Gustavo Zacarias [Fri, 24 Oct 2014 22:09:35 +0000 (19:09 -0300)]
coreutils: bump to version 8.23
Add hash file. patches 01 & 02 are now upstream so remove them.
Patch 01 isn't required any more (no posix_spawn usage).
Patch 02 is upstream.
Patch 03 retooled for 8.23 and renamed to 01.
Eric Benard [Mon, 20 Oct 2014 12:46:15 +0000 (14:46 +0200)]
mx53loco: update defconfig
- update bootloader to 2014.07 with fslc patches
- update kernel to 2.6.35_maintain branch which contains latest fixes
- defconfig was then generated by savedefconfig which suppressed and
reordered some items which explains the number of lines changed.
Francois Perrad [Fri, 17 Oct 2014 18:48:01 +0000 (20:48 +0200)]
intltool: fix build issues
This commit fixes various build failures caused by the host-perl
series.
Currently, the variables PERL and PERL5LIB are available only during
the configure step of host-intltool, but they are also needed when
running host-intltool, in all packages that depend on
host-intltool. Without them, host-intltool cannot work as it doesn't
find the libxml-parser-perl module installed in
$(HOST_DIR)/usr/lib/perl.
This commit therefore makes the PERL and PERL5LIB variables global, so
that all packages can access them.
Gustavo Zacarias [Mon, 20 Oct 2014 18:26:52 +0000 (15:26 -0300)]
package/eudev: tweak initscript
Make the udev initscript trigger in a coherent way, i.e. first set
subsystems, then device and then let the whole thing settle.
Otherwise for usb_modeswitch udev rules they never kick in since the
storage aspect (device) gets claimed first and the switch ignored.
Also set the settle timeout to a smaller value than the default
120 seconds to avoid stalling too much.
qt5connectivity selects the bluez-utils package but the dependencies are
not fully propagated so someone could do a static build, then select the
qt5connectivity package causing bluez-utils being auto-selected, which
is disabled for static builds, and then obtain a build failure like this
one:
checking for dlopen in -ldl... no
configure: error: dynamic linking loader is required
Yann E. MORIN [Mon, 20 Oct 2014 20:23:36 +0000 (22:23 +0200)]
infra/pkg-kconfig: fix saving the config file
When saving (aka updating) the configuration file of a kconfig package,
a subsequent call to "make" would rebuild the package, even though the
configuration did not actually change.
It took quite a while to understand why. But the reason is so simple and
obvious, when you think about it:
- $(@D)/.config depends on the config file $(BUSYBOX_CONFIG_FILE)
- busybox-update-config then copies $(@D)/.config back to the config
file $(BUSYBOX_CONFIG_FILE)
- so the config file is newer than $(@D)/.config
So, in the next run, pkg-kconfig believes that it has to rebuild
busybox. Bummer... :-/
So, the fix is very trivial, and just requires asking "cp" to preserve
timestamps, so the dependency does not kick in at the next run.
ngrep: update include path of libpcap so bpf.h can be found
The libpcap headers have moved from /usr/include to /usr/include/pcap,
although /usr/include/pcap.h still exists for backwards-compatibility
(but is a stub that simply includes pcap/pcap.h).
The ngrep package tries to find bpf.h based on the include statements in
pcap.h. If we point ngrep to /usr/include/pcap.h, bpf.h will not be
found and the associated functionality will not work.
The fix is to pass the new include path to the ngrep configure step.
Gustavo Zacarias [Tue, 21 Oct 2014 14:26:36 +0000 (11:26 -0300)]
scripts/mkusers: allow the creation of groups alone
Let mkusers create groups alone, useful for supplementary permissions in
udev/systemd for example where users can be added to later at runtime.
Use a magic string "-" to signal that user creation should be skipped.
Thomas Petazzoni [Sun, 19 Oct 2014 20:07:35 +0000 (22:07 +0200)]
ffmpeg: remove dangling reference to BR2_arm10
BR2_arm10 has been removed in commit d60489a6e5e733a20d18b128646088981da57867 ("arch: remove BR2_arm10t"),
so there is no point in keeping a reference to it in the ffmpeg
package.
Thomas Petazzoni [Fri, 17 Oct 2014 13:34:12 +0000 (15:34 +0200)]
pkg-stats: add statistics about hash files
As discussed during the Buildroot meeting, this commit extends the
pkg-stats script to include statistics about the number of packages
having vs. not having the hash file.
As of today, we have 104 packages with the hash file, and 1274
packages without.
Gustavo Zacarias [Sun, 19 Oct 2014 13:38:08 +0000 (10:38 -0300)]
tar: bump to version 1.28
Also add hash file and set eprefix to / otherwise when busybox is around
we've got duplicate tar applications (/bin/tar for busybox, /usr/bin/tar
for full-blown) normally making busybox still win in directory search.
[Thomas: slightly improve the comment that explains why we pass
--exec-prefix.]
Bernd Kuhls [Sun, 19 Oct 2014 09:59:01 +0000 (11:59 +0200)]
infra: Move --enable/--disable-debug to package/Makefile.in
A lot of packages ignored BR2_ENABLE_DEBUG. This patch simplifies the handling of
this option by adding the corresponding configure option to the global Makefile
for target packages.
For host packages --disable-debug is added to the global Makefile.
Karoly Kasza [Sun, 19 Oct 2014 13:30:08 +0000 (15:30 +0200)]
package/cdrkit: download from alternate site
www.cdrkit.org looks dead, the cdrkit archive always downloads from the
buildroot.org mirror. This patch points the downloader to the Debian
archive site.
Nathaniel Roach [Sun, 19 Oct 2014 10:10:29 +0000 (18:10 +0800)]
bandwidthd: Version bump to handle indirect deps properly
Thanks to Romain Naour, Baruch Siach and Thomas Petazzoni, I've
solved the build issues in certain cases for bandwidthd.
The issue was that libpcap and libpng would depend on other libraries
and upon testing for pcap/libpng the tests would fail due to not
being able to find the other libraries.
Bandwidthd's configure script now uses pkg-config for libpng
and pcap-config for libpcap. pcap-config's path is specified
in configure's arguments.
Some other tweaks have been made including the layout and flow
of configure.ac is a bit more logical, and the handling of postgres
and sqlite has been tweaked. Finally, the test for X.Org has been
removed.
[Thomas: re-add Nathaniel SoB line, which was misplaced in the commit
log, and therefore dropped when applying the patch.]
Thomas Petazzoni [Sun, 19 Oct 2014 11:10:58 +0000 (13:10 +0200)]
avahi: do not install S05avahi-setup.sh in systemd setup
The AVAHI_INSTALL_AUTOIPD was unconditionally installing
S05avahi-setup.sh to /etc/init.d/, even in systemd
configurations. Therefore, this commit:
* Moves the installation of S05avahi-setup.sh to a separate variable
called AVAHI_INSTALL_AUTOIPD_INIT_SYSV.
* Makes sure to use 'install -D' and a full destination path when
installing S05avahi-setup.sh.
* Moves the definition of AVAHI_INSTALL_AUTOIPD inside the
BR2_PACKAGE_AVAHI_AUTOIPD conditional, so that both the variable
definition and its addition to AVAHI_POST_INSTALL_TARGET_HOOKS are
enclosed in the conditional.
* Moves the S50avahi-daemon installation from AVAHI_INSTALL_INIT_SYSV
to AVAHI_INSTALL_DAEMON_INIT_SYSV.
* Uses 'install -D' to install S50avahi-daemon.
* Adds a AVAHI_INSTALL_INIT_SYSV definition which calls both
AVAHI_INSTALL_AUTOIPD_INIT_SYSV and AVAHI_INSTALL_DAEMON_INIT_SYSV.
Note: in a systemd configuration, there is no provision to do what
S05avahi-setup.sh is doing for sysv init configurations. Maybe this is
something that our systemd people should have a look at.
Thomas Petazzoni [Sun, 19 Oct 2014 11:01:04 +0000 (13:01 +0200)]
avahi: remove useless 'mkdir -p' in AVAHI_INSTALL_INIT_SYSTEMD
In AVAHI_INSTALL_INIT_SYSTEMD, since we're using 'install -D' to
install /usr/lib/tmpfiles.d/avahi.conf in $(TARGET_DIR), there's no
need to have a 'mkdir -p' before that, since 'install -D' creates the
directories as needed.
André Erdmann [Fri, 17 Oct 2014 19:10:42 +0000 (21:10 +0200)]
package/avahi: use upstream-provided systemd files
When using the buildroot-provided avahi-daemon.service file, bootup never
finishes, because multi-user.target is waiting for avahi-daemon to exit,
which is caused by "Type=oneshot" (in avahi-daemon.service).
Upstream's systemd files get already installed to /lib/systemd.
They're not an exact copy of S50avahi-daemon, but work flawlessly,
so use these units:
* avahi-daemon.service, auto-enabled (ln -fs in AVAHI_INSTALL_INIT_SYSTEMD)
* avahi-daemon.socket,
not auto-enabled, but a dependency of avahi-daemon.service
* avahi-dnsconfd.service, auto-enabled
[Thomas: use simpler absolute paths for the symbolic links instead of
relative paths. Suggested by Maxime Hadjinlian.]
Karoly Kasza [Sat, 18 Oct 2014 18:23:59 +0000 (20:23 +0200)]
fs/iso9660: add dependencies for make source
The "iso image" rootfs target is a special one, as it does not use the
ROOTFS_TARGET infrastructure. The absence of ROOTFS_*_DEPENDENCIES
variable makes "make source" to skip this target's dependencies
(namely host-cdrkit and it's children) obstructing an offline build.
[Thomas: add the rootfs-iso9660-show-depends target, so that 'make
graph-depends' doesn't break when iso9660 is selected.]
This installs the 'modplug123' command line tool to play tracker music
files (MOD, S3M, XM etc.) using libmodplug.
The modplugtools package can build two players: modplugplay that plays
directly via OSS and modplug123 that plays via libao. Since modplugplay
won't work without OSS support and modern kernels are likely to have no
OSS or OSS emulation, we build only modplug123.
If you have a system without ALSA support for some reason, you can
still use modplug123 to play over OSS via libao.
http://modplug-xmms.sourceforge.net/
[Thomas: add missing Config.in comment about the C++ dependency.]
Jörg Krause [Fri, 17 Oct 2014 22:36:33 +0000 (00:36 +0200)]
package/.mk files: remove --localstatedir=/var from autotools packages
Remove --localstatedir=/var from all autotools packages where it is no longer
needed.
Also remove --localstatedir=/var/lib/dhcp from package dhcp. localstatedir is
used by dhcp to set the default directory for the leases files. This can also
be done by setting --with-*-lease-file=/var/lib/dhcp/*, which is done in
dhcp.mk.
A custom --localstatedir is left in:
* proftpd.mk
* mysql.mk
This is safe to do:
One of the good thing with autoconf is that if you pass:
--localstatedir=/var ... --localstatedir=/var/something
Then /var/something will be used. So, we can set --localstatedir=/var
by default in the infrastructure, and still have certain packages doing
weird things override it. [Thanks to Thomas Petazzoni]
Jörg Krause [Fri, 17 Oct 2014 22:36:32 +0000 (00:36 +0200)]
package/pkg-autotools.mk: set --localstatedir=/var by default
`localstatedir` is a common autoconf variable like prefix and sysconfdir. At
this time 25 autotools and one generic package uses this Makefile variable. 22
of the autotools packages use --localstatedir=/var, the remaining use custom
settings.
Setting `localstatedir` to /var is consistent with the Filesystem Hierarchy
Standard (FHS). So it's time to move it as a common definition in the
autotools infrastructure.
Gustavo Zacarias [Sat, 18 Oct 2014 12:30:53 +0000 (09:30 -0300)]
libxml2: explicitly state zlib dir
Pass the zlib directory prefix to --with-zlib otherwise it expands into
absolute (distribution) directories leading to bad test results in
configure. Fixes:
http://autobuild.buildroot.net/results/303/303d6d165c5423944122bf430575c018479733f8/build-end.log
As stated by Maarten ter Huurne on the mailing list, the install of the
asound.h headers overwrite the one from the toolchains or kernel
headers.
Since tinyals's header is only a subset of the 'real' asound.h, there is
no need to overwrite it.
projects / buildroot-mgba.git / log