| 1 | /* |
| 2 | * Copyright (C) 2012 - David Goulet <dgoulet@efficios.com> |
| 3 | * |
| 4 | * This program is free software; you can redistribute it and/or modify it |
| 5 | * under the terms of the GNU General Public License, version 2 only, as |
| 6 | * published by the Free Software Foundation. |
| 7 | * |
| 8 | * This program is distributed in the hope that it will be useful, but WITHOUT |
| 9 | * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
| 10 | * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for |
| 11 | * more details. |
| 12 | * |
| 13 | * You should have received a copy of the GNU General Public License along with |
| 14 | * this program; if not, write to the Free Software Foundation, Inc., 51 |
| 15 | * Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. |
| 16 | */ |
| 17 | |
| 18 | #define _GNU_SOURCE |
| 19 | #include <assert.h> |
| 20 | #include <arpa/inet.h> |
| 21 | #include <netdb.h> |
| 22 | #include <stdlib.h> |
| 23 | #include <string.h> |
| 24 | #include <sys/socket.h> |
| 25 | |
| 26 | #include <common/common.h> |
| 27 | #include <common/defaults.h> |
| 28 | #include <common/utils.h> |
| 29 | |
| 30 | #include "uri.h" |
| 31 | |
| 32 | enum uri_proto_code { |
| 33 | P_NET, P_NET6, P_FILE, P_TCP, P_TCP6, |
| 34 | }; |
| 35 | |
| 36 | struct uri_proto { |
| 37 | const char *name; |
| 38 | const char *leading_string; |
| 39 | enum uri_proto_code code; |
| 40 | enum lttng_proto_type type; |
| 41 | enum lttng_dst_type dtype; |
| 42 | }; |
| 43 | |
| 44 | /* Supported protocols */ |
| 45 | static const struct uri_proto proto_uri[] = { |
| 46 | { .name = "file", .leading_string = "file://", .code = P_FILE, .type = 0, .dtype = LTTNG_DST_PATH }, |
| 47 | { .name = "net", .leading_string = "net://", .code = P_NET, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV4 }, |
| 48 | { .name = "net6", .leading_string = "net6://", .code = P_NET6, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV6 }, |
| 49 | { .name = "tcp", .leading_string = "tcp://", .code = P_TCP, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV4 }, |
| 50 | { .name = "tcp6", .leading_string = "tcp6://", .code = P_TCP6, .type = LTTNG_TCP, .dtype = LTTNG_DST_IPV6 }, |
| 51 | /* Invalid proto marking the end of the array. */ |
| 52 | { NULL, NULL, 0, 0, 0 } |
| 53 | }; |
| 54 | |
| 55 | /* |
| 56 | * Return pointer to the character in s matching one of the characters in |
| 57 | * accept. If nothing is found, return pointer to the end of string (eos). |
| 58 | */ |
| 59 | const inline char *strpbrk_or_eos(const char *s, const char *accept) |
| 60 | { |
| 61 | char *p = strpbrk(s, accept); |
| 62 | if (p == NULL) { |
| 63 | p = strchr(s, '\0'); |
| 64 | } |
| 65 | |
| 66 | return p; |
| 67 | } |
| 68 | |
| 69 | |
| 70 | /* |
| 71 | * Validate if proto is a supported protocol from proto_uri array. |
| 72 | */ |
| 73 | static const struct uri_proto *get_uri_proto(const char *uri_str) |
| 74 | { |
| 75 | const struct uri_proto *supported = NULL; |
| 76 | |
| 77 | /* Safety net */ |
| 78 | if (uri_str == NULL) { |
| 79 | goto end; |
| 80 | } |
| 81 | |
| 82 | for (supported = &proto_uri[0]; |
| 83 | supported->leading_string != NULL; ++supported) { |
| 84 | if (strncasecmp(uri_str, supported->leading_string, |
| 85 | strlen(supported->leading_string)) == 0) { |
| 86 | goto end; |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | /* Proto not found */ |
| 91 | return NULL; |
| 92 | |
| 93 | end: |
| 94 | return supported; |
| 95 | } |
| 96 | |
| 97 | /* |
| 98 | * Set network address from string into dst. Supports both IP string and |
| 99 | * hostname. |
| 100 | */ |
| 101 | static int set_ip_address(const char *addr, int af, char *dst, size_t size) |
| 102 | { |
| 103 | int ret; |
| 104 | unsigned char buf[sizeof(struct in6_addr)]; |
| 105 | struct hostent *record; |
| 106 | |
| 107 | assert(addr); |
| 108 | assert(dst); |
| 109 | |
| 110 | memset(dst, 0, size); |
| 111 | |
| 112 | /* Network protocol */ |
| 113 | ret = inet_pton(af, addr, buf); |
| 114 | if (ret < 1) { |
| 115 | /* We consider the dst to be an hostname or an invalid IP char */ |
| 116 | record = gethostbyname2(addr, af); |
| 117 | if (record == NULL) { |
| 118 | /* At this point, the IP or the hostname is bad */ |
| 119 | ERR("URI parse bad hostname %s for af %d", addr, af); |
| 120 | goto error; |
| 121 | } |
| 122 | |
| 123 | /* Translate IP to string */ |
| 124 | (void) inet_ntop(af, record->h_addr_list[0], dst, size); |
| 125 | } else { |
| 126 | if (size > 0) { |
| 127 | strncpy(dst, addr, size); |
| 128 | dst[size - 1] = '\0'; |
| 129 | } |
| 130 | } |
| 131 | |
| 132 | DBG2("IP address resolved to %s", dst); |
| 133 | |
| 134 | return 0; |
| 135 | |
| 136 | error: |
| 137 | return -1; |
| 138 | } |
| 139 | |
| 140 | /* |
| 141 | * Compare two URIs. |
| 142 | * |
| 143 | * Return 0 if equal else 1. |
| 144 | */ |
| 145 | int uri_compare(struct lttng_uri *uri1, struct lttng_uri *uri2) |
| 146 | { |
| 147 | return memcmp(uri1, uri2, sizeof(struct lttng_uri)); |
| 148 | } |
| 149 | |
| 150 | /* |
| 151 | * Free URI memory. |
| 152 | */ |
| 153 | void uri_free(struct lttng_uri *uri) |
| 154 | { |
| 155 | /* Safety check */ |
| 156 | if (uri != NULL) { |
| 157 | free(uri); |
| 158 | } |
| 159 | } |
| 160 | |
| 161 | /* |
| 162 | * Return an allocated URI. |
| 163 | */ |
| 164 | struct lttng_uri *uri_create(void) |
| 165 | { |
| 166 | struct lttng_uri *uri; |
| 167 | |
| 168 | uri = zmalloc(sizeof(struct lttng_uri)); |
| 169 | if (uri == NULL) { |
| 170 | PERROR("zmalloc uri"); |
| 171 | } |
| 172 | |
| 173 | return uri; |
| 174 | } |
| 175 | |
| 176 | /* |
| 177 | * Parses a string URI to a lttng_uri. This function can potentially return |
| 178 | * more than one URI in uris so the size of the array is returned and uris is |
| 179 | * allocated and populated. Caller must free(3) the array. |
| 180 | * |
| 181 | * This function can not detect the stream type of the URI so the caller has to |
| 182 | * make sure the correct type (stype) is set on the return URI(s). The default |
| 183 | * port must also be set by the caller if the returned URI has its port set to |
| 184 | * zero. |
| 185 | * |
| 186 | * NOTE: A good part of the following code was inspired from the "wget" source |
| 187 | * tree from the src/url.c file and url_parse() function. Also, the |
| 188 | * strpbrk_or_eos() function found above is also inspired by the same code. |
| 189 | * This code was originally licensed GPLv2 so we acknolwedge the Free Software |
| 190 | * Foundation here for the work and to make sure we are compliant with it. |
| 191 | */ |
| 192 | ssize_t uri_parse(const char *str_uri, struct lttng_uri **uris) |
| 193 | { |
| 194 | int ret, i = 0; |
| 195 | /* Size of the uris array. Default is 1 */ |
| 196 | ssize_t size = 1; |
| 197 | char subdir[PATH_MAX]; |
| 198 | unsigned int ctrl_port = 0; |
| 199 | unsigned int data_port = 0; |
| 200 | struct lttng_uri *tmp_uris; |
| 201 | char *addr_f = NULL; |
| 202 | const struct uri_proto *proto; |
| 203 | const char *purl, *addr_e, *addr_b, *subdir_b = NULL; |
| 204 | const char *seps = ":/\0"; |
| 205 | |
| 206 | /* |
| 207 | * The first part is the protocol portion of a maximum of 5 bytes for now. |
| 208 | * The second part is the hostname or IP address. The 255 bytes size is the |
| 209 | * limit found in the RFC 1035 for the total length of a domain name |
| 210 | * (https://www.ietf.org/rfc/rfc1035.txt). Finally, for the net:// |
| 211 | * protocol, two ports CAN be specified. |
| 212 | */ |
| 213 | |
| 214 | DBG3("URI string: %s", str_uri); |
| 215 | |
| 216 | proto = get_uri_proto(str_uri); |
| 217 | if (proto == NULL) { |
| 218 | ERR("URI parse unknown protocol %s", str_uri); |
| 219 | goto error; |
| 220 | } |
| 221 | |
| 222 | purl = str_uri; |
| 223 | |
| 224 | if (proto->code == P_NET || proto->code == P_NET6) { |
| 225 | /* Special case for net:// which requires two URI objects */ |
| 226 | size = 2; |
| 227 | } |
| 228 | |
| 229 | /* Allocate URI array */ |
| 230 | tmp_uris = zmalloc(sizeof(struct lttng_uri) * size); |
| 231 | if (tmp_uris == NULL) { |
| 232 | PERROR("zmalloc uri"); |
| 233 | goto error; |
| 234 | } |
| 235 | |
| 236 | memset(subdir, 0, sizeof(subdir)); |
| 237 | purl += strlen(proto->leading_string); |
| 238 | |
| 239 | /* Copy known value to the first URI. */ |
| 240 | tmp_uris[0].dtype = proto->dtype; |
| 241 | tmp_uris[0].proto = proto->type; |
| 242 | |
| 243 | if (proto->code == P_FILE) { |
| 244 | if (*purl != '/') { |
| 245 | ERR("Missing destination full path."); |
| 246 | goto free_error; |
| 247 | } |
| 248 | |
| 249 | strncpy(tmp_uris[0].dst.path, purl, sizeof(tmp_uris[0].dst.path)); |
| 250 | tmp_uris[0].dst.path[sizeof(tmp_uris[0].dst.path) - 1] = '\0'; |
| 251 | DBG3("URI file destination: %s", purl); |
| 252 | goto end; |
| 253 | } |
| 254 | |
| 255 | /* Assume we are at the beginning of an address or host of some sort. */ |
| 256 | addr_b = purl; |
| 257 | |
| 258 | /* |
| 259 | * Handle IPv6 address inside square brackets as mention by RFC 2732. IPv6 |
| 260 | * address that does not start AND end with brackets will be rejected even |
| 261 | * if valid. |
| 262 | * |
| 263 | * proto://[<addr>]... |
| 264 | * ^ |
| 265 | */ |
| 266 | if (*purl == '[') { |
| 267 | /* Address begins after '[' */ |
| 268 | addr_b = purl + 1; |
| 269 | addr_e = strchr(addr_b, ']'); |
| 270 | if (addr_e == NULL || addr_b == addr_e) { |
| 271 | ERR("Broken IPv6 address %s", addr_b); |
| 272 | goto free_error; |
| 273 | } |
| 274 | |
| 275 | /* Moving parsed URL pointer after the final bracket ']' */ |
| 276 | purl = addr_e + 1; |
| 277 | |
| 278 | /* |
| 279 | * The closing bracket must be followed by a seperator or NULL char. |
| 280 | */ |
| 281 | if (strchr(seps, *purl) == NULL) { |
| 282 | ERR("Unknown symbol after IPv6 address: %s", purl); |
| 283 | goto free_error; |
| 284 | } |
| 285 | } else { |
| 286 | purl = strpbrk_or_eos(purl, seps); |
| 287 | addr_e = purl; |
| 288 | } |
| 289 | |
| 290 | /* Check if we at least have a char for the addr or hostname. */ |
| 291 | if (addr_b == addr_e) { |
| 292 | ERR("No address or hostname detected."); |
| 293 | goto free_error; |
| 294 | } |
| 295 | |
| 296 | addr_f = utils_strdupdelim(addr_b, addr_e); |
| 297 | if (addr_f == NULL) { |
| 298 | goto free_error; |
| 299 | } |
| 300 | |
| 301 | /* |
| 302 | * Detect PORT after address. The net/net6 protocol allows up to two port |
| 303 | * so we can define the control and data port. |
| 304 | */ |
| 305 | while (*purl == ':') { |
| 306 | int port; |
| 307 | const char *port_b, *port_e; |
| 308 | char *port_f; |
| 309 | |
| 310 | /* Update pass counter */ |
| 311 | i++; |
| 312 | |
| 313 | /* |
| 314 | * Maximum of two ports is possible if P_NET/NET6. Bigger than that, |
| 315 | * two much stuff. |
| 316 | */ |
| 317 | if ((i == 2 && (proto->code != P_NET && proto->code != P_NET6)) |
| 318 | || i > 2) { |
| 319 | break; |
| 320 | } |
| 321 | |
| 322 | /* |
| 323 | * Move parsed URL to port value. |
| 324 | * proto://addr_host:PORT1:PORT2/foo/bar |
| 325 | * ^ |
| 326 | */ |
| 327 | ++purl; |
| 328 | port_b = purl; |
| 329 | purl = strpbrk_or_eos(purl, seps); |
| 330 | port_e = purl; |
| 331 | |
| 332 | if (port_b != port_e) { |
| 333 | port_f = utils_strdupdelim(port_b, port_e); |
| 334 | if (port_f == NULL) { |
| 335 | goto free_error; |
| 336 | } |
| 337 | |
| 338 | port = atoi(port_f); |
| 339 | if (port > 0xffff || port <= 0x0) { |
| 340 | ERR("Invalid port number %d", port); |
| 341 | free(port_f); |
| 342 | goto free_error; |
| 343 | } |
| 344 | free(port_f); |
| 345 | |
| 346 | if (i == 1) { |
| 347 | ctrl_port = port; |
| 348 | } else { |
| 349 | data_port = port; |
| 350 | } |
| 351 | } |
| 352 | }; |
| 353 | |
| 354 | /* Check for a valid subdir or trailing garbage */ |
| 355 | if (*purl == '/') { |
| 356 | /* |
| 357 | * Move to subdir value. |
| 358 | * proto://addr_host:PORT1:PORT2/foo/bar |
| 359 | * ^ |
| 360 | */ |
| 361 | ++purl; |
| 362 | subdir_b = purl; |
| 363 | } else if (*purl != '\0') { |
| 364 | ERR("Trailing characters not recognized: %s", purl); |
| 365 | goto free_error; |
| 366 | } |
| 367 | |
| 368 | /* We have enough valid information to create URI(s) object */ |
| 369 | |
| 370 | /* Copy generic information */ |
| 371 | tmp_uris[0].port = ctrl_port; |
| 372 | |
| 373 | /* Copy subdirectory if one. */ |
| 374 | if (subdir_b) { |
| 375 | strncpy(tmp_uris[0].subdir, subdir_b, sizeof(tmp_uris[0].subdir)); |
| 376 | tmp_uris[0].subdir[sizeof(tmp_uris[0].subdir) - 1] = '\0'; |
| 377 | } |
| 378 | |
| 379 | switch (proto->code) { |
| 380 | case P_NET: |
| 381 | ret = set_ip_address(addr_f, AF_INET, tmp_uris[0].dst.ipv4, |
| 382 | sizeof(tmp_uris[0].dst.ipv4)); |
| 383 | if (ret < 0) { |
| 384 | goto free_error; |
| 385 | } |
| 386 | |
| 387 | memcpy(tmp_uris[1].dst.ipv4, tmp_uris[0].dst.ipv4, sizeof(tmp_uris[1].dst.ipv4)); |
| 388 | |
| 389 | tmp_uris[1].dtype = proto->dtype; |
| 390 | tmp_uris[1].proto = proto->type; |
| 391 | tmp_uris[1].port = data_port; |
| 392 | break; |
| 393 | case P_NET6: |
| 394 | ret = set_ip_address(addr_f, AF_INET6, tmp_uris[0].dst.ipv6, |
| 395 | sizeof(tmp_uris[0].dst.ipv6)); |
| 396 | if (ret < 0) { |
| 397 | goto free_error; |
| 398 | } |
| 399 | |
| 400 | memcpy(tmp_uris[1].dst.ipv6, tmp_uris[0].dst.ipv6, sizeof(tmp_uris[1].dst.ipv6)); |
| 401 | |
| 402 | tmp_uris[1].dtype = proto->dtype; |
| 403 | tmp_uris[1].proto = proto->type; |
| 404 | tmp_uris[1].port = data_port; |
| 405 | break; |
| 406 | case P_TCP: |
| 407 | ret = set_ip_address(addr_f, AF_INET, tmp_uris[0].dst.ipv4, |
| 408 | sizeof(tmp_uris[0].dst.ipv4)); |
| 409 | if (ret < 0) { |
| 410 | goto free_error; |
| 411 | } |
| 412 | break; |
| 413 | case P_TCP6: |
| 414 | ret = set_ip_address(addr_f, AF_INET6, tmp_uris[0].dst.ipv6, |
| 415 | sizeof(tmp_uris[0].dst.ipv6)); |
| 416 | if (ret < 0) { |
| 417 | goto free_error; |
| 418 | } |
| 419 | break; |
| 420 | default: |
| 421 | goto free_error; |
| 422 | } |
| 423 | |
| 424 | end: |
| 425 | DBG3("URI dtype: %d, proto: %d, host: %s, subdir: %s, ctrl: %d, data: %d", |
| 426 | proto->dtype, proto->type, (addr_f == NULL) ? "" : addr_f, |
| 427 | (subdir_b == NULL) ? "" : subdir_b, ctrl_port, data_port); |
| 428 | |
| 429 | free(addr_f); |
| 430 | |
| 431 | *uris = tmp_uris; |
| 432 | return size; |
| 433 | |
| 434 | free_error: |
| 435 | free(addr_f); |
| 436 | free(tmp_uris); |
| 437 | error: |
| 438 | return -1; |
| 439 | } |