Mercurial > hg > dhcpcd
view net.c @ 1693:829716214dbc draft dhcpcd-5.6.0
Improve IPv6 RA support by allowing dhcpcd to manage the address and routes
instead of the kernel. dhcpcd will only do this if RA is disabled in the kernel
or dhcpcd has been instructed to do this via dhcpcd.conf(5) ipv6ra_own and
ipv6ra_own_default directives.
Send and process IPv6 Neighbor Solicitions and Adverts to prove router
reachability. If a router cannot be reached in this way then it is expired.
When debugging, all ND messages are displayed which will create a lot of log
spam.
To ease packaging, ./configure now accepts LDFLAGS and --enable-static.
| author | Roy Marples <roy@marples.name> |
|---|---|
| date | Thu, 05 Jul 2012 16:37:41 +0000 |
| parents | 22baa5ab2eea |
| children | 5331515903a6 |
line wrap: on
line source
/* * dhcpcd - DHCP client daemon * Copyright (c) 2006-2011 Roy Marples <roy@marples.name> * All rights reserved * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include <sys/types.h> #include <sys/ioctl.h> #include <sys/param.h> #include <sys/socket.h> #include <sys/time.h> #include <arpa/inet.h> #include <net/if.h> #include <net/if_arp.h> #ifdef AF_LINK # include <net/if_dl.h> # include <net/if_types.h> #endif #include <netinet/in_systm.h> #include <netinet/in.h> #include <netinet/ip.h> #define __FAVOR_BSD /* Nasty glibc hack so we can use BSD semantics for UDP */ #include <netinet/udp.h> #undef __FAVOR_BSD #ifdef AF_PACKET # include <netpacket/packet.h> #endif #ifdef SIOCGIFMEDIA # include <net/if_media.h> #endif #include <ctype.h> #include <errno.h> #include <ifaddrs.h> #include <fnmatch.h> #include <stddef.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <syslog.h> #include <unistd.h> #include "config.h" #include "common.h" #include "dhcp.h" #include "if-options.h" #include "ipv6rs.h" #include "net.h" #include "signals.h" static char hwaddr_buffer[(HWADDR_LEN * 3) + 1 + 1024]; int socket_afnet = -1; int inet_ntocidr(struct in_addr address) { int cidr = 0; uint32_t mask = htonl(address.s_addr); while (mask) { cidr++; mask <<= 1; } return cidr; } int inet_cidrtoaddr(int cidr, struct in_addr *addr) { int ocets; if (cidr < 1 || cidr > 32) { errno = EINVAL; return -1; } ocets = (cidr + 7) / 8; addr->s_addr = 0; if (ocets > 0) { memset(&addr->s_addr, 255, (size_t)ocets - 1); memset((unsigned char *)&addr->s_addr + (ocets - 1), (256 - (1 << (32 - cidr) % 8)), 1); } return 0; } uint32_t get_netmask(uint32_t addr) { uint32_t dst; if (addr == 0) return 0; dst = htonl(addr); if (IN_CLASSA(dst)) return ntohl(IN_CLASSA_NET); if (IN_CLASSB(dst)) return ntohl(IN_CLASSB_NET); if (IN_CLASSC(dst)) return ntohl(IN_CLASSC_NET); return 0; } char * hwaddr_ntoa(const unsigned char *hwaddr, size_t hwlen) { char *p = hwaddr_buffer; size_t i; for (i = 0; i < hwlen; i++) { if (i > 0) *p ++= ':'; p += snprintf(p, 3, "%.2x", hwaddr[i]); } *p ++= '\0'; return hwaddr_buffer; } size_t hwaddr_aton(unsigned char *buffer, const char *addr) { char c[3]; const char *p = addr; unsigned char *bp = buffer; size_t len = 0; c[2] = '\0'; while (*p) { c[0] = *p++; c[1] = *p++; /* Ensure that digits are hex */ if (isxdigit((unsigned char)c[0]) == 0 || isxdigit((unsigned char)c[1]) == 0) { errno = EINVAL; return 0; } /* We should have at least two entries 00:01 */ if (len == 0 && *p == '\0') { errno = EINVAL; return 0; } /* Ensure that next data is EOL or a seperator with data */ if (!(*p == '\0' || (*p == ':' && *(p + 1) != '\0'))) { errno = EINVAL; return 0; } if (*p) p++; if (bp) *bp++ = (unsigned char)strtol(c, NULL, 16); len++; } return len; } void free_interface(struct interface *iface) { if (!iface) return; ipv6rs_free(iface); if (iface->state) { free_options(iface->state->options); free(iface->state->old); free(iface->state->new); free(iface->state->offer); free(iface->state); } free(iface->buffer); free(iface->clientid); free(iface); } int carrier_status(struct interface *iface) { int ret; struct ifreq ifr; #ifdef SIOCGIFMEDIA struct ifmediareq ifmr; #endif #ifdef __linux__ char *p; #endif memset(&ifr, 0, sizeof(ifr)); strlcpy(ifr.ifr_name, iface->name, sizeof(ifr.ifr_name)); #ifdef __linux__ /* We can only test the real interface up */ if ((p = strchr(ifr.ifr_name, ':'))) *p = '\0'; #endif if (ioctl(socket_afnet, SIOCGIFFLAGS, &ifr) == -1) return -1; iface->flags = ifr.ifr_flags; ret = -1; #ifdef SIOCGIFMEDIA memset(&ifmr, 0, sizeof(ifmr)); strlcpy(ifmr.ifm_name, iface->name, sizeof(ifmr.ifm_name)); if (ioctl(socket_afnet, SIOCGIFMEDIA, &ifmr) != -1 && ifmr.ifm_status & IFM_AVALID) ret = (ifmr.ifm_status & IFM_ACTIVE) ? 1 : 0; #endif if (ret == -1) ret = (ifr.ifr_flags & IFF_RUNNING) ? 1 : 0; return ret; } int up_interface(struct interface *iface) { struct ifreq ifr; int retval = -1; #ifdef __linux__ char *p; #endif memset(&ifr, 0, sizeof(ifr)); strlcpy(ifr.ifr_name, iface->name, sizeof(ifr.ifr_name)); #ifdef __linux__ /* We can only bring the real interface up */ if ((p = strchr(ifr.ifr_name, ':'))) *p = '\0'; #endif if (ioctl(socket_afnet, SIOCGIFFLAGS, &ifr) == 0) { if ((ifr.ifr_flags & IFF_UP)) retval = 0; else { ifr.ifr_flags |= IFF_UP; if (ioctl(socket_afnet, SIOCSIFFLAGS, &ifr) == 0) retval = 0; } iface->flags = ifr.ifr_flags; } return retval; } struct interface * discover_interfaces(int argc, char * const *argv) { struct ifaddrs *ifaddrs, *ifa; char *p; int i, sdl_type; struct interface *ifp, *ifs, *ifl; #ifdef __linux__ char ifn[IF_NAMESIZE]; #endif #ifdef AF_LINK const struct sockaddr_dl *sdl; #ifdef IFLR_ACTIVE struct if_laddrreq iflr; int socket_aflink; socket_aflink = socket(AF_LINK, SOCK_DGRAM, 0); if (socket_aflink == -1) return NULL; memset(&iflr, 0, sizeof(iflr)); #endif #elif AF_PACKET const struct sockaddr_ll *sll; #endif if (getifaddrs(&ifaddrs) == -1) return NULL; ifs = ifl = NULL; for (ifa = ifaddrs; ifa; ifa = ifa->ifa_next) { if (ifa->ifa_addr != NULL) { #ifdef AF_LINK if (ifa->ifa_addr->sa_family != AF_LINK) continue; #elif AF_PACKET if (ifa->ifa_addr->sa_family != AF_PACKET) continue; #endif } /* It's possible for an interface to have >1 AF_LINK. * For our purposes, we use the first one. */ for (ifp = ifs; ifp; ifp = ifp->next) if (strcmp(ifp->name, ifa->ifa_name) == 0) break; if (ifp) continue; if (argc > 0) { for (i = 0; i < argc; i++) { #ifdef __linux__ /* Check the real interface name */ strlcpy(ifn, argv[i], sizeof(ifn)); p = strchr(ifn, ':'); if (p) *p = '\0'; if (strcmp(ifn, ifa->ifa_name) == 0) break; #else if (strcmp(argv[i], ifa->ifa_name) == 0) break; #endif } if (i == argc) continue; p = argv[i]; } else { /* -1 means we're discovering against a specific * interface, but we still need the below rules * to apply. */ if (argc == -1 && strcmp(argv[0], ifa->ifa_name) != 0) continue; for (i = 0; i < ifdc; i++) if (!fnmatch(ifdv[i], ifa->ifa_name, 0)) break; if (i < ifdc) continue; for (i = 0; i < ifac; i++) if (!fnmatch(ifav[i], ifa->ifa_name, 0)) break; if (ifac && i == ifac) continue; p = ifa->ifa_name; } ifp = xzalloc(sizeof(*ifp)); strlcpy(ifp->name, p, sizeof(ifp->name)); ifp->flags = ifa->ifa_flags; /* Bring the interface up if not already */ if (!(ifp->flags & IFF_UP) #ifdef SIOCGIFMEDIA && carrier_status(ifp) != -1 #endif ) { if (up_interface(ifp) == 0) options |= DHCPCD_WAITUP; else syslog(LOG_ERR, "%s: up_interface: %m", ifp->name); } sdl_type = 0; /* Don't allow loopback unless explicit */ if (ifp->flags & IFF_LOOPBACK) { if (argc == 0 && ifac == 0) { free_interface(ifp); continue; } } else if (ifa->ifa_addr != NULL) { #ifdef AF_LINK sdl = (const struct sockaddr_dl *)(void *)ifa->ifa_addr; #ifdef IFLR_ACTIVE /* We need to check for active address */ strlcpy(iflr.iflr_name, ifp->name, sizeof(iflr.iflr_name)); memcpy(&iflr.addr, ifa->ifa_addr, MIN(ifa->ifa_addr->sa_len, sizeof(iflr.addr))); iflr.flags = IFLR_PREFIX; iflr.prefixlen = sdl->sdl_alen * NBBY; if (ioctl(socket_aflink, SIOCGLIFADDR, &iflr) == -1 || !(iflr.flags & IFLR_ACTIVE)) { free_interface(ifp); continue; } #endif ifp->index = sdl->sdl_index; sdl_type = sdl->sdl_type; switch(sdl->sdl_type) { case IFT_BRIDGE: /* FALLTHROUGH */ case IFT_L2VLAN: /* FALLTHOUGH */ case IFT_L3IPVLAN: /* FALLTHROUGH */ case IFT_ETHER: ifp->family = ARPHRD_ETHER; break; case IFT_IEEE1394: ifp->family = ARPHRD_IEEE1394; break; #ifdef IFT_INFINIBAND case IFT_INFINIBAND: ifp->family = ARPHRD_INFINIBAND; break; #endif } ifp->hwlen = sdl->sdl_alen; #ifndef CLLADDR # define CLLADDR(s) ((const char *)((s)->sdl_data + (s)->sdl_nlen)) #endif memcpy(ifp->hwaddr, CLLADDR(sdl), ifp->hwlen); #elif AF_PACKET sll = (const struct sockaddr_ll *)(void *)ifa->ifa_addr; ifp->index = sll->sll_ifindex; ifp->family = sdl_type = sll->sll_hatype; ifp->hwlen = sll->sll_halen; if (ifp->hwlen != 0) memcpy(ifp->hwaddr, sll->sll_addr, ifp->hwlen); #endif } /* We only work on ethernet by default */ if (!(ifp->flags & IFF_POINTOPOINT) && ifp->family != ARPHRD_ETHER) { if (argc == 0 && ifac == 0) { free_interface(ifp); continue; } switch (ifp->family) { case ARPHRD_IEEE1394: /* FALLTHROUGH */ case ARPHRD_INFINIBAND: /* We don't warn for supported families */ break; default: syslog(LOG_WARNING, "%s: unsupported interface type %.2x" ", falling back to ethernet", ifp->name, sdl_type); ifp->family = ARPHRD_ETHER; break; } } /* Handle any platform init for the interface */ if (if_init(ifp) == -1) { syslog(LOG_ERR, "%s: if_init: %m", p); free_interface(ifp); continue; } /* Ensure that the MTU is big enough for DHCP */ if (get_mtu(ifp->name) < MTU_MIN && set_mtu(ifp->name, MTU_MIN) == -1) { syslog(LOG_ERR, "%s: set_mtu: %m", p); free_interface(ifp); continue; } /* We reserve the 100 range for virtual interfaces, if and when * we can work them out. */ ifp->metric = 200 + ifp->index; if (getifssid(ifp->name, ifp->ssid) != -1) { ifp->wireless = 1; ifp->metric += 100; } snprintf(ifp->leasefile, sizeof(ifp->leasefile), LEASEFILE, ifp->name); /* 0 is a valid fd, so init to -1 */ ifp->raw_fd = ifp->udp_fd = ifp->arp_fd = -1; if (ifl) ifl->next = ifp; else ifs = ifp; ifl = ifp; } freeifaddrs(ifaddrs); #ifdef IFLR_ACTIVE close(socket_aflink); #endif return ifs; } int do_address(const char *ifname, struct in_addr *addr, struct in_addr *net, struct in_addr *dst, int act) { struct ifaddrs *ifaddrs, *ifa; const struct sockaddr_in *a, *n, *d; int retval; if (getifaddrs(&ifaddrs) == -1) return -1; retval = 0; for (ifa = ifaddrs; ifa; ifa = ifa->ifa_next) { if (ifa->ifa_addr == NULL || ifa->ifa_addr->sa_family != AF_INET || strcmp(ifa->ifa_name, ifname) != 0) continue; a = (const struct sockaddr_in *)(void *)ifa->ifa_addr; n = (const struct sockaddr_in *)(void *)ifa->ifa_netmask; if (ifa->ifa_flags & IFF_POINTOPOINT) d = (const struct sockaddr_in *)(void *) ifa->ifa_dstaddr; else d = NULL; if (act == 1) { addr->s_addr = a->sin_addr.s_addr; net->s_addr = n->sin_addr.s_addr; if (dst) { if (ifa->ifa_flags & IFF_POINTOPOINT) dst->s_addr = d->sin_addr.s_addr; else dst->s_addr = INADDR_ANY; } retval = 1; break; } if (addr->s_addr == a->sin_addr.s_addr && (net == NULL || net->s_addr == n->sin_addr.s_addr)) { retval = 1; break; } } freeifaddrs(ifaddrs); return retval; } int do_mtu(const char *ifname, short int mtu) { struct ifreq ifr; int r; memset(&ifr, 0, sizeof(ifr)); strlcpy(ifr.ifr_name, ifname, sizeof(ifr.ifr_name)); ifr.ifr_mtu = mtu; r = ioctl(socket_afnet, mtu ? SIOCSIFMTU : SIOCGIFMTU, &ifr); if (r == -1) return -1; return ifr.ifr_mtu; } void free_routes(struct rt *routes) { struct rt *r; while (routes) { r = routes->next; free(routes); routes = r; } } int open_udp_socket(struct interface *iface) { int s; struct sockaddr_in sin; int n; #ifdef SO_BINDTODEVICE struct ifreq ifr; char *p; #endif if ((s = socket(PF_INET, SOCK_DGRAM, IPPROTO_UDP)) == -1) return -1; n = 1; if (setsockopt(s, SOL_SOCKET, SO_REUSEADDR, &n, sizeof(n)) == -1) goto eexit; #ifdef SO_BINDTODEVICE memset(&ifr, 0, sizeof(ifr)); strlcpy(ifr.ifr_name, iface->name, sizeof(ifr.ifr_name)); /* We can only bind to the real device */ p = strchr(ifr.ifr_name, ':'); if (p) *p = '\0'; if (setsockopt(s, SOL_SOCKET, SO_BINDTODEVICE, &ifr, sizeof(ifr)) == -1) goto eexit; #endif /* As we don't use this socket for receiving, set the * receive buffer to 1 */ n = 1; if (setsockopt(s, SOL_SOCKET, SO_RCVBUF, &n, sizeof(n)) == -1) goto eexit; memset(&sin, 0, sizeof(sin)); sin.sin_family = AF_INET; sin.sin_port = htons(DHCP_CLIENT_PORT); sin.sin_addr.s_addr = iface->addr.s_addr; if (bind(s, (struct sockaddr *)&sin, sizeof(sin)) == -1) goto eexit; iface->udp_fd = s; set_cloexec(s); return 0; eexit: close(s); return -1; } ssize_t send_packet(const struct interface *iface, struct in_addr to, const uint8_t *data, ssize_t len) { struct sockaddr_in sin; memset(&sin, 0, sizeof(sin)); sin.sin_family = AF_INET; sin.sin_addr.s_addr = to.s_addr; sin.sin_port = htons(DHCP_SERVER_PORT); return sendto(iface->udp_fd, data, len, 0, (struct sockaddr *)&sin, sizeof(sin)); } struct udp_dhcp_packet { struct ip ip; struct udphdr udp; struct dhcp_message dhcp; }; const size_t udp_dhcp_len = sizeof(struct udp_dhcp_packet); static uint16_t checksum(const void *data, uint16_t len) { const uint8_t *addr = data; uint32_t sum = 0; while (len > 1) { sum += addr[0] * 256 + addr[1]; addr += 2; len -= 2; } if (len == 1) sum += *addr * 256; sum = (sum >> 16) + (sum & 0xffff); sum += (sum >> 16); sum = htons(sum); return ~sum; } ssize_t make_udp_packet(uint8_t **packet, const uint8_t *data, size_t length, struct in_addr source, struct in_addr dest) { struct udp_dhcp_packet *udpp; struct ip *ip; struct udphdr *udp; udpp = xzalloc(sizeof(*udpp)); ip = &udpp->ip; udp = &udpp->udp; /* OK, this is important :) * We copy the data to our packet and then create a small part of the * ip structure and an invalid ip_len (basically udp length). * We then fill the udp structure and put the checksum * of the whole packet into the udp checksum. * Finally we complete the ip structure and ip checksum. * If we don't do the ordering like so then the udp checksum will be * broken, so find another way of doing it! */ memcpy(&udpp->dhcp, data, length); ip->ip_p = IPPROTO_UDP; ip->ip_src.s_addr = source.s_addr; if (dest.s_addr == 0) ip->ip_dst.s_addr = INADDR_BROADCAST; else ip->ip_dst.s_addr = dest.s_addr; udp->uh_sport = htons(DHCP_CLIENT_PORT); udp->uh_dport = htons(DHCP_SERVER_PORT); udp->uh_ulen = htons(sizeof(*udp) + length); ip->ip_len = udp->uh_ulen; udp->uh_sum = checksum(udpp, sizeof(*udpp)); ip->ip_v = IPVERSION; ip->ip_hl = sizeof(*ip) >> 2; ip->ip_id = arc4random() & UINT16_MAX; ip->ip_ttl = IPDEFTTL; ip->ip_len = htons(sizeof(*ip) + sizeof(*udp) + length); ip->ip_sum = checksum(ip, sizeof(*ip)); *packet = (uint8_t *)udpp; return sizeof(*ip) + sizeof(*udp) + length; } ssize_t get_udp_data(const uint8_t **data, const uint8_t *udp) { struct udp_dhcp_packet packet; memcpy(&packet, udp, sizeof(packet)); *data = udp + offsetof(struct udp_dhcp_packet, dhcp); return ntohs(packet.ip.ip_len) - sizeof(packet.ip) - sizeof(packet.udp); } int valid_udp_packet(const uint8_t *data, size_t data_len, struct in_addr *from, int noudpcsum) { struct udp_dhcp_packet packet; uint16_t bytes, udpsum; if (data_len < sizeof(packet.ip)) { if (from) from->s_addr = INADDR_ANY; errno = EINVAL; return -1; } memcpy(&packet, data, MIN(data_len, sizeof(packet))); if (from) from->s_addr = packet.ip.ip_src.s_addr; if (data_len > sizeof(packet)) { errno = EINVAL; return -1; } if (checksum(&packet.ip, sizeof(packet.ip)) != 0) { errno = EINVAL; return -1; } bytes = ntohs(packet.ip.ip_len); if (data_len < bytes) { errno = EINVAL; return -1; } if (noudpcsum == 0) { udpsum = packet.udp.uh_sum; packet.udp.uh_sum = 0; packet.ip.ip_hl = 0; packet.ip.ip_v = 0; packet.ip.ip_tos = 0; packet.ip.ip_len = packet.udp.uh_ulen; packet.ip.ip_id = 0; packet.ip.ip_off = 0; packet.ip.ip_ttl = 0; packet.ip.ip_sum = 0; if (udpsum && checksum(&packet, bytes) != udpsum) { errno = EINVAL; return -1; } } return 0; }
