1. Cross-compiling qt4-embedded and qt4-x11-free applications for AArch64 ilp32
(both big and little endian) fails with below error,
(snip)
In file included from ../3rdparty/javascriptcore/JavaScriptCore/runtime/Collector.h:31:0,
from ../3rdparty/javascriptcore/JavaScriptCore/runtime/JSGlobalData.h:32,
from ../3rdparty/javascriptcore/JavaScriptCore/interpreter/CallFrame.h:26,
from ../3rdparty/javascriptcore/JavaScriptCore/runtime/ClassInfo.h:26,
from ../3rdparty/javascriptcore/JavaScriptCore/runtime/JSObject.h:27,
from ../3rdparty/javascriptcore/JavaScriptCore/API/JSCallbackConstructor.h:30,
from ../3rdparty/javascriptcore/JavaScriptCore/API/JSCallbackConstructor.cpp:27:
../3rdparty/javascriptcore/JavaScriptCore/wtf/StdLibExtras.h: In instantiation of 'TO QTWTF::bitwise_cast(FROM) [with TO = int; FROM = double]':
../3rdparty/javascriptcore/JavaScriptCore/runtime/JSImmediate.h:52:49: required from here
../3rdparty/javascriptcore/JavaScriptCore/wtf/StdLibExtras.h:63:1: error: size of array is negative
../3rdparty/javascriptcore/JavaScriptCore/wtf/StdLibExtras.h: In instantiation of 'TO QTWTF::bitwise_cast(FROM) [with TO = double; FROM = int]':
../3rdparty/javascriptcore/JavaScriptCore/runtime/JSImmediate.h:57:47: required from here
../3rdparty/javascriptcore/JavaScriptCore/wtf/StdLibExtras.h:63:1: error: size of array is negative
make[1]: *** [obj/release/JSCallbackConstructor.o] Error 1
-- CUT --
the error is because of JSVALUE64 is being defined while compiling
programs for AArch64 ilp32. This makes equal condition checking of
intptr_t and double datatypes as false, and results in -1 (negative)
as array size in bitwise_cast function.
The patch defines JSVALUE64, if the programs are being compiled for
AArch64 64bit NOT for ilp32.
2. qt4_arch.inc is modified to set system architecture as aarch64 if the
TARGET_ARCH is other than aarch64. Without this modification, for
TARGET_ARCH equal to aarch64_32, aarch64be, aarch64be_32; the system
architecture is set as 'generic' and results in below error.
configure log:
(snip)
64-bit AMD 80x86 (x86_64)
'-opensource' is unsupported, using 'generic'
'x86_64' is supported
System architecture: 'generic'
Host architecture: 'x86_64'
-- CUT --
build error:
(snip)
In file included from ../../include/QtCore/qatomic_generic.h:1:0,
from ../../include/QtCore/../../src/corelib/arch/qatomic_arch.h:64,
from ../../include/QtCore/qatomic_arch.h:1,
from ../../include/QtCore/../../src/corelib/thread/qbasicatomic.h:227,
from ../../include/QtCore/qbasicatomic.h:1,
from ../../include/QtCore/../../src/corelib/thread/qatomic.h:46,
from ../../include/QtCore/qatomic.h:1,
from ../../include/QtCore/../../src/corelib/tools/qbytearray.h:45,
from ../../include/QtCore/qbytearray.h:1,
from ../../include/QtCore/../../src/corelib/tools/qstring.h:46,
from ../../include/QtCore/qstring.h:1,
from ../../include/QtCore/../../src/corelib/kernel/qobject.h:48,
from ../../include/QtCore/qobject.h:1,
from ../../include/QtCore/../../src/corelib/kernel/qcoreapplication.h:45,
from ../../include/QtCore/qcoreapplication.h:1,
from qdbusintegrator.cpp:42:
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h: In instantiation of 'bool QBasicAtomicPointer<T>::testAndSetOrdered(T*, T*) [with T = const QString]':
qdbusintegrator.cpp:76:1: required from here
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h:197:88: error: invalid conversion from 'const void*' to 'void*' [-fpermissive]
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h:108:20: error: initializing argument 2 of 'bool QBasicAtomicPointer_testAndSetOrdered(void* volatile*, void*, void*)' [-fpermissive]
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h:197:88: error: invalid conversion from 'const void*' to 'void*' [-fpermissive]
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h:108:20: error: initializing argument 3 of 'bool QBasicAtomicPointer_testAndSetOrdered(void* volatile*, void*, void*)' [-fpermissive]
make[1]: *** [.obj/release-shared-emb-auto/qdbusintegrator.o] Error 1
make[1]: *** Waiting for unfinished jobs....
-- CUT --
So, select system architecture (via QT_ARCH) as aarch64 for
AArch64 and its other variants
(From OE-Core rev: 5f191d86b8d60624d729b3b147c1dff818326dbd)
Signed-off-by: Krishnanjanappa, Jagadeesh <jagadeesh.krishnanjanappa@caviumnetworks.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add aarch64 support for qt4 packages.
Most of the patches are derived from debian:
http://anonscm.debian.org/cgit/pkg-kde/qt/qt4-x11.git/tree/debian/patches
Add arch check to set right QT_ARCH.
Disable webkit temporarily for qt4 and qt4-embedded that they fail to build:
| runtime/JSValueInlineMethods.h: In constructor
'JSC::JSValue::JSValue(const JSC::JSCell*)':
| runtime/JSValueInlineMethods.h:242:78: error: cast from 'JSC::JSCell*'
to 'int32_t {aka int}' loses precision [-fpermissive]
It has been fix by Qt5, but didn't backport to Qt4. There is an still open
defect for it:
https://bugreports.qt-project.org/browse/QTBUG-35442
(From OE-Core rev: 886e0d43f2a543a13ac8e3f1cb570ecae90339e0)
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In qt4's do_configure operation, it will refer to some variables that
are derived from 'd', however these variable values may be not correct
in multilib case since the extraction of these variables happens before
the multilib handler.
The fix is to move these python style functions back to shell style.
This fixes [YOCTO #2355]
[RP: Fix whitepace]
(From OE-Core rev: 977ba301a1063a84b865ddf7367c35827fbffc86)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Using "1" with getVar is bad coding style and "True" is preferred.
This patch is a sed over the meta directory of the form:
sed \
-e 's:\(\.getVar([^,()]*, \)1 *):\1True):g' \
-e 's:\(\.getVarFlag([^,()]*, [^,()]*, \)1 *):\1True):g' \
-i `grep -ril getVar *`
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There is this discrepency in spelling. Lets fix it in
core. There are lot of layers using SITEINFO_ENDIANNESS
This was shielded since meta-oe had its own copy of
siteinfo class. But that class has now been deleted in
favor of oe-core
(From OE-Core rev: 54a54778fad39931ac7d43daaf37ce7c1946a29b)
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Differences from meta-openembedded version:
* SRC_URI and S now come from qt-${PV}.inc since these are version specific
* Source checksums are also now in qt-${PV}.inc
* Remove do_compile as this is handled in qt-${PV}.inc
* Move contents of do_install_append from qt-${PV}.inc to do_install in
qt4.inc as this is the same for 4.6.3 and 4.7.1 and will get in the way of
do_install_append in qt-embedded.inc.
* Don't enable PostgreSQL, MySQL or SQLite 2.x plugins as we don't currently
have recipes for these DBMSs in Poky. These can be re-enabled easily when
or if we do.
* Use INC_PR in qt4-x11-free_4.6.3.bb
* Don't always specify -embedded config option in qt4.inc
* Don't add qte.sh to SRC_URI in qt-4.6.3.inc (this is embedded-specific)
Differences from what we have currently in Poky (plus the above):
* Set DESCRIPTION based on embedded/X11
* Move out arch-specific settings to qt4-arch.inc
* Add qt4x11.bbclass which can be inherited by application recipes to select
the X11 version (this makes more sense once the embedded version is added).
* Update HOMEPAGE
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Krishnanjanappa, Jagadeesh
Kai Kang
Dongxiao Xu
Richard Purdie
Khem Raj
Paul Eggleton