Launch the client management thread using lttng_thread
authorJérémie Galarneau <jeremie.galarneau@efficios.com>
Fri, 30 Nov 2018 19:29:32 +0000 (14:29 -0500)
committerJérémie Galarneau <jeremie.galarneau@efficios.com>
Wed, 5 Dec 2018 17:25:11 +0000 (12:25 -0500)
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
src/bin/lttng-sessiond/Makefile.am
src/bin/lttng-sessiond/client.c [new file with mode: 0644]
src/bin/lttng-sessiond/client.h [new file with mode: 0644]
src/bin/lttng-sessiond/globals.c
src/bin/lttng-sessiond/load-session-thread.c
src/bin/lttng-sessiond/load-session-thread.h
src/bin/lttng-sessiond/lttng-sessiond.h
src/bin/lttng-sessiond/main.c
src/bin/lttng-sessiond/thread-utils.c

index 43962e7dffc69fd8e2896744d9420017f54adc84..fa10b79fa10f89952bb610ab22d8bef4e9636191 100644 (file)
@@ -45,7 +45,8 @@ lttng_sessiond_SOURCES = utils.c utils.h \
                        thread-utils.c \
                        process-utils.c \
                        thread.c thread.h \
-                       health.c
+                       health.c \
+                       client.c client.h
 
 if HAVE_LIBLTTNG_UST_CTL
 lttng_sessiond_SOURCES += trace-ust.c ust-registry.c ust-app.c \
diff --git a/src/bin/lttng-sessiond/client.c b/src/bin/lttng-sessiond/client.c
new file mode 100644 (file)
index 0000000..119062c
--- /dev/null
@@ -0,0 +1,2556 @@
+/*
+ * Copyright (C) 2011 - David Goulet <david.goulet@polymtl.ca>
+ *                      Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
+ *               2013 - 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.
+ *
+ * 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 <stddef.h>
+#include <pthread.h>
+#include <signal.h>
+#include <sys/stat.h>
+#include <common/compat/getenv.h>
+#include <common/unix.h>
+#include <common/utils.h>
+#include <lttng/userspace-probe-internal.h>
+#include <lttng/event-internal.h>
+
+#include "client.h"
+#include "lttng-sessiond.h"
+#include "cmd.h"
+#include "kernel.h"
+#include "save.h"
+#include "health-sessiond.h"
+#include "testpoint.h"
+#include "utils.h"
+
+static bool is_root;
+
+static struct thread_state {
+       pthread_cond_t cond;
+       pthread_mutex_t lock;
+       bool is_running;
+} thread_state = {
+       .cond = PTHREAD_COND_INITIALIZER,
+       .lock = PTHREAD_MUTEX_INITIALIZER,
+       .is_running = false
+};
+
+void set_thread_state_running(void)
+{
+       pthread_mutex_lock(&thread_state.lock);
+        thread_state.is_running = true;
+       pthread_cond_broadcast(&thread_state.cond);
+       pthread_mutex_unlock(&thread_state.lock);
+}
+
+static void wait_thread_state_running(void)
+{
+       pthread_mutex_lock(&thread_state.lock);
+       while (!thread_state.is_running) {
+               pthread_cond_wait(&thread_state.cond,
+                               &thread_state.lock);
+       }
+       pthread_mutex_unlock(&thread_state.lock);
+}
+
+/*
+ * Setup the outgoing data buffer for the response (llm) by allocating the
+ * right amount of memory and copying the original information from the lsm
+ * structure.
+ *
+ * Return 0 on success, negative value on error.
+ */
+static int setup_lttng_msg(struct command_ctx *cmd_ctx,
+       const void *payload_buf, size_t payload_len,
+       const void *cmd_header_buf, size_t cmd_header_len)
+{
+       int ret = 0;
+       const size_t header_len = sizeof(struct lttcomm_lttng_msg);
+       const size_t cmd_header_offset = header_len;
+       const size_t payload_offset = cmd_header_offset + cmd_header_len;
+       const size_t total_msg_size = header_len + cmd_header_len + payload_len;
+
+       cmd_ctx->llm = zmalloc(total_msg_size);
+
+       if (cmd_ctx->llm == NULL) {
+               PERROR("zmalloc");
+               ret = -ENOMEM;
+               goto end;
+       }
+
+       /* Copy common data */
+       cmd_ctx->llm->cmd_type = cmd_ctx->lsm->cmd_type;
+       cmd_ctx->llm->pid = cmd_ctx->lsm->domain.attr.pid;
+       cmd_ctx->llm->cmd_header_size = cmd_header_len;
+       cmd_ctx->llm->data_size = payload_len;
+       cmd_ctx->lttng_msg_size = total_msg_size;
+
+       /* Copy command header */
+       if (cmd_header_len) {
+               memcpy(((uint8_t *) cmd_ctx->llm) + cmd_header_offset, cmd_header_buf,
+                       cmd_header_len);
+       }
+
+       /* Copy payload */
+       if (payload_len) {
+               memcpy(((uint8_t *) cmd_ctx->llm) + payload_offset, payload_buf,
+                       payload_len);
+       }
+
+end:
+       return ret;
+}
+
+/*
+ * Start the thread_manage_consumer. This must be done after a lttng-consumerd
+ * exec or it will fails.
+ */
+static int spawn_consumer_thread(struct consumer_data *consumer_data)
+{
+       int ret, clock_ret;
+       struct timespec timeout;
+
+       /*
+        * Make sure we set the readiness flag to 0 because we are NOT ready.
+        * This access to consumer_thread_is_ready does not need to be
+        * protected by consumer_data.cond_mutex (yet) since the consumer
+        * management thread has not been started at this point.
+        */
+       consumer_data->consumer_thread_is_ready = 0;
+
+       /* Setup pthread condition */
+       ret = pthread_condattr_init(&consumer_data->condattr);
+       if (ret) {
+               errno = ret;
+               PERROR("pthread_condattr_init consumer data");
+               goto error;
+       }
+
+       /*
+        * Set the monotonic clock in order to make sure we DO NOT jump in time
+        * between the clock_gettime() call and the timedwait call. See bug #324
+        * for a more details and how we noticed it.
+        */
+       ret = pthread_condattr_setclock(&consumer_data->condattr, CLOCK_MONOTONIC);
+       if (ret) {
+               errno = ret;
+               PERROR("pthread_condattr_setclock consumer data");
+               goto error;
+       }
+
+       ret = pthread_cond_init(&consumer_data->cond, &consumer_data->condattr);
+       if (ret) {
+               errno = ret;
+               PERROR("pthread_cond_init consumer data");
+               goto error;
+       }
+
+       ret = pthread_create(&consumer_data->thread, default_pthread_attr(),
+                       thread_manage_consumer, consumer_data);
+       if (ret) {
+               errno = ret;
+               PERROR("pthread_create consumer");
+               ret = -1;
+               goto error;
+       }
+
+       /* We are about to wait on a pthread condition */
+       pthread_mutex_lock(&consumer_data->cond_mutex);
+
+       /* Get time for sem_timedwait absolute timeout */
+       clock_ret = lttng_clock_gettime(CLOCK_MONOTONIC, &timeout);
+       /*
+        * Set the timeout for the condition timed wait even if the clock gettime
+        * call fails since we might loop on that call and we want to avoid to
+        * increment the timeout too many times.
+        */
+       timeout.tv_sec += DEFAULT_SEM_WAIT_TIMEOUT;
+
+       /*
+        * The following loop COULD be skipped in some conditions so this is why we
+        * set ret to 0 in order to make sure at least one round of the loop is
+        * done.
+        */
+       ret = 0;
+
+       /*
+        * Loop until the condition is reached or when a timeout is reached. Note
+        * that the pthread_cond_timedwait(P) man page specifies that EINTR can NOT
+        * be returned but the pthread_cond(3), from the glibc-doc, says that it is
+        * possible. This loop does not take any chances and works with both of
+        * them.
+        */
+       while (!consumer_data->consumer_thread_is_ready && ret != ETIMEDOUT) {
+               if (clock_ret < 0) {
+                       PERROR("clock_gettime spawn consumer");
+                       /* Infinite wait for the consumerd thread to be ready */
+                       ret = pthread_cond_wait(&consumer_data->cond,
+                                       &consumer_data->cond_mutex);
+               } else {
+                       ret = pthread_cond_timedwait(&consumer_data->cond,
+                                       &consumer_data->cond_mutex, &timeout);
+               }
+       }
+
+       /* Release the pthread condition */
+       pthread_mutex_unlock(&consumer_data->cond_mutex);
+
+       if (ret != 0) {
+               errno = ret;
+               if (ret == ETIMEDOUT) {
+                       int pth_ret;
+
+                       /*
+                        * Call has timed out so we kill the kconsumerd_thread and return
+                        * an error.
+                        */
+                       ERR("Condition timed out. The consumer thread was never ready."
+                                       " Killing it");
+                       pth_ret = pthread_cancel(consumer_data->thread);
+                       if (pth_ret < 0) {
+                               PERROR("pthread_cancel consumer thread");
+                       }
+               } else {
+                       PERROR("pthread_cond_wait failed consumer thread");
+               }
+               /* Caller is expecting a negative value on failure. */
+               ret = -1;
+               goto error;
+       }
+
+       pthread_mutex_lock(&consumer_data->pid_mutex);
+       if (consumer_data->pid == 0) {
+               ERR("Consumerd did not start");
+               pthread_mutex_unlock(&consumer_data->pid_mutex);
+               goto error;
+       }
+       pthread_mutex_unlock(&consumer_data->pid_mutex);
+
+       return 0;
+
+error:
+       return ret;
+}
+
+/*
+ * Fork and exec a consumer daemon (consumerd).
+ *
+ * Return pid if successful else -1.
+ */
+static pid_t spawn_consumerd(struct consumer_data *consumer_data)
+{
+       int ret;
+       pid_t pid;
+       const char *consumer_to_use;
+       const char *verbosity;
+       struct stat st;
+
+       DBG("Spawning consumerd");
+
+       pid = fork();
+       if (pid == 0) {
+               /*
+                * Exec consumerd.
+                */
+               if (config.verbose_consumer) {
+                       verbosity = "--verbose";
+               } else if (lttng_opt_quiet) {
+                       verbosity = "--quiet";
+               } else {
+                       verbosity = "";
+               }
+
+               switch (consumer_data->type) {
+               case LTTNG_CONSUMER_KERNEL:
+                       /*
+                        * Find out which consumerd to execute. We will first try the
+                        * 64-bit path, then the sessiond's installation directory, and
+                        * fallback on the 32-bit one,
+                        */
+                       DBG3("Looking for a kernel consumer at these locations:");
+                       DBG3("  1) %s", config.consumerd64_bin_path.value ? : "NULL");
+                       DBG3("  2) %s/%s", INSTALL_BIN_PATH, DEFAULT_CONSUMERD_FILE);
+                       DBG3("  3) %s", config.consumerd32_bin_path.value ? : "NULL");
+                       if (stat(config.consumerd64_bin_path.value, &st) == 0) {
+                               DBG3("Found location #1");
+                               consumer_to_use = config.consumerd64_bin_path.value;
+                       } else if (stat(INSTALL_BIN_PATH "/" DEFAULT_CONSUMERD_FILE, &st) == 0) {
+                               DBG3("Found location #2");
+                               consumer_to_use = INSTALL_BIN_PATH "/" DEFAULT_CONSUMERD_FILE;
+                       } else if (config.consumerd32_bin_path.value &&
+                                       stat(config.consumerd32_bin_path.value, &st) == 0) {
+                               DBG3("Found location #3");
+                               consumer_to_use = config.consumerd32_bin_path.value;
+                       } else {
+                               DBG("Could not find any valid consumerd executable");
+                               ret = -EINVAL;
+                               goto error;
+                       }
+                       DBG("Using kernel consumer at: %s",  consumer_to_use);
+                       (void) execl(consumer_to_use,
+                               "lttng-consumerd", verbosity, "-k",
+                               "--consumerd-cmd-sock", consumer_data->cmd_unix_sock_path,
+                               "--consumerd-err-sock", consumer_data->err_unix_sock_path,
+                               "--group", config.tracing_group_name.value,
+                               NULL);
+                       break;
+               case LTTNG_CONSUMER64_UST:
+               {
+                       if (config.consumerd64_lib_dir.value) {
+                               char *tmp;
+                               size_t tmplen;
+                               char *tmpnew;
+
+                               tmp = lttng_secure_getenv("LD_LIBRARY_PATH");
+                               if (!tmp) {
+                                       tmp = "";
+                               }
+                               tmplen = strlen(config.consumerd64_lib_dir.value) + 1 /* : */ + strlen(tmp);
+                               tmpnew = zmalloc(tmplen + 1 /* \0 */);
+                               if (!tmpnew) {
+                                       ret = -ENOMEM;
+                                       goto error;
+                               }
+                               strcat(tmpnew, config.consumerd64_lib_dir.value);
+                               if (tmp[0] != '\0') {
+                                       strcat(tmpnew, ":");
+                                       strcat(tmpnew, tmp);
+                               }
+                               ret = setenv("LD_LIBRARY_PATH", tmpnew, 1);
+                               free(tmpnew);
+                               if (ret) {
+                                       ret = -errno;
+                                       goto error;
+                               }
+                       }
+                       DBG("Using 64-bit UST consumer at: %s",  config.consumerd64_bin_path.value);
+                       (void) execl(config.consumerd64_bin_path.value, "lttng-consumerd", verbosity, "-u",
+                                       "--consumerd-cmd-sock", consumer_data->cmd_unix_sock_path,
+                                       "--consumerd-err-sock", consumer_data->err_unix_sock_path,
+                                       "--group", config.tracing_group_name.value,
+                                       NULL);
+                       break;
+               }
+               case LTTNG_CONSUMER32_UST:
+               {
+                       if (config.consumerd32_lib_dir.value) {
+                               char *tmp;
+                               size_t tmplen;
+                               char *tmpnew;
+
+                               tmp = lttng_secure_getenv("LD_LIBRARY_PATH");
+                               if (!tmp) {
+                                       tmp = "";
+                               }
+                               tmplen = strlen(config.consumerd32_lib_dir.value) + 1 /* : */ + strlen(tmp);
+                               tmpnew = zmalloc(tmplen + 1 /* \0 */);
+                               if (!tmpnew) {
+                                       ret = -ENOMEM;
+                                       goto error;
+                               }
+                               strcat(tmpnew, config.consumerd32_lib_dir.value);
+                               if (tmp[0] != '\0') {
+                                       strcat(tmpnew, ":");
+                                       strcat(tmpnew, tmp);
+                               }
+                               ret = setenv("LD_LIBRARY_PATH", tmpnew, 1);
+                               free(tmpnew);
+                               if (ret) {
+                                       ret = -errno;
+                                       goto error;
+                               }
+                       }
+                       DBG("Using 32-bit UST consumer at: %s",  config.consumerd32_bin_path.value);
+                       (void) execl(config.consumerd32_bin_path.value, "lttng-consumerd", verbosity, "-u",
+                                       "--consumerd-cmd-sock", consumer_data->cmd_unix_sock_path,
+                                       "--consumerd-err-sock", consumer_data->err_unix_sock_path,
+                                       "--group", config.tracing_group_name.value,
+                                       NULL);
+                       break;
+               }
+               default:
+                       ERR("unknown consumer type");
+                       errno = 0;
+               }
+               if (errno != 0) {
+                       PERROR("Consumer execl()");
+               }
+               /* Reaching this point, we got a failure on our execl(). */
+               exit(EXIT_FAILURE);
+       } else if (pid > 0) {
+               ret = pid;
+       } else {
+               PERROR("start consumer fork");
+               ret = -errno;
+       }
+error:
+       return ret;
+}
+
+/*
+ * Spawn the consumerd daemon and session daemon thread.
+ */
+static int start_consumerd(struct consumer_data *consumer_data)
+{
+       int ret;
+
+       /*
+        * Set the listen() state on the socket since there is a possible race
+        * between the exec() of the consumer daemon and this call if place in the
+        * consumer thread. See bug #366 for more details.
+        */
+       ret = lttcomm_listen_unix_sock(consumer_data->err_sock);
+       if (ret < 0) {
+               goto error;
+       }
+
+       pthread_mutex_lock(&consumer_data->pid_mutex);
+       if (consumer_data->pid != 0) {
+               pthread_mutex_unlock(&consumer_data->pid_mutex);
+               goto end;
+       }
+
+       ret = spawn_consumerd(consumer_data);
+       if (ret < 0) {
+               ERR("Spawning consumerd failed");
+               pthread_mutex_unlock(&consumer_data->pid_mutex);
+               goto error;
+       }
+
+       /* Setting up the consumer_data pid */
+       consumer_data->pid = ret;
+       DBG2("Consumer pid %d", consumer_data->pid);
+       pthread_mutex_unlock(&consumer_data->pid_mutex);
+
+       DBG2("Spawning consumer control thread");
+       ret = spawn_consumer_thread(consumer_data);
+       if (ret < 0) {
+               ERR("Fatal error spawning consumer control thread");
+               goto error;
+       }
+
+end:
+       return 0;
+
+error:
+       /* Cleanup already created sockets on error. */
+       if (consumer_data->err_sock >= 0) {
+               int err;
+
+               err = close(consumer_data->err_sock);
+               if (err < 0) {
+                       PERROR("close consumer data error socket");
+               }
+       }
+       return ret;
+}
+
+/*
+ * Copy consumer output from the tracing session to the domain session. The
+ * function also applies the right modification on a per domain basis for the
+ * trace files destination directory.
+ *
+ * Should *NOT* be called with RCU read-side lock held.
+ */
+static int copy_session_consumer(int domain, struct ltt_session *session)
+{
+       int ret;
+       const char *dir_name;
+       struct consumer_output *consumer;
+
+       assert(session);
+       assert(session->consumer);
+
+       switch (domain) {
+       case LTTNG_DOMAIN_KERNEL:
+               DBG3("Copying tracing session consumer output in kernel session");
+               /*
+                * XXX: We should audit the session creation and what this function
+                * does "extra" in order to avoid a destroy since this function is used
+                * in the domain session creation (kernel and ust) only. Same for UST
+                * domain.
+                */
+               if (session->kernel_session->consumer) {
+                       consumer_output_put(session->kernel_session->consumer);
+               }
+               session->kernel_session->consumer =
+                       consumer_copy_output(session->consumer);
+               /* Ease our life a bit for the next part */
+               consumer = session->kernel_session->consumer;
+               dir_name = DEFAULT_KERNEL_TRACE_DIR;
+               break;
+       case LTTNG_DOMAIN_JUL:
+       case LTTNG_DOMAIN_LOG4J:
+       case LTTNG_DOMAIN_PYTHON:
+       case LTTNG_DOMAIN_UST:
+               DBG3("Copying tracing session consumer output in UST session");
+               if (session->ust_session->consumer) {
+                       consumer_output_put(session->ust_session->consumer);
+               }
+               session->ust_session->consumer =
+                       consumer_copy_output(session->consumer);
+               /* Ease our life a bit for the next part */
+               consumer = session->ust_session->consumer;
+               dir_name = DEFAULT_UST_TRACE_DIR;
+               break;
+       default:
+               ret = LTTNG_ERR_UNKNOWN_DOMAIN;
+               goto error;
+       }
+
+       /* Append correct directory to subdir */
+       strncat(consumer->subdir, dir_name,
+                       sizeof(consumer->subdir) - strlen(consumer->subdir) - 1);
+       DBG3("Copy session consumer subdir %s", consumer->subdir);
+
+       ret = LTTNG_OK;
+
+error:
+       return ret;
+}
+
+/*
+ * Create an UST session and add it to the session ust list.
+ *
+ * Should *NOT* be called with RCU read-side lock held.
+ */
+static int create_ust_session(struct ltt_session *session,
+               struct lttng_domain *domain)
+{
+       int ret;
+       struct ltt_ust_session *lus = NULL;
+
+       assert(session);
+       assert(domain);
+       assert(session->consumer);
+
+       switch (domain->type) {
+       case LTTNG_DOMAIN_JUL:
+       case LTTNG_DOMAIN_LOG4J:
+       case LTTNG_DOMAIN_PYTHON:
+       case LTTNG_DOMAIN_UST:
+               break;
+       default:
+               ERR("Unknown UST domain on create session %d", domain->type);
+               ret = LTTNG_ERR_UNKNOWN_DOMAIN;
+               goto error;
+       }
+
+       DBG("Creating UST session");
+
+       lus = trace_ust_create_session(session->id);
+       if (lus == NULL) {
+               ret = LTTNG_ERR_UST_SESS_FAIL;
+               goto error;
+       }
+
+       lus->uid = session->uid;
+       lus->gid = session->gid;
+       lus->output_traces = session->output_traces;
+       lus->snapshot_mode = session->snapshot_mode;
+       lus->live_timer_interval = session->live_timer;
+       session->ust_session = lus;
+       if (session->shm_path[0]) {
+               strncpy(lus->root_shm_path, session->shm_path,
+                       sizeof(lus->root_shm_path));
+               lus->root_shm_path[sizeof(lus->root_shm_path) - 1] = '\0';
+               strncpy(lus->shm_path, session->shm_path,
+                       sizeof(lus->shm_path));
+               lus->shm_path[sizeof(lus->shm_path) - 1] = '\0';
+               strncat(lus->shm_path, "/ust",
+                       sizeof(lus->shm_path) - strlen(lus->shm_path) - 1);
+       }
+       /* Copy session output to the newly created UST session */
+       ret = copy_session_consumer(domain->type, session);
+       if (ret != LTTNG_OK) {
+               goto error;
+       }
+
+       return LTTNG_OK;
+
+error:
+       free(lus);
+       session->ust_session = NULL;
+       return ret;
+}
+
+/*
+ * Create a kernel tracer session then create the default channel.
+ */
+static int create_kernel_session(struct ltt_session *session)
+{
+       int ret;
+
+       DBG("Creating kernel session");
+
+       ret = kernel_create_session(session, kernel_tracer_fd);
+       if (ret < 0) {
+               ret = LTTNG_ERR_KERN_SESS_FAIL;
+               goto error;
+       }
+
+       /* Code flow safety */
+       assert(session->kernel_session);
+
+       /* Copy session output to the newly created Kernel session */
+       ret = copy_session_consumer(LTTNG_DOMAIN_KERNEL, session);
+       if (ret != LTTNG_OK) {
+               goto error;
+       }
+
+       session->kernel_session->uid = session->uid;
+       session->kernel_session->gid = session->gid;
+       session->kernel_session->output_traces = session->output_traces;
+       session->kernel_session->snapshot_mode = session->snapshot_mode;
+
+       return LTTNG_OK;
+
+error:
+       trace_kernel_destroy_session(session->kernel_session);
+       session->kernel_session = NULL;
+       return ret;
+}
+
+/*
+ * Count number of session permitted by uid/gid.
+ */
+static unsigned int lttng_sessions_count(uid_t uid, gid_t gid)
+{
+       unsigned int i = 0;
+       struct ltt_session *session;
+       const struct ltt_session_list *session_list = session_get_list();
+
+       DBG("Counting number of available session for UID %d GID %d",
+                       uid, gid);
+       cds_list_for_each_entry(session, &session_list->head, list) {
+               if (!session_get(session)) {
+                       continue;
+               }
+               session_lock(session);
+               /* Only count the sessions the user can control. */
+               if (session_access_ok(session, uid, gid) &&
+                               !session->destroyed) {
+                       i++;
+               }
+               session_unlock(session);
+               session_put(session);
+       }
+       return i;
+}
+
+static int receive_userspace_probe(struct command_ctx *cmd_ctx, int sock,
+               int *sock_error, struct lttng_event *event)
+{
+       int fd, ret;
+       struct lttng_userspace_probe_location *probe_location;
+       const struct lttng_userspace_probe_location_lookup_method *lookup = NULL;
+       struct lttng_dynamic_buffer probe_location_buffer;
+       struct lttng_buffer_view buffer_view;
+
+       /*
+        * Create a buffer to store the serialized version of the probe
+        * location.
+        */
+       lttng_dynamic_buffer_init(&probe_location_buffer);
+       ret = lttng_dynamic_buffer_set_size(&probe_location_buffer,
+                       cmd_ctx->lsm->u.enable.userspace_probe_location_len);
+       if (ret) {
+               ret = LTTNG_ERR_NOMEM;
+               goto error;
+       }
+
+       /*
+        * Receive the probe location.
+        */
+       ret = lttcomm_recv_unix_sock(sock, probe_location_buffer.data,
+                       probe_location_buffer.size);
+       if (ret <= 0) {
+               DBG("Nothing recv() from client var len data... continuing");
+               *sock_error = 1;
+               lttng_dynamic_buffer_reset(&probe_location_buffer);
+               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
+               goto error;
+       }
+
+       buffer_view = lttng_buffer_view_from_dynamic_buffer(
+                       &probe_location_buffer, 0, probe_location_buffer.size);
+
+       /*
+        * Extract the probe location from the serialized version.
+        */
+       ret = lttng_userspace_probe_location_create_from_buffer(
+                               &buffer_view, &probe_location);
+       if (ret < 0) {
+               WARN("Failed to create a userspace probe location from the received buffer");
+               lttng_dynamic_buffer_reset( &probe_location_buffer);
+               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
+               goto error;
+       }
+
+       /*
+        * Receive the file descriptor to the target binary from the client.
+        */
+       DBG("Receiving userspace probe target FD from client ...");
+       ret = lttcomm_recv_fds_unix_sock(sock, &fd, 1);
+       if (ret <= 0) {
+               DBG("Nothing recv() from client userspace probe fd... continuing");
+               *sock_error = 1;
+               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
+               goto error;
+       }
+
+       /*
+        * Set the file descriptor received from the client through the unix
+        * socket in the probe location.
+        */
+       lookup = lttng_userspace_probe_location_get_lookup_method(probe_location);
+       if (!lookup) {
+               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
+               goto error;
+       }
+
+       /*
+        * From the kernel tracer's perspective, all userspace probe event types
+        * are all the same: a file and an offset.
+        */
+       switch (lttng_userspace_probe_location_lookup_method_get_type(lookup)) {
+       case LTTNG_USERSPACE_PROBE_LOCATION_LOOKUP_METHOD_TYPE_FUNCTION_ELF:
+               ret = lttng_userspace_probe_location_function_set_binary_fd(
+                               probe_location, fd);
+               break;
+       case LTTNG_USERSPACE_PROBE_LOCATION_LOOKUP_METHOD_TYPE_TRACEPOINT_SDT:
+               ret = lttng_userspace_probe_location_tracepoint_set_binary_fd(
+                               probe_location, fd);
+               break;
+       default:
+               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
+               goto error;
+       }
+
+       if (ret) {
+               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
+               goto error;
+       }
+
+       /* Attach the probe location to the event. */
+       ret = lttng_event_set_userspace_probe_location(event, probe_location);
+       if (ret) {
+               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
+               goto error;
+       }
+
+       lttng_dynamic_buffer_reset(&probe_location_buffer);
+error:
+       return ret;
+}
+
+/*
+ * Join consumer thread
+ */
+static int join_consumer_thread(struct consumer_data *consumer_data)
+{
+       void *status;
+
+       /* Consumer pid must be a real one. */
+       if (consumer_data->pid > 0) {
+               int ret;
+               ret = kill(consumer_data->pid, SIGTERM);
+               if (ret) {
+                       PERROR("Error killing consumer daemon");
+                       return ret;
+               }
+               return pthread_join(consumer_data->thread, &status);
+       } else {
+               return 0;
+       }
+}
+
+/*
+ * Version of setup_lttng_msg() without command header.
+ */
+static int setup_lttng_msg_no_cmd_header(struct command_ctx *cmd_ctx,
+       void *payload_buf, size_t payload_len)
+{
+       return setup_lttng_msg(cmd_ctx, payload_buf, payload_len, NULL, 0);
+}
+
+/*
+ * Free memory of a command context structure.
+ */
+static void clean_command_ctx(struct command_ctx **cmd_ctx)
+{
+       DBG("Clean command context structure");
+       if (*cmd_ctx) {
+               if ((*cmd_ctx)->llm) {
+                       free((*cmd_ctx)->llm);
+               }
+               if ((*cmd_ctx)->lsm) {
+                       free((*cmd_ctx)->lsm);
+               }
+               free(*cmd_ctx);
+               *cmd_ctx = NULL;
+       }
+}
+
+/*
+ * Check if the current kernel tracer supports the session rotation feature.
+ * Return 1 if it does, 0 otherwise.
+ */
+static int check_rotate_compatible(void)
+{
+       int ret = 1;
+
+       if (kernel_tracer_version.major != 2 || kernel_tracer_version.minor < 11) {
+               DBG("Kernel tracer version is not compatible with the rotation feature");
+               ret = 0;
+       }
+
+       return ret;
+}
+
+/*
+ * Send data on a unix socket using the liblttsessiondcomm API.
+ *
+ * Return lttcomm error code.
+ */
+static int send_unix_sock(int sock, void *buf, size_t len)
+{
+       /* Check valid length */
+       if (len == 0) {
+               return -1;
+       }
+
+       return lttcomm_send_unix_sock(sock, buf, len);
+}
+
+/*
+ * Process the command requested by the lttng client within the command
+ * context structure. This function make sure that the return structure (llm)
+ * is set and ready for transmission before returning.
+ *
+ * Return any error encountered or 0 for success.
+ *
+ * "sock" is only used for special-case var. len data.
+ *
+ * Should *NOT* be called with RCU read-side lock held.
+ */
+static int process_client_msg(struct command_ctx *cmd_ctx, int sock,
+               int *sock_error)
+{
+       int ret = LTTNG_OK;
+       int need_tracing_session = 1;
+       int need_domain;
+
+       DBG("Processing client command %d", cmd_ctx->lsm->cmd_type);
+
+       assert(!rcu_read_ongoing());
+
+       *sock_error = 0;
+
+       switch (cmd_ctx->lsm->cmd_type) {
+       case LTTNG_CREATE_SESSION:
+       case LTTNG_CREATE_SESSION_SNAPSHOT:
+       case LTTNG_CREATE_SESSION_LIVE:
+       case LTTNG_DESTROY_SESSION:
+       case LTTNG_LIST_SESSIONS:
+       case LTTNG_LIST_DOMAINS:
+       case LTTNG_START_TRACE:
+       case LTTNG_STOP_TRACE:
+       case LTTNG_DATA_PENDING:
+       case LTTNG_SNAPSHOT_ADD_OUTPUT:
+       case LTTNG_SNAPSHOT_DEL_OUTPUT:
+       case LTTNG_SNAPSHOT_LIST_OUTPUT:
+       case LTTNG_SNAPSHOT_RECORD:
+       case LTTNG_SAVE_SESSION:
+       case LTTNG_SET_SESSION_SHM_PATH:
+       case LTTNG_REGENERATE_METADATA:
+       case LTTNG_REGENERATE_STATEDUMP:
+       case LTTNG_REGISTER_TRIGGER:
+       case LTTNG_UNREGISTER_TRIGGER:
+       case LTTNG_ROTATE_SESSION:
+       case LTTNG_ROTATION_GET_INFO:
+       case LTTNG_ROTATION_SET_SCHEDULE:
+       case LTTNG_SESSION_LIST_ROTATION_SCHEDULES:
+               need_domain = 0;
+               break;
+       default:
+               need_domain = 1;
+       }
+
+       if (config.no_kernel && need_domain
+                       && cmd_ctx->lsm->domain.type == LTTNG_DOMAIN_KERNEL) {
+               if (!is_root) {
+                       ret = LTTNG_ERR_NEED_ROOT_SESSIOND;
+               } else {
+                       ret = LTTNG_ERR_KERN_NA;
+               }
+               goto error;
+       }
+
+       /* Deny register consumer if we already have a spawned consumer. */
+       if (cmd_ctx->lsm->cmd_type == LTTNG_REGISTER_CONSUMER) {
+               pthread_mutex_lock(&kconsumer_data.pid_mutex);
+               if (kconsumer_data.pid > 0) {
+                       ret = LTTNG_ERR_KERN_CONSUMER_FAIL;
+                       pthread_mutex_unlock(&kconsumer_data.pid_mutex);
+                       goto error;
+               }
+               pthread_mutex_unlock(&kconsumer_data.pid_mutex);
+       }
+
+       /*
+        * Check for command that don't needs to allocate a returned payload. We do
+        * this here so we don't have to make the call for no payload at each
+        * command.
+        */
+       switch(cmd_ctx->lsm->cmd_type) {
+       case LTTNG_LIST_SESSIONS:
+       case LTTNG_LIST_TRACEPOINTS:
+       case LTTNG_LIST_TRACEPOINT_FIELDS:
+       case LTTNG_LIST_DOMAINS:
+       case LTTNG_LIST_CHANNELS:
+       case LTTNG_LIST_EVENTS:
+       case LTTNG_LIST_SYSCALLS:
+       case LTTNG_LIST_TRACKER_PIDS:
+       case LTTNG_DATA_PENDING:
+       case LTTNG_ROTATE_SESSION:
+       case LTTNG_ROTATION_GET_INFO:
+       case LTTNG_SESSION_LIST_ROTATION_SCHEDULES:
+               break;
+       default:
+               /* Setup lttng message with no payload */
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, NULL, 0);
+               if (ret < 0) {
+                       /* This label does not try to unlock the session */
+                       goto init_setup_error;
+               }
+       }
+
+       /* Commands that DO NOT need a session. */
+       switch (cmd_ctx->lsm->cmd_type) {
+       case LTTNG_CREATE_SESSION:
+       case LTTNG_CREATE_SESSION_SNAPSHOT:
+       case LTTNG_CREATE_SESSION_LIVE:
+       case LTTNG_LIST_SESSIONS:
+       case LTTNG_LIST_TRACEPOINTS:
+       case LTTNG_LIST_SYSCALLS:
+       case LTTNG_LIST_TRACEPOINT_FIELDS:
+       case LTTNG_SAVE_SESSION:
+       case LTTNG_REGISTER_TRIGGER:
+       case LTTNG_UNREGISTER_TRIGGER:
+               need_tracing_session = 0;
+               break;
+       default:
+               DBG("Getting session %s by name", cmd_ctx->lsm->session.name);
+               /*
+                * We keep the session list lock across _all_ commands
+                * for now, because the per-session lock does not
+                * handle teardown properly.
+                */
+               session_lock_list();
+               cmd_ctx->session = session_find_by_name(cmd_ctx->lsm->session.name);
+               if (cmd_ctx->session == NULL) {
+                       ret = LTTNG_ERR_SESS_NOT_FOUND;
+                       goto error;
+               } else {
+                       /* Acquire lock for the session */
+                       session_lock(cmd_ctx->session);
+               }
+               break;
+       }
+
+       /*
+        * Commands that need a valid session but should NOT create one if none
+        * exists. Instead of creating one and destroying it when the command is
+        * handled, process that right before so we save some round trip in useless
+        * code path.
+        */
+       switch (cmd_ctx->lsm->cmd_type) {
+       case LTTNG_DISABLE_CHANNEL:
+       case LTTNG_DISABLE_EVENT:
+               switch (cmd_ctx->lsm->domain.type) {
+               case LTTNG_DOMAIN_KERNEL:
+                       if (!cmd_ctx->session->kernel_session) {
+                               ret = LTTNG_ERR_NO_CHANNEL;
+                               goto error;
+                       }
+                       break;
+               case LTTNG_DOMAIN_JUL:
+               case LTTNG_DOMAIN_LOG4J:
+               case LTTNG_DOMAIN_PYTHON:
+               case LTTNG_DOMAIN_UST:
+                       if (!cmd_ctx->session->ust_session) {
+                               ret = LTTNG_ERR_NO_CHANNEL;
+                               goto error;
+                       }
+                       break;
+               default:
+                       ret = LTTNG_ERR_UNKNOWN_DOMAIN;
+                       goto error;
+               }
+       default:
+               break;
+       }
+
+       if (!need_domain) {
+               goto skip_domain;
+       }
+
+       /*
+        * Check domain type for specific "pre-action".
+        */
+       switch (cmd_ctx->lsm->domain.type) {
+       case LTTNG_DOMAIN_KERNEL:
+               if (!is_root) {
+                       ret = LTTNG_ERR_NEED_ROOT_SESSIOND;
+                       goto error;
+               }
+
+               /* Consumer is in an ERROR state. Report back to client */
+               if (uatomic_read(&kernel_consumerd_state) == CONSUMER_ERROR) {
+                       ret = LTTNG_ERR_NO_KERNCONSUMERD;
+                       goto error;
+               }
+
+               /* Need a session for kernel command */
+               if (need_tracing_session) {
+                       if (cmd_ctx->session->kernel_session == NULL) {
+                               ret = create_kernel_session(cmd_ctx->session);
+                               if (ret < 0) {
+                                       ret = LTTNG_ERR_KERN_SESS_FAIL;
+                                       goto error;
+                               }
+                       }
+
+                       /* Start the kernel consumer daemon */
+                       pthread_mutex_lock(&kconsumer_data.pid_mutex);
+                       if (kconsumer_data.pid == 0 &&
+                                       cmd_ctx->lsm->cmd_type != LTTNG_REGISTER_CONSUMER) {
+                               pthread_mutex_unlock(&kconsumer_data.pid_mutex);
+                               ret = start_consumerd(&kconsumer_data);
+                               if (ret < 0) {
+                                       ret = LTTNG_ERR_KERN_CONSUMER_FAIL;
+                                       goto error;
+                               }
+                               uatomic_set(&kernel_consumerd_state, CONSUMER_STARTED);
+                       } else {
+                               pthread_mutex_unlock(&kconsumer_data.pid_mutex);
+                       }
+
+                       /*
+                        * The consumer was just spawned so we need to add the socket to
+                        * the consumer output of the session if exist.
+                        */
+                       ret = consumer_create_socket(&kconsumer_data,
+                                       cmd_ctx->session->kernel_session->consumer);
+                       if (ret < 0) {
+                               goto error;
+                       }
+               }
+
+               break;
+       case LTTNG_DOMAIN_JUL:
+       case LTTNG_DOMAIN_LOG4J:
+       case LTTNG_DOMAIN_PYTHON:
+       case LTTNG_DOMAIN_UST:
+       {
+               if (!ust_app_supported()) {
+                       ret = LTTNG_ERR_NO_UST;
+                       goto error;
+               }
+               /* Consumer is in an ERROR state. Report back to client */
+               if (uatomic_read(&ust_consumerd_state) == CONSUMER_ERROR) {
+                       ret = LTTNG_ERR_NO_USTCONSUMERD;
+                       goto error;
+               }
+
+               if (need_tracing_session) {
+                       /* Create UST session if none exist. */
+                       if (cmd_ctx->session->ust_session == NULL) {
+                               ret = create_ust_session(cmd_ctx->session,
+                                               &cmd_ctx->lsm->domain);
+                               if (ret != LTTNG_OK) {
+                                       goto error;
+                               }
+                       }
+
+                       /* Start the UST consumer daemons */
+                       /* 64-bit */
+                       pthread_mutex_lock(&ustconsumer64_data.pid_mutex);
+                       if (config.consumerd64_bin_path.value &&
+                                       ustconsumer64_data.pid == 0 &&
+                                       cmd_ctx->lsm->cmd_type != LTTNG_REGISTER_CONSUMER) {
+                               pthread_mutex_unlock(&ustconsumer64_data.pid_mutex);
+                               ret = start_consumerd(&ustconsumer64_data);
+                               if (ret < 0) {
+                                       ret = LTTNG_ERR_UST_CONSUMER64_FAIL;
+                                       uatomic_set(&ust_consumerd64_fd, -EINVAL);
+                                       goto error;
+                               }
+
+                               uatomic_set(&ust_consumerd64_fd, ustconsumer64_data.cmd_sock);
+                               uatomic_set(&ust_consumerd_state, CONSUMER_STARTED);
+                       } else {
+                               pthread_mutex_unlock(&ustconsumer64_data.pid_mutex);
+                       }
+
+                       /*
+                        * Setup socket for consumer 64 bit. No need for atomic access
+                        * since it was set above and can ONLY be set in this thread.
+                        */
+                       ret = consumer_create_socket(&ustconsumer64_data,
+                                       cmd_ctx->session->ust_session->consumer);
+                       if (ret < 0) {
+                               goto error;
+                       }
+
+                       /* 32-bit */
+                       pthread_mutex_lock(&ustconsumer32_data.pid_mutex);
+                       if (config.consumerd32_bin_path.value &&
+                                       ustconsumer32_data.pid == 0 &&
+                                       cmd_ctx->lsm->cmd_type != LTTNG_REGISTER_CONSUMER) {
+                               pthread_mutex_unlock(&ustconsumer32_data.pid_mutex);
+                               ret = start_consumerd(&ustconsumer32_data);
+                               if (ret < 0) {
+                                       ret = LTTNG_ERR_UST_CONSUMER32_FAIL;
+                                       uatomic_set(&ust_consumerd32_fd, -EINVAL);
+                                       goto error;
+                               }
+
+                               uatomic_set(&ust_consumerd32_fd, ustconsumer32_data.cmd_sock);
+                               uatomic_set(&ust_consumerd_state, CONSUMER_STARTED);
+                       } else {
+                               pthread_mutex_unlock(&ustconsumer32_data.pid_mutex);
+                       }
+
+                       /*
+                        * Setup socket for consumer 32 bit. No need for atomic access
+                        * since it was set above and can ONLY be set in this thread.
+                        */
+                       ret = consumer_create_socket(&ustconsumer32_data,
+                                       cmd_ctx->session->ust_session->consumer);
+                       if (ret < 0) {
+                               goto error;
+                       }
+               }
+               break;
+       }
+       default:
+               break;
+       }
+skip_domain:
+
+       /* Validate consumer daemon state when start/stop trace command */
+       if (cmd_ctx->lsm->cmd_type == LTTNG_START_TRACE ||
+                       cmd_ctx->lsm->cmd_type == LTTNG_STOP_TRACE) {
+               switch (cmd_ctx->lsm->domain.type) {
+               case LTTNG_DOMAIN_NONE:
+                       break;
+               case LTTNG_DOMAIN_JUL:
+               case LTTNG_DOMAIN_LOG4J:
+               case LTTNG_DOMAIN_PYTHON:
+               case LTTNG_DOMAIN_UST:
+                       if (uatomic_read(&ust_consumerd_state) != CONSUMER_STARTED) {
+                               ret = LTTNG_ERR_NO_USTCONSUMERD;
+                               goto error;
+                       }
+                       break;
+               case LTTNG_DOMAIN_KERNEL:
+                       if (uatomic_read(&kernel_consumerd_state) != CONSUMER_STARTED) {
+                               ret = LTTNG_ERR_NO_KERNCONSUMERD;
+                               goto error;
+                       }
+                       break;
+               default:
+                       ret = LTTNG_ERR_UNKNOWN_DOMAIN;
+                       goto error;
+               }
+       }
+
+       /*
+        * Check that the UID or GID match that of the tracing session.
+        * The root user can interact with all sessions.
+        */
+       if (need_tracing_session) {
+               if (!session_access_ok(cmd_ctx->session,
+                               LTTNG_SOCK_GET_UID_CRED(&cmd_ctx->creds),
+                               LTTNG_SOCK_GET_GID_CRED(&cmd_ctx->creds)) ||
+                               cmd_ctx->session->destroyed) {
+                       ret = LTTNG_ERR_EPERM;
+                       goto error;
+               }
+       }
+
+       /*
+        * Send relayd information to consumer as soon as we have a domain and a
+        * session defined.
+        */
+       if (cmd_ctx->session && need_domain) {
+               /*
+                * Setup relayd if not done yet. If the relayd information was already
+                * sent to the consumer, this call will gracefully return.
+                */
+               ret = cmd_setup_relayd(cmd_ctx->session);
+               if (ret != LTTNG_OK) {
+                       goto error;
+               }
+       }
+
+       /* Process by command type */
+       switch (cmd_ctx->lsm->cmd_type) {
+       case LTTNG_ADD_CONTEXT:
+       {
+               /*
+                * An LTTNG_ADD_CONTEXT command might have a supplementary
+                * payload if the context being added is an application context.
+                */
+               if (cmd_ctx->lsm->u.context.ctx.ctx ==
+                               LTTNG_EVENT_CONTEXT_APP_CONTEXT) {
+                       char *provider_name = NULL, *context_name = NULL;
+                       size_t provider_name_len =
+                                       cmd_ctx->lsm->u.context.provider_name_len;
+                       size_t context_name_len =
+                                       cmd_ctx->lsm->u.context.context_name_len;
+
+                       if (provider_name_len == 0 || context_name_len == 0) {
+                               /*
+                                * Application provider and context names MUST
+                                * be provided.
+                                */
+                               ret = -LTTNG_ERR_INVALID;
+                               goto error;
+                       }
+
+                       provider_name = zmalloc(provider_name_len + 1);
+                       if (!provider_name) {
+                               ret = -LTTNG_ERR_NOMEM;
+                               goto error;
+                       }
+                       cmd_ctx->lsm->u.context.ctx.u.app_ctx.provider_name =
+                                       provider_name;
+
+                       context_name = zmalloc(context_name_len + 1);
+                       if (!context_name) {
+                               ret = -LTTNG_ERR_NOMEM;
+                               goto error_add_context;
+                       }
+                       cmd_ctx->lsm->u.context.ctx.u.app_ctx.ctx_name =
+                                       context_name;
+
+                       ret = lttcomm_recv_unix_sock(sock, provider_name,
+                                       provider_name_len);
+                       if (ret < 0) {
+                               goto error_add_context;
+                       }
+
+                       ret = lttcomm_recv_unix_sock(sock, context_name,
+                                       context_name_len);
+                       if (ret < 0) {
+                               goto error_add_context;
+                       }
+               }
+
+               /*
+                * cmd_add_context assumes ownership of the provider and context
+                * names.
+                */
+               ret = cmd_add_context(cmd_ctx->session,
+                               cmd_ctx->lsm->domain.type,
+                               cmd_ctx->lsm->u.context.channel_name,
+                               &cmd_ctx->lsm->u.context.ctx,
+                               kernel_poll_pipe[1]);
+
+               cmd_ctx->lsm->u.context.ctx.u.app_ctx.provider_name = NULL;
+               cmd_ctx->lsm->u.context.ctx.u.app_ctx.ctx_name = NULL;
+error_add_context:
+               free(cmd_ctx->lsm->u.context.ctx.u.app_ctx.provider_name);
+               free(cmd_ctx->lsm->u.context.ctx.u.app_ctx.ctx_name);
+               if (ret < 0) {
+                       goto error;
+               }
+               break;
+       }
+       case LTTNG_DISABLE_CHANNEL:
+       {
+               ret = cmd_disable_channel(cmd_ctx->session, cmd_ctx->lsm->domain.type,
+                               cmd_ctx->lsm->u.disable.channel_name);
+               break;
+       }
+       case LTTNG_DISABLE_EVENT:
+       {
+
+               /*
+                * FIXME: handle filter; for now we just receive the filter's
+                * bytecode along with the filter expression which are sent by
+                * liblttng-ctl and discard them.
+                *
+                * This fixes an issue where the client may block while sending
+                * the filter payload and encounter an error because the session
+                * daemon closes the socket without ever handling this data.
+                */
+               size_t count = cmd_ctx->lsm->u.disable.expression_len +
+                       cmd_ctx->lsm->u.disable.bytecode_len;
+
+               if (count) {
+                       char data[LTTNG_FILTER_MAX_LEN];
+
+                       DBG("Discarding disable event command payload of size %zu", count);
+                       while (count) {
+                               ret = lttcomm_recv_unix_sock(sock, data,
+                                       count > sizeof(data) ? sizeof(data) : count);
+                               if (ret < 0) {
+                                       goto error;
+                               }
+
+                               count -= (size_t) ret;
+                       }
+               }
+               /* FIXME: passing packed structure to non-packed pointer */
+               ret = cmd_disable_event(cmd_ctx->session, cmd_ctx->lsm->domain.type,
+                               cmd_ctx->lsm->u.disable.channel_name,
+                               &cmd_ctx->lsm->u.disable.event);
+               break;
+       }
+       case LTTNG_ENABLE_CHANNEL:
+       {
+               cmd_ctx->lsm->u.channel.chan.attr.extended.ptr =
+                               (struct lttng_channel_extended *) &cmd_ctx->lsm->u.channel.extended;
+               ret = cmd_enable_channel(cmd_ctx->session, &cmd_ctx->lsm->domain,
+                               &cmd_ctx->lsm->u.channel.chan,
+                               kernel_poll_pipe[1]);
+               break;
+       }
+       case LTTNG_TRACK_PID:
+       {
+               ret = cmd_track_pid(cmd_ctx->session,
+                               cmd_ctx->lsm->domain.type,
+                               cmd_ctx->lsm->u.pid_tracker.pid);
+               break;
+       }
+       case LTTNG_UNTRACK_PID:
+       {
+               ret = cmd_untrack_pid(cmd_ctx->session,
+                               cmd_ctx->lsm->domain.type,
+                               cmd_ctx->lsm->u.pid_tracker.pid);
+               break;
+       }
+       case LTTNG_ENABLE_EVENT:
+       {
+               struct lttng_event *ev = NULL;
+               struct lttng_event_exclusion *exclusion = NULL;
+               struct lttng_filter_bytecode *bytecode = NULL;
+               char *filter_expression = NULL;
+
+               /* Handle exclusion events and receive it from the client. */
+               if (cmd_ctx->lsm->u.enable.exclusion_count > 0) {
+                       size_t count = cmd_ctx->lsm->u.enable.exclusion_count;
+
+                       exclusion = zmalloc(sizeof(struct lttng_event_exclusion) +
+                                       (count * LTTNG_SYMBOL_NAME_LEN));
+                       if (!exclusion) {
+                               ret = LTTNG_ERR_EXCLUSION_NOMEM;
+                               goto error;
+                       }
+
+                       DBG("Receiving var len exclusion event list from client ...");
+                       exclusion->count = count;
+                       ret = lttcomm_recv_unix_sock(sock, exclusion->names,
+                                       count * LTTNG_SYMBOL_NAME_LEN);
+                       if (ret <= 0) {
+                               DBG("Nothing recv() from client var len data... continuing");
+                               *sock_error = 1;
+                               free(exclusion);
+                               ret = LTTNG_ERR_EXCLUSION_INVAL;
+                               goto error;
+                       }
+               }
+
+               /* Get filter expression from client. */
+               if (cmd_ctx->lsm->u.enable.expression_len > 0) {
+                       size_t expression_len =
+                               cmd_ctx->lsm->u.enable.expression_len;
+
+                       if (expression_len > LTTNG_FILTER_MAX_LEN) {
+                               ret = LTTNG_ERR_FILTER_INVAL;
+                               free(exclusion);
+                               goto error;
+                       }
+
+                       filter_expression = zmalloc(expression_len);
+                       if (!filter_expression) {
+                               free(exclusion);
+                               ret = LTTNG_ERR_FILTER_NOMEM;
+                               goto error;
+                       }
+
+                       /* Receive var. len. data */
+                       DBG("Receiving var len filter's expression from client ...");
+                       ret = lttcomm_recv_unix_sock(sock, filter_expression,
+                               expression_len);
+                       if (ret <= 0) {
+                               DBG("Nothing recv() from client var len data... continuing");
+                               *sock_error = 1;
+                               free(filter_expression);
+                               free(exclusion);
+                               ret = LTTNG_ERR_FILTER_INVAL;
+                               goto error;
+                       }
+               }
+
+               /* Handle filter and get bytecode from client. */
+               if (cmd_ctx->lsm->u.enable.bytecode_len > 0) {
+                       size_t bytecode_len = cmd_ctx->lsm->u.enable.bytecode_len;
+
+                       if (bytecode_len > LTTNG_FILTER_MAX_LEN) {
+                               ret = LTTNG_ERR_FILTER_INVAL;
+                               free(filter_expression);
+                               free(exclusion);
+                               goto error;
+                       }
+
+                       bytecode = zmalloc(bytecode_len);
+                       if (!bytecode) {
+                               free(filter_expression);
+                               free(exclusion);
+                               ret = LTTNG_ERR_FILTER_NOMEM;
+                               goto error;
+                       }
+
+                       /* Receive var. len. data */
+                       DBG("Receiving var len filter's bytecode from client ...");
+                       ret = lttcomm_recv_unix_sock(sock, bytecode, bytecode_len);
+                       if (ret <= 0) {
+                               DBG("Nothing recv() from client var len data... continuing");
+                               *sock_error = 1;
+                               free(filter_expression);
+                               free(bytecode);
+                               free(exclusion);
+                               ret = LTTNG_ERR_FILTER_INVAL;
+                               goto error;
+                       }
+
+                       if ((bytecode->len + sizeof(*bytecode)) != bytecode_len) {
+                               free(filter_expression);
+                               free(bytecode);
+                               free(exclusion);
+                               ret = LTTNG_ERR_FILTER_INVAL;
+                               goto error;
+                       }
+               }
+
+               ev = lttng_event_copy(&cmd_ctx->lsm->u.enable.event);
+               if (!ev) {
+                       DBG("Failed to copy event: %s",
+                                       cmd_ctx->lsm->u.enable.event.name);
+                       free(filter_expression);
+                       free(bytecode);
+                       free(exclusion);
+                       ret = LTTNG_ERR_NOMEM;
+                       goto error;
+               }
+
+
+               if (cmd_ctx->lsm->u.enable.userspace_probe_location_len > 0) {
+                       /* Expect a userspace probe description. */
+                       ret = receive_userspace_probe(cmd_ctx, sock, sock_error, ev);
+                       if (ret) {
+                               free(filter_expression);
+                               free(bytecode);
+                               free(exclusion);
+                               lttng_event_destroy(ev);
+                               goto error;
+                       }
+               }
+
+               ret = cmd_enable_event(cmd_ctx->session, &cmd_ctx->lsm->domain,
+                               cmd_ctx->lsm->u.enable.channel_name,
+                               ev,
+                               filter_expression, bytecode, exclusion,
+                               kernel_poll_pipe[1]);
+               lttng_event_destroy(ev);
+               break;
+       }
+       case LTTNG_LIST_TRACEPOINTS:
+       {
+               struct lttng_event *events;
+               ssize_t nb_events;
+
+               session_lock_list();
+               nb_events = cmd_list_tracepoints(cmd_ctx->lsm->domain.type, &events);
+               session_unlock_list();
+               if (nb_events < 0) {
+                       /* Return value is a negative lttng_error_code. */
+                       ret = -nb_events;
+                       goto error;
+               }
+
+               /*
+                * Setup lttng message with payload size set to the event list size in
+                * bytes and then copy list into the llm payload.
+                */
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, events,
+                       sizeof(struct lttng_event) * nb_events);
+               free(events);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_LIST_TRACEPOINT_FIELDS:
+       {
+               struct lttng_event_field *fields;
+               ssize_t nb_fields;
+
+               session_lock_list();
+               nb_fields = cmd_list_tracepoint_fields(cmd_ctx->lsm->domain.type,
+                               &fields);
+               session_unlock_list();
+               if (nb_fields < 0) {
+                       /* Return value is a negative lttng_error_code. */
+                       ret = -nb_fields;
+                       goto error;
+               }
+
+               /*
+                * Setup lttng message with payload size set to the event list size in
+                * bytes and then copy list into the llm payload.
+                */
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, fields,
+                               sizeof(struct lttng_event_field) * nb_fields);
+               free(fields);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_LIST_SYSCALLS:
+       {
+               struct lttng_event *events;
+               ssize_t nb_events;
+
+               nb_events = cmd_list_syscalls(&events);
+               if (nb_events < 0) {
+                       /* Return value is a negative lttng_error_code. */
+                       ret = -nb_events;
+                       goto error;
+               }
+
+               /*
+                * Setup lttng message with payload size set to the event list size in
+                * bytes and then copy list into the llm payload.
+                */
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, events,
+                       sizeof(struct lttng_event) * nb_events);
+               free(events);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_LIST_TRACKER_PIDS:
+       {
+               int32_t *pids = NULL;
+               ssize_t nr_pids;
+
+               nr_pids = cmd_list_tracker_pids(cmd_ctx->session,
+                               cmd_ctx->lsm->domain.type, &pids);
+               if (nr_pids < 0) {
+                       /* Return value is a negative lttng_error_code. */
+                       ret = -nr_pids;
+                       goto error;
+               }
+
+               /*
+                * Setup lttng message with payload size set to the event list size in
+                * bytes and then copy list into the llm payload.
+                */
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, pids,
+                       sizeof(int32_t) * nr_pids);
+               free(pids);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_SET_CONSUMER_URI:
+       {
+               size_t nb_uri, len;
+               struct lttng_uri *uris;
+
+               nb_uri = cmd_ctx->lsm->u.uri.size;
+               len = nb_uri * sizeof(struct lttng_uri);
+
+               if (nb_uri == 0) {
+                       ret = LTTNG_ERR_INVALID;
+                       goto error;
+               }
+
+               uris = zmalloc(len);
+               if (uris == NULL) {
+                       ret = LTTNG_ERR_FATAL;
+                       goto error;
+               }
+
+               /* Receive variable len data */
+               DBG("Receiving %zu URI(s) from client ...", nb_uri);
+               ret = lttcomm_recv_unix_sock(sock, uris, len);
+               if (ret <= 0) {
+                       DBG("No URIs received from client... continuing");
+                       *sock_error = 1;
+                       ret = LTTNG_ERR_SESSION_FAIL;
+                       free(uris);
+                       goto error;
+               }
+
+               ret = cmd_set_consumer_uri(cmd_ctx->session, nb_uri, uris);
+               free(uris);
+               if (ret != LTTNG_OK) {
+                       goto error;
+               }
+
+
+               break;
+       }
+       case LTTNG_START_TRACE:
+       {
+               /*
+                * On the first start, if we have a kernel session and we have
+                * enabled time or size-based rotations, we have to make sure
+                * the kernel tracer supports it.
+                */
+               if (!cmd_ctx->session->has_been_started && \
+                               cmd_ctx->session->kernel_session && \
+                               (cmd_ctx->session->rotate_timer_period || \
+                                       cmd_ctx->session->rotate_size) && \
+                               !check_rotate_compatible()) {
+                       DBG("Kernel tracer version is not compatible with the rotation feature");
+                       ret = LTTNG_ERR_ROTATION_WRONG_VERSION;
+                       goto error;
+               }
+               ret = cmd_start_trace(cmd_ctx->session);
+               break;
+       }
+       case LTTNG_STOP_TRACE:
+       {
+               ret = cmd_stop_trace(cmd_ctx->session);
+               break;
+       }
+       case LTTNG_CREATE_SESSION:
+       {
+               size_t nb_uri, len;
+               struct lttng_uri *uris = NULL;
+
+               nb_uri = cmd_ctx->lsm->u.uri.size;
+               len = nb_uri * sizeof(struct lttng_uri);
+
+               if (nb_uri > 0) {
+                       uris = zmalloc(len);
+                       if (uris == NULL) {
+                               ret = LTTNG_ERR_FATAL;
+                               goto error;
+                       }
+
+                       /* Receive variable len data */
+                       DBG("Waiting for %zu URIs from client ...", nb_uri);
+                       ret = lttcomm_recv_unix_sock(sock, uris, len);
+                       if (ret <= 0) {
+                               DBG("No URIs received from client... continuing");
+                               *sock_error = 1;
+                               ret = LTTNG_ERR_SESSION_FAIL;
+                               free(uris);
+                               goto error;
+                       }
+
+                       if (nb_uri == 1 && uris[0].dtype != LTTNG_DST_PATH) {
+                               DBG("Creating session with ONE network URI is a bad call");
+                               ret = LTTNG_ERR_SESSION_FAIL;
+                               free(uris);
+                               goto error;
+                       }
+               }
+
+               ret = cmd_create_session_uri(cmd_ctx->lsm->session.name, uris, nb_uri,
+                       &cmd_ctx->creds, 0);
+
+               free(uris);
+
+               break;
+       }
+       case LTTNG_DESTROY_SESSION:
+       {
+               ret = cmd_destroy_session(cmd_ctx->session,
+                               notification_thread_handle);
+               break;
+       }
+       case LTTNG_LIST_DOMAINS:
+       {
+               ssize_t nb_dom;
+               struct lttng_domain *domains = NULL;
+
+               nb_dom = cmd_list_domains(cmd_ctx->session, &domains);
+               if (nb_dom < 0) {
+                       /* Return value is a negative lttng_error_code. */
+                       ret = -nb_dom;
+                       goto error;
+               }
+
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, domains,
+                       nb_dom * sizeof(struct lttng_domain));
+               free(domains);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_LIST_CHANNELS:
+       {
+               ssize_t payload_size;
+               struct lttng_channel *channels = NULL;
+
+               payload_size = cmd_list_channels(cmd_ctx->lsm->domain.type,
+                               cmd_ctx->session, &channels);
+               if (payload_size < 0) {
+                       /* Return value is a negative lttng_error_code. */
+                       ret = -payload_size;
+                       goto error;
+               }
+
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, channels,
+                       payload_size);
+               free(channels);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_LIST_EVENTS:
+       {
+               ssize_t nb_event;
+               struct lttng_event *events = NULL;
+               struct lttcomm_event_command_header cmd_header;
+               size_t total_size;
+
+               memset(&cmd_header, 0, sizeof(cmd_header));
+               /* Extended infos are included at the end of events */
+               nb_event = cmd_list_events(cmd_ctx->lsm->domain.type,
+                       cmd_ctx->session, cmd_ctx->lsm->u.list.channel_name,
+                       &events, &total_size);
+
+               if (nb_event < 0) {
+                       /* Return value is a negative lttng_error_code. */
+                       ret = -nb_event;
+                       goto error;
+               }
+
+               cmd_header.nb_events = nb_event;
+               ret = setup_lttng_msg(cmd_ctx, events, total_size,
+                       &cmd_header, sizeof(cmd_header));
+               free(events);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_LIST_SESSIONS:
+       {
+               unsigned int nr_sessions;
+               void *sessions_payload;
+               size_t payload_len;
+
+               session_lock_list();
+               nr_sessions = lttng_sessions_count(
+                               LTTNG_SOCK_GET_UID_CRED(&cmd_ctx->creds),
+                               LTTNG_SOCK_GET_GID_CRED(&cmd_ctx->creds));
+               payload_len = sizeof(struct lttng_session) * nr_sessions;
+               sessions_payload = zmalloc(payload_len);
+
+               if (!sessions_payload) {
+                       session_unlock_list();
+                       ret = -ENOMEM;
+                       goto setup_error;
+               }
+
+               cmd_list_lttng_sessions(sessions_payload,
+                       LTTNG_SOCK_GET_UID_CRED(&cmd_ctx->creds),
+                       LTTNG_SOCK_GET_GID_CRED(&cmd_ctx->creds));
+               session_unlock_list();
+
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, sessions_payload,
+                       payload_len);
+               free(sessions_payload);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_REGISTER_CONSUMER:
+       {
+               struct consumer_data *cdata;
+
+               switch (cmd_ctx->lsm->domain.type) {
+               case LTTNG_DOMAIN_KERNEL:
+                       cdata = &kconsumer_data;
+                       break;
+               default:
+                       ret = LTTNG_ERR_UND;
+                       goto error;
+               }
+
+               ret = cmd_register_consumer(cmd_ctx->session, cmd_ctx->lsm->domain.type,
+                               cmd_ctx->lsm->u.reg.path, cdata);
+               break;
+       }
+       case LTTNG_DATA_PENDING:
+       {
+               int pending_ret;
+               uint8_t pending_ret_byte;
+
+               pending_ret = cmd_data_pending(cmd_ctx->session);
+
+               /*
+                * FIXME
+                *
+                * This function may returns 0 or 1 to indicate whether or not
+                * there is data pending. In case of error, it should return an
+                * LTTNG_ERR code. However, some code paths may still return
+                * a nondescript error code, which we handle by returning an
+                * "unknown" error.
+                */
+               if (pending_ret == 0 || pending_ret == 1) {
+                       /*
+                        * ret will be set to LTTNG_OK at the end of
+                        * this function.
+                        */
+               } else if (pending_ret < 0) {
+                       ret = LTTNG_ERR_UNK;
+                       goto setup_error;
+               } else {
+                       ret = pending_ret;
+                       goto setup_error;
+               }
+
+               pending_ret_byte = (uint8_t) pending_ret;
+
+               /* 1 byte to return whether or not data is pending */
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx,
+                       &pending_ret_byte, 1);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_SNAPSHOT_ADD_OUTPUT:
+       {
+               struct lttcomm_lttng_output_id reply;
+
+               ret = cmd_snapshot_add_output(cmd_ctx->session,
+                               &cmd_ctx->lsm->u.snapshot_output.output, &reply.id);
+               if (ret != LTTNG_OK) {
+                       goto error;
+               }
+
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, &reply,
+                       sizeof(reply));
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               /* Copy output list into message payload */
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_SNAPSHOT_DEL_OUTPUT:
+       {
+               ret = cmd_snapshot_del_output(cmd_ctx->session,
+                               &cmd_ctx->lsm->u.snapshot_output.output);
+               break;
+       }
+       case LTTNG_SNAPSHOT_LIST_OUTPUT:
+       {
+               ssize_t nb_output;
+               struct lttng_snapshot_output *outputs = NULL;
+
+               nb_output = cmd_snapshot_list_outputs(cmd_ctx->session, &outputs);
+               if (nb_output < 0) {
+                       ret = -nb_output;
+                       goto error;
+               }
+
+               assert((nb_output > 0 && outputs) || nb_output == 0);
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, outputs,
+                               nb_output * sizeof(struct lttng_snapshot_output));
+               free(outputs);
+
+               if (ret < 0) {
+                       goto setup_error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_SNAPSHOT_RECORD:
+       {
+               ret = cmd_snapshot_record(cmd_ctx->session,
+                               &cmd_ctx->lsm->u.snapshot_record.output,
+                               cmd_ctx->lsm->u.snapshot_record.wait);
+               break;
+       }
+       case LTTNG_CREATE_SESSION_SNAPSHOT:
+       {
+               size_t nb_uri, len;
+               struct lttng_uri *uris = NULL;
+
+               nb_uri = cmd_ctx->lsm->u.uri.size;
+               len = nb_uri * sizeof(struct lttng_uri);
+
+               if (nb_uri > 0) {
+                       uris = zmalloc(len);
+                       if (uris == NULL) {
+                               ret = LTTNG_ERR_FATAL;
+                               goto error;
+                       }
+
+                       /* Receive variable len data */
+                       DBG("Waiting for %zu URIs from client ...", nb_uri);
+                       ret = lttcomm_recv_unix_sock(sock, uris, len);
+                       if (ret <= 0) {
+                               DBG("No URIs received from client... continuing");
+                               *sock_error = 1;
+                               ret = LTTNG_ERR_SESSION_FAIL;
+                               free(uris);
+                               goto error;
+                       }
+
+                       if (nb_uri == 1 && uris[0].dtype != LTTNG_DST_PATH) {
+                               DBG("Creating session with ONE network URI is a bad call");
+                               ret = LTTNG_ERR_SESSION_FAIL;
+                               free(uris);
+                               goto error;
+                       }
+               }
+
+               ret = cmd_create_session_snapshot(cmd_ctx->lsm->session.name, uris,
+                               nb_uri, &cmd_ctx->creds);
+               free(uris);
+               break;
+       }
+       case LTTNG_CREATE_SESSION_LIVE:
+       {
+               size_t nb_uri, len;
+               struct lttng_uri *uris = NULL;
+
+               nb_uri = cmd_ctx->lsm->u.uri.size;
+               len = nb_uri * sizeof(struct lttng_uri);
+
+               if (nb_uri > 0) {
+                       uris = zmalloc(len);
+                       if (uris == NULL) {
+                               ret = LTTNG_ERR_FATAL;
+                               goto error;
+                       }
+
+                       /* Receive variable len data */
+                       DBG("Waiting for %zu URIs from client ...", nb_uri);
+                       ret = lttcomm_recv_unix_sock(sock, uris, len);
+                       if (ret <= 0) {
+                               DBG("No URIs received from client... continuing");
+                               *sock_error = 1;
+                               ret = LTTNG_ERR_SESSION_FAIL;
+                               free(uris);
+                               goto error;
+                       }
+
+                       if (nb_uri == 1 && uris[0].dtype != LTTNG_DST_PATH) {
+                               DBG("Creating session with ONE network URI is a bad call");
+                               ret = LTTNG_ERR_SESSION_FAIL;
+                               free(uris);
+                               goto error;
+                       }
+               }
+
+               ret = cmd_create_session_uri(cmd_ctx->lsm->session.name, uris,
+                               nb_uri, &cmd_ctx->creds, cmd_ctx->lsm->u.session_live.timer_interval);
+               free(uris);
+               break;
+       }
+       case LTTNG_SAVE_SESSION:
+       {
+               ret = cmd_save_sessions(&cmd_ctx->lsm->u.save_session.attr,
+                       &cmd_ctx->creds);
+               break;
+       }
+       case LTTNG_SET_SESSION_SHM_PATH:
+       {
+               ret = cmd_set_session_shm_path(cmd_ctx->session,
+                               cmd_ctx->lsm->u.set_shm_path.shm_path);
+               break;
+       }
+       case LTTNG_REGENERATE_METADATA:
+       {
+               ret = cmd_regenerate_metadata(cmd_ctx->session);
+               break;
+       }
+       case LTTNG_REGENERATE_STATEDUMP:
+       {
+               ret = cmd_regenerate_statedump(cmd_ctx->session);
+               break;
+       }
+       case LTTNG_REGISTER_TRIGGER:
+       {
+               ret = cmd_register_trigger(cmd_ctx, sock,
+                               notification_thread_handle);
+               break;
+       }
+       case LTTNG_UNREGISTER_TRIGGER:
+       {
+               ret = cmd_unregister_trigger(cmd_ctx, sock,
+                               notification_thread_handle);
+               break;
+       }
+       case LTTNG_ROTATE_SESSION:
+       {
+               struct lttng_rotate_session_return rotate_return;
+
+               DBG("Client rotate session \"%s\"", cmd_ctx->session->name);
+
+               memset(&rotate_return, 0, sizeof(rotate_return));
+               if (cmd_ctx->session->kernel_session && !check_rotate_compatible()) {
+                       DBG("Kernel tracer version is not compatible with the rotation feature");
+                       ret = LTTNG_ERR_ROTATION_WRONG_VERSION;
+                       goto error;
+               }
+
+               ret = cmd_rotate_session(cmd_ctx->session, &rotate_return);
+               if (ret < 0) {
+                       ret = -ret;
+                       goto error;
+               }
+
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, &rotate_return,
+                               sizeof(rotate_return));
+               if (ret < 0) {
+                       ret = -ret;
+                       goto error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_ROTATION_GET_INFO:
+       {
+               struct lttng_rotation_get_info_return get_info_return;
+
+               memset(&get_info_return, 0, sizeof(get_info_return));
+               ret = cmd_rotate_get_info(cmd_ctx->session, &get_info_return,
+                               cmd_ctx->lsm->u.get_rotation_info.rotation_id);
+               if (ret < 0) {
+                       ret = -ret;
+                       goto error;
+               }
+
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, &get_info_return,
+                               sizeof(get_info_return));
+               if (ret < 0) {
+                       ret = -ret;
+                       goto error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       case LTTNG_ROTATION_SET_SCHEDULE:
+       {
+               bool set_schedule;
+               enum lttng_rotation_schedule_type schedule_type;
+               uint64_t value;
+
+               if (cmd_ctx->session->kernel_session && !check_rotate_compatible()) {
+                       DBG("Kernel tracer version does not support session rotations");
+                       ret = LTTNG_ERR_ROTATION_WRONG_VERSION;
+                       goto error;
+               }
+
+               set_schedule = cmd_ctx->lsm->u.rotation_set_schedule.set == 1;
+               schedule_type = (enum lttng_rotation_schedule_type) cmd_ctx->lsm->u.rotation_set_schedule.type;
+               value = cmd_ctx->lsm->u.rotation_set_schedule.value;
+
+               ret = cmd_rotation_set_schedule(cmd_ctx->session,
+                               set_schedule,
+                               schedule_type,
+                               value,
+                               notification_thread_handle);
+               if (ret != LTTNG_OK) {
+                       goto error;
+               }
+
+               break;
+       }
+       case LTTNG_SESSION_LIST_ROTATION_SCHEDULES:
+       {
+               struct lttng_session_list_schedules_return schedules = {
+                       .periodic.set = !!cmd_ctx->session->rotate_timer_period,
+                       .periodic.value = cmd_ctx->session->rotate_timer_period,
+                       .size.set = !!cmd_ctx->session->rotate_size,
+                       .size.value = cmd_ctx->session->rotate_size,
+               };
+
+               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, &schedules,
+                               sizeof(schedules));
+               if (ret < 0) {
+                       ret = -ret;
+                       goto error;
+               }
+
+               ret = LTTNG_OK;
+               break;
+       }
+       default:
+               ret = LTTNG_ERR_UND;
+               break;
+       }
+
+error:
+       if (cmd_ctx->llm == NULL) {
+               DBG("Missing llm structure. Allocating one.");
+               if (setup_lttng_msg_no_cmd_header(cmd_ctx, NULL, 0) < 0) {
+                       goto setup_error;
+               }
+       }
+       /* Set return code */
+       cmd_ctx->llm->ret_code = ret;
+setup_error:
+       if (cmd_ctx->session) {
+               session_unlock(cmd_ctx->session);
+               session_put(cmd_ctx->session);
+       }
+       if (need_tracing_session) {
+               session_unlock_list();
+       }
+init_setup_error:
+       assert(!rcu_read_ongoing());
+       return ret;
+}
+
+static int create_client_sock(void)
+{
+       int ret, client_sock;
+       const mode_t old_umask = umask(0);
+
+       /* Create client tool unix socket */
+       client_sock = lttcomm_create_unix_sock(config.client_unix_sock_path.value);
+       if (client_sock < 0) {
+               ERR("Create unix sock failed: %s", config.client_unix_sock_path.value);
+               ret = -1;
+               goto end;
+       }
+
+       /* Set the cloexec flag */
+       ret = utils_set_fd_cloexec(client_sock);
+       if (ret < 0) {
+               ERR("Unable to set CLOEXEC flag to the client Unix socket (fd: %d). "
+                               "Continuing but note that the consumer daemon will have a "
+                               "reference to this socket on exec()", client_sock);
+       }
+
+       /* File permission MUST be 660 */
+       ret = chmod(config.client_unix_sock_path.value, S_IRUSR | S_IWUSR | S_IRGRP | S_IWGRP);
+       if (ret < 0) {
+               ERR("Set file permissions failed: %s", config.client_unix_sock_path.value);
+               PERROR("chmod");
+               goto end;
+       }
+       DBG("Created client socket (fd = %i)", client_sock);
+       ret = client_sock;
+end:
+       umask(old_umask);
+       return ret;
+}
+
+static void cleanup_client_thread(void *data)
+{
+       struct lttng_pipe *quit_pipe = data;
+
+       lttng_pipe_destroy(quit_pipe);
+}
+
+/*
+ * This thread manage all clients request using the unix client socket for
+ * communication.
+ */
+static void *thread_manage_clients(void *data)
+{
+       int sock = -1, ret, i, pollfd, err = -1;
+       int sock_error;
+       uint32_t revents, nb_fd;
+       struct command_ctx *cmd_ctx = NULL;
+       struct lttng_poll_event events;
+       int client_sock = -1;
+       struct lttng_pipe *quit_pipe = data;
+       const int thread_quit_pipe_fd = lttng_pipe_get_readfd(quit_pipe);
+
+       DBG("[thread] Manage client started");
+
+       is_root = (getuid() == 0);
+
+       client_sock = create_client_sock();
+       if (client_sock < 0) {
+               goto error_listen;
+       }
+
+       rcu_register_thread();
+
+       health_register(health_sessiond, HEALTH_SESSIOND_TYPE_CMD);
+
+       health_code_update();
+
+       ret = lttcomm_listen_unix_sock(client_sock);
+       if (ret < 0) {
+               goto error_listen;
+       }
+
+       /*
+        * Pass 2 as size here for the thread quit pipe and client_sock. Nothing
+        * more will be added to this poll set.
+        */
+       ret = lttng_poll_create(&events, 2, LTTNG_CLOEXEC);
+       if (ret < 0) {
+               goto error_create_poll;
+       }
+
+       /* Add the application registration socket */
+       ret = lttng_poll_add(&events, client_sock, LPOLLIN | LPOLLPRI);
+       if (ret < 0) {
+               goto error;
+       }
+
+       /* Add thread quit pipe */
+       ret = lttng_poll_add(&events, thread_quit_pipe_fd, LPOLLIN | LPOLLERR);
+       if (ret < 0) {
+               goto error;
+       }
+
+       /* This testpoint is after we signal readiness to the parent. */
+       if (testpoint(sessiond_thread_manage_clients)) {
+               goto error;
+       }
+
+       if (testpoint(sessiond_thread_manage_clients_before_loop)) {
+               goto error;
+       }
+
+       health_code_update();
+
+       /* Set state as running. */
+        set_thread_state_running();
+
+       while (1) {
+               const struct cmd_completion_handler *cmd_completion_handler;
+
+               DBG("Accepting client command ...");
+
+               /* Inifinite blocking call, waiting for transmission */
+       restart:
+               health_poll_entry();
+               ret = lttng_poll_wait(&events, -1);
+               health_poll_exit();
+               if (ret < 0) {
+                       /*
+                        * Restart interrupted system call.
+                        */
+                       if (errno == EINTR) {
+                               goto restart;
+                       }
+                       goto error;
+               }
+
+               nb_fd = ret;
+
+               for (i = 0; i < nb_fd; i++) {
+                       revents = LTTNG_POLL_GETEV(&events, i);
+                       pollfd = LTTNG_POLL_GETFD(&events, i);
+
+                       health_code_update();
+
+                       if (!revents) {
+                               /* No activity for this FD (poll implementation). */
+                               continue;
+                       }
+
+                       if (pollfd == thread_quit_pipe_fd) {
+                               err = 0;
+                               goto exit;
+                       } else {
+                               /* Event on the registration socket */
+                               if (revents & LPOLLIN) {
+                                       continue;
+                               } else if (revents & (LPOLLERR | LPOLLHUP | LPOLLRDHUP)) {
+                                       ERR("Client socket poll error");
+                                       goto error;
+                               } else {
+                                       ERR("Unexpected poll events %u for sock %d", revents, pollfd);
+                                       goto error;
+                               }
+                       }
+               }
+
+               DBG("Wait for client response");
+
+               health_code_update();
+
+               sock = lttcomm_accept_unix_sock(client_sock);
+               if (sock < 0) {
+                       goto error;
+               }
+
+               /*
+                * Set the CLOEXEC flag. Return code is useless because either way, the
+                * show must go on.
+                */
+               (void) utils_set_fd_cloexec(sock);
+
+               /* Set socket option for credentials retrieval */
+               ret = lttcomm_setsockopt_creds_unix_sock(sock);
+               if (ret < 0) {
+                       goto error;
+               }
+
+               /* Allocate context command to process the client request */
+               cmd_ctx = zmalloc(sizeof(struct command_ctx));
+               if (cmd_ctx == NULL) {
+                       PERROR("zmalloc cmd_ctx");
+                       goto error;
+               }
+
+               /* Allocate data buffer for reception */
+               cmd_ctx->lsm = zmalloc(sizeof(struct lttcomm_session_msg));
+               if (cmd_ctx->lsm == NULL) {
+                       PERROR("zmalloc cmd_ctx->lsm");
+                       goto error;
+               }
+
+               cmd_ctx->llm = NULL;
+               cmd_ctx->session = NULL;
+
+               health_code_update();
+
+               /*
+                * Data is received from the lttng client. The struct
+                * lttcomm_session_msg (lsm) contains the command and data request of
+                * the client.
+                */
+               DBG("Receiving data from client ...");
+               ret = lttcomm_recv_creds_unix_sock(sock, cmd_ctx->lsm,
+                               sizeof(struct lttcomm_session_msg), &cmd_ctx->creds);
+               if (ret <= 0) {
+                       DBG("Nothing recv() from client... continuing");
+                       ret = close(sock);
+                       if (ret) {
+                               PERROR("close");
+                       }
+                       sock = -1;
+                       clean_command_ctx(&cmd_ctx);
+                       continue;
+               }
+
+               health_code_update();
+
+               // TODO: Validate cmd_ctx including sanity check for
+               // security purpose.
+
+               rcu_thread_online();
+               /*
+                * This function dispatch the work to the kernel or userspace tracer
+                * libs and fill the lttcomm_lttng_msg data structure of all the needed
+                * informations for the client. The command context struct contains
+                * everything this function may needs.
+                */
+               ret = process_client_msg(cmd_ctx, sock, &sock_error);
+               rcu_thread_offline();
+               if (ret < 0) {
+                       ret = close(sock);
+                       if (ret) {
+                               PERROR("close");
+                       }
+                       sock = -1;
+                       /*
+                        * TODO: Inform client somehow of the fatal error. At
+                        * this point, ret < 0 means that a zmalloc failed
+                        * (ENOMEM). Error detected but still accept
+                        * command, unless a socket error has been
+                        * detected.
+                        */
+                       clean_command_ctx(&cmd_ctx);
+                       continue;
+               }
+
+               cmd_completion_handler = cmd_pop_completion_handler();
+               if (cmd_completion_handler) {
+                       enum lttng_error_code completion_code;
+
+                       completion_code = cmd_completion_handler->run(
+                                       cmd_completion_handler->data);
+                       if (completion_code != LTTNG_OK) {
+                               clean_command_ctx(&cmd_ctx);
+                               continue;
+                       }
+               }
+
+               health_code_update();
+
+               DBG("Sending response (size: %d, retcode: %s (%d))",
+                               cmd_ctx->lttng_msg_size,
+                               lttng_strerror(-cmd_ctx->llm->ret_code),
+                               cmd_ctx->llm->ret_code);
+               ret = send_unix_sock(sock, cmd_ctx->llm, cmd_ctx->lttng_msg_size);
+               if (ret < 0) {
+                       ERR("Failed to send data back to client");
+               }
+
+               /* End of transmission */
+               ret = close(sock);
+               if (ret) {
+                       PERROR("close");
+               }
+               sock = -1;
+
+               clean_command_ctx(&cmd_ctx);
+
+               health_code_update();
+       }
+
+exit:
+error:
+       if (sock >= 0) {
+               ret = close(sock);
+               if (ret) {
+                       PERROR("close");
+               }
+       }
+
+       lttng_poll_clean(&events);
+       clean_command_ctx(&cmd_ctx);
+
+error_listen:
+error_create_poll:
+       unlink(config.client_unix_sock_path.value);
+       if (client_sock >= 0) {
+               ret = close(client_sock);
+               if (ret) {
+                       PERROR("close");
+               }
+       }
+
+       if (err) {
+               health_error();
+               ERR("Health error occurred in %s", __func__);
+       }
+
+       health_unregister(health_sessiond);
+
+       DBG("Client thread dying");
+
+       rcu_unregister_thread();
+
+       /*
+        * Since we are creating the consumer threads, we own them, so we need
+        * to join them before our thread exits.
+        */
+       ret = join_consumer_thread(&kconsumer_data);
+       if (ret) {
+               errno = ret;
+               PERROR("join_consumer");
+       }
+
+       ret = join_consumer_thread(&ustconsumer32_data);
+       if (ret) {
+               errno = ret;
+               PERROR("join_consumer ust32");
+       }
+
+       ret = join_consumer_thread(&ustconsumer64_data);
+       if (ret) {
+               errno = ret;
+               PERROR("join_consumer ust64");
+       }
+       return NULL;
+}
+
+static
+bool shutdown_client_thread(void *thread_data)
+{
+       struct lttng_pipe *client_quit_pipe = thread_data;
+       const int write_fd = lttng_pipe_get_writefd(client_quit_pipe);
+
+       return notify_thread_pipe(write_fd) == 1;
+}
+
+struct lttng_thread *launch_client_thread(void)
+{
+       struct lttng_pipe *client_quit_pipe;
+       struct lttng_thread *thread;
+
+       client_quit_pipe = lttng_pipe_open(FD_CLOEXEC);
+       if (!client_quit_pipe) {
+               goto error;
+       }
+
+       thread = lttng_thread_create("Client management",
+                       thread_manage_clients,
+                       shutdown_client_thread,
+                       cleanup_client_thread,
+                       client_quit_pipe);
+       if (!thread) {
+               goto error;
+       }
+
+       /*
+        * This thread is part of the threads that need to be fully
+        * initialized before the session daemon is marked as "ready".
+        */
+       wait_thread_state_running();
+
+       return thread;
+error:
+       cleanup_client_thread(client_quit_pipe);
+       return NULL;
+}
diff --git a/src/bin/lttng-sessiond/client.h b/src/bin/lttng-sessiond/client.h
new file mode 100644 (file)
index 0000000..f821398
--- /dev/null
@@ -0,0 +1,27 @@
+/*
+ * Copyright (C) 2011 - David Goulet <david.goulet@polymtl.ca>
+ *                      Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
+ *               2013 - 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.
+ *
+ * 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.
+ */
+
+#ifndef CLIENT_SESSIOND_H
+#define CLIENT_SESSIOND_H
+
+#include "thread.h"
+
+struct lttng_thread *launch_client_thread(void);
+
+#endif /* CLIENT_SESSIOND_H */
index 63fd799386581059d80f4235593f489a1f76f4fc..82a51512ad4e1dea80cc263a92be347d6aa45765 100644 (file)
@@ -31,6 +31,8 @@ struct notification_thread_handle *notification_thread_handle;
 struct lttng_ht *agent_apps_ht_by_sock = NULL;
 
 int kernel_tracer_fd = -1;
+struct lttng_kernel_tracer_version kernel_tracer_version;
+struct lttng_kernel_tracer_abi_version kernel_tracer_abi_version;
 
 int apps_cmd_notify_pipe[2] = { -1, -1 };
 int kernel_poll_pipe[2] = { -1, -1 };
@@ -39,3 +41,39 @@ pid_t ppid;
 pid_t child_ppid;
 
 struct sessiond_config config;
+
+struct consumer_data kconsumer_data = {
+       .type = LTTNG_CONSUMER_KERNEL,
+       .err_sock = -1,
+       .cmd_sock = -1,
+       .channel_monitor_pipe = -1,
+       .pid_mutex = PTHREAD_MUTEX_INITIALIZER,
+       .lock = PTHREAD_MUTEX_INITIALIZER,
+       .cond = PTHREAD_COND_INITIALIZER,
+       .cond_mutex = PTHREAD_MUTEX_INITIALIZER,
+};
+
+struct consumer_data ustconsumer64_data = {
+       .type = LTTNG_CONSUMER64_UST,
+       .err_sock = -1,
+       .cmd_sock = -1,
+       .channel_monitor_pipe = -1,
+       .pid_mutex = PTHREAD_MUTEX_INITIALIZER,
+       .lock = PTHREAD_MUTEX_INITIALIZER,
+       .cond = PTHREAD_COND_INITIALIZER,
+       .cond_mutex = PTHREAD_MUTEX_INITIALIZER,
+};
+
+struct consumer_data ustconsumer32_data = {
+       .type = LTTNG_CONSUMER32_UST,
+       .err_sock = -1,
+       .cmd_sock = -1,
+       .channel_monitor_pipe = -1,
+       .pid_mutex = PTHREAD_MUTEX_INITIALIZER,
+       .lock = PTHREAD_MUTEX_INITIALIZER,
+       .cond = PTHREAD_COND_INITIALIZER,
+       .cond_mutex = PTHREAD_MUTEX_INITIALIZER,
+};
+
+enum consumerd_state ust_consumerd_state;
+enum consumerd_state kernel_consumerd_state;
index a6c393c9f994a912230d46f5c7f926ee84d040f6..a93a799f6072f023519596d2e8e7e550da6ea855 100644 (file)
@@ -30,15 +30,6 @@ void load_session_destroy_data(struct load_session_thread_data *data)
        if (!data) {
                return;
        }
-
-       if (data->sem_initialized) {
-               int ret;
-
-               ret = sem_destroy(&data->message_thread_ready);
-               if (ret) {
-                       PERROR("sem_destroy message_thread_ready");
-               }
-       }
 }
 
 /*
@@ -50,7 +41,6 @@ void load_session_destroy_data(struct load_session_thread_data *data)
  */
 int load_session_init_data(struct load_session_thread_data **data)
 {
-       int ret;
        struct load_session_thread_data *_data = NULL;
 
        assert(data);
@@ -64,12 +54,6 @@ int load_session_init_data(struct load_session_thread_data **data)
                PERROR("zmalloc load session info");
                goto error;
        }
-       ret = sem_init(&_data->message_thread_ready, 0, 0);
-       if (ret) {
-               PERROR("sem_init message_thread_ready");
-               goto error;
-       }
-       _data->sem_initialized = 1;
 
        *data = _data;
        return 0;
@@ -90,19 +74,12 @@ void *thread_load_session(void *data)
 
        DBG("[load-session-thread] Load session");
 
-       ret = sem_wait(&info->message_thread_ready);
-       if (ret) {
-               PERROR("sem_wait message_thread_ready");
-               goto end;
-       }
-
        /* Override existing session and autoload also. */
        ret = config_load_session(info->path, NULL, 1, 1, NULL);
        if (ret) {
                ERR("Session load failed: %s", error_get_str(ret));
        }
 
-end:
        sessiond_signal_parents();
        return NULL;
 }
index b7cd763400b4d63fe63a116e833175f28ae5ebb7..aaaea91890d97375db05745b189c0fb16df7accf 100644 (file)
 
 /* Data passed to the thread. */
 struct load_session_thread_data {
-       /* Flag if the sem_init() has been done successfully on the sem. */
-       unsigned int sem_initialized:1;
-
-       /*
-        * The load session thread waits on that semaphore which the client thread
-        * will do a sem_post() to unblock it.
-        */
-       sem_t message_thread_ready;
-
        /* Path where the sessions are located. */
        const char *path;
 };
index 9c5fcd32cd22930c19f2bcd9ab35f4f38931c730..243138a9f043311eb611318b0b2d058eda69587f 100644 (file)
 #include "notification-thread.h"
 #include "sessiond-config.h"
 
+/*
+ * Consumer daemon state which is changed when spawning it, killing it or in
+ * case of a fatal error.
+ */
+enum consumerd_state {
+       CONSUMER_STARTED = 1,
+       CONSUMER_STOPPED = 2,
+       CONSUMER_ERROR   = 3,
+};
+
+/*
+ * This consumer daemon state is used to validate if a client command will be
+ * able to reach the consumer. If not, the client is informed. For instance,
+ * doing a "lttng start" when the consumer state is set to ERROR will return an
+ * error to the client.
+ *
+ * The following example shows a possible race condition of this scheme:
+ *
+ * consumer thread error happens
+ *                                    client cmd arrives
+ *                                    client cmd checks state -> still OK
+ * consumer thread exit, sets error
+ *                                    client cmd try to talk to consumer
+ *                                    ...
+ *
+ * However, since the consumer is a different daemon, we have no way of making
+ * sure the command will reach it safely even with this state flag. This is why
+ * we consider that up to the state validation during command processing, the
+ * command is safe. After that, we can not guarantee the correctness of the
+ * client request vis-a-vis the consumer.
+ */
+extern enum consumerd_state ust_consumerd_state;
+extern enum consumerd_state kernel_consumerd_state;
+
 extern const char default_home_dir[],
        default_tracing_group[],
        default_ust_sock_dir[],
@@ -38,6 +72,8 @@ extern const char default_home_dir[],
 
 /* Set in main.c at boot time of the daemon */
 extern int kernel_tracer_fd;
+extern struct lttng_kernel_tracer_version kernel_tracer_version;
+extern struct lttng_kernel_tracer_abi_version kernel_tracer_abi_version;
 
 /* Notification thread handle. */
 extern struct notification_thread_handle *notification_thread_handle;
@@ -126,6 +162,11 @@ extern pid_t ppid;
 /* Internal parent PID use with daemonize. */
 extern pid_t child_ppid;
 
+/* Consumer daemon specific control data. */
+extern struct consumer_data ustconsumer32_data;
+extern struct consumer_data ustconsumer64_data;
+extern struct consumer_data kconsumer_data;
+
 int sessiond_init_thread_quit_pipe(void);
 int sessiond_check_thread_quit_pipe(int fd, uint32_t events);
 int sessiond_wait_for_quit_pipe(unsigned int timeout_us);
@@ -139,4 +180,6 @@ void sessiond_signal_parents(void);
 void sessiond_set_client_thread_state(bool running);
 void sessiond_wait_client_thread_stopped(void);
 
+void *thread_manage_consumer(void *data);
+
 #endif /* _LTT_SESSIOND_H */
index c82e0095bc1e476b015a164a61975a661d39ff95..1641ebeb876dfe2bc820ce6b0ea5d608148656c1 100644 (file)
@@ -50,7 +50,6 @@
 #include <common/daemonize.h>
 #include <common/config/session-config.h>
 #include <common/dynamic-buffer.h>
-#include <lttng/userspace-probe-internal.h>
 #include <lttng/event-internal.h>
 
 #include "lttng-sessiond.h"
@@ -83,6 +82,7 @@
 #include "sessiond-config.h"
 #include "timer.h"
 #include "thread.h"
+#include "client.h"
 
 static const char *help_msg =
 #ifdef LTTNG_EMBED_HELP
@@ -98,44 +98,6 @@ static int lockfile_fd = -1;
 /* Set to 1 when a SIGUSR1 signal is received. */
 static int recv_child_signal;
 
-static struct lttng_kernel_tracer_version kernel_tracer_version;
-static struct lttng_kernel_tracer_abi_version kernel_tracer_abi_version;
-
-/*
- * Consumer daemon specific control data. Every value not initialized here is
- * set to 0 by the static definition.
- */
-static struct consumer_data kconsumer_data = {
-       .type = LTTNG_CONSUMER_KERNEL,
-       .err_sock = -1,
-       .cmd_sock = -1,
-       .channel_monitor_pipe = -1,
-       .pid_mutex = PTHREAD_MUTEX_INITIALIZER,
-       .lock = PTHREAD_MUTEX_INITIALIZER,
-       .cond = PTHREAD_COND_INITIALIZER,
-       .cond_mutex = PTHREAD_MUTEX_INITIALIZER,
-};
-static struct consumer_data ustconsumer64_data = {
-       .type = LTTNG_CONSUMER64_UST,
-       .err_sock = -1,
-       .cmd_sock = -1,
-       .channel_monitor_pipe = -1,
-       .pid_mutex = PTHREAD_MUTEX_INITIALIZER,
-       .lock = PTHREAD_MUTEX_INITIALIZER,
-       .cond = PTHREAD_COND_INITIALIZER,
-       .cond_mutex = PTHREAD_MUTEX_INITIALIZER,
-};
-static struct consumer_data ustconsumer32_data = {
-       .type = LTTNG_CONSUMER32_UST,
-       .err_sock = -1,
-       .cmd_sock = -1,
-       .channel_monitor_pipe = -1,
-       .pid_mutex = PTHREAD_MUTEX_INITIALIZER,
-       .lock = PTHREAD_MUTEX_INITIALIZER,
-       .cond = PTHREAD_COND_INITIALIZER,
-       .cond_mutex = PTHREAD_MUTEX_INITIALIZER,
-};
-
 /* Command line options */
 static const struct option long_options[] = {
        { "client-sock", required_argument, 0, 'c' },
@@ -175,8 +137,6 @@ static const char *config_ignore_options[] = { "help", "version", "config" };
 /* Shared between threads */
 static int dispatch_thread_exit;
 
-/* Sockets and FDs */
-static int client_sock = -1;
 static int apps_sock = -1;
 
 /*
@@ -189,7 +149,6 @@ static int apps_cmd_pipe[2] = { -1, -1 };
 static pthread_t apps_thread;
 static pthread_t apps_notify_thread;
 static pthread_t reg_apps_thread;
-static pthread_t client_thread;
 static pthread_t kernel_thread;
 static pthread_t dispatch_thread;
 static pthread_t agent_reg_thread;
@@ -208,40 +167,6 @@ static struct ust_cmd_queue ust_cmd_queue;
 
 static const char *module_proc_lttng = "/proc/lttng";
 
-/*
- * Consumer daemon state which is changed when spawning it, killing it or in
- * case of a fatal error.
- */
-enum consumerd_state {
-       CONSUMER_STARTED = 1,
-       CONSUMER_STOPPED = 2,
-       CONSUMER_ERROR   = 3,
-};
-
-/*
- * This consumer daemon state is used to validate if a client command will be
- * able to reach the consumer. If not, the client is informed. For instance,
- * doing a "lttng start" when the consumer state is set to ERROR will return an
- * error to the client.
- *
- * The following example shows a possible race condition of this scheme:
- *
- * consumer thread error happens
- *                                    client cmd arrives
- *                                    client cmd checks state -> still OK
- * consumer thread exit, sets error
- *                                    client cmd try to talk to consumer
- *                                    ...
- *
- * However, since the consumer is a different daemon, we have no way of making
- * sure the command will reach it safely even with this state flag. This is why
- * we consider that up to the state validation during command processing, the
- * command is safe. After that, we can not guarantee the correctness of the
- * client request vis-a-vis the consumer.
- */
-static enum consumerd_state ust_consumerd_state;
-static enum consumerd_state kernel_consumerd_state;
-
 /* Load session thread information to operate. */
 static struct load_session_thread_data *load_info;
 
@@ -466,39 +391,6 @@ static void sessiond_cleanup_options(void)
        run_as_destroy_worker();
 }
 
-/*
- * Send data on a unix socket using the liblttsessiondcomm API.
- *
- * Return lttcomm error code.
- */
-static int send_unix_sock(int sock, void *buf, size_t len)
-{
-       /* Check valid length */
-       if (len == 0) {
-               return -1;
-       }
-
-       return lttcomm_send_unix_sock(sock, buf, len);
-}
-
-/*
- * Free memory of a command context structure.
- */
-static void clean_command_ctx(struct command_ctx **cmd_ctx)
-{
-       DBG("Clean command context structure");
-       if (*cmd_ctx) {
-               if ((*cmd_ctx)->llm) {
-                       free((*cmd_ctx)->llm);
-               }
-               if ((*cmd_ctx)->lsm) {
-                       free((*cmd_ctx)->lsm);
-               }
-               free(*cmd_ctx);
-               *cmd_ctx = NULL;
-       }
-}
-
 /*
  * Notify UST applications using the shm mmap futex.
  */
@@ -524,62 +416,6 @@ error:
        return -1;
 }
 
-/*
- * Setup the outgoing data buffer for the response (llm) by allocating the
- * right amount of memory and copying the original information from the lsm
- * structure.
- *
- * Return 0 on success, negative value on error.
- */
-static int setup_lttng_msg(struct command_ctx *cmd_ctx,
-       const void *payload_buf, size_t payload_len,
-       const void *cmd_header_buf, size_t cmd_header_len)
-{
-       int ret = 0;
-       const size_t header_len = sizeof(struct lttcomm_lttng_msg);
-       const size_t cmd_header_offset = header_len;
-       const size_t payload_offset = cmd_header_offset + cmd_header_len;
-       const size_t total_msg_size = header_len + cmd_header_len + payload_len;
-
-       cmd_ctx->llm = zmalloc(total_msg_size);
-
-       if (cmd_ctx->llm == NULL) {
-               PERROR("zmalloc");
-               ret = -ENOMEM;
-               goto end;
-       }
-
-       /* Copy common data */
-       cmd_ctx->llm->cmd_type = cmd_ctx->lsm->cmd_type;
-       cmd_ctx->llm->pid = cmd_ctx->lsm->domain.attr.pid;
-       cmd_ctx->llm->cmd_header_size = cmd_header_len;
-       cmd_ctx->llm->data_size = payload_len;
-       cmd_ctx->lttng_msg_size = total_msg_size;
-
-       /* Copy command header */
-       if (cmd_header_len) {
-               memcpy(((uint8_t *) cmd_ctx->llm) + cmd_header_offset, cmd_header_buf,
-                       cmd_header_len);
-       }
-
-       /* Copy payload */
-       if (payload_len) {
-               memcpy(((uint8_t *) cmd_ctx->llm) + payload_offset, payload_buf,
-                       payload_len);
-       }
-
-end:
-       return ret;
-}
-
-/*
- * Version of setup_lttng_msg() without command header.
- */
-static int setup_lttng_msg_no_cmd_header(struct command_ctx *cmd_ctx,
-       void *payload_buf, size_t payload_len)
-{
-       return setup_lttng_msg(cmd_ctx, payload_buf, payload_len, NULL, 0);
-}
 /*
  * Update the kernel poll set of all channel fd available over all tracing
  * session. Add the wakeup pipe at the end of the set.
@@ -939,7 +775,7 @@ static void signal_consumer_condition(struct consumer_data *data, int state)
 /*
  * This thread manage the consumer error sent back to the session daemon.
  */
-static void *thread_manage_consumer(void *data)
+void *thread_manage_consumer(void *data)
 {
        int sock = -1, i, ret, pollfd, err = -1, should_quit = 0;
        uint32_t revents, nb_fd;
@@ -2088,371 +1924,6 @@ error_testpoint:
        return NULL;
 }
 
-/*
- * Start the thread_manage_consumer. This must be done after a lttng-consumerd
- * exec or it will fails.
- */
-static int spawn_consumer_thread(struct consumer_data *consumer_data)
-{
-       int ret, clock_ret;
-       struct timespec timeout;
-
-       /*
-        * Make sure we set the readiness flag to 0 because we are NOT ready.
-        * This access to consumer_thread_is_ready does not need to be
-        * protected by consumer_data.cond_mutex (yet) since the consumer
-        * management thread has not been started at this point.
-        */
-       consumer_data->consumer_thread_is_ready = 0;
-
-       /* Setup pthread condition */
-       ret = pthread_condattr_init(&consumer_data->condattr);
-       if (ret) {
-               errno = ret;
-               PERROR("pthread_condattr_init consumer data");
-               goto error;
-       }
-
-       /*
-        * Set the monotonic clock in order to make sure we DO NOT jump in time
-        * between the clock_gettime() call and the timedwait call. See bug #324
-        * for a more details and how we noticed it.
-        */
-       ret = pthread_condattr_setclock(&consumer_data->condattr, CLOCK_MONOTONIC);
-       if (ret) {
-               errno = ret;
-               PERROR("pthread_condattr_setclock consumer data");
-               goto error;
-       }
-
-       ret = pthread_cond_init(&consumer_data->cond, &consumer_data->condattr);
-       if (ret) {
-               errno = ret;
-               PERROR("pthread_cond_init consumer data");
-               goto error;
-       }
-
-       ret = pthread_create(&consumer_data->thread, default_pthread_attr(),
-                       thread_manage_consumer, consumer_data);
-       if (ret) {
-               errno = ret;
-               PERROR("pthread_create consumer");
-               ret = -1;
-               goto error;
-       }
-
-       /* We are about to wait on a pthread condition */
-       pthread_mutex_lock(&consumer_data->cond_mutex);
-
-       /* Get time for sem_timedwait absolute timeout */
-       clock_ret = lttng_clock_gettime(CLOCK_MONOTONIC, &timeout);
-       /*
-        * Set the timeout for the condition timed wait even if the clock gettime
-        * call fails since we might loop on that call and we want to avoid to
-        * increment the timeout too many times.
-        */
-       timeout.tv_sec += DEFAULT_SEM_WAIT_TIMEOUT;
-
-       /*
-        * The following loop COULD be skipped in some conditions so this is why we
-        * set ret to 0 in order to make sure at least one round of the loop is
-        * done.
-        */
-       ret = 0;
-
-       /*
-        * Loop until the condition is reached or when a timeout is reached. Note
-        * that the pthread_cond_timedwait(P) man page specifies that EINTR can NOT
-        * be returned but the pthread_cond(3), from the glibc-doc, says that it is
-        * possible. This loop does not take any chances and works with both of
-        * them.
-        */
-       while (!consumer_data->consumer_thread_is_ready && ret != ETIMEDOUT) {
-               if (clock_ret < 0) {
-                       PERROR("clock_gettime spawn consumer");
-                       /* Infinite wait for the consumerd thread to be ready */
-                       ret = pthread_cond_wait(&consumer_data->cond,
-                                       &consumer_data->cond_mutex);
-               } else {
-                       ret = pthread_cond_timedwait(&consumer_data->cond,
-                                       &consumer_data->cond_mutex, &timeout);
-               }
-       }
-
-       /* Release the pthread condition */
-       pthread_mutex_unlock(&consumer_data->cond_mutex);
-
-       if (ret != 0) {
-               errno = ret;
-               if (ret == ETIMEDOUT) {
-                       int pth_ret;
-
-                       /*
-                        * Call has timed out so we kill the kconsumerd_thread and return
-                        * an error.
-                        */
-                       ERR("Condition timed out. The consumer thread was never ready."
-                                       " Killing it");
-                       pth_ret = pthread_cancel(consumer_data->thread);
-                       if (pth_ret < 0) {
-                               PERROR("pthread_cancel consumer thread");
-                       }
-               } else {
-                       PERROR("pthread_cond_wait failed consumer thread");
-               }
-               /* Caller is expecting a negative value on failure. */
-               ret = -1;
-               goto error;
-       }
-
-       pthread_mutex_lock(&consumer_data->pid_mutex);
-       if (consumer_data->pid == 0) {
-               ERR("Consumerd did not start");
-               pthread_mutex_unlock(&consumer_data->pid_mutex);
-               goto error;
-       }
-       pthread_mutex_unlock(&consumer_data->pid_mutex);
-
-       return 0;
-
-error:
-       return ret;
-}
-
-/*
- * Join consumer thread
- */
-static int join_consumer_thread(struct consumer_data *consumer_data)
-{
-       void *status;
-
-       /* Consumer pid must be a real one. */
-       if (consumer_data->pid > 0) {
-               int ret;
-               ret = kill(consumer_data->pid, SIGTERM);
-               if (ret) {
-                       PERROR("Error killing consumer daemon");
-                       return ret;
-               }
-               return pthread_join(consumer_data->thread, &status);
-       } else {
-               return 0;
-       }
-}
-
-/*
- * Fork and exec a consumer daemon (consumerd).
- *
- * Return pid if successful else -1.
- */
-static pid_t spawn_consumerd(struct consumer_data *consumer_data)
-{
-       int ret;
-       pid_t pid;
-       const char *consumer_to_use;
-       const char *verbosity;
-       struct stat st;
-
-       DBG("Spawning consumerd");
-
-       pid = fork();
-       if (pid == 0) {
-               /*
-                * Exec consumerd.
-                */
-               if (config.verbose_consumer) {
-                       verbosity = "--verbose";
-               } else if (lttng_opt_quiet) {
-                       verbosity = "--quiet";
-               } else {
-                       verbosity = "";
-               }
-
-               switch (consumer_data->type) {
-               case LTTNG_CONSUMER_KERNEL:
-                       /*
-                        * Find out which consumerd to execute. We will first try the
-                        * 64-bit path, then the sessiond's installation directory, and
-                        * fallback on the 32-bit one,
-                        */
-                       DBG3("Looking for a kernel consumer at these locations:");
-                       DBG3("  1) %s", config.consumerd64_bin_path.value ? : "NULL");
-                       DBG3("  2) %s/%s", INSTALL_BIN_PATH, DEFAULT_CONSUMERD_FILE);
-                       DBG3("  3) %s", config.consumerd32_bin_path.value ? : "NULL");
-                       if (stat(config.consumerd64_bin_path.value, &st) == 0) {
-                               DBG3("Found location #1");
-                               consumer_to_use = config.consumerd64_bin_path.value;
-                       } else if (stat(INSTALL_BIN_PATH "/" DEFAULT_CONSUMERD_FILE, &st) == 0) {
-                               DBG3("Found location #2");
-                               consumer_to_use = INSTALL_BIN_PATH "/" DEFAULT_CONSUMERD_FILE;
-                       } else if (config.consumerd32_bin_path.value &&
-                                       stat(config.consumerd32_bin_path.value, &st) == 0) {
-                               DBG3("Found location #3");
-                               consumer_to_use = config.consumerd32_bin_path.value;
-                       } else {
-                               DBG("Could not find any valid consumerd executable");
-                               ret = -EINVAL;
-                               goto error;
-                       }
-                       DBG("Using kernel consumer at: %s",  consumer_to_use);
-                       (void) execl(consumer_to_use,
-                               "lttng-consumerd", verbosity, "-k",
-                               "--consumerd-cmd-sock", consumer_data->cmd_unix_sock_path,
-                               "--consumerd-err-sock", consumer_data->err_unix_sock_path,
-                               "--group", config.tracing_group_name.value,
-                               NULL);
-                       break;
-               case LTTNG_CONSUMER64_UST:
-               {
-                       if (config.consumerd64_lib_dir.value) {
-                               char *tmp;
-                               size_t tmplen;
-                               char *tmpnew;
-
-                               tmp = lttng_secure_getenv("LD_LIBRARY_PATH");
-                               if (!tmp) {
-                                       tmp = "";
-                               }
-                               tmplen = strlen(config.consumerd64_lib_dir.value) + 1 /* : */ + strlen(tmp);
-                               tmpnew = zmalloc(tmplen + 1 /* \0 */);
-                               if (!tmpnew) {
-                                       ret = -ENOMEM;
-                                       goto error;
-                               }
-                               strcat(tmpnew, config.consumerd64_lib_dir.value);
-                               if (tmp[0] != '\0') {
-                                       strcat(tmpnew, ":");
-                                       strcat(tmpnew, tmp);
-                               }
-                               ret = setenv("LD_LIBRARY_PATH", tmpnew, 1);
-                               free(tmpnew);
-                               if (ret) {
-                                       ret = -errno;
-                                       goto error;
-                               }
-                       }
-                       DBG("Using 64-bit UST consumer at: %s",  config.consumerd64_bin_path.value);
-                       (void) execl(config.consumerd64_bin_path.value, "lttng-consumerd", verbosity, "-u",
-                                       "--consumerd-cmd-sock", consumer_data->cmd_unix_sock_path,
-                                       "--consumerd-err-sock", consumer_data->err_unix_sock_path,
-                                       "--group", config.tracing_group_name.value,
-                                       NULL);
-                       break;
-               }
-               case LTTNG_CONSUMER32_UST:
-               {
-                       if (config.consumerd32_lib_dir.value) {
-                               char *tmp;
-                               size_t tmplen;
-                               char *tmpnew;
-
-                               tmp = lttng_secure_getenv("LD_LIBRARY_PATH");
-                               if (!tmp) {
-                                       tmp = "";
-                               }
-                               tmplen = strlen(config.consumerd32_lib_dir.value) + 1 /* : */ + strlen(tmp);
-                               tmpnew = zmalloc(tmplen + 1 /* \0 */);
-                               if (!tmpnew) {
-                                       ret = -ENOMEM;
-                                       goto error;
-                               }
-                               strcat(tmpnew, config.consumerd32_lib_dir.value);
-                               if (tmp[0] != '\0') {
-                                       strcat(tmpnew, ":");
-                                       strcat(tmpnew, tmp);
-                               }
-                               ret = setenv("LD_LIBRARY_PATH", tmpnew, 1);
-                               free(tmpnew);
-                               if (ret) {
-                                       ret = -errno;
-                                       goto error;
-                               }
-                       }
-                       DBG("Using 32-bit UST consumer at: %s",  config.consumerd32_bin_path.value);
-                       (void) execl(config.consumerd32_bin_path.value, "lttng-consumerd", verbosity, "-u",
-                                       "--consumerd-cmd-sock", consumer_data->cmd_unix_sock_path,
-                                       "--consumerd-err-sock", consumer_data->err_unix_sock_path,
-                                       "--group", config.tracing_group_name.value,
-                                       NULL);
-                       break;
-               }
-               default:
-                       ERR("unknown consumer type");
-                       errno = 0;
-               }
-               if (errno != 0) {
-                       PERROR("Consumer execl()");
-               }
-               /* Reaching this point, we got a failure on our execl(). */
-               exit(EXIT_FAILURE);
-       } else if (pid > 0) {
-               ret = pid;
-       } else {
-               PERROR("start consumer fork");
-               ret = -errno;
-       }
-error:
-       return ret;
-}
-
-/*
- * Spawn the consumerd daemon and session daemon thread.
- */
-static int start_consumerd(struct consumer_data *consumer_data)
-{
-       int ret;
-
-       /*
-        * Set the listen() state on the socket since there is a possible race
-        * between the exec() of the consumer daemon and this call if place in the
-        * consumer thread. See bug #366 for more details.
-        */
-       ret = lttcomm_listen_unix_sock(consumer_data->err_sock);
-       if (ret < 0) {
-               goto error;
-       }
-
-       pthread_mutex_lock(&consumer_data->pid_mutex);
-       if (consumer_data->pid != 0) {
-               pthread_mutex_unlock(&consumer_data->pid_mutex);
-               goto end;
-       }
-
-       ret = spawn_consumerd(consumer_data);
-       if (ret < 0) {
-               ERR("Spawning consumerd failed");
-               pthread_mutex_unlock(&consumer_data->pid_mutex);
-               goto error;
-       }
-
-       /* Setting up the consumer_data pid */
-       consumer_data->pid = ret;
-       DBG2("Consumer pid %d", consumer_data->pid);
-       pthread_mutex_unlock(&consumer_data->pid_mutex);
-
-       DBG2("Spawning consumer control thread");
-       ret = spawn_consumer_thread(consumer_data);
-       if (ret < 0) {
-               ERR("Fatal error spawning consumer control thread");
-               goto error;
-       }
-
-end:
-       return 0;
-
-error:
-       /* Cleanup already created sockets on error. */
-       if (consumer_data->err_sock >= 0) {
-               int err;
-
-               err = close(consumer_data->err_sock);
-               if (err < 0) {
-                       PERROR("close consumer data error socket");
-               }
-       }
-       return ret;
-}
-
 /*
  * Setup necessary data for kernel tracer action.
  */
@@ -2528,2009 +1999,20 @@ error:
        }
 }
 
-
-/*
- * Copy consumer output from the tracing session to the domain session. The
- * function also applies the right modification on a per domain basis for the
- * trace files destination directory.
- *
- * Should *NOT* be called with RCU read-side lock held.
- */
-static int copy_session_consumer(int domain, struct ltt_session *session)
+static int string_match(const char *str1, const char *str2)
 {
-       int ret;
-       const char *dir_name;
-       struct consumer_output *consumer;
-
-       assert(session);
-       assert(session->consumer);
-
-       switch (domain) {
-       case LTTNG_DOMAIN_KERNEL:
-               DBG3("Copying tracing session consumer output in kernel session");
-               /*
-                * XXX: We should audit the session creation and what this function
-                * does "extra" in order to avoid a destroy since this function is used
-                * in the domain session creation (kernel and ust) only. Same for UST
-                * domain.
-                */
-               if (session->kernel_session->consumer) {
-                       consumer_output_put(session->kernel_session->consumer);
-               }
-               session->kernel_session->consumer =
-                       consumer_copy_output(session->consumer);
-               /* Ease our life a bit for the next part */
-               consumer = session->kernel_session->consumer;
-               dir_name = DEFAULT_KERNEL_TRACE_DIR;
-               break;
-       case LTTNG_DOMAIN_JUL:
-       case LTTNG_DOMAIN_LOG4J:
-       case LTTNG_DOMAIN_PYTHON:
-       case LTTNG_DOMAIN_UST:
-               DBG3("Copying tracing session consumer output in UST session");
-               if (session->ust_session->consumer) {
-                       consumer_output_put(session->ust_session->consumer);
-               }
-               session->ust_session->consumer =
-                       consumer_copy_output(session->consumer);
-               /* Ease our life a bit for the next part */
-               consumer = session->ust_session->consumer;
-               dir_name = DEFAULT_UST_TRACE_DIR;
-               break;
-       default:
-               ret = LTTNG_ERR_UNKNOWN_DOMAIN;
-               goto error;
-       }
-
-       /* Append correct directory to subdir */
-       strncat(consumer->subdir, dir_name,
-                       sizeof(consumer->subdir) - strlen(consumer->subdir) - 1);
-       DBG3("Copy session consumer subdir %s", consumer->subdir);
-
-       ret = LTTNG_OK;
-
-error:
-       return ret;
+       return (str1 && str2) && !strcmp(str1, str2);
 }
 
 /*
- * Create an UST session and add it to the session ust list.
+ * Take an option from the getopt output and set it in the right variable to be
+ * used later.
  *
- * Should *NOT* be called with RCU read-side lock held.
+ * Return 0 on success else a negative value.
  */
-static int create_ust_session(struct ltt_session *session,
-               struct lttng_domain *domain)
+static int set_option(int opt, const char *arg, const char *optname)
 {
-       int ret;
-       struct ltt_ust_session *lus = NULL;
-
-       assert(session);
-       assert(domain);
-       assert(session->consumer);
-
-       switch (domain->type) {
-       case LTTNG_DOMAIN_JUL:
-       case LTTNG_DOMAIN_LOG4J:
-       case LTTNG_DOMAIN_PYTHON:
-       case LTTNG_DOMAIN_UST:
-               break;
-       default:
-               ERR("Unknown UST domain on create session %d", domain->type);
-               ret = LTTNG_ERR_UNKNOWN_DOMAIN;
-               goto error;
-       }
-
-       DBG("Creating UST session");
-
-       lus = trace_ust_create_session(session->id);
-       if (lus == NULL) {
-               ret = LTTNG_ERR_UST_SESS_FAIL;
-               goto error;
-       }
-
-       lus->uid = session->uid;
-       lus->gid = session->gid;
-       lus->output_traces = session->output_traces;
-       lus->snapshot_mode = session->snapshot_mode;
-       lus->live_timer_interval = session->live_timer;
-       session->ust_session = lus;
-       if (session->shm_path[0]) {
-               strncpy(lus->root_shm_path, session->shm_path,
-                       sizeof(lus->root_shm_path));
-               lus->root_shm_path[sizeof(lus->root_shm_path) - 1] = '\0';
-               strncpy(lus->shm_path, session->shm_path,
-                       sizeof(lus->shm_path));
-               lus->shm_path[sizeof(lus->shm_path) - 1] = '\0';
-               strncat(lus->shm_path, "/ust",
-                       sizeof(lus->shm_path) - strlen(lus->shm_path) - 1);
-       }
-       /* Copy session output to the newly created UST session */
-       ret = copy_session_consumer(domain->type, session);
-       if (ret != LTTNG_OK) {
-               goto error;
-       }
-
-       return LTTNG_OK;
-
-error:
-       free(lus);
-       session->ust_session = NULL;
-       return ret;
-}
-
-/*
- * Create a kernel tracer session then create the default channel.
- */
-static int create_kernel_session(struct ltt_session *session)
-{
-       int ret;
-
-       DBG("Creating kernel session");
-
-       ret = kernel_create_session(session, kernel_tracer_fd);
-       if (ret < 0) {
-               ret = LTTNG_ERR_KERN_SESS_FAIL;
-               goto error;
-       }
-
-       /* Code flow safety */
-       assert(session->kernel_session);
-
-       /* Copy session output to the newly created Kernel session */
-       ret = copy_session_consumer(LTTNG_DOMAIN_KERNEL, session);
-       if (ret != LTTNG_OK) {
-               goto error;
-       }
-
-       session->kernel_session->uid = session->uid;
-       session->kernel_session->gid = session->gid;
-       session->kernel_session->output_traces = session->output_traces;
-       session->kernel_session->snapshot_mode = session->snapshot_mode;
-
-       return LTTNG_OK;
-
-error:
-       trace_kernel_destroy_session(session->kernel_session);
-       session->kernel_session = NULL;
-       return ret;
-}
-
-/*
- * Count number of session permitted by uid/gid.
- */
-static unsigned int lttng_sessions_count(uid_t uid, gid_t gid)
-{
-       unsigned int i = 0;
-       struct ltt_session *session;
-       const struct ltt_session_list *session_list = session_get_list();
-
-       DBG("Counting number of available session for UID %d GID %d",
-                       uid, gid);
-       cds_list_for_each_entry(session, &session_list->head, list) {
-               if (!session_get(session)) {
-                       continue;
-               }
-               session_lock(session);
-               /* Only count the sessions the user can control. */
-               if (session_access_ok(session, uid, gid) &&
-                               !session->destroyed) {
-                       i++;
-               }
-               session_unlock(session);
-               session_put(session);
-       }
-       return i;
-}
-
-static int receive_userspace_probe(struct command_ctx *cmd_ctx, int sock,
-               int *sock_error, struct lttng_event *event)
-{
-       int fd, ret;
-       struct lttng_userspace_probe_location *probe_location;
-       const struct lttng_userspace_probe_location_lookup_method *lookup = NULL;
-       struct lttng_dynamic_buffer probe_location_buffer;
-       struct lttng_buffer_view buffer_view;
-
-       /*
-        * Create a buffer to store the serialized version of the probe
-        * location.
-        */
-       lttng_dynamic_buffer_init(&probe_location_buffer);
-       ret = lttng_dynamic_buffer_set_size(&probe_location_buffer,
-                       cmd_ctx->lsm->u.enable.userspace_probe_location_len);
-       if (ret) {
-               ret = LTTNG_ERR_NOMEM;
-               goto error;
-       }
-
-       /*
-        * Receive the probe location.
-        */
-       ret = lttcomm_recv_unix_sock(sock, probe_location_buffer.data,
-                       probe_location_buffer.size);
-       if (ret <= 0) {
-               DBG("Nothing recv() from client var len data... continuing");
-               *sock_error = 1;
-               lttng_dynamic_buffer_reset(&probe_location_buffer);
-               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
-               goto error;
-       }
-
-       buffer_view = lttng_buffer_view_from_dynamic_buffer(
-                       &probe_location_buffer, 0, probe_location_buffer.size);
-
-       /*
-        * Extract the probe location from the serialized version.
-        */
-       ret = lttng_userspace_probe_location_create_from_buffer(
-                               &buffer_view, &probe_location);
-       if (ret < 0) {
-               WARN("Failed to create a userspace probe location from the received buffer");
-               lttng_dynamic_buffer_reset( &probe_location_buffer);
-               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
-               goto error;
-       }
-
-       /*
-        * Receive the file descriptor to the target binary from the client.
-        */
-       DBG("Receiving userspace probe target FD from client ...");
-       ret = lttcomm_recv_fds_unix_sock(sock, &fd, 1);
-       if (ret <= 0) {
-               DBG("Nothing recv() from client userspace probe fd... continuing");
-               *sock_error = 1;
-               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
-               goto error;
-       }
-
-       /*
-        * Set the file descriptor received from the client through the unix
-        * socket in the probe location.
-        */
-       lookup = lttng_userspace_probe_location_get_lookup_method(probe_location);
-       if (!lookup) {
-               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
-               goto error;
-       }
-
-       /*
-        * From the kernel tracer's perspective, all userspace probe event types
-        * are all the same: a file and an offset.
-        */
-       switch (lttng_userspace_probe_location_lookup_method_get_type(lookup)) {
-       case LTTNG_USERSPACE_PROBE_LOCATION_LOOKUP_METHOD_TYPE_FUNCTION_ELF:
-               ret = lttng_userspace_probe_location_function_set_binary_fd(
-                               probe_location, fd);
-               break;
-       case LTTNG_USERSPACE_PROBE_LOCATION_LOOKUP_METHOD_TYPE_TRACEPOINT_SDT:
-               ret = lttng_userspace_probe_location_tracepoint_set_binary_fd(
-                               probe_location, fd);
-               break;
-       default:
-               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
-               goto error;
-       }
-
-       if (ret) {
-               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
-               goto error;
-       }
-
-       /* Attach the probe location to the event. */
-       ret = lttng_event_set_userspace_probe_location(event, probe_location);
-       if (ret) {
-               ret = LTTNG_ERR_PROBE_LOCATION_INVAL;
-               goto error;
-       }
-
-       lttng_dynamic_buffer_reset(&probe_location_buffer);
-error:
-       return ret;
-}
-
-/*
- * Check if the current kernel tracer supports the session rotation feature.
- * Return 1 if it does, 0 otherwise.
- */
-static int check_rotate_compatible(void)
-{
-       int ret = 1;
-
-       if (kernel_tracer_version.major != 2 || kernel_tracer_version.minor < 11) {
-               DBG("Kernel tracer version is not compatible with the rotation feature");
-               ret = 0;
-       }
-
-       return ret;
-}
-
-/*
- * Process the command requested by the lttng client within the command
- * context structure. This function make sure that the return structure (llm)
- * is set and ready for transmission before returning.
- *
- * Return any error encountered or 0 for success.
- *
- * "sock" is only used for special-case var. len data.
- *
- * Should *NOT* be called with RCU read-side lock held.
- */
-static int process_client_msg(struct command_ctx *cmd_ctx, int sock,
-               int *sock_error)
-{
-       int ret = LTTNG_OK;
-       int need_tracing_session = 1;
-       int need_domain;
-
-       DBG("Processing client command %d", cmd_ctx->lsm->cmd_type);
-
-       assert(!rcu_read_ongoing());
-
-       *sock_error = 0;
-
-       switch (cmd_ctx->lsm->cmd_type) {
-       case LTTNG_CREATE_SESSION:
-       case LTTNG_CREATE_SESSION_SNAPSHOT:
-       case LTTNG_CREATE_SESSION_LIVE:
-       case LTTNG_DESTROY_SESSION:
-       case LTTNG_LIST_SESSIONS:
-       case LTTNG_LIST_DOMAINS:
-       case LTTNG_START_TRACE:
-       case LTTNG_STOP_TRACE:
-       case LTTNG_DATA_PENDING:
-       case LTTNG_SNAPSHOT_ADD_OUTPUT:
-       case LTTNG_SNAPSHOT_DEL_OUTPUT:
-       case LTTNG_SNAPSHOT_LIST_OUTPUT:
-       case LTTNG_SNAPSHOT_RECORD:
-       case LTTNG_SAVE_SESSION:
-       case LTTNG_SET_SESSION_SHM_PATH:
-       case LTTNG_REGENERATE_METADATA:
-       case LTTNG_REGENERATE_STATEDUMP:
-       case LTTNG_REGISTER_TRIGGER:
-       case LTTNG_UNREGISTER_TRIGGER:
-       case LTTNG_ROTATE_SESSION:
-       case LTTNG_ROTATION_GET_INFO:
-       case LTTNG_ROTATION_SET_SCHEDULE:
-       case LTTNG_SESSION_LIST_ROTATION_SCHEDULES:
-               need_domain = 0;
-               break;
-       default:
-               need_domain = 1;
-       }
-
-       if (config.no_kernel && need_domain
-                       && cmd_ctx->lsm->domain.type == LTTNG_DOMAIN_KERNEL) {
-               if (!is_root) {
-                       ret = LTTNG_ERR_NEED_ROOT_SESSIOND;
-               } else {
-                       ret = LTTNG_ERR_KERN_NA;
-               }
-               goto error;
-       }
-
-       /* Deny register consumer if we already have a spawned consumer. */
-       if (cmd_ctx->lsm->cmd_type == LTTNG_REGISTER_CONSUMER) {
-               pthread_mutex_lock(&kconsumer_data.pid_mutex);
-               if (kconsumer_data.pid > 0) {
-                       ret = LTTNG_ERR_KERN_CONSUMER_FAIL;
-                       pthread_mutex_unlock(&kconsumer_data.pid_mutex);
-                       goto error;
-               }
-               pthread_mutex_unlock(&kconsumer_data.pid_mutex);
-       }
-
-       /*
-        * Check for command that don't needs to allocate a returned payload. We do
-        * this here so we don't have to make the call for no payload at each
-        * command.
-        */
-       switch(cmd_ctx->lsm->cmd_type) {
-       case LTTNG_LIST_SESSIONS:
-       case LTTNG_LIST_TRACEPOINTS:
-       case LTTNG_LIST_TRACEPOINT_FIELDS:
-       case LTTNG_LIST_DOMAINS:
-       case LTTNG_LIST_CHANNELS:
-       case LTTNG_LIST_EVENTS:
-       case LTTNG_LIST_SYSCALLS:
-       case LTTNG_LIST_TRACKER_PIDS:
-       case LTTNG_DATA_PENDING:
-       case LTTNG_ROTATE_SESSION:
-       case LTTNG_ROTATION_GET_INFO:
-       case LTTNG_SESSION_LIST_ROTATION_SCHEDULES:
-               break;
-       default:
-               /* Setup lttng message with no payload */
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, NULL, 0);
-               if (ret < 0) {
-                       /* This label does not try to unlock the session */
-                       goto init_setup_error;
-               }
-       }
-
-       /* Commands that DO NOT need a session. */
-       switch (cmd_ctx->lsm->cmd_type) {
-       case LTTNG_CREATE_SESSION:
-       case LTTNG_CREATE_SESSION_SNAPSHOT:
-       case LTTNG_CREATE_SESSION_LIVE:
-       case LTTNG_LIST_SESSIONS:
-       case LTTNG_LIST_TRACEPOINTS:
-       case LTTNG_LIST_SYSCALLS:
-       case LTTNG_LIST_TRACEPOINT_FIELDS:
-       case LTTNG_SAVE_SESSION:
-       case LTTNG_REGISTER_TRIGGER:
-       case LTTNG_UNREGISTER_TRIGGER:
-               need_tracing_session = 0;
-               break;
-       default:
-               DBG("Getting session %s by name", cmd_ctx->lsm->session.name);
-               /*
-                * We keep the session list lock across _all_ commands
-                * for now, because the per-session lock does not
-                * handle teardown properly.
-                */
-               session_lock_list();
-               cmd_ctx->session = session_find_by_name(cmd_ctx->lsm->session.name);
-               if (cmd_ctx->session == NULL) {
-                       ret = LTTNG_ERR_SESS_NOT_FOUND;
-                       goto error;
-               } else {
-                       /* Acquire lock for the session */
-                       session_lock(cmd_ctx->session);
-               }
-               break;
-       }
-
-       /*
-        * Commands that need a valid session but should NOT create one if none
-        * exists. Instead of creating one and destroying it when the command is
-        * handled, process that right before so we save some round trip in useless
-        * code path.
-        */
-       switch (cmd_ctx->lsm->cmd_type) {
-       case LTTNG_DISABLE_CHANNEL:
-       case LTTNG_DISABLE_EVENT:
-               switch (cmd_ctx->lsm->domain.type) {
-               case LTTNG_DOMAIN_KERNEL:
-                       if (!cmd_ctx->session->kernel_session) {
-                               ret = LTTNG_ERR_NO_CHANNEL;
-                               goto error;
-                       }
-                       break;
-               case LTTNG_DOMAIN_JUL:
-               case LTTNG_DOMAIN_LOG4J:
-               case LTTNG_DOMAIN_PYTHON:
-               case LTTNG_DOMAIN_UST:
-                       if (!cmd_ctx->session->ust_session) {
-                               ret = LTTNG_ERR_NO_CHANNEL;
-                               goto error;
-                       }
-                       break;
-               default:
-                       ret = LTTNG_ERR_UNKNOWN_DOMAIN;
-                       goto error;
-               }
-       default:
-               break;
-       }
-
-       if (!need_domain) {
-               goto skip_domain;
-       }
-
-       /*
-        * Check domain type for specific "pre-action".
-        */
-       switch (cmd_ctx->lsm->domain.type) {
-       case LTTNG_DOMAIN_KERNEL:
-               if (!is_root) {
-                       ret = LTTNG_ERR_NEED_ROOT_SESSIOND;
-                       goto error;
-               }
-
-               /* Kernel tracer check */
-               if (kernel_tracer_fd == -1) {
-                       /* Basically, load kernel tracer modules */
-                       ret = init_kernel_tracer();
-                       if (ret != 0) {
-                               goto error;
-                       }
-               }
-
-               /* Consumer is in an ERROR state. Report back to client */
-               if (uatomic_read(&kernel_consumerd_state) == CONSUMER_ERROR) {
-                       ret = LTTNG_ERR_NO_KERNCONSUMERD;
-                       goto error;
-               }
-
-               /* Need a session for kernel command */
-               if (need_tracing_session) {
-                       if (cmd_ctx->session->kernel_session == NULL) {
-                               ret = create_kernel_session(cmd_ctx->session);
-                               if (ret < 0) {
-                                       ret = LTTNG_ERR_KERN_SESS_FAIL;
-                                       goto error;
-                               }
-                       }
-
-                       /* Start the kernel consumer daemon */
-                       pthread_mutex_lock(&kconsumer_data.pid_mutex);
-                       if (kconsumer_data.pid == 0 &&
-                                       cmd_ctx->lsm->cmd_type != LTTNG_REGISTER_CONSUMER) {
-                               pthread_mutex_unlock(&kconsumer_data.pid_mutex);
-                               ret = start_consumerd(&kconsumer_data);
-                               if (ret < 0) {
-                                       ret = LTTNG_ERR_KERN_CONSUMER_FAIL;
-                                       goto error;
-                               }
-                               uatomic_set(&kernel_consumerd_state, CONSUMER_STARTED);
-                       } else {
-                               pthread_mutex_unlock(&kconsumer_data.pid_mutex);
-                       }
-
-                       /*
-                        * The consumer was just spawned so we need to add the socket to
-                        * the consumer output of the session if exist.
-                        */
-                       ret = consumer_create_socket(&kconsumer_data,
-                                       cmd_ctx->session->kernel_session->consumer);
-                       if (ret < 0) {
-                               goto error;
-                       }
-               }
-
-               break;
-       case LTTNG_DOMAIN_JUL:
-       case LTTNG_DOMAIN_LOG4J:
-       case LTTNG_DOMAIN_PYTHON:
-       case LTTNG_DOMAIN_UST:
-       {
-               if (!ust_app_supported()) {
-                       ret = LTTNG_ERR_NO_UST;
-                       goto error;
-               }
-               /* Consumer is in an ERROR state. Report back to client */
-               if (uatomic_read(&ust_consumerd_state) == CONSUMER_ERROR) {
-                       ret = LTTNG_ERR_NO_USTCONSUMERD;
-                       goto error;
-               }
-
-               if (need_tracing_session) {
-                       /* Create UST session if none exist. */
-                       if (cmd_ctx->session->ust_session == NULL) {
-                               ret = create_ust_session(cmd_ctx->session,
-                                               &cmd_ctx->lsm->domain);
-                               if (ret != LTTNG_OK) {
-                                       goto error;
-                               }
-                       }
-
-                       /* Start the UST consumer daemons */
-                       /* 64-bit */
-                       pthread_mutex_lock(&ustconsumer64_data.pid_mutex);
-                       if (config.consumerd64_bin_path.value &&
-                                       ustconsumer64_data.pid == 0 &&
-                                       cmd_ctx->lsm->cmd_type != LTTNG_REGISTER_CONSUMER) {
-                               pthread_mutex_unlock(&ustconsumer64_data.pid_mutex);
-                               ret = start_consumerd(&ustconsumer64_data);
-                               if (ret < 0) {
-                                       ret = LTTNG_ERR_UST_CONSUMER64_FAIL;
-                                       uatomic_set(&ust_consumerd64_fd, -EINVAL);
-                                       goto error;
-                               }
-
-                               uatomic_set(&ust_consumerd64_fd, ustconsumer64_data.cmd_sock);
-                               uatomic_set(&ust_consumerd_state, CONSUMER_STARTED);
-                       } else {
-                               pthread_mutex_unlock(&ustconsumer64_data.pid_mutex);
-                       }
-
-                       /*
-                        * Setup socket for consumer 64 bit. No need for atomic access
-                        * since it was set above and can ONLY be set in this thread.
-                        */
-                       ret = consumer_create_socket(&ustconsumer64_data,
-                                       cmd_ctx->session->ust_session->consumer);
-                       if (ret < 0) {
-                               goto error;
-                       }
-
-                       /* 32-bit */
-                       pthread_mutex_lock(&ustconsumer32_data.pid_mutex);
-                       if (config.consumerd32_bin_path.value &&
-                                       ustconsumer32_data.pid == 0 &&
-                                       cmd_ctx->lsm->cmd_type != LTTNG_REGISTER_CONSUMER) {
-                               pthread_mutex_unlock(&ustconsumer32_data.pid_mutex);
-                               ret = start_consumerd(&ustconsumer32_data);
-                               if (ret < 0) {
-                                       ret = LTTNG_ERR_UST_CONSUMER32_FAIL;
-                                       uatomic_set(&ust_consumerd32_fd, -EINVAL);
-                                       goto error;
-                               }
-
-                               uatomic_set(&ust_consumerd32_fd, ustconsumer32_data.cmd_sock);
-                               uatomic_set(&ust_consumerd_state, CONSUMER_STARTED);
-                       } else {
-                               pthread_mutex_unlock(&ustconsumer32_data.pid_mutex);
-                       }
-
-                       /*
-                        * Setup socket for consumer 32 bit. No need for atomic access
-                        * since it was set above and can ONLY be set in this thread.
-                        */
-                       ret = consumer_create_socket(&ustconsumer32_data,
-                                       cmd_ctx->session->ust_session->consumer);
-                       if (ret < 0) {
-                               goto error;
-                       }
-               }
-               break;
-       }
-       default:
-               break;
-       }
-skip_domain:
-
-       /* Validate consumer daemon state when start/stop trace command */
-       if (cmd_ctx->lsm->cmd_type == LTTNG_START_TRACE ||
-                       cmd_ctx->lsm->cmd_type == LTTNG_STOP_TRACE) {
-               switch (cmd_ctx->lsm->domain.type) {
-               case LTTNG_DOMAIN_NONE:
-                       break;
-               case LTTNG_DOMAIN_JUL:
-               case LTTNG_DOMAIN_LOG4J:
-               case LTTNG_DOMAIN_PYTHON:
-               case LTTNG_DOMAIN_UST:
-                       if (uatomic_read(&ust_consumerd_state) != CONSUMER_STARTED) {
-                               ret = LTTNG_ERR_NO_USTCONSUMERD;
-                               goto error;
-                       }
-                       break;
-               case LTTNG_DOMAIN_KERNEL:
-                       if (uatomic_read(&kernel_consumerd_state) != CONSUMER_STARTED) {
-                               ret = LTTNG_ERR_NO_KERNCONSUMERD;
-                               goto error;
-                       }
-                       break;
-               default:
-                       ret = LTTNG_ERR_UNKNOWN_DOMAIN;
-                       goto error;
-               }
-       }
-
-       /*
-        * Check that the UID or GID match that of the tracing session.
-        * The root user can interact with all sessions.
-        */
-       if (need_tracing_session) {
-               if (!session_access_ok(cmd_ctx->session,
-                               LTTNG_SOCK_GET_UID_CRED(&cmd_ctx->creds),
-                               LTTNG_SOCK_GET_GID_CRED(&cmd_ctx->creds)) ||
-                               cmd_ctx->session->destroyed) {
-                       ret = LTTNG_ERR_EPERM;
-                       goto error;
-               }
-       }
-
-       /*
-        * Send relayd information to consumer as soon as we have a domain and a
-        * session defined.
-        */
-       if (cmd_ctx->session && need_domain) {
-               /*
-                * Setup relayd if not done yet. If the relayd information was already
-                * sent to the consumer, this call will gracefully return.
-                */
-               ret = cmd_setup_relayd(cmd_ctx->session);
-               if (ret != LTTNG_OK) {
-                       goto error;
-               }
-       }
-
-       /* Process by command type */
-       switch (cmd_ctx->lsm->cmd_type) {
-       case LTTNG_ADD_CONTEXT:
-       {
-               /*
-                * An LTTNG_ADD_CONTEXT command might have a supplementary
-                * payload if the context being added is an application context.
-                */
-               if (cmd_ctx->lsm->u.context.ctx.ctx ==
-                               LTTNG_EVENT_CONTEXT_APP_CONTEXT) {
-                       char *provider_name = NULL, *context_name = NULL;
-                       size_t provider_name_len =
-                                       cmd_ctx->lsm->u.context.provider_name_len;
-                       size_t context_name_len =
-                                       cmd_ctx->lsm->u.context.context_name_len;
-
-                       if (provider_name_len == 0 || context_name_len == 0) {
-                               /*
-                                * Application provider and context names MUST
-                                * be provided.
-                                */
-                               ret = -LTTNG_ERR_INVALID;
-                               goto error;
-                       }
-
-                       provider_name = zmalloc(provider_name_len + 1);
-                       if (!provider_name) {
-                               ret = -LTTNG_ERR_NOMEM;
-                               goto error;
-                       }
-                       cmd_ctx->lsm->u.context.ctx.u.app_ctx.provider_name =
-                                       provider_name;
-
-                       context_name = zmalloc(context_name_len + 1);
-                       if (!context_name) {
-                               ret = -LTTNG_ERR_NOMEM;
-                               goto error_add_context;
-                       }
-                       cmd_ctx->lsm->u.context.ctx.u.app_ctx.ctx_name =
-                                       context_name;
-
-                       ret = lttcomm_recv_unix_sock(sock, provider_name,
-                                       provider_name_len);
-                       if (ret < 0) {
-                               goto error_add_context;
-                       }
-
-                       ret = lttcomm_recv_unix_sock(sock, context_name,
-                                       context_name_len);
-                       if (ret < 0) {
-                               goto error_add_context;
-                       }
-               }
-
-               /*
-                * cmd_add_context assumes ownership of the provider and context
-                * names.
-                */
-               ret = cmd_add_context(cmd_ctx->session,
-                               cmd_ctx->lsm->domain.type,
-                               cmd_ctx->lsm->u.context.channel_name,
-                               &cmd_ctx->lsm->u.context.ctx,
-                               kernel_poll_pipe[1]);
-
-               cmd_ctx->lsm->u.context.ctx.u.app_ctx.provider_name = NULL;
-               cmd_ctx->lsm->u.context.ctx.u.app_ctx.ctx_name = NULL;
-error_add_context:
-               free(cmd_ctx->lsm->u.context.ctx.u.app_ctx.provider_name);
-               free(cmd_ctx->lsm->u.context.ctx.u.app_ctx.ctx_name);
-               if (ret < 0) {
-                       goto error;
-               }
-               break;
-       }
-       case LTTNG_DISABLE_CHANNEL:
-       {
-               ret = cmd_disable_channel(cmd_ctx->session, cmd_ctx->lsm->domain.type,
-                               cmd_ctx->lsm->u.disable.channel_name);
-               break;
-       }
-       case LTTNG_DISABLE_EVENT:
-       {
-
-               /*
-                * FIXME: handle filter; for now we just receive the filter's
-                * bytecode along with the filter expression which are sent by
-                * liblttng-ctl and discard them.
-                *
-                * This fixes an issue where the client may block while sending
-                * the filter payload and encounter an error because the session
-                * daemon closes the socket without ever handling this data.
-                */
-               size_t count = cmd_ctx->lsm->u.disable.expression_len +
-                       cmd_ctx->lsm->u.disable.bytecode_len;
-
-               if (count) {
-                       char data[LTTNG_FILTER_MAX_LEN];
-
-                       DBG("Discarding disable event command payload of size %zu", count);
-                       while (count) {
-                               ret = lttcomm_recv_unix_sock(sock, data,
-                                       count > sizeof(data) ? sizeof(data) : count);
-                               if (ret < 0) {
-                                       goto error;
-                               }
-
-                               count -= (size_t) ret;
-                       }
-               }
-               /* FIXME: passing packed structure to non-packed pointer */
-               ret = cmd_disable_event(cmd_ctx->session, cmd_ctx->lsm->domain.type,
-                               cmd_ctx->lsm->u.disable.channel_name,
-                               &cmd_ctx->lsm->u.disable.event);
-               break;
-       }
-       case LTTNG_ENABLE_CHANNEL:
-       {
-               cmd_ctx->lsm->u.channel.chan.attr.extended.ptr =
-                               (struct lttng_channel_extended *) &cmd_ctx->lsm->u.channel.extended;
-               ret = cmd_enable_channel(cmd_ctx->session, &cmd_ctx->lsm->domain,
-                               &cmd_ctx->lsm->u.channel.chan,
-                               kernel_poll_pipe[1]);
-               break;
-       }
-       case LTTNG_TRACK_PID:
-       {
-               ret = cmd_track_pid(cmd_ctx->session,
-                               cmd_ctx->lsm->domain.type,
-                               cmd_ctx->lsm->u.pid_tracker.pid);
-               break;
-       }
-       case LTTNG_UNTRACK_PID:
-       {
-               ret = cmd_untrack_pid(cmd_ctx->session,
-                               cmd_ctx->lsm->domain.type,
-                               cmd_ctx->lsm->u.pid_tracker.pid);
-               break;
-       }
-       case LTTNG_ENABLE_EVENT:
-       {
-               struct lttng_event *ev = NULL;
-               struct lttng_event_exclusion *exclusion = NULL;
-               struct lttng_filter_bytecode *bytecode = NULL;
-               char *filter_expression = NULL;
-
-               /* Handle exclusion events and receive it from the client. */
-               if (cmd_ctx->lsm->u.enable.exclusion_count > 0) {
-                       size_t count = cmd_ctx->lsm->u.enable.exclusion_count;
-
-                       exclusion = zmalloc(sizeof(struct lttng_event_exclusion) +
-                                       (count * LTTNG_SYMBOL_NAME_LEN));
-                       if (!exclusion) {
-                               ret = LTTNG_ERR_EXCLUSION_NOMEM;
-                               goto error;
-                       }
-
-                       DBG("Receiving var len exclusion event list from client ...");
-                       exclusion->count = count;
-                       ret = lttcomm_recv_unix_sock(sock, exclusion->names,
-                                       count * LTTNG_SYMBOL_NAME_LEN);
-                       if (ret <= 0) {
-                               DBG("Nothing recv() from client var len data... continuing");
-                               *sock_error = 1;
-                               free(exclusion);
-                               ret = LTTNG_ERR_EXCLUSION_INVAL;
-                               goto error;
-                       }
-               }
-
-               /* Get filter expression from client. */
-               if (cmd_ctx->lsm->u.enable.expression_len > 0) {
-                       size_t expression_len =
-                               cmd_ctx->lsm->u.enable.expression_len;
-
-                       if (expression_len > LTTNG_FILTER_MAX_LEN) {
-                               ret = LTTNG_ERR_FILTER_INVAL;
-                               free(exclusion);
-                               goto error;
-                       }
-
-                       filter_expression = zmalloc(expression_len);
-                       if (!filter_expression) {
-                               free(exclusion);
-                               ret = LTTNG_ERR_FILTER_NOMEM;
-                               goto error;
-                       }
-
-                       /* Receive var. len. data */
-                       DBG("Receiving var len filter's expression from client ...");
-                       ret = lttcomm_recv_unix_sock(sock, filter_expression,
-                               expression_len);
-                       if (ret <= 0) {
-                               DBG("Nothing recv() from client var len data... continuing");
-                               *sock_error = 1;
-                               free(filter_expression);
-                               free(exclusion);
-                               ret = LTTNG_ERR_FILTER_INVAL;
-                               goto error;
-                       }
-               }
-
-               /* Handle filter and get bytecode from client. */
-               if (cmd_ctx->lsm->u.enable.bytecode_len > 0) {
-                       size_t bytecode_len = cmd_ctx->lsm->u.enable.bytecode_len;
-
-                       if (bytecode_len > LTTNG_FILTER_MAX_LEN) {
-                               ret = LTTNG_ERR_FILTER_INVAL;
-                               free(filter_expression);
-                               free(exclusion);
-                               goto error;
-                       }
-
-                       bytecode = zmalloc(bytecode_len);
-                       if (!bytecode) {
-                               free(filter_expression);
-                               free(exclusion);
-                               ret = LTTNG_ERR_FILTER_NOMEM;
-                               goto error;
-                       }
-
-                       /* Receive var. len. data */
-                       DBG("Receiving var len filter's bytecode from client ...");
-                       ret = lttcomm_recv_unix_sock(sock, bytecode, bytecode_len);
-                       if (ret <= 0) {
-                               DBG("Nothing recv() from client var len data... continuing");
-                               *sock_error = 1;
-                               free(filter_expression);
-                               free(bytecode);
-                               free(exclusion);
-                               ret = LTTNG_ERR_FILTER_INVAL;
-                               goto error;
-                       }
-
-                       if ((bytecode->len + sizeof(*bytecode)) != bytecode_len) {
-                               free(filter_expression);
-                               free(bytecode);
-                               free(exclusion);
-                               ret = LTTNG_ERR_FILTER_INVAL;
-                               goto error;
-                       }
-               }
-
-               ev = lttng_event_copy(&cmd_ctx->lsm->u.enable.event);
-               if (!ev) {
-                       DBG("Failed to copy event: %s",
-                                       cmd_ctx->lsm->u.enable.event.name);
-                       free(filter_expression);
-                       free(bytecode);
-                       free(exclusion);
-                       ret = LTTNG_ERR_NOMEM;
-                       goto error;
-               }
-
-
-               if (cmd_ctx->lsm->u.enable.userspace_probe_location_len > 0) {
-                       /* Expect a userspace probe description. */
-                       ret = receive_userspace_probe(cmd_ctx, sock, sock_error, ev);
-                       if (ret) {
-                               free(filter_expression);
-                               free(bytecode);
-                               free(exclusion);
-                               lttng_event_destroy(ev);
-                               goto error;
-                       }
-               }
-
-               ret = cmd_enable_event(cmd_ctx->session, &cmd_ctx->lsm->domain,
-                               cmd_ctx->lsm->u.enable.channel_name,
-                               ev,
-                               filter_expression, bytecode, exclusion,
-                               kernel_poll_pipe[1]);
-               lttng_event_destroy(ev);
-               break;
-       }
-       case LTTNG_LIST_TRACEPOINTS:
-       {
-               struct lttng_event *events;
-               ssize_t nb_events;
-
-               session_lock_list();
-               nb_events = cmd_list_tracepoints(cmd_ctx->lsm->domain.type, &events);
-               session_unlock_list();
-               if (nb_events < 0) {
-                       /* Return value is a negative lttng_error_code. */
-                       ret = -nb_events;
-                       goto error;
-               }
-
-               /*
-                * Setup lttng message with payload size set to the event list size in
-                * bytes and then copy list into the llm payload.
-                */
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, events,
-                       sizeof(struct lttng_event) * nb_events);
-               free(events);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_LIST_TRACEPOINT_FIELDS:
-       {
-               struct lttng_event_field *fields;
-               ssize_t nb_fields;
-
-               session_lock_list();
-               nb_fields = cmd_list_tracepoint_fields(cmd_ctx->lsm->domain.type,
-                               &fields);
-               session_unlock_list();
-               if (nb_fields < 0) {
-                       /* Return value is a negative lttng_error_code. */
-                       ret = -nb_fields;
-                       goto error;
-               }
-
-               /*
-                * Setup lttng message with payload size set to the event list size in
-                * bytes and then copy list into the llm payload.
-                */
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, fields,
-                               sizeof(struct lttng_event_field) * nb_fields);
-               free(fields);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_LIST_SYSCALLS:
-       {
-               struct lttng_event *events;
-               ssize_t nb_events;
-
-               nb_events = cmd_list_syscalls(&events);
-               if (nb_events < 0) {
-                       /* Return value is a negative lttng_error_code. */
-                       ret = -nb_events;
-                       goto error;
-               }
-
-               /*
-                * Setup lttng message with payload size set to the event list size in
-                * bytes and then copy list into the llm payload.
-                */
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, events,
-                       sizeof(struct lttng_event) * nb_events);
-               free(events);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_LIST_TRACKER_PIDS:
-       {
-               int32_t *pids = NULL;
-               ssize_t nr_pids;
-
-               nr_pids = cmd_list_tracker_pids(cmd_ctx->session,
-                               cmd_ctx->lsm->domain.type, &pids);
-               if (nr_pids < 0) {
-                       /* Return value is a negative lttng_error_code. */
-                       ret = -nr_pids;
-                       goto error;
-               }
-
-               /*
-                * Setup lttng message with payload size set to the event list size in
-                * bytes and then copy list into the llm payload.
-                */
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, pids,
-                       sizeof(int32_t) * nr_pids);
-               free(pids);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_SET_CONSUMER_URI:
-       {
-               size_t nb_uri, len;
-               struct lttng_uri *uris;
-
-               nb_uri = cmd_ctx->lsm->u.uri.size;
-               len = nb_uri * sizeof(struct lttng_uri);
-
-               if (nb_uri == 0) {
-                       ret = LTTNG_ERR_INVALID;
-                       goto error;
-               }
-
-               uris = zmalloc(len);
-               if (uris == NULL) {
-                       ret = LTTNG_ERR_FATAL;
-                       goto error;
-               }
-
-               /* Receive variable len data */
-               DBG("Receiving %zu URI(s) from client ...", nb_uri);
-               ret = lttcomm_recv_unix_sock(sock, uris, len);
-               if (ret <= 0) {
-                       DBG("No URIs received from client... continuing");
-                       *sock_error = 1;
-                       ret = LTTNG_ERR_SESSION_FAIL;
-                       free(uris);
-                       goto error;
-               }
-
-               ret = cmd_set_consumer_uri(cmd_ctx->session, nb_uri, uris);
-               free(uris);
-               if (ret != LTTNG_OK) {
-                       goto error;
-               }
-
-
-               break;
-       }
-       case LTTNG_START_TRACE:
-       {
-               /*
-                * On the first start, if we have a kernel session and we have
-                * enabled time or size-based rotations, we have to make sure
-                * the kernel tracer supports it.
-                */
-               if (!cmd_ctx->session->has_been_started && \
-                               cmd_ctx->session->kernel_session && \
-                               (cmd_ctx->session->rotate_timer_period || \
-                                       cmd_ctx->session->rotate_size) && \
-                               !check_rotate_compatible()) {
-                       DBG("Kernel tracer version is not compatible with the rotation feature");
-                       ret = LTTNG_ERR_ROTATION_WRONG_VERSION;
-                       goto error;
-               }
-               ret = cmd_start_trace(cmd_ctx->session);
-               break;
-       }
-       case LTTNG_STOP_TRACE:
-       {
-               ret = cmd_stop_trace(cmd_ctx->session);
-               break;
-       }
-       case LTTNG_CREATE_SESSION:
-       {
-               size_t nb_uri, len;
-               struct lttng_uri *uris = NULL;
-
-               nb_uri = cmd_ctx->lsm->u.uri.size;
-               len = nb_uri * sizeof(struct lttng_uri);
-
-               if (nb_uri > 0) {
-                       uris = zmalloc(len);
-                       if (uris == NULL) {
-                               ret = LTTNG_ERR_FATAL;
-                               goto error;
-                       }
-
-                       /* Receive variable len data */
-                       DBG("Waiting for %zu URIs from client ...", nb_uri);
-                       ret = lttcomm_recv_unix_sock(sock, uris, len);
-                       if (ret <= 0) {
-                               DBG("No URIs received from client... continuing");
-                               *sock_error = 1;
-                               ret = LTTNG_ERR_SESSION_FAIL;
-                               free(uris);
-                               goto error;
-                       }
-
-                       if (nb_uri == 1 && uris[0].dtype != LTTNG_DST_PATH) {
-                               DBG("Creating session with ONE network URI is a bad call");
-                               ret = LTTNG_ERR_SESSION_FAIL;
-                               free(uris);
-                               goto error;
-                       }
-               }
-
-               ret = cmd_create_session_uri(cmd_ctx->lsm->session.name, uris, nb_uri,
-                       &cmd_ctx->creds, 0);
-
-               free(uris);
-
-               break;
-       }
-       case LTTNG_DESTROY_SESSION:
-       {
-               ret = cmd_destroy_session(cmd_ctx->session,
-                               notification_thread_handle);
-               break;
-       }
-       case LTTNG_LIST_DOMAINS:
-       {
-               ssize_t nb_dom;
-               struct lttng_domain *domains = NULL;
-
-               nb_dom = cmd_list_domains(cmd_ctx->session, &domains);
-               if (nb_dom < 0) {
-                       /* Return value is a negative lttng_error_code. */
-                       ret = -nb_dom;
-                       goto error;
-               }
-
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, domains,
-                       nb_dom * sizeof(struct lttng_domain));
-               free(domains);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_LIST_CHANNELS:
-       {
-               ssize_t payload_size;
-               struct lttng_channel *channels = NULL;
-
-               payload_size = cmd_list_channels(cmd_ctx->lsm->domain.type,
-                               cmd_ctx->session, &channels);
-               if (payload_size < 0) {
-                       /* Return value is a negative lttng_error_code. */
-                       ret = -payload_size;
-                       goto error;
-               }
-
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, channels,
-                       payload_size);
-               free(channels);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_LIST_EVENTS:
-       {
-               ssize_t nb_event;
-               struct lttng_event *events = NULL;
-               struct lttcomm_event_command_header cmd_header;
-               size_t total_size;
-
-               memset(&cmd_header, 0, sizeof(cmd_header));
-               /* Extended infos are included at the end of events */
-               nb_event = cmd_list_events(cmd_ctx->lsm->domain.type,
-                       cmd_ctx->session, cmd_ctx->lsm->u.list.channel_name,
-                       &events, &total_size);
-
-               if (nb_event < 0) {
-                       /* Return value is a negative lttng_error_code. */
-                       ret = -nb_event;
-                       goto error;
-               }
-
-               cmd_header.nb_events = nb_event;
-               ret = setup_lttng_msg(cmd_ctx, events, total_size,
-                       &cmd_header, sizeof(cmd_header));
-               free(events);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_LIST_SESSIONS:
-       {
-               unsigned int nr_sessions;
-               void *sessions_payload;
-               size_t payload_len;
-
-               session_lock_list();
-               nr_sessions = lttng_sessions_count(
-                               LTTNG_SOCK_GET_UID_CRED(&cmd_ctx->creds),
-                               LTTNG_SOCK_GET_GID_CRED(&cmd_ctx->creds));
-               payload_len = sizeof(struct lttng_session) * nr_sessions;
-               sessions_payload = zmalloc(payload_len);
-
-               if (!sessions_payload) {
-                       session_unlock_list();
-                       ret = -ENOMEM;
-                       goto setup_error;
-               }
-
-               cmd_list_lttng_sessions(sessions_payload,
-                       LTTNG_SOCK_GET_UID_CRED(&cmd_ctx->creds),
-                       LTTNG_SOCK_GET_GID_CRED(&cmd_ctx->creds));
-               session_unlock_list();
-
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, sessions_payload,
-                       payload_len);
-               free(sessions_payload);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_REGISTER_CONSUMER:
-       {
-               struct consumer_data *cdata;
-
-               switch (cmd_ctx->lsm->domain.type) {
-               case LTTNG_DOMAIN_KERNEL:
-                       cdata = &kconsumer_data;
-                       break;
-               default:
-                       ret = LTTNG_ERR_UND;
-                       goto error;
-               }
-
-               ret = cmd_register_consumer(cmd_ctx->session, cmd_ctx->lsm->domain.type,
-                               cmd_ctx->lsm->u.reg.path, cdata);
-               break;
-       }
-       case LTTNG_DATA_PENDING:
-       {
-               int pending_ret;
-               uint8_t pending_ret_byte;
-
-               pending_ret = cmd_data_pending(cmd_ctx->session);
-
-               /*
-                * FIXME
-                *
-                * This function may returns 0 or 1 to indicate whether or not
-                * there is data pending. In case of error, it should return an
-                * LTTNG_ERR code. However, some code paths may still return
-                * a nondescript error code, which we handle by returning an
-                * "unknown" error.
-                */
-               if (pending_ret == 0 || pending_ret == 1) {
-                       /*
-                        * ret will be set to LTTNG_OK at the end of
-                        * this function.
-                        */
-               } else if (pending_ret < 0) {
-                       ret = LTTNG_ERR_UNK;
-                       goto setup_error;
-               } else {
-                       ret = pending_ret;
-                       goto setup_error;
-               }
-
-               pending_ret_byte = (uint8_t) pending_ret;
-
-               /* 1 byte to return whether or not data is pending */
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx,
-                       &pending_ret_byte, 1);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_SNAPSHOT_ADD_OUTPUT:
-       {
-               struct lttcomm_lttng_output_id reply;
-
-               ret = cmd_snapshot_add_output(cmd_ctx->session,
-                               &cmd_ctx->lsm->u.snapshot_output.output, &reply.id);
-               if (ret != LTTNG_OK) {
-                       goto error;
-               }
-
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, &reply,
-                       sizeof(reply));
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               /* Copy output list into message payload */
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_SNAPSHOT_DEL_OUTPUT:
-       {
-               ret = cmd_snapshot_del_output(cmd_ctx->session,
-                               &cmd_ctx->lsm->u.snapshot_output.output);
-               break;
-       }
-       case LTTNG_SNAPSHOT_LIST_OUTPUT:
-       {
-               ssize_t nb_output;
-               struct lttng_snapshot_output *outputs = NULL;
-
-               nb_output = cmd_snapshot_list_outputs(cmd_ctx->session, &outputs);
-               if (nb_output < 0) {
-                       ret = -nb_output;
-                       goto error;
-               }
-
-               assert((nb_output > 0 && outputs) || nb_output == 0);
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, outputs,
-                               nb_output * sizeof(struct lttng_snapshot_output));
-               free(outputs);
-
-               if (ret < 0) {
-                       goto setup_error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_SNAPSHOT_RECORD:
-       {
-               ret = cmd_snapshot_record(cmd_ctx->session,
-                               &cmd_ctx->lsm->u.snapshot_record.output,
-                               cmd_ctx->lsm->u.snapshot_record.wait);
-               break;
-       }
-       case LTTNG_CREATE_SESSION_SNAPSHOT:
-       {
-               size_t nb_uri, len;
-               struct lttng_uri *uris = NULL;
-
-               nb_uri = cmd_ctx->lsm->u.uri.size;
-               len = nb_uri * sizeof(struct lttng_uri);
-
-               if (nb_uri > 0) {
-                       uris = zmalloc(len);
-                       if (uris == NULL) {
-                               ret = LTTNG_ERR_FATAL;
-                               goto error;
-                       }
-
-                       /* Receive variable len data */
-                       DBG("Waiting for %zu URIs from client ...", nb_uri);
-                       ret = lttcomm_recv_unix_sock(sock, uris, len);
-                       if (ret <= 0) {
-                               DBG("No URIs received from client... continuing");
-                               *sock_error = 1;
-                               ret = LTTNG_ERR_SESSION_FAIL;
-                               free(uris);
-                               goto error;
-                       }
-
-                       if (nb_uri == 1 && uris[0].dtype != LTTNG_DST_PATH) {
-                               DBG("Creating session with ONE network URI is a bad call");
-                               ret = LTTNG_ERR_SESSION_FAIL;
-                               free(uris);
-                               goto error;
-                       }
-               }
-
-               ret = cmd_create_session_snapshot(cmd_ctx->lsm->session.name, uris,
-                               nb_uri, &cmd_ctx->creds);
-               free(uris);
-               break;
-       }
-       case LTTNG_CREATE_SESSION_LIVE:
-       {
-               size_t nb_uri, len;
-               struct lttng_uri *uris = NULL;
-
-               nb_uri = cmd_ctx->lsm->u.uri.size;
-               len = nb_uri * sizeof(struct lttng_uri);
-
-               if (nb_uri > 0) {
-                       uris = zmalloc(len);
-                       if (uris == NULL) {
-                               ret = LTTNG_ERR_FATAL;
-                               goto error;
-                       }
-
-                       /* Receive variable len data */
-                       DBG("Waiting for %zu URIs from client ...", nb_uri);
-                       ret = lttcomm_recv_unix_sock(sock, uris, len);
-                       if (ret <= 0) {
-                               DBG("No URIs received from client... continuing");
-                               *sock_error = 1;
-                               ret = LTTNG_ERR_SESSION_FAIL;
-                               free(uris);
-                               goto error;
-                       }
-
-                       if (nb_uri == 1 && uris[0].dtype != LTTNG_DST_PATH) {
-                               DBG("Creating session with ONE network URI is a bad call");
-                               ret = LTTNG_ERR_SESSION_FAIL;
-                               free(uris);
-                               goto error;
-                       }
-               }
-
-               ret = cmd_create_session_uri(cmd_ctx->lsm->session.name, uris,
-                               nb_uri, &cmd_ctx->creds, cmd_ctx->lsm->u.session_live.timer_interval);
-               free(uris);
-               break;
-       }
-       case LTTNG_SAVE_SESSION:
-       {
-               ret = cmd_save_sessions(&cmd_ctx->lsm->u.save_session.attr,
-                       &cmd_ctx->creds);
-               break;
-       }
-       case LTTNG_SET_SESSION_SHM_PATH:
-       {
-               ret = cmd_set_session_shm_path(cmd_ctx->session,
-                               cmd_ctx->lsm->u.set_shm_path.shm_path);
-               break;
-       }
-       case LTTNG_REGENERATE_METADATA:
-       {
-               ret = cmd_regenerate_metadata(cmd_ctx->session);
-               break;
-       }
-       case LTTNG_REGENERATE_STATEDUMP:
-       {
-               ret = cmd_regenerate_statedump(cmd_ctx->session);
-               break;
-       }
-       case LTTNG_REGISTER_TRIGGER:
-       {
-               ret = cmd_register_trigger(cmd_ctx, sock,
-                               notification_thread_handle);
-               break;
-       }
-       case LTTNG_UNREGISTER_TRIGGER:
-       {
-               ret = cmd_unregister_trigger(cmd_ctx, sock,
-                               notification_thread_handle);
-               break;
-       }
-       case LTTNG_ROTATE_SESSION:
-       {
-               struct lttng_rotate_session_return rotate_return;
-
-               DBG("Client rotate session \"%s\"", cmd_ctx->session->name);
-
-               memset(&rotate_return, 0, sizeof(rotate_return));
-               if (cmd_ctx->session->kernel_session && !check_rotate_compatible()) {
-                       DBG("Kernel tracer version is not compatible with the rotation feature");
-                       ret = LTTNG_ERR_ROTATION_WRONG_VERSION;
-                       goto error;
-               }
-
-               ret = cmd_rotate_session(cmd_ctx->session, &rotate_return);
-               if (ret < 0) {
-                       ret = -ret;
-                       goto error;
-               }
-
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, &rotate_return,
-                               sizeof(rotate_return));
-               if (ret < 0) {
-                       ret = -ret;
-                       goto error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_ROTATION_GET_INFO:
-       {
-               struct lttng_rotation_get_info_return get_info_return;
-
-               memset(&get_info_return, 0, sizeof(get_info_return));
-               ret = cmd_rotate_get_info(cmd_ctx->session, &get_info_return,
-                               cmd_ctx->lsm->u.get_rotation_info.rotation_id);
-               if (ret < 0) {
-                       ret = -ret;
-                       goto error;
-               }
-
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, &get_info_return,
-                               sizeof(get_info_return));
-               if (ret < 0) {
-                       ret = -ret;
-                       goto error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       case LTTNG_ROTATION_SET_SCHEDULE:
-       {
-               bool set_schedule;
-               enum lttng_rotation_schedule_type schedule_type;
-               uint64_t value;
-
-               if (cmd_ctx->session->kernel_session && !check_rotate_compatible()) {
-                       DBG("Kernel tracer version does not support session rotations");
-                       ret = LTTNG_ERR_ROTATION_WRONG_VERSION;
-                       goto error;
-               }
-
-               set_schedule = cmd_ctx->lsm->u.rotation_set_schedule.set == 1;
-               schedule_type = (enum lttng_rotation_schedule_type) cmd_ctx->lsm->u.rotation_set_schedule.type;
-               value = cmd_ctx->lsm->u.rotation_set_schedule.value;
-
-               ret = cmd_rotation_set_schedule(cmd_ctx->session,
-                               set_schedule,
-                               schedule_type,
-                               value,
-                               notification_thread_handle);
-               if (ret != LTTNG_OK) {
-                       goto error;
-               }
-
-               break;
-       }
-       case LTTNG_SESSION_LIST_ROTATION_SCHEDULES:
-       {
-               struct lttng_session_list_schedules_return schedules = {
-                       .periodic.set = !!cmd_ctx->session->rotate_timer_period,
-                       .periodic.value = cmd_ctx->session->rotate_timer_period,
-                       .size.set = !!cmd_ctx->session->rotate_size,
-                       .size.value = cmd_ctx->session->rotate_size,
-               };
-
-               ret = setup_lttng_msg_no_cmd_header(cmd_ctx, &schedules,
-                               sizeof(schedules));
-               if (ret < 0) {
-                       ret = -ret;
-                       goto error;
-               }
-
-               ret = LTTNG_OK;
-               break;
-       }
-       default:
-               ret = LTTNG_ERR_UND;
-               break;
-       }
-
-error:
-       if (cmd_ctx->llm == NULL) {
-               DBG("Missing llm structure. Allocating one.");
-               if (setup_lttng_msg_no_cmd_header(cmd_ctx, NULL, 0) < 0) {
-                       goto setup_error;
-               }
-       }
-       /* Set return code */
-       cmd_ctx->llm->ret_code = ret;
-setup_error:
-       if (cmd_ctx->session) {
-               session_unlock(cmd_ctx->session);
-               session_put(cmd_ctx->session);
-       }
-       if (need_tracing_session) {
-               session_unlock_list();
-       }
-init_setup_error:
-       assert(!rcu_read_ongoing());
-       return ret;
-}
-
-/*
- * This thread manage all clients request using the unix client socket for
- * communication.
- */
-static void *thread_manage_clients(void *data)
-{
-       int sock = -1, ret, i, pollfd, err = -1;
-       int sock_error;
-       uint32_t revents, nb_fd;
-       struct command_ctx *cmd_ctx = NULL;
-       struct lttng_poll_event events;
-
-       DBG("[thread] Manage client started");
-
-       rcu_register_thread();
-
-       health_register(health_sessiond, HEALTH_SESSIOND_TYPE_CMD);
-
-       health_code_update();
-
-       ret = lttcomm_listen_unix_sock(client_sock);
-       if (ret < 0) {
-               goto error_listen;
-       }
-
-       /*
-        * Pass 2 as size here for the thread quit pipe and client_sock. Nothing
-        * more will be added to this poll set.
-        */
-       ret = sessiond_set_thread_pollset(&events, 2);
-       if (ret < 0) {
-               goto error_create_poll;
-       }
-
-       /* Add the application registration socket */
-       ret = lttng_poll_add(&events, client_sock, LPOLLIN | LPOLLPRI);
-       if (ret < 0) {
-               goto error;
-       }
-
-       ret = sem_post(&load_info->message_thread_ready);
-       if (ret) {
-               PERROR("sem_post message_thread_ready");
-               goto error;
-       }
-
-       /*
-        * Wait until all support threads are initialized before accepting
-        * commands.
-        */
-       while (uatomic_read(&lttng_sessiond_ready) != 0) {
-               /*
-                * If a support thread failed to launch, it may signal that
-                * we must exit and the sessiond would never be marked as
-                * "ready".
-                *
-                * The timeout is set to 1ms, which serves as a way to
-                * pace down this check.
-                */
-               ret = sessiond_wait_for_quit_pipe(1000);
-               if (ret > 0) {
-                       goto exit;
-               }
-       }
-       /*
-        * This barrier is paired with the one in sessiond_notify_ready() to
-        * ensure that loads accessing data initialized by the other threads,
-        * on which this thread was waiting, are not performed before this point.
-        *
-        * Note that this could be a 'read' memory barrier, but a full barrier
-        * is used in case the code changes. The performance implications of
-        * this choice are minimal since this is a slow path.
-        */
-       cmm_smp_mb();
-
-       /* This testpoint is after we signal readiness to the parent. */
-       if (testpoint(sessiond_thread_manage_clients)) {
-               goto error;
-       }
-
-       if (testpoint(sessiond_thread_manage_clients_before_loop)) {
-               goto error;
-       }
-
-       health_code_update();
-
-       /* Set state as running. */
-       sessiond_set_client_thread_state(true);
-
-       while (1) {
-               const struct cmd_completion_handler *cmd_completion_handler;
-
-               DBG("Accepting client command ...");
-
-               /* Inifinite blocking call, waiting for transmission */
-       restart:
-               health_poll_entry();
-               ret = lttng_poll_wait(&events, -1);
-               health_poll_exit();
-               if (ret < 0) {
-                       /*
-                        * Restart interrupted system call.
-                        */
-                       if (errno == EINTR) {
-                               goto restart;
-                       }
-                       goto error;
-               }
-
-               nb_fd = ret;
-
-               for (i = 0; i < nb_fd; i++) {
-                       /* Fetch once the poll data */
-                       revents = LTTNG_POLL_GETEV(&events, i);
-                       pollfd = LTTNG_POLL_GETFD(&events, i);
-
-                       health_code_update();
-
-                       if (!revents) {
-                               /* No activity for this FD (poll implementation). */
-                               continue;
-                       }
-
-                       /* Thread quit pipe has been closed. Killing thread. */
-                       ret = sessiond_check_thread_quit_pipe(pollfd, revents);
-                       if (ret) {
-                               err = 0;
-                               goto exit;
-                       }
-
-                       /* Event on the registration socket */
-                       if (pollfd == client_sock) {
-                               if (revents & LPOLLIN) {
-                                       continue;
-                               } else if (revents & (LPOLLERR | LPOLLHUP | LPOLLRDHUP)) {
-                                       ERR("Client socket poll error");
-                                       goto error;
-                               } else {
-                                       ERR("Unexpected poll events %u for sock %d", revents, pollfd);
-                                       goto error;
-                               }
-                       }
-               }
-
-               DBG("Wait for client response");
-
-               health_code_update();
-
-               sock = lttcomm_accept_unix_sock(client_sock);
-               if (sock < 0) {
-                       goto error;
-               }
-
-               /*
-                * Set the CLOEXEC flag. Return code is useless because either way, the
-                * show must go on.
-                */
-               (void) utils_set_fd_cloexec(sock);
-
-               /* Set socket option for credentials retrieval */
-               ret = lttcomm_setsockopt_creds_unix_sock(sock);
-               if (ret < 0) {
-                       goto error;
-               }
-
-               /* Allocate context command to process the client request */
-               cmd_ctx = zmalloc(sizeof(struct command_ctx));
-               if (cmd_ctx == NULL) {
-                       PERROR("zmalloc cmd_ctx");
-                       goto error;
-               }
-
-               /* Allocate data buffer for reception */
-               cmd_ctx->lsm = zmalloc(sizeof(struct lttcomm_session_msg));
-               if (cmd_ctx->lsm == NULL) {
-                       PERROR("zmalloc cmd_ctx->lsm");
-                       goto error;
-               }
-
-               cmd_ctx->llm = NULL;
-               cmd_ctx->session = NULL;
-
-               health_code_update();
-
-               /*
-                * Data is received from the lttng client. The struct
-                * lttcomm_session_msg (lsm) contains the command and data request of
-                * the client.
-                */
-               DBG("Receiving data from client ...");
-               ret = lttcomm_recv_creds_unix_sock(sock, cmd_ctx->lsm,
-                               sizeof(struct lttcomm_session_msg), &cmd_ctx->creds);
-               if (ret <= 0) {
-                       DBG("Nothing recv() from client... continuing");
-                       ret = close(sock);
-                       if (ret) {
-                               PERROR("close");
-                       }
-                       sock = -1;
-                       clean_command_ctx(&cmd_ctx);
-                       continue;
-               }
-
-               health_code_update();
-
-               // TODO: Validate cmd_ctx including sanity check for
-               // security purpose.
-
-               rcu_thread_online();
-               /*
-                * This function dispatch the work to the kernel or userspace tracer
-                * libs and fill the lttcomm_lttng_msg data structure of all the needed
-                * informations for the client. The command context struct contains
-                * everything this function may needs.
-                */
-               ret = process_client_msg(cmd_ctx, sock, &sock_error);
-               rcu_thread_offline();
-               if (ret < 0) {
-                       ret = close(sock);
-                       if (ret) {
-                               PERROR("close");
-                       }
-                       sock = -1;
-                       /*
-                        * TODO: Inform client somehow of the fatal error. At
-                        * this point, ret < 0 means that a zmalloc failed
-                        * (ENOMEM). Error detected but still accept
-                        * command, unless a socket error has been
-                        * detected.
-                        */
-                       clean_command_ctx(&cmd_ctx);
-                       continue;
-               }
-
-               cmd_completion_handler = cmd_pop_completion_handler();
-               if (cmd_completion_handler) {
-                       enum lttng_error_code completion_code;
-
-                       completion_code = cmd_completion_handler->run(
-                                       cmd_completion_handler->data);
-                       if (completion_code != LTTNG_OK) {
-                               clean_command_ctx(&cmd_ctx);
-                               continue;
-                       }
-               }
-
-               health_code_update();
-
-               DBG("Sending response (size: %d, retcode: %s (%d))",
-                               cmd_ctx->lttng_msg_size,
-                               lttng_strerror(-cmd_ctx->llm->ret_code),
-                               cmd_ctx->llm->ret_code);
-               ret = send_unix_sock(sock, cmd_ctx->llm, cmd_ctx->lttng_msg_size);
-               if (ret < 0) {
-                       ERR("Failed to send data back to client");
-               }
-
-               /* End of transmission */
-               ret = close(sock);
-               if (ret) {
-                       PERROR("close");
-               }
-               sock = -1;
-
-               clean_command_ctx(&cmd_ctx);
-
-               health_code_update();
-       }
-
-exit:
-error:
-       if (sock >= 0) {
-               ret = close(sock);
-               if (ret) {
-                       PERROR("close");
-               }
-       }
-
-       lttng_poll_clean(&events);
-       clean_command_ctx(&cmd_ctx);
-
-error_listen:
-error_create_poll:
-       unlink(config.client_unix_sock_path.value);
-       if (client_sock >= 0) {
-               ret = close(client_sock);
-               if (ret) {
-                       PERROR("close");
-               }
-       }
-
-       if (err) {
-               health_error();
-               ERR("Health error occurred in %s", __func__);
-       }
-
-       health_unregister(health_sessiond);
-
-       DBG("Client thread dying");
-
-       rcu_unregister_thread();
-
-       /*
-        * Since we are creating the consumer threads, we own them, so we need
-        * to join them before our thread exits.
-        */
-       ret = join_consumer_thread(&kconsumer_data);
-       if (ret) {
-               errno = ret;
-               PERROR("join_consumer");
-       }
-
-       ret = join_consumer_thread(&ustconsumer32_data);
-       if (ret) {
-               errno = ret;
-               PERROR("join_consumer ust32");
-       }
-
-       ret = join_consumer_thread(&ustconsumer64_data);
-       if (ret) {
-               errno = ret;
-               PERROR("join_consumer ust64");
-       }
-
-       /* Set state as non-running. */
-       sessiond_set_client_thread_state(false);
-       return NULL;
-}
-
-static int string_match(const char *str1, const char *str2)
-{
-       return (str1 && str2) && !strcmp(str1, str2);
-}
-
-/*
- * Take an option from the getopt output and set it in the right variable to be
- * used later.
- *
- * Return 0 on success else a negative value.
- */
-static int set_option(int opt, const char *arg, const char *optname)
-{
-       int ret = 0;
+       int ret = 0;
 
        if (string_match(optname, "client-sock") || opt == 'c') {
                if (!arg || *arg == '\0') {
@@ -5030,9 +2512,7 @@ end:
 }
 
 /*
- * Creates the two needed socket by the daemon.
- *         apps_sock - The communication socket for all UST apps.
- *         client_sock - The communication of the cli tool (lttng).
+ * Creates the application socket.
  */
 static int init_daemon_socket(void)
 {
@@ -5041,30 +2521,6 @@ static int init_daemon_socket(void)
 
        old_umask = umask(0);
 
-       /* Create client tool unix socket */
-       client_sock = lttcomm_create_unix_sock(config.client_unix_sock_path.value);
-       if (client_sock < 0) {
-               ERR("Create unix sock failed: %s", config.client_unix_sock_path.value);
-               ret = -1;
-               goto end;
-       }
-
-       /* Set the cloexec flag */
-       ret = utils_set_fd_cloexec(client_sock);
-       if (ret < 0) {
-               ERR("Unable to set CLOEXEC flag to the client Unix socket (fd: %d). "
-                               "Continuing but note that the consumer daemon will have a "
-                               "reference to this socket on exec()", client_sock);
-       }
-
-       /* File permission MUST be 660 */
-       ret = chmod(config.client_unix_sock_path.value, S_IRUSR | S_IWUSR | S_IRGRP | S_IWGRP);
-       if (ret < 0) {
-               ERR("Set file permissions failed: %s", config.client_unix_sock_path.value);
-               PERROR("chmod");
-               goto end;
-       }
-
        /* Create the application unix socket */
        apps_sock = lttcomm_create_unix_sock(config.apps_unix_sock_path.value);
        if (apps_sock < 0) {
@@ -5090,8 +2546,8 @@ static int init_daemon_socket(void)
                goto end;
        }
 
-       DBG3("Session daemon client socket %d and application socket %d created",
-                       client_sock, apps_sock);
+       DBG3("Session daemon application socket %d created",
+                       apps_sock);
 
 end:
        umask(old_umask);
@@ -5448,12 +2904,6 @@ static void destroy_all_sessions_and_wait(void)
                return;
        }
 
-       /*
-        * Ensure that the client thread is no longer accepting new commands,
-        * which could cause new sessions to be created.
-        */
-       sessiond_wait_client_thread_stopped();
-
        session_lock_list();
        /* Initiate the destruction of all sessions. */
        cds_list_for_each_entry_safe(session, tmp,
@@ -5497,6 +2947,7 @@ int main(int argc, char **argv)
        struct rotation_thread_handle *rotation_thread_handle = NULL;
        /* Queue of rotation jobs populated by the sessiond-timer. */
        struct rotation_thread_timer_queue *rotation_timer_queue = NULL;
+       struct lttng_thread *client_thread = NULL;
 
        init_kernel_workarounds();
 
@@ -5887,13 +3338,9 @@ int main(int argc, char **argv)
        }
 
        /* Create thread to manage the client socket */
-       ret = pthread_create(&client_thread, default_pthread_attr(),
-                       thread_manage_clients, (void *) NULL);
-       if (ret) {
-               errno = ret;
-               PERROR("pthread_create clients");
+       client_thread = launch_client_thread();
+       if (!client_thread) {
                retval = -1;
-               stop_threads();
                goto exit_client;
        }
 
@@ -5991,6 +3438,17 @@ int main(int argc, char **argv)
 
        /* Initiate teardown once activity occurs on the quit pipe. */
        sessiond_wait_for_quit_pipe(-1U);
+
+       /*
+        * Ensure that the client thread is no longer accepting new commands,
+        * which could cause new sessions to be created.
+        */
+       if (!lttng_thread_shutdown(client_thread)) {
+               ERR("Failed to shutdown the client thread, continuing teardown");
+               lttng_thread_put(client_thread);
+               client_thread = NULL;
+       }
+
        destroy_all_sessions_and_wait();
 exit_load_session:
 
@@ -6047,20 +3505,16 @@ exit_reg_apps:
                retval = -1;
        }
 exit_dispatch:
-
-       ret = pthread_join(client_thread, &status);
-       if (ret) {
-               errno = ret;
-               PERROR("pthread_join");
-               retval = -1;
-       }
-
 exit_client:
 exit_rotation:
 exit_notification:
        lttng_thread_list_shutdown_orphans();
 exit_health:
 exit_init_data:
+       if (client_thread) {
+               lttng_thread_put(client_thread);
+       }
+
        /*
         * Wait for all pending call_rcu work to complete before tearing
         * down data structures. call_rcu worker may be trying to
index 3a8ddb3303639f75d86be3e1b7da7a232d2771ec..549cd69f867e23998220f01e5e80174520b83014 100644 (file)
  */
 static int thread_quit_pipe[2] = { -1, -1 };
 
-/*
- * Allows threads to query the state of the client thread.
- */
-static struct client_thread_state {
-       pthread_cond_t cond;
-       pthread_mutex_t lock;
-       bool is_running;
-} client_thread_state = {
-       .cond = PTHREAD_COND_INITIALIZER,
-       .lock = PTHREAD_MUTEX_INITIALIZER,
-       .is_running = false
-};
-
 /*
  * Init thread quit pipe.
  *
@@ -136,24 +123,6 @@ void sessiond_close_quit_pipe(void)
        utils_close_pipe(thread_quit_pipe);
 }
 
-void sessiond_set_client_thread_state(bool running)
-{
-       pthread_mutex_lock(&client_thread_state.lock);
-       client_thread_state.is_running = running;
-       pthread_cond_broadcast(&client_thread_state.cond);
-       pthread_mutex_unlock(&client_thread_state.lock);
-}
-
-void sessiond_wait_client_thread_stopped(void)
-{
-       pthread_mutex_lock(&client_thread_state.lock);
-       while (client_thread_state.is_running) {
-               pthread_cond_wait(&client_thread_state.cond,
-                               &client_thread_state.lock);
-       }
-       pthread_mutex_unlock(&client_thread_state.lock);
-}
-
 static
 int __sessiond_set_thread_pollset(struct lttng_poll_event *events, size_t size,
                int *a_pipe)
This page took 0.139771 seconds and 5 git commands to generate.