buildroot.git
14 years agolibglib2: add patch to rework clock_gettime() test
Thomas Petazzoni [Thu, 12 Aug 2010 13:59:14 +0000 (15:59 +0200)]
libglib2: add patch to rework clock_gettime() test

The test for clock_gettime() in configure.in doesn't work properly
when a previous package has loaded the shared configuration cache with
informations about the availability of clock_gettime. A package such
as ctorrent does so, which means that compiling ctorrent *then*
libglib2 currently fails.

According to people on the Autoconf mailing list, the libglib2 test is
likely the one that needs to be fixed. The problem is that the
AC_CHECK_FUNCS() test assumes that if it finds clock_gettime() it
means that there's no need to add any -lrt flag to the
build. Unfortunately, due to the shared configuration cache, this test
is already done with -lrt, so the test succeeds, and libglib2 does not
know that it needs to add -lrt to G_THREAD_LIBS and
G_THREAD_LIBS_FOR_GTHREAD.

So instead, we remplace the test with an AC_SEARCH_LIBS() test,
followed by a test on the result of this AC_SEARCH_LIBS() test to add
the necessary -lrt to G_THREAD_LIBS and
G_THREAD_LIBS_FOR_GTHREAD. Therefore, even if the result for the
AC_SEARCH_LIBS() test is cached due to the prior execution ctorrent
./configure script, libglib2 ./configure will properly add -lrt to the
appropriate variables.

Obviously, as this patch modifies the configure.in file, we enable the
autoreconf step for the libglib2 package.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoimagemagick: don't create useless debugging file
Thomas Petazzoni [Thu, 12 Aug 2010 13:57:02 +0000 (15:57 +0200)]
imagemagick: don't create useless debugging file

For some reason, the imagemagick Buildroot .mk file creates a
"datefile" file in the Buildroot source directory, probably an ancient
debugging thing that has been left here for no reason. Let's get rid
of it.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoUpdate busybox 1.17.1 fixes
Gustavo Zacarias [Wed, 11 Aug 2010 14:07:01 +0000 (11:07 -0300)]
Update busybox 1.17.1 fixes

New shell_common fix and sed fix.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
14 years agoPrevent C++ + locale + uClibc 0.9.31 + gcc 4.2 to be selected
Thomas Petazzoni [Wed, 11 Aug 2010 14:00:20 +0000 (16:00 +0200)]
Prevent C++ + locale + uClibc 0.9.31 + gcc 4.2 to be selected

The problem fixed by 60f945e47a15e10f0e777f69b05492b6f7ba918d is in
fact not limited to the AVR32 architecture, as reported by Will Newton
on the list. The issue is the combination uClibc 0.9.31 with gcc 4.2,
C++ support and locales.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoMark CRIS architecture as deprecated
Thomas Petazzoni [Wed, 11 Aug 2010 13:31:33 +0000 (15:31 +0200)]
Mark CRIS architecture as deprecated

The CRIS architecture support in Buildroot hasn't been updated since a
long time. Even a toolchain with recent kernel headers does not build
due to missing patches.

Moreover, the CRIS architecture has been discontinued by Axis, as
visible at http://www.axis.com/products/dev/index.htm. We will remove
it from Buildroot at the next release cycle.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoAdd the traditional powerpc-link-with-math-lib patch to gcc 4.4.4
Thomas Petazzoni [Wed, 11 Aug 2010 11:25:50 +0000 (13:25 +0200)]
Add the traditional powerpc-link-with-math-lib patch to gcc 4.4.4

Patch taken from Crosstool-NG patchset.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoDetect early if an UTF-8 locale is needed
Thomas Petazzoni [Wed, 11 Aug 2010 11:10:54 +0000 (13:10 +0200)]
Detect early if an UTF-8 locale is needed

Check in toolchain/dependencies/dependencies.sh if an UTF-8 locale is
properly present on the system before trying to build a locale enabled
toolchain. As this test is only needed when a locale enabled toolchain
is going to be built, we pass the configuration file path to the
dependencies.sh script so that it can grep for the current value of
various options.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoMake uClibc gen_wc8bit shows an error when no locale support available
Thomas Petazzoni [Wed, 11 Aug 2010 10:59:46 +0000 (12:59 +0200)]
Make uClibc gen_wc8bit shows an error when no locale support available

When no UTF-8 locale is available on the host system, uClibc can't
generate some stuff it needs to compile a C library with locale
support. Unfortunately, as gen_wc8bit message is shown on stdout and
the stdout of gen_wc8bit is redirected to a file, the user don't see
anything, as reported at
http://lists.busybox.net/pipermail/buildroot/2010-May/034177.html.

Those two patches fix the problem for uClibc 0.9.31 and 0.9.30.3. It
has been submitted upstream:
 http://lists.uclibc.org/pipermail/uclibc/2010-August/044256.html

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoAdd dependency from util-linux on ncurses
Thomas Petazzoni [Tue, 10 Aug 2010 17:15:26 +0000 (19:15 +0200)]
Add dependency from util-linux on ncurses

util-linux can build without ncurses, but when ncurses is available,
additional features can be built (such as the more
command). Therefore, in util-linux.mk, when ncurses is available, mark
it as a dependency.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoFix util-linux build on MIPS
Thomas Petazzoni [Tue, 10 Aug 2010 17:08:10 +0000 (19:08 +0200)]
Fix util-linux build on MIPS

As reported in bug #635, util-linux doesn't build due to missing
constant definitions related to the a.out binary format. We fix this
by hardcoding these constant definitions, as done in newer versions of
util-linux.

Obviously, the long term fix is to upgrade to util-linux-ng, but this
is probably not acceptable for 2010.08.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoupdate for 2010.08-rc1
Peter Korsgaard [Fri, 30 Jul 2010 10:35:45 +0000 (12:35 +0200)]
update for 2010.08-rc1

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoRemove gdb plugins from target
Malte Starostik [Sat, 24 Jul 2010 19:29:56 +0000 (21:29 +0200)]
Remove gdb plugins from target

* Don't install gdb plugins unless BR2_TARGET_GDB is set

Signed-off-by: Malte Starostik <m-starostik@versanet.de>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoGlobally remove aclocal directories from target
Malte Starostik [Sat, 24 Jul 2010 11:21:25 +0000 (13:21 +0200)]
Globally remove aclocal directories from target

* Remove $(TARGET_DIR)/usr/share/aclocal from target-finalize when not
  installing devfiles and
* Remove some (now) redundant cleanup from individual packages

Signed-off-by: Malte Starostik <m-starostik@versanet.de>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoPurge libglib2 dev files from target
Malte Starostik [Sat, 24 Jul 2010 11:21:28 +0000 (13:21 +0200)]
Purge libglib2 dev files from target

* Don't install libglib2 development binaries and to target unless
  BR2_HAVE_DEVFILES is set

Signed-off-by: Malte Starostik <m-starostik@versanet.de>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoCHANGES: add a bunch of issues resolved by Thomas
Peter Korsgaard [Fri, 30 Jul 2010 08:42:02 +0000 (10:42 +0200)]
CHANGES: add a bunch of issues resolved by Thomas

Thanks!

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoCHANGES: update with changes from fixes-20100729 branch
Peter Korsgaard [Fri, 30 Jul 2010 08:24:08 +0000 (10:24 +0200)]
CHANGES: update with changes from fixes-20100729 branch

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoMerge branch 'fixes-20100729' of git://git.busybox.net/~tpetazzoni/git/buildroot
Peter Korsgaard [Fri, 30 Jul 2010 08:21:40 +0000 (10:21 +0200)]
Merge branch 'fixes-20100729' of git://git.busybox.net/~tpetazzoni/git/buildroot

14 years agogcc: move <tuple>/lib* symlink handling up to gcc-intermediate
Peter Korsgaard [Fri, 30 Jul 2010 08:02:23 +0000 (10:02 +0200)]
gcc: move <tuple>/lib* symlink handling up to gcc-intermediate

The <tuple>/lib* symlinking added by 3c77bab2eeace needs to
be moved up to the gcc-intermediate step now the NPTL stuff is merged,
otherwise 64bit builds fails (lib64 already created).

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agotoolchain: enforce --disable-multilib
Peter Korsgaard [Fri, 30 Jul 2010 07:31:39 +0000 (09:31 +0200)]
toolchain: enforce --disable-multilib

Since 5575d205c (toolchain: remove multilib) we were no longer passing
--disable-multilib, which broke builds for multilib-capable archs (like
x86-64, ppc, ..).

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agolinux: bump 2.6.34.x version
Peter Korsgaard [Thu, 29 Jul 2010 21:50:27 +0000 (23:50 +0200)]
linux: bump 2.6.34.x version

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agobusybox: 1.17.1 mktemp fix
Peter Korsgaard [Thu, 29 Jul 2010 21:47:44 +0000 (23:47 +0200)]
busybox: 1.17.1 mktemp fix

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoxerces: only add -liconv when locale are disabled
Thomas Petazzoni [Thu, 29 Jul 2010 20:03:33 +0000 (22:03 +0200)]
xerces: only add -liconv when locale are disabled

The iconv library can only be present when locale are disabled in the
toolchain. When locale are enabled in the toolchain, iconv is directly
implemented by the C library.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoxerces: fix download location
Thomas Petazzoni [Thu, 29 Jul 2010 19:52:18 +0000 (21:52 +0200)]
xerces: fix download location

apache.jumper.nu does not work anymore, so use archive.apache.org
instead.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agolibcurl: bump version to fix build issue
Thomas Petazzoni [Thu, 29 Jul 2010 16:16:49 +0000 (18:16 +0200)]
libcurl: bump version to fix build issue

The ./configure script of libcurl includes <arpa_inet.h> instead of
<arpa/inet.h> when testing for inet_pton(). The test fails, but it
doesn't prevent libcurl to build as it can work without inet_pton().

However, it fills the configure cache with the fact that inet_pton()
does not exist. And later, tcpreplay reads this from the configure
cache and fails to build, because tcpreplay really need inet_pton().

Unfortunately, just fixing the .m4 file doesn't work because the
autoreconfiguration of the package fails. Since the fix for this
problem is already upstream, the easiest solution is therefore to bump
libcurl.

The libcurl-7.19.2-fix-ssl-no-verbose.patch patch is no longer needed.

Since we're patching a m4 file, we must autoreconfigure the package.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agotoolchain: mark uClibc 0.9.31 + locale + C++ as broken
Thomas Petazzoni [Thu, 29 Jul 2010 14:03:27 +0000 (16:03 +0200)]
toolchain: mark uClibc 0.9.31 + locale + C++ as broken

It fails to build with:

ctype_members.cc: In constructor 'std::ctype_byname<_CharT>::ctype_byname(const char*, size_t) [with _CharT = char]':
ctype_members.cc:59: error: invalid use of incomplete type 'struct __uclibc_locale_struct'
/home/test/avr32-br/usr/avr32-unknown-linux-uclibc/sys-include/bits/uClibc_locale.h:85: error: forward declaration of 'struct __uclibc_locale_struct'
ctype_members.cc:60: error: invalid use of incomplete type 'struct __uclibc_locale_struct'
/home/test/avr32-br/usr/avr32-unknown-linux-uclibc/sys-include/bits/uClibc_locale.h:85: error: forward declaration of 'struct __uclibc_locale_struct'
ctype_members.cc:61: error: invalid use of incomplete type 'struct __uclibc_locale_struct'
/home/test/avr32-br/usr/avr32-unknown-linux-uclibc/sys-include/bits/uClibc_locale.h:85: error: forward declaration of 'struct __uclibc_locale_struct'
make[5]: *** [ctype_members.lo] Error 1

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agocairo: PDF support is needed when SVG support is enabled
Thomas Petazzoni [Thu, 29 Jul 2010 13:56:48 +0000 (15:56 +0200)]
cairo: PDF support is needed when SVG support is enabled

As soon as PostScript, PNG or SVG support is enabled, PDF support is
required for Cairo to build properly. Otherwise, you get build
failures such as:

.libs/cairo-type3-glyph-surface.o: In function `_cairo_type3_glyph_surface_set_stream':
/home/thomas/local/buildroot-dl/cairo-1.8.10/src/cairo-type3-glyph-surface.c:337: undefined reference to `_cairo_pdf_operators_set_stream'
/home/thomas/local/buildroot-dl/cairo-1.8.10/src/cairo-type3-glyph-surface.c:337: undefined reference to `_cairo_pdf_operators_set_stream'

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agolinux: more mistake detection
Thomas Petazzoni [Thu, 29 Jul 2010 13:40:28 +0000 (15:40 +0200)]
linux: more mistake detection

Just as we do for U-Boot, error out in the Linux kernel makefile when
the defconfig name or the configuration file path are not
correct. What prompted me to implement this was a report on IRC from
an user using BR 2010.05 and not understand why the kernel build
process was failing. It was because he just forgot to set the path of
the configuration file.

Of course, it doesn't catch all mistakes (like pointing to a
non-existing defconfig or to a non-existing configuration file), but
it at least catches basic mistakes.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agou-boot: fix custom patch handling
Thomas Petazzoni [Thu, 29 Jul 2010 13:38:27 +0000 (15:38 +0200)]
u-boot: fix custom patch handling

When U-Boot is enabled and no custom patch directory has been set,
then the current test:

  ifneq ($(strip $(BR2_TARGET_UBOOT_CUSTOM_PATCH_DIR)),"")

works. However, when U-Boot is not enabled, but still gets compiled
because mkimage is needed to build the kernel,
BR2_TARGET_UBOOT_CUSTOM_PATCH_DIR is completely empty. It does not
even have quotes. So the test in fact needs to be:

 ifneq ($(call qstrip,$(BR2_TARGET_UBOOT_CUSTOM_PATCH_DIR)),)

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agosamba: remove swat documentation when not needed
Thomas Petazzoni [Thu, 29 Jul 2010 11:41:46 +0000 (13:41 +0200)]
samba: remove swat documentation when not needed

When SWAT (the Web administration tool of Samba) is enabled, which is
the default when one enables samba in Buildroot, a lot of
documentation gets installed in /usr/swat (~15 MB). This patch fixes
this by removing the documentation when BR2_HAVE_DOCUMENTATION is not
set.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoavahi: pkg-config is needed
Thomas Petazzoni [Thu, 29 Jul 2010 09:57:15 +0000 (11:57 +0200)]
avahi: pkg-config is needed

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agospeech-tools: fix build failure by using Debian version and patches
Thomas Petazzoni [Thu, 29 Jul 2010 09:14:00 +0000 (11:14 +0200)]
speech-tools: fix build failure by using Debian version and patches

The upstream version of speech-tools does not build with GCC >= 4.3,
mainly due to changes in how C++ headers are included. This is fixed
in Debian, so let's use the Debian version and patches.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agodmalloc: pass -fPIC when compiling
Thomas Petazzoni [Thu, 29 Jul 2010 08:18:19 +0000 (10:18 +0200)]
dmalloc: pass -fPIC when compiling

This is needed to avoid:

/home/test/mips-4.4/bin/mips-linux-gnu-ld --sysroot=/home/test/outputs/test-35/staging -shared --whole-archive -soname libdmallocxx.so -o libdmallocxx.so.t libdmallocxx.a
/home/test/mips-4.4/bin/mips-linux-gnu-ld: libdmalloc.a(arg_check.o): relocation R_MIPS_HI16 against `_dmalloc_flags' can not be used when making a shared object; recompile with -fPIC

It is fixed through a patch to Makefile.in instead of passing a CFLAGS
variable to ./configure environment in order to avoid cluttering the
configuration cache with incorrect values.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoiostat: migrate to gentargets
Gustavo Zacarias [Thu, 22 Jul 2010 15:50:54 +0000 (12:50 -0300)]
iostat: migrate to gentargets

Also make the cpu counting routine more reliable (for ARM it's
"Processor" in cpuinfo rather than "processor").

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agouClibc: remove old 0.9.28 support
Peter Korsgaard [Thu, 29 Jul 2010 15:02:29 +0000 (17:02 +0200)]
uClibc: remove old 0.9.28 support

Not supported upstream and needs complicated workaround for the NPTL stuff.

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agogcc-4.2.4: Add patch to accept --with-abi=aapcs-linux
Khem Raj [Wed, 28 Jul 2010 07:11:39 +0000 (00:11 -0700)]
gcc-4.2.4: Add patch to accept --with-abi=aapcs-linux

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoAdd support for uclibc NPTL toolchain.
Khem Raj [Wed, 28 Jul 2010 07:11:37 +0000 (00:11 -0700)]
Add support for uclibc NPTL toolchain.

This patch modifies current toolchain build sequence so that
NPTL enabled toolchain can be built. The new sequence works
well with linuxthreads as well.

It introduces a new pass for gcc cross compilation. The new
sequence is binutils->gcc-initial->linux-headers -> uclibc-configured
(some cheats to generate phony shared libc.so and libm.o)
-> gcc-intermediate(with shared lib support) -> uclibc -> gcc-final

I also added a new sample config arm_nptl_toolchain_defconfig which
builds the toolchain and busybox.

I have only tried it on arm. However it should work for other
architectures which support NPTL on uclibc e.g. mips, sh, x86, ppc, x86_64

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agocdrkit: fix TARGET_CC/TARGET_CFLAGS for CMake
Thomas Petazzoni [Thu, 29 Jul 2010 08:10:35 +0000 (10:10 +0200)]
cdrkit: fix TARGET_CC/TARGET_CFLAGS for CMake

Since the reorganization of the variables in package/Makefile.in,
TARGET_CC and TARGET_CXX now directly contain the --sysroot= option in
addition to the compiler path. This is due to some ./configure scripts
using just $(TARGET_CC) for some tests instead of $(TARGET_CC)
$(TARGET_CFLAGS).

However, in the case of CMake, this fails as CMake really only wants
the path of the compiler in its CMAKE_C_COMPILER and
CMAKE_CXX_COMPILER variables. So here, we recompute proper values for
CMake by removing the --sysroot option from the compiler variables and
re-adding it to the flags variables.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agolibeXosip: add dependency on host-pkg-config and remove useless flags
Thomas Petazzoni [Thu, 29 Jul 2010 08:03:32 +0000 (10:03 +0200)]
libeXosip: add dependency on host-pkg-config and remove useless flags

When libeXosip fails to use pkg-config to find libosip, it defaults to
thinking that libosip is installed in $(prefix)/lib and
$(prefix)/include, which is of course wrong. There was an attempt to
fix this by passing OSIP_CFLAGS and OSIP_LIBS variables to libeXosip
./configure script, but it still does not work:

checking pkg-config is at least version 0.9.0... ./configure: line 21035: /home/test/outputs/test-41/host/usr/bin/pkg-config: No such file or directory
no
checking for OSIP... configure: WARNING: assuming osip can be found in -I${prefix}/include and -L${exec_prefix}/lib

Therefore, the correct fix is to depend on host-pkg-config.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agolibiconv: add an error when both libiconv and locale are enabled
Thomas Petazzoni [Thu, 29 Jul 2010 07:50:59 +0000 (09:50 +0200)]
libiconv: add an error when both libiconv and locale are enabled

This error should never show up if all Buildroot dependencies are
correct. However, rather than failing horribly later on, catch this
particular case early on and error out.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoxerces: fix dependency on iconv
Thomas Petazzoni [Thu, 29 Jul 2010 07:49:42 +0000 (09:49 +0200)]
xerces: fix dependency on iconv

All "select BR2_PACKAGE_LIBICONV" must use the "if !BR2_ENABLE_LOCALE"
condition, otherwise we can end up with a toolchain suppoting locales
*and* the libiconv package being compiled, which confuses other
packages. Example with glib:

gconvert.c:52:2: error: #error GNU libiconv in use but included
iconv.h not from libiconv

In addition to that, in xerces.mk, we add the dependency on libiconv
when it is available, to make sure it gets compiled before xerces.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agogdb: disallow GDB_HOST on external toolchain builds
Thomas Petazzoni [Thu, 29 Jul 2010 07:38:20 +0000 (09:38 +0200)]
gdb: disallow GDB_HOST on external toolchain builds

The cross-gdb is supposed to be part of the external toolchain, so
Buildroot does not need to build it. Moreover, GDB_HOST build
currently fail with:

ln -snf ../../bin/arm-unknown-linux-gnueabi-gdb \
                /home/test/outputs/test-48/staging/usr/arm-unknown-linux-gnueabi/bin/gdb
ln: creating symbolic link `/home/test/outputs/test-48/staging/usr/arm-unknown-linux-gnueabi/bin/gdb': No such file or directory

And even worse: they overwrite the cross-gdb of the external
toolchain!

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoFix computation of REAL_GNU_TARGET_NAME
Thomas Petazzoni [Thu, 29 Jul 2010 07:30:40 +0000 (09:30 +0200)]
Fix computation of REAL_GNU_TARGET_NAME

The current computation of REAL_GNU_TARGET_NAME is incorrect for
non-ARM glibc platforms because it generates something such as
mipsel-unknown-linux- as the REAL_GNU_TARGET_NAME.

So we correct this by :

 * Adding "gnu" in the suffix when glibc is used, so that in the
   previous case we will have mipsel-unknown-linux-gnu

 * Improving the ARM_EABI code to correctly append "eabi" when glibc
   is selected, so that we have arm-unknown-linux-gnueabi, and to
   append "gnueabi" when uclibc is selected, so that we have
   arm-unknown-linux-uclibcgnueabi. The little trick here is that LIBC
   and ABI aren't completely orthogonal on ARM.

This fixes problems such as :

checking host system type... Invalid configuration
`mipsel-unknown-linux-': machine `mipsel-unknown-linux' not recognized

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agooprofile: disable with external toolchain
Thomas Petazzoni [Thu, 29 Jul 2010 06:45:20 +0000 (08:45 +0200)]
oprofile: disable with external toolchain

oprofile depends on binutils_target, but binutils_target fails to
build with external toolchains because the binutils version has not
been choosen. As the fix is not trivial, let's just disable oprofile
in external toolchain builds for the moment.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agodependencies: add svn as a mandatory tool
Thomas Petazzoni [Thu, 29 Jul 2010 06:37:56 +0000 (08:37 +0200)]
dependencies: add svn as a mandatory tool

Now that two packages (tremor and libsvgtiny) are being downloaded
from svn, svn becomes a mandatory tool to run Buildroot.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agonetsnmp: depend on openssl when the package is on to ensure build order
Gustavo Zacarias [Thu, 29 Jul 2010 12:15:17 +0000 (09:15 -0300)]
netsnmp: depend on openssl when the package is on to ensure build order

The netsnmp package should depend on openssl when using it.
Otherwise netsnmp might get built before openssl and poison the
configure cache since it's not a mandatory dependency.

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoCHANGES: #2257 / #2263 are resolved
Peter Korsgaard [Thu, 29 Jul 2010 07:10:25 +0000 (09:10 +0200)]
CHANGES: #2257 / #2263 are resolved

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agotoolchain: remove multilib
Thomas Petazzoni [Wed, 28 Jul 2010 10:07:04 +0000 (12:07 +0200)]
toolchain: remove multilib

Supporting multilib is much more than just passing --enable-multilib
to gcc. You have to actually build the C library several times (once
for each multilib variant you want to support in your toolchain), and
to pass MULTILIB_OPTIONS/MULTILIB_EXCEPTIONS values to gcc to let it
know the set of multilib variants you're interested in.

Since we'll probably never support multilib toolchains in Buildroot,
just get rid of this BR2_ENABLE_MULTILIB option.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agogcc: remove option on SJLJ exceptions
Thomas Petazzoni [Wed, 28 Jul 2010 10:02:30 +0000 (12:02 +0200)]
gcc: remove option on SJLJ exceptions

This is a very advanced option, and it seems, according to
http://choices.cs.uiuc.edu/exceptions.pdf that SJLJ exceptions aren't
really interesting.

Users really interested by this can always use the
BR2_EXTRA_GCC_CONFIG_OPTIONS is they want.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoCHANGES: add note about GTK+ on directfb being deprecated
Peter Korsgaard [Wed, 28 Jul 2010 14:27:09 +0000 (16:27 +0200)]
CHANGES: add note about GTK+ on directfb being deprecated

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoMerge branch 'deprecate-gtk-on-dfb' of git://git.busybox.net/~tpetazzoni/git/buildroot
Peter Korsgaard [Wed, 28 Jul 2010 14:24:10 +0000 (16:24 +0200)]
Merge branch 'deprecate-gtk-on-dfb' of git://git.busybox.net/~tpetazzoni/git/buildroot

14 years agoCHANGES: update with changes from the 'various-things' branch
Peter Korsgaard [Wed, 28 Jul 2010 14:21:44 +0000 (16:21 +0200)]
CHANGES: update with changes from the 'various-things' branch

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoMerge branch 'various-things' of git://git.busybox.net/~tpetazzoni/git/buildroot
Peter Korsgaard [Wed, 28 Jul 2010 14:20:46 +0000 (16:20 +0200)]
Merge branch 'various-things' of git://git.busybox.net/~tpetazzoni/git/buildroot

14 years agotoolchain: move buildroot config files
Yann E. MORIN [Tue, 27 Jul 2010 22:08:16 +0000 (00:08 +0200)]
toolchain: move buildroot config files

Handle the internal toolchain backend mechanism the
same way we handle other backends.

Signed-off-by: Yann E. MORIN <yann.morin.1998@anciens.enib.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agotoolchain: move makefile includes
Yann E. MORIN [Tue, 27 Jul 2010 22:08:15 +0000 (00:08 +0200)]
toolchain: move makefile includes

Including a bunch of Makefiles with wildcard makes it impossible to add
new toolchain backends. Avoid that by namely including needed files.

The external toolchain still needs to include all the toolchain/*/*.mk
sub-makefiles, as they are needed to build a toolchain that runs on the
target. It is to be noted that the cross-toolchain is not built in this
case, as the make-targets to build the cross-toolchain are not present
in the $(BASE_TARGETS) variable, which is later used to create the
dependency rules.

Also, the comment 'Explicit ordering' has been removed, as it is mis-
leading. It is make's responsibility to create the proper ordering based
on the dependency rules it finds in the Makefiles

Signed-off-by: Yann E. MORIN <yann.morin.1998@anciens.enib.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agotoolchain: rename external toolchain dir
Yann E. MORIN [Tue, 27 Jul 2010 22:08:14 +0000 (00:08 +0200)]
toolchain: rename external toolchain dir

Rename the external toolchain directory.
When new backends are here, it will be easier to sort them out
if they are all prefixed the same way.

Signed-off-by: Yann E. MORIN <yann.morin.1998@anciens.enib.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agotoolchain: move helper functions from external toolchain
Yann E. MORIN [Tue, 27 Jul 2010 22:08:13 +0000 (00:08 +0200)]
toolchain: move helper functions from external toolchain

The helper functions used for external toolchains may also be useful
to alternate toolchain backends (currently, the external toolchain is
the sole user).

Signed-off-by: Yann E. MORIN <yann.morin.1998@anciens.enib.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agolibgtk2: mark DirectFB backend as deprecated
Thomas Petazzoni [Wed, 28 Jul 2010 10:13:45 +0000 (12:13 +0200)]
libgtk2: mark DirectFB backend as deprecated

libgtk2 on DirectFB is deprecated because it is no longer supported in
recent versions of Gtk. We will remove support for Gtk over DirectFB
in the next Buildroot version unless support for DirectFB in mainline
Gtk is improved in the mean time.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoBump samba to 3.3.13 [CVE-2010-2063]
Gustavo Zacarias [Wed, 21 Jul 2010 13:58:00 +0000 (10:58 -0300)]
Bump samba to 3.3.13 [CVE-2010-2063]

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
14 years agoRemove unused X11_PREFIX option
Thomas Petazzoni [Tue, 27 Jul 2010 21:46:27 +0000 (23:46 +0200)]
Remove unused X11_PREFIX option

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoConvert net-snmp to autotargets
Gustavo Zacarias [Tue, 20 Jul 2010 18:30:07 +0000 (15:30 -0300)]
Convert net-snmp to autotargets

Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
14 years agozlib: Bump to 1.2.5
Paul Jones [Mon, 7 Jun 2010 12:33:06 +0000 (22:33 +1000)]
zlib: Bump to 1.2.5

Closes #1981

Signed-off-by: Paul Jones <paul@pauljones.id.au>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoDeprecate the Alpha, IA64 and Sparc64 support
Thomas Petazzoni [Mon, 26 Jul 2010 16:15:11 +0000 (18:15 +0200)]
Deprecate the Alpha, IA64 and Sparc64 support

Those architectures don't exist anymore (Alpha, IA64) or aren't widely
used for embedded systems running Linux. Moreover, no clear Buildroot
maintainer has stepped in to maintain these architectures, so it's
better to not pretend that we support them.

The goal is to mark them as deprecated in 2010.08 and remove them in
2010.11.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agolibglib2: bump to 2.24.1
Lionel Landwerlin [Tue, 27 Jul 2010 21:16:26 +0000 (23:16 +0200)]
libglib2: bump to 2.24.1

Signed-off-by: Lionel Landwerlin <llandwerlin@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoMerge branch 'avr32-toolchain-fix' of git://git.busybox.net/~tpetazzoni/git/buildroot
Peter Korsgaard [Tue, 27 Jul 2010 21:06:16 +0000 (23:06 +0200)]
Merge branch 'avr32-toolchain-fix' of git://git.busybox.net/~tpetazzoni/git/buildroot

14 years agoCHANGES: update with changes from various-bumps branch
Peter Korsgaard [Tue, 27 Jul 2010 21:01:29 +0000 (23:01 +0200)]
CHANGES: update with changes from various-bumps branch

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agotoolchain: remove redundant and incorrect --with-build-time-tools option
Thomas Petazzoni [Thu, 22 Jul 2010 13:44:36 +0000 (15:44 +0200)]
toolchain: remove redundant and incorrect --with-build-time-tools option

This option is already part of the gcc configure options through the
BR2_CONFIGURE_BUILD_TOOLS variable (in toolchain/Makefile.in).

Additionnally, the value that was passed in the AVR32 specific case
was incorrect: it was $(STAGING_DIR)/$(REAL_GNU_TARGET_NAME)/bin
instead of $(STAGING_DIR)/usr/$(REAL_GNU_TARGET_NAME)/bin.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agotoolchain: Remove now-unused variables
Thomas Petazzoni [Thu, 22 Jul 2010 13:43:37 +0000 (15:43 +0200)]
toolchain: Remove now-unused variables

The variable BR2_SYSROOT_STAGING_DESTDIR is no longer used, since now
the prefix for gcc is already set to the correct location.

The variable BR2_SYSROOT_TARGET_DESTDIR was already unused.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agobinutils,gcc: use correct --prefix
Thomas Petazzoni [Thu, 22 Jul 2010 13:42:27 +0000 (15:42 +0200)]
binutils,gcc: use correct --prefix

The cross binutils and cross gcc are actually going to be executed
from $(STAGING_DIR)/usr, so the correct prefix is $(STAGING_DIR)/usr
and not /usr.

This also fixes what is known as the "AVR32 toolchain build failure",
which was due to the fact that the prefix directory wasn't writable
(since it was /usr).

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoMerge branch 'various-bumps' of git://git.busybox.net/~tpetazzoni/git/buildroot
Peter Korsgaard [Tue, 27 Jul 2010 20:52:19 +0000 (22:52 +0200)]
Merge branch 'various-bumps' of git://git.busybox.net/~tpetazzoni/git/buildroot

14 years agoCreate <tuple>/lib -> <sysroot>/lib symlink before installing cross gcc
Thomas Petazzoni [Tue, 27 Jul 2010 14:25:15 +0000 (16:25 +0200)]
Create <tuple>/lib -> <sysroot>/lib symlink before installing cross gcc

This commit solves bug #1051. The problem in this bug in that WebKit
compiles a sample C program, which uses WebKit. As WebKit is written
in C++, even though the program it built with CROSS-gcc, it must be
linked with libstdc++. However, CROSS-gcc can't find the libstdc++ has
it's hidden inside <sysroot>/<tuple>/lib.

Therefore, this commit creates a symbolic link <sysroot>/<tuple>/lib
-> <sysroot>/lib before running the CROSS-gcc installation. While this
may look like a hack, this is the solution used by both Crosstool-NG
and OpenWRT.

Moreover, with this symbolic link in place, I think bug #1741 may also
be solved. The problem in this bug is that the linker tries to link
against /lib/libc.so.0. This is due to the fact that the linker finds
a libc.so script file in the original toolchain location and not
inside the copy of the toolchain sysroot in $(STAGING_DIR). As the
script file is found outside of the current toolchain sysroot, ld
considers the script has non-sysrooted, and therefore doesn't prefix
all paths found in the script file (such as /lib/libc.so.0) with the
sysroot path, leading to the failure.

So, in details, this commit :

 * Adds a BR2_ARCH_IS_64 invisible config knob that is used to know if
   the arch is a 64 bits architecture or not.

 * Creates the <sysroot>/<tuple>/lib -> <sysroot>/lib symbolic link,
   and the <sysroot>/<tuple>/lib64 -> <sysroot>/lib64 symbolic link if
   needed.

 * Fixes the external toolchain sysroot detection code so that the
   'sed' replacement is done *after* the readlink -f evaluation.

I have tested this by building ARM, x86 and x86_64 toolchains with
Buildroot, and then use these toolchains as external toolchains to
build a full X.org/Gtk/WebKit/Midori stack. I have also done a
complete ARM Buildroot internal toolchain build with the same full
X.org/Gtk/WebKit/Midori stack.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agowebkit: Fix build failure on uClibc
Thomas Petazzoni [Mon, 26 Jul 2010 13:21:09 +0000 (15:21 +0200)]
webkit: Fix build failure on uClibc

uClibc does not define pthread_getattr_np(), so we need to workaround
this using a patch.

Fixes bug #1405.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agowebkit: the X11 variant needs libXt
Thomas Petazzoni [Thu, 22 Jul 2010 15:03:28 +0000 (17:03 +0200)]
webkit: the X11 variant needs libXt

Fixes bug #829.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agomidori: make sure it can't be selected if webkit can't be enabled
Thomas Petazzoni [Thu, 22 Jul 2010 14:03:28 +0000 (16:03 +0200)]
midori: make sure it can't be selected if webkit can't be enabled

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoxserver: Add xkeyboard-config as a dependency
Thomas Petazzoni [Thu, 22 Jul 2010 14:22:32 +0000 (16:22 +0200)]
xserver: Add xkeyboard-config as a dependency

xkeyboard-config is required to install keymap files to the target,
without which the X server does not start.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoxserver: tell the X server where fonts are installed
Thomas Petazzoni [Thu, 22 Jul 2010 13:47:48 +0000 (15:47 +0200)]
xserver: tell the X server where fonts are installed

All fonts are installed in /usr/share/fonts/X11 and not in
/usr/lib/X11/fonts/ where the X server looks by default.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoxkeyboard-config: add dependency on host-xapp_xkbcomp
Thomas Petazzoni [Thu, 22 Jul 2010 13:37:16 +0000 (15:37 +0200)]
xkeyboard-config: add dependency on host-xapp_xkbcomp

In order to get xkeyboard-config to build, it needs xkbcomp to be
built on the host, i.e host-xapp_xkbcomp. In turn, xapp_xkbcomp needs
xlib_libX11 and xlib_libxkbfile, which in turn need a few libraries.

This commit adds all the relevant dependencies to build everything
that is needed to get xkeyboard-config to build correctly.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agolibsoup: bump version to 2.31.2
Thomas Petazzoni [Mon, 19 Jul 2010 09:53:41 +0000 (11:53 +0200)]
libsoup: bump version to 2.31.2

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agocairo: bump to 1.8.10
Thomas Petazzoni [Mon, 19 Jul 2010 08:45:45 +0000 (10:45 +0200)]
cairo: bump to 1.8.10

This commit bumps Cairo to 1.8.10 and updates the DirectFB support
patch. The work has been done by Lionel Landwerlin
<llandwerlin@gmail.com>.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoicu: add quotes around TARGET_CXX
Thomas Petazzoni [Mon, 19 Jul 2010 07:55:13 +0000 (09:55 +0200)]
icu: add quotes around TARGET_CXX

Now that TARGET_CXX contains a --sysroot= option and therefore spaces,
it needs to be used with quotes.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agomidori: bump to 0.2.6
Thomas Petazzoni [Mon, 19 Jul 2010 07:21:22 +0000 (09:21 +0200)]
midori: bump to 0.2.6

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agowebkit: bump to webkitgtk 1.2.3
Thomas Petazzoni [Mon, 19 Jul 2010 07:12:15 +0000 (09:12 +0200)]
webkit: bump to webkitgtk 1.2.3

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoMerge branch 'fix-gxx-on-target' of git://git.busybox.net/~tpetazzoni/git/buildroot
Peter Korsgaard [Tue, 27 Jul 2010 20:46:18 +0000 (22:46 +0200)]
Merge branch 'fix-gxx-on-target' of git://git.busybox.net/~tpetazzoni/git/buildroot

14 years agoMerge branch 'fix-gcc-on-target' of git://git.busybox.net/~tpetazzoni/git/buildroot
Peter Korsgaard [Tue, 27 Jul 2010 20:43:04 +0000 (22:43 +0200)]
Merge branch 'fix-gcc-on-target' of git://git.busybox.net/~tpetazzoni/git/buildroot

14 years agopackage: add libgail package
Lionel Landwerlin [Tue, 13 Apr 2010 23:06:08 +0000 (01:06 +0200)]
package: add libgail package

This package include a patch for directfb support. This patch has been
proposed for upstream integration
(https://bugzilla.gnome.org/show_bug.cgi?id=614199).

Signed-off-by: Lionel Landwerlin <llandwerlin@gmail.com>
14 years agopango: bump to 1.28
Lionel Landwerlin [Tue, 13 Apr 2010 23:06:13 +0000 (01:06 +0200)]
pango: bump to 1.28

Signed-off-by: Lionel Landwerlin <llandwerlin@gmail.com>
14 years agoatk: bump to 1.30
Lionel Landwerlin [Tue, 13 Apr 2010 23:06:11 +0000 (01:06 +0200)]
atk: bump to 1.30

Signed-off-by: Lionel Landwerlin <llandwerlin@gmail.com>
14 years agolibglib2: bump to 2.24.0
Lionel Landwerlin [Tue, 13 Apr 2010 23:06:10 +0000 (01:06 +0200)]
libglib2: bump to 2.24.0

Signed-off-by: Lionel Landwerlin <llandwerlin@gmail.com>
14 years agotarget-g++: fix build
Thomas Petazzoni [Tue, 27 Jul 2010 20:32:02 +0000 (22:32 +0200)]
target-g++: fix build

Just as we did to fix target-gcc, pass CXX_FOR_TARGET when building
target g++, and remove useless copies of g++ and c++.

Tested on ARM by compiling a simple C++ program using <iostream> on
the target and running it.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agotarget-gcc: remove useless copies of gcc
Thomas Petazzoni [Tue, 27 Jul 2010 16:21:44 +0000 (18:21 +0200)]
target-gcc: remove useless copies of gcc

When doing the "make install" of target, three identical copies of gcc
are installed in $(TARGET_DIR)/usr/bin:

  039adcc582c365f12ba6fc5f96098128  arm-unknown-linux-uclibcgnueabi-gcc
  039adcc582c365f12ba6fc5f96098128  arm-unknown-linux-uclibcgnueabi-gcc-4.3.5
  039adcc582c365f12ba6fc5f96098128  gcc

This patch removes the first two copies and keeps only the common "gcc" one.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agotarget-gcc: no need to strip binaries, remove .la files and doc
Thomas Petazzoni [Tue, 27 Jul 2010 16:20:15 +0000 (18:20 +0200)]
target-gcc: no need to strip binaries, remove .la files and doc

This is done in a global way by the target-finalize target of the main
Makefile.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agotarget-gcc: fix build
Thomas Petazzoni [Tue, 27 Jul 2010 16:17:26 +0000 (18:17 +0200)]
target-gcc: fix build

Now that $(STAGING_DIR)/usr/bin is no longer in the PATH, we need to
pass the absolute paths to $(TARGET_CC) when building the target gcc
compiler.

This commit fixes the target gcc build problem reported on the list. I
have successfully been able to build a target gcc for ARM, use it to
compile a hello world application on the target and run this
application.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agotarget-gcc: Get rid of TARGET_GCC_FLAGS
Thomas Petazzoni [Tue, 27 Jul 2010 16:16:59 +0000 (18:16 +0200)]
target-gcc: Get rid of TARGET_GCC_FLAGS

This variable is used only once, so let's just hardcode its value at
its call site.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
14 years agoat: fix configure cache issue with flex
Peter Korsgaard [Tue, 27 Jul 2010 07:45:26 +0000 (09:45 +0200)]
at: fix configure cache issue with flex

Make sure flex is built before at if enabled, so it correctly detects
it and doesn't populate tgt-config.cache with invalid values.

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agognuchess: fix configure cache issue with readline
Peter Korsgaard [Tue, 27 Jul 2010 07:25:39 +0000 (09:25 +0200)]
gnuchess: fix configure cache issue with readline

Make sure readline is built before gnuchess if enabled, so it correctly
detects it and doesn't populate tgt-config.cache with invalid values.

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agodiffutils: fix configure cache issue with BR2_NEEDS_GETTEXT
Peter Korsgaard [Tue, 27 Jul 2010 07:19:45 +0000 (09:19 +0200)]
diffutils: fix configure cache issue with BR2_NEEDS_GETTEXT

Like flex, diffutils doesn't NEED gettext/libintl, but it's configure script
checks for it, so make sure those a built before diffutils, otherwise it
will populate tgt-config.cache with invalid values, breaking the build of
other packages needing it (like libglib2).

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoCHANGES: fix typo
Peter Korsgaard [Tue, 27 Jul 2010 07:08:01 +0000 (09:08 +0200)]
CHANGES: fix typo

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoDeprecate pcmcia and microwin packages
Thomas Petazzoni [Sun, 25 Jul 2010 14:53:52 +0000 (16:53 +0200)]
Deprecate pcmcia and microwin packages

It isn't worth the effort to convert these packages to the new package
infrastructure since they are deprecated.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agosetserial: convert to autotargets and fix manpages install
Thomas Petazzoni [Sun, 25 Jul 2010 14:53:51 +0000 (16:53 +0200)]
setserial: convert to autotargets and fix manpages install

Close #2269

[Peter: fix Makefile patch to not strip wrong file, rename hayes patch]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agotftpd: convert to autotargets
Thomas Petazzoni [Sun, 25 Jul 2010 14:53:50 +0000 (16:53 +0200)]
tftpd: convert to autotargets

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
14 years agoltrace: convert to gentargets and bump to 0.5.3
Thomas Petazzoni [Sun, 25 Jul 2010 14:53:49 +0000 (16:53 +0200)]
ltrace: convert to gentargets and bump to 0.5.3

[Peter: remove unneeded 0.5 USE_DEMANGLE patch]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>