src/common/uri.c

   1 /*
   2  * Copyright (C) 2012 David Goulet <dgoulet@efficios.com>
   3  *
   4  * SPDX-License-Identifier: GPL-2.0-only
   5  *
   6  */
   7
   8 #define _LGPL_SOURCE
   9 #include <assert.h>
  10 #include <arpa/inet.h>
  11 #include <common/compat/netdb.h>
  12 #include <stdlib.h>
  13 #include <string.h>
  14 #include <sys/socket.h>
  15
  16 #include <common/common.h>
  17 #include <common/defaults.h>
  18 #include <common/utils.h>
  19
  20 #include "uri.h"
  21
  22 #define LOOPBACK_ADDR_IPV4 "127.0.0.1"
  23 #define LOOPBACK_ADDR_IPV6 "::1"
  24
  25 enum uri_proto_code {
  26         P_NET, P_NET6, P_FILE, P_TCP, P_TCP6,
  27 };
  28
  29 struct uri_proto {
  30         const char *name;
  31         const char *leading_string;
  32         enum uri_proto_code code;
  33         enum lttng_proto_type type;
  34         enum lttng_dst_type dtype;
  35 };
  36
  37 /* Supported protocols */
  38 static const struct uri_proto proto_uri[] = {
  39         { .name = "file", .leading_string = "file://", .code = P_FILE, .type = 0, .dtype = LTTNG_DST_PATH },
  40         { .name = "net", .leading_string = "net://", .code = P_NET, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV4 },
  41         { .name = "net4", .leading_string = "net4://", .code = P_NET, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV4 },
  42         { .name = "net6", .leading_string = "net6://", .code = P_NET6, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV6 },
  43         { .name = "tcp", .leading_string = "tcp://", .code = P_TCP, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV4 },
  44         { .name = "tcp4", .leading_string = "tcp4://", .code = P_TCP, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV4 },
  45         { .name = "tcp6", .leading_string = "tcp6://", .code = P_TCP6, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV6 },
  46         /* Invalid proto marking the end of the array. */
  47         { NULL, NULL, 0, 0, 0 }
  48 };
  49
  50 /*
  51  * Return pointer to the character in s matching one of the characters in
  52  * accept. If nothing is found, return pointer to the end of string (eos).
  53  */
  54 static inline const char *strpbrk_or_eos(const char *s, const char *accept)
  55 {
  56         char *p = strpbrk(s, accept);
  57         if (p == NULL) {
  58                 p = strchr(s, '\0');
  59         }
  60
  61         return p;
  62 }
  63
  64 /*
  65  * Validate if proto is a supported protocol from proto_uri array.
  66  */
  67 static const struct uri_proto *get_uri_proto(const char *uri_str)
  68 {
  69         const struct uri_proto *supported = NULL;
  70
  71         /* Safety net */
  72         if (uri_str == NULL) {
  73                 goto end;
  74         }
  75
  76         for (supported = &proto_uri[0];
  77                         supported->leading_string != NULL; ++supported) {
  78                 if (strncasecmp(uri_str, supported->leading_string,
  79                                         strlen(supported->leading_string)) == 0) {
  80                         goto end;
  81                 }
  82         }
  83
  84         /* Proto not found */
  85         return NULL;
  86
  87 end:
  88         return supported;
  89 }
  90
  91 /*
  92  * Set network address from string into dst. Supports both IP string and
  93  * hostname.
  94  */
  95 static int set_ip_address(const char *addr, int af, char *dst, size_t size)
  96 {
  97         int ret;
  98         unsigned char buf[sizeof(struct in6_addr)];
  99         struct hostent *record;
 100
 101         assert(addr);
 102         assert(dst);
 103
 104         memset(dst, 0, size);
 105
 106         /* Network protocol */
 107         ret = inet_pton(af, addr, buf);
 108         if (ret < 1) {
 109                 /* We consider the dst to be an hostname or an invalid IP char */
 110                 record = lttng_gethostbyname2(addr, af);
 111                 if (record) {
 112                         /* Translate IP to string */
 113                         if (!inet_ntop(af, record->h_addr_list[0], dst, size)) {
 114                                 PERROR("inet_ntop");
 115                                 goto error;
 116                         }
 117                 } else if (!strcmp(addr, "localhost") &&
 118                                 (af == AF_INET || af == AF_INET6)) {
 119                         /*
 120                          * Some systems may not have "localhost" defined in
 121                          * accordance with IETF RFC 6761. According to this RFC,
 122                          * applications may recognize "localhost" names as
 123                          * special and resolve to the appropriate loopback
 124                          * address.
 125                          *
 126                          * We choose to use the system name resolution API first
 127                          * to honor its network configuration. If this fails, we
 128                          * resolve to the appropriate loopback address. This is
 129                          * done to accommodates systems which may want to start
 130                          * tracing before their network configured.
 131                          */
 132                         const char *loopback_addr = af == AF_INET ?
 133                                         LOOPBACK_ADDR_IPV4 : LOOPBACK_ADDR_IPV6;
 134                         const size_t loopback_addr_len = af == AF_INET ?
 135                                         sizeof(LOOPBACK_ADDR_IPV4) :
 136                                         sizeof(LOOPBACK_ADDR_IPV6);
 137
 138                         DBG2("Could not resolve localhost address, using fallback");
 139                         if (loopback_addr_len > size) {
 140                                 ERR("Could not resolve localhost address; destination string is too short");
 141                                 goto error;
 142                         }
 143                         strcpy(dst, loopback_addr);
 144                 } else {
 145                         /* At this point, the IP or the hostname is bad */
 146                         goto error;
 147                 }
 148         } else {
 149                 if (size > 0) {
 150                         strncpy(dst, addr, size);
 151                         dst[size - 1] = '\0';
 152                 }
 153         }
 154
 155         DBG2("IP address resolved to %s", dst);
 156         return 0;
 157
 158 error:
 159         ERR("URI parse bad hostname %s for af %d", addr, af);
 160         return -1;
 161 }
 162
 163 /*
 164  * Set default URI attribute which is basically the given stream type and the
 165  * default port if none is set in the URI.
 166  */
 167 static void set_default_uri_attr(struct lttng_uri *uri,
 168                 enum lttng_stream_type stype)
 169 {
 170         uri->stype = stype;
 171         if (uri->dtype != LTTNG_DST_PATH && uri->port == 0) {
 172                 uri->port = (stype == LTTNG_STREAM_CONTROL) ?
 173                         DEFAULT_NETWORK_CONTROL_PORT : DEFAULT_NETWORK_DATA_PORT;
 174         }
 175 }
 176
 177 /*
 178  * Compare two URL destination.
 179  *
 180  * Return 0 is equal else is not equal.
 181  */
 182 static int compare_destination(struct lttng_uri *ctrl, struct lttng_uri *data)
 183 {
 184         int ret;
 185
 186         assert(ctrl);
 187         assert(data);
 188
 189         switch (ctrl->dtype) {
 190         case LTTNG_DST_IPV4:
 191                 ret = strncmp(ctrl->dst.ipv4, data->dst.ipv4, sizeof(ctrl->dst.ipv4));
 192                 break;
 193         case LTTNG_DST_IPV6:
 194                 ret = strncmp(ctrl->dst.ipv6, data->dst.ipv6, sizeof(ctrl->dst.ipv6));
 195                 break;
 196         default:
 197                 ret = -1;
 198                 break;
 199         }
 200
 201         return ret;
 202 }
 203
 204 /*
 205  * Build a string URL from a lttng_uri object.
 206  */
 207 LTTNG_HIDDEN
 208 int uri_to_str_url(struct lttng_uri *uri, char *dst, size_t size)
 209 {
 210         int ipver, ret;
 211         const char *addr;
 212         char proto[5], port[7];
 213
 214         assert(uri);
 215         assert(dst);
 216
 217         if (uri->dtype == LTTNG_DST_PATH) {
 218                 ipver = 0;
 219                 addr = uri->dst.path;
 220                 (void) snprintf(proto, sizeof(proto), "file");
 221                 (void) snprintf(port, sizeof(port), "%s", "");
 222         } else {
 223                 ipver = (uri->dtype == LTTNG_DST_IPV4) ? 4 : 6;
 224                 addr = (ipver == 4) ?  uri->dst.ipv4 : uri->dst.ipv6;
 225                 (void) snprintf(proto, sizeof(proto), "tcp%d", ipver);
 226                 (void) snprintf(port, sizeof(port), ":%d", uri->port);
 227         }
 228
 229         ret = snprintf(dst, size, "%s://%s%s%s%s/%s", proto,
 230                         (ipver == 6) ? "[" : "", addr, (ipver == 6) ? "]" : "",
 231                         port, uri->subdir);
 232         if (ret < 0) {
 233                 PERROR("snprintf uri to url");
 234         }
 235
 236         return ret;
 237 }
 238
 239 /*
 240  * Compare two URIs.
 241  *
 242  * Return 0 if equal else 1.
 243  */
 244 LTTNG_HIDDEN
 245 int uri_compare(struct lttng_uri *uri1, struct lttng_uri *uri2)
 246 {
 247         return memcmp(uri1, uri2, sizeof(struct lttng_uri));
 248 }
 249
 250 /*
 251  * Free URI memory.
 252  */
 253 LTTNG_HIDDEN
 254 void uri_free(struct lttng_uri *uri)
 255 {
 256         free(uri);
 257 }
 258
 259 /*
 260  * Parses a string URI to a lttng_uri. This function can potentially return
 261  * more than one URI in uris so the size of the array is returned and uris is
 262  * allocated and populated. Caller must free(3) the array.
 263  *
 264  * This function can not detect the stream type of the URI so the caller has to
 265  * make sure the correct type (stype) is set on the return URI(s). The default
 266  * port must also be set by the caller if the returned URI has its port set to
 267  * zero.
 268  *
 269  * NOTE: A good part of the following code was inspired from the "wget" source
 270  * tree from the src/url.c file and url_parse() function. Also, the
 271  * strpbrk_or_eos() function found above is also inspired by the same code.
 272  * This code was originally licensed GPLv2 so we acknolwedge the Free Software
 273  * Foundation here for the work and to make sure we are compliant with it.
 274  */
 275 LTTNG_HIDDEN
 276 ssize_t uri_parse(const char *str_uri, struct lttng_uri **uris)
 277 {
 278         int ret, i = 0;
 279         /* Size of the uris array. Default is 1 */
 280         ssize_t size = 1;
 281         char subdir[PATH_MAX];
 282         unsigned int ctrl_port = 0;
 283         unsigned int data_port = 0;
 284         struct lttng_uri *tmp_uris;
 285         char *addr_f = NULL;
 286         const struct uri_proto *proto;
 287         const char *purl, *addr_e, *addr_b, *subdir_b = NULL;
 288         const char *seps = ":/\0";
 289
 290         /*
 291          * The first part is the protocol portion of a maximum of 5 bytes for now.
 292          * The second part is the hostname or IP address. The 255 bytes size is the
 293          * limit found in the RFC 1035 for the total length of a domain name
 294          * (https://www.ietf.org/rfc/rfc1035.txt). Finally, for the net://
 295          * protocol, two ports CAN be specified.
 296          */
 297
 298         DBG3("URI string: %s", str_uri);
 299
 300         proto = get_uri_proto(str_uri);
 301         if (proto == NULL) {
 302                 ERR("URI parse unknown protocol %s", str_uri);
 303                 goto error;
 304         }
 305
 306         purl = str_uri;
 307
 308         if (proto->code == P_NET || proto->code == P_NET6) {
 309                 /* Special case for net:// which requires two URI objects */
 310                 size = 2;
 311         }
 312
 313         /* Allocate URI array */
 314         tmp_uris = zmalloc(sizeof(struct lttng_uri) * size);
 315         if (tmp_uris == NULL) {
 316                 PERROR("zmalloc uri");
 317                 goto error;
 318         }
 319
 320         memset(subdir, 0, sizeof(subdir));
 321         purl += strlen(proto->leading_string);
 322
 323         /* Copy known value to the first URI. */
 324         tmp_uris[0].dtype = proto->dtype;
 325         tmp_uris[0].proto = proto->type;
 326
 327         if (proto->code == P_FILE) {
 328                 if (*purl != '/') {
 329                         ERR("Missing destination full path.");
 330                         goto free_error;
 331                 }
 332
 333                 strncpy(tmp_uris[0].dst.path, purl, sizeof(tmp_uris[0].dst.path));
 334                 tmp_uris[0].dst.path[sizeof(tmp_uris[0].dst.path) - 1] = '\0';
 335                 DBG3("URI file destination: %s", purl);
 336                 goto end;
 337         }
 338
 339         /* Assume we are at the beginning of an address or host of some sort. */
 340         addr_b = purl;
 341
 342         /*
 343          * Handle IPv6 address inside square brackets as mention by RFC 2732. IPv6
 344          * address that does not start AND end with brackets will be rejected even
 345          * if valid.
 346          *
 347          * proto://[<addr>]...
 348          *         ^
 349          */
 350         if (*purl == '[') {
 351                 /* Address begins after '[' */
 352                 addr_b = purl + 1;
 353                 addr_e = strchr(addr_b, ']');
 354                 if (addr_e == NULL || addr_b == addr_e) {
 355                         ERR("Broken IPv6 address %s", addr_b);
 356                         goto free_error;
 357                 }
 358
 359                 /* Moving parsed URL pointer after the final bracket ']' */
 360                 purl = addr_e + 1;
 361
 362                 /*
 363                  * The closing bracket must be followed by a seperator or NULL char.
 364                  */
 365                 if (strchr(seps, *purl) == NULL) {
 366                         ERR("Unknown symbol after IPv6 address: %s", purl);
 367                         goto free_error;
 368                 }
 369         } else {
 370                 purl = strpbrk_or_eos(purl, seps);
 371                 addr_e = purl;
 372         }
 373
 374         /* Check if we at least have a char for the addr or hostname. */
 375         if (addr_b == addr_e) {
 376                 ERR("No address or hostname detected.");
 377                 goto free_error;
 378         }
 379
 380         addr_f = utils_strdupdelim(addr_b, addr_e);
 381         if (addr_f == NULL) {
 382                 goto free_error;
 383         }
 384
 385         /*
 386          * Detect PORT after address. The net/net6 protocol allows up to two port
 387          * so we can define the control and data port.
 388          */
 389         while (*purl == ':') {
 390                 const char *port_b, *port_e;
 391                 char *port_f;
 392
 393                 /* Update pass counter */
 394                 i++;
 395
 396                 /*
 397                  * Maximum of two ports is possible if P_NET/NET6. Bigger than that,
 398                  * two much stuff.
 399                  */
 400                 if ((i == 2 && (proto->code != P_NET && proto->code != P_NET6))
 401                                 || i > 2) {
 402                         break;
 403                 }
 404
 405                 /*
 406                  * Move parsed URL to port value.
 407                  * proto://addr_host:PORT1:PORT2/foo/bar
 408                  *                   ^
 409                  */
 410                 ++purl;
 411                 port_b = purl;
 412                 purl = strpbrk_or_eos(purl, seps);
 413                 port_e = purl;
 414
 415                 if (port_b != port_e) {
 416                         int port;
 417
 418                         port_f = utils_strdupdelim(port_b, port_e);
 419                         if (port_f == NULL) {
 420                                 goto free_error;
 421                         }
 422
 423                         port = atoi(port_f);
 424                         if (port > 0xffff || port <= 0x0) {
 425                                 ERR("Invalid port number %d", port);
 426                                 free(port_f);
 427                                 goto free_error;
 428                         }
 429                         free(port_f);
 430
 431                         if (i == 1) {
 432                                 ctrl_port = port;
 433                         } else {
 434                                 data_port = port;
 435                         }
 436                 }
 437         };
 438
 439         /* Check for a valid subdir or trailing garbage */
 440         if (*purl == '/') {
 441                 /*
 442                  * Move to subdir value.
 443                  * proto://addr_host:PORT1:PORT2/foo/bar
 444                  *                               ^
 445                  */
 446                 ++purl;
 447                 subdir_b = purl;
 448         } else if (*purl != '\0') {
 449                 ERR("Trailing characters not recognized: %s", purl);
 450                 goto free_error;
 451         }
 452
 453         /* We have enough valid information to create URI(s) object */
 454
 455         /* Copy generic information */
 456         tmp_uris[0].port = ctrl_port;
 457
 458         /* Copy subdirectory if one. */
 459         if (subdir_b) {
 460                 strncpy(tmp_uris[0].subdir, subdir_b, sizeof(tmp_uris[0].subdir));
 461                 tmp_uris[0].subdir[sizeof(tmp_uris[0].subdir) - 1] = '\0';
 462         }
 463
 464         switch (proto->code) {
 465         case P_NET:
 466                 ret = set_ip_address(addr_f, AF_INET, tmp_uris[0].dst.ipv4,
 467                                 sizeof(tmp_uris[0].dst.ipv4));
 468                 if (ret < 0) {
 469                         goto free_error;
 470                 }
 471
 472                 memcpy(tmp_uris[1].dst.ipv4, tmp_uris[0].dst.ipv4, sizeof(tmp_uris[1].dst.ipv4));
 473
 474                 tmp_uris[1].dtype = proto->dtype;
 475                 tmp_uris[1].proto = proto->type;
 476                 tmp_uris[1].port = data_port;
 477                 break;
 478         case P_NET6:
 479                 ret = set_ip_address(addr_f, AF_INET6, tmp_uris[0].dst.ipv6,
 480                                 sizeof(tmp_uris[0].dst.ipv6));
 481                 if (ret < 0) {
 482                         goto free_error;
 483                 }
 484
 485                 memcpy(tmp_uris[1].dst.ipv6, tmp_uris[0].dst.ipv6, sizeof(tmp_uris[1].dst.ipv6));
 486
 487                 tmp_uris[1].dtype = proto->dtype;
 488                 tmp_uris[1].proto = proto->type;
 489                 tmp_uris[1].port = data_port;
 490                 break;
 491         case P_TCP:
 492                 ret = set_ip_address(addr_f, AF_INET, tmp_uris[0].dst.ipv4,
 493                                 sizeof(tmp_uris[0].dst.ipv4));
 494                 if (ret < 0) {
 495                         goto free_error;
 496                 }
 497                 break;
 498         case P_TCP6:
 499                 ret = set_ip_address(addr_f, AF_INET6, tmp_uris[0].dst.ipv6,
 500                                 sizeof(tmp_uris[0].dst.ipv6));
 501                 if (ret < 0) {
 502                         goto free_error;
 503                 }
 504                 break;
 505         default:
 506                 goto free_error;
 507         }
 508
 509 end:
 510         DBG3("URI dtype: %d, proto: %d, host: %s, subdir: %s, ctrl: %d, data: %d",
 511                         proto->dtype, proto->type, (addr_f == NULL) ? "" : addr_f,
 512                         (subdir_b == NULL) ? "" : subdir_b, ctrl_port, data_port);
 513
 514         free(addr_f);
 515
 516         *uris = tmp_uris;
 517         return size;
 518
 519 free_error:
 520         free(addr_f);
 521         free(tmp_uris);
 522 error:
 523         return -1;
 524 }
 525
 526 /*
 527  * Parse a string URL and creates URI(s) returning the size of the populated
 528  * array.
 529  */
 530 LTTNG_HIDDEN
 531 ssize_t uri_parse_str_urls(const char *ctrl_url, const char *data_url,
 532                 struct lttng_uri **uris)
 533 {
 534         unsigned int equal = 1, idx = 0;
 535         /* Add the "file://" size to the URL maximum size */
 536         char url[PATH_MAX + 7];
 537         ssize_t size_ctrl = 0, size_data = 0, size;
 538         struct lttng_uri *ctrl_uris = NULL, *data_uris = NULL;
 539         struct lttng_uri *tmp_uris = NULL;
 540
 541         /* No URL(s) is allowed. This means that the consumer will be disabled. */
 542         if (ctrl_url == NULL && data_url == NULL) {
 543                 return 0;
 544         }
 545
 546         /* Check if URLs are equal and if so, only use the control URL */
 547         if ((ctrl_url && *ctrl_url != '\0') && (data_url && *data_url != '\0')) {
 548                 equal = !strcmp(ctrl_url, data_url);
 549         }
 550
 551         /*
 552          * Since we allow the str_url to be a full local filesystem path, we are
 553          * going to create a valid file:// URL if it's the case.
 554          *
 555          * Check if first character is a '/' or else reject the URL.
 556          */
 557         if (ctrl_url && ctrl_url[0] == '/') {
 558                 int ret;
 559
 560                 ret = snprintf(url, sizeof(url), "file://%s", ctrl_url);
 561                 if (ret < 0) {
 562                         PERROR("snprintf file url");
 563                         goto parse_error;
 564                 } else if (ret >= sizeof(url)) {
 565                         PERROR("snprintf file url is too long");
 566                         goto parse_error;
 567
 568                 }
 569                 ctrl_url = url;
 570         }
 571
 572         /* Parse the control URL if there is one */
 573         if (ctrl_url && *ctrl_url != '\0') {
 574                 size_ctrl = uri_parse(ctrl_url, &ctrl_uris);
 575                 if (size_ctrl < 1) {
 576                         ERR("Unable to parse the URL %s", ctrl_url);
 577                         goto parse_error;
 578                 }
 579
 580                 /* At this point, we know there is at least one URI in the array */
 581                 set_default_uri_attr(&ctrl_uris[0], LTTNG_STREAM_CONTROL);
 582
 583                 if (ctrl_uris[0].dtype == LTTNG_DST_PATH &&
 584                                 (data_url && *data_url != '\0')) {
 585                         ERR("Cannot have a data URL when destination is file://");
 586                         goto error;
 587                 }
 588
 589                 /* URL are not equal but the control URL uses a net:// protocol */
 590                 if (size_ctrl == 2) {
 591                         if (!equal) {
 592                                 ERR("Control URL uses the net:// protocol and the data URL is "
 593                                                 "different. Not allowed.");
 594                                 goto error;
 595                         } else {
 596                                 set_default_uri_attr(&ctrl_uris[1], LTTNG_STREAM_DATA);
 597                                 /*
 598                                  * The data_url and ctrl_url are equal and the ctrl_url
 599                                  * contains a net:// protocol so we just skip the data part.
 600                                  */
 601                                 data_url = NULL;
 602                         }
 603                 }
 604         }
 605
 606         if (data_url && *data_url != '\0') {
 607                 int ret;
 608
 609                 /* We have to parse the data URL in this case */
 610                 size_data = uri_parse(data_url, &data_uris);
 611                 if (size_data < 1) {
 612                         ERR("Unable to parse the URL %s", data_url);
 613                         goto error;
 614                 } else if (size_data == 2) {
 615                         ERR("Data URL can not be set with the net[4|6]:// protocol");
 616                         goto error;
 617                 }
 618
 619                 set_default_uri_attr(&data_uris[0], LTTNG_STREAM_DATA);
 620
 621                 if (ctrl_uris) {
 622                         ret = compare_destination(&ctrl_uris[0], &data_uris[0]);
 623                         if (ret != 0) {
 624                                 ERR("Control and data destination mismatch");
 625                                 goto error;
 626                         }
 627                 }
 628         }
 629
 630         /* Compute total size */
 631         size = size_ctrl + size_data;
 632
 633         tmp_uris = zmalloc(sizeof(struct lttng_uri) * size);
 634         if (tmp_uris == NULL) {
 635                 PERROR("zmalloc uris");
 636                 goto error;
 637         }
 638
 639         if (ctrl_uris) {
 640                 /* It's possible the control URIs array contains more than one URI */
 641                 memcpy(tmp_uris, ctrl_uris, sizeof(struct lttng_uri) * size_ctrl);
 642                 ++idx;
 643                 free(ctrl_uris);
 644         }
 645
 646         if (data_uris) {
 647                 memcpy(&tmp_uris[idx], data_uris, sizeof(struct lttng_uri));
 648                 free(data_uris);
 649         }
 650
 651         *uris = tmp_uris;
 652
 653         return size;
 654
 655 error:
 656         free(ctrl_uris);
 657         free(data_uris);
 658         free(tmp_uris);
 659 parse_error:
 660         return -1;
 661 }