git.efficios.com Git - deliverable/tracecompass.git/log

tmf: Use a CountDownLatch instead of Object.wait() in TmfEventThread

The current way to manage paused requests in TmfEventThread uses
a wait()/notify() on a custom Object. Since this object is not used
for anything else, we can replace its with a CountDownLatch, which
does the same thing without requiring an additional synchronized lock.

Change-Id: Ie69612769e3554d6e71d030230b409d8a0145450
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36516
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: test reading all downloaded traces once

Change-Id: I28e5feef958ab974f411d43ee6e98ed74ac0ca9b
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36179
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Make ITmfTrace.getEventAspects() @NonNull

It's easier to deal with empty collections than dealing with
null/non-null references (for Collection.addAll() and such).

Change-Id: I9e8d0091e099027b5da0c15881b73130b4ef2e9d
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/37121
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Hudson CI

ss: remove e.printStackTrace()s from HT_IO

Change-Id: Ieab2def800f88e50548d8f94b1a39c41e401247c
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/37047
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

tmf.analysis: Restore the "getCPU" for data-driven analyses

XML data-driven analyses used to use ITmfEvent.getSource() to get
the event's CPU. This was misleading, because trace types were free
to define "source" as something other than CPUs.

Now instead use a CpuAspect to get the CPU of an event. This
should restore support for getting the CPU of CTF traces, as it
did before.

Also added a new method to TmfTraceUtils to get the aspects matching
a specific class. The analysis modules will make use of this to get
the "cpu" aspects.

Change-Id: I88437b4a419e81aef36f2aa5298ff54f87242b23
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36366
Tested-by: Hudson CI

ctf: add trace to test ctf

this trace is the one from bug 446190, it contains headers that are 64 bit aligned.

Change-Id: I1ee228ee4f0635b940032f14bf6a080dc8a5c8f2
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36178
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

ctf: make uint_5t and 27t aligned on 1 bit

Change-Id: Idb8d1c96fcf64929f8b49af47910b49e86c2e0c3
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36162
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Add a TmfTraceUtils class for advanced getter methods

For instance, the various getModuleFromClass() methods should not
be part of the base interface, the trace should only expose a basic
getter. The more fancy functions can however be implemented in a
separate utility class.

Change-Id: I8d957971abfbcc7e3afe81f5db2dba43cbebe84d
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/37218
Tested-by: Hudson CI

ctf: improve testing of EventHeaderDeclaration

Add more subtle test cases

Change-Id: I5a61426a9f4358af522c84de5ae4a8a4ca0e773e
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36161
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

ctf: add test coverage

Change-Id: I405cded362e408fe50e1871143ba5e07069a62d3
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36160
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

ctf: add isBinaryEquivalent() to IDeclaration

This makes EventHeader substitutions more reliable.
Fixes bug 446190

Change-Id: I06f8c0fc99b6a9d07377d57198985d39fb63bc9b
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/35984
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

rcp: Change workspace directory to .tracecompass

This was supposed to be part of the big patchset of renames,
but got lost somehow.

Also updated the PLUGIN_ID while at it.

Change-Id: I05664f5697fd655b250884a4ed3e985eeed7a2ff
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36726
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: add equals and hashcodes to Declarations

Change-Id: I15d8030d2f75979012b87caae4dfc08bbda3ee7a
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36093
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf.ui: remove duplication in TimeCompressionBar

Change-Id: Ice353af65b57ef71b85082e55fea67a20c61f4af
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/37054
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

TMF: Add automatic execution of module dependencies

Change-Id: I461805133715db09f47293af73e9ef5952797298
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35991
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf.ctf: Rework the event aspects of CTF traces

- Move the "channel" aspect from LttngKernel to CtfTmfTrace
(it can apply to any CTF trace).
- Add a CPU aspect for CTF traces.

CTF events know the CPU they were taken on (previously saved in
event.getSource()). Now we can define an aspect to display it
in its own column in the event table.

Change-Id: I603cfd4cc4ca61964ec59ea17293900e7aa1f42a
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36365
Tested-by: Hudson CI

tmf: Remove the ITmfEventTableColumns extension point

Add a getEventAspects() method in ITmfTrace, which allows trace types
to define which aspects can be relevant for their own events.

This becomes redundant with the ITmfEventTableColumn extension point:
now that we build columns directly from aspects, the TmfEventsEditor
can simply call trace.getEventAspects() instead of reading extension
point data. The trace type is already defined in a core extension point,
so we'll pass the event aspect information through that now.

The combination of careful equals() methods in aspects and the use
of a Set in the TmfEventsEditor will allow "merging" aspects/columns
together, like was already the case.

Change-Id: Ic5b6126a50b1b1fdb4b2d5d23eb6a47a57555594
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36364
Tested-by: Hudson CI

ctf: make Declarations have some NonNull annotations

Change-Id: I2732bad6d7e390223c61fa08af98b3ae649baa79
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36573
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

LTTng: use tracer_name for perf-CTF traces

Domain will be set to kernel for both, perf and lttng traces. The
tracer_name will be set to perf if the trace is generated by perf and
otherwise lttng-modules if created by thet lttng tool.

Change-Id: I76e0c44efebe8be4e9f3a571884d63b932ac33d3
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/37212
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

TMF: Add an experiment utility class to get data from experiment's traces

Change-Id: Ifbc9e4a773bca227150f136bfcbe2a1410a91de5
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/36822
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: Correctly handle packet sizes > 2^31 bits

We use long's for packet sizes, so we should be able to handle
those, just need to make sure the value does not get truncated
to Integer.MAX_VALUE (or even worse, becomes negative!)

Change-Id: If3d6d39f765d8baf43297efaed1681473e1ffeaf
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/37129
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

xml: Remove duplication caused by d2e628

Remove a repeated switch case and create a helper to do this task.

Change-Id: I5ec17a09fd337107cd0e745e3e7d90bb8bba8300
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/36936
Tested-by: Hudson CI

tmf: Introduce the core concept of event aspects

An "event aspect" is a piece of information that can be extracted,
directly or indirectly, from a trace event. Simple examples could be
timestamp or event fields. But it could also be something indirect,
like a state system request at the timestamp of the given event.

This patch introduces the base ITmfEventAspect interface, some basic
implementations, and simplifies the TmfEventTableColumns to use aspects
instead, since part of it is now redundant.

Fixes bug #451411.

Change-Id: Ib045439d6084a6e6c7c276f1eddccfb13fe01be0
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36363
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

TMF: Call postDrawEvent also for transparent events in time graph control

The color of an event may be transparent, but it may contain data that needs to
be displayed through the postDrawEvent callback, so it should be called before
returning.

Change-Id: Ic672a1aeb781c89b9da2da9670e2b6f4968c3c43
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/36702
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Hudson CI

rcp: Update Trace Compass RCP branding

Uses the new icons and splash screen.

Happens to fix bug #446385, since we use a splash screen(!)

Change-Id: I3c5cb0f77b00f7c9513ad2161f0185df90a108eb
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36500
Tested-by: Hudson CI

tmf.ui clean up CustomXmlTraceDefinition a bit

Change-Id: Ia8c96e588307b892d08f8a5cc8e723f7a723cf53
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36871
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

lttng.control.ui: various cleanup fixes.

Fix empty nested blocks
Fix inner assignments
Fix utility constructor
Fix magic numbers
Try to slightly address complexity

Change-Id: I6f16be7d31d559b2b3f8189bdab5c4c44fded327
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/37013
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-by: Markus Schorn <markus.schorn@windriver.com>

Remove redundant modifiers in Interfaces

Change-Id: Iac7a27f1fa623dd16a8784fad25292356bc16010
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36527
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

TMF: Add some XML basic test cases and fix a bug

A bug where if the previous value was not existent, it was unboxed at -1 while
it should start at 0.

Change-Id: I58fda5925010b675bcab15cf49b69b303e6e9a9e
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/36344
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

TMF: Rework some test code for XML analysis to make it easier to add test cases

Change-Id: Ica13bc36c1c5312dd588ef87670f435ef01a76f2
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/36343
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

Bug 448058: Replace RSE by org.eclipse.remote

Change-Id: I9ee1b7e572839d66a5bacff8468e7da4dae235b4
Signed-off-by: Markus Schorn <markus.schorn@windriver.com>
Reviewed-on: https://git.eclipse.org/r/36632
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf: Remove source and reference from ITmfEvent

Those two notions were originally meant to have every trace type give
them a specific meaning, but to have some "common" concepts that
would be shown in the default event table.

Now that we have the possibility to aggregate columns in the event
table from multiple trace types, there is no need for those vague
concepts anymore. Every trace is free to implement and show any
specific item it wants. The base interface should be for the absolute
minimum that is required by the framework.

In fact, the notions of "source" and "reference" already meant different
things for different trace types (the Source of a CTF event was the source
CPU, while the source of a pcap event was the source address). So these
should *not* be aggregated in the same event columns since they do not
represent the same concepts.

Note: For the trace types that did make use of the source and reference
fields, this patch simply adds new fields to those traces, with the
names "source" and "reference" for now. But now that they will be decoupled
from the interface, we will be free to rename them accordingly.

Change-Id: I84f9729edee64233893779ca9c375a0a157bfc84
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36131
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

Make colors in SDPrintDialogUI static

Change-Id: I1bf8ebb95bca348884fa7dfe490886208a4afa0d
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36813
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Bug 449206: NullPointerException in HistogramDataModel

- make sure countEvent() isn't interrupted by clear()

- make sure events are not counted after histogram request is cancelled

- fix integer overflow calculating bucket move offset

Change-Id: I9e32032b3efed93a22c69d5cd1e2a99ab5550a11
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/36584
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

Remove tmf.ui magic numbers

Change-Id: Ia9499f01e3c184b878b19bdfdabb3a265b2a261b
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36525
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

ctf: make CTFStreamInput no longer autoclosable

This does not hold any resources. A rule of thumb is "if and only if it ends in reader, it is autoclosable"

Change-Id: I5c02a139eeda663f0ef721151cb6cbc4206a0448
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36371
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

CTF: Remove autoclosable from CTFTrace

Metadata is now autoclosable so the trace doesn't have to be.

Change-Id: I376c21f0888d57c55f41e83a15762c786a982b06
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36367
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tracecompass: remove duplicate test case

Change-Id: I03a546a5cc1bf436d5dcd5b8883c9ecd1631ab30
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36664
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

lttng: Fix Resource View for traces without sched_switch events

The Resource View assumes that if the base quark of a resource is
present, the "Status" sub-attribute will always be there. This may
not be the case, for example if the trace does not have sched_switch
events enabled, but has shed_wakeup.

Change-Id: Iaf25cf3d8a5b8ee686281dc3064f241229c2c245
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36492
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

ctf: Further updates to the CTF-testsuite tests

The stress/metadata/pass directory is now populated with scripts
to generate traces, update the test case accordingly.

Also skip the new regression/stream/pass/integer-large-size test,
we fail it at the moment. Keeps the build working until we can
look into it.

Change-Id: I5b6798de773354dce7abb93d43c649c0ef9fd258
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36596
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

Fix magic numbers in BTFTrace

Change-Id: Iea52a720801134faf1ebde2f69ac3c1f6ec71687
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36524
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

ctf: Update ctf-testsuite test for the new stream stress tests

Won't run anything yet, because the test traces need to be generated
with a Python script. The temerary ones may generate them manually,
then the test suite will run them.

Change-Id: I9db27da958d803eb8b4530ec6fc528d55af85631
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36546
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: Port the ctf-testsuite test to java.nio.Files/Path

Allows using some fancy DirectoryStream filter stuff.

Change-Id: I3aeb5259202e8f80da4fefcf6a5efc0c9b153d00
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36545
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: Use a ConcurrentHashMap in LexicalScope

The fChildren map in LexicalScope is normally written once, then
read over and over and over by many (possibly concurrent) threads.

For this use case, a ConcurrentHashMap is more appropriate than a
Collections.synchronizedMap(). A synchronized map will require readers
to obtain the lock, while a ConcurrentHashMap won't.

Change-Id: Ib0e826798427e0fc81c70e70aa4211b00a9e55e3
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36515
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

lttng/live: get last timestamp first

Change-Id: I0c3989ee4a421b0555adc873e44d3771f0a599cc
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36250
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

TMF: Synchronize an experiment instead of collection of traces

That will make it compatible with streaming since one full trace does not
have to be read before the others. It will also accelerate the time necessary
to get results, as matches will be found earlier. The overhead of reading the
experiment instead of the traces decreases performance though.

Change-Id: I7c94c21e63781b2cae0faead915be8959170a992
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35350
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

tmf: Bug 434646: Control Flow view not cleared when opening UST trace

Change-Id: I691ac18b2c6fb14b3eba5e7719822ecb1fe57698
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/36170
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: fix IllegalArgumentException when doing auto-detection

with trace type that doesn't exist (anymore). Since it is determined
during runtime which trace types exists this should not throw
an exception.

Change-Id: I0fa7a01351331b350eb2da85ab9624e96346cb2c
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36228
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf: add (internal) API for handling name clashes during import

Change-Id: Ice0d488b313f7c3e303e953b80f018f4c906ebf5
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36227
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf: Add missing NonNull import to tmf.ui.tests

Change-Id: Iec621cb9eb70773e71433ebb7498d6329d7842ef
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36374
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Consolidate constructors in TmfEvent

The default constructor can be used by the ITmfEvent extension
point, so should remain available, but should not be used by
normal code.

Change-Id: Id0e5e6d35d5593852b2a89672efea4fcbcd5b948
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36130
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

XML state system: Add "eventName" to the current stateattribute types.

We already have the "eventName" as a type for statevalue, but sometimes we need
it as a valid stateattribute type as well, e.g., in counting the number of each
event type.

Change-Id: Ie699221d71cee1686c6865bcffc3b38a991a0731
Signed-off-by: Naser Ezzati <n.ezzati@polymtl.ca>
Reviewed-on: https://git.eclipse.org/r/36059
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>

tmf: Add some nonNull annotation to the tmf.core.analysis package

And some more of those where required.

Change-Id: I369b0c5a0a2db8b3fccc6f6b97a2ba272b14fb28
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35752
Tested-by: Hudson CI

ctf: Use new location for the test traces

Simply change them from .../test-traces/ to .../test-traces/ctf/
We will eventually put other trace types there (pcap for instance),
so it would make it a bit cleaner to have each in their own directory.

Change-Id: Ie5ce2098e47a65872759d99a359c20219e6710b5
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36249
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Move TmfExperiment to its own package

With many experiment-specific concepts, and more on the way, it's
making more sense to put TmfExperiment in its own package.

Change-Id: I1ae03f693b075dd6e521fc20fb22d70b0c619551
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36208
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Consolidate constructors in TmfExperiment

Change-Id: I3844cde0e651052b626eeeb90e367a03b1743a4d
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36207
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

Disable "warn for deprecated in deprecated code" warnings in test plugins

Change-Id: Ie22c99b600e0c6c162e94cd6f7d62cc311bffb3c
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36248
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

lttng: Experimental support for perf-CTF traces

Implement initial support for perf kernel traces in CTF format.
They can re-use the event layout capabilities introduced previously,
so that we can add this support without implementing a completely new
trace type and accompanying analyses.

Also rename the "LTTng Kernel" trace type and analysis to "Linux Kernel",
as they are now tracer-agnostic.

The perf-in-CTF format is not set in stone yet, so the settings used
here will most probably have to be updated in the future.

Change-Id: I099f6e21caf40b9bfbe1a305c51947f1a264124b
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36200
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Hudson CI

lttng: Fix system calls for LTTng 2.6 kernel traces

The "syscall entry" and "syscall exit" events were renamed in LTTng 2.6.
This patch adds a new trace layout, which extends the base LTTng definitions,
but updates the system call event names.

"syscall exit" also had to be changed to a prefix, because LTTng 2.6 now lists
the syscall name in exits too. We don't make use of it though, so it shouldn't
affect other tracers who don't.

Also required updating the CFV/RV to only display the non-prefix part of the
syscall name, no matter what layout is used underneat.

Change-Id: I9061b0dff92daa11894d951c09ed3dffe80e4d1b
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36199
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Hudson CI

lttng: Add interface to abstract the event layout away

It would be possible to run the kernel analyses using events coming
from different tracers (or different versions of the same tracer).
This adds a IKernelAnalysisEventLayout, which specifies know concepts
in the Linux kernel, and the "support implementation" of each tracer
(or version) can extend it.

This will allow re-using the same modules and state providers, no matter
which actual tracer took the kernel trace. Also avoids having to define
separate trace types.

Also took the chance to rename some classes and packages to align them
better (i.e., the "CPU usage" and "kernel" analyses now use the same
naming scheme).

Change-Id: I0f073060071991e206011810bbb50dd826c688a8
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36185
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

lttng.kernel: Rename analysis packages and classes

There are two analyses in the kernel plugin right now, "kernel" and
"cpu usage". Simply rename the packages and classes to align them
better.

Change-Id: I9b7f86d2b9bad2d3d286d234eacea4ad51143bd4
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36290
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

TMF: Add possibility to set autoExpand level in time graph views

Change-Id: Idfeea90916fe1ebb7f60bb8a4893e730ef4a9544
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/36066
Tested-by: Hudson CI

ctf: Remove deprecated datatypes

Change-Id: I65f2e72c21ecd44993f969b1192e9f5e973be89f
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36092
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

Use archive.eclipse.org for the test traces

Change-Id: I871e2ec18b2d385bf9289d33d0b368919cc9da21
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36203
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

lttng: Do not rebuild the SS in every single state system test

At some point we changed the state system tests to rebuild the whole
thing in the @Before, because at that point AssumeTrue() inside a
@BeforeClass did not work very well. Unfortunately this made running
the tests much longer.

Instead, really build the analysis/state system once per test class,
but only keep the AssumeTrue() in the main @Before that is ran every
test. That check will test a static boolean set by the @BeforeClass.

tl;dr Make the state system tests ~20 times faster.

Change-Id: I96a137e0d2ec871f92efe586ba6456aca0a61129
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36184
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

Add some missing NonNull annotation "imports"

Now needed indirectly by some analysis modules in plugins that
didn't have it before.

Change-Id: I56bbf6e39290b83f9dce7a6852e7529fbd46c15b
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/36206
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Add element to its parent's children automatically and remove setChildren

Also, add a few useful setters.

Change-Id: I8ae7cff34d9a066d0ce9d17ef555e5539a3dc55c
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/35916
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36177

Change versions to 0.1.0

Change-Id: I58549b2ca6ef460cc3a0e073c62194a3053ac7fa
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36166
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf.ctf: Split the classes into proper packages

Change-Id: Ice64f1c902bc75715f02516ee7dd82b1175c8a05
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35616
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

tmf.ctf: Move each CtfIteratorManager into its own trace object

Which allows removing the "iterator manager manager" completely.

Some extra synchronization had to be added to CTFTraceReader and
CtfIteratorManager, because we've now lost the all-synchronized
manager-manager.

Change-Id: I0e11afa5865f52f6c2eba0ceb1773f23b7a9ac1d
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35615
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

tmf.ctf: Rename CtfIteratorManager classes

This patch simply renames existing constructs related to TMF-CTF iterators
to better represent what they do, to help future refactoring:

CtfIteratorManager -> CtfIteratorCEO .It is a "meta-manager" that manages
all the existing, now-called CtfIteratorManager's. Silly name, but don't
worry it's temporary ;)

CtfTraceManager -> CtfIteratorManager. This is the object that manages all
the iterators of one trace. It should eventually be moved to inside the trace.

Change-Id: Ie5483844856610907d3e332307eb0844209a93ef
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35614
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

Update some feature names to include 'Trace Compass'

Change-Id: I218c81b37a03f2e2da7e3977eaf24601133db10f
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36157
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

Add support for signing the jars when building the repo

Also update siteDestination to actual real path on the server.

Change-Id: Ib74f39ff0a7a48a75326396fc69fc3b960556a1d
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/36080
Tested-by: Hudson CI

XML state system: Add support of different comparison operators to the XML.

This will add 6 new comparison operators to the xml conditional statements.
These operators are: eq, ne, ge, gt, le, lt.

Change-Id: Ice9e87503eec6fea5a5d2281ffd8e42e59c8224a
Signed-off-by: Naser Ezzati <n.ezzati@polymtl.ca>
Reviewed-on: https://git.eclipse.org/r/35681
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>

tmf: Extract a method so that the manifest can be embedded in other xml files

Change-Id: Id8537482e006b9471f7b676989c91f08144f1aae
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/35819
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/35920

tmf: Add package-info to internal.tmf.core.analysis package

Change-Id: Iac1fdff332e4fc4bdcc9889ec4790a44adbcba61
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35745
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Have ITmfTrace#getHostId return type be NonNull

This will avoid a lot of null checks in analysis

Change-Id: Ic5730cab105ac9652dde4dc71a4524d946cbaf59
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35611
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

TMF: Set the name field of TmfComponent @NonNull

Change-Id: Id043e1efc4d070e043ec92d967fdb262271f59dc
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35904
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

TMF: Do not send objects in TmfCoreTracer#traceComponent and #traceRequest

These tracing methods were sometimes called in constructors and called methods
from the object that it is not necessarily safe to use when the object is not
finished building. This patch fixes this.

Change-Id: I64acee3adb850b8e17f0be3c8a0bbe5d54e97bf8
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35902
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

TMF: Add support to more than one match event definition per trace

Change-Id: I1da9129e616a5c831e19d9592c09b60396fb1ebf
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35491
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Hudson CI

TMF: Remove some deprecated class from synchronization package

Change-Id: I117758ebca01e5066e4e2fb4410301c8be40e78d
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35484
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

TMF: Improve the serialization of trace synchronization

Made some fields transient, so they are not serialized and initialize the
transient fields in the readObject method. Also added serialization unit tests.

Change-Id: I103b01b36efa66d27c37afba35935e415f4f0cc4
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35345
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

TMF: Have ITmfEvent#getTrace() return NonNull

Change-Id: If3eef82eeb4c79d6d9d8e09393fc6276f485b885
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35612
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

TMF: Update .options files to the tracecompass name

Change-Id: I744bd1bfbd73d7e6ccd715e6c9e3448231c15760
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35757
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

lttng: Fix a bug with lttng version when compiled from source

The output of the lttng version command is something like this:

lttng version heads/stable-2.5-0-g3910e60 - Fumisterie

The current code didn't recognize this as version.

Change-Id: Id480402ba6de0803e88d4561d1604ad464f80926
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35495
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

Tmf: Fix a bug in CFV and Xml time graph views where entries have no parent

If a child entry starts before the parent entry, like it happens at the
beginning of a trace, then in time graph views, they are not associated with
the parent.

Change-Id: I3fb72049248c13aecdc41e88420eaf4dfae24911
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/35342
Tested-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Fix legend image leak in Histogram view

Change-Id: Ic1262f2c191bec7c2e69a0b30c12d3b114f60d25
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/35607
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Hudson CI

lttng: add some stress test for opening and closing of traces

This will trigger concurrent starting and cancelling of analyses.
After fixing bug 447434 (commit e1c415) this test should run without
exceptions.

Change-Id: I9c7a35d29e0c4a406511ec1e8032b3d818498b94
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/35508
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

State system: Make ITmfStateValue Comparable

It is required to compare two state values based on different comparison
operators. It will be useful for the XML analysis.

Change-Id: I1fcb9525a41b11a648922271b5c2e06d6bd309ac
Signed-off-by: Naser Ezzati <n.ezzati@polymtl.ca>
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35414
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Hudson CI

Update top-level README file

The readme still had some references to linuxtools.

Change-Id: Ibf8371f6474f8c14d33b708a310fc98fa21be4bf
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35586
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

Update maven profile names

"tracingRcp" was used back in Linux Tools to differentiate between
the different RCPs in the tree. Now we can simply call the profile
"build rcp".

There doesn't seem to be any guidelines for profile names... but I
find names-with-dashes more readable than camelCaseNames.

Change-Id: Ie18af97adf162d1f2dd6af1845ba9f5869afbbea
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35585
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

tmf: Bump checkpoint index version due to incompatible timestamps

TmfTimestamp serialization was incompatible due to removal of precision.

Change-Id: Iace77186f730971e0296c582097fc179f332d985
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/35603
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

ss: Cleanup TransientState a bit

- Remove assertions (see bug 409346)
- Add more null annotations
- Update comments

Change-Id: I1d2edb3524cefa720365926b6d589141872779b2
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35566
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Hudson CI

ss: Also search in the ThreadedHistoryBackend's interval queue

If a view, like a live-enabled time graph view, makes a lot of queries
close to the "current end time" while the history is being built, it
is very possible to miss some intervals due to them being in the
ThreadedHistoryBackend's interval queue. We should also search in that
queue for possible intervals.

Since this case is still relatively rare, it's usually faster to query
the history tree first, then go look through the queue, rather than
checking in the queue first every time (which can only be explored
in O(n)).

Fixes bug #443127.

Change-Id: I82e389cf95ce04002f61629d5a0d317a35a931ff
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35504
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Allow reordering columns in event table

Change-Id: I7885463d90bcd223abe0ac915721490038c22923
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/35357
Tested-by: Hudson CI

tmf: Add unit tests for time offset

Change-Id: Ib44c6acb6f8bb0e3d23a583e30e6fec7071686b3
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/35558
Tested-by: Hudson CI

tmf: Update user guide for time offsetting

Change-Id: I4cd45175c8030bf18dd3013bf5fc91b21676e09c
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/35557
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

tmf: Change Offset Dialog advanced mode layout

Change-Id: I06d67eed0b6a8c8cc341a36079baa069507eb065
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/35556
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

tmf: Remove the precision concept from ITmfTimestamp

Timestamp precision is a problem in tracers, not in a viewer.
We already have the notion of scale which kind-of does the same
thing (if a timestamp is specified up to milliseconds, we can
consider it precise up to milliseconds).

Not only this removes big chunks of unused code, it should speed
up timestamp comparisons a little bit.

Change-Id: I55253721396cb05610503abd81861f522d3ee223
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/35426
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>