{
int ret;
char *metadata_str = NULL;
- size_t len, offset;
+ size_t len, offset, new_metadata_len_sent;
ssize_t ret_val;
+ uint64_t metadata_key;
assert(registry);
assert(socket);
+ metadata_key = registry->metadata_key;
+
/*
* Means that no metadata was assigned to the session. This can
* happens if no start has been done previously.
*/
- if (!registry->metadata_key) {
+ if (!metadata_key) {
return 0;
}
offset = registry->metadata_len_sent;
len = registry->metadata_len - registry->metadata_len_sent;
+ new_metadata_len_sent = registry->metadata_len;
if (len == 0) {
DBG3("No metadata to push for metadata key %" PRIu64,
registry->metadata_key);
ret_val = -ENOMEM;
goto error;
}
- /* Copy what we haven't send out. */
+ /* Copy what we haven't sent out. */
memcpy(metadata_str, registry->metadata + offset, len);
- registry->metadata_len_sent += len;
push_data:
- ret = consumer_push_metadata(socket, registry->metadata_key,
+ pthread_mutex_unlock(®istry->lock);
+ /*
+ * We need to unlock the registry while we push metadata to
+ * break a circular dependency between the consumerd metadata
+ * lock and the sessiond registry lock. Indeed, pushing metadata
+ * to the consumerd awaits that it gets pushed all the way to
+ * relayd, but doing so requires grabbing the metadata lock. If
+ * a concurrent metadata request is being performed by
+ * consumerd, this can try to grab the registry lock on the
+ * sessiond while holding the metadata lock on the consumer
+ * daemon. Those push and pull schemes are performed on two
+ * different bidirectionnal communication sockets.
+ */
+ ret = consumer_push_metadata(socket, metadata_key,
metadata_str, len, offset);
+ pthread_mutex_lock(®istry->lock);
if (ret < 0) {
/*
* There is an acceptable race here between the registry
*/
if (ret == -LTTCOMM_CONSUMERD_CHANNEL_FAIL) {
ret = 0;
+ } else {
+ ERR("Error pushing metadata to consumer");
}
-
- /*
- * Update back the actual metadata len sent since it
- * failed here.
- */
- registry->metadata_len_sent -= len;
ret_val = ret;
goto error_push;
+ } else {
+ /*
+ * Metadata may have been concurrently pushed, since
+ * we're not holding the registry lock while pushing to
+ * consumer. This is handled by the fact that we send
+ * the metadata content, size, and the offset at which
+ * that metadata belongs. This may arrive out of order
+ * on the consumer side, and the consumer is able to
+ * deal with overlapping fragments. The consumer
+ * supports overlapping fragments, which must be
+ * contiguous starting from offset 0. We keep the
+ * largest metadata_len_sent value of the concurrent
+ * send.
+ */
+ registry->metadata_len_sent =
+ max_t(size_t, registry->metadata_len_sent,
+ new_metadata_len_sent);
}
-
free(metadata_str);
return len;
}
pthread_mutex_unlock(&ua_sess->lock);
+ consumer_output_put(ua_sess->consumer);
+
call_rcu(&ua_sess->rcu_head, delete_ust_app_session_rcu);
}
/* Send channel to the application. */
ret = ust_consumer_send_channel_to_ust(app, ua_sess, ua_chan);
- if (ret < 0) {
+ if (ret == -EPIPE || ret == -LTTNG_UST_ERR_EXITING) {
+ ret = -ENOTCONN; /* Caused by app exiting. */
+ goto error;
+ } else if (ret < 0) {
goto error;
}
/* Send all streams to application. */
cds_list_for_each_entry_safe(stream, stmp, &ua_chan->streams.head, list) {
ret = ust_consumer_send_stream_to_ust(app, ua_chan, stream);
- if (ret < 0) {
+ if (ret == -EPIPE || ret == -LTTNG_UST_ERR_EXITING) {
+ ret = -ENOTCONN; /* Caused by app exiting. */
+ goto error;
+ } else if (ret < 0) {
goto error;
}
/* We don't need the stream anymore once sent to the tracer. */
ua_sess->egid = usess->gid;
ua_sess->buffer_type = usess->buffer_type;
ua_sess->bits_per_long = app->bits_per_long;
+
/* There is only one consumer object per session possible. */
+ consumer_output_get(usess->consumer);
ua_sess->consumer = usess->consumer;
+
ua_sess->output_traces = usess->output_traces;
ua_sess->live_timer_interval = usess->live_timer_interval;
copy_channel_attr_to_ustctl(&ua_sess->metadata_attr,
lttng_ht_add_unique_str(ua_sess->channels, &ua_chan->node);
}
+ return;
error:
- return;
+ consumer_output_put(ua_sess->consumer);
}
/*
/* Send channel to the application. */
ret = ust_consumer_send_channel_to_ust(app, ua_sess, ua_chan);
- if (ret < 0) {
+ if (ret == -EPIPE || ret == -LTTNG_UST_ERR_EXITING) {
+ ret = -ENOTCONN; /* Caused by app exiting. */
+ goto error;
+ } else if (ret < 0) {
goto error;
}
ret = ust_consumer_send_stream_to_ust(app, ua_chan, &stream);
if (ret < 0) {
(void) release_ust_app_stream(-1, &stream);
+ if (ret == -EPIPE || ret == -LTTNG_UST_ERR_EXITING) {
+ ret = -ENOTCONN; /* Caused by app exiting. */
+ goto error_stream_unlock;
+ } else if (ret < 0) {
+ goto error_stream_unlock;
+ }
goto error_stream_unlock;
}
/* Send buffers to the application. */
ret = send_channel_uid_to_ust(reg_chan, app, ua_sess, ua_chan);
if (ret < 0) {
- /*
- * Don't report error to the console, since it may be
- * caused by application concurrently exiting.
- */
+ if (ret != -ENOTCONN) {
+ ERR("Error sending channel to application");
+ }
goto error;
}
ret = send_channel_pid_to_ust(app, ua_sess, ua_chan);
if (ret < 0) {
- /*
- * Don't report error to the console, since it may be
- * caused by application concurrently exiting.
- */
+ if (ret != -ENOTCONN) {
+ ERR("Error sending channel to application");
+ }
goto error;
}
* need and send it to the application. This MUST be called with a RCU read
* side lock acquired.
*
- * Return 0 on success or else a negative value.
+ * Return 0 on success or else a negative value. Returns -ENOTCONN if
+ * the application exited concurrently.
*/
static int do_create_channel(struct ust_app *app,
struct ltt_ust_session *usess, struct ust_app_session *ua_sess,
*
* Called with UST app session lock and RCU read-side lock held.
*
- * Return 0 on success or else a negative value.
+ * Return 0 on success or else a negative value. Returns -ENOTCONN if
+ * the application exited concurrently.
*/
static int create_ust_app_channel(struct ust_app_session *ua_sess,
struct ltt_ust_channel *uchan, struct ust_app *app,
* or a timeout on it. We can't inform the caller that for a
* specific app, the session failed so lets continue here.
*/
+ ret = 0; /* Not an error. */
continue;
case -ENOMEM:
default:
}
pthread_mutex_unlock(&ua_sess->lock);
if (ret < 0) {
- if (ret == -ENOMEM) {
- /* No more memory is a fatal error. Stop right now. */
- goto error_rcu_unlock;
- }
/* Cleanup the created session if it's the case. */
if (created) {
destroy_app_session(app, ua_sess);
}
+ switch (ret) {
+ case -ENOTCONN:
+ /*
+ * The application's socket is not valid. Either a bad socket
+ * or a timeout on it. We can't inform the caller that for a
+ * specific app, the session failed so lets continue here.
+ */
+ ret = 0; /* Not an error. */
+ continue;
+ case -ENOMEM:
+ default:
+ goto error_rcu_unlock;
+ }
}
}
/* Lookup channel in the ust app session */
lttng_ht_lookup(ua_sess->channels, (void *)uchan->name, &uiter);
ua_chan_node = lttng_ht_iter_get_node_str(&uiter);
- /* If the channel is not found, there is a code flow error */
- assert(ua_chan_node);
+ /*
+ * It is possible that the channel cannot be found is
+ * the channel/event creation occurs concurrently with
+ * an application exit.
+ */
+ if (!ua_chan_node) {
+ pthread_mutex_unlock(&ua_sess->lock);
+ continue;
+ }
ua_chan = caa_container_of(ua_chan_node, struct ust_app_channel, node);
ret = run_as_mkdir_recursive(usess->consumer->dst.trace_path,
S_IRWXU | S_IRWXG, ua_sess->euid, ua_sess->egid);
if (ret < 0) {
- if (ret != -EEXIST) {
+ if (errno != EEXIST) {
ERR("Trace directory creation error");
goto error_unlock;
}
cds_lfht_for_each_entry(ua_sess->channels->ht, &iter.iter, ua_chan,
node.node) {
ret = do_create_channel(app, usess, ua_sess, ua_chan);
- if (ret < 0) {
+ if (ret < 0 && ret != -ENOTCONN) {
/*
- * Stop everything. On error, the application failed, no more
- * file descriptor are available or ENOMEM so stopping here is
- * the only thing we can do for now.
+ * Stop everything. On error, the application
+ * failed, no more file descriptor are available
+ * or ENOMEM so stopping here is the only thing
+ * we can do for now. The only exception is
+ * -ENOTCONN, which indicates that the application
+ * has exit.
*/
goto error_unlock;
}
projects / lttng-tools.git / blobdiff