Code cleanup.

This allows us to query for the root zone DNSKEY RRset and validate
it, thus automatically handling KSK rollover.

CHANGELOG

@@ -17,6 +17,56 @@ version 2.69
 	    dnsmasq, [fe80::] with the link-local address. 
 	    Thanks to Tsachi Kimeldorfer for championing this.
 
+	    DNSSEC validation and caching. Dnsmasq needs to be
+	    compiled with this enabled, with 
+	    
+	    make dnsmasq COPTS=-DHAVE_DNSSEC
+	    
+	    this add dependencies on the nettle crypto library and the 
+	    gmp maths library. It's possible to have these linked
+	    statically with
+	    
+	    make dnsmasq COPTS='-DHAVE_DNSSEC -DHAVE_DNSSEC_STATIC'
+	    
+	    which bloats the dnsmasq binary to over a megabyte, but
+	    saves the size of the shared libraries which are five
+	    times that size.
+	    To enable, DNSSEC, you will need a set of
+	    trust-anchors. Now that the TLDs are signed, this can be
+	    the keys for the root zone, and for convenience they are
+	    included in trust-anchors.conf in the dnsmasq
+	    distribution. You should of course check that these are
+	    legitimate and up-to-date. So, adding
+	    
+	    conf-file=/path/to/trust-anchors.conf
+	    dnssec
+
+	    to your config is all thats needed to get things
+	    working. The upstream nameservers have to be DNSSEC-capable
+	    too, of course. Many ISP nameservers aren't, but the
+	    Google public nameservers (8.8.8.8 and 8.8.4.4) are.
+	    When DNSSEC is configured, dnsmasq validates any queries 
+	    for domains which are signed. Query results which are 
+	    bogus are replaced with SERVFAIL replies, and results 
+	    which are correctly signed have the AD bit set. In 
+	    addition, and just as importantly, dnsmasq supplies 
+	    correct DNSSEC information to clients which are doing 
+	    their own validation, and caches DNSKEY, DS and RRSIG
+	    records, which significantly improve the performance of 
+	    downstream validators. Setting --log-queries will show 
+	    DNSSEC in action.
+
+	    The development of DNSSEC in dnsmasq was started by 
+	    Giovanni Bajo, to whom huge thanks are owed. It has been
+	    supported by Comcast, whose techfund grant has allowed for 
+	    an invaluable period of full-time work to get it to 
+	    a workable state.
+ 
+	    Add --rev-server. Thanks to Dave Taht for suggesting this.
+	    
+	    Add --servers-file. Allows dynamic update of upstream servers 
+	    full access to configuration. 
+
  	    
 version 2.68
             Use random addresses for DHCPv6 temporary address

Makefile

@@ -61,6 +61,7 @@ lua_cflags =    `echo $(COPTS) | $(top)/bld/pkg-wrapper HAVE_LUASCRIPT $(PKG_CON
 lua_libs =      `echo $(COPTS) | $(top)/bld/pkg-wrapper HAVE_LUASCRIPT $(PKG_CONFIG) --libs lua5.1` 
 nettle_cflags = `echo $(COPTS) | $(top)/bld/pkg-wrapper HAVE_DNSSEC $(PKG_CONFIG) --cflags nettle hogweed`
 nettle_libs =   `echo $(COPTS) | $(top)/bld/pkg-wrapper HAVE_DNSSEC $(PKG_CONFIG) --libs nettle hogweed`
+gmp_libs =      `echo $(COPTS) | $(top)/bld/pkg-wrapper HAVE_DNSSEC $(PKG_CONFIG) --copy -lgmp`
 sunos_libs =    `if uname | grep SunOS >/dev/null 2>&1; then echo -lsocket -lnsl -lposix4; fi`
 version =     -DVERSION='\"`$(top)/bld/get-version $(top)`\"'
 
@@ -77,7 +78,7 @@ all : $(BUILDDIR)
 	@cd $(BUILDDIR) && $(MAKE) \
  top="$(top)" \
  build_cflags="$(version) $(dbus_cflags) $(idn_cflags) $(ct_cflags) $(lua_cflags) $(nettle_cflags)" \
- build_libs="$(dbus_libs) $(idn_libs) $(ct_libs) $(lua_libs) $(sunos_libs) $(nettle_libs)" \
+ build_libs="$(dbus_libs) $(idn_libs) $(ct_libs) $(lua_libs) $(sunos_libs) $(nettle_libs) $(gmp_libs)" \
  -f $(top)/Makefile dnsmasq 
 
 mostly_clean :
@@ -101,7 +102,7 @@ all-i18n : $(BUILDDIR)
  top="$(top)" \
  i18n=-DLOCALEDIR=\'\"$(LOCALEDIR)\"\' \
  build_cflags="$(version) $(dbus_cflags) $(ct_cflags) $(lua_cflags) $(nettle_cflags) `$(PKG_CONFIG) --cflags libidn`" \
- build_libs="$(dbus_libs) $(ct_libs) $(lua_libs) $(sunos_libs) $(nettle_libs) `$(PKG_CONFIG) --libs libidn`"  \
+ build_libs="$(dbus_libs) $(ct_libs) $(lua_libs) $(sunos_libs) $(nettle_libs) $(gmp_libs) `$(PKG_CONFIG) --libs libidn`"  \
  -f $(top)/Makefile dnsmasq
 	for f in `cd $(PO); echo *.po`; do \
 		cd $(top) && cd $(BUILDDIR) && $(MAKE) top="$(top)" -f $(top)/Makefile $${f%.po}.mo; \

bld/pkg-wrapper

@@ -9,19 +9,24 @@ shift
 
 in=`cat`
 
-if grep "^\#[[:space:]]*define[[:space:]]*${search}_STATIC" config.h >/dev/null 2>&1 || \
-   echo $in | grep ${search}_STATIC >/dev/null 2>&1; then
-   if [ $op = "--libs" ]; then
-     pkg=`$pkg --static $op $*`
-     echo "-Wl,-Bstatic $pkg -Wl,-Bdynamic"
-     exit 0
-   fi
-fi
-
 if grep "^\#[[:space:]]*define[[:space:]]*$search" config.h >/dev/null 2>&1 || \
     echo $in | grep $search >/dev/null 2>&1; then
-    pkg=`$pkg $op $*`
-    echo "$pkg"
 
+    if [ $op = "--copy" ]; then
+	pkg="$*"
+    else
+	pkg=`$pkg $op $*` 
+    fi
+
+    if grep "^\#[[:space:]]*define[[:space:]]*${search}_STATIC" config.h >/dev/null 2>&1 || \
+	echo $in | grep ${search}_STATIC >/dev/null 2>&1; then
+	if [ $op = "--libs" ] || [ $op = "--copy" ]; then
+	    echo "-Wl,-Bstatic $pkg -Wl,-Bdynamic"
+	else
+	    echo "$pkg" 
+	fi
+    else
+	echo "$pkg"
+    fi
 fi
 

contrib/try-all-ns/dnsmasq-2.68-try-all-ns

@@ -0,0 +1,29 @@
+From: Jesse Glick <jglick@cloudbees.com>
+To: dnsmasq-discuss@lists.thekelleys.org.uk
+Subject: Re: [Dnsmasq-discuss] Ability to delegate to one server but fall
+ back to another after NXDOMAIN?
+
+
+On Wed, Jan 15, 2014 at 12:30 PM, Simon Kelley <simon@thekelleys.org.uk> wrote:
+> > There's a (very old) patch in contrib/try-all-ns that would make a starting point
+This does not apply against trunk, so I tried to rework it. The
+following appears to do what I expect:
+
+diff --git a/src/forward.c b/src/forward.c
+index 8167229..76070b5 100644
+--- a/src/forward.c
++++ b/src/forward.c
+@@ -610,7 +610,11 @@ void reply_query(int fd, int family, time_t now)
+
+   if ((RCODE(header) == SERVFAIL || RCODE(header) == REFUSED) &&
+       !option_bool(OPT_ORDER) &&
+-      forward->forwardall == 0)
++      forward->forwardall == 0 ||
++      /* try each in turn */
++      RCODE(header) == NXDOMAIN &&
++      option_bool(OPT_ORDER) &&
++      server->next != NULL)
+     /* for broken servers, attempt to send to another one. */
+     {
+       unsigned char *pheader;
+

debian/changelog

@@ -1,3 +1,9 @@
+dnsmasq (2.69-1) unstable; urgency=low
+
+   * New upstream.
+
+ -- Simon Kelley <simon@thekelleys.org.uk>  Tue, 4 Feb 2014 16:28:12 +0000
+
 dnsmasq (2.68-1) unstable; urgency=low
 
    * New upstream. (closes: #730553)

debian/control

@@ -1,7 +1,8 @@
 Source: dnsmasq
 Section: net
 Priority: optional
-Build-depends: gettext, libnetfilter-conntrack-dev [linux-any], libidn11-dev, libdbus-1-dev (>=0.61)
+Build-depends: gettext, libnetfilter-conntrack-dev [linux-any],
+               libidn11-dev, libdbus-1-dev (>=0.61), libgmp-dev, nettle-dev (>=2.4-3)
 Maintainer: Simon Kelley <simon@thekelleys.org.uk>
 Standards-Version: 3.9.3
 

debian/rules

@@ -19,6 +19,10 @@ LDFLAGS = $(shell dpkg-buildflags --get LDFLAGS)
 
 DEB_COPTS = $(COPTS)
 
+# The nettle library in Debian is too old to include
+# ECC support.
+DEB_COPTS += -DNO_NETTLE_ECC
+
 TARGET = install-i18n
 
 DEB_BUILD_ARCH_OS := $(shell dpkg-architecture -qDEB_BUILD_ARCH_OS)
@@ -77,7 +81,7 @@ ifneq (,$(filter uselua,$(DEB_BUILD_OPTIONS)))
      DEB_COPTS += -DHAVE_LUASCRIPT
 endif
 
-ifneq (,$(filter usednssec,$(DEB_BUILD_OPTIONS)))
+ifeq (,$(filter nodnssec,$(DEB_BUILD_OPTIONS)))
      DEB_COPTS += -DHAVE_DNSSEC
 endif
 
@@ -126,12 +130,14 @@ binary-arch:	checkroot
 	        -d debian/base/usr/share/doc/$(package) \
 		-d debian/base/usr/share/doc/$(package)/examples \
 	        -d debian/base/var/run \
+		-d debian/base/usr/share/$(package) \
 	        -d debian/base/var/lib/misc
 	make $(TARGET) PREFIX=/usr DESTDIR=`pwd`/debian/base CFLAGS="$(CFLAGS)" LDFLAGS="$(LDFLAGS)" COPTS="$(DEB_COPTS)" CC=gcc
 ifeq (,$(findstring nodocs,$(DEB_BUILD_OPTIONS)))
 	install -m 644 doc.html debian/base/usr/share/doc/$(package)/.
 	install -m 644 setup.html debian/base/usr/share/doc/$(package)/.
 	install -m 644 dnsmasq.conf.example debian/base/usr/share/doc/$(package)/examples/.
+	install -m 644 trust-anchors.conf debian/base/usr/share/$(package)/.
 	install -m 644 FAQ debian/base/usr/share/doc/$(package)/.
 	gzip -9 debian/base/usr/share/doc/$(package)/FAQ
 	install -m 644 CHANGELOG debian/base/usr/share/doc/$(package)/changelog

dnsmasq.conf.example

@@ -20,6 +20,10 @@
 # Never forward addresses in the non-routed address spaces.
 #bogus-priv
 
+# Uncomment these to enable DNSSEC validation and caching:
+# (Requires dnsmasq to be built with DNSSEC option.)
+#conf-file=%%PREFIX%%/share/dnsmasq/trust-anchors.conf
+#dnssec
 
 # Uncomment this to filter useless windows-originated DNS requests
 # which can trigger dial-on-demand links needlessly.

man/dnsmasq.8

@@ -13,7 +13,10 @@ Dnsmasq accepts DNS queries and either answers them from a small, local,
 cache or forwards them to a real, recursive, DNS server. It loads the
 contents of /etc/hosts so that local hostnames
 which do not appear in the global DNS can be resolved and also answers
-DNS queries for DHCP configured hosts. It can also act as the authoritative DNS server for one or more domains, allowing local names to appear in the global DNS.
+DNS queries for DHCP configured hosts. It can also act as the
+authoritative DNS server for one or more domains, allowing local names
+to appear in the global DNS. It can be configured to do DNSSEC
+validation.
 .PP
 The dnsmasq DHCP server supports static address assignments and multiple
 networks. It automatically
@@ -412,6 +415,14 @@ source address specified but the port may be specified directly as
 part of the source address. Forcing queries to an interface is not
 implemented on all platforms supported by dnsmasq.
 .TP
+.B --rev-server=<ip-address>/<prefix-len>,<ipaddr>[#<port>][@<source-ip>|<interface>[#<port>]]
+This is functionally the same as 
+.B --server, 
+but provides some syntactic sugar to make specifying address-to-name queries easier. For example
+.B --rev-server=1.2.3.0/24,192.168.0.1
+is exactly equivalent to 
+.B --server=/3.2.1.in-addr.arpa/192.168.0.1
+.TP
 .B \-A, --address=/<domain>/[domain/]<ipaddr>
 Specify an IP address to return for any host in the given domains.
 Queries in the domains are never forwarded and always replied to
@@ -587,12 +598,15 @@ validation by clients more efficient. Note that validation by clients is the mos
 clients unable to do validation, use of the AD bit set by dnsmasq is useful, provided that the network between 
 the dnsmasq server and the client is trusted. Dnsmasq must be compiled with HAVE_DNSSEC enabled, and DNSSEC
 trust anchors provided, see 
-.B --dnsskey.
+.B --trust-anchor.
 Because the DNSSEC validation process uses the cache, it is not permitted to reduce the cache size below the default when DNSSEC is enabled.
 .TP
-.B --dnskey=[<class>],<domain>,<flags>,<algorithm>,<base64-key>
+.B --trust-anchor=[<class>],<domain>,<key-tag>,<algorithm>,<digest-type>,<digest>
-Provide DNSKEY records to act a trust anchors for DNSSEC validation. Typically these will be the keys for root zone,
+Provide DS records to act a trust anchors for DNSSEC
-but trust anchors for limited domains are also possible. 
+validation. Typically these will be the DS record(s) for Zone Signing
+key(s) of the root zone,
+but trust anchors for limited domains are also possible. The current
+root-zone trust anchors may be donwloaded from https://data.iana.org/root-anchors/root-anchors.xml 
 .TP
 .B --proxy-dnssec
 Copy the DNSSEC Authenticated Data bit from upstream servers to downstream clients and cache it.  This is an 
@@ -601,7 +615,10 @@ dnsmasq and the upstream servers, and the trustworthiness of the upstream server
 .TP
 .B --dnssec-debug
 Set debugging mode for the DNSSEC validation, set the Checking Disabled bit on upstream queries, 
-and don't convert BOGUS replies to SERVFAIL responses. 
+and don't convert replies which do not validate to responses with
+a return code of SERVFAIL. Note that
+setting this may affect DNS behaviour in bad ways, it is not an
+extra-logging flag and should not be set in production.
 .TP
 .B --auth-zone=<domain>[,<subnet>[/<prefix length>][,<subnet>[/<prefix length>].....]]
 Define a DNS zone for which dnsmasq acts as authoritative server. Locally defined DNS records which are in the domain
@@ -1661,6 +1678,13 @@ files. If extension(s) are given, any files which end in those
 extensions are skipped. Any files whose names end in ~ or start with . or start and end
 with # are always skipped. This flag may be given on the command
 line or in a configuration file.
+.TP
+.B --servers-file=<file>
+A special case of 
+.B --conf-file
+which differs in two respects. Firstly, only --server and --rev-server are allowed 
+in the configuration file included. Secondly, the file is re-read and the configuration
+therein is updated when dnsmasq recieves SIGHUP.
 .SH CONFIG FILE
 At startup, dnsmasq reads
 .I /etc/dnsmasq.conf,

src/cache.c

@@ -55,7 +55,9 @@ static const struct {
   { 41,  "OPT" },
   { 43,  "DS" },
   { 46,  "RRSIG" },
+  { 47,  "NSEC" },
   { 48,  "DNSKEY" },
+  { 50,  "NSEC3" },
   { 249, "TKEY" },
   { 250, "TSIG" },
   { 251, "IXFR" },
@@ -446,19 +448,21 @@ struct crec *cache_insert(char *name, struct all_addr *addr,
   int freed_all = flags & F_REVERSE;
   int free_avail = 0;
 
-  if (daemon->max_cache_ttl != 0 && daemon->max_cache_ttl < ttl)
+  /* Don't log DNSSEC records here, done elsewhere */
-    ttl = daemon->max_cache_ttl;
-
-  /* Don't log keys here, done elsewhere */
   if (flags & (F_IPV4 | F_IPV6 | F_CNAME))
-    log_query(flags | F_UPSTREAM, name, addr, NULL);
+    {
+      log_query(flags | F_UPSTREAM, name, addr, NULL);
+      /* Don;t mess with TTL for DNSSEC records. */
+      if (daemon->max_cache_ttl != 0 && daemon->max_cache_ttl < ttl)
+	ttl = daemon->max_cache_ttl;
+    }
 
   /* if previous insertion failed give up now. */
   if (insert_error)
     return NULL;
   
   /* First remove any expired entries and entries for the name/address we
-     are currently inserting. Fail is we attempt to delete a name from
+     are currently inserting. Fail if we attempt to delete a name from
      /etc/hosts or DHCP. */
   if (!cache_scan_free(name, addr, now, flags))
     {
@@ -592,9 +596,12 @@ void cache_end_insert(void)
   new_chain = NULL;
 }
 
-struct crec *cache_find_by_name(struct crec *crecp, char *name, time_t now, unsigned short prot)
+struct crec *cache_find_by_name(struct crec *crecp, char *name, time_t now, unsigned int prot)
 {
   struct crec *ans;
+  int no_rr = prot & F_NO_RR;
+
+  prot &= ~F_NO_RR;
   
   if (crecp) /* iterating */
     ans = crecp->next;
@@ -643,7 +650,7 @@ struct crec *cache_find_by_name(struct crec *crecp, char *name, time_t now, unsi
 		    }
 		  else
 		    {
-		      if (!insert)
+		      if (!insert && !no_rr)
 			{
 			  insert = up;
 			  ins_flags = crecp->flags & (F_REVERSE | F_IMMORTAL);
@@ -683,7 +690,7 @@ struct crec *cache_find_by_name(struct crec *crecp, char *name, time_t now, unsi
 }
 
 struct crec *cache_find_by_addr(struct crec *crecp, struct all_addr *addr, 
-				time_t now, unsigned short prot)
+				time_t now, unsigned int prot)
 {
   struct crec *ans;
 #ifdef HAVE_IPV6
@@ -980,7 +987,7 @@ void cache_reload(void)
   struct cname *a;
   struct interface_name *intr;
 #ifdef HAVE_DNSSEC
-  struct dnskey *key;
+  struct ds_config *ds;
 #endif
 
   cache_inserted = cache_live_freed = 0;
@@ -1026,17 +1033,17 @@ void cache_reload(void)
 	}
 
 #ifdef HAVE_DNSSEC
-  for (key = daemon->dnskeys; key; key = key->next)
+  for (ds = daemon->ds; ds; ds = ds->next)
     if ((cache = whine_malloc(sizeof(struct crec))) &&
-	(cache->addr.key.keydata = blockdata_alloc(key->key, key->keylen)))
+	(cache->addr.ds.keydata = blockdata_alloc(ds->digest, ds->digestlen)))
       {
-	cache->flags = F_FORWARD | F_IMMORTAL | F_DNSKEY | F_CONFIG | F_NAMEP;
+	cache->flags = F_FORWARD | F_IMMORTAL | F_DS | F_CONFIG | F_NAMEP;
-	cache->name.namep = key->name;
+	cache->name.namep = ds->name;
-	cache->addr.key.keylen = key->keylen;
+	cache->addr.ds.keylen = ds->digestlen;
-	cache->addr.key.algo = key->algo;
+	cache->addr.ds.algo = ds->algo;
-	cache->addr.key.flags = key->flags;
+	cache->addr.ds.keytag = ds->keytag;
-	cache->addr.key.keytag = dnskey_keytag(key->algo, key->flags, (unsigned char *)key->key, key->keylen);
+	cache->addr.ds.digest = ds->digest_type;
-	cache->uid = key->class;
+	cache->uid = ds->class;
 	cache_hash(cache);
       }
 #endif
@@ -1298,12 +1305,10 @@ void dump_cache(time_t now)
 	      {
 		if (cache->flags & F_DNSKEY)
 		  {
-		    char tp[20];
 		    /* RRSIG */
-		    querystr("", tp, cache->addr.sig.type_covered);
 		    a = daemon->addrbuff;
 		    sprintf(a, "%5u %3u %s", cache->addr.sig.keytag,
-			    cache->addr.sig.algo, tp);
+			    cache->addr.sig.algo, querystr("", cache->addr.sig.type_covered));
 		  }
 		else
 		  {
@@ -1379,14 +1384,45 @@ char *record_source(int index)
   return "<unknown>";
 }
 
-void querystr(char *desc, char *str, unsigned short type)
+char *querystr(char *desc, unsigned short type)
 {
   unsigned int i;
+  int len = 10; /* strlen("type=xxxxx") */
+  const char *types = NULL;
+  static char *buff = NULL;
+  static int bufflen = 0;
 
-  sprintf(str, "%s[type=%d]", desc, type); 
   for (i = 0; i < (sizeof(typestr)/sizeof(typestr[0])); i++)
     if (typestr[i].type == type)
-      sprintf(str,"%s[%s]", desc, typestr[i].name);
+      {
+	types = typestr[i].name;
+	len = strlen(types);
+	break;
+      }
+
+  len += 3; /* braces, terminator */
+  len += strlen(desc);
+
+  if (!buff || bufflen < len)
+    {
+      if (buff)
+	free(buff);
+      else if (len < 20)
+	len = 20;
+      
+      buff = whine_malloc(len);
+      bufflen = len;
+    }
+
+  if (buff)
+    {
+      if (types)
+	sprintf(buff, "%s[%s]", desc, types);
+      else
+	sprintf(buff, "%s[type=%d]", desc, type);
+    }
+
+  return buff ? buff : "";
 }
 
 void log_query(unsigned int flags, char *name, struct all_addr *addr, char *arg)

src/config.h

@@ -19,6 +19,7 @@
 #define CHILD_LIFETIME 150 /* secs 'till terminated (RFC1035 suggests > 120s) */
 #define EDNS_PKTSZ 4096 /* default max EDNS.0 UDP packet from RFC5625 */
 #define KEYBLOCK_LEN 35 /* choose to mininise fragmentation when storing DNSSEC keys */
+#define DNSSEC_WORK 50 /* Max number of queries to validate one question */
 #define TIMEOUT 10 /* drop UDP queries after TIMEOUT seconds */
 #define FORWARD_TEST 50 /* try all servers every 50 queries */
 #define FORWARD_TIME 20 /* or 20 seconds */

src/dbus.c

@@ -108,108 +108,6 @@ static void remove_watch(DBusWatch *watch, void *data)
   w = data; /* no warning */
 }
 
-static void add_update_server(union mysockaddr *addr,
-                              union mysockaddr *source_addr,
-			      const char *interface,
-                              const char *domain)
-{
-  struct server *serv;
-
-  /* See if there is a suitable candidate, and unmark */
-  for (serv = daemon->servers; serv; serv = serv->next)
-    if ((serv->flags & SERV_FROM_DBUS) &&
-       (serv->flags & SERV_MARK))
-      {
-	if (domain)
-	  {
-	    if (!(serv->flags & SERV_HAS_DOMAIN) || !hostname_isequal(domain, serv->domain))
-	      continue;
-	  }
-	else
-	  {
-	    if (serv->flags & SERV_HAS_DOMAIN)
-	      continue;
-	  }
-	
-	serv->flags &= ~SERV_MARK;
-
-        break;
-      }
-  
-  if (!serv && (serv = whine_malloc(sizeof (struct server))))
-    {
-      /* Not found, create a new one. */
-      memset(serv, 0, sizeof(struct server));
-      
-      if (domain && !(serv->domain = whine_malloc(strlen(domain)+1)))
-	{
-	  free(serv);
-          serv = NULL;
-        }
-      else
-        {
-	  serv->next = daemon->servers;
-	  daemon->servers = serv;
-	  serv->flags = SERV_FROM_DBUS;
-	  if (domain)
-	    {
-	      strcpy(serv->domain, domain);
-	      serv->flags |= SERV_HAS_DOMAIN;
-	    }
-	}
-    }
-  
-  if (serv)
-    {
-      if (interface)
-	strcpy(serv->interface, interface);
-      else
-	serv->interface[0] = 0;
-            
-      if (source_addr->in.sin_family == AF_INET &&
-          addr->in.sin_addr.s_addr == 0 &&
-          serv->domain)
-        serv->flags |= SERV_NO_ADDR;
-      else
-        {
-          serv->flags &= ~SERV_NO_ADDR;
-          serv->addr = *addr;
-          serv->source_addr = *source_addr;
-        }
-    }
-}
-
-static void mark_dbus(void)
-{
-  struct server *serv;
-
-  /* mark everything from DBUS */
-  for (serv = daemon->servers; serv; serv = serv->next)
-    if (serv->flags & SERV_FROM_DBUS)
-      serv->flags |= SERV_MARK;
-}
-
-static void cleanup_dbus()
-{
-  struct server *serv, *tmp, **up;
-
-  /* unlink and free anything still marked. */
-  for (serv = daemon->servers, up = &daemon->servers; serv; serv = tmp) 
-    {
-      tmp = serv->next;
-      if (serv->flags & SERV_MARK)
-       {
-         server_gone(serv);
-         *up = serv->next;
-         if (serv->domain)
-	   free(serv->domain);
-	 free(serv);
-       }
-      else 
-       up = &serv->next;
-    }
-}
-
 static void dbus_read_servers(DBusMessage *message)
 {
   DBusMessageIter iter;
@@ -218,7 +116,7 @@ static void dbus_read_servers(DBusMessage *message)
   
   dbus_message_iter_init(message, &iter);
 
-  mark_dbus();
+  mark_servers(SERV_FROM_DBUS);
   
   while (1)
     {
@@ -289,13 +187,13 @@ static void dbus_read_servers(DBusMessage *message)
 	  domain = NULL;
 	
 	if (!skip)
-	  add_update_server(&addr, &source_addr, NULL, domain);
+	  add_update_server(SERV_FROM_DBUS, &addr, &source_addr, NULL, domain);
      
       } while (dbus_message_iter_get_arg_type(&iter) == DBUS_TYPE_STRING); 
     }
    
   /* unlink and free anything still marked. */
-  cleanup_dbus();
+  cleanup_servers();
 }
 
 static DBusMessage* dbus_read_servers_ex(DBusMessage *message, int strings)
@@ -319,7 +217,7 @@ static DBusMessage* dbus_read_servers_ex(DBusMessage *message, int strings)
                                     strings ? "Expected array of string" : "Expected array of string arrays");
      }
  
-  mark_dbus();
+  mark_servers(SERV_FROM_DBUS);
 
   /* array_iter points to each "as" element in the outer array */
   dbus_message_iter_recurse(&iter, &array_iter);
@@ -327,6 +225,7 @@ static DBusMessage* dbus_read_servers_ex(DBusMessage *message, int strings)
     {
       const char *str = NULL;
       union  mysockaddr addr, source_addr;
+      int flags = 0;
       char interface[IF_NAMESIZE];
       char *str_addr, *str_domain = NULL;
 
@@ -416,16 +315,19 @@ static DBusMessage* dbus_read_servers_ex(DBusMessage *message, int strings)
       memset(&interface, 0, sizeof(interface));
 
       /* parse the IP address */
-      addr_err = parse_server(str_addr, &addr, &source_addr, (char *) &interface, NULL);
+      if ((addr_err = parse_server(str_addr, &addr, &source_addr, (char *) &interface, &flags)))
-
+	{
-      if (addr_err)
-        {
           error = dbus_message_new_error_printf(message, DBUS_ERROR_INVALID_ARGS,
                                                 "Invalid IP address '%s': %s",
                                                 str, addr_err);
           break;
         }
       
+      /* 0.0.0.0 for server address == NULL, for Dbus */
+      if (addr.in.sin_family == AF_INET &&
+          addr.in.sin_addr.s_addr == 0)
+        flags |= SERV_NO_ADDR;
+      
       if (strings)
 	{
 	  char *p;
@@ -439,7 +341,7 @@ static DBusMessage* dbus_read_servers_ex(DBusMessage *message, int strings)
 	    else 
 	      p = NULL;
 	    
-	    add_update_server(&addr, &source_addr, interface, str_domain);
+	    add_update_server(flags | SERV_FROM_DBUS, &addr, &source_addr, interface, str_domain);
 	  } while ((str_domain = p));
 	}
       else
@@ -454,7 +356,7 @@ static DBusMessage* dbus_read_servers_ex(DBusMessage *message, int strings)
 	      dbus_message_iter_get_basic(&string_iter, &str);
 	    dbus_message_iter_next (&string_iter);
 	    
-	    add_update_server(&addr, &source_addr, interface, str);
+	    add_update_server(flags | SERV_FROM_DBUS, &addr, &source_addr, interface, str);
 	  } while (dbus_message_iter_get_arg_type(&string_iter) == DBUS_TYPE_STRING);
 	}
 	 
@@ -462,7 +364,7 @@ static DBusMessage* dbus_read_servers_ex(DBusMessage *message, int strings)
       dbus_message_iter_next(&array_iter);
     }
 
-  cleanup_dbus();
+  cleanup_servers();
 
   if (dup)
     free(dup);

src/dns-protocol.h

@@ -68,6 +68,7 @@
 #define T_RRSIG         46
 #define T_NSEC          47
 #define T_DNSKEY        48
+#define T_NSEC3         50
 #define	T_TKEY		249		
 #define	T_TSIG		250
 #define T_AXFR          252

src/dnsmasq.c

@@ -98,7 +98,10 @@ int main (int argc, char **argv)
   
 #ifdef HAVE_DNSSEC
   if (option_bool(OPT_DNSSEC_VALID))
-    daemon->keyname = safe_malloc(MAXDNAME);
+    {
+      daemon->keyname = safe_malloc(MAXDNAME);
+      daemon->workspacename = safe_malloc(MAXDNAME);
+    }
 #endif
 
 #ifdef HAVE_DHCP
@@ -144,7 +147,7 @@ int main (int argc, char **argv)
   if (option_bool(OPT_DNSSEC_VALID))
     {
 #ifdef HAVE_DNSSEC
-      if (!daemon->dnskeys)
+      if (!daemon->ds)
 	die(_("No trust anchors provided for DNSSEC"), NULL, EC_BADCONF);
       
       if (daemon->cachesize < CACHESIZ)
@@ -1114,7 +1117,7 @@ static void async_event(int pipe, time_t now)
 {
   pid_t p;
   struct event_desc ev;
-  int i;
+  int i, check = 0;
   char *msg;
   
   /* NOTE: the memory used to return msg is leaked: use msgs in events only
@@ -1125,11 +1128,25 @@ static void async_event(int pipe, time_t now)
       {
       case EVENT_RELOAD:
 	clear_cache_and_reload(now);
-	if (daemon->port != 0 && daemon->resolv_files && option_bool(OPT_NO_POLL))
+
+	if (daemon->port != 0)
 	  {
-	    reload_servers(daemon->resolv_files->name);
+	    if (daemon->resolv_files && option_bool(OPT_NO_POLL))
-	    check_servers();
+	      {
+		reload_servers(daemon->resolv_files->name);
+		check = 1;
+	      }
+
+	    if (daemon->servers_file)
+	      {
+		read_servers_file();
+		check = 1;
+	      }
+
+	    if (check)
+	      check_servers();
 	  }
+
 #ifdef HAVE_DHCP
 	rerun_scripts();
 #endif

src/dnsmasq.h

@@ -295,10 +295,10 @@ struct cname {
   struct cname *next;
 }; 
 
-struct dnskey {
+struct ds_config {
-  char *name, *key;
+  char *name, *digest;
-  int keylen, class, algo, flags;
+  int digestlen, class, algo, keytag, digest_type;
-  struct dnskey *next;
+  struct ds_config *next;
 };
 
 #define ADDRLIST_LITERAL 1
@@ -423,6 +423,7 @@ struct crec {
 #define F_DNSSEC    (1u<<22)
 #define F_KEYTAG    (1u<<23)
 #define F_SECSTAT   (1u<<24)
+#define F_NO_RR     (1u<<25)
 
 
 /* struct sockaddr is not large enough to hold any address,
@@ -455,6 +456,7 @@ union mysockaddr {
 #define SERV_COUNTED         512  /* workspace for log code */
 #define SERV_USE_RESOLV     1024  /* forward this domain in the normal way */
 #define SERV_NO_REBIND      2048  /* inhibit dns-rebind protection */
+#define SERV_FROM_FILE      4096  /* read from --servers-file */
 
 struct serverfd {
   int fd;
@@ -532,12 +534,14 @@ struct hostsfile {
 #define STAT_NEED_DS            4
 #define STAT_NEED_KEY           5
 #define STAT_TRUNCATED          6
+#define STAT_SECURE_WILDCARD    7
 
 #define FREC_NOREBIND           1
 #define FREC_CHECKING_DISABLED  2
 #define FREC_HAS_SUBNET         4
 #define FREC_DNSKEY_QUERY       8
 #define FREC_DS_QUERY          16
+#define FREC_AD_QUESTION       32
 
 #ifdef HAVE_DNSSEC
 #define HASH_SIZE 20 /* SHA-1 digest size */
@@ -559,7 +563,7 @@ struct frec {
   time_t time;
   unsigned char *hash[HASH_SIZE];
 #ifdef HAVE_DNSSEC 
-  int class;
+  int class, work_counter;
   struct blockdata *stash; /* Saved reply, whilst we validate */
   size_t stash_len;
   struct frec *dependent; /* Query awaiting internally-generated DNSKEY or DS query */
@@ -862,6 +866,7 @@ extern struct daemon {
   unsigned int options, options2;
   struct resolvc default_resolv, *resolv_files;
   time_t last_resolv;
+  char *servers_file;
   struct mx_srv_record *mxnames;
   struct naptr *naptr;
   struct txt_record *txt, *rr;
@@ -928,7 +933,7 @@ extern struct daemon {
   struct prefix_class *prefix_classes;
 #endif
 #ifdef HAVE_DNSSEC
-  struct dnskey *dnskeys;
+  struct ds_config *ds;
 #endif
 
   /* globally used stuff for DNS */
@@ -937,6 +942,7 @@ extern struct daemon {
   char *namebuff; /* MAXDNAME size buffer */
 #ifdef HAVE_DNSSEC
   char *keyname; /* MAXDNAME size buffer */
+  char *workspacename; /* ditto */
 #endif
   unsigned int local_answer, queries_forwarded, auth_answer;
   struct frec *frec_list;
@@ -990,12 +996,12 @@ extern struct daemon {
 void cache_init(void);
 void log_query(unsigned int flags, char *name, struct all_addr *addr, char *arg); 
 char *record_source(int index);
-void querystr(char *desc, char *str, unsigned short type);
+char *querystr(char *desc, unsigned short type);
 struct crec *cache_find_by_addr(struct crec *crecp,
 				struct all_addr *addr, time_t now, 
-				unsigned short prot);
+				unsigned int prot);
 struct crec *cache_find_by_name(struct crec *crecp, 
-				char *name, time_t now, unsigned short  prot);
+				char *name, time_t now, unsigned int prot);
 void cache_end_insert(void);
 void cache_start_insert(void);
 struct crec *cache_insert(char *name, struct all_addr *addr,
@@ -1031,6 +1037,7 @@ int extract_name(struct dns_header *header, size_t plen, unsigned char **pp,
                  char *name, int isExtract, int extrabytes);
 unsigned char *skip_name(unsigned char *ansp, struct dns_header *header, size_t plen, int extrabytes);
 unsigned char *skip_questions(struct dns_header *header, size_t plen);
+unsigned char *skip_section(unsigned char *ansp, int count, struct dns_header *header, size_t plen);
 unsigned int extract_request(struct dns_header *header, size_t qlen, 
 			       char *name, unsigned short *typep);
 size_t setup_reply(struct dns_header *header, size_t  qlen,
@@ -1040,7 +1047,8 @@ int extract_addresses(struct dns_header *header, size_t qlen, char *namebuff,
 		      time_t now, char **ipsets, int is_sign, int checkrebind,
 		      int no_cache, int secure, int *doctored);
 size_t answer_request(struct dns_header *header, char *limit, size_t qlen,  
-		   struct in_addr local_addr, struct in_addr local_netmask, time_t now);
+		      struct in_addr local_addr, struct in_addr local_netmask, 
+		      time_t now, int *ad_reqd);
 int check_for_bogus_wildcard(struct dns_header *header, size_t qlen, char *name, 
 			     struct bogus_addr *addr, time_t now);
 unsigned char *find_pseudoheader(struct dns_header *header, size_t plen,
@@ -1104,9 +1112,6 @@ void prettyprint_time(char *buf, unsigned int t);
 int prettyprint_addr(union mysockaddr *addr, char *buf);
 int parse_hex(char *in, unsigned char *out, int maxlen, 
 	      unsigned int *wildcard_mask, int *mac_type);
-#ifdef HAVE_DNSSEC
-int parse_base64(char *in, char *out);
-#endif
 int memcmp_masked(unsigned char *a, unsigned char *b, int len, 
 		  unsigned int mask);
 int expand_buf(struct iovec *iov, size_t size);
@@ -1130,6 +1135,7 @@ void read_opts (int argc, char **argv, char *compile_opts);
 char *option_string(int prot, unsigned int opt, unsigned char *val, 
 		    int opt_len, char *buf, int buf_len);
 void reread_dhcp(void);
+void read_servers_file(void);
 void set_option_bool(unsigned int opt);
 void reset_option_bool(unsigned int opt);
 struct hostsfile *expand_filelist(struct hostsfile *list);
@@ -1153,6 +1159,13 @@ int local_bind(int fd, union mysockaddr *addr, char *intname, int is_tcp);
 int random_sock(int family);
 void pre_allocate_sfds(void);
 int reload_servers(char *fname);
+void mark_servers(int flag);
+void cleanup_servers(void);
+void add_update_server(int flags,
+		       union mysockaddr *addr,
+		       union mysockaddr *source_addr,
+		       const char *interface,
+		       const char *domain);
 void check_servers(void);
 int enumerate_interfaces(int reset);
 void create_wildcard_listeners(void);

src/dnssec.c

@@ -21,6 +21,10 @@
 
 #include <nettle/rsa.h>
 #include <nettle/dsa.h>
+#ifndef NO_NETTLE_ECC
+#  include <nettle/ecdsa.h>
+#  include <nettle/ecc-curve.h>
+#endif
 #include <nettle/nettle-meta.h>
 #include <gmp.h>
 
@@ -208,9 +212,83 @@ static int dsa_verify(struct blockdata *key_data, unsigned int key_len, unsigned
   return nettle_dsa_sha1_verify_digest(key, digest, sig_struct);
 } 
  
-static int verify(struct blockdata *key_data, unsigned int key_len, unsigned char *sig, size_t sig_len,
+#ifndef NO_NETTLE_ECC
-		  unsigned char *digest, int algo)
+static int dnsmasq_ecdsa_verify(struct blockdata *key_data, unsigned int key_len, 
+				unsigned char *sig, size_t sig_len,
+				unsigned char *digest, size_t digest_len, int algo)
 {
+  unsigned char *p;
+  unsigned int t;
+  struct ecc_point *key;
+
+  static struct ecc_point *key_256 = NULL, *key_384 = NULL;
+  static mpz_t x, y;
+  static struct dsa_signature *sig_struct;
+  
+  if (!sig_struct)
+    {
+      if (!(sig_struct = whine_malloc(sizeof(struct dsa_signature))))
+	return 0;
+      
+      nettle_dsa_signature_init(sig_struct);
+      mpz_init(x);
+      mpz_init(y);
+    }
+  
+  switch (algo)
+    {
+    case 13:
+      if (!key_256)
+	{
+	  if (!(key_256 = whine_malloc(sizeof(struct ecc_point))))
+	    return 0;
+	  
+	  nettle_ecc_point_init(key_256, &nettle_secp_256r1);
+	}
+      
+      key = key_256;
+      t = 32;
+      break;
+      
+    case 14:
+      if (!key_384)
+	{
+	  if (!(key_384 = whine_malloc(sizeof(struct ecc_point))))
+	    return 0;
+	  
+	  nettle_ecc_point_init(key_384, &nettle_secp_384r1);
+	}
+      
+      key = key_384;
+      t = 48;
+      break;
+        
+    default:
+      return 0;
+    }
+  
+  if (sig_len != 2*t || key_len != 2*t ||
+      (p = blockdata_retrieve(key_data, key_len, NULL)))
+    return 0;
+  
+  mpz_import(x, t , 1, 1, 0, 0, p);
+  mpz_import(y, t , 1, 1, 0, 0, p + t);
+
+  if (!ecc_point_set(key, x, y))
+    return 0;
+  
+  mpz_import(sig_struct->r, t, 1, 1, 0, 0, sig);
+  mpz_import(sig_struct->s, t, 1, 1, 0, 0, sig + t);
+  
+  return nettle_ecdsa_verify(key, digest_len, digest, sig_struct);
+} 
+#endif 
+
+static int verify(struct blockdata *key_data, unsigned int key_len, unsigned char *sig, size_t sig_len,
+		  unsigned char *digest, size_t digest_len, int algo)
+{
+  (void)digest_len;
+
   switch (algo)
     {
     case 1: case 5: case 7: case 8: case 10:
@@ -218,6 +296,11 @@ static int verify(struct blockdata *key_data, unsigned int key_len, unsigned cha
       
     case 3: case 6: 
       return dsa_verify(key_data, key_len, sig, sig_len, digest, algo);
+ 
+#ifndef NO_NETTLE_ECC   
+    case 13: case 14:
+      return dnsmasq_ecdsa_verify(key_data, key_len, sig, sig_len, digest, digest_len, algo);
+#endif
     }
   
   return 0;
@@ -415,6 +498,8 @@ static void sort_rrset(struct dns_header *header, size_t plen, u16 *rr_desc, int
 	  int rdlen1, rdlen2, left1, left2, len1, len2, len, rc;
 	  u16 *dp1, *dp2;
 	  unsigned char *end1, *end2;
+	  /* Note that these have been determined to be OK previously,
+	     so we don't need to check for NULL return here. */
 	  unsigned char *p1 = skip_name(rrset[i], header, plen, 10);
 	  unsigned char *p2 = skip_name(rrset[i+1], header, plen, 10);
 	  
@@ -476,12 +561,15 @@ static void sort_rrset(struct dns_header *header, size_t plen, u16 *rr_desc, int
 /* Validate a single RRset (class, type, name) in the supplied DNS reply 
    Return code:
    STAT_SECURE   if it validates.
+   STAT_SECURE_WILDCARD if it validates and is the result of wildcard expansion.
    STAT_INSECURE can't validate (no RRSIG, bad packet).
    STAT_BOGUS    signature is wrong.
    STAT_NEED_KEY need DNSKEY to complete validation (name is returned in keyname)
 
    if key is non-NULL, use that key, which has the algo and tag given in the params of those names,
    otherwise find the key in the cache.
+
+   name is unchanged on exit. keyname is used as workspace and trashed.
 */
 static int validate_rrset(time_t now, struct dns_header *header, size_t plen, int class, 
 			  int type, char *name, char *keyname, struct blockdata *key, int keylen, int algo_in, int keytag_in)
@@ -505,7 +593,7 @@ static int validate_rrset(time_t now, struct dns_header *header, size_t plen, in
        j != 0; j--) 
     {
       unsigned char *pstart, *pdata;
-      int stype, sclass, ttl;
+      int stype, sclass;
 
       pstart = p;
       
@@ -514,7 +602,7 @@ static int validate_rrset(time_t now, struct dns_header *header, size_t plen, in
       
       GETSHORT(stype, p);
       GETSHORT(sclass, p);
-      GETLONG(ttl, p);
+      p += 4; /* TTL */
       
       pdata = p;
 
@@ -531,6 +619,11 @@ static int validate_rrset(time_t now, struct dns_header *header, size_t plen, in
 		{
 		  unsigned char **new;
 
+		  /* Protect against insane/maliciuos queries which bloat the workspace
+		     and eat CPU in the sort */
+		  if (rrsetidx >= 100)
+		    return STAT_INSECURE; 
+
 		  /* expand */
 		  if (!(new = whine_malloc((rrset_sz + 5) * sizeof(unsigned char **))))
 		    return STAT_INSECURE;
@@ -589,18 +682,18 @@ static int validate_rrset(time_t now, struct dns_header *header, size_t plen, in
     return STAT_INSECURE; 
   
   /* Sort RRset records into canonical order. 
-     Note that at this point keyname and name buffs are
+     Note that at this point keyname and daemon->workspacename buffs are
      unused, and used as workspace by the sort. */
-  sort_rrset(header, plen, rr_desc, rrsetidx, rrset, name, keyname);
+  sort_rrset(header, plen, rr_desc, rrsetidx, rrset, daemon->workspacename, keyname);
          
   /* Now try all the sigs to try and find one which validates */
   for (j = 0; j <sigidx; j++)
     {
-      unsigned char *psav, *sig;
+      unsigned char *psav, *sig, *digest;
       int i, wire_len, sig_len;
       const struct nettle_hash *hash;
       void *ctx;
-      unsigned char *digest;
+      char *name_start;
       u32 nsigttl;
       
       p = sigs[j];
@@ -618,6 +711,31 @@ static int validate_rrset(time_t now, struct dns_header *header, size_t plen, in
       if (!extract_name(header, plen, &p, keyname, 1, 0))
 	return STAT_INSECURE;
 
+      /* RFC 4035 5.3.1 says that the Signer's Name field MUST equal
+	 the name of the zone containing the RRset. We can't tell that
+	 for certain, but we can check that  the RRset name is equal to
+	 or encloses the signers name, which should be enough to stop 
+	 an attacker using signatures made with the key of an unrelated 
+	 zone he controls. Note that the root key is always allowed. */
+      if (*keyname != 0)
+	{
+	  int failed = 0;
+	  
+	  for (name_start = name; !hostname_isequal(name_start, keyname); )
+	    if ((name_start = strchr(name_start, '.')))
+	      name_start++; /* chop a label off and try again */
+	    else
+	      {
+		failed = 1;
+		break;
+	      }
+
+	  /* Bad sig, try another */
+	  if (failed)
+	    continue;
+	}
+      
+      /* Other 5.3.1 checks */
       if (!check_date_range(sig_inception, sig_expiration) ||
 	  labels > name_labels ||
 	  !(hash = hash_find(algo_digest_name(algo))) ||
@@ -642,13 +760,14 @@ static int validate_rrset(time_t now, struct dns_header *header, size_t plen, in
 	{
 	  int seg;
 	  unsigned char *end, *cp;
-	  char *name_start = name;
 	  u16 len, *dp;
 	  
 	  p = rrset[i];
 	  if (!extract_name(header, plen, &p, name, 1, 10)) 
 	    return STAT_INSECURE;
 
+	  name_start = name;
+	  
 	  /* if more labels than in RRsig name, hash *.<no labels in rrsig labels field>  4035 5.3.2 */
 	  if (labels < name_labels)
 	    {
@@ -698,7 +817,7 @@ static int validate_rrset(time_t now, struct dns_header *header, size_t plen, in
       if (key)
 	{
 	  if (algo_in == algo && keytag_in == key_tag &&
-	      verify(key, keylen, sig, sig_len, digest, algo))
+	      verify(key, keylen, sig, sig_len, digest, hash->digest_size, algo))
 	    return STAT_SECURE;
 	}
       else
@@ -708,8 +827,8 @@ static int validate_rrset(time_t now, struct dns_header *header, size_t plen, in
 	    if (crecp->addr.key.algo == algo && 
 		crecp->addr.key.keytag == key_tag &&
 		crecp->uid == class &&
-		verify(crecp->addr.key.keydata, crecp->addr.key.keylen, sig, sig_len, digest, algo))
+		verify(crecp->addr.key.keydata, crecp->addr.key.keylen, sig, sig_len, digest, hash->digest_size, algo))
-	      return STAT_SECURE;
+	      return (labels < name_labels) ? STAT_SECURE_WILDCARD : STAT_SECURE;
 	}
     }
 
@@ -741,7 +860,7 @@ int dnssec_validate_by_ds(time_t now, struct dns_header *header, size_t plen, ch
   GETSHORT(qclass, p);
   
   if (qtype != T_DNSKEY || qclass != class || ntohs(header->ancount) == 0)
-    return STAT_INSECURE;
+    return STAT_BOGUS;
 
    /* See if we have cached a DS record which validates this key */
   if (!(crecp = cache_find_by_name(NULL, name, now, F_DS)))
@@ -775,7 +894,7 @@ int dnssec_validate_by_ds(time_t now, struct dns_header *header, size_t plen, ch
       
       GETSHORT(flags, p);
       if (*p++ != 3)
-	return STAT_INSECURE;
+	return STAT_BOGUS;
       algo = *p++;
       keytag = dnskey_keytag(algo, flags, p, rdlen - 4);
       key = NULL;
@@ -865,7 +984,7 @@ int dnssec_validate_by_ds(time_t now, struct dns_header *header, size_t plen, ch
 		  
 		  GETSHORT(flags, p);
 		  if (*p++ != 3)
-		    return STAT_INSECURE;
+		    return STAT_BOGUS;
 		  algo = *p++;
 		  keytag = dnskey_keytag(algo, flags, p, rdlen - 4);
 		  
@@ -961,7 +1080,7 @@ int dnssec_validate_ds(time_t now, struct dns_header *header, size_t plen, char
   GETSHORT(qclass, p);
 
   if (qtype != T_DS || qclass != class || ntohs(header->ancount) == 0)
-    return STAT_INSECURE;
+    return STAT_BOGUS;
   
   val = dnssec_validate_reply(now, header, plen, name, keyname, NULL);
   
@@ -1039,14 +1158,454 @@ static int hostname_cmp(const char *a, const char *b)
     }
 }
 
+/* Find all the NSEC or NSEC3 records in a reply.
+   return an array of pointers to them. */
+static int find_nsec_records(struct dns_header *header, size_t plen, unsigned char ***nsecsetp, int *nsecsetl, int class_reqd)
+{
+  static unsigned char **nsecset = NULL;
+  static int nsecset_sz = 0;
+  
+  int type_found = -1;
+  unsigned char *p = skip_questions(header, plen);
+  int type, class, rdlen, i, nsecs_found;
+
+  /* Move to NS section */
+  if (!p || !(p = skip_section(p, ntohs(header->ancount), header, plen)))
+    return 0;
+  
+  for (nsecs_found = 0, i = ntohs(header->nscount); i != 0; i--)
+    {
+      unsigned char *pstart = p;
+      
+      if (!(p = skip_name(p, header, plen, 10)))
+	return 0;
+      
+      GETSHORT(type, p); 
+      GETSHORT(class, p);
+      p += 4; /* TTL */
+      GETSHORT(rdlen, p);
+
+      if (class == class_reqd && (type == T_NSEC || type == T_NSEC3))
+	{
+	  /* No mixed NSECing 'round here, thankyouverymuch */
+	  if (type_found == T_NSEC && type == T_NSEC3)
+	    return 0;
+	  if (type_found == T_NSEC3 && type == T_NSEC)
+	    return 0;
+
+	  type_found = type;
+
+	  if (nsecs_found == nsecset_sz)
+	    {
+	      unsigned char **new;
+	      
+	      /* Protect against insane/malicious queries which bloat the workspace
+		 and eat CPU in the sort */
+	      if (nsecs_found >= 100)
+		return 0; 
+
+	      /* expand */
+	      if (!(new = whine_malloc((nsecset_sz + 5) * sizeof(unsigned char **))))
+		return 0;
+		  
+	      if (nsecset)
+		{
+		  memcpy(new, nsecset, nsecset_sz * sizeof(unsigned char **));
+		  free(nsecset);
+		}
+	      
+	      nsecset = new;
+	      nsecset_sz += 5;
+	    }
+
+	  nsecset[nsecs_found++] = pstart;
+	}
+     
+      if (!ADD_RDLEN(header, p, plen, rdlen))
+	return 0;
+    }
+  
+  *nsecsetp = nsecset;
+  *nsecsetl = nsecs_found;
+  
+  return type_found;
+}
+
+static int prove_non_existance_nsec(struct dns_header *header, size_t plen, unsigned char **nsecs, int nsec_count,
+				    char *workspace1, char *workspace2, char *name, int type)
+{
+  int i, rc, rdlen;
+  unsigned char *p, *psave;
+  int offset = (type & 0xff) >> 3;
+  int mask = 0x80 >> (type & 0x07);
+  
+  /* Find NSEC record that proves name doesn't exist */
+  for (i = 0; i < nsec_count; i++)
+    {
+      p = nsecs[i];
+      if (!extract_name(header, plen, &p, workspace1, 1, 10))
+	return STAT_INSECURE;
+      p += 8; /* class, type, TTL */
+      GETSHORT(rdlen, p);
+      psave = p;
+      if (!extract_name(header, plen, &p, workspace2, 1, 10))
+	return STAT_INSECURE;
+      
+      rc = hostname_cmp(workspace1, name);
+      
+      if (rc == 0)
+	{
+	  /* 4035 para 5.4. Last sentence */
+	  if (type == T_NSEC || type == T_RRSIG)
+	    return STAT_SECURE;
+
+	  /* NSEC with the same name as the RR we're testing, check
+	     that the type in question doesn't appear in the type map */
+	  rdlen -= p - psave;
+	  /* rdlen is now length of type map, and p points to it */
+	  
+	  while (rdlen >= 2)
+	    {
+	      if (!CHECK_LEN(header, p, plen, rdlen))
+		return STAT_INSECURE;
+	      
+	      if (p[0] == type >> 8)
+		{
+		  /* Does the NSEC say our type exists? */
+		  if (offset < p[1] && (p[offset+2] & mask) != 0)
+		    return STAT_BOGUS;
+		  
+		  break; /* finshed checking */
+		}
+	      
+	      rdlen -= p[1];
+	      p +=  p[1];
+	    }
+	  
+	  return STAT_SECURE;
+	}
+      else if (rc == -1)
+	{
+	  /* Normal case, name falls between NSEC name and next domain name,
+	     wrap around case, name falls between NSEC name (rc == -1) and end */
+	  if (hostname_cmp(workspace2, name) == 1 || hostname_cmp(workspace1, workspace2) == 1)
+	    return STAT_SECURE;
+	}
+      else 
+	{
+	  /* wrap around case, name falls between start and next domain name */
+	  if (hostname_cmp(workspace1, workspace2) == 1 && hostname_cmp(workspace2, name) == 1)
+	    return STAT_SECURE;
+	}
+    }
+  
+  return STAT_BOGUS;
+}
+
+/* return digest length, or zero on error */
+static int hash_name(char *in, unsigned char **out, struct nettle_hash const *hash, 
+		     unsigned char *salt, int salt_len, int iterations)
+{
+  void *ctx;
+  unsigned char *digest;
+  int i;
+
+  if (!hash_init(hash, &ctx, &digest))
+    return 0;
+ 
+  hash->update(ctx, to_wire(in), (unsigned char *)in);
+  hash->update(ctx, salt_len, salt);
+  hash->digest(ctx, hash->digest_size, digest);
+
+  for(i = 0; i < iterations; i++)
+    {
+      hash->update(ctx, hash->digest_size, digest);
+      hash->update(ctx, salt_len, salt);
+      hash->digest(ctx, hash->digest_size, digest);
+    }
+   
+  from_wire(in);
+
+  *out = digest;
+  return hash->digest_size;
+}
+
+/* Decode base32 to first "." or end of string */
+static int base32_decode(char *in, unsigned char *out)
+{
+  int oc, on, c, mask, i;
+  unsigned char *p = out;
+ 
+  for (c = *in, oc = 0, on = 0; c != 0 && c != '.'; c = *++in) 
+    {
+      if (c >= '0' && c <= '9')
+	c -= '0';
+      else if (c >= 'a' && c <= 'v')
+	c -= 'a', c += 10;
+      else if (c >= 'A' && c <= 'V')
+	c -= 'A', c += 10;
+      else
+	return 0;
+      
+      for (mask = 0x10, i = 0; i < 5; i++)
+        {
+	  if (c & mask)
+	    oc |= 1;
+	  mask = mask >> 1;
+	  if (((++on) & 7) == 0)
+	    *p++ = oc;
+	  oc = oc << 1;
+	}
+    }
+  
+  if ((on & 7) != 0)
+    return 0;
+
+  return p - out;
+}
+
+static int prove_non_existance_nsec3(struct dns_header *header, size_t plen, unsigned char **nsecs, int nsec_count,
+				     char *workspace1, char *workspace2, char *name, int type)
+{
+  unsigned char *salt, *p, *digest;
+  int digest_len, i, iterations, salt_len, hash_len, base32_len, algo = 0;
+  struct nettle_hash const *hash;
+  char *closest_encloser, *next_closest, *wildcard;
+ 
+  /* Look though the NSEC3 records to find the first one with 
+     an algorithm we support (currently only algo == 1).
+
+     Take the algo, iterations, and salt of that record
+     as the ones we're going to use, and prune any 
+     that don't match. */
+  
+  for (i = 0; i < nsec_count; i++)
+    {
+      if (!(p = skip_name(nsecs[i], header, plen, 15)))
+	return STAT_INSECURE; /* bad packet */
+      
+      p += 10; /* type, class, TTL, rdlen */
+      algo = *p++;
+      
+      if (algo == 1)
+	break; /* known algo */
+    }
+
+  /* No usable NSEC3s */
+  if (i == nsec_count)
+    return STAT_BOGUS;
+
+  p++; /* flags */
+  GETSHORT (iterations, p);
+  salt_len = *p++;
+  salt = p;
+  if (!CHECK_LEN(header, salt, plen, salt_len))
+    return STAT_INSECURE; /* bad packet */
+    
+  /* Now prune so we only have NSEC3 records with same iterations, salt and algo */
+  for (i = 0; i < nsec_count; i++)
+    {
+      unsigned char *nsec3p = nsecs[i];
+      int this_iter;
+
+      nsecs[i] = NULL; /* Speculative, will be restored if OK. */
+      
+      if (!(p = skip_name(nsec3p, header, plen, 15)))
+	return STAT_INSECURE; /* bad packet */
+      
+      p += 10; /* type, class, TTL, rdlen */
+      
+      if (*p++ != algo)
+	continue;
+ 
+      p++; /* flags */
+      
+      GETSHORT(this_iter, p);
+      if (this_iter != iterations)
+	continue;
+
+      if (salt_len != *p++)
+	continue;
+      
+      if (!CHECK_LEN(header, p, plen, salt_len))
+	return STAT_INSECURE; /* bad packet */
+
+      if (memcmp(p, salt, salt_len) != 0)
+	continue;
+
+      /* All match, put the pointer back */
+      nsecs[i] = nsec3p;
+    }
+
+  /* Algo is checked as 1 above */
+  if (!(hash = hash_find("sha1")))
+    return STAT_INSECURE;
+
+  /* Now, we need the "closest encloser NSEC3" */
+  closest_encloser = name;
+  next_closest = NULL;
+
+  do
+    {
+      if (*closest_encloser == '.')
+	closest_encloser++;
+
+      if ((digest_len = hash_name(closest_encloser, &digest, hash, salt, salt_len, iterations)) == 0)
+	return STAT_INSECURE;
+      
+      for (i = 0; i < nsec_count; i++)
+	if ((p = nsecs[i]))
+	  {
+	    if (!extract_name(header, plen, &p, workspace1, 1, 0) ||
+		!(base32_len = base32_decode(workspace1, (unsigned char *)workspace2)))
+	      return STAT_INSECURE;
+	  
+	    if (digest_len == base32_len &&
+		memcmp(digest, workspace2, digest_len) == 0)
+	      break; /* Gotit */
+	  }
+      
+      if (i != nsec_count)
+	break;
+      
+      next_closest = closest_encloser;
+    }
+  while ((closest_encloser = strchr(closest_encloser, '.')));
+  
+  /* No usable NSEC3s */
+  if (i == nsec_count)
+    return STAT_BOGUS;
+  
+  if (!next_closest)
+    {
+      /* We found an NSEC3 whose hashed name exactly matches the query, so
+	 Now we just need to check the type map. p points to the RR data for the record. */
+      int rdlen;
+      unsigned char *psave;
+      int offset = (type & 0xff) >> 3;
+      int mask = 0x80 >> (type & 0x07);
+      
+      p += 8; /* class, type, TTL */
+      GETSHORT(rdlen, p);
+      psave = p;
+      p += 5 + salt_len; /* algo, flags, iterations, salt_len, salt */
+      hash_len = *p++;
+      if (!CHECK_LEN(header, p, plen, hash_len))
+	return STAT_INSECURE; /* bad packet */
+      p += hash_len;
+      rdlen -= p - psave;
+      
+      while (rdlen >= 2)
+	{
+	  if (!CHECK_LEN(header, p, plen, rdlen))
+	    return STAT_INSECURE;
+	  
+	  if (p[0] == type >> 8)
+	    {
+	      /* Does the NSEC3 say our type exists? */
+	      if (offset < p[1] && (p[offset+2] & mask) != 0)
+		return STAT_BOGUS;
+	      
+	      break; /* finshed checking */
+	    }
+	  
+	  rdlen -= p[1];
+	  p +=  p[1];
+	}
+      
+      return STAT_SECURE;
+    }
+
+  /* Look for NSEC3 that proves the non-existance of the next-closest encloser */
+  if ((digest_len = hash_name(next_closest, &digest, hash, salt, salt_len, iterations)) == 0)
+    return STAT_INSECURE;
+
+  for (i = 0; i < nsec_count; i++)
+    if ((p = nsecs[i]))
+      {
+       	if (!extract_name(header, plen, &p, workspace1, 1, 0) ||
+	    !(base32_len = base32_decode(workspace1, (unsigned char *)workspace2)))
+	  return STAT_INSECURE;
+	   
+	p += 15 + salt_len; /* class, type, TTL, rdlen, algo, flags, iterations, salt_len, salt */
+	hash_len = *p++; /* p now points to next hashed name */
+ 
+	if (!CHECK_LEN(header, p, plen, hash_len))
+	  return STAT_INSECURE;
+	
+	if (digest_len == base32_len && hash_len == base32_len)
+	  {
+	    if (memcmp(workspace2, digest, digest_len) <= 0)
+	      {
+		/* Normal case, hash falls between NSEC3 name-hash and next domain name-hash,
+		   wrap around case, name-hash falls between NSEC3 name-hash and end */
+		if (memcmp(p, digest, digest_len) > 0 || memcmp(workspace2, p, digest_len) > 0)
+		  return STAT_SECURE;
+	      }
+	    else 
+	      {
+		/* wrap around case, name falls between start and next domain name */
+		if (memcmp(workspace2, p, digest_len) > 0 && memcmp(p, digest, digest_len) > 0)
+		  return STAT_SECURE;
+	      }
+	  }
+      }
+  
+  /* Finally, check that there's no seat of wildcard synthesis */
+  if (!(wildcard = strchr(next_closest, '.')) || wildcard == next_closest)
+    return STAT_BOGUS;
+  
+  wildcard--;
+  *wildcard = '*';
+  
+  if ((digest_len = hash_name(wildcard, &digest, hash, salt, salt_len, iterations)) == 0)
+    return STAT_INSECURE;
+  
+  for (i = 0; i < nsec_count; i++)
+    if ((p = nsecs[i]))
+      {
+	if (!extract_name(header, plen, &p, workspace1, 1, 0) ||
+	    !(base32_len = base32_decode(workspace1, (unsigned char *)workspace2)))
+	  return STAT_INSECURE;
+	   
+	p += 15 + salt_len; /* class, type, TTL, rdlen, algo, flags, iterations, salt_len, salt */
+	hash_len = *p++; /* p now points to next hashed name */
+ 
+	if (!CHECK_LEN(header, p, plen, hash_len))
+	  return STAT_INSECURE;
+	
+	if (digest_len == base32_len && hash_len == base32_len)
+	  {
+	    if (memcmp(workspace2, digest, digest_len) <= 0)
+	      {
+		/* Normal case, hash falls between NSEC3 name-hash and next domain name-hash,
+		   wrap around case, name-hash falls between NSEC3 name-hash and end */
+		if (memcmp(p, digest, digest_len) > 0 || memcmp(workspace2, p, digest_len) > 0)
+		  return STAT_SECURE;
+	      }
+	    else 
+	      {
+		/* wrap around case, name falls between start and next domain name */
+		if (memcmp(workspace2, p, digest_len) > 0 && memcmp(p, digest, digest_len) > 0)
+		  return STAT_SECURE;
+	      }
+	  }
+      }
+  
+  return STAT_BOGUS;
+}
     
 /* Validate all the RRsets in the answer and authority sections of the reply (4035:3.2.3) */
 /* Returns are the same as validate_rrset, plus the class if the missing key is in *class */
 int dnssec_validate_reply(time_t now, struct dns_header *header, size_t plen, char *name, char *keyname, int *class)
 {
-  unsigned char *ans_start, *p1, *p2;
+  unsigned char *ans_start, *p1, *p2, **nsecs;
   int type1, class1, rdlen1, type2, class2, rdlen2;
-  int i, j, rc, have_nsec, have_nsec_equal, cname_count = 5;
+  int i, j, rc, nsec_count, cname_count = 10;
+  int nsec_type = 0;
+
+  if (RCODE(header) == SERVFAIL)
+    return STAT_BOGUS;
   
   if ((RCODE(header) != NXDOMAIN && RCODE(header) != NOERROR) || ntohs(header->qdcount) != 1)
     return STAT_INSECURE;
@@ -1094,7 +1653,33 @@ int dnssec_validate_reply(time_t now, struct dns_header *header, size_t plen, ch
 	      struct blockdata *key;
 	      struct crec *crecp;
 	      
-	      if ((rc = validate_rrset(now, header, plen, class1, type1, name, keyname, NULL, 0, 0, 0)) != STAT_SECURE)
+	      rc = validate_rrset(now, header, plen, class1, type1, name, keyname, NULL, 0, 0, 0);
+	      
+	      if (rc == STAT_SECURE_WILDCARD)
+		{
+		  /* An attacker replay a wildcard answer with a different
+		     answer and overlay a genuine RR. To prove this
+		     hasn't happened, the answer must prove that
+		     the gennuine record doesn't exist. Check that here. */
+		  if (!nsec_type)
+		    {
+		      nsec_type = find_nsec_records(header, plen, &nsecs, &nsec_count, class1);
+		      
+		      if (nsec_type == 0)
+			return STAT_INSECURE; /* Bad packet */
+		      if (nsec_type == -1)
+			return STAT_BOGUS; /* No NSECs */
+		    }
+		  
+		  if (nsec_type == T_NSEC)
+		    rc = prove_non_existance_nsec(header, plen, nsecs, nsec_count, daemon->workspacename, keyname, name, type1);
+		  else
+		    rc = prove_non_existance_nsec3(header, plen, nsecs, nsec_count, daemon->workspacename, keyname, name, type1);
+
+		  if (rc != STAT_SECURE)
+		    return rc;
+		} 
+	      else if (rc != STAT_SECURE)
 		{
 		  if (class)
 		    *class = class1; /* Class for DS or DNSKEY */
@@ -1163,6 +1748,7 @@ int dnssec_validate_reply(time_t now, struct dns_header *header, size_t plen, ch
 			       type_covered == T_DNSKEY || type_covered == T_PTR)) 
 			    {
 			      a.addr.dnssec.type = type_covered;
+			      a.addr.dnssec.class = class1;
 			      
 			      algo = *p2++;
 			      p2 += 13; /* labels, orig_ttl, expiration, inception */
@@ -1229,7 +1815,7 @@ int dnssec_validate_reply(time_t now, struct dns_header *header, size_t plen, ch
 	{
 	  /* Do we have an answer for the question? */
 	  if (type1 == type2)
-	    return RCODE(header) == NXDOMAIN ? STAT_INSECURE : STAT_SECURE;
+	    return RCODE(header) == NXDOMAIN ? STAT_BOGUS : STAT_SECURE;
 	  else if (type2 == T_CNAME)
 	    {
 	      /* looped CNAMES */
@@ -1246,117 +1832,25 @@ int dnssec_validate_reply(time_t now, struct dns_header *header, size_t plen, ch
 	return STAT_INSECURE;
     }
 
-  /* NXDOMAIN or NODATA reply, look for NSEC records to support that.
+  /* NXDOMAIN or NODATA reply, prove that (name, class1, type1) can't exist */
-     At this point, p1 points to the start of the auth section.
+  
-     Use keyname as workspace */
+  /* First marshall the NSEC records, if we've not done it previously */
-  for (have_nsec = 0, have_nsec_equal = 0, p2 = NULL, rdlen2 = 0, j = ntohs(header->nscount); j != 0; j--) 
+  if (!nsec_type)
     {
-      unsigned char *nsec_start = p1;
+      nsec_type = find_nsec_records(header, plen, &nsecs, &nsec_count, class1);
-      if (!extract_name(header, plen, &p1, keyname, 1, 10))
-	return STAT_INSECURE; /* bad packet */
       
-      GETSHORT(type2, p1); 
+      if (nsec_type == 0)
-      GETSHORT(class2, p1);
+	return STAT_INSECURE; /* Bad packet */
-      p1 += 4; /* TTL */
+      if (nsec_type == -1)
-      GETSHORT(rdlen1, p1);
+	return STAT_BOGUS; /* No NSECs */
-      
-      if (class1 == class2 && type2 == T_NSEC)
-	{
-	  have_nsec = 1;
-	  rc = hostname_cmp(name, keyname);
-
-	  if (rc >= 0)
-	    {
-	      if (p2)
-		{
-		  unsigned char *psave = p2;
-		  /* new NSEC is smaller than name,
-		     is it bigger than previous one? */
-		  
-		  /* get previous one into name buffer */
-		  if (!extract_name(header, plen, &psave, name, 1, 0))
-		    return STAT_INSECURE; /* bad packet */
-
-		  if (hostname_cmp(name, keyname) < 0)
-		    {
-		      p2 = nsec_start;
-		      rdlen2 = rdlen1;
-		    }
-		  
-		  /* restore query name */
-		  psave = (unsigned char *)(header+1);
-		  if (!extract_name(header, plen, &psave, name, 1, 0))
-		    return STAT_INSECURE;
-		}
-	      else
-		{
-		  /* There was no previous best candidate */
-		  p2 = nsec_start;
-		  rdlen2 = rdlen1; 
-		}
-	    }
-	  
-	  if (rc == 0)
-	    have_nsec_equal = 1;
-	}
-	  
-      if (!ADD_RDLEN(header, p1, plen, rdlen1))
-	return STAT_INSECURE;
     }
   
-  
+  if (nsec_type == T_NSEC)
-  if (p2)
+    return prove_non_existance_nsec(header, plen, nsecs, nsec_count, daemon->workspacename, keyname, name, type1);
-    {
+  else
-      unsigned char *psave;
+    return prove_non_existance_nsec3(header, plen, nsecs, nsec_count, daemon->workspacename, keyname, name, type1);
-      p2 = skip_name(p2, header, plen, 0);
-      p2 += 10; /* type, class, ttl, rdlen */
-      psave = p2;
-      extract_name(header, plen, &p2, keyname, 1, 0);
-      rdlen2 -= p2 - psave;
-    }
-
-  /* At this point, have_nsec is set if there's at least one NSEC
-     have_nsec_equal is set if there's an NSEC with the same name as the query;
-     p2 points to the type bit maps of the biggest NSEC smaller than or equal to the query
-     or NULL if the query is smaller than all of them.
-     Keyname holds the next domain name for that NSEC.
-     rdlen2 is the length of the bitmap field */
-  
-
-  if (RCODE(header) == NOERROR && have_nsec_equal)
-    {
-      int offset = (type1 & 0xff) >> 3;
-      int mask = 0x80 >> (type1 & 0x07);
-      
-      while (rdlen2 >= 2)
-	{
-	  if (p2[0] == type1 >> 8)
-	    {
-	      /* Does the NSEC say our type exists? */
-	      if (offset < p2[1] &&
-		  (p2[offset+2] & mask) != 0)
-		return STAT_INSECURE;
-	      
-	      break; /* finshed checking */
-	    }
-	  
-	  rdlen2 -= p2[1];
-	  p2 +=  p2[1];
-	}
-      
-      return STAT_SECURE;
-    }
-
-  if (RCODE(header) == NXDOMAIN && have_nsec)
-    {
-      if (!p2 || hostname_cmp(name, keyname) < 0)
-	return STAT_SECURE; /* Before the first, or in a proven gap */
-    }
-
-  return STAT_INSECURE;
 }
 
-
 /* Compute keytag (checksum to quickly index a key). See RFC4034 */
 int dnskey_keytag(int alg, int flags, unsigned char *key, int keylen)
 {
@@ -1368,12 +1862,12 @@ int dnskey_keytag(int alg, int flags, unsigned char *key, int keylen)
     }
   else
     {
-      unsigned long ac;
+      unsigned long ac = flags + 0x300 + alg;
       int i;
 
-      ac = ((htons(flags) >> 8) | ((htons(flags) << 8) & 0xff00)) + 0x300 + alg;
       for (i = 0; i < keylen; ++i)
         ac += (i & 1) ? key[i] : key[i] << 8;
+
       ac += (ac >> 16) & 0xffff;
       return ac & 0xffff;
     }
@@ -1382,9 +1876,7 @@ int dnskey_keytag(int alg, int flags, unsigned char *key, int keylen)
 size_t dnssec_generate_query(struct dns_header *header, char *end, char *name, int class, int type, union mysockaddr *addr)
 {
   unsigned char *p;
-  char types[20];
+  char *types = querystr("dnssec-query", type);
-  
-  querystr("dnssec-query", types, type);
 
   if (addr->sa.sa_family == AF_INET) 
     log_query(F_DNSSEC | F_IPV4, name, (struct all_addr *)&addr->in.sin_addr, types);

src/forward.c

@@ -234,7 +234,8 @@ static unsigned int search_servers(time_t now, struct all_addr **addrpp,
 
 static int forward_query(int udpfd, union mysockaddr *udpaddr,
 			 struct all_addr *dst_addr, unsigned int dst_iface,
-			 struct dns_header *header, size_t plen, time_t now, struct frec *forward)
+			 struct dns_header *header, size_t plen, time_t now, 
+			 struct frec *forward, int ad_reqd)
 {
   char *domain = NULL;
   int type = 0, norebind = 0;
@@ -249,16 +250,13 @@ static int forward_query(int udpfd, union mysockaddr *udpaddr,
 #endif
  unsigned int gotname = extract_request(header, plen, daemon->namebuff, NULL);
 
-  /* RFC 4035: sect 4.6 para 2 */
-  header->hb4 &= ~HB4_AD;
-  
   /* may be no servers available. */
   if (!daemon->servers)
     forward = NULL;
   else if (forward || (hash && (forward = lookup_frec_by_sender(ntohs(header->id), udpaddr, hash))))
     {
 #ifdef HAVE_DNSSEC
-      /* If we've already got an answer to this query, but we're awaiting keys for vaildation,
+      /* If we've already got an answer to this query, but we're awaiting keys for validation,
 	 there's no point retrying the query, retry the key query instead...... */
       if (forward->blocking_query)
 	{
@@ -334,6 +332,11 @@ static int forward_query(int udpfd, union mysockaddr *udpaddr,
 	    forward->flags |= FREC_NOREBIND;
 	  if (header->hb4 & HB4_CD)
 	    forward->flags |= FREC_CHECKING_DISABLED;
+	  if (ad_reqd)
+	    forward->flags |= FREC_AD_QUESTION;
+#ifdef HAVE_DNSSEC
+	  forward->work_counter = DNSSEC_WORK;
+#endif
 
 	  header->id = htons(forward->new_id);
 	  
@@ -503,13 +506,15 @@ static int forward_query(int udpfd, union mysockaddr *udpaddr,
 }
 
 static size_t process_reply(struct dns_header *header, time_t now, struct server *server, size_t n, int check_rebind, 
-			    int no_cache, int cache_secure, int check_subnet, union mysockaddr *query_source)
+			    int no_cache, int cache_secure, int ad_reqd, int check_subnet, union mysockaddr *query_source)
 {
   unsigned char *pheader, *sizep;
   char **sets = 0;
   int munged = 0, is_sign;
   size_t plen; 
 
+  (void)ad_reqd;
+
 #ifdef HAVE_IPSET
   /* Similar algorithm to search_servers. */
   struct ipsets *ipset_pos;
@@ -535,15 +540,13 @@ static size_t process_reply(struct dns_header *header, time_t now, struct server
 
   if ((pheader = find_pseudoheader(header, n, &plen, &sizep, &is_sign)))
     {
-      if (!is_sign)
+      unsigned short udpsz;
-	{
+      unsigned char *psave = sizep;
-	  unsigned short udpsz;
-	  unsigned char *psave = sizep;
       
-	  GETSHORT(udpsz, sizep);
+      GETSHORT(udpsz, sizep);
-	  if (udpsz > daemon->edns_pktsz)
+
-	    PUTSHORT(daemon->edns_pktsz, psave);
+      if (!is_sign && udpsz > daemon->edns_pktsz)
-	}
+	PUTSHORT(daemon->edns_pktsz, psave);
       
       if (check_subnet && !check_source(header, plen, pheader, query_source))
 	{
@@ -619,7 +622,7 @@ static size_t process_reply(struct dns_header *header, time_t now, struct server
   if (option_bool(OPT_DNSSEC_VALID))
     header->hb4 &= ~HB4_AD;
   
-  if (!(header->hb4 & HB4_CD) && cache_secure)
+  if (!(header->hb4 & HB4_CD) && ad_reqd && cache_secure)
     header->hb4 |= HB4_AD;
 #endif
 
@@ -705,7 +708,7 @@ void reply_query(int fd, int family, time_t now)
 	  if ((nn = resize_packet(header, (size_t)n, pheader, plen)))
 	    {
 	      header->hb3 &= ~(HB3_QR | HB3_TC);
-	      forward_query(-1, NULL, NULL, 0, header, nn, now, forward);
+	      forward_query(-1, NULL, NULL, 0, header, nn, now, forward, 0);
 	      return;
 	    }
 	}
@@ -780,10 +783,26 @@ void reply_query(int fd, int family, time_t now)
 	     answer aside, whilst we get that. */     
 	  if (status == STAT_NEED_DS || status == STAT_NEED_KEY)
 	    {
-	      struct frec *new;
+	      struct frec *new, *orig;
 	      
-	      if ((new = get_new_frec(now, NULL, 1)))
+	      /* Free any saved query */
+	      if (forward->stash)
+		blockdata_free(forward->stash);
+	      
+	      /* Now save reply pending receipt of key data */
+	      if (!(forward->stash = blockdata_alloc((char *)header, n)))
+		return;
+	      forward->stash_len = n;
+	      
+	    anotherkey:	      
+	      /* Find the original query that started it all.... */
+	      for (orig = forward; orig->dependent; orig = orig->dependent);
+
+	      if (--orig->work_counter == 0 || !(new = get_new_frec(now, NULL, 1)))
+		status = STAT_INSECURE;
+	      else
 		{
+		  int fd;
 		  struct frec *next = new->next;
 		  *new = *forward; /* copy everything, then overwrite */
 		  new->next = next;
@@ -794,80 +813,67 @@ void reply_query(int fd, int family, time_t now)
 #endif
 		  new->flags &= ~(FREC_DNSKEY_QUERY | FREC_DS_QUERY);
 		  
-		  /* Free any saved query */
+		  new->dependent = forward; /* to find query awaiting new one. */
-		  if (forward->stash)
+		  forward->blocking_query = new; /* for garbage cleaning */
-		    blockdata_free(forward->stash);
+		  /* validate routines leave name of required record in daemon->keyname */
-		  
+		  if (status == STAT_NEED_KEY)
-		  /* Now save reply pending receipt of key data */
+		    {
-		  if (!(forward->stash = blockdata_alloc((char *)header, n)))
+		      new->flags |= FREC_DNSKEY_QUERY; 
-		    free_frec(new); /* malloc failure, unwind */
+		      nn = dnssec_generate_query(header, ((char *) header) + daemon->packet_buff_sz,
+						 daemon->keyname, forward->class, T_DNSKEY, &server->addr);
+		    }
 		  else 
 		    {
-		      int fd;
+		      new->flags |= FREC_DS_QUERY;
+		      nn = dnssec_generate_query(header,((char *) header) + daemon->packet_buff_sz,
+						 daemon->keyname, forward->class, T_DS, &server->addr);
+		    }
+		  if ((hash = hash_questions(header, nn, daemon->namebuff)))
+		    memcpy(new->hash, hash, HASH_SIZE);
+		  new->new_id = get_id();
+		  header->id = htons(new->new_id);
+		  /* Save query for retransmission */
+		  new->stash = blockdata_alloc((char *)header, nn);
+		  new->stash_len = nn;
 		  
-		      forward->stash_len = n;
+		  /* Don't resend this. */
+		  daemon->srv_save = NULL;
 		  
-		      new->dependent = forward; /* to find query awaiting new one. */
+		  if (server->sfd)
-		      forward->blocking_query = new; /* for garbage cleaning */
+		    fd = server->sfd->fd;
-		      /* validate routines leave name of required record in daemon->keyname */
+		  else
-		      if (status == STAT_NEED_KEY)
+		    {
-			{
+		      fd = -1;
-			  new->flags |= FREC_DNSKEY_QUERY; 
-			  nn = dnssec_generate_query(header, ((char *) header) + daemon->packet_buff_sz,
-						     daemon->keyname, forward->class, T_DNSKEY, &server->addr);
-			}
-		      else 
-			{
-			  new->flags |= FREC_DS_QUERY;
-			  nn = dnssec_generate_query(header,((char *) header) + daemon->packet_buff_sz,
-						     daemon->keyname, forward->class, T_DS, &server->addr);
-			}
-		      if ((hash = hash_questions(header, nn, daemon->namebuff)))
-			memcpy(new->hash, hash, HASH_SIZE);
-		      new->new_id = get_id();
-		      header->id = htons(new->new_id);
-		      /* Save query for retransmission */
-		      new->stash = blockdata_alloc((char *)header, nn);
-		      new->stash_len = nn;
-
-		      /* Don't resend this. */
-		      daemon->srv_save = NULL;
-	
-		      if (server->sfd)
-			fd = server->sfd->fd;
-		      else
-			{
-			  fd = -1;
 #ifdef HAVE_IPV6
-			  if (server->addr.sa.sa_family == AF_INET6)
+		      if (server->addr.sa.sa_family == AF_INET6)
-			    {
-			      if (new->rfd6 || (new->rfd6 = allocate_rfd(AF_INET6)))
-				fd = new->rfd6->fd;
-			    }
-			  else
-#endif
-			    {
-			      if (new->rfd4 || (new->rfd4 = allocate_rfd(AF_INET)))
-				fd = new->rfd4->fd;
-			    }
-			}
-		      
-		      if (fd != -1)
 			{
-			  while (sendto(fd, (char *)header, nn, 0, &server->addr.sa, sa_len(&server->addr)) == -1 && retry_send()); 
+			  if (new->rfd6 || (new->rfd6 = allocate_rfd(AF_INET6)))
-			  server->queries++;
+			    fd = new->rfd6->fd;
+			}
+		      else
+#endif
+			{
+			  if (new->rfd4 || (new->rfd4 = allocate_rfd(AF_INET)))
+			    fd = new->rfd4->fd;
 			}
 		    }
-		}
 		  
-	      return;
+		  if (fd != -1)
+		    {
+		      while (sendto(fd, (char *)header, nn, 0, &server->addr.sa, sa_len(&server->addr)) == -1 && retry_send()); 
+		      server->queries++;
+		    }
+		  
+		  return;
+		}
 	    }
 	  
 	  /* Ok, we reached far enough up the chain-of-trust that we can validate something.
 	     Now wind back down, pulling back answers which wouldn't previously validate
-	     and validate them with the new data. Failure to find needed data here is an internal error.
+	     and validate them with the new data. Note that if an answer needs multiple
-	     Once we get to the original answer (FREC_DNSSEC_QUERY not set) and it validates,
+	     keys to validate, we may find another key is needed, in which case we set off
-	     return it to the original requestor. */
+	     down another branch of the tree. Once we get to the original answer 
+	     (FREC_DNSSEC_QUERY not set) and it validates, return it to the original requestor. */
 	  while (forward->dependent)
 	    {
 	      struct frec *prev = forward->dependent;
@@ -887,18 +893,23 @@ void reply_query(int fd, int family, time_t now)
 		    status = dnssec_validate_reply(now, header, n, daemon->namebuff, daemon->keyname, &forward->class);	
 		  
 		  if (status == STAT_NEED_DS || status == STAT_NEED_KEY)
-		    {
+		    goto anotherkey;
-		      my_syslog(LOG_ERR, _("Unexpected missing data for DNSSEC validation"));
-		      status = STAT_INSECURE;
-		    }
 		}
 	    }
 	  
 	  if (status == STAT_TRUNCATED)
 	    header->hb3 |= HB3_TC;
 	  else
-	    log_query(F_KEYTAG | F_SECSTAT, "result", NULL, 
+	    {
-		      status == STAT_SECURE ? "SECURE" : (status == STAT_INSECURE ? "INSECURE" : "BOGUS"));
+	      char *result;
+	      
+	      if (forward->work_counter == 0)
+		result = "ABANDONED";
+	      else
+		result = (status == STAT_SECURE ? "SECURE" : (status == STAT_INSECURE ? "INSECURE" : "BOGUS"));
+	      
+	      log_query(F_KEYTAG | F_SECSTAT, "result", NULL, result);
+	    }
 	  
 	  no_cache_dnssec = 0;
 	  
@@ -906,17 +917,17 @@ void reply_query(int fd, int family, time_t now)
 	    cache_secure = 1;
 	  else if (status == STAT_BOGUS)
 	    no_cache_dnssec = 1;
-	  
-	  /* restore CD bit to the value in the query */
-	  if (forward->flags & FREC_CHECKING_DISABLED)
-	    header->hb4 |= HB4_CD;
-	  else
-	    header->hb4 &= ~HB4_CD;
 	}
 #endif     
       
+      /* restore CD bit to the value in the query */
+      if (forward->flags & FREC_CHECKING_DISABLED)
+	header->hb4 |= HB4_CD;
+      else
+	header->hb4 &= ~HB4_CD;
+      
       if ((nn = process_reply(header, now, server, (size_t)n, check_rebind, no_cache_dnssec, cache_secure,
-			      forward->flags & FREC_HAS_SUBNET, &forward->source)))
+			      forward->flags & FREC_AD_QUESTION, forward->flags & FREC_HAS_SUBNET, &forward->source)))
 	{
 	  header->id = htons(forward->orig_id);
 	  header->hb4 |= HB4_RA; /* recursion if available */
@@ -1117,12 +1128,10 @@ void receive_query(struct listener *listen, time_t now)
   
   if (extract_request(header, (size_t)n, daemon->namebuff, &type))
     {
-      char types[20];
 #ifdef HAVE_AUTH
       struct auth_zone *zone;
 #endif
-
+      char *types = querystr(auth_dns ? "auth" : "query", type);
-      querystr(auth_dns ? "auth" : "query", types, type);
       
       if (listen->family == AF_INET) 
 	log_query(F_QUERY | F_IPV4 | F_FORWARD, daemon->namebuff, 
@@ -1160,8 +1169,9 @@ void receive_query(struct listener *listen, time_t now)
   else
 #endif
     {
+      int ad_reqd;
       m = answer_request(header, ((char *) header) + daemon->packet_buff_sz, (size_t)n, 
-			 dst_addr_4, netmask, now);
+			 dst_addr_4, netmask, now, &ad_reqd);
       
       if (m >= 1)
 	{
@@ -1170,7 +1180,7 @@ void receive_query(struct listener *listen, time_t now)
 	  daemon->local_answer++;
 	}
       else if (forward_query(listen->fd, &source_addr, &dst_addr, if_index,
-			     header, (size_t)n, now, NULL))
+			     header, (size_t)n, now, NULL, ad_reqd))
 	daemon->queries_forwarded++;
       else
 	daemon->local_answer++;
@@ -1178,59 +1188,72 @@ void receive_query(struct listener *listen, time_t now)
 }
 
 #ifdef HAVE_DNSSEC
-static int tcp_key_recurse(time_t now, int status, int class, char *keyname, struct server *server)
+static int tcp_key_recurse(time_t now, int status, struct dns_header *header, size_t n, 
+			   int class, char *name, char *keyname, struct server *server, int *keycount)
 {
   /* Recurse up the key heirarchy */
-  size_t n; 
-  unsigned char *packet = whine_malloc(65536 + MAXDNAME + RRFIXEDSZ + sizeof(u16));
-  unsigned char *payload = &packet[2];
-  struct dns_header *header = (struct dns_header *)payload;
-  u16 *length = (u16 *)packet;
   int new_status;
-  unsigned char c1, c2;
 
-  n = dnssec_generate_query(header, ((char *) header) + 65536, keyname, class, 
+  /* limit the amount of work we do, to avoid cycling forever on loops in the DNS */
-			    status == STAT_NEED_KEY ? T_DNSKEY : T_DS, &server->addr);
+  if (--(*keycount) == 0)
+    return STAT_INSECURE;
   
-  *length = htons(n);
+  if (status == STAT_NEED_KEY)
-  
+    new_status = dnssec_validate_by_ds(now, header, n, name, keyname, class);
-  if (!read_write(server->tcpfd, packet, n + sizeof(u16), 0) ||
+  else if (status == STAT_NEED_DS)
-      !read_write(server->tcpfd, &c1, 1, 1) ||
+    new_status = dnssec_validate_ds(now, header, n, name, keyname, class);
-      !read_write(server->tcpfd, &c2, 1, 1) ||
-      !read_write(server->tcpfd, payload, (c1 << 8) | c2, 1))
-    {
-      close(server->tcpfd);
-      server->tcpfd = -1;
-      new_status = STAT_INSECURE;
-    }
   else
+    new_status = dnssec_validate_reply(now, header, n, name, keyname, &class);
+  
+  /* Can't validate because we need a key/DS whose name now in keyname.
+     Make query for same, and recurse to validate */
+  if (new_status == STAT_NEED_DS || new_status == STAT_NEED_KEY)
     {
-      n = (c1 << 8) | c2;
+      size_t m; 
+      unsigned char *packet = whine_malloc(65536 + MAXDNAME + RRFIXEDSZ + sizeof(u16));
+      unsigned char *payload = &packet[2];
+      struct dns_header *new_header = (struct dns_header *)payload;
+      u16 *length = (u16 *)packet;
+      unsigned char c1, c2;
        
-      if (status == STAT_NEED_KEY)
+      if (!packet)
-	new_status = dnssec_validate_by_ds(now, header, n, daemon->namebuff, daemon->keyname, class);
+	return STAT_INSECURE;
+
+    another_tcp_key:
+      m = dnssec_generate_query(new_header, ((char *) new_header) + 65536, keyname, class, 
+				new_status == STAT_NEED_KEY ? T_DNSKEY : T_DS, &server->addr);
+      
+      *length = htons(m);
+      
+      if (!read_write(server->tcpfd, packet, m + sizeof(u16), 0) ||
+	  !read_write(server->tcpfd, &c1, 1, 1) ||
+	  !read_write(server->tcpfd, &c2, 1, 1) ||
+	  !read_write(server->tcpfd, payload, (c1 << 8) | c2, 1))
+	new_status = STAT_INSECURE;
       else
-	new_status = dnssec_validate_ds(now, header, n, daemon->namebuff, daemon->keyname, class);
-      
-      if (new_status == STAT_NEED_DS || new_status == STAT_NEED_KEY)
 	{
-	  if ((new_status = tcp_key_recurse(now, new_status, class, daemon->keyname, server) == STAT_SECURE))
+	  m = (c1 << 8) | c2;
+	  
+	  if (tcp_key_recurse(now, new_status, new_header, m, class, name, keyname, server, keycount) == STAT_SECURE)
 	    {
-	      if (status ==  STAT_NEED_KEY)
+	      /* Reached a validated record, now try again at this level.
-		new_status = dnssec_validate_by_ds(now, header, n, daemon->namebuff, daemon->keyname, class);
+		 Note that we may get ANOTHER NEED_* if an answer needs more than one key.
+		 If so, go round again. */
+	      
+	      if (status == STAT_NEED_KEY)
+		new_status = dnssec_validate_by_ds(now, header, n, name, keyname, class);
+	      else if (status == STAT_NEED_DS)
+		new_status = dnssec_validate_ds(now, header, n, name, keyname, class);
 	      else
-		new_status = dnssec_validate_ds(now, header, n, daemon->namebuff, daemon->keyname, class);
+		new_status = dnssec_validate_reply(now, header, n, name, keyname, &class);
 
 	      if (new_status == STAT_NEED_DS || new_status == STAT_NEED_KEY)
-		{
+		goto another_tcp_key;
-		  my_syslog(LOG_ERR, _("Unexpected missing data for DNSSEC validation"));
-		  status = STAT_INSECURE;
-		}
 	    }
 	}
-    }
 
-  free(packet);
+      free(packet);
+    }
   
   return new_status;
 }
@@ -1249,7 +1272,7 @@ unsigned char *tcp_request(int confd, time_t now,
 #ifdef HAVE_AUTH
   int local_auth = 0;
 #endif
-  int checking_disabled, check_subnet, no_cache_dnssec = 0, cache_secure = 0;
+  int checking_disabled, ad_question, check_subnet, no_cache_dnssec = 0, cache_secure = 0;
   size_t m;
   unsigned short qtype;
   unsigned int gotname;
@@ -1285,16 +1308,12 @@ unsigned char *tcp_request(int confd, time_t now,
       if ((checking_disabled = header->hb4 & HB4_CD))
 	no_cache_dnssec = 1;
        
-      /* RFC 4035: sect 4.6 para 2 */
-      header->hb4 &= ~HB4_AD;
-      
       if ((gotname = extract_request(header, (unsigned int)size, daemon->namebuff, &qtype)))
 	{
-	  char types[20];
 #ifdef HAVE_AUTH
 	  struct auth_zone *zone;
 #endif
-	  querystr(auth_dns ? "auth" : "query", types, qtype);
+	  char *types = querystr(auth_dns ? "auth" : "query", qtype);
 	  
 	  if (peer_addr.sa.sa_family == AF_INET) 
 	    log_query(F_QUERY | F_IPV4 | F_FORWARD, daemon->namebuff, 
@@ -1331,7 +1350,7 @@ unsigned char *tcp_request(int confd, time_t now,
 	{
 	  /* m > 0 if answered from cache */
 	  m = answer_request(header, ((char *) header) + 65536, (size_t)size, 
-			     dst_addr_4, netmask, now);
+			     dst_addr_4, netmask, now, &ad_question);
 	  
 	  /* Do this by steam now we're not in the select() loop */
 	  check_log_writer(NULL); 
@@ -1412,7 +1431,10 @@ unsigned char *tcp_request(int confd, time_t now,
 			  if (option_bool(OPT_DNSSEC_VALID))
 			    {
 			      size = add_do_bit(header, size, ((char *) header) + 65536);
-			      header->hb4 |= HB4_CD;
+			      /* For debugging, set Checking Disabled, otherwise, have the upstream check too,
+				 this allows it to select auth servers when one is returning bad data. */
+			      if (option_bool(OPT_DNSSEC_DEBUG))
+				header->hb4 |= HB4_CD;
 			    }
 #endif
 			  
@@ -1463,18 +1485,16 @@ unsigned char *tcp_request(int confd, time_t now,
 #ifdef HAVE_DNSSEC
 		      if (option_bool(OPT_DNSSEC_VALID) && !checking_disabled)
 			{
-			  int class, status;
+			  int keycount = DNSSEC_WORK; /* Limit to number of DNSSEC questions, to catch loops and avoid filling cache. */
+			  int status = tcp_key_recurse(now, STAT_TRUNCATED, header, m, 0, daemon->namebuff, daemon->keyname, last_server, &keycount);
+			  char *result;
 
-			  status = dnssec_validate_reply(now, header, m, daemon->namebuff, daemon->keyname, &class);
+			  if (keycount == 0)
+			    result = "ABANDONED";
+			  else
+			    result = (status == STAT_SECURE ? "SECURE" : (status == STAT_INSECURE ? "INSECURE" : "BOGUS"));
 			  
-			  if (status == STAT_NEED_DS || status == STAT_NEED_KEY)
+			  log_query(F_KEYTAG | F_SECSTAT, "result", NULL, result);
-			    {
-			      if ((status = tcp_key_recurse(now, status, class, daemon->keyname, last_server)) == STAT_SECURE)
-				status = dnssec_validate_reply(now, header, m, daemon->namebuff, daemon->keyname, &class);
-			    }
-
-			  log_query(F_KEYTAG | F_SECSTAT, "result", NULL, 
-				    status == STAT_SECURE ? "SECURE" : (status == STAT_INSECURE ? "INSECURE" : "BOGUS"));
 			  
 			  if (status == STAT_BOGUS)
 			    no_cache_dnssec = 1;
@@ -1513,7 +1533,7 @@ unsigned char *tcp_request(int confd, time_t now,
 
 		      m = process_reply(header, now, last_server, (unsigned int)m, 
 					option_bool(OPT_NO_REBIND) && !norebind, no_cache_dnssec,
-					cache_secure, check_subnet, &peer_addr); 
+					cache_secure, ad_question, check_subnet, &peer_addr); 
 		      
 		      break;
 		    }

src/network.c

@@ -1254,87 +1254,187 @@ void pre_allocate_sfds(void)
       }  
 }
 
+void mark_servers(int flag)
+{
+  struct server *serv;
+
+  /* mark everything with argument flag */
+  for (serv = daemon->servers; serv; serv = serv->next)
+    if (serv->flags & flag)
+      serv->flags |= SERV_MARK;
+}
+
+void cleanup_servers(void)
+{
+  struct server *serv, *tmp, **up;
+
+  /* unlink and free anything still marked. */
+  for (serv = daemon->servers, up = &daemon->servers; serv; serv = tmp) 
+    {
+      tmp = serv->next;
+      if (serv->flags & SERV_MARK)
+       {
+         server_gone(serv);
+         *up = serv->next;
+         if (serv->domain)
+	   free(serv->domain);
+	 free(serv);
+       }
+      else 
+       up = &serv->next;
+    }
+}
+
+void add_update_server(int flags,
+		       union mysockaddr *addr,
+		       union mysockaddr *source_addr,
+		       const char *interface,
+		       const char *domain)
+{
+  struct server *serv, *next = NULL;
+  char *domain_str = NULL;
+  
+  /* See if there is a suitable candidate, and unmark */
+  for (serv = daemon->servers; serv; serv = serv->next)
+    if (serv->flags & SERV_MARK)
+      {
+	if (domain)
+	  {
+	    if (!(serv->flags & SERV_HAS_DOMAIN) || !hostname_isequal(domain, serv->domain))
+	      continue;
+	  }
+	else
+	  {
+	    if (serv->flags & SERV_HAS_DOMAIN)
+	      continue;
+	  }
+	
+        break;
+      }
+
+  if (serv)
+    {
+      domain_str = serv->domain;
+      next = serv->next;
+    }
+  else if ((serv = whine_malloc(sizeof (struct server))))
+    {
+      /* Not found, create a new one. */
+      if (domain && !(domain_str = whine_malloc(strlen(domain)+1)))
+	{
+	  free(serv);
+          serv = NULL;
+        }
+      else
+        {
+	  struct server *s;
+	  /* Add to the end of the chain, for order */
+	  if (!daemon->servers)
+	    daemon->servers = serv;
+	  else
+	    {
+	      for (s = daemon->servers; s->next; s = s->next);
+	      s->next = serv;
+	    }
+	  if (domain)
+	    strcpy(domain_str, domain);
+	}
+    }
+  
+  if (serv)
+    {
+      memset(serv, 0, sizeof(struct server));
+      serv->flags = flags;
+      serv->domain = domain_str;
+      serv->next = next;
+      serv->queries = serv->failed_queries = 0;
+      
+      if (domain)
+	serv->flags |= SERV_HAS_DOMAIN;
+      
+      if (interface)
+	strcpy(serv->interface, interface);      
+      if (addr)
+	serv->addr = *addr;
+      if (source_addr)
+	serv->source_addr = *source_addr;
+    }
+}
 
 void check_servers(void)
 {
   struct irec *iface;
-  struct server *new, *tmp, *ret = NULL;
+  struct server *serv;
   int port = 0;
 
   /* interface may be new since startup */
   if (!option_bool(OPT_NOWILD))
     enumerate_interfaces(0);
   
-  for (new = daemon->servers; new; new = tmp)
+  for (serv = daemon->servers; serv; serv = serv->next)
     {
-      tmp = new->next;
+       if (!(serv->flags & (SERV_LITERAL_ADDRESS | SERV_NO_ADDR | SERV_USE_RESOLV | SERV_NO_REBIND)))
-      
-      if (!(new->flags & (SERV_LITERAL_ADDRESS | SERV_NO_ADDR | SERV_USE_RESOLV | SERV_NO_REBIND)))
 	{
-	  port = prettyprint_addr(&new->addr, daemon->namebuff);
+	  port = prettyprint_addr(&serv->addr, daemon->namebuff);
 
 	  /* 0.0.0.0 is nothing, the stack treats it like 127.0.0.1 */
-	  if (new->addr.sa.sa_family == AF_INET &&
+	  if (serv->addr.sa.sa_family == AF_INET &&
-	      new->addr.in.sin_addr.s_addr == 0)
+	      serv->addr.in.sin_addr.s_addr == 0)
 	    {
-	      free(new);
+	      serv->flags |= SERV_MARK;
 	      continue;
 	    }
 
 	  for (iface = daemon->interfaces; iface; iface = iface->next)
-	    if (sockaddr_isequal(&new->addr, &iface->addr))
+	    if (sockaddr_isequal(&serv->addr, &iface->addr))
 	      break;
 	  if (iface)
 	    {
 	      my_syslog(LOG_WARNING, _("ignoring nameserver %s - local interface"), daemon->namebuff);
-	      free(new);
+	      serv->flags |= SERV_MARK;
 	      continue;
 	    }
 	  
 	  /* Do we need a socket set? */
-	  if (!new->sfd && 
+	  if (!serv->sfd && 
-	      !(new->sfd = allocate_sfd(&new->source_addr, new->interface)) &&
+	      !(serv->sfd = allocate_sfd(&serv->source_addr, serv->interface)) &&
 	      errno != 0)
 	    {
 	      my_syslog(LOG_WARNING, 
 			_("ignoring nameserver %s - cannot make/bind socket: %s"),
 			daemon->namebuff, strerror(errno));
-	      free(new);
+	      serv->flags |= SERV_MARK;
 	      continue;
 	    }
 	}
       
-      /* reverse order - gets it right. */
+      if (!(serv->flags & SERV_NO_REBIND))
-      new->next = ret;
-      ret = new;
-      
-      if (!(new->flags & SERV_NO_REBIND))
 	{
-	  if (new->flags & (SERV_HAS_DOMAIN | SERV_FOR_NODOTS | SERV_USE_RESOLV))
+	  if (serv->flags & (SERV_HAS_DOMAIN | SERV_FOR_NODOTS | SERV_USE_RESOLV))
 	    {
 	      char *s1, *s2;
-	      if (!(new->flags & SERV_HAS_DOMAIN))
+	      if (!(serv->flags & SERV_HAS_DOMAIN))
 		s1 = _("unqualified"), s2 = _("names");
-	      else if (strlen(new->domain) == 0)
+	      else if (strlen(serv->domain) == 0)
 		s1 = _("default"), s2 = "";
 	      else
-		s1 = _("domain"), s2 = new->domain;
+		s1 = _("domain"), s2 = serv->domain;
 	      
-	      if (new->flags & SERV_NO_ADDR)
+	      if (serv->flags & SERV_NO_ADDR)
 		my_syslog(LOG_INFO, _("using local addresses only for %s %s"), s1, s2);
-	      else if (new->flags & SERV_USE_RESOLV)
+	      else if (serv->flags & SERV_USE_RESOLV)
 		my_syslog(LOG_INFO, _("using standard nameservers for %s %s"), s1, s2);
-	      else if (!(new->flags & SERV_LITERAL_ADDRESS))
+	      else if (!(serv->flags & SERV_LITERAL_ADDRESS))
 		my_syslog(LOG_INFO, _("using nameserver %s#%d for %s %s"), daemon->namebuff, port, s1, s2);
 	    }
-	  else if (new->interface[0] != 0)
+	  else if (serv->interface[0] != 0)
-	    my_syslog(LOG_INFO, _("using nameserver %s#%d(via %s)"), daemon->namebuff, port, new->interface); 
+	    my_syslog(LOG_INFO, _("using nameserver %s#%d(via %s)"), daemon->namebuff, port, serv->interface); 
 	  else
 	    my_syslog(LOG_INFO, _("using nameserver %s#%d"), daemon->namebuff, port); 
 	}
     }
 
-  daemon->servers = ret;
+  cleanup_servers();
 }
 
 /* Return zero if no servers found, in that case we keep polling.
@@ -1343,9 +1443,6 @@ int reload_servers(char *fname)
 {
   FILE *f;
   char *line;
-  struct server *old_servers = NULL;
-  struct server *new_servers = NULL;
-  struct server *serv;
   int gotone = 0;
 
   /* buff happens to be MAXDNAME long... */
@@ -1355,26 +1452,7 @@ int reload_servers(char *fname)
       return 0;
     }
    
-  /* move old servers to free list - we can reuse the memory 
+  mark_servers(SERV_FROM_RESOLV);
-     and not risk malloc if there are the same or fewer new servers. 
-     Servers which were specced on the command line go to the new list. */
-  for (serv = daemon->servers; serv;)
-    {
-      struct server *tmp = serv->next;
-      if (serv->flags & SERV_FROM_RESOLV)
-	{
-	  serv->next = old_servers;
-	  old_servers = serv; 
-	  /* forward table rules reference servers, so have to blow them away */
-	  server_gone(serv);
-	}
-      else
-	{
-	  serv->next = new_servers;
-	  new_servers = serv;
-	}
-      serv = tmp;
-    }
     
   while ((line = fgets(daemon->namebuff, MAXDNAME, f)))
     {
@@ -1434,38 +1512,12 @@ int reload_servers(char *fname)
 	continue;
 #endif 
 
-      if (old_servers)
+      add_update_server(SERV_FROM_RESOLV, &addr, &source_addr, NULL, NULL);
-	{
-	  serv = old_servers;
-	  old_servers = old_servers->next;
-	}
-      else if (!(serv = whine_malloc(sizeof (struct server))))
-	continue;
-      
-      /* this list is reverse ordered: 
-	 it gets reversed again in check_servers */
-      serv->next = new_servers;
-      new_servers = serv;
-      serv->addr = addr;
-      serv->source_addr = source_addr;
-      serv->domain = NULL;
-      serv->interface[0] = 0;
-      serv->sfd = NULL;
-      serv->flags = SERV_FROM_RESOLV;
-      serv->queries = serv->failed_queries = 0;
       gotone = 1;
     }
   
-  /* Free any memory not used. */
-  while (old_servers)
-    {
-      struct server *tmp = old_servers->next;
-      free(old_servers);
-      old_servers = tmp;
-    }
-
-  daemon->servers = new_servers;
   fclose(f);
+  cleanup_servers();
 
   return gotone;
 }

src/option.c

@@ -139,8 +139,10 @@ struct myoption {
 #define LOPT_QUIET_DHCP6  327
 #define LOPT_QUIET_RA     328
 #define LOPT_SEC_VALID    329
-#define LOPT_DNSKEY       330
+#define LOPT_TRUST_ANCHOR 330
 #define LOPT_DNSSEC_DEBUG 331
+#define LOPT_REV_SERV     332
+#define LOPT_SERVERS_FILE 333
 
 #ifdef HAVE_GETOPT_LONG
 static const struct option opts[] =  
@@ -157,6 +159,7 @@ static const struct myoption opts[] =
     { "user", 2, 0, 'u' },
     { "group", 2, 0, 'g' },
     { "resolv-file", 2, 0, 'r' },
+    { "servers-file", 1, 0, LOPT_SERVERS_FILE },
     { "mx-host", 1, 0, 'm' },
     { "mx-target", 1, 0, 't' },
     { "cache-size", 2, 0, 'c' },
@@ -178,6 +181,7 @@ static const struct myoption opts[] =
     { "pid-file", 2, 0, 'x' },
     { "strict-order", 0, 0, 'o' },
     { "server", 1, 0, 'S' },
+    { "rev-server", 1, 0, LOPT_REV_SERV },
     { "local", 1, 0, LOPT_LOCAL },
     { "address", 1, 0, 'A' },
     { "conf-file", 2, 0, 'C' },
@@ -277,7 +281,7 @@ static const struct myoption opts[] =
     { "ipset", 1, 0, LOPT_IPSET },
     { "synth-domain", 1, 0, LOPT_SYNTH },
     { "dnssec", 0, 0, LOPT_SEC_VALID },
-    { "dnskey", 1, 0, LOPT_DNSKEY },
+    { "trust-anchor", 1, 0, LOPT_TRUST_ANCHOR },
     { "dnssec-debug", 0, 0, LOPT_DNSSEC_DEBUG },
 #ifdef OPTION6_PREFIX_CLASS 
     { "dhcp-prefix-class", 1, 0, LOPT_PREF_CLSS },
@@ -347,7 +351,9 @@ static struct {
   { 'Q', ARG_ONE, "<integer>", gettext_noop("Force the originating port for upstream DNS queries."), NULL },
   { 'R', OPT_NO_RESOLV, NULL, gettext_noop("Do NOT read resolv.conf."), NULL },
   { 'r', ARG_DUP, "<path>", gettext_noop("Specify path to resolv.conf (defaults to %s)."), RESOLVFILE }, 
+  { LOPT_SERVERS_FILE, ARG_ONE, "<path>", gettext_noop("Specify path to file with server= options"), NULL },
   { 'S', ARG_DUP, "/<domain>/<ipaddr>", gettext_noop("Specify address(es) of upstream servers with optional domains."), NULL },
+  { LOPT_REV_SERV, ARG_DUP, "<addr>/<prefix>,<ipaddr>", gettext_noop("Specify address of upstream servers for reverse address queries"), NULL },
   { LOPT_LOCAL, ARG_DUP, "/<domain>/", gettext_noop("Never forward queries to specified domains."), NULL },
   { 's', ARG_DUP, "<domain>[,<range>]", gettext_noop("Specify the domain to be assigned in DHCP leases."), NULL },
   { 't', ARG_ONE, "<host_name>", gettext_noop("Specify default target in an MX record."), NULL },
@@ -430,7 +436,7 @@ static struct {
   { LOPT_IPSET, ARG_DUP, "/<domain>/<ipset>[,<ipset>...]", gettext_noop("Specify ipsets to which matching domains should be added"), NULL },
   { LOPT_SYNTH, ARG_DUP, "<domain>,<range>,[<prefix>]", gettext_noop("Specify a domain and address range for synthesised names"), NULL },
   { LOPT_SEC_VALID, OPT_DNSSEC_VALID, NULL, gettext_noop("Activate DNSSEC validation"), NULL },
-  { LOPT_DNSKEY, ARG_DUP, "<domain>,<algo>,<key>", gettext_noop("Specify trust anchor DNSKEY"), NULL },
+  { LOPT_TRUST_ANCHOR, ARG_DUP, "<domain>,[<class>],...", gettext_noop("Specify trust anchor key digest."), NULL },
   { LOPT_DNSSEC_DEBUG, OPT_DNSSEC_DEBUG, NULL, gettext_noop("Disable upstream checking for DNSSEC debugging."), NULL },
 #ifdef OPTION6_PREFIX_CLASS 
   { LOPT_PREF_CLSS, ARG_DUP, "set:tag,<class>", gettext_noop("Specify DHCPv6 prefix class"), NULL },
@@ -591,6 +597,18 @@ static int atoi_check16(char *a, int *res)
   return 1;
 }
 
+#ifdef HAVE_DNSSEC
+static int atoi_check8(char *a, int *res)
+{
+  if (!(atoi_check(a, res)) ||
+      *res < 0 ||
+      *res > 0xff)
+    return 0;
+
+  return 1;
+}
+#endif
+	
 static void add_txt(char *name, char *txt)
 {
   size_t len = strlen(txt);
@@ -674,6 +692,13 @@ char *parse_server(char *arg, union mysockaddr *addr, union mysockaddr *source_a
   char *scope_id;
 #endif
   
+  if (!arg || strlen(arg) == 0)
+    {
+      *flags |= SERV_NO_ADDR;
+      *interface = 0;
+      return NULL;
+    }
+
   if ((source = split_chr(arg, '@')) && /* is there a source. */
       (portno = split_chr(source, '#')) &&
       !atoi_check16(portno, &source_port))
@@ -752,6 +777,54 @@ char *parse_server(char *arg, union mysockaddr *addr, union mysockaddr *source_a
   return NULL;
 }
 
+static struct server *add_rev4(struct in_addr addr, int msize)
+{
+  struct server *serv = opt_malloc(sizeof(struct server));
+  in_addr_t  a = ntohl(addr.s_addr) >> 8;
+  char *p;
+
+  memset(serv, 0, sizeof(struct server));
+  p = serv->domain = opt_malloc(25); /* strlen("xxx.yyy.zzz.in-addr.arpa")+1 */
+  
+  if (msize == 24)
+    p += sprintf(p, "%d.", a & 0xff);
+  a = a >> 8;
+  if (msize != 8)
+    p += sprintf(p, "%d.", a & 0xff);
+  a = a >> 8;
+  p += sprintf(p, "%d.in-addr.arpa", a & 0xff);
+  
+  serv->flags = SERV_HAS_DOMAIN;
+  serv->next = daemon->servers;
+  daemon->servers = serv;
+
+  return serv;
+
+}
+
+static struct server *add_rev6(struct in6_addr *addr, int msize)
+{
+  struct server *serv = opt_malloc(sizeof(struct server));
+  char *p;
+  int i;
+				  
+  memset(serv, 0, sizeof(struct server));
+  p = serv->domain = opt_malloc(73); /* strlen("32*<n.>ip6.arpa")+1 */
+  
+  for (i = msize-1; i >= 0; i -= 4)
+    { 
+      int dig = ((unsigned char *)addr)[i>>3];
+      p += sprintf(p, "%.1x.", (i>>2) & 1 ? dig & 15 : dig >> 4);
+    }
+  p += sprintf(p, "ip6.arpa");
+  
+  serv->flags = SERV_HAS_DOMAIN;
+  serv->next = daemon->servers;
+  daemon->servers = serv;
+  
+  return serv;
+}
+
 #ifdef HAVE_DHCP
 
 static int is_tag_prefix(char *arg)
@@ -1315,7 +1388,7 @@ void reset_option_bool(unsigned int opt)
     daemon->options2 &= ~(1u << (opt - 32));
 }
 
-static int one_opt(int option, char *arg, char *errstr, char *gen_err, int command_line)
+static int one_opt(int option, char *arg, char *errstr, char *gen_err, int command_line, int servers_only)
 {      
   int i;
   char *comma;
@@ -1519,6 +1592,10 @@ static int one_opt(int option, char *arg, char *errstr, char *gen_err, int comma
 	break;
       }
 
+    case LOPT_SERVERS_FILE:
+      daemon->servers_file = opt_string_alloc(arg);
+      break;
+      
     case 'm':  /* --mx-host */
       {
 	int pref = 1;
@@ -1814,34 +1891,11 @@ static int one_opt(int option, char *arg, char *errstr, char *gen_err, int comma
 				ret_err(gen_err);
 			      else
 				{
-				  /* generate the equivalent of
+				   /* generate the equivalent of
-				     local=/<domain>/
+				      local=/<domain>/
-				     local=/xxx.yyy.zzz.in-addr.arpa/ */
+				      local=/xxx.yyy.zzz.in-addr.arpa/ */
-				  struct server *serv = opt_malloc(sizeof(struct server));
+				  struct server *serv = add_rev4(new->start, msize);
-				  in_addr_t a = ntohl(new->start.s_addr) >> 8;
+				  serv->flags |= SERV_NO_ADDR;
-				  char *p;
-				  
-				  memset(serv, 0, sizeof(struct server));
-				  serv->domain = d;
-				  serv->flags = SERV_HAS_DOMAIN | SERV_NO_ADDR;
-				  serv->next = daemon->servers;
-				  daemon->servers = serv;
-				  
-				  serv = opt_malloc(sizeof(struct server));
-				  memset(serv, 0, sizeof(struct server));
-				  p = serv->domain = opt_malloc(25); /* strlen("xxx.yyy.zzz.in-addr.arpa")+1 */
-				  
-				  if (msize == 24)
-				    p += sprintf(p, "%d.", a & 0xff);
-				  a = a >> 8;
-				  if (msize != 8)
-				    p += sprintf(p, "%d.", a & 0xff);
-				  a = a >> 8;
-				  p += sprintf(p, "%d.in-addr.arpa", a & 0xff);
-				  
-				  serv->flags = SERV_HAS_DOMAIN | SERV_NO_ADDR;
-				  serv->next = daemon->servers;
-				  daemon->servers = serv;
 				}
 			    }
 			}
@@ -1877,29 +1931,8 @@ static int one_opt(int option, char *arg, char *errstr, char *gen_err, int comma
 				  /* generate the equivalent of
 				     local=/<domain>/
 				     local=/xxx.yyy.zzz.ip6.arpa/ */
-				  struct server *serv = opt_malloc(sizeof(struct server));
+				  struct server *serv = add_rev6(&new->start6, msize);
-				  char *p;
+				  serv->flags |= SERV_NO_ADDR;
-				  
-				  memset(serv, 0, sizeof(struct server));
-				  serv->domain = d;
-				  serv->flags = SERV_HAS_DOMAIN | SERV_NO_ADDR;
-				  serv->next = daemon->servers;
-				  daemon->servers = serv;
-
-				  serv = opt_malloc(sizeof(struct server));
-				  memset(serv, 0, sizeof(struct server));
-				  p = serv->domain = opt_malloc(73); /* strlen("32*<n.>ip6.arpa")+1 */
-				  
-				  for (i = msize-1; i >= 0; i -= 4)
-				    { 
-				      int dig = ((unsigned char *)&new->start6)[i>>3];
-				      p += sprintf(p, "%.1x.", (i>>2) & 1 ? dig & 15 : dig >> 4);
-				    }
-				  p += sprintf(p, "ip6.arpa");
-				  
-				  serv->flags = SERV_HAS_DOMAIN | SERV_NO_ADDR;
-				  serv->next = daemon->servers;
-				  daemon->servers = serv;
 				}
 			    }
 			}
@@ -2126,6 +2159,9 @@ static int one_opt(int option, char *arg, char *errstr, char *gen_err, int comma
 	    memset(newlist, 0, sizeof(struct server));
 	  }
 	
+	if (servers_only && option == 'S')
+	  newlist->flags |= SERV_FROM_FILE;
+	
 	if (option == 'A')
 	  {
 	    newlist->flags |= SERV_LITERAL_ADDRESS;
@@ -2170,6 +2206,40 @@ static int one_opt(int option, char *arg, char *errstr, char *gen_err, int comma
 	break;
       }
 
+    case LOPT_REV_SERV: /* --rev-server */
+      {
+	char *string;
+	int size;
+	struct server *serv;
+	struct in_addr addr4;
+#ifdef HAVE_IPV6
+	struct in6_addr addr6;
+#endif
+ 
+	unhide_metas(arg);
+	if (!arg || !(comma=split(arg)) || !(string = split_chr(arg, '/')) || !atoi_check(string, &size))
+	  ret_err(gen_err);
+
+	if (inet_pton(AF_INET, arg, &addr4))
+	  serv = add_rev4(addr4, size);
+#ifdef HAVE_IPV6
+	else if (inet_pton(AF_INET6, arg, &addr6))
+	  serv = add_rev6(&addr6, size);
+#endif
+	else
+	  ret_err(gen_err);
+ 
+	string = parse_server(comma, &serv->addr, &serv->source_addr, serv->interface, &serv->flags);
+	
+	if (string)
+	  ret_err(string);
+	
+	if (servers_only)
+	  serv->flags |= SERV_FROM_FILE;
+	
+	break;
+      }
+
     case LOPT_IPSET: /* --ipset */
 #ifndef HAVE_IPSET
       ret_err(_("recompile with HAVE_IPSET defined to enable ipset directives"));
@@ -3675,10 +3745,11 @@ static int one_opt(int option, char *arg, char *errstr, char *gen_err, int comma
       }
 
 #ifdef HAVE_DNSSEC
-    case LOPT_DNSKEY:
+    case LOPT_TRUST_ANCHOR:
       {
-	struct dnskey *new = opt_malloc(sizeof(struct dnskey));
+	struct ds_config *new = opt_malloc(sizeof(struct ds_config));
-      	char *key64, *algo = NULL;
+      	char *cp, *cp1, *keyhex, *digest, *algo = NULL;
+	int len;
 	
 	new->class = C_IN;
 
@@ -3700,19 +3771,29 @@ static int one_opt(int option, char *arg, char *errstr, char *gen_err, int comma
 	      }
 	  }
 		  
-       	if (!comma || !algo || !(key64 = split(algo)) ||
+       	if (!comma || !algo || !(digest = split(algo)) || !(keyhex = split(digest)) ||
-	    !atoi_check16(comma, &new->flags) || !atoi_check16(algo, &new->algo) ||
+	    !atoi_check16(comma, &new->keytag) || 
+	    !atoi_check8(algo, &new->algo) ||
+	    !atoi_check8(digest, &new->digest_type) ||
 	    !(new->name = canonicalise_opt(arg)))
-	  ret_err(_("bad DNSKEY"));
+	  ret_err(_("bad trust anchor"));
 	    
 	/* Upper bound on length */
-	new->key = opt_malloc((3*strlen(key64)/4)+1);
+	len = (2*strlen(keyhex))+1;
-	unhide_metas(key64);
+	new->digest = opt_malloc(len);
-	if ((new->keylen = parse_base64(key64, new->key)) == -1)
+	unhide_metas(keyhex);
-	  ret_err(_("bad base64 in DNSKEY"));
+	/* 4034: "Whitespace is allowed within digits" */
+	for (cp = keyhex; *cp; )
+	  if (isspace(*cp))
+	    for (cp1 = cp; *cp1; cp1++)
+	      *cp1 = *(cp1+1);
+	  else
+	    cp++;
+	if ((new->digestlen = parse_hex(keyhex, (unsigned char *)new->digest, len, NULL, NULL)) == -1)
+	  ret_err(_("bad HEX in trust anchor"));
 	
-	new->next = daemon->dnskeys;
+	new->next = daemon->ds;
-	daemon->dnskeys = new;
+	daemon->ds = new;
 	
 	break;
       }
@@ -3733,12 +3814,13 @@ static void read_file(char *file, FILE *f, int hard_opt)
   
   while (fgets(buff, MAXDNAME, f))
     {
-      int white, i, option = hard_opt;
+      int white, i;
+      volatile int option = (hard_opt == LOPT_REV_SERV) ? 0 : hard_opt;
       char *errmess, *p, *arg = NULL, *start;
       size_t len;
 
       /* Memory allocation failure longjmps here if mem_recover == 1 */ 
-      if (option != 0)
+      if (option != 0 || hard_opt == LOPT_REV_SERV)
 	{
 	  if (setjmp(mem_jmp))
 	    continue;
@@ -3839,13 +3921,15 @@ static void read_file(char *file, FILE *f, int hard_opt)
 	    errmess = _("extraneous parameter");
 	  else if (opts[i].has_arg == 1 && !arg)
 	    errmess = _("missing parameter");
+	  else if (hard_opt == LOPT_REV_SERV && option != 'S' && option != LOPT_REV_SERV)
+	    errmess = _("illegal option");
 	}
 
     oops:
       if (errmess)
 	strcpy(daemon->namebuff, errmess);
 	  
-      if (errmess || !one_opt(option, arg, buff, _("error"), 0))
+      if (errmess || !one_opt(option, arg, buff, _("error"), 0, hard_opt == LOPT_REV_SERV))
 	{
 	  sprintf(daemon->namebuff + strlen(daemon->namebuff), _(" at line %d of %s"), lineno, file);
 	  if (hard_opt != 0)
@@ -4027,6 +4111,22 @@ struct hostsfile *expand_filelist(struct hostsfile *list)
   return list;
 }
 
+void read_servers_file(void)
+{
+  FILE *f;
+
+  if (!(f = fopen(daemon->servers_file, "r")))
+    {
+       my_syslog(LOG_ERR, _("cannot read %s: %s"), daemon->servers_file, strerror(errno));
+       return;
+    }
+  
+  mark_servers(SERV_FROM_FILE);
+  cleanup_servers();
+  
+  read_file(daemon->servers_file, f, LOPT_REV_SERV);
+}
+ 
 
 #ifdef HAVE_DHCP
 void reread_dhcp(void)
@@ -4220,9 +4320,9 @@ void read_opts(int argc, char **argv, char *compile_opts)
       else
 	{
 #ifdef HAVE_GETOPT_LONG
-	  if (!one_opt(option, arg, daemon->namebuff, _("try --help"), 1))
+	  if (!one_opt(option, arg, daemon->namebuff, _("try --help"), 1, 0))
 #else 
-	  if (!one_opt(option, arg, daemon->namebuff, _("try -w"), 1)) 
+	    if (!one_opt(option, arg, daemon->namebuff, _("try -w"), 1, 0)) 
 #endif  
 	    die(_("bad command line options: %s"), daemon->namebuff, EC_BADCONF);
 	}

src/rfc1035.c

@@ -337,7 +337,7 @@ unsigned char *skip_questions(struct dns_header *header, size_t plen)
   return ansp;
 }
 
-static unsigned char *skip_section(unsigned char *ansp, int count, struct dns_header *header, size_t plen)
+unsigned char *skip_section(unsigned char *ansp, int count, struct dns_header *header, size_t plen)
 {
   int i, rdlen;
   
@@ -1242,7 +1242,7 @@ int check_for_local_domain(char *name, time_t now)
   struct ptr_record *ptr;
   struct naptr *naptr;
 
-  if ((crecp = cache_find_by_name(NULL, name, now, F_IPV4 | F_IPV6 | F_CNAME)) &&
+  if ((crecp = cache_find_by_name(NULL, name, now, F_IPV4 | F_IPV6 | F_CNAME | F_NO_RR)) &&
       (crecp->flags & (F_HOSTS | F_DHCP | F_CONFIG)))
     return 1;
   
@@ -1452,7 +1452,8 @@ static unsigned long crec_ttl(struct crec *crecp, time_t now)
 
 /* return zero if we can't answer from cache, or packet size if we can */
 size_t answer_request(struct dns_header *header, char *limit, size_t qlen,  
-		      struct in_addr local_addr, struct in_addr local_netmask, time_t now) 
+		      struct in_addr local_addr, struct in_addr local_netmask, 
+		      time_t now, int *ad_reqd) 
 {
   char *name = daemon->namebuff;
   unsigned char *p, *ansp, *pheader;
@@ -1468,10 +1469,13 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
   struct mx_srv_record *rec;
   size_t len;
  
-  /* Don't return AD set even for local data if checking disabled. */
+  /* Don't return AD set if checking disabled. */
   if (header->hb4 & HB4_CD)
     sec_data = 0;
   
+  /* RFC 6840 5.7 */
+  *ad_reqd = header->hb4 & HB4_AD;
+  
   /* If there is an RFC2671 pseudoheader then it will be overwritten by
      partial replies, so we have to do a dry run to see if we can answer
      the query. We check to see if the do bit is set, if so we always
@@ -1490,6 +1494,7 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
       GETSHORT(flags, pheader);
       
       sec_reqd = flags & 0x8000; /* do bit */ 
+      *ad_reqd = 1;
 
       /* If our client is advertising a larger UDP packet size
 	 than we allow, trim it so that we don't get an overlarge
@@ -1551,37 +1556,33 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
 	}
 
 #ifdef HAVE_DNSSEC
-      if (option_bool(OPT_DNSSEC_VALID) && (qtype == T_DNSKEY || qtype == T_DS || qtype == T_RRSIG))
+      if (option_bool(OPT_DNSSEC_VALID) && (qtype == T_DNSKEY || qtype == T_DS))
 	{
-	  int gotone = 0, have_rrsig = 0;
+	  int gotone = 0;
 	  struct blockdata *keydata;
 
 	  /* Do we have RRSIG? Can't do DS or DNSKEY otherwise. */
-	  crecp = NULL;
+	  if (sec_reqd)
-	  while ((crecp = cache_find_by_name(crecp, name, now, F_DNSKEY | F_DS)))
-	    if (crecp->uid == qclass && (qtype == T_RRSIG || crecp->addr.sig.type_covered == qtype))
-	      {
-		have_rrsig = 1;
-		break;
-	      }
-	  
-	  if (qtype == T_RRSIG && have_rrsig)
 	    {
-	      ans = gotone = 1;
-	      auth = 0;
-	    }
-	  else if (qtype == T_DS && have_rrsig)
-	    {
-	      auth = 0;
 	      crecp = NULL;
-	      while ((crecp = cache_find_by_name(crecp, name, now, F_DS)))
+	      while ((crecp = cache_find_by_name(crecp, name, now, F_DNSKEY | F_DS)))
-		if (crecp->uid == qclass)
+		if (crecp->uid == qclass && crecp->addr.sig.type_covered == qtype)
-		  {
+		  break;
-		    ans = gotone = 1;
+	    }
-		    if (!dryrun && (keydata = blockdata_retrieve(crecp->addr.ds.keydata, crecp->addr.ds.keylen, NULL)))
+	  
+	  if (!sec_reqd || crecp)
+	    {
+	      if (qtype == T_DS)
+		{
+		  crecp = NULL;
+		  while ((crecp = cache_find_by_name(crecp, name, now, F_DS)))
+		    if (crecp->uid == qclass)
 		      {
-			struct all_addr a;
+			gotone = 1;
-			a.addr.keytag =  crecp->addr.ds.keytag;
+			if (!dryrun && (keydata = blockdata_retrieve(crecp->addr.ds.keydata, crecp->addr.ds.keylen, NULL)))
+			  {			     			      
+			    struct all_addr a;
+			    a.addr.keytag =  crecp->addr.ds.keytag;
 			    log_query(F_KEYTAG | (crecp->flags & F_CONFIG), name, &a, "DS keytag %u");
 			    if (add_resource_record(header, limit, &trunc, nameoffset, &ansp, 
 						    crec_ttl(crecp, now), &nameoffset,
@@ -1589,20 +1590,16 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
 						    crecp->addr.ds.keytag, crecp->addr.ds.algo, crecp->addr.ds.digest, crecp->addr.ds.keylen, keydata))
 			      anscount++;
 			    
+			  } 
 		      }
-		  }
+		}
-	    }
+	      else /* DNSKEY */
-	  else if (qtype == T_DNSKEY)
+		{
-	    {
+		  crecp = NULL;
-	      crecp = NULL;
+		  while ((crecp = cache_find_by_name(crecp, name, now, F_DNSKEY)))
-	      while ((crecp = cache_find_by_name(crecp, name, now, F_DNSKEY)))
+		    if (crecp->uid == qclass)
-		if (crecp->uid == qclass)
-		  {
-		    if ((crecp->flags & F_CONFIG) || have_rrsig) /* Return configured keys without an RRISG */
 		      {
-			if (!(crecp->flags & F_CONFIG))
+			gotone = 1;
-			  auth = 0, gotone = 1;
-			ans = 1;
 			if (!dryrun && (keydata = blockdata_retrieve(crecp->addr.key.keydata, crecp->addr.key.keylen, NULL)))
 			  {			     			      
 			    struct all_addr a;
@@ -1615,30 +1612,27 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
 			      anscount++;
 			  }
 		      }
-		  }
+		}
 	    }
 	  
 	  /* Now do RRSIGs */
 	  if (gotone)
 	    {
-	      crecp = NULL;
+	      ans = 1;
-	      while ((crecp = cache_find_by_name(crecp, name, now, F_DNSKEY | F_DS)))
+	      auth = 0;
-		if (crecp->uid == qclass && (qtype == T_RRSIG || (sec_reqd && crecp->addr.sig.type_covered == qtype)) &&
+	      if (!dryrun && sec_reqd)
-		    !dryrun &&
+		{
-		    (keydata = blockdata_retrieve(crecp->addr.sig.keydata, crecp->addr.sig.keylen, NULL)))
+		  crecp = NULL;
-		  {
+		  while ((crecp = cache_find_by_name(crecp, name, now, F_DNSKEY | F_DS)))
-		    if (qtype == T_RRSIG)
+		    if (crecp->uid == qclass && crecp->addr.sig.type_covered == qtype &&
+			(keydata = blockdata_retrieve(crecp->addr.sig.keydata, crecp->addr.sig.keylen, NULL)))
 		      {
-			char types[20];
-			querystr("rrsig", types, crecp->addr.sig.type_covered);
-			log_query(F_RRNAME, name, NULL, types);
-		      }
-		    if ((keydata = blockdata_retrieve(crecp->addr.sig.keydata, crecp->addr.sig.keylen, NULL)) &&
 			add_resource_record(header, limit, &trunc, nameoffset, &ansp, 
 					    crec_ttl(crecp, now), &nameoffset,
-					    T_RRSIG, qclass, "t", crecp->addr.sig.keylen, keydata))
+					    T_RRSIG, qclass, "t", crecp->addr.sig.keylen, keydata);
-		      anscount++;
+			anscount++;
-		  }
+		      }
+		}
 	    }
 	}
 #endif	     
@@ -1742,24 +1736,25 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
 		      else if (crecp->flags & F_DNSSECOK)
 			{
 			  int gotsig = 0;
+			  struct crec *rr_crec = NULL;
 
-			  crecp = NULL;
+			  while ((rr_crec = cache_find_by_name(rr_crec, name, now, F_DS | F_DNSKEY)))
-			  while ((crecp = cache_find_by_name(crecp, name, now, F_DS | F_DNSKEY)))
 			    {
-			      if (crecp->addr.sig.type_covered == T_PTR && crecp->uid == C_IN)
+			      if (rr_crec->addr.sig.type_covered == T_PTR && rr_crec->uid == C_IN)
 				{
-				  char *sigdata = blockdata_retrieve(crecp->addr.sig.keydata, crecp->addr.sig.keylen, NULL);
+				  char *sigdata = blockdata_retrieve(rr_crec->addr.sig.keydata, rr_crec->addr.sig.keylen, NULL);
 				  gotsig = 1;
 				  
 				  if (!dryrun && 
 				      add_resource_record(header, limit, &trunc, nameoffset, &ansp, 
-							  crecp->ttd - now, &nameoffset,
+							  rr_crec->ttd - now, &nameoffset,
 							  T_RRSIG, C_IN, "t", crecp->addr.sig.keylen, sigdata))
 				    anscount++;
 				}
 			    } 
-			  /* Need to re-run original cache search */
+			  
-			  crecp = gotsig ? cache_find_by_addr(NULL, &addr, now, is_arpa) : NULL;
+			  if (!gotsig)
+			    crecp = NULL;
 			}
 #endif
 		    }
@@ -1923,7 +1918,7 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
 		}
 
 	    cname_restart:
-	      if ((crecp = cache_find_by_name(NULL, name, now, flag | F_CNAME)))
+	      if ((crecp = cache_find_by_name(NULL, name, now, flag | F_CNAME | (dryrun ? F_NO_RR : 0))))
 		{
 		  int localise = 0;
 		  
@@ -1954,7 +1949,7 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
 		      else if (crecp->flags & F_DNSSECOK)
 			{
 			  /* We're returning validated data, need to return the RRSIG too. */
-			  
+			  struct crec *rr_crec = NULL;
 			  int sigtype = type;
 			  /* The signature may have expired even though the data is still in cache, 
 			     forward instead of answering from cache if so. */
@@ -1963,23 +1958,23 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
 			  if (crecp->flags & F_CNAME)
 			    sigtype = T_CNAME;
 			  
-			  crecp = NULL;
+			  while ((rr_crec = cache_find_by_name(rr_crec, name, now, F_DS | F_DNSKEY)))
-			  while ((crecp = cache_find_by_name(crecp, name, now, F_DS | F_DNSKEY)))
 			    {
-			      if (crecp->addr.sig.type_covered == sigtype && crecp->uid == C_IN)
+			      if (rr_crec->addr.sig.type_covered == sigtype && rr_crec->uid == C_IN)
 				{
-				  char *sigdata = blockdata_retrieve(crecp->addr.sig.keydata, crecp->addr.sig.keylen, NULL);
+				  char *sigdata = blockdata_retrieve(rr_crec->addr.sig.keydata, rr_crec->addr.sig.keylen, NULL);
 				  gotsig = 1;
 				  
 				  if (!dryrun && 
 				      add_resource_record(header, limit, &trunc, nameoffset, &ansp, 
-							  crecp->ttd - now, &nameoffset,
+							  rr_crec->ttd - now, &nameoffset,
-							  T_RRSIG, C_IN, "t", crecp->addr.sig.keylen, sigdata))
+							  T_RRSIG, C_IN, "t", rr_crec->addr.sig.keylen, sigdata))
 				    anscount++;
 				}
 			    }
-			  /* Need to re-run original cache search */
+			  
-			  crecp = gotsig ? cache_find_by_name(NULL, name, now, flag | F_CNAME) : NULL;
+			  if (!gotsig)
+			    crecp = NULL;
 			}
 #endif
 		    }		 
@@ -2072,7 +2067,7 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
 	  if (qtype == T_CNAME || qtype == T_ANY)
 	    {
 	      if ((crecp = cache_find_by_name(NULL, name, now, F_CNAME)) &&
-		  (qtype == T_CNAME || (crecp->flags & (F_HOSTS | F_DHCP | F_CONFIG))))
+		  (qtype == T_CNAME || (crecp->flags & (F_HOSTS | F_DHCP | F_CONFIG  | (dryrun ? F_NO_RR : 0)))))
 		{
 		  if (!(crecp->flags & F_DNSSECOK))
 		    sec_data = 0;
@@ -2111,7 +2106,7 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
 		  }
 	      
 	      if (!found && (option_bool(OPT_SELFMX) || option_bool(OPT_LOCALMX)) && 
-		  cache_find_by_name(NULL, name, now, F_HOSTS | F_DHCP))
+		  cache_find_by_name(NULL, name, now, F_HOSTS | F_DHCP | F_NO_RR))
 		{ 
 		  ans = 1;
 		  if (!dryrun)
@@ -2264,16 +2259,15 @@ size_t answer_request(struct dns_header *header, char *limit, size_t qlen,
   header->nscount = htons(0);
   header->arcount = htons(addncount);
 
-  header->hb4 &= ~HB4_AD;
   len = ansp - (unsigned char *)header;
   
   if (have_pseudoheader)
-    {
+    len = add_pseudoheader(header, len, (unsigned char *)limit, 0, NULL, 0, sec_reqd);
-      len = add_pseudoheader(header, len, (unsigned char *)limit, 0, NULL, 0, sec_reqd);
-      if (sec_reqd && sec_data)
-	header->hb4 |= HB4_AD;
   
-    }
+  if (*ad_reqd && sec_data)
+    header->hb4 |= HB4_AD;
+  else
+    header->hb4 &= ~HB4_AD;
   
   return len;
 }

src/util.c

@@ -482,66 +482,6 @@ int parse_hex(char *in, unsigned char *out, int maxlen,
   return i;
 }
 
-#ifdef HAVE_DNSSEC
-static int charval(char c)
-{
-  if (c >= 'A' && c <= 'Z')
-    return c - 'A';
-  
-  if (c >= 'a' && c <= 'z')
-    return c - 'a' + 26;
-
-  if (c >= '0' && c <= '9')
-    return c - '0' + 52;
-
-  if (c == '+')
-    return 62;
-  
-  if (c == '/')
-    return 63;
-
-  if (c == '=')
-    return -1;
-
-  return -2;
-}
-
-int parse_base64(char *in, char *out)
-{
-  char *p = out;
-  int i, val[4];
-
-  while (*in)
-    {
-      for (i = 0; i < 4; i++)
-	{ 
-	  while (*in == ' ') 
-	    in++;
-	  if (*in == 0)
-	    return -1;
-	  if ((val[i] = charval(*in++)) == -2)
-	    return -1;
-	}
-          
-      while (*in == ' ') 
-	in++;
-      
-      if (val[1] == -1)
-	return -1; /* too much padding */
-      
-      *p++ = (val[0] << 2) | (val[1] >> 4);
-      
-      if (val[2] != -1)
-	*p++ = (val[1] << 4) | ( val[2] >> 2);
-
-      if (val[3] != -1)
-	*p++ = (val[2] << 6) | val[3];
-    }
-
-  return p - out;
-}
-#endif
-
 /* return 0 for no match, or (no matched octets) + 1 */
 int memcmp_masked(unsigned char *a, unsigned char *b, int len, unsigned int mask)
 {

trust-anchors.conf

@@ -1,8 +1,9 @@
-# The root DNSSEC trust anchors, valid as at 30/01/2014
+# The root DNSSEC trust anchor, valid as at 30/01/2014
+
+# Note that this is a DS record (ie a hash of the root Zone Signing Key) 
+# If was downloaded from https://data.iana.org/root-anchors/root-anchors.xml
+
+trust-anchor=.,19036,8,2,49AAC11D7B6F6446702E54A1607371607A1A41855200FD2CE1CDDE32F24E8FB5
 
 
-dnskey=.,257,8,AwEAAagAIKlVZrpC6Ia7gEzahOR+9W29euxhJhVVLOyQbSEW0O8gcCjF FVQUTf6v58fLjwBd0YI0EzrAcQqBGCzh/RStIoO8g0NfnfL2MTJRkxoX bfDaUeVPQuYEhg37NZWAJQ9VnMVDxP/VHL496M/QZxkjf5/Efucp2gaD X6RS6CXpoY68LsvPVjR0ZSwzz1apAzvN9dlzEheX7ICJBBtuA6G3LQpz W5hOA2hzCTMjJPJ8LbqF6dsV6DoBQzgul0sGIcGOYl7OyQdXfZ57relS Qageu+ipAdTTJ25AsRTAoub8ONGcLmqrAmRLKBP1dfwhYB4N7knNnulq QxA+Uk1ihz0=
-dnskey=.,256,8,AwEAAb8sU6pbYMWRbkRnEuEZw9NSir707TkOcF+UL1XiK4NDJOvXRyX1 95Am5dQ7bRnnuySZ3daf37vvjUUhuIWUAQ4stht8nJfYxVQXDYjSpGH5 I6Hf/0CZEoNP6cNvrQ7AFmKkmv00xWExKQjbvnRPI4bqpMwtHVzn6Wyb BZ6kuqED
-dnskey=.,256,8,AwEAAYRU41/8smgAvuSojEP4jaj5Yll7WPaUKpYvnz2pnX2VIvRn4jsy Jns80bloenG6X9ebJVy2CFtZQLKHP8DcKmIFotdgs2HolyocY1am/+33 4RtzusM2ojkhjn1FRGtuSE9s2TSz1ISv0yVnFyu+EP/ZkiWnDfWeVrJI SEWBEr4V
-