aboutsummaryrefslogtreecommitdiffstats
path: root/tests/auto/qml
Commit message (Collapse)AuthorAgeFilesLines
* Remove extra addrefMichael Brasser2017-12-293-0/+44
| | | | | | | | | The code was previously updated to use QQmlRefPointer, so we shouldn't explicitly addref. This allows more components to be correctly trimmed when needed. Change-Id: I15a961cfc456eeab5c791c8a282cc7e2852912cb Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Use potentially intercepted URL as ID for compilation unitsUlf Hermann2017-12-212-0/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | We generally have to pass a URL and a file name everywhere because the logical URL might be something else than the actual file being loaded. For example a QQmlFileSelector might modify the URL to be loaded for a specific file. This resulting URL, however, should not be used to resolve further URLs defined in the file loaded that way. As we need to access QQmlTypeLoader::m_url as string more often now, cache it and avoid frequent translations between QUrl and QString. Furthermore, QQmlDataBlob's URLs are changed to follow the same semantics. The finalUrl is the one that should be used to resolve further URLs, the url is the one used to load the content, and subject to any redirects or interceptions. This changes the semantics of URL redirects. Previously a redirected URL was used as the base URL for furher URL resolution. This doesn't work because redirection occurs after interception and interception should not influence the resolution of further URLs. We now use the original URL as base URL for resolution of further URLs and rely on the server to redirect those, too. Task-number: QTBUG-61209 Change-Id: I93822f820bed2515995de3cb118099218b510ca4 Reviewed-by: Michael Brasser <michael.brasser@live.com>
* Add the source directory as import paths in tst_qqmlimportUlf Hermann2017-12-111-3/+3
| | | | | | | | On shadow builds the QML components won't be found in the build directory. Change-Id: I0f524063537669f68e64fc6c04a6756ed1d130fd Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Pass the source directory in the plugindump testUlf Hermann2017-12-112-1/+2
| | | | | | | | On shadow builds this fails when just passing ".", because "." is the build directory. Change-Id: Iee84b73f2c4e5c8663d84d53b31f658501244dc9 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Add a means to unregister custom qml typesAndy Shaw2017-12-013-0/+209
| | | | | | | | | | | | | | | In cases where Qt is used in a plugin it is possible that a plugin will be unloaded while Qt itself is still loaded and as a result there is a chance that there will be conflicting types registered. Therefore, to ensure that plugins correctly clean up after themselves cleanly, we need to add a means to unregister qml types. This is intended to only be used when the user knows what they are doing. Task-number: QTBUG-56521 Task-number: QTBUG-56532 Change-Id: Ie396e522385004e6e9f3841e04f8072ff29cb15b Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* More fine-grained deferred property executionJ-P Nurmi2017-11-251-0/+152
| | | | | | | | | | | | | | | | This allows Qt Quick Controls 2 to defer the execution of certain building blocks until needed. For example, a button control can defer its background item so that the default background is not executed at all when replaced by a custom background. First of all, this gives a massive performance boost for customized controls. Secondly, this avoids the most burning issue in QQC2, problems with asynchronous incubation ("Object destroyed during incubation"). Task-number: QTBUG-50992 Change-Id: If3616c9dac70e3a474a20070ad0452874d267164 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Re-enable QML memory profilingUlf Hermann2017-11-233-1/+40
| | | | | | Task-number: QTBUG-64674 Change-Id: I48ed1a51f66ef8d55cc026f140d270baaca04fbf Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Don't reject plugin-only qmldir filesUlf Hermann2017-11-222-12/+41
| | | | | | | | | | | | On QQmlImportsPrivate::updateQmldirContent we need to check if the new module has actually been established after figuring out that it doesn't have any components or scripts. If it has, then we shouldn't fail, as obviously a plugin has been loaded. We don't need to check the component and script versions in that case, as plugins don't have separate versions. Change-Id: Ie328b59038fe65c3f6a2eeecfe969927bba6cd68 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Fix URL interception for qmldir filesUlf Hermann2017-11-223-0/+262
| | | | | | | | | | | | | | | | | We need to intercept the URL when it is created. This relieves us of the need to hack around in it when actually retrieving the content of the qmldir file and prevents the futile attempt to load remote qmldir files via the code path that should load local ones (or vice versa). The back and forth conversion between URLs and strings is unfortunate, but can only be solved by using QUrl rather than QString where we actually mean URL. This would be a bigger change which is unsuitable for 5.9. Mind that nothing changes for code that doesn't use URL interceptors. Task-number: QTBUG-36773 Change-Id: I6bff3ae352009fdc0a17ec209691c7b390367f11 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* QQmlEngineDebugService: Check QML contexts for validityUlf Hermann2017-11-051-0/+39
| | | | | | | | | We should not operate on invalid QML contexts and once we have established a context to be valid we don't have to check the result of QQmlContextData::get anymore. Change-Id: I9106115ddf925c3572048f1fd334bdfd9a9cfca7 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Fix outdated BSD license header in tests to use GPL-EXCEPTKai Koehne2017-10-231-29/+17
| | | | | Change-Id: I64a7b7cb7fa6c5fe53019ed42b19989a0f8a0da2 Reviewed-by: Jani Heikkinen <jani.heikkinen@qt.io>
* QQmlTypeLoader: Drain events before shutting down the threadUlf Hermann2017-10-203-0/+52
| | | | | | | | | | We take references to types when sending events to the other thread. If we don't process the events, the references are kept, which leads to memory leaks. Therefore, when shutting down a QML engine, we have to make sure the event queues are emptied. Change-Id: Id8b0440029cfd7d03a9e540747eaedbcaa7c9ff3 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Tell QQmlImportInstance::resolveType what kind of type we wantUlf Hermann2017-10-192-0/+32
| | | | | | | | | | | | | | | | | | | | | | | | | | | In QQmlTypeData::resolveTypes() we know if we're looking at a reference to a composite singleton type, or some other type reference. When we call resolveType() we expect the correct type to be returned, not only based on URL, but also based on its singleton property. QQmlTypeData::resolveType() eventually invokes QQmlImportInstance::resolveType() which will call fetchOrCreateTypeForUrl(), passing a parameter on whether the result should be a composite singleton. When operating on a qmldir component the component itself encodes this. When fetching a type from a local file without qmldir, we currently assume that it isn't a singleton, no matter QQmlTypeData::resolveTypes() has determined. This means that actual singletons loaded this way later get refused by the sanity check. In order to fix this, pass the information about the expected singleton property on to QQmlImportInstance. This is done using QQmlType::RegistrationType, which gets another entry for "any type". If the expected type is CompositeSingletonType QQmlTypeData::resolveType() will not create a non-singleton type. If it is any specific other type, it will not create a composite singleton. And if it is AnyRegistrationType, it will behave as it previously did. Change-Id: I6b7e082b63582e0aed946bb3d19077b94c7a45f7 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Fix execution of deferred propertiesJ-P Nurmi2017-10-185-0/+102
| | | | | | | | | | | When deferred properties were assigned in multiple contexts, only the outermost context was executed. Any deferred property assignments in other inner contexts were never executed. Collect the deferred data to a container to be able to execute them all. Task-number: QTBUG-63200 Change-Id: I88fab27c1f81b5188430ada086dcc19842507e99 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Fix outdated BSD license headerKai Koehne2017-10-175-10/+60
| | | | | Change-Id: Icc08925454445fc9497fb3bfd2c26efe90605983 Reviewed-by: Jani Heikkinen <jani.heikkinen@qt.io>
* ObjectModel: provide unique move IDsJ-P Nurmi2017-09-261-5/+5
| | | | | | | | | | | | QQuickItemViewPrivate::removeItem() uses a QHash to store items that were removed due to a move. If the move IDs are not unique, multiple buffered moves end up overriding each other. This results to leaked items that are never released. Task-number: QTBUG-62607 Change-Id: I7e7e7fcd6b1b0aa50ed55643ba5674e98536f89f Reviewed-by: Mitch Curtis <mitch.curtis@qt.io> Reviewed-by: Michael Brasser <michael.brasser@live.com>
* Fix qmlClearTypeRegistrations() not dropping all registrationsSimon Hausmann2017-09-192-9/+52
| | | | | | | | | | | | | In commit 48c09a85ce397979c7e706e3694c879ffe456e09 we added the undeletableTypes container to hold a reference on C++ registered types to keep the indices returned by the public qmlRegisterType() API stable. Since qmlClearTypeRegistrations() is API that also resets those indices, we must also clear the undeletableTypes container to avoid leaking memory. Change-Id: I2038c00913f894d58aca3714d64d497493585326 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Fall back to the ObjectWrapper for model advanceIteratorAndy Shaw2017-09-151-0/+23
| | | | | | | | | | | | When falling back to the QObjectWrapper it will add in the extra parts added when the roles were added to the object created by the model to hold the data being returned. This was causing the last entry to be duplicated and causing extra work too. Task-number: QTBUG-54285 Task-number: QTBUG-62156 Change-Id: I2907477277df8d16db4491a4999f004433e4205c Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Error out when compiling signal handlers with arguments in qml filesSimon Hausmann2017-09-121-2/+33
| | | | | | | | | | | | | | | Ahead of time we cannot tell whether the use of "arguments" in a signal hander refers to the JS arguments object or a potential arguments signal parameter. Resolving that requires access to information we currently don't have. The QML engine has it at run-time (in SignalHandlerConverter) and that's why it works there accordingly. However when generating caches ahead of time, let's rather produce an error message with a hint how to work around it instead of producing differing behavior at run-time. Task-number: QTBUG-60011 Change-Id: I9e460bd467dbb5998f12a44c439223ea44e7bbad Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Fix qml cache invalidation when changing dependent C++ registered QML singletonsSimon Hausmann2017-09-121-0/+61
| | | | | | | | | | | | | | | | | | When a qml file uses a qml singleton, we need to reliably detect when the singleton changes and re-generate the cache of the qml file using it. This is a scenario covered and fixed by commit 5b94de09cc738837d1539e28b3c0dccd17c18d29, with the exception that currently QML singletons registered via qmlRegisterSingleton were not added to the list of dependent singletons for a qml file. We can fix this by extending findCompositeSingletons() to also cover the singletons that do not originate from a qmldir file. [ChangeLog][Qt][Qml] Fixed bug where sometimes changes to a qml singleton would not propagate to the users or cause crashes. Task-number: QTBUG-62243 Change-Id: I16c3d9ba65fd82e898a29b946c341907751135a9 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Fix reuse of regexp objects by regexp literalsAllan Sandfeld Jensen2017-09-081-0/+23
| | | | | | | | | | | | | Accoding to the standard the regexp objects created by literals should be separate objects as if calling new. We were violating that by caching the same object for every instance of a literal. This also fixes a problem with leaking values of lastIndex between separate instances of the same global regexp literal. Task-number: QTBUG-62175 Change-Id: Ib22e9ee68de1d1209fbd4212e72f576bc059d245 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Get rid of the root object index variableSimon Hausmann2017-09-082-3/+3
| | | | | | | | This is a follow-up to the parent commit to remove the variable that is really a constant (zero). Change-Id: I8fc20027c5c7b871269b814cb8b93636e94be267 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Disable deferring when referenced as a grouped propertyJ-P Nurmi2017-09-065-0/+49
| | | | | | | | | This allows us to fix QTBUG-50992 - the issue with most votes in QQC2. Task-number: QTBUG-63036 Change-Id: I996cd1128582b80e0c8480ae143d682c1e8eb8fe Reviewed-by: Simon Hausmann <simon.hausmann@qt.io> Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
* Fix crashes with closures created in QML componentsLars Knoll2017-09-063-0/+47
| | | | | | | | | | | | | | | When closures created inside QML components are called after the surrounding component (and consequently QML context) has been destroyed, we are in a somewhat limited environment. Initially we would just crash as the calling QML context is not valid anymore. We can alleviate that by introducing reference counting on the context and letting the QML context wrapper keep a strong reference. This avoids the crashes and also ensures that at least imports continue to be accessible within these contexts (as the singleton test case demonstrates). Task-number: QTBUG-61781 Change-Id: I893f171842d01b0863d95a02ea738adc2620e236 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Stabilize the threadSignal() testSimon Hausmann2017-09-063-14/+22
| | | | | | | | | | | | | | | | | | | | We've seen the case in the CI where we delete the worker thread object before it has had a chance (resulting in a crash). This patch attempts to stabilize this by waiting for the thread to terminate properly. In addition QSignalSpy's connection to the done(QString) signal is forced to be direct, which means the spy's internal list is accessed from the gui thread (via QCOMPARE) at the same time as the thread may be emitting the signal and calling the signalspy's slot (metacall), which helgrind complains about (rightly so). I don't see any purpose in connecting to the signal, so let's remove that code. The test continues to cover the threading code in QQmlData::signalEmitted, once as the thread is triggered via C++ and once via QML (doIt invocation). Change-Id: I5e8a4ae65e2d0890a26491d25c73de1ba33a6668 Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>
* Make sure QQmlDebugProcess correctly reports session as startedUlf Hermann2017-08-177-15/+4
| | | | | | | | | | | If the "Waiting" output appears before we wait, the event loop could run forever. Also, the timeout of 5s was too low. We increase it to 15s. Remove the blacklists and other workarounds in turn. Task-number: QTQAINFRA-1334 Change-Id: Ib1032a8e57ab8dada3e56163ebab1523a7357aeb Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Bump tests262 submodule to include .gitattributes fixSimon Hausmann2017-08-071-0/+0
| | | | | | | | This way we exclude the .gitignore file in the source archives. Task-number: QTBUG-59991 Change-Id: Ic0c2cf9031b890ec2a1d1208bcb65415ad0483fb Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Move the engine pointer from the property cache to the VME meta objectLars Knoll2017-08-021-12/+6
| | | | | | | | | This is where it belongs, and it makes the PropertyCache independent of the engine used. Task-number: QTBUG-61536 Change-Id: I21c2674ee3e2895abd2418764d140b154b47b868 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Use QQmlType by valueLars Knoll2017-08-022-55/+55
| | | | | | | | | | | QQmlType is now refcounted, and we need to use it by value, to control it's lifetime properly. This is required, so we can clean up the QQmlMetaTypeData cache on engine destruction and with trimComponentCache() Task-number: QTBUG-61536 Change-Id: If86391c86ea20a646ded7c9925d8f743f628fb91 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Ignore debugger failures with GCC on WindowsSimon Hausmann2017-07-2811-3/+53
| | | | | | | | | | | | | | | Capturing the output of the started process mysteriously fails on Windows with gcc. While the nested event loop is run, the readyRead() signal is never emitted. Only after the timer fired and the event loop is terminated, we receive the output we were expecting. Some tests needed adapting to the initializing sub-function failing in QVERIFY/QCOMPARE calls, in order for the process to not crash and the blacklisting to work. Task-number: QTQAINFRA-1334 Change-Id: I4804d94580e7db65595137d19d7b75d75c243257 Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@qt.io>
* Fix ObjectModel::move() to mark the changes as movesJ-P Nurmi2017-06-291-0/+44
| | | | | | | | | | | | | QQmlObjectModel::move() created a QQmlChangeSet with moveId -1, which made item views and controls see the changes as removals and insertions, because QQmlChangeSet::Change::isMove() returned false. Consequently, item views did not update the current index when the current item was moved. Task-number: QTBUG-60894 Change-Id: I4a64b7670c1fae12337995627437cc83efb9f1ef Reviewed-by: Michael Brasser <michael.brasser@live.com> Reviewed-by: Robin Burchell <robin.burchell@crimson.no>
* Fix loading QML caches for qrc:/// urlsAlexander Volkov2017-06-133-0/+64
| | | | | | | | | | | | | | | Running examples/quick/window/window produces debug messages: "QML source file has moved to a different location." This is because QQmlFile::urlToLocalFileOrQrc(const QString &) overload is incompatible with QQmlFile::urlToLocalFileOrQrc(const QUrl &) when it deals with qrc:/// urls. For example it returns ":///window/window.qml" while the QUrl overload returns ":/window/window.qml". Thus the comparison of source paths in CompilationUnit::loadFromDisk() fails. Fix the incompatibility and add a test. Change-Id: I20449b8cf13d715d88860f2cd413ab39c893f3ef Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* tst_qqmlengine::clearComponentCache(): Use QTemporaryDirFriedemann Kleint2017-06-082-6/+26
| | | | | | | | | The test used to create a file temp.qml in the working directory. Put it into a temporary directory. Change-Id: I0720a4b4c652c83656505a5dc979660b94503717 Reviewed-by: Oliver Wolff <oliver.wolff@qt.io> Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Also run the ecmascript testsuite in the interpreterErik Verbruggen2017-06-081-1/+1
| | | | | | Change-Id: Ibc3e67273bf01dccfad132b53aef6e5241883d97 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io> Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* Merge remote-tracking branch 'origin/5.9.0' into 5.9Liang Qi2017-05-318-153/+99
|\ | | | | | | Change-Id: I3f77593ca944114534fff5df26bbb09150ee4400
| * QML Settings: fix JS array handlingJ-P Nurmi2017-05-194-153/+65
| | | | | | | | | | | | | | | | | | | | | | Before Qt 5.4, JS arrays were passed as QVariantLists. Since Qt 5.4, they are passed as QJSValues instead. Use QJSValue::toVariant() (the same way as QQuickItemView::setModel(QVariant) which was fixed in cf959b4b) to convert JS values to QSettings-compatible variants. Task-number: QTBUG-45316 Change-Id: Icc6f8ad09bfef089d9efcf5b90e3783bb3f73a9f Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
| * QQmlIRBuilder: Only query type name cache for type namesRobin Burchell2017-05-194-0/+34
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The behavior here was always incorrect: type names must start with an uppercase letter, so querying the type name cache with a lowercase string is wrong. However, this was turned into a larger problem by making more extensive use of QQmlTypeNameCache in e74a1d0b342f2c95dc3a543c8c9ec07fd52d8fe0, as it contained a lot of new types (including composite types, which previously were only in the cache if they were singletons). Task-number: QTBUG-60547 Change-Id: I40be2d535e99d3e1af250d995d7149ecbe2965d7 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* | Set LC_TIME in ecmascripttests/test262.pyJüri Valdmann2017-05-231-0/+1
| | | | | | | | | | | | | | | | One test fails with LC_TIME=et_EE.UTF-8. The script already sets LANG, but LC_TIME has higher precedence. Change-Id: Ifdb37a1a9d69415d34883343e761d23c6353ef24 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* | Get rid of the old way of defining builtin functionsLars Knoll2017-05-191-4/+4
| | | | | | | | | | | | | | | | | | | | The old calling convention used for builtin functions is very inefficient. It was still being used in a few places. Clean those up and convert them to the new and much more effiecient calling convention. Change-Id: I6b769c6185df7e9be1e80709330fc1ca868576c1 Reviewed-by: Robin Burchell <robin.burchell@crimson.no>
* | Move the prototype into the internal classLars Knoll2017-05-191-1/+1
| | | | | | | | | | | | | | | | | | | | This saves another pointer on all Objects. Currently introduces a slight performance regression on some of the v8 benchmarks, that needs addressing. Change-Id: I87de8e1d198d2683f4e903c467ce2a60ba542243 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* | Fix QML Connections element ignoring the enabled propertyThomas McGuire2017-05-172-0/+32
|/ | | | | | | | | | | The enabled property was ignored if it was set before componentComplete() was called. [ChangeLog][QtQml] Fixed the QML Connections element ignoring the initial state of the enabled property Change-Id: I40c92fcb30f0bb8ca406f248b3bde2fced5ab58f Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Fix build of tst_qqmlengineSzabolcs David2017-05-091-0/+1
| | | | | | | | | | | | It fixes the following compile error: tst_qqmlengine.cpp:279:51: error: variable ‘QCryptographicHash md5’ has initializer but incomplete type tst_qqmlengine.cpp:279:28: error: incomplete type ‘QCryptographicHash’ used in nested name specifier Change-Id: I647a5a487d79f201118bfbddd757aa9cece180e4 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io> Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
* Fix licensingJani Heikkinen2017-05-042-33/+23
| | | | | | | | | - Remove unused license files - Switch old LGPLv21 license headers with GPL-EXCEPT one Task-number: QTBUG-57147 Change-Id: Ib59c3e2e39bfe0038db795af85dc75028564efa3 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
* QQmlComponent: Fix heap buffer overflow with bogus inputPeter Hartmann2017-05-031-0/+12
| | | | | Change-Id: I8a725018a5aeb39df370f856cd77d887faa511e3 Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Fix concurrent loading of the same qmldir from different scriptsAndy Shaw2017-05-024-0/+35
| | | | | | | | | | | | | | | | QQmlQmldirData keeps a pointer to a QQmlScript::Import, and an integer priority. Each Blob that is waiting on it was setting its own import and priority even though the QQmlQmldirData itself was shared. This resulted in whichever one began loading last succeeding to load, and the rest failing. This change instead stores the import and priority data per-dependent Blob Fix was originally done by Josh Faust <jfaust@suitabletech.com>. I added the test. Task-number: QTBUG-30469 Change-Id: Id3d15569a999a7c22eeb12b431e5daf1ddae51dc Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* Merge "Merge remote-tracking branch 'origin/5.8' into 5.9" into refs/staging/5.9Simon Hausmann2017-04-272-0/+23
|\
| * Merge remote-tracking branch 'origin/5.8' into 5.9Simon Hausmann2017-04-272-0/+23
| |\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Conflicts: src/qtqmlglobal.h src/qtqmlglobal_p.h src/jsruntime/qv4global_p.h src/qml/compiler/compiler.pri src/qml/compiler/qv4ssa.cpp src/qmldevtools/qtqmldevtoolsglobal_p.h tests/auto/qml/qqmlecmascript/tst_qqmlecmascript.cpp Change-Id: I55c5d015b2cb1053b83b9c61caaf004fb49ee486
| | * V4: Fix issues with very small loops5.8Erik Verbruggen2017-04-182-0/+24
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Loops consisting of just a single basic block (e.g. a do-while loop with no nested loops or if statements) have a back-edge to themselves. There were 2 problems: - loop detection would create LoopInfo for any loop header referred to by blocks inside the loop (and a 1 block loop doesn't have body blocks), nor would it mark the loop header as such - when splitting critical edges, the newly inserted block would not be marked as part of the loop This is a problem specifically for 1 block loops: the block ends with a CJUMP, so the back-edge is a critical edge. So the new block inserted by edge splitting wouldn't be marked as belonging to the loop. The end result was that the life-time intervals for temporaries that are defined before the loop, but that are used inside the loop, and not after the loop, would have their life-time ended before the loop ends (instead of spanning the whole loop, *including* the back-edge). This in turns could lead to the stack/register allocator re-using the storage for that temporary, resulting in strange things happening. Task-number: QTBUG-59012 Change-Id: Ic946c73913711272efea2151cb85350412ca2fde Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* | | QQmlApplicationEngine: Fix using invalid pointersRainer Keller2017-04-261-0/+19
|/ / | | | | | | | | | | | | | | | | | | When a root object was deleted before QQmlApplicationEngine the invalid pointers stayed in the list of root objects leading to crashes when destructing QQmlApplicationEngine. Root objects are watched for destruction and removed from the list. Change-Id: I1babab54dbb7d7b16ed883ada5b3e420ca8690cb Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
* | Add an autotest for GC crashes on some platformsRobin Burchell2017-04-212-1/+39
| | | | | | | | | | | | | | | | | | Related to QTBUG-59977, where crashes were seen on some platforms. Try to prevent this from creeping out and inconveniencing other modules again. Task-number: QTBUG-59977 Change-Id: I89730c7882e4d87e6049f087724988a7aa52d5a5 Reviewed-by: Lars Knoll <lars.knoll@qt.io>
generated by cgit v1.2.3 (git 2.25.1) at 2024-05-18 12:00:16 +0000