Write DPKG_ARCH to /etc/apt/apt.conf

Apt does not recognize the architecture if a different one is set with
DPKG_ARCH (e.g. armel). This patch writes the correct architecture to
/etc/apt/apt.conf.

(From OE-Core rev: 81b8c36641994dc7a4e025f2d43f9ce57d04b6f0)

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

relocate_sdk.py: allow relocate_sdk.py to work with python 2.4.x

Avoid the chicken / egg problem of an SDK that provides a working
python but requires that version of python to extract itself.  The
RHEL 5.x systems and some other enterprise Linux systems ship with
python 2.4.x as the default python.  We need to at least be able to
extract work executables even if we never use the the host provided
python again.

(From OE-Core rev: e1d42db8749b0b965ddc6cfba4f3b93ee96ed4f4)

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

populate_sdk_base.bbclass: Improve debugging capabilities for SDK installer

After having to debug the SDK installer a few times in
addition to the relocation code the following patch was created
to improve the capabilities around debugging the SDK installer.

1) Add a verbose mode -D which set a set -x to see what
   the SDK installer is doing.

2) Add a mode -S to save the relocation scripts for the purpose
   of debugging them in conjunction with -D

3) Add a mode -R to not execute the relocation scripts for the
   purpose of debugging the relocations.

(From OE-Core rev: 0e6dd19b9736d2a8ae7c0f0ab124337d579b8f06)

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

relocate_sdk.py: Fix corruption of sdk binaries

There are two cases of corruption that the relocate_sdk.py was not correctly
dealing with.

1) SDK Extras should be left alone
   Extra external binaries included in an SDK that were linked against the
   host's version of /usr/lib/ld-so.so should not get a relocation applied.
   In the case that was discovered these were LSB compliant binaries that
   already worked on many hosts.

2) If the interp section is too small generate an error
   In the case of the qemu user code, it was using its own .ld file
   to link the executables which overrides the default in the nativesdk
   binutils.  This generated host executables which had a interp section
   that was too small to relocate.

   Now the relocate_sdk.py will print an error and continue on such that
   the error can be fixed by a developer without having to do the
   difficult task of debugging why it is crashing or not loading correctly.

(From OE-Core rev: 3752a9c6d772b39bbe04d62ef4d3527b4c7198c1)

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oe-git-proxy*: Remove previous git proxy solutions

The new oe-git-proxy should address all git proxying needs, remove
the previous scripts.

V2: Separate the removal of the old scripts into their own patch

(From OE-Core rev: 75738ac47b9ca11daa94820c9c5f829937397da7)

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

oe-git-proxy: Use socat instead of BSD nc

BSD nc was commonly available on the current distros until Fedora 18
appears to have dropped it. socat appears to be a reasonable replacement
with availability on Fedora and Ubuntu and going back some time as well.

Update the script to use the socat syntax.
Simplify the logic a bit by using exec for the no-proxy-needed cases.

(From OE-Core rev: 795b1ea370b8a1d9152c171a50e80bd0b4b8dc60)

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

oe-git-proxy: Add a new comprehensive git proxy script

oe-git-proxy.sh is a simple tool to be used via GIT_PROXY_COMMAND. It
uses BSD netcat to make SOCKS5 or HTTPS proxy connections. It uses
ALL_PROXY to determine the proxy server, protocol, and port. It uses
NO_PROXY to skip using the proxy for a comma delimited list of hosts,
host globs (*.example.com), IPs, or CIDR masks (192.168.1.0/24). It is
known to work with both bash and dash shells.

V2: Implement recommendations by Enrico Scholz:
    o Use exec for the nc calls
    o Use "$@" instead of $* to avoid quoting issues inherent with $*
    o Use bash explicitly and simplify some of the string manipulations
    Also:
    o Drop the .sh in the name per Otavio Salvador
    o Remove a stray debug statement

V3: Implement recommendations by Otavio Salvador
    o GPL license blurb
    o Fix minor typo in comment block

(From OE-Core rev: 62867f56da0e0904f0108f113324c2432659fbac)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Cc: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Cc: Otavio Salvador <otavio@ossystems.com.br>
git-proxy cleanup

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

base.bbclass: Remove generate_git_config()

git no longer supports the use of GIT_CONFIG which defeats the purpose of
GIT_CORE_CONFIG and the generate_git_config() function. Remove it.

(From OE-Core rev: c485322fa2b89eb90efd88969d0c73575f128af7)

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

oe-buildenv-internal: Add upper and lower case proxy vars to BB_ENV_EXTRAWHITE

Applications are inconsistent in their use of upper and lower case proxy
variables. Curl, for example, specifies NO_PROXY (not no_proxy) in the
man page (changed in 2009 [1]). Avoid proxy issues by ensuring both the
upper and lower case versions of each proxy variable are available in
the environment for the fetcher commands.

Add FTPS_PROXY and ftps_proxy to the list as well.

1. http://curl.haxx.se/mail/tracker-2009-04/0012.html

(From OE-Core rev: 684c6512850ceb108e52af634be98eaacb8351e1)

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

oe-buildenv-internal: Remove GIT variables from BB_ENV_EXTRAWHITE

The following variables perform no function outside of bitbake:

    GIT_CONFIG
    GIT_PROXY_HOST
    GIT_PROXY_PORT
    GIT_PROXY_IGNORE

GIT_CONFIG only affects the git-config command which is not relevant to
the fetcher. This was previously used with the OE GIT_CORE_CONFIG
variable which would provide a basic git config to use instead of the
user's config. This usage was deprecated by git for over a year now:

http://git.661346.n2.nabble.com/Overriding-gitconfig-using-GIT-CONFIG-td6680977

GIT_PROXY_HOST and GIT_PROXY_PORT are not used by git.

GIT_PROXY_IGNORE was an OE construct used to create the custom git
config and had no meaning outside of the OE environment. It is not used
by git.

Remove these variables from the fetcher environment.

Users wishing to configure git to work with a proxy should define the
GIT_PROXY_COMMAND environment variable to use an external script.
NO_PROXY can be used within this script to skip the proxy for certain
hosts.

(From OE-Core rev: ea0284a8cc1b531e115b7fdbfa18852f55573f00)

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

qt4: use system SQLite

We've always been depending on external SQLite, however by default Qt
builds using its own internal copy of SQLite. Add the configure option
to actually use the external SQLite library and make it clearer which
version is in use.

Fixes [YOCTO #2514].

(From OE-Core rev: 33e0afd2fffdef4dccfce6383bc646f975972b1e)

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

gcc-cross-canadian: enable multilib support

(From OE-Core rev: ee7331e3e7d0b38f3e19ccf7e394537f5f653a22)

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

multilib.bbclass: save multilib variables before executing the gcc-cross-canadian statements

(From OE-Core rev: 45528026c190c7c3b121e9fb65747d050b2bb21a)

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

systemd.bbclass: use PACKAGESPLITFUNCS instead of populate_packages_prepend

This is cleaner and leads to more accurate profiles.

(From OE-Core rev: eed7294ba9aedf47af5c64ff11777015e59f48ef)

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

fontconfig: rename the patches directory

Rename the patches directory to files as we only have one version of fontconfig,
so the hassle of moving these files every upgrade can be avoided.

(From OE-Core rev: 67a279f517fc43e418482cc104458875264141f6)

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

fontconfig: drop explicit -native recipe

Instead, use BBCLASSEXTEND=native in fontconfig.

We can drop the installation of fc-lang/fc-glyphname, they are not used by
fontconfig's build anymore as it ships the generated files in the tarball.

(From OE-Core rev: d5ccc6cbbbdc7502d35ddcbbc2bfc2d4657cbe78)

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

fontconfig: remove unexplainable and broken build tweaking

The build flag twiddling in do_configure_append was unexplainable and clearly
wrong (native flags for a cross build), remove it all.

Parallel make appears to be working now, so enable it.

The tarball doesn't contain read only sources, remove the unpack hacking.

The pkgconfig doesn't need patching as the freetype link lines are the same.

Don't need to specify where freetype is, it's found automatically.

Merge fontconfig-util-dbg into fontconfig-dbg.

Don't export HASDOCBOOK, --disable-docs is sufficient to turn off documentation
building.

(From OE-Core rev: ba3db5f5e7055d597b21d8b5f8e60bc3c4f12bd2)

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

lib/oe/path.py: support missing directory components in realpath()

Some use cases in OE operate on symlinks which dangling path components.
Assume that these are directories instead of raising ENOENT.

(From OE-Core rev: a96e2c84f24c15b77ee1fbc1f998b8b4796b8664)

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

package.bbclass: support dangling path components

Commit ec2aab09769f4b6817d74d2175afa2b7c7598750 introduced a regression
on packages which contain symlinks with unresolvable path components
(e.g. lsof-dbg).  While assigning a variable, an exception was raised
and the exception handler accessed this variable.

Patch deals both with the dangling path components by assuming them as
valid directories and by avoiding the broken assignment.

(From OE-Core rev: 579369b0fb27fad6d628746a50b9b798078500f6)

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

libiconv: Remove RPATH from binaries

Modify libtool to not add RPATH.
This solves https://bugzilla.yoctoproject.org/show_bug.cgi?id=2669

(From OE-Core rev: 273c437813b19577d68fa49ec04ea42154ad70c9)

Signed-off-by: Björn Stenberg <bjst@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glib-2.0: Fix ptest to build with uclibc.

Only install eglibc-specific dependencies when building for eglibc.
Tweak a test case that won't build with uclibc.

(From OE-Core rev: e28e04e26893416d577ee54e03019c03865e1bf6)

Signed-off-by: Björn Stenberg <bjst@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib: implemented oe.path.realpath()

Various parts of the buildsystem have to work with symlinks. Resolving
them is not trivial because they are always relative to a sysroot
directory.

Patch adds a function which returns the destination of a symlink by
assuming a given path as the / toplevel directory.  A testsuite was
added too.

(From OE-Core rev: 76e0bd7f8e3a3bd052a6e329f88e2d8099e899c4)

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

package.bbclass: use oe.path.realpath()

oe.path.realpath() provides are common and more correct implementation
for resolving symlinks within sysroot. Use it.

Old implementation suffered from lot of problems; e.g.

* redundant code

* calls 'os.stat()' which references files on host; this can give wrong
  results about existing/non-existing and can cause EPERM (instead of
  the catched ENONENT) exceptions

* does not deal with special cases like '..' leaving the sysroot.

(From OE-Core rev: ec2aab09769f4b6817d74d2175afa2b7c7598750)

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

update-alternatives.bblcass: use oe.path.realpath()

oe.path.realpath() provides are common and more correct implementation
for resolving symlinks within sysroot. Use it.

(From OE-Core rev: 2fa5cc0d08e855e24a4497601d7cc73b9f2aa550)

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

ncurses.inc: Fix display corruption and add fallback for sstate compiled paths

CQID: WIND00402979

Display corruption was occurring on 64 bit hosts using menuconfig for
the kernel and busybox with the ncurses-native or ncurses-nativesdk
because the configure arguments were not sufficient vs the expected
use from the upstream source.

Also changed in this commit is to provide a fallback to the hard
compiled paths for the terminfo and termcap.  Eventually this needs to
be fixed another way if we want a truly portable SDK because any SDK
that is relocated or native binaries pulled out of the sstate can end
up referencing terminfo files that no longer are in the same location.
Because the host system has terminfo files that ncurses will happily
use we might as well have a fall back to buy some time for a better
fix, vs having thing just not work occasionally when using sstate.

(From OE-Core rev: c820ae11ce369002063bad8b11ee95e1882c99bb)

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wpa-supplicant: Enabling with systemd

(From OE-Core rev: bd1e384017954e883f56379560afc2f9947ebbd6)

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

connman: Enabling with systemd

(From OE-Core rev: 352b3e3663fc4ccab2d8240176f4b085db726e82)

Signed-off-by: Radu Moisan <radu.moisan@intel.com>
Signed-off-by: Sander van Grieken <sander@outrightsolutions.nl>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

update-rc.d: don't do anything if systemd.bbclass is inherited

We need the update-rc.d class to work when systemd is being used so that
packages that only have SysV init scripts still work.  However if a recipe
supports both we don't want to install SysV and systemd files under systemd.

To solve this, before doing real work in update-rc.d check if the systemd class
has been inherited and don't do anything if it has.

(From OE-Core rev: 0273a22fec3c9360df2510b759c5bf9af610551f)

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

avahi: Enabling with systemd

(From OE-Core rev: 6d89c6744a98dce1fee6a21f1dad1305f6bf6fb5)

Signed-off-by: Radu Moisan <radu.moisan@intel.com>
Signed-off-by: Sander van Grieken <sander@outrightsolutions.nl>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd.bbclass: helper class for recipes with systemd units

This class adds postinst/prerm scripts to start/stop/enable/disable the services
as relevant, and some magic to ensure the service files are installed.

Based on (but not the same as) the systemd.bbclass in meta-systemd, so thanks to
the following for their work there:

Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Khem Raj <raj.khem@gmail.com>
Martin Jansa <Martin.Jansa@gmail.com>
Andreas Müller <schnitzeltony@googlemail.com>
Koen Kooi <koen@dominion.thruhere.net>

(From OE-Core rev: f4bf51612f8be1d3dd340fc456f3fa08fcfa34ef)

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

nfs-utils: Create the default statdpath

packaging these extra directories is needed
Otherwise rpcbind wont start as it
expects /var/lib/nfs/statd to exist.

This fixes the issue where automounter fails
to mount since rpcbind did not start correctly

Failed to open directory sm: No such file or directory

(From OE-Core rev: edd7b7f9f1cea921bdc9480c2a120f6abc6b0c9c)

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

buildhistory: record more R* variables

Add RPROVIDES, RREPLACES, RCONFLICTS and RSUGGESTS to the list of
tracked variables. Of these, RPROVIDES is always output, whereas the
others are only output if they have a value (since it is more common
that they don't).

Implements [YOCTO #3391].

(From OE-Core rev: 564d76bed7b96d381d6438df81c0d5b4f5a7b2b0)

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

buildhistory_analysis: handle more R* variables

Report changes to RPROVIDES, RREPLACES, and RCONFLICTS. As RSUGGESTS
isn't widely used and isn't of huge concern if it changes, it is not
reported by default.

Implements [YOCTO #3391].

(From OE-Core rev: d20011571db96da79a8a0e056c6cef8e4c083608)

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

python-smartpm: show friendly error if YAML output requested without PyYAML

Instead of a python backtrace, tell the user they need to install PyYAML
if they wish to use the --yaml output options.

Fixes [YOCTO #3768].

(From OE-Core rev: 69caf24112c11609eb351bea09817029bca0ff2e)

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

runqemu: add option to make the VNC server publically available

If the qemu is running on a headless machine, a VNC server that only allows
connections from localhost isn't too useful.

Add a "vncpublic" option to bind a VNC server to 0.0.0.0, so it's publically
available.

(From OE-Core rev: 883666821ec46483bbfb9b3cb84c5afa8118a553)

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

make-3.82: Add patch from git to fix parallel make race

See: http://savannah.gnu.org/bugs/?30653

This only shows up for projects which have a number of intermediate
targets.  For me, systemd fails to build without this patch, and does
build consistently with it.

WebKitGtk+ is another known affected project.

(From OE-Core rev: 1747a70f95ead49985eeaf16c28e818ed5b109cd)

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

udev: Cleanup .inc file

* udev-acl and udev-concolekit have no files so remove these packages
* there's nothing in sbindir so remove useless sbindir = "${base_sbindir}"
* there is no udevinfo or udevtest so remove these from udev-utils
* udevadm is installed in bindir so modify it's path in FILES_udev-utils

(From OE-Core rev: 25e01e11586f422f3659318796cb847cb701b03e)

Signed-off-by: Andrei Gherzan <andrei@gherzan.ro>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev: PR bump

(From OE-Core rev: 098134533650ba16ec5eb03e7c604af43652e7bc)

Signed-off-by: Andrei Gherzan <andrei@gherzan.ro>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev: Modify init script to use the correct path of udevadm

udevadm is installed in /usr/bin not in /usr/sbin. Init script modified
accordingly.

(From OE-Core rev: b8ba1e3db44d2443e0071d4923101280151ccd03)

Signed-off-by: Andrei Gherzan <andrei@gherzan.ro>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: bitbake/utils: Improve environment handling to allow UIs access to original environment

We need to empty out the environment whilst we build the cooker but
we need the environment for the UIs since hob uses DISPLAY and other
session variables.

This patch adapts the utils functions to return removed environment
components so we can reinject them for use by the UI, allowing hob
to work again.

(Bitbake rev: fc330d810099c57fefd4e706159a73ad8401d97c)

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

dbus-native: Don't install dbus-launch

dbus-launch from dbus-native has no X support so lets not install it in
case the host has a more featured and useful version. It can interfere
with running X utils with STAGING_BINDIR_NATIVE is in PATH and we don'
use it anywhere.

(From OE-Core rev: adfa83bfa1ccb52b1a5d086aff36fe27271d0e59)

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

dbus: Drop old 1.5.12 version

We have a newer default version...

(From OE-Core rev: 1ffa1b1b23cbd7fbcfe95abc6c045eef93f96f2a)

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

udev: Add start/stop/status/restart support to init script

[ CQID: WIND00388860 ]

Add the ability to use "stop", "start", "restart" and "status"
with the udev initscript.

(From OE-Core rev: bc6193824ec3eb1f3c590a93ccb79c8588b74412)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
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>

initscripts: Add status function to /etc/init.d/functions

[ CQID: WIND00388860 ]

Many initscripts want a simple way to display status information.
Add the 'status' function to the functions file.

(From OE-Core rev: adcb39845b6d3af9472fa5051a1d918344eb6bda)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
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>

libcgroup: fix failure for absolute path

[ CQID: WIND00401091 ]

While using sstate or some other conditions, there may be no image
directory in the builddir, so the absolute path of libcgroup.so.1
will cause build failures.

(From OE-Core rev: 0b58d98567a1c0531ad8396c4464402da21a9665)

Signed-off-by: Xin Ouyang <Xin.Ouyang@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
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>

rt-tests: fix cyclictest reported 1 sec latencies

[ CQID: WIND00391769 ]

Huge latencies reported close to 1 second when certain options
are used in cyclictest.

Extend the original commit to remove the 1 second hardcoded
timer values from the RELTIME and ITIMER options.  Use the
the actual interval instead.

(From OE-Core rev: d1a3f3e2a6a74ba9b7ad5554648faebaf1b73c9f)

Signed-off-by: Jim Somerville <Jim.Somerville@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
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>

coreutils: remove dependency on coreutils-native

[ CQID: WIND00397793 ]

It's not reasonable to build such a low-level package. Remove the
dependency so that we always use the host version.

The 8.14 recipe doesn't have such dependency so needs no change.

(From OE-Core rev: 194c902c28291d564cef71bdd9588afa5b72d9f5)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
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>

eglibc-nativesdk: Fix buffer overrun with a relocated SDK

When ld-linux-*.so.2 is relocated to a path that is longer than the
original fixed location, the dynamic loader will crash in open_path
because it implicitly assumes that max_dirnamelen is a fixed size that
never changes.

The allocated buffer will not be large enough to contain the directory
path string which is larger than the fixed location provided at build
time.

(From OE-Core rev: 8ebd85d29eb1a9c0c0d3cd79e7dda8b857c27bbb)

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
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>

ghostscript: fix endianness detection

[ CQID: WIND00394747 ]

The original patch removes endianness detection completely, causing lcms
to be built against the wrong endian. Instead, pass the correct endian
through the recipe using SITEINFO_ENDIANNESS.

(From OE-Core rev: 0a07ec7d13becc7ce87c502e75a006300a7090a5)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
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>

openssl: Add mips64 configure support.

Add mips64 configure support but assume mips(32) userspace.

(From OE-Core rev: 7d775b071b902ee0de6391b2c30d36e3003643e1)

Signed-off-by: Randy MacLeod <Randy.MacLeod@windriver.com>
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>

libart-lgpl: add art_config for mips64/mips64el

[ CQID: WIND00389505 ]

Generated by gen_art_config.sh

(From OE-Core rev: e900b4a37a12f15a2e734d86984fc613540890f7)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
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>

ghostscript: add pregenerated objarch for mipsel/mips64/mips64el

[ CQID: WIND00389504 ]

Generated by building and running base/genarch.c.

(From OE-Core rev: 78a13ba170c1de6d7ef077854e3e34e18d17099f)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
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>

populate_sdk_base.bbclass: add execution permission for self-extracting archive

[ CQID: WIND00392947 ]

It is not good user experience that the self-extracting archive (.sh file)
has no execution permission by default.

(From OE-Core rev: 3d6537f6ab1ce98075461b9a5d49885c2454417f)

Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
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>

scripts/create-recipe: Bugfixes for create-recipe easy_install handling.

1. Tell easy_install to always download, regardless of python
   install content.
2. Support https/ftp et.c. URLs provided by easy_install

(From OE-Core rev: 18e3654894175af0f51049cf2dcf42295bfbc905)

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

qemu: disable smartcard support

* sometimes it's autodetected and fails to build:
  | /usr/bin/ld: libcacard/.libs/cac.o: Relocations in generic ELF (EM:40)
  | libcacard/.libs/cac.o: could not read symbols: File in wrong format
  | collect2: ld returned 1 exit status
  | make[1]: *** [libcacard.la] Error 1

(From OE-Core rev: acb0e56d79cfaa606ccd0a075a7c78ede172da86)

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>

Revert: update-rc.d: disable update-rc.d.bbclass when systemd enabled

This was just wrong - when systemd is being used there'll still be packages that
use SysV-style init scripts, which systemd supports fine.

This reverts commit b94227f7290796f6ebbe5c5ff1680b9b689022b1.

(From OE-Core rev: 3f50b61c77406f87d36437cca53573f86f314641)

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>

systemd_197.bb: Point to the place where we keep the /etc/rcN.d files

By default /etc/rc.d is searched by systemd but we will keep rcN.d directories
inside /etc.

(From OE-Core rev: 659b146ef51c4873c67f227bd39f2368c28a022b)

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

pseudo.inc: Fix sqlite libdir again, pseudo 1.4.3

This updates to pseudo 1.4.3. Changes:

1. A couple of minor tweaks to reduce difficulties using SDKs built
   on slightly more recent machines on older machines; specifically,
   avoiding getting @GLIBC_2.7 symbol references for sscanf(), fscanf(),
   and open2().
2. Revision of the logic determining the library directory to use for
   sqlite's library files.

The latter is a source of difficulty because it's come up a few times
that we may want pseudo to use lib64 for libpseudo.so, but bitbake's
usual setup would have libsqlite3.a in lib regardless of bit width.
Cleaned up previous design a bit by providing a distinct setting for
sqlite-lib, which defaults to the same library directory used for other
things. Adjusted build to use this new setting. (This ends up being
${baselib}; on targets, that might not be lib, but for native builds
it generally is, and for SDK builds it appears to do the right thing.)

Testing: Successful build of meta-toolchain for both 64-bit and 32-bit
SDKMACHINE, and builds with NO32LIBS = "0" also succeeded. Also builds
for multilib targets.

(From OE-Core rev: ae8811bb26fba2e71d7280f6d6c4f5cec6a2871b)

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

create-pull-request: Error message on missing -u

The script was erroring out without a hint on what failed.

(From OE-Core rev: 72266cfa3a12a19a94d9176ecca9d080658dbf2e)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

base.bbclass: remove redundant setting of MAKE

oe_runmake() tried to set MAKE a second time, remove it since this is
(nowadays?) redundant WRT the normal export MAKE= in the generated
scripts.

(From OE-Core rev: b9c83b22b555349314191ba60346b01a9252a812)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

uclibc: turn on kernel module support

(From OE-Core rev: 14c08a97949b3a05221a556902fd7fb868d1f843)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

uclibc: add c6x arch

(From OE-Core rev: 4f6f4a621a16d79830ba3b417f66c9aa44c89d55)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

base.bbclass: Commentary typo fix

(From OE-Core rev: 3f0c1f6f001bbe958a0589da7db52f96a2d9d369)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

yocto-bsp: prepend includes in machine.scc files with machine

The names of the -user files were changed to have the machine
prepended, but the includes weren't - fix the includes.

(From meta-yocto rev: c430d6a0d126df7a51c0f585665de6aebbeac028)

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

qemu.bbclass: fix segfaults when running through pseudo

qemu user binaries sometimes segfault when running them through pseudo.
So, set PSEUDO_UNLOAD to 1 before running any qemu binary.

[YOCTO #3788]

(From OE-Core rev: 688e9485980de0f29aa00e24ce53a3efd3a3a7cc)

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

qemu.bbclass: Use the correct qemu binary in multilib cases

For example with a lib32 multilib, we need to still use the 64 bit
qemu binary in case we do encounter a 64 bit binary.

(From OE-Core rev: e8ec13a26217bf473504ae4aab22b134dd9dffff)

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

image.bbclass: Add MLPREFIX to DEPENDS

Add MLPREFIX to DEPENDS to ensure the correct qemuwrapper is dependended upon. Its searched
for in PATH so we need to do this to ensure the correct version is present.

(From OE-Core rev: 41163fc5e6662251ec264fd5194a649342d11de1)

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

qemuwrapper: Don't use STAGING_BINDIR_CROSS directly

We need the files to be tracked by sstate so we need to jump through
some hoops to ensure this happens. The cross bindir directory
isn't staged automatically so we need to handle this outselves.

(From OE-Core rev: 743d5233747d0a107490b31ea3da151ea1ace3ae)

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

sanity.bbclass: Check for DISPLAY for qemutests in BB_ORIGENV now

(From OE-Core rev: e43f8f917e4069d74b467dfd6aa29ff762e83a5e)

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

directfb: Really disable mesa

Mesa was disabled in the .inc but not in the main recipe where a separate
EXTRA_OECONF is used. Fix disabling mesa there too to avoid build failures.

(From OE-Core rev: 722a5eaa6bed6ab8c8a1caa066cd2eba7acda5ee)

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

bitbake: prserv/serv.py: Fix logging in daemon mode

In deamon mode we need to ensure the logging module is sending log data to the
log file. These changes ensure this happens correctly.

(Bitbake rev: bb53b47710ca4579e20284668cb354f734c3d502)

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

bitbake: bitbake: Always use separate process for PR Service

Using the threading module interacts badly with multiprocessing used elsewhere
in bitbake under certain machine loads. This was leading to bitbake hanging on
Ctrl+C when the PR Server was being used.

This patch converts it to always use the daemonize code which
then means the threading code isn't required.

[YOCTO #3742]

(Bitbake rev: 2d0bbd9398ab839bd2d1e29e50b25d52efb1ce2a)

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

imagetest-qemu: Adapt to bitbake environment changes

Load environment variables like DISPLAY from BB_ORIGENV after recent
bitbake changes.

(From OE-Core rev: 8ad8b34b5ac36d805bb10c120f3388e7dce83b98)

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

terminal.bbclass: Use BB_ORIGENV to restore user environmental variables

(From OE-Core rev: 131643695d7ba9d0c32bf8e192e7e3942f50f318)

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

bitbake.conf: Update hashconfig after BB_ORIGENV changes

We should ignore BB_ORIGENV from the hash for the config data. There
are also a number of variables which no longer make it into the data
store so we can drop these from the hash whitelist.

(From OE-Core rev: cb21af00f9321ea48e533089dbffbb1a9665cb92)

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

package.bbclass: Replace undefined 'src' with valid variable

After rewrite of split_and_strip_files src var is no longer
available here where it is used so we should replace it with
a valid value

(From OE-Core rev: b07e79ea049d341a5a0a7e4bb72151f49336c5bc)

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

package.bbclass: Ensure package_get_auto_pr is run at the correct point

This needs to run before PKGR and other variables using PRAUTO are used,
including the expansion optimisations.

(From OE-Core rev: eb09ebc9680eb50e23778dcc1c2697d13cf073ae)

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

bitbake-prserv-tool: show error when export file does not exist

* otherwise it shows error about failing import

(From OE-Core rev: 65b2f068719f4cd6e1bd438e8714c2977bd93535)

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

initiscripts: Fix populate-volatiles.sh whitespace

(From OE-Core rev: d69935fbef2b5a8b187e58ac5ade0ffb9f95f803)

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

insane.bbclass: Add missing R* variables to pkgvarcheck

* check also RSUGGESTS, RCONFLICTS, RPROVIDES, RREPLACES

(From OE-Core rev: 63d9926dde562fd2447fc26482745c3b6f232c12)

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

libxscrnsaver: improve RREPLACES to add libxss only for PN

(From OE-Core rev: 0bb5572536a39fabd29c307004e76202afd5b6b4)

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

connman: be explicit when enabling ofono

(From OE-Core rev: 9c5fa8da28f86e88327356e784f2bb8281da2cca)

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

package.bbclass: return list of packages created in do_split_packages

* sometimes it's useful to do something with packages created
  by do_split_packages later in do_package_prepend, e.g. in:
  http://lists.linuxtogo.org/pipermail/openembedded-devel/2013-February/043824.html
  I have .bbclass which adds some postinst, postrm and RDEPENDS,
  but ttf-mplus is using do_split_packages to create those packages

(From OE-Core rev: 5aa52c6882d2929b2e530c4fa297c6d3e97d4f9f)

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

sanity.conf: Update minimum bitbake version

We need this for the multiprocessing pool issues in python < 2.7.3 whic
we now use in do_package.

(From OE-Core rev: 027dc0ec80b570c79f4adb0deabb63bcf3c49b78)

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

bitbake: Update version to 1.17.0

(Bitbake rev: d96ef9398356b2a61f7cff765821616bf1e55492)

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

bitbake: fetch2: Remove broken git variables from the environment

The following variables perform no function outside of bitbake:

    GIT_CONFIG
    GIT_PROXY_HOST
    GIT_PROXY_PORT
    GIT_PROXY_IGNORE

GIT_CONFIG only affects the git-config command which is not relevant to
the fetcher. This was previously used with the OE GIT_CORE_CONFIG
variable which would provide a basic git config to use instead of the
user's config. This usage was deprecated by git for over a year now:

http://git.661346.n2.nabble.com/Overriding-gitconfig-using-GIT-CONFIG-td6680977.html

GIT_PROXY_HOST and GIT_PROXY_PORT are not used by git.

GIT_PROXY_IGNORE was an OE construct used to create the custom git
config and had no meaning outside of the OE environment. It is not used
by git.

Remove these variables from the fetcher environment.

Users wishing to configure git to work with a proxy should define the
GIT_PROXY_COMMAND environment variable to use an external script.
NO_PROXY can be used within this script to skip the proxy for certain
hosts.

(Bitbake rev: e60270bdce6b8c2f8da1a4838aa374da9db3c86a)

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

bitbake: fetch2: Export upper and lower case environment variables

Applications are inconsistent in their use of upper and lower case proxy
variables. Curl, for example, specifies NO_PROXY (not no_proxy) in the
man page (changed in 2009 [1]). Avoid proxy issues by ensuring both the
upper and lower case versions of each proxy variable is available in the
environment for the fetcher commands.

Add FTPS_PROXY and ftps_proxy to the list as well.

1. http://curl.haxx.se/mail/tracker-2009-04/0012.html

(Bitbake rev: c3e6b2c5ec81d5ad7dcf606fff16fd5552bd267c)

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

bitbake: cooker/utils: Drop custom 'interactive' variables list and store environment in BB_ORIGENV instead

Maintining hardcoded lists of user environmental variables is hacky,
replace these with the savedenv datastore. Allow access to that
through the BB_ORIGENV variable.

(Bitbake rev: 0a99563a4ea270594fd9a61da46f9387fb79dc66)

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

bitbake: compat/utils: Add copy of python multiprocessing pool for pre 2.7.3 issues

python 2.7 shows hangs with issues in its pool implmenetation. Rather than
try and hack around these, add a copy of the working pool implementation
to the compat module from 2.7.3.

(Bitbake rev: c9eb742637131e8dbd526d2ad9b458abea0a2d87)

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

scripts/create-recipe: Python improvements for create-recipe.

1. Added ability to parse .zip files.
2. Added optional automatic dependency resolving for python
   recipes(easy_install wrapper).
3. Fixed a few name/version bugs.

Give it a whirl by:
create-recipe -r https://launchpad.net/nova/folsom/2012.2.3/+download/nova-2012.2.3.tar.gz

Saves me some time unwinding python dependencies, and creating template recipes.

(From OE-Core rev: 1a491a4dde0d3618f8815182d12c21f76b64de5a)

Signed-off-by: David Nyström <david.nystrom@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package.bbclass: Pre-expand some variables to save time

(From OE-Core rev: fc5bff0145d8f5db1c09be61f5de209ac4016ed1)

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

package.bbclass: Better document the different phases of operation

Add headers to document the different phases of do_package and
make the steps clearer.

(From OE-Core rev: b6438c94035a014902ec89af63ff3787cd8c67f6)

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

package.bbclass: Add PACKAGESPLITFUNCS variable

Prepending to populate_packages is rather ugly and means its hard to trace
errors and also profiling informaiton is summed together in one function.

This patch starts to split out the prepends to become separate functions
to avoid these issues. This is generally a neater way to write functions
than prepending to where there can sometimes be variable scope issues
and we've been bitten by whitespace issues in the past.

(From OE-Core rev: 4f9963d1d82ee896fe9491d6a8b32be42cd06f14)

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

package.bbclass: Simplify empty directory removal

Rather than an exec() per directory, we might as well exec one command and
be done with it.

(From OE-Core rev: 82ae9cfb09ee5c0aa6402c972d71e2b64d1ce8bc)

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

package.bbclass: Various minor performance tweaks

* Call getVar outside the loop
* Drop unneeded PATH export (bitbake does this already)
* Drop unused variable
* Simplify if statement nesting
* Simplify variable expandion to a getVar call (expand would just call getVar)

(From OE-Core rev: 52b506145bcddc133ca93a8c9f7343de69d10907)

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

package.bbclass: Make use of cleandirs and dirs function flags

We can use the cleandirs and dirs flags for the fuctions to handle
directory cleaning and creation at the bitbake level rather than
using these calls within the functions

(From OE-Core rev: 4b31d6f6f0a2a6b9e504ffae0d3b2099cbd7dddc)

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

kernel.bbclass: Improve populate_packages_prepend

Small performance tweaks for populate_packages_prepend:

* Compile the regexps once at the start
* Don't keep importing a module which is already imported
* No need to check PKG is set, we'd have failed long before now if it wasn't
* Don't export PATH, bitbake takes care of this at the task level

(From OE-Core rev: e9d43d7b4d2cfb22b21f3814c2401a699c78b025)

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

package.bbclass: Rewrite split_and_strip_files

The split_and_strip_files funciton was hard to follow and its usage of prefixes
to strings was unusual. This rewrites it to use a list of hardlinks, symlinks and
elffiles where each list is iterated over at the correct point.

This means we can avoid creating dandling symlinks for example so we can simply
delete the cleanup code for this.

The isfile() check is also removed which gives a significant improvement in speed.
Its uneeded since os.walk will have already checked things in files are files.

(From OE-Core rev: 0cd295d8cdc8cc39d6b6c7d26ea8a2a10a979d7c)

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

package.bbclass: Fix up bb.mkdirhier/bb.copyfile usage

These are in bb.utils so lets the correct function and avoid the overhead
of the fixup/warning for the deprecated usage.

(From OE-Core rev: d17329db4842c50af1a3d7f5f20e692c89913fba)

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

update-rc.d: Drop OVERRIDES code

The data store copy and overrides is overkill given the small number
of accesses that are being made. This simplifies the code.

(From OE-Core rev: 72c1fd72d3b479c728e249eaa763116d352e945b)

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