json: Add conditionals to avoid locking if Jansson is thread safe.
[asterisk/asterisk.git] / main / acl.c
index f3fd509..c131551 100644 (file)
@@ -1,7 +1,7 @@
 /*
  * Asterisk -- An open source telephony toolkit.
  *
- * Copyright (C) 1999 - 2006, Digium, Inc.
+ * Copyright (C) 1999 - 2012, Digium, Inc.
  *
  * Mark Spencer <markster@digium.com>
  *
  * \author Mark Spencer <markster@digium.com>
  */
 
+/*** MODULEINFO
+       <support_level>core</support_level>
+ ***/
+
 #include "asterisk.h"
 
-ASTERISK_FILE_VERSION(__FILE__, "$Revision$")
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <sys/time.h>
-#include <signal.h>
-#include <errno.h>
-#include <unistd.h>
-#include <netinet/in.h>
-#include <arpa/inet.h>
-#include <sys/socket.h>
-#include <netdb.h>
-#include <net/if.h>
-#include <netinet/in.h>
-#include <netinet/in_systm.h>
-#include <netinet/ip.h>
-#include <sys/ioctl.h>
+#include "asterisk/network.h"
 
-#if defined(__OpenBSD__) || defined(__NetBSD__) || defined(__FreeBSD__)
+#if defined(__OpenBSD__) || defined(__NetBSD__) || defined(__FreeBSD__) || defined(__Darwin__)
 #include <fcntl.h>
 #include <net/route.h>
 #endif
 
 #if defined(SOLARIS)
 #include <sys/sockio.h>
-#endif
-
-/* netinet/ip.h may not define the following (See RFCs 791 and 1349) */
-#if !defined(IPTOS_LOWCOST)
-#define       IPTOS_LOWCOST           0x02
-#endif
-
-#if !defined(IPTOS_MINCOST)
-#define       IPTOS_MINCOST           IPTOS_LOWCOST
+#include <net/if.h>
+#elif defined(HAVE_GETIFADDRS)
+#include <ifaddrs.h>
 #endif
 
 #include "asterisk/acl.h"
-#include "asterisk/logger.h"
 #include "asterisk/channel.h"
-#include "asterisk/options.h"
 #include "asterisk/utils.h"
 #include "asterisk/lock.h"
 #include "asterisk/srv.h"
 
-struct ast_ha {
-       /* Host access rule */
-       struct in_addr netaddr;
-       struct in_addr netmask;
-       int sense;
-       struct ast_ha *next;
-};
+#if (!defined(SOLARIS) && !defined(HAVE_GETIFADDRS))
+static int get_local_address(struct ast_sockaddr *ourip)
+{
+       return -1;
+}
+#else
+static void score_address(const struct sockaddr_in *sin, struct in_addr *best_addr, int *best_score)
+{
+       const char *address;
+       int score;
 
-/* Default IP - if not otherwise set, don't breathe garbage */
-static struct in_addr __ourip = { 0x00000000 };
+       address = ast_inet_ntoa(sin->sin_addr);
 
-struct my_ifreq {
-       char ifrn_name[IFNAMSIZ];       /* Interface name, e.g. "eth0", "ppp0", etc.  */
-       struct sockaddr_in ifru_addr;
-};
+       /* RFC 1700 alias for the local network */
+       if (address[0] == '0') {
+               score = -25;
+       /* RFC 1700 localnet */
+       } else if (strncmp(address, "127", 3) == 0) {
+               score = -20;
+       /* RFC 1918 non-public address space */
+       } else if (strncmp(address, "10.", 3) == 0) {
+               score = -5;
+       /* RFC 1918 non-public address space */
+       } else if (strncmp(address, "172", 3) == 0) {
+               /* 172.16.0.0 - 172.19.255.255, but not 172.160.0.0 - 172.169.255.255 */
+               if (address[4] == '1' && address[5] >= '6' && address[6] == '.') {
+                       score = -5;
+               /* 172.20.0.0 - 172.29.255.255, but not 172.200.0.0 - 172.255.255.255 nor 172.2.0.0 - 172.2.255.255 */
+               } else if (address[4] == '2' && address[6] == '.') {
+                       score = -5;
+               /* 172.30.0.0 - 172.31.255.255, but not 172.3.0.0 - 172.3.255.255 */
+               } else if (address[4] == '3' && (address[5] == '0' || address[5] == '1')) {
+                       score = -5;
+               /* All other 172 addresses are public */
+               } else {
+                       score = 0;
+               }
+       /* RFC 2544 Benchmark test range (198.18.0.0 - 198.19.255.255, but not 198.180.0.0 - 198.199.255.255) */
+       } else if (strncmp(address, "198.1", 5) == 0 && address[5] >= '8' && address[6] == '.') {
+               score = -10;
+       /* RFC 1918 non-public address space */
+       } else if (strncmp(address, "192.168", 7) == 0) {
+               score = -5;
+       /* RFC 3330 Zeroconf network */
+       } else if (strncmp(address, "169.254", 7) == 0) {
+               /*!\note Better score than a test network, but not quite as good as RFC 1918
+                * address space.  The reason is that some Linux distributions automatically
+                * configure a Zeroconf address before trying DHCP, so we want to prefer a
+                * DHCP lease to a Zeroconf address.
+                */
+               score = -10;
+       /* RFC 3330 Test network */
+       } else if (strncmp(address, "192.0.2.", 8) == 0) {
+               score = -15;
+       /* Every other address should be publically routable */
+       } else {
+               score = 0;
+       }
+
+       if (score > *best_score) {
+               *best_score = score;
+               memcpy(best_addr, &sin->sin_addr, sizeof(*best_addr));
+       }
+}
+
+static int get_local_address(struct ast_sockaddr *ourip)
+{
+       int s, res = -1;
+#ifdef SOLARIS
+       struct lifreq *ifr = NULL;
+       struct lifnum ifn;
+       struct lifconf ifc;
+       struct sockaddr_in *sa;
+       char *buf = NULL;
+       int bufsz, x;
+#endif /* SOLARIS */
+#if defined(__OpenBSD__) || defined(__NetBSD__) || defined(__FreeBSD__) || defined(__linux__) || defined(__Darwin__) || defined(__GLIBC__)
+       struct ifaddrs *ifap, *ifaphead;
+       int rtnerr;
+       const struct sockaddr_in *sin;
+#endif /* BSD_OR_LINUX */
+       struct in_addr best_addr;
+       int best_score = -100;
+       memset(&best_addr, 0, sizeof(best_addr));
+
+#if defined(__OpenBSD__) || defined(__NetBSD__) || defined(__FreeBSD__) || defined(__linux__) || defined(__Darwin__) || defined(__GLIBC__)
+       rtnerr = getifaddrs(&ifaphead);
+       if (rtnerr) {
+               perror(NULL);
+               return -1;
+       }
+#endif /* BSD_OR_LINUX */
+
+       s = socket(AF_INET, SOCK_STREAM, 0);
+
+       if (s > 0) {
+#if defined(__OpenBSD__) || defined(__NetBSD__) || defined(__FreeBSD__) || defined(__linux__) || defined(__Darwin__) || defined(__GLIBC__)
+               for (ifap = ifaphead; ifap; ifap = ifap->ifa_next) {
+
+                       if (ifap->ifa_addr && ifap->ifa_addr->sa_family == AF_INET) {
+                               sin = (const struct sockaddr_in *) ifap->ifa_addr;
+                               score_address(sin, &best_addr, &best_score);
+                               res = 0;
+
+                               if (best_score == 0) {
+                                       break;
+                               }
+                       }
+               }
+#endif /* BSD_OR_LINUX */
+
+               /* There is no reason whatsoever that this shouldn't work on Linux or BSD also. */
+#ifdef SOLARIS
+               /* Get a count of interfaces on the machine */
+               ifn.lifn_family = AF_INET;
+               ifn.lifn_flags = 0;
+               ifn.lifn_count = 0;
+               if (ioctl(s, SIOCGLIFNUM, &ifn) < 0) {
+                       close(s);
+                       return -1;
+               }
+
+               bufsz = ifn.lifn_count * sizeof(struct lifreq);
+               if (!(buf = ast_malloc(bufsz))) {
+                       close(s);
+                       return -1;
+               }
+               memset(buf, 0, bufsz);
+
+               /* Get a list of interfaces on the machine */
+               ifc.lifc_len = bufsz;
+               ifc.lifc_buf = buf;
+               ifc.lifc_family = AF_INET;
+               ifc.lifc_flags = 0;
+               if (ioctl(s, SIOCGLIFCONF, &ifc) < 0) {
+                       close(s);
+                       ast_free(buf);
+                       return -1;
+               }
+
+               for (ifr = ifc.lifc_req, x = 0; x < ifn.lifn_count; ifr++, x++) {
+                       sa = (struct sockaddr_in *)&(ifr->lifr_addr);
+                       score_address(sa, &best_addr, &best_score);
+                       res = 0;
+
+                       if (best_score == 0) {
+                               break;
+                       }
+               }
+
+               ast_free(buf);
+#endif /* SOLARIS */
+
+               close(s);
+       }
+#if defined(__OpenBSD__) || defined(__NetBSD__) || defined(__FreeBSD__) || defined(__linux__) || defined(__Darwin__)
+       freeifaddrs(ifaphead);
+#endif /* BSD_OR_LINUX */
+
+       if (res == 0 && ourip) {
+               ast_sockaddr_setnull(ourip);
+               ourip->ss.ss_family = AF_INET;
+               ((struct sockaddr_in *)&ourip->ss)->sin_addr = best_addr;
+       }
+       return res;
+}
+#endif /* HAVE_GETIFADDRS */
 
 /* Free HA structure */
 void ast_free_ha(struct ast_ha *ha)
@@ -93,15 +224,37 @@ void ast_free_ha(struct ast_ha *ha)
        while (ha) {
                hal = ha;
                ha = ha->next;
-               free(hal);
+               ast_free(hal);
+       }
+}
+
+/* Free ACL list structure */
+struct ast_acl_list *ast_free_acl_list(struct ast_acl_list *acl_list)
+{
+       struct ast_acl *current;
+
+       if (!acl_list) {
+               return NULL;
+       }
+
+       AST_LIST_LOCK(acl_list);
+       while ((current = AST_LIST_REMOVE_HEAD(acl_list, list))) {
+               ast_free_ha(current->acl);
+               ast_free(current);
        }
+       AST_LIST_UNLOCK(acl_list);
+
+       AST_LIST_HEAD_DESTROY(acl_list);
+       ast_free(acl_list);
+
+       return NULL;
 }
 
 /* Copy HA structure */
-static void ast_copy_ha(struct ast_ha *from, struct ast_ha *to)
+void ast_copy_ha(const struct ast_ha *from, struct ast_ha *to)
 {
-       memcpy(&to->netaddr, &from->netaddr, sizeof(from->netaddr));
-       memcpy(&to->netmask, &from->netmask, sizeof(from->netmask));
+       ast_sockaddr_copy(&to->addr, &from->addr);
+       ast_sockaddr_copy(&to->netmask, &from->netmask);
        to->sense = from->sense;
 }
 
@@ -110,7 +263,7 @@ static struct ast_ha *ast_duplicate_ha(struct ast_ha *original)
 {
        struct ast_ha *new_ha;
 
-       if ((new_ha = ast_malloc(sizeof(*new_ha)))) {
+       if ((new_ha = ast_calloc(1, sizeof(*new_ha)))) {
                /* Copy from original to new object */
                ast_copy_ha(original, new_ha);
        }
@@ -124,125 +277,588 @@ struct ast_ha *ast_duplicate_ha_list(struct ast_ha *original)
 {
        struct ast_ha *start = original;
        struct ast_ha *ret = NULL;
-       struct ast_ha *link, *prev = NULL;
+       struct ast_ha *current, *prev = NULL;
 
        while (start) {
-               link = ast_duplicate_ha(start);  /* Create copy of this object */
-               if (prev)
-                       prev->next = link;              /* Link previous to this object */
+               current = ast_duplicate_ha(start);  /* Create copy of this object */
+               if (!current) {
+                       ast_free_ha(ret);
+
+                       return NULL;
+               }
+
+               if (prev) {
+                       prev->next = current;           /* Link previous to this object */
+               }
+
+               if (!ret) {
+                       ret = current;                  /* Save starting point */
+               }
+
+               start = start->next;                /* Go to next object */
+               prev = current;                     /* Save pointer to this object */
+       }
+       return ret;                             /* Return start of list */
+}
 
-               if (!ret)
-                       ret = link;             /* Save starting point */
+static int acl_new(struct ast_acl **pointer, const char *name) {
+       struct ast_acl *acl;
+       if (!(acl = ast_calloc(1, sizeof(*acl)))) {
+               return 1;
+       }
+
+       *pointer = acl;
+       ast_copy_string(acl->name, name, ACL_NAME_LENGTH);
+       return 0;
+}
+
+struct ast_acl_list *ast_duplicate_acl_list(struct ast_acl_list *original)
+{
+       struct ast_acl_list *clone;
+       struct ast_acl *current_cursor;
+       struct ast_acl *current_clone;
+
+       /* Early return if we receive a duplication request for a NULL original. */
+       if (!original) {
+               return NULL;
+       }
+
+       if (!(clone = ast_calloc(1, sizeof(*clone)))) {
+               ast_log(LOG_ERROR, "Failed to allocate ast_acl_list struct while cloning an ACL\n");
+               return NULL;
+       }
+       AST_LIST_HEAD_INIT(clone);
+
+       AST_LIST_LOCK(original);
+
+       AST_LIST_TRAVERSE(original, current_cursor, list) {
+               if ((acl_new(&current_clone, current_cursor->name))) {
+                       ast_log(LOG_ERROR, "Failed to allocate ast_acl struct while cloning an ACL.\n");
+                       ast_free_acl_list(clone);
+                       clone = NULL;
+                       break;
+               }
+
+               /* Copy data from original ACL to clone ACL */
+               current_clone->acl = ast_duplicate_ha_list(current_cursor->acl);
 
-               start = start->next;            /* Go to next object */
-               prev = link;                    /* Save pointer to this object */
+               current_clone->is_invalid = current_cursor->is_invalid;
+               current_clone->is_realtime = current_cursor->is_realtime;
+
+               AST_LIST_INSERT_TAIL(clone, current_clone, list);
+
+               if (current_cursor->acl && !current_clone->acl) {
+                       /* Deal with failure after adding to clone so we don't have to free
+                        * current_clone separately. */
+                       ast_log(LOG_ERROR, "Failed to duplicate HA list while cloning ACL.\n");
+                       ast_free_acl_list(clone);
+                       clone = NULL;
+                       break;
+               }
        }
-       return ret;                     /* Return start of list */
+
+       AST_LIST_UNLOCK(original);
+
+       return clone;
 }
 
-struct ast_ha *ast_append_ha(char *sense, char *stuff, struct ast_ha *path)
+/*!
+ * \brief
+ * Parse a netmask in CIDR notation
+ *
+ * \details
+ * For a mask of an IPv4 address, this should be a number between 0 and 32. For
+ * a mask of an IPv6 address, this should be a number between 0 and 128. This
+ * function creates an IPv6 ast_sockaddr from the given netmask. For masks of
+ * IPv4 addresses, this is accomplished by adding 96 to the original netmask.
+ *
+ * \param[out] addr The ast_sockaddr produced from the CIDR netmask
+ * \param is_v4 Tells if the address we are masking is IPv4.
+ * \param mask_str The CIDR mask to convert
+ * \retval -1 Failure
+ * \retval 0 Success
+ */
+static int parse_cidr_mask(struct ast_sockaddr *addr, int is_v4, const char *mask_str)
+{
+       int mask;
+
+       if (sscanf(mask_str, "%30d", &mask) != 1) {
+               return -1;
+       }
+
+       if (is_v4) {
+               struct sockaddr_in sin;
+               if (mask < 0 || mask > 32) {
+                       return -1;
+               }
+               memset(&sin, 0, sizeof(sin));
+               sin.sin_family = AF_INET;
+               /* If mask is 0, then we already have the
+                * appropriate all 0s address in sin from
+                * the above memset.
+                */
+               if (mask != 0) {
+                       sin.sin_addr.s_addr = htonl(0xFFFFFFFF << (32 - mask));
+               }
+               ast_sockaddr_from_sin(addr, &sin);
+       } else {
+               struct sockaddr_in6 sin6;
+               int i;
+               if (mask < 0 || mask > 128) {
+                       return -1;
+               }
+               memset(&sin6, 0, sizeof(sin6));
+               sin6.sin6_family = AF_INET6;
+               for (i = 0; i < 4; ++i) {
+                       /* Once mask reaches 0, we don't have
+                        * to explicitly set anything anymore
+                        * since sin6 was zeroed out already
+                        */
+                       if (mask > 0) {
+                               V6_WORD(&sin6, i) = htonl(0xFFFFFFFF << (mask < 32 ? (32 - mask) : 0));
+                               mask -= mask < 32 ? mask : 32;
+                       }
+               }
+               memcpy(&addr->ss, &sin6, sizeof(sin6));
+               addr->len = sizeof(sin6);
+       }
+
+       return 0;
+}
+
+void ast_append_acl(const char *sense, const char *stuff, struct ast_acl_list **path, int *error, int *named_acl_flag)
+{
+       struct ast_acl *acl = NULL;
+       struct ast_acl *current;
+       struct ast_acl_list *working_list;
+
+       char *tmp, *list;
+
+       /* If the ACL list is currently uninitialized, it must be initialized. */
+       if (*path == NULL) {
+               struct ast_acl_list *list;
+               list = ast_calloc(1, sizeof(*list));
+               if (!list) {
+                       /* Allocation Error */
+                       if (error) {
+                               *error = 1;
+                       }
+                       return;
+               }
+
+               AST_LIST_HEAD_INIT(list);
+               *path = list;
+       }
+
+       working_list = *path;
+
+       AST_LIST_LOCK(working_list);
+
+       /* First we need to determine if we will need to add a new ACL node or if we can use an existing one. */
+       if (strncasecmp(sense, "a", 1)) {
+               /* The first element in the path should be the unnamed, base ACL. If that's the case, we use it. If not,
+                * we have to make one and link it up appropriately. */
+               current = AST_LIST_FIRST(working_list);
+
+               if (!current || !ast_strlen_zero(current->name)) {
+                       if (acl_new(&acl, "")) {
+                               if (error) {
+                                       *error = 1;
+                               }
+                               AST_LIST_UNLOCK(working_list);
+                               return;
+                       }
+                       // Need to INSERT the ACL at the head here.
+                       AST_LIST_INSERT_HEAD(working_list, acl, list);
+               } else {
+                       /* If the first element was already the unnamed base ACL, we just use that one. */
+                       acl = current;
+               }
+
+               /* With the proper ACL set for modification, we can just pass this off to the ast_ha append function. */
+               acl->acl = ast_append_ha(sense, stuff, acl->acl, error);
+
+               AST_LIST_UNLOCK(working_list);
+               return;
+       }
+
+       /* We are in ACL append mode, so we know we'll be adding one or more named ACLs. */
+       list = ast_strdupa(stuff);
+
+       while ((tmp = strsep(&list, ","))) {
+               struct ast_ha *named_ha;
+               int already_included = 0;
+
+               /* Remove leading whitespace from the string in case the user put spaces between items */
+               tmp = ast_skip_blanks(tmp);
+
+               /* The first step is to check for a duplicate */
+               AST_LIST_TRAVERSE(working_list, current, list) {
+                       if (!strcasecmp(current->name, tmp)) { /* ACL= */
+                               /* Inclusion of the same ACL multiple times isn't a catastrophic error, but it will raise the error flag and skip the entry. */
+                               ast_log(LOG_ERROR, "Named ACL '%s' occurs multiple times in ACL definition. "
+                                                  "Please update your ACL configuration.\n", tmp);
+                               if (error) {
+                                       *error = 1;
+                               }
+                               already_included = 1;
+                               break;
+                       }
+               }
+
+               if (already_included) {
+                       continue;
+               }
+
+               if (acl_new(&acl, tmp)) {
+                       /* This is a catastrophic allocation error and we'll return immediately if this happens. */
+                       if (error) {
+                               *error = 1;
+                       }
+                       AST_LIST_UNLOCK(working_list);
+                       return;
+               }
+
+               /* Attempt to grab the Named ACL we are looking for. */
+               named_ha = ast_named_acl_find(tmp, &acl->is_realtime, &acl->is_invalid);
+
+               /* Set the ACL's ast_ha to the duplicated named ACL retrieved above. */
+               acl->acl = named_ha;
+
+               /* Raise the named_acl_flag since we are adding a named ACL to the ACL container. */
+               if (named_acl_flag) {
+                       *named_acl_flag = 1;
+               }
+
+               /* Now insert the new ACL at the end of the list. */
+               AST_LIST_INSERT_TAIL(working_list, acl, list);
+       }
+
+       AST_LIST_UNLOCK(working_list);
+}
+
+int ast_acl_list_is_empty(struct ast_acl_list *acl_list)
+{
+       struct ast_acl *head;
+
+       if (!acl_list) {
+               return 1;
+       }
+
+       AST_LIST_LOCK(acl_list);
+       head = AST_LIST_FIRST(acl_list);
+       AST_LIST_UNLOCK(acl_list);
+
+       if (head) {
+               return 0;
+       }
+
+       return 1;
+}
+
+/*!
+ * \internal
+ * \brief Used by ast_append_ha to avoid ast_strdupa in a loop.
+ *
+ * \note This function is only called at debug level 3 and higher.
+ */
+static void debug_ha_sense_appended(struct ast_ha *ha)
+{
+       const char *parsed_mask = ast_strdupa(ast_sockaddr_stringify(&ha->netmask));
+
+       ast_log(LOG_DEBUG, "%s/%s sense %u appended to ACL\n",
+               ast_sockaddr_stringify(&ha->addr),
+               parsed_mask,
+               ha->sense);
+}
+
+struct ast_ha *ast_append_ha(const char *sense, const char *stuff, struct ast_ha *path, int *error)
 {
        struct ast_ha *ha;
-       char *nm = "255.255.255.255";
-       char tmp[256];
        struct ast_ha *prev = NULL;
        struct ast_ha *ret;
-       int x, z;
-       unsigned int y;
+       char *tmp, *list = ast_strdupa(stuff);
+       char *address = NULL, *mask = NULL;
+       int addr_is_v4;
+       int allowing = strncasecmp(sense, "p", 1) ? AST_SENSE_DENY : AST_SENSE_ALLOW;
 
        ret = path;
        while (path) {
                prev = path;
                path = path->next;
        }
-       if ((ha = ast_malloc(sizeof(*ha)))) {
-               ast_copy_string(tmp, stuff, sizeof(tmp));
-               nm = strchr(tmp, '/');
-               if (!nm) {
-                       nm = "255.255.255.255";
+
+       while ((tmp = strsep(&list, ","))) {
+               if (!(ha = ast_calloc(1, sizeof(*ha)))) {
+                       if (error) {
+                               *error = 1;
+                       }
+                       return ret;
+               }
+
+               address = strsep(&tmp, "/");
+               if (!address) {
+                       address = tmp;
+               } else {
+                       mask = tmp;
+               }
+
+               if (*address == '!') {
+                       ha->sense = (allowing == AST_SENSE_DENY) ? AST_SENSE_ALLOW : AST_SENSE_DENY;
+                       address++;
                } else {
-                       *nm = '\0';
-                       nm++;
-               }
-               if (!strchr(nm, '.')) {
-                       if ((sscanf(nm, "%d", &x) == 1) && (x >= 0) && (x <= 32)) {
-                               y = 0;
-                               for (z = 0; z < x; z++) {
-                                       y >>= 1;
-                                       y |= 0x80000000;
+                       ha->sense = allowing;
+               }
+
+               if (!ast_sockaddr_parse(&ha->addr, address, PARSE_PORT_FORBID)) {
+                       ast_log(LOG_WARNING, "Invalid IP address: %s\n", address);
+                       ast_free_ha(ha);
+                       if (error) {
+                               *error = 1;
+                       }
+                       return ret;
+               }
+
+               /* If someone specifies an IPv4-mapped IPv6 address,
+                * we just convert this to an IPv4 ACL
+                */
+               if (ast_sockaddr_ipv4_mapped(&ha->addr, &ha->addr)) {
+                       ast_log(LOG_NOTICE, "IPv4-mapped ACL network address specified. "
+                               "Converting to an IPv4 ACL network address.\n");
+               }
+
+               addr_is_v4 = ast_sockaddr_is_ipv4(&ha->addr);
+
+               if (!mask) {
+                       parse_cidr_mask(&ha->netmask, addr_is_v4, addr_is_v4 ? "32" : "128");
+               } else if (strchr(mask, ':') || strchr(mask, '.')) {
+                       int mask_is_v4;
+                       /* Mask is of x.x.x.x or x:x:x:x:x:x:x:x variety */
+                       if (!ast_sockaddr_parse(&ha->netmask, mask, PARSE_PORT_FORBID)) {
+                               ast_log(LOG_WARNING, "Invalid netmask: %s\n", mask);
+                               ast_free_ha(ha);
+                               if (error) {
+                                       *error = 1;
                                }
-                               ha->netmask.s_addr = htonl(y);
+                               return ret;
                        }
-               } else if (!inet_aton(nm, &ha->netmask)) {
-                       ast_log(LOG_WARNING, "%s is not a valid netmask\n", nm);
-                       free(ha);
-                       return path;
-               }
-               if (!inet_aton(tmp, &ha->netaddr)) {
-                       ast_log(LOG_WARNING, "%s is not a valid IP\n", tmp);
-                       free(ha);
-                       return path;
-               }
-               ha->netaddr.s_addr &= ha->netmask.s_addr;
-               if (!strncasecmp(sense, "p", 1)) {
-                       ha->sense = AST_SENSE_ALLOW;
-               } else {
-                       ha->sense = AST_SENSE_DENY;
+                       /* If someone specifies an IPv4-mapped IPv6 netmask,
+                        * we just convert this to an IPv4 ACL
+                        */
+                       if (ast_sockaddr_ipv4_mapped(&ha->netmask, &ha->netmask)) {
+                               ast_log(LOG_NOTICE, "IPv4-mapped ACL netmask specified. "
+                                       "Converting to an IPv4 ACL netmask.\n");
+                       }
+                       mask_is_v4 = ast_sockaddr_is_ipv4(&ha->netmask);
+                       if (addr_is_v4 ^ mask_is_v4) {
+                               ast_log(LOG_WARNING, "Address and mask are not using same address scheme.\n");
+                               ast_free_ha(ha);
+                               if (error) {
+                                       *error = 1;
+                               }
+                               return ret;
+                       }
+               } else if (parse_cidr_mask(&ha->netmask, addr_is_v4, mask)) {
+                       ast_log(LOG_WARNING, "Invalid CIDR netmask: %s\n", mask);
+                       ast_free_ha(ha);
+                       if (error) {
+                               *error = 1;
+                       }
+                       return ret;
                }
-               ha->next = NULL;
+
+               if (ast_sockaddr_apply_netmask(&ha->addr, &ha->netmask, &ha->addr)) {
+                       /* This shouldn't happen because ast_sockaddr_parse would
+                        * have failed much earlier on an unsupported address scheme
+                        */
+                       char *failmask = ast_strdupa(ast_sockaddr_stringify(&ha->netmask));
+                       char *failaddr = ast_strdupa(ast_sockaddr_stringify(&ha->addr));
+                       ast_log(LOG_WARNING, "Unable to apply netmask %s to address %s\n", failmask, failaddr);
+                       ast_free_ha(ha);
+                       if (error) {
+                               *error = 1;
+                       }
+                       return ret;
+               }
+
                if (prev) {
                        prev->next = ha;
                } else {
                        ret = ha;
                }
+               prev = ha;
+
+               if (DEBUG_ATLEAST(3)) {
+                       debug_ha_sense_appended(ha);
+               }
        }
-       ast_log(LOG_DEBUG, "%s/%s appended to acl for peer\n", stuff, nm);
+
        return ret;
 }
 
-int ast_apply_ha(struct ast_ha *ha, struct sockaddr_in *sin)
+void ast_ha_join(const struct ast_ha *ha, struct ast_str **buf)
+{
+       for (; ha; ha = ha->next) {
+               ast_str_append(buf, 0, "%s%s/",
+                       ha->sense == AST_SENSE_ALLOW ? "!" : "",
+                       ast_sockaddr_stringify_addr(&ha->addr));
+               /* Separated to avoid duplicating stringified addresses. */
+               ast_str_append(buf, 0, "%s", ast_sockaddr_stringify_addr(&ha->netmask));
+               if (ha->next) {
+                       ast_str_append(buf, 0, ",");
+               }
+       }
+}
+
+void ast_ha_join_cidr(const struct ast_ha *ha, struct ast_str **buf)
+{
+       for (; ha; ha = ha->next) {
+               const char *addr = ast_sockaddr_stringify_addr(&ha->addr);
+               ast_str_append(buf, 0, "%s%s/%d",
+                              ha->sense == AST_SENSE_ALLOW ? "!" : "",
+                              addr, ast_sockaddr_cidr_bits(&ha->netmask));
+               if (ha->next) {
+                       ast_str_append(buf, 0, ",");
+               }
+       }
+}
+
+enum ast_acl_sense ast_apply_acl(struct ast_acl_list *acl_list, const struct ast_sockaddr *addr, const char *purpose)
+{
+       struct ast_acl *acl;
+
+       /* If the list is NULL, there are no rules, so we'll allow automatically. */
+       if (!acl_list) {
+               return AST_SENSE_ALLOW;
+       }
+
+       AST_LIST_LOCK(acl_list);
+
+       AST_LIST_TRAVERSE(acl_list, acl, list) {
+               if (acl->is_invalid) {
+                       /* In this case, the baseline ACL shouldn't ever trigger this, but if that somehow happens, it'll still be shown. */
+                       ast_log(LOG_WARNING, "%sRejecting '%s' due to use of an invalid ACL '%s'.\n", purpose ? purpose : "", ast_sockaddr_stringify_addr(addr),
+                                       ast_strlen_zero(acl->name) ? "(BASELINE)" : acl->name);
+                       AST_LIST_UNLOCK(acl_list);
+                       return AST_SENSE_DENY;
+               }
+
+               if (acl->acl) {
+                       if (ast_apply_ha(acl->acl, addr) == AST_SENSE_DENY) {
+                               ast_log(LOG_NOTICE, "%sRejecting '%s' due to a failure to pass ACL '%s'\n", purpose ? purpose : "", ast_sockaddr_stringify_addr(addr),
+                                               ast_strlen_zero(acl->name) ? "(BASELINE)" : acl->name);
+                               AST_LIST_UNLOCK(acl_list);
+                               return AST_SENSE_DENY;
+                       }
+               }
+       }
+
+       AST_LIST_UNLOCK(acl_list);
+
+       return AST_SENSE_ALLOW;
+}
+
+enum ast_acl_sense ast_apply_ha(const struct ast_ha *ha, const struct ast_sockaddr *addr)
 {
        /* Start optimistic */
-       int res = AST_SENSE_ALLOW;
-       while (ha) {
+       enum ast_acl_sense res = AST_SENSE_ALLOW;
+       const struct ast_ha *current_ha;
+
+       for (current_ha = ha; current_ha; current_ha = current_ha->next) {
+               struct ast_sockaddr result;
+               struct ast_sockaddr mapped_addr;
+               const struct ast_sockaddr *addr_to_use;
+#if 0  /* debugging code */
                char iabuf[INET_ADDRSTRLEN];
                char iabuf2[INET_ADDRSTRLEN];
                /* DEBUG */
-               ast_copy_string(iabuf, ast_inet_ntoa(sin->sin_addr), sizeof(iabuf));
-               ast_copy_string(iabuf2, ast_inet_ntoa(ha->netaddr), sizeof(iabuf2));
-               ast_log(LOG_DEBUG, "##### Testing %s with %s\n", iabuf, iabuf2);
+               ast_copy_string(iabuf, ast_sockaddr_stringify(addr), sizeof(iabuf));
+               ast_copy_string(iabuf2, ast_sockaddr_stringify(&current_ha->addr), sizeof(iabuf2));
+               ast_debug(1, "##### Testing %s with %s\n", iabuf, iabuf2);
+#endif
+               if (ast_sockaddr_is_ipv4(&current_ha->addr)) {
+                       if (ast_sockaddr_is_ipv6(addr)) {
+                               if (ast_sockaddr_is_ipv4_mapped(addr)) {
+                                       /* IPv4 ACLs apply to IPv4-mapped addresses */
+                                       if (!ast_sockaddr_ipv4_mapped(addr, &mapped_addr)) {
+                                               ast_log(LOG_ERROR, "%s provided to ast_sockaddr_ipv4_mapped could not be converted. That shouldn't be possible.\n",
+                                                       ast_sockaddr_stringify(addr));
+                                               continue;
+                                       }
+                                       addr_to_use = &mapped_addr;
+                               } else {
+                                       /* An IPv4 ACL does not apply to an IPv6 address */
+                                       continue;
+                               }
+                       } else {
+                               /* Address is IPv4 and ACL is IPv4. No biggie */
+                               addr_to_use = addr;
+                       }
+               } else {
+                       if (ast_sockaddr_is_ipv6(addr) && !ast_sockaddr_is_ipv4_mapped(addr)) {
+                               addr_to_use = addr;
+                       } else {
+                               /* Address is IPv4 or IPv4 mapped but ACL is IPv6. Skip */
+                               continue;
+                       }
+               }
+
                /* For each rule, if this address and the netmask = the net address
                   apply the current rule */
-               if ((sin->sin_addr.s_addr & ha->netmask.s_addr) == ha->netaddr.s_addr)
-                       res = ha->sense;
-               ha = ha->next;
+               if (ast_sockaddr_apply_netmask(addr_to_use, &current_ha->netmask, &result)) {
+                       /* Unlikely to happen since we know the address to be IPv4 or IPv6 */
+                       continue;
+               }
+               if (!ast_sockaddr_cmp_addr(&result, &current_ha->addr)) {
+                       res = current_ha->sense;
+               }
        }
        return res;
 }
 
-int ast_get_ip_or_srv(struct sockaddr_in *sin, const char *value, const char *service)
+static int resolve_first(struct ast_sockaddr *addr, const char *name, int flag,
+                        int family)
+{
+       struct ast_sockaddr *addrs;
+       int addrs_cnt;
+
+       addrs_cnt = ast_sockaddr_resolve(&addrs, name, flag, family);
+       if (addrs_cnt > 0) {
+               if (addrs_cnt > 1) {
+                       ast_debug(1, "Multiple addresses. Using the first only\n");
+               }
+               ast_sockaddr_copy(addr, &addrs[0]);
+               ast_free(addrs);
+       } else {
+               ast_log(LOG_WARNING, "Unable to lookup '%s'\n", name);
+               return -1;
+       }
+
+       return 0;
+}
+
+int ast_get_ip_or_srv(struct ast_sockaddr *addr, const char *hostname, const char *service)
 {
-       struct hostent *hp;
-       struct ast_hostent ahp;
        char srv[256];
        char host[256];
-       int tportno = ntohs(sin->sin_port);
-       if (inet_aton(value, &sin->sin_addr))
-               return 0;
+       int srv_ret = 0;
+       int tportno;
+
        if (service) {
-               snprintf(srv, sizeof(srv), "%s.%s", service, value);
-               if (ast_get_srv(NULL, host, sizeof(host), &tportno, srv) > 0) {
-                       sin->sin_port = htons(tportno);
-                       value = host;
+               snprintf(srv, sizeof(srv), "%s.%s", service, hostname);
+               if ((srv_ret = ast_get_srv(NULL, host, sizeof(host), &tportno, srv)) > 0) {
+                       hostname = host;
                }
        }
-       hp = ast_gethostbyname(value, &ahp);
-       if (hp) {
-               memcpy(&sin->sin_addr, hp->h_addr, sizeof(sin->sin_addr));
-       } else {
-               ast_log(LOG_WARNING, "Unable to lookup '%s'\n", value);
+
+       if (resolve_first(addr, hostname, PARSE_PORT_FORBID, addr->ss.ss_family) != 0) {
                return -1;
        }
+
+       if (srv_ret > 0) {
+               ast_sockaddr_set_port(addr, tportno);
+       }
+
        return 0;
 }
 
@@ -277,149 +893,136 @@ static const struct dscp_codepoint dscp_pool1[] = {
        { "EF", 0x2E },
 };
 
+int ast_str2cos(const char *value, unsigned int *cos)
+{
+       int fval;
+
+       if (sscanf(value, "%30d", &fval) == 1) {
+               if (fval < 8) {
+                   *cos = fval;
+                   return 0;
+               }
+       }
+
+       return -1;
+}
+
 int ast_str2tos(const char *value, unsigned int *tos)
 {
        int fval;
        unsigned int x;
 
-       if (sscanf(value, "%i", &fval) == 1) {
+       if (sscanf(value, "%30i", &fval) == 1) {
                *tos = fval & 0xFF;
                return 0;
        }
 
-       for (x = 0; x < sizeof(dscp_pool1) / sizeof(dscp_pool1[0]); x++) {
+       for (x = 0; x < ARRAY_LEN(dscp_pool1); x++) {
                if (!strcasecmp(value, dscp_pool1[x].name)) {
                        *tos = dscp_pool1[x].space << 2;
                        return 0;
                }
        }
 
-       if (!strcasecmp(value, "lowdelay"))
-               *tos = IPTOS_LOWDELAY;
-       else if (!strcasecmp(value, "throughput"))
-               *tos = IPTOS_THROUGHPUT;
-       else if (!strcasecmp(value, "reliability"))
-               *tos = IPTOS_RELIABILITY;
-       else if (!strcasecmp(value, "mincost"))
-               *tos = IPTOS_MINCOST;
-       else if (!strcasecmp(value, "none"))
-               *tos = 0;
-       else
-               return -1;
-
-       ast_log(LOG_WARNING, "TOS value %s is deprecated. Please see doc/ip-tos.txt for more information.\n", value);
-
-       return 0;
+       return -1;
 }
 
 const char *ast_tos2str(unsigned int tos)
 {
        unsigned int x;
 
-       switch (tos) {
-       case 0:
-               return "none";
-       case IPTOS_LOWDELAY:
-               return "lowdelay";
-       case IPTOS_THROUGHPUT:
-               return "throughput";
-       case IPTOS_RELIABILITY:
-               return "reliability";
-       case IPTOS_MINCOST:
-               return "mincost";
-       default:
-               for (x = 0; x < sizeof(dscp_pool1) / sizeof(dscp_pool1[0]); x++) {
-                       if (dscp_pool1[x].space == (tos >> 2))
-                               return dscp_pool1[x].name;
+       for (x = 0; x < ARRAY_LEN(dscp_pool1); x++) {
+               if (dscp_pool1[x].space == (tos >> 2)) {
+                       return dscp_pool1[x].name;
                }
        }
 
        return "unknown";
 }
 
-int ast_get_ip(struct sockaddr_in *sin, const char *value)
-{
-       return ast_get_ip_or_srv(sin, value, NULL);
-}
-
-/* iface is the interface (e.g. eth0); address is the return value */
-int ast_lookup_iface(char *iface, struct in_addr *address)
+int ast_get_ip(struct ast_sockaddr *addr, const char *hostname)
 {
-       int mysock, res = 0;
-       struct my_ifreq ifreq;
-
-       memset(&ifreq, 0, sizeof(ifreq));
-       ast_copy_string(ifreq.ifrn_name, iface, sizeof(ifreq.ifrn_name));
-
-       mysock = socket(PF_INET, SOCK_DGRAM, IPPROTO_IP);
-       res = ioctl(mysock, SIOCGIFADDR, &ifreq);
-
-       close(mysock);
-       if (res < 0) {
-               ast_log(LOG_WARNING, "Unable to get IP of %s: %s\n", iface, strerror(errno));
-               memcpy((char *)address, (char *)&__ourip, sizeof(__ourip));
-               return -1;
-       } else {
-               memcpy((char *)address, (char *)&ifreq.ifru_addr.sin_addr, sizeof(ifreq.ifru_addr.sin_addr));
-               return 0;
-       }
+       return ast_get_ip_or_srv(addr, hostname, NULL);
 }
 
-int ast_ouraddrfor(struct in_addr *them, struct in_addr *us)
+int ast_ouraddrfor(const struct ast_sockaddr *them, struct ast_sockaddr *us)
 {
+       /*
+        * We must create the errno string before creating the address
+        * string because it could wipe out errno on the error return
+        * paths.
+        */
+       const char *sock_err;
+       int port;
        int s;
-       struct sockaddr_in sin;
-       socklen_t slen;
 
-       s = socket(PF_INET, SOCK_DGRAM, 0);
+       /* Preserve our original address port */
+       port = ast_sockaddr_port(us);
+
+       s = socket(ast_sockaddr_is_ipv6(them) ? AF_INET6 : AF_INET, SOCK_DGRAM, 0);
        if (s < 0) {
-               ast_log(LOG_WARNING, "Cannot create socket\n");
+               sock_err = ast_strdupa(strerror(errno));
+               ast_log(LOG_ERROR, "Cannot create socket to %s: %s\n",
+                       ast_sockaddr_stringify_addr(them), sock_err);
                return -1;
        }
-       sin.sin_family = AF_INET;
-       sin.sin_port = 5060;
-       sin.sin_addr = *them;
-       if (connect(s, (struct sockaddr *)&sin, sizeof(sin))) {
-               ast_log(LOG_WARNING, "Cannot connect\n");
+
+       if (ast_connect(s, them)) {
+               sock_err = ast_strdupa(strerror(errno));
+               ast_log(LOG_WARNING, "Cannot connect to %s: %s\n",
+                       ast_sockaddr_stringify_addr(them), sock_err);
                close(s);
                return -1;
        }
-       slen = sizeof(sin);
-       if (getsockname(s, (struct sockaddr *)&sin, &slen)) {
-               ast_log(LOG_WARNING, "Cannot get socket name\n");
+       if (ast_getsockname(s, us)) {
+               sock_err = ast_strdupa(strerror(errno));
+               ast_log(LOG_WARNING, "Cannot get socket name for connection to %s: %s\n",
+                       ast_sockaddr_stringify_addr(them), sock_err);
                close(s);
                return -1;
        }
        close(s);
-       *us = sin.sin_addr;
+
+       ast_sockaddr_set_port(us, port);
+
+       ast_debug(3, "For destination '%s', our source address is '%s'.\n",
+               ast_strdupa(ast_sockaddr_stringify_addr(them)),
+               ast_strdupa(ast_sockaddr_stringify_addr(us)));
+
        return 0;
 }
 
-int ast_find_ourip(struct in_addr *ourip, struct sockaddr_in bindaddr)
+int ast_find_ourip(struct ast_sockaddr *ourip, const struct ast_sockaddr *bindaddr, int family)
 {
        char ourhost[MAXHOSTNAMELEN] = "";
-       struct ast_hostent ahp;
-       struct hostent *hp;
-       struct in_addr saddr;
+       struct ast_sockaddr root;
+       int res, port = ast_sockaddr_port(ourip);
 
        /* just use the bind address if it is nonzero */
-       if (ntohl(bindaddr.sin_addr.s_addr)) {
-               memcpy(ourip, &bindaddr.sin_addr, sizeof(*ourip));
+       if (!ast_sockaddr_is_any(bindaddr)) {
+               ast_sockaddr_copy(ourip, bindaddr);
+               ast_debug(3, "Attached to given IP address\n");
                return 0;
        }
        /* try to use our hostname */
        if (gethostname(ourhost, sizeof(ourhost) - 1)) {
                ast_log(LOG_WARNING, "Unable to get hostname\n");
        } else {
-               hp = ast_gethostbyname(ourhost, &ahp);
-               if (hp) {
-                       memcpy(ourip, hp->h_addr, sizeof(*ourip));
+               if (resolve_first(ourip, ourhost, PARSE_PORT_FORBID, family) == 0) {
+                       /* reset port since resolve_first wipes this out */
+                       ast_sockaddr_set_port(ourip, port);
                        return 0;
                }
        }
+       ast_debug(3, "Trying to check A.ROOT-SERVERS.NET and get our IP address for that connection\n");
        /* A.ROOT-SERVERS.NET. */
-       if (inet_aton("198.41.0.4", &saddr) && !ast_ouraddrfor(&saddr, ourip))
+       if (!resolve_first(&root, "A.ROOT-SERVERS.NET", PARSE_PORT_FORBID, 0) &&
+           !ast_ouraddrfor(&root, ourip)) {
+               /* reset port since resolve_first wipes this out */
+               ast_sockaddr_set_port(ourip, port);
                return 0;
-       return -1;
+       }
+       res = get_local_address(ourip);
+       ast_sockaddr_set_port(ourip, port);
+       return res;
 }
-