1 /* 2 * request - HTTP request processing 3 * 4 * Fully-rewritten from original EXCEPT for request_check_hostname() 5 * Copyright(c) 2018 Glenn Strauss gstrauss()gluelogic.com All rights reserved 6 * License: BSD 3-clause (same as lighttpd) 7 */ 8 #include "first.h" 9 10 #include "request.h" 11 #include "burl.h" 12 #include "http_header.h" 13 #include "http_kv.h" 14 #include "log.h" 15 #include "sock_addr.h" 16 17 #include <limits.h> 18 #include <stdint.h> 19 #include <stdlib.h> 20 #include <string.h> 21 22 static int request_check_hostname(buffer * const host) { 23 enum { DOMAINLABEL, TOPLABEL } stage = TOPLABEL; 24 size_t i; 25 int label_len = 0; 26 size_t host_len, hostport_len; 27 char *colon; 28 int is_ip = -1; /* -1 don't know yet, 0 no, 1 yes */ 29 int level = 0; 30 31 /* 32 * hostport = host [ ":" port ] 33 * host = hostname | IPv4address | IPv6address 34 * hostname = *( domainlabel "." ) toplabel [ "." ] 35 * domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum 36 * toplabel = alpha | alpha *( alphanum | "-" ) alphanum 37 * IPv4address = 1*digit "." 1*digit "." 1*digit "." 1*digit 38 * IPv6address = "[" ... "]" 39 * port = *digit 40 */ 41 42 /* IPv6 address */ 43 if (host->ptr[0] == '[') { 44 char *c = host->ptr + 1; 45 int colon_cnt = 0; 46 47 /* check the address inside [...] */ 48 for (; *c && *c != ']'; c++) { 49 if (*c == ':') { 50 if (++colon_cnt > 7) { 51 return -1; 52 } 53 } else if (!light_isxdigit(*c) && '.' != *c) { 54 return -1; 55 } 56 } 57 58 /* missing ] */ 59 if (!*c) { 60 return -1; 61 } 62 63 /* check port */ 64 if (*(c+1) == ':') { 65 for (c += 2; *c; c++) { 66 if (!light_isdigit(*c)) { 67 return -1; 68 } 69 } 70 } 71 else if ('\0' != *(c+1)) { 72 /* only a port is allowed to follow [...] */ 73 return -1; 74 } 75 return 0; 76 } 77 78 hostport_len = host_len = buffer_string_length(host); 79 80 if (NULL != (colon = memchr(host->ptr, ':', host_len))) { 81 char *c = colon + 1; 82 83 /* check portnumber */ 84 for (; *c; c++) { 85 if (!light_isdigit(*c)) return -1; 86 } 87 88 /* remove the port from the host-len */ 89 host_len = colon - host->ptr; 90 } 91 92 /* Host is empty */ 93 if (host_len == 0) return -1; 94 95 /* if the hostname ends in a "." strip it */ 96 if (host->ptr[host_len-1] == '.') { 97 /* shift port info one left */ 98 if (NULL != colon) memmove(colon-1, colon, hostport_len - host_len); 99 buffer_string_set_length(host, --hostport_len); 100 if (--host_len == 0) return -1; 101 } 102 103 104 /* scan from the right and skip the \0 */ 105 for (i = host_len; i-- > 0; ) { 106 const char c = host->ptr[i]; 107 108 switch (stage) { 109 case TOPLABEL: 110 if (c == '.') { 111 /* only switch stage, if this is not the last character */ 112 if (i != host_len - 1) { 113 if (label_len == 0) { 114 return -1; 115 } 116 117 /* check the first character at right of the dot */ 118 if (is_ip == 0) { 119 if (!light_isalnum(host->ptr[i+1])) { 120 return -1; 121 } 122 } else if (!light_isdigit(host->ptr[i+1])) { 123 is_ip = 0; 124 } else if ('-' == host->ptr[i+1]) { 125 return -1; 126 } else { 127 /* just digits */ 128 is_ip = 1; 129 } 130 131 stage = DOMAINLABEL; 132 133 label_len = 0; 134 level++; 135 } else if (i == 0) { 136 /* just a dot and nothing else is evil */ 137 return -1; 138 } 139 } else if (i == 0) { 140 /* the first character of the hostname */ 141 if (!light_isalnum(c)) { 142 return -1; 143 } 144 label_len++; 145 } else { 146 if (c != '-' && !light_isalnum(c)) { 147 return -1; 148 } 149 if (is_ip == -1) { 150 if (!light_isdigit(c)) is_ip = 0; 151 } 152 label_len++; 153 } 154 155 break; 156 case DOMAINLABEL: 157 if (is_ip == 1) { 158 if (c == '.') { 159 if (label_len == 0) { 160 return -1; 161 } 162 163 label_len = 0; 164 level++; 165 } else if (!light_isdigit(c)) { 166 return -1; 167 } else { 168 label_len++; 169 } 170 } else { 171 if (c == '.') { 172 if (label_len == 0) { 173 return -1; 174 } 175 176 /* c is either - or alphanum here */ 177 if ('-' == host->ptr[i+1]) { 178 return -1; 179 } 180 181 label_len = 0; 182 level++; 183 } else if (i == 0) { 184 if (!light_isalnum(c)) { 185 return -1; 186 } 187 label_len++; 188 } else { 189 if (c != '-' && !light_isalnum(c)) { 190 return -1; 191 } 192 label_len++; 193 } 194 } 195 196 break; 197 } 198 } 199 200 /* a IP has to consist of 4 parts */ 201 if (is_ip == 1 && level != 3) { 202 return -1; 203 } 204 205 if (label_len == 0) { 206 return -1; 207 } 208 209 return 0; 210 } 211 212 int http_request_host_normalize(buffer * const b, const int scheme_port) { 213 /* 214 * check for and canonicalize numeric IP address and portnum (optional) 215 * (IP address may be followed by ":portnum" (optional)) 216 * - IPv6: "[...]" 217 * - IPv4: "x.x.x.x" 218 * - IPv4: 12345678 (32-bit decimal number) 219 * - IPv4: 012345678 (32-bit octal number) 220 * - IPv4: 0x12345678 (32-bit hex number) 221 * 222 * allow any chars (except ':' and '\0' and stray '[' or ']') 223 * (other code may check chars more strictly or more pedantically) 224 * ':' delimits (optional) port at end of string 225 * "[]" wraps IPv6 address literal 226 * '\0' should have been rejected earlier were it present 227 * 228 * any chars includes, but is not limited to: 229 * - allow '-' any where, even at beginning of word 230 * (security caution: might be confused for cmd flag if passed to shell) 231 * - allow all-digit TLDs 232 * (might be mistaken for IPv4 addr by inet_aton() 233 * unless non-digits appear in subdomain) 234 */ 235 236 /* Note: not using getaddrinfo() since it does not support "[]" around IPv6 237 * and is not as lenient as inet_aton() and inet_addr() for IPv4 strings. 238 * Not using inet_pton() (when available) on IPv4 for similar reasons. */ 239 240 const char * const p = b->ptr; 241 const size_t blen = buffer_string_length(b); 242 long port = 0; 243 244 if (*p != '[') { 245 char * const colon = (char *)memchr(p, ':', blen); 246 if (colon) { 247 if (*p == ':') return -1; /*(empty host then port, or naked IPv6)*/ 248 if (colon[1] != '\0') { 249 char *e; 250 port = strtol(colon+1, &e, 0); /*(allow decimal, octal, hex)*/ 251 if (0 < port && port <= USHRT_MAX && *e == '\0') { 252 /* valid port */ 253 } else { 254 return -1; 255 } 256 } /*(else ignore stray colon at string end)*/ 257 buffer_string_set_length(b, (size_t)(colon - p)); /*(remove port str)*/ 258 } 259 260 if (light_isdigit(*p)) do { 261 /* (IPv4 address literal or domain starting w/ digit (e.g. 3com))*/ 262 /* (check one-element cache of normalized IPv4 address string) */ 263 static struct { char s[INET_ADDRSTRLEN]; size_t n; } laddr; 264 size_t n = colon ? (size_t)(colon - p) : blen; 265 sock_addr addr; 266 if (n == laddr.n && 0 == memcmp(p, laddr.s, n)) break; 267 if (1 == sock_addr_inet_pton(&addr, p, AF_INET, 0)) { 268 sock_addr_inet_ntop_copy_buffer(b, &addr); 269 n = buffer_string_length(b); 270 if (n < sizeof(laddr.s)) memcpy(laddr.s, b->ptr, (laddr.n = n)); 271 } 272 } while (0); 273 } else do { /* IPv6 addr */ 274 #if defined(HAVE_IPV6) && defined(HAVE_INET_PTON) 275 276 /* (check one-element cache of normalized IPv4 address string) */ 277 static struct { char s[INET6_ADDRSTRLEN]; size_t n; } laddr; 278 sock_addr addr; 279 char *bracket = b->ptr+blen-1; 280 char *percent = strchr(b->ptr+1, '%'); 281 size_t len; 282 int rc; 283 char buf[INET6_ADDRSTRLEN+16]; /*(+16 for potential %interface name)*/ 284 if (blen <= 2) return -1; /*(invalid "[]")*/ 285 if (*bracket != ']') { 286 bracket = (char *)memchr(b->ptr+1, ']', blen-1); 287 if (NULL == bracket || bracket[1] != ':' || bracket - b->ptr == 1){ 288 return -1; 289 } 290 if (bracket[2] != '\0') { /*(ignore stray colon at string end)*/ 291 char *e; 292 port = strtol(bracket+2, &e, 0); /*(allow decimal, octal, hex)*/ 293 if (0 < port && port <= USHRT_MAX && *e == '\0') { 294 /* valid port */ 295 } else { 296 return -1; 297 } 298 } 299 } 300 301 len = (size_t)((percent ? percent : bracket) - (b->ptr+1)); 302 if (laddr.n == len && 0 == memcmp(laddr.s, b->ptr+1, len)) { 303 /* truncate after ']' and re-add normalized port, if needed */ 304 buffer_string_set_length(b, (size_t)(bracket - b->ptr + 1)); 305 break; 306 } 307 308 *bracket = '\0';/*(terminate IPv6 string)*/ 309 if (percent) *percent = '\0'; /*(remove %interface from address)*/ 310 rc = sock_addr_inet_pton(&addr, b->ptr+1, AF_INET6, 0); 311 if (percent) *percent = '%'; /*(restore %interface)*/ 312 *bracket = ']'; /*(restore bracket)*/ 313 if (1 != rc) return -1; 314 315 sock_addr_inet_ntop(&addr, buf, sizeof(buf)); 316 len = strlen(buf); 317 if (percent) { 318 if (percent > bracket) return -1; 319 if (len + (size_t)(bracket - percent) >= sizeof(buf)) return -1; 320 if (len < sizeof(laddr.s)) memcpy(laddr.s, buf, (laddr.n = len)); 321 memcpy(buf+len, percent, (size_t)(bracket - percent)); 322 len += (size_t)(bracket - percent); 323 } 324 buffer_string_set_length(b, 1); /* truncate after '[' */ 325 buffer_append_str2(b, buf, len, CONST_STR_LEN("]")); 326 327 #else 328 329 return -1; 330 331 #endif 332 } while (0); 333 334 if (0 != port && port != scheme_port) { 335 buffer_append_string_len(b, CONST_STR_LEN(":")); 336 buffer_append_int(b, (int)port); 337 } 338 339 return 0; 340 } 341 342 int http_request_host_policy (buffer * const b, const unsigned int http_parseopts, const int scheme_port) { 343 return (((http_parseopts & HTTP_PARSEOPT_HOST_STRICT) 344 && 0 != request_check_hostname(b)) 345 || ((http_parseopts & HTTP_PARSEOPT_HOST_NORMALIZE) 346 && 0 != http_request_host_normalize(b, scheme_port))); 347 } 348 349 __attribute_const__ 350 static int request_uri_is_valid_char(const unsigned char c) { 351 return (c > 32 && c != 127 && c != 255); 352 } 353 354 __attribute_cold__ 355 __attribute_noinline__ 356 static int http_request_header_line_invalid(request_st * const restrict r, const int status, const char * const restrict msg) { 357 if (r->conf.log_request_header_on_error) { 358 if (msg) log_error(r->conf.errh, __FILE__, __LINE__, "%s", msg); 359 } 360 return status; 361 } 362 363 __attribute_cold__ 364 __attribute_noinline__ 365 static int http_request_header_char_invalid(request_st * const restrict r, const char ch, const char * const restrict msg) { 366 if (r->conf.log_request_header_on_error) { 367 if ((unsigned char)ch > 32 && ch != 127) { 368 log_error(r->conf.errh, __FILE__, __LINE__, "%s ('%c')", msg, ch); 369 } 370 else { 371 log_error(r->conf.errh, __FILE__, __LINE__, "%s (0x%x)", msg, ch); 372 } 373 } 374 return 400; 375 } 376 377 378 int64_t 379 li_restricted_strtoint64 (const char *v, const uint32_t vlen, const char ** const err) 380 { 381 /* base 10 strtoll() parsing exactly vlen chars and requiring digits 0-9 */ 382 /* rejects negative numbers and considers values > INT64_MAX an error */ 383 /* note: errno is not set; detect error if *err != v+vlen upon return */ 384 /*(caller must check 0 == vlen if that is to be an error for caller)*/ 385 int64_t rv = 0; 386 uint32_t i; 387 for (i = 0; i < vlen; ++i) { 388 const uint8_t c = ((uint8_t *)v)[i] - '0'; /*(unsigned; underflow ok)*/ 389 if (c > 9) break; 390 if (rv > INT64_MAX/10) break; 391 rv *= 10; 392 if (rv > INT64_MAX - c) break; 393 rv += c; 394 } 395 *err = v+i; 396 return rv; 397 } 398 399 400 /* add header to list of headers 401 * certain headers are also parsed 402 * might drop a header if deemed unnecessary/broken 403 * 404 * returns 0 on success, HTTP status on error 405 */ 406 static int http_request_parse_single_header(request_st * const restrict r, const enum http_header_e id, const char * const restrict k, const size_t klen, const char * const restrict v, const size_t vlen) { 407 buffer **saveb = NULL; 408 409 /* 410 * Note: k might not be '\0'-terminated 411 * Note: v is not '\0'-terminated 412 * With lighttpd HTTP/1.1 parser, v ends with whitespace 413 * (one of '\r' '\n' ' ' '\t') 414 * With lighttpd HTTP/2 parser, v should not be accessed beyond vlen 415 * (care must be taken to avoid libc funcs which expect z-strings) 416 */ 417 /*assert(vlen);*//*(caller must not call this func with 0 klen or 0 vlen)*/ 418 419 switch (id) { 420 /*case HTTP_HEADER_OTHER:*/ 421 default: 422 break; 423 case HTTP_HEADER_HOST: 424 if (!light_btst(r->rqst_htags, HTTP_HEADER_HOST)) { 425 saveb = &r->http_host; 426 if (vlen >= 1024) { /*(expecting < 256)*/ 427 return http_request_header_line_invalid(r, 400, "uri-authority too long -> 400"); 428 } 429 } 430 else if (NULL != r->http_host 431 && buffer_is_equal_string(r->http_host, v, vlen)) { 432 /* ignore all Host: headers if match authority in request line */ 433 return 0; /* ignore header */ 434 } 435 else { 436 return http_request_header_line_invalid(r, 400, "duplicate Host header -> 400"); 437 } 438 break; 439 case HTTP_HEADER_CONNECTION: 440 /* "Connection: close" is common case if header is present */ 441 if ((vlen == 5 && buffer_eq_icase_ssn(v, CONST_STR_LEN("close"))) 442 || http_header_str_contains_token(v,vlen,CONST_STR_LEN("close"))) { 443 r->keep_alive = 0; 444 break; 445 } 446 if (http_header_str_contains_token(v,vlen,CONST_STR_LEN("keep-alive"))){ 447 r->keep_alive = 1; 448 break; 449 } 450 break; 451 case HTTP_HEADER_CONTENT_TYPE: 452 if (light_btst(r->rqst_htags, HTTP_HEADER_CONTENT_TYPE)) { 453 return http_request_header_line_invalid(r, 400, "duplicate Content-Type header -> 400"); 454 } 455 break; 456 case HTTP_HEADER_IF_NONE_MATCH: 457 /* if dup, only the first one will survive */ 458 if (light_btst(r->rqst_htags, HTTP_HEADER_IF_NONE_MATCH)) { 459 return 0; /* ignore header */ 460 } 461 break; 462 case HTTP_HEADER_CONTENT_LENGTH: 463 if (!light_btst(r->rqst_htags, HTTP_HEADER_CONTENT_LENGTH)) { 464 /*(trailing whitespace was removed from vlen)*/ 465 /*(not using strtoll() since v might not be z-string)*/ 466 const char *err; 467 off_t clen = (off_t)li_restricted_strtoint64(v, vlen, &err); 468 if (err == v+vlen) { 469 /* (set only if not set to -1 by Transfer-Encoding: chunked) */ 470 if (0 == r->reqbody_length) r->reqbody_length = clen; 471 } 472 else { 473 return http_request_header_line_invalid(r, 400, "invalid Content-Length header -> 400"); 474 } 475 } 476 else { 477 return http_request_header_line_invalid(r, 400, "duplicate Content-Length header -> 400"); 478 } 479 break; 480 case HTTP_HEADER_HTTP2_SETTINGS: 481 if (light_btst(r->rqst_htags, HTTP_HEADER_HTTP2_SETTINGS)) { 482 return http_request_header_line_invalid(r, 400, "duplicate HTTP2-Settings header -> 400"); 483 } 484 break; 485 case HTTP_HEADER_IF_MODIFIED_SINCE: 486 if (light_btst(r->rqst_htags, HTTP_HEADER_IF_MODIFIED_SINCE)) { 487 /* Proxies sometimes send dup headers 488 * if they are the same we ignore the second 489 * if not, we raise an error */ 490 const buffer *vb = 491 http_header_request_get(r, HTTP_HEADER_IF_MODIFIED_SINCE, 492 CONST_STR_LEN("If-Modified-Since")); 493 if (vb && buffer_eq_icase_slen(vb, v, vlen)) { 494 /* ignore it if they are the same */ 495 return 0; /* ignore header */ 496 } 497 else { 498 return http_request_header_line_invalid(r, 400, "duplicate If-Modified-Since header -> 400"); 499 } 500 } 501 break; 502 case HTTP_HEADER_TRANSFER_ENCODING: 503 if (HTTP_VERSION_1_1 != r->http_version) { 504 return http_request_header_line_invalid(r, 400, 505 HTTP_VERSION_1_0 == r->http_version 506 ? "HTTP/1.0 with Transfer-Encoding (bad HTTP/1.0 proxy?) -> 400" 507 : "HTTP/2 with Transfer-Encoding is invalid -> 400"); 508 } 509 510 if (!buffer_eq_icase_ss(v, vlen, CONST_STR_LEN("chunked"))) { 511 /* Transfer-Encoding might contain additional encodings, 512 * which are not currently supported by lighttpd */ 513 return http_request_header_line_invalid(r, 501, NULL); /* Not Implemented */ 514 } 515 r->reqbody_length = -1; 516 517 /* Transfer-Encoding is a hop-by-hop header, 518 * which must not be blindly forwarded to backends */ 519 return 0; /* skip header */ 520 } 521 522 http_header_request_append(r, id, k, klen, v, vlen); 523 524 if (saveb) { 525 *saveb = http_header_request_get(r, id, k, klen); 526 } 527 528 return 0; 529 } 530 531 __attribute_cold__ 532 static int http_request_parse_proto_loose(request_st * const restrict r, const char * const restrict ptr, const size_t len, const unsigned int http_parseopts) { 533 const char * proto = memchr(ptr, ' ', len); 534 if (NULL == proto) 535 return http_request_header_line_invalid(r, 400, "incomplete request line -> 400"); 536 proto = memchr(proto+1, ' ', len - (proto+1 - ptr)); 537 if (NULL == proto) 538 return http_request_header_line_invalid(r, 400, "incomplete request line -> 400"); 539 ++proto; 540 541 if (proto[0]=='H' && proto[1]=='T' && proto[2]=='T' && proto[3]=='P' && proto[4] == '/') { 542 if (proto[5] == '1' && proto[6] == '.' && (proto[7] == '1' || proto[7] == '0')) { 543 /* length already checked before calling this routine */ 544 /* (len != (size_t)(proto - ptr + 8)) */ 545 if (http_parseopts & HTTP_PARSEOPT_HEADER_STRICT) /*(http_header_strict)*/ 546 return http_request_header_line_invalid(r, 400, "incomplete request line -> 400"); 547 r->http_version = (proto[7] == '1') ? HTTP_VERSION_1_1 : HTTP_VERSION_1_0; 548 } 549 else 550 return http_request_header_line_invalid(r, 505, "unknown HTTP version -> 505"); 551 } 552 else 553 return http_request_header_line_invalid(r, 400, "unknown protocol -> 400"); 554 555 /* keep-alive default: HTTP/1.1 -> true; HTTP/1.0 -> false */ 556 r->keep_alive = (HTTP_VERSION_1_0 != r->http_version); 557 558 return 0; 559 } 560 561 __attribute_cold__ 562 static const char * http_request_parse_reqline_uri(request_st * const restrict r, const char * const restrict uri, const size_t len, const unsigned int http_parseopts) { 563 const char *nuri; 564 if ((len > 7 && buffer_eq_icase_ssn(uri, "http://", 7) 565 && NULL != (nuri = memchr(uri + 7, '/', len-7))) 566 || 567 (len > 8 && buffer_eq_icase_ssn(uri, "https://", 8) 568 && NULL != (nuri = memchr(uri + 8, '/', len-8)))) { 569 const char * const host = uri + (uri[4] == ':' ? 7 : 8); 570 const size_t hostlen = nuri - host; 571 if (0 == hostlen || hostlen >= 1024) { /*(expecting < 256)*/ 572 http_request_header_line_invalid(r, 400, "uri-authority empty or too long -> 400"); 573 return NULL; 574 } 575 /* Insert as host header */ 576 http_header_request_set(r, HTTP_HEADER_HOST, CONST_STR_LEN("Host"), host, hostlen); 577 r->http_host = http_header_request_get(r, HTTP_HEADER_HOST, CONST_STR_LEN("Host")); 578 return nuri; 579 } else if (!(http_parseopts & HTTP_PARSEOPT_HEADER_STRICT) /*(!http_header_strict)*/ 580 || (HTTP_METHOD_CONNECT == r->http_method && (uri[0] == ':' || light_isdigit(uri[0]))) 581 || (HTTP_METHOD_OPTIONS == r->http_method && uri[0] == '*' && 1 == len)) { 582 /* (permitted) */ 583 return uri; 584 } else { 585 http_request_header_line_invalid(r, 400, "request-URI parse error -> 400"); 586 return NULL; 587 } 588 } 589 590 591 __attribute_cold__ 592 __attribute_noinline__ 593 static int http_request_parse_header_other(request_st * const restrict r, const char * const restrict k, const int klen, const unsigned int http_header_strict); 594 595 596 int 597 http_request_validate_pseudohdrs (request_st * const restrict r, const int scheme, const unsigned int http_parseopts) 598 { 599 /* :method is required to indicate method 600 * CONNECT method must have :method and :authority 601 * All other methods must have at least :method :scheme :path */ 602 603 if (HTTP_METHOD_UNSET == r->http_method) 604 return http_request_header_line_invalid(r, 400, 605 "missing pseudo-header method -> 400"); 606 607 if (HTTP_METHOD_CONNECT != r->http_method) { 608 if (!scheme) 609 return http_request_header_line_invalid(r, 400, 610 "missing pseudo-header scheme -> 400"); 611 612 if (buffer_string_is_empty(&r->target)) 613 return http_request_header_line_invalid(r, 400, 614 "missing pseudo-header path -> 400"); 615 616 const char * const uri = r->target.ptr; 617 if (*uri != '/') { /* (common case: (*uri == '/')) */ 618 if (uri[0] != '*' || uri[1] != '\0' 619 || HTTP_METHOD_OPTIONS != r->http_method) 620 return http_request_header_line_invalid(r, 400, 621 "invalid pseudo-header path -> 400"); 622 } 623 } 624 else { /* HTTP_METHOD_CONNECT */ 625 if (NULL == r->http_host) 626 return http_request_header_line_invalid(r, 400, 627 "missing pseudo-header authority -> 400"); 628 if (!buffer_string_is_empty(&r->target) || scheme) 629 return http_request_header_line_invalid(r, 400, 630 "invalid pseudo-header with CONNECT -> 400"); 631 /*(reuse uri and ulen to assign to r->target)*/ 632 buffer_copy_buffer(&r->target, r->http_host); 633 } 634 buffer_copy_buffer(&r->target_orig, &r->target); 635 636 /* r->http_host, if set, is checked with http_request_host_policy() 637 * in http_request_parse() */ 638 639 /* copied and modified from end of http_request_parse_reqline() */ 640 641 /* check uri for invalid characters */ 642 const unsigned int http_header_strict = 643 (http_parseopts & HTTP_PARSEOPT_HEADER_STRICT); 644 if (http_header_strict 645 && (http_parseopts & HTTP_PARSEOPT_URL_NORMALIZE_CTRLS_REJECT)) 646 return 0; /* URI will be checked in http_request_parse_target() */ 647 648 const uint32_t ulen = buffer_string_length(&r->target); 649 const uint8_t * const uri = (uint8_t *)r->target.ptr; 650 if (http_header_strict) { 651 for (uint32_t i = 0; i < ulen; ++i) { 652 if (!request_uri_is_valid_char(uri[i])) 653 return http_request_header_char_invalid(r, uri[i], 654 "invalid character in URI -> 400"); 655 } 656 } 657 else { 658 if (NULL != memchr(uri, '\0', ulen)) 659 return http_request_header_char_invalid(r, '\0', 660 "invalid character in header -> 400"); 661 } 662 663 return 0; 664 } 665 666 667 int 668 http_request_parse_header (request_st * const restrict r, http_header_parse_ctx * const restrict hpctx) 669 { 670 /* Note: k and v might not be '\0' terminated strings; 671 * care must be taken to avoid libc funcs which expect z-strings */ 672 const char * const restrict k = hpctx->k; 673 const char * const restrict v = hpctx->v; 674 const uint32_t klen = hpctx->klen; 675 const uint32_t vlen = hpctx->vlen; 676 677 if (0 == klen) 678 return http_request_header_line_invalid(r, 400, 679 "invalid header key -> 400"); 680 681 if ((hpctx->hlen += klen + vlen + 4) > hpctx->max_request_field_size) { 682 /*(configurable with server.max-request-field-size; default 8k)*/ 683 #if 1 /* emit to error log for people sending large headers */ 684 log_error(r->conf.errh, __FILE__, __LINE__, 685 "oversized request header -> 431"); 686 return 431; /* Request Header Fields Too Large */ 687 #else 688 /* 431 Request Header Fields Too Large */ 689 return http_request_header_line_invalid(r, 431, 690 "oversized request header -> 431"); 691 #endif 692 } 693 694 if (2 == klen && k[0] == 't' && k[1] == 'e' 695 && !buffer_eq_icase_ss(v, vlen, CONST_STR_LEN("trailers"))) 696 return http_request_header_line_invalid(r, 400, 697 "invalid TE header value with HTTP/2 -> 400"); 698 699 if (!hpctx->trailers) { 700 if (*k == ':') { 701 /* HTTP/2 request pseudo-header fields */ 702 if (!hpctx->pseudo) /*(pseudo header after non-pseudo header)*/ 703 return http_request_header_line_invalid(r, 400, 704 "invalid pseudo-header -> 400"); 705 if (0 == vlen) 706 return http_request_header_line_invalid(r, 400, 707 "invalid header value -> 400"); 708 switch (klen-1) { 709 case 4: 710 if (0 == memcmp(k+1, "path", 4)) { 711 if (!buffer_string_is_empty(&r->target)) 712 return http_request_header_line_invalid(r, 400, 713 "repeated pseudo-header -> 400"); 714 buffer_copy_string_len(&r->target, v, vlen); 715 return 0; 716 } 717 break; 718 case 6: 719 if (0 == memcmp(k+1, "method", 6)) { 720 if (HTTP_METHOD_UNSET != r->http_method) 721 return http_request_header_line_invalid(r, 400, 722 "repeated pseudo-header -> 400"); 723 r->http_method = get_http_method_key(v, vlen); 724 if (HTTP_METHOD_UNSET >= r->http_method) 725 return http_request_header_line_invalid(r, 501, 726 "unknown http-method -> 501"); 727 return 0; 728 } 729 else if (0 == memcmp(k+1, "scheme", 6)) { 730 if (hpctx->scheme) 731 return http_request_header_line_invalid(r, 400, 732 "repeated pseudo-header -> 400"); 733 switch (vlen) {/*(validated, but then ignored)*/ 734 case 5: /* "https" */ 735 if (v[4]!='s') break; 736 __attribute_fallthrough__ 737 case 4: /* "http" */ 738 if (v[0]=='h' && v[1]=='t' && v[2]=='t' && v[3]=='p') { 739 hpctx->scheme = 1; 740 return 0; 741 } 742 break; 743 default: 744 break; 745 } 746 return http_request_header_line_invalid(r, 400, 747 "unknown pseudo-header scheme -> 400"); 748 } 749 break; 750 case 9: 751 if (0 == memcmp(k+1, "authority", 9)) { 752 if (r->http_host) 753 return http_request_header_line_invalid(r, 400, 754 "repeated pseudo-header -> 400"); 755 if (vlen >= 1024) /*(expecting < 256)*/ 756 return http_request_header_line_invalid(r, 400, 757 "invalid pseudo-header authority too long -> 400"); 758 /* insert as host header */ 759 http_header_request_set(r, HTTP_HEADER_HOST, 760 CONST_STR_LEN("host"), v, vlen); 761 r->http_host = 762 http_header_request_get(r, HTTP_HEADER_HOST, 763 CONST_STR_LEN("Host")); 764 return 0; 765 } 766 break; 767 default: 768 break; 769 } 770 return http_request_header_line_invalid(r, 400, 771 "invalid pseudo-header -> 400"); 772 } 773 else { /*(non-pseudo headers)*/ 774 if (hpctx->pseudo) { /*(transition to non-pseudo headers)*/ 775 hpctx->pseudo = 0; 776 int status = 777 http_request_validate_pseudohdrs(r, hpctx->scheme, 778 hpctx->http_parseopts); 779 if (0 != status) return status; 780 } 781 if (0 == vlen) 782 return 0; 783 784 const unsigned int http_header_strict = 785 (hpctx->http_parseopts & HTTP_PARSEOPT_HEADER_STRICT); 786 787 for (uint32_t j = 0; j < klen; ++j) { 788 if (light_islower(k[j]) || k[j] == '-') 789 continue; /*(common cases)*/ 790 if (light_isupper(k[j])) 791 return 400; 792 if (0 != http_request_parse_header_other(r, k+j, klen-j, 793 http_header_strict)) 794 return 400; 795 break; 796 } 797 798 if (http_header_strict) { 799 for (uint32_t j = 0; j < vlen; ++j) { 800 if ((((uint8_t *)v)[j] < 32 && v[j] != '\t') || v[j]==127) 801 return http_request_header_char_invalid(r, v[j], 802 "invalid character in header -> 400"); 803 } 804 } 805 else { 806 if (NULL != memchr(v, '\0', vlen)) 807 return http_request_header_char_invalid(r, '\0', 808 "invalid character in header -> 400"); 809 } 810 811 const enum http_header_e id = 812 hpctx->id ? hpctx->id : http_header_hkey_get_lc(k, klen); 813 return http_request_parse_single_header(r, id, k, klen, v, vlen); 814 } 815 } 816 else { /*(trailers)*/ 817 /* ignore trailers (after required HPACK decoding) if streaming 818 * request body to backend since headers have already been sent 819 * to backend via Common Gateway Interface (CGI) (CGI, FastCGI, 820 * SCGI, etc) or HTTP/1.1 (proxy) (mod_proxy does not currently 821 * support using HTTP/2 to connect to backends) */ 822 #if 0 /* (if needed, save flag in hpctx instead of fdevent.h dependency)*/ 823 if (r->conf.stream_request_body & FDEVENT_STREAM_REQUEST) 824 return 0; 825 #endif 826 /* Note: do not unconditionally merge into headers since if 827 * headers had already been sent to backend, then mod_accesslog 828 * logging of request headers might be inaccurate. 829 * Many simple backends do not support HTTP/1.1 requests sending 830 * Transfer-Encoding: chunked, and even those that do might not 831 * handle trailers. Some backends do not even support HTTP/1.1. 832 * For all these reasons, ignore trailers if streaming request 833 * body to backend. Revisit in future if adding support for 834 * connecting to backends using HTTP/2 (with explicit config 835 * option to force connecting to backends using HTTP/2) */ 836 837 /* XXX: TODO: request trailers not handled if streaming reqbody 838 * XXX: must ensure that trailers are not disallowed field-names 839 */ 840 841 #if 0 842 if (0 == vlen) 843 return 0; 844 #endif 845 846 return 0; 847 } 848 } 849 850 851 static int http_request_parse_reqline(request_st * const restrict r, const char * const restrict ptr, const unsigned short * const restrict hoff, const unsigned int http_parseopts) { 852 size_t len = hoff[2]; 853 854 /* parse the first line of the request 855 * <method> <uri> <protocol>\r\n 856 * */ 857 if (len < 13) /* minimum len with (!http_header_strict): "x x HTTP/1.0\n" */ 858 return http_request_header_line_invalid(r, 400, "invalid request line (too short) -> 400"); 859 if (ptr[len-2] == '\r') 860 len-=2; 861 else if (!(http_parseopts & HTTP_PARSEOPT_HEADER_STRICT)) /*(!http_header_strict)*/ 862 len-=1; 863 else 864 return http_request_header_line_invalid(r, 400, "missing CR before LF in header -> 400"); 865 866 /* 867 * RFC7230: 868 * HTTP-version = HTTP-name "/" DIGIT "." DIGIT 869 * HTTP-name = %x48.54.54.50 ; "HTTP", case-sensitive 870 */ 871 872 /* protocol is expected to be " HTTP/1.1" or " HTTP/1.0" at end of line */ 873 union proto_un { 874 char c[8]; 875 uint64_t u; 876 }; 877 static const union proto_un http_1_1 = {{'H','T','T','P','/','1','.','1'}}; 878 static const union proto_un http_1_0 = {{'H','T','T','P','/','1','.','0'}}; 879 const char *p = ptr + len - 8; 880 union proto_un proto8; 881 proto8.c[0]=p[0]; proto8.c[1]=p[1]; proto8.c[2]=p[2]; proto8.c[3]=p[3]; 882 proto8.c[4]=p[4]; proto8.c[5]=p[5]; proto8.c[6]=p[6]; proto8.c[7]=p[7]; 883 if (p[-1] == ' ' && http_1_1.u == proto8.u) { 884 r->http_version = HTTP_VERSION_1_1; 885 r->keep_alive = 1; /* keep-alive default: HTTP/1.1 -> true */ 886 } 887 else if (p[-1] == ' ' && http_1_0.u == proto8.u) { 888 r->http_version = HTTP_VERSION_1_0; 889 r->keep_alive = 0; /* keep-alive default: HTTP/1.0 -> false */ 890 } 891 else { 892 int status = http_request_parse_proto_loose(r,ptr,len,http_parseopts); 893 if (0 != status) return status; 894 /*(space char must exist if http_request_parse_proto_loose() succeeds)*/ 895 for (p = ptr + len - 9; p[-1] != ' '; --p) ; 896 } 897 898 /* method is expected to be a short string in the general case */ 899 size_t i = 0; 900 while (ptr[i] != ' ') ++i; 901 #if 0 /*(space must exist if protocol was parsed successfully)*/ 902 while (i < len && ptr[i] != ' ') ++i; 903 if (ptr[i] != ' ') 904 return http_request_header_line_invalid(r, 400, "incomplete request line -> 400"); 905 #endif 906 907 r->http_method = get_http_method_key(ptr, i); 908 if (HTTP_METHOD_UNSET >= r->http_method) 909 return http_request_header_line_invalid(r, 501, "unknown http-method -> 501"); 910 911 const char *uri = ptr + i + 1; 912 913 if (uri == p) 914 return http_request_header_line_invalid(r, 400, "no uri specified -> 400"); 915 len = (size_t)(p - uri - 1); 916 917 if (*uri != '/') { /* (common case: (*uri == '/')) */ 918 uri = http_request_parse_reqline_uri(r, uri, len, http_parseopts); 919 if (NULL == uri) return 400; 920 len = (size_t)(p - uri - 1); 921 } 922 923 if (0 == len) 924 return http_request_header_line_invalid(r, 400, "no uri specified -> 400"); 925 926 /* check uri for invalid characters */ 927 if (http_parseopts & HTTP_PARSEOPT_HEADER_STRICT) { /* http_header_strict */ 928 if ((http_parseopts & HTTP_PARSEOPT_URL_NORMALIZE_CTRLS_REJECT)) { 929 /* URI will be checked in http_request_parse_target() */ 930 } 931 else { 932 for (i = 0; i < len; ++i) { 933 if (!request_uri_is_valid_char(uri[i])) 934 return http_request_header_char_invalid(r, uri[i], "invalid character in URI -> 400"); 935 } 936 } 937 } 938 else { 939 /* check entire set of request headers for '\0' */ 940 if (NULL != memchr(ptr, '\0', hoff[hoff[0]])) 941 return http_request_header_char_invalid(r, '\0', "invalid character in header -> 400"); 942 } 943 944 buffer_copy_string_len(&r->target, uri, len); 945 buffer_copy_string_len(&r->target_orig, uri, len); 946 return 0; 947 } 948 949 int http_request_parse_target(request_st * const r, int scheme_port) { 950 /* URI is parsed into components at start of request and may 951 * also be re-parsed upon HANDLER_COMEBACK during the request 952 * r->target is expected to be a "/url-part?query-part" 953 * (and *not* a fully-qualified URI starting https://...) 954 * r->uri.authority is expected to be parsed elsewhere into r->http_host 955 */ 956 957 /** 958 * prepare strings 959 * 960 * - uri.path 961 * - uri.query 962 * 963 */ 964 965 /** 966 * Name according to RFC 2396 967 * 968 * - scheme 969 * - authority 970 * - path 971 * - query 972 * 973 * (scheme)://(authority)(path)?(query)#fragment 974 * 975 */ 976 977 /* take initial scheme value from connection-level state 978 * (request r->uri.scheme can be overwritten for later, 979 * for example by mod_extforward or mod_magnet) */ 980 buffer_copy_string_len(&r->uri.scheme, "https", scheme_port == 443 ? 5 : 4); 981 982 buffer * const target = &r->target; 983 if (r->http_method == HTTP_METHOD_CONNECT 984 || (r->http_method == HTTP_METHOD_OPTIONS 985 && target->ptr[0] == '*' 986 && target->ptr[1] == '\0')) { 987 /* CONNECT ... (or) OPTIONS * ... */ 988 buffer_copy_buffer(&r->uri.path, target); 989 buffer_clear(&r->uri.query); 990 return 0; 991 } 992 993 char *qstr; 994 if (r->conf.http_parseopts & HTTP_PARSEOPT_URL_NORMALIZE) { 995 /*uint32_t len = (uint32_t)buffer_string_length(target);*/ 996 int qs = burl_normalize(target, r->tmp_buf, r->conf.http_parseopts); 997 if (-2 == qs) 998 return http_request_header_line_invalid(r, 400, 999 "invalid character in URI -> 400"); /* Bad Request */ 1000 qstr = (-1 == qs) ? NULL : target->ptr+qs; 1001 #if 0 /* future: might enable here, or below for all requests */ 1002 /* (Note: total header size not recalculated on HANDLER_COMEBACK 1003 * even if other request headers changed during processing) 1004 * (If (0 != r->loops_per_request), then the generated 1005 * request is too large. Should a different error be returned?) */ 1006 r->rqst_header_len -= len; 1007 len = buffer_string_length(target); 1008 r->rqst_header_len += len; 1009 if (len > MAX_HTTP_REQUEST_URI) { 1010 return 414; /* 414 URI Too Long */ 1011 } 1012 if (r->rqst_header_len > MAX_HTTP_REQUEST_HEADER) { 1013 log_error(r->conf.errh, __FILE__, __LINE__, 1014 "request header fields too large: %u -> 431", 1015 r->rqst_header_len); 1016 return 431; /* Request Header Fields Too Large */ 1017 } 1018 #endif 1019 } 1020 else { 1021 size_t rlen = buffer_string_length(target); 1022 qstr = memchr(target->ptr, '#', rlen);/* discard fragment */ 1023 if (qstr) { 1024 rlen = (size_t)(qstr - target->ptr); 1025 buffer_string_set_length(target, rlen); 1026 } 1027 qstr = memchr(target->ptr, '?', rlen); 1028 } 1029 1030 /** extract query string from target */ 1031 const char * const pstr = target->ptr; 1032 const uint32_t rlen = buffer_string_length(target); 1033 uint32_t plen; 1034 if (NULL != qstr) { 1035 plen = (uint32_t)(qstr - pstr); 1036 buffer_copy_string_len(&r->uri.query, qstr + 1, rlen - plen - 1); 1037 } 1038 else { 1039 plen = rlen; 1040 buffer_clear(&r->uri.query); 1041 } 1042 buffer_copy_string_len(&r->uri.path, pstr, plen); 1043 1044 /* decode url to path 1045 * 1046 * - decode url-encodings (e.g. %20 -> ' ') 1047 * - remove path-modifiers (e.g. /../) 1048 */ 1049 1050 buffer_urldecode_path(&r->uri.path); 1051 buffer_path_simplify(&r->uri.path, &r->uri.path); 1052 if (r->uri.path.ptr[0] != '/') 1053 return http_request_header_line_invalid(r, 400, 1054 "uri-path does not begin with '/' -> 400"); /* Bad Request */ 1055 1056 return 0; 1057 } 1058 1059 __attribute_cold__ 1060 __attribute_noinline__ 1061 static int http_request_parse_header_other(request_st * const restrict r, const char * const restrict k, const int klen, const unsigned int http_header_strict) { 1062 for (int i = 0; i < klen; ++i) { 1063 if (light_isalpha(k[i]) || k[i] == '-') continue; /*(common cases)*/ 1064 /** 1065 * 1*<any CHAR except CTLs or separators> 1066 * CTLs == 0-31 + 127, CHAR = 7-bit ascii (0..127) 1067 * 1068 */ 1069 switch(k[i]) { 1070 case ' ': 1071 case '\t': 1072 return http_request_header_line_invalid(r, 400, "WS character in key -> 400"); 1073 case '(': 1074 case ')': 1075 case '<': 1076 case '>': 1077 case '@': 1078 case ',': 1079 case ';': 1080 case '\\': 1081 case '\"': 1082 case '/': 1083 case '[': 1084 case ']': 1085 case '?': 1086 case '=': 1087 case '{': 1088 case '}': 1089 return http_request_header_char_invalid(r, k[i], "invalid character in header key -> 400"); 1090 default: 1091 if (http_header_strict ? (k[i] < 32 || ((unsigned char *)k)[i] >= 127) : k[i] == '\0') 1092 return http_request_header_char_invalid(r, k[i], "invalid character in header key -> 400"); 1093 break; /* ok */ 1094 } 1095 } 1096 return 0; 1097 } 1098 1099 static int http_request_parse_headers(request_st * const restrict r, char * const restrict ptr, const unsigned short * const restrict hoff, const unsigned int http_parseopts) { 1100 const unsigned int http_header_strict = (http_parseopts & HTTP_PARSEOPT_HEADER_STRICT); 1101 1102 #if 0 /*(not checked here; will later result in invalid label for HTTP header)*/ 1103 int i = hoff[2]; 1104 1105 if (ptr[i] == ' ' || ptr[i] == '\t') { 1106 return http_request_header_line_invalid(r, 400, "WS at the start of first line -> 400"); 1107 } 1108 #endif 1109 1110 for (int i = 2; i < hoff[0]; ++i) { 1111 const char *k = ptr + hoff[i]; 1112 /* one past last line hoff[hoff[0]] is to final "\r\n" */ 1113 char *end = ptr + hoff[i+1]; 1114 1115 const char *colon = memchr(k, ':', end - k); 1116 if (NULL == colon) 1117 return http_request_header_line_invalid(r, 400, "invalid header missing ':' -> 400"); 1118 1119 const char *v = colon + 1; 1120 1121 /* RFC7230 Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing 1122 * 3.2.4. Field Parsing 1123 * [...] 1124 * No whitespace is allowed between the header field-name and colon. In 1125 * the past, differences in the handling of such whitespace have led to 1126 * security vulnerabilities in request routing and response handling. A 1127 * server MUST reject any received request message that contains 1128 * whitespace between a header field-name and colon with a response code 1129 * of 400 (Bad Request). A proxy MUST remove any such whitespace from a 1130 * response message before forwarding the message downstream. 1131 */ 1132 /* (line k[-1] is always preceded by a '\n', 1133 * including first header after request-line, 1134 * so no need to check colon != k) */ 1135 if (colon[-1] == ' ' || colon[-1] == '\t') { 1136 if (http_header_strict) { 1137 return http_request_header_line_invalid(r, 400, "invalid whitespace between field-name and colon -> 400"); 1138 } 1139 else { 1140 /* remove trailing whitespace from key(if !http_header_strict)*/ 1141 do { --colon; } while (colon[-1] == ' ' || colon[-1] == '\t'); 1142 } 1143 } 1144 1145 const int klen = (int)(colon - k); 1146 if (0 == klen) 1147 return http_request_header_line_invalid(r, 400, "invalid header key -> 400"); 1148 const enum http_header_e id = http_header_hkey_get(k, klen); 1149 1150 if (id == HTTP_HEADER_OTHER) { 1151 for (int j = 0; j < klen; ++j) { 1152 if (light_isalpha(k[j]) || k[j] == '-') continue; /*(common cases)*/ 1153 if (0 != http_request_parse_header_other(r, k+j, klen-j, http_header_strict)) 1154 return 400; 1155 break; 1156 } 1157 } 1158 1159 /* remove leading whitespace from value */ 1160 while (*v == ' ' || *v == '\t') ++v; 1161 1162 for (; i+1 <= hoff[0]; ++i) { 1163 end = ptr + hoff[i+1]; 1164 if (end[0] != ' ' && end[0] != '\t') break; 1165 1166 /* line folding */ 1167 #ifdef __COVERITY__ 1168 force_assert(end - k >= 2); 1169 #endif 1170 if (end[-2] == '\r') 1171 end[-2] = ' '; 1172 else if (http_header_strict) 1173 return http_request_header_line_invalid(r, 400, "missing CR before LF in header -> 400"); 1174 end[-1] = ' '; 1175 } 1176 #ifdef __COVERITY__ 1177 /*(buf holding k has non-zero request-line, so end[-2] valid)*/ 1178 force_assert(end >= k + 2); 1179 #endif 1180 if (end[-2] == '\r') 1181 --end; 1182 else if (http_header_strict) 1183 return http_request_header_line_invalid(r, 400, "missing CR before LF in header -> 400"); 1184 /* remove trailing whitespace from value (+ remove '\r\n') */ 1185 /* (line k[-1] is always preceded by a '\n', 1186 * including first header after request-line, 1187 * so no need to check (end != k)) */ 1188 do { --end; } while (end[-1] == ' ' || end[-1] == '\t'); 1189 1190 const int vlen = (int)(end - v); 1191 /* empty header-fields are not allowed by HTTP-RFC, we just ignore them */ 1192 if (vlen <= 0) continue; /* ignore header */ 1193 1194 if (http_header_strict) { 1195 for (int j = 0; j < vlen; ++j) { 1196 if ((((unsigned char *)v)[j] < 32 && v[j] != '\t') || v[j]==127) 1197 return http_request_header_char_invalid(r, v[j], "invalid character in header -> 400"); 1198 } 1199 } /* else URI already checked in http_request_parse_reqline() for any '\0' */ 1200 1201 int status = http_request_parse_single_header(r, id, k, (size_t)klen, v, (size_t)vlen); 1202 if (0 != status) return status; 1203 } 1204 1205 return 0; 1206 } 1207 1208 1209 static int 1210 http_request_parse (request_st * const restrict r, const int scheme_port) 1211 { 1212 int status = http_request_parse_target(r, scheme_port); 1213 if (0 != status) return status; 1214 1215 /*(r->http_host might not be set until after parsing request headers)*/ 1216 buffer_copy_buffer(&r->uri.authority, r->http_host);/*(copy even if empty)*/ 1217 buffer_to_lower(&r->uri.authority); 1218 1219 /* post-processing */ 1220 const unsigned int http_parseopts = r->conf.http_parseopts; 1221 1222 /* check hostname field if it is set */ 1223 if (r->http_host) { 1224 if (0 != http_request_host_policy(r->http_host, 1225 http_parseopts, scheme_port)) 1226 return http_request_header_line_invalid(r, 400, "Invalid Hostname -> 400"); 1227 } 1228 else { 1229 if (r->http_version >= HTTP_VERSION_1_1) 1230 return http_request_header_line_invalid(r, 400, "HTTP/1.1 but Host missing -> 400"); 1231 } 1232 1233 if (0 == r->reqbody_length) { 1234 /* POST requires Content-Length (or Transfer-Encoding) 1235 * (-1 == r->reqbody_length when Transfer-Encoding: chunked)*/ 1236 if (HTTP_METHOD_POST == r->http_method 1237 && !light_btst(r->rqst_htags, HTTP_HEADER_CONTENT_LENGTH)) { 1238 return http_request_header_line_invalid(r, 411, "POST-request, but content-length missing -> 411"); 1239 } 1240 } 1241 else { 1242 /* (-1 == r->reqbody_length when Transfer-Encoding: chunked)*/ 1243 if (-1 == r->reqbody_length 1244 && light_btst(r->rqst_htags, HTTP_HEADER_CONTENT_LENGTH)) { 1245 /* RFC7230 Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing 1246 * 3.3.3. Message Body Length 1247 * [...] 1248 * If a message is received with both a Transfer-Encoding and a 1249 * Content-Length header field, the Transfer-Encoding overrides the 1250 * Content-Length. Such a message might indicate an attempt to 1251 * perform request smuggling (Section 9.5) or response splitting 1252 * (Section 9.4) and ought to be handled as an error. A sender MUST 1253 * remove the received Content-Length field prior to forwarding such 1254 * a message downstream. 1255 */ 1256 const unsigned int http_header_strict = 1257 (http_parseopts & HTTP_PARSEOPT_HEADER_STRICT); 1258 if (http_header_strict) { 1259 return http_request_header_line_invalid(r, 400, "invalid Transfer-Encoding + Content-Length -> 400"); 1260 } 1261 else { 1262 /* ignore Content-Length */ 1263 http_header_request_unset(r, HTTP_HEADER_CONTENT_LENGTH, CONST_STR_LEN("Content-Length")); 1264 } 1265 } 1266 if (http_method_get_or_head(r->http_method) 1267 && !(http_parseopts & HTTP_PARSEOPT_METHOD_GET_BODY)) { 1268 return http_request_header_line_invalid(r, 400, "GET/HEAD with content-length -> 400"); 1269 } 1270 } 1271 1272 return 0; 1273 } 1274 1275 1276 static int 1277 http_request_parse_hoff (request_st * const restrict r, char * const restrict hdrs, const unsigned short * const restrict hoff, const int scheme_port) 1278 { 1279 /* 1280 * Request: "^(GET|POST|HEAD|...) ([^ ]+(\\?[^ ]+|)) (HTTP/1\\.[01])$" 1281 * Header : "^([-a-zA-Z]+): (.+)$" 1282 * End : "^$" 1283 */ 1284 1285 int status; 1286 const unsigned int http_parseopts = r->conf.http_parseopts; 1287 1288 status = http_request_parse_reqline(r, hdrs, hoff, http_parseopts); 1289 if (0 != status) return status; 1290 1291 status = http_request_parse_headers(r, hdrs, hoff, http_parseopts); 1292 if (0 != status) return status; 1293 1294 return http_request_parse(r, scheme_port); 1295 } 1296 1297 1298 static void 1299 http_request_headers_fin (request_st * const restrict r) 1300 { 1301 if (0 == r->http_status) { 1302 #if 0 1303 r->conditional_is_valid = (1 << COMP_SERVER_SOCKET) 1304 | (1 << COMP_HTTP_SCHEME) 1305 | (1 << COMP_HTTP_HOST) 1306 | (1 << COMP_HTTP_REMOTE_IP) 1307 | (1 << COMP_HTTP_REQUEST_METHOD) 1308 | (1 << COMP_HTTP_URL) 1309 | (1 << COMP_HTTP_QUERY_STRING) 1310 | (1 << COMP_HTTP_REQUEST_HEADER); 1311 #else 1312 /* all config conditions are valid after parsing header 1313 * (set all bits; remove dependency on plugin_config.h) */ 1314 r->conditional_is_valid = ~0u; 1315 #endif 1316 } 1317 else { 1318 r->keep_alive = 0; 1319 r->reqbody_length = 0; 1320 } 1321 } 1322 1323 1324 void 1325 http_request_headers_process (request_st * const restrict r, char * const restrict hdrs, const unsigned short * const restrict hoff, const int scheme_port) 1326 { 1327 r->http_status = http_request_parse_hoff(r, hdrs, hoff, scheme_port); 1328 1329 http_request_headers_fin(r); 1330 1331 if (__builtin_expect( (0 != r->http_status), 0)) { 1332 if (r->conf.log_request_header_on_error) { 1333 /*(http_request_parse_headers() modifies hdrs only to 1334 * undo line-wrapping in-place using spaces)*/ 1335 log_error(r->conf.errh, __FILE__, __LINE__, 1336 "request-header:\n%.*s", (int)r->rqst_header_len, hdrs); 1337 } 1338 } 1339 } 1340 1341 1342 void 1343 http_request_headers_process_h2 (request_st * const restrict r, const int scheme_port) 1344 { 1345 if (0 == r->http_status) 1346 r->http_status = http_request_parse(r, scheme_port); 1347 1348 if (0 == r->http_status) { 1349 if (light_btst(r->rqst_htags, HTTP_HEADER_CONNECTION)) 1350 r->http_status = http_request_header_line_invalid(r, 400, 1351 "invalid Connection header with HTTP/2 -> 400"); 1352 } 1353 1354 http_request_headers_fin(r); 1355 1356 /* limited; headers not collected into a single buf for HTTP/2 */ 1357 if (__builtin_expect( (0 != r->http_status), 0)) { 1358 if (r->conf.log_request_header_on_error) { 1359 log_error(r->conf.errh, __FILE__, __LINE__, 1360 "request-header:\n:authority: %s\n:method: %s\n:path: %s", 1361 r->http_host ? r->http_host->ptr : "", 1362 (HTTP_METHOD_UNSET != r->http_method) 1363 ? get_http_method_name(r->http_method) 1364 : "", 1365 !buffer_string_is_empty(&r->target) ? r->target.ptr : ""); 1366 } 1367 } 1368 1369 /* ignore Upgrade if using HTTP/2 */ 1370 if (light_btst(r->rqst_htags, HTTP_HEADER_UPGRADE)) 1371 http_header_request_unset(r, HTTP_HEADER_UPGRADE, 1372 CONST_STR_LEN("upgrade")); 1373 /* XXX: should filter out other hop-by-hop connection headers, too */ 1374 } 1375