Commit | Line | Data |
---|---|---|
2f713615 ID |
1 | // SPDX-License-Identifier: GPL-2.0 |
2 | #include <linux/ceph/ceph_debug.h> | |
3 | ||
4 | #include <linux/bvec.h> | |
5 | #include <linux/crc32c.h> | |
6 | #include <linux/net.h> | |
7 | #include <linux/socket.h> | |
8 | #include <net/sock.h> | |
9 | ||
10 | #include <linux/ceph/ceph_features.h> | |
11 | #include <linux/ceph/decode.h> | |
12 | #include <linux/ceph/libceph.h> | |
13 | #include <linux/ceph/messenger.h> | |
14 | ||
15 | /* static tag bytes (protocol control messages) */ | |
16 | static char tag_msg = CEPH_MSGR_TAG_MSG; | |
17 | static char tag_ack = CEPH_MSGR_TAG_ACK; | |
18 | static char tag_keepalive = CEPH_MSGR_TAG_KEEPALIVE; | |
19 | static char tag_keepalive2 = CEPH_MSGR_TAG_KEEPALIVE2; | |
20 | ||
21 | /* | |
22 | * If @buf is NULL, discard up to @len bytes. | |
23 | */ | |
24 | static int ceph_tcp_recvmsg(struct socket *sock, void *buf, size_t len) | |
25 | { | |
26 | struct kvec iov = {buf, len}; | |
27 | struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL }; | |
28 | int r; | |
29 | ||
30 | if (!buf) | |
31 | msg.msg_flags |= MSG_TRUNC; | |
32 | ||
de4eda9d | 33 | iov_iter_kvec(&msg.msg_iter, ITER_DEST, &iov, 1, len); |
2f713615 ID |
34 | r = sock_recvmsg(sock, &msg, msg.msg_flags); |
35 | if (r == -EAGAIN) | |
36 | r = 0; | |
37 | return r; | |
38 | } | |
39 | ||
40 | static int ceph_tcp_recvpage(struct socket *sock, struct page *page, | |
41 | int page_offset, size_t length) | |
42 | { | |
1eb9cd15 | 43 | struct bio_vec bvec; |
2f713615 ID |
44 | struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL }; |
45 | int r; | |
46 | ||
47 | BUG_ON(page_offset + length > PAGE_SIZE); | |
1eb9cd15 | 48 | bvec_set_page(&bvec, page, length, page_offset); |
de4eda9d | 49 | iov_iter_bvec(&msg.msg_iter, ITER_DEST, &bvec, 1, length); |
2f713615 ID |
50 | r = sock_recvmsg(sock, &msg, msg.msg_flags); |
51 | if (r == -EAGAIN) | |
52 | r = 0; | |
53 | return r; | |
54 | } | |
55 | ||
56 | /* | |
57 | * write something. @more is true if caller will be sending more data | |
58 | * shortly. | |
59 | */ | |
60 | static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov, | |
61 | size_t kvlen, size_t len, bool more) | |
62 | { | |
63 | struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL }; | |
64 | int r; | |
65 | ||
66 | if (more) | |
67 | msg.msg_flags |= MSG_MORE; | |
68 | else | |
69 | msg.msg_flags |= MSG_EOR; /* superfluous, but what the hell */ | |
70 | ||
71 | r = kernel_sendmsg(sock, &msg, iov, kvlen, len); | |
72 | if (r == -EAGAIN) | |
73 | r = 0; | |
74 | return r; | |
75 | } | |
76 | ||
77 | /* | |
78 | * @more: either or both of MSG_MORE and MSG_SENDPAGE_NOTLAST | |
79 | */ | |
80 | static int ceph_tcp_sendpage(struct socket *sock, struct page *page, | |
81 | int offset, size_t size, int more) | |
82 | { | |
83 | ssize_t (*sendpage)(struct socket *sock, struct page *page, | |
84 | int offset, size_t size, int flags); | |
85 | int flags = MSG_DONTWAIT | MSG_NOSIGNAL | more; | |
86 | int ret; | |
87 | ||
88 | /* | |
89 | * sendpage cannot properly handle pages with page_count == 0, | |
90 | * we need to fall back to sendmsg if that's the case. | |
91 | * | |
92 | * Same goes for slab pages: skb_can_coalesce() allows | |
93 | * coalescing neighboring slab objects into a single frag which | |
94 | * triggers one of hardened usercopy checks. | |
95 | */ | |
96 | if (sendpage_ok(page)) | |
97 | sendpage = sock->ops->sendpage; | |
98 | else | |
99 | sendpage = sock_no_sendpage; | |
100 | ||
101 | ret = sendpage(sock, page, offset, size, flags); | |
102 | if (ret == -EAGAIN) | |
103 | ret = 0; | |
104 | ||
105 | return ret; | |
106 | } | |
107 | ||
108 | static void con_out_kvec_reset(struct ceph_connection *con) | |
109 | { | |
a56dd9bf | 110 | BUG_ON(con->v1.out_skip); |
2f713615 | 111 | |
a56dd9bf ID |
112 | con->v1.out_kvec_left = 0; |
113 | con->v1.out_kvec_bytes = 0; | |
114 | con->v1.out_kvec_cur = &con->v1.out_kvec[0]; | |
2f713615 ID |
115 | } |
116 | ||
117 | static void con_out_kvec_add(struct ceph_connection *con, | |
118 | size_t size, void *data) | |
119 | { | |
a56dd9bf | 120 | int index = con->v1.out_kvec_left; |
2f713615 | 121 | |
a56dd9bf ID |
122 | BUG_ON(con->v1.out_skip); |
123 | BUG_ON(index >= ARRAY_SIZE(con->v1.out_kvec)); | |
2f713615 | 124 | |
a56dd9bf ID |
125 | con->v1.out_kvec[index].iov_len = size; |
126 | con->v1.out_kvec[index].iov_base = data; | |
127 | con->v1.out_kvec_left++; | |
128 | con->v1.out_kvec_bytes += size; | |
2f713615 ID |
129 | } |
130 | ||
131 | /* | |
132 | * Chop off a kvec from the end. Return residual number of bytes for | |
133 | * that kvec, i.e. how many bytes would have been written if the kvec | |
134 | * hadn't been nuked. | |
135 | */ | |
136 | static int con_out_kvec_skip(struct ceph_connection *con) | |
137 | { | |
2f713615 ID |
138 | int skip = 0; |
139 | ||
a56dd9bf ID |
140 | if (con->v1.out_kvec_bytes > 0) { |
141 | skip = con->v1.out_kvec_cur[con->v1.out_kvec_left - 1].iov_len; | |
142 | BUG_ON(con->v1.out_kvec_bytes < skip); | |
143 | BUG_ON(!con->v1.out_kvec_left); | |
144 | con->v1.out_kvec_bytes -= skip; | |
145 | con->v1.out_kvec_left--; | |
2f713615 ID |
146 | } |
147 | ||
148 | return skip; | |
149 | } | |
150 | ||
151 | static size_t sizeof_footer(struct ceph_connection *con) | |
152 | { | |
153 | return (con->peer_features & CEPH_FEATURE_MSG_AUTH) ? | |
154 | sizeof(struct ceph_msg_footer) : | |
155 | sizeof(struct ceph_msg_footer_old); | |
156 | } | |
157 | ||
158 | static void prepare_message_data(struct ceph_msg *msg, u32 data_len) | |
159 | { | |
160 | /* Initialize data cursor */ | |
161 | ||
162 | ceph_msg_data_cursor_init(&msg->cursor, msg, data_len); | |
163 | } | |
164 | ||
165 | /* | |
166 | * Prepare footer for currently outgoing message, and finish things | |
167 | * off. Assumes out_kvec* are already valid.. we just add on to the end. | |
168 | */ | |
169 | static void prepare_write_message_footer(struct ceph_connection *con) | |
170 | { | |
171 | struct ceph_msg *m = con->out_msg; | |
172 | ||
173 | m->footer.flags |= CEPH_MSG_FOOTER_COMPLETE; | |
174 | ||
175 | dout("prepare_write_message_footer %p\n", con); | |
176 | con_out_kvec_add(con, sizeof_footer(con), &m->footer); | |
177 | if (con->peer_features & CEPH_FEATURE_MSG_AUTH) { | |
178 | if (con->ops->sign_message) | |
179 | con->ops->sign_message(m); | |
180 | else | |
181 | m->footer.sig = 0; | |
182 | } else { | |
183 | m->old_footer.flags = m->footer.flags; | |
184 | } | |
a56dd9bf ID |
185 | con->v1.out_more = m->more_to_follow; |
186 | con->v1.out_msg_done = true; | |
2f713615 ID |
187 | } |
188 | ||
189 | /* | |
190 | * Prepare headers for the next outgoing message. | |
191 | */ | |
192 | static void prepare_write_message(struct ceph_connection *con) | |
193 | { | |
194 | struct ceph_msg *m; | |
195 | u32 crc; | |
196 | ||
197 | con_out_kvec_reset(con); | |
a56dd9bf | 198 | con->v1.out_msg_done = false; |
2f713615 ID |
199 | |
200 | /* Sneak an ack in there first? If we can get it into the same | |
201 | * TCP packet that's a good thing. */ | |
202 | if (con->in_seq > con->in_seq_acked) { | |
203 | con->in_seq_acked = con->in_seq; | |
204 | con_out_kvec_add(con, sizeof (tag_ack), &tag_ack); | |
a56dd9bf ID |
205 | con->v1.out_temp_ack = cpu_to_le64(con->in_seq_acked); |
206 | con_out_kvec_add(con, sizeof(con->v1.out_temp_ack), | |
207 | &con->v1.out_temp_ack); | |
2f713615 ID |
208 | } |
209 | ||
210 | ceph_con_get_out_msg(con); | |
211 | m = con->out_msg; | |
212 | ||
213 | dout("prepare_write_message %p seq %lld type %d len %d+%d+%zd\n", | |
214 | m, con->out_seq, le16_to_cpu(m->hdr.type), | |
215 | le32_to_cpu(m->hdr.front_len), le32_to_cpu(m->hdr.middle_len), | |
216 | m->data_length); | |
217 | WARN_ON(m->front.iov_len != le32_to_cpu(m->hdr.front_len)); | |
218 | WARN_ON(m->data_length != le32_to_cpu(m->hdr.data_len)); | |
219 | ||
220 | /* tag + hdr + front + middle */ | |
221 | con_out_kvec_add(con, sizeof (tag_msg), &tag_msg); | |
a56dd9bf | 222 | con_out_kvec_add(con, sizeof(con->v1.out_hdr), &con->v1.out_hdr); |
2f713615 ID |
223 | con_out_kvec_add(con, m->front.iov_len, m->front.iov_base); |
224 | ||
225 | if (m->middle) | |
226 | con_out_kvec_add(con, m->middle->vec.iov_len, | |
227 | m->middle->vec.iov_base); | |
228 | ||
229 | /* fill in hdr crc and finalize hdr */ | |
230 | crc = crc32c(0, &m->hdr, offsetof(struct ceph_msg_header, crc)); | |
231 | con->out_msg->hdr.crc = cpu_to_le32(crc); | |
a56dd9bf | 232 | memcpy(&con->v1.out_hdr, &con->out_msg->hdr, sizeof(con->v1.out_hdr)); |
2f713615 ID |
233 | |
234 | /* fill in front and middle crc, footer */ | |
235 | crc = crc32c(0, m->front.iov_base, m->front.iov_len); | |
236 | con->out_msg->footer.front_crc = cpu_to_le32(crc); | |
237 | if (m->middle) { | |
238 | crc = crc32c(0, m->middle->vec.iov_base, | |
239 | m->middle->vec.iov_len); | |
240 | con->out_msg->footer.middle_crc = cpu_to_le32(crc); | |
241 | } else | |
242 | con->out_msg->footer.middle_crc = 0; | |
243 | dout("%s front_crc %u middle_crc %u\n", __func__, | |
244 | le32_to_cpu(con->out_msg->footer.front_crc), | |
245 | le32_to_cpu(con->out_msg->footer.middle_crc)); | |
246 | con->out_msg->footer.flags = 0; | |
247 | ||
248 | /* is there a data payload? */ | |
249 | con->out_msg->footer.data_crc = 0; | |
250 | if (m->data_length) { | |
251 | prepare_message_data(con->out_msg, m->data_length); | |
a56dd9bf | 252 | con->v1.out_more = 1; /* data + footer will follow */ |
2f713615 ID |
253 | } else { |
254 | /* no, queue up footer too and be done */ | |
255 | prepare_write_message_footer(con); | |
256 | } | |
257 | ||
258 | ceph_con_flag_set(con, CEPH_CON_F_WRITE_PENDING); | |
259 | } | |
260 | ||
261 | /* | |
262 | * Prepare an ack. | |
263 | */ | |
264 | static void prepare_write_ack(struct ceph_connection *con) | |
265 | { | |
266 | dout("prepare_write_ack %p %llu -> %llu\n", con, | |
267 | con->in_seq_acked, con->in_seq); | |
268 | con->in_seq_acked = con->in_seq; | |
269 | ||
270 | con_out_kvec_reset(con); | |
271 | ||
272 | con_out_kvec_add(con, sizeof (tag_ack), &tag_ack); | |
273 | ||
a56dd9bf ID |
274 | con->v1.out_temp_ack = cpu_to_le64(con->in_seq_acked); |
275 | con_out_kvec_add(con, sizeof(con->v1.out_temp_ack), | |
276 | &con->v1.out_temp_ack); | |
2f713615 | 277 | |
a56dd9bf | 278 | con->v1.out_more = 1; /* more will follow.. eventually.. */ |
2f713615 ID |
279 | ceph_con_flag_set(con, CEPH_CON_F_WRITE_PENDING); |
280 | } | |
281 | ||
282 | /* | |
283 | * Prepare to share the seq during handshake | |
284 | */ | |
285 | static void prepare_write_seq(struct ceph_connection *con) | |
286 | { | |
287 | dout("prepare_write_seq %p %llu -> %llu\n", con, | |
288 | con->in_seq_acked, con->in_seq); | |
289 | con->in_seq_acked = con->in_seq; | |
290 | ||
291 | con_out_kvec_reset(con); | |
292 | ||
a56dd9bf ID |
293 | con->v1.out_temp_ack = cpu_to_le64(con->in_seq_acked); |
294 | con_out_kvec_add(con, sizeof(con->v1.out_temp_ack), | |
295 | &con->v1.out_temp_ack); | |
2f713615 ID |
296 | |
297 | ceph_con_flag_set(con, CEPH_CON_F_WRITE_PENDING); | |
298 | } | |
299 | ||
300 | /* | |
301 | * Prepare to write keepalive byte. | |
302 | */ | |
303 | static void prepare_write_keepalive(struct ceph_connection *con) | |
304 | { | |
305 | dout("prepare_write_keepalive %p\n", con); | |
306 | con_out_kvec_reset(con); | |
307 | if (con->peer_features & CEPH_FEATURE_MSGR_KEEPALIVE2) { | |
308 | struct timespec64 now; | |
309 | ||
310 | ktime_get_real_ts64(&now); | |
311 | con_out_kvec_add(con, sizeof(tag_keepalive2), &tag_keepalive2); | |
a56dd9bf ID |
312 | ceph_encode_timespec64(&con->v1.out_temp_keepalive2, &now); |
313 | con_out_kvec_add(con, sizeof(con->v1.out_temp_keepalive2), | |
314 | &con->v1.out_temp_keepalive2); | |
2f713615 ID |
315 | } else { |
316 | con_out_kvec_add(con, sizeof(tag_keepalive), &tag_keepalive); | |
317 | } | |
318 | ceph_con_flag_set(con, CEPH_CON_F_WRITE_PENDING); | |
319 | } | |
320 | ||
321 | /* | |
322 | * Connection negotiation. | |
323 | */ | |
324 | ||
325 | static int get_connect_authorizer(struct ceph_connection *con) | |
326 | { | |
327 | struct ceph_auth_handshake *auth; | |
328 | int auth_proto; | |
329 | ||
330 | if (!con->ops->get_authorizer) { | |
a56dd9bf ID |
331 | con->v1.auth = NULL; |
332 | con->v1.out_connect.authorizer_protocol = CEPH_AUTH_UNKNOWN; | |
333 | con->v1.out_connect.authorizer_len = 0; | |
2f713615 ID |
334 | return 0; |
335 | } | |
336 | ||
a56dd9bf | 337 | auth = con->ops->get_authorizer(con, &auth_proto, con->v1.auth_retry); |
2f713615 ID |
338 | if (IS_ERR(auth)) |
339 | return PTR_ERR(auth); | |
340 | ||
a56dd9bf ID |
341 | con->v1.auth = auth; |
342 | con->v1.out_connect.authorizer_protocol = cpu_to_le32(auth_proto); | |
343 | con->v1.out_connect.authorizer_len = | |
344 | cpu_to_le32(auth->authorizer_buf_len); | |
2f713615 ID |
345 | return 0; |
346 | } | |
347 | ||
348 | /* | |
349 | * We connected to a peer and are saying hello. | |
350 | */ | |
351 | static void prepare_write_banner(struct ceph_connection *con) | |
352 | { | |
353 | con_out_kvec_add(con, strlen(CEPH_BANNER), CEPH_BANNER); | |
354 | con_out_kvec_add(con, sizeof (con->msgr->my_enc_addr), | |
355 | &con->msgr->my_enc_addr); | |
356 | ||
a56dd9bf | 357 | con->v1.out_more = 0; |
2f713615 ID |
358 | ceph_con_flag_set(con, CEPH_CON_F_WRITE_PENDING); |
359 | } | |
360 | ||
361 | static void __prepare_write_connect(struct ceph_connection *con) | |
362 | { | |
a56dd9bf ID |
363 | con_out_kvec_add(con, sizeof(con->v1.out_connect), |
364 | &con->v1.out_connect); | |
365 | if (con->v1.auth) | |
366 | con_out_kvec_add(con, con->v1.auth->authorizer_buf_len, | |
367 | con->v1.auth->authorizer_buf); | |
2f713615 | 368 | |
a56dd9bf | 369 | con->v1.out_more = 0; |
2f713615 ID |
370 | ceph_con_flag_set(con, CEPH_CON_F_WRITE_PENDING); |
371 | } | |
372 | ||
373 | static int prepare_write_connect(struct ceph_connection *con) | |
374 | { | |
375 | unsigned int global_seq = ceph_get_global_seq(con->msgr, 0); | |
376 | int proto; | |
377 | int ret; | |
378 | ||
379 | switch (con->peer_name.type) { | |
380 | case CEPH_ENTITY_TYPE_MON: | |
381 | proto = CEPH_MONC_PROTOCOL; | |
382 | break; | |
383 | case CEPH_ENTITY_TYPE_OSD: | |
384 | proto = CEPH_OSDC_PROTOCOL; | |
385 | break; | |
386 | case CEPH_ENTITY_TYPE_MDS: | |
387 | proto = CEPH_MDSC_PROTOCOL; | |
388 | break; | |
389 | default: | |
390 | BUG(); | |
391 | } | |
392 | ||
393 | dout("prepare_write_connect %p cseq=%d gseq=%d proto=%d\n", con, | |
a56dd9bf | 394 | con->v1.connect_seq, global_seq, proto); |
2f713615 | 395 | |
a56dd9bf ID |
396 | con->v1.out_connect.features = |
397 | cpu_to_le64(from_msgr(con->msgr)->supported_features); | |
398 | con->v1.out_connect.host_type = cpu_to_le32(CEPH_ENTITY_TYPE_CLIENT); | |
399 | con->v1.out_connect.connect_seq = cpu_to_le32(con->v1.connect_seq); | |
400 | con->v1.out_connect.global_seq = cpu_to_le32(global_seq); | |
401 | con->v1.out_connect.protocol_version = cpu_to_le32(proto); | |
402 | con->v1.out_connect.flags = 0; | |
2f713615 ID |
403 | |
404 | ret = get_connect_authorizer(con); | |
405 | if (ret) | |
406 | return ret; | |
407 | ||
408 | __prepare_write_connect(con); | |
409 | return 0; | |
410 | } | |
411 | ||
412 | /* | |
413 | * write as much of pending kvecs to the socket as we can. | |
414 | * 1 -> done | |
415 | * 0 -> socket full, but more to do | |
416 | * <0 -> error | |
417 | */ | |
418 | static int write_partial_kvec(struct ceph_connection *con) | |
419 | { | |
420 | int ret; | |
421 | ||
a56dd9bf ID |
422 | dout("write_partial_kvec %p %d left\n", con, con->v1.out_kvec_bytes); |
423 | while (con->v1.out_kvec_bytes > 0) { | |
424 | ret = ceph_tcp_sendmsg(con->sock, con->v1.out_kvec_cur, | |
425 | con->v1.out_kvec_left, | |
426 | con->v1.out_kvec_bytes, | |
427 | con->v1.out_more); | |
2f713615 ID |
428 | if (ret <= 0) |
429 | goto out; | |
a56dd9bf ID |
430 | con->v1.out_kvec_bytes -= ret; |
431 | if (!con->v1.out_kvec_bytes) | |
2f713615 ID |
432 | break; /* done */ |
433 | ||
434 | /* account for full iov entries consumed */ | |
a56dd9bf ID |
435 | while (ret >= con->v1.out_kvec_cur->iov_len) { |
436 | BUG_ON(!con->v1.out_kvec_left); | |
437 | ret -= con->v1.out_kvec_cur->iov_len; | |
438 | con->v1.out_kvec_cur++; | |
439 | con->v1.out_kvec_left--; | |
2f713615 ID |
440 | } |
441 | /* and for a partially-consumed entry */ | |
442 | if (ret) { | |
a56dd9bf ID |
443 | con->v1.out_kvec_cur->iov_len -= ret; |
444 | con->v1.out_kvec_cur->iov_base += ret; | |
2f713615 ID |
445 | } |
446 | } | |
a56dd9bf | 447 | con->v1.out_kvec_left = 0; |
2f713615 ID |
448 | ret = 1; |
449 | out: | |
450 | dout("write_partial_kvec %p %d left in %d kvecs ret = %d\n", con, | |
a56dd9bf | 451 | con->v1.out_kvec_bytes, con->v1.out_kvec_left, ret); |
2f713615 ID |
452 | return ret; /* done! */ |
453 | } | |
454 | ||
455 | /* | |
456 | * Write as much message data payload as we can. If we finish, queue | |
457 | * up the footer. | |
458 | * 1 -> done, footer is now queued in out_kvec[]. | |
459 | * 0 -> socket full, but more to do | |
460 | * <0 -> error | |
461 | */ | |
462 | static int write_partial_message_data(struct ceph_connection *con) | |
463 | { | |
464 | struct ceph_msg *msg = con->out_msg; | |
465 | struct ceph_msg_data_cursor *cursor = &msg->cursor; | |
466 | bool do_datacrc = !ceph_test_opt(from_msgr(con->msgr), NOCRC); | |
467 | int more = MSG_MORE | MSG_SENDPAGE_NOTLAST; | |
468 | u32 crc; | |
469 | ||
470 | dout("%s %p msg %p\n", __func__, con, msg); | |
471 | ||
472 | if (!msg->num_data_items) | |
473 | return -EINVAL; | |
474 | ||
475 | /* | |
476 | * Iterate through each page that contains data to be | |
477 | * written, and send as much as possible for each. | |
478 | * | |
479 | * If we are calculating the data crc (the default), we will | |
480 | * need to map the page. If we have no pages, they have | |
481 | * been revoked, so use the zero page. | |
482 | */ | |
483 | crc = do_datacrc ? le32_to_cpu(msg->footer.data_crc) : 0; | |
484 | while (cursor->total_resid) { | |
485 | struct page *page; | |
486 | size_t page_offset; | |
487 | size_t length; | |
488 | int ret; | |
489 | ||
490 | if (!cursor->resid) { | |
491 | ceph_msg_data_advance(cursor, 0); | |
492 | continue; | |
493 | } | |
494 | ||
da4ab869 | 495 | page = ceph_msg_data_next(cursor, &page_offset, &length); |
2f713615 ID |
496 | if (length == cursor->total_resid) |
497 | more = MSG_MORE; | |
498 | ret = ceph_tcp_sendpage(con->sock, page, page_offset, length, | |
499 | more); | |
500 | if (ret <= 0) { | |
501 | if (do_datacrc) | |
502 | msg->footer.data_crc = cpu_to_le32(crc); | |
503 | ||
504 | return ret; | |
505 | } | |
506 | if (do_datacrc && cursor->need_crc) | |
507 | crc = ceph_crc32c_page(crc, page, page_offset, length); | |
508 | ceph_msg_data_advance(cursor, (size_t)ret); | |
509 | } | |
510 | ||
511 | dout("%s %p msg %p done\n", __func__, con, msg); | |
512 | ||
513 | /* prepare and queue up footer, too */ | |
514 | if (do_datacrc) | |
515 | msg->footer.data_crc = cpu_to_le32(crc); | |
516 | else | |
517 | msg->footer.flags |= CEPH_MSG_FOOTER_NOCRC; | |
518 | con_out_kvec_reset(con); | |
519 | prepare_write_message_footer(con); | |
520 | ||
521 | return 1; /* must return > 0 to indicate success */ | |
522 | } | |
523 | ||
524 | /* | |
525 | * write some zeros | |
526 | */ | |
527 | static int write_partial_skip(struct ceph_connection *con) | |
528 | { | |
529 | int more = MSG_MORE | MSG_SENDPAGE_NOTLAST; | |
530 | int ret; | |
531 | ||
a56dd9bf ID |
532 | dout("%s %p %d left\n", __func__, con, con->v1.out_skip); |
533 | while (con->v1.out_skip > 0) { | |
534 | size_t size = min(con->v1.out_skip, (int)PAGE_SIZE); | |
2f713615 | 535 | |
a56dd9bf | 536 | if (size == con->v1.out_skip) |
2f713615 ID |
537 | more = MSG_MORE; |
538 | ret = ceph_tcp_sendpage(con->sock, ceph_zero_page, 0, size, | |
539 | more); | |
540 | if (ret <= 0) | |
541 | goto out; | |
a56dd9bf | 542 | con->v1.out_skip -= ret; |
2f713615 ID |
543 | } |
544 | ret = 1; | |
545 | out: | |
546 | return ret; | |
547 | } | |
548 | ||
549 | /* | |
550 | * Prepare to read connection handshake, or an ack. | |
551 | */ | |
552 | static void prepare_read_banner(struct ceph_connection *con) | |
553 | { | |
554 | dout("prepare_read_banner %p\n", con); | |
a56dd9bf | 555 | con->v1.in_base_pos = 0; |
2f713615 ID |
556 | } |
557 | ||
558 | static void prepare_read_connect(struct ceph_connection *con) | |
559 | { | |
560 | dout("prepare_read_connect %p\n", con); | |
a56dd9bf | 561 | con->v1.in_base_pos = 0; |
2f713615 ID |
562 | } |
563 | ||
564 | static void prepare_read_ack(struct ceph_connection *con) | |
565 | { | |
566 | dout("prepare_read_ack %p\n", con); | |
a56dd9bf | 567 | con->v1.in_base_pos = 0; |
2f713615 ID |
568 | } |
569 | ||
570 | static void prepare_read_seq(struct ceph_connection *con) | |
571 | { | |
572 | dout("prepare_read_seq %p\n", con); | |
a56dd9bf ID |
573 | con->v1.in_base_pos = 0; |
574 | con->v1.in_tag = CEPH_MSGR_TAG_SEQ; | |
2f713615 ID |
575 | } |
576 | ||
577 | static void prepare_read_tag(struct ceph_connection *con) | |
578 | { | |
579 | dout("prepare_read_tag %p\n", con); | |
a56dd9bf ID |
580 | con->v1.in_base_pos = 0; |
581 | con->v1.in_tag = CEPH_MSGR_TAG_READY; | |
2f713615 ID |
582 | } |
583 | ||
584 | static void prepare_read_keepalive_ack(struct ceph_connection *con) | |
585 | { | |
586 | dout("prepare_read_keepalive_ack %p\n", con); | |
a56dd9bf | 587 | con->v1.in_base_pos = 0; |
2f713615 ID |
588 | } |
589 | ||
590 | /* | |
591 | * Prepare to read a message. | |
592 | */ | |
593 | static int prepare_read_message(struct ceph_connection *con) | |
594 | { | |
595 | dout("prepare_read_message %p\n", con); | |
596 | BUG_ON(con->in_msg != NULL); | |
a56dd9bf | 597 | con->v1.in_base_pos = 0; |
2f713615 ID |
598 | con->in_front_crc = con->in_middle_crc = con->in_data_crc = 0; |
599 | return 0; | |
600 | } | |
601 | ||
602 | static int read_partial(struct ceph_connection *con, | |
603 | int end, int size, void *object) | |
604 | { | |
a56dd9bf ID |
605 | while (con->v1.in_base_pos < end) { |
606 | int left = end - con->v1.in_base_pos; | |
2f713615 ID |
607 | int have = size - left; |
608 | int ret = ceph_tcp_recvmsg(con->sock, object + have, left); | |
609 | if (ret <= 0) | |
610 | return ret; | |
a56dd9bf | 611 | con->v1.in_base_pos += ret; |
2f713615 ID |
612 | } |
613 | return 1; | |
614 | } | |
615 | ||
616 | /* | |
617 | * Read all or part of the connect-side handshake on a new connection | |
618 | */ | |
619 | static int read_partial_banner(struct ceph_connection *con) | |
620 | { | |
621 | int size; | |
622 | int end; | |
623 | int ret; | |
624 | ||
a56dd9bf | 625 | dout("read_partial_banner %p at %d\n", con, con->v1.in_base_pos); |
2f713615 ID |
626 | |
627 | /* peer's banner */ | |
628 | size = strlen(CEPH_BANNER); | |
629 | end = size; | |
a56dd9bf | 630 | ret = read_partial(con, end, size, con->v1.in_banner); |
2f713615 ID |
631 | if (ret <= 0) |
632 | goto out; | |
633 | ||
a56dd9bf | 634 | size = sizeof(con->v1.actual_peer_addr); |
2f713615 | 635 | end += size; |
a56dd9bf | 636 | ret = read_partial(con, end, size, &con->v1.actual_peer_addr); |
2f713615 ID |
637 | if (ret <= 0) |
638 | goto out; | |
a56dd9bf | 639 | ceph_decode_banner_addr(&con->v1.actual_peer_addr); |
2f713615 | 640 | |
a56dd9bf | 641 | size = sizeof(con->v1.peer_addr_for_me); |
2f713615 | 642 | end += size; |
a56dd9bf | 643 | ret = read_partial(con, end, size, &con->v1.peer_addr_for_me); |
2f713615 ID |
644 | if (ret <= 0) |
645 | goto out; | |
a56dd9bf | 646 | ceph_decode_banner_addr(&con->v1.peer_addr_for_me); |
2f713615 ID |
647 | |
648 | out: | |
649 | return ret; | |
650 | } | |
651 | ||
652 | static int read_partial_connect(struct ceph_connection *con) | |
653 | { | |
654 | int size; | |
655 | int end; | |
656 | int ret; | |
657 | ||
a56dd9bf | 658 | dout("read_partial_connect %p at %d\n", con, con->v1.in_base_pos); |
2f713615 | 659 | |
a56dd9bf | 660 | size = sizeof(con->v1.in_reply); |
2f713615 | 661 | end = size; |
a56dd9bf | 662 | ret = read_partial(con, end, size, &con->v1.in_reply); |
2f713615 ID |
663 | if (ret <= 0) |
664 | goto out; | |
665 | ||
a56dd9bf ID |
666 | if (con->v1.auth) { |
667 | size = le32_to_cpu(con->v1.in_reply.authorizer_len); | |
668 | if (size > con->v1.auth->authorizer_reply_buf_len) { | |
2f713615 | 669 | pr_err("authorizer reply too big: %d > %zu\n", size, |
a56dd9bf | 670 | con->v1.auth->authorizer_reply_buf_len); |
2f713615 ID |
671 | ret = -EINVAL; |
672 | goto out; | |
673 | } | |
674 | ||
675 | end += size; | |
676 | ret = read_partial(con, end, size, | |
a56dd9bf | 677 | con->v1.auth->authorizer_reply_buf); |
2f713615 ID |
678 | if (ret <= 0) |
679 | goto out; | |
680 | } | |
681 | ||
682 | dout("read_partial_connect %p tag %d, con_seq = %u, g_seq = %u\n", | |
a56dd9bf ID |
683 | con, con->v1.in_reply.tag, |
684 | le32_to_cpu(con->v1.in_reply.connect_seq), | |
685 | le32_to_cpu(con->v1.in_reply.global_seq)); | |
2f713615 ID |
686 | out: |
687 | return ret; | |
688 | } | |
689 | ||
690 | /* | |
691 | * Verify the hello banner looks okay. | |
692 | */ | |
693 | static int verify_hello(struct ceph_connection *con) | |
694 | { | |
a56dd9bf | 695 | if (memcmp(con->v1.in_banner, CEPH_BANNER, strlen(CEPH_BANNER))) { |
2f713615 ID |
696 | pr_err("connect to %s got bad banner\n", |
697 | ceph_pr_addr(&con->peer_addr)); | |
698 | con->error_msg = "protocol error, bad banner"; | |
699 | return -1; | |
700 | } | |
701 | return 0; | |
702 | } | |
703 | ||
704 | static int process_banner(struct ceph_connection *con) | |
705 | { | |
706 | struct ceph_entity_addr *my_addr = &con->msgr->inst.addr; | |
707 | ||
708 | dout("process_banner on %p\n", con); | |
709 | ||
710 | if (verify_hello(con) < 0) | |
711 | return -1; | |
712 | ||
713 | /* | |
714 | * Make sure the other end is who we wanted. note that the other | |
715 | * end may not yet know their ip address, so if it's 0.0.0.0, give | |
716 | * them the benefit of the doubt. | |
717 | */ | |
a56dd9bf | 718 | if (memcmp(&con->peer_addr, &con->v1.actual_peer_addr, |
2f713615 | 719 | sizeof(con->peer_addr)) != 0 && |
a56dd9bf ID |
720 | !(ceph_addr_is_blank(&con->v1.actual_peer_addr) && |
721 | con->v1.actual_peer_addr.nonce == con->peer_addr.nonce)) { | |
2f713615 ID |
722 | pr_warn("wrong peer, want %s/%u, got %s/%u\n", |
723 | ceph_pr_addr(&con->peer_addr), | |
724 | le32_to_cpu(con->peer_addr.nonce), | |
a56dd9bf ID |
725 | ceph_pr_addr(&con->v1.actual_peer_addr), |
726 | le32_to_cpu(con->v1.actual_peer_addr.nonce)); | |
2f713615 ID |
727 | con->error_msg = "wrong peer at address"; |
728 | return -1; | |
729 | } | |
730 | ||
731 | /* | |
732 | * did we learn our address? | |
733 | */ | |
734 | if (ceph_addr_is_blank(my_addr)) { | |
735 | memcpy(&my_addr->in_addr, | |
a56dd9bf ID |
736 | &con->v1.peer_addr_for_me.in_addr, |
737 | sizeof(con->v1.peer_addr_for_me.in_addr)); | |
2f713615 ID |
738 | ceph_addr_set_port(my_addr, 0); |
739 | ceph_encode_my_addr(con->msgr); | |
740 | dout("process_banner learned my addr is %s\n", | |
741 | ceph_pr_addr(my_addr)); | |
742 | } | |
743 | ||
744 | return 0; | |
745 | } | |
746 | ||
747 | static int process_connect(struct ceph_connection *con) | |
748 | { | |
749 | u64 sup_feat = from_msgr(con->msgr)->supported_features; | |
750 | u64 req_feat = from_msgr(con->msgr)->required_features; | |
a56dd9bf | 751 | u64 server_feat = le64_to_cpu(con->v1.in_reply.features); |
2f713615 ID |
752 | int ret; |
753 | ||
a56dd9bf | 754 | dout("process_connect on %p tag %d\n", con, con->v1.in_tag); |
2f713615 | 755 | |
a56dd9bf ID |
756 | if (con->v1.auth) { |
757 | int len = le32_to_cpu(con->v1.in_reply.authorizer_len); | |
2f713615 ID |
758 | |
759 | /* | |
760 | * Any connection that defines ->get_authorizer() | |
761 | * should also define ->add_authorizer_challenge() and | |
762 | * ->verify_authorizer_reply(). | |
763 | * | |
764 | * See get_connect_authorizer(). | |
765 | */ | |
a56dd9bf ID |
766 | if (con->v1.in_reply.tag == |
767 | CEPH_MSGR_TAG_CHALLENGE_AUTHORIZER) { | |
2f713615 | 768 | ret = con->ops->add_authorizer_challenge( |
a56dd9bf | 769 | con, con->v1.auth->authorizer_reply_buf, len); |
2f713615 ID |
770 | if (ret < 0) |
771 | return ret; | |
772 | ||
773 | con_out_kvec_reset(con); | |
774 | __prepare_write_connect(con); | |
775 | prepare_read_connect(con); | |
776 | return 0; | |
777 | } | |
778 | ||
779 | if (len) { | |
780 | ret = con->ops->verify_authorizer_reply(con); | |
781 | if (ret < 0) { | |
782 | con->error_msg = "bad authorize reply"; | |
783 | return ret; | |
784 | } | |
785 | } | |
786 | } | |
787 | ||
a56dd9bf | 788 | switch (con->v1.in_reply.tag) { |
2f713615 ID |
789 | case CEPH_MSGR_TAG_FEATURES: |
790 | pr_err("%s%lld %s feature set mismatch," | |
791 | " my %llx < server's %llx, missing %llx\n", | |
792 | ENTITY_NAME(con->peer_name), | |
793 | ceph_pr_addr(&con->peer_addr), | |
794 | sup_feat, server_feat, server_feat & ~sup_feat); | |
795 | con->error_msg = "missing required protocol features"; | |
796 | return -1; | |
797 | ||
798 | case CEPH_MSGR_TAG_BADPROTOVER: | |
799 | pr_err("%s%lld %s protocol version mismatch," | |
800 | " my %d != server's %d\n", | |
801 | ENTITY_NAME(con->peer_name), | |
802 | ceph_pr_addr(&con->peer_addr), | |
a56dd9bf ID |
803 | le32_to_cpu(con->v1.out_connect.protocol_version), |
804 | le32_to_cpu(con->v1.in_reply.protocol_version)); | |
2f713615 ID |
805 | con->error_msg = "protocol version mismatch"; |
806 | return -1; | |
807 | ||
808 | case CEPH_MSGR_TAG_BADAUTHORIZER: | |
a56dd9bf | 809 | con->v1.auth_retry++; |
2f713615 | 810 | dout("process_connect %p got BADAUTHORIZER attempt %d\n", con, |
a56dd9bf ID |
811 | con->v1.auth_retry); |
812 | if (con->v1.auth_retry == 2) { | |
2f713615 ID |
813 | con->error_msg = "connect authorization failure"; |
814 | return -1; | |
815 | } | |
816 | con_out_kvec_reset(con); | |
817 | ret = prepare_write_connect(con); | |
818 | if (ret < 0) | |
819 | return ret; | |
820 | prepare_read_connect(con); | |
821 | break; | |
822 | ||
823 | case CEPH_MSGR_TAG_RESETSESSION: | |
824 | /* | |
825 | * If we connected with a large connect_seq but the peer | |
826 | * has no record of a session with us (no connection, or | |
827 | * connect_seq == 0), they will send RESETSESION to indicate | |
828 | * that they must have reset their session, and may have | |
829 | * dropped messages. | |
830 | */ | |
831 | dout("process_connect got RESET peer seq %u\n", | |
a56dd9bf | 832 | le32_to_cpu(con->v1.in_reply.connect_seq)); |
2f713615 ID |
833 | pr_info("%s%lld %s session reset\n", |
834 | ENTITY_NAME(con->peer_name), | |
835 | ceph_pr_addr(&con->peer_addr)); | |
836 | ceph_con_reset_session(con); | |
837 | con_out_kvec_reset(con); | |
838 | ret = prepare_write_connect(con); | |
839 | if (ret < 0) | |
840 | return ret; | |
841 | prepare_read_connect(con); | |
842 | ||
843 | /* Tell ceph about it. */ | |
844 | mutex_unlock(&con->mutex); | |
845 | if (con->ops->peer_reset) | |
846 | con->ops->peer_reset(con); | |
847 | mutex_lock(&con->mutex); | |
848 | if (con->state != CEPH_CON_S_V1_CONNECT_MSG) | |
849 | return -EAGAIN; | |
850 | break; | |
851 | ||
852 | case CEPH_MSGR_TAG_RETRY_SESSION: | |
853 | /* | |
854 | * If we sent a smaller connect_seq than the peer has, try | |
855 | * again with a larger value. | |
856 | */ | |
857 | dout("process_connect got RETRY_SESSION my seq %u, peer %u\n", | |
a56dd9bf ID |
858 | le32_to_cpu(con->v1.out_connect.connect_seq), |
859 | le32_to_cpu(con->v1.in_reply.connect_seq)); | |
860 | con->v1.connect_seq = le32_to_cpu(con->v1.in_reply.connect_seq); | |
2f713615 ID |
861 | con_out_kvec_reset(con); |
862 | ret = prepare_write_connect(con); | |
863 | if (ret < 0) | |
864 | return ret; | |
865 | prepare_read_connect(con); | |
866 | break; | |
867 | ||
868 | case CEPH_MSGR_TAG_RETRY_GLOBAL: | |
869 | /* | |
870 | * If we sent a smaller global_seq than the peer has, try | |
871 | * again with a larger value. | |
872 | */ | |
873 | dout("process_connect got RETRY_GLOBAL my %u peer_gseq %u\n", | |
a56dd9bf ID |
874 | con->v1.peer_global_seq, |
875 | le32_to_cpu(con->v1.in_reply.global_seq)); | |
2f713615 | 876 | ceph_get_global_seq(con->msgr, |
a56dd9bf | 877 | le32_to_cpu(con->v1.in_reply.global_seq)); |
2f713615 ID |
878 | con_out_kvec_reset(con); |
879 | ret = prepare_write_connect(con); | |
880 | if (ret < 0) | |
881 | return ret; | |
882 | prepare_read_connect(con); | |
883 | break; | |
884 | ||
885 | case CEPH_MSGR_TAG_SEQ: | |
886 | case CEPH_MSGR_TAG_READY: | |
887 | if (req_feat & ~server_feat) { | |
888 | pr_err("%s%lld %s protocol feature mismatch," | |
889 | " my required %llx > server's %llx, need %llx\n", | |
890 | ENTITY_NAME(con->peer_name), | |
891 | ceph_pr_addr(&con->peer_addr), | |
892 | req_feat, server_feat, req_feat & ~server_feat); | |
893 | con->error_msg = "missing required protocol features"; | |
894 | return -1; | |
895 | } | |
896 | ||
897 | WARN_ON(con->state != CEPH_CON_S_V1_CONNECT_MSG); | |
898 | con->state = CEPH_CON_S_OPEN; | |
a56dd9bf ID |
899 | con->v1.auth_retry = 0; /* we authenticated; clear flag */ |
900 | con->v1.peer_global_seq = | |
901 | le32_to_cpu(con->v1.in_reply.global_seq); | |
902 | con->v1.connect_seq++; | |
2f713615 ID |
903 | con->peer_features = server_feat; |
904 | dout("process_connect got READY gseq %d cseq %d (%d)\n", | |
a56dd9bf ID |
905 | con->v1.peer_global_seq, |
906 | le32_to_cpu(con->v1.in_reply.connect_seq), | |
907 | con->v1.connect_seq); | |
908 | WARN_ON(con->v1.connect_seq != | |
909 | le32_to_cpu(con->v1.in_reply.connect_seq)); | |
2f713615 | 910 | |
a56dd9bf | 911 | if (con->v1.in_reply.flags & CEPH_MSG_CONNECT_LOSSY) |
2f713615 ID |
912 | ceph_con_flag_set(con, CEPH_CON_F_LOSSYTX); |
913 | ||
914 | con->delay = 0; /* reset backoff memory */ | |
915 | ||
a56dd9bf | 916 | if (con->v1.in_reply.tag == CEPH_MSGR_TAG_SEQ) { |
2f713615 ID |
917 | prepare_write_seq(con); |
918 | prepare_read_seq(con); | |
919 | } else { | |
920 | prepare_read_tag(con); | |
921 | } | |
922 | break; | |
923 | ||
924 | case CEPH_MSGR_TAG_WAIT: | |
925 | /* | |
926 | * If there is a connection race (we are opening | |
927 | * connections to each other), one of us may just have | |
928 | * to WAIT. This shouldn't happen if we are the | |
929 | * client. | |
930 | */ | |
931 | con->error_msg = "protocol error, got WAIT as client"; | |
932 | return -1; | |
933 | ||
934 | default: | |
935 | con->error_msg = "protocol error, garbage tag during connect"; | |
936 | return -1; | |
937 | } | |
938 | return 0; | |
939 | } | |
940 | ||
941 | /* | |
942 | * read (part of) an ack | |
943 | */ | |
944 | static int read_partial_ack(struct ceph_connection *con) | |
945 | { | |
a56dd9bf | 946 | int size = sizeof(con->v1.in_temp_ack); |
2f713615 ID |
947 | int end = size; |
948 | ||
a56dd9bf | 949 | return read_partial(con, end, size, &con->v1.in_temp_ack); |
2f713615 ID |
950 | } |
951 | ||
952 | /* | |
953 | * We can finally discard anything that's been acked. | |
954 | */ | |
955 | static void process_ack(struct ceph_connection *con) | |
956 | { | |
a56dd9bf | 957 | u64 ack = le64_to_cpu(con->v1.in_temp_ack); |
2f713615 | 958 | |
a56dd9bf | 959 | if (con->v1.in_tag == CEPH_MSGR_TAG_ACK) |
2f713615 ID |
960 | ceph_con_discard_sent(con, ack); |
961 | else | |
962 | ceph_con_discard_requeued(con, ack); | |
963 | ||
964 | prepare_read_tag(con); | |
965 | } | |
966 | ||
967 | static int read_partial_message_section(struct ceph_connection *con, | |
968 | struct kvec *section, | |
969 | unsigned int sec_len, u32 *crc) | |
970 | { | |
971 | int ret, left; | |
972 | ||
973 | BUG_ON(!section); | |
974 | ||
975 | while (section->iov_len < sec_len) { | |
976 | BUG_ON(section->iov_base == NULL); | |
977 | left = sec_len - section->iov_len; | |
978 | ret = ceph_tcp_recvmsg(con->sock, (char *)section->iov_base + | |
979 | section->iov_len, left); | |
980 | if (ret <= 0) | |
981 | return ret; | |
982 | section->iov_len += ret; | |
983 | } | |
984 | if (section->iov_len == sec_len) | |
985 | *crc = crc32c(0, section->iov_base, section->iov_len); | |
986 | ||
987 | return 1; | |
988 | } | |
989 | ||
990 | static int read_partial_msg_data(struct ceph_connection *con) | |
991 | { | |
038b8d1d | 992 | struct ceph_msg_data_cursor *cursor = &con->in_msg->cursor; |
2f713615 ID |
993 | bool do_datacrc = !ceph_test_opt(from_msgr(con->msgr), NOCRC); |
994 | struct page *page; | |
995 | size_t page_offset; | |
996 | size_t length; | |
997 | u32 crc = 0; | |
998 | int ret; | |
999 | ||
2f713615 ID |
1000 | if (do_datacrc) |
1001 | crc = con->in_data_crc; | |
1002 | while (cursor->total_resid) { | |
1003 | if (!cursor->resid) { | |
1004 | ceph_msg_data_advance(cursor, 0); | |
1005 | continue; | |
1006 | } | |
1007 | ||
da4ab869 | 1008 | page = ceph_msg_data_next(cursor, &page_offset, &length); |
2f713615 ID |
1009 | ret = ceph_tcp_recvpage(con->sock, page, page_offset, length); |
1010 | if (ret <= 0) { | |
1011 | if (do_datacrc) | |
1012 | con->in_data_crc = crc; | |
1013 | ||
1014 | return ret; | |
1015 | } | |
1016 | ||
1017 | if (do_datacrc) | |
1018 | crc = ceph_crc32c_page(crc, page, page_offset, ret); | |
1019 | ceph_msg_data_advance(cursor, (size_t)ret); | |
1020 | } | |
1021 | if (do_datacrc) | |
1022 | con->in_data_crc = crc; | |
1023 | ||
1024 | return 1; /* must return > 0 to indicate success */ | |
1025 | } | |
1026 | ||
038b8d1d ID |
1027 | static int read_partial_msg_data_bounce(struct ceph_connection *con) |
1028 | { | |
1029 | struct ceph_msg_data_cursor *cursor = &con->in_msg->cursor; | |
1030 | struct page *page; | |
1031 | size_t off, len; | |
1032 | u32 crc; | |
1033 | int ret; | |
1034 | ||
1035 | if (unlikely(!con->bounce_page)) { | |
1036 | con->bounce_page = alloc_page(GFP_NOIO); | |
1037 | if (!con->bounce_page) { | |
1038 | pr_err("failed to allocate bounce page\n"); | |
1039 | return -ENOMEM; | |
1040 | } | |
1041 | } | |
1042 | ||
1043 | crc = con->in_data_crc; | |
1044 | while (cursor->total_resid) { | |
1045 | if (!cursor->resid) { | |
1046 | ceph_msg_data_advance(cursor, 0); | |
1047 | continue; | |
1048 | } | |
1049 | ||
da4ab869 | 1050 | page = ceph_msg_data_next(cursor, &off, &len); |
038b8d1d ID |
1051 | ret = ceph_tcp_recvpage(con->sock, con->bounce_page, 0, len); |
1052 | if (ret <= 0) { | |
1053 | con->in_data_crc = crc; | |
1054 | return ret; | |
1055 | } | |
1056 | ||
1057 | crc = crc32c(crc, page_address(con->bounce_page), ret); | |
1058 | memcpy_to_page(page, off, page_address(con->bounce_page), ret); | |
1059 | ||
1060 | ceph_msg_data_advance(cursor, ret); | |
1061 | } | |
1062 | con->in_data_crc = crc; | |
1063 | ||
1064 | return 1; /* must return > 0 to indicate success */ | |
1065 | } | |
1066 | ||
2f713615 ID |
1067 | /* |
1068 | * read (part of) a message. | |
1069 | */ | |
1070 | static int read_partial_message(struct ceph_connection *con) | |
1071 | { | |
1072 | struct ceph_msg *m = con->in_msg; | |
1073 | int size; | |
1074 | int end; | |
1075 | int ret; | |
1076 | unsigned int front_len, middle_len, data_len; | |
1077 | bool do_datacrc = !ceph_test_opt(from_msgr(con->msgr), NOCRC); | |
1078 | bool need_sign = (con->peer_features & CEPH_FEATURE_MSG_AUTH); | |
1079 | u64 seq; | |
1080 | u32 crc; | |
1081 | ||
1082 | dout("read_partial_message con %p msg %p\n", con, m); | |
1083 | ||
1084 | /* header */ | |
a56dd9bf | 1085 | size = sizeof(con->v1.in_hdr); |
2f713615 | 1086 | end = size; |
a56dd9bf | 1087 | ret = read_partial(con, end, size, &con->v1.in_hdr); |
2f713615 ID |
1088 | if (ret <= 0) |
1089 | return ret; | |
1090 | ||
a56dd9bf ID |
1091 | crc = crc32c(0, &con->v1.in_hdr, offsetof(struct ceph_msg_header, crc)); |
1092 | if (cpu_to_le32(crc) != con->v1.in_hdr.crc) { | |
2f713615 | 1093 | pr_err("read_partial_message bad hdr crc %u != expected %u\n", |
a56dd9bf | 1094 | crc, con->v1.in_hdr.crc); |
2f713615 ID |
1095 | return -EBADMSG; |
1096 | } | |
1097 | ||
a56dd9bf | 1098 | front_len = le32_to_cpu(con->v1.in_hdr.front_len); |
2f713615 ID |
1099 | if (front_len > CEPH_MSG_MAX_FRONT_LEN) |
1100 | return -EIO; | |
a56dd9bf | 1101 | middle_len = le32_to_cpu(con->v1.in_hdr.middle_len); |
2f713615 ID |
1102 | if (middle_len > CEPH_MSG_MAX_MIDDLE_LEN) |
1103 | return -EIO; | |
a56dd9bf | 1104 | data_len = le32_to_cpu(con->v1.in_hdr.data_len); |
2f713615 ID |
1105 | if (data_len > CEPH_MSG_MAX_DATA_LEN) |
1106 | return -EIO; | |
1107 | ||
1108 | /* verify seq# */ | |
a56dd9bf | 1109 | seq = le64_to_cpu(con->v1.in_hdr.seq); |
2f713615 ID |
1110 | if ((s64)seq - (s64)con->in_seq < 1) { |
1111 | pr_info("skipping %s%lld %s seq %lld expected %lld\n", | |
1112 | ENTITY_NAME(con->peer_name), | |
1113 | ceph_pr_addr(&con->peer_addr), | |
1114 | seq, con->in_seq + 1); | |
a56dd9bf ID |
1115 | con->v1.in_base_pos = -front_len - middle_len - data_len - |
1116 | sizeof_footer(con); | |
1117 | con->v1.in_tag = CEPH_MSGR_TAG_READY; | |
2f713615 ID |
1118 | return 1; |
1119 | } else if ((s64)seq - (s64)con->in_seq > 1) { | |
1120 | pr_err("read_partial_message bad seq %lld expected %lld\n", | |
1121 | seq, con->in_seq + 1); | |
1122 | con->error_msg = "bad message sequence # for incoming message"; | |
1123 | return -EBADE; | |
1124 | } | |
1125 | ||
1126 | /* allocate message? */ | |
1127 | if (!con->in_msg) { | |
1128 | int skip = 0; | |
1129 | ||
a56dd9bf | 1130 | dout("got hdr type %d front %d data %d\n", con->v1.in_hdr.type, |
2f713615 | 1131 | front_len, data_len); |
a56dd9bf | 1132 | ret = ceph_con_in_msg_alloc(con, &con->v1.in_hdr, &skip); |
2f713615 ID |
1133 | if (ret < 0) |
1134 | return ret; | |
1135 | ||
9d5ae6f3 | 1136 | BUG_ON((!con->in_msg) ^ skip); |
2f713615 ID |
1137 | if (skip) { |
1138 | /* skip this message */ | |
1139 | dout("alloc_msg said skip message\n"); | |
a56dd9bf ID |
1140 | con->v1.in_base_pos = -front_len - middle_len - |
1141 | data_len - sizeof_footer(con); | |
1142 | con->v1.in_tag = CEPH_MSGR_TAG_READY; | |
2f713615 ID |
1143 | con->in_seq++; |
1144 | return 1; | |
1145 | } | |
1146 | ||
1147 | BUG_ON(!con->in_msg); | |
1148 | BUG_ON(con->in_msg->con != con); | |
1149 | m = con->in_msg; | |
1150 | m->front.iov_len = 0; /* haven't read it yet */ | |
1151 | if (m->middle) | |
1152 | m->middle->vec.iov_len = 0; | |
1153 | ||
1154 | /* prepare for data payload, if any */ | |
1155 | ||
1156 | if (data_len) | |
1157 | prepare_message_data(con->in_msg, data_len); | |
1158 | } | |
1159 | ||
1160 | /* front */ | |
1161 | ret = read_partial_message_section(con, &m->front, front_len, | |
1162 | &con->in_front_crc); | |
1163 | if (ret <= 0) | |
1164 | return ret; | |
1165 | ||
1166 | /* middle */ | |
1167 | if (m->middle) { | |
1168 | ret = read_partial_message_section(con, &m->middle->vec, | |
1169 | middle_len, | |
1170 | &con->in_middle_crc); | |
1171 | if (ret <= 0) | |
1172 | return ret; | |
1173 | } | |
1174 | ||
1175 | /* (page) data */ | |
1176 | if (data_len) { | |
038b8d1d ID |
1177 | if (!m->num_data_items) |
1178 | return -EIO; | |
1179 | ||
1180 | if (ceph_test_opt(from_msgr(con->msgr), RXBOUNCE)) | |
1181 | ret = read_partial_msg_data_bounce(con); | |
1182 | else | |
1183 | ret = read_partial_msg_data(con); | |
2f713615 ID |
1184 | if (ret <= 0) |
1185 | return ret; | |
1186 | } | |
1187 | ||
1188 | /* footer */ | |
1189 | size = sizeof_footer(con); | |
1190 | end += size; | |
1191 | ret = read_partial(con, end, size, &m->footer); | |
1192 | if (ret <= 0) | |
1193 | return ret; | |
1194 | ||
1195 | if (!need_sign) { | |
1196 | m->footer.flags = m->old_footer.flags; | |
1197 | m->footer.sig = 0; | |
1198 | } | |
1199 | ||
1200 | dout("read_partial_message got msg %p %d (%u) + %d (%u) + %d (%u)\n", | |
1201 | m, front_len, m->footer.front_crc, middle_len, | |
1202 | m->footer.middle_crc, data_len, m->footer.data_crc); | |
1203 | ||
1204 | /* crc ok? */ | |
1205 | if (con->in_front_crc != le32_to_cpu(m->footer.front_crc)) { | |
1206 | pr_err("read_partial_message %p front crc %u != exp. %u\n", | |
1207 | m, con->in_front_crc, m->footer.front_crc); | |
1208 | return -EBADMSG; | |
1209 | } | |
1210 | if (con->in_middle_crc != le32_to_cpu(m->footer.middle_crc)) { | |
1211 | pr_err("read_partial_message %p middle crc %u != exp %u\n", | |
1212 | m, con->in_middle_crc, m->footer.middle_crc); | |
1213 | return -EBADMSG; | |
1214 | } | |
1215 | if (do_datacrc && | |
1216 | (m->footer.flags & CEPH_MSG_FOOTER_NOCRC) == 0 && | |
1217 | con->in_data_crc != le32_to_cpu(m->footer.data_crc)) { | |
1218 | pr_err("read_partial_message %p data crc %u != exp. %u\n", m, | |
1219 | con->in_data_crc, le32_to_cpu(m->footer.data_crc)); | |
1220 | return -EBADMSG; | |
1221 | } | |
1222 | ||
1223 | if (need_sign && con->ops->check_message_signature && | |
1224 | con->ops->check_message_signature(m)) { | |
1225 | pr_err("read_partial_message %p signature check failed\n", m); | |
1226 | return -EBADMSG; | |
1227 | } | |
1228 | ||
1229 | return 1; /* done! */ | |
1230 | } | |
1231 | ||
1232 | static int read_keepalive_ack(struct ceph_connection *con) | |
1233 | { | |
1234 | struct ceph_timespec ceph_ts; | |
1235 | size_t size = sizeof(ceph_ts); | |
1236 | int ret = read_partial(con, size, size, &ceph_ts); | |
1237 | if (ret <= 0) | |
1238 | return ret; | |
1239 | ceph_decode_timespec64(&con->last_keepalive_ack, &ceph_ts); | |
1240 | prepare_read_tag(con); | |
1241 | return 1; | |
1242 | } | |
1243 | ||
1244 | /* | |
1245 | * Read what we can from the socket. | |
1246 | */ | |
1247 | int ceph_con_v1_try_read(struct ceph_connection *con) | |
1248 | { | |
1249 | int ret = -1; | |
1250 | ||
1251 | more: | |
1252 | dout("try_read start %p state %d\n", con, con->state); | |
1253 | if (con->state != CEPH_CON_S_V1_BANNER && | |
1254 | con->state != CEPH_CON_S_V1_CONNECT_MSG && | |
1255 | con->state != CEPH_CON_S_OPEN) | |
1256 | return 0; | |
1257 | ||
1258 | BUG_ON(!con->sock); | |
1259 | ||
a56dd9bf ID |
1260 | dout("try_read tag %d in_base_pos %d\n", con->v1.in_tag, |
1261 | con->v1.in_base_pos); | |
2f713615 ID |
1262 | |
1263 | if (con->state == CEPH_CON_S_V1_BANNER) { | |
1264 | ret = read_partial_banner(con); | |
1265 | if (ret <= 0) | |
1266 | goto out; | |
1267 | ret = process_banner(con); | |
1268 | if (ret < 0) | |
1269 | goto out; | |
1270 | ||
1271 | con->state = CEPH_CON_S_V1_CONNECT_MSG; | |
1272 | ||
1273 | /* | |
1274 | * Received banner is good, exchange connection info. | |
1275 | * Do not reset out_kvec, as sending our banner raced | |
1276 | * with receiving peer banner after connect completed. | |
1277 | */ | |
1278 | ret = prepare_write_connect(con); | |
1279 | if (ret < 0) | |
1280 | goto out; | |
1281 | prepare_read_connect(con); | |
1282 | ||
1283 | /* Send connection info before awaiting response */ | |
1284 | goto out; | |
1285 | } | |
1286 | ||
1287 | if (con->state == CEPH_CON_S_V1_CONNECT_MSG) { | |
1288 | ret = read_partial_connect(con); | |
1289 | if (ret <= 0) | |
1290 | goto out; | |
1291 | ret = process_connect(con); | |
1292 | if (ret < 0) | |
1293 | goto out; | |
1294 | goto more; | |
1295 | } | |
1296 | ||
1297 | WARN_ON(con->state != CEPH_CON_S_OPEN); | |
1298 | ||
a56dd9bf | 1299 | if (con->v1.in_base_pos < 0) { |
2f713615 ID |
1300 | /* |
1301 | * skipping + discarding content. | |
1302 | */ | |
a56dd9bf | 1303 | ret = ceph_tcp_recvmsg(con->sock, NULL, -con->v1.in_base_pos); |
2f713615 ID |
1304 | if (ret <= 0) |
1305 | goto out; | |
a56dd9bf ID |
1306 | dout("skipped %d / %d bytes\n", ret, -con->v1.in_base_pos); |
1307 | con->v1.in_base_pos += ret; | |
1308 | if (con->v1.in_base_pos) | |
2f713615 ID |
1309 | goto more; |
1310 | } | |
a56dd9bf | 1311 | if (con->v1.in_tag == CEPH_MSGR_TAG_READY) { |
2f713615 ID |
1312 | /* |
1313 | * what's next? | |
1314 | */ | |
a56dd9bf | 1315 | ret = ceph_tcp_recvmsg(con->sock, &con->v1.in_tag, 1); |
2f713615 ID |
1316 | if (ret <= 0) |
1317 | goto out; | |
a56dd9bf ID |
1318 | dout("try_read got tag %d\n", con->v1.in_tag); |
1319 | switch (con->v1.in_tag) { | |
2f713615 ID |
1320 | case CEPH_MSGR_TAG_MSG: |
1321 | prepare_read_message(con); | |
1322 | break; | |
1323 | case CEPH_MSGR_TAG_ACK: | |
1324 | prepare_read_ack(con); | |
1325 | break; | |
1326 | case CEPH_MSGR_TAG_KEEPALIVE2_ACK: | |
1327 | prepare_read_keepalive_ack(con); | |
1328 | break; | |
1329 | case CEPH_MSGR_TAG_CLOSE: | |
1330 | ceph_con_close_socket(con); | |
1331 | con->state = CEPH_CON_S_CLOSED; | |
1332 | goto out; | |
1333 | default: | |
1334 | goto bad_tag; | |
1335 | } | |
1336 | } | |
a56dd9bf | 1337 | if (con->v1.in_tag == CEPH_MSGR_TAG_MSG) { |
2f713615 ID |
1338 | ret = read_partial_message(con); |
1339 | if (ret <= 0) { | |
1340 | switch (ret) { | |
1341 | case -EBADMSG: | |
1342 | con->error_msg = "bad crc/signature"; | |
1343 | fallthrough; | |
1344 | case -EBADE: | |
1345 | ret = -EIO; | |
1346 | break; | |
1347 | case -EIO: | |
1348 | con->error_msg = "io error"; | |
1349 | break; | |
1350 | } | |
1351 | goto out; | |
1352 | } | |
a56dd9bf | 1353 | if (con->v1.in_tag == CEPH_MSGR_TAG_READY) |
2f713615 ID |
1354 | goto more; |
1355 | ceph_con_process_message(con); | |
1356 | if (con->state == CEPH_CON_S_OPEN) | |
1357 | prepare_read_tag(con); | |
1358 | goto more; | |
1359 | } | |
a56dd9bf ID |
1360 | if (con->v1.in_tag == CEPH_MSGR_TAG_ACK || |
1361 | con->v1.in_tag == CEPH_MSGR_TAG_SEQ) { | |
2f713615 ID |
1362 | /* |
1363 | * the final handshake seq exchange is semantically | |
1364 | * equivalent to an ACK | |
1365 | */ | |
1366 | ret = read_partial_ack(con); | |
1367 | if (ret <= 0) | |
1368 | goto out; | |
1369 | process_ack(con); | |
1370 | goto more; | |
1371 | } | |
a56dd9bf | 1372 | if (con->v1.in_tag == CEPH_MSGR_TAG_KEEPALIVE2_ACK) { |
2f713615 ID |
1373 | ret = read_keepalive_ack(con); |
1374 | if (ret <= 0) | |
1375 | goto out; | |
1376 | goto more; | |
1377 | } | |
1378 | ||
1379 | out: | |
1380 | dout("try_read done on %p ret %d\n", con, ret); | |
1381 | return ret; | |
1382 | ||
1383 | bad_tag: | |
a56dd9bf | 1384 | pr_err("try_read bad tag %d\n", con->v1.in_tag); |
2f713615 ID |
1385 | con->error_msg = "protocol error, garbage tag"; |
1386 | ret = -1; | |
1387 | goto out; | |
1388 | } | |
1389 | ||
1390 | /* | |
1391 | * Write something to the socket. Called in a worker thread when the | |
1392 | * socket appears to be writeable and we have something ready to send. | |
1393 | */ | |
1394 | int ceph_con_v1_try_write(struct ceph_connection *con) | |
1395 | { | |
1396 | int ret = 1; | |
1397 | ||
1398 | dout("try_write start %p state %d\n", con, con->state); | |
1399 | if (con->state != CEPH_CON_S_PREOPEN && | |
1400 | con->state != CEPH_CON_S_V1_BANNER && | |
1401 | con->state != CEPH_CON_S_V1_CONNECT_MSG && | |
1402 | con->state != CEPH_CON_S_OPEN) | |
1403 | return 0; | |
1404 | ||
1405 | /* open the socket first? */ | |
1406 | if (con->state == CEPH_CON_S_PREOPEN) { | |
1407 | BUG_ON(con->sock); | |
1408 | con->state = CEPH_CON_S_V1_BANNER; | |
1409 | ||
1410 | con_out_kvec_reset(con); | |
1411 | prepare_write_banner(con); | |
1412 | prepare_read_banner(con); | |
1413 | ||
1414 | BUG_ON(con->in_msg); | |
a56dd9bf | 1415 | con->v1.in_tag = CEPH_MSGR_TAG_READY; |
2f713615 ID |
1416 | dout("try_write initiating connect on %p new state %d\n", |
1417 | con, con->state); | |
1418 | ret = ceph_tcp_connect(con); | |
1419 | if (ret < 0) { | |
1420 | con->error_msg = "connect error"; | |
1421 | goto out; | |
1422 | } | |
1423 | } | |
1424 | ||
1425 | more: | |
a56dd9bf | 1426 | dout("try_write out_kvec_bytes %d\n", con->v1.out_kvec_bytes); |
2f713615 ID |
1427 | BUG_ON(!con->sock); |
1428 | ||
1429 | /* kvec data queued? */ | |
a56dd9bf | 1430 | if (con->v1.out_kvec_left) { |
2f713615 ID |
1431 | ret = write_partial_kvec(con); |
1432 | if (ret <= 0) | |
1433 | goto out; | |
1434 | } | |
a56dd9bf | 1435 | if (con->v1.out_skip) { |
2f713615 ID |
1436 | ret = write_partial_skip(con); |
1437 | if (ret <= 0) | |
1438 | goto out; | |
1439 | } | |
1440 | ||
1441 | /* msg pages? */ | |
1442 | if (con->out_msg) { | |
a56dd9bf | 1443 | if (con->v1.out_msg_done) { |
2f713615 ID |
1444 | ceph_msg_put(con->out_msg); |
1445 | con->out_msg = NULL; /* we're done with this one */ | |
1446 | goto do_next; | |
1447 | } | |
1448 | ||
1449 | ret = write_partial_message_data(con); | |
1450 | if (ret == 1) | |
1451 | goto more; /* we need to send the footer, too! */ | |
1452 | if (ret == 0) | |
1453 | goto out; | |
1454 | if (ret < 0) { | |
1455 | dout("try_write write_partial_message_data err %d\n", | |
1456 | ret); | |
1457 | goto out; | |
1458 | } | |
1459 | } | |
1460 | ||
1461 | do_next: | |
1462 | if (con->state == CEPH_CON_S_OPEN) { | |
1463 | if (ceph_con_flag_test_and_clear(con, | |
1464 | CEPH_CON_F_KEEPALIVE_PENDING)) { | |
1465 | prepare_write_keepalive(con); | |
1466 | goto more; | |
1467 | } | |
1468 | /* is anything else pending? */ | |
1469 | if (!list_empty(&con->out_queue)) { | |
1470 | prepare_write_message(con); | |
1471 | goto more; | |
1472 | } | |
1473 | if (con->in_seq > con->in_seq_acked) { | |
1474 | prepare_write_ack(con); | |
1475 | goto more; | |
1476 | } | |
1477 | } | |
1478 | ||
1479 | /* Nothing to do! */ | |
1480 | ceph_con_flag_clear(con, CEPH_CON_F_WRITE_PENDING); | |
1481 | dout("try_write nothing else to write.\n"); | |
1482 | ret = 0; | |
1483 | out: | |
1484 | dout("try_write done on %p ret %d\n", con, ret); | |
1485 | return ret; | |
1486 | } | |
1487 | ||
1488 | void ceph_con_v1_revoke(struct ceph_connection *con) | |
1489 | { | |
1490 | struct ceph_msg *msg = con->out_msg; | |
1491 | ||
a56dd9bf | 1492 | WARN_ON(con->v1.out_skip); |
2f713615 | 1493 | /* footer */ |
a56dd9bf ID |
1494 | if (con->v1.out_msg_done) { |
1495 | con->v1.out_skip += con_out_kvec_skip(con); | |
2f713615 ID |
1496 | } else { |
1497 | WARN_ON(!msg->data_length); | |
a56dd9bf | 1498 | con->v1.out_skip += sizeof_footer(con); |
2f713615 ID |
1499 | } |
1500 | /* data, middle, front */ | |
1501 | if (msg->data_length) | |
a56dd9bf | 1502 | con->v1.out_skip += msg->cursor.total_resid; |
2f713615 | 1503 | if (msg->middle) |
a56dd9bf ID |
1504 | con->v1.out_skip += con_out_kvec_skip(con); |
1505 | con->v1.out_skip += con_out_kvec_skip(con); | |
2f713615 ID |
1506 | |
1507 | dout("%s con %p out_kvec_bytes %d out_skip %d\n", __func__, con, | |
a56dd9bf | 1508 | con->v1.out_kvec_bytes, con->v1.out_skip); |
2f713615 ID |
1509 | } |
1510 | ||
1511 | void ceph_con_v1_revoke_incoming(struct ceph_connection *con) | |
1512 | { | |
a56dd9bf ID |
1513 | unsigned int front_len = le32_to_cpu(con->v1.in_hdr.front_len); |
1514 | unsigned int middle_len = le32_to_cpu(con->v1.in_hdr.middle_len); | |
1515 | unsigned int data_len = le32_to_cpu(con->v1.in_hdr.data_len); | |
2f713615 ID |
1516 | |
1517 | /* skip rest of message */ | |
a56dd9bf | 1518 | con->v1.in_base_pos = con->v1.in_base_pos - |
2f713615 ID |
1519 | sizeof(struct ceph_msg_header) - |
1520 | front_len - | |
1521 | middle_len - | |
1522 | data_len - | |
1523 | sizeof(struct ceph_msg_footer); | |
1524 | ||
a56dd9bf | 1525 | con->v1.in_tag = CEPH_MSGR_TAG_READY; |
2f713615 ID |
1526 | con->in_seq++; |
1527 | ||
a56dd9bf | 1528 | dout("%s con %p in_base_pos %d\n", __func__, con, con->v1.in_base_pos); |
2f713615 ID |
1529 | } |
1530 | ||
1531 | bool ceph_con_v1_opened(struct ceph_connection *con) | |
1532 | { | |
a56dd9bf | 1533 | return con->v1.connect_seq; |
2f713615 ID |
1534 | } |
1535 | ||
1536 | void ceph_con_v1_reset_session(struct ceph_connection *con) | |
1537 | { | |
a56dd9bf ID |
1538 | con->v1.connect_seq = 0; |
1539 | con->v1.peer_global_seq = 0; | |
2f713615 ID |
1540 | } |
1541 | ||
1542 | void ceph_con_v1_reset_protocol(struct ceph_connection *con) | |
1543 | { | |
a56dd9bf | 1544 | con->v1.out_skip = 0; |
2f713615 | 1545 | } |