The strongswan configure.in uses the AC_LIB_PREFIX macro, which adds
--with-lib-prefix and --without-lib-prefix options, and which, by
default assumes that adding ${prefix}/lib to LDFLAGS and
${prefix}/include to CPPFLAGS is a good idea. Obviously, when
cross-compiling, it is definitely not a good idea.
In the specific case of strongswan, the result is that when testing if
the backtrace() function was available, the small C program was being
built and linked with -L/usr/lib. So when the host architecture and
target architecture are identical, it may find the backtrace()
function in the host C library, without looking at the target C
library:
configure:16457: /home/test/outputs/a637f916962b6136dd6dd4f4b9ff4e1cab568ef3/output/host/usr/bin/x86_64-unknown-linux-uclibc-gcc -o conftest -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -pipe -Os -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -L/usr/lib conftest.c >&5
/home/test/outputs/a637f916962b6136dd6dd4f4b9ff4e1cab568ef3/output/host/opt/ext-toolchain/bin/../lib/gcc/x86_64-unknown-linux-uclibc/4.6.3/../../../../x86_64-unknown-linux-uclibc/bin/ld: warning: libc.so.0, needed by /home/test/outputs/a637f916962b6136dd6dd4f4b9ff4e1cab568ef3/output/host/opt/ext-toolchain/bin/../lib/gcc/x86_64-unknown-linux-uclibc/4.6.3/../../../../x86_64-unknown-linux-uclibc/lib/../lib64/libgcc_s.so, may conflict with libc.so.6
Passing --without-lib-prefix prevents this default behavior from
happening, and no stupid value is added to CPPFLAGS or LDFLAGS.
Fixes:
http://autobuild.buildroot.net/results/a637f916962b6136dd6dd4f4b9ff4e1cab568ef3/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The MesaLib 7.10.3 tarball has been moved on ftp.freedesktop.org from
pub/mesa/7.10.3/ to pub/mesa/old_versions/7.x/7.10.3/.
A bump to a newer release is only coming after Buildroot 2013.08, so for now
let us just change URL to let the download work properly.
Reported-by: Ryan Wilkins <ryan@deadfrog.net>
Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Some parts of redis fail to build when ccache is enabled, due to one of our
own patches. The construct
make CC=$(CC) target1 target2
would result in
make CC=ccache <tuple>-gcc target1 target2
and here <tuple>-gcc would be treated as a target and the compiler used
would be just 'ccache'.
Reported-by: Johan Sagaert <sagaert.johan@skynet.be>
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Since 5628776c4a
"toolchain-external: fix lib64 symlinks"
lib64 is a symlink to lib so there is no reason to copy the libraries
to lib64 anymore. This fixes multiple building problems
for buildroot toolchains:
[...]
mv: 'output/host/usr/mips64-buildroot-linux-uclibc/sysroot/lib64/libuClibc-0.9.33.2.so'
and 'output/host/usr/mips64-buildroot-linux-uclibc/sysroot/lib/libuClibc-0.9.33.2.so'
are the same file
mv: 'output/host/usr/mips64-buildroot-linux-uclibc/sysroot/lib64/libutil-0.9.33.2.so'
and 'output/host/usr/mips64-buildroot-linux-uclibc/sysroot/lib/libutil-0.9.33.2.so'
are the same file
mv: 'output/host/usr/mips64-buildroot-linux-uclibc/sysroot/lib64/libutil.so.0'
and 'output/host/usr/mips64-buildroot-linux-uclibc/sysroot/lib/libutil.so.0'
are the same file
rmdir: failed to remove
'output/host/usr/mips64-buildroot-linux-uclibc/sysroot/lib64': Not a directory
[...]
Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
minidlna will detect and use the libiconv library if
the package is selected. In that case, we need to link
against -liconv.
Fixes the following linking problem:
tagutils/tagutils.o: In function `do_iconv.constprop.6':
tagutils.c:(.text+0x4d5c): undefined reference to `libiconv_open'
tagutils.c:(.text+0x4da0): undefined reference to `libiconv'
tagutils.c:(.text+0x4e08): undefined reference to `libiconv_close'
[Peter: wrap COMMON_LIBS line]
Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The NEON intrinsics used by Pulseaudio are only available in either
-mfloat-abi=softfp or -mfloat-abi=hard, not in
-mfloat-abi=soft. Therefore having NEON support in the processor is
not sufficient, we also should *not* be using soft-float.
Moreover, looking at BR2_ARM_ENABLE_NEON to know if the processor has
NEON support is incorrect. This option is only here to allow the user
to tell whether the processor has NEON support or not, for the ARM
cores that only have optional NEON support. Instead, the
BR2_ARM_CPU_HAS_NEON option really indicates whether NEON is available
or not (it is either set automatically by the ARM cores that always
have NEON support, or when BR2_ARM_ENABLE_NEON is enabled by the user
to confirm that his ARM processor has NEON support).
This fixes build failures such as
http://autobuild.buildroot.org/results/675/675c96059be348b594cc92980bef391126931c83/build-end.log
[Peter: add comment about reason for soft abi check]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The current code to set ABI on ARM does the following:
ifeq ($(BR2_arm)$(BR2_armeb),y)
... set ABI without 'hf' suffix ...
else ifeq ($(BR2_ARM_EABIHF),y)
... set ABI with 'hf' suffix ...
endif
But since $(BR2_arm)$(BR2_armeb) will always be 'y' in the cases where
BR2_ARM_EABIHF is 'y', it means that the 'else' part of the condition
will never be used.
Fix this by appending 'hf' to the ABI variable when BR2_ARM_EABIHF is
selected.
[Peter: put EABIhf handling under arm/armeb conditional for consistency]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Several of the lines in S50sshd script have a strange mix of spaces
and tabs, that at least do not look consistent with neighboring lines.
This patch makes the spacing consistent, and also strips the trailing
spaces.
Signed-off-by: Danomi Manchego <danomimanchego123@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The $(TARGET_DIR) variable is required when building
python-setuptools for the target otherwise the build system detects
the host installation which leads to permission error problems
like these:
Setuptools installation detected at /usr/lib64/python2.7/site-packages
Renaming /usr/lib64/python2.7/site-packages/setuptools-0.9.8-py2.7.egg-info to
/usr/lib64/python2.7/site-packages/setuptools-0.9.8-py2.7.egg-info.OLD.1377005697.88
OSError: [Errno 13] Permission denied
Moreover, remove the PYTHONPATH variable for host variant since it's
not needed.
Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Tested-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Just like 'attr', 'acl' doesn't use automake to control the
build/installation of its components, and the static-only installation
process was not installing libacl.a. We add a patch that fixes this.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
When attr is built static-only, it forgets to install its libattr.a
file, which leads to the build failure of packages such as 'acl' that
rely on attr.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
The Qt5 upstream URL has changed, which leads to build failures in the
autobuilders.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
See Source/WTF/wtf/Platform.h. Webkit is only supported for
arm, armeb, i386, mips, mipsel, powerpc, sh, sparc, x86_64.
[Thomas P: propagate dependency to the midori package.]
Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
