bitbake: lib/bb/runqueue.py: fix exceptions with -k and failed targets

If a target dependency is marked as failed and yet we are continuing on
because -k has been specified, don't try to access the dependency's data
in taskData.build_targets since it will have been removed. This fixes
"IndexError: list index out of range" errors in this situation.

Also, do not print the "unhandled exception" message when SystemExit is
raised since we will have reported the actual error already in this
case (e.g. when -k has been specified and some targets failed).

Fixes [YOCTO #3133].

(Bitbake rev: 70eebc184eb1ab3678be87bed019b5beadecdc89)

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

bitbake: hob: fix Gtk-WARNINGs due to invalid markup on Back button

You can't use markup characters (e.g '<' or '>') in the labels for many
widgets - you must use the appropriate entities instead.

(Bitbake rev: 54a16ac999d4a2c4c3f8a4531e8c6fabc39a4147)

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

bitbake: hob: remove confirmation dialog on close

This is not necessary for modern applications - instead we just need to
check if we're in the middle of a build and if so, do the same thing as
pressing the "Stop" button.

(Bitbake rev: a79eb5d918239db1dade8134743e6142a4854930)

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

bitbake: hob: allow configuring default machine using HOB_MACHINE

Allow specifying HOB_MACHINE in local.conf to set the initially
selected machine. With this set, Hob will select the specified machine
and then jump straight into parsing recipes. If you do wish to change
the selected machine with HOB_MACHINE set you still can - you just need
to stop the parsing process first.

Fixes [YOCTO #3148].

(Bitbake rev: c3b623dc7d546a1ededdb532dcbcba4a6230bc65)

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

bitbake: hob: don't show error dialog for errors during building

During building we already report errors in a special tab and
indicate when the build has failed; bringing up a dialog was a
regression introduced in bitbake revision
5bab81b124087d63d6eb62a861e1241714fcd483.

Fixes [YOCTO #3151].

(Bitbake rev: cf0a67d62f631aa48d1afc3fbdd0f73995b1c401)

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

bitbake: hob: The title of the packages screen depends on the screen you arrive from

- If you arrive to the packages screen from the recipes screen, the title
should say: 'Step 2 of 2: Edit packages'
- If you arrive to the packages screen from the image details screen, the
title should say: 'Edit packages'
- The title of the recipes screen should say 'Step 1 of 2: Edit recipes'

[YOCTO #2982]

(Bitbake rev: c366f4314c29b873a486daa9a0a4e29bb4225dd6)

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

Fix Upstream-Status

These were not getting fixed by orignal committer!

(From OE-Core rev: 7db73c70351939c4be9867981a8cf97148bbe57e)

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

binutils-2.22: Disable recent gold backports from 2.22 branch

This patch has been causing some regressions on gold.
e.g. systemd based images segfault and uclibc based images
dont boot. There has been few other reports on the mailing
list. Considering this lets withdraw this patch.

(From OE-Core rev: ecbe671de1553956f83798e1c6fa3ec2fc6a7b4e)

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>

initscripts/sysfs.sh: mount debugfs if present

debugfs is another kernel virtual file system that should be mounted
if configured, so if it's configured into the kernel, mount it.

(From OE-Core rev: 55c4d3c55e4c3a7c2cda6d006cf7b78567bd3298)

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

rootfs_ipk: Avoid leaving run-postinsts around if online package management is disabled

If all postinsts have already been run during rootfs construction then
there's no point in having run-postinsts in the installed system.
Clean it up at the same time that update-rc.d and suchlike are being
removed.

(From OE-Core rev: b260cf9fbeb6f029c1ce45e77edd03968caa8288)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu: work with tap device names that end with a colon

On Fedora systems (and likely others), ifconfig returns interface
names that end with a colon. Make sure we strip the colon off the
tap device name before using it.

This fixes [YOCTO #3028]

(From OE-Core rev: 85ed217b603a86113dda11d952850e8ceed30795)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemu-native: fix build on hosts without libX11 installed

(From OE-Core rev: 6bb6ca6164b6b5f082fe05c30974463c2aa1c170)

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

libcap: respect ${base_sbindir}

Need to pass the path explicitly to "make install" to ensure that the binaries end up
in the right locations.

(From OE-Core rev: 3dffb9f577b794e8b78fb884adf9a15c45965bc4)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sqlite3: enable USE_PREAD

This gives a small but measureable performance improvement for
I/O heavy workloads.

(From OE-Core rev: e4bd65e68c3d0dd798ff69c2e9491e5b2dcafdc3)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/multilib: prevent multilib extension of nativesdk recipes

It isn't supported to mix multilib and nativesdk in the same target, so
explicitly skip multilib processing if nativesdk is inherited. As a
bonus this fixes a bunch of related "missing file" warnings from the
file checksum code during parsing because BPN was not correctly stripped
for these targets.

Second half of the fix for [YOCTO #3146].

(From OE-Core rev: d9a1eb5054d487affb94431374a9cb1a735e2122)

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

classes/multilib: ensure MLPREFIX is set for image recipes

We need MLPREFIX to be set so that oe.utils.prune_suffix() (as used for
the value of BPN) can derive the bare name from the multilib-extended
name for image recipes. BPN being set correctly avoids missing file
warnings during parse from the file checksum code for (unusual) images
that set SRC_URI, such as build-appliance-image.

First half of the fix for [YOCTO #3146].

(From OE-Core rev: ddec9a1b45159c75e97e92abe9a940268acd84b2)

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

pulseaudio: Ensure openssl is disabled

We don't DEPEND on openssl but configure will try to use it if
pkg-config thinks it might be installed.  This can lead to failing
and/or nondeterministic builds, so let's force it off.

(From OE-Core rev: 1cdcd754651f8d519ac8e4ba0d241d784e2a0090)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

autotools: Remove special handling for autoconf* and automake*

For reasons that are now shrouded in obscurity, autotools.bbclass
has long contained a special heuristic to avoid attempting to run
autoreconf when building autoconf or automake themselves.  However,
the wildcard test against PN which is used there is problematic when
trying to build another package whose name happens to start with
"autoconf", and in any case it is silly to do this test at runtime
for every package.  The individual recipes for autoconf and automake
can just as easily suppress the behaviour that they don't want by
providing a custom do_configure() method which just runs configure.

(From OE-Core rev: a87db6f8dea71cbb7ead9285ff8af0e28cf75604)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

dbus: Remove hardcoded reference to /usr in System V startup script

Use ${bindir} to locate the binary instead.

(From OE-Core rev: 8cf6f87bd753e1c84a018ddb92a97eed7bd79a28)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

arch-armv7a.inc: Don't disable vectorization

We have been adding this option to paper over a bug in old toolchain
http://hardwarebug.org/2008/11/28/codesourcery-fails-again/
e.g. is one but these have been weeded out. Therefore let gcc
take the default vectorization optimizations

(From OE-Core rev: e4336ab56db1e07a7f3dc08d3a4de3593b0fad22)

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

linux-yocto/3.4: Update fri2 and sys940x to emgd-1.14

Bumping the meta SRCREV to pickup a newer emgd for the fri2 and
sys940x.

(From OE-Core rev: abb23d987184f76fb922af15508f2e251399e3ed)

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

lsbtest: update list file

Update the lsbtest's list file that we can download the latest
sub-packages of LSB suite 4.1.0 and then run the test with them.

(From OE-Core rev: a2f81aa58c4753412afc0227c34f08134b6a3d2d)

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

util-linux: Ensure that ${sbindir} is respected

The configure script uses a hard-coded value for ${usrsbin_execdir},
which is the path that we know as ${sbindir}.  Adjust configure to take
this from the environment if it's set there, and have do_configure()
pass it in.

(From OE-Core rev: 6fdca45ec85e226f570917d2d1aaa2aa39ab6b42)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bash: Make it possible to run bash 3.2.48 instead of 4.2.

bash-3.2.48 did not provide the linking from sh to bash, making it unusable.
Moving the license part out of the bash.inc file, and into bash_4.2.bb file makes
us able to use that file also for bash_3.2.48.bb, which makes maintaining both
at the same time a lot easier.

(From OE-Core rev: e7b82cb4d107bfbfa5c939d406dd6ce6615b24e1)

Signed-off-by: Martin Ertsaas <mertsas@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

adt-installer: add sudo when relocating symlinks

This is needed if installation is done in a directory that needs root
privileges.

(From OE-Core rev: 28823486ba8ce4d88bbad3cea696ce9fba0cc165)

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

gcc: Use 4.7.2 release tarball

This avoids the SVN or git fetcher issues for gcc
and the tar is mirrored around the world so it will
not be slow

Fixes [YOCTO #2908]

(From OE-Core rev: 5e03d1e83d0536a2fc69a88d3e5407108836203f)

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

bitbake.conf/gcc-common.inc: Fix STAMPCLEAN expression

The globs used for STAMPCLEAN were too greedy matching gcc-cross-initial
stamps for gcc-cross for example. This patch resolves that problem making
the assumption that PV starts with something numeric. This assumption
should hold in most cases and has a better failure case that the current
situation.

(From OE-Core rev: d7fbc70b6c6ac629d2a23ac16ab45461f88b4b26)

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

insane.bbclass: Improve ability to detect enabled tests

Define an ALL_QA variable which can be used to determine which tests to
run. Improve the libdir test to work in the case it is set to raise an
error rather than a warning.

(From OE-Core rev: 069992a502658e8e44b870601e2e189cd9596ec9)

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

insane.bbclass: add library dir sanity check

Check in ${PKGD} for libraries in wrong locations.
Trigger a warning if so.
Eg. Catch recipe installing /lib/bar.so when ${base_libdir}="lib32"
or installing in /usr/lib64 when ${libdir}="/usr/lib"

[Yocto #2038]

(From OE-Core rev: 534fa3a55de19f249583207aaeec58fec8154a1d)

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

linux-yocto/3.4: use profiling and latencytop features in yocto bsps

Updating the meta SRCREVs to pickup the following configuration change:

    594994c meta: use profiling and latencytop features in yocto bsps

(From OE-Core rev: df0197810e72337093ea6ced1c8b28eeeef8a7a8)

Signed-off-by: Tom Zanussi <tom.zanussi@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>

SDK: allow toolchain installation from another directory

This patch will allow one to run the installer from another directory
than the one where it's actually located.

Suppose the installer is in /home/user/test/my/sdk and the current
directory is in a different place. With this patch, one can run the
installer like this:

$ sh ~/test/my/sdk/poky-eglibc-x86_64-arm-toolchain-gmae-1.2+snapshot-20120920.sh

[YOCTO #3135]

(From OE-Core rev: 3c7aac33cb63dc63b989db4e9d7389a7f4d3c18d)

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>

SDK: relocate symlinks too

The directory usr/libexec/ in the SDK sysroot contains the default
symlinks to the toolchain binaries and these, too, need to point to the
correct toolchain path.

[YOCTO #3090]

(From OE-Core rev: 6e4923c0c9b218271fd44d78df9987b5cabb1c03)

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>

SDK: remove references to Poky distro from tarball installer

The installer should be generic.

(From OE-Core rev: eba5cdf10aa49da844ac141f00b79327da0cb8c3)

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>

adt-installer: ensure directory exists before copying/removing

If the installation is done in a directory which already contains a
valid installation, opkg will not install anything and the moving the
contents of /install/dir/opt/poky/1.2 (for example) to /install/dir will
throw some errors. However, the install directory will not be affected.
This patch will ensure that the /install/dir/opt/poky/1.2 exists.

(From OE-Core rev: 1cd01533cbec0e9ed61bbd33ecdf5dc306a32eec)

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>

SDK: fix installation into symlinked directories

The SDK installation scripts should not canonicalize symlinked
directories because the entire relocation would be done to the directory
to which the symlink points. Instead, if the installation is a symlink,
use that path to relocate the binaries.

For example, if we have the following symlink: /opt/sdk -> ~/my/test/sdk
the binaries will be relocated to /opt/sdk not ~/my/test/sdk as it is
done now.

[YOCTO #3102]

(From OE-Core rev: 9e6a25e2e9a7f37c3baa0b2949a43ac4127868da)

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>

adt-installer: fix package installation issue

When the cross canadian toolchains are installed, for different
architectures, they might contain common files. This leads to
installation failures since the opkg, by default, does not overwrite
files. This issue happens, for example, for binutils packages (that
contain the same locale files) or gdb (which installs some syscalls xml
files). The locale files could be removed from the binutils
cross-canadian package but we cannot do the same for the syscalls GDB
files which are used by GDB to display user friendly names for the
syscall numbers. Hence, the best solution is to force opkg to overwrite
these files.

[YOCTO #3109]

(From OE-Core rev: 3396545467df05421c3adeb4b5ec532fa95dcb06)

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>

intltool: include intltool.m4 and add missing rdepends

- include /usr/share/aclocal/intltool.m4 into the intltool
package (the files from intltool-dev must be included into the
main package, as intltool is a development tool)
- add missing rdepends: gettext-dev, libxml-parser-perl

[YOCTO #2597]

(From OE-Core rev: f5f3d3fb14e983af114afc6425dc339053927f25)

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>

opkg svn: respect to the arch priority

This is for fixing the problem:
1) bitbake core-image-sato-sdk with MACHINE=qemux86
2) bitbake core-image-sato with with MACHINE=crownbay

The qemux86's PACKAGE_ARCH is i586, the crownbay's is core2, but several
i586 packages will be installed into crownbay's rootfs though there are
core2 packages. For example, there are:

xserver-xorg*_1.11.2-r7_i586.ipk
xserver-xorg*_1.9.3-r1_core2.ipk

The crownbay.conf says:
PREFERRED_VERSION_xserver-xorg ?= "1.9.3"

What the crownbay's image needs is xserver-xorg*_1.9.3-r1_core2.ipk, but
the xserver-xorg*_1.11.2-r7_i586.ipk will be installed, this is
incorrect.

This is caused by opkg's selecting mechanism: when more than one
candidate is found, it will use the higher version one and ignore the
arch priority.

we have several conf files which set the PREFERRED_VERSION_pkg = "..." ,
but there is no such a mechanism which can let us tell the opkg to
install the preferred version. When the preferred version is higher,
this is OK, but if the preferred version is lower, there would be
problems:

1) Most of the packages are core2 in the image, but several of them are
   i586, though we have built the core2 ones, this seems strange.

2) What's worse is that the image may not work since the preferred
   version pkg is not installed.

We have set the arch priority clearly in the opkg.conf, I think that
respect to the arch priority is reasonable during the image generation.

Add the "--select-higher-version" option to let the user have another
choice, the default is no.

[YOCTO #2575]

(From OE-Core rev: 0a80a02644f624443cef8cc4f604edb5ef8e6975)

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

gdb: Upgrade 7.4 -> 7.5

This is a simple upgrade. Dropping the unneeded patches
and adding --disable-werror to configure since thats is
what one of the patch was doing which was dropped.

(From OE-Core rev: 452f26b6d189b9fafba644e41921091925fb6a47)

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>

opkg-nogpg: drop SRCREV

* use SRCREV from opkg_svn.bb, because with 596 patches we have won't apply

(From OE-Core rev: 2bf865a9b89eb88c3f7c754362315841195cd8ca)

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>

opkg: add patch to fix SIGSEGV when printing status file

* it was triggered by c02364f36e228835ea5d7fd4e1d347fd451f8544 when new package had 2 entries in Provides and old version just 1

(From OE-Core rev: d98d6ec9425bd8764405c9812cddfcfd2a2b025b)

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>

opkg: replace local patches with git patches submitted upstream

(From OE-Core rev: 1f1ae93d8cd5140028e86d92483e349868b4f3f6)

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>

opkg 0.1.8: remove it since it doesn't work

Remove opkg_0.1.8.bb and the related files since it doesn't work:

- It doesn't support the "--force_postinstall" option which is used by
  package_ipk.bbclass.

- It still doesn't work after remove the "--force_postinstall" option,
  it can't install the packages in complementary_pkgs.txt.

[YOCTO #3136]

(From OE-Core rev: 526da07578de2c6261d21bc339bca0d3b94d93cf)

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

qemux86-64: Support for KVM, paravirt and virtio added

    KVM, paravirtualization and virtio drivers are now activated
    in runqemu using the kvm option flag for qemux86-64.
    Host CPU features are also exported to guest OS (Yocto Linux).
    Usage example: runqemu qemux86-64 core-image-x11 kvm

    Implements [YOCTO #2550].

(From OE-Core rev: cbb6431b3ee9128ea15c9ae0a19e7d2998ffc561)

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

qemux86: Support for KVM, paravirt and virtio added

KVM, paravirtualization and virtio drivers are now activated
in runqemu using the kvm option flag for qemux86.
Host CPU features are also exported to guest OS (Yocto Linux).
Usage example: runqemu qemux86 core-image-x11 kvm

Implements [YOCTO #2550].

(From OE-Core rev: a35d03e2eb905de4eadc9c7df5b50bff1fb7f897)

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

runqemu: show bitbake errors to user

In certain edge cases, bitbake may fail to run and cause setup_tmpdir()
within runqemu to fail, and not give the user a helpful error message.
Catch this case and show the user the output of bitbake -e.

This fixes [YOCTO #3112]

(From OE-Core rev: 465d7b6e66b5a55706535e194b3e44e11ee542c6)

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

librsvg: make the libcroco dependency a PACKAGECONFIG option

Default to enabling it as we were build-depending on it already.  If a user
needs the disk space and the limitations imposed by not using libcroco are
acceptable they can override this.

(From OE-Core rev: e177f1475c55c7d0bf3e2752e6502a7e8577a075)

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>

mkefidisk.sh: Add script to do an EFI install on the host

Sometimes it is convenient to prepare a bootable image from the
host rather than using a live-image to install to a disk on the
target.

This script takes a live image as input, partitions a device, and
performs the installation just as the installer would if run on
the target.

(From OE-Core rev: 7225c6739f9f1e51741a42437692868165aa1dfe)

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

rpm 5.4.9: DEPENDS on bison-native

The rpm should depend on bison-native, otherwise errors when "bitbake
rpm-native" in a fresh build:

| make[4]: Entering directory `/path/to/rpm-native-5.4.9-r46/rpm-5.4.9/syck/lib'
| bison -d -t -v -p syck -o gram.c gram.y
| make[4]: bison: Command not found

Basically, both the rpm-native and rpm should depend on bison-native,
but don't need depend on bison, but it seems that it isn't necessary
to add another depend line:

DEPENDS_virtclass-native = "libpcre-native ... bison-native"

So just add it to the DEPENDS.

[YOCTO #3123]

(From OE-Core rev: 839faed2e7ef554668f647732c7ee1c8d339c123)

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

gtk+: enable gtk+-native

This is needed in order to run postinst scriplets at do_rootfs time
rather than first boot time.

(From OE-Core rev: 9db6ea4981fe22dd430c13d9f5a838ad92d1afe6)

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

Replace "echo -e" with "printf" to have the same behavior in dash or bash

oe-core removed the prerequisite to have sh as bash. POSIX doesn't define
any options and furthermore allows 'echo -e' to be the default behavior.
This means that in dash 'echo -e' will actually print '-e' and interpret
backslashes by default. We use instead 'printf' builtin command with or
without '\n' to simulate 'echo -e' or 'echo -n'.
'printf' needs format while 'echo' can be used without any arguments. So
'echo >' was replaced by 'printf "" >'.
'echo' without '-n' flag adds a new line by default so to keep the same
behavior of two new lines while using 'echo "\n"', 'printf "\n\n"' is
used.

[YOCTO #3138]

(From OE-Core rev: a19880ad10ccb5d7d909dcf9de5c3dc58a0ebcd3)

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

byacc: update config.{sub, guess} before running configure

(From OE-Core rev: 6d8aeb0c9b939082cc8d54a940d615b33d81348d)

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

xcursor-transparent-theme: switch SRC_URI as matchbox-project is dead

(From OE-Core rev: bcd5ea36bd5e0dbb063f45fbc6d5d3272841fb0f)

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

bitbake: hob: Don't set busy cursor on the root window, just our window

[ YOCTO #3127 ]

(Bitbake rev: 5ef9d98b343b9ed05167e5471eb9f7f12e97b045)

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

bitbake: hob: add a top bar when building process is stopped

When a build was stopped, it wasn't obvious what to do next.
Now, on the page it appers a top bar with 3 buttons: "edit image",
"open log", "build new image"

[YOCTO #2537]
(Bitbake rev: a6afd15b7a40919313e26777b514ae44b587a0f6)

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

bitbake: cooker: Clarify package/recipe wording

When the '-s' option is run, change the heading above the list of recipes
to say "Recipe Name" instead of "Package Name".

(Bitbake rev: d1c3a9de875fb488a56ab5cb1d2f8e2f24f31d69)

Signed-off-by: Trevor Woerner <twoerner@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob/packageselectionpage: "Cancel" button returns to "Image configuration" screen

Once package building completes, you can customise the list of
packages that will go into the final image. Once you have made
the changes you need, you can either build your image, or you
can exit the process and go back to the 'Image configuration'
screen by selecting 'Cancel'.

[YOCTO #3105]
(Bitbake rev: c5fd1824c9794923576ec1e747536c0430542fd1)

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

bitbake.conf/gcc: Add clean masks for stamp files

This takes advantage of new bitbake functionality to clean up stale stamp
files when creating new stamp files.

[YOCTO #2961]

(From OE-Core rev: e21b6c04e512a3bc2339a20045b7041f1d26e859)

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

bitbake: hob: report event handler failures

If an event handler failed we were not recieving an error message -
parsing just "froze" at 99%. This is because we were expecting a
CommandFailure event and this never happened in the case of
RequestPackageInfo which is where the failure was occurring.

This also required tweaking the error formatting slightly, taking the
return value of the format function rather than the message property
since the latter only seems to contain the first line without the
traceback in the case of event handler failure. Other error cases were
tested and their message formatting is unaffected by this change.

Final part of the fix for [YOCTO #2651].

(Bitbake rev: 5bab81b124087d63d6eb62a861e1241714fcd483)

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

bitbake: hob/settings: Remove stray "distro" combobox from simple settings

The "distro" combobox was moved to advanced settings, but it
was also present in simple settings. This patch removed it
from simple settings.

(Bitbake rev: 6d56bec4464da14d7fde0e60946be43293ad6e52)

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

bitbake: build/siggen: Add support for stamp 'clean' masks

Currently when we execute a task, we don't remove other potentially stale
stamps. This can mean if you switch between two different versions of a
recipe without a clean, the build can get very confused.

This patch adds in functionality to allow a wildcard expression of stamp
files to be removed when creating a new stamp file. This patch adds in
the core of the code to enable this but it also requires metadata support
to enable it.

When writing this improvement I went through several different options but
this was the only way I could find to allow things like noexec tasks to
function correctly (where stamps need to be created without the data store).

[YOCTO #2961]

(Bitbake rev: e026469b307522e5b6a680e0ae5587749d33dcae)

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

bitbake: fetch2/cvs: Fix parameter spacing

Add in misssing space between the parameters. Reported by
Jate Sujjavanich <Jate.Sujjavanich@myfuelmaster.com>.

(Bitbake rev: 55382f0aac84b8f81cad0b82053c0b8295c33e54)

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

bitbake: fetch2/cvs: Clean up various data store references

The code in the CVS fetcher is elderly and there are simpler ways of
using the data store. This updates to use the modern APIs.

(Bitbake rev: 78eee8c70a80997293df99475153aed0b2ad0a17)

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

bitbake: fetch2/cvs: Fix localdata variable reference

The localdata variable was removed, fix up a lost reference to this.

(Bitbake rev: 02ccc1396005ce0b7a2150a5ce12b723df21d464)

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

bitbake: hob: Fix settings dialog issues

Fix some issues with the settings dialog that were introduced as
a result of merging the fix for #2162.

[YOCTO #3117]

(Bitbake rev: a363f59579e01cb7bd39be2ce73f22c875c62ce7)

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

bitbake: hob: rename 'View log' to 'Open log'

Rename all the 'View log' buttons to 'Open log' for
consistency.

[YOCTO #3045]

(Bitbake rev: 7bc9b0c1c2544b494959b13ac79ac3e52edb4fe3)

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

bitbake: compat, event: use OrderedDict from py2.7 for the event handlers

This ensures that our event handlers get run in registration order, making the
behavior more deterministic. I pulled in the python2.7 OrderedDict to avoid
essentially reimplementing a version of it ourselves, figuring we can drop it
when we bump our required python version next.

(Bitbake rev: 44aa0b0537d3fbd1272015e7677948f84d8c0607)

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

image_types: fix squashfs-lzma image creation

This removes the dependency on a seperate binary which we
don't seem to have. So, use mksquashfs's -comp lzma to
replace that functionality

[YOCTO #3126]

(From OE-Core rev: 8e82713724dfcb40f2ae24a166ec94f50f8b4cd0)

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

rootfs_rpm: Add Multilib prefix to installed_packages list

RPM does not name it's packages with the Multilib prefix,
but the rootfs_rpm class keeps track of the Multilib prefixs
in a list. Use that list to re-attach the prefix for use with
the license bbclass, buildhistory bbclass will also use this
and make it more accurate between multilib and non-multilib.
Use the embedded "Platform" information to ensure we get all
the correct matching.

(From OE-Core rev: f72abd80b0cc9d27aad2e31ecb548b4ab0fd8f67)

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

eglibc: Do not use fsqrt in libm when building for fsl ppc with fpu

(From OE-Core rev: f06097f4581e4c728c2950a86e025384e4bdcdf0)

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>

eglibc: Fix fcntl.h for powerpc

This fix is needed for systemd to work on powerpc

(From OE-Core rev: 76f3a1979ea166238e26a2569fb06a4a403bd864)

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>

nfs-utils: add x32 patch to fix nfsctl issue

nfsservctl syscall does not exist for x32, so return an error.

(From OE-Core rev: fddcb9dd086cfb396255ae5c8f717a39c6b9c4b0)

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

xserver-xorg: Modify RREPLACES for RCONFLICTS

fix bad runtime dependency that was causing -exa to be a suitable candidate for xserver-xorg, thus resulting in no X server in some situations

(From OE-Core rev: 467c59495d83748d35846e8b37548182fea99cbf)

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

util-linux: Fix bloken swapoff symlink

There were 2 issues with this symlink.
1. Is was installed in base_bindidir but packaged in bindir. Fixed to
be packaged in base_bindir
2. The symlink swapoff was created to point to swapon. The problem is
that swapoff is an alternative so it would end up pointing to
swapoff.util-linux which was an inexistent file. The fix is to create
a symlink swapoff.util-linux to swapon.util-linux.

(From OE-Core rev: 0ff32e8fb5463a23af9966afcb58eb00772af65b)

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

grub: disable lzma, device-mapper, zfs and nvpair

check-if-liblzma-is-disabled.patch: added
- add support for the --enable_liblzma option

[YOCTO #2750]

(From OE-Core rev: 1773a98b68dd223b016fe408b022c2c5475669c2)

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>

libtelepathy: PR bump to rebuild after libffi5 -> libffi6

(From OE-Core rev: 08dbc9790074fdb315d286849afbbda72d19cbd6)

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>

recipes: few more PR bumps to rebuild after libffi5 -> libffi6

(From OE-Core rev: e95835742d2f27820d654c309bd9828305276c90)

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>

recipes: bump PR to rebuild after libffi5 -> libffi6

(From OE-Core rev: 211200fb98a72ba815e7c411fbebfd781879064c)

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>

patch.bbclass: increase security

- Use mkdtemp for generating temp dir names
- Use bb.utils.remove for removing temp dirs
- Add comment for explaining the "patch" workaround

[YOCTO #3070]

(From OE-Core rev: fbe9fc4d5ece1e66b03b4c4bce9b7ffad3b5b138)

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

bitbake: Implement 'settings' dialog as designed

[YOCTO #2162]

(Bitbake rev: ac75b06744e73399ca1fbda322ef851ae5754b0a)

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

bitbake: cooker: Fix priority for virtual recipes

When making filename comparisons for recipes for priority calculations,
we need to split off any virtual prefix. Without this, BBCLASSEXTEND
version of recipes don't follow the priority settings they should.

[YOCTO #2933]

(Bitbake rev: 055b72a230e6b0b1cababd65372c62d9ddce385e)

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

bitbake: hob/builddetailspage: Add tooltips to the build failed notification

[YOCTO #3046]

(Bitbake rev: 69ad4ebd1379e804d0753bd4ee704b602b5efcc4)

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

bitbake: lib/bb/event: improve handling of event queue on exit

If BitBake exits before a UI handler (server) has been registered, we
print the event queue; if there are any errors or other non-debug
messages just print these and suppress the rest of the message queue.

This improves the output when sanity check failures occur with OE-Core
by avoiding printing a long stream of uninformative debug messages.

(Bitbake rev: 8668a94cb1841798636b68fe123400d6b81f6574)

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

bitbake: hob: format error messages properly

Error messages that use arguments need to be formatted properly, or we
don't get the full message. Use a formatter to do this when an error
occurs.

Partial fix for [YOCTO #2983].

(Bitbake rev: 6783538884adecd914909a9ab4ca73c27575f3ad)

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

bitbake: hob/imagedetailspage: change "FileCreated" label to "Files created"

[YOCTO #2998]

(Bitbake rev: ca2464561d54d59d1146359e41eb08201954fc21)

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

bitbake: Fixed hob proxy tab tooltip text per review suggestions [YOCTO #2499]

(Bitbake rev: 485e69d41e220ed4e8efc89a357a8d395e44da9f)

Signed-off-by: Jessica Zhang <jessica.zhang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob: sort base image drop-down list

Sort the list of base images to make it easier to find a specific image
in the list. Note that "Create your own image" still remains the last
item in the list.

(Bitbake rev: db16f575a774de7d9d44b4bc727b252de5d0f34d)

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

bitbake: lib/bb/command.py: ensure setVariable only sets values as strings

This is the interface Hob uses to set variable values in many instances,
and at the moment it is possible that some of the values it passes are
not strings. If a non-string value gets into the datastore it can
trigger exceptions during parsing when we attempt to expand the variable
and substitute in the non-string value.

This fixes using the meta-ti layer within Hob - it currently has a
reference to BB_NUMBER_THREADS within a shell function and since this
is a variable that Hob was setting from its configuration as an integer,
due to the above this was triggering an ExpansionError.

Fixes [YOCTO #2875].

(Bitbake rev: 855b71d8a8e468bfeff9e1a6699d79d68ab27aa1)

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

bitbake: hob: don't reorder layers list

We cannot reorder this list - it must stay in the order shown in the
dialog (which may in future be configurable by the user).

Fixes [YOCTO #2649].

(Bitbake rev: eca0352195d2d8ae8ef15baab9737884ec674a46)

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

linux-yocto/3.4: make uprobes select perf events

uprobes depends on functionality provided by perf events. After
uprobes was enabled in the standard kernel the mpc8315 board showed
link errors due to missing perf event functions.

This problem isn't isolated to the board or powerpc arch, but all
other boards have PERF_EVENTS enabled. To fix this, we make UPROBE_EVENT
select PERF_EVENTS, and any new boards will be protected from the
same failure.

We also update the configuration fragments since CONFIG_UPROBES depends on
CONFIG_PERF_EVENTS being set, so PERF_EVENTS needs to be added whenever
uprobes are enabled.

[YOCTO #3111]

(From OE-Core rev: b681b74624d1c8c4c98b2a121828e010fc5c3a25)

Signed-off-by: Tom Zanussi <tom.zanussi@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>

bitbake.conf: Assign SRCPV so that it will be tracked correcting in the sstate checksum

Currently, SRCPV is just listed as having a value of ${@bb.fetch2.get_srcrev(d)}
which isn't helpful. This can mean that if PV changes, two recipes can have the
same sstate checksum despite having different PV values since the PV value itself
isn't tracked anywhere.

Adding this line means that the real PV value is expanded and recorded in the sstate
checksum, meaning the sstate packages no longer overlap. This is critical in ensuring
consistent builds for revipes using SRCPV.

(From OE-Core rev: a9fffadec4fb60547257cb3d7496b6e39ed07be8)

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

opkg: Fix package dependency issue for preinsts

When processing dependencies, we need to look for both the SW_INSTALL and
SW_UNKNOWN states. If we don't do this, dependencies can be missed
and preinst scripts can run before dependencies are all installed.

This leads to package installation errors for packages like dbus-1
and associated user permission errors.

(From OE-Core rev: 119ef2789484222b94559675a09adc399f3b6bf0)

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>

taglib: Update to v1.8

Patches not needed anymore - they switched to cmake.
LGPL license was replaced with the actual LGPL 2.1 file.
License section in audioproperties.h file was modified as it includes the
new address of Free Software Foundation.
libtag static library is not built by default anymore and if cmake is
instructed to build static library than shared library is deactivated.
So actually this is a switch now.

(From OE-Core rev: 312efe73dad8a9baf32578bd11a1654219d759df)

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

binutils-2.22: Backport PR fixes from 2.22 branch

These are fixes mainly cherrypicks for mips/ppc/x86
mainly fixing PRs in ld and gold

(From OE-Core rev: f098cfc24bae8e0685bcae53ea4fdc3326ddc6c4)

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

linux-yocto: virtio and KVM guest configuration

Updating the meta branch SRCREV to pick up virtio and kvm guest
configuration fragments.

  79947f1 meta: add paravirtualized KVM guest config fragment
  3ed86ed meta: add MMIO support in virtio config fragment

(From OE-Core rev: b6b5b501fbe7158f190e887c3edc1214bb3671ed)

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: fix kernel configuration audit for custom yocto kernels

It was reported that the kernel configuration checks for custom yocto
kernels had the following output:

  NOTE: validating kernel configuration
  grep: /meta-series: No such file or directory
  grep: /meta-series: No such file or directory
  WARNING: Can't find any BSP hardware or required configuration fragments.
  WARNING: Looked at //cfg///hdw_frags.txt and //cfg///required_frags.txt in directory: //cfg//
  NOTE: Tasks Summary: Attempted 375 tasks of which 367 didn't need to be rerun and all succeeded.

which is not inspire confidence in the output of the process.

Completely inhibiting the check is one option to remove the messages,
but that removes the ability see output, which can help move users to
a better or more fully configured linux-yocto based kernel.

To fix this, we have to ensure that the path to the meta-series is
always valid, and that the tools can deal with not all files existing
in the audit directory.

Since custom yocto kernels do not set KMETA (they don't have a meta branch),
we ensure that a default of 'meta' is passed to the audit ('meta' is always
valid), and that kconf_check itself can deal with an incomplete set of
input audit files.

The net result is output like this (using a defconfig with invalid options
for the kernel being built):

  NOTE: validating kernel configuration
  This BSP sets 19 invalid/obsolete kernel options.
  These config options are not offered anywhere within this kernel.
  The full list can be found in your kernel src dir at:
  meta/cfg/standard/qemux86/invalid.cfg

  There were 1 instances of config fragment errors.
  The full list can be found in your kernel src dir at:
  meta/cfg/standard/qemux86/fragment_errors.txt

  The full list can be found in your kernel src dir at:
  meta/cfg/standard/qemux86/missing_required.cfg

(From OE-Core rev: 4d1b7dae063ee4c35c426306d0e22f11ce112c72)

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>

pseudo_1.4.1.bb: update to pseudo 1.4.1, fixing 32-bit host problems

There were a number of cases where pseudo used plain old stat()
to get dev/inode data for files; on 32-bit hosts, this could fail
if the files were over 2GB, causing pseudo to prevent removing of
large files. This is fixed in 1.4.1.

(From OE-Core rev: 056eddc4299d10ddafe0da3dc768bd80d0c1b96b)

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>

scripts/combo-layer: ensure we validate branch/revision on init

If both branch and last_revision are specified for a component when
combo-layer init is run, ensure that the specified revision is actually
on the specified branch and error out if not. Also ensure that the error
message mentions the component.

(From OE-Core rev: e498257ecbec94cec181d73bda57d44335b4dee0)

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

scripts/combo-layer: use last_revision if specified in init

If last_revision is specified for a component when running combo-layer
init, then use that revision instead of the latest revision on the
branch. Also, remove unnecessary git checkout during init since we
specify the revision to all calls to git when dealing with the component
repositories.

Fixes [YOCTO #3040].

(From OE-Core rev: ff8277cd133e9a02b131977078cff61fa587a1af)

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

libffi upgrade to 3.0.11

Changes :

- Added ax_append_flags.m4 and ax_check_compile_flag.m4 to the m4 directory.
  The files were missing and aclocal.m4 was generated without those two macros.

- Added a new license md5 checksum to the recipe because the old LICENSE file
  differs from the new one here :

        OLD : libffi - Copyright (c) 1996-2011

        NEW : libffi - Copyright (c) 1996-2012

(From OE-Core rev: 3e40136e8bd13b17b6d88b6acfb5ed162bb8d96a)

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>