2 * net/tipc/msg.c: TIPC message header routines
4 * Copyright (c) 2000-2006, 2014-2015, Ericsson AB
5 * Copyright (c) 2005, 2010-2011, Wind River Systems
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the names of the copyright holders nor the names of its
17 * contributors may be used to endorse or promote products derived from
18 * this software without specific prior written permission.
20 * Alternatively, this software may be distributed under the terms of the
21 * GNU General Public License ("GPL") version 2 as published by the Free
22 * Software Foundation.
24 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
25 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
28 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
31 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
32 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
33 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34 * POSSIBILITY OF SUCH DAMAGE.
41 #include "name_table.h"
43 #define MAX_FORWARD_SIZE 1024
44 #define BUF_HEADROOM (LL_MAX_HEADER + 48)
45 #define BUF_TAILROOM 16
47 static unsigned int align(unsigned int i
)
53 * tipc_buf_acquire - creates a TIPC message buffer
54 * @size: message size (including TIPC header)
56 * Returns a new buffer with data pointers set to the specified size.
58 * NOTE: Headroom is reserved to allow prepending of a data link header.
59 * There may also be unrequested tailroom present at the buffer's end.
61 struct sk_buff
*tipc_buf_acquire(u32 size
, gfp_t gfp
)
64 unsigned int buf_size
= (BUF_HEADROOM
+ size
+ 3) & ~3u;
66 skb
= alloc_skb_fclone(buf_size
, gfp
);
68 skb_reserve(skb
, BUF_HEADROOM
);
75 void tipc_msg_init(u32 own_node
, struct tipc_msg
*m
, u32 user
, u32 type
,
80 msg_set_user(m
, user
);
81 msg_set_hdr_sz(m
, hsize
);
82 msg_set_size(m
, hsize
);
83 msg_set_prevnode(m
, own_node
);
84 msg_set_type(m
, type
);
85 if (hsize
> SHORT_H_SIZE
) {
86 msg_set_orignode(m
, own_node
);
87 msg_set_destnode(m
, dnode
);
91 struct sk_buff
*tipc_msg_create(uint user
, uint type
,
92 uint hdr_sz
, uint data_sz
, u32 dnode
,
93 u32 onode
, u32 dport
, u32 oport
, int errcode
)
98 buf
= tipc_buf_acquire(hdr_sz
+ data_sz
, GFP_ATOMIC
);
103 tipc_msg_init(onode
, msg
, user
, type
, hdr_sz
, dnode
);
104 msg_set_size(msg
, hdr_sz
+ data_sz
);
105 msg_set_origport(msg
, oport
);
106 msg_set_destport(msg
, dport
);
107 msg_set_errcode(msg
, errcode
);
108 if (hdr_sz
> SHORT_H_SIZE
) {
109 msg_set_orignode(msg
, onode
);
110 msg_set_destnode(msg
, dnode
);
115 /* tipc_buf_append(): Append a buffer to the fragment list of another buffer
116 * @*headbuf: in: NULL for first frag, otherwise value returned from prev call
117 * out: set when successful non-complete reassembly, otherwise NULL
118 * @*buf: in: the buffer to append. Always defined
119 * out: head buf after successful complete reassembly, otherwise NULL
120 * Returns 1 when reassembly complete, otherwise 0
122 int tipc_buf_append(struct sk_buff
**headbuf
, struct sk_buff
**buf
)
124 struct sk_buff
*head
= *headbuf
;
125 struct sk_buff
*frag
= *buf
;
126 struct sk_buff
*tail
= NULL
;
127 struct tipc_msg
*msg
;
136 fragid
= msg_type(msg
);
138 skb_pull(frag
, msg_hdr_sz(msg
));
140 if (fragid
== FIRST_FRAGMENT
) {
143 if (unlikely(skb_unclone(frag
, GFP_ATOMIC
)))
145 head
= *headbuf
= frag
;
147 TIPC_SKB_CB(head
)->tail
= NULL
;
148 if (skb_is_nonlinear(head
)) {
149 skb_walk_frags(head
, tail
) {
150 TIPC_SKB_CB(head
)->tail
= tail
;
153 skb_frag_list_init(head
);
161 if (skb_try_coalesce(head
, frag
, &headstolen
, &delta
)) {
162 kfree_skb_partial(frag
, headstolen
);
164 tail
= TIPC_SKB_CB(head
)->tail
;
165 if (!skb_has_frag_list(head
))
166 skb_shinfo(head
)->frag_list
= frag
;
169 head
->truesize
+= frag
->truesize
;
170 head
->data_len
+= frag
->len
;
171 head
->len
+= frag
->len
;
172 TIPC_SKB_CB(head
)->tail
= frag
;
175 if (fragid
== LAST_FRAGMENT
) {
176 TIPC_SKB_CB(head
)->validated
= false;
177 if (unlikely(!tipc_msg_validate(&head
)))
180 TIPC_SKB_CB(head
)->tail
= NULL
;
189 *buf
= *headbuf
= NULL
;
193 /* tipc_msg_validate - validate basic format of received message
195 * This routine ensures a TIPC message has an acceptable header, and at least
196 * as much data as the header indicates it should. The routine also ensures
197 * that the entire message header is stored in the main fragment of the message
198 * buffer, to simplify future access to message header fields.
200 * Note: Having extra info present in the message header or data areas is OK.
201 * TIPC will ignore the excess, under the assumption that it is optional info
202 * introduced by a later release of the protocol.
204 bool tipc_msg_validate(struct sk_buff
**_skb
)
206 struct sk_buff
*skb
= *_skb
;
207 struct tipc_msg
*hdr
;
210 /* Ensure that flow control ratio condition is satisfied */
211 if (unlikely(skb
->truesize
/ buf_roundup_len(skb
) >= 4)) {
212 skb
= skb_copy_expand(skb
, BUF_HEADROOM
, 0, GFP_ATOMIC
);
219 if (unlikely(TIPC_SKB_CB(skb
)->validated
))
221 if (unlikely(!pskb_may_pull(skb
, MIN_H_SIZE
)))
224 hsz
= msg_hdr_sz(buf_msg(skb
));
225 if (unlikely(hsz
< MIN_H_SIZE
) || (hsz
> MAX_H_SIZE
))
227 if (unlikely(!pskb_may_pull(skb
, hsz
)))
231 if (unlikely(msg_version(hdr
) != TIPC_VERSION
))
235 if (unlikely(msz
< hsz
))
237 if (unlikely((msz
- hsz
) > TIPC_MAX_USER_MSG_SIZE
))
239 if (unlikely(skb
->len
< msz
))
242 TIPC_SKB_CB(skb
)->validated
= true;
247 * tipc_msg_build - create buffer chain containing specified header and data
248 * @mhdr: Message header, to be prepended to data
250 * @dsz: Total length of user data
251 * @pktmax: Max packet size that can be used
252 * @list: Buffer or chain of buffers to be returned to caller
254 * Note that the recursive call we are making here is safe, since it can
255 * logically go only one further level down.
257 * Returns message data size or errno: -ENOMEM, -EFAULT
259 int tipc_msg_build(struct tipc_msg
*mhdr
, struct msghdr
*m
, int offset
,
260 int dsz
, int pktmax
, struct sk_buff_head
*list
)
262 int mhsz
= msg_hdr_sz(mhdr
);
263 struct tipc_msg pkthdr
;
264 int msz
= mhsz
+ dsz
;
273 msg_set_size(mhdr
, msz
);
275 /* No fragmentation needed? */
276 if (likely(msz
<= pktmax
)) {
277 skb
= tipc_buf_acquire(msz
, GFP_KERNEL
);
279 /* Fall back to smaller MTU if node local message */
280 if (unlikely(!skb
)) {
281 if (pktmax
!= MAX_MSG_SIZE
)
283 rc
= tipc_msg_build(mhdr
, m
, offset
, dsz
, FB_MTU
, list
);
286 if (tipc_msg_assemble(list
))
291 __skb_queue_tail(list
, skb
);
292 skb_copy_to_linear_data(skb
, mhdr
, mhsz
);
293 pktpos
= skb
->data
+ mhsz
;
294 if (copy_from_iter_full(pktpos
, dsz
, &m
->msg_iter
))
300 /* Prepare reusable fragment header */
301 tipc_msg_init(msg_prevnode(mhdr
), &pkthdr
, MSG_FRAGMENTER
,
302 FIRST_FRAGMENT
, INT_H_SIZE
, msg_destnode(mhdr
));
303 msg_set_size(&pkthdr
, pktmax
);
304 msg_set_fragm_no(&pkthdr
, pktno
);
305 msg_set_importance(&pkthdr
, msg_importance(mhdr
));
307 /* Prepare first fragment */
308 skb
= tipc_buf_acquire(pktmax
, GFP_KERNEL
);
312 __skb_queue_tail(list
, skb
);
314 skb_copy_to_linear_data(skb
, &pkthdr
, INT_H_SIZE
);
315 pktpos
+= INT_H_SIZE
;
316 pktrem
-= INT_H_SIZE
;
317 skb_copy_to_linear_data_offset(skb
, INT_H_SIZE
, mhdr
, mhsz
);
325 if (!copy_from_iter_full(pktpos
, pktrem
, &m
->msg_iter
)) {
334 /* Prepare new fragment: */
335 if (drem
< (pktmax
- INT_H_SIZE
))
336 pktsz
= drem
+ INT_H_SIZE
;
339 skb
= tipc_buf_acquire(pktsz
, GFP_KERNEL
);
345 __skb_queue_tail(list
, skb
);
346 msg_set_type(&pkthdr
, FRAGMENT
);
347 msg_set_size(&pkthdr
, pktsz
);
348 msg_set_fragm_no(&pkthdr
, ++pktno
);
349 skb_copy_to_linear_data(skb
, &pkthdr
, INT_H_SIZE
);
350 pktpos
= skb
->data
+ INT_H_SIZE
;
351 pktrem
= pktsz
- INT_H_SIZE
;
354 msg_set_type(buf_msg(skb
), LAST_FRAGMENT
);
357 __skb_queue_purge(list
);
358 __skb_queue_head_init(list
);
363 * tipc_msg_bundle(): Append contents of a buffer to tail of an existing one
364 * @skb: the buffer to append to ("bundle")
365 * @msg: message to be appended
366 * @mtu: max allowable size for the bundle buffer
367 * Consumes buffer if successful
368 * Returns true if bundling could be performed, otherwise false
370 bool tipc_msg_bundle(struct sk_buff
*skb
, struct tipc_msg
*msg
, u32 mtu
)
372 struct tipc_msg
*bmsg
;
374 unsigned int msz
= msg_size(msg
);
376 u32 max
= mtu
- INT_H_SIZE
;
378 if (likely(msg_user(msg
) == MSG_FRAGMENTER
))
383 bsz
= msg_size(bmsg
);
387 if (unlikely(msg_user(msg
) == TUNNEL_PROTOCOL
))
389 if (unlikely(msg_user(msg
) == BCAST_PROTOCOL
))
391 if (unlikely(msg_user(bmsg
) != MSG_BUNDLER
))
393 if (unlikely(skb_tailroom(skb
) < (pad
+ msz
)))
395 if (unlikely(max
< (start
+ msz
)))
397 if ((msg_importance(msg
) < TIPC_SYSTEM_IMPORTANCE
) &&
398 (msg_importance(bmsg
) == TIPC_SYSTEM_IMPORTANCE
))
401 skb_put(skb
, pad
+ msz
);
402 skb_copy_to_linear_data_offset(skb
, start
, msg
, msz
);
403 msg_set_size(bmsg
, start
+ msz
);
404 msg_set_msgcnt(bmsg
, msg_msgcnt(bmsg
) + 1);
409 * tipc_msg_extract(): extract bundled inner packet from buffer
410 * @skb: buffer to be extracted from.
411 * @iskb: extracted inner buffer, to be returned
412 * @pos: position in outer message of msg to be extracted.
413 * Returns position of next msg
414 * Consumes outer buffer when last packet extracted
415 * Returns true when when there is an extracted buffer, otherwise false
417 bool tipc_msg_extract(struct sk_buff
*skb
, struct sk_buff
**iskb
, int *pos
)
419 struct tipc_msg
*hdr
, *ihdr
;
423 if (unlikely(skb_linearize(skb
)))
427 if (unlikely(*pos
> (msg_data_sz(hdr
) - MIN_H_SIZE
)))
430 ihdr
= (struct tipc_msg
*)(msg_data(hdr
) + *pos
);
431 imsz
= msg_size(ihdr
);
433 if ((*pos
+ imsz
) > msg_data_sz(hdr
))
436 *iskb
= tipc_buf_acquire(imsz
, GFP_ATOMIC
);
440 skb_copy_to_linear_data(*iskb
, ihdr
, imsz
);
441 if (unlikely(!tipc_msg_validate(iskb
)))
454 * tipc_msg_make_bundle(): Create bundle buf and append message to its tail
455 * @list: the buffer chain, where head is the buffer to replace/append
456 * @skb: buffer to be created, appended to and returned in case of success
457 * @msg: message to be appended
458 * @mtu: max allowable size for the bundle buffer, inclusive header
459 * @dnode: destination node for message. (Not always present in header)
460 * Returns true if success, otherwise false
462 bool tipc_msg_make_bundle(struct sk_buff
**skb
, struct tipc_msg
*msg
,
465 struct sk_buff
*_skb
;
466 struct tipc_msg
*bmsg
;
467 u32 msz
= msg_size(msg
);
468 u32 max
= mtu
- INT_H_SIZE
;
470 if (msg_user(msg
) == MSG_FRAGMENTER
)
472 if (msg_user(msg
) == TUNNEL_PROTOCOL
)
474 if (msg_user(msg
) == BCAST_PROTOCOL
)
479 _skb
= tipc_buf_acquire(max
, GFP_ATOMIC
);
483 skb_trim(_skb
, INT_H_SIZE
);
484 bmsg
= buf_msg(_skb
);
485 tipc_msg_init(msg_prevnode(msg
), bmsg
, MSG_BUNDLER
, 0,
488 msg_set_importance(bmsg
, TIPC_CRITICAL_IMPORTANCE
);
490 msg_set_importance(bmsg
, TIPC_SYSTEM_IMPORTANCE
);
491 msg_set_seqno(bmsg
, msg_seqno(msg
));
492 msg_set_ack(bmsg
, msg_ack(msg
));
493 msg_set_bcast_ack(bmsg
, msg_bcast_ack(msg
));
494 tipc_msg_bundle(_skb
, msg
, mtu
);
500 * tipc_msg_reverse(): swap source and destination addresses and add error code
501 * @own_node: originating node id for reversed message
502 * @skb: buffer containing message to be reversed; will be consumed
503 * @err: error code to be set in message, if any
504 * Replaces consumed buffer with new one when successful
505 * Returns true if success, otherwise false
507 bool tipc_msg_reverse(u32 own_node
, struct sk_buff
**skb
, int err
)
509 struct sk_buff
*_skb
= *skb
;
510 struct tipc_msg
*_hdr
, *hdr
;
513 if (skb_linearize(_skb
))
515 _hdr
= buf_msg(_skb
);
516 dlen
= min_t(uint
, msg_data_sz(_hdr
), MAX_FORWARD_SIZE
);
517 hlen
= msg_hdr_sz(_hdr
);
519 if (msg_dest_droppable(_hdr
))
521 if (msg_errcode(_hdr
))
524 /* Never return SHORT header */
525 if (hlen
== SHORT_H_SIZE
)
528 /* Don't return data along with SYN+, - sender has a clone */
529 if (msg_is_syn(_hdr
) && err
== TIPC_ERR_OVERLOAD
)
532 /* Allocate new buffer to return */
533 *skb
= tipc_buf_acquire(hlen
+ dlen
, GFP_ATOMIC
);
536 memcpy((*skb
)->data
, _skb
->data
, msg_hdr_sz(_hdr
));
537 memcpy((*skb
)->data
+ hlen
, msg_data(_hdr
), dlen
);
539 /* Build reverse header in new buffer */
541 msg_set_hdr_sz(hdr
, hlen
);
542 msg_set_errcode(hdr
, err
);
543 msg_set_non_seq(hdr
, 0);
544 msg_set_origport(hdr
, msg_destport(_hdr
));
545 msg_set_destport(hdr
, msg_origport(_hdr
));
546 msg_set_destnode(hdr
, msg_prevnode(_hdr
));
547 msg_set_prevnode(hdr
, own_node
);
548 msg_set_orignode(hdr
, own_node
);
549 msg_set_size(hdr
, hlen
+ dlen
);
559 bool tipc_msg_skb_clone(struct sk_buff_head
*msg
, struct sk_buff_head
*cpy
)
561 struct sk_buff
*skb
, *_skb
;
563 skb_queue_walk(msg
, skb
) {
564 _skb
= skb_clone(skb
, GFP_ATOMIC
);
566 __skb_queue_purge(cpy
);
567 pr_err_ratelimited("Failed to clone buffer chain\n");
570 __skb_queue_tail(cpy
, _skb
);
576 * tipc_msg_lookup_dest(): try to find new destination for named message
577 * @skb: the buffer containing the message.
578 * @err: error code to be used by caller if lookup fails
579 * Does not consume buffer
580 * Returns true if a destination is found, false otherwise
582 bool tipc_msg_lookup_dest(struct net
*net
, struct sk_buff
*skb
, int *err
)
584 struct tipc_msg
*msg
= buf_msg(skb
);
586 u32 onode
= tipc_own_addr(net
);
588 if (!msg_isdata(msg
))
592 if (msg_errcode(msg
))
594 *err
= TIPC_ERR_NO_NAME
;
595 if (skb_linearize(skb
))
598 if (msg_reroute_cnt(msg
))
600 dnode
= tipc_scope2node(net
, msg_lookup_scope(msg
));
601 dport
= tipc_nametbl_translate(net
, msg_nametype(msg
),
602 msg_nameinst(msg
), &dnode
);
605 msg_incr_reroute_cnt(msg
);
607 msg_set_prevnode(msg
, onode
);
608 msg_set_destnode(msg
, dnode
);
609 msg_set_destport(msg
, dport
);
612 if (!skb_cloned(skb
))
618 /* tipc_msg_assemble() - assemble chain of fragments into one message
620 bool tipc_msg_assemble(struct sk_buff_head
*list
)
622 struct sk_buff
*skb
, *tmp
= NULL
;
624 if (skb_queue_len(list
) == 1)
627 while ((skb
= __skb_dequeue(list
))) {
629 if (tipc_buf_append(&tmp
, &skb
)) {
630 __skb_queue_tail(list
, skb
);
636 __skb_queue_purge(list
);
637 __skb_queue_head_init(list
);
638 pr_warn("Failed do assemble buffer\n");
642 /* tipc_msg_reassemble() - clone a buffer chain of fragments and
643 * reassemble the clones into one message
645 bool tipc_msg_reassemble(struct sk_buff_head
*list
, struct sk_buff_head
*rcvq
)
647 struct sk_buff
*skb
, *_skb
;
648 struct sk_buff
*frag
= NULL
;
649 struct sk_buff
*head
= NULL
;
652 /* Copy header if single buffer */
653 if (skb_queue_len(list
) == 1) {
654 skb
= skb_peek(list
);
655 hdr_len
= skb_headroom(skb
) + msg_hdr_sz(buf_msg(skb
));
656 _skb
= __pskb_copy(skb
, hdr_len
, GFP_ATOMIC
);
659 __skb_queue_tail(rcvq
, _skb
);
663 /* Clone all fragments and reassemble */
664 skb_queue_walk(list
, skb
) {
665 frag
= skb_clone(skb
, GFP_ATOMIC
);
669 if (tipc_buf_append(&head
, &frag
))
674 __skb_queue_tail(rcvq
, frag
);
677 pr_warn("Failed do clone local mcast rcv buffer\n");
682 bool tipc_msg_pskb_copy(u32 dst
, struct sk_buff_head
*msg
,
683 struct sk_buff_head
*cpy
)
685 struct sk_buff
*skb
, *_skb
;
687 skb_queue_walk(msg
, skb
) {
688 _skb
= pskb_copy(skb
, GFP_ATOMIC
);
690 __skb_queue_purge(cpy
);
693 msg_set_destnode(buf_msg(_skb
), dst
);
694 __skb_queue_tail(cpy
, _skb
);
699 /* tipc_skb_queue_sorted(); sort pkt into list according to sequence number
700 * @list: list to be appended to
701 * @seqno: sequence number of buffer to add
702 * @skb: buffer to add
704 void __tipc_skb_queue_sorted(struct sk_buff_head
*list
, u16 seqno
,
707 struct sk_buff
*_skb
, *tmp
;
709 if (skb_queue_empty(list
) || less(seqno
, buf_seqno(skb_peek(list
)))) {
710 __skb_queue_head(list
, skb
);
714 if (more(seqno
, buf_seqno(skb_peek_tail(list
)))) {
715 __skb_queue_tail(list
, skb
);
719 skb_queue_walk_safe(list
, _skb
, tmp
) {
720 if (more(seqno
, buf_seqno(_skb
)))
722 if (seqno
== buf_seqno(_skb
))
724 __skb_queue_before(list
, _skb
, skb
);
730 void tipc_skb_reject(struct net
*net
, int err
, struct sk_buff
*skb
,
731 struct sk_buff_head
*xmitq
)
733 if (tipc_msg_reverse(tipc_own_addr(net
), &skb
, err
))
734 __skb_queue_tail(xmitq
, skb
);