src/common/relayd/relayd.c

   1 /*
   2  * Copyright (C) 2012 - David Goulet <dgoulet@efficios.com>
   3  *
   4  * This program is free software; you can redistribute it and/or modify it
   5  * under the terms of the GNU General Public License, version 2 only, as
   6  * published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful, but WITHOUT
   9  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  10  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
  11  * more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along with
  14  * this program; if not, write to the Free Software Foundation, Inc., 51
  15  * Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  16  */
  17
  18 #define _LGPL_SOURCE
  19 #include <assert.h>
  20 #include <stdio.h>
  21 #include <stdlib.h>
  22 #include <string.h>
  23 #include <sys/stat.h>
  24 #include <inttypes.h>
  25
  26 #include <common/common.h>
  27 #include <common/defaults.h>
  28 #include <common/compat/endian.h>
  29 #include <common/compat/string.h>
  30 #include <common/sessiond-comm/relayd.h>
  31 #include <common/index/ctf-index.h>
  32
  33 #include "relayd.h"
  34
  35 /*
  36  * Send command. Fill up the header and append the data.
  37  */
  38 static int send_command(struct lttcomm_relayd_sock *rsock,
  39                 enum lttcomm_relayd_command cmd, const void *data, size_t size,
  40                 int flags)
  41 {
  42         int ret;
  43         struct lttcomm_relayd_hdr header;
  44         char *buf;
  45         uint64_t buf_size = sizeof(header);
  46
  47         if (rsock->sock.fd < 0) {
  48                 return -ECONNRESET;
  49         }
  50
  51         if (data) {
  52                 buf_size += size;
  53         }
  54
  55         buf = zmalloc(buf_size);
  56         if (buf == NULL) {
  57                 PERROR("zmalloc relayd send command buf");
  58                 ret = -1;
  59                 goto alloc_error;
  60         }
  61
  62         memset(&header, 0, sizeof(header));
  63         header.cmd = htobe32(cmd);
  64         header.data_size = htobe64(size);
  65
  66         /* Zeroed for now since not used. */
  67         header.cmd_version = 0;
  68         header.circuit_id = 0;
  69
  70         /* Prepare buffer to send. */
  71         memcpy(buf, &header, sizeof(header));
  72         if (data) {
  73                 memcpy(buf + sizeof(header), data, size);
  74         }
  75
  76         DBG3("Relayd sending command %d of size %" PRIu64, (int) cmd, buf_size);
  77         ret = rsock->sock.ops->sendmsg(&rsock->sock, buf, buf_size, flags);
  78         if (ret < 0) {
  79                 PERROR("Failed to send command %d of size %" PRIu64,
  80                                 (int) cmd, buf_size);
  81                 ret = -errno;
  82                 goto error;
  83         }
  84 error:
  85         free(buf);
  86 alloc_error:
  87         return ret;
  88 }
  89
  90 /*
  91  * Receive reply data on socket. This MUST be call after send_command or else
  92  * could result in unexpected behavior(s).
  93  */
  94 static int recv_reply(struct lttcomm_relayd_sock *rsock, void *data, size_t size)
  95 {
  96         int ret;
  97
  98         if (rsock->sock.fd < 0) {
  99                 return -ECONNRESET;
 100         }
 101
 102         DBG3("Relayd waiting for reply of size %zu", size);
 103
 104         ret = rsock->sock.ops->recvmsg(&rsock->sock, data, size, 0);
 105         if (ret <= 0 || ret != size) {
 106                 if (ret == 0) {
 107                         /* Orderly shutdown. */
 108                         DBG("Socket %d has performed an orderly shutdown", rsock->sock.fd);
 109                 } else {
 110                         DBG("Receiving reply failed on sock %d for size %zu with ret %d",
 111                                         rsock->sock.fd, size, ret);
 112                 }
 113                 /* Always return -1 here and the caller can use errno. */
 114                 ret = -1;
 115                 goto error;
 116         }
 117
 118 error:
 119         return ret;
 120 }
 121
 122 /*
 123  * Starting from 2.11, RELAYD_CREATE_SESSION payload (session_name & hostname)
 124  * have no length restriction on the sender side.
 125  * Length for both payloads is stored in the msg struct. A new dynamic size
 126  * payload size is introduced.
 127  */
 128 static int relayd_create_session_2_11(struct lttcomm_relayd_sock *rsock,
 129                 char *session_name, char *hostname,
 130                 int session_live_timer, unsigned int snapshot,
 131                 uint64_t sessiond_session_id, const lttng_uuid sessiond_uuid)
 132 {
 133         int ret;
 134         struct lttcomm_relayd_create_session_2_11 *msg = NULL;
 135         size_t session_name_len;
 136         size_t hostname_len;
 137         size_t msg_length;
 138
 139         /* The two names are sent with a '\0' delimiter between them. */
 140         session_name_len = strlen(session_name) + 1;
 141         hostname_len = strlen(hostname) + 1;
 142
 143         msg_length = sizeof(*msg) + session_name_len + hostname_len;
 144         msg = zmalloc(msg_length);
 145         if (!msg) {
 146                 PERROR("zmalloc create_session_2_11 command message");
 147                 ret = -1;
 148                 goto error;
 149         }
 150
 151         assert(session_name_len <= UINT32_MAX);
 152         msg->session_name_len = htobe32(session_name_len);
 153
 154         assert(hostname_len <= UINT32_MAX);
 155         msg->hostname_len = htobe32(hostname_len);
 156
 157         if (lttng_strncpy(msg->names, session_name, session_name_len)) {
 158                 ret = -1;
 159                 goto error;
 160         }
 161         if (lttng_strncpy(msg->names + session_name_len, hostname, hostname_len)) {
 162                 ret = -1;
 163                 goto error;
 164         }
 165
 166         msg->live_timer = htobe32(session_live_timer);
 167         msg->snapshot = !!snapshot;
 168
 169         lttng_uuid_copy(msg->sessiond_uuid, sessiond_uuid);
 170         msg->session_id = htobe64(sessiond_session_id);
 171
 172         /* Send command */
 173         ret = send_command(rsock, RELAYD_CREATE_SESSION, msg, msg_length, 0);
 174         if (ret < 0) {
 175                 goto error;
 176         }
 177 error:
 178         free(msg);
 179         return ret;
 180 }
 181 /*
 182  * From 2.4 to 2.10, RELAYD_CREATE_SESSION takes additional parameters to
 183  * support the live reading capability.
 184  */
 185 static int relayd_create_session_2_4(struct lttcomm_relayd_sock *rsock,
 186                 char *session_name, char *hostname, int session_live_timer,
 187                 unsigned int snapshot)
 188 {
 189         int ret;
 190         struct lttcomm_relayd_create_session_2_4 msg;
 191
 192         if (lttng_strncpy(msg.session_name, session_name,
 193                         sizeof(msg.session_name))) {
 194                 ret = -1;
 195                 goto error;
 196         }
 197         if (lttng_strncpy(msg.hostname, hostname, sizeof(msg.hostname))) {
 198                 ret = -1;
 199                 goto error;
 200         }
 201         msg.live_timer = htobe32(session_live_timer);
 202         msg.snapshot = htobe32(snapshot);
 203
 204         /* Send command */
 205         ret = send_command(rsock, RELAYD_CREATE_SESSION, &msg, sizeof(msg), 0);
 206         if (ret < 0) {
 207                 goto error;
 208         }
 209
 210 error:
 211         return ret;
 212 }
 213
 214 /*
 215  * RELAYD_CREATE_SESSION from 2.1 to 2.3.
 216  */
 217 static int relayd_create_session_2_1(struct lttcomm_relayd_sock *rsock)
 218 {
 219         int ret;
 220
 221         /* Send command */
 222         ret = send_command(rsock, RELAYD_CREATE_SESSION, NULL, 0, 0);
 223         if (ret < 0) {
 224                 goto error;
 225         }
 226
 227 error:
 228         return ret;
 229 }
 230
 231 /*
 232  * Send a RELAYD_CREATE_SESSION command to the relayd with the given socket and
 233  * set session_id of the relayd if we have a successful reply from the relayd.
 234  *
 235  * On success, return 0 else a negative value which is either an errno error or
 236  * a lttng error code from the relayd.
 237  */
 238 int relayd_create_session(struct lttcomm_relayd_sock *rsock, uint64_t *relayd_session_id,
 239                 char *session_name, char *hostname, int session_live_timer,
 240                 unsigned int snapshot, uint64_t sessiond_session_id,
 241                 const lttng_uuid sessiond_uuid)
 242 {
 243         int ret;
 244         struct lttcomm_relayd_status_session reply;
 245
 246         assert(rsock);
 247         assert(relayd_session_id);
 248
 249         DBG("Relayd create session");
 250
 251         if (rsock->minor < 4) {
 252                 /* From 2.1 to 2.3 */
 253                 ret = relayd_create_session_2_1(rsock);
 254         } else if (rsock->minor >= 4 && rsock->minor < 11) {
 255                 /* From 2.4 to 2.10 */
 256                 ret = relayd_create_session_2_4(rsock, session_name,
 257                                 hostname, session_live_timer, snapshot);
 258         } else {
 259                 /* From 2.11 to ... */
 260                 ret = relayd_create_session_2_11(rsock, session_name,
 261                                 hostname, session_live_timer, snapshot,
 262                                 sessiond_session_id, sessiond_uuid);
 263         }
 264
 265         if (ret < 0) {
 266                 goto error;
 267         }
 268
 269         /* Receive response */
 270         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
 271         if (ret < 0) {
 272                 goto error;
 273         }
 274
 275         reply.session_id = be64toh(reply.session_id);
 276         reply.ret_code = be32toh(reply.ret_code);
 277
 278         /* Return session id or negative ret code. */
 279         if (reply.ret_code != LTTNG_OK) {
 280                 ret = -1;
 281                 ERR("Relayd create session replied error %d", reply.ret_code);
 282                 goto error;
 283         } else {
 284                 ret = 0;
 285                 *relayd_session_id = reply.session_id;
 286         }
 287
 288         DBG("Relayd session created with id %" PRIu64, reply.session_id);
 289
 290 error:
 291         return ret;
 292 }
 293
 294 static int relayd_add_stream_2_1(struct lttcomm_relayd_sock *rsock,
 295                 const char *channel_name, const char *pathname)
 296 {
 297         int ret;
 298         struct lttcomm_relayd_add_stream msg;
 299
 300         memset(&msg, 0, sizeof(msg));
 301         if (lttng_strncpy(msg.channel_name, channel_name,
 302                                 sizeof(msg.channel_name))) {
 303                 ret = -1;
 304                 goto error;
 305         }
 306
 307         if (lttng_strncpy(msg.pathname, pathname,
 308                                 sizeof(msg.pathname))) {
 309                 ret = -1;
 310                 goto error;
 311         }
 312
 313         /* Send command */
 314         ret = send_command(rsock, RELAYD_ADD_STREAM, (void *) &msg, sizeof(msg), 0);
 315         if (ret < 0) {
 316                 ret = -1;
 317                 goto error;
 318         }
 319         ret = 0;
 320 error:
 321         return ret;
 322 }
 323
 324 static int relayd_add_stream_2_2(struct lttcomm_relayd_sock *rsock,
 325                 const char *channel_name, const char *pathname,
 326                 uint64_t tracefile_size, uint64_t tracefile_count)
 327 {
 328         int ret;
 329         struct lttcomm_relayd_add_stream_2_2 msg;
 330
 331         memset(&msg, 0, sizeof(msg));
 332         /* Compat with relayd 2.2 to 2.10 */
 333         if (lttng_strncpy(msg.channel_name, channel_name,
 334                                 sizeof(msg.channel_name))) {
 335                 ret = -1;
 336                 goto error;
 337         }
 338         if (lttng_strncpy(msg.pathname, pathname,
 339                                 sizeof(msg.pathname))) {
 340                 ret = -1;
 341                 goto error;
 342         }
 343         msg.tracefile_size = htobe64(tracefile_size);
 344         msg.tracefile_count = htobe64(tracefile_count);
 345
 346         /* Send command */
 347         ret = send_command(rsock, RELAYD_ADD_STREAM, (void *) &msg, sizeof(msg), 0);
 348         if (ret < 0) {
 349                 goto error;
 350         }
 351         ret = 0;
 352 error:
 353         return ret;
 354 }
 355
 356 static int relayd_add_stream_2_11(struct lttcomm_relayd_sock *rsock,
 357                 const char *channel_name, const char *pathname,
 358                 uint64_t tracefile_size, uint64_t tracefile_count,
 359                 uint64_t trace_archive_id)
 360 {
 361         int ret;
 362         struct lttcomm_relayd_add_stream_2_11 *msg = NULL;
 363         size_t channel_name_len;
 364         size_t pathname_len;
 365         size_t msg_length;
 366
 367         /* The two names are sent with a '\0' delimiter between them. */
 368         channel_name_len = strlen(channel_name) + 1;
 369         pathname_len = strlen(pathname) + 1;
 370
 371         msg_length = sizeof(*msg) + channel_name_len + pathname_len;
 372         msg = zmalloc(msg_length);
 373         if (!msg) {
 374                 PERROR("zmalloc add_stream_2_11 command message");
 375                 ret = -1;
 376                 goto error;
 377         }
 378
 379         assert(channel_name_len <= UINT32_MAX);
 380         msg->channel_name_len = htobe32(channel_name_len);
 381
 382         assert(pathname_len <= UINT32_MAX);
 383         msg->pathname_len = htobe32(pathname_len);
 384
 385         if (lttng_strncpy(msg->names, channel_name, channel_name_len)) {
 386                 ret = -1;
 387                 goto error;
 388         }
 389         if (lttng_strncpy(msg->names + channel_name_len, pathname, pathname_len)) {
 390                 ret = -1;
 391                 goto error;
 392         }
 393
 394         msg->tracefile_size = htobe64(tracefile_size);
 395         msg->tracefile_count = htobe64(tracefile_count);
 396         msg->trace_archive_id = htobe64(trace_archive_id);
 397
 398         /* Send command */
 399         ret = send_command(rsock, RELAYD_ADD_STREAM, (void *) msg, msg_length, 0);
 400         if (ret < 0) {
 401                 goto error;
 402         }
 403         ret = 0;
 404 error:
 405         free(msg);
 406         return ret;
 407 }
 408
 409 /*
 410  * Add stream on the relayd and assign stream handle to the stream_id argument.
 411  *
 412  * On success return 0 else return ret_code negative value.
 413  */
 414 int relayd_add_stream(struct lttcomm_relayd_sock *rsock, const char *channel_name,
 415                 const char *pathname, uint64_t *stream_id,
 416                 uint64_t tracefile_size, uint64_t tracefile_count,
 417                 uint64_t trace_archive_id)
 418 {
 419         int ret;
 420         struct lttcomm_relayd_status_stream reply;
 421
 422         /* Code flow error. Safety net. */
 423         assert(rsock);
 424         assert(channel_name);
 425         assert(pathname);
 426
 427         DBG("Relayd adding stream for channel name %s", channel_name);
 428
 429         /* Compat with relayd 2.1 */
 430         if (rsock->minor == 1) {
 431                 /* For 2.1 */
 432                 ret = relayd_add_stream_2_1(rsock, channel_name, pathname);
 433
 434         } else if (rsock->minor > 1 && rsock->minor < 11) {
 435                 /* From 2.2 to 2.10 */
 436                 ret = relayd_add_stream_2_2(rsock, channel_name, pathname,
 437                                 tracefile_size, tracefile_count);
 438         } else {
 439                 /* From 2.11 to ...*/
 440                 ret = relayd_add_stream_2_11(rsock, channel_name, pathname,
 441                                 tracefile_size, tracefile_count,
 442                                 trace_archive_id);
 443         }
 444
 445         if (ret) {
 446                 ret = -1;
 447                 goto error;
 448         }
 449
 450         /* Waiting for reply */
 451         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
 452         if (ret < 0) {
 453                 goto error;
 454         }
 455
 456         /* Back to host bytes order. */
 457         reply.handle = be64toh(reply.handle);
 458         reply.ret_code = be32toh(reply.ret_code);
 459
 460         /* Return session id or negative ret code. */
 461         if (reply.ret_code != LTTNG_OK) {
 462                 ret = -1;
 463                 ERR("Relayd add stream replied error %d", reply.ret_code);
 464         } else {
 465                 /* Success */
 466                 ret = 0;
 467                 *stream_id = reply.handle;
 468         }
 469
 470         DBG("Relayd stream added successfully with handle %" PRIu64,
 471                         reply.handle);
 472
 473 error:
 474         return ret;
 475 }
 476
 477 /*
 478  * Inform the relay that all the streams for the current channel has been sent.
 479  *
 480  * On success return 0 else return ret_code negative value.
 481  */
 482 int relayd_streams_sent(struct lttcomm_relayd_sock *rsock)
 483 {
 484         int ret;
 485         struct lttcomm_relayd_generic_reply reply;
 486
 487         /* Code flow error. Safety net. */
 488         assert(rsock);
 489
 490         DBG("Relayd sending streams sent.");
 491
 492         /* This feature was introduced in 2.4, ignore it for earlier versions. */
 493         if (rsock->minor < 4) {
 494                 ret = 0;
 495                 goto end;
 496         }
 497
 498         /* Send command */
 499         ret = send_command(rsock, RELAYD_STREAMS_SENT, NULL, 0, 0);
 500         if (ret < 0) {
 501                 goto error;
 502         }
 503
 504         /* Waiting for reply */
 505         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
 506         if (ret < 0) {
 507                 goto error;
 508         }
 509
 510         /* Back to host bytes order. */
 511         reply.ret_code = be32toh(reply.ret_code);
 512
 513         /* Return session id or negative ret code. */
 514         if (reply.ret_code != LTTNG_OK) {
 515                 ret = -1;
 516                 ERR("Relayd streams sent replied error %d", reply.ret_code);
 517                 goto error;
 518         } else {
 519                 /* Success */
 520                 ret = 0;
 521         }
 522
 523         DBG("Relayd streams sent success");
 524
 525 error:
 526 end:
 527         return ret;
 528 }
 529
 530 /*
 531  * Check version numbers on the relayd.
 532  * If major versions are compatible, we assign minor_to_use to the
 533  * minor version of the procotol we are going to use for this session.
 534  *
 535  * Return 0 if the two daemons are compatible, LTTNG_ERR_RELAYD_VERSION_FAIL
 536  * otherwise, or a negative value on network errors.
 537  */
 538 int relayd_version_check(struct lttcomm_relayd_sock *rsock)
 539 {
 540         int ret;
 541         struct lttcomm_relayd_version msg;
 542
 543         /* Code flow error. Safety net. */
 544         assert(rsock);
 545
 546         DBG("Relayd version check for major.minor %u.%u", rsock->major,
 547                         rsock->minor);
 548
 549         memset(&msg, 0, sizeof(msg));
 550         /* Prepare network byte order before transmission. */
 551         msg.major = htobe32(rsock->major);
 552         msg.minor = htobe32(rsock->minor);
 553
 554         /* Send command */
 555         ret = send_command(rsock, RELAYD_VERSION, (void *) &msg, sizeof(msg), 0);
 556         if (ret < 0) {
 557                 goto error;
 558         }
 559
 560         /* Receive response */
 561         ret = recv_reply(rsock, (void *) &msg, sizeof(msg));
 562         if (ret < 0) {
 563                 goto error;
 564         }
 565
 566         /* Set back to host bytes order */
 567         msg.major = be32toh(msg.major);
 568         msg.minor = be32toh(msg.minor);
 569
 570         /*
 571          * Only validate the major version. If the other side is higher,
 572          * communication is not possible. Only major version equal can talk to each
 573          * other. If the minor version differs, the lowest version is used by both
 574          * sides.
 575          */
 576         if (msg.major != rsock->major) {
 577                 /* Not compatible */
 578                 ret = LTTNG_ERR_RELAYD_VERSION_FAIL;
 579                 DBG2("Relayd version is NOT compatible. Relayd version %u != %u (us)",
 580                                 msg.major, rsock->major);
 581                 goto error;
 582         }
 583
 584         /*
 585          * If the relayd's minor version is higher, it will adapt to our version so
 586          * we can continue to use the latest relayd communication data structure.
 587          * If the received minor version is higher, the relayd should adapt to us.
 588          */
 589         if (rsock->minor > msg.minor) {
 590                 rsock->minor = msg.minor;
 591         }
 592
 593         /* Version number compatible */
 594         DBG2("Relayd version is compatible, using protocol version %u.%u",
 595                         rsock->major, rsock->minor);
 596         ret = 0;
 597
 598 error:
 599         return ret;
 600 }
 601
 602 /*
 603  * Add stream on the relayd and assign stream handle to the stream_id argument.
 604  *
 605  * On success return 0 else return ret_code negative value.
 606  */
 607 int relayd_send_metadata(struct lttcomm_relayd_sock *rsock, size_t len)
 608 {
 609         int ret;
 610
 611         /* Code flow error. Safety net. */
 612         assert(rsock);
 613
 614         DBG("Relayd sending metadata of size %zu", len);
 615
 616         /* Send command */
 617         ret = send_command(rsock, RELAYD_SEND_METADATA, NULL, len, 0);
 618         if (ret < 0) {
 619                 goto error;
 620         }
 621
 622         DBG2("Relayd metadata added successfully");
 623
 624         /*
 625          * After that call, the metadata data MUST be sent to the relayd so the
 626          * receive size on the other end matches the len of the metadata packet
 627          * header. This is why we don't wait for a reply here.
 628          */
 629
 630 error:
 631         return ret;
 632 }
 633
 634 /*
 635  * Connect to relay daemon with an allocated lttcomm_relayd_sock.
 636  */
 637 int relayd_connect(struct lttcomm_relayd_sock *rsock)
 638 {
 639         /* Code flow error. Safety net. */
 640         assert(rsock);
 641
 642         if (!rsock->sock.ops) {
 643                 /*
 644                  * Attempting a connect on a non-initialized socket.
 645                  */
 646                 return -ECONNRESET;
 647         }
 648
 649         DBG3("Relayd connect ...");
 650
 651         return rsock->sock.ops->connect(&rsock->sock);
 652 }
 653
 654 /*
 655  * Close relayd socket with an allocated lttcomm_relayd_sock.
 656  *
 657  * If no socket operations are found, simply return 0 meaning that everything
 658  * is fine. Without operations, the socket can not possibly be opened or used.
 659  * This is possible if the socket was allocated but not created. However, the
 660  * caller could simply use it to store a valid file descriptor for instance
 661  * passed over a Unix socket and call this to cleanup but still without a valid
 662  * ops pointer.
 663  *
 664  * Return the close returned value. On error, a negative value is usually
 665  * returned back from close(2).
 666  */
 667 int relayd_close(struct lttcomm_relayd_sock *rsock)
 668 {
 669         int ret;
 670
 671         /* Code flow error. Safety net. */
 672         assert(rsock);
 673
 674         /* An invalid fd is fine, return success. */
 675         if (rsock->sock.fd < 0) {
 676                 ret = 0;
 677                 goto end;
 678         }
 679
 680         DBG3("Relayd closing socket %d", rsock->sock.fd);
 681
 682         if (rsock->sock.ops) {
 683                 ret = rsock->sock.ops->close(&rsock->sock);
 684         } else {
 685                 /* Default call if no specific ops found. */
 686                 ret = close(rsock->sock.fd);
 687                 if (ret < 0) {
 688                         PERROR("relayd_close default close");
 689                 }
 690         }
 691         rsock->sock.fd = -1;
 692
 693 end:
 694         return ret;
 695 }
 696
 697 /*
 698  * Send data header structure to the relayd.
 699  */
 700 int relayd_send_data_hdr(struct lttcomm_relayd_sock *rsock,
 701                 struct lttcomm_relayd_data_hdr *hdr, size_t size)
 702 {
 703         int ret;
 704
 705         /* Code flow error. Safety net. */
 706         assert(rsock);
 707         assert(hdr);
 708
 709         if (rsock->sock.fd < 0) {
 710                 return -ECONNRESET;
 711         }
 712
 713         DBG3("Relayd sending data header of size %zu", size);
 714
 715         /* Again, safety net */
 716         if (size == 0) {
 717                 size = sizeof(struct lttcomm_relayd_data_hdr);
 718         }
 719
 720         /* Only send data header. */
 721         ret = rsock->sock.ops->sendmsg(&rsock->sock, hdr, size, 0);
 722         if (ret < 0) {
 723                 ret = -errno;
 724                 goto error;
 725         }
 726
 727         /*
 728          * The data MUST be sent right after that command for the receive on the
 729          * other end to match the size in the header.
 730          */
 731
 732 error:
 733         return ret;
 734 }
 735
 736 /*
 737  * Send close stream command to the relayd.
 738  */
 739 int relayd_send_close_stream(struct lttcomm_relayd_sock *rsock, uint64_t stream_id,
 740                 uint64_t last_net_seq_num)
 741 {
 742         int ret;
 743         struct lttcomm_relayd_close_stream msg;
 744         struct lttcomm_relayd_generic_reply reply;
 745
 746         /* Code flow error. Safety net. */
 747         assert(rsock);
 748
 749         DBG("Relayd closing stream id %" PRIu64, stream_id);
 750
 751         memset(&msg, 0, sizeof(msg));
 752         msg.stream_id = htobe64(stream_id);
 753         msg.last_net_seq_num = htobe64(last_net_seq_num);
 754
 755         /* Send command */
 756         ret = send_command(rsock, RELAYD_CLOSE_STREAM, (void *) &msg, sizeof(msg), 0);
 757         if (ret < 0) {
 758                 goto error;
 759         }
 760
 761         /* Receive response */
 762         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
 763         if (ret < 0) {
 764                 goto error;
 765         }
 766
 767         reply.ret_code = be32toh(reply.ret_code);
 768
 769         /* Return session id or negative ret code. */
 770         if (reply.ret_code != LTTNG_OK) {
 771                 ret = -1;
 772                 ERR("Relayd close stream replied error %d", reply.ret_code);
 773         } else {
 774                 /* Success */
 775                 ret = 0;
 776         }
 777
 778         DBG("Relayd close stream id %" PRIu64 " successfully", stream_id);
 779
 780 error:
 781         return ret;
 782 }
 783
 784 /*
 785  * Check for data availability for a given stream id.
 786  *
 787  * Return 0 if NOT pending, 1 if so and a negative value on error.
 788  */
 789 int relayd_data_pending(struct lttcomm_relayd_sock *rsock, uint64_t stream_id,
 790                 uint64_t last_net_seq_num)
 791 {
 792         int ret;
 793         struct lttcomm_relayd_data_pending msg;
 794         struct lttcomm_relayd_generic_reply reply;
 795
 796         /* Code flow error. Safety net. */
 797         assert(rsock);
 798
 799         DBG("Relayd data pending for stream id %" PRIu64, stream_id);
 800
 801         memset(&msg, 0, sizeof(msg));
 802         msg.stream_id = htobe64(stream_id);
 803         msg.last_net_seq_num = htobe64(last_net_seq_num);
 804
 805         /* Send command */
 806         ret = send_command(rsock, RELAYD_DATA_PENDING, (void *) &msg,
 807                         sizeof(msg), 0);
 808         if (ret < 0) {
 809                 goto error;
 810         }
 811
 812         /* Receive response */
 813         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
 814         if (ret < 0) {
 815                 goto error;
 816         }
 817
 818         reply.ret_code = be32toh(reply.ret_code);
 819
 820         /* Return session id or negative ret code. */
 821         if (reply.ret_code >= LTTNG_OK) {
 822                 ERR("Relayd data pending replied error %d", reply.ret_code);
 823         }
 824
 825         /* At this point, the ret code is either 1 or 0 */
 826         ret = reply.ret_code;
 827
 828         DBG("Relayd data is %s pending for stream id %" PRIu64,
 829                         ret == 1 ? "" : "NOT", stream_id);
 830
 831 error:
 832         return ret;
 833 }
 834
 835 /*
 836  * Check on the relayd side for a quiescent state on the control socket.
 837  */
 838 int relayd_quiescent_control(struct lttcomm_relayd_sock *rsock,
 839                 uint64_t metadata_stream_id)
 840 {
 841         int ret;
 842         struct lttcomm_relayd_quiescent_control msg;
 843         struct lttcomm_relayd_generic_reply reply;
 844
 845         /* Code flow error. Safety net. */
 846         assert(rsock);
 847
 848         DBG("Relayd checking quiescent control state");
 849
 850         memset(&msg, 0, sizeof(msg));
 851         msg.stream_id = htobe64(metadata_stream_id);
 852
 853         /* Send command */
 854         ret = send_command(rsock, RELAYD_QUIESCENT_CONTROL, &msg, sizeof(msg), 0);
 855         if (ret < 0) {
 856                 goto error;
 857         }
 858
 859         /* Receive response */
 860         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
 861         if (ret < 0) {
 862                 goto error;
 863         }
 864
 865         reply.ret_code = be32toh(reply.ret_code);
 866
 867         /* Return session id or negative ret code. */
 868         if (reply.ret_code != LTTNG_OK) {
 869                 ret = -1;
 870                 ERR("Relayd quiescent control replied error %d", reply.ret_code);
 871                 goto error;
 872         }
 873
 874         /* Control socket is quiescent */
 875         return 0;
 876
 877 error:
 878         return ret;
 879 }
 880
 881 /*
 882  * Begin a data pending command for a specific session id.
 883  */
 884 int relayd_begin_data_pending(struct lttcomm_relayd_sock *rsock, uint64_t id)
 885 {
 886         int ret;
 887         struct lttcomm_relayd_begin_data_pending msg;
 888         struct lttcomm_relayd_generic_reply reply;
 889
 890         /* Code flow error. Safety net. */
 891         assert(rsock);
 892
 893         DBG("Relayd begin data pending");
 894
 895         memset(&msg, 0, sizeof(msg));
 896         msg.session_id = htobe64(id);
 897
 898         /* Send command */
 899         ret = send_command(rsock, RELAYD_BEGIN_DATA_PENDING, &msg, sizeof(msg), 0);
 900         if (ret < 0) {
 901                 goto error;
 902         }
 903
 904         /* Receive response */
 905         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
 906         if (ret < 0) {
 907                 goto error;
 908         }
 909
 910         reply.ret_code = be32toh(reply.ret_code);
 911
 912         /* Return session id or negative ret code. */
 913         if (reply.ret_code != LTTNG_OK) {
 914                 ret = -1;
 915                 ERR("Relayd begin data pending replied error %d", reply.ret_code);
 916                 goto error;
 917         }
 918
 919         return 0;
 920
 921 error:
 922         return ret;
 923 }
 924
 925 /*
 926  * End a data pending command for a specific session id.
 927  *
 928  * Return 0 on success and set is_data_inflight to 0 if no data is being
 929  * streamed or 1 if it is the case.
 930  */
 931 int relayd_end_data_pending(struct lttcomm_relayd_sock *rsock, uint64_t id,
 932                 unsigned int *is_data_inflight)
 933 {
 934         int ret, recv_ret;
 935         struct lttcomm_relayd_end_data_pending msg;
 936         struct lttcomm_relayd_generic_reply reply;
 937
 938         /* Code flow error. Safety net. */
 939         assert(rsock);
 940
 941         DBG("Relayd end data pending");
 942
 943         memset(&msg, 0, sizeof(msg));
 944         msg.session_id = htobe64(id);
 945
 946         /* Send command */
 947         ret = send_command(rsock, RELAYD_END_DATA_PENDING, &msg, sizeof(msg), 0);
 948         if (ret < 0) {
 949                 goto error;
 950         }
 951
 952         /* Receive response */
 953         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
 954         if (ret < 0) {
 955                 goto error;
 956         }
 957
 958         recv_ret = be32toh(reply.ret_code);
 959         if (recv_ret < 0) {
 960                 ret = recv_ret;
 961                 goto error;
 962         }
 963
 964         *is_data_inflight = recv_ret;
 965
 966         DBG("Relayd end data pending is data inflight: %d", recv_ret);
 967
 968         return 0;
 969
 970 error:
 971         return ret;
 972 }
 973
 974 /*
 975  * Send index to the relayd.
 976  */
 977 int relayd_send_index(struct lttcomm_relayd_sock *rsock,
 978                 struct ctf_packet_index *index, uint64_t relay_stream_id,
 979                 uint64_t net_seq_num)
 980 {
 981         int ret;
 982         struct lttcomm_relayd_index msg;
 983         struct lttcomm_relayd_generic_reply reply;
 984
 985         /* Code flow error. Safety net. */
 986         assert(rsock);
 987
 988         if (rsock->minor < 4) {
 989                 DBG("Not sending indexes before protocol 2.4");
 990                 ret = 0;
 991                 goto error;
 992         }
 993
 994         DBG("Relayd sending index for stream ID %" PRIu64, relay_stream_id);
 995
 996         memset(&msg, 0, sizeof(msg));
 997         msg.relay_stream_id = htobe64(relay_stream_id);
 998         msg.net_seq_num = htobe64(net_seq_num);
 999
1000         /* The index is already in big endian. */
1001         msg.packet_size = index->packet_size;
1002         msg.content_size = index->content_size;
1003         msg.timestamp_begin = index->timestamp_begin;
1004         msg.timestamp_end = index->timestamp_end;
1005         msg.events_discarded = index->events_discarded;
1006         msg.stream_id = index->stream_id;
1007
1008         if (rsock->minor >= 8) {
1009                 msg.stream_instance_id = index->stream_instance_id;
1010                 msg.packet_seq_num = index->packet_seq_num;
1011         }
1012
1013         /* Send command */
1014         ret = send_command(rsock, RELAYD_SEND_INDEX, &msg,
1015                 lttcomm_relayd_index_len(lttng_to_index_major(rsock->major,
1016                                                                 rsock->minor),
1017                                 lttng_to_index_minor(rsock->major, rsock->minor)),
1018                                 0);
1019         if (ret < 0) {
1020                 goto error;
1021         }
1022
1023         /* Receive response */
1024         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
1025         if (ret < 0) {
1026                 goto error;
1027         }
1028
1029         reply.ret_code = be32toh(reply.ret_code);
1030
1031         /* Return session id or negative ret code. */
1032         if (reply.ret_code != LTTNG_OK) {
1033                 ret = -1;
1034                 ERR("Relayd send index replied error %d", reply.ret_code);
1035         } else {
1036                 /* Success */
1037                 ret = 0;
1038         }
1039
1040 error:
1041         return ret;
1042 }
1043
1044 /*
1045  * Ask the relay to reset the metadata trace file (regeneration).
1046  */
1047 int relayd_reset_metadata(struct lttcomm_relayd_sock *rsock,
1048                 uint64_t stream_id, uint64_t version)
1049 {
1050         int ret;
1051         struct lttcomm_relayd_reset_metadata msg;
1052         struct lttcomm_relayd_generic_reply reply;
1053
1054         /* Code flow error. Safety net. */
1055         assert(rsock);
1056
1057         /* Should have been prevented by the sessiond. */
1058         if (rsock->minor < 8) {
1059                 ERR("Metadata regeneration unsupported before 2.8");
1060                 ret = -1;
1061                 goto error;
1062         }
1063
1064         DBG("Relayd reset metadata stream id %" PRIu64, stream_id);
1065
1066         memset(&msg, 0, sizeof(msg));
1067         msg.stream_id = htobe64(stream_id);
1068         msg.version = htobe64(version);
1069
1070         /* Send command */
1071         ret = send_command(rsock, RELAYD_RESET_METADATA, (void *) &msg, sizeof(msg), 0);
1072         if (ret < 0) {
1073                 goto error;
1074         }
1075
1076         /* Receive response */
1077         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
1078         if (ret < 0) {
1079                 goto error;
1080         }
1081
1082         reply.ret_code = be32toh(reply.ret_code);
1083
1084         /* Return session id or negative ret code. */
1085         if (reply.ret_code != LTTNG_OK) {
1086                 ret = -1;
1087                 ERR("Relayd reset metadata replied error %d", reply.ret_code);
1088         } else {
1089                 /* Success */
1090                 ret = 0;
1091         }
1092
1093         DBG("Relayd reset metadata stream id %" PRIu64 " successfully", stream_id);
1094
1095 error:
1096         return ret;
1097 }
1098
1099 int relayd_rotate_stream(struct lttcomm_relayd_sock *rsock, uint64_t stream_id,
1100                 const char *new_pathname, uint64_t new_chunk_id,
1101                 uint64_t seq_num)
1102 {
1103         int ret;
1104         struct lttcomm_relayd_rotate_stream *msg = NULL;
1105         struct lttcomm_relayd_generic_reply reply;
1106         size_t len;
1107         int msg_len;
1108
1109         /* Code flow error. Safety net. */
1110         assert(rsock);
1111
1112         DBG("Sending rotate stream id %" PRIu64 " command to relayd", stream_id);
1113
1114         /* Account for the trailing NULL. */
1115         len = lttng_strnlen(new_pathname, LTTNG_PATH_MAX) + 1;
1116         if (len > LTTNG_PATH_MAX) {
1117                 ERR("Path used in relayd rotate stream command exceeds the maximal allowed length");
1118                 ret = -1;
1119                 goto error;
1120         }
1121
1122         msg_len = offsetof(struct lttcomm_relayd_rotate_stream, new_pathname) + len;
1123         msg = zmalloc(msg_len);
1124         if (!msg) {
1125                 PERROR("Failed to allocate relayd rotate stream command of %d bytes",
1126                                 msg_len);
1127                 ret = -1;
1128                 goto error;
1129         }
1130
1131         if (lttng_strncpy(msg->new_pathname, new_pathname, len)) {
1132                 ret = -1;
1133                 ERR("Failed to copy relayd rotate stream command's new path name");
1134                 goto error;
1135         }
1136
1137         msg->pathname_length = htobe32(len);
1138         msg->stream_id = htobe64(stream_id);
1139         msg->new_chunk_id = htobe64(new_chunk_id);
1140         /*
1141          * The seq_num is invalid for metadata streams, but it is ignored on
1142          * the relay.
1143          */
1144         msg->rotate_at_seq_num = htobe64(seq_num);
1145
1146         /* Send command. */
1147         ret = send_command(rsock, RELAYD_ROTATE_STREAM, (void *) msg, msg_len, 0);
1148         if (ret < 0) {
1149                 ERR("Send rotate command");
1150                 goto error;
1151         }
1152
1153         /* Receive response. */
1154         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
1155         if (ret < 0) {
1156                 ERR("Receive rotate reply");
1157                 goto error;
1158         }
1159
1160         reply.ret_code = be32toh(reply.ret_code);
1161
1162         /* Return session id or negative ret code. */
1163         if (reply.ret_code != LTTNG_OK) {
1164                 ret = -1;
1165                 ERR("Relayd rotate stream replied error %d", reply.ret_code);
1166         } else {
1167                 /* Success. */
1168                 ret = 0;
1169                 DBG("Relayd rotated stream id %" PRIu64 " successfully", stream_id);
1170         }
1171
1172 error:
1173         free(msg);
1174         return ret;
1175 }
1176
1177 int relayd_rotate_rename(struct lttcomm_relayd_sock *rsock,
1178                 const char *old_path, const char *new_path)
1179 {
1180         int ret;
1181         struct lttcomm_relayd_rotate_rename *msg = NULL;
1182         struct lttcomm_relayd_generic_reply reply;
1183         size_t old_path_length, new_path_length;
1184         size_t msg_length;
1185
1186         /* Code flow error. Safety net. */
1187         assert(rsock);
1188
1189         DBG("Relayd rename chunk %s to %s", old_path, new_path);
1190
1191         /* The two paths are sent with a '\0' delimiter between them. */
1192         old_path_length = strlen(old_path) + 1;
1193         new_path_length = strlen(new_path) + 1;
1194
1195         msg_length = sizeof(*msg) + old_path_length + new_path_length;
1196         msg = zmalloc(msg_length);
1197         if (!msg) {
1198                 PERROR("zmalloc rotate-rename command message");
1199                 ret = -1;
1200                 goto error;
1201         }
1202
1203         assert(old_path_length <= UINT32_MAX);
1204         msg->old_path_length = htobe32(old_path_length);
1205
1206         assert(new_path_length <= UINT32_MAX);
1207         msg->new_path_length = htobe32(new_path_length);
1208
1209         strcpy(msg->paths, old_path);
1210         strcpy(msg->paths + old_path_length, new_path);
1211
1212         /* Send command */
1213         ret = send_command(rsock, RELAYD_ROTATE_RENAME, (const void *) msg,
1214                         msg_length, 0);
1215         if (ret < 0) {
1216                 goto error;
1217         }
1218
1219         /* Receive response */
1220         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
1221         if (ret < 0) {
1222                 goto error;
1223         }
1224
1225         reply.ret_code = be32toh(reply.ret_code);
1226
1227         /* Return session id or negative ret code. */
1228         if (reply.ret_code != LTTNG_OK) {
1229                 ret = -1;
1230                 ERR("Relayd rotate rename replied error %d", reply.ret_code);
1231         } else {
1232                 /* Success */
1233                 ret = 0;
1234         }
1235
1236         DBG("Relayd rotate rename completed successfully");
1237
1238 error:
1239         free(msg);
1240         return ret;
1241 }
1242
1243 int relayd_rotate_pending(struct lttcomm_relayd_sock *rsock, uint64_t chunk_id)
1244 {
1245         int ret;
1246         struct lttcomm_relayd_rotate_pending msg;
1247         struct lttcomm_relayd_rotate_pending_reply reply;
1248
1249         /* Code flow error. Safety net. */
1250         assert(rsock);
1251
1252         DBG("Querying relayd for rotate pending with chunk_id %" PRIu64,
1253                         chunk_id);
1254
1255         memset(&msg, 0, sizeof(msg));
1256         msg.chunk_id = htobe64(chunk_id);
1257
1258         /* Send command */
1259         ret = send_command(rsock, RELAYD_ROTATE_PENDING, (void *) &msg,
1260                         sizeof(msg), 0);
1261         if (ret < 0) {
1262                 goto error;
1263         }
1264
1265         /* Receive response */
1266         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
1267         if (ret < 0) {
1268                 goto error;
1269         }
1270
1271         reply.generic.ret_code = be32toh(reply.generic.ret_code);
1272
1273         /* Return session id or negative ret code. */
1274         if (reply.generic.ret_code != LTTNG_OK) {
1275                 ret = -reply.generic.ret_code;
1276                 ERR("Relayd rotate pending replied with error %d", ret);
1277                 goto error;
1278         } else {
1279                 /* No error, just rotate pending state */
1280                 if (reply.is_pending == 0 || reply.is_pending == 1) {
1281                         ret = reply.is_pending;
1282                         DBG("Relayd rotate pending command completed successfully with result \"%s\"",
1283                                         ret ? "rotation pending" : "rotation NOT pending");
1284                 } else {
1285                         ret = -LTTNG_ERR_UNK;
1286                 }
1287         }
1288
1289 error:
1290         return ret;
1291 }
1292
1293 int relayd_mkdir(struct lttcomm_relayd_sock *rsock, const char *path)
1294 {
1295         int ret;
1296         struct lttcomm_relayd_mkdir *msg;
1297         struct lttcomm_relayd_generic_reply reply;
1298         size_t len;
1299
1300         /* Code flow error. Safety net. */
1301         assert(rsock);
1302
1303         DBG("Relayd mkdir path %s", path);
1304
1305         len = strlen(path) + 1;
1306         msg = zmalloc(sizeof(msg->length) + len);
1307         if (!msg) {
1308                 PERROR("Alloc mkdir msg");
1309                 ret = -1;
1310                 goto error;
1311         }
1312         msg->length = htobe32((uint32_t) len);
1313
1314         if (lttng_strncpy(msg->path, path, len)) {
1315                 ret = -1;
1316                 goto error;
1317         }
1318
1319         /* Send command */
1320         ret = send_command(rsock, RELAYD_MKDIR, (void *) msg,
1321                         sizeof(msg->length) + len, 0);
1322         if (ret < 0) {
1323                 goto error;
1324         }
1325
1326         /* Receive response */
1327         ret = recv_reply(rsock, (void *) &reply, sizeof(reply));
1328         if (ret < 0) {
1329                 goto error;
1330         }
1331
1332         reply.ret_code = be32toh(reply.ret_code);
1333
1334         /* Return session id or negative ret code. */
1335         if (reply.ret_code != LTTNG_OK) {
1336                 ret = -1;
1337                 ERR("Relayd mkdir replied error %d", reply.ret_code);
1338         } else {
1339                 /* Success */
1340                 ret = 0;
1341         }
1342
1343         DBG("Relayd mkdir completed successfully");
1344
1345 error:
1346         free(msg);
1347         return ret;
1348 }