git.sesse.net Git - cubemap/blob - httpinput.cpp

   1 #include <assert.h>
   2 #include <errno.h>
   3 #include <fcntl.h>
   4 #include <math.h>
   5 #include <netdb.h>
   6 #include <netinet/in.h>
   7 #include <poll.h>
   8 #include <stdint.h>
   9 #include <stdlib.h>
  10 #include <string.h>
  11 #include <sys/ioctl.h>
  12 #include <sys/socket.h>
  13 #include <sys/time.h>
  14 #include <sys/types.h>
  15 #include <sys/wait.h>
  16 #include <time.h>
  17 #include <unistd.h>
  18 #include <math.h>
  19 #include <spawn.h>
  20 #include <map>
  21 #include <string>
  22 #include <utility>
  23 #include <vector>
  24
  25 #include "httpinput.h"
  26 #include "log.h"
  27 #include "metacube2.h"
  28 #include "parse.h"
  29 #include "serverpool.h"
  30 #include "state.pb.h"
  31 #include "stream.h"
  32 #include "timespec.h"
  33 #include "util.h"
  34 #include "version.h"
  35
  36 using namespace std;
  37
  38 namespace {
  39
  40 string host_header(const string &host, const string &port)
  41 {
  42         if (port == "http" || atoi(port.c_str()) == 80) {
  43                 return host;
  44         } else {
  45                 return host + ":" + port;
  46         }
  47 }
  48
  49 }  // namespace
  50
  51 extern ServerPool *servers;
  52
  53 HTTPInput::HTTPInput(const string &url, Input::Encoding encoding)
  54         : state(NOT_CONNECTED),
  55           url(url),
  56           encoding(encoding)
  57 {
  58         stats.url = url;
  59         stats.bytes_received = 0;
  60         stats.data_bytes_received = 0;
  61         stats.metadata_bytes_received = 0;
  62         stats.connect_time = -1;
  63         stats.latency_sec = HUGE_VAL;
  64 }
  65
  66 HTTPInput::HTTPInput(const InputProto &serialized)
  67         : state(State(serialized.state())),
  68           url(serialized.url()),
  69           encoding(serialized.is_metacube_encoded() ?
  70                    Input::INPUT_ENCODING_METACUBE :
  71                    Input::INPUT_ENCODING_RAW),
  72           request(serialized.request()),
  73           request_bytes_sent(serialized.request_bytes_sent()),
  74           response(serialized.response()),
  75           http_header(serialized.http_header()),
  76           stream_header(serialized.stream_header()),
  77           has_metacube_header(serialized.has_metacube_header()),
  78           sock(serialized.sock()),
  79           child_pid(serialized.child_pid())
  80 {
  81         // Set back the close-on-exec flag for the socket.
  82         // (This can't leak into a child, since we haven't been started yet.)
  83         fcntl(sock, F_SETFD, 1);
  84
  85         pending_data.resize(serialized.pending_data().size());
  86         memcpy(&pending_data[0], serialized.pending_data().data(), serialized.pending_data().size());
  87
  88         string protocol, user;
  89         parse_url(url, &protocol, &user, &host, &port, &path);  // Don't care if it fails.
  90
  91         stats.url = url;
  92         stats.bytes_received = serialized.bytes_received();
  93         stats.data_bytes_received = serialized.data_bytes_received();
  94         stats.metadata_bytes_received = serialized.metadata_bytes_received();
  95         if (serialized.has_connect_time()) {
  96                 stats.connect_time = serialized.connect_time();
  97         } else {
  98                 stats.connect_time = time(nullptr);
  99         }
 100         if (serialized.has_latency_sec()) {
 101                 stats.latency_sec = serialized.latency_sec();
 102         } else {
 103                 stats.latency_sec = HUGE_VAL;
 104         }
 105
 106         last_verbose_connection.tv_sec = -3600;
 107         last_verbose_connection.tv_nsec = 0;
 108 }
 109
 110 void HTTPInput::close_socket()
 111 {
 112         if (sock != -1) {
 113                 safe_close(sock);
 114                 sock = -1;
 115         }
 116         if (child_pid != -1) {
 117                 // Kill the child process group, forcibly.
 118                 // TODO: Consider using a pidfd on newer kernels, so that we're guaranteed
 119                 // never to kill the wrong process.
 120                 kill(-child_pid, SIGKILL);
 121         }
 122         child_pid = -1;
 123
 124         lock_guard<mutex> lock(stats_mutex);
 125         stats.connect_time = -1;
 126 }
 127
 128 InputProto HTTPInput::serialize() const
 129 {
 130         // Unset the close-on-exec flag for the socket.
 131         // (This can't leak into a child, since there's only one thread left.)
 132         fcntl(sock, F_SETFD, 0);
 133
 134         InputProto serialized;
 135         serialized.set_state(state);
 136         serialized.set_url(url);
 137         serialized.set_request(request);
 138         serialized.set_request_bytes_sent(request_bytes_sent);
 139         serialized.set_response(response);
 140         serialized.set_http_header(http_header);
 141         serialized.set_stream_header(stream_header);
 142         serialized.set_pending_data(string(pending_data.begin(), pending_data.end()));
 143         serialized.set_has_metacube_header(has_metacube_header);
 144         serialized.set_sock(sock);
 145         serialized.set_child_pid(child_pid);
 146         serialized.set_bytes_received(stats.bytes_received);
 147         serialized.set_data_bytes_received(stats.data_bytes_received);
 148         if (isfinite(stats.latency_sec)) {
 149                 serialized.set_latency_sec(stats.latency_sec);
 150         }
 151         serialized.set_connect_time(stats.connect_time);
 152         if (encoding == Input::INPUT_ENCODING_METACUBE) {
 153                 serialized.set_is_metacube_encoded(true);
 154         } else {
 155                 assert(encoding == Input::INPUT_ENCODING_RAW);
 156                 serialized.set_is_metacube_encoded(false);
 157         }
 158         return serialized;
 159 }
 160
 161 int HTTPInput::lookup_and_connect(const string &host, const string &port)
 162 {
 163         addrinfo *ai;
 164         int err = getaddrinfo(host.c_str(), port.c_str(), nullptr, &ai);
 165         if (err != 0) {
 166                 if (!suppress_logging) {
 167                         log(WARNING, "[%s] Lookup of '%s' failed (%s).",
 168                                 url.c_str(), host.c_str(), gai_strerror(err));
 169                 }
 170                 return -1;
 171         }
 172
 173         addrinfo *base_ai = ai;
 174
 175         // Connect to everything in turn until we have a socket.
 176         for ( ; ai && !should_stop(); ai = ai->ai_next) {
 177                 // Now do a non-blocking connect. This is important because we want to be able to be
 178                 // woken up, even though it's rather cumbersome.
 179                 int sock = socket(ai->ai_family, SOCK_STREAM | SOCK_NONBLOCK | SOCK_CLOEXEC, IPPROTO_TCP);
 180                 if (sock == -1) {
 181                         // Could be e.g. EPROTONOSUPPORT. The show must go on.
 182                         continue;
 183                 }
 184
 185                 // Do a non-blocking connect.
 186                 do {
 187                         err = connect(sock, ai->ai_addr, ai->ai_addrlen);
 188                 } while (err == -1 && errno == EINTR);
 189
 190                 if (err == -1 && errno != EINPROGRESS) {
 191                         log_perror("connect");
 192                         safe_close(sock);
 193                         continue;
 194                 }
 195
 196                 // Wait for the connect to complete, or an error to happen.
 197                 for ( ;; ) {
 198                         bool complete = wait_for_activity(sock, POLLIN | POLLOUT, nullptr);
 199                         if (should_stop()) {
 200                                 safe_close(sock);
 201                                 return -1;
 202                         }
 203                         if (complete) {
 204                                 break;
 205                         }
 206                 }
 207
 208                 // Check whether it ended in an error or not.
 209                 socklen_t err_size = sizeof(err);
 210                 if (getsockopt(sock, SOL_SOCKET, SO_ERROR, &err, &err_size) == -1) {
 211                         log_perror("getsockopt");
 212                         safe_close(sock);
 213                         continue;
 214                 }
 215
 216                 errno = err;
 217
 218                 if (err == 0) {
 219                         // Successful connect.
 220                         freeaddrinfo(base_ai);
 221                         return sock;
 222                 }
 223
 224                 safe_close(sock);
 225         }
 226
 227         // Give the last one as error.
 228         if (!suppress_logging) {
 229                 log(WARNING, "[%s] Connect to '%s' failed (%s)",
 230                         url.c_str(), host.c_str(), strerror(errno));
 231         }
 232         freeaddrinfo(base_ai);
 233         return -1;
 234 }
 235
 236 int HTTPInput::open_child_process(const string &cmdline)
 237 {
 238         int devnullfd = open("/dev/null", O_RDONLY | O_CLOEXEC);
 239         if (devnullfd == -1) {
 240                 log_perror("/dev/null");
 241                 return -1;
 242         }
 243
 244         int pipefd[2];
 245         if (pipe2(pipefd, O_CLOEXEC) == -1) {
 246                 log_perror("pipe2()");
 247                 close(devnullfd);
 248                 return -1;
 249         }
 250
 251         // Point stdout to us, stdin to /dev/null, and stderr remains where it is
 252         // (probably the systemd log). All other file descriptors should be marked
 253         // as close-on-exec, and should thus not leak into the child.
 254         posix_spawn_file_actions_t actions;
 255         posix_spawn_file_actions_init(&actions);
 256         posix_spawn_file_actions_adddup2(&actions, devnullfd, 0);
 257         posix_spawn_file_actions_adddup2(&actions, pipefd[1], 1);
 258
 259         // Make the process a leader of its own process group, so that we can easily
 260         // kill it and any of its child processes (unless it's started new process
 261         // groups itself, of course).
 262         posix_spawnattr_t attr;
 263         posix_spawnattr_init(&attr);
 264         posix_spawnattr_setflags(&attr, POSIX_SPAWN_SETPGROUP);
 265         posix_spawnattr_setpgroup(&attr, 0);
 266
 267         char * const argv[] = {
 268                 strdup("/bin/sh"),
 269                 strdup("-c"),
 270                 strdup(path.c_str()),
 271                 nullptr
 272         };
 273         int err = posix_spawn(&child_pid, "/bin/sh", &actions, &attr, argv, /*envp=*/nullptr);
 274         posix_spawn_file_actions_destroy(&actions);
 275         posix_spawnattr_destroy(&attr);
 276         free(argv[0]);
 277         free(argv[1]);
 278         free(argv[2]);
 279         close(pipefd[1]);
 280
 281         if (err == 0) {
 282                 return pipefd[0];
 283         } else {
 284                 child_pid = -1;
 285                 log_perror(cmdline.c_str());
 286                 close(pipefd[0]);
 287                 return -1;
 288         }
 289 }
 290
 291 bool HTTPInput::parse_response(const string &request)
 292 {
 293         vector<string> lines = split_lines(response);
 294         if (lines.empty()) {
 295                 if (!suppress_logging) {
 296                         log(WARNING, "[%s] Empty HTTP response from input.", url.c_str());
 297                 }
 298                 return false;
 299         }
 300
 301         vector<string> first_line_tokens = split_tokens(lines[0]);
 302         if (first_line_tokens.size() < 2) {
 303                 if (!suppress_logging) {
 304                         log(WARNING, "[%s] Malformed response line '%s' from input.",
 305                                 url.c_str(), lines[0].c_str());
 306                 }
 307                 return false;
 308         }
 309
 310         int response = atoi(first_line_tokens[1].c_str());
 311         if (response != 200) {
 312                 if (!suppress_logging) {
 313                         log(WARNING, "[%s] Non-200 response '%s' from input.",
 314                                 url.c_str(), lines[0].c_str());
 315                 }
 316                 return false;
 317         }
 318
 319         HTTPHeaderMultimap parameters = extract_headers(lines, url);
 320
 321         // Remove “Content-encoding: metacube”.
 322         const auto encoding_it = parameters.find("Content-Encoding");
 323         if (encoding_it != parameters.end() && encoding_it->second == "metacube") {
 324                 parameters.erase(encoding_it);
 325         }
 326
 327         // Change “Server: foo” to “Server: metacube/0.1 (reflecting: foo)”
 328         // XXX: Use a Via: instead?
 329         if (parameters.count("Server") == 0) {
 330                 parameters.insert(make_pair("Server", SERVER_IDENTIFICATION));
 331         } else {
 332                 for (auto &key_and_value : parameters) {
 333                         if (key_and_value.first != "Server") {
 334                                 continue;
 335                         }
 336                         key_and_value.second = SERVER_IDENTIFICATION " (reflecting: " + key_and_value.second + ")";
 337                 }
 338         }
 339
 340         // Erase “Connection: close”; we'll set it on the sending side if needed.
 341         parameters.erase("Connection");
 342
 343         // Construct the new HTTP header.
 344         http_header = "HTTP/1.0 200 OK\r\n";
 345         for (const auto &key_and_value : parameters) {
 346                 http_header.append(key_and_value.first + ": " + key_and_value.second + "\r\n");
 347         }
 348
 349         for (int stream_index : stream_indices) {
 350                 servers->set_header(stream_index, http_header, stream_header);
 351         }
 352
 353         return true;
 354 }
 355
 356 void HTTPInput::do_work()
 357 {
 358         timespec last_activity;
 359
 360         // TODO: Make the timeout persist across restarts.
 361         if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 362                 int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &last_activity);
 363                 assert(err != -1);
 364         }
 365
 366         while (!should_stop()) {
 367                 if (state == SENDING_REQUEST || state == RECEIVING_HEADER || state == RECEIVING_DATA) {
 368                         // Give the socket 30 seconds since last activity before we time out.
 369                         static const int timeout_secs = 30;
 370
 371                         timespec now;
 372                         int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &now);
 373                         assert(err != -1);
 374
 375                         timespec elapsed = clock_diff(last_activity, now);
 376                         if (elapsed.tv_sec >= timeout_secs) {
 377                                 // Timeout!
 378                                 if (!suppress_logging) {
 379                                         log(ERROR, "[%s] Timeout after %d seconds, closing.", url.c_str(), elapsed.tv_sec);
 380                                 }
 381                                 state = CLOSING_SOCKET;
 382                                 continue;
 383                         }
 384
 385                         // Basically calculate (30 - (now - last_activity)) = (30 + (last_activity - now)).
 386                         // Add a second of slack to account for differences between clocks.
 387                         timespec timeout = clock_diff(now, last_activity);
 388                         timeout.tv_sec += timeout_secs + 1;
 389                         assert(timeout.tv_sec > 0 || (timeout.tv_sec >= 0 && timeout.tv_nsec > 0));
 390
 391                         bool activity = wait_for_activity(sock, (state == SENDING_REQUEST) ? POLLOUT : POLLIN, &timeout);
 392                         if (activity) {
 393                                 err = clock_gettime(CLOCK_MONOTONIC_COARSE, &last_activity);
 394                                 assert(err != -1);
 395                         } else {
 396                                 // OK. Most likely, should_stop was set, or we have timed out.
 397                                 continue;
 398                         }
 399                 }
 400
 401                 switch (state) {
 402                 case NOT_CONNECTED: {
 403                         // Reap any exited children.
 404                         int wstatus, err;
 405                         do {
 406                                 err = waitpid(-1, &wstatus, WNOHANG);
 407                                 if (err == -1) {
 408                                         if (errno == EINTR) {
 409                                                 continue;
 410                                         }
 411                                         if (errno == ECHILD) {
 412                                                 break;
 413                                         }
 414                                         log_perror("waitpid");
 415                                         break;
 416                                 }
 417                         } while (err != 0);
 418                         child_pid = -1;
 419
 420                         request.clear();
 421                         request_bytes_sent = 0;
 422                         response.clear();
 423                         pending_data.clear();
 424                         has_metacube_header = false;
 425                         for (int stream_index : stream_indices) {
 426                                 // Don't zero out the header; it might still be of use to HLS clients.
 427                                 servers->set_unavailable(stream_index);
 428                         }
 429
 430                         string protocol;
 431                         {
 432                                 string user;  // Thrown away.
 433                                 if (!parse_url(url, &protocol, &user, &host, &port, &path)) {
 434                                         if (!suppress_logging) {
 435                                                 log(WARNING, "[%s] Failed to parse URL '%s'", url.c_str(), url.c_str());
 436                                         }
 437                                         break;
 438                                 }
 439
 440                                 // Remove the brackets around IPv6 address literals.
 441                                 // TODO: See if we can join this with the code in parse_ip_address(),
 442                                 // or maybe even more it into parse_url().
 443                                 if (!host.empty() && host[0] == '[' && host[host.size() - 1] == ']') {
 444                                         host = host.substr(1, host.size() - 2);
 445                                 }
 446                         }
 447
 448                         if (suppress_logging) {
 449                                 // See if there's more than one minute since last time we made a connection
 450                                 // with logging enabled. If so, turn it on again.
 451                                 timespec now;
 452                                 int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &now);
 453                                 assert(err != -1);
 454
 455                                 double elapsed = now.tv_sec - last_verbose_connection.tv_sec +
 456                                         1e-9 * (now.tv_nsec - last_verbose_connection.tv_nsec);
 457                                 if (elapsed > 60.0) {
 458                                         suppress_logging = false;
 459                                 }
 460                         }
 461                         if (!suppress_logging) {
 462                                 int err = clock_gettime(CLOCK_MONOTONIC_COARSE, &last_verbose_connection);
 463                                 assert(err != -1);
 464                         }
 465                         ++num_connection_attempts;
 466                         if (protocol == "pipe") {
 467                                 sock = open_child_process(path.c_str());
 468
 469                                 if (sock != -1) {
 470                                         // Construct a minimal HTTP header.
 471                                         http_header = "HTTP/1.0 200 OK\r\n";
 472                                         for (int stream_index : stream_indices) {
 473                                                 servers->set_header(stream_index, http_header, stream_header);
 474                                         }
 475                                         state = RECEIVING_DATA;
 476                                 }
 477                         } else {
 478                                 sock = lookup_and_connect(host, port);
 479                                 if (sock != -1) {
 480                                         // Yay, successful connect.
 481                                         state = SENDING_REQUEST;
 482                                         request = "GET " + path + " HTTP/1.0\r\nHost: " + host_header(host, port) + "\r\nUser-Agent: cubemap\r\n\r\n";
 483                                         request_bytes_sent = 0;
 484                                 }
 485                         }
 486                         if (sock != -1) {
 487                                 lock_guard<mutex> lock(stats_mutex);
 488                                 stats.connect_time = time(nullptr);
 489                                 clock_gettime(CLOCK_MONOTONIC_COARSE, &last_activity);
 490                         }
 491                         break;
 492                 }
 493                 case SENDING_REQUEST: {
 494                         size_t to_send = request.size() - request_bytes_sent;
 495                         int ret;
 496
 497                         do {
 498                                 ret = write(sock, request.data() + request_bytes_sent, to_send);
 499                         } while (ret == -1 && errno == EINTR);
 500
 501                         if (ret == -1) {
 502                                 log_perror("write");
 503                                 state = CLOSING_SOCKET;
 504                                 continue;
 505                         }
 506
 507                         assert(ret >= 0);
 508                         request_bytes_sent += ret;
 509
 510                         if (request_bytes_sent == request.size()) {
 511                                 state = RECEIVING_HEADER;
 512                         }
 513                         break;
 514                 }
 515                 case RECEIVING_HEADER: {
 516                         char buf[4096];
 517                         int ret;
 518
 519                         do {
 520                                 ret = read(sock, buf, sizeof(buf));
 521                         } while (ret == -1 && errno == EINTR);
 522
 523                         if (ret == -1) {
 524                                 log_perror("read");
 525                                 state = CLOSING_SOCKET;
 526                                 continue;
 527                         }
 528
 529                         if (ret == 0) {
 530                                 // This really shouldn't happen...
 531                                 if (!suppress_logging) {
 532                                         log(ERROR, "[%s] Socket unexpectedly closed while reading header",
 533                                                    url.c_str());
 534                                 }
 535                                 state = CLOSING_SOCKET;
 536                                 continue;
 537                         }
 538
 539                         RequestParseStatus status = wait_for_double_newline(&response, buf, ret);
 540
 541                         if (status == RP_OUT_OF_SPACE) {
 542                                 if (!suppress_logging) {
 543                                         log(WARNING, "[%s] Server sent overlong HTTP response!", url.c_str());
 544                                 }
 545                                 state = CLOSING_SOCKET;
 546                                 continue;
 547                         } else if (status == RP_NOT_FINISHED_YET) {
 548                                 continue;
 549                         }
 550
 551                         // OK, so we're fine, but there might be some of the actual data after the response.
 552                         // We'll need to deal with that separately.
 553                         string extra_data;
 554                         if (status == RP_EXTRA_DATA) {
 555                                 char *ptr = static_cast<char *>(
 556                                         memmem(response.data(), response.size(), "\r\n\r\n", 4));
 557                                 assert(ptr != nullptr);
 558                                 extra_data = string(ptr + 4, &response[0] + response.size());
 559                                 response.resize(ptr - response.data());
 560                         }
 561
 562                         if (!parse_response(response)) {
 563                                 state = CLOSING_SOCKET;
 564                                 continue;
 565                         }
 566
 567                         if (!extra_data.empty()) {
 568                                 process_data(&extra_data[0], extra_data.size());
 569                         }
 570
 571                         if (!suppress_logging) {
 572                                 if (encoding == Input::INPUT_ENCODING_RAW) {
 573                                         log(INFO, "[%s] Connected to '%s', receiving raw data.",
 574                                                    url.c_str(), url.c_str());
 575                                 } else {
 576                                         assert(encoding == Input::INPUT_ENCODING_METACUBE);
 577                                         log(INFO, "[%s] Connected to '%s', receiving data.",
 578                                                    url.c_str(), url.c_str());
 579                                 }
 580                         }
 581                         state = RECEIVING_DATA;
 582                         break;
 583                 }
 584                 case RECEIVING_DATA: {
 585                         char buf[4096];
 586                         int ret;
 587
 588                         do {
 589                                 ret = read(sock, buf, sizeof(buf));
 590                         } while (ret == -1 && errno == EINTR);
 591
 592                         if (ret == -1) {
 593                                 log_perror("read");
 594                                 state = CLOSING_SOCKET;
 595                                 continue;
 596                         }
 597
 598                         if (ret == 0) {
 599                                 // This really shouldn't happen...
 600                                 if (!suppress_logging) {
 601                                         log(ERROR, "[%s] Socket unexpectedly closed while reading data",
 602                                                    url.c_str());
 603                                 }
 604                                 state = CLOSING_SOCKET;
 605                                 continue;
 606                         }
 607
 608                         num_connection_attempts = 0;  // Reset, since we have a successful read.
 609                         if (suppress_logging) {
 610                                 // This was suppressed earlier, so print it out now.
 611                                 if (encoding == Input::INPUT_ENCODING_RAW) {
 612                                         log(INFO, "[%s] Connected to '%s', receiving raw data.",
 613                                                    url.c_str(), url.c_str());
 614                                 } else {
 615                                         assert(encoding == Input::INPUT_ENCODING_METACUBE);
 616                                         log(INFO, "[%s] Connected to '%s', receiving data.",
 617                                                    url.c_str(), url.c_str());
 618                                 }
 619                                 suppress_logging = false;
 620                         }
 621
 622                         process_data(buf, ret);
 623                         break;
 624                 }
 625                 case CLOSING_SOCKET: {
 626                         close_socket();
 627                         state = NOT_CONNECTED;
 628                         break;
 629                 }
 630                 default:
 631                         assert(false);
 632                 }
 633
 634                 // If we are still in NOT_CONNECTED, either something went wrong,
 635                 // or the connection just got closed.
 636                 // The earlier steps have already given the error message, if any.
 637                 if (state == NOT_CONNECTED && !should_stop()) {
 638                         if (!suppress_logging) {
 639                                 log(INFO, "[%s] Waiting 0.2 seconds and restarting...", url.c_str());
 640                         }
 641
 642                         if (num_connection_attempts >= 3 && !suppress_logging) {
 643                                 log(INFO, "[%s] %d failed connection attempts, suppressing logging for one minute.",
 644                                         url.c_str(), num_connection_attempts);
 645                                 suppress_logging = true;
 646                         }
 647                         timespec timeout_ts;
 648                         timeout_ts.tv_sec = 0;
 649                         timeout_ts.tv_nsec = 200000000;
 650                         wait_for_wakeup(&timeout_ts);
 651                 }
 652         }
 653 }
 654
 655 void HTTPInput::process_data(char *ptr, size_t bytes)
 656 {
 657         {
 658                 lock_guard<mutex> lock(stats_mutex);
 659                 stats.bytes_received += bytes;
 660         }
 661
 662         if (encoding == Input::INPUT_ENCODING_RAW) {
 663                 for (int stream_index : stream_indices) {
 664                         servers->add_data(stream_index, ptr, bytes, /*metacube_flags=*/0, /*pts=*/RationalPTS());
 665                 }
 666                 return;
 667         }
 668
 669         assert(encoding == Input::INPUT_ENCODING_METACUBE);
 670         pending_data.insert(pending_data.end(), ptr, ptr + bytes);
 671
 672         for ( ;; ) {
 673                 // If we don't have enough data (yet) for even the Metacube header, just return.
 674                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 675                         return;
 676                 }
 677
 678                 // Make sure we have the Metacube sync header at the start.
 679                 // We may need to skip over junk data (it _should_ not happen, though).
 680                 if (!has_metacube_header) {
 681                         char *ptr = static_cast<char *>(
 682                                 memmem(pending_data.data(), pending_data.size(),
 683                                        METACUBE2_SYNC, strlen(METACUBE2_SYNC)));
 684                         if (ptr == nullptr) {
 685                                 // OK, so we didn't find the sync marker. We know then that
 686                                 // we do not have the _full_ marker in the buffer, but we
 687                                 // could have N-1 bytes. Drop everything before that,
 688                                 // and then give up.
 689                                 drop_pending_data(pending_data.size() - (strlen(METACUBE2_SYNC) - 1));
 690                                 return;
 691                         } else {
 692                                 // Yay, we found the header. Drop everything (if anything) before it.
 693                                 drop_pending_data(ptr - pending_data.data());
 694                                 has_metacube_header = true;
 695
 696                                 // Re-check that we have the entire header; we could have dropped data.
 697                                 if (pending_data.size() < sizeof(metacube2_block_header)) {
 698                                         return;
 699                                 }
 700                         }
 701                 }
 702
 703                 // Now it's safe to read the header.
 704                 metacube2_block_header hdr;
 705                 memcpy(&hdr, pending_data.data(), sizeof(hdr));
 706                 assert(memcmp(hdr.sync, METACUBE2_SYNC, sizeof(hdr.sync)) == 0);
 707                 uint32_t size = ntohl(hdr.size);
 708                 uint16_t flags = ntohs(hdr.flags);
 709                 uint16_t expected_csum = metacube2_compute_crc(&hdr);
 710
 711                 if (expected_csum != ntohs(hdr.csum)) {
 712                         log(WARNING, "[%s] Metacube checksum failed (expected 0x%x, got 0x%x), "
 713                                 "not reading block claiming to be %d bytes (flags=%x).",
 714                                 url.c_str(), expected_csum, ntohs(hdr.csum),
 715                                 size, flags);
 716
 717                         // Drop only the first byte, and let the rest of the code handle resync.
 718                         pending_data.erase(pending_data.begin(), pending_data.begin() + 1);
 719                         has_metacube_header = false;
 720                         continue;
 721                 }
 722                 if (size > 10485760) {
 723                         log(WARNING, "[%s] Metacube block of %d bytes (flags=%x); corrupted header??",
 724                                 url.c_str(), size, flags);
 725                 }
 726
 727                 // See if we have the entire block. If not, wait for more data.
 728                 if (pending_data.size() < sizeof(metacube2_block_header) + size) {
 729                         return;
 730                 }
 731
 732                 // See if this is a metadata block. If so, we don't want to send it on,
 733                 // but rather process it ourselves.
 734                 // TODO: Keep metadata when sending on to other Metacube users.
 735                 if (flags & METACUBE_FLAGS_METADATA) {
 736                         {
 737                                 lock_guard<mutex> lock(stats_mutex);
 738                                 stats.metadata_bytes_received += size;
 739                         }
 740                         process_metacube_metadata_block(hdr, pending_data.data() + sizeof(hdr), size);
 741                 } else {
 742                         // Send this block on to the servers.
 743                         {
 744                                 lock_guard<mutex> lock(stats_mutex);
 745                                 stats.data_bytes_received += size;
 746                         }
 747                         char *inner_data = pending_data.data() + sizeof(metacube2_block_header);
 748                         if (flags & METACUBE_FLAGS_HEADER) {
 749                                 stream_header = string(inner_data, inner_data + size);
 750                                 for (int stream_index : stream_indices) {
 751                                         servers->set_header(stream_index, http_header, stream_header);
 752                                 }
 753                         }
 754                         for (int stream_index : stream_indices) {
 755                                 servers->add_data(stream_index, inner_data, size, flags, next_block_pts);
 756                         }
 757                         next_block_pts = RationalPTS();
 758                 }
 759
 760                 // Consume the block. This isn't the most efficient way of dealing with things
 761                 // should we have many blocks, but these routines don't need to be too efficient
 762                 // anyway.
 763                 pending_data.erase(pending_data.begin(), pending_data.begin() + sizeof(metacube2_block_header) + size);
 764                 has_metacube_header = false;
 765         }
 766 }
 767
 768 void HTTPInput::drop_pending_data(size_t num_bytes)
 769 {
 770         if (num_bytes == 0) {
 771                 return;
 772         }
 773         log(WARNING, "[%s] Dropping %lld junk bytes; not a Metacube2 stream, or data was dropped from the middle of the stream.",
 774                 url.c_str(), (long long)num_bytes);
 775         assert(pending_data.size() >= num_bytes);
 776         pending_data.erase(pending_data.begin(), pending_data.begin() + num_bytes);
 777 }
 778
 779 void HTTPInput::add_destination(int stream_index)
 780 {
 781         stream_indices.push_back(stream_index);
 782         servers->set_header(stream_index, http_header, stream_header);
 783 }
 784
 785 InputStats HTTPInput::get_stats() const
 786 {
 787         lock_guard<mutex> lock(stats_mutex);
 788         return stats;
 789 }
 790
 791 void HTTPInput::process_metacube_metadata_block(const metacube2_block_header &hdr, const char *payload, uint32_t payload_size)
 792 {
 793         if (payload_size < sizeof(uint64_t)) {
 794                 log(WARNING, "[%s] Undersized Metacube metadata block (%d bytes); corrupted header?",
 795                         url.c_str(), payload_size);
 796                 return;
 797         }
 798
 799         uint64_t type = be64toh(*(const uint64_t *)payload);
 800         if (type == METACUBE_METADATA_TYPE_ENCODER_TIMESTAMP) {
 801                 timespec now;
 802                 clock_gettime(CLOCK_REALTIME, &now);
 803
 804                 const metacube2_timestamp_packet *pkt = (const metacube2_timestamp_packet *)payload;
 805                 if (payload_size != sizeof(*pkt)) {
 806                         log(WARNING, "[%s] Metacube timestamp block of wrong size (%d bytes); ignoring.",
 807                                 url.c_str(), payload_size);
 808                         return;
 809                 }
 810
 811                 double elapsed = now.tv_sec - be64toh(pkt->tv_sec) +
 812                         1e-9 * (now.tv_nsec - long(be64toh(pkt->tv_nsec)));
 813                 {
 814                         lock_guard<mutex> lock(stats_mutex);
 815                         stats.latency_sec = elapsed;
 816                 }
 817         } else if (type == METACUBE_METADATA_TYPE_NEXT_BLOCK_PTS) {
 818                 const metacube2_pts_packet *pkt = (const metacube2_pts_packet *)payload;
 819                 if (payload_size != sizeof(*pkt)) {
 820                         log(WARNING, "[%s] Metacube pts block of wrong size (%d bytes); ignoring.",
 821                                 url.c_str(), payload_size);
 822                         return;
 823                 }
 824                 next_block_pts.pts = be64toh(pkt->pts);
 825                 next_block_pts.timebase_num = be64toh(pkt->timebase_num);
 826                 next_block_pts.timebase_den = be64toh(pkt->timebase_den);
 827         } else {
 828                 // Unknown metadata block, ignore
 829                 log(INFO, "[%s] Metadata block %llu\n", url.c_str(), type);
 830                 return;
 831         }
 832 }