summaryrefslogtreecommitdiff
path: root/net/rds/rds.h
blob: aab5e949fa93682dd0addb8b11b4348406262a42 (plain)
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
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
#ifndef _RDS_RDS_H
#define _RDS_RDS_H

#include <net/sock.h>
#include <linux/scatterlist.h>
#include <linux/highmem.h>
#include <rdma/rdma_cm.h>
#include <linux/mutex.h>
#include <linux/rds.h>

#include "info.h"

/*
 * RDS Network protocol version
 */
#define RDS_PROTOCOL_3_0	0x0300
#define RDS_PROTOCOL_3_1	0x0301
#define RDS_PROTOCOL_VERSION	RDS_PROTOCOL_3_1
#define RDS_PROTOCOL_MAJOR(v)	((v) >> 8)
#define RDS_PROTOCOL_MINOR(v)	((v) & 255)
#define RDS_PROTOCOL(maj, min)	(((maj) << 8) | min)

/*
 * XXX randomly chosen, but at least seems to be unused:
 * #               18464-18768 Unassigned
 * We should do better.  We want a reserved port to discourage unpriv'ed
 * userspace from listening.
 */
#define RDS_PORT	18634

#ifdef ATOMIC64_INIT
#define KERNEL_HAS_ATOMIC64
#endif

#ifdef DEBUG
#define rdsdebug(fmt, args...) pr_debug("%s(): " fmt, __func__ , ##args)
#else
/* sigh, pr_debug() causes unused variable warnings */
static inline void __attribute__ ((format (printf, 1, 2)))
rdsdebug(char *fmt, ...)
{
}
#endif

/* XXX is there one of these somewhere? */
#define ceil(x, y) \
	({ unsigned long __x = (x), __y = (y); (__x + __y - 1) / __y; })

#define RDS_FRAG_SHIFT	12
#define RDS_FRAG_SIZE	((unsigned int)(1 << RDS_FRAG_SHIFT))

#define RDS_CONG_MAP_BYTES	(65536 / 8)
#define RDS_CONG_MAP_LONGS	(RDS_CONG_MAP_BYTES / sizeof(unsigned long))
#define RDS_CONG_MAP_PAGES	(PAGE_ALIGN(RDS_CONG_MAP_BYTES) / PAGE_SIZE)
#define RDS_CONG_MAP_PAGE_BITS	(PAGE_SIZE * 8)

struct rds_cong_map {
	struct rb_node		m_rb_node;
	__be32			m_addr;
	wait_queue_head_t	m_waitq;
	struct list_head	m_conn_list;
	unsigned long		m_page_addrs[RDS_CONG_MAP_PAGES];
};


/*
 * This is how we will track the connection state:
 * A connection is always in one of the following
 * states. Updates to the state are atomic and imply
 * a memory barrier.
 */
enum {
	RDS_CONN_DOWN = 0,
	RDS_CONN_CONNECTING,
	RDS_CONN_DISCONNECTING,
	RDS_CONN_UP,
	RDS_CONN_ERROR,
};

/* Bits for c_flags */
#define RDS_LL_SEND_FULL	0
#define RDS_RECONNECT_PENDING	1
#define RDS_IN_XMIT		2

struct rds_connection {
	struct hlist_node	c_hash_node;
	__be32			c_laddr;
	__be32			c_faddr;
	unsigned int		c_loopback:1;
	struct rds_connection	*c_passive;

	struct rds_cong_map	*c_lcong;
	struct rds_cong_map	*c_fcong;

	struct rds_message	*c_xmit_rm;
	unsigned long		c_xmit_sg;
	unsigned int		c_xmit_hdr_off;
	unsigned int		c_xmit_data_off;
	unsigned int		c_xmit_atomic_sent;
	unsigned int		c_xmit_rdma_sent;
	unsigned int		c_xmit_data_sent;

	spinlock_t		c_lock;		/* protect msg queues */
	u64			c_next_tx_seq;
	struct list_head	c_send_queue;
	struct list_head	c_retrans;

	u64			c_next_rx_seq;

	struct rds_transport	*c_trans;
	void			*c_transport_data;

	atomic_t		c_state;
	unsigned long		c_flags;
	unsigned long		c_reconnect_jiffies;
	struct delayed_work	c_send_w;
	struct delayed_work	c_recv_w;
	struct delayed_work	c_conn_w;
	struct work_struct	c_down_w;
	struct mutex		c_cm_lock;	/* protect conn state & cm */
	wait_queue_head_t	c_waitq;

	struct list_head	c_map_item;
	unsigned long		c_map_queued;

	unsigned int		c_unacked_packets;
	unsigned int		c_unacked_bytes;

	/* Protocol version */
	unsigned int		c_version;
};

#define RDS_FLAG_CONG_BITMAP	0x01
#define RDS_FLAG_ACK_REQUIRED	0x02
#define RDS_FLAG_RETRANSMITTED	0x04
#define RDS_MAX_ADV_CREDIT	255

/*
 * Maximum space available for extension headers.
 */
#define RDS_HEADER_EXT_SPACE	16

struct rds_header {
	__be64	h_sequence;
	__be64	h_ack;
	__be32	h_len;
	__be16	h_sport;
	__be16	h_dport;
	u8	h_flags;
	u8	h_credit;
	u8	h_padding[4];
	__sum16	h_csum;

	u8	h_exthdr[RDS_HEADER_EXT_SPACE];
};

/*
 * Reserved - indicates end of extensions
 */
#define RDS_EXTHDR_NONE		0

/*
 * This extension header is included in the very
 * first message that is sent on a new connection,
 * and identifies the protocol level. This will help
 * rolling updates if a future change requires breaking
 * the protocol.
 * NB: This is no longer true for IB, where we do a version
 * negotiation during the connection setup phase (protocol
 * version information is included in the RDMA CM private data).
 */
#define RDS_EXTHDR_VERSION	1
struct rds_ext_header_version {
	__be32			h_version;
};

/*
 * This extension header is included in the RDS message
 * chasing an RDMA operation.
 */
#define RDS_EXTHDR_RDMA		2
struct rds_ext_header_rdma {
	__be32			h_rdma_rkey;
};

/*
 * This extension header tells the peer about the
 * destination <R_Key,offset> of the requested RDMA
 * operation.
 */
#define RDS_EXTHDR_RDMA_DEST	3
struct rds_ext_header_rdma_dest {
	__be32			h_rdma_rkey;
	__be32			h_rdma_offset;
};

#define __RDS_EXTHDR_MAX	16 /* for now */

struct rds_incoming {
	atomic_t		i_refcount;
	struct list_head	i_item;
	struct rds_connection	*i_conn;
	struct rds_header	i_hdr;
	unsigned long		i_rx_jiffies;
	__be32			i_saddr;

	rds_rdma_cookie_t	i_rdma_cookie;
};

struct rds_mr {
	struct rb_node		r_rb_node;
	atomic_t		r_refcount;
	u32			r_key;

	/* A copy of the creation flags */
	unsigned int		r_use_once:1;
	unsigned int		r_invalidate:1;
	unsigned int		r_write:1;

	/* This is for RDS_MR_DEAD.
	 * It would be nice & consistent to make this part of the above
	 * bit field here, but we need to use test_and_set_bit.
	 */
	unsigned long		r_state;
	struct rds_sock		*r_sock; /* back pointer to the socket that owns us */
	struct rds_transport	*r_trans;
	void			*r_trans_private;
};

/* Flags for mr->r_state */
#define RDS_MR_DEAD		0

static inline rds_rdma_cookie_t rds_rdma_make_cookie(u32 r_key, u32 offset)
{
	return r_key | (((u64) offset) << 32);
}

static inline u32 rds_rdma_cookie_key(rds_rdma_cookie_t cookie)
{
	return cookie;
}

static inline u32 rds_rdma_cookie_offset(rds_rdma_cookie_t cookie)
{
	return cookie >> 32;
}

/* atomic operation types */
#define RDS_ATOMIC_TYPE_CSWP		0
#define RDS_ATOMIC_TYPE_FADD		1

/*
 * m_sock_item and m_conn_item are on lists that are serialized under
 * conn->c_lock.  m_sock_item has additional meaning in that once it is empty
 * the message will not be put back on the retransmit list after being sent.
 * messages that are canceled while being sent rely on this.
 *
 * m_inc is used by loopback so that it can pass an incoming message straight
 * back up into the rx path.  It embeds a wire header which is also used by
 * the send path, which is kind of awkward.
 *
 * m_sock_item indicates the message's presence on a socket's send or receive
 * queue.  m_rs will point to that socket.
 *
 * m_daddr is used by cancellation to prune messages to a given destination.
 *
 * The RDS_MSG_ON_SOCK and RDS_MSG_ON_CONN flags are used to avoid lock
 * nesting.  As paths iterate over messages on a sock, or conn, they must
 * also lock the conn, or sock, to remove the message from those lists too.
 * Testing the flag to determine if the message is still on the lists lets
 * us avoid testing the list_head directly.  That means each path can use
 * the message's list_head to keep it on a local list while juggling locks
 * without confusing the other path.
 *
 * m_ack_seq is an optional field set by transports who need a different
 * sequence number range to invalidate.  They can use this in a callback
 * that they pass to rds_send_drop_acked() to see if each message has been
 * acked.  The HAS_ACK_SEQ flag can be used to detect messages which haven't
 * had ack_seq set yet.
 */
#define RDS_MSG_ON_SOCK		1
#define RDS_MSG_ON_CONN		2
#define RDS_MSG_HAS_ACK_SEQ	3
#define RDS_MSG_ACK_REQUIRED	4
#define RDS_MSG_RETRANSMITTED	5
#define RDS_MSG_MAPPED		6
#define RDS_MSG_PAGEVEC		7

struct rds_message {
	atomic_t		m_refcount;
	struct list_head	m_sock_item;
	struct list_head	m_conn_item;
	struct rds_incoming	m_inc;
	u64			m_ack_seq;
	__be32			m_daddr;
	unsigned long		m_flags;

	/* Never access m_rs without holding m_rs_lock.
	 * Lock nesting is
	 *  rm->m_rs_lock
	 *   -> rs->rs_lock
	 */
	spinlock_t		m_rs_lock;
	wait_queue_head_t	m_flush_wait;

	struct rds_sock		*m_rs;

	/* cookie to send to remote, in rds header */
	rds_rdma_cookie_t	m_rdma_cookie;

	unsigned int		m_used_sgs;
	unsigned int		m_total_sgs;

	void			*m_final_op;

	struct {
		struct rm_atomic_op {
			int			op_type;
			uint64_t		op_swap_add;
			uint64_t		op_compare;

			u32			op_rkey;
			u64			op_remote_addr;
			unsigned int		op_notify:1;
			unsigned int		op_recverr:1;
			unsigned int		op_mapped:1;
			unsigned int		op_silent:1;
			unsigned int		op_active:1;
			struct scatterlist	*op_sg;
			struct rds_notifier	*op_notifier;

			struct rds_mr		*op_rdma_mr;
		} atomic;
		struct rm_rdma_op {
			u32			op_rkey;
			u64			op_remote_addr;
			unsigned int		op_write:1;
			unsigned int		op_fence:1;
			unsigned int		op_notify:1;
			unsigned int		op_recverr:1;
			unsigned int		op_mapped:1;
			unsigned int		op_silent:1;
			unsigned int		op_active:1;
			unsigned int		op_bytes;
			unsigned int		op_nents;
			unsigned int		op_count;
			struct scatterlist	*op_sg;
			struct rds_notifier	*op_notifier;

			struct rds_mr		*op_rdma_mr;
		} rdma;
		struct rm_data_op {
			unsigned int		op_active:1;
			unsigned int		op_nents;
			unsigned int		op_count;
			struct scatterlist	*op_sg;
		} data;
	};
};

/*
 * The RDS notifier is used (optionally) to tell the application about
 * completed RDMA operations. Rather than keeping the whole rds message
 * around on the queue, we allocate a small notifier that is put on the
 * socket's notifier_list. Notifications are delivered to the application
 * through control messages.
 */
struct rds_notifier {
	struct list_head	n_list;
	uint64_t		n_user_token;
	int			n_status;
};

/**
 * struct rds_transport -  transport specific behavioural hooks
 *
 * @xmit: .xmit is called by rds_send_xmit() to tell the transport to send
 *        part of a message.  The caller serializes on the send_sem so this
 *        doesn't need to be reentrant for a given conn.  The header must be
 *        sent before the data payload.  .xmit must be prepared to send a
 *        message with no data payload.  .xmit should return the number of
 *        bytes that were sent down the connection, including header bytes.
 *        Returning 0 tells the caller that it doesn't need to perform any
 *        additional work now.  This is usually the case when the transport has
 *        filled the sending queue for its connection and will handle
 *        triggering the rds thread to continue the send when space becomes
 *        available.  Returning -EAGAIN tells the caller to retry the send
 *        immediately.  Returning -ENOMEM tells the caller to retry the send at
 *        some point in the future.
 *
 * @conn_shutdown: conn_shutdown stops traffic on the given connection.  Once
 *                 it returns the connection can not call rds_recv_incoming().
 *                 This will only be called once after conn_connect returns
 *                 non-zero success and will The caller serializes this with
 *                 the send and connecting paths (xmit_* and conn_*).  The
 *                 transport is responsible for other serialization, including
 *                 rds_recv_incoming().  This is called in process context but
 *                 should try hard not to block.
 */

#define RDS_TRANS_IB	0
#define RDS_TRANS_IWARP	1
#define RDS_TRANS_TCP	2
#define RDS_TRANS_COUNT	3

struct rds_transport {
	char			t_name[TRANSNAMSIZ];
	struct list_head	t_item;
	struct module		*t_owner;
	unsigned int		t_prefer_loopback:1;
	unsigned int		t_type;

	int (*laddr_check)(__be32 addr);
	int (*conn_alloc)(struct rds_connection *conn, gfp_t gfp);
	void (*conn_free)(void *data);
	int (*conn_connect)(struct rds_connection *conn);
	void (*conn_shutdown)(struct rds_connection *conn);
	void (*xmit_prepare)(struct rds_connection *conn);
	void (*xmit_complete)(struct rds_connection *conn);
	int (*xmit)(struct rds_connection *conn, struct rds_message *rm,
		    unsigned int hdr_off, unsigned int sg, unsigned int off);
	int (*xmit_rdma)(struct rds_connection *conn, struct rm_rdma_op *op);
	int (*xmit_atomic)(struct rds_connection *conn, struct rm_atomic_op *op);
	int (*recv)(struct rds_connection *conn);
	int (*inc_copy_to_user)(struct rds_incoming *inc, struct iovec *iov,
				size_t size);
	void (*inc_free)(struct rds_incoming *inc);

	int (*cm_handle_connect)(struct rdma_cm_id *cm_id,
				 struct rdma_cm_event *event);
	int (*cm_initiate_connect)(struct rdma_cm_id *cm_id);
	void (*cm_connect_complete)(struct rds_connection *conn,
				    struct rdma_cm_event *event);

	unsigned int (*stats_info_copy)(struct rds_info_iterator *iter,
					unsigned int avail);
	void (*exit)(void);
	void *(*get_mr)(struct scatterlist *sg, unsigned long nr_sg,
			struct rds_sock *rs, u32 *key_ret);
	void (*sync_mr)(void *trans_private, int direction);
	void (*free_mr)(void *trans_private, int invalidate);
	void (*flush_mrs)(void);
};

struct rds_sock {
	struct sock		rs_sk;

	u64			rs_user_addr;
	u64			rs_user_bytes;

	/*
	 * bound_addr used for both incoming and outgoing, no INADDR_ANY
	 * support.
	 */
	struct hlist_node	rs_bound_node;
	__be32			rs_bound_addr;
	__be32			rs_conn_addr;
	__be16			rs_bound_port;
	__be16			rs_conn_port;
	struct rds_transport    *rs_transport;

	/*
	 * rds_sendmsg caches the conn it used the last time around.
	 * This helps avoid costly lookups.
	 */
	struct rds_connection	*rs_conn;

	/* flag indicating we were congested or not */
	int			rs_congested;
	/* seen congestion (ENOBUFS) when sending? */
	int			rs_seen_congestion;

	/* rs_lock protects all these adjacent members before the newline */
	spinlock_t		rs_lock;
	struct list_head	rs_send_queue;
	u32			rs_snd_bytes;
	int			rs_rcv_bytes;
	struct list_head	rs_notify_queue;	/* currently used for failed RDMAs */

	/* Congestion wake_up. If rs_cong_monitor is set, we use cong_mask
	 * to decide whether the application should be woken up.
	 * If not set, we use rs_cong_track to find out whether a cong map
	 * update arrived.
	 */
	uint64_t		rs_cong_mask;
	uint64_t		rs_cong_notify;
	struct list_head	rs_cong_list;
	unsigned long		rs_cong_track;

	/*
	 * rs_recv_lock protects the receive queue, and is
	 * used to serialize with rds_release.
	 */
	rwlock_t		rs_recv_lock;
	struct list_head	rs_recv_queue;

	/* just for stats reporting */
	struct list_head	rs_item;

	/* these have their own lock */
	spinlock_t		rs_rdma_lock;
	struct rb_root		rs_rdma_keys;

	/* Socket options - in case there will be more */
	unsigned char		rs_recverr,
				rs_cong_monitor;
};

static inline struct rds_sock *rds_sk_to_rs(const struct sock *sk)
{
	return container_of(sk, struct rds_sock, rs_sk);
}
static inline struct sock *rds_rs_to_sk(struct rds_sock *rs)
{
	return &rs->rs_sk;
}

/*
 * The stack assigns sk_sndbuf and sk_rcvbuf to twice the specified value
 * to account for overhead.  We don't account for overhead, we just apply
 * the number of payload bytes to the specified value.
 */
static inline int rds_sk_sndbuf(struct rds_sock *rs)
{
	return rds_rs_to_sk(rs)->sk_sndbuf / 2;
}
static inline int rds_sk_rcvbuf(struct rds_sock *rs)
{
	return rds_rs_to_sk(rs)->sk_rcvbuf / 2;
}

struct rds_statistics {
	uint64_t	s_conn_reset;
	uint64_t	s_recv_drop_bad_checksum;
	uint64_t	s_recv_drop_old_seq;
	uint64_t	s_recv_drop_no_sock;
	uint64_t	s_recv_drop_dead_sock;
	uint64_t	s_recv_deliver_raced;
	uint64_t	s_recv_delivered;
	uint64_t	s_recv_queued;
	uint64_t	s_recv_immediate_retry;
	uint64_t	s_recv_delayed_retry;
	uint64_t	s_recv_ack_required;
	uint64_t	s_recv_rdma_bytes;
	uint64_t	s_recv_ping;
	uint64_t	s_send_queue_empty;
	uint64_t	s_send_queue_full;
	uint64_t	s_send_lock_contention;
	uint64_t	s_send_lock_queue_raced;
	uint64_t	s_send_immediate_retry;
	uint64_t	s_send_delayed_retry;
	uint64_t	s_send_drop_acked;
	uint64_t	s_send_ack_required;
	uint64_t	s_send_queued;
	uint64_t	s_send_rdma;
	uint64_t	s_send_rdma_bytes;
	uint64_t	s_send_pong;
	uint64_t	s_page_remainder_hit;
	uint64_t	s_page_remainder_miss;
	uint64_t	s_copy_to_user;
	uint64_t	s_copy_from_user;
	uint64_t	s_cong_update_queued;
	uint64_t	s_cong_update_received;
	uint64_t	s_cong_send_error;
	uint64_t	s_cong_send_blocked;
};

/* af_rds.c */
void rds_sock_addref(struct rds_sock *rs);
void rds_sock_put(struct rds_sock *rs);
void rds_wake_sk_sleep(struct rds_sock *rs);
static inline void __rds_wake_sk_sleep(struct sock *sk)
{
	wait_queue_head_t *waitq = sk_sleep(sk);

	if (!sock_flag(sk, SOCK_DEAD) && waitq)
		wake_up(waitq);
}
extern wait_queue_head_t rds_poll_waitq;


/* bind.c */
int rds_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
void rds_remove_bound(struct rds_sock *rs);
struct rds_sock *rds_find_bound(__be32 addr, __be16 port);

/* cong.c */
int rds_cong_get_maps(struct rds_connection *conn);
void rds_cong_add_conn(struct rds_connection *conn);
void rds_cong_remove_conn(struct rds_connection *conn);
void rds_cong_set_bit(struct rds_cong_map *map, __be16 port);
void rds_cong_clear_bit(struct rds_cong_map *map, __be16 port);
int rds_cong_wait(struct rds_cong_map *map, __be16 port, int nonblock, struct rds_sock *rs);
void rds_cong_queue_updates(struct rds_cong_map *map);
void rds_cong_map_updated(struct rds_cong_map *map, uint64_t);
int rds_cong_updated_since(unsigned long *recent);
void rds_cong_add_socket(struct rds_sock *);
void rds_cong_remove_socket(struct rds_sock *);
void rds_cong_exit(void);
struct rds_message *rds_cong_update_alloc(struct rds_connection *conn);

/* conn.c */
int rds_conn_init(void);
void rds_conn_exit(void);
struct rds_connection *rds_conn_create(__be32 laddr, __be32 faddr,
				       struct rds_transport *trans, gfp_t gfp);
struct rds_connection *rds_conn_create_outgoing(__be32 laddr, __be32 faddr,
			       struct rds_transport *trans, gfp_t gfp);
void rds_conn_shutdown(struct rds_connection *conn);
void rds_conn_destroy(struct rds_connection *conn);
void rds_conn_reset(struct rds_connection *conn);
void rds_conn_drop(struct rds_connection *conn);
void rds_conn_connect_if_down(struct rds_connection *conn);
void rds_for_each_conn_info(struct socket *sock, unsigned int len,
			  struct rds_info_iterator *iter,
			  struct rds_info_lengths *lens,
			  int (*visitor)(struct rds_connection *, void *),
			  size_t item_len);
void __rds_conn_error(struct rds_connection *conn, const char *, ...)
				__attribute__ ((format (printf, 2, 3)));
#define rds_conn_error(conn, fmt...) \
	__rds_conn_error(conn, KERN_WARNING "RDS: " fmt)

static inline int
rds_conn_transition(struct rds_connection *conn, int old, int new)
{
	return atomic_cmpxchg(&conn->c_state, old, new) == old;
}

static inline int
rds_conn_state(struct rds_connection *conn)
{
	return atomic_read(&conn->c_state);
}

static inline int
rds_conn_up(struct rds_connection *conn)
{
	return atomic_read(&conn->c_state) == RDS_CONN_UP;
}

static inline int
rds_conn_connecting(struct rds_connection *conn)
{
	return atomic_read(&conn->c_state) == RDS_CONN_CONNECTING;
}

/* message.c */
struct rds_message *rds_message_alloc(unsigned int nents, gfp_t gfp);
struct scatterlist *rds_message_alloc_sgs(struct rds_message *rm, int nents);
int rds_message_copy_from_user(struct rds_message *rm, struct iovec *first_iov,
					       size_t total_len);
struct rds_message *rds_message_map_pages(unsigned long *page_addrs, unsigned int total_len);
void rds_message_populate_header(struct rds_header *hdr, __be16 sport,
				 __be16 dport, u64 seq);
int rds_message_add_extension(struct rds_header *hdr,
			      unsigned int type, const void *data, unsigned int len);
int rds_message_next_extension(struct rds_header *hdr,
			       unsigned int *pos, void *buf, unsigned int *buflen);
int rds_message_add_version_extension(struct rds_header *hdr, unsigned int version);
int rds_message_get_version_extension(struct rds_header *hdr, unsigned int *version);
int rds_message_add_rdma_dest_extension(struct rds_header *hdr, u32 r_key, u32 offset);
int rds_message_inc_copy_to_user(struct rds_incoming *inc,
				 struct iovec *first_iov, size_t size);
void rds_message_inc_free(struct rds_incoming *inc);
void rds_message_addref(struct rds_message *rm);
void rds_message_put(struct rds_message *rm);
void rds_message_wait(struct rds_message *rm);
void rds_message_unmapped(struct rds_message *rm);

static inline void rds_message_make_checksum(struct rds_header *hdr)
{
	hdr->h_csum = 0;
	hdr->h_csum = ip_fast_csum((void *) hdr, sizeof(*hdr) >> 2);
}

static inline int rds_message_verify_checksum(const struct rds_header *hdr)
{
	return !hdr->h_csum || ip_fast_csum((void *) hdr, sizeof(*hdr) >> 2) == 0;
}


/* page.c */
int rds_page_remainder_alloc(struct scatterlist *scat, unsigned long bytes,
			     gfp_t gfp);
int rds_page_copy_user(struct page *page, unsigned long offset,
		       void __user *ptr, unsigned long bytes,
		       int to_user);
#define rds_page_copy_to_user(page, offset, ptr, bytes) \
	rds_page_copy_user(page, offset, ptr, bytes, 1)
#define rds_page_copy_from_user(page, offset, ptr, bytes) \
	rds_page_copy_user(page, offset, ptr, bytes, 0)
void rds_page_exit(void);

/* recv.c */
void rds_inc_init(struct rds_incoming *inc, struct rds_connection *conn,
		  __be32 saddr);
void rds_inc_addref(struct rds_incoming *inc);
void rds_inc_put(struct rds_incoming *inc);
void rds_recv_incoming(struct rds_connection *conn, __be32 saddr, __be32 daddr,
		       struct rds_incoming *inc, gfp_t gfp, enum km_type km);
int rds_recvmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg,
		size_t size, int msg_flags);
void rds_clear_recv_queue(struct rds_sock *rs);
int rds_notify_queue_get(struct rds_sock *rs, struct msghdr *msg);
void rds_inc_info_copy(struct rds_incoming *inc,
		       struct rds_info_iterator *iter,
		       __be32 saddr, __be32 daddr, int flip);

/* send.c */
int rds_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg,
		size_t payload_len);
void rds_send_reset(struct rds_connection *conn);
int rds_send_xmit(struct rds_connection *conn);
struct sockaddr_in;
void rds_send_drop_to(struct rds_sock *rs, struct sockaddr_in *dest);
typedef int (*is_acked_func)(struct rds_message *rm, uint64_t ack);
void rds_send_drop_acked(struct rds_connection *conn, u64 ack,
			 is_acked_func is_acked);
void rds_send_remove_from_sock(struct list_head *messages, int status);
int rds_send_pong(struct rds_connection *conn, __be16 dport);
struct rds_message *rds_send_get_message(struct rds_connection *,
					 struct rm_rdma_op *);

/* rdma.c */
void rds_rdma_unuse(struct rds_sock *rs, u32 r_key, int force);
int rds_get_mr(struct rds_sock *rs, char __user *optval, int optlen);
int rds_get_mr_for_dest(struct rds_sock *rs, char __user *optval, int optlen);
int rds_free_mr(struct rds_sock *rs, char __user *optval, int optlen);
void rds_rdma_drop_keys(struct rds_sock *rs);
int rds_rdma_extra_size(struct rds_rdma_args *args);
int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
			  struct cmsghdr *cmsg);
int rds_cmsg_rdma_dest(struct rds_sock *rs, struct rds_message *rm,
			  struct cmsghdr *cmsg);
int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
			  struct cmsghdr *cmsg);
int rds_cmsg_rdma_map(struct rds_sock *rs, struct rds_message *rm,
			  struct cmsghdr *cmsg);
void rds_rdma_free_op(struct rm_rdma_op *ro);
void rds_atomic_free_op(struct rm_atomic_op *ao);
void rds_rdma_send_complete(struct rds_message *rm, int wc_status);
void rds_atomic_send_complete(struct rds_message *rm, int wc_status);
int rds_cmsg_atomic(struct rds_sock *rs, struct rds_message *rm,
		    struct cmsghdr *cmsg);

extern void __rds_put_mr_final(struct rds_mr *mr);
static inline void rds_mr_put(struct rds_mr *mr)
{
	if (atomic_dec_and_test(&mr->r_refcount))
		__rds_put_mr_final(mr);
}

/* stats.c */
DECLARE_PER_CPU_SHARED_ALIGNED(struct rds_statistics, rds_stats);
#define rds_stats_inc_which(which, member) do {		\
	per_cpu(which, get_cpu()).member++;		\
	put_cpu();					\
} while (0)
#define rds_stats_inc(member) rds_stats_inc_which(rds_stats, member)
#define rds_stats_add_which(which, member, count) do {		\
	per_cpu(which, get_cpu()).member += count;	\
	put_cpu();					\
} while (0)
#define rds_stats_add(member, count) rds_stats_add_which(rds_stats, member, count)
int rds_stats_init(void);
void rds_stats_exit(void);
void rds_stats_info_copy(struct rds_info_iterator *iter,
			 uint64_t *values, const char *const *names,
			 size_t nr);

/* sysctl.c */
int rds_sysctl_init(void);
void rds_sysctl_exit(void);
extern unsigned long rds_sysctl_sndbuf_min;
extern unsigned long rds_sysctl_sndbuf_default;
extern unsigned long rds_sysctl_sndbuf_max;
extern unsigned long rds_sysctl_reconnect_min_jiffies;
extern unsigned long rds_sysctl_reconnect_max_jiffies;
extern unsigned int  rds_sysctl_max_unacked_packets;
extern unsigned int  rds_sysctl_max_unacked_bytes;
extern unsigned int  rds_sysctl_ping_enable;
extern unsigned long rds_sysctl_trace_flags;
extern unsigned int  rds_sysctl_trace_level;

/* threads.c */
int rds_threads_init(void);
void rds_threads_exit(void);
extern struct workqueue_struct *rds_wq;
void rds_queue_reconnect(struct rds_connection *conn);
void rds_connect_worker(struct work_struct *);
void rds_shutdown_worker(struct work_struct *);
void rds_send_worker(struct work_struct *);
void rds_recv_worker(struct work_struct *);
void rds_connect_complete(struct rds_connection *conn);

/* transport.c */
int rds_trans_register(struct rds_transport *trans);
void rds_trans_unregister(struct rds_transport *trans);
struct rds_transport *rds_trans_get_preferred(__be32 addr);
void rds_trans_put(struct rds_transport *trans);
unsigned int rds_trans_stats_info_copy(struct rds_info_iterator *iter,
				       unsigned int avail);
int rds_trans_init(void);
void rds_trans_exit(void);

#endif