]> code.ossystems Code Review - openembedded-core.git/log
openembedded-core.git
13 years agopseudo: PR bump.
Lianhao Lu [Mon, 23 Apr 2012 10:10:43 +0000 (18:10 +0800)]
pseudo: PR bump.

Bump PR value due to the commit
c6c701f424aeb502d20ff02d02712e56f4e259a5.

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoscripts/hob: disable sanity checks when launching
Joshua Lock [Mon, 23 Apr 2012 21:36:10 +0000 (14:36 -0700)]
scripts/hob: disable sanity checks when launching

This enables us to use the GUI to change any settings which might cause
sanity checks to fail, such as the proxy configuration.

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agosanity.bbclass: add variable to disable the sanity checks
Joshua Lock [Mon, 23 Apr 2012 21:36:09 +0000 (14:36 -0700)]
sanity.bbclass: add variable to disable the sanity checks

It's useful for Hob to be able to disable the sanity checks completely
without marking them as passed so that the user can get into the GUI to
configure their settings, etc.

Add a variable, DISABLE_SANITY_CHECKS, to do so.

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopython: Fix various contamination issues leading to broken/missing c modules
Richard Purdie [Mon, 23 Apr 2012 21:46:42 +0000 (22:46 +0100)]
python: Fix various contamination issues leading to broken/missing c modules

The move of libcrypto to /lib instead of /usr/lib has broken the _hashlib module
compilation. There were also a number of other failing modules which should
have been building correctly. This turned out partly to be the /lib issue
but also due to a number of native paths creeping into compiler commandlines.

These changes add in /lib as part of the searh directory and remove
a number of host contamination issues within setup.py. Post release we
should really further go through this file and just delete large sections
of it as its hard to be sure what strange paths python is injecting as
search paths.

This patch also fixes issues where re-execution of the compile task
would corrupt the Makefile in various ways, again leading to puzzling
paths within the configuration.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agobitbake.conf: Add a STAGING_BASELIBDIR variable that recipes can use to find base_libdir
Richard Purdie [Mon, 23 Apr 2012 21:45:36 +0000 (22:45 +0100)]
bitbake.conf: Add a STAGING_BASELIBDIR variable that recipes can use to find base_libdir

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoself-hosted-image: Update poky revision to point at the 1.2 release branch
Richard Purdie [Wed, 18 Apr 2012 14:58:45 +0000 (15:58 +0100)]
self-hosted-image: Update poky revision to point at the 1.2 release branch

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoinitramfs-live-boot: Disable unionfs until its issue with the system rootdir are...
Richard Purdie [Sun, 22 Apr 2012 14:47:30 +0000 (15:47 +0100)]
initramfs-live-boot: Disable unionfs until its issue with the system rootdir are resolved

There are issues with the current unionfs when making a union mount over "/".
Until these are resolved we can't use unionfs for live booting so disable this
temporarily as a workaround.

unionfs is usable in other circumstances.

[YOCTO #2331 workaround]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopseudo: Drop nativesdk wrapper and link against old memcpy symbol
Richard Purdie [Sun, 22 Apr 2012 14:46:17 +0000 (15:46 +0100)]
pseudo: Drop nativesdk wrapper and link against old memcpy symbol

The -nativesdk pseudo wrapper setting LD_LIBRARY_PATH turned out to be a
bad idea since it can mix up different libc and lib-dl verisons which
may or may not work depending on the phase of the moon.

As an alternative to solving the original problem, this patch drops the
symbol version requirement on memcpy which allows pseudo to work with
libc's back to 2.7 which should be sufficient for our supported targets
using nativesdk.

[YOCTO #2299]
[YOCTO #2351]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopseudo: Fix bashisms
Richard Purdie [Thu, 19 Apr 2012 22:01:09 +0000 (23:01 +0100)]
pseudo: Fix bashisms

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopackage.bbclass: Ensure kernel modules get stripped
Richard Purdie [Thu, 19 Apr 2012 12:17:57 +0000 (13:17 +0100)]
package.bbclass: Ensure kernel modules get stripped

Kernel modules are not marked as executable but we do expect to strip them.
This patch adds in missing code to ensure we do this. Without this images
are getting sigificantly bloated in size.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogettext.bbclass: Ensure we don't overwrite other DEPENDS_GETTEXT values
Richard Purdie [Thu, 19 Apr 2012 07:48:25 +0000 (08:48 +0100)]
gettext.bbclass: Ensure we don't overwrite other DEPENDS_GETTEXT values

In particular, this overwrites the value from cross-canadian.bbclass in
some cases which isn't the desired behaviour and unnecessarily
complicates/breaks the dependency chain.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoqemu: Backport a patch to solve SSE2 instruction emulation issues
Richard Purdie [Thu, 19 Apr 2012 07:44:11 +0000 (08:44 +0100)]
qemu: Backport a patch to solve SSE2 instruction emulation issues

This fix addresses various issues seen in qemux86-64 images:
 * scroll bars in matchbox-terminal not working
 * files not appearing in pcmanfm
 * warnings on the console from glib/gobject about invalid gdouble values

Its due to an emulation issue in qemu which the backported patch fixes.

I managed to debug it to a specific function, Khem found the qemu patch
to backport, thanks Khem!

[YOCTO #1906]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoself-hosted-image: Update poky revision to point at the 1.2 release branch
Richard Purdie [Wed, 18 Apr 2012 14:58:45 +0000 (15:58 +0100)]
self-hosted-image: Update poky revision to point at the 1.2 release branch

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopackageinfo.bbclass: A workaround for RPM architecture renaming
Dongxiao Xu [Wed, 18 Apr 2012 12:48:16 +0000 (20:48 +0800)]
packageinfo.bbclass: A workaround for RPM architecture renaming

For beagleboard platform, the PACKAGE_ARCH for certain recipes is
"armv7a-vfp-neon", however, the architecture label in RPM file name is
"armv7a" due to a potential bug in RPM backend.

This commit is a workaround to make Hob work in this case.

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolsb: Fix PR bump, fix bashisms, correct dependency issue, tweak package description
Richard Purdie [Wed, 18 Apr 2012 11:06:48 +0000 (12:06 +0100)]
lsb: Fix PR bump, fix bashisms, correct dependency issue, tweak package description

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoclasses/sanity: import regular expression module in SuSE distro check
Paul Eggleton [Wed, 18 Apr 2012 09:56:10 +0000 (10:56 +0100)]
classes/sanity: import regular expression module in SuSE distro check

We call re.sub here, so we need to "import re" or an error occurs
(tested on OpenSuSE 12.1).

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoclasses/sanity: trim trailing newline when reading /etc/redhat-release
Paul Eggleton [Wed, 18 Apr 2012 09:56:09 +0000 (10:56 +0100)]
classes/sanity: trim trailing newline when reading /etc/redhat-release

Any trailing whitespace needs to be stripped before comparing it to the
SANITY_TESTED_DISTROS list.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolibproxy: remove dependency on gconf when x11 not in DISTRO_FEATURES
Paul Eggleton [Tue, 17 Apr 2012 23:18:35 +0000 (00:18 +0100)]
libproxy: remove dependency on gconf when x11 not in DISTRO_FEATURES

This allows libzypp to successfully build when x11 is not in
DISTRO_FEATURES (avoiding the libzypp -> libproxy -> gconf -> gtk+
dependency chain which without X results in gtk+ failing at
do_configure).

Fixes [YOCTO #2320].

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoqemugl: make postinst run at first boot
Scott Garman [Tue, 17 Apr 2012 23:10:30 +0000 (23:10 +0000)]
qemugl: make postinst run at first boot

The previous postinst script was not working for images which use
RPM because do_rootfs could allow qemugl to be installed after
libgl1, and the postinst needs to run after libgl1 is installed.

Hence, it's being changed to run at first boot instead of during
do_rootfs.

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoself-hosted-image: remove BB_NO_NETWORK and update Poky SRCREV
Saul Wold [Tue, 17 Apr 2012 22:02:13 +0000 (15:02 -0700)]
self-hosted-image: remove BB_NO_NETWORK and update Poky SRCREV

Remove BB_NO_NETWORK because we can not guarentee that all
downloads will be available in the self-hosted image for
the possible targets being built.  If this image is run with
out networking, failures may be seen and BB_NO_NETWORK
should be set via the Hob

Update the Poky SRCREV so it has the latest set of fixes and
patches.

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agotask-self-hosted: Add LSB for distro info
Saul Wold [Tue, 17 Apr 2012 22:02:12 +0000 (15:02 -0700)]
task-self-hosted: Add LSB for distro info

The lsb package is needed to enable the distro checking
in the self hosted image. Currently there is no standard
identification in the self-hosted image so use the existing
lsb package which provides lsb_release that parses the
/etc/release file

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolsb: Add DISTRO Info to /etc/lsb-release file also add util-linux to DEPENDS
Saul Wold [Tue, 17 Apr 2012 22:02:11 +0000 (15:02 -0700)]
lsb: Add DISTRO Info to /etc/lsb-release file also add util-linux to DEPENDS

Add the DISTRO* fields to the /etc/lsb-release file so the distro
parsing code can find standard information about the Poky Distro
when using the Self-Hosted Image.

The lsb_release command uses getopt which is provided by util-linux

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolinux-yocto/meta: remove kernel config audit warnings
Bruce Ashfield [Tue, 17 Apr 2012 22:01:46 +0000 (18:01 -0400)]
linux-yocto/meta: remove kernel config audit warnings

Updating the meta SRCREVs to pickup the following meta change for the
3.0 and 3.2 kernels:

[
    meta: Clean up BSPs kernel config

    Clean up some QEMU and non-x86 BSPs kernel config, including

    qemuarm
    qemuppc
    qemux86
    beagleboard
    mpc8315e_rdb

    Only obsolete/invalid kernel configs are removed.

Signed-off-by: Yang Shi <yang.shi@windriver.com>
]

With this commit, the configuration audit for the qemu and hardware
reference boards is (largely) warning free.

Signed-off-by: Yang Shi <yang.shi@windriver.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolinux-yocto/meta: beagleboard: disable CONFIG_PREEMPT
Bruce Ashfield [Tue, 17 Apr 2012 22:01:45 +0000 (18:01 -0400)]
linux-yocto/meta: beagleboard: disable CONFIG_PREEMPT

Updating the meta SRCREV for both the 3.0 and 3.2 kernel trees to
pickup the beagleboard xM boot fix:

[
    meta/beagleboard: disable CONFIG_PREEMPT

    The boot hangs with the message:
    mmc0: error -110 whilst initialising SD card

    The MMC driver has issues initializing when PREEMPT is enabled (either forced
    or voluntary). Unplugging and then plugging the card back will reset the
    driver and continue booting. Alternatively, disable preemption.
]

[YOCTO: #1892]

Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolinux-yocto: .diff is a valid patch extension
Bruce Ashfield [Tue, 17 Apr 2012 22:01:44 +0000 (18:01 -0400)]
linux-yocto: .diff is a valid patch extension

In fixing an existing patch migration bug, the list of valid extensions
got out of sync from the core patch class. As a result, valid patches
were not being applied to the tree.

Updating the tools to migrate .diff files fixes the issue.

Also in this fix is the removal of .patch in the find_sccs() routine, since
it will never be returned by patch.bbclass when all non-patches are
requested, it is simply confusing.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agomatchbox-wm-2: Fix incorrect update-alternatives
Mark Hatle [Tue, 17 Apr 2012 22:05:52 +0000 (17:05 -0500)]
matchbox-wm-2: Fix incorrect update-alternatives

Update-alternatives was pointing to matchbox-session, and should have
been pointing to matchbox-window-manager.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolrzsz: sz, sx and sb were linked incorrectly
Mark Hatle [Mon, 16 Apr 2012 23:16:29 +0000 (18:16 -0500)]
lrzsz: sz, sx and sb were linked incorrectly

The sz, sx and sb links were created incorrectly to lrz, they should
point to lsz.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolibproxy: Allow to build in non-gplv3 configuration
Richard Purdie [Tue, 17 Apr 2012 13:06:56 +0000 (13:06 +0000)]
libproxy: Allow to build in non-gplv3 configuration

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogst-plugins-bad: Add missing PR bump after curl change
Richard Purdie [Tue, 17 Apr 2012 11:33:01 +0000 (12:33 +0100)]
gst-plugins-bad: Add missing PR bump after curl change

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoqemu: Disable curl explicitly, its not in DEPENDS
Richard Purdie [Tue, 17 Apr 2012 11:30:01 +0000 (11:30 +0000)]
qemu: Disable curl explicitly, its not in DEPENDS

There is an error when build meta-toolchain-sdk on Ubuntu 10.10:

| Installing NATIVESDK packages
| Processing task-sdk-host-nativesdk...
| Processing task-cross-canadian-mips...
| error: Failed dependencies:
|       libcurl.so.4 is needed by qemu-nativesdk-0.15.1-r5.i686
NOTE: package meta-toolchain-gmae-1.0-r7: task do_populate_sdk: Failed
ERROR: Task 21 (/buildarea2/lyang1/poky/meta/recipes-core/meta/meta-toolchain-gmae.bb, do_populate_sdk) failed with exit code '1'

This is because qemu would depend on curl if it happened to be
available at build time and could confuse the -native and -nativesdk
librbaries.

[YOCTO #2305]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agosanity.bbclass: If the sanity tests fail, don't mark them as complete
Richard Purdie [Tue, 17 Apr 2012 11:16:18 +0000 (11:16 +0000)]
sanity.bbclass: If the sanity tests fail, don't mark them as complete

If the sanity tests fail, we still were writing out the stamp which means
they'd get skipped the next time we run bitbake. This is clearly wrong
and we should only write out the stamp file if the sanity tests complete
successfully.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoperl: fix re-execution of compile task
Venkata ramana gollamudi [Tue, 17 Apr 2012 09:04:15 +0000 (09:04 +0000)]
perl: fix re-execution of compile task

After building perl package, re-execution of compile task recursively
 substitutes the path, making it an invalid path.

Fixed to prevent recursive substitution.

Similar case as [Yocto #2194]

Signed-off-by: Venkata Ramana Gollamudi <ramana.gollamudi@huawei.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoavahi-daemon: fix missing init script links
Andreas Oberritter [Tue, 17 Apr 2012 10:37:17 +0000 (12:37 +0200)]
avahi-daemon: fix missing init script links

* update-rc.d wasn't executed when the rootfs was built, because
  pkg_postinst_avahi-daemon exited early. It wasn't run either
  on first boot, because the exit code was 0.

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agobitbake.conf: Add PARALLEL_MAKE to BB_HASHBASE_WHITELIST
Richard Purdie [Mon, 16 Apr 2012 12:19:10 +0000 (13:19 +0100)]
bitbake.conf: Add PARALLEL_MAKE to BB_HASHBASE_WHITELIST

The value of PARALLEL_MAKE shouldn't affect sstate checksums. In general it
doesn't as the value is injected as a task override and the sstate checksum
doesn't track these. That isn't the case for linux-yocto in particular so
adding it the list of variables to ignore is useful.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolibzypp: add libproxy to DEPENDS
Saul Wold [Mon, 16 Apr 2012 22:08:36 +0000 (15:08 -0700)]
libzypp: add libproxy to DEPENDS

libzypp uses proxy.h to handle proxies

[YOCTO #2306]

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agodistrodata.bbclass: fix comment typo
Otavio Salvador [Wed, 11 Apr 2012 03:19:51 +0000 (03:19 +0000)]
distrodata.bbclass: fix comment typo

Fixes:

  - xf86-intput-synaptics-12.6.9.tar.gz
  + xf86-input-synaptics-12.6.9.tar.gz

Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoflex: backport Debian patches to fix generated code warnings
Otavio Salvador [Wed, 11 Apr 2012 03:19:52 +0000 (03:19 +0000)]
flex: backport Debian patches to fix generated code warnings

The generated parser had warnings regarding signess and return check
which makes Linux Kernel's perf tool from 3.4 release to fail without
those patches.

Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoconnman-gnome: fix segfault on connect
Paul Eggleton [Mon, 16 Apr 2012 21:37:16 +0000 (22:37 +0100)]
connman-gnome: fix segfault on connect

Fixes a segfault in the properties application when connecting to a WiFi
service succeeds.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoautomake-nativesdk: fix builds on fedora 17
Robert Yang [Mon, 16 Apr 2012 09:18:49 +0000 (17:18 +0800)]
automake-nativesdk: fix builds on fedora 17

Generally distros keep perl at /usr/bin/perl
Fedora 17 also has /bin/perl

this causes automake-nativesdk build on such distros to put perl
interpreter path in the perl scripts as /bin/perl

But we set perl location for target as /usr/bin/perl

This mismatch of perl path causes failure of rootfs image creation
like this:

| error: Failed dependencies:
|       /bin/perl is needed by automake-nativesdk-1.11.2-r2.x86_64
NOTE: package meta-toolchain-gmae-1.0-r7: task do_populate_sdk: Failed

The second sed command is for such a case:

eval 'case $# in 0) exec /bin/perl -S "$0";; *) exec /bin/perl -S "$0" "$@";; esac'

This line has two "/bin/perl" and we can't use a line number to tell
sed which line it is since the line numbers in different files are
defferent.

[YOCTO #2304]

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
13 years agoautoconf-nativesdk: fix builds on fedora 17
Robert Yang [Mon, 16 Apr 2012 09:15:58 +0000 (17:15 +0800)]
autoconf-nativesdk: fix builds on fedora 17

Generally distros keep perl at /usr/bin/perl
Fedora 17 also has /bin/perl

this causes autoconf-nativesdk build on such distros to put perl
interpreter path in the perl scripts as /bin/perl

But we set perl location for target as /usr/bin/perl

This mismatch of perl path causes failure of rootfs image creation
like this:

| error: Failed dependencies:
|       /bin/perl is needed by autoconf-nativesdk-2.68-r6.x86_64.rpm
NOTE: package meta-toolchain-gmae-1.0-r7: task do_populate_sdk: Failed

The second sed command is for such a case:

eval 'case $# in 0) exec /bin/perl -S "$0";; *) exec /bin/perl -S "$0" "$@";; esac'

This line has two "/bin/perl" and we can't use a line number to tell
sed which line it is since the line numbers in different files are
defferent.

[YOCTO #2304]

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
13 years agoglib-2.0: fix builds on fedora 17
Robert Yang [Mon, 16 Apr 2012 08:36:45 +0000 (16:36 +0800)]
glib-2.0: fix builds on fedora 17

Generally distros keep perl at /usr/bin/perl
Fedora 17 also has /bin/perl

this causes glib-2.0 build on such distros to put perl interpreter path in
the perl scripts as /bin/perl

But we set perl location for target as /usr/bin/perl

This mismatch of perl path causes failure of rootfs image creation
like this:

| error: Failed dependencies:
|       /bin/perl is needed by libglib-2.0-utils-2.30.3-r0.ppc603e
NOTE: package meta-toolchain-gmae-1.0-r7: task do_populate_sdk: Failed

[YOCTO #2304]

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
13 years agoperl: fix re-execution of patch/configure
Paul Eggleton [Mon, 16 Apr 2012 16:43:02 +0000 (17:43 +0100)]
perl: fix re-execution of patch/configure

The perl recipe's do_configure() does some manual sed replacements over
the entire WORKDIR. Fix the following issues with this:

* Skip patches, which fixes re-execution of do_patch after do_configure
  has run once
* Ensure that the replacement operation does nothing if do_configure is
  re-executed
* Avoid unnecessarily modifying /usr/include paths within documentation
  that will end up being packaged
* Fix a quoting issue in the expression used in the grep command that
  ended up causing files that did not contain .*\.h after /usr/include
  to be matched and modified.

The files modified during do_configure have been compared before and
after this patch to ensure there are no unexpected changes. Some
/usr/include paths that are not within documentation are no longer
being substituted however these are all within comments or scripts for
other Unix operating systems that are not applicable.

[YOCTO #2194]

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agocore-image-sato-sdk: typo in DESCRIPTION
Beth Flanagan [Mon, 16 Apr 2012 20:10:32 +0000 (13:10 -0700)]
core-image-sato-sdk: typo in DESCRIPTION

Fix for a minor type in DESCRIPTION

Signed-off-by: Beth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agocore-image-rt-*.bb: require munges DESCRIPTION
Beth Flanagan [Mon, 16 Apr 2012 20:10:31 +0000 (13:10 -0700)]
core-image-rt-*.bb: require munges DESCRIPTION

The require of core-image-minimal overrides the description.
The best fix for this would be conditional setting of DESCRIPTION
in the required images for all image descriptions, but that opens
a potential can of worms. Moving the require fixes this.

Signed-off-by: Beth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoautogen-native: Fix sstate relocation issue when reusing libguile
Richard Purdie [Mon, 16 Apr 2012 21:08:18 +0000 (22:08 +0100)]
autogen-native: Fix sstate relocation issue when reusing libguile

This was found to cause issues on the Yocto autobuilders and fixes do_compile
failures when guile-native has been relocated.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agobitbake.conf: Include abi_version.conf in a early stage
Dongxiao Xu [Mon, 16 Apr 2012 07:52:01 +0000 (15:52 +0800)]
bitbake.conf: Include abi_version.conf in a early stage

We may define new OELAYOUT_ABI variable in ${DISTRO}.conf, therefore we
need to move the abi_version.conf before that.

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoMultilib: Fix RDEPENDS in libc-package.bbclass, useradd.bbclass...
Zhai Edwin [Mon, 16 Apr 2012 06:50:00 +0000 (14:50 +0800)]
Multilib: Fix RDEPENDS in libc-package.bbclass, useradd.bbclass...

MLPREFIX is needed in RDEPENDS for multilib build

Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoopkg-config-base/poky-feed-config-opkg: Always use ALL_MULTILIB_PACKAGE_ARCHS to...
Richard Purdie [Mon, 16 Apr 2012 11:31:45 +0000 (11:31 +0000)]
opkg-config-base/poky-feed-config-opkg: Always use ALL_MULTILIB_PACKAGE_ARCHS to build arch list

We need consistent configuration files for opkg so we might as well always use
the full list. This is equal to PACKAGE_ARCHS in the non-multilib case.

This fixes various multilib failures with ipk multilibs.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agodropbear: Allow tasks to be safely re-executed
Richard Purdie [Mon, 16 Apr 2012 10:32:46 +0000 (11:32 +0100)]
dropbear: Allow tasks to be safely re-executed

Re-running the debug_patch task would cause the build to fail. This patch
moves the extra patch handling directly into SRC_URI and removes the need
for the separate task, allowing safe re-execution of each task.

[YOCTO #2194]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agodocbook-utils-native: Fix re-execution of the unpack task
Richard Purdie [Mon, 16 Apr 2012 10:05:02 +0000 (11:05 +0100)]
docbook-utils-native: Fix re-execution of the unpack task

If the do_unpack task is re-executed, the sed_done stamp was not removed, the
sed commands wouldn't re-run and the build would fail in do_compile. This
patch ensures the stamp is in ${S} and that we clean that directory
when unpack runs so it is a build from sctatch at that point.

[YOCTO #2194]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agosgml-common: Fix re-execution of the compile task
Richard Purdie [Mon, 16 Apr 2012 09:49:45 +0000 (10:49 +0100)]
sgml-common: Fix re-execution of the compile task

The sed command run during do_compile isn't idempotent. This change fixes it
so that it is and the compile task can be re-executed.

[YOCTO #2194]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agowget: Fix re-execution issue
Richard Purdie [Mon, 16 Apr 2012 09:48:35 +0000 (10:48 +0100)]
wget: Fix re-execution issue

The patch in this recipe patches both configure and configure.in which leads
to build failures once configure has executed and regenerated configure.

The fix to make patch re-executable is not to patch configure and
only patch configure.in.

[YOCTO #2194]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopython: Fix re-execution problem due to patching configure and configure.in
Richard Purdie [Mon, 16 Apr 2012 09:47:56 +0000 (10:47 +0100)]
python: Fix re-execution problem due to patching configure and configure.in

There was a patch patching both configure and configure.in. Since the recipes
all autoreconf, this is pointless and means subsequent patch tasks fail.

The fix is to only patch configure.in.

[YOCTO #2194]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogcc-configure-common.inc: Stop gcc looking at build system paths
Richard Purdie [Sun, 15 Apr 2012 17:05:53 +0000 (18:05 +0100)]
gcc-configure-common.inc: Stop gcc looking at build system paths

There were puzzling failures when you make a force recompile of any gcc
component. The error was in do_configure with cross-compilation badness
being detected in config.log files.

gcc is different in that many of the config.log files are generated during
the do_compile phase. This means this host contamination issue has always
been present but only shows up on a rebuild.

The fix is to force the appropriate configuration variable to "none required"
then gcc won't look in the bad locations.

[YOCTO #2279]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogst-plugins-bad: add curl as build dependency
Dongxiao Xu [Sat, 14 Apr 2012 10:44:14 +0000 (18:44 +0800)]
gst-plugins-bad: add curl as build dependency

gst-plugins-bad needs curl.h to finish its build, so add curl as its
dependency.

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agonet-tools: Fix do_patch to be re-executable
Richard Purdie [Sun, 15 Apr 2012 16:06:47 +0000 (17:06 +0100)]
net-tools: Fix do_patch to be re-executable

This recipe's do_patch was not able to be re-executed due to its custom
patching functions. This fixes things so that it can be re-executed.

It also fixes the unpack task re-execution by ensuring ${S} is clean
before the unpack starts.

[YOCTO #2194]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoman-pages: Fix to allow re-execution of tasks
Richard Purdie [Sun, 15 Apr 2012 15:32:29 +0000 (16:32 +0100)]
man-pages: Fix to allow re-execution of tasks

The all target in the makefile triggers "screen" and "remove" targets,
neither of which do anything useful/sane for our usecases. The simplest
way to get the makefile to behave is to only use the install target.

Clean up the recipe a bit whilst making these changes and simplify things.

[YOCTO #2194]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopseudo: Ensure the correct libraries are used at runtime
Richard Purdie [Sun, 15 Apr 2012 15:00:42 +0000 (16:00 +0100)]
pseudo: Ensure the correct libraries are used at runtime

There can be a conflict between the nativesdk libc and the host system's
libc. It is assumed the nativesdk version is of an equal or higher version.
This is a particular issue for pseudo if its loading a system binary
since the system's libc might be used of an older verison which would
then confuse libpseudo.so when loaded as a preload.

To avoid this, set LD_LIBRARY_PATH so the nativesdk libc is always
used.

Since we now use --without-rpath, we can remove the MAKEOPTS RPATH workaround.

[YOCTO #2299]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agounfs-server: Fix do_configure so it can be re-executed safely
Richard Purdie [Sun, 15 Apr 2012 13:02:02 +0000 (14:02 +0100)]
unfs-server: Fix do_configure so it can be re-executed safely

[YOCTO #2194]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoguile: Fix binary relocation issue for precompiled guile objects
Richard Purdie [Sun, 15 Apr 2012 11:02:18 +0000 (12:02 +0100)]
guile: Fix binary relocation issue for precompiled guile objects

When guile-native was relocated, there were messages like:

;;; note: source file /srv/home/pokybuild/yocto-autobuilder/yocto-slave/emenlow/build/build/tmp/sysroots/x86_64-linux/usr/share/guile/2.0/ice-9/eval.scm
;;;       newer than compiled /srv/home/pokybuild/yocto-autobuilder/yocto-slave/fri2/build/build/tmp/sysroots/x86_64-linux/usr/lib/guile/2.0/ccache/ice-9/eval.go

and this confuses things like the autogen-native "guile --version"
check. This patch ensures the wrapper script sets the necessary
variables correctly.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoconnman-gnome: connman 0.79 API fixes
Paul Eggleton [Sun, 15 Apr 2012 02:01:57 +0000 (03:01 +0100)]
connman-gnome: connman 0.79 API fixes

Fix connman-gnome to work with connman 0.79, which made a number of
fairly serious DBus API changes. Also switch over to the newly
repopulated git repo on kernel.org in which the two previous patches
have been merged.

Fixes [YOCTO #2202].

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogrub: Add missing dependency on autogen-native to avoid build failures
Richard Purdie [Sat, 14 Apr 2012 22:42:16 +0000 (23:42 +0100)]
grub: Add missing dependency on autogen-native to avoid build failures

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agodistro_tracking: add entries for several new recipes
Scott Garman [Sat, 14 Apr 2012 17:38:01 +0000 (10:38 -0700)]
distro_tracking: add entries for several new recipes

Added distro tracking fileds for the following recipes:

qt-mobility-x11
gst-plugin-bluetooth
alsa-state
ocf-linux
avahi-ui
libjson
systemtap-uprobes
xf86-video-fbdev
u-boot-fw-utils
qt4-graphics-system
qt-mobility-embedded
lttng-tools
shadow-securetty

Also fixed a few minor typos in others (spelling of bluetooth and months)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agodistro_tracking: add some additional DISTRO_ALIASES
Saul Wold [Sat, 14 Apr 2012 17:38:00 +0000 (10:38 -0700)]
distro_tracking: add some additional DISTRO_ALIASES

ltp
rgb
libksba
stat
directfb-example
libfakekey
builder
core-gtk-directfb
task-core-tools-*
task-self-hosted
sato-screenshot
libxxf86misc
qmmp
portmap

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agomulitlib.bbclass: Ensure correct value of ALL_MULTILIB_PACKAGE_ARCHS is preserved
Richard Purdie [Sat, 14 Apr 2012 21:46:17 +0000 (22:46 +0100)]
mulitlib.bbclass: Ensure correct value of ALL_MULTILIB_PACKAGE_ARCHS is preserved

The value of ALL_MULTILIB_PACKAGE_ARCHS needs to be consistent both
in multilib extended recipes and in normal context. If this isn't the
case it can lead to inconsistent configuration files at a minimum.

This patch ensures the value is preserved during the class extension code
since computing it after that point is hard.

[YOCTO #2290]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoupdate-alternatives: Fix for compatibility with multilib recipes
Richard Purdie [Sat, 14 Apr 2012 17:31:24 +0000 (18:31 +0100)]
update-alternatives: Fix for compatibility with multilib recipes

The PN == BPN test was designed to exclude native/nativesdk but also unfortunately
excludes multilib unintentionally. This change fixes this and allows multilib images
with alternatives to funciton correctly.

[YOCTO #2214]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoexternal-poky-toolchain: This is broken and confusing now, remove it
Richard Purdie [Sat, 14 Apr 2012 17:30:42 +0000 (18:30 +0100)]
external-poky-toolchain: This is broken and confusing now, remove it

This avoids a number of bogus preferred providers warnings too.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agosystemtap: Ensure FILESPATH is set so systemtap-uprobes can find patches
Richard Purdie [Sat, 14 Apr 2012 13:30:34 +0000 (14:30 +0100)]
systemtap: Ensure FILESPATH is set so systemtap-uprobes can find patches

This allows systemtap-uprobes to build again after the recently
added patch broke the builds.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoeglibc: fix builds on fedora 17 alpha
Nitin A Kamble [Sat, 14 Apr 2012 08:59:05 +0000 (01:59 -0700)]
eglibc: fix builds on fedora 17 alpha

Generally distros keep perl at /sur/bin/perl
Fedora 17 alpha also has /bin/perl

this causes eglibc build on such distros to put perl interpreter path in
the perl scripts as /bin/perl

But we set perl location for target as /usr/bin/perl

This mismatch of perl path causes failure of rootfs image creation
like this:

| error: Failed dependencies:
|       /bin/perl is needed by eglibc-utils-2.13-r23+svnr15508.i586
NOTE: package core-image-sato-1.0-r0: task do_rootfs: Failed
ERROR: Task 8
(/home/nitin/prj/poky.git/meta/recipes-sato/images/core-image-sato.bb,
do_rootfs) failed with exit code '1'

This Fixes bug : [YOCTO #2286]

RP: Updated to better account for -nativesdk
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogrub-1.99: fix configure error
Nitin A Kamble [Sat, 14 Apr 2012 08:59:08 +0000 (01:59 -0700)]
grub-1.99: fix configure error

This commit deals with this configure error.

configure:20306: checking whether `ccache i586-poky-linux-gcc  -m32
-march=core2 -msse3 -mtune=generic -mfpmath=sse
--sysroot=/builddisk/build/build0/tmp/sysroots/emenlow' generates calls
to `__enable_execute_stack()'
configure:20320: ccache i586-poky-linux-gcc  -m32    -march=core2 -msse3
-mtune=generic -mfpmath=sse
--sysroot=/builddisk/build/build0/tmp/sysroots/emenlow -O2 -pipe -g
-feliminate-unused-debug-types -Wall -W -Wshadow -Wpointer-arith
-Wmissing-prototypes -Wundef -Wstrict-prototypes -g -falign-jumps=1
-falign-loops=1 -falign-functions=1 -mno-mmx -mno-sse -mno-sse2
-mno-3dnow -mfpmath=387 -fno-dwarf2-cfi-asm -m32 -fno-stack-protector
-mno-stack-arg-probe -Werror -Wno-trampolines -falign-loops=1 -S
conftest.c
conftest.c:308:6: error: no previous prototype for 'g'
[-Werror=missing-prototypes]
cc1: all warnings being treated as errors
configure:20323: $? = 1
configure:20327: error: ccache i586-poky-linux-gcc  -m32    -march=core2
-msse3 -mtune=generic -mfpmath=sse
--sysroot=/builddisk/build/build0/tmp/sysroots/emenlow failed to produce
assembly code

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogrub-efi-native, grub: fix build with gcc 4.7
Nitin A Kamble [Sat, 14 Apr 2012 08:59:07 +0000 (01:59 -0700)]
grub-efi-native, grub: fix build with gcc 4.7

This fixes bug [YOCTO #2293]

These build failure caused by gcc4.7 is fixed with a backport of a
grub-1.99 patch from fedora 17 alpha plus two more new patches

| gcc -DHAVE_CONFIG_H -I. -I..  -Wall -W -I../include -I../include
-DGRUB_MACHINE_EFI=1 -DGRUB_MACHINE=I386_EFI -nostdinc -isystem
/usr/lib/gcc/x86_64-redhat-linux/4.7.0/include
-DGRUB_FILE=\"commands/efi/acpi.c\" -I. -I. -I.. -I.. -I../include
-I../include
-isystem/home/nitin/builds/build0/tmp/sysroots/x86_64-linux/usr/include
-O2 -pipe -g -feliminate-unused-debug-types -Wall -W -Wshadow
-Wpointer-arith -Wmissing-prototypes -Wundef -Wstrict-prototypes -g
-falign-jumps=1 -falign-loops=1 -falign-functions=1 -mno-mmx -mno-sse
-mno-sse2 -mno-3dnow -fno-dwarf2-cfi-asm -m32 -fno-stack-protector
-mno-stack-arg-probe -Werror -Wno-trampolines          -ffreestanding
-isystem/home/nitin/builds/build0/tmp/sysroots/x86_64-linux/usr/include
-O2 -pipe -c -o commands/efi/acpi_module-acpi.o `test -f
'commands/efi/acpi.c' || echo './'`commands/efi/acpi.c
| gcc: error: unrecognized command line option '-melf_i386'
| make[3]: *** [trig.module] Error 1

| make[3]: Entering directory
`/home/nitin/builds/build0/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r7/grub-1.99/grub-core'
| gcc -DHAVE_CONFIG_H -I. -I..  -Wall -W -I../include -I../include
-DGRUB_MACHINE_EFI=1 -DGRUB_MACHINE=I386_EFI -nostdinc -isystem
/usr/lib/gcc/x86_64-redhat-linux/4.7.0/include
-DGRUB_FILE=\"fs/btrfs.c\" -I.
-I. -I.. -I.. -I../include -I../include
-isystem/home/nitin/builds/build0/tmp/sysroots/x86_64-linux/usr/include
-O2
-pipe -g -feliminate-unused-debug-types -Wall -W -Wshadow
-Wpointer-arith
-Wmissing-prototypes -Wundef -Wstrict-prototypes -g -falign-jumps=1
-falign-loops=1 -falign-functions=1 -mno-mmx -mno-sse -mno-sse2
-mno-3dnow
-fno-dwarf2-cfi-asm -m32 -fno-stack-protector -mno-stack-arg-probe
-Werror
-Wno-trampolines          -ffreestanding
-isystem/home/nitin/builds/build0/tmp/sysroots/x86_64-linux/usr/include
-O2
-pipe -c -o fs/btrfs_module-btrfs.o `test -f 'fs/btrfs.c' || echo
'./'`fs/btrfs.c
| fs/btrfs.c: In function 'grub_btrfs_read_logical':
| fs/btrfs.c:791:5: error: 'err' may be used uninitialized in this
function
[-Werror=maybe-uninitialized]
| fs/btrfs.c:592:18: note: 'err' was declared here
| cc1: all warnings being treated as errors
| make[3]: *** [fs/btrfs_module-btrfs.o] Error 1

| gcc -DHAVE_CONFIG_H -I. -I..  -Wall -W -I../include -I../include
-DGRUB_MACHINE_EFI=1 -DGRUB_MACHINE=I386_EFI -nostdinc -isystem
/usr/lib/gcc/x86_64-redhat-linux/4.7.0/include
-DGRUB_FILE=\"fs/zfs/zfs.c\" -I. -I. -I.. -I.. -I../include -I../include
-isystem/home/nitin/builds/build0/tmp/sysroots/x86_64-linux/usr/include
-O2 -pipe -g -feliminate-unused-debug-types -Wall -W -Wshadow
-Wpointer-arith -Wmissing-prototypes -Wundef -Wstrict-prototypes -g
-falign-jumps=1 -falign-loops=1 -falign-functions=1 -mno-mmx -mno-sse
-mno-sse2 -mno-3dnow -fno-dwarf2-cfi-asm -m32 -fno-stack-protector
-mno-stack-arg-probe -Werror -Wno-trampolines          -ffreestanding
-isystem/home/nitin/builds/build0/tmp/sysroots/x86_64-linux/usr/include
-O2 -pipe -c -o fs/zfs/zfs_module-zfs.o `test -f 'fs/zfs/zfs.c' || echo
'./'`fs/zfs/zfs.c
| fs/zfs/zfs.c: In function 'get_filesystem_dnode':
| fs/zfs/zfs.c:1449:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:1449:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c: In function 'make_mdn':
| fs/zfs/zfs.c:1478:3: error: dereferencing type-punned pointer will
break strict-alERROR: Function failed: do_compile (see
/home/nitin/builds/build0/tmp/work/x86_64-linux/grub-efi-i586-native-1.99-r7/temp/log.do_compile.9293
for further information)
| iasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c: In function 'dnode_get_fullpath':
| fs/zfs/zfs.c:1554:3: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:1554:3: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:1571:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:1571:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c: In function 'grub_zfs_open':
| fs/zfs/zfs.c:2234:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2234:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c: In function 'fill_fs_info':
| fs/zfs/zfs.c:2362:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2362:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2395:3: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2395:3: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c: In function 'iterate_zap':
| fs/zfs/zfs.c:2420:5: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2420:5: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c: In function 'grub_zfs_dir':
| fs/zfs/zfs.c:2483:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2483:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2484:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2484:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2502:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| fs/zfs/zfs.c:2502:7: error: dereferencing type-punned pointer will
break strict-aliasing rules [-Werror=strict-aliasing]
| cc1: all warnings being treated as errors
| make[3]: *** [fs/zfs/zfs_module-zfs.o] Error 1

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogrub-1.99: use gcc-cross for building the target binaries
Nitin A Kamble [Sat, 14 Apr 2012 08:59:06 +0000 (01:59 -0700)]
grub-1.99: use gcc-cross for building the target binaries

It was using distro gcc to build binaries for target. This got detected
on fedora 17 alpha, on which it hit an gcc-4.7 issue.

This Fixes Bug: [Yocto #2291]

More information in the patch header

Thanks to Saul Wold for suggesting to remove the --target parameter from
EXTRA_OECONF var.

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agonspr: fix package spliting
Dexuan Cui [Fri, 13 Apr 2012 10:04:09 +0000 (18:04 +0800)]
nspr: fix package spliting

Here /usr/lib/lib*.so files are binaries rather than symbol links.
We should package them into ${PN} rather than ${PN}-dev, or else,
when a package, that rdepends on nspr, is packaged, we get a
"non-dev package rdepends on nspr-dev" ERROR.

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agosystemtap: disable document generation by default
Tom Zanussi [Fri, 13 Apr 2012 19:56:03 +0000 (14:56 -0500)]
systemtap: disable document generation by default

Building the systemtap documentation adds significantly to the build
time, so disable it by default.

Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agosystemtap: fix docproc build error
Tom Zanussi [Fri, 13 Apr 2012 19:56:02 +0000 (14:56 -0500)]
systemtap: fix docproc build error

When building docs in systemtap, docproc is used to generate the
tapset documentation, but it gets built for the target, while it needs
to be build for the host instead.  This change causes that to happen.

Fixes [YOCTO #2193].

Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolinux-yocto: allow .cfg, .scc, .patch and defconfigs to be processed in order
Bruce Ashfield [Fri, 13 Apr 2012 20:55:44 +0000 (16:55 -0400)]
linux-yocto: allow .cfg, .scc, .patch and defconfigs to be processed in order

During testing/extension of the linux-yocto-tiny kernel it was found that
defconfigs were not always properly applied. This was due to two issues:

  - not being able to fully control the order of objects applied to the
    git tree on the SRC_URI
  - defconfigs triggering --allnoconfig before being applied

To fix this, the recipe space code that previously detected and generated
automatic features moves back to the kernel tools (where it was before) and
is updated to also process .cfg and defconfigs. Moving this back to the
tools allow other recipes to automatically benefit from the additional
support.

The second issue is addressed by allowing configme to take --alldefconfig
when a recipe wishes to pass a defconfig and override the default
behaviour.

Fixes [YOCTO: 2250]

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolinux-yocto/3.2: add igb support to romley
Bruce Ashfield [Fri, 13 Apr 2012 20:55:43 +0000 (16:55 -0400)]
linux-yocto/3.2: add igb support to romley

Updating the 3.2 recipe SRCREVs to pickup the following meta change:

[
    meta: Add igb.scc to Romley

    Romley machine has 82580 Giga bit Ethernet Controller.
    Add the relavent Nic driver to it.

Signed-off-by: Kishore Bodke <kishore.k.bodke@intel.com>
]

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoRevert "eglibc: fix perl path in target scripts on fedora 17 alpha"
Richard Purdie [Fri, 13 Apr 2012 21:38:14 +0000 (22:38 +0100)]
Revert "eglibc: fix perl path in target scripts on fedora 17 alpha"

This reverts commit 7c95d554c431451f975e1e5d0336d1fb7d0ce7ae.

eglibc uses the PERL to execute things as well as encoding the path
in target scripts. We will therefore have to resort to sed on the scripts
at do_install time to fix the problems originally reported. This patch
causes various weird build failures and needs to be removed.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agorpm-native: Compile python rpm module (with-python)
Andrei Gherzan [Wed, 11 Apr 2012 14:26:17 +0000 (17:26 +0300)]
rpm-native: Compile python rpm module (with-python)

This module is needed by python scripts from createrepo native package.

[YOCTO #2134]

RP: Add missing DEPENDS on python-native
Signed-off-by: Andrei Gherzan <andrei@gherzan.ro>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopackage-index: Force NATIVE python to use modules from STAGING_DIR_NATIVE
Andrei Gherzan [Wed, 11 Apr 2012 14:26:16 +0000 (17:26 +0300)]
package-index: Force NATIVE python to use modules from STAGING_DIR_NATIVE

Export PYTHONHOME to ${STAGING_DIR_NATIVE}/usr. In this way python will search
for modules in native sysroot.

[YOCTO #2134]

Signed-off-by: Andrei Gherzan <andrei@gherzan.ro>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agocreaterepo: Python scripts should use the python interpreter from env
Andrei Gherzan [Wed, 11 Apr 2012 14:26:15 +0000 (17:26 +0300)]
createrepo: Python scripts should use the python interpreter from env

Added path to modify the hardcoded path of python interpreter from python
scripts.
The python interpreter should be the one from native sysroot and not the
one from host machine.

[YOCTO #2134]

Signed-off-by: Andrei Gherzan <andrei@gherzan.ro>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoeglibc: fix perl path in target scripts on fedora 17 alpha
Nitin A Kamble [Fri, 13 Apr 2012 04:43:44 +0000 (21:43 -0700)]
eglibc: fix perl path in target scripts on fedora 17 alpha

fedora 17 has /usr/bin/perl as well as /bin/perl, which results in
 the perl path set in thetarget  perl scripts as /bin/perl
While we install perl on target at ${bindir}/perl i.e. /usr/bin/perl
Hence the target perl scripts are broken when build is done on fedora 17.
Work around the issue by providing cached value of perl path to eglibc

This fixes bug [YOCTO #2286]

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoeglibc: fix re-execution of task
Venkata ramana gollamudi [Fri, 13 Apr 2012 11:44:49 +0000 (11:44 +0000)]
eglibc: fix re-execution of task

Task do_patch_append calling do_fix_ia_headers is removing files using "rm" not "rm -f".
So first time execution of patch task is success, while re-execution of patch task
fails as it tries to remove the files already removed.

So changed "rm" to "rm -f".

[Yocto #2194]

Signed-off-by: Venkata Ramana Gollamudi <ramana.gollamudi@huawei.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoboost: fix re-execution of task
Venkata ramana gollamudi [Fri, 13 Apr 2012 11:42:46 +0000 (11:42 +0000)]
boost: fix re-execution of task

After building boost package, re-execution of boostconfig task followed by
re-execution of compile task is giving following error
"error: duplicate initialization of gcc with the following parameters" during compilation
It is because multiple entries of gcc are being added during boostconfig re-execution
there by failing the compilation.

The patch fixes adding multiple "Using gcc" entries into /tools/build/v2/user-config.jam

[Yocto #2194]

Signed-off-by: Venkata Ramana Gollamudi <ramana.gollamudi@huawei.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoPR bump packages with gdbm in DEPENDS
Andrei Gherzan [Fri, 13 Apr 2012 10:34:19 +0000 (13:34 +0300)]
PR bump packages with gdbm in DEPENDS

This is done because of this change in gdbm:
"gdbm: Package compat libs in gdbm-compat"

Signed-off-by: Andrei Gherzan <andrei@gherzan.ro>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogdbm: Package compat libs in gdbm-compat
Andrei Gherzan [Fri, 13 Apr 2012 10:34:18 +0000 (13:34 +0300)]
gdbm: Package compat libs in gdbm-compat

In order to avoid breaking packages which depend on old package name libgdbm4 (>= 1.10),
compat libs are packaged into a separate package named gdbm-compat.

Signed-off-by: Andrei Gherzan <andrei@gherzan.ro>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agoRevert "default-distrovars: remove NO32LIBS setting"
Richard Purdie [Fri, 13 Apr 2012 14:50:00 +0000 (15:50 +0100)]
Revert "default-distrovars: remove NO32LIBS setting"

We still need this value to keep the sanity testing working. The other
part of the change is still valid.

This reverts commit cb940f46efacddc7200581edba034e4201abb94a.

13 years agoopkg-utils: Update to include a fix for version comparision issues
Richard Purdie [Fri, 13 Apr 2012 13:56:05 +0000 (14:56 +0100)]
opkg-utils: Update to include a fix for version comparision issues

This brings the version comparision within opkg-utils into sync with that
in opkg itself.

[YOCTO #2233]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agogcc-4.6: Add fix for relocation problem and ccache
Richard Purdie [Fri, 13 Apr 2012 13:13:48 +0000 (14:13 +0100)]
gcc-4.6: Add fix for relocation problem and ccache

If the toolchain is reused from sstate and ccache is installed, build failures
were occuring due to gcc trying to access the original sysroot rather than the
new one, particularly if the old sysroot existed but was not readable by the
current user.

This turns out of the an issue inside gcc to do with preservation of the sysroot
option. See the gcc patch for more details. It only triggers when preprocessed
sources are used which happens when ccache is used.

The same issue occurs with c++ and c++-cpp-output so the same fix is applied there.

[YOCTO #2074]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agodefault-distrovars: remove NO32LIBS setting
Paul Eggleton [Fri, 13 Apr 2012 12:40:17 +0000 (13:40 +0100)]
default-distrovars: remove NO32LIBS setting

The ??= assignment in pseudo.inc effectively nullifies this ??=
assignment here, so remove it.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopseudo: default NO32LIBS to 1
Paul Eggleton [Fri, 13 Apr 2012 12:35:16 +0000 (13:35 +0100)]
pseudo: default NO32LIBS to 1

If this value is not set to 1, then systems with some 32-bit libraries
but no 32-bit version of libgcc installed will have pseudo-native fail
at do_compile. It should only really be set to 0 by those who know what
they are doing.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agobase.bbclass: Fix PACKAGECONFIG issues with native and nativesdk BBCLASSEXTEND recipe...
Richard Purdie [Thu, 12 Apr 2012 13:04:18 +0000 (14:04 +0100)]
base.bbclass: Fix PACKAGECONFIG issues with native and nativesdk BBCLASSEXTEND recipes (and multilib)

This patch fixes up the issues that were being seen where BBCLASSEXTEND and
PACKAGECONFIG were interacting badly. It also ensures PACKAGECONFIG interacts
properly with multilib builds.

Ideally some of this code will be abstracted into lib/oe/classextend.py but
at this point in release more invasive changes like this are inappropriate.

This patch also removed empty strings from expressions rather than
passing them around as this was complicating the additional code
unnecessarily.

The patch was verified against the OE-Core metadata where the return values of
expandFilter() were sanity checked by hand for native/nativesdk and
multilib combinations.

[YOCTO #2225]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agomklibs-native 0.1.33: include unistd.h to fix build for gcc 4.7
Robert Yang [Thu, 12 Apr 2012 14:49:02 +0000 (07:49 -0700)]
mklibs-native 0.1.33: include unistd.h to fix build for gcc 4.7

Many of the standard C++ library include files have been edited to no
longer include <unistd.h> to remove namespace pollution in gcc 4.7, so
we need include <unistd.h> manually sometimes.

More information:
http://gcc.gnu.org/gcc-4.7/porting_to.html

[YOCTO #2278]

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
[Updated patch header - sgw]

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agorpm: Ensure that we check both providename and filepaths
Mark Hatle [Thu, 12 Apr 2012 21:21:23 +0000 (16:21 -0500)]
rpm: Ensure that we check both providename and filepaths

In lib/rpmts.c we need to make sure to look for missing dependencies,
such as /bin/sh, in both the providename and filepaths DB.  Previously
the system would key off an initial '/' and only look in the filepaths DB.

This causes problems when a package such as "bash", has a Provides: /bin/sh.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopseudo: Tell pseudo to avoid specifying an RPATH
Mark Hatle [Thu, 12 Apr 2012 21:21:22 +0000 (16:21 -0500)]
pseudo: Tell pseudo to avoid specifying an RPATH

[Yocto #2251]

Add --without-rpath to avoid embedding rpaths into the pseudo
components.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopackage_rpm.bbclass: Set tmppath for rpm to somewhere which won't conflict with the...
Richard Purdie [Thu, 12 Apr 2012 20:16:41 +0000 (20:16 +0000)]
package_rpm.bbclass: Set tmppath for rpm to somewhere which won't conflict with the rootfs

If the rootfs has a specific layout in mind for /var/tmp, it should be
free to create this. Preinst scripts trigger rpm to create its tmppath
directory which could potentially conflict with this.

We're seeing issues if the provider of the /var/tmp wants to use a symlink
and a preinst script executes before it has been installed. This change
places tmppath as used by rpm somewhere safe which is already automatically
cleaned up too.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolibunistring: Fix parallel make issue
Richard Purdie [Thu, 12 Apr 2012 08:42:59 +0000 (09:42 +0100)]
libunistring: Fix parallel make issue

See patch header for the full description of the parallel make issue this resolves.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agolinux-dtb: Ensure dtb files are covered by sstate
Richard Purdie [Thu, 12 Apr 2012 09:18:58 +0000 (10:18 +0100)]
linux-dtb: Ensure dtb files are covered by sstate

The dtb files were not being installed into a location compatible with sstate and
the do_deploy task. This means in builds just using sstate, the dtb files disappeared.

This patch fixes the code to use the correct location for deploy files.

[YOCTO #2190]

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopackage_rpm.bbclass: Use the correct macros file to avoid empty solvedb path issues
Richard Purdie [Thu, 12 Apr 2012 07:21:38 +0000 (08:21 +0100)]
package_rpm.bbclass: Use the correct macros file to avoid empty solvedb path issues

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
13 years agopackage_rpm: Fix useradd preinst ordering issues
Richard Purdie [Wed, 11 Apr 2012 21:31:20 +0000 (22:31 +0100)]
package_rpm: Fix useradd preinst ordering issues

We were already having occasional ordering issues with package_rpm.
Fixing the ldconfig postinstall issue pushed rpm over the cliff and
totally broke rpm builds with the packages getting installed in
effectively a random order and the useradd preinstalls getting executed
out of order and breaking.

The only explanation I can find for this is that rpm is special. It will
happily run a preinst for a package without any of that package's
dependencies being present regardless of whether there are any circular
dependency issues or not. I attempted various ways of solving this such
as ordering the total_solution.manifest in creative ways but the bottom
line is RPM ignores this. It takes little account of any request to
ignore /bin/sh dependencies for the purposes of constructing the final
image.

The end result is we're having to install the base-passwd, base-files
and shadow packages first (if there is a request to install them), then
install any other packages.

It this wasn't in the middle of a release I'd be rewriting this bbclass
file, its horrible.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>