src/common/kernel-consumer/kernel-consumer.cpp

   1 /*
   2  * Copyright (C) 2011 EfficiOS Inc.
   3  * Copyright (C) 2011 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
   4  * Copyright (C) 2017 Jérémie Galarneau <jeremie.galarneau@efficios.com>
   5  *
   6  * SPDX-License-Identifier: GPL-2.0-only
   7  *
   8  */
   9
  10 #define _LGPL_SOURCE
  11 #include <poll.h>
  12 #include <pthread.h>
  13 #include <stdlib.h>
  14 #include <string.h>
  15 #include <sys/mman.h>
  16 #include <sys/socket.h>
  17 #include <sys/types.h>
  18 #include <inttypes.h>
  19 #include <unistd.h>
  20 #include <sys/stat.h>
  21 #include <stdint.h>
  22
  23 #include <bin/lttng-consumerd/health-consumerd.hpp>
  24 #include <common/common.hpp>
  25 #include <common/kernel-ctl/kernel-ctl.hpp>
  26 #include <common/sessiond-comm/sessiond-comm.hpp>
  27 #include <common/sessiond-comm/relayd.hpp>
  28 #include <common/compat/fcntl.hpp>
  29 #include <common/compat/endian.hpp>
  30 #include <common/pipe.hpp>
  31 #include <common/relayd/relayd.hpp>
  32 #include <common/utils.hpp>
  33 #include <common/consumer/consumer-stream.hpp>
  34 #include <common/index/index.hpp>
  35 #include <common/consumer/consumer-timer.hpp>
  36 #include <common/optional.hpp>
  37 #include <common/buffer-view.hpp>
  38 #include <common/consumer/consumer.hpp>
  39 #include <common/consumer/metadata-bucket.hpp>
  40
  41 #include "kernel-consumer.hpp"
  42
  43 extern struct lttng_consumer_global_data the_consumer_data;
  44 extern int consumer_poll_timeout;
  45
  46 /*
  47  * Take a snapshot for a specific fd
  48  *
  49  * Returns 0 on success, < 0 on error
  50  */
  51 int lttng_kconsumer_take_snapshot(struct lttng_consumer_stream *stream)
  52 {
  53         int ret = 0;
  54         int infd = stream->wait_fd;
  55
  56         ret = kernctl_snapshot(infd);
  57         /*
  58          * -EAGAIN is not an error, it just means that there is no data to
  59          *  be read.
  60          */
  61         if (ret != 0 && ret != -EAGAIN) {
  62                 PERROR("Getting sub-buffer snapshot.");
  63         }
  64
  65         return ret;
  66 }
  67
  68 /*
  69  * Sample consumed and produced positions for a specific fd.
  70  *
  71  * Returns 0 on success, < 0 on error.
  72  */
  73 int lttng_kconsumer_sample_snapshot_positions(
  74                 struct lttng_consumer_stream *stream)
  75 {
  76         LTTNG_ASSERT(stream);
  77
  78         return kernctl_snapshot_sample_positions(stream->wait_fd);
  79 }
  80
  81 /*
  82  * Get the produced position
  83  *
  84  * Returns 0 on success, < 0 on error
  85  */
  86 int lttng_kconsumer_get_produced_snapshot(struct lttng_consumer_stream *stream,
  87                 unsigned long *pos)
  88 {
  89         int ret;
  90         int infd = stream->wait_fd;
  91
  92         ret = kernctl_snapshot_get_produced(infd, pos);
  93         if (ret != 0) {
  94                 PERROR("kernctl_snapshot_get_produced");
  95         }
  96
  97         return ret;
  98 }
  99
 100 /*
 101  * Get the consumerd position
 102  *
 103  * Returns 0 on success, < 0 on error
 104  */
 105 int lttng_kconsumer_get_consumed_snapshot(struct lttng_consumer_stream *stream,
 106                 unsigned long *pos)
 107 {
 108         int ret;
 109         int infd = stream->wait_fd;
 110
 111         ret = kernctl_snapshot_get_consumed(infd, pos);
 112         if (ret != 0) {
 113                 PERROR("kernctl_snapshot_get_consumed");
 114         }
 115
 116         return ret;
 117 }
 118
 119 static
 120 int get_current_subbuf_addr(struct lttng_consumer_stream *stream,
 121                 const char **addr)
 122 {
 123         int ret;
 124         unsigned long mmap_offset;
 125         const char *mmap_base = (const char *) stream->mmap_base;
 126
 127         ret = kernctl_get_mmap_read_offset(stream->wait_fd, &mmap_offset);
 128         if (ret < 0) {
 129                 PERROR("Failed to get mmap read offset");
 130                 goto error;
 131         }
 132
 133         *addr = mmap_base + mmap_offset;
 134 error:
 135         return ret;
 136 }
 137
 138 /*
 139  * Take a snapshot of all the stream of a channel
 140  * RCU read-side lock must be held across this function to ensure existence of
 141  * channel.
 142  *
 143  * Returns 0 on success, < 0 on error
 144  */
 145 static int lttng_kconsumer_snapshot_channel(
 146                 struct lttng_consumer_channel *channel,
 147                 uint64_t key, char *path, uint64_t relayd_id,
 148                 uint64_t nb_packets_per_stream)
 149 {
 150         int ret;
 151         struct lttng_consumer_stream *stream;
 152
 153         DBG("Kernel consumer snapshot channel %" PRIu64, key);
 154
 155         /* Prevent channel modifications while we perform the snapshot.*/
 156         pthread_mutex_lock(&channel->lock);
 157
 158         rcu_read_lock();
 159
 160         /* Splice is not supported yet for channel snapshot. */
 161         if (channel->output != CONSUMER_CHANNEL_MMAP) {
 162                 ERR("Unsupported output type for channel \"%s\": mmap output is required to record a snapshot",
 163                                 channel->name);
 164                 ret = -1;
 165                 goto end;
 166         }
 167
 168         cds_list_for_each_entry(stream, &channel->streams.head, send_node) {
 169                 unsigned long consumed_pos, produced_pos;
 170
 171                 health_code_update();
 172
 173                 /*
 174                  * Lock stream because we are about to change its state.
 175                  */
 176                 pthread_mutex_lock(&stream->lock);
 177
 178                 LTTNG_ASSERT(channel->trace_chunk);
 179                 if (!lttng_trace_chunk_get(channel->trace_chunk)) {
 180                         /*
 181                          * Can't happen barring an internal error as the channel
 182                          * holds a reference to the trace chunk.
 183                          */
 184                         ERR("Failed to acquire reference to channel's trace chunk");
 185                         ret = -1;
 186                         goto end_unlock;
 187                 }
 188                 LTTNG_ASSERT(!stream->trace_chunk);
 189                 stream->trace_chunk = channel->trace_chunk;
 190
 191                 /*
 192                  * Assign the received relayd ID so we can use it for streaming. The streams
 193                  * are not visible to anyone so this is OK to change it.
 194                  */
 195                 stream->net_seq_idx = relayd_id;
 196                 channel->relayd_id = relayd_id;
 197                 if (relayd_id != (uint64_t) -1ULL) {
 198                         ret = consumer_send_relayd_stream(stream, path);
 199                         if (ret < 0) {
 200                                 ERR("sending stream to relayd");
 201                                 goto end_unlock;
 202                         }
 203                 } else {
 204                         ret = consumer_stream_create_output_files(stream,
 205                                         false);
 206                         if (ret < 0) {
 207                                 goto end_unlock;
 208                         }
 209                         DBG("Kernel consumer snapshot stream (%" PRIu64 ")",
 210                                         stream->key);
 211                 }
 212
 213                 ret = kernctl_buffer_flush_empty(stream->wait_fd);
 214                 if (ret < 0) {
 215                         /*
 216                          * Doing a buffer flush which does not take into
 217                          * account empty packets. This is not perfect
 218                          * for stream intersection, but required as a
 219                          * fall-back when "flush_empty" is not
 220                          * implemented by lttng-modules.
 221                          */
 222                         ret = kernctl_buffer_flush(stream->wait_fd);
 223                         if (ret < 0) {
 224                                 ERR("Failed to flush kernel stream");
 225                                 goto end_unlock;
 226                         }
 227                         goto end_unlock;
 228                 }
 229
 230                 ret = lttng_kconsumer_take_snapshot(stream);
 231                 if (ret < 0) {
 232                         ERR("Taking kernel snapshot");
 233                         goto end_unlock;
 234                 }
 235
 236                 ret = lttng_kconsumer_get_produced_snapshot(stream, &produced_pos);
 237                 if (ret < 0) {
 238                         ERR("Produced kernel snapshot position");
 239                         goto end_unlock;
 240                 }
 241
 242                 ret = lttng_kconsumer_get_consumed_snapshot(stream, &consumed_pos);
 243                 if (ret < 0) {
 244                         ERR("Consumerd kernel snapshot position");
 245                         goto end_unlock;
 246                 }
 247
 248                 consumed_pos = consumer_get_consume_start_pos(consumed_pos,
 249                                 produced_pos, nb_packets_per_stream,
 250                                 stream->max_sb_size);
 251
 252                 while ((long) (consumed_pos - produced_pos) < 0) {
 253                         ssize_t read_len;
 254                         unsigned long len, padded_len;
 255                         const char *subbuf_addr;
 256                         struct lttng_buffer_view subbuf_view;
 257
 258                         health_code_update();
 259                         DBG("Kernel consumer taking snapshot at pos %lu", consumed_pos);
 260
 261                         ret = kernctl_get_subbuf(stream->wait_fd, &consumed_pos);
 262                         if (ret < 0) {
 263                                 if (ret != -EAGAIN) {
 264                                         PERROR("kernctl_get_subbuf snapshot");
 265                                         goto end_unlock;
 266                                 }
 267                                 DBG("Kernel consumer get subbuf failed. Skipping it.");
 268                                 consumed_pos += stream->max_sb_size;
 269                                 stream->chan->lost_packets++;
 270                                 continue;
 271                         }
 272
 273                         ret = kernctl_get_subbuf_size(stream->wait_fd, &len);
 274                         if (ret < 0) {
 275                                 ERR("Snapshot kernctl_get_subbuf_size");
 276                                 goto error_put_subbuf;
 277                         }
 278
 279                         ret = kernctl_get_padded_subbuf_size(stream->wait_fd, &padded_len);
 280                         if (ret < 0) {
 281                                 ERR("Snapshot kernctl_get_padded_subbuf_size");
 282                                 goto error_put_subbuf;
 283                         }
 284
 285                         ret = get_current_subbuf_addr(stream, &subbuf_addr);
 286                         if (ret) {
 287                                 goto error_put_subbuf;
 288                         }
 289
 290                         subbuf_view = lttng_buffer_view_init(
 291                                         subbuf_addr, 0, padded_len);
 292                         read_len = lttng_consumer_on_read_subbuffer_mmap(
 293                                         stream, &subbuf_view,
 294                                         padded_len - len);
 295                         /*
 296                          * We write the padded len in local tracefiles but the data len
 297                          * when using a relay. Display the error but continue processing
 298                          * to try to release the subbuffer.
 299                          */
 300                         if (relayd_id != (uint64_t) -1ULL) {
 301                                 if (read_len != len) {
 302                                         ERR("Error sending to the relay (ret: %zd != len: %lu)",
 303                                                         read_len, len);
 304                                 }
 305                         } else {
 306                                 if (read_len != padded_len) {
 307                                         ERR("Error writing to tracefile (ret: %zd != len: %lu)",
 308                                                         read_len, padded_len);
 309                                 }
 310                         }
 311
 312                         ret = kernctl_put_subbuf(stream->wait_fd);
 313                         if (ret < 0) {
 314                                 ERR("Snapshot kernctl_put_subbuf");
 315                                 goto end_unlock;
 316                         }
 317                         consumed_pos += stream->max_sb_size;
 318                 }
 319
 320                 if (relayd_id == (uint64_t) -1ULL) {
 321                         if (stream->out_fd >= 0) {
 322                                 ret = close(stream->out_fd);
 323                                 if (ret < 0) {
 324                                         PERROR("Kernel consumer snapshot close out_fd");
 325                                         goto end_unlock;
 326                                 }
 327                                 stream->out_fd = -1;
 328                         }
 329                 } else {
 330                         close_relayd_stream(stream);
 331                         stream->net_seq_idx = (uint64_t) -1ULL;
 332                 }
 333                 lttng_trace_chunk_put(stream->trace_chunk);
 334                 stream->trace_chunk = NULL;
 335                 pthread_mutex_unlock(&stream->lock);
 336         }
 337
 338         /* All good! */
 339         ret = 0;
 340         goto end;
 341
 342 error_put_subbuf:
 343         ret = kernctl_put_subbuf(stream->wait_fd);
 344         if (ret < 0) {
 345                 ERR("Snapshot kernctl_put_subbuf error path");
 346         }
 347 end_unlock:
 348         pthread_mutex_unlock(&stream->lock);
 349 end:
 350         rcu_read_unlock();
 351         pthread_mutex_unlock(&channel->lock);
 352         return ret;
 353 }
 354
 355 /*
 356  * Read the whole metadata available for a snapshot.
 357  * RCU read-side lock must be held across this function to ensure existence of
 358  * metadata_channel.
 359  *
 360  * Returns 0 on success, < 0 on error
 361  */
 362 static int lttng_kconsumer_snapshot_metadata(
 363                 struct lttng_consumer_channel *metadata_channel,
 364                 uint64_t key, char *path, uint64_t relayd_id,
 365                 struct lttng_consumer_local_data *ctx)
 366 {
 367         int ret, use_relayd = 0;
 368         ssize_t ret_read;
 369         struct lttng_consumer_stream *metadata_stream;
 370
 371         LTTNG_ASSERT(ctx);
 372
 373         DBG("Kernel consumer snapshot metadata with key %" PRIu64 " at path %s",
 374                         key, path);
 375
 376         rcu_read_lock();
 377
 378         metadata_stream = metadata_channel->metadata_stream;
 379         LTTNG_ASSERT(metadata_stream);
 380
 381         metadata_stream->read_subbuffer_ops.lock(metadata_stream);
 382         LTTNG_ASSERT(metadata_channel->trace_chunk);
 383         LTTNG_ASSERT(metadata_stream->trace_chunk);
 384
 385         /* Flag once that we have a valid relayd for the stream. */
 386         if (relayd_id != (uint64_t) -1ULL) {
 387                 use_relayd = 1;
 388         }
 389
 390         if (use_relayd) {
 391                 ret = consumer_send_relayd_stream(metadata_stream, path);
 392                 if (ret < 0) {
 393                         goto error_snapshot;
 394                 }
 395         } else {
 396                 ret = consumer_stream_create_output_files(metadata_stream,
 397                                 false);
 398                 if (ret < 0) {
 399                         goto error_snapshot;
 400                 }
 401         }
 402
 403         do {
 404                 health_code_update();
 405
 406                 ret_read = lttng_consumer_read_subbuffer(metadata_stream, ctx, true);
 407                 if (ret_read < 0) {
 408                         ERR("Kernel snapshot reading metadata subbuffer (ret: %zd)",
 409                                         ret_read);
 410                         ret = ret_read;
 411                         goto error_snapshot;
 412                 }
 413         } while (ret_read > 0);
 414
 415         if (use_relayd) {
 416                 close_relayd_stream(metadata_stream);
 417                 metadata_stream->net_seq_idx = (uint64_t) -1ULL;
 418         } else {
 419                 if (metadata_stream->out_fd >= 0) {
 420                         ret = close(metadata_stream->out_fd);
 421                         if (ret < 0) {
 422                                 PERROR("Kernel consumer snapshot metadata close out_fd");
 423                                 /*
 424                                  * Don't go on error here since the snapshot was successful at this
 425                                  * point but somehow the close failed.
 426                                  */
 427                         }
 428                         metadata_stream->out_fd = -1;
 429                         lttng_trace_chunk_put(metadata_stream->trace_chunk);
 430                         metadata_stream->trace_chunk = NULL;
 431                 }
 432         }
 433
 434         ret = 0;
 435 error_snapshot:
 436         metadata_stream->read_subbuffer_ops.unlock(metadata_stream);
 437         consumer_stream_destroy(metadata_stream, NULL);
 438         metadata_channel->metadata_stream = NULL;
 439         rcu_read_unlock();
 440         return ret;
 441 }
 442
 443 /*
 444  * Receive command from session daemon and process it.
 445  *
 446  * Return 1 on success else a negative value or 0.
 447  */
 448 int lttng_kconsumer_recv_cmd(struct lttng_consumer_local_data *ctx,
 449                 int sock, struct pollfd *consumer_sockpoll)
 450 {
 451         int ret_func;
 452         enum lttcomm_return_code ret_code = LTTCOMM_CONSUMERD_SUCCESS;
 453         struct lttcomm_consumer_msg msg;
 454
 455         health_code_update();
 456
 457         {
 458                 ssize_t ret_recv;
 459
 460                 ret_recv = lttcomm_recv_unix_sock(sock, &msg, sizeof(msg));
 461                 if (ret_recv != sizeof(msg)) {
 462                         if (ret_recv > 0) {
 463                                 lttng_consumer_send_error(ctx,
 464                                                 LTTCOMM_CONSUMERD_ERROR_RECV_CMD);
 465                                 ret_recv = -1;
 466                         }
 467                         return ret_recv;
 468                 }
 469         }
 470
 471         health_code_update();
 472
 473         /* Deprecated command */
 474         LTTNG_ASSERT(msg.cmd_type != LTTNG_CONSUMER_STOP);
 475
 476         health_code_update();
 477
 478         /* relayd needs RCU read-side protection */
 479         rcu_read_lock();
 480
 481         switch (msg.cmd_type) {
 482         case LTTNG_CONSUMER_ADD_RELAYD_SOCKET:
 483         {
 484                 uint32_t major = msg.u.relayd_sock.major;
 485                 uint32_t minor = msg.u.relayd_sock.minor;
 486                 enum lttcomm_sock_proto protocol = (enum lttcomm_sock_proto)
 487                                 msg.u.relayd_sock.relayd_socket_protocol;
 488
 489                 /* Session daemon status message are handled in the following call. */
 490                 consumer_add_relayd_socket(msg.u.relayd_sock.net_index,
 491                                 msg.u.relayd_sock.type, ctx, sock,
 492                                 consumer_sockpoll, msg.u.relayd_sock.session_id,
 493                                 msg.u.relayd_sock.relayd_session_id, major,
 494                                 minor, protocol);
 495                 goto end_nosignal;
 496         }
 497         case LTTNG_CONSUMER_ADD_CHANNEL:
 498         {
 499                 struct lttng_consumer_channel *new_channel;
 500                 int ret_send_status, ret_add_channel = 0;
 501                 const uint64_t chunk_id = msg.u.channel.chunk_id.value;
 502
 503                 health_code_update();
 504
 505                 /* First send a status message before receiving the fds. */
 506                 ret_send_status = consumer_send_status_msg(sock, ret_code);
 507                 if (ret_send_status < 0) {
 508                         /* Somehow, the session daemon is not responding anymore. */
 509                         goto error_fatal;
 510                 }
 511
 512                 health_code_update();
 513
 514                 DBG("consumer_add_channel %" PRIu64, msg.u.channel.channel_key);
 515                 new_channel = consumer_allocate_channel(msg.u.channel.channel_key,
 516                                 msg.u.channel.session_id,
 517                                 msg.u.channel.chunk_id.is_set ?
 518                                                 &chunk_id : NULL,
 519                                 msg.u.channel.pathname,
 520                                 msg.u.channel.name,
 521                                 msg.u.channel.relayd_id, msg.u.channel.output,
 522                                 msg.u.channel.tracefile_size,
 523                                 msg.u.channel.tracefile_count, 0,
 524                                 msg.u.channel.monitor,
 525                                 msg.u.channel.live_timer_interval,
 526                                 msg.u.channel.is_live,
 527                                 NULL, NULL);
 528                 if (new_channel == NULL) {
 529                         lttng_consumer_send_error(ctx, LTTCOMM_CONSUMERD_OUTFD_ERROR);
 530                         goto end_nosignal;
 531                 }
 532                 new_channel->nb_init_stream_left = msg.u.channel.nb_init_streams;
 533                 switch (msg.u.channel.output) {
 534                 case LTTNG_EVENT_SPLICE:
 535                         new_channel->output = CONSUMER_CHANNEL_SPLICE;
 536                         break;
 537                 case LTTNG_EVENT_MMAP:
 538                         new_channel->output = CONSUMER_CHANNEL_MMAP;
 539                         break;
 540                 default:
 541                         ERR("Channel output unknown %d", msg.u.channel.output);
 542                         goto end_nosignal;
 543                 }
 544
 545                 /* Translate and save channel type. */
 546                 switch (msg.u.channel.type) {
 547                 case CONSUMER_CHANNEL_TYPE_DATA:
 548                 case CONSUMER_CHANNEL_TYPE_METADATA:
 549                         new_channel->type = (consumer_channel_type) msg.u.channel.type;
 550                         break;
 551                 default:
 552                         abort();
 553                         goto end_nosignal;
 554                 };
 555
 556                 health_code_update();
 557
 558                 if (ctx->on_recv_channel != NULL) {
 559                         int ret_recv_channel =
 560                                         ctx->on_recv_channel(new_channel);
 561                         if (ret_recv_channel == 0) {
 562                                 ret_add_channel = consumer_add_channel(
 563                                                 new_channel, ctx);
 564                         } else if (ret_recv_channel < 0) {
 565                                 goto end_nosignal;
 566                         }
 567                 } else {
 568                         ret_add_channel =
 569                                         consumer_add_channel(new_channel, ctx);
 570                 }
 571                 if (msg.u.channel.type == CONSUMER_CHANNEL_TYPE_DATA &&
 572                                 !ret_add_channel) {
 573                         int monitor_start_ret;
 574
 575                         DBG("Consumer starting monitor timer");
 576                         consumer_timer_live_start(new_channel,
 577                                         msg.u.channel.live_timer_interval);
 578                         monitor_start_ret = consumer_timer_monitor_start(
 579                                         new_channel,
 580                                         msg.u.channel.monitor_timer_interval);
 581                         if (monitor_start_ret < 0) {
 582                                 ERR("Starting channel monitoring timer failed");
 583                                 goto end_nosignal;
 584                         }
 585                 }
 586
 587                 health_code_update();
 588
 589                 /* If we received an error in add_channel, we need to report it. */
 590                 if (ret_add_channel < 0) {
 591                         ret_send_status = consumer_send_status_msg(
 592                                         sock, ret_add_channel);
 593                         if (ret_send_status < 0) {
 594                                 goto error_fatal;
 595                         }
 596                         goto end_nosignal;
 597                 }
 598
 599                 goto end_nosignal;
 600         }
 601         case LTTNG_CONSUMER_ADD_STREAM:
 602         {
 603                 int fd;
 604                 struct lttng_pipe *stream_pipe;
 605                 struct lttng_consumer_stream *new_stream;
 606                 struct lttng_consumer_channel *channel;
 607                 int alloc_ret = 0;
 608                 int ret_send_status, ret_poll, ret_get_max_subbuf_size;
 609                 ssize_t ret_pipe_write, ret_recv;
 610
 611                 /*
 612                  * Get stream's channel reference. Needed when adding the stream to the
 613                  * global hash table.
 614                  */
 615                 channel = consumer_find_channel(msg.u.stream.channel_key);
 616                 if (!channel) {
 617                         /*
 618                          * We could not find the channel. Can happen if cpu hotplug
 619                          * happens while tearing down.
 620                          */
 621                         ERR("Unable to find channel key %" PRIu64, msg.u.stream.channel_key);
 622                         ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND;
 623                 }
 624
 625                 health_code_update();
 626
 627                 /* First send a status message before receiving the fds. */
 628                 ret_send_status = consumer_send_status_msg(sock, ret_code);
 629                 if (ret_send_status < 0) {
 630                         /* Somehow, the session daemon is not responding anymore. */
 631                         goto error_add_stream_fatal;
 632                 }
 633
 634                 health_code_update();
 635
 636                 if (ret_code != LTTCOMM_CONSUMERD_SUCCESS) {
 637                         /* Channel was not found. */
 638                         goto error_add_stream_nosignal;
 639                 }
 640
 641                 /* Blocking call */
 642                 health_poll_entry();
 643                 ret_poll = lttng_consumer_poll_socket(consumer_sockpoll);
 644                 health_poll_exit();
 645                 if (ret_poll) {
 646                         goto error_add_stream_fatal;
 647                 }
 648
 649                 health_code_update();
 650
 651                 /* Get stream file descriptor from socket */
 652                 ret_recv = lttcomm_recv_fds_unix_sock(sock, &fd, 1);
 653                 if (ret_recv != sizeof(fd)) {
 654                         lttng_consumer_send_error(ctx, LTTCOMM_CONSUMERD_ERROR_RECV_FD);
 655                         ret_func = ret_recv;
 656                         goto end;
 657                 }
 658
 659                 health_code_update();
 660
 661                 /*
 662                  * Send status code to session daemon only if the recv works. If the
 663                  * above recv() failed, the session daemon is notified through the
 664                  * error socket and the teardown is eventually done.
 665                  */
 666                 ret_send_status = consumer_send_status_msg(sock, ret_code);
 667                 if (ret_send_status < 0) {
 668                         /* Somehow, the session daemon is not responding anymore. */
 669                         goto error_add_stream_nosignal;
 670                 }
 671
 672                 health_code_update();
 673
 674                 pthread_mutex_lock(&channel->lock);
 675                 new_stream = consumer_stream_create(
 676                                 channel,
 677                                 channel->key,
 678                                 fd,
 679                                 channel->name,
 680                                 channel->relayd_id,
 681                                 channel->session_id,
 682                                 channel->trace_chunk,
 683                                 msg.u.stream.cpu,
 684                                 &alloc_ret,
 685                                 channel->type,
 686                                 channel->monitor);
 687                 if (new_stream == NULL) {
 688                         switch (alloc_ret) {
 689                         case -ENOMEM:
 690                         case -EINVAL:
 691                         default:
 692                                 lttng_consumer_send_error(ctx, LTTCOMM_CONSUMERD_OUTFD_ERROR);
 693                                 break;
 694                         }
 695                         pthread_mutex_unlock(&channel->lock);
 696                         goto error_add_stream_nosignal;
 697                 }
 698
 699                 new_stream->wait_fd = fd;
 700                 ret_get_max_subbuf_size = kernctl_get_max_subbuf_size(
 701                                 new_stream->wait_fd, &new_stream->max_sb_size);
 702                 if (ret_get_max_subbuf_size < 0) {
 703                         pthread_mutex_unlock(&channel->lock);
 704                         ERR("Failed to get kernel maximal subbuffer size");
 705                         goto error_add_stream_nosignal;
 706                 }
 707
 708                 consumer_stream_update_channel_attributes(new_stream,
 709                                 channel);
 710
 711                 /*
 712                  * We've just assigned the channel to the stream so increment the
 713                  * refcount right now. We don't need to increment the refcount for
 714                  * streams in no monitor because we handle manually the cleanup of
 715                  * those. It is very important to make sure there is NO prior
 716                  * consumer_del_stream() calls or else the refcount will be unbalanced.
 717                  */
 718                 if (channel->monitor) {
 719                         uatomic_inc(&new_stream->chan->refcount);
 720                 }
 721
 722                 /*
 723                  * The buffer flush is done on the session daemon side for the kernel
 724                  * so no need for the stream "hangup_flush_done" variable to be
 725                  * tracked. This is important for a kernel stream since we don't rely
 726                  * on the flush state of the stream to read data. It's not the case for
 727                  * user space tracing.
 728                  */
 729                 new_stream->hangup_flush_done = 0;
 730
 731                 health_code_update();
 732
 733                 pthread_mutex_lock(&new_stream->lock);
 734                 if (ctx->on_recv_stream) {
 735                         int ret_recv_stream = ctx->on_recv_stream(new_stream);
 736                         if (ret_recv_stream < 0) {
 737                                 pthread_mutex_unlock(&new_stream->lock);
 738                                 pthread_mutex_unlock(&channel->lock);
 739                                 consumer_stream_free(new_stream);
 740                                 goto error_add_stream_nosignal;
 741                         }
 742                 }
 743                 health_code_update();
 744
 745                 if (new_stream->metadata_flag) {
 746                         channel->metadata_stream = new_stream;
 747                 }
 748
 749                 /* Do not monitor this stream. */
 750                 if (!channel->monitor) {
 751                         DBG("Kernel consumer add stream %s in no monitor mode with "
 752                                         "relayd id %" PRIu64, new_stream->name,
 753                                         new_stream->net_seq_idx);
 754                         cds_list_add(&new_stream->send_node, &channel->streams.head);
 755                         pthread_mutex_unlock(&new_stream->lock);
 756                         pthread_mutex_unlock(&channel->lock);
 757                         goto end_add_stream;
 758                 }
 759
 760                 /* Send stream to relayd if the stream has an ID. */
 761                 if (new_stream->net_seq_idx != (uint64_t) -1ULL) {
 762                         int ret_send_relayd_stream;
 763
 764                         ret_send_relayd_stream = consumer_send_relayd_stream(
 765                                         new_stream, new_stream->chan->pathname);
 766                         if (ret_send_relayd_stream < 0) {
 767                                 pthread_mutex_unlock(&new_stream->lock);
 768                                 pthread_mutex_unlock(&channel->lock);
 769                                 consumer_stream_free(new_stream);
 770                                 goto error_add_stream_nosignal;
 771                         }
 772
 773                         /*
 774                          * If adding an extra stream to an already
 775                          * existing channel (e.g. cpu hotplug), we need
 776                          * to send the "streams_sent" command to relayd.
 777                          */
 778                         if (channel->streams_sent_to_relayd) {
 779                                 int ret_send_relayd_streams_sent;
 780
 781                                 ret_send_relayd_streams_sent =
 782                                                 consumer_send_relayd_streams_sent(
 783                                                                 new_stream->net_seq_idx);
 784                                 if (ret_send_relayd_streams_sent < 0) {
 785                                         pthread_mutex_unlock(&new_stream->lock);
 786                                         pthread_mutex_unlock(&channel->lock);
 787                                         goto error_add_stream_nosignal;
 788                                 }
 789                         }
 790                 }
 791                 pthread_mutex_unlock(&new_stream->lock);
 792                 pthread_mutex_unlock(&channel->lock);
 793
 794                 /* Get the right pipe where the stream will be sent. */
 795                 if (new_stream->metadata_flag) {
 796                         consumer_add_metadata_stream(new_stream);
 797                         stream_pipe = ctx->consumer_metadata_pipe;
 798                 } else {
 799                         consumer_add_data_stream(new_stream);
 800                         stream_pipe = ctx->consumer_data_pipe;
 801                 }
 802
 803                 /* Visible to other threads */
 804                 new_stream->globally_visible = 1;
 805
 806                 health_code_update();
 807
 808                 ret_pipe_write = lttng_pipe_write(
 809                                 stream_pipe, &new_stream, sizeof(new_stream));
 810                 if (ret_pipe_write < 0) {
 811                         ERR("Consumer write %s stream to pipe %d",
 812                                         new_stream->metadata_flag ? "metadata" : "data",
 813                                         lttng_pipe_get_writefd(stream_pipe));
 814                         if (new_stream->metadata_flag) {
 815                                 consumer_del_stream_for_metadata(new_stream);
 816                         } else {
 817                                 consumer_del_stream_for_data(new_stream);
 818                         }
 819                         goto error_add_stream_nosignal;
 820                 }
 821
 822                 DBG("Kernel consumer ADD_STREAM %s (fd: %d) %s with relayd id %" PRIu64,
 823                                 new_stream->name, fd, new_stream->chan->pathname, new_stream->relayd_stream_id);
 824 end_add_stream:
 825                 break;
 826 error_add_stream_nosignal:
 827                 goto end_nosignal;
 828 error_add_stream_fatal:
 829                 goto error_fatal;
 830         }
 831         case LTTNG_CONSUMER_STREAMS_SENT:
 832         {
 833                 struct lttng_consumer_channel *channel;
 834                 int ret_send_status;
 835
 836                 /*
 837                  * Get stream's channel reference. Needed when adding the stream to the
 838                  * global hash table.
 839                  */
 840                 channel = consumer_find_channel(msg.u.sent_streams.channel_key);
 841                 if (!channel) {
 842                         /*
 843                          * We could not find the channel. Can happen if cpu hotplug
 844                          * happens while tearing down.
 845                          */
 846                         ERR("Unable to find channel key %" PRIu64,
 847                                         msg.u.sent_streams.channel_key);
 848                         ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND;
 849                 }
 850
 851                 health_code_update();
 852
 853                 /*
 854                  * Send status code to session daemon.
 855                  */
 856                 ret_send_status = consumer_send_status_msg(sock, ret_code);
 857                 if (ret_send_status < 0 ||
 858                                 ret_code != LTTCOMM_CONSUMERD_SUCCESS) {
 859                         /* Somehow, the session daemon is not responding anymore. */
 860                         goto error_streams_sent_nosignal;
 861                 }
 862
 863                 health_code_update();
 864
 865                 /*
 866                  * We should not send this message if we don't monitor the
 867                  * streams in this channel.
 868                  */
 869                 if (!channel->monitor) {
 870                         goto end_error_streams_sent;
 871                 }
 872
 873                 health_code_update();
 874                 /* Send stream to relayd if the stream has an ID. */
 875                 if (msg.u.sent_streams.net_seq_idx != (uint64_t) -1ULL) {
 876                         int ret_send_relay_streams;
 877
 878                         ret_send_relay_streams = consumer_send_relayd_streams_sent(
 879                                         msg.u.sent_streams.net_seq_idx);
 880                         if (ret_send_relay_streams < 0) {
 881                                 goto error_streams_sent_nosignal;
 882                         }
 883                         channel->streams_sent_to_relayd = true;
 884                 }
 885 end_error_streams_sent:
 886                 break;
 887 error_streams_sent_nosignal:
 888                 goto end_nosignal;
 889         }
 890         case LTTNG_CONSUMER_UPDATE_STREAM:
 891         {
 892                 rcu_read_unlock();
 893                 return -ENOSYS;
 894         }
 895         case LTTNG_CONSUMER_DESTROY_RELAYD:
 896         {
 897                 uint64_t index = msg.u.destroy_relayd.net_seq_idx;
 898                 struct consumer_relayd_sock_pair *relayd;
 899                 int ret_send_status;
 900
 901                 DBG("Kernel consumer destroying relayd %" PRIu64, index);
 902
 903                 /* Get relayd reference if exists. */
 904                 relayd = consumer_find_relayd(index);
 905                 if (relayd == NULL) {
 906                         DBG("Unable to find relayd %" PRIu64, index);
 907                         ret_code = LTTCOMM_CONSUMERD_RELAYD_FAIL;
 908                 }
 909
 910                 /*
 911                  * Each relayd socket pair has a refcount of stream attached to it
 912                  * which tells if the relayd is still active or not depending on the
 913                  * refcount value.
 914                  *
 915                  * This will set the destroy flag of the relayd object and destroy it
 916                  * if the refcount reaches zero when called.
 917                  *
 918                  * The destroy can happen either here or when a stream fd hangs up.
 919                  */
 920                 if (relayd) {
 921                         consumer_flag_relayd_for_destroy(relayd);
 922                 }
 923
 924                 health_code_update();
 925
 926                 ret_send_status = consumer_send_status_msg(sock, ret_code);
 927                 if (ret_send_status < 0) {
 928                         /* Somehow, the session daemon is not responding anymore. */
 929                         goto error_fatal;
 930                 }
 931
 932                 goto end_nosignal;
 933         }
 934         case LTTNG_CONSUMER_DATA_PENDING:
 935         {
 936                 int32_t ret_data_pending;
 937                 uint64_t id = msg.u.data_pending.session_id;
 938                 ssize_t ret_send;
 939
 940                 DBG("Kernel consumer data pending command for id %" PRIu64, id);
 941
 942                 ret_data_pending = consumer_data_pending(id);
 943
 944                 health_code_update();
 945
 946                 /* Send back returned value to session daemon */
 947                 ret_send = lttcomm_send_unix_sock(sock, &ret_data_pending,
 948                                 sizeof(ret_data_pending));
 949                 if (ret_send < 0) {
 950                         PERROR("send data pending ret code");
 951                         goto error_fatal;
 952                 }
 953
 954                 /*
 955                  * No need to send back a status message since the data pending
 956                  * returned value is the response.
 957                  */
 958                 break;
 959         }
 960         case LTTNG_CONSUMER_SNAPSHOT_CHANNEL:
 961         {
 962                 struct lttng_consumer_channel *channel;
 963                 uint64_t key = msg.u.snapshot_channel.key;
 964                 int ret_send_status;
 965
 966                 channel = consumer_find_channel(key);
 967                 if (!channel) {
 968                         ERR("Channel %" PRIu64 " not found", key);
 969                         ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND;
 970                 } else {
 971                         if (msg.u.snapshot_channel.metadata == 1) {
 972                                 int ret_snapshot;
 973
 974                                 ret_snapshot = lttng_kconsumer_snapshot_metadata(
 975                                                 channel, key,
 976                                                 msg.u.snapshot_channel.pathname,
 977                                                 msg.u.snapshot_channel.relayd_id,
 978                                                 ctx);
 979                                 if (ret_snapshot < 0) {
 980                                         ERR("Snapshot metadata failed");
 981                                         ret_code = LTTCOMM_CONSUMERD_SNAPSHOT_FAILED;
 982                                 }
 983                         } else {
 984                                 int ret_snapshot;
 985
 986                                 ret_snapshot = lttng_kconsumer_snapshot_channel(
 987                                                 channel, key,
 988                                                 msg.u.snapshot_channel.pathname,
 989                                                 msg.u.snapshot_channel.relayd_id,
 990                                                 msg.u.snapshot_channel
 991                                                                 .nb_packets_per_stream);
 992                                 if (ret_snapshot < 0) {
 993                                         ERR("Snapshot channel failed");
 994                                         ret_code = LTTCOMM_CONSUMERD_SNAPSHOT_FAILED;
 995                                 }
 996                         }
 997                 }
 998                 health_code_update();
 999
1000                 ret_send_status = consumer_send_status_msg(sock, ret_code);
1001                 if (ret_send_status < 0) {
1002                         /* Somehow, the session daemon is not responding anymore. */
1003                         goto end_nosignal;
1004                 }
1005                 break;
1006         }
1007         case LTTNG_CONSUMER_DESTROY_CHANNEL:
1008         {
1009                 uint64_t key = msg.u.destroy_channel.key;
1010                 struct lttng_consumer_channel *channel;
1011                 int ret_send_status;
1012
1013                 channel = consumer_find_channel(key);
1014                 if (!channel) {
1015                         ERR("Kernel consumer destroy channel %" PRIu64 " not found", key);
1016                         ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND;
1017                 }
1018
1019                 health_code_update();
1020
1021                 ret_send_status = consumer_send_status_msg(sock, ret_code);
1022                 if (ret_send_status < 0) {
1023                         /* Somehow, the session daemon is not responding anymore. */
1024                         goto end_destroy_channel;
1025                 }
1026
1027                 health_code_update();
1028
1029                 /* Stop right now if no channel was found. */
1030                 if (!channel) {
1031                         goto end_destroy_channel;
1032                 }
1033
1034                 /*
1035                  * This command should ONLY be issued for channel with streams set in
1036                  * no monitor mode.
1037                  */
1038                 LTTNG_ASSERT(!channel->monitor);
1039
1040                 /*
1041                  * The refcount should ALWAYS be 0 in the case of a channel in no
1042                  * monitor mode.
1043                  */
1044                 LTTNG_ASSERT(!uatomic_sub_return(&channel->refcount, 1));
1045
1046                 consumer_del_channel(channel);
1047 end_destroy_channel:
1048                 goto end_nosignal;
1049         }
1050         case LTTNG_CONSUMER_DISCARDED_EVENTS:
1051         {
1052                 ssize_t ret;
1053                 uint64_t count;
1054                 struct lttng_consumer_channel *channel;
1055                 uint64_t id = msg.u.discarded_events.session_id;
1056                 uint64_t key = msg.u.discarded_events.channel_key;
1057
1058                 DBG("Kernel consumer discarded events command for session id %"
1059                                 PRIu64 ", channel key %" PRIu64, id, key);
1060
1061                 channel = consumer_find_channel(key);
1062                 if (!channel) {
1063                         ERR("Kernel consumer discarded events channel %"
1064                                         PRIu64 " not found", key);
1065                         count = 0;
1066                 } else {
1067                         count = channel->discarded_events;
1068                 }
1069
1070                 health_code_update();
1071
1072                 /* Send back returned value to session daemon */
1073                 ret = lttcomm_send_unix_sock(sock, &count, sizeof(count));
1074                 if (ret < 0) {
1075                         PERROR("send discarded events");
1076                         goto error_fatal;
1077                 }
1078
1079                 break;
1080         }
1081         case LTTNG_CONSUMER_LOST_PACKETS:
1082         {
1083                 ssize_t ret;
1084                 uint64_t count;
1085                 struct lttng_consumer_channel *channel;
1086                 uint64_t id = msg.u.lost_packets.session_id;
1087                 uint64_t key = msg.u.lost_packets.channel_key;
1088
1089                 DBG("Kernel consumer lost packets command for session id %"
1090                                 PRIu64 ", channel key %" PRIu64, id, key);
1091
1092                 channel = consumer_find_channel(key);
1093                 if (!channel) {
1094                         ERR("Kernel consumer lost packets channel %"
1095                                         PRIu64 " not found", key);
1096                         count = 0;
1097                 } else {
1098                         count = channel->lost_packets;
1099                 }
1100
1101                 health_code_update();
1102
1103                 /* Send back returned value to session daemon */
1104                 ret = lttcomm_send_unix_sock(sock, &count, sizeof(count));
1105                 if (ret < 0) {
1106                         PERROR("send lost packets");
1107                         goto error_fatal;
1108                 }
1109
1110                 break;
1111         }
1112         case LTTNG_CONSUMER_SET_CHANNEL_MONITOR_PIPE:
1113         {
1114                 int channel_monitor_pipe;
1115                 int ret_send_status, ret_set_channel_monitor_pipe;
1116                 ssize_t ret_recv;
1117
1118                 ret_code = LTTCOMM_CONSUMERD_SUCCESS;
1119                 /* Successfully received the command's type. */
1120                 ret_send_status = consumer_send_status_msg(sock, ret_code);
1121                 if (ret_send_status < 0) {
1122                         goto error_fatal;
1123                 }
1124
1125                 ret_recv = lttcomm_recv_fds_unix_sock(
1126                                 sock, &channel_monitor_pipe, 1);
1127                 if (ret_recv != sizeof(channel_monitor_pipe)) {
1128                         ERR("Failed to receive channel monitor pipe");
1129                         goto error_fatal;
1130                 }
1131
1132                 DBG("Received channel monitor pipe (%d)", channel_monitor_pipe);
1133                 ret_set_channel_monitor_pipe =
1134                                 consumer_timer_thread_set_channel_monitor_pipe(
1135                                                 channel_monitor_pipe);
1136                 if (!ret_set_channel_monitor_pipe) {
1137                         int flags;
1138                         int ret_fcntl;
1139
1140                         ret_code = LTTCOMM_CONSUMERD_SUCCESS;
1141                         /* Set the pipe as non-blocking. */
1142                         ret_fcntl = fcntl(channel_monitor_pipe, F_GETFL, 0);
1143                         if (ret_fcntl == -1) {
1144                                 PERROR("fcntl get flags of the channel monitoring pipe");
1145                                 goto error_fatal;
1146                         }
1147                         flags = ret_fcntl;
1148
1149                         ret_fcntl = fcntl(channel_monitor_pipe, F_SETFL,
1150                                         flags | O_NONBLOCK);
1151                         if (ret_fcntl == -1) {
1152                                 PERROR("fcntl set O_NONBLOCK flag of the channel monitoring pipe");
1153                                 goto error_fatal;
1154                         }
1155                         DBG("Channel monitor pipe set as non-blocking");
1156                 } else {
1157                         ret_code = LTTCOMM_CONSUMERD_ALREADY_SET;
1158                 }
1159                 ret_send_status = consumer_send_status_msg(sock, ret_code);
1160                 if (ret_send_status < 0) {
1161                         goto error_fatal;
1162                 }
1163                 break;
1164         }
1165         case LTTNG_CONSUMER_ROTATE_CHANNEL:
1166         {
1167                 struct lttng_consumer_channel *channel;
1168                 uint64_t key = msg.u.rotate_channel.key;
1169                 int ret_send_status;
1170
1171                 DBG("Consumer rotate channel %" PRIu64, key);
1172
1173                 channel = consumer_find_channel(key);
1174                 if (!channel) {
1175                         ERR("Channel %" PRIu64 " not found", key);
1176                         ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND;
1177                 } else {
1178                         /*
1179                          * Sample the rotate position of all the streams in this channel.
1180                          */
1181                         int ret_rotate_channel;
1182
1183                         ret_rotate_channel = lttng_consumer_rotate_channel(
1184                                         channel, key,
1185                                         msg.u.rotate_channel.relayd_id);
1186                         if (ret_rotate_channel < 0) {
1187                                 ERR("Rotate channel failed");
1188                                 ret_code = LTTCOMM_CONSUMERD_ROTATION_FAIL;
1189                         }
1190
1191                         health_code_update();
1192                 }
1193
1194                 ret_send_status = consumer_send_status_msg(sock, ret_code);
1195                 if (ret_send_status < 0) {
1196                         /* Somehow, the session daemon is not responding anymore. */
1197                         goto error_rotate_channel;
1198                 }
1199                 if (channel) {
1200                         /* Rotate the streams that are ready right now. */
1201                         int ret_rotate;
1202
1203                         ret_rotate = lttng_consumer_rotate_ready_streams(
1204                                         channel, key);
1205                         if (ret_rotate < 0) {
1206                                 ERR("Rotate ready streams failed");
1207                         }
1208                 }
1209                 break;
1210 error_rotate_channel:
1211                 goto end_nosignal;
1212         }
1213         case LTTNG_CONSUMER_CLEAR_CHANNEL:
1214         {
1215                 struct lttng_consumer_channel *channel;
1216                 uint64_t key = msg.u.clear_channel.key;
1217                 int ret_send_status;
1218
1219                 channel = consumer_find_channel(key);
1220                 if (!channel) {
1221                         DBG("Channel %" PRIu64 " not found", key);
1222                         ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND;
1223                 } else {
1224                         int ret_clear_channel;
1225
1226                         ret_clear_channel =
1227                                         lttng_consumer_clear_channel(channel);
1228                         if (ret_clear_channel) {
1229                                 ERR("Clear channel failed");
1230                                 ret_code = (lttcomm_return_code) ret_clear_channel;
1231                         }
1232
1233                         health_code_update();
1234                 }
1235
1236                 ret_send_status = consumer_send_status_msg(sock, ret_code);
1237                 if (ret_send_status < 0) {
1238                         /* Somehow, the session daemon is not responding anymore. */
1239                         goto end_nosignal;
1240                 }
1241
1242                 break;
1243         }
1244         case LTTNG_CONSUMER_INIT:
1245         {
1246                 int ret_send_status;
1247                 lttng_uuid sessiond_uuid;
1248
1249                 std::copy(std::begin(msg.u.init.sessiond_uuid), std::end(msg.u.init.sessiond_uuid),
1250                                 sessiond_uuid.begin());
1251
1252                 ret_code = lttng_consumer_init_command(ctx,
1253                                 sessiond_uuid);
1254                 health_code_update();
1255                 ret_send_status = consumer_send_status_msg(sock, ret_code);
1256                 if (ret_send_status < 0) {
1257                         /* Somehow, the session daemon is not responding anymore. */
1258                         goto end_nosignal;
1259                 }
1260                 break;
1261         }
1262         case LTTNG_CONSUMER_CREATE_TRACE_CHUNK:
1263         {
1264                 const struct lttng_credentials credentials = {
1265                         .uid = LTTNG_OPTIONAL_INIT_VALUE(msg.u.create_trace_chunk.credentials.value.uid),
1266                         .gid = LTTNG_OPTIONAL_INIT_VALUE(msg.u.create_trace_chunk.credentials.value.gid),
1267                 };
1268                 const bool is_local_trace =
1269                                 !msg.u.create_trace_chunk.relayd_id.is_set;
1270                 const uint64_t relayd_id =
1271                                 msg.u.create_trace_chunk.relayd_id.value;
1272                 const char *chunk_override_name =
1273                                 *msg.u.create_trace_chunk.override_name ?
1274                                         msg.u.create_trace_chunk.override_name :
1275                                         NULL;
1276                 struct lttng_directory_handle *chunk_directory_handle = NULL;
1277
1278                 /*
1279                  * The session daemon will only provide a chunk directory file
1280                  * descriptor for local traces.
1281                  */
1282                 if (is_local_trace) {
1283                         int chunk_dirfd;
1284                         int ret_send_status;
1285                         ssize_t ret_recv;
1286
1287                         /* Acnowledge the reception of the command. */
1288                         ret_send_status = consumer_send_status_msg(
1289                                         sock, LTTCOMM_CONSUMERD_SUCCESS);
1290                         if (ret_send_status < 0) {
1291                                 /* Somehow, the session daemon is not responding anymore. */
1292                                 goto end_nosignal;
1293                         }
1294
1295                         ret_recv = lttcomm_recv_fds_unix_sock(
1296                                         sock, &chunk_dirfd, 1);
1297                         if (ret_recv != sizeof(chunk_dirfd)) {
1298                                 ERR("Failed to receive trace chunk directory file descriptor");
1299                                 goto error_fatal;
1300                         }
1301
1302                         DBG("Received trace chunk directory fd (%d)",
1303                                         chunk_dirfd);
1304                         chunk_directory_handle = lttng_directory_handle_create_from_dirfd(
1305                                         chunk_dirfd);
1306                         if (!chunk_directory_handle) {
1307                                 ERR("Failed to initialize chunk directory handle from directory file descriptor");
1308                                 if (close(chunk_dirfd)) {
1309                                         PERROR("Failed to close chunk directory file descriptor");
1310                                 }
1311                                 goto error_fatal;
1312                         }
1313                 }
1314
1315                 ret_code = lttng_consumer_create_trace_chunk(
1316                                 !is_local_trace ? &relayd_id : NULL,
1317                                 msg.u.create_trace_chunk.session_id,
1318                                 msg.u.create_trace_chunk.chunk_id,
1319                                 (time_t) msg.u.create_trace_chunk
1320                                                 .creation_timestamp,
1321                                 chunk_override_name,
1322                                 msg.u.create_trace_chunk.credentials.is_set ?
1323                                                 &credentials :
1324                                                 NULL,
1325                                 chunk_directory_handle);
1326                 lttng_directory_handle_put(chunk_directory_handle);
1327                 goto end_msg_sessiond;
1328         }
1329         case LTTNG_CONSUMER_CLOSE_TRACE_CHUNK:
1330         {
1331                 enum lttng_trace_chunk_command_type close_command =
1332                                 (lttng_trace_chunk_command_type) msg.u.close_trace_chunk.close_command.value;
1333                 const uint64_t relayd_id =
1334                                 msg.u.close_trace_chunk.relayd_id.value;
1335                 struct lttcomm_consumer_close_trace_chunk_reply reply;
1336                 char path[LTTNG_PATH_MAX];
1337                 ssize_t ret_send;
1338
1339                 ret_code = lttng_consumer_close_trace_chunk(
1340                                 msg.u.close_trace_chunk.relayd_id.is_set ?
1341                                                 &relayd_id :
1342                                                 NULL,
1343                                 msg.u.close_trace_chunk.session_id,
1344                                 msg.u.close_trace_chunk.chunk_id,
1345                                 (time_t) msg.u.close_trace_chunk.close_timestamp,
1346                                 msg.u.close_trace_chunk.close_command.is_set ?
1347                                                 &close_command :
1348                                                 NULL, path);
1349                 reply.ret_code = ret_code;
1350                 reply.path_length = strlen(path) + 1;
1351                 ret_send = lttcomm_send_unix_sock(sock, &reply, sizeof(reply));
1352                 if (ret_send != sizeof(reply)) {
1353                         goto error_fatal;
1354                 }
1355                 ret_send = lttcomm_send_unix_sock(
1356                                 sock, path, reply.path_length);
1357                 if (ret_send != reply.path_length) {
1358                         goto error_fatal;
1359                 }
1360                 goto end_nosignal;
1361         }
1362         case LTTNG_CONSUMER_TRACE_CHUNK_EXISTS:
1363         {
1364                 const uint64_t relayd_id =
1365                                 msg.u.trace_chunk_exists.relayd_id.value;
1366
1367                 ret_code = lttng_consumer_trace_chunk_exists(
1368                                 msg.u.trace_chunk_exists.relayd_id.is_set ?
1369                                                 &relayd_id : NULL,
1370                                 msg.u.trace_chunk_exists.session_id,
1371                                 msg.u.trace_chunk_exists.chunk_id);
1372                 goto end_msg_sessiond;
1373         }
1374         case LTTNG_CONSUMER_OPEN_CHANNEL_PACKETS:
1375         {
1376                 const uint64_t key = msg.u.open_channel_packets.key;
1377                 struct lttng_consumer_channel *channel =
1378                                 consumer_find_channel(key);
1379
1380                 if (channel) {
1381                         pthread_mutex_lock(&channel->lock);
1382                         ret_code = lttng_consumer_open_channel_packets(channel);
1383                         pthread_mutex_unlock(&channel->lock);
1384                 } else {
1385                         WARN("Channel %" PRIu64 " not found", key);
1386                         ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND;
1387                 }
1388
1389                 health_code_update();
1390                 goto end_msg_sessiond;
1391         }
1392         default:
1393                 goto end_nosignal;
1394         }
1395
1396 end_nosignal:
1397         /*
1398          * Return 1 to indicate success since the 0 value can be a socket
1399          * shutdown during the recv() or send() call.
1400          */
1401         ret_func = 1;
1402         goto end;
1403 error_fatal:
1404         /* This will issue a consumer stop. */
1405         ret_func = -1;
1406         goto end;
1407 end_msg_sessiond:
1408         /*
1409          * The returned value here is not useful since either way we'll return 1 to
1410          * the caller because the session daemon socket management is done
1411          * elsewhere. Returning a negative code or 0 will shutdown the consumer.
1412          */
1413         {
1414                 int ret_send_status;
1415
1416                 ret_send_status = consumer_send_status_msg(sock, ret_code);
1417                 if (ret_send_status < 0) {
1418                         goto error_fatal;
1419                 }
1420         }
1421
1422         ret_func = 1;
1423
1424 end:
1425         health_code_update();
1426         rcu_read_unlock();
1427         return ret_func;
1428 }
1429
1430 /*
1431  * Sync metadata meaning request them to the session daemon and snapshot to the
1432  * metadata thread can consumer them.
1433  *
1434  * Metadata stream lock MUST be acquired.
1435  */
1436 enum sync_metadata_status lttng_kconsumer_sync_metadata(
1437                 struct lttng_consumer_stream *metadata)
1438 {
1439         int ret;
1440         enum sync_metadata_status status;
1441
1442         LTTNG_ASSERT(metadata);
1443
1444         ret = kernctl_buffer_flush(metadata->wait_fd);
1445         if (ret < 0) {
1446                 ERR("Failed to flush kernel stream");
1447                 status = SYNC_METADATA_STATUS_ERROR;
1448                 goto end;
1449         }
1450
1451         ret = kernctl_snapshot(metadata->wait_fd);
1452         if (ret < 0) {
1453                 if (errno == EAGAIN) {
1454                         /* No new metadata, exit. */
1455                         DBG("Sync metadata, no new kernel metadata");
1456                         status = SYNC_METADATA_STATUS_NO_DATA;
1457                 } else {
1458                         ERR("Sync metadata, taking kernel snapshot failed.");
1459                         status = SYNC_METADATA_STATUS_ERROR;
1460                 }
1461         } else {
1462                 status = SYNC_METADATA_STATUS_NEW_DATA;
1463         }
1464
1465 end:
1466         return status;
1467 }
1468
1469 static
1470 int extract_common_subbuffer_info(struct lttng_consumer_stream *stream,
1471                 struct stream_subbuffer *subbuf)
1472 {
1473         int ret;
1474
1475         ret = kernctl_get_subbuf_size(
1476                         stream->wait_fd, &subbuf->info.data.subbuf_size);
1477         if (ret) {
1478                 goto end;
1479         }
1480
1481         ret = kernctl_get_padded_subbuf_size(
1482                         stream->wait_fd, &subbuf->info.data.padded_subbuf_size);
1483         if (ret) {
1484                 goto end;
1485         }
1486
1487 end:
1488         return ret;
1489 }
1490
1491 static
1492 int extract_metadata_subbuffer_info(struct lttng_consumer_stream *stream,
1493                 struct stream_subbuffer *subbuf)
1494 {
1495         int ret;
1496
1497         ret = extract_common_subbuffer_info(stream, subbuf);
1498         if (ret) {
1499                 goto end;
1500         }
1501
1502         ret = kernctl_get_metadata_version(
1503                         stream->wait_fd, &subbuf->info.metadata.version);
1504         if (ret) {
1505                 goto end;
1506         }
1507
1508 end:
1509         return ret;
1510 }
1511
1512 static
1513 int extract_data_subbuffer_info(struct lttng_consumer_stream *stream,
1514                 struct stream_subbuffer *subbuf)
1515 {
1516         int ret;
1517
1518         ret = extract_common_subbuffer_info(stream, subbuf);
1519         if (ret) {
1520                 goto end;
1521         }
1522
1523         ret = kernctl_get_packet_size(
1524                         stream->wait_fd, &subbuf->info.data.packet_size);
1525         if (ret < 0) {
1526                 PERROR("Failed to get sub-buffer packet size");
1527                 goto end;
1528         }
1529
1530         ret = kernctl_get_content_size(
1531                         stream->wait_fd, &subbuf->info.data.content_size);
1532         if (ret < 0) {
1533                 PERROR("Failed to get sub-buffer content size");
1534                 goto end;
1535         }
1536
1537         ret = kernctl_get_timestamp_begin(
1538                         stream->wait_fd, &subbuf->info.data.timestamp_begin);
1539         if (ret < 0) {
1540                 PERROR("Failed to get sub-buffer begin timestamp");
1541                 goto end;
1542         }
1543
1544         ret = kernctl_get_timestamp_end(
1545                         stream->wait_fd, &subbuf->info.data.timestamp_end);
1546         if (ret < 0) {
1547                 PERROR("Failed to get sub-buffer end timestamp");
1548                 goto end;
1549         }
1550
1551         ret = kernctl_get_events_discarded(
1552                         stream->wait_fd, &subbuf->info.data.events_discarded);
1553         if (ret) {
1554                 PERROR("Failed to get sub-buffer events discarded count");
1555                 goto end;
1556         }
1557
1558         ret = kernctl_get_sequence_number(stream->wait_fd,
1559                         &subbuf->info.data.sequence_number.value);
1560         if (ret) {
1561                 /* May not be supported by older LTTng-modules. */
1562                 if (ret != -ENOTTY) {
1563                         PERROR("Failed to get sub-buffer sequence number");
1564                         goto end;
1565                 }
1566         } else {
1567                 subbuf->info.data.sequence_number.is_set = true;
1568         }
1569
1570         ret = kernctl_get_stream_id(
1571                         stream->wait_fd, &subbuf->info.data.stream_id);
1572         if (ret < 0) {
1573                 PERROR("Failed to get stream id");
1574                 goto end;
1575         }
1576
1577         ret = kernctl_get_instance_id(stream->wait_fd,
1578                         &subbuf->info.data.stream_instance_id.value);
1579         if (ret) {
1580                 /* May not be supported by older LTTng-modules. */
1581                 if (ret != -ENOTTY) {
1582                         PERROR("Failed to get stream instance id");
1583                         goto end;
1584                 }
1585         } else {
1586                 subbuf->info.data.stream_instance_id.is_set = true;
1587         }
1588 end:
1589         return ret;
1590 }
1591
1592 static
1593 enum get_next_subbuffer_status get_subbuffer_common(
1594                 struct lttng_consumer_stream *stream,
1595                 struct stream_subbuffer *subbuffer)
1596 {
1597         int ret;
1598         enum get_next_subbuffer_status status;
1599
1600         ret = kernctl_get_next_subbuf(stream->wait_fd);
1601         switch (ret) {
1602         case 0:
1603                 status = GET_NEXT_SUBBUFFER_STATUS_OK;
1604                 break;
1605         case -ENODATA:
1606                 case -EAGAIN:
1607                 /*
1608                  * The caller only expects -ENODATA when there is no data to
1609                  * read, but the kernel tracer returns -EAGAIN when there is
1610                  * currently no data for a non-finalized stream, and -ENODATA
1611                  * when there is no data for a finalized stream. Those can be
1612                  * combined into a -ENODATA return value.
1613                  */
1614                 status = GET_NEXT_SUBBUFFER_STATUS_NO_DATA;
1615                 goto end;
1616         default:
1617                 status = GET_NEXT_SUBBUFFER_STATUS_ERROR;
1618                 goto end;
1619         }
1620
1621         ret = stream->read_subbuffer_ops.extract_subbuffer_info(
1622                 stream, subbuffer);
1623         if (ret) {
1624                 status = GET_NEXT_SUBBUFFER_STATUS_ERROR;
1625         }
1626 end:
1627         return status;
1628 }
1629
1630 static
1631 enum get_next_subbuffer_status get_next_subbuffer_splice(
1632                 struct lttng_consumer_stream *stream,
1633                 struct stream_subbuffer *subbuffer)
1634 {
1635         const enum get_next_subbuffer_status status =
1636                         get_subbuffer_common(stream, subbuffer);
1637
1638         if (status != GET_NEXT_SUBBUFFER_STATUS_OK) {
1639                 goto end;
1640         }
1641
1642         subbuffer->buffer.fd = stream->wait_fd;
1643 end:
1644         return status;
1645 }
1646
1647 static
1648 enum get_next_subbuffer_status get_next_subbuffer_mmap(
1649                 struct lttng_consumer_stream *stream,
1650                 struct stream_subbuffer *subbuffer)
1651 {
1652         int ret;
1653         enum get_next_subbuffer_status status;
1654         const char *addr;
1655
1656         status = get_subbuffer_common(stream, subbuffer);
1657         if (status != GET_NEXT_SUBBUFFER_STATUS_OK) {
1658                 goto end;
1659         }
1660
1661         ret = get_current_subbuf_addr(stream, &addr);
1662         if (ret) {
1663                 status = GET_NEXT_SUBBUFFER_STATUS_ERROR;
1664                 goto end;
1665         }
1666
1667         subbuffer->buffer.buffer = lttng_buffer_view_init(
1668                         addr, 0, subbuffer->info.data.padded_subbuf_size);
1669 end:
1670         return status;
1671 }
1672
1673 static
1674 enum get_next_subbuffer_status get_next_subbuffer_metadata_check(struct lttng_consumer_stream *stream,
1675                 struct stream_subbuffer *subbuffer)
1676 {
1677         int ret;
1678         const char *addr;
1679         bool coherent;
1680         enum get_next_subbuffer_status status;
1681
1682         ret = kernctl_get_next_subbuf_metadata_check(stream->wait_fd,
1683                         &coherent);
1684         if (ret) {
1685                 goto end;
1686         }
1687
1688         ret = stream->read_subbuffer_ops.extract_subbuffer_info(
1689                         stream, subbuffer);
1690         if (ret) {
1691                 goto end;
1692         }
1693
1694         LTTNG_OPTIONAL_SET(&subbuffer->info.metadata.coherent, coherent);
1695
1696         ret = get_current_subbuf_addr(stream, &addr);
1697         if (ret) {
1698                 goto end;
1699         }
1700
1701         subbuffer->buffer.buffer = lttng_buffer_view_init(
1702                         addr, 0, subbuffer->info.data.padded_subbuf_size);
1703         DBG("Got metadata packet with padded_subbuf_size = %lu, coherent = %s",
1704                         subbuffer->info.metadata.padded_subbuf_size,
1705                         coherent ? "true" : "false");
1706 end:
1707         /*
1708          * The caller only expects -ENODATA when there is no data to read, but
1709          * the kernel tracer returns -EAGAIN when there is currently no data
1710          * for a non-finalized stream, and -ENODATA when there is no data for a
1711          * finalized stream. Those can be combined into a -ENODATA return value.
1712          */
1713         switch (ret) {
1714         case 0:
1715                 status = GET_NEXT_SUBBUFFER_STATUS_OK;
1716                 break;
1717         case -ENODATA:
1718         case -EAGAIN:
1719                 /*
1720                  * The caller only expects -ENODATA when there is no data to
1721                  * read, but the kernel tracer returns -EAGAIN when there is
1722                  * currently no data for a non-finalized stream, and -ENODATA
1723                  * when there is no data for a finalized stream. Those can be
1724                  * combined into a -ENODATA return value.
1725                  */
1726                 status = GET_NEXT_SUBBUFFER_STATUS_NO_DATA;
1727                 break;
1728         default:
1729                 status = GET_NEXT_SUBBUFFER_STATUS_ERROR;
1730                 break;
1731         }
1732
1733         return status;
1734 }
1735
1736 static
1737 int put_next_subbuffer(struct lttng_consumer_stream *stream,
1738                 struct stream_subbuffer *subbuffer __attribute__((unused)))
1739 {
1740         const int ret = kernctl_put_next_subbuf(stream->wait_fd);
1741
1742         if (ret) {
1743                 if (ret == -EFAULT) {
1744                         PERROR("Error in unreserving sub buffer");
1745                 } else if (ret == -EIO) {
1746                         /* Should never happen with newer LTTng versions */
1747                         PERROR("Reader has been pushed by the writer, last sub-buffer corrupted");
1748                 }
1749         }
1750
1751         return ret;
1752 }
1753
1754 static
1755 bool is_get_next_check_metadata_available(int tracer_fd)
1756 {
1757         const int ret = kernctl_get_next_subbuf_metadata_check(tracer_fd, NULL);
1758         const bool available = ret != -ENOTTY;
1759
1760         if (ret == 0) {
1761                 /* get succeeded, make sure to put the subbuffer. */
1762                 kernctl_put_subbuf(tracer_fd);
1763         }
1764
1765         return available;
1766 }
1767
1768 static
1769 int signal_metadata(struct lttng_consumer_stream *stream,
1770                 struct lttng_consumer_local_data *ctx __attribute__((unused)))
1771 {
1772         ASSERT_LOCKED(stream->metadata_rdv_lock);
1773         return pthread_cond_broadcast(&stream->metadata_rdv) ? -errno : 0;
1774 }
1775
1776 static
1777 int lttng_kconsumer_set_stream_ops(
1778                 struct lttng_consumer_stream *stream)
1779 {
1780         int ret = 0;
1781
1782         if (stream->metadata_flag && stream->chan->is_live) {
1783                 DBG("Attempting to enable metadata bucketization for live consumers");
1784                 if (is_get_next_check_metadata_available(stream->wait_fd)) {
1785                         DBG("Kernel tracer supports get_next_subbuffer_metadata_check, metadata will be accumulated until a coherent state is reached");
1786                         stream->read_subbuffer_ops.get_next_subbuffer =
1787                                         get_next_subbuffer_metadata_check;
1788                         ret = consumer_stream_enable_metadata_bucketization(
1789                                         stream);
1790                         if (ret) {
1791                                 goto end;
1792                         }
1793                 } else {
1794                         /*
1795                          * The kernel tracer version is too old to indicate
1796                          * when the metadata stream has reached a "coherent"
1797                          * (parseable) point.
1798                          *
1799                          * This means that a live viewer may see an incoherent
1800                          * sequence of metadata and fail to parse it.
1801                          */
1802                         WARN("Kernel tracer does not support get_next_subbuffer_metadata_check which may cause live clients to fail to parse the metadata stream");
1803                         metadata_bucket_destroy(stream->metadata_bucket);
1804                         stream->metadata_bucket = NULL;
1805                 }
1806
1807                 stream->read_subbuffer_ops.on_sleep = signal_metadata;
1808         }
1809
1810         if (!stream->read_subbuffer_ops.get_next_subbuffer) {
1811                 if (stream->chan->output == CONSUMER_CHANNEL_MMAP) {
1812                         stream->read_subbuffer_ops.get_next_subbuffer =
1813                                         get_next_subbuffer_mmap;
1814                 } else {
1815                         stream->read_subbuffer_ops.get_next_subbuffer =
1816                                         get_next_subbuffer_splice;
1817                 }
1818         }
1819
1820         if (stream->metadata_flag) {
1821                 stream->read_subbuffer_ops.extract_subbuffer_info =
1822                                 extract_metadata_subbuffer_info;
1823         } else {
1824                 stream->read_subbuffer_ops.extract_subbuffer_info =
1825                                 extract_data_subbuffer_info;
1826                 if (stream->chan->is_live) {
1827                         stream->read_subbuffer_ops.send_live_beacon =
1828                                         consumer_flush_kernel_index;
1829                 }
1830         }
1831
1832         stream->read_subbuffer_ops.put_next_subbuffer = put_next_subbuffer;
1833 end:
1834         return ret;
1835 }
1836
1837 int lttng_kconsumer_on_recv_stream(struct lttng_consumer_stream *stream)
1838 {
1839         int ret;
1840
1841         LTTNG_ASSERT(stream);
1842
1843         /*
1844          * Don't create anything if this is set for streaming or if there is
1845          * no current trace chunk on the parent channel.
1846          */
1847         if (stream->net_seq_idx == (uint64_t) -1ULL && stream->chan->monitor &&
1848                         stream->chan->trace_chunk) {
1849                 ret = consumer_stream_create_output_files(stream, true);
1850                 if (ret) {
1851                         goto error;
1852                 }
1853         }
1854
1855         if (stream->output == LTTNG_EVENT_MMAP) {
1856                 /* get the len of the mmap region */
1857                 unsigned long mmap_len;
1858
1859                 ret = kernctl_get_mmap_len(stream->wait_fd, &mmap_len);
1860                 if (ret != 0) {
1861                         PERROR("kernctl_get_mmap_len");
1862                         goto error_close_fd;
1863                 }
1864                 stream->mmap_len = (size_t) mmap_len;
1865
1866                 stream->mmap_base = mmap(NULL, stream->mmap_len, PROT_READ,
1867                                 MAP_PRIVATE, stream->wait_fd, 0);
1868                 if (stream->mmap_base == MAP_FAILED) {
1869                         PERROR("Error mmaping");
1870                         ret = -1;
1871                         goto error_close_fd;
1872                 }
1873         }
1874
1875         ret = lttng_kconsumer_set_stream_ops(stream);
1876         if (ret) {
1877                 goto error_close_fd;
1878         }
1879
1880         /* we return 0 to let the library handle the FD internally */
1881         return 0;
1882
1883 error_close_fd:
1884         if (stream->out_fd >= 0) {
1885                 int err;
1886
1887                 err = close(stream->out_fd);
1888                 LTTNG_ASSERT(!err);
1889                 stream->out_fd = -1;
1890         }
1891 error:
1892         return ret;
1893 }
1894
1895 /*
1896  * Check if data is still being extracted from the buffers for a specific
1897  * stream. Consumer data lock MUST be acquired before calling this function
1898  * and the stream lock.
1899  *
1900  * Return 1 if the traced data are still getting read else 0 meaning that the
1901  * data is available for trace viewer reading.
1902  */
1903 int lttng_kconsumer_data_pending(struct lttng_consumer_stream *stream)
1904 {
1905         int ret;
1906
1907         LTTNG_ASSERT(stream);
1908
1909         if (stream->endpoint_status != CONSUMER_ENDPOINT_ACTIVE) {
1910                 ret = 0;
1911                 goto end;
1912         }
1913
1914         ret = kernctl_get_next_subbuf(stream->wait_fd);
1915         if (ret == 0) {
1916                 /* There is still data so let's put back this subbuffer. */
1917                 ret = kernctl_put_subbuf(stream->wait_fd);
1918                 LTTNG_ASSERT(ret == 0);
1919                 ret = 1;   /* Data is pending */
1920                 goto end;
1921         }
1922
1923         /* Data is NOT pending and ready to be read. */
1924         ret = 0;
1925
1926 end:
1927         return ret;
1928 }