Age | Commit message (Collapse) | Author |
|
Prepare for the scarthgap release.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
6badeda ("mingw32: Add WINDRES export for SDK") attempted to fix the GCC
13 Canadian cross compile for MinGW host, but used the broad sdkming32
override, which made it apply to all target recipes. This caused build
errors in other recipes. Tighten the scope of the variables to only
apply when doing a Canadian cross compile.
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
Amend 6c54d16058ed8fb911c44df93b5732ae693b9803 and add WINDMC
to be ignored from hash, otherwise it contaminates sstate cache
for every recipe.
Signed-off-by: Samuli Piippo <samuli.piippo@qt.io>
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
GCC-13 has added code where windres is called on a rc file which
includes winuser.h, this requires to call windres with -I option to
point to right header directory duting gcc-candian-cross builds
Fixes
../../../../../../../work-shared/gcc-13.1.0-r0/gcc-13.1.0/gcc/config/i386/utf8-mingw32.rc:1:10: fatal error: winuser.h: No such file or directory
| 1 | #include <winuser.h>
| | ^~~~~~~~~~~
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
Adjust a comment.
Signed-off-by: Joe Slater <joe.slater@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This is needed by qemu 8.0.0
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Use the new addpylib functionality to ensure our library code is seen by
bitbake.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The changes to split classes into global vs. image specific contexts
has broken the inclusion of `wine` and `wineserver` host tools for
testing MinGW SDKs. This is because testsdk is an image specific class
and therefore it's inclusion is not detected globally and the wine host
tools are not present so the SDK tests fail.
Resolve this by using HOSTTOOLS_NONFATAL which will include the tools if
they exist, but won't fail if they are not present. This does mean that
users will now not know they need wine "up front" when doing a build,
but it will instead fail later when they actually try to test the SDK,
but there isn't really a better way to fix this.
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Add a dynamic BBFILES pattern so that patches for openembedded-layer
are conditionally applied only if meta-oe is present.
Signed-off-by: Sinan Kaya <okaya@kernel.org>
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This converts the layer to the new override syntax. This was done using
<oe-core>/scripts/contrib/convert-overrides.py <this-layer>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Disables debuginfod when using MingGW. This feature brings in
unbuildable dependencies and can't be used.
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Certain recipes require the "glibc-utils" recipe to include Glibc utility tools.
However, the native version is incompatible with the Windows host, and thus these
recipes will cause the Windows SDK to fail to build, particularly the kernel
development recipes.
[YOCTO #14286]
Signed-off-by: amann <anmolbir.mann@windriver.com>
Signed-off-by: dreyna <david.reyna@windriver.com>
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Set SDK_VENDOR to '-w64', which makes the host triplet match what GCC
expect to find when using mingw32-w64. This enables features that are
not functional in the classic mingw32, but have been implemented in the
mingw32-w64.
Disable 32bit libs from the runtime component when compiling for 64bit,
which were enabled as a side effect of the GCC config change.
Signed-off-by: Samuli Piippo <samuli.piippo@qt.io>
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Adds the framework for testing SDKs that ties into the oeqa test
framework. This allows commands like:
$ bitbake -c testsdk ...
to be run for MinGW SDKs.
The test framework currently executes all tests under Wine in lieu of
having access to actual Windows machines.
[YOCTO #13020]
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
Use the winpthreads library available in the mingw64 and
change the thread model from win32 to posix to get access
to c++11 features.
Signed-off-by: Samuli Piippo <samuli.piippo@qt.io>
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
GCCPIE defaults to "--enable-default-pie" in security_flags.inc,
which breaks Windows binaries.
Signed-off-by: Samuli Piippo <samuli.piippo@qt.io>
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
|
|
Fix below problem:
MACHINE=intel-x86-64
SDKMACHINE = "i686-mingw32"
bitbake freetype
do_compile failed with below error:
x86_64-wrs-linux-libtool: compile: x86_64-wrs-linux-windres
--include-dir=work/corei7-64-wrs-linux/freetype/2.9.1-r0/recipe-sysroot/usr/include
/work/corei7-64-wrs-linux/freetype/2.9.1-r0/freetype-2.9.1/src/base/ftver.rc
-o work/corei7-64-wrs-linux/freetype/2.9.1-r0/build/.libs/ftver.o
work/corei7-64-wrs-linux/freetype/2.9.1-r0/build//x86_64-wrs-linux-libtool:
line 1752: x86_64-wrs-linux-windres: command not found
we don't want to use windres for target build, only export WINDRES/RC when build
SDK. SDK_OS = "mingw32", so use this override to fix it.
Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We need to set this so that the override for nativesdk builds using MinGW don't
think they're using glibc.
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
|
|
The MinGW runtime doesn't provide libintl, so set gettext as the preferred
provider for libintl.
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
|
|
Use explicit mingw32 override for
PREFERRED_PROVIDER_virtual/nativesdk-${SDK_PREFIX}libc-initial
to avoid multiple providers and consequently broken builds.
Allow building static libraries, disabled by default.
Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
WINDRES and RC values are exported when building for a mingw32 target. These
values, due to being exported, end up in the signature hash for everything in
the system. By adding these to the whitelist we prevent contamination and
allow the system to properly reuse existing -native and other sstate-cache.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We need to ensure that WINDRES searches the correct include directory
for header files. Now that a default sysroot is specified to an invalid
location, it highlights that we need to specify the location manually and
explictly.
autotools appears to use "RC" in preference to "WINDRES" so we need
to export that to ensure it doesn't just find the utility and use it
without an include directory.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
mingw builds use WINDRES so export this tool in the same manner as the other
tools for mingw builds.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Nowadays this is already set in OE-Core, in tcmode-default.inc.
Fixes:
WARNING: Variable key
PREFERRED_PROVIDER_virtual/${SDK_PREFIX}binutils-crosssdk
(binutils-crosssdk-${SDK_ARCH}) replaces original key
PREFERRED_PROVIDER_virtual/x86_64-oesdk-mingw32-binutils-crosssdk
(binutils-crosssdk-${SDK_ARCH}).
Signed-off-by: Jacob Kroon <jacob.kroon@mikrodidakt.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Adding the SDKTAROPTS can result in either a failure or an endless loop when
generating the SDK tarball. This is due to potential recursive symlinks
within the cross compiler directory structure.
It also means files from places /var/log on the build system can be pulled
into the sdk, dependning on the chosen rootfs of the target system this
is built against.
Disabling it is therefore the best opton, when post processing the tarball,
the symlinks can be dealt with accordingly but the current approach is just
dangerous.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
|