babeltrace.git
8 years agoCleanup: lift function call outside of condition
Jérémie Galarneau [Mon, 22 Feb 2016 20:18:02 +0000 (15:18 -0500)] 
Cleanup: lift function call outside of condition

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add bt_ctf_clock_ns_from_value() utility
Philippe Proulx [Fri, 19 Feb 2016 02:34:12 +0000 (21:34 -0500)] 
ir: add bt_ctf_clock_ns_from_value() utility

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: move clock value accessors to writer header
Philippe Proulx [Fri, 19 Feb 2016 02:16:59 +0000 (21:16 -0500)] 
ir: move clock value accessors to writer header

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agotest_ctf_writer.c: test writer/non-writer clock value accessors
Philippe Proulx [Fri, 19 Feb 2016 02:15:39 +0000 (21:15 -0500)] 
test_ctf_writer.c: test writer/non-writer clock value accessors

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: move bt_ctf_stream_class_set_clock() to writer header
Philippe Proulx [Fri, 19 Feb 2016 02:15:05 +0000 (21:15 -0500)] 
ir: move bt_ctf_stream_class_set_clock() to writer header

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: disable clock value accessors in non-writer mode
Philippe Proulx [Fri, 19 Feb 2016 02:00:31 +0000 (21:00 -0500)] 
ir: disable clock value accessors in non-writer mode

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add bt_ctf_event_get_clock_value()
Philippe Proulx [Tue, 16 Feb 2016 07:23:26 +0000 (02:23 -0500)] 
ir: add bt_ctf_event_get_clock_value()

This patch adds a hash table of clocks to their current values (in
cycles) to the stream object. bt_ctf_stream_update_clock_value() can
be called to update the registered clock value of a given stream
from the current value of a given integer field, of which the type
is mapped to a clock. The function handles the case where the size of
the field is less than 64 bits and can make the current clock value
wrap once per call.

This patch also adds a similar hash table to the event object.
bt_ctf_event_register_stream_clock_values() is to be called by
internal code to take a snapshot of all the registered clock values
of the event's stream. The copied clock values can be retrieved by
the user later using bt_ctf_event_get_clock_value(), which should
return the value that a given clock had when the event was emitted.

This is intended to be used by sink components since the value
of a clock changes as events are created, whereas it is common
situation to hold an event and later need the clock values when
this event was emitted.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: bt_ctf_stream_append_event(): do not append event with parent
Philippe Proulx [Thu, 18 Feb 2016 00:21:26 +0000 (19:21 -0500)] 
ir: bt_ctf_stream_append_event(): do not append event with parent

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: remove unused event header/context fields from stream
Philippe Proulx [Thu, 18 Feb 2016 00:12:23 +0000 (19:12 -0500)] 
ir: remove unused event header/context fields from stream

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add optional name property to stream
Philippe Proulx [Wed, 17 Feb 2016 23:57:04 +0000 (18:57 -0500)] 
ir: add optional name property to stream

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agotest_ctf_writer.c: test bt_ctf_stream_create()
Philippe Proulx [Wed, 17 Feb 2016 21:55:22 +0000 (16:55 -0500)] 
test_ctf_writer.c: test bt_ctf_stream_create()

This new test function tests the creation of both a writer
(when the stream class's trace was itself created by a writer)
and a non-writer stream.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: make bt_ctf_stream_create() public
Philippe Proulx [Wed, 17 Feb 2016 21:52:05 +0000 (16:52 -0500)] 
ir: make bt_ctf_stream_create() public

This patch also removes bt_ctf_trace_create_stream(), so that the
only way to create a stream in non-writer mode is using
bt_ctf_stream_create(). Only the stream class is needed, and since
it needs to be part of a trace before calling the function, the
created stream gets associated to this trace on creation.

The function also knows if the stream class's trace was created by
a CTF writer thanks to a new flag. With this, it's able to create
a stream file and set the stream's FD if it's a writer stream.
Otherwise the FD is set to -1 and this stream cannot be passed to
stream writer functions like bt_ctf_stream_append_event() and
bt_ctf_stream_append_discarded_events(), for example.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agotest_ctf_writer.c: test stream class clock -> trace clock
Philippe Proulx [Tue, 16 Feb 2016 22:20:54 +0000 (17:20 -0500)] 
test_ctf_writer.c: test stream class clock -> trace clock

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: bt_ctf_trace_add_stream_class(): add stream class's clock
Philippe Proulx [Tue, 16 Feb 2016 19:50:21 +0000 (14:50 -0500)] 
ir: bt_ctf_trace_add_stream_class(): add stream class's clock

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: remove unused bt_ctf_event_copy()
Philippe Proulx [Tue, 16 Feb 2016 21:48:08 +0000 (16:48 -0500)] 
ir: remove unused bt_ctf_event_copy()

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: verify stream mode in stream functions
Philippe Proulx [Mon, 15 Feb 2016 21:24:01 +0000 (16:24 -0500)] 
ir: verify stream mode in stream functions

Some functions are disabled in CTF writer mode and vice versa.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: freeze event on bt_ctf_stream_append_event()
Philippe Proulx [Sat, 13 Feb 2016 02:15:16 +0000 (21:15 -0500)] 
ir: freeze event on bt_ctf_stream_append_event()

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add bt_ctf_event_freeze()
Philippe Proulx [Sat, 13 Feb 2016 02:12:03 +0000 (21:12 -0500)] 
ir: add bt_ctf_event_freeze()

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add bt_ctf_field_freeze()
Philippe Proulx [Sat, 13 Feb 2016 01:50:40 +0000 (20:50 -0500)] 
ir: add bt_ctf_field_freeze()

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: remove bt_ctf_trace_get() and bt_ctf_trace_put()
Philippe Proulx [Sat, 13 Feb 2016 01:01:15 +0000 (20:01 -0500)] 
ir: remove bt_ctf_trace_get() and bt_ctf_trace_put()

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: move writer-specific declarations to writer header
Philippe Proulx [Fri, 19 Feb 2016 23:07:25 +0000 (18:07 -0500)] 
ir: move writer-specific declarations to writer header

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: remove event header accessors from stream
Philippe Proulx [Sat, 13 Feb 2016 00:45:05 +0000 (19:45 -0500)] 
ir: remove event header accessors from stream

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add bt_ctf_event_set_packet()
Philippe Proulx [Fri, 12 Feb 2016 23:55:50 +0000 (18:55 -0500)] 
ir: add bt_ctf_event_set_packet()

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add public bt_ctf_packet object
Philippe Proulx [Wed, 17 Feb 2016 22:07:29 +0000 (17:07 -0500)] 
ir: add public bt_ctf_packet object

The purpose of a bt_ctf_packet object is to hold the packet
header and packet context fields.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: move the stream event ctx field to the event
Philippe Proulx [Fri, 12 Feb 2016 00:27:42 +0000 (19:27 -0500)] 
ir: move the stream event ctx field to the event

Just like the event header, the stream event context field has
its type defined in the stream class, but its data can be different
in each event. The current strategy to sample and copy the current
stream's stream event context and append the copies to an array
is fine as far as CTF writer goes, but this array would not be
populated strictly using the event in the context of Babeltrace
components.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: remove unused event_headers member from struct bt_ctf_stream
Philippe Proulx [Thu, 11 Feb 2016 23:40:18 +0000 (18:40 -0500)] 
ir: remove unused event_headers member from struct bt_ctf_stream

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: validate integer FT's mapped clock against signedness
Philippe Proulx [Thu, 11 Feb 2016 23:38:10 +0000 (18:38 -0500)] 
ir: validate integer FT's mapped clock against signedness

It is invalid for a signed integer field type to be mapped to a clock.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: clock: use value in cycles instead of ns
Philippe Proulx [Tue, 16 Feb 2016 01:13:02 +0000 (20:13 -0500)] 
ir: clock: use value in cycles instead of ns

The value of a clock in CTF is in cycles, not in nanoseconds.
Here, bt_ctf_clock_get_time() and bt_ctf_clock_set_time() are left
as utilities to converter from/to the clock value in cycles.
Note that this conversion, unless the frequency is exactly 1 GHz,
implies casting 64-bit integers to doubles so it is possible to
lose accuracy.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: freeze mapped clock when freezing integer FT
Philippe Proulx [Thu, 11 Feb 2016 23:23:26 +0000 (18:23 -0500)] 
ir: freeze mapped clock when freezing integer FT

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: freeze clocks when freezing a trace
Philippe Proulx [Thu, 11 Feb 2016 23:15:25 +0000 (18:15 -0500)] 
ir: freeze clocks when freezing a trace

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: split event files into event and event-class files
Philippe Proulx [Tue, 9 Feb 2016 01:05:55 +0000 (20:05 -0500)] 
ir: split event files into event and event-class files

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: rename event-fields/event-types -> fields/field-types
Philippe Proulx [Tue, 9 Feb 2016 00:32:00 +0000 (19:32 -0500)] 
ir: rename event-fields/event-types -> fields/field-types

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: rename ctf_type_id -> bt_ctf_type_id
Philippe Proulx [Mon, 8 Feb 2016 21:03:22 +0000 (16:03 -0500)] 
ir: rename ctf_type_id -> bt_ctf_type_id

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: rename ctf_string_encoding -> bt_ctf_string_encoding
Philippe Proulx [Mon, 8 Feb 2016 08:43:06 +0000 (03:43 -0500)] 
ir: rename ctf_string_encoding -> bt_ctf_string_encoding

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: rename bt_ctf_field_path_get_root()
Philippe Proulx [Mon, 8 Feb 2016 08:26:40 +0000 (03:26 -0500)] 
ir: rename bt_ctf_field_path_get_root()

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: rename bt_ctf_node -> bt_ctf_scope
Philippe Proulx [Mon, 8 Feb 2016 08:24:13 +0000 (03:24 -0500)] 
ir: rename bt_ctf_node -> bt_ctf_scope

"scope" is more conforming to the CTF nomenclature than "node".

Note that this changes the Babeltrace 1.x API, but this API is
expected to be hidden from the user starting from Babeltrace 2.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add public bt_ctf_field_path object
Philippe Proulx [Mon, 8 Feb 2016 22:56:15 +0000 (17:56 -0500)] 
ir: add public bt_ctf_field_path object

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: rename bt_ctf_field_type_variant_set_tag()
Philippe Proulx [Mon, 8 Feb 2016 07:43:49 +0000 (02:43 -0500)] 
ir: rename bt_ctf_field_type_variant_set_tag()

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoAdd validation/resolving tests
Philippe Proulx [Mon, 8 Feb 2016 06:41:51 +0000 (01:41 -0500)] 
Add validation/resolving tests

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agotest_bt_values.c: fix top comment
Philippe Proulx [Mon, 8 Feb 2016 06:40:44 +0000 (01:40 -0500)] 
test_bt_values.c: fix top comment

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: freeze attributes when freezing an event
Philippe Proulx [Mon, 8 Feb 2016 07:38:25 +0000 (02:38 -0500)] 
ir: freeze attributes when freezing an event

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoFix scan-build warning of potential NULL dereference
Jérémie Galarneau [Fri, 19 Feb 2016 05:26:35 +0000 (00:26 -0500)] 
Fix scan-build warning of potential NULL dereference

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: refactor FT validation and resolving
Philippe Proulx [Mon, 8 Feb 2016 07:21:40 +0000 (02:21 -0500)] 
ir: refactor FT validation and resolving

This patch refactors the CTF IR field type validation and the
resolving of variant/sequence field types.

First, validation is completely decoupled from the objects
containing the field types. In other words, validation does not
know the exact event class, stream class, or trace in which the
field types to validate are or are supposed to be. Only field types
are given to the validation function, which is isolated in its
own file.

To make the life of the resolving engine easier, all scopes are
deep-copied before reaching it. This is done during the validation
process. This ensures that no two field types in the whole
hierarchy refer to the same object in memory. This greatly simplifies
the resolving process, while adding insignificant overhead since
this is only done once. The copies do not replace the original types
until we are sure that the object to be validated (event class, stream
class, or trace) is valid and will be frozen. When it is the case, a
valid flag is set, which avoids validating the object again in the
future when calling the same function. If the object is not valid,
then the copies of its field types are discarded, and the function
fails.

Validation is done at three important places:

  * bt_ctf_event_create(): when creating an event, because we don't
    want the user to obtain an event linked to an invalid event
    class. It is possible to create an event from an even class
    which has a parent stream class which has no parent trace. In
    this case, no variant/sequence field type in all the field types
    of the stream class and event class can be resolved to the
    trace packet header field type.
  * bt_ctf_stream_class_add_event_class(): when adding an event class
    to a stream class AND when the stream class has a parent trace,
    because in this case the whole hierarchy exists and everything
    is frozen, so it makes sense that everything in there should be
    valid. If the stream class has no parent trace yet, then the
    validation will be done when adding it to a trace using the
    following function. This is because the added event class could
    contain a variant/sequence field type which resolves to a field
    type of the trace packet header field type, contained in the trace
    object.
  * bt_ctf_trace_add_stream_class(): when adding a stream class to
    a trace, because in this case both the trace and the stream
    class will be frozen, and hence they should be valid because they
    cannot be changed anymore.

The test_ctf_writer test is updated here to follow those changes. Since
the field types can be copied when calling one of the above functions,
we cannot create a field out of an original field type and then compare
its address to a field obtained from the created event or stream: we must
put the original field types and obtain the new ones from the event class
or stream class. The contents of field types are compared using
bt_ctf_field_type_compare() instead.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: allow to set variant FT's tag FT even when frozen
Philippe Proulx [Mon, 8 Feb 2016 07:14:51 +0000 (02:14 -0500)] 
ir: allow to set variant FT's tag FT even when frozen

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add internal bt_ctf_field_path_clear()
Philippe Proulx [Mon, 8 Feb 2016 07:11:20 +0000 (02:11 -0500)] 
ir: add internal bt_ctf_field_path_clear()

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: validate field types on field creation
Philippe Proulx [Mon, 8 Feb 2016 07:05:35 +0000 (02:05 -0500)] 
ir: validate field types on field creation

This patch removes the validation that is done when adding
fields to structure/variant field types, and when creating
array/sequence field types with specific element types. This
is needed because a variant with an unset tag type may be
created and thus exist in any of those compound types. This
variant will be resolved when the time comes.

A valid field type is still required when creating a concrete
field out of it and when serializing it.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agovalues: fix indentation
Philippe Proulx [Thu, 4 Feb 2016 22:11:41 +0000 (17:11 -0500)] 
values: fix indentation

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoref.h: fix doc
Philippe Proulx [Thu, 4 Feb 2016 18:22:27 +0000 (13:22 -0500)] 
ref.h: fix doc

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agovalues: add `_empty` prefix when relevant
Philippe Proulx [Thu, 4 Feb 2016 08:11:07 +0000 (03:11 -0500)] 
values: add `_empty` prefix when relevant

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoBT_MOVE(): call bt_put(_dst)
Philippe Proulx [Mon, 1 Feb 2016 13:21:00 +0000 (08:21 -0500)] 
BT_MOVE(): call bt_put(_dst)

It makes no semantical sense to move a reference from a variable
to another without first putting the destination, if it has one.
Otherwise this would be a lost reference, unless it was moved
elsewhere or put manually using bt_put() previously, which is why
BT_MOVE() should always be used for move semantics and BT_PUT() is
preferred to bt_put() to avoid this situation.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoFix: bt_ctf_field_type_floating_point_copy(): set self references
Philippe Proulx [Sat, 30 Jan 2016 06:22:57 +0000 (01:22 -0500)] 
Fix: bt_ctf_field_type_floating_point_copy(): set self references

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoAdd bt_ctf_field_type_compare(): field type deep comparison
Philippe Proulx [Sat, 30 Jan 2016 06:19:01 +0000 (01:19 -0500)] 
Add bt_ctf_field_type_compare(): field type deep comparison

The new bt_ctf_field_type_compare() function compares two field
types recursively.

The tag field type of variant field types are not compared; only
the tag strings are compared.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: event-types-internal.h: fix comment
Philippe Proulx [Sat, 30 Jan 2016 06:29:39 +0000 (01:29 -0500)] 
ir: event-types-internal.h: fix comment

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: add user byte order to integer/float types
Philippe Proulx [Sat, 30 Jan 2016 06:27:18 +0000 (01:27 -0500)] 
ir: add user byte order to integer/float types

The `byte_order` field of an integer/float field type gets
replaced by `BIG_ENDIAN` or `LITTLE_ENDIAN` if it's set to
0 (native) once the type becomes the descendant of a
trace. However, once this is set, there is no way to know
that the user intended this field type's byte order to be
native. The metadata string, for example, will never contain
a byte order set to `native` with this behaviour.

This patch adds a `user_byte_order` field to the internal
integer and floating point number field type structures. This
field is the one set by and returned to the user using the
CTF IR API. Internally, the declaration's `byte_order` field
is still used for caching the resolved byte order.

This patch also modifies bt_ctf_stream_class_set_byte_order()
so that it cannot fail. This is an internal, hidden function
which should not be called with invalid parameters. Assertions
are inserted to verify this.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agotest_ctf_writer.c: assert() unchecked return values
Philippe Proulx [Sat, 30 Jan 2016 05:56:30 +0000 (00:56 -0500)] 
test_ctf_writer.c: assert() unchecked return values

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoir: bt_ctf_stream_class_set_byte_order(): EC is already frozen
Philippe Proulx [Fri, 31 Jul 2015 18:22:25 +0000 (14:22 -0400)] 
ir: bt_ctf_stream_class_set_byte_order(): EC is already frozen

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoPython bindings: remove unnecessary semi-colons
Jérémie Galarneau [Wed, 17 Feb 2016 03:36:24 +0000 (22:36 -0500)] 
Python bindings: remove unnecessary semi-colons

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoPython bindings: missing error check in Clock precision setter
Jérémie Galarneau [Wed, 17 Feb 2016 03:27:21 +0000 (22:27 -0500)] 
Python bindings: missing error check in Clock precision setter

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoPython bindings: accomodate API changes introduced for negative time
Jérémie Galarneau [Wed, 17 Feb 2016 03:26:26 +0000 (22:26 -0500)] 
Python bindings: accomodate API changes introduced for negative time

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoRemove stale comment from event destructor
Jérémie Galarneau [Mon, 15 Feb 2016 19:46:11 +0000 (14:46 -0500)] 
Remove stale comment from event destructor

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoHandle negative time and offset from Epoch
Mathieu Desnoyers [Tue, 9 Feb 2016 18:27:24 +0000 (13:27 -0500)] 
Handle negative time and offset from Epoch

Handle cases where a trace have a negative offset from Epoch.
If Epoch is arbitrary (e.g. embedded system starting at 0, without any
network access), the "0" can be used as correlation point between
various components, and some components could start before the
correlation point. Therefore, especially in traces where the time is
meant to be shown in nanoseconds or cycles from the correlation point,
it also makes sense to have a negative time value.

It introduces API-breaking changes in the C and Python APIs, since we
need to be able to return negative time values, which were previously
used as errors (-1ULL).

The --offset and --offset-ns command line parameters can now take
negative offset (seconds and nanoseconds) values too.

The [sec.ns] format is used as fallback so we don't attempt to pass
a negative time value to POSIX time-formatting APIs.

This also fixes an inaccurate return value in an error path of
bt_ctf_event_populate_event_header().

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoFix: CTF parser: accept signed constant literals
Mathieu Desnoyers [Tue, 9 Feb 2016 18:27:23 +0000 (13:27 -0500)] 
Fix: CTF parser: accept signed constant literals

Those are inaccurately rejected by the parser due to this bug. This
affects environment variables and loglevel values.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoFix: out-of-bound memory access
Mathieu Desnoyers [Tue, 9 Feb 2016 18:27:22 +0000 (13:27 -0500)] 
Fix: out-of-bound memory access

We need to check for end of stream before updating the current packet
index, else we access memory beyond the last stream packet index array
element. This fixes valgrind errors which appear at commit

480ef05 Support ctf index version 1.1

due to change of the element size, but actually caused by commit

2654fe9 Fix: events discarded timing inaccuracy

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoFix out-of-tree tests: seek tests can't find traces
Jérémie Galarneau [Wed, 10 Feb 2016 16:25:33 +0000 (11:25 -0500)] 
Fix out-of-tree tests: seek tests can't find traces

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoTests: Fix test plans and do not skip on fail
Michael Jeanson [Mon, 25 Jan 2016 21:00:41 +0000 (16:00 -0500)] 
Tests: Fix test plans and do not skip on fail

Use fixed count test plans in all tests and make sure we do not skip
some tests on initialisation failures.

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoFix: Check return value of fpathconf
Michael Jeanson [Fri, 4 Dec 2015 22:03:12 +0000 (17:03 -0500)] 
Fix: Check return value of fpathconf

Current glibc has a bug in fpathconf(fd, _PC_NAME_MAX) where it will
fail with a 32bit userland on a 64bit kernel and where the filesystem
has a large block count, see glibc bug #18675.

In any case, we should check this return value because on a failure we
we don't allocate enough memory for dirent and then overflow on the
readdir_r call.

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoFix: BABELTRACE_HAVE_LIBC_UUID misspelled as BT_HAVE_LIBC_UUID
Michael Jeanson [Thu, 5 Nov 2015 17:52:37 +0000 (12:52 -0500)] 
Fix: BABELTRACE_HAVE_LIBC_UUID misspelled as BT_HAVE_LIBC_UUID

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoFix: check for socket lib before function checks
Michael Jeanson [Tue, 9 Feb 2016 23:50:22 +0000 (18:50 -0500)] 
Fix: check for socket lib before function checks

On platforms that require extra libraries to link socket apps,
running this macro before the function checks will lead to
socket functions being correctly detected.

Signed-off by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoPort: Add OSX libuuid compat
Michael Jeanson [Thu, 5 Nov 2015 17:51:55 +0000 (12:51 -0500)] 
Port: Add OSX libuuid compat

OSX has the libuuid symbols built in the system libraries.

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoPort: Add OSX mman.h compat
Michael Jeanson [Thu, 5 Nov 2015 17:51:54 +0000 (12:51 -0500)] 
Port: Add OSX mman.h compat

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoPort: Add OSX endian.h compat
Alexis Martin [Thu, 5 Nov 2015 17:51:53 +0000 (12:51 -0500)] 
Port: Add OSX endian.h compat

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoCleanup: remove unused label 'end'
Michael Jeanson [Thu, 15 Oct 2015 20:40:22 +0000 (16:40 -0400)] 
Cleanup: remove unused label 'end'

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoFix: Wrong variable checked for allocation failure
Jérémie Galarneau [Tue, 26 Jan 2016 20:45:38 +0000 (15:45 -0500)] 
Fix: Wrong variable checked for allocation failure

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoImplement new CTF-IR reference counting scheme
Jérémie Galarneau [Tue, 19 Jan 2016 18:40:30 +0000 (13:40 -0500)] 
Implement new CTF-IR reference counting scheme

See doc/ref-counting.md for the rationale and description of this
architecture change.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoUse bt_put instead of bt_ctf_field_put as release callback
Jérémie Galarneau [Tue, 19 Jan 2016 18:39:37 +0000 (13:39 -0500)] 
Use bt_put instead of bt_ctf_field_put as release callback

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoClean-up bt_put() implementation
Jérémie Galarneau [Tue, 19 Jan 2016 18:24:12 +0000 (13:24 -0500)] 
Clean-up bt_put() implementation

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoObject: Acquire reference to parent on first acquired reference
Jérémie Galarneau [Tue, 19 Jan 2016 18:22:59 +0000 (13:22 -0500)] 
Object: Acquire reference to parent on first acquired reference

See doc/ref-counting.md

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoTests: Add CTF-IR reference counting test
Jérémie Galarneau [Tue, 5 Jan 2016 23:07:05 +0000 (18:07 -0500)] 
Tests: Add CTF-IR reference counting test

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoReturn pointer when using bt_get()
Jérémie Galarneau [Mon, 30 Nov 2015 14:03:04 +0000 (09:03 -0500)] 
Return pointer when using bt_get()

bt_get() now returns the pointer it was passed to replace code
of the form:

bt_get(my_thingy);
another_object->thingy = my_thingy;

to the shorter form

another_object->thingy = bt_get(my_thingy);

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoAdd internal bt_object reference count accessor
Jérémie Galarneau [Mon, 30 Nov 2015 14:01:16 +0000 (09:01 -0500)] 
Add internal bt_object reference count accessor

This accessor is added to facilitate the testing and validation
of the bt_object reference counting and aggregation mechanism.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoDocs: Document reference counting scheme implemented by Object
Jérémie Galarneau [Mon, 23 Nov 2015 23:25:38 +0000 (18:25 -0500)] 
Docs: Document reference counting scheme implemented by Object

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoAdd parented object memory management to the Object interface
Jérémie Galarneau [Mon, 23 Nov 2015 23:14:10 +0000 (18:14 -0500)] 
Add parented object memory management to the Object interface

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
8 years agoDo not use g_list_free_full
Jonathan Rajotte [Mon, 4 Jan 2016 22:04:23 +0000 (17:04 -0500)] 
Do not use g_list_free_full

Babeltrace supports glib all the way back to 2.22 and g_list_free_full
was introduced in 2.28.0 (and backported to 2.27.2).

The replacement is equivalent to the behaviour offered by a modern glib.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoUpdate version to 2.0.0-pre
Jérémie Galarneau [Wed, 4 Nov 2015 21:33:52 +0000 (16:33 -0500)] 
Update version to 2.0.0-pre

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoOnly compare "bison -y" to the basename of YACC variable
Jonathan Rajotte [Thu, 22 Oct 2015 21:07:02 +0000 (17:07 -0400)] 
Only compare "bison -y" to the basename of YACC variable

In the event that the YACC variable is set to a full path
this test fail even if it should not.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoPython bindings: work around Python 3.5 behaviour change
Jérémie Galarneau [Tue, 27 Oct 2015 20:10:03 +0000 (16:10 -0400)] 
Python bindings: work around Python 3.5 behaviour change

Python 3.5 changes the StopIteration exception clearing behaviour
when a generator finishes its iteration. This causes the
interpreter to errounously consider SWIG clean-up functions as
having "set an error".

This hack explicitly allocates and cleans up struct bt_iter_pos
instead of relying on SWIG auto-generated code which manages
the lifetime of temporary objects.

An investigation of the cause of this change is under way, but
at least this makes the bindings usable on Python 3.5 which is
being rolled-out in some distros.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoOutput a warning if packets are lost
Julien Desfossez [Thu, 20 Aug 2015 01:19:58 +0000 (21:19 -0400)] 
Output a warning if packets are lost

Depends on the packet_seq_num fields available in the CTF index v1.1
(produced by LTTng 2.8).
Same limitation as the events discarded information: if a stream is
split in multiple files, the counters might not report the appropriate
information for now.

Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoClean-up: Use memset() to initialize packet_index
Jérémie Galarneau [Mon, 26 Oct 2015 16:11:30 +0000 (12:11 -0400)] 
Clean-up: Use memset() to initialize packet_index

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoSupport ctf index version 1.1
Julien Desfossez [Thu, 20 Aug 2015 01:19:57 +0000 (21:19 -0400)] 
Support ctf index version 1.1

Produced by LTTng as of 2.8 (no compatibility breakage with 1.0)

Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoCleanup: Don't hardcode include paths unnecessarily
Jérémie Galarneau [Sat, 17 Oct 2015 19:19:05 +0000 (15:19 -0400)] 
Cleanup: Don't hardcode include paths unnecessarily

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoFix: include ctf-scanner-symbols.h to prefix yy_* symbols
Jérémie Galarneau [Sat, 17 Oct 2015 19:18:40 +0000 (15:18 -0400)] 
Fix: include ctf-scanner-symbols.h to prefix yy_* symbols

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoFix: missing includes break the out-of-tree build
Jérémie Galarneau [Fri, 16 Oct 2015 19:31:28 +0000 (15:31 -0400)] 
Fix: missing includes break the out-of-tree build

Addresses out-of-tree build breakage introduced by
commit 3842465694945829d76452ff83924aa0103c6293

Reported-by: Nathan Lynch <Nathan_Lynch@mentor.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoFix: memstream compat layer requires use of babeltrace_close_memstream
Mathieu Desnoyers [Thu, 15 Oct 2015 16:08:48 +0000 (12:08 -0400)] 
Fix: memstream compat layer requires use of babeltrace_close_memstream

In compatibility mode, the babeltrace compat layer around memstream
copies the file content into the buffer only at close.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agocompat send no SIGPIPE: multithread-safe
Mathieu Desnoyers [Thu, 15 Oct 2015 16:07:46 +0000 (12:07 -0400)] 
compat send no SIGPIPE: multithread-safe

The current implementation of the no-SIGPIPE send in the compatibility
layer has side-effects on multithreaded processes due to use of
sigaction(). Although multithread-safety is not strictly needed since
Babeltrace is single-threaded for now, there is no reason to keep this
limitation deeply rooted in a compatibility layer.

Use the multithreaded-safe algorithm to catch SIGPIPE implemented in
LTTng-UST for the write() system call for platforms that do not have
MSG_NOSIGNAL. It was originally implented in LTTng-UST as part of the
ring buffer wakeup. This is a re-implementation of this same algorithm
under MIT license. It uses signal masks and sigtimedwait.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoFix: truncated value in fgetc() usage
Jérémie Galarneau [Thu, 15 Oct 2015 17:43:06 +0000 (13:43 -0400)] 
Fix: truncated value in fgetc() usage

Reported by Coverity Scan:
CID 1327417:  API usage errors  (CHAR_IO)
Assigning the return value of "fgetc" to char "c" truncates
its value.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoFix: handle EINTR return value for bt_posix_fallocate
Mathieu Desnoyers [Tue, 13 Oct 2015 20:04:23 +0000 (16:04 -0400)] 
Fix: handle EINTR return value for bt_posix_fallocate

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoConfigure: Add missing checks
Michael Jeanson [Tue, 13 Oct 2015 19:29:58 +0000 (15:29 -0400)] 
Configure: Add missing checks

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoImplement bt_getline
Mathieu Desnoyers [Tue, 13 Oct 2015 18:56:44 +0000 (14:56 -0400)] 
Implement bt_getline

Implement our own getline to remove dependency from glibc 2.10+ and to
increase portability to other operating systems.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoFix: initialize getline linesize argument to 0
Mathieu Desnoyers [Tue, 13 Oct 2015 18:44:05 +0000 (14:44 -0400)] 
Fix: initialize getline linesize argument to 0

It works by accident because getline() reallocates based on the initial
pointer, but doing this falls into a blind spot of the Linux man pages.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
9 years agoFix: warning, may be used uninitialized
Michael Jeanson [Fri, 9 Oct 2015 21:16:37 +0000 (17:16 -0400)] 
Fix: warning, may be used uninitialized

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
This page took 0.042894 seconds and 4 git commands to generate.