package_rpm.bbclass: Add support for PACKAGE_EXCLUDE to RPM installs

Using the new smart exclude mechanism an error will be generated in the
excluded package is required for the image to be generated.

(From OE-Core rev: 238c46c03d0d9c35523c78b94bfebb57904bac5a)

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>

python-smartpm: Add support for excluding package from the install

Update smart to support a mechanism for excluding specific packages from the
install process.  An error will be generated if this package is required.

(From OE-Core rev: 87660d636c2ebe76cd9dff2a334f135def9a0cf3)

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>

image.bbclass: Add basic support for PACKAGE_EXCLUDE

Add the foundation for the PACKAGE_EXCLUDE support.

As part of this work, it was noticed that the PACKAGE_INSTALL and
PACKAGE_INSTALL_ATTEMPTONLY were still using he 'normal' version for
dependencies.  This should no longer be necessary as of the change in the way
the complementary package groups (dev, dbg, ptest and others) are defined.
By making this change the dependency tree is more correct than before, and
gives the ability for manipulating PACKAGE_INSTALL and
PACKAGE_INSTALL_ATTEMPTONLY, while adjusting the dependencies at the same
time.

Warning messages will be generated if the user is trying to exclude a
package that was previously in the PACKAGE_INSTALL or
PACKAGE_INSTALL_ATTEMPTONLY variables.

(See additional commits for package manager specific support.)

Add documentation on PACKAGE_EXCLUDE and related variables.

(From OE-Core rev: 208d4d5ef7c5ead35dc27b7808f92ed377377aa4)

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>

busybox.inc: Avoid error when SYSLOG is not enabled

Add -f to rm of ${D}${sysconfdir}/syslog-startup.conf.${BPN} so as to
not error out when the busybox config used does not have SYSLOG enabled
and DISTRO_FEATURES does not contain sysvinit.

(From OE-Core rev: 4c5756149754d0b18b14595db335f8f5e14cc0a3)

Signed-off-by: Florin Sarbu <florin.sarbu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob: changes to image combo box from image configuration page

In some cases when a new machine is selected, the image combo
shows the same option. Fixed this issue, because the image
combo box should be reseted.

Fix the counter for the options in the combobox. This bug
was introduced by the templates functionality. The combo box had some
last changes, and I forgot about this counter.

[YOCTO #4858 & #5000]
(Bitbake rev: 457fd80ee6b1b2bcef463e3a83e048da2f8bf805)

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

bitbake: propertydialog.py : Scrollable windows for long tooltips

In some cases, the length of the description and the brought
in by field was too big. That led to the size of the property
dialog exceeding Hob's size. For long tooltips we use
scrollable windows now.

[HOB #4321]

(Bitbake rev: 78ecabf19bf01e5a662b6e2b865cd93bf47d962b)

Signed-off-by: Andrei Dinu <andrei.adrianx.dinu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: build.py: Provide useful diagnostics when exiting.

Running scripts with 'set -e' produces silent failures with no
diagnostic. Add an exit handler which produces diagnostics, including
details of what was running if the shell seems to be bash.

(Bitbake rev: e213e6a4c297a4f1c22eed15bd7b4cbc0e9eab4f)

Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

genext2fs: fix memory corruption on powerpc

    * fix memory corruption errors on powerpc64
    e.g.
    |$ genext2fs -b 30000 -d . -i 8192 /tmp/tstrootfs
    |  Segmentation fault

(From OE-Core rev: d98f08a7ad95d0b17846276b028a6614f16b6846)

Signed-off-by: Chunrong Guo <B40290@freescale.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-dtb.inc: Fix dtb generation for kernels newer than 3.8

The 3.8 kernel has change the default directory where the dtb file is
stored. The change has been done at:

,----[ Quote of 3.8 kernel change ]
| commit 499cd8298628eeabf0eb5eb6525d4faa0eec80d8
| Author: Grant Likely <grant.likely@secretlab.ca>
| Date:   Tue Nov 27 16:29:11 2012 -0700
|
|     ARM: dt: change .dtb build rules to build in dts directory
|
|     The current rules have the .dtb files build in a different directory
|     from the .dts files. The only reason for this is that it was what
|     PowerPC has done historically. This patch changes ARM to use the generic
|     dtb rule which builds .dtb files in the same directory as the source .dts.
|
|     Cc: Russell King <linux@arm.linux.org.uk>
|     Cc: Arnd Bergmann <arnd@arndb.de>
|     Acked-by: Olof Johansson <olof@lixom.net>
|     Cc: linux-arm-kernel@lists.infradead.org
|     Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
|     [swarren: added rm command for old stale .dtb files]
|     Signed-off-by: Stephen Warren <swarren@nvidia.com>
|     Signed-off-by: Rob Herring <rob.herring@calxeda.com>
`----

This change adds support for both places to backward and forward
compatibility.

(From OE-Core rev: 0ec3710b8dcae311e8d9d676d5f1c6843a81383b)

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>

coreutils: allow for acl support

CQID: 430353

Define PACKAGECONFIG info for acl assuming it
might be a DISTRO_FEATURE.

(From OE-Core rev: a8455a13554088613d4576a74b19294a8b49ff88)

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>

coreutils: add PACKAGECONFIG info for acl support

Updates GPLv2 version of coreutils.  They also want
patches for other packages that might use acl.

CQID: 430353

Add configuration data to enable acl support if
it is a distro feature.

(From OE-Core rev: 8030fdc198c1037ae458899eebd14a4fae04c49a)

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>

initscripts: remove obsolete device_table.txt

device_table.txt was once used by an init script named 'devices' to
create the basic files under /dev. However, it's no longer used now.
The devices init script has been removed, and makedevs command has been
disabled by default in busybox.

Besides, considering the device managers and devtmpfs filesystem, this
file is not likely to be used again. So we remove it.

(From OE-Core rev: cc3842255ade0ad6fb312f06f5e5e1edd9767c94)

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>

init-install.sh: fix to handle the boot partition correctly

Previously, the boot partition was created for the target hard drive
but there was no corresponding entry for it in /etc/fstab. Besides,
even if the boot partition was mounted, it would just result in odd
directory hierarchy like /boot/boot/grub. However, what we really need
is /boot/grub. This patch fixes this problem.

Besides, for future maintance work, this patch also renames some of the
intermediate directories. It uses more descriptive names like /tgt_root
and /src_root. The name of /ssd is dropped.

[YOCTO #5018]

(From OE-Core rev: aa67b1333b4774e1845f562085f7048df65a644f)

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>

init-install.sh: improve hard drive searching process

Previously, only unremovable hard drives are searched and are treated
as candidates of target disks to intall into.

However, it's possible that we're going to install the live image into
a removable media such as an USB. This patch enables this possibility.

In addition, this patch presents more information about the hard drives
so that user may have more knowledge about which hard drive they are
going to install their image into.

[YOCTO #5018]

(From OE-Core rev: 358f0584d779825307eec08c023b5ff14e72cf9e)

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>

tslib: update to 1.1

Drop patches merged upstream. 32bitBE-support.patch wasn't merged, but
no longer applies and similar changes look to have been made; tslib 1.1
works properly on qemumips without it, so this has also been dropped.

(From OE-Core rev: 2e92d845b433f3a1805c310ccda54cfc7dd8b1e1)

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>

default-distrovars.inc: Add 'DISTRO_FEATURES_DEFAULT' variable

When making distributions based on the default distro-less config, it
is useful to be able to reuse the default DISTRO_FEATURES options
without the need of duplication. The new variable,
DISTRO_FEATURES_DEFAULT, allow this reuse and customization.

So distros can include 'default-distrovars.inc' and use:

,----[ Use example ]
| DISTRO_FEATURES ?= "${DISTRO_FEATURES_DEFAULT} myfeature"
`----

(From OE-Core rev: 660ec04786162ff7f40aa78eb154dc4b5bf6ed9f)

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>

useradd.bbclass: add missing functions to preinst

The preinst scripts are calling bbnote, bbwarn and bbfatal functions,
but these functions are not written to the preinst scripts.

This patch writes these missing functions to the preinst scripts to
avoid the 'not found' errors.

(From OE-Core rev: ed45f00ef76d189611cda2cb922fa7eb8f86d9cc)

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>

gettext: update to upstream version 0.18.3

(From OE-Core rev: 5487373672c55d87fbe90117c802d26d99099d03)

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>

gtk+: update to upstream version 2.24.20

(From OE-Core rev: 176e5c5bfdeb2529cc40e35870ac49d75439dfeb)

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 2.36.4

(From OE-Core rev: 8414bb5a7f2780fc067f1fdc30a56b568cbb7d82)

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>

curl: update to upstream version 7.32.0

(From OE-Core rev: 4ec7eeea49be0bddb688b5bda5d423c4c7f4695d)

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>

pixman: update to upstream version 0.30.2

(From OE-Core rev: 054db541535967e573f59e32f5b8e2387020231f)

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: use trim_version() to get the source directory

(From OE-Core rev: 0c510da33a0a1a7b0bee8ce9caaf028e1235c291)

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>

hello-mod: Add comment for kernel module package naming

modules bbclass renames kernel module packages with "kernel-module-" prefix
automatically.

(From OE-Core rev: 6640f312062ad6648459f0b75086e9a8614300a9)

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

Revert "hello-mod: Ensure the produced package name begins with kernel-module-"

[YOCTO #4286]

The package runtime mapping rename issue is already fixed by
commit 0bc564af07c1bae8112f834a60aea3b72af7de13, the "kernel-module-" workaround
is not necessary anymore for out-of-tree module package, so revert that commit.

This reverts commit 71aafc214fe407b7620e747c11dfb8326c867b1c.

(From OE-Core rev: b0676d2ffceec3027aee8d2d450ae6318085341e)

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

texinfo: handle correctly @enumerate specification greater than 10

Backport the patch to fix the error when @enumerate is greater than 10:
mpatrol.texi:6356: bad argument to @enumerate

(From OE-Core rev: 9dca06b16fc10743ed879221fadb836733a9332c)

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>

uclibc: Upgrade to latest on git

It brings fixes needed for python3 to work on uclibc/64bit targets

(From OE-Core rev: 01777e78639888d437b103ebafefccd932631bfd)

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>

prelink: update to latest cross-prelink

Sync to latest cross-prelink, as of Aug 1, 2013.

Drop the PR number, no longer needed.

(From OE-Core rev: b59d256349cc0ac19357158be8e63bd52ab9fb51)

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>

qemuppc: Change default tune to 74xx

We use mac99 as platform for qemuppc
lets choose a tuning thats appropriate for it

(From OE-Core rev: 9b5572b8014f23747f18a7e0ca30c7094c524920)

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>

tune-ppc7400.inc: Add tune file

This is appropriate tune for mac99/g4 platform
that we use for emulating qemuppc

(From OE-Core rev: af10ecb57a5eb12c65975043d419f7506ef89b99)

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>

linux-yocto-3.8/meta: enable ALTIVEC for qemuppc

As part of the qemuppc tuning activities, we are aligning on ppc74xx, and
as a result we can enable ALTIVEC support in the base BSP config.

[YOCTO #1914]

(From OE-Core rev: 81c5c93fb0589dc24c10a4d3722da72d4774db22)

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>

linux-yocto-3.8/meta: update drm and intel power management settings

Updating the meta branch SRCREV to import the following config
changes:

  f706bd4 drm-emgd.cfg: convert some config options from y to m
  5995fa5 meta: features/power/intel.cfg

(From OE-Core rev: 1c9703c9e8b79f5c225c585c154416c1e7e6899f)

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>

cronie: fix out of tree build

(From OE-Core rev: fd9e591f266e1a6c183e77f24e50d31e0d52bdd5)

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

qemu: Fix bit extraction for MTFSFI and MTFSF

Power ISA 2.05 enhancements introduced regression
in mtfsfi implementation. Fixed thusly

[YOCTO #4854]

(From OE-Core rev: b8952942aed77473d3b44a17112cbf6a9e83eff3)

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

buildhistory-collect-srcrevs: match new buildhistory-diff command line parsing

* Default buildhistory directory to buildhistory/ under the current
  directory and require an option to set it
* Show a description in the help output

(From OE-Core rev: 64aa7d0b53f6ad45ab1a2121e917d7a512097407)

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

buildhistory-diff: improve command-line handling

Improve command-line argument handling of buildhistory-diff to make it
easier to use.

* Default buildhistory directory to buildhistory/ under the current
  directory and require an option to set it (since most users will
  likely run buildhistory-diff from the build directory and keep
  BUILDHISTORY_DIR at its default location)
* Default from-revision to "build-minus-1" to get the difference from
  the previous build with no arguments
* Allow from/to revisions to be specified by from..to (since git accepts
  this form).

(From OE-Core rev: 5e2be70e89820ffc74208d225fe4414fe5182050)

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

classes/buildhistory: avoid mangling names in dot graphs for images

We were mangling names here to avoid characters that are invalid for dot
node names, but if you just quote all names that isn't necessary and we
retain the original naming, allowing easy searching of the graph files.

(From OE-Core rev: 99efdad886aa5063a68912846c00a46ba9b15536)

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

classes/buildhistory: tag last 3 build revisions

This makes it easier to obtain the difference from the last three
builds particularly where a single build might account for more than one
commit in the buildhistory git repository (e.g. if package and image
changes occurred).

(From OE-Core rev: 9512f3787295d662678c2943da31197c8df39e99)

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

classes/buildhistory: show command line in commit message

Record the bitbake command line in the commit message as a further
context indication.

(From OE-Core rev: 34fb802b2170f0fe208e54e7a4d18f663db72f90)

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

classes/buildhistory: record size of installed package not compressed archive

* usually it's more important to know how much space will each
  package take on target device then size of compressed package
* example for libewebkit0 with 4 different architectures, interesting
  that om_gta02 .ipk is bigger but it's smaller when installed

  before:
  MACHINE     DEFAULTTUNE       SIZE (.ipk file)
  om_gta04    cortexa8t-neon    15996 KiB libewebkit0
  qemux86_64  x86-64            16992 KiB libewebkit0
  spitz       xscale            16148 KiB libewebkit0
  om_gta02    arm920t           16260 KiB libewebkit0

  after:
  MACHINE     DEFAULTTUNE       SIZE (installed)
  om_gta04    cortexa8t-neon    60544 KiB libewebkit0
  qemux86_64  x86-64            63720 KiB libewebkit0
  spitz       xscale            60588 KiB libewebkit0
  om_gta02    arm920t           56268 KiB libewebkit0

(From OE-Core rev: 85e4a77138381a6086d5ebd3a28cb5a94bc26a19)

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

classes/buildhistory: remove default of BUILDHISTORY_CHECKVERBACKWARDS

Since this variable is no longer used, we don't need to set a default
for it.

(From OE-Core rev: a5230835c539781b5b035dc6d0be3cac5a5bd305)

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

cronie: upgrade to 1.4.11

LIC_FILES_CHKSUM changed because cron.h doesn't exist anymore.
Replaced it with cron.c.

(From OE-Core rev: c1f3eabece0bb998c46367c06e3433696f97225c)

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

sudo: upgrade to 1.8.7

Removed a patch because the changes were merged upstream.
Also, the license had some modifications in two files.

(From OE-Core rev: 13ba4490f6422109f934ed36809bd52d44577574)

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

bitbake: hob/retrieveimagedialog: change labels

Some labels needed to be changed in order to respect the convention.

[YOCTO #4999]
(Bitbake rev: 110b485eac6adea242297b7942da8710403c465a)

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

bitbake: taskdata: report close matches with NoProvider errors

Assuming there is no known reason why an item is not provided, show
close matches on the assumption that it might have been a typo or
other mistake.

(Bitbake rev: ed81b0856b4a3892b53d39871eaaa6273390ea75)

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

bitbake: runqueue: report close matches for an invalid task name

Help to pick up mistakes such as "bitbake -c cleanstate xyz" (instead
of "bitbake -c cleansstate xyz".)

(Bitbake rev: 15c3db1cffdffd85641c6b12e77f19ce7a553472)

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

texinfo: add perl RDEPENDS and fix up packaging

This new version of texinfo uses perl for the conversion process.

The perl modules are installed in ${datadir}/texinfo which was getting
pulled into the -doc package, since FILE_${PN} is a full override, add
it back.

(From OE-Core rev: f0b6734182a5505fad7d26bc67a2141e51bc7a9d)

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

openssh: fix for read-only rootfs

If the rootfs is read-only and the ssh keys are not available at system
start-up, the init script will generate ssh keys into /etc/ssh, thus
causing a 'read-only file system' error.

In order for Yocto based image to work correctly for read-only rootfs,
we use the following logic for openssh.

If the rootfs is read-only and there are pre-generated keys under /etc/ssh,
we use the pre-generated keys. Note the pre-generated keys are mainly for
debugging or development purpose.
If the rootfs is read-only and there are no pre-generated keys under
/etc/ssh, we use /var/run/ssh as the location for ssh keys. That is, at
system boot-up, the generated ssh keys will put into /var/run/ssh.

[YOCTO #4887]

(From OE-Core rev: 2ed44745024f04aa4e00ddba3009153c6b47c8e9)

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>

irda-utils: fix for read-only rootfs

The init script for irda writes configuration items to /etc/sysconfig/irda
if that file is not available in system. But it's actually not necessary,
the behavior doesn't change whether the init script writes to the file or not.

Considering it issues error messages in case of a read-only rootfs, I delete
the writing process.

[YOCTO #4103]
[YOCTO #4886]

(From OE-Core rev: f88a101bc0caa7b486527f0d337406651cbaeb0d)

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>

populate-volatile.sh: use 'cp -a' to avoid potential problem

Previously, dead links in target directory will not be copied.
This is incorrect as dead links are not uncommon in our rootfs.
So we use '-a' option instead.

(From OE-Core rev: 742440441222e0627abbdd3eb2ee16401e8f4adf)

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>

runqemu-internal: fix to start X correctly in live images

The QEMUOPTIONS for ISOFS was not complete, leading to failures when
trying to start X in live images.

This patch fixes this problem.

[YOCTO #4103]
[YOCTO #4884]

(From OE-Core rev: 08947869917dc5a9dfff05b0ee19279f60cf6d2b)

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>

populate-volatile.sh: don't spawn background process at rootfs time

If we're building a read-only rootfs, we'll get the following error now
and then.

    tar: .: file changed as we read it

The root cause is that we spawn background process at rootfs time.
When the tar command is running, it's possible that files under rootfs
are changed by background processes, thus this error.

[YOCTO #4937]

(From OE-Core rev: 4bd419f95868d5b8707a45cee5f6c5c6a840a65b)

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>

populate-volatile.sh: use $ROOT_DIR/var/volatile/tmp as TMPDIR

It's possible that a 'No such directory' error occurs when doing
check_requirement in populate-volatile.sh at rootfs time. This is
because the $ROOT_DIR/var/tmp might be a dead link.

Use $ROOT_DIR/var/volatile/tmp as the TMPDIR instead to avoid this
error.

[YOCTO #4883]
[YOCTO #4103]

(From OE-Core rev: 7c2c36a97bab46c73a3a1fb743ad3ec67a4c072e)

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>

udev: remove implicit dependency on initscripts

At some point, the udev was modified to source the functions from
initscripts or lsbinitscripts. This dependency is actually not needed.
If we use udev in a system where initscripts from oe-core is not available,
there will be errors.

This patch fixes this problem by removing the implicit dependency.

[YOCTO #4882]
[YOCTO #4103]

(From OE-Core rev: 72d6825c24f4c3e4a7a907cf0a09e2e8f7720ae8)

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>

init-live.sh: make $ROOT_MOUNT/media writable when necessary

If the live image is mounted as read-only, we cannot make necessary
directories under $ROOT_MOUNT/media, so trying to move the mount points
lead to errors.

So in case that no unification filesystem mechanism is available in kernel
and the rootfs is mounted as read-only, we mount tmpfs on $ROOT_MOUNT/media
so that it's possible to make necessary directories under it.

[YOCTO #4881]
[YOCTO #4103]

(From OE-Core rev: aeeb3418ff08dfd29edc0ce8a41cb6887d4e11fe)

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>

device_table-minimal.txt: use user/group names instead of uid/gid

Instead of hard coding the uid/gid, we prefer the way of using user/group
names.

Note the way of using uid/gid is still supported.

[YOCTO #1159]

(From OE-Core rev: c5ef0294a9b8d178896a47c9f5d6e3dd6797e343)

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>

image.bbclass: create device table after package installation

Now that the makedevs supports using user/group names in the devcie
table files, and it uses passwd and group files under the rootfs which
is provided by the base-passwd package, we should let package installation
finish first, so that makedevs can get a correct mapping from user/group
names to uid/gid.

The check for existence of ${IMAGE_ROOTFS}/dev is removed. This is because
do_rootfs doesn't have 'nostamp' flag any more, so the do_rootfs task will
not be rerun for every build. Checking for the existence of ${IMAGE_ROOTFS}/dev
is not necessary any more. Besides, as base-files package also installs the
/dev directory, this checking does not serve as a good criteria.

[YOCTO #1159]

(From OE-Core rev: d073ca77ba886c7912abd3ec0640881c00aea3bb)

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>

makedevs: support using user/group names in device table files

Compared to hard coding the numeric group and user ids in the device
table files, the way of using user/group names is preferred.

This patch adds the ability to makedevs to correctly deal with device
table files with user/group names in them.

To maintain backward compatibility, the way of using uid/gid is still
supported.

[YOCTO #1159]

(From OE-Core rev: 1fcf718e3a1e50446ab61972069566e5016bc625)

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>

wipe-sysroots: don't assume TMPDIR is under BUILDDIR

The previous code used the environment variable BUILDDIR and assumed that TMPDIR
was a subdirectory. This often isn't the case, so instead ask bitbake where the
directories we're about to delete are.

(From OE-Core rev: 29491a72acac81ebb7e9ecfbc9392fbeb9a7ea26)

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>

qemu, default-providers: Add mesa as default virtual/egl

* it's safer to select it consistently with virtual/libgl* providers

(From OE-Core rev: d9321da1bf01175a9e3721607df31055e3765bc6)

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>

cmake.bbclass: Don't use packages from the native build machine

(From OE-Core rev: 044037d8add3556a90ac4da5991e1e5975d62e25)

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

matchbox-wm: fix icon handling on 64-bit hosts

Xlib has a historical quirk where 32-bit values are returned in longs (but
always hidden through a cast).  On 32-bit machines this doesn't matter but on
64-bit this leads to problems.  Bump SRCREV to integrate a fix from upstream.

Also, drop the conditional configure_fix.patch, it was integrated long ago.

[ YOCTO #4917 ]

(From OE-Core rev: 09c609afbfb6e9a2504166d0fe8c7ae3d8ae9ebd)

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>

libmatchbox: upgrade to 1.11

Remove check.m4 as libmatchbox uses the check.pc instead of a m4 file now.

(From OE-Core rev: 4f51bf23346947cd47e7ac9e1290e799f3c20708)

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>

xrestop: add dependency on ncurses

* fails without it
  | xrestop.c:50:20: fatal error: curses.h: No such file or directory
  |  #include <curses.h>
  |                     ^
  | compilation terminated.

(From OE-Core rev: c709fcd7e5f81707ece5f7d085c8c43eed165f7e)

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>

xeyes: add dependency on libxrender

* fails without:
  | configure:10604: error: Package requirements (xrender >= 0.4) were not met:
  | No package 'xrender' found

(From OE-Core rev: ceb69dce2cff26153c6e17614a934b9fa8f9f2a5)

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>

gst-plugins-gl: add dependency on glew

* fails without:
  | ./gstglshader.h:30:21: fatal error: GL/glew.h: No such file or directory

(From OE-Core rev: a3c58aaeb84c711431e78458afabed2f7054d937)

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>

at-spi2-core: add intltool-native dependency

* configure fails without it
  at-spi2-core/2.8.0-r0/temp/run.do_configure.372: intltoolize: not found

(From OE-Core rev: 483abbe91c1320d6c6e3e7a5824de80e427aa50e)

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>

nfs-utils: Replace spaces with tabs in shell function

As per OE-Core convention, indentation should be TAB for shell.

(From OE-Core rev: 91b6d2cf08c5c6ad58f0f9a85a3536a9034342b8)

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

rpcbind: Replace spaces with tabs in shell function

As per OE-Core convention, indentation should be TAB for shell.

(From OE-Core rev: a92376f2daf10baf45d99a3de9502b52cbce8b7e)

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

lighttpd: Replace spaces with tabs in shell function

As per OE-Core convention, indentation should be TAB for shell.

(From OE-Core rev: d73a0d00554cc197d2eb7e95432252af267d7bca)

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

opkg.inc: remove the obsolete POSTLOG variable

With the current implementation of postinst logging mechanism, the
location for log file is configured via the POSTINST_LOGFILE variable.

The POSTLOG variable is obsolete now, thus removing it.

(From OE-Core rev: 6aaf2d5c51eb4ee6acbb85fd4aa5f44406907bc2)

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>

cogl: GL/GLX feature cleanup

Separate GL and GLX flags don't make sense, as on Linux GL means GLX and GLX
means GL.

So, default to GLESv2 and GLX if X11 is enabled.  EGL on X11 doesn't have
feature parity yet so leave it disabled by default.

(From OE-Core rev: e5f8f6c6191505d731c4672353e811d6729d424a)

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>

linux-dtb: Use kernel build system to generate the dtb files

As the Linux kernel, unconditionally, builds the dtc application and
it is the compatible version with the DeviceTree files shipped within
the kernel it is better to use it and the kernel build system to
generate the dtb files.

Some DeviceTree files rely on CPP and kernel headers to be able to
generate the dtb binary contents and it is harder to replicate it
outside of Linux kernel build system so we /use/ it.

To comply with these assumptions we need to use the dtb file when
calling 'make' instead of pointing to the DeviceTree source file; the
code has been made backward compatible but it is advised to move to
the new definition to avoid warnings as:

,----[ Original definition ]
| KERNEL_DEVICETREE = "${S}/arch/arm/boot/dts/imx6q-sabresd.dts"
`----

Becomes:

,----[ New definition ]
| KERNEL_DEVICETREE = "imx6q-sabresd.dtb"
`----

(From OE-Core rev: 72980d5bb465f0640ed451d1ebb9c5d2a210ad0c)

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>

systemtap: upgrade to 2.3

Disabled javac and jar.

(From OE-Core rev: 0308bf2619734f07814f55c0adae17937f77afbd)

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

meta-skeleton: layer.conf Should also have .bbappend in example

This will allow for an example recipe with a .bbappend, such as
busybox or kernel configure fragments.

(From OE-Core rev: 2a2a2540b37880cdd866f01d8e4077cbd0378749)

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

meta-skeleton: Add busybox config fragment example

This shows an example of the config fragment support that
both the linux-yocto and busybox recipes use. This example
is specific to busybox.

By adding busybox CONFIG options into a .cfg file and then
adding that .cfg file to SRC_URI the merge_config.sh script will
correctly handle these CONFIG options during the do_configure task.
The merge_config.sh script uses a last-in wins method.

For more details about this, see the Kernel Development Manual Section 2.2.3
http://www.yoctoproject.org/docs/current/kernel-dev/kernel-dev.html#changing-the-configuration

(From OE-Core rev: 57662d4f813d5795cac1529633db80a09efdb089)

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

qemu: Skip it for mips64 target

We should not skip it always since same recipes provide
qemu-native too so user class-target override to make it
only skip for target recipes

(From OE-Core rev: 23d633cc65cf31486263f120c71df82186cfdd22)

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

dpkg: fix include header caused compile error

Build dpkg-native on Fedora 19, it fails with:

/usr/include/c++/4.8.1/cstdlib: In function ‘long long int std::abs(long long int)’:
/usr/include/c++/4.8.1/cstdlib:174:20: error: declaration of C function ‘long long int std::abs(long long int)’ conflicts with
   abs(long long __x) { return __builtin_llabs (__x); }
                    ^
/usr/include/c++/4.8.1/cstdlib:166:3: error: previous declaration ‘long int std::abs(long int)’ here
   abs(long __i) { return __builtin_labs(__i); }
   ^

That because header cstdlib is included in a 'extern "C"' block that gcc
4.8 doesn't support. Fix it by move the header file out of the 'extern "C"'
block.

(From OE-Core rev: 7de61ecc3efc43c625dde9a66f5c05e980a82e34)

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

oe-setup-builddir: substitute ##OEROOT## when using templates

oe-setup-builddir substitutes ##COREBASE## which seems like it should have the
same value as the variable ${COREBASE}.  In reality it doesn't as ##COREBASE##
is substituted with the value of $OEROOT (the location of the oe-init-build-env
script), whereas ${COREBASE} is set by oe-core to the parent directory of meta/.
If oe-core's meta/ isn't a top-level directory then ##COREBASE## and ${COREBASE}
have different values, which can lead to confusion.

To resolve this, deprecate (but still substitute) ##COREBASE## and substitute
 ##OEROOT# for $OEROOT.

(From OE-Core rev: 1890783928dd1c73105fae32fb6c588afc287ff6)

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>

procps: Add cgroup support

(From OE-Core rev: 767383ac80bd52a51d52655f95bb503cd0bb827b)

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

blacklist.bbclass: Avoid blacklist specific handle in base.bbclass

base.bbclass had code which handled the PNBLACKLIST in case of
multilib use. This is better to be done in the blacklist.bbclass so it
has all logic in a single place.

(From OE-Core rev: 04f0fefeccc1e7e3af8a1f741350492ae3f171fc)

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>

opkg: Fix possible installation of GPLv3 header

opkg is GPLv2+ licensed but it has optional support for sha256 checksums which
was GPLv3+ licensed. This code is not built unless '--enable-sha256' is passed
to the configure script, the default is equivalent to '--disable-sha256'.

However, the header 'sha256.h', which is GPLv3+ licensed, is in the list of
header files to be installed and thus could end up in the libopkg-dev package.
As this header is installed to '/usr/include/libopkg' it is very unlikely that
it will ever be used. However, if you're uncomfortable with GPLv3 code going
anywhere near your target filesystem you won't want this to happen.

The simplest solution is to replace the sha256 implementation in opkg with the
implementation from coreutils-6.9 which is licensed under GPLv2+. This is
committed to the opkg subversion repository as r652/r653.

The only intervening commit between r650 (previous SRCREV) and this is r651,
which integrates 'obsolete_automake_macros.patch' into the opkg sources. Thus
this patch isn't needed in oe-core anymore.

(Note: Before 873689bbabba25e7be5c12317c04519a7bc8d0ef, this header is only
installed if opkg is built in its source tree (ie. ${B}=${S}). After that commit
the header will always be installed)

(From OE-Core rev: 3c6a8a39d820f14f9eb3df3d719cef2c469769da)

Signed-off-by: Paul Barker <paul@paulbarker.me.uk>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kernel.bbclass: Correct post(inst|rm) package association

Fixes [YOCTO #4991]

The kernel image is installed as part of the kernel-image package, but
the symlink creation/removal via alternatives is being done in
pkg_post(inst|rm)_kernel-base.

Move the postinst alternatives logic into the kernel-image functions.

(From OE-Core rev: 35f538b117e3387354d2dab1f22c3de28ab1322b)

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

groff: Add GPLv2 version

Apparrently a LICENSE change was missed, so we add the GPLv2 version back in

(From OE-Core rev: 002818712fab5c6325b1f7205512945ea87ad76c)

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

groff: Fix License to be GPLv3

(From OE-Core rev: 4090c4d9b204118ec9826db02b0739ce60563dd4)

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

image.bbclass: Move runtime_mapping_rename to avoid conflict w/ multilib

[YOCTO #4993]

Move the runtime_mapping_rename into a prefunc for the do_rootfs function.
Otherwise doing it in the python section could occur BEFORE the multilib
classes renaming.

If the package 'b' is a kernel module, then lib32-b and b should both
point to the same package.  The runtime_mapping code will do this
automatically.

Before if you ran: bitbake lib32-<image>

It may do:

start PACKAGE_INSTALL (a b c)
remap  (a b c)
MULTILIB naming (lib32-a lib32-b lib32-c)

What we want is:

start PACKAGE_INSTALL (a b c)
MULTILIB naming (lib32-a lib32-b lib32-c)
remap (lib32-a b lib32-c)

(From OE-Core rev: 836662c9a9c175521dbcd29cdfc0a7c144d8770f)

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>

rpm: Enable compatibility with older RPM packages that have invalid platforms

Some LSB packages appear to have the platform set to '%{_target_platform}'
which is not a valid platform field.  This causes a failure of the type:

warning: package lsb-test-core-4.1.15-1.x86_64 is intended for a %{_target_platform} platform

When we detect an invalid platform, fall back and try to construct a new
platform name that may be valid based on the arch and os contents of the
package.  (This should only ever be needed by invalid or older RPM packages.)

(From OE-Core rev: 6513fa327aeb7e9fdd313290c205917952eed226)

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>

rpm: Change references from RPM_VENDOR_POKY to RPM_VENDOR_OE

Change the #define references to match RPM_VENDOR_OE.

(From OE-Core rev: a84ecc5ad158a7529a904785de25ebfedf5767a7)

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>

linux-dtb.inc: Replace /boot/ with /${KERNEL_IMAGEDEST}/

Devicetree files were installed hard-coded in /boot. When KERNEL_IMAGEDEST
is anything else but "boot", the postinstall script and the file locations
no longer match and the postinstall will fail.

Replace "boot" with "${KERNEL_IMAGEDEST}" to fix this problem, and to allow
the devicetree files to be installed in another location.

(From OE-Core rev: 0f589b9a38397fdf55025062a45889b19d1c83c4)

Signed-off-by: Mike Looijmans <mike.looijmans@topic.nl>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub_0.97: Fix to correctly ship files under /usr/lib

The grub_fix_for_automake-1.12.patch replaced pkglib with pkgdata to
make grub_0.97 build with automake-1.12. However, it forgot to set up
the pkgdatadir, thus causing grub_0.97 not shipping files under /usr/lib.
This in turn resulted in an unworkable grub.

This patch fixes this problem by setting up the pkgdatadir correctly.

[YOCTO #4997]

(From OE-Core rev: 883b1b396328e6cd67dcb4ca6fd8975b6e716c0a)

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>

lib/oeqa/runtime: add new systemd tests

New systemd runtime tests for enable/disable service, start/stop service and list services.

(From OE-Core rev: 6386dc718f85210c9b6b9f69878ec9a7847b78de)

Signed-off-by: Alexandru Palalau <alexandrux.palalau@intel.com>
Signed-off-by: Stefan Stanacar <stefanx.stanacar@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/oeqa/runtime: rework syslog test

Add separate tests for restarting syslog and using logger, and
skip the configuration test for systemd images which always fail
because syslog's systemd service doesn't read a config by default
(see YB#4860).

(From OE-Core rev: c75f3e2385dde44ee96e33f4e5d064894dfb7d52)

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

oeqa/utils/qemurunner: get ip old fashioned way and use tcp serial console

The way we read data from the serial console was unreliable and blocking (AutoBuilder
seems to hit that often), so change the serial console type from unix socket to tcp
and reverse the connection - don't let qemu act as server (wait for a connection).
So now the serial console is used to save the boot log and make sure that we reached
the login prompt. Until a better way is found this should solve some of the AutoBuilder
failures (one being YB#4904).

Also we need to use the same method as the old qemuimagetest to get the ip
(from the qemu process arguments), because that it's more reliable.
The first version used here was to log into the target and use the output of
"ip addr show eth0" but then systemd decides that it should rename interfaces,
so that was changed to get the ip of the interface that has the default gw,
but if there is no default gw we'll get the loopback ip and we end up trying to
ssh into the host machine (some recent AutoBuilder runs showed that).

Changed in V2:
 - use -ww for ps, as output might get truncated

(From OE-Core rev: 55e78185110937b7e2b143cf1020426d8df58b72)

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

classes/testimage: create a link for qemu boot log

The full name for the log is confusing when there are multiple files.
Also move the ssh log path stuff where it's needed.

(From OE-Core rev: 1dbef61a0776ec6c9ac9209442bb4c346e706d7d)

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

mesa: fix and enable out-of-tree builds

One patch (submitted upstream) for when Gallium is enabled, and another
(inappropriate for upstream) to fix out-of-tree builds with
0003-EGL-Mutate-NativeDisplayType-depending-on-config.

(From OE-Core rev: fbc7092f0ae07538d4363679b1597ba4e556d1a8)

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>

gst-ffmpeg: set SUMMARY instead of DESCRIPTION

We only have a short description, so set SUMMARY and DESCRIPTION will be
defaulted from it.

(From OE-Core rev: 33fc0caa169c8d0c707977e387bda6a23630ee12)

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>

gst-ffmpeg: enable using yasm during build

A recipe is now available for this, and it should enhance performance on
x86/x86-64.

(From OE-Core rev: 39aab68c1d60cb071bbd0c5d21b94222c89209a3)

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>

gst-ffmpeg: merge in bbappend from meta-oe

* Enable external libav for better optimisations/additional bugfixes
  (internal ffmpeg copy is quite old), default enabled but can be
  disabled using PACKAGECONFIG
* Add a PACKAGECONFIG for orc, disabled by default in line with other
  gstreamer recipes
* Bump PR to r7 so the bbappend can be dropped without PR going
  backwards

(From OE-Core rev: aab668fbba25d3e590e4182224b7b064d7705c5b)

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>

libav: add from meta-oe, update and tweak

Changes from the meta-oe recipe:
* Update stable recipe to 0.8.8
* Update git recipe to the tag for 9.8 (for now)
* Switch over to a tarball for the release version recipe
* Add LICENSE_FLAGS = "commercial"
* Set SUMMARY instead of DESCRIPTION
* Add yasm-native to DEPENDS since there is now a recipe for it
* Remove libvpx from DEPENDS and add a PACKAGECONFIG option for it,
  disabled by default since it wasn't actually being enabled
* Add a PACKAGECONFIG option for x11 to enable/disable x11grab, and
  add the proper DEPENDS if so (still defaults to enabled)
* Add a number of other PACKAGECONFIG options, replacing some old
  comments as well as offering the ability to disable x264.
* Hide text relocation warning when building for i586 (PIC can't be
  enabled for 32-bit x86).
* Drop PR

Notes for the git recipe:
* This hasn't been able to be built recently in meta-oe since there was
  a circular dependency between libav and libpostproc. libpostproc is
  part of libav 0.8.x but was split out in 9+ and is not needed at all
  anymore by libav itself, so this dependency was removed.
* Additionally the recipe was filtering out the option to enable
  libpostproc but this option wasn't being added by the inc file and
  thus the filter wasn't doing anything, so I dropped this as well.

(From OE-Core rev: 9a670f780b1f1204d426017ff9a95842ad85800e)

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>

x264: add from meta-oe, update and tweak

This is required by the default configuration of libav being brought over
from meta-oe. Changes from the meta-oe recipe:

* Update to the latest revision from the stable branch (upstream does
  not seem to provide stable releases.)
* Add LICENSE_FLAGS = "commercial"
* Enable PIC to fix text relocation warnings and disable warning for
  i586 (since PIC can't be used there)
* Make SUMMARY value slightly shorter
* Indent SRC_URI consistently with other recipes in OE-Core

(From OE-Core rev: a3ca077285003fbc04d134e875a58a745271e47f)

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>