update to NSD 4.1.0, ok sthen@

23 files changed, 749 insertions, 173 deletions

diff --git a/usr.sbin/nsd/compat/fake-rfc2553.c b/usr.sbin/nsd/compat/fake-rfc2553.c
index 91ddf8a809b..0f0f34f1fb2 100644
--- a/usr.sbin/nsd/compat/fake-rfc2553.c
+++ b/usr.sbin/nsd/compat/fake-rfc2553.c
@@ -120,12 +120,10 @@ addrinfo *malloc_ai(int port, u_long addr, const struct addrinfo *hints)
 {
 	struct addrinfo *ai;
 
-	ai = malloc(sizeof(*ai) + sizeof(struct sockaddr_in));
+	ai = calloc(1, sizeof(*ai) + sizeof(struct sockaddr_in));
 	if (ai == NULL)
 		return (NULL);
 	
-	memset(ai, '\0', sizeof(*ai) + sizeof(struct sockaddr_in));
-	
 	ai->ai_addr = (struct sockaddr *)(ai + 1);
 	/* XXX -- ssh doesn't use sa_len */
 	ai->ai_addrlen = sizeof(struct sockaddr_in);
diff --git a/usr.sbin/nsd/configlexer.lex b/usr.sbin/nsd/configlexer.lex
index c9f71fc38f4..87b96b3c3e7 100644
--- a/usr.sbin/nsd/configlexer.lex
+++ b/usr.sbin/nsd/configlexer.lex
@@ -14,6 +14,9 @@
 #include <errno.h>
 #include <string.h>
 #include <strings.h>
+#ifdef HAVE_GLOB_H
+# include <glob.h>
+#endif
 
 #include "options.h"
 #include "configyyrename.h"
@@ -27,22 +30,40 @@ void c_error(const char *message);
 #endif
 
 struct inc_state {
-	const char* filename;
+	char* filename;
 	int line;
+	YY_BUFFER_STATE buffer;
+	struct inc_state* next;
 };
-static struct inc_state parse_stack[MAXINCLUDES];
-static YY_BUFFER_STATE include_stack[MAXINCLUDES];
-static int config_include_stack_ptr = 0;
+static struct inc_state* config_include_stack = NULL;
+static int inc_depth = 0;
+static int inc_prev = 0;
+static int num_args = 0;
+
+void init_cfg_parse(void)
+{
+	config_include_stack = NULL;
+	inc_depth = 0;
+	inc_prev = 0;
+	num_args = 0;
+}
 
 static void config_start_include(const char* filename)
 {
 	FILE *input;
+	struct inc_state* s;
+	char* nm;
+	if(inc_depth++ > 10000000) {
+		c_error_msg("too many include files");
+		return;
+	}
 	if(strlen(filename) == 0) {
 		c_error_msg("empty include file name");
 		return;
 	}
-	if(config_include_stack_ptr >= MAXINCLUDES) {
-		c_error_msg("includes nested too deeply, skipped (>%d)", MAXINCLUDES);
+	s = (struct inc_state*)malloc(sizeof(*s));
+	if(!s) {
+		c_error_msg("include %s: malloc failure", filename);
 		return;
 	}
 	if (cfg_parser->chroot) {
@@ -54,29 +75,89 @@ static void config_start_include(const char* filename)
 		}
 		filename += l - 1; /* strip chroot without trailing slash */
 	}
+	nm = strdup(filename);
+	if(!nm) {
+		c_error_msg("include %s: strdup failure", filename);
+		free(s);
+		return;
+	}
 	input = fopen(filename, "r");
 	if(!input) {
 		c_error_msg("cannot open include file '%s': %s",
 			filename, strerror(errno));
+		free(s);
+		free(nm);
 		return;
 	}
 	LEXOUT(("switch_to_include_file(%s) ", filename));
-	parse_stack[config_include_stack_ptr].filename = cfg_parser->filename;
-	parse_stack[config_include_stack_ptr].line = cfg_parser->line;
-	include_stack[config_include_stack_ptr] = YY_CURRENT_BUFFER;
-	cfg_parser->filename = region_strdup(cfg_parser->opt->region, filename);
+	s->filename = cfg_parser->filename;
+	s->line = cfg_parser->line;
+	s->buffer = YY_CURRENT_BUFFER;
+	s->next = config_include_stack;
+	config_include_stack = s;
+
+	cfg_parser->filename = nm;
 	cfg_parser->line = 1;
 	yy_switch_to_buffer(yy_create_buffer(input, YY_BUF_SIZE));
-	++config_include_stack_ptr;
+}
+
+static void config_start_include_glob(const char* filename)
+{
+	 /* check for wildcards */
+#ifdef HAVE_GLOB
+	 glob_t g;
+	 size_t i;
+	 int r, flags;
+	 if(!(!strchr(filename, '*') && !strchr(filename, '?') &&
+		 !strchr(filename, '[') && !strchr(filename, '{') &&
+		 !strchr(filename, '~'))) {
+		 flags = 0
+#ifdef GLOB_ERR
+		 	 | GLOB_ERR
+#endif
+#ifdef GLOB_NOSORT
+			 | GLOB_NOSORT
+#endif
+#ifdef GLOB_BRACE
+			 | GLOB_BRACE
+#endif
+#ifdef GLOB_TILDE
+			 | GLOB_TILDE
+#endif
+		;
+		memset(&g, 0, sizeof(g));
+		r = glob(filename, flags, NULL, &g);
+		if(r) {
+			/* some error */
+			globfree(&g);
+			if(r == GLOB_NOMATCH)
+				return; /* no matches for pattern */
+			config_start_include(filename); /* let original deal with it */
+			return;
+		}
+		/* process files found, if any */
+		for(i=0; i<(size_t)g.gl_pathc; i++) {
+			config_start_include(g.gl_pathv[i]);
+		}
+		globfree(&g);
+		return;
+	 }
+#endif /* HAVE_GLOB */
+	 config_start_include(filename);
 }
 
 static void config_end_include(void)
 {
-	--config_include_stack_ptr;
-	cfg_parser->filename = parse_stack[config_include_stack_ptr].filename;
-	cfg_parser->line = parse_stack[config_include_stack_ptr].line;
+	struct inc_state* s = config_include_stack;
+	--inc_depth;
+	if(!s) return;
+	free(cfg_parser->filename);
+	cfg_parser->filename = s->filename;
+	cfg_parser->line = s->line;
 	yy_delete_buffer(YY_CURRENT_BUFFER);
-	yy_switch_to_buffer(include_stack[config_include_stack_ptr]);
+	yy_switch_to_buffer(s->buffer);
+	config_include_stack = s->next;
+	free(s);
 }
 
 #ifndef yy_set_bol /* compat definition, for flex 2.4.6 */
@@ -178,6 +259,9 @@ rrl-ipv6-prefix-length{COLON}	{ LEXOUT(("v(%s) ", yytext)); return VAR_RRL_IPV6_
 rrl-whitelist-ratelimit{COLON}	{ LEXOUT(("v(%s) ", yytext)); return VAR_RRL_WHITELIST_RATELIMIT;}
 rrl-whitelist{COLON}	{ LEXOUT(("v(%s) ", yytext)); return VAR_RRL_WHITELIST;}
 zonefiles-check{COLON}	{ LEXOUT(("v(%s) ", yytext)); return VAR_ZONEFILES_CHECK;}
+zonefiles-write{COLON}	{ LEXOUT(("v(%s) ", yytext)); return VAR_ZONEFILES_WRITE;}
+log-time-ascii{COLON}	{ LEXOUT(("v(%s) ", yytext)); return VAR_LOG_TIME_ASCII;}
+round-robin{COLON}	{ LEXOUT(("v(%s) ", yytext)); return VAR_ROUND_ROBIN;}
 {NEWLINE}		{ LEXOUT(("NL\n")); cfg_parser->line++;}
 
 	/* Quoted strings. Strip leading and ending quotes */
@@ -207,7 +291,7 @@ include{COLON}		{ LEXOUT(("v(%s) ", yytext)); BEGIN(include); }
 <include>\"		{ LEXOUT(("IQS ")); BEGIN(include_quoted); }
 <include>{UNQUOTEDLETTER}*	{
 	LEXOUT(("Iunquotedstr(%s) ", yytext));
-	config_start_include(yytext);
+	config_start_include_glob(yytext);
 	BEGIN(INITIAL);
 }
 <include_quoted><<EOF>>	{
@@ -219,12 +303,12 @@ include{COLON}		{ LEXOUT(("v(%s) ", yytext)); BEGIN(include); }
 <include_quoted>\"	{
 	LEXOUT(("IQE "));
 	yytext[yyleng - 1] = '\0';
-	config_start_include(yytext);
+	config_start_include_glob(yytext);
 	BEGIN(INITIAL);
 }
 <INITIAL><<EOF>>	{
 	yy_set_bol(1); /* Set beginning of line, so "^" rules match.  */
-	if (config_include_stack_ptr == 0) {
+	if (!config_include_stack) {
 		yyterminate();
 	} else {
 		fclose(yyin);
diff --git a/usr.sbin/nsd/dbaccess.c b/usr.sbin/nsd/dbaccess.c
index 4ecbf7a990b..3a60d517665 100644
--- a/usr.sbin/nsd/dbaccess.c
+++ b/usr.sbin/nsd/dbaccess.c
@@ -32,8 +32,8 @@
 #include "nsd.h"
 
 static time_t udb_time = 0;
-static unsigned udb_rrsets = 0;
-static unsigned udb_rrset_count = 0;
+static unsigned long udb_rrsets = 0;
+static unsigned long udb_rrset_count = 0;
 
 void
 namedb_close(struct namedb* db)
@@ -193,7 +193,8 @@ static void read_node_elem(udb_base* udb, namedb_type* db,
 		if(++udb_rrsets % ZONEC_PCT_COUNT == 0 && time(NULL) > udb_time + ZONEC_PCT_TIME) {
 			udb_time = time(NULL);
 			VERBOSITY(1, (LOG_INFO, "read %s %d %%",
-				zone->opts->name, udb_rrsets*100/udb_rrset_count));
+				zone->opts->name,
+				(int)(udb_rrsets*((unsigned long)100)/udb_rrset_count)));
 		}
 	}
 	region_free_all(dname_region);
@@ -268,6 +269,9 @@ namedb_zone_create(namedb_type* db, const dname_type* dname,
 	zone->dshashtree = NULL;
 #endif
 	zone->opts = zo;
+	zone->filename = NULL;
+	zone->logstr = NULL;
+	zone->mtime = 0;
 	zone->is_secure = 0;
 	zone->is_changed = 0;
 	zone->is_ok = 1;
@@ -302,9 +306,16 @@ namedb_zone_delete(namedb_type* db, zone_type* zone)
 	hash_tree_delete(db->region, zone->wchashtree);
 	hash_tree_delete(db->region, zone->dshashtree);
 #endif
+	if(zone->filename)
+		region_recycle(db->region, zone->filename,
+			strlen(zone->filename)+1);
+	if(zone->logstr)
+		region_recycle(db->region, zone->logstr,
+			strlen(zone->logstr)+1);
 	region_recycle(db->region, zone, sizeof(zone_type));
 }
 
+#ifdef HAVE_MMAP
 /** read a zone */
 static void
 read_zone(udb_base* udb, namedb_type* db, nsd_options_t* opt,
@@ -334,7 +345,9 @@ read_zone(udb_base* udb, namedb_type* db, nsd_options_t* opt,
 	prehash_zone_complete(db, zone);
 #endif
 }
+#endif /* HAVE_MMAP */
 
+#ifdef HAVE_MMAP
 /** read zones from nsd.db */
 static void
 read_zones(udb_base* udb, namedb_type* db, nsd_options_t* opt,
@@ -350,12 +363,15 @@ read_zones(udb_base* udb, namedb_type* db, nsd_options_t* opt,
 		udb_radix_next(udb, &n); /* store in case n is deleted */
 		read_zone(udb, db, opt, dname_region, &z);
 		udb_ptr_zero(&z, udb);
+		if(nsd.signal_hint_shutdown) break;
 	}
 	udb_ptr_unlink(&ztree, udb);
 	udb_ptr_unlink(&n, udb);
 	udb_ptr_unlink(&z, udb);
 }
+#endif /* HAVE_MMAP */
 
+#ifdef HAVE_MMAP
 /** try to read the udb file or fail */
 static int
 try_read_udb(namedb_type* db, int fd, const char* filename,
@@ -392,6 +408,7 @@ try_read_udb(namedb_type* db, int fd, const char* filename,
 	region_destroy(dname_region);
 	return 1;
 }
+#endif /* HAVE_MMAP */
 
 struct namedb *
 namedb_open (const char* filename, nsd_options_t* opt)
@@ -405,15 +422,6 @@ namedb_open (const char* filename, nsd_options_t* opt)
 	region_type* db_region;
 	int fd;
 
-	/* attempt to open, if does not exist, create a new one */
-	fd = open(filename, O_RDWR);
-	if(fd == -1) {
-		if(errno != ENOENT) {
-			log_msg(LOG_ERR, "%s: %s", filename, strerror(errno));
-			return NULL;
-		}
-	}
-
 #ifdef USE_MMAP_ALLOC
 	db_region = region_create_custom(mmap_alloc, mmap_free, MMAP_ALLOC_CHUNK_SIZE,
 		MMAP_ALLOC_LARGE_OBJECT_SIZE, MMAP_ALLOC_INITIAL_CLEANUP_SIZE, 1);
@@ -427,15 +435,38 @@ namedb_open (const char* filename, nsd_options_t* opt)
 	db->zonetree = radix_tree_create(db->region);
 	db->diff_skip = 0;
 	db->diff_pos = 0;
+	zonec_setup_parser(db);
 
 	if (gettimeofday(&(db->diff_timestamp), NULL) != 0) {
 		log_msg(LOG_ERR, "unable to load %s: cannot initialize"
 				 "timestamp", filename);
 		region_destroy(db_region);
-		close(fd);
 		return NULL;
         }
 
+	/* in dbless mode there is no file to read or mmap */
+	if(filename == NULL || filename[0] == 0) {
+		db->udb = NULL;
+		return db;
+	}
+
+#ifndef HAVE_MMAP
+	/* no mmap() system call, use dbless mode */
+	VERBOSITY(1, (LOG_INFO, "no mmap(), ignoring database %s", filename));
+	db->udb = NULL;
+	(void)fd; (void)opt;
+	return db;
+#else /* HAVE_MMAP */
+
+	/* attempt to open, if does not exist, create a new one */
+	fd = open(filename, O_RDWR);
+	if(fd == -1) {
+		if(errno != ENOENT) {
+			log_msg(LOG_ERR, "%s: %s", filename, strerror(errno));
+			region_destroy(db_region);
+			return NULL;
+		}
+	}
 	/* attempt to read the file (if it exists) */
 	if(fd != -1) {
 		if(!try_read_udb(db, fd, filename, opt))
@@ -453,8 +484,8 @@ namedb_open (const char* filename, nsd_options_t* opt)
 			return NULL;
 		}
 	}
-	zonec_setup_parser(db);
 	return db;
+#endif /* HAVE_MMAP */
 }
 
 /** the the file mtime stat (or nonexist or error) */
@@ -480,7 +511,7 @@ namedb_read_zonefile(struct nsd* nsd, struct zone* zone, udb_base* taskudb,
 	int nonexist = 0;
 	unsigned int errors;
 	const char* fname;
-	if(!nsd->db || !nsd->db->udb || !zone || !zone->opts || !zone->opts->pattern->zonefile)
+	if(!nsd->db || !zone || !zone->opts || !zone->opts->pattern->zonefile)
 		return;
 	fname = config_make_zonefile(zone->opts, nsd);
 	if(!file_get_mtime(fname, &mtime, &nonexist)) {
@@ -493,16 +524,21 @@ namedb_read_zonefile(struct nsd* nsd, struct zone* zone, udb_base* taskudb,
 		if(taskudb) task_new_soainfo(taskudb, last_task, zone, 0);
 		return;
 	} else {
-		const char* zone_fname = udb_zone_get_file_str(nsd->db->udb,
-			dname_name(domain_dname(zone->apex)), domain_dname(
-			zone->apex)->name_size);
+		const char* zone_fname = zone->filename;
+		time_t zone_mtime = zone->mtime;
+		if(nsd->db->udb) {
+			zone_fname = udb_zone_get_file_str(nsd->db->udb,
+				dname_name(domain_dname(zone->apex)),
+				domain_dname(zone->apex)->name_size);
+			zone_mtime = (time_t)udb_zone_get_mtime(nsd->db->udb,
+				dname_name(domain_dname(zone->apex)),
+				domain_dname(zone->apex)->name_size);
+		}
 		/* if no zone_fname, then it was acquired in zone transfer,
 		 * see if the file is newer than the zone transfer
 		 * (regardless if this is a different file), because the
 		 * zone transfer is a different content source too */
-		if(!zone_fname && udb_zone_get_mtime(nsd->db->udb,
-			dname_name(domain_dname(zone->apex)), domain_dname(
-			zone->apex)->name_size) >= (uint64_t)mtime) {
+		if(!zone_fname && zone_mtime >= mtime) {
 			VERBOSITY(3, (LOG_INFO, "zonefile %s is older than "
 				"zone transfer in memory", fname));
 			return;
@@ -510,10 +546,7 @@ namedb_read_zonefile(struct nsd* nsd, struct zone* zone, udb_base* taskudb,
 		/* if zone_fname, then the file was acquired from reading it,
 		 * and see if filename changed or mtime newer to read it */
 		} else if(zone_fname && fname &&
-		   strcmp(zone_fname, fname) == 0 &&
-		   udb_zone_get_mtime(nsd->db->udb, dname_name(domain_dname(
-			zone->apex)), domain_dname(zone->apex)->name_size)
-			>= (uint64_t)mtime) {
+		   strcmp(zone_fname, fname) == 0 && zone_mtime >= mtime) {
 			VERBOSITY(3, (LOG_INFO, "zonefile %s is not modified",
 				fname));
 			return;
@@ -532,8 +565,6 @@ namedb_read_zonefile(struct nsd* nsd, struct zone* zone, udb_base* taskudb,
 #endif /* NSEC3 */
 	errors = zonec_read(zone->opts->name, fname, zone);
 	if(errors > 0) {
-		region_type* dname_region;
-		udb_ptr z;
 		log_msg(LOG_ERR, "zone %s file %s read with %u errors",
 			zone->opts->name, fname, errors);
 		/* wipe (partial) zone from memory */
@@ -546,32 +577,57 @@ namedb_read_zonefile(struct nsd* nsd, struct zone* zone, udb_base* taskudb,
 		nsec3_clear_precompile(nsd->db, zone);
 		zone->nsec3_param = NULL;
 #endif /* NSEC3 */
-		/* see if we can revert to the udb stored version */
-		if(!udb_zone_search(nsd->db->udb, &z, dname_name(domain_dname(
-			zone->apex)), domain_dname(zone->apex)->name_size)) {
-			/* tell that zone contents has been lost */
-			if(taskudb) task_new_soainfo(taskudb, last_task, zone, 0);
-			return;
+		if(nsd->db->udb) {
+			region_type* dname_region;
+			udb_ptr z;
+			/* see if we can revert to the udb stored version */
+			if(!udb_zone_search(nsd->db->udb, &z, dname_name(domain_dname(
+				zone->apex)), domain_dname(zone->apex)->name_size)) {
+				/* tell that zone contents has been lost */
+				if(taskudb) task_new_soainfo(taskudb, last_task, zone, 0);
+				return;
+			}
+			/* read from udb */
+			dname_region = region_create(xalloc, free);
+			udb_rrsets = 0;
+			udb_rrset_count = ZONE(&z)->rrset_count;
+			udb_time = time(NULL);
+			read_zone_data(nsd->db->udb, nsd->db, dname_region, &z, zone);
+			region_destroy(dname_region);
+			udb_ptr_unlink(&z, nsd->db->udb);
+		} else {
+			if(zone->filename)
+				region_recycle(nsd->db->region, zone->filename,
+					strlen(zone->filename)+1);
+			zone->filename = NULL;
+			if(zone->logstr)
+				region_recycle(nsd->db->region, zone->logstr,
+					strlen(zone->logstr)+1);
+			zone->logstr = NULL;
 		}
-		/* read from udb */
-		dname_region = region_create(xalloc, free);
-		udb_rrsets = 0;
-		udb_rrset_count = ZONE(&z)->rrset_count;
-		udb_time = time(NULL);
-		read_zone_data(nsd->db->udb, nsd->db, dname_region, &z, zone);
-		region_destroy(dname_region);
-		udb_ptr_unlink(&z, nsd->db->udb);
 	} else {
 		VERBOSITY(1, (LOG_INFO, "zone %s read with no errors",
 			zone->opts->name));
 		zone->is_ok = 1;
 		zone->is_changed = 0;
 		/* store zone into udb */
-		if(!write_zone_to_udb(nsd->db->udb, zone, mtime, fname)) {
-			log_msg(LOG_ERR, "failed to store zone in db");
+		if(nsd->db->udb) {
+			if(!write_zone_to_udb(nsd->db->udb, zone, mtime, fname)) {
+				log_msg(LOG_ERR, "failed to store zone in db");
+			} else {
+				VERBOSITY(2, (LOG_INFO, "zone %s written to db",
+					zone->opts->name));
+			}
 		} else {
-			VERBOSITY(2, (LOG_INFO, "zone %s written to db",
-				zone->opts->name));
+			zone->mtime = mtime;
+			if(zone->filename)
+				region_recycle(nsd->db->region, zone->filename,
+					strlen(zone->filename)+1);
+			zone->filename = region_strdup(nsd->db->region, fname);
+			if(zone->logstr)
+				region_recycle(nsd->db->region, zone->logstr,
+					strlen(zone->logstr)+1);
+			zone->logstr = NULL;
 		}
 	}
 	if(taskudb) task_new_soainfo(taskudb, last_task, zone, 0);
@@ -600,5 +656,6 @@ void namedb_check_zonefiles(struct nsd* nsd, nsd_options_t* opt,
 	/* check all zones in opt, create if not exist in main db */
 	RBTREE_FOR(zo, zone_options_t*, opt->zone_options) {
 		namedb_check_zonefile(nsd, taskudb, last_task, zo);
+		if(nsd->signal_hint_shutdown) break;
 	}
 }
diff --git a/usr.sbin/nsd/dbcreate.c b/usr.sbin/nsd/dbcreate.c
index 92f3976e4bc..24866af774b 100644
--- a/usr.sbin/nsd/dbcreate.c
+++ b/usr.sbin/nsd/dbcreate.c
@@ -116,7 +116,7 @@ write_zone(udb_base* udb, udb_ptr* z, zone_type* zone)
 	/* write all domains in the zone */
 	domain_type* walk;
 	rrset_type* rrset;
-	int n = 0, c = 0;
+	unsigned long n = 0, c = 0;
 	time_t t = time(NULL);
 
 	/* count domains: for pct logging */
@@ -138,7 +138,7 @@ write_zone(udb_base* udb, udb_ptr* z, zone_type* zone)
 		if(++c % ZONEC_PCT_COUNT == 0 && time(NULL) > t + ZONEC_PCT_TIME) {
 			t = time(NULL);
 			VERBOSITY(1, (LOG_INFO, "write %s %d %%",
-				zone->opts->name, c*100/n));
+				zone->opts->name, (int)(c*((unsigned long)100)/n)));
 		}
 	}
 	return 1;
@@ -185,16 +185,20 @@ print_rrs(FILE* out, struct zone* zone)
 	rrset_type *rrset;
 	domain_type *domain = zone->apex;
 	region_type* region = region_create(xalloc, free);
+	region_type* rr_region = region_create(xalloc, free);
+	buffer_type* rr_buffer = buffer_create(region, MAX_RDLENGTH);
 	struct state_pretty_rr* state = create_pretty_rr(region);
 	/* first print the SOA record for the zone */
 	if(zone->soa_rrset) {
 		size_t i;
 		for(i=0; i < zone->soa_rrset->rr_count; i++) {
-			if(!print_rr(out, state, &zone->soa_rrset->rrs[i])){
+			if(!print_rr(out, state, &zone->soa_rrset->rrs[i],
+				rr_region, rr_buffer)){
 				log_msg(LOG_ERR, "There was an error "
 				   "printing SOARR to zone %s",
 				   zone->opts->name);
 				region_destroy(region);
+				region_destroy(rr_region);
 				return 0;
 			}
 		}
@@ -208,11 +212,13 @@ print_rrs(FILE* out, struct zone* zone)
 			if(rrset->zone != zone || rrset == zone->soa_rrset)
 				continue;
 			for(i=0; i < rrset->rr_count; i++) {
-				if(!print_rr(out, state, &rrset->rrs[i])){
+				if(!print_rr(out, state, &rrset->rrs[i],
+					rr_region, rr_buffer)){
 					log_msg(LOG_ERR, "There was an error "
 					   "printing RR to zone %s",
 					   zone->opts->name);
 					region_destroy(region);
+					region_destroy(rr_region);
 					return 0;
 				}
 			}
@@ -220,6 +226,7 @@ print_rrs(FILE* out, struct zone* zone)
 		domain = domain_next(domain);
 	}
 	region_destroy(region);
+	region_destroy(rr_region);
 	return 1;
 }
 
@@ -241,11 +248,7 @@ static int
 write_to_zonefile(zone_type* zone, const char* filename, const char* logs)
 {
 	time_t now = time(0);
-	FILE *out;
-	VERBOSITY(1, (LOG_INFO, "writing zone %s to file %s",
-		zone->opts->name, filename));
-
-	out = fopen(filename, "w");
+	FILE *out = fopen(filename, "w");
 	if(!out) {
 		log_msg(LOG_ERR, "cannot write zone %s file %s: %s",
 			zone->opts->name, filename, strerror(errno));
@@ -261,7 +264,11 @@ write_to_zonefile(zone_type* zone, const char* filename, const char* logs)
 		fclose(out);
 		return 0;
 	}
-	fclose(out);
+	if(fclose(out) != 0) {
+		log_msg(LOG_ERR, "cannot write zone %s to file %s: fclose: %s",
+			zone->opts->name, filename, strerror(errno));
+		return 0;
+	}
 	return 1;
 }
 
@@ -328,7 +335,7 @@ namedb_write_zonefile(struct nsd* nsd, zone_options_t* zopt)
 	if(!zopt->pattern->zonefile)
 		return;
 	zone = namedb_find_zone(nsd->db, (const dname_type*)zopt->node.key);
-	if(!zone || !zone->apex)
+	if(!zone || !zone->apex || !zone->soa_rrset)
 		return;
 	/* write if file does not exist, or if changed */
 	/* so, determine filename, create directory components, check exist*/
@@ -344,33 +351,55 @@ namedb_write_zonefile(struct nsd* nsd, zone_options_t* zopt)
 		char logs[4096];
 		char bakfile[4096];
 		udb_ptr zudb;
-		if(!udb_zone_search(nsd->db->udb, &zudb,
-			dname_name(domain_dname(zone->apex)),
-			domain_dname(zone->apex)->name_size))
-			return; /* zone does not exist in db */
+		if(nsd->db->udb) {
+			if(!udb_zone_search(nsd->db->udb, &zudb,
+				dname_name(domain_dname(zone->apex)),
+				domain_dname(zone->apex)->name_size))
+				return; /* zone does not exist in db */
+		}
 		/* write to zfile~ first, then rename if that works */
 		snprintf(bakfile, sizeof(bakfile), "%s~", zfile);
-		if(ZONE(&zudb)->log_str.data) {
+		if(nsd->db->udb && ZONE(&zudb)->log_str.data) {
 			udb_ptr s;
 			udb_ptr_new(&s, nsd->db->udb, &ZONE(&zudb)->log_str);
 			strlcpy(logs, (char*)udb_ptr_data(&s), sizeof(logs));
 			udb_ptr_unlink(&s, nsd->db->udb);
+		} else if(zone->logstr) {
+			strlcpy(logs, zone->logstr, sizeof(logs));
 		} else logs[0] = 0;
+		VERBOSITY(1, (LOG_INFO, "writing zone %s to file %s",
+			zone->opts->name, zfile));
 		if(!write_to_zonefile(zone, bakfile, logs)) {
-			udb_ptr_unlink(&zudb, nsd->db->udb);
+			if(nsd->db->udb)
+				udb_ptr_unlink(&zudb, nsd->db->udb);
+			(void)unlink(bakfile); /* delete failed file */
 			return; /* error already printed */
 		}
 		if(rename(bakfile, zfile) == -1) {
 			log_msg(LOG_ERR, "rename(%s to %s) failed: %s",
 				bakfile, zfile, strerror(errno));
-			udb_ptr_unlink(&zudb, nsd->db->udb);
+			if(nsd->db->udb)
+				udb_ptr_unlink(&zudb, nsd->db->udb);
+			(void)unlink(bakfile); /* delete failed file */
 			return;
 		}
 		zone->is_changed = 0;
-		ZONE(&zudb)->mtime = (uint64_t)time(0);
-		ZONE(&zudb)->is_changed = 0;
-		udb_zone_set_log_str(nsd->db->udb, &zudb, NULL);
-		udb_ptr_unlink(&zudb, nsd->db->udb);
+		if(nsd->db->udb) {
+			ZONE(&zudb)->mtime = (uint64_t)time(0);
+			ZONE(&zudb)->is_changed = 0;
+			udb_zone_set_log_str(nsd->db->udb, &zudb, NULL);
+			udb_ptr_unlink(&zudb, nsd->db->udb);
+		} else {
+			zone->mtime = time(0);
+			if(zone->filename)
+				region_recycle(nsd->db->region, zone->filename,
+					strlen(zone->filename)+1);
+			zone->filename = region_strdup(nsd->db->region, zfile);
+			if(zone->logstr)
+				region_recycle(nsd->db->region, zone->logstr,
+					strlen(zone->logstr)+1);
+			zone->logstr = NULL;
+		}
 	}
 }
 
diff --git a/usr.sbin/nsd/difffile.c b/usr.sbin/nsd/difffile.c
index 5b9bd789845..5510714f364 100644
--- a/usr.sbin/nsd/difffile.c
+++ b/usr.sbin/nsd/difffile.c
@@ -235,6 +235,28 @@ has_data_below(domain_type* top)
 	return 0;
 }
 
+/** check if domain with 0 rrsets has become empty (nonexist) */
+static void
+rrset_zero_nonexist_check(domain_type* domain)
+{
+	/* is the node now an empty node (completely deleted) */
+	if(domain->rrsets == 0) {
+		/* if there is no data below it, it becomes non existing.
+		   also empty nonterminals above it become nonexisting */
+		/* check for data below this node. */
+		if(!has_data_below(domain)) {
+			/* nonexist this domain and all parent empty nonterminals */
+			domain_type* p = domain;
+			while(p != NULL && p->rrsets == 0) {
+				if(has_data_below(p))
+					break;
+				p->is_existing = 0;
+				p = p->parent;
+			}
+		}
+	}
+}
+
 /** remove rrset.  Adjusts zone params.  Does not remove domain */
 static void
 rrset_delete(namedb_type* db, domain_type* domain, rrset_type* rrset)
@@ -277,23 +299,6 @@ rrset_delete(namedb_type* db, domain_type* domain, rrset_type* rrset)
 		sizeof(rr_type) * rrset->rr_count);
 	rrset->rr_count = 0;
 	region_recycle(db->region, rrset, sizeof(rrset_type));
-
-	/* is the node now an empty node (completely deleted) */
-	if(domain->rrsets == 0) {
-		/* if there is no data below it, it becomes non existing.
-		   also empty nonterminals above it become nonexisting */
-		/* check for data below this node. */
-		if(!has_data_below(domain)) {
-			/* nonexist this domain and all parent empty nonterminals */
-			domain_type* p = domain;
-			while(p != NULL && p->rrsets == 0) {
-				if(has_data_below(p))
-					break;
-				p->is_existing = 0;
-				p = p->parent;
-			}
-		}
-	}
 }
 
 static int
@@ -303,9 +308,13 @@ rdatas_equal(rdata_atom_type *a, rdata_atom_type *b, int num, uint16_t type,
 	int k, start, end;
 	start = 0;
 	end = num;
+	/**
+	 * SOA RDATA comparisons in XFR are more lenient,
+	 * only serial rdata is checked.
+	 **/
 	if (type == TYPE_SOA) {
 		start = 2;
-		end = 2;
+		end = 3;
 	}
 	for(k = start; k < end; k++)
 	{
@@ -447,8 +456,8 @@ nsec3_delete_rr_trigger(namedb_type* db, rr_type* rr, zone_type* zone,
 	else if(rr->type == TYPE_NSEC3PARAM && rr == zone->nsec3_param) {
 		/* clear trees, wipe hashes, wipe precompile */
 		nsec3_clear_precompile(db, zone);
-		/* pick up new nsec3param from udb */
-		nsec3_find_zone_param(db, zone, udbz);
+		/* pick up new nsec3param (from udb, or avoid deleted rr) */
+		nsec3_find_zone_param(db, zone, udbz, rr);
 		/* if no more NSEC3, done */
 		if(!zone->nsec3_param)
 			return;
@@ -550,7 +559,7 @@ nsec3_add_rr_trigger(namedb_type* db, rr_type* rr, zone_type* zone,
 		prehash_add(db->domains, rr->owner);
 	} else if(!zone->nsec3_param && rr->type == TYPE_NSEC3PARAM) {
 		/* see if this means NSEC3 chain can be used */
-		nsec3_find_zone_param(db, zone, udbz);
+		nsec3_find_zone_param(db, zone, udbz, NULL);
 		if(!zone->nsec3_param)
 			return;
 		nsec3_zone_trees_create(db->region, zone);
@@ -660,7 +669,8 @@ delete_RR(namedb_type* db, const dname_type* dname,
 			return 1; /* not fatal error */
 		}
 		/* delete the normalized RR from the udb */
-		udb_del_rr(db->udb, udbz, &rrset->rrs[rrnum]);
+		if(db->udb)
+			udb_del_rr(db->udb, udbz, &rrset->rrs[rrnum]);
 #ifdef NSEC3
 		/* process triggers for RR deletions */
 		nsec3_delete_rr_trigger(db, &rrset->rrs[rrnum], zone, udbz);
@@ -671,6 +681,8 @@ delete_RR(namedb_type* db, const dname_type* dname,
 		if(rrset->rr_count == 1) {
 			/* delete entire rrset */
 			rrset_delete(db, domain, rrset);
+			/* check if domain is now nonexisting (or parents) */
+			rrset_zero_nonexist_check(domain);
 #ifdef NSEC3
 			/* cleanup nsec3 */
 			nsec3_delete_rrset_trigger(db, domain, zone, type);
@@ -812,9 +824,11 @@ add_RR(namedb_type* db, const dname_type* dname,
 	}
 
 	/* write the just-normalized RR to the udb */
-	if(!udb_write_rr(db->udb, udbz, &rrset->rrs[rrset->rr_count - 1])) {
-		log_msg(LOG_ERR, "could not add RR to nsd.db, disk-space?");
-		return 0;
+	if(db->udb) {
+		if(!udb_write_rr(db->udb, udbz, &rrset->rrs[rrset->rr_count - 1])) {
+			log_msg(LOG_ERR, "could not add RR to nsd.db, disk-space?");
+			return 0;
+		}
 	}
 #ifdef NSEC3
 	if(rrset_added) {
@@ -876,6 +890,7 @@ delete_zone_rrs(namedb_type* db, zone_type* zone)
 {
 	rrset_type *rrset;
 	domain_type *domain = zone->apex, *next;
+	int nonexist_check = 0;
 	/* go through entire tree below the zone apex (incl subzones) */
 	while(domain && domain_is_subdomain(domain, zone->apex))
 	{
@@ -887,6 +902,9 @@ delete_zone_rrs(namedb_type* db, zone_type* zone)
 			rrset_lower_usage(db, rrset);
 			/* rrset del does not delete our domain(yet) */
 			rrset_delete(db, domain, rrset);
+			/* no rrset_zero_nonexist_check, do that later */
+			if(domain->rrsets == 0)
+				nonexist_check = 1;
 		}
 		/* the delete upcoming could delete parents, but nothing next
 		 * or after the domain so store next ptr */
@@ -896,10 +914,27 @@ delete_zone_rrs(namedb_type* db, zone_type* zone)
 		domain = next;
 	}
 
+	/* check if data deleteions have created nonexisting domain entries,
+	 * but after deleting domains so the checks are faster */
+	if(nonexist_check) {
+		DEBUG(DEBUG_XFRD, 1, (LOG_INFO, "axfrdel: zero rrset check"));
+		domain = zone->apex;
+		while(domain && domain_is_subdomain(domain, zone->apex))
+		{
+			/* the interesting domains should be existing==1
+			 * and rrsets==0, speeding up out processing of
+			 * sub-zones, since we only spuriously check empty
+			 * nonterminals */
+			if(domain->is_existing)
+				rrset_zero_nonexist_check(domain);
+			domain = domain_next(domain);
+		}
+	}
+
 	DEBUG(DEBUG_XFRD, 1, (LOG_INFO, "axfrdel: recyclebin holds %lu bytes",
 		(unsigned long) region_get_recycle_size(db->region)));
 #ifndef NDEBUG
-	if(nsd_debug_level >= 1)
+	if(nsd_debug_level >= 2)
 		region_log_stats(db->region);
 #endif
 
@@ -1087,7 +1122,8 @@ apply_ixfr(namedb_type* db, FILE *in, const char* zone, uint32_t serialno,
 			nsec3_hash_tree_clear(zone_db);
 #endif
 			delete_zone_rrs(db, zone_db);
-			udb_zone_clear(db->udb, udbz);
+			if(db->udb)
+				udb_zone_clear(db->udb, udbz);
 #ifdef NSEC3
 			nsec3_clear_precompile(db, zone_db);
 			zone_db->nsec3_param = NULL;
@@ -1117,7 +1153,8 @@ apply_ixfr(namedb_type* db, FILE *in, const char* zone, uint32_t serialno,
 				nsec3_hash_tree_clear(zone_db);
 #endif
 				delete_zone_rrs(db, zone_db);
-				udb_zone_clear(db->udb, udbz);
+				if(db->udb)
+					udb_zone_clear(db->udb, udbz);
 #ifdef NSEC3
 				nsec3_clear_precompile(db, zone_db);
 				zone_db->nsec3_param = NULL;
@@ -1277,33 +1314,35 @@ apply_ixfr_for_zone(nsd_type* nsd, zone_type* zonedb, FILE* in,
 		udb_ptr z;
 
 		DEBUG(DEBUG_XFRD,1, (LOG_INFO, "processing xfr: %s", zone_buf));
-		if(udb_base_get_userflags(nsd->db->udb) != 0) {
-			log_msg(LOG_ERR, "database corrupted, cannot update");
-			xfrd_unlink_xfrfile(nsd, xfrfilenr);
-			exit(1);
-		}
-		/* all parts were checked by xfrd before commit */
-		if(!udb_zone_search(nsd->db->udb, &z, dname_name(apex),
-			apex->name_size)) {
-			/* create it */
-			if(!udb_zone_create(nsd->db->udb, &z, dname_name(apex),
+		if(nsd->db->udb) {
+			if(udb_base_get_userflags(nsd->db->udb) != 0) {
+				log_msg(LOG_ERR, "database corrupted, cannot update");
+				xfrd_unlink_xfrfile(nsd, xfrfilenr);
+				exit(1);
+			}
+			/* all parts were checked by xfrd before commit */
+			if(!udb_zone_search(nsd->db->udb, &z, dname_name(apex),
 				apex->name_size)) {
-				/* out of disk space perhaps */
-				log_msg(LOG_ERR, "could not udb_create_zone "
-					"%s, disk space full?", log_buf);
-				return 0;
+				/* create it */
+				if(!udb_zone_create(nsd->db->udb, &z, dname_name(apex),
+					apex->name_size)) {
+					/* out of disk space perhaps */
+					log_msg(LOG_ERR, "could not udb_create_zone "
+						"%s, disk space full?", log_buf);
+					return 0;
+				}
 			}
+			/* set the udb dirty until we are finished applying changes */
+			udb_base_set_userflags(nsd->db->udb, 1);
 		}
-		/* set the udb dirty until we are finished applying changes */
-		udb_base_set_userflags(nsd->db->udb, 1);
 		/* read and apply all of the parts */
 		for(i=0; i<num_parts; i++) {
 			int ret;
 			DEBUG(DEBUG_XFRD,2, (LOG_INFO, "processing xfr: apply part %d", (int)i));
 			ret = apply_ixfr(nsd->db, in, zone_buf, new_serial, opt,
 				i, num_parts, &is_axfr, &delete_mode,
-				&rr_count, &z, &zonedb, patname_buf, &num_bytes,
-				&softfail);
+				&rr_count, (nsd->db->udb?&z:NULL), &zonedb,
+				patname_buf, &num_bytes, &softfail);
 			if(ret == 0) {
 				log_msg(LOG_ERR, "bad ixfr packet part %d in diff file for %s", (int)i, zone_buf);
 				xfrd_unlink_xfrfile(nsd, xfrfilenr);
@@ -1313,7 +1352,8 @@ apply_ixfr_for_zone(nsd_type* nsd, zone_type* zonedb, FILE* in,
 				break;
 			}
 		}
-		udb_base_set_userflags(nsd->db->udb, 0);
+		if(nsd->db->udb)
+			udb_base_set_userflags(nsd->db->udb, 0);
 		/* read the final log_str: but do not fail on it */
 		if(!diff_read_str(in, log_buf, sizeof(log_buf))) {
 			log_msg(LOG_ERR, "could not read log for transfer %s",
@@ -1324,11 +1364,23 @@ apply_ixfr_for_zone(nsd_type* nsd, zone_type* zonedb, FILE* in,
 		if(zonedb) prehash_zone(nsd->db, zonedb);
 #endif /* NSEC3 */
 		zonedb->is_changed = 1;
-		ZONE(&z)->is_changed = 1;
-		ZONE(&z)->mtime = time_end_0;
-		udb_zone_set_log_str(nsd->db->udb, &z, log_buf);
-		udb_zone_set_file_str(nsd->db->udb, &z, NULL);
-		udb_ptr_unlink(&z, nsd->db->udb);
+		if(nsd->db->udb) {
+			ZONE(&z)->is_changed = 1;
+			ZONE(&z)->mtime = time_end_0;
+			udb_zone_set_log_str(nsd->db->udb, &z, log_buf);
+			udb_zone_set_file_str(nsd->db->udb, &z, NULL);
+			udb_ptr_unlink(&z, nsd->db->udb);
+		} else {
+			zonedb->mtime = time_end_0;
+			if(zonedb->logstr)
+				region_recycle(nsd->db->region, zonedb->logstr,
+					strlen(zonedb->logstr)+1);
+			zonedb->logstr = region_strdup(nsd->db->region, log_buf);
+			if(zonedb->filename)
+				region_recycle(nsd->db->region, zonedb->filename,
+					strlen(zonedb->filename)+1);
+			zonedb->filename = NULL;
+		}
 		if(softfail && taskudb && !is_axfr) {
 			log_msg(LOG_ERR, "Failed to apply IXFR cleanly "
 				"(deletes nonexistent RRs, adds existing RRs). "
@@ -1809,7 +1861,6 @@ task_process_add_zone(struct nsd* nsd, udb_base* udb, udb_ptr* last_task,
 static void
 task_process_del_zone(struct nsd* nsd, struct task_list_d* task)
 {
-	udb_ptr udbz;
 	zone_type* zone;
 	zone_options_t* zopt;
 	DEBUG(DEBUG_IPC,1, (LOG_INFO, "delzone task %s", dname_to_string(
@@ -1822,10 +1873,13 @@ task_process_del_zone(struct nsd* nsd, struct task_list_d* task)
 	nsec3_hash_tree_clear(zone);
 #endif
 	delete_zone_rrs(nsd->db, zone);
-	if(udb_zone_search(nsd->db->udb, &udbz, dname_name(task->zname),
-		task->zname->name_size)) {
-		udb_zone_delete(nsd->db->udb, &udbz);
-		udb_ptr_unlink(&udbz, nsd->db->udb);
+	if(nsd->db->udb) {
+		udb_ptr udbz;
+		if(udb_zone_search(nsd->db->udb, &udbz, dname_name(task->zname),
+			task->zname->name_size)) {
+			udb_zone_delete(nsd->db->udb, &udbz);
+			udb_ptr_unlink(&udbz, nsd->db->udb);
+		}
 	}
 #ifdef NSEC3
 	nsec3_clear_precompile(nsd->db, zone);
@@ -1910,6 +1964,7 @@ task_process_apply_xfr(struct nsd* nsd, udb_base* udb, udb_ptr *last_task,
 	if(!zone) {
 		/* assume the zone has been deleted and a zone transfer was
 		 * still waiting to be processed */
+		xfrd_unlink_xfrfile(nsd, TASKLIST(task)->yesno);
 		return;
 	}
 	/* apply the XFR */
@@ -1919,6 +1974,7 @@ task_process_apply_xfr(struct nsd* nsd, udb_base* udb, udb_ptr *last_task,
 		/* could not open file to update */
 		/* there is no reply to xfrd failed-update,
 		 * because xfrd has a scan for apply-failures. */
+		xfrd_unlink_xfrfile(nsd, TASKLIST(task)->yesno);
 		return;
 	}
 	/* read and apply zone transfer */
diff --git a/usr.sbin/nsd/ipc.c b/usr.sbin/nsd/ipc.c
index 141b0f3a83d..e1bf2d35b92 100644
--- a/usr.sbin/nsd/ipc.c
+++ b/usr.sbin/nsd/ipc.c
@@ -224,10 +224,23 @@ child_is_done(struct nsd* nsd, int fd)
 	for(i=0; i<nsd->child_count; ++i)
 		if(nsd->children[i].child_fd == fd) {
 			nsd->children[i].child_fd = -1;
-			nsd->children[i].has_exited = 1;
 			nsd->children[i].handler->fd = -1;
-			DEBUG(DEBUG_IPC,1, (LOG_INFO, "server %d is done",
-				(int)nsd->children[i].pid));
+			if(nsd->children[i].need_to_exit) {
+				DEBUG(DEBUG_IPC,1, (LOG_INFO, "server %d is done",
+					(int)nsd->children[i].pid));
+				nsd->children[i].has_exited = 1;
+			} else {
+				log_msg(LOG_WARNING,
+				       "server %d died unexpectedly, restarting",
+				       (int)nsd->children[i].pid);
+				/* this child is now going to be re-forked as
+				 * a subprocess of this server-main, and if a
+				 * reload is in progress the other children
+				 * are subprocesses of reload.  Until the
+				 * reload is done and they are all reforked. */
+				nsd->children[i].pid = -1;
+				nsd->restart_children = 1;
+			}
 		}
 	parent_check_all_children_exited(nsd);
 }
@@ -480,6 +493,7 @@ parent_handle_reload_command(netio_type *ATTR_UNUSED(netio),
 			handler->fd = -1;
 		}
 		log_msg(LOG_ERR, "handle_reload_cmd: reload closed cmd channel");
+		nsd->reload_failed = 1;
 		return;
 	}
 	switch (mode) {
diff --git a/usr.sbin/nsd/namedb.h b/usr.sbin/nsd/namedb.h
index 34de36be6fb..76f78dfa648 100644
--- a/usr.sbin/nsd/namedb.h
+++ b/usr.sbin/nsd/namedb.h
@@ -125,6 +125,9 @@ struct zone
 	rbtree_t* dshashtree; /* tree, ds-parent-hash domains */
 #endif
 	struct zone_options* opts;
+	char*        filename; /* set if read from file, which file */
+	char*        logstr; /* set for zone xfer, the log string */
+	time_t       mtime; /* time of last modification */
 	unsigned     is_secure : 1; /* zone uses DNSSEC */
 	unsigned     is_ok : 1; /* zone has not expired. */
 	unsigned     is_changed : 1; /* zone was changed by AXFR */
diff --git a/usr.sbin/nsd/nsd-checkzone.8.in b/usr.sbin/nsd/nsd-checkzone.8.in
new file mode 100644
index 00000000000..1ab76bc5e2f
--- /dev/null
+++ b/usr.sbin/nsd/nsd-checkzone.8.in
@@ -0,0 +1,33 @@
+.TH "nsd\-checkzone" "8" "@date@" "NLnet Labs" "nsd @version@"
+.\" Copyright (c) 2014, NLnet Labs. All rights reserved.
+.\" See LICENSE for the license.
+.SH "NAME"
+.B nsd\-checkzone
+\- NSD zone file syntax checker.
+.SH "SYNOPSIS"
+.B nsd\-checkzone
+.RB [ \-h ]
+.I zonename
+.I zonefile
+.SH "DESCRIPTION"
+.B nsd\-checkzone
+reads a DNS zone file and checks it for errors.  It prints errors to
+stderr.  On failure it exits with nonzero exit status.
+.P
+This is used to check files before feeding them to the nsd(8) daemon.
+.SH "OPTIONS"
+.TP
+.B \-h
+Print usage help information and exit.
+.TP
+.I zonename
+The name of the zone to check, eg. "example.com".
+.TP
+.I zonefile
+The file to read, eg. "zones/example.com.zone.signed".
+.SH "SEE ALSO"
+\fInsd\fR(8), \fInsd-checkconf\fR(8)
+.SH "AUTHORS"
+.B NSD
+was written by NLnet Labs and RIPE NCC joint team. Please see
+CREDITS file in the distribution for further details.
diff --git a/usr.sbin/nsd/nsd-checkzone.c b/usr.sbin/nsd/nsd-checkzone.c
new file mode 100644
index 00000000000..95bb4e72836
--- /dev/null
+++ b/usr.sbin/nsd/nsd-checkzone.c
@@ -0,0 +1,147 @@
+/*
+ * nsd-checkzone.c -- nsd-checkzone(8) checks zones for syntax errors
+ *
+ * Copyright (c) 2013, NLnet Labs. All rights reserved.
+ *
+ * See LICENSE for the license.
+ *
+ */
+
+#include "config.h"
+
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <time.h>
+#include <unistd.h>
+#include <errno.h>
+
+#include "nsd.h"
+#include "options.h"
+#include "util.h"
+#include "zonec.h"
+
+static void error(const char *format, ...) ATTR_FORMAT(printf, 1, 2);
+struct nsd nsd;
+
+/*
+ * Print the help text.
+ *
+ */
+static void
+usage (void)
+{
+	fprintf(stderr, "Usage: nsd-checkzone <zone name> <zone file>\n");
+	fprintf(stderr, "Version %s. Report bugs to <%s>.\n",
+		PACKAGE_VERSION, PACKAGE_BUGREPORT);
+}
+
+/*
+ * Something went wrong, give error messages and exit.
+ *
+ */
+static void
+error(const char *format, ...)
+{
+	va_list args;
+	va_start(args, format);
+	log_vmsg(LOG_ERR, format, args);
+	va_end(args);
+	exit(1);
+}
+
+static void
+check_zone(struct nsd* nsd, const char* name, const char* fname)
+{
+	const dname_type* dname;
+	zone_options_t* zo;
+	zone_type* zone;
+	unsigned errors;
+
+	/* init*/
+	nsd->db = namedb_open("", nsd->options);
+	dname = dname_parse(nsd->options->region, name);
+	if(!dname) {
+		/* parse failure */
+		error("cannot parse zone name '%s'", name);
+	}
+	zo = zone_options_create(nsd->options->region);
+	memset(zo, 0, sizeof(*zo));
+	zo->node.key = dname;
+	zo->name = name;
+	zone = namedb_zone_create(nsd->db, dname, zo);
+
+	/* read the zone */
+	errors = zonec_read(name, fname, zone);
+	if(errors > 0) {
+		printf("zone %s file %s has %u errors\n", name, fname, errors);
+		exit(1);
+	}
+	printf("zone %s is ok\n", name);
+	namedb_close(nsd->db);
+}
+
+/* dummy functions to link */
+int writepid(struct nsd * ATTR_UNUSED(nsd))
+{
+	        return 0;
+}
+void unlinkpid(const char * ATTR_UNUSED(file))
+{
+}
+void bind8_stats(struct nsd * ATTR_UNUSED(nsd))
+{
+}
+
+void sig_handler(int ATTR_UNUSED(sig))
+{
+}
+
+extern char *optarg;
+extern int optind;
+
+int
+main(int argc, char *argv[])
+{
+	/* Scratch variables... */
+	int c;
+	struct nsd nsd;
+	memset(&nsd, 0, sizeof(nsd));
+
+	log_init("nsd-checkzone");
+
+	/* Parse the command line... */
+	while ((c = getopt(argc, argv, "h")) != -1) {
+		switch (c) {
+		case 'h':
+			usage();
+			exit(0);
+		case '?':
+		default:
+			usage();
+			exit(1);
+		}
+	}
+	argc -= optind;
+	argv += optind;
+
+	/* Commandline parse error */
+	if (argc != 2) {
+		fprintf(stderr, "wrong number of arguments.\n");
+		usage();
+		exit(1);
+	}
+
+	nsd.options = nsd_options_create(region_create_custom(xalloc, free,
+		DEFAULT_CHUNK_SIZE, DEFAULT_LARGE_OBJECT_SIZE,
+		DEFAULT_INITIAL_CLEANUP_SIZE, 1));
+	if (verbosity == 0)
+		verbosity = nsd.options->verbosity;
+
+	check_zone(&nsd, argv[0], argv[1]);
+	region_destroy(nsd.options->region);
+	/* yylex_destroy(); but, not available in all versions of flex */
+
+	exit(0);
+}
diff --git a/usr.sbin/nsd/nsd-control.8.in b/usr.sbin/nsd/nsd-control.8.in
index b8a8701bea5..2ee2e4c8d0a 100644
--- a/usr.sbin/nsd/nsd-control.8.in
+++ b/usr.sbin/nsd/nsd-control.8.in
@@ -1,8 +1,7 @@
-.TH "nsd\-control" "8" "Mar 14, 2014" "NLnet Labs" "nsd 4.0.3"
+.TH "nsd\-control" "8" "Sep  4, 2014" "NLnet Labs" "nsd 4.1.0"
 .\" Copyright (c) 2011, NLnet Labs. All rights reserved.
 .\" See LICENSE for the license.
 .SH "NAME"
-.LP
 .B nsd\-control,
 .B nsd\-control\-setup
 \- NSD remote server control utility.
diff --git a/usr.sbin/nsd/nsd-mem.c b/usr.sbin/nsd/nsd-mem.c
index e47bb9add1e..05710c42576 100644
--- a/usr.sbin/nsd/nsd-mem.c
+++ b/usr.sbin/nsd/nsd-mem.c
@@ -26,6 +26,7 @@
 #include "util.h"
 
 static void error(const char *format, ...) ATTR_FORMAT(printf, 1, 2);
+struct nsd nsd;
 
 /*
  * Print the help text.
@@ -104,9 +105,11 @@ account_zone(struct namedb* db, struct zone_mem* zmem)
 {
 	zmem->data = region_get_mem(db->region);
 	zmem->data_unused = region_get_mem_unused(db->region);
-	zmem->udb_data = (size_t)db->udb->alloc->disk->stat_data;
-	zmem->udb_overhead = (size_t)(db->udb->alloc->disk->stat_alloc -
-		db->udb->alloc->disk->stat_data);
+	if(db->udb) {
+		zmem->udb_data = (size_t)db->udb->alloc->disk->stat_data;
+		zmem->udb_overhead = (size_t)(db->udb->alloc->disk->stat_alloc -
+			db->udb->alloc->disk->stat_data);
+	}
 	zmem->domaincount = db->domains->nametree->count;
 }
 
@@ -239,7 +242,9 @@ check_mem(nsd_options_t* opt)
 	char df[512];
 	memset(&totmem, 0, sizeof(totmem));
 	snprintf(tf, sizeof(tf), "./nsd-mem-task-%u.db", (unsigned)getpid());
-	snprintf(df, sizeof(df), "./nsd-mem-db-%u.db", (unsigned)getpid());
+	if(opt->database == NULL || opt->database[0] == 0)
+		df[0] = 0;
+	else snprintf(df, sizeof(df), "./nsd-mem-db-%u.db", (unsigned)getpid());
 
 	/* read all zones and account memory */
 	RBTREE_FOR(zo, zone_options_t*, opt->zone_options) {
@@ -252,10 +257,12 @@ check_mem(nsd_options_t* opt)
 	print_tot_mem(&totmem);
 
 	/* final advice */
-	printf("\nFinal advice estimate:\n");
-	printf("(The partial mmap causes reload&AXFR to take longer(disk access))\n");
-	pretty_mem(totmem.ram + totmem.disk, "data and big mmap");
-	pretty_mem(totmem.ram + totmem.disk/6, "data and partial mmap");
+	if(opt->database != NULL && opt->database[0] != 0) {
+		printf("\nFinal advice estimate:\n");
+		printf("(The partial mmap causes reload&AXFR to take longer(disk access))\n");
+		pretty_mem(totmem.ram + totmem.disk, "data and big mmap");
+		pretty_mem(totmem.ram + totmem.disk/6, "data and partial mmap");
+	}
 }
 
 /* dummy functions to link */
diff --git a/usr.sbin/nsd/nsd.conf.sample.in b/usr.sbin/nsd/nsd.conf.sample.in
index aa0ab38a9fc..d6154526c07 100644
--- a/usr.sbin/nsd/nsd.conf.sample.in
+++ b/usr.sbin/nsd/nsd.conf.sample.in
@@ -8,7 +8,7 @@
 
 # This is a comment.
 # Sample configuration file
-# include: "file" # include that file's text over here.
+# include: "file" # include that file's text over here.  Globbed, "*.conf"
 
 # options for the nsd server
 server:
@@ -55,6 +55,7 @@ server:
 	# zonelistfile: "@zonelistfile@"
 
 	# the database to use
+	# if set to "" then no disk-database is used, less memory usage.
 	# database: "@dbfile@"
 
 	# log messages to file. Default to stderr and syslog (with
@@ -66,7 +67,7 @@ server:
 
 	# The file where secondary zone refresh and expire timeouts are kept.
 	# If you delete this file, all secondary zones are forced to be 
-	# 'refreshing' (as if nsd got a notify).
+	# 'refreshing' (as if nsd got a notify).  Set to "" to disable.
 	# xfrdfile: "@xfrdfile@"
 
 	# The directory where zone transfers are stored, in a subdir of it.
@@ -98,13 +99,24 @@ server:
 	# ipv6-edns-size: 4096
 
 	# statistics are produced every number of seconds. Prints to log.
+	# Default is 0, meaning no statistics are produced.
 	# statistics: 3600
 
 	# Number of seconds between reloads triggered by xfrd.
 	# xfrd-reload-timeout: 1
+	
+	# log timestamp in ascii (y-m-d h:m:s.msec), yes is default.
+	# log-time-ascii: yes
+
+	# round robin rotation of records in the answer.
+	# round-robin: no
 
 	# check mtime of all zone files on start and sighup
 	# zonefiles-check: yes
+	
+	# write changed zonefiles to disk, every N seconds.
+	# default is 0(disabled) or 3600(if database is "").
+	# zonefiles-write: 3600
 
 	# RRLconfig
 	# Response Rate Limiting, size of the hashtable. Default 1000000.
diff --git a/usr.sbin/nsd/nsd.h b/usr.sbin/nsd/nsd.h
index 955fc4fbae2..14e2dae55f4 100644
--- a/usr.sbin/nsd/nsd.h
+++ b/usr.sbin/nsd/nsd.h
@@ -158,6 +158,8 @@ struct	nsd
 
 	size_t            child_count;
 	struct nsd_child *children;
+	int	restart_children;
+	int	reload_failed;
 
 	/* NULL if this is the parent process. */
 	struct nsd_child *this_child;
@@ -224,6 +226,8 @@ struct	nsd
 	struct nsd_options* options;
 };
 
+extern struct nsd nsd;
+
 /* nsd.c */
 pid_t readpid(const char *file);
 int writepid(struct nsd *nsd);
diff --git a/usr.sbin/nsd/nsec3.h b/usr.sbin/nsd/nsec3.h
index 96c4367ff33..663776f220b 100644
--- a/usr.sbin/nsd/nsec3.h
+++ b/usr.sbin/nsd/nsec3.h
@@ -92,7 +92,7 @@ int nsec3_condition_hash(struct domain* d, struct zone* z);
 int nsec3_condition_dshash(struct domain* d, struct zone* z);
 /* set nsec3param for this zone or NULL if no NSEC3 available */
 void nsec3_find_zone_param(struct namedb* db, struct zone* zone,
-	struct udb_ptr* z);
+	struct udb_ptr* z, struct rr* avoid_rr);
 /* hash domain and wcchild, and lookup nsec3 in tree, and precompile */
 void nsec3_precompile_domain(struct namedb* db, struct domain* domain,
 	struct zone* zone, struct region* tmpregion);
diff --git a/usr.sbin/nsd/options.c b/usr.sbin/nsd/options.c
index 060e5eba2bc..c7ef346ec7a 100644
--- a/usr.sbin/nsd/options.c
+++ b/usr.sbin/nsd/options.c
@@ -53,6 +53,8 @@ nsd_options_create(region_type* region)
 	opt->identity = 0;
 	opt->nsid = 0;
 	opt->logfile = 0;
+	opt->log_time_ascii = 1;
+	opt->round_robin = 0; /* also packet.h::round_robin */
 	opt->server_count = 1;
 	opt->tcp_count = 100;
 	opt->tcp_query_count = 0;
@@ -78,6 +80,9 @@ nsd_options_create(region_type* region)
 	opt->rrl_whitelist_ratelimit = RRL_WLIST_LIMIT/2;
 #endif
 	opt->zonefiles_check = 1;
+	if(opt->database == NULL || opt->database[0] == 0)
+		opt->zonefiles_write = ZONEFILES_WRITE_INTERVAL;
+	else	opt->zonefiles_write = 0;
 	opt->xfrd_reload_timeout = 1;
 	opt->control_enable = 0;
 	opt->control_interface = NULL;
@@ -128,7 +133,7 @@ parse_options_file(nsd_options_t* opt, const char* file,
 	}
 	cfg_parser->err = err;
 	cfg_parser->err_arg = err_arg;
-	cfg_parser->filename = file;
+	cfg_parser->filename = (char*)file;
 	cfg_parser->line = 1;
 	cfg_parser->errors = 0;
 	cfg_parser->server_settings_seen = 0;
@@ -238,12 +243,12 @@ parse_options_file(nsd_options_t* opt, const char* file,
 		if(err) {
 			char m[MAXSYSLOGMSGLEN];
 			snprintf(m, sizeof(m), "read %s failed: %d errors in "
-				"configuration file\n", cfg_parser->filename,
+				"configuration file\n", file,
 				cfg_parser->errors);
 			err(err_arg, m);
 		} else {
 			fprintf(stderr, "read %s failed: %d errors in "
-				"configuration file\n", cfg_parser->filename,
+				"configuration file\n", file,
 				cfg_parser->errors);
 		}
 		return 0;
diff --git a/usr.sbin/nsd/options.h b/usr.sbin/nsd/options.h
index 062ac2842c8..f36b11f242f 100644
--- a/usr.sbin/nsd/options.h
+++ b/usr.sbin/nsd/options.h
@@ -80,6 +80,9 @@ struct nsd_options {
 	const char* nsid;
 	int xfrd_reload_timeout;
 	int zonefiles_check;
+	int zonefiles_write;
+	int log_time_ascii;
+	int round_robin;
 
         /** remote control section. enable toggle. */
 	int control_enable;
@@ -226,11 +229,14 @@ struct zonelist_bucket {
 	struct zonelist_free* list;
 };
 
+/* default zonefile write interval if database is "", in seconds */
+#define ZONEFILES_WRITE_INTERVAL 3600
+
 /*
  * Used during options parsing
  */
 struct config_parser_state {
-	const char* filename;
+	char* filename;
 	const char* chroot;
 	int line;
 	int errors;
diff --git a/usr.sbin/nsd/packet.c b/usr.sbin/nsd/packet.c
index 72ae4cedf03..08d91b269d7 100644
--- a/usr.sbin/nsd/packet.c
+++ b/usr.sbin/nsd/packet.c
@@ -15,6 +15,8 @@
 #include "query.h"
 #include "rdata.h"
 
+int round_robin = 0;
+
 static void
 encode_dname(query_type *q, domain_type *domain)
 {
@@ -130,13 +132,30 @@ packet_encode_rrset(query_type *query,
 				section == AUTHORITY_SECTION ||
 				section == OPTIONAL_AUTHORITY_SECTION);
 #endif
+	static int round_robin_off = 0;
+	int do_robin = (round_robin && section == ANSWER_SECTION &&
+		query->qtype != TYPE_AXFR && query->qtype != TYPE_IXFR);
+	uint16_t start;
 	rrset_type *rrsig;
 
 	assert(rrset->rr_count > 0);
 
 	truncation_mark = buffer_position(query->packet);
 
-	for (i = 0; i < rrset->rr_count; ++i) {
+	if(do_robin && rrset->rr_count)
+		start = (uint16_t)(round_robin_off++ % rrset->rr_count);
+	else	start = 0;
+	for (i = start; i < rrset->rr_count; ++i) {
+		if (packet_encode_rr(query, owner, &rrset->rrs[i],
+			rrset->rrs[i].ttl)) {
+			++added;
+		} else {
+			all_added = 0;
+			start = 0;
+			break;
+		}
+	}
+	for (i = 0; i < start; ++i) {
 		if (packet_encode_rr(query, owner, &rrset->rrs[i],
 			rrset->rrs[i].ttl)) {
 			++added;
diff --git a/usr.sbin/nsd/packet.h b/usr.sbin/nsd/packet.h
index 47558394e0d..b07c6204c5c 100644
--- a/usr.sbin/nsd/packet.h
+++ b/usr.sbin/nsd/packet.h
@@ -143,6 +143,9 @@ struct query;
 #define IPV4_MINIMAL_RESPONSE_SIZE 1480	 /* Recommended minimal edns size for IPv4 */
 #define IPV6_MINIMAL_RESPONSE_SIZE 1220	 /* Recommended minimal edns size for IPv6 */
 
+/* use round robin rotation */
+extern int round_robin;
+
 /*
  * Encode RR with OWNER as owner name into QUERY.  Returns the number
  * of RRs successfully encoded.
diff --git a/usr.sbin/nsd/remote.c b/usr.sbin/nsd/remote.c
index 4b3659878f9..47664294d2d 100644
--- a/usr.sbin/nsd/remote.c
+++ b/usr.sbin/nsd/remote.c
@@ -512,7 +512,11 @@ static void
 remote_accept_callback(int fd, short event, void* arg)
 {
 	struct daemon_remote *rc = (struct daemon_remote*)arg;
+#ifdef INET6
 	struct sockaddr_storage addr;
+#else
+	struct sockaddr_in addr;
+#endif
 	socklen_t addrlen;
 	int newfd;
 	struct rc_state* n;
@@ -869,6 +873,7 @@ force_transfer_zone(xfrd_zone_t* zone)
 	/* pretend we not longer have it and force any
 	 * zone to be downloaded (even same serial, w AXFR) */
 	zone->soa_disk_acquired = 0;
+	zone->soa_nsd_acquired = 0;
 	xfrd_handle_notify_and_start_xfr(zone, NULL);
 }
 
@@ -1080,6 +1085,7 @@ do_stats(struct daemon_remote* rc, int peek, struct rc_state* rs)
 static void
 do_addzone(SSL* ssl, xfrd_state_t* xfrd, char* arg)
 {
+	const dname_type* dname;
 	zone_options_t* zopt;
 	char* arg2 = NULL;
 	if(!find_arg2(ssl, arg, &arg2))
@@ -1095,10 +1101,28 @@ do_addzone(SSL* ssl, xfrd_state_t* xfrd, char* arg)
 
 	/* check that the pattern exists */
 	if(!rbtree_search(xfrd->nsd->options->patterns, arg2)) {
-		(void)ssl_printf(ssl, "error pattern does not exist\n");
+		(void)ssl_printf(ssl, "error pattern %s does not exist\n",
+			arg2);
 		return;
 	}
 
+	dname = dname_parse(xfrd->region, arg);
+	if(!dname) {
+		(void)ssl_printf(ssl, "error cannot parse zone name\n");
+		return;
+	}
+
+	/* see if zone is a duplicate */
+	if( (zopt=zone_options_find(xfrd->nsd->options, dname)) ) {
+		region_recycle(xfrd->region, (void*)dname,
+			dname_total_size(dname));
+		(void)ssl_printf(ssl, "zone %s already exists\n", arg);
+		send_ok(ssl); /* a nop operation */
+		return;
+	}
+	region_recycle(xfrd->region, (void*)dname, dname_total_size(dname));
+	dname = NULL;
+
 	/* add to zonelist and adds to config in memory */
 	zopt = zone_list_add(xfrd->nsd->options, arg, arg2);
 	if(!zopt) {
diff --git a/usr.sbin/nsd/udb.c b/usr.sbin/nsd/udb.c
index c3b1de0347c..9b31fcfc310 100644
--- a/usr.sbin/nsd/udb.c
+++ b/usr.sbin/nsd/udb.c
@@ -168,11 +168,15 @@ udb_base_create_fd(const char* fname, int fd, udb_walk_relptr_func walkfunc,
 		goto fail;
 	}
 	udb->base_size = (size_t)g.fsize;
+#ifdef HAVE_MMAP
 	/* note the size_t casts must be there for portability, on some
 	 * systems the layout of memory is otherwise broken. */
 	udb->base = mmap(NULL, (size_t)udb->base_size,
 		(int)PROT_READ|PROT_WRITE, (int)MAP_SHARED,
 		(int)udb->fd, (off_t)0);
+#else
+	udb->base = MAP_FAILED; errno = ENOSYS;
+#endif
 	if(udb->base == MAP_FAILED) {
 		udb->base = NULL;
 		log_msg(LOG_ERR, "mmap(size %u) error: %s",
@@ -297,7 +301,9 @@ udb_base_shrink(udb_base* udb, uint64_t nsize)
 	udb->glob_data->fsize = nsize;
 	/* sync, does not *seem* to be required on Linux, but it is
 	   certainly required on OpenBSD.  Otherwise changed data is lost. */
+#ifdef HAVE_MMAP
 	msync(udb->base, udb->base_size, MS_ASYNC);
+#endif
 	if(ftruncate(udb->fd, (off_t)nsize) != 0) {
 		log_msg(LOG_ERR, "%s: ftruncate(%u) %s", udb->fname,
 			(unsigned)nsize, strerror(errno));
@@ -320,9 +326,11 @@ void udb_base_close(udb_base* udb)
 		udb->fd = -1;
 	}
 	if(udb->base) {
+#ifdef HAVE_MMAP
 		if(munmap(udb->base, udb->base_size) == -1) {
 			log_msg(LOG_ERR, "munmap: %s", strerror(errno));
 		}
+#endif
 		udb->base = NULL;
 	}
 }
@@ -354,10 +362,15 @@ void udb_base_free_keep_mmap(udb_base* udb)
 
 void udb_base_sync(udb_base* udb, int wait)
 {
+	if(!udb) return;
+#ifdef HAVE_MMAP
 	if(msync(udb->base, udb->base_size, wait?MS_SYNC:MS_ASYNC) != 0) {
 		log_msg(LOG_ERR, "msync(%s) error %s",
 			udb->fname, strerror(errno));
 	}
+#else
+	(void)wait;
+#endif
 }
 
 /** hash a chunk pointer */
@@ -499,6 +512,7 @@ uint8_t udb_base_get_userflags(udb_base* udb)
 static void*
 udb_base_remap(udb_base* udb, udb_alloc* alloc, uint64_t nsize)
 {
+#ifdef HAVE_MMAP
 	void* nb;
 	/* for use with valgrind, do not use mremap, but the other version */
 #ifdef MREMAP_MAYMOVE
@@ -545,6 +559,10 @@ udb_base_remap(udb_base* udb, udb_alloc* alloc, uint64_t nsize)
 	}
 	udb->base_size = nsize;
 	return nb;
+#else /* HAVE_MMAP */
+	(void)udb; (void)alloc; (void)nsize;
+	return NULL;
+#endif /* HAVE_MMAP */
 }
 
 void
@@ -1531,7 +1549,7 @@ coagulate_and_push(void* base, udb_alloc* alloc, udb_void last, int exp,
 }
 
 /** attempt to compact the data and move free space to the end */
-static int
+int
 udb_alloc_compact(void* base, udb_alloc* alloc)
 {
 	udb_void last;
@@ -1540,6 +1558,9 @@ udb_alloc_compact(void* base, udb_alloc* alloc)
 	uint64_t at = alloc->disk->nextgrow;
 	udb_void xl_start = 0;
 	uint64_t xl_sz = 0;
+	if(alloc->udb->inhibit_compact)
+		return 1;
+	alloc->udb->useful_compact = 0;
 	while(at > alloc->udb->glob_data->hsize) {
 		/* grab last entry */
 		exp = (int)*((uint8_t*)UDB_REL(base, at-1));
@@ -1657,6 +1678,21 @@ udb_alloc_compact(void* base, udb_alloc* alloc)
 	return 1;
 }
 
+int
+udb_compact(udb_base* udb)
+{
+	if(!udb) return 1;
+	if(!udb->useful_compact) return 1;
+	DEBUG(DEBUG_DBACCESS, 1, (LOG_INFO, "Compacting database..."));
+	return udb_alloc_compact(udb->base, udb->alloc);
+}
+
+void udb_compact_inhibited(udb_base* udb, int inhibit)
+{
+	if(!udb) return;
+	udb->inhibit_compact = inhibit;
+}
+
 #ifdef UDB_CHECK
 /** check that rptrs are really zero before free */
 void udb_check_rptr_zero(void* base, udb_rel_ptr* p, void* arg)
@@ -1733,6 +1769,10 @@ int udb_alloc_free(udb_alloc* alloc, udb_void r, size_t sz)
 	if(fp->exp == UDB_EXP_XL) {
 		udb_free_xl(base, alloc, f, (udb_xl_chunk_d*)fp, sz);
 		/* compact */
+		if(alloc->udb->inhibit_compact) {
+			alloc->udb->useful_compact = 1;
+			return 1;
+		}
 		return udb_alloc_compact(base, alloc);
 	}
 	/* it is a regular chunk of 2**exp size */
@@ -1776,6 +1816,10 @@ int udb_alloc_free(udb_alloc* alloc, udb_void r, size_t sz)
 	}
 	alloc->udb->glob_data->dirty_alloc = udb_dirty_clean;
 	/* compact */
+	if(alloc->udb->inhibit_compact) {
+		alloc->udb->useful_compact = 1;
+		return 1;
+	}
 	return udb_alloc_compact(base, alloc);
 }
 
diff --git a/usr.sbin/nsd/udb.h b/usr.sbin/nsd/udb.h
index 691b78c817f..049760dd961 100644
--- a/usr.sbin/nsd/udb.h
+++ b/usr.sbin/nsd/udb.h
@@ -204,6 +204,11 @@ struct udb_base {
 	udb_walk_relptr_func* walkfunc;
 	/** user data for walkfunc */
 	void* walkarg;
+
+	/** compaction is inhibited */
+	int inhibit_compact;
+	/** compaction is useful; deletions performed. */
+	int useful_compact;
 };
 
 typedef enum udb_chunk_type udb_chunk_type;
@@ -549,6 +554,22 @@ udb_void udb_alloc_realloc(udb_alloc* alloc, udb_void r, size_t osz,
  */
 int udb_alloc_grow(udb_alloc* alloc, size_t sz, size_t num);
 
+/** 
+ * attempt to compact the data and move free space to the end
+ * can shrink the db, which calls sync on the db (for portability).
+ * @param udb: the udb base.
+ * @return 0 on failure (to remap the (possibly) changed udb base).
+ */
+int udb_compact(udb_base* udb);
+
+/** 
+ * set the udb to inhibit or uninhibit compaction.  Does not perform
+ * the compaction itself if enabled, for that call udb_compact.
+ * @param udb: the udb base
+ * @param inhibit: 0 or 1.
+ */
+void udb_compact_inhibited(udb_base* udb, int inhibit);
+
 /**
  * Set the alloc type for a newly alloced piece of data
  * @param alloc: the udb space allocator.
diff --git a/usr.sbin/nsd/xfrd.h b/usr.sbin/nsd/xfrd.h
index 83cb93fa51b..5c731f6c085 100644
--- a/usr.sbin/nsd/xfrd.h
+++ b/usr.sbin/nsd/xfrd.h
@@ -73,6 +73,14 @@ struct xfrd_state {
 	time_t reload_cmd_last_sent;
 	uint8_t can_send_reload;
 	pid_t reload_pid;
+	/* timeout for lost sigchild and reaping children */
+	struct event child_timer;
+	int child_timer_added;
+
+	/* timeout event for zonefiles_write events */
+	struct event write_timer;
+	/* set to 1 if zones have received xfrs since the last write_timer */
+	int write_zonefile_needed;
 
 	/* communication channel with server_main */
 	struct event ipc_handler;
@@ -230,7 +238,8 @@ enum xfrd_packet_result {
 extern xfrd_state_t* xfrd;
 
 /* start xfrd, new start. Pass socket to server_main. */
-void xfrd_init(int socket, struct nsd* nsd, int shortsoa, int reload_active);
+void xfrd_init(int socket, struct nsd* nsd, int shortsoa, int reload_active,
+	pid_t nsd_pid);
 
 /* add new slave zone, dname(from zone_opt) and given options */
 void xfrd_init_slave_zone(xfrd_state_t* xfrd, zone_options_t* zone_opt);
diff --git a/usr.sbin/nsd/zlexer.lex b/usr.sbin/nsd/zlexer.lex
index bcb5661ab80..56ad724e108 100644
--- a/usr.sbin/nsd/zlexer.lex
+++ b/usr.sbin/nsd/zlexer.lex
@@ -107,7 +107,9 @@ parser_pop_stringbuf(void)
 SPACE   [ \t]
 LETTER  [a-zA-Z]
 NEWLINE [\n\r]
-ZONESTR [^ \t\n\r();.\"\$]
+ZONESTR [^ \t\n\r();.\"\$]|\\.|\\\n
+CHARSTR [^ \t\n\r();.]|\\.|\\\n
+QUOTE   \"
 DOLLAR  \$
 COMMENT ;
 DOT     \.
@@ -281,7 +283,7 @@ ANY     [^\"\n\\]|\\.
 }
 
 	/* Quoted strings.  Strip leading and ending quotes.  */
-\"			{ BEGIN(quotedstring); LEXOUT(("\" ")); }
+{QUOTE}			{ BEGIN(quotedstring); LEXOUT(("\" ")); }
 <quotedstring><<EOF>> 	{
 	zc_error("EOF inside quoted string");
 	BEGIN(INITIAL);
@@ -290,14 +292,14 @@ ANY     [^\"\n\\]|\\.
 }
 <quotedstring>{ANY}*	{ LEXOUT(("STR ")); yymore(); }
 <quotedstring>\n 	{ ++parser->line; yymore(); }
-<quotedstring>\" {
+<quotedstring>{QUOTE} {
 	LEXOUT(("\" "));
 	BEGIN(INITIAL);
 	yytext[yyleng - 1] = '\0';
 	return parse_token(STR, yytext, &lexer_state);
 }
 
-({ZONESTR}|\\.|\\\n)+ {
+{ZONESTR}({CHARSTR})* {
 	/* Any allowed word.  */
 	return parse_token(STR, yytext, &lexer_state);
 }