aboutsummaryrefslogtreecommitdiffstats
path: root/fs/ocfs2/cluster
diff options
context:
space:
mode:
authorZach Brown <zach.brown@oracle.com>2005-12-15 17:31:23 -0500
committerJoel Becker <joel.becker@oracle.com>2006-01-03 14:45:46 -0500
commit98211489d4147e41b11703e4245846d60b3acce4 (patch)
treef3c9c6b8df5bb001db79bc6314d8cbb5e127b45b /fs/ocfs2/cluster
parenta7f6a5fb4bde142b622706e2006ba33f793e13ed (diff)
[PATCH] OCFS2: The Second Oracle Cluster Filesystem
Node messaging via tcp. Used by the dlm and the file system for point to point communication between nodes. Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com> Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>
Diffstat (limited to 'fs/ocfs2/cluster')
-rw-r--r--fs/ocfs2/cluster/quorum.c315
-rw-r--r--fs/ocfs2/cluster/quorum.h36
-rw-r--r--fs/ocfs2/cluster/sys.c124
-rw-r--r--fs/ocfs2/cluster/sys.h33
-rw-r--r--fs/ocfs2/cluster/tcp.c1829
-rw-r--r--fs/ocfs2/cluster/tcp.h113
-rw-r--r--fs/ocfs2/cluster/tcp_internal.h174
7 files changed, 2624 insertions, 0 deletions
diff --git a/fs/ocfs2/cluster/quorum.c b/fs/ocfs2/cluster/quorum.c
new file mode 100644
index 000000000000..7bba98fbfc15
--- /dev/null
+++ b/fs/ocfs2/cluster/quorum.c
@@ -0,0 +1,315 @@
1/* -*- mode: c; c-basic-offset: 8; -*-
2 *
3 * vim: noexpandtab sw=8 ts=8 sts=0:
4 *
5 * Copyright (C) 2005 Oracle. All rights reserved.
6 *
7 * This program is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public
9 * License as published by the Free Software Foundation; either
10 * version 2 of the License, or (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public
18 * License along with this program; if not, write to the
19 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
20 * Boston, MA 021110-1307, USA.
21 */
22
23/* This quorum hack is only here until we transition to some more rational
24 * approach that is driven from userspace. Honest. No foolin'.
25 *
26 * Imagine two nodes lose network connectivity to each other but they're still
27 * up and operating in every other way. Presumably a network timeout indicates
28 * that a node is broken and should be recovered. They can't both recover each
29 * other and both carry on without serialising their access to the file system.
30 * They need to decide who is authoritative. Now extend that problem to
31 * arbitrary groups of nodes losing connectivity between each other.
32 *
33 * So we declare that a node which has given up on connecting to a majority
34 * of nodes who are still heartbeating will fence itself.
35 *
36 * There are huge opportunities for races here. After we give up on a node's
37 * connection we need to wait long enough to give heartbeat an opportunity
38 * to declare the node as truly dead. We also need to be careful with the
39 * race between when we see a node start heartbeating and when we connect
40 * to it.
41 *
42 * So nodes that are in this transtion put a hold on the quorum decision
43 * with a counter. As they fall out of this transition they drop the count
44 * and if they're the last, they fire off the decision.
45 */
46#include <linux/kernel.h>
47#include <linux/slab.h>
48#include <linux/workqueue.h>
49
50#include "heartbeat.h"
51#include "nodemanager.h"
52#define MLOG_MASK_PREFIX ML_QUORUM
53#include "masklog.h"
54#include "quorum.h"
55
56static struct o2quo_state {
57 spinlock_t qs_lock;
58 struct work_struct qs_work;
59 int qs_pending;
60 int qs_heartbeating;
61 unsigned long qs_hb_bm[BITS_TO_LONGS(O2NM_MAX_NODES)];
62 int qs_connected;
63 unsigned long qs_conn_bm[BITS_TO_LONGS(O2NM_MAX_NODES)];
64 int qs_holds;
65 unsigned long qs_hold_bm[BITS_TO_LONGS(O2NM_MAX_NODES)];
66} o2quo_state;
67
68/* this is horribly heavy-handed. It should instead flip the file
69 * system RO and call some userspace script. */
70static void o2quo_fence_self(void)
71{
72 /* panic spins with interrupts enabled. with preempt
73 * threads can still schedule, etc, etc */
74 o2hb_stop_all_regions();
75 panic("ocfs2 is very sorry to be fencing this system by panicing\n");
76}
77
78/* Indicate that a timeout occured on a hearbeat region write. The
79 * other nodes in the cluster may consider us dead at that time so we
80 * want to "fence" ourselves so that we don't scribble on the disk
81 * after they think they've recovered us. This can't solve all
82 * problems related to writeout after recovery but this hack can at
83 * least close some of those gaps. When we have real fencing, this can
84 * go away as our node would be fenced externally before other nodes
85 * begin recovery. */
86void o2quo_disk_timeout(void)
87{
88 o2quo_fence_self();
89}
90
91static void o2quo_make_decision(void *arg)
92{
93 int quorum;
94 int lowest_hb, lowest_reachable = 0, fence = 0;
95 struct o2quo_state *qs = &o2quo_state;
96
97 spin_lock(&qs->qs_lock);
98
99 lowest_hb = find_first_bit(qs->qs_hb_bm, O2NM_MAX_NODES);
100 if (lowest_hb != O2NM_MAX_NODES)
101 lowest_reachable = test_bit(lowest_hb, qs->qs_conn_bm);
102
103 mlog(0, "heartbeating: %d, connected: %d, "
104 "lowest: %d (%sreachable)\n", qs->qs_heartbeating,
105 qs->qs_connected, lowest_hb, lowest_reachable ? "" : "un");
106
107 if (!test_bit(o2nm_this_node(), qs->qs_hb_bm) ||
108 qs->qs_heartbeating == 1)
109 goto out;
110
111 if (qs->qs_heartbeating & 1) {
112 /* the odd numbered cluster case is straight forward --
113 * if we can't talk to the majority we're hosed */
114 quorum = (qs->qs_heartbeating + 1)/2;
115 if (qs->qs_connected < quorum) {
116 mlog(ML_ERROR, "fencing this node because it is "
117 "only connected to %u nodes and %u is needed "
118 "to make a quorum out of %u heartbeating nodes\n",
119 qs->qs_connected, quorum,
120 qs->qs_heartbeating);
121 fence = 1;
122 }
123 } else {
124 /* the even numbered cluster adds the possibility of each half
125 * of the cluster being able to talk amongst themselves.. in
126 * that case we're hosed if we can't talk to the group that has
127 * the lowest numbered node */
128 quorum = qs->qs_heartbeating / 2;
129 if (qs->qs_connected < quorum) {
130 mlog(ML_ERROR, "fencing this node because it is "
131 "only connected to %u nodes and %u is needed "
132 "to make a quorum out of %u heartbeating nodes\n",
133 qs->qs_connected, quorum,
134 qs->qs_heartbeating);
135 fence = 1;
136 }
137 else if ((qs->qs_connected == quorum) &&
138 !lowest_reachable) {
139 mlog(ML_ERROR, "fencing this node because it is "
140 "connected to a half-quorum of %u out of %u "
141 "nodes which doesn't include the lowest active "
142 "node %u\n", quorum, qs->qs_heartbeating,
143 lowest_hb);
144 fence = 1;
145 }
146 }
147
148out:
149 spin_unlock(&qs->qs_lock);
150 if (fence)
151 o2quo_fence_self();
152}
153
154static void o2quo_set_hold(struct o2quo_state *qs, u8 node)
155{
156 assert_spin_locked(&qs->qs_lock);
157
158 if (!test_and_set_bit(node, qs->qs_hold_bm)) {
159 qs->qs_holds++;
160 mlog_bug_on_msg(qs->qs_holds == O2NM_MAX_NODES,
161 "node %u\n", node);
162 mlog(0, "node %u, %d total\n", node, qs->qs_holds);
163 }
164}
165
166static void o2quo_clear_hold(struct o2quo_state *qs, u8 node)
167{
168 assert_spin_locked(&qs->qs_lock);
169
170 if (test_and_clear_bit(node, qs->qs_hold_bm)) {
171 mlog(0, "node %u, %d total\n", node, qs->qs_holds - 1);
172 if (--qs->qs_holds == 0) {
173 if (qs->qs_pending) {
174 qs->qs_pending = 0;
175 schedule_work(&qs->qs_work);
176 }
177 }
178 mlog_bug_on_msg(qs->qs_holds < 0, "node %u, holds %d\n",
179 node, qs->qs_holds);
180 }
181}
182
183/* as a node comes up we delay the quorum decision until we know the fate of
184 * the connection. the hold will be droped in conn_up or hb_down. it might be
185 * perpetuated by con_err until hb_down. if we already have a conn, we might
186 * be dropping a hold that conn_up got. */
187void o2quo_hb_up(u8 node)
188{
189 struct o2quo_state *qs = &o2quo_state;
190
191 spin_lock(&qs->qs_lock);
192
193 qs->qs_heartbeating++;
194 mlog_bug_on_msg(qs->qs_heartbeating == O2NM_MAX_NODES,
195 "node %u\n", node);
196 mlog_bug_on_msg(test_bit(node, qs->qs_hb_bm), "node %u\n", node);
197 set_bit(node, qs->qs_hb_bm);
198
199 mlog(0, "node %u, %d total\n", node, qs->qs_heartbeating);
200
201 if (!test_bit(node, qs->qs_conn_bm))
202 o2quo_set_hold(qs, node);
203 else
204 o2quo_clear_hold(qs, node);
205
206 spin_unlock(&qs->qs_lock);
207}
208
209/* hb going down releases any holds we might have had due to this node from
210 * conn_up, conn_err, or hb_up */
211void o2quo_hb_down(u8 node)
212{
213 struct o2quo_state *qs = &o2quo_state;
214
215 spin_lock(&qs->qs_lock);
216
217 qs->qs_heartbeating--;
218 mlog_bug_on_msg(qs->qs_heartbeating < 0,
219 "node %u, %d heartbeating\n",
220 node, qs->qs_heartbeating);
221 mlog_bug_on_msg(!test_bit(node, qs->qs_hb_bm), "node %u\n", node);
222 clear_bit(node, qs->qs_hb_bm);
223
224 mlog(0, "node %u, %d total\n", node, qs->qs_heartbeating);
225
226 o2quo_clear_hold(qs, node);
227
228 spin_unlock(&qs->qs_lock);
229}
230
231/* this tells us that we've decided that the node is still heartbeating
232 * even though we've lost it's conn. it must only be called after conn_err
233 * and indicates that we must now make a quorum decision in the future,
234 * though we might be doing so after waiting for holds to drain. Here
235 * we'll be dropping the hold from conn_err. */
236void o2quo_hb_still_up(u8 node)
237{
238 struct o2quo_state *qs = &o2quo_state;
239
240 spin_lock(&qs->qs_lock);
241
242 mlog(0, "node %u\n", node);
243
244 qs->qs_pending = 1;
245 o2quo_clear_hold(qs, node);
246
247 spin_unlock(&qs->qs_lock);
248}
249
250/* This is analagous to hb_up. as a node's connection comes up we delay the
251 * quorum decision until we see it heartbeating. the hold will be droped in
252 * hb_up or hb_down. it might be perpetuated by con_err until hb_down. if
253 * it's already heartbeating we we might be dropping a hold that conn_up got.
254 * */
255void o2quo_conn_up(u8 node)
256{
257 struct o2quo_state *qs = &o2quo_state;
258
259 spin_lock(&qs->qs_lock);
260
261 qs->qs_connected++;
262 mlog_bug_on_msg(qs->qs_connected == O2NM_MAX_NODES,
263 "node %u\n", node);
264 mlog_bug_on_msg(test_bit(node, qs->qs_conn_bm), "node %u\n", node);
265 set_bit(node, qs->qs_conn_bm);
266
267 mlog(0, "node %u, %d total\n", node, qs->qs_connected);
268
269 if (!test_bit(node, qs->qs_hb_bm))
270 o2quo_set_hold(qs, node);
271 else
272 o2quo_clear_hold(qs, node);
273
274 spin_unlock(&qs->qs_lock);
275}
276
277/* we've decided that we won't ever be connecting to the node again. if it's
278 * still heartbeating we grab a hold that will delay decisions until either the
279 * node stops heartbeating from hb_down or the caller decides that the node is
280 * still up and calls still_up */
281void o2quo_conn_err(u8 node)
282{
283 struct o2quo_state *qs = &o2quo_state;
284
285 spin_lock(&qs->qs_lock);
286
287 if (test_bit(node, qs->qs_conn_bm)) {
288 qs->qs_connected--;
289 mlog_bug_on_msg(qs->qs_connected < 0,
290 "node %u, connected %d\n",
291 node, qs->qs_connected);
292
293 clear_bit(node, qs->qs_conn_bm);
294 }
295
296 mlog(0, "node %u, %d total\n", node, qs->qs_connected);
297
298 if (test_bit(node, qs->qs_hb_bm))
299 o2quo_set_hold(qs, node);
300
301 spin_unlock(&qs->qs_lock);
302}
303
304void o2quo_init(void)
305{
306 struct o2quo_state *qs = &o2quo_state;
307
308 spin_lock_init(&qs->qs_lock);
309 INIT_WORK(&qs->qs_work, o2quo_make_decision, NULL);
310}
311
312void o2quo_exit(void)
313{
314 flush_scheduled_work();
315}
diff --git a/fs/ocfs2/cluster/quorum.h b/fs/ocfs2/cluster/quorum.h
new file mode 100644
index 000000000000..6649cc6f67c9
--- /dev/null
+++ b/fs/ocfs2/cluster/quorum.h
@@ -0,0 +1,36 @@
1/* -*- mode: c; c-basic-offset: 8; -*-
2 * vim: noexpandtab sw=8 ts=8 sts=0:
3 *
4 * Copyright (C) 2005 Oracle. All rights reserved.
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
10 *
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * General Public License for more details.
15 *
16 * You should have received a copy of the GNU General Public
17 * License along with this program; if not, write to the
18 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
19 * Boston, MA 021110-1307, USA.
20 *
21 */
22
23#ifndef O2CLUSTER_QUORUM_H
24#define O2CLUSTER_QUORUM_H
25
26void o2quo_init(void);
27void o2quo_exit(void);
28
29void o2quo_hb_up(u8 node);
30void o2quo_hb_down(u8 node);
31void o2quo_hb_still_up(u8 node);
32void o2quo_conn_up(u8 node);
33void o2quo_conn_err(u8 node);
34void o2quo_disk_timeout(void);
35
36#endif /* O2CLUSTER_QUORUM_H */
diff --git a/fs/ocfs2/cluster/sys.c b/fs/ocfs2/cluster/sys.c
new file mode 100644
index 000000000000..f1e99461bb0d
--- /dev/null
+++ b/fs/ocfs2/cluster/sys.c
@@ -0,0 +1,124 @@
1/* -*- mode: c; c-basic-offset: 8; -*-
2 * vim: noexpandtab sw=8 ts=8 sts=0:
3 *
4 * sys.c
5 *
6 * OCFS2 cluster sysfs interface
7 *
8 * Copyright (C) 2005 Oracle. All rights reserved.
9 *
10 * This program is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public
12 * License as published by the Free Software Foundation,
13 * version 2 of the License.
14 *
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * General Public License for more details.
19 *
20 * You should have received a copy of the GNU General Public
21 * License along with this program; if not, write to the
22 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
23 * Boston, MA 021110-1307, USA.
24 *
25 */
26
27#include <linux/kernel.h>
28#include <linux/module.h>
29#include <linux/kobject.h>
30#include <linux/sysfs.h>
31
32#include "ocfs2_nodemanager.h"
33#include "masklog.h"
34#include "sys.h"
35
36struct o2cb_attribute {
37 struct attribute attr;
38 ssize_t (*show)(char *buf);
39 ssize_t (*store)(const char *buf, size_t count);
40};
41
42#define O2CB_ATTR(_name, _mode, _show, _store) \
43struct o2cb_attribute o2cb_attr_##_name = __ATTR(_name, _mode, _show, _store)
44
45#define to_o2cb_subsys(k) container_of(to_kset(k), struct subsystem, kset)
46#define to_o2cb_attr(_attr) container_of(_attr, struct o2cb_attribute, attr)
47
48static ssize_t o2cb_interface_revision_show(char *buf)
49{
50 return snprintf(buf, PAGE_SIZE, "%u\n", O2NM_API_VERSION);
51}
52
53O2CB_ATTR(interface_revision, S_IFREG | S_IRUGO, o2cb_interface_revision_show, NULL);
54
55static struct attribute *o2cb_attrs[] = {
56 &o2cb_attr_interface_revision.attr,
57 NULL,
58};
59
60static ssize_t
61o2cb_show(struct kobject * kobj, struct attribute * attr, char * buffer);
62static ssize_t
63o2cb_store(struct kobject * kobj, struct attribute * attr,
64 const char * buffer, size_t count);
65static struct sysfs_ops o2cb_sysfs_ops = {
66 .show = o2cb_show,
67 .store = o2cb_store,
68};
69
70static struct kobj_type o2cb_subsys_type = {
71 .default_attrs = o2cb_attrs,
72 .sysfs_ops = &o2cb_sysfs_ops,
73};
74
75/* gives us o2cb_subsys */
76decl_subsys(o2cb, NULL, NULL);
77
78static ssize_t
79o2cb_show(struct kobject * kobj, struct attribute * attr, char * buffer)
80{
81 struct o2cb_attribute *o2cb_attr = to_o2cb_attr(attr);
82 struct subsystem *sbs = to_o2cb_subsys(kobj);
83
84 BUG_ON(sbs != &o2cb_subsys);
85
86 if (o2cb_attr->show)
87 return o2cb_attr->show(buffer);
88 return -EIO;
89}
90
91static ssize_t
92o2cb_store(struct kobject * kobj, struct attribute * attr,
93 const char * buffer, size_t count)
94{
95 struct o2cb_attribute *o2cb_attr = to_o2cb_attr(attr);
96 struct subsystem *sbs = to_o2cb_subsys(kobj);
97
98 BUG_ON(sbs != &o2cb_subsys);
99
100 if (o2cb_attr->store)
101 return o2cb_attr->store(buffer, count);
102 return -EIO;
103}
104
105void o2cb_sys_shutdown(void)
106{
107 mlog_sys_shutdown();
108 subsystem_unregister(&o2cb_subsys);
109}
110
111int o2cb_sys_init(void)
112{
113 int ret;
114
115 o2cb_subsys.kset.kobj.ktype = &o2cb_subsys_type;
116 ret = subsystem_register(&o2cb_subsys);
117 if (ret)
118 return ret;
119
120 ret = mlog_sys_init(&o2cb_subsys);
121 if (ret)
122 subsystem_unregister(&o2cb_subsys);
123 return ret;
124}
diff --git a/fs/ocfs2/cluster/sys.h b/fs/ocfs2/cluster/sys.h
new file mode 100644
index 000000000000..d66b8ab0045e
--- /dev/null
+++ b/fs/ocfs2/cluster/sys.h
@@ -0,0 +1,33 @@
1/* -*- mode: c; c-basic-offset: 8; -*-
2 * vim: noexpandtab sw=8 ts=8 sts=0:
3 *
4 * sys.h
5 *
6 * Function prototypes for o2cb sysfs interface
7 *
8 * Copyright (C) 2005 Oracle. All rights reserved.
9 *
10 * This program is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public
12 * License as published by the Free Software Foundation,
13 * version 2 of the License.
14 *
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * General Public License for more details.
19 *
20 * You should have received a copy of the GNU General Public
21 * License along with this program; if not, write to the
22 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
23 * Boston, MA 021110-1307, USA.
24 *
25 */
26
27#ifndef O2CLUSTER_SYS_H
28#define O2CLUSTER_SYS_H
29
30void o2cb_sys_shutdown(void);
31int o2cb_sys_init(void);
32
33#endif /* O2CLUSTER_SYS_H */
diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
new file mode 100644
index 000000000000..35d92c01a972
--- /dev/null
+++ b/fs/ocfs2/cluster/tcp.c
@@ -0,0 +1,1829 @@
1/* -*- mode: c; c-basic-offset: 8; -*-
2 *
3 * vim: noexpandtab sw=8 ts=8 sts=0:
4 *
5 * Copyright (C) 2004 Oracle. All rights reserved.
6 *
7 * This program is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public
9 * License as published by the Free Software Foundation; either
10 * version 2 of the License, or (at your option) any later version.
11 *
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public
18 * License along with this program; if not, write to the
19 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
20 * Boston, MA 021110-1307, USA.
21 *
22 * ----
23 *
24 * Callers for this were originally written against a very simple synchronus
25 * API. This implementation reflects those simple callers. Some day I'm sure
26 * we'll need to move to a more robust posting/callback mechanism.
27 *
28 * Transmit calls pass in kernel virtual addresses and block copying this into
29 * the socket's tx buffers via a usual blocking sendmsg. They'll block waiting
30 * for a failed socket to timeout. TX callers can also pass in a poniter to an
31 * 'int' which gets filled with an errno off the wire in response to the
32 * message they send.
33 *
34 * Handlers for unsolicited messages are registered. Each socket has a page
35 * that incoming data is copied into. First the header, then the data.
36 * Handlers are called from only one thread with a reference to this per-socket
37 * page. This page is destroyed after the handler call, so it can't be
38 * referenced beyond the call. Handlers may block but are discouraged from
39 * doing so.
40 *
41 * Any framing errors (bad magic, large payload lengths) close a connection.
42 *
43 * Our sock_container holds the state we associate with a socket. It's current
44 * framing state is held there as well as the refcounting we do around when it
45 * is safe to tear down the socket. The socket is only finally torn down from
46 * the container when the container loses all of its references -- so as long
47 * as you hold a ref on the container you can trust that the socket is valid
48 * for use with kernel socket APIs.
49 *
50 * Connections are initiated between a pair of nodes when the node with the
51 * higher node number gets a heartbeat callback which indicates that the lower
52 * numbered node has started heartbeating. The lower numbered node is passive
53 * and only accepts the connection if the higher numbered node is heartbeating.
54 */
55
56#include <linux/kernel.h>
57#include <linux/jiffies.h>
58#include <linux/slab.h>
59#include <linux/idr.h>
60#include <linux/kref.h>
61#include <net/tcp.h>
62
63#include <asm/uaccess.h>
64
65#include "heartbeat.h"
66#include "tcp.h"
67#include "nodemanager.h"
68#define MLOG_MASK_PREFIX ML_TCP
69#include "masklog.h"
70#include "quorum.h"
71
72#include "tcp_internal.h"
73
74/*
75 * The linux network stack isn't sparse endian clean.. It has macros like
76 * ntohs() which perform the endian checks and structs like sockaddr_in
77 * which aren't annotated. So __force is found here to get the build
78 * clean. When they emerge from the dark ages and annotate the code
79 * we can remove these.
80 */
81
82#define SC_NODEF_FMT "node %s (num %u) at %u.%u.%u.%u:%u"
83#define SC_NODEF_ARGS(sc) sc->sc_node->nd_name, sc->sc_node->nd_num, \
84 NIPQUAD(sc->sc_node->nd_ipv4_address), \
85 ntohs(sc->sc_node->nd_ipv4_port)
86
87/*
88 * In the following two log macros, the whitespace after the ',' just
89 * before ##args is intentional. Otherwise, gcc 2.95 will eat the
90 * previous token if args expands to nothing.
91 */
92#define msglog(hdr, fmt, args...) do { \
93 typeof(hdr) __hdr = (hdr); \
94 mlog(ML_MSG, "[mag %u len %u typ %u stat %d sys_stat %d " \
95 "key %08x num %u] " fmt, \
96 be16_to_cpu(__hdr->magic), be16_to_cpu(__hdr->data_len), \
97 be16_to_cpu(__hdr->msg_type), be32_to_cpu(__hdr->status), \
98 be32_to_cpu(__hdr->sys_status), be32_to_cpu(__hdr->key), \
99 be32_to_cpu(__hdr->msg_num) , ##args); \
100} while (0)
101
102#define sclog(sc, fmt, args...) do { \
103 typeof(sc) __sc = (sc); \
104 mlog(ML_SOCKET, "[sc %p refs %d sock %p node %u page %p " \
105 "pg_off %zu] " fmt, __sc, \
106 atomic_read(&__sc->sc_kref.refcount), __sc->sc_sock, \
107 __sc->sc_node->nd_num, __sc->sc_page, __sc->sc_page_off , \
108 ##args); \
109} while (0)
110
111static rwlock_t o2net_handler_lock = RW_LOCK_UNLOCKED;
112static struct rb_root o2net_handler_tree = RB_ROOT;
113
114static struct o2net_node o2net_nodes[O2NM_MAX_NODES];
115
116/* XXX someday we'll need better accounting */
117static struct socket *o2net_listen_sock = NULL;
118
119/*
120 * listen work is only queued by the listening socket callbacks on the
121 * o2net_wq. teardown detaches the callbacks before destroying the workqueue.
122 * quorum work is queued as sock containers are shutdown.. stop_listening
123 * tears down all the node's sock containers, preventing future shutdowns
124 * and queued quroum work, before canceling delayed quorum work and
125 * destroying the work queue.
126 */
127static struct workqueue_struct *o2net_wq;
128static struct work_struct o2net_listen_work;
129
130static struct o2hb_callback_func o2net_hb_up, o2net_hb_down;
131#define O2NET_HB_PRI 0x1
132
133static struct o2net_handshake *o2net_hand;
134static struct o2net_msg *o2net_keep_req, *o2net_keep_resp;
135
136static int o2net_sys_err_translations[O2NET_ERR_MAX] =
137 {[O2NET_ERR_NONE] = 0,
138 [O2NET_ERR_NO_HNDLR] = -ENOPROTOOPT,
139 [O2NET_ERR_OVERFLOW] = -EOVERFLOW,
140 [O2NET_ERR_DIED] = -EHOSTDOWN,};
141
142/* can't quite avoid *all* internal declarations :/ */
143static void o2net_sc_connect_completed(void *arg);
144static void o2net_rx_until_empty(void *arg);
145static void o2net_shutdown_sc(void *arg);
146static void o2net_listen_data_ready(struct sock *sk, int bytes);
147static void o2net_sc_send_keep_req(void *arg);
148static void o2net_idle_timer(unsigned long data);
149static void o2net_sc_postpone_idle(struct o2net_sock_container *sc);
150
151static inline int o2net_sys_err_to_errno(enum o2net_system_error err)
152{
153 int trans;
154 BUG_ON(err >= O2NET_ERR_MAX);
155 trans = o2net_sys_err_translations[err];
156
157 /* Just in case we mess up the translation table above */
158 BUG_ON(err != O2NET_ERR_NONE && trans == 0);
159 return trans;
160}
161
162static struct o2net_node * o2net_nn_from_num(u8 node_num)
163{
164 BUG_ON(node_num >= ARRAY_SIZE(o2net_nodes));
165 return &o2net_nodes[node_num];
166}
167
168static u8 o2net_num_from_nn(struct o2net_node *nn)
169{
170 BUG_ON(nn == NULL);
171 return nn - o2net_nodes;
172}
173
174/* ------------------------------------------------------------ */
175
176static int o2net_prep_nsw(struct o2net_node *nn, struct o2net_status_wait *nsw)
177{
178 int ret = 0;
179
180 do {
181 if (!idr_pre_get(&nn->nn_status_idr, GFP_ATOMIC)) {
182 ret = -EAGAIN;
183 break;
184 }
185 spin_lock(&nn->nn_lock);
186 ret = idr_get_new(&nn->nn_status_idr, nsw, &nsw->ns_id);
187 if (ret == 0)
188 list_add_tail(&nsw->ns_node_item,
189 &nn->nn_status_list);
190 spin_unlock(&nn->nn_lock);
191 } while (ret == -EAGAIN);
192
193 if (ret == 0) {
194 init_waitqueue_head(&nsw->ns_wq);
195 nsw->ns_sys_status = O2NET_ERR_NONE;
196 nsw->ns_status = 0;
197 }
198
199 return ret;
200}
201
202static void o2net_complete_nsw_locked(struct o2net_node *nn,
203 struct o2net_status_wait *nsw,
204 enum o2net_system_error sys_status,
205 s32 status)
206{
207 assert_spin_locked(&nn->nn_lock);
208
209 if (!list_empty(&nsw->ns_node_item)) {
210 list_del_init(&nsw->ns_node_item);
211 nsw->ns_sys_status = sys_status;
212 nsw->ns_status = status;
213 idr_remove(&nn->nn_status_idr, nsw->ns_id);
214 wake_up(&nsw->ns_wq);
215 }
216}
217
218static void o2net_complete_nsw(struct o2net_node *nn,
219 struct o2net_status_wait *nsw,
220 u64 id, enum o2net_system_error sys_status,
221 s32 status)
222{
223 spin_lock(&nn->nn_lock);
224 if (nsw == NULL) {
225 if (id > INT_MAX)
226 goto out;
227
228 nsw = idr_find(&nn->nn_status_idr, id);
229 if (nsw == NULL)
230 goto out;
231 }
232
233 o2net_complete_nsw_locked(nn, nsw, sys_status, status);
234
235out:
236 spin_unlock(&nn->nn_lock);
237 return;
238}
239
240static void o2net_complete_nodes_nsw(struct o2net_node *nn)
241{
242 struct list_head *iter, *tmp;
243 unsigned int num_kills = 0;
244 struct o2net_status_wait *nsw;
245
246 assert_spin_locked(&nn->nn_lock);
247
248 list_for_each_safe(iter, tmp, &nn->nn_status_list) {
249 nsw = list_entry(iter, struct o2net_status_wait, ns_node_item);
250 o2net_complete_nsw_locked(nn, nsw, O2NET_ERR_DIED, 0);
251 num_kills++;
252 }
253
254 mlog(0, "completed %d messages for node %u\n", num_kills,
255 o2net_num_from_nn(nn));
256}
257
258static int o2net_nsw_completed(struct o2net_node *nn,
259 struct o2net_status_wait *nsw)
260{
261 int completed;
262 spin_lock(&nn->nn_lock);
263 completed = list_empty(&nsw->ns_node_item);
264 spin_unlock(&nn->nn_lock);
265 return completed;
266}
267
268/* ------------------------------------------------------------ */
269
270static void sc_kref_release(struct kref *kref)
271{
272 struct o2net_sock_container *sc = container_of(kref,
273 struct o2net_sock_container, sc_kref);
274 sclog(sc, "releasing\n");
275
276 if (sc->sc_sock) {
277 sock_release(sc->sc_sock);
278 sc->sc_sock = NULL;
279 }
280
281 o2nm_node_put(sc->sc_node);
282 sc->sc_node = NULL;
283
284 kfree(sc);
285}
286
287static void sc_put(struct o2net_sock_container *sc)
288{
289 sclog(sc, "put\n");
290 kref_put(&sc->sc_kref, sc_kref_release);
291}
292static void sc_get(struct o2net_sock_container *sc)
293{
294 sclog(sc, "get\n");
295 kref_get(&sc->sc_kref);
296}
297static struct o2net_sock_container *sc_alloc(struct o2nm_node *node)
298{
299 struct o2net_sock_container *sc, *ret = NULL;
300 struct page *page = NULL;
301
302 page = alloc_page(GFP_NOFS);
303 sc = kcalloc(1, sizeof(*sc), GFP_NOFS);
304 if (sc == NULL || page == NULL)
305 goto out;
306
307 kref_init(&sc->sc_kref);
308 o2nm_node_get(node);
309 sc->sc_node = node;
310
311 INIT_WORK(&sc->sc_connect_work, o2net_sc_connect_completed, sc);
312 INIT_WORK(&sc->sc_rx_work, o2net_rx_until_empty, sc);
313 INIT_WORK(&sc->sc_shutdown_work, o2net_shutdown_sc, sc);
314 INIT_WORK(&sc->sc_keepalive_work, o2net_sc_send_keep_req, sc);
315
316 init_timer(&sc->sc_idle_timeout);
317 sc->sc_idle_timeout.function = o2net_idle_timer;
318 sc->sc_idle_timeout.data = (unsigned long)sc;
319
320 sclog(sc, "alloced\n");
321
322 ret = sc;
323 sc->sc_page = page;
324 sc = NULL;
325 page = NULL;
326
327out:
328 if (page)
329 __free_page(page);
330 kfree(sc);
331
332 return ret;
333}
334
335/* ------------------------------------------------------------ */
336
337static void o2net_sc_queue_work(struct o2net_sock_container *sc,
338 struct work_struct *work)
339{
340 sc_get(sc);
341 if (!queue_work(o2net_wq, work))
342 sc_put(sc);
343}
344static void o2net_sc_queue_delayed_work(struct o2net_sock_container *sc,
345 struct work_struct *work,
346 int delay)
347{
348 sc_get(sc);
349 if (!queue_delayed_work(o2net_wq, work, delay))
350 sc_put(sc);
351}
352static void o2net_sc_cancel_delayed_work(struct o2net_sock_container *sc,
353 struct work_struct *work)
354{
355 if (cancel_delayed_work(work))
356 sc_put(sc);
357}
358
359static void o2net_set_nn_state(struct o2net_node *nn,
360 struct o2net_sock_container *sc,
361 unsigned valid, int err)
362{
363 int was_valid = nn->nn_sc_valid;
364 int was_err = nn->nn_persistent_error;
365 struct o2net_sock_container *old_sc = nn->nn_sc;
366
367 assert_spin_locked(&nn->nn_lock);
368
369 /* the node num comparison and single connect/accept path should stop
370 * an non-null sc from being overwritten with another */
371 BUG_ON(sc && nn->nn_sc && nn->nn_sc != sc);
372 mlog_bug_on_msg(err && valid, "err %d valid %u\n", err, valid);
373 mlog_bug_on_msg(valid && !sc, "valid %u sc %p\n", valid, sc);
374
375 /* we won't reconnect after our valid conn goes away for
376 * this hb iteration.. here so it shows up in the logs */
377 if (was_valid && !valid && err == 0)
378 err = -ENOTCONN;
379
380 mlog(ML_CONN, "node %u sc: %p -> %p, valid %u -> %u, err %d -> %d\n",
381 o2net_num_from_nn(nn), nn->nn_sc, sc, nn->nn_sc_valid, valid,
382 nn->nn_persistent_error, err);
383
384 nn->nn_sc = sc;
385 nn->nn_sc_valid = valid ? 1 : 0;
386 nn->nn_persistent_error = err;
387
388 /* mirrors o2net_tx_can_proceed() */
389 if (nn->nn_persistent_error || nn->nn_sc_valid)
390 wake_up(&nn->nn_sc_wq);
391
392 if (!was_err && nn->nn_persistent_error) {
393 o2quo_conn_err(o2net_num_from_nn(nn));
394 queue_delayed_work(o2net_wq, &nn->nn_still_up,
395 msecs_to_jiffies(O2NET_QUORUM_DELAY_MS));
396 }
397
398 if (was_valid && !valid) {
399 mlog(ML_NOTICE, "no longer connected to " SC_NODEF_FMT "\n",
400 SC_NODEF_ARGS(old_sc));
401 o2net_complete_nodes_nsw(nn);
402 }
403
404 if (!was_valid && valid) {
405 o2quo_conn_up(o2net_num_from_nn(nn));
406 /* this is a bit of a hack. we only try reconnecting
407 * when heartbeating starts until we get a connection.
408 * if that connection then dies we don't try reconnecting.
409 * the only way to start connecting again is to down
410 * heartbeat and bring it back up. */
411 cancel_delayed_work(&nn->nn_connect_expired);
412 mlog(ML_NOTICE, "%s " SC_NODEF_FMT "\n",
413 o2nm_this_node() > sc->sc_node->nd_num ?
414 "connected to" : "accepted connection from",
415 SC_NODEF_ARGS(sc));
416 }
417
418 /* trigger the connecting worker func as long as we're not valid,
419 * it will back off if it shouldn't connect. This can be called
420 * from node config teardown and so needs to be careful about
421 * the work queue actually being up. */
422 if (!valid && o2net_wq) {
423 unsigned long delay;
424 /* delay if we're withing a RECONNECT_DELAY of the
425 * last attempt */
426 delay = (nn->nn_last_connect_attempt +
427 msecs_to_jiffies(O2NET_RECONNECT_DELAY_MS))
428 - jiffies;
429 if (delay > msecs_to_jiffies(O2NET_RECONNECT_DELAY_MS))
430 delay = 0;
431 mlog(ML_CONN, "queueing conn attempt in %lu jiffies\n", delay);
432 queue_delayed_work(o2net_wq, &nn->nn_connect_work, delay);
433 }
434
435 /* keep track of the nn's sc ref for the caller */
436 if ((old_sc == NULL) && sc)
437 sc_get(sc);
438 if (old_sc && (old_sc != sc)) {
439 o2net_sc_queue_work(old_sc, &old_sc->sc_shutdown_work);
440 sc_put(old_sc);
441 }
442}
443
444/* see o2net_register_callbacks() */
445static void o2net_data_ready(struct sock *sk, int bytes)
446{
447 void (*ready)(struct sock *sk, int bytes);
448
449 read_lock(&sk->sk_callback_lock);
450 if (sk->sk_user_data) {
451 struct o2net_sock_container *sc = sk->sk_user_data;
452 sclog(sc, "data_ready hit\n");
453 do_gettimeofday(&sc->sc_tv_data_ready);
454 o2net_sc_queue_work(sc, &sc->sc_rx_work);
455 ready = sc->sc_data_ready;
456 } else {
457 ready = sk->sk_data_ready;
458 }
459 read_unlock(&sk->sk_callback_lock);
460
461 ready(sk, bytes);
462}
463
464/* see o2net_register_callbacks() */
465static void o2net_state_change(struct sock *sk)
466{
467 void (*state_change)(struct sock *sk);
468 struct o2net_sock_container *sc;
469
470 read_lock(&sk->sk_callback_lock);
471 sc = sk->sk_user_data;
472 if (sc == NULL) {
473 state_change = sk->sk_state_change;
474 goto out;
475 }
476
477 sclog(sc, "state_change to %d\n", sk->sk_state);
478
479 state_change = sc->sc_state_change;
480
481 switch(sk->sk_state) {
482 /* ignore connecting sockets as they make progress */
483 case TCP_SYN_SENT:
484 case TCP_SYN_RECV:
485 break;
486 case TCP_ESTABLISHED:
487 o2net_sc_queue_work(sc, &sc->sc_connect_work);
488 break;
489 default:
490 o2net_sc_queue_work(sc, &sc->sc_shutdown_work);
491 break;
492 }
493out:
494 read_unlock(&sk->sk_callback_lock);
495 state_change(sk);
496}
497
498/*
499 * we register callbacks so we can queue work on events before calling
500 * the original callbacks. our callbacks our careful to test user_data
501 * to discover when they've reaced with o2net_unregister_callbacks().
502 */
503static void o2net_register_callbacks(struct sock *sk,
504 struct o2net_sock_container *sc)
505{
506 write_lock_bh(&sk->sk_callback_lock);
507
508 /* accepted sockets inherit the old listen socket data ready */
509 if (sk->sk_data_ready == o2net_listen_data_ready) {
510 sk->sk_data_ready = sk->sk_user_data;
511 sk->sk_user_data = NULL;
512 }
513
514 BUG_ON(sk->sk_user_data != NULL);
515 sk->sk_user_data = sc;
516 sc_get(sc);
517
518 sc->sc_data_ready = sk->sk_data_ready;
519 sc->sc_state_change = sk->sk_state_change;
520 sk->sk_data_ready = o2net_data_ready;
521 sk->sk_state_change = o2net_state_change;
522
523 write_unlock_bh(&sk->sk_callback_lock);
524}
525
526static int o2net_unregister_callbacks(struct sock *sk,
527 struct o2net_sock_container *sc)
528{
529 int ret = 0;
530
531 write_lock_bh(&sk->sk_callback_lock);
532 if (sk->sk_user_data == sc) {
533 ret = 1;
534 sk->sk_user_data = NULL;
535 sk->sk_data_ready = sc->sc_data_ready;
536 sk->sk_state_change = sc->sc_state_change;
537 }
538 write_unlock_bh(&sk->sk_callback_lock);
539
540 return ret;
541}
542
543/*
544 * this is a little helper that is called by callers who have seen a problem
545 * with an sc and want to detach it from the nn if someone already hasn't beat
546 * them to it. if an error is given then the shutdown will be persistent
547 * and pending transmits will be canceled.
548 */
549static void o2net_ensure_shutdown(struct o2net_node *nn,
550 struct o2net_sock_container *sc,
551 int err)
552{
553 spin_lock(&nn->nn_lock);
554 if (nn->nn_sc == sc)
555 o2net_set_nn_state(nn, NULL, 0, err);
556 spin_unlock(&nn->nn_lock);
557}
558
559/*
560 * This work queue function performs the blocking parts of socket shutdown. A
561 * few paths lead here. set_nn_state will trigger this callback if it sees an
562 * sc detached from the nn. state_change will also trigger this callback
563 * directly when it sees errors. In that case we need to call set_nn_state
564 * ourselves as state_change couldn't get the nn_lock and call set_nn_state
565 * itself.
566 */
567static void o2net_shutdown_sc(void *arg)
568{
569 struct o2net_sock_container *sc = arg;
570 struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
571
572 sclog(sc, "shutting down\n");
573
574 /* drop the callbacks ref and call shutdown only once */
575 if (o2net_unregister_callbacks(sc->sc_sock->sk, sc)) {
576 /* we shouldn't flush as we're in the thread, the
577 * races with pending sc work structs are harmless */
578 del_timer_sync(&sc->sc_idle_timeout);
579 o2net_sc_cancel_delayed_work(sc, &sc->sc_keepalive_work);
580 sc_put(sc);
581 sc->sc_sock->ops->shutdown(sc->sc_sock,
582 RCV_SHUTDOWN|SEND_SHUTDOWN);
583 }
584
585 /* not fatal so failed connects before the other guy has our
586 * heartbeat can be retried */
587 o2net_ensure_shutdown(nn, sc, 0);
588 sc_put(sc);
589}
590
591/* ------------------------------------------------------------ */
592
593static int o2net_handler_cmp(struct o2net_msg_handler *nmh, u32 msg_type,
594 u32 key)
595{
596 int ret = memcmp(&nmh->nh_key, &key, sizeof(key));
597
598 if (ret == 0)
599 ret = memcmp(&nmh->nh_msg_type, &msg_type, sizeof(msg_type));
600
601 return ret;
602}
603
604static struct o2net_msg_handler *
605o2net_handler_tree_lookup(u32 msg_type, u32 key, struct rb_node ***ret_p,
606 struct rb_node **ret_parent)
607{
608 struct rb_node **p = &o2net_handler_tree.rb_node;
609 struct rb_node *parent = NULL;
610 struct o2net_msg_handler *nmh, *ret = NULL;
611 int cmp;
612
613 while (*p) {
614 parent = *p;
615 nmh = rb_entry(parent, struct o2net_msg_handler, nh_node);
616 cmp = o2net_handler_cmp(nmh, msg_type, key);
617
618 if (cmp < 0)
619 p = &(*p)->rb_left;
620 else if (cmp > 0)
621 p = &(*p)->rb_right;
622 else {
623 ret = nmh;
624 break;
625 }
626 }
627
628 if (ret_p != NULL)
629 *ret_p = p;
630 if (ret_parent != NULL)
631 *ret_parent = parent;
632
633 return ret;
634}
635
636static void o2net_handler_kref_release(struct kref *kref)
637{
638 struct o2net_msg_handler *nmh;
639 nmh = container_of(kref, struct o2net_msg_handler, nh_kref);
640
641 kfree(nmh);
642}
643
644static void o2net_handler_put(struct o2net_msg_handler *nmh)
645{
646 kref_put(&nmh->nh_kref, o2net_handler_kref_release);
647}
648
649/* max_len is protection for the handler func. incoming messages won't
650 * be given to the handler if their payload is longer than the max. */
651int o2net_register_handler(u32 msg_type, u32 key, u32 max_len,
652 o2net_msg_handler_func *func, void *data,
653 struct list_head *unreg_list)
654{
655 struct o2net_msg_handler *nmh = NULL;
656 struct rb_node **p, *parent;
657 int ret = 0;
658
659 if (max_len > O2NET_MAX_PAYLOAD_BYTES) {
660 mlog(0, "max_len for message handler out of range: %u\n",
661 max_len);
662 ret = -EINVAL;
663 goto out;
664 }
665
666 if (!msg_type) {
667 mlog(0, "no message type provided: %u, %p\n", msg_type, func);
668 ret = -EINVAL;
669 goto out;
670
671 }
672 if (!func) {
673 mlog(0, "no message handler provided: %u, %p\n",
674 msg_type, func);
675 ret = -EINVAL;
676 goto out;
677 }
678
679 nmh = kcalloc(1, sizeof(struct o2net_msg_handler), GFP_NOFS);
680 if (nmh == NULL) {
681 ret = -ENOMEM;
682 goto out;
683 }
684
685 nmh->nh_func = func;
686 nmh->nh_func_data = data;
687 nmh->nh_msg_type = msg_type;
688 nmh->nh_max_len = max_len;
689 nmh->nh_key = key;
690 /* the tree and list get this ref.. they're both removed in
691 * unregister when this ref is dropped */
692 kref_init(&nmh->nh_kref);
693 INIT_LIST_HEAD(&nmh->nh_unregister_item);
694
695 write_lock(&o2net_handler_lock);
696 if (o2net_handler_tree_lookup(msg_type, key, &p, &parent))
697 ret = -EEXIST;
698 else {
699 rb_link_node(&nmh->nh_node, parent, p);
700 rb_insert_color(&nmh->nh_node, &o2net_handler_tree);
701 list_add_tail(&nmh->nh_unregister_item, unreg_list);
702
703 mlog(ML_TCP, "registered handler func %p type %u key %08x\n",
704 func, msg_type, key);
705 /* we've had some trouble with handlers seemingly vanishing. */
706 mlog_bug_on_msg(o2net_handler_tree_lookup(msg_type, key, &p,
707 &parent) == NULL,
708 "couldn't find handler we *just* registerd "
709 "for type %u key %08x\n", msg_type, key);
710 }
711 write_unlock(&o2net_handler_lock);
712 if (ret)
713 goto out;
714
715out:
716 if (ret)
717 kfree(nmh);
718
719 return ret;
720}
721EXPORT_SYMBOL_GPL(o2net_register_handler);
722
723void o2net_unregister_handler_list(struct list_head *list)
724{
725 struct list_head *pos, *n;
726 struct o2net_msg_handler *nmh;
727
728 write_lock(&o2net_handler_lock);
729 list_for_each_safe(pos, n, list) {
730 nmh = list_entry(pos, struct o2net_msg_handler,
731 nh_unregister_item);
732 mlog(ML_TCP, "unregistering handler func %p type %u key %08x\n",
733 nmh->nh_func, nmh->nh_msg_type, nmh->nh_key);
734 rb_erase(&nmh->nh_node, &o2net_handler_tree);
735 list_del_init(&nmh->nh_unregister_item);
736 kref_put(&nmh->nh_kref, o2net_handler_kref_release);
737 }
738 write_unlock(&o2net_handler_lock);
739}
740EXPORT_SYMBOL_GPL(o2net_unregister_handler_list);
741
742static struct o2net_msg_handler *o2net_handler_get(u32 msg_type, u32 key)
743{
744 struct o2net_msg_handler *nmh;
745
746 read_lock(&o2net_handler_lock);
747 nmh = o2net_handler_tree_lookup(msg_type, key, NULL, NULL);
748 if (nmh)
749 kref_get(&nmh->nh_kref);
750 read_unlock(&o2net_handler_lock);
751
752 return nmh;
753}
754
755/* ------------------------------------------------------------ */
756
757static int o2net_recv_tcp_msg(struct socket *sock, void *data, size_t len)
758{
759 int ret;
760 mm_segment_t oldfs;
761 struct kvec vec = {
762 .iov_len = len,
763 .iov_base = data,
764 };
765 struct msghdr msg = {
766 .msg_iovlen = 1,
767 .msg_iov = (struct iovec *)&vec,
768 .msg_flags = MSG_DONTWAIT,
769 };
770
771 oldfs = get_fs();
772 set_fs(get_ds());
773 ret = sock_recvmsg(sock, &msg, len, msg.msg_flags);
774 set_fs(oldfs);
775
776 return ret;
777}
778
779static int o2net_send_tcp_msg(struct socket *sock, struct kvec *vec,
780 size_t veclen, size_t total)
781{
782 int ret;
783 mm_segment_t oldfs;
784 struct msghdr msg = {
785 .msg_iov = (struct iovec *)vec,
786 .msg_iovlen = veclen,
787 };
788
789 if (sock == NULL) {
790 ret = -EINVAL;
791 goto out;
792 }
793
794 oldfs = get_fs();
795 set_fs(get_ds());
796 ret = sock_sendmsg(sock, &msg, total);
797 set_fs(oldfs);
798 if (ret != total) {
799 mlog(ML_ERROR, "sendmsg returned %d instead of %zu\n", ret,
800 total);
801 if (ret >= 0)
802 ret = -EPIPE; /* should be smarter, I bet */
803 goto out;
804 }
805
806 ret = 0;
807out:
808 if (ret < 0)
809 mlog(0, "returning error: %d\n", ret);
810 return ret;
811}
812
813static void o2net_sendpage(struct o2net_sock_container *sc,
814 void *kmalloced_virt,
815 size_t size)
816{
817 struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
818 ssize_t ret;
819
820
821 ret = sc->sc_sock->ops->sendpage(sc->sc_sock,
822 virt_to_page(kmalloced_virt),
823 (long)kmalloced_virt & ~PAGE_MASK,
824 size, MSG_DONTWAIT);
825 if (ret != size) {
826 mlog(ML_ERROR, "sendpage of size %zu to " SC_NODEF_FMT
827 " failed with %zd\n", size, SC_NODEF_ARGS(sc), ret);
828 o2net_ensure_shutdown(nn, sc, 0);
829 }
830}
831
832static void o2net_init_msg(struct o2net_msg *msg, u16 data_len, u16 msg_type, u32 key)
833{
834 memset(msg, 0, sizeof(struct o2net_msg));
835 msg->magic = cpu_to_be16(O2NET_MSG_MAGIC);
836 msg->data_len = cpu_to_be16(data_len);
837 msg->msg_type = cpu_to_be16(msg_type);
838 msg->sys_status = cpu_to_be32(O2NET_ERR_NONE);
839 msg->status = 0;
840 msg->key = cpu_to_be32(key);
841}
842
843static int o2net_tx_can_proceed(struct o2net_node *nn,
844 struct o2net_sock_container **sc_ret,
845 int *error)
846{
847 int ret = 0;
848
849 spin_lock(&nn->nn_lock);
850 if (nn->nn_persistent_error) {
851 ret = 1;
852 *sc_ret = NULL;
853 *error = nn->nn_persistent_error;
854 } else if (nn->nn_sc_valid) {
855 kref_get(&nn->nn_sc->sc_kref);
856
857 ret = 1;
858 *sc_ret = nn->nn_sc;
859 *error = 0;
860 }
861 spin_unlock(&nn->nn_lock);
862
863 return ret;
864}
865
866int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *caller_vec,
867 size_t caller_veclen, u8 target_node, int *status)
868{
869 int ret, error = 0;
870 struct o2net_msg *msg = NULL;
871 size_t veclen, caller_bytes = 0;
872 struct kvec *vec = NULL;
873 struct o2net_sock_container *sc = NULL;
874 struct o2net_node *nn = o2net_nn_from_num(target_node);
875 struct o2net_status_wait nsw = {
876 .ns_node_item = LIST_HEAD_INIT(nsw.ns_node_item),
877 };
878
879 if (o2net_wq == NULL) {
880 mlog(0, "attempt to tx without o2netd running\n");
881 ret = -ESRCH;
882 goto out;
883 }
884
885 if (caller_veclen == 0) {
886 mlog(0, "bad kvec array length\n");
887 ret = -EINVAL;
888 goto out;
889 }
890
891 caller_bytes = iov_length((struct iovec *)caller_vec, caller_veclen);
892 if (caller_bytes > O2NET_MAX_PAYLOAD_BYTES) {
893 mlog(0, "total payload len %zu too large\n", caller_bytes);
894 ret = -EINVAL;
895 goto out;
896 }
897
898 if (target_node == o2nm_this_node()) {
899 ret = -ELOOP;
900 goto out;
901 }
902
903 ret = wait_event_interruptible(nn->nn_sc_wq,
904 o2net_tx_can_proceed(nn, &sc, &error));
905 if (!ret && error)
906 ret = error;
907 if (ret)
908 goto out;
909
910 veclen = caller_veclen + 1;
911 vec = kmalloc(sizeof(struct kvec) * veclen, GFP_ATOMIC);
912 if (vec == NULL) {
913 mlog(0, "failed to %zu element kvec!\n", veclen);
914 ret = -ENOMEM;
915 goto out;
916 }
917
918 msg = kmalloc(sizeof(struct o2net_msg), GFP_ATOMIC);
919 if (!msg) {
920 mlog(0, "failed to allocate a o2net_msg!\n");
921 ret = -ENOMEM;
922 goto out;
923 }
924
925 o2net_init_msg(msg, caller_bytes, msg_type, key);
926
927 vec[0].iov_len = sizeof(struct o2net_msg);
928 vec[0].iov_base = msg;
929 memcpy(&vec[1], caller_vec, caller_veclen * sizeof(struct kvec));
930
931 ret = o2net_prep_nsw(nn, &nsw);
932 if (ret)
933 goto out;
934
935 msg->msg_num = cpu_to_be32(nsw.ns_id);
936
937 /* finally, convert the message header to network byte-order
938 * and send */
939 ret = o2net_send_tcp_msg(sc->sc_sock, vec, veclen,
940 sizeof(struct o2net_msg) + caller_bytes);
941 msglog(msg, "sending returned %d\n", ret);
942 if (ret < 0) {
943 mlog(0, "error returned from o2net_send_tcp_msg=%d\n", ret);
944 goto out;
945 }
946
947 /* wait on other node's handler */
948 wait_event(nsw.ns_wq, o2net_nsw_completed(nn, &nsw));
949
950 /* Note that we avoid overwriting the callers status return
951 * variable if a system error was reported on the other
952 * side. Callers beware. */
953 ret = o2net_sys_err_to_errno(nsw.ns_sys_status);
954 if (status && !ret)
955 *status = nsw.ns_status;
956
957 mlog(0, "woken, returning system status %d, user status %d\n",
958 ret, nsw.ns_status);
959out:
960 if (sc)
961 sc_put(sc);
962 if (vec)
963 kfree(vec);
964 if (msg)
965 kfree(msg);
966 o2net_complete_nsw(nn, &nsw, 0, 0, 0);
967 return ret;
968}
969EXPORT_SYMBOL_GPL(o2net_send_message_vec);
970
971int o2net_send_message(u32 msg_type, u32 key, void *data, u32 len,
972 u8 target_node, int *status)
973{
974 struct kvec vec = {
975 .iov_base = data,
976 .iov_len = len,
977 };
978 return o2net_send_message_vec(msg_type, key, &vec, 1,
979 target_node, status);
980}
981EXPORT_SYMBOL_GPL(o2net_send_message);
982
983static int o2net_send_status_magic(struct socket *sock, struct o2net_msg *hdr,
984 enum o2net_system_error syserr, int err)
985{
986 struct kvec vec = {
987 .iov_base = hdr,
988 .iov_len = sizeof(struct o2net_msg),
989 };
990
991 BUG_ON(syserr >= O2NET_ERR_MAX);
992
993 /* leave other fields intact from the incoming message, msg_num
994 * in particular */
995 hdr->sys_status = cpu_to_be32(syserr);
996 hdr->status = cpu_to_be32(err);
997 hdr->magic = cpu_to_be16(O2NET_MSG_STATUS_MAGIC); // twiddle the magic
998 hdr->data_len = 0;
999
1000 msglog(hdr, "about to send status magic %d\n", err);
1001 /* hdr has been in host byteorder this whole time */
1002 return o2net_send_tcp_msg(sock, &vec, 1, sizeof(struct o2net_msg));
1003}
1004
1005/* this returns -errno if the header was unknown or too large, etc.
1006 * after this is called the buffer us reused for the next message */
1007static int o2net_process_message(struct o2net_sock_container *sc,
1008 struct o2net_msg *hdr)
1009{
1010 struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
1011 int ret = 0, handler_status;
1012 enum o2net_system_error syserr;
1013 struct o2net_msg_handler *nmh = NULL;
1014
1015 msglog(hdr, "processing message\n");
1016
1017 o2net_sc_postpone_idle(sc);
1018
1019 switch(be16_to_cpu(hdr->magic)) {
1020 case O2NET_MSG_STATUS_MAGIC:
1021 /* special type for returning message status */
1022 o2net_complete_nsw(nn, NULL,
1023 be32_to_cpu(hdr->msg_num),
1024 be32_to_cpu(hdr->sys_status),
1025 be32_to_cpu(hdr->status));
1026 goto out;
1027 case O2NET_MSG_KEEP_REQ_MAGIC:
1028 o2net_sendpage(sc, o2net_keep_resp,
1029 sizeof(*o2net_keep_resp));
1030 goto out;
1031 case O2NET_MSG_KEEP_RESP_MAGIC:
1032 goto out;
1033 case O2NET_MSG_MAGIC:
1034 break;
1035 default:
1036 msglog(hdr, "bad magic\n");
1037 ret = -EINVAL;
1038 goto out;
1039 break;
1040 }
1041
1042 /* find a handler for it */
1043 handler_status = 0;
1044 nmh = o2net_handler_get(be16_to_cpu(hdr->msg_type),
1045 be32_to_cpu(hdr->key));
1046 if (!nmh) {
1047 mlog(ML_TCP, "couldn't find handler for type %u key %08x\n",
1048 be16_to_cpu(hdr->msg_type), be32_to_cpu(hdr->key));
1049 syserr = O2NET_ERR_NO_HNDLR;
1050 goto out_respond;
1051 }
1052
1053 syserr = O2NET_ERR_NONE;
1054
1055 if (be16_to_cpu(hdr->data_len) > nmh->nh_max_len)
1056 syserr = O2NET_ERR_OVERFLOW;
1057
1058 if (syserr != O2NET_ERR_NONE)
1059 goto out_respond;
1060
1061 do_gettimeofday(&sc->sc_tv_func_start);
1062 sc->sc_msg_key = be32_to_cpu(hdr->key);
1063 sc->sc_msg_type = be16_to_cpu(hdr->msg_type);
1064 handler_status = (nmh->nh_func)(hdr, sizeof(struct o2net_msg) +
1065 be16_to_cpu(hdr->data_len),
1066 nmh->nh_func_data);
1067 do_gettimeofday(&sc->sc_tv_func_stop);
1068
1069out_respond:
1070 /* this destroys the hdr, so don't use it after this */
1071 ret = o2net_send_status_magic(sc->sc_sock, hdr, syserr,
1072 handler_status);
1073 hdr = NULL;
1074 mlog(0, "sending handler status %d, syserr %d returned %d\n",
1075 handler_status, syserr, ret);
1076
1077out:
1078 if (nmh)
1079 o2net_handler_put(nmh);
1080 return ret;
1081}
1082
1083static int o2net_check_handshake(struct o2net_sock_container *sc)
1084{
1085 struct o2net_handshake *hand = page_address(sc->sc_page);
1086 struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
1087
1088 if (hand->protocol_version != cpu_to_be64(O2NET_PROTOCOL_VERSION)) {
1089 mlog(ML_NOTICE, SC_NODEF_FMT " advertised net protocol "
1090 "version %llu but %llu is required, disconnecting\n",
1091 SC_NODEF_ARGS(sc),
1092 (unsigned long long)be64_to_cpu(hand->protocol_version),
1093 O2NET_PROTOCOL_VERSION);
1094
1095 /* don't bother reconnecting if its the wrong version. */
1096 o2net_ensure_shutdown(nn, sc, -ENOTCONN);
1097 return -1;
1098 }
1099
1100 sc->sc_handshake_ok = 1;
1101
1102 spin_lock(&nn->nn_lock);
1103 /* set valid and queue the idle timers only if it hasn't been
1104 * shut down already */
1105 if (nn->nn_sc == sc) {
1106 o2net_sc_postpone_idle(sc);
1107 o2net_set_nn_state(nn, sc, 1, 0);
1108 }
1109 spin_unlock(&nn->nn_lock);
1110
1111 /* shift everything up as though it wasn't there */
1112 sc->sc_page_off -= sizeof(struct o2net_handshake);
1113 if (sc->sc_page_off)
1114 memmove(hand, hand + 1, sc->sc_page_off);
1115
1116 return 0;
1117}
1118
1119/* this demuxes the queued rx bytes into header or payload bits and calls
1120 * handlers as each full message is read off the socket. it returns -error,
1121 * == 0 eof, or > 0 for progress made.*/
1122static int o2net_advance_rx(struct o2net_sock_container *sc)
1123{
1124 struct o2net_msg *hdr;
1125 int ret = 0;
1126 void *data;
1127 size_t datalen;
1128
1129 sclog(sc, "receiving\n");
1130 do_gettimeofday(&sc->sc_tv_advance_start);
1131
1132 /* do we need more header? */
1133 if (sc->sc_page_off < sizeof(struct o2net_msg)) {
1134 data = page_address(sc->sc_page) + sc->sc_page_off;
1135 datalen = sizeof(struct o2net_msg) - sc->sc_page_off;
1136 ret = o2net_recv_tcp_msg(sc->sc_sock, data, datalen);
1137 if (ret > 0) {
1138 sc->sc_page_off += ret;
1139
1140 /* this working relies on the handshake being
1141 * smaller than the normal message header */
1142 if (sc->sc_page_off >= sizeof(struct o2net_handshake)&&
1143 !sc->sc_handshake_ok && o2net_check_handshake(sc)) {
1144 ret = -EPROTO;
1145 goto out;
1146 }
1147
1148 /* only swab incoming here.. we can
1149 * only get here once as we cross from
1150 * being under to over */
1151 if (sc->sc_page_off == sizeof(struct o2net_msg)) {
1152 hdr = page_address(sc->sc_page);
1153 if (be16_to_cpu(hdr->data_len) >
1154 O2NET_MAX_PAYLOAD_BYTES)
1155 ret = -EOVERFLOW;
1156 }
1157 }
1158 if (ret <= 0)
1159 goto out;
1160 }
1161
1162 if (sc->sc_page_off < sizeof(struct o2net_msg)) {
1163 /* oof, still don't have a header */
1164 goto out;
1165 }
1166
1167 /* this was swabbed above when we first read it */
1168 hdr = page_address(sc->sc_page);
1169
1170 msglog(hdr, "at page_off %zu\n", sc->sc_page_off);
1171
1172 /* do we need more payload? */
1173 if (sc->sc_page_off - sizeof(struct o2net_msg) < be16_to_cpu(hdr->data_len)) {
1174 /* need more payload */
1175 data = page_address(sc->sc_page) + sc->sc_page_off;
1176 datalen = (sizeof(struct o2net_msg) + be16_to_cpu(hdr->data_len)) -
1177 sc->sc_page_off;
1178 ret = o2net_recv_tcp_msg(sc->sc_sock, data, datalen);
1179 if (ret > 0)
1180 sc->sc_page_off += ret;
1181 if (ret <= 0)
1182 goto out;
1183 }
1184
1185 if (sc->sc_page_off - sizeof(struct o2net_msg) == be16_to_cpu(hdr->data_len)) {
1186 /* we can only get here once, the first time we read
1187 * the payload.. so set ret to progress if the handler
1188 * works out. after calling this the message is toast */
1189 ret = o2net_process_message(sc, hdr);
1190 if (ret == 0)
1191 ret = 1;
1192 sc->sc_page_off = 0;
1193 }
1194
1195out:
1196 sclog(sc, "ret = %d\n", ret);
1197 do_gettimeofday(&sc->sc_tv_advance_stop);
1198 return ret;
1199}
1200
1201/* this work func is triggerd by data ready. it reads until it can read no
1202 * more. it interprets 0, eof, as fatal. if data_ready hits while we're doing
1203 * our work the work struct will be marked and we'll be called again. */
1204static void o2net_rx_until_empty(void *arg)
1205{
1206 struct o2net_sock_container *sc = arg;
1207 int ret;
1208
1209 do {
1210 ret = o2net_advance_rx(sc);
1211 } while (ret > 0);
1212
1213 if (ret <= 0 && ret != -EAGAIN) {
1214 struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
1215 sclog(sc, "saw error %d, closing\n", ret);
1216 /* not permanent so read failed handshake can retry */
1217 o2net_ensure_shutdown(nn, sc, 0);
1218 }
1219
1220 sc_put(sc);
1221}
1222
1223static int o2net_set_nodelay(struct socket *sock)
1224{
1225 int ret, val = 1;
1226 mm_segment_t oldfs;
1227
1228 oldfs = get_fs();
1229 set_fs(KERNEL_DS);
1230
1231 /*
1232 * Dear unsuspecting programmer,
1233 *
1234 * Don't use sock_setsockopt() for SOL_TCP. It doesn't check its level
1235 * argument and assumes SOL_SOCKET so, say, your TCP_NODELAY will
1236 * silently turn into SO_DEBUG.
1237 *
1238 * Yours,
1239 * Keeper of hilariously fragile interfaces.
1240 */
1241 ret = sock->ops->setsockopt(sock, SOL_TCP, TCP_NODELAY,
1242 (char __user *)&val, sizeof(val));
1243
1244 set_fs(oldfs);
1245 return ret;
1246}
1247
1248/* ------------------------------------------------------------ */
1249
1250/* called when a connect completes and after a sock is accepted. the
1251 * rx path will see the response and mark the sc valid */
1252static void o2net_sc_connect_completed(void *arg)
1253{
1254 struct o2net_sock_container *sc = arg;
1255
1256 mlog(ML_MSG, "sc sending handshake with ver %llu id %llx\n",
1257 (unsigned long long)O2NET_PROTOCOL_VERSION,
1258 (unsigned long long)be64_to_cpu(o2net_hand->connector_id));
1259
1260 o2net_sendpage(sc, o2net_hand, sizeof(*o2net_hand));
1261 sc_put(sc);
1262}
1263
1264/* this is called as a work_struct func. */
1265static void o2net_sc_send_keep_req(void *arg)
1266{
1267 struct o2net_sock_container *sc = arg;
1268
1269 o2net_sendpage(sc, o2net_keep_req, sizeof(*o2net_keep_req));
1270 sc_put(sc);
1271}
1272
1273/* socket shutdown does a del_timer_sync against this as it tears down.
1274 * we can't start this timer until we've got to the point in sc buildup
1275 * where shutdown is going to be involved */
1276static void o2net_idle_timer(unsigned long data)
1277{
1278 struct o2net_sock_container *sc = (struct o2net_sock_container *)data;
1279 struct timeval now;
1280
1281 do_gettimeofday(&now);
1282
1283 mlog(ML_NOTICE, "connection to " SC_NODEF_FMT " has been idle for 10 "
1284 "seconds, shutting it down.\n", SC_NODEF_ARGS(sc));
1285 mlog(ML_NOTICE, "here are some times that might help debug the "
1286 "situation: (tmr %ld.%ld now %ld.%ld dr %ld.%ld adv "
1287 "%ld.%ld:%ld.%ld func (%08x:%u) %ld.%ld:%ld.%ld)\n",
1288 sc->sc_tv_timer.tv_sec, sc->sc_tv_timer.tv_usec,
1289 now.tv_sec, now.tv_usec,
1290 sc->sc_tv_data_ready.tv_sec, sc->sc_tv_data_ready.tv_usec,
1291 sc->sc_tv_advance_start.tv_sec, sc->sc_tv_advance_start.tv_usec,
1292 sc->sc_tv_advance_stop.tv_sec, sc->sc_tv_advance_stop.tv_usec,
1293 sc->sc_msg_key, sc->sc_msg_type,
1294 sc->sc_tv_func_start.tv_sec, sc->sc_tv_func_start.tv_usec,
1295 sc->sc_tv_func_stop.tv_sec, sc->sc_tv_func_stop.tv_usec);
1296
1297 o2net_sc_queue_work(sc, &sc->sc_shutdown_work);
1298}
1299
1300static void o2net_sc_postpone_idle(struct o2net_sock_container *sc)
1301{
1302 o2net_sc_cancel_delayed_work(sc, &sc->sc_keepalive_work);
1303 o2net_sc_queue_delayed_work(sc, &sc->sc_keepalive_work,
1304 O2NET_KEEPALIVE_DELAY_SECS * HZ);
1305 do_gettimeofday(&sc->sc_tv_timer);
1306 mod_timer(&sc->sc_idle_timeout,
1307 jiffies + (O2NET_IDLE_TIMEOUT_SECS * HZ));
1308}
1309
1310/* this work func is kicked whenever a path sets the nn state which doesn't
1311 * have valid set. This includes seeing hb come up, losing a connection,
1312 * having a connect attempt fail, etc. This centralizes the logic which decides
1313 * if a connect attempt should be made or if we should give up and all future
1314 * transmit attempts should fail */
1315static void o2net_start_connect(void *arg)
1316{
1317 struct o2net_node *nn = arg;
1318 struct o2net_sock_container *sc = NULL;
1319 struct o2nm_node *node = NULL;
1320 struct socket *sock = NULL;
1321 struct sockaddr_in myaddr = {0, }, remoteaddr = {0, };
1322 int ret = 0;
1323
1324 /* if we're greater we initiate tx, otherwise we accept */
1325 if (o2nm_this_node() <= o2net_num_from_nn(nn))
1326 goto out;
1327
1328 /* watch for racing with tearing a node down */
1329 node = o2nm_get_node_by_num(o2net_num_from_nn(nn));
1330 if (node == NULL) {
1331 ret = 0;
1332 goto out;
1333 }
1334
1335 spin_lock(&nn->nn_lock);
1336 /* see if we already have one pending or have given up */
1337 if (nn->nn_sc || nn->nn_persistent_error)
1338 arg = NULL;
1339 spin_unlock(&nn->nn_lock);
1340 if (arg == NULL) /* *shrug*, needed some indicator */
1341 goto out;
1342
1343 nn->nn_last_connect_attempt = jiffies;
1344
1345 sc = sc_alloc(node);
1346 if (sc == NULL) {
1347 mlog(0, "couldn't allocate sc\n");
1348 ret = -ENOMEM;
1349 goto out;
1350 }
1351
1352 ret = sock_create(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
1353 if (ret < 0) {
1354 mlog(0, "can't create socket: %d\n", ret);
1355 goto out;
1356 }
1357 sc->sc_sock = sock; /* freed by sc_kref_release */
1358
1359 sock->sk->sk_allocation = GFP_ATOMIC;
1360
1361 myaddr.sin_family = AF_INET;
1362 myaddr.sin_port = (__force u16)htons(0); /* any port */
1363
1364 ret = sock->ops->bind(sock, (struct sockaddr *)&myaddr,
1365 sizeof(myaddr));
1366 if (ret) {
1367 mlog(0, "bind failed: %d\n", ret);
1368 goto out;
1369 }
1370
1371 ret = o2net_set_nodelay(sc->sc_sock);
1372 if (ret) {
1373 mlog(ML_ERROR, "setting TCP_NODELAY failed with %d\n", ret);
1374 goto out;
1375 }
1376
1377 o2net_register_callbacks(sc->sc_sock->sk, sc);
1378
1379 spin_lock(&nn->nn_lock);
1380 /* handshake completion will set nn->nn_sc_valid */
1381 o2net_set_nn_state(nn, sc, 0, 0);
1382 spin_unlock(&nn->nn_lock);
1383
1384 remoteaddr.sin_family = AF_INET;
1385 remoteaddr.sin_addr.s_addr = (__force u32)node->nd_ipv4_address;
1386 remoteaddr.sin_port = (__force u16)node->nd_ipv4_port;
1387
1388 ret = sc->sc_sock->ops->connect(sc->sc_sock,
1389 (struct sockaddr *)&remoteaddr,
1390 sizeof(remoteaddr),
1391 O_NONBLOCK);
1392 if (ret == -EINPROGRESS)
1393 ret = 0;
1394
1395out:
1396 if (ret) {
1397 mlog(ML_NOTICE, "connect attempt to " SC_NODEF_FMT " failed "
1398 "with errno %d\n", SC_NODEF_ARGS(sc), ret);
1399 /* 0 err so that another will be queued and attempted
1400 * from set_nn_state */
1401 if (sc)
1402 o2net_ensure_shutdown(nn, sc, 0);
1403 }
1404 if (sc)
1405 sc_put(sc);
1406 if (node)
1407 o2nm_node_put(node);
1408
1409 return;
1410}
1411
1412static void o2net_connect_expired(void *arg)
1413{
1414 struct o2net_node *nn = arg;
1415
1416 spin_lock(&nn->nn_lock);
1417 if (!nn->nn_sc_valid) {
1418 mlog(ML_ERROR, "no connection established with node %u after "
1419 "%u seconds, giving up and returning errors.\n",
1420 o2net_num_from_nn(nn), O2NET_IDLE_TIMEOUT_SECS);
1421
1422 o2net_set_nn_state(nn, NULL, 0, -ENOTCONN);
1423 }
1424 spin_unlock(&nn->nn_lock);
1425}
1426
1427static void o2net_still_up(void *arg)
1428{
1429 struct o2net_node *nn = arg;
1430
1431 o2quo_hb_still_up(o2net_num_from_nn(nn));
1432}
1433
1434/* ------------------------------------------------------------ */
1435
1436void o2net_disconnect_node(struct o2nm_node *node)
1437{
1438 struct o2net_node *nn = o2net_nn_from_num(node->nd_num);
1439
1440 /* don't reconnect until it's heartbeating again */
1441 spin_lock(&nn->nn_lock);
1442 o2net_set_nn_state(nn, NULL, 0, -ENOTCONN);
1443 spin_unlock(&nn->nn_lock);
1444
1445 if (o2net_wq) {
1446 cancel_delayed_work(&nn->nn_connect_expired);
1447 cancel_delayed_work(&nn->nn_connect_work);
1448 cancel_delayed_work(&nn->nn_still_up);
1449 flush_workqueue(o2net_wq);
1450 }
1451}
1452
1453static void o2net_hb_node_down_cb(struct o2nm_node *node, int node_num,
1454 void *data)
1455{
1456 o2quo_hb_down(node_num);
1457
1458 if (node_num != o2nm_this_node())
1459 o2net_disconnect_node(node);
1460}
1461
1462static void o2net_hb_node_up_cb(struct o2nm_node *node, int node_num,
1463 void *data)
1464{
1465 struct o2net_node *nn = o2net_nn_from_num(node_num);
1466
1467 o2quo_hb_up(node_num);
1468
1469 /* ensure an immediate connect attempt */
1470 nn->nn_last_connect_attempt = jiffies -
1471 (msecs_to_jiffies(O2NET_RECONNECT_DELAY_MS) + 1);
1472
1473 if (node_num != o2nm_this_node()) {
1474 /* heartbeat doesn't work unless a local node number is
1475 * configured and doing so brings up the o2net_wq, so we can
1476 * use it.. */
1477 queue_delayed_work(o2net_wq, &nn->nn_connect_expired,
1478 O2NET_IDLE_TIMEOUT_SECS * HZ);
1479
1480 /* believe it or not, accept and node hearbeating testing
1481 * can succeed for this node before we got here.. so
1482 * only use set_nn_state to clear the persistent error
1483 * if that hasn't already happened */
1484 spin_lock(&nn->nn_lock);
1485 if (nn->nn_persistent_error)
1486 o2net_set_nn_state(nn, NULL, 0, 0);
1487 spin_unlock(&nn->nn_lock);
1488 }
1489}
1490
1491void o2net_unregister_hb_callbacks(void)
1492{
1493 int ret;
1494
1495 ret = o2hb_unregister_callback(&o2net_hb_up);
1496 if (ret < 0)
1497 mlog(ML_ERROR, "Status return %d unregistering heartbeat up "
1498 "callback!\n", ret);
1499
1500 ret = o2hb_unregister_callback(&o2net_hb_down);
1501 if (ret < 0)
1502 mlog(ML_ERROR, "Status return %d unregistering heartbeat down "
1503 "callback!\n", ret);
1504}
1505
1506int o2net_register_hb_callbacks(void)
1507{
1508 int ret;
1509
1510 o2hb_setup_callback(&o2net_hb_down, O2HB_NODE_DOWN_CB,
1511 o2net_hb_node_down_cb, NULL, O2NET_HB_PRI);
1512 o2hb_setup_callback(&o2net_hb_up, O2HB_NODE_UP_CB,
1513 o2net_hb_node_up_cb, NULL, O2NET_HB_PRI);
1514
1515 ret = o2hb_register_callback(&o2net_hb_up);
1516 if (ret == 0)
1517 ret = o2hb_register_callback(&o2net_hb_down);
1518
1519 if (ret)
1520 o2net_unregister_hb_callbacks();
1521
1522 return ret;
1523}
1524
1525/* ------------------------------------------------------------ */
1526
1527static int o2net_accept_one(struct socket *sock)
1528{
1529 int ret, slen;
1530 struct sockaddr_in sin;
1531 struct socket *new_sock = NULL;
1532 struct o2nm_node *node = NULL;
1533 struct o2net_sock_container *sc = NULL;
1534 struct o2net_node *nn;
1535
1536 BUG_ON(sock == NULL);
1537 ret = sock_create_lite(sock->sk->sk_family, sock->sk->sk_type,
1538 sock->sk->sk_protocol, &new_sock);
1539 if (ret)
1540 goto out;
1541
1542 new_sock->type = sock->type;
1543 new_sock->ops = sock->ops;
1544 ret = sock->ops->accept(sock, new_sock, O_NONBLOCK);
1545 if (ret < 0)
1546 goto out;
1547
1548 new_sock->sk->sk_allocation = GFP_ATOMIC;
1549
1550 ret = o2net_set_nodelay(new_sock);
1551 if (ret) {
1552 mlog(ML_ERROR, "setting TCP_NODELAY failed with %d\n", ret);
1553 goto out;
1554 }
1555
1556 slen = sizeof(sin);
1557 ret = new_sock->ops->getname(new_sock, (struct sockaddr *) &sin,
1558 &slen, 1);
1559 if (ret < 0)
1560 goto out;
1561
1562 node = o2nm_get_node_by_ip((__force __be32)sin.sin_addr.s_addr);
1563 if (node == NULL) {
1564 mlog(ML_NOTICE, "attempt to connect from unknown node at "
1565 "%u.%u.%u.%u:%d\n", NIPQUAD(sin.sin_addr.s_addr),
1566 ntohs((__force __be16)sin.sin_port));
1567 ret = -EINVAL;
1568 goto out;
1569 }
1570
1571 if (o2nm_this_node() > node->nd_num) {
1572 mlog(ML_NOTICE, "unexpected connect attempted from a lower "
1573 "numbered node '%s' at " "%u.%u.%u.%u:%d with num %u\n",
1574 node->nd_name, NIPQUAD(sin.sin_addr.s_addr),
1575 ntohs((__force __be16)sin.sin_port), node->nd_num);
1576 ret = -EINVAL;
1577 goto out;
1578 }
1579
1580 /* this happens all the time when the other node sees our heartbeat
1581 * and tries to connect before we see their heartbeat */
1582 if (!o2hb_check_node_heartbeating_from_callback(node->nd_num)) {
1583 mlog(ML_CONN, "attempt to connect from node '%s' at "
1584 "%u.%u.%u.%u:%d but it isn't heartbeating\n",
1585 node->nd_name, NIPQUAD(sin.sin_addr.s_addr),
1586 ntohs((__force __be16)sin.sin_port));
1587 ret = -EINVAL;
1588 goto out;
1589 }
1590
1591 nn = o2net_nn_from_num(node->nd_num);
1592
1593 spin_lock(&nn->nn_lock);
1594 if (nn->nn_sc)
1595 ret = -EBUSY;
1596 else
1597 ret = 0;
1598 spin_unlock(&nn->nn_lock);
1599 if (ret) {
1600 mlog(ML_NOTICE, "attempt to connect from node '%s' at "
1601 "%u.%u.%u.%u:%d but it already has an open connection\n",
1602 node->nd_name, NIPQUAD(sin.sin_addr.s_addr),
1603 ntohs((__force __be16)sin.sin_port));
1604 goto out;
1605 }
1606
1607 sc = sc_alloc(node);
1608 if (sc == NULL) {
1609 ret = -ENOMEM;
1610 goto out;
1611 }
1612
1613 sc->sc_sock = new_sock;
1614 new_sock = NULL;
1615
1616 spin_lock(&nn->nn_lock);
1617 o2net_set_nn_state(nn, sc, 0, 0);
1618 spin_unlock(&nn->nn_lock);
1619
1620 o2net_register_callbacks(sc->sc_sock->sk, sc);
1621 o2net_sc_queue_work(sc, &sc->sc_rx_work);
1622
1623 o2net_sendpage(sc, o2net_hand, sizeof(*o2net_hand));
1624
1625out:
1626 if (new_sock)
1627 sock_release(new_sock);
1628 if (node)
1629 o2nm_node_put(node);
1630 if (sc)
1631 sc_put(sc);
1632 return ret;
1633}
1634
1635static void o2net_accept_many(void *arg)
1636{
1637 struct socket *sock = arg;
1638 while (o2net_accept_one(sock) == 0)
1639 cond_resched();
1640}
1641
1642static void o2net_listen_data_ready(struct sock *sk, int bytes)
1643{
1644 void (*ready)(struct sock *sk, int bytes);
1645
1646 read_lock(&sk->sk_callback_lock);
1647 ready = sk->sk_user_data;
1648 if (ready == NULL) { /* check for teardown race */
1649 ready = sk->sk_data_ready;
1650 goto out;
1651 }
1652
1653 /* ->sk_data_ready is also called for a newly established child socket
1654 * before it has been accepted and the acceptor has set up their
1655 * data_ready.. we only want to queue listen work for our listening
1656 * socket */
1657 if (sk->sk_state == TCP_LISTEN) {
1658 mlog(ML_TCP, "bytes: %d\n", bytes);
1659 queue_work(o2net_wq, &o2net_listen_work);
1660 }
1661
1662out:
1663 read_unlock(&sk->sk_callback_lock);
1664 ready(sk, bytes);
1665}
1666
1667static int o2net_open_listening_sock(__be16 port)
1668{
1669 struct socket *sock = NULL;
1670 int ret;
1671 struct sockaddr_in sin = {
1672 .sin_family = PF_INET,
1673 .sin_addr = { .s_addr = (__force u32)htonl(INADDR_ANY) },
1674 .sin_port = (__force u16)port,
1675 };
1676
1677 ret = sock_create(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
1678 if (ret < 0) {
1679 mlog(ML_ERROR, "unable to create socket, ret=%d\n", ret);
1680 goto out;
1681 }
1682
1683 sock->sk->sk_allocation = GFP_ATOMIC;
1684
1685 write_lock_bh(&sock->sk->sk_callback_lock);
1686 sock->sk->sk_user_data = sock->sk->sk_data_ready;
1687 sock->sk->sk_data_ready = o2net_listen_data_ready;
1688 write_unlock_bh(&sock->sk->sk_callback_lock);
1689
1690 o2net_listen_sock = sock;
1691 INIT_WORK(&o2net_listen_work, o2net_accept_many, sock);
1692
1693 sock->sk->sk_reuse = 1;
1694 ret = sock->ops->bind(sock, (struct sockaddr *)&sin, sizeof(sin));
1695 if (ret < 0) {
1696 mlog(ML_ERROR, "unable to bind socket to port %d, ret=%d\n",
1697 ntohs(port), ret);
1698 goto out;
1699 }
1700
1701 ret = sock->ops->listen(sock, 64);
1702 if (ret < 0) {
1703 mlog(ML_ERROR, "unable to listen on port %d, ret=%d\n",
1704 ntohs(port), ret);
1705 }
1706
1707out:
1708 if (ret) {
1709 o2net_listen_sock = NULL;
1710 if (sock)
1711 sock_release(sock);
1712 }
1713 return ret;
1714}
1715
1716/*
1717 * called from node manager when we should bring up our network listening
1718 * socket. node manager handles all the serialization to only call this
1719 * once and to match it with o2net_stop_listening(). note,
1720 * o2nm_this_node() doesn't work yet as we're being called while it
1721 * is being set up.
1722 */
1723int o2net_start_listening(struct o2nm_node *node)
1724{
1725 int ret = 0;
1726
1727 BUG_ON(o2net_wq != NULL);
1728 BUG_ON(o2net_listen_sock != NULL);
1729
1730 mlog(ML_KTHREAD, "starting o2net thread...\n");
1731 o2net_wq = create_singlethread_workqueue("o2net");
1732 if (o2net_wq == NULL) {
1733 mlog(ML_ERROR, "unable to launch o2net thread\n");
1734 return -ENOMEM; /* ? */
1735 }
1736
1737 ret = o2net_open_listening_sock(node->nd_ipv4_port);
1738 if (ret) {
1739 destroy_workqueue(o2net_wq);
1740 o2net_wq = NULL;
1741 } else
1742 o2quo_conn_up(node->nd_num);
1743
1744 return ret;
1745}
1746
1747/* again, o2nm_this_node() doesn't work here as we're involved in
1748 * tearing it down */
1749void o2net_stop_listening(struct o2nm_node *node)
1750{
1751 struct socket *sock = o2net_listen_sock;
1752 size_t i;
1753
1754 BUG_ON(o2net_wq == NULL);
1755 BUG_ON(o2net_listen_sock == NULL);
1756
1757 /* stop the listening socket from generating work */
1758 write_lock_bh(&sock->sk->sk_callback_lock);
1759 sock->sk->sk_data_ready = sock->sk->sk_user_data;
1760 sock->sk->sk_user_data = NULL;
1761 write_unlock_bh(&sock->sk->sk_callback_lock);
1762
1763 for (i = 0; i < ARRAY_SIZE(o2net_nodes); i++) {
1764 struct o2nm_node *node = o2nm_get_node_by_num(i);
1765 if (node) {
1766 o2net_disconnect_node(node);
1767 o2nm_node_put(node);
1768 }
1769 }
1770
1771 /* finish all work and tear down the work queue */
1772 mlog(ML_KTHREAD, "waiting for o2net thread to exit....\n");
1773 destroy_workqueue(o2net_wq);
1774 o2net_wq = NULL;
1775
1776 sock_release(o2net_listen_sock);
1777 o2net_listen_sock = NULL;
1778
1779 o2quo_conn_err(node->nd_num);
1780}
1781
1782/* ------------------------------------------------------------ */
1783
1784int o2net_init(void)
1785{
1786 unsigned long i;
1787
1788 o2quo_init();
1789
1790 o2net_hand = kcalloc(1, sizeof(struct o2net_handshake), GFP_KERNEL);
1791 o2net_keep_req = kcalloc(1, sizeof(struct o2net_msg), GFP_KERNEL);
1792 o2net_keep_resp = kcalloc(1, sizeof(struct o2net_msg), GFP_KERNEL);
1793 if (!o2net_hand || !o2net_keep_req || !o2net_keep_resp) {
1794 kfree(o2net_hand);
1795 kfree(o2net_keep_req);
1796 kfree(o2net_keep_resp);
1797 return -ENOMEM;
1798 }
1799
1800 o2net_hand->protocol_version = cpu_to_be64(O2NET_PROTOCOL_VERSION);
1801 o2net_hand->connector_id = cpu_to_be64(1);
1802
1803 o2net_keep_req->magic = cpu_to_be16(O2NET_MSG_KEEP_REQ_MAGIC);
1804 o2net_keep_resp->magic = cpu_to_be16(O2NET_MSG_KEEP_RESP_MAGIC);
1805
1806 for (i = 0; i < ARRAY_SIZE(o2net_nodes); i++) {
1807 struct o2net_node *nn = o2net_nn_from_num(i);
1808
1809 spin_lock_init(&nn->nn_lock);
1810 INIT_WORK(&nn->nn_connect_work, o2net_start_connect, nn);
1811 INIT_WORK(&nn->nn_connect_expired, o2net_connect_expired, nn);
1812 INIT_WORK(&nn->nn_still_up, o2net_still_up, nn);
1813 /* until we see hb from a node we'll return einval */
1814 nn->nn_persistent_error = -ENOTCONN;
1815 init_waitqueue_head(&nn->nn_sc_wq);
1816 idr_init(&nn->nn_status_idr);
1817 INIT_LIST_HEAD(&nn->nn_status_list);
1818 }
1819
1820 return 0;
1821}
1822
1823void o2net_exit(void)
1824{
1825 o2quo_exit();
1826 kfree(o2net_hand);
1827 kfree(o2net_keep_req);
1828 kfree(o2net_keep_resp);
1829}
diff --git a/fs/ocfs2/cluster/tcp.h b/fs/ocfs2/cluster/tcp.h
new file mode 100644
index 000000000000..a6f4585501c8
--- /dev/null
+++ b/fs/ocfs2/cluster/tcp.h
@@ -0,0 +1,113 @@
1/* -*- mode: c; c-basic-offset: 8; -*-
2 * vim: noexpandtab sw=8 ts=8 sts=0:
3 *
4 * tcp.h
5 *
6 * Function prototypes
7 *
8 * Copyright (C) 2004 Oracle. All rights reserved.
9 *
10 * This program is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public
12 * License as published by the Free Software Foundation; either
13 * version 2 of the License, or (at your option) any later version.
14 *
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * General Public License for more details.
19 *
20 * You should have received a copy of the GNU General Public
21 * License along with this program; if not, write to the
22 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
23 * Boston, MA 021110-1307, USA.
24 *
25 */
26
27#ifndef O2CLUSTER_TCP_H
28#define O2CLUSTER_TCP_H
29
30#include <linux/socket.h>
31#ifdef __KERNEL__
32#include <net/sock.h>
33#include <linux/tcp.h>
34#else
35#include <sys/socket.h>
36#endif
37#include <linux/inet.h>
38#include <linux/in.h>
39
40struct o2net_msg
41{
42 __be16 magic;
43 __be16 data_len;
44 __be16 msg_type;
45 __be16 pad1;
46 __be32 sys_status;
47 __be32 status;
48 __be32 key;
49 __be32 msg_num;
50 __u8 buf[0];
51};
52
53typedef int (o2net_msg_handler_func)(struct o2net_msg *msg, u32 len, void *data);
54
55#define O2NET_MAX_PAYLOAD_BYTES (4096 - sizeof(struct o2net_msg))
56
57/* TODO: figure this out.... */
58static inline int o2net_link_down(int err, struct socket *sock)
59{
60 if (sock) {
61 if (sock->sk->sk_state != TCP_ESTABLISHED &&
62 sock->sk->sk_state != TCP_CLOSE_WAIT)
63 return 1;
64 }
65
66 if (err >= 0)
67 return 0;
68 switch (err) {
69 /* ????????????????????????? */
70 case -ERESTARTSYS:
71 case -EBADF:
72 /* When the server has died, an ICMP port unreachable
73 * message prompts ECONNREFUSED. */
74 case -ECONNREFUSED:
75 case -ENOTCONN:
76 case -ECONNRESET:
77 case -EPIPE:
78 return 1;
79 }
80 return 0;
81}
82
83enum {
84 O2NET_DRIVER_UNINITED,
85 O2NET_DRIVER_READY,
86};
87
88int o2net_init_tcp_sock(struct inode *inode);
89int o2net_send_message(u32 msg_type, u32 key, void *data, u32 len,
90 u8 target_node, int *status);
91int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *vec,
92 size_t veclen, u8 target_node, int *status);
93int o2net_broadcast_message(u32 msg_type, u32 key, void *data, u32 len,
94 struct inode *group);
95
96int o2net_register_handler(u32 msg_type, u32 key, u32 max_len,
97 o2net_msg_handler_func *func, void *data,
98 struct list_head *unreg_list);
99void o2net_unregister_handler_list(struct list_head *list);
100
101struct o2nm_node;
102int o2net_register_hb_callbacks(void);
103void o2net_unregister_hb_callbacks(void);
104int o2net_start_listening(struct o2nm_node *node);
105void o2net_stop_listening(struct o2nm_node *node);
106void o2net_disconnect_node(struct o2nm_node *node);
107
108int o2net_init(void);
109void o2net_exit(void);
110int o2net_proc_init(struct proc_dir_entry *parent);
111void o2net_proc_exit(struct proc_dir_entry *parent);
112
113#endif /* O2CLUSTER_TCP_H */
diff --git a/fs/ocfs2/cluster/tcp_internal.h b/fs/ocfs2/cluster/tcp_internal.h
new file mode 100644
index 000000000000..ff9e2e2104c2
--- /dev/null
+++ b/fs/ocfs2/cluster/tcp_internal.h
@@ -0,0 +1,174 @@
1/* -*- mode: c; c-basic-offset: 8; -*-
2 * vim: noexpandtab sw=8 ts=8 sts=0:
3 *
4 * Copyright (C) 2005 Oracle. All rights reserved.
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
10 *
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * General Public License for more details.
15 *
16 * You should have received a copy of the GNU General Public
17 * License along with this program; if not, write to the
18 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
19 * Boston, MA 021110-1307, USA.
20 */
21
22#ifndef O2CLUSTER_TCP_INTERNAL_H
23#define O2CLUSTER_TCP_INTERNAL_H
24
25#define O2NET_MSG_MAGIC ((u16)0xfa55)
26#define O2NET_MSG_STATUS_MAGIC ((u16)0xfa56)
27#define O2NET_MSG_KEEP_REQ_MAGIC ((u16)0xfa57)
28#define O2NET_MSG_KEEP_RESP_MAGIC ((u16)0xfa58)
29
30/* same as hb delay, we're waiting for another node to recognize our hb */
31#define O2NET_RECONNECT_DELAY_MS O2HB_REGION_TIMEOUT_MS
32
33/* we're delaying our quorum decision so that heartbeat will have timed
34 * out truly dead nodes by the time we come around to making decisions
35 * on their number */
36#define O2NET_QUORUM_DELAY_MS ((o2hb_dead_threshold + 2) * O2HB_REGION_TIMEOUT_MS)
37
38#define O2NET_KEEPALIVE_DELAY_SECS 5
39#define O2NET_IDLE_TIMEOUT_SECS 10
40
41/*
42 * This version number represents quite a lot, unfortunately. It not
43 * only represents the raw network message protocol on the wire but also
44 * locking semantics of the file system using the protocol. It should
45 * be somewhere else, I'm sure, but right now it isn't.
46 *
47 * New in version 2:
48 * - full 64 bit i_size in the metadata lock lvbs
49 * - introduction of "rw" lock and pushing meta/data locking down
50 */
51#define O2NET_PROTOCOL_VERSION 2ULL
52struct o2net_handshake {
53 __be64 protocol_version;
54 __be64 connector_id;
55};
56
57struct o2net_node {
58 /* this is never called from int/bh */
59 spinlock_t nn_lock;
60
61 /* set the moment an sc is allocated and a connect is started */
62 struct o2net_sock_container *nn_sc;
63 /* _valid is only set after the handshake passes and tx can happen */
64 unsigned nn_sc_valid:1;
65 /* if this is set tx just returns it */
66 int nn_persistent_error;
67
68 /* threads waiting for an sc to arrive wait on the wq for generation
69 * to increase. it is increased when a connecting socket succeeds
70 * or fails or when an accepted socket is attached. */
71 wait_queue_head_t nn_sc_wq;
72
73 struct idr nn_status_idr;
74 struct list_head nn_status_list;
75
76 /* connects are attempted from when heartbeat comes up until either hb
77 * goes down, the node is unconfigured, no connect attempts succeed
78 * before O2NET_CONN_IDLE_DELAY, or a connect succeeds. connect_work
79 * is queued from set_nn_state both from hb up and from itself if a
80 * connect attempt fails and so can be self-arming. shutdown is
81 * careful to first mark the nn such that no connects will be attempted
82 * before canceling delayed connect work and flushing the queue. */
83 struct work_struct nn_connect_work;
84 unsigned long nn_last_connect_attempt;
85
86 /* this is queued as nodes come up and is canceled when a connection is
87 * established. this expiring gives up on the node and errors out
88 * transmits */
89 struct work_struct nn_connect_expired;
90
91 /* after we give up on a socket we wait a while before deciding
92 * that it is still heartbeating and that we should do some
93 * quorum work */
94 struct work_struct nn_still_up;
95};
96
97struct o2net_sock_container {
98 struct kref sc_kref;
99 /* the next two are vaild for the life time of the sc */
100 struct socket *sc_sock;
101 struct o2nm_node *sc_node;
102
103 /* all of these sc work structs hold refs on the sc while they are
104 * queued. they should not be able to ref a freed sc. the teardown
105 * race is with o2net_wq destruction in o2net_stop_listening() */
106
107 /* rx and connect work are generated from socket callbacks. sc
108 * shutdown removes the callbacks and then flushes the work queue */
109 struct work_struct sc_rx_work;
110 struct work_struct sc_connect_work;
111 /* shutdown work is triggered in two ways. the simple way is
112 * for a code path calls ensure_shutdown which gets a lock, removes
113 * the sc from the nn, and queues the work. in this case the
114 * work is single-shot. the work is also queued from a sock
115 * callback, though, and in this case the work will find the sc
116 * still on the nn and will call ensure_shutdown itself.. this
117 * ends up triggering the shutdown work again, though nothing
118 * will be done in that second iteration. so work queue teardown
119 * has to be careful to remove the sc from the nn before waiting
120 * on the work queue so that the shutdown work doesn't remove the
121 * sc and rearm itself.
122 */
123 struct work_struct sc_shutdown_work;
124
125 struct timer_list sc_idle_timeout;
126 struct work_struct sc_keepalive_work;
127
128 unsigned sc_handshake_ok:1;
129
130 struct page *sc_page;
131 size_t sc_page_off;
132
133 /* original handlers for the sockets */
134 void (*sc_state_change)(struct sock *sk);
135 void (*sc_data_ready)(struct sock *sk, int bytes);
136
137 struct timeval sc_tv_timer;
138 struct timeval sc_tv_data_ready;
139 struct timeval sc_tv_advance_start;
140 struct timeval sc_tv_advance_stop;
141 struct timeval sc_tv_func_start;
142 struct timeval sc_tv_func_stop;
143 u32 sc_msg_key;
144 u16 sc_msg_type;
145};
146
147struct o2net_msg_handler {
148 struct rb_node nh_node;
149 u32 nh_max_len;
150 u32 nh_msg_type;
151 u32 nh_key;
152 o2net_msg_handler_func *nh_func;
153 o2net_msg_handler_func *nh_func_data;
154 struct kref nh_kref;
155 struct list_head nh_unregister_item;
156};
157
158enum o2net_system_error {
159 O2NET_ERR_NONE = 0,
160 O2NET_ERR_NO_HNDLR,
161 O2NET_ERR_OVERFLOW,
162 O2NET_ERR_DIED,
163 O2NET_ERR_MAX
164};
165
166struct o2net_status_wait {
167 enum o2net_system_error ns_sys_status;
168 s32 ns_status;
169 int ns_id;
170 wait_queue_head_t ns_wq;
171 struct list_head ns_node_item;
172};
173
174#endif /* O2CLUSTER_TCP_INTERNAL_H */
generated by cgit v1.2.2 (git 2.25.0) at 2024-12-22 08:45:13 -0500