| 1 | /* |
| 2 | * Copyright (C) 2011 - Julien Desfossez <julien.desfossez@polymtl.ca> |
| 3 | * Mathieu Desnoyers <mathieu.desnoyers@efficios.com> |
| 4 | * Copyright (C) 2017 - Jérémie Galarneau <jeremie.galarneau@efficios.com> |
| 5 | * |
| 6 | * This program is free software; you can redistribute it and/or modify |
| 7 | * it under the terms of the GNU General Public License, version 2 only, |
| 8 | * as published by the Free Software Foundation. |
| 9 | * |
| 10 | * This program is distributed in the hope that it will be useful, |
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 13 | * GNU General Public License for more details. |
| 14 | * |
| 15 | * You should have received a copy of the GNU General Public License along |
| 16 | * with this program; if not, write to the Free Software Foundation, Inc., |
| 17 | * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. |
| 18 | */ |
| 19 | |
| 20 | #include <stdint.h> |
| 21 | #define _LGPL_SOURCE |
| 22 | #include <assert.h> |
| 23 | #include <poll.h> |
| 24 | #include <pthread.h> |
| 25 | #include <stdlib.h> |
| 26 | #include <string.h> |
| 27 | #include <sys/mman.h> |
| 28 | #include <sys/socket.h> |
| 29 | #include <sys/types.h> |
| 30 | #include <inttypes.h> |
| 31 | #include <unistd.h> |
| 32 | #include <sys/stat.h> |
| 33 | |
| 34 | #include <bin/lttng-consumerd/health-consumerd.h> |
| 35 | #include <common/common.h> |
| 36 | #include <common/kernel-ctl/kernel-ctl.h> |
| 37 | #include <common/sessiond-comm/sessiond-comm.h> |
| 38 | #include <common/sessiond-comm/relayd.h> |
| 39 | #include <common/compat/fcntl.h> |
| 40 | #include <common/compat/endian.h> |
| 41 | #include <common/pipe.h> |
| 42 | #include <common/relayd/relayd.h> |
| 43 | #include <common/utils.h> |
| 44 | #include <common/consumer/consumer-stream.h> |
| 45 | #include <common/index/index.h> |
| 46 | #include <common/consumer/consumer-timer.h> |
| 47 | #include <common/optional.h> |
| 48 | |
| 49 | #include "kernel-consumer.h" |
| 50 | |
| 51 | extern struct lttng_consumer_global_data consumer_data; |
| 52 | extern int consumer_poll_timeout; |
| 53 | |
| 54 | /* |
| 55 | * Take a snapshot for a specific fd |
| 56 | * |
| 57 | * Returns 0 on success, < 0 on error |
| 58 | */ |
| 59 | int lttng_kconsumer_take_snapshot(struct lttng_consumer_stream *stream) |
| 60 | { |
| 61 | int ret = 0; |
| 62 | int infd = stream->wait_fd; |
| 63 | |
| 64 | ret = kernctl_snapshot(infd); |
| 65 | /* |
| 66 | * -EAGAIN is not an error, it just means that there is no data to |
| 67 | * be read. |
| 68 | */ |
| 69 | if (ret != 0 && ret != -EAGAIN) { |
| 70 | PERROR("Getting sub-buffer snapshot."); |
| 71 | } |
| 72 | |
| 73 | return ret; |
| 74 | } |
| 75 | |
| 76 | /* |
| 77 | * Sample consumed and produced positions for a specific fd. |
| 78 | * |
| 79 | * Returns 0 on success, < 0 on error. |
| 80 | */ |
| 81 | int lttng_kconsumer_sample_snapshot_positions( |
| 82 | struct lttng_consumer_stream *stream) |
| 83 | { |
| 84 | assert(stream); |
| 85 | |
| 86 | return kernctl_snapshot_sample_positions(stream->wait_fd); |
| 87 | } |
| 88 | |
| 89 | /* |
| 90 | * Get the produced position |
| 91 | * |
| 92 | * Returns 0 on success, < 0 on error |
| 93 | */ |
| 94 | int lttng_kconsumer_get_produced_snapshot(struct lttng_consumer_stream *stream, |
| 95 | unsigned long *pos) |
| 96 | { |
| 97 | int ret; |
| 98 | int infd = stream->wait_fd; |
| 99 | |
| 100 | ret = kernctl_snapshot_get_produced(infd, pos); |
| 101 | if (ret != 0) { |
| 102 | PERROR("kernctl_snapshot_get_produced"); |
| 103 | } |
| 104 | |
| 105 | return ret; |
| 106 | } |
| 107 | |
| 108 | /* |
| 109 | * Get the consumerd position |
| 110 | * |
| 111 | * Returns 0 on success, < 0 on error |
| 112 | */ |
| 113 | int lttng_kconsumer_get_consumed_snapshot(struct lttng_consumer_stream *stream, |
| 114 | unsigned long *pos) |
| 115 | { |
| 116 | int ret; |
| 117 | int infd = stream->wait_fd; |
| 118 | |
| 119 | ret = kernctl_snapshot_get_consumed(infd, pos); |
| 120 | if (ret != 0) { |
| 121 | PERROR("kernctl_snapshot_get_consumed"); |
| 122 | } |
| 123 | |
| 124 | return ret; |
| 125 | } |
| 126 | |
| 127 | static |
| 128 | int get_current_subbuf_addr(struct lttng_consumer_stream *stream, |
| 129 | const char **addr) |
| 130 | { |
| 131 | int ret; |
| 132 | unsigned long mmap_offset; |
| 133 | const char *mmap_base = stream->mmap_base; |
| 134 | |
| 135 | ret = kernctl_get_mmap_read_offset(stream->wait_fd, &mmap_offset); |
| 136 | if (ret < 0) { |
| 137 | PERROR("Failed to get mmap read offset"); |
| 138 | goto error; |
| 139 | } |
| 140 | |
| 141 | *addr = mmap_base + mmap_offset; |
| 142 | error: |
| 143 | return ret; |
| 144 | } |
| 145 | |
| 146 | /* |
| 147 | * Take a snapshot of all the stream of a channel |
| 148 | * RCU read-side lock must be held across this function to ensure existence of |
| 149 | * channel. The channel lock must be held by the caller. |
| 150 | * |
| 151 | * Returns 0 on success, < 0 on error |
| 152 | */ |
| 153 | static int lttng_kconsumer_snapshot_channel( |
| 154 | struct lttng_consumer_channel *channel, |
| 155 | uint64_t key, char *path, uint64_t relayd_id, |
| 156 | uint64_t nb_packets_per_stream, |
| 157 | struct lttng_consumer_local_data *ctx) |
| 158 | { |
| 159 | int ret; |
| 160 | struct lttng_consumer_stream *stream; |
| 161 | |
| 162 | DBG("Kernel consumer snapshot channel %" PRIu64, key); |
| 163 | |
| 164 | rcu_read_lock(); |
| 165 | |
| 166 | /* Splice is not supported yet for channel snapshot. */ |
| 167 | if (channel->output != CONSUMER_CHANNEL_MMAP) { |
| 168 | ERR("Unsupported output type for channel \"%s\": mmap output is required to record a snapshot", |
| 169 | channel->name); |
| 170 | ret = -1; |
| 171 | goto end; |
| 172 | } |
| 173 | |
| 174 | cds_list_for_each_entry(stream, &channel->streams.head, send_node) { |
| 175 | unsigned long consumed_pos, produced_pos; |
| 176 | |
| 177 | health_code_update(); |
| 178 | |
| 179 | /* |
| 180 | * Lock stream because we are about to change its state. |
| 181 | */ |
| 182 | pthread_mutex_lock(&stream->lock); |
| 183 | |
| 184 | assert(channel->trace_chunk); |
| 185 | if (!lttng_trace_chunk_get(channel->trace_chunk)) { |
| 186 | /* |
| 187 | * Can't happen barring an internal error as the channel |
| 188 | * holds a reference to the trace chunk. |
| 189 | */ |
| 190 | ERR("Failed to acquire reference to channel's trace chunk"); |
| 191 | ret = -1; |
| 192 | goto end_unlock; |
| 193 | } |
| 194 | assert(!stream->trace_chunk); |
| 195 | stream->trace_chunk = channel->trace_chunk; |
| 196 | |
| 197 | /* |
| 198 | * Assign the received relayd ID so we can use it for streaming. The streams |
| 199 | * are not visible to anyone so this is OK to change it. |
| 200 | */ |
| 201 | stream->net_seq_idx = relayd_id; |
| 202 | channel->relayd_id = relayd_id; |
| 203 | if (relayd_id != (uint64_t) -1ULL) { |
| 204 | ret = consumer_send_relayd_stream(stream, path); |
| 205 | if (ret < 0) { |
| 206 | ERR("sending stream to relayd"); |
| 207 | goto end_unlock; |
| 208 | } |
| 209 | } else { |
| 210 | ret = consumer_stream_create_output_files(stream, |
| 211 | false); |
| 212 | if (ret < 0) { |
| 213 | goto end_unlock; |
| 214 | } |
| 215 | DBG("Kernel consumer snapshot stream (%" PRIu64 ")", |
| 216 | stream->key); |
| 217 | } |
| 218 | |
| 219 | ret = kernctl_buffer_flush_empty(stream->wait_fd); |
| 220 | if (ret < 0) { |
| 221 | /* |
| 222 | * Doing a buffer flush which does not take into |
| 223 | * account empty packets. This is not perfect |
| 224 | * for stream intersection, but required as a |
| 225 | * fall-back when "flush_empty" is not |
| 226 | * implemented by lttng-modules. |
| 227 | */ |
| 228 | ret = kernctl_buffer_flush(stream->wait_fd); |
| 229 | if (ret < 0) { |
| 230 | ERR("Failed to flush kernel stream"); |
| 231 | goto end_unlock; |
| 232 | } |
| 233 | goto end_unlock; |
| 234 | } |
| 235 | |
| 236 | ret = lttng_kconsumer_take_snapshot(stream); |
| 237 | if (ret < 0) { |
| 238 | ERR("Taking kernel snapshot"); |
| 239 | goto end_unlock; |
| 240 | } |
| 241 | |
| 242 | ret = lttng_kconsumer_get_produced_snapshot(stream, &produced_pos); |
| 243 | if (ret < 0) { |
| 244 | ERR("Produced kernel snapshot position"); |
| 245 | goto end_unlock; |
| 246 | } |
| 247 | |
| 248 | ret = lttng_kconsumer_get_consumed_snapshot(stream, &consumed_pos); |
| 249 | if (ret < 0) { |
| 250 | ERR("Consumerd kernel snapshot position"); |
| 251 | goto end_unlock; |
| 252 | } |
| 253 | |
| 254 | consumed_pos = consumer_get_consume_start_pos(consumed_pos, |
| 255 | produced_pos, nb_packets_per_stream, |
| 256 | stream->max_sb_size); |
| 257 | |
| 258 | while ((long) (consumed_pos - produced_pos) < 0) { |
| 259 | ssize_t read_len; |
| 260 | unsigned long len, padded_len; |
| 261 | const char *subbuf_addr; |
| 262 | |
| 263 | health_code_update(); |
| 264 | |
| 265 | DBG("Kernel consumer taking snapshot at pos %lu", consumed_pos); |
| 266 | |
| 267 | ret = kernctl_get_subbuf(stream->wait_fd, &consumed_pos); |
| 268 | if (ret < 0) { |
| 269 | if (ret != -EAGAIN) { |
| 270 | PERROR("kernctl_get_subbuf snapshot"); |
| 271 | goto end_unlock; |
| 272 | } |
| 273 | DBG("Kernel consumer get subbuf failed. Skipping it."); |
| 274 | consumed_pos += stream->max_sb_size; |
| 275 | stream->chan->lost_packets++; |
| 276 | continue; |
| 277 | } |
| 278 | |
| 279 | ret = kernctl_get_subbuf_size(stream->wait_fd, &len); |
| 280 | if (ret < 0) { |
| 281 | ERR("Snapshot kernctl_get_subbuf_size"); |
| 282 | goto error_put_subbuf; |
| 283 | } |
| 284 | |
| 285 | ret = kernctl_get_padded_subbuf_size(stream->wait_fd, &padded_len); |
| 286 | if (ret < 0) { |
| 287 | ERR("Snapshot kernctl_get_padded_subbuf_size"); |
| 288 | goto error_put_subbuf; |
| 289 | } |
| 290 | |
| 291 | ret = get_current_subbuf_addr(stream, &subbuf_addr); |
| 292 | if (ret) { |
| 293 | goto error_put_subbuf; |
| 294 | } |
| 295 | |
| 296 | read_len = lttng_consumer_on_read_subbuffer_mmap(ctx, |
| 297 | stream, subbuf_addr, len, |
| 298 | padded_len - len, NULL); |
| 299 | /* |
| 300 | * We write the padded len in local tracefiles but the data len |
| 301 | * when using a relay. Display the error but continue processing |
| 302 | * to try to release the subbuffer. |
| 303 | */ |
| 304 | if (relayd_id != (uint64_t) -1ULL) { |
| 305 | if (read_len != len) { |
| 306 | ERR("Error sending to the relay (ret: %zd != len: %lu)", |
| 307 | read_len, len); |
| 308 | } |
| 309 | } else { |
| 310 | if (read_len != padded_len) { |
| 311 | ERR("Error writing to tracefile (ret: %zd != len: %lu)", |
| 312 | read_len, padded_len); |
| 313 | } |
| 314 | } |
| 315 | |
| 316 | ret = kernctl_put_subbuf(stream->wait_fd); |
| 317 | if (ret < 0) { |
| 318 | ERR("Snapshot kernctl_put_subbuf"); |
| 319 | goto end_unlock; |
| 320 | } |
| 321 | consumed_pos += stream->max_sb_size; |
| 322 | } |
| 323 | |
| 324 | if (relayd_id == (uint64_t) -1ULL) { |
| 325 | if (stream->out_fd >= 0) { |
| 326 | ret = close(stream->out_fd); |
| 327 | if (ret < 0) { |
| 328 | PERROR("Kernel consumer snapshot close out_fd"); |
| 329 | goto end_unlock; |
| 330 | } |
| 331 | stream->out_fd = -1; |
| 332 | } |
| 333 | } else { |
| 334 | close_relayd_stream(stream); |
| 335 | stream->net_seq_idx = (uint64_t) -1ULL; |
| 336 | } |
| 337 | lttng_trace_chunk_put(stream->trace_chunk); |
| 338 | stream->trace_chunk = NULL; |
| 339 | pthread_mutex_unlock(&stream->lock); |
| 340 | } |
| 341 | |
| 342 | /* All good! */ |
| 343 | ret = 0; |
| 344 | goto end; |
| 345 | |
| 346 | error_put_subbuf: |
| 347 | ret = kernctl_put_subbuf(stream->wait_fd); |
| 348 | if (ret < 0) { |
| 349 | ERR("Snapshot kernctl_put_subbuf error path"); |
| 350 | } |
| 351 | end_unlock: |
| 352 | pthread_mutex_unlock(&stream->lock); |
| 353 | end: |
| 354 | rcu_read_unlock(); |
| 355 | return ret; |
| 356 | } |
| 357 | |
| 358 | /* |
| 359 | * Read the whole metadata available for a snapshot. |
| 360 | * RCU read-side lock must be held across this function to ensure existence of |
| 361 | * metadata_channel. The channel lock must be held by the caller. |
| 362 | * |
| 363 | * Returns 0 on success, < 0 on error |
| 364 | */ |
| 365 | static int lttng_kconsumer_snapshot_metadata( |
| 366 | struct lttng_consumer_channel *metadata_channel, |
| 367 | uint64_t key, char *path, uint64_t relayd_id, |
| 368 | struct lttng_consumer_local_data *ctx) |
| 369 | { |
| 370 | int ret, use_relayd = 0; |
| 371 | ssize_t ret_read; |
| 372 | struct lttng_consumer_stream *metadata_stream; |
| 373 | |
| 374 | assert(ctx); |
| 375 | |
| 376 | DBG("Kernel consumer snapshot metadata with key %" PRIu64 " at path %s", |
| 377 | key, path); |
| 378 | |
| 379 | rcu_read_lock(); |
| 380 | |
| 381 | metadata_stream = metadata_channel->metadata_stream; |
| 382 | assert(metadata_stream); |
| 383 | |
| 384 | pthread_mutex_lock(&metadata_stream->lock); |
| 385 | assert(metadata_channel->trace_chunk); |
| 386 | assert(metadata_stream->trace_chunk); |
| 387 | |
| 388 | /* Flag once that we have a valid relayd for the stream. */ |
| 389 | if (relayd_id != (uint64_t) -1ULL) { |
| 390 | use_relayd = 1; |
| 391 | } |
| 392 | |
| 393 | if (use_relayd) { |
| 394 | ret = consumer_send_relayd_stream(metadata_stream, path); |
| 395 | if (ret < 0) { |
| 396 | goto error_snapshot; |
| 397 | } |
| 398 | } else { |
| 399 | ret = consumer_stream_create_output_files(metadata_stream, |
| 400 | false); |
| 401 | if (ret < 0) { |
| 402 | goto error_snapshot; |
| 403 | } |
| 404 | } |
| 405 | |
| 406 | do { |
| 407 | health_code_update(); |
| 408 | |
| 409 | ret_read = lttng_kconsumer_read_subbuffer(metadata_stream, ctx); |
| 410 | if (ret_read < 0) { |
| 411 | if (ret_read != -EAGAIN) { |
| 412 | ERR("Kernel snapshot reading metadata subbuffer (ret: %zd)", |
| 413 | ret_read); |
| 414 | ret = ret_read; |
| 415 | goto error_snapshot; |
| 416 | } |
| 417 | /* ret_read is negative at this point so we will exit the loop. */ |
| 418 | continue; |
| 419 | } |
| 420 | } while (ret_read >= 0); |
| 421 | |
| 422 | if (use_relayd) { |
| 423 | close_relayd_stream(metadata_stream); |
| 424 | metadata_stream->net_seq_idx = (uint64_t) -1ULL; |
| 425 | } else { |
| 426 | if (metadata_stream->out_fd >= 0) { |
| 427 | ret = close(metadata_stream->out_fd); |
| 428 | if (ret < 0) { |
| 429 | PERROR("Kernel consumer snapshot metadata close out_fd"); |
| 430 | /* |
| 431 | * Don't go on error here since the snapshot was successful at this |
| 432 | * point but somehow the close failed. |
| 433 | */ |
| 434 | } |
| 435 | metadata_stream->out_fd = -1; |
| 436 | lttng_trace_chunk_put(metadata_stream->trace_chunk); |
| 437 | metadata_stream->trace_chunk = NULL; |
| 438 | } |
| 439 | } |
| 440 | |
| 441 | ret = 0; |
| 442 | error_snapshot: |
| 443 | pthread_mutex_unlock(&metadata_stream->lock); |
| 444 | cds_list_del(&metadata_stream->send_node); |
| 445 | consumer_stream_destroy(metadata_stream, NULL); |
| 446 | metadata_channel->metadata_stream = NULL; |
| 447 | rcu_read_unlock(); |
| 448 | return ret; |
| 449 | } |
| 450 | |
| 451 | /* |
| 452 | * Receive command from session daemon and process it. |
| 453 | * |
| 454 | * Return 1 on success else a negative value or 0. |
| 455 | */ |
| 456 | int lttng_kconsumer_recv_cmd(struct lttng_consumer_local_data *ctx, |
| 457 | int sock, struct pollfd *consumer_sockpoll) |
| 458 | { |
| 459 | ssize_t ret; |
| 460 | enum lttcomm_return_code ret_code = LTTCOMM_CONSUMERD_SUCCESS; |
| 461 | struct lttcomm_consumer_msg msg; |
| 462 | |
| 463 | health_code_update(); |
| 464 | |
| 465 | ret = lttcomm_recv_unix_sock(sock, &msg, sizeof(msg)); |
| 466 | if (ret != sizeof(msg)) { |
| 467 | if (ret > 0) { |
| 468 | lttng_consumer_send_error(ctx, LTTCOMM_CONSUMERD_ERROR_RECV_CMD); |
| 469 | ret = -1; |
| 470 | } |
| 471 | return ret; |
| 472 | } |
| 473 | |
| 474 | health_code_update(); |
| 475 | |
| 476 | /* Deprecated command */ |
| 477 | assert(msg.cmd_type != LTTNG_CONSUMER_STOP); |
| 478 | |
| 479 | health_code_update(); |
| 480 | |
| 481 | /* relayd needs RCU read-side protection */ |
| 482 | rcu_read_lock(); |
| 483 | |
| 484 | switch (msg.cmd_type) { |
| 485 | case LTTNG_CONSUMER_ADD_RELAYD_SOCKET: |
| 486 | { |
| 487 | /* Session daemon status message are handled in the following call. */ |
| 488 | consumer_add_relayd_socket(msg.u.relayd_sock.net_index, |
| 489 | msg.u.relayd_sock.type, ctx, sock, consumer_sockpoll, |
| 490 | &msg.u.relayd_sock.sock, msg.u.relayd_sock.session_id, |
| 491 | msg.u.relayd_sock.relayd_session_id); |
| 492 | goto end_nosignal; |
| 493 | } |
| 494 | case LTTNG_CONSUMER_ADD_CHANNEL: |
| 495 | { |
| 496 | struct lttng_consumer_channel *new_channel; |
| 497 | int ret_recv; |
| 498 | const uint64_t chunk_id = msg.u.channel.chunk_id.value; |
| 499 | |
| 500 | health_code_update(); |
| 501 | |
| 502 | /* First send a status message before receiving the fds. */ |
| 503 | ret = consumer_send_status_msg(sock, ret_code); |
| 504 | if (ret < 0) { |
| 505 | /* Somehow, the session daemon is not responding anymore. */ |
| 506 | goto error_fatal; |
| 507 | } |
| 508 | |
| 509 | health_code_update(); |
| 510 | |
| 511 | DBG("consumer_add_channel %" PRIu64, msg.u.channel.channel_key); |
| 512 | new_channel = consumer_allocate_channel(msg.u.channel.channel_key, |
| 513 | msg.u.channel.session_id, |
| 514 | msg.u.channel.chunk_id.is_set ? |
| 515 | &chunk_id : NULL, |
| 516 | msg.u.channel.pathname, |
| 517 | msg.u.channel.name, |
| 518 | msg.u.channel.relayd_id, msg.u.channel.output, |
| 519 | msg.u.channel.tracefile_size, |
| 520 | msg.u.channel.tracefile_count, 0, |
| 521 | msg.u.channel.monitor, |
| 522 | msg.u.channel.live_timer_interval, |
| 523 | NULL, NULL); |
| 524 | if (new_channel == NULL) { |
| 525 | lttng_consumer_send_error(ctx, LTTCOMM_CONSUMERD_OUTFD_ERROR); |
| 526 | goto end_nosignal; |
| 527 | } |
| 528 | new_channel->nb_init_stream_left = msg.u.channel.nb_init_streams; |
| 529 | switch (msg.u.channel.output) { |
| 530 | case LTTNG_EVENT_SPLICE: |
| 531 | new_channel->output = CONSUMER_CHANNEL_SPLICE; |
| 532 | break; |
| 533 | case LTTNG_EVENT_MMAP: |
| 534 | new_channel->output = CONSUMER_CHANNEL_MMAP; |
| 535 | break; |
| 536 | default: |
| 537 | ERR("Channel output unknown %d", msg.u.channel.output); |
| 538 | goto end_nosignal; |
| 539 | } |
| 540 | |
| 541 | /* Translate and save channel type. */ |
| 542 | switch (msg.u.channel.type) { |
| 543 | case CONSUMER_CHANNEL_TYPE_DATA: |
| 544 | case CONSUMER_CHANNEL_TYPE_METADATA: |
| 545 | new_channel->type = msg.u.channel.type; |
| 546 | break; |
| 547 | default: |
| 548 | assert(0); |
| 549 | goto end_nosignal; |
| 550 | }; |
| 551 | |
| 552 | health_code_update(); |
| 553 | |
| 554 | if (ctx->on_recv_channel != NULL) { |
| 555 | ret_recv = ctx->on_recv_channel(new_channel); |
| 556 | if (ret_recv == 0) { |
| 557 | ret = consumer_add_channel(new_channel, ctx); |
| 558 | } else if (ret_recv < 0) { |
| 559 | goto end_nosignal; |
| 560 | } |
| 561 | } else { |
| 562 | ret = consumer_add_channel(new_channel, ctx); |
| 563 | } |
| 564 | if (msg.u.channel.type == CONSUMER_CHANNEL_TYPE_DATA && !ret) { |
| 565 | int monitor_start_ret; |
| 566 | |
| 567 | DBG("Consumer starting monitor timer"); |
| 568 | consumer_timer_live_start(new_channel, |
| 569 | msg.u.channel.live_timer_interval); |
| 570 | monitor_start_ret = consumer_timer_monitor_start( |
| 571 | new_channel, |
| 572 | msg.u.channel.monitor_timer_interval); |
| 573 | if (monitor_start_ret < 0) { |
| 574 | ERR("Starting channel monitoring timer failed"); |
| 575 | goto end_nosignal; |
| 576 | } |
| 577 | |
| 578 | } |
| 579 | |
| 580 | health_code_update(); |
| 581 | |
| 582 | /* If we received an error in add_channel, we need to report it. */ |
| 583 | if (ret < 0) { |
| 584 | ret = consumer_send_status_msg(sock, ret); |
| 585 | if (ret < 0) { |
| 586 | goto error_fatal; |
| 587 | } |
| 588 | goto end_nosignal; |
| 589 | } |
| 590 | |
| 591 | goto end_nosignal; |
| 592 | } |
| 593 | case LTTNG_CONSUMER_ADD_STREAM: |
| 594 | { |
| 595 | int fd; |
| 596 | struct lttng_pipe *stream_pipe; |
| 597 | struct lttng_consumer_stream *new_stream; |
| 598 | struct lttng_consumer_channel *channel; |
| 599 | int alloc_ret = 0; |
| 600 | |
| 601 | /* |
| 602 | * Get stream's channel reference. Needed when adding the stream to the |
| 603 | * global hash table. |
| 604 | */ |
| 605 | channel = consumer_find_channel(msg.u.stream.channel_key); |
| 606 | if (!channel) { |
| 607 | /* |
| 608 | * We could not find the channel. Can happen if cpu hotplug |
| 609 | * happens while tearing down. |
| 610 | */ |
| 611 | ERR("Unable to find channel key %" PRIu64, msg.u.stream.channel_key); |
| 612 | ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND; |
| 613 | } |
| 614 | |
| 615 | health_code_update(); |
| 616 | |
| 617 | /* First send a status message before receiving the fds. */ |
| 618 | ret = consumer_send_status_msg(sock, ret_code); |
| 619 | if (ret < 0) { |
| 620 | /* Somehow, the session daemon is not responding anymore. */ |
| 621 | goto error_add_stream_fatal; |
| 622 | } |
| 623 | |
| 624 | health_code_update(); |
| 625 | |
| 626 | if (ret_code != LTTCOMM_CONSUMERD_SUCCESS) { |
| 627 | /* Channel was not found. */ |
| 628 | goto error_add_stream_nosignal; |
| 629 | } |
| 630 | |
| 631 | /* Blocking call */ |
| 632 | health_poll_entry(); |
| 633 | ret = lttng_consumer_poll_socket(consumer_sockpoll); |
| 634 | health_poll_exit(); |
| 635 | if (ret) { |
| 636 | goto error_add_stream_fatal; |
| 637 | } |
| 638 | |
| 639 | health_code_update(); |
| 640 | |
| 641 | /* Get stream file descriptor from socket */ |
| 642 | ret = lttcomm_recv_fds_unix_sock(sock, &fd, 1); |
| 643 | if (ret != sizeof(fd)) { |
| 644 | lttng_consumer_send_error(ctx, LTTCOMM_CONSUMERD_ERROR_RECV_FD); |
| 645 | goto end; |
| 646 | } |
| 647 | |
| 648 | health_code_update(); |
| 649 | |
| 650 | /* |
| 651 | * Send status code to session daemon only if the recv works. If the |
| 652 | * above recv() failed, the session daemon is notified through the |
| 653 | * error socket and the teardown is eventually done. |
| 654 | */ |
| 655 | ret = consumer_send_status_msg(sock, ret_code); |
| 656 | if (ret < 0) { |
| 657 | /* Somehow, the session daemon is not responding anymore. */ |
| 658 | goto error_add_stream_nosignal; |
| 659 | } |
| 660 | |
| 661 | health_code_update(); |
| 662 | |
| 663 | pthread_mutex_lock(&channel->lock); |
| 664 | new_stream = consumer_allocate_stream(channel->key, |
| 665 | fd, |
| 666 | channel->name, |
| 667 | channel->relayd_id, |
| 668 | channel->session_id, |
| 669 | channel->trace_chunk, |
| 670 | msg.u.stream.cpu, |
| 671 | &alloc_ret, |
| 672 | channel->type, |
| 673 | channel->monitor); |
| 674 | if (new_stream == NULL) { |
| 675 | switch (alloc_ret) { |
| 676 | case -ENOMEM: |
| 677 | case -EINVAL: |
| 678 | default: |
| 679 | lttng_consumer_send_error(ctx, LTTCOMM_CONSUMERD_OUTFD_ERROR); |
| 680 | break; |
| 681 | } |
| 682 | pthread_mutex_unlock(&channel->lock); |
| 683 | goto error_add_stream_nosignal; |
| 684 | } |
| 685 | |
| 686 | new_stream->chan = channel; |
| 687 | new_stream->wait_fd = fd; |
| 688 | ret = kernctl_get_max_subbuf_size(new_stream->wait_fd, |
| 689 | &new_stream->max_sb_size); |
| 690 | if (ret < 0) { |
| 691 | pthread_mutex_unlock(&channel->lock); |
| 692 | ERR("Failed to get kernel maximal subbuffer size"); |
| 693 | goto error_add_stream_nosignal; |
| 694 | } |
| 695 | |
| 696 | consumer_stream_update_channel_attributes(new_stream, |
| 697 | channel); |
| 698 | switch (channel->output) { |
| 699 | case CONSUMER_CHANNEL_SPLICE: |
| 700 | new_stream->output = LTTNG_EVENT_SPLICE; |
| 701 | ret = utils_create_pipe(new_stream->splice_pipe); |
| 702 | if (ret < 0) { |
| 703 | pthread_mutex_unlock(&channel->lock); |
| 704 | goto error_add_stream_nosignal; |
| 705 | } |
| 706 | break; |
| 707 | case CONSUMER_CHANNEL_MMAP: |
| 708 | new_stream->output = LTTNG_EVENT_MMAP; |
| 709 | break; |
| 710 | default: |
| 711 | ERR("Stream output unknown %d", channel->output); |
| 712 | pthread_mutex_unlock(&channel->lock); |
| 713 | goto error_add_stream_nosignal; |
| 714 | } |
| 715 | |
| 716 | /* |
| 717 | * We've just assigned the channel to the stream so increment the |
| 718 | * refcount right now. We don't need to increment the refcount for |
| 719 | * streams in no monitor because we handle manually the cleanup of |
| 720 | * those. It is very important to make sure there is NO prior |
| 721 | * consumer_del_stream() calls or else the refcount will be unbalanced. |
| 722 | */ |
| 723 | if (channel->monitor) { |
| 724 | uatomic_inc(&new_stream->chan->refcount); |
| 725 | } |
| 726 | |
| 727 | /* |
| 728 | * The buffer flush is done on the session daemon side for the kernel |
| 729 | * so no need for the stream "hangup_flush_done" variable to be |
| 730 | * tracked. This is important for a kernel stream since we don't rely |
| 731 | * on the flush state of the stream to read data. It's not the case for |
| 732 | * user space tracing. |
| 733 | */ |
| 734 | new_stream->hangup_flush_done = 0; |
| 735 | |
| 736 | health_code_update(); |
| 737 | |
| 738 | pthread_mutex_lock(&new_stream->lock); |
| 739 | if (ctx->on_recv_stream) { |
| 740 | ret = ctx->on_recv_stream(new_stream); |
| 741 | if (ret < 0) { |
| 742 | pthread_mutex_unlock(&new_stream->lock); |
| 743 | pthread_mutex_unlock(&channel->lock); |
| 744 | consumer_stream_free(new_stream); |
| 745 | goto error_add_stream_nosignal; |
| 746 | } |
| 747 | } |
| 748 | health_code_update(); |
| 749 | |
| 750 | if (new_stream->metadata_flag) { |
| 751 | channel->metadata_stream = new_stream; |
| 752 | } |
| 753 | |
| 754 | /* Do not monitor this stream. */ |
| 755 | if (!channel->monitor) { |
| 756 | DBG("Kernel consumer add stream %s in no monitor mode with " |
| 757 | "relayd id %" PRIu64, new_stream->name, |
| 758 | new_stream->net_seq_idx); |
| 759 | cds_list_add(&new_stream->send_node, &channel->streams.head); |
| 760 | pthread_mutex_unlock(&new_stream->lock); |
| 761 | pthread_mutex_unlock(&channel->lock); |
| 762 | goto end_add_stream; |
| 763 | } |
| 764 | |
| 765 | /* Send stream to relayd if the stream has an ID. */ |
| 766 | if (new_stream->net_seq_idx != (uint64_t) -1ULL) { |
| 767 | ret = consumer_send_relayd_stream(new_stream, |
| 768 | new_stream->chan->pathname); |
| 769 | if (ret < 0) { |
| 770 | pthread_mutex_unlock(&new_stream->lock); |
| 771 | pthread_mutex_unlock(&channel->lock); |
| 772 | consumer_stream_free(new_stream); |
| 773 | goto error_add_stream_nosignal; |
| 774 | } |
| 775 | |
| 776 | /* |
| 777 | * If adding an extra stream to an already |
| 778 | * existing channel (e.g. cpu hotplug), we need |
| 779 | * to send the "streams_sent" command to relayd. |
| 780 | */ |
| 781 | if (channel->streams_sent_to_relayd) { |
| 782 | ret = consumer_send_relayd_streams_sent( |
| 783 | new_stream->net_seq_idx); |
| 784 | if (ret < 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 = lttng_pipe_write(stream_pipe, &new_stream, sizeof(new_stream)); |
| 809 | if (ret < 0) { |
| 810 | ERR("Consumer write %s stream to pipe %d", |
| 811 | new_stream->metadata_flag ? "metadata" : "data", |
| 812 | lttng_pipe_get_writefd(stream_pipe)); |
| 813 | if (new_stream->metadata_flag) { |
| 814 | consumer_del_stream_for_metadata(new_stream); |
| 815 | } else { |
| 816 | consumer_del_stream_for_data(new_stream); |
| 817 | } |
| 818 | goto error_add_stream_nosignal; |
| 819 | } |
| 820 | |
| 821 | DBG("Kernel consumer ADD_STREAM %s (fd: %d) %s with relayd id %" PRIu64, |
| 822 | new_stream->name, fd, new_stream->chan->pathname, new_stream->relayd_stream_id); |
| 823 | end_add_stream: |
| 824 | break; |
| 825 | error_add_stream_nosignal: |
| 826 | goto end_nosignal; |
| 827 | error_add_stream_fatal: |
| 828 | goto error_fatal; |
| 829 | } |
| 830 | case LTTNG_CONSUMER_STREAMS_SENT: |
| 831 | { |
| 832 | struct lttng_consumer_channel *channel; |
| 833 | |
| 834 | /* |
| 835 | * Get stream's channel reference. Needed when adding the stream to the |
| 836 | * global hash table. |
| 837 | */ |
| 838 | channel = consumer_find_channel(msg.u.sent_streams.channel_key); |
| 839 | if (!channel) { |
| 840 | /* |
| 841 | * We could not find the channel. Can happen if cpu hotplug |
| 842 | * happens while tearing down. |
| 843 | */ |
| 844 | ERR("Unable to find channel key %" PRIu64, |
| 845 | msg.u.sent_streams.channel_key); |
| 846 | ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND; |
| 847 | } |
| 848 | |
| 849 | health_code_update(); |
| 850 | |
| 851 | /* |
| 852 | * Send status code to session daemon. |
| 853 | */ |
| 854 | ret = consumer_send_status_msg(sock, ret_code); |
| 855 | if (ret < 0 || ret_code != LTTCOMM_CONSUMERD_SUCCESS) { |
| 856 | /* Somehow, the session daemon is not responding anymore. */ |
| 857 | goto error_streams_sent_nosignal; |
| 858 | } |
| 859 | |
| 860 | health_code_update(); |
| 861 | |
| 862 | /* |
| 863 | * We should not send this message if we don't monitor the |
| 864 | * streams in this channel. |
| 865 | */ |
| 866 | if (!channel->monitor) { |
| 867 | goto end_error_streams_sent; |
| 868 | } |
| 869 | |
| 870 | health_code_update(); |
| 871 | /* Send stream to relayd if the stream has an ID. */ |
| 872 | if (msg.u.sent_streams.net_seq_idx != (uint64_t) -1ULL) { |
| 873 | ret = consumer_send_relayd_streams_sent( |
| 874 | msg.u.sent_streams.net_seq_idx); |
| 875 | if (ret < 0) { |
| 876 | goto error_streams_sent_nosignal; |
| 877 | } |
| 878 | channel->streams_sent_to_relayd = true; |
| 879 | } |
| 880 | end_error_streams_sent: |
| 881 | break; |
| 882 | error_streams_sent_nosignal: |
| 883 | goto end_nosignal; |
| 884 | } |
| 885 | case LTTNG_CONSUMER_UPDATE_STREAM: |
| 886 | { |
| 887 | rcu_read_unlock(); |
| 888 | return -ENOSYS; |
| 889 | } |
| 890 | case LTTNG_CONSUMER_DESTROY_RELAYD: |
| 891 | { |
| 892 | uint64_t index = msg.u.destroy_relayd.net_seq_idx; |
| 893 | struct consumer_relayd_sock_pair *relayd; |
| 894 | |
| 895 | DBG("Kernel consumer destroying relayd %" PRIu64, index); |
| 896 | |
| 897 | /* Get relayd reference if exists. */ |
| 898 | relayd = consumer_find_relayd(index); |
| 899 | if (relayd == NULL) { |
| 900 | DBG("Unable to find relayd %" PRIu64, index); |
| 901 | ret_code = LTTCOMM_CONSUMERD_RELAYD_FAIL; |
| 902 | } |
| 903 | |
| 904 | /* |
| 905 | * Each relayd socket pair has a refcount of stream attached to it |
| 906 | * which tells if the relayd is still active or not depending on the |
| 907 | * refcount value. |
| 908 | * |
| 909 | * This will set the destroy flag of the relayd object and destroy it |
| 910 | * if the refcount reaches zero when called. |
| 911 | * |
| 912 | * The destroy can happen either here or when a stream fd hangs up. |
| 913 | */ |
| 914 | if (relayd) { |
| 915 | consumer_flag_relayd_for_destroy(relayd); |
| 916 | } |
| 917 | |
| 918 | health_code_update(); |
| 919 | |
| 920 | ret = consumer_send_status_msg(sock, ret_code); |
| 921 | if (ret < 0) { |
| 922 | /* Somehow, the session daemon is not responding anymore. */ |
| 923 | goto error_fatal; |
| 924 | } |
| 925 | |
| 926 | goto end_nosignal; |
| 927 | } |
| 928 | case LTTNG_CONSUMER_DATA_PENDING: |
| 929 | { |
| 930 | int32_t ret; |
| 931 | uint64_t id = msg.u.data_pending.session_id; |
| 932 | |
| 933 | DBG("Kernel consumer data pending command for id %" PRIu64, id); |
| 934 | |
| 935 | ret = consumer_data_pending(id); |
| 936 | |
| 937 | health_code_update(); |
| 938 | |
| 939 | /* Send back returned value to session daemon */ |
| 940 | ret = lttcomm_send_unix_sock(sock, &ret, sizeof(ret)); |
| 941 | if (ret < 0) { |
| 942 | PERROR("send data pending ret code"); |
| 943 | goto error_fatal; |
| 944 | } |
| 945 | |
| 946 | /* |
| 947 | * No need to send back a status message since the data pending |
| 948 | * returned value is the response. |
| 949 | */ |
| 950 | break; |
| 951 | } |
| 952 | case LTTNG_CONSUMER_SNAPSHOT_CHANNEL: |
| 953 | { |
| 954 | struct lttng_consumer_channel *channel; |
| 955 | uint64_t key = msg.u.snapshot_channel.key; |
| 956 | |
| 957 | channel = consumer_find_channel(key); |
| 958 | if (!channel) { |
| 959 | ERR("Channel %" PRIu64 " not found", key); |
| 960 | ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND; |
| 961 | } else { |
| 962 | pthread_mutex_lock(&channel->lock); |
| 963 | if (msg.u.snapshot_channel.metadata == 1) { |
| 964 | ret = lttng_kconsumer_snapshot_metadata(channel, key, |
| 965 | msg.u.snapshot_channel.pathname, |
| 966 | msg.u.snapshot_channel.relayd_id, ctx); |
| 967 | if (ret < 0) { |
| 968 | ERR("Snapshot metadata failed"); |
| 969 | ret_code = LTTCOMM_CONSUMERD_SNAPSHOT_FAILED; |
| 970 | } |
| 971 | } else { |
| 972 | ret = lttng_kconsumer_snapshot_channel(channel, key, |
| 973 | msg.u.snapshot_channel.pathname, |
| 974 | msg.u.snapshot_channel.relayd_id, |
| 975 | msg.u.snapshot_channel.nb_packets_per_stream, |
| 976 | ctx); |
| 977 | if (ret < 0) { |
| 978 | ERR("Snapshot channel failed"); |
| 979 | ret_code = LTTCOMM_CONSUMERD_SNAPSHOT_FAILED; |
| 980 | } |
| 981 | } |
| 982 | pthread_mutex_unlock(&channel->lock); |
| 983 | } |
| 984 | health_code_update(); |
| 985 | |
| 986 | ret = consumer_send_status_msg(sock, ret_code); |
| 987 | if (ret < 0) { |
| 988 | /* Somehow, the session daemon is not responding anymore. */ |
| 989 | goto end_nosignal; |
| 990 | } |
| 991 | break; |
| 992 | } |
| 993 | case LTTNG_CONSUMER_DESTROY_CHANNEL: |
| 994 | { |
| 995 | uint64_t key = msg.u.destroy_channel.key; |
| 996 | struct lttng_consumer_channel *channel; |
| 997 | |
| 998 | channel = consumer_find_channel(key); |
| 999 | if (!channel) { |
| 1000 | ERR("Kernel consumer destroy channel %" PRIu64 " not found", key); |
| 1001 | ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND; |
| 1002 | } |
| 1003 | |
| 1004 | health_code_update(); |
| 1005 | |
| 1006 | ret = consumer_send_status_msg(sock, ret_code); |
| 1007 | if (ret < 0) { |
| 1008 | /* Somehow, the session daemon is not responding anymore. */ |
| 1009 | goto end_destroy_channel; |
| 1010 | } |
| 1011 | |
| 1012 | health_code_update(); |
| 1013 | |
| 1014 | /* Stop right now if no channel was found. */ |
| 1015 | if (!channel) { |
| 1016 | goto end_destroy_channel; |
| 1017 | } |
| 1018 | |
| 1019 | /* |
| 1020 | * This command should ONLY be issued for channel with streams set in |
| 1021 | * no monitor mode. |
| 1022 | */ |
| 1023 | assert(!channel->monitor); |
| 1024 | |
| 1025 | /* |
| 1026 | * The refcount should ALWAYS be 0 in the case of a channel in no |
| 1027 | * monitor mode. |
| 1028 | */ |
| 1029 | assert(!uatomic_sub_return(&channel->refcount, 1)); |
| 1030 | |
| 1031 | consumer_del_channel(channel); |
| 1032 | end_destroy_channel: |
| 1033 | goto end_nosignal; |
| 1034 | } |
| 1035 | case LTTNG_CONSUMER_DISCARDED_EVENTS: |
| 1036 | { |
| 1037 | ssize_t ret; |
| 1038 | uint64_t count; |
| 1039 | struct lttng_consumer_channel *channel; |
| 1040 | uint64_t id = msg.u.discarded_events.session_id; |
| 1041 | uint64_t key = msg.u.discarded_events.channel_key; |
| 1042 | |
| 1043 | DBG("Kernel consumer discarded events command for session id %" |
| 1044 | PRIu64 ", channel key %" PRIu64, id, key); |
| 1045 | |
| 1046 | channel = consumer_find_channel(key); |
| 1047 | if (!channel) { |
| 1048 | ERR("Kernel consumer discarded events channel %" |
| 1049 | PRIu64 " not found", key); |
| 1050 | count = 0; |
| 1051 | } else { |
| 1052 | count = channel->discarded_events; |
| 1053 | } |
| 1054 | |
| 1055 | health_code_update(); |
| 1056 | |
| 1057 | /* Send back returned value to session daemon */ |
| 1058 | ret = lttcomm_send_unix_sock(sock, &count, sizeof(count)); |
| 1059 | if (ret < 0) { |
| 1060 | PERROR("send discarded events"); |
| 1061 | goto error_fatal; |
| 1062 | } |
| 1063 | |
| 1064 | break; |
| 1065 | } |
| 1066 | case LTTNG_CONSUMER_LOST_PACKETS: |
| 1067 | { |
| 1068 | ssize_t ret; |
| 1069 | uint64_t count; |
| 1070 | struct lttng_consumer_channel *channel; |
| 1071 | uint64_t id = msg.u.lost_packets.session_id; |
| 1072 | uint64_t key = msg.u.lost_packets.channel_key; |
| 1073 | |
| 1074 | DBG("Kernel consumer lost packets command for session id %" |
| 1075 | PRIu64 ", channel key %" PRIu64, id, key); |
| 1076 | |
| 1077 | channel = consumer_find_channel(key); |
| 1078 | if (!channel) { |
| 1079 | ERR("Kernel consumer lost packets channel %" |
| 1080 | PRIu64 " not found", key); |
| 1081 | count = 0; |
| 1082 | } else { |
| 1083 | count = channel->lost_packets; |
| 1084 | } |
| 1085 | |
| 1086 | health_code_update(); |
| 1087 | |
| 1088 | /* Send back returned value to session daemon */ |
| 1089 | ret = lttcomm_send_unix_sock(sock, &count, sizeof(count)); |
| 1090 | if (ret < 0) { |
| 1091 | PERROR("send lost packets"); |
| 1092 | goto error_fatal; |
| 1093 | } |
| 1094 | |
| 1095 | break; |
| 1096 | } |
| 1097 | case LTTNG_CONSUMER_SET_CHANNEL_MONITOR_PIPE: |
| 1098 | { |
| 1099 | int channel_monitor_pipe; |
| 1100 | |
| 1101 | ret_code = LTTCOMM_CONSUMERD_SUCCESS; |
| 1102 | /* Successfully received the command's type. */ |
| 1103 | ret = consumer_send_status_msg(sock, ret_code); |
| 1104 | if (ret < 0) { |
| 1105 | goto error_fatal; |
| 1106 | } |
| 1107 | |
| 1108 | ret = lttcomm_recv_fds_unix_sock(sock, &channel_monitor_pipe, |
| 1109 | 1); |
| 1110 | if (ret != sizeof(channel_monitor_pipe)) { |
| 1111 | ERR("Failed to receive channel monitor pipe"); |
| 1112 | goto error_fatal; |
| 1113 | } |
| 1114 | |
| 1115 | DBG("Received channel monitor pipe (%d)", channel_monitor_pipe); |
| 1116 | ret = consumer_timer_thread_set_channel_monitor_pipe( |
| 1117 | channel_monitor_pipe); |
| 1118 | if (!ret) { |
| 1119 | int flags; |
| 1120 | |
| 1121 | ret_code = LTTCOMM_CONSUMERD_SUCCESS; |
| 1122 | /* Set the pipe as non-blocking. */ |
| 1123 | ret = fcntl(channel_monitor_pipe, F_GETFL, 0); |
| 1124 | if (ret == -1) { |
| 1125 | PERROR("fcntl get flags of the channel monitoring pipe"); |
| 1126 | goto error_fatal; |
| 1127 | } |
| 1128 | flags = ret; |
| 1129 | |
| 1130 | ret = fcntl(channel_monitor_pipe, F_SETFL, |
| 1131 | flags | O_NONBLOCK); |
| 1132 | if (ret == -1) { |
| 1133 | PERROR("fcntl set O_NONBLOCK flag of the channel monitoring pipe"); |
| 1134 | goto error_fatal; |
| 1135 | } |
| 1136 | DBG("Channel monitor pipe set as non-blocking"); |
| 1137 | } else { |
| 1138 | ret_code = LTTCOMM_CONSUMERD_ALREADY_SET; |
| 1139 | } |
| 1140 | ret = consumer_send_status_msg(sock, ret_code); |
| 1141 | if (ret < 0) { |
| 1142 | goto error_fatal; |
| 1143 | } |
| 1144 | break; |
| 1145 | } |
| 1146 | case LTTNG_CONSUMER_ROTATE_CHANNEL: |
| 1147 | { |
| 1148 | struct lttng_consumer_channel *channel; |
| 1149 | uint64_t key = msg.u.rotate_channel.key; |
| 1150 | |
| 1151 | DBG("Consumer rotate channel %" PRIu64, key); |
| 1152 | |
| 1153 | channel = consumer_find_channel(key); |
| 1154 | if (!channel) { |
| 1155 | ERR("Channel %" PRIu64 " not found", key); |
| 1156 | ret_code = LTTCOMM_CONSUMERD_CHAN_NOT_FOUND; |
| 1157 | } else { |
| 1158 | /* |
| 1159 | * Sample the rotate position of all the streams in this channel. |
| 1160 | */ |
| 1161 | ret = lttng_consumer_rotate_channel(channel, key, |
| 1162 | msg.u.rotate_channel.relayd_id, |
| 1163 | msg.u.rotate_channel.metadata, |
| 1164 | ctx); |
| 1165 | if (ret < 0) { |
| 1166 | ERR("Rotate channel failed"); |
| 1167 | ret_code = LTTCOMM_CONSUMERD_ROTATION_FAIL; |
| 1168 | } |
| 1169 | |
| 1170 | health_code_update(); |
| 1171 | } |
| 1172 | ret = consumer_send_status_msg(sock, ret_code); |
| 1173 | if (ret < 0) { |
| 1174 | /* Somehow, the session daemon is not responding anymore. */ |
| 1175 | goto error_rotate_channel; |
| 1176 | } |
| 1177 | if (channel) { |
| 1178 | /* Rotate the streams that are ready right now. */ |
| 1179 | ret = lttng_consumer_rotate_ready_streams( |
| 1180 | channel, key, ctx); |
| 1181 | if (ret < 0) { |
| 1182 | ERR("Rotate ready streams failed"); |
| 1183 | } |
| 1184 | } |
| 1185 | break; |
| 1186 | error_rotate_channel: |
| 1187 | goto end_nosignal; |
| 1188 | } |
| 1189 | case LTTNG_CONSUMER_INIT: |
| 1190 | { |
| 1191 | ret_code = lttng_consumer_init_command(ctx, |
| 1192 | msg.u.init.sessiond_uuid); |
| 1193 | health_code_update(); |
| 1194 | ret = consumer_send_status_msg(sock, ret_code); |
| 1195 | if (ret < 0) { |
| 1196 | /* Somehow, the session daemon is not responding anymore. */ |
| 1197 | goto end_nosignal; |
| 1198 | } |
| 1199 | break; |
| 1200 | } |
| 1201 | case LTTNG_CONSUMER_CREATE_TRACE_CHUNK: |
| 1202 | { |
| 1203 | const struct lttng_credentials credentials = { |
| 1204 | .uid = msg.u.create_trace_chunk.credentials.value.uid, |
| 1205 | .gid = msg.u.create_trace_chunk.credentials.value.gid, |
| 1206 | }; |
| 1207 | const bool is_local_trace = |
| 1208 | !msg.u.create_trace_chunk.relayd_id.is_set; |
| 1209 | const uint64_t relayd_id = |
| 1210 | msg.u.create_trace_chunk.relayd_id.value; |
| 1211 | const char *chunk_override_name = |
| 1212 | *msg.u.create_trace_chunk.override_name ? |
| 1213 | msg.u.create_trace_chunk.override_name : |
| 1214 | NULL; |
| 1215 | LTTNG_OPTIONAL(struct lttng_directory_handle) chunk_directory_handle = |
| 1216 | LTTNG_OPTIONAL_INIT; |
| 1217 | |
| 1218 | /* |
| 1219 | * The session daemon will only provide a chunk directory file |
| 1220 | * descriptor for local traces. |
| 1221 | */ |
| 1222 | if (is_local_trace) { |
| 1223 | int chunk_dirfd; |
| 1224 | |
| 1225 | /* Acnowledge the reception of the command. */ |
| 1226 | ret = consumer_send_status_msg(sock, |
| 1227 | LTTCOMM_CONSUMERD_SUCCESS); |
| 1228 | if (ret < 0) { |
| 1229 | /* Somehow, the session daemon is not responding anymore. */ |
| 1230 | goto end_nosignal; |
| 1231 | } |
| 1232 | |
| 1233 | ret = lttcomm_recv_fds_unix_sock(sock, &chunk_dirfd, 1); |
| 1234 | if (ret != sizeof(chunk_dirfd)) { |
| 1235 | ERR("Failed to receive trace chunk directory file descriptor"); |
| 1236 | goto error_fatal; |
| 1237 | } |
| 1238 | |
| 1239 | DBG("Received trace chunk directory fd (%d)", |
| 1240 | chunk_dirfd); |
| 1241 | ret = lttng_directory_handle_init_from_dirfd( |
| 1242 | &chunk_directory_handle.value, |
| 1243 | chunk_dirfd); |
| 1244 | if (ret) { |
| 1245 | ERR("Failed to initialize chunk directory handle from directory file descriptor"); |
| 1246 | if (close(chunk_dirfd)) { |
| 1247 | PERROR("Failed to close chunk directory file descriptor"); |
| 1248 | } |
| 1249 | goto error_fatal; |
| 1250 | } |
| 1251 | chunk_directory_handle.is_set = true; |
| 1252 | } |
| 1253 | |
| 1254 | ret_code = lttng_consumer_create_trace_chunk( |
| 1255 | !is_local_trace ? &relayd_id : NULL, |
| 1256 | msg.u.create_trace_chunk.session_id, |
| 1257 | msg.u.create_trace_chunk.chunk_id, |
| 1258 | (time_t) msg.u.create_trace_chunk |
| 1259 | .creation_timestamp, |
| 1260 | chunk_override_name, |
| 1261 | msg.u.create_trace_chunk.credentials.is_set ? |
| 1262 | &credentials : |
| 1263 | NULL, |
| 1264 | chunk_directory_handle.is_set ? |
| 1265 | &chunk_directory_handle.value : |
| 1266 | NULL); |
| 1267 | |
| 1268 | if (chunk_directory_handle.is_set) { |
| 1269 | lttng_directory_handle_fini( |
| 1270 | &chunk_directory_handle.value); |
| 1271 | } |
| 1272 | goto end_msg_sessiond; |
| 1273 | } |
| 1274 | case LTTNG_CONSUMER_CLOSE_TRACE_CHUNK: |
| 1275 | { |
| 1276 | enum lttng_trace_chunk_command_type close_command = |
| 1277 | msg.u.close_trace_chunk.close_command.value; |
| 1278 | const uint64_t relayd_id = |
| 1279 | msg.u.close_trace_chunk.relayd_id.value; |
| 1280 | struct lttcomm_consumer_close_trace_chunk_reply reply; |
| 1281 | char path[LTTNG_PATH_MAX]; |
| 1282 | |
| 1283 | ret_code = lttng_consumer_close_trace_chunk( |
| 1284 | msg.u.close_trace_chunk.relayd_id.is_set ? |
| 1285 | &relayd_id : |
| 1286 | NULL, |
| 1287 | msg.u.close_trace_chunk.session_id, |
| 1288 | msg.u.close_trace_chunk.chunk_id, |
| 1289 | (time_t) msg.u.close_trace_chunk.close_timestamp, |
| 1290 | msg.u.close_trace_chunk.close_command.is_set ? |
| 1291 | &close_command : |
| 1292 | NULL, path); |
| 1293 | reply.ret_code = ret_code; |
| 1294 | reply.path_length = strlen(path) + 1; |
| 1295 | ret = lttcomm_send_unix_sock(sock, &reply, sizeof(reply)); |
| 1296 | if (ret != sizeof(reply)) { |
| 1297 | goto error_fatal; |
| 1298 | } |
| 1299 | ret = lttcomm_send_unix_sock(sock, path, reply.path_length); |
| 1300 | if (ret != reply.path_length) { |
| 1301 | goto error_fatal; |
| 1302 | } |
| 1303 | goto end_nosignal; |
| 1304 | } |
| 1305 | case LTTNG_CONSUMER_TRACE_CHUNK_EXISTS: |
| 1306 | { |
| 1307 | const uint64_t relayd_id = |
| 1308 | msg.u.trace_chunk_exists.relayd_id.value; |
| 1309 | |
| 1310 | ret_code = lttng_consumer_trace_chunk_exists( |
| 1311 | msg.u.trace_chunk_exists.relayd_id.is_set ? |
| 1312 | &relayd_id : NULL, |
| 1313 | msg.u.trace_chunk_exists.session_id, |
| 1314 | msg.u.trace_chunk_exists.chunk_id); |
| 1315 | goto end_msg_sessiond; |
| 1316 | } |
| 1317 | default: |
| 1318 | goto end_nosignal; |
| 1319 | } |
| 1320 | |
| 1321 | end_nosignal: |
| 1322 | /* |
| 1323 | * Return 1 to indicate success since the 0 value can be a socket |
| 1324 | * shutdown during the recv() or send() call. |
| 1325 | */ |
| 1326 | ret = 1; |
| 1327 | goto end; |
| 1328 | error_fatal: |
| 1329 | /* This will issue a consumer stop. */ |
| 1330 | ret = -1; |
| 1331 | goto end; |
| 1332 | end_msg_sessiond: |
| 1333 | /* |
| 1334 | * The returned value here is not useful since either way we'll return 1 to |
| 1335 | * the caller because the session daemon socket management is done |
| 1336 | * elsewhere. Returning a negative code or 0 will shutdown the consumer. |
| 1337 | */ |
| 1338 | ret = consumer_send_status_msg(sock, ret_code); |
| 1339 | if (ret < 0) { |
| 1340 | goto error_fatal; |
| 1341 | } |
| 1342 | ret = 1; |
| 1343 | end: |
| 1344 | health_code_update(); |
| 1345 | rcu_read_unlock(); |
| 1346 | return ret; |
| 1347 | } |
| 1348 | |
| 1349 | /* |
| 1350 | * Populate index values of a kernel stream. Values are set in big endian order. |
| 1351 | * |
| 1352 | * Return 0 on success or else a negative value. |
| 1353 | */ |
| 1354 | static int get_index_values(struct ctf_packet_index *index, int infd) |
| 1355 | { |
| 1356 | int ret; |
| 1357 | uint64_t packet_size, content_size, timestamp_begin, timestamp_end, |
| 1358 | events_discarded, stream_id, stream_instance_id, |
| 1359 | packet_seq_num; |
| 1360 | |
| 1361 | ret = kernctl_get_timestamp_begin(infd, ×tamp_begin); |
| 1362 | if (ret < 0) { |
| 1363 | PERROR("kernctl_get_timestamp_begin"); |
| 1364 | goto error; |
| 1365 | } |
| 1366 | |
| 1367 | ret = kernctl_get_timestamp_end(infd, ×tamp_end); |
| 1368 | if (ret < 0) { |
| 1369 | PERROR("kernctl_get_timestamp_end"); |
| 1370 | goto error; |
| 1371 | } |
| 1372 | |
| 1373 | ret = kernctl_get_events_discarded(infd, &events_discarded); |
| 1374 | if (ret < 0) { |
| 1375 | PERROR("kernctl_get_events_discarded"); |
| 1376 | goto error; |
| 1377 | } |
| 1378 | |
| 1379 | ret = kernctl_get_content_size(infd, &content_size); |
| 1380 | if (ret < 0) { |
| 1381 | PERROR("kernctl_get_content_size"); |
| 1382 | goto error; |
| 1383 | } |
| 1384 | |
| 1385 | ret = kernctl_get_packet_size(infd, &packet_size); |
| 1386 | if (ret < 0) { |
| 1387 | PERROR("kernctl_get_packet_size"); |
| 1388 | goto error; |
| 1389 | } |
| 1390 | |
| 1391 | ret = kernctl_get_stream_id(infd, &stream_id); |
| 1392 | if (ret < 0) { |
| 1393 | PERROR("kernctl_get_stream_id"); |
| 1394 | goto error; |
| 1395 | } |
| 1396 | |
| 1397 | ret = kernctl_get_instance_id(infd, &stream_instance_id); |
| 1398 | if (ret < 0) { |
| 1399 | if (ret == -ENOTTY) { |
| 1400 | /* Command not implemented by lttng-modules. */ |
| 1401 | stream_instance_id = -1ULL; |
| 1402 | } else { |
| 1403 | PERROR("kernctl_get_instance_id"); |
| 1404 | goto error; |
| 1405 | } |
| 1406 | } |
| 1407 | |
| 1408 | ret = kernctl_get_sequence_number(infd, &packet_seq_num); |
| 1409 | if (ret < 0) { |
| 1410 | if (ret == -ENOTTY) { |
| 1411 | /* Command not implemented by lttng-modules. */ |
| 1412 | packet_seq_num = -1ULL; |
| 1413 | ret = 0; |
| 1414 | } else { |
| 1415 | PERROR("kernctl_get_sequence_number"); |
| 1416 | goto error; |
| 1417 | } |
| 1418 | } |
| 1419 | index->packet_seq_num = htobe64(index->packet_seq_num); |
| 1420 | |
| 1421 | *index = (typeof(*index)) { |
| 1422 | .offset = index->offset, |
| 1423 | .packet_size = htobe64(packet_size), |
| 1424 | .content_size = htobe64(content_size), |
| 1425 | .timestamp_begin = htobe64(timestamp_begin), |
| 1426 | .timestamp_end = htobe64(timestamp_end), |
| 1427 | .events_discarded = htobe64(events_discarded), |
| 1428 | .stream_id = htobe64(stream_id), |
| 1429 | .stream_instance_id = htobe64(stream_instance_id), |
| 1430 | .packet_seq_num = htobe64(packet_seq_num), |
| 1431 | }; |
| 1432 | |
| 1433 | error: |
| 1434 | return ret; |
| 1435 | } |
| 1436 | /* |
| 1437 | * Sync metadata meaning request them to the session daemon and snapshot to the |
| 1438 | * metadata thread can consumer them. |
| 1439 | * |
| 1440 | * Metadata stream lock MUST be acquired. |
| 1441 | * |
| 1442 | * Return 0 if new metadatda is available, EAGAIN if the metadata stream |
| 1443 | * is empty or a negative value on error. |
| 1444 | */ |
| 1445 | int lttng_kconsumer_sync_metadata(struct lttng_consumer_stream *metadata) |
| 1446 | { |
| 1447 | int ret; |
| 1448 | |
| 1449 | assert(metadata); |
| 1450 | |
| 1451 | ret = kernctl_buffer_flush(metadata->wait_fd); |
| 1452 | if (ret < 0) { |
| 1453 | ERR("Failed to flush kernel stream"); |
| 1454 | goto end; |
| 1455 | } |
| 1456 | |
| 1457 | ret = kernctl_snapshot(metadata->wait_fd); |
| 1458 | if (ret < 0) { |
| 1459 | if (ret != -EAGAIN) { |
| 1460 | ERR("Sync metadata, taking kernel snapshot failed."); |
| 1461 | goto end; |
| 1462 | } |
| 1463 | DBG("Sync metadata, no new kernel metadata"); |
| 1464 | /* No new metadata, exit. */ |
| 1465 | ret = ENODATA; |
| 1466 | goto end; |
| 1467 | } |
| 1468 | |
| 1469 | end: |
| 1470 | return ret; |
| 1471 | } |
| 1472 | |
| 1473 | static |
| 1474 | int update_stream_stats(struct lttng_consumer_stream *stream) |
| 1475 | { |
| 1476 | int ret; |
| 1477 | uint64_t seq, discarded; |
| 1478 | |
| 1479 | ret = kernctl_get_sequence_number(stream->wait_fd, &seq); |
| 1480 | if (ret < 0) { |
| 1481 | if (ret == -ENOTTY) { |
| 1482 | /* Command not implemented by lttng-modules. */ |
| 1483 | seq = -1ULL; |
| 1484 | stream->sequence_number_unavailable = true; |
| 1485 | } else { |
| 1486 | PERROR("kernctl_get_sequence_number"); |
| 1487 | goto end; |
| 1488 | } |
| 1489 | } |
| 1490 | |
| 1491 | /* |
| 1492 | * Start the sequence when we extract the first packet in case we don't |
| 1493 | * start at 0 (for example if a consumer is not connected to the |
| 1494 | * session immediately after the beginning). |
| 1495 | */ |
| 1496 | if (stream->last_sequence_number == -1ULL) { |
| 1497 | stream->last_sequence_number = seq; |
| 1498 | } else if (seq > stream->last_sequence_number) { |
| 1499 | stream->chan->lost_packets += seq - |
| 1500 | stream->last_sequence_number - 1; |
| 1501 | } else { |
| 1502 | /* seq <= last_sequence_number */ |
| 1503 | ERR("Sequence number inconsistent : prev = %" PRIu64 |
| 1504 | ", current = %" PRIu64, |
| 1505 | stream->last_sequence_number, seq); |
| 1506 | ret = -1; |
| 1507 | goto end; |
| 1508 | } |
| 1509 | stream->last_sequence_number = seq; |
| 1510 | |
| 1511 | ret = kernctl_get_events_discarded(stream->wait_fd, &discarded); |
| 1512 | if (ret < 0) { |
| 1513 | PERROR("kernctl_get_events_discarded"); |
| 1514 | goto end; |
| 1515 | } |
| 1516 | if (discarded < stream->last_discarded_events) { |
| 1517 | /* |
| 1518 | * Overflow has occurred. We assume only one wrap-around |
| 1519 | * has occurred. |
| 1520 | */ |
| 1521 | stream->chan->discarded_events += (1ULL << (CAA_BITS_PER_LONG - 1)) - |
| 1522 | stream->last_discarded_events + discarded; |
| 1523 | } else { |
| 1524 | stream->chan->discarded_events += discarded - |
| 1525 | stream->last_discarded_events; |
| 1526 | } |
| 1527 | stream->last_discarded_events = discarded; |
| 1528 | ret = 0; |
| 1529 | |
| 1530 | end: |
| 1531 | return ret; |
| 1532 | } |
| 1533 | |
| 1534 | /* |
| 1535 | * Check if the local version of the metadata stream matches with the version |
| 1536 | * of the metadata stream in the kernel. If it was updated, set the reset flag |
| 1537 | * on the stream. |
| 1538 | */ |
| 1539 | static |
| 1540 | int metadata_stream_check_version(int infd, struct lttng_consumer_stream *stream) |
| 1541 | { |
| 1542 | int ret; |
| 1543 | uint64_t cur_version; |
| 1544 | |
| 1545 | ret = kernctl_get_metadata_version(infd, &cur_version); |
| 1546 | if (ret < 0) { |
| 1547 | if (ret == -ENOTTY) { |
| 1548 | /* |
| 1549 | * LTTng-modules does not implement this |
| 1550 | * command. |
| 1551 | */ |
| 1552 | ret = 0; |
| 1553 | goto end; |
| 1554 | } |
| 1555 | ERR("Failed to get the metadata version"); |
| 1556 | goto end; |
| 1557 | } |
| 1558 | |
| 1559 | if (stream->metadata_version == cur_version) { |
| 1560 | ret = 0; |
| 1561 | goto end; |
| 1562 | } |
| 1563 | |
| 1564 | DBG("New metadata version detected"); |
| 1565 | stream->metadata_version = cur_version; |
| 1566 | stream->reset_metadata_flag = 1; |
| 1567 | ret = 0; |
| 1568 | |
| 1569 | end: |
| 1570 | return ret; |
| 1571 | } |
| 1572 | |
| 1573 | /* |
| 1574 | * Consume data on a file descriptor and write it on a trace file. |
| 1575 | * The stream and channel locks must be held by the caller. |
| 1576 | */ |
| 1577 | ssize_t lttng_kconsumer_read_subbuffer(struct lttng_consumer_stream *stream, |
| 1578 | struct lttng_consumer_local_data *ctx) |
| 1579 | { |
| 1580 | unsigned long len, subbuf_size, padding; |
| 1581 | int err, write_index = 1, rotation_ret; |
| 1582 | ssize_t ret = 0; |
| 1583 | int infd = stream->wait_fd; |
| 1584 | struct ctf_packet_index index = {}; |
| 1585 | |
| 1586 | DBG("In read_subbuffer (infd : %d)", infd); |
| 1587 | |
| 1588 | /* |
| 1589 | * If the stream was flagged to be ready for rotation before we extract the |
| 1590 | * next packet, rotate it now. |
| 1591 | */ |
| 1592 | if (stream->rotate_ready) { |
| 1593 | DBG("Rotate stream before extracting data"); |
| 1594 | rotation_ret = lttng_consumer_rotate_stream(ctx, stream); |
| 1595 | if (rotation_ret < 0) { |
| 1596 | ERR("Stream rotation error"); |
| 1597 | ret = -1; |
| 1598 | goto error; |
| 1599 | } |
| 1600 | } |
| 1601 | |
| 1602 | /* Get the next subbuffer */ |
| 1603 | err = kernctl_get_next_subbuf(infd); |
| 1604 | if (err != 0) { |
| 1605 | /* |
| 1606 | * This is a debug message even for single-threaded consumer, |
| 1607 | * because poll() have more relaxed criterions than get subbuf, |
| 1608 | * so get_subbuf may fail for short race windows where poll() |
| 1609 | * would issue wakeups. |
| 1610 | */ |
| 1611 | DBG("Reserving sub buffer failed (everything is normal, " |
| 1612 | "it is due to concurrency)"); |
| 1613 | ret = err; |
| 1614 | goto error; |
| 1615 | } |
| 1616 | |
| 1617 | /* Get the full subbuffer size including padding */ |
| 1618 | err = kernctl_get_padded_subbuf_size(infd, &len); |
| 1619 | if (err != 0) { |
| 1620 | PERROR("Getting sub-buffer len failed."); |
| 1621 | err = kernctl_put_subbuf(infd); |
| 1622 | if (err != 0) { |
| 1623 | if (err == -EFAULT) { |
| 1624 | PERROR("Error in unreserving sub buffer\n"); |
| 1625 | } else if (err == -EIO) { |
| 1626 | /* Should never happen with newer LTTng versions */ |
| 1627 | PERROR("Reader has been pushed by the writer, last sub-buffer corrupted."); |
| 1628 | } |
| 1629 | ret = err; |
| 1630 | goto error; |
| 1631 | } |
| 1632 | ret = err; |
| 1633 | goto error; |
| 1634 | } |
| 1635 | |
| 1636 | if (!stream->metadata_flag) { |
| 1637 | ret = get_index_values(&index, infd); |
| 1638 | if (ret < 0) { |
| 1639 | err = kernctl_put_subbuf(infd); |
| 1640 | if (err != 0) { |
| 1641 | if (err == -EFAULT) { |
| 1642 | PERROR("Error in unreserving sub buffer\n"); |
| 1643 | } else if (err == -EIO) { |
| 1644 | /* Should never happen with newer LTTng versions */ |
| 1645 | PERROR("Reader has been pushed by the writer, last sub-buffer corrupted."); |
| 1646 | } |
| 1647 | ret = err; |
| 1648 | goto error; |
| 1649 | } |
| 1650 | goto error; |
| 1651 | } |
| 1652 | ret = update_stream_stats(stream); |
| 1653 | if (ret < 0) { |
| 1654 | err = kernctl_put_subbuf(infd); |
| 1655 | if (err != 0) { |
| 1656 | if (err == -EFAULT) { |
| 1657 | PERROR("Error in unreserving sub buffer\n"); |
| 1658 | } else if (err == -EIO) { |
| 1659 | /* Should never happen with newer LTTng versions */ |
| 1660 | PERROR("Reader has been pushed by the writer, last sub-buffer corrupted."); |
| 1661 | } |
| 1662 | ret = err; |
| 1663 | goto error; |
| 1664 | } |
| 1665 | goto error; |
| 1666 | } |
| 1667 | } else { |
| 1668 | write_index = 0; |
| 1669 | ret = metadata_stream_check_version(infd, stream); |
| 1670 | if (ret < 0) { |
| 1671 | err = kernctl_put_subbuf(infd); |
| 1672 | if (err != 0) { |
| 1673 | if (err == -EFAULT) { |
| 1674 | PERROR("Error in unreserving sub buffer\n"); |
| 1675 | } else if (err == -EIO) { |
| 1676 | /* Should never happen with newer LTTng versions */ |
| 1677 | PERROR("Reader has been pushed by the writer, last sub-buffer corrupted."); |
| 1678 | } |
| 1679 | ret = err; |
| 1680 | goto error; |
| 1681 | } |
| 1682 | goto error; |
| 1683 | } |
| 1684 | } |
| 1685 | |
| 1686 | switch (stream->chan->output) { |
| 1687 | case CONSUMER_CHANNEL_SPLICE: |
| 1688 | /* |
| 1689 | * XXX: The lttng-modules splice "actor" does not handle copying |
| 1690 | * partial pages hence only using the subbuffer size without the |
| 1691 | * padding makes the splice fail. |
| 1692 | */ |
| 1693 | subbuf_size = len; |
| 1694 | padding = 0; |
| 1695 | |
| 1696 | /* splice the subbuffer to the tracefile */ |
| 1697 | ret = lttng_consumer_on_read_subbuffer_splice(ctx, stream, subbuf_size, |
| 1698 | padding, &index); |
| 1699 | /* |
| 1700 | * XXX: Splice does not support network streaming so the return value |
| 1701 | * is simply checked against subbuf_size and not like the mmap() op. |
| 1702 | */ |
| 1703 | if (ret != subbuf_size) { |
| 1704 | /* |
| 1705 | * display the error but continue processing to try |
| 1706 | * to release the subbuffer |
| 1707 | */ |
| 1708 | ERR("Error splicing to tracefile (ret: %zd != len: %lu)", |
| 1709 | ret, subbuf_size); |
| 1710 | write_index = 0; |
| 1711 | } |
| 1712 | break; |
| 1713 | case CONSUMER_CHANNEL_MMAP: |
| 1714 | { |
| 1715 | const char *subbuf_addr; |
| 1716 | |
| 1717 | /* Get subbuffer size without padding */ |
| 1718 | err = kernctl_get_subbuf_size(infd, &subbuf_size); |
| 1719 | if (err != 0) { |
| 1720 | PERROR("Getting sub-buffer len failed."); |
| 1721 | err = kernctl_put_subbuf(infd); |
| 1722 | if (err != 0) { |
| 1723 | if (err == -EFAULT) { |
| 1724 | PERROR("Error in unreserving sub buffer\n"); |
| 1725 | } else if (err == -EIO) { |
| 1726 | /* Should never happen with newer LTTng versions */ |
| 1727 | PERROR("Reader has been pushed by the writer, last sub-buffer corrupted."); |
| 1728 | } |
| 1729 | ret = err; |
| 1730 | goto error; |
| 1731 | } |
| 1732 | ret = err; |
| 1733 | goto error; |
| 1734 | } |
| 1735 | |
| 1736 | ret = get_current_subbuf_addr(stream, &subbuf_addr); |
| 1737 | if (ret) { |
| 1738 | goto error_put_subbuf; |
| 1739 | } |
| 1740 | |
| 1741 | /* Make sure the tracer is not gone mad on us! */ |
| 1742 | assert(len >= subbuf_size); |
| 1743 | |
| 1744 | padding = len - subbuf_size; |
| 1745 | |
| 1746 | /* write the subbuffer to the tracefile */ |
| 1747 | ret = lttng_consumer_on_read_subbuffer_mmap(ctx, stream, |
| 1748 | subbuf_addr, |
| 1749 | subbuf_size, |
| 1750 | padding, &index); |
| 1751 | /* |
| 1752 | * The mmap operation should write subbuf_size amount of data when |
| 1753 | * network streaming or the full padding (len) size when we are _not_ |
| 1754 | * streaming. |
| 1755 | */ |
| 1756 | if ((ret != subbuf_size && stream->net_seq_idx != (uint64_t) -1ULL) || |
| 1757 | (ret != len && stream->net_seq_idx == (uint64_t) -1ULL)) { |
| 1758 | /* |
| 1759 | * Display the error but continue processing to try to release the |
| 1760 | * subbuffer. This is a DBG statement since this is possible to |
| 1761 | * happen without being a critical error. |
| 1762 | */ |
| 1763 | DBG("Error writing to tracefile " |
| 1764 | "(ret: %zd != len: %lu != subbuf_size: %lu)", |
| 1765 | ret, len, subbuf_size); |
| 1766 | write_index = 0; |
| 1767 | } |
| 1768 | break; |
| 1769 | } |
| 1770 | default: |
| 1771 | ERR("Unknown output method"); |
| 1772 | ret = -EPERM; |
| 1773 | } |
| 1774 | error_put_subbuf: |
| 1775 | err = kernctl_put_next_subbuf(infd); |
| 1776 | if (err != 0) { |
| 1777 | if (err == -EFAULT) { |
| 1778 | PERROR("Error in unreserving sub buffer\n"); |
| 1779 | } else if (err == -EIO) { |
| 1780 | /* Should never happen with newer LTTng versions */ |
| 1781 | PERROR("Reader has been pushed by the writer, last sub-buffer corrupted."); |
| 1782 | } |
| 1783 | ret = err; |
| 1784 | goto error; |
| 1785 | } |
| 1786 | |
| 1787 | /* Write index if needed. */ |
| 1788 | if (!write_index) { |
| 1789 | goto rotate; |
| 1790 | } |
| 1791 | |
| 1792 | if (stream->chan->live_timer_interval && !stream->metadata_flag) { |
| 1793 | /* |
| 1794 | * In live, block until all the metadata is sent. |
| 1795 | */ |
| 1796 | pthread_mutex_lock(&stream->metadata_timer_lock); |
| 1797 | assert(!stream->missed_metadata_flush); |
| 1798 | stream->waiting_on_metadata = true; |
| 1799 | pthread_mutex_unlock(&stream->metadata_timer_lock); |
| 1800 | |
| 1801 | err = consumer_stream_sync_metadata(ctx, stream->session_id); |
| 1802 | |
| 1803 | pthread_mutex_lock(&stream->metadata_timer_lock); |
| 1804 | stream->waiting_on_metadata = false; |
| 1805 | if (stream->missed_metadata_flush) { |
| 1806 | stream->missed_metadata_flush = false; |
| 1807 | pthread_mutex_unlock(&stream->metadata_timer_lock); |
| 1808 | (void) consumer_flush_kernel_index(stream); |
| 1809 | } else { |
| 1810 | pthread_mutex_unlock(&stream->metadata_timer_lock); |
| 1811 | } |
| 1812 | if (err < 0) { |
| 1813 | goto error; |
| 1814 | } |
| 1815 | } |
| 1816 | |
| 1817 | err = consumer_stream_write_index(stream, &index); |
| 1818 | if (err < 0) { |
| 1819 | goto error; |
| 1820 | } |
| 1821 | |
| 1822 | rotate: |
| 1823 | /* |
| 1824 | * After extracting the packet, we check if the stream is now ready to be |
| 1825 | * rotated and perform the action immediately. |
| 1826 | */ |
| 1827 | rotation_ret = lttng_consumer_stream_is_rotate_ready(stream); |
| 1828 | if (rotation_ret == 1) { |
| 1829 | rotation_ret = lttng_consumer_rotate_stream(ctx, stream); |
| 1830 | if (rotation_ret < 0) { |
| 1831 | ERR("Stream rotation error"); |
| 1832 | ret = -1; |
| 1833 | goto error; |
| 1834 | } |
| 1835 | } else if (rotation_ret < 0) { |
| 1836 | ERR("Checking if stream is ready to rotate"); |
| 1837 | ret = -1; |
| 1838 | goto error; |
| 1839 | } |
| 1840 | |
| 1841 | error: |
| 1842 | return ret; |
| 1843 | } |
| 1844 | |
| 1845 | int lttng_kconsumer_on_recv_stream(struct lttng_consumer_stream *stream) |
| 1846 | { |
| 1847 | int ret; |
| 1848 | |
| 1849 | assert(stream); |
| 1850 | |
| 1851 | /* |
| 1852 | * Don't create anything if this is set for streaming or if there is |
| 1853 | * no current trace chunk on the parent channel. |
| 1854 | */ |
| 1855 | if (stream->net_seq_idx == (uint64_t) -1ULL && stream->chan->monitor && |
| 1856 | stream->chan->trace_chunk) { |
| 1857 | ret = consumer_stream_create_output_files(stream, true); |
| 1858 | if (ret) { |
| 1859 | goto error; |
| 1860 | } |
| 1861 | } |
| 1862 | |
| 1863 | if (stream->output == LTTNG_EVENT_MMAP) { |
| 1864 | /* get the len of the mmap region */ |
| 1865 | unsigned long mmap_len; |
| 1866 | |
| 1867 | ret = kernctl_get_mmap_len(stream->wait_fd, &mmap_len); |
| 1868 | if (ret != 0) { |
| 1869 | PERROR("kernctl_get_mmap_len"); |
| 1870 | goto error_close_fd; |
| 1871 | } |
| 1872 | stream->mmap_len = (size_t) mmap_len; |
| 1873 | |
| 1874 | stream->mmap_base = mmap(NULL, stream->mmap_len, PROT_READ, |
| 1875 | MAP_PRIVATE, stream->wait_fd, 0); |
| 1876 | if (stream->mmap_base == MAP_FAILED) { |
| 1877 | PERROR("Error mmaping"); |
| 1878 | ret = -1; |
| 1879 | goto error_close_fd; |
| 1880 | } |
| 1881 | } |
| 1882 | |
| 1883 | /* we return 0 to let the library handle the FD internally */ |
| 1884 | return 0; |
| 1885 | |
| 1886 | error_close_fd: |
| 1887 | if (stream->out_fd >= 0) { |
| 1888 | int err; |
| 1889 | |
| 1890 | err = close(stream->out_fd); |
| 1891 | assert(!err); |
| 1892 | stream->out_fd = -1; |
| 1893 | } |
| 1894 | error: |
| 1895 | return ret; |
| 1896 | } |
| 1897 | |
| 1898 | /* |
| 1899 | * Check if data is still being extracted from the buffers for a specific |
| 1900 | * stream. Consumer data lock MUST be acquired before calling this function |
| 1901 | * and the stream lock. |
| 1902 | * |
| 1903 | * Return 1 if the traced data are still getting read else 0 meaning that the |
| 1904 | * data is available for trace viewer reading. |
| 1905 | */ |
| 1906 | int lttng_kconsumer_data_pending(struct lttng_consumer_stream *stream) |
| 1907 | { |
| 1908 | int ret; |
| 1909 | |
| 1910 | assert(stream); |
| 1911 | |
| 1912 | if (stream->endpoint_status != CONSUMER_ENDPOINT_ACTIVE) { |
| 1913 | ret = 0; |
| 1914 | goto end; |
| 1915 | } |
| 1916 | |
| 1917 | ret = kernctl_get_next_subbuf(stream->wait_fd); |
| 1918 | if (ret == 0) { |
| 1919 | /* There is still data so let's put back this subbuffer. */ |
| 1920 | ret = kernctl_put_subbuf(stream->wait_fd); |
| 1921 | assert(ret == 0); |
| 1922 | ret = 1; /* Data is pending */ |
| 1923 | goto end; |
| 1924 | } |
| 1925 | |
| 1926 | /* Data is NOT pending and ready to be read. */ |
| 1927 | ret = 0; |
| 1928 | |
| 1929 | end: |
| 1930 | return ret; |
| 1931 | } |