atomic_t *a = is_net ? &mdev->pp_in_use_by_net : &mdev->pp_in_use;
int i;
+ if (page == NULL)
+ return;
+
if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count)
i = page_chain_free(page);
else {
unsigned int data_size, gfp_t gfp_mask) __must_hold(local)
{
struct drbd_peer_request *peer_req;
- struct page *page;
+ struct page *page = NULL;
unsigned nr_pages = (data_size + PAGE_SIZE -1) >> PAGE_SHIFT;
if (drbd_insert_fault(mdev, DRBD_FAULT_AL_EE))
return NULL;
}
- page = drbd_alloc_pages(mdev, nr_pages, (gfp_mask & __GFP_WAIT));
- if (!page)
- goto fail;
+ if (data_size) {
+ page = drbd_alloc_pages(mdev, nr_pages, (gfp_mask & __GFP_WAIT));
+ if (!page)
+ goto fail;
+ }
drbd_clear_interval(&peer_req->i);
peer_req->i.size = data_size;
drbd_free_net_peer_req(mdev, peer_req);
/* possible callbacks here:
- * e_end_block, and e_end_resync_block, e_send_discard_write.
+ * e_end_block, and e_end_resync_block, e_send_superseded.
* all ignore the last argument.
*/
list_for_each_entry_safe(peer_req, t, &work_list, w.list) {
static int drbd_recv(struct drbd_tconn *tconn, void *buf, size_t size)
{
- mm_segment_t oldfs;
- struct kvec iov = {
- .iov_base = buf,
- .iov_len = size,
- };
- struct msghdr msg = {
- .msg_iovlen = 1,
- .msg_iov = (struct iovec *)&iov,
- .msg_flags = MSG_WAITALL | MSG_NOSIGNAL
- };
int rv;
- oldfs = get_fs();
- set_fs(KERNEL_DS);
+ rv = drbd_recv_short(tconn->data.socket, buf, size, 0);
- for (;;) {
- rv = sock_recvmsg(tconn->data.socket, &msg, size, msg.msg_flags);
- if (rv == size)
- break;
+ if (rv < 0) {
+ if (rv == -ECONNRESET)
+ conn_info(tconn, "sock was reset by peer\n");
+ else if (rv != -ERESTARTSYS)
+ conn_err(tconn, "sock_recvmsg returned %d\n", rv);
+ } else if (rv == 0) {
+ if (test_bit(DISCONNECT_SENT, &tconn->flags)) {
+ long t;
+ rcu_read_lock();
+ t = rcu_dereference(tconn->net_conf)->ping_timeo * HZ/10;
+ rcu_read_unlock();
- /* Note:
- * ECONNRESET other side closed the connection
- * ERESTARTSYS (on sock) we got a signal
- */
+ t = wait_event_timeout(tconn->ping_wait, tconn->cstate < C_WF_REPORT_PARAMS, t);
- if (rv < 0) {
- if (rv == -ECONNRESET)
- conn_info(tconn, "sock was reset by peer\n");
- else if (rv != -ERESTARTSYS)
- conn_err(tconn, "sock_recvmsg returned %d\n", rv);
- break;
- } else if (rv == 0) {
- conn_info(tconn, "sock was shut down by peer\n");
- break;
- } else {
- /* signal came in, or peer/link went down,
- * after we read a partial message
- */
- /* D_ASSERT(signal_pending(current)); */
- break;
+ if (t)
+ goto out;
}
- };
-
- set_fs(oldfs);
+ conn_info(tconn, "sock was shut down by peer\n");
+ }
if (rv != size)
conn_request_state(tconn, NS(conn, C_BROKEN_PIPE), CS_HARD);
+out:
return rv;
}
};
-static void incomming_connection(struct sock *sk)
+static void drbd_incoming_connection(struct sock *sk)
{
struct accept_wait_data *ad = sk->sk_user_data;
- struct drbd_tconn *tconn = ad->tconn;
-
- if (sk->sk_state != TCP_ESTABLISHED)
- conn_warn(tconn, "unexpected tcp state change. sk_state = %d\n", sk->sk_state);
+ void (*state_change)(struct sock *sk);
- write_lock_bh(&sk->sk_callback_lock);
- sk->sk_state_change = ad->original_sk_state_change;
- sk->sk_user_data = NULL;
- write_unlock_bh(&sk->sk_callback_lock);
-
- sk->sk_state_change(sk);
- complete(&ad->door_bell);
+ state_change = ad->original_sk_state_change;
+ if (sk->sk_state == TCP_ESTABLISHED)
+ complete(&ad->door_bell);
+ state_change(sk);
}
static int prepare_listen_socket(struct drbd_tconn *tconn, struct accept_wait_data *ad)
goto out;
}
- s_listen->sk->sk_reuse = 1; /* SO_REUSEADDR */
+ s_listen->sk->sk_reuse = SK_CAN_REUSE; /* SO_REUSEADDR */
drbd_setbufsize(s_listen, sndbuf_size, rcvbuf_size);
what = "bind before listen";
ad->s_listen = s_listen;
write_lock_bh(&s_listen->sk->sk_callback_lock);
ad->original_sk_state_change = s_listen->sk->sk_state_change;
- s_listen->sk->sk_state_change = incomming_connection;
+ s_listen->sk->sk_state_change = drbd_incoming_connection;
s_listen->sk->sk_user_data = ad;
write_unlock_bh(&s_listen->sk->sk_callback_lock);
return -EIO;
}
+static void unregister_state_change(struct sock *sk, struct accept_wait_data *ad)
+{
+ write_lock_bh(&sk->sk_callback_lock);
+ sk->sk_state_change = ad->original_sk_state_change;
+ sk->sk_user_data = NULL;
+ write_unlock_bh(&sk->sk_callback_lock);
+}
+
static struct socket *drbd_wait_for_connect(struct drbd_tconn *tconn, struct accept_wait_data *ad)
{
int timeo, connect_int, err = 0;
}
}
+ if (s_estab)
+ unregister_state_change(s_estab->sk, ad);
+
return s_estab;
}
struct drbd_socket sock, msock;
struct drbd_conf *mdev;
struct net_conf *nc;
- int vnr, timeout, try, h, ok;
+ int vnr, timeout, h, ok;
bool discard_my_data;
enum drbd_state_rv rv;
struct accept_wait_data ad = {
.door_bell = COMPLETION_INITIALIZER_ONSTACK(ad.door_bell),
};
+ clear_bit(DISCONNECT_SENT, &tconn->flags);
if (conn_request_state(tconn, NS(conn, C_WF_CONNECTION), CS_VERBOSE) < SS_SUCCESS)
return -2;
msock.rbuf = tconn->meta.rbuf;
msock.socket = NULL;
- clear_bit(DISCARD_CONCURRENT, &tconn->flags);
-
/* Assume that the peer only understands protocol 80 until we know better. */
tconn->agreed_pro_version = 80;
do {
struct socket *s;
- for (try = 0;;) {
- /* 3 tries, this should take less than a second! */
- s = drbd_try_connect(tconn);
- if (s || ++try >= 3)
- break;
- /* give the other side time to call bind() & listen() */
- schedule_timeout_interruptible(HZ / 10);
- }
-
+ s = drbd_try_connect(tconn);
if (s) {
if (!sock.socket) {
sock.socket = s;
send_first_packet(tconn, &sock, P_INITIAL_DATA);
} else if (!msock.socket) {
+ clear_bit(RESOLVE_CONFLICTS, &tconn->flags);
msock.socket = s;
send_first_packet(tconn, &msock, P_INITIAL_META);
} else {
retry:
s = drbd_wait_for_connect(tconn, &ad);
if (s) {
- try = receive_first_packet(tconn, s);
+ int fp = receive_first_packet(tconn, s);
drbd_socket_okay(&sock.socket);
drbd_socket_okay(&msock.socket);
- switch (try) {
+ switch (fp) {
case P_INITIAL_DATA:
if (sock.socket) {
conn_warn(tconn, "initial packet S crossed\n");
sock_release(sock.socket);
+ sock.socket = s;
+ goto randomize;
}
sock.socket = s;
break;
case P_INITIAL_META:
+ set_bit(RESOLVE_CONFLICTS, &tconn->flags);
if (msock.socket) {
conn_warn(tconn, "initial packet M crossed\n");
sock_release(msock.socket);
+ msock.socket = s;
+ goto randomize;
}
msock.socket = s;
- set_bit(DISCARD_CONCURRENT, &tconn->flags);
break;
default:
conn_warn(tconn, "Error receiving initial packet\n");
sock_release(s);
+randomize:
if (random32() & 1)
goto retry;
}
goto out_release_sockets;
}
- if (sock.socket && &msock.socket) {
- ok = drbd_socket_okay(&sock.socket);
- ok = drbd_socket_okay(&msock.socket) && ok;
- if (ok)
- break;
- }
- } while (1);
+ ok = drbd_socket_okay(&sock.socket);
+ ok = drbd_socket_okay(&msock.socket) && ok;
+ } while (!ok);
if (ad.s_listen)
sock_release(ad.s_listen);
- sock.socket->sk->sk_reuse = 1; /* SO_REUSEADDR */
- msock.socket->sk->sk_reuse = 1; /* SO_REUSEADDR */
+ sock.socket->sk->sk_reuse = SK_CAN_REUSE; /* SO_REUSEADDR */
+ msock.socket->sk->sk_reuse = SK_CAN_REUSE; /* SO_REUSEADDR */
sock.socket->sk->sk_allocation = GFP_NOIO;
msock.socket->sk->sk_allocation = GFP_NOIO;
data_size -= dgs;
}
- if (!expect(data_size != 0))
- return NULL;
if (!expect(IS_ALIGNED(data_size, 512)))
return NULL;
if (!expect(data_size <= DRBD_MAX_BIO_SIZE))
if (!peer_req)
return NULL;
+ if (!data_size)
+ return peer_req;
+
ds = data_size;
page = peer_req->pages;
page_chain_for_each(page) {
continue;
/* as it is RQ_POSTPONED, this will cause it to
* be queued on the retry workqueue. */
- __req_mod(req, DISCARD_WRITE, NULL);
+ __req_mod(req, CONFLICT_RESOLVED, NULL);
}
}
return err;
}
-static int e_send_discard_write(struct drbd_work *w, int unused)
+static int e_send_superseded(struct drbd_work *w, int unused)
{
- return e_send_ack(w, P_DISCARD_WRITE);
+ return e_send_ack(w, P_SUPERSEDED);
}
static int e_send_retry_write(struct drbd_work *w, int unused)
struct drbd_tconn *tconn = w->mdev->tconn;
return e_send_ack(w, tconn->agreed_pro_version >= 100 ?
- P_RETRY_WRITE : P_DISCARD_WRITE);
+ P_RETRY_WRITE : P_SUPERSEDED);
}
static bool seq_greater(u32 a, u32 b)
/*
* We only need to keep track of the last packet_seq number of our peer
- * if we are in dual-primary mode and we have the discard flag set; see
+ * if we are in dual-primary mode and we have the resolve-conflicts flag set; see
* handle_write_conflicts().
*/
tp = rcu_dereference(mdev->tconn->net_conf)->two_primaries;
rcu_read_unlock();
- return tp && test_bit(DISCARD_CONCURRENT, &tconn->flags);
+ return tp && test_bit(RESOLVE_CONFLICTS, &tconn->flags);
}
static void update_peer_seq(struct drbd_conf *mdev, unsigned int peer_seq)
struct drbd_peer_request *peer_req)
{
struct drbd_tconn *tconn = mdev->tconn;
- bool resolve_conflicts = test_bit(DISCARD_CONCURRENT, &tconn->flags);
+ bool resolve_conflicts = test_bit(RESOLVE_CONFLICTS, &tconn->flags);
sector_t sector = peer_req->i.sector;
const unsigned int size = peer_req->i.size;
struct drbd_interval *i;
if (resolve_conflicts) {
/*
* If the peer request is fully contained within the
- * overlapping request, it can be discarded; otherwise,
- * it will be retried once all overlapping requests
- * have completed.
+ * overlapping request, it can be considered overwritten
+ * and thus superseded; otherwise, it will be retried
+ * once all overlapping requests have completed.
*/
- bool discard = i->sector <= sector && i->sector +
+ bool superseded = i->sector <= sector && i->sector +
(i->size >> 9) >= sector + (size >> 9);
if (!equal)
"assuming %s came first\n",
(unsigned long long)i->sector, i->size,
(unsigned long long)sector, size,
- discard ? "local" : "remote");
+ superseded ? "local" : "remote");
inc_unacked(mdev);
- peer_req->w.cb = discard ? e_send_discard_write :
+ peer_req->w.cb = superseded ? e_send_superseded :
e_send_retry_write;
list_add_tail(&peer_req->w.list, &mdev->done_ee);
wake_asender(mdev->tconn);
!(req->rq_state & RQ_POSTPONED)) {
/*
* Wait for the node with the discard flag to
- * decide if this request will be discarded or
- * retried. Requests that are discarded will
+ * decide if this request has been superseded
+ * or needs to be retried.
+ * Requests that have been superseded will
* disappear from the write_requests tree.
*
* In addition, wait for the conflicting
dp_flags = be32_to_cpu(p->dp_flags);
rw |= wire_flags_to_bio(mdev, dp_flags);
+ if (peer_req->pages == NULL) {
+ D_ASSERT(peer_req->i.size == 0);
+ D_ASSERT(dp_flags & DP_FLUSH);
+ }
if (dp_flags & DP_MAY_SET_IN_SYNC)
peer_req->flags |= EE_MAY_SET_IN_SYNC;
"Using discard-least-changes instead\n");
case ASB_DISCARD_ZERO_CHG:
if (ch_peer == 0 && ch_self == 0) {
- rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
+ rv = test_bit(RESOLVE_CONFLICTS, &mdev->tconn->flags)
? -1 : 1;
break;
} else {
rv = 1;
else /* ( ch_self == ch_peer ) */
/* Well, then use something else. */
- rv = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags)
+ rv = test_bit(RESOLVE_CONFLICTS, &mdev->tconn->flags)
? -1 : 1;
break;
case ASB_DISCARD_LOCAL:
if ((mdev->ldev->md.uuid[UI_BITMAP] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1)) &&
(mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START + 1] & ~((u64)1))) {
dev_info(DEV, "was SyncSource, missed the resync finished event, corrected myself:\n");
- drbd_uuid_set_bm(mdev, 0UL);
+ drbd_uuid_move_history(mdev);
+ mdev->ldev->md.uuid[UI_HISTORY_START] = mdev->ldev->md.uuid[UI_BITMAP];
+ mdev->ldev->md.uuid[UI_BITMAP] = 0;
drbd_uuid_dump(mdev, "self", mdev->ldev->md.uuid,
mdev->state.disk >= D_NEGOTIATING ? drbd_bm_total_weight(mdev) : 0, 0);
case 1: /* self_pri && !peer_pri */ return 1;
case 2: /* !self_pri && peer_pri */ return -1;
case 3: /* self_pri && peer_pri */
- dc = test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags);
+ dc = test_bit(RESOLVE_CONFLICTS, &mdev->tconn->flags);
return dc ? -1 : 1;
}
}
if (mdev->tconn->agreed_pro_version < 91)
return -1091;
- _drbd_uuid_set(mdev, UI_BITMAP, mdev->ldev->md.uuid[UI_HISTORY_START]);
- _drbd_uuid_set(mdev, UI_HISTORY_START, mdev->ldev->md.uuid[UI_HISTORY_START + 1]);
+ __drbd_uuid_set(mdev, UI_BITMAP, mdev->ldev->md.uuid[UI_HISTORY_START]);
+ __drbd_uuid_set(mdev, UI_HISTORY_START, mdev->ldev->md.uuid[UI_HISTORY_START + 1]);
dev_info(DEV, "Last syncUUID did not get through, corrected:\n");
drbd_uuid_dump(mdev, "self", mdev->ldev->md.uuid,
mydisk = mdev->new_state_tmp.disk;
dev_info(DEV, "drbd_sync_handshake:\n");
+
+ spin_lock_irq(&mdev->ldev->md.uuid_lock);
drbd_uuid_dump(mdev, "self", mdev->ldev->md.uuid, mdev->comm_bm_set, 0);
drbd_uuid_dump(mdev, "peer", mdev->p_uuid,
mdev->p_uuid[UI_SIZE], mdev->p_uuid[UI_FLAGS]);
hg = drbd_uuid_compare(mdev, &rule_nr);
+ spin_unlock_irq(&mdev->ldev->md.uuid_lock);
dev_info(DEV, "uuid_compare()=%d by rule %d\n", hg, rule_nr);
mask.i = be32_to_cpu(p->mask);
val.i = be32_to_cpu(p->val);
- if (test_bit(DISCARD_CONCURRENT, &mdev->tconn->flags) &&
+ if (test_bit(RESOLVE_CONFLICTS, &mdev->tconn->flags) &&
mutex_is_locked(mdev->state_mutex)) {
drbd_send_sr_reply(mdev, SS_CONCURRENT_ST_CHG);
return 0;
mask.i = be32_to_cpu(p->mask);
val.i = be32_to_cpu(p->val);
- if (test_bit(DISCARD_CONCURRENT, &tconn->flags) &&
+ if (test_bit(RESOLVE_CONFLICTS, &tconn->flags) &&
mutex_is_locked(&tconn->cstate_mutex)) {
conn_send_sr_reply(tconn, SS_CONCURRENT_ST_CHG);
return 0;
* already decided to close the connection again,
* we must not "re-establish" it here. */
if (os.conn <= C_TEAR_DOWN)
- return false;
+ return -ECONNRESET;
/* If this is the "end of sync" confirmation, usually the peer disk
* transitions from D_INCONSISTENT to D_UP_TO_DATE. For empty (0 bits
}
}
+ /* explicit verify finished notification, stop sector reached. */
+ if (os.conn == C_VERIFY_T && os.disk == D_UP_TO_DATE &&
+ peer_state.conn == C_CONNECTED && real_peer_disk == D_UP_TO_DATE) {
+ ov_out_of_sync_print(mdev);
+ drbd_resync_finished(mdev);
+ return 0;
+ }
+
/* peer says his disk is inconsistent, while we think it is uptodate,
* and this happens while the peer still thinks we have a sync going on,
* but we think we are already done with the sync.
barr.w.cb = w_prev_work_done;
barr.w.tconn = tconn;
init_completion(&barr.done);
- drbd_queue_work(&tconn->data.work, &barr.w);
+ drbd_queue_work(&tconn->sender_work, &barr.w);
wait_for_completion(&barr.done);
}
conn_err(tconn, "ASSERTION FAILED: tconn->current_epoch->list not empty\n");
/* ok, no more ee's on the fly, it is safe to reset the epoch_size */
atomic_set(&tconn->current_epoch->epoch_size, 0);
+ tconn->send.seen_any_write_yet = false;
conn_info(tconn, "Connection closed\n");
necessary to reclain net_ee in drbd_finish_peer_reqs(). */
drbd_flush_workqueue(mdev);
+ /* need to do it again, drbd_finish_peer_reqs() may have populated it
+ * again via drbd_try_clear_on_disk_bm(). */
+ drbd_rs_cancel_all(mdev);
+
kfree(mdev->p_uuid);
mdev->p_uuid = NULL;
case P_RECV_ACK:
what = RECV_ACKED_BY_PEER;
break;
- case P_DISCARD_WRITE:
- what = DISCARD_WRITE;
+ case P_SUPERSEDED:
+ what = CONFLICT_RESOLVED;
break;
case P_RETRY_WRITE:
what = POSTPONE_WRITE;
if (w) {
w->cb = w_ov_finished;
w->mdev = mdev;
- drbd_queue_work_front(&mdev->tconn->data.work, w);
+ drbd_queue_work(&mdev->tconn->sender_work, w);
} else {
dev_err(DEV, "kmalloc(w) failed.");
ov_out_of_sync_print(mdev);
[P_RECV_ACK] = { sizeof(struct p_block_ack), got_BlockAck },
[P_WRITE_ACK] = { sizeof(struct p_block_ack), got_BlockAck },
[P_RS_WRITE_ACK] = { sizeof(struct p_block_ack), got_BlockAck },
- [P_DISCARD_WRITE] = { sizeof(struct p_block_ack), got_BlockAck },
+ [P_SUPERSEDED] = { sizeof(struct p_block_ack), got_BlockAck },
[P_NEG_ACK] = { sizeof(struct p_block_ack), got_NegAck },
[P_NEG_DREPLY] = { sizeof(struct p_block_ack), got_NegDReply },
[P_NEG_RS_DREPLY] = { sizeof(struct p_block_ack), got_NegRSDReply },
received += rv;
buf += rv;
} else if (rv == 0) {
+ if (test_bit(DISCONNECT_SENT, &tconn->flags)) {
+ long t;
+ rcu_read_lock();
+ t = rcu_dereference(tconn->net_conf)->ping_timeo * HZ/10;
+ rcu_read_unlock();
+
+ t = wait_event_timeout(tconn->ping_wait,
+ tconn->cstate < C_WF_REPORT_PARAMS,
+ t);
+ if (t)
+ break;
+ }
conn_err(tconn, "meta connection shut down by peer.\n");
goto reconnect;
} else if (rv == -EAGAIN) {
if (0) {
reconnect:
conn_request_state(tconn, NS(conn, C_NETWORK_FAILURE), CS_HARD);
+ conn_md_sync(tconn);
}
if (0) {
disconnect: