traceroute: fix traceroute6 -I (icmp mode)

function                                             old     new   delta
common_traceroute_main                              3530    3544     +14

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

1 files changed, 135 insertions, 71 deletions

diff --git a/networking/traceroute.c b/networking/traceroute.c
index 7dde10524..ddde922bf 100644
--- a/networking/traceroute.c
+++ b/networking/traceroute.c
@@ -414,9 +414,32 @@ struct globals {
         waittime = 5; \
 } while (0)
 
-#define outicmp ((struct icmp *)(outip + 1))
-#define outudp  ((struct udphdr *)(outip + 1))
-
+#define outudp   ((struct udphdr *)(outip + 1))
+#define outudp6  ((struct udphdr *)(((struct ip6_hdr*)outip) + 1))
+#define outicmp  ((struct icmp *)(outip + 1))
+#define outicmp6 ((struct icmp *)(((struct ip6_hdr*)outip) + 1))
+/* NB: for icmp echo, IPv4 and IPv6 fields are the same size and offset:
+ * struct icmp:
+ *      uint8_t  icmp_type;
+ *      uint8_t  icmp_code;
+ *      uint16_t icmp_cksum;
+ *      uint16_t icmp_id;
+ *      uint16_t icmp_seq;
+ * struct icmp6_hdr:
+ *      uint8_t  icmp6_type;
+ *      uint8_t  icmp6_code;
+ *      uint16_t icmp6_cksum;
+ *      uint16_t icmp6_id;
+ *      uint16_t icmp6_seq;
+ * therefore both outicmp and outicmp6 are pointers to *IPv4* icmp struct.
+ * SIZEOF_ICMP_HDR == 8 is the same for both, as well.
+ * However, values of these pointers are not the same (since IPv6 IP header is larger),
+ * and icmp_type constants are not the same:
+ * #define ICMP_ECHO               8
+ * #define ICMP_ECHOREPLY          0
+ * #define ICMP6_ECHO_REQUEST    128
+ * #define ICMP6_ECHO_REPLY      129
+ */
 
 static int
 wait_for_reply(len_and_sockaddr *from_lsa, struct sockaddr *to, unsigned *timestamp_us, int *left_ms)
@@ -446,14 +469,16 @@ send_probe(int seq, int ttl)
 {
         int len, res;
         void *out;
+        struct icmp *icp;
 
         /* Payload */
 #if ENABLE_TRACEROUTE6
         if (dest_lsa->u.sa.sa_family == AF_INET6) {
-                struct outdata6_t *pkt = (struct outdata6_t *) outdata;
+                struct outdata6_t *pkt = (void *) outdata;
                 pkt->ident6 = htonl(ident);
                 pkt->seq6   = htonl(seq);
                 /*gettimeofday(&pkt->tv, &tz);*/
+                icp = outicmp6;
         } else
 #endif
         {
@@ -461,19 +486,23 @@ send_probe(int seq, int ttl)
                 outdata->ttl = ttl;
 // UNUSED: was storing gettimeofday's result there, but never ever checked it
                 /*memcpy(&outdata->tv, tp, sizeof(outdata->tv));*/
-
-                if (option_mask32 & OPT_USE_ICMP) {
-                        outicmp->icmp_seq = htons(seq);
-
-                        /* Always calculate checksum for icmp packets */
-                        outicmp->icmp_cksum = 0;
-                        outicmp->icmp_cksum = inet_cksum(
-                                        outicmp,
-                                        ((char*)outip + packlen) - (char*)outicmp
-                        );
-                        if (outicmp->icmp_cksum == 0)
-                                outicmp->icmp_cksum = 0xffff;
-                }
+                icp = outicmp;
+        }
+        out = outdata;
+        if (option_mask32 & OPT_USE_ICMP) {
+                out = icp;
+                /*icp->icmp_type = ICMP[6]_ECHO; - already set */
+                /*icp->icmp_code = 0; - already set */
+                /*icp->icmp_id = ident; - already set */
+                icp->icmp_seq = htons(seq);
+                /* Always calculate checksum for icmp packets */
+                icp->icmp_cksum = 0;
+                icp->icmp_cksum = inet_cksum(
+                                icp,
+                                ((char*)outip + packlen) - (char*)icp
+                );
+                if (icp->icmp_cksum == 0)
+                        icp->icmp_cksum = 0xffff;
         }
 
 //BUG! verbose is (x & OPT_VERBOSE), not a counter!
@@ -502,7 +531,6 @@ send_probe(int seq, int ttl)
         }
 #endif
 
-        out = outdata;
 #if ENABLE_TRACEROUTE6
         if (dest_lsa->u.sa.sa_family == AF_INET6) {
                 res = setsockopt_int(sndsock, SOL_IPV6, IPV6_UNICAST_HOPS, ttl);
@@ -516,8 +544,6 @@ send_probe(int seq, int ttl)
                 if (res != 0)
                         bb_perror_msg_and_die("setsockopt(%s) %d", "TTL", ttl);
 #endif
-                if (option_mask32 & OPT_USE_ICMP)
-                        out = outicmp;
         }
 
         if (!(option_mask32 & OPT_USE_ICMP)) {
@@ -579,7 +605,12 @@ packet4_ok(int read_len, const struct sockaddr_in *from, int seq)
         int hlen;
         const struct ip *ip;
 
+        /* NB: reads from (AF_INET, SOCK_RAW, IPPROTO_ICMP) socket
+         * return the entire IP packet (IOW: they do not strip IP header).
+         * This differs from (AF_INET6, SOCK_RAW, IPPROTO_ICMPV6) sockets!?
+         */
         ip = (struct ip *) recv_pkt;
+
         hlen = ip->ip_hl << 2;
         if (read_len < hlen + ICMP_MINLEN) {
 #if ENABLE_FEATURE_TRACEROUTE_VERBOSE
@@ -598,9 +629,20 @@ packet4_ok(int read_len, const struct sockaddr_in *from, int seq)
         if (code == ICMP_UNREACH_NEEDFRAG)
                 pmtu = ntohs(icp->icmp_nextmtu);
 
+        if ((option_mask32 & OPT_USE_ICMP)
+         && type == ICMP_ECHOREPLY
+         && icp->icmp_id == htons(ident)
+         && icp->icmp_seq == htons(seq)
+        ) {
+                /* In UDP mode, when we reach the machine, we (usually)
+                 * would get "port unreachable" - in ICMP we got "echo reply".
+                 * Simulate "port unreachable" for caller:
+                 */
+                return ICMP_UNREACH_PORT+1;
+        }
+
         if ((type == ICMP_TIMXCEED && code == ICMP_TIMXCEED_INTRANS)
          || type == ICMP_UNREACH
-         || type == ICMP_ECHOREPLY
         ) {
                 const struct ip *hip;
                 const struct udphdr *up;
@@ -610,14 +652,6 @@ packet4_ok(int read_len, const struct sockaddr_in *from, int seq)
                 if (option_mask32 & OPT_USE_ICMP) {
                         struct icmp *hicmp;
 
-                        /* XXX */
-                        if (type == ICMP_ECHOREPLY
-                         && icp->icmp_id == htons(ident)
-                         && icp->icmp_seq == htons(seq)
-                        ) {
-                                return ICMP_UNREACH_PORT+1;
-                        }
-
                         hicmp = (struct icmp *)((unsigned char *)hip + hlen);
                         if (hlen + SIZEOF_ICMP_HDR <= read_len
                          && hip->ip_p == IPPROTO_ICMP
@@ -648,6 +682,7 @@ packet4_ok(int read_len, const struct sockaddr_in *from, int seq)
                 printf("\n%d bytes from %s to "
                        "%s: icmp type %d (%s) code %d\n",
                         read_len, inet_ntoa(from->sin_addr),
+//BUG: inet_ntoa() returns static buf! x2 is NONO!
                         inet_ntoa(ip->ip_dst),
                         type, pr_type(type), icp->icmp_code);
                 for (i = 4; i < read_len; i += sizeof(*lp))
@@ -673,11 +708,27 @@ packet_ok(int read_len, len_and_sockaddr *from_lsa,
         if (from_lsa->u.sa.sa_family == AF_INET)
                 return packet4_ok(read_len, &from_lsa->u.sin, seq);
 
+        /* NB: reads from (AF_INET6, SOCK_RAW, IPPROTO_ICMPV6) socket
+         * return only ICMP packet (IOW: they strip IPv6 header).
+         * This differs from (AF_INET, SOCK_RAW, IPPROTO_ICMP) sockets!?
+         */
         icp = (struct icmp6_hdr *) recv_pkt;
 
         type = icp->icmp6_type;
         code = icp->icmp6_code;
 
+        if ((option_mask32 & OPT_USE_ICMP)
+         && type == ICMP6_ECHO_REPLY
+         && icp->icmp6_id == htons(ident)
+         && icp->icmp6_seq == htons(seq)
+        ) {
+                /* In UDP mode, when we reach the machine, we (usually)
+                 * would get "port unreachable" - in ICMP we got "echo reply".
+                 * Simulate "port unreachable" for caller:
+                 */
+                return (ICMP6_DST_UNREACH_NOPORT << 8) + 1;
+        }
+
         if ((type == ICMP6_TIME_EXCEEDED && code == ICMP6_TIME_EXCEED_TRANSIT)
          || type == ICMP6_DST_UNREACH
         ) {
@@ -787,8 +838,13 @@ print(int read_len, const struct sockaddr *from, const struct sockaddr *to)
         if (verbose) {
                 char *ina = xmalloc_sockaddr2dotted_noport(to);
 #if ENABLE_TRACEROUTE6
+                /* NB: reads from (AF_INET, SOCK_RAW, IPPROTO_ICMP) socket
+                 * return the entire IP packet (IOW: they do not strip IP header).
+                 * Reads from (AF_INET6, SOCK_RAW, IPPROTO_ICMPV6) do strip IPv6
+                 * header and return only ICMP6 packet. Weird.
+                 */
                 if (to->sa_family == AF_INET6) {
-                        read_len -= sizeof(struct ip6_hdr);
+                        /* read_len -= sizeof(struct ip6_hdr); - WRONG! */
                 } else
 #endif
                 {
@@ -844,6 +900,9 @@ common_traceroute_main(int op, char **argv)
         struct sockaddr *lastaddr;
         struct sockaddr *to;
 
+        /* Ensure the socket fds won't be 0, 1 or 2 */
+        bb_sanitize_stdio();
+
         INIT_G();
 
         op |= getopt32(argv, "^"
@@ -885,12 +944,14 @@ common_traceroute_main(int op, char **argv)
 
         /* Process destination and optional packet size */
         minpacket = sizeof(struct ip)
-                        + SIZEOF_ICMP_HDR
+                        + sizeof(struct udphdr)
                         + sizeof(struct outdata_t);
-        if (!(op & OPT_USE_ICMP))
+        if (op & OPT_USE_ICMP) {
                 minpacket = sizeof(struct ip)
-                        + sizeof(struct udphdr)
+                        + SIZEOF_ICMP_HDR
                         + sizeof(struct outdata_t);
+                port = 0; /* on ICMP6 sockets, sendto(ipv6.nonzero_port) throws EINVAL */
+        }
 #if ENABLE_TRACEROUTE6
         af = AF_UNSPEC;
         if (op & OPT_IPV4)
@@ -899,10 +960,15 @@ common_traceroute_main(int op, char **argv)
                 af = AF_INET6;
         dest_lsa = xhost_and_af2sockaddr(argv[0], port, af);
         af = dest_lsa->u.sa.sa_family;
-        if (af == AF_INET6)
+        if (af == AF_INET6) {
                 minpacket = sizeof(struct ip6_hdr)
-                        + sizeof(struct udphdr)
-                        + sizeof(struct outdata6_t);
+                                + sizeof(struct udphdr)
+                                + sizeof(struct outdata6_t);
+                if (op & OPT_USE_ICMP)
+                        minpacket = sizeof(struct ip6_hdr)
+                                + SIZEOF_ICMP_HDR
+                                + sizeof(struct outdata6_t);
+        }
 #else
         dest_lsa = xhost2sockaddr(argv[0], port);
 #endif
@@ -910,13 +976,10 @@ common_traceroute_main(int op, char **argv)
         if (argv[1])
                 packlen = xatoul_range(argv[1], minpacket, 32 * 1024);
 
-        /* Ensure the socket fds won't be 0, 1 or 2 */
-        bb_sanitize_stdio();
-
 #if ENABLE_TRACEROUTE6
         if (af == AF_INET6) {
                 xmove_fd(xsocket(AF_INET6, SOCK_RAW, IPPROTO_ICMPV6), rcvsock);
-                setsockopt_1(rcvsock, SOL_IPV6, IPV6_RECVPKTINFO);
+                setsockopt_1(rcvsock, SOL_IPV6, IPV6_RECVPKTINFO); //why?
         } else
 #endif
         {
@@ -934,7 +997,10 @@ common_traceroute_main(int op, char **argv)
         if (af == AF_INET6) {
                 if (setsockopt_int(rcvsock, SOL_RAW, IPV6_CHECKSUM, 2) != 0)
                         bb_perror_msg_and_die("setsockopt(%s)", "IPV6_CHECKSUM");
-                xmove_fd(xsocket(af, SOCK_DGRAM, 0), sndsock);
+                if (op & OPT_USE_ICMP)
+                        xmove_fd(xsocket(AF_INET6, SOCK_RAW, IPPROTO_ICMPV6), sndsock);
+                else
+                        xmove_fd(xsocket(AF_INET6, SOCK_DGRAM, 0), sndsock);
         } else
 #endif
         {
@@ -943,6 +1009,7 @@ common_traceroute_main(int op, char **argv)
                 else
                         xmove_fd(xsocket(AF_INET, SOCK_DGRAM, 0), sndsock);
         }
+//TODO xmove_fd to here!
 
 #ifdef SO_SNDBUF
         if (setsockopt_SOL_SOCKET_int(sndsock, SO_SNDBUF, packlen) != 0) {
@@ -970,21 +1037,25 @@ common_traceroute_main(int op, char **argv)
         ident = getpid();
 
         if (!ENABLE_TRACEROUTE6 || af == AF_INET) {
+                outdata = (void*)(outudp + 1);
                 if (op & OPT_USE_ICMP) {
                         ident |= 0x8000;
                         outicmp->icmp_type = ICMP_ECHO;
+                        /*outicmp->icmp_code = 0; - set by xzalloc */
                         outicmp->icmp_id = htons(ident);
-                        outdata = (struct outdata_t *)((char *)outicmp + SIZEOF_ICMP_HDR);
-                } else {
-                        outdata = (struct outdata_t *)(outudp + 1);
+                        outdata = (void*)((char *)outicmp + SIZEOF_ICMP_HDR);
                 }
         }
 #if ENABLE_TRACEROUTE6
         if (af == AF_INET6) {
-                outdata = (void*)((char*)outip
-                                + sizeof(struct ip6_hdr)
-                                + sizeof(struct udphdr)
-                                );
+                outdata = (void*)(outudp6 + 1);
+                if (op & OPT_USE_ICMP) {
+                        ident |= 0x8000;
+                        outicmp6->icmp_type = ICMP6_ECHO_REQUEST;
+                        /*outicmp->icmp_code = 0; - set by xzalloc */
+                        outicmp6->icmp_id = htons(ident);
+                        outdata = (void*)((char *)outicmp6 + SIZEOF_ICMP_HDR);
+                }
         }
 #endif
 
@@ -1013,26 +1084,16 @@ common_traceroute_main(int op, char **argv)
 //TODO: why we don't do it for IPv4?
                 len_and_sockaddr *source_lsa;
 
-                int probe_fd = xsocket(af, SOCK_DGRAM, 0);
-                if (op & OPT_DEVICE)
-                        setsockopt_bindtodevice(probe_fd, device);
-                set_nport(&dest_lsa->u.sa, htons(1025));
-                /* dummy connect. makes kernel pick source IP (and port) */
-                xconnect(probe_fd, &dest_lsa->u.sa, dest_lsa->len);
                 set_nport(&dest_lsa->u.sa, htons(port));
-
-                /* read IP and port */
-                source_lsa = get_sock_lsa(probe_fd);
+                /* Connect makes kernel pick source IP and port */
+                xconnect(sndsock, &dest_lsa->u.sa, dest_lsa->len);
+                /* Read IP and port */
+                source_lsa = get_sock_lsa(sndsock);
                 if (source_lsa == NULL)
                         bb_simple_error_msg_and_die("can't get probe addr");
-
-                close(probe_fd);
-
-                /* bind our sockets to this IP (but not port) */
+                /* bind our recv socket to this IP (but not port) */
                 set_nport(&source_lsa->u.sa, 0);
-                xbind(sndsock, &source_lsa->u.sa, source_lsa->len);
                 xbind(rcvsock, &source_lsa->u.sa, source_lsa->len);
-
                 free(source_lsa);
         }
 #endif
@@ -1067,10 +1128,9 @@ common_traceroute_main(int op, char **argv)
                         unsigned t1;
                         unsigned t2;
                         int left_ms;
-                        struct ip *ip;
 
                         fflush_all();
-                        if (probe != 0 && pausemsecs > 0)
+                        if (probe != 0)
                                 msleep(pausemsecs);
 
                         send_probe(++seq, ttl);
@@ -1099,15 +1159,18 @@ common_traceroute_main(int op, char **argv)
                                 }
 
                                 print_delta_ms(t1, t2);
-                                ip = (struct ip *)recv_pkt;
 
-                                if (from_lsa->u.sa.sa_family == AF_INET)
-                                        if (op & OPT_TTL_FLAG)
+                                if (from_lsa->u.sa.sa_family == AF_INET) {
+                                        if (op & OPT_TTL_FLAG) {
+                                                struct ip *ip = (struct ip *)recv_pkt;
                                                 printf(" (%d)", ip->ip_ttl);
+                                        }
+                                }
 
-                                /* time exceeded in transit */
+                                /* Got a "time exceeded in transit" icmp message? */
                                 if (icmp_code == -1)
                                         break;
+
                                 icmp_code--;
                                 switch (icmp_code) {
 #if ENABLE_TRACEROUTE6
@@ -1115,12 +1178,13 @@ common_traceroute_main(int op, char **argv)
                                         got_there = 1;
                                         break;
 #endif
-                                case ICMP_UNREACH_PORT:
+                                case ICMP_UNREACH_PORT: {
+                                        struct ip *ip = (struct ip *)recv_pkt;
                                         if (ip->ip_ttl <= 1)
                                                 printf(" !");
                                         got_there = 1;
                                         break;
-
+                                }
                                 case ICMP_UNREACH_NET:
 #if ENABLE_TRACEROUTE6 && (ICMP6_DST_UNREACH_NOROUTE != ICMP_UNREACH_NET)
                                 case ICMP6_DST_UNREACH_NOROUTE << 8: