4 Copyright (C) Andrew Tridgell 2006
6 This program is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 3 of the License, or
9 (at your option) any later version.
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, see <http://www.gnu.org/licenses/>.
21 #include "lib/tevent/tevent.h"
22 #include "lib/tdb/include/tdb.h"
23 #include "system/network.h"
24 #include "system/filesys.h"
25 #include "system/wait.h"
26 #include "system/shmem.h"
27 #include "../include/ctdb_private.h"
29 int LogLevel = DEBUG_NOTICE;
30 int this_log_level = 0;
33 return error string for last error
35 const char *ctdb_errstr(struct ctdb_context *ctdb)
42 remember an error message
44 void ctdb_set_error(struct ctdb_context *ctdb, const char *fmt, ...)
47 talloc_free(ctdb->err_msg);
49 ctdb->err_msg = talloc_vasprintf(ctdb, fmt, ap);
50 DEBUG(DEBUG_ERR,("ctdb error: %s\n", ctdb->err_msg));
55 a fatal internal error occurred - no hope for recovery
57 void ctdb_fatal(struct ctdb_context *ctdb, const char *msg)
59 DEBUG(DEBUG_ALERT,("ctdb fatal error: %s\n", msg));
66 int ctdb_parse_address(struct ctdb_context *ctdb,
67 TALLOC_CTX *mem_ctx, const char *str,
68 struct ctdb_address *address)
73 se = getservbyname("ctdb", "tcp");
76 address->address = talloc_strdup(mem_ctx, str);
77 CTDB_NO_MEMORY(ctdb, address->address);
80 address->port = CTDB_PORT;
82 address->port = ntohs(se->s_port);
89 check if two addresses are the same
91 bool ctdb_same_address(struct ctdb_address *a1, struct ctdb_address *a2)
93 return strcmp(a1->address, a2->address) == 0 && a1->port == a2->port;
98 hash function for mapping data to a VNN - taken from tdb
100 uint32_t ctdb_hash(const TDB_DATA *key)
102 return tdb_jenkins_hash(discard_const(key));
106 a type checking varient of idr_find
108 static void *_idr_find_type(struct idr_context *idp, int id, const char *type, const char *location)
110 void *p = idr_find(idp, id);
111 if (p && talloc_check_name(p, type) == NULL) {
112 DEBUG(DEBUG_ERR,("%s idr_find_type expected type %s but got %s\n",
113 location, type, talloc_get_name(p)));
119 uint32_t ctdb_reqid_new(struct ctdb_context *ctdb, void *state)
121 int id = idr_get_new_above(ctdb->idr, state, ctdb->lastid+1, INT_MAX);
123 DEBUG(DEBUG_DEBUG, ("Reqid wrap!\n"));
124 id = idr_get_new(ctdb->idr, state, INT_MAX);
130 void *_ctdb_reqid_find(struct ctdb_context *ctdb, uint32_t reqid, const char *type, const char *location)
134 p = _idr_find_type(ctdb->idr, reqid, type, location);
136 DEBUG(DEBUG_WARNING, ("Could not find idr:%u\n",reqid));
143 void ctdb_reqid_remove(struct ctdb_context *ctdb, uint32_t reqid)
147 ret = idr_remove(ctdb->idr, reqid);
149 DEBUG(DEBUG_ERR, ("Removing idr that does not exist\n"));
155 form a ctdb_rec_data record from a key/data pair
157 note that header may be NULL. If not NULL then it is included in the data portion
160 struct ctdb_rec_data *ctdb_marshall_record(TALLOC_CTX *mem_ctx, uint32_t reqid,
162 struct ctdb_ltdb_header *header,
166 struct ctdb_rec_data *d;
168 length = offsetof(struct ctdb_rec_data, data) + key.dsize +
169 data.dsize + (header?sizeof(*header):0);
170 d = (struct ctdb_rec_data *)talloc_size(mem_ctx, length);
176 d->keylen = key.dsize;
177 memcpy(&d->data[0], key.dptr, key.dsize);
179 d->datalen = data.dsize + sizeof(*header);
180 memcpy(&d->data[key.dsize], header, sizeof(*header));
181 memcpy(&d->data[key.dsize+sizeof(*header)], data.dptr, data.dsize);
183 d->datalen = data.dsize;
184 memcpy(&d->data[key.dsize], data.dptr, data.dsize);
190 /* helper function for marshalling multiple records */
191 struct ctdb_marshall_buffer *ctdb_marshall_add(TALLOC_CTX *mem_ctx,
192 struct ctdb_marshall_buffer *m,
196 struct ctdb_ltdb_header *header,
199 struct ctdb_rec_data *r;
200 size_t m_size, r_size;
201 struct ctdb_marshall_buffer *m2;
203 r = ctdb_marshall_record(mem_ctx, reqid, key, header, data);
210 m = talloc_zero_size(mem_ctx, offsetof(struct ctdb_marshall_buffer, data));
217 m_size = talloc_get_size(m);
218 r_size = talloc_get_size(r);
220 m2 = talloc_realloc_size(mem_ctx, m, m_size + r_size);
226 memcpy(m_size + (uint8_t *)m2, r, r_size);
235 /* we've finished marshalling, return a data blob with the marshalled records */
236 TDB_DATA ctdb_marshall_finish(struct ctdb_marshall_buffer *m)
239 data.dptr = (uint8_t *)m;
240 data.dsize = talloc_get_size(m);
245 loop over a marshalling buffer
247 - pass r==NULL to start
248 - loop the number of times indicated by m->count
250 struct ctdb_rec_data *ctdb_marshall_loop_next(struct ctdb_marshall_buffer *m, struct ctdb_rec_data *r,
252 struct ctdb_ltdb_header *header,
253 TDB_DATA *key, TDB_DATA *data)
256 r = (struct ctdb_rec_data *)&m->data[0];
258 r = (struct ctdb_rec_data *)(r->length + (uint8_t *)r);
266 key->dptr = &r->data[0];
267 key->dsize = r->keylen;
270 data->dptr = &r->data[r->keylen];
271 data->dsize = r->datalen;
272 if (header != NULL) {
273 data->dptr += sizeof(*header);
274 data->dsize -= sizeof(*header);
278 if (header != NULL) {
279 if (r->datalen < sizeof(*header)) {
282 *header = *(struct ctdb_ltdb_header *)&r->data[r->keylen];
289 if possible, make this task very high priority
291 void ctdb_high_priority(struct ctdb_context *ctdb)
294 if (nice(-20) == -1 && errno != 0) {
295 DEBUG(DEBUG_WARNING,("Unable to renice self: %s\n",
298 DEBUG(DEBUG_NOTICE,("Scheduler says I'm nice: %i\n",
299 getpriority(PRIO_PROCESS, getpid())));
304 make ourselves slightly nicer: eg. a ctdb child.
306 void ctdb_reduce_priority(struct ctdb_context *ctdb)
309 if (nice(10) == -1 && errno != 0) {
310 DEBUG(DEBUG_WARNING,("Unable to lower priority: %s\n",
315 void set_nonblocking(int fd)
318 v = fcntl(fd, F_GETFL, 0);
319 fcntl(fd, F_SETFL, v | O_NONBLOCK);
322 void set_close_on_exec(int fd)
325 v = fcntl(fd, F_GETFD, 0);
326 fcntl(fd, F_SETFD, v | FD_CLOEXEC);
330 bool parse_ipv4(const char *s, unsigned port, struct sockaddr_in *sin)
332 sin->sin_family = AF_INET;
333 sin->sin_port = htons(port);
335 if (inet_pton(AF_INET, s, &sin->sin_addr) != 1) {
336 DEBUG(DEBUG_ERR, (__location__ " Failed to translate %s into sin_addr\n", s));
343 static bool parse_ipv6(const char *s, const char *ifaces, unsigned port, ctdb_sock_addr *saddr)
345 saddr->ip6.sin6_family = AF_INET6;
346 saddr->ip6.sin6_port = htons(port);
347 saddr->ip6.sin6_flowinfo = 0;
348 saddr->ip6.sin6_scope_id = 0;
350 if (inet_pton(AF_INET6, s, &saddr->ip6.sin6_addr) != 1) {
351 DEBUG(DEBUG_ERR, (__location__ " Failed to translate %s into sin6_addr\n", s));
355 if (ifaces && IN6_IS_ADDR_LINKLOCAL(&saddr->ip6.sin6_addr)) {
356 if (strchr(ifaces, ',')) {
357 DEBUG(DEBUG_ERR, (__location__ " Link local address %s "
358 "is specified for multiple ifaces %s\n",
362 saddr->ip6.sin6_scope_id = if_nametoindex(ifaces);
370 bool parse_ip_port(const char *addr, ctdb_sock_addr *saddr)
372 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
378 s = talloc_strdup(tmp_ctx, addr);
380 DEBUG(DEBUG_ERR, (__location__ " Failed strdup()\n"));
381 talloc_free(tmp_ctx);
387 DEBUG(DEBUG_ERR, (__location__ " This addr: %s does not contain a port number\n", s));
388 talloc_free(tmp_ctx);
392 port = strtoul(p+1, &endp, 10);
393 if (endp == NULL || *endp != 0) {
394 /* trailing garbage */
395 DEBUG(DEBUG_ERR, (__location__ " Trailing garbage after the port in %s\n", s));
396 talloc_free(tmp_ctx);
402 /* now is this a ipv4 or ipv6 address ?*/
403 ret = parse_ip(s, NULL, port, saddr);
405 talloc_free(tmp_ctx);
412 bool parse_ip(const char *addr, const char *ifaces, unsigned port, ctdb_sock_addr *saddr)
417 /* now is this a ipv4 or ipv6 address ?*/
418 p = index(addr, ':');
420 ret = parse_ipv4(addr, port, &saddr->ip);
422 ret = parse_ipv6(addr, ifaces, port, saddr);
431 bool parse_ip_mask(const char *str, const char *ifaces, ctdb_sock_addr *addr, unsigned *mask)
433 TALLOC_CTX *tmp_ctx = talloc_new(NULL);
439 s = talloc_strdup(tmp_ctx, str);
441 DEBUG(DEBUG_ERR, (__location__ " Failed strdup()\n"));
442 talloc_free(tmp_ctx);
448 DEBUG(DEBUG_ERR, (__location__ " This addr: %s does not contain a mask\n", s));
449 talloc_free(tmp_ctx);
453 *mask = strtoul(p+1, &endp, 10);
454 if (endp == NULL || *endp != 0) {
455 /* trailing garbage */
456 DEBUG(DEBUG_ERR, (__location__ " Trailing garbage after the mask in %s\n", s));
457 talloc_free(tmp_ctx);
463 /* now is this a ipv4 or ipv6 address ?*/
464 ret = parse_ip(s, ifaces, 0, addr);
466 talloc_free(tmp_ctx);
471 This is used to canonicalize a ctdb_sock_addr structure.
473 void ctdb_canonicalize_ip(const ctdb_sock_addr *ip, ctdb_sock_addr *cip)
475 char prefix[12] = { 0,0,0,0,0,0,0,0,0,0,0xff,0xff };
477 memcpy(cip, ip, sizeof (*cip));
479 if ( (ip->sa.sa_family == AF_INET6)
480 && !memcmp(&ip->ip6.sin6_addr, prefix, 12)) {
481 memset(cip, 0, sizeof(*cip));
482 #ifdef HAVE_SOCK_SIN_LEN
483 cip->ip.sin_len = sizeof(*cip);
485 cip->ip.sin_family = AF_INET;
486 cip->ip.sin_port = ip->ip6.sin6_port;
487 memcpy(&cip->ip.sin_addr, &ip->ip6.sin6_addr.s6_addr32[3], 4);
491 bool ctdb_same_ip(const ctdb_sock_addr *tip1, const ctdb_sock_addr *tip2)
493 ctdb_sock_addr ip1, ip2;
495 ctdb_canonicalize_ip(tip1, &ip1);
496 ctdb_canonicalize_ip(tip2, &ip2);
498 if (ip1.sa.sa_family != ip2.sa.sa_family) {
502 switch (ip1.sa.sa_family) {
504 return ip1.ip.sin_addr.s_addr == ip2.ip.sin_addr.s_addr;
506 return !memcmp(&ip1.ip6.sin6_addr.s6_addr[0],
507 &ip2.ip6.sin6_addr.s6_addr[0],
510 DEBUG(DEBUG_ERR, (__location__ " CRITICAL Can not compare sockaddr structures of type %u\n", ip1.sa.sa_family));
518 compare two ctdb_sock_addr structures
520 bool ctdb_same_sockaddr(const ctdb_sock_addr *ip1, const ctdb_sock_addr *ip2)
522 return ctdb_same_ip(ip1, ip2) && ip1->ip.sin_port == ip2->ip.sin_port;
525 char *ctdb_addr_to_str(ctdb_sock_addr *addr)
527 static char cip[128] = "";
529 switch (addr->sa.sa_family) {
531 inet_ntop(addr->ip.sin_family, &addr->ip.sin_addr, cip, sizeof(cip));
534 inet_ntop(addr->ip6.sin6_family, &addr->ip6.sin6_addr, cip, sizeof(cip));
537 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family %u\n", addr->sa.sa_family));
543 unsigned ctdb_addr_to_port(ctdb_sock_addr *addr)
545 switch (addr->sa.sa_family) {
547 return ntohs(addr->ip.sin_port);
550 return ntohs(addr->ip6.sin6_port);
553 DEBUG(DEBUG_ERR, (__location__ " ERROR, unknown family %u\n", addr->sa.sa_family));
559 void ctdb_block_signal(int signum)
563 sigaddset(&set,signum);
564 sigprocmask(SIG_BLOCK,&set,NULL);
567 void ctdb_unblock_signal(int signum)
571 sigaddset(&set,signum);
572 sigprocmask(SIG_UNBLOCK,&set,NULL);
575 struct debug_levels debug_levels[] = {
576 {DEBUG_EMERG, "EMERG"},
577 {DEBUG_ALERT, "ALERT"},
578 {DEBUG_CRIT, "CRIT"},
580 {DEBUG_WARNING, "WARNING"},
581 {DEBUG_NOTICE, "NOTICE"},
582 {DEBUG_INFO, "INFO"},
583 {DEBUG_DEBUG, "DEBUG"},
587 const char *get_debug_by_level(int32_t level)
591 for (i=0; debug_levels[i].description != NULL; i++) {
592 if (debug_levels[i].level == level) {
593 return debug_levels[i].description;
599 int32_t get_debug_by_desc(const char *desc)
603 for (i=0; debug_levels[i].description != NULL; i++) {
604 if (!strcmp(debug_levels[i].description, desc)) {
605 return debug_levels[i].level;
612 /* we don't lock future pages here; it would increase the chance that
613 * we'd fail to mmap later on. */
614 void ctdb_lockdown_memory(struct ctdb_context *ctdb)
617 /* Extra stack, please! */
619 memset(dummy, 0, sizeof(dummy));
621 if (ctdb->valgrinding) {
625 /* Avoid compiler optimizing out dummy. */
626 mlock(dummy, sizeof(dummy));
627 if (mlockall(MCL_CURRENT) != 0) {
628 DEBUG(DEBUG_WARNING,("Failed to lock memory: %s'\n",
634 const char *ctdb_eventscript_call_names[] = {