Philippe Proulx [Mon, 24 Apr 2017 20:29:39 +0000 (16:29 -0400)]
text.pretty: handle AGAIN status
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 24 Apr 2017 18:42:08 +0000 (14:42 -0400)]
utils.muxer: fix unhandled connected ports during notif. iter. init.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 24 Apr 2017 18:41:08 +0000 (14:41 -0400)]
text.pretty: fix handle_notification()
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 18 Apr 2017 19:02:26 +0000 (15:02 -0400)]
utils.muxer: fix behaviour with ports connected during next/init ops
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 10 Apr 2017 19:20:56 +0000 (15:20 -0400)]
Add basic utils.muxer tests
The tests in tests/plugins/test-utils-muxer.c create custom source and
sink components and connects them to a utils.muxer filter component
within a new graph.
The components do as follows:
* Source component: Depending on the current test, its notification
iterators generate specific sequences of notifications.
* Sink component: For each notification, and some specific notification
iterator statuses ("again", "end"), create a test event and append it
to a global array of test events.
Each test creates a graph, runs it, and then compares the global array
of test events against a static list of expected test events.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 20:30:33 +0000 (16:30 -0400)]
Add utils.muxer component class
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 10 Apr 2017 19:18:14 +0000 (15:18 -0400)]
bt_plugin_create_append_all_from_dir(): use printf_verbose() instead of error
It's not a fatal error when this function cannot open a directory: it
can simply mean that this directory does not exist, but that's okay
because the function does not load plugins in this case, and it returns
an error code.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 7 Apr 2017 17:55:56 +0000 (13:55 -0400)]
Event notification: validate that CC in the CC prio map have a value
Instead of making sure that all the trace's clock classes have a value
and a priority in an event, instead make sure that all the clock classes
found in the event notification's clock class priority map have a value
in the event and that this clock class is part of the event's trace.
This makes it possible for an event to have only clock values for
relevant clock classes, not all the trace's, or even to have no value
at all.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 7 Apr 2017 17:55:31 +0000 (13:55 -0400)]
event-class-internal.h: fix include warning
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 20:48:21 +0000 (16:48 -0400)]
text.pretty: remove unused debug info options
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 20:43:26 +0000 (16:43 -0400)]
text.pretty: use clock class with highest priority to print the timestamp
Also, if there's no clock class in the event notification's clock class
priority map, then do not print any timestamp.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 20:19:57 +0000 (16:19 -0400)]
Add bt_plugin_find_component_class()
This patch also renames bt_plugin_create_from_name() to bt_plugin_find()
which seems more intuitive. bt_plugin_find_component_class() simply
calls bt_plugin_find() and then gets one of its component classes.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 18:41:12 +0000 (14:41 -0400)]
Fix: ctf: notif-iter.c: update_clock() uses the wrong FT sometimes
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 18:40:29 +0000 (14:40 -0400)]
ctf.fs: make clock classes absolute when it's an LTTng metadata
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 18:37:44 +0000 (14:37 -0400)]
Use the same values for different status codes
The weird values are based on the Linux error numbers
(EAGAIN, EINVAL, etc.).
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 07:31:55 +0000 (03:31 -0400)]
Add bt_component_class_is_source/filter/sink()
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 07:29:48 +0000 (03:29 -0400)]
Add bt_port_is_input(), bt_port_is_output()
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 07:27:19 +0000 (03:27 -0400)]
Add bt_component_is_source(), bt_component_is_filter(), bt_component_is_sink()
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 07:24:03 +0000 (03:24 -0400)]
Fix a few include warnings
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 06:16:16 +0000 (02:16 -0400)]
plugins/ctf/fs/metadata.c: remove invalid return code assignment
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 06:10:42 +0000 (02:10 -0400)]
bt_component_create(): create convenient empty map parameters when NULL
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 6 Apr 2017 04:54:01 +0000 (00:54 -0400)]
Add BT_NOTIFICATION_ITERATOR_STATUS_AGAIN status
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 5 Apr 2017 21:00:48 +0000 (17:00 -0400)]
test_graph_topo: test both "port connected" and "accept port connection"
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 5 Apr 2017 20:11:36 +0000 (16:11 -0400)]
Add "port connected" component class method
What used to be "accept port connection" is now split in "accept port
connection" and "port connected". The latter confirms that the two ports
are connected after they both accepted the connection. Thus in the
"accept port connection" method, the ports are not connected yet: this
is not where the component can create a notification iterator because
the connection object does not exist.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 4 Apr 2017 22:04:44 +0000 (18:04 -0400)]
Rename text.text sink CC to text.pretty
The `text` plugin is expected to contain more than the `text` sink: it
could eventually read and write many plain text formats.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 4 Apr 2017 18:54:09 +0000 (14:54 -0400)]
Rename lib/component/ -> lib/graph/ to match include/babeltrace/graph/
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 4 Apr 2017 18:49:10 +0000 (14:49 -0400)]
Add `-internal` suffix to all internal header files
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 4 Apr 2017 18:11:14 +0000 (14:11 -0400)]
Add bt_version_get_*() functions to dynamically get the lib's version
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 24 Mar 2017 01:54:02 +0000 (21:54 -0400)]
.gitignore: ignore gcov files
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 24 Mar 2017 01:39:22 +0000 (21:39 -0400)]
ctf plugin: add CTF metadata decoder API
The new plugins/ctf/common/metadata/decoder.h header exposes the CTF
metadata decoder API. This is a common API used by both the ctf.fs and
ctf.lttng-live component classes.
Once you create a CTF metadata decoder object with
ctf_metadata_decoder_create(), you can call
ctf_metadata_decoder_decode() as many times as needed to decode chunks
of metadata. Those chunks can be packetized or not, however the
resulting metadata text needs to be "complete", that is, zero or more
top-level TSDL blocks ("declarations"). If ctf_metadata_decoder_decode()
cannot decode because the metadata text is incomplete, the function
returns CTF_METADATA_DECODER_STATUS_INCOMPLETE. It is then the caller's
responsibility to keep the existing metadata and append more metadata
to it and then call ctf_metadata_decoder_decode() again with this,
until the function returns CTF_METADATA_DECODER_STATUS_OK or another
error.
At anytime you can get the CTF metadata decoder's associated CTF IR
trace object with ctf_metadata_decoder_get_trace(). The trace object is
initially empty: as you call ctf_metadata_decoder_decode(), the decoder
makes it grow by adding other CTF IR objects to it: stream classes,
event classes to new or existing stream classes, clock classes, etc.
The public ctf_metadata_decoder_is_packetized() and
ctf_metadata_decoder_packetized_file_stream_to_buf() functions exist
only for the `metadata-info` ctf.fs query for the moment.
Also in this patch:
* The identical btr/print.h and notif-iter/print.h is moved to
metadata/print.h.
* A fix is added to the AST-to-CTF IR visitor to pop the current
declaration scope at the end of visit_event_decl().
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 4 Apr 2017 12:53:32 +0000 (08:53 -0400)]
tests/lib/common.c: add missing includes
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 4 Apr 2017 00:23:17 +0000 (20:23 -0400)]
bt_notification_event_create(): validate and freeze event
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 4 Apr 2017 00:13:24 +0000 (20:13 -0400)]
Add inactivity notification
The new inactivity notification indicates that there's no activity to be
expected on the iterator pointing to it until a given time. You specify
the given time with one or more clock values which are associated to
clock classes. Those clock classes must be mapped to a priority in the
notification's clock class priority map.
This notification is similar to the event notification, except that it
does not contain any context or payload. Its purpose is to make the
graph more efficient with live sources which know that a given iterator
won't deliver any nofication which "occur" before a given time. For
example, a muxer can add this notification to its internal heap, just
like it adds event notifications. Most sinks should ignore this
notification.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 3 Apr 2017 23:41:35 +0000 (19:41 -0400)]
Add frozen property to notification objects
A frozen notification cannot be modified (except for its reference
count). Currently, no notification can be modified after its creation
anyway, but we're excepting some.
The notification is frozen by bt_notification_iterator_next() when the
component's "next" method is successful.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 3 Apr 2017 23:28:53 +0000 (19:28 -0400)]
ir: bt_ctf_event_set_clock_value(): validate clock value's class
On bt_ctf_event_set_clock_value(): validate that the clock class of a
clock value is part of the event's trace object.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 3 Apr 2017 22:32:01 +0000 (18:32 -0400)]
ir: make sure you can't add a SC to a trace with a native BO
When you create a trace object with bt_ctf_trace_create(), its native
byte order is set to BT_CTF_BYTE_ORDER_NATIVE, which essentially means
"unset". Before freezing the trace (which only happens in
bt_ctf_trace_add_stream_class()), make sure that the trace's native byte
order is a real byte order, not still BT_CTF_BYTE_ORDER_NATIVE (unset).
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 3 Apr 2017 22:05:19 +0000 (18:05 -0400)]
CTF writer: restore native byte order is the CPU's native byte order
To remain compatible with the first experiments of CTF writer, let's
restore this legacy behaviour in which a native byte order means the
native byte order of the CPU running the Babeltrace library.
This only applies to bt_ctf_writer_set_byte_order() when you pass
BT_CTF_BYTE_ORDER_NATIVE: bt_ctf_trace_set_native_byte_order() does
not accept BT_CTF_BYTE_ORDER_NATIVE. Thus, when you create a CTF writer
object with bt_ctf_writer_create(), it sets its trace's native byte
order to the CPU's native byte order.
perf is a CTF writer user known to not call
bt_ctf_writer_set_byte_order() manually: in this case the CTF IR trace
would be invalid because its native byte order would still be set to
BT_CTF_BYTE_ORDER_NATIVE.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 3 Apr 2017 22:01:06 +0000 (18:01 -0400)]
Rename bt_ctf_trace_set_byte_order() -> bt_ctf_trace_set_native_byte_order()
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 3 Apr 2017 21:52:57 +0000 (17:52 -0400)]
test_bt_ctf_field_type_validation: fix compiler warnings
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 3 Apr 2017 21:48:51 +0000 (17:48 -0400)]
Rename bt_ctf_type_id -> bt_ctf_field_type_id (and the enumerators)
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Mon, 3 Apr 2017 20:39:45 +0000 (16:39 -0400)]
fix Makefile for libbabeltrace-ctf
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Mon, 3 Apr 2017 20:32:28 +0000 (16:32 -0400)]
fix warnings
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Mon, 3 Apr 2017 20:30:38 +0000 (16:30 -0400)]
fix missing libbabeltrace-ctf for libctfcopytrace
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Fri, 31 Mar 2017 17:34:58 +0000 (13:34 -0400)]
fixes after rebase on eepp-jgalar/fixes-mar-7
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Mon, 27 Mar 2017 19:32:52 +0000 (15:32 -0400)]
Text: handle output file
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Mon, 27 Mar 2017 19:11:38 +0000 (15:11 -0400)]
text: escape control sequences characters
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Tue, 21 Mar 2017 17:41:28 +0000 (13:41 -0400)]
Support for --clock-offset and --clock-offset-ns
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Mon, 20 Mar 2017 20:13:44 +0000 (16:13 -0400)]
Text output identical with babeltrace 1
Various fixes to make sure the text output stays the same as babeltrace
1. Tested all -f and -n options combinations with no differences.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Fri, 3 Mar 2017 19:16:54 +0000 (14:16 -0500)]
fix: missing equal sign when printing events without colors
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Tue, 31 Jan 2017 18:12:38 +0000 (13:12 -0500)]
debug-info filter plugin
This plugin can be inserted in the graph to add the debugging
informations to events that have the necessary context informations (see
the README for usage informations).
Compared to the previous implementation, this version does not only
output the debug informations to the text output, but it adds the
debug_info structure into the CTF events. That way, the plugins and
sinks after this one see the debug_info as part of the event contexts.
The default name of the structure added to the events is "debug_info",
but can be overridden. If this structure already exists in the event
context, it is not added again.
If a trace does not have the ip and vpid event contexts in its stream
class, the trace is copied without adding the "debug_info" structure.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Thu, 2 Mar 2017 22:06:14 +0000 (17:06 -0500)]
Make bt_ctf_field_structure_set_field public
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Thu, 23 Mar 2017 15:25:00 +0000 (11:25 -0400)]
cleanup error paths in trimmer, writer and ctfcopytrace
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Wed, 22 Feb 2017 21:49:29 +0000 (16:49 -0500)]
trimmer: Update the bounds of the trimmed packets
Update the timestamp_begin field of the first packet of each stream and
the timestamp_end field of the last packet to reflect the fact that the
packet content got truncated by the user. This is important for the
packet intersection feature afterwards. This requires that we create new
packets and copy the events to assign them to the new packets.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Mon, 27 Feb 2017 19:53:41 +0000 (14:53 -0500)]
Fix graph facilities handling in writer component class
Commit
43e5ad310e9ea54e86dd3444e23c2d3c69827bc9 missed a few details to
make it work properly.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Wed, 15 Mar 2017 20:04:31 +0000 (16:04 -0400)]
fix trimmer compiled library name
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Wed, 15 Feb 2017 19:10:46 +0000 (14:10 -0500)]
writer: create event classes lazily
Instead of copying all the event classes when we copy the stream
classes, we now create them only when needed. That way, we can use this
plugin in live where new events can get created after the trace has
already been opened.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Thu, 2 Mar 2017 22:05:39 +0000 (17:05 -0500)]
libcopytrace: export ctf_copy_event_header
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Tue, 14 Feb 2017 17:03:00 +0000 (12:03 -0500)]
Override clock fields in copied traces
Some tracers like LTTng use a compression scheme on the timestamp field
to avoid writing the full 64-bit timestamp at every event. When the
trace is complete, the reader has all the information to understand this
compression, but if a trace is filtered or truncated, there might be
some state information missing.
Since we don't know in advance if a compression scheme is used in a
trace, we now force the copied traces to have a full 64-bit timestamp.
The user of this library can decide whether or not to override the type
of the clock fields.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Tue, 14 Feb 2017 17:00:28 +0000 (12:00 -0500)]
Create a library to copy a CTF trace
Move all the code from the writer plugin to a standalone plugin library
so it can be used by other plugins.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Tue, 14 Feb 2017 17:31:28 +0000 (12:31 -0500)]
Allow to set the size of an integer type
Add bt_ctf_field_type_integer_set_size to the API.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Fri, 20 Jan 2017 21:11:01 +0000 (16:11 -0500)]
Fix writer: leak of the event classes
We have to release the writer_event_class once it is added to the
writer_stream_class. This also highlighted a problem with the ownership
of the stream, stream_class and writer objects which is now fixed.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Fri, 20 Jan 2017 21:08:58 +0000 (16:08 -0500)]
Fix writer: missing put ref to values and types
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Julien Desfossez [Wed, 18 Jan 2017 18:35:44 +0000 (13:35 -0500)]
Fix writer: leak of the header fields
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Mon, 3 Apr 2017 18:10:34 +0000 (14:10 -0400)]
Rename converter/ -> cli/ (it's more than a converter now)
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 31 Mar 2017 05:12:08 +0000 (01:12 -0400)]
Event notification: pass CC priority map on creation
An event notification must have an associated clock class priority map
to indicate the ranks of its event's clock values.
As of this patch, a ctf.fs source component always set the priority of
all the parsed clock classes to 0.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Sat, 1 Apr 2017 01:29:42 +0000 (21:29 -0400)]
Remove Babeltrace 1 files and reorganize the tree
This is one of the patches that you won't easily forget.
This patch removes the remaining legacy files, moving them if we still
need them, and reorganizes the tree as such:
* formats/ is removed
* formats/ctf/ir/ is moved to lib/ctf-ir/
* formats/ctf/writer/ is moved to lib/ctf-writer/
* types/ is removed
Because some of the CTF IR and CTF writer objects are so coupled, both
components are in libbabeltrace now. The build system creates a symbolic
link from libbabeltrace-ctf (where CTF writer used to reside) to
libbabeltrace, so that `-lbabeltrace-ctf` still works for the current
users of CTF writer.
converter/babeltrace-log.c still exists, but it's not build as of this
patch because it uses the legacy API. We need to change this to use CTF
writer and add it to the Makefile again.
Some debug info files are gone because they also use the legacy API.
They are updated anyway as part of the upcoming debug info component
class.
The Python bindings are not covered by this patch: a subsequent patch
should fix them at once.
Some tests which used the legacy API are removed.
The legacy include files are removed, except for
include/babeltrace/ctf/event.h which could be included by CTF writer
users. The file simply includes all the CTF writer header files now.
The functions to serialize integer and floating point number fields
are moved to lib/ctf-writer/serialize.c where a light
`struct bt_ctf_stream_pos` exists along with a few helper functions
for this renamed legacy object. The serialization functions use BT 2
objects now so that the dependency on the legacy API can be dropped.
The serialization functions accept a native byte order parameter now.
When a field type has a "native" byte order, the real (passed) native
byte order must be used. This ensures that:
* CTF IR objects do not need to propagate the trace's native byte order
to all the contained field types.
* Two references to the same field type which are part of two different
trace objects can have a native byte order without causing byte order
overwriting bugs.
The CTF IR code is simplified with this change. Also the
bt_ctf_trace_set_byte_order() function refuses the
BT_CTF_BYTE_ORDER_NATIVE byte order.
I also added internal inline functions to quickly access the parent
of some CTF IR object without getting a new reference:
* bt_ctf_event_borrow_event_class()
* bt_ctf_event_class_borrow_stream_class()
* bt_ctf_stream_class_borrow_trace()
Since the CTF parser test in formats/ctf/metadata/ was a legacy tool,
it's removed in this patch. It's not used anymore by test_ctf_writer
(only the converter runs to validate the generated trace).
Leaks are fixed in test_ctf_ir_ref and test_trace_listener.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 31 Mar 2017 05:10:31 +0000 (01:10 -0400)]
ctf.fs: bt_ctf_notif_iter_create(): assert() that all medops exist
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 31 Mar 2017 03:16:18 +0000 (23:16 -0400)]
trace.h: update API doc
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 31 Mar 2017 02:42:42 +0000 (22:42 -0400)]
tests: add clock class priority map tests
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 31 Mar 2017 02:28:59 +0000 (22:28 -0400)]
Add clock class priority map object
This new object is a simple map of clock classes to their priorities
against the other clock classes of the same map. See the header
file's documentation for more information (clock-class-priority-map.h).
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 29 Mar 2017 21:44:06 +0000 (17:44 -0400)]
ctf.fs: split streams, one per port
The ports are named `traceN-stream-STREAM`, where `N` is the index of
the trace (starts at 0) and `STREAM` is the base name of the stream
file.
The heap which used to be in this component class is expected to be
moved to the utils.muxer filter component class.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 29 Mar 2017 20:16:13 +0000 (16:16 -0400)]
Add user data for private port
A private port belongs to a private component, so it's okay to set
custom user data within a port. The user must make sure to release the
private user data whatever happens. The easiest solution is to make
a port's private data owned by the component's private data.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 28 Mar 2017 16:01:49 +0000 (12:01 -0400)]
Add graph topology tests
The new tests verify that the component methods and graph listeners are
called by the graph framework in reaction to specific events with the
expected parameters and in the expected order.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 29 Mar 2017 15:55:11 +0000 (11:55 -0400)]
Call a single "ports connected/disconnected" graph listener instead of two
It is reasonable to believe that a user implementing a
"ports connected" or "ports disconnected" graph listener wants both
the upstream and downstream ports, instead of getting called two times
with one and the other.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 29 Mar 2017 06:27:53 +0000 (02:27 -0400)]
Fix: call acting component's "port disconnected" method too
When a port is disconnected, call both upstream and downstream
components. It's okay to get notified as a component even if you
caused this notification.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 29 Mar 2017 02:31:45 +0000 (22:31 -0400)]
bt_graph_connect() -> bt_graph_connect_ports()
This is more explicit.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 29 Mar 2017 02:04:59 +0000 (22:04 -0400)]
Rename <babeltrace/component/...> -> <babeltrace/graph/...>
Also put all the graph-related header files directly in this directory,
not in subdirectories.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 29 Mar 2017 00:15:40 +0000 (20:15 -0400)]
Update notification iterator's "init" function signature
As per this project's standard way of calling methods, the self object
should always be the first parameter. In the cast of
bt_component_class_notification_iterator_init_method, we don't need to
explicitly pass the bt_private_component object since you can get it
from the notification iterator with
bt_private_notification_iterator_get_private_component().
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 28 Mar 2017 23:57:34 +0000 (19:57 -0400)]
Remove the need to implement the notification iterator's "get" method
The order of a notification iterator's "get" method is to return the
"current notification", set by the last "next" method call. Since all
source and filter plugins do the same in their "get" method, let's
remove the need to implement it at all.
The bt_notification_iterator_next() and
bt_notification_iterator_get_notification() public functions are still
available for the iterator's user, but the iterator object itself keeps
the current notification returned by the last "next" method call.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 28 Mar 2017 23:24:01 +0000 (19:24 -0400)]
"destroy" method -> "finalize" method
"Finalize" is more symmetric with "initialize" ("init"), and "destroy"
implies that the object is completely destroyed, which might not be the
case eventually.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 28 Mar 2017 23:09:31 +0000 (19:09 -0400)]
Accept port connection method: take other port as parameter
This can be useful, as we already have it at this point when calling the
method, to inspect the other port and its component.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 28 Mar 2017 18:14:42 +0000 (14:14 -0400)]
Visibility: split graph API into public and private interfaces
The goal here is to avoid that a component or a graph user calls methods
that do not logically apply to itself, but should according to the API's
intention. For example, a component should not be able to remove the
ports of another component, and a graph user should not be able to
create a notification iterator directly.
Therefore the existing bt_component, bt_port, bt_connection, and
bt_notification_iterator objects are considered "public": any user can
call their API.
The user-defined methods (callbacks) now accept private versions of
those objects:
* bt_private_component
* bt_private_port
* bt_private_connection
* bt_private_notification_iterator
This ensures that only the user which is called back with those private
objects can call specific, private methods, which start with the
`bt_private_` prefix.
For example, any user can get a sink component's input port by name with
bt_component_sink_get_input_port(). This function accepts a bt_component
object parameter and returns a bt_port object. However, only the called
back user can get its private version of the port with the equivalent
bt_private_component_sink_get_input_private_port(). This function
accepts a bt_private_component object parameter and returns a
bt_private_port object. The private port API has the
bt_private_port_remove_from_component() function: thus only the owner
component of a port can remove it from itself.
Each private API has a bt_OBJ_from_private_OBJ() function which
returns a _new reference_ to the equivalent public object of a private
object to be able to call public functions.
With all this, if a user calls a private function with a public object,
the compiler emits a warning similar to:
warning: assignment from incompatible pointer type
With GCC and Clang, you can make this warning an error with:
-Werror=incompatible-pointer-types
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Sat, 25 Mar 2017 05:27:24 +0000 (01:27 -0400)]
Add graph event listeners
With this patch you can add event listeners to a graph:
* Port added to a graph's component. Note that this is only called when
the component belongs to a graph. It is not called for existing ports
when the component becomes part of a graph (with bt_graph_connect()).
* Port removed from a graph's component.
* Graph component's port connected.
* Graph component's port disconnected.
All those listeners return `void` because they cannot "fail". In the
worst case they should just abort. Another way would be to set a flag
at some place and read it once you have called bt_graph_run() or
bt_graph_consume().
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 23 Mar 2017 21:12:40 +0000 (17:12 -0400)]
Allow a component to remove a port and any user to disconnect one
This patch allows a component to remove one of its own ports at anytime.
Removing a connected port has the side effect of also disconnecting it.
This patch also allows any user (graph's creator or component, for
example) to disconnect a component's port.
A component can remove one of its port with
bt_port_remove_from_component(). This function only needs the port
object because the removing component is its parent. Note that this
function does not check in any way that the removing component is the
parent of the port to remove from its parent, so any component or even
the graph's user could remove any component's port. This is prohibited
by the API's contract, but it's not enforced. It should be documented
accordingly.
You can disconnect a port with bt_port_disconnect(). When you call this
function, both upstream and downstream ports of the port's current
connection are disconnected, that is, they become available again for
bt_graph_connect().
When a port is disconnected with bt_port_remove_from_component(), the
parent component's of the connection's other port is called back ("port
disconnected" method) to notify it that one of its ports lost a
connection. Then the component is free to keep this port as an available
port, or to remove it too.
When a port is disconnected with bt_port_disconnect(), both
components are called back ("port disconnected method").
The existing iterators on a connection which is dropped (ports are
disconnected) are left as is: they still work.
The "new connection" optional method is renamed to "accept port
connection": this method can return
BT_COMPONENT_STATUS_REFUSE_PORT_CONNECTION to explicitly refuse the
new connection.
With this patch, only one connection per port is allowed. This
simplifies the model. One-to-many connections can be replaced with
a "tee" filter. Thus this patch removes the concept of a maximum number
of connections per port (it's always one).
With this patch, a component can add a new port at anytime, not just
during its initialization.
Also in this patch: the input and output ports are part of the
bt_component object. The public API is not changed. This avoids some
code duplication.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 8 Mar 2017 19:00:07 +0000 (14:00 -0500)]
Put Python plugin support in a separate shared object
This is to make the work of packagers easier for libbabeltrace not
to depend on libpython, even if you need Python plugin support.
With this patch, python.c's constructor tries to open the new Python
plugin provider shared object. If it fails, Python plugin support is
disabled, but shared object plugins can still be loaded.
The new BUILT_IN_PYTHON_PLUGIN_SUPPORT configure variable, if set to 1,
makes the build embed the Python plugin provider into libbabeltrace
(like before this patch).
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 7 Mar 2017 15:04:58 +0000 (10:04 -0500)]
Add `babeltrace convert` argument tests
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 3 Mar 2017 05:13:36 +0000 (00:13 -0500)]
CLI: add `run` command and make `convert` command use it
This patch adds the babeltrace(1) command `run` which is essentially the
same as the current `convert` command without the magic stuff (legacy
options and their interesting constraints), plus this:
* You can set the name of a component by prefixing the --source,
--filter, or --sink option's argument with `NAME:`. You can escape
`.`, `:`, and `\` with `\`. Example:
--source=instance:ctf.fs --sink=my-text:text.text
--filter='my\.name:my\.plugin.my\.comp\.cls'
This is the equivalent of:
--source=ctf.fs --name=instance --sink=text.text --name=my-text
--filter='my\.plugin.my\.comp\.cls' --name my.name
The --name option still exists: it sets the current name of the
current component, so that, with this:
--sink=a:b.c --name=salut --name=bonjour
the effective name of this component is `bonjour`.
* You can use the new --key and --value options to add one string
initialization parameter to the current component. The name of this
parameter is given by the last --key option's argument, and its value
by the --value option's argument. Examples:
--source=ctf.fs --key=path --value ~/lttng-traces/...
--key=path --source=ctf.fs --value ~/lttng-traces/...
This is useful to not care about the specific escaping required by the
--params option format for a string parameter, to use any parameter
name, and to use shell expansion for the value (tilde expansion for
example, as --path is not a `run` command's option).
The `convert` command is now a specialization of the `run` command,
specific to a _conversion graph_ (whereas the terminology for what you
build with the `run` command is a _processing graph_). The conversion
graph that the `convert` command builds always has the following
topology:
sources -> utils.muxer -> [utils.trimmer] -> lttng-utils.debug-info -.
.-------------------------------------------------------------'
'-> [user filters] -> sinks
There's no more --connect option in the `convert` command. When you run
the `convert` command, it prepares the equivalent `run` command
arguments and invokes the latter directly. However if `-o ctf-metadata`
exists, a new "print CTF metadata" command configuration is created, and
if `-o lttng-live` exists with a URL missing a session name, a new
"print LTTng-live sessions" command configuration is created.
New `convert` options are:
--color=(always | auto | never)
Adds the `color` initialization parameter set to the corresponding
value to the implicit text.text sink component.
--no-debug-info
Do not add an implicit lttng-utils.debug-info filter component to
the filter chain. The lttng-utils.debug-info filter always exists
otherwise.
--run-args
Instead of invoking the `run` command, prints to the standard output
the equivalent `run` command arguments, formatted for a shell. This
means you can copy the whole output and use it as is to execute
`babeltrace run` without escaping, so that the two following
commands are equivalent:
babeltrace convert [...]
babeltrace run $(babeltrace convert --run-args [...])
You can also use this trick to augment the conversion graph with the
`run` command, for example:
babeltrace run --base-params common-to-each-convert-comp=23
$(babeltrace convert --run-args /path/to/trace
--source=src:my.source [...])
--reset-base-params
--sink=special-sink:my.sink
--connect=src.some-port:special-sink
--run-args-0
Like --run-args, but does not escape the individual arguments for a
shell, and uses the null character as the argument delimiter. This
is useful to get the list of equivalent raw arguments outside a
shell (in a user program), or to use `xargs -0`.
--url=URL
Set the `url` initialization parameter of the current user component
to URL.
Lost `convert` options are:
* --base-params
* --connect
* --reset-base-params
Note that --clock-force-correlate sets the boolean `force-correlate`
initialization parameter of the implicit utils.muxer filter to true.
The necessary conditions to instantiate implicit components are now:
* Implicit ctf.fs source: any of:
* Leftover argument and no --input-format
* Leftover argument and --input-format=ctf
* Leftover argument and --clock-offset
* Leftover argument and --clock-offset-ns
* Leftover argument and --stream-intersection
* Implicit ctf.lttng-live source: any of:
* Leftover argument and --input-format=lttng-live
* Implicit utils.muxer filter: always exists
* Implicit utils.trimmer filter: any of:
* --begin
* --end
* --timerange
* Implicit lttng-utils.debug-info filter:
* No --no-debug-info
* Implicit utils.dummy sink:
* --output-format=dummy
* Implicit text.text sink: any of:
* No --output-format and no --sink
* --output-format=text
* --clock-cycles
* --clock-date
* --clock-gmt
* --clock-seconds
* --color
* --fields
* --names
* --no-delta
* --output
The default names of the implicit components, given by the `convert`
command, to be used by the `run` command, are:
* Implicit ctf.fs source: `ctf`
* Implicit ctf.lttng-live source: `lttng-live`
* Implicit utils.muxer filter: always exists `mux`
* Implicit utils.trimmer filter: `trim`
* Implicit lttng-utils.debug-info filter: `debug-info`
* Implicit utils.dummy sink: `dummy`
* Implicit text.text sink: `text`
If any name is already taken by an explicit user component (--source,
--filter, and --sink), the rule is to try `NAME-0`, then `NAME-1`, then
`NAME-2`, etc. The same rule applies to unnamed user components.
Implicit components are always automatically named after automatically
naming the unnamed user components.
This patch also adds the following:
* Connection cycle detection in babeltrace-cfg-connect.c.
* bt_common_string_until() which returns a substring from a given point
to one of a set of terminating characters, possibly with escapable
characters.
* bt_common_shell_quote() which quotes a string (if necessary) for a
shell.
* bt_common_string_is_printable() which checks if a string contains
only printable characters, including whitespaces.
* bt_common_parse_lttng_live_url() to parse a LTTng-live URL. This is
needed in both the LTTng-live plugin itself and babeltrace-cfg.c to
know if the command is a `run` command or a "print LTTng-live
sessions" command.
* The --name, --source, --filter, --sink, and --connect arguments are
checked with bt_common_string_is_printable().
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 10 Mar 2017 21:59:52 +0000 (16:59 -0500)]
Fix: missing reference release in field reset
A missing bt_put() on in reset_structure_field() will cause
the reset fields to be leaked. This function is used on fields
which are automatically managed by the writer such as
timestamp_begin, timestamp_end, packet_size, and content_size.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 10 Mar 2017 19:29:15 +0000 (14:29 -0500)]
Fix: bt_connection_get_*_port() must take a reference on port
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 10 Mar 2017 19:19:43 +0000 (14:19 -0500)]
Fix: only remove newly added components on connection error
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 10 Mar 2017 19:16:46 +0000 (14:16 -0500)]
Fix: only add components to graph if they are not already present
bt_graph_connect() creates a connection and adds the components
on both ends of the connection to the graph. This fix ensures
that the components are not added a second time if multiple
connections involving the same components are established.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 10 Mar 2017 19:11:29 +0000 (14:11 -0500)]
Fix: release the reference held by a child to its former parent
The bt_object class' bt_object_set_parent() increments the reference
count of the parent since the child now holds a reference to it.
This assumes that a child is externally visible when this call is
made (as is always the case currently).
The parent of a child can be changed in the future. In this case,
we want to ensure the reference it held to its former parent is
released.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 22 Feb 2017 15:17:45 +0000 (10:17 -0500)]
babeltrace-cfg.c: do not allow duplicate -v option
Otherwise, `babeltrace -vvv list-plugins` passes and tries to open the
`list-plugins` directory with ctf.fs.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 22 Feb 2017 15:20:46 +0000 (10:20 -0500)]
babeltrace-cfg.c: add "Command line error" prefix to error messages
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 22 Feb 2017 15:26:00 +0000 (10:26 -0500)]
babeltrace-cfg.c: do not infer text.text sink with -v
Having the -v option passed to the `convert` command is not a
sufficient criterion to infer a text.text sink component, because -v
is also a general option which applies to the whole command. Without
this patch, you can't run this:
babeltrace -v /path/to/trace -o my.sink
because the configurator sees both an implicit text.text sink (caused by
-v) and an explicit my.sink sink.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 22 Feb 2017 16:08:32 +0000 (11:08 -0500)]
babeltrace(1): add --filter option to create filter components
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 22 Feb 2017 18:08:12 +0000 (13:08 -0500)]
babeltrace(1): escape '.' in PLUGIN.COMPCLS arguments
Now you can have dots in your plugin or component class names:
babeltrace --source 'hello\.world.my\.component\.class'
Also, escape `\` with `\\`.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Wed, 22 Feb 2017 18:18:53 +0000 (13:18 -0500)]
common.c: bt_common_colors_supported(): add a few supported terminal prefixes
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 23 Feb 2017 14:36:05 +0000 (09:36 -0500)]
Return component port counts by parameter
The rest of the API separates the return of success/error from
the return of the item count. This change makes the component
API consistent with the style used elsewhere.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 23 Feb 2017 13:45:01 +0000 (08:45 -0500)]
Remove component prefix from graph, connection and port filenames
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
This page took 0.057324 seconds and 4 git commands to generate.