gcc-configure-common: Address problems with gengtype

The gengtype patch we apply to gcc aims to ensure that the build and host
config headers don't get confused. We're seeing build failures where
both headers have been included, likely due to a race over the configuration
files.

It seems the gengtype-lex.c file isn't being regenerated when it should
and the unconditional inclusion of bconfig.h is resulting in these issues.

The fix is therefore to remove the file, forcing its regeneration.

[YOCTO #6393]

(From OE-Core rev: dd649374b30eb2d9980dce6eae95db0563593ef7)

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

kernel-module-split: Add support for KERNEL_MODULE_AUTOLOAD and KERNEL_MODULE_PROBECONF

The current module_autoload_* and module_conf_* variables are error
both ugly and error prone. They aren't registered in the task checksums
so changes to them aren't reflected in the build. This turns out to
be near impossible to fix with the current variable format in any
sensible way :(.

This patch replace module_autoload with the list of variables in
KERNEL_MODULE_AUTOLOAD which is a much simpler and usable API. An
error is printed if an old style variable is encountered. It should
be simple to convert to this.

module_conf_* are harder to deal with since there is data associated
with it, it isn't simply a flag. We need a list of variables that are set
in order to be able to correctly handle the task checksum so we add
KERNEL_MODULE_PROBECONF for this purpose and error if the user hasn't
added a module to it when they should have.

[YOCTO #5786]

(From OE-Core rev: 6f8b5be646be0f3e15e215907547f11d2a23d81b)

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

binconfig-disabled: Add class and use

This adds a binconfig-disabled class which can be used by recipes where
a -config file is installed but we wish to disable it and just rely on
the .pc files instead.

Rather than simply deleting it, we make the script "exit 1" so that it
can be found in PATH and raise a build error rather than something
silently falling back to the build system for example.

Rather than randomly finding -config files, this adds in the
specification of a list of binconfig scripts which is more deterministic
and maintainable moving forward.

This patch converts various users in OE-Core to use this, a world build
of OE-Core tests out ok with this change. There will likely be issues in
other layers however, hence this being a RFT.

(From OE-Core rev: 5870bd272b0b077d0826fb900b251884c1c05061)

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

alsa-tools: Add missing pkgconfg dependency

(From OE-Core rev: 12bab2d828836c8926f753caff80b61dbe6390a5)

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

neard: fix more parallel build issues

One fix, another appears...  root-cause to a magic dependency hiding at the
bottom of the Makefile, and add all binaries to it.

[ YOCTO #6416 ]

(From OE-Core rev: 4625eb0793ff59e4414017df0371ee9b89f47b38)

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

native.bbclass: Properly define directories

For most users this commit will have no effect. But if you come across the idea of giving
different names for paths, you'll get some troubles.

When a recipe inherit native, properly define bindir, sbindir, includedir, sysconfdir, datadir
(using xxxdir_native definitions from meta/conf/bitbake.conf).

For example, edit "${BASE_WORKDIR}/${MULTIMACH_TARGET_SYS}/quilt-native/temp/log.do_configure"
and see what are the arguments given by oe_runconf.

Notice that ${docdir}, ${mandir}, ${infodir}, ${localstatedir} have no associated _native definition.

(From OE-Core rev: 15345ddd4be6a0b041b3d6caaad48d46b22142e9)

Signed-off-by: Matthieu Crapet <Matthieu.Crapet@ingenico.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake.conf/qemu: Move QEMU_OPTIONS to qemu.bbclass

The QEMU_OPTIONS variables belong in qemu.bbclass so move them there. The
only users of them inherit qemu.bbclass. There is no point in pushing
these into every recipe.

(From OE-Core rev: 5824293de37919e89f60192836997281933e23d6)

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

qemuwrapper-cross: Use QEMU_OPTIONS

The correct cpu options are needed in order to correctly run some CPU
types. This information is available in QEMU_OPTIONS, use it. This
avoids architectures like qemuppc failing postinstalls.

(From OE-Core rev: 3d9c0cbf1f9b9802d7374c4fa1672c26fc5db5cb)

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

bitbake.conf: Add QEMU_OPTION for ppc7400 as used by qemuppc

Currently, qemuppc prints warnings about gdk-pixbuf postinstalls
not working due to illegal instructions. This is due to qemu
running with the wrong cpu type. Add an option for ppc7400 so
that qemuppc works correctly.

(From OE-Core rev: 5995fdbe81799f1ecf5de722cb2eb95ccb2aa860)

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

git: don't hardcode the full path to perl

Hardcoded paths to perl don't make sense, use from the environment instead.

[Patch taken from meta-mentor by RP]

(From OE-Core rev: 8072f26f7304ff5367d5be357037644cb1f6241e)

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

buildstats-summary.bbclass: Import useful bbclass from meta-mentor

This class summarizes sstate reuse at the end of the build, so you know how
much of your build was done from scratch.

(From OE-Core rev: 0069c06cc9c929de7e7d29b0381fcb36049a4401)

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

git: revert to wrapper based relocation

It seems that there are multiple problems with the upstream RUNTIME_PREFIX
mechanism at this time. It doesn't canonicalize argv[0] to an absolute path,
breaking calls via the PATH, for example. In addition, it doesn't seem to
locate template_dir via the runtime prefix even when specified as relative.

Revert this for now to the previous wrapper-based mechanism, but tweaked
slightly to avoid hardcoding the sysroot path into the wrapper (based on the
bits in the rpm recipe).

[YOCTO #6211]

[Pulled from meta-mentor by RP]

(From OE-Core rev: 85ce11e7b5402cc443adb8007c0e5d01f914fa74)

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

busybox: handle syslog related files properly

If CONFIG_SYSLOGD is not enabled, then the related service file should
not be installed as it will always fail at system start-up. The error
message is as following.

[FAILED] Failed to start System Logging Service.

The same logic applies to CONFIG_KLOGD.

So we should first check the configuration before we install these
service files.

[YOCTO #5302]

(From OE-Core rev: b44e291a87539fbb8e6da1a16c56f425a417e7bd)

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>

curl: control ipv6 support based on DISTRO feature

By default ipv6 is auto detected for native builds but disabled when
cross compiling.

This commit adds a PACKAGECONFIG option, unconditionally enabled for
native and nativesdk builds and controlled by the ipv6 DISTRO feature
for target builds.

(From OE-Core rev: f8377e96b353f8cf4a5812fa14c1c0405f769096)

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

gstreamer1.0-rtsp-server: depends on libcgroup and gstreamer1.0-plugins-base

Otherwise may do_compile error:
test-cgroups.c:43:23:: fatal error: libcgroup.h: No such file or directory

and:

configure: No package 'gstreamer-plugins-base-1.0' found

(From OE-Core rev: bef49e8e8bf0a7f900f4ad44c2cbb6dec1d5d180)

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>

stat: use update-alt for stat

Three different recipes provide the stat program, busybox, coreutils and stat.
Ensure that they are installed to the same place and use update-alternative
with the correct priorities to have the correct binary installed.

[YOCTO #6415]

(From OE-Core rev: 2abc776393f8b5574dd9cf614ff1ae4b460e4d8c)

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

e2fsprogs: use update-alt for chattr

Both busybox and e2fsprogs provide chattr, ensure that they are delivered
to the same location and use update-alternatives to ensure the correct
links are there.

[YOCTO #6407]

(From OE-Core rev: 23f1dddbf9cf783d90040b67978d1291b16a13de)

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

texinfo-dummy-native: Fix dummy scripts failing when paths contain capital Es

This was an issue in the command-line argument parsing. It was the cause of
the bug reported on the OE-core mailing list by Denys Dmytriyenko.

(From OE-Core rev: cbdf390806cf9fb7c0b0141a54abde372514b1cb)

Signed-off-by: Max Eliaser <max.eliaser@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

default-distrovars.inc: Remove unused variable 'LIMIT_BUILT_LOCALES'

This variable is a left-over from early glibc recipes.
It is not referenced anymore in oe-core, and its functionality
has been replaced with 'GLIBC_GENERATE_LOCALES'.

Also remove the reference to it in local.conf.sample.extended.

(From OE-Core rev: d668245991d1369e2906d1605c749c62274c0620)

Signed-off-by: Jacob Kroon <jacob.kroon@mikrodidakt.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

dbus: fixed mode if systemd service file

systemd complains with

|  Configuration file /lib/systemd/system/dbus.socket is marked executable. Please remove executable permission bits. Proceeding anyway.
|  Configuration file /lib/systemd/system/dbus.service is marked executable. Please remove executable permission bits. Proceeding anyway.

else.

(From OE-Core rev: 67e966c2576b360ee132989bee32d3080305a099)

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

dbus: disable usage of dbus-glib explicitly

on rebuilds, dbus-native can detect an already existing dbus-glib and
tries to link some tests against it.  As files were unstaged and due
to libtool insanities this fails with:

| /usr/bin/grep: .../sysroots/x86_64-oe-linux/usr/lib/libdbus-1.la: No such file or directory
| sed: can't read .../sysroots/x86_64-oe-linux/usr/lib/libdbus-1.la: No such file or directory
| x86_64-oe-linux-libtool: link: `.../sysroots/x86_64-oe-linux/usr/lib/libdbus-1.la' is not a valid libtool archive
| make[3]: *** [libdbus-testutils.la] Error 1
| make[3]: Leaving directory `.../work/x86_64-oe-linux/dbus-native/1.8.2-r0/build/test'

Make builds predicatable (and working) by disabling usage of dbus-glib
explicitly.

(From OE-Core rev: 0eeeb40d2af99448c3b94047f26d33a9983a221b)

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

bluez5: Force obex.service in /usr/lib

Upstream bluez installs the obex.service file into /usr/lib
regardless of the multilib settings as does the current systemd
recipe.  Make sure it gets packaged properly.

(From OE-Core rev: 8b66eccc05a62f1e2267c1bdf1086d8328bb962c)

Signed-off-by: Drew Moseley <drew_moseley@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

elfutils: Fix debugedit failure in grub

We've seeing occasional debugedit failures in grub during do_package
which in turn are coming from section alignment failures from libelf.
The failures occur when gold is used to link grub instead of ld.bfd.

"readelf -e uhci.module" shows:

Section Headers:
  [Nr] Name              Type            Addr     Off    Size   ES Flg Lk Inf Al
  [12] .note.GNU-stack   PROGBITS        00000000 0010ce 000000 00      0   0  1

in a good build and:

Section Headers:
  [Nr] Name              Type            Addr     Off    Size   ES Flg Lk Inf Al
  [24] .note.GNU-stack   PROGBITS        00000000 009180 000000 00      0   0  0

in a bad build. The problem is the "Al" (alignment) change from 1 to 0.
If its 0, debugedit complains.

As far as I can tell, the alignment of a zero length section is not
an issue and the check in libelf should check the section size and only
give alignment errors if there is some data to align.

(From OE-Core rev: b34447fa5223b4e0be49594aaf0254defd69bbd1)

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

git: fixed typo in 'libexecdir' varname

variable is called 'libexecdir', not 'libxecdir'.

(From OE-Core rev: 92fc79690374aceaa3b821013cfe25604b1db18a)

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

scripts/test-remote-image: Add script for running runtime tests on remotely built images

YB: #6254

Adding a new script that will fetch image files from a remote images repository.
These images will then be used for local runtime testing.

Use the '-h' option for more details on usage.

(From OE-Core rev: 5ebe9c57efb9715d58691d7fa75ccf06fb5d4b18)

Signed-off-by: Corneliu Stoicescu <corneliux.stoicescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

controllers/masterimage.py: Make testimage kernel naming pattern universal

YB: #6254

The QemuTargetControl class does not specify any kernel naming but the runqemu script it uses uses the " KERNEL_IMAGETYPE + MACHINE + '.bin' " naming scheme.
Modifying the other major target controller class, MasterImageHardwareTarget, to use the same kernel naming scheme.

This is usefull also to outside scripts that want to anticipate the kernel file name for all target controllers.

(From OE-Core rev: e8666e91a9633da6a560d5a9510bb53d0251b16d)

Signed-off-by: Corneliu Stoicescu <corneliux.stoicescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/controllers/testtargetloader.py: add 'import os'

YB: #6254

This module uses os but relies on other modules to import it. Adding 'import os' in order to be self-sustained.

(From OE-Core rev: 26e4d5212ec5b2bcfdb0f42bbed31f468a17aca4)

Signed-off-by: Corneliu Stoicescu <corneliux.stoicescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

targetcontrol.py: Add a classmethod to get extra files needed by the target controllers

YB: #6254

Add a new classmethod that can be used by outside scripts to get the extra files needed by the target controllers.
An outside script can predict rootfs, manifest and kernel files needed by a target controller, but sometimes there are other files needed.

(From OE-Core rev: fea627022473cfb73299d0988628962ad8e80f89)

Signed-off-by: Corneliu Stoicescu <corneliux.stoicescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

core-image-testmaster.bb: add bzip2 to image

YB: #6375

Some test image fstypes require bzip2 decompression.

(From OE-Core rev: 7b81fce3a80f7a1b94015a1da95d2349dac577ab)

Signed-off-by: Corneliu Stoicescu <corneliux.stoicescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

masterimage.py: enable dynamical determination of rootfs type

YB: #6375

Adding support for get_image_fstype() in the MasterImageHardwareTarget and GummibootTarget classes.

(From OE-Core rev: 06ed50a2b8c311e56ac9f1c6f2145bc020d5d500)

Signed-off-by: Corneliu Stoicescu <corneliux.stoicescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

targetcontrol.py: make possible dynamical determination of rootfs type

YB: #6375

Added a new method get_image_fstype() that autodetermines what fstype to use for the rootfs file.
This method uses a new list variable 'supported_image_fstypes' that contains image fstypes supported by the target controller.

This method is also a classmethod which means outside scripts can get the image fstype.

(From OE-Core rev: 39d5aa5c9f2916700f81d15adc220a30c6b120d1)

Signed-off-by: Corneliu Stoicescu <corneliux.stoicescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kexec-tools: Refine kdump device_tree sort

The commit b02d735bf was to rearrange the device-tree entries, and
assumed that these entries are sorted in the ascending order. but
acctually when I was validating kexec and kdump, the order of
serial node still is changed. So the patch is to sort these entries
by the directory name in ascending order.

(From OE-Core rev: c8722b510f779cd20757477a7f7a7a2a35b9a9c5)

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

oprofile: Determine the root home directory dynamically

This commit detects the root home directory dynamically with changes to
the opcontrol script and the oprofile gui app source.

The commit replaces an earlier fix that detected and adjusted a
'non-standard' root home directory at build time.  The advantage of this
patch is that the oprofile tools are adjusted to the current run-time
path to ~root, not the build time path.

(From OE-Core rev: 4fa57b42aa7ec3f77606ab3d3b5814f638c859d5)

Signed-off-by: Dave Lerner <dave.lerner@windriver.com>
Signed-off-by: Chong Lu <Chong.Lu@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

perf: split packging

* some fundamental perf commands can work
  without the dependency on perl, python or bash
  make them separate packages and RSUGGEST them

* bump PR

The patch was sponsored by sysmocom

(From OE-Core rev: 6d1c0cde05dbab5ca84b1bbd8abeecf8df49e37b)

Signed-off-by: Henning Heinold <henning@itconsulting-heinold.de>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

perf: fix broken shell comparsion in do_install

The patch was sponsored by sysmocom

(From OE-Core rev: 17bf25adad715754de589d2ae54f97c07f5dafde)

Signed-off-by: Henning Heinold <henning@itconsulting-heinold.de>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nfs-utils: Do not pass CFLAGS to gcc while building

Do not pass CFLAGS/LDFLAGS to gcc while building, The needed flags has
been passed by xxx_CFLAGS=$(CFLAGS_FOR_BUILD).

(From OE-Core rev: e4b01d651cf94185cfb285a7b64292b26b74c6bb)

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

sed: enable ptest support

Install sed test suite and run it as ptest.

(From OE-Core rev: de8cc3192eefafa639a9f34fcc35799be3286191)

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

v4 cups: Shouldn't link against libgcrypt without using gcrypt functions

Backported from http://www.cups.org/strfiles.php/3308/cups-no-gcrypt.patch

This addresses the cryto dependency seen during build.

(From OE-Core rev: e5f9166302baad837244e5a93bedb2797ab17e57)

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

texinfo.bbclass: native/cross uses dummy texinfo; target uses host's Texinfo.

To unpack that to more than a single line: -native and -cross recipes are made
to use the dummy Texinfo utilities provided by texinfo-dummy-native if they
invoke those utilities at build time. The target-architecture (cross-compiled)
recipes still use the genuine Texinfo utilites. Right now, they still use
the host system's Texinfo utilities, but could be made to use the
texinfo-native recipe we already ship with some config file changes.

(From OE-Core rev: 160087f754eabf5da90fb51997e19d2e585aac4a)

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

texinfo-dummy-native: Create recipe w/ scripts to stand in for Texinfo utils.

More work toward eliminating the dependency on the host system's Texinfo--
Python scripts that understand the same command-line options as the Texinfo
utilities, and create blank output files if appropriate, but don't actually
do any of the work done by those utilities.

This will be necessary to avoid circular dependencies when we start explicitly
tracking dependencies on texinfo-native; i.e. texinfo-native ->
autoconf-native -> texinfo-native. If we have all native recipes that inherit
texinfo.bbclass depend on texinfo-dummy-native instead of texinfo-native, the
cycle is broken.

It may also provide a performance gain by skipping the actual work of
formatting and generating documentation files.

(From OE-Core rev: 0d8d5a2d44988d32a5c8b995202a12ac106ba93c)

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

texinfo: Use texinfo-replacement-native as an alias for texinfo-native.

If texinfo-native is added to ASSUME_PROVIDED, this will be needed for two
reasons. First, a world build will still build the native texinfo recipe, so
it will still get test coverage and will not rot. Second, when the target
architecture texinfo recipe is built, the native recipe must be built first
so its makedoc binary can be extracted, ASSUME_PROVIDED or no.

(From OE-Core rev: af5cd3ad123f45170d1ba06a8f5c9ee4ce5c082e)

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

rootfs.py: change the logic in _uninstall_unneeded

Previously, if we have 'package-management' and 'read-only-rootfs'
both in IMAGE_FEATRUES, we would meet the following error at system
start-up.

rm: can't remove '/etc/rcS.d/S99run-postinsts': Read-only file system

However, what's really expected is that when there's no postinstall
script at system start-up, the /etc/rcS.d/S99run-postinsts should not
even be there.

Whether or not to remove the init script symlinks to run-postinsts should
not depend on whether we have 'package-management' in IMAGE_FEATURES; rather,
it should only depend on whether we have any postinstall script left to run
at system start-up.

This patch changes the _uninstall_unneeded function based on the logic
stated above.

[YOCTO #6257]

(From OE-Core rev: 6c2f7ecee754ff3f29fdde17c0363f5d138057ff)

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: toaster: build control functionality

We add the build control functionality to toaster.

* The bldcontrol application gains bbcontroller classes
that know how to manage a localhost build environment.

* The toaster UI now detects it is running under build
environment controller, and update the build controller
database and will shut down the bitbake server once
the build is complete.

* The toaster script can now run in standalone mode,
launching the build controller and the web interface instead
of just monitoring the build, as in the interactive mode.

* A fixture with the default build controller entry for
localhost is provided.

[YOCTO #5490]
[YOCTO #5491]
[YOCTO #5492]
[YOCTO #5493]
[YOCTO #5494]
[YOCTO #5537]

(Bitbake rev: 10988bd77c8c7cefad3b88744bc5d8a7e3c1f4cf)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toastergui: fix built ETA calculation

We fix the calculation showing the estimated time
to build completion.

(Bitbake rev: dc1daae51ef5201475b5c1a69d966ae57b66dcb6)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: xmlrpc: add support for token reusing

We add support to specify a connection token in the command line
and in the environment variable BBTOKEN.

When a client registers to a bitbake server, that client
will have exclusive access to the server. The client is identified
by a server-supplied token. If a client terminates, we cannot
reconnect to the server as the token is lost.

This patch adds the capability to specify the connection token
in the command line for xmlrpc clients. This allows us
to have bitbake work as an already-authenticated
client with the server and resume sending commands to a server.

(Bitbake rev: db5390940c0afbcdc9fbcf1225761968ae51d4a7)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: add function to get the database url

We add a function that returns the DATABASE_URL
for the current 'default' database settings. This
is useful to be able to start other toaster instances
with the same database settigns as the currently running
instance.

(Bitbake rev: 272a4bba0804bb6b5e0d498d3453321b5ed1dc76)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: create models for bldcontrol and enable it

We create the model classes that store information
about triggering builds, and the available build
environments.

We add a fixture with a default build environment
for build control, using a "build/" directory under
the poky checkout directory.

We enable the bldcontrol in toaster starting script
and in the toaster settings as to allow the actual database
to be kept in sync with the source code.

(Bitbake rev: d4bfe9059f765f11244b97e324c0131f32f8e400)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: initial bldcontrol application

The build features of Toaster will be contained into
a separate application, as to modularize the Toaster
project and provide multiple options for deployment.

This patch adds the application as a barebone Django
application in the Toaster project.

(Bitbake rev: 08556b79b7b2af08aaeedf8733b1b8996f387c4e)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: bb/utils: fix contains_any()

"set1 in set2" doesn't do what you'd expect, and if it did do a subset test
that's not the logic required by contains_any().

Instead get the intersection of checkvalues and val and check if the resulting
set is empty or not (by evaluating it in a boolean context), which tells us if
there are any elements in common.

Based on a patch by Valentin Popa <valentin.popa@intel.com>.

(Bitbake rev: 2e742c03e8dfdfa67899e7f5d579ed14bd87e139)

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

bitbake: bb/tests/data.py: add tests for contains() and contains_any()

contains_any is buggy so write a test case to demonstrate this.

(Bitbake rev: 25d10c349aab77bf99745c0a90dd8f9b90abccac)

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

bitbake: cache.py: print debug info when EXCLUDE_FROM_WORLD

This gives us an easy way to find out which recipes have been excluded
from world when there are many layers.

(Bitbake rev: cf33d4bfc53ae8dc6353fa295e1acd789d4fc2d2)

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

rpm: Fix cpio 32 bit overflow issues on 64 bit inode filesystems

When building on XFS filesystems, the resulting rpms can be corrupted
with the same inode number being used for multiple hardlinked files.
There are two fixes, one to stop rpm crashing when accessing a broken
binary rpm, the other to stop generating them in the first places. Full
descriptions in the patch headers.

(From OE-Core rev: d20d3476157b7c949b0077cad0ab1e8716d6162a)

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

bitbake: toaster: add Project class models to orm

This patch adds the Project class models to the
toaster GUI data model. There is no specified use,
and the link in the Build class is optional.

(Bitbake rev: 38f77a43d16e202d1f40632b27cb70dd20d02a5e)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: comment out bldcontrol migration command

This fixes toaster startup failing after the preparatory patches
for bldcontrol application have been merged.

I slipped a bldcontrol-specific line into the toaster startup
script and this prevents toaster from starting. We comment
out this line until the bldcontrol application have been merged.

(Bitbake rev: d095eec6fc958d0aeb514cdc206734617fd7c930)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Revert "fontconfig: enable 64 bit file support"

Fontconfig doesn't need to be passed large file flags through CPPFLAGS because
it's configure script uses AC_SYS_LARGEFILE.

This reverts commit 5a49a24d35794fcf9e3895469a67db992a52305e.

(From OE-Core rev: 84a4326473ce448c9bdd9a8b79383c319f04d339)

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

update-rc.d: Allow to use different initscripts provider

* until now all recipes were respecting VIRTUAL-RUNTIME_initscripts
  variable but commit bba835fed88c3bd5bb5bd58962034aef57c408d8
  hardcoded "initscripts" runtime dependency

(From OE-Core rev: 8b4256758ba55dcabe80dd1bf3884cdf1cc39909)

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

nfs-utils: unset CFLAGS for testlk build

testlk is built with host gcc at do_compile stage, which leads to unrecognized
some flags for special architecture. So unset CFLAGS for testlk to make sure
it passed.

(From OE-Core rev: 2461336b9432cf8379bda19d425ba00e542fbab7)

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

oe-core/init-install.sh: do not overwrite /etc/mtab if the link already exist

Overwriting of /etc/mtab would fail as below if the /etc/mtab link already
exist during installation phase, this patch fix this problem by checking
existance of the link before try to overwrite it.

Error message during installation if the /etc/mtab exists:
"cat: /proc/mounts: input file is output file

(From OE-Core rev: 26a5121e966f465386da4ead40cc558fd877ce2b)

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

oprofile: Add PTEST regression testing to oprofile

The commit changes the oe-core oprofile recipe by adding the PTEST
interface, implementing do_compile_ptest() and do_install_ptest().

The install routine adds several files that only the regression tests
require. Also, a patch is necessary to inhibit executing the tests on
the build host and to point to the correct target directory for the
libutil++ test 'file-manip-tests'.

That test requires the recipe to set SRCDIR in the build environment
instead of determining the value from build-time host paths as for a
self-hosted build.

(From OE-Core rev: ac5a9835075a04726a64dbd669b9c89270b23865)

Signed-off-by: Dave Lerner <dave.lerner@windriver.com>
Signed-off-by: Chong Lu <Chong.Lu@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gawk: remove dgawk and pgawk packages, merge gawk-common into gawk

The dgawk and pgawk commands no longer exist in gawk 4.1 onwards, replaced by
options to gawk.

Remove the dgawk and pgawk binary packages, and merge gawk-common into gawk as
it is needless splitting.

(From OE-Core rev: 20f749de439aa962f32a3f4f5977fd44c9e76e33)

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>

gawk: fix module paths

(From OE-Core rev: 64b62f6a03fa6405f309d798880dbfe3a3c18f84)

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>

libgcrypt: fix do_configure failed

Fixed:
rm: cannot remove `/path/libgcrypt-1.6.1/m4/*.m4': No such file or directory

We would meet this error if we stop the configure and run again.

(From OE-Core rev: 2d13380c6546bb41afa53227f1571bd2908ceca2)

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>

python-gst: link python shared library to config directory

python-gst needs python shared library in python config directory.

(From OE-Core rev: 7a3b7d70a0cc4cdef81bb63fdac7de8f1309d1fc)

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

xorg-font-common.inc: remove shell debug option

'set -x' is added to package postinstall scripts. It causes trace
information are printed when image first time boots such as core-image-sato.

Remove it.

(From OE-Core rev: 38ac187881266e6349084ee983046e311fb293e9)

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

oeqa/utils/commands.py: add support for postconfig option

Adding support for postconfig option to the bitbake() and related methods.

This enables us to use 'bitbake -R postconfig_file <command>'.

Usage: bitbake(cmd, postconfig="some confguration")

'postconfig_file' would contain what we add in 'postconfig'

Other methods affected: get_bb_env(), get_bb_var()

(From OE-Core rev: 4fe771940a8f59a0d5f1541978d6d9ff73b222f4)

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

gawk: fix libexec of the old 3.1.5 version

${libexec} doesn't necessarily equals to ${libdir}/${BPN}. So fix
this problem by using ${libdir}/${BPN} in FILES variable.

(From OE-Core rev: 6df0e8b99fda8ef40862f2e92d85bc3cd371615f)

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>

libgcrypt: workaround ICE failure on mips with '-O -g'

Hit a ICE and could reduce it to the following minimal example:

1. Only the size of array assigned with 2 caused the issue:
$ cat > mipgcc-test.c << END

int main (int argc, char **argv)
{
        char *pStrArry[ARRAY_SIZE_MAX] = {"hello"};
        int i = 0;

        while(pStrArry[i] && i<ARRAY_SIZE_MAX)
        {
                printf("%s\n", pStrArry[i]);
                i++;
        }

        return 0;
}

END

2. Only -O1 and -g on mips caused the issue:
$ mips-poky-linux-gcc -O1 -g -o mipgcc-test mipgcc-test.c
mipgcc-test.c: In function 'main':
mipgcc-test.c:18:1: internal compiler error: in dwarf2out_var_location, at dwarf2out.c:20810
 }
 ^
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions

[YOCTO #6034]

3. The quick workround is trying to enlarge the size of array with larger
than 2.

4. File a bug to GNU, but it could not be reproduced on there environment.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60643

(From OE-Core rev: 4af0c70c70809c8f03d7ba14745d79e3c6e35b2e)

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

xorg-xserver: disable DRI3 by default

Add a PACKAGECONFIG for DRI3 but don't enable it yet.

Based on a patch by Valentin Popa <valentin.popa@intel.com>.

[YOCTO #6402]

(From OE-Core rev: 898d60f1901745b861ecab91b3ceceb9a1cfc8b1)

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

mesa: upgrade to 10.1.3

libdricore was removed and replaced with a megadriver with hardlinks from the
old driver names.

Add PACKAGECONFIG for DRI3 but disable by default as there are currently
regressions compared to DRI2.

License is unchanged.

(From OE-Core rev: 36c6909d4410e5ecb44800aafadd8593978e1b24)

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

mesa: add build-dependency on libxml2-native

glapi/gen needs the libxml2 Python module, but this isn't listed as a build
dependency.

(From OE-Core rev: 12414a65bfa9222be0fe0ab7ced2a02e1e5ed2ed)

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

glib-2.0: fix paths to test binaries

The tests are installed to $libexecdir not $libdir/glib-2.0.  By default these
are the same location but they can be changed independently.

(From OE-Core rev: 73d774fa7456842690af95206b77162d5f1bcb37)

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

gdk-pixbuf: fix paths to test binaries

The tests are installed to $libexecdir not $libdir/gdk-pixbuf.  By default these
are the same location but they can be changed independently.

(From OE-Core rev: 56dd1a61f52cdf66fdb3e9510c70b02224475d4a)

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

openssl: fix ptests

Add some missing dependencies and fix the Makefile in order to get most
of the ptest tests working (specifically test_bn, test_verify, test_cms,
test_srp and test_heartbeat). test_verify still fails for unknown
reasons (perhaps some of the now expired certificates weren't meant to
have expired as far as the test is concerned?) but at least it has the
certificates to run now.

(From OE-Core rev: c679ec81c19dd2b5e366b713801785ce0ba5b49a)

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

openssl: upgrade to 1.0.1h

Fixes the following security issues:

* CVE-2014-0224
* CVE-2014-0221
* CVE-2014-0195
* CVE-2014-3470

The patch for CVE-2010-5298, CVE-2014-0198 and a fix for building the
documentation are integrated upstream in this release and so were
dropped. Additionally, a patch from upstream was added in order to
fix a failure during do_compile_ptest_base.

A similar upgrade was also submitted by Yao Xinpan <yaoxp@cn.fujitsu.com>
and Lei Maohui <leimaohui@cn.fujitsu.com>.

(From OE-Core rev: a3e80de6d423c272a287bf3538196b48ac5ddec1)

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

bitbake: bitbake: show wildcard appends for bitbake-layers

The 'bitbake-layers show-appends' command can use the built-in
get_file_appends() method which correctly identifies bbappends that
include a '%' wildcard in their filename.

(Bitbake rev: 2732dbae67c1945b668c38cc4cc5678c4aafe3d6)

Signed-off-by: Ash Charles <ashcharles@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toasterui: fix django settings environment value

Previously, the buildinfohelper only set a django settings module
environment variable if none were set.

This may lead to problems when the starting the toasterui
from an already existing Django environment.

As such, we always override the variable to provide the
correct name for the local Django settings module.

(Bitbake rev: 8271e61a2fbddd3fc49556829675478d7505d58f)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: add project related models

We introduce the notion of a project in Toaster as the item
that holds the specification for triggering a build: the set
of layers used, the set of configuration variable values, and
the set of targets to be build.

Builds triggered through Toaster will be associated with a
Project, and they will be configured based on the project
settings at the moment when the build is ordered.

(Bitbake rev: 0bf0251ae05442ae260c6099b389bf765c4fef26)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: update toaster to run in managed mode

We disable bitbake self start to prevent race condition.

ToasterUI will shutdown the server when the build is done
if running in managed mode.

We fix usage of kill server flag in the bitbake binary.

(Bitbake rev: 30159dbda3a40fa596302f91c705cb5f148c97a9)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: startup script standalone mode

The toaster starter script was designed to be sourced in
a build environment and set up the build recording environment
to be used in an interactive mode.

For the standalone web server mode, we modify the toaster
script to be run directly from the checked-out sources,
without a build environment set up, and run the web server
alone. In the standalone mode, the build environemnts and
all build activities are controled through the web interface.

(Bitbake rev: c1db4ccf27bedcbab2f03e7539fdb11b042c4fb9)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: startup script noweb mode

We add an option to the startup script, named "noweb" that
will start toaster without the embedded web server.

This is useful to start the system for build-only environments,
where the web server code is running on a different machine.

(Bitbake rev: c39838201301b9732581288a93783400bebe6591)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: read database settings from the environment

We add the capability to read the database settings for
Toaster from the environment. The DATABASE_URL is intepreted
and used to override the default settings.

This capability is essential for easy deployment of Toaster
in a managed hosted environment, and for creating
build environments with custom database settings.

(Bitbake rev: d16d19dafb83448fc214fce4fbdc2bcbf4bf9ce3)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: uievent: retry on handler registration failure

The registration of a remote UI event handler may fail
if the server cooker is currently in some certain states.
This may happen, for example, when a remote UI is started
very fast after the bitbake server is started, and the
server hadn't time to finish initial configuration parsing.

Rather than fail outright, we have the remote UI event retry
registration for five time at one-second intervals,
in the hope it will succeed.

(Bitbake rev: c3d520c92ae4ae80d31926a416456df510654b6a)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: xmlrpc: client - remove fatal errors

When we use the XMLRPC client API to connect to a bitbake server,
we want to receive errors from the API instead of having the
API exiting without warning.

Thus the "bb.fatal" calls have been replaced with "bb.warn" calls,
and we re-raise the original exception for handling by the
original caller.

The bitbake starting script has been modified to properly test
for failures in calling the client API and handle them.

Additional error handling added in the client, as to prevent
fatal crashes.

(Bitbake rev: eb63f08c33644f64752aaae2146a000956ce894a)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: bitbake: move configuration reading code

The configuration reading code should live in the
main bitbake entry point, and the server modules should
be supplied with correct configuration instead of attempting
to parse from configuration files.

This patch moves the endpoint address reading from XMLRPC
to the bitbake main script.

(Bitbake rev: ac5753274ff932e1d6f073ab4dab7bd6fe5355a1)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: check if BBLAYERS is valid before use

If wic is running as raw mode, it's better to check if BBLAYERS
is valid before inspect it.

No functional changes.

(From OE-Core rev: a3ee9cc7aebaecfa2223552a2c1865a9337de664)

Signed-off-by: João Henrique Ferreira de Freitas <joaohf@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: default plugin type directory should be added only once

Fix 'for' statement identention so plugin type directory will be
added only once in layers_dirs list.

No functional changes.

(From OE-Core rev: ba88329115a3d6f964febcbf554af8391e1b84a1)

Signed-off-by: João Henrique Ferreira de Freitas <joaohf@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gnutls: upgrade to 3.2.15

(From OE-Core rev: f82523e8afd1fc9b86ccabff01dbb781bac9b6f5)

Signed-off-by: Valentin Popa <valentin.popa@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: put udhcpc in background

If the interface is slow to come up udhcpc will continue in the background

[YOCTO #6339]

(From OE-Core rev: 45af0ba08dbc676be41fd29e9877fe820b531f7c)

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

update-rc.d/useradd: Add additional dependecies

These dependcies are needed to ensure that thier packages are created
correctly since these classes have runtime dependiences in their packages
but they are not actually created yet at rootfs time.

[YOCTO #6072]

(From OE-Core rev: bba835fed88c3bd5bb5bd58962034aef57c408d8)

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

generate-manifest-2.7.py: Add importlib

importlib isn't currently included in any of the python packages, so
create a new one for recipes that require it.

Cc: Paul Eggleton <paul.eggleton@linux.intel.com>
(From OE-Core rev: b3dae96d9fdb4e26101f6f7edc6e65989375a5a2)

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

autogen-native: upgrade from 2.18.2 to 2.18.3

Upgrade autogen-native from 2.18.2 to 2.18.3.
A patch is made to fix the compilation error.

(From OE-Core rev: 84052c30c7e4b845543c9704945170a55734343e)

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

file: upgrade from 5.17 to 5.18

Upgrade file from 5.17 to 5.18.

(From OE-Core rev: 6167e415e63366c606c8366abda7edfa21d68b58)

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

grub-2.00: remove our 40_custom

We have this in recipes-bsp/grub/grub/40_custom:

[snip]
menuentry "Linux" {
    set root=(hd0,1)
    linux /vmlinuz root=__ROOTFS__ rw __CONSOLE__ __VIDEO_MODE__ __VGA_MODE__ quiet
}
[snip]

These lines are only for initrdscripts/files/init-install.sh, the side
effect is that it would make the target's grub-mkconfig doesn't work
well since the 40_custom will be installed to /etc/grub.d/40_custom, the
grub-mkconfig will run the 40_custom, and there will always be a
'menuentry "Linux"' menu in grub.cfg no matter it is valid or not, we
can do this in init-install.sh rather than grub to fix the problem,
which is also much simpler.

We have done the related work in init-install.sh, now we need remove our own
40_custom, and use grub's own 40_custom which is the right one.

(From OE-Core rev: e5f6dc48930c8ff35df5cff3550ec2ee86641faf)

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

grub-2.00: remove our 40_custom

We have this in recipes-bsp/grub/grub/40_custom:

[snip]
menuentry "Linux" {
    set root=(hd0,1)
    linux /vmlinuz root=__ROOTFS__ rw __CONSOLE__ __VIDEO_MODE__ __VGA_MODE__ quiet
}
[snip]

These lines are only for initrdscripts/files/init-install.sh, the side
effect is that it would make the target's grub-mkconfig doesn't work
well since the 40_custom will be installed to /etc/grub.d/40_custom, the
grub-mkconfig will run the 40_custom, and there will always be a
'menuentry "Linux"' menu in grub.cfg no matter it is valid or not, we
can do this in init-install.sh rather than grub to fix the problem,
which is also much simpler.

We have done the related work in init-install.sh, now we need remove our own
40_custom, and use grub's own 40_custom which is the right one.

(From OE-Core rev: 4fad24fef5552bd0e4a5ea0982573ba8e4523570)

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

fontconfig: enable 64 bit file support

When running 32bit code on a large filessytem with 64bit inodes, the
fontcache was not being created correctly because an EOVERFLOW was being
returned from the fstat when reading the 64 inode on a 32bit system.

The fontcache is created at rootfs time on the host system via qemu.

[YOCTO #6338]

(From OE-Core rev: 9f363f9e2a56d847efc37dfba38764ae393fbfd6)

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

freetype: enable 64 bit file support

When running 32bit code on a large filessytem with 64bit inodes, the
fontcache was not being created correctly because an EOVERFLOW was being
returned from the fstat when reading the 64 inode on a 32bit system.

The fontcache is created at rootfs time on the host system via qemu.

[YOCTO #6338]

(From OE-Core rev: 65121c8ab37907938b1891b5ee8fb44fb4f76d8f)

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

autoconf: don't disable the autoheader warnings

The usual autoheader warning is due to AC_DEFINE variables not having a
description.  This results in no variable being defined in config.h, which leads
to code behaving as if the test failed when it actually succeeded.

This patch was introduced to OpenEmbedded back in 2004:

http://git.openembedded.org/openembedded/commit/?id=5eab06d132cb2895fd579f5cedffbb54c27794f8

There is no rationale for the patch so I suspect this is due to the warnings
being fatal and the submitter not understanding that the problem is more than
cosmetic.

(From OE-Core rev: de5fb9d7f60763082327ddeca71842c00a2fa23e)

(From OE-Core rev: dd9c3d7bc946ff44e0ca90f4e345711d6ad21728)

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

libxml2: fix python packaging for nativesdk

We enable the python module in nativesdk-libxml2, but the python binary
used is in the native sysroot and thus you get the module installed in
the wrong path. Even with that fixed the python files are still
unpackaged, so create an ${PN}-python package and add them to it. (This
does not affect the libxml target build at all since python is disabled
for that.)

(From OE-Core rev: e3d06aa104065748367e1479138f824da5d9951f)

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>

libarchive: avoid dependency on e2fsprogs

libarchive's configure script looks for ext2fs/ext2_fs.h in order to use
some defines for file attributes support if present (but doesn't link to
any additional libraries.) There is no configure option to disable this,
and if e2fsprogs is rebuilding between do_configure and do_compile you
can currently get a failure. Because it doesn't need anything else from
e2fsprogs, and e2fsprogs isn't currently buildable for nativesdk anyway,
copy the headers in from e2fsprogs-native which we're likely to have
built already (and add it to DEPENDS just to be sure we have.)

Fixes [YOCTO #6268].

(From OE-Core rev: ad754e46ad477acfbe7543187a5c38bc333b8612)

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>

neard: fix for parallel build

for neard tools/snep-send object might cause a
parallel build failure,due to undetected
dependency on dbus.h header file.

Patch will be submitted upstream.

Fixes [YOCTO #6389].

(From OE-Core rev: ae55abd60ef217ad1a957102d80b06857ea8ebcd)

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