bitbake: fetch2/__init__.py: Add NoChecksumError exception

Without the new exception, when the system is configured to use premirrors,
but not allow network access (via BB_NO_NETWORK), when a recipe was lacking a
checksum the wrong error message(s) were being generated.

Instead of complaining about trying to perform network access, if the system
was able to find the item in the premirror, it should inform the user of the
lack of checksums, and the two SRC_URI fields they should use to update their
recipe.

(Bitbake rev: cb10e9c03a3f96d94e27e18330009616dde5e2b3)

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

guile: updated to 2.0.6

Some patches (disable_goops_optimizations, fix_cross_compilation) don't
seem to be needed anymore. Others (change-install-data-hook, mark-unused-modules)
were updated to work on 2.0.6. Tested by building and running guile under
QEMU with core-image-sato-sdk and also builing guile on MIPS.

(From OE-Core rev: 310f169d3d89f3a4fc6a540974a30c7eb565db3a)

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

busybox: udhcpc shouldn't be a service

udhcpc is invoked from the networking service, so it shouldn't exist
standalone. If /etc/init.d/busybox-udhcpc is present, it will cause
udhcpc to be started two times.

And since the script doesn't support the stop action, it will cause
udhcpc to be started even when system is being shutdown.

Remove /etc/init.d/busybox-udhcpc to resolve the issues.

[YOCTO #2840]

(From OE-Core rev: 8329c4679ec0ce319d2a81d755a0da5b05474688)

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob: deal event DiskFull

Part of [Yocto #2168]

When bitbake runqueue is teminated by disk monitor, it will send event
DiskFull. Update to handle it.

(Bitbake rev: a691de10d8b8de552628b277e78d47c607ee3a46)

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

bitbake: monitordisk: fire event DISKFULL when terminate build

Part of [Yocto #2168]

Add a event DiskFull to descript the termination by disk monitor.

Update check() to fire the event DiskFull when terminates the build.
This could help UIs to deal this scenario and show more information to
end user.

(Bitbake rev: 0a0fca3d94a1db6458ae21501a66461a334410ed)

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

dbus: Ensure dbus-nativesdk doesn't RPROVIDE dbus-x11

dbus-x11 should not RPROVIDE dbus-x11 as this is incorrect and confuses
builds. This fixes the nativesdk case.

(From OE-Core rev: 09e3ebdb77c090ddc45fed7c6acc1591cd4b3595)

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

python-setuptools: Fix setuptools install libdir != /usr/lib

--install-lib= is required to be passed via distutils install
otherwise the install location gets set to whatever the python-native
location is.

(From OE-Core rev: d8ca8b21c6959e772fa1b437ce498b7759fae0ab)

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

package.bbclass: change RPATHs for nativesdk packages

Change binaries RPATHs, to include $ORIGIN, to make them relocatable.

(From OE-Core rev: f6362c5248914485fc8bf3c3f0ca6cc962bf753e)

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

relocatable.bbclass: split it up, to reuse code

Most of the code in relocatable.bbclass will be used for relocating the
SDK binaries. So, create another class chrpath.bbclass that will contain
the core of the relocatable.bbclass, so we can reuse it.

(From OE-Core rev: b50677b1641b201fd69942fd82a360907338234d)

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

populate_sdk_base.bbclass: create self-extracting archive

In order for the SDK to be relocatable, the user would need to call a
setup script to change the binaries acordingly. Having an auto-extracting
archive has the advantage of being more user friendly and the user does
not have to call the setup script separately after extracting the SDK.
It is called automatically.

(From OE-Core rev: d4e976849da845d595db7029d8089946ae16d982)

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

scripts: add script for relocating the SDK

This script will be embedded in the SDK tarball and will be called by
the SDK installer. It replaces the interpreter path in all binaries and
it also changes the ld.so.cache and SYSDIRS in dynamic loader.

(From OE-Core rev: 3d366f4953962566f33a3d77d65ed0bd2c48f922)

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

binutils: relocatable SDK: change PT_INTERP section size

This patch is needed so that all SDK binaries have the PT_INTERP section
size set to 4096 (max path size in Linux) in order to be able to parse
the binaries later and change the interpreter to the path we want.

(From OE-Core rev: 9bf657bbdfe7c1a9dff1cc121eab496e3e407197)

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

eglibc: relocatable SDK changes

Modifying the dynamic loader path in all binaries by the SDK installer
would not be possible because we cannot know in advance the SDK target
location. Hence, the PT_INTERP section size has been set to 4096 (which
is the maximum path lengh in Linux).

Also, for the dynamic loader SYSDIRS and ld.so.cache paths, special
4096 bytes sections were allocated so that we can easily parse the ELF
binary at install time and modify the paths.

(From OE-Core rev: 6be55d36333e0c81a4926332d036246d28d995c4)

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

kmod: Upgrade to upstream version 9.0

(From OE-Core rev: 9fcbd447e99af5228cccb6a84dd6398c61574caf)

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

send-pull-request: Make -a really imply -c

The switch statement does not fall through as the comment suggests.
This results in -a not implying -c as was intended.

Add the two lines from -c to -a to achieve the intended behavior.

(From OE-Core rev: 319d2cc8e60450106b665ee4172e2e6dc90e14b5)

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

qt4: fix qt4-native build failure on SLED-11

This issue is that with C++ compiler process an old version of kernel
header file, coincidently that file has a variable named 'new':

* 'embedded/qkbdtty_qws.cpp' include 'linux/vt.h';
* '/usr/include/linux/vt.h' has below code on SLED-11.x:
    + unsigned int new;

On mostly hosts it has been changed to: new -> newev.

[YOCTO #2845]

(From OE-Core rev: bb4b71552da0d1f4396955f35db5819f88a7d2ae)

Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sip.bbclass: fix sip-native dependency

(From OE-Core rev: 5f1af539b3d34e6689a69d09c69d9099eee983d4)

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

python-dbus: updated to 1.1.1

(From OE-Core rev: 191f06e8886209341a1822f5078fdf410dc86202)

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

watchdog: Upgrade to upstream v5.12

sourceforge.net link is broken at the time of this patch,
so I added a mirror.
License changed to GPLv2+

(From OE-Core rev: 1452b5e60eff29e7423ecdd51d2d942e88dcc263)

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

usbutils: Upgrade usbutils to upstream v006

(From OE-Core rev: 375cddcc2241016a3114d707b1938c2190642a4f)

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

u-boot.inc: fix regexp used when ld-is-gold

* with old regexp it wasn't reentrant, causing stuff like this:
  -LD     = $(CROSS_COMPILE)ld
  +LD     = $(CROSS_COMPILE)ld.bfd.bfd
  -LDR    = $(CROSS_COMPILE)ldr
  +LDR    = $(CROSS_COMPILE)ld.bfd.bfdr
  when do_compile was reexecuted, also breaking LDR variable and
  http://git.yoctoproject.org/cgit/cgit.cgi/meta-ti/tree/recipes-bsp/u-boot/u-boot/0001-config-Always-use-GNU-ld.patch
  when it was used together with u-boot.inc from oe-core (meta-ti is using own u-boot.inc)

* This patch is also better solution to ld-is-gold problem then regexp in metadata.

(From OE-Core rev: f78044f85ab1a0acce852a7032fc0c81285cd4c1)

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

gcc-configure-target: Set native-system-header-dir for target gcc

Without this set, in the case host=target (which it does for on-target
gcc), the native header directory is set to the target sysroot with
no prefix. This means it would look for sdt.h on the build system
instead of in the target headers and this can lead to build failures
and is host contamination of the build.

The correct fix is to explicitly set the native header directory to
the correct location and then the headers get detected correctly.

(From OE-Core rev: 404d2d490fc347203e89d274530c17fb5f0aa20f)

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

core-image.bbclass: Fix typo in comment, "IMAGE_FEAETURES"

(From OE-Core rev: af274b1b266b5aa661823bd702eacc62cc3aefff)

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

createrepo 0.4.11: create wrapper for python script

The native python is installed to
${STAGING_BINDIR_NATIVE}/python-native/python, so we should use wrapper
for the python script.

[YOCTO #2822]

(From OE-Core rev: 1e96a8460b93ba9c0c9a6c4ea4ffd6a19302831c)

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

createrepo 0.4.11: add rpm-createsolvedb.py

Move scripts/rpm-createsolvedb.py to
meta/recipes-support/createrepo/createrepo/ since we should wrap it to
use the native python.

[YOCTO #2822]

(From OE-Core rev: 72d673bef385e756bd858f9eca7fe419efaceb39)

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

imagetest-qemu.bbclass: Fix whitespace issues

(From OE-Core rev: 674d65c94478a1ff33ef9d9c1e49f677091301f4)

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

package_rpm.bbclass: Accomodate dash when using arrays

we were assigning local variable to an array coming through
positional arguments. local is a non posix contruct thats
also supported by dash luckily but operates differently in this
case it exapnds the array before assignment. so

local pkgs="$@"

turns into pkgs=locale-base-en-us locale-base-en-gb

ant we see errors

run.do_rootfs.25593: 932: local: locale-base-en-gb: bad variable name

So lets not use defining and assigning local in one go
first define a local and then the assignment

(From OE-Core rev: 4029ce1a15e1dbd374444ee77ccf2a915e463b7b)

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

scripts/oe-pkgdata-util: exclude unpackaged in glob output

Check for .packaged file and exclude packages if this file does not
exist - this avoids attempting to install empty packages during
complementary package installation within do_rootfs that didn't end up
being created (and failing as a result).

(From OE-Core rev: 4a85d8a4026cf1d1603513ed9780f80c603cc611)

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

combo-layer: ensure init works with split local config

If the local configuration is already split out, ensure the init action
works properly and does not error in the case that last_revision is not
set. Additionally, if the local configuration is within the repository,
prevent it from being committed and add it to .gitignore.

(From OE-Core rev: de339b0cb201035e27df1128ccf526937b8325ec)

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

dbus: include dbus-launch in the main dbus package

Followed suggestions from Bugz 2261:

2) make the virtual/libx11 DEPENDS conditional based on the x11 distro feature.
This makes the build dependencies reflect the feature list.

3) remove dbus-x11, meaning that dbus-launch with its potential X11 dependency
is now back in dbus where is belongs.

4) make dbus provide dbus-x11, for compatibility.

Fixes [Yocto #2261]

(From OE-Core rev: c02364f36e228835ea5d7fd4e1d347fd451f8544)

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

glib-2.0-nativesdk: Fix build on ubuntu 11.10

When you have a distro which has older version of
glib-2.0 installed then it does not have compile_resources
installed which essentially glib-2.0 nativesdk is
looking for on build host. So like in a cross build
we disable the configure check which will force it
to build an internal copy and use it.

This problem may not be seen on newer distros like
ubuntu 12.04 since they have newer version of glib-2.0
installed

(From OE-Core rev: 69bd94173294c1ae2cf5e300ada7116e99a5b0c7)

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

package.bbclass: fix TypeError in runstrip

* some packages have .ko files which are not elf, without this change
  it fails with TypeError, with this change only runstip fails and
  reports where:
  ERROR: runstrip: ''arm-oe-linux-gnueabi-strip'  '/OE/shr-core/tmp-eglibc/work/armv4t-oe-linux-gnueabi/emacs-23.4-r0/package/usr/share/emacs/23.4/etc/tutorials/TUTORIAL.ko'' strip command failed

(From OE-Core rev: a834ab8a6d53cae504fa112a89bab93d726539ec)

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

slang: change char type to signed char in macros

[YOCTO 2726]

The patch of converting char type to signed char is only applied to arm,
in fact it should be applied to powerpc too since powerpc has the same
assumption for char type. To x86 and mips, this patch has not impact.

(From OE-Core rev: 8c6a9bd7b4fea62b5765caecad660d372379e412)

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

scripts/bitbake: unbreak the git version comparison

With the current code, we're calling awk to do a floating point comparison
between '1.7.0.4' and '1.7.5' (on an ubuntu 10.04 LTS machine). These clearly
aren't proper floating point numbers, and the comparison is incorrect. It's
returning true for 1.7.0.4 >= 1.7.5.

Instead of using a floating point comparison for this, call out to python and
let it do it.

(From OE-Core rev: f28f6267271edbbef16caec323e9ba76e2216723)

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

trace-cmd: Update for python-native changes

Add pythonnative to the inherits list

(From OE-Core rev: b3ef5a986e3bc1a29ba2b4a25d063a0267e2f72f)

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

lib/oe/lsb.py: Map unknown distributions to 'Unknown'

(From OE-Core rev: e12df2ca5c71ad672ebbfefbc47cabbac3333d61)

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

combo-layer: allow splitting out local config

Allow splitting the local parts of the configuration (mostly
local_repo_dir and last_revision, although there is no limitation) to
a side-by-side -local.conf file, with component sections optionally
tagged with the combo layer branch name. This effectively allows you to:

 * avoid polluting the history by committing the updated last revision
   to the combo repository for every update
 * avoid putting local repo paths into the combo repository
 * manage multiple branches of the combo repository whilst avoiding the
   possibility of mixing the configuration for one branch with another.

An example split configuration (note, values may be artificial):

------------------- combo-layer.conf -------------------
[bitbake]
src_uri = git://git.openembedded.org/bitbake
dest_dir = bitbake
hook = scripts/combo-layer-hook-default.sh

[oe-core]
src_uri = git://git.openembedded.org/openembedded-core
dest_dir = .
hook = scripts/combo-layer-hook-default.sh
--------------------------------------------------------

---------------- combo-layer-local.conf ----------------
[bitbake]
local_repo_dir = ../repos/bitbake

[oe-core]
local_repo_dir = ../repos/oe-core

[bitbake|master]
branch = master
last_revision = db689a99beffea1a285cdfc74a58fe73f1666987

[oe-core|master]
branch = master
last_revision = 121a1499a81706366acc0081272a6bff634d4d62

[bitbake|denzil]
branch = 1.12
last_revision = 24b631acdaa143a4de39c6e1328849660c66f219

[oe-core|denzil]
branch = denzil
last_revision = 741146fa90f28f7ce8d82ee7f7e254872d519724
--------------------------------------------------------

It is assumed that the local config file will be added to .gitignore.

(From OE-Core rev: f0065d7a6973628803a17c57f2265512aba3234c)

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

combo-layer: check that last_revision is valid

If the user edits the configuration file by hand and sets last_revision,
we need to ensure that the revision is valid and on the specified
branch.

(From OE-Core rev: 05382932257257247b8c18bc06e9c0039d134d06)

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

combo-layer: improve patch list handling and output

* Ignore blank lines in patch list
* Don't fail in interactive mode if patch list is deleted
* Show patch counter
* Show relative path for patches
* Print headings before applying patch list for each component

Also change to using a "with" block to read the patch list so it gets
closed properly when we're finished.

Fixes [YOCTO #2455].

(From OE-Core rev: 65461d7c35fdadb5b008052798731dce19ed187f)

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

combo-layer: drop to a shell when apply fails during update

If applying a patch fails during the update process, drop to a shell
instead of exiting; at that point the user can manually apply the patch,
do nothing and "exit" to skip it, or "exit 1" to abort the process.

(From OE-Core rev: c82b28982c4f630c130c827a7da3ac0454cd93b6)

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

combo-layer: ignore patch-* temp directories in dirty check

Make the dirty repo check somewhat less strict by ignoring old
patch directories created by this tool.

(From OE-Core rev: fea477ac55e2555c5bb0aad36db641aaa27aa915)

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

combo-layer: allow component pull to be done separately

* Add a -n option to disable component repo pull during update
* Add a 'pull' action to pull the component repos only

(From OE-Core rev: 61983b2191253b24117b63f586d5aac00c7eb48e)

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

combo-layer: remove &> bashism

&> does not work with dash - use > xxxx 2>&1 instead.

(From OE-Core rev: 5582cc58e42402c4680877e669d8c1ce058e7098)

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

qt4: make font packaging specific to embedded version

Qt 4's lib/fonts files are intended only for installation with
Qt Embedded [1] so only install and package them with the embedded
version.

Fixes [YOCTO #2809].

[1] http://lists.qt-project.org/pipermail/interest/2012-July/003062.html

(From OE-Core rev: 64369219e3e03cacfaef8df6e3eec507140bdd7d)

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

qt4: allow recipes building commercial edition

To enable building the commercial edition of Qt (through additional
recipes that are *not* provided by OE-Core) we need to tweak a few
things:

* Don't make recipes that inherit qt4x11.bbclass depend on qt4-x11-free
  - instead add qt4-x11 to DEPENDS and then have qt4-x11-free include
  this in its PROVIDES. A commercial equivalent recipe should do the
  same.
* Add a QT_LICENSE_FILE variable that can be used to specify the license
  file required by the commercial edition.
* Add a QT_LICENSE_FLAGS variable which the recipe can set to select the
  license option being used. The default of "-opensource" retains the
  current behaviour; a commercial recipe should set it to "-commercial".

Fixes [YOCTO #2505].

(From OE-Core rev: 2be0058fc4acddab611637656183accd052b40eb)

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

classes/populate_sdk_rpm: add depends on rpmresolve-native

As of OE-Core revision 72d1048a8381fa4a8c4c0d082047536727b4be47,
rpmresolve is now required for complementary package installation during
do_populate_sdk.

(From OE-Core rev: 8a71c1a507289d38a0f675584ab206abfa9aad9a)

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

classes/populate_sdk_base: don't break existing do_populate_sdk depends

We set do_populate_sdk[depends] in populate_sdk_*.bbclass, but since
these are inherited at the top of populate_sdk_base.bbclass, those
values are wiped out by using = at the end of the latter class. Use +=
to avoid this.

(From OE-Core rev: a87b5d6cab34579cf3abe03f4663d9a72e218397)

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

kernel.bbclass: Copy bounds.h only if it exists, needed for 2.6.x.

Linux 2.6.x kernels did not (all) have the bounds.h file, so copy
only iff exists.

(From OE-Core rev: 02ac0d1b65389e1779d5f95047f761d7a82ef7a4)

Signed-off-by: Leon Woestenberg <leon@sidebranch.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libxcb: Update for python-native changes

Add pythonnative to the inherits list

(From OE-Core rev: ab0187c13b2b0a041bf3d98c3a53bd3f45a624de)

Signed-off-by: Morgan Little <morgan.little@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xcb-proto: Update for python-native changes

Add pythonnative to the inherits list

(From OE-Core rev: 529346325b5df7bf316823f98f7332be4d019215)

Signed-off-by: Morgan Little <morgan.little@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libdrm: add DEFAULT_PREFERENCE = -1 to git recipe

* recipe from tarball should be default version

(From OE-Core rev: ad42ff66d8a1e57dc8c43b936235eca6bd75b79a)

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

socat 1.7.2: add readline dependency

[YOCTO #2844]

socat has dependency on readline

(From OE-Core rev: dc84ac6bc1c496501fe3cd4b4dfbc1b7ed973663)

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

guile:meta/Makefile change install-data-hook to install-exec-hook

In guile meta/Makefile, what install-data-hook do
must be run after install-binSCRIPTS.so it's a exec hook, not a data hook.
If keep these exec-hook in data-hook,when we run make with multi-jobbing ,
install-data-hook and install-binSCRIPTS will be separately
processed in different thread, no any dependence.
that will cause error sometimes.

[YOCTO #2796]

(From OE-Core rev: 018d5b87bdf00e85fe34e6276bd99e118843d704)

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

qt-mobility, xserver-xorg, ofono, bluez4, gst-plugins-good: bump PR to rebuild after libudev0 soname change to libudev1

* this patch depends on http://patchwork.openembedded.org/patch/32085/

(From OE-Core rev: 2792569605b50bdad8b2b8398c823dd08006daf1)

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

bison: move remove-gets.patch to BASE_SRC_URI, it's needed for bison-native too if host has (e)glibc-2.16

(From OE-Core rev: 9009c29fff9ca8619d5195fdf1ce09479c87b754)

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

classes/package: ensure pkgdata is cleaned out first

If do_package needs to be re-run and packages are removed/renamed as a
result, we want the corresponding stale pkgdata files to no longer be
present; for that we need to ensure all of the old pkgdata files are
cleaned out prior to creating the new ones. This is particularly
important now we rely on pkgdata more heavily (for complementary
packages, license manifest, etc.).

(From OE-Core rev: 5dee47b476a6f715ab02af5e758441a2a934eea3)

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

pseudo.inc/pseudo_1.4.bb: update pseudo to 1.4

This update replaces the half-baked --arch logic with the use
of $CFLAGS to pick compiler flags, on the grounds that it makes
a lot more sense for the build system to pick flags than for
pseudo to try to guess what they should be; this should allow
pseudo to at least compile for targets, and possibly run on
them.

This doesn't solve the problem of guessing how to forcibly
build the 32-bit variant on hosts, because we really don't
have a general solution for that. There's no idiom for "given
this set of compiler flags and this architecture, what flags
would you use to request a 32-bit compile instead?" So we
basically ignore that for now. If someone comes along trying
to use the build system to build pseudo-native on a 64-bit
host that also supports 32-bit binaries and isn't x86, we
will revisit this.

(From OE-Core rev: 711fcb4f10e2cefd7ff6e1921d87d1cad840d0c8)

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

usbutils 0.91: Fix NULL pointer crash.

[YOCTO #2847]

Before use usbbuslist, we should check if it is valid.

(From OE-Core rev: fc6c6ce73105518f4106da469ed81d53d860a290)

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

libpciaccess: lpci access does not depend on X

So let's remove the X dependencies from DEPENDS and white list it in
xorg-lib-common.inc.

(From OE-Core rev: e82b4fe82689c760c4910ebfc9aed6e3d9e1a2a1)

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

libdrm: Bump libdrm_git to 2.4.37

(From OE-Core rev: 9325ecf5698071fbe962d28377d30ab6f39b22ac)

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

libdrm: Put the license md5sum in the common include

With this, libdrm_git now does have LIC_FILES_CHKSUM as well.

(From OE-Core rev: ff86dba803ea4a8c66be57113e33de8ac92a21d1)

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

libdrm: libdrm does not depend on xcb any more

(From OE-Core rev: ed49240504e6ebbe294ba3b7c20c13308bb2768e)

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

drm: move to PR/INC_PR, starting at r1 as libdrm_git was at r1

(From OE-Core rev: a5f227d1325572d31f55bb8e8f0e8226687253bb)

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

ppp: bump PR to rebuild against libpcap1 instead libpcap

(From OE-Core rev: 62ca78c78504bf0c7441da7777d5a7834bfded37)

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

buildhistory: improve performance of image info collection

Reduce the number of calls to the packaging tool, especially in the case
of rpm, using helper utilities to gather the required information more
efficiently where possible.

(From OE-Core rev: d0b8a98c5b46c305afd389fc862b3bf0c6f1eaab)

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

classes/image: add staticdev-pkgs IMAGE_FEATURES feature

Add a staticdev-pkgs feature that can be added to IMAGE_FEATURES in
order to install all staticdev packages.

Fixes [YOCTO #2531].

(From OE-Core rev: 3ba9c0757eb51a0bb5873f4faae023587a33cc1d)

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

Rework installation of dev, dbg, doc, and locale packages

Use a similar mechanism that was previously used to install locales at
rootfs generation time to install other "complementary" packages (e.g.
*-dev packages) - i.e. install all of the explicitly requested packages
and their dependencies, then get a list of the packages that were
installed, and use that list to install the complementary packages. This
has been implemented by using a list of globs which should make it
easier to extend in future.

The previous locale package installation code assumed that the locale
packages did not have any dependencies that were not already installed;
now that we are installing non-locale packages this is no longer
correct. In practice only the rpm backend actually made use of this
assumption, so it needed to be changed to call into the existing package
backend code to do the complementary package installation rather than
calling rpm directly.

This fixes the doc-pkgs IMAGE_FEATURES feature to work correctly, and
also ensures that all dev/dbg packages get installed for
dev-pkgs/dbg-pkgs respectively even if the dependency chains between
those packages was not ensuring that already.

The code has also been adapted to work correctly with the new
SDK-from-image functionality. To that end, an SDKIMAGE_FEATURES variable
has been added to allow specifying what extra image features should go
into the SDK (extra, because by virtue of installing all of the packages
in the image into the target part of the SDK, we already include all of
IMAGE_FEATURES) with a default value of "dev-pkgs dbg-pkgs".

Fixes [YOCTO #2614].

(From OE-Core rev: 72d1048a8381fa4a8c4c0d082047536727b4be47)

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

classes/rootfs_rpm: improve speed of RPM rootfs construction

Improve the performance of the RPM backend during do_rootfs by
performing most of the package name to file resolution in a separate
utility written in C, processing the entire list of packages at once
rather than running rpm on the command line which loads the RPM database
for every package.

(From OE-Core rev: 9135d351ba7cb21e50239d2b310565680bd4fdca)

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

gconf: Clean up introspection disabling by using gobject-introspection-stub

(From OE-Core rev: 3861dc6f52c8c3abe925302dadba15d90efbd6b5)

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

polkit/vte/gupnp/gssdp: Replace introspection hacks with gobject-introspection-stub

(From OE-Core rev: dcd076a727de57677efc7130793fa0cda03bb146)

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

gtk-doc-stub: Add PROVIDES for gobject-introspection-stub and fix unpackaged files

(From OE-Core rev: 3c0c1f9b826478af075fbae7b6366726c454702c)

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

gobject-introspection: Drop the recipe, it has never worked and if it does cross compile, we can add it back

(From OE-Core rev: cc95d5472936f24f8375f4366062606595dd8124)

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

eds-tools: Drop this recipe, its inapproriate in OE-Core

(From OE-Core rev: 302a1251be7ec81fc1d270d9d7b6db6d140d4692)

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

world-broken.inc: Remove clutter-box2d from the exclude list, it works

(From OE-Core rev: f7a156ef72b4aa605d21c1ab148f0da4ddd924c8)

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

package.bbclass/eglibc-locale: Ensure the correct PN is used for locale packages

(From OE-Core rev: fc4b138948656d482603b80f8c8e86fdbb185df4)

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

sstate.bbclass: Add a string representing the lsb release to native/cross sstate

This patch adds the lsb name and revision to the path used for sstate files.
This means that reuse of sstate files between different distributions is restricted
by default. The behaviour can be configured using mirror urls, for example:

SSTATE_MIRRORS = "file://Ubuntu-11.10/(.*) file://Ubuntu/\1 \n"

would map Ubuntu 11.10 to a more generic "Ubuntu" named sstate feed.

Usually, more modern distros have increased libc versions for example
so whilst more older native/cross sstate packages will usually work on newer
distros, the opposite is not true. This patch allows development of policy
to better handle this although no default policy is currently being used.

(From OE-Core rev: 9360fc318cec5c873b17f59b817dd6312c81ee27)

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

sstate: Add a two character subdirectory to the sstate directory layout

Currently all sstate files are placed into one directory. This does not scale and
causes a variety of filesystem issues. This patch adds a two character subdirectory
to the layout (based on the first two characters of the hash) so that files
can be split into several directories.

This should help performance of sstate in most cases.

(From OE-Core rev: 5d45c7b025f6635b2232d7bf92b7c1aba350396b)

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

base.bbclass: Set the NATIVELSBSTRING variable to represent the distro we're running on

This can then be used by the sstate code to mark native and cross packages
as being specific to a given distro.

(From OE-Core rev: 8556eb98be8bd9f02ee11a0d8a889c1895c86460)

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

lib/oe/lsb: Add basic LSB functions

This code was written by Christopher Larson <chris_larson@mentor.com> and
allows generation of the LSB release data based upon the lsb_release
command. It also includes a helper function to generate a string
representing a given distribution.

(From OE-Core rev: 458bdfe1cc4872e901ea3c38f497bfea6cf4c8cb)

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

bzip2-native: handling native path issue

follow Richard's approach, modify bzip2-native recipe to install bzip2 package
binaries to STAGING_BINDIR_NATIVE/bzip2-native to avoid problems when other
package is doing upack to reference bzip2. libbz2.so* still installs to
STAGING_LIBDIR_NATIVE.

change python-native to depends on bzip2-replacement-native instead of
bzip2-full-native and add EXTRANATIVEPATH for bzip2-native.

Didn't add bzip2native.bbclass as python-native is the only user so far.

(From OE-Core rev: 5bf3541836457465aa76577bfb41cdf6316213c9)

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

image.bbclass: Fix a couple innocuous typoes, should cause no functional change

"error_promt" -> "error_prompt" changed in both places
"subfolers" -> "subfolders" since it's not referenced anyway

(From OE-Core rev: cc1f824778c023686b4ea75c64a182a138ff2267)

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/package: reduce dbg package dependencies

Make dbg package dependencies added via DEPCHAIN_POST less aggressive -
only add dependencies on dbg packages for shared library dependencies.
This avoids dragging in extraneous packages (such as eglibc-dbg forcing
bash-dbg to be installed) whilst preserving the ability to easily debug
into shared libraries in use by an application by just requesting the
installation of the single dbg package for that application.

For task recipes however we preserve the previous behaviour, since when
you install task-xxx-dbg you expect the dbg packages for every package
in the task to be installed. Unfortunately not all of our tasks inherit
from task.bbclass so we just use a name match - this should be tidied up
later.

Fixes [YOCTO #2599].

(From OE-Core rev: 352522d474cb75992d7865545b6fbe4e157a5f99)

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>

gtk-doc.bbclass: unbreak dependency-chain for native recipes

see [1] & [2] for discussion of this patch

[1] http://lists.linuxtogo.org/pipermail/openembedded-core/2012-July/026606.html
[2] http://lists.linuxtogo.org/pipermail/openembedded-core/2012-July/026611.html

(From OE-Core rev: f9ae930552bc5f1d59f207d4cd0e2b1b4f811dbc)

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

u-boot.inc: update linker arguments to pass --sysroot arg

If we are building from sstate-cache it's possible to be building
from another folder on another machine, therefore the linker requires
that a proper --sysroot is passed too it so it can find things like
libgcc.a and avoid errors such as:

| arm-poky-linux-gnueabi-gcc  -g  -O2  -fno-common -ffixed-r8 -msoft-float   -D__KERNEL__ -DCONFIG_SYS_TEXT_BASE=0x80008000 -I/local/yocto/upstream/label/ubuntu1204-64b/machine/beagleboard/poky/edison/tmp/work/beagleboard-poky-linux-gnueabi/u-boot-v2011.06+git5+b1af6f532e0d348b153d5c148369229d24af361a-r0/git/include -fno-builtin -ffreestanding -nostdinc -isystem /local/yocto/upstream/label/ubuntu1204-64b/machine/beagleboard/poky/edison/tmp/sysroots/x86_64-linux/usr/bin/armv7a-vfp-neon-poky-linux-gnueabi/../../lib/armv7a-vfp-neon-poky-linux-gnueabi/gcc/arm-poky-linux-gnueabi/4.6.3/include -pipe  -DCONFIG_ARM -D__ARM__ -marm  -mabi=aapcs-linux -mno-thumb-interwork -march=armv5 -Wall -Wstrict-prototypes -fno-stack-protector -fno-toplevel-reorder   -o hello_world.o hello_world.c -c
| arm-poky-linux-gnueabi-gcc  -g  -O2  -fno-common -ffixed-r8 -msoft-float   -D__KERNEL__ -DCONFIG_SYS_TEXT_BASE=0x80008000 -I/local/yocto/upstream/label/ubuntu1204-64b/machine/beagleboard/poky/edison/tmp/work/beagleboard-poky-linux-gnueabi/u-boot-v2011.06+git5+b1af6f532e0d348b153d5c148369229d24af361a-r0/git/include -fno-builtin -ffreestanding -nostdinc -isystem /local/yocto/upstream/label/ubuntu1204-64b/machine/beagleboard/poky/edison/tmp/sysroots/x86_64-linux/usr/bin/armv7a-vfp-neon-poky-linux-gnueabi/../../lib/armv7a-vfp-neon-poky-linux-gnueabi/gcc/arm-poky-linux-gnueabi/4.6.3/include -pipe  -DCONFIG_ARM -D__ARM__ -marm  -mabi=aapcs-linux -mno-thumb-interwork -march=armv5 -Wall -Wstrict-prototypes -fno-stack-protector -fno-toplevel-reorder   -o stubs.o stubs.c -c
| arm-poky-linux-gnueabi-ld  -r -o libstubs.o  stubs.o
| arm-poky-linux-gnueabi-ld -g -Ttext 0x80300000 \
|  -o hello_world -e hello_world hello_world.o libstubs.o \
|  -L. -lgcc
| arm-poky-linux-gnueabi-ld: cannot find -lgcc
| make[1]: *** [hello_world] Error 1

(From OE-Core rev: 923eb657f24a1feaa93172eda8f87252e86bfea2)

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

insserv: define do_install_virtclass-native()

[YOCTO #2812]

Since files installation dirs are DESTDIR/etc/, DESTDIR/sbin etc
in Makefile, not DESTDIR/${sysconfdir}/, DESTDIR/${sbindir}/, for
the native package, a prefix can not be inserted which is needed
when do the next step of install and do_populate_sysroot.

Now we define do_install_virtclass-native in which DESTDIR is
${D}/${STAGING_DIR_NATIVE}

(From OE-Core rev: bb80a5f6f383d874b314af4263695e5a6b2d84cb)

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

Update intltool to version 0.50.0

Uprev intltool to newer 0.50.0 version

(From OE-Core rev: c17d1f146e43a4acaad5de629c33404426870356)

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

bluez4: use PACKAGECONFIG to control 'pie'

This lets us control whether we build position independent executables. As, as
far as I know no distros are adding a 'pie' distro feature, the effect of this
commit for most will be to disable pie in bluez4. I think this is best to
ensure consistent behavior among the recipes with regard to this.

This also sidesteps a build failure I've seen with bluez4 failing due to
libudev.a not having been built fPIC.

(From OE-Core rev: 06b5d57e89f2dd20fb3f4cb9fd362e2d428f493d)

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>

bitbake.conf: override -cpu for qemu on ppc with SPE

We use -mtune flags which lets our application have
SPE intruction in them which is a good thing but when
running qemu especially in user mode these application
wont work with default powerpc emulation. Therefore we
have to add proper -cpu option to let qemu emulate for
a given cpu.

(From OE-Core rev: 695598415b4e9e17def7f04332bdbc0d24cc1efc)

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>

cmake: Upgrade recipes to 2.8.8

Copyright year change means a change in checksum
...

 CMake - Cross Platform Makefile Generator
-Copyright 2000-2009 Kitware, Inc., Insight Software Consortium
+Copyright 2000-2011 Kitware, Inc., Insight Software Consortium
 All rights reserved.

(From OE-Core rev: db6a69f132b3ad2bf1108eccd945123969347154)

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>

qemu: Fix illegal instruction errors on e500 emulation

I caught this when running user mode qemu for ppc/e500 applications
which had SPE instructions in them will abort with illegal instruction
error all the time.

The patch is already applied upstream we needed a backport into 0.15.x

(From OE-Core rev: febec229b27279345b756d2fd83f3766915fcd67)

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>

pkgconfig: Fix packaging issue

The pkg.m4 file should be packaged as part of the base package
instead of the -dev package, it had been in the base package, but
I seemed to have missed that it moved during my PACKAGES reorder,
and this caused the meta-toolchain to break

[YOCTO #1908]

(From OE-Core rev: d2dba9aebef323a57824b41d09991850c703be35)

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

eglibc: Add patch to add sys/resource.h to wait.h

This is needed for the older kernel perf tools to build correctly, as
noted in the patch new perf tools have sys/resource.h included directly.

(From OE-Core rev: 21a57b98ff5091a5563c00093081297a6c586208)

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

tcmode-default: Bump EGLIBC Version to 2.16

(From OE-Core rev: 08f57808b82599ed32fdc2fb9ec2c6491465c520)

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

usbutils: avoid dependency on bash

By virtue of having #!/bin/bash, the usb-devices script declared that it
requires bash, however manual examination, checkbashisms and tests with
dash and busybox show that it doesn't contain any bashisms, so change
the script's header and remove the RDEPENDS on bash.

Additionally, update-usbids.sh was manually checked for bashisms, run
through checkbashisms and tested with busybox (although it did not have
a bash-specific header).

(From OE-Core rev: 651b223c5fc93c7504e304e954b9ae4640ed47c6)

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

classes/package: do not add self-dependencies to dev/dbg packages

Avoid adding dependencies to dev and dbg packages on themselves via the
DEPCHAIN_* handling code. These don't actually cause any problems, but
they do complicate dependency graphs.

(From OE-Core rev: ecdf7874470cff24cb23b7fd2723096b15aead31)

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

guile: Add continuation marker to create_wrapper

This was causing the x86_64-linux-guile script to not have all the
environment bits needed

V2: Forgot PR Bump!

[YOCTO #2835]

(From OE-Core rev: be89b81bdad0489d43d33342243b5d1da76ec3c4)

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

base/perlnative/pythonnative/gzipnative: Add a new variable to enable easier handling of native path issues

If something is listed in ASSUME_PROVIDED but also manually built (like bzip2-native and
bzip2-replacement-native), we need to ensure that the installed binaries are only accessed
for things which list the provider in DEPENDS.

This patch adds a generic mechanism to handle this instead of everything reinventing the
wheel. EXTRANATIVEPATH += "perl-native" will ensure that ${STAGING_BINDIR_NATIVE}/perl-native
is added to PATH.

This means that:

a) Such providers should install to ${bindir}/xxx-native
b) Should PROVIDE += "xxxx-replacement-native"
c) That users should DEPEND on xxx-replacement-native and have EXTRANATIVEPATH += "xxx-native"

This patch enables this at the core level, the bzip2 recipe still needs adjusting to work
like this along with adjustment of the users of bzip2-replacement-native (python-native?).

(From OE-Core rev: 14b70cd222519e5bccaca955334787f123d9fc54)

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

staging.bbclass: Add support for invalidtasks parameter (see bitbake change)

(From OE-Core rev: d3ae2fc896fd866aac9f064dccbff971c324f43c)

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

bitbake: fetch/local: Allow preservation of path components in relative file:// urls

This enhances the fetcher to allow preservation of the path component in urls
like: file://xxx/yyy/somefile.patch.

(Bitbake rev: e49a656a499355a5c6e7eb00bf5b8f1795e8dddb)

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

bitbake: runqueue.py: Wipe out the stamp cache between setscene and main task execution

The stamp files can change during setscene and the cache should be cleared to
account for this.

(Bitbake rev: 5ec12f586a50fce675b268965b3dc487aaa96c43)

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