Makefile's target "install-magickincarchHEADERS" and "magick-install-data-local"
install both the same file (magick-baseconfig.h) in the same time...
The problem can be reproduced with:
mkdir /tmp/bar
/usr/bin/install -c -m 644 foo /tmp/bar &
/usr/bin/install -c -m 644 foo /tmp/bar/foo
/usr/bin/install: cannot create regular file '/tmp/bar/foo' : File exists
So, remove one of them.
Fixes:
http://autobuild.buildroot.net/results/d34/d34077ce582866c50bbd90de10bbe593e39463f1/build-end.log
Signed-off-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
By design, building duma on Blackfin and other non-MMU platforms is
quite useless: it uses page mappings to detect buffer overrun and
use-after-free programming errors. Therefore, this commit simply marks
duma as not available on non-MMU platforms.
Fixes:
http://autobuild.buildroot.org/results/951/9519d7abb45c6a789874ec6bb84418c31a98916b/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
There will soon be new options to the graph-depends script, which we
can only sanely pass via environment variables.
Currently, we use such an environment variable to pass the maximum depth
of the dependency graph; the name of that variable is explicit that it
contains just the depth.
However, there has been so far no release of Buildroot which would make
use of that variable, so no user should have come to rely on it.
Rename that variable so it is less specific, and more generic, so it can
be used to pass more options to graph-depends.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
We need to tell the duma Makefile what OS we're gonna run on, or it
uses default values.
It turns out the default values are the same as for OS=linux (and so
this change should be a no-op), but it's just better to force it, and
it avoids a warning during the build.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Even though pulseaudio has some HAVE_FORK compile-time conditionals to
avoid using fork(), those parts of pulseaudio are not using it. Since
using pulseaudio on !MMU platforms is fairly unlikely, we simply
disable it when no MMU is available, and propagate this dependency to
the appropriate locations.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
fluxbox uses the fork() system call, not available on noMMU
platforms. It's also fairly unlikely to have to use a full-blown
X.org window manager on such platforms.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
php-imagick selects imagemagick, and imagemagick 'depends on
BR2_USE_MMU', but this dependency was not propagated to
php-imagick. This commit fixes that.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
zyre selects filemq, and filemq 'depends on BR2_USE_MMU', but this
dependency was not propagated to zyre. This commit fixes that.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
sane-backends currently doesn't build with the ADI external
toolchains, due to an internal compiler failure. However, using the
internal toolchain backend for Blackfin goes past this internal
compiler failure, but raises the problem that sane-backends uses
fork(). Sane-backends has a --enable-pthreads option that allows to
use threads instead of forking in some parts of sane, but there are
still some other areas of sane-backends that continue to use fork().
Since the probability of using sane-backends on a noMMU system is
fairly small, we'll just disable this package for now on such
architectures, until someone cares enough to investigate deeper.
Fixes:
http://autobuild.buildroot.org/results/136/136100c3a2d843b7915fdd83d23d8b301fee577e/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
elfutils depends on !BR2_avr32, but this dependency was not properly
propagated to the comment about toolchain options. This commit fixes
that.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
printf is POSIX-compliant, echo -e/n is not.
Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Acked-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
printf is POSIX-compliant, echo -e/n is not.
Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Acked-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
No functional change, but cleaner.
Remove the double --disable-manuals and move the debug handling together
with the other optional configure flags.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
ldconfig breaks links created by gpu-viv-bin-mx6q.
fixes
http://autobuild.buildroot.net/results/ab5/ab57dd3729d529366808cbfc802f49360b2ec2b9/
[Peter: use NETSNMP_INSTALL_*_OPT, only install initscript for sysv init]
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Tested-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Qt5Quick1 module doesn't need OpenGL libs and declarative module.
This module can be used to port qt4 qml apps to qt5. Patch has been tested
with linuxfb output plugin.
Signed-off-by: Wojciech Slenska <wojciech.slenska@gmail.com>
Reviewed-by: Fatih Aşıcı <fatih.asici@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This package fails to compile for MIPS64 n32 platforms. We are in
conversations with upstream to fix this problem. Meanwhile we disable
this package until we fix it.
Fixes:
http://autobuild.buildroot.net/results/0e3/0e3f1482d6f2f9bddc53d4e78b575120a2729e1d/
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The --with-fontdir switch has been substituted by the --with-fontrootdir
switch a few versions ago. The use of the old configuration switch was
causing error messages like the following:
_FontTransOpen: Unable to Parse address ${prefix}/share/fonts/X11/*/
[dix] Could not init font path element ${prefix}/share/fonts/X11/*/, removing from list!
Signed-off-by: Filippo Sironi <filippo.sironi@gmail.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
this patch fixes :
luajit: error while loading shared libraries: libluajit-5.1.so.2: cannot open shared object file: No such file or directory
Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Two fixes for xtensa are still applicable to gdb-7.5.1, which is now the
default; copy them from 7.4.1. This fixes native gdb build for xtensa.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
jack2 uses Waf 1.6.11. The waf configure program combines a part of
python script with a compiled blob. According to Waf documentation,
waf currently supports python interpreter from version 2.3 to 3.1.
However, Buildroot provides python-3.4. So, let's uses host python2
interpreter for building jack2.
This patch also cleanup whitespace, replacing wrongly used spaces with
tabs.
Fixed:
http://autobuild.buildroot.net/results/913/913bb1836076be9f201bf6641c7614708d6916bc/
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The autobuilders are not able to build glibc on AArch64 since we moved
to the internal toolchain backend as the default for AArch64. The
reason is that the cross-ld segfaults while linking of the glibc
utilities. Unfortunately, the issue has so far been impossible to
reproduce outside of the autobuilder CI loop itself: even on the same
machine, in the same chroot, the issue does not appear.
Today, Will Newton helped me analyze a bit further the problem. In the
logs of the build machine, we have found that the following segfault
notification matched the dates of the autobuilder failures:
/var/log/syslog:May 15 10:39:20 sd-56966 kernel: [767023.832598] ld[6846]: segfault at 1b2c002 ip 00000000004caf80 sp 00007fffaa308df8 error 4 in ld[400000+1b9000]
/var/log/syslog.2.gz:May 13 07:58:24 sd-56966 kernel: [585094.434264] ld[17761]: segfault at 154b002 ip 00000000004caf80 sp 00007fff0d400168 error 4 in ld[400000+1b9000]
/var/log/syslog.3.gz:May 12 20:24:08 sd-56966 kernel: [543558.227915] ld[14841]: segfault at 2706002 ip 00000000004caf80 sp 00007fffc08d41c8 error 4 in ld[400000+1b9000]
/var/log/syslog.4.gz:May 11 19:51:59 sd-56966 kernel: [455484.523033] ld[50162]: segfault at 1f00002 ip 00000000004caf80 sp 00007fff14a02328 error 4 in ld[400000+1b9000]
/var/log/syslog.4.gz:May 12 05:13:37 sd-56966 kernel: [489085.120089] ld[32067]: segfault at 2430002 ip 00000000004caf80 sp 00007fffc448db48 error 4 in ld[400000+1b9000]
/var/log/syslog.6.gz:May 9 15:20:38 sd-56966 kernel: [266948.197141] ld[13483]: segfault at ff0002 ip 00000000004caf80 sp 00007fff7e4fe948 error 4 in ld[400000+1b9000]
/var/log/syslog.7.gz:May 8 16:45:08 sd-56966 kernel: [185853.688607] ld[1757]: segfault at 20b1002 ip 00000000004caf80 sp 00007fffd07d5ae8 error 4 in ld[400000+1b9000]
See
http://autobuild.buildroot.org/?reason=glibc-2.18-svnr23787&arch=aarch64
for the autobuilder results.
So, the segfault always happens while accessing an address 2 bytes
after the beginning of a page: it looks like a buffer overflow, which
gets triggered because the object is placed towards the end of the
previous page, and crosses over the next, unallocated page.
Moreover, we looked at the 0x4caf80 address in an AArch64 cross-ld
built by Buildroot, and it turns out that it is part of the
iterative_hash() function. This nicely correlates with the patch that
Will Newton had pointed in
http://lists.busybox.net/pipermail/buildroot/2014-May/095881.html as a
fix for invalid memory accesses reported by Valgrind in the AArch64
cross-ld, specifically around the same area of 'ld'.
Therefore, even though we have not been able to reproduce the issue
outside of the autobuilders, and cannot confirm that the patch is
fixing the issue, I propose to integrate this binutils patch into
Buildroot. It is anyway an upstream binutils patch, which fixes a real
problem.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The lsof Configure script builds a test program using the host toolchain to
test whether glibc is used. This test is broken in cross compilation
environment. Set LINUX_CLIB to avoid the test. This should give the correct
result even for non glibc toolchains, since all libc variants we support have
the netinet/tcp.h header.
Fixes:
http://autobuild.buildroot.net/results/a1f/a1f0572dbf968c21f70b35cefff7ef7a1d9a348a/
Cc: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Tested-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since commit 6c5c08b854,
openpowerlink package is rebuilt at every make call because
.stamp_downloaded is missing in the build directory.[D
The culprit is OPENPOWERLINK_EXTRACT_CMDS that remove and
replace the build directory.
unzip extract a directory "openPOWERLINK-V1.08.4" and
Buildroot expect a directory "openpowerlink-V1.08.4"
It is easier to use git repository instead of zip archive and
avoids rename the directory openPOWERLINK to openpowerlink.
Signed-off-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The test program any2ppm can run as daemon. This feature can be disabled
at compile time, if the required headers are not present. However the
support for fork() is not checked.
Fixes http://autobuild.buildroot.net/results/4f2/4f271d55d7060d412aa336cfc2da3be3538d3594/.
Signed-off-by: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since libatomic_ops does not implement real atomic operations for
ARMv4 and ARMv5, libunwind must define AO_REQUIRE_CAS do indicate it
requires compare-and-swap operations, even if not available as real
atomic operations for the current architecture. In this case,
libatomic_ops will rely on emulated atomic operations, which also
require linking against libatomic_ops, which was until now not done by
libunwind.
This fixes the mysterious ltrace build issue:
http://autobuild.buildroot.org/results/e1b/e1b330abfa2d80f3f30bc3359428ea429c690eb8/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Even though flann build system uses CMake, it always tries to build a
shared library, even if BUILD_SHARED_LIBS=OFF is passed. It could
probably be fixed in flann's CMakeLists.txt, but that's not really
worth the effort if upstream doesn't support it.
Therefore, we simply disallow the flann package in purely static
builds.
Fixes:
http://autobuild.buildroot.org/results/a4e/a4ec0e9f28cd12ce770718fb407bbb4dc93b528b/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The "_" symbol prefix added by the blackfin toolchain causes "undefined
symbol" build failures when the compiler is trying to link a binary
with libblas and/or liblapack.
This patch also disables clapack reverse dependencies: armadillo.
Fixes:
http://autobuild.buildroot.net/results/a28df0ba10803e6e54c2f8160bbd8190cba4d690/
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Building crda on NIOS II triggers the infamous _gp linker issue. Since
nobody cared, let's disable this package on NIOS II for the moment.
Fixes:
http://autobuild.buildroot.net/results/272/272f18410e4855ab6b3bfd8a44243c1ebecf6abe/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
cppcms uses some thread functions that are not available in certain
uClibc thread implementations. Even though a bit more restrictive than
necessary, adding a NPTL dependency is the easiest solution, and is
quite logical for a relatively large and complex package such as
cppcms.
Fixes:
http://autobuild.buildroot.net/results/a26/a26574419aacbea4140dfca1d503bcab599edd71/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Building aiccu does not work on NIOS II due to the _gp problem, and
nobody is interested to fix it for now, so we simply disallow this
package. While we're at it, we propagate the AVR32 exclusion to the
comment.
Fixes:
http://autobuild.buildroot.net/results/786/7864cd5789a9f9b9f0603d738f4e63822d8469dd/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This fixes the following build error often seen during linux kernel
linking step with relaxation enabled:
dangerous relocation: call8: misaligned call target: (.text.unlikely+0x63)
Backported from: a35d5e823fdfe8a6e7e05ca8e3fb8bb5697335b1
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
SYSSTAT_MAKE_OPT is a string used to pass options to make, '+=' operator
doesn't have any special meaning inside it, so CFLAGS+=... is passed to
shell, overwriting previous CFLAGS value.
Replace CFLAGS+="..." with CFLAGS="$(TARGET_CFLAGS) ...".
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
EBTABLES_K64U32 is a string passed directly to the configure script,
'+=' operator doesn't have any special meaningi inside it, so
CFLAGS+=-DKERNEL_64_USERSPACE_32 is passed to shell, overwriting previous
CFLAGS value.
Replace CFLAGS+="-DKERNEL_64_USERSPACE_32" with -DKERNEL_64_USERSPACE_32
and pass CFLAGS="$(TARGET_CFLAGS) $(EBTABLES_K64U32)".
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
LIBCGROUP_CONF_ENV is a string used to pass environment variables to the
confgiure script, '+=' operator doesn't have any special meaning inside
it, so CFLAGS+=... is passed to shell, overwriting previous CFLAGS
value. Replace CFLAGS+="..." with CFLAGS="$(TARGET_CFLAGS) ...".
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
SQLCIPHER_CONF_ENV is a string used to pass environment variables to the
confgiure script, '+=' operator doesn't have any special meaning inside
it, so CFLAGS+=... is passed to shell, overwriting previous CFLAGS
value. Replace CFLAGS+="..." with CFLAGS="$(TARGET_CFLAGS) ...".
Fixes:
http://autobuild.buildroot.net/results/dbf/dbf947ad6442fa8e57201ffcc96871361bf39ad7/
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
To speedup building disable ccmake because it's not needed by buildroot.
Also disabling ccmake get rid of the ncurses optional unspecified
dependency for the sake of reproducible builds.
Signed-off-by: Fabio Porcedda <fabio.porcedda@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Jerzy Grzegorek <jerzy.grzegorek@trzebnica.net>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Luca Ceresoli <luca@lucaceresoli.net>
Cc: Samuel Martin <s.martin49@gmail.com>
Cc: Alexander Potashev <aspotashev@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The host-protobuf package can only build on certain architectures. For
example, host-protobuf does not build on PowerPC platforms, causing
build failures on the autobuilders. Since we don't really care about
build platforms other than x86 and x86-64, this commit restricts the
build of host-protobuf to these architectures only, even though if
technically it could build on ARM and MIPS.
Fixes:
http://autobuild.buildroot.org/results/9b9/9b9b20b0e4694b11425cdc2382650cdc2774e400/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The dependency is propagated to XBMC, even though XBMC currently
requires an (e)glibc toolchain, which always have dynamic library
support. This is just in case one day XBMC can be built against
uClibc.
Fixes:
http://autobuild.buildroot.org/results/3fa/3fae35ebcf14d6f38284d8318f61f9b99998794c/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The libmbus tarball is no longer available at
http://www.freescada.com/public-dist. Apparently, the new upstream is
at http://www.rscada.se/libmbus/ and tarballs are at
http://www.rscada.se/public-dist/. Without this patch, the current
download of libmbus-0.8.0.tar.gz downloads an HTML document with an
error, which cannot be extracted, obviously.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
radvd needs some C library definitions not available in old uClibc
versions. Since we don't care much about AVR32, we simply disallow the
radvd package on this architecture.
Fixes:
http://autobuild.buildroot.org/results/f03/f0367a30e8d772eb5e955a85b853ec3e43fe54b3/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
U-Boot v2014.04 introduced significant changes in its build infrastrcture.
Among things related to buildroot are:
1. Special new target ("tools-only") was added for building host tools alone
2. Tools cross-building for target is no longer possible with substitution of
HOSTxx with TARGETxx because host binaries won't be built. Instead we just set
CROSS_COMPILE and CROSS_BUILD_TOOLS variables.
3. Most of make targets now require U-Boot to be configured before building.
So for building generic "fw_printenv" we now need to make "env" target config-
independent.
4. HOSTCPPFLAGS are not used anywhere, so dropping
"uboot-tools-02-hostcflags-override-fix.patch"
Also due to lincese boilerplate change in sources
"uboot-tools-01-drop-configh-from-tools.patch" required subtle changes.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Cc: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The "make freetype-patch" fails because does not found autogen.sh,
because autogen.sh available only for the configure stage.
To fix this add host-automake as a prerequisite to freetype-patch.
This change also fix the support to top-level parallel make.
Signed-off-by: Fabio Porcedda <fabio.porcedda@gmail.com>
Cc: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
protobuf 2.5.0 adds an atomic operations implementation that is limited
to a few architectures only. mips64el and armeb don't work either.
Fixes (among others)
http://autobuild.buildroot.net/results/ae9fa612d0b6c9d593504206d5cedd480ad2547e/
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Allow stereo modes to be selected for output; enable NTSC output.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Undocumented changes (probably related to HDMI).
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Update the package to use patches that have been applied upstream and
update their status comment accordingly.
f_fsid checks have been fixed via upstream patch that allows the cache
to work properly on the result.
Builtin heimdal tools can be worked around via the --bundled-libraries
option to disable those components (even if they are not libraries it
can be done that way).
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
During the CMake bootstrap phase, the {C,LD}FLAGS set in the bootstrap
environment are not forwarded/converted as CMake flags.
The CMake build contains a bootstrap phase building a minimal CMake
program using a standard Makfile, then reconfigures itself with this
minimal program.
On system with no ncurses installed, and because the prefix option
points to $(HOST_DIR)/usr, if host-cmake was built after host-ncurses,
then ncurses libraries and headers are correctly found (in the host
tree) during the second configuration (because of the prefix). However,
it fails at building ccmake (the curses interface) because the
CMAKE_C_FLAGS, CMAKE_CXX_FLAGS and CMAKE_EXE_LINKER_FLAGS do not
point to the host tree.
Because these flags are needed when running the bootstrap script,
this patch makes sure the same flags are set when running the second
configuration.
Reported-by: Fabio Porcedda <fabio.porcedda@gmail.com>
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Cc: Fabio Porcedda <fabio.porcedda@gmail.com>
Cc: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The make_hash host tool, should be built during a PRE_BUILD_HOOKS and
not a POST_CONFIGURE_HOOKS.
Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Add PRE_*_HOOKS to all the different steps through which a package may go.
This will help avoid using POST_*_HOOKS to do tasks that should be done
in the PRE_*_HOOKS of the next step.
Otherwise, when the user would do a make foo-re<step>, this would not do
what was really intented, the POST_*_HOOK of the preceding step not
being executed.
Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
[ThomasDS: rebase, add images hooks to manual]
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
With the new XBMC, on RaspberryPi, libcec throws an asserts in RPi
userland's code which forbid XBMC Gotham to start.
Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The toolchain currently doesn't build for nommu ARM and is in need of
serious work.
Problem is there are no emulation targets and real ARM(7TDMI/720T/740T)
hardware that's capable of running linux (enough memory, having a
memory controller...) is VERY rare and uses very old versions to
make it usable.
The ARM nommu focus should go into Cortex M series processors that are
obtainable at reasonable cost on modern hardware that has external
memory controllers.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Forcibly disable mysql support when we don't want it since ulogd's
configure script can pick up the host mysql_config thus leading to a
build failure.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Detect when BR2_PREFER_STATIC_LIBS is selected and modify the host
pkg-config wrapper to append the --static option in that case.
Fixes:
http://autobuild.buildroot.net/results/161/161446dde7e8e774773eb2b34fd555f5ac22dd02/
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Tested-by: Baruch Siach <baruch@tkos.co.il>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Tested-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Features:
Supports parsing a request encoded by GET/POST method
Supports parsing multipart/form-data encoding.(in-memory and direct disk)
Supports COOKIE handling.
Supports Session management.
Supports FastCGI
[Peter: drop invalid patch, add patches for make install + configure paths]
Signed-off-by: Sagaert Johan <sagaert.johan@skynet.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
On PowerPC, clapack uses the _fpu_control symbol, which isn't defined
on uClibc, so this commit disallows the selection of clapack on
PowerPC uClibc configurations. PowerPC/glibc and ARM/uClibc
configurations have been tested to build correctly.
Fixes:
http://autobuild.buildroot.org/results/d1d/d1d721fa697956218dfc2c865dfb61911cf2600e/build-end.log
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
[Peter: license on single line, use PKG-INFO as license file]
Signed-off-by: Ryan Barnett <rjbarnet@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Correct license information after switch to correct package on commit
5aa4f71b51.
Signed-off-by: Ryan Barnett <rjbarnet@rockwellcollins.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
While the Blackfin external uClibc toolchains have <glob.h>, they
don't implement all the functions needed by popt, causing build
failures. One solution would have been to disable popt with such
toolchains, but this would have meant propagating this additional
dependency to the numerous reverse dependencies of popt. Instead, this
commit chooses to make popt build on Blackfin external toolchains by
disabling the usage of <glob.h> functions altogether.
Fixes:
http://autobuild.buildroot.org/results/bfc/bfcb2ed27debafc584e133f5ae11ad2061ad2b16/build-end.loghttp://autobuild.buildroot.org/results/410/410c8f3a0c7ca2c7098a47c30088038411635ae4/build-end.log
and gazillion of similar failures.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Disable this package for MIPS because it needs IRIX headers and
libraries.
Fixes:
http://autobuild.buildroot.net/results/370/370f964441722675820c644403dde7fbc683a315/
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Reviewed-by: Markos Chandras <Markos.Chandras@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
On the systems with old dblatex (for example 3.0) the build step will
fail, because dblatex doesn't support '-q' parameter. This patch
overrides dblatex check.
Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
A build issue affects libsanitizer on musl toolchains, even with
previous versions of gcc such as 4.8.x, so we disable building
libsanitizer when working with musl.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The musl C library does not yet implemented the getpwent_r() function,
needed by the Busybox password code when CONFIG_USE_BB_PWD_GRP is
disabled. So we enable it when the musl C library is used.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
For the moment, the musl support is not in mainline gcc, so it
requires a few patches. We have integrated those patches only for gcc
4.7 and gcc 4.8 at the moment, so only allow those gcc versions when
the musl library is selected.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Tarball download from github is preferred, because:
- download size is smaller
- git clone may be blocked in some corporate environments
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
kernel commit cf4cca10 (perf tools: Move libelf check config into config/Makefile)
removes the NO_LIBELF check from the top-level Makefile
for newer kernels, if config/Makefile exists, we can assume that NO_LIBELF is
also there
Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
Reviewed-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Tested-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This commit slightly improves the external toolchain backend, and the
gdb build logic to create a file named
$(STAGING_DIR)/usr/share/buildroot/gdbinit which can be used as a
gdbinit file using gdb -x option. This allows gdb to automatically use
the proper sysroot to find libraries.
The initial insight for this patch comes from the report of Oded
Hanson <OHanson@xsightsys.com>, who found an issue with the Eclipse
Buildroot plugin, which was setting a solib-path in gdb, but not a
sysroot. Setting a solib-path was enough to find shared libraries, but
not the dynamic linker. And since Eclipse doesn't allow to set the
sysroot in any other way than giving a gdbinit file, it makes sense to
have Buildroot generate a gdbinit file (which can be used in other
situations than Eclipse).
To achieve this, this commit introduces a gen_gdbinit_file helper in
toolchain/helpers.mk, and uses it for the internal toolchain and
external toolchain backends.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
[ThomasDS: minor updates in commit message]
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes:
http://autobuild.buildroot.net/results/6dd/6dd6954e5ab8c422070e53b0e616e1b1a821c626/
Also, remove the Blackfin toolchain dependencies as these are redundant now
since Blackfin is a noMMU arch.
[Peter: propagate to filemq as well]
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Prior to commit f64a1b850f ("cups:
convert to autotargets"), cups was explicitely configured to
put its modifiable single-machine data under /var. After that
commit, the --localstatedir was not specified, so it remained at
it's default value, "PREFIX/var" - making cups use "/usr/var".
Using /var to store state files seems preferable to /usr/var,
as /usr might be read-only on some systems. This patch
re-introduces the --localstatedir specification to build cups
to use /var.
Signed-off-by: Danomi Manchego <danomimanchego123@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Armadillo is a C++ linear algebra library.
[Peter: needs largefile, sort Config.in alphabetically]
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
As for the tzdata package, install the missing files:
- zone.tab : {country-code,coordinates,timezone} tuples
- iso3166.tab: {country-code,country} tuples
Reported-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This patch also call autoreconf in order to use buildroot's libtool as
buildroot's libtool patch doesn't apply to the upstream ltmain.sh.
[Peter: clarified the need for AUTORECONF]
Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This package provides BLAS and LAPACK libraries.
Though it is common to find implementation of these two libraries in
Fortran, this package provides a C-implementation for both, because:
- Fortran support has been deprecated in Buildroot since the 2013.11
release;
- most of the external toolchains do not provide a Fortran compiler.
Often BLAS build-systems build some test programs and run them to
generate some source files or adjust some build optimizations, naively
assuming they are building the library for the build-machine. This does
not play well when cross-compiling.
This implementation has this defect too, by building and running a tool
generating a header.
However, the build-system allows to pass an empty header.
So, we have to patch the CMake to build the generator (but never
install it) and correctly support building with and without this header
provided by the user.
Also, some CMake patches are needed to fix the build and install rules.
[Peter: needs largefile, fix _LICENSE_FILES, tweak patch desc]
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The OpenCV highgui module links against QtGui and QtTest.
The latter is built by default but was never installed.
Thanks to the previous patch on Qt, this can be rectified.
So, update the OpenCV Config.in with the missing Qt module selection.
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When building Qt in a minimal configuration, a couple of Qt modules
(QtNetwork, QtSql, QtTest and QtXml) are built in addition to QtCore,
so they are installed in the sysroot but they are not copied into the
target tree.
Among these modules that could be built but optionally installed, only
QtTest had no option to allow being installed in the target fs.
Thus, programs linking against QtTest would successfully build, but
would miserably fail at runtime due to this missing dependency.
So far, the only package triggering this issue is OpenCV (the highgui
module built with Qt support).
A followup patch will add this missing dependency in OpenCV using this
new option.
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
- update version to 1.31
- correct license to GPLv2+
- change download url to official repository
- remove evtest-capture support (dropped since 1.31)
- change broken project URL to official repository
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The flite configure script detects the availability of alsa-lib. Add alsa-lib
to flite dependencies to make sure alsa-lib is always detected when
available.
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
At runtime, libao will load all plugins, even ones that will never be used.
Since libao will prefer ALSA over OSS by default anyway, it is pretty safe
to exclude the OSS plugin.
There is no configure option for disabling the OSS plugin, so we remove
the shared object in a post-install hook.
Also disable alsa oss emulation workaround as it no longer will be used with
alsa.
[Peter: explain why --disable-broken-oss is passed]
Signed-Off-By: Maarten ter Huurne <maarten@treewalker.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
UPX is a free, portable, extendable, high-performance executable
packer for several executable formats.
This is needed for syslinux 6.x.
Note: upx is a host-only package. Although it supports quite a few
target architectures, the only use so far will be for use by syslinux,
which already has an x86-on-x86 requirement anyway. So we currently do
not care to have target-dependencies on host-upx.
Signed-off-by: Frank Hunleth <fhunleth@troodon-software.com>
[yann.morin.1998@free.fr: builds fine in parallel; fix build to
locate libucl; fix installation; don't add a menuconfig entry]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Tested-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
UCL implements a number of compression algorithms that achieve an
excellent compression ratio while allowing *very* fast decompression.
Decompression requires no additional memory.
This is needed for UPX, which is needed for syslinux 6.x.
Signed-off-by: Frank Hunleth <fhunleth@troodon-software.com>
[yann.morin.1998@free.fr: don't add a menuconfig entry, since it
currently is for internal use only]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Tested-by: Romain Naour <romain.naour@openwide.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
zeromp's configure script correctly detects the fork availability, but
unconditionally build test_fork program whatever the fork() availability.
This patch fixes the build-system by disabling test_fork when fork is
not available (e.g. when !BR2_USE_MMU).
Fixes:
http://autobuild.buildroot.org/results/359/3599cc3b7bf2bb22a78961cd84d21cb03cbd7015/
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
[yann.morin.1998@free.fr: remove Makefile.in hunks]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
[yann.morin.1998@free.fr: tested against the failed bfin config, as well as
a i686 target]
Tested-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This consists of two files:
- avahi-daemon.service:
It start/stop the daemon avahi, it's a copy of the S50avahi-daemon
script.
- avahi_tmpfiles.conf:
This one gets installed in /usr/lib/tmpfiles.d/ to be used by
systemd-tmpfiles which will create the correct directory structure
with the appropriate rights.
This is the equivalent of the S05avahi-setup.sh script.
Signed-off-by: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
gcc support was added in version 4.6:
http://gcc.gnu.org/gcc-4.6/changes.html
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since libwesockets in written in pure C, no C++, there is no reason to
disable it or make the configuration failed when the toolchain has no
C++ support.
Unfortunately, CMake defaults sets C and C++ as languages used by the
project. This makes the configuration failed if no C++ compiler is found.
This patch just make the CMake code of libwebsockets reflecting the real
status of the project with regard to the languages used.
Should fix:
http://autobuild.buildroot.org/results/fef/fef98ef289d44d13ec61025c50febba1e835b421/
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Building tests fails mainly because of the cross-compilation, so they
are disabled for the moment.
Signed-off-by: Hadrien Boutteville <hadrien.boutteville@gmail.com>
Reviewed-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Switch to gcc 4.9.x for microblaze since it's a better target than
4.8.x, and also add a build patch that fixes (e)glibc build issues.
Hence disable 4.8.x for microblaze.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Commit a5e92d92e7 converted uClibc to the
package infrastructure, but while doing so removed support for the
documented uclibc-update-config make target. This make target copies the
uclibc .config file back to the original (typically version-controlled)
configuration file.
As this was an unintentional removal, this patch re-adds the
uclibc-update-config target.
Reported-by: Dan Moulding <dan.moulding@rackwareinc.com>
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Kernel headers versions 3.6.x and 3.7.x have been deprecated since 2013.05
and thus can be removed in 2014.05.
An automatic selection of 3.8.x headers is performed in the legacy menu.
An existing automatic selection of 3.6.x headers is modified to select
3.8.x.
As this patch removes the last occurrances of BR2_DEPRECATED_SINCE_2013_05,
the symbol is removed too.
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The vala target package has been deprecated since 2013.05 and thus can be
removed in 2014.05. The host vala support is left untouched.
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This will resolve uClibc build error for configurations without large file
support.
Signed-off-by: Anton Kolesov <Anton.Kolesov@synopsys.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
bump to v1.23-chrome32-firefox24
The latest releases use the cmake build infrastructure.
Converted the makefile to the cmake infrastructure.
The autotools infrastructure is no longer supported.
[Peter: correct license data, ensure it gets built after openssl if enabled]
Signed-off-by: Sagaert Johan <sagaert.johan@skynet.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since qt's version has been bumped to 4.8.6 we can add support for
webkit on MIPS64 architectures (both big and little endian). It wasn't
possible on 4.8.5 because QtScript wasn't supported on MIPS64, but that
has been fixed on 4.8.6.
With reference to:
https://codereview.qt-project.org/#change,73022
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Reviewed-by: Markos Chandras <Markos.Chandras@imgtec.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Somehow the 'else' part got dropped from commit 3f82e9dbcd (use default gcc
4.8.2 for microblaze), breaking download for "normal" architectures.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
According to the imlib2 people, if you don't support dlopen(), you
have a "crap box":
image.h:16:4: warning: #warning "your crap box doesn't define RTLD_LOCAL !?"
So, let's disable imlib2 when doing static linking.
Fixes:
http://autobuild.buildroot.net/results/b5e/b5ee859409cd626b769e8a159026e60ec6f29180/
[Peter: fix imlib2 comment as pointed out by Thomas]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The toolchain is not up to the task, fixes:
http://autobuild.buildroot.net/results/21f/21f5fd1fc8415abff1d1178843586956ef1fd1e3/
Also AVR32s are low on resources (RAM, no proper storage port, most of
them short of flash) to handle it.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
There is no need for Xilinx Git.
Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Upstream gcc 4.8.2 works fine for microblaze, no need for
Xilinx Git.
Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Upstream binutils 2.24 works fine for microblaze, no need for
Xilinx Git.
[Peter: disable older versions as suggested by Gustavo]
Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Acked-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
