4 * Copyright (c) 2016 Tom Herbert <tom@herbertland.com>
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2
8 * as published by the Free Software Foundation.
11 #include <linux/bpf.h>
12 #include <linux/errno.h>
13 #include <linux/errqueue.h>
14 #include <linux/file.h>
16 #include <linux/kernel.h>
17 #include <linux/module.h>
18 #include <linux/net.h>
19 #include <linux/netdevice.h>
20 #include <linux/poll.h>
21 #include <linux/rculist.h>
22 #include <linux/skbuff.h>
23 #include <linux/socket.h>
24 #include <linux/uaccess.h>
25 #include <linux/workqueue.h>
26 #include <net/strparser.h>
27 #include <net/netns/generic.h>
31 static struct workqueue_struct *strp_wq;
34 /* Internal cb structure. struct strp_rx_msg must be first for passing
37 struct strp_rx_msg strp;
42 static inline struct _strp_rx_msg *_strp_rx_msg(struct sk_buff *skb)
44 return (struct _strp_rx_msg *)((void *)skb->cb +
45 offsetof(struct qdisc_skb_cb, data));
49 static void strp_abort_rx_strp(struct strparser *strp, int err)
51 struct sock *csk = strp->sk;
53 /* Unrecoverable error in receive */
55 del_timer(&strp->rx_msg_timer);
62 /* Report an error on the lower socket */
64 csk->sk_error_report(csk);
67 static void strp_start_rx_timer(struct strparser *strp)
69 if (strp->sk->sk_rcvtimeo)
70 mod_timer(&strp->rx_msg_timer, strp->sk->sk_rcvtimeo);
74 static void strp_parser_err(struct strparser *strp, int err,
75 read_descriptor_t *desc)
78 kfree_skb(strp->rx_skb_head);
79 strp->rx_skb_head = NULL;
80 strp->cb.abort_parser(strp, err);
83 /* Lower socket lock held */
84 static int strp_tcp_recv(read_descriptor_t *desc, struct sk_buff *orig_skb,
85 unsigned int orig_offset, size_t orig_len)
87 struct strparser *strp = (struct strparser *)desc->arg.data;
88 struct _strp_rx_msg *rxm;
89 struct sk_buff *head, *skb;
90 size_t eaten = 0, cand_len;
93 bool cloned_orig = false;
98 head = strp->rx_skb_head;
100 /* Message already in progress */
102 rxm = _strp_rx_msg(head);
103 if (unlikely(rxm->early_eaten)) {
104 /* Already some number of bytes on the receive sock
105 * data saved in rx_skb_head, just indicate they
108 eaten = orig_len <= rxm->early_eaten ?
109 orig_len : rxm->early_eaten;
110 rxm->early_eaten -= eaten;
115 if (unlikely(orig_offset)) {
116 /* Getting data with a non-zero offset when a message is
117 * in progress is not expected. If it does happen, we
118 * need to clone and pull since we can't deal with
119 * offsets in the skbs for a message expect in the head.
121 orig_skb = skb_clone(orig_skb, GFP_ATOMIC);
123 STRP_STATS_INCR(strp->stats.rx_mem_fail);
124 desc->error = -ENOMEM;
127 if (!pskb_pull(orig_skb, orig_offset)) {
128 STRP_STATS_INCR(strp->stats.rx_mem_fail);
130 desc->error = -ENOMEM;
137 if (!strp->rx_skb_nextp) {
138 /* We are going to append to the frags_list of head.
139 * Need to unshare the frag_list.
141 err = skb_unclone(head, GFP_ATOMIC);
143 STRP_STATS_INCR(strp->stats.rx_mem_fail);
148 if (unlikely(skb_shinfo(head)->frag_list)) {
149 /* We can't append to an sk_buff that already
150 * has a frag_list. We create a new head, point
151 * the frag_list of that to the old head, and
152 * then are able to use the old head->next for
153 * appending to the message.
155 if (WARN_ON(head->next)) {
156 desc->error = -EINVAL;
160 skb = alloc_skb(0, GFP_ATOMIC);
162 STRP_STATS_INCR(strp->stats.rx_mem_fail);
163 desc->error = -ENOMEM;
166 skb->len = head->len;
167 skb->data_len = head->len;
168 skb->truesize = head->truesize;
169 *_strp_rx_msg(skb) = *_strp_rx_msg(head);
170 strp->rx_skb_nextp = &head->next;
171 skb_shinfo(skb)->frag_list = head;
172 strp->rx_skb_head = skb;
176 &skb_shinfo(head)->frag_list;
181 while (eaten < orig_len) {
182 /* Always clone since we will consume something */
183 skb = skb_clone(orig_skb, GFP_ATOMIC);
185 STRP_STATS_INCR(strp->stats.rx_mem_fail);
186 desc->error = -ENOMEM;
190 cand_len = orig_len - eaten;
192 head = strp->rx_skb_head;
195 strp->rx_skb_head = head;
196 /* Will set rx_skb_nextp on next packet if needed */
197 strp->rx_skb_nextp = NULL;
198 rxm = _strp_rx_msg(head);
199 memset(rxm, 0, sizeof(*rxm));
200 rxm->strp.offset = orig_offset + eaten;
202 /* Unclone since we may be appending to an skb that we
203 * already share a frag_list with.
205 err = skb_unclone(skb, GFP_ATOMIC);
207 STRP_STATS_INCR(strp->stats.rx_mem_fail);
212 rxm = _strp_rx_msg(head);
213 *strp->rx_skb_nextp = skb;
214 strp->rx_skb_nextp = &skb->next;
215 head->data_len += skb->len;
216 head->len += skb->len;
217 head->truesize += skb->truesize;
220 if (!rxm->strp.full_len) {
223 len = (*strp->cb.parse_msg)(strp, head);
226 /* Need more header to determine length */
227 if (!rxm->accum_len) {
228 /* Start RX timer for new message */
229 strp_start_rx_timer(strp);
231 rxm->accum_len += cand_len;
233 STRP_STATS_INCR(strp->stats.rx_need_more_hdr);
234 WARN_ON(eaten != orig_len);
236 } else if (len < 0) {
237 if (len == -ESTRPIPE && rxm->accum_len) {
239 strp->rx_unrecov_intr = 1;
241 strp->rx_interrupted = 1;
243 strp_parser_err(strp, err, desc);
245 } else if (len > strp->sk->sk_rcvbuf) {
246 /* Message length exceeds maximum allowed */
247 STRP_STATS_INCR(strp->stats.rx_msg_too_big);
248 strp_parser_err(strp, -EMSGSIZE, desc);
250 } else if (len <= (ssize_t)head->len -
251 skb->len - rxm->strp.offset) {
252 /* Length must be into new skb (and also
255 STRP_STATS_INCR(strp->stats.rx_bad_hdr_len);
256 strp_parser_err(strp, -EPROTO, desc);
260 rxm->strp.full_len = len;
263 extra = (ssize_t)(rxm->accum_len + cand_len) -
267 /* Message not complete yet. */
268 if (rxm->strp.full_len - rxm->accum_len >
270 /* Don't have the whole messages in the socket
271 * buffer. Set strp->rx_need_bytes to wait for
272 * the rest of the message. Also, set "early
273 * eaten" since we've already buffered the skb
274 * but don't consume yet per tcp_read_sock.
277 if (!rxm->accum_len) {
278 /* Start RX timer for new message */
279 strp_start_rx_timer(strp);
282 strp->rx_need_bytes = rxm->strp.full_len -
284 rxm->accum_len += cand_len;
285 rxm->early_eaten = cand_len;
286 STRP_STATS_ADD(strp->stats.rx_bytes, cand_len);
287 desc->count = 0; /* Stop reading socket */
290 rxm->accum_len += cand_len;
292 WARN_ON(eaten != orig_len);
296 /* Positive extra indicates ore bytes than needed for the
300 WARN_ON(extra > cand_len);
302 eaten += (cand_len - extra);
304 /* Hurray, we have a new message! */
305 del_timer(&strp->rx_msg_timer);
306 strp->rx_skb_head = NULL;
307 STRP_STATS_INCR(strp->stats.rx_msgs);
309 /* Give skb to upper layer */
310 strp->cb.rcv_msg(strp, head);
312 if (unlikely(strp->rx_paused)) {
313 /* Upper layer paused strp */
321 STRP_STATS_ADD(strp->stats.rx_bytes, eaten);
326 static int default_read_sock_done(struct strparser *strp, int err)
331 /* Called with lock held on lower socket */
332 static int strp_tcp_read_sock(struct strparser *strp)
334 read_descriptor_t desc;
336 desc.arg.data = strp;
338 desc.count = 1; /* give more than one skb per call */
340 /* sk should be locked here, so okay to do tcp_read_sock */
341 tcp_read_sock(strp->sk, &desc, strp_tcp_recv);
343 desc.error = strp->cb.read_sock_done(strp, desc.error);
348 /* Lower sock lock held */
349 void strp_tcp_data_ready(struct strparser *strp)
351 struct sock *csk = strp->sk;
353 if (unlikely(strp->rx_stopped))
356 /* This check is needed to synchronize with do_strp_rx_work.
357 * do_strp_rx_work acquires a process lock (lock_sock) whereas
358 * the lock held here is bh_lock_sock. The two locks can be
359 * held by different threads at the same time, but bh_lock_sock
360 * allows a thread in BH context to safely check if the process
361 * lock is held. In this case, if the lock is held, queue work.
363 if (sock_owned_by_user(csk)) {
364 queue_work(strp_wq, &strp->rx_work);
371 if (strp->rx_need_bytes) {
372 if (tcp_inq(csk) >= strp->rx_need_bytes)
373 strp->rx_need_bytes = 0;
378 if (strp_tcp_read_sock(strp) == -ENOMEM)
379 queue_work(strp_wq, &strp->rx_work);
381 EXPORT_SYMBOL_GPL(strp_tcp_data_ready);
383 static void do_strp_rx_work(struct strparser *strp)
385 read_descriptor_t rd_desc;
386 struct sock *csk = strp->sk;
388 /* We need the read lock to synchronize with strp_tcp_data_ready. We
389 * need the socket lock for calling tcp_read_sock.
393 if (unlikely(csk->sk_user_data != strp))
396 if (unlikely(strp->rx_stopped))
402 rd_desc.arg.data = strp;
404 if (strp_tcp_read_sock(strp) == -ENOMEM)
405 queue_work(strp_wq, &strp->rx_work);
411 static void strp_rx_work(struct work_struct *w)
413 do_strp_rx_work(container_of(w, struct strparser, rx_work));
416 static void strp_rx_msg_timeout(unsigned long arg)
418 struct strparser *strp = (struct strparser *)arg;
420 /* Message assembly timed out */
421 STRP_STATS_INCR(strp->stats.rx_msg_timeouts);
423 strp->cb.abort_parser(strp, ETIMEDOUT);
424 release_sock(strp->sk);
427 int strp_init(struct strparser *strp, struct sock *csk,
428 struct strp_callbacks *cb)
430 if (!cb || !cb->rcv_msg || !cb->parse_msg)
433 memset(strp, 0, sizeof(*strp));
437 setup_timer(&strp->rx_msg_timer, strp_rx_msg_timeout,
438 (unsigned long)strp);
440 INIT_WORK(&strp->rx_work, strp_rx_work);
442 strp->cb.rcv_msg = cb->rcv_msg;
443 strp->cb.parse_msg = cb->parse_msg;
444 strp->cb.read_sock_done = cb->read_sock_done ? : default_read_sock_done;
445 strp->cb.abort_parser = cb->abort_parser ? : strp_abort_rx_strp;
449 EXPORT_SYMBOL_GPL(strp_init);
451 /* strp must already be stopped so that strp_tcp_recv will no longer be called.
452 * Note that strp_done is not called with the lower socket held.
454 void strp_done(struct strparser *strp)
456 WARN_ON(!strp->rx_stopped);
458 del_timer_sync(&strp->rx_msg_timer);
459 cancel_work_sync(&strp->rx_work);
461 if (strp->rx_skb_head) {
462 kfree_skb(strp->rx_skb_head);
463 strp->rx_skb_head = NULL;
466 EXPORT_SYMBOL_GPL(strp_done);
468 void strp_stop(struct strparser *strp)
470 strp->rx_stopped = 1;
472 EXPORT_SYMBOL_GPL(strp_stop);
474 void strp_check_rcv(struct strparser *strp)
476 queue_work(strp_wq, &strp->rx_work);
478 EXPORT_SYMBOL_GPL(strp_check_rcv);
480 static int __init strp_mod_init(void)
482 strp_wq = create_singlethread_workqueue("kstrp");
487 static void __exit strp_mod_exit(void)
490 module_init(strp_mod_init);
491 module_exit(strp_mod_exit);
492 MODULE_LICENSE("GPL");