Some packages installs libOpenCL without declaring
BR2_PACKAGE_PROVIDES_LIBOPENCL (e.g. imx-gpu-viv). ImageMagick will detect
the library and will require libtool. Since libtool is not in dependencies,
build might fail.
To prevent that situation, explicitly disable opencl support for target and host.
Signed-off-by: Julien Olivain <juju@cotds.org>
[Peter: drop unneeded ax_cv_check_cl_libcl=no]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When the fis and gadgetfs-test packages were removed, their respective
options in Config.in.legacy were placed before the "Legacy options
removed in 2020.02", while they should have been placed after. Let's
fix that.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Following commit 0dcb5513ef
("package/refpolicy: remove dependency on policycoreutils"), we have a
build failure on some configurations:
Makefile:571: *** libselinux is in the dependency chain of busybox that has added it to its _DEPENDENCIES variable without selecting it or depending on it from Config.in. Stop.
This is because refpolicy selects the busybox SELinux support when
Busybox is enabled, which it turns selects libselinux, but we no
longer pay attention to the libselinux dependencies while doing this.
Since it's quite weird to have refpolicy mess with Busybox SELinux
support, this commit changes the logic to have Busybox automatically
enable its SELinux support as soon as SELinux support is enabled,
while still allowing it to be disabled.
Fixes:
http://autobuild.buildroot.net/results/5d8fda7c488a03c14942d87467d501acd633d24a/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
If a policy is built that is newer than the kernel can support, the
libsepol will fail to load that policy.
Indeed, a user can manually select the policy version in the config
as-is. However, it is not a friendly solution. The best solution available
is to set a default policy version based off of the toolchain header kernel
version. While a user may have a toolchain that has older kernel headers than
the built kernel, it is still better than setting the default to the maximum
available version that SELinux can support.
The following defaults policy versions are as follows for the given toolchain
headers:
31 >= 4.13
30 >= 4.3
29 >= 3.14
28 >= 3.5
26 >= 2.6
default 25
Note: Version 27 was never released.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
The policy version has to be a number, as such, set the type to int.
Due to the type change, we can't any longer do the legacy handling of
re-using the refpolicy policy version.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Currently, a user sets a policy version via the refpolicy package.
Having the option here has a few disadvantages:
- The Refpolicy package is not technically needed to use SELinux.
- When building a modular policy, Refpolicy will ignore the version string
and build the highest version possible which will cause libsemanage to
possibly fail when loading the policy.
Specifying a manual policy version in /etc/selinux/semanage.conf
forces libsemanage to load a specific policy version, which fixes the
above issue. However, because refpolicy currently defines the policy
version, libsemanage does not have a way to determine the policy
version, as refpolicy is not a dependency of libsemanage.
To work around these limitations, move the policy version number
selection to libsepol, as a system using SELinux always requires this
library.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Remove gadgetfs-test as gadgetfs has been deprecated in favour of
functionfs.
Signed-off-by: Mark Corbin <mark@dibsco.co.uk>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Also add myself as DEVELOPER, as waf is currently orphan,
and I am the last one to fiddle with it.
Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
The CMakeLists.txt file uses the CMP0091 which is an MSVC runtime
library flag abstraction macro.
Because we are not building ninja for Microsoft, it is safe to remove
this macro and set the minimum version to 3.10.
Fixes:
http://autobuild.buildroot.net/results/992b34c5625ec733d8dce678aa7a7540c4768ca1
Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Other changes:
- Depend on host-python3, as python2 support was removed.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Only host-policycoreutils is needed to build refpolicy. Remove the uneeded
target package.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Remove fis as RedBoot hasn't been updated for over 10 years.
Signed-off-by: Mark Corbin <mark@dibsco.co.uk>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
[yann.morin.1998@free.fr: rebase libplist after bump]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
This was missed when LLVM and Clang were updated.
Signed-off-by: Michael Drake <michael.drake@codethink.co.uk>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Upstream changes include:
- Fix loading FreeBSD kernels with multiple PT_LOAD sections.
- Use autotools to configure and build kexec-lite
- Add support for kexec_file_load
The packaging is adjusted to account for the change in build systems.
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Other changes:
- Add --with-stdc++lib=dynamic to openjdk.mk or else openjdk will fail to
build because it defaults to looking for a static libstdc++ library.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Convert patch 0005-Fix-installation-of-class-headers.patch to git
format and re-number it to 0004-Fix-installation-of-class-headers.patch.
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
