Gustavo Zacarias [Mon, 21 Feb 2011 19:25:06 +0000 (16:25 -0300)]
quagga: fix build with snmp support
Closes #3277
Quagga's snmp support tries to link with libcrypto from openssl by
default.
That's a bad no no since snmp might have ssl support disabled
and openssl won't be around.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The makedevs script did not always generate the requested set of device names / minor number series.
* If start != 0, then requesting (count) devices would generate only (count - start)
* If start != 0 && increment != 1, then requesting minors starting with (minor) would generate minors starting with (minor + (start * (increment - 1)))
This patch fixes the code and updates the usage text with extra examples.
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
OpenSSL's build system tries to be too wise for it's own good when
guessing what libdir should be.
This causes problems like the one reported in bug #3205 so just specify
libdir to point to /lib (since it's prefixed it would finally be
/usr/lib) since it should be present on 32 and 64 bit targets.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Peter Korsgaard [Wed, 23 Feb 2011 21:37:47 +0000 (22:37 +0100)]
metacity: fix build, bump version
Metacity no longer built after libgtk2 was updated, so fix it by bumping
the version. We unfortunately cannot move to the latest version, as
these have extra dependencies we don't have in BR yet
(libcanberra-gtk + gnome-doc-utils).
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Peter Korsgaard [Wed, 23 Feb 2011 15:46:07 +0000 (16:46 +0100)]
u-boot: drop old 2009.xx versions
We use 'make tools' to build a mkimage for the host if needed by the
kernel (uImage) without having u-boot configured, but that only works
since the 2010.03 release:
Peter Korsgaard [Tue, 22 Feb 2011 23:40:11 +0000 (00:40 +0100)]
package/multimedia: mark festival packages as broken
After spending a couple of hours on it today, I'm still unable to get
speech-tools to build and install properly with a uClibc toolchain.
So I'm giving up. Mark the entire thing broken as the text-to-speech
stuff is unlikely to be used very much on the kind of small systems
we're targetting.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Peter Korsgaard [Sun, 20 Feb 2011 22:31:38 +0000 (23:31 +0100)]
nfs-utils: needs largefile support
Even though ./configure has a --disable-largefile, functions like
statfs64 is used in several places, breaking the build on toolchains
without largefile support.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Peter Korsgaard [Sat, 19 Feb 2011 22:37:40 +0000 (23:37 +0100)]
libmicrohttpd: make https support optional
Libmicrohttpd works nicely without https support, so make it optional.
Also ensure the cross-libgcrypt-config is found, rather than whatever
is installed on the host.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Peter Korsgaard [Fri, 18 Feb 2011 13:17:34 +0000 (14:17 +0100)]
olsr: fix build
Pass TARGET_CONFIGURE_OPTS in the environment rather than as make arguments,
so the olsr build system can append to CFLAGS / LDFLAGS - Otherwise build
fails when the wrong linker flags are used to create the plugin .so's.
Also pass ARCH, so the olsr build system choses the right libdir / PIC
settings for the target, rather than for the host.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Peter Korsgaard [Fri, 18 Feb 2011 11:30:00 +0000 (12:30 +0100)]
taglib: fix build
The old libtool version gets confused about rpath, which breaks the build.
FIx it by up upgrading to 1.5.22 (package is not compatible with libtool
2.x, so we cannot just autoreconf).
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Nate Ford [Fri, 11 Feb 2011 09:26:24 +0000 (10:26 +0100)]
package: add xmlstarlet package
[Peter: don't link statically, only depend on iconv if !BR_ENABLE_LOCALE] Signed-off-by: Nate Ford <nford@westpond.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Bjørn Forsman [Thu, 10 Feb 2011 16:47:15 +0000 (17:47 +0100)]
mplayer: bump version to fix build error
Bump mplayer to the first version that fixes this issue:
file.c:23:5: warning: "HAVE_MMAP" is not defined
file.c:25:7: warning: "HAVE_MAPVIEWOFFILE" is not defined
file.c:77:5: warning: "HAVE_MMAP" is not defined
file.c:87:7: warning: "HAVE_MAPVIEWOFFILE" is not defined
file.c:124:5: warning: "HAVE_MMAP" is not defined
file.c:126:7: warning: "HAVE_MAPVIEWOFFILE" is not defined
file.c: In function 'av_file_map':
file.c:50: error: 'HAVE_MMAP' undeclared (first use in this function)
file.c:50: error: (Each undeclared identifier is reported only once
file.c:50: error: for each function it appears in.)
make[2]: *** [file.o] Error 1
Signed-off-by: Bjørn Forsman <bjorn.forsman@gmail.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Gustavo Zacarias [Thu, 10 Feb 2011 14:19:09 +0000 (11:19 -0300)]
target gcc: fix recompilation issue
gcc_target wants uclibc_target which is now first stage uclibc (not
real, from the NPTL build stage addition).
This triggers gcc-final to be rebuilt (albeit cached) and reinstalled,
thus touching all the stamps around and triggering a rebuild of
gcc_target.
Switch to $(STAMP_DIR)/gcc_libs_target_installed instead so that we only
depend on the last stage of cross gcc being installed to rebuild.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The FindQt4 module of CMake insists on having uic, even if the QtGui
module isn't part of the requirements to build the program. This isn't
correct, as Qt doesn't build/install the uic program when QtGui is
disabled (uic is used to generate some UI code).
This has been fixed upstream in
http://cmake.org/gitweb?p=cmake.git;a=commit;h=43cb9b8276a70d153d56a69d5c61daaf2bc51b78. This
commit will be part of the upcoming 2.8.4 CMake release, but in the
mean time, let's include it in Buildroot.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The Qt build system installs all host programs (uic, rcc, moc, qmake,
etc.) in the same base directory as the headers and libraries, but
that's not great for Buildroot.
So we now move the host programs and the mkspecs files to the
$(HOST_DIR), and create a qt.conf, which tells qmake where everything
is located.
Tested with Qmake and Cmake as build system for a simple Qt
application.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com> Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
[Peter: add tidsp-binaries to _DEPENDENCIES] Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com> Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com> Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
We already pass the LD variable to openssl in order to use gcc as the
driver for the link process, instead of directly using the ld
linker. However, we were not passing LDFLAGS so that the compiler
flags are passed, which means that with multilib toolchains, the
incorrect library variant could be used at link time, leading to
invalid binaries (partly ARMv4, partly ARMv5) or broken compilation
(when the build took place in soft-float, but the link stage takes
place against hard-float libraries).
This fixes a problem reported on IRC by amo-ej1 when compiling ssh on
PowerPC e500v2 with a CodeSourcery toolchain ("crtbegin.o uses hard
float, sshd uses soft float").
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Mike Frysinger [Mon, 7 Feb 2011 05:49:12 +0000 (00:49 -0500)]
gdb: add support for Blackfin gdbserver
This requires removing "deprecated" markings from gdb-6.6, but this isn't
that big of a deal. That is the last version with Blackfin support at the
moment and we're in the process of getting mainlined.
[Peter: only mark as undeprecated on bfin] Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Mike Frysinger <vapier@gentoo.org> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Mike Frysinger [Mon, 7 Feb 2011 05:49:11 +0000 (00:49 -0500)]
initial support for Blackfin processors
[Peter: don't allow MMU on bfin] Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Mike Frysinger <vapier@gentoo.org> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Mike Frysinger [Mon, 7 Feb 2011 05:49:10 +0000 (00:49 -0500)]
debugging: do not require no stripping
The stripping options operate on the final image and not the intermediate
builds, so requiring stripping to be disabled just to enable debugging
options doesn't make much sense. Especially when working with gdbserver:
only the host needs the debugging information to be available. The board
can run & debug perfectly fine without it.
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Mike Frysinger <vapier@gentoo.org> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Bjørn Forsman [Sun, 6 Feb 2011 19:02:04 +0000 (20:02 +0100)]
docs: fix $(PKG)_SITE_METHOD names (UPPER CASE => lower case)
Currently, the documentation says $(PKG)_SITE_METHOD should have upper
case names such as SVN or GIT, but the implementation only accepts lower
case svn or git. This fixes the documentation.
Signed-off-by: Bjørn Forsman <bjorn.forsman@gmail.com> Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Daniel Nyström <daniel.nystrom@timeterminal.se> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Mark Wickham [Fri, 4 Feb 2011 18:45:46 +0000 (19:45 +0100)]
python: fix modules_lib_dirs typo in patch-2.7-005-stagng-headers-libs.patch
Closes #3169
Typo in modules_lib_dirs section of patch specified modules_include_dirs
instead of modules_lib_dirs. This matters if PYTHON_MODULES_LIB
is not passed into the script.
Signed-off-by: Mark Wickham <markw@digi.com> Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Peter Korsgaard [Thu, 3 Feb 2011 19:44:45 +0000 (20:44 +0100)]
bzip2: don't hide on !BR2_PACKAGE_BUSYBOX_SHOW_OTHERS
Quite a lot of packages select bzip2 (for its libbz2 library), so don't
hide it even though busybox DOES have a bzip2 applet - Otherwise
kconfig complains.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Allow the user to define HOST_DIR in the config menu.
This way when building an internal toolchain a separate (maybe shared)
output directory may be defined and the toolchain can be used by
multiple users and/or projects.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar> Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>