libzypp: move package-manger to it own package

[YOCTO #1234]

This moves a couple of files to a pkgmgt package so that
they will not be installed by default. This removes the
Add/Remove Software Icon from rpm based rootfs.

(From OE-Core rev: 3be290aac4da885fe5bc31797c4a689196c63b7a)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hob: correctly set the selected image when loading a recipe

When the user saves their recipe based on an existing image type, loads it
in a newly run hob instance and clicks bake they should not be asked about
building packages vs an empty image up.

Partially addresses [YOCTO #1468]

(Bitbake rev: f4be83aae77a713e2b5a2d094651639b6e641031)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hob: correctly handle an exception

It doesn't matter if we can't remove the temprorary file, for some reason,
so catch the exception and ignore it.

Partially addresses [YOCTO #1468]

(Bitbake rev: 4394e38b038e1bc9845adf01d73363157d98c96d)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ui/crumbs/hobeventhandler: don't check BBPATH and BBFILES each build

There's no need to check the BBPATH and BBFILES are set correctly each
build when running multiple builds for one launch of the UI.

Partially addresses [YOCTO #1468]

(Bitbake rev: 39ed18e70e9f3a13b522b4ea02bf6f4bdb7de89c)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ui/crumbs/hobeventhandler: fix test for BBFILES

It seems we have a race whereby the image_dir variable may not be set
before it's tested for, since the variable is always the same set it in the
initialiser.

Partially addresses [YOCTO #1468]

(Bitbake rev: 09f5aed2edb0b8f4a6570e1041b5db66cf2d17cc)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

dpkg: fix pkg_postinst_dpkg, don't supply {bindir}/update-alternatives

[YOCTO #1070].

The pkg_postinst_dpkg seems buggy, e.g., for dpkg-native,
${IMAGE_ROOTFS}/${sysconfdir}/rcS.d doesn't exist.
I think OE's version(do_install_prepend) at
http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/dpkg/dpkg.inc
is correct, so let's change to OE's version.

And, bth dpkg and update-alternatives-cworth supply
/usr/bin/update-alternatives and this causes conflict. e.g., bug 1070.
We can resolve the issue by ask dpkg not to supply the binary.

(From OE-Core rev: c7af358ec427487ccc053316d2f68f6121b97181)

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

multilib: Remove recipe from multilib.conf that inherits allarch

Recipes like update-rc.d and qemu-config inherit "allarch", thus we
shouldn't add multilib BBCLASSEXTEND for them in multilib.conf.

Besides, we need to add multilib packages as the RPROVIDER contents for
those recipes, in order to avoid the NoProvider error when parsing.

[YOCTO #1471]

(From OE-Core rev: 329d864f9bbf94ad3aae8df43d63fe10e4237e4f)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

base.bbclass: do not expand PREFERRED_PROVIDER for kernel recipe

With Richard's commit f9c36392, we only build one kernel for a system,
thus we shouldn't extend PREFERRED_PROVIDER for virtual/kernel.

[YOCTO #1471]

(From OE-Core rev: 90ad70fe13a9f8409387ca70289998bdca649aeb)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Revert "pango: use qemu to generate pango.modules during rootfs construction"

This reverts commit 62381fd3a135318794f14dc3740b04facd32446e.

This commit has issues with the rpm rootfs generation which need
resolving before it can be accepted.

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

opensp: Fix hard path in native nsgmls.

This relocation issues caused following bug and introduce one partial fix bb024d:
[YOCTO #1440]

But the environment val hacking is not working all the time, as nsgmsl may
append wrong path in some case. This fix use search directory "-D" as solution.
And more important, change the iputils' way to make doc: Use the sysid in
current directory instead of parent directory, as "../" is also appended to the
search directory then lead searching failure.

(From OE-Core rev: a5aeb873b943300e18493031ffa9e9f8ab56c2d2)

Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
CC: Saul Would <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqueue.py: Ensure fakeroot variables are reflected in the datastore

Without this, variables can be set to one thing in one part of the environment and something
different in another part. This change ensures the datastore and the environment
are consistent.

(Bitbake rev: 459addf13721a6847406f215650fa1882fb83ea9)

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

shadow: add patch for useradd lossage

This fixes (or, at least, papers over) a failure in do_install for recipes which
inherit useradd.bbclass.  Rewinding optind in this way is not entirely portable
but in practice it seems to work on GNU-ish build hosts at least.

(From OE-Core rev: 8fce8180c802ad187c4df44c17207bfb026ce6c7)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu: improve auto-detection of rootfs filenames

This refactors the way rootfs filenames are auto-detected when you
run the runqemu script without an explicit rootfs filename argument.
It allows the script to use rootfs files generated by hob, and when
there are mutliple rootfs files to choose from, it will pick the
most recently created one.

Fixes [YOCTO #1437].

(From OE-Core rev: 094453f443ed592d814dfb4182a0a02f0a2552e4)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake.conf: Default to explicitly setting the variable to disable pseudo

If we don't do this, sometimes pseudo can end up enabling itself when
it isn't required. Setting this value into the environment explicitly
ensures that does not happen.

(From OE-Core rev: 72252799e8c51a633a231a2cd1fe797b7faae713)

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

runqemu-internal: Hide some harmless warning messages

If sudo is used in the pseudo environment, as done in image tests when
the user hasn't pre-setup the tap device, ensure the LD_PRELOAD error
message isn't seen by the user.

(From OE-Core rev: 86234ac514cbd33a0058f3b74f158daeda325c0d)

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

imagetest-qemu/runqueue: Since we no longer support BUILDDIR, use TMPDIR

Commit 993672fa2739794a6dd0dbd7bb232fa60522b897 removed the BUILDDIR
support from runqueue which broke the imagetest-qemu integration. We now
therefore need to set TMPDIR and pass this through the environment to
ensure the runqueue script finds the right locations without running
bitbake directly.

This patch also adds a sleep to the qemu command in the error case so that
this remains on the screen for a period of time so the user can see errors
more easily.

This change unbreaks automated testing failures on the autobuilder.

(From OE-Core rev: de470333dbdeea444199340e4cd458c13fed6a5a)

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

Revert "opensp: Fix sstate relocation issue"

This patch causes the linuxdoc-tools-native to fail, cause it's
overriding the SGML_CATALOG_FILES, need to find a better way to
set the CATALOG_FILE information for opensp / nsgmls

(From OE-Core rev: e9ecf5f22b5aa1f58b3104f7d058d04ab08e8839)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
This reverts commit bb024d7a8c771c189b94b1ae177e29f3c318e236.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

terminal.bbclass: Improve environment handling

Currently the environment handling for terminals is inconsistent. There
are two fixes here:

a) Ensure the environment is setup before all oe.terminal call

b) Actually set the environment before the spawn calls since we need
variables like DISPLAY when the commands are being executed, not just
within the terminal environment. If this doesn't happen, DISPLAY can end
up not set with the errors that brings with it when trying to run X
commands.

(From OE-Core rev: 6b680c8da02017bdb17a4826f1819fe0e5f68652)

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

runqueue: Ensure task environment is correct

This fixes two problems:

a) Variables which were in the parent environment but not set as "export"
   variables in the datastore could end up in the task environment

b) oe.environ.update() can't cope with the generator returned by
   bb.data.exported_vars()

Whilst the updated code isn't as neat, it does do the expected thing,
sets the environment correctly and stops unwanted values leaking into
the task environment.

(Bitbake rev: 67e5e23034c5ec2b9efcca935242830306c0048d)

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

python-native: add link for python2

Add link for python2 as some configuration tools search from a python2
and could find the host python instead of the python-native.

(From OE-Core rev: ae4db0bacb9d40489499f77905f26502f3bcaa19)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

base.bbclass: fix substring matching in COMMERCIAL_LICENSE

Previously, if for example you had a package called "mx", and a second
package called "libomxil" listed in COMMERCIAL_LICENSE (without mx being
listed there), it would match mx as being commercially licensed because
mx is a substring of libomxil. Fix the search to ensure it only matches
the listed package name exactly.

(From OE-Core rev: b23e51e6c961cf3f7e2aaf89648fecce78424c99)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

avahi: only use gtk if DISTRO_FEATURES includes x11

Otherwise avahi isn't buildable for an x11-less DISTRO.

(From OE-Core rev: fa8dad209247b780590bc8476b5b16e6608d7907)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

useradd.bbclass: use correct value for $D in postinst functions

This corrects the location of the password file used during package installation.

See http://lists.linuxtogo.org/pipermail/openembedded-core/2011-September/009183.html and subsequent discussion.

(From OE-Core rev: 848bb277769af5b094031aeb54d287c158256724)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

resolvconf: switch to 1.59 as 1.58 was removed from debian ftp

(From OE-Core rev: 11b466546533c6c2ad1f24d61d02a5bf6c1329e1)

Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sstate: Avoid contention between sstate packages

If several sstate packages are decompressing at the same time, they can
conflict over the "fixmepath". If two packages try to write to this at
the same time it results in an error. To avoid this we remove the file
once we've finished processing it.

(From OE-Core rev: c08d7093bf654ffedb155c5627972e9ecfeb0b60)

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

gettext: use included libxml and libcroco

[YOCTO #1400]

By using the included libxml and libcroco, we do not create additional
on this core library. There was an ordering issue that when one library
was not built, but gettext detected the other library a compliation would
fail, this will force it to always use the included libraries.

(From OE-Core rev: 0c12210f63f9705af98bcf5de8ccd715026bace3)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc: add patch for ice 50099, which caused lttng-ust not to build

[YOCTO #1381]

This patch came from from GCC Bugzilla via Khem

Cc: Khem Raj <raj.khem@gmail.com>
(From OE-Core rev: 61dac2f6f68bc46d8f3f6f7a8757924f103c7c54)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta/recipes.txt: correct Qt capitalisation

"Qt" is correct according to the documentation.

(From OE-Core rev: 429be858090d1c40589cb269a70ed1515119ac1f)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity.bbclass: re-enable DISTRO check

If DISTRO has been specified, ensure it is valid. (Unset or empty string
is valid for DISTRO in OE-core by the use of defaultsetup.conf.)

(From OE-Core rev: 895aa6b8ee9636a9f9d716c62c0adf951e1ff138)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

core-image.bbclass: zap root password if debug-tweaks not enabled

If you do not have debug-tweaks in IMAGE_FEATURES, then zap the root
password so that you can't log in as root without a password in an image
potentially intended for a production system.

Also mention debug-tweaks in the comments listing IMAGE_FEATURES in this
file.

(From OE-Core rev: 5ddda44efc97b4d6cf863033348e57e5cfdcd392)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hal: Add virtual/kernel as hal's build dependency

hal may need some kernel headers, thus we need virtual/kernel as its
dependency.

(From OE-Core rev: b5db50e280e2547f5428440be5446c4ad267aeef)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

multilib: Remove the multilib extension for linux-yocto

Kernel should be unique in a multilib system, so we remove the
linux-yocto's extension in multilib.conf

(From OE-Core rev: db78d1289705a38f0069f84101296e9309d67fa3)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

multilib_global.bbclass: handle kernel-module-* for multilib

bitbake would report failed dependency of kernel-module-* when testing
multilib. kernel-module-* are recommended by some other recipes.

Do not extend name for kernel-module-* related packages.

[YOCTO #1456]

(From OE-Core rev: 30ac343888dc801614922045b374537c6c54f312)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

telepathy-python: avoid duplicated installation of errors.py

telepathy-python install fails on latest Fedora 16 due to the tighter
check of duplicated file installation.

Remove one errors.py entry in Makefile.am to fix this issue.

[YOCTO #1453]

(From OE-Core rev: d84fd7c1e90f080ea0c702cba9a9f81a2c465ee6)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cpan.bbclass: fix native perl issue

(From OE-Core rev: acfe8c657dd2496ae8669ce69def96a0b04d061c)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

prelink: Update to latest upstream version

[YOCTO #1374] - Fixes underlying issue
[YOCTO #1331] - Detects bad GNU_HASH tables

Update to the latest upstream version.  This includes a number of
generic fixes, as well as a complete refresh of the prelink-rtld.

(Prelink-rtld is similar to the ld.so/ldd functionality on a normal
system.)

The new prelink-rtld update syncs to eglibc-2.13 and enables new
symbol resolution, include GNU_UNIQUE, that was causing problems on
x86_64 and potentially other architectures with C++.

An assert has been added to also detect a bad GNU_HASH situation
which was detected on some PPC64 systems.  The underlying problem is
within the toolchain, however prelink was making it worse leaving the
problem undetected.

(From OE-Core rev: 4f401134694bbc1de532aaaab6c9a761649725a1)

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glib-networking: Ensure certificates are consistently disabled

Currently configure is looking whether the build system has certificates
to determine the certificate paths. This is clear contamination so
this patch disables that functionality to ensure builds are consistent.

We could equally set this option to a consistent path if we start
installing certificates.

[YOCTO #1451]

(From OE-Core rev: 658887961dbb1eea17d9a1ceeef4fe38cb8c33ba)

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

package_deb.bbclass: put the stderr into log.do_xxx-attemptonly.xxx

Currently the stderr msgs go into log.do_rootfs and can cause a fatal failure
in "log_check rootfs".

However, the err msg could be spurious since we *only attempt* to install the
the packages, which may not exist actually. E.g., without this patch, we can
get a fatal ERROR due to
E: Couldn't find package task-core-tools-dev.

(From OE-Core rev: 5f00bda874ccaebdb00714b5cb5227462d43f8fb)

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

populate_sdk_deb.bbclass: add the missing parameter - for tar

(From OE-Core rev: a47ceddd6eb91e7564e6e48ddbb5d490f3d2767d)

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: update to 1.8.5

(From OE-Core rev: 77571c4d224d5bcc6be944fe3d00309dafe1d512)

Signed-off-by: Denis Carikli <denis@eukrea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

site/x86_64: Add entries to fix conf error on x86_64

Add entries for ashldi3, bswapsi2... to fix autoconf error of grub on x86_64

[YOCTO #1434] got fixed.

(From OE-Core rev: dba3c3de225f7f8008e638b65b8f31dd0ba91e21)

Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xserver-nodm-init: Change hidraw mode to allow normal user access

USB HID like touch screen could be presented as hidraw* device, this change
make user under rootless X can access them.

(From OE-Core rev: ddb54d4a467b51f5f8f305f532bd5ea64df92876)

Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

initramfs-live-install: Be consistent with grub.

[YOCTO #1428]
Because the underlying grub 0.97 the initramfs-live-install RDEPDNEDS on
only supports 32bit x86 target, set the COMPATIBLE_HOST accordingly.

(From OE-Core rev: d128b4f2746ee8525352ce982a9c2a5e422630f2)

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

web: Update SRCREV to address ccs_parser_parse_stylesheet arg mismatch [YOCTO #1454]

(From OE-Core rev: 5e558237fc22baf1a638eb0e98eb08ff5e2870b4)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mtd-utils: update to version to 1.4.6

mtd-utils 1.4.6 is the lastest release of mtd-utils
  at this time.

(From OE-Core rev: e43bec6c7dce3c6f9af5a0c5b7e5c6f740f89996)

Signed-off-by: Denis Carikli <denis@eukrea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

pango: use qemu to generate pango.modules during rootfs construction

This allows pango to be installed in an image with read-only-rootfs.

(From OE-Core rev: 62381fd3a135318794f14dc3740b04facd32446e)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto: re-enable utrace

Upating the meta data and machine branch contents with the up
to date merge of utrace.

(From OE-Core rev: cb8efc0e1ec00892b46325aabfb1b4020a46c078)

Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto: re-enable systemtap

[
  Author: Tom Zanussi <tom.zanussi@intel.com>
  Date:   Fri Sep 2 23:45:12 2011 -0500

    meta: re-enable systemtap feature for linux-yocto

    Re-enable the 'systemtap feature' that turns on the kernel options required
    for systemtap, a system-wide tracing tool.

    (From OE-Core rev: fdaaddbf2a673c7030c3e15e2f65cb13af3db720)

Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
]

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

icu-3.6: fix building with big make -j

icu-3.6 buildsystem contains small problem which causes some parallel
builds to fail. Fix that.

(From OE-Core rev: 41b7c955cfdc7b4d5c667c8a8742237d5edaca4f)

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qt-demo-init: fix init script

qt-demo-init initscript lacks some quotes, which makes it bave not fulle
correct. Fix that.

(From OE-Core rev: 84395cd31afd78a84d82a80297fe06f10406d57b)

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image_types: Fix ubi images creation

Without that commit ubinize.cfg lack a volume name value,
  and the related ubinize.cfg line looks like that:
    vol_name=
  which result in a broken ubi image,which after beeing flashed produce
  the following error:
    UBI error: vtbl_check: volume table check failed: record 0, error 11
  wich result in a kernel panic because the rootfs can't be mounted.

[RP: Moved from bitbake.conf to the .bbclass file]
(From OE-Core rev: 534dcaba0517ebfb6e48b6b436960a74e642a556)

Signed-off-by: Denis Carikli <denis@eukrea.com>
Acked-by: Anders Darander <anders@chargestorm.se>
Acked-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/hob: notify the user when the GUI won't launch immediately

If the wrapper script needs to build pseudo before we can launch hob we need
to notify the user so they aren't shocked by the action of launching a GUI
and seeing a bunch of text whiz by on the console.

Fixes [YOCTO #1435]

(From OE-Core rev: a160f4dd48b91c5e6f8c290c7572e29a39a3e693)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/hob: update to match recent hob changes

hob now uses both a pre and post file, update the wrapper script to generate
and use both of these.

Addresses [YOCTO #1281]

(From OE-Core rev: b68f90b765e7c8923033ee7ff7746f39a2e91ff7)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto: clean configuration for v3.0.4

Fixes [YOCTO #940]

Since v3.0.4 is likely the last stable update in the the release
timeframe a configuration audit was performed. This updates the
SRCREV to remove obselete, and improperly defined configuration
items. With this, all qemu* BSPs configure with no warnings.

(From OE-Core rev: d19eaf7ca92c6632d42d12120800028f77ef21aa)

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto: update to v3.0.4

The v3.0.4 stable kernel is available and it can now be merged
into linux-yocto. Build and boot tested on all qemu* machines.

(From OE-Core rev: 0f3750ee83b8fe012cced2e376ec780a6e7f189b)

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto: generalize kernel config search pattern

After constructing a kernel configuration file it then needs
to be located in the tree so it can be audited against the
final .config. The previous string that was used for the search
pattern contains the kernel version. If the recipe space kernel
version and internal tree version are out of sync, this will
cause the constructed config to not be found. By removing the
version from the search string, we can still find out config and
gracefully adapt to minor version skew.

(From OE-Core rev: f072ddc43828ebe8df4dd7433726775dd547580b)

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image_types.bbclass: use per-image dirs for extX.gz images generation

Currently image_types.bbclass uses deploy/images/${MACHINE}tpm.gz dir for
temporary files when generating extX.gz images. This can lead to races
and strange errors during image generation. To overcome this use
deploy/images/${MACHINE}tpm.gz-${PN} dir for temporary files.

(From OE-Core rev: 3016f8f13e86034cc1c5686fc01a3bf39eb33d79)

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu: standardize ability to specify custom qemu/kenel boot options

The old manner of specifying custom options to QEMU in this script
using angle brackets was a frequent source of confusion. Meanwhile,
Otavio Salvador added a decent method of specifying custom kernel
boot options to this script. This patch documents the bootparams
option and adds a similar way of specifying custom QEMU options
using qemuparams="".

This fixes [YOCTO #1019]

(From OE-Core rev: 1038df14a96d789b3f9e9e1692305ba1fea67886)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

task-core-tools: add systemtap to qemuarm

Systemtap now works with arm, so include it in task-core-tools-profile
for qemuarm.

(From OE-Core rev: 69a776eab96776ab33191cd6eeda4560e3f3f357)

Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemtap: upgrade to 1.6

Also enable for arm, since systemtap now works on arm and remove the
gcc-4.6 compile fix patch since the problems it addresses have been
fixed upstream.

(From OE-Core rev: 09fa8e139b0c06f961927ac192d644bebdf31abf)

Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qt4e-demo-image: Ensure a shell is present

"/bin/sh" is needed by qt4e-demo-image, So I add busybox for fixing this problem.

[YOCTO #999]

(From OE-Core rev: 5f06d5012bfa8b2bd83092e41ab29babdfd4605e)

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

git: fix sstate relocation for git-native

Create a wrapper for git to set GIT_EXEC_PATH and GIT_TEMPLATE_DIR so
that git-native is relocatable.

Fixes specific example given in [YOCTO #1137]

(From OE-Core rev: 9a3e25b5a3aa5e3b6ba543c4f789f66b240b38a7)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/bb/siggen.py: return a string from noop get_taskhash

OpenEmbedded is expecting to get a string from get_taskhash, but noop siggen
returns just 0 (number), so OE classes/sstate.bbclass barfs badly. Fix that.

(Bitbake rev: 24272dae15ccf641ece11ef5a6e2bfa3ebb6f5f9)

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hob: fix segfault on second build

Some internal lists were not being cleared, resulting in incorrect
program flow on the second build, causing a structure to be accessed
incorrectly which resulted in a segfault.

Fixes [YOCTO #1332]

(Bitbake rev: 71ac7fda51ed80e9ef6d1a3bca653683893e4770)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hob: clear out temporary file list after deleting

If you don't clear out files_to_clean after the files get deleted and
then you run a second build, it will try to delete the files from the
first build and you will get a "No such file or directory" error.

(Bitbake rev: 069d85cde01d14f8da31ad5cbd843a4d99628d42)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hob: disable removal of packages

It's felt that the stability of package deselection is not sufficient for
the upcoming release and thus package removal should be disabled.

I'd actually like to see this patch, or its effects, reverted as soon as
the release bits have been frozen so that this issue can continue to be
worked on.

(Bitbake rev: 73b6ff4654d10baae59d83e8568d58d989e99dd7)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ui/crumbs/runningbuild: mask run_buildstats failure

The buildstats handler causes an exception with: "'NoneType' object has no
attribute 'startswith'" early a build via hob, leaving a glaring red row
which means nothing to the user.
Mask this error until such a time as we have opportunity to correctly
diagnose and fix the root problem.

Workaround fix for [YOCTO #1433]

(Bitbake rev: b0cce5b52a20c0dad5ec0c4053f437cae89b6137)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ui/crumbs/hobprefs: disable 'build toolchain with headers'

The generic meta-toolchain-sdk we are currently building when this option
is enabled is likely unsuitable for the majority of images built with hob.
Remove this option from the Preferences UI until such a time as we can
correctly implement this feature to include the library headers for the
selected packages.

Addresses [YOCTO #1302]

(Bitbake rev: 3157967d08266b8d1ac563ba609ac3027b60d040)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hob: add a test to ensure hob is run with the required pre and post files

hob requires pre and post configuration files to store configuration values
in, whilst this should (and will) be fixed long-term for so long as we
require these files we should alert the user should they run without them.

Fixes [YOCTO #1383]

(Bitbake rev: bb3e9113074ea1254aa03a247a1a1070682df9c8)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hob: reflect defaultsetup being default distro

If no value is set for DISTRO the defaultsetup policy is used, reflect this
in the UI by having defaultsetup selected in the Distribution combo when no
other DISTRO is set.

(Bitbake rev: 126267c545ede65042959d134ea75c0345577747)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hob: use both pre and post files for hob configuration

We need to set various variables *before* parse begins, the simplest way
to ensure this is to use a pre configuration file for the relevant
configuration entries.

This series adapts hob to use both pre and post files to store its
configuration. Any variables which affect initial parse are set in the pre
file and all others in the post file.

Unfortunately this requires hob related code to have even more hard-coded
data as to what is relevant but this is the simplest way to solve issues
with variables and parse order at this time.

Addresses [YOCTO #1281]

(Bitbake rev: 02ab0e11d8dd42f5ca440b3d8d2073e23f55113a)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

opensp: Fix sstate relocation issue

[YOCTO #1440]

This fixes a problem where the native nsgmls has a path hard coded into
it, therefor add a wrapper to set the environment with the correct
paths for catalog files.

(From OE-Core rev: ce236234831ebc05ffdf1e0a368692ac2818a388)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

groff: Fix sstate relocation issue

[YOCTO #1439]

This fixes a problem where the native groff has a path hard coded into
it, therefor add a wrapper to set the command line with the correct
paths for fonts and tmac directories.

(From OE-Core rev: 9106e2922ba5972e3b87436372a12c2e04e9eb65)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

populate_sdk: We need to ensure that the SDK sysroot reflects PACKAGE_ARCH

If we don't do this, the SDK target sysroot is named generically even
when it contains package architecture specific optimisations.

(From OE-Core rev: 0616557a8c29b42bae0ffd5fd665a046810047e4)

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

sanity.bbclass: fix splitting on newlines of SANITY_TESTED_DISTROS

This variable should be split with \n sequences and these need to be
specified literally in the string. A corrected version of the example
given in the original commit (OE-core rev
75e3875341ddc8940e9ee2ccbbb2ec18194a68e6):

SANITY_TESTED_DISTROS = " \
        Ubuntu 11.04 \n \
        Fedora release 14 (Laughlin) \n \
        "

(From OE-Core rev: cfc72d5796b6f83a01e06f3a1f044869db2d5d18)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

useradd.bbclass: remove hardcoded reference to /usr/bin

Otherwise the class doesn't work if ${bindir} is set to a different value;
likewise for /var vs ${localstatedir}.

(From OE-Core rev: 21371df16917cd82642b39763793783d61ee5516)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc: Fix multilib baselib confusion

Commit 35fa8dc5f7da90fdd40091a3c3600d3fcd232922 changed the gcc recipes to use
baselib for the compiler location. This is fine as long as baselib happens to
match the platform multilib definition which is enabled at the time.

This patch fixes things so that gcc will honour whatever ${base_libdir} is
set to re-allowing suitable customisation of the system layout.

[YOCTO #1362]

(From OE-Core rev: bc5f293b151b9ba0d6660814d88ee5041efce318)

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

multilib: Only build one kernel

For a given system we only want one kernel to be built. This change makes
the main kernel recipe provide all of the provides of the various enabled
multilibs hence allowing it to fulfil all the appropriate dependencies.

To make this work a global multilib class file needed to be created.

This patch also enables this multi provider functionality for "allarch"
packages.

[YOCTO #1361]

(From OE-Core rev: 2fd257f6c610624f05c8dd3fe1486364af04696f)

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

multilib: extend multilib prefix for libsdl

libsdl is required by sato image, so extend it for multilib.

(From OE-Core rev: 88abab9ba9632e87f3a081915a69cbc5c36b4d3c)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

multilib: Using different sysroot for multilib recipes

Thinking of the senario that, if we already built out a 64bit image
along with the full toolchain bootstrapped, then we need to build some
32bit libraries, which needs lib32 versions of gcc and eglibc. These
toolchain recipes will bootstrap again in the same sysroot, resulting
that lib32-gcc-cross-initial will find some macros owned by eglibc have
already been defined and thus it includes non-existed headers that
provided by later lib32-eglibc.

The solution for the above issue is to use different sysroot for
multilib recipes, here we add ${MLPREFIX} in front of the machine
specific sysroot directory name.

[YOCTO #1372]

(From OE-Core rev: a1508ad1aec2d2f9ee040aa217c33193cd5bd871)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

rootfs_rpm: setting DEFAULTTUNE for multilib archs

To get the MULTILIB_PACKAGE_ARCHS, we need to get the corresponding
DEFAULTTUNE value. This fixes the multilib arch directory missing issue
in solvedb-ml_archs.conf.

(From OE-Core rev: 54306ff373e13696637b547fa1514e0ef8633248)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hal: Remove build time dependency of virtual/kernel

hal has runtime dependency on kernel, but not build time. Remove it from
"DEPENDS" list.

Also fix a wrong PACKAGE_ARCH setting when building multilib lib32-hal,
because ":=" will be extended immediately which is not the right value.
Using TUNE_PKGARCH instead.

(From OE-Core rev: 74646a2b2d7d452dfe95b08940389a686e8addcb)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

dropbear: add missing files

(From OE-Core rev: 4bc94f1896aad7f540ac520cd69edf3e96029319)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tzcode: Update to 2011i

(From OE-Core rev: 2c0c78fc777a7dbe098f2518ecbaf24d03227f5d)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sato recipes: Update PR because of libowl conversion to git

(From OE-Core rev: 8216888c1caba6d469a5443ba74fce0f1392792e)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

o-hand recipes: convert from svn.o-hand.org to git.yoctoproject.org

(From OE-Core rev: 4e7218c19775ef81caed27742e9d203af10b7416)

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

util-linux: fix packaging of readprofile

We rename readprofile to readprofile.util-linux so we need to use that binary
name in the FILES entry for the readprofile package.

(From OE-Core rev: 55168655ec95e8eff70f90a462ed0a8d87d8d87f)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake/event: Allow event handlers to quietly raise SkipPackage events

(Bitbake rev: 2a7c92bdadf9a86d9ea2ea0c128108e38e0e97e5)

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

fetch2/git: Allow to specify the name of the checkout directory

(Bitbake rev: 639db8c766cada7180f9447f51303f9b30d7e817)

Signed-off-by: Holger Hans Peter Freyther <holger@moiji-mobile.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts: Show sensible warning messages if expected binaries don't exist

[YOCTO #1438]

(From OE-Core rev: 6b5706d1f9ce7a3fd4d8f819ff8f3fd789665647)

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

scripts: Don't show errors from which ifconfig failing

(From OE-Core rev: 06625096f897235ed85f0d9a1355497f92938454)

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

Fix powerpc term and provide more template for different arches

(From OE-Core rev: bf9e9961ec4e7b2d10f25b550b902df62b3939b1)

Signed-off-by: Jessica Zhang <jessica.zhang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Revert "Fix adt-installer for consistent naming for powerpc and add all arch template for installation"

This reverts commit ae4ad20edbb6ea023155f7ae2b00871d479e370d.

(From OE-Core rev: 13d713dccd7fedb5a223f1292227e177d7a44164)

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

libcap: only enable pam if DISTRO_FEATURES requests it

(From OE-Core rev: 9864b0a8253922e044f61506a4a8e9064aac2bd7)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

shadow: Fix rpm failed dependencies when install

There are some links become invalid after rename, so that failed dependencies
detected when install rpm package. This commit update links to resolve it.

[YOCTO #1158] got fixed.

(From OE-Core rev: 48ec20d1331eb665d9fc1a06bdb1ea79e4513159)

Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

dropbear: Change the path to find configuration file and add a configuration file for dropbear

dropbear will check "/etc/pam.d/sshd" which comes from package "openssh" \
When enabling pam supporting. But if we only install dropbear \
package without package "openssh", then "dropbear" will not \
find a configuration file.
The changes are as follow for fixing this bug:

- Change the path to find configuration file (/etc/pam.d/sshd --> /etc/pam.d/dropbear)
- Add a configuration file "/etc/pam.d/dropbear"

(From OE-Core rev: 48dcb8fc7b669b27160dde33079f40551853702b)

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qt-demo-init: Fix bug 999

[YOCTO #999]
qt4e-demo-image needs qt-demo-init when starting qtdemoE.
qt-demo-init was pulled from Openembeded.

(From OE-Core rev: e21e8b502ab2f982836cf1f7a30e33bff1bd5b7b)

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

connman-gnome: Fix WiFi security display issue

Latest connman-0.75 change the WiFi security type, causing the
incompatibility of connman-gnome. Fix connman-gnome accordingly to show
the correct security informtion.

[YOCTO #1343]

(From OE-Core rev: 38589c37923ecfe34f1dbd5f12d89d55dfb11e2a)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gdb: Further S vs. PV fixups

(From OE-Core rev: 5c9ae426a058908be2e43d249c284e122b8a5444)

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

xserver-nodm: Disable dbus-wait as it causes users too much confusion

In the event the xserver fails to start the dbus-wait has to time out and
this causes many users a lot of confusion. If we wish to reinstate this,
we need to find a safer mechanism to do it where X failing to start cancels
the timeout (sends a dbus event at that point?) The comments are left in the
file as an example in case some user does wish to enable is.

(From OE-Core rev: 0471b17b061e57231387ef90c95fc0c34fc0e66b)

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