blob: 82eec8d38471ee4f5accadc18219ed1755a2d5b3 [file] [log] [blame]
Willy Tarreaubaaee002006-06-26 02:48:02 +02001/*
2 * General purpose functions.
3 *
Willy Tarreau348238b2010-01-18 15:05:57 +01004 * Copyright 2000-2010 Willy Tarreau <w@1wt.eu>
Willy Tarreaubaaee002006-06-26 02:48:02 +02005 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version
9 * 2 of the License, or (at your option) any later version.
10 *
11 */
12
Baruch Siache1651b22020-07-24 07:52:20 +030013#if (defined(__ELF__) && !defined(__linux__)) || defined(USE_DL)
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +010014#define _GNU_SOURCE
15#include <dlfcn.h>
16#include <link.h>
17#endif
18
devnexen@gmail.comc4e52322021-08-17 12:55:49 +010019#if defined(__FreeBSD__)
20#include <elf.h>
21#include <dlfcn.h>
22extern void *__elf_aux_vector;
23#endif
24
David Carlierbd2cced2021-08-17 08:44:25 +010025#if defined(__NetBSD__)
26#include <sys/exec_elf.h>
27#include <dlfcn.h>
28#endif
29
Willy Tarreau2e74c3f2007-12-02 18:45:09 +010030#include <ctype.h>
Willy Tarreau16e01562016-08-09 16:46:18 +020031#include <errno.h>
Willy Tarreaubaaee002006-06-26 02:48:02 +020032#include <netdb.h>
Willy Tarreau9a7bea52012-04-27 11:16:50 +020033#include <stdarg.h>
Willy Tarreaudd2f85e2012-09-02 22:34:23 +020034#include <stdio.h>
Willy Tarreaubaaee002006-06-26 02:48:02 +020035#include <stdlib.h>
36#include <string.h>
Thierry Fournier93127942016-01-20 18:49:45 +010037#include <time.h>
Willy Tarreau16e01562016-08-09 16:46:18 +020038#include <unistd.h>
Willy Tarreau127f9662007-12-06 00:53:51 +010039#include <sys/socket.h>
Willy Tarreau37101052019-05-20 16:48:20 +020040#include <sys/stat.h>
41#include <sys/types.h>
Willy Tarreau127f9662007-12-06 00:53:51 +010042#include <sys/un.h>
Willy Tarreaubaaee002006-06-26 02:48:02 +020043#include <netinet/in.h>
44#include <arpa/inet.h>
45
David Carlier43a56852022-03-04 15:50:48 +000046#if defined(__linux__) && defined(__GLIBC__) && (__GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ >= 16))
Willy Tarreau30053062020-08-20 16:39:14 +020047#include <sys/auxv.h>
48#endif
49
Willy Tarreau48fbcae2020-06-03 18:09:46 +020050#include <import/eb32sctree.h>
Willy Tarreaub2551052020-06-09 09:07:15 +020051#include <import/eb32tree.h>
Willy Tarreau0dc9e6d2022-08-01 11:55:57 +020052#include <import/ebmbtree.h>
Willy Tarreau48fbcae2020-06-03 18:09:46 +020053
Willy Tarreau4c7e4b72020-05-27 12:58:42 +020054#include <haproxy/api.h>
Christopher Faulet6b0a0fb2022-04-04 11:29:28 +020055#include <haproxy/applet.h>
Willy Tarreauc13ed532020-06-02 10:22:45 +020056#include <haproxy/chunk.h>
Willy Tarreau7c18b542020-06-11 09:23:02 +020057#include <haproxy/dgram.h>
Willy Tarreauf268ee82020-06-04 17:05:57 +020058#include <haproxy/global.h>
Willy Tarreau86416052020-06-04 09:20:54 +020059#include <haproxy/hlua.h>
Willy Tarreau213e9902020-06-04 14:58:24 +020060#include <haproxy/listener.h>
Willy Tarreau7a00efb2020-06-02 17:02:59 +020061#include <haproxy/namespace.h>
Christopher Faulet9553de72021-02-26 09:12:50 +010062#include <haproxy/net_helper.h>
Willy Tarreau5fc93282020-09-16 18:25:03 +020063#include <haproxy/protocol.h>
Willy Tarreau8f6da642023-03-10 12:04:02 +010064#include <haproxy/quic_sock.h>
Emeric Brunc9437992021-02-12 19:42:55 +010065#include <haproxy/resolvers.h>
Willy Tarreau5edca2f2022-05-27 09:25:10 +020066#include <haproxy/sc_strm.h>
Willy Tarreau586f71b2020-12-11 15:54:36 +010067#include <haproxy/sock.h>
Willy Tarreau209108d2020-06-04 20:30:20 +020068#include <haproxy/ssl_sock.h>
William Lallemand3aeb3f92021-08-21 23:59:56 +020069#include <haproxy/ssl_utils.h>
Willy Tarreaucb086c62022-05-27 09:47:12 +020070#include <haproxy/stconn.h>
Willy Tarreaucea0e1b2020-06-04 17:25:40 +020071#include <haproxy/task.h>
Willy Tarreau48fbcae2020-06-03 18:09:46 +020072#include <haproxy/tools.h>
Willy Tarreau08093cc2022-11-24 08:09:12 +010073#include <haproxy/xxhash.h>
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +010074
Thierry Fournier93127942016-01-20 18:49:45 +010075/* This macro returns false if the test __x is false. Many
76 * of the following parsing function must be abort the processing
77 * if it returns 0, so this macro is useful for writing light code.
78 */
79#define RET0_UNLESS(__x) do { if (!(__x)) return 0; } while (0)
80
Erwan Le Goas9c766372022-09-14 17:40:24 +020081/* Define the number of line of hash_word */
Erwan Le Goas5eef1582022-09-29 10:25:31 +020082#define NB_L_HASH_WORD 15
Erwan Le Goas9c766372022-09-14 17:40:24 +020083
Willy Tarreau08093cc2022-11-24 08:09:12 +010084/* return the hash of a string and length for a given key. All keys are valid. */
85#define HA_ANON(key, str, len) (XXH32(str, len, key) & 0xFFFFFF)
86
Willy Tarreau56adcf22012-12-23 18:00:29 +010087/* enough to store NB_ITOA_STR integers of :
Willy Tarreau72d759c2007-10-25 12:14:10 +020088 * 2^64-1 = 18446744073709551615 or
89 * -2^63 = -9223372036854775808
Willy Tarreaue7239b52009-03-29 13:41:58 +020090 *
91 * The HTML version needs room for adding the 25 characters
92 * '<span class="rls"></span>' around digits at positions 3N+1 in order
93 * to add spacing at up to 6 positions : 18 446 744 073 709 551 615
Willy Tarreau72d759c2007-10-25 12:14:10 +020094 */
Christopher Faulet99bca652017-11-14 16:47:26 +010095THREAD_LOCAL char itoa_str[NB_ITOA_STR][171];
96THREAD_LOCAL int itoa_idx = 0; /* index of next itoa_str to use */
Willy Tarreaubaaee002006-06-26 02:48:02 +020097
Willy Tarreau588297f2014-06-16 15:16:40 +020098/* sometimes we'll need to quote strings (eg: in stats), and we don't expect
99 * to quote strings larger than a max configuration line.
100 */
Christopher Faulet99bca652017-11-14 16:47:26 +0100101THREAD_LOCAL char quoted_str[NB_QSTR][QSTR_SIZE + 1];
102THREAD_LOCAL int quoted_idx = 0;
Willy Tarreau588297f2014-06-16 15:16:40 +0200103
Willy Tarreau06e69b52021-03-02 14:01:35 +0100104/* thread-local PRNG state. It's modified to start from a different sequence
105 * on all threads upon startup. It must not be used or anything beyond getting
106 * statistical values as it's 100% predictable.
107 */
108THREAD_LOCAL unsigned int statistical_prng_state = 2463534242U;
109
Willy Tarreau5b3cd952022-07-18 13:58:17 +0200110/* set to true if this is a static build */
111int build_is_static = 0;
112
Erwan Le Goas9c766372022-09-14 17:40:24 +0200113/* A global static table to store hashed words */
114static THREAD_LOCAL char hash_word[NB_L_HASH_WORD][20];
115static THREAD_LOCAL int index_hash = 0;
116
Willy Tarreaubaaee002006-06-26 02:48:02 +0200117/*
William Lallemande7340ec2012-01-24 11:15:39 +0100118 * unsigned long long ASCII representation
119 *
120 * return the last char '\0' or NULL if no enough
121 * space in dst
122 */
123char *ulltoa(unsigned long long n, char *dst, size_t size)
124{
125 int i = 0;
126 char *res;
127
128 switch(n) {
129 case 1ULL ... 9ULL:
130 i = 0;
131 break;
132
133 case 10ULL ... 99ULL:
134 i = 1;
135 break;
136
137 case 100ULL ... 999ULL:
138 i = 2;
139 break;
140
141 case 1000ULL ... 9999ULL:
142 i = 3;
143 break;
144
145 case 10000ULL ... 99999ULL:
146 i = 4;
147 break;
148
149 case 100000ULL ... 999999ULL:
150 i = 5;
151 break;
152
153 case 1000000ULL ... 9999999ULL:
154 i = 6;
155 break;
156
157 case 10000000ULL ... 99999999ULL:
158 i = 7;
159 break;
160
161 case 100000000ULL ... 999999999ULL:
162 i = 8;
163 break;
164
165 case 1000000000ULL ... 9999999999ULL:
166 i = 9;
167 break;
168
169 case 10000000000ULL ... 99999999999ULL:
170 i = 10;
171 break;
172
173 case 100000000000ULL ... 999999999999ULL:
174 i = 11;
175 break;
176
177 case 1000000000000ULL ... 9999999999999ULL:
178 i = 12;
179 break;
180
181 case 10000000000000ULL ... 99999999999999ULL:
182 i = 13;
183 break;
184
185 case 100000000000000ULL ... 999999999999999ULL:
186 i = 14;
187 break;
188
189 case 1000000000000000ULL ... 9999999999999999ULL:
190 i = 15;
191 break;
192
193 case 10000000000000000ULL ... 99999999999999999ULL:
194 i = 16;
195 break;
196
197 case 100000000000000000ULL ... 999999999999999999ULL:
198 i = 17;
199 break;
200
201 case 1000000000000000000ULL ... 9999999999999999999ULL:
202 i = 18;
203 break;
204
205 case 10000000000000000000ULL ... ULLONG_MAX:
206 i = 19;
207 break;
208 }
209 if (i + 2 > size) // (i + 1) + '\0'
210 return NULL; // too long
211 res = dst + i + 1;
212 *res = '\0';
213 for (; i >= 0; i--) {
214 dst[i] = n % 10ULL + '0';
215 n /= 10ULL;
216 }
217 return res;
218}
219
220/*
221 * unsigned long ASCII representation
222 *
223 * return the last char '\0' or NULL if no enough
224 * space in dst
225 */
226char *ultoa_o(unsigned long n, char *dst, size_t size)
227{
228 int i = 0;
229 char *res;
230
231 switch (n) {
232 case 0U ... 9UL:
233 i = 0;
234 break;
235
236 case 10U ... 99UL:
237 i = 1;
238 break;
239
240 case 100U ... 999UL:
241 i = 2;
242 break;
243
244 case 1000U ... 9999UL:
245 i = 3;
246 break;
247
248 case 10000U ... 99999UL:
249 i = 4;
250 break;
251
252 case 100000U ... 999999UL:
253 i = 5;
254 break;
255
256 case 1000000U ... 9999999UL:
257 i = 6;
258 break;
259
260 case 10000000U ... 99999999UL:
261 i = 7;
262 break;
263
264 case 100000000U ... 999999999UL:
265 i = 8;
266 break;
267#if __WORDSIZE == 32
268
269 case 1000000000ULL ... ULONG_MAX:
270 i = 9;
271 break;
272
273#elif __WORDSIZE == 64
274
275 case 1000000000ULL ... 9999999999UL:
276 i = 9;
277 break;
278
279 case 10000000000ULL ... 99999999999UL:
280 i = 10;
281 break;
282
283 case 100000000000ULL ... 999999999999UL:
284 i = 11;
285 break;
286
287 case 1000000000000ULL ... 9999999999999UL:
288 i = 12;
289 break;
290
291 case 10000000000000ULL ... 99999999999999UL:
292 i = 13;
293 break;
294
295 case 100000000000000ULL ... 999999999999999UL:
296 i = 14;
297 break;
298
299 case 1000000000000000ULL ... 9999999999999999UL:
300 i = 15;
301 break;
302
303 case 10000000000000000ULL ... 99999999999999999UL:
304 i = 16;
305 break;
306
307 case 100000000000000000ULL ... 999999999999999999UL:
308 i = 17;
309 break;
310
311 case 1000000000000000000ULL ... 9999999999999999999UL:
312 i = 18;
313 break;
314
315 case 10000000000000000000ULL ... ULONG_MAX:
316 i = 19;
317 break;
318
319#endif
320 }
321 if (i + 2 > size) // (i + 1) + '\0'
322 return NULL; // too long
323 res = dst + i + 1;
324 *res = '\0';
325 for (; i >= 0; i--) {
326 dst[i] = n % 10U + '0';
327 n /= 10U;
328 }
329 return res;
330}
331
332/*
333 * signed long ASCII representation
334 *
335 * return the last char '\0' or NULL if no enough
336 * space in dst
337 */
338char *ltoa_o(long int n, char *dst, size_t size)
339{
340 char *pos = dst;
341
342 if (n < 0) {
343 if (size < 3)
344 return NULL; // min size is '-' + digit + '\0' but another test in ultoa
345 *pos = '-';
346 pos++;
347 dst = ultoa_o(-n, pos, size - 1);
348 } else {
349 dst = ultoa_o(n, dst, size);
350 }
351 return dst;
352}
353
354/*
355 * signed long long ASCII representation
356 *
357 * return the last char '\0' or NULL if no enough
358 * space in dst
359 */
360char *lltoa(long long n, char *dst, size_t size)
361{
362 char *pos = dst;
363
364 if (n < 0) {
365 if (size < 3)
366 return NULL; // min size is '-' + digit + '\0' but another test in ulltoa
367 *pos = '-';
368 pos++;
369 dst = ulltoa(-n, pos, size - 1);
370 } else {
371 dst = ulltoa(n, dst, size);
372 }
373 return dst;
374}
375
376/*
377 * write a ascii representation of a unsigned into dst,
378 * return a pointer to the last character
379 * Pad the ascii representation with '0', using size.
380 */
381char *utoa_pad(unsigned int n, char *dst, size_t size)
382{
383 int i = 0;
384 char *ret;
385
386 switch(n) {
387 case 0U ... 9U:
388 i = 0;
389 break;
390
391 case 10U ... 99U:
392 i = 1;
393 break;
394
395 case 100U ... 999U:
396 i = 2;
397 break;
398
399 case 1000U ... 9999U:
400 i = 3;
401 break;
402
403 case 10000U ... 99999U:
404 i = 4;
405 break;
406
407 case 100000U ... 999999U:
408 i = 5;
409 break;
410
411 case 1000000U ... 9999999U:
412 i = 6;
413 break;
414
415 case 10000000U ... 99999999U:
416 i = 7;
417 break;
418
419 case 100000000U ... 999999999U:
420 i = 8;
421 break;
422
423 case 1000000000U ... 4294967295U:
424 i = 9;
425 break;
426 }
427 if (i + 2 > size) // (i + 1) + '\0'
428 return NULL; // too long
Aurelien DARRAGONe3177af2022-11-22 11:42:07 +0100429 i = size - 2; // padding - '\0'
William Lallemande7340ec2012-01-24 11:15:39 +0100430
431 ret = dst + i + 1;
432 *ret = '\0';
433 for (; i >= 0; i--) {
434 dst[i] = n % 10U + '0';
435 n /= 10U;
436 }
437 return ret;
438}
439
440/*
Willy Tarreaubaaee002006-06-26 02:48:02 +0200441 * copies at most <size-1> chars from <src> to <dst>. Last char is always
442 * set to 0, unless <size> is 0. The number of chars copied is returned
443 * (excluding the terminating zero).
444 * This code has been optimized for size and speed : on x86, it's 45 bytes
445 * long, uses only registers, and consumes only 4 cycles per char.
446 */
447int strlcpy2(char *dst, const char *src, int size)
448{
449 char *orig = dst;
450 if (size) {
451 while (--size && (*dst = *src)) {
452 src++; dst++;
453 }
454 *dst = 0;
455 }
456 return dst - orig;
457}
458
459/*
Willy Tarreau72d759c2007-10-25 12:14:10 +0200460 * This function simply returns a locally allocated string containing
Willy Tarreaubaaee002006-06-26 02:48:02 +0200461 * the ascii representation for number 'n' in decimal.
462 */
Emeric Brun3a7fce52010-01-04 14:54:38 +0100463char *ultoa_r(unsigned long n, char *buffer, int size)
Willy Tarreaubaaee002006-06-26 02:48:02 +0200464{
465 char *pos;
466
Willy Tarreau72d759c2007-10-25 12:14:10 +0200467 pos = buffer + size - 1;
Willy Tarreaubaaee002006-06-26 02:48:02 +0200468 *pos-- = '\0';
469
470 do {
471 *pos-- = '0' + n % 10;
472 n /= 10;
Willy Tarreau72d759c2007-10-25 12:14:10 +0200473 } while (n && pos >= buffer);
Willy Tarreaubaaee002006-06-26 02:48:02 +0200474 return pos + 1;
475}
476
Willy Tarreau91092e52007-10-25 16:58:42 +0200477/*
Willy Tarreaue7239b52009-03-29 13:41:58 +0200478 * This function simply returns a locally allocated string containing
Thierry FOURNIER763a5d82015-07-06 23:09:52 +0200479 * the ascii representation for number 'n' in decimal.
480 */
481char *lltoa_r(long long int in, char *buffer, int size)
482{
483 char *pos;
484 int neg = 0;
485 unsigned long long int n;
486
487 pos = buffer + size - 1;
488 *pos-- = '\0';
489
490 if (in < 0) {
491 neg = 1;
492 n = -in;
493 }
494 else
495 n = in;
496
497 do {
498 *pos-- = '0' + n % 10;
499 n /= 10;
500 } while (n && pos >= buffer);
501 if (neg && pos > buffer)
502 *pos-- = '-';
503 return pos + 1;
504}
505
506/*
507 * This function simply returns a locally allocated string containing
Thierry FOURNIER1480bd82015-06-06 19:14:59 +0200508 * the ascii representation for signed number 'n' in decimal.
509 */
510char *sltoa_r(long n, char *buffer, int size)
511{
512 char *pos;
513
514 if (n >= 0)
515 return ultoa_r(n, buffer, size);
516
517 pos = ultoa_r(-n, buffer + 1, size - 1) - 1;
518 *pos = '-';
519 return pos;
520}
521
522/*
523 * This function simply returns a locally allocated string containing
Willy Tarreaue7239b52009-03-29 13:41:58 +0200524 * the ascii representation for number 'n' in decimal, formatted for
525 * HTML output with tags to create visual grouping by 3 digits. The
526 * output needs to support at least 171 characters.
527 */
528const char *ulltoh_r(unsigned long long n, char *buffer, int size)
529{
530 char *start;
531 int digit = 0;
532
533 start = buffer + size;
534 *--start = '\0';
535
536 do {
537 if (digit == 3 && start >= buffer + 7)
538 memcpy(start -= 7, "</span>", 7);
539
540 if (start >= buffer + 1) {
541 *--start = '0' + n % 10;
542 n /= 10;
543 }
544
545 if (digit == 3 && start >= buffer + 18)
546 memcpy(start -= 18, "<span class=\"rls\">", 18);
547
548 if (digit++ == 3)
549 digit = 1;
550 } while (n && start > buffer);
551 return start;
552}
553
554/*
Willy Tarreau91092e52007-10-25 16:58:42 +0200555 * This function simply returns a locally allocated string containing the ascii
556 * representation for number 'n' in decimal, unless n is 0 in which case it
557 * returns the alternate string (or an empty string if the alternate string is
558 * NULL). It use is intended for limits reported in reports, where it's
559 * desirable not to display anything if there is no limit. Warning! it shares
560 * the same vector as ultoa_r().
561 */
562const char *limit_r(unsigned long n, char *buffer, int size, const char *alt)
563{
564 return (n) ? ultoa_r(n, buffer, size) : (alt ? alt : "");
565}
566
Willy Tarreau56d1d8d2021-05-08 10:28:53 +0200567/* Trims the first "%f" float in a string to its minimum number of digits after
568 * the decimal point by trimming trailing zeroes, even dropping the decimal
569 * point if not needed. The string is in <buffer> of length <len>, and the
570 * number is expected to start at or after position <num_start> (the first
571 * point appearing there is considered). A NUL character is always placed at
572 * the end if some trimming occurs. The new buffer length is returned.
573 */
574size_t flt_trim(char *buffer, size_t num_start, size_t len)
575{
576 char *end = buffer + len;
577 char *p = buffer + num_start;
578 char *trim;
579
580 do {
581 if (p >= end)
582 return len;
583 trim = p++;
584 } while (*trim != '.');
585
586 /* For now <trim> is on the decimal point. Let's look for any other
587 * meaningful digit after it.
588 */
589 while (p < end) {
590 if (*p++ != '0')
591 trim = p;
592 }
593
594 if (trim < end)
595 *trim = 0;
596
597 return trim - buffer;
598}
599
Willy Tarreauae03d262021-05-08 07:35:00 +0200600/*
601 * This function simply returns a locally allocated string containing
602 * the ascii representation for number 'n' in decimal with useless trailing
603 * zeroes trimmed.
604 */
605char *ftoa_r(double n, char *buffer, int size)
606{
607 flt_trim(buffer, 0, snprintf(buffer, size, "%f", n));
608 return buffer;
609}
610
Willy Tarreau588297f2014-06-16 15:16:40 +0200611/* returns a locally allocated string containing the quoted encoding of the
612 * input string. The output may be truncated to QSTR_SIZE chars, but it is
613 * guaranteed that the string will always be properly terminated. Quotes are
614 * encoded by doubling them as is commonly done in CSV files. QSTR_SIZE must
615 * always be at least 4 chars.
616 */
617const char *qstr(const char *str)
618{
619 char *ret = quoted_str[quoted_idx];
620 char *p, *end;
621
622 if (++quoted_idx >= NB_QSTR)
623 quoted_idx = 0;
624
625 p = ret;
626 end = ret + QSTR_SIZE;
627
628 *p++ = '"';
629
630 /* always keep 3 chars to support passing "" and the ending " */
631 while (*str && p < end - 3) {
632 if (*str == '"') {
633 *p++ = '"';
634 *p++ = '"';
635 }
636 else
637 *p++ = *str;
638 str++;
639 }
640 *p++ = '"';
641 return ret;
642}
643
Robert Tsai81ae1952007-12-05 10:47:29 +0100644/*
Willy Tarreaubaaee002006-06-26 02:48:02 +0200645 * Returns non-zero if character <s> is a hex digit (0-9, a-f, A-F), else zero.
646 *
647 * It looks like this one would be a good candidate for inlining, but this is
648 * not interesting because it around 35 bytes long and often called multiple
649 * times within the same function.
650 */
651int ishex(char s)
652{
653 s -= '0';
654 if ((unsigned char)s <= 9)
655 return 1;
656 s -= 'A' - '0';
657 if ((unsigned char)s <= 5)
658 return 1;
659 s -= 'a' - 'A';
660 if ((unsigned char)s <= 5)
661 return 1;
662 return 0;
663}
664
Willy Tarreau3ca1a882015-01-15 18:43:49 +0100665/* rounds <i> down to the closest value having max 2 digits */
666unsigned int round_2dig(unsigned int i)
667{
668 unsigned int mul = 1;
669
670 while (i >= 100) {
671 i /= 10;
672 mul *= 10;
673 }
674 return i * mul;
675}
676
Willy Tarreau2e74c3f2007-12-02 18:45:09 +0100677/*
678 * Checks <name> for invalid characters. Valid chars are [A-Za-z0-9_:.-]. If an
679 * invalid character is found, a pointer to it is returned. If everything is
680 * fine, NULL is returned.
681 */
682const char *invalid_char(const char *name)
683{
684 if (!*name)
685 return name;
686
687 while (*name) {
Willy Tarreau90807112020-02-25 08:16:33 +0100688 if (!isalnum((unsigned char)*name) && *name != '.' && *name != ':' &&
Willy Tarreau2e74c3f2007-12-02 18:45:09 +0100689 *name != '_' && *name != '-')
690 return name;
691 name++;
692 }
693 return NULL;
694}
Willy Tarreaubaaee002006-06-26 02:48:02 +0200695
696/*
Frédéric Lécailleb82f7422017-04-13 18:24:23 +0200697 * Checks <name> for invalid characters. Valid chars are [_.-] and those
698 * accepted by <f> function.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +0200699 * If an invalid character is found, a pointer to it is returned.
700 * If everything is fine, NULL is returned.
701 */
Frédéric Lécailleb82f7422017-04-13 18:24:23 +0200702static inline const char *__invalid_char(const char *name, int (*f)(int)) {
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +0200703
704 if (!*name)
705 return name;
706
707 while (*name) {
Willy Tarreau90807112020-02-25 08:16:33 +0100708 if (!f((unsigned char)*name) && *name != '.' &&
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +0200709 *name != '_' && *name != '-')
710 return name;
711
712 name++;
713 }
714
715 return NULL;
716}
717
718/*
Frédéric Lécailleb82f7422017-04-13 18:24:23 +0200719 * Checks <name> for invalid characters. Valid chars are [A-Za-z0-9_.-].
720 * If an invalid character is found, a pointer to it is returned.
721 * If everything is fine, NULL is returned.
722 */
723const char *invalid_domainchar(const char *name) {
724 return __invalid_char(name, isalnum);
725}
726
727/*
728 * Checks <name> for invalid characters. Valid chars are [A-Za-z_.-].
729 * If an invalid character is found, a pointer to it is returned.
730 * If everything is fine, NULL is returned.
731 */
732const char *invalid_prefix_char(const char *name) {
Thierry Fournierf7b7c3e2018-03-26 11:54:39 +0200733 return __invalid_char(name, isalnum);
Frédéric Lécailleb82f7422017-04-13 18:24:23 +0200734}
735
736/*
Willy Tarreauc120c8d2013-03-10 19:27:44 +0100737 * converts <str> to a struct sockaddr_storage* provided by the caller. The
Willy Tarreau24709282013-03-10 21:32:12 +0100738 * caller must have zeroed <sa> first, and may have set sa->ss_family to force
739 * parse a specific address format. If the ss_family is 0 or AF_UNSPEC, then
740 * the function tries to guess the address family from the syntax. If the
741 * family is forced and the format doesn't match, an error is returned. The
Willy Tarreaufab5a432011-03-04 15:31:53 +0100742 * string is assumed to contain only an address, no port. The address can be a
743 * dotted IPv4 address, an IPv6 address, a host name, or empty or "*" to
744 * indicate INADDR_ANY. NULL is returned if the host part cannot be resolved.
745 * The return address will only have the address family and the address set,
746 * all other fields remain zero. The string is not supposed to be modified.
Thierry FOURNIER58639a02014-11-25 12:02:25 +0100747 * The IPv6 '::' address is IN6ADDR_ANY. If <resolve> is non-zero, the hostname
748 * is resolved, otherwise only IP addresses are resolved, and anything else
Willy Tarreauecde7df2016-11-02 22:37:03 +0100749 * returns NULL. If the address contains a port, this one is preserved.
Willy Tarreaubaaee002006-06-26 02:48:02 +0200750 */
Thierry FOURNIER58639a02014-11-25 12:02:25 +0100751struct sockaddr_storage *str2ip2(const char *str, struct sockaddr_storage *sa, int resolve)
Willy Tarreaubaaee002006-06-26 02:48:02 +0200752{
Willy Tarreaufab5a432011-03-04 15:31:53 +0100753 struct hostent *he;
mildisff5d5102015-10-26 18:50:08 +0100754 /* max IPv6 length, including brackets and terminating NULL */
755 char tmpip[48];
Willy Tarreauecde7df2016-11-02 22:37:03 +0100756 int port = get_host_port(sa);
mildisff5d5102015-10-26 18:50:08 +0100757
758 /* check IPv6 with square brackets */
759 if (str[0] == '[') {
760 size_t iplength = strlen(str);
761
762 if (iplength < 4) {
763 /* minimal size is 4 when using brackets "[::]" */
764 goto fail;
765 }
766 else if (iplength >= sizeof(tmpip)) {
767 /* IPv6 literal can not be larger than tmpip */
768 goto fail;
769 }
770 else {
771 if (str[iplength - 1] != ']') {
772 /* if address started with bracket, it should end with bracket */
773 goto fail;
774 }
775 else {
776 memcpy(tmpip, str + 1, iplength - 2);
777 tmpip[iplength - 2] = '\0';
778 str = tmpip;
779 }
780 }
781 }
Willy Tarreaufab5a432011-03-04 15:31:53 +0100782
Willy Tarreaufab5a432011-03-04 15:31:53 +0100783 /* Any IPv6 address */
784 if (str[0] == ':' && str[1] == ':' && !str[2]) {
Willy Tarreau24709282013-03-10 21:32:12 +0100785 if (!sa->ss_family || sa->ss_family == AF_UNSPEC)
786 sa->ss_family = AF_INET6;
787 else if (sa->ss_family != AF_INET6)
788 goto fail;
Willy Tarreauecde7df2016-11-02 22:37:03 +0100789 set_host_port(sa, port);
Willy Tarreauc120c8d2013-03-10 19:27:44 +0100790 return sa;
Willy Tarreaufab5a432011-03-04 15:31:53 +0100791 }
792
Willy Tarreau24709282013-03-10 21:32:12 +0100793 /* Any address for the family, defaults to IPv4 */
Willy Tarreaufab5a432011-03-04 15:31:53 +0100794 if (!str[0] || (str[0] == '*' && !str[1])) {
Willy Tarreau24709282013-03-10 21:32:12 +0100795 if (!sa->ss_family || sa->ss_family == AF_UNSPEC)
796 sa->ss_family = AF_INET;
Willy Tarreauecde7df2016-11-02 22:37:03 +0100797 set_host_port(sa, port);
Willy Tarreauc120c8d2013-03-10 19:27:44 +0100798 return sa;
Willy Tarreaufab5a432011-03-04 15:31:53 +0100799 }
800
801 /* check for IPv6 first */
Willy Tarreau24709282013-03-10 21:32:12 +0100802 if ((!sa->ss_family || sa->ss_family == AF_UNSPEC || sa->ss_family == AF_INET6) &&
803 inet_pton(AF_INET6, str, &((struct sockaddr_in6 *)sa)->sin6_addr)) {
Willy Tarreauc120c8d2013-03-10 19:27:44 +0100804 sa->ss_family = AF_INET6;
Willy Tarreauecde7df2016-11-02 22:37:03 +0100805 set_host_port(sa, port);
Willy Tarreauc120c8d2013-03-10 19:27:44 +0100806 return sa;
Willy Tarreaufab5a432011-03-04 15:31:53 +0100807 }
808
809 /* then check for IPv4 */
Willy Tarreau24709282013-03-10 21:32:12 +0100810 if ((!sa->ss_family || sa->ss_family == AF_UNSPEC || sa->ss_family == AF_INET) &&
811 inet_pton(AF_INET, str, &((struct sockaddr_in *)sa)->sin_addr)) {
Willy Tarreauc120c8d2013-03-10 19:27:44 +0100812 sa->ss_family = AF_INET;
Willy Tarreauecde7df2016-11-02 22:37:03 +0100813 set_host_port(sa, port);
Willy Tarreauc120c8d2013-03-10 19:27:44 +0100814 return sa;
Willy Tarreaufab5a432011-03-04 15:31:53 +0100815 }
816
Thierry FOURNIER58639a02014-11-25 12:02:25 +0100817 if (!resolve)
818 return NULL;
819
Emeric Brund30e9a12020-12-23 18:49:16 +0100820 if (!resolv_hostname_validation(str, NULL))
Baptiste Assmanna68ca962015-04-14 01:15:08 +0200821 return NULL;
822
David du Colombierd5f43282011-03-17 10:40:16 +0100823#ifdef USE_GETADDRINFO
Nenad Merdanovic88afe032014-04-14 15:56:58 +0200824 if (global.tune.options & GTUNE_USE_GAI) {
David du Colombierd5f43282011-03-17 10:40:16 +0100825 struct addrinfo hints, *result;
Tim Duesterhus7d58b4d2018-01-21 22:11:17 +0100826 int success = 0;
David du Colombierd5f43282011-03-17 10:40:16 +0100827
828 memset(&result, 0, sizeof(result));
829 memset(&hints, 0, sizeof(hints));
Willy Tarreau24709282013-03-10 21:32:12 +0100830 hints.ai_family = sa->ss_family ? sa->ss_family : AF_UNSPEC;
David du Colombierd5f43282011-03-17 10:40:16 +0100831 hints.ai_socktype = SOCK_DGRAM;
Dmitry Sivachenkoeab7f392015-10-02 01:01:58 +0200832 hints.ai_flags = 0;
David du Colombierd5f43282011-03-17 10:40:16 +0100833 hints.ai_protocol = 0;
834
835 if (getaddrinfo(str, NULL, &hints, &result) == 0) {
Willy Tarreau24709282013-03-10 21:32:12 +0100836 if (!sa->ss_family || sa->ss_family == AF_UNSPEC)
837 sa->ss_family = result->ai_family;
Tim Duesterhus7d58b4d2018-01-21 22:11:17 +0100838 else if (sa->ss_family != result->ai_family) {
839 freeaddrinfo(result);
Willy Tarreau24709282013-03-10 21:32:12 +0100840 goto fail;
Tim Duesterhus7d58b4d2018-01-21 22:11:17 +0100841 }
Willy Tarreau24709282013-03-10 21:32:12 +0100842
David du Colombierd5f43282011-03-17 10:40:16 +0100843 switch (result->ai_family) {
844 case AF_INET:
Willy Tarreauc120c8d2013-03-10 19:27:44 +0100845 memcpy((struct sockaddr_in *)sa, result->ai_addr, result->ai_addrlen);
Willy Tarreauecde7df2016-11-02 22:37:03 +0100846 set_host_port(sa, port);
Tim Duesterhus7d58b4d2018-01-21 22:11:17 +0100847 success = 1;
848 break;
David du Colombierd5f43282011-03-17 10:40:16 +0100849 case AF_INET6:
Willy Tarreauc120c8d2013-03-10 19:27:44 +0100850 memcpy((struct sockaddr_in6 *)sa, result->ai_addr, result->ai_addrlen);
Willy Tarreauecde7df2016-11-02 22:37:03 +0100851 set_host_port(sa, port);
Tim Duesterhus7d58b4d2018-01-21 22:11:17 +0100852 success = 1;
853 break;
David du Colombierd5f43282011-03-17 10:40:16 +0100854 }
855 }
856
Sean Carey58ea0392013-02-15 23:39:18 +0100857 if (result)
858 freeaddrinfo(result);
Tim Duesterhus7d58b4d2018-01-21 22:11:17 +0100859
860 if (success)
861 return sa;
Willy Tarreaufab5a432011-03-04 15:31:53 +0100862 }
David du Colombierd5f43282011-03-17 10:40:16 +0100863#endif
Nenad Merdanovic88afe032014-04-14 15:56:58 +0200864 /* try to resolve an IPv4/IPv6 hostname */
865 he = gethostbyname(str);
866 if (he) {
867 if (!sa->ss_family || sa->ss_family == AF_UNSPEC)
868 sa->ss_family = he->h_addrtype;
869 else if (sa->ss_family != he->h_addrtype)
870 goto fail;
871
872 switch (sa->ss_family) {
873 case AF_INET:
874 ((struct sockaddr_in *)sa)->sin_addr = *(struct in_addr *) *(he->h_addr_list);
Willy Tarreauecde7df2016-11-02 22:37:03 +0100875 set_host_port(sa, port);
Nenad Merdanovic88afe032014-04-14 15:56:58 +0200876 return sa;
877 case AF_INET6:
878 ((struct sockaddr_in6 *)sa)->sin6_addr = *(struct in6_addr *) *(he->h_addr_list);
Willy Tarreauecde7df2016-11-02 22:37:03 +0100879 set_host_port(sa, port);
Nenad Merdanovic88afe032014-04-14 15:56:58 +0200880 return sa;
881 }
882 }
883
David du Colombierd5f43282011-03-17 10:40:16 +0100884 /* unsupported address family */
Willy Tarreau24709282013-03-10 21:32:12 +0100885 fail:
Willy Tarreaufab5a432011-03-04 15:31:53 +0100886 return NULL;
887}
888
889/*
Willy Tarreaud4448bc2013-02-20 15:55:15 +0100890 * Converts <str> to a locally allocated struct sockaddr_storage *, and a port
891 * range or offset consisting in two integers that the caller will have to
892 * check to find the relevant input format. The following format are supported :
893 *
894 * String format | address | port | low | high
895 * addr | <addr> | 0 | 0 | 0
896 * addr: | <addr> | 0 | 0 | 0
897 * addr:port | <addr> | <port> | <port> | <port>
898 * addr:pl-ph | <addr> | <pl> | <pl> | <ph>
899 * addr:+port | <addr> | <port> | 0 | <port>
900 * addr:-port | <addr> |-<port> | <port> | 0
901 *
902 * The detection of a port range or increment by the caller is made by
903 * comparing <low> and <high>. If both are equal, then port 0 means no port
904 * was specified. The caller may pass NULL for <low> and <high> if it is not
905 * interested in retrieving port ranges.
906 *
907 * Note that <addr> above may also be :
908 * - empty ("") => family will be AF_INET and address will be INADDR_ANY
909 * - "*" => family will be AF_INET and address will be INADDR_ANY
910 * - "::" => family will be AF_INET6 and address will be IN6ADDR_ANY
911 * - a host name => family and address will depend on host name resolving.
912 *
Willy Tarreau24709282013-03-10 21:32:12 +0100913 * A prefix may be passed in before the address above to force the family :
914 * - "ipv4@" => force address to resolve as IPv4 and fail if not possible.
915 * - "ipv6@" => force address to resolve as IPv6 and fail if not possible.
916 * - "unix@" => force address to be a path to a UNIX socket even if the
917 * path does not start with a '/'
Willy Tarreauccfccef2014-05-10 01:49:15 +0200918 * - 'abns@' -> force address to belong to the abstract namespace (Linux
919 * only). These sockets are just like Unix sockets but without
920 * the need for an underlying file system. The address is a
921 * string. Technically it's like a Unix socket with a zero in
922 * the first byte of the address.
Willy Tarreau40aa0702013-03-10 23:51:38 +0100923 * - "fd@" => an integer must follow, and is a file descriptor number.
Willy Tarreau24709282013-03-10 21:32:12 +0100924 *
mildisff5d5102015-10-26 18:50:08 +0100925 * IPv6 addresses can be declared with or without square brackets. When using
926 * square brackets for IPv6 addresses, the port separator (colon) is optional.
927 * If not using square brackets, and in order to avoid any ambiguity with
928 * IPv6 addresses, the last colon ':' is mandatory even when no port is specified.
929 * NULL is returned if the address cannot be parsed. The <low> and <high> ports
930 * are always initialized if non-null, even for non-IP families.
Willy Tarreaud393a622013-03-04 18:22:00 +0100931 *
932 * If <pfx> is non-null, it is used as a string prefix before any path-based
933 * address (typically the path to a unix socket).
Willy Tarreau40aa0702013-03-10 23:51:38 +0100934 *
Willy Tarreau72b8c1f2015-09-08 15:50:19 +0200935 * if <fqdn> is non-null, it will be filled with :
936 * - a pointer to the FQDN of the server name to resolve if there's one, and
937 * that the caller will have to free(),
938 * - NULL if there was an explicit address that doesn't require resolution.
939 *
Willy Tarreaucd3a55912020-09-04 15:30:46 +0200940 * Hostnames are only resolved if <opts> has PA_O_RESOLVE. Otherwise <fqdn> is
941 * still honored so it is possible for the caller to know whether a resolution
942 * failed by clearing this flag and checking if <fqdn> was filled, indicating
943 * the need for a resolution.
Thierry FOURNIER7fe3be72015-09-26 20:03:36 +0200944 *
Willy Tarreau40aa0702013-03-10 23:51:38 +0100945 * When a file descriptor is passed, its value is put into the s_addr part of
Willy Tarreaua5b325f2020-09-04 16:44:20 +0200946 * the address when cast to sockaddr_in and the address family is
947 * AF_CUST_EXISTING_FD.
Willy Tarreaua93e5c72020-09-15 14:01:16 +0200948 *
Willy Tarreau5fc93282020-09-16 18:25:03 +0200949 * The matching protocol will be set into <proto> if non-null.
950 *
Willy Tarreaua93e5c72020-09-15 14:01:16 +0200951 * Any known file descriptor is also assigned to <fd> if non-null, otherwise it
952 * is forced to -1.
Willy Tarreaufab5a432011-03-04 15:31:53 +0100953 */
Willy Tarreau5fc93282020-09-16 18:25:03 +0200954struct sockaddr_storage *str2sa_range(const char *str, int *port, int *low, int *high, int *fd,
955 struct protocol **proto, char **err,
956 const char *pfx, char **fqdn, unsigned int opts)
Willy Tarreaufab5a432011-03-04 15:31:53 +0100957{
Christopher Faulet1bc04c72017-10-29 20:14:08 +0100958 static THREAD_LOCAL struct sockaddr_storage ss;
David du Colombier6f5ccb12011-03-10 22:26:24 +0100959 struct sockaddr_storage *ret = NULL;
Willy Tarreau5fc93282020-09-16 18:25:03 +0200960 struct protocol *new_proto = NULL;
Willy Tarreau24709282013-03-10 21:32:12 +0100961 char *back, *str2;
Willy Tarreaud4448bc2013-02-20 15:55:15 +0100962 char *port1, *port2;
963 int portl, porth, porta;
Willy Tarreauccfccef2014-05-10 01:49:15 +0200964 int abstract = 0;
Willy Tarreaua93e5c72020-09-15 14:01:16 +0200965 int new_fd = -1;
Willy Tarreaue3b45182021-10-27 17:28:55 +0200966 enum proto_type proto_type;
967 int ctrl_type;
Willy Tarreaud4448bc2013-02-20 15:55:15 +0100968
969 portl = porth = porta = 0;
Willy Tarreau72b8c1f2015-09-08 15:50:19 +0200970 if (fqdn)
971 *fqdn = NULL;
Willy Tarreaubaaee002006-06-26 02:48:02 +0200972
Willy Tarreaudad36a32013-03-11 01:20:04 +0100973 str2 = back = env_expand(strdup(str));
Willy Tarreaudf350f12013-03-01 20:22:54 +0100974 if (str2 == NULL) {
Thierry Fournier3e83b092023-05-23 17:58:03 +0200975 memprintf(err, "out of memory in '%s'", __FUNCTION__);
Willy Tarreaud5191e72010-02-09 20:50:45 +0100976 goto out;
Willy Tarreaudf350f12013-03-01 20:22:54 +0100977 }
Willy Tarreaubaaee002006-06-26 02:48:02 +0200978
Willy Tarreau9f69f462015-09-08 16:01:25 +0200979 if (!*str2) {
Thierry Fournier3e83b092023-05-23 17:58:03 +0200980 memprintf(err, "'%s' resolves to an empty address (environment variable missing?)", str);
Willy Tarreau9f69f462015-09-08 16:01:25 +0200981 goto out;
982 }
983
Willy Tarreau24709282013-03-10 21:32:12 +0100984 memset(&ss, 0, sizeof(ss));
985
Willy Tarreaue835bd82020-09-16 11:35:47 +0200986 /* prepare the default socket types */
Willy Tarreauf23b1bc2021-03-23 18:36:37 +0100987 if ((opts & (PA_O_STREAM|PA_O_DGRAM)) == PA_O_DGRAM ||
Willy Tarreaue3b45182021-10-27 17:28:55 +0200988 ((opts & (PA_O_STREAM|PA_O_DGRAM)) == (PA_O_DGRAM|PA_O_STREAM) && (opts & PA_O_DEFAULT_DGRAM))) {
989 proto_type = PROTO_TYPE_DGRAM;
990 ctrl_type = SOCK_DGRAM;
991 } else {
992 proto_type = PROTO_TYPE_STREAM;
993 ctrl_type = SOCK_STREAM;
994 }
Willy Tarreaue835bd82020-09-16 11:35:47 +0200995
996 if (strncmp(str2, "stream+", 7) == 0) {
997 str2 += 7;
Willy Tarreaue3b45182021-10-27 17:28:55 +0200998 proto_type = PROTO_TYPE_STREAM;
999 ctrl_type = SOCK_STREAM;
Willy Tarreaue835bd82020-09-16 11:35:47 +02001000 }
1001 else if (strncmp(str2, "dgram+", 6) == 0) {
1002 str2 += 6;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001003 proto_type = PROTO_TYPE_DGRAM;
1004 ctrl_type = SOCK_DGRAM;
Willy Tarreaue835bd82020-09-16 11:35:47 +02001005 }
Willy Tarreau40725a42023-01-16 13:55:27 +01001006 else if (strncmp(str2, "quic+", 5) == 0) {
1007 str2 += 5;
1008 proto_type = PROTO_TYPE_DGRAM;
1009 ctrl_type = SOCK_STREAM;
1010 }
Willy Tarreaue835bd82020-09-16 11:35:47 +02001011
Willy Tarreau24709282013-03-10 21:32:12 +01001012 if (strncmp(str2, "unix@", 5) == 0) {
1013 str2 += 5;
Willy Tarreauccfccef2014-05-10 01:49:15 +02001014 abstract = 0;
Willy Tarreau24709282013-03-10 21:32:12 +01001015 ss.ss_family = AF_UNIX;
1016 }
Emeric Brunce325c42021-04-02 17:05:09 +02001017 else if (strncmp(str2, "uxdg@", 5) == 0) {
1018 str2 += 5;
1019 abstract = 0;
1020 ss.ss_family = AF_UNIX;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001021 proto_type = PROTO_TYPE_DGRAM;
1022 ctrl_type = SOCK_DGRAM;
Emeric Brunce325c42021-04-02 17:05:09 +02001023 }
1024 else if (strncmp(str2, "uxst@", 5) == 0) {
1025 str2 += 5;
1026 abstract = 0;
1027 ss.ss_family = AF_UNIX;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001028 proto_type = PROTO_TYPE_STREAM;
1029 ctrl_type = SOCK_STREAM;
Emeric Brunce325c42021-04-02 17:05:09 +02001030 }
Willy Tarreauccfccef2014-05-10 01:49:15 +02001031 else if (strncmp(str2, "abns@", 5) == 0) {
1032 str2 += 5;
1033 abstract = 1;
1034 ss.ss_family = AF_UNIX;
1035 }
Emeric Brunce325c42021-04-02 17:05:09 +02001036 else if (strncmp(str2, "ip@", 3) == 0) {
1037 str2 += 3;
1038 ss.ss_family = AF_UNSPEC;
1039 }
Willy Tarreau24709282013-03-10 21:32:12 +01001040 else if (strncmp(str2, "ipv4@", 5) == 0) {
1041 str2 += 5;
1042 ss.ss_family = AF_INET;
1043 }
1044 else if (strncmp(str2, "ipv6@", 5) == 0) {
1045 str2 += 5;
1046 ss.ss_family = AF_INET6;
1047 }
Emeric Brunce325c42021-04-02 17:05:09 +02001048 else if (strncmp(str2, "tcp4@", 5) == 0) {
1049 str2 += 5;
1050 ss.ss_family = AF_INET;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001051 proto_type = PROTO_TYPE_STREAM;
1052 ctrl_type = SOCK_STREAM;
Emeric Brunce325c42021-04-02 17:05:09 +02001053 }
Emeric Brun3835c0d2020-07-07 09:46:09 +02001054 else if (strncmp(str2, "udp4@", 5) == 0) {
1055 str2 += 5;
1056 ss.ss_family = AF_INET;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001057 proto_type = PROTO_TYPE_DGRAM;
1058 ctrl_type = SOCK_DGRAM;
Emeric Brun3835c0d2020-07-07 09:46:09 +02001059 }
Emeric Brunce325c42021-04-02 17:05:09 +02001060 else if (strncmp(str2, "tcp6@", 5) == 0) {
1061 str2 += 5;
1062 ss.ss_family = AF_INET6;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001063 proto_type = PROTO_TYPE_STREAM;
1064 ctrl_type = SOCK_STREAM;
Emeric Brunce325c42021-04-02 17:05:09 +02001065 }
Emeric Brun3835c0d2020-07-07 09:46:09 +02001066 else if (strncmp(str2, "udp6@", 5) == 0) {
1067 str2 += 5;
1068 ss.ss_family = AF_INET6;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001069 proto_type = PROTO_TYPE_DGRAM;
1070 ctrl_type = SOCK_DGRAM;
Emeric Brun3835c0d2020-07-07 09:46:09 +02001071 }
Emeric Brunce325c42021-04-02 17:05:09 +02001072 else if (strncmp(str2, "tcp@", 4) == 0) {
1073 str2 += 4;
1074 ss.ss_family = AF_UNSPEC;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001075 proto_type = PROTO_TYPE_STREAM;
1076 ctrl_type = SOCK_STREAM;
Emeric Brunce325c42021-04-02 17:05:09 +02001077 }
Emeric Brun3835c0d2020-07-07 09:46:09 +02001078 else if (strncmp(str2, "udp@", 4) == 0) {
1079 str2 += 4;
1080 ss.ss_family = AF_UNSPEC;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001081 proto_type = PROTO_TYPE_DGRAM;
1082 ctrl_type = SOCK_DGRAM;
Emeric Brun3835c0d2020-07-07 09:46:09 +02001083 }
Frédéric Lécaille10caf652020-11-23 11:36:57 +01001084 else if (strncmp(str2, "quic4@", 6) == 0) {
1085 str2 += 6;
1086 ss.ss_family = AF_INET;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001087 proto_type = PROTO_TYPE_DGRAM;
Frédéric Lécaille10caf652020-11-23 11:36:57 +01001088 ctrl_type = SOCK_STREAM;
1089 }
1090 else if (strncmp(str2, "quic6@", 6) == 0) {
1091 str2 += 6;
1092 ss.ss_family = AF_INET6;
Willy Tarreaue3b45182021-10-27 17:28:55 +02001093 proto_type = PROTO_TYPE_DGRAM;
Frédéric Lécaille10caf652020-11-23 11:36:57 +01001094 ctrl_type = SOCK_STREAM;
1095 }
Willy Tarreau5a7beed2020-09-04 16:54:05 +02001096 else if (strncmp(str2, "fd@", 3) == 0) {
1097 str2 += 3;
1098 ss.ss_family = AF_CUST_EXISTING_FD;
1099 }
1100 else if (strncmp(str2, "sockpair@", 9) == 0) {
1101 str2 += 9;
1102 ss.ss_family = AF_CUST_SOCKPAIR;
1103 }
Willy Tarreau24709282013-03-10 21:32:12 +01001104 else if (*str2 == '/') {
1105 ss.ss_family = AF_UNIX;
1106 }
1107 else
1108 ss.ss_family = AF_UNSPEC;
1109
Willy Tarreau5a7beed2020-09-04 16:54:05 +02001110 if (ss.ss_family == AF_CUST_SOCKPAIR) {
Willy Tarreaua215be22020-09-16 10:14:16 +02001111 struct sockaddr_storage ss2;
1112 socklen_t addr_len;
William Lallemand2fe7dd02018-09-11 16:51:29 +02001113 char *endptr;
1114
Willy Tarreaua93e5c72020-09-15 14:01:16 +02001115 new_fd = strtol(str2, &endptr, 10);
1116 if (!*str2 || new_fd < 0 || *endptr) {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001117 memprintf(err, "file descriptor '%s' is not a valid integer in '%s'", str2, str);
William Lallemand2fe7dd02018-09-11 16:51:29 +02001118 goto out;
1119 }
Willy Tarreaua93e5c72020-09-15 14:01:16 +02001120
Willy Tarreaua215be22020-09-16 10:14:16 +02001121 /* just verify that it's a socket */
1122 addr_len = sizeof(ss2);
1123 if (getsockname(new_fd, (struct sockaddr *)&ss2, &addr_len) == -1) {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001124 memprintf(err, "cannot use file descriptor '%d' : %s.", new_fd, strerror(errno));
Willy Tarreaua215be22020-09-16 10:14:16 +02001125 goto out;
1126 }
1127
Willy Tarreaua93e5c72020-09-15 14:01:16 +02001128 ((struct sockaddr_in *)&ss)->sin_addr.s_addr = new_fd;
1129 ((struct sockaddr_in *)&ss)->sin_port = 0;
William Lallemand2fe7dd02018-09-11 16:51:29 +02001130 }
Willy Tarreau5a7beed2020-09-04 16:54:05 +02001131 else if (ss.ss_family == AF_CUST_EXISTING_FD) {
Willy Tarreau40aa0702013-03-10 23:51:38 +01001132 char *endptr;
1133
Willy Tarreaua93e5c72020-09-15 14:01:16 +02001134 new_fd = strtol(str2, &endptr, 10);
1135 if (!*str2 || new_fd < 0 || *endptr) {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001136 memprintf(err, "file descriptor '%s' is not a valid integer in '%s'", str2, str);
Willy Tarreau40aa0702013-03-10 23:51:38 +01001137 goto out;
1138 }
Willy Tarreaua93e5c72020-09-15 14:01:16 +02001139
Willy Tarreau6edc7222020-09-15 17:41:56 +02001140 if (opts & PA_O_SOCKET_FD) {
1141 socklen_t addr_len;
1142 int type;
1143
1144 addr_len = sizeof(ss);
1145 if (getsockname(new_fd, (struct sockaddr *)&ss, &addr_len) == -1) {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001146 memprintf(err, "cannot use file descriptor '%d' : %s.", new_fd, strerror(errno));
Willy Tarreau6edc7222020-09-15 17:41:56 +02001147 goto out;
1148 }
1149
1150 addr_len = sizeof(type);
1151 if (getsockopt(new_fd, SOL_SOCKET, SO_TYPE, &type, &addr_len) != 0 ||
Willy Tarreaue3b45182021-10-27 17:28:55 +02001152 (type == SOCK_STREAM) != (proto_type == PROTO_TYPE_STREAM)) {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001153 memprintf(err, "socket on file descriptor '%d' is of the wrong type.", new_fd);
Willy Tarreau6edc7222020-09-15 17:41:56 +02001154 goto out;
1155 }
1156
1157 porta = portl = porth = get_host_port(&ss);
1158 } else if (opts & PA_O_RAW_FD) {
1159 ((struct sockaddr_in *)&ss)->sin_addr.s_addr = new_fd;
1160 ((struct sockaddr_in *)&ss)->sin_port = 0;
1161 } else {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001162 memprintf(err, "a file descriptor is not acceptable here in '%s'", str);
Willy Tarreau6edc7222020-09-15 17:41:56 +02001163 goto out;
1164 }
Willy Tarreau40aa0702013-03-10 23:51:38 +01001165 }
1166 else if (ss.ss_family == AF_UNIX) {
Willy Tarreau8daa9202019-06-16 18:16:33 +02001167 struct sockaddr_un *un = (struct sockaddr_un *)&ss;
Willy Tarreau15586382013-03-04 19:48:14 +01001168 int prefix_path_len;
1169 int max_path_len;
Willy Tarreau94ef3f32014-04-14 14:49:00 +02001170 int adr_len;
Willy Tarreau15586382013-03-04 19:48:14 +01001171
1172 /* complete unix socket path name during startup or soft-restart is
1173 * <unix_bind_prefix><path>.<pid>.<bak|tmp>
1174 */
Willy Tarreauccfccef2014-05-10 01:49:15 +02001175 prefix_path_len = (pfx && !abstract) ? strlen(pfx) : 0;
Willy Tarreau8daa9202019-06-16 18:16:33 +02001176 max_path_len = (sizeof(un->sun_path) - 1) -
Willy Tarreau327ea5a2020-02-11 06:43:37 +01001177 (abstract ? 0 : prefix_path_len + 1 + 5 + 1 + 3);
Willy Tarreau15586382013-03-04 19:48:14 +01001178
Willy Tarreau94ef3f32014-04-14 14:49:00 +02001179 adr_len = strlen(str2);
1180 if (adr_len > max_path_len) {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001181 memprintf(err, "socket path '%s' too long (max %d)", str, max_path_len);
Willy Tarreau15586382013-03-04 19:48:14 +01001182 goto out;
1183 }
1184
Willy Tarreauccfccef2014-05-10 01:49:15 +02001185 /* when abstract==1, we skip the first zero and copy all bytes except the trailing zero */
Willy Tarreau8daa9202019-06-16 18:16:33 +02001186 memset(un->sun_path, 0, sizeof(un->sun_path));
Willy Tarreau94ef3f32014-04-14 14:49:00 +02001187 if (prefix_path_len)
Willy Tarreau8daa9202019-06-16 18:16:33 +02001188 memcpy(un->sun_path, pfx, prefix_path_len);
1189 memcpy(un->sun_path + prefix_path_len + abstract, str2, adr_len + 1 - abstract);
Willy Tarreau15586382013-03-04 19:48:14 +01001190 }
Willy Tarreau24709282013-03-10 21:32:12 +01001191 else { /* IPv4 and IPv6 */
mildisff5d5102015-10-26 18:50:08 +01001192 char *end = str2 + strlen(str2);
1193 char *chr;
Willy Tarreau72b8c1f2015-09-08 15:50:19 +02001194
mildisff5d5102015-10-26 18:50:08 +01001195 /* search for : or ] whatever comes first */
1196 for (chr = end-1; chr > str2; chr--) {
1197 if (*chr == ']' || *chr == ':')
1198 break;
1199 }
1200
1201 if (*chr == ':') {
1202 /* Found a colon before a closing-bracket, must be a port separator.
1203 * This guarantee backward compatibility.
1204 */
Willy Tarreau7f96a842020-09-15 11:12:44 +02001205 if (!(opts & PA_O_PORT_OK)) {
1206 memprintf(err, "port specification not permitted here in '%s'", str);
1207 goto out;
1208 }
mildisff5d5102015-10-26 18:50:08 +01001209 *chr++ = '\0';
1210 port1 = chr;
1211 }
1212 else {
1213 /* Either no colon and no closing-bracket
1214 * or directly ending with a closing-bracket.
1215 * However, no port.
1216 */
Willy Tarreau7f96a842020-09-15 11:12:44 +02001217 if (opts & PA_O_PORT_MAND) {
1218 memprintf(err, "missing port specification in '%s'", str);
1219 goto out;
1220 }
Willy Tarreauc120c8d2013-03-10 19:27:44 +01001221 port1 = "";
mildisff5d5102015-10-26 18:50:08 +01001222 }
Willy Tarreaubaaee002006-06-26 02:48:02 +02001223
Willy Tarreau90807112020-02-25 08:16:33 +01001224 if (isdigit((unsigned char)*port1)) { /* single port or range */
Willy Tarreauc120c8d2013-03-10 19:27:44 +01001225 port2 = strchr(port1, '-');
Willy Tarreau7f96a842020-09-15 11:12:44 +02001226 if (port2) {
1227 if (!(opts & PA_O_PORT_RANGE)) {
1228 memprintf(err, "port range not permitted here in '%s'", str);
1229 goto out;
1230 }
Willy Tarreauc120c8d2013-03-10 19:27:44 +01001231 *port2++ = '\0';
Willy Tarreau7f96a842020-09-15 11:12:44 +02001232 }
Willy Tarreauc120c8d2013-03-10 19:27:44 +01001233 else
1234 port2 = port1;
1235 portl = atoi(port1);
1236 porth = atoi(port2);
Willy Tarreau7f96a842020-09-15 11:12:44 +02001237
1238 if (portl < !!(opts & PA_O_PORT_MAND) || portl > 65535) {
1239 memprintf(err, "invalid port '%s'", port1);
1240 goto out;
1241 }
1242
1243 if (porth < !!(opts & PA_O_PORT_MAND) || porth > 65535) {
1244 memprintf(err, "invalid port '%s'", port2);
1245 goto out;
1246 }
1247
1248 if (portl > porth) {
1249 memprintf(err, "invalid port range '%d-%d'", portl, porth);
1250 goto out;
1251 }
1252
Willy Tarreauc120c8d2013-03-10 19:27:44 +01001253 porta = portl;
1254 }
1255 else if (*port1 == '-') { /* negative offset */
Willy Tarreau7f96a842020-09-15 11:12:44 +02001256 if (!(opts & PA_O_PORT_OFS)) {
1257 memprintf(err, "port offset not permitted here in '%s'", str);
1258 goto out;
1259 }
Willy Tarreauc120c8d2013-03-10 19:27:44 +01001260 portl = atoi(port1 + 1);
1261 porta = -portl;
1262 }
1263 else if (*port1 == '+') { /* positive offset */
Willy Tarreau7f96a842020-09-15 11:12:44 +02001264 if (!(opts & PA_O_PORT_OFS)) {
1265 memprintf(err, "port offset not permitted here in '%s'", str);
1266 goto out;
1267 }
Willy Tarreauc120c8d2013-03-10 19:27:44 +01001268 porth = atoi(port1 + 1);
1269 porta = porth;
1270 }
1271 else if (*port1) { /* other any unexpected char */
Thierry Fournier3e83b092023-05-23 17:58:03 +02001272 memprintf(err, "invalid character '%c' in port number '%s' in '%s'", *port1, port1, str);
Willy Tarreauc120c8d2013-03-10 19:27:44 +01001273 goto out;
1274 }
Willy Tarreau7f96a842020-09-15 11:12:44 +02001275 else if (opts & PA_O_PORT_MAND) {
1276 memprintf(err, "missing port specification in '%s'", str);
1277 goto out;
1278 }
Willy Tarreauceccdd72016-11-02 22:27:10 +01001279
1280 /* first try to parse the IP without resolving. If it fails, it
1281 * tells us we need to keep a copy of the FQDN to resolve later
1282 * and to enable DNS. In this case we can proceed if <fqdn> is
Willy Tarreaucd3a55912020-09-04 15:30:46 +02001283 * set or if PA_O_RESOLVE is set, otherwise it's an error.
Willy Tarreauceccdd72016-11-02 22:27:10 +01001284 */
1285 if (str2ip2(str2, &ss, 0) == NULL) {
Willy Tarreaucd3a55912020-09-04 15:30:46 +02001286 if ((!(opts & PA_O_RESOLVE) && !fqdn) ||
1287 ((opts & PA_O_RESOLVE) && str2ip2(str2, &ss, 1) == NULL)) {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001288 memprintf(err, "invalid address: '%s' in '%s'", str2, str);
Willy Tarreauceccdd72016-11-02 22:27:10 +01001289 goto out;
1290 }
Willy Tarreau72b8c1f2015-09-08 15:50:19 +02001291
Willy Tarreauceccdd72016-11-02 22:27:10 +01001292 if (fqdn) {
1293 if (str2 != back)
1294 memmove(back, str2, strlen(str2) + 1);
1295 *fqdn = back;
1296 back = NULL;
1297 }
Willy Tarreau72b8c1f2015-09-08 15:50:19 +02001298 }
Willy Tarreauceccdd72016-11-02 22:27:10 +01001299 set_host_port(&ss, porta);
Willy Tarreaue4c58c82013-03-06 15:28:17 +01001300 }
Willy Tarreaufab5a432011-03-04 15:31:53 +01001301
Willy Tarreaue835bd82020-09-16 11:35:47 +02001302 if (ctrl_type == SOCK_STREAM && !(opts & PA_O_STREAM)) {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001303 memprintf(err, "stream-type address not acceptable in '%s'", str);
Willy Tarreaue835bd82020-09-16 11:35:47 +02001304 goto out;
1305 }
1306 else if (ctrl_type == SOCK_DGRAM && !(opts & PA_O_DGRAM)) {
Thierry Fournier3e83b092023-05-23 17:58:03 +02001307 memprintf(err, "dgram-type address not acceptable in '%s'", str);
Willy Tarreaue835bd82020-09-16 11:35:47 +02001308 goto out;
1309 }
1310
Willy Tarreau65ec4e32020-09-16 19:17:08 +02001311 if (proto || (opts & PA_O_CONNECT)) {
Willy Tarreau5fc93282020-09-16 18:25:03 +02001312 /* Note: if the caller asks for a proto, we must find one,
Emeric Brun26754902021-04-07 14:26:44 +02001313 * except if we inherit from a raw FD (family == AF_CUST_EXISTING_FD)
1314 * orif we return with an fqdn that will resolve later,
Willy Tarreau5fc93282020-09-16 18:25:03 +02001315 * in which case the address is not known yet (this is only
1316 * for servers actually).
1317 */
Willy Tarreaub2ffc992020-09-16 21:37:31 +02001318 new_proto = protocol_lookup(ss.ss_family,
Willy Tarreaue3b45182021-10-27 17:28:55 +02001319 proto_type,
Willy Tarreauaf9609b2020-09-16 22:04:46 +02001320 ctrl_type == SOCK_DGRAM);
Willy Tarreaub2ffc992020-09-16 21:37:31 +02001321
Emeric Brun26754902021-04-07 14:26:44 +02001322 if (!new_proto && (!fqdn || !*fqdn) && (ss.ss_family != AF_CUST_EXISTING_FD)) {
Willy Tarreau3d7b4682022-05-20 17:32:35 +02001323 memprintf(err, "unsupported %s protocol for %s family %d address '%s'%s",
Willy Tarreau2b049b82022-05-20 17:28:30 +02001324 (ctrl_type == SOCK_DGRAM) ? "datagram" : "stream",
1325 (proto_type == PROTO_TYPE_DGRAM) ? "datagram" : "stream",
1326 ss.ss_family,
Willy Tarreau3d7b4682022-05-20 17:32:35 +02001327 str,
Willy Tarreau3d7b4682022-05-20 17:32:35 +02001328#ifndef USE_QUIC
Tim Duesterhus147eeb22022-05-22 12:40:58 +02001329 (ctrl_type == SOCK_STREAM && proto_type == PROTO_TYPE_DGRAM)
1330 ? "; QUIC is not compiled in if this is what you were looking for."
1331 : ""
Willy Tarreau3d7b4682022-05-20 17:32:35 +02001332#else
1333 ""
1334#endif
Tim Duesterhus147eeb22022-05-22 12:40:58 +02001335 );
Willy Tarreau5fc93282020-09-16 18:25:03 +02001336 goto out;
1337 }
Willy Tarreau65ec4e32020-09-16 19:17:08 +02001338
1339 if ((opts & PA_O_CONNECT) && new_proto && !new_proto->connect) {
1340 memprintf(err, "connect() not supported for this protocol family %d used by address '%s'", ss.ss_family, str);
1341 goto out;
1342 }
Willy Tarreau5fc93282020-09-16 18:25:03 +02001343 }
1344
Willy Tarreauc120c8d2013-03-10 19:27:44 +01001345 ret = &ss;
Willy Tarreaud5191e72010-02-09 20:50:45 +01001346 out:
Willy Tarreau48ef4c92017-01-06 18:32:38 +01001347 if (port)
1348 *port = porta;
Willy Tarreaud4448bc2013-02-20 15:55:15 +01001349 if (low)
1350 *low = portl;
1351 if (high)
1352 *high = porth;
Willy Tarreaua93e5c72020-09-15 14:01:16 +02001353 if (fd)
1354 *fd = new_fd;
Willy Tarreau5fc93282020-09-16 18:25:03 +02001355 if (proto)
1356 *proto = new_proto;
Willy Tarreau24709282013-03-10 21:32:12 +01001357 free(back);
Willy Tarreaud5191e72010-02-09 20:50:45 +01001358 return ret;
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02001359}
1360
Thayne McCombs92149f92020-11-20 01:28:26 -07001361/* converts <addr> and <port> into a string representation of the address and port. This is sort
1362 * of an inverse of str2sa_range, with some restrictions. The supported families are AF_INET,
1363 * AF_INET6, AF_UNIX, and AF_CUST_SOCKPAIR. If the family is unsopported NULL is returned.
1364 * If map_ports is true, then the sign of the port is included in the output, to indicate it is
1365 * relative to the incoming port. AF_INET and AF_INET6 will be in the form "<addr>:<port>".
1366 * AF_UNIX will either be just the path (if using a pathname) or "abns@<path>" if it is abstract.
1367 * AF_CUST_SOCKPAIR will be of the form "sockpair@<fd>".
1368 *
1369 * The returned char* is allocated, and it is the responsibility of the caller to free it.
1370 */
1371char * sa2str(const struct sockaddr_storage *addr, int port, int map_ports)
1372{
1373 char buffer[INET6_ADDRSTRLEN];
1374 char *out = NULL;
1375 const void *ptr;
1376 const char *path;
1377
1378 switch (addr->ss_family) {
1379 case AF_INET:
1380 ptr = &((struct sockaddr_in *)addr)->sin_addr;
1381 break;
1382 case AF_INET6:
1383 ptr = &((struct sockaddr_in6 *)addr)->sin6_addr;
1384 break;
1385 case AF_UNIX:
1386 path = ((struct sockaddr_un *)addr)->sun_path;
1387 if (path[0] == '\0') {
1388 const int max_length = sizeof(struct sockaddr_un) - offsetof(struct sockaddr_un, sun_path) - 1;
1389 return memprintf(&out, "abns@%.*s", max_length, path+1);
1390 } else {
1391 return strdup(path);
1392 }
1393 case AF_CUST_SOCKPAIR:
1394 return memprintf(&out, "sockpair@%d", ((struct sockaddr_in *)addr)->sin_addr.s_addr);
1395 default:
1396 return NULL;
1397 }
Tim Duesterhus22535a52022-05-23 09:30:49 +02001398 if (inet_ntop(addr->ss_family, ptr, buffer, sizeof(buffer)) == NULL) {
1399 BUG_ON(errno == ENOSPC);
1400 return NULL;
1401 }
Thayne McCombs92149f92020-11-20 01:28:26 -07001402 if (map_ports)
1403 return memprintf(&out, "%s:%+d", buffer, port);
1404 else
1405 return memprintf(&out, "%s:%d", buffer, port);
1406}
1407
1408
Willy Tarreau2937c0d2010-01-26 17:36:17 +01001409/* converts <str> to a struct in_addr containing a network mask. It can be
1410 * passed in dotted form (255.255.255.0) or in CIDR form (24). It returns 1
Jarno Huuskonen577d5ac2017-05-21 17:32:21 +03001411 * if the conversion succeeds otherwise zero.
Willy Tarreau2937c0d2010-01-26 17:36:17 +01001412 */
1413int str2mask(const char *str, struct in_addr *mask)
1414{
1415 if (strchr(str, '.') != NULL) { /* dotted notation */
1416 if (!inet_pton(AF_INET, str, mask))
1417 return 0;
1418 }
1419 else { /* mask length */
1420 char *err;
1421 unsigned long len = strtol(str, &err, 10);
1422
1423 if (!*str || (err && *err) || (unsigned)len > 32)
1424 return 0;
Tim Duesterhus8575f722018-01-25 16:24:48 +01001425
1426 len2mask4(len, mask);
Willy Tarreau2937c0d2010-01-26 17:36:17 +01001427 }
1428 return 1;
1429}
1430
Tim Duesterhus47185172018-01-25 16:24:49 +01001431/* converts <str> to a struct in6_addr containing a network mask. It can be
Tim Duesterhus5e642862018-02-20 17:02:18 +01001432 * passed in quadruplet form (ffff:ffff::) or in CIDR form (64). It returns 1
Tim Duesterhus47185172018-01-25 16:24:49 +01001433 * if the conversion succeeds otherwise zero.
1434 */
1435int str2mask6(const char *str, struct in6_addr *mask)
1436{
1437 if (strchr(str, ':') != NULL) { /* quadruplet notation */
1438 if (!inet_pton(AF_INET6, str, mask))
1439 return 0;
1440 }
1441 else { /* mask length */
1442 char *err;
1443 unsigned long len = strtol(str, &err, 10);
1444
1445 if (!*str || (err && *err) || (unsigned)len > 128)
1446 return 0;
1447
1448 len2mask6(len, mask);
1449 }
1450 return 1;
1451}
1452
Thierry FOURNIERb0504632013-12-14 15:39:02 +01001453/* convert <cidr> to struct in_addr <mask>. It returns 1 if the conversion
1454 * succeeds otherwise zero.
1455 */
1456int cidr2dotted(int cidr, struct in_addr *mask) {
1457
1458 if (cidr < 0 || cidr > 32)
1459 return 0;
1460
1461 mask->s_addr = cidr ? htonl(~0UL << (32 - cidr)) : 0;
1462 return 1;
1463}
1464
Thierry Fournier70473a52016-02-17 17:12:14 +01001465/* Convert mask from bit length form to in_addr form.
1466 * This function never fails.
1467 */
1468void len2mask4(int len, struct in_addr *addr)
1469{
1470 if (len >= 32) {
1471 addr->s_addr = 0xffffffff;
1472 return;
1473 }
1474 if (len <= 0) {
1475 addr->s_addr = 0x00000000;
1476 return;
1477 }
1478 addr->s_addr = 0xffffffff << (32 - len);
1479 addr->s_addr = htonl(addr->s_addr);
1480}
1481
1482/* Convert mask from bit length form to in6_addr form.
1483 * This function never fails.
1484 */
1485void len2mask6(int len, struct in6_addr *addr)
1486{
1487 len2mask4(len, (struct in_addr *)&addr->s6_addr[0]); /* msb */
1488 len -= 32;
1489 len2mask4(len, (struct in_addr *)&addr->s6_addr[4]);
1490 len -= 32;
1491 len2mask4(len, (struct in_addr *)&addr->s6_addr[8]);
1492 len -= 32;
1493 len2mask4(len, (struct in_addr *)&addr->s6_addr[12]); /* lsb */
1494}
1495
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02001496/*
Willy Tarreaud077a8e2007-05-08 18:28:09 +02001497 * converts <str> to two struct in_addr* which must be pre-allocated.
Willy Tarreaubaaee002006-06-26 02:48:02 +02001498 * The format is "addr[/mask]", where "addr" cannot be empty, and mask
Ilya Shipitsin856aabc2020-04-16 23:51:34 +05001499 * is optional and either in the dotted or CIDR notation.
Willy Tarreaubaaee002006-06-26 02:48:02 +02001500 * Note: "addr" can also be a hostname. Returns 1 if OK, 0 if error.
1501 */
Thierry FOURNIERfc7ac7b2014-02-11 15:23:04 +01001502int str2net(const char *str, int resolve, struct in_addr *addr, struct in_addr *mask)
Willy Tarreaubaaee002006-06-26 02:48:02 +02001503{
Willy Tarreau8aeae4a2007-06-17 11:42:08 +02001504 __label__ out_free, out_err;
1505 char *c, *s;
1506 int ret_val;
Willy Tarreaubaaee002006-06-26 02:48:02 +02001507
Willy Tarreau8aeae4a2007-06-17 11:42:08 +02001508 s = strdup(str);
1509 if (!s)
1510 return 0;
1511
Willy Tarreaubaaee002006-06-26 02:48:02 +02001512 memset(mask, 0, sizeof(*mask));
1513 memset(addr, 0, sizeof(*addr));
Willy Tarreaubaaee002006-06-26 02:48:02 +02001514
Willy Tarreau8aeae4a2007-06-17 11:42:08 +02001515 if ((c = strrchr(s, '/')) != NULL) {
Willy Tarreaubaaee002006-06-26 02:48:02 +02001516 *c++ = '\0';
1517 /* c points to the mask */
Willy Tarreau2937c0d2010-01-26 17:36:17 +01001518 if (!str2mask(c, mask))
1519 goto out_err;
Willy Tarreaubaaee002006-06-26 02:48:02 +02001520 }
1521 else {
Willy Tarreauebd61602006-12-30 11:54:15 +01001522 mask->s_addr = ~0U;
Willy Tarreaubaaee002006-06-26 02:48:02 +02001523 }
Willy Tarreau8aeae4a2007-06-17 11:42:08 +02001524 if (!inet_pton(AF_INET, s, addr)) {
Willy Tarreaubaaee002006-06-26 02:48:02 +02001525 struct hostent *he;
1526
Thierry FOURNIERfc7ac7b2014-02-11 15:23:04 +01001527 if (!resolve)
1528 goto out_err;
1529
Willy Tarreau8aeae4a2007-06-17 11:42:08 +02001530 if ((he = gethostbyname(s)) == NULL) {
1531 goto out_err;
Willy Tarreaubaaee002006-06-26 02:48:02 +02001532 }
1533 else
1534 *addr = *(struct in_addr *) *(he->h_addr_list);
1535 }
Willy Tarreau8aeae4a2007-06-17 11:42:08 +02001536
1537 ret_val = 1;
1538 out_free:
1539 free(s);
1540 return ret_val;
1541 out_err:
1542 ret_val = 0;
1543 goto out_free;
Willy Tarreaubaaee002006-06-26 02:48:02 +02001544}
1545
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001546
1547/*
Willy Tarreau6d20e282012-04-27 22:49:47 +02001548 * converts <str> to two struct in6_addr* which must be pre-allocated.
1549 * The format is "addr[/mask]", where "addr" cannot be empty, and mask
Ilya Shipitsin856aabc2020-04-16 23:51:34 +05001550 * is an optional number of bits (128 being the default).
Willy Tarreau6d20e282012-04-27 22:49:47 +02001551 * Returns 1 if OK, 0 if error.
1552 */
1553int str62net(const char *str, struct in6_addr *addr, unsigned char *mask)
1554{
1555 char *c, *s;
1556 int ret_val = 0;
1557 char *err;
1558 unsigned long len = 128;
1559
1560 s = strdup(str);
1561 if (!s)
1562 return 0;
1563
1564 memset(mask, 0, sizeof(*mask));
1565 memset(addr, 0, sizeof(*addr));
1566
1567 if ((c = strrchr(s, '/')) != NULL) {
1568 *c++ = '\0'; /* c points to the mask */
1569 if (!*c)
1570 goto out_free;
1571
1572 len = strtoul(c, &err, 10);
1573 if ((err && *err) || (unsigned)len > 128)
1574 goto out_free;
1575 }
1576 *mask = len; /* OK we have a valid mask in <len> */
1577
1578 if (!inet_pton(AF_INET6, s, addr))
1579 goto out_free;
1580
1581 ret_val = 1;
1582 out_free:
1583 free(s);
1584 return ret_val;
1585}
1586
1587
1588/*
Willy Tarreau12e10272021-03-25 11:34:40 +01001589 * Parse IPv4 address found in url. Return the number of bytes parsed. It
1590 * expects exactly 4 numbers between 0 and 255 delimited by dots, and returns
1591 * zero in case of mismatch.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001592 */
David du Colombier6f5ccb12011-03-10 22:26:24 +01001593int url2ipv4(const char *addr, struct in_addr *dst)
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001594{
1595 int saw_digit, octets, ch;
1596 u_char tmp[4], *tp;
1597 const char *cp = addr;
1598
1599 saw_digit = 0;
1600 octets = 0;
1601 *(tp = tmp) = 0;
1602
1603 while (*addr) {
Willy Tarreau12e10272021-03-25 11:34:40 +01001604 unsigned char digit = (ch = *addr) - '0';
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001605 if (digit > 9 && ch != '.')
1606 break;
Willy Tarreau12e10272021-03-25 11:34:40 +01001607 addr++;
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001608 if (digit <= 9) {
1609 u_int new = *tp * 10 + digit;
1610 if (new > 255)
1611 return 0;
1612 *tp = new;
1613 if (!saw_digit) {
1614 if (++octets > 4)
1615 return 0;
1616 saw_digit = 1;
1617 }
1618 } else if (ch == '.' && saw_digit) {
1619 if (octets == 4)
1620 return 0;
1621 *++tp = 0;
1622 saw_digit = 0;
1623 } else
1624 return 0;
1625 }
1626
1627 if (octets < 4)
1628 return 0;
1629
1630 memcpy(&dst->s_addr, tmp, 4);
Willy Tarreau12e10272021-03-25 11:34:40 +01001631 return addr - cp;
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001632}
1633
1634/*
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001635 * Resolve destination server from URL. Convert <str> to a sockaddr_storage.
Ilya Shipitsin856aabc2020-04-16 23:51:34 +05001636 * <out> contain the code of the detected scheme, the start and length of
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001637 * the hostname. Actually only http and https are supported. <out> can be NULL.
1638 * This function returns the consumed length. It is useful if you parse complete
1639 * url like http://host:port/path, because the consumed length corresponds to
1640 * the first character of the path. If the conversion fails, it returns -1.
1641 *
1642 * This function tries to resolve the DNS name if haproxy is in starting mode.
1643 * So, this function may be used during the configuration parsing.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001644 */
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001645int url2sa(const char *url, int ulen, struct sockaddr_storage *addr, struct split_url *out)
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001646{
1647 const char *curr = url, *cp = url;
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001648 const char *end;
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001649 int ret, url_code = 0;
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001650 unsigned long long int http_code = 0;
1651 int default_port;
1652 struct hostent *he;
1653 char *p;
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001654
1655 /* Firstly, try to find :// pattern */
1656 while (curr < url+ulen && url_code != 0x3a2f2f) {
1657 url_code = ((url_code & 0xffff) << 8);
1658 url_code += (unsigned char)*curr++;
1659 }
1660
1661 /* Secondly, if :// pattern is found, verify parsed stuff
1662 * before pattern is matching our http pattern.
1663 * If so parse ip address and port in uri.
1664 *
1665 * WARNING: Current code doesn't support dynamic async dns resolver.
1666 */
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001667 if (url_code != 0x3a2f2f)
1668 return -1;
1669
1670 /* Copy scheme, and utrn to lower case. */
1671 while (cp < curr - 3)
1672 http_code = (http_code << 8) + *cp++;
1673 http_code |= 0x2020202020202020ULL; /* Turn everything to lower case */
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001674
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001675 /* HTTP or HTTPS url matching */
1676 if (http_code == 0x2020202068747470ULL) {
1677 default_port = 80;
1678 if (out)
1679 out->scheme = SCH_HTTP;
1680 }
1681 else if (http_code == 0x2020206874747073ULL) {
1682 default_port = 443;
1683 if (out)
1684 out->scheme = SCH_HTTPS;
1685 }
1686 else
1687 return -1;
1688
1689 /* If the next char is '[', the host address is IPv6. */
1690 if (*curr == '[') {
1691 curr++;
1692
1693 /* Check trash size */
1694 if (trash.size < ulen)
1695 return -1;
1696
1697 /* Look for ']' and copy the address in a trash buffer. */
Willy Tarreau843b7cb2018-07-13 10:54:26 +02001698 p = trash.area;
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001699 for (end = curr;
1700 end < url + ulen && *end != ']';
1701 end++, p++)
1702 *p = *end;
1703 if (*end != ']')
1704 return -1;
1705 *p = '\0';
1706
1707 /* Update out. */
1708 if (out) {
1709 out->host = curr;
1710 out->host_len = end - curr;
1711 }
1712
1713 /* Try IPv6 decoding. */
Willy Tarreau843b7cb2018-07-13 10:54:26 +02001714 if (!inet_pton(AF_INET6, trash.area, &((struct sockaddr_in6 *)addr)->sin6_addr))
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001715 return -1;
1716 end++;
1717
1718 /* Decode port. */
William Lallemand3d7a9182022-03-25 17:37:51 +01001719 if (end < url + ulen && *end == ':') {
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001720 end++;
1721 default_port = read_uint(&end, url + ulen);
1722 }
1723 ((struct sockaddr_in6 *)addr)->sin6_port = htons(default_port);
1724 ((struct sockaddr_in6 *)addr)->sin6_family = AF_INET6;
1725 return end - url;
1726 }
1727 else {
William Lallemand8a913742022-02-18 16:13:12 +01001728 /* we need to copy the string into the trash because url2ipv4
1729 * needs a \0 at the end of the string */
1730 if (trash.size < ulen)
1731 return -1;
1732
1733 memcpy(trash.area, curr, ulen - (curr - url));
1734 trash.area[ulen - (curr - url)] = '\0';
1735
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001736 /* We are looking for IP address. If you want to parse and
1737 * resolve hostname found in url, you can use str2sa_range(), but
1738 * be warned this can slow down global daemon performances
1739 * while handling lagging dns responses.
1740 */
William Lallemand8a913742022-02-18 16:13:12 +01001741 ret = url2ipv4(trash.area, &((struct sockaddr_in *)addr)->sin_addr);
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001742 if (ret) {
1743 /* Update out. */
1744 if (out) {
1745 out->host = curr;
1746 out->host_len = ret;
1747 }
1748
William Lallemandb938b772022-03-24 21:59:03 +01001749 curr += ret;
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001750
1751 /* Decode port. */
William Lallemand3d7a9182022-03-25 17:37:51 +01001752 if (curr < url + ulen && *curr == ':') {
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001753 curr++;
1754 default_port = read_uint(&curr, url + ulen);
1755 }
1756 ((struct sockaddr_in *)addr)->sin_port = htons(default_port);
1757
1758 /* Set family. */
1759 ((struct sockaddr_in *)addr)->sin_family = AF_INET;
1760 return curr - url;
1761 }
1762 else if (global.mode & MODE_STARTING) {
1763 /* The IPv4 and IPv6 decoding fails, maybe the url contain name. Try to execute
1764 * synchronous DNS request only if HAProxy is in the start state.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001765 */
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001766
1767 /* look for : or / or end */
1768 for (end = curr;
1769 end < url + ulen && *end != '/' && *end != ':';
1770 end++);
Willy Tarreau843b7cb2018-07-13 10:54:26 +02001771 memcpy(trash.area, curr, end - curr);
1772 trash.area[end - curr] = '\0';
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001773
1774 /* try to resolve an IPv4/IPv6 hostname */
Willy Tarreau843b7cb2018-07-13 10:54:26 +02001775 he = gethostbyname(trash.area);
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001776 if (!he)
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001777 return -1;
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001778
1779 /* Update out. */
1780 if (out) {
1781 out->host = curr;
1782 out->host_len = end - curr;
1783 }
1784
1785 /* Decode port. */
William Lallemand3d7a9182022-03-25 17:37:51 +01001786 if (end < url + ulen && *end == ':') {
Thierry FOURNIER9f95e402014-03-21 14:51:46 +01001787 end++;
1788 default_port = read_uint(&end, url + ulen);
1789 }
1790
1791 /* Copy IP address, set port and family. */
1792 switch (he->h_addrtype) {
1793 case AF_INET:
1794 ((struct sockaddr_in *)addr)->sin_addr = *(struct in_addr *) *(he->h_addr_list);
1795 ((struct sockaddr_in *)addr)->sin_port = htons(default_port);
1796 ((struct sockaddr_in *)addr)->sin_family = AF_INET;
1797 return end - url;
1798
1799 case AF_INET6:
1800 ((struct sockaddr_in6 *)addr)->sin6_addr = *(struct in6_addr *) *(he->h_addr_list);
1801 ((struct sockaddr_in6 *)addr)->sin6_port = htons(default_port);
1802 ((struct sockaddr_in6 *)addr)->sin6_family = AF_INET6;
1803 return end - url;
1804 }
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001805 }
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001806 }
Alexandre Cassen5eb1a902007-11-29 15:43:32 +01001807 return -1;
1808}
1809
Willy Tarreau631f01c2011-09-05 00:36:48 +02001810/* Tries to convert a sockaddr_storage address to text form. Upon success, the
1811 * address family is returned so that it's easy for the caller to adapt to the
1812 * output format. Zero is returned if the address family is not supported. -1
1813 * is returned upon error, with errno set. AF_INET, AF_INET6 and AF_UNIX are
1814 * supported.
1815 */
Willy Tarreaud5ec4bf2019-04-25 17:48:16 +02001816int addr_to_str(const struct sockaddr_storage *addr, char *str, int size)
Willy Tarreau631f01c2011-09-05 00:36:48 +02001817{
1818
Willy Tarreaud5ec4bf2019-04-25 17:48:16 +02001819 const void *ptr;
Willy Tarreau631f01c2011-09-05 00:36:48 +02001820
1821 if (size < 5)
1822 return 0;
1823 *str = '\0';
1824
1825 switch (addr->ss_family) {
1826 case AF_INET:
1827 ptr = &((struct sockaddr_in *)addr)->sin_addr;
1828 break;
1829 case AF_INET6:
1830 ptr = &((struct sockaddr_in6 *)addr)->sin6_addr;
1831 break;
1832 case AF_UNIX:
1833 memcpy(str, "unix", 5);
1834 return addr->ss_family;
1835 default:
1836 return 0;
1837 }
1838
1839 if (inet_ntop(addr->ss_family, ptr, str, size))
1840 return addr->ss_family;
1841
1842 /* failed */
1843 return -1;
1844}
1845
Simon Horman75ab8bd2014-06-16 09:39:41 +09001846/* Tries to convert a sockaddr_storage port to text form. Upon success, the
1847 * address family is returned so that it's easy for the caller to adapt to the
1848 * output format. Zero is returned if the address family is not supported. -1
1849 * is returned upon error, with errno set. AF_INET, AF_INET6 and AF_UNIX are
1850 * supported.
1851 */
Willy Tarreaud5ec4bf2019-04-25 17:48:16 +02001852int port_to_str(const struct sockaddr_storage *addr, char *str, int size)
Simon Horman75ab8bd2014-06-16 09:39:41 +09001853{
1854
1855 uint16_t port;
1856
1857
Willy Tarreaud7dad1b2017-01-06 16:46:22 +01001858 if (size < 6)
Simon Horman75ab8bd2014-06-16 09:39:41 +09001859 return 0;
1860 *str = '\0';
1861
1862 switch (addr->ss_family) {
1863 case AF_INET:
1864 port = ((struct sockaddr_in *)addr)->sin_port;
1865 break;
1866 case AF_INET6:
1867 port = ((struct sockaddr_in6 *)addr)->sin6_port;
1868 break;
1869 case AF_UNIX:
1870 memcpy(str, "unix", 5);
1871 return addr->ss_family;
1872 default:
1873 return 0;
1874 }
1875
1876 snprintf(str, size, "%u", ntohs(port));
1877 return addr->ss_family;
1878}
1879
Willy Tarreau16e01562016-08-09 16:46:18 +02001880/* check if the given address is local to the system or not. It will return
1881 * -1 when it's not possible to know, 0 when the address is not local, 1 when
1882 * it is. We don't want to iterate over all interfaces for this (and it is not
1883 * portable). So instead we try to bind in UDP to this address on a free non
1884 * privileged port and to connect to the same address, port 0 (connect doesn't
1885 * care). If it succeeds, we own the address. Note that non-inet addresses are
1886 * considered local since they're most likely AF_UNIX.
1887 */
1888int addr_is_local(const struct netns_entry *ns,
1889 const struct sockaddr_storage *orig)
1890{
1891 struct sockaddr_storage addr;
1892 int result;
1893 int fd;
1894
1895 if (!is_inet_addr(orig))
1896 return 1;
1897
1898 memcpy(&addr, orig, sizeof(addr));
1899 set_host_port(&addr, 0);
1900
1901 fd = my_socketat(ns, addr.ss_family, SOCK_DGRAM, IPPROTO_UDP);
1902 if (fd < 0)
1903 return -1;
1904
1905 result = -1;
1906 if (bind(fd, (struct sockaddr *)&addr, get_addr_len(&addr)) == 0) {
1907 if (connect(fd, (struct sockaddr *)&addr, get_addr_len(&addr)) == -1)
1908 result = 0; // fail, non-local address
1909 else
1910 result = 1; // success, local address
1911 }
1912 else {
1913 if (errno == EADDRNOTAVAIL)
1914 result = 0; // definitely not local :-)
1915 }
1916 close(fd);
1917
1918 return result;
1919}
1920
Willy Tarreaubaaee002006-06-26 02:48:02 +02001921/* will try to encode the string <string> replacing all characters tagged in
1922 * <map> with the hexadecimal representation of their ASCII-code (2 digits)
1923 * prefixed by <escape>, and will store the result between <start> (included)
1924 * and <stop> (excluded), and will always terminate the string with a '\0'
Aurelien DARRAGONebcab992024-04-09 11:44:54 +02001925 * before <stop>. If bytes are missing between <start> and <stop>, then the
1926 * conversion will be incomplete and truncated.
Willy Tarreaubaaee002006-06-26 02:48:02 +02001927 * The input string must also be zero-terminated.
Aurelien DARRAGONebcab992024-04-09 11:44:54 +02001928 *
1929 * Return the address of the \0 character, or NULL on error
Willy Tarreaubaaee002006-06-26 02:48:02 +02001930 */
1931const char hextab[16] = "0123456789ABCDEF";
1932char *encode_string(char *start, char *stop,
Willy Tarreau1bfd6022019-06-07 11:10:07 +02001933 const char escape, const long *map,
Willy Tarreaubaaee002006-06-26 02:48:02 +02001934 const char *string)
1935{
1936 if (start < stop) {
1937 stop--; /* reserve one byte for the final '\0' */
1938 while (start < stop && *string != '\0') {
Willy Tarreau1bfd6022019-06-07 11:10:07 +02001939 if (!ha_bit_test((unsigned char)(*string), map))
Willy Tarreaubaaee002006-06-26 02:48:02 +02001940 *start++ = *string;
1941 else {
1942 if (start + 3 >= stop)
1943 break;
1944 *start++ = escape;
1945 *start++ = hextab[(*string >> 4) & 15];
1946 *start++ = hextab[*string & 15];
1947 }
1948 string++;
1949 }
1950 *start = '\0';
Aurelien DARRAGONebcab992024-04-09 11:44:54 +02001951 return start;
Willy Tarreaubaaee002006-06-26 02:48:02 +02001952 }
Aurelien DARRAGONebcab992024-04-09 11:44:54 +02001953 return NULL;
Willy Tarreaubaaee002006-06-26 02:48:02 +02001954}
1955
Thierry FOURNIERe059ec92014-03-17 12:01:13 +01001956/*
1957 * Same behavior as encode_string() above, except that it encodes chunk
1958 * <chunk> instead of a string.
1959 */
1960char *encode_chunk(char *start, char *stop,
Willy Tarreau1bfd6022019-06-07 11:10:07 +02001961 const char escape, const long *map,
Willy Tarreau83061a82018-07-13 11:56:34 +02001962 const struct buffer *chunk)
Thierry FOURNIERe059ec92014-03-17 12:01:13 +01001963{
Willy Tarreau843b7cb2018-07-13 10:54:26 +02001964 char *str = chunk->area;
1965 char *end = chunk->area + chunk->data;
Thierry FOURNIERe059ec92014-03-17 12:01:13 +01001966
1967 if (start < stop) {
1968 stop--; /* reserve one byte for the final '\0' */
1969 while (start < stop && str < end) {
Willy Tarreau1bfd6022019-06-07 11:10:07 +02001970 if (!ha_bit_test((unsigned char)(*str), map))
Thierry FOURNIERe059ec92014-03-17 12:01:13 +01001971 *start++ = *str;
1972 else {
1973 if (start + 3 >= stop)
1974 break;
1975 *start++ = escape;
1976 *start++ = hextab[(*str >> 4) & 15];
1977 *start++ = hextab[*str & 15];
1978 }
1979 str++;
1980 }
1981 *start = '\0';
Aurelien DARRAGONebcab992024-04-09 11:44:54 +02001982 return start;
Thierry FOURNIERe059ec92014-03-17 12:01:13 +01001983 }
Aurelien DARRAGONebcab992024-04-09 11:44:54 +02001984 return NULL;
Thierry FOURNIERe059ec92014-03-17 12:01:13 +01001985}
1986
Dragan Dosen0edd1092016-02-12 13:23:02 +01001987/*
1988 * Tries to prefix characters tagged in the <map> with the <escape>
Aurelien DARRAGONc5bff8e2022-09-20 14:33:18 +02001989 * character. The input <string> is processed until string_stop
1990 * is reached or NULL-byte is encountered. The result will
Dragan Dosen1a5d0602016-07-22 16:00:31 +02001991 * be stored between <start> (included) and <stop> (excluded). This
1992 * function will always try to terminate the resulting string with a '\0'
Aurelien DARRAGONebcab992024-04-09 11:44:54 +02001993 * before <stop>.
1994 *
1995 * Return the address of the \0 character, or NULL on error
Dragan Dosen1a5d0602016-07-22 16:00:31 +02001996 */
1997char *escape_string(char *start, char *stop,
Willy Tarreau1bfd6022019-06-07 11:10:07 +02001998 const char escape, const long *map,
Aurelien DARRAGONc5bff8e2022-09-20 14:33:18 +02001999 const char *string, const char *string_stop)
Dragan Dosen1a5d0602016-07-22 16:00:31 +02002000{
2001 if (start < stop) {
2002 stop--; /* reserve one byte for the final '\0' */
Aurelien DARRAGONc5bff8e2022-09-20 14:33:18 +02002003 while (start < stop && string < string_stop && *string != '\0') {
Willy Tarreau1bfd6022019-06-07 11:10:07 +02002004 if (!ha_bit_test((unsigned char)(*string), map))
Dragan Dosen1a5d0602016-07-22 16:00:31 +02002005 *start++ = *string;
2006 else {
2007 if (start + 2 >= stop)
2008 break;
2009 *start++ = escape;
2010 *start++ = *string;
2011 }
2012 string++;
2013 }
2014 *start = '\0';
Aurelien DARRAGONebcab992024-04-09 11:44:54 +02002015 return start;
Dragan Dosen1a5d0602016-07-22 16:00:31 +02002016 }
Aurelien DARRAGONebcab992024-04-09 11:44:54 +02002017 return NULL;
Dragan Dosen1a5d0602016-07-22 16:00:31 +02002018}
2019
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002020/* Check a string for using it in a CSV output format. If the string contains
2021 * one of the following four char <">, <,>, CR or LF, the string is
2022 * encapsulated between <"> and the <"> are escaped by a <""> sequence.
2023 * <str> is the input string to be escaped. The function assumes that
2024 * the input string is null-terminated.
2025 *
2026 * If <quote> is 0, the result is returned escaped but without double quote.
Willy Tarreau898529b2016-01-06 18:07:04 +01002027 * It is useful if the escaped string is used between double quotes in the
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002028 * format.
2029 *
Emeric Brun36e54272024-01-23 15:44:32 +01002030 * printf("..., \"%s\", ...\r\n", csv_enc(str, 0, 0, &trash));
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002031 *
Willy Tarreaub631c292016-01-08 10:04:08 +01002032 * If <quote> is 1, the converter puts the quotes only if any reserved character
2033 * is present. If <quote> is 2, the converter always puts the quotes.
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002034 *
Emeric Brun36e54272024-01-23 15:44:32 +01002035 * If <oneline> is not 0, CRs are skipped and LFs are replaced by spaces.
2036 * This re-format multi-lines strings to only one line. The purpose is to
2037 * allow a line by line parsing but also to keep the output compliant with
2038 * the CLI witch uses LF to defines the end of the response.
2039 *
2040 * If <oneline> is 2, In addition to previous action, the trailing spaces are
2041 * removed.
2042 *
Willy Tarreau83061a82018-07-13 11:56:34 +02002043 * <output> is a struct buffer used for storing the output string.
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002044 *
Willy Tarreau898529b2016-01-06 18:07:04 +01002045 * The function returns the converted string on its output. If an error
2046 * occurs, the function returns an empty string. This type of output is useful
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002047 * for using the function directly as printf() argument.
2048 *
2049 * If the output buffer is too short to contain the input string, the result
2050 * is truncated.
Willy Tarreau898529b2016-01-06 18:07:04 +01002051 *
Willy Tarreaub631c292016-01-08 10:04:08 +01002052 * This function appends the encoding to the existing output chunk, and it
2053 * guarantees that it starts immediately at the first available character of
2054 * the chunk. Please use csv_enc() instead if you want to replace the output
2055 * chunk.
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002056 */
Emeric Brun36e54272024-01-23 15:44:32 +01002057const char *csv_enc_append(const char *str, int quote, int oneline, struct buffer *output)
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002058{
Willy Tarreau843b7cb2018-07-13 10:54:26 +02002059 char *end = output->area + output->size;
2060 char *out = output->area + output->data;
Willy Tarreau898529b2016-01-06 18:07:04 +01002061 char *ptr = out;
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002062
Willy Tarreaub631c292016-01-08 10:04:08 +01002063 if (quote == 1) {
2064 /* automatic quoting: first verify if we'll have to quote the string */
2065 if (!strpbrk(str, "\n\r,\""))
2066 quote = 0;
2067 }
2068
2069 if (quote)
2070 *ptr++ = '"';
2071
Willy Tarreau898529b2016-01-06 18:07:04 +01002072 while (*str && ptr < end - 2) { /* -2 for reserving space for <"> and \0. */
Emeric Brun36e54272024-01-23 15:44:32 +01002073 if (oneline) {
2074 if (*str == '\n' ) {
2075 /* replace LF by a space */
2076 *ptr++ = ' ';
2077 str++;
2078 continue;
2079 }
2080 else if (*str == '\r' ) {
2081 /* skip CR */
2082 str++;
2083 continue;
2084 }
2085 }
Willy Tarreau898529b2016-01-06 18:07:04 +01002086 *ptr = *str;
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002087 if (*str == '"') {
Willy Tarreau898529b2016-01-06 18:07:04 +01002088 ptr++;
2089 if (ptr >= end - 2) {
2090 ptr--;
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002091 break;
2092 }
Willy Tarreau898529b2016-01-06 18:07:04 +01002093 *ptr = '"';
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002094 }
Willy Tarreau898529b2016-01-06 18:07:04 +01002095 ptr++;
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002096 str++;
2097 }
2098
Emeric Brun36e54272024-01-23 15:44:32 +01002099 if (oneline == 2) {
2100 /* remove trailing spaces */
2101 while (ptr > out && *(ptr - 1) == ' ')
2102 ptr--;
2103 }
2104
Willy Tarreaub631c292016-01-08 10:04:08 +01002105 if (quote)
2106 *ptr++ = '"';
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002107
Willy Tarreau898529b2016-01-06 18:07:04 +01002108 *ptr = '\0';
Willy Tarreau843b7cb2018-07-13 10:54:26 +02002109 output->data = ptr - output->area;
Willy Tarreau898529b2016-01-06 18:07:04 +01002110 return out;
Thierry FOURNIERddea6262015-05-28 16:00:28 +02002111}
2112
Willy Tarreaubf9c2fc2011-05-31 18:06:18 +02002113/* Decode an URL-encoded string in-place. The resulting string might
2114 * be shorter. If some forbidden characters are found, the conversion is
Thierry FOURNIER5068d962013-10-04 16:27:27 +02002115 * aborted, the string is truncated before the issue and a negative value is
2116 * returned, otherwise the operation returns the length of the decoded string.
Willy Tarreau62ba9ba2020-04-23 17:54:47 +02002117 * If the 'in_form' argument is non-nul the string is assumed to be part of
2118 * an "application/x-www-form-urlencoded" encoded string, and the '+' will be
2119 * turned to a space. If it's zero, this will only be done after a question
2120 * mark ('?').
Willy Tarreaubf9c2fc2011-05-31 18:06:18 +02002121 */
Willy Tarreau62ba9ba2020-04-23 17:54:47 +02002122int url_decode(char *string, int in_form)
Willy Tarreaubf9c2fc2011-05-31 18:06:18 +02002123{
2124 char *in, *out;
Thierry FOURNIER5068d962013-10-04 16:27:27 +02002125 int ret = -1;
Willy Tarreaubf9c2fc2011-05-31 18:06:18 +02002126
2127 in = string;
2128 out = string;
2129 while (*in) {
2130 switch (*in) {
2131 case '+' :
Willy Tarreau62ba9ba2020-04-23 17:54:47 +02002132 *out++ = in_form ? ' ' : *in;
Willy Tarreaubf9c2fc2011-05-31 18:06:18 +02002133 break;
2134 case '%' :
2135 if (!ishex(in[1]) || !ishex(in[2]))
2136 goto end;
2137 *out++ = (hex2i(in[1]) << 4) + hex2i(in[2]);
2138 in += 2;
2139 break;
Willy Tarreau62ba9ba2020-04-23 17:54:47 +02002140 case '?':
2141 in_form = 1;
Willy Tarreau7de8de02022-11-14 07:20:09 +01002142 __fallthrough;
Willy Tarreaubf9c2fc2011-05-31 18:06:18 +02002143 default:
2144 *out++ = *in;
2145 break;
2146 }
2147 in++;
2148 }
Thierry FOURNIER5068d962013-10-04 16:27:27 +02002149 ret = out - string; /* success */
Willy Tarreaubf9c2fc2011-05-31 18:06:18 +02002150 end:
2151 *out = 0;
2152 return ret;
2153}
Willy Tarreaubaaee002006-06-26 02:48:02 +02002154
Willy Tarreau6911fa42007-03-04 18:06:08 +01002155unsigned int str2ui(const char *s)
2156{
2157 return __str2ui(s);
2158}
2159
2160unsigned int str2uic(const char *s)
2161{
2162 return __str2uic(s);
2163}
2164
2165unsigned int strl2ui(const char *s, int len)
2166{
2167 return __strl2ui(s, len);
2168}
2169
2170unsigned int strl2uic(const char *s, int len)
2171{
2172 return __strl2uic(s, len);
2173}
2174
Willy Tarreau4ec83cd2010-10-15 23:19:55 +02002175unsigned int read_uint(const char **s, const char *end)
2176{
2177 return __read_uint(s, end);
2178}
2179
Thierry FOURNIER763a5d82015-07-06 23:09:52 +02002180/* This function reads an unsigned integer from the string pointed to by <s> and
2181 * returns it. The <s> pointer is adjusted to point to the first unread char. The
2182 * function automatically stops at <end>. If the number overflows, the 2^64-1
2183 * value is returned.
2184 */
2185unsigned long long int read_uint64(const char **s, const char *end)
2186{
2187 const char *ptr = *s;
2188 unsigned long long int i = 0, tmp;
2189 unsigned int j;
2190
2191 while (ptr < end) {
2192
2193 /* read next char */
2194 j = *ptr - '0';
2195 if (j > 9)
2196 goto read_uint64_end;
2197
2198 /* add char to the number and check overflow. */
2199 tmp = i * 10;
2200 if (tmp / 10 != i) {
2201 i = ULLONG_MAX;
2202 goto read_uint64_eat;
2203 }
2204 if (ULLONG_MAX - tmp < j) {
2205 i = ULLONG_MAX;
2206 goto read_uint64_eat;
2207 }
2208 i = tmp + j;
2209 ptr++;
2210 }
2211read_uint64_eat:
2212 /* eat each numeric char */
2213 while (ptr < end) {
2214 if ((unsigned int)(*ptr - '0') > 9)
2215 break;
2216 ptr++;
2217 }
2218read_uint64_end:
2219 *s = ptr;
2220 return i;
2221}
2222
2223/* This function reads an integer from the string pointed to by <s> and returns
2224 * it. The <s> pointer is adjusted to point to the first unread char. The function
2225 * automatically stops at <end>. Il the number is bigger than 2^63-2, the 2^63-1
2226 * value is returned. If the number is lowest than -2^63-1, the -2^63 value is
2227 * returned.
2228 */
2229long long int read_int64(const char **s, const char *end)
2230{
2231 unsigned long long int i = 0;
2232 int neg = 0;
2233
2234 /* Look for minus char. */
2235 if (**s == '-') {
2236 neg = 1;
2237 (*s)++;
2238 }
2239 else if (**s == '+')
2240 (*s)++;
2241
2242 /* convert as positive number. */
2243 i = read_uint64(s, end);
2244
2245 if (neg) {
2246 if (i > 0x8000000000000000ULL)
2247 return LLONG_MIN;
2248 return -i;
2249 }
2250 if (i > 0x7fffffffffffffffULL)
2251 return LLONG_MAX;
2252 return i;
2253}
2254
Willy Tarreau6911fa42007-03-04 18:06:08 +01002255/* This one is 7 times faster than strtol() on athlon with checks.
2256 * It returns the value of the number composed of all valid digits read,
2257 * and can process negative numbers too.
2258 */
2259int strl2ic(const char *s, int len)
2260{
2261 int i = 0;
Willy Tarreau3f0c9762007-10-25 09:42:24 +02002262 int j, k;
Willy Tarreau6911fa42007-03-04 18:06:08 +01002263
2264 if (len > 0) {
2265 if (*s != '-') {
2266 /* positive number */
2267 while (len-- > 0) {
2268 j = (*s++) - '0';
Willy Tarreau3f0c9762007-10-25 09:42:24 +02002269 k = i * 10;
Willy Tarreau6911fa42007-03-04 18:06:08 +01002270 if (j > 9)
2271 break;
Willy Tarreau3f0c9762007-10-25 09:42:24 +02002272 i = k + j;
Willy Tarreau6911fa42007-03-04 18:06:08 +01002273 }
2274 } else {
2275 /* negative number */
2276 s++;
2277 while (--len > 0) {
2278 j = (*s++) - '0';
Willy Tarreau3f0c9762007-10-25 09:42:24 +02002279 k = i * 10;
Willy Tarreau6911fa42007-03-04 18:06:08 +01002280 if (j > 9)
2281 break;
Willy Tarreau3f0c9762007-10-25 09:42:24 +02002282 i = k - j;
Willy Tarreau6911fa42007-03-04 18:06:08 +01002283 }
2284 }
2285 }
2286 return i;
2287}
2288
2289
2290/* This function reads exactly <len> chars from <s> and converts them to a
2291 * signed integer which it stores into <ret>. It accurately detects any error
2292 * (truncated string, invalid chars, overflows). It is meant to be used in
2293 * applications designed for hostile environments. It returns zero when the
2294 * number has successfully been converted, non-zero otherwise. When an error
2295 * is returned, the <ret> value is left untouched. It is yet 5 to 40 times
2296 * faster than strtol().
2297 */
2298int strl2irc(const char *s, int len, int *ret)
2299{
2300 int i = 0;
2301 int j;
2302
2303 if (!len)
2304 return 1;
2305
2306 if (*s != '-') {
2307 /* positive number */
2308 while (len-- > 0) {
2309 j = (*s++) - '0';
2310 if (j > 9) return 1; /* invalid char */
2311 if (i > INT_MAX / 10) return 1; /* check for multiply overflow */
2312 i = i * 10;
2313 if (i + j < i) return 1; /* check for addition overflow */
2314 i = i + j;
2315 }
2316 } else {
2317 /* negative number */
2318 s++;
2319 while (--len > 0) {
2320 j = (*s++) - '0';
2321 if (j > 9) return 1; /* invalid char */
2322 if (i < INT_MIN / 10) return 1; /* check for multiply overflow */
2323 i = i * 10;
2324 if (i - j > i) return 1; /* check for subtract overflow */
2325 i = i - j;
2326 }
2327 }
2328 *ret = i;
2329 return 0;
2330}
2331
2332
2333/* This function reads exactly <len> chars from <s> and converts them to a
2334 * signed integer which it stores into <ret>. It accurately detects any error
2335 * (truncated string, invalid chars, overflows). It is meant to be used in
2336 * applications designed for hostile environments. It returns zero when the
2337 * number has successfully been converted, non-zero otherwise. When an error
2338 * is returned, the <ret> value is left untouched. It is about 3 times slower
William Dauchy060ffc82021-02-06 20:47:51 +01002339 * than strl2irc().
Willy Tarreau6911fa42007-03-04 18:06:08 +01002340 */
Willy Tarreau6911fa42007-03-04 18:06:08 +01002341
2342int strl2llrc(const char *s, int len, long long *ret)
2343{
2344 long long i = 0;
2345 int j;
2346
2347 if (!len)
2348 return 1;
2349
2350 if (*s != '-') {
2351 /* positive number */
2352 while (len-- > 0) {
2353 j = (*s++) - '0';
2354 if (j > 9) return 1; /* invalid char */
2355 if (i > LLONG_MAX / 10LL) return 1; /* check for multiply overflow */
2356 i = i * 10LL;
2357 if (i + j < i) return 1; /* check for addition overflow */
2358 i = i + j;
2359 }
2360 } else {
2361 /* negative number */
2362 s++;
2363 while (--len > 0) {
2364 j = (*s++) - '0';
2365 if (j > 9) return 1; /* invalid char */
2366 if (i < LLONG_MIN / 10LL) return 1; /* check for multiply overflow */
2367 i = i * 10LL;
2368 if (i - j > i) return 1; /* check for subtract overflow */
2369 i = i - j;
2370 }
2371 }
2372 *ret = i;
2373 return 0;
2374}
2375
Thierry FOURNIER511e9472014-01-23 17:40:34 +01002376/* This function is used with pat_parse_dotted_ver(). It converts a string
2377 * composed by two number separated by a dot. Each part must contain in 16 bits
2378 * because internally they will be represented as a 32-bit quantity stored in
2379 * a 64-bit integer. It returns zero when the number has successfully been
2380 * converted, non-zero otherwise. When an error is returned, the <ret> value
2381 * is left untouched.
2382 *
2383 * "1.3" -> 0x0000000000010003
2384 * "65535.65535" -> 0x00000000ffffffff
2385 */
2386int strl2llrc_dotted(const char *text, int len, long long *ret)
2387{
2388 const char *end = &text[len];
2389 const char *p;
2390 long long major, minor;
2391
2392 /* Look for dot. */
2393 for (p = text; p < end; p++)
2394 if (*p == '.')
2395 break;
2396
2397 /* Convert major. */
2398 if (strl2llrc(text, p - text, &major) != 0)
2399 return 1;
2400
2401 /* Check major. */
2402 if (major >= 65536)
2403 return 1;
2404
2405 /* Convert minor. */
2406 minor = 0;
2407 if (p < end)
2408 if (strl2llrc(p + 1, end - (p + 1), &minor) != 0)
2409 return 1;
2410
2411 /* Check minor. */
2412 if (minor >= 65536)
2413 return 1;
2414
2415 /* Compose value. */
2416 *ret = (major << 16) | (minor & 0xffff);
2417 return 0;
2418}
2419
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002420/* This function parses a time value optionally followed by a unit suffix among
2421 * "d", "h", "m", "s", "ms" or "us". It converts the value into the unit
2422 * expected by the caller. The computation does its best to avoid overflows.
2423 * The value is returned in <ret> if everything is fine, and a NULL is returned
2424 * by the function. In case of error, a pointer to the error is returned and
2425 * <ret> is left untouched. Values are automatically rounded up when needed.
Willy Tarreau9faebe32019-06-07 19:00:37 +02002426 * Values resulting in values larger than or equal to 2^31 after conversion are
2427 * reported as an overflow as value PARSE_TIME_OVER. Non-null values resulting
2428 * in an underflow are reported as an underflow as value PARSE_TIME_UNDER.
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002429 */
2430const char *parse_time_err(const char *text, unsigned *ret, unsigned unit_flags)
2431{
Willy Tarreau9faebe32019-06-07 19:00:37 +02002432 unsigned long long imult, idiv;
2433 unsigned long long omult, odiv;
2434 unsigned long long value, result;
Christopher Fauletc20ad0d2020-12-11 09:23:07 +01002435 const char *str = text;
2436
2437 if (!isdigit((unsigned char)*text))
2438 return text;
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002439
2440 omult = odiv = 1;
2441
2442 switch (unit_flags & TIME_UNIT_MASK) {
2443 case TIME_UNIT_US: omult = 1000000; break;
2444 case TIME_UNIT_MS: omult = 1000; break;
2445 case TIME_UNIT_S: break;
2446 case TIME_UNIT_MIN: odiv = 60; break;
2447 case TIME_UNIT_HOUR: odiv = 3600; break;
2448 case TIME_UNIT_DAY: odiv = 86400; break;
2449 default: break;
2450 }
2451
2452 value = 0;
2453
2454 while (1) {
2455 unsigned int j;
2456
2457 j = *text - '0';
2458 if (j > 9)
2459 break;
2460 text++;
2461 value *= 10;
2462 value += j;
2463 }
2464
2465 imult = idiv = 1;
2466 switch (*text) {
2467 case '\0': /* no unit = default unit */
2468 imult = omult = idiv = odiv = 1;
Christopher Fauletc20ad0d2020-12-11 09:23:07 +01002469 goto end;
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002470 case 's': /* second = unscaled unit */
2471 break;
2472 case 'u': /* microsecond : "us" */
2473 if (text[1] == 's') {
2474 idiv = 1000000;
2475 text++;
Thayne McCombsa6838052021-04-02 14:12:43 -06002476 break;
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002477 }
Christopher Fauletc20ad0d2020-12-11 09:23:07 +01002478 return text;
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002479 case 'm': /* millisecond : "ms" or minute: "m" */
2480 if (text[1] == 's') {
2481 idiv = 1000;
2482 text++;
2483 } else
2484 imult = 60;
2485 break;
2486 case 'h': /* hour : "h" */
2487 imult = 3600;
2488 break;
2489 case 'd': /* day : "d" */
2490 imult = 86400;
2491 break;
2492 default:
2493 return text;
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002494 }
Christopher Fauletc20ad0d2020-12-11 09:23:07 +01002495 if (*(++text) != '\0') {
2496 ha_warning("unexpected character '%c' after the timer value '%s', only "
2497 "(us=microseconds,ms=milliseconds,s=seconds,m=minutes,h=hours,d=days) are supported."
2498 " This will be reported as an error in next versions.\n", *text, str);
2499 }
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002500
Christopher Fauletc20ad0d2020-12-11 09:23:07 +01002501 end:
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002502 if (omult % idiv == 0) { omult /= idiv; idiv = 1; }
2503 if (idiv % omult == 0) { idiv /= omult; omult = 1; }
2504 if (imult % odiv == 0) { imult /= odiv; odiv = 1; }
2505 if (odiv % imult == 0) { odiv /= imult; imult = 1; }
2506
Willy Tarreau9faebe32019-06-07 19:00:37 +02002507 result = (value * (imult * omult) + (idiv * odiv - 1)) / (idiv * odiv);
2508 if (result >= 0x80000000)
2509 return PARSE_TIME_OVER;
2510 if (!result && value)
2511 return PARSE_TIME_UNDER;
2512 *ret = result;
Willy Tarreaua0d37b62007-12-02 22:00:35 +01002513 return NULL;
2514}
Willy Tarreau6911fa42007-03-04 18:06:08 +01002515
Emeric Brun39132b22010-01-04 14:57:24 +01002516/* this function converts the string starting at <text> to an unsigned int
2517 * stored in <ret>. If an error is detected, the pointer to the unexpected
Joseph Herlant32b83272018-11-15 11:58:28 -08002518 * character is returned. If the conversion is successful, NULL is returned.
Emeric Brun39132b22010-01-04 14:57:24 +01002519 */
2520const char *parse_size_err(const char *text, unsigned *ret) {
2521 unsigned value = 0;
2522
Christopher Faulet82635a02020-12-11 09:30:45 +01002523 if (!isdigit((unsigned char)*text))
2524 return text;
2525
Emeric Brun39132b22010-01-04 14:57:24 +01002526 while (1) {
2527 unsigned int j;
2528
2529 j = *text - '0';
2530 if (j > 9)
2531 break;
2532 if (value > ~0U / 10)
2533 return text;
2534 value *= 10;
2535 if (value > (value + j))
2536 return text;
2537 value += j;
2538 text++;
2539 }
2540
2541 switch (*text) {
2542 case '\0':
2543 break;
2544 case 'K':
2545 case 'k':
2546 if (value > ~0U >> 10)
2547 return text;
2548 value = value << 10;
2549 break;
2550 case 'M':
2551 case 'm':
2552 if (value > ~0U >> 20)
2553 return text;
2554 value = value << 20;
2555 break;
2556 case 'G':
2557 case 'g':
2558 if (value > ~0U >> 30)
2559 return text;
2560 value = value << 30;
2561 break;
2562 default:
2563 return text;
2564 }
2565
Godbach58048a22015-01-28 17:36:16 +08002566 if (*text != '\0' && *++text != '\0')
2567 return text;
2568
Emeric Brun39132b22010-01-04 14:57:24 +01002569 *ret = value;
2570 return NULL;
2571}
2572
Willy Tarreau126d4062013-12-03 17:50:47 +01002573/*
2574 * Parse binary string written in hexadecimal (source) and store the decoded
Ilya Shipitsin856aabc2020-04-16 23:51:34 +05002575 * result into binstr and set binstrlen to the length of binstr. Memory for
Willy Tarreau126d4062013-12-03 17:50:47 +01002576 * binstr is allocated by the function. In case of error, returns 0 with an
Ilya Shipitsin856aabc2020-04-16 23:51:34 +05002577 * error message in err. In success case, it returns the consumed length.
Willy Tarreau126d4062013-12-03 17:50:47 +01002578 */
2579int parse_binary(const char *source, char **binstr, int *binstrlen, char **err)
2580{
2581 int len;
2582 const char *p = source;
2583 int i,j;
Thierry FOURNIER9645d422013-12-06 19:59:28 +01002584 int alloc;
Willy Tarreau126d4062013-12-03 17:50:47 +01002585
2586 len = strlen(source);
2587 if (len % 2) {
2588 memprintf(err, "an even number of hex digit is expected");
2589 return 0;
2590 }
2591
2592 len = len >> 1;
Thierry FOURNIER9645d422013-12-06 19:59:28 +01002593
Willy Tarreau126d4062013-12-03 17:50:47 +01002594 if (!*binstr) {
Tim Duesterhuse52b6e52020-09-12 20:26:43 +02002595 *binstr = calloc(len, sizeof(**binstr));
Thierry FOURNIER9645d422013-12-06 19:59:28 +01002596 if (!*binstr) {
2597 memprintf(err, "out of memory while loading string pattern");
2598 return 0;
2599 }
2600 alloc = 1;
Willy Tarreau126d4062013-12-03 17:50:47 +01002601 }
Thierry FOURNIER9645d422013-12-06 19:59:28 +01002602 else {
2603 if (*binstrlen < len) {
Joseph Herlant76dbe782018-11-15 12:01:22 -08002604 memprintf(err, "no space available in the buffer. expect %d, provides %d",
Thierry FOURNIER9645d422013-12-06 19:59:28 +01002605 len, *binstrlen);
2606 return 0;
2607 }
2608 alloc = 0;
2609 }
2610 *binstrlen = len;
Willy Tarreau126d4062013-12-03 17:50:47 +01002611
2612 i = j = 0;
2613 while (j < len) {
2614 if (!ishex(p[i++]))
2615 goto bad_input;
2616 if (!ishex(p[i++]))
2617 goto bad_input;
2618 (*binstr)[j++] = (hex2i(p[i-2]) << 4) + hex2i(p[i-1]);
2619 }
Thierry FOURNIERee330af2014-01-21 11:36:14 +01002620 return len << 1;
Willy Tarreau126d4062013-12-03 17:50:47 +01002621
2622bad_input:
2623 memprintf(err, "an hex digit is expected (found '%c')", p[i-1]);
Willy Tarreau61cfdf42021-02-20 10:46:51 +01002624 if (alloc)
2625 ha_free(binstr);
Willy Tarreau126d4062013-12-03 17:50:47 +01002626 return 0;
2627}
2628
Willy Tarreau946ba592009-05-10 15:41:18 +02002629/* copies at most <n> characters from <src> and always terminates with '\0' */
2630char *my_strndup(const char *src, int n)
2631{
2632 int len = 0;
2633 char *ret;
2634
2635 while (len < n && src[len])
2636 len++;
2637
Vincent Bernat3c2f2f22016-04-03 13:48:42 +02002638 ret = malloc(len + 1);
Willy Tarreau946ba592009-05-10 15:41:18 +02002639 if (!ret)
2640 return ret;
2641 memcpy(ret, src, len);
2642 ret[len] = '\0';
2643 return ret;
2644}
2645
Baptiste Assmannbb77c8e2013-10-06 23:24:13 +02002646/*
2647 * search needle in haystack
2648 * returns the pointer if found, returns NULL otherwise
2649 */
2650const void *my_memmem(const void *haystack, size_t haystacklen, const void *needle, size_t needlelen)
2651{
2652 const void *c = NULL;
2653 unsigned char f;
2654
2655 if ((haystack == NULL) || (needle == NULL) || (haystacklen < needlelen))
2656 return NULL;
2657
2658 f = *(char *)needle;
2659 c = haystack;
2660 while ((c = memchr(c, f, haystacklen - (c - haystack))) != NULL) {
2661 if ((haystacklen - (c - haystack)) < needlelen)
2662 return NULL;
2663
2664 if (memcmp(c, needle, needlelen) == 0)
2665 return c;
2666 ++c;
2667 }
2668 return NULL;
2669}
2670
Ilya Shipitsinc02a23f2020-05-06 00:53:22 +05002671/* get length of the initial segment consisting entirely of bytes in <accept> */
Christopher Faulet5eb96cb2020-04-15 10:23:01 +02002672size_t my_memspn(const void *str, size_t len, const void *accept, size_t acceptlen)
2673{
2674 size_t ret = 0;
2675
2676 while (ret < len && memchr(accept, *((int *)str), acceptlen)) {
2677 str++;
2678 ret++;
2679 }
2680 return ret;
2681}
2682
Ilya Shipitsinc02a23f2020-05-06 00:53:22 +05002683/* get length of the initial segment consisting entirely of bytes not in <rejcet> */
Christopher Faulet5eb96cb2020-04-15 10:23:01 +02002684size_t my_memcspn(const void *str, size_t len, const void *reject, size_t rejectlen)
2685{
2686 size_t ret = 0;
2687
2688 while (ret < len) {
2689 if(memchr(reject, *((int *)str), rejectlen))
2690 return ret;
2691 str++;
2692 ret++;
2693 }
2694 return ret;
2695}
2696
Willy Tarreau482b00d2009-10-04 22:48:42 +02002697/* This function returns the first unused key greater than or equal to <key> in
2698 * ID tree <root>. Zero is returned if no place is found.
2699 */
2700unsigned int get_next_id(struct eb_root *root, unsigned int key)
2701{
2702 struct eb32_node *used;
2703
2704 do {
2705 used = eb32_lookup_ge(root, key);
2706 if (!used || used->key > key)
2707 return key; /* key is available */
2708 key++;
2709 } while (key);
2710 return key;
2711}
2712
Willy Tarreau9c1e15d2017-11-15 18:51:29 +01002713/* dump the full tree to <file> in DOT format for debugging purposes. Will
2714 * optionally highlight node <subj> if found, depending on operation <op> :
2715 * 0 : nothing
2716 * >0 : insertion, node/leaf are surrounded in red
2717 * <0 : removal, node/leaf are dashed with no background
2718 * Will optionally add "desc" as a label on the graph if set and non-null.
2719 */
2720void eb32sc_to_file(FILE *file, struct eb_root *root, const struct eb32sc_node *subj, int op, const char *desc)
Willy Tarreaued3cda02017-11-15 15:04:05 +01002721{
2722 struct eb32sc_node *node;
2723 unsigned long scope = -1;
2724
Willy Tarreau9c1e15d2017-11-15 18:51:29 +01002725 fprintf(file, "digraph ebtree {\n");
2726
2727 if (desc && *desc) {
2728 fprintf(file,
2729 " fontname=\"fixed\";\n"
2730 " fontsize=8;\n"
2731 " label=\"%s\";\n", desc);
2732 }
2733
Willy Tarreaued3cda02017-11-15 15:04:05 +01002734 fprintf(file,
Willy Tarreau6c7f4de2017-11-15 17:49:54 +01002735 " node [fontname=\"fixed\" fontsize=8 shape=\"box\" style=\"filled\" color=\"black\" fillcolor=\"white\"];\n"
2736 " edge [fontname=\"fixed\" fontsize=8 style=\"solid\" color=\"magenta\" dir=\"forward\"];\n"
Willy Tarreaued3cda02017-11-15 15:04:05 +01002737 " \"%lx_n\" [label=\"root\\n%lx\"]\n", (long)eb_root_to_node(root), (long)root
2738 );
2739
Willy Tarreau6c7f4de2017-11-15 17:49:54 +01002740 fprintf(file, " \"%lx_n\" -> \"%lx_%c\" [taillabel=\"L\"];\n",
Willy Tarreaued3cda02017-11-15 15:04:05 +01002741 (long)eb_root_to_node(root),
2742 (long)eb_root_to_node(eb_clrtag(root->b[0])),
Willy Tarreaued3cda02017-11-15 15:04:05 +01002743 eb_gettag(root->b[0]) == EB_LEAF ? 'l' : 'n');
2744
2745 node = eb32sc_first(root, scope);
2746 while (node) {
2747 if (node->node.node_p) {
2748 /* node part is used */
Willy Tarreau9c1e15d2017-11-15 18:51:29 +01002749 fprintf(file, " \"%lx_n\" [label=\"%lx\\nkey=%u\\nscope=%lx\\nbit=%d\" fillcolor=\"lightskyblue1\" %s];\n",
2750 (long)node, (long)node, node->key, node->node_s, node->node.bit,
2751 (node == subj) ? (op < 0 ? "color=\"red\" style=\"dashed\"" : op > 0 ? "color=\"red\"" : "") : "");
Willy Tarreaued3cda02017-11-15 15:04:05 +01002752
Willy Tarreau6c7f4de2017-11-15 17:49:54 +01002753 fprintf(file, " \"%lx_n\" -> \"%lx_n\" [taillabel=\"%c\"];\n",
Willy Tarreaued3cda02017-11-15 15:04:05 +01002754 (long)node,
2755 (long)eb_root_to_node(eb_clrtag(node->node.node_p)),
Willy Tarreau6c7f4de2017-11-15 17:49:54 +01002756 eb_gettag(node->node.node_p) ? 'R' : 'L');
Willy Tarreaued3cda02017-11-15 15:04:05 +01002757
Willy Tarreau6c7f4de2017-11-15 17:49:54 +01002758 fprintf(file, " \"%lx_n\" -> \"%lx_%c\" [taillabel=\"L\"];\n",
Willy Tarreaued3cda02017-11-15 15:04:05 +01002759 (long)node,
2760 (long)eb_root_to_node(eb_clrtag(node->node.branches.b[0])),
Willy Tarreaued3cda02017-11-15 15:04:05 +01002761 eb_gettag(node->node.branches.b[0]) == EB_LEAF ? 'l' : 'n');
2762
Willy Tarreau6c7f4de2017-11-15 17:49:54 +01002763 fprintf(file, " \"%lx_n\" -> \"%lx_%c\" [taillabel=\"R\"];\n",
Willy Tarreaued3cda02017-11-15 15:04:05 +01002764 (long)node,
2765 (long)eb_root_to_node(eb_clrtag(node->node.branches.b[1])),
Willy Tarreaued3cda02017-11-15 15:04:05 +01002766 eb_gettag(node->node.branches.b[1]) == EB_LEAF ? 'l' : 'n');
2767 }
2768
Willy Tarreau9c1e15d2017-11-15 18:51:29 +01002769 fprintf(file, " \"%lx_l\" [label=\"%lx\\nkey=%u\\nscope=%lx\\npfx=%u\" fillcolor=\"yellow\" %s];\n",
2770 (long)node, (long)node, node->key, node->leaf_s, node->node.pfx,
2771 (node == subj) ? (op < 0 ? "color=\"red\" style=\"dashed\"" : op > 0 ? "color=\"red\"" : "") : "");
Willy Tarreaued3cda02017-11-15 15:04:05 +01002772
Willy Tarreau6c7f4de2017-11-15 17:49:54 +01002773 fprintf(file, " \"%lx_l\" -> \"%lx_n\" [taillabel=\"%c\"];\n",
Willy Tarreaued3cda02017-11-15 15:04:05 +01002774 (long)node,
2775 (long)eb_root_to_node(eb_clrtag(node->node.leaf_p)),
Willy Tarreau6c7f4de2017-11-15 17:49:54 +01002776 eb_gettag(node->node.leaf_p) ? 'R' : 'L');
Willy Tarreaued3cda02017-11-15 15:04:05 +01002777 node = eb32sc_next(node, scope);
2778 }
2779 fprintf(file, "}\n");
2780}
2781
Willy Tarreau0dc9e6d2022-08-01 11:55:57 +02002782/* dump the full tree to <file> in DOT format for debugging purposes. Will
2783 * optionally highlight node <subj> if found, depending on operation <op> :
2784 * 0 : nothing
2785 * >0 : insertion, node/leaf are surrounded in red
2786 * <0 : removal, node/leaf are dashed with no background
2787 * Will optionally add "desc" as a label on the graph if set and non-null. The
2788 * key is printed as a u32 hex value. A full-sized hex dump would be better but
2789 * is left to be implemented.
2790 */
2791void ebmb_to_file(FILE *file, struct eb_root *root, const struct ebmb_node *subj, int op, const char *desc)
2792{
2793 struct ebmb_node *node;
2794
2795 fprintf(file, "digraph ebtree {\n");
2796
2797 if (desc && *desc) {
2798 fprintf(file,
2799 " fontname=\"fixed\";\n"
2800 " fontsize=8;\n"
2801 " label=\"%s\";\n", desc);
2802 }
2803
2804 fprintf(file,
2805 " node [fontname=\"fixed\" fontsize=8 shape=\"box\" style=\"filled\" color=\"black\" fillcolor=\"white\"];\n"
2806 " edge [fontname=\"fixed\" fontsize=8 style=\"solid\" color=\"magenta\" dir=\"forward\"];\n"
2807 " \"%lx_n\" [label=\"root\\n%lx\"]\n", (long)eb_root_to_node(root), (long)root
2808 );
2809
2810 fprintf(file, " \"%lx_n\" -> \"%lx_%c\" [taillabel=\"L\"];\n",
2811 (long)eb_root_to_node(root),
2812 (long)eb_root_to_node(eb_clrtag(root->b[0])),
2813 eb_gettag(root->b[0]) == EB_LEAF ? 'l' : 'n');
2814
2815 node = ebmb_first(root);
2816 while (node) {
2817 if (node->node.node_p) {
2818 /* node part is used */
2819 fprintf(file, " \"%lx_n\" [label=\"%lx\\nkey=%#x\\nbit=%d\" fillcolor=\"lightskyblue1\" %s];\n",
2820 (long)node, (long)node, read_u32(node->key), node->node.bit,
2821 (node == subj) ? (op < 0 ? "color=\"red\" style=\"dashed\"" : op > 0 ? "color=\"red\"" : "") : "");
2822
2823 fprintf(file, " \"%lx_n\" -> \"%lx_n\" [taillabel=\"%c\"];\n",
2824 (long)node,
2825 (long)eb_root_to_node(eb_clrtag(node->node.node_p)),
2826 eb_gettag(node->node.node_p) ? 'R' : 'L');
2827
2828 fprintf(file, " \"%lx_n\" -> \"%lx_%c\" [taillabel=\"L\"];\n",
2829 (long)node,
2830 (long)eb_root_to_node(eb_clrtag(node->node.branches.b[0])),
2831 eb_gettag(node->node.branches.b[0]) == EB_LEAF ? 'l' : 'n');
2832
2833 fprintf(file, " \"%lx_n\" -> \"%lx_%c\" [taillabel=\"R\"];\n",
2834 (long)node,
2835 (long)eb_root_to_node(eb_clrtag(node->node.branches.b[1])),
2836 eb_gettag(node->node.branches.b[1]) == EB_LEAF ? 'l' : 'n');
2837 }
2838
2839 fprintf(file, " \"%lx_l\" [label=\"%lx\\nkey=%#x\\npfx=%u\" fillcolor=\"yellow\" %s];\n",
2840 (long)node, (long)node, read_u32(node->key), node->node.pfx,
2841 (node == subj) ? (op < 0 ? "color=\"red\" style=\"dashed\"" : op > 0 ? "color=\"red\"" : "") : "");
2842
2843 fprintf(file, " \"%lx_l\" -> \"%lx_n\" [taillabel=\"%c\"];\n",
2844 (long)node,
2845 (long)eb_root_to_node(eb_clrtag(node->node.leaf_p)),
2846 eb_gettag(node->node.leaf_p) ? 'R' : 'L');
2847 node = ebmb_next(node);
2848 }
2849 fprintf(file, "}\n");
2850}
2851
Willy Tarreau348238b2010-01-18 15:05:57 +01002852/* This function compares a sample word possibly followed by blanks to another
2853 * clean word. The compare is case-insensitive. 1 is returned if both are equal,
2854 * otherwise zero. This intends to be used when checking HTTP headers for some
2855 * values. Note that it validates a word followed only by blanks but does not
2856 * validate a word followed by blanks then other chars.
2857 */
2858int word_match(const char *sample, int slen, const char *word, int wlen)
2859{
2860 if (slen < wlen)
2861 return 0;
2862
2863 while (wlen) {
2864 char c = *sample ^ *word;
2865 if (c && c != ('A' ^ 'a'))
2866 return 0;
2867 sample++;
2868 word++;
2869 slen--;
2870 wlen--;
2871 }
2872
2873 while (slen) {
2874 if (*sample != ' ' && *sample != '\t')
2875 return 0;
2876 sample++;
2877 slen--;
2878 }
2879 return 1;
2880}
Willy Tarreau482b00d2009-10-04 22:48:42 +02002881
Willy Tarreaud54bbdc2009-09-07 11:00:31 +02002882/* Converts any text-formatted IPv4 address to a host-order IPv4 address. It
2883 * is particularly fast because it avoids expensive operations such as
2884 * multiplies, which are optimized away at the end. It requires a properly
Ilya Shipitsin856aabc2020-04-16 23:51:34 +05002885 * formatted address though (3 points).
Willy Tarreaud54bbdc2009-09-07 11:00:31 +02002886 */
2887unsigned int inetaddr_host(const char *text)
2888{
2889 const unsigned int ascii_zero = ('0' << 24) | ('0' << 16) | ('0' << 8) | '0';
2890 register unsigned int dig100, dig10, dig1;
2891 int s;
2892 const char *p, *d;
2893
2894 dig1 = dig10 = dig100 = ascii_zero;
2895 s = 24;
2896
2897 p = text;
2898 while (1) {
2899 if (((unsigned)(*p - '0')) <= 9) {
2900 p++;
2901 continue;
2902 }
2903
2904 /* here, we have a complete byte between <text> and <p> (exclusive) */
2905 if (p == text)
2906 goto end;
2907
2908 d = p - 1;
2909 dig1 |= (unsigned int)(*d << s);
2910 if (d == text)
2911 goto end;
2912
2913 d--;
2914 dig10 |= (unsigned int)(*d << s);
2915 if (d == text)
2916 goto end;
2917
2918 d--;
2919 dig100 |= (unsigned int)(*d << s);
2920 end:
2921 if (!s || *p != '.')
2922 break;
2923
2924 s -= 8;
2925 text = ++p;
2926 }
2927
2928 dig100 -= ascii_zero;
2929 dig10 -= ascii_zero;
2930 dig1 -= ascii_zero;
2931 return ((dig100 * 10) + dig10) * 10 + dig1;
2932}
2933
2934/*
2935 * Idem except the first unparsed character has to be passed in <stop>.
2936 */
2937unsigned int inetaddr_host_lim(const char *text, const char *stop)
2938{
2939 const unsigned int ascii_zero = ('0' << 24) | ('0' << 16) | ('0' << 8) | '0';
2940 register unsigned int dig100, dig10, dig1;
2941 int s;
2942 const char *p, *d;
2943
2944 dig1 = dig10 = dig100 = ascii_zero;
2945 s = 24;
2946
2947 p = text;
2948 while (1) {
2949 if (((unsigned)(*p - '0')) <= 9 && p < stop) {
2950 p++;
2951 continue;
2952 }
2953
2954 /* here, we have a complete byte between <text> and <p> (exclusive) */
2955 if (p == text)
2956 goto end;
2957
2958 d = p - 1;
2959 dig1 |= (unsigned int)(*d << s);
2960 if (d == text)
2961 goto end;
2962
2963 d--;
2964 dig10 |= (unsigned int)(*d << s);
2965 if (d == text)
2966 goto end;
2967
2968 d--;
2969 dig100 |= (unsigned int)(*d << s);
2970 end:
2971 if (!s || p == stop || *p != '.')
2972 break;
2973
2974 s -= 8;
2975 text = ++p;
2976 }
2977
2978 dig100 -= ascii_zero;
2979 dig10 -= ascii_zero;
2980 dig1 -= ascii_zero;
2981 return ((dig100 * 10) + dig10) * 10 + dig1;
2982}
2983
2984/*
2985 * Idem except the pointer to first unparsed byte is returned into <ret> which
2986 * must not be NULL.
2987 */
Willy Tarreau74172752010-10-15 23:21:42 +02002988unsigned int inetaddr_host_lim_ret(char *text, char *stop, char **ret)
Willy Tarreaud54bbdc2009-09-07 11:00:31 +02002989{
2990 const unsigned int ascii_zero = ('0' << 24) | ('0' << 16) | ('0' << 8) | '0';
2991 register unsigned int dig100, dig10, dig1;
2992 int s;
Willy Tarreau74172752010-10-15 23:21:42 +02002993 char *p, *d;
Willy Tarreaud54bbdc2009-09-07 11:00:31 +02002994
2995 dig1 = dig10 = dig100 = ascii_zero;
2996 s = 24;
2997
2998 p = text;
2999 while (1) {
3000 if (((unsigned)(*p - '0')) <= 9 && p < stop) {
3001 p++;
3002 continue;
3003 }
3004
3005 /* here, we have a complete byte between <text> and <p> (exclusive) */
3006 if (p == text)
3007 goto end;
3008
3009 d = p - 1;
3010 dig1 |= (unsigned int)(*d << s);
3011 if (d == text)
3012 goto end;
3013
3014 d--;
3015 dig10 |= (unsigned int)(*d << s);
3016 if (d == text)
3017 goto end;
3018
3019 d--;
3020 dig100 |= (unsigned int)(*d << s);
3021 end:
3022 if (!s || p == stop || *p != '.')
3023 break;
3024
3025 s -= 8;
3026 text = ++p;
3027 }
3028
3029 *ret = p;
3030 dig100 -= ascii_zero;
3031 dig10 -= ascii_zero;
3032 dig1 -= ascii_zero;
3033 return ((dig100 * 10) + dig10) * 10 + dig1;
3034}
3035
Willy Tarreauf0b38bf2010-06-06 13:22:23 +02003036/* Convert a fixed-length string to an IP address. Returns 0 in case of error,
3037 * or the number of chars read in case of success. Maybe this could be replaced
3038 * by one of the functions above. Also, apparently this function does not support
3039 * hosts above 255 and requires exactly 4 octets.
Willy Tarreau075415a2013-12-12 11:29:39 +01003040 * The destination is only modified on success.
Willy Tarreauf0b38bf2010-06-06 13:22:23 +02003041 */
3042int buf2ip(const char *buf, size_t len, struct in_addr *dst)
3043{
3044 const char *addr;
3045 int saw_digit, octets, ch;
3046 u_char tmp[4], *tp;
3047 const char *cp = buf;
3048
3049 saw_digit = 0;
3050 octets = 0;
3051 *(tp = tmp) = 0;
3052
3053 for (addr = buf; addr - buf < len; addr++) {
3054 unsigned char digit = (ch = *addr) - '0';
3055
3056 if (digit > 9 && ch != '.')
3057 break;
3058
3059 if (digit <= 9) {
3060 u_int new = *tp * 10 + digit;
3061
3062 if (new > 255)
3063 return 0;
3064
3065 *tp = new;
3066
3067 if (!saw_digit) {
3068 if (++octets > 4)
3069 return 0;
3070 saw_digit = 1;
3071 }
3072 } else if (ch == '.' && saw_digit) {
3073 if (octets == 4)
3074 return 0;
3075
3076 *++tp = 0;
3077 saw_digit = 0;
3078 } else
3079 return 0;
3080 }
3081
3082 if (octets < 4)
3083 return 0;
3084
3085 memcpy(&dst->s_addr, tmp, 4);
3086 return addr - cp;
3087}
3088
Thierry FOURNIERd559dd82013-11-22 16:16:59 +01003089/* This function converts the string in <buf> of the len <len> to
3090 * struct in6_addr <dst> which must be allocated by the caller.
3091 * This function returns 1 in success case, otherwise zero.
Willy Tarreau075415a2013-12-12 11:29:39 +01003092 * The destination is only modified on success.
Thierry FOURNIERd559dd82013-11-22 16:16:59 +01003093 */
Thierry FOURNIERd559dd82013-11-22 16:16:59 +01003094int buf2ip6(const char *buf, size_t len, struct in6_addr *dst)
3095{
Thierry FOURNIERcd659912013-12-11 12:33:54 +01003096 char null_term_ip6[INET6_ADDRSTRLEN + 1];
Willy Tarreau075415a2013-12-12 11:29:39 +01003097 struct in6_addr out;
Thierry FOURNIERd559dd82013-11-22 16:16:59 +01003098
Thierry FOURNIERcd659912013-12-11 12:33:54 +01003099 if (len > INET6_ADDRSTRLEN)
Thierry FOURNIERd559dd82013-11-22 16:16:59 +01003100 return 0;
3101
3102 memcpy(null_term_ip6, buf, len);
3103 null_term_ip6[len] = '\0';
3104
Willy Tarreau075415a2013-12-12 11:29:39 +01003105 if (!inet_pton(AF_INET6, null_term_ip6, &out))
Thierry FOURNIERd559dd82013-11-22 16:16:59 +01003106 return 0;
3107
Willy Tarreau075415a2013-12-12 11:29:39 +01003108 *dst = out;
Thierry FOURNIERd559dd82013-11-22 16:16:59 +01003109 return 1;
3110}
3111
Willy Tarreauacf95772010-06-14 19:09:21 +02003112/* To be used to quote config arg positions. Returns the short string at <ptr>
3113 * surrounded by simple quotes if <ptr> is valid and non-empty, or "end of line"
3114 * if ptr is NULL or empty. The string is locally allocated.
3115 */
3116const char *quote_arg(const char *ptr)
3117{
Christopher Faulet1bc04c72017-10-29 20:14:08 +01003118 static THREAD_LOCAL char val[32];
Willy Tarreauacf95772010-06-14 19:09:21 +02003119 int i;
3120
3121 if (!ptr || !*ptr)
3122 return "end of line";
3123 val[0] = '\'';
Willy Tarreaude2dd6b2013-01-24 02:14:42 +01003124 for (i = 1; i < sizeof(val) - 2 && *ptr; i++)
Willy Tarreauacf95772010-06-14 19:09:21 +02003125 val[i] = *ptr++;
3126 val[i++] = '\'';
3127 val[i] = '\0';
3128 return val;
3129}
3130
Willy Tarreau5b180202010-07-18 10:40:48 +02003131/* returns an operator among STD_OP_* for string <str> or < 0 if unknown */
3132int get_std_op(const char *str)
3133{
3134 int ret = -1;
3135
3136 if (*str == 'e' && str[1] == 'q')
3137 ret = STD_OP_EQ;
3138 else if (*str == 'n' && str[1] == 'e')
3139 ret = STD_OP_NE;
3140 else if (*str == 'l') {
3141 if (str[1] == 'e') ret = STD_OP_LE;
3142 else if (str[1] == 't') ret = STD_OP_LT;
3143 }
3144 else if (*str == 'g') {
3145 if (str[1] == 'e') ret = STD_OP_GE;
3146 else if (str[1] == 't') ret = STD_OP_GT;
3147 }
3148
3149 if (ret == -1 || str[2] != '\0')
3150 return -1;
3151 return ret;
3152}
3153
Willy Tarreau4c14eaa2010-11-24 14:01:45 +01003154/* hash a 32-bit integer to another 32-bit integer */
3155unsigned int full_hash(unsigned int a)
3156{
3157 return __full_hash(a);
3158}
3159
Willy Tarreauf3241112019-02-26 09:56:22 +01003160/* Return the bit position in mask <m> of the nth bit set of rank <r>, between
3161 * 0 and LONGBITS-1 included, starting from the left. For example ranks 0,1,2,3
3162 * for mask 0x55 will be 6, 4, 2 and 0 respectively. This algorithm is based on
3163 * a popcount variant and is described here :
3164 * https://graphics.stanford.edu/~seander/bithacks.html
3165 */
3166unsigned int mask_find_rank_bit(unsigned int r, unsigned long m)
3167{
3168 unsigned long a, b, c, d;
3169 unsigned int s;
3170 unsigned int t;
3171
3172 a = m - ((m >> 1) & ~0UL/3);
3173 b = (a & ~0UL/5) + ((a >> 2) & ~0UL/5);
3174 c = (b + (b >> 4)) & ~0UL/0x11;
3175 d = (c + (c >> 8)) & ~0UL/0x101;
3176
3177 r++; // make r be 1..64
3178
3179 t = 0;
3180 s = LONGBITS;
3181 if (s > 32) {
Willy Tarreau9b6be3b2019-03-18 16:31:18 +01003182 unsigned long d2 = (d >> 16) >> 16;
3183 t = d2 + (d2 >> 16);
Willy Tarreauf3241112019-02-26 09:56:22 +01003184 s -= ((t - r) & 256) >> 3; r -= (t & ((t - r) >> 8));
3185 }
3186
3187 t = (d >> (s - 16)) & 0xff;
3188 s -= ((t - r) & 256) >> 4; r -= (t & ((t - r) >> 8));
3189 t = (c >> (s - 8)) & 0xf;
3190 s -= ((t - r) & 256) >> 5; r -= (t & ((t - r) >> 8));
3191 t = (b >> (s - 4)) & 0x7;
3192 s -= ((t - r) & 256) >> 6; r -= (t & ((t - r) >> 8));
3193 t = (a >> (s - 2)) & 0x3;
3194 s -= ((t - r) & 256) >> 7; r -= (t & ((t - r) >> 8));
3195 t = (m >> (s - 1)) & 0x1;
3196 s -= ((t - r) & 256) >> 8;
3197
3198 return s - 1;
3199}
3200
3201/* Same as mask_find_rank_bit() above but makes use of pre-computed bitmaps
3202 * based on <m>, in <a..d>. These ones must be updated whenever <m> changes
3203 * using mask_prep_rank_map() below.
3204 */
3205unsigned int mask_find_rank_bit_fast(unsigned int r, unsigned long m,
3206 unsigned long a, unsigned long b,
3207 unsigned long c, unsigned long d)
3208{
3209 unsigned int s;
3210 unsigned int t;
3211
3212 r++; // make r be 1..64
3213
3214 t = 0;
3215 s = LONGBITS;
3216 if (s > 32) {
Willy Tarreau9b6be3b2019-03-18 16:31:18 +01003217 unsigned long d2 = (d >> 16) >> 16;
3218 t = d2 + (d2 >> 16);
Willy Tarreauf3241112019-02-26 09:56:22 +01003219 s -= ((t - r) & 256) >> 3; r -= (t & ((t - r) >> 8));
3220 }
3221
3222 t = (d >> (s - 16)) & 0xff;
3223 s -= ((t - r) & 256) >> 4; r -= (t & ((t - r) >> 8));
3224 t = (c >> (s - 8)) & 0xf;
3225 s -= ((t - r) & 256) >> 5; r -= (t & ((t - r) >> 8));
3226 t = (b >> (s - 4)) & 0x7;
3227 s -= ((t - r) & 256) >> 6; r -= (t & ((t - r) >> 8));
3228 t = (a >> (s - 2)) & 0x3;
3229 s -= ((t - r) & 256) >> 7; r -= (t & ((t - r) >> 8));
3230 t = (m >> (s - 1)) & 0x1;
3231 s -= ((t - r) & 256) >> 8;
3232
3233 return s - 1;
3234}
3235
3236/* Prepare the bitmaps used by the fast implementation of the find_rank_bit()
3237 * above.
3238 */
3239void mask_prep_rank_map(unsigned long m,
3240 unsigned long *a, unsigned long *b,
3241 unsigned long *c, unsigned long *d)
3242{
3243 *a = m - ((m >> 1) & ~0UL/3);
3244 *b = (*a & ~0UL/5) + ((*a >> 2) & ~0UL/5);
3245 *c = (*b + (*b >> 4)) & ~0UL/0x11;
3246 *d = (*c + (*c >> 8)) & ~0UL/0x101;
3247}
3248
Willy Tarreauc7a8a3c2022-06-21 20:19:54 +02003249/* Returns the position of one bit set in <v>, starting at position <bit>, and
3250 * searching in other halves if not found. This is intended to be used to
3251 * report the position of one bit set among several based on a counter or a
3252 * random generator while preserving a relatively good distribution so that
3253 * values made of holes in the middle do not see one of the bits around the
3254 * hole being returned much more often than the other one. It can be seen as a
3255 * disturbed ffsl() where the initial search starts at bit <bit>. The look up
3256 * is performed in O(logN) time for N bit words, yielding a bit among 64 in
3257 * about 16 cycles. Its usage differs from the rank find function in that the
3258 * bit passed doesn't need to be limited to the value's popcount, making the
3259 * function easier to use for random picking, and twice as fast. Passing value
3260 * 0 for <v> makes no sense and -1 is returned in this case.
3261 */
3262int one_among_mask(unsigned long v, int bit)
3263{
3264 /* note, these masks may be produced by ~0UL/((1UL<<scale)+1) but
3265 * that's more expensive.
3266 */
3267 static const unsigned long halves[] = {
3268 (unsigned long)0x5555555555555555ULL,
3269 (unsigned long)0x3333333333333333ULL,
3270 (unsigned long)0x0F0F0F0F0F0F0F0FULL,
3271 (unsigned long)0x00FF00FF00FF00FFULL,
3272 (unsigned long)0x0000FFFF0000FFFFULL,
3273 (unsigned long)0x00000000FFFFFFFFULL
3274 };
3275 unsigned long halfword = ~0UL;
3276 int scope = 0;
3277 int mirror;
3278 int scale;
3279
3280 if (!v)
3281 return -1;
3282
3283 /* we check if the exact bit is set or if it's present in a mirror
3284 * position based on the current scale we're checking, in which case
3285 * it's returned with its current (or mirrored) value. Otherwise we'll
3286 * make sure there's at least one bit in the half we're in, and will
3287 * scale down to a smaller scope and try again, until we find the
3288 * closest bit.
3289 */
3290 for (scale = (sizeof(long) > 4) ? 5 : 4; scale >= 0; scale--) {
3291 halfword >>= (1UL << scale);
3292 scope |= (1UL << scale);
3293 mirror = bit ^ (1UL << scale);
3294 if (v & ((1UL << bit) | (1UL << mirror)))
3295 return (v & (1UL << bit)) ? bit : mirror;
3296
3297 if (!((v >> (bit & scope)) & halves[scale] & halfword))
3298 bit = mirror;
3299 }
3300 return bit;
3301}
3302
David du Colombier4f92d322011-03-24 11:09:31 +01003303/* Return non-zero if IPv4 address is part of the network,
Willy Tarreaueec1d382016-07-13 11:59:39 +02003304 * otherwise zero. Note that <addr> may not necessarily be aligned
3305 * while the two other ones must.
David du Colombier4f92d322011-03-24 11:09:31 +01003306 */
Willy Tarreaueec1d382016-07-13 11:59:39 +02003307int in_net_ipv4(const void *addr, const struct in_addr *mask, const struct in_addr *net)
David du Colombier4f92d322011-03-24 11:09:31 +01003308{
Willy Tarreaueec1d382016-07-13 11:59:39 +02003309 struct in_addr addr_copy;
3310
3311 memcpy(&addr_copy, addr, sizeof(addr_copy));
3312 return((addr_copy.s_addr & mask->s_addr) == (net->s_addr & mask->s_addr));
David du Colombier4f92d322011-03-24 11:09:31 +01003313}
3314
3315/* Return non-zero if IPv6 address is part of the network,
Willy Tarreaueec1d382016-07-13 11:59:39 +02003316 * otherwise zero. Note that <addr> may not necessarily be aligned
3317 * while the two other ones must.
David du Colombier4f92d322011-03-24 11:09:31 +01003318 */
Willy Tarreaueec1d382016-07-13 11:59:39 +02003319int in_net_ipv6(const void *addr, const struct in6_addr *mask, const struct in6_addr *net)
David du Colombier4f92d322011-03-24 11:09:31 +01003320{
3321 int i;
Willy Tarreaueec1d382016-07-13 11:59:39 +02003322 struct in6_addr addr_copy;
David du Colombier4f92d322011-03-24 11:09:31 +01003323
Willy Tarreaueec1d382016-07-13 11:59:39 +02003324 memcpy(&addr_copy, addr, sizeof(addr_copy));
David du Colombier4f92d322011-03-24 11:09:31 +01003325 for (i = 0; i < sizeof(struct in6_addr) / sizeof(int); i++)
Willy Tarreaueec1d382016-07-13 11:59:39 +02003326 if (((((int *)&addr_copy)[i] & ((int *)mask)[i])) !=
David du Colombier4f92d322011-03-24 11:09:31 +01003327 (((int *)net)[i] & ((int *)mask)[i]))
3328 return 0;
3329 return 1;
3330}
3331
3332/* RFC 4291 prefix */
3333const char rfc4291_pfx[] = { 0x00, 0x00, 0x00, 0x00,
3334 0x00, 0x00, 0x00, 0x00,
3335 0x00, 0x00, 0xFF, 0xFF };
3336
Joseph Herlant32b83272018-11-15 11:58:28 -08003337/* Map IPv4 address on IPv6 address, as specified in RFC 3513.
Thierry FOURNIER4a04dc32013-11-28 16:33:15 +01003338 * Input and output may overlap.
3339 */
David du Colombier4f92d322011-03-24 11:09:31 +01003340void v4tov6(struct in6_addr *sin6_addr, struct in_addr *sin_addr)
3341{
Thierry FOURNIER4a04dc32013-11-28 16:33:15 +01003342 struct in_addr tmp_addr;
3343
3344 tmp_addr.s_addr = sin_addr->s_addr;
David du Colombier4f92d322011-03-24 11:09:31 +01003345 memcpy(sin6_addr->s6_addr, rfc4291_pfx, sizeof(rfc4291_pfx));
Thierry FOURNIER4a04dc32013-11-28 16:33:15 +01003346 memcpy(sin6_addr->s6_addr+12, &tmp_addr.s_addr, 4);
David du Colombier4f92d322011-03-24 11:09:31 +01003347}
3348
Joseph Herlant32b83272018-11-15 11:58:28 -08003349/* Map IPv6 address on IPv4 address, as specified in RFC 3513.
David du Colombier4f92d322011-03-24 11:09:31 +01003350 * Return true if conversion is possible and false otherwise.
3351 */
3352int v6tov4(struct in_addr *sin_addr, struct in6_addr *sin6_addr)
3353{
3354 if (memcmp(sin6_addr->s6_addr, rfc4291_pfx, sizeof(rfc4291_pfx)) == 0) {
3355 memcpy(&(sin_addr->s_addr), &(sin6_addr->s6_addr[12]),
3356 sizeof(struct in_addr));
3357 return 1;
3358 }
3359
3360 return 0;
3361}
3362
Amaury Denoyelle21e611d2022-12-01 17:46:45 +01003363/* compare two struct sockaddr_storage, including port if <check_port> is true,
3364 * and return:
Baptiste Assmann08b24cf2016-01-23 23:39:12 +01003365 * 0 (true) if the addr is the same in both
3366 * 1 (false) if the addr is not the same in both
3367 * -1 (unable) if one of the addr is not AF_INET*
3368 */
Amaury Denoyelle21e611d2022-12-01 17:46:45 +01003369int ipcmp(struct sockaddr_storage *ss1, struct sockaddr_storage *ss2, int check_port)
Baptiste Assmann08b24cf2016-01-23 23:39:12 +01003370{
3371 if ((ss1->ss_family != AF_INET) && (ss1->ss_family != AF_INET6))
3372 return -1;
3373
3374 if ((ss2->ss_family != AF_INET) && (ss2->ss_family != AF_INET6))
3375 return -1;
3376
3377 if (ss1->ss_family != ss2->ss_family)
3378 return 1;
3379
3380 switch (ss1->ss_family) {
3381 case AF_INET:
Amaury Denoyelle21e611d2022-12-01 17:46:45 +01003382 return (memcmp(&((struct sockaddr_in *)ss1)->sin_addr,
Baptiste Assmann08b24cf2016-01-23 23:39:12 +01003383 &((struct sockaddr_in *)ss2)->sin_addr,
Amaury Denoyelle21e611d2022-12-01 17:46:45 +01003384 sizeof(struct in_addr)) != 0) ||
3385 (check_port && get_net_port(ss1) != get_net_port(ss2));
Baptiste Assmann08b24cf2016-01-23 23:39:12 +01003386 case AF_INET6:
Amaury Denoyelle21e611d2022-12-01 17:46:45 +01003387 return (memcmp(&((struct sockaddr_in6 *)ss1)->sin6_addr,
Baptiste Assmann08b24cf2016-01-23 23:39:12 +01003388 &((struct sockaddr_in6 *)ss2)->sin6_addr,
Amaury Denoyelle21e611d2022-12-01 17:46:45 +01003389 sizeof(struct in6_addr)) != 0) ||
3390 (check_port && get_net_port(ss1) != get_net_port(ss2));
Baptiste Assmann08b24cf2016-01-23 23:39:12 +01003391 }
3392
3393 return 1;
3394}
3395
Christopher Faulet9553de72021-02-26 09:12:50 +01003396/* compare a struct sockaddr_storage to a struct net_addr and return :
3397 * 0 (true) if <addr> is matching <net>
3398 * 1 (false) if <addr> is not matching <net>
3399 * -1 (unable) if <addr> or <net> is not AF_INET*
3400 */
3401int ipcmp2net(const struct sockaddr_storage *addr, const struct net_addr *net)
3402{
3403 if ((addr->ss_family != AF_INET) && (addr->ss_family != AF_INET6))
3404 return -1;
3405
3406 if ((net->family != AF_INET) && (net->family != AF_INET6))
3407 return -1;
3408
3409 if (addr->ss_family != net->family)
3410 return 1;
3411
3412 if (addr->ss_family == AF_INET &&
3413 (((struct sockaddr_in *)addr)->sin_addr.s_addr & net->addr.v4.mask.s_addr) == net->addr.v4.ip.s_addr)
3414 return 0;
3415 else {
3416 const struct in6_addr *addr6 = &(((const struct sockaddr_in6*)addr)->sin6_addr);
3417 const struct in6_addr *nip6 = &net->addr.v6.ip;
3418 const struct in6_addr *nmask6 = &net->addr.v6.mask;
3419
3420 if ((read_u32(&addr6->s6_addr[0]) & read_u32(&nmask6->s6_addr[0])) == read_u32(&nip6->s6_addr[0]) &&
3421 (read_u32(&addr6->s6_addr[4]) & read_u32(&nmask6->s6_addr[4])) == read_u32(&nip6->s6_addr[4]) &&
3422 (read_u32(&addr6->s6_addr[8]) & read_u32(&nmask6->s6_addr[8])) == read_u32(&nip6->s6_addr[8]) &&
3423 (read_u32(&addr6->s6_addr[12]) & read_u32(&nmask6->s6_addr[12])) == read_u32(&nip6->s6_addr[12]))
3424 return 0;
3425 }
3426
3427 return 1;
3428}
3429
Baptiste Assmann08396c82016-01-31 00:27:17 +01003430/* copy IP address from <source> into <dest>
Willy Tarreaudc3a9e82016-11-04 18:47:01 +01003431 * The caller must allocate and clear <dest> before calling.
3432 * The source must be in either AF_INET or AF_INET6 family, or the destination
3433 * address will be undefined. If the destination address used to hold a port,
3434 * it is preserved, so that this function can be used to switch to another
3435 * address family with no risk. Returns a pointer to the destination.
Baptiste Assmann08396c82016-01-31 00:27:17 +01003436 */
3437struct sockaddr_storage *ipcpy(struct sockaddr_storage *source, struct sockaddr_storage *dest)
3438{
Willy Tarreaudc3a9e82016-11-04 18:47:01 +01003439 int prev_port;
3440
3441 prev_port = get_net_port(dest);
3442 memset(dest, 0, sizeof(*dest));
Baptiste Assmann08396c82016-01-31 00:27:17 +01003443 dest->ss_family = source->ss_family;
3444
3445 /* copy new addr and apply it */
3446 switch (source->ss_family) {
3447 case AF_INET:
3448 ((struct sockaddr_in *)dest)->sin_addr.s_addr = ((struct sockaddr_in *)source)->sin_addr.s_addr;
Willy Tarreaudc3a9e82016-11-04 18:47:01 +01003449 ((struct sockaddr_in *)dest)->sin_port = prev_port;
Baptiste Assmann08396c82016-01-31 00:27:17 +01003450 break;
3451 case AF_INET6:
3452 memcpy(((struct sockaddr_in6 *)dest)->sin6_addr.s6_addr, ((struct sockaddr_in6 *)source)->sin6_addr.s6_addr, sizeof(struct in6_addr));
Willy Tarreaudc3a9e82016-11-04 18:47:01 +01003453 ((struct sockaddr_in6 *)dest)->sin6_port = prev_port;
Baptiste Assmann08396c82016-01-31 00:27:17 +01003454 break;
3455 }
3456
3457 return dest;
3458}
3459
William Lallemand421f5b52012-02-06 18:15:57 +01003460char *human_time(int t, short hz_div) {
3461 static char rv[sizeof("24855d23h")+1]; // longest of "23h59m" and "59m59s"
3462 char *p = rv;
Willy Tarreau761b3d52014-04-14 14:53:06 +02003463 char *end = rv + sizeof(rv);
William Lallemand421f5b52012-02-06 18:15:57 +01003464 int cnt=2; // print two numbers
3465
3466 if (unlikely(t < 0 || hz_div <= 0)) {
Willy Tarreau761b3d52014-04-14 14:53:06 +02003467 snprintf(p, end - p, "?");
William Lallemand421f5b52012-02-06 18:15:57 +01003468 return rv;
3469 }
3470
3471 if (unlikely(hz_div > 1))
3472 t /= hz_div;
3473
3474 if (t >= DAY) {
Willy Tarreau761b3d52014-04-14 14:53:06 +02003475 p += snprintf(p, end - p, "%dd", t / DAY);
William Lallemand421f5b52012-02-06 18:15:57 +01003476 cnt--;
3477 }
3478
3479 if (cnt && t % DAY / HOUR) {
Willy Tarreau761b3d52014-04-14 14:53:06 +02003480 p += snprintf(p, end - p, "%dh", t % DAY / HOUR);
William Lallemand421f5b52012-02-06 18:15:57 +01003481 cnt--;
3482 }
3483
3484 if (cnt && t % HOUR / MINUTE) {
Willy Tarreau761b3d52014-04-14 14:53:06 +02003485 p += snprintf(p, end - p, "%dm", t % HOUR / MINUTE);
William Lallemand421f5b52012-02-06 18:15:57 +01003486 cnt--;
3487 }
3488
3489 if ((cnt && t % MINUTE) || !t) // also display '0s'
Willy Tarreau761b3d52014-04-14 14:53:06 +02003490 p += snprintf(p, end - p, "%ds", t % MINUTE / SEC);
William Lallemand421f5b52012-02-06 18:15:57 +01003491
3492 return rv;
3493}
3494
3495const char *monthname[12] = {
3496 "Jan", "Feb", "Mar", "Apr", "May", "Jun",
3497 "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
3498};
3499
3500/* date2str_log: write a date in the format :
3501 * sprintf(str, "%02d/%s/%04d:%02d:%02d:%02d.%03d",
3502 * tm.tm_mday, monthname[tm.tm_mon], tm.tm_year+1900,
3503 * tm.tm_hour, tm.tm_min, tm.tm_sec, (int)date.tv_usec/1000);
3504 *
3505 * without using sprintf. return a pointer to the last char written (\0) or
3506 * NULL if there isn't enough space.
3507 */
Willy Tarreauf16cb412018-09-04 19:08:48 +02003508char *date2str_log(char *dst, const struct tm *tm, const struct timeval *date, size_t size)
William Lallemand421f5b52012-02-06 18:15:57 +01003509{
3510
3511 if (size < 25) /* the size is fixed: 24 chars + \0 */
3512 return NULL;
3513
3514 dst = utoa_pad((unsigned int)tm->tm_mday, dst, 3); // day
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003515 if (!dst)
3516 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003517 *dst++ = '/';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003518
William Lallemand421f5b52012-02-06 18:15:57 +01003519 memcpy(dst, monthname[tm->tm_mon], 3); // month
3520 dst += 3;
3521 *dst++ = '/';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003522
William Lallemand421f5b52012-02-06 18:15:57 +01003523 dst = utoa_pad((unsigned int)tm->tm_year+1900, dst, 5); // year
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003524 if (!dst)
3525 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003526 *dst++ = ':';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003527
William Lallemand421f5b52012-02-06 18:15:57 +01003528 dst = utoa_pad((unsigned int)tm->tm_hour, dst, 3); // hour
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003529 if (!dst)
3530 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003531 *dst++ = ':';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003532
William Lallemand421f5b52012-02-06 18:15:57 +01003533 dst = utoa_pad((unsigned int)tm->tm_min, dst, 3); // minutes
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003534 if (!dst)
3535 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003536 *dst++ = ':';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003537
William Lallemand421f5b52012-02-06 18:15:57 +01003538 dst = utoa_pad((unsigned int)tm->tm_sec, dst, 3); // secondes
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003539 if (!dst)
3540 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003541 *dst++ = '.';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003542
Willy Tarreau7d9421d2020-02-29 09:08:02 +01003543 dst = utoa_pad((unsigned int)(date->tv_usec/1000)%1000, dst, 4); // milliseconds
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003544 if (!dst)
3545 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003546 *dst = '\0';
3547
3548 return dst;
3549}
3550
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003551/* Base year used to compute leap years */
3552#define TM_YEAR_BASE 1900
3553
3554/* Return the difference in seconds between two times (leap seconds are ignored).
3555 * Retrieved from glibc 2.18 source code.
3556 */
3557static int my_tm_diff(const struct tm *a, const struct tm *b)
3558{
3559 /* Compute intervening leap days correctly even if year is negative.
3560 * Take care to avoid int overflow in leap day calculations,
3561 * but it's OK to assume that A and B are close to each other.
3562 */
3563 int a4 = (a->tm_year >> 2) + (TM_YEAR_BASE >> 2) - ! (a->tm_year & 3);
3564 int b4 = (b->tm_year >> 2) + (TM_YEAR_BASE >> 2) - ! (b->tm_year & 3);
3565 int a100 = a4 / 25 - (a4 % 25 < 0);
3566 int b100 = b4 / 25 - (b4 % 25 < 0);
3567 int a400 = a100 >> 2;
3568 int b400 = b100 >> 2;
3569 int intervening_leap_days = (a4 - b4) - (a100 - b100) + (a400 - b400);
3570 int years = a->tm_year - b->tm_year;
3571 int days = (365 * years + intervening_leap_days
3572 + (a->tm_yday - b->tm_yday));
3573 return (60 * (60 * (24 * days + (a->tm_hour - b->tm_hour))
3574 + (a->tm_min - b->tm_min))
3575 + (a->tm_sec - b->tm_sec));
3576}
3577
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003578/* Return the GMT offset for a specific local time.
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003579 * Both t and tm must represent the same time.
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003580 * The string returned has the same format as returned by strftime(... "%z", tm).
3581 * Offsets are kept in an internal cache for better performances.
3582 */
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003583const char *get_gmt_offset(time_t t, struct tm *tm)
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003584{
3585 /* Cache offsets from GMT (depending on whether DST is active or not) */
Christopher Faulet1bc04c72017-10-29 20:14:08 +01003586 static THREAD_LOCAL char gmt_offsets[2][5+1] = { "", "" };
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003587
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003588 char *gmt_offset;
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003589 struct tm tm_gmt;
3590 int diff;
3591 int isdst = tm->tm_isdst;
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003592
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003593 /* Pretend DST not active if its status is unknown */
3594 if (isdst < 0)
3595 isdst = 0;
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003596
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003597 /* Fetch the offset and initialize it if needed */
3598 gmt_offset = gmt_offsets[isdst & 0x01];
3599 if (unlikely(!*gmt_offset)) {
3600 get_gmtime(t, &tm_gmt);
3601 diff = my_tm_diff(tm, &tm_gmt);
3602 if (diff < 0) {
3603 diff = -diff;
3604 *gmt_offset = '-';
3605 } else {
3606 *gmt_offset = '+';
3607 }
Willy Tarreaue112c8a2019-10-29 10:16:11 +01003608 diff %= 86400U;
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003609 diff /= 60; /* Convert to minutes */
3610 snprintf(gmt_offset+1, 4+1, "%02d%02d", diff/60, diff%60);
3611 }
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003612
Willy Tarreaue112c8a2019-10-29 10:16:11 +01003613 return gmt_offset;
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003614}
3615
William Lallemand421f5b52012-02-06 18:15:57 +01003616/* gmt2str_log: write a date in the format :
3617 * "%02d/%s/%04d:%02d:%02d:%02d +0000" without using snprintf
3618 * return a pointer to the last char written (\0) or
3619 * NULL if there isn't enough space.
3620 */
3621char *gmt2str_log(char *dst, struct tm *tm, size_t size)
3622{
Yuxans Yao4e25b012012-10-19 10:36:09 +08003623 if (size < 27) /* the size is fixed: 26 chars + \0 */
William Lallemand421f5b52012-02-06 18:15:57 +01003624 return NULL;
3625
3626 dst = utoa_pad((unsigned int)tm->tm_mday, dst, 3); // day
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003627 if (!dst)
3628 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003629 *dst++ = '/';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003630
William Lallemand421f5b52012-02-06 18:15:57 +01003631 memcpy(dst, monthname[tm->tm_mon], 3); // month
3632 dst += 3;
3633 *dst++ = '/';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003634
William Lallemand421f5b52012-02-06 18:15:57 +01003635 dst = utoa_pad((unsigned int)tm->tm_year+1900, dst, 5); // year
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003636 if (!dst)
3637 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003638 *dst++ = ':';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003639
William Lallemand421f5b52012-02-06 18:15:57 +01003640 dst = utoa_pad((unsigned int)tm->tm_hour, dst, 3); // hour
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003641 if (!dst)
3642 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003643 *dst++ = ':';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003644
William Lallemand421f5b52012-02-06 18:15:57 +01003645 dst = utoa_pad((unsigned int)tm->tm_min, dst, 3); // minutes
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003646 if (!dst)
3647 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003648 *dst++ = ':';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003649
William Lallemand421f5b52012-02-06 18:15:57 +01003650 dst = utoa_pad((unsigned int)tm->tm_sec, dst, 3); // secondes
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003651 if (!dst)
3652 return NULL;
William Lallemand421f5b52012-02-06 18:15:57 +01003653 *dst++ = ' ';
3654 *dst++ = '+';
3655 *dst++ = '0';
3656 *dst++ = '0';
3657 *dst++ = '0';
3658 *dst++ = '0';
3659 *dst = '\0';
3660
3661 return dst;
3662}
3663
Yuxans Yao4e25b012012-10-19 10:36:09 +08003664/* localdate2str_log: write a date in the format :
3665 * "%02d/%s/%04d:%02d:%02d:%02d +0000(local timezone)" without using snprintf
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003666 * Both t and tm must represent the same time.
3667 * return a pointer to the last char written (\0) or
3668 * NULL if there isn't enough space.
Yuxans Yao4e25b012012-10-19 10:36:09 +08003669 */
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003670char *localdate2str_log(char *dst, time_t t, struct tm *tm, size_t size)
Yuxans Yao4e25b012012-10-19 10:36:09 +08003671{
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003672 const char *gmt_offset;
Yuxans Yao4e25b012012-10-19 10:36:09 +08003673 if (size < 27) /* the size is fixed: 26 chars + \0 */
3674 return NULL;
3675
Benoit GARNIERe2e5bde2016-03-27 03:04:16 +02003676 gmt_offset = get_gmt_offset(t, tm);
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003677
Yuxans Yao4e25b012012-10-19 10:36:09 +08003678 dst = utoa_pad((unsigned int)tm->tm_mday, dst, 3); // day
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003679 if (!dst)
3680 return NULL;
Yuxans Yao4e25b012012-10-19 10:36:09 +08003681 *dst++ = '/';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003682
Yuxans Yao4e25b012012-10-19 10:36:09 +08003683 memcpy(dst, monthname[tm->tm_mon], 3); // month
3684 dst += 3;
3685 *dst++ = '/';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003686
Yuxans Yao4e25b012012-10-19 10:36:09 +08003687 dst = utoa_pad((unsigned int)tm->tm_year+1900, dst, 5); // year
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003688 if (!dst)
3689 return NULL;
Yuxans Yao4e25b012012-10-19 10:36:09 +08003690 *dst++ = ':';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003691
Yuxans Yao4e25b012012-10-19 10:36:09 +08003692 dst = utoa_pad((unsigned int)tm->tm_hour, dst, 3); // hour
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003693 if (!dst)
3694 return NULL;
Yuxans Yao4e25b012012-10-19 10:36:09 +08003695 *dst++ = ':';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003696
Yuxans Yao4e25b012012-10-19 10:36:09 +08003697 dst = utoa_pad((unsigned int)tm->tm_min, dst, 3); // minutes
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003698 if (!dst)
3699 return NULL;
Yuxans Yao4e25b012012-10-19 10:36:09 +08003700 *dst++ = ':';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003701
Yuxans Yao4e25b012012-10-19 10:36:09 +08003702 dst = utoa_pad((unsigned int)tm->tm_sec, dst, 3); // secondes
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003703 if (!dst)
3704 return NULL;
Yuxans Yao4e25b012012-10-19 10:36:09 +08003705 *dst++ = ' ';
Willy Tarreau4eee38a2019-02-12 11:26:29 +01003706
Benoit GARNIERb413c2a2016-03-27 11:08:03 +02003707 memcpy(dst, gmt_offset, 5); // Offset from local time to GMT
Yuxans Yao4e25b012012-10-19 10:36:09 +08003708 dst += 5;
3709 *dst = '\0';
3710
3711 return dst;
3712}
3713
Willy Tarreaucb1949b2017-07-19 19:05:29 +02003714/* Returns the number of seconds since 01/01/1970 0:0:0 GMT for GMT date <tm>.
3715 * It is meant as a portable replacement for timegm() for use with valid inputs.
3716 * Returns undefined results for invalid dates (eg: months out of range 0..11).
3717 */
3718time_t my_timegm(const struct tm *tm)
3719{
3720 /* Each month has 28, 29, 30 or 31 days, or 28+N. The date in the year
3721 * is thus (current month - 1)*28 + cumulated_N[month] to count the
3722 * sum of the extra N days for elapsed months. The sum of all these N
3723 * days doesn't exceed 30 for a complete year (366-12*28) so it fits
3724 * in a 5-bit word. This means that with 60 bits we can represent a
3725 * matrix of all these values at once, which is fast and efficient to
3726 * access. The extra February day for leap years is not counted here.
3727 *
3728 * Jan : none = 0 (0)
3729 * Feb : Jan = 3 (3)
3730 * Mar : Jan..Feb = 3 (3 + 0)
3731 * Apr : Jan..Mar = 6 (3 + 0 + 3)
3732 * May : Jan..Apr = 8 (3 + 0 + 3 + 2)
3733 * Jun : Jan..May = 11 (3 + 0 + 3 + 2 + 3)
3734 * Jul : Jan..Jun = 13 (3 + 0 + 3 + 2 + 3 + 2)
3735 * Aug : Jan..Jul = 16 (3 + 0 + 3 + 2 + 3 + 2 + 3)
3736 * Sep : Jan..Aug = 19 (3 + 0 + 3 + 2 + 3 + 2 + 3 + 3)
3737 * Oct : Jan..Sep = 21 (3 + 0 + 3 + 2 + 3 + 2 + 3 + 3 + 2)
3738 * Nov : Jan..Oct = 24 (3 + 0 + 3 + 2 + 3 + 2 + 3 + 3 + 2 + 3)
3739 * Dec : Jan..Nov = 26 (3 + 0 + 3 + 2 + 3 + 2 + 3 + 3 + 2 + 3 + 2)
3740 */
3741 uint64_t extra =
3742 ( 0ULL << 0*5) + ( 3ULL << 1*5) + ( 3ULL << 2*5) + /* Jan, Feb, Mar, */
3743 ( 6ULL << 3*5) + ( 8ULL << 4*5) + (11ULL << 5*5) + /* Apr, May, Jun, */
3744 (13ULL << 6*5) + (16ULL << 7*5) + (19ULL << 8*5) + /* Jul, Aug, Sep, */
3745 (21ULL << 9*5) + (24ULL << 10*5) + (26ULL << 11*5); /* Oct, Nov, Dec, */
3746
3747 unsigned int y = tm->tm_year + 1900;
3748 unsigned int m = tm->tm_mon;
3749 unsigned long days = 0;
3750
3751 /* days since 1/1/1970 for full years */
3752 days += days_since_zero(y) - days_since_zero(1970);
3753
3754 /* days for full months in the current year */
3755 days += 28 * m + ((extra >> (m * 5)) & 0x1f);
3756
3757 /* count + 1 after March for leap years. A leap year is a year multiple
3758 * of 4, unless it's multiple of 100 without being multiple of 400. 2000
3759 * is leap, 1900 isn't, 1904 is.
3760 */
3761 if ((m > 1) && !(y & 3) && ((y % 100) || !(y % 400)))
3762 days++;
3763
3764 days += tm->tm_mday - 1;
3765 return days * 86400ULL + tm->tm_hour * 3600 + tm->tm_min * 60 + tm->tm_sec;
3766}
3767
Thierry Fournier93127942016-01-20 18:49:45 +01003768/* This function check a char. It returns true and updates
3769 * <date> and <len> pointer to the new position if the
3770 * character is found.
3771 */
3772static inline int parse_expect_char(const char **date, int *len, char c)
3773{
3774 if (*len < 1 || **date != c)
3775 return 0;
3776 (*len)--;
3777 (*date)++;
3778 return 1;
3779}
3780
3781/* This function expects a string <str> of len <l>. It return true and updates.
3782 * <date> and <len> if the string matches, otherwise, it returns false.
3783 */
3784static inline int parse_strcmp(const char **date, int *len, char *str, int l)
3785{
3786 if (*len < l || strncmp(*date, str, l) != 0)
3787 return 0;
3788 (*len) -= l;
3789 (*date) += l;
3790 return 1;
3791}
3792
3793/* This macro converts 3 chars name in integer. */
3794#define STR2I3(__a, __b, __c) ((__a) * 65536 + (__b) * 256 + (__c))
3795
3796/* day-name = %x4D.6F.6E ; "Mon", case-sensitive
3797 * / %x54.75.65 ; "Tue", case-sensitive
3798 * / %x57.65.64 ; "Wed", case-sensitive
3799 * / %x54.68.75 ; "Thu", case-sensitive
3800 * / %x46.72.69 ; "Fri", case-sensitive
3801 * / %x53.61.74 ; "Sat", case-sensitive
3802 * / %x53.75.6E ; "Sun", case-sensitive
3803 *
3804 * This array must be alphabetically sorted
3805 */
3806static inline int parse_http_dayname(const char **date, int *len, struct tm *tm)
3807{
3808 if (*len < 3)
3809 return 0;
3810 switch (STR2I3((*date)[0], (*date)[1], (*date)[2])) {
3811 case STR2I3('M','o','n'): tm->tm_wday = 1; break;
3812 case STR2I3('T','u','e'): tm->tm_wday = 2; break;
3813 case STR2I3('W','e','d'): tm->tm_wday = 3; break;
3814 case STR2I3('T','h','u'): tm->tm_wday = 4; break;
3815 case STR2I3('F','r','i'): tm->tm_wday = 5; break;
3816 case STR2I3('S','a','t'): tm->tm_wday = 6; break;
3817 case STR2I3('S','u','n'): tm->tm_wday = 7; break;
3818 default: return 0;
3819 }
3820 *len -= 3;
3821 *date += 3;
3822 return 1;
3823}
3824
3825/* month = %x4A.61.6E ; "Jan", case-sensitive
3826 * / %x46.65.62 ; "Feb", case-sensitive
3827 * / %x4D.61.72 ; "Mar", case-sensitive
3828 * / %x41.70.72 ; "Apr", case-sensitive
3829 * / %x4D.61.79 ; "May", case-sensitive
3830 * / %x4A.75.6E ; "Jun", case-sensitive
3831 * / %x4A.75.6C ; "Jul", case-sensitive
3832 * / %x41.75.67 ; "Aug", case-sensitive
3833 * / %x53.65.70 ; "Sep", case-sensitive
3834 * / %x4F.63.74 ; "Oct", case-sensitive
3835 * / %x4E.6F.76 ; "Nov", case-sensitive
3836 * / %x44.65.63 ; "Dec", case-sensitive
3837 *
3838 * This array must be alphabetically sorted
3839 */
3840static inline int parse_http_monthname(const char **date, int *len, struct tm *tm)
3841{
3842 if (*len < 3)
3843 return 0;
3844 switch (STR2I3((*date)[0], (*date)[1], (*date)[2])) {
3845 case STR2I3('J','a','n'): tm->tm_mon = 0; break;
3846 case STR2I3('F','e','b'): tm->tm_mon = 1; break;
3847 case STR2I3('M','a','r'): tm->tm_mon = 2; break;
3848 case STR2I3('A','p','r'): tm->tm_mon = 3; break;
3849 case STR2I3('M','a','y'): tm->tm_mon = 4; break;
3850 case STR2I3('J','u','n'): tm->tm_mon = 5; break;
3851 case STR2I3('J','u','l'): tm->tm_mon = 6; break;
3852 case STR2I3('A','u','g'): tm->tm_mon = 7; break;
3853 case STR2I3('S','e','p'): tm->tm_mon = 8; break;
3854 case STR2I3('O','c','t'): tm->tm_mon = 9; break;
3855 case STR2I3('N','o','v'): tm->tm_mon = 10; break;
3856 case STR2I3('D','e','c'): tm->tm_mon = 11; break;
3857 default: return 0;
3858 }
3859 *len -= 3;
3860 *date += 3;
3861 return 1;
3862}
3863
3864/* day-name-l = %x4D.6F.6E.64.61.79 ; "Monday", case-sensitive
3865 * / %x54.75.65.73.64.61.79 ; "Tuesday", case-sensitive
3866 * / %x57.65.64.6E.65.73.64.61.79 ; "Wednesday", case-sensitive
3867 * / %x54.68.75.72.73.64.61.79 ; "Thursday", case-sensitive
3868 * / %x46.72.69.64.61.79 ; "Friday", case-sensitive
3869 * / %x53.61.74.75.72.64.61.79 ; "Saturday", case-sensitive
3870 * / %x53.75.6E.64.61.79 ; "Sunday", case-sensitive
3871 *
3872 * This array must be alphabetically sorted
3873 */
3874static inline int parse_http_ldayname(const char **date, int *len, struct tm *tm)
3875{
3876 if (*len < 6) /* Minimum length. */
3877 return 0;
3878 switch (STR2I3((*date)[0], (*date)[1], (*date)[2])) {
3879 case STR2I3('M','o','n'):
3880 RET0_UNLESS(parse_strcmp(date, len, "Monday", 6));
3881 tm->tm_wday = 1;
3882 return 1;
3883 case STR2I3('T','u','e'):
3884 RET0_UNLESS(parse_strcmp(date, len, "Tuesday", 7));
3885 tm->tm_wday = 2;
3886 return 1;
3887 case STR2I3('W','e','d'):
3888 RET0_UNLESS(parse_strcmp(date, len, "Wednesday", 9));
3889 tm->tm_wday = 3;
3890 return 1;
3891 case STR2I3('T','h','u'):
3892 RET0_UNLESS(parse_strcmp(date, len, "Thursday", 8));
3893 tm->tm_wday = 4;
3894 return 1;
3895 case STR2I3('F','r','i'):
3896 RET0_UNLESS(parse_strcmp(date, len, "Friday", 6));
3897 tm->tm_wday = 5;
3898 return 1;
3899 case STR2I3('S','a','t'):
3900 RET0_UNLESS(parse_strcmp(date, len, "Saturday", 8));
3901 tm->tm_wday = 6;
3902 return 1;
3903 case STR2I3('S','u','n'):
3904 RET0_UNLESS(parse_strcmp(date, len, "Sunday", 6));
3905 tm->tm_wday = 7;
3906 return 1;
3907 }
3908 return 0;
3909}
3910
3911/* This function parses exactly 1 digit and returns the numeric value in "digit". */
3912static inline int parse_digit(const char **date, int *len, int *digit)
3913{
3914 if (*len < 1 || **date < '0' || **date > '9')
3915 return 0;
3916 *digit = (**date - '0');
3917 (*date)++;
3918 (*len)--;
3919 return 1;
3920}
3921
3922/* This function parses exactly 2 digits and returns the numeric value in "digit". */
3923static inline int parse_2digit(const char **date, int *len, int *digit)
3924{
3925 int value;
3926
3927 RET0_UNLESS(parse_digit(date, len, &value));
3928 (*digit) = value * 10;
3929 RET0_UNLESS(parse_digit(date, len, &value));
3930 (*digit) += value;
3931
3932 return 1;
3933}
3934
3935/* This function parses exactly 4 digits and returns the numeric value in "digit". */
3936static inline int parse_4digit(const char **date, int *len, int *digit)
3937{
3938 int value;
3939
3940 RET0_UNLESS(parse_digit(date, len, &value));
3941 (*digit) = value * 1000;
3942
3943 RET0_UNLESS(parse_digit(date, len, &value));
3944 (*digit) += value * 100;
3945
3946 RET0_UNLESS(parse_digit(date, len, &value));
3947 (*digit) += value * 10;
3948
3949 RET0_UNLESS(parse_digit(date, len, &value));
3950 (*digit) += value;
3951
3952 return 1;
3953}
3954
3955/* time-of-day = hour ":" minute ":" second
3956 * ; 00:00:00 - 23:59:60 (leap second)
3957 *
3958 * hour = 2DIGIT
3959 * minute = 2DIGIT
3960 * second = 2DIGIT
3961 */
3962static inline int parse_http_time(const char **date, int *len, struct tm *tm)
3963{
3964 RET0_UNLESS(parse_2digit(date, len, &tm->tm_hour)); /* hour 2DIGIT */
3965 RET0_UNLESS(parse_expect_char(date, len, ':')); /* expect ":" */
3966 RET0_UNLESS(parse_2digit(date, len, &tm->tm_min)); /* min 2DIGIT */
3967 RET0_UNLESS(parse_expect_char(date, len, ':')); /* expect ":" */
3968 RET0_UNLESS(parse_2digit(date, len, &tm->tm_sec)); /* sec 2DIGIT */
3969 return 1;
3970}
3971
3972/* From RFC7231
3973 * https://tools.ietf.org/html/rfc7231#section-7.1.1.1
3974 *
3975 * IMF-fixdate = day-name "," SP date1 SP time-of-day SP GMT
3976 * ; fixed length/zone/capitalization subset of the format
3977 * ; see Section 3.3 of [RFC5322]
3978 *
3979 *
3980 * date1 = day SP month SP year
3981 * ; e.g., 02 Jun 1982
3982 *
3983 * day = 2DIGIT
3984 * year = 4DIGIT
3985 *
3986 * GMT = %x47.4D.54 ; "GMT", case-sensitive
3987 *
3988 * time-of-day = hour ":" minute ":" second
3989 * ; 00:00:00 - 23:59:60 (leap second)
3990 *
3991 * hour = 2DIGIT
3992 * minute = 2DIGIT
3993 * second = 2DIGIT
3994 *
3995 * DIGIT = decimal 0-9
3996 */
3997int parse_imf_date(const char *date, int len, struct tm *tm)
3998{
David Carlier327298c2016-11-20 10:42:38 +00003999 /* tm_gmtoff, if present, ought to be zero'ed */
4000 memset(tm, 0, sizeof(*tm));
4001
Thierry Fournier93127942016-01-20 18:49:45 +01004002 RET0_UNLESS(parse_http_dayname(&date, &len, tm)); /* day-name */
4003 RET0_UNLESS(parse_expect_char(&date, &len, ',')); /* expect "," */
4004 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4005 RET0_UNLESS(parse_2digit(&date, &len, &tm->tm_mday)); /* day 2DIGIT */
4006 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4007 RET0_UNLESS(parse_http_monthname(&date, &len, tm)); /* Month */
4008 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4009 RET0_UNLESS(parse_4digit(&date, &len, &tm->tm_year)); /* year = 4DIGIT */
4010 tm->tm_year -= 1900;
4011 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4012 RET0_UNLESS(parse_http_time(&date, &len, tm)); /* Parse time. */
4013 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4014 RET0_UNLESS(parse_strcmp(&date, &len, "GMT", 3)); /* GMT = %x47.4D.54 ; "GMT", case-sensitive */
4015 tm->tm_isdst = -1;
Thierry Fournier93127942016-01-20 18:49:45 +01004016 return 1;
4017}
4018
4019/* From RFC7231
4020 * https://tools.ietf.org/html/rfc7231#section-7.1.1.1
4021 *
4022 * rfc850-date = day-name-l "," SP date2 SP time-of-day SP GMT
4023 * date2 = day "-" month "-" 2DIGIT
4024 * ; e.g., 02-Jun-82
4025 *
4026 * day = 2DIGIT
4027 */
4028int parse_rfc850_date(const char *date, int len, struct tm *tm)
4029{
4030 int year;
4031
David Carlier327298c2016-11-20 10:42:38 +00004032 /* tm_gmtoff, if present, ought to be zero'ed */
4033 memset(tm, 0, sizeof(*tm));
4034
Thierry Fournier93127942016-01-20 18:49:45 +01004035 RET0_UNLESS(parse_http_ldayname(&date, &len, tm)); /* Read the day name */
4036 RET0_UNLESS(parse_expect_char(&date, &len, ',')); /* expect "," */
4037 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4038 RET0_UNLESS(parse_2digit(&date, &len, &tm->tm_mday)); /* day 2DIGIT */
4039 RET0_UNLESS(parse_expect_char(&date, &len, '-')); /* expect "-" */
4040 RET0_UNLESS(parse_http_monthname(&date, &len, tm)); /* Month */
4041 RET0_UNLESS(parse_expect_char(&date, &len, '-')); /* expect "-" */
4042
4043 /* year = 2DIGIT
4044 *
4045 * Recipients of a timestamp value in rfc850-(*date) format, which uses a
4046 * two-digit year, MUST interpret a timestamp that appears to be more
4047 * than 50 years in the future as representing the most recent year in
4048 * the past that had the same last two digits.
4049 */
4050 RET0_UNLESS(parse_2digit(&date, &len, &tm->tm_year));
4051
4052 /* expect SP */
4053 if (!parse_expect_char(&date, &len, ' ')) {
4054 /* Maybe we have the date with 4 digits. */
4055 RET0_UNLESS(parse_2digit(&date, &len, &year));
4056 tm->tm_year = (tm->tm_year * 100 + year) - 1900;
4057 /* expect SP */
4058 RET0_UNLESS(parse_expect_char(&date, &len, ' '));
4059 } else {
4060 /* I fix 60 as pivot: >60: +1900, <60: +2000. Note that the
4061 * tm_year is the number of year since 1900, so for +1900, we
4062 * do nothing, and for +2000, we add 100.
4063 */
4064 if (tm->tm_year <= 60)
4065 tm->tm_year += 100;
4066 }
4067
4068 RET0_UNLESS(parse_http_time(&date, &len, tm)); /* Parse time. */
4069 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4070 RET0_UNLESS(parse_strcmp(&date, &len, "GMT", 3)); /* GMT = %x47.4D.54 ; "GMT", case-sensitive */
4071 tm->tm_isdst = -1;
Thierry Fournier93127942016-01-20 18:49:45 +01004072
4073 return 1;
4074}
4075
4076/* From RFC7231
4077 * https://tools.ietf.org/html/rfc7231#section-7.1.1.1
4078 *
4079 * asctime-date = day-name SP date3 SP time-of-day SP year
4080 * date3 = month SP ( 2DIGIT / ( SP 1DIGIT ))
4081 * ; e.g., Jun 2
4082 *
4083 * HTTP-date is case sensitive. A sender MUST NOT generate additional
4084 * whitespace in an HTTP-date beyond that specifically included as SP in
4085 * the grammar.
4086 */
4087int parse_asctime_date(const char *date, int len, struct tm *tm)
4088{
David Carlier327298c2016-11-20 10:42:38 +00004089 /* tm_gmtoff, if present, ought to be zero'ed */
4090 memset(tm, 0, sizeof(*tm));
4091
Thierry Fournier93127942016-01-20 18:49:45 +01004092 RET0_UNLESS(parse_http_dayname(&date, &len, tm)); /* day-name */
4093 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4094 RET0_UNLESS(parse_http_monthname(&date, &len, tm)); /* expect month */
4095 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4096
4097 /* expect SP and 1DIGIT or 2DIGIT */
4098 if (parse_expect_char(&date, &len, ' '))
4099 RET0_UNLESS(parse_digit(&date, &len, &tm->tm_mday));
4100 else
4101 RET0_UNLESS(parse_2digit(&date, &len, &tm->tm_mday));
4102
4103 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4104 RET0_UNLESS(parse_http_time(&date, &len, tm)); /* Parse time. */
4105 RET0_UNLESS(parse_expect_char(&date, &len, ' ')); /* expect SP */
4106 RET0_UNLESS(parse_4digit(&date, &len, &tm->tm_year)); /* year = 4DIGIT */
4107 tm->tm_year -= 1900;
4108 tm->tm_isdst = -1;
Thierry Fournier93127942016-01-20 18:49:45 +01004109 return 1;
4110}
4111
4112/* From RFC7231
4113 * https://tools.ietf.org/html/rfc7231#section-7.1.1.1
4114 *
4115 * HTTP-date = IMF-fixdate / obs-date
4116 * obs-date = rfc850-date / asctime-date
4117 *
4118 * parses an HTTP date in the RFC format and is accepted
4119 * alternatives. <date> is the strinf containing the date,
4120 * len is the len of the string. <tm> is filled with the
4121 * parsed time. We must considers this time as GMT.
4122 */
4123int parse_http_date(const char *date, int len, struct tm *tm)
4124{
4125 if (parse_imf_date(date, len, tm))
4126 return 1;
4127
4128 if (parse_rfc850_date(date, len, tm))
4129 return 1;
4130
4131 if (parse_asctime_date(date, len, tm))
4132 return 1;
4133
4134 return 0;
4135}
4136
Willy Tarreau4deeb102021-01-29 10:47:52 +01004137/* print the time <ns> in a short form (exactly 7 chars) at the end of buffer
4138 * <out>. "-" is printed if the value is zero, "inf" if larger than 1000 years.
4139 * It returns the new buffer length, or 0 if it doesn't fit. The value will be
4140 * surrounded by <pfx> and <sfx> respectively if not NULL.
4141 */
4142int print_time_short(struct buffer *out, const char *pfx, uint64_t ns, const char *sfx)
4143{
4144 double val = ns; // 52 bits of mantissa keep ns accuracy over 52 days
4145 const char *unit;
4146
4147 if (!pfx)
4148 pfx = "";
4149 if (!sfx)
4150 sfx = "";
4151
4152 do {
4153 unit = " - "; if (val <= 0.0) break;
4154 unit = "ns"; if (val < 1000.0) break;
4155 unit = "us"; val /= 1000.0; if (val < 1000.0) break;
4156 unit = "ms"; val /= 1000.0; if (val < 1000.0) break;
4157 unit = "s "; val /= 1000.0; if (val < 60.0) break;
4158 unit = "m "; val /= 60.0; if (val < 60.0) break;
4159 unit = "h "; val /= 60.0; if (val < 24.0) break;
4160 unit = "d "; val /= 24.0; if (val < 365.0) break;
4161 unit = "yr"; val /= 365.0; if (val < 1000.0) break;
4162 unit = " inf "; val = 0.0; break;
4163 } while (0);
4164
4165 if (val <= 0.0)
4166 return chunk_appendf(out, "%s%7s%s", pfx, unit, sfx);
4167 else if (val < 10.0)
4168 return chunk_appendf(out, "%s%1.3f%s%s", pfx, val, unit, sfx);
4169 else if (val < 100.0)
4170 return chunk_appendf(out, "%s%2.2f%s%s", pfx, val, unit, sfx);
4171 else
4172 return chunk_appendf(out, "%s%3.1f%s%s", pfx, val, unit, sfx);
4173}
4174
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004175/* Dynamically allocates a string of the proper length to hold the formatted
4176 * output. NULL is returned on error. The caller is responsible for freeing the
4177 * memory area using free(). The resulting string is returned in <out> if the
4178 * pointer is not NULL. A previous version of <out> might be used to build the
4179 * new string, and it will be freed before returning if it is not NULL, which
4180 * makes it possible to build complex strings from iterative calls without
4181 * having to care about freeing intermediate values, as in the example below :
4182 *
4183 * memprintf(&err, "invalid argument: '%s'", arg);
4184 * ...
4185 * memprintf(&err, "parser said : <%s>\n", *err);
4186 * ...
4187 * free(*err);
4188 *
4189 * This means that <err> must be initialized to NULL before first invocation.
4190 * The return value also holds the allocated string, which eases error checking
4191 * and immediate consumption. If the output pointer is not used, NULL must be
Willy Tarreaueb6cead2012-09-20 19:43:14 +02004192 * passed instead and it will be ignored. The returned message will then also
4193 * be NULL so that the caller does not have to bother with freeing anything.
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004194 *
4195 * It is also convenient to use it without any free except the last one :
4196 * err = NULL;
4197 * if (!fct1(err)) report(*err);
4198 * if (!fct2(err)) report(*err);
4199 * if (!fct3(err)) report(*err);
4200 * free(*err);
Christopher Faulet93a518f2017-10-24 11:25:33 +02004201 *
4202 * memprintf relies on memvprintf. This last version can be called from any
4203 * function with variadic arguments.
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004204 */
Christopher Faulet93a518f2017-10-24 11:25:33 +02004205char *memvprintf(char **out, const char *format, va_list orig_args)
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004206{
4207 va_list args;
4208 char *ret = NULL;
4209 int allocated = 0;
4210 int needed = 0;
4211
Willy Tarreaueb6cead2012-09-20 19:43:14 +02004212 if (!out)
4213 return NULL;
4214
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004215 do {
Willy Tarreaue0609f52019-03-29 19:13:23 +01004216 char buf1;
4217
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004218 /* vsnprintf() will return the required length even when the
4219 * target buffer is NULL. We do this in a loop just in case
4220 * intermediate evaluations get wrong.
4221 */
Christopher Faulet93a518f2017-10-24 11:25:33 +02004222 va_copy(args, orig_args);
Willy Tarreaue0609f52019-03-29 19:13:23 +01004223 needed = vsnprintf(ret ? ret : &buf1, allocated, format, args);
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004224 va_end(args);
Willy Tarreau1b2fed62013-04-01 22:48:54 +02004225 if (needed < allocated) {
4226 /* Note: on Solaris 8, the first iteration always
4227 * returns -1 if allocated is zero, so we force a
4228 * retry.
4229 */
4230 if (!allocated)
4231 needed = 0;
4232 else
4233 break;
4234 }
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004235
Willy Tarreau1b2fed62013-04-01 22:48:54 +02004236 allocated = needed + 1;
Hubert Verstraete831962e2016-06-28 22:44:26 +02004237 ret = my_realloc2(ret, allocated);
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004238 } while (ret);
4239
4240 if (needed < 0) {
4241 /* an error was encountered */
Willy Tarreau61cfdf42021-02-20 10:46:51 +01004242 ha_free(&ret);
Willy Tarreau9a7bea52012-04-27 11:16:50 +02004243 }
4244
4245 if (out) {
4246 free(*out);
4247 *out = ret;
4248 }
4249
4250 return ret;
4251}
William Lallemand421f5b52012-02-06 18:15:57 +01004252
Christopher Faulet93a518f2017-10-24 11:25:33 +02004253char *memprintf(char **out, const char *format, ...)
4254{
4255 va_list args;
4256 char *ret = NULL;
4257
4258 va_start(args, format);
4259 ret = memvprintf(out, format, args);
4260 va_end(args);
4261
4262 return ret;
4263}
4264
Willy Tarreau21c705b2012-09-14 11:40:36 +02004265/* Used to add <level> spaces before each line of <out>, unless there is only one line.
4266 * The input argument is automatically freed and reassigned. The result will have to be
Willy Tarreau70eec382012-10-10 08:56:47 +02004267 * freed by the caller. It also supports being passed a NULL which results in the same
4268 * output.
Willy Tarreau21c705b2012-09-14 11:40:36 +02004269 * Example of use :
4270 * parse(cmd, &err); (callee: memprintf(&err, ...))
4271 * fprintf(stderr, "Parser said: %s\n", indent_error(&err));
4272 * free(err);
4273 */
4274char *indent_msg(char **out, int level)
4275{
4276 char *ret, *in, *p;
4277 int needed = 0;
4278 int lf = 0;
4279 int lastlf = 0;
4280 int len;
4281
Willy Tarreau70eec382012-10-10 08:56:47 +02004282 if (!out || !*out)
4283 return NULL;
4284
Willy Tarreau21c705b2012-09-14 11:40:36 +02004285 in = *out - 1;
4286 while ((in = strchr(in + 1, '\n')) != NULL) {
4287 lastlf = in - *out;
4288 lf++;
4289 }
4290
4291 if (!lf) /* single line, no LF, return it as-is */
4292 return *out;
4293
4294 len = strlen(*out);
4295
4296 if (lf == 1 && lastlf == len - 1) {
4297 /* single line, LF at end, strip it and return as-is */
4298 (*out)[lastlf] = 0;
4299 return *out;
4300 }
4301
4302 /* OK now we have at least one LF, we need to process the whole string
4303 * as a multi-line string. What we'll do :
4304 * - prefix with an LF if there is none
4305 * - add <level> spaces before each line
4306 * This means at most ( 1 + level + (len-lf) + lf*<1+level) ) =
4307 * 1 + level + len + lf * level = 1 + level * (lf + 1) + len.
4308 */
4309
4310 needed = 1 + level * (lf + 1) + len + 1;
4311 p = ret = malloc(needed);
4312 in = *out;
4313
4314 /* skip initial LFs */
4315 while (*in == '\n')
4316 in++;
4317
4318 /* copy each line, prefixed with LF and <level> spaces, and without the trailing LF */
4319 while (*in) {
4320 *p++ = '\n';
4321 memset(p, ' ', level);
4322 p += level;
4323 do {
4324 *p++ = *in++;
4325 } while (*in && *in != '\n');
4326 if (*in)
4327 in++;
4328 }
4329 *p = 0;
4330
4331 free(*out);
4332 *out = ret;
4333
4334 return ret;
4335}
4336
Willy Tarreaua2c99112019-08-21 13:17:37 +02004337/* makes a copy of message <in> into <out>, with each line prefixed with <pfx>
4338 * and end of lines replaced with <eol> if not 0. The first line to indent has
4339 * to be indicated in <first> (starts at zero), so that it is possible to skip
4340 * indenting the first line if it has to be appended after an existing message.
4341 * Empty strings are never indented, and NULL strings are considered empty both
4342 * for <in> and <pfx>. It returns non-zero if an EOL was appended as the last
4343 * character, non-zero otherwise.
4344 */
4345int append_prefixed_str(struct buffer *out, const char *in, const char *pfx, char eol, int first)
4346{
4347 int bol, lf;
4348 int pfxlen = pfx ? strlen(pfx) : 0;
4349
4350 if (!in)
4351 return 0;
4352
4353 bol = 1;
4354 lf = 0;
4355 while (*in) {
4356 if (bol && pfxlen) {
4357 if (first > 0)
4358 first--;
4359 else
4360 b_putblk(out, pfx, pfxlen);
4361 bol = 0;
4362 }
4363
4364 lf = (*in == '\n');
4365 bol |= lf;
4366 b_putchr(out, (lf && eol) ? eol : *in);
4367 in++;
4368 }
4369 return lf;
4370}
4371
Willy Tarreau9d22e562019-03-29 18:49:09 +01004372/* removes environment variable <name> from the environment as found in
4373 * environ. This is only provided as an alternative for systems without
4374 * unsetenv() (old Solaris and AIX versions). THIS IS NOT THREAD SAFE.
Ilya Shipitsin856aabc2020-04-16 23:51:34 +05004375 * The principle is to scan environ for each occurrence of variable name
Willy Tarreau9d22e562019-03-29 18:49:09 +01004376 * <name> and to replace the matching pointers with the last pointer of
4377 * the array (since variables are not ordered).
4378 * It always returns 0 (success).
4379 */
4380int my_unsetenv(const char *name)
4381{
4382 extern char **environ;
4383 char **p = environ;
4384 int vars;
4385 int next;
4386 int len;
4387
4388 len = strlen(name);
4389 for (vars = 0; p[vars]; vars++)
4390 ;
4391 next = 0;
4392 while (next < vars) {
4393 if (strncmp(p[next], name, len) != 0 || p[next][len] != '=') {
4394 next++;
4395 continue;
4396 }
4397 if (next < vars - 1)
4398 p[next] = p[vars - 1];
4399 p[--vars] = NULL;
4400 }
4401 return 0;
4402}
4403
Willy Tarreaudad36a32013-03-11 01:20:04 +01004404/* Convert occurrences of environment variables in the input string to their
4405 * corresponding value. A variable is identified as a series of alphanumeric
4406 * characters or underscores following a '$' sign. The <in> string must be
4407 * free()able. NULL returns NULL. The resulting string might be reallocated if
Willy Tarreau9cb2ca42024-05-31 18:52:51 +02004408 * some expansion is made (an NULL will be returned on failure). Variable names
4409 * may also be enclosed into braces if needed (eg: to concatenate alphanum
4410 * characters).
Willy Tarreaudad36a32013-03-11 01:20:04 +01004411 */
4412char *env_expand(char *in)
4413{
4414 char *txt_beg;
4415 char *out;
4416 char *txt_end;
4417 char *var_beg;
4418 char *var_end;
4419 char *value;
4420 char *next;
4421 int out_len;
4422 int val_len;
4423
4424 if (!in)
4425 return in;
4426
4427 value = out = NULL;
4428 out_len = 0;
4429
4430 txt_beg = in;
4431 do {
4432 /* look for next '$' sign in <in> */
4433 for (txt_end = txt_beg; *txt_end && *txt_end != '$'; txt_end++);
4434
4435 if (!*txt_end && !out) /* end and no expansion performed */
4436 return in;
4437
4438 val_len = 0;
4439 next = txt_end;
4440 if (*txt_end == '$') {
4441 char save;
4442
4443 var_beg = txt_end + 1;
4444 if (*var_beg == '{')
4445 var_beg++;
4446
4447 var_end = var_beg;
Willy Tarreau90807112020-02-25 08:16:33 +01004448 while (isalnum((unsigned char)*var_end) || *var_end == '_') {
Willy Tarreaudad36a32013-03-11 01:20:04 +01004449 var_end++;
4450 }
4451
4452 next = var_end;
4453 if (*var_end == '}' && (var_beg > txt_end + 1))
4454 next++;
4455
4456 /* get value of the variable name at this location */
4457 save = *var_end;
4458 *var_end = '\0';
4459 value = getenv(var_beg);
4460 *var_end = save;
4461 val_len = value ? strlen(value) : 0;
4462 }
4463
Hubert Verstraete831962e2016-06-28 22:44:26 +02004464 out = my_realloc2(out, out_len + (txt_end - txt_beg) + val_len + 1);
Willy Tarreau9cb2ca42024-05-31 18:52:51 +02004465 if (!out)
4466 goto leave;
4467
Willy Tarreaudad36a32013-03-11 01:20:04 +01004468 if (txt_end > txt_beg) {
4469 memcpy(out + out_len, txt_beg, txt_end - txt_beg);
4470 out_len += txt_end - txt_beg;
4471 }
4472 if (val_len) {
4473 memcpy(out + out_len, value, val_len);
4474 out_len += val_len;
4475 }
4476 out[out_len] = 0;
4477 txt_beg = next;
4478 } while (*txt_beg);
4479
4480 /* here we know that <out> was allocated and that we don't need <in> anymore */
4481 free(in);
Willy Tarreau9cb2ca42024-05-31 18:52:51 +02004482leave:
Willy Tarreaudad36a32013-03-11 01:20:04 +01004483 return out;
4484}
4485
de Lafond Guillaume88c278f2013-04-15 19:27:10 +02004486
4487/* same as strstr() but case-insensitive and with limit length */
4488const char *strnistr(const char *str1, int len_str1, const char *str2, int len_str2)
4489{
4490 char *pptr, *sptr, *start;
Willy Tarreauc8746532014-05-28 23:05:07 +02004491 unsigned int slen, plen;
4492 unsigned int tmp1, tmp2;
de Lafond Guillaume88c278f2013-04-15 19:27:10 +02004493
4494 if (str1 == NULL || len_str1 == 0) // search pattern into an empty string => search is not found
4495 return NULL;
4496
4497 if (str2 == NULL || len_str2 == 0) // pattern is empty => every str1 match
4498 return str1;
4499
4500 if (len_str1 < len_str2) // pattern is longer than string => search is not found
4501 return NULL;
4502
4503 for (tmp1 = 0, start = (char *)str1, pptr = (char *)str2, slen = len_str1, plen = len_str2; slen >= plen; start++, slen--) {
Willy Tarreauf278eec2020-07-05 21:46:32 +02004504 while (toupper((unsigned char)*start) != toupper((unsigned char)*str2)) {
de Lafond Guillaume88c278f2013-04-15 19:27:10 +02004505 start++;
4506 slen--;
4507 tmp1++;
4508
4509 if (tmp1 >= len_str1)
4510 return NULL;
4511
4512 /* if pattern longer than string */
4513 if (slen < plen)
4514 return NULL;
4515 }
4516
4517 sptr = start;
4518 pptr = (char *)str2;
4519
4520 tmp2 = 0;
Willy Tarreauf278eec2020-07-05 21:46:32 +02004521 while (toupper((unsigned char)*sptr) == toupper((unsigned char)*pptr)) {
de Lafond Guillaume88c278f2013-04-15 19:27:10 +02004522 sptr++;
4523 pptr++;
4524 tmp2++;
4525
4526 if (*pptr == '\0' || tmp2 == len_str2) /* end of pattern found */
4527 return start;
4528 if (*sptr == '\0' || tmp2 == len_str1) /* end of string found and the pattern is not fully found */
4529 return NULL;
4530 }
4531 }
4532 return NULL;
4533}
4534
Willy Tarreau3ff476e2022-03-30 10:02:56 +02004535/* Returns true if s1 < s2 < s3 otherwise zero. Both s1 and s3 may be NULL and
4536 * in this case only non-null strings are compared. This allows to pass initial
4537 * values in iterators and in sort functions.
4538 */
4539int strordered(const char *s1, const char *s2, const char *s3)
4540{
4541 return (!s1 || strcmp(s1, s2) < 0) && (!s3 || strcmp(s2, s3) < 0);
4542}
4543
Thierry FOURNIER317e1c42014-08-12 10:20:47 +02004544/* This function read the next valid utf8 char.
4545 * <s> is the byte srray to be decode, <len> is its length.
4546 * The function returns decoded char encoded like this:
4547 * The 4 msb are the return code (UTF8_CODE_*), the 4 lsb
4548 * are the length read. The decoded character is stored in <c>.
4549 */
4550unsigned char utf8_next(const char *s, int len, unsigned int *c)
4551{
4552 const unsigned char *p = (unsigned char *)s;
4553 int dec;
4554 unsigned char code = UTF8_CODE_OK;
4555
4556 if (len < 1)
4557 return UTF8_CODE_OK;
4558
4559 /* Check the type of UTF8 sequence
4560 *
4561 * 0... .... 0x00 <= x <= 0x7f : 1 byte: ascii char
4562 * 10.. .... 0x80 <= x <= 0xbf : invalid sequence
4563 * 110. .... 0xc0 <= x <= 0xdf : 2 bytes
4564 * 1110 .... 0xe0 <= x <= 0xef : 3 bytes
4565 * 1111 0... 0xf0 <= x <= 0xf7 : 4 bytes
4566 * 1111 10.. 0xf8 <= x <= 0xfb : 5 bytes
4567 * 1111 110. 0xfc <= x <= 0xfd : 6 bytes
4568 * 1111 111. 0xfe <= x <= 0xff : invalid sequence
4569 */
4570 switch (*p) {
4571 case 0x00 ... 0x7f:
4572 *c = *p;
4573 return UTF8_CODE_OK | 1;
4574
4575 case 0x80 ... 0xbf:
4576 *c = *p;
4577 return UTF8_CODE_BADSEQ | 1;
4578
4579 case 0xc0 ... 0xdf:
4580 if (len < 2) {
4581 *c = *p;
4582 return UTF8_CODE_BADSEQ | 1;
4583 }
4584 *c = *p & 0x1f;
4585 dec = 1;
4586 break;
4587
4588 case 0xe0 ... 0xef:
4589 if (len < 3) {
4590 *c = *p;
4591 return UTF8_CODE_BADSEQ | 1;
4592 }
4593 *c = *p & 0x0f;
4594 dec = 2;
4595 break;
4596
4597 case 0xf0 ... 0xf7:
4598 if (len < 4) {
4599 *c = *p;
4600 return UTF8_CODE_BADSEQ | 1;
4601 }
4602 *c = *p & 0x07;
4603 dec = 3;
4604 break;
4605
4606 case 0xf8 ... 0xfb:
4607 if (len < 5) {
4608 *c = *p;
4609 return UTF8_CODE_BADSEQ | 1;
4610 }
4611 *c = *p & 0x03;
4612 dec = 4;
4613 break;
4614
4615 case 0xfc ... 0xfd:
4616 if (len < 6) {
4617 *c = *p;
4618 return UTF8_CODE_BADSEQ | 1;
4619 }
4620 *c = *p & 0x01;
4621 dec = 5;
4622 break;
4623
4624 case 0xfe ... 0xff:
4625 default:
4626 *c = *p;
4627 return UTF8_CODE_BADSEQ | 1;
4628 }
4629
4630 p++;
4631
4632 while (dec > 0) {
4633
4634 /* need 0x10 for the 2 first bits */
4635 if ( ( *p & 0xc0 ) != 0x80 )
4636 return UTF8_CODE_BADSEQ | ((p-(unsigned char *)s)&0xffff);
4637
4638 /* add data at char */
4639 *c = ( *c << 6 ) | ( *p & 0x3f );
4640
4641 dec--;
4642 p++;
4643 }
4644
4645 /* Check ovelong encoding.
4646 * 1 byte : 5 + 6 : 11 : 0x80 ... 0x7ff
4647 * 2 bytes : 4 + 6 + 6 : 16 : 0x800 ... 0xffff
4648 * 3 bytes : 3 + 6 + 6 + 6 : 21 : 0x10000 ... 0x1fffff
4649 */
Thierry FOURNIER9e7ec082015-03-12 19:32:38 +01004650 if (( *c <= 0x7f && (p-(unsigned char *)s) > 1) ||
Thierry FOURNIER317e1c42014-08-12 10:20:47 +02004651 (*c >= 0x80 && *c <= 0x7ff && (p-(unsigned char *)s) > 2) ||
4652 (*c >= 0x800 && *c <= 0xffff && (p-(unsigned char *)s) > 3) ||
4653 (*c >= 0x10000 && *c <= 0x1fffff && (p-(unsigned char *)s) > 4))
4654 code |= UTF8_CODE_OVERLONG;
4655
4656 /* Check invalid UTF8 range. */
4657 if ((*c >= 0xd800 && *c <= 0xdfff) ||
4658 (*c >= 0xfffe && *c <= 0xffff))
4659 code |= UTF8_CODE_INVRANGE;
4660
4661 return code | ((p-(unsigned char *)s)&0x0f);
4662}
4663
Maxime de Roucydc887852016-05-13 23:52:54 +02004664/* append a copy of string <str> (in a wordlist) at the end of the list <li>
4665 * On failure : return 0 and <err> filled with an error message.
4666 * The caller is responsible for freeing the <err> and <str> copy
4667 * memory area using free()
4668 */
4669int list_append_word(struct list *li, const char *str, char **err)
4670{
4671 struct wordlist *wl;
4672
4673 wl = calloc(1, sizeof(*wl));
4674 if (!wl) {
4675 memprintf(err, "out of memory");
4676 goto fail_wl;
4677 }
4678
4679 wl->s = strdup(str);
4680 if (!wl->s) {
4681 memprintf(err, "out of memory");
4682 goto fail_wl_s;
4683 }
4684
Willy Tarreau2b718102021-04-21 07:32:39 +02004685 LIST_APPEND(li, &wl->list);
Maxime de Roucydc887852016-05-13 23:52:54 +02004686
4687 return 1;
4688
4689fail_wl_s:
4690 free(wl->s);
4691fail_wl:
4692 free(wl);
4693 return 0;
4694}
4695
Willy Tarreau37101052019-05-20 16:48:20 +02004696/* indicates if a memory location may safely be read or not. The trick consists
4697 * in performing a harmless syscall using this location as an input and letting
4698 * the operating system report whether it's OK or not. For this we have the
4699 * stat() syscall, which will return EFAULT when the memory location supposed
4700 * to contain the file name is not readable. If it is readable it will then
4701 * either return 0 if the area contains an existing file name, or -1 with
4702 * another code. This must not be abused, and some audit systems might detect
4703 * this as abnormal activity. It's used only for unsafe dumps.
4704 */
4705int may_access(const void *ptr)
4706{
4707 struct stat buf;
4708
4709 if (stat(ptr, &buf) == 0)
4710 return 1;
4711 if (errno == EFAULT)
4712 return 0;
4713 return 1;
4714}
4715
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004716/* print a string of text buffer to <out>. The format is :
4717 * Non-printable chars \t, \n, \r and \e are * encoded in C format.
4718 * Other non-printable chars are encoded "\xHH". Space, '\', and '=' are also escaped.
4719 * Print stopped if null char or <bsize> is reached, or if no more place in the chunk.
4720 */
Willy Tarreau83061a82018-07-13 11:56:34 +02004721int dump_text(struct buffer *out, const char *buf, int bsize)
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004722{
4723 unsigned char c;
Tim Duesterhus18795d42021-08-29 00:58:22 +02004724 size_t ptr = 0;
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004725
Tim Duesterhus18795d42021-08-29 00:58:22 +02004726 while (ptr < bsize && buf[ptr]) {
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004727 c = buf[ptr];
Willy Tarreau90807112020-02-25 08:16:33 +01004728 if (isprint((unsigned char)c) && isascii((unsigned char)c) && c != '\\' && c != ' ' && c != '=') {
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004729 if (out->data > out->size - 1)
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004730 break;
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004731 out->area[out->data++] = c;
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004732 }
4733 else if (c == '\t' || c == '\n' || c == '\r' || c == '\e' || c == '\\' || c == ' ' || c == '=') {
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004734 if (out->data > out->size - 2)
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004735 break;
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004736 out->area[out->data++] = '\\';
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004737 switch (c) {
4738 case ' ': c = ' '; break;
4739 case '\t': c = 't'; break;
4740 case '\n': c = 'n'; break;
4741 case '\r': c = 'r'; break;
4742 case '\e': c = 'e'; break;
4743 case '\\': c = '\\'; break;
4744 case '=': c = '='; break;
4745 }
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004746 out->area[out->data++] = c;
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004747 }
4748 else {
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004749 if (out->data > out->size - 4)
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004750 break;
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004751 out->area[out->data++] = '\\';
4752 out->area[out->data++] = 'x';
4753 out->area[out->data++] = hextab[(c >> 4) & 0xF];
4754 out->area[out->data++] = hextab[c & 0xF];
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004755 }
4756 ptr++;
4757 }
4758
4759 return ptr;
4760}
4761
4762/* print a buffer in hexa.
4763 * Print stopped if <bsize> is reached, or if no more place in the chunk.
4764 */
Willy Tarreau83061a82018-07-13 11:56:34 +02004765int dump_binary(struct buffer *out, const char *buf, int bsize)
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004766{
4767 unsigned char c;
4768 int ptr = 0;
4769
4770 while (ptr < bsize) {
4771 c = buf[ptr];
4772
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004773 if (out->data > out->size - 2)
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004774 break;
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004775 out->area[out->data++] = hextab[(c >> 4) & 0xF];
4776 out->area[out->data++] = hextab[c & 0xF];
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004777
4778 ptr++;
4779 }
4780 return ptr;
4781}
4782
Willy Tarreau9fc5dcb2019-05-20 16:13:40 +02004783/* Appends into buffer <out> a hex dump of memory area <buf> for <len> bytes,
4784 * prepending each line with prefix <pfx>. The output is *not* initialized.
4785 * The output will not wrap pas the buffer's end so it is more optimal if the
4786 * caller makes sure the buffer is aligned first. A trailing zero will always
4787 * be appended (and not counted) if there is room for it. The caller must make
Willy Tarreau37101052019-05-20 16:48:20 +02004788 * sure that the area is dumpable first. If <unsafe> is non-null, the memory
4789 * locations are checked first for being readable.
Willy Tarreau9fc5dcb2019-05-20 16:13:40 +02004790 */
Willy Tarreau37101052019-05-20 16:48:20 +02004791void dump_hex(struct buffer *out, const char *pfx, const void *buf, int len, int unsafe)
Willy Tarreau9fc5dcb2019-05-20 16:13:40 +02004792{
4793 const unsigned char *d = buf;
4794 int i, j, start;
4795
4796 d = (const unsigned char *)(((unsigned long)buf) & -16);
4797 start = ((unsigned long)buf) & 15;
4798
4799 for (i = 0; i < start + len; i += 16) {
4800 chunk_appendf(out, (sizeof(void *) == 4) ? "%s%8p: " : "%s%16p: ", pfx, d + i);
4801
Willy Tarreau37101052019-05-20 16:48:20 +02004802 // 0: unchecked, 1: checked safe, 2: danger
4803 unsafe = !!unsafe;
4804 if (unsafe && !may_access(d + i))
4805 unsafe = 2;
4806
Willy Tarreau9fc5dcb2019-05-20 16:13:40 +02004807 for (j = 0; j < 16; j++) {
Willy Tarreau37101052019-05-20 16:48:20 +02004808 if ((i + j < start) || (i + j >= start + len))
Willy Tarreau9fc5dcb2019-05-20 16:13:40 +02004809 chunk_strcat(out, "'' ");
Willy Tarreau37101052019-05-20 16:48:20 +02004810 else if (unsafe > 1)
4811 chunk_strcat(out, "** ");
4812 else
4813 chunk_appendf(out, "%02x ", d[i + j]);
Willy Tarreau9fc5dcb2019-05-20 16:13:40 +02004814
4815 if (j == 7)
4816 chunk_strcat(out, "- ");
4817 }
4818 chunk_strcat(out, " ");
4819 for (j = 0; j < 16; j++) {
Willy Tarreau37101052019-05-20 16:48:20 +02004820 if ((i + j < start) || (i + j >= start + len))
Willy Tarreau9fc5dcb2019-05-20 16:13:40 +02004821 chunk_strcat(out, "'");
Willy Tarreau37101052019-05-20 16:48:20 +02004822 else if (unsafe > 1)
4823 chunk_strcat(out, "*");
Willy Tarreau90807112020-02-25 08:16:33 +01004824 else if (isprint((unsigned char)d[i + j]))
Willy Tarreau37101052019-05-20 16:48:20 +02004825 chunk_appendf(out, "%c", d[i + j]);
4826 else
4827 chunk_strcat(out, ".");
Willy Tarreau9fc5dcb2019-05-20 16:13:40 +02004828 }
4829 chunk_strcat(out, "\n");
4830 }
4831}
4832
Willy Tarreau762fb3e2020-03-03 15:57:10 +01004833/* dumps <pfx> followed by <n> bytes from <addr> in hex form into buffer <buf>
4834 * enclosed in brackets after the address itself, formatted on 14 chars
4835 * including the "0x" prefix. This is meant to be used as a prefix for code
4836 * areas. For example:
4837 * "0x7f10b6557690 [48 c7 c0 0f 00 00 00 0f]"
4838 * It relies on may_access() to know if the bytes are dumpable, otherwise "--"
4839 * is emitted. A NULL <pfx> will be considered empty.
4840 */
4841void dump_addr_and_bytes(struct buffer *buf, const char *pfx, const void *addr, int n)
4842{
4843 int ok = 0;
4844 int i;
4845
4846 chunk_appendf(buf, "%s%#14lx [", pfx ? pfx : "", (long)addr);
4847
4848 for (i = 0; i < n; i++) {
4849 if (i == 0 || (((long)(addr + i) ^ (long)(addr)) & 4096))
4850 ok = may_access(addr + i);
4851 if (ok)
4852 chunk_appendf(buf, "%02x%s", ((uint8_t*)addr)[i], (i<n-1) ? " " : "]");
4853 else
4854 chunk_appendf(buf, "--%s", (i<n-1) ? " " : "]");
4855 }
4856}
4857
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004858/* print a line of text buffer (limited to 70 bytes) to <out>. The format is :
4859 * <2 spaces> <offset=5 digits> <space or plus> <space> <70 chars max> <\n>
4860 * which is 60 chars per line. Non-printable chars \t, \n, \r and \e are
4861 * encoded in C format. Other non-printable chars are encoded "\xHH". Original
4862 * lines are respected within the limit of 70 output chars. Lines that are
4863 * continuation of a previous truncated line begin with "+" instead of " "
4864 * after the offset. The new pointer is returned.
4865 */
Willy Tarreau83061a82018-07-13 11:56:34 +02004866int dump_text_line(struct buffer *out, const char *buf, int bsize, int len,
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004867 int *line, int ptr)
4868{
4869 int end;
4870 unsigned char c;
4871
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004872 end = out->data + 80;
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004873 if (end > out->size)
4874 return ptr;
4875
4876 chunk_appendf(out, " %05d%c ", ptr, (ptr == *line) ? ' ' : '+');
4877
4878 while (ptr < len && ptr < bsize) {
4879 c = buf[ptr];
Willy Tarreau90807112020-02-25 08:16:33 +01004880 if (isprint((unsigned char)c) && isascii((unsigned char)c) && c != '\\') {
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004881 if (out->data > end - 2)
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004882 break;
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004883 out->area[out->data++] = c;
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004884 } else if (c == '\t' || c == '\n' || c == '\r' || c == '\e' || c == '\\') {
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004885 if (out->data > end - 3)
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004886 break;
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004887 out->area[out->data++] = '\\';
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004888 switch (c) {
4889 case '\t': c = 't'; break;
4890 case '\n': c = 'n'; break;
4891 case '\r': c = 'r'; break;
4892 case '\e': c = 'e'; break;
4893 case '\\': c = '\\'; break;
4894 }
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004895 out->area[out->data++] = c;
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004896 } else {
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004897 if (out->data > end - 5)
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004898 break;
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004899 out->area[out->data++] = '\\';
4900 out->area[out->data++] = 'x';
4901 out->area[out->data++] = hextab[(c >> 4) & 0xF];
4902 out->area[out->data++] = hextab[c & 0xF];
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004903 }
4904 if (buf[ptr++] == '\n') {
4905 /* we had a line break, let's return now */
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004906 out->area[out->data++] = '\n';
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004907 *line = ptr;
4908 return ptr;
4909 }
4910 }
4911 /* we have an incomplete line, we return it as-is */
Willy Tarreau843b7cb2018-07-13 10:54:26 +02004912 out->area[out->data++] = '\n';
Willy Tarreau97c2ae12016-11-22 18:00:20 +01004913 return ptr;
4914}
4915
Willy Tarreau0ebb5112016-12-05 00:10:57 +01004916/* displays a <len> long memory block at <buf>, assuming first byte of <buf>
Willy Tarreaued936c52017-04-27 18:03:20 +02004917 * has address <baseaddr>. String <pfx> may be placed as a prefix in front of
4918 * each line. It may be NULL if unused. The output is emitted to file <out>.
Willy Tarreau0ebb5112016-12-05 00:10:57 +01004919 */
Willy Tarreaued936c52017-04-27 18:03:20 +02004920void debug_hexdump(FILE *out, const char *pfx, const char *buf,
4921 unsigned int baseaddr, int len)
Willy Tarreau0ebb5112016-12-05 00:10:57 +01004922{
Willy Tarreau73459792017-04-11 07:58:08 +02004923 unsigned int i;
4924 int b, j;
Willy Tarreau0ebb5112016-12-05 00:10:57 +01004925
4926 for (i = 0; i < (len + (baseaddr & 15)); i += 16) {
4927 b = i - (baseaddr & 15);
Willy Tarreaued936c52017-04-27 18:03:20 +02004928 fprintf(out, "%s%08x: ", pfx ? pfx : "", i + (baseaddr & ~15));
Willy Tarreau0ebb5112016-12-05 00:10:57 +01004929 for (j = 0; j < 8; j++) {
4930 if (b + j >= 0 && b + j < len)
4931 fprintf(out, "%02x ", (unsigned char)buf[b + j]);
4932 else
4933 fprintf(out, " ");
4934 }
4935
4936 if (b + j >= 0 && b + j < len)
4937 fputc('-', out);
4938 else
4939 fputc(' ', out);
4940
4941 for (j = 8; j < 16; j++) {
4942 if (b + j >= 0 && b + j < len)
4943 fprintf(out, " %02x", (unsigned char)buf[b + j]);
4944 else
4945 fprintf(out, " ");
4946 }
4947
4948 fprintf(out, " ");
4949 for (j = 0; j < 16; j++) {
4950 if (b + j >= 0 && b + j < len) {
4951 if (isprint((unsigned char)buf[b + j]))
4952 fputc((unsigned char)buf[b + j], out);
4953 else
4954 fputc('.', out);
4955 }
4956 else
4957 fputc(' ', out);
4958 }
4959 fputc('\n', out);
4960 }
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01004961}
4962
Willy Tarreaubb869862020-04-16 10:52:41 +02004963/* Tries to report the executable path name on platforms supporting this. If
4964 * not found or not possible, returns NULL.
4965 */
4966const char *get_exec_path()
4967{
4968 const char *ret = NULL;
4969
David Carlier43a56852022-03-04 15:50:48 +00004970#if defined(__linux__) && defined(__GLIBC__) && (__GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ >= 16))
Willy Tarreaubb869862020-04-16 10:52:41 +02004971 long execfn = getauxval(AT_EXECFN);
4972
4973 if (execfn && execfn != ENOENT)
4974 ret = (const char *)execfn;
devnexen@gmail.comc4e52322021-08-17 12:55:49 +01004975#elif defined(__FreeBSD__)
4976 Elf_Auxinfo *auxv;
4977 for (auxv = __elf_aux_vector; auxv->a_type != AT_NULL; ++auxv) {
4978 if (auxv->a_type == AT_EXECPATH) {
4979 ret = (const char *)auxv->a_un.a_ptr;
4980 break;
4981 }
4982 }
David Carlierbd2cced2021-08-17 08:44:25 +01004983#elif defined(__NetBSD__)
4984 AuxInfo *auxv;
4985 for (auxv = _dlauxinfo(); auxv->a_type != AT_NULL; ++auxv) {
4986 if (auxv->a_type == AT_SUN_EXECNAME) {
4987 ret = (const char *)auxv->a_v;
4988 break;
4989 }
4990 }
David Carlier7198c702022-05-14 17:15:49 +01004991#elif defined(__sun)
4992 ret = getexecname();
Willy Tarreaubb869862020-04-16 10:52:41 +02004993#endif
4994 return ret;
4995}
4996
Baruch Siache1651b22020-07-24 07:52:20 +03004997#if (defined(__ELF__) && !defined(__linux__)) || defined(USE_DL)
Willy Tarreau9133e482020-03-04 10:19:36 +01004998/* calls dladdr() or dladdr1() on <addr> and <dli>. If dladdr1 is available,
4999 * also returns the symbol size in <size>, otherwise returns 0 there.
5000 */
5001static int dladdr_and_size(const void *addr, Dl_info *dli, size_t *size)
5002{
5003 int ret;
Willy Tarreau7b2108c2021-08-30 10:15:35 +02005004#if defined(__GLIBC__) && (__GLIBC__ > 2 || (__GLIBC__ == 2 && __GLIBC_MINOR__ >= 3)) // most detailed one
Willy Tarreauf3d5c4b2022-01-28 09:42:29 +01005005 const ElfW(Sym) *sym __attribute__((may_alias));
Willy Tarreau9133e482020-03-04 10:19:36 +01005006
5007 ret = dladdr1(addr, dli, (void **)&sym, RTLD_DL_SYMENT);
5008 if (ret)
5009 *size = sym ? sym->st_size : 0;
5010#else
David Carlierae5c42f2021-12-31 08:15:29 +00005011#if defined(__sun)
5012 ret = dladdr((void *)addr, dli);
5013#else
Willy Tarreau9133e482020-03-04 10:19:36 +01005014 ret = dladdr(addr, dli);
David Carlierae5c42f2021-12-31 08:15:29 +00005015#endif
Willy Tarreau9133e482020-03-04 10:19:36 +01005016 *size = 0;
5017#endif
5018 return ret;
5019}
Willy Tarreau64192392021-05-05 09:06:21 +02005020
Willy Tarreau5b3cd952022-07-18 13:58:17 +02005021/* Sets build_is_static to true if we detect a static build. Some older glibcs
5022 * tend to crash inside dlsym() in static builds, but tests show that at least
5023 * dladdr() still works (and will fail to resolve anything of course). Thus we
5024 * try to determine if we're on a static build to avoid calling dlsym() in this
5025 * case.
Willy Tarreau288dc1d2022-07-16 13:49:34 +02005026 */
Willy Tarreau5b3cd952022-07-18 13:58:17 +02005027void check_if_static_build()
Willy Tarreau288dc1d2022-07-16 13:49:34 +02005028{
Willy Tarreau5b3cd952022-07-18 13:58:17 +02005029 Dl_info dli = { };
5030 size_t size = 0;
5031
5032 /* Now let's try to be smarter */
5033 if (!dladdr_and_size(&main, &dli, &size))
5034 build_is_static = 1;
5035 else
5036 build_is_static = 0;
Willy Tarreau288dc1d2022-07-16 13:49:34 +02005037}
5038
Willy Tarreau5b3cd952022-07-18 13:58:17 +02005039INITCALL0(STG_PREPARE, check_if_static_build);
Willy Tarreau288dc1d2022-07-16 13:49:34 +02005040
Willy Tarreau64192392021-05-05 09:06:21 +02005041/* Tries to retrieve the address of the first occurrence symbol <name>.
5042 * Note that NULL in return is not always an error as a symbol may have that
5043 * address in special situations.
5044 */
5045void *get_sym_curr_addr(const char *name)
5046{
5047 void *ptr = NULL;
5048
5049#ifdef RTLD_DEFAULT
Willy Tarreau5b3cd952022-07-18 13:58:17 +02005050 if (!build_is_static)
Willy Tarreau288dc1d2022-07-16 13:49:34 +02005051 ptr = dlsym(RTLD_DEFAULT, name);
Willy Tarreau64192392021-05-05 09:06:21 +02005052#endif
5053 return ptr;
5054}
5055
5056
5057/* Tries to retrieve the address of the next occurrence of symbol <name>
5058 * Note that NULL in return is not always an error as a symbol may have that
5059 * address in special situations.
5060 */
5061void *get_sym_next_addr(const char *name)
5062{
5063 void *ptr = NULL;
5064
5065#ifdef RTLD_NEXT
Willy Tarreau5b3cd952022-07-18 13:58:17 +02005066 if (!build_is_static)
Willy Tarreau288dc1d2022-07-16 13:49:34 +02005067 ptr = dlsym(RTLD_NEXT, name);
Willy Tarreau9133e482020-03-04 10:19:36 +01005068#endif
Willy Tarreau64192392021-05-05 09:06:21 +02005069 return ptr;
5070}
5071
5072#else /* elf & linux & dl */
5073
5074/* no possible resolving on other platforms at the moment */
5075void *get_sym_curr_addr(const char *name)
5076{
5077 return NULL;
5078}
5079
5080void *get_sym_next_addr(const char *name)
5081{
5082 return NULL;
5083}
5084
5085#endif /* elf & linux & dl */
Willy Tarreau9133e482020-03-04 10:19:36 +01005086
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005087/* Tries to append to buffer <buf> some indications about the symbol at address
5088 * <addr> using the following form:
5089 * lib:+0xoffset (unresolvable address from lib's base)
5090 * main+0xoffset (unresolvable address from main (+/-))
5091 * lib:main+0xoffset (unresolvable lib address from main (+/-))
5092 * name (resolved exact exec address)
5093 * lib:name (resolved exact lib address)
5094 * name+0xoffset/0xsize (resolved address within exec symbol)
5095 * lib:name+0xoffset/0xsize (resolved address within lib symbol)
5096 *
5097 * The file name (lib or executable) is limited to what lies between the last
5098 * '/' and the first following '.'. An optional prefix <pfx> is prepended before
5099 * the output if not null. The file is not dumped when it's the same as the one
Baruch Siache1651b22020-07-24 07:52:20 +03005100 * that contains the "main" symbol, or when __ELF__ && USE_DL are not set.
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005101 *
5102 * The symbol's base address is returned, or NULL when unresolved, in order to
5103 * allow the caller to match it against known ones.
5104 */
Willy Tarreau45fd1032021-01-20 14:37:59 +01005105const void *resolve_sym_name(struct buffer *buf, const char *pfx, const void *addr)
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005106{
5107 const struct {
5108 const void *func;
5109 const char *name;
5110 } fcts[] = {
5111 { .func = process_stream, .name = "process_stream" },
5112 { .func = task_run_applet, .name = "task_run_applet" },
Willy Tarreau462b9892022-05-18 18:06:53 +02005113 { .func = sc_conn_io_cb, .name = "sc_conn_io_cb" },
Willy Tarreau586f71b2020-12-11 15:54:36 +01005114 { .func = sock_conn_iocb, .name = "sock_conn_iocb" },
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005115 { .func = dgram_fd_handler, .name = "dgram_fd_handler" },
5116 { .func = listener_accept, .name = "listener_accept" },
Willy Tarreaud597ec22021-01-29 14:29:06 +01005117 { .func = manage_global_listener_queue, .name = "manage_global_listener_queue" },
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005118 { .func = poller_pipe_io_handler, .name = "poller_pipe_io_handler" },
5119 { .func = mworker_accept_wrapper, .name = "mworker_accept_wrapper" },
Willy Tarreau02922e12021-01-29 12:27:57 +01005120 { .func = session_expire_embryonic, .name = "session_expire_embryonic" },
Willy Tarreaufb5401f2021-01-29 12:25:23 +01005121#ifdef USE_THREAD
5122 { .func = accept_queue_process, .name = "accept_queue_process" },
5123#endif
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005124#ifdef USE_LUA
5125 { .func = hlua_process_task, .name = "hlua_process_task" },
5126#endif
Ilya Shipitsinbdec3ba2020-11-14 01:56:34 +05005127#ifdef SSL_MODE_ASYNC
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005128 { .func = ssl_async_fd_free, .name = "ssl_async_fd_free" },
5129 { .func = ssl_async_fd_handler, .name = "ssl_async_fd_handler" },
5130#endif
Willy Tarreau8f6da642023-03-10 12:04:02 +01005131#ifdef USE_QUIC
5132 { .func = quic_conn_sock_fd_iocb, .name = "quic_conn_sock_fd_iocb" },
5133#endif
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005134 };
5135
Baruch Siache1651b22020-07-24 07:52:20 +03005136#if (defined(__ELF__) && !defined(__linux__)) || defined(USE_DL)
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005137 Dl_info dli, dli_main;
Willy Tarreau9133e482020-03-04 10:19:36 +01005138 size_t size;
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005139 const char *fname, *p;
5140#endif
5141 int i;
5142
5143 if (pfx)
5144 chunk_appendf(buf, "%s", pfx);
5145
5146 for (i = 0; i < sizeof(fcts) / sizeof(fcts[0]); i++) {
5147 if (addr == fcts[i].func) {
5148 chunk_appendf(buf, "%s", fcts[i].name);
5149 return addr;
5150 }
5151 }
5152
Baruch Siache1651b22020-07-24 07:52:20 +03005153#if (defined(__ELF__) && !defined(__linux__)) || defined(USE_DL)
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005154 /* Now let's try to be smarter */
Willy Tarreau9133e482020-03-04 10:19:36 +01005155 if (!dladdr_and_size(addr, &dli, &size))
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005156 goto unknown;
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005157
5158 /* 1. prefix the library name if it's not the same object as the one
5159 * that contains the main function. The name is picked between last '/'
5160 * and first following '.'.
5161 */
5162 if (!dladdr(main, &dli_main))
5163 dli_main.dli_fbase = NULL;
5164
5165 if (dli_main.dli_fbase != dli.dli_fbase) {
5166 fname = dli.dli_fname;
5167 p = strrchr(fname, '/');
5168 if (p++)
5169 fname = p;
5170 p = strchr(fname, '.');
5171 if (!p)
5172 p = fname + strlen(fname);
5173
5174 chunk_appendf(buf, "%.*s:", (int)(long)(p - fname), fname);
5175 }
5176
5177 /* 2. symbol name */
5178 if (dli.dli_sname) {
5179 /* known, dump it and return symbol's address (exact or relative) */
5180 chunk_appendf(buf, "%s", dli.dli_sname);
5181 if (addr != dli.dli_saddr) {
5182 chunk_appendf(buf, "+%#lx", (long)(addr - dli.dli_saddr));
Willy Tarreau9133e482020-03-04 10:19:36 +01005183 if (size)
5184 chunk_appendf(buf, "/%#lx", (long)size);
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005185 }
5186 return dli.dli_saddr;
5187 }
5188 else if (dli_main.dli_fbase != dli.dli_fbase) {
5189 /* unresolved symbol from a known library, report relative offset */
5190 chunk_appendf(buf, "+%#lx", (long)(addr - dli.dli_fbase));
5191 return NULL;
5192 }
Baruch Siache1651b22020-07-24 07:52:20 +03005193#endif /* __ELF__ && !__linux__ || USE_DL */
Willy Tarreaueb8b1ca2020-03-03 17:09:08 +01005194 unknown:
5195 /* unresolved symbol from the main file, report relative offset to main */
5196 if ((void*)addr < (void*)main)
5197 chunk_appendf(buf, "main-%#lx", (long)((void*)main - addr));
5198 else
5199 chunk_appendf(buf, "main+%#lx", (long)(addr - (void*)main));
5200 return NULL;
Willy Tarreau0ebb5112016-12-05 00:10:57 +01005201}
5202
Willy Tarreau6ab7b212021-12-28 09:57:10 +01005203/* On systems where this is supported, let's provide a possibility to enumerate
5204 * the list of object files. The output is appended to a buffer initialized by
5205 * the caller, with one name per line. A trailing zero is always emitted if data
5206 * are written. Only real objects are dumped (executable and .so libs). The
5207 * function returns non-zero if it dumps anything. These functions do not make
5208 * use of the trash so that it is possible for the caller to call them with the
5209 * trash on input. The output format may be platform-specific but at least one
5210 * version must emit raw object file names when argument is zero.
5211 */
5212#if defined(HA_HAVE_DUMP_LIBS)
5213# if defined(HA_HAVE_DL_ITERATE_PHDR)
5214/* the private <data> we pass below is a dump context initialized like this */
5215struct dl_dump_ctx {
5216 struct buffer *buf;
5217 int with_addr;
5218};
5219
5220static int dl_dump_libs_cb(struct dl_phdr_info *info, size_t size, void *data)
5221{
5222 struct dl_dump_ctx *ctx = data;
5223 const char *fname;
5224 size_t p1, p2, beg, end;
5225 int idx;
5226
5227 if (!info || !info->dlpi_name)
5228 goto leave;
5229
5230 if (!*info->dlpi_name)
5231 fname = get_exec_path();
5232 else if (strchr(info->dlpi_name, '/'))
5233 fname = info->dlpi_name;
5234 else
5235 /* else it's a VDSO or similar and we're not interested */
5236 goto leave;
5237
5238 if (!ctx->with_addr)
5239 goto dump_name;
5240
5241 /* virtual addresses are relative to the load address and are per
5242 * pseudo-header, so we have to scan them all to find the furthest
5243 * one from the beginning. In this case we only dump entries if
5244 * they have at least one section.
5245 */
5246 beg = ~0; end = 0;
5247 for (idx = 0; idx < info->dlpi_phnum; idx++) {
5248 if (!info->dlpi_phdr[idx].p_memsz)
5249 continue;
5250 p1 = info->dlpi_phdr[idx].p_vaddr;
5251 if (p1 < beg)
5252 beg = p1;
5253 p2 = p1 + info->dlpi_phdr[idx].p_memsz - 1;
5254 if (p2 > end)
5255 end = p2;
5256 }
5257
5258 if (!idx)
5259 goto leave;
5260
5261 chunk_appendf(ctx->buf, "0x%012llx-0x%012llx (0x%07llx) ",
5262 (ullong)info->dlpi_addr + beg,
5263 (ullong)info->dlpi_addr + end,
5264 (ullong)(end - beg + 1));
5265 dump_name:
5266 chunk_appendf(ctx->buf, "%s\n", fname);
5267 leave:
5268 return 0;
5269}
5270
5271/* dumps lib names and optionally address ranges */
5272int dump_libs(struct buffer *output, int with_addr)
5273{
5274 struct dl_dump_ctx ctx = { .buf = output, .with_addr = with_addr };
5275 size_t old_data = output->data;
5276
5277 dl_iterate_phdr(dl_dump_libs_cb, &ctx);
5278 return output->data != old_data;
5279}
5280# else // no DL_ITERATE_PHDR
5281# error "No dump_libs() function for this platform"
5282# endif
5283#else // no HA_HAVE_DUMP_LIBS
5284
5285/* unsupported platform: do not dump anything */
5286int dump_libs(struct buffer *output, int with_addr)
5287{
5288 return 0;
5289}
5290
5291#endif // HA_HAVE_DUMP_LIBS
5292
Frédéric Lécaille3b717162019-02-25 15:04:22 +01005293/*
5294 * Allocate an array of unsigned int with <nums> as address from <str> string
Ilya Shipitsin46a030c2020-07-05 16:36:08 +05005295 * made of integer separated by dot characters.
Frédéric Lécaille3b717162019-02-25 15:04:22 +01005296 *
5297 * First, initializes the value with <sz> as address to 0 and initializes the
5298 * array with <nums> as address to NULL. Then allocates the array with <nums> as
5299 * address updating <sz> pointed value to the size of this array.
5300 *
5301 * Returns 1 if succeeded, 0 if not.
5302 */
5303int parse_dotted_uints(const char *str, unsigned int **nums, size_t *sz)
5304{
5305 unsigned int *n;
5306 const char *s, *end;
5307
5308 s = str;
5309 *sz = 0;
5310 end = str + strlen(str);
5311 *nums = n = NULL;
5312
5313 while (1) {
5314 unsigned int r;
5315
5316 if (s >= end)
5317 break;
5318
5319 r = read_uint(&s, end);
5320 /* Expected characters after having read an uint: '\0' or '.',
5321 * if '.', must not be terminal.
5322 */
Christopher Faulet4b524122021-02-11 10:42:41 +01005323 if (*s != '\0'&& (*s++ != '.' || s == end)) {
5324 free(n);
Frédéric Lécaille3b717162019-02-25 15:04:22 +01005325 return 0;
Christopher Faulet4b524122021-02-11 10:42:41 +01005326 }
Frédéric Lécaille3b717162019-02-25 15:04:22 +01005327
Frédéric Lécaille12a71842019-02-26 18:19:48 +01005328 n = my_realloc2(n, (*sz + 1) * sizeof *n);
Frédéric Lécaille3b717162019-02-25 15:04:22 +01005329 if (!n)
5330 return 0;
5331
5332 n[(*sz)++] = r;
5333 }
5334 *nums = n;
5335
5336 return 1;
5337}
5338
Willy Tarreau4d589e72019-08-23 19:02:26 +02005339
5340/* returns the number of bytes needed to encode <v> as a varint. An inline
5341 * version exists for use with constants (__varint_bytes()).
5342 */
5343int varint_bytes(uint64_t v)
5344{
5345 int len = 1;
5346
5347 if (v >= 240) {
5348 v = (v - 240) >> 4;
5349 while (1) {
5350 len++;
5351 if (v < 128)
5352 break;
5353 v = (v - 128) >> 7;
5354 }
5355 }
5356 return len;
5357}
5358
Willy Tarreau52bf8392020-03-08 00:42:37 +01005359
5360/* Random number generator state, see below */
Willy Tarreau1544c142020-03-12 00:31:18 +01005361static uint64_t ha_random_state[2] ALIGNED(2*sizeof(uint64_t));
Willy Tarreau52bf8392020-03-08 00:42:37 +01005362
5363/* This is a thread-safe implementation of xoroshiro128** described below:
5364 * http://prng.di.unimi.it/
5365 * It features a 2^128 long sequence, returns 64 high-quality bits on each call,
5366 * supports fast jumps and passes all common quality tests. It is thread-safe,
5367 * uses a double-cas on 64-bit architectures supporting it, and falls back to a
5368 * local lock on other ones.
5369 */
5370uint64_t ha_random64()
5371{
Willy Tarreau1544c142020-03-12 00:31:18 +01005372 uint64_t old[2] ALIGNED(2*sizeof(uint64_t));
5373 uint64_t new[2] ALIGNED(2*sizeof(uint64_t));
Willy Tarreau52bf8392020-03-08 00:42:37 +01005374
5375#if defined(USE_THREAD) && (!defined(HA_CAS_IS_8B) || !defined(HA_HAVE_CAS_DW))
5376 static HA_SPINLOCK_T rand_lock;
5377
5378 HA_SPIN_LOCK(OTHER_LOCK, &rand_lock);
5379#endif
5380
5381 old[0] = ha_random_state[0];
5382 old[1] = ha_random_state[1];
5383
5384#if defined(USE_THREAD) && defined(HA_CAS_IS_8B) && defined(HA_HAVE_CAS_DW)
5385 do {
5386#endif
Willy Tarreau52bf8392020-03-08 00:42:37 +01005387 new[1] = old[0] ^ old[1];
5388 new[0] = rotl64(old[0], 24) ^ new[1] ^ (new[1] << 16); // a, b
5389 new[1] = rotl64(new[1], 37); // c
5390
5391#if defined(USE_THREAD) && defined(HA_CAS_IS_8B) && defined(HA_HAVE_CAS_DW)
5392 } while (unlikely(!_HA_ATOMIC_DWCAS(ha_random_state, old, new)));
5393#else
5394 ha_random_state[0] = new[0];
5395 ha_random_state[1] = new[1];
5396#if defined(USE_THREAD)
5397 HA_SPIN_UNLOCK(OTHER_LOCK, &rand_lock);
5398#endif
5399#endif
Willy Tarreaub2475a12021-05-09 10:26:14 +02005400 return rotl64(old[0] * 5, 7) * 9;
Willy Tarreau52bf8392020-03-08 00:42:37 +01005401}
5402
5403/* seeds the random state using up to <len> bytes from <seed>, starting with
5404 * the first non-zero byte.
5405 */
5406void ha_random_seed(const unsigned char *seed, size_t len)
5407{
5408 size_t pos;
5409
5410 /* the seed must not be all zeroes, so we pre-fill it with alternating
5411 * bits and overwrite part of them with the block starting at the first
5412 * non-zero byte from the seed.
5413 */
5414 memset(ha_random_state, 0x55, sizeof(ha_random_state));
5415
5416 for (pos = 0; pos < len; pos++)
5417 if (seed[pos] != 0)
5418 break;
5419
5420 if (pos == len)
5421 return;
5422
5423 seed += pos;
5424 len -= pos;
5425
5426 if (len > sizeof(ha_random_state))
5427 len = sizeof(ha_random_state);
5428
5429 memcpy(ha_random_state, seed, len);
5430}
5431
5432/* This causes a jump to (dist * 2^96) places in the pseudo-random sequence,
5433 * and is equivalent to calling ha_random64() as many times. It is used to
5434 * provide non-overlapping sequences of 2^96 numbers (~7*10^28) to up to 2^32
5435 * different generators (i.e. different processes after a fork). The <dist>
5436 * argument is the distance to jump to and is used in a loop so it rather not
5437 * be too large if the processing time is a concern.
5438 *
5439 * BEWARE: this function is NOT thread-safe and must not be called during
5440 * concurrent accesses to ha_random64().
5441 */
5442void ha_random_jump96(uint32_t dist)
5443{
5444 while (dist--) {
5445 uint64_t s0 = 0;
5446 uint64_t s1 = 0;
5447 int b;
5448
5449 for (b = 0; b < 64; b++) {
5450 if ((0xd2a98b26625eee7bULL >> b) & 1) {
5451 s0 ^= ha_random_state[0];
5452 s1 ^= ha_random_state[1];
5453 }
5454 ha_random64();
5455 }
5456
5457 for (b = 0; b < 64; b++) {
5458 if ((0xdddf9b1090aa7ac1ULL >> b) & 1) {
5459 s0 ^= ha_random_state[0];
5460 s1 ^= ha_random_state[1];
5461 }
5462 ha_random64();
5463 }
5464 ha_random_state[0] = s0;
5465 ha_random_state[1] = s1;
5466 }
5467}
5468
Willy Tarreauee3bcdd2020-03-08 17:48:17 +01005469/* Generates an RFC4122 UUID into chunk <output> which must be at least 37
5470 * bytes large.
5471 */
5472void ha_generate_uuid(struct buffer *output)
5473{
5474 uint32_t rnd[4];
5475 uint64_t last;
5476
5477 last = ha_random64();
5478 rnd[0] = last;
5479 rnd[1] = last >> 32;
5480
5481 last = ha_random64();
5482 rnd[2] = last;
5483 rnd[3] = last >> 32;
5484
5485 chunk_printf(output, "%8.8x-%4.4x-%4.4x-%4.4x-%12.12llx",
5486 rnd[0],
5487 rnd[1] & 0xFFFF,
5488 ((rnd[1] >> 16u) & 0xFFF) | 0x4000, // highest 4 bits indicate the uuid version
5489 (rnd[2] & 0x3FFF) | 0x8000, // the highest 2 bits indicate the UUID variant (10),
5490 (long long)((rnd[2] >> 14u) | ((uint64_t) rnd[3] << 18u)) & 0xFFFFFFFFFFFFull);
5491}
5492
5493
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005494/* only used by parse_line() below. It supports writing in place provided that
5495 * <in> is updated to the next location before calling it. In that case, the
5496 * char at <in> may be overwritten.
5497 */
5498#define EMIT_CHAR(x) \
5499 do { \
5500 char __c = (char)(x); \
5501 if ((opts & PARSE_OPT_INPLACE) && out+outpos > in) \
5502 err |= PARSE_ERR_OVERLAP; \
5503 if (outpos >= outmax) \
5504 err |= PARSE_ERR_TOOLARGE; \
5505 if (!err) \
5506 out[outpos] = __c; \
5507 outpos++; \
5508 } while (0)
5509
Ilya Shipitsin46a030c2020-07-05 16:36:08 +05005510/* Parse <in>, copy it into <out> split into isolated words whose pointers
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005511 * are put in <args>. If more than <outlen> bytes have to be emitted, the
5512 * extraneous ones are not emitted but <outlen> is updated so that the caller
5513 * knows how much to realloc. Similarly, <args> are not updated beyond <nbargs>
5514 * but the returned <nbargs> indicates how many were found. All trailing args
Willy Tarreau61dd44b2020-06-25 07:35:42 +02005515 * up to <nbargs> point to the trailing zero, and as long as <nbargs> is > 0,
5516 * it is guaranteed that at least one arg will point to the zero. It is safe
5517 * to call it with a NULL <args> if <nbargs> is 0.
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005518 *
5519 * <out> may overlap with <in> provided that it never goes further, in which
5520 * case the parser will accept to perform in-place parsing and unquoting/
5521 * unescaping but only if environment variables do not lead to expansion that
5522 * causes overlapping, otherwise the input string being destroyed, the error
5523 * will not be recoverable. Note that even during out-of-place <in> will
5524 * experience temporary modifications in-place for variable resolution and must
5525 * be writable, and will also receive zeroes to delimit words when using
5526 * in-place copy. Parsing options <opts> taken from PARSE_OPT_*. Return value
5527 * is zero on success otherwise a bitwise-or of PARSE_ERR_*. Upon error, the
5528 * starting point of the first invalid character sequence or unmatched
5529 * quote/brace is reported in <errptr> if not NULL. When using in-place parsing
5530 * error reporting might be difficult since zeroes will have been inserted into
5531 * the string. One solution for the caller may consist in replacing all args
5532 * delimiters with spaces in this case.
5533 */
Maximilian Mader29c6cd72021-06-06 00:50:21 +02005534uint32_t parse_line(char *in, char *out, size_t *outlen, char **args, int *nbargs, uint32_t opts, const char **errptr)
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005535{
5536 char *quote = NULL;
5537 char *brace = NULL;
Amaury Denoyellefa41cb62020-10-01 14:32:35 +02005538 char *word_expand = NULL;
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005539 unsigned char hex1, hex2;
5540 size_t outmax = *outlen;
Willy Tarreau61dd44b2020-06-25 07:35:42 +02005541 int argsmax = *nbargs - 1;
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005542 size_t outpos = 0;
5543 int squote = 0;
5544 int dquote = 0;
5545 int arg = 0;
5546 uint32_t err = 0;
5547
5548 *nbargs = 0;
5549 *outlen = 0;
5550
Willy Tarreau61dd44b2020-06-25 07:35:42 +02005551 /* argsmax may be -1 here, protecting args[] from any write */
5552 if (arg < argsmax)
5553 args[arg] = out;
5554
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005555 while (1) {
5556 if (*in >= '-' && *in != '\\') {
5557 /* speedup: directly send all regular chars starting
5558 * with '-', '.', '/', alnum etc...
5559 */
5560 EMIT_CHAR(*in++);
5561 continue;
5562 }
5563 else if (*in == '\0' || *in == '\n' || *in == '\r') {
5564 /* end of line */
5565 break;
5566 }
5567 else if (*in == '#' && (opts & PARSE_OPT_SHARP) && !squote && !dquote) {
5568 /* comment */
5569 break;
5570 }
5571 else if (*in == '"' && !squote && (opts & PARSE_OPT_DQUOTE)) { /* double quote outside single quotes */
5572 if (dquote) {
5573 dquote = 0;
5574 quote = NULL;
5575 }
5576 else {
5577 dquote = 1;
5578 quote = in;
5579 }
5580 in++;
5581 continue;
5582 }
5583 else if (*in == '\'' && !dquote && (opts & PARSE_OPT_SQUOTE)) { /* single quote outside double quotes */
5584 if (squote) {
5585 squote = 0;
5586 quote = NULL;
5587 }
5588 else {
5589 squote = 1;
5590 quote = in;
5591 }
5592 in++;
5593 continue;
5594 }
5595 else if (*in == '\\' && !squote && (opts & PARSE_OPT_BKSLASH)) {
5596 /* first, we'll replace \\, \<space>, \#, \r, \n, \t, \xXX with their
5597 * C equivalent value but only when they have a special meaning and within
5598 * double quotes for some of them. Other combinations left unchanged (eg: \1).
5599 */
5600 char tosend = *in;
5601
5602 switch (in[1]) {
5603 case ' ':
5604 case '\\':
5605 tosend = in[1];
5606 in++;
5607 break;
5608
5609 case 't':
5610 tosend = '\t';
5611 in++;
5612 break;
5613
5614 case 'n':
5615 tosend = '\n';
5616 in++;
5617 break;
5618
5619 case 'r':
5620 tosend = '\r';
5621 in++;
5622 break;
5623
5624 case '#':
5625 /* escaping of "#" only if comments are supported */
5626 if (opts & PARSE_OPT_SHARP)
5627 in++;
5628 tosend = *in;
5629 break;
5630
5631 case '\'':
5632 /* escaping of "'" only outside single quotes and only if single quotes are supported */
5633 if (opts & PARSE_OPT_SQUOTE && !squote)
5634 in++;
5635 tosend = *in;
5636 break;
5637
5638 case '"':
5639 /* escaping of '"' only outside single quotes and only if double quotes are supported */
5640 if (opts & PARSE_OPT_DQUOTE && !squote)
5641 in++;
5642 tosend = *in;
5643 break;
5644
5645 case '$':
5646 /* escaping of '$' only inside double quotes and only if env supported */
5647 if (opts & PARSE_OPT_ENV && dquote)
5648 in++;
5649 tosend = *in;
5650 break;
5651
5652 case 'x':
5653 if (!ishex(in[2]) || !ishex(in[3])) {
5654 /* invalid or incomplete hex sequence */
5655 err |= PARSE_ERR_HEX;
5656 if (errptr)
5657 *errptr = in;
5658 goto leave;
5659 }
Willy Tarreauf278eec2020-07-05 21:46:32 +02005660 hex1 = toupper((unsigned char)in[2]) - '0';
5661 hex2 = toupper((unsigned char)in[3]) - '0';
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005662 if (hex1 > 9) hex1 -= 'A' - '9' - 1;
5663 if (hex2 > 9) hex2 -= 'A' - '9' - 1;
5664 tosend = (hex1 << 4) + hex2;
5665 in += 3;
5666 break;
5667
5668 default:
5669 /* other combinations are not escape sequences */
5670 break;
5671 }
5672
5673 in++;
5674 EMIT_CHAR(tosend);
5675 }
5676 else if (isspace((unsigned char)*in) && !squote && !dquote) {
5677 /* a non-escaped space is an argument separator */
5678 while (isspace((unsigned char)*in))
5679 in++;
5680 EMIT_CHAR(0);
5681 arg++;
5682 if (arg < argsmax)
5683 args[arg] = out + outpos;
5684 else
5685 err |= PARSE_ERR_TOOMANY;
5686 }
5687 else if (*in == '$' && (opts & PARSE_OPT_ENV) && (dquote || !(opts & PARSE_OPT_DQUOTE))) {
5688 /* environment variables are evaluated anywhere, or only
5689 * inside double quotes if they are supported.
5690 */
5691 char *var_name;
5692 char save_char;
Willy Tarreaua46f1af2021-05-06 10:25:11 +02005693 const char *value;
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005694
5695 in++;
5696
5697 if (*in == '{')
5698 brace = in++;
5699
Willy Tarreaua46f1af2021-05-06 10:25:11 +02005700 if (!isalpha((unsigned char)*in) && *in != '_' && *in != '.') {
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005701 /* unacceptable character in variable name */
5702 err |= PARSE_ERR_VARNAME;
5703 if (errptr)
5704 *errptr = in;
5705 goto leave;
5706 }
5707
5708 var_name = in;
Willy Tarreaua46f1af2021-05-06 10:25:11 +02005709 if (*in == '.')
5710 in++;
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005711 while (isalnum((unsigned char)*in) || *in == '_')
5712 in++;
5713
5714 save_char = *in;
5715 *in = '\0';
Willy Tarreaua46f1af2021-05-06 10:25:11 +02005716 if (unlikely(*var_name == '.')) {
5717 /* internal pseudo-variables */
5718 if (strcmp(var_name, ".LINE") == 0)
5719 value = ultoa(global.cfg_curr_line);
5720 else if (strcmp(var_name, ".FILE") == 0)
5721 value = global.cfg_curr_file;
5722 else if (strcmp(var_name, ".SECTION") == 0)
5723 value = global.cfg_curr_section;
5724 else {
5725 /* unsupported internal variable name */
5726 err |= PARSE_ERR_VARNAME;
5727 if (errptr)
5728 *errptr = var_name;
5729 goto leave;
5730 }
5731 } else {
5732 value = getenv(var_name);
5733 }
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005734 *in = save_char;
5735
Amaury Denoyellefa41cb62020-10-01 14:32:35 +02005736 /* support for '[*]' sequence to force word expansion,
5737 * only available inside braces */
5738 if (*in == '[' && brace && (opts & PARSE_OPT_WORD_EXPAND)) {
5739 word_expand = in++;
5740
5741 if (*in++ != '*' || *in++ != ']') {
5742 err |= PARSE_ERR_WRONG_EXPAND;
5743 if (errptr)
5744 *errptr = word_expand;
5745 goto leave;
5746 }
5747 }
5748
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005749 if (brace) {
Willy Tarreauec347b12021-11-18 17:42:50 +01005750 if (*in == '-') {
5751 /* default value starts just after the '-' */
5752 if (!value)
5753 value = in + 1;
5754
5755 while (*in && *in != '}')
5756 in++;
5757 if (!*in)
5758 goto no_brace;
5759 *in = 0; // terminate the default value
5760 }
5761 else if (*in != '}') {
5762 no_brace:
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005763 /* unmatched brace */
5764 err |= PARSE_ERR_BRACE;
5765 if (errptr)
5766 *errptr = brace;
5767 goto leave;
5768 }
Willy Tarreauec347b12021-11-18 17:42:50 +01005769
5770 /* brace found, skip it */
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005771 in++;
5772 brace = NULL;
5773 }
5774
5775 if (value) {
Amaury Denoyellefa41cb62020-10-01 14:32:35 +02005776 while (*value) {
5777 /* expand as individual parameters on a space character */
Willy Tarreaufe2cc412020-10-01 18:04:40 +02005778 if (word_expand && isspace((unsigned char)*value)) {
Amaury Denoyellefa41cb62020-10-01 14:32:35 +02005779 EMIT_CHAR(0);
5780 ++arg;
5781 if (arg < argsmax)
5782 args[arg] = out + outpos;
5783 else
5784 err |= PARSE_ERR_TOOMANY;
5785
5786 /* skip consecutive spaces */
Willy Tarreaufe2cc412020-10-01 18:04:40 +02005787 while (isspace((unsigned char)*++value))
Amaury Denoyellefa41cb62020-10-01 14:32:35 +02005788 ;
5789 } else {
5790 EMIT_CHAR(*value++);
5791 }
5792 }
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005793 }
Frédéric Lécaille51fad642023-11-30 09:26:05 +01005794 else {
5795 /* An unmatched environment variable was parsed.
5796 * Let's skip the trailing double-quote character
5797 * and spaces.
5798 */
5799 if (likely(*var_name != '.') && *in == '"') {
5800 in++;
5801 while (isspace((unsigned char)*in))
5802 in++;
5803 if (dquote) {
5804 dquote = 0;
5805 quote = NULL;
5806 }
5807 }
5808 }
Amaury Denoyellefa41cb62020-10-01 14:32:35 +02005809 word_expand = NULL;
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005810 }
5811 else {
5812 /* any other regular char */
5813 EMIT_CHAR(*in++);
5814 }
5815 }
5816
5817 /* end of output string */
5818 EMIT_CHAR(0);
Erwan Le Goas8a6767d2022-09-23 15:06:34 +02005819
Willy Tarreau94ab1392022-10-03 08:27:55 +02005820 /* Don't add an empty arg after trailing spaces. Note that args[arg]
5821 * may contain some distances relative to NULL if <out> was NULL, or
5822 * pointers beyond the end of <out> in case <outlen> is too short, thus
5823 * we must not dereference it.
Erwan Le Goas8a6767d2022-09-23 15:06:34 +02005824 */
Willy Tarreau94ab1392022-10-03 08:27:55 +02005825 if (arg < argsmax && args[arg] != out + outpos - 1)
Erwan Le Goas8a6767d2022-09-23 15:06:34 +02005826 arg++;
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005827
5828 if (quote) {
5829 /* unmatched quote */
5830 err |= PARSE_ERR_QUOTE;
5831 if (errptr)
5832 *errptr = quote;
5833 goto leave;
5834 }
5835 leave:
5836 *nbargs = arg;
5837 *outlen = outpos;
5838
Willy Tarreau61dd44b2020-06-25 07:35:42 +02005839 /* empty all trailing args by making them point to the trailing zero,
5840 * at least the last one in any case.
5841 */
5842 if (arg > argsmax)
5843 arg = argsmax;
5844
5845 while (arg >= 0 && arg <= argsmax)
Willy Tarreauc8d167b2020-06-16 16:27:26 +02005846 args[arg++] = out + outpos - 1;
5847
5848 return err;
5849}
5850#undef EMIT_CHAR
5851
Willy Tarreauc54e5ad2020-06-25 09:15:40 +02005852/* This is used to sanitize an input line that's about to be used for error reporting.
5853 * It will adjust <line> to print approximately <width> chars around <pos>, trying to
5854 * preserve the beginning, with leading or trailing "..." when the line is truncated.
5855 * If non-printable chars are present in the output. It returns the new offset <pos>
5856 * in the modified line. Non-printable characters are replaced with '?'. <width> must
5857 * be at least 6 to support two "..." otherwise the result is undefined. The line
5858 * itself must have at least 7 chars allocated for the same reason.
5859 */
5860size_t sanitize_for_printing(char *line, size_t pos, size_t width)
5861{
5862 size_t shift = 0;
5863 char *out = line;
5864 char *in = line;
5865 char *end = line + width;
5866
5867 if (pos >= width) {
5868 /* if we have to shift, we'll be out of context, so let's
5869 * try to put <pos> at the center of width.
5870 */
5871 shift = pos - width / 2;
5872 in += shift + 3;
5873 end = out + width - 3;
5874 out[0] = out[1] = out[2] = '.';
5875 out += 3;
5876 }
5877
5878 while (out < end && *in) {
5879 if (isspace((unsigned char)*in))
5880 *out++ = ' ';
5881 else if (isprint((unsigned char)*in))
5882 *out++ = *in;
5883 else
5884 *out++ = '?';
5885 in++;
5886 }
5887
5888 if (end < line + width) {
5889 out[0] = out[1] = out[2] = '.';
5890 out += 3;
5891 }
5892
5893 *out++ = 0;
5894 return pos - shift;
5895}
Willy Tarreau06e69b52021-03-02 14:01:35 +01005896
Willy Tarreaue33c4b32021-03-12 18:59:31 +01005897/* Update array <fp> with the fingerprint of word <word> by counting the
Willy Tarreauba2c4452021-03-12 09:01:52 +01005898 * transitions between characters. <fp> is a 1024-entries array indexed as
5899 * 32*from+to. Positions for 'from' and 'to' are:
Willy Tarreau9294e882021-03-15 09:34:27 +01005900 * 1..26=letter, 27=digit, 28=other/begin/end.
5901 * Row "from=0" is used to mark the character's presence. Others unused.
Willy Tarreauba2c4452021-03-12 09:01:52 +01005902 */
Willy Tarreaue33c4b32021-03-12 18:59:31 +01005903void update_word_fingerprint(uint8_t *fp, const char *word)
Willy Tarreauba2c4452021-03-12 09:01:52 +01005904{
5905 const char *p;
5906 int from, to;
5907 int c;
5908
Willy Tarreauba2c4452021-03-12 09:01:52 +01005909 from = 28; // begin
5910 for (p = word; *p; p++) {
5911 c = tolower(*p);
5912 switch(c) {
Willy Tarreau9294e882021-03-15 09:34:27 +01005913 case 'a'...'z': to = c - 'a' + 1; break;
5914 case 'A'...'Z': to = tolower(c) - 'a' + 1; break;
5915 case '0'...'9': to = 27; break;
5916 default: to = 28; break;
Willy Tarreauba2c4452021-03-12 09:01:52 +01005917 }
Willy Tarreau9294e882021-03-15 09:34:27 +01005918 fp[to] = 1;
Willy Tarreauba2c4452021-03-12 09:01:52 +01005919 fp[32 * from + to]++;
5920 from = to;
5921 }
5922 to = 28; // end
5923 fp[32 * from + to]++;
5924}
5925
Erwan Le Goas9c766372022-09-14 17:40:24 +02005926/* This function hashes a word, scramble is the anonymizing key, returns
5927 * the hashed word when the key (scramble) != 0, else returns the word.
5928 * This function can be called NB_L_HASH_WORD times in a row, don't call
5929 * it if you called it more than NB_L_HASH_WORD.
5930 */
5931const char *hash_anon(uint32_t scramble, const char *string2hash, const char *prefix, const char *suffix)
5932{
5933 index_hash++;
Erwan Le Goasd2605cf2022-09-21 16:24:23 +02005934 if (index_hash == NB_L_HASH_WORD)
Erwan Le Goas9c766372022-09-14 17:40:24 +02005935 index_hash = 0;
5936
5937 /* don't hash empty strings */
5938 if (!string2hash[0] || (string2hash[0] == ' ' && string2hash[1] == 0))
5939 return string2hash;
5940
5941 if (scramble != 0) {
5942 snprintf(hash_word[index_hash], sizeof(hash_word[index_hash]), "%s%06x%s",
5943 prefix, HA_ANON(scramble, string2hash, strlen(string2hash)), suffix);
5944 return hash_word[index_hash];
5945 }
5946 else
5947 return string2hash;
5948}
5949
5950/* This function hashes or not an ip address ipstring, scramble is the anonymizing
5951 * key, returns the hashed ip with his port or ipstring when there is nothing to hash.
Erwan Le Goas5eef1582022-09-29 10:25:31 +02005952 * Put hasport equal 0 to point out ipstring has no port, else put an other int.
5953 * Without port, return a simple hash or ipstring.
Erwan Le Goas9c766372022-09-14 17:40:24 +02005954 */
Erwan Le Goas5eef1582022-09-29 10:25:31 +02005955const char *hash_ipanon(uint32_t scramble, char *ipstring, int hasport)
Erwan Le Goas9c766372022-09-14 17:40:24 +02005956{
5957 char *errmsg = NULL;
5958 struct sockaddr_storage *sa;
Erwan Le Goas5eef1582022-09-29 10:25:31 +02005959 struct sockaddr_storage ss;
Erwan Le Goas9c766372022-09-14 17:40:24 +02005960 char addr[46];
5961 int port;
5962
5963 index_hash++;
Erwan Le Goasd2605cf2022-09-21 16:24:23 +02005964 if (index_hash == NB_L_HASH_WORD) {
Erwan Le Goas9c766372022-09-14 17:40:24 +02005965 index_hash = 0;
5966 }
5967
Erwan Le Goas5eef1582022-09-29 10:25:31 +02005968 if (scramble == 0) {
Erwan Le Goas9c766372022-09-14 17:40:24 +02005969 return ipstring;
5970 }
Christopher Faulet015bbc22022-09-29 11:53:07 +02005971 if (strcmp(ipstring, "localhost") == 0 ||
5972 strcmp(ipstring, "stdout") == 0 ||
5973 strcmp(ipstring, "stderr") == 0 ||
5974 strncmp(ipstring, "fd@", 3) == 0 ||
5975 strncmp(ipstring, "sockpair@", 9) == 0) {
Erwan Le Goas5eef1582022-09-29 10:25:31 +02005976 return ipstring;
5977 }
Erwan Le Goas9c766372022-09-14 17:40:24 +02005978 else {
Erwan Le Goas5eef1582022-09-29 10:25:31 +02005979 if (hasport == 0) {
5980 memset(&ss, 0, sizeof(ss));
5981 if (str2ip2(ipstring, &ss, 1) == NULL) {
5982 return HA_ANON_STR(scramble, ipstring);
5983 }
5984 sa = &ss;
Erwan Le Goas9c766372022-09-14 17:40:24 +02005985 }
5986 else {
Erwan Le Goas5eef1582022-09-29 10:25:31 +02005987 sa = str2sa_range(ipstring, NULL, NULL, NULL, NULL, NULL, &errmsg, NULL, NULL,
Christopher Faulet7e50e4b2022-09-29 11:46:34 +02005988 PA_O_PORT_OK | PA_O_STREAM | PA_O_DGRAM | PA_O_XPRT | PA_O_CONNECT |
5989 PA_O_PORT_RANGE | PA_O_PORT_OFS | PA_O_RESOLVE);
Erwan Le Goas5eef1582022-09-29 10:25:31 +02005990 if (sa == NULL) {
5991 return HA_ANON_STR(scramble, ipstring);
5992 }
5993 }
5994 addr_to_str(sa, addr, sizeof(addr));
5995 port = get_host_port(sa);
Erwan Le Goas9c766372022-09-14 17:40:24 +02005996
Erwan Le Goas5eef1582022-09-29 10:25:31 +02005997 switch(sa->ss_family) {
5998 case AF_INET:
5999 if (strncmp(addr, "127", 3) == 0 || strncmp(addr, "255", 3) == 0 || strncmp(addr, "0", 1) == 0) {
6000 return ipstring;
6001 }
6002 else {
6003 if (port != 0) {
6004 snprintf(hash_word[index_hash], sizeof(hash_word[index_hash]), "IPV4(%06x):%d", HA_ANON(scramble, addr, strlen(addr)), port);
6005 return hash_word[index_hash];
Erwan Le Goas9c766372022-09-14 17:40:24 +02006006 }
6007 else {
Erwan Le Goas5eef1582022-09-29 10:25:31 +02006008 snprintf(hash_word[index_hash], sizeof(hash_word[index_hash]), "IPV4(%06x)", HA_ANON(scramble, addr, strlen(addr)));
6009 return hash_word[index_hash];
Erwan Le Goas9c766372022-09-14 17:40:24 +02006010 }
Erwan Le Goas5eef1582022-09-29 10:25:31 +02006011 }
6012 break;
Erwan Le Goas9c766372022-09-14 17:40:24 +02006013
Erwan Le Goas5eef1582022-09-29 10:25:31 +02006014 case AF_INET6:
6015 if (strcmp(addr, "::1") == 0) {
6016 return ipstring;
6017 }
6018 else {
6019 if (port != 0) {
6020 snprintf(hash_word[index_hash], sizeof(hash_word[index_hash]), "IPV6(%06x):%d", HA_ANON(scramble, addr, strlen(addr)), port);
6021 return hash_word[index_hash];
Erwan Le Goas9c766372022-09-14 17:40:24 +02006022 }
6023 else {
Erwan Le Goas5eef1582022-09-29 10:25:31 +02006024 snprintf(hash_word[index_hash], sizeof(hash_word[index_hash]), "IPV6(%06x)", HA_ANON(scramble, addr, strlen(addr)));
6025 return hash_word[index_hash];
Erwan Le Goas9c766372022-09-14 17:40:24 +02006026 }
Erwan Le Goas5eef1582022-09-29 10:25:31 +02006027 }
6028 break;
Erwan Le Goas9c766372022-09-14 17:40:24 +02006029
Erwan Le Goas5eef1582022-09-29 10:25:31 +02006030 case AF_UNIX:
6031 return HA_ANON_STR(scramble, ipstring);
6032 break;
6033
6034 default:
6035 return ipstring;
6036 break;
6037 };
Erwan Le Goas9c766372022-09-14 17:40:24 +02006038 }
6039 return ipstring;
6040}
6041
Willy Tarreaue33c4b32021-03-12 18:59:31 +01006042/* Initialize array <fp> with the fingerprint of word <word> by counting the
6043 * transitions between characters. <fp> is a 1024-entries array indexed as
6044 * 32*from+to. Positions for 'from' and 'to' are:
6045 * 0..25=letter, 26=digit, 27=other, 28=begin, 29=end, others unused.
6046 */
6047void make_word_fingerprint(uint8_t *fp, const char *word)
6048{
6049 memset(fp, 0, 1024);
6050 update_word_fingerprint(fp, word);
6051}
6052
Willy Tarreauba2c4452021-03-12 09:01:52 +01006053/* Return the distance between two word fingerprints created by function
6054 * make_word_fingerprint(). It's a positive integer calculated as the sum of
Willy Tarreau714c4c12021-03-15 09:44:53 +01006055 * the differences between each location.
Willy Tarreauba2c4452021-03-12 09:01:52 +01006056 */
6057int word_fingerprint_distance(const uint8_t *fp1, const uint8_t *fp2)
6058{
6059 int i, k, dist = 0;
6060
6061 for (i = 0; i < 1024; i++) {
6062 k = (int)fp1[i] - (int)fp2[i];
Willy Tarreau714c4c12021-03-15 09:44:53 +01006063 dist += abs(k);
Willy Tarreauba2c4452021-03-12 09:01:52 +01006064 }
6065 return dist;
6066}
6067
William Lallemand3aeb3f92021-08-21 23:59:56 +02006068/*
6069 * This function compares the loaded openssl version with a string <version>
6070 * This function use the same return code as compare_current_version:
6071 *
6072 * -1 : the version in argument is older than the current openssl version
6073 * 0 : the version in argument is the same as the current openssl version
6074 * 1 : the version in argument is newer than the current openssl version
6075 *
6076 * Or some errors:
6077 * -2 : openssl is not available on this process
6078 * -3 : the version in argument is not parsable
6079 */
6080int openssl_compare_current_version(const char *version)
6081{
6082#ifdef USE_OPENSSL
6083 int numversion;
6084
6085 numversion = openssl_version_parser(version);
6086 if (numversion == 0)
6087 return -3;
6088
6089 if (numversion < OPENSSL_VERSION_NUMBER)
6090 return -1;
6091 else if (numversion > OPENSSL_VERSION_NUMBER)
6092 return 1;
6093 else
6094 return 0;
6095#else
6096 return -2;
6097#endif
6098}
6099
Remi Tricot-Le Bretonb01179a2021-10-11 15:34:12 +02006100/*
6101 * This function compares the loaded openssl name with a string <name>
6102 * This function returns 0 if the OpenSSL name starts like the passed parameter,
6103 * 1 otherwise.
6104 */
6105int openssl_compare_current_name(const char *name)
6106{
6107#ifdef USE_OPENSSL
6108 int name_len = 0;
6109 const char *openssl_version = OpenSSL_version(OPENSSL_VERSION);
6110
6111 if (name) {
6112 name_len = strlen(name);
6113 if (strlen(name) <= strlen(openssl_version))
6114 return strncmp(openssl_version, name, name_len);
6115 }
6116#endif
6117 return 1;
6118}
6119
Willy Tarreau40dde2d2022-06-19 16:41:59 +02006120#if defined(RTLD_DEFAULT) || defined(RTLD_NEXT)
6121/* redefine dlopen() so that we can detect unexpected replacement of some
6122 * critical symbols, typically init/alloc/free functions coming from alternate
6123 * libraries. When called, a tainted flag is set (TAINTED_SHARED_LIBS).
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006124 * It's important to understand that the dynamic linker will present the
6125 * first loaded of each symbol to all libs, so that if haproxy is linked
6126 * with a new lib that uses a static inline or a #define to replace an old
6127 * function, and a dependency was linked against an older version of that
6128 * lib that had a function there, that lib would use all of the newer
6129 * versions of the functions that are already loaded in haproxy, except
6130 * for that unique function which would continue to be the old one. This
6131 * creates all sort of problems when init code allocates smaller structs
6132 * than required for example but uses new functions on them, etc. Thus what
6133 * we do here is to try to detect API consistency: we take a fingerprint of
6134 * a number of known functions, and verify that if they change in a loaded
6135 * library, either there all appeared or all disappeared, but not partially.
6136 * We can check up to 64 symbols that belong to individual groups that are
6137 * checked together.
Willy Tarreau40dde2d2022-06-19 16:41:59 +02006138 */
6139void *dlopen(const char *filename, int flags)
6140{
6141 static void *(*_dlopen)(const char *filename, int flags);
Willy Tarreau177aed52022-06-19 16:49:51 +02006142 struct {
6143 const char *name;
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006144 uint64_t bit, grp;
Willy Tarreau177aed52022-06-19 16:49:51 +02006145 void *curr, *next;
6146 } check_syms[] = {
Willy Tarreau997ad152023-04-22 19:47:19 +02006147 /* openssl's libcrypto checks: group bits 0x1f */
6148 { .name="OPENSSL_init", .bit = 0x0000000000000001, .grp = 0x000000000000001f, }, // openssl 1.0 / 1.1 / 3.0
6149 { .name="OPENSSL_init_crypto", .bit = 0x0000000000000002, .grp = 0x000000000000001f, }, // openssl 1.1 / 3.0
6150 { .name="ENGINE_init", .bit = 0x0000000000000004, .grp = 0x000000000000001f, }, // openssl 1.x / 3.x with engine
6151 { .name="EVP_CIPHER_CTX_init", .bit = 0x0000000000000008, .grp = 0x000000000000001f, }, // openssl 1.0
6152 { .name="HMAC_Init", .bit = 0x0000000000000010, .grp = 0x000000000000001f, }, // openssl 1.x
6153
6154 /* openssl's libssl checks: group bits 0x3e0 */
6155 { .name="OPENSSL_init_ssl", .bit = 0x0000000000000020, .grp = 0x00000000000003e0, }, // openssl 1.1 / 3.0
6156 { .name="SSL_library_init", .bit = 0x0000000000000040, .grp = 0x00000000000003e0, }, // openssl 1.x
6157 { .name="SSL_is_quic", .bit = 0x0000000000000080, .grp = 0x00000000000003e0, }, // quictls
6158 { .name="SSL_CTX_new_ex", .bit = 0x0000000000000100, .grp = 0x00000000000003e0, }, // openssl 3.x
6159 { .name="SSL_CTX_get0_security_ex_data", .bit = 0x0000000000000200, .grp = 0x00000000000003e0, }, // openssl 1.x / 3.x
6160
Willy Tarreau177aed52022-06-19 16:49:51 +02006161 /* insert only above, 0 must be the last one */
6162 { 0 },
6163 };
6164 const char *trace;
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006165 uint64_t own_fp, lib_fp; // symbols fingerprints
Willy Tarreau177aed52022-06-19 16:49:51 +02006166 void *addr;
Willy Tarreau40dde2d2022-06-19 16:41:59 +02006167 void *ret;
Willy Tarreau177aed52022-06-19 16:49:51 +02006168 int sym = 0;
Willy Tarreau40dde2d2022-06-19 16:41:59 +02006169
6170 if (!_dlopen) {
6171 _dlopen = get_sym_next_addr("dlopen");
6172 if (!_dlopen || _dlopen == dlopen) {
6173 _dlopen = NULL;
6174 return NULL;
6175 }
6176 }
6177
Willy Tarreau177aed52022-06-19 16:49:51 +02006178 /* save a few pointers to critical symbols. We keep a copy of both the
6179 * current and the next value, because we might already have replaced
Willy Tarreau58912b82023-03-22 15:47:51 +01006180 * some of them in an inconsistent way (i.e. not all), and we're only
Willy Tarreau177aed52022-06-19 16:49:51 +02006181 * interested in verifying that a loaded library doesn't come with a
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006182 * completely different definition that would be incompatible. We'll
6183 * keep a fingerprint of our own symbols.
Willy Tarreau177aed52022-06-19 16:49:51 +02006184 */
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006185 own_fp = 0;
Willy Tarreau177aed52022-06-19 16:49:51 +02006186 for (sym = 0; check_syms[sym].name; sym++) {
6187 check_syms[sym].curr = get_sym_curr_addr(check_syms[sym].name);
6188 check_syms[sym].next = get_sym_next_addr(check_syms[sym].name);
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006189 if (check_syms[sym].curr || check_syms[sym].next)
6190 own_fp |= check_syms[sym].bit;
Willy Tarreau177aed52022-06-19 16:49:51 +02006191 }
Willy Tarreau40dde2d2022-06-19 16:41:59 +02006192
6193 /* now open the requested lib */
6194 ret = _dlopen(filename, flags);
6195 if (!ret)
6196 return ret;
6197
6198 mark_tainted(TAINTED_SHARED_LIBS);
6199
Willy Tarreau177aed52022-06-19 16:49:51 +02006200 /* and check that critical symbols didn't change */
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006201 lib_fp = 0;
Willy Tarreau177aed52022-06-19 16:49:51 +02006202 for (sym = 0; check_syms[sym].name; sym++) {
Willy Tarreau177aed52022-06-19 16:49:51 +02006203 addr = dlsym(ret, check_syms[sym].name);
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006204 if (addr)
6205 lib_fp |= check_syms[sym].bit;
6206 }
Willy Tarreau177aed52022-06-19 16:49:51 +02006207
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006208 if (lib_fp != own_fp) {
6209 /* let's check what changed: */
6210 uint64_t mask = 0;
Willy Tarreau177aed52022-06-19 16:49:51 +02006211
Willy Tarreauc3b297d2023-03-22 17:01:32 +01006212 for (sym = 0; check_syms[sym].name; sym++) {
6213 mask = check_syms[sym].grp;
6214
6215 /* new group of symbols. If they all appeared together
6216 * their use will be consistent. If none appears, it's
6217 * just that the lib doesn't use them. If some appear
6218 * or disappear, it means the lib relies on a different
6219 * dependency and will end up with a mix.
6220 */
6221 if (!(own_fp & mask) || !(lib_fp & mask) ||
6222 (own_fp & mask) == (lib_fp & mask))
6223 continue;
6224
6225 /* let's report a symbol that really changes */
6226 if (!((own_fp ^ lib_fp) & check_syms[sym].bit))
6227 continue;
6228
6229 /* OK it's clear that this symbol was redefined */
6230 mark_tainted(TAINTED_REDEFINITION);
6231
6232 trace = hlua_show_current_location("\n ");
6233 ha_warning("dlopen(): shared library '%s' brings a different and inconsistent definition of symbol '%s'. The process cannot be trusted anymore!%s%s\n",
6234 filename, check_syms[sym].name,
6235 trace ? " Suspected call location: \n " : "",
6236 trace ? trace : "");
6237 }
Willy Tarreau177aed52022-06-19 16:49:51 +02006238 }
6239
Willy Tarreau40dde2d2022-06-19 16:41:59 +02006240 return ret;
6241}
6242#endif
6243
Willy Tarreau06e69b52021-03-02 14:01:35 +01006244static int init_tools_per_thread()
6245{
6246 /* Let's make each thread start from a different position */
Willy Tarreaue936aaf2024-03-01 16:17:47 +01006247 statistical_prng_state += ha_random32();
Willy Tarreau06e69b52021-03-02 14:01:35 +01006248 if (!statistical_prng_state)
6249 statistical_prng_state++;
6250 return 1;
6251}
6252REGISTER_PER_THREAD_INIT(init_tools_per_thread);
Willy Tarreauc54e5ad2020-06-25 09:15:40 +02006253
Willy Tarreaubaaee002006-06-26 02:48:02 +02006254/*
6255 * Local variables:
6256 * c-indent-level: 8
6257 * c-basic-offset: 8
6258 * End:
6259 */