pre-release

Originally when vgdisplay encountered an exported VG it issued a
WARNING.  Commit d6b1de30 replaced this with an error message
but still exited with success (incorrect).  A backtrace was recently
added in commit b193809987.

As vgdisplay already states that the VG is exported in its output,
just drop these messages completely.

VERSION

@@ -1 +1 @@
-2.02.128(2)-git (2015-08-10)
+2.02.129(2)-git (2015-08-26)

VERSION_DM

@@ -1 +1 @@
-1.02.105-git (2015-08-10)
+1.02.106-git (2015-08-26)

WHATS_NEW

@@ -1,10 +1,22 @@
-Version 2.02.128 - 
+Version 2.02.129 - 26th August 2015
 ===================================
+  Drop error message when vgdisplay encounters an exported VG. (2.02.27)
+  Fix shared library generation to stop exporting internal functions.(2.02.120)
+  Accept --cachemode with lvconvert.
+  Fix and improve reporting properties of cache-pool.
+  Enable usage of --cachepolicy and --cachesetting with lvconvert.
+  Don't allow to reduce size of thin-pool metadata.
+  Fix debug buffer overflows in cmirrord logging.
+  Add --foreground and --help to cmirrord.
+
+Version 2.02.128 - 17th August 2015
+===================================
+  Allocation setting cache_pool_cachemode is replaced by cache_mode.
+  Don't attempt to close config file that couldn't be opened.
   Check for valid cache mode in validation of cache segment.
-  Enhance internal API cache_set_mode() and cache_set_policy().
-  Enhance toollib's get_cache_params().
-  Runtime detect presence of cache smq policy.
-  Add demo cache-mq  and cache-smq profiles.
+  Change internal interface handling cache mode and policy.
+  When no cache policy specified, prefer smq (if available) over mq.
+  Add demo cache-mq and cache-smq profiles.
   Add cmd profilable allocation/cache_policy,cache_settings,cache_mode.
   Require cache_check 0.5.4 for use of --clear-needs-check-flag.
   Fix lvmetad udev rules to not override SYSTEMD_WANTS, add the service instead.

WHATS_NEW_DM

@@ -1,6 +1,24 @@
-Version 1.02.105 - 
+Version 1.02.106 - 26th August 2015
 ===================================
+  Add 'precise' column to statistics reports.
+  Add --precise switch to 'dmstats create' to request nanosecond counters.
+  Add precise argument to dm_stats_create_region().
+  Add support to libdm-stats for precise_timestamps
+
+Version 1.02.105 - 17th August 2015
+===================================
+  Fix 'dmstats list -o all' segfault.
+  Separate dmstats statistics fields from region information fields.
+  Add interval and interval_ns fields to dmstats reports.
+  Do not include internal glibc headers in libdm-timestamp.c (1.02.104)
+  Exit immediately if no device is supplied to dmsetup wipe_table.
+  Suppress dmsetup report headings when no data is output. (1.02.104)
+  Adjust dmsetup usage/help output selection to match command invoked.
+  Fix dmsetup -o all to select correct fields in splitname report.
+  Restructure internal dmsetup argument handling across all commands.
+  Add dm_report_is_empty() to indicate there is no data awaiting output.
   Add more arg validation for dm_tree_node_add_cache_target().
+  Add --alldevices switch to replace use of --force for stats create / delete.
 
 Version 1.02.104 - 10th August 2015
 ===================================

conf/lvmlocal.conf.in

@@ -24,34 +24,33 @@ local {
 
 	# Configuration option local/system_id.
 	# Defines the local system ID for lvmlocal mode.
-	# This is used when global/system_id_source is set
-	# to 'lvmlocal' in the main configuration file,
-	# e.g. lvm.conf.
-	# When used, it must be set to a unique value
-	# among all hosts sharing access to the storage,
+	# This is used when global/system_id_source is set to 'lvmlocal' in the
+	# main configuration file, e.g. lvm.conf. When used, it must be set to
+	# a unique value among all hosts sharing access to the storage,
 	# e.g. a host name.
-	# Example:
-	# Set no system ID.
+	# 
+	# Example
+	# Set no system ID:
 	# system_id = ""
-	# Example:
-	# Set the system_id to the string 'host1'.
+	# Set the system_id to a specific name:
 	# system_id = "host1"
+	# 
+	# This configuration option has an automatic default value.
 	# system_id = ""
 
 	# Configuration option local/extra_system_ids.
 	# A list of extra VG system IDs the local host can access.
-	# VGs with the system IDs listed here (in addition
-	# to the host's own system ID) can be fully accessed
-	# by the local host.  (These are system IDs that the
-	# host sees in VGs, not system IDs that identify the
-	# local host, which is determined by system_id_source.)
-	# Use this only after consulting 'man lvmsystemid'
-	# to be certain of correct usage and possible dangers.
+	# VGs with the system IDs listed here (in addition to the host's own
+	# system ID) can be fully accessed by the local host. (These are
+	# system IDs that the host sees in VGs, not system IDs that identify
+	# the local host, which is determined by system_id_source.)
+	# Use this only after consulting 'man lvmsystemid' to be certain of
+	# correct usage and possible dangers.
 	# This configuration option does not have a default value defined.
 
 	# Configuration option local/host_id.
 	# The lvmlockd sanlock host_id.
-	# This must be a unique among all hosts,
-	# and must be between 1 and 2000.
+	# This must be unique among all hosts, and must be between 1 and 2000.
+	# This configuration option has an automatic default value.
 	# host_id = 0
 }

configure

@@ -665,6 +665,7 @@ PYTHON_LIBDIRS
 PYTHON_INCDIRS
 PYTHON_BINDINGS
 PTHREAD_LIBS
+M_LIBS
 POOL
 PKGCONFIG
 OCFDIR
@@ -12569,6 +12570,50 @@ if [ \( "$LVM1" = shared -o "$POOL" = shared -o "$CLUSTER" = shared \
 	as_fn_error $? "Features cannot be 'shared' when building statically" "$LINENO" 5
 fi
 
+################################################################################
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for log10 in -lm" >&5
+$as_echo_n "checking for log10 in -lm... " >&6; }
+if ${ac_cv_lib_m_log10+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lm  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char log10 ();
+int
+main ()
+{
+return log10 ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_lib_m_log10=yes
+else
+  ac_cv_lib_m_log10=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_m_log10" >&5
+$as_echo "$ac_cv_lib_m_log10" >&6; }
+if test "x$ac_cv_lib_m_log10" = xyes; then :
+  M_LIBS="-lm"
+else
+  hard_bailout
+fi
+
+
 ################################################################################
 { $as_echo "$as_me:${as_lineno-$LINENO}: checking for pthread_mutex_lock in -lpthread" >&5
 $as_echo_n "checking for pthread_mutex_lock in -lpthread... " >&6; }
@@ -14151,6 +14196,7 @@ LVM_LIBAPI=`echo "$VER" | $AWK -F '[()]' '{print $2}'`
 
 
 
+
 
 
 ################################################################################

configure.in

@@ -1537,6 +1537,10 @@ if [[ \( "$LVM1" = shared -o "$POOL" = shared -o "$CLUSTER" = shared \
 	AC_MSG_ERROR([Features cannot be 'shared' when building statically])
 fi
 
+################################################################################
+AC_CHECK_LIB(m, log10,
+	[M_LIBS="-lm"], hard_bailout)
+
 ################################################################################
 AC_CHECK_LIB([pthread], [pthread_mutex_lock],
 	[PTHREAD_LIBS="-lpthread"], hard_bailout)
@@ -1984,6 +1988,7 @@ AC_SUBST(OCF)
 AC_SUBST(OCFDIR)
 AC_SUBST(PKGCONFIG)
 AC_SUBST(POOL)
+AC_SUBST(M_LIBS)
 AC_SUBST(PTHREAD_LIBS)
 AC_SUBST(PYTHON)
 AC_SUBST(PYTHON_BINDINGS)

daemons/clvmd/clvmd.c

@@ -855,12 +855,12 @@ static void main_loop(int cmd_timeout)
 		int quorate = clops->is_quorate();
 		int client_count = 0;
 		int max_fd = 0;
+		struct local_client *lastfd = &local_client_head;
+		struct local_client *nextfd = local_client_head.next;
 
 		/* Wait on the cluster FD and all local sockets/pipes */
 		local_client_head.fd = clops->get_main_cluster_fd();
 		FD_ZERO(&in);
-		struct local_client *lastfd = &local_client_head;
-		struct local_client *nextfd = local_client_head.next;
 
 		for (thisfd = &local_client_head; thisfd; thisfd = thisfd->next) {
 			client_count++;

daemons/cmirrord/clogd.c

@@ -15,6 +15,7 @@
 #include "link_mon.h"
 #include "local.h"
 
+#include <getopt.h>
 #include <errno.h>
 #include <fcntl.h>
 #include <sys/socket.h>
@@ -32,14 +33,49 @@ static void daemonize(void);
 static void init_all(void);
 static void cleanup_all(void);
 
-int main(int argc __attribute__((unused)), char *argv[] __attribute__((unused)))
+static void usage (FILE *dest)
 {
-	daemonize();
+	fprintf (dest, "Usage: cmirrord [options]\n"
+		 "   -f, --foreground    stay in the foreground, log to the terminal\n"
+		 "   -h, --help          print this help\n");
+}
+
+int main(int argc, char *argv[])
+{
+	int foreground_mode = 0;
+	struct option longopts[] = {
+		{ "foreground", no_argument, NULL, 'f' },
+		{ "help"      , no_argument, NULL, 'h' },
+		{ 0, 0, 0, 0 }
+	};
+	int opt;
+
+	while ((opt = getopt_long (argc, argv, "fh", longopts, NULL)) != -1) {
+		switch (opt) {
+		case 'f':
+			foreground_mode = 1;
+			break;
+		case 'h':
+			usage (stdout);
+			exit (0);
+		default:
+			usage (stderr);
+			exit (2);
+		}
+	}
+	if (optind < argc) {
+		usage (stderr);
+		exit (2);
+	}
+
+	if (!foreground_mode)
+		daemonize();
 
 	init_all();
 
 	/* Parent can now exit, we're ready to handle requests */
-	kill(getppid(), SIGTERM);
+	if (!foreground_mode)
+		kill(getppid(), SIGTERM);
 
 	LOG_PRINT("Starting cmirrord:");
 	LOG_PRINT(" Built: "__DATE__" "__TIME__"\n");
@@ -209,6 +245,16 @@ static void daemonize(void)
 	}
 
 	LOG_OPEN("cmirrord", LOG_PID, LOG_DAEMON);
+}
+
+/*
+ * init_all
+ *
+ * Initialize modules.  Exit on failure.
+ */
+static void init_all(void)
+{
+	int r;
 
 	(void) dm_prepare_selinux_context(CMIRRORD_PIDFILE, S_IFREG);
 	if (dm_create_lockfile(CMIRRORD_PIDFILE) == 0)
@@ -227,16 +273,6 @@ static void daemonize(void)
 	signal(SIGUSR2, &sig_handler);
 	sigemptyset(&signal_mask);
 	signal_received = 0;
-}
-
-/*
- * init_all
- *
- * Initialize modules.  Exit on failure.
- */
-static void init_all(void)
-{
-	int r;
 
 	if ((r = init_local()) ||
 	    (r = init_cluster())) {

daemons/cmirrord/cluster.c

@@ -104,10 +104,11 @@ static SaVersionT version = { 'B', 1, 1 };
 #endif
 
 #define DEBUGGING_HISTORY 100
+#define DEBUGGING_BUFLEN 128
 #define LOG_SPRINT(cc, f, arg...) do {				\
 		cc->idx++;					\
 		cc->idx = cc->idx % DEBUGGING_HISTORY;		\
-		sprintf(cc->debugging[cc->idx], f, ## arg);	\
+		snprintf(cc->debugging[cc->idx], DEBUGGING_BUFLEN, f, ## arg); \
 	} while (0)
 
 static int log_resp_rec = 0;
@@ -150,7 +151,7 @@ struct clog_cpg {
 	uint32_t checkpoint_requesters[MAX_CHECKPOINT_REQUESTERS];
 	struct checkpoint_data *checkpoint_list;
 	int idx;
-	char debugging[DEBUGGING_HISTORY][128];
+	char debugging[DEBUGGING_HISTORY][DEBUGGING_BUFLEN];
 };
 
 static struct dm_list clog_cpg_list;
@@ -1294,7 +1295,9 @@ static void cpg_join_callback(struct clog_cpg *match,
 	uint32_t my_pid = (uint32_t)getpid();
 	uint32_t lowest = match->lowest_id;
 	struct clog_request *rq;
-	char dbuf[32] = { 0 };
+	char dbuf[64] = { 0 };
+	char *dbuf_p = dbuf;
+	size_t dbuf_rem = sizeof dbuf;
 
 	/* Assign my_cluster_id */
 	if ((my_cluster_id == 0xDEAD) && (joined->pid == my_pid))
@@ -1310,9 +1313,17 @@ static void cpg_join_callback(struct clog_cpg *match,
 	if (joined->nodeid == my_cluster_id)
 		goto out;
 
-	for (i = 0; i < member_list_entries - 1; i++)
-		sprintf(dbuf+strlen(dbuf), "%u-", member_list[i].nodeid);
-	sprintf(dbuf+strlen(dbuf), "(%u)", joined->nodeid);
+	for (i = 0; i < member_list_entries - 1; i++) {
+		int written = snprintf(dbuf_p, dbuf_rem, "%u-", member_list[i].nodeid);
+		if (written < 0) continue; /* impossible */
+		if ((unsigned)written >= dbuf_rem) {
+			dbuf_rem = 0;
+			break;
+		}
+		dbuf_rem -= written;
+		dbuf_p += written;
+	}
+	snprintf(dbuf_p, dbuf_rem, "(%u)", joined->nodeid);
 	LOG_COND(log_checkpoint, "[%s] Joining node, %u needs checkpoint [%s]",
 		 SHORT_UUID(match->name.value), joined->nodeid, dbuf);
 

daemons/cmirrord/functions.c

@@ -32,12 +32,13 @@
 #define LOG_OFFSET 2
 
 #define RESYNC_HISTORY 50
+#define RESYNC_BUFLEN 128
 //static char resync_history[RESYNC_HISTORY][128];
 //static int idx = 0;
 #define LOG_SPRINT(_lc, f, arg...) do {					\
 		lc->idx++;						\
 		lc->idx = lc->idx % RESYNC_HISTORY;			\
-		sprintf(lc->resync_history[lc->idx], f, ## arg);	\
+		snprintf(lc->resync_history[lc->idx], RESYNC_BUFLEN, f, ## arg); \
 	} while (0)
 
 struct log_header {
@@ -88,7 +89,7 @@ struct log_c {
 	size_t disk_size;       /* size of disk_buffer in bytes */
 	void *disk_buffer;      /* aligned memory for O_DIRECT */
 	int idx;
-	char resync_history[RESYNC_HISTORY][128];
+	char resync_history[RESYNC_HISTORY][RESYNC_BUFLEN];
 };
 
 struct mark_entry {

daemons/lvmlockd/lvmlockctl.c

@@ -151,13 +151,12 @@ static void format_info_r(char *line, char *r_name_out, char *r_type_out)
 	sscanf(line, "info=r name=%s type=%s mode=%s %s version=%u",
 	       r_name, r_type, mode, sh_count, &ver);
 
-	/* when mode is not un, wait and print each lk line */
+	strcpy(r_name_out, r_name);
+	strcpy(r_type_out, r_type);
 
-	if (strcmp(mode, "un")) {
-		strcpy(r_name_out, r_name);
-		strcpy(r_type_out, r_type);
+	/* when mode is not un, wait and print each lk line */
+	if (strcmp(mode, "un"))
 		return;
-	}
 
 	/* when mode is un, there will be no lk lines, so print now */
 
@@ -229,7 +228,7 @@ static void format_info_r_action(char *line, char *r_name, char *r_type)
 	find_client_info(client_id, &pid, cl_name);
 
 	if (strcmp(op, "lock")) {
-		printf("OP %s pid %u (%s)", op, pid, cl_name);
+		printf("OP %s pid %u (%s)\n", op, pid, cl_name);
 		return;
 	}
 
@@ -380,6 +379,7 @@ static int setup_dump_socket(void)
 
 	rv = bind(s, (struct sockaddr *) &dump_addr, dump_addrlen);
 	if (rv < 0) {
+		rv = -errno;
 		if (!close(s))
 			log_error("failed to close dump socket");
 		return rv;
@@ -393,6 +393,7 @@ static int do_dump(const char *req_name)
 	daemon_reply reply;
 	int result;
 	int fd, rv = 0;
+	int count = 0;
 
 	fd = setup_dump_socket();
 	if (fd < 0) {
@@ -423,13 +424,18 @@ static int do_dump(const char *req_name)
 
 	memset(dump_buf, 0, sizeof(dump_buf));
 
-	rv = recvfrom(fd, dump_buf, dump_len, MSG_WAITALL,
+retry:
+	rv = recvfrom(fd, dump_buf + count, dump_len - count, MSG_WAITALL,
 		      (struct sockaddr *)&dump_addr, &dump_addrlen);
 	if (rv < 0) {
 		log_error("recvfrom error %d %d", rv, errno);
 		rv = -errno;
 		goto out;
 	}
+	count += rv;
+
+	if (count < dump_len)
+		goto retry;
 
 	rv = 0;
 	if ((info && dump) || !strcmp(req_name, "dump"))

daemons/lvmlockd/lvmlockd-core.c

@@ -265,6 +265,10 @@ static int alloc_new_structs; /* used for initializing in setup_structs */
 static int add_lock_action(struct action *act);
 static int str_to_lm(const char *str);
 static int clear_lockspace_inactive(char *name);
+static int setup_dump_socket(void);
+static void send_dump_buf(int fd, int dump_len);
+static int dump_info(int *dump_len);
+static int dump_log(int *dump_len);
 
 static int _syslog_name_to_num(const char *name)
 {
@@ -739,6 +743,10 @@ static const char *op_str(int x)
 		return "kill_vg";
 	case LD_OP_DROP_VG:
 		return "drop_vg";
+	case LD_OP_DUMP_LOG:
+		return "dump_log";
+	case LD_OP_DUMP_INFO:
+		return "dump_info";
 	default:
 		return "op_unknown";
 	};
@@ -1283,7 +1291,7 @@ static int res_unlock(struct lockspace *ls, struct resource *r,
 	}
 
 	if (act->op != LD_OP_CLOSE)
-		log_error("S %s R %s res_unlock no locks", ls->name, r->name);
+		log_debug("S %s R %s res_unlock no locks", ls->name, r->name);
 	return -ENOENT;
 
 do_unlock:
@@ -3414,6 +3422,8 @@ static void client_send_result(struct client *cl, struct action *act)
 {
 	response res;
 	char result_flags[128];
+	int dump_len = 0;
+	int dump_fd = -1;
 
 	if (cl->dead) {
 		log_debug("client send %d skip dead", cl->id);
@@ -3498,6 +3508,33 @@ static void client_send_result(struct client *cl, struct action *act)
 					  "lv_lock_args = %s", lv_args,
 					  "result_flags = %s", result_flags[0] ? result_flags : "none",
 					  NULL);
+
+	} else if (act->op == LD_OP_DUMP_LOG || act->op == LD_OP_DUMP_INFO) {
+		/*
+		 * lvmlockctl creates the unix socket then asks us to write to it.
+		 * FIXME: move processing this to a new dedicated query thread to
+		 * avoid having a large data dump interfere with normal operation
+		 * of the client thread?
+		 */
+
+		dump_fd = setup_dump_socket();
+		if (dump_fd < 0)
+			act->result = dump_fd;
+		else if (act->op == LD_OP_DUMP_LOG)
+			act->result = dump_log(&dump_len);
+		else if (act->op == LD_OP_DUMP_INFO)
+			act->result = dump_info(&dump_len);
+		else
+			act->result = -EINVAL;
+
+		log_debug("send %s[%d.%u] dump result %d dump_len %d",
+			  cl->name[0] ? cl->name : "client", cl->pid, cl->id,
+			  act->result, dump_len);
+
+		res = daemon_reply_simple("OK",
+					  "result = %d", act->result,
+					  "dump_len = %d", dump_len,
+					  NULL);
 	} else {
 		/*
 		 * A normal reply.
@@ -3521,6 +3558,12 @@ static void client_send_result(struct client *cl, struct action *act)
 	buffer_destroy(&res.buffer);
 
 	client_resume(cl);
+
+	if (dump_fd >= 0) {
+		/* To avoid deadlock, send data here after the reply. */
+		send_dump_buf(dump_fd, dump_len);
+		close(dump_fd);
+	}
 }
 
 /* called from client_thread */
@@ -3529,6 +3572,13 @@ static void client_purge(struct client *cl)
 	struct lockspace *ls;
 	struct action *act;
 
+	/*
+	 * If the client made no lock requests, there can be
+	 * no locks to release for it.
+	 */
+	if (!cl->lock_ops)
+		return;
+
 	pthread_mutex_lock(&lockspaces_mutex);
 	list_for_each_entry(ls, &lockspaces, list) {
 		if (!(act = alloc_action()))
@@ -3887,23 +3937,43 @@ static int setup_dump_socket(void)
 	return s;
 }
 
-static int send_dump_buf(int fd, int dump_len)
+#define MAX_SEND_LEN 65536
+#define RESEND_DELAY_US 1000
+#define RESEND_DELAY_US_MAX 500000
+
+static void send_dump_buf(int fd, int dump_len)
 {
 	int pos = 0;
 	int ret;
+	int send_len;
+	int delay = 0;
 
-retry:
-	ret = sendto(fd, dump_buf + pos, dump_len - pos, MSG_DONTWAIT | MSG_NOSIGNAL,
+	if (!dump_len)
+		return;
+repeat:
+	if (dump_len - pos < MAX_SEND_LEN)
+		send_len = dump_len - pos;
+	else
+		send_len = MAX_SEND_LEN;
+
+	ret = sendto(fd, dump_buf + pos, send_len, MSG_NOSIGNAL | MSG_DONTWAIT,
 		     (struct sockaddr *)&dump_addr, dump_addrlen);
-	if (ret <= 0)
-		return ret;
+	if (ret < 0) {
+		if ((errno == EAGAIN || errno == EINTR) && (delay < RESEND_DELAY_US_MAX)) {
+			usleep(RESEND_DELAY_US);
+			delay += RESEND_DELAY_US;
+			goto repeat;
+		}
+		log_error("send_dump_buf delay %d errno %d", delay, errno);
+		return;
+	}
 
 	pos += ret;
 
 	if (pos < dump_len)
-		goto retry;
+		goto repeat;
 
-	return 0;
+	log_debug("send_dump_buf delay %d total %d", delay, pos);
 }
 
 static int print_structs(const char *prefix, int pos, int len)
@@ -4186,8 +4256,7 @@ static void client_recv_action(struct client *cl)
 		return;
 	}
 
-	if (op == LD_OP_HELLO || op == LD_OP_QUIT ||
-	    op == LD_OP_DUMP_INFO || op == LD_OP_DUMP_LOG) {
+	if (op == LD_OP_HELLO || op == LD_OP_QUIT) {
 
 		/*
 		 * FIXME: add the client command name to the hello messages
@@ -4208,37 +4277,11 @@ static void client_recv_action(struct client *cl)
 
 		buffer_init(&res.buffer);
 
-		if (op == LD_OP_DUMP_INFO || op == LD_OP_DUMP_LOG) {
-			int dump_len = 0;
-			int fd;
-
-			fd = setup_dump_socket();
-			if (fd < 0)
-				result = fd;
-			else if (op == LD_OP_DUMP_INFO)
-				result = dump_info(&dump_len);
-			else if (op == LD_OP_DUMP_LOG)
-				result = dump_log(&dump_len);
-			else
-				result = -EINVAL;
-
-			res = daemon_reply_simple("OK",
-					  "result = %d", result,
-					  "dump_len = %d", dump_len,
-					  NULL);
-			if (fd >= 0) {
-				send_dump_buf(fd, dump_len);
-				close(fd);
-			}
-
-		} else {
-			res = daemon_reply_simple("OK",
+		res = daemon_reply_simple("OK",
 					  "result = %d", result,
 					  "protocol = %s", lvmlockd_protocol,
 					  "version = %d", lvmlockd_protocol_version,
 					  NULL);
-		}
-
 		buffer_write(cl->fd, &res.buffer);
 		buffer_destroy(&res.buffer);
 		dm_config_destroy(req.cft);
@@ -4345,6 +4388,9 @@ static void client_recv_action(struct client *cl)
 		goto out;
 	}
 
+	if (act->op == LD_OP_LOCK && act->mode != LD_LK_UN)
+		cl->lock_ops = 1;
+
 	switch (act->op) {
 	case LD_OP_START:
 		rv = add_lockspace(act);
@@ -4352,6 +4398,12 @@ static void client_recv_action(struct client *cl)
 	case LD_OP_STOP:
 		rv = rem_lockspace(act);
 		break;
+	case LD_OP_DUMP_LOG:
+	case LD_OP_DUMP_INFO:
+		/* The client thread reply will copy and send the dump. */
+		add_client_result(act);
+		rv = 0;
+		break;
 	case LD_OP_INIT:
 	case LD_OP_START_WAIT:
 	case LD_OP_STOP_ALL:

daemons/lvmlockd/lvmlockd-internal.h

@@ -85,6 +85,7 @@ struct client {
 	unsigned int recv : 1;
 	unsigned int dead : 1;
 	unsigned int poll_ignore : 1;
+	unsigned int lock_ops : 1;
 	char name[MAX_NAME+1];
 };
 

daemons/lvmpolld/lvmpolld-core.c

@@ -585,12 +585,16 @@ static int spawn_detached_thread(struct lvmpolld_lv *pdlv)
 	int r;
 	pthread_attr_t attr;
 
-	pthread_attr_init(&attr);
-	pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
+	if (pthread_attr_init(&attr) != 0)
+		return 0;
+
+	if (pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED) != 0)
+		return 0;
 
 	r = pthread_create(&pdlv->tid, &attr, fork_and_poll, (void *)pdlv);
 
-	pthread_attr_destroy(&attr);
+	if (pthread_attr_destroy(&attr) != 0)
+		return 0;
 
 	return !r;
 }

lib/config/config.c

@@ -23,7 +23,6 @@
 #include "toolcontext.h"
 #include "lvm-file.h"
 #include "memlock.h"
-#include "segtype.h"
 
 #include <sys/stat.h>
 #include <sys/mman.h>
@@ -582,8 +581,11 @@ int config_file_read(struct dm_config_tree *cft)
 		if (!(cf->dev = dev_create_file(filename, NULL, NULL, 1)))
 			return_0;
 
-		if (!dev_open_readonly_buffered(cf->dev))
+		if (!dev_open_readonly_buffered(cf->dev)) {
+			dev_destroy_file(cf->dev);
+			cf->dev = NULL;
 			return_0;
+		}
 	}
 
 	r = config_file_read_fd(cft, cf->dev, 0, (size_t) info.st_size, 0, 0,
@@ -1692,6 +1694,11 @@ static int _out_prefix_fn(const struct dm_config_node *cn, const char *line, voi
 		if (cfg_def->comment) {
 			int pos = 0;
 			while (_copy_one_line(cfg_def->comment, commentline, &pos, strlen(cfg_def->comment))) {
+				if ((commentline[0] == '#') && (strlen(commentline) == 1)) {
+					if (!out->tree_spec->withspaces)
+						continue;
+					commentline[0] = '\0';
+				}
 				fprintf(out->fp, "%s# %s\n", line, commentline);
 				/* withsummary prints only the first comment line. */
 				if (!out->tree_spec->withcomments)
@@ -1714,6 +1721,9 @@ static int _out_prefix_fn(const struct dm_config_node *cn, const char *line, voi
 		if (cfg_def->flags & CFG_DEFAULT_UNDEFINED)
 			fprintf(out->fp, "%s# This configuration %s does not have a default value defined.\n", line, node_type_name);
 
+		if (cfg_def->flags & CFG_DEFAULT_COMMENTED)
+			fprintf(out->fp, "%s# This configuration %s has an automatic default value.\n", line, node_type_name);
+
 		if ((out->tree_spec->type == CFG_DEF_TREE_FULL) &&
 		    (out->tree_spec->check_status[cn->id] & CFG_USED))
 			fprintf(out->fp, "%s# Value defined in existing configuration has been used for this setting.\n", line);
@@ -2416,27 +2426,3 @@ int get_default_allocation_cache_pool_chunk_size_CFG(struct cmd_context *cmd, st
 {
 	return DEFAULT_CACHE_POOL_CHUNK_SIZE * 2;
 }
-
-const char *get_default_allocation_cache_policy_CFG(struct cmd_context *cmd, struct profile *profile)
-{
-	const struct segment_type *segtype = get_segtype_from_string(cmd, "cache");
-	unsigned attr = ~0;
-
-	if (!segtype ||
-	    !segtype->ops->target_present ||
-	    !segtype->ops->target_present(cmd, NULL, &attr)) {
-		log_warn("WARNING: Cannot detect default cache policy, using \""
-			 DEFAULT_CACHE_POLICY "\".");
-		return DEFAULT_CACHE_POLICY;
-	}
-
-	if (attr & CACHE_FEATURE_POLICY_SMQ)
-		return "smq";
-
-	if (attr & CACHE_FEATURE_POLICY_MQ)
-		return "mq";
-
-	log_warn("WARNING: Default cache policy not available.");
-
-	return NULL;
-}

lib/device/dev-cache.c

@@ -71,6 +71,16 @@ static void _dev_init(struct device *dev, int max_error_count)
 	dm_list_init(&dev->open_list);
 }
 
+void dev_destroy_file(struct device *dev)
+{
+	if (!(dev->flags & DEV_ALLOCED))
+		return;
+
+	dm_free((void *) dm_list_item(dev->aliases.n, struct dm_str_list)->str);
+	dm_free(dev->aliases.n);
+	dm_free(dev);
+}
+
 struct device *dev_create_file(const char *filename, struct device *dev,
 			       struct dm_str_list *alias, int use_malloc)
 {

lib/device/dev-io.c

@@ -586,12 +586,8 @@ static void _close(struct device *dev)
 
 	log_debug_devs("Closed %s", dev_name(dev));
 
-	if (dev->flags & DEV_ALLOCED) {
-		dm_free((void *) dm_list_item(dev->aliases.n, struct dm_str_list)->
-			 str);
-		dm_free(dev->aliases.n);
-		dm_free(dev);
-	}
+	if (dev->flags & DEV_ALLOCED)
+		dev_destroy_file(dev);
 }
 
 static int _dev_close(struct device *dev, int immediate)

lib/device/device.h

@@ -123,6 +123,7 @@ void dev_flush(struct device *dev);
 
 struct device *dev_create_file(const char *filename, struct device *dev,
 			       struct dm_str_list *alias, int use_malloc);
+void dev_destroy_file(struct device *dev);
 
 /* Return a valid device name from the alias list; NULL otherwise */
 const char *dev_name_confirmed(struct device *dev, int quiet);

lib/format_text/format-text.c

@@ -453,8 +453,9 @@ static struct raw_locn *_find_vg_rlocn(struct device_area *dev_area,
 				   "not match expected name %s.", vgname);
 
       bad:
-	if ((info = lvmcache_info_from_pvid(dev_area->dev->pvid, 0)))
-		lvmcache_update_vgname_and_id(info, &vgsummary_orphan);
+	if ((info = lvmcache_info_from_pvid(dev_area->dev->pvid, 0)) &&
+	    !lvmcache_update_vgname_and_id(info, &vgsummary_orphan))
+		stack;
 
 	return NULL;
 }

lib/label/label.c

@@ -105,7 +105,8 @@ static void _update_lvmcache_orphan(struct lvmcache_info *info)
 
         memcpy(&vgsummary_orphan.vgid, lvmcache_fmt(info)->orphan_vg_name, strlen(lvmcache_fmt(info)->orphan_vg_name));
 
-	lvmcache_update_vgname_and_id(info, &vgsummary_orphan);
+	if (!lvmcache_update_vgname_and_id(info, &vgsummary_orphan))
+		stack;
 }
 
 static struct labeller *_find_labeller(struct device *dev, char *buf,

lib/metadata/cache_manip.c

@@ -434,6 +434,34 @@ int lv_is_cache_origin(const struct logical_volume *lv)
 	return seg && lv_is_cache(seg->lv) && !lv_is_pending_delete(seg->lv) && (seg_lv(seg, 0) == lv);
 }
 
+static const char *_get_default_cache_policy(struct cmd_context *cmd)
+{
+	const struct segment_type *segtype = get_segtype_from_string(cmd, "cache");
+	unsigned attr = ~0;
+        const char *def = NULL;
+
+	if (!segtype ||
+	    !segtype->ops->target_present ||
+	    !segtype->ops->target_present(cmd, NULL, &attr)) {
+		log_warn("WARNING: Cannot detect default cache policy, using \""
+			 DEFAULT_CACHE_POLICY "\".");
+		return DEFAULT_CACHE_POLICY;
+	}
+
+	if (attr & CACHE_FEATURE_POLICY_SMQ)
+		def = "smq";
+	else if (attr & CACHE_FEATURE_POLICY_MQ)
+		def = "mq";
+	else {
+		log_error("Default cache policy is not available.");
+		return NULL;
+	}
+
+	log_debug_metadata("Detected default cache_policy \"%s\".", def);
+
+	return def;
+}
+
 int cache_set_policy(struct lv_segment *seg, const char *name,
 		     const struct dm_config_tree *settings)
 {
@@ -451,8 +479,11 @@ int cache_set_policy(struct lv_segment *seg, const char *name,
 			log_error("Failed to duplicate policy name.");
 			return 0;
 		}
-	} else if (!seg->policy_name && passed_seg_is_cache)
-		seg->policy_name = find_config_tree_str(seg->lv->vg->cmd, allocation_cache_policy_CFG, NULL);
+	} else if (!seg->policy_name && passed_seg_is_cache) {
+		if (!(seg->policy_name = find_config_tree_str(seg->lv->vg->cmd, allocation_cache_policy_CFG, NULL)) &&
+		    !(seg->policy_name = _get_default_cache_policy(seg->lv->vg->cmd)))
+			return_0;
+	}
 
 	if (settings) {
 		if (!seg->policy_name) {

lib/metadata/lv_manip.c

@@ -5105,6 +5105,11 @@ static int _lvresize_check_type(struct cmd_context *cmd, const struct logical_vo
 		}
 	}
 
+	if ((lp->resize == LV_REDUCE) && lv_is_thin_pool_metadata(lv)) {
+		log_error("Thin pool metadata volumes cannot be reduced.");
+		return 0;
+	}
+
 	if (lv_is_thin_volume(lv) && first_seg(lv)->external_lv &&
 	    (lp->resize == LV_EXTEND)) {
 		/* Validate thin target supports bigger size of thin volume then external origin */
@@ -5296,7 +5301,7 @@ int lv_resize(struct cmd_context *cmd, struct logical_volume *lv,
 		 */
 		inactive = 1;
 		if (!activate_lv_excl(cmd, lock_lv)) {
-			log_error("Failed to activate %s.", lock_lv->name);
+			log_error("Failed to activate %s.", display_lvname(lock_lv));
 			return 0;
 		}
 	}
@@ -5317,12 +5322,12 @@ int lv_resize(struct cmd_context *cmd, struct logical_volume *lv,
 		backup(vg);
 
 		if (inactive && !deactivate_lv(cmd, lock_lv)) {
-			log_error("Problem deactivating %s.", lock_lv->name);
+			log_error("Problem deactivating %s.", display_lvname(lock_lv));
 			return 0;
 		}
 	}
 
-	log_print_unless_silent("Logical volume %s successfully resized", lp->lv_name);
+	log_print_unless_silent("Logical volume %s successfully resized.", lp->lv_name);
 
 	if (lp->resizefs && (lp->resize == LV_EXTEND) &&
 	    !_fsadm_cmd(cmd, vg, lp, FSADM_CMD_RESIZE, NULL))

lib/metadata/raid_manip.c

@@ -21,7 +21,6 @@
 #include "activate.h"
 #include "lv_alloc.h"
 #include "lvm-string.h"
-#include "lvmlockd.h"
 
 static int _lv_is_raid_with_tracking(const struct logical_volume *lv,
 				     struct logical_volume **tracking)

lib/metadata/thin_manip.c

@@ -63,6 +63,10 @@ int attach_pool_message(struct lv_segment *pool_seg, dm_thin_message_t type,
 
 	tmsg->type = type;
 
+	/* If the 1st message is add in non-read-only mode, modify transaction_id */
+	if (!no_update && dm_list_empty(&pool_seg->thin_messages))
+		pool_seg->transaction_id++;
+
 	dm_list_add(&pool_seg->thin_messages, &tmsg->list);
 
 	log_debug_metadata("Added %s message.",
@@ -476,9 +480,6 @@ int update_pool_lv(struct logical_volume *lv, int activate)
 
 	dm_list_init(&(first_seg(lv)->thin_messages));
 
-	/* thin-pool target transaction is finished, increase lvm2 TID */
-	first_seg(lv)->transaction_id++;
-
 	if (!vg_write(lv->vg) || !vg_commit(lv->vg))
 		return_0;
 

lib/metadata/vg.c

@@ -20,7 +20,6 @@
 #include "toolcontext.h"
 #include "lvmcache.h"
 #include "archiver.h"
-#include "lvmlockd.h"
 
 struct volume_group *alloc_vg(const char *pool_name, struct cmd_context *cmd,
 			      const char *vg_name)

lib/misc/lib.h

@@ -25,14 +25,57 @@
 #define _GNU_SOURCE
 #define _FILE_OFFSET_BITS 64
 
+/*
+ * Symbol export control macros
+ *
+ *   DM_EXPORT_SYMBOL(func,ver)
+ *   DM_EXPORT_SYMBOL_BASE(func,ver)
+ *
+ * For functions that have multiple implementations these macros control
+ * symbol export and versioning.
+ *
+ * Function definitions that exist in only one version never need to use
+ * these macros.
+ *
+ * Backwards compatible implementations must include a version tag of
+ * the form "_v1_02_104" as a suffix to the function name and use the
+ * macro DM_EXPORT_SYMBOL to export the function and bind it to the
+ * specified version string.
+ *
+ * Since versioning is only available when compiling with GCC the entire
+ * compatibility version should be enclosed in '#if defined(__GNUC__)',
+ * for example:
+ *
+ *   int dm_foo(int bar)
+ *   {
+ *     return bar;
+ *   }
+ *
+ *   #if defined(__GNUC__)
+ *   // Backward compatible dm_foo() version 1.02.104
+ *   int dm_foo_v1_02_104(void);
+ *   int dm_foo_v1_02_104(void)
+ *   {
+ *     return 0;
+ *   }
+ *   DM_EXPORT_SYMBOL(dm_foo,1_02_104)
+ *   #endif
+ *
+ * A prototype for the compatibility version is required as these
+ * functions must not be declared static.
+ *
+ * The DM_EXPORT_SYMBOL_BASE macro is only used to export the base
+ * versions of library symbols prior to the introduction of symbol
+ * versioning: it must never be used for new symbols.
+ */
 #if defined(__GNUC__)
-#define DM_EXPORTED_SYMBOL(func, ver) \
-	__asm__(".symver " #func "_v" #ver ", " #func "@@DM_" #ver )
-#define DM_EXPORTED_SYMBOL_BASE(func) \
+#define DM_EXPORT_SYMBOL(func, ver) \
+	__asm__(".symver " #func "_v" #ver ", " #func "@DM_" #ver )
+#define DM_EXPORT_SYMBOL_BASE(func) \
 	__asm__(".symver " #func "_base, " #func "@Base" )
 #else
-#define DM_EXPORTED_SYMBOL(func, ver)
-#define DM_EXPORTED_SYMBOL_BASE(func)
+#define DM_EXPORT_SYMBOL(func, ver)
+#define DM_EXPORT_SYMBOL_BASE(func)
 #endif
 
 

lib/misc/util.h

@@ -37,11 +37,13 @@
 
 /* Define some portable printing types */
 #define PRIsize_t "zu"
+#define PRIssize_t "zd"
 #define PRIptrdiff_t "td"
 #define PRIpid_t PRId32
 
 /* For convenience */
 #define FMTsize_t "%" PRIsize_t
+#define FMTssize_t "%" PRIssize_t
 #define FMTptrdiff_t "%" PRIptrdiff_t
 #define FMTpid_t "%" PRIpid_t
 

lib/report/report.c

@@ -1344,7 +1344,7 @@ static int _cache_settings_disp(struct dm_report *rh, struct dm_pool *mem,
 
 	if (seg_is_cache(seg))
 		seg = first_seg(seg->pool_lv);
-	else {
+	else if (!seg_is_cache_pool(seg)) {
 		dm_list_init(&dummy_list);
 		return _field_set_string_list(rh, field, &dummy_list, private, 0);
 		/* TODO: once we have support for STR_LIST reserved values, replace with:
@@ -1384,7 +1384,7 @@ static int _cache_policy_disp(struct dm_report *rh, struct dm_pool *mem,
 
 	if (seg_is_cache(seg))
 		seg = first_seg(seg->pool_lv);
-	else
+	else if (!seg_is_cache_pool(seg) || !seg->policy_name)
 		return _field_set_value(field, GET_FIRST_RESERVED_NAME(cache_policy_undef),
 					GET_FIELD_RESERVED_VALUE(cache_policy_undef));
 
@@ -2078,7 +2078,7 @@ static int _cachemode_disp(struct dm_report *rh, struct dm_pool *mem,
 	if (seg_is_cache(seg))
 		seg = first_seg(seg->pool_lv);
 
-	if (seg_is_cache_pool(seg)) {
+	if (seg_is_cache_pool(seg) && cache_mode_is_set(seg)) {
 		if (!(cachemode_str = get_cache_mode_name(seg)))
 			return_0;
 

lib/thin/thin.c

@@ -295,7 +295,7 @@ static int _thin_pool_add_target_line(struct dev_manager *dm,
 	}
 
 	if (!dm_tree_node_add_thin_pool_target(node, len,
-					       seg->transaction_id + (laopts->send_messages ? 1 : 0),
+					       seg->transaction_id,
 					       metadata_dlid, pool_dlid,
 					       seg->chunk_size, seg->low_water_mark,
 					       seg->zero_new_blocks ? 0 : 1))
@@ -345,7 +345,7 @@ static int _thin_pool_add_target_line(struct dev_manager *dm,
 				 */
 				if (!lv_thin_pool_transaction_id(seg->lv, &transaction_id))
 					return_0; /* Thin pool should exist and work */
-				if (transaction_id != seg->transaction_id) {
+				if ((transaction_id + 1) != seg->transaction_id) {
 					log_error("Can't create snapshot %s as origin %s is not suspended.",
 						  lmsg->u.lv->name, origin->name);
 					return 0;
@@ -373,11 +373,11 @@ static int _thin_pool_add_target_line(struct dev_manager *dm,
 
 	if (!dm_list_empty(&seg->thin_messages)) {
 		/* Messages were passed, modify transaction_id as the last one */
-		log_debug_activation("Thin pool set transaction id %" PRIu64 ".", seg->transaction_id + 1);
+		log_debug_activation("Thin pool set transaction id %" PRIu64 ".", seg->transaction_id);
 		if (!dm_tree_node_add_thin_pool_message(node,
 							DM_THIN_MESSAGE_SET_TRANSACTION_ID,
-							seg->transaction_id,
-							seg->transaction_id + 1))
+							seg->transaction_id - 1,
+							seg->transaction_id))
 			return_0;
 	}
 

libdaemon/server/daemon-server.c

@@ -539,6 +539,7 @@ void daemon_start(daemon_state s)
 	log_state _log = { { 0 } };
 	thread_state _threads = { .next = NULL };
 	unsigned timeout_count = 0;
+	fd_set in;
 
 	/*
 	 * Switch to C locale to avoid reading large locale-archive file used by
@@ -623,7 +624,6 @@ void daemon_start(daemon_state s)
 
 	while (!_shutdown_requested && !failed) {
 		_reset_timeout(s);
-		fd_set in;
 		FD_ZERO(&in);
 		FD_SET(s.socket_fd, &in);
 		if (select(FD_SETSIZE, &in, NULL, NULL, _get_timeout(s)) < 0 && errno != EINTR)

libdm/.exported_symbols.Base

@@ -179,7 +179,6 @@ dm_task_destroy
 dm_task_enable_checks
 dm_task_get_deps
 dm_task_get_driver_version
-dm_task_get_info
 dm_task_get_info_with_deferred_remove
 dm_task_get_message_response
 dm_task_get_name

libdm/.exported_symbols.DM_1_02_104

@@ -6,7 +6,6 @@ dm_stats_bind_uuid
 dm_stats_buffer_destroy
 dm_stats_clear_region
 dm_stats_create
-dm_stats_create_region
 dm_stats_delete_region
 dm_stats_destroy
 dm_stats_get_area_start
@@ -72,7 +71,6 @@ dm_task_get_ioctl_timestamp
 dm_task_set_record_timestamp
 dm_timestamp_alloc
 dm_timestamp_compare
-dm_timestamp_copy
 dm_timestamp_delta
 dm_timestamp_destroy
 dm_timestamp_get

libdm/.exported_symbols.DM_1_02_105

@@ -0,0 +1,4 @@
+dm_report_is_empty
+dm_stats_get_area_offset
+dm_stats_get_current_area_offset
+dm_timestamp_copy

libdm/.exported_symbols.DM_1_02_106

@@ -0,0 +1,5 @@
+dm_message_supports_precise_timestamps
+dm_stats_create_region
+dm_stats_driver_supports_precise
+dm_stats_get_current_region_precise_timestamps
+dm_stats_get_region_precise_timestamps

libdm/Makefile.in

@@ -55,7 +55,7 @@ include $(top_builddir)/make.tmpl
 
 CFLAGS += $(UDEV_CFLAGS) $(VALGRIND_CFLAGS)
 
-LIBS += $(SELINUX_LIBS) $(UDEV_LIBS) $(PTHREAD_LIBS)
+LIBS += $(SELINUX_LIBS) $(UDEV_LIBS) $(PTHREAD_LIBS) $(M_LIBS)
 
 device-mapper: all
 

libdm/ioctl/libdm-iface.c

@@ -604,6 +604,20 @@ static int dm_inactive_supported(void)
 	return inactive_supported;
 }
 
+int dm_message_supports_precise_timestamps(void)
+{
+	/*
+	 * 4.32.0 supports "precise_timestamps" and "histogram:" options
+	 * to @stats_create messages but lacks the ability to report
+	 * these properties via a subsequent @stats_list: require at
+	 * least 4.33.0 in order to use these features.
+	 */
+	if (dm_check_version() && _dm_version >= 4)
+		if (_dm_version_minor >= 33)
+			return 1;
+	return 0;
+}
+
 void *dm_get_next_target(struct dm_task *dmt, void *next,
 			 uint64_t *start, uint64_t *length,
 			 char **target_type, char **params)
@@ -667,13 +681,7 @@ int dm_format_dev(char *buf, int bufsize, uint32_t dev_major,
 	return 1;
 }
 
-#if defined(__GNUC__)
-int dm_task_get_info_v1_02_97(struct dm_task *dmt, struct dm_info *info);
-DM_EXPORTED_SYMBOL(dm_task_get_info, 1_02_97);
-int dm_task_get_info_v1_02_97(struct dm_task *dmt, struct dm_info *info)
-#else
 int dm_task_get_info(struct dm_task *dmt, struct dm_info *info)
-#endif
 {
 	if (!dmt->dmi.v4)
 		return 0;
@@ -2126,12 +2134,12 @@ void dm_lib_exit(void)
  */
 
 int dm_task_get_info_base(struct dm_task *dmt, struct dm_info *info);
-DM_EXPORTED_SYMBOL_BASE(dm_task_get_info);
+DM_EXPORT_SYMBOL_BASE(dm_task_get_info);
 int dm_task_get_info_base(struct dm_task *dmt, struct dm_info *info)
 {
 	struct dm_info new_info;
 
-	if (!dm_task_get_info_v1_02_97(dmt, &new_info))
+	if (!dm_task_get_info(dmt, &new_info))
 		return 0;
 
 	memcpy(info, &new_info, offsetof(struct dm_info, deferred_remove));
@@ -2144,7 +2152,7 @@ int dm_task_get_info_with_deferred_remove(struct dm_task *dmt, struct dm_info *i
 {
 	struct dm_info new_info;
 
-	if (!dm_task_get_info_v1_02_97(dmt, &new_info))
+	if (!dm_task_get_info(dmt, &new_info))
 		return 0;
 
 	memcpy(info, &new_info, offsetof(struct dm_info, internal_suspend));

libdm/libdevmapper.h

@@ -464,6 +464,38 @@ int dm_stats_bind_name(struct dm_stats *dms, const char *name);
  */
 int dm_stats_bind_uuid(struct dm_stats *dms, const char *uuid);
 
+/*
+ * Test whether the running kernel supports the precise_timestamps
+ * feature. Presence of this feature also implies histogram support.
+ * The library will check this call internally and fails any attempt
+ * to use nanosecond counters or histograms on kernels that fail to
+ * meet this check.
+ */
+int dm_message_supports_precise_timestamps(void);
+
+/*
+ * Precise timetamps support.
+ * 
+ * Test for the presence of precise_timestamps support.
+ */
+int dm_stats_driver_supports_precise(void);
+
+/*
+ * Returns 1 if the specified region has the precise_timestamps feature
+ * enabled (i.e. produces nanosecond-precision counter values) or 0 for
+ * a region using the default milisecond precision.
+ */
+int dm_stats_get_region_precise_timestamps(const struct dm_stats *dms,
+					   uint64_t region_id);
+
+/*
+ * Returns 1 if the region at the current cursor location has the
+ * precise_timestamps feature enabled (i.e. produces
+ * nanosecond-precision counter values) or 0 for a region using the
+ * default milisecond precision.
+ */
+int dm_stats_get_current_region_precise_timestamps(const struct dm_stats *dms);
+
 #define DM_STATS_ALL_PROGRAMS ""
 /*
  * Parse the response from a @stats_list message. dm_stats_list will
@@ -526,6 +558,9 @@ int dm_stats_populate(struct dm_stats *dms, const char *program_id,
  *
  *   dm_stats_create_region(dms, 1024, 1 << 11, -2, p, a);
  *
+ * If precise is non-zero attempt to create a region with nanosecond
+ * precision counters using the kernel precise_timestamps feature.
+ *
  * program_id is an optional string argument that identifies the
  * program creating the region. If program_id is NULL or the empty
  * string the default program_id stored in the handle will be used.
@@ -539,6 +574,7 @@ int dm_stats_populate(struct dm_stats *dms, const char *program_id,
  */
 int dm_stats_create_region(struct dm_stats *dms, uint64_t *region_id,
 			   uint64_t start, uint64_t len, int64_t step,
+			   int precise,
 			   const char *program_id, const char *aux_data);
 
 /*
@@ -697,26 +733,36 @@ int dm_stats_set_program_id(struct dm_stats *dms, int allow_empty,
  *
  * All values are returned in units of 512b sectors.
  */
-uint64_t dm_stats_get_region_start(const struct dm_stats *dms, uint64_t *start,
-				   uint64_t region_id);
-uint64_t dm_stats_get_region_len(const struct dm_stats *dms, uint64_t *len,
-				 uint64_t region_id);
-uint64_t dm_stats_get_region_area_len(const struct dm_stats *dms,
-				      uint64_t *area_len, uint64_t region_id);
+int dm_stats_get_region_start(const struct dm_stats *dms, uint64_t *start,
+			      uint64_t region_id);
+
+int dm_stats_get_region_len(const struct dm_stats *dms, uint64_t *len,
+			    uint64_t region_id);
+
+int dm_stats_get_region_area_len(const struct dm_stats *dms,
+				 uint64_t *len, uint64_t region_id);
 
 /*
- * Area properties: start and length.
+ * Area properties: start, offset and length.
  *
  * The area length is always equal to the area length of the region
  * that contains it and is obtained from dm_stats_get_region_area_len().
  *
- * The start offset of an area is a function of the area_id and the
- * containing region's start and area length.
+ * The start of an area is a function of the area_id and the containing
+ * region's start and area length: it gives the absolute offset into the
+ * containing device of the beginning of the area.
+ *
+ * The offset expresses the area's relative offset into the current
+ * region. I.e. the area start minus the start offset of the containing
+ * region.
  *
  * All values are returned in units of 512b sectors.
  */
-uint64_t dm_stats_get_area_start(const struct dm_stats *dms, uint64_t *start,
-				 uint64_t region_id, uint64_t area_id);
+int dm_stats_get_area_start(const struct dm_stats *dms, uint64_t *start,
+			    uint64_t region_id, uint64_t area_id);
+
+int dm_stats_get_area_offset(const struct dm_stats *dms, uint64_t *offset,
+			     uint64_t region_id, uint64_t area_id);
 
 /*
  * Retrieve program_id and aux_data for a specific region. Only valid
@@ -856,14 +902,14 @@ uint64_t dm_stats_get_current_area(const struct dm_stats *dms);
  *
  * All values are returned in units of 512b sectors.
  */
-uint64_t dm_stats_get_current_region_start(const struct dm_stats *dms,
-					   uint64_t *start);
+int dm_stats_get_current_region_start(const struct dm_stats *dms,
+				      uint64_t *start);
 
-uint64_t dm_stats_get_current_region_len(const struct dm_stats *dms,
-					 uint64_t *len);
+int dm_stats_get_current_region_len(const struct dm_stats *dms,
+				    uint64_t *len);
 
-uint64_t dm_stats_get_current_region_area_len(const struct dm_stats *dms,
-					      uint64_t *area_len);
+int dm_stats_get_current_region_area_len(const struct dm_stats *dms,
+					 uint64_t *area_len);
 
 /*
  * Current area properties: start and length.
@@ -873,10 +919,13 @@ uint64_t dm_stats_get_current_region_area_len(const struct dm_stats *dms,
  *
  * All values are returned in units of 512b sectors.
  */
-uint64_t dm_stats_get_current_area_start(const struct dm_stats *dms,
-					 uint64_t *start);
+int dm_stats_get_current_area_start(const struct dm_stats *dms,
+				    uint64_t *start);
 
-uint64_t dm_stats_get_current_area_len(const struct dm_stats *dms,
+int dm_stats_get_current_area_offset(const struct dm_stats *dms,
+				     uint64_t *offset);
+
+int dm_stats_get_current_area_len(const struct dm_stats *dms,
 				       uint64_t *start);
 
 /*
@@ -2396,6 +2445,11 @@ int dm_report_object_is_selected(struct dm_report *rh, void *object, int do_outp
  */
 int dm_report_compact_fields(struct dm_report *rh);
 
+/*
+ * Returns 1 if there is no data waiting to be output.
+ */
+int dm_report_is_empty(struct dm_report *rh);
+
 int dm_report_output(struct dm_report *rh);
 
 /*

libdm/libdm-deptree.c

@@ -1586,7 +1586,7 @@ static int _node_send_messages(struct dm_tree_node *dnode,
 	}
 
 	/* Error if there are no stacked messages or id mismatches */
-	if (trans_id != (seg->transaction_id - have_messages)) {
+	if ((trans_id + 1) != seg->transaction_id) {
 		log_error("Thin pool %s transaction_id is %" PRIu64 ", while expected %" PRIu64 ".",
 			  _node_name(dnode), trans_id, seg->transaction_id - have_messages);
 		return 0;

libdm/libdm-report.c

@@ -257,7 +257,7 @@ static int _selected_disp(struct dm_report *rh,
 			  const void *data,
 			  void *private __attribute__((unused)))
 {
-	struct row *row = (struct row *)data;
+	const struct row *row = (const struct row *)data;
 	return dm_report_field_int(rh, field, &row->selected);
 }
 
@@ -1387,11 +1387,11 @@ static int _dbl_less_or_equal(double d1, double d2)
 	return _dbl_less(d1, d2) || _dbl_equal(d1, d2);
 }
 
-#define _uint64 *(uint64_t *)
-#define _uint64arr(var,index) ((uint64_t *)var)[index]
+#define _uint64 *(const uint64_t *)
+#define _uint64arr(var,index) ((const uint64_t *)var)[index]
 #define _str (const char *)
-#define _dbl *(double *)
-#define _dblarr(var,index) ((double *)var)[index]
+#define _dbl *(const double *)
+#define _dblarr(var,index) ((const double *)var)[index]
 
 static int _do_check_value_is_strictly_reserved(unsigned type, const void *res_val, int res_range,
 						const void *val, struct field_selection *fs)
@@ -1790,7 +1790,7 @@ static int _compare_selection_field(struct dm_report *rh,
 				r = _cmp_field_int(rh, f->props->field_num, field_id, *(const uint64_t *) f->sort_value, fs);
 				break;
 			case DM_REPORT_FIELD_TYPE_SIZE:
-				r = _cmp_field_double(rh, f->props->field_num, field_id, *(double *) f->sort_value, fs);
+				r = _cmp_field_double(rh, f->props->field_num, field_id, *(const double *) f->sort_value, fs);
 				break;
 			case DM_REPORT_FIELD_TYPE_STRING:
 				r = _cmp_field_string(rh, f->props->field_num, field_id, (const char *) f->sort_value, fs);
@@ -2446,8 +2446,8 @@ static const char *_tok_value_regex(struct dm_report *rh,
 
 static int _str_list_item_cmp(const void *a, const void *b)
 {
-	const struct dm_str_list **item_a = (const struct dm_str_list **) a;
-	const struct dm_str_list **item_b = (const struct dm_str_list **) b;
+	const struct dm_str_list * const *item_a = (const struct dm_str_list * const *) a;
+	const struct dm_str_list * const *item_b = (const struct dm_str_list * const *) b;
 
 	return strcmp((*item_a)->str, (*item_b)->str);
 }
@@ -3357,7 +3357,7 @@ static struct field_selection *_create_field_selection(struct dm_report *rh,
 		memcpy(s, v, len);
 		s[len] = '\0';
 
-		fs->value->v.r = dm_regex_create(rh->selection->mem, (const char **) &s, 1);
+		fs->value->v.r = dm_regex_create(rh->selection->mem, (const char * const *) &s, 1);
 		dm_free(s);
 		if (!fs->value->v.r) {
 			log_error("dm_report: failed to create regex "
@@ -3377,7 +3377,7 @@ static struct field_selection *_create_field_selection(struct dm_report *rh,
 				if (rvw->value) {
 					fs->value->v.s = (const char *) rvw->value;
 					if (rvw->reserved->type & DM_REPORT_FIELD_RESERVED_VALUE_RANGE)
-						fs->value->next->v.s = (((const char **) rvw->value)[1]);
+						fs->value->next->v.s = (((const char * const *) rvw->value)[1]);
 					dm_pool_free(rh->selection->mem, s);
 				} else {
 					fs->value->v.s = s;
@@ -3389,9 +3389,9 @@ static struct field_selection *_create_field_selection(struct dm_report *rh,
 				break;
 			case DM_REPORT_FIELD_TYPE_NUMBER:
 				if (rvw->value) {
-					fs->value->v.i = *(uint64_t *) rvw->value;
+					fs->value->v.i = *(const uint64_t *) rvw->value;
 					if (rvw->reserved->type & DM_REPORT_FIELD_RESERVED_VALUE_RANGE)
-						fs->value->next->v.i = (((uint64_t *) rvw->value)[1]);
+						fs->value->next->v.i = (((const uint64_t *) rvw->value)[1]);
 				} else {
 					if (((fs->value->v.i = strtoull(s, NULL, 10)) == ULLONG_MAX) &&
 						 (errno == ERANGE)) {
@@ -3407,16 +3407,16 @@ static struct field_selection *_create_field_selection(struct dm_report *rh,
 				break;
 			case DM_REPORT_FIELD_TYPE_SIZE:
 				if (rvw->value) {
-					fs->value->v.d = *(double *) rvw->value;
+					fs->value->v.d = *(const double *) rvw->value;
 					if (rvw->reserved->type & DM_REPORT_FIELD_RESERVED_VALUE_RANGE)
-						fs->value->next->v.d = (((double *) rvw->value)[1]);
+						fs->value->next->v.d = (((const double *) rvw->value)[1]);
 				} else {
 					fs->value->v.d = strtod(s, NULL);
 					if (errno == ERANGE) {
 						log_error(_out_of_range_msg, s, field_id);
 						goto error;
 					}
-					if (custom && (factor = *((uint64_t *)custom)))
+					if (custom && (factor = *((const uint64_t *)custom)))
 						fs->value->v.d *= factor;
 					fs->value->v.d /= 512; /* store size in sectors! */
 					if (_check_value_is_strictly_reserved(rh, field_num, DM_REPORT_FIELD_TYPE_SIZE, &fs->value->v.d, NULL)) {
@@ -3428,9 +3428,9 @@ static struct field_selection *_create_field_selection(struct dm_report *rh,
 				break;
 			case DM_REPORT_FIELD_TYPE_PERCENT:
 				if (rvw->value) {
-					fs->value->v.i = *(uint64_t *) rvw->value;
+					fs->value->v.i = *(const uint64_t *) rvw->value;
 					if (rvw->reserved->type & DM_REPORT_FIELD_RESERVED_VALUE_RANGE)
-						fs->value->next->v.i = (((uint64_t *) rvw->value)[1]);
+						fs->value->next->v.i = (((const uint64_t *) rvw->value)[1]);
 				} else {
 					fs->value->v.d = strtod(s, NULL);
 					if ((errno == ERANGE) || (fs->value->v.d < 0) || (fs->value->v.d > 100)) {
@@ -3455,9 +3455,9 @@ static struct field_selection *_create_field_selection(struct dm_report *rh,
 				break;
 			case DM_REPORT_FIELD_TYPE_TIME:
 				if (rvw->value) {
-					fs->value->v.t = *(time_t *) rvw->value;
+					fs->value->v.t = *(const time_t *) rvw->value;
 					if (rvw->reserved->type & DM_REPORT_FIELD_RESERVED_VALUE_RANGE)
-						fs->value->next->v.t = (((time_t *) rvw->value)[1]);
+						fs->value->next->v.t = (((const time_t *) rvw->value)[1]);
 				} else {
 					tval = (struct time_value *) custom;
 					fs->value->v.t = tval->t1;
@@ -4305,6 +4305,11 @@ static int _output_as_columns(struct dm_report *rh)
 	return 0;
 }
 
+int dm_report_is_empty(struct dm_report *rh)
+{
+	return dm_list_empty(&rh->rows) ? 1 : 0;
+}
+
 int dm_report_output(struct dm_report *rh)
 {
 	if (dm_list_empty(&rh->rows))

libdm/libdm-stats.c

@@ -13,14 +13,14 @@
  */
 
 #include "dmlib.h"
-#include "libdm-targets.h"
-#include "libdm-common.h"
 
 #define DM_STATS_REGION_NOT_PRESENT UINT64_MAX
 
 #define NSEC_PER_MSEC   1000000L
 #define NSEC_PER_SEC    1000000000L
 
+#define PRECISE_ARG "precise_timestamps"
+
 /*
  * See Documentation/device-mapper/statistics.txt for full descriptions
  * of the device-mapper statistics counter fields.
@@ -63,7 +63,8 @@ struct dm_stats {
 	uint64_t nr_regions; /* total number of present regions */
 	uint64_t max_region; /* size of the regions table */
 	uint64_t interval_ns;  /* sampling interval in nanoseconds */
-	uint64_t timescale; /* sample value multiplier */
+	uint64_t timescale; /* default sample value multiplier */
+	int precise; /* use precise_timestamps when creating regions */
 	struct dm_stats_region *regions;
 	/* statistics cursor */
 	uint64_t cur_region;
@@ -111,8 +112,9 @@ struct dm_stats *dm_stats_create(const char *program_id)
 	dms->name = NULL;
 	dms->uuid = NULL;
 
-	/* all regions currently use msec precision */
+	/* by default all regions use msec precision */
 	dms->timescale = NSEC_PER_MSEC;
+	dms->precise = 0;
 
 	dms->nr_regions = DM_STATS_REGION_NOT_PRESENT;
 	dms->max_region = DM_STATS_REGION_NOT_PRESENT;
@@ -227,6 +229,19 @@ int dm_stats_bind_uuid(struct dm_stats *dms, const char *uuid)
 	return 1;
 }
 
+static int _stats_check_precise_timestamps(const struct dm_stats *dms)
+{
+	/* Already checked? */
+	if (dms && dms->precise)
+		return 1;
+	return dm_message_supports_precise_timestamps();
+}
+
+int dm_stats_driver_supports_precise(void)
+{
+	return _stats_check_precise_timestamps(NULL);
+}
+
 static struct dm_task *_stats_send_message(struct dm_stats *dms, char *msg)
 {
 	struct dm_task *dmt;
@@ -251,24 +266,55 @@ out:
 
 static int _stats_parse_list_region(struct dm_stats_region *region, char *line)
 {
-	/* FIXME: the kernel imposes no length limit here */
-	char program_id[256], aux_data[256];
+	char *p = NULL, string_data[4096]; /* FIXME: add dm_sscanf with %ms? */
+	const char *program_id, *aux_data, *stats_args;
+	const char *empty_string = "";
 	int r;
 
-	/* line format:
-	 * <region_id>: <start_sector>+<length> <step> <program_id> <aux_data>
-	 */
-	r = sscanf(line, FMTu64 ": " FMTu64 "+" FMTu64 " " FMTu64 "%255s %255s",
-		   &region->region_id, &region->start, &region->len, &region->step,
-		   program_id, aux_data);
+	memset(string_data, 0, sizeof(string_data));
 
-	if (r != 6)
+	/*
+	 * Parse fixed fields, line format:
+	 *
+	 * <region_id>: <start_sector>+<length> <step> <string data>
+	 *
+	 * Maximum string data size is 4096 - 1 bytes.
+	 */
+	r = sscanf(line, FMTu64 ": " FMTu64 "+" FMTu64 " " FMTu64 " %4095c",
+		   &region->region_id, &region->start, &region->len,
+		   &region->step, string_data);
+
+	if (r != 5)
 		return_0;
 
-	if (!strcmp(program_id, "-"))
-		program_id[0] = '\0';
-	if (!strcmp(aux_data, "-"))
-		aux_data[0] = '\0';
+	/* program_id is guaranteed to be first. */
+	program_id = string_data;
+
+	/*
+	 * FIXME: support embedded '\ ' in string data:
+	 *   s/strchr/_find_unescaped_space()/
+	 */
+	if ((p = strchr(string_data, ' '))) {
+		/* terminate program_id string. */
+		*p = '\0';
+		if (!strcmp(program_id, "-"))
+			program_id = empty_string;
+		aux_data = p + 1;
+		if ((p = strchr(aux_data, ' '))) {
+			/* terminate aux_data string. */
+			*p = '\0';
+			if (!strcmp(aux_data, "-"))
+				aux_data = empty_string;
+			stats_args = p + 1;
+		} else
+			stats_args = empty_string;
+	} else
+		aux_data = stats_args = empty_string;
+
+	if (strstr(stats_args, PRECISE_ARG))
+		region->timescale = 1;
+	else
+		region->timescale = NSEC_PER_MSEC;
 
 	if (!(region->program_id = dm_strdup(program_id)))
 		return_0;
@@ -504,14 +550,16 @@ out:
 
 static uint64_t _nr_areas(uint64_t len, uint64_t step)
 {
+	/* Default is one area. */
+	if (!len || !step)
+		return 1;
 	/*
 	 * drivers/md/dm-stats.c::message_stats_create()
 	 * A region may be sub-divided into areas with their own counters.
-	 * If step is non-zero, divide len into that many areas, otherwise
-	 * treat the entire region as a single area. Any partial area at the
-	 * end of the region is treated as an additional complete area.
+	 * Any partial area at the end of the region is treated as an
+	 * additional complete area.
 	 */
-	return (len / (step ? : len)) + !!(len % step);
+	return (len + step - 1) / step;
 }
 
 static uint64_t _nr_areas_region(struct dm_stats_region *region)
@@ -621,14 +669,17 @@ uint64_t dm_stats_get_nr_areas(const struct dm_stats *dms)
 	return nr_areas;
 }
 
-int dm_stats_create_region(struct dm_stats *dms, uint64_t *region_id,
-			   uint64_t start, uint64_t len, int64_t step,
-			   const char *program_id, const char *aux_data)
+static int _stats_create_region(struct dm_stats *dms, uint64_t *region_id,
+				uint64_t start, uint64_t len, int64_t step,
+				int precise,
+				const char *program_id,	const char *aux_data)
 {
 	struct dm_task *dmt = NULL;
 	char msg[1024], range[64];
 	const char *err_fmt = "Could not prepare @stats_create %s.";
-	const char *resp;
+	const char *precise_str = PRECISE_ARG;
+	const char *resp, *opt_args = NULL;
+	int nr_opt = 0, r = 0; /* number of optional args. */
 
 	if (!_stats_bound(dms))
 		return_0;
@@ -640,16 +691,35 @@ int dm_stats_create_region(struct dm_stats *dms, uint64_t *region_id,
 		if (!dm_snprintf(range, sizeof(range), FMTu64 "+" FMTu64,
 				 start, len)) {
 			log_error(err_fmt, "range");
-			goto out;
+			return 0;
 		}
 	}
 
-	if (!dm_snprintf(msg, sizeof(msg), "@stats_create %s %s" FMTu64 " %s %s",
-			 (start || len) ? range : "-",
+	if (precise < 0)
+		precise = dms->precise;
+
+	if (precise)
+		nr_opt++;
+	else
+		precise_str = "";
+
+	if (nr_opt) {
+		if ((dm_asprintf((char **)&opt_args, "%d %s", nr_opt,
+				 precise_str)) < 0) {
+			log_error(err_fmt, PRECISE_ARG " option.");
+			return 0;
+		}
+	} else
+		opt_args = dm_strdup("");
+
+	if (!dm_snprintf(msg, sizeof(msg), "@stats_create %s %s" FMTu64
+			 " %s %s %s", (start || len) ? range : "-",
 			 (step < 0) ? "/" : "",
-			 (uint64_t)llabs(step), program_id, aux_data)) {
+			 (uint64_t)llabs(step),
+			 opt_args, program_id, aux_data)) {
 		log_error(err_fmt, "message");
-		goto out;
+		dm_free((void *) opt_args);
+		return 0;
 	}
 
 	if (!(dmt = _stats_send_message(dms, msg)))
@@ -668,15 +738,46 @@ int dm_stats_create_region(struct dm_stats *dms, uint64_t *region_id,
 			goto_out;
 	}
 
-	dm_task_destroy(dmt);
+	r = 1;
 
-	return 1;
 out:
 	if(dmt)
 		dm_task_destroy(dmt);
-	return 0;
+	dm_free((void *) opt_args);
+	return r;
 }
 
+int dm_stats_create_region(struct dm_stats *dms, uint64_t *region_id,
+			   uint64_t start, uint64_t len, int64_t step,
+			   int precise,
+			   const char *program_id,
+			   const char *aux_data)
+{
+	/* Nanosecond counters need precise_timestamps. */
+	if (precise && !_stats_check_precise_timestamps(dms))
+		return 0;
+
+	return _stats_create_region(dms, region_id, start, len, step,
+				    precise,
+				    program_id, aux_data);
+}
+
+/* Backward compatible dm_stats_create_region() */
+#if defined(__GNUC__)
+int dm_stats_create_region_v1_02_104(struct dm_stats *dms, uint64_t *region_id,
+				     uint64_t start, uint64_t len, int64_t step,
+				     const char *program_id, const char *aux_data);
+int dm_stats_create_region_v1_02_104(struct dm_stats *dms, uint64_t *region_id,
+				     uint64_t start, uint64_t len, int64_t step,
+				     const char *program_id, const char *aux_data)
+{
+	return _stats_create_region(dms, region_id, start, len, step,
+				    dms->precise,
+				    program_id, aux_data);
+}
+DM_EXPORT_SYMBOL(dm_stats_create_region, 1_02_104);
+#endif
+
 int dm_stats_delete_region(struct dm_stats *dms, uint64_t region_id)
 {
 	struct dm_task *dmt;
@@ -815,7 +916,7 @@ static int _dm_stats_populate_region(struct dm_stats *dms, uint64_t region_id,
 	if (!_stats_bound(dms))
 		return_0;
 
-	if (!_stats_parse_region(dms->mem, resp, region, dms->timescale)) {
+	if (!_stats_parse_region(dms->mem, resp, region, region->timescale)) {
 		log_error("Could not parse @stats_print message response.");
 		return 0;
 	}
@@ -1280,7 +1381,7 @@ uint64_t dm_stats_get_current_area(const struct dm_stats *dms)
 	return dms->cur_area;
 }
 
-uint64_t dm_stats_get_region_start(const struct dm_stats *dms, uint64_t *start,
+int dm_stats_get_region_start(const struct dm_stats *dms, uint64_t *start,
 			      uint64_t region_id)
 {
 	if (!dms || !dms->regions)
@@ -1289,7 +1390,7 @@ uint64_t dm_stats_get_region_start(const struct dm_stats *dms, uint64_t *start,
 	return 1;
 }
 
-uint64_t dm_stats_get_region_len(const struct dm_stats *dms, uint64_t *len,
+int dm_stats_get_region_len(const struct dm_stats *dms, uint64_t *len,
 			    uint64_t region_id)
 {
 	if (!dms || !dms->regions)
@@ -1298,51 +1399,69 @@ uint64_t dm_stats_get_region_len(const struct dm_stats *dms, uint64_t *len,
 	return 1;
 }
 
-uint64_t dm_stats_get_region_area_len(const struct dm_stats *dms, uint64_t *step,
-			    uint64_t region_id)
+int dm_stats_get_region_area_len(const struct dm_stats *dms, uint64_t *len,
+				 uint64_t region_id)
 {
 	if (!dms || !dms->regions)
 		return_0;
-	*step = dms->regions[region_id].step;
+	*len = dms->regions[region_id].step;
 	return 1;
 }
 
-uint64_t dm_stats_get_current_region_start(const struct dm_stats *dms,
-					   uint64_t *start)
+int dm_stats_get_current_region_start(const struct dm_stats *dms,
+				      uint64_t *start)
 {
 	return dm_stats_get_region_start(dms, start, dms->cur_region);
 }
 
-uint64_t dm_stats_get_current_region_len(const struct dm_stats *dms,
-					 uint64_t *len)
+int dm_stats_get_current_region_len(const struct dm_stats *dms,
+				    uint64_t *len)
 {
 	return dm_stats_get_region_len(dms, len, dms->cur_region);
 }
 
-uint64_t dm_stats_get_current_region_area_len(const struct dm_stats *dms,
-					      uint64_t *step)
+int dm_stats_get_current_region_area_len(const struct dm_stats *dms,
+					 uint64_t *step)
 {
 	return dm_stats_get_region_area_len(dms, step, dms->cur_region);
 }
 
-uint64_t dm_stats_get_area_start(const struct dm_stats *dms, uint64_t *start,
-				 uint64_t region_id, uint64_t area_id)
+int dm_stats_get_area_start(const struct dm_stats *dms, uint64_t *start,
+			    uint64_t region_id, uint64_t area_id)
 {
+	struct dm_stats_region *region;
 	if (!dms || !dms->regions)
 		return_0;
-	*start = dms->regions[region_id].step * area_id;
+	region = &dms->regions[region_id];
+	*start = region->start + region->step * area_id;
 	return 1;
 }
 
-uint64_t dm_stats_get_current_area_start(const struct dm_stats *dms,
-					 uint64_t *start)
+int dm_stats_get_area_offset(const struct dm_stats *dms, uint64_t *offset,
+			     uint64_t region_id, uint64_t area_id)
+{
+	if (!dms || !dms->regions)
+		return_0;
+	*offset = dms->regions[region_id].step * area_id;
+	return 1;
+}
+
+int dm_stats_get_current_area_start(const struct dm_stats *dms,
+				    uint64_t *start)
 {
 	return dm_stats_get_area_start(dms, start,
 				       dms->cur_region, dms->cur_area);
 }
 
-uint64_t dm_stats_get_current_area_len(const struct dm_stats *dms,
-				       uint64_t *len)
+int dm_stats_get_current_area_offset(const struct dm_stats *dms,
+					  uint64_t *offset)
+{
+	return dm_stats_get_area_offset(dms, offset,
+				       dms->cur_region, dms->cur_area);
+}
+
+int dm_stats_get_current_area_len(const struct dm_stats *dms,
+				  uint64_t *len)
 {
 	return dm_stats_get_region_area_len(dms, len, dms->cur_region);
 }
@@ -1370,3 +1489,15 @@ const char *dm_stats_get_current_region_aux_data(const struct dm_stats *dms)
 {
 	return dm_stats_get_region_aux_data(dms, dms->cur_region);
 }
+
+int dm_stats_get_region_precise_timestamps(const struct dm_stats *dms,
+					   uint64_t region_id)
+{
+	struct dm_stats_region *region = &dms->regions[region_id];
+	return region->timescale == 1;
+}
+
+int dm_stats_get_current_region_precise_timestamps(const struct dm_stats *dms)
+{
+	return dm_stats_get_region_precise_timestamps(dms, dms->cur_region);
+}

libdm/libdm-timestamp.c

@@ -34,7 +34,6 @@
 #ifdef HAVE_REALTIME
 
 #include <time.h>
-#include <bits/time.h>
 
 struct dm_timestamp {
 	struct timespec t;

make.tmpl.in

@@ -54,6 +54,7 @@ LDDEPS += @LDDEPS@
 LIB_SUFFIX = @LIB_SUFFIX@
 LVMINTERNAL_LIBS = -llvm-internal $(DAEMON_LIBS) $(UDEV_LIBS) $(DL_LIBS) $(BLKID_LIBS)
 DL_LIBS = @DL_LIBS@
+M_LIBS = @M_LIBS@
 PTHREAD_LIBS = @PTHREAD_LIBS@
 READLINE_LIBS = @READLINE_LIBS@
 SELINUX_LIBS = @SELINUX_LIBS@
@@ -501,10 +502,10 @@ else
 	set -e;\
 	R=$$(sort $^ | uniq -u);\
 	test -z "$$R" || { echo "Mismatch between symbols in shared library and lists in .exported_symbols.* files: $$R"; false; } ;\
-	for i in $(EXPORTED_SYMBOLS); do\
+	for i in $$(echo $(EXPORTED_SYMBOLS) | tr ' ' '\n' | sort -rnt_ -k5 ); do\
 		echo "$${i##*.} {"; echo "	global:";\
 		$(SED) "s/^/		/;s/$$/;/" $$i;\
-		test "$$i" = Base && { echo "	local:"; echo "		*;"; };\
+		test "$${i##*.}" = Base && { echo "	local:"; echo "		*;"; };\
 		echo "};";\
 	done > $@
 endif

man/cmirrord.8.in

@@ -3,7 +3,7 @@
 cmirrord \(em cluster mirror log daemon
 
 .SH SYNOPSIS
-.B cmirrord
+\fBcmirrord\fR [\fB\-f\fR] [\fB\-h\fR]
 
 .SH DESCRIPTION
 cmirrord is the daemon that tracks mirror log information in a cluster.
@@ -24,6 +24,12 @@ there are still active cluster mirrors, however, the signal will be
 ignored.  Active cluster mirrors should be shutdown before stopping the
 cluster mirror log daemon.
 
+.SH OPTIONS
+.IP "\fB\-f\fR, \fB\-\-foreground\fR" 4
+Do not fork and log to the terminal.
+.IP "\fB\-h\fR, \fB\-\-help\fR" 4
+Print usage.
+
 .SH SEE ALSO
 .BR lvm (8)
 .BR clvmd (8)

man/dmstats.8.in

@@ -27,10 +27,12 @@ dmstats \(em device-mapper statistics management
 .br
 .B dmstats create
 .I device_name
+.RB [ \-\-alldevices ]
 .RB [[ \-\-areas
 .IR nr_areas ]
 .RB |[ \-\-areasize
 .IR area_size ]]
+.RB [ \-\-precise ]
 .RB [[ \-\-start
 .IR start_sector ]
 .RB [ \-\-length
@@ -43,7 +45,7 @@ dmstats \(em device-mapper statistics management
 .br
 .B dmstats delete
 .I device_name
-.RB [ \-\-force ]
+.RB [ \-\-alldevices ]
 .RB [ \-\-allregions
 .RB | \-\-regionid
 .IR id ]
@@ -62,6 +64,8 @@ dmstats \(em device-mapper statistics management
 .RB [ \-\-units
 .IR units ]
 .RB [ \-\-nosuffix ]
+.RB [ \-\-nosuffix ]
+.RB [ \-v | \-\-verbose \ [ \-v | \-\-verbose ]
 .br
 .B dmstats print
 .RI [ device_name ]
@@ -115,10 +119,14 @@ when run as 'dmsetup stats'.
 
 When no device argument is given dmstats will by default operate on all
 device-mapper devices present. The \fBcreate\fP and \fBdelete\fP
-commands require the use of \fB--force\fP when used in this way.
+commands require the use of \fB--alldevices\fP when used in this way.
 
 .SH OPTIONS
 .TP
+.B \-\-alldevices
+If no device arguments are given allow operation on all devices when
+creating or deleting regions.
+.TP
 .B \-\-allprograms
 Include regions from all program IDs for list and report operations.
 .TP
@@ -174,6 +182,10 @@ Specify which report fields to display.
 Sort output according to the list of fields given. Precede any
 sort_field with - for a reverse sort on that column.
 .TP
+.BR \-\-precise
+Attempt to use nanosecond precision counters when creating new
+statistics regions.
+.TP
 .B \-\-programid \fIid
 Specify a program ID string. When creating new statistics regions this
 string is stored with the region. Subsequent operations may supply a
@@ -233,6 +245,7 @@ regions (with the exception of in-flight IO counters).
 .IR nr_areas ]
 .RB [ \-\-areasize
 .IR area_size ]
+.RB [ \-\-precise ]
 .RB [[ \-\-start
 .IR start_sector ]
 .RB [ \-\-length
@@ -252,6 +265,9 @@ at an arbitrary offset into the device. The \fB\-\-segments\fP option
 causes a new region to be created for each target in the corresponding
 device-mapper device's table.
 
+If the \fB\-\-precise\fP option is used the command will attempt to
+create a region using nanosecond precision counters.
+
 An optional \fBprogram_id\fP or \fBaux_data\fP string may be associated
 with the region. A \fBprogram_id\fP may then be used to select regions
 for subsequent list, print, and report operations. The \fBaux_data\fP
@@ -259,7 +275,7 @@ stores an arbitrary string and is not used by dmstats or the
 device-mapper kernel statistics subsystem.
 
 By default dmstats creates regions with a \fBprogram_id\fP of
-"DMSTATS1".
+"dmstats".
 
 On success the \fBregion_id\fP of the newly created region is printed to
 stdout.
@@ -267,7 +283,7 @@ stdout.
 .TP
 .B delete
 .I [ device_name ]
-.RB [ \-\-force ]
+.RB [ \-\-alldevices ]
 .RB [ \-\-allregions
 .RB | \-\-regionid
 .IR id ]
@@ -282,7 +298,8 @@ of subsequent list, print, or report operations.
 All regions registered on a device may be removed using
 \fB\-\-allregions\fP.
 
-To remove all regions on all devices \fB\-\-force\fP must be used.
+To remove all regions on all devices both \fB--allregions\fP and
+\fB\-\-alldevices\fP must be used.
 .br
 .TP
 .B help
@@ -296,11 +313,15 @@ the list of report fields.
 .RI [ device_name ]
 .RB [ \-\-allprograms ]
 .RB [ \-\-programid
+.RB [ \-v | \-\-verbose \ [ \-v | \-\-verbose ]]
 .IR id ]
 .br
 List the statistics regions registered on the device. If the
 \fB\-\-allprograms\fP switch is given all regions will be listed
 regardless of region program ID values.
+
+If \fB\-v\fP or \fB\-\-verbose\fP is given the report will include
+a row of information for each area contained in each region displayed.
 .br
 .TP
 .B print
@@ -495,6 +516,20 @@ The program ID value associated with this region.
 .br
 The auxiliary data value associated with this region.
 .br
+.HP
+.B interval_ns
+.br
+The estimated interval over which the current counter values have
+accumulated. The vaulue is reported as an interger expressed in units
+of nanoseconds.
+.br
+.HP
+.B interval
+.br
+The estimated interval over which the current counter values have
+accumulated. The value is reported as a real number in units of
+seconds.
+.br
 .SS Basic counters
 Basic counters provide access to the raw counter data from the kernel,
 allowing further processing to be carried out by another program.
@@ -565,7 +600,7 @@ Create a whole-device region with one area on vg00/lvol1
 .br
 # dmstats create vg00/lvol1
 .br
-Created region: 0
+vg00/lvol1: Created new region with 1 area(s) as region ID 0
 .br
 .br
 
@@ -575,7 +610,7 @@ Create a 32M region 1G into device d0
 .br
 # dmstats create --start 1G --length 32M d0
 .br
-Created region: 2
+d0: Created new region with 1 area(s) as region ID 0
 .br
 
 
@@ -584,24 +619,24 @@ Create a whole-device region with 8 areas on every device
 .br
 # dmstats create --areas 8
 .br
-Created region: 0
+vg00/lvol1: Created new region with 8 area(s) as region ID 0
 .br
-Created region: 0
+vg00/lvol2: Created new region with 8 area(s) as region ID 0
 .br
-Created region: 0
+vg00/lvol3: Created new region with 8 area(s) as region ID 0
 .br
-Created region: 2
+vg01/lvol0: Created new region with 8 area(s) as region ID 2
 .br
-Created region: 0
+vg01/lvol1: Created new region with 8 area(s) as region ID 0
 .br
-Created region: 0
+vg00/lvol2: Created new region with 8 area(s) as region ID 1
 .br
 .br
 
 Delete all regions on all devices
 .br
 .br
-# dmstats delete --allregions --force
+# dmstats delete --alldevices --allregions
 .br
 .br
 
@@ -611,7 +646,7 @@ using dmsetup
 .br
 # dmsetup stats create --areasize 10G vg00/lvol1
 .br
-Created region: 1
+vg00/lvol1: Created new region with 5 area(s) as region ID 1
 .br
 .br
 
@@ -619,7 +654,7 @@ Create a 1GiB region with 16 areas at the start of vg00/lvol1
 .br
 # dmstats create --start 0 --len 1G --areas=16 vg00/lvol1
 .br
-Created region: 2
+vg00/lvol1: Created new region with 16 area(s) as region ID 0
 .br
 .br
 
@@ -627,13 +662,13 @@ List the statistics regions registered on vg00/lvol1
 .br
 # dmstats list vg00/lvol1
 .br
-RegionID    RegStart    RegLen  AreaSize ProgramID AuxData
+Name             RgID  RStart RSize  #Areas ASize  ProgID
 .br
-0                  0 104857600  20971520 dmstats
+vg00-lvol1           0      0 61.00g      1 61.00g dmstats
 .br
-1                  0 104857600  20971520 dmstats
+vg00-lvol1           1 61.00g 19.20g      1 19.20g dmstats
 .br
-2                  0   2097152    131072 dmstats
+vg00-lvol1           2 80.20g  2.14g      1  2.14g dmstats
 .br
 .br
 
@@ -642,19 +677,15 @@ Display five statistics reports for vg00/lvol1 at an interval of one second
 .br
 # dmstats report --interval 1 --count 5 vg00/lvol1
 .br
-Name             RgID  ArID  RRqM/s   WRqM/s   R/s    W/s   RSz/s   WSz/s AvRqSz QSize SvcTm Util% AWait
+# dmstats report
 .br
-vg00-lvol1       0     0     0.00     0.00   8.00  0.00  48.00k     0  6.00k  0.00  5.50  4.40  6.62
+Name             RgID  ArID  AStart ASize  RRqM/s   WRqM/s   R/s   W/s    RSz/s WSz/s   AvRqSz  QSize Util%      AWait RdAWa WrAWa
 .br
-vg00-lvol1       0     1     0.00     0.00  22.00  0.00 624.00k     0 28.00k  0.00  5.23 11.50  5.36
+vg_hex-lv_home       0     0      0 61.00g     0.00     0.00  0.00 218.00     0   1.04m   4.50k  2.97      81.70 13.62  0.00 13.62
 .br
-vg00-lvol1       0     2     0.00     0.00 353.00  0.00   1.84m     0  5.00k  0.00  1.34 47.40  1.33
+vg_hex-lv_home       1     0 61.00g 19.20g     0.00     0.00  0.00   5.00     0 548.00k 109.50k  0.14      11.00 27.40  0.00 27.40
 .br
-vg00-lvol1       0     3     0.00     0.00  73.00  0.00 592.00k     0  8.00k  0.00  2.10 15.30  2.10
-.br
-vg00-lvol1       0     4     0.00     0.00   5.00  0.00  52.00k     0 10.00k  0.00  4.00  2.00  4.00
-.br
-[...]
+vg_hex-lv_home       2     0 80.20g  2.14g     0.00     0.00  0.00  14.00     0   1.15m  84.00k  0.39      18.70 27.71  0.00 27.71
 .br
 .br
 
@@ -663,11 +694,11 @@ Create one region for reach target contained in device vg00/lvol1
 .br
 # dmstats create --segments vg00/lvol1
 .br
-Created region: 0
+Created new region with 1 area(s) as region ID 0
 .br
-Created region: 1
+Created new region with 1 area(s) as region ID 1
 .br
-Created region: 2
+Created new region with 1 area(s) as region ID 2
 .br
 .br
 

man/lvconvert.8.in

@@ -228,7 +228,7 @@ For more information on cache pool LVs and cache LVs, see \fBlvmcache\fP(7).
 .TP
 .B \-\-cachepolicy \fIpolicy
 Only applicable to cached LVs; see also \fBlvmcache(7)\fP. Sets
-the cache policy. \fImq\fP is the basic policy name. \fIsqm\fP is more advanced
+the cache policy. \fImq\fP is the basic policy name. \fIsmq\fP is more advanced
 version available in newer kernels.
 .TP
 .BR \-\-cachepool " " \fICachePoolLV
@@ -326,7 +326,7 @@ operation from its arguments.
 Separates \fICacheLogicalVolume\fP from cache pool.
 Before the logical volume becomes uncached, cache is flushed.
 The cache pool volume is then left unused and
-could be used e.g. to caching another volume.
+could be used e.g. for caching another volume.
 See also the option \fB\-\-uncache\fP for uncaching and removing
 cache pool with one command.
 .TP

man/lvcreate.8.in

@@ -200,7 +200,7 @@ cache pool LV.
 .TP
 .B \-\-cachepolicy \fIpolicy
 Only applicable to cached LVs; see also \fBlvmcache(7)\fP. Sets
-the cache policy. \fImq\fP is the basic policy name. \fIsqm\fP is more advanced
+the cache policy. \fImq\fP is the basic policy name. \fIsmq\fP is more advanced
 version available in newer kernels.
 .TP
 .IR \fB\-\-cachepool " " CachePoolLogicalVolume { Name | Path }

spec/packages.inc

@@ -164,6 +164,8 @@ fi
 %config(noreplace) %verify(not md5 mtime size) %{_sysconfdir}/lvm/profile/metadata_profile_template.profile
 %config(noreplace) %verify(not md5 mtime size) %{_sysconfdir}/lvm/profile/thin-generic.profile
 %config(noreplace) %verify(not md5 mtime size) %{_sysconfdir}/lvm/profile/thin-performance.profile
+%config(noreplace) %verify(not md5 mtime size) %{_sysconfdir}/lvm/profile/cache-mq.profile
+%config(noreplace) %verify(not md5 mtime size) %{_sysconfdir}/lvm/profile/cache-smq.profile
 %dir %{_sysconfdir}/lvm/backup
 %dir %{_sysconfdir}/lvm/cache
 %dir %{_sysconfdir}/lvm/archive

test/lib/check.sh

@@ -308,6 +308,12 @@ lv_field() {
 		die "lv_field: lv=$1, field=\"$2\", actual=\"$actual\", expected=\"$3\""
 }
 
+lva_field() {
+	local actual=$(get lva_field "$1" "$2" "${@:4}")
+	test "$actual" = "$3" || \
+		die "lva_field: lv=$1, field=\"$2\", actual=\"$actual\", expected=\"$3\""
+}
+
 lv_attr_bit() {
 	local actual=$(get lv_field "$2" lv_attr "${@:4}")
 	local offset=$1

test/lib/get.sh

@@ -42,6 +42,11 @@ lv_field() {
 	trim_ "$r"
 }
 
+lva_field() {
+	local r=$(lvs -a --config 'log{prefix=""}' --noheadings -o "$2" "${@:3}" "$1")
+	trim_ "$r"
+}
+
 lv_devices() {
 	lv_field "$1" devices -a "${@:2}" | sed 's/([^)]*)//g; s/,/\n/g'
 }

test/shell/dlm-remove.sh

@@ -15,6 +15,10 @@ test_description='Remove the dlm test setup'
 
 [ -z "$LVM_TEST_LOCK_TYPE_DLM" ] && skip;
 
+# FIXME: collect debug logs (only if a test failed?)
+# lvmlockctl -d > lvmlockd-debug.txt
+# dlm_tool dump > dlm-debug.txt
+
 systemctl stop dlm
 systemctl stop corosync
 killall lvmlockd

test/shell/lvconvert-cache-smq.sh

@@ -0,0 +1,32 @@
+#!/bin/sh
+# Copyright (C) 2014 Red Hat, Inc. All rights reserved.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# 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., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+# Exercise conversion of cache and cache pool
+
+. lib/inittest
+
+test -e LOCAL_LVMPOLLD && skip
+
+aux have_cache 1 8 0 || skip
+
+aux prepare_vg 5 80
+
+lvcreate --type cache-pool -an -v -L 2 -n cpool $vg
+
+lvcreate -H --cachepolicy smq -L 4 -n corigin --cachepool $vg/cpool
+
+check lv_field $vg/corigin cache_policy "smq"
+
+lvconvert --splitcache $vg/corigin
+
+lvs -o+cache_policy -a $vg
+
+vgremove -f $vg

test/shell/lvconvert-cache.sh

@@ -1,5 +1,5 @@
 #!/bin/sh
-# Copyright (C) 2014 Red Hat, Inc. All rights reserved.
+# Copyright (C) 2014-2015 Red Hat, Inc. All rights reserved.
 #
 # This copyrighted material is made available to anyone wishing to use,
 # modify, copy, or redistribute it subject to the terms and conditions
@@ -37,11 +37,25 @@ lvcreate -an -Zn -L 8 -n $lv3 $vg
 lvcreate -an -Zn -L 8 -n $lv4 $vg
 lvcreate -an -Zn -L 16 -n $lv5 $vg
 
-lvconvert --yes --type cache-pool --cachepool $vg/$lv1
+# check validation of cachemode arg works
+fail lvconvert --yes --type cache-pool --cachemode writethroughX --cachepool $vg/$lv1
+
+# by default no cache settings are attached to converted cache-pool
+lvconvert --yes --type cache-pool --chunksize 256 $vg/$lv1
 check inactive $vg ${lv1}_cdata
-lvconvert --yes --type cache-pool --chunksize 256 $vg/$lv2
+check lv_field $vg/$lv1 cache_mode ""
+check lv_field $vg/$lv1 cache_policy ""
+check lv_field $vg/$lv1 cache_settings ""
+check lv_field $vg/$lv1 chunk_size "256.00k"
+
+# but allow to set them when specified explicitely on command line
+lvconvert --yes --type cache-pool --cachemode writeback --cachepolicy mq \
+	--cachesettings sequential_threshold=1234 --cachesettings random_threshold=56 \
+	--cachepool $vg/$lv2
 check inactive $vg ${lv2}_cdata
-check lv_field $vg/$lv2 chunk_size "256.00k"
+check lv_field $vg/$lv2 cache_mode "writeback"
+check lv_field $vg/$lv2 cache_policy "mq"
+check lv_field $vg/$lv2 cache_settings "random_threshold=56,sequential_threshold=1234"
 
 # Check swap of cache pool metadata
 lvconvert --yes --type cache-pool --poolmetadata $lv4 $vg/$lv3
@@ -58,7 +72,8 @@ lvremove -ff $vg
 
 lvcreate -L 2 -n $lv1 $vg
 lvcreate --type cache-pool -l 1 -n ${lv1}_cachepool $vg
-lvconvert --cache --cachepool $vg/${lv1}_cachepool $vg/$lv1
+lvconvert --cache --cachepool $vg/${lv1}_cachepool --cachemode writeback $vg/$lv1
+check lv_field $vg/$lv1 cache_mode "writeback"
 dmsetup table ${vg}-$lv1 | grep cache  # ensure it is loaded in kernel
 
 #lvconvert --cachepool $vg/${lv1}_cachepool $vg/$lv1

test/shell/lvextend-thin-metadata-dmeventd.sh

@@ -63,8 +63,9 @@ vgcreate -s 1M $vg $(cat DEVICES)
 # Testing dmeventd autoresize
 lvcreate -L200M -V500M -n thin -T $vg/pool 2>&1 | tee out
 not grep "WARNING: Sum" out
+lvcreate -V2M -n thin2 $vg/pool
 lvcreate -L2M -n $lv1 $vg
-lvchange -an $vg/thin $vg/pool
+lvchange -an $vg/thin $vg/thin2 $vg/pool
 
 # Prepare some fake metadata with unmatching id
 # Transaction_id is lower by 1 and there are no message -> ERROR
@@ -86,7 +87,7 @@ grep expected out
 check inactive $vg pool_tmeta
 
 # Prepare some fake metadata prefilled to ~81% (>70%)
-fake_metadata_ 400 1 >data
+fake_metadata_ 400 2 >data
 "$LVM_TEST_THIN_RESTORE_CMD" -i data -o "$DM_DEV_DIR/mapper/$vg-$lv1"
 
 # Swap volume with restored fake metadata

test/shell/lvmlockd-lv-types.sh

@@ -0,0 +1,164 @@
+#!/bin/sh
+# Copyright (C) 2008-2012 Red Hat, Inc. All rights reserved.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# 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., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+test_description='Check lvmlockd lock_args for different LV types'
+
+. lib/inittest
+
+[ -z "$LVM_TEST_LVMLOCKD" ] && skip;
+
+if test -n "$LVM_TEST_LOCK_TYPE_SANLOCK" ; then
+LOCKARGS1="1.0.0:70254592"
+LOCKARGS2="1.0.0:71303168"
+LOCKARGS3="1.0.0:72351744"
+fi
+
+if test -n "$LVM_TEST_LOCK_TYPE_DLM" ; then
+LOCKARGS1="dlm"
+LOCKARGS2="dlm"
+LOCKARGS3="dlm"
+fi
+
+aux prepare_devs 5
+
+vgcreate --shared $vg "$dev1" "$dev2" "$dev3" "$dev4" "$dev5"
+
+#
+# thin pool, thin lv, thin snap
+#
+
+lvcreate -L 8M -n pool1 $vg
+check lva_field $vg/pool1 lockargs $LOCKARGS1
+
+lvcreate -L 8M -n pool1_meta $vg
+check lva_field $vg/pool1_meta lockargs $LOCKARGS2
+
+lvconvert -y --type thin-pool --poolmetadata $vg/pool1_meta $vg/pool1
+check lva_field $vg/pool1 lockargs $LOCKARGS3
+check lva_field $vg/pool1_tdata lockargs ""
+check lva_field $vg/pool1_tmeta lockargs ""
+
+lvcreate -n thin1 -V 1G --thinpool $vg/pool1
+check lva_field $vg/thin1 lockargs ""
+
+lvcreate -s -n snap1 $vg/thin1
+check lva_field $vg/snap1 lockargs ""
+
+lvchange -ay -K $vg/snap1
+
+lvchange -an $vg/snap1
+lvchange -an $vg/thin1
+lvchange -an $vg/pool1
+lvremove $vg/snap1
+lvremove $vg/thin1
+lvremove $vg/pool1
+
+# the first sanlock lock should be found and reused
+lvcreate -L 8M -n lv1 $vg
+check lva_field $vg/lv1 lockargs $LOCKARGS1
+
+lvchange -an $vg/lv1
+lvremove $vg/lv1
+
+
+#
+# with automatic metadata lv
+#
+
+lvcreate -L 8M -n pool2 $vg
+check lva_field $vg/pool2 lockargs $LOCKARGS1
+
+lvconvert -y --type thin-pool $vg/pool2
+check lva_field $vg/pool2 lockargs $LOCKARGS2
+check lva_field $vg/pool2_tdata lockargs ""
+check lva_field $vg/pool2_tmeta lockargs ""
+
+lvcreate -n thin2 -V 1G --thinpool $vg/pool2
+check lva_field $vg/thin2 lockargs ""
+
+lvchange -an $vg/thin2
+lvchange -an $vg/pool2
+lvremove $vg/thin2
+lvremove $vg/pool2
+
+
+#
+# cache pool, cache lv
+#
+
+lvcreate -L 8M -n cache1 $vg
+check lva_field $vg/cache1 lockargs $LOCKARGS1
+
+lvcreate -L 8M -n cache1_meta $vg
+check lva_field $vg/cache1_meta lockargs $LOCKARGS2
+
+lvconvert -y --type cache-pool --poolmetadata $vg/cache1_meta $vg/cache1
+check lva_field $vg/cache1 lockargs ""
+check lva_field $vg/cache1_cdata lockargs ""
+check lva_field $vg/cache1_cmeta lockargs ""
+
+lvcreate -n lv1 -L 8M $vg
+check lva_field $vg/lv1 lockargs $LOCKARGS1
+
+lvconvert -y --type cache --cachepool $vg/cache1 $vg/lv1
+check lva_field $vg/lv1 lockargs $LOCKARGS1
+check lva_field $vg/cache1 lockargs ""
+check lva_field $vg/cache1_cdata lockargs ""
+check lva_field $vg/cache1_cmeta lockargs ""
+
+lvconvert --splitcache $vg/lv1
+check lva_field $vg/lv1 lockargs $LOCKARGS1
+check lva_field $vg/cache1 lockargs ""
+check lva_field $vg/cache1_cdata lockargs ""
+check lva_field $vg/cache1_cmeta lockargs ""
+
+lvchange -an $vg/cache1
+lvchange -an $vg/lv1
+lvremove $vg/cache1
+lvremove $vg/lv1
+
+#
+# cow snap
+#
+
+lvcreate -n lv2 -L 8M $vg
+check lva_field $vg/lv2 lockargs $LOCKARGS1
+
+lvcreate -s -n lv2snap -L 8M $vg/lv2
+check lva_field $vg/lv2 lockargs $LOCKARGS1
+check lva_field $vg/lv2snap lockargs ""
+
+lvchange -y -an $vg/lv2
+lvremove $vg/lv2snap
+lvremove $vg/lv2
+
+#
+# mirror
+#
+
+lvcreate --type mirror -m 1 -n lv3 -L 8M $vg
+check lva_field $vg/lv3 lockargs $LOCKARGS1
+
+lvchange -an $vg/lv3
+lvremove $vg/lv3
+
+#
+# raid1
+#
+
+lvcreate --type raid1 -m 1 -n lv4 -L 8M $vg
+check lva_field $vg/lv4 lockargs $LOCKARGS1
+
+lvchange -an $vg/lv4
+lvremove $vg/lv4
+
+vgremove $vg
+

test/shell/process-each-pv-nomda-all.sh

@@ -13,8 +13,6 @@ test_description='Test process_each_pv with zero mda'
 
 . lib/inittest
 
-test -e LOCAL_LVMPOLLD && skip
-
 aux prepare_devs 14
 
 # for vg1
@@ -41,9 +39,9 @@ pvcreate "$dev14" --metadatacopies 0
 # dev12
 # dev13
 
-vgcreate $vg1 "$dev10"
-vgcreate $vg2 "$dev2" "$dev3" "$dev4" "$dev5"
-vgcreate $vg3 "$dev6" "$dev7" "$dev8" "$dev9"
+vgcreate $SHARED $vg1 "$dev10"
+vgcreate $SHARED $vg2 "$dev2" "$dev3" "$dev4" "$dev5"
+vgcreate $SHARED $vg3 "$dev6" "$dev7" "$dev8" "$dev9"
 
 pvs -a | tee err
 grep "$dev10" err
@@ -59,3 +57,5 @@ grep "$dev11" err
 grep "$dev12" err
 grep "$dev13" err
 grep "$dev14" err
+
+vgremove $vg1 $vg2 $vg3

test/shell/process-each-pv-nomda.sh

@@ -13,15 +13,15 @@ test_description='Test process_each_pv with zero mda'
 
 . lib/inittest
 
-test -e LOCAL_LVMPOLLD && skip
-
 aux prepare_devs 2
 
 pvcreate "$dev1" --metadatacopies 0
 pvcreate "$dev2"
 
-vgcreate $vg1 "$dev1" "$dev2"
+vgcreate $SHARED $vg1 "$dev1" "$dev2"
 
 pvdisplay -a -C | tee err
 grep "$dev1" err
 grep "$dev2" err
+
+vgremove $vg1

test/shell/sanlock-prepare.sh

@@ -73,6 +73,7 @@ prepare_lvmlockd_sanlock() {
 
 GL_DEV="/dev/mapper/GL_DEV"
 GL_FILE="$PWD/gl_file.img"
+dmsetup remove GL_DEV || true
 rm -f "$GL_FILE"
 dd if=/dev/zero of="$GL_FILE" bs=$((1024*1024)) count=1024 2> /dev/null
 GL_LOOP=$(losetup -f "$GL_FILE" --show)
@@ -80,7 +81,7 @@ echo "0 `blockdev --getsize $GL_LOOP` linear $GL_LOOP 0" | dmsetup create GL_DEV
 
 prepare_lvmlockd_sanlock
 
-vgcreate --config 'devices { global_filter=["a|GL_DEV|", "r|.*|"] filter=["a|GL_DEV|", "r|.*|"]}' --lock-type sanlock --lock-gl enable --lock-opt wait glvg $GL_DEV
+vgcreate --config 'devices { global_filter=["a|GL_DEV|", "r|.*|"] filter=["a|GL_DEV|", "r|.*|"]}' --lock-type sanlock glvg $GL_DEV
 
 vgs --config 'devices { global_filter=["a|GL_DEV|", "r|.*|"] filter=["a|GL_DEV|", "r|.*|"]}' -o+locktype,lockargs glvg
 

test/shell/sanlock-remove.sh

@@ -15,14 +15,25 @@ test_description='Remove the sanlock test setup'
 
 [ -z "$LVM_TEST_LOCK_TYPE_SANLOCK" ] && skip;
 
+# FIMXME: get this to run after a test fails
+
 # Removes the VG with the global lock that was created by
 # the corresponding create script.
 
 vgremove --config 'devices { global_filter=["a|GL_DEV|", "r|.*|"] filter=["a|GL_DEV|", "r|.*|"]}' glvg
 
+# FIXME: collect debug logs (only if a test failed?)
+# lvmlockctl -d > lvmlockd-debug.txt
+# sanlock log_dump > sanlock-debug.txt
 
 killall lvmlockd
 killall sanlock
 
-dmsetup remove GL_DEV
-# dmsetup remove glvg-lvmlock
+killall -9 lvmlockd
+killall -9 sanlock
+
+# FIXME: dmsetup remove LVMTEST*-lvmlock
+
+dmsetup remove glvg-lvmlock || true
+dmsetup remove GL_DEV || true
+

test/shell/thin-volume-list.sh

@@ -29,8 +29,11 @@ lvcreate -T -L8M $vg/pool -V10M -n $lv1
 # skip $vg from activation
 aux lvmconf "activation/volume_list = [ \"$vg1\" ]"
 
-# We cannot pass - pool volume cannot be manipulated
-not lvcreate -V10 -n $lv2 -T $vg/pool
+# We still could pass - since pool is still active
+lvcreate -V10 -n $lv2 -T $vg/pool
+
+# but $lv2 is not active
+check inactive $vg $lv2
 
 vgchange -an $vg
 

tools/commands.h

@@ -275,6 +275,8 @@ xx(lvconvert,
    "[--type cache[-pool]|-H|--cache]\n"
    "\t[--cachepool CacheDataLogicalVolume[Path]]\n"
    "\t[--cachemode CacheMode]\n"
+   "\t[--cachepolicy policy]\n"
+   "\t[--cachesettings key=value]\n"
    "\t[--chunksize size]\n"
    "\t[--poolmetadataspare {y|n}]]\n"
    "\t[{--poolmetadata CacheMetadataLogicalVolume[Path] |\n"
@@ -282,7 +284,8 @@ xx(lvconvert,
    COMMON_OPTS
    "\t[Cache|CacheDataPool]LogicalVolume[Path] [PhysicalVolumePath...]\n\n",
 
-   alloc_ARG, background_ARG, cache_ARG, cachemode_ARG, cachepool_ARG, chunksize_ARG,
+   alloc_ARG, background_ARG, cache_ARG, cachemode_ARG,
+   cachepool_ARG, cachepolicy_ARG, cachesettings_ARG, chunksize_ARG,
    corelog_ARG, discards_ARG, force_ARG, interval_ARG, merge_ARG, mirrorlog_ARG,
    mirrors_ARG, name_ARG, noudevsync_ARG, originname_ARG, poolmetadata_ARG,
    poolmetadatasize_ARG, poolmetadataspare_ARG, readahead_ARG, regionsize_ARG,
@@ -301,6 +304,8 @@ xx(lvcreate,
    "\t[--alloc AllocationPolicy]\n"
    "\t[-H|--cache\n"
    "\t  [--cachemode {writeback|writethrough}]\n"
+   "\t  [--cachepolicy policy]\n"
+   "\t  [--cachesettings key=value]\n"
    "\t[--cachepool CachePoolLogicalVolume{Name|Path}]\n"
    "\t[-c|--chunksize ChunkSize]\n"
    "\t[-C|--contiguous {y|n}]\n"

tools/lvconvert.c

@@ -2002,7 +2002,7 @@ static int _lvconvert_uncache(struct cmd_context *cmd,
 		return 0;
 	}
 
-	if (!lv_remove_single(cmd, first_seg(lv)->pool_lv, lp->force, 0))
+	if (!lv_remove_single(cmd, first_seg(lv)->pool_lv, (force_t) lp->force, 0))
 		return_0;
 
 	log_print_unless_silent("Logical volume %s is not cached.", display_lvname(lv));
@@ -2430,7 +2430,7 @@ static int _lvconvert_pool_repair(struct cmd_context *cmd,
 			 * Scan only the 1st. line for transation id.
 			 * Watch out, if the thin_dump format changes
 			 */
-			if ((fgets(meta_path, sizeof(meta_path), f) > 0) &&
+			if (fgets(meta_path, sizeof(meta_path), f) &&
 			    (trans_id_str = strstr(meta_path, "transaction=\"")) &&
 			    (sscanf(trans_id_str + 13, FMTu64, &trans_id) == 1) &&
 			    (trans_id != first_seg(pool_lv)->transaction_id) &&
@@ -2828,7 +2828,7 @@ static int _lvconvert_pool(struct cmd_context *cmd,
 
 		if (!metadata_lv) {
 			if (arg_from_list_is_set(cmd, "is invalid with existing pool",
-						 cachemode_ARG, chunksize_ARG, discards_ARG,
+						 chunksize_ARG, discards_ARG,
 						 zero_ARG, poolmetadatasize_ARG, -1))
 				return_0;
 
@@ -3069,6 +3069,10 @@ mda_write:
 	seg->discards = lp->discards;
 	seg->zero_new_blocks = lp->zero ? 1 : 0;
 
+	if (lp->cache_mode &&
+	    !cache_set_mode(seg, lp->cache_mode))
+		return_0;
+
 	if ((lp->policy_name || lp->policy_settings) &&
 	    !cache_set_policy(seg, lp->policy_name, lp->policy_settings))
 		return_0;
@@ -3450,7 +3454,8 @@ bad:
 	unlock_vg(cmd, lp->vg_name);
 
 	/* Unlock here so it's not held during polling. */
-	lockd_vg(cmd, lp->vg_name, "un", 0, &lockd_state);
+	if (!lockd_vg(cmd, lp->vg_name, "un", 0, &lockd_state))
+		stack;
 
 	release_vg(vg);
 out:

tools/lvcreate.c

@@ -574,7 +574,7 @@ static int _read_cache_params(struct cmd_context *cmd,
 			      &lp->cache_mode,
 			      &lp->policy_name,
 			      &lp->policy_settings))
-		return_NULL;
+		return_0;
 
 	return 1;
 }

tools/polldaemon.c

@@ -202,7 +202,8 @@ int wait_for_single_lv(struct cmd_context *cmd, struct poll_operation_id *id,
 
 		unlock_and_release_vg(cmd, vg, vg->name);
 
-		lockd_vg(cmd, id->vg_name, "un", 0, &lockd_state);
+		if (!lockd_vg(cmd, id->vg_name, "un", 0, &lockd_state))
+			stack;
 
 		/*
 		 * FIXME Sleeping after testing, while preferred, also works around
@@ -225,7 +226,9 @@ int wait_for_single_lv(struct cmd_context *cmd, struct poll_operation_id *id,
 out:
 	if (vg)
 		unlock_and_release_vg(cmd, vg, vg->name);
-	lockd_vg(cmd, id->vg_name, "un", 0, &lockd_state);
+	if (!lockd_vg(cmd, id->vg_name, "un", 0, &lockd_state))
+		stack;
+
 	return ret;
 }
 

tools/pvmove.c

@@ -712,7 +712,8 @@ out_ret:
 	 * for some time monitoring the progress, and we don not want
 	 * or need the lockd lock held over that.
 	 */
-	lockd_vg(cmd, vg_name, "un", 0, &lockd_state);
+	if (!lockd_vg(cmd, vg_name, "un", 0, &lockd_state))
+		stack;
 
 	return r;
 }
@@ -762,7 +763,8 @@ static int _read_poll_id_from_pvname(struct cmd_context *cmd, const char *pv_nam
 
 	unlock_and_release_vg(cmd, vg, vg_name);
 out:
-	lockd_vg(cmd, vg_name, "un", 0, &lockd_state);
+	if (!lockd_vg(cmd, vg_name, "un", 0, &lockd_state))
+		stack;
 	free_pv_fid(pv);
 	return ret;
 }

tools/toollib.c

@@ -1957,7 +1957,8 @@ static int _process_vgnameid_list(struct cmd_context *cmd, uint32_t flags,
 			unlock_vg(cmd, vg_name);
 endvg:
 		release_vg(vg);
-		lockd_vg(cmd, vg_name, "un", 0, &lockd_state);
+		if (!lockd_vg(cmd, vg_name, "un", 0, &lockd_state))
+			stack;
 	}
 
 	/* the VG is selected if at least one LV is selected */
@@ -2438,7 +2439,8 @@ static int _process_lv_vgnameid_list(struct cmd_context *cmd, uint32_t flags,
 		unlock_vg(cmd, vg_name);
 endvg:
 		release_vg(vg);
-		lockd_vg(cmd, vg_name, "un", 0, &lockd_state);
+		if (!lockd_vg(cmd, vg_name, "un", 0, &lockd_state))
+			stack;
 	}
 
 	return ret_max;
@@ -2942,7 +2944,8 @@ static int _process_pvs_in_vgs(struct cmd_context *cmd, uint32_t flags,
 			unlock_vg(cmd, vg->name);
 endvg:
 		release_vg(vg);
-		lockd_vg(cmd, vg_name, "un", 0, &lockd_state);
+		if (!lockd_vg(cmd, vg_name, "un", 0, &lockd_state))
+			stack;
 
 		/* Quit early when possible. */
 		if (!process_all_pvs && dm_list_empty(arg_tags) && dm_list_empty(arg_devices))

tools/vgchange.c

@@ -1227,7 +1227,8 @@ int vgchange(struct cmd_context *cmd, int argc, char **argv)
 	if (arg_is_set(cmd, lockstart_ARG)) {
 		const char *start_opt = arg_str_value(cmd, lockopt_ARG, NULL);
 
-		lockd_gl(cmd, "un", 0);
+		if (!lockd_gl(cmd, "un", 0))
+			stack;
 
 		if (!start_opt || !strcmp(start_opt, "auto")) {
 			log_print_unless_silent("Starting locking.  Waiting until locks are ready...");

tools/vgdisplay.c

@@ -22,8 +22,6 @@ static int vgdisplay_single(struct cmd_context *cmd, const char *vg_name,
 	if (arg_count(cmd, activevolumegroups_ARG) && !lvs_in_vg_activated(vg))
 		return ECMD_PROCESSED;
 
-	vg_check_status(vg, EXPORTED_VG);
-
 	if (arg_count(cmd, colon_ARG)) {
 		vgdisplay_colons(vg);
 		return ECMD_PROCESSED;

udev/Makefile.in

@@ -47,7 +47,7 @@ BLKID_RULE=IMPORT{program}=\"${SBIN}\/blkid -o udev -p \$$tempnode\"
 endif
 
 ifeq ("@UDEV_SYSTEMD_BACKGROUND_JOBS@", "yes")
-PVSCAN_RULE=ACTION\!=\"remove\", ENV{LVM_PV_GONE}==\"1\", RUN\+=\"@bindir@/systemd-run $(LVM_EXEC)\/lvm pvscan --cache \$$major\:\$$minor\", GOTO=\"lvm_end\"\nENV{SYSTEMD_ALIAS}=\"\/dev\/block\/\$$major:\$$minor\"\nENV{ID_MODEL}=\"LVM PV \$$env{ID_FS_UUID_ENC} on \/dev\/\$$name\"\nENV{SYSTEMD_WANTS}+=\"lvm2-pvscan@\$$major:\$$minor.service\"
+PVSCAN_RULE=ACTION\!=\"remove\", ENV{LVM_PV_GONE}==\"1\", RUN\+=\"@bindir@/systemd-run $(LVM_EXEC)\/lvm pvscan --cache \$$major\:\$$minor\", GOTO=\"lvm_end\"\nENV{SYSTEMD_ALIAS}=\"\/dev\/block\/\$$major:\$$minor\"\nENV{ID_MODEL}=\"LVM PV \$$env{ID_FS_UUID_ENC} on \/dev\/\$$name\"\nENV{SYSTEMD_WANTS}\+=\"lvm2-pvscan@\$$major:\$$minor.service\"
 else
 PVSCAN_RULE=RUN\+\=\"$(LVM_EXEC)/lvm pvscan --background --cache --activate ay --major \$$major --minor \$$minor\", ENV{LVM_SCANNED}=\"1\"
 endif