fps = net->ibn_fmr_ps[cpt];
rc = kiblnd_fmr_pool_map(fps, tx, rd, nob, 0, &tx->tx_fmr);
if (rc != 0) {
- CERROR("Can't map %u pages: %d\n", nob, rc);
+ CERROR("Can't map %u bytes (%u/%u)s: %d\n", nob,
+ tx->tx_nfrags, rd->rd_nfrags, rc);
return rc;
}
return -EINVAL;
}
-static int kiblnd_setup_rd_iov(struct lnet_ni *ni, struct kib_tx *tx,
- struct kib_rdma_desc *rd, unsigned int niov,
- struct kvec *iov, int offset, int nob)
-{
- struct kib_net *net = ni->ni_data;
- struct page *page;
- struct scatterlist *sg;
- unsigned long vaddr;
- int fragnob;
- int page_offset;
- unsigned int max_niov;
-
- LASSERT (nob > 0);
- LASSERT (niov > 0);
- LASSERT (net != NULL);
-
- while (offset >= iov->iov_len) {
- offset -= iov->iov_len;
- niov--;
- iov++;
- LASSERT (niov > 0);
- }
-
- max_niov = niov;
-
- sg = tx->tx_frags;
- do {
- LASSERT(niov > 0);
-
- vaddr = ((unsigned long)iov->iov_base) + offset;
- page_offset = vaddr & (PAGE_SIZE - 1);
- page = lnet_kvaddr_to_page(vaddr);
- if (page == NULL) {
- CERROR("Can't find page\n");
- return -EFAULT;
- }
-
- fragnob = min((int)(iov->iov_len - offset), nob);
- fragnob = min(fragnob, (int)PAGE_SIZE - page_offset);
-
- /*
- * We're allowed to start at a non-aligned page offset in
- * the first fragment and end at a non-aligned page offset
- * in the last fragment.
- */
- if ((fragnob < (int)PAGE_SIZE - page_offset) &&
- (niov < max_niov) && nob > fragnob) {
- CDEBUG(D_NET, "fragnob %d < available page %d: with"
- " remaining %d iovs with %d nob left\n",
- fragnob, (int)PAGE_SIZE - page_offset, niov,
- nob);
- tx->tx_gaps = true;
- }
-
- sg_set_page(sg, page, fragnob, page_offset);
- sg = sg_next(sg);
- if (!sg) {
- CERROR("lacking enough sg entries to map tx\n");
- return -EFAULT;
- }
-
- if (offset + fragnob < iov->iov_len) {
- offset += fragnob;
- } else {
- offset = 0;
- iov++;
- niov--;
- }
- nob -= fragnob;
- } while (nob > 0);
-
- return kiblnd_map_tx(ni, tx, rd, sg - tx->tx_frags);
-}
-
static int kiblnd_setup_rd_kiov(struct lnet_ni *ni, struct kib_tx *tx,
struct kib_rdma_desc *rd, int nkiov,
- lnet_kiov_t *kiov, int offset, int nob)
+ struct bio_vec *kiov, int offset, int nob)
{
struct kib_net *net = ni->ni_data;
struct scatterlist *sg;
LASSERT(nkiov > 0);
LASSERT(net != NULL);
- while (offset >= kiov->kiov_len) {
- offset -= kiov->kiov_len;
+ while (offset >= kiov->bv_len) {
+ offset -= kiov->bv_len;
nkiov--;
kiov++;
LASSERT(nkiov > 0);
do {
LASSERT(nkiov > 0);
- fragnob = min((int)(kiov->kiov_len - offset), nob);
+ fragnob = min((int)(kiov->bv_len - offset), nob);
/*
* We're allowed to start at a non-aligned page offset in
* the first fragment and end at a non-aligned page offset
* in the last fragment.
*/
- if ((fragnob < (int)(kiov->kiov_len - offset)) &&
+ if ((fragnob < (int)(kiov->bv_len - offset)) &&
nkiov < max_nkiov && nob > fragnob) {
CDEBUG(D_NET, "fragnob %d < available page %d: with"
" remaining %d kiovs with %d nob left\n",
- fragnob, (int)(kiov->kiov_len - offset),
+ fragnob, (int)(kiov->bv_len - offset),
nkiov, nob);
tx->tx_gaps = true;
}
- sg_set_page(sg, kiov->kiov_page, fragnob,
- kiov->kiov_offset + offset);
+ sg_set_page(sg, kiov->bv_page, fragnob,
+ kiov->bv_offset + offset);
sg = sg_next(sg);
if (!sg) {
CERROR("lacking enough sg entries to map tx\n");
!list_empty(&conn->ibc_tx_queue_rsrvd)) {
tx = list_entry(conn->ibc_tx_queue_rsrvd.next,
struct kib_tx, tx_list);
- list_del(&tx->tx_list);
- list_add_tail(&tx->tx_list, &conn->ibc_tx_queue);
+ list_move_tail(&tx->tx_list, &conn->ibc_tx_queue);
conn->ibc_reserved_credits--;
}
static void
kiblnd_tx_complete(struct kib_tx *tx, int status)
{
- int failed = (status != IB_WC_SUCCESS);
+ int failed = (status != IB_WC_SUCCESS);
struct kib_conn *conn = tx->tx_conn;
- int idle;
+ int idle;
- LASSERT (tx->tx_sending > 0);
+ if (tx->tx_sending <= 0) {
+ CERROR("Received an event on a freed tx: %p status %d\n",
+ tx, tx->tx_status);
+ return;
+ }
- if (failed) {
- if (conn->ibc_state == IBLND_CONN_ESTABLISHED)
+ if (failed) {
+ if (conn->ibc_state == IBLND_CONN_ESTABLISHED)
CNETERR("Tx -> %s cookie %#llx"
- " sending %d waiting %d: failed %d\n",
- libcfs_nid2str(conn->ibc_peer->ibp_nid),
- tx->tx_cookie, tx->tx_sending, tx->tx_waiting,
- status);
+ " sending %d waiting %d: failed %d\n",
+ libcfs_nid2str(conn->ibc_peer->ibp_nid),
+ tx->tx_cookie, tx->tx_sending, tx->tx_waiting,
+ status);
- kiblnd_close_conn(conn, -EIO);
- } else {
- kiblnd_peer_alive(conn->ibc_peer);
- }
+ kiblnd_close_conn(conn, -EIO);
+ } else {
+ kiblnd_peer_alive(conn->ibc_peer);
+ }
spin_lock(&conn->ibc_lock);
break;
}
- sge_nob = MIN(MIN(kiblnd_rd_frag_size(srcrd, srcidx),
- kiblnd_rd_frag_size(dstrd, dstidx)), resid);
+ sge_nob = min3(kiblnd_rd_frag_size(srcrd, srcidx),
+ kiblnd_rd_frag_size(dstrd, dstidx),
+ resid);
sge = &tx->tx_sge[tx->tx_nsge];
sge->addr = kiblnd_rd_frag_addr(srcrd, srcidx);
return;
}
- timeout_ns = lnet_get_lnd_timeout() * NSEC_PER_SEC;
+ timeout_ns = kiblnd_timeout() * NSEC_PER_SEC;
tx->tx_queued = 1;
tx->tx_deadline = ktime_add_ns(ktime_get(), timeout_ns);
spin_unlock(&conn->ibc_lock);
}
-static int kiblnd_resolve_addr(struct rdma_cm_id *cmid,
- struct sockaddr_in *srcaddr,
- struct sockaddr_in *dstaddr,
- int timeout_ms)
+static int
+kiblnd_resolve_addr_cap(struct rdma_cm_id *cmid,
+ struct sockaddr_in *srcaddr,
+ struct sockaddr_in *dstaddr,
+ int timeout_ms)
{
unsigned short port;
int rc;
+ LASSERT(capable(CAP_NET_BIND_SERVICE));
+
/* allow the port to be reused */
rc = rdma_set_reuseaddr(cmid, 1);
if (rc != 0) {
return rc;
}
+static int
+kiblnd_resolve_addr(struct rdma_cm_id *cmid,
+ struct sockaddr_in *srcaddr,
+ struct sockaddr_in *dstaddr,
+ int timeout_ms)
+{
+ const struct cred *old_creds = NULL;
+ struct cred *new_creds;
+ int rc;
+
+ if (!capable(CAP_NET_BIND_SERVICE)) {
+ new_creds = prepare_creds();
+ if (!new_creds)
+ return -ENOMEM;
+
+ cap_raise(new_creds->cap_effective, CAP_NET_BIND_SERVICE);
+ old_creds = override_creds(new_creds);
+ }
+
+ rc = kiblnd_resolve_addr_cap(cmid, srcaddr, dstaddr, timeout_ms);
+
+ if (old_creds)
+ revert_creds(old_creds);
+
+ return rc;
+}
+
static void
kiblnd_connect_peer(struct kib_peer_ni *peer_ni)
{
if (*kiblnd_tunables.kib_use_priv_port) {
rc = kiblnd_resolve_addr(cmid, &srcaddr, &dstaddr,
- lnet_get_lnd_timeout() * 1000);
+ kiblnd_timeout() * 1000);
} else {
rc = rdma_resolve_addr(cmid,
(struct sockaddr *)&srcaddr,
(struct sockaddr *)&dstaddr,
- lnet_get_lnd_timeout() * 1000);
+ kiblnd_timeout() * 1000);
}
if (rc != 0) {
/* Can't initiate address resolution: */
bool
kiblnd_reconnect_peer(struct kib_peer_ni *peer_ni)
{
- rwlock_t *glock = &kiblnd_data.kib_global_lock;
- char *reason = NULL;
- struct list_head txs;
- unsigned long flags;
-
- INIT_LIST_HEAD(&txs);
+ rwlock_t *glock = &kiblnd_data.kib_global_lock;
+ char *reason = NULL;
+ LIST_HEAD(txs);
+ unsigned long flags;
write_lock_irqsave(glock, flags);
if (peer_ni->ibp_reconnecting == 0) {
int target_is_router = lntmsg->msg_target_is_router;
int routing = lntmsg->msg_routing;
unsigned int payload_niov = lntmsg->msg_niov;
- struct kvec *payload_iov = lntmsg->msg_iov;
- lnet_kiov_t *payload_kiov = lntmsg->msg_kiov;
+ struct bio_vec *payload_kiov = lntmsg->msg_kiov;
unsigned int payload_offset = lntmsg->msg_offset;
unsigned int payload_nob = lntmsg->msg_len;
struct kib_msg *ibmsg;
/* Thread context */
LASSERT (!in_interrupt());
- /* payload is either all vaddrs or all pages */
- LASSERT (!(payload_kiov != NULL && payload_iov != NULL));
switch (type) {
default:
ibmsg = tx->tx_msg;
rd = &ibmsg->ibm_u.get.ibgm_rd;
- if ((lntmsg->msg_md->md_options & LNET_MD_KIOV) == 0)
- rc = kiblnd_setup_rd_iov(ni, tx, rd,
- lntmsg->msg_md->md_niov,
- lntmsg->msg_md->md_iov.iov,
- 0, lntmsg->msg_md->md_length);
- else
- rc = kiblnd_setup_rd_kiov(ni, tx, rd,
- lntmsg->msg_md->md_niov,
- lntmsg->msg_md->md_iov.kiov,
- 0, lntmsg->msg_md->md_length);
+ rc = kiblnd_setup_rd_kiov(ni, tx, rd,
+ lntmsg->msg_md->md_niov,
+ lntmsg->msg_md->md_kiov,
+ 0, lntmsg->msg_md->md_length);
if (rc != 0) {
CERROR("Can't setup GET sink for %s: %d\n",
libcfs_nid2str(target.nid), rc);
return -ENOMEM;
}
- if (payload_kiov == NULL)
- rc = kiblnd_setup_rd_iov(ni, tx, tx->tx_rd,
- payload_niov, payload_iov,
- payload_offset, payload_nob);
- else
- rc = kiblnd_setup_rd_kiov(ni, tx, tx->tx_rd,
- payload_niov, payload_kiov,
- payload_offset, payload_nob);
+ rc = kiblnd_setup_rd_kiov(ni, tx, tx->tx_rd,
+ payload_niov, payload_kiov,
+ payload_offset, payload_nob);
if (rc != 0) {
CERROR("Can't setup PUT src for %s: %d\n",
libcfs_nid2str(target.nid), rc);
ibmsg = tx->tx_msg;
ibmsg->ibm_u.immediate.ibim_hdr = *hdr;
- if (payload_kiov != NULL)
- lnet_copy_kiov2flat(IBLND_MSG_SIZE, ibmsg,
- offsetof(struct kib_msg, ibm_u.immediate.ibim_payload),
- payload_niov, payload_kiov,
- payload_offset, payload_nob);
- else
- lnet_copy_iov2flat(IBLND_MSG_SIZE, ibmsg,
- offsetof(struct kib_msg, ibm_u.immediate.ibim_payload),
- payload_niov, payload_iov,
- payload_offset, payload_nob);
+ lnet_copy_kiov2flat(IBLND_MSG_SIZE, ibmsg,
+ offsetof(struct kib_msg,
+ ibm_u.immediate.ibim_payload),
+ payload_niov, payload_kiov,
+ payload_offset, payload_nob);
nob = offsetof(struct kib_immediate_msg, ibim_payload[payload_nob]);
kiblnd_init_tx_msg(ni, tx, IBLND_MSG_IMMEDIATE, nob);
kiblnd_reply(struct lnet_ni *ni, struct kib_rx *rx, struct lnet_msg *lntmsg)
{
struct lnet_process_id target = lntmsg->msg_target;
- unsigned int niov = lntmsg->msg_niov;
- struct kvec *iov = lntmsg->msg_iov;
- lnet_kiov_t *kiov = lntmsg->msg_kiov;
- unsigned int offset = lntmsg->msg_offset;
- unsigned int nob = lntmsg->msg_len;
+ unsigned int niov = lntmsg->msg_niov;
+ struct bio_vec *kiov = lntmsg->msg_kiov;
+ unsigned int offset = lntmsg->msg_offset;
+ unsigned int nob = lntmsg->msg_len;
struct kib_tx *tx;
- int rc;
+ int rc;
tx = kiblnd_get_idle_tx(ni, rx->rx_conn->ibc_peer->ibp_nid);
if (tx == NULL) {
if (nob == 0)
rc = 0;
- else if (kiov == NULL)
- rc = kiblnd_setup_rd_iov(ni, tx, tx->tx_rd,
- niov, iov, offset, nob);
else
rc = kiblnd_setup_rd_kiov(ni, tx, tx->tx_rd,
niov, kiov, offset, nob);
int
kiblnd_recv(struct lnet_ni *ni, void *private, struct lnet_msg *lntmsg,
- int delayed, unsigned int niov, struct kvec *iov, lnet_kiov_t *kiov,
+ int delayed, unsigned int niov, struct bio_vec *kiov,
unsigned int offset, unsigned int mlen, unsigned int rlen)
{
struct kib_rx *rx = private;
LASSERT (mlen <= rlen);
LASSERT (!in_interrupt());
- /* Either all pages or all vaddrs */
- LASSERT (!(kiov != NULL && iov != NULL));
switch (rxmsg->ibm_type) {
default:
break;
}
- if (kiov != NULL)
- lnet_copy_flat2kiov(niov, kiov, offset,
- IBLND_MSG_SIZE, rxmsg,
- offsetof(struct kib_msg, ibm_u.immediate.ibim_payload),
- mlen);
- else
- lnet_copy_flat2iov(niov, iov, offset,
- IBLND_MSG_SIZE, rxmsg,
- offsetof(struct kib_msg, ibm_u.immediate.ibim_payload),
- mlen);
+ lnet_copy_flat2kiov(niov, kiov, offset,
+ IBLND_MSG_SIZE, rxmsg,
+ offsetof(struct kib_msg,
+ ibm_u.immediate.ibim_payload),
+ mlen);
lnet_finalize(lntmsg, 0);
break;
txmsg = tx->tx_msg;
rd = &txmsg->ibm_u.putack.ibpam_rd;
- if (kiov == NULL)
- rc = kiblnd_setup_rd_iov(ni, tx, rd,
- niov, iov, offset, mlen);
- else
- rc = kiblnd_setup_rd_kiov(ni, tx, rd,
- niov, kiov, offset, mlen);
+ rc = kiblnd_setup_rd_kiov(ni, tx, rd,
+ niov, kiov, offset, mlen);
if (rc != 0) {
CERROR("Can't setup PUT sink for %s: %d\n",
libcfs_nid2str(conn->ibc_peer->ibp_nid), rc);
void
kiblnd_abort_txs(struct kib_conn *conn, struct list_head *txs)
{
- struct list_head zombies = LIST_HEAD_INIT(zombies);
- struct list_head *tmp;
- struct list_head *nxt;
+ LIST_HEAD(zombies);
+ struct kib_tx *nxt;
struct kib_tx *tx;
spin_lock(&conn->ibc_lock);
- list_for_each_safe(tmp, nxt, txs) {
- tx = list_entry(tmp, struct kib_tx, tx_list);
-
+ list_for_each_entry_safe(tx, nxt, txs, tx_list) {
if (txs == &conn->ibc_active_txs) {
LASSERT(!tx->tx_queued);
LASSERT(tx->tx_waiting ||
*/
if (tx->tx_sending == 0) {
tx->tx_queued = 0;
- list_del(&tx->tx_list);
- list_add(&tx->tx_list, &zombies);
+ list_move(&tx->tx_list, &zombies);
}
}
kiblnd_peer_connect_failed(struct kib_peer_ni *peer_ni, int active,
int error)
{
- struct list_head zombies = LIST_HEAD_INIT(zombies);
+ LIST_HEAD(zombies);
unsigned long flags;
LASSERT (error != 0);
{
struct kib_peer_ni *peer_ni = conn->ibc_peer;
struct kib_tx *tx;
- struct list_head txs;
+ LIST_HEAD(txs);
unsigned long flags;
int active;
(conn->ibc_state == IBLND_CONN_PASSIVE_WAIT &&
peer_ni->ibp_accepting > 0));
- LIBCFS_FREE(conn->ibc_connvars, sizeof(*conn->ibc_connvars));
- conn->ibc_connvars = NULL;
+ LIBCFS_FREE(conn->ibc_connvars, sizeof(*conn->ibc_connvars));
+ conn->ibc_connvars = NULL;
- if (status != 0) {
- /* failed to establish connection */
- kiblnd_peer_connect_failed(peer_ni, active, status);
- kiblnd_finalise_conn(conn);
- return;
- }
+ if (status != 0) {
+ /* failed to establish connection */
+ kiblnd_peer_connect_failed(peer_ni, active, status);
+ kiblnd_finalise_conn(conn);
+ return;
+ }
- /* connection established */
+ /* connection established */
write_lock_irqsave(&kiblnd_data.kib_global_lock, flags);
+ /* reset retry count */
+ peer_ni->ibp_retries = 0;
+
conn->ibc_last_send = ktime_get();
- kiblnd_set_conn_state(conn, IBLND_CONN_ESTABLISHED);
- kiblnd_peer_alive(peer_ni);
+ kiblnd_set_conn_state(conn, IBLND_CONN_ESTABLISHED);
+ kiblnd_peer_alive(peer_ni);
/* Add conn to peer_ni's list and nuke any dangling conns from a different
* peer_ni instance... */
}
/* grab pending txs while I have the lock */
- INIT_LIST_HEAD(&txs);
list_splice_init(&peer_ni->ibp_tx_queue, &txs);
if (!kiblnd_peer_active(peer_ni) || /* peer_ni has been deleted */
{
int rc;
+#ifdef HAVE_RDMA_REJECT_4ARGS
+ rc = rdma_reject(cmid, rej, sizeof(*rej), IB_CM_REJ_CONSUMER_DEFINED);
+#else
rc = rdma_reject(cmid, rej, sizeof(*rej));
+#endif
if (rc != 0)
CWARN("Error %d sending reject\n", rc);
}
write_lock_irqsave(glock, flags);
- /* retry connection if it's still needed and no other connection
- * attempts (active or passive) are in progress
- * NB: reconnect is still needed even when ibp_tx_queue is
- * empty if ibp_version != version because reconnect may be
- * initiated by kiblnd_query() */
+ /* retry connection if it's still needed and no other connection
+ * attempts (active or passive) are in progress
+ * NB: reconnect is still needed even when ibp_tx_queue is
+ * empty if ibp_version != version because reconnect may be
+ * initiated.
+ */
reconnect = (!list_empty(&peer_ni->ibp_tx_queue) ||
peer_ni->ibp_version != version) &&
peer_ni->ibp_connecting &&
goto out;
}
- switch (why) {
- default:
- reason = "Unknown";
- break;
+ if (peer_ni->ibp_retries > *kiblnd_tunables.kib_retry_count) {
+ reason = "retry count exceeded due to no listener";
+ goto out;
+ }
+
+ switch (why) {
+ default:
+ reason = "Unknown";
+ break;
case IBLND_REJECT_RDMA_FRAGS: {
struct lnet_ioctl_config_o2iblnd_tunables *tunables;
IBLND_REJECT_CONN_STALE, NULL);
break;
- case IB_CM_REJ_INVALID_SERVICE_ID:
+ case IB_CM_REJ_INVALID_SERVICE_ID:
+ peer_ni->ibp_retries++;
kiblnd_check_reconnect(conn, IBLND_MSG_VERSION, 0,
IBLND_REJECT_INVALID_SRV_ID, NULL);
- CNETERR("%s rejected: no listener at %d\n",
- libcfs_nid2str(peer_ni->ibp_nid),
- *kiblnd_tunables.kib_service);
- break;
+ CNETERR("%s rejected: no listener at %d\n",
+ libcfs_nid2str(peer_ni->ibp_nid),
+ *kiblnd_tunables.kib_service);
+ break;
case IB_CM_REJ_CONSUMER_DEFINED:
if (priv_nob >= offsetof(struct kib_rej, ibr_padding)) {
rc = event->status;
} else {
rc = rdma_resolve_route(
- cmid, lnet_get_lnd_timeout() * 1000);
+ cmid, kiblnd_timeout() * 1000);
if (rc == 0) {
struct kib_net *net = peer_ni->ibp_ni->ni_data;
struct kib_dev *dev = net->ibn_dev;
static void
kiblnd_check_conns (int idx)
{
- struct list_head closes = LIST_HEAD_INIT(closes);
- struct list_head checksends = LIST_HEAD_INIT(checksends);
- struct list_head timedout_txs = LIST_HEAD_INIT(timedout_txs);
+ LIST_HEAD(closes);
+ LIST_HEAD(checksends);
+ LIST_HEAD(timedout_txs);
struct list_head *peers = &kiblnd_data.kib_peers[idx];
struct list_head *ptmp;
struct kib_peer_ni *peer_ni;
int peer_index = 0;
unsigned long deadline = jiffies;
- cfs_block_allsigs();
-
init_waitqueue_entry(&wait, current);
kiblnd_data.kib_connd = current;
* connection within (n+1)/n times the timeout
* interval. */
- lnd_timeout = lnet_get_lnd_timeout();
+ lnd_timeout = kiblnd_timeout();
if (lnd_timeout > n * p)
chunk = (chunk * n * p) / lnd_timeout;
if (chunk == 0)
schedule_timeout(timeout);
- set_current_state(TASK_RUNNING);
remove_wait_queue(&kiblnd_data.kib_connd_waitq, &wait);
spin_lock_irqsave(lock, flags);
}
unsigned long flags;
struct ib_wc wc;
int did_something;
- int busy_loops = 0;
int rc;
- cfs_block_allsigs();
-
init_waitqueue_entry(&wait, current);
sched = kiblnd_data.kib_scheds[KIB_THREAD_CPT(id)];
spin_lock_irqsave(&sched->ibs_lock, flags);
while (!kiblnd_data.kib_shutdown) {
- if (busy_loops++ >= IBLND_RESCHED) {
+ if (need_resched()) {
spin_unlock_irqrestore(&sched->ibs_lock, flags);
cond_resched();
- busy_loops = 0;
spin_lock_irqsave(&sched->ibs_lock, flags);
}
spin_unlock_irqrestore(&sched->ibs_lock, flags);
schedule();
- busy_loops = 0;
remove_wait_queue(&sched->ibs_waitq, &wait);
set_current_state(TASK_RUNNING);
LASSERT(*kiblnd_tunables.kib_dev_failover != 0);
- cfs_block_allsigs();
-
init_waitqueue_entry(&wait, current);
write_lock_irqsave(glock, flags);
git://git.whamcloud.com / fs / lustre-release.git / blobdiff