terminal.py: use unique ids for screen sessions

When running multiple sets of builds on the same
system, it is hard to distinguish which build belongs
to which screen session and you can end up resuming
the wrong session.

The simple solution is to just append the process
id to the screen session invocation to make each
unique.

(From OE-Core rev: 1677b736bca5dc46db522da1874459d2de77209d)

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

bitbake.conf: add variables to exclude from hash.

The parallelism flags should not change the parse hash.

(From OE-Core rev: 6bfd4a6abd7488e663598620f4436ac49183528f)

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

kernel-yocto: don't require meta branch for custom repos

The commit a9f11849 [linux-yocto: explicitly export KMETA to scripts]
allows the meta branch name to be changed by exporting it to all
phases of the build.

But if a custom kernel without a meta branch is built, we end up
passing an empty string to the creation scripts, which breaks the
build since input is expected.

Inhibiting the export of KMETA to the creation scripts when empty
fixes the problem.

(From OE-Core rev: 61162fbe01993659301fe2e821bf9c3e801206d8)

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

kernel-yocto: set master branch to a defined SRCREV

To support custom repositories that set a SRCREV and that only have
a single master branch, do_validate_branches needs a special case
for 'master'. We can't delete and recreate the branch, since you
cannot delete the current branch, instead we must reset the branch
to the proper SRCREV.

(From OE-Core rev: de5bb5879fa3282c46dc1ede36af34eaab8f647f)

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

kernel.bbclass: fix INC directory for SLANG

In the kernel's tools/perf/Makefile CFLAGS was previously hard coded to
contain "-I/usr/include/slang" to work with hosts that have
"/usr/include/slang/slang.h" as well as hosts that have
"/usr/include/slang.h". This path can cause compile warnings like:

  cc1: warning: '/usr/include/slang' doesn't exists.

  or

  cc1: warning: include location "/usr/include/slang" is unsafe for
  cross-compilation [-Wpoison-system-directories]

Then in some cases warnings become errors if WERROR is enabled hence
build errors.

In coordination with a kernel fix, we can fix this error for all
kernels by modifying the perf Makefile within the staged kernel
source.

(From OE-Core rev: 4e0daf05ed04e9fb4343032c6290a379f53b89f3)

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

grub-efi: fix Nothing PROVIDES 'grub-efi-x86_64-native'

"ERROR: Nothing PROVIDES 'grub-efi-x86_64-native'" is returned when
trying to build meta-intel/meta-cedartrail with DEFAULTTUNE = "core2-64"
Used TRANSLATED_TARGET_ARCH instead of TARGET_ARCH in grub-efi.bbclass

[YOCTO #3013]

(From OE-Core rev: 20e2aaacb36113604c4ea7c40dd0bbbf224b8360)

Signed-off-by: Mihai Lindner <mihaix.lindner@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kmod-native_git.bb: fix builds for hosts with older libc

kmod will fail to build with the following error because O_CLOEXEC is
not defined:

| libkmod/libkmod-module.c: In function 'kmod_module_get_initstate':
| libkmod/libkmod-module.c:1640: error: 'O_CLOEXEC' undeclared (first use in this function)
| libkmod/libkmod-module.c:1640: error: (Each undeclared identifier is reported only once
| libkmod/libkmod-module.c:1640: error: for each function it appears in.)
| libkmod/libkmod-module.c: In function 'kmod_module_get_refcnt':
| libkmod/libkmod-module.c:1754: error: 'O_CLOEXEC' undeclared (first use in this function)
| libkmod/libkmod-module.c: In function 'kmod_module_get_sections':
| libkmod/libkmod-module.c:1913: error: 'O_CLOEXEC' undeclared (first use in this function)
| libkmod/libkmod-file.c: In function 'kmod_file_open':
| libkmod/libkmod-file.c:282: error: 'O_CLOEXEC' undeclared (first use in this function)
| libkmod/libkmod-file.c:282: error: (Each undeclared identifier is reported only once
| libkmod/libkmod-file.c:282: error: for each function it appears in.)

Since we are only using kmod-native for depmod, and it's a non-threaded
user of this libary being built this should be safe to override O_CLOEXEC.

Keep in mind this is ONLY effecting the native builds and not what is
being shipped in the root file system.

(From OE-Core rev: edcb57fe308979cc5b92359d064e32bb70d4ad76)

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

base-passwd: upgrade to 3.5.26

The S variable is now the default one.

(From OE-Core rev: a8b1081791af60873c6fd26cbf298cfe1886f97d)

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

gdk-pixbuf: call gtk-update-icon-cache with -t in postinstall script

gtk-update-icon-cache is called for each subdir of ${datadir}/icons,
but there are some themes without an index.theme file such as
xcursor-transparent, so add -t(--ignore-theme-index) option to avoid
errors of gtk-update-icon-cache.

(From OE-Core rev: bab34e5fa734a98cb5199db4ebc95a6634a38431)

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

wget: upgrade to 1.14

remove-gets.patch: removed
- included in the new version

(From OE-Core rev: 396af9e2c8a8b7ad8d49566350d79f05898734a7)

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

npth: upgrade to 0.91

(From OE-Core rev: 99ef4adf7d5be6a565441505b5ad3308f67b7138)

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

kconfig-frontends: upgrade to 3.5

(From OE-Core rev: 512fdcbaea71711f75a7389c14319dbf865cdfaf)

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

dhcp : upgrade to 4.2.4-P1

(From OE-Core rev: 5f68e2c4fb746ff798f9747776cc1f7620fb39dd)

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

subversion: update to 1.7.6

Dropped --without-apache option as it does not exists.

Added patch from subversion-users ML to not build mod_dontdothat.

(From OE-Core rev: c79fb25161b958b07fbfa965768754d51717d616)

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

runqemu: Move the KVM check after the MACHINE/KERNEL Checks

The MACHINE/KERNEL Checks setup MACHINE is it's unset, the KVM checks valid
MACHINE is qemux86 or qemux86-64 and fail if it's unset!

[YOCTO #2970]

(From OE-Core rev: 8d5e6999caff50a4b7d9a9ba69f9875285270459)

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

build-appliance-image: Update SRCREV for beta build

(From OE-Core rev: 1c4d7a04ff47c5b5e6fa4f689f60f7178192873b)

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

task-core-lsb: Don't pull in eglibc-pic via RDEPENDS

The task-core-lsb-runtime-add subpackage includes eglibc-pic as an
RDEPEND.  That's incorrect, eglibc-pic is nothing but 22MB of static
libraries and should never appear as a runtime dependency.

(From OE-Core rev: 79545d068659c943ac9aa925a7da0eee82207b8b)

Signed-off-by: Andy Ross <andy.ross@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libcap: fetch from Debian

Original location was kernel.org and it was not restored after server
was hacked.

(From OE-Core rev: 149b8bebe21ebf2f126d3ab2a21c036c556a63c7)

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

build-appliance-image: Add vmx* files and build zip file

This commit adds the vmx* files needed to setup a VMware image,
this also packages the vmdk along with the vmx files.

(From OE-Core rev: 968cfc0b630fb409430a46b1512d6bf0de225ad1)

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

build-appliance-image: Update SRCREV to Denzil 1.2.1

(From OE-Core rev: 3aeb67524cecfe4890fb8e76d75a82ed037c65f0)

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

iputils: chmod in pkg_postinst are redundant

With pseudo the cmod in pkg_postinst is actaully redundant since
the do_install uses install -m 4555 and pseudo is able to track that
between the install time and the rootfs package installation, so the
perms are correct.

[YOCTO #2894]

(From OE-Core rev: a2bc81032c85548f7c72dc76e6020ab4df9af6de)

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

mc: upgrade to 4.8.4

(From OE-Core rev: 2c57834687d9cc1c2f14d701e5f35e5b7c779aad)

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

dbus-glib: upgrade to 0.100

test-service-glib-include-glib-only.patch: removed
- no more compiling errors

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

libnfsidmap: upgrade to 0.25

fix-ac-prereq.patch: adapted to the new version

(From OE-Core rev: fdd9df341f47ef03ea9cb319205d75418c94cd01)

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

foomatic-filters: upgrade to 4.0.17

(From OE-Core rev: ff64550810754f9f061015507719e3082272cbb3)

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

telepathy-idle: upgrade to 0.1.12

Add patch for fixing the _gen/svc-gtk-doc.h target in
the makefiles.

(From OE-Core rev: a39cad99a95dd5e31213e4661304f515dc48628d)

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

telepathy-glib: update to 0.19.6

The 0001-Fix-a-race-in-_gen-reentrant-methods.list patch is
included in the 0.19.6 version.

(From OE-Core rev: 712c365bf745de3fe927fcc24da505009d17a3e5)

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

distutils/steuptools: Fix files layout and unbreak builds

The last two distutils changes progressivly broke the builds. Firstly they
moved things from the site_packages directory to being higher up the tree
which introduced package QA warnings as a side effect. Secondly, it interacts
badly with setuptools which passes in --root=${D} itself.

This patch restores the original directory layout, hence fixing the QA
warnings and also passes extra options to setuptools to deal with the
--root option it passes.

(From OE-Core rev: bed18d5df7915e4127a538be9c7550e185c8c850)

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

mutter: Explictly disable libcanberra

Its not listed in DEPENDS, this ensures build reproducibility.

(From OE-Core rev: a6029573fba6badc077486f2e3eab8df46e455ad)

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

gcc-4.7: Add fix for libtool rpath problems

This avoids problems with libstdc++ having bad rpaths (/usr/lib/../.lib)
in its .la file. See the patch for more information.

(From OE-Core rev: bd27e81f1fa49e2770da9a4a65a83e9d4c3a0dd0)

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

libtool: update rpath normalization to use builtin

Use the built-in normalization function instead of the sed hack.

(From OE-Core rev: ba8263e5dc520f5024fc76d8bd2e10fe0564b0e2)

Signed-off-by: Andy Ross <andy.ross@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ld: -rpath must search under sysroot

The -rpath argument would search the host filesystem for libraries,
even when a sysroot was defined.  For cross toolchains with targets
compatible with the host architecture this can find incorrect
libraries.  Leave -rpath-link unmodified, as build systems in the wild
are already using this to point to host directories.

[YOCTO #2965]

(From OE-Core rev: bccea580f1abb762d231f785a4e60c9cd368dcdf)

Signed-off-by: Andy Ross <andy.ross@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xev: Add missing dependencies on libxrandr and xproto

(From OE-Core rev: dd56ca611671233b8eec78aaa7e24b232654bc92)

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

external-python-tarball: Blacklist recipe if ipk is not in the PACKAGE_CLASSES list

This is needed because ipk tools and variables (specically
IPKGCONF_TARGET and IPKGCONF_SDK) are used from the rootfs_ipk.bbclass

I tried to inherit directly but it still fails to expand those.

[YOCTO #2814]

(From OE-Core rev: 161043782682cccd2efc2b349e8248ae084857c7)

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

libiconv: skip recipe if using eglibc

libiconv is provided for use with uClibc - if you build it together with
eglibc (which already PROVIDES virtual/libiconv) you can end up with
dependency problems during do_rootfs.

(From OE-Core rev: f3e2ccff952f148522a09c09e0dea92e59bab5b6)

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>

cups: disable pdftops

This ensure a reliable built without host contamination, this
will also disable the usage of ghostscript.

[YOCTO #2966]

(From OE-Core rev: 90bccaa290d2bc04c7ea1bbeb2cddb3509d0d380)

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

task-base: pull in the bluez asound module when appropriate

If the machine and distro both want alsa, then have task-base-bluetooth pull
in the libasound-module-bluez package.

(From OE-Core rev: 4d7b8da48a1df57bf8e08eff368564f7b47a5222)

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

busybox: Upgrade to upstream 1.20.2

Updated defconfig file from generated .config
Disabled CONFIG_LSOF,CONFIG_CROSS_COMPILER_PREFIX, and CONFIG_EXTRA_CFLAGS
for backwards compatibility.
Added busybox-1.20.2-kernel_ver.patch from upstream.

(From OE-Core rev: e5f2cc805509a23e160c6f05e6776d6d32e3ce98)

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

grub-efi-native: Update to version 2.00

Added -Wno-unused-result because of a read() with no result, and causing
error due to -Werror
Updated grub recipe because of a patch name change
Removed unnecessary patches (merged upstream).

(From OE-Core rev: 88cd5d99918a896f515e8bb59b7238f3ad50adff)

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

grub: Update to upstream stable 2.00

Removed unnecessary patches (merged upstream).

(From OE-Core rev: d888f50d5cfdf5b3881f81c26f0349ef41c06197)

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

scripts/combo-layer: specify branch when getting current revision

Handle the case during update where the configured branch is not
currently checked out in the component repository by just specifying
it in the places where it was not previously.

(From OE-Core rev: a3ddf39af7e7cd47156677d8ae753964ee582745)

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>

grub-1.99: Add flex-native to DEPENDS

[YOCTO #2916]

On some systems the host version of flex may be too old to work properly,
instead we depend on flex-native to ensure that we have the proper version
available to us.  (Flex version 2.5.35 or newer is requried.)

(From OE-Core rev: b53b4c4d6dad745b2eb0ab804f7975146a81aedf)

Signed-off-by: Xin Ouyang <Xin.Ouyang@windriver.com>
Clarified commit message

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

eglibc-2.15: Delete

eglibc 2.15 is not buildable with current toolchain build sequence
and is not used as default on OE-Core. So lets remove it.

(From OE-Core rev: d8c47eeb09d1bc2a6a7a335cc94658f6bdfe4026)

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

linux-yocto: Remove parens from COMPATIBLE_MACHINE regexp

This helps in appending to this regexp from bbappends
coming from other layers who want to leverage linux-yocto
with minimum tweaks by using a standard prefefined machine
from linux-yocto

(From OE-Core rev: 32ba716e71fb7a16f13c83ab6c8cc51de56b3be2)

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

runqemu: Honor KERNEL and ROOTFS settings from environment

Expand the usage to be able to specify KERNEL and ROOTFS
on commandline. This helps in using the script for booting
images that are essentially not part of OE-Core

(From OE-Core rev: c985b02130658dd64581ecf14b16e2c70d1d8db5)

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

mesa: clean up PACKAGECONFIG

Use machine overrides instead of machine features.

(From OE-Core rev: 821a38d71d7e15f5871f44a3ee08268d05ef610e)

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

bitbake: ast: Store anonymous python function contents in the datstore

This is useful if we need to disable part of one during a backtrace
for debugging purposes.

(Bitbake rev: 80a0c1b06a30a6ba9977c29fac0437a208d8cbbc)

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

bitbake: ast: Extract text variable in PythonMethodNode

(Bitbake rev: 2e98f740b4a57a3467b1a00b1ebc1aaee33a8ff0)

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

bitbake: ast: Rename PythonMethodNode define variable to modulename

It was hard for me to understand what the define variable was, modulename
is hopefully a bit better.

(Bitbake rev: 79f9f46319de85f85613ebe248c327f5852225ba)

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

bitbake: methodpool: Clean up the parsed module list handling to be slightly less insane

This removes some dubious functions and replaces them with a simpler, cleaner
API which better describes what the code is doing. Unused code/variables are
removed and comments tweaked.

(Bitbake rev: f1e943aebfb84e98253f3f21646d6765c4fa1d66)

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

bitbake: methodpool: Remove unused check_insert_method function

(Bitbake rev: 4664333e55ffc9d618f2c2f074200923209fd6c4)

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

bitbake: cooker: Add explict handling of BBHandledException for parsing failures

This removes some unnecessary tracebacks

(Bitbake rev: db0ff7b4d47fce8322dd2350a2b1a6f60ef61d25)

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

bitbake: cooker: Ensure parsing failures stop the build

Currently parsing failures still allow bitbake to continue on and try
and execute a build. This is clearly a bad idea and this patch adds in
more correct error handling and stops the build.

The use of sys.exit is nasty but this patches other usage in this function
so is at least consisent and its better than the current situation of
trying to execure a half parsed set of recipes. There are probably better
ways this could be improved to use to stop the build.

(Bitbake rev: 22756e9c0f1da33ba2c6e881b214577a610b7986)

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

bitbake: utils.py: Once we've printed errors messages, suppress the backtrace

Adding a generic backtrace to better_compile and better_exec is pointless,
therefore reraise the exception as a bb.BBHandledException so the
generic code doesn't confuse the user even more.

(Bitbake rev: b3d97130e1e70fe969399277dcd7cccd888103d6)

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

bitbake: utils.py: Try harder to extract good traceback information by querying the datastore

Currently as soon as execution passes outside the code fragment being
executed by better_exec, we don't get any good traceback information,
just a likely obscure reference to some function name which may
or may not be identifiable.

This patch adds code to query the datastore if present, allowing a more
meaningful back trace to be displayed in many cases.

[YOCTO #2981]

(Bitbake rev: 0edf8431f9ff52581afe0d3ef525c59909af02ba)

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

bitbake: cooker: fix traceback when using -b with skipped recipe

If a recipe is skipped during parsing for whatever reason, check and
report this as an error rather than trying to use the data that is sent
back and failing.

Fixes [YOCTO #2976].

(Bitbake rev: a324df40243fa55ccc89fd5970d46f25330d0a0d)

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

gmp-4.2.1: Set CC_FOR_BUILD to ensure the host system compiler is used, not the target one

This addresses errors like:

| NOTE: make -j 16
| i586-poky-linux-gcc  -m32   -march=i586 --sysroot=/srv/home/pokybuild/yocto-autobuilder/yocto-slave/nightly-non-gpl3/build/build/tmp/sysroots/qemux86 `test -f 'gen-fac_ui.c' || echo './'`gen-fac_ui.c -o gen-fac_ui
| i586-poky-linux-gcc  -m32   -march=i586 --sysroot=/srv/home/pokybuild/yocto-autobuilder/yocto-slave/nightly-non-gpl3/build/build/tmp/sysroots/qemux86 `test -f 'gen-fib.c' || echo './'`gen-fib.c -o gen-fib
| i586-poky-linux-gcc  -m32   -march=i586 --sysroot=/srv/home/pokybuild/yocto-autobuilder/yocto-slave/nightly-non-gpl3/build/build/tmp/sysroots/qemux86 `test -f 'gen-bases.c' || echo './'`gen-bases.c -o gen-bases -lm
| i586-poky-linux-gcc  -m32   -march=i586 --sysroot=/srv/home/pokybuild/yocto-autobuilder/yocto-slave/nightly-non-gpl3/build/build/tmp/sysroots/qemux86 `test -f 'gen-psqr.c' || echo './'`gen-psqr.c -o gen-psqr -lm
| ./gen-fac_ui 32 0 >mpz/fac_ui.h || (rm -f mpz/fac_ui.h; exit 1)
| ./gen-bases header 32 0 >mp_bases.h || (rm -f mp_bases.h; exit 1)
| ./gen-bases table 32 0 >mpn/mp_bases.c || (rm -f mpn/mp_bases.c; exit 1)
| ./gen-fib header 32 0 >fib_table.h || (rm -f fib_table.h; exit 1)
| ./gen-bases: error while loading shared libraries: libgcc_s.so.1: cannot open shared object file: No such file or directory
| ./gen-bases: error while loading shared libraries: libgcc_s.so.1: cannot open shared object file: No such file or directory
| ./gen-fib table 32 0 >mpn/fib_table.c || (rm -f mpn/fib_table.c; exit 1)
| make: *** [mp_bases.h] Error 1

and matches the fix for the other gmp version.

[YOCTO #2992]

(From OE-Core rev: 24d31cd30080401271936422e6bd77727a16e0a6)

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

libtool: normalize link paths before considering for RPATH

Libtool may be passed link paths of the form "/usr/lib/../lib", which fool
its detection code into thinking it should be included as an RPATH in
the generated binary.  Normalize before comparision.

(From OE-Core rev: e3b0925f4c861393e436deb6b1912f9f1f325d1e)

Signed-off-by: Andy Ross <andy.ross@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

distutils.bblass: change order of args to install step

This let's the user override install-lib argument again if it needs
to be something else, otherwise things like python-setuptools
won't be able to modify the install-lib dir

This fixes a new issue exposed by my previous distutils patch
that fixed the python modules default install location. Also,
it removes running the install step twice which was inadvertant

(From OE-Core rev: 3b23feca31480cc56f55301fd0274e622c40b522)

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

README: Clarify where to send patches

There is still a great deal of confusion over where patches for the
various components of the poky repository should be sent. Attempt to
mitigate this by including a blurb in the poky README similar to what
other layers and repositories do.

V2: Call out the special care required when dealing with the scripts
    directory per Saul Wold's recommendation.

V3: Recommend applying patches to oe-core prior to sending patches to
    the oe-core mailing list.

(From meta-yocto rev: d2747240e75ca0d43f569161694799c3d5d23014)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
CC: Richard Purdie <richard.purdie@intel.com>
CC: Paul Gortmaker <paul.gortmaker@windriver.com>
CC: Paul Eggleton <paul.eggleton@linux.intel.com>
CC: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

conf/bitbake.conf: allow cache reuse between bitbake, bitbake-layers and hob

BB_ENV_EXTRAWHITE is set by the bitbake wrapper script but not by
bitbake-layers, and thus it was affecting the data hash and thus
invalidating the cache. Any variables passed through from the
environment by being added to this variable should themselves
influence the hash already if they are not already excluded from,
the data hash, so BB_ENV_EXTRAWHITE can be excluded.

Similarly, DISABLE_SANITY_CHECKS is set by the hob script in order to
have sanity checks run later, but not set by the bitbake wrapper script.
It doesn't actually affect the build itself and so its value can also
be excluded from the data hash.

These changes prevent unnecessary reparsing between executions of
bitbake, bitbake-layers, and the initial pseudo build within the hob
script.

Fixes [YOCTO #2680].

(From OE-Core rev: 84bcacd5b4737520f0fc76333f0165ef5b87c207)

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

gcc: bump PR

* missing in 30617bde61a3b0a0944b49a0c9fb7159dacbb19f, eglibc-initial fails to compile in incremental build with OEBasic

(From OE-Core rev: 495ccece87b56d696e92c5cdc39b79fd9813caa0)

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

bitbake: event/ast: Use better_exec instead of simple_exec

This improves the stacktraces dumped by bitbake when for example anonymous
python functions fail.

Also default to passing code strings to better_exec to match the behaviour of
simple_exec to aid the transition.

(Bitbake rev: 7e8205929ae953731a6854ea80b197847cff5771)

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

bitbake: ui: Improve error message if bitbake cannot import python curses module

On some SuSE systems, the curses python module is not installed by default.
Instead of a python failure, we want a nicer error message.

(On SuSE systems the package is typically python-curses.)

(Bitbake rev: 65a5845ac942d0aa6838c295e41b656f9d2a98bb)

Signed-off-by: Konrad Scherer <Konrad.Scherer@windriver.com>
Signed-off-by: Jeff Polk <jeff.polk@windriver.com>
Reword commit message, rebase to latest bitbake.

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

bitbake: bitbake-layers: use parsable title for show-appends

Print title text surrounded by === as we do with other subcommands so
that you can filter it out in scripts. (You will also want to filter out
lines starting with "WARNING:" if you just want the list).

(Bitbake rev: ba15ba16a5d863886bcfd5b3f0bdfff2a6de11a0)

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

bitbake: bitbake-layers: avoid full parse for show-layers subcommand

We don't actually need to parse all recipes just to show the configured
layers, so just parse the configuration instead.

(Bitbake rev: 406477cbae066b6379873e266cb79801e545a61c)

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

bitbake: bitbake/cooker: minor refactor of parseConfiguration()

All calls to parseConfiguration set self.status and call
self.handleCollections() afterwards, so just do this inside
parseConfiguration() itself.

(Bitbake rev: 3c2322d8a6ce15e20adb07a61aa321d884a9bcca)

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

bitbake: data_smart: skip all interal (underscored) flags in getVarFlags

(Bitbake rev: 9f631e29a2eebb96a8291839dd8b39aa9126a10e)

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

bitbake: data-smart: rename the 'content' internal variable to '_content'

(Bitbake rev: 684cf09aed09ec82c8afb99895f92d73cd0519df)

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

bitbake: Hob wizard-like reorganization

Hob is now more context-sensitive regarding
user changes/options. Also, the workflow have been
streamlined and resembles more of a wizard.
Beautified some hardcoded values.
Fixed typo.
New streamlined Image Configuration page.
Build and/or Edit image buttons presence
is context sensitive.
Recipes and packages tabs selected automatically
based on custom image or pre-defined target image
(included or all).
Context sensitive Back button.

Fixes [YOCTO 2165]

(Bitbake rev: b48cd7dcf57b1abc8c5b46ced11d4f57bf06e557)

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

bitbake: bitbake: hob: create console log directory before writing

The default value of BB_CONSOLELOG in OE-Core now points to a directory
that might not exist prior to the first build, so ensure it is created
first.

(Bitbake rev: 660821769b9dfb89086291ffc835e9b5444509be)

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

classes/package: document do_packages_split arguments

This function takes quite a number of arguments and can be tricky to use
properly; this is not made easier if it is undocumented, so document all
of the arguments. (No functional changes, comments only.)

(From OE-Core rev: 324fbcc047819b6948d13eda627ec898cb1f1eb8)

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

gcc-cross-intermediate, gcc-crosssdk-intermediate: Remove

These recipes arent used anymore so delete them

(From OE-Core rev: 4ef8960c8d2876914bb78cbdce5fae3c5297e942)

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

gcc-crosssdk-initial_4.7.bb: Add --with-native-system-header-dir to EXTRA_OECONF

gcc-crosssdk-intermediate will go away and -initial will
assume its responsibility therefore transfer the option

(From OE-Core rev: 7f6883b225f9a0fdc44be2d1cb66ddc744174fbc)

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

uclibc: Use gcc-cross-initial as staging compiler

Do not build fake shared libraries during uclibc-initial phase

Let uclibc depend on gcc-cross-initial instead of gcc-cross-intermediate
thereby set stage to remove gcc-cross-intermediate

Let uclibc depend on uclibc-initial so that boostrap sysroot
gets populated properly

(From OE-Core rev: b954c2d7cd978adc059d672425c197eaaf57e852)

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

uclibc.inc,uclibc-config.inc: Untabify python snippets

To appease bitbake

(From OE-Core rev: cf975073a11c93f4a9fb5bdd72c16dc0ca9c3c54)

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

eglibc: Fix eglibc-initial and let eglibc depend on it

eglibc-initial does not need to install fake shared libs
anymore so drop it.

eglibc now should depend on eglibc-initial so that bootstrap
sysroot gets populated properly.

Drop references to gcc intermediate from glibc testing
scripts.

(From OE-Core rev: bddc141e7cec18744902787ab53c21bf28e71f78)

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

gcc-cross: Make gcc-cross-initial as the only intermediate gcc stage

Now glibc can be compiled with gcc-cross-initial therefore prepare
the stage to drop gcc-cross-intermediate

Also drop arm-nolibfloat.patch should not be needed anymore
half of changes in this patch are meant for OABI which we dont
use anymore

(From OE-Core rev: 30617bde61a3b0a0944b49a0c9fb7159dacbb19f)

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

eglibc-2.16: Update to top of 2.16 branch

Some translations and tst-gettext fixes

(From OE-Core rev: 0ad2bde39f03531cbef114253d566a36525399df)

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

eglibc-2.16: Remove build dependency on libgcc_s and libgcc_eh

This patch backports two patches from glibc trunk essentially
dropping the requirements to have libgcc_s and libgcc_eh
this will simplify the toolchain bootstrap sequence.

(From OE-Core rev: 9591c93c20af236fdc1a99b548435b6df75af497)

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

coreutils: fix to build with acl support

Fix the issue that coreutils was built without acl support:

configure: WARNING: libacl development library was not found or not usable.
configure: WARNING: GNU coreutils will be built without ACL support.

which will cause ls and cp commands don't have the ACL related behaviors:
ls -l: no plus sign(+) after the permission string for files that have ACL
cp -p: can not preserves the ACLs.

[YOCTO #2959]

(From OE-Core rev: b30ef39cc26dd0b6ed08bd2b1a6394de6b540720)

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

package_rpm.bbclass: fix and enhance the incremental rpm generation

The incremental rpm generation usually broke when package_rpm.bbclass
changed, change its implementation to make it more stable:

* It depended on the previous and current saved manifest files in
  the past, it would break when the manifest changed. Now query the
  previous and current installed pkgs from rootfs/var/lib and
  rootfs/install/, this would be more reliable, the manifest's change
  would not affect it any more.

* Add explanations before package_install_internal_rpm to explain what
  does the function do.

* Remove an unwanted "awk '{print $1}'".

[YOCTO #2906]

(From OE-Core rev: 5d8ca498e09438bd91654fa8b8b2c970956d88e3)

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

insane.bbclass: Fix RPATH warning in the face of funny path strings

In toolchain edge cases it's possible for the RPATH of a library to be
set to something like "/usr/lib/../lib".  This should be detected as
"/usr/lib" and generate a warning.

(From OE-Core rev: 72a924d0686389d648338efd3f675fc85ee2d181)

Signed-off-by: Andy Ross <andy.ross@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

formfactor: Clean up errors on machines without framebuffers

If fbset can't find a framebuffer we end up with a mess on the console.
We might as well avoid doing this on machines without a framebuffer.

(From OE-Core rev: f5c39ea822b440ae4bdd38be94fcf8a2814d2229)

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

meta/classes: Various python whitespace fixes

It was pointed out we have a number of weird indentations in the python functions.
This patch cleans up 3, 7 and other weird indentations for the core bbclass files.

It also fixes some wierd (odd) shell function indentation which my searches picked up.

(From OE-Core rev: 8385d6d74624000d68814f4e3266d47bc8885942)

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

qemu: Fix broken accidental path move

(From OE-Core rev: 04568d1f18250d531aad5d286157d2d559083520)

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

sstate-cache-management.sh: update for the SSTATE_MIRRORS

Several fixes:
* We have put the sstate file to SSTATE_DIR/??/ currently, but the
  sstate file on the SSTATE_MIRRORS or the obsolete one is still in
  SSTATE_DIR/ (no subdir), update the script to support manage them.

* Remove the related ".done" file in the SSTATE_DIR.

* Add a "-L, --follow-symlink" which will remove both the symbol link and
  the destination file

* Change the "ls -u file_list" (access time) to "ls -t file_list"
  (change tiem), since the "ls -u" and readlink will change the
  symlink's access time, which would make the result inconsistent.
  A solution is save the access time before every "ls -u" and "readlink",
  save it back after the command, but this would cause performance lost
  since it needs check each file and modify the symlink's status. Use
  the "-t" doesn't cause much different.

[YOCTO #2897]

(From OE-Core rev: 209ec08787981345a7f62b10a8a5c2ace0887c8e)

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

classes/sanity: remove texi2html from required host utilities

texi2html is not actually required to build world of OE-Core anymore,
so we don't need to mandate it. The only difference without it (verified
with buildhistory) is that groff doesn't produce HTML documentation; the
rest of its docs are still produced and packaged and no other packages
are affected.

Part of the work towards [YOCTO #2423].

(From OE-Core rev: aa1c4519f8e549b43a7d8bc51c5342d3409b464b)

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

kernel.bbclass: Preserve generated headers only if they exist

* I have 2.6.37 kernel without this directory and do_install now fails
  after 813256bf7bb6e26d542d5f769e2802564116ebe5

(From OE-Core rev: 9a83afe0794e96590aee374555efea91b3085b91)

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

scripts/runqueue-internal: Drop enable-gl option

This is no longer necessary since we've dropped the GL passthrough patches.

(From OE-Core rev: b6bc580f1cf56c2e771e87dfeadbb1e5d8416fce)

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

qemu-git: Drop GL passthrough support

These patches are a maintenance nightmare and impacting our abaility to
keep up to date with qemu. They are also a source of various bugs.

Remove them until someone is willing to step up and maintain them,
or upstream gains GL support.

(From OE-Core rev: 30af78f0db16b9f51666341c9dad0123ccf8ac85)

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

qemu-0.15: Drop GL passthrough support

These patches are a maintenance nightmare and impacting our abaility to
keep up to date with qemu. They are also a source of various bugs.

Remove them until someone is willing to step up and maintain them,
or upstream gains GL support.

(From OE-Core rev: e9a6df98458d9147227659d3888eff01589f2f76)

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

libpam: Fix missing DESTDIR for a mkdir causing build failures

(From OE-Core rev: c39e823138cbf4210e17bdb95ca322ec0a6c8f78)

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

randrproto: upgrade to 1.4.0

License change due to addition of missing Copyright

(From OE-Core rev: 4e8df7f9a3f506d69c85b051cad3426fb1b18fb9)

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

avahi: fix packaging of avahi-autoipd

- Avahi's "avahi-autoipd" file is in the dhcp/ folder, not dhcp3/
- udhcpc files were being installed but not being shipped in avahi-autoipd
- do_install override for avahi-autoipd was not being called and is not
  required, so join it into regular do_install.
- While at it, fixup string quoting in description.

This makes avahi-autoipd work with simple ifup/down scripts.

(From OE-Core rev: 172e663f1b00733cf956cb118d7a5e1a252c1671)

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

libxrandr: upgrade to 1.4.0

Only some spaces changed in the license file.

(From OE-Core rev: a1c2330a514a8fa627c55678afb34aa724734a56)

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

libpam: Update recipes from 1.1.5 -> 1.1.6

Drop include-sys-resource.patch already fixed upstream
LIC_FILE_CHKSUM change is due to deletion of space in
COPYING file see
http://git.fedorahosted.org/cgit/linux-pam.git/commit/COPYING?id=1814aec611a5f9e03eceee81237ad3a3f51c954a

(From OE-Core rev: 619092b699bfd79e060755fa41645cac7ac4fd0d)

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

unifdef-native: Correct LICENSE

The header in sourcefile says its BSD licensed
not GPL therefore make the correction

(From OE-Core rev: 06bd5b3d10d479fce02cd1d5821faeb9afb4ebac)

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

core: Prefer mesa-dri as virtual/libopengles1/2 provider

Wihtout it, you have both mesa-dri and mesa-xlib as providers. Let's
prefer the accelerated version.

(From meta-yocto rev: 13f847744aa842ef38b3f2adf3764425319b5507)

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>

linux-yocto/3.0: update meta SRCREV

Bumping the meta SRCREV to pickup this fix:

    meta: rename virto.scc to virtio.scc

    The virtio configuration block is misnamed. BSPs that include it with
    the proper name, now throw an error (as they should). So fixing the
    name of the fragment fixes the build.

(From OE-Core rev: c6b335099224d590109a8bcda22bf54c597c7eee)

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