Maciej Borzecki [Thu, 14 Aug 2014 11:34:24 +0000 (13:34 +0200)]
cml1.bbclass: fix diffconfig diff command
Commit 6a7b98993350d0d24eae0058ae26ae19cfdf7c4c introduced a typo in
command, a whitespace was missing between --new-line-format and path to
original file. As a result, diffconfig task did not work, diff failed
with exit status 2, resulting in empty fragment.cfg.
Signed-off-by: Maciej Borzecki <maciej.borzecki@open-rnd.pl> Signed-off-by: Maciek Borzecki <maciek.borzecki@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Thu, 14 Aug 2014 14:26:53 +0000 (15:26 +0100)]
qemu: Add PACKAGECONFIG for glx
With the configuration for sdl and glx, the virtual/libx11 in DEPENDS
no longer makes any sense at all, the dependency comes through the
various graphics backends. Therefore drop it. The glx PACKAGECONFIG
is added for completeness. We avoid a floating dependency on mesa.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Thu, 14 Aug 2014 14:25:52 +0000 (15:25 +0100)]
qemu: Add alsa PACKAGECONFIG and cleanup native/nativesdk DEPENDS
Converting this dependency to become a PACKAGECONFIG makes sense. Since
there is a "," in the configuration option, its not quite as straightforward
as normal.
Also clean up the native and nativesdk DEPENDS and set the PACKAGECONFIG
entries to match the old behaviour by default.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Thu, 14 Aug 2014 13:38:26 +0000 (14:38 +0100)]
site: Add common-darwin
We need this to ensure there are some correct values when building with
meta-darwin. This matches the addition of a common-mingw site file in
the core.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Thu, 14 Aug 2014 13:34:42 +0000 (14:34 +0100)]
glib-2.0: Allow correct printf configuration on darwin/mingw
mingw and darwin builds need to change the value of printf from no
to yes. Whilst you can do this from a bbappend, its currently a bit ugly
and assumes knowledge of the other conf arguments. Fixing this in the glib
recipe directly is cleaner in this case.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Thu, 14 Aug 2014 13:33:33 +0000 (14:33 +0100)]
populate_sdk_base: Move toolchain installation script to a separate file
It makes sense to move the script to a separate file, making the bbclass
clearer and allowing the end user to more easily customise or replace
it. There are no functionality changes, just the addition of some
substitution variables.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Thu, 14 Aug 2014 13:31:40 +0000 (14:31 +0100)]
classes: Combine rootfs_<pkgtype> and populate_sdk_<pkgtype>
There is no real reason to have these separate files any longer. It does
mean in meta-toolchain type recipes some extra variables are defined but
it also means the common code can be grouped and maintained together
which I believe is more beneficial. We therefore merge the classes.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Martin Jansa [Mon, 11 Aug 2014 14:09:13 +0000 (16:09 +0200)]
SIGGEN_EXCLUDERECIPES_ABISAFE: add initscripts
* update-rc.d now adds initscripts dependency even to allarch recipes
making them effectively TUNE_PKGARCH, "fix" it by excluding
initscripts from all signatures
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Adds 'docs' (for html doc generation) and 'api-docs' (for API doc generation)
configurations and leaves them both disabled by default. This avoids
autodetected dependency upon docutils.
Fixes [YOCTO #6530]
Signed-off-by: Christopher Larson <chris_larson@mentor.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Martin Jansa [Sun, 10 Aug 2014 18:41:53 +0000 (20:41 +0200)]
sstate-sysroot-cruft.sh: Improve to use it from CI
* strip tmpdir prefix, so that we have shorter paths which aren't
builder specific
* use '#' for regexp delimiter so that we don't need to prefix
forward slashes in paths
* extend default whitelist to cover typical cases
* add parameter for external whitelist file
* use number of found paths as return code, so that CI can easily
report error when new untracked files are found
* use .txt suffix for all output files, so that they can be easily
viewed in browser
* add populate_sysroot task, because somewhere between dora and daisy
the populate-sysroot files in sstate-control were renamed to have
underscore instead of dash
* only few entries not covered by this default whitelist were found
in world build (but I'll leave these for people to whitelist, because
they are not generated in most builds)
* [^/]*/home/builder
home directory from meta/recipes-graphics/builder/builder_0.1.bb
* [^/]*/usr/src/kernel/patches
* [^/]*/usr/lib/gdk-pixbuf-2.0/.*/loaders.cache
3 places are using this, not sure which one creates it
meta/recipes-gnome/gdk-pixbuf/gdk-pixbuf_2.30.8.bb:
GDK_PIXBUF_MODULE_FILE=${STAGING_LIBDIR_NATIVE}/gdk-pixbuf-2.0/${LIBV}/loaders.cache
meta/recipes-gnome/gtk+/gtk-update-icon-cache-native_3.4.4.bb:
GDK_PIXBUF_MODULE_FILE=${STAGING_LIBDIR_NATIVE}/gdk-pixbuf-2.0/2.10.0/loaders.cache
scripts/postinst-intercepts/update_pixbuf_cache:
>$GDK_PIXBUF_MODULEDIR/../loaders.cache && \
sed -i -e "s:$D::g" $GDK_PIXBUF_MODULEDIR/../loaders.cache
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Marek Vasut [Thu, 7 Aug 2014 15:25:46 +0000 (17:25 +0200)]
qt4: Avoid duplicate flags in the g++-unix.conf
The removed flags from this g++.conf file, which is installed to the
target as g++-unix.conf are added by gcc-base.conf . Do not add those
flags twice.
Signed-off-by: Marek Vasut <marex@denx.de> Cc: Eric Bénard <eric@eukrea.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Yang [Thu, 7 Aug 2014 13:40:55 +0000 (06:40 -0700)]
trace-cmd: fix recompile error
Fixed:
make: *** No rule to make target `/path/to/sysroot/4.9.0/include/stddef.h', needed by `parse-events.o'. Stop.
make: *** Waiting for unfinished jobs....
ERROR: oe_runmake failed
This happens when upgrade gcc from 4.9.0 to 4.9.1, and the
.parse-events.d isn't regenerated when recompile, the content of it are:
Martin Jansa [Sat, 9 Aug 2014 16:07:59 +0000 (18:07 +0200)]
libarchive: add PACKAGECONFIG for nettle
* fixes following floating dependencies:
libarchive/libarchive/latest lost dependency on nettle
libarchive/libarchive-bin/latest lost dependency on libxml2 nettle
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Martin Jansa [Sat, 9 Aug 2014 16:07:58 +0000 (18:07 +0200)]
mpeg2dec: add PACKAGECONFIG for x11 and fix dependencies
* the configure script checks for Xext and Xv when X/libvo is enabled
* fixes following warnings:
WARN: mpeg2dec: mpeg2dec rdepends on libxext but it isn't a build dependency?
WARN: mpeg2dec: mpeg2dec rdepends on libxv but it isn't a build dependency?
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Tom Zanussi [Fri, 8 Aug 2014 20:56:55 +0000 (15:56 -0500)]
wic: Bump the version to 2.0
This represents a complete rework, and if we assume the previous
version should have been 1.0 (instead of a silly .1x version), then
this should be 2.0.
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Tom Zanussi [Fri, 8 Aug 2014 18:43:45 +0000 (13:43 -0500)]
wic: Update/rename configure-related code
The wic code inherited a basic image-creation flow based on
image-configuration, but wic doesn't actually configure anything, so
rename parts of the code dealing with configuration to something more
appropriate.
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Tom Zanussi [Fri, 8 Aug 2014 18:28:54 +0000 (13:28 -0500)]
wic: Update/rename install-related code
The wic code inherited a basic image-creation flow based on installing
packages, but wic doesn't actually install anything, so rename parts
of the code dealing with installing to something more appropriate.
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Tom Zanussi [Fri, 8 Aug 2014 15:50:07 +0000 (10:50 -0500)]
wic: Update cleanup/unmount-related code
The wic code inherited a basic image-creation flow based on mounting
loop devices, but wic doesn't actually mount or unmount anything, so
get rid of unmount() and consolidate whatever it did do with
cleanup().
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Tom Zanussi [Fri, 8 Aug 2014 15:09:12 +0000 (10:09 -0500)]
wic: Update/rename/delete mount-related code
The wic code inherited a basic image-creation flow based on mounting
loop devices, but wic doesn't actually mount anything, so rename parts
of the code dealing with mounting to something more appropriate, and
remove related unused code.
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Tom Zanussi [Sun, 3 Aug 2014 23:19:29 +0000 (18:19 -0500)]
wic: Remove proxy support
wic doesn't use proxy support, so remove it.
Also remove anything related to proxies in misc and conf, and while
we're at it, remove all the obviously unneeded code from those files -
it's easier to just remove it than to figure out the callchain to the
proxy code usages.
Basically the extra stuff relates to packaging, images, and config
files we don't use.
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Tom Zanussi [Thu, 31 Jul 2014 18:55:24 +0000 (13:55 -0500)]
wic: Make exec_cmd() error out instead of warn
The reason exec_cmd() warns but doesn't error out (broken parted)
doesn't really make sense, since the parted invocations don't even use
exec_cmd(). It really should just fail since by not doing so it's
actually enabling invalid images in some cases.
Also, since the return code is now always zero, there's no point in
having a return code, so remove it. This represents a change in the
API, so we also need to update all callers.
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Cristian Iorga [Wed, 6 Aug 2014 07:43:25 +0000 (10:43 +0300)]
qemu: upgrade to 2.1
QEMU 2.1 comes with fixes and improvements.
See http://wiki.qemu.org/ChangeLog/2.1 for details.
- Added config for quorum support, depending on gnutls.
- pcie_better_hotplug_support.patch removed,
integrated upstream.
- Qemu-Arm-versatilepb-Add-memory-size-checking.patch updated
to 2.1 source code.
- no-strip.patch removed, no longer necessary due to code changes.
Signed-off-by: Cristian Iorga <cristian.iorga@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Yasir-Khan [Fri, 8 Aug 2014 15:41:08 +0000 (20:41 +0500)]
trace-cmd: Update to version 2.3.2
Current trace-cmd version 1.2 throws "recorder error in
splice input" error while recording a trace on kernel
version 3.10. Update the trace-cmd version to latest 2.3.2
to resolve this error and drop the redundant patches.
Max Eliaser [Mon, 28 Jul 2014 17:13:26 +0000 (10:13 -0700)]
bootchart2: create recipe for bootchart2
This recipe creates packages for the bootchart2 system-wide profiler daemon
and related utilities. It fetches the Git revision immediately past the one
corresponding to the 0.14.6 release of bootchart2. (0.14.6 had a systemd-
related bug that was corrected right after it was tagged.)
The recipe contains three packages:
* bootchart2 - The daemon itself.
* pybootchartgui - Python program to visualize and display the data
collected by bootchart2 or compatible daemons such as the original
bootchart.
* bootchartd-stop-initscript - A SysV init script to stop data collection
when booting completes.
Depending on how you wish to use bootchart2, you may not end up having all
three of those packages installed on your image.
There is also a bootchart2-native variant, which is intended solely to provide
a native version of the pybootchartgui utility. The non-cross-compiled version
of the bootchart2 daemon has not been tested at all, don't use it.
The recipe is extensively documented. Read the comments at the beginning of
bootchart2_git.bb or else you'll have no idea how to use it.
This recipe is based on a recipe from meta-WebOS. The WebOS people had some
extra code (including patches against the bootchart2 code) to support the
Upstart init system. However, since upstream Poky does not support Upstart,
that stuff is being left behind. The WebOS people can write a bbappend to re-
add it.
Original recipe written by Wonhong Kwon of LG. Upstreamed as part of the
solution to [YOCTO #5893].
Signed-off-by: Max Eliaser <max.eliaser@intel.com> Signed-off-by: Saul Wold <sgw@linux.intel.com>
Chen Qi [Thu, 7 Aug 2014 06:58:12 +0000 (02:58 -0400)]
v86d: fix for systemd to load uvesafb module correctly
The /etc/init.d/fbsetup script doesn't have any effect in a systemd
image. Its purpose is to load the uvesafb kernel module at boot.
This functionality could be achieved by adding a configuration file
under /etc/modules-load.d/ directory which would be parsed by the
systemd-modules-load.service.
Robert Yang [Thu, 7 Aug 2014 03:05:22 +0000 (20:05 -0700)]
syslinux: fix reinstall error
Fixed:
make: *** No rule to make target `/path/to/sysroot/4.9.0/include/stdarg.h', needed by `cpio.o'. Stop.
make: *** Waiting for unfinished jobs....
ERROR: oe_runmake failed
This happens when upgrade gcc from 4.9.0 to 4.9.1, and the .cpio.o.d isn't
regenerated when recompile (the compile happens when do_install), the content
of it are:
Robert Yang [Thu, 7 Aug 2014 03:05:22 +0000 (20:05 -0700)]
u-boot-mkimage: fix recompile error
Fixed:
make: *** No rule to make target `/path/to/sysroot/4.9.0/include/stddef.h', needed by `crc32.o'. Stop.
make: *** Waiting for unfinished jobs....
ERROR: oe_runmake failed
This happens when upgrade gcc from 4.9.0 to 4.9.1, and the .depend isn't
regenerated when recompile, the content of it are:
Robert Yang [Thu, 7 Aug 2014 03:05:22 +0000 (20:05 -0700)]
kernelshark: fix recompile error
Fixed:
make: *** No rule to make target `/path/to/sysroot/4.9.0/include/stddef.h', needed by `parse-events.o'. Stop.
make: *** Waiting for unfinished jobs....
ERROR: oe_runmake failed
This happens when upgrade gcc from 4.9.0 to 4.9.1, and the
.parse-events.d isn't regenerated when recompile, the content of it are:
Scot Salmon [Mon, 4 Aug 2014 15:40:30 +0000 (10:40 -0500)]
sysvinit: allow stack size configuration from rcS
For certain swap/overcommit settings (e.g. when overcommit is disabled
on a real-time system), we need to limit the stack size used by
initscripts. When the STACK_SIZE environment variable is set (usually
in /etc/default/rcS), ulimit the stack size to the value specified.
Make the stack size ulimit a soft limit, which allows the user to
increase the stack size where required without having to run the
respective application as root.
Robert Yang [Wed, 6 Aug 2014 15:43:55 +0000 (08:43 -0700)]
boot-directdisk.bbclass: use local HDDDIR
Use $HDDDIR rather than ${HDDDIR} in build_boot_dd(), otherwise may
errors:
- Set these in local.conf:
IMAGE_FSTYPES += "live"
IMAGE_FSTYPES += "vmdk"
AUTO_SYSLINUXMENU = "1"
NOHDD = "1"
$ bitbake core-image-sato
DEBUG: Executing shell function build_boot_dd
install: cannot create regular file
`/path/to/core-image-sato-1.0/hddimg//vesamenu.c32': No such file or directory
WARNING: exit code 1 from a shell command.
This because it uses the ${HDDDIR} which is set in bootimg.bbclass, use
local HDDDIR which is set in build_boot_dd() will fix the problem.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Saul Wold <sgw@linux.intel.com>
