oe-selftest: test building wic image by bitbake

Added test case to verify building of wic-image-minimal recipe
and produced artifacts: manifest and bzipped partitioned image.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic-image-minimal: add wic image recipe and .wks

Added example of recipe and .wks file to create partitioned image.

This image is using quite complex partitioning scheme.
It uses its own rootfs to populate two partitions in two different ways.
It also uses core-image-minimal rootfs to populate another partition.

This is how wic reports about artifacts used to create this image:
  ROOTFS_DIR: tmp/work/qemux86_64-poky-linux/wic-image-minimal/1.0-r0/rootfs
  ROOTFS_DIR["/core"]: tmp/work/qemux86_64-poky-linux/core-image-minimal/1.0-r0/rootfs
  ROOTFS_DIR["/backup"]: tmp/work/qemux86_64-poky-linux/wic-image-minimal/1.0-r0/rootfs
  BOOTIMG_DIR: tmp/sysroots/qemux86-64/usr/share
  KERNEL_DIR: tmp/deploy/images/qemux86-64
  NATIVE_SYSROOT: tmp/sysroots/x86_64-linux

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oe-selftest: test generation of <image>.env

Added test case to check if <image>.env file is generated
and contains bitbake variables used in wic code.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image.py: write bitbake variables to .env file

Write set of bitbake variables used by wic into
build/tmp/sysroots/<machine>/imagedata/<image>.env

List of variables is defined in WICVARS variable in
meta/classes/image_types.bbclass.

This is needed for wic to be able to get bitbake variables without
running 'bitbake -e'.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/utils/dump: Handle empty commandlist gracefully

If the commandlist isn't available, the code currently gives a backtrace.
At least stop doing that and return more gracefully.

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

image_types.bbclass: add wic image type

wic image type is used to produce partitioned images.

Image configuration should be stored in either <recipe>.<machine>.wks
or <recipe>.wks file.
.wks file should be put to the same location as image recipe
and have the same name.

[YOCTO #7672]

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: implement --vars option

This option is used to point wic to the directory with .env
files containing list of bitbake variables and their values.

If this option is used wic will get bitbake variables from
files instead of parsing 'bitbake -e' output.

The main reason for this is to support new mode, when bitbake
runs wic to produce wic images. In this case wic can't run bitbake
again as it's locked, so it will get variables from .env files.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: implement getting variables from .env files

Added functionality of getting variables from <image>.env files to
BitbakeVars class. env files will be parsed if the directory with
env files is known, i.e. when vars_dir attribute is set.
Otherwise 'bitbake -e' output will be parsed.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: set default image

Set BitbakeVars.default_image when wic is called with -e option.
This makes get_bitbake_var API to use provided image as a default
source of variables.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: add default_image attribute to BitbakeVars

New attribute is used when bitbake variable is requested without
specifying image name. The attribute should be set from outside,
for example when wic is called with '-e <image>' option.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: create new method _parse_line

Moved code that parses one line of 'bitbake -e' output
to separate method _parse_line.

This method will be also used later to parse lines of .env files.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: add BitbakeVars class

Moved code of getting bitbake variables into separate class.

Created singleton object of this class in the module namespace.

Preserved existing API get_bitbake_var.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto-3.14: Use a revision thats on the branch for qemuppc

At a guess, Bruce's scripts have mixed up the -rt and on -rt versions
of the qemuppc branches. Set this to a revision on the
standard/qemuppc branch.

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

linux-yocto/3.14/4.1: fix ARM boot with gcc 5.x

Booting qemuarm when the kernel was built with gcc 5.x would result in
a boot hang (or at least no visible output).

Updating the SRCREVs for the following change:

    Author: Jianchuan Wang
    Email: jianchuan.wang@windriver.com
    Subject: Omit to optimize vsprintf.c/kasprintf.c
    Date: Wed, 26 Aug 2015 13:54:57 +0800

    Add "-O0" for vsprintf.c/kasprintf.c

Signed-off-by: Jianchuan Wang <jianchuan.wang@windriver.com>
While this may not be the final fix (it still has to go usptream and
be better explained), it gets us booting, so is good enough for the
time being.

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

oeqa/runtime/multilib: add test for libc

Add a basic test to verify that /lib/libc.so.6 and /lib32/libc.so.6 have the
right ELF class.

Signed-off-by: Ross Burton <ross.burton@intel.com>

oetest: Fix regresion when testing real hardware

This fix the regresion introduced in commit
9c72c1a5aa0b49d3895bbefee7a264adfcc6f4ca
when testing with real hardware. This regression
happens when a test in real hardware fails.

[YOCTO #8203]

Signed-off-by: Mariano Lopez <mariano.lopez@linux.intel.com>

dump: allow to have datastore vars on dump commands

This allows to have datastore variables in the dump
commands and will get the data when a new instance
it's created.

Also this remove special cases from the commands.

[YOCTO #8118]

Signed-off-by: Mariano Lopez <mariano.lopez@linux.intel.com>

dump: Created new classes for dump host and target

It makes sense to separate the dump commands from the
oeRuntimeTest class, this way it can be used in all
the test context.

These are the changes included in this patch:

    - Created classes: BaseDumper, HostDumper, TargetDumper
    - Create an instance of HostDumper in imagetest.bbclass
      and add it to TestContext class, this way any class
      that have access to the TestContext would be able
      to dump logs from the host
    - Create an instance of TargetDumper in QemuTarget
      class after get the runner, this way it is
      accessible during the tests.

[YOCTO #8118]

Signed-off-by: Mariano Lopez <mariano.lopez@linux.intel.com>

classes/buildhistory: add build result to commit message

We have the command in the commit message, we might as well have the
build result as well (succeeded/failed and whether or not it was
interrupted by the user). The interrupted part relies upon a change to
BitBake to extend the BuildCompleted event to include an attribute for
that, but will not fail if the attribute is not present.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

classes/buildhistory: include metadata revisions in commit message

We do already commit these into the repository itself, but have them in
the commit message as well as a reference. As part of this, refactor out
running "git commit" into a separate function so we don't have to
duplicate the code in the two places we call it.

Implements [YOCTO #7966].

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

classes/buildhistory: handle additional files at recipe level

Avoid an error when attempting to remove previous data if it's not a
subdirectory - we were assuming that anything that wasn't named "latest"
or "latest_srcrev" had to be a directory. This makes it possible to have
a buildhistory_emit_pkghistory_append which writes additional files at
the recipe level.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

classes/buildhistory: fix permissions in package file listing

If we want the correct file permissions to show up here as they would on
the target, we need to run the command under pseudo. Normally we'd set
the fakeroot varflag on the function and that would be enough, but it
turns out that setting fakeroot on a non-task function that you execute
using bb.build.exec_func() isn't working at the moment. Work around this
by simply using FAKEROOTENV and FAKEROOTCMD. Unfortunately that means we
have to duplicate the command for the two cases but I couldn't find a
better means of doing that that actually works.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

classes/buildhistory: tweak buildhistory_list_pkg_files

* Avoid using ${...} for shell variables (since they could be
  expanded as bitbake variables if present)
* Use files-in-package.txt rather than files-in-<packagename>.txt; the
  file is already in a subdirectory named with the package name and this
  naming is consistent with that of files-in-image.txt.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

classes/buildhistory: indent recently added function consistently

Shell functions use tabs in this file.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

classes/buildhistory: exclude . in file listings

We don't care about the permissions on the top-level directory in which
the files are contained, just everything under it; this also avoids
lists with just this entry in it for empty packages. Affects file
listings for both images and packages.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

classes/buildhistory: ensure we push when "no changes" commits are made

If there aren't any changes, we still make a commit to the buildhistory
repo, but this wasn't being pushed if BUILDHISTORY_PUSH_REPO is set.
Move the push to the end to make it unconditional.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

image.py: add script output to the rootfs log

Let's add output of image creation script to the bitbake log
as it can contain useful information.

One good example of such an information is wic report about
artifacts and .wks file used for image creation.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

image.py: set bitbake variable ROOTFS_SIZE

This variable is going to be used by wic to set partition
size. Setting it in image.py makes it possible for wic to
use it without calculating it again.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

wic: deferred call of hlp.get_wic_plugins_help()

If get_wic_plugins_help is called from wic main module
it calls git_bitbake_var at some point. This fails when
wic is called from bitbake as 'bitbake -e' can't be
run.

Moved call of this method to help.py in order to call it
later, when BitbakeVariables singleton is properly initialized
to get variables from .env files.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

wic: rename variable

Renamed variable help -> hlp as 'help' is a name of Python
built-in function.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

wic: remove undescore from function name

Renamed __exec_cmd -> _exec_cmd as double underscores cause
strange behaviour when function is called in class method.
Python complains that __exec_cmd method(!!!) of the same class
doesn't exist.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

python-distribute, python-setuptools: Migrate to good old python-setuptools

setuptools the old king is ruling again, distribute fork has been merged
into setuptools starting 0.7x, so lets move to use it as well

Remove licence ambiguity we need to say which verison of PSF and Zope license is in use

Fixes [YOCTO #8201]

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

testimage: Run commands in target and host when test fails

This patch modify three files altought two of them
are minimal modifications. This version includes
the changes proposed by Paul.

testimage.bbclass:
    Create new vars for easy modification of the dump
directory and commands to be run on host and target
when a test fails
    TESTIMAGE_DUMP_DIR: Directory to save the dumps
    testimage_dump_target: Commands to run on target
    testimage_dump_host: Commands to run on host

oetest.py:
    - Allow to use the vars defined in testimage class
    - Now able to run commands in the host and dump the
      results
    - Fix an issue with the condition where to run the
      dump commands (Before it run the commands every
      test after a failure, now it runs the commands only
      in tests that failed)
    - Fix the output to stdout

[YOCTO #8118]

(From OE-Core rev: 26fe645457633f90bb5ddbb12f5f7b9ca4a06cc5)

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

byacc: fix compilation on Ubuntu 14.04

Ubuntu defaults to passing _FORTIFY_SOURCE=2 which breaks byacc as it doesn't
pass enough arguments to open():

inlined from 'open_tmpfile' at byacc-20150711/main.c:588:5:
/usr/include/x86_64-linux-gnu/bits/fcntl2.h:50:24: error: call to '__open_missing_mode' declared with attribute error:
open with O_CREAT in second argument needs 3 arguments

Add a mode of 0666 to fix this.

Signed-off-by: Ross Burton <ross.burton@intel.com>

autotools.bbclass: Allow dependency tracking option to be overridden

CONFIGUREOPT_DEPTRACK can not be overridden since it is set by =

Signed-off-by: Roy Li <rongqing.li@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

runqemu-internal: Fixes unary operator expected in new TCPSERIAL_PORTNUM

If $TCPSERIAL_PORTNUM is empty string causes an error because
expands the expresion to,

$TCPSERIAL_PORTNUM == "" -> == ""

Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

linux-yocto-dev: update to 4.2-rc

The -dev kernel is now tracking 4.2 content, so we update the
linux-yocto-dev recipes PV to match.

(From OE-Core rev: fc22f6ee38731e60b2bc15640fa697e5af663422)

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

linux-yocto/4.1: CIRRUS config and quark thermal support

Updating the meta SRCREVs for the following changes:

  92977854ff52 graphics: disable CIRRUS DRM for qemumips and qemuppc
  57f4dc4dfe16 intel-quark: Enable thermal support

(From OE-Core rev: 5c9d414fc8e88cd5ef8e7e530ba4c5e788e03bb7)

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

linux-yocto/4.1: enable DRM_CIRRUS_QEMU and EXT4_USE_FOR_EXT23

Updating the meta SRCREV to enable cirrus graphics emulation and wider
compatibility for the ext4 driver.

[YOCTO: #7348]
[YOCTO: #6667]

(From OE-Core rev: 530c51e5354d5cd233b7015a3d0dfe94cb9cbaa1)

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

linux-yocto/4.1: update to v4.1.6 and v4.1.6-rt5

Bumping the SRCREVs to import the latest 4.1-stable and 4.1-rt changes.

(From OE-Core rev: 21fd4093bbd0177453a1e749d825cf510746f201)

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

linux-yocto/3.14: cleanups and gcc5 ARM build fixes

Updating the 3.14 SRCREVs to match the latest kernel meta data updates
and also to merge four patches Richard Purdie located that fix the
gcc 5.x ARM build (we still have boot issues, but building is the
first step).

(From OE-Core rev: 24a888ddd04e44f8a069364bfbde06871ad33ae8)

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

linux-yocto/4.1: quark configuration and thermal support

Updating the 4.1 SRECREVs to reflect the following changes:

  74159a303c21 thermal: intel Quark SoC X1000 DTS thermal driver

  833214b33303 intel-quark-tiny: Add tiny KTYPE .scc file
  45ecab18d462 amd.cfg: Add X86_AMD_PLAFORM_DEVICE to default
  9a31d2cbc251 intel-quark-standard: add mass storage and boot live support
  4ea43922558a bsp: add coretemp to intel-common-drivers

(From OE-Core rev: e1aa475dd362d3545911b0677404ced644e394eb)

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

linux-yocto/3.19/4.1: add coretemp to intel-common-drivers

Updating the 3.19 and 4.1 meta SRCREVs to pull in the coretemp
configuration values.

Partial fix for [YOCTO #8107].

(From OE-Core rev: 4511961afca854d8006c0d058f46f8ba46f277c4)

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

linux-yocto/4.1: fix qemuarm preempt-rt configuration

With the recent changes to the kernel meta data handling and the
removal of BSP branches, we need to shuffle some patches around to
match.

This prevents patches from being applied twice, and the qemuarm BSP
can be configured for preempt-rt.

[YOCTO: #8122]

(From OE-Core rev: b6b7a80440521a8e82cfe6c56dddedf061de6208)

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

lib/oe/patch: handle patches that delete files with older git

When running "git add" on a directory with older git versions, deleted
files aren't added to what is to be committed unless you explicitly
specify the -A option. The result of this is that when applying a patch
from a recipe which doesn't apply with "git am" (and we fall back to
applying the patch through other means then "git add" following by
a "git commit") these deletes weren't committed with the patch, leaving
them sitting deleted but not committed at the end.

This should fix test_devtool_modify_native (which unwittingly exercises
this scenario by attempting to run "devtool modify" on apt-native) on
machines with older git versions.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

os-release: Exclude DATETIME from sstate signature

I'm torn over this, I can see someone wanting this to vary in each
build. What pushed me to believe its incorrect in its current form
is that it changes for every MACHINE build, making a complete mess
if you try and generate package feeds using it.

The alternative would be to give up on allarch in this case and make
it MACHINE specific which at least would then be more consistent when
the package makes it to a package feed. If someone wants to do that,
the can propose another patch but this change at least make package
feed usage 'sane' again and avoids the perpetual rebuilds.

Incidentally, its worth noting that changes in source metadata revision
used for the build and included in this recipe would still trigger
rebuilds which is likely the common use case people actually care about.

(From OE-Core rev: 46de6f6eb9c4387298fafb48fb9c36abc3ff48a4)

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

layer.conf: Add several allarch dependency exclusions

These are dependencies that our allarch packages have in OE-Core that cause
those allarch packages to rebuild every time MACHINE changes.

With these changes, OE-Core allarch packages all have a common sstate
signatures and no longer rebuild.

(From OE-Core rev: 63bff90fa4fb4a95e8c79f9f8e5dd90ae1dfc69d)

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

package_*.bbclass: Ensure OVERRIDES doesn't change sstate signature

Changes to OVERRIDES (e.g. from changing MACHINE) should not change
the sstate signatures of do_package_write_*. Exclude the variable
explicitly in the package classes to avoid this.

(From OE-Core rev: 5826a9260138c437f87ba1a9f84d5c08442b997d)

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

oeqa: Optimise sstate tests

We can use "none" here instead of printdiff for some small performance
gains. This also means we can remove the ignore exit code hack which
leads to safer code.

Also drop an unused variable.

(From OE-Core rev: e2b8d6fa45b1ae530be2ffabf48ea61d4c467066)

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

jpeg: update to version 9a

Previous webkit version was blocking this; latest webkit works fine.

(From OE-Core rev: 5d9d179023857c53801e24b18f9b3051d9c84145)

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

xf86-video-intel: fix missing fstat() prototype

What I'm guessing is an updated systemd exposes less system headers, so
xf86-video-intel was using fstat() without the required includes.

Backport a patch from upstream to fix this problem.

(From OE-Core rev: 3f38c912384d7398516f689bc5934d36cbfdd4d0)

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

libatomic-ops: move to recipes-support, fix recipe name

- This recipe is useful for more than just pulseaudio, so move it to
  recipes-support.
- Rename to the correct upstream name, which corresponds to the library name.
  Keep a PROVIDES of libatomics-ops for compatibility.

(From OE-Core rev: 5014de67fa6da1672626e3ec92fc51430fca3262)

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

libunwind: always depend on libatomics-ops

- Depend on libatomics-ops unconditionally for all archs, as it's used by more
  than just arm, and currently the configure script auto-detects and enables
  its use based on the existence of its header, resulting in non-deterministic
  builds.
- Drop the ldflags bits which link against libatomic_ops, because libunwind
  doesn't actually link against it, nor does it need to, as it currently uses
  all inline functions from the header file.

(From OE-Core rev: 83bb89502e2f05b8b78b14532ad96abc400ab264)

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

nativesdk.bbclass: set sbindir to bindir

The sbindir vs bindir separation really doesn't mean much in this context, and
toolchain-scripts.bbclass doesn't add sbindir to the PATH, so set the former
to the latter.

(From OE-Core rev: 51a19850700bd96e3d2e104e1626aceb73a48e72)

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

update-rc.d.bbclass: explicitly dep on initscripts

update-rc.d doesn't just want an initscripts package, it wants
initscripts-functions. In the case where VIRTUAL-RUNTIME_initscripts is set to
nothing, update-rc.d still needs initscripts-functions to satisfy its runtime
dependencies. Further, we shouldn't be using a *runtime* virtual in a *build
time* variable. This should resolve image creation failures with
configurations without sysv initscripts when a recipe inherits update-rc.d and
doesn't inherit systemd.

(From OE-Core rev: e6642ae0849e5ac3174c643a7693af25d829d6d1)

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

image.bbclass: re-run do_rootfs when feature packages change

Currently, FEATURE_PACKAGE_<feature> isn't in any vardeps, so changing the
packages for a feature won't change the checksum for do_rootfs. Rather than
explicitly adding those to vardeps, just use the expanded form of
FEATURE_INSTALL and FEATURE_INSTALL_OPTIONAL, so the actual list of packages
from the features goes into the checksum.

(From OE-Core rev: fdd1669e04bd8219344b1896b9d9c6a187e4f84e)

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

bitbake.conf: set PATCHRESOLVE to noop by default

Automatically spawning a devshell, rather than doing so on demand, is
questionable default behavior, and is potentially problematic in headless
builds. Further, there are problems with the patch resolver today. Default to
noop, and the user can always opt-in to use of the patch resolver by setting
PATCHRESOLVE to user.

(From OE-Core rev: 7964936dd1fb202373e58048c19a91d4b27cdfd6)

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

bitbake.conf: set USE_NLS based on DISTRO_FEATURES

If our libc doesn't support locales, we don't need gettext nls bits enabled.

(From OE-Core rev: f1bc8afa6ee584a81fb65bcf77e5ae1a8889f47c)

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

bitbake.conf: unexport SHELL

The user's shell shouldn't be allowed to affect the build (and it can break
the flock command and possibly more, if the user's shell isn't POSIX
compliant).

(From OE-Core rev: fc5e1cfcc3ab7acfb6e7e12cb2cf7fa4699ae7b3)

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

systemd: Cache/define tool paths for target

Recently, systemd has added a patch

f00929ad622c978f8ad83590a15a765b4beecac9

where it now pokes at the system to find
out the path of mount/umount tools, this caused the builds to fail
because it ended up with identifying these programs from native sysroot
but they were actually meant for target, this lead to boot failures due
to none of mount worked because the paths are encoded into systemd
binaries during build time.

Correct few others while here, these are not yet detected wrongly in my
build, because those binaries are not found in my native sysroot but if
some one staged the native providers of these packages they will fail too.

(From OE-Core rev: 9d87fbc87276673d958a65d476d06fec96add8e7)

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

core-image-minimal: Add 4M extra space when using systemd

qemu images fail to rebuild the hwdb due to no space letf on device

 systemd-hwdb-update.service - Rebuild Hardware Database
   Loaded: loaded (/lib/systemd/system/systemd-hwdb-update.service;
   static; vendor preset: enabled)
      Active: failed (Result: exit-code) since Sat 2015-08-15 21:39:28
      UTC; 28s ago
      Docs: man:hwdb(7)
            man:systemd-hwdb(8)
        Process: 92 ExecStart=/bin/systemd-hwdb update
(code=exited, status=1/FAILURE)
 Main PID: 92 (code=exited, status=1/FAILURE)

 Aug 15 21:39:28 qemux86 systemd-hwdb[92]:
 Failure writing database //etc/udev/hwdb.bin:
 No space left on device
 Aug 15 21:39:28 qemux86 systemd[1]:
 systemd-hwdb-update.service: Main process
 exited, code=exited, status=1/FAILURE
 Aug 15 21:39:28 qemux86 systemd[1]: Failed to
 start Rebuild Hardware Database.

(From OE-Core rev: 4086c539d47bc837e775972016e1844a3676c8e7)

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

diffstat: Add TOOLCHAIN_OPTIONS to Linker calls

configure in this case does a novel thing where it drops all additional
compiler options from CC variable into CFLAGS which in OE's case include
the options added by TOOLCHAIN_OPTIONS and then it does exactly same for
LD as well, the problem starts when we miss the --sysroot options and
ABI options which are part of CC variables usually in OE, in the end it
assigns LD = CC but doesnt add newly separated CFLAGS to LDFLAGS hence
the compile still works since all those needed options are getting into
compiler cmdline but link step fails especially when using gold, since
it does not find --sysroot option on cmdline it starts to complain about
missing libraries and paths

This shows up with clang more so because clang does not have implicit
sysroot that it silently passes down to linker like gcc does when
confgured with sysroot

We see errors like

arm-angstrom-linux-gnueabi-clang -c -I.
-I/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/work/armv7at2hf-vfp-neon-angstrom-linux-gnueabi/diffstat/1.60-r0/diffstat-1.60
-DHAVE_CONFIG_H   -D__extern_always_inline=inline -D_GNU_SOURCE -O2
-pipe -g -feliminate-unused-debug-types  -march=armv7-a -mthumb
-mfloat-abi=hard -mfpu=neon -no-integrated-as
--sysroot=/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/raspberrypi2
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/work/armv7at2hf-vfp-neon-angstrom-linux-gnueabi/diffstat/1.60-r0/diffstat-1.60/diffstat.c
NOTE: make -j 16
arm-angstrom-linux-gnueabi-clang -Wl,-O1 -Wl,--hash-style=gnu
-Wl,--as-needed -o diffstat diffstat.o
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/x86_64-linux/usr/bin/arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi-ld:
error: cannot open crt1.o: No such file or directory
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/x86_64-linux/usr/bin/arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi-ld:
error: cannot open crti.o: No such file or directory
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/x86_64-linux/usr/bin/arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi-ld:
error: cannot open crtbegin.o: No such file or directory
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/x86_64-linux/usr/bin/arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi-ld:
error: cannot open crtend.o: No such file or directory
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/x86_64-linux/usr/bin/arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi-ld:
error: cannot open crtn.o: No such file or directory
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/x86_64-linux/usr/bin/arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi-ld:
error: cannot find -lgcc
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/x86_64-linux/usr/bin/arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi-ld:
error: cannot find -lgcc
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/x86_64-linux/usr/bin/arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi-ld:
error: cannot find -lgcc
/mnt/home/kraj/work/angstrom/build/tmp-angstrom-glibc/sysroots/x86_64-linux/usr/bin/arm-angstrom-linux-gnueabi/arm-angstrom-linux-gnueabi-ld:
error: cannot find -lgcc

(From OE-Core rev: ebee06dbf195b74791feb63c23a637324c1656e8)

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

liburcu: Fix atomic function parameter types in calls

__sync_val_compare_and_swap family specifically asks for proper types of
pointer type parameters, gcc does not notice it but clang does and sends
out errors calling it out.

(From OE-Core rev: 2767b6be71ca809f4a39ba9b8707b311b8334434)

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

systemd: Remove exporting special CPP

This is no more needed.
it was done long ago while systemd lived in meta-openembedded
http://lists.openembedded.org/pipermail/openembedded-commits/2012-August/141061.html
The accompanying patch has been applied to systemd already so we were
not needing to set CPP for sometime now.

as a nice side effect it helps compiling systemd with clang

(From OE-Core rev: b816e3f520bf71c9b681ccea30c8eefd62fb20a2)

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

bitbake.conf: Define fallback DISTRO_NAME

When using OE-Core only we miss this define and its now used in recipes
like os-release, which goes undefined and when booting we see messages
like

Welcome to ${DISTRO_NAME} nodistro.0!

This would change to
Welcome to OpenEmbedded nodistro.0!

Remove all trailing whitespaces while touching this file

(From OE-Core rev: cf2383a63975ac7a51729a313fe0a52b559d2c61)

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

mdadm: Upgrade mdadm_3.3.2.bb -> mdadm_3.3.4.bb

Drop upstreamed patch for gcc5 compatibility
Forward port x32 patch

(From OE-Core rev: 2782ae4dfb417ba377f861283792d7d9929f2ae3)

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

psplash: Turn psplash_fb_plot_pixel() into static inline

This function is not used anywhere except psplash-fb.c so make it static
inline function which is portable across compilers

Fixes issues like
psplash-fb.o: In function `psplash_fb_draw_rect':
|
/mnt/home/kraj/work/angstrom/sources/openembedded-core/build/tmp-glibc/work/i586-oe-linux/psplash/0.1+gitAUTOINC+14c8f7b705-r15/git/psplash-fb.c:363:
undefined reference to `psplash_fb_plot_pixel'

(From OE-Core rev: 2429932c2dc1f81bf04f4377911094e2c8a04a19)

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

libvorbis: Update libvorbis_1.3.4.bb -> libvorbis_1.3.5.bb

Licence years have changes causing checksum changes

-Copyright (c) 2002-2008 Xiph.org Foundation
+Copyright (c) 2002-2015 Xiph.org Foundation

(From OE-Core rev: cb034175054ae31f7065677ba36b6a45dc067be0)

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

json-c: Include math_compat.c for NAN/INF values

The backport patch also fixes the warning which was being work arounded
by disabling unused-but-set-variable

(From OE-Core rev: ebce241d0badff0a12b4c03a428720e0b4a59f32)

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

sysvinit: Fix makefile bug found with clang

This is due to specifying .h files on linker cmdline
clang driver is picky about it, and its not entirely correct
either

(From OE-Core rev: de45b5e68faeefe3d68818d456f280b98f397634)

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

cross-canadian.bbclass: support for TCLIBC="baremetal"

Allow "baremetal" builds.

(From OE-Core rev: 0cd3121058ea620c74622f1200c8040696b4d1d8)

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

libgcc-common.inc: overrides for libc-baremetal

Added TCLIBC="baremetal" specific overrides.

(From OE-Core rev: 044f989b34e46f919e6b28860d7a1b396b2c3808)

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

gcc-cross.inc: overrides for libc-baremetal

Modifications to to support build for TCLIBC="baremetal"

(From OE-Core rev: 9e7da158e0e7628c22013fbee34270f7d2db621f)

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

gcc-cross-canadian.inc: overrides for libc-baremetal

Modify configure arguments (EXTRA_OECONF) to support build for
TCLIBC="baremetal"

(From OE-Core rev: 87e81e469fd941fa1b41ded3e02644695cafd2dc)

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

tclibc-baremetal.inc: baremetal specific distro policy file

Support for TCLIBC="baremetal".
Allows building "baremetal" toolchains.

(From OE-Core rev: f1972eea0145ca54f0c087c0f29fd2e54d6b95e8)

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

alsa-lib: Lift the limitation to use gcc only cross compiler

alsa-lib configure tries to see that if CC variabled doesnt contain
'gcc' then it tries to guess and redefine CC to be something gcc, which
doesnt work with OE since we define CC to have sysroot and options
attached to it. So all tests fail when configure constructs the CC for
cross compiler like this. This portion is not needed at all

(From OE-Core rev: 33a7eb64cded61601c6ad6883e4d64ebe9ff547c)

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

devtool: implement build-image plugin

Implemented new plugin to build image from workspace packages.

Plugin creates <image>.bbappend file, adds
all workspace packages to the image using IMAGE_INSTALL_append
variable in bbappend file. After that it runs 'bitbake <image>'.

(From OE-Core rev: 00bc43868da3ea2a4532215d3abef8e150c7b2e5)

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

devtool: make add_md5 a public API

Moved _add_md5 function from standard.py to __init__.py to
be able to call it from other modules.

(From OE-Core rev: ee38bb20dc7ba21dac782d8d13383f81dfedef55)

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

harfbuzz: upgrade to 1.0.2

Bugfix release.

(From OE-Core rev: 71dbd7cc2333fdf830f0f483774faeadbb0e2236)

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

btrfs-tools: 4.1.1 -> 4.1.2

* Rename btrfs-tools_git.bb to btrfs-tools_4.1.2.bb
* Remove PV in the recipe.

(From OE-Core rev: 4b55d7e02cecc4f9cd5028185f9452f0c90cbf89)

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

tune-octeon.inc: add BASE_LIB settings

Provide BASE_LIB settings for octeon* tunes that follow the practice of
mips64/mips64-n32 tunes (lib64 for N64 ABI, lib32 for N32 ABI).

(From OE-Core rev: 2b52312174e52886b0a978ece41f66b4fb455604)

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

tune-octeon.inc: correct packaging suffix

Octeon II/III binaries can contain instructions that are not compatible
with MIPS64 processors. Thus Octeon II/III packages should go to
separate directories. Set MIPSPKGSFX_VARIANT_tune-* to Octeon-specific
values and update PACKAGE_EXTRA_ARCHS_tune-* accordingly.

(From OE-Core rev: 69798449a8c1049728674dd352cf828063974cd0)

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

oprofile: 1.0.0 -> 1.1.0

* Remove backport patch filemode-fix.patch.
* Update --with-kernel=${STAGING_DIR_HOST}/${prefix} to find kernel
  headers (linux/*.h) to fix the error:
| checking kernel supports perf_events... unknown -- perf_event.h not found
| ERROR: You requested to build oprofile with '--with-kernel=/buildarea/lyang1/test_f2/tmp/work-shared/qemux86/kernel-source',
| but headers were not accessible at the given location.
| Be sure you have run the following command from within your kernel source tree:
|      make headers_install INSTALL_HDR_PATH=<kernel-hdrs-install-dir>
| Then pass <kernel-hdrs-install-dir> to oprofile's '--with-kernel' configure option.
| configure: error: Unable to build oprofile. Exiting.

(From OE-Core rev: b2200152deef0aec3cd8fe55bb73d6e8fccfe159)

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

gnu-efi: 3.0.2 -> 3.0.3

(From OE-Core rev: c731ecab0a6cb31198ec8270dc9381b14964dd3c)

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

gnupg: 2.1.6 -> 2.1.7

(From OE-Core rev: 78b159bf753c43c78bb0f283add7d447b96e08f1)

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

qemu-native: Enable temporary debug info as default.

We recently enable backtrace print rev[1] for address
qemu_cpu_kick_thread() issue but without debug information
isn't useful.

[YOCTO #8143]

[1] 73a876e4ff04ebbbcd3bde62efe02146ab601e25

Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd: Increase devices timeout in QEMU machines to avoid failures on serial-getty

Systemd serial-getty is failing because dev-tty ends with timeout, systemd
uses a default timeout of 90 secs that it's reached on AB's production environment
due to high I/O loads (nfs).

When use QEMU is used without KVM support, the machine code is executed by TCG
(software code execution) that is dependent on devices layer and causes locks
 between TCG/Devices increasing the machine code execution time [1].

QEMU don't support configuration of device timeout always uses a default one that is
90 secs, so the next patch increases the device timeout to 240 secs [2] in order give
enough time to get devices ready. It ONLY applies on QEMU machines.

[YOCTO #8141]

[1] http://blog.vmsplice.net/2011/03/qemu-internals-overall-architecture-and.html
[2] https://bugzilla.yoctoproject.org/show_bug.cgi?id=8141#c10

Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc-source: exlude from world

Exclude all versions of gcc-source from world builds so that:
   bitbake -c <stage> world
will work. gcc-source deletes most bitbake build stages
since it is a source-only package.

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

Empty image: filesystem allocation

Increase sparse image block size when ROOTFS_SIZE is smaller than
the minimum needed for ext4 to fit into it.

[YOCTO #7664]

Signed-off-by: Alex Franco <alejandro.franco@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Empty image: package list splitting and iteration

A few short fixes to splitting/iteration done over package lists
in license.bbclass, package_manager.py and rootfs.py.

[YOCTO #7664]

Signed-off-by: Alex Franco <alejandro.franco@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/runtime/multilib: fix and improve multilib test

Replace the previous shell magic with a Python function to parse the readelf
output, and fix the package names to include the lib32- prefix.

[ YOCTO #8219 ]

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

qemuarm64.conf: Make the second serial console /dev/hvc0

Since the qemu for aarch64 must use a virtual console for the second
serial port rather than emulating actual hardware, make sure the correct
device is specified so that a tty is actually started.

Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sysvinit-inittab: Run ttys on ttys that don't have tty in the name

On qemuarm64 the second serial port in SERIAL_CONSOLES will be hvc0.
Since that doesn't have tty in the name, a correct label didn't get
added to inittab.

This change makes both names with tty and without work.

Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bind: remove dependency on bind in bind-dev

bind doesn't ship shared libraries only static libraries, so the default
dependency on PN from PN-dev is pointless and means that an image with bind-dev
installed (via dhcp-dev's automatic dependency) ends up with named installed and
started on boot which is rarely intended.

If and when we ship bind's shared libraries we should ensure that the libraries
go into a separate package.

Also remove an old comment about --enable-exportlib which isn't supported by
configure anymore.

[ YOCTO #8216 ]

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

qemurunner: Don't loop on EWOULDBLOCK in logging thread.

EAGAIN/EWOULDBLOCK can be followed by no data. So don't tight loop
waiting for data.

Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu-internal: Make sure tcpserial is always last

If this is not the case, sometimes the additional tcpserial will be
enumerated as ttyS0, which is not what we want. Because then it would be
the console, and qemurunner would not log things properly.

Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runtime/syslog.py: fix syslog test crash

This patch fixes the ability to correctly identify syslog's package
name for the built image. It is derived from modifying oeqa/oetest.py
for [YOCTO #8170]

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

oeqa/oetest.py: add better package search for hasPackage()

Modified hasPackage() to split the content of pacakage manifest file
in containing lines and search at the begining of each line the
existance of the needed pkg.

[YOCTO #8170]

Signed-off-by: Costin Constantin <costin.c.constantin@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemurunner: In the logging thread retry on EAGAIN, EWOULDBLOCK

On a nonblocking socket an exception can be generated for the EAGAIN
and EWOULDBLOCK errnos. Since these aren't actually errors make sure to
retry rather than bailing out.

Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libnotify: update to version 0.7.6

New webkit requires at least version 0.7

Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>