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

tmf: Fix incorrect histogram state when concurrently cleared during drag

- Reset the offset when drag zoom interrupted by clear

- Update selection when drag selection interrupted by clear

- Fix overflow in min and max offset calculation

Change-Id: I6eea70d713d4b46e75d254b9374631623c023600
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19605
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Hudson CI

lttng: Fix maven tests on Mac

Cocoa requires the Display to be created on the main thread
so -XstartOnFirstThread needs to be specified in the VM args.
Also use -Dorg.eclipse.swt.internal.carbon.smallFonts to make it
consistent with the fonts in the SDK.

Change-Id: If695baab41a35165f2cdc1120f088b5d6ba61354
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19888
Tested-by: Hudson CI

ctf: Add unittests for the CTF Lexer and Parser.

This is a first draft of unittests for the CTF Lexer and Parser.
The coverage of these tests is still incomplete. More unittests
are to come.

Some tests are commented and need to fix the grammars.
The plan is to fix them in future commits to limits the
amount of refactoring.

Change-Id: I2d8e06560314ef22710985915b726a74974bbb1c
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19485
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: simplify trace environment lookup

Environment is an immutable data structure.
Wrap it into an unmodifiable collection.

Change-Id: I29ab6d4671eae2c9efd59412f80186a10b9b4952
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19622
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

lttng: Update Callstack View documentation

Add information about the Import Mapping File action, and document
the process for LTTng-UST traces.

Change-Id: I2db3c7e8021ed99bd7af9cbc7ed6a38fe06233b0
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/19845
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
IP-Clean: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Enable the finish button when default target project is selected

The previous fix for the batch import "target project" page introduces
a default selection but the user still has to click on the list again
to enable the finish button.

Change-Id: I492cc1ff115d6208aa577a721f04d96ebd2a4e2c
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19881
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Select a target project by default in batch import

Select the first list item by default in the last page of
the batch import (project selection). On Mac, no item is
selected by default. Since an item is now selected by default,
I changed the behavior of the page's message so that is only
displays an error when nothing is selected. Also, the Eclipse
UI guideline specifies that a page at initialization shouldn't
display an error, see
http://www.eclipse.org/articles/Article-UI-Guidelines/Contents.html#Wizards

Change-Id: I770fb9151f3c74a35dbf2d4e94db08f1301817cb
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19822
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: modify data-structure in CTF-Trace holding streams.

CTF allows multiple events to have the same ID as long as
they are not part of the same stream.

Thus, a unique event is represented by <stream-id, event-id>
and not only by event-id.

In this case, the CTF-Trace contains multiple streams
but maps all event-id into the same map (without stream-id).

Back-references are always a bad design.
In a future patch-set, the plan is to remove
CTF-Trace from CTF-Stream.

Change-Id: If5fc50d8541b152938a96447087d88c78e35e959
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19619
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

lttng: Handle threads in statedump events

In the statedump of a kernel thread (tid), we used to only check the
'ppid' field to get the paren thread. This works only for *processes*,
not for threads.

For threads we should use the 'pid' field to indicate which is the
parent TID.

Thanks to David Goulet for pointing this out!

Change-Id: I6b309d671ce43958ca9c0b43c5912877dca14e38
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/19728
Tested-by: Hudson CI
Reviewed-by: Florian Wininger <florian.wininger@polymtl.ca>

lttng: Add SWTChart to target defintions

Change-Id: Idd7728afa8293d236e74729b1d12ff5b67d5f0ee
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19732
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>

ctf: Generate synthetic trace in test traces folder instead of temp dir

I was having problems with the synthetic trace now that the Luna branch
is merged into master because the index files got generated into the
synthetic trace folder and the folder would not get deleted on exit
because of that. Then the next test run, the trace would not get
generated because the folder already exists. I think it makes sense to
move the trace to the test traces folder and only generate it once.

Change-Id: I03e815c50e22adf2d5355856b8031cc7a8b4c9b1
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19680
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

tmf: Avoid HistoryTree to expose its internal HT_IO node.

The HT_IO node is an abstraction of the file accesses for
a tree node. It should not be exposed to others classes.
The history tree must act as a proxy for requests to HT_IO.
This will ease modifications and maintenance of HT_IO.

HT_IO node should not keep a back reference to the HistoryTree.
This is a bad design. HT_IO node needs to know about HTConfig
to perform its tasks.

The HT_IO class contains method to read from disk.
The concept of a LastBranch is related to the HistoryTree.

Thus, readFromMemory is the same than calling

HistoryTree.readNode(...)

and readFromDisk is the same than calling

HT_IO.readNode(...)

Each layer is reponsible of its own caching mechanism.

Change-Id: Ia835a03efa321cdefd170de002833e044d144ee0
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19557
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

lttng: Update @since annotations for 3.0

Change-Id: I53ec925b49cb24485f0fbaf2c6441b2914f1e452
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/19458
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
IP-Clean: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Hudson CI

lttng: Re-enable API tracking

This reverts commit 26595e5ced3af74051513822158484e57315aa16
(and adds API errors to the new tmf.analysis.xml.* plugins).

Change-Id: Id3d93a594bdbadb8973698df26a5683b1d02c71e
Reviewed-on: https://git.eclipse.org/r/19457
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
IP-Clean: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

ctf: Remove a backward dependency into StreamInput to CTF-Trace.

This dependency is useless.
In fact, the creation of the indexes seems also to be useless.
I didn't find any use of it.

I think the notion of StreamPacketInputIndex should be an class
only use by StreamInput.

Change-Id: I7dc236a64ad9c8a33e0d0f2b701d8794ef3f9226
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19650
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

ctf: tiny code refactoring in the metadata.

This pathch-set is a tiny cleanup before more refactoring.

The metadata are data and the "parsing/decoding/../" and
any actions on those data should be isolated in "internals"
and only accessors exposed to the user should be visible.

Patches to come soon.

Change-Id: Ie197726e16e0373e9a312b22ca9116968463990f
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19585
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

tmf: Remove back-reference from HT-Node to HT-Tree

The HT-Node needs to know about the layout of the node (size, n-child)
but not about the tree in which it belongs to. Any back reference
is typically a bad design. It complicated unittests and maintenance.

  HT-Tree --->  HTNode--\
   \                     \
    \---------------------> HTConfig

This patch-set is one of multiples to refactor the HT-Tree API.
The API refactoring will lead us to a easier algorithmic refactoring.

Change-Id: Ifa11cec6a89b9b1c698e0a2b2b18ac5ec70a7db0
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19555
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

ctf: isolate the abstraction of memory map byte-buffer.

The concept of the memory map mechanism should not be exposed
outside the StreamInput. This is internal to StreamInput.
The API should exposed : give me a byte buffer (I don't care how).

Change-Id: I832d2776845a07173f4ef7cb47fd9c75c7db73bf
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19652
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: avoid array creation to iterate through scope types.

The array creation is useless in this case.
It come from the exposed API of the scope which expose an "Array"
instead of an "Set".

The collection is not important to iterate through.

Change-Id: Ica313c1cbb4d58529389fe15dfd71f153a11b00b
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19591
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Code cleanup in the history tree

The method preOrderPrint wasn't thread safe because it was using
a "global" field. This is not a big issue, but it's cleaner this
way.

I did a cleanup pass on this file.

Change-Id: I6255be52d3b2a8f84ef95a093578e10489c234eb
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19560
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

ctf: Some little refactoring in the ctf lexer.

Mostly only moving rules aroung to ease the reading.

The rule to match c++-like comment do not need to
take care of '\r'. It's a caractere like the others,
and the comment will be tuncante to the '\n'.

Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Change-Id: I7fdb6dbd05729c6502ee987dec47b3cfd65806bc
Reviewed-on: https://git.eclipse.org/r/19411
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: better enum and variant verification

Check if enums are valid integers
Check if variants are not using unspecified tags
Check for variant tags recursively
Check for out of range variants
Check for non-intersection between variant and tag

Change-Id: Idd6a98673d8bb3b0a7a1649893a2e21f230a1bcc
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19410
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

tmf: Bring Java 7-specific errors back down to warnings

If we were to switch to Java 7 today, thousands of *errors* would
show up because of the 1.7-specific project settings set to generate
errors. Tone them down to warnings, until they are actually fixed.

Change-Id: I65720d2e3ea3849d5450e2677e25710ff2833909
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/18932

lttng: Add the 4.3 JDT annotations to the 4.2 and 3.8 targets

Since we know that this is not a runtime dependency, we can safely
add it to the target definitions, so the code actually compiles when
we're using these targets.

Change-Id: I8fd8540ac4e9cd3cc4ae94290bd4b2a3ecbf0a22
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/18990
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Fix concurrency issues in histogram view

- Make sure signals sent to itself with the signal throttler are
executed in the UI thread, so that Histogram.clear() doesn't
concurrently set fScaledData to null

- Clear the drag state when Histogram.clear() is called so that
subsequent mouse operations don't need to protect from clear during drag

- Prevent the trace updated signal from constantly resetting the
histogram range so that trace indexing doesn't interfere with drag move
and drag zoom

- Fix time range in full trace histogram being offset incorrectly when
new trace events cause the scale to change during dragging

- Fix mouse wheel zoom out snapped to current zoom start time in time
range histogram when trace updated

Change-Id: I1c6e71d9c7f26dec93192768e99306fdbb978dd1
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19341

tmf: Fix mouse actions disabled when no visible events in histogram

Change conditions from zero events to empty time window to detect unused
histogram and disable mouse actions.

Also removes some unnecessary checks.

Change-Id: I761473005d6b004101741c535e8ec98a2bc0870f
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19324
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

ctf: simplify getMinValue/getMaxValue computation.

The BigNumber divide is useless in this case.

Change-Id: Ibec0c081430b93ff4873db3bf9de3a2eece76f29
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19416
Tested-by: Hudson CI

ctf: more restrictive parameters accepted as sequence lengths

The parser now only accepts unsigned integer variables as valid sequence lengths.
This patch also changes the API of DeclarationScope to be more clearly named.

Change-Id: I70cc6022ad99d1340f3c5a30546e489d2428832e
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19403
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

tmf: Add a TODO the the HTNode cache size

At Florian's request.

Change-Id: Ifbe9fc2e2ce878789716ecf4854c2dd6f1a6be66
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

tmf: Add a HTNode cache to decrease the number of disk access

On startup, a cache of 256 entries give this ratio:

7778/8000 [97.23%]

After running a while:

42375/42600 [99.47%]

Change-Id: Icf2f6af3c9f40d4532196b18a87f2aa728341a58
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Signed-off-by: Florian Wininger <florian.wininger@polymtl.ca>
Reviewed-on: https://git.eclipse.org/r/19408
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

tmf: Implement a cache of TmfStateValue

This patch provides a caching implementation in the TmfStateValue factory.
TmfStateValue are immutable and can be shared.

By changing the size of the cache, we observed this cache ratio.

size         hit      total     ratio
  2 count:  7369956/15551000 [0.4739216770625683]
  4 count: 11742495/15530000 [0.7561168705730843]
  8 count: 14622383/15545000 [0.9406486330009649]
16 count: 15007948/15552000 [0.965017232510288]
32 count: 15372868/15543000 [0.9890541079585665]
64 count: 15432506/15523000 [0.9941703279005347]
128 count: 15488720/15534000 [0.9970851036436205]

Which seems to be an excellent ratio for a really small cache.
The same idea is used with Integer.valueOf(...).

Change-Id: I3701b900035d83b19531101ab60e7292899d8cef
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19030
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

TMF: Add new XML analysis project

Empty project where the data-driven XML analysis will be added

Change-Id: I3fc3c1ab468c601127451990bbb87dc26881f547
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/18583
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>

ctf: Re-indent the CTF grammar to ease reading.

This patch-set does not change any rules.
I tried to apply the same Java indent rules.
A few rules are still badly indented.

I don't get why you keep all enter/exit debugging in the grammar.
You have antlr option to do this:

-trace have all rules call traceIn/traceOut.
-traceParser have parser rules call traceIn/traceOut.
-traceLexer have lexer rules call traceIn/traceOut.
-traceTreeParser have tree walker rules call traceIn/traceOut.

Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Change-Id: I5077cdc23fe8c76fb17fd6ce0569bdd66d2e9ad7
Reviewed-on: https://git.eclipse.org/r/19334
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Hudson CI

TMF: Bug 422846, trace resource is refreshed after analysis

Change-Id: I2c9301eb5f6eb18139e0af475ad639429b6126bd
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/19116
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
IP-Clean: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Hudson CI

ctf: fix a misleading comment on string alignment.

Signed-off-by: etienne.bergeron@gmail.com <etienne.bergeron@gmail.com>
Change-Id: I4efeaf273c8695a0aa8f72ba57592dd8d0e8b649
Reviewed-on: https://git.eclipse.org/r/19320
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

Merge branch 'master' into lttng-luna

tmf: Fix NullPointerExeptions in Histogram view

These NPEs happens when clicking or dragging the mouse in the
histogram windows before any data is available in the scaled data
instance.

Change-Id: I57efdb03cd53e2198580fa6c0ed047fc2dc4c3ac
Signed-off-by: Bernd Hufmann <Bernd.Hufmann@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19268
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
IP-Clean: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Hudson CI

Bump build system to build 3.0.0.

Subprojects parent poms are losing their versions as they are always the
same as the parent pom version and thus only introduce noise.

Change-Id: I6388f33c5e68d8c061cf57c7c09612817e189e30
Signed-off-by: Alexander Kurtakov <akurtako@redhat.com>
Reviewed-on: https://git.eclipse.org/r/19310
Tested-by: Hudson CI

ctf: Add support for arrays or any type.

A null pointer exception occured for events that contained an array
of something else than "int". For example, an array of struct:

fields := struct {
  ...
  struct {
    string arg_name;
    string arg_value;
  } arguments[4];
};

With this fix, all arrays of CTF types are supported.
Sample output in the "Properties" view: http://i.imgur.com/31tqbgG.png

Change-Id: I3f2be314d1fd2bda8c95b208d018227477572fa8
Signed-off-by: François Doray <fdoray.eclipse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19029
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: Make CtfTmfEventFieldTest easier to maintain.

Before this patch, a test event was created by filling a buffer with 0x02
bytes. Then, 0x00 bytes were added at magic positions to terminate the
strings used in the tests. That made the code hard to read and maintain. In
particular, it was hard to add new values in the test event without breaking
the null-terminated strings. It was also hard to understand, for example, that
0x02020202 is the float 9.551467814359616E-38.

With this patch, values in the right format are added explicitly to the test
event. It's easier to understand the tests and to add values for new tests.

Change-Id: I5d482442a10f09627b3ed2b32a9a8cc68c0f5cdc
Signed-off-by: Francois Doray <fdoray.eclipse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19192
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: fix multiples coding style issues.

There is no behavior change in this patch.

Change-Id: I892fe351754363768210ba4bd333e39e339533ca
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19242
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: Hide internal functionalities of StringDefinition.

The StringDefinition should not expose its internal data structures.
The StringBuffer is a temporary buffer used by read. Nothing outside
the 'read' method should ever play with this object.

Change-Id: I2edda57f9208f4d1a56a3aa619545f803ea4278d
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19235
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

ctf: Improve int decoding in IOStructGen

Now handles "0x10" and "010" better, by using Long.decode() instead of
Long.parseLong().

Change-Id: I25d9cd239063674d52a73f00b3f4af90e1ddff33
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19272
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

ctf: support traces with no content but a packet size

Change-Id: I979e3826580db246263cfcb219ee7f9cba968922
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19023
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

tmf: Fix Linux drawing weirdness in Histogram

- Set anti-aliasing to off when drawing histogram bars because it's on
by default in Linux

- Change order of drawing histogram bars and draw one pixel too far
because Linux doesn't draw last pixel of a line

- Clamp rounded rectangle arc size to rectangle width because Linux
doesn't

Change-Id: I574c80ecf15cda8655b3fbb8b34b80fd0f5a6d0c
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19263
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Hudson CI

ctf: avoid temporary array creation when parsing CTF-Array.

The temporary List<long> is useless and the conversion function too.

Change-Id: I42bb50536deb1529998287fd9dae34c5f009c194
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19240
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: Succeed at failing on OOME in tests

There is a test that fails with an OutOfMemoryError. This catches the
failure and thus makes the test succeed at failing.

Change-Id: If4b1a14bf135da3fae9f356631440acede56673c
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/19001
Tested-by: Hudson CI

ctf: Fix out of bounds positioning of the bitbuffer

rename private fields to follow fNameHere style

Change-Id: Ie579863b352a99e70cc9c1887d647aaa6018c9d4
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19024
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

tmf: Update documentation for Open File, legacy import removal

Change-Id: Ib2c636e0e75ab8923a12728bec12260bd051ce7d
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19090
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

ctf: Fix decoding of an invalid packet id type.

When an event id is not present, it is assumed to be zero.
Otherwise, it must be a _valid_ integer.

The current implementation accepts a "struct" as an id and map to the event id zero.

This fix throw an exception on an invalid id.

Change-Id: Iff6f6bd0173048dd3fb9b00f564948d7ebd55f4f
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19166
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

ctf: improve code clarity by merging if

Change-Id: I8ca0351762d36a5e25ba861e9fbc90ed14f5620a
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19160

ctf: Cleanup BitBufferIntTest

Update Javadoc, use correct order for (expected, actual) in asserts,
remove unneeded variables, etc. Hopefully they'll start looking less
like copy-pasted CodePro stubs.

Change-Id: Ie73748217201e93f9101a7c74928022857b93153
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/19161
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: remove false positive in the case of cross-packet events

Change-Id: I8ba4d1edee641aa52c03c2f7742f672700b3282f
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19006
Tested-by: Hudson CI
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

ctf: make float use readlong for 64 bit floats

Change-Id: I15dd34cd4a5068894155bac33a8ada58d35fcfca
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19015
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

tmf: Import trace package without manifest

Change-Id: I4e97dcc25734b6a5a91ffb35bfa38a94246b4b85
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/18591

tmf: Import and export multiple traces in a package

Change-Id: I935a1fcf76518bcad02940f8cc687389dae89673
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/18533

ctf: Remove unused alignment comparison

This code just do nothing.
What was the original purpose?

Change-Id: I0bf24dcdf5cb607c1af0a3326dbab60d51c6fbd4
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19167
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

tmf: Fix a method name typo in CtfTmfTrace.

There is not behavior change in this patch.

Change-Id: Ib1c043ca13528e7b307275f3728ffab14ca2ac1a
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19092
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

ctf: Fix Integer 33-63 bit handling

Thanks Etienne Bergeron for the initial patch.

Change-Id: I388280ef111fe3fd50729780f13e984cbf648886
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/18909
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Tested-by: Hudson CI

tmf: Update Javadoc in CtfIterator

And removed the clone() method, it wasn't used and clone() is bad.

Change-Id: Ib82eb78a50fe4ad1a0836cb98f99f4cc0da9606a
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/19155
Tested-by: Hudson CI

ctf: BitBuffer.java api overhaul

* Now 3 getters are public: get(length, extendSign), getInt(), and getLong().

Change-Id: I6848b2bde4e37580c1327f03cd025ea2bae4e134
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/18908
Tested-by: Hudson CI
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-by: Etienne Bergeron <etienne.bergeron@gmail.com>

ctf: Throw CTFReaderException in the BitBuffer API

Instead of the hard-to-catch BufferOverflowException. This ensures
that users of the library handle cases where the trace reader fails
to read correctly (corrupted traces, twisted test traces, etc.)

Change-Id: If79e6f4fa9650c08e66f22c91a4cf4d250551e40
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/19091
Tested-by: Hudson CI

tmf: Bug 416660: -1 long state value is not long

- Do not return nullValue when creating a new int or long with a value
of -1. This ensures that these values can be used and be differentiated
from a null value.

- Change occurences of -1 null value check to use isNull() method.

Change-Id: I7c9e80963736e701a0b30f5f3009a0c7084527bb
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/16396
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI

tmf: Cleanup TmfStateValue to remove 1-level of indirection.

This is a refactoring to diminush the memory usage of the
StateSystem.

In the current implementation, each TmfStateValue has a
"box" for an other object (i.e. Integer, Double, String, ...).
We propose to remove this indirection and use the raw type inside
the corresponding type.

As an example, IntegerStateValue now contains an "int" instead
of an "Integer".

Change-Id: Iaa75172be42174ebd6dc7d9bdd434631077c0146
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19027
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

Merge branch 'master' into luna

Conflicts:
lttng/org.eclipse.linuxtools.ctf.core/src/org/eclipse/linuxtools/ctf/core/event/types/Definition.java

Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Change-Id: I59e02fdb4d9f58edd817dcacb6852458097a1ca2

ctf: add early exit to alignRead

Change-Id: Ic1467ec9cdb6820e306bd92ec3137ed185c72dd8
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/18630
Tested-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
IP-Clean: Patrick Tasse <patrick.tasse@gmail.com>

ctf: improve alignment performance

Change-Id: I1ace35dc0002285874f1d0c3d9a841d2fb9900fd
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/18589
Tested-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
IP-Clean: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Fix NPE in the TimeGraphControl.

On TMF startup, the variable fEventColorMap may be <null>.

for (Color color : fEventColorMap) {

Change-Id: I83db13bffa93c4c1fbbcbc21198f3660b56866b3
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/19025
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Hudson CI
Reviewed-by: Genevieve Bastien <gbastien+lttng@versatic.net>

tmf: Add Open Trace file, directory under Traces folder

Change-Id: I4acb8121c8cb2915117f1bf5a501bc246a93877d
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/18695
Tested-by: Hudson CI

tmf: Remove legacy import wizard

Change-Id: I5915c26efd4f076959ea791d8b7d33b2cec1ce56
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/18694
Tested-by: Hudson CI

ctf: Catch ANTLR runtime exceptions in Metadata parsing

Change-Id: Ia50ee58b56a6e50b213e008950322c2e04231c28
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/19000

ctf: fix CTF-testsuite to only test traces and not scripts

Change-Id: I7575b2cbe1eae5bc9019496515c081d51f5a6bdd
Signed-off-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/19011
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

Merge branch 'master' into lttng-luna

Conflicts:
lttng/org.eclipse.linuxtools.tmf.core.tests/src/org/eclipse/linuxtools/tmf/core/tests/request/TmfSchedulerTest.java

Change-Id: I54ee583ea1075e7b50a0008b1940c00c7412a41a

lttng: Update user guide for histogram changes

Change-Id: Ia9b5cdab10f6444acd035f9f57ddac6031dbb8d3
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18938
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Hudson CI

TMF: TimeGraphPresentationProvider can notify listeners of color change

Bug 419282: If a presentation provider see its state table change and modify
colors, it fires a notification to listeners
The TimeGraphControl implements that listening interface.

Change-Id: I985576b10e0f5dcff154c555c307433193065d81
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/18764
Tested-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
IP-Clean: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Disable scheduler executionOrder test

That unit test has always been prone to multi-threaded uncertainty,
but recently it has been failing a lot more, which is very annoying
for the CI. I suspect it's because of the recent changes to the CTF
parser, some requests may end up finishing faster than they did before,
and before another request could be scheduled.

I suggest disabling it until we rewrite it to be more permissive (or
better, add testing mechanisms to the scheduler, and test the order
of requests being *sent*, not the order in which they finish).

Change-Id: I70db04feb9f86534f5d12623cf33f06d56b3a737
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/18942
Tested-by: Hudson CI
Reviewed-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>

TMF: add support of cut/copy/paste/dnd in FilterView

Signed-off-by: Xavier Raynaud <xavier.raynaud@kalray.eu>
Change-Id: I67f15cd9c1dcb046de8dc6eef196012be14d9265
Reviewed-on: https://git.eclipse.org/r/17019
Tested-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
IP-Clean: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

TMF: Add assert exists and dispose to traces in synchronization tests

Change-Id: Ida426ccb6abe6404aa6c443dc5ae38b8d175f745
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/18862
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

Merge branch 'master' into lttng-luna

ctf: Fix multiple coding style and typo issues while reading code.

Change-Id: If63330f4c86e7cd87cf979e1709a81fb7c8958ab
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18541
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Hudson CI

ctf: Plug the unsigned utils comparator into Stream comparator.

This patch-set fixes a pending TODO found randomly.
It modifies the code to use the unsigned comparator and avoid
bugs for large timestamp values.

The comparator no longer accept null values.

Change-Id: Idee0e56edfe85ba01a27fb79d1076736fee24874
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/18785
Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Tested-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
IP-Clean: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Fix core tests failing on Windows

2 tests did not dispose the index correctly. Solution: make sure
fCheckpointCollection gets assigned when recreating the collection so
that it gets disposed in the tearDown.

Change-Id: I6a79b62cdabf136790bcf76ca606e71c31cdd81b
Signed-off-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>
Reviewed-on: https://git.eclipse.org/r/18481
Tested-by: Hudson CI
Reviewed-by: Patrick Tasse <patrick.tasse@gmail.com>
IP-Clean: Patrick Tasse <patrick.tasse@gmail.com>
Tested-by: Patrick Tasse <patrick.tasse@gmail.com>

tmf: Add timeouts to potentially hanging tests

Change-Id: Ia429dad124844abbac44afa644250238c8a60f5a
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/18863
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Reviewed-by: Etienne Bergeron <etienne.bergeron@gmail.com>

tmf: Remove unused TmfSortedArray.

This data-structure seems unused.
It's implemented with an insertion sort, and won't scale.

Change-Id: I49ae518a384817c4fd537c989240454e00f6f38b
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18875
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

[ctf] Fix coding style issues in BitBuffer.

Fix potential issue with an undefined behavior in CTF.
- Reading a integer of zero bit at the end of the buffer may be valid.
- Change the behavior to accept it.

Change-Id: I22fccc4e2ffee3bbf796e52b516b00f7fc5e31ac
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18707
Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

[ctf] Fix binary search for a long sequence of same timestamps.

The binary search algorithm does not need a third case
(when values are equals) if there is a guarantee to remove
one element in the search space at each iteration.
The actual binary search implementation performs a sequential
search on elements with the same timestamp (to find the first
one).

The ETW2CTF traces produce many debug events for the debugging
information at the module load timestamp. Which is the worse
case for the actual implementation.

We changed the algorithm to use the timestamp end of a packet
instead of the beginning. We changed the way to choose a
middle element to ease the recursion by using only two cases.

Change-Id: I4f16d43b9533f8f1449cdb3c4c213bcb9f962daf
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18654
Reviewed-by: Matthias Nick <Matthias.Nick@bsiag.com>
Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Tested-by: Hudson CI
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

TMF: Check presence of fields for TCP event synchronization

Traces using an old version of the TCP addons module would get a NPE when
getting the key to match packets.

Change-Id: Ib0470834807b1933f5f46069c5377115764b9dda
Signed-off-by: Geneviève Bastien <gbastien+lttng@versatic.net>
Reviewed-on: https://git.eclipse.org/r/18848
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>

ctf: Replace Long.compare() with Java 6 compatible method

We still technically support Java 6, and Long.compare() was added in 7.

Change-Id: I565b40bbc11c1c1c3b05a4097cc0e619598d0553
Signed-off-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-on: https://git.eclipse.org/r/18841
Reviewed-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>

tmf: Update histogram text controls

- 'Current Event' text control replaces with 'Selection Start' and
'Selection End' controls
- Text control layout updated
- Added link button to force single selection time
- Added selection span to histogram tool tip when mouse hovers over
selection range
- Fixed text control to support keypad Enter key
- Fixed to ignore invalid window span value
- Externalized tool tip messages

Change-Id: I2158dd579b2e508b3c2d0332aedc6b34626e6feb
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18634
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Hudson CI
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Update histogram layout

- Both histograms fill all available space with no minimum height
- The max nb event text control grows in width according to its maximum
(watermark) text extent but never shrinks, eliminating layout jitter
(bug 375990)

Change-Id: I1ca835434f3e0fca4196aa7f23eb29dead2f50db
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18355
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Support drag zoom in histogram

- Right button drag is used to select a new zoom window in both
histograms
- Shift-right button drag or click is used to modify the zoom window in
full range histogram
- Zoom is cancelled if the selected range duration is zero
- New time range is propagated when the mouse button is released
- Move draw time range window method to base class

Change-Id: If46c1edd5c8f67a1b48ebc2ebbe2527613ba927c
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18310
Tested-by: Hudson CI
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Support drag move in time range histogram

- Add support for bucket display offset in histogram
- Middle button or Ctrl-left button are used to drag the window in time
range histogram
- New time range is propagated when the mouse button is released
- Remove restriction of selection buckets to visible range in data model
- Adjust histogram tool tip according to bucket display offset
- Prevent histogram tool tip when mouse is not over visible time range

Change-Id: I9ee3663511b57ac32a668cc25cf97e903a72e60f
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18309
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Support drag selection in histogram

- Left button drag makes a range selection in both histograms
- Middle button or Ctrl-left button are used to drag the window in full
trace histogram
- Middle-click or Shift-left-click on full trace histogram now centers
the window on the mouse position
- Selection is only propagated when the mouse button is released
- When moving the mouse only redraw instead of refreshing the whole
histogram
- Fix range selection start on Shift-click
- Limit selection range to visible time range
- Draw selection region background corrected when end time is before
start time
- Prevent mouse wheel zoom when dragging in progress

Change-Id: I393e0e72cd2f445be2bd75bb90090f27abc2a2da
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18308
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Marc-Andre Laperle <marc-andre.laperle@ericsson.com>

tmf: Fix time range histogram to show its full range

- Set the end time of the time range histogram to the time range end
time (instead of the last event time)
- Display the time range text controls even when model has no events
- Snap selection to the end time
- Draw the delimiter to the right of the end time instead of last event
time
- Fill the area to the right of the delimiter with the background color
- Draw selection range anywhere in the histogram range regardless of
last event time

Change-Id: I913d22a20d79a3ca138902acd0d9cc815918cbc9
Signed-off-by: Patrick Tasse <patrick.tasse@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18307
Tested-by: Hudson CI
Reviewed-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
IP-Clean: Bernd Hufmann <bernd.hufmann@ericsson.com>
Tested-by: Bernd Hufmann <bernd.hufmann@ericsson.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

ctf: Remove unused StreamInputReaderComparator.

This is a clone of StreamInputReaderTimestampComparator.

Change-Id: Ib640cbafe8105bad2c18a3f5e35136e0a5693597
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18796
Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>

[ctf] Simplify the logic of the unsigned long comparator.

To understand this modification, let assume the "char 8-bits" domain.

unsigned [0 .. 255]
signed [-128 .. 127]

By receiving the parameters encoded in a signed number, the method receives

left:  [0..127,-128..-1]  (which represents [0..127,128..255])
right: [0..127,-128..-1]  (which represents [0..127,128..255])

By definition (on an unconstrained domain), this assertion holds

left < right  <==>  left + k < right + k

Thus, the idea is to rotate the domain by k to allow a signed operator to
be able to compare to full domain.
In this case k must be -128.

By rotating the domain, left and right range become [-128..-1,0..127],
and are now comparable with the signed operator.

Change-Id: I92b27ab00e8f14102a04e085ef807e211e39a7f0
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18784
Tested-by: Hudson CI
Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Reviewed-by: Xavier Raynaud <xavier.raynaud@kalray.eu>
IP-Clean: Xavier Raynaud <xavier.raynaud@kalray.eu>
Tested-by: Xavier Raynaud <xavier.raynaud@kalray.eu>

TMF: make BufferedRandomAccessFile fields private, and final if possible.

Change-Id: I213410e6e4cc5fe96e24c88d6eac61d6bb2b1b23
Signed-off-by: Xavier Raynaud <xavier.raynaud@kalray.eu>
Reviewed-on: https://git.eclipse.org/r/18711
Tested-by: Hudson CI
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
IP-Clean: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Reviewed-by: Etienne Bergeron <etienne.bergeron@gmail.com>

Avoid multiple seek to the same location.

There is multiple instance of seeking to the same location twice.
To solve this issue the seek method check if the current location is the same.
Here is the explanation:
[CtfTmfContext.java]
    public void setLocation(ITmfLocation location) {
        curLocation = (CtfLocation) location;
        if (curLocation != null) {
            getIterator().seek(curLocation.getLocationInfo());
        }
    }
GetIterator  --> will force a seek to curLocation
.seek        --> force an other seek to curLocation
Here are the called method:
    private CtfIterator getIterator() {
        return CtfIteratorManager.getIterator(fTrace, this);
    }
    public static synchronized CtfIterator getIterator(final CtfTmfTrace trace,
            final CtfTmfContext ctx) {
        return map.get(trace).getIterator(ctx);
    }
    public CtfIterator getIterator(final CtfTmfContext context) {
            [...]
            if (context.getLocation() != null) {
                final CtfLocationInfo location = (CtfLocationInfo) context.getLocation().getLocationInfo();
                retVal.seek(location);   /// FORCE SEEK!!!!!!!!!!!!
            }
Change-Id: Ib5ad6d05eb47045660c7ab5076930a30e829c7a0
Signed-off-by: Etienne Bergeron <etienne.bergeron@gmail.com>
Reviewed-on: https://git.eclipse.org/r/18461
Reviewed-by: François Doray <fdoray.eclipse@gmail.com>
Tested-by: Hudson CI
Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Reviewed-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
IP-Clean: Matthew Khouzam <matthew.khouzam@ericsson.com>
Tested-by: Matthew Khouzam <matthew.khouzam@ericsson.com>
Reviewed-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>
Tested-by: Alexandre Montplaisir <alexmonthy@voxpopuli.im>