When a foreign event loop that does not enter an alertable wait state
is running (which is also the case when a native dialog window is
modal), pipe handlers would freeze temporarily due to their APC
callbacks not being invoked.
We address this problem by moving the I/O callbacks to the Windows
thread pool, and only posting completion events to the main loop
from there. That makes the actual I/O completely independent from
any main loop, while the signal delivery works also with foreign
loops (because Qt event delivery uses Windows messages, which foreign
loops typically handle correctly).
As a nice side effect, performance (and in particular scalability)
is improved.
Several other approaches have been tried:
1) Using QWinEventNotifier was about a quarter slower and scaled much
worse. Additionally, it also required a rather egregious hack to
handle the (pathological) case of a single thread talking to both
ends of a QLocalSocket synchronously.
2) Queuing APCs from the thread pool to the main thread and also
posting wake-up events to its event loop, and handling I/O on the
main thread; this performed roughly like this solution, but scaled
half as well, and the separate wake-up path was still deemed hacky.
3) Only posting wake-up events to the main thread from the thread pool,
and still handling I/O on the main thread; this still performed
comparably to 2), and the pathological case was not handled at all.
4) Using this approach for reads and that of 3) for writes was slightly
faster with big amounts of data, but scaled slightly worse, and the
diverging implementations were deemed not desirable.
Fixes: QTBUG-64443
Change-Id: I66443c3021d6ba98639a214c3e768be97d2cf14b
Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Avoid painting errors with dpr < 1, also for the PassThrough
mode.
This limits the minimum high-dpi scale factor to 1,
for the code path which determines the scale factor
based on screen DPI.
Pick-to: 6.1
Task-number: QTBUG-89948
Change-Id: I14b3f130f0ae141d5f05c87437f926a9f76d1dec
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
After QPoint(F), it's now the time of QSize(F) and QVectorND,
which can be unambiguously decomposed.
[ChangeLog][QtCore][QSize] QSize is now usable in a structured
binding declaration.
[ChangeLog][QtCore][QSizeF] QSizeF is now usable in a structured
binding declaration.
[ChangeLog][QtGui][QVector2D] QVector2D is now usable in a
structured binding declaration.
[ChangeLog][QtGui][QVector3D] QVector3D is now usable in a
structured binding declaration.
[ChangeLog][QtGui][QVector4D] QVector4D is now usable in a
structured binding declaration.
Change-Id: I67bb152f4210f2be27607179cd2ec522174cc483
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Only RGB444 and RGB666 were treated slighlty different from the rest,
but the test had a few additional mistakes.
Pick-to: 6.1 6.0 5.15
Change-Id: I4728b4036affedfffce8bca5c1e7be3869344fbe
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
When changing transferfunction the look-up-tables needs to be
regenerated.
Pick-to: 6.1 6.0 5.15
Change-Id: I83ca5fe570f85d478a374f52c0a82db84e70c3b8
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
The emulation detection has been usable only on qtbase tests, move it to
QTest so that it can be used in other modules as well.
Pick-to: 6.1
Change-Id: I4b2321b7856414d7b1cfd5e6b1405a633c6bb878
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
A few formats were not treating the input QColor correctly. Fixed and
added more exhaustive test.
Pick-to: 6.1 6.0 5.15
Change-Id: I872aeeb45e518f9a34b4ac35642264821f9927f2
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Calculate the effective width of the hyphen better, and compare with
ceiled sizes.
Pick-to: 6.1 6.0
Fixes: QTBUG-90698
Change-Id: I7ed2eb44c54240ecb2f8a38e5acf1f32608b2bfb
Reviewed-by: Eskil Abrahamsen Blomfeldt <eskil.abrahamsen-blomfeldt@qt.io>
Different font was used when running on QEMU ARMv7 and the second page
was never reached.
Pick-to: 6.1
Task-number: QTQAINFRA-4127
Change-Id: Ic85b76661cf3642b69e6e1b21e8062d7c36231e3
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
As described in 3279c8e7d7, we can't
depend on the widths of decoration to always be the same for
drawText() and equivalent drawStaticText/drawGlyphRun.
This is typically visible as an off-by-one when using the default
fonts on OpenSUSE.
Since this test was actually made to test positions of RTL glyphs,
we can just disable the decoration and simplify the text.
Fixes: QTBUG-89086
Change-Id: I139fe3e1c5e98d8b1d7e0e7c19645fd4717d1d95
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
This was added by 9ff76c27b9 on
the basis that it signifies a shaping error and would later assert
or crash.
But the line is easily reachable by user code. If Harfbuzz returns
0 glyphs, it just means it is unable to shape the string, for instance
if the input string only contains default ignorables (like a ZWJ)
and does not have any appropriate glyph to use for replacement.
Qt expects there to always be at least one glyph in the output
(num_glyphs == 0 is used to indicate shaping is not yet done), so
to avoid asserts later on, we simply populate the output with a
single 0 token, which is a required entry in the font that is
reserved for representing unrepresentable characters.
This also adds a test and therefore a zero-width joiner to the test
font to reproduce the issue.
[ChangeLog][QtGui][Text] Fixed a possible crash with certain fonts
when shaping strings consisting only of control characters.
Fixes: QTBUG-89155
Pick-to: 5.15
Pick-to: 6.0
Change-Id: Ia0dd6a04844c9be90dcab6c464bebe339a3dab11
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
With the updated QEMU and toolchain, these tests now pass.
Change-Id: Icb74562a0e6422cd4564f63db991aa431e0e3119
Reviewed-by: Assam Boudjelthia <assam.boudjelthia@qt.io>
Make use of the device pixel ratio in the QIcon paint method so the @nx
hi-dpi pixmaps are selected when appropriate when painting to a
QPainter.
Pick-to: 6.0
Fixes: QTBUG-90042
Change-Id: I53995a2285ef879e3c4fddb9f8da702e256a260f
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
Initialize the offscreen platform plugin with no screens,
create QGuiApplication object.
Not much of the high-dpi related Qt API can be used
in this configuration, but at least Qt should not crash
on startup.
Task-number: QTBUG-71034
Change-Id: I6620843c3bd8b692c5c2419b1ba290e16175ba5b
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Add QRhi APIs to retrieve and reload the contents of the "pipeline
cache".
The only API where there is a true pipeline cache is object is Vulkan
(VkPipelineCache). For OpenGL, the other backend where we support this,
it is simulated with program binaries. The Qt 5 style OpenGL program
binary disk cache continues to work like before, but one has now the
option to do things in a more modern, graphics API agnostic way, that
leads to generating a single blob instead of a large set of files in
some system location, allowing easier "pre-baking" of the cache content.
It is expected that Qt Quick exposes the two new functions in form
if QSG_RHI_ environment variables, thus allowing easy testing and
cache file generation.
As an example for the performance improvements this can give, consider
Vulkan, where we do not have any existing persistent caching mechanism
in place:
Running BenchmarkDemoQt6.exe --scene flythrough --mode demo creates 18
QRhiGraphicsPipeline objects from Qt Quick and Qt Quick 3D.
The total time spent in QRhiGraphicsPipeline::create() during application
startup for these 18 pipelines is 35-40 ms on a given Windows (NVIDIA)
system.
When exporting the pipeline cache contents to a file, and then, in a
subsequent run, reloading the cache contents, this is reduced to 5-7 ms
on the same system, meaning we get a 6-7x improvement.
The generated data is always specific to a given Qt version, RHI
backend, graphics device, and driver version. Much of the implementation
consists of adding and verifying the appropriate header to the blobs
retrieved from the driver, to allow gracefully ignoring data that was
generated with a device or driver that differs from the one used at
run time. This should provide robustness, even if the Vulkan or OpenGL
implementation is for some reason not prepared to identity and reject
incompatible cache/program blobs.
Fixes: QTBUG-90398
Change-Id: I67b197f393562434f372c7b7377f638abab85cb3
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
...when the platform does not support Vulkan. The version is left
at the default 0 then, so checking for >= 1.0 is wrong.
This allows the test to pass with the offscreen platform plugin.
Change-Id: I5afba8f1e703e4fa0ff41da91d18f5fabfb54868
Reviewed-by: Andreas Buhr <andreas.buhr@qt.io>
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
This also needs improvements to qvkgen. What we get with this patch
are the Vulkan 1.1 and 1.2 core API's additional 11 instance-level
and 30 device-level commands present in QVulkanFunctions and
QVulkanDeviceFunctions.
All of these are attempted to be resolved upon construction. When the
implementation does not return a valid function pointer for some of them
(e.g. because it is a Vulkan 1.0 instance or physical device), calling
the corresponding wrapper functions will lead to unspecified behavior.
This is in line with how QOpenGLExtraFunctions works. The simple
autotest added to exercise some Vulkan 1.1 APIs demonstrates this in
action.
The member functions in the generated qvulkan(device)functions header
and source files are ifdefed by VK_VERSION_1_{0,1,2}. This is essential
because otherwise a Qt build made on a system with Vulkan 1.2
headers would cause compilation breaks in application build environments
with Vulkan 1.0/1.1 headers when including qvulkanfunctions.h (due to
missing the 1.1/1.2 types and constants, some of which are used in the
function prototypes). In practice this should be alright - the only
caveat to keep in mind is that the Qt builds meant to be distributed
to a wide variety of systems need to be made with a sufficiently new
version of the Vulkan headers installed, just to ensure that the
1.1 and 1.2 wrapper functions are compiled into the Qt libraries.
Task-number: QTBUG-90219
Change-Id: I48360a8a2e915d2709fe82993f65e99b2ccd5d53
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Instead, have a static function in QRhiVulkanInitParams then Qt Quick
and anyone else who creates a QVulkanInstance that is then used in
combination with QRhi can query.
Change-Id: I046e0d84541fc00f5487a7527c97be262221527f
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
...as described in the Vulkan >= 1.1 spec. One can now call
supportedApiVersion() (before create(), similarly to the other
supported* functions) to determine the available Vulkan
(instance-level) version.
Fixes: QTBUG-90333
Change-Id: Ibe8482402b7f07e4abc48c88252ff0365e4e2faa
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
QGuiApplicationPrivate::processMouseEvent() sends a
QWindowSystemInterfacePrivate::TouchEvent if the mouse event is not
accepted and AA_SynthesizeTouchForUnhandledMouseEvents is enabled.
A QPA TouchEvent always contains native touch points, which is why
it calls QWindowSystemInterfacePrivate::fromNativeTouchPoints to
translate the QMouseEvent's device-independent position back to the
raw position that it would have had if it came from a real touchscreen.
Therefore we must give that function touchpoints that are actually in
native coordinates.
It may be that some of this transformation could be avoided entirely,
but here we prove that the existing way works correctly, by adding
coordinate checking to the tst_QWindow::mouseToTouchTranslation() test.
Pick-to: 6.0
Task-number: QTBUG-86165
Change-Id: I4c9ca2b11e9eb76d79712c187db3eb9865da581a
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Android and Wasm do not build the offscreen platform
plugin (see src/plugins/platforms/CMakeLists.txt).
Skip building the tst_qhighdpi test as well in this
case. Remove the BLACKLIST entry.
Task-number: QTBUG-88505
Change-Id: I172198c8c24759b14f73ad07260c449fc6ab893f
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
From QRhi's perspective this consists of two things:
- A shader with samplerExternalOES in it cannot go through the standard
pipeline. Rather, a QShader with suitable GLSL code in it has to be
constructed manually. As this is something useful as an autotest
anyway, add a test case to the qshader autotest that demonstrates
this.
- When it comes to correctly calling glBindTexture, add a QRhiTexture
flag. The expectation is that an OpenGL-only client sets this in
combination with QRhiTexture::createFrom(), thus wrapping an existing
texture that then gets bound to the GL_TEXTURE_EXTERNAL_OES target
instead of our usual GL_TEXTURE_2D.
For completeness we also add a SamplerExternalOES variable type to
QShaderDescription, but the sampler type is not actually used by the
QRhi OpenGL backend, as it is the QRhiTexture that defines the
texture target.
Change-Id: I36b52325deb3703b59186ee3d726d0c3015bfc4b
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
...to the extent it is sensible. We have to make compromises still,
meaning some fields will only be applicable with certain APIs.
Most of this is already shown upon QRhi::create() as info debug
prints, when enabled. Now expose it all through the QRhi API as
well.
This is useful for printing in qtdiag, and, while it should be
avoided as much as possible, to make decisions about disabling
3D rendering features depending on the driver and GPU in use.
Change-Id: Iebe1e192965c928b82a094d1c7c50ddf4b38b9a2
Reviewed-by: Andy Nichols <andy.nichols@qt.io>
Those serve no purpose anymore, now that the .pro files are gone.
Task-number: QTBUG-88742
Change-Id: I39943327b8c9871785b58e9973e4e7602371793e
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
QTextDocument and the text editor classes suggest to override
their loadResource() methods to provide data associated with
a text document. This approach has the following drawbacks:
- it requires subclassing
- there is no way to set a global resource provider
- QLabel is missing virtual loadResource() method and
it can't be added without breaking ABI
QUrlResourceProvider is designed to solve these issues.
One should create a derived class that implements
QUrlResourceProvider::resource(). The objects of the derived
class then can be set for any text document.
The default resource provider can be set with
QUrlResourceProvider::setDefaultProvider().
This change also adds QLabel::setResourceProvider(),
which doesn't break ABI.
[ChangeLog][QtGui][Text] Introduced QUrlResourceProvider that allows to
load resources for HTML. It is intended to replace the use of
QTextDocument::loadResource().
Change-Id: Iaf19b229f522a73508f20715257450fe58f68daf
Reviewed-by: Konstantin Ritt <ritt.ks@gmail.com>
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
With the introduction of QKeyCombination, the result of |'ing together
a set of modifiers and a key goes always through the same QKeySequence
constructor, no matter the order.
The implicit conversion through int when the wrong order is used
causes a compiler warning as that conversion is deprecated. So remove
that test case.
Pick-to: 6.0
Change-Id: I04e27bcd51723ee0efc77e52e45ca3eb8bac5fc7
Reviewed-by: Mitch Curtis <mitch.curtis@qt.io>
Most tests fail when running on QEMU ARMv7, but not on target HW or
QEMU ARM64.
Task-number: QTBUG-89819
Change-Id: I686268c200cce1a44a717b80c2970f608be44636
Reviewed-by: Shawn Rutledge <shawn.rutledge@qt.io>
Remove the qmake project files for most of Qt.
Leave the qmake project files for examples, because we still test those
in the CI to ensure qmake does not regress.
Also leave the qmake project files for utils and other minor parts that
lack CMake project files.
Task-number: QTBUG-88742
Change-Id: I6cdf059e6204816f617f9624f3ea9822703f73cc
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
Blacklist qfont tests that don't work correctly with qemu.
Task-number: QTQAINFRA-3836
Change-Id: I7f5d4b07428febb0adb82d38203226b5c56a2ad5
Reviewed-by: Tony Sarajärvi <tony.sarajarvi@qt.io>
Complete search and replace of QtTest and QtTest/QtTest with QTest, as
QtTest includes the whole module. Replace all such instances with
correct header includes. See Jira task for more discussion.
Fixes: QTBUG-88831
Change-Id: I981cfae18a1cabcabcabee376016b086d9d01f44
Pick-to: 6.0
Reviewed-by: Volker Hilsheimer <volker.hilsheimer@qt.io>
Apparently some library definitions went overboard, link them directly.
Pick-to: 6.0
Change-Id: I009737f7e3edff5619241b700a627dc4e25e6018
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
This is the most basic way to represent custom transfer functions.
Change-Id: I529fb647ece82c03e85ef77b056d9daf13fe5a61
Reviewed-by: Morten Johan Sørvig <morten.sorvig@qt.io>
Reviewed-by: Eirik Aavitsland <eirik.aavitsland@qt.io>
QColors were not premultiplied before being set.
Pick-to: 6.0 5.15 5.12
Change-Id: Id3765b6932a72374ddfd788fae4bb628a4edf0b7
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
If the first font in the families list happens to have been
disambiguated because of duplicates, two things went wrong:
1. hasFamily() would return false for the font family, because
it does not disambiguate when checking for the family name and
only checks if the families list contains the exact string.
2. Adding aliases to the full disambiguated string is not supported,
only the family name.
The first issue has been reported separately as QTBUG-89068.
The test failure is fixed by just avoiding the fonts that
are ambiguous in the test, as it really doesn't matter which
font we pick.
Fixes: QTBUG-89008
Pick-to: 5.15 6.0
Change-Id: I829778c2e7bb6090475c34dcf9cdce58862729d6
Reviewed-by: Liang Qi <liang.qi@qt.io>
This patch disables four failing unit tests when executed with the
offscreen backend.
Change-Id: Ie67341b886984e6de19cd8dd8a8a237a620a1b7a
Reviewed-by: Fabian Kosmale <fabian.kosmale@qt.io>
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
Reviewed-by: Laszlo Agocs <laszlo.agocs@qt.io>
