X-Git-Url: https://git.lttng.org/?a=blobdiff_plain;f=src%2Fbin%2Flttng-sessiond%2Fcmd.c;h=01dc3bb7486471844ec358d6dfba5dee25e3f424;hb=33b9609d08bf77e0afcb34d8d14b6fee44ffa0aa;hp=5a122d8d3c8dcf7e7f361b34f69f46c4680c0d22;hpb=1f4962443f25c371e4b54e97f9eb867d67cbf88e;p=lttng-tools.git

diff --git a/src/bin/lttng-sessiond/cmd.c b/src/bin/lttng-sessiond/cmd.c
index 5a122d8d3..01dc3bb74 100644
--- a/src/bin/lttng-sessiond/cmd.c
+++ b/src/bin/lttng-sessiond/cmd.c
@@ -1,21 +1,14 @@
 /*
- * Copyright (C) 2012 - David Goulet <dgoulet@efficios.com>
- * Copyright (C) 2016 - Jérémie Galarneau <jeremie.galarneau@efficios.com>
+ * Copyright (C) 2012 David Goulet <dgoulet@efficios.com>
+ * Copyright (C) 2016 Jérémie Galarneau <jeremie.galarneau@efficios.com>
  *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License, version 2 only, as
- * published by the Free Software Foundation.
+ * SPDX-License-Identifier: GPL-2.0-only
  *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program; if not, write to the Free Software Foundation, Inc., 51
- * Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  */
 
+#include "bin/lttng-sessiond/tracker.h"
+#include "lttng/lttng-error.h"
+#include "lttng/tracker.h"
 #define _LGPL_SOURCE
 #include <assert.h>
 #include <inttypes.h>
@@ -118,6 +111,10 @@ static int cmd_enable_event_internal(struct ltt_session *session,
 		struct lttng_filter_bytecode *filter,
 		struct lttng_event_exclusion *exclusion,
 		int wpipe);
+static int cmd_enable_channel_internal(struct ltt_session *session,
+		const struct lttng_domain *domain,
+		const struct lttng_channel *_attr,
+		int wpipe);
 
 /*
  * Create a session path used by list_lttng_sessions for the case that the
@@ -288,118 +285,6 @@ end:
 	return ret;
 }
 
-/*
- * Fill lttng_channel array of all channels.
- */
-static ssize_t list_lttng_channels(enum lttng_domain_type domain,
-		struct ltt_session *session, struct lttng_channel *channels,
-		struct lttng_channel_extended *chan_exts)
-{
-	int i = 0, ret = 0;
-	struct ltt_kernel_channel *kchan;
-
-	DBG("Listing channels for session %s", session->name);
-
-	switch (domain) {
-	case LTTNG_DOMAIN_KERNEL:
-		/* Kernel channels */
-		if (session->kernel_session != NULL) {
-			cds_list_for_each_entry(kchan,
-					&session->kernel_session->channel_list.head, list) {
-				uint64_t discarded_events, lost_packets;
-				struct lttng_channel_extended *extended;
-
-				extended = (struct lttng_channel_extended *)
-						kchan->channel->attr.extended.ptr;
-
-				ret = get_kernel_runtime_stats(session, kchan,
-						&discarded_events, &lost_packets);
-				if (ret < 0) {
-					goto end;
-				}
-				/* Copy lttng_channel struct to array */
-				memcpy(&channels[i], kchan->channel, sizeof(struct lttng_channel));
-				channels[i].enabled = kchan->enabled;
-				chan_exts[i].discarded_events =
-						discarded_events;
-				chan_exts[i].lost_packets = lost_packets;
-				chan_exts[i].monitor_timer_interval =
-						extended->monitor_timer_interval;
-				chan_exts[i].blocking_timeout = 0;
-				i++;
-			}
-		}
-		break;
-	case LTTNG_DOMAIN_UST:
-	{
-		struct lttng_ht_iter iter;
-		struct ltt_ust_channel *uchan;
-
-		rcu_read_lock();
-		cds_lfht_for_each_entry(session->ust_session->domain_global.channels->ht,
-				&iter.iter, uchan, node.node) {
-			uint64_t discarded_events = 0, lost_packets = 0;
-
-			if (lttng_strncpy(channels[i].name, uchan->name,
-					LTTNG_SYMBOL_NAME_LEN)) {
-				break;
-			}
-			channels[i].attr.overwrite = uchan->attr.overwrite;
-			channels[i].attr.subbuf_size = uchan->attr.subbuf_size;
-			channels[i].attr.num_subbuf = uchan->attr.num_subbuf;
-			channels[i].attr.switch_timer_interval =
-				uchan->attr.switch_timer_interval;
-			channels[i].attr.read_timer_interval =
-				uchan->attr.read_timer_interval;
-			channels[i].enabled = uchan->enabled;
-			channels[i].attr.tracefile_size = uchan->tracefile_size;
-			channels[i].attr.tracefile_count = uchan->tracefile_count;
-
-			/*
-			 * Map enum lttng_ust_output to enum lttng_event_output.
-			 */
-			switch (uchan->attr.output) {
-			case LTTNG_UST_MMAP:
-				channels[i].attr.output = LTTNG_EVENT_MMAP;
-				break;
-			default:
-				/*
-				 * LTTNG_UST_MMAP is the only supported UST
-				 * output mode.
-				 */
-				assert(0);
-				break;
-			}
-
-			chan_exts[i].monitor_timer_interval =
-					uchan->monitor_timer_interval;
-			chan_exts[i].blocking_timeout =
-				uchan->attr.u.s.blocking_timeout;
-
-			ret = get_ust_runtime_stats(session, uchan,
-					&discarded_events, &lost_packets);
-			if (ret < 0) {
-				break;
-			}
-			chan_exts[i].discarded_events = discarded_events;
-			chan_exts[i].lost_packets = lost_packets;
-			i++;
-		}
-		rcu_read_unlock();
-		break;
-	}
-	default:
-		break;
-	}
-
-end:
-	if (ret < 0) {
-		return -LTTNG_ERR_FATAL;
-	} else {
-		return LTTNG_OK;
-	}
-}
-
 static int increment_extended_len(const char *filter_expression,
 		struct lttng_event_exclusion *exclusion,
 		const struct lttng_userspace_probe_location *probe_location,
@@ -517,7 +402,6 @@ static int list_lttng_agent_events(struct agent *agt,
 
 	rcu_read_lock();
 	nb_event = lttng_ht_get_count(agt->events);
-	rcu_read_unlock();
 	if (nb_event == 0) {
 		ret = nb_event;
 		*total_size = 0;
@@ -531,7 +415,6 @@ static int list_lttng_agent_events(struct agent *agt,
 	 * This is only valid because the commands which add events are
 	 * processed in the same thread as the listing.
 	 */
-	rcu_read_lock();
 	cds_lfht_for_each_entry(agt->events->ht, &iter.iter, event, node.node) {
 		ret = increment_extended_len(event->filter_expression, NULL, NULL,
 				&extended_len);
@@ -541,7 +424,6 @@ static int list_lttng_agent_events(struct agent *agt,
 			goto error;
 		}
 	}
-	rcu_read_unlock();
 
 	*total_size = nb_event * sizeof(*tmp_events) + extended_len;
 	tmp_events = zmalloc(*total_size);
@@ -554,7 +436,6 @@ static int list_lttng_agent_events(struct agent *agt,
 	extended_at = ((uint8_t *) tmp_events) +
 		nb_event * sizeof(struct lttng_event);
 
-	rcu_read_lock();
 	cds_lfht_for_each_entry(agt->events->ht, &iter.iter, event, node.node) {
 		strncpy(tmp_events[i].name, event->name, sizeof(tmp_events[i].name));
 		tmp_events[i].name[sizeof(tmp_events[i].name) - 1] = '\0';
@@ -823,8 +704,7 @@ end:
 	return nb_event;
 
 error:
-	/* Negate the error code to differentiate the size from an error */
-	return -ret;
+	return ret;
 }
 
 /*
@@ -998,6 +878,8 @@ static enum lttng_error_code create_connect_relayd(struct lttng_uri *uri,
 
 	/* Create socket for control stream. */
 	if (uri->stype == LTTNG_STREAM_CONTROL) {
+		uint64_t result_flags;
+
 		DBG3("Creating relayd stream socket from URI");
 
 		/* Check relayd version */
@@ -1012,6 +894,16 @@ static enum lttng_error_code create_connect_relayd(struct lttng_uri *uri,
 		}
 		consumer->relay_major_version = rsock->major;
 		consumer->relay_minor_version = rsock->minor;
+		ret = relayd_get_configuration(rsock, 0,
+				&result_flags);
+		if (ret < 0) {
+			ERR("Unable to get relayd configuration");
+			status = LTTNG_ERR_RELAYD_CONNECT_FAIL;
+			goto close_sock;
+		}
+		if (result_flags & LTTCOMM_RELAYD_CONFIGURATION_FLAG_CLEAR_ALLOWED) {
+			consumer->relay_allows_clear = true;
+		}
 	} else if (uri->stype == LTTNG_STREAM_DATA) {
 		DBG3("Creating relayd data socket from URI");
 	} else {
@@ -1185,6 +1077,7 @@ int cmd_setup_relayd(struct ltt_session *session)
 
 	DBG("Setting relayd for session %s", session->name);
 
+	rcu_read_lock();
 	if (session->current_trace_chunk) {
 		enum lttng_trace_chunk_status status = lttng_trace_chunk_get_id(
 				session->current_trace_chunk, &current_chunk_id.value);
@@ -1198,8 +1091,6 @@ int cmd_setup_relayd(struct ltt_session *session)
 		}
 	}
 
-	rcu_read_lock();
-
 	if (usess && usess->consumer && usess->consumer->type == CONSUMER_DST_NET
 			&& usess->consumer->enabled) {
 		/* For each consumer socket, send relayd sockets */
@@ -1225,6 +1116,8 @@ int cmd_setup_relayd(struct ltt_session *session)
 			usess->consumer->relay_major_version;
 		session->consumer->relay_minor_version =
 			usess->consumer->relay_minor_version;
+		session->consumer->relay_allows_clear =
+			usess->consumer->relay_allows_clear;
 	}
 
 	if (ksess && ksess->consumer && ksess->consumer->type == CONSUMER_DST_NET
@@ -1251,6 +1144,8 @@ int cmd_setup_relayd(struct ltt_session *session)
 			ksess->consumer->relay_major_version;
 		session->consumer->relay_minor_version =
 			ksess->consumer->relay_minor_version;
+		session->consumer->relay_allows_clear =
+			ksess->consumer->relay_allows_clear;
 	}
 
 error:
@@ -1261,7 +1156,7 @@ error:
 /*
  * Start a kernel session by opening all necessary streams.
  */
-static int start_kernel_session(struct ltt_kernel_session *ksess)
+int start_kernel_session(struct ltt_kernel_session *ksess)
 {
 	int ret;
 	struct ltt_kernel_channel *kchan;
@@ -1323,6 +1218,53 @@ error:
 	return ret;
 }
 
+int stop_kernel_session(struct ltt_kernel_session *ksess)
+{
+	struct ltt_kernel_channel *kchan;
+	bool error_occurred = false;
+	int ret;
+
+	if (!ksess || !ksess->active) {
+		return LTTNG_OK;
+	}
+	DBG("Stopping kernel tracing");
+
+	ret = kernel_stop_session(ksess);
+	if (ret < 0) {
+		ret = LTTNG_ERR_KERN_STOP_FAIL;
+		goto error;
+	}
+
+	kernel_wait_quiescent();
+
+	/* Flush metadata after stopping (if exists) */
+	if (ksess->metadata_stream_fd >= 0) {
+		ret = kernel_metadata_flush_buffer(ksess->metadata_stream_fd);
+		if (ret < 0) {
+			ERR("Kernel metadata flush failed");
+			error_occurred = true;
+		}
+	}
+
+	/* Flush all buffers after stopping */
+	cds_list_for_each_entry(kchan, &ksess->channel_list.head, list) {
+		ret = kernel_flush_buffer(kchan);
+		if (ret < 0) {
+			ERR("Kernel flush buffer error");
+			error_occurred = true;
+		}
+	}
+
+	ksess->active = 0;
+	if (error_occurred) {
+		ret = LTTNG_ERR_UNK;
+	} else {
+		ret = LTTNG_OK;
+	}
+error:
+	return ret;
+}
+
 /*
  * Command LTTNG_DISABLE_CHANNEL processed by the client thread.
  */
@@ -1380,156 +1322,99 @@ error:
 }
 
 /*
- * Command LTTNG_TRACK_PID processed by the client thread.
+ * Command LTTNG_ENABLE_CHANNEL processed by the client thread.
  *
- * Called with session lock held.
+ * The wpipe arguments is used as a notifier for the kernel thread.
  */
-int cmd_track_pid(struct ltt_session *session, enum lttng_domain_type domain,
-		int pid)
+int cmd_enable_channel(struct command_ctx *cmd_ctx, int sock, int wpipe)
 {
 	int ret;
+	size_t channel_len;
+	ssize_t sock_recv_len;
+	struct lttng_channel *channel = NULL;
+	struct lttng_buffer_view view;
+	struct lttng_dynamic_buffer channel_buffer;
 
-	rcu_read_lock();
-
-	switch (domain) {
-	case LTTNG_DOMAIN_KERNEL:
-	{
-		struct ltt_kernel_session *ksess;
-
-		ksess = session->kernel_session;
-
-		ret = kernel_track_pid(ksess, pid);
-		if (ret != LTTNG_OK) {
-			goto error;
-		}
-
-		kernel_wait_quiescent();
-		break;
+	lttng_dynamic_buffer_init(&channel_buffer);
+	channel_len = (size_t) cmd_ctx->lsm->u.channel.length;
+	ret = lttng_dynamic_buffer_set_size(&channel_buffer, channel_len);
+	if (ret) {
+		ret = LTTNG_ERR_NOMEM;
+		goto end;
 	}
-	case LTTNG_DOMAIN_UST:
-	{
-		struct ltt_ust_session *usess;
-
-		usess = session->ust_session;
 
-		ret = trace_ust_track_pid(usess, pid);
-		if (ret != LTTNG_OK) {
-			goto error;
-		}
-		break;
-	}
-	default:
-		ret = LTTNG_ERR_UNKNOWN_DOMAIN;
-		goto error;
+	sock_recv_len = lttcomm_recv_unix_sock(sock, channel_buffer.data,
+			channel_len);
+	if (sock_recv_len < 0 || sock_recv_len != channel_len) {
+		ERR("Failed to receive \"enable channel\" command payload");
+		ret = LTTNG_ERR_INVALID;
+		goto end;
 	}
 
-	ret = LTTNG_OK;
-
-error:
-	rcu_read_unlock();
-	return ret;
-}
-
-/*
- * Command LTTNG_UNTRACK_PID processed by the client thread.
- *
- * Called with session lock held.
- */
-int cmd_untrack_pid(struct ltt_session *session, enum lttng_domain_type domain,
-		int pid)
-{
-	int ret;
-
-	rcu_read_lock();
-
-	switch (domain) {
-	case LTTNG_DOMAIN_KERNEL:
-	{
-		struct ltt_kernel_session *ksess;
-
-		ksess = session->kernel_session;
-
-		ret = kernel_untrack_pid(ksess, pid);
-		if (ret != LTTNG_OK) {
-			goto error;
-		}
-
-		kernel_wait_quiescent();
-		break;
+	view = lttng_buffer_view_from_dynamic_buffer(&channel_buffer, 0, channel_len);
+	if (!lttng_buffer_view_is_valid(&view)) {
+		ret = LTTNG_ERR_INVALID;
+		goto end;
 	}
-	case LTTNG_DOMAIN_UST:
-	{
-		struct ltt_ust_session *usess;
-
-		usess = session->ust_session;
 
-		ret = trace_ust_untrack_pid(usess, pid);
-		if (ret != LTTNG_OK) {
-			goto error;
-		}
-		break;
-	}
-	default:
-		ret = LTTNG_ERR_UNKNOWN_DOMAIN;
-		goto error;
+	if (lttng_channel_create_from_buffer(&view, &channel) != channel_len) {
+		ERR("Invalid channel payload received in \"enable channel\" command");
+		ret = LTTNG_ERR_INVALID;
+		goto end;
 	}
 
-	ret = LTTNG_OK;
+	ret = cmd_enable_channel_internal(cmd_ctx->session,
+			ALIGNED_CONST_PTR(cmd_ctx->lsm->domain), channel,
+			wpipe);
 
-error:
-	rcu_read_unlock();
+end:
+	lttng_dynamic_buffer_reset(&channel_buffer);
+	lttng_channel_destroy(channel);
 	return ret;
 }
 
-/*
- * Command LTTNG_ENABLE_CHANNEL processed by the client thread.
- *
- * The wpipe arguments is used as a notifier for the kernel thread.
- */
-int cmd_enable_channel(struct ltt_session *session,
-		const struct lttng_domain *domain, const struct lttng_channel *_attr, int wpipe)
+static int cmd_enable_channel_internal(struct ltt_session *session,
+		const struct lttng_domain *domain,
+		const struct lttng_channel *_attr,
+		int wpipe)
 {
 	int ret;
 	struct ltt_ust_session *usess = session->ust_session;
 	struct lttng_ht *chan_ht;
 	size_t len;
-	struct lttng_channel attr;
+	struct lttng_channel *attr = NULL;
 
 	assert(session);
 	assert(_attr);
 	assert(domain);
 
-	attr = *_attr;
-	len = lttng_strnlen(attr.name, sizeof(attr.name));
+	attr = lttng_channel_copy(_attr);
+	if (!attr) {
+		ret = -LTTNG_ERR_NOMEM;
+		goto end;
+	}
+
+	len = lttng_strnlen(attr->name, sizeof(attr->name));
 
 	/* Validate channel name */
-	if (attr.name[0] == '.' ||
-		memchr(attr.name, '/', len) != NULL) {
+	if (attr->name[0] == '.' ||
+		memchr(attr->name, '/', len) != NULL) {
 		ret = LTTNG_ERR_INVALID_CHANNEL_NAME;
 		goto end;
 	}
 
-	DBG("Enabling channel %s for session %s", attr.name, session->name);
+	DBG("Enabling channel %s for session %s", attr->name, session->name);
 
 	rcu_read_lock();
 
-	/*
-	 * Don't try to enable a channel if the session has been started at
-	 * some point in time before. The tracer does not allow it.
-	 */
-	if (session->has_been_started) {
-		ret = LTTNG_ERR_TRACE_ALREADY_STARTED;
-		goto error;
-	}
-
 	/*
 	 * If the session is a live session, remove the switch timer, the
 	 * live timer does the same thing but sends also synchronisation
 	 * beacons for inactive streams.
 	 */
 	if (session->live_timer > 0) {
-		attr.attr.live_timer_interval = session->live_timer;
-		attr.attr.switch_timer_interval = 0;
+		attr->attr.live_timer_interval = session->live_timer;
+		attr->attr.switch_timer_interval = 0;
 	}
 
 	/* Check for feature support */
@@ -1541,8 +1426,8 @@ int cmd_enable_channel(struct ltt_session *session,
 			WARN("Kernel tracer does not support buffer monitoring. "
 					"Setting the monitor interval timer to 0 "
 					"(disabled) for channel '%s' of session '%s'",
-					attr.name, session->name);
-			lttng_channel_set_monitor_timer_interval(&attr, 0);
+					attr->name, session->name);
+			lttng_channel_set_monitor_timer_interval(attr, 0);
 		}
 		break;
 	}
@@ -1567,16 +1452,26 @@ int cmd_enable_channel(struct ltt_session *session,
 	{
 		struct ltt_kernel_channel *kchan;
 
-		kchan = trace_kernel_get_channel_by_name(attr.name,
-				session->kernel_session);
+		kchan = trace_kernel_get_channel_by_name(
+				attr->name, session->kernel_session);
 		if (kchan == NULL) {
+			/*
+			 * Don't try to create a channel if the session has been started at
+			 * some point in time before. The tracer does not allow it.
+			 */
+			if (session->has_been_started) {
+				ret = LTTNG_ERR_TRACE_ALREADY_STARTED;
+				goto error;
+			}
+
 			if (session->snapshot.nb_output > 0 ||
 					session->snapshot_mode) {
 				/* Enforce mmap output for snapshot sessions. */
-				attr.attr.output = LTTNG_EVENT_MMAP;
+				attr->attr.output = LTTNG_EVENT_MMAP;
 			}
-			ret = channel_kernel_create(session->kernel_session, &attr, wpipe);
-			if (attr.name[0] != '\0') {
+			ret = channel_kernel_create(
+					session->kernel_session, attr, wpipe);
+			if (attr->name[0] != '\0') {
 				session->kernel_session->has_non_default_channel = 1;
 			}
 		} else {
@@ -1606,50 +1501,265 @@ int cmd_enable_channel(struct ltt_session *session,
 		 * adhered to.
 		 */
 		if (domain->type == LTTNG_DOMAIN_JUL) {
-			if (strncmp(attr.name, DEFAULT_JUL_CHANNEL_NAME,
+			if (strncmp(attr->name, DEFAULT_JUL_CHANNEL_NAME,
 					LTTNG_SYMBOL_NAME_LEN)) {
 				ret = LTTNG_ERR_INVALID_CHANNEL_NAME;
 				goto error;
 			}
 		} else if (domain->type == LTTNG_DOMAIN_LOG4J) {
-			if (strncmp(attr.name, DEFAULT_LOG4J_CHANNEL_NAME,
+			if (strncmp(attr->name, DEFAULT_LOG4J_CHANNEL_NAME,
 					LTTNG_SYMBOL_NAME_LEN)) {
 				ret = LTTNG_ERR_INVALID_CHANNEL_NAME;
 				goto error;
 			}
 		} else if (domain->type == LTTNG_DOMAIN_PYTHON) {
-			if (strncmp(attr.name, DEFAULT_PYTHON_CHANNEL_NAME,
+			if (strncmp(attr->name, DEFAULT_PYTHON_CHANNEL_NAME,
 					LTTNG_SYMBOL_NAME_LEN)) {
 				ret = LTTNG_ERR_INVALID_CHANNEL_NAME;
 				goto error;
 			}
 		}
 
-		chan_ht = usess->domain_global.channels;
+		chan_ht = usess->domain_global.channels;
+
+		uchan = trace_ust_find_channel_by_name(chan_ht, attr->name);
+		if (uchan == NULL) {
+			/*
+			 * Don't try to create a channel if the session has been started at
+			 * some point in time before. The tracer does not allow it.
+			 */
+			if (session->has_been_started) {
+				ret = LTTNG_ERR_TRACE_ALREADY_STARTED;
+				goto error;
+			}
+
+			ret = channel_ust_create(usess, attr, domain->buf_type);
+			if (attr->name[0] != '\0') {
+				usess->has_non_default_channel = 1;
+			}
+		} else {
+			ret = channel_ust_enable(usess, uchan);
+		}
+		break;
+	}
+	default:
+		ret = LTTNG_ERR_UNKNOWN_DOMAIN;
+		goto error;
+	}
+
+	if (ret == LTTNG_OK && attr->attr.output != LTTNG_EVENT_MMAP) {
+		session->has_non_mmap_channel = true;
+	}
+error:
+	rcu_read_unlock();
+end:
+	lttng_channel_destroy(attr);
+	return ret;
+}
+
+enum lttng_error_code cmd_process_attr_tracker_get_tracking_policy(
+		struct ltt_session *session,
+		enum lttng_domain_type domain,
+		enum lttng_process_attr process_attr,
+		enum lttng_tracking_policy *policy)
+{
+	enum lttng_error_code ret_code = LTTNG_OK;
+	const struct process_attr_tracker *tracker;
+
+	switch (domain) {
+	case LTTNG_DOMAIN_KERNEL:
+		if (!session->kernel_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
+		}
+		tracker = kernel_get_process_attr_tracker(
+				session->kernel_session, process_attr);
+		break;
+	case LTTNG_DOMAIN_UST:
+		if (!session->ust_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
+		}
+		tracker = trace_ust_get_process_attr_tracker(
+				session->ust_session, process_attr);
+		break;
+	default:
+		ret_code = LTTNG_ERR_UNSUPPORTED_DOMAIN;
+		goto end;
+	}
+	if (tracker) {
+		*policy = process_attr_tracker_get_tracking_policy(tracker);
+	} else {
+		ret_code = LTTNG_ERR_INVALID;
+	}
+end:
+	return ret_code;
+}
+
+enum lttng_error_code cmd_process_attr_tracker_set_tracking_policy(
+		struct ltt_session *session,
+		enum lttng_domain_type domain,
+		enum lttng_process_attr process_attr,
+		enum lttng_tracking_policy policy)
+{
+	enum lttng_error_code ret_code = LTTNG_OK;
+
+	switch (policy) {
+	case LTTNG_TRACKING_POLICY_INCLUDE_SET:
+	case LTTNG_TRACKING_POLICY_EXCLUDE_ALL:
+	case LTTNG_TRACKING_POLICY_INCLUDE_ALL:
+		break;
+	default:
+		ret_code = LTTNG_ERR_INVALID;
+		goto end;
+	}
+
+	switch (domain) {
+	case LTTNG_DOMAIN_KERNEL:
+		if (!session->kernel_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
+		}
+		ret_code = kernel_process_attr_tracker_set_tracking_policy(
+				session->kernel_session, process_attr, policy);
+		break;
+	case LTTNG_DOMAIN_UST:
+		if (!session->ust_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
+		}
+		ret_code = trace_ust_process_attr_tracker_set_tracking_policy(
+				session->ust_session, process_attr, policy);
+		break;
+	default:
+		ret_code = LTTNG_ERR_UNSUPPORTED_DOMAIN;
+		break;
+	}
+end:
+	return ret_code;
+}
+
+enum lttng_error_code cmd_process_attr_tracker_inclusion_set_add_value(
+		struct ltt_session *session,
+		enum lttng_domain_type domain,
+		enum lttng_process_attr process_attr,
+		const struct process_attr_value *value)
+{
+	enum lttng_error_code ret_code = LTTNG_OK;
+
+	switch (domain) {
+	case LTTNG_DOMAIN_KERNEL:
+		if (!session->kernel_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
+		}
+		ret_code = kernel_process_attr_tracker_inclusion_set_add_value(
+				session->kernel_session, process_attr, value);
+		break;
+	case LTTNG_DOMAIN_UST:
+		if (!session->ust_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
+		}
+		ret_code = trace_ust_process_attr_tracker_inclusion_set_add_value(
+				session->ust_session, process_attr, value);
+		break;
+	default:
+		ret_code = LTTNG_ERR_UNSUPPORTED_DOMAIN;
+		break;
+	}
+end:
+	return ret_code;
+}
+
+enum lttng_error_code cmd_process_attr_tracker_inclusion_set_remove_value(
+		struct ltt_session *session,
+		enum lttng_domain_type domain,
+		enum lttng_process_attr process_attr,
+		const struct process_attr_value *value)
+{
+	enum lttng_error_code ret_code = LTTNG_OK;
+
+	switch (domain) {
+	case LTTNG_DOMAIN_KERNEL:
+		if (!session->kernel_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
+		}
+		ret_code = kernel_process_attr_tracker_inclusion_set_remove_value(
+				session->kernel_session, process_attr, value);
+		break;
+	case LTTNG_DOMAIN_UST:
+		if (!session->ust_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
+		}
+		ret_code = trace_ust_process_attr_tracker_inclusion_set_remove_value(
+				session->ust_session, process_attr, value);
+		break;
+	default:
+		ret_code = LTTNG_ERR_UNSUPPORTED_DOMAIN;
+		break;
+	}
+end:
+	return ret_code;
+}
+
+enum lttng_error_code cmd_process_attr_tracker_get_inclusion_set(
+		struct ltt_session *session,
+		enum lttng_domain_type domain,
+		enum lttng_process_attr process_attr,
+		struct lttng_process_attr_values **values)
+{
+	enum lttng_error_code ret_code = LTTNG_OK;
+	const struct process_attr_tracker *tracker;
+	enum process_attr_tracker_status status;
 
-		uchan = trace_ust_find_channel_by_name(chan_ht, attr.name);
-		if (uchan == NULL) {
-			ret = channel_ust_create(usess, &attr, domain->buf_type);
-			if (attr.name[0] != '\0') {
-				usess->has_non_default_channel = 1;
-			}
-		} else {
-			ret = channel_ust_enable(usess, uchan);
+	switch (domain) {
+	case LTTNG_DOMAIN_KERNEL:
+		if (!session->kernel_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
 		}
+		tracker = kernel_get_process_attr_tracker(
+				session->kernel_session, process_attr);
+		break;
+	case LTTNG_DOMAIN_UST:
+		if (!session->ust_session) {
+			ret_code = LTTNG_ERR_INVALID;
+			goto end;
+		}
+		tracker = trace_ust_get_process_attr_tracker(
+				session->ust_session, process_attr);
 		break;
-	}
 	default:
-		ret = LTTNG_ERR_UNKNOWN_DOMAIN;
-		goto error;
+		ret_code = LTTNG_ERR_UNSUPPORTED_DOMAIN;
+		goto end;
 	}
 
-	if (ret == LTTNG_OK && attr.attr.output != LTTNG_EVENT_MMAP) {
-		session->has_non_mmap_channel = true;
+	if (!tracker) {
+		ret_code = LTTNG_ERR_INVALID;
+		goto end;
 	}
-error:
-	rcu_read_unlock();
+
+	status = process_attr_tracker_get_inclusion_set(tracker, values);
+	switch (status) {
+	case PROCESS_ATTR_TRACKER_STATUS_OK:
+		ret_code = LTTNG_OK;
+		break;
+	case PROCESS_ATTR_TRACKER_STATUS_INVALID_TRACKING_POLICY:
+		ret_code = LTTNG_ERR_PROCESS_ATTR_TRACKER_INVALID_TRACKING_POLICY;
+		break;
+	case PROCESS_ATTR_TRACKER_STATUS_ERROR:
+		ret_code = LTTNG_ERR_NOMEM;
+		break;
+	default:
+		ret_code = LTTNG_ERR_UNK;
+		break;
+	}
+
 end:
-	return ret;
+	return ret_code;
 }
 
 /*
@@ -2070,7 +2180,8 @@ static int _cmd_enable_event(struct ltt_session *session,
 				goto error;
 			}
 
-			ret = cmd_enable_channel(session, domain, attr, wpipe);
+			ret = cmd_enable_channel_internal(
+					session, domain, attr, wpipe);
 			if (ret != LTTNG_OK) {
 				goto error;
 			}
@@ -2209,7 +2320,8 @@ static int _cmd_enable_event(struct ltt_session *session,
 				goto error;
 			}
 
-			ret = cmd_enable_channel(session, domain, attr, wpipe);
+			ret = cmd_enable_channel_internal(
+					session, domain, attr, wpipe);
 			if (ret != LTTNG_OK) {
 				goto error;
 			}
@@ -2510,57 +2622,6 @@ ssize_t cmd_list_syscalls(struct lttng_event **events)
 	return syscall_table_list(events);
 }
 
-/*
- * Command LTTNG_LIST_TRACKER_PIDS processed by the client thread.
- *
- * Called with session lock held.
- */
-ssize_t cmd_list_tracker_pids(struct ltt_session *session,
-		enum lttng_domain_type domain, int32_t **pids)
-{
-	int ret;
-	ssize_t nr_pids = 0;
-
-	switch (domain) {
-	case LTTNG_DOMAIN_KERNEL:
-	{
-		struct ltt_kernel_session *ksess;
-
-		ksess = session->kernel_session;
-		nr_pids = kernel_list_tracker_pids(ksess, pids);
-		if (nr_pids < 0) {
-			ret = LTTNG_ERR_KERN_LIST_FAIL;
-			goto error;
-		}
-		break;
-	}
-	case LTTNG_DOMAIN_UST:
-	{
-		struct ltt_ust_session *usess;
-
-		usess = session->ust_session;
-		nr_pids = trace_ust_list_tracker_pids(usess, pids);
-		if (nr_pids < 0) {
-			ret = LTTNG_ERR_UST_LIST_FAIL;
-			goto error;
-		}
-		break;
-	}
-	case LTTNG_DOMAIN_LOG4J:
-	case LTTNG_DOMAIN_JUL:
-	case LTTNG_DOMAIN_PYTHON:
-	default:
-		ret = LTTNG_ERR_UND;
-		goto error;
-	}
-
-	return nr_pids;
-
-error:
-	/* Return negative value to differentiate return code */
-	return -ret;
-}
-
 /*
  * Command LTTNG_START_TRACE processed by the client thread.
  *
@@ -2574,6 +2635,8 @@ int cmd_start_trace(struct ltt_session *session)
 	struct ltt_ust_session *usess;
 	const bool session_rotated_after_last_stop =
 			session->rotated_after_last_stop;
+	const bool session_cleared_after_last_stop =
+			session->cleared_after_last_stop;
 
 	assert(session);
 
@@ -2584,7 +2647,8 @@ int cmd_start_trace(struct ltt_session *session)
 	/* Is the session already started? */
 	if (session->active) {
 		ret = LTTNG_ERR_TRACE_ALREADY_STARTED;
-		goto error;
+		/* Perform nothing */
+		goto end;
 	}
 
 	if (session->rotation_state == LTTNG_ROTATION_STATE_ONGOING &&
@@ -2620,6 +2684,7 @@ int cmd_start_trace(struct ltt_session *session)
 
 	session->active = 1;
 	session->rotated_after_last_stop = false;
+	session->cleared_after_last_stop = false;
 	if (session->output_traces && !session->current_trace_chunk) {
 		if (!session->has_been_started) {
 			struct lttng_trace_chunk *trace_chunk;
@@ -2653,7 +2718,8 @@ int cmd_start_trace(struct ltt_session *session)
 			 * was produced as the session was stopped, so the
 			 * rotation should happen on reception of the command.
 			 */
-			ret = cmd_rotate_session(session, NULL, true);
+			ret = cmd_rotate_session(session, NULL, true,
+					LTTNG_TRACE_CHUNK_COMMAND_TYPE_NO_OPERATION);
 			if (ret != LTTNG_OK) {
 				goto error;
 			}
@@ -2679,6 +2745,16 @@ int cmd_start_trace(struct ltt_session *session)
 		}
 	}
 
+	/*
+	 * Open a packet in every stream of the session to ensure that viewers
+	 * can correctly identify the boundaries of the periods during which
+	 * tracing was active for this session.
+	 */
+	ret = session_open_packets(session);
+	if (ret != LTTNG_OK) {
+		goto error;
+	}
+
 	/*
 	 * Clear the flag that indicates that a rotation was done while the
 	 * session was stopped.
@@ -2707,7 +2783,10 @@ error:
 		/* Restore initial state on error. */
 		session->rotated_after_last_stop =
 				session_rotated_after_last_stop;
+		session->cleared_after_last_stop =
+				session_cleared_after_last_stop;
 	}
+end:
 	return ret;
 }
 
@@ -2717,14 +2796,12 @@ error:
 int cmd_stop_trace(struct ltt_session *session)
 {
 	int ret;
-	struct ltt_kernel_channel *kchan;
 	struct ltt_kernel_session *ksession;
 	struct ltt_ust_session *usess;
-	bool error_occurred = false;
 
 	assert(session);
 
-	DBG("Begin stop session %s (id %" PRIu64 ")", session->name, session->id);
+	DBG("Begin stop session \"%s\" (id %" PRIu64 ")", session->name, session->id);
 	/* Short cut */
 	ksession = session->kernel_session;
 	usess = session->ust_session;
@@ -2735,39 +2812,9 @@ int cmd_stop_trace(struct ltt_session *session)
 		goto error;
 	}
 
-	/* Kernel tracer */
-	if (ksession && ksession->active) {
-		DBG("Stop kernel tracing");
-
-		ret = kernel_stop_session(ksession);
-		if (ret < 0) {
-			ret = LTTNG_ERR_KERN_STOP_FAIL;
-			goto error;
-		}
-
-		kernel_wait_quiescent();
-
-		/* Flush metadata after stopping (if exists) */
-		if (ksession->metadata_stream_fd >= 0) {
-			ret = kernel_metadata_flush_buffer(ksession->metadata_stream_fd);
-			if (ret < 0) {
-				ERR("Kernel metadata flush failed");
-				error_occurred = true;
-			}
-		}
-
-		/* Flush all buffers after stopping */
-		cds_list_for_each_entry(kchan, &ksession->channel_list.head, list) {
-			ret = kernel_flush_buffer(kchan);
-			if (ret < 0) {
-				ERR("Kernel flush buffer error");
-				error_occurred = true;
-			}
-		}
-
-		ksession->active = 0;
-		DBG("Kernel session stopped %s (id %" PRIu64 ")", session->name,
-				session->id);
+	ret = stop_kernel_session(ksession);
+	if (ret != LTTNG_OK) {
+		goto error;
 	}
 
 	if (usess && usess->active) {
@@ -2778,9 +2825,11 @@ int cmd_stop_trace(struct ltt_session *session)
 		}
 	}
 
+	DBG("Completed stop session \"%s\" (id %" PRIu64 ")", session->name,
+			session->id);
 	/* Flag inactive after a successful stop. */
 	session->active = 0;
-	ret = !error_occurred ? LTTNG_OK : LTTNG_ERR_UNK;
+	ret = LTTNG_OK;
 
 error:
 	return ret;
@@ -3213,6 +3262,7 @@ void cmd_destroy_session_reply(const struct ltt_session *session,
 	payload_size_before_location = payload.size;
 	comm_ret = lttng_trace_archive_location_serialize(location,
 			&payload);
+	lttng_trace_archive_location_put(location);
 	if (comm_ret < 0) {
 		ERR("Failed to serialize the location of the trace archive produced during the destruction of session \"%s\"",
 				session->name);
@@ -3290,14 +3340,13 @@ int cmd_destroy_session(struct ltt_session *session,
 		session->rotate_size = 0;
 	}
 
-	if (session->most_recent_chunk_id.is_set &&
-			session->most_recent_chunk_id.value != 0 &&
-			session->current_trace_chunk && session->output_traces) {
+	if (session->rotated && session->current_trace_chunk && session->output_traces) {
 		/*
 		 * Perform a last rotation on destruction if rotations have
 		 * occurred during the session's lifetime.
 		 */
-		ret = cmd_rotate_session(session, NULL, false);
+		ret = cmd_rotate_session(session, NULL, false,
+			LTTNG_TRACE_CHUNK_COMMAND_TYPE_MOVE_TO_COMPLETED);
 		if (ret != LTTNG_OK) {
 			ERR("Failed to perform an implicit rotation as part of the destruction of session \"%s\": %s",
 					session->name, lttng_strerror(-ret));
@@ -3316,10 +3365,19 @@ int cmd_destroy_session(struct ltt_session *session,
 		 * emitted and no renaming of the current trace chunk takes
 		 * place.
 		 */
-		ret = cmd_rotate_session(session, NULL, true);
-		if (ret != LTTNG_OK) {
+		ret = cmd_rotate_session(session, NULL, true,
+			LTTNG_TRACE_CHUNK_COMMAND_TYPE_NO_OPERATION);
+		/*
+		 * Rotation operations may not be supported by the kernel
+		 * tracer. Hence, do not consider this implicit rotation as
+		 * a session destruction error. The library has already stopped
+		 * the session and waited for pending data; there is nothing
+		 * left to do but complete the destruction of the session.
+		 */
+		if (ret != LTTNG_OK &&
+				ret != -LTTNG_ERR_ROTATION_NOT_AVAILABLE_KERNEL) {
 			ERR("Failed to perform a quiet rotation as part of the destruction of session \"%s\": %s",
-					session->name, lttng_strerror(-ret));
+			    session->name, lttng_strerror(ret));
 			destruction_last_error = -ret;
 		}
 	}
@@ -3557,67 +3615,113 @@ error:
 /*
  * Command LTTNG_LIST_CHANNELS processed by the client thread.
  */
-ssize_t cmd_list_channels(enum lttng_domain_type domain,
-		struct ltt_session *session, struct lttng_channel **channels)
+enum lttng_error_code cmd_list_channels(enum lttng_domain_type domain,
+		struct ltt_session *session,
+		struct lttng_dynamic_buffer *buffer,
+		uint32_t *nb_channel)
 {
-	ssize_t nb_chan = 0, payload_size = 0, ret;
+	int ret = 0;
+	uint32_t i = 0;
+
+	assert(session);
+	assert(buffer);
+	assert(nb_channel);
+
+	DBG("Listing channels for session %s", session->name);
 
 	switch (domain) {
 	case LTTNG_DOMAIN_KERNEL:
+	{
+		/* Kernel channels */
+		struct ltt_kernel_channel *kchan;
 		if (session->kernel_session != NULL) {
-			nb_chan = session->kernel_session->channel_count;
-		}
-		DBG3("Number of kernel channels %zd", nb_chan);
-		if (nb_chan <= 0) {
-			ret = -LTTNG_ERR_KERN_CHAN_NOT_FOUND;
-			goto end;
-		}
-		break;
-	case LTTNG_DOMAIN_UST:
-		if (session->ust_session != NULL) {
-			rcu_read_lock();
-			nb_chan = lttng_ht_get_count(
-				session->ust_session->domain_global.channels);
-			rcu_read_unlock();
-		}
-		DBG3("Number of UST global channels %zd", nb_chan);
-		if (nb_chan < 0) {
-			ret = -LTTNG_ERR_UST_CHAN_NOT_FOUND;
-			goto end;
+			cds_list_for_each_entry(kchan,
+					&session->kernel_session->channel_list.head, list) {
+				uint64_t discarded_events, lost_packets;
+				struct lttng_channel_extended *extended;
+
+				extended = (struct lttng_channel_extended *)
+						kchan->channel->attr.extended.ptr;
+
+				ret = get_kernel_runtime_stats(session, kchan,
+						&discarded_events, &lost_packets);
+				if (ret < 0) {
+					goto end;
+				}
+
+				/*
+				 * Update the discarded_events and lost_packets
+				 * count for the channel
+				 */
+				extended->discarded_events = discarded_events;
+				extended->lost_packets = lost_packets;
+
+				ret = lttng_channel_serialize(
+						kchan->channel, buffer);
+				if (ret) {
+					ret = -1;
+					goto end;
+				}
+
+				i++;
+			}
 		}
 		break;
-	default:
-		ret = -LTTNG_ERR_UND;
-		goto end;
 	}
+	case LTTNG_DOMAIN_UST:
+	{
+		struct lttng_ht_iter iter;
+		struct ltt_ust_channel *uchan;
 
-	if (nb_chan > 0) {
-		const size_t channel_size = sizeof(struct lttng_channel) +
-			sizeof(struct lttng_channel_extended);
-		struct lttng_channel_extended *channel_exts;
+		rcu_read_lock();
+		cds_lfht_for_each_entry(session->ust_session->domain_global.channels->ht,
+				&iter.iter, uchan, node.node) {
+			uint64_t discarded_events = 0, lost_packets = 0;
+			struct lttng_channel *channel = NULL;
+			struct lttng_channel_extended *extended;
 
-		payload_size = nb_chan * channel_size;
-		*channels = zmalloc(payload_size);
-		if (*channels == NULL) {
-			ret = -LTTNG_ERR_FATAL;
-			goto end;
-		}
+			channel = trace_ust_channel_to_lttng_channel(uchan);
+			if (!channel) {
+				ret = -1;
+				break;
+			}
 
-		channel_exts = ((void *) *channels) +
-				(nb_chan * sizeof(struct lttng_channel));
-		ret = list_lttng_channels(domain, session, *channels, channel_exts);
-		if (ret != LTTNG_OK) {
-			free(*channels);
-			*channels = NULL;
-			goto end;
+			extended = (struct lttng_channel_extended *)
+						   channel->attr.extended.ptr;
+
+			ret = get_ust_runtime_stats(session, uchan,
+					&discarded_events, &lost_packets);
+			if (ret < 0) {
+				lttng_channel_destroy(channel);
+				break;
+			}
+
+			extended->discarded_events = discarded_events;
+			extended->lost_packets = lost_packets;
+
+			ret = lttng_channel_serialize(channel, buffer);
+			if (ret) {
+				ret = -1;
+				break;
+			}
+
+			i++;
 		}
-	} else {
-		*channels = NULL;
+		rcu_read_unlock();
+		break;
+	}
+	default:
+		break;
 	}
 
-	ret = payload_size;
 end:
-	return ret;
+	if (ret < 0) {
+		*nb_channel = 0;
+		return LTTNG_ERR_FATAL;
+	} else {
+		*nb_channel = i;
+		return LTTNG_OK;
+	}
 }
 
 /*
@@ -4541,7 +4645,7 @@ int64_t get_session_nb_packets_per_stream(const struct ltt_session *session,
 		}
 		cur_nb_packets++;
 	}
-	if (!cur_nb_packets) {
+	if (!cur_nb_packets && size_left != max_size) {
 		/* Not enough room to grab one packet of each stream, error. */
 		return -1;
 	}
@@ -4588,6 +4692,11 @@ enum lttng_error_code snapshot_record(struct ltt_session *session,
 				consumer_copy_output(snapshot_output->consumer);
 		strcpy(snapshot_kernel_consumer_output->chunk_path,
 			snapshot_chunk_name);
+
+		/* Copy the original domain subdir. */
+		strcpy(snapshot_kernel_consumer_output->domain_subdir,
+				original_kernel_consumer_output->domain_subdir);
+
 		ret = consumer_copy_sockets(snapshot_kernel_consumer_output,
 				original_kernel_consumer_output);
 		if (ret < 0) {
@@ -4610,6 +4719,11 @@ enum lttng_error_code snapshot_record(struct ltt_session *session,
 				consumer_copy_output(snapshot_output->consumer);
 		strcpy(snapshot_ust_consumer_output->chunk_path,
 			snapshot_chunk_name);
+
+		/* Copy the original domain subdir. */
+		strcpy(snapshot_ust_consumer_output->domain_subdir,
+				original_ust_consumer_output->domain_subdir);
+
 		ret = consumer_copy_sockets(snapshot_ust_consumer_output,
 				original_ust_consumer_output);
 		if (ret < 0) {
@@ -4654,7 +4768,7 @@ enum lttng_error_code snapshot_record(struct ltt_session *session,
 			snapshot_output->max_size);
 	if (nb_packets_per_stream < 0) {
 		ret_code = LTTNG_ERR_MAX_SIZE_INVALID;
-		goto error;
+		goto error_close_trace_chunk;
 	}
 
 	if (session->kernel_session) {
@@ -4662,7 +4776,7 @@ enum lttng_error_code snapshot_record(struct ltt_session *session,
 				snapshot_kernel_consumer_output, session,
 				wait, nb_packets_per_stream);
 		if (ret_code != LTTNG_OK) {
-			goto error;
+			goto error_close_trace_chunk;
 		}
 	}
 
@@ -4671,12 +4785,19 @@ enum lttng_error_code snapshot_record(struct ltt_session *session,
 				snapshot_ust_consumer_output, session,
 				wait, nb_packets_per_stream);
 		if (ret_code != LTTNG_OK) {
-			goto error;
+			goto error_close_trace_chunk;
 		}
 	}
 
-	if (session_close_trace_chunk(
-			    session, session->current_trace_chunk, NULL, NULL)) {
+error_close_trace_chunk:
+	if (session_set_trace_chunk(session, NULL, &snapshot_trace_chunk)) {
+		ERR("Failed to release the current trace chunk of session \"%s\"",
+				session->name);
+		ret_code = LTTNG_ERR_UNK;
+	}
+
+	if (session_close_trace_chunk(session, snapshot_trace_chunk,
+			LTTNG_TRACE_CHUNK_COMMAND_TYPE_NO_OPERATION, NULL)) {
 		/*
 		 * Don't goto end; make sure the chunk is closed for the session
 		 * to allow future snapshots.
@@ -4685,11 +4806,9 @@ enum lttng_error_code snapshot_record(struct ltt_session *session,
 				session->name);
 		ret_code = LTTNG_ERR_CLOSE_TRACE_CHUNK_FAIL_CONSUMER;
 	}
-	if (session_set_trace_chunk(session, NULL, NULL)) {
-		ERR("Failed to release the current trace chunk of session \"%s\"",
-				session->name);
-		ret_code = LTTNG_ERR_UNK;
-	}
+
+	lttng_trace_chunk_put(snapshot_trace_chunk);
+	snapshot_trace_chunk = NULL;
 error:
 	if (original_ust_consumer_output) {
 		session->ust_session->consumer = original_ust_consumer_output;
@@ -4870,7 +4989,8 @@ int cmd_set_session_shm_path(struct ltt_session *session,
  */
 int cmd_rotate_session(struct ltt_session *session,
 		struct lttng_rotate_session_return *rotate_return,
-		bool quiet_rotation)
+		bool quiet_rotation,
+		enum lttng_trace_chunk_command_type command)
 {
 	int ret;
 	uint64_t ongoing_rotation_chunk_id;
@@ -4908,6 +5028,12 @@ int cmd_rotate_session(struct ltt_session *session,
 		goto end;
 	}
 
+	/* Unsupported feature in lttng-modules before 2.8 (lack of sequence number). */
+	if (session->kernel_session && !kernel_supports_ring_buffer_packet_sequence_number()) {
+		cmd_ret = LTTNG_ERR_ROTATION_NOT_AVAILABLE_KERNEL;
+		goto end;
+	}
+
 	if (session->rotation_state == LTTNG_ROTATION_STATE_ONGOING) {
 		DBG("Refusing to launch a rotation; a rotation is already in progress for session %s",
 				session->name);
@@ -4926,7 +5052,16 @@ int cmd_rotate_session(struct ltt_session *session,
 		goto end;
 	}
 
-	session->rotation_state = LTTNG_ROTATION_STATE_ONGOING;
+	/*
+	 * After a stop followed by a clear, disallow following rotations a they would
+	 * generate empty chunks.
+	 */
+	if (session->cleared_after_last_stop) {
+		DBG("Session \"%s\" was already cleared after stop, refusing rotation",
+				session->name);
+		cmd_ret = LTTNG_ERR_ROTATION_AFTER_STOP_CLEAR;
+		goto end;
+	}
 
 	if (session->active) {
 		new_trace_chunk = session_create_new_trace_chunk(session, NULL,
@@ -4951,11 +5086,6 @@ int cmd_rotate_session(struct ltt_session *session,
 		goto error;
 	}
 
-	assert(chunk_being_archived);
-	chunk_status = lttng_trace_chunk_get_id(chunk_being_archived,
-			&ongoing_rotation_chunk_id);
-	assert(chunk_status == LTTNG_TRACE_CHUNK_STATUS_OK);
-
 	if (session->kernel_session) {
 		cmd_ret = kernel_rotate_session(session);
 		if (cmd_ret != LTTNG_OK) {
@@ -4971,12 +5101,28 @@ int cmd_rotate_session(struct ltt_session *session,
 		}
 	}
 
+	if (!session->active) {
+		session->rotated_after_last_stop = true;
+	}
+
+	if (!chunk_being_archived) {
+		DBG("Rotating session \"%s\" from a \"NULL\" trace chunk to a new trace chunk, skipping completion check",
+				session->name);
+		if (failed_to_rotate) {
+			cmd_ret = rotation_fail_code;
+			goto error;
+		}
+		cmd_ret = LTTNG_OK;
+		goto end;
+	}
+
+	session->rotation_state = LTTNG_ROTATION_STATE_ONGOING;
+	chunk_status = lttng_trace_chunk_get_id(chunk_being_archived,
+			&ongoing_rotation_chunk_id);
+	assert(chunk_status == LTTNG_TRACE_CHUNK_STATUS_OK);
+
 	ret = session_close_trace_chunk(session, chunk_being_archived,
-			quiet_rotation ?
-					NULL :
-					&((enum lttng_trace_chunk_command_type){
-							LTTNG_TRACE_CHUNK_COMMAND_TYPE_MOVE_TO_COMPLETED}),
-			session->last_chunk_path);
+		command, session->last_chunk_path);
 	if (ret) {
 		cmd_ret = LTTNG_ERR_CLOSE_TRACE_CHUNK_FAIL_CONSUMER;
 		goto error;
@@ -4995,10 +5141,6 @@ int cmd_rotate_session(struct ltt_session *session,
 		goto error;
 	}
 
-	if (!session->active) {
-		session->rotated_after_last_stop = true;
-	}
-
 	if (rotate_return) {
 		rotate_return->rotation_id = ongoing_rotation_chunk_id;
 	}