From 25e63f1e56f5acdcf91893a1b92ad1e0f2f552d8 Mon Sep 17 00:00:00 2001 From: Simon Kelley Date: Wed, 25 Nov 2020 21:17:52 +0000 Subject: Handle caching with EDNS options better. If we add the EDNS client subnet option, or the client's MAC address, then the reply we get back may very depending on that. Since the cache is ignorant of such things, it's not safe to cache such replies. This patch determines when a dangerous EDNS option is being added and disables caching. Note that for much the same reason, we can't combine multiple queries for the same question when dangerous EDNS options are being added, and the code now handles that in the same way. This query combining is required for security against cache poisoning, so disabling the cache has a security function as well as a correctness one. --- man/dnsmasq.8 | 4 +-- src/dnsmasq.h | 3 ++- src/edns0.c | 75 ++++++++++++++++++++++++++++++++------------------- src/forward.c | 41 ++++++++++++++++++---------- 4 files changed, 78 insertions(+), 45 deletions(-) --- a/man/dnsmasq.8 +++ b/man/dnsmasq.8 @@ -690,8 +690,8 @@ still marks the request so that no upstr address information either. The default is zero for both IPv4 and IPv6. Note that upstream nameservers may be configured to return different results based on this information, but the dnsmasq cache -does not take account. If a dnsmasq instance is configured such that -different results may be encountered, caching should be disabled. +does not take account. Caching is therefore disabled for such replies, +unless the subnet address being added is constant. For example, .B --add-subnet=24,96 --- a/src/dnsmasq.h +++ b/src/dnsmasq.h @@ -644,6 +644,7 @@ struct hostsfile { #define FREC_TEST_PKTSZ 256 #define FREC_HAS_EXTRADATA 512 #define FREC_HAS_PHEADER 1024 +#define FREC_NO_CACHE 2048 #define HASH_SIZE 32 /* SHA-256 digest size */ @@ -1628,7 +1629,7 @@ size_t add_pseudoheader(struct dns_heade unsigned short udp_sz, int optno, unsigned char *opt, size_t optlen, int set_do, int replace); size_t add_do_bit(struct dns_header *header, size_t plen, unsigned char *limit); size_t add_edns0_config(struct dns_header *header, size_t plen, unsigned char *limit, - union mysockaddr *source, time_t now, int *check_subnet); + union mysockaddr *source, time_t now, int *check_subnet, int *cacheable); int check_source(struct dns_header *header, size_t plen, unsigned char *pseudoheader, union mysockaddr *peer); /* arp.c */ --- a/src/edns0.c +++ b/src/edns0.c @@ -264,7 +264,8 @@ static void encoder(unsigned char *in, c out[3] = char64(in[2]); } -static size_t add_dns_client(struct dns_header *header, size_t plen, unsigned char *limit, union mysockaddr *l3, time_t now) +static size_t add_dns_client(struct dns_header *header, size_t plen, unsigned char *limit, + union mysockaddr *l3, time_t now, int *cacheablep) { int maclen, replace = 2; /* can't get mac address, just delete any incoming. */ unsigned char mac[DHCP_CHADDR_MAX]; @@ -273,6 +274,7 @@ static size_t add_dns_client(struct dns_ if ((maclen = find_mac(l3, mac, 1, now)) == 6) { replace = 1; + *cacheablep = 0; if (option_bool(OPT_MAC_HEX)) print_mac(encode, mac, maclen); @@ -288,14 +290,18 @@ static size_t add_dns_client(struct dns_ } -static size_t add_mac(struct dns_header *header, size_t plen, unsigned char *limit, union mysockaddr *l3, time_t now) +static size_t add_mac(struct dns_header *header, size_t plen, unsigned char *limit, + union mysockaddr *l3, time_t now, int *cacheablep) { int maclen; unsigned char mac[DHCP_CHADDR_MAX]; if ((maclen = find_mac(l3, mac, 1, now)) != 0) - plen = add_pseudoheader(header, plen, limit, PACKETSZ, EDNS0_OPTION_MAC, mac, maclen, 0, 0); - + { + *cacheablep = 0; + plen = add_pseudoheader(header, plen, limit, PACKETSZ, EDNS0_OPTION_MAC, mac, maclen, 0, 0); + } + return plen; } @@ -313,17 +319,18 @@ static void *get_addrp(union mysockaddr return &addr->in.sin_addr; } -static size_t calc_subnet_opt(struct subnet_opt *opt, union mysockaddr *source) +static size_t calc_subnet_opt(struct subnet_opt *opt, union mysockaddr *source, int *cacheablep) { /* http://tools.ietf.org/html/draft-vandergaast-edns-client-subnet-02 */ int len; void *addrp = NULL; int sa_family = source->sa.sa_family; - + int cacheable = 0; + opt->source_netmask = 0; opt->scope_netmask = 0; - + if (source->sa.sa_family == AF_INET6 && daemon->add_subnet6) { opt->source_netmask = daemon->add_subnet6->mask; @@ -331,6 +338,7 @@ static size_t calc_subnet_opt(struct sub { sa_family = daemon->add_subnet6->addr.sa.sa_family; addrp = get_addrp(&daemon->add_subnet6->addr, sa_family); + cacheable = 1; } else addrp = &source->in6.sin6_addr; @@ -343,6 +351,7 @@ static size_t calc_subnet_opt(struct sub { sa_family = daemon->add_subnet4->addr.sa.sa_family; addrp = get_addrp(&daemon->add_subnet4->addr, sa_family); + cacheable = 1; /* Address is constant */ } else addrp = &source->in.sin_addr; @@ -350,8 +359,6 @@ static size_t calc_subnet_opt(struct sub opt->family = htons(sa_family == AF_INET6 ? 2 : 1); - len = 0; - if (addrp && opt->source_netmask != 0) { len = ((opt->source_netmask - 1) >> 3) + 1; @@ -359,18 +366,26 @@ static size_t calc_subnet_opt(struct sub if (opt->source_netmask & 7) opt->addr[len-1] &= 0xff << (8 - (opt->source_netmask & 7)); } + else + { + cacheable = 1; /* No address ever supplied. */ + len = 0; + } + + if (cacheablep) + *cacheablep = cacheable; return len + 4; } -static size_t add_source_addr(struct dns_header *header, size_t plen, unsigned char *limit, union mysockaddr *source) +static size_t add_source_addr(struct dns_header *header, size_t plen, unsigned char *limit, union mysockaddr *source, int *cacheable) { /* http://tools.ietf.org/html/draft-vandergaast-edns-client-subnet-02 */ int len; struct subnet_opt opt; - len = calc_subnet_opt(&opt, source); + len = calc_subnet_opt(&opt, source, cacheable); return add_pseudoheader(header, plen, (unsigned char *)limit, PACKETSZ, EDNS0_OPTION_CLIENT_SUBNET, (unsigned char *)&opt, len, 0, 0); } @@ -383,18 +398,18 @@ int check_source(struct dns_header *head unsigned char *p; int code, i, rdlen; - calc_len = calc_subnet_opt(&opt, peer); - - if (!(p = skip_name(pseudoheader, header, plen, 10))) - return 1; - - p += 8; /* skip UDP length and RCODE */ + calc_len = calc_subnet_opt(&opt, peer, NULL); - GETSHORT(rdlen, p); - if (!CHECK_LEN(header, p, plen, rdlen)) - return 1; /* bad packet */ - - /* check if option there */ + if (!(p = skip_name(pseudoheader, header, plen, 10))) + return 1; + + p += 8; /* skip UDP length and RCODE */ + + GETSHORT(rdlen, p); + if (!CHECK_LEN(header, p, plen, rdlen)) + return 1; /* bad packet */ + + /* check if option there */ for (i = 0; i + 4 < rdlen; i += len + 4) { GETSHORT(code, p); @@ -412,24 +427,28 @@ int check_source(struct dns_header *head return 1; } +/* Set *check_subnet if we add a client subnet option, which needs to checked + in the reply. Set *cacheable to zero if we add an option which the answer + may depend on. */ size_t add_edns0_config(struct dns_header *header, size_t plen, unsigned char *limit, - union mysockaddr *source, time_t now, int *check_subnet) + union mysockaddr *source, time_t now, int *check_subnet, int *cacheable) { *check_subnet = 0; - + *cacheable = 1; + if (option_bool(OPT_ADD_MAC)) - plen = add_mac(header, plen, limit, source, now); + plen = add_mac(header, plen, limit, source, now, cacheable); if (option_bool(OPT_MAC_B64) || option_bool(OPT_MAC_HEX)) - plen = add_dns_client(header, plen, limit, source, now); - + plen = add_dns_client(header, plen, limit, source, now, cacheable); + if (daemon->dns_client_id) plen = add_pseudoheader(header, plen, limit, PACKETSZ, EDNS0_OPTION_NOMCPEID, (unsigned char *)daemon->dns_client_id, strlen(daemon->dns_client_id), 0, 1); if (option_bool(OPT_CLIENT_SUBNET)) { - plen = add_source_addr(header, plen, limit, source); + plen = add_source_addr(header, plen, limit, source, cacheable); *check_subnet = 1; } --- a/src/forward.c +++ b/src/forward.c @@ -344,13 +344,10 @@ static int forward_query(int udpfd, unio { /* Query from new source, but the same query may be in progress from another source. If so, just add this client to the - list that will get the reply. + list that will get the reply.*/ - Note that is the EDNS client subnet option is in use, we can't do this, - as the clients (and therefore query EDNS options) will be different - for each query. The EDNS subnet code has checks to avoid - attacks in this case. */ - if (!option_bool(OPT_CLIENT_SUBNET) && (forward = lookup_frec_by_query(hash, fwd_flags))) + if (!option_bool(OPT_ADD_MAC) && !option_bool(OPT_MAC_B64) && + (forward = lookup_frec_by_query(hash, fwd_flags))) { /* Note whine_malloc() zeros memory. */ if (!daemon->free_frec_src && @@ -447,18 +444,21 @@ static int forward_query(int udpfd, unio if (!flags && forward) { struct server *firstsentto = start; - int subnet, forwarded = 0; + int subnet, cacheable, forwarded = 0; size_t edns0_len; unsigned char *pheader; /* If a query is retried, use the log_id for the retry when logging the answer. */ forward->frec_src.log_id = daemon->log_id; - plen = add_edns0_config(header, plen, ((unsigned char *)header) + PACKETSZ, &forward->frec_src.source, now, &subnet); + plen = add_edns0_config(header, plen, ((unsigned char *)header) + PACKETSZ, &forward->frec_src.source, now, &subnet, &cacheable); if (subnet) forward->flags |= FREC_HAS_SUBNET; - + + if (!cacheable) + forward->flags |= FREC_NO_CACHE; + #ifdef HAVE_DNSSEC if (option_bool(OPT_DNSSEC_VALID) && do_dnssec) { @@ -642,7 +642,7 @@ static size_t process_reply(struct dns_h } } #endif - + if ((pheader = find_pseudoheader(header, n, &plen, &sizep, &is_sign, NULL))) { /* Get extended RCODE. */ @@ -1244,6 +1244,11 @@ void reply_query(int fd, int family, tim header->hb4 |= HB4_CD; else header->hb4 &= ~HB4_CD; + + /* Never cache answers which are contingent on the source or MAC address EDSN0 option, + since the cache is ignorant of such things. */ + if (forward->flags & FREC_NO_CACHE) + no_cache_dnssec = 1; if ((nn = process_reply(header, now, forward->sentto, (size_t)n, check_rebind, no_cache_dnssec, cache_secure, bogusanswer, forward->flags & FREC_AD_QUESTION, forward->flags & FREC_DO_QUESTION, @@ -1788,7 +1793,7 @@ unsigned char *tcp_request(int confd, ti int local_auth = 0; #endif int checking_disabled, do_bit, added_pheader = 0, have_pseudoheader = 0; - int check_subnet, no_cache_dnssec = 0, cache_secure = 0, bogusanswer = 0; + int check_subnet, cacheable, no_cache_dnssec = 0, cache_secure = 0, bogusanswer = 0; size_t m; unsigned short qtype; unsigned int gotname; @@ -1959,7 +1964,7 @@ unsigned char *tcp_request(int confd, ti char *domain = NULL; unsigned char *oph = find_pseudoheader(header, size, NULL, NULL, NULL, NULL); - size = add_edns0_config(header, size, ((unsigned char *) header) + 65536, &peer_addr, now, &check_subnet); + size = add_edns0_config(header, size, ((unsigned char *) header) + 65536, &peer_addr, now, &check_subnet, &cacheable); if (gotname) flags = search_servers(now, &addrp, gotname, daemon->namebuff, &type, &domain, &norebind); @@ -2122,6 +2127,11 @@ unsigned char *tcp_request(int confd, ti break; } + /* Never cache answers which are contingent on the source or MAC address EDSN0 option, + since the cache is ignorant of such things. */ + if (!cacheable) + no_cache_dnssec = 1; + m = process_reply(header, now, last_server, (unsigned int)m, option_bool(OPT_NO_REBIND) && !norebind, no_cache_dnssec, cache_secure, bogusanswer, ad_reqd, do_bit, added_pheader, check_subnet, &peer_addr); @@ -2385,10 +2395,13 @@ static struct frec *lookup_frec_by_query struct frec *f; /* FREC_DNSKEY and FREC_DS_QUERY are never set in flags, so the test below - ensures that no frec created for internal DNSSEC query can be returned here. */ + ensures that no frec created for internal DNSSEC query can be returned here. + + Similarly FREC_NO_CACHE is never set in flags, so a query which is + contigent on a particular source address EDNS0 option will never be matched. */ #define FLAGMASK (FREC_CHECKING_DISABLED | FREC_AD_QUESTION | FREC_DO_QUESTION \ - | FREC_HAS_PHEADER | FREC_DNSKEY_QUERY | FREC_DS_QUERY) + | FREC_HAS_PHEADER | FREC_DNSKEY_QUERY | FREC_DS_QUERY | FREC_NO_CACHE) for(f = daemon->frec_list; f; f = f->next) if (f->sentto &&