tune-cortexa7: add tunes for ARM Cortex-A7

http://www.arm.com/products/processors/cortex-a/cortex-a7.php

(From OE-Core rev: d6d1752c1f75fa97049d9e46b82e71d1ebbeb05f)

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

tune-cortexa15: add tunes for ARM Cortex-A15

http://www.arm.com/products/processors/cortex-a/cortex-a15.php

(From OE-Core rev: 513180af7f67b614dd0dbd8d9d77c057c45190b8)

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

shadow: add build dependency on shadow-native

The postinst requires patched pwconv and grpconv from shadow-native,
without these the rootfs creation uses the host binaries and fails.

(From OE-Core rev: 4baf25420bafb1f3c063efb053d653bd37e01a62)

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

systemd: set the location of the kill binary

Otherwise systemd uses AC_PROG_PATH and finds it in the sysroot, which won't
work on the target.

(From OE-Core rev: 54760c783089c98cc6eda1933393ac3cb49bcaed)

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: remove --with-distro option, it's not used

(From OE-Core rev: 18f4d510798d571127f24139b3121e444b673b38)

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: remove unused cached configure variables

(From OE-Core rev: 4a6be4d202526403aaf9130e615773bf7cb7852c)

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: make gcrypt support (for signing the journal) optional

(From OE-Core rev: 9ca041ccc6af1378a1072bdde678c0e97faedde6)

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>

populate-volatile.sh: remove trailing whitespaces

Remove trailing whitespaces.

(From OE-Core rev: 082dca6f73858581863f103b467ff73cebe2c0da)

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

populate-volatile.sh: improve the handling of link config items

Previously, if there's a link config item in the config file like

    l root root 1777 /tmp /var/tmp

and /tmp has existed, the symlink will not be created correctly.

Another example is the /run directory. If /run directory has been
created by some recipe or script before populate-volatile.sh runs,
the symlink of /run to /var/run will not be created correctly.

This patch ensures that the system creates symlinks exactly as the
config file tells it.

[YOCTO #3404]
[YOCTO #3406]

(From OE-Core rev: a54287d8e3d75a727c8ed5654a822bda256b0849)

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

populate-volatile.sh: add ROOT_DIR variable to support running at rootfs time

For populate-volatile.sh script to run correctly both at rootfs time and
at system boot time, it needs to be aware of which situation it is now in.

We use the ROOT_DIR variable to indicate whether it is run at rootfs time or
not. ROOT_DIR being "/" indicates that this script is run at system boot time,
otherwise, it is run at rootfs time.

Also, we ignore failures when running this script at rootfs time.
For example, if ${ROOT_DIR}/var/dir1 is symlink to /var/volatile/dir1, it's
possible that the link is a dead link. So if we're going to create some file
under ${ROOT_DIR}/var/dir1, it will fail. But the failure does no harm,
because this script will always run at system boot time to set up the correct
files and directories.

[YOCTO #3406]

(From OE-Core rev: 45396e3edcce4a33fcbef6456f31811f30c26c63)

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

bitbake: Revert "cooker: parse using bb.compat.Pool"

Reverting the pool changes, terminate does not work reliably on
bb.compat.Pool :(

[YOCTO #3978]

This reverts commit 8af519a49a3374bd9004864ef31ca8aa328e9f34.

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

bitbake: Revert "cooker: now that we use a Pool, raise the exceptions"

Reverting the pool changes, terminate does not work reliably on
bb.compat.Pool :(

This reverts commit 9ca8e513dc142dd7aefadeb6db2ccdb00f38b39f.

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

libpcap: fix description variable

* The variable name was truncated.

(From OE-Core rev: 2217176a9e2b9b9ac3b1e42e0feecb2c84d60f49)

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

qt4.inc: package keyboard drivers

* Fixes the following packaging error, if compiled with
  -plugin-kbd-linuxinput:

| WARNING: For recipe qt4-embedded, the following files/directories were installed but not shipped in any package:
| WARNING:   /usr/lib/qtopia/plugins/kbddrivers/.debug
| WARNING:   /usr/lib/qtopia/plugins/kbddrivers/.debug/libqlinuxinputkbddriver.so

* No PR bump, because -plugin-kbd-linuxinput is not enabled
  by default.

(From OE-Core rev: f477717a6e9b5820b6e6e505d64adb5d3a64abc7)

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

avahi: move systemd support where it belongs

We don't want the systemd inherit in avahi-ui since that causes various
warnings since PACKAGES is rewritten.

(From OE-Core rev: 02492d43cd7095acb3cf29843a50444c28b4d994)

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

gmp: fix missing PR after update to 5.1.0a

(From OE-Core rev: 8f2ae03dd950e119d9c06cecfa57d69edcf293be)

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

gtk-doc: Fix B verses S issues

Fic the class to handle recipes which set B to somewhere other than S.

(From OE-Core rev: 9d88baea4f3668612c76201c306ec6efe72c281e)

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

slang: Fix .debug packaging warning after debug striping was fixed

(From OE-Core rev: 2cdec8044cef963a1cc376484b4c7fa29924d5dc)

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

sstate.bbclass: Enclose sstate search parameter with quotes

In case filenames have spaces, execution of the function
sstater_install will hang, so the print parameter %s must be
enclosed with quotes.

(From OE-Core rev: 545d7aa26dfefdc927e0f4e2cc37398ef2c63fa6)

Signed-off-by: Leonardo Sandoval <leonardo.sandoval@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

insane.bbclass: Add ALLOW_EMPTY to list of package specific variables

ALLOW_EMPTY should have a package specified so extend sanity checks to
cover it.

(From OE-Core rev: 02849a54d38c0b98db7ff6b52b4974ee0d88bfb4)

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

packagegroup: Set ALLOW_EMPTY on a per package basis

ALLOW_EMPTY should be set on a per package basis, this updates the code
to do this, avoiding warnings.

(From OE-Core rev: 8981ed20234c42dc1f5cdef802ebe7214bd55238)

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

package.bbclass: Ensure all .so files get stripped

It was realised that .so files which were not marked as executable were not
gettings stripped. This was wasting space in images. This patch ensures
they do get processed by the code correctly.

[YOCTO #3973]

(From OE-Core rev: 725354886ae3650a7a4875d4c0bffcfab7e8cc40)

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

avahi: fix installed but not shipped warning

The avahi recipe has a warning when build without systemd feature:

WARNING: QA Issue: avahi: Files/directories were installed but not shipped
  /usr/share/dbus-1

The /usr/share/dbus-1 is empty without systemd feature, there would be
/usr/share/dbus-1/system-services with systemd. Use rmdir to fix the
problem, and remove the ${datadir}/dbus-1/interfaces from the FILES list
which had been removed clearly by the do_install.

[YOCTO #3939]

(From OE-Core rev: 51cfcef15b59f7b97c871d9c0d6bf560c7b06c9b)

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

perf: disable PARALLEL_MAKE

Autobuilder builds periodically and now more frequently have been
failing because of a race between the perf build and the newly
separated libtraceevent - perf tries to link libtraceevent.a, which
hasn't finished building yet in those cases.

This disables the parallel build to prevent that.

(From OE-Core rev: cd1e98513016c01e32bdb175ec7225d7378d952c)

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

package.bbclass: Fix do_package variable dependencies

When the do_package function was split into different sections, the vardeps
were not updated to match meaning some function changes weren't reflected
in the sstate checksums. Fix this.

[YOCTO #3980]

(From OE-Core rev: 2e7ddbcca3f613c3b5725dcc3d457985d41e5f16)

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

alsa-utils: Remove bogus comment

alsaconf clearly is packaged, remove obsolete comment

(From OE-Core rev: 9420ffd8bea1dd862ec26443f6197cdc7617a25a)

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

populate_sdk_base/image: Fix races for variable mappings

When using the -c populate_sdk option, images are not generated quite as
they should be under certain circumstances. For example the dropbear
feature may not get replaced with openssh, leading to both being installed
with an appropriate rootfs failure.

This patch moves the remapping logic to later points in the code, ensuring
there is no conflict. The result is slightly simpler too as an added bonus.

[YOCTO #3749]

(From OE-Core rev: 90cfa16bd4a02ada84ef94f6ae6f182beb8bdc01)

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

sysprof: upgrade to 1.2.0

(From OE-Core rev: aef171f824659b255b0ed2f6430bb757fc26f9c2)

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

puzzles: upgrade to 9765

(From OE-Core rev: aea07f4010a00d288fab241fd755df4f1063c234)

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

ethtool: upgrade to 3.8

(From OE-Core rev: 81bdf523f01fe9e9e9053e20a687ba7de5041cbf)

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

gpgme: upgrade to 1.4.0

License md5 checkum change due to:
- copyright year modification
- space modifications

(From OE-Core rev: a8968a05e2cac5ce9de2a42f71c81437906e7508)

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

libpng : update to 1.60.0

(From OE-Core rev: aca6a67527d97b57b3d16848057bc29fff7c19d0)

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

git: updated to 1.8.1.4

(From OE-Core rev: 86e0415d81809e86aae210fa0fe9ef80d4431b56)

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

libtimedate-perl: updated to 2.30

(From OE-Core rev: c75d2ee3a284549aa983157d8ad8e3e2227961fa)

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

gmp: updated to 5.1.0a

(From OE-Core rev: 46a29592a781b20472e6d749ee8754ebe29f5a80)

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

lttng: updated lttng-ust and lttng-modules to 2.1.1

(From OE-Core rev: a895f5c535c9cbd71d2965c1197ffb31afd831a8)

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

qemu: Upgrade to 1.4.0 release

Tested qemu on x86-64 target and qemu-native
no obvious problems seen in testing

(From OE-Core rev: f479201fffda617e3530967a75ab350739574b4b)

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

apt: upgrade to 0.9.7.7

bb files: remove invalid config.sub/config.guess files (autotools will
generate the correct files)
allocate-larger-memory.patch: not needed anymore
fix-gcc-4.6-null-not-defined.patch: added
- include header that defines NULL
includes-fix.patch: not needed anymore
localefixes.patch: removed
- not necessary anymore (no compiling errors)
makerace.patch: adapted to the new version
no-ko-translation.patch: adapted to the new version
no-nls-dpkg.patch: added
- fix build without nls
noconfigure.patch: adapted to the new version
nodoc.patch: adapted to the new version (apply patch on all cases
because the doc generation needs docbook-xsl)
remove-redeclaration.patch: not needed anymore
truncate-filename.patch: moved
use-host.patch: adapted to the new version
no-curl.patch: adapted to the new version
disable-configure-in-makefile.patch: added
    - do not run configure at do_compile

(From OE-Core rev: 97403b14765331c8c48bb570c6b98f2809214a9a)

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

sstate: Add gtk-update-icon-cache-native to postinst recipes list

(From OE-Core rev: 59cd32a6661d76ee070810823614eb797d6cc153)

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

gtk-icon-cache: use virtual/gtk-update-icon-cache-native instead of gtk+-native

(From OE-Core rev: 135e512c09710d65155e345033e93f905de42b0d)

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

gtk+: mark as provider of virtual/gtk-update-icon-cache-native

(From OE-Core rev: e2f8a6576906d0383c19d3ebefb8aabe7078904e)

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

default-providers: add default virtual provider for gtk-update-icon-cache

Use a virtual provider instead of a hard dependency so that if gtk+-native is
required in some configuration, this provider can be changed and then
gtk+-native and gtk-update-icon-cache-native won't be both built and conflict in
the sysroot.

(From OE-Core rev: 73c5458c7f041157832123696814b02df2b55090)

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

gtk-update-icon-cache-native: add

This recipe builds natively just the gtk-update-icon-cache binary that is
required to run the gtk-icon-cache.bbclass postinstall scripts.

The advantage of doing this is it means running 400 less tasks which takes four
minutes on my machine, as the alternative is building GTK+ natively (and so
libX11, freetype, fontconfig...).

(From OE-Core rev: 8265ef7f3a3598ff4535da1e5f9329fcf236f776)

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

sstate.bbclass: consider postinstall dependencies too

When checking the dependencies in setscene_depvalid(), make sure we also
consider those dependencies needed when running the postinstalls on
host.

[YOCTO #3918]

(From OE-Core rev: 8de0616825ed1b238b3486077af6897834bcb62d)

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

apr: add libtool to SSTATE_SCAN_FILES

The hardcoded paths in libtool will cause apache2 compile fail at
link stage when the package apr is installed from sstate cache, so
add libtool to SSTATE_SCAN_FILES.

[ CQID: WIND00405372 ]

(From OE-Core rev: 9d590d714c67ffcf1cd3f3844f5b102f72c2d006)

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

documentation: Fix P entry

P is PN-PV, not PN-PF.

(From OE-Core rev: c48f774e0dc46c89a8a7b5c0045d6de7acdbb112)

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

buildhistory: Allow the version backwards check to be disabled

When the PR service is not enabled, the versions can go backwards and there
are cases we don't want to know about this. This patch adds a variable
allowing these checks to be turned off whilst still using buildhistory.

(From OE-Core rev: 49f33a42e93711c96fbb099f3e6ee7f092d7b356)

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

recipes: Fix ALLOW_EMPTY with no package specified

There are various usages of ALLOW_EMPTY with no packages specified. This
is not recommended syntax, nor is it likely to be supported in the future.
This patch improves the references in OE-Core, either removing them if they're
pointless (e.g. when PACKAGES="") or specifying which package it applies to.

(From OE-Core rev: fe81bd4f600877e63433184cfc2e22c33bb77db6)

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

kernel-arch.bbclass: don't pass extra args to KERNEL_{CC, LD, AR}

Commit a3ca06c3b45c791dd38dbf154ad0fc62b4063475 added extra args to
KERNEL_CC, these revert that bit while maintaining machine additions
in TARGET_??_KERNEL_ARCH

Producing incorrect instructions and generating the following errors:

   CC      arch/powerpc/kernel/init_task.o
| {standard input}: Assembler messages:
| {standard input}:384: Error: junk at end of line: `1'
| {standard input}:949: Error: junk at end of line: `1'
| {standard input}:1428: Error: junk at end of line: `1'
| {standard input}:1526: Error: junk at end of line: `1'
| {standard input}:1626: Error: junk at end of line: `1'
| {standard input}:1685: Error: junk at end of line: `1'
| {standard input}:1973: Error: junk at end of line: `1'
| {standard input}:2001: Error: junk at end of line: `1'
| {standard input}:2100: Error: junk at end of line: `1'
| {standard input}:2168: Error: junk at end of line: `1'
| {standard input}:2520: Error: junk at end of line: `1'
|   CC      arch/powerpc/lib/locks.o

(From OE-Core rev: 6ed278ab93e9c3bb291dd37837c27e120954afc7)

Signed-off-by: Matthew McClintock <msm@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oprofile: update patch for powerpc64

This is a more appropriate follow up patch from upstream. Also,
only powerpc64 requires libpfm4 currently for this specific
version of oprofile (x86, sparc can make use of libpfm but
don't make use of it here)

Additionally, this patch from upstream requires some more
patches to be pulled into oprofile

(From OE-Core rev: 2792e1f6a1d8969e0891334e6cd4e04f84f7e9ff)

Signed-off-by: Matthew McClintock <msm@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libpfm4_4.3.0.bb: use compatible host to limit arches

This is only test with powerpc64 currently, even though
it should work on other arches such as x86 and sparc. When
thos are testing and working this COMPATIBLE_HOST should
be updated

(From OE-Core rev: 09f57ae44e1f610e615463149fb45ccfab1d5135)

Signed-off-by: Matthew McClintock <msm@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc-common.inc: handle case where tune is not defined

(From OE-Core rev: d3eab8b17f8e50f99042a7a8f43db94640c53d41)

Signed-off-by: Matthew McClintock <msm@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

iproute2: Fix build failure on ppc

Latest version bump started failing for ppc32
attached patch is a backport from 3.9-rc1

(From OE-Core rev: 08be60d903ee97859295387b3418d5db85ab16e1)

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

package_rpm.bbclass: Optimise per file dependency handling

Currently the process for injecting the per file rpm dependencies into
rpmbuild is painfully slow. Its done through the repeated execution of
a script which has to return the correct value in each case. This continual
execution means the CPU usage of rpmbuild is low.

This patch allows the option of collapsing the per file dependencies to
a per package basis and injecting them through the .spec file. This removes
the execution overhead and allows rpmbuild to run at 100% of cpu.

Ultimately it would be nice to inject the per file dependencies through
the .spec file however that is not currently possible.

Since few people use the per file dependency information, this patch
goes for the faster approach. It can be enabled if anyone needs it although
I'd mention that its being used to us as this code may well go away in
the future if nobody complains.

(From OE-Core rev: be40f6d0bb80274366af00461112af65687a4de8)

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

kernel: extract functions for kernel modules to separate bbclass

* this way we can reuse the same functionality also for external modules
  including module_autoload_foo and module_conf_foo functionality
* MODULE_PACKAGES variable was removed (splited modules are now returned
  by do_split_packages
* KERNEL_MODULES_META_PACKAGE is used to append all splitted packages
  to RDEPENDS. In kernel.bbclass it's old "kernel-modules" in
  module.bbclass it defaults to ${PN} for upgrade path from
  single PN with all modules to PN depending on all new kernel-module-*

(From OE-Core rev: 51928b6b5ca0a46a9dcd754483a19af58b95fa18)

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

bitbake: siggen: add quotes around variable values to see whitespace

* without this change it's hard to see what's changed:
  bitbake-diffsigs sstate-cache/*/sstate-depmodwrapper-cross-cortexa8-vfp-neon-oe-linux-gnueabi-1.0-r0-cortexa8-vfp-neon-*_populate-sysroot.tgz.siginfo
  basehash changed from 62f28c13172d38af917716d6d4b15ac5 to 9cf38842bcc79e0a4356856a64141548
  Variable TUNE_CCARGS value changed from  -march=armv7-a    -mthumb -mthumb-interwork -mfloat-abi=softfp -mfpu=neon -mtune=cortex-a8  to  -march=armv7-a    -mthumb -mthumb-interwork -mfloat-abi=softfp -mfpu=neon -mtune=cortex-a8
  Hash for dependent task depmodwrapper-cross_1.0.bb.do_install changed from 4494cb6c117309a9b4c541ad86c0aac5 to 1e3bcf2d14e5bd30a38350b41bb98db3

* with this change, we can see extra trailing space in first version:
  basehash changed from 62f28c13172d38af917716d6d4b15ac5 to 9cf38842bcc79e0a4356856a64141548
  Variable TUNE_CCARGS value changed from ' -march=armv7-a    -mthumb -mthumb-interwork -mfloat-abi=softfp -mfpu=neon -mtune=cortex-a8 ' to ' -march=armv7-a    -mthumb -mthumb-interwork -mfloat-abi=softfp -mfpu=neon -mtune=cortex-a8'
  Hash for dependent task depmodwrapper-cross_1.0.bb.do_install changed from 4494cb6c117309a9b4c541ad86c0aac5 to 1e3bcf2d14e5bd30a38350b41bb98db3

(Bitbake rev: 7d85b1a383feaa072503d71d3baeb3bdcd1012a8)

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

btrfs-tools: added PV to git package recipe

(From OE-Core rev: 889a36dffb65e73e8fea125c0923803679adf2ae)

Signed-off-by: Emilia Ciobanu <emilia.maria.silvia.ciobanu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Python: Fix for CVE-2012-2135

Reference:http://bugs.python.org/issue14579

The utf-16 decoder in Python 3.1 through 3.3 does not update the
aligned_end variable after calling the unicode_decode_call_errorhandler
function, which allows remote attackers to obtain sensitive information
(process memory) or cause a denial of service (memory corruption and crash)
via unspecified vectors.

http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2012-2135

[YOCTO #3450]

(From OE-Core rev: f60d3efe93323b7056a9400a483e625a3fed4491)

Signed-off-by: yanjun.zhu <yanjun.zhu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

weston: recommend some reasonable fonts so the terminal works

(From OE-Core rev: 5d3e26d12706606f804118fb77f96f0ada025e36)

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

weston: add dependency on xkeyboard-config

(From OE-Core rev: 603c329448ec761cb8e2faa6bb06430b70bb7d05)

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

linux-yocto: remove 3.0 linux-yocto recipes

With the introduction of the 3.8 linux-yocto recipes, we remove
the 3.0 variant.

(From OE-Core rev: 6835f1ecd7012bd4616efa3129bcf03b5be542ca)

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

qemuimage-tests/dmesg: Add exception for error message on qemuarm with 3.8 kernel

With 3.8 kernels we get a harmless error message during kernel boot.
For now we can ignore. This allows the tests to pass and ensures we
can merge the 3.8 kernel.

(From OE-Core rev: da1e094c407353af2ab230c4867c9d8fd68e3161)

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

linux-yocto: introduce 3.8 kernel recipe

Introducing the linux-yocto 3.8 kernel recipe. The tools and branch structure
of this tree are the same as the previous linux-yocto recipes, while the meta
data and content have been updated for the 3.8 kernel.

build and boot tested for qemu*.

(From OE-Core rev: ecf287a80c5ace2c440f8d0089934d59d25dc3ec)

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

linux-libc-headers: update to v3.8

Now that the 3.8 kernel has been released we can bump the libc-headers
to that version and remove the 3.7 variant. Userspace compatibility is
maintained through kernel versions, we also make the single 3.8 version
the toolchain default.

(From OE-Core rev: 4f9ef639143d890e9d2e71fea3b461fcc8e3f678)

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

kern-tools: import configuration, controls and audit updates

Updating the kern-tools SRCREV to import the following fixes:

  commit 7f91d198d32fc90260e52724ef4aac0b997c1e8b

    kconf_check: fix new Kconfig detection

    One of the functions of the kernel configuration audit is to notify
    the user if Kconfig* files have been removed from the kernel, and
    also to notify of new Kconfig files.

    New Kconfig files should be classified as hardware or non-hardware to
    allow BSP audits to notify if boards are setting values that they
    shouldn't, hence why notifying about new "buckets" is important.

  commit c4f26a3296e0e1c3dbdd5ec8e2947d5443a9ffc2

    updateme/scc: allow config fragment exclusion

    It is common to need the features (patches, git operations) of a
    branch, but not want the kernel configuration fragments of a given
    branch. To allow this, we provide a new include flag "nocfg".

    When this flag is used, all of the configuration fragments included
    by the targetted feature will not be applied to the current build,
    with one exception, a base/critical fragment can force it's config
    values, since without them, the system would not be functional.

    Example:

       include ktypes/standard/standard.scc nocfg

  commit c7ec19d55aca6c4b17073c5362fce5be61a89d82

    scc: wrap git merge

    To allow for parameter validation and sanity checking, wrap "git merge"
    as a dedicated "merge" command instead of using the raw git fallback.

    This also makes it consistent with existing top level commands such
    as  'tag', 'branch', 'patch', etc.

    There are no changes to arguments, and existing 'git merge' commands
    continue to work with this change.

[YOCTO #3419]
[YOCTO #3421]

(From OE-Core rev: faf042b2c87874153a6b689479ab86e49804af8c)

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

chkconfig-alternatives-native: just append to PROVIDES

(From OE-Core rev: f0c600ce8971c5444198ffe4f596b671e3e668cf)

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

eglibc: move RPROVIDES to eglibc-package

* PROVIDES_${PN}-dbg doesn't look correct, it should be RPROVIDES
  and that was already in eglibc-package
* RPROVIDES_${PN}-dev was overwritten in eglibc-package, but add
  those 2 entries now

(From OE-Core rev: 2edf614729a71f9ec15d00eeeea557a97e673284)

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

bitbake.conf: move PERSISTENT_DIR outside TMPDIR

* PR service cache needs to be persistent between rebuilds
  having it in directory starting with tmp* does not help
  people to understand that it needs to be persistent, so
  move it to TOPDIR

(From OE-Core rev: 5d43f752429707b74dbf46ecb81ac76bad7bb715)

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

depmodwrapper-cross: Mark as machine specific

The kernel and its staging directory are machine specific so the wrapper
needs to be as well. Also take the opportunity to remove the default
dependencies of the recipe since its a script and doesn't need the cross
compiler. This gives the build a little more scheduing freedom.

(From OE-Core rev: 1cc2ac769baae19b54184df5f43a3e77c1743c29)

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

sanity.bbclass: always use oe.lsb.distro_identifier method

* even when /etc/redhat-release or /etc/SuSE-release exists
* don't read /etc/lsb-release manually, NATIVELSBSTRING is not
  reading it too

(From OE-Core rev: 3dba05d74821f08cd96a18351805c8812ddd626c)

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

package_deb.bbclass:fix meta-toolchain-sdk fail on do_populate_sdk

When build meta-toolchain-sdk in a newly created environment, there is
an error:
...
packagegroup-core-standalone-gmae-sdk-target set to manually installed.
You might want to run `apt-get -f install' to correct these:
The following packages have unmet dependencies:
avahi-dev: Depends: avahi (= 0.6.31-r6.1) but it is not going to be installed
E: Unmet dependencies. Try 'apt-get -f install' with no packages (or specify a
solution).
...
In this case, avahi was installed and then removed by dpkg (the reason is
unknown, only with log `Noting disappearance of avahi, which has been
completely replaced'), the uninstall was done by dpkg rather than apt,
and apt detected the package dependency was broken, so apt-get install failed.

Use `apt-get install -f' to correct the upper broken dependencies in place.
The removed avahi will be reinstalled.

[YOCTO #3720]

(From OE-Core rev: a66ed54490305380c549838f9a580528d5b49275)

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

package_deb.bbclass:fix the arch (replace "_" with "-") in deb package control

when build deb image, such as building meta-toolchain-sdk in x86_64 host,
there is warning like that:
...
'x86_64' is not a valid architecture name: character `_' not allowed
(only letters, digits and characters `-')
...

The params in deb package control file don't allow character `_', only
letters, digits and characters `-' allowed. Change the arch's "_" to "-"
in the deb package's control file at the control file creation time. Such
as `x86_64'-->`x86-64'

[YOCTO #3721]

(From OE-Core rev: 8487b352cabd8c8ae8a7d9e7e66489e4e964bd50)

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

package_rpm.bbclass: Fix translate_smart_to_oe arch comparison

When the OE arch is of the format "foo_bar-foobar" the previous
comparison routine did not selectively translate the '-' causing
a failed comparison.

In order to work around this issue, we -always- compare the
RPM translated version of the package architectures.

(From OE-Core rev: caf798da9502c1c4967d13a4450fe02b1e7b4850)

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

remake: added PV to git package recipe

(From OE-Core rev: fa6b65e1f6192baa284ab5549ba00df0bbedf53b)

Signed-off-by: Emilia Ciobanu <emilia.maria.silvia.ciobanu@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemu.inc: Non deterministic compile of qemu

When you using a qemuppc target and sstate you might end up with
the problem:

qemu-system-ppc: error while loading shared libraries:
   libfdt.so.1: cannot open shared object file: No such file or directory

The way you can force this to happen assuming you are using sstate is
as follows:

bitbake dtc-native
bitbake -c cleansstate qemu-native
bitbake qemu-native
bitbake -c clean dtc-native

Now go start qemu and it will fail.  The solution is to always build
the dtc libraries since they are used and needed by the qemuppc
simulator.

(From OE-Core rev: 79c620016b0be4dacc9fcfbd4baab5ea6c66a440)

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

Add scratch directory for NFS upcall state storage

When the /var/lib/nfs/v4recovery directory is not present, NFSD cannot
record the upcall state and cannot end the 90-second grace period on
startup.  This is true even when NFS4 is not being served according to
nfsstat.

More details and discussion here:
https://lkml.org/lkml/2012/6/11/206

(From OE-Core rev: c7cb5cb92770615ffd302f737d0fef3f2300c321)

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

zip: explictly disable bzip2 support

* we don't depends on bzip2, so make sure it's not autodetected in
  sysroot

(From OE-Core rev: e93186642708cb99bfba776012e345c42e9bd0d5)

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>

perf: add bash dependency

There are a number of scripts in the perf installation that use bash,
so we need to add a run-time dependency on bash for them. If not, we
can generate build errors like "no package provides /bin/bash".

Fixes [YOCTO #3951].

(From OE-Core rev: 8169adeba67180f062fa7c0105fc4283da3103c6)

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

systemtap: add python and bash dependencies

There are a number of scripts in the systemtap installation that use
python and bash, so we need to add run-time dependencies for them.  If
not, we can generate build errors like "no package provides
/usr/bin/python".

Fixes [YOCTO #3951].

(From OE-Core rev: 8a0d31af7008459d1affa0afe3ff193a78f9020d)

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

Help ncurses-native find terminfo on RHEL 6

[ CQID: WIND00405305 ]

When using ncurses from sstate cache it is not able to find terminfo
for screen on RHEL 6.  While RHEL 6 has /lib/terminfo, it is not
complete, the complete terminfo database is in /usr/share/terminfo.  This
results failure of programs linked to ncurses, such as menuconfig.

(From OE-Core rev: b2f5c8f7dfda0bc8f07561336c8667c864d9c3bb)

Signed-off-by: Jeffrey C Honig <jeffrey.honig@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

eglibc: fix ldd RTLDLIST

The sed expression for retrieving the previous RTLDLIST from ldd script
was also retrieving the double-quotes. Hence, we ended up with a
RTLDLIST like below and ldd would fail:

RTLDLIST=""/lib/ld-linux.so.2 /lib64/ld-linux-x86-64.so.2
/libx32/ld-linux-x32.so.2" /lib/ld-linux-x86-64.so.2"

[YOCTO #3903]

(From OE-Core rev: f2804dc9a9ba666de85b4efb1fd005fa985f4f8a)

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

rm_work: remove package and packages-split dirs

* as described in
  http://git.openembedded.org/openembedded-core/commit/?id=6107ee294afde395e39d084c33e8e94013c625a9

(From OE-Core rev: 4067afcda78d17058f2aa8d7f82173d181e0aae4)

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>

cups: remove old patch files

(From OE-Core rev: c3c51a5961f77d51e7f8eb1f0746d16576663bba)

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

pulseaudio: add config option to ensure valgrind is excluded

There have been occasions that pulseaudio sees valgrind and then later it's
not available, adding this setting ensures determinism by disabling it by
default

(From OE-Core rev: 61001cd4d8a36a980b6e6967cba6f0ea48fdaebc)

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

Add KERNEL_EXTRA_ARGS parameter

With recent kernel versions, some ARM configurations need may fail to
build with errors like this:

multiple load addresses: 0x80008000 0x80008000
This is incompatible with uImages
Specify LOADADDR on the commandline to build an uImage

We cannot pass this information in EXTRA_OEMAKE, as
"meta/classes/kernel.bbclass" explicitly ignores all EXTRA_OEMAKE
settings.  So add KERNEL_EXTRA_ARGS parameter so affected boards
can add for example

KERNEL_EXTRA_ARGS = "LOADADDR=0x80008000"

to their configuration.  We make this general enough so other needed
command line options can be passed as well when compiling the kernel.

(From OE-Core rev: e3cdd0aafb9081cd7f1f98490087c1157a3d7403)

Signed-off-by: Wolfgang Denk <wd@denx.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

module.bbclass: Create a new depmodwrapper to assist cross-installs

Previously the build path to STAGING_KERNEL_DIR was being embedded into the
package post install scripts.  We avoid this behavior by generating a special
depmodwrapper script.  This script contains that hard-coded path, ensuring
that re-use of the sstate-cache (and/or packages) will always run through the
wrapper generated by the current build with a checksum that includes
STAGING_KERNEL_DIR.

[ YOCTO #3962 ]

(From OE-Core rev: b18c61bae4d7161c087a004bba3c696006f7a2f6)

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

sanity: use lsb distro_identifier

* distro_identifier returns lsb_data['Distributor ID']-lsb_data['Release']
  which in some cases is different then lsb_release -d -s, e.g.
  Distributor ID: Ubuntu
  Description:    Ubuntu 12.04 LTS
  Release:        12.04
  Codename:       precise
  But we probably don't need to sanity list each point release in LTS
  and sstate is already using distro_identifier as prefix for native
  sstate archives
* This will need update to SANITY_TESTED_DISTROS (at least
  s/Ubuntu 12.04 LTS/Ubuntu-12.04/g etc), that's why sending as RFC
  first.

(From OE-Core rev: a32ea459ca168792161af2d521a14fe00cffb4b3)

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

package_rpm.bbclass: add more description for pre/post scriptlets

When write rpm spec file, one description line

"# pkgname - script_type"

is added to pre/post scriptlets for base package but no such line
writted for subpackage.

Add similiar line to subpackage to facilitate handling the pre/post
scriptlets.

(From OE-Core rev: 34f3926f1787f9a2efa2257a2aaea6ee92adf8a3)

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

package_rpm.bbclass: name postinst files with package name

When create rootfs, some post install scripts need be run on device.
They are saved under directory /etc/rpm-postinst and named with numbers
such as 100, 101 etc.

Update to name the postinst script file with its package name instead of
just number. That may be more easy to debug when there is a error.

[YOCTO #3218]

(From OE-Core rev: 9b2a008c25a7a3152ae2d8c64ae8de2534471d47)

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

perl-nativesdk:fix toolchain relocation issues

1, Use create_wrapper to create perl wrapper in the SDK.
2, Add perl.real to perl-nativesdk package.

[YOCTO #3338]

(From OE-Core rev: 643cdac63353527c1b5cb6eeabe75df8d0dc8346)

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

populate_sdk_base.bbclass:fix toolchain relocation issues

When run "autoreconf" in toolchain, there is an error if the host's perl's
version is not the same as the one in the SDK, the error says that the
executable perl mismatches the perl lib's version.

This is because most of the autotools' scripts use the "#!/usr/bin/perl -w"
which is host perl, but the gnu-configize uses "#! /usr/bin/env perl" which
invokes the perl wrapper in the SDK, and the wrapper will set the PERL5LIB to
the SDK which causes the mismatch. We can make all the perl scripts to use the
host perl or the SDK perl to fix this problem.

[YOCTO #3338]

(From OE-Core rev: 365aa1412a3d31128cb3df02959acdc7df5f2ebc)

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

matchbox-session-sato: do not rdepend on initscripts

Previously, matchbox-session-sato depended on initscripts, thus resulting
in initscripts package built and installed in case of systemd init manager.

But there is actually no dependency between them. So this patch removes
the dependency.

[YOCTO #3940]

(From OE-Core rev: 8fbf22ef856a68e4f2e9ddabe33334f60616e16c)

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

shadow: remove grpconv argc check

The patched version of grpconv takes arguments but the check on
argc was not removed. This patch removes this check which
otherwise results in a spurious warning during rootfs creation.

(From OE-Core rev: 845a24e78835e93807cfb810fa99715ac4d14e21)

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

kernel.bbclass: Add missing modules.order and modules.builtins

When running depmod, from kmod, warnings about missing modules.order
and modules.builtins occur.  Looking back it is not clear why these
files were being ignored, other then nobody knew what to do with them.

Add the files into the kernel-base package, if they exist, ensuring
they will always get installed if at least one module is installed.

[ YOCTO #3963 ]

(From OE-Core rev: bff4bda1fe86ddd218a192a14605924205a5300b)

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

Update pseudo to 1.5.1

pseudo 1.5's enable-force-async works great, unless you use a host
where, inexplicably, stat(2) reports inconsistent and changing values
for a file's size or times for some time unless a file has been fsynced,
in which case you might want a way to cause an fsync to work.

Also noticed that some recent changes never made it into the docs, so
I did a little cleanup there. And changed the way NDEBUG suppresses
pseudo's debug messages, so arguments to them with possible side
effects (like calls into functions in another translation unit) can be
omitted, which should drastically reduce computational time if anyone
ever uses NDEBUG.

(From OE-Core rev: 150174d52adefdefe62e2ed0598665481591e4c2)

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>

v86d: compile against userspace safe kernel headers

v86d was building directly against the staged kernel in the target
sysroot. This has been fine in the past, but with recent (3.7) changes
to user headers into the "uapi" include structure, there are no longer

Without the preprocesor protection, v86d fails to build against any
kernel with uapi header files.

v86d doesn't actually need the whole kernel tree, the exported headers
are enough to build a working binary. This change sets the v86d variable
to ${STAGING_DIR_HOST}/usr, and once the v86d build adds "include" to the
end of KDIR, we have a valid include path to the target sysroot headers.

This also works for pre-uapi kernel headers, so the change is safe for
most (if not all) kernel's with properly exported headers.

(From OE-Core rev: 8d4256b2c9f782368a16c1458e04ed61fa82691a)

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

linux-yocto/3.4: remove cedartrail machine

The cedartrail BSP is no longer support from yocto 1.4, so we remove
it from the tree.

(From OE-Core rev: a1ce9da1feb028bbfd861d075bc3b6fbe3300147)

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

kernel-yocto: allow building from fetcher source dir

The linux-yocto recipes themselves always set S="${WORKDIR}/linux" and
arrange for the fetcher default of ${WORKDIR}/git to be renamed before
building.

Part of this rename involves an assumption that the directory used by
the fetcher can be removed as part of the renaming process, or in fact
that renaming is required.

If a derived recipe uses S="${WORKDIR}/git", the checkout phase fails
since the kernel source is removed as part of the processing.

To fix this the code now detects this situation and does not clean the
source directory before renaming the fetcher default (and in fact does
not rename it at all).

(From OE-Core rev: e4ab5efea1a41297f63c96de97270136535b5f0b)

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

kernel-yocto: respect SRC_URI modified branch selection

Although the setting of KBRANCH is the suggested/primary way to interact with
the yocto kern-tools and the fetcher, some users may be more comfortable
modifying the SRC_URI branch parameter directly.

If they do, the tools will not force their branch and build output will be
different then they expect, in non obvious ways.

It's easy enough to detect this scenario, but checking the SRC_URI in the
same way that the git fetcher checks for the branch (and SRCREV). If we take
the value from the SRC_URI and use it directly in the patch/validate/update
routines, we'll stay consistent with KBRANCH if it is used, and also
automatically adapt to a manually changed branch parameter on the SRC_URI.

For all other users, there are no visible or behavioural changes as a result
of this change

(From OE-Core rev: cfce8643ed166b51d7178be173677ea6f527d453)

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

iproute2: don't use the kernel header tree

This should compile against the userspace safe kernel headers

(From OE-Core rev: e443887379fae4b4fd70d1cd6f7d0d7f064a2cb1)

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