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) {
144 if (skb_has_frag_list(frag) && __skb_linearize(frag))
146 frag = skb_unshare(frag, GFP_ATOMIC);
149 head = *headbuf = frag;
150 TIPC_SKB_CB(head)->tail = NULL;
157 if (skb_try_coalesce(head, frag, &headstolen, &delta)) {
158 kfree_skb_partial(frag, headstolen);
160 tail = TIPC_SKB_CB(head)->tail;
161 if (!skb_has_frag_list(head))
162 skb_shinfo(head)->frag_list = frag;
165 head->truesize += frag->truesize;
166 head->data_len += frag->len;
167 head->len += frag->len;
168 TIPC_SKB_CB(head)->tail = frag;
171 if (fragid == LAST_FRAGMENT) {
172 TIPC_SKB_CB(head)->validated = false;
173 if (unlikely(!tipc_msg_validate(&head)))
176 TIPC_SKB_CB(head)->tail = NULL;
185 *buf = *headbuf = NULL;
189 /* tipc_msg_validate - validate basic format of received message
191 * This routine ensures a TIPC message has an acceptable header, and at least
192 * as much data as the header indicates it should. The routine also ensures
193 * that the entire message header is stored in the main fragment of the message
194 * buffer, to simplify future access to message header fields.
196 * Note: Having extra info present in the message header or data areas is OK.
197 * TIPC will ignore the excess, under the assumption that it is optional info
198 * introduced by a later release of the protocol.
200 bool tipc_msg_validate(struct sk_buff **_skb)
202 struct sk_buff *skb = *_skb;
203 struct tipc_msg *hdr;
206 /* Ensure that flow control ratio condition is satisfied */
207 if (unlikely(skb->truesize / buf_roundup_len(skb) >= 4)) {
208 skb = skb_copy_expand(skb, BUF_HEADROOM, 0, GFP_ATOMIC);
215 if (unlikely(TIPC_SKB_CB(skb)->validated))
217 if (unlikely(!pskb_may_pull(skb, MIN_H_SIZE)))
220 hsz = msg_hdr_sz(buf_msg(skb));
221 if (unlikely(hsz < MIN_H_SIZE) || (hsz > MAX_H_SIZE))
223 if (unlikely(!pskb_may_pull(skb, hsz)))
227 if (unlikely(msg_version(hdr) != TIPC_VERSION))
231 if (unlikely(msz < hsz))
233 if (unlikely((msz - hsz) > TIPC_MAX_USER_MSG_SIZE))
235 if (unlikely(skb->len < msz))
238 TIPC_SKB_CB(skb)->validated = true;
243 * tipc_msg_fragment - build a fragment skb list for TIPC message
245 * @skb: TIPC message skb
246 * @hdr: internal msg header to be put on the top of the fragments
247 * @pktmax: max size of a fragment incl. the header
248 * @frags: returned fragment skb list
250 * Returns 0 if the fragmentation is successful, otherwise: -EINVAL
253 int tipc_msg_fragment(struct sk_buff *skb, const struct tipc_msg *hdr,
254 int pktmax, struct sk_buff_head *frags)
256 int pktno, nof_fragms, dsz, dmax, eat;
257 struct tipc_msg *_hdr;
258 struct sk_buff *_skb;
261 /* Non-linear buffer? */
262 if (skb_linearize(skb))
265 data = (u8 *)skb->data;
266 dsz = msg_size(buf_msg(skb));
267 dmax = pktmax - INT_H_SIZE;
268 if (dsz <= dmax || !dmax)
271 nof_fragms = dsz / dmax + 1;
272 for (pktno = 1; pktno <= nof_fragms; pktno++) {
273 if (pktno < nof_fragms)
277 /* Allocate a new fragment */
278 _skb = tipc_buf_acquire(INT_H_SIZE + eat, GFP_ATOMIC);
282 __skb_queue_tail(frags, _skb);
283 /* Copy header & data to the fragment */
284 skb_copy_to_linear_data(_skb, hdr, INT_H_SIZE);
285 skb_copy_to_linear_data_offset(_skb, INT_H_SIZE, data, eat);
287 /* Update the fragment's header */
288 _hdr = buf_msg(_skb);
289 msg_set_fragm_no(_hdr, pktno);
290 msg_set_nof_fragms(_hdr, nof_fragms);
291 msg_set_size(_hdr, INT_H_SIZE + eat);
296 __skb_queue_purge(frags);
297 __skb_queue_head_init(frags);
302 * tipc_msg_build - create buffer chain containing specified header and data
303 * @mhdr: Message header, to be prepended to data
305 * @dsz: Total length of user data
306 * @pktmax: Max packet size that can be used
307 * @list: Buffer or chain of buffers to be returned to caller
309 * Note that the recursive call we are making here is safe, since it can
310 * logically go only one further level down.
312 * Returns message data size or errno: -ENOMEM, -EFAULT
314 int tipc_msg_build(struct tipc_msg *mhdr, struct msghdr *m, int offset,
315 int dsz, int pktmax, struct sk_buff_head *list)
317 int mhsz = msg_hdr_sz(mhdr);
318 struct tipc_msg pkthdr;
319 int msz = mhsz + dsz;
328 msg_set_size(mhdr, msz);
330 /* No fragmentation needed? */
331 if (likely(msz <= pktmax)) {
332 skb = tipc_buf_acquire(msz, GFP_KERNEL);
334 /* Fall back to smaller MTU if node local message */
335 if (unlikely(!skb)) {
336 if (pktmax != MAX_MSG_SIZE)
338 rc = tipc_msg_build(mhdr, m, offset, dsz, FB_MTU, list);
341 if (tipc_msg_assemble(list))
346 __skb_queue_tail(list, skb);
347 skb_copy_to_linear_data(skb, mhdr, mhsz);
348 pktpos = skb->data + mhsz;
349 if (copy_from_iter_full(pktpos, dsz, &m->msg_iter))
355 /* Prepare reusable fragment header */
356 tipc_msg_init(msg_prevnode(mhdr), &pkthdr, MSG_FRAGMENTER,
357 FIRST_FRAGMENT, INT_H_SIZE, msg_destnode(mhdr));
358 msg_set_size(&pkthdr, pktmax);
359 msg_set_fragm_no(&pkthdr, pktno);
360 msg_set_importance(&pkthdr, msg_importance(mhdr));
362 /* Prepare first fragment */
363 skb = tipc_buf_acquire(pktmax, GFP_KERNEL);
367 __skb_queue_tail(list, skb);
369 skb_copy_to_linear_data(skb, &pkthdr, INT_H_SIZE);
370 pktpos += INT_H_SIZE;
371 pktrem -= INT_H_SIZE;
372 skb_copy_to_linear_data_offset(skb, INT_H_SIZE, mhdr, mhsz);
380 if (!copy_from_iter_full(pktpos, pktrem, &m->msg_iter)) {
389 /* Prepare new fragment: */
390 if (drem < (pktmax - INT_H_SIZE))
391 pktsz = drem + INT_H_SIZE;
394 skb = tipc_buf_acquire(pktsz, GFP_KERNEL);
400 __skb_queue_tail(list, skb);
401 msg_set_type(&pkthdr, FRAGMENT);
402 msg_set_size(&pkthdr, pktsz);
403 msg_set_fragm_no(&pkthdr, ++pktno);
404 skb_copy_to_linear_data(skb, &pkthdr, INT_H_SIZE);
405 pktpos = skb->data + INT_H_SIZE;
406 pktrem = pktsz - INT_H_SIZE;
409 msg_set_type(buf_msg(skb), LAST_FRAGMENT);
412 __skb_queue_purge(list);
413 __skb_queue_head_init(list);
418 * tipc_msg_bundle(): Append contents of a buffer to tail of an existing one
419 * @skb: the buffer to append to ("bundle")
420 * @msg: message to be appended
421 * @mtu: max allowable size for the bundle buffer
422 * Consumes buffer if successful
423 * Returns true if bundling could be performed, otherwise false
425 bool tipc_msg_bundle(struct sk_buff *skb, struct tipc_msg *msg, u32 mtu)
427 struct tipc_msg *bmsg;
429 unsigned int msz = msg_size(msg);
431 u32 max = mtu - INT_H_SIZE;
433 if (likely(msg_user(msg) == MSG_FRAGMENTER))
438 bsz = msg_size(bmsg);
442 if (unlikely(msg_user(msg) == TUNNEL_PROTOCOL))
444 if (unlikely(msg_user(msg) == BCAST_PROTOCOL))
446 if (unlikely(msg_user(bmsg) != MSG_BUNDLER))
448 if (unlikely(skb_tailroom(skb) < (pad + msz)))
450 if (unlikely(max < (start + msz)))
452 if ((msg_importance(msg) < TIPC_SYSTEM_IMPORTANCE) &&
453 (msg_importance(bmsg) == TIPC_SYSTEM_IMPORTANCE))
456 skb_put(skb, pad + msz);
457 skb_copy_to_linear_data_offset(skb, start, msg, msz);
458 msg_set_size(bmsg, start + msz);
459 msg_set_msgcnt(bmsg, msg_msgcnt(bmsg) + 1);
464 * tipc_msg_extract(): extract bundled inner packet from buffer
465 * @skb: buffer to be extracted from.
466 * @iskb: extracted inner buffer, to be returned
467 * @pos: position in outer message of msg to be extracted.
468 * Returns position of next msg
469 * Consumes outer buffer when last packet extracted
470 * Returns true when when there is an extracted buffer, otherwise false
472 bool tipc_msg_extract(struct sk_buff *skb, struct sk_buff **iskb, int *pos)
474 struct tipc_msg *hdr, *ihdr;
478 if (unlikely(skb_linearize(skb)))
482 if (unlikely(*pos > (msg_data_sz(hdr) - MIN_H_SIZE)))
485 ihdr = (struct tipc_msg *)(msg_data(hdr) + *pos);
486 imsz = msg_size(ihdr);
488 if ((*pos + imsz) > msg_data_sz(hdr))
491 *iskb = tipc_buf_acquire(imsz, GFP_ATOMIC);
495 skb_copy_to_linear_data(*iskb, ihdr, imsz);
496 if (unlikely(!tipc_msg_validate(iskb)))
509 * tipc_msg_make_bundle(): Create bundle buf and append message to its tail
510 * @list: the buffer chain, where head is the buffer to replace/append
511 * @skb: buffer to be created, appended to and returned in case of success
512 * @msg: message to be appended
513 * @mtu: max allowable size for the bundle buffer, inclusive header
514 * @dnode: destination node for message. (Not always present in header)
515 * Returns true if success, otherwise false
517 bool tipc_msg_make_bundle(struct sk_buff **skb, struct tipc_msg *msg,
520 struct sk_buff *_skb;
521 struct tipc_msg *bmsg;
522 u32 msz = msg_size(msg);
523 u32 max = mtu - INT_H_SIZE;
525 if (msg_user(msg) == MSG_FRAGMENTER)
527 if (msg_user(msg) == TUNNEL_PROTOCOL)
529 if (msg_user(msg) == BCAST_PROTOCOL)
534 _skb = tipc_buf_acquire(max, GFP_ATOMIC);
538 skb_trim(_skb, INT_H_SIZE);
539 bmsg = buf_msg(_skb);
540 tipc_msg_init(msg_prevnode(msg), bmsg, MSG_BUNDLER, 0,
542 msg_set_importance(bmsg, msg_importance(msg));
543 msg_set_seqno(bmsg, msg_seqno(msg));
544 msg_set_ack(bmsg, msg_ack(msg));
545 msg_set_bcast_ack(bmsg, msg_bcast_ack(msg));
546 tipc_msg_bundle(_skb, msg, mtu);
552 * tipc_msg_reverse(): swap source and destination addresses and add error code
553 * @own_node: originating node id for reversed message
554 * @skb: buffer containing message to be reversed; will be consumed
555 * @err: error code to be set in message, if any
556 * Replaces consumed buffer with new one when successful
557 * Returns true if success, otherwise false
559 bool tipc_msg_reverse(u32 own_node, struct sk_buff **skb, int err)
561 struct sk_buff *_skb = *skb;
562 struct tipc_msg *_hdr, *hdr;
565 if (skb_linearize(_skb))
567 _hdr = buf_msg(_skb);
568 dlen = min_t(uint, msg_data_sz(_hdr), MAX_FORWARD_SIZE);
569 hlen = msg_hdr_sz(_hdr);
571 if (msg_dest_droppable(_hdr))
573 if (msg_errcode(_hdr))
576 /* Never return SHORT header */
577 if (hlen == SHORT_H_SIZE)
580 /* Don't return data along with SYN+, - sender has a clone */
581 if (msg_is_syn(_hdr) && err == TIPC_ERR_OVERLOAD)
584 /* Allocate new buffer to return */
585 *skb = tipc_buf_acquire(hlen + dlen, GFP_ATOMIC);
588 memcpy((*skb)->data, _skb->data, msg_hdr_sz(_hdr));
589 memcpy((*skb)->data + hlen, msg_data(_hdr), dlen);
591 /* Build reverse header in new buffer */
593 msg_set_hdr_sz(hdr, hlen);
594 msg_set_errcode(hdr, err);
595 msg_set_non_seq(hdr, 0);
596 msg_set_origport(hdr, msg_destport(_hdr));
597 msg_set_destport(hdr, msg_origport(_hdr));
598 msg_set_destnode(hdr, msg_prevnode(_hdr));
599 msg_set_prevnode(hdr, own_node);
600 msg_set_orignode(hdr, own_node);
601 msg_set_size(hdr, hlen + dlen);
611 bool tipc_msg_skb_clone(struct sk_buff_head *msg, struct sk_buff_head *cpy)
613 struct sk_buff *skb, *_skb;
615 skb_queue_walk(msg, skb) {
616 _skb = skb_clone(skb, GFP_ATOMIC);
618 __skb_queue_purge(cpy);
619 pr_err_ratelimited("Failed to clone buffer chain\n");
622 __skb_queue_tail(cpy, _skb);
628 * tipc_msg_lookup_dest(): try to find new destination for named message
629 * @skb: the buffer containing the message.
630 * @err: error code to be used by caller if lookup fails
631 * Does not consume buffer
632 * Returns true if a destination is found, false otherwise
634 bool tipc_msg_lookup_dest(struct net *net, struct sk_buff *skb, int *err)
636 struct tipc_msg *msg = buf_msg(skb);
638 u32 onode = tipc_own_addr(net);
640 if (!msg_isdata(msg))
644 if (msg_errcode(msg))
646 *err = TIPC_ERR_NO_NAME;
647 if (skb_linearize(skb))
650 if (msg_reroute_cnt(msg))
652 dnode = tipc_scope2node(net, msg_lookup_scope(msg));
653 dport = tipc_nametbl_translate(net, msg_nametype(msg),
654 msg_nameinst(msg), &dnode);
657 msg_incr_reroute_cnt(msg);
659 msg_set_prevnode(msg, onode);
660 msg_set_destnode(msg, dnode);
661 msg_set_destport(msg, dport);
664 if (!skb_cloned(skb))
670 /* tipc_msg_assemble() - assemble chain of fragments into one message
672 bool tipc_msg_assemble(struct sk_buff_head *list)
674 struct sk_buff *skb, *tmp = NULL;
676 if (skb_queue_len(list) == 1)
679 while ((skb = __skb_dequeue(list))) {
681 if (tipc_buf_append(&tmp, &skb)) {
682 __skb_queue_tail(list, skb);
688 __skb_queue_purge(list);
689 __skb_queue_head_init(list);
690 pr_warn("Failed do assemble buffer\n");
694 /* tipc_msg_reassemble() - clone a buffer chain of fragments and
695 * reassemble the clones into one message
697 bool tipc_msg_reassemble(struct sk_buff_head *list, struct sk_buff_head *rcvq)
699 struct sk_buff *skb, *_skb;
700 struct sk_buff *frag = NULL;
701 struct sk_buff *head = NULL;
704 /* Copy header if single buffer */
705 if (skb_queue_len(list) == 1) {
706 skb = skb_peek(list);
707 hdr_len = skb_headroom(skb) + msg_hdr_sz(buf_msg(skb));
708 _skb = __pskb_copy(skb, hdr_len, GFP_ATOMIC);
711 __skb_queue_tail(rcvq, _skb);
715 /* Clone all fragments and reassemble */
716 skb_queue_walk(list, skb) {
717 frag = skb_clone(skb, GFP_ATOMIC);
721 if (tipc_buf_append(&head, &frag))
726 __skb_queue_tail(rcvq, frag);
729 pr_warn("Failed do clone local mcast rcv buffer\n");
734 bool tipc_msg_pskb_copy(u32 dst, struct sk_buff_head *msg,
735 struct sk_buff_head *cpy)
737 struct sk_buff *skb, *_skb;
739 skb_queue_walk(msg, skb) {
740 _skb = pskb_copy(skb, GFP_ATOMIC);
742 __skb_queue_purge(cpy);
745 msg_set_destnode(buf_msg(_skb), dst);
746 __skb_queue_tail(cpy, _skb);
751 /* tipc_skb_queue_sorted(); sort pkt into list according to sequence number
752 * @list: list to be appended to
753 * @seqno: sequence number of buffer to add
754 * @skb: buffer to add
756 void __tipc_skb_queue_sorted(struct sk_buff_head *list, u16 seqno,
759 struct sk_buff *_skb, *tmp;
761 if (skb_queue_empty(list) || less(seqno, buf_seqno(skb_peek(list)))) {
762 __skb_queue_head(list, skb);
766 if (more(seqno, buf_seqno(skb_peek_tail(list)))) {
767 __skb_queue_tail(list, skb);
771 skb_queue_walk_safe(list, _skb, tmp) {
772 if (more(seqno, buf_seqno(_skb)))
774 if (seqno == buf_seqno(_skb))
776 __skb_queue_before(list, _skb, skb);
782 void tipc_skb_reject(struct net *net, int err, struct sk_buff *skb,
783 struct sk_buff_head *xmitq)
785 if (tipc_msg_reverse(tipc_own_addr(net), &skb, err))
786 __skb_queue_tail(xmitq, skb);