image.bbclass: Enable the complementary install to be called w/ globbing params

If the image.bbclass is called with arguments, and these arguments are not
"populate_sdk", they will be passed in as the expected GLOBS.

This enables external components and scripting to use the
rootfs_install_complementary code.

(From OE-Core rev: f44c5f227a170290f567d0a0a24baaa870048788)

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

sstate: Append to EXTRASSTATEMAPS and add comment

Appending to EXTRA_SSTATEMAPS is better than just hardcoding a value. Also
add a comment about why this is necessary.

(From OE-Core rev: d4f4a57b8d564d57256017d937ed2eabf94c36ae)

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

bitbake: hob/packageselectionpage: restore selected packages

After "Cancel" action, selected packages are restored to default.

[YOCTO #2984]
(Bitbake rev: 81b0c0cd15cbd61285e6525f482412051371ea4c)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob: rename task -> packagegroup in recipe selection

This changes the filtering to use the inheritance of
packagegroup.bbclass to determine if a recipe is a package group.

Also makes the tab tooltip text generic; these recipes could come from
any enabled layer, not just the default ones.

(Bitbake rev: a3bf87a90198bf6127663c27d8be086dab04aaf9)

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

bitbake: hob: ensure error message text is properly escaped

Our poor implementation of markup escaping was causing invalid
markup, leading to the error dialog being blank. Use the glib markup
escaping function provided by PyGTK+ to do this properly and avoid the
blank error dialogs.

Partial fix for [YOCTO #2983].

(Bitbake rev: 563ea5233a5ab1629c51e802d04280692f96c596)

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

bitbake: hob: use correct semantics for dealing with pkgdata

Some of these values may or may not be overridden on a per-package
basis, so handle them accordingly.

(Bitbake rev: 56cee6a958843b03c5389d4a45245a04d1e03327)

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

classes/packageinfo: use better method to check if package exists

Instead of using a rather error-prone method of looking for output
package files in order to determine if a package got created, use the
.packaged file within pkgdata.

This fixes two separate issues:
* Some packages apparently not being found by this code e.g. all
  apm/apmd packages when using ipk packaging.
* Buggy implementation of this checking code which triggered an
  exception during the event handler if PKGV was overridden on a
  per-package basis (as it is with external-sourcery-toolchain), which
  blocked Hob from completing parsing at 99% - fixes [YOCTO #2651].

(From OE-Core rev: 48169c6bc44c546cecaa06207b6c36da558b81f7)

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

bitbake: tinfoil: Add file inadvertently not committed

(Bitbake rev: 44a3fb49e817be641090d5d1bce7b586af407d71)

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

packagegroup-core-x11-xserver: remove redundant PACKAGES statement

(From OE-Core rev: 8d16cbe934291557a26e61266417febcb2e8dfba)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

eglibc-initial-2.16: add kconfig-frontends-native to depends

* fixes:
  | make[1]: Entering directory `/OE/shr-core/tmp-eglibc/work/armv5te-oe-linux-gnueabi/eglibc-initial-2.16-r8+svnr20393/eglibc-2_16/libc'
  | make[1]: *** No rule to make target `/OE/shr-core/tmp-eglibc/sysroots/x86_64-linux/usr/bin/conf', needed by `config'.  Stop.
  | make[1]: Leaving directory `/OE/shr-core/tmp-eglibc/work/armv5te-oe-linux-gnueabi/eglibc-initial-2.16-r8+svnr20393/eglibc-2_16/libc'

* it's because, eglibc-initial.inc overwrites DEPENDS from eglibc_2.16.bb
  $ grep DEPENDS eglibc_2.16.bb
  DEPENDS += "gperf-native kconfig-frontends-native"
  $ grep DEPENDS eglibc-initial.inc
  DEPENDS = "linux-libc-headers virtual/${TARGET_PREFIX}gcc-initial"

  and it's included after eglibc_2.16.bb
  $ head -n 3 eglibc-initial_2.16.bb
  require eglibc_${PV}.bb
  require eglibc-initial.inc

(From OE-Core rev: 8616e16ea0f9536c431e203e19d7bdff6ca867bb)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: cooker: fix handling of exceptions during exception handling

If an exception occurs during handling another exception we were
getting a useless traceback such as the following, after which
BitBake froze:

ERROR: Command execution failed: Traceback (most recent call last):
  File "/home/user/poky/poky/bitbake/lib/bb/command.py", line 84, in runAsyncCommand
    self.cooker.updateCache()
  File "/home/user/poky/poky/bitbake/lib/bb/cooker.py", line 1207, in updateCache
    if not self.parser.parse_next():
  File "/home/user/poky/poky/bitbake/lib/bb/cooker.py", line 1694, in parse_next
    logger.error('Unable to parse %s', value.recipe,
AttributeError: 'exceptions.TypeError' object has no attribute 'recipe'

Fix this to print an actual traceback of the exception and exit
gracefully (well, as gracefully as possible under the circumstances).

The general fix for [YOCTO #2977].

(Bitbake rev: 675b237a284dff84e972546774b69e2f89afb360)

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

bitbake: fetch2: fix malformed URL causing a useless traceback

The implementation of NoMethodError and MalformedUrl was broken - if you
just set self.args in an exception class to a string it treats it as a
list and then fails later on with a TypeError due to the number of
arguments not matching up.

This nasty exception during exception handling was breaking the normal
exception flow (fixed separately), which meant that if you had a
malformed URL or invalid protocol in SRC_URI you would get the
following:

ERROR: Command execution failed: Traceback (most recent call last):
  File "/home/user/poky/poky/bitbake/lib/bb/command.py", line 84, in runAsyncCommand
    self.cooker.updateCache()
  File "/home/user/poky/poky/bitbake/lib/bb/cooker.py", line 1207, in updateCache
    if not self.parser.parse_next():
  File "/home/user/poky/poky/bitbake/lib/bb/cooker.py", line 1694, in parse_next
    logger.error('Unable to parse %s', value.recipe,
AttributeError: 'exceptions.TypeError' object has no attribute 'recipe'

A specific fix for [YOCTO #2977].

(Bitbake rev: 9d4150d99051d24ff218e8a43664ceaf524b19c7)

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

bitbake: hob/imageconfigurationpage: remove or_label reference

When or_label was removed, I forgot to remove also the references to it.

[YOCTO #3010]

(Bitbake rev: 4d208aaedd60f79a4277f501fdbf8c2afc32c250)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/oe/sstatesig.py: add signature data query function

Add a function that can be used from BitBake code which will find
signature data (sigdata/siginfo) files based on specified criteria, and
hook it into BitBake as bb.siggen.find_siginfo.

(From OE-Core rev: 9f0453c29891e32f8038c4bbc22ada28bfbf818a)

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

yocto-bsp: remove packagegroup-core-tools-profile.bbappend

The functionality previously added by these bbappends was already
handled in task-core-tools-profile.bb (now
packagegroup-core-tools-profile.bb), so remove this.

(From meta-yocto rev: e999a6639a711f5c9a64c69d6b89fb478566d34a)

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

runqemu-internal: New qemu calls it qemu-system-i386 for x86

New qemu calls the x86 system emulator to be qemu-system-i386
which is consistent now so change it in scripts

(From OE-Core rev: b1ccf0202ba66f9be76463df177f11719ab589e8)

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

qemu-git: Move to tip of git past 1.2 release

There are a lot of armv7 and sh4 fixes that
its worth moving to latest version. The patch
forward porting can happen later.

(From OE-Core rev: 1b91e597f3550c35605d6b15fd958376e3dde93d)

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

qemu: Update from 0.15 to 1.2

Forward port the patches which were not applied upstream

(From OE-Core rev: 0c1328a27881f1b3046ed527447608a9fa91b1ea)

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

Add package wget and perl modules.

LSB needs wget to download packages but wget provided by busybox doesn't
support some options such as '-N'.

LSB perl test 4.1.6-2 case all/tst_perlModPresent.pl,
../lib/Class/ISA/t/00_about_verbose and ../lib/Class/ISA/t/01_old_junk.t
fail because of lack of these modules, add them to make test pass.

File CORE/config.h which is provided by perl-dev and file
unicore/version which is provided by perl-doc are required by LSB perl
test cases.

Add perl-dev and perl-doc to packagegroups-core-lsb.

[Yocto #3030 #3031 #3052 #3054 #3055]

(From OE-Core rev: ac4a60a1c585bfe5bdce1556303d49bef2594070)

Signed-off-by: Kang Kai <kai.kang@windriver.com>
Rebased for packagegroup change -sgw
These perl libraries are being added directly to OE-Core for 4.1
LSB Complainace, when 5.0 comes out early next year (2013), we will
remove these changes.

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

libi18n-collate-perl: add it

This module is deprecated from perl 5.003_06.
But LSB perl test 4.1.6-2 still test it.
So add it.

[Yocto #3031]

(From OE-Core rev: 8228aa1f6bc1ee5ecddd78ce43d5ebfc0eed2d3c)

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

libfile-checktree-perl: add it

LSB perl test 4.1.6-2, all/tst_perlModPresent.pl 1 fails with:
test 88 'use File::CheckTree;' failed

Add it to fix this issue.

[Yocto #3031]

(From OE-Core rev: af5141135d0888ea1f72cd63c454bc82884ec567)

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

libenv-perl: add it

Just add it for LSB 4.1 perl tests.

[Yocto #3031]

(From OE-Core rev: 16be305db800c976e6241973e0cf176b954de8e4)

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

libdumpvalue-perl: add it

LSB perl test 4.1.6-2 case all/tst_perlModPresent.pl 1 fails with:
test 44 'use Dumpvalue;' failed

So add libdumpvalue-perl to fix it.

(From OE-Core rev: 0141967e4a721586c2c04252aae3b8e3732f53c3)

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

libpod-plainer-perl: add it

LSB perl test 4.1.6-2 case all/tst_perlModPresent.pl 1 fails with:
test 190 'use Pod::Plainer;' failed

Add libpod-plainer-perl to fix it.

[Yocto #3031]

(From OE-Core rev: 0c26ced44326280f65f574c6ffb2dc1392a2f79e)

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

perl: package modules Pod-Html and Tie-Hash-NamedCapture

Package modules Pod-Html and Tie-Hash-NamedCapture.

Without module Tie::Hash::NameCapture.pm, call "use English;" will fail.

Module Pod::Html is required by LSB perl test 4.1 case
all/tst_perlModPresent.pl 1.

[Yocto #3031]

(From OE-Core rev: e8d4386b48e169f126dd2fa018b3f44c8a42eef8)

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

libclass-isa-perl: add it

perl module Class::ISA is needed by LSB 4.1 perl test, so add it.
When LSB 5.0 release, it will be deprecated. But we need it now.

[Yocto #3030]

(From OE-Core rev: f2e8d7670d926aa88c3c38996d1a216812741dd2)

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

u-boot: Use fw_env.config if available.

* Add support for board specific fw_env.config file if available.

(From OE-Core rev: 5b6a8e987f9039aca63d97bc42421b78acab8a3d)

Signed-off-by: Franklin S. Cooper Jr <fcooper27jr@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

core-image: allow root login when debug-tweaks is enabled

This allows root to login over ssh with an empty password just like
dropbear when the debug-tweaks are enabled, it's important to disable
debug-tweaks for a production system as this will leave open a security
hole!

Thanks to Marc for the settings.
Cc: Marc Ferland <marc.ferland@gmail.com>
[Yocto #3078]

(From OE-Core rev: 13e6aa8bba6ab1ebba1efa23f94af379a8fcb6a9)

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

xserver-xorg: make DRI/GLX options respect opengl distro feature

If the distro doesn't have the opengl feature there's no point building the DRI
or GLX support, making the mesa-dri build dependency optional.

(From OE-Core rev: 73d02f6b121c8b0ed2d42de0bfd6c227fd4de41f)

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

xserver-xorg-lite: remove

Building xserver-xorg and not installing the DRI and GLX modules (and so not
Mesa) results in an increase of 16kb compared to this package.

This isn't worth the effort of maintaining two packages.

(From OE-Core rev: 586835801a11e514a10228be957713e1ce90dd44)

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

autoconf: use --warning=cross only if supported

Adapt autoconf to pass --warning=cross to automake only if
supported.

[YOCTO #842]

(From OE-Core rev: 16d1c8f076378d0878f332f83b7e1f5fcf16447d)

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

machines/x86: Drop redundant glibc configure knobs

nptl and thereby tls are not optional anymore

(From OE-Core rev: 1a4b277e47a8d624cde4c73713d036e230f3a523)

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

kconfig-frontends: Create symlinks for conf and mconf

eglibc calls out mconf and conf directly so lets create
symlinks to point to their kconfig- counterparts

(From OE-Core rev: 5857772e285ad6e39b7216bc542189e45f6fdbf7)

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

eglibc: Enable kconfig for option management

(From OE-Core rev: 13e2ccf6f4e71d674583894750f70865ebe5e4d1)

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

eglibc-2.16: Add kconfig infrastructure

This will let eglibc use kernel like option
management through kconfig

(From OE-Core rev: 4282b86072fd5a916d0d12082d6ba575bce691f2)

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

arch-armv4.inc: On armv4 add --fix-v4bx to linker flags for kernel

(From OE-Core rev: 2092e08ba81595c6aaedca8237f6717409eb53b6)

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

pixman: ignore NEON, IWMMXT, LOONGSON_MMI variables for class-native

* pixman-native can have different do_configure sstate checksums if it's built with armv4t machine and armv7a
  OE @ ~ $ bitbake/bin/bitbake-diffsigs
    before-mgmt/stamps.1346795706/nokia900/x86_64-linux/pixman-native-*.do_configure.sigdata.*
    after-mgmt/stamps.1346801508/om-gta02/x86_64-linux/pixman-native-*.do_configure.sigdata.*
  basehash changed from 27e577de60880a788c7aaba797ef83e0 to c6799807eb3e767daf1e75738fc753f7
  Variable NEON value changed from   to  --disable-arm-neon
* so if you start building with different machine then last time (wrt
  NEON setting) all recipes which depends on pixman-native will be rebuilt too
* this explains why sstate-cache-management.sh wanted to remove many
  native sstate packages when --stamps-dir option was used (see comment
  28 in https://bugzilla.yoctoproject.org/show_bug.cgi?id=2897)

(From OE-Core rev: 0b466e6677208aeefdfa15aa37bd4681eda166c8)

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

pixman: merge meta-oe append into oe-core

* neon configuration settings included
* patches were aligned to 0.27.2.

(From OE-Core rev: 97c547f3efc4bfd801a24f189ee3f38e5a017fb7)

Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package_rpm.bbclass: fix incremental rpm image generation

* Check ${target_rootfs}/etc/passwd rather than
  ${target_rootfs}${rpmlibdir} to make sure that it has been previously
  installed.

* Remove the "--nodeps" when incremental image generation, it should
  take care of the dependencies. Still use "--replacefiles --replacepkgs" in
  case there are conflicts.

[YOCTO #3047]

(From OE-Core rev: 2b3df2ec7979a49842df172be442a8794fe68fff)

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

lsb: update version

Update package lsb version to be same with current lsb test suit
version. Because when install the suit, it warns that need lsb
version >= 3.0 at least.

Drop the duplicated creating files under /etc/lsb-release.d.

Provides directories /etc/opt and /var/opt that they are required by
package lsb-dist-checker in lsb test suit.

(From OE-Core rev: 973e615ab4ee325ab568f84e001a5724f4b0dd01)

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

ltp: update

Update to latest version 20120903

(From OE-Core rev: e14a9921928c774d1817704a0a606d3ac7e4f989)

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

qemuimagetest: modifying the scenario file as per new test cases

modified scenario files.

(From OE-Core rev: dfd2ead41846c568d251a47c4baa2d9666e0c98f)

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

qemuimagetest: add basic function to check syslogd

one test case to check syslogd is executing on target.

(From OE-Core rev: 9286ea7a4eb85ba559d48135458f3b94da7a3866)

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

qemuimagetest: add basic function to check enough disk space

one test case to check disk space availability.

(From OE-Core rev: d7b549a72a91db41d8b7084b4b3efa162a62a880)

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

qemuimagetest: basic function to check bash exists

 one test case to check  bash command availability on qemu target.

Signed-off-by: veerabrahmam <veerabrahmamvr@huawei.com>
(From OE-Core rev: 357478b624b27fdfce25b6064b0f64717db75fa6)

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

bin_package.bbclass: binary package recipe class

This is used for the binary package recipe, it's been suggested that it
would be a useful feature to be able to easily take an RPM or similar
containing a software binary from a 3rd party software vendor and
integrate it into an image created by the build system.

* Brief introduction
  - The binary pkg can be .rpm, .deb, .ipk and other formats which can
    be unpacked by bitbake fetcher.

  - Let bitbake unpack the bianry package, just like unpack the source
    package.

  - Skip the do_configure and do_compile.

  - Install the files to ${D}

  - Other steps are similar to the source package's recipe.

* Note:
  - The "subdir" parameter in the SRC_URI is useful for the binary
    package recipe, so I added an example in the comment.

  - I have sent a patch to bitbake-devel mailing list to support
    unpack the .rpm, .ipk, and .deb files.

[YOCTO #1592]

(From OE-Core rev: 7037f52909b8226d2afed4ac73c902d410afc112)

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

package_rpm.bbclass: fix the arch (replace "-" with "_")

rpm can't use the "-" as the arch, which causes problem, e.g., when
MACHINE = "beagleboard":

* The arch should be armv7a-vfp-neon, but rpm only takes the armv7a,
  this is incorrect since it is mixed with real arch armv7a.

* The nativesdk's arch should be i686-nativesdk (or x86_64-nativesdk),
  but rpm only takes the i686 (or x86_64), this in incorrect since it is
  mixed with the arch i686 (or x86_64).

Replace "-" with "_" when rpm package and the rootfs generation would
fix the problem, I think this is fine since it doesn't change the tune's
arch, the package manager doesn't care about the arch's name, but it
needs a unify arch system to avoid confusing. This is similar to what we
have done on the deb which fixed the arch i486, i586 and so on to i386.

[YOCTO #2328]

(From OE-Core rev: fc985f511da86400e4fa7d17555216c12eb51666)

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

distrodata: replace Tab with Space

(From OE-Core rev: adb241958f125cc4c74ac5fbfc00674e7cd7305d)

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

classes/mirrors: remove bogus gnutls mirror

This mirror entry which maps to itself plus a slash, if matched, put the
fetcher into a circular loop until the stack space is exhausted. A patch
has been sent to fix this issue in BitBake, but we should remove the
bogus entry as well.

(Note that this entry does not actually trigger the issue with current
master because the gnutls recipe now uses GNU_MIRROR instead of
ftp.gnutls.org, thus the bogus mirror entry is not matched.)

(From OE-Core rev: 0de1827a9601143b090f751ea702fdb65a936b77)

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

bitbake: fetch2: replace double slashes in paths in encodeurl()

This ensures that if all a MIRRORS entry does is add a slash, this does
not result in a circular loop.

Fixes [YOCTO #3073].

(Bitbake rev: 57055d337a2c9997a6e5d5bdabaec396e3e128e9)

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

bitbake: hob: print message when DISPLAY is not set

If DISPLAY wasn't set, launching hob has printed a traceback
difficult to understand. Now, the exception is caught and it
shows a human message.

[YOCTO #2596]

(Bitbake rev: a41098a2dacbd903422ccdcd1885b0f351c7ddf3)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: Hob: change view of 'recipes' and 'packages' tables as ui design

changed the order of task tables,
cancel the 'description' column,
add the binb total number indicator, and so on

[YOCTO 2195]

(Bitbake rev: 6dc3263d60a6d35f9eebfcdbc2665201ee40b953)

Signed-off-by: Liming An <limingx.l.an@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: bitbake-diffsigs: allow specifying task & follow deps recursively

Add the ability to compare the two most recent runs of a specified task,
and follow dependent hash changes recursively. This enables you to trace
back and find exactly why a task was re-run after the fact.

Note that this relies on the metadata providing a function, hooked in
as bb.siggen.find_siginfo, which allows searching in the appropriate
places to find signature data files.

(Bitbake rev: cc70181659c07e04c205e17832846acf1ff31d28)

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

bitbake: lib/bb/siggen.py: make signature dump/compare functions return a list

These functions become a little bit more reusable if they return a list
containing the output rather than just printing it.

(Bitbake rev: a0ad2a947b71abcc0a1244cf139b9e9dfd8ee049)

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

bitbake: lib/bb/siggen.py: insert a colon between class and recipe name

before:
  virtual:nativeautomake_1.12.1.bb.do_compile
after:
  virtual:native:automake_1.12.1.bb.do_compile

This separation ensures that the key is readable, and if necessary,
parsable.

Unfortunately this invalidates previous native sstate signatures with
OE-Core - not much that can be done about that; however that occurs
frequently during the development cycle so it's par for the course.

(Bitbake rev: 5b96c32dad256090e9bda5af0f80c7dbcc90bde8)

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

bitbake: lib/bb/siggen: replace tabs with spaces

We had one section of the code mixing tabs with spaces, which is
particularly undesirable with python code.

(Bitbake rev: 8eaa093b179e03a6003a47220540b1bc73afca17)

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

bitbake: tinfoil: create simple interface for bitbake-based utilities

The code to initialise BitBake within bitbake-layers should be useful
for other utilities that need to query configuration or recipe
information, so refactor it out into its own class, "Tinfoil" (to
continue with our cooking metaphor).

(Bitbake rev: e5707e3938ace47c4a8d1fa2e81583fd4dc6b95d)

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

bitbake: crumbs/hig: Save toolchain in settings

Toolchain was not saved in the settings dialog ("Output" tab).

[YOCTO #2695]

(Bitbake rev: f8924b75d6ff7f093d73f4e3c0953e349960d5ff)

Signed-off-by: Bogdan Marinescu <bogdan.a.marinescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob: The 'run image' and 'deploy image' dialogs text and alignment corrections

-changed the text shown by both dialogs text
-make small tweaks to alignment

[YOCTO #2999]

(Bitbake rev: b193db13472908b8ec6c670da96ff3b0004e635b)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob/imagedetailspage: "Image ready" icon appers only after the image was generated

Now, "Your image is ready" icon doesn't appear when you come back on Imagedetails
page. It appears only after the image was generated.

[YOCTO #2984]

(Bitbake rev: de29bfc163471e4959483493a5e5b26f8a2cf8a0)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hobwidget: Button theme is taken from host, fix

- All buttons in the interface inherit a BaseHobButton that
  use the gtk settings for buttons from the host;
- Removed 'or' label between actions on image details page

[Yocto #3011]

(Bitbake rev: 1a8356b57f906cf575612eb52fc8d3a9824ff9a7)

Signed-off-by: Ioana Grigoropol <ioanax.grigoropol@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: fetch2: unpack rpm, ipk and deb binary package

* Unpack the ".rpm" binary package (only .src.rpm in the past)

* Unpack the .deb and .ipk binary package, their unpack commands are the same.

* This is useful for binary package recipe.

[YOCTO #1592]

(Bitbake rev: de7ceb9459574f33920ccc06255b533434f0ec25)

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob/recipe&packageselectionpage: Change "Back" button to "Cancel" button

"Back" button placed on RecipeSelectionPage and PackageSelectionPage
was changed to "Cancel" button to avoid any confusion.
Also, it was placed next to the other buttons on the page.

[YOCTO #3012]

(Bitbake rev: 1785b49a1b0b9698851d6e8aea94d1d2aa22c445)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob/imageconfigurationpage: Set secondary action for "Edit image" button

The image configuration screen should have only one primary action.
"Edit image" button has now secondary action, and also I have removed
"or" label.

[YOCTO #3010]

(Bitbake rev: f54191dac18b4e1100944cc6da86705c1e9c1683)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob/packageselectionpage: Add tooltips to 'Included' and 'All packages' tab and 'Search' field

For the 'Included' tab: "The packages currently included for your image"
For the 'All packages' tab: "All packages that have been built"
For the 'Search' field: "Enter a package name to find it"

[YOCTO #2322]

(Bitbake rev: 0828f352419127fb30dc4eb5f91feba84ea59202)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob/packageselectionpage: Correctly restore previously selected packages

"Back" button from "Packageselection" page now restores correctly previously
selected packages list. Till now "Back" button was implemented just to switch
pages, not to cancel changes you have made to packages list.

[YOCTO #2984]

(Bitbake rev: 1ad03d6a327eb3389f7b4d0d74d2e8ae8b50c3b6)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob2: remove class hcc

Because class hcc is useless, remove it.

(Bitbake rev: 08d4a0f76542e05755c298b3875ea373e5512e13)

Signed-off-by: Kang Kai <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob2: remove the hard-coded images map

[Yocto #2795]

When a new image type added, the hob will crash because the new type is
not in the hard-coded image dictionary.

For most of the image types, they are same with the image file's
extension name. So use variable "IMAGE_EXTENSION_difftype" to map the
image type which is diff with the image file extension name, such as
type "live". And the variable(s) will be set in image_types.bbclass.

(Bitbake rev: e7c84f056af9c613920d5adcd078a011e0387193)

Signed-off-by: Kang Kai <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image_types.bbclass: add variable for Hob

Add a new variable "IMAGE_EXTENSION_live" for Hob to map image type
"live" with real image file extension names.

This is for Hob to remove the hard-coded maps.

(From OE-Core rev: fe0973df7c72b1acec7feae03a4e13c1f49c8b1f)

Signed-off-by: Kang Kai <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

rsync: Add md5sum and sha256sum for the rsync_2.6.9

rsync_2.6.9 is the only rsync recipe in openembedded without GPLv3 license, but it lacked the
checksums for the fetcher.

(From OE-Core rev: 1016edcadb22c1655e1a902601118ec3c7332fea)

Signed-off-by: Martin Ertsaas <mertsas@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

conf/tune: add tune-ppce300c3

It has been pointed out several times that the yocto mpc8315e-rdb
reference was using the wrong tuning (603e), since it is actually
a e300c3 board.

This commit creates a e300c3 tune file based on the e300c2 variant
already in oe-core.

This commit also inhibits altivec in flac when this new tuning is
enabled and used by the mpc8315e-rdb

[YOCTO #1192]

(From OE-Core rev: 8663c7ba0530eb36728fe524ed0137e064cc1c5a)

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

gst-plugins-good: disable (uninstalled) examples

The examples pull in a GTK+ build dependency, so remove that too.

(From OE-Core rev: eef5cca4f364545759d8cf624f38e3cd6e8fd295)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/core-image: improve comments listing IMAGE_FEATURES

Ensure this listing is complete. (Comment change only.)

(From OE-Core rev: f9df0be3790ca837e43c41633d17e6ee851cb01d)

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

pulseaudio: remove ConsoleKit dependency

ConsoleKit is a runtime dependency for the ConsoleKit module, but there isn't a
build-time dependency.

(From OE-Core rev: 3e05e52821e6042708a37421423601e9fe59116e)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gssdp: build examples which require GTK+ only if we have X11

They just need GTK+ not libglade, so fix the depends.

The recipe also had multple checksums, so remove the old ones.

(From OE-Core rev: 68f320b2c4c6a6e284e6f6f24791a197f3464f36)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc-cross-initial: Reinstate limits.h generation fix

The removal of gcc-cross-intermediate inadvertently reintroduced a
problem with the limits.h file being incorrectly/inconsistently
generated.

http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=b0faebd1f07e1616004bd19664395932e7c2c48f
reverted part of:
http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=c8815d2f21849deb9359706f54dc71490773415e

This reintroduces the protected sysroot ensuring the limits.h file is
always correctly generated. To reproduce the bug, build gcc-cross, then
rebuild gcc-cross-initial and it will reference the limits.h file from
gcc-cross in the sysroot.

(From OE-Core rev: 71854f5c8c6850ed37777cad21acc92fdbea32fc)

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

matchbox-session: Should have high ALTERNATIE_PRIORITY than mini-x-session

By having a high priority it will be installed instead of the mini-x-session
when both are in a give set of packagegroups, as it stands now the
mini-x-session is installed by default on sato which is wrong!

(From OE-Core rev: d9fde0aaab26cbd657ffa56781754c85eec6f83a)

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

adt-installer: Allow changing YOCTOADT_REPO

YOCTOADT_REPO within adt-installer.conf was hardcoded to
http://adtrepo.yoctoproject.org/${SDK_VERSION}

This allows it to be overriden in local.conf by setting
ADTREPO = "http://adtrepo.yoctoproject.org/${SDK_VERSION}"

I'm also fixing a bad rm in do_populate_adt. The file name
that is being rm'd is wrong.

(From OE-Core rev: 610f8c0c0cab8e258919689fd641faec28b58caf)

Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

pulseaudio: fix pulseaudio-server RDEPENDS

* module-cork-music-on-phone was renamed to module-role-cork
  http://cgit.freedesktop.org/pulseaudio/pulseaudio/commit/?id=3c5cc345472302b9511c19244b3eceb4a3674d8c

(From OE-Core rev: 15a089fdc55faa0e2b14bb6ba019732e6608e2a5)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xserver-xorg-common: add RREPLACES on xserver-xorg to -module-exa

As the file moved from xserver-xorg to xserver-xorg-module-exa, the latter needs
to "replace" the former to ensure a smooth migration when upgrading the
packages.

(From OE-Core rev: 45a4e2064991555570f346e0662a57e61c947acf)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gsettings.bbclass: fix whitespace

As nothing in oe-core uses gsettings, it wasn't noticed that the whitespace is
wrong.

(From OE-Core rev: 1c2afda021f7abf88bc08400beba53c603912382)

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-pygtk: Upgrade to 2.24

This is needed for the build appliance and Hob also

(From OE-Core rev: e177b0dcf9b39130d4a2a4dd1ea5af72b3cc87a3)

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

packagegroup-core-x11: split machine specific parts to separate recipe packagegroup-core-x11-server

* add new packagegroup-core-x11-server to SIGGEN_EXCLUDERECIPES_ABISAFE,
  so that recipes depending on it are not rebuilt after every machine
  swtich
* allows to remove task-x11-server and task-x11 from meta-oe without
  loosing any functionality
* be carefull with default XSERVER value which does not have
  xf86-input-mouse and xf86-input-keyboard)
* VIRTUAL-RUNTIME_xserver_common which defaults to x11-common in oe-core
  and xserver-common in meta-oe's task-x11

(From OE-Core rev: a398c96706c119f298f57a929a317fcf8e0f5b92)

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

packagegroup-core-x11: add upgrade path from meta-oe task-x11*

This recipe was structured in order to replace task-x11 from meta-oe,
which it hopefully can now that a runtime upgrade path is in place.

(From OE-Core rev: 3b428debe9fdd4c9197f5ee7ab34fc9886f4a3cb)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

build-appliance-image: fix for packagegroup-apps-console removal

(From OE-Core rev: ca8ff97080b635ff9ba21c2a2f6d33cb211f7a83)

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

packagegroup-(base|core-basic): add RREPLACES/RCONFLICTS

These are needed to support upgrading from task-* to packagegroup-*
within existing target images at runtime.

Note: these settings will very likely be moved to a separate inc file
at some point in the future.

(From OE-Core rev: 1b07bd4c6b1ff70267b97f94b25026a8f8ad3314)

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

linux-yocto-rt: 3.4.9-rt17 remaining changes

The first import of 3.4.9-rt17 only picked up part of the import due
to a merge conflict with the modified files, and then SRCREVs being
generated from the pre-merge commit.

This SRCREV update picks up the rest of the modified files for full
-rt support.

(From OE-Core rev: d1ec0b64018dafaa6e47233b3465fbe4189e6280)

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

classes/qmake_base: support linux-gnuspe/linux-uclibcspe TARGET_OS

Fix borrowed from OE-Classic. This should fix build failures during
do_configure of Qt applications with the p1022ds machine from
meta-fsl-ppc, for example.

(From OE-Core rev: a19fc8e19a6cc6885a1e0616b1f42cc49c8f2c9f)

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

sanity.bbclass: Move back to running at ConfigParsed time

If we don't do this, users can get extremely confused errors since the sanity tests
happen too late (after parsing) and don't see the warnings.

Also cleanup messages about merging configuration file changes to give better
hints at where the updated files may be.

(From OE-Core rev: 7b8a68cda7ef8186e834b39e73ee12a55b33f85b)

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

build-appliance-image: Update to latest poky master

(From OE-Core rev: da9b66ddde02ec19f3f83f23eb2b595af9a57520)

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

packagegroup-*: add RREPLACES/RCONFLICTS

These are needed to support upgrading from task-* to packagegroup-*
within existing target images at runtime.

Note: these settings will very likely be moved out to a separate inc
file at some point in the future.

(From OE-Core rev: 5dd4de086d744c81d2275077ad0e0022204b0a68)

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

packagegroup-core-standalone-gmae-sdk-target: allow dbg package

meta-toolchain-gmae is still referring to the dbg package so we still
need to create this.

(From OE-Core rev: c65f8af02ebba371f96119baef0520c6685c7167)

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

scripts/cleanup-workdir: change autoconf-nativesdk to nativesdk-autoconf

the autoconf-nativesdk package name was recently changed to nativesdk-autoconf which in turn broke the cleanup-workdir script. Changed the package to the correct name.

(From OE-Core rev: 0df6904f8050c4305d90d9bb4c9cac279508d9ca)

Signed-off-by: Jack Mitchell <jack.mitchell@dbbroadcast.co.uk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto-rt: update to 3.4.9-rt17

Bumping the 3.4 kernel to -rt17.

As well as picking up the normal set of bug fixes, this update fixes
a boot issue that was introduced during merging of kernel.org -stable
updates into the -rt branches (in the timekeeping code).

(From OE-Core rev: 0c060a81ade685aee5492e0273f199850dffaf76)

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

udev: Remove files moved to udev-extraconf

(From OE-Core rev: 92f8605b81d373b66224b15355226f93ac5b3a46)

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

linux-yocto/3.2/3.4: arm: Fix linking errors with binutils 2.23

Updating the 3.2/3.4 linux-yocto SRCREVs to pickup the following fix:

    arm: Fix linking errors with binutils 2.23

    We end up with linking errors on arm using latest binutils
    because we are linking asm generated objects which use -march=all
    generating different attributes into object and use some c objects
    e.g. misc.o which use different march therefore generating a
    different set of attributes into object. When linking is done
    the ld complains since it finds incompatible attributes and ends
    up with errors like

    error: DIV usage mismatch between arch/arm/boot/compressed/head.o and
    output
    error: DIV usage mismatch between arch/arm/boot/compressed/misc.o and
    output
    error: DIV usage mismatch between arch/arm/boot/compressed/decompress.o
    and output
    error: DIV usage mismatch between arch/arm/boot/compressed/lib1funcs.o
    and output

    This patch fixes it by providing correct march to assembly routines
    which than matches with output of ld.

Signed-off-by: Khem Raj <raj.khem@gmail.com>
(From OE-Core rev: e67ffb708522a622d149641ea09b249259ed3fb4)

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

linux-yocto/3.4: v3.4.9, unionfs, perf and configuration changes

Updating the SRCREVs for the following updates:

  - v3.4.8, v3.4.9

  - bug fixes:
      a4c7a04 perf: add SLANG_INC for slang.h
      e32ab98 unionfs: restore FD_* utility defines
      9f0fe58 unionfs: update unionfs fork to align with mainline
      b3fcc2f codel: refine one condition to avoid a nul rec_inv_sqrt
      ef7cf65 fq_codel: should use qdisc backlog as threshold
      3f0ed1d net: codel: fix build errors
      4c06aa1 net/codel: Add missing #include <linux/prefetch.h>
      2888bf2 codel: use u16 field instead of 31bits for rec_inv_sqrt
      6edeb2e codel: use Newton method instead of sqrt() and divides
      2faecd6 fq_codel: Fair Queue Codel AQM
      edca95f netem: add ECN capability

  - config/BSP updates:

      463299b meta: bump kver to v3.4.9
      6b961c8 CrystalForest: Enable PCI IOV feature
      4bc2238 meta: Add new pci iov feature
      70346e6 mpc8315: remove now obsolete CONFIG_MISC_DEVICES

(From OE-Core rev: 696252de3a8e5c1bdc0e2c8c2c92cdeefd0babe2)

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

kern-tools: fix non-inheriting branch names

Importing the following tools SRCREV:

    kgit-meta: exclude explicit branches from name calculations

    kernel branches are constructed during patching of the tree by
    constructing a '/' based hierarchy of names as each branch
    directive is encountered.

    But if a "branch $name $branchpoint" is used, the entire branch
    name is supplied so no additions to the hierarchy should
    happen. As such, that type of branch command should not be part
    of branch name calculation and preparation.

(From OE-Core rev: a3884938233c8a2d6861b1d4e6be5b9824d3b131)

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

linux-yocto: check for bad SRCREV before branch checks

do_validate_branches ensures that the desired SRCREV is at
the tip of every branch that contains the revision. To do this,
it looks for containing branches and processes them.

This processing was mistakenly placed before the check for an
invalid SRCREV, hence a git error message is seen in the log
if a bad SRCREV is used, rather than a clear message.

reordering the checks, and fixing a check for master, ensures
that clear messages are generated in all cases.

(From OE-Core rev: 7e4518bf88af53b09536a3bafcd4c392a094023f)

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

linux-yocto: fix unapplied patch error message

When patches fail to apply, the status of all pending patches should
be exported to the logs and to the user. Currently, a missing export
of GUILT_BASE makes it look more like an internal error, than a 'normal'
patch failure:

    | [ERROR] unable to complete push
    | pending patches are:
    | Patches directory doesn't exist, try guilt-init

With this variable exported, we have this:

  | [INFO] validating against known patches  (qemux86-standard-meta)
  | error: patch failed: Makefile:2
  | error: Makefile: patch does not apply
  | To force apply this patch, use 'guilt push -f'
  | [ERROR] unable to complete push
  | pending patches are:
  | links/files/0002-makefile-patch.patch

(From OE-Core rev: b2b2512cbc4196fa0f814be3677517dab30e5b52)

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