lttng2: Simplify boolean expression

Change-Id: I368354e936be2af10b6abcb6160fab8e45a9ee44
Signed-off-by: Sonia Farrah <sonia.farrah@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/72099
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

linux.ui: Add Threads Presentation to View Menu of the CFV

This patch adds a threads presentation item to the view menu
(the triangular menu in the top right corner of the view) of
the control flow view, in order to select which presentation
mode is needed.

At the moment, the options offered are:
 * Hierarchical
 * Flat

Others may come later such as clamping by time...

These menus are the first step towards grouping threads by
affinity.

Change-Id: I65a71ad4a5eb6489f668f8b5806f12702f43edbc
Signed-off-by: Mahdi Zolnouri <mzolnouri@yahoo.com>
Signed-off-by: Samuel Gagnon <samuel.gagnon92@gmail.com>
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/68231
Reviewed-by: Hudson CI

tmf: make CallstackProvider entry and exit return TmfStates

This allows certain callstacks to return simpler long or int
states instead of a string. Expect SIGNIFICANT space savings.

In an example application, ls was traced (coreutils 8
with -finstrument-functions)

For a 149 mb trace with no lost events.

State system size
before:
233 mb
after:
25mb

Total time for trace compass fullscreen (1920x1080)

read the trace and build the state systems:
before:
61.5 sec
after:
44.0 sec

average seek time (100ms window, 10 random locations)
before:
1.2 sec
after:
0.6 s

Change-Id: Ic868829296d92d854cda7b78784b6ee3be4d5a51
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70719
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-by: Hudson CI

tmf: Bug492895: Avoid unnecessary refreshes during import operation

This is achieved by using the TmfWorkspaceModifyOperation which
defers the refresh to the end of the operation. Also, setting a trace
type won't refresh the project explorer anymore. The refresh will
be handled at the end of the operation.

Change-Id: I302d54d18b8640b833fc964d5fe74eca06057c15
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71908
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: verify that resource is not null in import operation

Change-Id: Id73e76b550b5ee1767007e618fecbaa9154842de
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71907
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Update progress monitor in TraceValidateAndImportOperation

Change-Id: Id62964aa5f1ada0ebfe2156a0b52070103721fce
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71906
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Add a method to indicate if a ODA is user-defined

This will be useful in diffentiating what analyses are considered
"built-in" and which ones were added manually by the user. For
example, some operations might be prevented by the manager (like
deleting non-user-defined analyses). Or a different icon could be
used for them.

Change-Id: I887291f0b754c26cb8aca1e758b1741cb7152f52
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Signed-off-by: Philippe Proulx <pproulx@efficios.com>
Reviewed-on: https://git.eclipse.org/r/71930
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Make IOnDemandAnalysis#execute() throw an exception

If execution of an analysis yields an error, it should be reported
to the user. This can be done using a custom exception.

Avoid @Nullable parameters while at it, and document what can be
used instead.

Change-Id: I499ec1938339f6fd6c00e5fb0b5ecf25f082f7ff
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/71563
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-by: Hudson CI

tmf: Introduce IOnDemandAnalysisReport and corresponding element

For now the ReportsElement will be responsible for tracking which
reports are currently available. This means they are not persisted
and will be lost on restart. A future improvement could be to move
this logic to the trace so they can be saved as supplementary files.

Change-Id: I6c896665c7e4f43155aa499e59dc93d296e3119e
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/71562
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

common: Add a DecimalUnitFormat formatter

This format implementation will shorten number by using
SI prefixes (k, M, G,...).

For the special cases of NaN and +- inf, no unit is appended.

Change-Id: I941fcd410f050666c93c5c9c0ecaf4d9f92e2ec3
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70718
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Return correct type in SourceLookupPropertySource

The "Line" value of the Source Lookup event property displays
correctly, but throws an AssertionFailedException upon clicking
on it.

This is due to the property source returning a Long object
instead of a String. It was using Long.valueOf(long) instead
of String.valueOf(long). Changing the return type highlights
it clearly.

Change-Id: Ie12813c67ef754b2bcf1bb38804b08565847dea8
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70316
Reviewed-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

lttng.ust: Split the function name into its own aspect/column

Introduce a new aspect for the Function Location. For now it will
be used to store the function name, but could eventually also
print the offset within the function once we can retrieve it.

Implement a new Callsite object for this analysis which will not
print the function name, since we will have it separately.

Change-Id: Ie7d1598a2bfebe690c3e82183e8ad16f62489b4e
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70308
Reviewed-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

releng: Target updates for Neon M7

Change-Id: I462f916bbd5560d1a418fa01609300670d31c008
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/72050
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

analysis.lami: Introduce LAMI integration plugins

The LAMI (LTTng-Analyses Machine Interface) specification [1]
defines a protocol for external scripts to receive commands and
return results. As its name implies, this is what LTTng-Analyses
will use.

This integration however should be generic to the protocol and
not specific to LTTng.

[1] https://github.com/lttng/lami-spec

Change-Id: I0c5399bf70a288ebdcf5ab96ce31f5765b8aecb4
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/71564
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Fix the actual end time of state system modules

State providers keep track of the latest event that was handled is returns
this time as the current end time.

State system analysis module use this time to determine if an analysis is ready
to be queried. Also on the way to fix bug 488757.

Change-Id: I58308da05c3105f0a528398622658c3fb3f2250f
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/69760
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf: Add the dependency level to the analyses' event requests

Change-Id: Ib665794aaba3ff56599a2db6041c01f0ee0149f7
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/71541
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

analysis.os.linux.core: Change the == operator with the equals() method

Change-Id: I1107ea2aa1fb01ac5191e347364cdab94653231d
Signed-off-by: Bruno Roy <bruno.roy@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71951
Reviewed-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.core: Add a dependency level to analyses

This dependency level can then be used by event requests

Change-Id: Ia359fe367b03e99d84b6cf9e213dba6f019981df
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/71540
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf.core: Remove overriding methods

Change-Id: Id80b36e999af51f31905c7ef1e812ee8ebf90610
Signed-off-by: Bruno Roy <bruno.roy@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71946
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Introduce dependency level for event requests

Requests with the same dependency level that are compatible otherwise
can be serviced by the event provider together. In TMF, the
TmfEventProvider will coalesce these requests together to a coalesced
event request so the trace is read once and the events are dispatched
to each sub-request.

If these request are not on the same dependency level, then each
request will be serviced separately by the event provider.

This can be useful, when a request has to wait for a condition to
happen in another event request. It can block its request till that
condition is met. If both request were coalesced the coalesced request
will deadlock by one request.

Side effects of this patch:
- Traces (TmfTrace) will be read concurrently when using different
dependency levels. The event request scheduler of the TmfEventProvider
will take care of scheduling each request.
- In an experiment, requests may not be serviced by the same
TmfEventProvider, some may be serviced by the TmfExperiment, some
by the children traces (TmfTrace).

Change-Id: I2b2e143bc8dc9cd8d1707b033715cbb05a529401
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70864
Reviewed-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf: bug 492744 Reverse CAUSE and EFFECT in event matching

The code worked OK, but what was called cause was actually the effect and
reversely.

Change-Id: I5cf80cc31fd355e12e2c4024709a6ad32d6033d4
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/71736
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

Add project model elements for on-demand analyses

Change-Id: I618f1591c06a87ef92ff361290129975fc26fe55
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70922
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Introduce base class and extension point for on-demand analyses

The on-demand analyses are analyses that are run manually when
the user requests it. As such, they are much more simple and
much less featureful than regular analyses.

The upcoming integration with the LTTng-Analyses scripts will
make use of this.

Change-Id: I012bcc3fe3ddbd586a8d23887a481d4bf927aadc
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70921
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

ss: make node cache static

The state system is being used in many places now. A kernel trace
can have over 10 state systems running simultaneously. Therefore,
the caches have been taking more and more space. A summary reading
of heap dumps showed that for a kernel trace now, 512 mb is needed
for these caches. This will not scale.

This patch makes ONE cache for all of trace compass. There is a tiny
heuristic, each location is offsetted by the object's hashcode. This
is to avoid too much cache thrashing du to temporal co-locality.

The seek times in trace compass (going from one place to another)
are accelerated by a factor of 3 and are noticeable on an i7 with
16gb of ram and an ssd.

Moreover, this patch improves scalability, by making trace compass
not run out of memory after opening 5 kernel traces simultaneously.

Change-Id: I9c6509b0e32c04025fc66c0fdda3d31e971584b4
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70600
Reviewed-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@efficios.com>

lttng.ust: Tweaks to BinaryCallsite

- Rename the aspect (and Event table column) to Binary Location.
- Remove the concept of function name, it is not part of the
  information available from the trace. Was not being used
  anyway.
- Print the offset differently depending on if the address is
  absolute or not:
  /usr/lib/mylib.so+0x123 for Position-Indpendant-Code (PIC) binaries
  /usr/myprog@0x401234 for non-PIC binaries

The differentiation PIC/PIE is not very robust at the moment
(we just check if the file name ends in ".so"), but it will be
improved due to a new field being added by UST soon.

Change-Id: Ib0014a77e14c6a88ae19f0c4b410d3675f7966d4
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70307
Reviewed-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Introduce "External Analyses" and "Reports" project elements

These will contain the upcoming LTTng-Analyses wrappers, along
with the reports they will produce.

"External Analyses" will be refered to as "on-demand analyses" in
the code, as this is what this sub-tree will eventually become.
We can use this name for now to better differentiate with the
"standard" analyses, which are still shown in the Project View.

Change-Id: I9ef5296f22a121a7b8bc85c4681a43e7a5ee5054
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70717
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Fix unnecessary unknown state at beginning of call stack event list

It could be seen as a single pixel state at the beginning of the trace.

Change-Id: I48a601f37931ac3b7e2fcf056e988e9ae169d514
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/71821
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Fix ClassCastException in Call Stack view ThreadIdComparator

Change-Id: I2ad81fc94b2253a96289ca13cc9ed4ef6954a6c4
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/71820
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Fix thread id of Call Stack view thread entries

It was inadvertently changed in commit 1a0ff02.

Change-Id: I6763b8e0aa275a6fdbd3743c3a6fbd2ffb348306
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/71528
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: change printStackTrace() to Activator.logError()

Change-Id: Ia940e255c5eb903c392fb794422fbf463972323f
Signed-off-by: Sonia Farrah <sonia.farrah@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71827
Reviewed-by: Hudson CI
Reviewed-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Tested-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf.core: Changed printStackTrace() for logError()

Change-Id: I2a82e9488fc598dc4c2f2211141e5b42070d5007
Signed-off-by: Bruno Roy <bruno.roy@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71829
Reviewed-by: Hudson CI
Reviewed-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Tested-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf.xml: Move all .core and .ui packages to internal

and remove the @since tags from the files.

Change-Id: I087db7e4f275e042ce6fa36809aeb1b4a9a8a712
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/71022
Reviewed-by: Hudson CI

lttng.ust: Add the build-ID to the key of cache of addr2line calls

Before calling addr2line, we should also verify that the buildId
we got from the trace matches the file on disk we are about to
look at.

This buildId is already present in the UstDebugInfoBinaryFile
objects returned by the analysis, it is just a matter of
passing it down to the FileOffsetMapper call.

This way we can use it as part of the key for the cache of
calls, so that eventually calling it on different binaries
yields separate calls.

The actual verification is not done at the moment, as this
would require calling a separate process (like "eu-readelf").
There is not much gain in doing it yet anyway, because we only
look for one possible file (the one at the expected path). If
eventually the analysis is extended to look through several
different files for a given path - using separate debug symbol
files for examples - then the surrounding code won't have to
be modified.

Change-Id: I11f42a4295c88f66ad4a685198e30181daf0a1ba
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70306
Reviewed-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Put analyses under their own node in the Project View

Moves all the analysis/views element under a trace to a new
sub-tree called "Views".

First part of implementing Proposal #2 at
https://wiki.eclipse.org/Trace_Compass/Design_Documents/Project_View#Proposal_2

Change-Id: Icfe0f779b61db1c0a84845b87273d33b8d9cf9dc
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/69071
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

analysis: Fix storing of fields of system call latency analysis

Right now all fields stored as map fieldName -> field value using
the toString() method on the field value (field.getValue().toString)
However, this cause a problem when the type the value is an object
other than string or a primitive type (e.g. long[]). This patches
changes this to uses field.toString() which creates a formatted string
of the value.

Change-Id: I6ff61eb70d7a7bb958b77797c3271c61cc430eef
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71698
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

xml analysis: internalize strings.

Stings are currently taking a major amount of memory. This patch reduces
the memory usage by making the strings internal (cached) so duplicates
are no longer copied in memory.

Change-Id: Id686ea8c0648ddf2a7d041b6975ffbab6607167c
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70908
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Hudson CI

tmf.core: null type safety (added type annotation)

Change-Id: I85a2851e6f68ef4a1e9904de41ab6b03f7b932e6
Signed-off-by: Florian Moesch <florian@moesch.org>
Reviewed-on: https://git.eclipse.org/r/71753
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

Skip irrelevant modules when using Sonarqube 5.x

sonar.skippedModules is depracated and will not work in more recent
versions of Sonarqube. We can define sonar.skip in respective modules
instead. Both properties can be defined without problem. To preserve
the pom-less build, a new profile is activated in the presence of the
sonar.skip.properties file. So each module we do not want to analyse
with maven have to contain this file.

Change-Id: I6bf8115540744e48a9431f63ba96c504917f35e3
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71270
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

Adjust code to latest JDT changes

Disable @NNBD on methods extending ITmfTrace.initTrace due to
unconstrained Class<> parameter in base class. See bug 485988.

Change-Id: I6bee371dec7c53063a2a619bda8c22117fe756de
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/71471
Reviewed-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Fix TmfProjectModelElement.equals()

Elements of different types but with equal path should not be considered
equal. This avoids clashes between trace names and 'virtual' resource
name used by some elements, such as module id for an analysis element.

Change-Id: Id2238a9e8fe4ba9b8daede86b3a2b54d0de3b815
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/71628
Reviewed-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@efficios.com>

lttng.ust: Implement a symbol provider for LTTng-UST traces

This new symbol provider will make use of the "Debug Info"
analysis, which means it will only be available for traces
taken with LTTng-UST >= 2.8.

It will call into the existing BinaryCallsite and SourceCallsite
aspects to get the function name of a given address.

The provider's configuration page will allow specifying the
"root directory" of a target image. This is useful in cases
where a trace is taken on a remote target, and an image of that
target is available locally.

Since paths are found directly in the trace, this means that a
trace with the proper information taken on the same system
should show the function names in the Callstack View without
the need for the user to configure anything!

Possible future improvements:
- Split the symbol provider extension point in core and ui parts
- Merge the FileOffsetMapper logic with the core parts of the
  symbol provider.
- Move the "current path prefix" configuration from the trace
  object to the core symbol provider.

References bug 484042

Change-Id: Ia346c1eab2b54a54ddf672c7ffafd0655a47ee34
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/69973
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

lttng.ust: Introduce a cache of the calls to addr2line

Since calling an external process is relatively costly, we can
cache the results of the calls so future lookups can reuse the
information.

Change-Id: I5d84be78d45b893ea5942bdd5d2459faf4574af5
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/69972
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

lttng : Add SWTBot test for TimeGraphFindDialog

The find dialog is tested on both Resources View and Control Flow View

Change-Id: Ic26c9c986cdc9422d9a69015472c3edad4cacda4
Signed-off-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70602
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Fix selectAndReveal() in timegraphViewer

This fix handles the selectAndReveal() as if it came from an internal
source (triggered by the user) since the method is called from the
FindDialog. The selection should not be ignored when a pending selection
change is awaiting to be notified.

Change-Id: I1be75843b3fcb463eefa2c7587226205424b9c62
Signed-off-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71484
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

doc: Update LTTng-UST Callstack documentation

* Document that the 'vpid' context should now be enabled (it is not
  mandatory, but it's ideal to have it).
* Mention both the "Import Binary" and "Import Mapping File" methods
* Mention that the import step is only necessary for traces taken
  with UST < 2.8.

Change-Id: I2fd76ba41ffd166a76c20ce7929a7b7769974b29
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70081
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Make Callstack View pass timestamps and PIDs to symbol provider

Now that the Callstack analysis and view track the PIDs of each
function entry/exit, it can pass them along with the timesteamps
to the symbol providers, so that provider implementations making
use of the new methods can resolve symbols in the view too.

Also add proper comparison logic for ProcessEntries, they can use
numerical comparaison of the PID they now carry.

Change-Id: I7404ab4b893878dbdd8e526a646014739cfbe228
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/69970
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Introduce the notion of PID to the Callstack analysis

The current Callstack View/Analysis does not handle PIDs (process IDs)
only TIDs (thread IDs).

This works fine with the basic use case of a trace with only one,
single-threaded process. But since callstacks are per thread, but
symbol resolution is per-process, the callstack analysis should be
augmented to know about both concepts.

Change-Id: I6a6ad5f92e18c3c161390237b8dea513dc9ce8cf
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/69870
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

lttng: cleanup callstack requirement test

Reduce code duplication using static enum data. The file is 200 lines
shorter and new test are easier to add.

Change-Id: I8a67470d7fab53ca1a46b052bee3aa1354a906fe
Signed-off-by: Francis Giraldeau <francis.giraldeau@gmail.com>
Reviewed-on: https://git.eclipse.org/r/71537
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

analysis: change the SHT file name for the kernel analysis

The name of the file was not explicit. It is now the ID of the analysis.

Change-Id: I0fac8475c4d459389ff1b9c718b82d4381655dce
Signed-off-by: Cédric Biancheri <cdc.biancheri@gmail.com>
Reviewed-on: https://git.eclipse.org/r/71375
Reviewed-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

lttng: Update SWTBot tests to use MI 3.0 (LTTng 2.8)

In addition to being more up to date, it will solve an issue encountered
when the tests run in a workspace with a long path.

Change-Id: Ie0e02e09454fb9b9ee95b4c35d9f0da041182205
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71485
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

lttng: Add ControlService tests for LTTng MI 3.0

The MI 3.0 schema was added but no tests were using it yet.
This patch adds tests to make use of this new protocol.

Change-Id: If31665a4bc94fd0392accda1a6eaf2382c944d47
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/71380
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

lttng: Add schema for LTTng MI 3.0 (LTTng 2.8)

This makes it possible to use the LTTng Control view with schema
validation enabled and LTTng 2.8. The MI service is modified to be able
to handle a different schema depending on the LTTng version.

Change-Id: I85ec15edc3f0ce942d8df8637eadba3968fef1ca
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70188
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf: Add a method to verify if analysis is ready at timestamp

First step to resolve bug 488757. This new method will return whether an
analysis is ready to be queried at a given timestamp.

Change-Id: I2a1f597d503fe74664df95d0bb5c4c16952b6376
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/69753
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf.core.tests: Add method to simplify XML Trace stub setup

Change-Id: I7c59f26d0bf89b85aa648a452641d3b3edae59b2
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/70034
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf.core: Internalize timestamp types

Move timestamp implementations to an internal package.

Change-Id: I2705c11a4a2cf91204854e726c798c37b30bec6a
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70246
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf.core: simplify timestamp implementations

This patch makes BIG_BANG and BIG_CRUNCH and ZERO special singleton
objects. They now normalize to themselves. Big bang and big crunch are
always the largest or smallest values.

It also internalizes TmfRealTimestamp inside TmfTimestamp as
TmfTimestamp$Impl.

The timestamp comparison is very streamlined. The performance
should be similar though.

One bug found that will never occur in the real world:
additions with -Long.MIN_VALUE gave errors as -Long.MIN_VALUE ==
LONG.MIN_VALUE.

Change-Id: I284ed1b8f0dd1f6d5299e172dd07128d53ac5945
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70245
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Bug 491714: Fix population of XmlTimeGraphView for experiments

With this patch the XmlTimeGraphView is populated with the content of
all traces part of an experiment.

When rebuilding the event list the view cache (event list, filters,
view context) is cleared. This is important because right now
there is only one instance of XmlTimeGraphView per trace even if there
are more than one XML analyses and views defined.

Change-Id: I535cace0089147bea4584c5af8b7a0c57157bd36
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70681
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

lttng: Fix not being able to get available context with LTTng 2.8

'lttng add-context -h' was replaced with 'lttng add-context --list' in LTTng 2.8

Bug: 491933
Change-Id: If277b86410a5d19cfa53531b03dfb5b7350c4c5b
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70894
Reviewed-by: Hudson CI

tmf: Add/update '-', '+' and '*' key handling in TimeGraphViewer

The handling of '-'and '+' key have been updated when the mouse is over
the namespace. Before they were used for time zooming regardless of the
mouse position. The handling of '*' key has been added when the mouse
is over the namespace.

When the mouse is over the namespace, pressing the
- key '-' will collapse recursively selected entry
- key '+' will expand selected entry
- key '*' will expand selected entry to the level with at least one
  collapsed entry

This patch is to align the keystroke handling to the TimeGraphCombo.

Change-Id: I964168bd1aa388430775055a39f625beaffb5985
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70583
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: fix conflicting find handler message in time graph views

Change-Id: Icf6ff1772acad9b86278e507b0cf987fba859857
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70896
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

lttng: Fix filters not shown with LTTng 2.8

In LTTng 2.8, a new MI field was added, filter_expression.
This needs to be handled by Trace Compass.

Bug: 491941
Change-Id: If1e796ca24d5542422f61f63c91cb4c37d702edd
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70905
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

lttng: Make schema validation optional

With this change, the schema validation is disabled by default except
for tests. The validation can be enabled easily as a Tracing option
in the launch configuration. To be able to enable this by default for
tests, it is also exposed in a system variable which could also be
used on the command-line.

Change-Id: I492f2877a7e066e14176a209159da17cea83c3f1
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70199
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Update setting of minimum size of dialog box

Change-Id: I6a5e722c5a435ea8d34eafe8312a429c710f0503
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70606
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-by: Hudson CI

lttng.ust: Retrieve the function name when calling addr2line

Add the -f flag to also query function/symbol names. This will
allow us to populate the "function name" field of the generated
callsite objects, which in turn will populate the Callstack View
appropriately.

Also pass the -C flag, which demangles C++ function names, without
apparantly affecting native names.

Change-Id: Ife6e8581347ab2f94558e258dc350a4c1b04c3fa
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/69971
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI

tmf: Update BasicSymbolPrefPage dialog description

"provider" is an internal implementation concept the user
won't know about.

Change-Id: I3be1167f32183f4ad78d4e2135fda50466d3a024
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70315
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI

tmf: Rework TmfTraceUtils#resolveIntEventAspectOfClassForEvent

The generic type bound now ensures the aspect class passsed to
resolveIntEventAspectOfClassForEvent() really resolves to an
Integer, which means we can skip the check.

Also rework both aspect-finding methods to use flat Stream operations.

Change-Id: I8a341021699a69ac21b7a147c9ea91d58619fa99
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/69856
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

Upgrade to Tycho 0.25.0

See https://wiki.eclipse.org/Tycho/Release_Notes/0.25

Change-Id: I309392787e2f1c8b041e4c83648b4a096e8032e4
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70296
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf.core: Introduce TmfTimestamp factory methods

Often times, a TmfTimestamp will be created using a scale of -9 or 0.
The result is a less efficient timestamp that takes more memory than
the nano or simple timestamp and is slower to compare. This patch makes
timestamps created always return the more efficient type.

At the moment, the factory will either create a new TmfNanoTimestamp if the
scale is NANOSECOND_SCALE, TmfSecondTimestamp if the scale is SECOND_SCALE,
or a plain old TmfTimestamp (TmfRealTimestamp).

This patch adds the following helper methods:
* create(value, scale) - create a timestamp with an arbitrary scale and value
* fromNanos(value) - create a nanosecond based timestamp (more efficient)
* fromMicros(value) - create a microsecond based timestamp
* fromMillis(value) - create a millisecond based timestamp
* fromSeconds(value) - create a second based timestamp (more efficient)

Also, TmfTimestamp is converted to an abstract class with "getScale()" and
"getValue()" needing to be implemented. This allows a timestamp like
TmfNanoTimestamp or TmfSecondTimestamp to not store an extra integer and
save 33% of their sizes.

Change-Id: I0996c9b4f626a70436b58e82446ed69d2e50af98
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/65312
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-by: Hudson CI

analysis: Add count column to latency statistics view

Change-Id: Id2db373022a6c1b78ddac2a6c7b580459c189f3a
Signed-off-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70895
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI

tmf.ui: callstack - replaced multiple single SS queries by full query

CallStackView would do a single query for every thread,
at the start of the statesystem, or at the end of the
statesystem.
This patch replaces all single queries in buildEntryList
by two full queries and reads to the result of those queries

Change-Id: Id59b355fd84bfc0b8e30ef3f3f375f8ca0dbe634
Signed-off-by: Loïc Prieur-Drevon <loic.prieurdrevon@gmail.com>
Reviewed-on: https://git.eclipse.org/r/70854
Reviewed-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf : Add pattern analysis behavior

This patch defines the behavior of the pattern analysis state provider
and defines all the models needed.

Change-Id: I6561feb02b06627ef5059d777a25a8bde56a70be
Signed-off-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/65751
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI

analysis: Bug 489573: Internalize Control Flow view implementation

Change-Id: I92e81e1b158cb5dbf1910f9d25b880482fdb3355
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/69007
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

lttng: Add analysis requirements for LTTng UST Call Stack analysis

With this the user gets notified when the call stack analysis cannot be
executed. The help text will give some details about this.

Change-Id: Ib3628606a74ae5e83b60fde81720f7035d7eae86
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/68235
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf: add events field analysis requirement class and test

This class allows to specify required events fields for one given event
type.

This class is useful for CTF, where optional event contexts are provided
as event fields.

Moreover, this will be useful for custom txt and xml trace types, where
for each event the event type is the same but the event fields might
be different.

Change-Id: I445976ea7b93ee95be887504bc223912c2cab298
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/68233
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

tmf: Add generics to ITmfEventAspect

The generic type can allow us to define the type of
the resolve() method's return value.

Change-Id: Id9bb9ddaf143fa6b3ed09bf32eab98e1718e7ee3
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/69569
Reviewed-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

analysis.os: Add layouted requirements to the CPU analysis

That patch attempts to bring back the concept of requirements to linux kernel
analyses. The requirements were disabled since the event layouts appeared.

Change-Id: I12b4befa69c94ca7205ea0db7670ee126788dc42
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/68015
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI

analysis.linux.ui: Make disk io display read before write

This is more inline with the typical views of hard disk usage.

An added bonus is that reads are blue and writes are red. It is better
to avoid red/green for accessibility reasons, and typically red == write.

Change-Id: I7397fdd72539646ba2fbe79f8258580bb45908d0
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70779
Reviewed-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

lttng.ui: move UST memory legend to left

This maximizes readability.

Change-Id: I36c08a5f16477e1c60752a2efd7b2c4698dc5589
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70778
Reviewed-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf.ui: make CommonXLineChartViewer hide the X axis label by default.

The X axis label is always "Time" for the common line chart viewer, this
patch saves some vertical space by not displaying it.

The legend is moved by default to the bottom, this saves a lot of space
due to time alignment.

Change-Id: I965290de363d2196140e03e60f3abe69fe35efb0
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70776
Reviewed-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf: Fix NPE in XmlPresentationProvider for undefined states

A NullPointerException was caused when creating a state tooltip for
state values that are not defined in the XML definition.

This patch avoids the NPE in this case. The state tooltip won't be
shown when this happens.

Change-Id: I07ca75d7bb39d3479c225b4b144e1acc119eae3b
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/68072
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

os.linux.ui: Remove redundant query in ResourcesPresentationProvider

getEventHoverToolTipInfo would do a full state query then
a single query for each CPU/IRQ.
This patch replaces the single query by reading the full
state results.

Change-Id: I10959152e094674b8276608dddb2fe03368f1d96
Signed-off-by: Loïc Prieur-Drevon <loic.prieurdrevon@gmail.com>
Reviewed-on: https://git.eclipse.org/r/70765
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>

releng: Fix Mylyn update site in targets

Mylyn 3.19 was release and no new milestones has been produced so
we can use that release until (if) there is a new one for Neon.

Change-Id: I150eb050bcbb576ac2deda102451e02575c78e66
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70830
Reviewed-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-by: Hudson CI

tmf: Make Analysis Requirements implement Predicates

This allows us to use requirements as filters.

Change-Id: Ib103e7fec613455e140d5c63c24e7d2f53248d9c
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70049
Reviewed-by: Hudson CI

tmf.common: Fix boundaries in data size formatter

This patch fixes the display of units on boundaries,for example
"1024" will be displayed as "1 KB" instead of "1024 B".

Change-Id: I3464d1d54e05254f3133d76faf2b75890690b766
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70698
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Hudson CI

tmf: Add an event names requirement

It is preferable to use inheritance instead of String-typed values.

Change-Id: I62eaa0f7a01ac459ed06c471379f8f20bcd9b4ef
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/68014
Reviewed-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

enable new performance benchmarks.

Change-Id: Iae2f6ec13f60fd5a203c08c1eeed63866a0ac74f
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70584
Reviewed-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>

linux: tid analysis benchmark

This benchmark is nearly identical to the kernel usage benchmark, but it uses
the tid analysis. It should be interesting to compare and contrast the two.

Change-Id: I6db6ce5142c518e2d79959a4724db56ee1401348
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70578
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Hudson CI

analysis.os: Benchmark some typical usages of Kernel Analysis

This adds benchmark of getting the thread running on a CPU from a kernel
analysis and making full queries on that analysis.

It paves the way to adding some more of those benchmarks.

Change-Id: I853a350a7647973b87e49dabaf2ac14b09cd0136
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/67444
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

Revert "ss: Add serialization logic to state values"

This reverts commit ce148788289534e20b56d0790a8ad93ce2e6473d.

Change-Id: I81e3a1f3557ef8edceaf930ba9b8d8cb3b23f200
Signed-off-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70711
Reviewed-by: Hudson CI

Remove MaxPermSize from all launchers

The Permanent Generation space was completely removed from Java 8,
using the MaxPermSize parameter only prints a warning and has no
effect on the JVM.

Change-Id: I1d378c0c09a9d8a2861e0d39bfe3bbdf37df0bad
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Reviewed-on: https://git.eclipse.org/r/70700
Reviewed-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf : Move preconditions at the same level as states in the XML file

Change-Id: Ib4a6c59d841831d4b91298970cda039272149e04
Signed-off-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70292
Reviewed-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Tested-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

os.linux: Bump KernelStateProvider version number

Previous commits changed the content generated by the
KernelStateProvider for process' priorities and exec
names. Bumping the version number allows to take advantage
of those changes without manually removing the history
tree.

Change-Id: Iad483b0f3c51fb53242264948f52e98709ebb1bc
Signed-off-by: Raphaël Beamonte <raphael.beamonte@gmail.com>
Reviewed-on: https://git.eclipse.org/r/70688
Reviewed-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Tested-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Reviewed-by: Hudson CI

doc : Add user guide for time graph Find dialog

Change-Id: I27f17a27373270ccbe7af302dd85308d14a8f8f5
Signed-off-by: Jean-Christian Kouame <jean-christian.kouame@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/70409
Reviewed-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

SchedSwitchHandler: rename private methods

The setNewProcessPrio and setNewProcessExecName methods
were respectively renamed setProcessPrio and
setProcessExecName as the 'New' part of the name can be
confusing with the 'next process' when handling a
sched_switch event. The parameters of those methods have
been renamed accordingly.

Change-Id: Id1c0f00f165e2d906cb7650d511b1bebc20d4b1e
Signed-off-by: Raphaël Beamonte <raphael.beamonte@gmail.com>
Reviewed-on: https://git.eclipse.org/r/70514
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Tested-by: Alexandre Montplaisir <alexmonthy@efficios.com>

SchedSwitchHandler: update the exec name of the former process

Update the SchedSwitchHandler class to update the exec name of
the former process as well as the one of the new one. This
allows to keep a 'sort of' up to date exec name in the absence
of other events updating it.

Change-Id: Ie0a1ebc5b64214150edb0b41b62ccaab4447b041
Signed-off-by: Raphaël Beamonte <raphael.beamonte@gmail.com>
Reviewed-on: https://git.eclipse.org/r/70513
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Tested-by: Alexandre Montplaisir <alexmonthy@efficios.com>

SchedSwitchHandler: update the priority of the former process

Update the SchedSwitchHandler class to update the priority of
the former process as well as the one of the new one. This
allows to keep a 'sort of' up to date priority in the absence
of other events updating it.

Change-Id: Ie0da99274b9c932a0784dbd737d3279f2cfa6165
Signed-off-by: Raphaël Beamonte <raphael.beamonte@gmail.com>
Reviewed-on: https://git.eclipse.org/r/70504
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@efficios.com>
Tested-by: Alexandre Montplaisir <alexmonthy@efficios.com>

analysis: Bug 489573: Internalize Resources view implementation

Change-Id: I35fc6e359b2727b19ad77e5788bb9e7c0f1dfb51
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/69008
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

analysis: Bug 489573: Internalize CPU Usage view implementation

Change-Id: I45e930b51b718f0fd469096f968b8484965b0971
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/69009
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Hudson CI

tmf: Bug 491548: Do not incrementally build full time graph event list

The full event list should be recomputed for the full range at every
iteration of the build thread. If built incrementally it can create too
many events in the full event list.

The build thread should not insert NullTimeEvent in the full event list.

Change-Id: I2523291197b2d93267b2aa799e6126ea74eea0f9
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/70519
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Update AbstractTimeGraphView and TimeGraphEntry API

AbstractTimeGraphView.buildEventList() is renamed to buildEntryList()
which is more representative of its implementation. The Javadoc of this
method and getEventList() is improved to better explain their use.

TimeGraphEntry methods setParent(), getParent(), getChildren() and
addChild(...) are changed to restrict the entries to TimeGraphEntry.
This avoids instanceof checks in AbstractTimeGraphView implementations,
and avoids unexpected behavior in case a child or parent entry is not of
type TimeGraphEntry.

Change-Id: Ibf1c3f3fad467ab7ff215fd2552bb3350f00ef7e
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/70185
Reviewed-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>