created as a non-mr lpni or we just created
+ * it, then make sure you indicate that this lpni is a primary mr
+ * capable peer.
+ *
+ * TODO: update flags if necessary
+ */
+ if (!peer->lp_multi_rail && peer->lp_primary_nid == nid)
+ peer->lp_multi_rail = true;
- /* if the peer is empty then remove it from the
- * the_lnet.ln_peers */
- if (list_empty(&peer->lp_peer_nets)) {
- list_del_init(&peer->lp_on_lnet_peer_list);
- LIBCFS_FREE(peer, sizeof(*peer));
+ return rc;
+}
+
+static int
+lnet_add_peer_ni_to_prim_lpni(lnet_nid_t key_nid, lnet_nid_t nid)
+{
+ struct lnet_peer *peer, *primary_peer;
+ struct lnet_peer_ni *lpni = NULL, *klpni = NULL;
+
+ LASSERT(key_nid != LNET_NID_ANY && nid != LNET_NID_ANY);
+
+ /*
+ * key nid must be created by this point. If not then this
+ * operation is not permitted
+ */
+ klpni = lnet_find_peer_ni_locked(key_nid);
+ if (!klpni)
+ return -ENOENT;
+
+ lnet_peer_ni_decref_locked(klpni);
+
+ primary_peer = klpni->lpni_peer_net->lpn_peer;
+
+ lpni = lnet_find_peer_ni_locked(nid);
+ if (lpni) {
+ lnet_peer_ni_decref_locked(lpni);
+
+ peer = lpni->lpni_peer_net->lpn_peer;
+ /*
+ * lpni already exists in the system but it belongs to
+ * a different peer. We can't re-added it
+ */
+ if (peer->lp_primary_nid != key_nid && peer->lp_multi_rail) {
+ CERROR("Cannot add NID %s owned by peer %s to peer %s\n",
+ libcfs_nid2str(lpni->lpni_nid),
+ libcfs_nid2str(peer->lp_primary_nid),
+ libcfs_nid2str(key_nid));
+ return -EEXIST;
+ } else if (peer->lp_primary_nid == key_nid) {
+ /*
+ * found a peer_ni that is already part of the
+ * peer. This is a no-op operation.
+ */
+ return 0;
}
+
+ /*
+ * TODO: else if (peer->lp_primary_nid != key_nid &&
+ * !peer->lp_multi_rail)
+ * peer is not an MR peer and it will be moved in the next
+ * step to klpni, so update its flags accordingly.
+ * lnet_move_peer_ni()
+ */
+
+ /*
+ * TODO: call lnet_update_peer() from here to update the
+ * flags. This is the case when the lpni you're trying to
+ * add is already part of the peer. This could've been
+ * added by the DD previously, so go ahead and do any
+ * updates to the state if necessary
+ */
+
}
+
+ /*
+ * When we get here we either have found an existing lpni, which
+ * we can switch to the new peer. Or we need to create one and
+ * add it to the new peer
+ */
+ return lnet_peer_setup_hierarchy(primary_peer, lpni, nid);
}
+/*
+ * lpni creation initiated due to traffic either sending or receiving.
+ */
static int
-lnet_build_peer_hierarchy(struct lnet_peer_ni *lpni)
+lnet_peer_ni_traffic_add(lnet_nid_t nid)
{
- struct lnet_peer *peer;
- struct lnet_peer_net *peer_net;
- __u32 lpni_net = LNET_NIDNET(lpni->lpni_nid);
+ struct lnet_peer_ni *lpni;
+ int rc = 0;
+
+ if (nid == LNET_NID_ANY)
+ return -EINVAL;
+
+ /* lnet_net_lock is not needed here because ln_api_lock is held */
+ lpni = lnet_find_peer_ni_locked(nid);
+ if (lpni) {
+ /*
+ * TODO: lnet_update_primary_nid() but not all of it
+ * only indicate if we're converting this to MR capable
+ * Can happen due to DD
+ */
+ lnet_peer_ni_decref_locked(lpni);
+ } else {
+ rc = lnet_peer_setup_hierarchy(NULL, NULL, nid);
+ }
- peer = NULL;
- peer_net = NULL;
+ return rc;
- LIBCFS_ALLOC(peer, sizeof(*peer));
- if (peer == NULL)
- return -ENOMEM;
+}
- LIBCFS_ALLOC(peer_net, sizeof(*peer_net));
- if (peer_net == NULL) {
- LIBCFS_FREE(peer, sizeof(*peer));
- return -ENOMEM;
+static int
+lnet_peer_ni_add_non_mr(lnet_nid_t nid)
+{
+ struct lnet_peer_ni *lpni;
+
+ lpni = lnet_find_peer_ni_locked(nid);
+ if (lpni) {
+ CERROR("Cannot add %s as non-mr when it already exists\n",
+ libcfs_nid2str(nid));
+ lnet_peer_ni_decref_locked(lpni);
+ return -EEXIST;
}
- INIT_LIST_HEAD(&peer->lp_on_lnet_peer_list);
- INIT_LIST_HEAD(&peer->lp_peer_nets);
- INIT_LIST_HEAD(&peer_net->lpn_on_peer_list);
- INIT_LIST_HEAD(&peer_net->lpn_peer_nis);
+ return lnet_peer_setup_hierarchy(NULL, NULL, nid);
+}
- /* build the hierarchy */
- peer_net->lpn_net_id = lpni_net;
- peer_net->lpn_peer = peer;
- lpni->lpni_peer_net = peer_net;
- peer->lp_primary_nid = lpni->lpni_nid;
- list_add_tail(&peer_net->lpn_on_peer_list, &peer->lp_peer_nets);
- list_add_tail(&lpni->lpni_on_peer_net_list, &peer_net->lpn_peer_nis);
- list_add_tail(&peer->lp_on_lnet_peer_list, &the_lnet.ln_peers);
+/*
+ * This API handles the following combinations:
+ * Create a primary NI if only the key_nid is provided
+ * Create or add an lpni to a primary NI. Primary NI must've already
+ * been created
+ * Create a non-MR peer.
+ */
+int
+lnet_add_peer_ni_to_peer(lnet_nid_t key_nid, lnet_nid_t nid, bool mr)
+{
+ /*
+ * Caller trying to setup an MR like peer hierarchy but
+ * specifying it to be non-MR. This is not allowed.
+ */
+ if (key_nid != LNET_NID_ANY &&
+ nid != LNET_NID_ANY && !mr)
+ return -EPERM;
+
+ /* Add the primary NID of a peer */
+ if (key_nid != LNET_NID_ANY &&
+ nid == LNET_NID_ANY && mr)
+ return lnet_add_prim_lpni(key_nid);
+
+ /* Add a NID to an existing peer */
+ if (key_nid != LNET_NID_ANY &&
+ nid != LNET_NID_ANY && mr)
+ return lnet_add_peer_ni_to_prim_lpni(key_nid, nid);
+
+ /* Add a non-MR peer NI */
+ if (((key_nid != LNET_NID_ANY &&
+ nid == LNET_NID_ANY) ||
+ (key_nid == LNET_NID_ANY &&
+ nid != LNET_NID_ANY)) && !mr)
+ return lnet_peer_ni_add_non_mr(key_nid != LNET_NID_ANY ?
+ key_nid : nid);
+
+ return 0;
+}
+
+int
+lnet_del_peer_ni_from_peer(lnet_nid_t key_nid, lnet_nid_t nid)
+{
+ lnet_nid_t local_nid;
+ struct lnet_peer *peer;
+ struct lnet_peer_ni *lpni;
+
+ if (key_nid == LNET_NID_ANY)
+ return -EINVAL;
+
+ local_nid = (nid != LNET_NID_ANY) ? nid : key_nid;
+
+ lpni = lnet_find_peer_ni_locked(local_nid);
+ if (!lpni)
+ return -EINVAL;
+ lnet_peer_ni_decref_locked(lpni);
+
+ peer = lpni->lpni_peer_net->lpn_peer;
+ LASSERT(peer != NULL);
+
+ if (peer->lp_primary_nid == lpni->lpni_nid) {
+ /*
+ * deleting the primary ni is equivalent to deleting the
+ * entire peer
+ */
+ lnet_net_lock(LNET_LOCK_EX);
+ lnet_peer_del_locked(peer);
+ lnet_net_unlock(LNET_LOCK_EX);
+
+ return 0;
+ }
+
+ lnet_net_lock(LNET_LOCK_EX);
+ lnet_peer_ni_del_locked(lpni);
+ lnet_net_unlock(LNET_LOCK_EX);
return 0;
}
LASSERT(atomic_read(&lpni->lpni_refcount) == 0);
LASSERT(lpni->lpni_rtr_refcount == 0);
LASSERT(list_empty(&lpni->lpni_txq));
- LASSERT(list_empty(&lpni->lpni_hashlist));
LASSERT(lpni->lpni_txqnob == 0);
- LASSERT(lpni->lpni_peer_net != NULL);
- LASSERT(lpni->lpni_peer_net->lpn_peer != NULL);
-
- ptable = the_lnet.ln_peer_tables[lpni->lpni_cpt];
- LASSERT(ptable->pt_number > 0);
- ptable->pt_number--;
lpni->lpni_net = NULL;
- lnet_try_destroy_peer_hierarchy_locked(lpni);
+ /* remove the peer ni from the zombie list */
+ ptable = the_lnet.ln_peer_tables[lpni->lpni_cpt];
+ spin_lock(&ptable->pt_zombie_lock);
+ list_del_init(&lpni->lpni_hashlist);
+ ptable->pt_zombies--;
+ spin_unlock(&ptable->pt_zombie_lock);
LIBCFS_FREE(lpni, sizeof(*lpni));
-
- LASSERT(ptable->pt_zombies > 0);
- ptable->pt_zombies--;
}
-int
-lnet_nid2peerni_locked(struct lnet_peer_ni **lpnip, lnet_nid_t nid, int cpt)
+struct lnet_peer_ni *
+lnet_nid2peerni_locked(lnet_nid_t nid, int cpt)
{
struct lnet_peer_table *ptable;
struct lnet_peer_ni *lpni = NULL;
- struct lnet_peer_ni *lpni2;
int cpt2;
- int rc = 0;
+ int rc;
- *lpnip = NULL;
if (the_lnet.ln_shutdown) /* it's shutting down */
- return -ESHUTDOWN;
+ return ERR_PTR(-ESHUTDOWN);
/*
* calculate cpt2 with the standard hash function
- * This cpt2 becomes the slot where we'll find or create the peer.
+ * This cpt2 is the slot where we'll find or create the peer.
*/
cpt2 = lnet_nid_cpt_hash(nid, LNET_CPT_NUMBER);
-
- /*
- * Any changes to the peer tables happen under exclusive write
- * lock. Any reads to the peer tables can be done via a standard
- * CPT read lock.
- */
- if (cpt != LNET_LOCK_EX) {
- lnet_net_unlock(cpt);
- lnet_net_lock(LNET_LOCK_EX);
- }
-
ptable = the_lnet.ln_peer_tables[cpt2];
lpni = lnet_get_peer_ni_locked(ptable, nid);
- if (lpni != NULL) {
- *lpnip = lpni;
- if (cpt != LNET_LOCK_EX) {
- lnet_net_unlock(LNET_LOCK_EX);
- lnet_net_lock(cpt);
- }
- return 0;
- }
+ if (lpni)
+ return lpni;
+ /* Slow path: serialized using the ln_api_mutex. */
+ lnet_net_unlock(cpt);
+ mutex_lock(&the_lnet.ln_api_mutex);
/*
- * take extra refcount in case another thread has shutdown LNet
- * and destroyed locks and peer-table before I finish the allocation
+ * Shutdown is only set under the ln_api_lock, so a single
+ * check here is sufficent.
+ *
+ * lnet_add_nid_to_peer() also handles the case where we've
+ * raced and a different thread added the NID.
*/
- ptable->pt_number++;
- lnet_net_unlock(LNET_LOCK_EX);
-
- LIBCFS_CPT_ALLOC(lpni, lnet_cpt_table(), cpt2, sizeof(*lpni));
-
- if (lpni == NULL) {
- rc = -ENOMEM;
- lnet_net_lock(cpt);
- goto out;
- }
-
- INIT_LIST_HEAD(&lpni->lpni_txq);
- INIT_LIST_HEAD(&lpni->lpni_rtrq);
- INIT_LIST_HEAD(&lpni->lpni_routes);
-
- lpni->lpni_alive = !lnet_peers_start_down(); /* 1 bit!! */
- lpni->lpni_last_alive = cfs_time_current(); /* assumes alive */
- lpni->lpni_ping_feats = LNET_PING_FEAT_INVAL;
- lpni->lpni_nid = nid;
- lpni->lpni_cpt = cpt2;
- atomic_set(&lpni->lpni_refcount, 2); /* 1 for caller; 1 for hash */
-
- rc = lnet_build_peer_hierarchy(lpni);
- if (rc != 0)
- goto out;
-
- lnet_net_lock(LNET_LOCK_EX);
-
if (the_lnet.ln_shutdown) {
- rc = -ESHUTDOWN;
- goto out;
+ lpni = ERR_PTR(-ESHUTDOWN);
+ goto out_mutex_unlock;
}
- lpni2 = lnet_get_peer_ni_locked(ptable, nid);
- if (lpni2 != NULL) {
- *lpnip = lpni2;
- goto out;
+ rc = lnet_peer_ni_traffic_add(nid);
+ if (rc) {
+ lpni = ERR_PTR(rc);
+ goto out_mutex_unlock;
}
- lpni->lpni_net = lnet_get_net_locked(LNET_NIDNET(lpni->lpni_nid));
- lpni->lpni_txcredits =
- lpni->lpni_mintxcredits =
- lpni->lpni_net->net_tunables.lct_peer_tx_credits;
- lpni->lpni_rtrcredits =
- lpni->lpni_minrtrcredits =
- lnet_peer_buffer_credits(lpni->lpni_net);
-
- list_add_tail(&lpni->lpni_hashlist,
- &ptable->pt_hash[lnet_nid2peerhash(nid)]);
- ptable->pt_version++;
- *lpnip = lpni;
+ lpni = lnet_get_peer_ni_locked(ptable, nid);
+ LASSERT(lpni);
- if (cpt != LNET_LOCK_EX) {
- lnet_net_unlock(LNET_LOCK_EX);
- lnet_net_lock(cpt);
- }
+out_mutex_unlock:
+ mutex_unlock(&the_lnet.ln_api_mutex);
+ lnet_net_lock(cpt);
- return 0;
-out:
- if (lpni != NULL) {
- lnet_try_destroy_peer_hierarchy_locked(lpni);
- LIBCFS_FREE(lpni, sizeof(*lpni));
- }
- ptable->pt_number--;
- if (cpt != LNET_LOCK_EX) {
- lnet_net_unlock(LNET_LOCK_EX);
- lnet_net_lock(cpt);
- }
- return rc;
+ return lpni;
}
void
{
char *aliveness = "NA";
struct lnet_peer_ni *lp;
- int rc;
int cpt;
cpt = lnet_cpt_of_nid(nid, NULL);
lnet_net_lock(cpt);
- rc = lnet_nid2peerni_locked(&lp, nid, cpt);
- if (rc != 0) {
+ lp = lnet_nid2peerni_locked(nid, cpt);
+ if (IS_ERR(lp)) {
lnet_net_unlock(cpt);
CDEBUG(D_WARNING, "No peer %s\n", libcfs_nid2str(nid));
return;
lnet_net_unlock(cpt);
}
-int lnet_get_peer_info(__u32 peer_index, __u64 *nid,
- char aliveness[LNET_MAX_STR_LEN],
- __u32 *cpt_iter, __u32 *refcount,
- __u32 *ni_peer_tx_credits, __u32 *peer_tx_credits,
- __u32 *peer_rtr_credits, __u32 *peer_min_rtr_credits,
- __u32 *peer_tx_qnob)
+int lnet_get_peer_ni_info(__u32 peer_index, __u64 *nid,
+ char aliveness[LNET_MAX_STR_LEN],
+ __u32 *cpt_iter, __u32 *refcount,
+ __u32 *ni_peer_tx_credits, __u32 *peer_tx_credits,
+ __u32 *peer_rtr_credits, __u32 *peer_min_rtr_credits,
+ __u32 *peer_tx_qnob)
{
struct lnet_peer_table *peer_table;
struct lnet_peer_ni *lp;
return found ? 0 : -ENOENT;
}
+
+int lnet_get_peer_info(__u32 idx, lnet_nid_t *primary_nid, lnet_nid_t *nid,
+ bool *mr, struct lnet_peer_ni_credit_info *peer_ni_info,
+ struct lnet_ioctl_element_stats *peer_ni_stats)
+{
+ struct lnet_peer_ni *lpni = NULL;
+ struct lnet_peer_net *lpn = NULL;
+ struct lnet_peer *lp = NULL;
+
+ lpni = lnet_get_peer_ni_idx_locked(idx, &lpn, &lp);
+
+ if (!lpni)
+ return -ENOENT;
+
+ *primary_nid = lp->lp_primary_nid;
+ *mr = lp->lp_multi_rail;
+ *nid = lpni->lpni_nid;
+ snprintf(peer_ni_info->cr_aliveness, LNET_MAX_STR_LEN, "NA");
+ if (lnet_isrouter(lpni) ||
+ lnet_peer_aliveness_enabled(lpni))
+ snprintf(peer_ni_info->cr_aliveness, LNET_MAX_STR_LEN,
+ lpni->lpni_alive ? "up" : "down");
+
+ peer_ni_info->cr_refcount = atomic_read(&lpni->lpni_refcount);
+ peer_ni_info->cr_ni_peer_tx_credits = (lpni->lpni_net != NULL) ?
+ lpni->lpni_net->net_tunables.lct_peer_tx_credits : 0;
+ peer_ni_info->cr_peer_tx_credits = lpni->lpni_txcredits;
+ peer_ni_info->cr_peer_rtr_credits = lpni->lpni_rtrcredits;
+ peer_ni_info->cr_peer_min_rtr_credits = lpni->lpni_mintxcredits;
+ peer_ni_info->cr_peer_tx_qnob = lpni->lpni_txqnob;
+
+ peer_ni_stats->send_count = atomic_read(&lpni->lpni_stats.send_count);
+ peer_ni_stats->recv_count = atomic_read(&lpni->lpni_stats.recv_count);
+ peer_ni_stats->drop_count = atomic_read(&lpni->lpni_stats.drop_count);
+
+ return 0;
+}
git://git.whamcloud.com / fs / lustre-release.git / blobdiff