Use the same approach we use for iOS, which is to set multiple
CMAKE_OSX_ARCHITECTURES values and let the clang front end
deal with lipo-ing the final libraries.
For now, Qt can be configured to build universal macOS libraries by
passing 2 architectures to CMake, either via:
-DCMAKE_OSX_ARCHITECTURES="x86_64;arm64"
or
-DCMAKE_OSX_ARCHITECTURES="arm64;x86_64"
Currently we recommend specifying the intel x86_64 arch as the first
one, to get an intel slice configuration that is comparable to a
non-universal intel build.
Specifying the arm64 slice first could pessimize optimizations and
reduce the feature set for the intel slice due to the limitation
that we run configure tests only once.
The first specified architecture is the one used to do all the
configure tests.
It 'mostly' defines the common feature set of both architecture
slices, with the excepion of some special handling for sse2 and
neon instructions.
In the future we might want to run at least the Qt architecture config
test for all specified architectures, so that we can extract all the
supported sub-arches and instruction sets in a reliable way.
For now, we use the same sse2 hack as for iOS simulator_and_device
builds, otherwise QtGui fails to link due to missing
qt_memfill32_sse2 and other symbols.
The hack is somewhat augmented to ensure that reconfiguration
still succeeds (same issue happened with iOS). Previously the sse2
feature condition was broken due to force setting the feature
to be ON. Now the condition also checks for a special
QT_FORCE_FEATURE_sse2 variable which we set internally.
Note that we shouldn't build for arm64e, because the binaries
get killed when running on AS with the following message:
kernel: exec_mach_imgact: not running binary built against
preview arm64e ABI.
Aslo, by default, we disable the arm64 slice for qt sql plugins,
mostly because the CI provisioned sql libraries that we depend on only
contain x86_64 slices, and trying to build the sql plugins for both
slices will fail with linker errors.
This behavior can be disabled for all targets marked by
qt_internal_force_macos_intel_arch, by setting the
QT_FORCE_MACOS_ALL_ARCHES CMake option to ON.
To disble it per-target one can set
QT_FORCE_MACOS_ALL_ARCHES_${target} to ON.
Task-number: QTBUG-85447
Change-Id: Iccb5dfcc1a21a8a8292bd3817df0ea46c3445f75
Reviewed-by: Tor Arne Vestbø <tor.arne.vestbo@qt.io>
A few configure defines get changed:
QMAKE_WASM_PTHREAD_POOL_SIZE is now QT_WASM_PTHREAD_POOL_SIZE
QMAKE_WASM_TOTAL_MEMORY is now QT_WASM_INITIAL_MEMORY
QMAKE_WASM_SOURCE_MAP_BASE is now QT_WASM_SOURCE_MAP_BASE
device-option EMSCRIPTEN_ASYNCIFY=1 is QT_EMSCRIPTEN_ASYNCIFY=1
To create source maps for debugging. use
device-option QT_WASM_SOURCE_MAP=1
Task-number: QTBUG-78647
Change-Id: If9f30cd7fb408c386d6d69b5f7b1beecf1ab44b5
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Whenever something went wrong with building the arch test or extracting
info from it, the failure messages lacked useful details.
It's especially hard to figure out what went wrong if the failure
happened in the CI.
Print a bunch of useful information in case of failure.
Change-Id: Iafe287f951880c1441e57924ae372d6a9029a9a7
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Reviewed-by: Alexey Edelev <alexey.edelev@qt.io>
I'll preface by saying, it's not clear to me why the behavior is
different.
Our qt_run_config_test_architecture function builds a target
executable, and then uses file(STRINGS) to try and parse out some
magic strings that we expect the executable to have (architecture,
abi, etc).
In qmake this was done by matching with a regular expression ending on
a \\0 null byte character.
In CMake, we do a string(FIND) and string(SUBSTRING until the end of
the line) on a *line* returned from file(STRINGS). Notably, I did not
find any regexp syntax provided by CMake to try and match a null byte
character.
Note the docs for file(STRINGS) implies to me that *lines* are
detected by looking for newline characters '\n'. The docs also
mention that binary data in the file is ignored. What's binary
data though?
If you open the executable with a hex editor, at least on macOS, the
strings we are interested in are indeed separated by null byte chars,
not newline chars.
On most platforms file(STRINGS) did end a line on the null byte
character.
Except, for some reason not when building Qt for the oss-fuzz project
with clang under Docker.
Calling message() on one of the lines prints a very long string with
null characters seemingly replaced with semicolons, and of course the
matched architecture string is wrong and fails configuration.
For *some reason*, if I add a "REGEX ==Qt=magic=Qt==" option to the
file(STRINGS) command, the captured output lines don't contain
semicolons even when building for oss-fuzz.
The extracted strings are then correct, and configuration succeeds.
Use the REGEX option workaround, with the aim to quickly fix Qt
building for oss-fuzz for 6.0.1 release.
Pick-to: 6.0
Fixes: QTBUG-89047
Change-Id: Iad11c1090c1187aadd39082f196050bf3290df95
Reviewed-by: Robert Loehning <robert.loehning@qt.io>
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
Add a new function that returns the minimum CMake version required to
build Qt. Pass that value to cmake_minimum_required() when building
qtbase and its standalone tests.
The minimum supported CMake version is read from qtbase/.cmake.conf
and its value should be updated when the need arises. It's the main
source of truth for all repos.
Provide a way to lower the minimum CMake version at configure time by
passing a value via QT_FORCE_MIN_CMAKE_VERSION.
This is not an officially supported way of building Qt. If the
specified version is lower than Qt's supported minimum, show a
warning.
Nevertheless the option is useful for testing how Qt builds with a
different minimum CMake version due to different policies being
enabled by default.
Issue warnings for CMake versions that are higher than the minimum
version but are known to cause issues when building Qt.
A counterpart change is needed in qt5 to ensure the minimum CMake
version is set at the proper time for top-level builds.
Ideally we would use the same 'check the CMake minimum version` code
in all our repositories, but that will cause lots of duplication because
we can't really find_package() the code and doing something like
include(../qtbase/foo.cmake) hardcodes assumptions about repo
locations.
So for now we don't bump the minimum version in child repo
cmake_minimum_required calls (qtsvg, qtdeclarative, etc).
Instead we record both the minimum supported version and the computed
minimum version (in case a different version was forced) in
QtBuildInternalsExtra.cmake.
Then we require qtbase's computed min version in
qt_build_repo_begin().
This won't set policies as cmake_minimum_required would, but at least
it propagates what minimum CMake version should be used for child
repos.
We might still have to bump the versions in child repos at some point.
Task-number: QTBUG-88086
Change-Id: Ida1c0d5d3e0fbb15d2aee9b68abab7a1648774b9
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
Sometimes the info we show in the configure summary might differ from
the information that CMake computes.
To more easily debug and fix such cases, print out various CMake info
variables like host and target info, compiler versions, sdk versions,
etc.
Change-Id: I37a11dfabe5369236af78684a09bd3cec3fdd398
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
To achieve this, we save the result of the subarch test in the cache
variable TEST_subarch_result and use this value as the right hand side
of the QT_CPU_FEATURES.xxx assignment. The Qt6HostInfo package now
sets the variable QT6_HOST_INFO_SUBARCHS which will be used for the
host_build scope in qmodule.pri when cross-building.
Change-Id: I2c25f205bfc0692aef0d6f43ff4e542d27e1b948
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
* Assume UTF-8 on all Unix like systems
* Export some functions to be able to compile QTextCodec once
moved to Qt5Compat.
Task-number: QTBUG-75665
Change-Id: I52ec47a848bc0ba72e9c7689668b1bcc5d736c29
Reviewed-by: Lars Knoll <lars.knoll@qt.io>
Teaches configurejson2cmake about summaries / reports, so things
like enabled features, configure sections, notes, etc.
Add relevant CMake API for adding summary sections and entries,
as well as configure reports. The commands record the passed data,
and the data is later evaluated when the summary needs to be printed.
This is needed, to ensure that all features are evaluated by the
time the summary is printed.
Some report and summary entries are not generated if they mention a
feature that is explicitly exclduded by configurejson2cmake's feature
mapping dictionary. This is to prevent CMake from failing at configure
time when trying to evaluate an unknown feature. We should re-enable
these in the future.
A few custom report types are skipped by configurejson2cmake (like
values of qmake CONFIG or buildParts).
These will have to be addressed a case-by-case basis if still needed.
Change-Id: I95d74ce34734d347681905f15a781f64b5bd5edc
Reviewed-by: Leander Beernaert <leander.beernaert@qt.io>
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
This allows doing debug_and_release builds with Ninja on all
platforms.
The "Ninja Multi-Config generator" is available starting with CMake
3.17.
Desired configurations can be set via CMAKE_CONFIGURATION_TYPES.
Possible values: "Release, Debug, RelWithDebInfo, MinRelSize".
For example -DCMAKE_CONFIGURATION_TYPES="Release;Debug".
The first configuration is the 'default' configuration which is
built when calling ninja with no arguments.
To build all targets of a certain configuration use "ninja all:Release"
or "ninja all:Debug".
To build all targets in all configurations use "ninja all:all".
Note that the first configuration influences which configuration of
tools will be used when building the libraries for all configurations.
In simple terms, when configured with
-DCMAKE_CONFIGURATION_TYPES="Release;Debug" the release version of moc
is used by AUTOMOC.
When configured with -DCMAKE_CONFIGURATION_TYPES="Debug;Release"
the debug version of moc is used by AUTOMOC.
Framework builds and Ninja Multi-Config don't currently work together
due to multiple bugs in CMake, which ends up generating an invalid ninja
file with duplicate rules. There are also issues with placement of the
debug artifacts.
This will be handled in a follow up patch after CMake is fixed.
Task-number: QTBUG-76899
Change-Id: If224adc0b71b7d1d6606738101536146aa866cd7
Reviewed-by: Qt CMake Build Bot
Reviewed-by: Leander Beernaert <leander.beernaert@qt.io>
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
Tested locally with the following configurations:
- iOS device builds (arm64)
- iOS simulator builds (x86_64)
- iOS simulator_and_device builds (fat arm64 and x86_64 archives)
All iOS builds currently require a custom vcpkg fork which contains
fixes for building the required 3rd party libraries.
qtsvg, qtdeclarative, qtgraphicaleffects and qtquickcontrols2
have also been tested to build successfully.
simulator_and_device builds are also supported, but require an umerged
patch in upstream CMake as well as further patches to vcpkg.
Task-number: QTBUG-75576
Change-Id: Icd29913fbbd52a60e07ea5253fd9c7af7f8ce44c
Reviewed-by: Cristian Adam <cristian.adam@qt.io>
Reviewed-by: Qt CMake Build Bot
Reviewed-by: Leander Beernaert <leander.beernaert@qt.io>
Rename internal APIs like extend_target to qt_extend_target.
Prefix apis with qt_ where required.
Keep old names for compatibility until all their usages are removed.
Change-Id: I9a13515a01857257a4c5be3a89253749d46a4f41
Reviewed-by: Qt CMake Build Bot
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Every time the qtbase project was reconfigured, the list of
config tests results was appended to a cache variable which was
never reset, resulting in multiple copies of the same values
being written to QtBuildInternalsExtra.cmake.
Make sure to clean the cache variable before appending to it.
Also change the variable to be all upper-case, to be consistent
with other cache variables.
Change-Id: Ic12046730a080595e19377981a726bc330641dc1
Reviewed-by: Qt CMake Build Bot
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Replace condition x86 to i386 to match other platforms.
Regenerate src/gui/CMakeLists.txt
Disable SSE4 on android x86 to match qmake.
Change-Id: Ic0d330206f2d70a79d72553aa3ff0f91ff58119c
Reviewed-by: Qt CMake Build Bot
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
This is a simple mis-understanding between try_compile assuming interest
in the launching part of the thing we try to compile and our
architecture test being interested in the binary.
Change-Id: Ie972b662b6f34699f566649bb2c1e29da35b58c3
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
Export the architecture configuration variables from QtBase. There are
other modules that require access to this information in order to
enable certain features (e.g: qml_jit in QtDeclarative).
Change-Id: If2c7f29ccb1c0b0a0db3d78ad133a2a6be12b5ad
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The run_config_test_architecture() function in QtBaseConfig was not
working correctly as we were not updating the length of the line
properly.
Change-Id: I29c5b45f1fa36d16cb2ef00950feb80987f6b176
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
We record extra CPU features such as neon, etc. in the sub-architecture
field, which may very well be empty. Allow for that and avoid cmake
warnings about indexed string access beyond the string boundaries.
Change-Id: I63e61c6427d156180039b8ac6f5b0f5f55c36ee8
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
The test erroneously succeeds on macOS with an XCode generator.
Explicitly set the test result to be turned off.
Change-Id: Id6fcb96f420f611517e81cc3697f1c88b508bd7c
Reviewed-by: Tobias Hunger <tobias.hunger@qt.io>
In qmake there are at least 2 things to know regarding
sub-architectures and instruction sets.
Which instruction sets does the compiler know to compile for,
represented by the various config.tests and features in
qtbase/configure.json.
And which instructions sets are enabled by the compiler by default,
represented by the configure.json "architecture" test and accessed
via QT_CPU_FEATURES.$$arch qmake argument.
Before this patch there was some mishandling of the above concepts
in CMake code.
The former can now be checked in CMake with via TEST_subarch_foo and
QT_FEATURE_foo (where foo is sse2, etc).
The latter can now be checked by
TEST_arch_${TEST_architecture_arch}_subarch_foo
(where foo is sse2, etc and the main arch is dynamyicall evaluated).
The configurejson2cmake script was adjusted to take care of the above
changes, and the cmake files were regenerated as well.
Change-Id: Ifbf558242e320cafae50da388eee56fa5de2a50c
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>
Improve conditions/enable/disable conditions for iconv related features.
These are detected a bit different from what qmake does, so adapt to that.
Change-Id: I7b3e4baf05dc324507f370d7f651a62f29e42a98
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@qt.io>
