systemd: depend on libgcrypt

Although libgcrypt support is optional, its development files are
are always required to expand AM_PATH_LIBGCRYPT.  Build might fails else
with

| libtoolize: copying file `m4/lt~obsolete.m4'
| configure.ac:446: warning: macro 'AM_PATH_LIBGCRYPT' not found in library
| ...
| configure.ac:446: error: possibly undefined macro: AM_PATH_LIBGCRYPT
| autoreconf: .../autoconf failed with exit status: 1
| + bbfatal 'autoreconf execution failed.'
| + echo 'ERROR: autoreconf execution failed.'

(From OE-Core rev: c7fe1d0085339e55566989b96d4d48412f0fa47d)

Signed-off-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev: disable systemd support

If we want systemd support we use systemd's udev, so disable systemd
support in this udev to avoid packaging the unit files.

(From OE-Core rev: 36aa37e693bcb76a96761847dbef6a015d96fd98)

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

udev: drop dependency on udev in libudev

libudev previously depended on udev, but this causes problems with multilib
if the user wants to install two variants of libudev as they'll pull in two
variants of udev, which will conflict.

Instead, remove the dependency and rely that the image pulls in udev in some
way, such as the commonly used packagegroup-core-boot.

(From OE-Core rev: 6168ea4e00fd1c8296fe770fa9e2ef00018f5621)

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

xserver-xf86-config: set DefaultDepth to 16 for qemumips

Apparently, the default qemumips color depth was set to 8 and the colors were not
displayed properly. cirrusfb driver doensn't seem to accept color depth
as a kernel parameter, so we have to do it here.

[YOCTO #4340]

(From OE-Core rev: 876e020be334a9350094dbd1a29b9e49eceed603)

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

perf: Ensure license is handled correctly

The do_populate_lic task has a race with the recipe since it relies on the
kernel being populated in the sysroot. This patch adds in the explicit missing
dependency.

[YOCTO #3534]

(From OE-Core rev: 578937213c6db9fa99981778ce61210e391b19a7)

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

psplash: Fix multilib build

The update-alternatives for multilibs are broken and nothing provides psplash in a
multilib build. This fixes the multilib code.

(From OE-Core rev: db1f6b24ddb2a19fb16b2ebb948bb3274b7ac295)

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

package_ipk: Ensure the status file exists

The postinstall for the opkg run-postinst hook checks for the existence of this
file. We therefore ensure it always exists during image generation.

(From OE-Core rev: 540cf355599e555615ed9684c3b480463588eb78)

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

package.bbclass: Add useradd variables to PACKAGEVARS

THe USERADD_PARAMS and similar group variables are package specific variables
which should get added to the vardeps of the packaging process. This
change also ensures they get remapped correctly by the multilib code.

(From OE-Core rev: 805add18e7b917aaef61c82c0b5f2e9682af1d0d)

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

multilib: Ensure we map the USERADD_PACKAGES variable

If we don't do this, multilib packages don't have any code
added to the postinstalls to handle user additions.

(From OE-Core rev: b10d17d1b03fd0564103a6998f218d0968d1032b)

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

gst-plugins-bad: Disable neon, its not on DEPENDS

Otherwise this setting "floats" and leads to non-determinstic builds.

(From OE-Core rev: 4eea01e9d775f2e56958a8bb194aadd8a242946d)

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

build-appliance: Update revision to dylan release

(From OE-Core rev: 2453504004cfb6c9b5030b557f421e580802a280)

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

systemd: Fix udevd init script so sysvinit in systemd mode works

Commit http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=f76d4b3549ca220fa4bf84db2756ab45e11d06a3
moved volatiles handling for /run to the udev code only. This breaks
sysvinit+systemd combined systems when building sysvinit images.

This patch hacks the udevd init script in systemd to provide
the missing symlink allowing the sysvinit images built with udevd
from systemd to work correctly.

(From OE-Core rev: 6c6e5dc0baccd0904f785b1f80f39b5f530779cf)

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

util-linux: fix systemd service in multilib

Use ${PN} when specifing service files so that they continue to get packaged
with multilib.

(From OE-Core rev: d394d27b66cbeb4e57017d2a49605243586477ac)

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

systemd: fix -dbg package with multilib

When libdir isn't /lib the files that are in nonarch_base_libdir were being left
out of the -dbg package.

(From OE-Core rev: 60bb2f795a0b3e83de5861185446992fb91ffba0)

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

busybox: hard-code the syslog service file

Previously this was using ${PN}-syslog.service, which changes with multilib.

(From OE-Core rev: ee36f997e68f5a995baf361191cef7a46ff51203)

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

udev: explicitly package nonarch_base_libdir

When using multilib this doesn't get caught by the default FILES_${PN}, so add
it.

(From OE-Core rev: 53f1540780e5d4e8035118a1920202735bd0a370)

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

systemd: Effectively revert the move to /sbin

This effectively reverts the move systemd's libexecdir to /sbin.  This caused
too many issues in other places and was not well enough tested this close to
release.

Rather than use base_libdir, the nonarch version is used to try and give some
chance of multilib not being totally broken in the release.  Also some variables
are used that mirror the systemd build system to reduce the risk of using
variables that "work" in the general case but will break with multilib.

(From OE-Core rev: e921d43e213f93da176fd27e48b557f802443dec)

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

seperatebuilddir: add systemd

(From OE-Core rev: 39f9e48558003601d4eec181e32fc76b68df5b5a)

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

systemd: fix out of tree builds

(From OE-Core rev: 79b7c56cb5d6a0b75e7df01fd3cfef4111bf69ae)

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

udev: remove explicit path to udevadm

(From OE-Core rev: cc0f22cd1e93cc25647add1a3339e150572e4fce)

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

alsa-utils: Use pkg-config instead of hardcoded udev paths

(From OE-Core rev: b92a3e9d093bc9421aa38a40bc6bfd559a16b3be)

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

packagegroup: Add init-manager sanity check

Currently, you can set VIRTUAL-RUNTIME_init_manager to an init system that
isn't in DISTRO_FEATURES. This leads to head scratching over unbootable images.

This adds a sanity check which ensures more valid systems are built.

(From OE-Core rev: c72ec4b52827f75351790eab483d258b2e87611a)

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

udev: Effectively revert the move to /sbin

This effectively reverts the changes made in b49ddeb11c0319b7ece278fd7cef6b0223114ca9
to move udev's libexecdir to sbin. The changes caused too many issues in
other places and were not well enough tested this close to a release.

Rather than use base_libdir, the nonarch version is used to try and give some
chance of multilib not being totally broken in the release.

(From OE-Core rev: c50e50fdafad378d75b7b74259a3d55ca0fe3d18)

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

pulseaudio: Set udev variables using pkg-config

(From OE-Core rev: 2c948d1cfdd070c8ce6bf0ee940a345a31407663)

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

bluez4: Set udev variables using pkg-config

(From OE-Core rev: b8b275c263493e114baefcef94c68743d89329b7)

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

pcmciautils: Use pkgconfig to set udev variables correctly

(From OE-Core rev: e177bad3c5f2d37fe7a0092c22cc3dfc556f3dea)

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

package.bbclass: Fix populate_packages for glob expansion issues

If we put a valid glob like "*/foo/*" into FILES, populate_packages
breaks with a "file exists" message. This is because the glob expansion
does not have "./" prefix however there may already be an entry in
the seen list which does have such a prefix. The easiest/simplest fix
right now is to add the prefix if it doesn't exist which only happens
for certain globs.

(From OE-Core rev: 138c5f3af892e33f576fc7dd268e122b179f82a1)

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

busybox: fix mount issue

When the mount command is not given the filesystem type to mount, then
it will try all the known filesystems. However, when a filesystem is not
supported by the kernel, the mount function call will return ENODEV.

The following patch, ecd90bc6aa63da2aef2513ac090d4b426b2b719c,
introduced a problem because it bailed out on ENODEV too. Instead it should
have only bailed out on ENOMEDIUM.

[YOCTO #4308]

(From OE-Core rev: 82320d2074572477f26887eddc51dc2a1dfea403)

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

augeas: Fix missing files with separate build directory

${datadir} was empty when building ${B} != ${S}, this patch addresses that
problem and means the -lenses package is no longer empty.

(From OE-Core rev: ee3b0d8fbe42a3c813039850886b5bf97544eb15)

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

libpam: backport patches from upstream

Backport patches from linux-pam git repo to fix test case
tst-pam_pwhistory1 failure.

[YOCTO #4107]

(From OE-Core rev: 65e4a9f050ae588ec794808315a206d94ca7a861)

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

consolekit: Fix ${S} != ${B} issues with pam enabled

When pam is enabled an extra plugin is build and it references an incorrect directory
when separate build directories are used. This patch corrects the issue.

(From OE-Core rev: a6e91db198ae5b77986de25331631d894c7b97d1)

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

systemd: Fix shell variable assignment spacing

(From OE-Core rev: e389d47524297b3def665a9a3668b6de09e5d20a)

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

systemd: use AC_CHECK_TOOL instead of AC_PATH_TOOL when checking objcopy, strings, gperf

* with external toolchain I have different HOST_PREFIX and HOST_SYS
  AC_PATH_TOOL is using HOST_SYS as prefix and fails to find objcopy
  then it tries objcopy without prefix which is found on host, but
  that objcopy does not work for arm libs
* with internal toolchain gperf is not prefixed with HOST_PREFIX, but
  fallback to "gperf" only finds the one in native sysroot first
* based on http://lists.linuxtogo.org/pipermail/openembedded-core/2013-April/037985.html

(From OE-Core rev: 6ae99136f1c1c59f4e55331c43cf0c0bac5abcdd)

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

separatebuilddir.inc: disable libmusicbrainz

(From OE-Core rev: 7863a3c7f2d619c63ddb25883a774af6c9eb3143)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.8: Makefile: add -grecord-gcc-switches if using -mfentry

    Commit a2546fae [ftrace: Add -mfentry to Makefile on function tracer]
    adds support for using -mfentry when possible, whenever the ftrace
    CONFIG_FUNCTION_TRACER option is enabled.

    This unfortunately causes bogus dwarf debuginfo to be generated:

      gcc Bug 54793 - the location of a formal_parameter is not started from a
                      function entry with -mfentry

      http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54793

    This can be fixed by applying the gcc patch above, which will be
    present in gcc-4.8, but in the meantime, the problem is currently
    being worked around in different ways by various tools.  One of those
    tools that we support in Yocto, Systemtap, requires the DW_AT_producer
    string contain -mfentry, which means that it's assuming
    record-gcc-switches is also used:

      SystemTap Bug 15123 - workaround for bad debuginfo for -mfentry
      http://sourceware.org/bugzilla/show_bug.cgi?id=15123

    The lack of -grecord-gcc-switches and therefore a missing -mfentry
    string in Yocto leads to the following Yocto bug:

      Bug 4099 - Crosstap script check fails
      https://bugzilla.yoctoproject.org/show_bug.cgi?id=4099

    Which is what this patch intends to fix.

    It only affects architectures that HAVE_FENTRY when
    CONFIG_FUNCTION_TRACER is turned on and in any case is just adding to
    the debuginfo, so there doesn't seem to be much risk to adding it.

(From OE-Core rev: c04f611aa4bf77867fe78a005d09434226feb62c)

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

qt4-embedded: bump PR to allow removal of meta-oe bbappend

The meta-oe bbappend only increments PRINC, if we bump INC_PR here we
can remove that.

(From OE-Core rev: 9e74a1c9cb753a426b55544cc41860edb98e8d33)

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

packagegroup-qte-toolchain-target: bump PR to allow removal of meta-oe bbappend

The meta-oe bbappend only increments PRINC, if we bump PR here we can
remove that.

(From OE-Core rev: 7aadfe2f293da665236aede5cb03f801d420d050)

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

qt4-x11-free: bump PR to allow removal of meta-oe bbappend

The meta-oe bbappend only increments PRINC, if we bump INC_PR here we
can remove that.

(From OE-Core rev: 976a06f41558d2246173414658d9f7b8eee193d6)

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

wpa-supplicant: don't call DBus init script directly

After installing Avahi we need DBus to reload it's configuration.  In a
pure-systemd image there isn't a DBus init script to reload, so cut out the
middleman and just sent SIGHUP to all running dbus-daemon processes instead.

(From OE-Core rev: d6fb028de172bb649b905b605f6ddc8402af859a)

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

hicolor-icon-theme: bump PR to allow removal of duplicate from meta-gnome

The recipe in meta-gnome has PR = "r1" so make it the same here and
avoid PR going backwards for existing users of meta-gnome when the
recipe there is removed.

(From OE-Core rev: 200685bde7ea222891e1dde27f086caea9216e48)

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

avahi: don't call DBus init script directly

After installing Avahi we need DBus to reload it's configuration.  In a
pure-systemd image there isn't a DBus init script to reload, so cut out the
middleman and just sent SIGHUP to all running dbus-daemon processes instead.

(From OE-Core rev: b3e468294a0723b3ceafe2022bf9d735eee64678)

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

xserver-xorg: fix segfaults for ARM SoCs

(From OE-Core rev: f45a03b510479f6c27b24a0fd8d0c661674495e1)

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

runqemu-internal: Drop distcc support

The distcc support is clearly unused and broken, might as well drop the
remaining code fragements.

(From OE-Core rev: 1a70a3225947aa45f3e1f377d50a5865aac64d2b)

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

cairo: fix builds with libpng 1.6

libpng 1.6 emits more warnings than before, and is also stricter with function
ordering.  Fix the function ordering when reading PNGs, and stop treating all
warnings as errors.

(From OE-Core rev: 55d00b750f5c45e583abef406c96416cd6a8caa7)

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

README.hardware: Update mpc8315 load address

The kernel size has grown to a point where the original load address might not work
due to memory region overlap. This updates the instructions to use larger memory
addresses which avoids the problem.

[YOCTO #2430]

(From meta-yocto rev: 6b7cd4aa50f91c4ebb8a61c9d7fea616be63f470)

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

initramfs-live-install*: fix the "install" boot option

The install boot option was giving the following error when one tried to
install the live image on a  permanent storage of a BSP.

cat: write error Invalid argument
Installation image failed
sh: can't access tty: job control turned off

Further digging into the issue, found out that the install script was trying
to do this:
 cat /proc/mounts > /etc/mtab

And in the base-files recipe the /etc/mtab is made soft link to /proc/mounts.
So the cat command was failing to write on /etc/mtab. As the contents of
the /proc/mounts is already reflected in the /etc/mtab file due to the
symlink-ing, there is no need for this step to recreate /etc/mtab in the
install script. So just removing this unnecessary step, which solves the
install issue of the live images.

Fixes this bug:
[YOCTO #4229]

(From OE-Core rev: f8663eac872882e94d956b1b604304e92b865766)

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mesa: don't fail if x11 isn't available

Patches backported from mesa Git and from mesa-dev mailing list.

(From OE-Core rev: f704bb42062f2ac15edaad36497a8d2815b8b8b2)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

openssl: update range information in man-section.patch

do_patch failed after upgrading to openssl-1.0.1e. Log:
| ERROR: Command Error: exit status: 1  Output:
| Applying patch man-section.patch
| patching file Makefile.org
| Hunk #1 succeeded at 160 (offset 26 lines).
| Hunk #2 succeeded at 626 (offset 19 lines).
| misordered hunks! output would be garbled
| Hunk #3 FAILED at 633.
| 1 out of 3 hunks FAILED -- rejects in file Makefile.org
| Patch man-section.patch does not apply (enforce with -f)
| ERROR: Function failed: patch_do_patch
| ERROR: Logfile of failure stored in:temp/log.do_patch.14679
| ERROR: Task 646 (virtual:native:openssl_1.0.1e.bb, do_patch) failed with exit code '1'

Change-Id: Ib63031fdbd09443e387ee57efa70381e0aca382c
(From OE-Core rev: ded738b945b6af6c73a1c5f1b4cd5ad1b6ac06c0)

Signed-off-by: Ting Liu <b28495@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kern-tools: fix custom repository BSP generation

Updating the SCRCREV to pick up the following fix

    updateme: use absolute path for generated BSP descriptions

    When a custom BSP is used, a top level BSP is generated by the tools and fed
    to the build system just as a user defined BSP would be located and
    passed. The location of the generated file is placed in the top_tgt file,
    which is used by subsequent stages. A relative path was being placed into
    top_tgt, which binds the build to a particular directory structure and
    working directory.

    The location of parts of the build have changed, and this relative path is
    no longer accurate. Changing it to an absolute path solve the build issues
    related to custom BSPs.

(From OE-Core rev: 2d7b2478a3d48a5686afde790c378ee2f69b8e59)

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

coreutils: add realpath to alternatives

New coreutils (8.15 onwards) build /usr/bin/realpath, which busybox also builds.
Add it to the alternatives handling to avoid file conflicts.

(From OE-Core rev: 826c18eb1c69f5e3689b5b0ef188f74ae930a050)

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

connman: Fix wrong INC_PR reset

Other recipe versions in other layers
may be using connman.inc, so by resetting
INC_PR they go backwards in version.
Set the INC_PR correctly.

(From OE-Core rev: 2d56006cd88233c7f1f15a44f72a04bd0e441b25)

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

initramfs-live-boot: explicitly depend on udev-extraconf

init-live.sh depends on udev performing automounting, which happens in
udev-extraconf.  Explicitly depend on it so that we always have it installed.

(From OE-Core rev: a608d74e69ca1efe5f2b176c000fb8212797d056)

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

bitbake: data: fix performance regression

BitBake commit 7c568132c54a21161de28907159f902462f1e2bb resulted in a
fairly serious performance regression during parsing, almost doubling
the time taken to do a full parse and almost certainly impacting
performance during building. The expandKeys function is called
frequently, and if we avoid using keys() and instead just use the normal
variable lookup mechanism, performance is restored.

(Bitbake rev: 034b392e9877309f15940b258fc2c16f16fb40b5)

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

bitbake: fetch2: rename file with bad checksum instead of removing it completely

* this can be useful when someone wan't to compare old file with
  bad checksum and new one

(Bitbake rev: 33c6b93597dd43ab03ce7b62ba3eeb1893a68c38)

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

bitbake: hob: Refine sorting mechanisms in Hob

-refine sorting functions for each column from recipe list page
and package list page
-sort correctly size column from packages list page
-set default sroting order and secondary sorting criteria
-make included on included recipes/packages no sortable

[YOCTO #2346 & #4194]
(Bitbake rev: 56822176e3d5e613654a46c983d2f979d7a9eebc)

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

bitbake: monitordisk.py: disable inode checking for btrfs

The btrfs doesn't have static inode, so disable the inode check for it,
the previouse patch has set it:

minInode = None

But this is incorrect, the minInode is just a temporary variable, it
should be:

self.devDict[k][2] = None

[YOCTO #3609]

(Bitbake rev: 7e45149c292bc92314af9b42962fde8f603a179f)

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

systemd: fix bootup with 'ro' by re-adding util-linux-mount to RDEPENDS

When using busybox mount the filesystem won't get remounted as 'rw' and bootup will fail.

This fixes the regression seen after switching from 'danny' to 'dylan'.

(From OE-Core rev: 15a91361a0b5a44161d2dbcf2be6240d86123ad9)

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

busybox: Add inetd related files

Add back inetd and inetd.conf files which are needed if
CONFIG_INETD is enabled in the defconfig. Grabbed these
files from oe-classic

This patch is based on the previous patch for denzil:
http://patches.openembedded.org/patch/33235/

(From OE-Core rev: 929c738787b6f513ce235ed5f7753408a570d632)

Signed-off-by: Ting Liu <b28495@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: detects customized configs when do_install

After using 'bitbake -c menuconfig busybox' to customize defconfig,
do_install fail to detect the changes. Grep configs in ${B}/.config
instead of ${WORKDIR}/defconfig.

(From OE-Core rev: 5c088c2d3b23026752649d077ee44fe3dbe26aa4)

Signed-off-by: Ting Liu <b28495@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

allarch: Drop various problematic allarch usages

In each of these cases allarch is used where the package in question has a
dependency on things which are not allach and change when MACHINE is changed.

This leads to a rebuild of the package each time MACHINE is switched and
the sstate checksum changes. The dependencies in question are not suited
be being marked as ABISAFE.

(From OE-Core rev: 087a680429efa713a98fbb89f927b046fe07f87c)

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

nfs-export-root: Update to use packagegroup naming

task packages were renamed to use packagegroup so fix this reference.

(From OE-Core rev: bcd68f11e479e8a3a95793ab2ed65202c0f71d84)

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

qemuwrapper-cross: Inhibit default dependencies

This wrapper script doesn't need a compile or any of the default system
dependencies so lets inhibit them. This also stops the script
being rebuild every time the toolchain changes.

(From OE-Core rev: 7f72562d274dd2c37ce9262bb7fb8a8a6a9a37df)

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

encodings: Set RDEPENDS correctly

The .inc file sets RDEPENDS for the general font case but the dependencies
don't apply to this recipe. This removes those dependencies, simplifying the
dependency chains a little.

(From OE-Core rev: 71b3a156c11d01565f546f33e3f1e1bea2fafdff)

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

ttf-bitstream-vera: Use fontcache class for postinstall

Using the fontcache class means we can run the postinstall at build time
so this is generally more efficient.

(From OE-Core rev: 00cc684885efa555f7eac7653482f72095b1c443)

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

update-alternatives: Ensure DEPENDS is correct in multilib case

Ensure that the DEPENDS we're adding is correct in the multilib case by
including MLPREFIX, fixing unnecessary dependencies in those images.

(From OE-Core rev: 1fa8909e24866ffada75daf63225c8daa000b9be)

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

kernel.bbclass: Ensure we have correct version information in deploy data

Currently the names used for the kernel in deploy will contain "AUTOINC"
instead of the final incremental numbering. This fixes the problem by
ensuring data is obtained from the PR service and using the PKG* variables
instead of PE/PV/PR directly.

[YOCTO #4293]

(From OE-Core rev: 1392f959cb8cd50b5a4492899e54f3ed68ef56d7)

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

oe.terminal: add tmux classes

This adds two new Terminal classes. It's separated into two, so that opening
a split inside a tmux window is preferred to the other terminal types, but
opening a tmux session is prioritized only slightly higher than screen.

- tmuxrunning: Open a new pane in the current running tmux window. Requires
  that the TMUX variable be added to the env whitelist to use it.
- tmux: Open a new tmux session

(From OE-Core rev: 10f64d202ceb230c3c79e09dce182ffce94d1117)

Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

rpm-postinsts: remove erroneous call to /etc/default/rcS

This doesn't work when the initscripts package is not installed (e.g.
when using systemd only) and is not even needed.

(From OE-Core rev: 6c972598c6da17fbec7a4582eb593c31f4283275)

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

udev-extraconf: Add -o silent to auto mount for mount.util-linux

This will silence some of the noisy output from mount.util-linux and the kernel
when trying to automount filesystems or devices. Busybox does not accept the silent
option, it uses a loud option instead.

[YOCTO #3935]

(From OE-Core rev: be218292ee3f05afe47545aa8e1625452e0cd614)

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

qemu: define fdt_t types in libfdt_env.h from qemu

* fixes
In file included from /home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/libfdt.h:55:0,
                 from /home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/work/x86_64-linux/qemu-native/1.4.0-r0/qemu-1.4.0/hw/arm/../../device_tree.c:28:
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:58:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:59:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:60:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:61:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:62:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:63:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:64:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:67:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:70:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:73:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:77:2: error: unknown type name 'fdt64_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:78:2: error: unknown type name 'fdt64_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:82:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:87:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:88:2: error: unknown type name 'fdt32_t'
/home/oe/setup-scripts/build/tmp-angstrom_next-uclibc/sysroots/x86_64-linux/usr/include/fdt.h:89:2: error: unknown type name 'fdt32_t'

(From OE-Core rev: dfb0c2cf9799d084a76aa92e243c743d7ff05db8)

Signed-off-by: Henning Heinold <heinold@inf.fu-berlin.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

initrdscripts: look for new systemd-udevd location

As per the previous commit, systemd-udevd is now in /sbin/systemd/.

(From OE-Core rev: ff0fd25206c3c75921d51cb80bcb6c94ca47b405)

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

systemd: move the binaries to /sbin/systemd

As with udev, placing binaries in /lib breaks our current multilib
implementation.  Change the rootlibexecdir to /sbin/systemd so that binaries
don't move in multilib situations.

(From OE-Core rev: d612ca261d12e89e96675c24d9d7456319179720)

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

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

The special key kernel-module- is necessary for the system to avoid a package
rename when installing a multilib image.  For example:

local.conf: IMAGE_INSTALL_append = " hello-mod"

bitbake lib32-core-image-minimal

The system will translate names and prepend 'lib32-', unless the package
begins with kernel-module-.

(From OE-Core rev: b847f87f4213db917d6760cd399c0afae996cf23)

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

package/image.bbclass: Fix multilib rprovides

allarch multilib recipes are meant to provide a list of different multilib variants.
Unfortunately since the pkgdata also has mappings for these, they get mapped back to
the original package name which means the effect is undone at package creation time
when the remapping code is called.

This patch adds in a conditional to break that chain meaning the packages get
the correct RPROVIDES and image builds work correctly with opkg.

[YOCTO #3453]

(From OE-Core rev: 1a1927f8a04fe0a2b3b853ebdd33ccb807f00b59)

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

classes/sanity: fix handling of bblayers.conf updating

Fix the fairly long-standing problem of treating a newer bblayers.conf
in the same manner as an older one (reporting that it had been updated
even if nothing was done). The recent work to do a reparse without
having to manually re-run bitbake turned this from an annoyance into an
endless loop, so it had to be fixed.

As part of fixing this the following changes have been made:
* Extensions are now implemented using a function list, so distro layers
  can add their own functions which should either succeed (indicating
  they have successfully updated the file) or raise an exception
  (indicating nothing could be done). The functions are called in
  succession until one succeeds, at which point we reparse.
* If we can't do the update, the error message now says "older/newer"
  instead of just "older" since we only know the version is different.

(From OE-Core rev: 46b00fdfc9d1e3dc180de087bae2682a1baa2954)

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

multilib.conf: Workaround opkg multilib issues

Multilib with opkg hasn't worked since the --force-overwrite option was dropped in
http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=88a9f6db3482623c9cb54fb03db1045051cec9f6

Since we merged the libexecdir changes, we also need the --force-maintainer flag
to avoid conf file conflicts.

Both these changes are suboptimal however the alternative is completely broken and
these changes only affect people who have multilibs enabled and use opkg.

[YOCTO #3453]

(From OE-Core rev: 6310e66cd62aab6109027a8dce9c56bf721bea92)

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

udev: Update initscript to check for devtmpfs

This is needed because the udev_182 now requires devtmpfs and will not work correctly
with out, so ensure that the kernel contains devtmpfs by checking /proc/filesystems.

[YOCTO #4125]

(From OE-Core rev: 4f85bb5254b3f4a9db8b419947d4bde424ce9617)

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

gdk-pixbuf: Fix libpng determinism issues

We now have libpng 1.6. If we build libpng12 as well as libpng 1.6, the 1.2
version gets preferred which is not desirable and does not give deterministic builds.

We really do want to use libpng since the item in DEPENDS will provide this so
manipulate the search list so the one we DEPEND on gets chosen. This was the cause of a
recent autobuilder failure.

(From OE-Core rev: ce1d262ea36da9a9fdeeefc0ddc69833801d4d2d)

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

oprofile: Add Upstream-Status and description to patch

(From OE-Core rev: 6af6df6c6507cae61dbd9d6994ff4162b54df451)

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

qemu: Fix typo in patch header

(From OE-Core rev: 6749cd024eaeda238cfe22ec2ab5f57da866607c)

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

kern-tools: fix non-local patch/config location

A regression was introduced when implementing the ability to restrict
configuration values via include directives. Only patch and config files that
were local to a feature directory could be found. While this doesn't impact
most users of the tools, it is an issue that needs to be fixed.

Additionally, the regex that detected flags passed to includes was not
specific enough, and unfortunately named feature files would match. This
resulted in features like standard-nocfg.scc inhibiting all configuration
items, even base configs.

This change also bumps the linux-yocto 3.4 and 3.8 PR values to ensure
that kernels will be rebuilt once this change is active.

(From OE-Core rev: ddce9f375c626ef2c86f48612b3d7a24e3111b0b)

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

mesa-demos: fix build with non-Mesa GL stacks

These patches from upstream allow mesa-demos to build and run against a non-Mesa
GL stack.  Thanks to Tom Zanussi for doing this work for EMGD in meta-intel, and
Otavio Salvador for confirming it also works for Freescale.

[ YOCTO #3469 ]

(From OE-Core rev: 60fabb6ea0474b19ad57873b402a608a92c5a5d4)

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

wayland: only build the scanner in wayland-native

We only build wayland-native for the scanner, so disable the bits we don't
actually need.  This gives us a small speed up but importantly should allow
wayland-native to compile on older hosts such as CentOS 5 which currently fails.

[ YOCTO #4245 ]

(From OE-Core rev: bfbe9b6a4fd7a8b5e5827847c2adff894e609e94)

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

sudo: update crypt.patch to use backport from upstream

Upstream closed my bug and rewrote the patch, so update our patch with a
backport from upstream.

(From OE-Core rev: 31327bac1e5438a0041638332698a1e1e91640ba)

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

xmodmap: fix compile with gcc 4.8

With gcc 4.8 there are compile errors:

xmodmap.c:289:5: error: implicit declaration of function 'asprintf' [-Werror=implicit-function-declaration]
(and more)

These have been fixed upstream so take the patch from git until 1.0.8 is
released.

(From OE-Core rev: 3a4ce4bd2b1ab7834edabbaf63acb18113cf1907)

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

kernel.bbclass: Optionally create lib dir during deploy

when we have multilib and a 64bit machine and initramfs then image/lib
directory will not be created and it will end up with errors like

| DEBUG: Executing python function sstate_task_prefunc
| DEBUG: Python function sstate_task_prefunc finished
| DEBUG: Executing shell function do_deploy
| tar: lib: Cannot stat: No such file or directory
| tar: Exiting with failure status due to previous errors

So in order to overcome this shortcoming lets mkdir -p the
lib directory so the modules can be happily installed.

(From OE-Core rev: 3f501204f375cc40a14597a25cb09faaba9ff802)

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

util-linux: use $PN in SYSTEMD_PACKAGES

Instead of using util-linux-uuidd in SYSTEMD_PACKAGES use ${PN}-uuidd, as in
multilib configurations util-linux-uuidd doesn't exist.

(From OE-Core rev: 1b840f8f0a52423a2a395b4ff35a6b24b05e6c0b)

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

systemd: use ${BPN} instead of ${PN} in FILES

Otherwise in multlib builds the wrong name is used and files don't get packaged
correctly.

(From OE-Core rev: 6b1e5db596a8ba55a8f7b54aa9ff41771f39b230)

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

wayland: add necessary dependencies to fix build error

add libffi into DEPENDS to fix following build error:
| checking for FFI... no
| configure: error: Package requirements (libffi) were not met:
|
| No package 'libffi' found

(From OE-Core rev: 23d6746efe1b3f31ad156db58fbc2767f750b712)

Signed-off-by: Zhenhua Luo <zhenhua.luo@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

opkg.inc: don't hardcode /var/lib/opkg

(From OE-Core rev: 9caf1f2540f0419c1301cc21777f97c6671fd844)

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

dpkg, opkg, rpm-postinsts: avoid repackaging when changing IMAGE_FEATURES

Recipes cannot depend on the value of IMAGE_FEATURES; in this case the
result is do_package task signatures changing every time IMAGE_FEATURES
changes, causing a large number of task re-executions. The
implementation of the log capturing really needs to be changed to
capture these in a different place and possibly not even conditional
upon IMAGE_FEATURES at all, but this will be invasive at this point in
the development cycle. For now, remove the variable dependencies to fix
the immediate problem.

Fixes [YOCTO #4246].

(From OE-Core rev: b4fbe4095de447ef4e426128bafaf8a292fa63e1)

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

sudo: handle glibc 2.17 crypt semantics

Staring from glibc 2.17 the crypt() function will error out and return NULL if
the seed or "correct" is invalid. The failure case for this is the sudo user
having a locked account in /etc/shadow, so their password is "!", which is an
invalid hash.  crypt() never returned NULL previously so this is crashing in
strcmp().

[ YOCTO #4241 ]

(From OE-Core rev: 06d7078f7631b92e8b789f8e94a3a346d8181ce6)

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

icecc: Allow to use this bbclass together with external toolchains

* original implementation by Antti Harju

(From OE-Core rev: abb5bd9f2d3583808b61a0832378e4db45022be6)

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

package_rpm.bbclass: fix build multilib image failed when PR Server enabled

1, In bitbake.conf
   PKGR ?= "${PR}${EXTENDPRAUTO}"
   EXTENDPKGV ?= "${EXTENDPKGEVER}${PKGV}-${PKGR}"
   RDEPENDS_${PN}-dev = "${PN} (= ${EXTENDPKGV})"

2, When PR Server is enabled, EXTENDPRAUTO is not none which means PKGR and PR
don't have the same value.

3, When multilib is enabled, RDEPENDS_${PN}-dev is not expanded correctly
which uses PR rather than PKGR in the versioned dependency string.

4, Make sure PKGR rather than PR in version string when do_package_rpm.

[YOCTO #4050]

(From OE-Core rev: cf53c606fc1bc81abb68b6851ae68916f92e1d84)

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

systemd: Fix path to systemd-analyze so it end up in the right package.

Since the upgrade to version 199 the location for systemd-analyze has change
this caused the systemd-analyze package to be empty and the binary was
shipped with the systemd package instead.

(From OE-Core rev: 997f39575dbf85600a67bfb815d715443c3fe279)

Signed-off-by: Erik Botö <erik.boto@pelagicore.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

pulseaudio: remove spurious cd in do_compile_prepend

This prepend was cding to ${S}, which then breaks base_do_compile as it assumes
it's in ${B}.  The cd is pointless as all of the operations use absolute paths,
so remove it.

The result of this was that base_do_compile was failing to find the makefiles,
so the compilation happened in do_install.

(From OE-Core rev: ac3a8ce0b672d1488c9074bde1a1d062e0c5fd33)

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

dpkg, opkg, rpm-postinst: fix overwriting the run-postinstall script

If multiple package managers are installed in the image, they will
overwrite each other's run-postinsts script, resulting in postinstalls
not beeing run at all at first boot.

What this patch does:
 * checks whether opkg/dpks/rpm is actually used to install
   the packages and, only after, creates the run-postinsts script;
 * brings dpkg recipe in sync with opkg: moves the script creation from
   do_install to postinstall;
 * move creation of run-postinsts script (rpm-postinsts recipe) to the
   postinstall scriptlet in order to better control the creation of the
   script according to the package manager used;

[YOCTO #4231]
[YOCTO #4179]

(From OE-Core rev: d7fd56df0a4954954d6d0764ae06beb869e6b99a)

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

ptest bug fixes

Move ${PN}-ptest to start of PACKAGES to ensure all ptest files are
packaged in the -ptest package.

Add QA exclusions to insane.bbclass to ensure -ptest packages can contain
any files they need.

Disable ptest for native packages.

Don't emit errors on missing _ptest functions.

(From OE-Core rev: 01bea4ef932e46eb2fcc8b4be7ff5e2b5b2a0978)

Signed-off-by: Björn Stenberg <bjst@enea.com>
Signed-off-by: Anders Roxell <anders.roxell@enea.com>
Signed-off-by: Josep Puigdemont <josep.puigdemont@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kernel.bbclass: do_strip: allow recipes to strip the kernel

Allow recipes to specify sections to be stripped from the kernel output
using KERNEL_IMAGE_STRIP_EXTRA_SECTIONS.  For example:

KERNEL_IMAGE_STRIP_EXTRA_SECTIONS = ".comment .unwanted"

The kernel output is stripped in place.

Since the toolchain does not give indication when the specified sections
are absent, we read the sections first and make this report by issuing a
warning to the developer.

The toolchain by default strips the image with the -s option (even
when -s is not specified):
-s --strip-all       Remove all symbol and relocation information

For example, these sections are always removed:
.debug_aranges
.debug_info
.debug_abbrev
.debug_line
.debug_frame
.debug_str
.debug_loc
.debug_ranges
.symtab
.strtab

In addition to these, the sections listed in
KERNEL_IMAGE_STRIP_EXTRA_SECTIONS will also be removed.

Only stripping of vmlinux (elf) is supported at this time.  A warning
will be given if the image type is not vmlinux.

Stripping the image could also be done in the kernel, but that would
only work for linux-yocto based kernels, so it's not the route we
decided to go.

[YOCTO 3515]

(From OE-Core rev: 5f6d33b05b4e7883f2728ca812cb5386d1e36989)

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

kern-tools: fix conditional configuration items

Variables defined in .scc files have two purposes:

   - Documentation in the meta-series
   - Variables that can be tested in sub sections and other features

The second part of this functionality was broken when fixing configuration
for tiny/small systems. As a result, arch tests were failing and configs were
dropped.  This restores the existing functionality.

(From OE-Core rev: 4170e458e0f700319f4e1023c0c6c2d803449566)

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