diff options
author | Paul Warren <pdw@ex-parrot.com> | 2017-01-05 21:16:43 +0000 |
---|---|---|
committer | Paul Warren <pdw@ex-parrot.com> | 2017-01-05 21:16:43 +0000 |
commit | 3fae1acf1478f36b16763660bbf93c7d32a6a028 (patch) | |
tree | 1da4a621fb29ad4d7749c5e6ed684fbd600af5b3 | |
parent | 9addd978c444aabfc2af6fa888436ac00770a4c8 (diff) | |
download | iftop-3fae1acf1478f36b16763660bbf93c7d32a6a028.zip |
Fix handling of packets that are too short to be valid - Frédéric Perrin <fperrin@brocade.com>
When packets that are too short to be valid IP packets happen to start
with 0x45 or 0x60, iftop will still try to read source and destination
addresses, which will usually just be random garbage.
Note the assumption about what libpcap guarantees in the comments to
handle_ip_packet():
* It is assumed that the snaplen (currently hard-coded to 1000) is
* big enough to always capture the IP header past the L2 encap, and
* that pcap never truncates the packet to less than snaplen; in
* other words, that pcaphdr->caplen = MIN(pcaphdr->len, snaplen).
-rw-r--r-- | ChangeLog | 3 | ||||
-rw-r--r-- | iftop.c | 91 |
2 files changed, 56 insertions, 38 deletions
@@ -6,6 +6,9 @@ Unattributed items are by Paul Warren and Chris Lightfoot. 1.0 +* Fix handling of short packets + Frédéric Perrin <fperrin@brocade.com> + * Fix MAC address display Kevin Darbyshire-Bryant <kevin@darbyshire-bryant.me.uk> @@ -249,16 +249,16 @@ void assign_addr_pair(addr_pair* ap, struct ip* iptr, int flip) { } } -static void handle_ip_packet(struct ip* iptr, int hw_dir) +static void handle_ip_packet(struct ip* iptr, int hw_dir, int pld_len) { int direction = 0; /* incoming */ + int len; history_type* ht; union { history_type **ht_pp; void **void_pp; } u_ht = { &ht }; addr_pair ap; - unsigned int len = 0; struct in6_addr scribdst; /* Scratch pad. */ struct in6_addr scribsrc; /* Scratch pad. */ /* Reinterpret packet type. */ @@ -268,7 +268,21 @@ static void handle_ip_packet(struct ip* iptr, int hw_dir) tick(0); - if( (IP_V(iptr) ==4 && options.netfilter == 0) + /* + * Sanity check: drop obviously short packets. + * pld_len comes from pcaphdr->len - sizeof(struct l2_header). + * + * It is assumed that the snaplen (currently hard-coded to 1000) is + * big enough to always capture the IP header past the L2 encap, and + * that pcap never truncates the packet to less than snaplen; in + * other words, that pcaphdr->caplen = MIN(pcaphdr->len, snaplen). + */ + if (pld_len < sizeof (struct ip)) + return; + if (IP_V(iptr) == 6 && pld_len < sizeof (struct ip6_hdr)) + return; + + if( (IP_V(iptr) == 4 && options.netfilter == 0) || (IP_V(iptr) == 6 && options.netfilter6 == 0) ) { /* * Net filter is off, so assign direction based on MAC address @@ -424,7 +438,6 @@ static void handle_ip_packet(struct ip* iptr, int hw_dir) break; } - if(hash_find(history, &ap, u_ht.void_pp) == HASH_STATUS_KEY_NOT_FOUND) { ht = history_create(); hash_insert(history, &ap, ht); @@ -434,19 +447,13 @@ static void handle_ip_packet(struct ip* iptr, int hw_dir) switch (options.bandwidth_unit) { case OPTION_BW_BITS: case OPTION_BW_BYTES: - switch (IP_V(iptr)) { - case 4: - len = ntohs(iptr->ip_len); - break; - case 6: - len = ntohs(ip6tr->ip6_plen) + 40; - default: - break; - } + len = pld_len; break; case OPTION_BW_PKTS: len = 1; break; + default: + return; } /* Update record */ @@ -476,7 +483,7 @@ static void handle_ip_packet(struct ip* iptr, int hw_dir) static void handle_raw_packet(unsigned char* args, const struct pcap_pkthdr* pkthdr, const unsigned char* packet) { - handle_ip_packet((struct ip*)packet, -1); + handle_ip_packet((struct ip*)packet, -1, pkthdr->len); } #ifdef DLT_PFLOG @@ -490,18 +497,19 @@ static void handle_pflog_packet(unsigned char* args, const struct pcap_pkthdr* p hdrlen = BPF_WORDALIGN(hdr->length); length -= hdrlen; packet += hdrlen; - handle_ip_packet((struct ip*)packet, -1); + handle_ip_packet((struct ip*)packet, -1, length); } #endif static void handle_null_packet(unsigned char* args, const struct pcap_pkthdr* pkthdr, const unsigned char* packet) { - handle_ip_packet((struct ip*)(packet + 4), -1); + handle_ip_packet((struct ip*)(packet + 4), -1, pkthdr->len); } -static void handle_llc_packet(const struct llc* llc, int dir) { - - struct ip* ip = (struct ip*)((void*)llc + sizeof(struct llc)); +static void handle_llc_packet(const struct llc* llc, int dir, int llclen) { + int hdrlen = sizeof(struct llc); + int pldlen = llclen - hdrlen; + struct ip* ip = (struct ip*)((void*)llc + hdrlen); /* Taken from tcpdump/print-llc.c */ if(llc->ssap == LLCSAP_SNAP && llc->dsap == LLCSAP_SNAP @@ -513,11 +521,11 @@ static void handle_llc_packet(const struct llc* llc, int dir) { switch(orgcode) { case OUI_ENCAP_ETHER: case OUI_CISCO_90: - handle_ip_packet(ip, dir); + handle_ip_packet(ip, dir, pldlen); break; case OUI_APPLETALK: if(et == ETHERTYPE_ATALK) { - handle_ip_packet(ip, dir); + handle_ip_packet(ip, dir, pldlen); } break; default:; @@ -529,33 +537,36 @@ static void handle_llc_packet(const struct llc* llc, int dir) { static void handle_tokenring_packet(unsigned char* args, const struct pcap_pkthdr* pkthdr, const unsigned char* packet) { struct token_header *trp; + int hdrlen = 0; int dir = -1; trp = (struct token_header *)packet; if(IS_SOURCE_ROUTED(trp)) { - packet += RIF_LENGTH(trp); + hdrlen += RIF_LENGTH(trp); } - packet += TOKEN_HDRLEN; + hdrlen += TOKEN_HDRLEN; + packet += hdrlen; if(memcmp(trp->token_shost, if_hw_addr, 6) == 0 ) { /* packet leaving this i/f */ dir = 1; - } - else if(memcmp(trp->token_dhost, if_hw_addr, 6) == 0 || memcmp("\xFF\xFF\xFF\xFF\xFF\xFF", trp->token_dhost, 6) == 0) { + } + else if(memcmp(trp->token_dhost, if_hw_addr, 6) == 0 || memcmp("\xFF\xFF\xFF\xFF\xFF\xFF", trp->token_dhost, 6) == 0) { /* packet entering this i/f */ dir = 0; } /* Only know how to deal with LLC encapsulated packets */ if(FRAME_TYPE(trp) == TOKEN_FC_LLC) { - handle_llc_packet((struct llc*)packet, dir); + handle_llc_packet((struct llc*)packet, dir, pkthdr->len - hdrlen); } } static void handle_ppp_packet(unsigned char* args, const struct pcap_pkthdr* pkthdr, const unsigned char* packet) { - register u_int caplen = pkthdr->caplen; - u_int proto; + register u_int length = pkthdr->len; + register u_int caplen = pkthdr->caplen; + u_int proto; if (caplen < 2) return; @@ -565,12 +576,14 @@ static void handle_ppp_packet(unsigned char* args, const struct pcap_pkthdr* pkt return; packet += 2; + length -= 2; proto = EXTRACT_16BITS(packet); packet += 2; + length -= 2; if(proto == PPP_IP || proto == ETHERTYPE_IP || proto == ETHERTYPE_IPV6) { - handle_ip_packet((struct ip*)packet, -1); + handle_ip_packet((struct ip*)packet, -1, length); } } } @@ -593,24 +606,25 @@ static void handle_cooked_packet(unsigned char *args, const struct pcap_pkthdr * dir=1; break; } - handle_ip_packet((struct ip*)(packet+SLL_HDR_LEN), dir); + handle_ip_packet((struct ip*)(packet+SLL_HDR_LEN), dir, + thdr->len - SLL_HDR_LEN); } #endif /* DLT_LINUX_SLL */ static void handle_eth_packet(unsigned char* args, const struct pcap_pkthdr* pkthdr, const unsigned char* packet) { struct ether_header *eptr; - int ether_type; - const unsigned char *payload; + int ether_type, hdrlen; + eptr = (struct ether_header*)packet; ether_type = ntohs(eptr->ether_type); - payload = packet + sizeof(struct ether_header); + hdrlen = sizeof(struct ether_header); if(ether_type == ETHERTYPE_8021Q) { struct vlan_8021q_header* vptr; - vptr = (struct vlan_8021q_header*)payload; + vptr = (struct vlan_8021q_header*) (packet + hdrlen); ether_type = ntohs(vptr->ether_type); - payload += sizeof(struct vlan_8021q_header); + hdrlen += sizeof(struct vlan_8021q_header); } if(ether_type == ETHERTYPE_IP || ether_type == ETHERTYPE_IPV6) { @@ -634,8 +648,8 @@ static void handle_eth_packet(unsigned char* args, const struct pcap_pkthdr* pkt } /* Distinguishing ip_hdr and ip6_hdr will be done later. */ - iptr = (struct ip*)(payload); /* alignment? */ - handle_ip_packet(iptr, dir); + iptr = (struct ip*) (packet + hdrlen); /* alignment? */ + handle_ip_packet(iptr, dir, pkthdr->len - hdrlen); } } @@ -648,7 +662,8 @@ static void handle_radiotap_packet(unsigned char* args, const struct pcap_pkthdr { /* 802.11 MAC header is = 34 bytes (not sure if that's universally true) */ /* We could try harder to figure out hardware direction from the MAC header */ - handle_ip_packet((struct ip*)(packet + ((struct radiotap_header *)packet)->it_len + 34),-1); + int hdrlen = ((struct radiotap_header *)packet)->it_len + 34; + handle_ip_packet((struct ip*)(packet + hdrlen), -1, pkthdr->len - hdrlen); } |