1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
|
/*
* Copyright (c) 2007 Oracle. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
* General Public License (GPL) Version 2, available from the file
* COPYING in the main directory of this source tree, or the
* OpenIB.org BSD license below:
*
* Redistribution and use in source and binary forms, with or
* without modification, are permitted provided that the following
* conditions are met:
*
* - Redistributions of source code must retain the above
* copyright notice, this list of conditions and the following
* disclaimer.
*
* - Redistributions in binary form must reproduce the above
* copyright notice, this list of conditions and the following
* disclaimer in the documentation and/or other materials
* provided with the distribution.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
*/
#include <linux/slab.h>
#include <linux/types.h>
#include <linux/rbtree.h>
#include <asm-generic/bitops/le.h>
#include "rds.h"
/*
* This file implements the receive side of the unconventional congestion
* management in RDS.
*
* Messages waiting in the receive queue on the receiving socket are accounted
* against the sockets SO_RCVBUF option value. Only the payload bytes in the
* message are accounted for. If the number of bytes queued equals or exceeds
* rcvbuf then the socket is congested. All sends attempted to this socket's
* address should return block or return -EWOULDBLOCK.
*
* Applications are expected to be reasonably tuned such that this situation
* very rarely occurs. An application encountering this "back-pressure" is
* considered a bug.
*
* This is implemented by having each node maintain bitmaps which indicate
* which ports on bound addresses are congested. As the bitmap changes it is
* sent through all the connections which terminate in the local address of the
* bitmap which changed.
*
* The bitmaps are allocated as connections are brought up. This avoids
* allocation in the interrupt handling path which queues messages on sockets.
* The dense bitmaps let transports send the entire bitmap on any bitmap change
* reasonably efficiently. This is much easier to implement than some
* finer-grained communication of per-port congestion. The sender does a very
* inexpensive bit test to test if the port it's about to send to is congested
* or not.
*/
/*
* Interaction with poll is a tad tricky. We want all processes stuck in
* poll to wake up and check whether a congested destination became uncongested.
* The really sad thing is we have no idea which destinations the application
* wants to send to - we don't even know which rds_connections are involved.
* So until we implement a more flexible rds poll interface, we have to make
* do with this:
* We maintain a global counter that is incremented each time a congestion map
* update is received. Each rds socket tracks this value, and if rds_poll
* finds that the saved generation number is smaller than the global generation
* number, it wakes up the process.
*/
static atomic_t rds_cong_generation = ATOMIC_INIT(0);
/*
* Congestion monitoring
*/
static LIST_HEAD(rds_cong_monitor);
static DEFINE_RWLOCK(rds_cong_monitor_lock);
/*
* Yes, a global lock. It's used so infrequently that it's worth keeping it
* global to simplify the locking. It's only used in the following
* circumstances:
*
* - on connection buildup to associate a conn with its maps
* - on map changes to inform conns of a new map to send
*
* It's sadly ordered under the socket callback lock and the connection lock.
* Receive paths can mark ports congested from interrupt context so the
* lock masks interrupts.
*/
static DEFINE_SPINLOCK(rds_cong_lock);
static struct rb_root rds_cong_tree = RB_ROOT;
static struct rds_cong_map *rds_cong_tree_walk(__be32 addr,
struct rds_cong_map *insert)
{
struct rb_node **p = &rds_cong_tree.rb_node;
struct rb_node *parent = NULL;
struct rds_cong_map *map;
while (*p) {
parent = *p;
map = rb_entry(parent, struct rds_cong_map, m_rb_node);
if (addr < map->m_addr)
p = &(*p)->rb_left;
else if (addr > map->m_addr)
p = &(*p)->rb_right;
else
return map;
}
if (insert) {
rb_link_node(&insert->m_rb_node, parent, p);
rb_insert_color(&insert->m_rb_node, &rds_cong_tree);
}
return NULL;
}
/*
* There is only ever one bitmap for any address. Connections try and allocate
* these bitmaps in the process getting pointers to them. The bitmaps are only
* ever freed as the module is removed after all connections have been freed.
*/
static struct rds_cong_map *rds_cong_from_addr(__be32 addr)
{
struct rds_cong_map *map;
struct rds_cong_map *ret = NULL;
unsigned long zp;
unsigned long i;
unsigned long flags;
map = kzalloc(sizeof(struct rds_cong_map), GFP_KERNEL);
if (!map)
return NULL;
map->m_addr = addr;
init_waitqueue_head(&map->m_waitq);
INIT_LIST_HEAD(&map->m_conn_list);
for (i = 0; i < RDS_CONG_MAP_PAGES; i++) {
zp = get_zeroed_page(GFP_KERNEL);
if (zp == 0)
goto out;
map->m_page_addrs[i] = zp;
}
spin_lock_irqsave(&rds_cong_lock, flags);
ret = rds_cong_tree_walk(addr, map);
spin_unlock_irqrestore(&rds_cong_lock, flags);
if (!ret) {
ret = map;
map = NULL;
}
out:
if (map) {
for (i = 0; i < RDS_CONG_MAP_PAGES && map->m_page_addrs[i]; i++)
free_page(map->m_page_addrs[i]);
kfree(map);
}
rdsdebug("map %p for addr %x\n", ret, be32_to_cpu(addr));
return ret;
}
/*
* Put the conn on its local map's list. This is called when the conn is
* really added to the hash. It's nested under the rds_conn_lock, sadly.
*/
void rds_cong_add_conn(struct rds_connection *conn)
{
unsigned long flags;
rdsdebug("conn %p now on map %p\n", conn, conn->c_lcong);
spin_lock_irqsave(&rds_cong_lock, flags);
list_add_tail(&conn->c_map_item, &conn->c_lcong->m_conn_list);
spin_unlock_irqrestore(&rds_cong_lock, flags);
}
void rds_cong_remove_conn(struct rds_connection *conn)
{
unsigned long flags;
rdsdebug("removing conn %p from map %p\n", conn, conn->c_lcong);
spin_lock_irqsave(&rds_cong_lock, flags);
list_del_init(&conn->c_map_item);
spin_unlock_irqrestore(&rds_cong_lock, flags);
}
int rds_cong_get_maps(struct rds_connection *conn)
{
conn->c_lcong = rds_cong_from_addr(conn->c_laddr);
conn->c_fcong = rds_cong_from_addr(conn->c_faddr);
if (!(conn->c_lcong && conn->c_fcong))
return -ENOMEM;
return 0;
}
void rds_cong_queue_updates(struct rds_cong_map *map)
{
struct rds_connection *conn;
unsigned long flags;
spin_lock_irqsave(&rds_cong_lock, flags);
list_for_each_entry(conn, &map->m_conn_list, c_map_item) {
if (!test_and_set_bit(0, &conn->c_map_queued)) {
rds_stats_inc(s_cong_update_queued);
rds_send_xmit(conn);
}
}
spin_unlock_irqrestore(&rds_cong_lock, flags);
}
void rds_cong_map_updated(struct rds_cong_map *map, uint64_t portmask)
{
rdsdebug("waking map %p for %pI4\n",
map, &map->m_addr);
rds_stats_inc(s_cong_update_received);
atomic_inc(&rds_cong_generation);
if (waitqueue_active(&map->m_waitq))
wake_up(&map->m_waitq);
if (waitqueue_active(&rds_poll_waitq))
wake_up_all(&rds_poll_waitq);
if (portmask && !list_empty(&rds_cong_monitor)) {
unsigned long flags;
struct rds_sock *rs;
read_lock_irqsave(&rds_cong_monitor_lock, flags);
list_for_each_entry(rs, &rds_cong_monitor, rs_cong_list) {
spin_lock(&rs->rs_lock);
rs->rs_cong_notify |= (rs->rs_cong_mask & portmask);
rs->rs_cong_mask &= ~portmask;
spin_unlock(&rs->rs_lock);
if (rs->rs_cong_notify)
rds_wake_sk_sleep(rs);
}
read_unlock_irqrestore(&rds_cong_monitor_lock, flags);
}
}
EXPORT_SYMBOL_GPL(rds_cong_map_updated);
int rds_cong_updated_since(unsigned long *recent)
{
unsigned long gen = atomic_read(&rds_cong_generation);
if (likely(*recent == gen))
return 0;
*recent = gen;
return 1;
}
/*
* We're called under the locking that protects the sockets receive buffer
* consumption. This makes it a lot easier for the caller to only call us
* when it knows that an existing set bit needs to be cleared, and vice versa.
* We can't block and we need to deal with concurrent sockets working against
* the same per-address map.
*/
void rds_cong_set_bit(struct rds_cong_map *map, __be16 port)
{
unsigned long i;
unsigned long off;
rdsdebug("setting congestion for %pI4:%u in map %p\n",
&map->m_addr, ntohs(port), map);
i = be16_to_cpu(port) / RDS_CONG_MAP_PAGE_BITS;
off = be16_to_cpu(port) % RDS_CONG_MAP_PAGE_BITS;
generic___set_le_bit(off, (void *)map->m_page_addrs[i]);
}
void rds_cong_clear_bit(struct rds_cong_map *map, __be16 port)
{
unsigned long i;
unsigned long off;
rdsdebug("clearing congestion for %pI4:%u in map %p\n",
&map->m_addr, ntohs(port), map);
i = be16_to_cpu(port) / RDS_CONG_MAP_PAGE_BITS;
off = be16_to_cpu(port) % RDS_CONG_MAP_PAGE_BITS;
generic___clear_le_bit(off, (void *)map->m_page_addrs[i]);
}
static int rds_cong_test_bit(struct rds_cong_map *map, __be16 port)
{
unsigned long i;
unsigned long off;
i =
|