rootfs_ipk, package_ipk: drop --force-overwrite

* it was introduced in 2010 by 8c3a7ebac8bed700bcc37f778d5a883cfeee8de8
  but doesn't seem needed anymore
* also such do_rootfs failure is good indication that runtime packages
  installed are not sane, ignoring this error in do_rootfs just
  transfers the issue to end user when he does "opkg upgrade" later
  on device
* tested on SHR images with whole feed included and only issue was
  gstreamer/gst-plugins-* locale packages (and there is patch for this
  on oe-core ML)

(From OE-Core rev: 1f7d3a2e9071ff78170c06ebf58c7400a5ee09bf)

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>

readline: apply upstream patches

Upstream does not make bugfix releases but releases patches
to apply on top of original releases. Fetch and apply those patches.

(From OE-Core rev: 30a38ea1ee933fb134a5ee9000298703cab93692)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glib-2.0: update to upstream version 2.34.3

(From OE-Core rev: db9b8ad76335613e2de1e568a14ae95c68af6288)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

initramfs-framework: udev: Handle alternative binary paths

This allows use of udev daemon for different installation destinations
so allowing use of udev's from systemd code for initramfs.

(From OE-Core rev: 96daf1b3105e17a67acb5027d0418b2ac28b6820)

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

net-tools: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. Old md5sum was calculated from
the beginning of the file, not from beginning of the license segment.

(From OE-Core rev: 5037fef73e2c4d31b06a4422ad1d54f777fe00fa)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glib-2.0: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. Old md5sum was calculated from
the beginning of the file, not from beginning of the license segment.

(From OE-Core rev: a6e2e5b592747a7e4a8465c8637154555ee020c8)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ltp: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. Old md5sum was calculated from
the beginning of the file, not from beginning of the license segment.

(From OE-Core rev: c538673caafddbb208d9d18301e84a6a891d7844)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xcb: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. Old md5sum was calculated from
the beginning of the file, not from beginning of the license segment.

(From OE-Core rev: 4e73f9ae050c01b1356c82e91a73d46cb90d5402)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libdaemon: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. Old md5sum was calculated from
the beginning of the file, not from beginning of the license segment.

(From OE-Core rev: 76c186e4bf70a9969a0ce46da2c6a3e8039d8463)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gmp: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. Old md5sum was calculated from
the beginning of the file, not from beginning of the license segment.

(From OE-Core rev: f9281283c993ab2af81f0316bac5fed7b49bdb83)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

coreutils: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. Old md5sum was calculated from
the beginning of the file, not from beginning of the license segment.

(From OE-Core rev: 883b7899fe5adce49e1c39cca9854e7f035e7a68)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

uclibc: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. This change has no effect on
md5sum as license segments from the beginning of the file.

(From OE-Core rev: 1699c382b68d183540b088fd7644b936682716fb)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

pong-clock: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. This change has no effect on
md5sum as license segments from the beginning of the file.

(From OE-Core rev: 33b28f40ed8765262c3b4ac676cec25bf0b0d523)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xorg-app: fix license segment md5sum boundary

Replaced incorrect "startline" with correct "beginline" for telling
where license segment in file begins. This change has no effect on
md5sum as license segments from the beginning of the file.

(From OE-Core rev: 9f3f13a0f761bed4110ef1013d711c1678e03741)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qt4-native: make qt4-native work with long building path

If the TMPDIR has more than 256 chars, building qt4-native fails.
This violates the 410 length limit of TMPDIR.

This patch makes building qt4-native succeed with a long building
path (410 for example) by extending its static arrays' sizes by
256 chars.

[YOCTO #2766]

(From OE-Core rev: a635553cf46a7e43e81eab1a1baeae98312881d2)

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

ghostscript: make ghostscript work with long building path

If TMPDIR has more than 256 chars, building ghostscript fails.
This violates the 410 length limit of TMPDIR.

This patch makes building ghostscript succeed by changing its
MAX_TOKEN from 256 to 512.

[YOCTO #2766]

(From OE-Core rev: 30a19b7c791abcbf09193c678866df66d470a6d9)

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

guile: touch compiled modules on sstate install

When guile is installed from sstate, if the compiled files get a modified time
older than the source files then guile will produce warnings like this:

;;; note: source file /data/poky-master/tmp/sysroots/x86_64-linux/usr/share/guile/2.0/ice-9/boot-9.scm
;;; newer than compiled /data/poky-master/tmp/sysroots/x86_64-linux/usr/lib/guile/2.0/ccache/ice-9/boot-9.go

Not staging the files and letting guile re-compile them on demand won't work:

;;; compiling /data/poky-master/tmp/sysroots/x86_64-linux/usr/share/guile/2.0/system/base/compile.scm
;;; it seems /data/poky-master/tmp/sysroots/x86_64-linux/usr/share/guile/2.0/system/base/compile.scm
;;; is part of the compiler; skipping auto-compilation

So, use a sstate postinst function to explicitly touch the compiled files after
the extraction to ensure that they are fresher.

[ YOCTO #3370 ]

(From OE-Core rev: cb8fbc332f730cbb4d9c9866ee7b95d62245d121)

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

guile: fix bashism in script

(From OE-Core rev: 3d5050d281e9535d1931848fe6c0adb4e7afd629)

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

pango: update to upstream version 1.30.1

(From OE-Core rev: 82720dd1aa2594cc308130da4db4ff600263ac74)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

initramfs-framework: init: Stop mount warnings during boot

For a completely quiet boot using recent versions of utilities, the
/etc/fstab and /var/lock need to be available.

(From OE-Core rev: d7a8154d575f918a0a20cb0e3a8f65d02ed32f30)

Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

rootfs_rpm.bbclass: normalize arch output of list_installed_packages

list_installed_packages should output uniform formatted archs for all
packaging system(deb, rpm and ipk), for they are used by common code to
grab the package infos. Otherwise, it will lead some -dbg packages to be
missed processing in come cases.

It's introduced by commit: fc985f51, package_rpm.bbclass: fix the arch
(replace "-" with "_").

Fixed by replacing "_" back to "-" in list_installed_packages, make it
output real arch values which can be safely used.

(From OE-Core rev: 554f2436435bf96c6735827f491c1b78b198c016)

Signed-off-by: Ming Liu <ming.liu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

populate_sdk_base: tarball installer: SDK overwrite warning

This patch contains two fixes:
  * if the user wants to install the SDK in a directory that already
    contains a SDK for the same architecture, a warning will be shown;
  * when the symbolic links are relocated use -n option. Otherwise,
    symbolic links to existing directories will be created in the
directory itself;

[YOCTO #3401]

(From OE-Core rev: cd035f170a2448b9ebcf0cfac5d4c0fa472b7049)

Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

subversion: update to 1.7.7

- fix WARNING: Failed to fetch URL http://www.apache.org/dist/subversion/subversion-1.7.6.tar.bz2
- subversion-1.7.6_mod_dontdothat_svnserve_only.patch doesn't seems to be useful,
  cc Marcin to get confirmation

(From OE-Core rev: 60ac9eccd6101967a89ab74344920b4b3ca8cd5f)

Signed-off-by: Eric Bénard <eric@eukrea.com>
Cc: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ccache: update to upstream version 3.1.8

(From OE-Core rev: a627d8b1ecec394a9e638975f4d39553dc3efba5)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

slang: explicity disable invoking onig module

We don't implement module onig in our recipes,
so we should explicity disable its invoke in slang
to avoid that slang's configure incorrectly tests
the existence of onig.

[YOCTO #2820]

(From OE-Core rev: 38c1b92f188b7e9b27d7ce3f6a8ffbdd7e19f90d)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

eglibc: Move compile with optimization handling to common code

The optimisation handling code is not version specific, so move the
fix code to the .inc.

(From OE-Core rev: 22dfc7d5738a20f015d9de693f0c2accfa685560)

Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libpng: update to upstream version 1.2.50

License md5sum changed only because libpng version numbers and
release dates mentioned in it.

(From OE-Core rev: b3f99e01c249c34d6083eec3159d5ba6b54df97f)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cairo: update to upstream version 1.12.8

(From OE-Core rev: 060a9d982bf5e46d30a52c3bc6f17263f750a013)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gtk+: do not prelight GtkButton in touchscreen mode

(From OE-Core rev: 08119d076853dbf5273c01bb095d2e1618475a57)

Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gstreamer, gst-plugins*: fix localdata

* all gst* packages were producing LC_MESSAGES/.mo instead of
  LC_MESSAGES/gst*.mo and it was leading to file conflicts between gst*
  packages too
* for more details see
  http://lists.linuxtogo.org/pipermail/openembedded-core/2012-November/032233.html
* buildhistory diff, confirms issue fixed
  https://github.com/shr-distribution/buildhistory/commit/f2c0888c0e08dfb33cc0cdf384621fc499d4ac04
* Thanks to Enrico for simplier solution

(From OE-Core rev: f50e2984d9411a059b86d6c158e9416fceb84c3d)

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

rootfs_ipk.bbclass: add missing --force_postinstall option

The force_postinstall option was missing and some packages were
configured on target rather than on host at rootfs time.

(From OE-Core rev: dfadfaa0b38678029ffebe14f15e2dbc148cb1fb)

Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: build/siggen.py: Avoid removing too many stamps when cleaning

The "*" part of the mask is to ensure we clean both any stamp, and any
setscene varient. It turns out we would also trample other tasks,
e.g. do_package_write could trample do_package_write_rpm. do_package also
tramples do_package_write_* but this is less of an issue since the other
tasks depend on it.

Rather than use the wildcard, we can just use a list instead.

[YOCTO #3484]

(Bitbake rev: c14d831ea3f625e9a47266a0c4e6deefc924ca5a)

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

bitbake: hob: when BBLAYERS_NON_REMOVABLE is not set hob shows an error

(Bitbake rev: a40ceda3b349c4461f4b7bc0e18cd966fff5e3cf)

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

openssh: CVE-2011-4327

A security flaw was found in the way ssh-keysign,
a ssh helper program for host based authentication,
attempted to retrieve enough entropy information on configurations that
lacked a built-in entropy pool in OpenSSL (a ssh-rand-helper program would
be executed to retrieve the entropy from the system environment).
A local attacker could use this flaw to obtain unauthorized access to host keys
via ptrace(2) process trace attached to the 'ssh-rand-helper' program.

https://bugzilla.redhat.com/show_bug.cgi?id=CVE-2011-4327
http://www.openssh.com/txt/portable-keysign-rand-helper.adv

[YOCTO #3493]

(From OE-Core rev: bdce08215396e5ab99ada5fa0f62c3b002a44582)

Signed-off-by: Li Wang <li.wang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

autogen: use pkg-config directly instead of guile-config

The autoconf macros in autogen use dpkg (!) and guile-config to determine
what/where Guile is.

If the build host has an installed guile, these can produce conflicting results.

More interestingly, if the Guile library source and compiled form have bad
timestamps (source newer than compiled) the configure scripts knows that Guile
is present but doesn't know what version it is, resulting in compile errors.

[ YOCTO #3370 (partially) ]

(From OE-Core rev: 8a4f07d5111feaa3114e039431785d6ad37529b2)

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

bitbake.conf: Change build output message to list BUILD_SYS, TARGET_SYS and NATIVELSBSTRING

The build summary is meant to reflect key configuration variables. Information
about the build system we're running on is important but currently missing
from the information displayed.

Printing TARGET_SYS removes the need to print TARGET_OS and TARGET_ARCH
and we add BUILD_SYS and NATIVELSBSTRING to show information about the
build system.

[YOCTO #3456]

(From OE-Core rev: 764cc1eb3043c84121f597d2271108b91052095e)

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

build-appliance-image: Updated to wget mixed-up commands fix

Fixes Hob network test failing inside BA.

(From OE-Core rev: 89884032c5c39d6343f7b30ed3e040052aeb87d9)

Signed-off-by: Cristian Iorga <cristian.iorga@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sysprof: Fixes undefined  reference to `rmb'

* Recipe already contains a patch for mips arch but not for mips64.
For mips64 arch 'mips' was not available in OVERRIDES, rather mips64
was there. So added the same patch for mips64 arch using mips64.

(From OE-Core rev: 5fa9f9b626daed83c8d31755040574c13ad25459)

Signed-off-by: Noor Ahsan <noor_ahsan@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: fetch2/wget: Fix for mixed-up wget commands

wget commands for check and resume were
mixed-up, leading to the following issues:

1. long running "NOTE: Preparing runqueue"
reason: objects were downloaded, not spidered on the mirror
2. Failing network test in Build Appliance, because wget 1.14
(in use in BA) will fail if a file already exists.
During the network connectivity test, index.php file was
actually downloaded, not spidered (checked for existence on
yoctoproject.org website), leading to wget failure.

(Bitbake rev: d7a5185cae975eaca50a9785c6605e895dc7bb51)

Signed-off-by: Cristian Iorga <cristian.iorga@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: fetch2/local: Fix bug introduced by expression ambiguity

The last changes introduced an error in some of the logic. Add brackets
to clarify the meaning of the expression and fix certain build failures.

(Bitbake rev: 87aea65bd5d553bd0495b0f1efe6d41d0bb2810f)

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

bitbake: hob: make some layers non removable

- there are some layers which cannot be removed; so ,I have used
a new variable called BBLAYERS_NON_REMOVABLE located in bblayers.conf,
which contains those layers which cannot be deleted
- "meta-hob" layer is added to this variable in hob code, like it's
added to BBLAYERS var

[YOCTO #3176]
(Bitbake rev: 05da1621eed4c6201cd65372d229f63ea8a44b6e)

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

bitbake: Update version to 1.17.0

(Bitbake rev: cc7fdbdc607df530f5539b162831bf9998eb48d8)

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

bitbake: fetch2: Avoid using FILESDIR in unpack

Currently there is code which uses FILESDIR in unpack to ensure
parent directories are created, leading to differing behaviour depending on
which search path is used to locate the directory.

This change standardises the code and takes the data from the fetcher in
question meaning we can standardise the code and deprecate FILESDIR.

(Bitbake rev: 1cccb3bd01ed82e4978acfef0fda1bd797eef72a)

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

bitbake: fetch2/local: Improve handling of wildcard matches

Currently wildcard matches end up working by FILESDIR being defined
in the metadata to a default of "." in FILESPATH which is hacky at best.

This patch adds the behaviour into the fetcher so its at least slightly
more explicit.

(Bitbake rev: 07b5f84133ac79aac4e939ea5f24390ad7f940a5)

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

packagegroup/allarch: Convert to use allarch class

Currently there is some odd behaviour of the packagegroup class in relation
to sstate since it sets PACKAGE_ARCH = "all" but does not use the allarch class
leading to it being undetected by sstate.

Previously it was not possible to use allarch as the recipe couldn't "undo"
settings made by the allarch class. Since this no longer happens when
PACKAGE_ARCH != all, we can use the allarch class.

This patch also fixes up one case we need to preserve TRANSLATED_TARGET_ARCH
and ensures sstate only assumes allarch when PACKAGE_ARCH is "all".

(From OE-Core rev: 591fa7c1ab9e9ff75fdce602c77ecdeda3a255d9)

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

allarch: Allow class to be included but overridden

We have cases where we'd like to inherit this class by default but allow
special cases to override it. This change makes the code of the class
conditional on PACKAGE_ARCH remaining set to "all", allowing it to be
overridden. packagegroup usage is one case this is desirable.

(From OE-Core rev: 7dd91402b719ac62b51088f234354f82bfa9c4b6)

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

base.bbclass: Drop P and PN from FILESPATH

In the interests of simplifying things, remove P and PN from FILESPATH,
instead relying on the BP and BPN versions which work in 99% of cases.

In any problematic case such as a -native only recipe, either the patch
directory can be renamed or the recipe can set FILESPATH specifically.

(From OE-Core rev: fb359583b659cda643973fa285002aaffb729a51)

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

bitbake.conf: Simplify FILESPATH

Files are very rarely, if ever placed in ${PF}. If a recipe needs to do this,
it can easily append to FILESPATH so it makes sense to drop this from the
default search path.

Equally, using FILE_DIR as part of the search path leads to 'bad' SRC_URI
entries and/or file layouts which are not preferred. I'm therefore of the
opinion we should also remove this from FILESPATH and encourage people to
cleanup any places this breaks my correcting the layouts to match the standard
or worst case adding to FILESPATH in recipes that need it.

These changes work towards making the system more friendly as users won't be
greeted with huge search paths we rearely use making the "correct" layout
more obvious.

(From OE-Core rev: 3efa13cd76bbd5611805021945fc9def88d9fd93)

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

bitbake.conf: Drop obsolete FILESDIR setting

FILESPATH is the preferred way of finding files now. Having a value
for FILESDIR which defaults to paths which will have already been
searched is pointless at best. This is the final step in letting
us drop FILESDIR support entirely from bitbake at some future date.

(From OE-Core rev: d6e5ceafcaef06b8a3f9acc2aa826a40a016f913)

(From OE-Core rev: 3bb5c6bd51c91ada7fc17451627b8954dbe9c09c)

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

sanity.conf: Increase minimum bitbake version requirement to 1.17.0 for FILESDIR updates

(From OE-Core rev: a85f990efd62e50e5ffaddf5c6d2703f18f11071)

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

automake: update to upstream version 1.12.5

(From OE-Core rev: b524595ec8feeb05aeedd360fca34536ac21faff)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tune-*: define more generic DEFAULTTUNE to share feed between machines

* this is mostly for backwards compatibility and to share binary feed
  like it was before, but now without missing different -mtune in it
* if you want to build some package with -mtune add something like this
  to your distro config
  DEFAULTTUNE_qemuarm_pn-openssl = "arm926ejs"
  DEFAULTTUNE_qemuarmx_pn-openssl = "xscale"
  be aware that if you do this you should do it also for all packages
  which depends on openssl because if you dont and you build e.g. dhcp,
  then dhcp build for arm926ejs (even with DEFAULTTUNE armv5te) will
  depend on openssl with arm926ejs, so dhcp in armv5te feed will be
  rebuild after each MACHINE switch.
* cortexm3, cortexr4, iwmmx and ep9312 are using own DEFAULTTUNE because
  they define also different -march
* shared feeds are
  armv4t: arm920t, arm9tdmi
  armv5te: arm926ejs, xscale
  armv7a-neon: cortexa8, cortexa9

(From OE-Core rev: a11bdc36a1be18cc5aa14682b2a2c9ee83141f51)

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

arch-arm: define different ARMPKGARCH when different CCARGS are used

* without this tune-xscale and tune-arm926ejs were both creating
  packages in armv5te feed, but each with different -mtune, with
  OEBasicHash enabled it was causing each package to rebuild with new
  -mtune after MACHINE switch, but that doesn't make sense with output
  stored in the same armv5te feed

* this makes different feed for each -mtune, but more generic one to be
  selected with DEFAULTTUNE

* tune-iwmmxt and tune-ep9312 were already using this, just move it
  bellow AVAILTUNES and use ARMPKGARCH_tune-foo syntax

* tune-cortexr4 and tune-cortexm3 are using armv7r/armv7m as ARMPKGARCH
  because there isn't another tune to use the same -march

(From OE-Core rev: cffda9a821a3b83a8529d643c567859e091c6846)

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

arm/arch-arm*: define ARMPKGARCH_tune-* for default tunes

* tune-foo is not valid override, for it to work I had to add
  ARMPKGARCH = "${ARMPKGARCH_tune-${DEFAULTTUNE}}"
  but that doesn't work without value defined for every supported
  DEFAULTTUNE value, otherwise it's expanded like this
  TUNE_PKGARCH (${ARMPKGARCH_tune-armv5te}te).

(From OE-Core rev: 31e4f2dee990ee7f5d7491b65565e71d7d580209)

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

bitbake: knotty/msg: Avoid usage of curses initscr/endwin to avoid terminal corruption

Using curses initscr/endwin causes screen corruption if for example you
suspend bitbake and resume it. This changes the code to use a less
invasive approach to determining colour availability on the terminal.

(Bitbake rev: 4548a8f037eaf8d47a77052acc3e9ec264ac41e0)

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

bitbake: hob/builder: Don't rerun sanity checks

Run the sanity check only once

[YOCTO #3377]

(Bitbake rev: 0db80d57d4d2b1bb97375444c439827450ff33d1)

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

bitbake: hob: showing when build fails because out of disk space

-to enable this in hob, first you have to enable this in bitbake using
BB_DISKMON_DIRS and/or BB_DISKMON_WARNINTERVAL vars
-created "restart the build" action on the build_fail_top_bar

[YOCTO #3357]
(Bitbake rev: 964ac25d153ff4da144963289a32066db0e28b89)

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

bitbake.conf: add TUNE_CCARGS[vardepvalue]

* we don't care about expression but value
* e.g. tune-xscale and tune-arm926ejs have different expression
  in TUNE_CCARGS but with the same DEFAULTTUNE the result is the same
  http://lists.linuxtogo.org/pipermail/openembedded-core/2012-September/030032.html

(From OE-Core rev: 03f1e34ea3ce80931e9c3cd2ab22824f28a7233b)

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

tune-cortexr4: fix march value

* probably copy&paste error from tune-cortexm3.conf
  commit 789dcb8e68a2ab9784ac10ab36815010c61af2fc
  Author: Richard Purdie <richard.purdie@linuxfoundation.org>
  Date:   Mon Jul 25 19:03:24 2011 +0100

    Add ARM tune file overhaul based largely on work from Mark Hatle

(From OE-Core rev: 4827232077e4a059d6aa818f89c83c42bb91949a)

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

tune-xscale: replace TUNE_CCARGS for webkit-gtk and cairo only with xscale in TUNE_FEATURES

* without this you'll get different sstate checksum for webkit-gtk and
  cairo even when you build them with DEFAULTTUNE == armv5te
* maybe this isn't needed at all anymore or if it is then it should be
  applied in arm-armv5.inc for all armv5te devices, not only xscale?

(From OE-Core rev: c51643a510da6d1c3426b3de8f18ae864cb073a4)

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

gst-plugins-base: add dep on libsm and libice

If libICE exists when this package configures, libICE and libSM are
added to the list of included libraries.  Adding libsm and libice to the
list of dependencies ensures they'll be present and makes the contents
of this package more deterministic.

(From OE-Core rev: 0d8adb1b57bf435f888c9357d40d1e2dae10d07f)

Signed-off-by: Gilbert Coville <gilbert_coville@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

license.bbclass: Collect LICENSE level packages

Some bad logic in license.bbclass misses certain package level
LICENSEs.

(From OE-Core rev: c5a171d5817233c0371e6f5b19f57f3c4b84f5ac)

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

archiver.bbclass: DISTRO is not required variable so deal with it

(From OE-Core rev: 3e7f411e6eb428f6d49a6f1a396e70a2bd1ceadc)

Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

eglibc: always compile with optimization.

eglibc fails to compile if someone tries to compile an entire image as -O0:
error "glibc cannot be compiled without optimization"
so in this case, force to use -O2 and give a note about it.

[YOCTO #3405]

(From OE-Core rev: 9ca1c6120fad5dcae1694e8e37331c1b903f1fd0)

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

python-pygtk: add gtk-types.defs into gdk.c dependence

gdk.c depends on gtk-types.defs but
gdk/Makefile.am miss this. This will cause
build error sometimes when built
with multi-jobbing, so add gtk-types.defs into
gdk.c dependence.

[YOCTO #3460]

(From OE-Core rev: 0e97ef30c3819e22f43d88e817e8a8b39ca30e5d)

Signed-off-by: Song.Li <Song.Li@windriver.com>
Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

db: update gnu-config files in do_configure()

(From OE-Core rev: c675b53b9f3f3d858e2fa93170f731656d3fc3f6)

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

strace: backport AArch64 support

This changeset updates existing OE patches to commits from upstream git
tree and adds everything needed to get AArch64 support working.

(From OE-Core rev: f67ad1c2634b3c7a46c43ebdbdffbe7a083e99ed)

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

curl: eliminate forced setting of -g0 when compiling

Do not invoke CURL_SET_COMPILER_DEBUG_OPTS in configure.ac.
This will allow debug options set in our CFLAGS to be
used.

(From OE-Core rev: ba151faad47e6874b295ebd9699ce154bc4ff741)

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

unzip: pay some attention to our CFLAGS

Makefile makes use of CFLAGS_NOOPT.  If we set that
when calling make we can enable options like -g.  The
Makefile will override any optimization to -O3.

(From OE-Core rev: 7f26794dc9f2e78ee8aed1e23752acb709345c6f)

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

zip: pay some attention to our CFLAGS

Makefile makes use of CFLAGS_NOOPT.  If we set that
when calling make we can options like -g.  The Makefile
will override any optimization to -O3.

Upstream-Status: Pending

(From OE-Core rev: df2c260f9cda2e291c72f7debe1e6d53846ce058)

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

mingetty: replace cflags hard-coded into Makefile

Add CFLAGS to EXTRA_OEMAKE to allow us control over
debugging and optimization.

(From OE-Core rev: 2e123bb1b3ee023a020798972821f32b1e0054e6)

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

iproute2: pass CFLAGS to Makefile\

Makefile computes CFLAGS, but we can see that our
defaults get included by using CCOPTS to pass them
to make.

Upstream-Status: Pending

(From OE-Core rev: 8d71f7d33c18bb0a975eb86d602bda42db4baa2c)

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

bind: make "/etc/init.d/bind stop" work

Add some configurations, make rndc command be able to controls
the named daemon.

(From OE-Core rev: 9ad8f7c2bae021269e9a8e35f9fc44a16d23bb6f)

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

procps: pass CFLAGS to make

EXTRA_OEMAKE in the recipe currently discards the
environment CFLAGS when setting CFLAGS passed to make.
We change that to include these options.

(From OE-Core rev: 67924616963fee9cb9f2405d93c74a8cc10b6eab)

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

meta-toolchain-qte: add --sysroot option to OE_QMAKE_(CC|CXX|LD)

When building a QT application using OE_QMAKE_CC/OE_QMAKE_CXX, the
--sysroot was not included and the compilation would fail. The user had
to add the option manually which was not very user friendly. This
happened only when installing the SDK in another location than the
default one. Since CC/CXX/LD had this option already included, reuse them.

[YOCTO #3409]

(From OE-Core rev: 758f56523daa7d8c8b459757c70b50421d28b8dd)

Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gtk+: update to upstream version 2.24.13

(From OE-Core rev: 77cd1218bbd0760f674811eb748037deb4478db5)

Signed-off-by: Marko Lindqvist <cazfi74@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xf86-video-omap: drop RPROVIDES/RREPLACES/RCONFLICTS

* without _${PN} suffix they are invalid:
  opkg install bash-4.2# opkg install xf86-video-omapfb
  Package xf86-video-omap-doc is already installed on root.

* it's not upgrade to omapfb driver but different driver,
  so let BSP maintainers decide which one works for their MACHINE

* without this xf86-video-omap (or -doc, -dbg, ...) can be pulled
  to image even when XSERVER clearly says xf86-video-omapfb

(From OE-Core rev: ef0f8611001c435abab60c464d31134ef027d7c2)

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

xf86-video-omapfb: revive driver which actually works and is tested on real devices

* 0006-omapfb-port-to-new-xserver-video-API.patch added to port it to
  new xserver video API
* other patches just updated headers to be able to git am them

(From OE-Core rev: 58e600ae3cc955ebe30b2756540404a267c7cb26)

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

xserver-xorg: disable dri2 too when building without glx PACKAGECONFIG

* it was enabled when dri2proto was built before xserver-xorg

(From OE-Core rev: 1668b4d4f373292ea4d611712ef56148ededfce5)

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

xf86-video-omap: add xf86driproto dependency, drop --enable-neon and improve DESCRIPTION

* xf86driproto was missing
* --enable-neon isn't supported by omap driver
* DESCRIPTION merged from meta-ti
  http://git.yoctoproject.org/cgit/cgit.cgi/meta-ti/tree/recipes-graphics/xorg-driver/xf86-video-omap_git.bb?id=410dc026f2ee24a2346e7563a83f0181c79809cf
* this driver also depends PACKAGECONFIG[glx] used in xserver-xorg,
  without it dri2.h sometimes isn't built (it is autodetected only
  when dri2proto is built before xserver-xorg) and without dri2.h it now
  passes do_configure but fails later in do_compile

(From OE-Core rev: 19951d2583b4ed08bbc871128d90aefd062e7da9)

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

perl: use the exported LDDLFLAGS in generate_config_sh script

The perl shared libraries did not have RPATHs set and that made
autoreconf fail when using the SDK. The LDDLFLAGS environment variable
was already exported in the recipe but was not used when generating the
config.sh.

[YOCTO #3338]

(From OE-Core rev: f6f5bdace473d0dd1dd5b8bdc7ebbb24fc6ee90d)

Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kern-tools: report missing config fragments by name

If a configuration fragment was missing, the previous error output
was not clear about the error:

  | [INFO] doing kernel configme
  | [INFO] Configuring target/machine combo: "standard/atom-pc"
  | [INFO] collecting configs in ./meta/meta-series
  | ERROR: could not sanitize configuration fragments
  |    errors are logged in ... linux/meta/cfg/standard/atom-pc/config.log

but we know the name of the missing fragment and can improve the error
message to be this:

  | [ERROR] kernel configuration fragment fragment 'virto.cfg' cannot be found
  | ERROR. A meta series could not be created for branch yocto/standard/common-pc/atom-pc
  | ERROR. Could not locate meta series for atom-pc
  | ERROR. Could not apply patches for atom-pc.
  |        Patch failures can be resolved in the devshell (bitbake -c devshell linux-yocto)

[YOCTO #3473]

(From OE-Core rev: 8e5dc511ffce4f4b512457dbc5d3dd510f6e4a95)

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

linux-yocto-3.4: gcc optimization config feature

Updating the meta SRCREV to include a x86 gcc optimization feature, and
its use by several BSPs:

  1c59807 meta: rangeley: Remove the shortcut path
  b5477d0 meta: crystalforest: Enable GCC inline compiler option
  ab2b874 meta: rangeley: Enable GCC inline compiler option
  8287750 meta: Add New feature for GCC optimizing

(From OE-Core rev: b042d1cce535fe88dd6f5b5c0cd1a608a27afbd6)

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>

opkg: bump SRCREV and drop applied patches

* only change upstream which wasn't in oe-core is
  http://code.google.com/p/opkg/source/detail?r=635
  and added testcase for that

(From OE-Core rev: 5fd1d515db5966f45a3b2f936f3c4225f59186e2)

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

lsbsetup: drop it

lsbsetup provides script LSB_Setup.sh to setup LSB test on image. But
script LSB_Test.sh provided by lsbtest replaces its function now. So
drop lsbsetup.

lsbsetup links /etc/localtime to HongKong zoneinfo file to make LSB
wcsftime test case pass(Yocto 1079). But the test case PASS without
this link, so remove the link.

Other 2 links are moved to packages lsb.

[Yocto 3278]

(From OE-Core rev: d991393207af90fe73de76f89ac6949ef291904e)

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

packagegroup-core-lsb: remove lsbsetup

package lsbsetup has been dropped, so remove it from packagegroup
core-lsb.

(From OE-Core rev: 9b6ca9bd8dcf7e7d3eed9f0120a4ffcac80e441f)

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

lsb: move links from lsbsetup to here

Because package lsbsetup is dropped, move the links created for LSB test
to package lsb.

(From OE-Core rev: 7659ad17799672d48c4d8614e830b7e60a1e971a)

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

gcc: Add --enable-linker-build-id

See https://fedoraproject.org/wiki/Releases/FeatureBuildId for the
benefits this brings.  As far as I can tell from searching the
discussion archives, there doesn't appear to be a reason not to enable
this, and the benefits are real.

Both the Red Hat Enterprise Linux 6, all Fedora, and Ubuntu Quantal
GCC builds are configured with this on.  I plan to use it in
gnome-ostree.

(From OE-Core rev: f3be2b1e5e100a953d6d7fbbb19a77a5c4d547e1)

Signed-off-by: Colin Walters <walters@verbum.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

insane.bbclass : Disable tests for unsafe references in binaries and scripts.

These test look for programs in / that depend on programs in /usr. After
a brief discussion in #oe, we decided these tests should be disabled so
we can focus on more serious QA issues.

If you are working on a system where / and /usr are on different partitions,
you should turn these tests back on and resolving the QA warnings.

(From OE-Core rev: 2fb58da56e8e7044de21fd10fe9164d204587236)

Signed-off-by: Philip Balister <philip@balister.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

binutils-crosssdk: do not set .interp size to 0x1000 for partial linked objects

When building the SDK, the final .interp section size should be set to a
bigger value (0x1000) in order to be able to change the dynamic loader's
path later. However, we shouldn't do that for partial linked objects
(when -r or -rU is used). That's because those objects will then have an
.interp section of 0x1000 even if it contains no data and when the final
linking is done we will end up with a "cannot move location counter
backwards" error. That's because the linker will try to squeeze all the data in
the .interp sections found in various partial linked objects into one 0x1000
bytes final .interp section.

[YOCTO #3264]

(From OE-Core rev: b25d0c5fe286e44ded46aefdcbe35ed259087759)

Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image classes: use PN for depends, not IMAGE_BASE_NAME

Some images override IMAGE_BASE_NAME in the recipe causing targets using image-{live,vmdk} to fail.

(From OE-Core rev: 7e000fef0bf917f27dcad66dd90fae6c155c4d1d)

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

gcc-cross: Explicitly depend on linux-libc-headers

gcc-cross cannot build without linux-libc-headers but doesn't explicitly depend on
it relying on the implied dependency through libc. With cases where pieces
can be installed through sstate, we now need this explicit dependency to
ensure builds with partial sstate work.

(From OE-Core rev: 65e5670ef429bb6c348decb1804e425f1c4d7c61)

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

sstate: Explicitly define populate_sysroot task relationships

Clean up and clarify the populate_sysroot task dependencies. Target sysroot
packages do need their dependencies installed, as do some target/cross
relationships. We can whitelist the *-initial dependencies as these are
never needed indirectly.

(From OE-Core rev: eeec307917234d97be2674beeadef71599fb1487)

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

sstate: Add a rule for target sysroot requirements from cross dependencies

For example gcc-cross depends on linux-libc-headers and needs it to be present
to build/work correctly.

(From OE-Core rev: 43ce7a1d86bf82d976ad241057a4207b1a340b3b)

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

build-appliance-image: Add explict dependency on do_unpack

The code called by do_rootfs explicitly needs the code obtained during
do_unpack. If built from sstate, it might not be present so we ensure
it is by adding an explicit dependency. This fixes build failures when
building from sstate.

(From OE-Core rev: 53d3c3caf1894e088ebf10fdf233cdf109b04da6)

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

scripts/sstate-cache-management.sh: Fix stamp handling after recent layout change

(From OE-Core rev: 8df3350d107322380582c807ccc92ea353b543ab)

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

scripts/cleanup-workdir: Adpat to new workdir layout

(From OE-Core rev: d967a498cb464c3858dc280db5e67b7e7b281b02)

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

Revert "kern-tools: report missing config fragments by name"

This reverts commit 46cc0d0a2f1486bf541c1a1b11075de3da396cc2 since
the revision in question isn't in the repository.

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

kern-tools: report missing config fragments by name

If a configuration fragment was missing, the previous error output
was not clear about the error:

  | [INFO] doing kernel configme
  | [INFO] Configuring target/machine combo: "standard/atom-pc"
  | [INFO] collecting configs in ./meta/meta-series
  | ERROR: could not sanitize configuration fragments
  |    errors are logged in ... linux/meta/cfg/standard/atom-pc/config.log

but we know the name of the missing fragment and can improve the error
message to be this:

  | [ERROR] kernel configuration fragment fragment 'virto.cfg' cannot be found
  | ERROR. A meta series could not be created for branch yocto/standard/common-pc/atom-pc
  | ERROR. Could not locate meta series for atom-pc
  | ERROR. Could not apply patches for atom-pc.
  |        Patch failures can be resolved in the devshell (bitbake -c devshell linux-yocto)

[YOCTO #3473]

(From OE-Core rev: 46cc0d0a2f1486bf541c1a1b11075de3da396cc2)

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

kernel-yocto: clarify KMETA branch comments

Having a meta branch is not strictly required by the tools (and
recipes like linux-yocto-custom do not have meta branches), but the
comments in the kernel-yocto.bbclass could lead someone to think that
it was required.

This commit clearifies the comment to the following:

  # We can fix up the kernel repository even if it wasn't a bare clone.
  # If KMETA is defined, the branch must exist, but a machine branch
  # can be missing since it may be created later by the tools.

[YOCTO #3422]

(From OE-Core rev: 421a2e2523a8f2c461479a1c0d44908cc1eaca6b)

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