update copyright years

Don't wait for the first response with setting the address.

Makefile.in

@@ -35,7 +35,7 @@ LDFLAGS = @LDFLAGS@
 
 EXTRA_OBJS = @EXTRA_OBJS@
 
-OBJS = array.o cmdparse.o conf.o local.o logging.o main.o memory.o \
+OBJS = array.o cmdparse.o conf.o local.o logging.o main.o memory.o quantiles.o \
        reference.o regress.o rtc.o samplefilt.o sched.o socket.o sources.o sourcestats.o \
        stubs.o smooth.o sys.o sys_null.o tempcomp.o util.o $(EXTRA_OBJS)
 

NEWS

@@ -1,3 +1,24 @@
+New in version 4.3
+==================
+
+Enhancements
+------------
+* Add local option to refclock directive to stabilise system clock
+  with more stable free-running clock (e.g. TCXO, OCXO)
+* Add maxdelayquant option to server/pool/peer directive to replace
+  maxdelaydevratio filter with long-term quantile-based filtering
+* Add selection option to log directive
+* Allow external PPS in PHC refclock without configurable pin
+* Don't accept first interleaved response to minimise error in delay
+* Don't use arc4random on Linux to avoid server performance loss
+* Improve filter option to better handle missing NTP samples
+* Improve stability with hardware timestamping and PHC refclock
+* Update seccomp filter
+
+Bug fixes
+---------
+* Fix waitsync command to reconnect when not getting response
+
 New in version 4.2
 ==================
 
@@ -16,6 +37,11 @@ Bug fixes
 ---------
 * Fix RTC support with 64-bit time_t on 32-bit Linux
 * Fix seccomp filter to work correctly with bind*device directives
+* Suppress kernel adjustments of system clock (dosynctodr) on illumos
+
+Other changes
+-------------
+* Switch Solaris support to illumos
 
 New in version 4.1
 ==================

README

@@ -28,7 +28,7 @@ What will chrony run on?
 ========================
 
 The software is known to work on Linux, FreeBSD, NetBSD, macOS and
-Solaris.  Closely related systems may work too.  Any other system will
+illumos.  Closely related systems may work too.  Any other system will
 likely require a porting exercise.
 
 How do I set it up?

candm.h

@@ -296,7 +296,8 @@ typedef struct {
   uint32_t flags;
   int32_t filter_length;
   uint32_t cert_set;
-  uint32_t reserved[2];
+  Float max_delay_quant;
+  uint32_t reserved[1];
   int32_t EOR;
 } REQ_NTP_Source;
 
@@ -733,7 +734,8 @@ typedef struct {
   uint32_t total_tx_count;
   uint32_t total_rx_count;
   uint32_t total_valid_count;
-  uint32_t reserved[4];
+  uint32_t total_good_count;
+  uint32_t reserved[3];
   int32_t EOR;
 } RPY_NTPData;
 

client.c

@@ -283,6 +283,9 @@ open_io(void)
     close_io();
   }
 
+  /* Start from the first address if called again */
+  address_index = 0;
+
   return 0;
 }
 
@@ -949,6 +952,8 @@ process_cmd_add_source(CMD_Request *msg, char *line)
           (data.params.sel_options & SRC_SELECT_REQUIRE ? REQ_ADDSRC_REQUIRE : 0));
       msg->data.ntp_source.filter_length = htonl(data.params.filter_length);
       msg->data.ntp_source.cert_set = htonl(data.params.cert_set);
+      msg->data.ntp_source.max_delay_quant =
+        UTI_FloatHostToNetwork(data.params.max_delay_quant);
       memset(msg->data.ntp_source.reserved, 0, sizeof (msg->data.ntp_source.reserved));
 
       result = 1;
@@ -2341,7 +2346,8 @@ process_cmd_ntpdata(char *line)
                  "RX timestamping : %N\n"
                  "Total TX        : %U\n"
                  "Total RX        : %U\n"
-                 "Total valid RX  : %U\n",
+                 "Total valid RX  : %U\n"
+                 "Total good RX   : %U\n",
                  UTI_IPToString(&remote_addr), (unsigned long)UTI_IPToRefid(&remote_addr),
                  ntohs(reply.data.ntp_data.remote_port),
                  UTI_IPToString(&local_addr), (unsigned long)UTI_IPToRefid(&local_addr),
@@ -2369,6 +2375,7 @@ process_cmd_ntpdata(char *line)
                  (unsigned long)ntohl(reply.data.ntp_data.total_tx_count),
                  (unsigned long)ntohl(reply.data.ntp_data.total_rx_count),
                  (unsigned long)ntohl(reply.data.ntp_data.total_valid_count),
+                 (unsigned long)ntohl(reply.data.ntp_data.total_good_count),
                  REPORT_END);
   }
 
@@ -2396,12 +2403,12 @@ process_cmd_selectdata(char *line)
   n_sources = ntohl(reply.data.n_sources.n_sources);
 
   if (verbose) {
-    printf(    "  .-- State: N - noselect, M - missing samples, d/D - large distance,\n");
+    printf(    "  . State: N - noselect, s - unsynchronised, M - missing samples,\n");
-    printf(    " /           ~ - jittery, w/W - waits for others, T - not trusted,\n");
+    printf(    " /         d/D - large distance, ~ - jittery, w/W - waits for others,\n");
-    printf(    "|            x - falseticker, P - not preferred, U - waits for update,\n");
+    printf(    "|          S - stale, O - orphan, T - not trusted, P - not preferred,\n");
-    printf(    "|            S - stale, O - orphan, + - combined, * - best.\n");
+    printf(    "|          U - waits for update,, x - falseticker, + - combined, * - best.\n");
-    printf(    "|        Effective options ------.  (N - noselect, P - prefer\n");
+    printf(    "|   Effective options   ---------.  (N - noselect, P - prefer\n");
-    printf(    "|       Configured options -.     \\  T - trust, R - require)\n");
+    printf(    "|   Configured options  ----.     \\  T - trust, R - require)\n");
     printf(    "|   Auth. enabled (Y/N) -.   \\     \\     Offset interval --.\n");
     printf(    "|                        |    |     |                       |\n");
   }
@@ -3235,44 +3242,46 @@ process_line(char *line)
   if (do_normal_submit) {
     ret = request_reply(&tx_message, &rx_message, RPY_NULL, 1);
   }
+
   fflush(stderr);
-  fflush(stdout);
+
+  if (fflush(stdout) != 0 || ferror(stdout) != 0) {
+    LOG(LOGS_ERR, "Could not write to stdout");
+
+    /* Return error for commands that print data */
+    if (!do_normal_submit)
+      return 0;
+  }
+
   return ret;
 }
 
 /* ================================================== */
 
+#define MAX_LINE_LENGTH 2048
+
 static int
 process_args(int argc, char **argv, int multi)
 {
-  int total_length, i, ret = 0;
+  char line[MAX_LINE_LENGTH];
-  char *line;
+  int i, l, ret = 0;
 
-  total_length = 0;
+  for (i = l = 0; i < argc; i++) {
-  for(i=0; i<argc; i++) {
+    l += snprintf(line + l, sizeof (line) - l, "%s ", argv[i]);
-    total_length += strlen(argv[i]) + 1;
+    if (l >= sizeof (line)) {
-  }
+      LOG(LOGS_ERR, "Command too long");
-
+      return 0;
-  line = (char *) Malloc((2 + total_length) * sizeof(char));
-
-  for (i = 0; i < argc; i++) {
-    line[0] = '\0';
-    if (multi) {
-      strcat(line, argv[i]);
-    } else {
-      for (; i < argc; i++) {
-        strcat(line, argv[i]);
-        if (i + 1 < argc)
-          strcat(line, " ");
-      }
     }
 
+    if (!multi && i + 1 < argc)
+      continue;
+
     ret = process_line(line);
     if (!ret || quit)
       break;
-  }
 
-  Free(line);
+    l = 0;
+  }
 
   return ret;
 }
@@ -3291,7 +3300,7 @@ static void
 display_gpl(void)
 {
     printf("chrony version %s\n"
-           "Copyright (C) 1997-2003, 2007, 2009-2021 Richard P. Curnow and others\n"
+           "Copyright (C) 1997-2003, 2007, 2009-2022 Richard P. Curnow and others\n"
            "chrony comes with ABSOLUTELY NO WARRANTY.  This is free software, and\n"
            "you are welcome to redistribute it under certain conditions.  See the\n"
            "GNU General Public License version 2 for details.\n\n",

clientlog.c

@@ -3,7 +3,7 @@
 
  **********************************************************************
  * Copyright (C) Richard P. Curnow  1997-2003
- * Copyright (C) Miroslav Lichvar  2009, 2015-2017
+ * Copyright (C) Miroslav Lichvar  2009, 2015-2017, 2021
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -751,7 +751,7 @@ static void
 int64_to_ntp64(uint64_t ts, NTP_int64 *ntp_ts)
 {
   ntp_ts->hi = htonl(ts >> 32);
-  ntp_ts->lo = htonl(ts & ((1ULL << 32) - 1));
+  ntp_ts->lo = htonl(ts);
 }
 
 /* ================================================== */

cmdmon.c

@@ -321,7 +321,8 @@ transmit_reply(int sock_fd, int request_length, SCK_Message *message)
 
 #if !defined(HAVE_IN_PKTINFO) && defined(IP_SENDSRCADDR)
   /* On FreeBSD a local IPv4 address cannot be specified on bound socket */
-  if (message->local_addr.ip.family == IPADDR_INET4 && (sock_fd != sock_fd4 || bound_sock_fd4))
+  if (message->addr_type == SCK_ADDR_IP && message->local_addr.ip.family == IPADDR_INET4 &&
+      (sock_fd != sock_fd4 || bound_sock_fd4))
     message->local_addr.ip.family = IPADDR_UNSPEC;
 #endif
 
@@ -756,6 +757,8 @@ handle_add_source(CMD_Request *rx_message, CMD_Reply *tx_message)
     UTI_FloatNetworkToHost(rx_message->data.ntp_source.max_delay_ratio);
   params.max_delay_dev_ratio =
     UTI_FloatNetworkToHost(rx_message->data.ntp_source.max_delay_dev_ratio);
+  params.max_delay_quant =
+    UTI_FloatNetworkToHost(rx_message->data.ntp_source.max_delay_quant);
   params.min_delay = UTI_FloatNetworkToHost(rx_message->data.ntp_source.min_delay);
   params.asymmetry = UTI_FloatNetworkToHost(rx_message->data.ntp_source.asymmetry);
   params.offset = UTI_FloatNetworkToHost(rx_message->data.ntp_source.offset);
@@ -1221,6 +1224,7 @@ handle_ntp_data(CMD_Request *rx_message, CMD_Reply *tx_message)
   tx_message->data.ntp_data.total_tx_count = htonl(report.total_tx_count);
   tx_message->data.ntp_data.total_rx_count = htonl(report.total_rx_count);
   tx_message->data.ntp_data.total_valid_count = htonl(report.total_valid_count);
+  tx_message->data.ntp_data.total_good_count = htonl(report.total_good_count);
   memset(tx_message->data.ntp_data.reserved, 0xff, sizeof (tx_message->data.ntp_data.reserved));
 }
 

cmdparse.c

@@ -3,7 +3,7 @@
 
  **********************************************************************
  * Copyright (C) Richard P. Curnow  1997-2003
- * Copyright (C) Miroslav Lichvar  2013-2014, 2016
+ * Copyright (C) Miroslav Lichvar  2013-2014, 2016, 2021
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -72,6 +72,7 @@ CPS_ParseNTPSourceAdd(char *line, CPS_NTP_Source *src)
   src->params.max_delay = SRC_DEFAULT_MAXDELAY;
   src->params.max_delay_ratio = SRC_DEFAULT_MAXDELAYRATIO;
   src->params.max_delay_dev_ratio = SRC_DEFAULT_MAXDELAYDEVRATIO;
+  src->params.max_delay_quant = 0.0;
   src->params.min_delay = 0.0;
   src->params.asymmetry = SRC_DEFAULT_ASYMMETRY;
   src->params.offset = 0.0;
@@ -140,6 +141,9 @@ CPS_ParseNTPSourceAdd(char *line, CPS_NTP_Source *src)
     } else if (!strcasecmp(cmd, "maxdelaydevratio")) {
       if (sscanf(line, "%lf%n", &src->params.max_delay_dev_ratio, &n) != 1)
         return 0;
+    } else if (!strcasecmp(cmd, "maxdelayquant")) {
+      if (sscanf(line, "%lf%n", &src->params.max_delay_quant, &n) != 1)
+        return 0;
     } else if (!strcasecmp(cmd, "maxpoll")) {
       if (sscanf(line, "%d%n", &src->params.maxpoll, &n) != 1)
         return 0;

conf.c

@@ -115,6 +115,7 @@ static int cmd_port = DEFAULT_CANDM_PORT;
 
 static int raw_measurements = 0;
 static int do_log_measurements = 0;
+static int do_log_selection = 0;
 static int do_log_statistics = 0;
 static int do_log_tracking = 0;
 static int do_log_rtc = 0;
@@ -861,7 +862,7 @@ static void
 parse_refclock(char *line)
 {
   int n, poll, dpoll, filter_length, pps_rate, min_samples, max_samples, sel_options;
-  int max_lock_age, pps_forced, stratum, tai;
+  int local, max_lock_age, pps_forced, stratum, tai;
   uint32_t ref_id, lock_ref_id;
   double offset, delay, precision, max_dispersion, pulse_width;
   char *p, *cmd, *name, *param;
@@ -871,6 +872,7 @@ parse_refclock(char *line)
   poll = 4;
   dpoll = 0;
   filter_length = 64;
+  local = 0;
   pps_forced = 0;
   pps_rate = 0;
   min_samples = SRC_DEFAULT_MINSAMPLES;
@@ -929,6 +931,9 @@ parse_refclock(char *line)
       if (sscanf(line, "%d%n", &filter_length, &n) != 1) {
         break;
       }
+    } else if (!strcasecmp(cmd, "local")) {
+      n = 0;
+      local = 1;
     } else if (!strcasecmp(cmd, "rate")) {
       if (sscanf(line, "%d%n", &pps_rate, &n) != 1)
         break;
@@ -995,6 +1000,7 @@ parse_refclock(char *line)
   refclock->driver_poll = dpoll;
   refclock->poll = poll;
   refclock->filter_length = filter_length;
+  refclock->local = local;
   refclock->pps_forced = pps_forced;
   refclock->pps_rate = pps_rate;
   refclock->min_samples = min_samples;
@@ -1027,6 +1033,8 @@ parse_log(char *line)
         raw_measurements = 1;
       } else if (!strcmp(log_name, "measurements")) {
         do_log_measurements = 1;
+      } else if (!strcmp(log_name, "selection")) {
+        do_log_selection = 1;
       } else if (!strcmp(log_name, "statistics")) {
         do_log_statistics = 1;
       } else if (!strcmp(log_name, "tracking")) {
@@ -1924,6 +1932,14 @@ CNF_GetLogMeasurements(int *raw)
 
 /* ================================================== */
 
+int
+CNF_GetLogSelection(void)
+{
+  return do_log_selection;
+}
+
+/* ================================================== */
+
 int
 CNF_GetLogStatistics(void)
 {

conf.h

@@ -58,6 +58,7 @@ extern char *CNF_GetLogDir(void);
 extern char *CNF_GetDumpDir(void);
 extern int CNF_GetLogBanner(void);
 extern int CNF_GetLogMeasurements(int *raw);
+extern int CNF_GetLogSelection(void);
 extern int CNF_GetLogStatistics(void);
 extern int CNF_GetLogTracking(void);
 extern int CNF_GetLogRtc(void);

configure

@@ -245,6 +245,7 @@ try_lockmem=0
 feat_asyncdns=1
 feat_forcednsretry=1
 try_clock_gettime=1
+try_arc4random=1
 try_recvmmsg=1
 feat_timestamping=1
 try_timestamping=0
@@ -421,6 +422,7 @@ case $OPERATINGSYSTEM in
         try_setsched=1
         try_lockmem=1
         try_phc=1
+        try_arc4random=0
         add_def LINUX
         echo "Configuring for " $SYSTEM
     ;;
@@ -467,7 +469,7 @@ case $OPERATINGSYSTEM in
     ;;
     SunOS)
         EXTRA_OBJECTS="sys_generic.o sys_solaris.o sys_timex.o sys_posix.o"
-        LIBS="$LIBS -lsocket -lnsl -lresolv"
+        LIBS="$LIBS -lsocket -lnsl -lkvm -lelf -lresolv"
         try_setsched=1
         try_lockmem=1
         add_def SOLARIS
@@ -479,7 +481,7 @@ case $OPERATINGSYSTEM in
           add_def FEAT_PRIVDROP
           priv_ops="ADJUSTTIMEX SETTIME BINDSOCKET"
         fi
-        echo "Configuring for Solaris (" $SYSTEM "SunOS version" $VERSION ")" 
+        echo "Configuring for illumos (" $SYSTEM "SunOS version" $VERSION ")"
     ;;                                                                        
     * )
         echo "error: $SYSTEM is not supported (yet?)"
@@ -671,12 +673,12 @@ fi
 
 if [ $try_clock_gettime = "1" ]; then
   if test_code 'clock_gettime()' 'time.h' '' '' \
-    'clock_gettime(CLOCK_REALTIME, NULL);'
+    'clock_gettime(CLOCK_REALTIME, (void *)1);'
   then
     add_def HAVE_CLOCK_GETTIME
   else
     if test_code 'clock_gettime() in -lrt' 'time.h' '' '-lrt' \
-      'clock_gettime(CLOCK_REALTIME, NULL);'
+      'clock_gettime(CLOCK_REALTIME, (void *)1);'
     then
       add_def HAVE_CLOCK_GETTIME
       EXTRA_LIBS="$EXTRA_LIBS -lrt"
@@ -702,11 +704,14 @@ then
   use_pthread=1
 fi
 
-if test_code 'arc4random_buf()' 'stdlib.h' '' '' 'arc4random_buf(NULL, 0);'; then
+if [ $try_arc4random = "1" ] && \
+  test_code 'arc4random_buf()' 'stdlib.h' '' '' \
+    'arc4random_buf((void *)1, 1);'
+then
   add_def HAVE_ARC4RANDOM
 else
   if test_code 'getrandom()' 'stdlib.h sys/random.h' '' '' \
-      'return getrandom(NULL, 256, 0);'; then
+      'return getrandom((void *)1, 1, 0);'; then
     add_def HAVE_GETRANDOM
   fi
 fi
@@ -900,7 +905,7 @@ if [ $feat_sechash = "1" ] && [ "x$HASH_LINK" = "x" ]  && [ $try_nettle = "1" ];
     add_def FEAT_SECHASH
 
     if test_code 'CMAC in nettle' 'nettle/cmac.h' "$test_cflags" "$test_link" \
-      'cmac128_update(NULL, NULL, NULL, 0, NULL);'
+      'cmac128_update((void *)1, (void *)2, (void *)3, 1, (void *)4);'
     then
       add_def HAVE_CMAC
       EXTRA_OBJECTS="$EXTRA_OBJECTS cmac_nettle.o"
@@ -925,7 +930,7 @@ fi
 
 if [ $feat_sechash = "1" ] && [ "x$HASH_LINK" = "x" ] && [ $try_tomcrypt = "1" ]; then
   if test_code 'tomcrypt' 'tomcrypt.h' '-I/usr/include/tomcrypt' '-ltomcrypt' \
-    'hash_memory_multi(find_hash("md5"), NULL, NULL, NULL, 0, NULL, 0);'
+    'hash_memory_multi(find_hash("md5"), (void *)1, (void *)2, (void *)3, 1, (void *)4, 1);'
   then
     HASH_OBJ="hash_tomcrypt.o"
     HASH_LINK="-ltomcrypt"
@@ -939,7 +944,7 @@ if [ $feat_sechash = "1" ] && [ "x$HASH_LINK" = "x" ] && [ $try_gnutls = "1" ];
   test_link="`pkg_config --libs gnutls`"
   if test_code 'gnutls' 'gnutls/crypto.h' \
     "$test_cflags" "$test_link" '
-      return gnutls_hash(NULL, NULL, 0);'
+      return gnutls_hash((void *)1, (void *)2, 1);'
   then
     HASH_OBJ="hash_gnutls.o"
     HASH_LINK="$test_link"
@@ -947,7 +952,7 @@ if [ $feat_sechash = "1" ] && [ "x$HASH_LINK" = "x" ] && [ $try_gnutls = "1" ];
     add_def FEAT_SECHASH
 
     if test_code 'CMAC in gnutls' 'gnutls/crypto.h' "$test_cflags" "$test_link" \
-      'return gnutls_hmac_init(NULL, GNUTLS_MAC_AES_CMAC_128, NULL, 0);'
+      'return gnutls_hmac_init((void *)1, GNUTLS_MAC_AES_CMAC_128, (void *)2, 0);'
     then
       add_def HAVE_CMAC
       EXTRA_OBJECTS="$EXTRA_OBJECTS cmac_gnutls.o"
@@ -970,13 +975,13 @@ if [ $feat_ntp = "1" ] && [ $feat_nts = "1" ] && [ $try_gnutls = "1" ]; then
   fi
   if test_code 'TLS1.3 in gnutls' 'gnutls/gnutls.h' \
     "$test_cflags" "$test_link $LIBS" '
-      return gnutls_init(NULL, 0) + GNUTLS_TLS1_3 +
+      return gnutls_init((void *)1, 0) + GNUTLS_TLS1_3 +
-        gnutls_priority_init2(NULL, "", NULL, GNUTLS_PRIORITY_INIT_DEF_APPEND) +
+        gnutls_priority_init2((void *)1, "", NULL, GNUTLS_PRIORITY_INIT_DEF_APPEND) +
-        gnutls_prf_rfc5705(NULL, 0, "", 0, "", 16, NULL);'
+        gnutls_prf_rfc5705((void *)1, 0, "", 0, "", 16, (void *)2);'
   then
     if test_code 'SIV in nettle' \
       'nettle/siv-cmac.h' "" "$LIBS" \
-      'siv_cmac_aes128_set_key(NULL, NULL);'
+      'siv_cmac_aes128_set_key((void *)1, (void *)2);'
     then
       EXTRA_OBJECTS="$EXTRA_OBJECTS siv_nettle.o"
       add_def HAVE_SIV
@@ -984,13 +989,19 @@ if [ $feat_ntp = "1" ] && [ $feat_nts = "1" ] && [ $try_gnutls = "1" ]; then
     else
       if test_code 'SIV in gnutls' 'gnutls/crypto.h' \
         "$test_cflags" "$test_link $LIBS" '
-          return gnutls_aead_cipher_init(NULL, GNUTLS_CIPHER_AES_128_SIV, NULL);'
+          return gnutls_aead_cipher_init((void *)1, GNUTLS_CIPHER_AES_128_SIV, (void *)2);'
       then
         EXTRA_OBJECTS="$EXTRA_OBJECTS siv_gnutls.o"
         add_def HAVE_SIV
+        if test_code 'gnutls_aead_cipher_set_key()' 'gnutls/crypto.h' \
+          "$test_cflags" "$test_link $LIBS" '
+            return gnutls_aead_cipher_set_key((void *)1, (void *)2);'
+        then
+          add_def HAVE_GNUTLS_AEAD_CIPHER_SET_KEY
+        fi
       else
         if test_code 'AES128 in nettle' 'nettle/aes.h' '' "$LIBS" \
-          'aes128_set_encrypt_key(NULL, NULL);'
+          'aes128_set_encrypt_key((void *)1, (void *)2);'
         then
           EXTRA_OBJECTS="$EXTRA_OBJECTS siv_nettle.o"
           add_def HAVE_SIV

doc/chrony.conf.adoc

@@ -3,7 +3,7 @@
 // Copyright (C) Richard P. Curnow  1997-2003
 // Copyright (C) Stephen Wadeley  2016
 // Copyright (C) Bryan Christianson  2017
-// Copyright (C) Miroslav Lichvar  2009-2021
+// Copyright (C) Miroslav Lichvar  2009-2022
 //
 // This program is free software; you can redistribute it and/or modify
 // it under the terms of version 2 of the GNU General Public License as
@@ -63,16 +63,28 @@ source. The client-server relationship is strictly hierarchical: a client might
 synchronise its system time to that of the server, but the server's system time
 will never be influenced by that of a client.
 +
+The server can be specified by its hostname or IP address. If the hostname cannot
+be resolved on start, *chronyd* will try it again in increasing intervals, and
+also when the <<chronyc.adoc#online,*online*>> command is issued in *chronyc*.
++
+The DNS record can change over time. The used address will be replaced with a
+newly resolved address when the server becomes unreachable (i.e. no valid
+response to last 8 requests), unsynchronised, a falseticker (i.e. does not
+agree with a majority of other sources), or the root distance is too large (the
+limit can be configured by the <<maxdistance,*maxdistance*>> directive). The
+automatic replacement happens at most once per 30 minutes. It can also be
+triggered manually for all sources by the <<chronyc.adoc#refresh,*refresh*>>
+command in *chronyc*.
++
 This directive can be used multiple times to specify multiple servers.
 +
-The directive is immediately followed by either the name of the
+The directive supports the following options:
-server, or its IP address. It supports the following options:
 +
 *minpoll* _poll_:::
 This option specifies the minimum interval between requests sent to the server
 as a power of 2 in seconds. For example, *minpoll 5* would mean that the
 polling interval should not drop below 32 seconds. The default is 6 (64
-seconds), the minimum is -6 (1/64th of a second), and the maximum is 24 (6
+seconds), the minimum is -7 (1/128th of a second), and the maximum is 24 (6
 months). Note that intervals shorter than 6 (64 seconds) should generally not
 be used with public servers on the Internet, because it might be considered
 abuse. A sub-second interval will be enabled only when the server is reachable
@@ -82,7 +94,7 @@ should be in a local network.
 This option specifies the maximum interval between requests sent to the server
 as a power of 2 in seconds. For example, *maxpoll 9* indicates that the polling
 interval should stay at or below 9 (512 seconds). The default is 10 (1024
-seconds), the minimum is -6 (1/64th of a second), and the maximum is 24 (6
+seconds), the minimum is -7 (1/128th of a second), and the maximum is 24 (6
 months).
 *iburst*:::
 With this option, *chronyd* will start with a burst of 4-8 requests in order to
@@ -138,18 +150,33 @@ behind a lot of packets related to a download of some sort).
 If the user knows that round trip delays above a certain level should cause the
 measurement to be ignored, this level can be defined with the *maxdelay*
 option. For example, *maxdelay 0.3* would indicate that measurements with a
-round-trip delay of 0.3 seconds or more should be ignored. The default value is
+round-trip delay greater than 0.3 seconds should be ignored. The default value
-3 seconds and the maximum value is 1000 seconds.
+is 3 seconds and the maximum value is 1000 seconds.
 *maxdelayratio* _ratio_:::
 This option is similar to the *maxdelay* option above. *chronyd* keeps a record
 of the minimum round-trip delay amongst the previous measurements that it has
-buffered. If a measurement has a round trip delay that is greater than the
+buffered. If a measurement has a round-trip delay that is greater than the
-specified ratio times the minimum delay, it will be rejected.
+specified ratio times the minimum delay, it will be rejected. By default, this
+test is disabled.
 *maxdelaydevratio* _ratio_:::
 If a measurement has a ratio of the increase in the round-trip delay from the
 minimum delay amongst the previous measurements to the standard deviation of
 the previous measurements that is greater than the specified ratio, it will be
 rejected. The default is 10.0.
+*maxdelayquant* _p_:::
+This option disables the *maxdelaydevratio* test and specifies the maximum
+acceptable delay as a quantile of the round-trip delay instead of a function of
+the minimum delay amongst the buffered measurements. If a measurement has a
+round-trip delay that is greater than a long-term estimate of the _p_-quantile,
+it will be rejected.
++
+The specified _p_ value should be between 0.05 and 0.95. For example,
+*maxdelayquant 0.2* would indicate that only measurements with the lowest 20
+percent of round-trip delays should be accepted. Note that it can take many
+measurements for the estimated quantile to reach the expected value. This
+option is intended for synchronisation in mostly static local networks with
+very short polling intervals and possibly combined with the *filter* option.
+By default, this test is disabled in favour of the *maxdelaydevratio* test.
 *mindelay* _delay_:::
 This option specifies a fixed minimum round-trip delay to be used instead of
 the minimum amongst the previous measurements. This can be useful in networks
@@ -176,11 +203,12 @@ Set the minimum number of samples kept for this source. This overrides the
 *maxsamples* _samples_:::
 Set the maximum number of samples kept for this source. This overrides the
 <<maxsamples,*maxsamples*>> directive.
-*filter* _samples_:::
+*filter* _polls_:::
 This option enables a median filter to reduce noise in NTP measurements. The
-filter will reduce the specified number of samples to a single sample. It is
+filter will process samples collected in the specified number of polls
-intended to be used with very short polling intervals in local networks where
+into a single sample. It is intended to be used with very short polling
-it is acceptable to generate a lot of NTP traffic.
+intervals in local networks where it is acceptable to generate a lot of NTP
+traffic.
 *offline*:::
 If the server will not be reachable when *chronyd* is started, the *offline*
 option can be specified. *chronyd* will not try to poll the server until it is
@@ -325,12 +353,6 @@ This option sets the desired number of sources to be used from the pool.
 sources responding to requests. The default value is 4 and the maximum value is
 16.
 +
-{blank}::
-When an NTP source is unreachable,
-marked as a falseticker, or has a distance larger than the limit set by the
-<<maxdistance,*maxdistance*>> directive, *chronyd* will try to replace the
-source with a newly resolved address of the name.
-+
 An example of the *pool* directive is
 +
 ----
@@ -511,8 +533,10 @@ Note that some PTP clocks cannot be configured to timestamp only assert or
 clear events, and it is necessary to use the *width* option to filter wrong
 PPS samples.
 *pin*=_index_::::
-This option specifies the index of the pin to which is connected the PPS
+This option specifies the index of the pin which should be enabled for the
-signal. The default value is 0.
+PPS timestamping. If the PHC does not have configurable pins (i.e. the channel
+function is fixed), the index needs to be set to -1 to disable the pin
+configuration. The default value is 0.
 *channel*=_index_::::
 This option specifies the index of the channel for the PPS mode. The default
 value is 0.
@@ -624,6 +648,13 @@ and that *chronyd* should correct its offset by the current TAI-UTC offset. The
 <<leapsectz,*leapsectz*>> directive must be used with this option and the
 database must be kept up to date in order for this correction to work as
 expected. This option does not make sense with PPS refclocks.
+*local*:::
+This option specifies that the reference clock is an unsynchronised clock which
+is more stable than the system clock (e.g. TCXO, OCXO, or atomic clock) and
+it should be used as a local standard to stabilise the system clock. The
+refclock will bypass the source selection. There should be at most one refclock
+specified with this option and it should have the shortest polling interval
+among all configured sources.
 *minsamples* _samples_:::
 Set the minimum number of samples kept for this source. This overrides the
 <<minsamples,*minsamples*>> directive.
@@ -1099,7 +1130,11 @@ clock will be off for a longer time. On Linux with the default
 *ignore*:::
 No correction is applied to the clock for the leap second. The clock will be
 corrected later in normal operation when new measurements are made and the
-estimated offset includes the one second error.
+estimated offset includes the one second error. This option is particularly
+useful when multiple *chronyd* instances are running on the system, one
+controlling the system clock and others started with the *-x* option, which
+should rely on the first instance to correct the system clock and ignore it for
+the correction of their own NTP clock running on top of the system clock.
 {blank}::
 +
 When serving time to NTP clients that cannot be configured to correct their
@@ -1206,12 +1241,16 @@ This would step the system clock if the adjustment is larger than 0.1 seconds, b
 only in the first three clock updates.
 
 [[maxchange]]*maxchange* _offset_ _start_ _ignore_::
-This directive sets the maximum allowed offset corrected on a clock update. The
+This directive sets the maximum offset to be accepted on a clock update. The
-check is performed only after the specified number of updates to allow a large
+offset is measured relative to the current estimate of the true time, which is
-initial adjustment of the system clock. When an offset larger than the
+different from the system time if a previous slew did not finish.
-specified maximum occurs, it will be ignored for the specified number of times
++
-and then *chronyd* will give up and exit (a negative value can be used to never
+The check is enabled after the specified number of clock updates to allow a
-exit). In both cases a message is sent to syslog.
+large initial offset to be corrected on start. Offsets larger than the
+specified maximum will be ignored for the specified number of times. Another
+large offset will cause *chronyd* to give up and exit. A negative value
+can be used to disable the limit to ignore all large offsets. A syslog message
+will be generated when an offset is ignored or it causes the exit.
 +
 An example of the use of this directive is:
 +
@@ -1238,7 +1277,7 @@ This directive specifies the maximum assumed drift (frequency error) of the
 system clock. It limits the frequency adjustment that *chronyd* is allowed to
 use to correct the measured drift. It is an additional limit to the maximum
 adjustment that can be set by the system driver (100000 ppm on Linux, 500 ppm
-on FreeBSD, NetBSD, and macOS 10.13+, 32500 ppm on Solaris).
+on FreeBSD, NetBSD, and macOS 10.13+, 32500 ppm on illumos).
 +
 By default, the maximum assumed drift is 500000 ppm, i.e. the adjustment is
 limited by the system driver rather than this directive.
@@ -1277,7 +1316,7 @@ all supported systems with the exception of macOS 12 or earlier).
 +
 For each system there is a maximum frequency offset of the clock that can be set
 by the driver. On Linux it is 100000 ppm, on FreeBSD, NetBSD and macOS 10.13+ it
-is 5000 ppm, and on Solaris it is 32500 ppm. Also, due to a kernel limitation,
+is 5000 ppm, and on illumos it is 32500 ppm. Also, due to a kernel limitation,
 setting *maxslewrate* on FreeBSD, NetBSD, macOS 10.13+ to a value between 500
 ppm and 5000 ppm will effectively set it to 500 ppm.
 +
@@ -2051,9 +2090,11 @@ from the example line above):
 . Stratum of remote computer. [2]
 . RFC 5905 tests 1 through 3 (1=pass, 0=fail) [111]
 . RFC 5905 tests 5 through 7 (1=pass, 0=fail) [111]
-. Tests for maximum delay, maximum delay ratio and maximum delay dev ratio,
+. Results of the *maxdelay*, *maxdelayratio*, and *maxdelaydevratio* (or
-  against defined parameters, and a test for synchronisation loop (1=pass,
+  *maxdelayquant*) tests, and a test for synchronisation loop (1=pass,
-  0=fail) [1111]
+  0=fail). The first test from these four also checks the server precision,
+  response time, and whether an interleaved response is acceptable for
+  synchronisation. [1111]
 . Local poll [10]
 . Remote poll [10]
 . '`Score`' (an internal score within each polling level used to decide when to
@@ -2127,6 +2168,71 @@ from the example line above):
   the source is more variable than the delay of packets sent from the source
   back. [0.00, i.e. no correction for asymmetry]
 +
+*selection*:::
+This option logs information about selection of sources for synchronisation to
+a file called _selection.log_. Note that the rate of entries written to this
+file grows quadratically with the number of specified sources (each measurement
+triggers the selection for all sources). An example line (which actually
+appears as a single line in the file) from the log file is shown below.
++
+----
+2022-05-01 02:01:20 203.0.113.15    * -----  377  1.00  \
+     4.228e+01 -1.575e-04  1.239e-04
+----
++
+The columns are as follows (the quantities in square brackets are the values
+from the example line above):
++
+. Date [2022-05-01]
+. Hour:Minute:Second. Note that the date-time pair is expressed in
+  UTC, not the local time zone. [02:01:20]
+. IP address or reference ID of the source. [203.0.113.15]
+. State of the source indicated with one of the following symbols. [*]
+{blank}::::
+Not considered selectable for synchronisation:
+* _N_ - has the *noselect* option.
+* _s_ - is not synchronised.
+* _M_ - does not have enough measurements.
+* _d_ - has a root distance larger than the maximum distance (configured by the
+        <<maxdistance,*maxdistance*>> directive).
+* _~_ - has a jitter larger than the maximum jitter (configured by the
+        <<maxjitter,*maxjitter*>> directive).
+* _w_ - waits for other sources to get out of the _M_ state.
+* _S_ - has older measurements than other sources.
+* _O_ - has a stratum equal or larger than the orphan stratum (configured by
+        the <<local,*local*>> directive).
+* _T_ - does not fully agree with sources that have the *trust* option.
+* _x_ - does not agree with other sources (falseticker).
+{blank}::::
+Considered selectable for synchronisation, but not currently used:
+* _W_ - waits for other sources to be selectable (required by the
+        <<minsources,*minsources*>> directive, or the *require* option of
+        another source).
+* _P_ - another selectable source is preferred due to the *prefer* option.
+* _U_ - waits for a new measurement (after selecting a different best source).
+* _D_ - has, or recently had, a root distance which is too large to be combined
+        with other sources (configured by the <<combinelimit,*combinelimit*>>
+        directive).
+{blank}::::
+Used for synchronisation of the local clock:
+* _+_ - combined with the best source.
+* _*_ - selected as the best source to update the reference data (e.g. root
+        delay, root dispersion).
+. Reachability register printed as an octal number. The register has 8 bits and
+  is updated on every received or missed packet from the source. A value of 377
+  indicates that a valid reply was received for all from the last eight
+  transmissions. [377]
+. Current score against the source in the _*_ state. The scoring system avoids
+  frequent reselection when multiple sources have a similar root distance. A
+  value larger than 1 indicates this source was better than the _*_ source in
+  recent selections. If the score reaches 10, the best source will be reselected
+  and the scores will be reset to 1. [1.00]
+. Interval since the last measurement of the source in seconds. [4.228e+01]
+. Lower endpoint of the interval which was expected to contain the true offset
+  of the local clock determined by the root distance of the source. [-1.575e-04]
+. Upper endpoint of the interval which was expected to contain the true offset
+  of the local clock determined by the root distance of the source. [1.239e-04]
++
 *tracking*:::
 This option logs changes to the estimate of the system's gain or loss rate, and
 any slews made, to a file called _tracking.log_. An example line (which
@@ -2381,13 +2487,20 @@ be enabled by the *xleave* option in the <<server,*server*>> or the
 +
 This directive is supported on Linux 3.19 and newer. The NIC must support HW
 timestamping, which can be verified with the *ethtool -T* command. The list of
-capabilities should include _SOF_TIMESTAMPING_RAW_HARDWARE_,
+capabilities should include _hardware-raw-clock_, _hardware-transmit_, and
-_SOF_TIMESTAMPING_TX_HARDWARE_, and _SOF_TIMESTAMPING_RX_HARDWARE_. Receive
+_hardware-receive_. The receive filter _all_, or _ntp_, is necessary for
-filter _HWTSTAMP_FILTER_ALL_, or _HWTSTAMP_FILTER_NTP_ALL_, is necessary for
+timestamping of received NTP packets. Timestamping of packets received on
-timestamping of received NTP packets. Timestamping of packets received on bridged
+bridged and bonded interfaces is supported on Linux 4.13 and newer. If HW
-and bonded interfaces is supported on Linux 4.13 and newer. When *chronyd* is
+timestamping does not work for received packets, *chronyd* will use kernel
-running, no other process (e.g. a PTP daemon) should be working with the NIC
+receive timestamps instead. Transmit-only HW timestamping can still be useful
-clock.
+to improve stability of the synchronisation.
++
+*chronyd* does not synchronise the NIC clock. It assumes the clock is running
+free. Multiple instances of *chronyd* can use the same interface with enabled
+HW timestamping. Applications which need HW timestamping with a synchronised
+clock (e.g. a PTP daemon) should use a virtual clock running on top of the
+physical clock created by writing to _/sys/class/ptp/ptpX/n_vclocks_. This
+feature is available on Linux 5.14 and newer.
 +
 If the kernel supports software timestamping, it will be enabled for all
 interfaces. The source of timestamps (i.e. hardware, kernel, or daemon) is
@@ -2445,7 +2558,8 @@ is selected by default. Some NICs can timestamp PTP packets only. By default,
 they will be configured with the _none_ filter and expected to provide hardware
 timestamps for transmitted packets only. Timestamping of PTP packets is useful
 with NTP-over-PTP enabled by the <<chrony.conf.adoc#ptpport,*ptpport*>>
-directive. Forcing timestamping of all packets with the _all_ filter could be
+directive, or when another application is receiving PTP packets on the
+interface. Forcing timestamping of all packets with the _all_ filter could be
 useful if the NIC supported both the _all_ and _ntp_ filters, and it should
 timestamp both NTP and PTP packets, or NTP packets on a different UDP port.
 {blank}::
@@ -2491,7 +2605,7 @@ The type is a name of a cryptographic hash function or cipher which is used to
 generate and verify the MAC. The default type is *MD5*, which is always
 supported.
 If *chronyd* was built with enabled support for hashing using a crypto library
-(nettle, nss, or libtomcrypt), the following functions are available: *MD5*,
+(Nettle, GnuTLS, NSS, or LibTomCrypt), the following functions are available: *MD5*,
 *SHA1*, *SHA256*, *SHA384*, *SHA512*. Depending on which library and version is
 *chronyd* using, some of the following hash functions and ciphers may
 also be available:
@@ -2524,7 +2638,7 @@ file when the <<chronyc.adoc#rekey,*rekey*>> command is issued by *chronyc*).
 [[lock_all]]*lock_all*::
 The *lock_all* directive will lock the *chronyd* process into RAM so that it
 will never be paged out. This can result in lower and more consistent latency.
-The directive is supported on Linux, FreeBSD, NetBSD, and Solaris.
+The directive is supported on Linux, FreeBSD, NetBSD, and illumos.
 
 [[pidfile]]*pidfile* _file_::
 Unless *chronyd* is started with the *-Q* option, it writes its process ID
@@ -2564,7 +2678,7 @@ ptpport 319
 ----
 
 [[sched_priority]]*sched_priority* _priority_::
-On Linux, FreeBSD, NetBSD, and Solaris, the *sched_priority* directive will
+On Linux, FreeBSD, NetBSD, and illumos, the *sched_priority* directive will
 select the SCHED_FIFO real-time scheduler at the specified priority (which must
 be between 0 and 100). On macOS, this option must have either a value of 0 (the
 default) to disable the thread time constraint policy or 1 for the policy to be
@@ -2590,7 +2704,7 @@ The *user* directive sets the name of the system user to which *chronyd* will
 switch after start in order to drop root privileges.
 +
 On Linux, *chronyd* needs to be compiled with support for the *libcap* library.
-On macOS, FreeBSD, NetBSD and Solaris *chronyd* forks into two processes.
+On macOS, FreeBSD, NetBSD and illumos *chronyd* forks into two processes.
 The child process retains root privileges, but can only perform a very limited
 range of privileged system calls on behalf of the parent.
 +

doc/chronyc.adoc

@@ -2,7 +2,7 @@
 //
 // Copyright (C) Richard P. Curnow  1997-2003
 // Copyright (C) Stephen Wadeley  2016
-// Copyright (C) Miroslav Lichvar  2009-2017, 2019-2020
+// Copyright (C) Miroslav Lichvar  2009-2017, 2019-2022
 //
 // This program is free software; you can redistribute it and/or modify
 // it under the terms of version 2 of the GNU General Public License as
@@ -99,12 +99,15 @@ With this option multiple commands can be specified. Each argument will be
 interpreted as a whole command.
 
 *-h* _host_::
-This option allows the user to specify which host (or comma-separated list of
+This option specifies the host to be contacted by *chronyc*. It can be
-addresses) running the *chronyd* program is to be contacted. This allows for
+specified with a hostname, IP address, or path to the local Unix domain socket.
-remote monitoring, without having to connect over SSH to the other host first.
+Multiple values can be specified as a comma-separated list to provide a
+fallback.
 +
-The default is to contact *chronyd* running on the same host where
+The default value is _@CHRONYRUNDIR@/chronyd.sock,127.0.0.1,::1_, i.e. the host
-*chronyc* is being run.
+where *chronyc* is being run. First, it tries to connect to the Unix domain
+socket and if that fails (e.g. due to running under a non-root user), it
+will try to connect to 127.0.0.1 and then ::1.
 
 *-p* _port_::
 This option allows the user to specify the UDP port number which the target
@@ -177,15 +180,19 @@ stratum-2 and is synchronised from a stratum-1).
 This is the time (UTC) at which the last measurement from the reference
 source was processed.
 *System time*:::
-In normal operation, *chronyd* by default never steps the system clock, because
+This is the current offset between the NTP clock and system clock. The NTP
-any jump in the time can have adverse consequences for certain application
+clock is a software (virtual) clock maintained by *chronyd*, which is
-programs. Instead, any error in the system clock is corrected by slightly
+synchronised to the configured time sources and provides time to NTP clients.
-speeding up or slowing down the system clock until the error has been removed,
+The system clock is synchronised to the NTP clock. To avoid steps in the
-and then returning to the system clock's normal speed. A consequence of this is
+system time, which might have adverse consequences for certain applications,
-that there will be a period when the system clock (as read by other programs)
+the system clock is normally corrected only by speeding up or slowing down (up
-will be different from *chronyd*'s estimate of the current true time (which it
+to the rate configured by the <<chrony.conf.adoc#maxslewrate,*maxslewrate*>>
-reports to NTP clients when it is operating as a server). The value reported
+directive). If the offset is too large, this correction will take a very long
-on this line is the difference due to this effect.
+time. A step can be forced by the <<makestep,*makestep*>> command, or the
+<<chrony.conf.adoc#makestep,*makestep*>> directive in the configuration file.
++
+Note that all other offsets reported by *chronyc* and most offsets in the log
+files are relative to the NTP clock, not the system clock.
 *Last offset*:::
 This is the estimated local offset on the last clock update. A positive value
 indicates the local time (as previously estimated true time) was ahead of the
@@ -448,6 +455,7 @@ states are reported.
 The following states indicate the source is not considered selectable for
 synchronisation:
 * _N_ - has the *noselect* option.
+* _s_ - is not synchronised.
 * _M_ - does not have enough measurements.
 * _d_ - has a root distance larger than the maximum distance (configured by the
         <<chrony.conf.adoc#maxdistance,*maxdistance*>> directive).
@@ -658,6 +666,7 @@ RX timestamping : Kernel
 Total TX        : 24
 Total RX        : 24
 Total valid RX  : 24
+Total good RX   : 22
 ----
 +
 The fields are explained as follows:
@@ -695,7 +704,8 @@ packets sent to the source is more variable than the delay of packets sent
 from the source back.
 *NTP tests*:::
 Results of RFC 5905 tests 1 through 3, 5 through 7, and tests for maximum
-delay, delay ratio, delay dev ratio, and synchronisation loop.
+delay, delay ratio, delay dev ratio (or delay quantile), and synchronisation
+loop.
 *Interleaved*:::
 This shows if the response was in the interleaved mode.
 *Authenticated*:::
@@ -710,7 +720,10 @@ The number of packets sent to the source.
 *Total RX*:::
 The number of all packets received from the source.
 *Total valid RX*:::
-The number of valid packets received from the source.
+The number of packets which passed the first two groups of NTP tests.
+*Total good RX*:::
+The number of packets which passed all three groups of NTP tests, i.e. the NTP
+measurement was accepted.
 
 [[add_peer]]*add peer* _name_ [_option_]...::
 The *add peer* command allows a new NTP peer to be added whilst

doc/chronyd.adoc

@@ -100,7 +100,7 @@ directive in the configuration file. This option is useful if you want to stop
 and restart *chronyd* briefly for any reason, e.g. to install a new version.
 However, it should be used only on systems where the kernel can maintain clock
 compensation whilst not under *chronyd*'s control (i.e. Linux, FreeBSD, NetBSD,
-Solaris, and macOS 10.13 or later).
+illumos, and macOS 10.13 or later).
 
 *-R*::
 When this option is used, the <<chrony.conf.adoc#initstepslew,*initstepslew*>>
@@ -141,7 +141,7 @@ after start in order to drop root privileges. It overrides the
 _@DEFAULT_USER@_.
 +
 On Linux, *chronyd* needs to be compiled with support for the *libcap* library.
-On macOS, FreeBSD, NetBSD and Solaris *chronyd* forks into two processes.
+On macOS, FreeBSD, NetBSD, and illumos *chronyd* forks into two processes.
 The child process retains root privileges, but can only perform a very limited
 range of privileged system calls on behalf of the parent.
 
@@ -181,7 +181,7 @@ limited.
 The filters cannot be enabled with the *mailonchange* directive.
 
 *-P* _priority_::
-On Linux, FreeBSD, NetBSD, and Solaris, this option will select the SCHED_FIFO
+On Linux, FreeBSD, NetBSD, and illumos this option will select the SCHED_FIFO
 real-time scheduler at the specified priority (which must be between 0 and
 100). On macOS, this option must have either a value of 0 to disable the thread
 time constraint policy or 1 for the policy to be enabled. Other systems do not
@@ -189,7 +189,7 @@ support this option. The default value is 0.
 
 *-m*::
 This option will lock *chronyd* into RAM so that it will never be paged out.
-This mode is only supported on Linux, FreeBSD, NetBSD, and Solaris.
+This mode is only supported on Linux, FreeBSD, NetBSD, and illumos.
 
 *-x*::
 This option disables the control of the system clock. *chronyd* will not try to

doc/faq.adoc

@@ -1,7 +1,7 @@
 // This file is part of chrony
 //
 // Copyright (C) Richard P. Curnow  1997-2003
-// Copyright (C) Miroslav Lichvar  2014-2016, 2020-2021
+// Copyright (C) Miroslav Lichvar  2014-2016, 2020-2022
 //
 // This program is free software; you can redistribute it and/or modify
 // it under the terms of version 2 of the GNU General Public License as
@@ -345,6 +345,15 @@ server ntp.local minpoll -6 maxpoll -6 filter 15 xleave
 hwtimestamp eth0 minpoll -6
 ----
 
+Since `chrony` version 4.3, the minimum `minpoll` is -7 and a filter using a
+long-term estimate of a delay quantile can be enabled by the `maxdelayquant`
+option to replace the default `maxdelaydevratio` filter, which is sensitive to
+outliers corrupting the minimum delay. For example:
+
+----
+server ntp.local minpoll -7 maxpoll -7 filter 31 maxdelayquant 0.3 xleave
+----
+
 As an experimental feature added in version 4.2, `chronyd` supports an NTPv4
 extension field containing an additional timestamp to enable frequency transfer
 and significantly improve stability of synchronisation. It can be enabled by
@@ -475,6 +484,64 @@ identically configured leap-smearing servers. Note that some clients can get
 leap seconds from other sources (e.g. with the `leapsectz` directive in
 `chrony`) and they will not work correctly with a leap smearing server.
 
+=== How should `chronyd` be configuration with `gpsd`?
+
+A GPS or other GNSS receiver can be used as a reference clock with `gpsd`. It
+can work as one or two separate time sources for each connected receiver. The
+first time source is based on timestamping of messages sent by the receiver.
+Typically, it is accurate to milliseconds. The other source is much more
+accurate. It is timestamping a pulse-per-second (PPS) signal, usually connected
+to a serial port (e.g. DCD pin) or GPIO pin.
+
+If the PPS signal is connected to the serial port which is receiving messages
+from the GPS/GNSS receiver, `gpsd` should detect and use it automatically. If
+it is connected to a GPIO pin, or another serial port, the PPS device needs to
+be specified on the command line as an additional data source. On Linux, the
+`ldattach` utility can be used to create a PPS device for a serial device.
+
+The message-based time source provided by `gpsd` is specified as a `SHM 0`
+refclock, or other even number if `gpsd` is configured with multiple receivers.
+
+The PPS-based time source is specified as a `SHM 1` refclock (or other odd
+number), or `SOCK /var/run/chrony.DEV.sock` where `DEV` is the name of the
+serial device (e.g. ttyS0).
+
+With `chronyd` and `gpsd` both supporting PPS, and `gpsd` providing two
+different refclocks for PPS, there are three different recommended
+configurations:
+
+----
+# First option
+refclock SOCK /var/run/chrony.ttyS0.sock refid GPS
+
+# Second option
+refclock SHM 1 refid GPS
+
+# Third option
+refclock PPS /dev/pps0 lock NMEA refid GPS
+refclock SHM 0 offset 0.5 delay 0.1 refid NMEA noselect
+----
+
+Each option has some advantages:
+
+* `SOCK` does not use polling (i.e. it can get samples earlier than `SHM`),
+  but it requires `gpsd` to be started after `chronyd` in order to connect to
+  its socket
+* `SOCK` and `SHM 1` can be more accurate than `PPS` if `gpsd` corrects for the
+  sawtooth error provided by the receiver in serial data
+* `PPS` can be used with higher PPS rates (specified by the `rate` option),
+  but it requires a second refclock or another time source to pair pulses
+  with seconds, and the `SHM 0` offset needs to be specified
+  <<using-pps-refclock,correctly>> to compensate for the message delay, while
+  `gpsd` can apply HW-specific information
+
+If the PPS signal is not available, or cannot be used for some reason, the only
+option is the message-based timing
+
+----
+refclock SHM 0 offset 0.5 delay 0.1 refid GPS
+----
+
 === Does `chrony` support PTP?
 
 No, the Precision Time Protocol (PTP) is not supported as a protocol for
@@ -498,6 +565,27 @@ transport for NTP messages (NTP over PTP) to enable hardware timestamping on
 hardware which can timestamp PTP packets only. It can be enabled by the
 `ptpport` directive.
 
+=== Why are client log records dropped before reaching `clientloglimit`?
+
+The number of dropped client log records reported by the `serverstats` command
+can be increasing before the number of clients reported by the `clients` command
+reaches the maximum value corresponding to the memory limit set by the
+`clientloglimit` directive.
+
+This is due to the design of the data structure keeping the client records. It
+is a hash table which can store only up to 16 colliding addresses per slot. If
+a slot has more collisions and the table already has the maximum size, the
+oldest record will be dropped and replaced by the new client.
+
+Note that the size of the table is always a power of two and it can only grow.
+The limit set by the `clientloglimit` directive takes into account that two
+copies of the table exist when it is being resized. This means the actual
+memory usage reported by `top` and other utilities can be significantly smaller
+than the limit even when the maximum number of records is used.
+
+The absolute maximum number of client records kept at the same time is
+16777216.
+
 === What happened to the `commandkey` and `generatecommandkey` directives?
 
 They were removed in version 2.2. Authentication is no longer supported in the
@@ -624,6 +712,18 @@ was not shut down for too long and the server's certificate was not renewed too
 close to its expiration, it should be sufficient for the time checks to
 succeed.
 
+If you run your own server, you can use a self-signed certificate covering
+all dates where the client can start (e.g. years 1970-2100). The certificate
+needs to be installed on the client and specified with the `ntstrustedcerts`
+directive. The server can have multiple names and certificates. To avoid
+trusting a certificate for too long, a new certificate can be added to the
+server periodically (e.g. once per year) and the client can have the server
+name and trusted certificate updated automatically (e.g. using a package
+repository, or a cron script downloading the files directly from the server
+over HTTPS). A client that was shut down for years will still be able to
+synchronise its clock and perform the update as long as the server keeps
+the old certificate.
+
 As a last resort, you can disable the time checks by the `nocerttimecheck`
 directive. This has some important security implications. To reduce the
 security risk, you can use the `nosystemcert` and `ntstrustedcerts` directives
@@ -692,12 +792,14 @@ frequently, you can effectively disable the test by setting the
 `maxdelaydevratio` option to a very large value (e.g. 1000000), or speed up the
 recovery by increasing the clock error rate with the `maxclockerror` directive.
 
+[[using-pps-refclock]]
 === Using a PPS reference clock?
 
 A pulse-per-second (PPS) reference clock requires a non-PPS time source to
 determine which second of UTC corresponds to each pulse. If it is another
 reference clock specified with the `lock` option in the `refclock` directive,
-the offset between the two reference clocks must be smaller than 0.2 seconds in
+the offset between the two reference clocks must be smaller than 0.4 seconds
+(0.2 seconds with `chrony` versions before 4.1) in
 order for the PPS reference clock to work. With NMEA reference clocks it is
 common to have a larger offset. It needs to be corrected with the `offset`
 option.
@@ -716,7 +818,7 @@ foo.example.net             7   3   200     -2.991     16.141   -107us   492us
 
 the offset of the NMEA source would need to be increased by about 0.504
 seconds. It does not have to be very accurate. As long as the offset of the
-NMEA reference clock stays below 0.2 seconds, the PPS reference clock should be
+NMEA reference clock stays below the limit, the PPS reference clock should be
 able to determine the seconds corresponding to the pulses and allow the samples
 to be used for synchronisation.
 
@@ -772,6 +874,10 @@ in parentheses) on the `Reference ID` line.
 Only by the source code. See _cmdmon.c_ (`chronyd` side) and _client.c_
 (`chronyc` side).
 
+Note that this protocol is not compatible with the mode 6 or mode 7 protocol
+supported by `ntpd`, i.e. the `ntpq` or `ntpdc` utility cannot be used to
+monitor `chronyd`, and `chronyc` cannot be used to monitor `ntpd`.
+
 == Real-time clock issues
 
 === What is the real-time clock (RTC)?
@@ -898,6 +1004,34 @@ timestamps (e.g. daemon timestamp vs kernel timestamp) for serving time and
 synchronisation of its own clock, which will cause the other computer to
 measure a significant offset.
 
+== Operation
+
+=== What clocks does `chronyd` use?
+
+There are several different clocks used by `chronyd`:
+
+* *System clock:* software clock maintained by the kernel. It is the main clock
+  used by applications running on the computer. It is synchronised by `chronyd`
+  to its NTP clock, unless started with the *-x* option.
+* *NTP clock:* software clock (virtual) based on the system clock and internal
+  to `chronyd`. It keeps the best estimate of the true time according to the
+  configured time sources, which is served to NTP clients unless time smoothing
+  is enabled by the *smoothtime* directive. The *System time* value in the
+  `tracking` report is the current offset between the system and NTP clock.
+* *Real-time clock (RTC):* hardware clock keeping time even when the
+  computer is turned off. It is used by the kernel to initialise the system
+  clock on boot and also by `chronyd` to compensate for its measured drift if
+  configured with the `rtcfile` directive and started with the `-s` option.
+  The clock can be kept accurate only by stepping enabled by the `rtcsync` or
+  `rtcautotrim` directive.
+* *Reference clock:* hardware clock used as a time source. It is specified by
+  the `refclock` directive.
+* *NIC clock (also known as PTP hardware clock):* hardware clock timestamping
+  packets received and transmitted by a network device specified by the
+  *hwtimestamp* directive. The clock is expected to be running free. It is not
+  synchronised by `chronyd`. Its offset is tracked relative to the NTP clock in
+  order to convert the hardware timestamps.
+
 == Operating systems
 
 === Does `chrony` support Windows?

doc/installation.adoc

@@ -27,7 +27,7 @@ The following libraries with their development files, and programs, are needed
 to enable optional features:
 
 * pkg-config: detection of development libraries
-* Nettle, NSS, or LibTomCrypt: secure hash functions (`SECHASH`)
+* Nettle, GnuTLS, NSS, or LibTomCrypt: secure hash functions (`SECHASH`)
 * libcap: dropping root privileges on Linux (`DROPROOT`)
 * libseccomp: system call filter on Linux (`SCFILTER`)
 * GnuTLS and Nettle: Network Time Security (`NTS`)

examples/chrony.nm-dispatcher.dhcp

@@ -1,8 +1,7 @@
 #!/bin/sh
 # This is a NetworkManager dispatcher script for chronyd to update
-# its NTP sources passed from DHCP options. Note that this script is
+# its NTP sources with servers from DHCP options passed by NetworkManager
-# specific to NetworkManager-dispatcher due to use of the
+# in the DHCP4_NTP_SERVERS and DHCP6_DHCP6_NTP_SERVERS environment variables.
-# DHCP4_NTP_SERVERS environment variable.
 
 export LC_ALL=C
 
@@ -10,17 +9,23 @@ interface=$1
 action=$2
 
 chronyc=/usr/bin/chronyc
-default_server_options=iburst
+server_options=iburst
 server_dir=/var/run/chrony-dhcp
 
 dhcp_server_file=$server_dir/$interface.sources
-# DHCP4_NTP_SERVERS is passed from DHCP options by NetworkManager.
+dhcp_ntp_servers="$DHCP4_NTP_SERVERS $DHCP6_DHCP6_NTP_SERVERS"
-nm_dhcp_servers=$DHCP4_NTP_SERVERS
 
 add_servers_from_dhcp() {
     rm -f "$dhcp_server_file"
-    for server in $nm_dhcp_servers; do
+    for server in $dhcp_ntp_servers; do
-        echo "server $server $default_server_options" >> "$dhcp_server_file"
+        # Check for invalid characters (from the DHCPv6 NTP FQDN suboption)
+        len1=$(printf '%s' "$server" | wc -c)
+        len2=$(printf '%s' "$server" | tr -d -c 'A-Za-z0-9:.-' | wc -c)
+        if [ "$len1" -ne "$len2" ] || [ "$len2" -lt 1 ] || [ "$len2" -gt 255 ]; then
+          continue
+        fi
+
+        printf 'server %s %s\n' "$server" "$server_options" >> "$dhcp_server_file"
     done
     $chronyc reload sources > /dev/null 2>&1 || :
 }
@@ -34,10 +39,11 @@ clear_servers_from_dhcp() {
 
 mkdir -p $server_dir
 
-if [ "$action" = "up" ] || [ "$action" = "dhcp4-change" ]; then
+case "$action" in
-    add_servers_from_dhcp
+    up|dhcp4-change|dhcp6-change)
-elif [ "$action" = "down" ]; then
+        add_servers_from_dhcp;;
-    clear_servers_from_dhcp
+    down)
-fi
+        clear_servers_from_dhcp;;
+esac
 
 exit 0

examples/chrony.nm-dispatcher.onoffline

@@ -7,8 +7,18 @@ export LC_ALL=C
 
 chronyc=/usr/bin/chronyc
 
-# For NetworkManager consider only up/down events
+# For NetworkManager consider only selected events
-[ $# -ge 2 ] && [ "$2" != "up" ] && [ "$2" != "down" ] && exit 0
+if [ $# -ge 2 ]; then
+    case "$2" in
+        up|down|connectivity-change)
+            ;;
+        dhcp6-change)
+            # No other action is reported for routable IPv6
+            ;;
+        *)
+            exit 0;;
+    esac
+fi
 
 # Note: for networkd-dispatcher routable.d ~= on and off.d ~= off
 

hwclock.c

@@ -2,7 +2,7 @@
   chronyd/chronyc - Programs for keeping computer clocks accurate.
 
  **********************************************************************
- * Copyright (C) Miroslav Lichvar  2016-2018
+ * Copyright (C) Miroslav Lichvar  2016-2018, 2022
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -33,6 +33,7 @@
 #include "local.h"
 #include "logging.h"
 #include "memory.h"
+#include "quantiles.h"
 #include "regress.h"
 #include "util.h"
 
@@ -43,6 +44,13 @@
 /* Maximum acceptable frequency offset of the clock */
 #define MAX_FREQ_OFFSET (2.0 / 3.0)
 
+/* Quantiles for filtering readings by delay */
+#define DELAY_QUANT_MIN_K 1
+#define DELAY_QUANT_MAX_K 2
+#define DELAY_QUANT_Q 10
+#define DELAY_QUANT_REPEAT 7
+#define DELAY_QUANT_MIN_STEP 1.0e-9
+
 struct HCL_Instance_Record {
   /* HW and local reference timestamp */
   struct timespec hw_ref;
@@ -64,12 +72,18 @@ struct HCL_Instance_Record {
   /* Minimum interval between samples */
   double min_separation;
 
+  /* Expected precision of readings */
+  double precision;
+
   /* Flag indicating the offset and frequency values are valid */
   int valid_coefs;
 
   /* Estimated offset and frequency of HW clock relative to local clock */
   double offset;
   double frequency;
+
+  /* Estimated quantiles of reading delay */
+  QNT_Instance delay_quants;
 };
 
 /* ================================================== */
@@ -92,7 +106,7 @@ handle_slew(struct timespec *raw, struct timespec *cooked, double dfreq,
 /* ================================================== */
 
 HCL_Instance
-HCL_CreateInstance(int min_samples, int max_samples, double min_separation)
+HCL_CreateInstance(int min_samples, int max_samples, double min_separation, double precision)
 {
   HCL_Instance clock;
 
@@ -110,6 +124,10 @@ HCL_CreateInstance(int min_samples, int max_samples, double min_separation)
   clock->n_samples = 0;
   clock->valid_coefs = 0;
   clock->min_separation = min_separation;
+  clock->precision = precision;
+  clock->delay_quants = QNT_CreateInstance(DELAY_QUANT_MIN_K, DELAY_QUANT_MAX_K,
+                                           DELAY_QUANT_Q, DELAY_QUANT_REPEAT,
+                                           DELAY_QUANT_MIN_STEP);
 
   LCL_AddParameterChangeHandler(handle_slew, clock);
 
@@ -121,6 +139,7 @@ HCL_CreateInstance(int min_samples, int max_samples, double min_separation)
 void HCL_DestroyInstance(HCL_Instance clock)
 {
   LCL_RemoveParameterChangeHandler(handle_slew, clock);
+  QNT_DestroyInstance(clock->delay_quants);
   Free(clock->y_data);
   Free(clock->x_data);
   Free(clock);
@@ -140,6 +159,94 @@ HCL_NeedsNewSample(HCL_Instance clock, struct timespec *now)
 
 /* ================================================== */
 
+int
+HCL_ProcessReadings(HCL_Instance clock, int n_readings, struct timespec tss[][3],
+                    struct timespec *hw_ts, struct timespec *local_ts, double *err)
+{
+  double delay, raw_delay, min_delay, low_delay, high_delay, e, pred_err;
+  double delay_sum, hw_sum, local_sum, local_prec, freq;
+  int i, min_reading, combined;
+  struct timespec ts1, ts2;
+
+  if (n_readings < 1)
+    return 0;
+
+  /* Work out the current correction multiplier needed to get cooked delays */
+  LCL_CookTime(&tss[0][0], &ts1, NULL);
+  LCL_CookTime(&tss[n_readings - 1][2], &ts2, NULL);
+  if (UTI_CompareTimespecs(&tss[0][0], &tss[n_readings - 1][2]) < 0)
+    freq = UTI_DiffTimespecsToDouble(&ts1, &ts2) /
+           UTI_DiffTimespecsToDouble(&tss[0][0], &tss[n_readings - 1][2]);
+  else
+    freq = 1.0;
+
+  for (i = 0; i < n_readings; i++) {
+    delay = freq * UTI_DiffTimespecsToDouble(&tss[i][2], &tss[i][0]);
+
+    if (delay < 0.0) {
+      /* Step in the middle of a reading? */
+      DEBUG_LOG("Bad reading delay=%e", delay);
+      return 0;
+    }
+
+    if (i == 0 || min_delay > delay) {
+      min_delay = delay;
+      min_reading = i;
+    }
+
+    QNT_Accumulate(clock->delay_quants, delay);
+  }
+
+  local_prec = LCL_GetSysPrecisionAsQuantum();
+
+  low_delay = QNT_GetQuantile(clock->delay_quants, DELAY_QUANT_MIN_K);
+  high_delay = QNT_GetQuantile(clock->delay_quants, DELAY_QUANT_MAX_K);
+  low_delay = MIN(low_delay, high_delay);
+  high_delay = MAX(high_delay, low_delay + local_prec);
+
+  /* Combine readings with delay in the expected interval */
+  for (i = combined = 0, delay_sum = hw_sum = local_sum = 0.0; i < n_readings; i++) {
+    raw_delay = UTI_DiffTimespecsToDouble(&tss[i][2], &tss[i][0]);
+    delay = freq * raw_delay;
+
+    if (delay < low_delay || delay > high_delay)
+      continue;
+
+    delay_sum += delay;
+    hw_sum += UTI_DiffTimespecsToDouble(&tss[i][1], &tss[0][1]);
+    local_sum += UTI_DiffTimespecsToDouble(&tss[i][0], &tss[0][0]) + raw_delay / 2.0;
+    combined++;
+  }
+
+  DEBUG_LOG("Combined %d readings lo=%e hi=%e", combined, low_delay, high_delay);
+
+  if (combined > 0) {
+    UTI_AddDoubleToTimespec(&tss[0][1], hw_sum / combined, hw_ts);
+    UTI_AddDoubleToTimespec(&tss[0][0], local_sum / combined, local_ts);
+    *err = MAX(delay_sum / combined / 2.0, clock->precision);
+    return 1;
+  }
+
+  /* Accept the reading with minimum delay if its interval does not contain
+     the current offset predicted from previous samples */
+
+  *hw_ts = tss[min_reading][1];
+  UTI_AddDoubleToTimespec(&tss[min_reading][0], min_delay / freq / 2.0, local_ts);
+  *err = MAX(min_delay / 2.0, clock->precision);
+
+  pred_err = 0.0;
+  LCL_CookTime(local_ts, &ts1, NULL);
+  if (!HCL_CookTime(clock, hw_ts, &ts2, &e) ||
+      ((pred_err = UTI_DiffTimespecsToDouble(&ts1, &ts2)) > *err)) {
+    DEBUG_LOG("Accepted reading err=%e prerr=%e", *err, pred_err);
+    return 1;
+  }
+
+  return 0;
+}
+
+/* ================================================== */
+
 void
 HCL_AccumulateSample(HCL_Instance clock, struct timespec *hw_ts,
                      struct timespec *local_ts, double err)

hwclock.h

@@ -30,7 +30,7 @@ typedef struct HCL_Instance_Record *HCL_Instance;
 
 /* Create a new HW clock instance */
 extern HCL_Instance HCL_CreateInstance(int min_samples, int max_samples,
-                                       double min_separation);
+                                       double min_separation, double precision);
 
 /* Destroy a HW clock instance */
 extern void HCL_DestroyInstance(HCL_Instance clock);
@@ -38,6 +38,11 @@ extern void HCL_DestroyInstance(HCL_Instance clock);
 /* Check if a new sample should be accumulated at this time */
 extern int HCL_NeedsNewSample(HCL_Instance clock, struct timespec *now);
 
+/* Process new readings of the HW clock in form of (sys, hw, sys) triplets and
+   produce a sample which can be accumulated */
+extern int HCL_ProcessReadings(HCL_Instance clock, int n_readings, struct timespec tss[][3],
+                               struct timespec *hw_ts, struct timespec *local_ts, double *err);
+
 /* Accumulate a new sample */
 extern void HCL_AccumulateSample(HCL_Instance clock, struct timespec *hw_ts,
                                  struct timespec *local_ts, double err);

local.c

@@ -563,6 +563,8 @@ void
 LCL_NotifyExternalTimeStep(struct timespec *raw, struct timespec *cooked,
     double offset, double dispersion)
 {
+  LCL_CancelOffsetCorrection();
+
   /* Dispatch to all handlers */
   invoke_parameter_change_handlers(raw, cooked, 0.0, offset, LCL_ChangeUnknownStep);
 
@@ -628,6 +630,24 @@ LCL_AccumulateFrequencyAndOffset(double dfreq, double doffset, double corr_rate)
 
 /* ================================================== */
 
+int
+LCL_AccumulateFrequencyAndOffsetNoHandlers(double dfreq, double doffset, double corr_rate)
+{
+  ChangeListEntry *first_handler;
+  int r;
+
+  first_handler = change_list.next;
+  change_list.next = &change_list;
+
+  r = LCL_AccumulateFrequencyAndOffset(dfreq, doffset, corr_rate);
+
+  change_list.next = first_handler;
+
+  return r;
+}
+
+/* ================================================== */
+
 void
 lcl_InvokeDispersionNotifyHandlers(double dispersion)
 {

local.h

@@ -173,6 +173,11 @@ extern void LCL_NotifyLeap(int leap);
    a slew, in one easy step */
 extern int LCL_AccumulateFrequencyAndOffset(double dfreq, double doffset, double corr_rate);
 
+/* Same as the routine above, except it does not call the registered
+   parameter change handlers */
+extern int LCL_AccumulateFrequencyAndOffsetNoHandlers(double dfreq, double doffset,
+                                                      double corr_rate);
+
 /* Routine to read the system precision as a log to base 2 value. */
 extern int LCL_GetSysPrecisionAsLog(void);
 

main.c

@@ -166,6 +166,8 @@ signal_cleanup(int x)
 static void
 quit_timeout(void *arg)
 {
+  LOG(LOGS_INFO, "Timeout reached");
+
   /* Return with non-zero status if the clock is not synchronised */
   exit_status = REF_GetOurStratum() >= NTP_MAX_STRATUM;
   SCH_QuitProgram();

ntp_core.c

@@ -3,7 +3,7 @@
 
  **********************************************************************
  * Copyright (C) Richard P. Curnow  1997-2003
- * Copyright (C) Miroslav Lichvar  2009-2020
+ * Copyright (C) Miroslav Lichvar  2009-2022
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -35,6 +35,7 @@
 #include "ntp_ext.h"
 #include "ntp_io.h"
 #include "memory.h"
+#include "quantiles.h"
 #include "sched.h"
 #include "reference.h"
 #include "local.h"
@@ -196,8 +197,12 @@ struct NCR_Instance_Record {
 
   SRC_Instance source;
 
+  /* Optional long-term quantile estimate of peer delay */
+  QNT_Instance delay_quant;
+
   /* Optional median filter for NTP measurements */
   SPF_Instance filter;
+  int filter_count;
 
   int burst_good_samples_to_go;
   int burst_total_samples_to_go;
@@ -265,8 +270,12 @@ static ARR_Instance broadcasts;
 #define MAX_MAXDELAYRATIO 1.0e6
 #define MAX_MAXDELAYDEVRATIO 1.0e6
 
+/* Parameters for the peer delay quantile */
+#define DELAY_QUANT_Q 100
+#define DELAY_QUANT_REPEAT 7
+
 /* Minimum and maximum allowed poll interval */
-#define MIN_POLL -6
+#define MIN_POLL -7
 #define MAX_POLL 24
 
 /* Enable sub-second polling intervals only when the peer delay is not
@@ -318,6 +327,7 @@ static void transmit_timeout(void *arg);
 static double get_transmit_delay(NCR_Instance inst, int on_tx, double last_tx);
 static double get_separation(int poll);
 static int parse_packet(NTP_Packet *packet, int length, NTP_PacketInfo *info);
+static void process_sample(NCR_Instance inst, NTP_Sample *sample);
 static void set_connectivity(NCR_Instance inst, SRC_Connectivity connectivity);
 
 /* ================================================== */
@@ -546,6 +556,16 @@ take_offline(NCR_Instance inst)
 
 /* ================================================== */
 
+static void
+reset_report(NCR_Instance inst)
+{
+  memset(&inst->report, 0, sizeof (inst->report));
+  inst->report.remote_addr = inst->remote_addr.ip_addr;
+  inst->report.remote_port = inst->remote_addr.port;
+}
+
+/* ================================================== */
+
 NCR_Instance
 NCR_CreateInstance(NTP_Remote_Address *remote_addr, NTP_Source_Type type,
                    SourceParameters *params, const char *name)
@@ -640,9 +660,16 @@ NCR_CreateInstance(NTP_Remote_Address *remote_addr, NTP_Source_Type type,
                                          params->min_samples, params->max_samples,
                                          params->min_delay, params->asymmetry);
 
+  if (params->max_delay_quant > 0.0) {
+    int k = round(CLAMP(0.05, params->max_delay_quant, 0.95) * DELAY_QUANT_Q);
+    result->delay_quant = QNT_CreateInstance(k, k, DELAY_QUANT_Q, DELAY_QUANT_REPEAT,
+                                             LCL_GetSysPrecisionAsQuantum() / 2.0);
+  } else {
+    result->delay_quant = NULL;
+  }
+
   if (params->filter_length >= 1)
-    result->filter = SPF_CreateInstance(params->filter_length, params->filter_length,
+    result->filter = SPF_CreateInstance(1, params->filter_length, NTP_MAX_DISPERSION, 0.0);
-                                        NTP_MAX_DISPERSION, 0.0);
   else
     result->filter = NULL;
 
@@ -650,17 +677,18 @@ NCR_CreateInstance(NTP_Remote_Address *remote_addr, NTP_Source_Type type,
   result->tx_timeout_id = 0;
   result->tx_suspended = 1;
   result->opmode = MD_OFFLINE;
-  result->local_poll = result->minpoll;
+  result->local_poll = MAX(result->minpoll, MIN_NONLAN_POLL);
   result->poll_score = 0.0;
   zero_local_timestamp(&result->local_tx);
   result->burst_good_samples_to_go = 0;
   result->burst_total_samples_to_go = 0;
-  memset(&result->report, 0, sizeof (result->report));
   
   NCR_ResetInstance(result);
 
   set_connectivity(result, params->connectivity);
 
+  reset_report(result);
+
   return result;
 }
 
@@ -676,6 +704,8 @@ NCR_DestroyInstance(NCR_Instance instance)
   if (instance->mode == MODE_ACTIVE)
     NIO_CloseServerSocket(instance->local_addr.sock_fd);
 
+  if (instance->delay_quant)
+    QNT_DestroyInstance(instance->delay_quant);
   if (instance->filter)
     SPF_DestroyInstance(instance->filter);
 
@@ -732,8 +762,11 @@ NCR_ResetInstance(NCR_Instance instance)
   UTI_ZeroNtp64(&instance->init_remote_ntp_tx);
   zero_local_timestamp(&instance->init_local_rx);
 
+  if (instance->delay_quant)
+    QNT_Reset(instance->delay_quant);
   if (instance->filter)
     SPF_DropSamples(instance->filter);
+  instance->filter_count = 0;
 }
 
 /* ================================================== */
@@ -755,7 +788,6 @@ NCR_ResetPoll(NCR_Instance instance)
 void
 NCR_ChangeRemoteAddress(NCR_Instance inst, NTP_Remote_Address *remote_addr, int ntp_only)
 {
-  memset(&inst->report, 0, sizeof (inst->report));
   NCR_ResetInstance(inst);
 
   if (!ntp_only)
@@ -776,6 +808,8 @@ NCR_ChangeRemoteAddress(NCR_Instance inst, NTP_Remote_Address *remote_addr, int
   SRC_SetRefid(inst->source, UTI_IPToRefid(&remote_addr->ip_addr),
                &inst->remote_addr.ip_addr);
   SRC_ResetInstance(inst->source);
+
+  reset_report(inst);
 }
 
 /* ================================================== */
@@ -783,6 +817,8 @@ NCR_ChangeRemoteAddress(NCR_Instance inst, NTP_Remote_Address *remote_addr, int
 static void
 adjust_poll(NCR_Instance inst, double adj)
 {
+  NTP_Sample last_sample;
+
   inst->poll_score += adj;
 
   if (inst->poll_score >= 1.0) {
@@ -808,7 +844,9 @@ adjust_poll(NCR_Instance inst, double adj)
      or it is not in a local network according to the measured delay */
   if (inst->local_poll < MIN_NONLAN_POLL &&
       (!SRC_IsReachable(inst->source) ||
-       SST_MinRoundTripDelay(SRC_GetSourcestats(inst->source)) > MAX_LAN_PEER_DELAY))
+       (SST_MinRoundTripDelay(SRC_GetSourcestats(inst->source)) > MAX_LAN_PEER_DELAY &&
+        (!inst->filter || !SPF_GetLastSample(inst->filter, &last_sample) ||
+         last_sample.peer_delay > MAX_LAN_PEER_DELAY))))
     inst->local_poll = MIN_NONLAN_POLL;
 }
 
@@ -1353,6 +1391,9 @@ transmit_timeout(void *arg)
     }
 
     SRC_UpdateReachability(inst->source, 0);
+
+    /* Count missing samples for the sample filter */
+    process_sample(inst, NULL);
   }
 
   /* With auto_offline take the source offline if sending failed */
@@ -1541,6 +1582,22 @@ check_delay_ratio(NCR_Instance inst, SST_Stats stats,
 
 /* ================================================== */
 
+static int
+check_delay_quant(NCR_Instance inst, double delay)
+{
+  double quant;
+
+  quant = QNT_GetQuantile(inst->delay_quant, QNT_GetMinK(inst->delay_quant));
+
+  if (delay <= quant)
+    return 1;
+
+  DEBUG_LOG("maxdelayquant: delay=%e quant=%e", delay, quant);
+  return 0;
+}
+
+/* ================================================== */
+
 static int
 check_delay_dev_ratio(NCR_Instance inst, SST_Stats stats,
                       struct timespec *sample_time, double offset, double delay)
@@ -1626,33 +1683,29 @@ check_sync_loop(NCR_Instance inst, NTP_Packet *message, NTP_Local_Address *local
 static void
 process_sample(NCR_Instance inst, NTP_Sample *sample)
 {
-  double estimated_offset, error_in_estimate, filtered_sample_ago;
+  double estimated_offset, error_in_estimate;
   NTP_Sample filtered_sample;
-  int filtered_samples;
 
-  /* Accumulate the sample to the median filter if it is enabled.  When the
+  /* Accumulate the sample to the median filter if enabled and wait for
-     filter produces a result, check if it is not too old, i.e. the filter did
+     the configured number of samples before processing (NULL indicates
-     not miss too many samples due to missing responses or failing tests. */
+     a missing sample) */
   if (inst->filter) {
-    SPF_AccumulateSample(inst->filter, sample);
+    if (sample)
+      SPF_AccumulateSample(inst->filter, sample);
 
-    filtered_samples = SPF_GetNumberOfSamples(inst->filter);
+    if (++inst->filter_count < SPF_GetMaxSamples(inst->filter))
+      return;
 
     if (!SPF_GetFilteredSample(inst->filter, &filtered_sample))
       return;
 
-    filtered_sample_ago = UTI_DiffTimespecsToDouble(&sample->time, &filtered_sample.time);
-
-    if (filtered_sample_ago > SOURCE_REACH_BITS / 2 * filtered_samples *
-                              UTI_Log2ToDouble(inst->local_poll)) {
-      DEBUG_LOG("filtered sample dropped ago=%f poll=%d", filtered_sample_ago,
-                inst->local_poll);
-      return;
-    }
-
     sample = &filtered_sample;
+    inst->filter_count = 0;
   }
 
+  if (!sample)
+    return;
+
   /* Get the estimated offset predicted from previous samples.  The
      convention here is that positive means local clock FAST of
      reference, i.e. backwards to the way that 'offset' is defined. */
@@ -1907,12 +1960,17 @@ process_response(NCR_Instance inst, NTP_Local_Address *local_addr,
 
     /* Test A requires that the minimum estimate of the peer delay is not
        larger than the configured maximum, in both client modes that the server
-       processing time is sane, and in interleaved symmetric mode that the
+       processing time is sane, in interleaved client/server mode that the
+       previous response was not in basic mode (which prevents using timestamps
+       that minimise delay error), and in interleaved symmetric mode that the
        measured delay and intervals between remote timestamps don't indicate
        a missed response */
     testA = sample.peer_delay - sample.peer_dispersion <= inst->max_delay &&
             precision <= inst->max_delay &&
             !(inst->mode == MODE_CLIENT && response_time > MAX_SERVER_INTERVAL) &&
+            !(inst->mode == MODE_CLIENT && interleaved_packet &&
+              UTI_IsZeroTimespec(&inst->prev_local_tx.ts) &&
+              UTI_CompareTimespecs(&local_transmit.ts, &inst->local_tx.ts) == 0) &&
             !(inst->mode == MODE_ACTIVE && interleaved_packet &&
               (sample.peer_delay > 0.5 * prev_remote_poll_interval ||
                UTI_CompareNtp64(&message->receive_ts, &message->transmit_ts) <= 0 ||
@@ -1925,12 +1983,18 @@ process_response(NCR_Instance inst, NTP_Local_Address *local_addr,
        administrator-defined value */
     testB = check_delay_ratio(inst, stats, &sample.time, sample.peer_delay);
 
-    /* Test C requires that the ratio of the increase in delay from the minimum
+    /* Test C either requires that the delay is less than an estimate of an
+       administrator-defined quantile, or (if the quantile is not specified)
+       it requires that the ratio of the increase in delay from the minimum
        one in the stats data register to the standard deviation of the offsets
        in the register is less than an administrator-defined value or the
        difference between measured offset and predicted offset is larger than
        the increase in delay */
-    testC = check_delay_dev_ratio(inst, stats, &sample.time, sample.offset, sample.peer_delay);
+    if (inst->delay_quant)
+      testC = check_delay_quant(inst, sample.peer_delay);
+    else
+      testC = check_delay_dev_ratio(inst, stats, &sample.time, sample.offset,
+                                    sample.peer_delay);
 
     /* Test D requires that the source is not synchronised to us and is not us
        to prevent a synchronisation loop */
@@ -1972,7 +2036,7 @@ process_response(NCR_Instance inst, NTP_Local_Address *local_addr,
     updated_timestamps = 2;
 
     /* If available, update the monotonic timestamp and accumulate the offset.
-       This needs to be done here to no lose changes in remote_ntp_rx in
+       This needs to be done here to not lose changes in remote_ntp_rx in
        symmetric mode when there are multiple responses per request. */
     if (ef_exp1 && !UTI_IsZeroNtp64(&ef_exp1->mono_receive_ts)) {
       inst->remote_mono_epoch = ntohl(ef_exp1->mono_epoch);
@@ -2063,6 +2127,9 @@ process_response(NCR_Instance inst, NTP_Local_Address *local_addr,
       }
 
       SRC_UpdateStatus(inst->source, MAX(inst->remote_stratum, inst->min_stratum), pkt_leap);
+
+      if (inst->delay_quant)
+        QNT_Accumulate(inst->delay_quant, sample.peer_delay);
     }
 
     if (good_packet) {
@@ -2088,6 +2155,9 @@ process_response(NCR_Instance inst, NTP_Local_Address *local_addr,
     } else {
       /* Slowly increase the polling interval if we can't get a good response */
       adjust_poll(inst, testD ? 0.02 : 0.1);
+
+      /* Count missing samples for the sample filter */
+      process_sample(inst, NULL);
     }
 
     /* If in client mode, no more packets are expected to be coming from the
@@ -2121,9 +2191,7 @@ process_response(NCR_Instance inst, NTP_Local_Address *local_addr,
     }
 
     /* Update the NTP report */
-    inst->report.remote_addr = inst->remote_addr.ip_addr;
     inst->report.local_addr = inst->local_addr.ip_addr;
-    inst->report.remote_port = inst->remote_addr.port;
     inst->report.leap = pkt_leap;
     inst->report.version = pkt_version;
     inst->report.mode = NTP_LVM_TO_MODE(message->lvm);
@@ -2148,6 +2216,8 @@ process_response(NCR_Instance inst, NTP_Local_Address *local_addr,
     inst->report.rx_tss_char = tss_chars[local_receive.source];
 
     inst->report.total_valid_count++;
+    if (good_packet)
+      inst->report.total_good_count++;
   }
 
   /* Do measurement logging */

ntp_io.c

@@ -4,7 +4,7 @@
  **********************************************************************
  * Copyright (C) Richard P. Curnow  1997-2003
  * Copyright (C) Timo Teras  2009
- * Copyright (C) Miroslav Lichvar  2009, 2013-2016, 2018-2020
+ * Copyright (C) Miroslav Lichvar  2009, 2013-2016, 2018-2021
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -594,7 +594,7 @@ NIO_SendPacket(NTP_Packet *packet, NTP_Remote_Address *remote_addr,
 #if !defined(HAVE_IN_PKTINFO) && defined(IP_SENDSRCADDR)
   /* On FreeBSD a local IPv4 address cannot be specified on bound socket */
   if (message.local_addr.ip.family == IPADDR_INET4 &&
-      (local_addr->sock_fd != server_sock_fd4 || bound_server_sock_fd4))
+      (bound_server_sock_fd4 || !NIO_IsServerSocket(local_addr->sock_fd)))
     message.local_addr.ip.family = IPADDR_UNSPEC;
 #endif
 

ntp_io_linux.c

@@ -2,7 +2,7 @@
   chronyd/chronyc - Programs for keeping computer clocks accurate.
 
  **********************************************************************
- * Copyright (C) Miroslav Lichvar  2016-2019
+ * Copyright (C) Miroslav Lichvar  2016-2019, 2021-2022
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -59,8 +59,6 @@ struct Interface {
   /* Start of UDP data at layer 2 for IPv4 and IPv6 */
   int l2_udp4_ntp_start;
   int l2_udp6_ntp_start;
-  /* Precision of PHC readings */
-  double precision;
   /* Compensation of errors in TX and RX timestamping */
   double tx_comp;
   double rx_comp;
@@ -68,7 +66,7 @@ struct Interface {
 };
 
 /* Number of PHC readings per HW clock sample */
-#define PHC_READINGS 10
+#define PHC_READINGS 25
 
 /* Minimum interval between PHC readings */
 #define MIN_PHC_POLL -6
@@ -244,12 +242,12 @@ add_interface(CNF_HwTsInterface *conf_iface)
   iface->l2_udp4_ntp_start = 42;
   iface->l2_udp6_ntp_start = 62;
 
-  iface->precision = conf_iface->precision;
   iface->tx_comp = conf_iface->tx_comp;
   iface->rx_comp = conf_iface->rx_comp;
 
   iface->clock = HCL_CreateInstance(conf_iface->min_samples, conf_iface->max_samples,
-                                    UTI_Log2ToDouble(MAX(conf_iface->minpoll, MIN_PHC_POLL)));
+                                    UTI_Log2ToDouble(MAX(conf_iface->minpoll, MIN_PHC_POLL)),
+                                    conf_iface->precision);
 
   LOG(LOGS_INFO, "Enabled HW timestamping %son %s",
       ts_config.rx_filter == HWTSTAMP_FILTER_NONE ? "(TX only) " : "", iface->name);
@@ -566,19 +564,22 @@ process_hw_timestamp(struct Interface *iface, struct timespec *hw_ts,
                      int l2_length)
 {
   struct timespec sample_phc_ts, sample_sys_ts, sample_local_ts, ts;
+  struct timespec phc_readings[PHC_READINGS][3];
   double rx_correction, ts_delay, phc_err, local_err;
+  int n_readings;
 
   if (HCL_NeedsNewSample(iface->clock, &local_ts->ts)) {
-    if (!SYS_Linux_GetPHCSample(iface->phc_fd, iface->phc_nocrossts, iface->precision,
+    n_readings = SYS_Linux_GetPHCReadings(iface->phc_fd, iface->phc_nocrossts,
-                                &iface->phc_mode, &sample_phc_ts, &sample_sys_ts,
+                                          &iface->phc_mode, PHC_READINGS, phc_readings);
-                                &phc_err))
+    if (n_readings > 0 &&
-      return;
+        HCL_ProcessReadings(iface->clock, n_readings, phc_readings,
+                             &sample_phc_ts, &sample_sys_ts, &phc_err)) {
+      LCL_CookTime(&sample_sys_ts, &sample_local_ts, &local_err);
+      HCL_AccumulateSample(iface->clock, &sample_phc_ts, &sample_local_ts,
+                           phc_err + local_err);
 
-    LCL_CookTime(&sample_sys_ts, &sample_local_ts, &local_err);
+      update_interface_speed(iface);
-    HCL_AccumulateSample(iface->clock, &sample_phc_ts, &sample_local_ts,
+    }
-                         phc_err + local_err);
-
-    update_interface_speed(iface);
   }
 
   /* We need to transpose RX timestamps as hardware timestamps are normally

ntp_sources.c

@@ -353,7 +353,6 @@ add_source(NTP_Remote_Address *remote_addr, char *name, NTP_Source_Type type,
       record_lock = 1;
 
       record = get_record(slot);
-      assert(!name || !UTI_IsStringIP(name));
       record->name = Strdup(name ? name : UTI_IPToString(&remote_addr->ip_addr));
       record->data = NCR_CreateInstance(remote_addr, type, params, record->name);
       record->remote_addr = NCR_GetRemoteAddress(record->data);
@@ -698,21 +697,25 @@ static int get_unused_pool_id(void)
 
 /* ================================================== */
 
+static uint32_t
+get_next_conf_id(uint32_t *conf_id)
+{
+  last_conf_id++;
+
+  if (conf_id)
+    *conf_id = last_conf_id;
+
+  return last_conf_id;
+}
+
+/* ================================================== */
+
 NSR_Status
 NSR_AddSource(NTP_Remote_Address *remote_addr, NTP_Source_Type type,
               SourceParameters *params, uint32_t *conf_id)
 {
-  NSR_Status s;
+  return add_source(remote_addr, NULL, type, params, INVALID_POOL,
-
+                    get_next_conf_id(conf_id));
-  s = add_source(remote_addr, NULL, type, params, INVALID_POOL, last_conf_id + 1);
-  if (s != NSR_Success)
-    return s;
-
-  last_conf_id++;
-  if (conf_id)
-    *conf_id = last_conf_id;
-
-  return s;
 }
 
 /* ================================================== */
@@ -725,11 +728,13 @@ NSR_AddSourceByName(char *name, int port, int pool, NTP_Source_Type type,
   struct SourcePool *sp;
   NTP_Remote_Address remote_addr;
   int i, new_sources, pool_id;
+  uint32_t cid;
 
   /* If the name is an IP address, add the source with the address directly */
   if (UTI_StringToIP(name, &remote_addr.ip_addr)) {
     remote_addr.port = port;
-    return NSR_AddSource(&remote_addr, type, params, conf_id);
+    return add_source(&remote_addr, name, type, params, INVALID_POOL,
+                      get_next_conf_id(conf_id));
   }
 
   /* Make sure the name is at least printable and has no spaces */
@@ -770,14 +775,12 @@ NSR_AddSourceByName(char *name, int port, int pool, NTP_Source_Type type,
 
   append_unresolved_source(us);
 
-  last_conf_id++;
+  cid = get_next_conf_id(conf_id);
-  if (conf_id)
-    *conf_id = last_conf_id;
 
   for (i = 0; i < new_sources; i++) {
     if (i > 0)
       remote_addr.ip_addr.addr.id = ++last_address_id;
-    if (add_source(&remote_addr, name, type, params, us->pool_id, last_conf_id) != NSR_Success)
+    if (add_source(&remote_addr, name, type, params, us->pool_id, cid) != NSR_Success)
       return NSR_TooManySources;
   }
 
@@ -1100,8 +1103,10 @@ NSR_ProcessRx(NTP_Remote_Address *remote_addr, NTP_Local_Address *local_addr,
 
   assert(initialised);
 
-  /* Must match IP address AND port number */
+  /* Avoid unnecessary lookup if the packet cannot be a response from our
-  if (find_slot2(remote_addr, &slot) == 2) {
+     source.  Otherwise, it must match both IP address and port number. */
+  if (NTP_LVM_TO_MODE(message->lvm) != MODE_CLIENT &&
+      find_slot2(remote_addr, &slot) == 2) {
     record = get_record(slot);
 
     if (!NCR_ProcessRxKnown(record->data, local_addr, rx_ts, message, length))
@@ -1137,8 +1142,10 @@ NSR_ProcessTx(NTP_Remote_Address *remote_addr, NTP_Local_Address *local_addr,
   SourceRecord *record;
   int slot;
 
-  /* Must match IP address AND port number */
+  /* Avoid unnecessary lookup if the packet cannot be a request to our
-  if (find_slot2(remote_addr, &slot) == 2) {
+     source.  Otherwise, it must match both IP address and port number. */
+  if (NTP_LVM_TO_MODE(message->lvm) != MODE_SERVER &&
+      find_slot2(remote_addr, &slot) == 2) {
     record = get_record(slot);
     NCR_ProcessTxKnown(record->data, local_addr, tx_ts, message, length);
   } else {

nts_ke_session.c

@@ -594,13 +594,13 @@ handle_step(struct timespec *raw, struct timespec *cooked, double dfreq,
 
 static int gnutls_initialised = 0;
 
-static void
+static int
 init_gnutls(void)
 {
   int r;
 
   if (gnutls_initialised)
-    return;
+    return 1;
 
   r = gnutls_global_init();
   if (r < 0)
@@ -611,8 +611,12 @@ init_gnutls(void)
   r = gnutls_priority_init2(&priority_cache,
                             "-VERS-SSL3.0:-VERS-TLS1.0:-VERS-TLS1.1:-VERS-TLS1.2:-VERS-DTLS-ALL",
                             NULL, GNUTLS_PRIORITY_INIT_DEF_APPEND);
-  if (r < 0)
+  if (r < 0) {
-    LOG_FATAL("Could not initialise %s : %s", "priority cache", gnutls_strerror(r));
+    LOG(LOGS_ERR, "Could not initialise %s : %s",
+        "priority cache for TLS", gnutls_strerror(r));
+    gnutls_global_deinit();
+    return 0;
+  }
 
   /* Use our clock instead of the system clock in certificate verification */
   gnutls_global_set_time_function(get_time);
@@ -621,6 +625,8 @@ init_gnutls(void)
   DEBUG_LOG("Initialised");
 
   LCL_AddParameterChangeHandler(handle_step, NULL);
+
+  return 1;
 }
 
 /* ================================================== */
@@ -649,7 +655,8 @@ create_credentials(const char **certs, const char **keys, int n_certs_keys,
   gnutls_certificate_credentials_t credentials = NULL;
   int i, r;
 
-  init_gnutls();
+  if (!init_gnutls())
+    return NULL;
 
   r = gnutls_certificate_allocate_credentials(&credentials);
   if (r < 0)

quantiles.c

@@ -0,0 +1,209 @@
+/*
+  chronyd/chronyc - Programs for keeping computer clocks accurate.
+
+ **********************************************************************
+ * Copyright (C) Miroslav Lichvar  2022
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+ *
+ **********************************************************************
+
+  =======================================================================
+
+  Estimation of quantiles using the Frugal-2U streaming algorithm
+  (https://arxiv.org/pdf/1407.1121v1.pdf)
+  */
+
+#include "config.h"
+
+#include "logging.h"
+#include "memory.h"
+#include "quantiles.h"
+#include "regress.h"
+#include "util.h"
+
+/* Maximum number of repeated estimates for stabilisation */
+#define MAX_REPEAT 64
+
+struct Quantile {
+  double est;
+  double step;
+  int sign;
+};
+
+struct QNT_Instance_Record {
+  struct Quantile *quants;
+  int n_quants;
+  int repeat;
+  int q;
+  int min_k;
+  double min_step;
+  int n_set;
+};
+
+/* ================================================== */
+
+QNT_Instance
+QNT_CreateInstance(int min_k, int max_k, int q, int repeat, double min_step)
+{
+  QNT_Instance inst;
+  long seed;
+
+  if (q < 2 || min_k > max_k || min_k < 1 || max_k >= q ||
+      repeat < 1 || repeat > MAX_REPEAT || min_step <= 0.0)
+    assert(0);
+
+  inst = MallocNew(struct QNT_Instance_Record);
+  inst->n_quants = (max_k - min_k + 1) * repeat;
+  inst->quants = MallocArray(struct Quantile, inst->n_quants);
+  inst->repeat = repeat;
+  inst->q = q;
+  inst->min_k = min_k;
+  inst->min_step = min_step;
+
+  QNT_Reset(inst);
+
+  /* Seed the random number generator, which will not be isolated from
+     other instances and other random() users */
+  UTI_GetRandomBytes(&seed, sizeof (seed));
+  srandom(seed);
+
+  return inst;
+}
+
+/* ================================================== */
+
+void
+QNT_DestroyInstance(QNT_Instance inst)
+{
+  Free(inst->quants);
+  Free(inst);
+}
+
+/* ================================================== */
+
+void
+QNT_Reset(QNT_Instance inst)
+{
+  int i;
+
+  inst->n_set = 0;
+
+  for (i = 0; i < inst->n_quants; i++) {
+    inst->quants[i].est = 0.0;
+    inst->quants[i].step = inst->min_step;
+    inst->quants[i].sign = 1;
+  }
+}
+
+/* ================================================== */
+
+static void
+insert_initial_value(QNT_Instance inst, double value)
+{
+  int i, j, r = inst->repeat;
+
+  if (inst->n_set * r >= inst->n_quants)
+    assert(0);
+
+  /* Keep the initial estimates repeated and ordered */
+  for (i = inst->n_set; i > 0 && inst->quants[(i - 1) * r].est > value; i--) {
+    for (j = 0; j < r; j++)
+      inst->quants[i * r + j].est = inst->quants[(i - 1) * r].est;
+  }
+
+  for (j = 0; j < r; j++)
+    inst->quants[i * r + j].est = value;
+  inst->n_set++;
+
+  /* Duplicate the largest value in unset quantiles */
+  for (i = inst->n_set * r; i < inst->n_quants; i++)
+    inst->quants[i].est = inst->quants[i - 1].est;
+}
+
+/* ================================================== */
+
+static void
+update_estimate(struct Quantile *quantile, double value, double p, double rand,
+                double min_step)
+{
+  if (value > quantile->est && rand > (1.0 - p)) {
+    quantile->step += quantile->sign > 0 ? min_step : -min_step;
+    quantile->est += quantile->step > 0.0 ? fabs(quantile->step) : min_step;
+    if (quantile->est > value) {
+      quantile->step += value - quantile->est;
+      quantile->est = value;
+    }
+    if (quantile->sign < 0 && quantile->step > min_step)
+      quantile->step = min_step;
+    quantile->sign = 1;
+  } else if (value < quantile->est && rand > p) {
+    quantile->step += quantile->sign < 0 ? min_step : -min_step;
+    quantile->est -= quantile->step > 0.0 ? fabs(quantile->step) : min_step;
+    if (quantile->est < value) {
+      quantile->step += quantile->est - value;
+      quantile->est = value;
+    }
+    if (quantile->sign > 0 && quantile->step > min_step)
+      quantile->step = min_step;
+    quantile->sign = -1;
+  }
+}
+
+/* ================================================== */
+
+void
+QNT_Accumulate(QNT_Instance inst, double value)
+{
+  double p, rand;
+  int i;
+
+  /* Initialise the estimates with first received values */
+  if (inst->n_set * inst->repeat < inst->n_quants) {
+    insert_initial_value(inst, value);
+    return;
+  }
+
+  for (i = 0; i < inst->n_quants; i++) {
+    p = (double)(i / inst->repeat + inst->min_k) / inst->q;
+    rand = (double)random() / ((1U << 31) - 1);
+
+    update_estimate(&inst->quants[i], value, p, rand, inst->min_step);
+  }
+}
+
+/* ================================================== */
+
+int
+QNT_GetMinK(QNT_Instance inst)
+{
+  return inst->min_k;
+}
+
+/* ================================================== */
+
+double
+QNT_GetQuantile(QNT_Instance inst, int k)
+{
+  double estimates[MAX_REPEAT];
+  int i;
+
+  if (k < inst->min_k || k - inst->min_k >= inst->n_quants)
+    assert(0);
+
+  for (i = 0; i < inst->repeat; i++)
+    estimates[i] = inst->quants[(k - inst->min_k) * inst->repeat + i].est;
+
+  return RGR_FindMedian(estimates, inst->repeat);
+}

quantiles.h

@@ -0,0 +1,41 @@
+/*
+  chronyd/chronyc - Programs for keeping computer clocks accurate.
+
+ **********************************************************************
+ * Copyright (C) Miroslav Lichvar  2022
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+ *
+ **********************************************************************
+
+  =======================================================================
+
+  Header file for estimation of quantiles.
+
+  */
+
+#ifndef GOT_QUANTILES_H
+#define GOT_QUANTILES_H
+
+typedef struct QNT_Instance_Record *QNT_Instance;
+
+extern QNT_Instance QNT_CreateInstance(int min_k, int max_k, int q, int repeat, double min_step);
+extern void QNT_DestroyInstance(QNT_Instance inst);
+
+extern void QNT_Reset(QNT_Instance inst);
+extern void QNT_Accumulate(QNT_Instance inst, double value);
+extern int QNT_GetMinK(QNT_Instance inst);
+extern double QNT_GetQuantile(QNT_Instance inst, int k);
+
+#endif

refclock.c

@@ -2,7 +2,7 @@
   chronyd/chronyc - Programs for keeping computer clocks accurate.
 
  **********************************************************************
- * Copyright (C) Miroslav Lichvar  2009-2011, 2013-2014, 2016-2019
+ * Copyright (C) Miroslav Lichvar  2009-2011, 2013-2014, 2016-2019, 2022
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -55,21 +55,6 @@ struct FilterSample {
   struct timespec sample_time;
 };
 
-struct MedianFilter {
-  int length;
-  int index;
-  int used;
-  int last;
-  int avg_var_n;
-  double avg_var;
-  double max_var;
-  struct FilterSample *samples;
-  int *selected;
-  double *x_data;
-  double *y_data;
-  double *w_data;
-};
-
 struct RCL_Instance_Record {
   RefclockDriver *driver;
   void *data;
@@ -79,6 +64,7 @@ struct RCL_Instance_Record {
   int driver_polled;
   int poll;
   int leap_status;
+  int local;
   int pps_forced;
   int pps_rate;
   int pps_active;
@@ -190,6 +176,7 @@ RCL_AddRefclock(RefclockParameters *params)
   inst->poll = params->poll;
   inst->driver_polled = 0;
   inst->leap_status = LEAP_Normal;
+  inst->local = params->local;
   inst->pps_forced = params->pps_forced;
   inst->pps_rate = params->pps_rate;
   inst->pps_active = 0;
@@ -231,6 +218,13 @@ RCL_AddRefclock(RefclockParameters *params)
     inst->ref_id = (uint32_t)ref[0] << 24 | ref[1] << 16 | ref[2] << 8 | ref[3];
   }
 
+  if (inst->local) {
+    inst->pps_forced = 1;
+    inst->lock_ref = inst->ref_id;
+    inst->leap_status = LEAP_Unsynchronised;
+    inst->max_lock_age = MAX(inst->max_lock_age, 3);
+  }
+
   if (inst->driver->poll) {
     int max_samples;
 
@@ -300,7 +294,7 @@ RCL_StartRefclocks(void)
         break;
       }
 
-      if (lock_index == -1 || lock_index == i)
+      if (lock_index == -1 || (lock_index == i && !inst->local))
         LOG(LOGS_WARN, "Invalid lock refid %s", UTI_RefidToString(inst->lock_ref));
     }
 
@@ -440,20 +434,24 @@ accumulate_sample(RCL_Instance instance, struct timespec *sample_time, double of
 }
 
 int
-RCL_AddSample(RCL_Instance instance, struct timespec *sample_time, double offset, int leap)
+RCL_AddSample(RCL_Instance instance, struct timespec *sample_time,
+              struct timespec *ref_time, int leap)
 {
-  double correction, dispersion;
+  double correction, dispersion, raw_offset, offset;
   struct timespec cooked_time;
 
   if (instance->pps_forced)
-    return RCL_AddPulse(instance, sample_time, -offset);
+    return RCL_AddPulse(instance, sample_time,
+                        1.0e-9 * (sample_time->tv_nsec - ref_time->tv_nsec));
+
+  raw_offset = UTI_DiffTimespecsToDouble(ref_time, sample_time);
 
   LCL_GetOffsetCorrection(sample_time, &correction, &dispersion);
   UTI_AddDoubleToTimespec(sample_time, correction, &cooked_time);
   dispersion += instance->precision;
 
   /* Make sure the timestamp and offset provided by the driver are sane */
-  if (!UTI_IsTimeOffsetSane(sample_time, offset) ||
+  if (!UTI_IsTimeOffsetSane(sample_time, raw_offset) ||
       !valid_sample_time(instance, &cooked_time))
     return 0;
 
@@ -468,18 +466,24 @@ RCL_AddSample(RCL_Instance instance, struct timespec *sample_time, double offset
       return 0;
   }
 
+  /* Calculate offset = raw_offset - correction + instance->offset
+     in parts to avoid loss of precision if there are large differences */
+  offset = ref_time->tv_sec - sample_time->tv_sec -
+           (time_t)correction + (time_t)instance->offset;
+  offset += 1.0e-9 * (ref_time->tv_nsec - sample_time->tv_nsec) -
+            (correction - (time_t)correction) + (instance->offset - (time_t)instance->offset);
+
   if (instance->tai && !convert_tai_offset(sample_time, &offset)) {
     DEBUG_LOG("refclock sample ignored unknown TAI offset");
     return 0;
   }
 
-  if (!accumulate_sample(instance, &cooked_time,
+  if (!accumulate_sample(instance, &cooked_time, offset, dispersion))
-                         offset - correction + instance->offset, dispersion))
     return 0;
 
   instance->pps_active = 0;
 
-  log_sample(instance, &cooked_time, 0, 0, offset, offset - correction + instance->offset, dispersion);
+  log_sample(instance, &cooked_time, 0, 0, raw_offset, offset, dispersion);
 
   /* for logging purposes */
   if (!instance->driver->poll)
@@ -558,16 +562,30 @@ RCL_AddCookedPulse(RCL_Instance instance, struct timespec *cooked_time,
     lock_refclock = get_refclock(instance->lock_ref);
 
     if (!SPF_GetLastSample(lock_refclock->filter, &ref_sample)) {
-      DEBUG_LOG("refclock pulse ignored no ref sample");
+      if (instance->local) {
-      return 0;
+        /* Make the first sample in order to lock to itself */
+        ref_sample.time = *cooked_time;
+        ref_sample.offset = offset;
+        ref_sample.peer_delay = ref_sample.peer_dispersion = 0;
+        ref_sample.root_delay = ref_sample.root_dispersion = 0;
+      } else {
+        DEBUG_LOG("refclock pulse ignored no ref sample");
+        return 0;
+      }
     }
 
     ref_sample.root_dispersion += SPF_GetAvgSampleDispersion(lock_refclock->filter);
 
     sample_diff = UTI_DiffTimespecsToDouble(cooked_time, &ref_sample.time);
     if (fabs(sample_diff) >= (double)instance->max_lock_age / rate) {
-      DEBUG_LOG("refclock pulse ignored samplediff=%.9f",
+      DEBUG_LOG("refclock pulse ignored samplediff=%.9f", sample_diff);
-          sample_diff);
+
+      /* Restart the local mode */
+      if (instance->local) {
+        LOG(LOGS_WARN, "Local refclock lost lock");
+        SPF_DropSamples(instance->filter);
+        SRC_ResetInstance(instance->source);
+      }
       return 0;
     }
 
@@ -693,6 +711,52 @@ pps_stratum(RCL_Instance instance, struct timespec *ts)
   return 0;
 }
 
+static void
+get_local_stats(RCL_Instance inst, struct timespec *ref, double *freq, double *offset)
+{
+  double offset_sd, freq_sd, skew, root_delay, root_disp;
+  SST_Stats stats = SRC_GetSourcestats(inst->source);
+
+  if (SST_Samples(stats) < SST_GetMinSamples(stats)) {
+    UTI_ZeroTimespec(ref);
+    return;
+  }
+
+  SST_GetTrackingData(stats, ref, offset, &offset_sd, freq, &freq_sd,
+                      &skew, &root_delay, &root_disp);
+}
+
+static void
+follow_local(RCL_Instance inst, struct timespec *prev_ref_time, double prev_freq,
+             double prev_offset)
+{
+  SST_Stats stats = SRC_GetSourcestats(inst->source);
+  double freq, dfreq, offset, doffset, elapsed;
+  struct timespec now, ref_time;
+
+  get_local_stats(inst, &ref_time, &freq, &offset);
+
+  if (UTI_IsZeroTimespec(prev_ref_time) || UTI_IsZeroTimespec(&ref_time))
+    return;
+
+  dfreq = (freq - prev_freq) / (1.0 - prev_freq);
+  elapsed = UTI_DiffTimespecsToDouble(&ref_time, prev_ref_time);
+  doffset = offset - elapsed * prev_freq - prev_offset;
+
+  if (!REF_AdjustReference(doffset, dfreq))
+    return;
+
+  LCL_ReadCookedTime(&now, NULL);
+  SST_SlewSamples(stats, &now, dfreq, doffset);
+  SPF_SlewSamples(inst->filter, &now, dfreq, doffset);
+
+  /* Keep the offset close to zero to not lose precision */
+  if (fabs(offset) >= 1.0) {
+    SST_CorrectOffset(stats, -round(offset));
+    SPF_CorrectOffset(inst->filter, -round(offset));
+  }
+}
+
 static void
 poll_timeout(void *arg)
 {
@@ -713,17 +777,28 @@ poll_timeout(void *arg)
     inst->driver_polled = 0;
 
     if (SPF_GetFilteredSample(inst->filter, &sample)) {
+      double local_freq, local_offset;
+      struct timespec local_ref_time;
+
       /* Handle special case when PPS is used with the local reference */
       if (inst->pps_active && inst->lock_ref == -1)
         stratum = pps_stratum(inst, &sample.time);
       else
         stratum = inst->stratum;
 
+      if (inst->local) {
+        get_local_stats(inst, &local_ref_time, &local_freq, &local_offset);
+        inst->leap_status = LEAP_Unsynchronised;
+      }
+
       SRC_UpdateReachability(inst->source, 1);
       SRC_UpdateStatus(inst->source, stratum, inst->leap_status);
       SRC_AccumulateSample(inst->source, &sample);
       SRC_SelectSource(inst->source);
 
+      if (inst->local)
+        follow_local(inst, &local_ref_time, local_freq, local_offset);
+
       log_sample(inst, &sample.time, 1, 0, 0.0, sample.offset, sample.peer_dispersion);
     } else {
       SRC_UpdateReachability(inst->source, 0);

refclock.h

@@ -37,6 +37,7 @@ typedef struct {
   int driver_poll;
   int poll;
   int filter_length;
+  int local;
   int pps_forced;
   int pps_rate;
   int min_samples;
@@ -74,7 +75,8 @@ extern void *RCL_GetDriverData(RCL_Instance instance);
 extern char *RCL_GetDriverParameter(RCL_Instance instance);
 extern void RCL_CheckDriverOptions(RCL_Instance instance, const char **options);
 extern char *RCL_GetDriverOption(RCL_Instance instance, char *name);
-extern int RCL_AddSample(RCL_Instance instance, struct timespec *sample_time, double offset, int leap);
+extern int RCL_AddSample(RCL_Instance instance, struct timespec *sample_time,
+                         struct timespec *ref_time, int leap);
 extern int RCL_AddPulse(RCL_Instance instance, struct timespec *pulse_time, double second);
 extern int RCL_AddCookedPulse(RCL_Instance instance, struct timespec *cooked_time,
                               double second, double dispersion, double raw_correction);

refclock_phc.c

@@ -75,13 +75,15 @@ static int phc_initialise(RCL_Instance instance)
   phc->nocrossts = RCL_GetDriverOption(instance, "nocrossts") ? 1 : 0;
   phc->extpps = RCL_GetDriverOption(instance, "extpps") ? 1 : 0;
 
+  phc->clock = HCL_CreateInstance(0, 16, UTI_Log2ToDouble(RCL_GetDriverPoll(instance)),
+                                  RCL_GetPrecision(instance));
+
   if (phc->extpps) {
     s = RCL_GetDriverOption(instance, "pin");
     phc->pin = s ? atoi(s) : 0;
     s = RCL_GetDriverOption(instance, "channel");
     phc->channel = s ? atoi(s) : 0;
     rising_edge = RCL_GetDriverOption(instance, "clear") ? 0 : 1;
-    phc->clock = HCL_CreateInstance(0, 16, UTI_Log2ToDouble(RCL_GetDriverPoll(instance)));
 
     if (!SYS_Linux_SetPHCExtTimestamping(phc->fd, phc->pin, phc->channel,
                                          rising_edge, !rising_edge, 1))
@@ -90,7 +92,6 @@ static int phc_initialise(RCL_Instance instance)
     SCH_AddFileHandler(phc->fd, SCH_FILE_INPUT, read_ext_pulse, instance);
   } else {
     phc->pin = phc->channel = 0;
-    phc->clock = NULL;
   }
 
   RCL_SetDriverData(instance, phc);
@@ -106,9 +107,9 @@ static void phc_finalise(RCL_Instance instance)
   if (phc->extpps) {
     SCH_RemoveFileHandler(phc->fd);
     SYS_Linux_SetPHCExtTimestamping(phc->fd, phc->pin, phc->channel, 0, 0, 0);
-    HCL_DestroyInstance(phc->clock);
   }
 
+  HCL_DestroyInstance(phc->clock);
   close(phc->fd);
   Free(phc);
 }
@@ -139,29 +140,35 @@ static void read_ext_pulse(int fd, int event, void *anything)
                      UTI_DiffTimespecsToDouble(&phc_ts, &local_ts));
 }
 
+#define PHC_READINGS 25
+
 static int phc_poll(RCL_Instance instance)
 {
+  struct timespec phc_ts, sys_ts, local_ts, readings[PHC_READINGS][3];
   struct phc_instance *phc;
-  struct timespec phc_ts, sys_ts, local_ts;
+  double phc_err, local_err;
-  double offset, phc_err, local_err;
+  int n_readings;
 
   phc = (struct phc_instance *)RCL_GetDriverData(instance);
 
-  if (!SYS_Linux_GetPHCSample(phc->fd, phc->nocrossts, RCL_GetPrecision(instance),
+  n_readings = SYS_Linux_GetPHCReadings(phc->fd, phc->nocrossts, &phc->mode,
-                              &phc->mode, &phc_ts, &sys_ts, &phc_err))
+                                        PHC_READINGS, readings);
+  if (n_readings < 1)
     return 0;
 
-  if (phc->extpps) {
+  if (!HCL_ProcessReadings(phc->clock, n_readings, readings, &phc_ts, &sys_ts, &phc_err))
-    LCL_CookTime(&sys_ts, &local_ts, &local_err);
-    HCL_AccumulateSample(phc->clock, &phc_ts, &local_ts, phc_err + local_err);
     return 0;
-  }
 
-  offset = UTI_DiffTimespecsToDouble(&phc_ts, &sys_ts);
+  LCL_CookTime(&sys_ts, &local_ts, &local_err);
+  HCL_AccumulateSample(phc->clock, &phc_ts, &local_ts, phc_err + local_err);
 
-  DEBUG_LOG("PHC offset: %+.9f err: %.9f", offset, phc_err);
+  if (phc->extpps)
+    return 0;
 
-  return RCL_AddSample(instance, &sys_ts, offset, LEAP_Normal);
+  DEBUG_LOG("PHC offset: %+.9f err: %.9f",
+            UTI_DiffTimespecsToDouble(&phc_ts, &sys_ts), phc_err);
+
+  return RCL_AddSample(instance, &sys_ts, &phc_ts, LEAP_Normal);
 }
 
 RefclockDriver RCL_PHC_driver = {

refclock_shm.c

@@ -95,7 +95,6 @@ static int shm_poll(RCL_Instance instance)
 {
   struct timespec receive_ts, clock_ts;
   struct shmTime t, *shm;
-  double offset;
 
   shm = (struct shmTime *)RCL_GetDriverData(instance);
 
@@ -124,9 +123,8 @@ static int shm_poll(RCL_Instance instance)
 
   UTI_NormaliseTimespec(&clock_ts);
   UTI_NormaliseTimespec(&receive_ts);
-  offset = UTI_DiffTimespecsToDouble(&clock_ts, &receive_ts);
 
-  return RCL_AddSample(instance, &receive_ts, offset, t.leap);
+  return RCL_AddSample(instance, &receive_ts, &clock_ts, t.leap);
 }
 
 RefclockDriver RCL_SHM_driver = {

refclock_sock.c

@@ -60,8 +60,8 @@ struct sock_sample {
 
 static void read_sample(int sockfd, int event, void *anything)
 {
+  struct timespec sys_ts, ref_ts;
   struct sock_sample sample;
-  struct timespec ts;
   RCL_Instance instance;
   int s;
 
@@ -86,13 +86,18 @@ static void read_sample(int sockfd, int event, void *anything)
     return;
   }
 
-  UTI_TimevalToTimespec(&sample.tv, &ts);
+  UTI_TimevalToTimespec(&sample.tv, &sys_ts);
-  UTI_NormaliseTimespec(&ts);
+  UTI_NormaliseTimespec(&sys_ts);
+
+  if (!UTI_IsTimeOffsetSane(&sys_ts, sample.offset))
+    return;
+
+  UTI_AddDoubleToTimespec(&sys_ts, sample.offset, &ref_ts);
 
   if (sample.pulse) {
-    RCL_AddPulse(instance, &ts, sample.offset);
+    RCL_AddPulse(instance, &sys_ts, sample.offset);
   } else {
-    RCL_AddSample(instance, &ts, sample.offset, sample.leap);
+    RCL_AddSample(instance, &sys_ts, &ref_ts, sample.leap);
   }
 }
 

reference.c

@@ -3,7 +3,7 @@
 
  **********************************************************************
  * Copyright (C) Richard P. Curnow  1997-2003
- * Copyright (C) Miroslav Lichvar  2009-2018, 2020
+ * Copyright (C) Miroslav Lichvar  2009-2018, 2020, 2022
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -150,6 +150,9 @@ static SCH_TimeoutID fb_drift_timeout_id;
 static double last_ref_update;
 static double last_ref_update_interval;
 
+static double last_ref_adjustment;
+static int ref_adjustments;
+
 /* ================================================== */
 
 static NTP_Leap get_tz_leap(time_t when, int *tai_offset);
@@ -286,6 +289,8 @@ REF_Initialise(void)
   UTI_ZeroTimespec(&our_ref_time);
   last_ref_update = 0.0;
   last_ref_update_interval = 0.0;
+  last_ref_adjustment = 0.0;
+  ref_adjustments = 0;
 
   LCL_AddParameterChangeHandler(handle_slew, NULL);
 
@@ -960,6 +965,27 @@ fuzz_ref_time(struct timespec *ts)
 
 /* ================================================== */
 
+static double
+get_correction_rate(double offset_sd, double update_interval)
+{
+  /* We want to correct the offset quickly, but we also want to keep the
+     frequency error caused by the correction itself low.
+
+     Define correction rate as the area of the region bounded by the graph of
+     offset corrected in time.  Set the rate so that the time needed to correct
+     an offset equal to the current sourcestats stddev will be equal to the
+     update interval multiplied by the correction time ratio (assuming linear
+     adjustment).  The offset and the time needed to make the correction are
+     inversely proportional.
+
+     This is only a suggestion and it's up to the system driver how the
+     adjustment will be executed. */
+
+  return correction_time_ratio * 0.5 * offset_sd * update_interval;
+}
+
+/* ================================================== */
+
 void
 REF_SetReference(int stratum, NTP_Leap leap, int combined_sources,
                  uint32_t ref_id, IPAddr *ref_ip, struct timespec *ref_time,
@@ -969,7 +995,7 @@ REF_SetReference(int stratum, NTP_Leap leap, int combined_sources,
 {
   double uncorrected_offset, accumulate_offset, step_offset;
   double residual_frequency, local_abs_frequency;
-  double elapsed, mono_now, update_interval, correction_rate, orig_root_distance;
+  double elapsed, mono_now, update_interval, orig_root_distance;
   struct timespec now, raw_now;
   int manual;
 
@@ -1024,21 +1050,6 @@ REF_SetReference(int stratum, NTP_Leap leap, int combined_sources,
   last_ref_update_interval = update_interval;
   last_offset = offset;
 
-  /* We want to correct the offset quickly, but we also want to keep the
-     frequency error caused by the correction itself low.
-
-     Define correction rate as the area of the region bounded by the graph of
-     offset corrected in time. Set the rate so that the time needed to correct
-     an offset equal to the current sourcestats stddev will be equal to the
-     update interval multiplied by the correction time ratio (assuming linear
-     adjustment). The offset and the time needed to make the correction are
-     inversely proportional.
-
-     This is only a suggestion and it's up to the system driver how the
-     adjustment will be executed. */
-
-  correction_rate = correction_time_ratio * 0.5 * offset_sd * update_interval;
-
   /* Check if the clock should be stepped */
   if (is_step_limit_reached(offset, uncorrected_offset)) {
     /* Cancel the uncorrected offset and correct the total offset by step */
@@ -1050,7 +1061,8 @@ REF_SetReference(int stratum, NTP_Leap leap, int combined_sources,
   }
 
   /* Adjust the clock */
-  LCL_AccumulateFrequencyAndOffset(frequency, accumulate_offset, correction_rate);
+  LCL_AccumulateFrequencyAndOffset(frequency, accumulate_offset,
+                                   get_correction_rate(offset_sd, update_interval));
     
   maybe_log_offset(offset, raw_now.tv_sec);
 
@@ -1095,6 +1107,27 @@ REF_SetReference(int stratum, NTP_Leap leap, int combined_sources,
       avg2_moving = 1;
     avg2_offset = SQUARE(offset);
   }
+
+  ref_adjustments = 0;
+}
+
+/* ================================================== */
+
+int
+REF_AdjustReference(double offset, double frequency)
+{
+  double adj_corr_rate, ref_corr_rate, mono_now;
+
+  mono_now = SCH_GetLastEventMonoTime();
+  ref_adjustments++;
+
+  adj_corr_rate = get_correction_rate(fabs(offset), mono_now - last_ref_adjustment);
+  ref_corr_rate = get_correction_rate(our_offset_sd, last_ref_update_interval) /
+                  ref_adjustments;
+  last_ref_adjustment = mono_now;
+
+  return LCL_AccumulateFrequencyAndOffsetNoHandlers(frequency, offset,
+                                                    MAX(adj_corr_rate, ref_corr_rate));
 }
 
 /* ================================================== */

reference.h

@@ -162,6 +162,10 @@ extern void REF_SetManualReference
 extern void
 REF_SetUnsynchronised(void);
 
+/* Make a small correction of the clock without updating the reference
+   parameters and calling the clock change handlers */
+extern int REF_AdjustReference(double offset, double frequency);
+
 /* Announce a leap second before the full reference update */
 extern void REF_UpdateLeapStatus(NTP_Leap leap);
 

reports.h

@@ -174,6 +174,7 @@ typedef struct {
   uint32_t total_tx_count;
   uint32_t total_rx_count;
   uint32_t total_valid_count;
+  uint32_t total_good_count;
 } RPT_NTPReport;
 
 typedef struct {

samplefilt.c

@@ -162,6 +162,14 @@ SPF_GetNumberOfSamples(SPF_Instance filter)
 
 /* ================================================== */
 
+int
+SPF_GetMaxSamples(SPF_Instance filter)
+{
+  return filter->max_samples;
+}
+
+/* ================================================== */
+
 double
 SPF_GetAvgSampleDispersion(SPF_Instance filter)
 {
@@ -170,11 +178,21 @@ SPF_GetAvgSampleDispersion(SPF_Instance filter)
 
 /* ================================================== */
 
-void
+static void
-SPF_DropSamples(SPF_Instance filter)
+drop_samples(SPF_Instance filter, int keep_last)
 {
   filter->index = -1;
   filter->used = 0;
+  if (!keep_last)
+    filter->last = -1;
+}
+
+/* ================================================== */
+
+void
+SPF_DropSamples(SPF_Instance filter)
+{
+  drop_samples(filter, 0);
 }
 
 /* ================================================== */
@@ -399,17 +417,40 @@ SPF_GetFilteredSample(SPF_Instance filter, NTP_Sample *sample)
 
   n = select_samples(filter);
 
+  DEBUG_LOG("selected %d from %d samples", n, filter->used);
+
   if (n < 1)
     return 0;
 
   if (!combine_selected_samples(filter, n, sample))
     return 0;
 
-  SPF_DropSamples(filter);
+  drop_samples(filter, 1);
 
   return 1;
 }
 
+/* ================================================== */
+
+static int
+get_first_last(SPF_Instance filter, int *first, int *last)
+{
+  if (filter->last < 0)
+    return 0;
+
+  /* Always slew the last sample as it may be returned even if no new
+     samples were accumulated */
+  if (filter->used > 0) {
+    *first = 0;
+    *last = filter->used - 1;
+  } else {
+    *first = *last = filter->last;
+  }
+
+  return 1;
+}
+
+
 /* ================================================== */
 
 void
@@ -418,18 +459,9 @@ SPF_SlewSamples(SPF_Instance filter, struct timespec *when, double dfreq, double
   int i, first, last;
   double delta_time;
 
-  if (filter->last < 0)
+  if (!get_first_last(filter, &first, &last))
     return;
 
-  /* Always slew the last sample as it may be returned even if no new
-     samples were accumulated */
-  if (filter->used > 0) {
-    first = 0;
-    last = filter->used - 1;
-  } else {
-    first = last = filter->last;
-  }
-
   for (i = first; i <= last; i++) {
     UTI_AdjustTimespec(&filter->samples[i].time, when, &filter->samples[i].time,
                        &delta_time, dfreq, doffset);
@@ -439,6 +471,20 @@ SPF_SlewSamples(SPF_Instance filter, struct timespec *when, double dfreq, double
 
 /* ================================================== */
 
+void
+SPF_CorrectOffset(SPF_Instance filter, double doffset)
+{
+  int i, first, last;
+
+  if (!get_first_last(filter, &first, &last))
+    return;
+
+  for (i = first; i <= last; i++)
+    filter->samples[i].offset -= doffset;
+}
+
+/* ================================================== */
+
 void
 SPF_AddDispersion(SPF_Instance filter, double dispersion)
 {

samplefilt.h

@@ -39,11 +39,13 @@ extern void SPF_DestroyInstance(SPF_Instance filter);
 extern int SPF_AccumulateSample(SPF_Instance filter, NTP_Sample *sample);
 extern int SPF_GetLastSample(SPF_Instance filter, NTP_Sample *sample);
 extern int SPF_GetNumberOfSamples(SPF_Instance filter);
+extern int SPF_GetMaxSamples(SPF_Instance filter);
 extern double SPF_GetAvgSampleDispersion(SPF_Instance filter);
 extern void SPF_DropSamples(SPF_Instance filter);
 extern int SPF_GetFilteredSample(SPF_Instance filter, NTP_Sample *sample);
 extern void SPF_SlewSamples(SPF_Instance filter, struct timespec *when,
                             double dfreq, double doffset);
+extern void SPF_CorrectOffset(SPF_Instance filter, double doffset);
 extern void SPF_AddDispersion(SPF_Instance filter, double dispersion);
 
 #endif

siv_gnutls.c

@@ -165,17 +165,29 @@ SIV_SetKey(SIV_Instance instance, const unsigned char *key, int length)
   datum.data = (unsigned char *)key;
   datum.size = length;
 
-  /* Initialise a new cipher with the provided key (gnutls does not seem to
+#ifdef HAVE_GNUTLS_AEAD_CIPHER_SET_KEY
-     have a function to change the key directly) */
+  if (instance->cipher) {
+    r = gnutls_aead_cipher_set_key(instance->cipher, &datum);
+    if (r < 0) {
+      DEBUG_LOG("Could not set cipher key : %s", gnutls_strerror(r));
+      return 0;
+    }
+
+    return 1;
+  }
+#endif
+
+  /* Initialise a new cipher with the provided key */
   r = gnutls_aead_cipher_init(&cipher, instance->algorithm, &datum);
   if (r < 0) {
     DEBUG_LOG("Could not initialise %s : %s", "cipher", gnutls_strerror(r));
     return 0;
   }
 
-  /* Replace the previous cipher */
+  /* Destroy the previous cipher (if its key could not be changed directly) */
   if (instance->cipher)
     gnutls_aead_cipher_deinit(instance->cipher);
+
   instance->cipher = cipher;
 
   return 1;

socket.c

@@ -505,6 +505,8 @@ bind_unix_address(int sock_fd, const char *addr, int flags)
 {
   union sockaddr_all saddr;
 
+  memset(&saddr, 0, sizeof (saddr));
+
   if (snprintf(saddr.un.sun_path, sizeof (saddr.un.sun_path), "%s", addr) >=
       sizeof (saddr.un.sun_path)) {
     DEBUG_LOG("Unix socket path %s too long", addr);
@@ -537,6 +539,8 @@ connect_unix_address(int sock_fd, const char *addr)
 {
   union sockaddr_all saddr;
 
+  memset(&saddr, 0, sizeof (saddr));
+
   if (snprintf(saddr.un.sun_path, sizeof (saddr.un.sun_path), "%s", addr) >=
       sizeof (saddr.un.sun_path)) {
     DEBUG_LOG("Unix socket path %s too long", addr);

sources.c

@@ -3,7 +3,7 @@
 
  **********************************************************************
  * Copyright (C) Richard P. Curnow  1997-2003
- * Copyright (C) Miroslav Lichvar  2011-2016, 2018, 2020
+ * Copyright (C) Miroslav Lichvar  2011-2016, 2018, 2020-2021
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -68,6 +68,7 @@ struct SelectInfo {
 typedef enum {
   SRC_OK,               /* OK so far, not a final status! */
   SRC_UNSELECTABLE,     /* Has noselect option set */
+  SRC_UNSYNCHRONISED,   /* Provides samples but not unsynchronised */
   SRC_BAD_STATS,        /* Doesn't have valid stats data */
   SRC_BAD_DISTANCE,     /* Has root distance longer than allowed maximum */
   SRC_JITTERY,          /* Had std dev larger than allowed maximum */
@@ -177,6 +178,8 @@ static double reselect_distance;
 static double stratum_weight;
 static double combine_limit;
 
+static LOG_FileID logfileid;
+
 /* Identifier of the dump file */
 #define DUMP_IDENTIFIER "SRC0\n"
 
@@ -188,6 +191,7 @@ static void slew_sources(struct timespec *raw, struct timespec *cooked, double d
                          double doffset, LCL_ChangeType change_type, void *anything);
 static void add_dispersion(double dispersion, void *anything);
 static char *source_to_string(SRC_Instance inst);
+static char get_status_char(SRC_Status status);
 
 /* ================================================== */
 /* Initialisation function */
@@ -207,6 +211,10 @@ void SRC_Initialise(void) {
 
   LCL_AddParameterChangeHandler(slew_sources, NULL);
   LCL_AddDispersionNotifyHandler(add_dispersion, NULL);
+
+  logfileid = CNF_GetLogSelection() ? LOG_FileOpen("selection",
+              "   Date (UTC) Time     IP Address   S EOpts Reach Score Last sample  Low end   High end")
+              : -1;
 }
 
 /* ================================================== */
@@ -639,13 +647,33 @@ source_to_string(SRC_Instance inst)
 static void
 mark_source(SRC_Instance inst, SRC_Status status)
 {
+  struct timespec now;
+
   inst->status = status;
 
-  DEBUG_LOG("%s status=%d options=%x reach=%o/%d updates=%d distant=%d leap=%d vote=%d lo=%f hi=%f",
+  DEBUG_LOG("%s status=%c options=%x reach=%o/%d updates=%d distant=%d leap=%d vote=%d lo=%f hi=%f",
-            source_to_string(inst), (int)inst->status, (unsigned int)inst->sel_options,
+            source_to_string(inst), get_status_char(inst->status),
-            (unsigned int)inst->reachability, inst->reachability_size, inst->updates,
+            (unsigned int)inst->sel_options, (unsigned int)inst->reachability,
+            inst->reachability_size, inst->updates,
             inst->distant, (int)inst->leap, inst->leap_vote,
             inst->sel_info.lo_limit, inst->sel_info.hi_limit);
+
+  if (logfileid == -1)
+    return;
+
+  SCH_GetLastEventTime(&now, NULL, NULL);
+
+  LOG_FileWrite(logfileid,
+                "%s %-15s %c -%c%c%c%c %4o %5.2f %10.3e %10.3e %10.3e",
+                UTI_TimeToLogForm(now.tv_sec), source_to_string(inst),
+                get_status_char(inst->status),
+                inst->sel_options & SRC_SELECT_NOSELECT ? 'N' : '-',
+                inst->sel_options & SRC_SELECT_PREFER ? 'P' : '-',
+                inst->sel_options & SRC_SELECT_TRUST ? 'T' : '-',
+                inst->sel_options & SRC_SELECT_REQUIRE ? 'R' : '-',
+                (unsigned int)inst->reachability, inst->sel_score,
+                inst->sel_info.last_sample_ago,
+                inst->sel_info.lo_limit, inst->sel_info.hi_limit);
 }
 
 /* ================================================== */
@@ -722,8 +750,8 @@ combine_sources(int n_sel_sources, struct timespec *ref_time, double *offset,
     offset_weight = 1.0 / sources[index]->sel_info.root_distance;
     frequency_weight = 1.0 / SQUARE(src_frequency_sd);
 
-    DEBUG_LOG("combining index=%d oweight=%e offset=%e osd=%e fweight=%e freq=%e fsd=%e skew=%e",
+    DEBUG_LOG("combining %s oweight=%e offset=%e osd=%e fweight=%e freq=%e fsd=%e skew=%e",
-              index, offset_weight, src_offset, src_offset_sd,
+              source_to_string(sources[index]), offset_weight, src_offset, src_offset_sd,
               frequency_weight, src_frequency, src_frequency_sd, src_skew);
 
     sum_offset_weight += offset_weight;
@@ -815,6 +843,12 @@ SRC_SelectSource(SRC_Instance updated_inst)
       continue;
     }
 
+    /* Ignore sources which are not synchronised */
+    if (sources[i]->leap == LEAP_Unsynchronised) {
+      mark_source(sources[i], SRC_UNSYNCHRONISED);
+      continue;
+    }
+
     si = &sources[i]->sel_info;
     SST_GetSelectionData(sources[i]->stats, &now,
                          &si->lo_limit, &si->hi_limit, &si->root_distance,
@@ -1642,6 +1676,8 @@ get_status_char(SRC_Status status)
   switch (status) {
     case SRC_UNSELECTABLE:
       return 'N';
+    case SRC_UNSYNCHRONISED:
+      return 's';
     case SRC_BAD_STATS:
       return 'M';
     case SRC_BAD_DISTANCE:

sourcestats.c

@@ -211,8 +211,8 @@ SST_CreateInstance(uint32_t refid, IPAddr *addr, int min_samples, int max_sample
   SST_Stats inst;
   inst = MallocNew(struct SST_Stats_Record);
 
-  inst->min_samples = min_samples;
+  inst->max_samples = max_samples > 0 ? CLAMP(1, max_samples, MAX_SAMPLES) : MAX_SAMPLES;
-  inst->max_samples = max_samples;
+  inst->min_samples = CLAMP(1, min_samples, inst->max_samples);
   inst->fixed_min_delay = min_delay;
   inst->fixed_asymmetry = asymmetry;
 
@@ -698,7 +698,8 @@ SST_GetSelectionData(SST_Stats inst, struct timespec *now,
 
   /* If maxsamples is too small to have a successful regression, enable the
      selection as a special case for a fast update/print-once reference mode */
-  if (!*select_ok && inst->n_samples < 3 && inst->n_samples == inst->max_samples) {
+  if (!*select_ok && inst->n_samples < MIN_SAMPLES_FOR_REGRESS &&
+      inst->n_samples == inst->max_samples) {
     *std_dev = CNF_GetMaxJitter();
     *select_ok = 1;
   }
@@ -814,7 +815,7 @@ SST_PredictOffset(SST_Stats inst, struct timespec *when)
 {
   double elapsed;
   
-  if (inst->n_samples < 3) {
+  if (inst->n_samples < MIN_SAMPLES_FOR_REGRESS) {
     /* We don't have any useful statistics, and presumably the poll
        interval is minimal.  We can't do any useful prediction other
        than use the latest sample or zero if we don't have any samples */
@@ -930,6 +931,7 @@ SST_LoadFromFile(SST_Stats inst, FILE *in)
 
     /* Make sure the samples are sane and they are in order */
     if (!UTI_IsTimeOffsetSane(&inst->sample_times[i], -inst->offsets[i]) ||
+        UTI_CompareTimespecs(&now, &inst->sample_times[i]) < 0 ||
         !(fabs(inst->peer_delays[i]) < 1.0e6 && fabs(inst->peer_dispersions[i]) < 1.0e6 &&
           fabs(inst->root_delays[i]) < 1.0e6 && fabs(inst->root_dispersions[i]) < 1.0e6) ||
         (i > 0 && UTI_CompareTimespecs(&inst->sample_times[i],
@@ -985,6 +987,14 @@ SST_Samples(SST_Stats inst)
 
 /* ================================================== */
 
+int
+SST_GetMinSamples(SST_Stats inst)
+{
+  return inst->min_samples;
+}
+
+/* ================================================== */
+
 void
 SST_DoSourcestatsReport(SST_Stats inst, RPT_SourcestatsReport *report, struct timespec *now)
 {

sourcestats.h

@@ -133,6 +133,8 @@ extern void SST_DoSourcestatsReport(SST_Stats inst, RPT_SourcestatsReport *repor
 
 extern int SST_Samples(SST_Stats inst);
 
+extern int SST_GetMinSamples(SST_Stats inst);
+
 extern double SST_GetJitterAsymmetry(SST_Stats inst);
 
 #endif /* GOT_SOURCESTATS_H */

srcparams.h

@@ -61,6 +61,7 @@ typedef struct {
   double max_delay;
   double max_delay_ratio;
   double max_delay_dev_ratio;
+  double max_delay_quant;
   double min_delay;
   double asymmetry;
   double offset;

sys_generic.c

@@ -73,13 +73,28 @@ static double slew_freq;
 /* Time (raw) of last update of slewing frequency and offset */
 static struct timespec slew_start;
 
-/* Limits for the slew timeout */
+/* Limits for the slew length */
-#define MIN_SLEW_TIMEOUT 1.0
+#define MIN_SLEW_DURATION 1.0
-#define MAX_SLEW_TIMEOUT 1.0e4
+#define MAX_SLEW_DURATION 1.0e4
 
 /* Scheduler timeout ID for ending of the currently running slew */
 static SCH_TimeoutID slew_timeout_id;
 
+/* Scheduled duration of the currently running slew */
+static double slew_duration;
+
+/* Expected delay in ending of the slew due to process scheduling and
+   execution time, tracked as a decaying maximum value */
+static double slew_excess_duration;
+
+/* Maximum accepted excess duration to ignore large jumps after resuming
+   suspended system and other reasons (which should be handled in the
+   scheduler), a constant to determine the minimum slew duration to avoid
+   oscillations due to the excess, and the decay constant */
+#define MAX_SLEW_EXCESS_DURATION 100.0
+#define MIN_SLEW_DURATION_EXCESS_RATIO 5.0
+#define SLEW_EXCESS_DURATION_DECAY 0.9
+
 /* Suggested offset correction rate (correction time * offset) */
 static double correction_rate;
 
@@ -109,12 +124,7 @@ static void
 handle_step(struct timespec *raw, struct timespec *cooked, double dfreq,
             double doffset, LCL_ChangeType change_type, void *anything)
 {
-  if (change_type == LCL_ChangeUnknownStep) {
+  if (change_type == LCL_ChangeStep) {
-    /* Reset offset and slewing */
-    slew_start = *raw;
-    offset_register = 0.0;
-    update_slew();
-  } else if (change_type == LCL_ChangeStep) {
     UTI_AddDoubleToTimespec(&slew_start, -doffset, &slew_start);
   }
 }
@@ -169,8 +179,8 @@ clamp_freq(double freq)
 static void
 update_slew(void)
 {
+  double old_slew_freq, total_freq, corr_freq, duration, excess_duration;
   struct timespec now, end_of_slew;
-  double old_slew_freq, total_freq, corr_freq, duration;
 
   /* Remove currently running timeout */
   SCH_RemoveTimeout(slew_timeout_id);
@@ -183,13 +193,25 @@ update_slew(void)
 
   stop_fastslew(&now);
 
-  /* Estimate how long should the next slew take */
+  /* Update the maximum excess duration, decaying even when the slew did
+     not time out (i.e. frequency was set or offset accrued), but add a small
+     value to avoid denormals */
+  slew_excess_duration = (slew_excess_duration + 1.0e-9) * SLEW_EXCESS_DURATION_DECAY;
+  excess_duration = duration - slew_duration;
+  if (slew_excess_duration < excess_duration &&
+      excess_duration <= MAX_SLEW_EXCESS_DURATION)
+    slew_excess_duration = excess_duration;
+
+  /* Calculate the duration of the new slew, considering the current correction
+     rate and previous delays in stopping of the slew */
   if (fabs(offset_register) < MIN_OFFSET_CORRECTION) {
-    duration = MAX_SLEW_TIMEOUT;
+    duration = MAX_SLEW_DURATION;
   } else {
     duration = correction_rate / fabs(offset_register);
-    if (duration < MIN_SLEW_TIMEOUT)
+    if (duration < MIN_SLEW_DURATION)
-      duration = MIN_SLEW_TIMEOUT;
+      duration = MIN_SLEW_DURATION;
+    if (duration < MIN_SLEW_DURATION_EXCESS_RATIO * slew_excess_duration)
+      duration = MIN_SLEW_DURATION_EXCESS_RATIO * slew_excess_duration;
   }
 
   /* Get frequency offset needed to slew the offset in the duration
@@ -232,23 +254,25 @@ update_slew(void)
      maximum timeout and try again on the next update. */
   if (fabs(offset_register) < MIN_OFFSET_CORRECTION ||
       offset_register * slew_freq <= 0.0) {
-    duration = MAX_SLEW_TIMEOUT;
+    duration = MAX_SLEW_DURATION;
   } else {
     duration = offset_register / slew_freq;
-    if (duration < MIN_SLEW_TIMEOUT)
+    if (duration < MIN_SLEW_DURATION)
-      duration = MIN_SLEW_TIMEOUT;
+      duration = MIN_SLEW_DURATION;
-    else if (duration > MAX_SLEW_TIMEOUT)
+    else if (duration > MAX_SLEW_DURATION)
-      duration = MAX_SLEW_TIMEOUT;
+      duration = MAX_SLEW_DURATION;
   }
 
   /* Restart timer for the next update */
   UTI_AddDoubleToTimespec(&now, duration, &end_of_slew);
   slew_timeout_id = SCH_AddTimeout(&end_of_slew, handle_end_of_slew, NULL);
   slew_start = now;
+  slew_duration = duration;
 
-  DEBUG_LOG("slew offset=%e corr_rate=%e base_freq=%f total_freq=%f slew_freq=%e duration=%f slew_error=%e",
+  DEBUG_LOG("slew offset=%e corr_rate=%e base_freq=%f total_freq=%f slew_freq=%e"
-      offset_register, correction_rate, base_freq, total_freq, slew_freq,
+            " duration=%f excess=%f slew_error=%e",
-      duration, slew_error);
+            offset_register, correction_rate, base_freq, total_freq, slew_freq,
+            slew_duration, slew_excess_duration, slew_error);
 }
 
 /* ================================================== */
@@ -385,6 +409,7 @@ SYS_Generic_CompleteFreqDriver(double max_set_freq_ppm, double max_set_freq_dela
   base_freq = (*drv_read_freq)();
   slew_freq = 0.0;
   offset_register = 0.0;
+  slew_excess_duration = 0.0;
 
   max_corr_freq = CNF_GetMaxSlewRate() / 1.0e6;
 

sys_linux.c

@@ -497,6 +497,9 @@ SYS_Linux_EnableSystemCallFilter(int level, SYS_ProcessContext context)
     SCMP_SYS(getrlimit),
     SCMP_SYS(getuid),
     SCMP_SYS(getuid32),
+#ifdef __NR_rseq
+    SCMP_SYS(rseq),
+#endif
     SCMP_SYS(rt_sigaction),
     SCMP_SYS(rt_sigreturn),
     SCMP_SYS(rt_sigprocmask),
@@ -791,73 +794,25 @@ SYS_Linux_CheckKernelVersion(int req_major, int req_minor)
 
 #if defined(FEAT_PHC) || defined(HAVE_LINUX_TIMESTAMPING)
 
-#define PHC_READINGS 10
-
 static int
-process_phc_readings(struct timespec ts[][3], int n, double precision,
+get_phc_readings(int phc_fd, int max_samples, struct timespec ts[][3])
-                     struct timespec *phc_ts, struct timespec *sys_ts, double *err)
 {
-  double min_delay = 0.0, delays[PTP_MAX_SAMPLES], phc_sum, sys_sum, sys_prec;
-  int i, combined;
-
-  if (n > PTP_MAX_SAMPLES)
-    return 0;
-
-  for (i = 0; i < n; i++) {
-    delays[i] = UTI_DiffTimespecsToDouble(&ts[i][2], &ts[i][0]);
-
-    if (delays[i] < 0.0) {
-      /* Step in the middle of a PHC reading? */
-      DEBUG_LOG("Bad PTP_SYS_OFFSET sample delay=%e", delays[i]);
-      return 0;
-    }
-
-    if (!i || delays[i] < min_delay)
-      min_delay = delays[i];
-  }
-
-  sys_prec = LCL_GetSysPrecisionAsQuantum();
-
-  /* Combine best readings */
-  for (i = combined = 0, phc_sum = sys_sum = 0.0; i < n; i++) {
-    if (delays[i] > min_delay + MAX(sys_prec, precision))
-      continue;
-
-    phc_sum += UTI_DiffTimespecsToDouble(&ts[i][1], &ts[0][1]);
-    sys_sum += UTI_DiffTimespecsToDouble(&ts[i][0], &ts[0][0]) + delays[i] / 2.0;
-    combined++;
-  }
-
-  assert(combined);
-
-  UTI_AddDoubleToTimespec(&ts[0][1], phc_sum / combined, phc_ts);
-  UTI_AddDoubleToTimespec(&ts[0][0], sys_sum / combined, sys_ts);
-  *err = MAX(min_delay / 2.0, precision);
-
-  return 1;
-}
-
-/* ================================================== */
-
-static int
-get_phc_sample(int phc_fd, double precision, struct timespec *phc_ts,
-               struct timespec *sys_ts, double *err)
-{
-  struct timespec ts[PHC_READINGS][3];
   struct ptp_sys_offset sys_off;
   int i;
 
+  max_samples = CLAMP(0, max_samples, PTP_MAX_SAMPLES);
+
   /* Silence valgrind */
   memset(&sys_off, 0, sizeof (sys_off));
 
-  sys_off.n_samples = PHC_READINGS;
+  sys_off.n_samples = max_samples;
 
   if (ioctl(phc_fd, PTP_SYS_OFFSET, &sys_off)) {
     DEBUG_LOG("ioctl(%s) failed : %s", "PTP_SYS_OFFSET", strerror(errno));
     return 0;
   }
 
-  for (i = 0; i < PHC_READINGS; i++) {
+  for (i = 0; i < max_samples; i++) {
     ts[i][0].tv_sec = sys_off.ts[i * 2].sec;
     ts[i][0].tv_nsec = sys_off.ts[i * 2].nsec;
     ts[i][1].tv_sec = sys_off.ts[i * 2 + 1].sec;
@@ -866,31 +821,31 @@ get_phc_sample(int phc_fd, double precision, struct timespec *phc_ts,
     ts[i][2].tv_nsec = sys_off.ts[i * 2 + 2].nsec;
   }
 
-  return process_phc_readings(ts, PHC_READINGS, precision, phc_ts, sys_ts, err);
+  return max_samples;
 }
 
 /* ================================================== */
 
 static int
-get_extended_phc_sample(int phc_fd, double precision, struct timespec *phc_ts,
+get_extended_phc_readings(int phc_fd, int max_samples, struct timespec ts[][3])
-                        struct timespec *sys_ts, double *err)
 {
 #ifdef PTP_SYS_OFFSET_EXTENDED
-  struct timespec ts[PHC_READINGS][3];
   struct ptp_sys_offset_extended sys_off;
   int i;
 
+  max_samples = CLAMP(0, max_samples, PTP_MAX_SAMPLES);
+
   /* Silence valgrind */
   memset(&sys_off, 0, sizeof (sys_off));
 
-  sys_off.n_samples = PHC_READINGS;
+  sys_off.n_samples = max_samples;
 
   if (ioctl(phc_fd, PTP_SYS_OFFSET_EXTENDED, &sys_off)) {
     DEBUG_LOG("ioctl(%s) failed : %s", "PTP_SYS_OFFSET_EXTENDED", strerror(errno));
     return 0;
   }
 
-  for (i = 0; i < PHC_READINGS; i++) {
+  for (i = 0; i < max_samples; i++) {
     ts[i][0].tv_sec = sys_off.ts[i][0].sec;
     ts[i][0].tv_nsec = sys_off.ts[i][0].nsec;
     ts[i][1].tv_sec = sys_off.ts[i][1].sec;
@@ -899,7 +854,7 @@ get_extended_phc_sample(int phc_fd, double precision, struct timespec *phc_ts,
     ts[i][2].tv_nsec = sys_off.ts[i][2].nsec;
   }
 
-  return process_phc_readings(ts, PHC_READINGS, precision, phc_ts, sys_ts, err);
+  return max_samples;
 #else
   return 0;
 #endif
@@ -908,12 +863,14 @@ get_extended_phc_sample(int phc_fd, double precision, struct timespec *phc_ts,
 /* ================================================== */
 
 static int
-get_precise_phc_sample(int phc_fd, double precision, struct timespec *phc_ts,
+get_precise_phc_readings(int phc_fd, int max_samples, struct timespec ts[][3])
-		       struct timespec *sys_ts, double *err)
 {
 #ifdef PTP_SYS_OFFSET_PRECISE
   struct ptp_sys_offset_precise sys_off;
 
+  if (max_samples < 1)
+    return 0;
+
   /* Silence valgrind */
   memset(&sys_off, 0, sizeof (sys_off));
 
@@ -923,11 +880,11 @@ get_precise_phc_sample(int phc_fd, double precision, struct timespec *phc_ts,
     return 0;
   }
 
-  phc_ts->tv_sec = sys_off.device.sec;
+  ts[0][0].tv_sec = sys_off.sys_realtime.sec;
-  phc_ts->tv_nsec = sys_off.device.nsec;
+  ts[0][0].tv_nsec = sys_off.sys_realtime.nsec;
-  sys_ts->tv_sec = sys_off.sys_realtime.sec;
+  ts[0][1].tv_sec = sys_off.device.sec;
-  sys_ts->tv_nsec = sys_off.sys_realtime.nsec;
+  ts[0][1].tv_nsec = sys_off.device.nsec;
-  *err = MAX(LCL_GetSysPrecisionAsQuantum(), precision);
+  ts[0][2] = ts[0][0];
 
   return 1;
 #else
@@ -971,23 +928,23 @@ SYS_Linux_OpenPHC(const char *path, int phc_index)
 /* ================================================== */
 
 int
-SYS_Linux_GetPHCSample(int fd, int nocrossts, double precision, int *reading_mode,
+SYS_Linux_GetPHCReadings(int fd, int nocrossts, int *reading_mode, int max_readings,
-                       struct timespec *phc_ts, struct timespec *sys_ts, double *err)
+                         struct timespec tss[][3])
 {
-  if ((*reading_mode == 2 || !*reading_mode) && !nocrossts &&
+  int r = 0;
-      get_precise_phc_sample(fd, precision, phc_ts, sys_ts, err)) {
+
+  if ((*reading_mode == 2 || *reading_mode == 0) && !nocrossts &&
+      (r = get_precise_phc_readings(fd, max_readings, tss)) > 0) {
     *reading_mode = 2;
-    return 1;
+  } else if ((*reading_mode == 3 || *reading_mode == 0) &&
-  } else if ((*reading_mode == 3 || !*reading_mode) &&
+             (r = get_extended_phc_readings(fd, max_readings, tss)) > 0) {
-      get_extended_phc_sample(fd, precision, phc_ts, sys_ts, err)) {
     *reading_mode = 3;
-    return 1;
+  } else if ((*reading_mode == 1 || *reading_mode == 0) &&
-  } else if ((*reading_mode == 1 || !*reading_mode) &&
+             (r = get_phc_readings(fd, max_readings, tss)) > 0) {
-      get_phc_sample(fd, precision, phc_ts, sys_ts, err)) {
     *reading_mode = 1;
-    return 1;
   }
-  return 0;
+
+  return r;
 }
 
 /* ================================================== */
@@ -1005,7 +962,7 @@ SYS_Linux_SetPHCExtTimestamping(int fd, int pin, int channel,
   pin_desc.func = enable ? PTP_PF_EXTTS : PTP_PF_NONE;
   pin_desc.chan = channel;
 
-  if (ioctl(fd, PTP_PIN_SETFUNC, &pin_desc)) {
+  if (pin >= 0 && ioctl(fd, PTP_PIN_SETFUNC, &pin_desc)) {
     DEBUG_LOG("ioctl(%s) failed : %s", "PTP_PIN_SETFUNC", strerror(errno));
     return 0;
   }

sys_linux.h

@@ -41,8 +41,8 @@ extern int SYS_Linux_CheckKernelVersion(int req_major, int req_minor);
 
 extern int SYS_Linux_OpenPHC(const char *path, int phc_index);
 
-extern int SYS_Linux_GetPHCSample(int fd, int nocrossts, double precision, int *reading_mode,
+extern int SYS_Linux_GetPHCReadings(int fd, int nocrossts, int *reading_mode, int max_readings,
-                                  struct timespec *phc_ts, struct timespec *sys_ts, double *err);
+                                    struct timespec tss[][3]);
 
 extern int SYS_Linux_SetPHCExtTimestamping(int fd, int pin, int channel,
                                            int rising, int falling, int enable);

sys_solaris.c

@@ -21,23 +21,54 @@
 
   =======================================================================
 
-  Driver file for Solaris operating system
+  Driver file for illumos operating system (previously Solaris)
   */
 
 #include "config.h"
 
 #include "sysincl.h"
 
+#include "logging.h"
 #include "privops.h"
 #include "sys_solaris.h"
 #include "sys_timex.h"
 #include "util.h"
 
+#include <kvm.h>
+#include <nlist.h>
+
+/* ================================================== */
+
+static void
+set_dosynctodr(int on_off)
+{
+  struct nlist nl[] = { {"dosynctodr"}, {NULL} };
+  kvm_t *kt;
+
+  kt = kvm_open(NULL, NULL, NULL, O_RDWR, NULL);
+  if (!kt)
+    LOG_FATAL("Could not open kvm");
+
+  if (kvm_nlist(kt, nl) < 0 || !nl[0].n_value)
+    LOG_FATAL("Could not get dosynctodr address");
+
+  if (kvm_kwrite(kt, nl[0].n_value, &on_off, sizeof (on_off)) < 0)
+    LOG_FATAL("Could not write to dosynctodr");
+
+  kvm_close(kt);
+}
+
 /* ================================================== */
 
 void
 SYS_Solaris_Initialise(void)
 {
+  /* The kernel keeps the system clock and hardware clock synchronised to each
+     other.  The dosynctodr variable needs to be set to zero to prevent the
+     the system clock from following the hardware clock when the system clock
+     is not adjusted by adjtime() or ntp_adjtime(modes=MOD_OFFSET). */
+  set_dosynctodr(0);
+
   /* The kernel allows the frequency to be set in the full range off int32_t */
   SYS_Timex_InitialiseWithFunctions(32500, 1.0 / 100, NULL, NULL, NULL,
                                     0.0, 0.0, NULL, NULL);

test/compilation/003-sanitizers

@@ -2,6 +2,8 @@
 # Run the unit and simulation tests with different compiler sanitizers
 # and under valgrind
 
+valgrind_opts="--leak-check=full --errors-for-leak-kinds=definite"
+
 cd ../..
 
 if [ "$(uname -sm)" != "Linux x86_64" ]; then
@@ -75,7 +77,7 @@ do
         pushd test/unit || exit 1
         make "$@" || exit 1
         if [ "$san_options" = "" ]; then
-          make check TEST_WRAPPER="valgrind --error-exitcode=1" || exit 1
+          make check TEST_WRAPPER="valgrind $valgrind_opts --error-exitcode=1" || exit 1
         else
           make check || exit 1
         fi
@@ -87,7 +89,7 @@ do
         pushd test/simulation || exit 1
         export CLKNETSIM_RANDOM_SEED=101
         if [ "$arch_opts" = "" -a "$san_options" = "" ]; then
-          CLKNETSIM_CLIENT_WRAPPER=valgrind ./run -i 1 || exit 1
+          CLKNETSIM_CLIENT_WRAPPER="valgrind $valgrind_opts" ./run -i 1 || exit 1
         elif [ "$CC" = "gcc" ] && ! echo $CFLAGS | grep -q "-static-libasan"; then
           libasan=$(ldd ../../chronyd | grep -o '/.*lib.*/libasan.so.[0-9]')
           CLKNETSIM_PRELOAD=$libasan ./run -i 1 || exit 1

test/simulation/008-ntpera

@@ -29,7 +29,7 @@ echo "$ntp_start" | grep -q '-' && test_skip
 
 for time_offset in -1e-1 1e-1; do
 	for start_offset in 0 "2^32 - $limit"; do
-		export CLKNETSIM_START_DATE=$(awk "BEGIN {print $ntp_start + $start_offset}")
+		export CLKNETSIM_START_DATE=$(awk "BEGIN {printf \"%.0f\", $ntp_start + $start_offset}")
 		run_test || test_fail
 		check_chronyd_exit || test_fail
 		check_source_selection || test_fail
@@ -38,7 +38,7 @@ for time_offset in -1e-1 1e-1; do
 	done
 
 	for start_offset in -$limit "2^32"; do
-		export CLKNETSIM_START_DATE=$(awk "BEGIN {print $ntp_start + $start_offset}")
+		export CLKNETSIM_START_DATE=$(awk "BEGIN {printf \"%.0f\", $ntp_start + $start_offset}")
 		run_test || test_fail
 		check_chronyd_exit || test_fail
 		check_source_selection || test_fail

test/simulation/101-poll

@@ -27,4 +27,30 @@ for poll in $(seq 1 14); do
 	check_sync || test_fail
 done
 
+min_sync_time=$default_min_sync_time
+max_sync_time=$default_max_sync_time
+client_max_min_out_interval=$default_client_max_min_out_interval
+client_min_mean_out_interval=$default_client_min_mean_out_interval
+
+limit=10
+
+for poll in $(seq -7 2 -1); do
+	client_server_options="minpoll $poll maxpoll $poll"
+
+	base_delay=1e-4
+
+	run_test || test_fail
+	check_chronyd_exit || test_fail
+	check_source_selection || test_fail
+	check_file_messages "	2	1	" \
+		$[2**-poll * limit * 9 / 10] $[2**-poll * limit] log.packets || test_fail
+
+	base_delay=2e-2
+
+	run_test || test_fail
+	check_chronyd_exit || test_fail
+	check_source_selection || test_fail
+	check_file_messages "	2	1	" $[limit * 9 / 10] $limit log.packets || test_fail
+done
+
 test_pass

test/simulation/106-refclock

@@ -7,6 +7,9 @@ check_config_h 'FEAT_REFCLOCK 1' || test_skip
 check_config_h 'FEAT_PHC 1' || test_skip
 check_config_h 'FEAT_CMDMON 1' || test_skip
 
+export CLKNETSIM_PHC_DELAY=1e-6
+export CLKNETSIM_PHC_JITTER=1e-7
+
 servers=0
 limit=1000
 refclock_jitter=$jitter
@@ -15,7 +18,7 @@ max_sync_time=70
 chronyc_start=70
 chronyc_conf="tracking"
 
-for refclock in "SHM 0" "PHC /dev/ptp0"; do
+for refclock in "SHM 0" "PHC /dev/ptp0" "PHC /dev/ptp0:nocrossts"; do
 	client_conf="refclock $refclock stratum 3 delay 1e-3 refid GPS
 logdir tmp
 log refclocks"
@@ -32,7 +35,11 @@ Root delay      : 0.001000000 seconds
 Update interval : 16\.. seconds
 .*$" || test_fail
 
-	check_file_messages "20.* GPS.*[0-9] N " 997 1001 refclocks.log || test_fail
+	if echo "$refclock" | grep -q 'PHC.*nocrossts'; then
+		check_file_messages "20.* GPS.*[0-9] N " 650 750 refclocks.log || test_fail
+	else
+		check_file_messages "20.* GPS.*[0-9] N " 997 1001 refclocks.log || test_fail
+	fi
 	check_file_messages "20.* GPS.*- N " 61 63 refclocks.log || test_fail
 	rm -f tmp/refclocks.log
 done
@@ -106,4 +113,31 @@ Root delay      : 0\.000000001 seconds
 	rm -f tmp/refclocks.log
 fi
 
+refclock_offset="(+ 0.399 (sum 1e-3))"
+refclock_jitter=1e-6
+servers=1
+freq_offset="(* 1e-4 (sine 1000))"
+base_delay="(* -1.0 (equal 0.1 (min time 5000) 5000))"
+client_server_options="minpoll 4 maxpoll 4 filter 5 minsamples 64"
+client_conf="
+refclock PHC /dev/ptp0 local poll 2
+logdir tmp
+log refclocks tracking"
+chronyc_conf=""
+limit=10000
+max_sync_time=5000
+time_max_limit=1e-3
+time_rms_limit=5e-4
+freq_max_limit=2e-5
+freq_rms_limit=5e-6
+
+run_test || test_fail
+check_chronyd_exit || test_fail
+check_sync || test_fail
+
+check_file_messages "20.* PHC0 .* [0-9] ? " 9999 10001 refclocks.log || test_fail
+check_file_messages "20.* PHC0 .* - ? " 2499 2501 refclocks.log || test_fail
+check_file_messages "20.* PHC0 " 0 0 tracking.log || test_fail
+rm -f tmp/refclocks.log tmp/tracking.log
+
 test_pass

test/simulation/110-chronyc

@@ -102,7 +102,7 @@ limit=1
 for chronyc_conf in \
 	"accheck 1.2.3.4" \
 	"add peer 10.0.0.0 minpoll 2 maxpoll 6" \
-	"add server 10.0.0.0 minpoll 6 maxpoll 10 iburst burst key 1 certset 2 maxdelay 1e-3 maxdelayratio 10.0 maxdelaydevratio 10.0 mindelay 1e-4 asymmetry 0.5 offset 1e-5 minsamples 6 maxsamples 6 filter 3 offline auto_offline prefer noselect trust require xleave polltarget 20 port 123 presend 7 minstratum 3 version 4 nts ntsport 4460 copy extfield F323" \
+	"add server 10.0.0.0 minpoll 6 maxpoll 10 iburst burst key 1 certset 2 maxdelay 1e-3 maxdelayratio 10.0 maxdelaydevratio 10.0 maxdelayquant 0.5 mindelay 1e-4 asymmetry 0.5 offset 1e-5 minsamples 6 maxsamples 6 filter 3 offline auto_offline prefer noselect trust require xleave polltarget 20 port 123 presend 7 minstratum 3 version 4 nts ntsport 4460 copy extfield F323" \
 	"add server node1.net1.clk" \
 	"allow 1.2.3.4" \
 	"allow 1.2" \
@@ -231,6 +231,7 @@ RX timestamping : Kernel
 Total TX        : 1
 Total RX        : 1
 Total valid RX  : 1
+Total good RX   : 0
 S Name/IP Address        Auth COpts EOpts Last Score     Interval  Leap
 =======================================================================
 M node1\.net1\.clk            N ----- -----    0   1\.0    \+0ns    \+0ns  N

test/simulation/118-maxdelay

@@ -25,4 +25,18 @@ for client_server_options in "maxpoll 6 maxdelay 2e-5"; do
 	check_sync && test_fail
 done
 
+min_sync_time=10
+client_conf="
+logdir tmp
+log rawmeasurements"
+client_server_options="minpoll 2 maxpoll 2 maxdelayquant 0.1"
+
+run_test || test_fail
+check_chronyd_exit || test_fail
+check_packet_interval || test_fail
+check_sync || test_fail
+
+check_file_messages "20.*123\.1.* 111 111 1111" 200 500 measurements.log || test_fail
+check_file_messages "20.*123\.1.* 111 111 1101" 2000 2300 measurements.log || test_fail
+
 test_pass

test/simulation/122-xleave

@@ -15,10 +15,11 @@ check_chronyd_exit || test_fail
 check_source_selection || test_fail
 check_sync || test_fail
 
-check_file_messages "111 111 1111.* 4I [DKH] [DKH]\$" 0 0 measurements.log || test_fail
+check_file_messages "111 111 .111.* 4I [DKH] [DKH]\$" 0 0 measurements.log || test_fail
 rm -f tmp/measurements.log
 
 server_conf=""
+max_sync_time=270
 
 run_test || test_fail
 check_chronyd_exit || test_fail
@@ -28,6 +29,7 @@ check_sync || test_fail
 
 check_file_messages "111 111 1111.* 4B [DKH] [DKH]\$" 2 2 measurements.log || test_fail
 check_file_messages "111 111 1111.* 4I [DKH] [DKH]\$" 30 200 measurements.log || test_fail
+check_file_messages "111 111 0111.* 4I [DKH] [DKH]\$" 1 1 measurements.log || test_fail
 rm -f tmp/measurements.log
 
 clients=2

test/simulation/127-filter

@@ -4,7 +4,7 @@
 
 test_start "filter option"
 
-client_server_options="minpoll 4 maxpoll 4 filter 15"
+client_server_options="minpoll 4 maxpoll 4 filter 15 maxdelay 3.5e-4"
 min_sync_time=710
 max_sync_time=720
 client_max_min_out_interval=16.1
@@ -16,4 +16,28 @@ check_source_selection || test_fail
 check_packet_interval || test_fail
 check_sync || test_fail
 
+base_delay="(+ 1e-4 (* -1 (equal 0.3 (uniform) 0.0)))"
+client_server_options="minpoll 4 maxpoll 4 filter 3"
+min_sync_time=130
+
+run_test || test_fail
+check_chronyd_exit || test_fail
+check_packet_interval || test_fail
+check_sync || test_fail
+
+limit=10
+client_server_options="minpoll -6 maxpoll -6 filter 1"
+
+base_delay=1e-4
+
+run_test || test_fail
+check_chronyd_exit || test_fail
+check_file_messages "	2	1	" 590 640 log.packets || test_fail
+
+base_delay=2e-2
+
+run_test || test_fail
+check_chronyd_exit || test_fail
+check_file_messages "	2	1	" 9 10 log.packets || test_fail
+
 test_pass

test/simulation/133-hwtimestamp

@@ -7,32 +7,53 @@ test_start "hwtimestamp directive"
 check_config_h 'HAVE_LINUX_TIMESTAMPING 1' || test_skip
 
 export CLKNETSIM_TIMESTAMPING=2
+export CLKNETSIM_PHC_DELAY=1e-6
+export CLKNETSIM_PHC_JITTER=1e-7
+export CLKNETSIM_PHC_JITTER_ASYM=0.4
 
 refclock_jitter=1e-8
 refclock_offset=10.0
 min_sync_time=4
 max_sync_time=20
+time_rms_limit=1e-7
+freq_rms_limit=3e-8
+jitter=1e-8
+freq_offset=1e-5
 limit=200
-server_conf="hwtimestamp eth0"
+server_conf="
-client_server_options="minpoll 0 maxpoll 0 minsamples 32 xleave"
+clockprecision 1e-9
+hwtimestamp eth0"
+client_server_options="minpoll 0 maxpoll 0 xleave"
 client_chronyd_options="-d"
 
-for client_conf in "hwtimestamp eth0" "hwtimestamp eth0
+for client_conf in \
-acquisitionport 123"; do
+		"hwtimestamp eth0 nocrossts
+		clockprecision 1e-9" \
+		"hwtimestamp eth0
+		clockprecision 1e-9
+		acquisitionport 123"; do
 	run_test || test_fail
 	check_chronyd_exit || test_fail
 	check_source_selection || test_fail
 	check_sync || test_fail
 
 	if check_config_h 'FEAT_DEBUG 1'; then
+		check_log_messages "Accepted reading" 0 2 || test_fail
+		check_log_messages "Combined .* readings" 190 220 || test_fail
 		check_log_messages "HW clock samples" 190 200 || test_fail
 		check_log_messages "HW clock reset" 0 0 || test_fail
+		check_log_messages "Missing TX timestamp" 1 1 || test_fail
 		check_log_messages "Received message.*tss=KH" 195 200 || test_fail
 		check_log_messages "Received error.*message.*tss=KH" 195 200 || test_fail
 		check_log_messages "Updated RX timestamp.*tss=1" 1 1 || test_fail
 		check_log_messages "Updated RX timestamp.*tss=2" 195 200 || test_fail
-		check_log_messages "update_tx_timestamp.*Updated" 50 140 || test_fail
+		if echo "$client_conf" | grep -q nocrossts; then
-		check_log_messages "update_tx_timestamp.*Unacceptable" 50 140 || test_fail
+			check_log_messages "update_tx_timestamp.*Updated" 180 200 || test_fail
+			check_log_messages "update_tx_timestamp.*Unacceptable" 0 10 || test_fail
+		else
+			check_log_messages "update_tx_timestamp.*Updated" 50 140 || test_fail
+			check_log_messages "update_tx_timestamp.*Unacceptable" 50 140 || test_fail
+		fi
 	fi
 done
 

test/simulation/134-log

@@ -11,7 +11,7 @@ client_server_options="maxpoll 6"
 client_conf="refclock PHC /dev/ptp0 dpoll 4 poll 6 noselect
 logbanner 10
 logdir tmp
-log tracking rawmeasurements measurements statistics rtc refclocks tempcomp
+log tracking rawmeasurements measurements selection statistics rtc refclocks tempcomp
 tempcomp tmp/tempcomp 64 0.0 0.0 0.0 0.0"
 
 echo 0.0 > tmp/tempcomp
@@ -26,6 +26,8 @@ check_file_messages "=============" 31 33 \
 	tracking.log measurements.log tempcomp.log || test_fail
 check_file_messages "20.*192\.168\.123\.1" 150 160 \
 	tracking.log measurements.log statistics.log || test_fail
+check_file_messages "20.*PHC0 * N " 300 320 selection.log || test_fail
+check_file_messages "20.*192\.168\.123\.1 *[M*]" 300 320 selection.log || test_fail
 check_file_messages "20.*PHC0" 150 160 statistics.log || test_fail
 check_file_messages "20.*PHC0" 750 800 refclocks.log || test_fail
 check_file_messages "20.* 0\.0000" 150 160 tempcomp.log || test_fail

test/system/007-cmdmon

@@ -95,12 +95,13 @@ TX timestamping : (Daemon|Kernel)
 RX timestamping : (Daemon|Kernel)
 Total TX        : [0-9]+
 Total RX        : [0-9]+
-Total valid RX  : [0-9]+$" || test_fail
+Total valid RX  : [0-9]+
+Total good RX   : [0-9]+$" || test_fail
 
 run_chronyc "selectdata" || test_fail
 check_chronyc_output "^S Name/IP Address        Auth COpts EOpts Last Score     Interval  Leap
 =======================================================================
-M 127\.0\.0\.1                 N ----- -----    0   1\.0    \+0ns    \+0ns  \?$" || test_fail
+s 127\.0\.0\.1                 N ----- -----    0   1\.0    \+0ns    \+0ns  \?$" || test_fail
 
 run_chronyc "serverstats" || test_fail
 check_chronyc_output "^NTP packets received       : [0-9]+

test/unit/clientlog.c

@@ -1,6 +1,6 @@
 /*
  **********************************************************************
- * Copyright (C) Miroslav Lichvar  2016
+ * Copyright (C) Miroslav Lichvar  2016, 2021
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as

test/unit/hwclock.c

@@ -1,6 +1,6 @@
 /*
  **********************************************************************
- * Copyright (C) Miroslav Lichvar  2016-2018
+ * Copyright (C) Miroslav Lichvar  2016-2018, 2022
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -18,21 +18,29 @@
  **********************************************************************
  */
 
-#include <hwclock.c>
+#include <config.h>
 #include "test.h"
 
+#if defined(FEAT_PHC) || defined(HAVE_LINUX_TIMESTAMPING)
+
+#include <hwclock.c>
+
+#define MAX_READINGS 20
+
 void
 test_unit(void)
 {
   struct timespec start_hw_ts, start_local_ts, hw_ts, local_ts, ts;
+  struct timespec readings[MAX_READINGS][3];
   HCL_Instance clock;
-  double freq, jitter, interval, dj, sum;
+  double freq, jitter, interval, dj, err, sum;
-  int i, j, k, count;
+  int i, j, k, l, new_sample, n_readings, count;
 
   LCL_Initialise();
+  TST_RegisterDummyDrivers();
 
   for (i = 1; i <= 8; i++) {
-    clock = HCL_CreateInstance(random() % (1 << i), 1 << i, 1.0);
+    clock = HCL_CreateInstance(random() % (1 << i), 1 << i, 1.0, 1e-9);
 
     for (j = 0, count = 0, sum = 0.0; j < 100; j++) {
       UTI_ZeroTimespec(&start_hw_ts);
@@ -48,11 +56,14 @@ test_unit(void)
 
       clock->n_samples = 0;
       clock->valid_coefs = 0;
+      QNT_Reset(clock->delay_quants);
+
+      new_sample = 0;
 
       for (k = 0; k < 100; k++) {
         UTI_AddDoubleToTimespec(&start_hw_ts, k * interval * freq, &hw_ts);
         UTI_AddDoubleToTimespec(&start_local_ts, k * interval, &local_ts);
-        if (HCL_CookTime(clock, &hw_ts, &ts, NULL)) {
+        if (HCL_CookTime(clock, &hw_ts, &ts, NULL) && new_sample) {
           dj = fabs(UTI_DiffTimespecsToDouble(&ts, &local_ts) / jitter);
           DEBUG_LOG("delta/jitter %f", dj);
           if (clock->n_samples >= clock->max_samples / 2)
@@ -63,10 +74,25 @@ test_unit(void)
 
         UTI_AddDoubleToTimespec(&start_hw_ts, k * interval * freq + TST_GetRandomDouble(-jitter, jitter), &hw_ts);
 
-        if (HCL_NeedsNewSample(clock, &local_ts))
+        if (HCL_NeedsNewSample(clock, &local_ts)) {
-          HCL_AccumulateSample(clock, &hw_ts, &local_ts, 2.0 * jitter);
+          n_readings = random() % MAX_READINGS + 1;
+          for (l = 0; l < n_readings; l++) {
+            UTI_AddDoubleToTimespec(&local_ts, -TST_GetRandomDouble(0.0, jitter / 10.0), &readings[l][0]);
+            readings[l][1] = hw_ts;
+            UTI_AddDoubleToTimespec(&local_ts, TST_GetRandomDouble(0.0, jitter / 10.0), &readings[l][2]);
+          }
 
-        TEST_CHECK(clock->valid_coefs || clock->n_samples < 2);
+          UTI_ZeroTimespec(&hw_ts);
+          UTI_ZeroTimespec(&local_ts);
+          if (HCL_ProcessReadings(clock, n_readings, readings, &hw_ts, &local_ts, &err)) {
+            HCL_AccumulateSample(clock, &hw_ts, &local_ts, 2.0 * jitter);
+            new_sample = 1;
+          } else {
+            new_sample = 0;
+          }
+        }
+
+        TEST_CHECK(clock->valid_coefs == (clock->n_samples >= 2));
 
         if (!clock->valid_coefs)
           continue;
@@ -82,3 +108,10 @@ test_unit(void)
 
   LCL_Finalise();
 }
+#else
+void
+test_unit(void)
+{
+  TEST_REQUIRE(0);
+}
+#endif

test/unit/ntp_core.c

@@ -385,7 +385,7 @@ add_dummy_auth(NTP_AuthMode auth_mode, uint32_t key_id, NTP_Packet *packet, NTP_
 void
 test_unit(void)
 {
-  char source_line[] = "127.0.0.1 maxdelaydevratio 1e6";
+  char source_line[] = "127.0.0.1 maxdelaydevratio 1e6 noselect";
   char conf[][100] = {
     "allow",
     "port 0",
@@ -411,6 +411,7 @@ test_unit(void)
   NCR_Initialise();
   REF_Initialise();
   KEY_Initialise();
+  CLG_Initialise();
 
   CNF_SetupAccessRestrictions();
 
@@ -422,6 +423,9 @@ test_unit(void)
     source.params.version = random() % 4 + 1;
 
     UTI_ZeroTimespec(&current_time);
+#if HAVE_LONG_TIME_T
+    advance_time(NTP_ERA_SPLIT);
+#endif
     advance_time(TST_GetRandomDouble(1.0, 1e9));
 
     TST_GetRandomAddress(&remote_addr.ip_addr, IPADDR_UNSPEC, -1);
@@ -492,7 +496,10 @@ test_unit(void)
 
       send_request(inst1);
       process_request(&remote_addr);
-      proc_response(inst1, 1, 1, 1, 0);
+      proc_response(inst1,
+                    !source.params.interleaved || source.params.version != 4 ||
+                      inst1->mode == MODE_ACTIVE || j != 2,
+                    1, 1, 0);
       advance_time(1 << inst1->local_poll);
     }
 
@@ -595,6 +602,7 @@ test_unit(void)
   TEST_CHECK(info.auth.mac.length == 72);
   TEST_CHECK(info.auth.mac.key_id == 300);
 
+  CLG_Finalise();
   KEY_Finalise();
   REF_Finalise();
   NCR_Finalise();

test/unit/ntp_sources.c

@@ -110,7 +110,7 @@ change_remote_address(NCR_Instance inst, NTP_Remote_Address *remote_addr, int nt
 void
 test_unit(void)
 {
-  char source_line[] = "127.0.0.1 offline", conf[] = "port 0", name[64], msg[1];
+  char source_line[] = "127.0.0.1 offline", conf[] = "port 0", name[64];
   int i, j, k, slot, found, pool, prev_n;
   uint32_t hash = 0, conf_id;
   NTP_Remote_Address addrs[256], addr;
@@ -120,6 +120,7 @@ test_unit(void)
   RPT_ActivityReport report;
   CPS_NTP_Source source;
   NSR_Status status;
+  NTP_Packet msg;
 
   CNF_Initialise(0, 0);
   CNF_ParseLine(NULL, 1, conf);
@@ -272,12 +273,14 @@ test_unit(void)
 
         switch (random() % 5) {
           case 0:
+            msg.lvm = NTP_LVM(0, NTP_VERSION, random() % 2 ? MODE_CLIENT : MODE_SERVER);
             NSR_ProcessTx(get_record(slot)->remote_addr, &local_addr,
-                          &local_ts, (NTP_Packet *)msg, 0);
+                          &local_ts, &msg, 0);
             break;
           case 1:
+            msg.lvm = NTP_LVM(0, NTP_VERSION, random() % 2 ? MODE_CLIENT : MODE_SERVER);
             NSR_ProcessRx(get_record(slot)->remote_addr, &local_addr,
-                          &local_ts, (NTP_Packet *)msg, 0);
+                          &local_ts, &msg, 0);
             break;
           case 2:
             NSR_HandleBadSource(&get_record(slot)->remote_addr->ip_addr);

test/unit/quantiles.c

@@ -0,0 +1,68 @@
+/*
+ **********************************************************************
+ * Copyright (C) Miroslav Lichvar  2022
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+ *
+ **********************************************************************
+ */
+
+#include <local.h>
+#include "test.h"
+
+#include <quantiles.c>
+
+void
+test_unit(void)
+{
+  int i, j, k, min_k, max_k, q, r, in_order, out_order;
+  QNT_Instance inst;
+  double x;
+
+  in_order = out_order = 0;
+
+  for (i = 0; i < 100; i++) {
+    r = random() % 10 + 1;
+    q = random() % 20 + 2;
+    do {
+      min_k = random() % (q - 1) + 1;
+      max_k = random() % (q - 1) + 1;
+    } while (min_k > max_k);
+
+    inst = QNT_CreateInstance(min_k, max_k, q, r, 1e-9);
+
+    TEST_CHECK(min_k == QNT_GetMinK(inst));
+
+    for (j = 0; j < 3000; j++) {
+      x = TST_GetRandomDouble(0.0, 2e-6);
+      QNT_Accumulate(inst, x);
+      for (k = min_k; k < max_k; k++)
+        if (j < max_k - min_k) {
+          TEST_CHECK(QNT_GetQuantile(inst, k) <= QNT_GetQuantile(inst, k + 1));
+        } else if (j > 1000) {
+          if (QNT_GetQuantile(inst, k) <= QNT_GetQuantile(inst, k + 1))
+            in_order++;
+          else
+            out_order++;
+        }
+    }
+
+    QNT_Reset(inst);
+    TEST_CHECK(inst->n_set == 0);
+
+    QNT_DestroyInstance(inst);
+  }
+
+  TEST_CHECK(in_order > 100 * out_order);
+}

test/unit/samplefilt.c

@@ -45,6 +45,8 @@ test_unit(void)
 
     filter = SPF_CreateInstance(min_samples, max_samples, 2.0, combine_ratio);
 
+    TEST_CHECK(max_samples == SPF_GetMaxSamples(filter));
+
     for (j = 0, sum_count = 0, sum_err = 0.0; j < 100; j++) {
       DEBUG_LOG("iteration %d/%d", i, j);
 
@@ -69,6 +71,7 @@ test_unit(void)
         TEST_CHECK(!memcmp(&sample_in, &sample_out, sizeof (sample_in)));
 
         SPF_SlewSamples(filter, &sample_in.time, 0.0, 0.0);
+        SPF_CorrectOffset(filter, 0.0);
         SPF_AddDispersion(filter, 0.0);
 
         if (k + 1 < min_samples)
@@ -102,6 +105,7 @@ test_unit(void)
 
       } else {
         SPF_DropSamples(filter);
+        TEST_CHECK(filter->last < 0);
       }
 
       TEST_CHECK(SPF_GetNumberOfSamples(filter) == 0);

test/unit/sources.c

@@ -1,6 +1,6 @@
 /*
  **********************************************************************
- * Copyright (C) Miroslav Lichvar  2016, 2018
+ * Copyright (C) Miroslav Lichvar  2016, 2018, 2022
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as
@@ -95,12 +95,14 @@ test_unit(void)
         double passed_lo = DBL_MAX, passed_hi = DBL_MIN;
 
         SRC_SelectSource(srcs[k]);
-        DEBUG_LOG("source %d status %u", k, sources[k]->status);
+        DEBUG_LOG("source %d status %c", k, get_status_char(sources[k]->status));
 
         for (l = 0; l <= j; l++) {
           TEST_CHECK(sources[l]->status > SRC_OK && sources[l]->status <= SRC_SELECTED);
           if (sources[l]->sel_options & SRC_SELECT_NOSELECT) {
             TEST_CHECK(sources[l]->status == SRC_UNSELECTABLE);
+          } else if (sources[l]->leap == LEAP_Unsynchronised) {
+            TEST_CHECK(sources[l]->status == SRC_UNSYNCHRONISED);
           } else if (sources[l]->status != SRC_BAD_DISTANCE) {
             if (sources[l]->status >= SRC_NONPREFERRED) {
               passed++;
@@ -149,6 +151,51 @@ test_unit(void)
     }
   }
 
+  for (i = 0; i < 16; i++) {
+    DEBUG_LOG("iteration %d", i);
+
+    for (j = 0; j < sizeof (srcs) / sizeof (srcs[0]); j++) {
+      TEST_CHECK(n_sources == j);
+
+      srcs[j] = create_source(SRC_NTP, &addrs[j], 0, 0);
+      SRC_UpdateReachability(srcs[j], 1);
+
+      samples = 8;
+      for (k = 0; k < samples; k++) {
+        SCH_GetLastEventTime(&sample.time, NULL, NULL);
+        UTI_AddDoubleToTimespec(&sample.time, k - samples, &sample.time);
+        if (j != 0)
+          UTI_AddDoubleToTimespec(&sample.time, -i * (1.0e-3 / LCL_GetMaxClockError()), &sample.time);
+
+        sample.offset = (k % 2) * 1e-8;
+        sample.peer_delay = sample.root_delay = 2.0e-3 * (j + 1);
+        sample.peer_dispersion = sample.root_dispersion = 4.0e-3;
+
+        SRC_AccumulateSample(srcs[j], &sample);
+        SRC_UpdateStatus(srcs[j], 1, LEAP_Normal);
+      }
+    }
+
+    SRC_SelectSource(srcs[0]);
+
+    for (j = 0; j < sizeof (srcs) / sizeof (srcs[0]); j++) {
+      DEBUG_LOG("%d %c %f", j, get_status_char(srcs[j]->status),
+                srcs[j]->sel_info.root_distance);
+      if (j == 0)
+        TEST_CHECK(sources[j]->status == SRC_SELECTED);
+      else if (j < 11 - i)
+        TEST_CHECK(sources[j]->status == SRC_UNSELECTED);
+      else
+        TEST_CHECK(sources[j]->status == SRC_DISTANT);
+    }
+
+    for (j = 0; j < sizeof (srcs) / sizeof (srcs[0]); j++) {
+      SCH_GetLastEventTime(&sample.time, NULL, NULL);
+      SRC_ReportSource(j, &report, &sample.time);
+      SRC_DestroyInstance(srcs[j]);
+    }
+  }
+
   TEST_CHECK(CNF_GetAuthSelectMode() == SRC_AUTHSELECT_MIX);
 
   for (i = 0; i < 1000; i++) {

util.c

@@ -3,7 +3,7 @@
 
  **********************************************************************
  * Copyright (C) Richard P. Curnow  1997-2003
- * Copyright (C) Miroslav Lichvar  2009, 2012-2020
+ * Copyright (C) Miroslav Lichvar  2009, 2012-2021
  * 
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of version 2 of the GNU General Public License as