boost/libxslt: Enable nativesdk to unbreak <bitbake universe>

When run <make universe.fetchall>, the make will be broken because
nativesdk-boost and nativesdk-libxslt don't exist.
nativesdk-boost is depended by nativesdk-curlpp.
nativesdk-libxslt is depended by nativesdk-python-lxml.
In condition that meta-openembedded is compiled in.

Add nativesdk building for boost and libxslt.

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

wic: Add plugin for single partition disk

The wic plugin creates a disk image containig one ext2/3/4 partition.
No additional boot partition is required. Syslinux is installed into
the image. The target device is a legacy BIOS PC.

Purpose of this plugin:
Other avaliable plugins create a fat partition for /boot and an ext
partition for rootfs. Current linux-yocto kernel packages are not
compatible with this disk layout. The boot partition is not mounted
by default, hence the kernel is installed into rootfs and not into
boot partition. A kernel update ends up in a bricked device. The old
kernel which is still in boot likely does not even boot with updated
kernel modules from /. Even if the boot partition is mounted during
the kernel update the update will fail. The kernel package installs
a symbolic link which is not supported by the fat partition.
Creating just one ext partition for boot and rootfs solves all issues
related to package based kernel updates on the device.

The plugin depends on syslinux-nomtools a user space installer for
syslinux on ext filesystems.
Thanks to Robert Yang who implemented syslinux-nomtools and supported
the implementation of this plugin.

Signed-off-by: Adrian Freihofer <adrian.freihofer@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic tests: Move wic tests to oe-selftest infra

Modified existing wic test modules to work with oe-selftest
infrastructure.

$ oe-selftest --run-tests wic
...
test01_help (oeqa.selftest.wic.Wic)
Test wic --help ... ok
test02_createhelp (oeqa.selftest.wic.Wic)
Test wic create --help ... ok
test03_listhelp (oeqa.selftest.wic.Wic)
Test wic list --help ... ok
test04_build_image_name (oeqa.selftest.wic.Wic)
Test wic create directdisk --image-name core-image-minimal ... ok
test05_build_artifacts (oeqa.selftest.wic.Wic)
Test wic create directdisk providing all artifacts. ... ok

----------------------------------------------------------------------
Ran 5 tests in 6.434s

OK

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

wic tests: Add docstrings to test methods

Docstrings make nosetests output more clear:

Test wic create --help ... ok
Test wic --help ... ok
Test wic list --help ... ok
Test wic create directdisk providing all artifacts. ... ok
Test wic create directdisk --image-name core-image-minimal ... ok

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

wic tests: Add functional tests for directdisk

Added 2 simple functional test cases to check if
wic is able to produce images using directdisk.wks

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

qt4-embedded: add checking for AArch64 ilp32

1. Cross-compiling qt4-embedded and qt4-x11-free applications for AArch64 ilp32
   (both big and little endian) fails with below error,

(snip)
In file included from ../3rdparty/javascriptcore/JavaScriptCore/runtime/Collector.h:31:0,
                 from ../3rdparty/javascriptcore/JavaScriptCore/runtime/JSGlobalData.h:32,
                 from ../3rdparty/javascriptcore/JavaScriptCore/interpreter/CallFrame.h:26,
                 from ../3rdparty/javascriptcore/JavaScriptCore/runtime/ClassInfo.h:26,
                 from ../3rdparty/javascriptcore/JavaScriptCore/runtime/JSObject.h:27,
                 from ../3rdparty/javascriptcore/JavaScriptCore/API/JSCallbackConstructor.h:30,
                 from ../3rdparty/javascriptcore/JavaScriptCore/API/JSCallbackConstructor.cpp:27:
../3rdparty/javascriptcore/JavaScriptCore/wtf/StdLibExtras.h: In instantiation of 'TO QTWTF::bitwise_cast(FROM) [with TO = int; FROM = double]':
../3rdparty/javascriptcore/JavaScriptCore/runtime/JSImmediate.h:52:49: required from here
../3rdparty/javascriptcore/JavaScriptCore/wtf/StdLibExtras.h:63:1: error: size of array is negative
../3rdparty/javascriptcore/JavaScriptCore/wtf/StdLibExtras.h: In instantiation of 'TO QTWTF::bitwise_cast(FROM) [with TO = double; FROM = int]':
../3rdparty/javascriptcore/JavaScriptCore/runtime/JSImmediate.h:57:47: required from here
../3rdparty/javascriptcore/JavaScriptCore/wtf/StdLibExtras.h:63:1: error: size of array is negative
make[1]: *** [obj/release/JSCallbackConstructor.o] Error 1
-- CUT --

    the error is because of JSVALUE64 is being defined while compiling
    programs for AArch64 ilp32. This makes equal condition checking of
    intptr_t and double datatypes as false, and results in -1 (negative)
    as array size in bitwise_cast function.
    The patch defines JSVALUE64, if the programs are being compiled for
    AArch64 64bit NOT for ilp32.

2. qt4_arch.inc is modified to set system architecture as aarch64 if the
   TARGET_ARCH is other than aarch64. Without this modification, for
   TARGET_ARCH equal to aarch64_32, aarch64be, aarch64be_32; the system
   architecture is set as 'generic' and results in below error.

configure log:
(snip)
64-bit AMD 80x86 (x86_64)
'-opensource' is unsupported, using 'generic'
'x86_64' is supported
System architecture: 'generic'
Host architecture: 'x86_64'
-- CUT --

build error:
(snip)
In file included from ../../include/QtCore/qatomic_generic.h:1:0,
                 from ../../include/QtCore/../../src/corelib/arch/qatomic_arch.h:64,
                 from ../../include/QtCore/qatomic_arch.h:1,
                 from ../../include/QtCore/../../src/corelib/thread/qbasicatomic.h:227,
                 from ../../include/QtCore/qbasicatomic.h:1,
                 from ../../include/QtCore/../../src/corelib/thread/qatomic.h:46,
                 from ../../include/QtCore/qatomic.h:1,
                 from ../../include/QtCore/../../src/corelib/tools/qbytearray.h:45,
                 from ../../include/QtCore/qbytearray.h:1,
                 from ../../include/QtCore/../../src/corelib/tools/qstring.h:46,
                 from ../../include/QtCore/qstring.h:1,
                 from ../../include/QtCore/../../src/corelib/kernel/qobject.h:48,
                 from ../../include/QtCore/qobject.h:1,
                 from ../../include/QtCore/../../src/corelib/kernel/qcoreapplication.h:45,
                 from ../../include/QtCore/qcoreapplication.h:1,
                 from qdbusintegrator.cpp:42:
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h: In instantiation of 'bool QBasicAtomicPointer<T>::testAndSetOrdered(T*, T*) [with T = const QString]':
qdbusintegrator.cpp:76:1:   required from here
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h:197:88: error: invalid conversion from 'const void*' to 'void*' [-fpermissive]
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h:108:20: error: initializing argument 2 of 'bool QBasicAtomicPointer_testAndSetOrdered(void* volatile*, void*, void*)' [-fpermissive]
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h:197:88: error: invalid conversion from 'const void*' to 'void*' [-fpermissive]
../../include/QtCore/../../src/corelib/arch/qatomic_generic.h:108:20: error: initializing argument 3 of 'bool QBasicAtomicPointer_testAndSetOrdered(void* volatile*, void*, void*)' [-fpermissive]
make[1]: *** [.obj/release-shared-emb-auto/qdbusintegrator.o] Error 1
make[1]: *** Waiting for unfinished jobs....
-- CUT --

   So, select system architecture (via QT_ARCH) as aarch64 for
   AArch64 and its other variants

Signed-off-by: Krishnanjanappa, Jagadeesh <jagadeesh.krishnanjanappa@caviumnetworks.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libcap: Avoid passing "-e" to make

oe-core 51540b64f62234c145fc32cfa3fbbaaebbeece08 altered libcap.inc (at the
time) to append to EXTRA_OEMAKE rather than assign to it. The default value
for EXTRA_OEMAKE contains "-e". This means that the change caused "-e" to
be passed to make for the first time.

Unfortunately passing "-e" subtly changes the behaviour of libcap's
Make.Rules under recursive make when prefix="" (which it is for us since
we're using meta-micro.)

Without "-e" the prefix comes from the command line in both the parent and
submakes. This takes precedence over any attempt to reassign it with a
simple "=" operation so the headers are correctly installed in (empty
string)/include.

With "-e" the prefix still comes from the command line in the parent make
but from the environment in the submake. The attempt to assign it fails in
the parent make as before, but not in the submake so the headers are
installed incorrectly in /usr/include.

In all four cases the "ifdef prefix" else clause is executed.

So, let's assign EXTRA_OEMAKE in order to avoid using "-e" at all.

Signed-off-by: Mike Crowe <mac@mcrowe.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/utils: Allow ~ in bblayers

Bitbake can parse ~ in bblayer's paths.
Added this functionality to oeqa code.

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

libtasn1: Upgrade 4.0 -> 4.5

* Remove configure.ac-patch as unnecessary

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gnutls: Upgrade 3.3.12 -> 3.3.14

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oe-selftest: devtool: add some tests for devtool add -f

Tests the following aspects of "devtool add" with the -f option:
* Works for a git and https URI
* Extracts to the appropriate directory
* Uses the correct recipe file name with and without -V
* Sets SRC_URI and S appropriately with and without -V
* Sets SRCREV if rev= is specified in the URI

A lot of this functionality relies on "recipetool create" which
"devtool add" wraps around, so the associated behaviour of that is also
being tested here.

Refactor out the code to check a recipe for variable values / inherits
at the same time so we can use it to check the generated recipe.

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

recipetool: allow specifying SRCREV via rev= in URI

Provide a means to set SRCREV immediately by using rev= in the URI; if
it is specified then it is removed before setting SRC_URI and SRCREV is
set to the same value instead.

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

devtool: add: use the appropriate file naming and versioning for SCM recipes

* Recipes that fetch from git, svn or hg by OpenEmbedded convention
  should normally be named with this as a suffix, since PV is meant to
  be set appropriately within the recipe, so follow this. In order to
  make this work we need to be able to have the version independent from
  the file name, so add a -V option to recipetool create to allow this
  to be specified.
* If -V is specified on the devtool add command line, ensure at PV gets
  set to include this version.

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

devtool: add: add an option to fetch remote source

Add a -f/--fetch option to fetch a remote URI (into the already
specified source tree path) and set this as SRC_URI within the recipe.
This simply wraps around the existing functionality in recipetool.

Implements [YOCTO #7644].

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

recipetool: ensure git clone is standalone when extracting

If -x is specified and the specified URI was a git repository, we need
to ensure that the resulting clone is a sandalone and not one that has
pointers into the temporary fetch location or DL_DIR (since the git
fetcher does a local clone with -s). Split out the code from devtool
that already does this for "devtool modify -x" and reuse that.

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

recipetool: avoid second-level subdir when extracting

When -x was specified, we were getting the normal unpack subdirectory
which we don't really want - if there's only one subdirectory unpacked
then we should effectively copy just it to the extraction path, not as a
subdirectory under it.

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

package: Fix license exclusion packaging errors

Currently the license exclusion code removes packages from PACKAGES
pre population of the package directories. This means that the FILES
entries for some packages are not seen and invariably results in
packaging errors.

Instead, remove the packages from PACKAGES post population of the packages
so the usual FILES entries work as expected but the file are not placed
into any packages and no packages containing embargoed licenses are generated.

This avoids errors from gcc-runtime with GPLv3 exclusion like:

ERROR: QA Issue: gcc-runtime: Files/directories were installed but not shipped in any package:
  /usr/share
  /usr/src
  /usr/share/gcc-4.9.2
  /usr/share/gcc-4.9.2/python
Please set FILES such that these items are packaged. Alternatively if they are unneeded, avoid installing them or delete them within do_install. [installed-vs-shipped]

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

subversion: remove 1.6.X recipes

Removing the 1.6.X recipes, since there is a new version 1.8.X recipes,
and hope that all projects already upgraded their premirror caches to
use new format

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

pigz: set ALTERNATIVE for nativesdk package to empty

Set ALTERNATIVE for nativesdk package to empty to avoid warnings like below.

     WARNING: pigz: NOT adding alternative provide xxx/bin/gunzip: xxx/bin/pigz does not exist

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

elfutils: Security Advisory - CVE-2015-0255

Directory traversal vulnerability in the read_long_names function in
libelf/elf_begin.c in elfutils 0.152 and 0.161 allows remote attackers
to write to arbitrary files to the root directory via a / (slash) in a
crafted archive, as demonstrated using the ar program.

https://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-9447

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

linux-firmware: add NO_GENERIC_LICENSE for all licenses

An alternative approach with NO_GENERIC_LICENSE has been added
in license.bbclass to allow copying non-generic license,
add it for all firmware licenses.

Signed-off-by: Jackie Huang <jackie.huang@windriver.com>

Revert "common-licenses: Add common license files for linux-firmware"

This reverts commit 9021ee03e0f9bd36b0ce63fa29dd010e07241fdb

license.bbclass: allow copying license not in common licenses

Some package like linux-firmware has many licenses that aren't in any
way common, and new ones will be added from time to time, in order to
avoid adding bunch of such common license files that are only applicable
to a specific package, NO_GENERIC_LICENSE is added to allow copying license
not in common licenses, it should be used in the recipe as:

NO_GENERIC_LICENSE[<license_name>] = "<license file in fetched source>"

e.g.
NO_GENERIC_LICENSE[Firmware-Abilis] = "LICENCE.Abilis.txt"

Signed-off-by: Jackie Huang <jackie.huang@windriver.com>

pcmciautils: fix for parallel build

Fixed:
Compiling lex_config.c.
  src/lex_config.l:34:25: fatal error: yacc_config.h: No such file or directory

There was a patch for fixing the paralle issue before, so modify the
patch again.

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

bison: fix parallel issue

Fixed:
rm -f src/yacc src/yacc.tmp
echo '#! /bin/sh' >src/yacc.tmp
/bin/bash: src/yacc.tmp: No such file or directory
Makefile:6670: recipe for target 'src/yacc' failed

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

rpcbind: 0.2.2 -> 0.2.3

* Updated 0001-Avoid-use-of-glibc-sys-cdefs.h-header.patch
* Removed 0002-uclibc-rpcsvc-defines.patch since it is already in the
  source.

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

ltp: 20150119 -> 20150420

The COPYING's md5sum is changed by this commit:
commit 945f9c69af665044448b0eb6816656acc84fca77
Author: Ken Dreyer <kdreyer@redhat.com>
Date:   Mon Jan 26 14:02:46 2015 -0700

    update GPLv2 text in COPYING

    The FSF has issued a couple of tiny updates to the GPLv2. The main
    change is a new mailing address for the FSF headquarters.

    This license text was taken verbatim from
    https://www.gnu.org/licenses/old-licenses/gpl-2.0.txt

The main content are the same.

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

git: 2.3.5 -> 2.3.7

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

man-pages: 3.82 -> 3.83

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

dpkg: upgrade to 1.17.25

upgrade to fix two CVE defects: CVE-2014-8625 and CVE-2015-0840

http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-8625

Multiple format string vulnerabilities in the parse_error_msg
function in parsehelp.c in dpkg before 1.17.22 allow remote attackers
to cause a denial of service (crash) and possibly execute arbitrary
code via format string specifiers in the (1) package or (2)
architecture name.

http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2015-0840

The dpkg-source command in Debian dpkg before 1.16.16 and 1.17.x before
1.17.25 allows remote attackers to bypass signature verification
via a crafted Debian source control file (.dsc).

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

curl: Fix wrong assumption about sizeof off_t on largefile systems

This issue was reported on poky ml as well see
https://lists.yoctoproject.org/pipermail/poky/2013-December/009435.html

Change-Id: Iedf22467889893111fde0433e411fd0546a38546
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub2: fix initrd size restriction bug

The current grub2 fails on loading large initrd file (> 500M) since
the initrd size is added to the addr_min and causes the failure.
Fix it by picking a patch from grub2 upstream.

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

elfutils: enable fix-build-gcc-4.8.patch

The patch fixes a warning seen with gcc 4.8 (especially on ubuntu 13.10)

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

texinfo: remove enumerate_greater_than_ten.patch

It is a backport patch, and verified that the patch is in the source.

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

libaio: remove libaio-generic.patch

It is already in the source.

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

lttng-modules: remove bio-bvec-iter.patch

It is aready in the source.

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

openssl: remove 3 patches

Removed:
 - openssl-avoid-NULL-pointer-dereference-in-dh_pub_encode.patch
 - upgate-vegsion-script-for-1.0.2.patch
   Since they are already in the source.
 - make-targets.patch
   It removed test dir from DIRS, which is not needed any more since we
   need build it.

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

kmod: remove 0001-Makefile.am-fix-parallel-build-problem.patch

Confirmed with the author Qi, it isn't needed.

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

logrotate: remove logrotate-CVE-2011-1548.patch

It is a backport patch, and verified that the patch is in the source.

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

libxml2: remove libxml2-CVE-2014-3660.patch

It is a backport patch, and verified that the patch is in the source.

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

nspr: remove nspr-CVE-2014-1545.patch

It is a backport patch, and verified that the patch is in the source.

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

python3: remove sys_platform_is_now_always_linux2.patch

It is aready in the source.

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

python-numpy: remove 2 dangling patches

Removed:
- unbreak-assumptions.diff
  This patch changs the dir to /non-existant-dir, the source code has
  changed the dir to /deadir, so it is not needed any more.

- trycompile.diff
  There is no try_compile or try_run in numpy/core/setup.py any more, so
  assumed that it is not needed.

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

mesa: Upgrade 10.4.4 -> 10.5.4

* Remove patches that are no longer needed
* git build depends on python module mako. Inherit pythonnative
  for this
* source directory changed: default S is now correct in mesa recipe,
  but still needs to be set in mesa-gl

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

default-distrovars: Clean up gcc license issues

There are several entries here which are not needed with the
modern license handling code:

gcc-source - moved to direct handling in base.bbclass
             (due to version appended to the name)
libgcc - Listed as GPLv3 exception for its packages
libgcc-initial - Listed as GPLv3 exception
gcc-runtime - Indivisual packages listed as GPLv3 exception where appropriate

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

libgcc-initial: Correctly reflect libgcc LICENSE

In an effort to clean up some of the license handling, correctly set the
LICENSE of libgcc-initial to be the same as libgcc which has a GPLv3
exception.

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

gcc-shared-source: Add PV to PN

This means you can have one gcc version for some gcc recipes
(e.g. crosssdk/nativesdk) and another gcc version for target code.

Also remove the preferred version entry from the default toolchains
list since the version issue is now handled automatically.

We also need to specifically handle gcc-source in the license handling
code since expanding ${PV} in the base class isn't possible. Since
gcc-source doesn't generate any packages directly this shouldn't be
an issue and whitelisting in this way is easiest (and matches the
rest of the toolchain handling).

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

xserver-xorg: use GCC 5 patch from upstream

Upstream has already fixed the GCC 5 problem, so use the patch from upstream.

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

split_and_strip_files: regroup hardlinks to make build deterministic

Reverted 7c0fd561bad0250a00cef63e3d787573112a59cf

Created separate group of hardlinks for the files inside
the same package. This should prevent stripped files to be
populated outside of package directories.

This turns out not to be straightforward and has overlap with the
other hardlink handling code in this area. The code is condensed
into a more concise and documented form.

[Original patch from Ed with tweaks from RP]

[YOCTO #7586]

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

lib/oe/package: Improve strip subprocess handling

Currently if the strip process fails, we get a message but don't know why. This adds
code to show the return value and any error output.

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

lib/oeqa: Add sdk tests for gcc/perl/python

Enhance testing of the generated SDK tarballs by adding tests for
gcc/perl/python based on the existing runtime tests.

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

autotools: Fix find races on source directory

In a similar way to http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=aa1438b56f30515f9c31b306decef7f562dda81f
there are more find races in the autotools class.

For recipes with PACKAGES_remove = "${PN}", the find which removes
.la files can race against deletion of other directories in WORKDIR
e.g.:
find: '/home/autobuilder/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/build/tmp/work/qemux86_64-poky-linux/init-ifupdown/1.0-r7/sstate-build-populate_lic': No such file or directory
| WARNING: /home/autobuilder/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/build/tmp/work/qemux86_64-poky-linux/init-ifupdown/1.0-r7/temp/run.do_configure.6558:1 exit 1 from
| find /home/autobuilder/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/build/tmp/work/qemux86_64-poky-linux/init-ifupdown/1.0-r7 -name \*.la -delete

Fix the remaining races in the same way.

[YOCTO #7522]

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

testimage: Improve sdk handling of TEST_SUITES

Currently TEST_SUITES is used for both target image and sdk versions which
can be confusing. This introduces TEST_SUITES_SDK for the sdk version of
the code so that the different test sets can be specified independently.

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

gpgme: Fix to use pkg-config in .m4 files

pkg-config is less error prone than -config files so switch to
using it (we already do for most of the rest of the gpg stack).

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

oeqa/targetbuild: Use DL_DIR for downloads if present

This isn't a test of wget so if the files we need are present in DL_DIR,
use them from there and save a bit of speed/bandwidth and skip the wget.

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

sstate: Use SSTATE_DIR for FILESPATH

FILESPATH was only being overridden in one fetch location, it should be
equally handled in both.

Also use SSTATE_DIR as FILESPATH so that mirror urls which do remapping
can search the local SSTATE_DIR for other paths.

Also ensure that MIRRORS is removed in both locations, previously
it was only unset in one but both codepaths should be consistent.

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

testimage: Fix sdk test manifest handling

Fix SDK_MANIFEST -> SDK_TARGET_MANIFEST and add support for host
version too which is useful in SDK QA tests.

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

libxfont: Security Advisory - libxfont - CVE-2015-1804

bdfReadCharacters: ensure metrics fit into xCharInfo struct

We use 32-bit ints to read from the bdf file, but then try to stick
into a 16-bit int in the xCharInfo struct, so make sure they won't
overflow that range.

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

libxfont: Security Advisory - libxfont - CVE-2015-1803

bdfReadCharacters: bailout if a char's bitmap cannot be read

Previously would charge on ahead with a NULL pointer in ci->bits, and
then crash later in FontCharInkMetrics() trying to access the bits.

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

libxfont: Security Advisory - libxfont - CVE-2015-1802

bdfReadProperties: property count needs range check

Avoid integer overflow or underflow when allocating memory arrays
by multiplying the number of properties reported for a BDF font.

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

kern-tools: unify meta directory detection

It is possible that recipe specific tasks, or build processes drop
files into the kernel source directory. These files can cause problems
with the meta data detection in the kern-tools.

With this change, we have a single unified meta data detection routine,
that logs the result in a new file ".metadir", which subsequent scripts
can find, and use, thereby avoid repeating the same check many times.

We also enhance the check to look for a sentinel file in a proper meta
directory, to avoid false positives when an unexpected kernel process
leaves an uncommitted directory in the kernel dir.

[YOCTO: #7441]

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

libarchive: Security Advisory - libarchive - CVE-2015-2304

libarchive: Updated libarchive packages fix security vulnerability

Alexander Cherepanov discovered that bsdcpio, an implementation of the "cpio"
program part of the libarchive project, is susceptible to a directory
traversal vulnerability via absolute paths.

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

less: fix CVE-2014-9488

An out of bounds read access in the UTF-8 decoding can be triggered with
a malformed file in the tool less. The access happens in the function
is_utf8_well_formed due to a truncated multibyte character in the sample
file.

The bug does not crash less, it can only be made visible by running less
with valgrind or compiling it with Address Sanitizer.

Version 475 of less contains a fix for this issue. The file version.c
contains some entry mentioning this issue (without any credit):

 - v475 3/2/15 Fix possible buffer overrun with invalid UTF-8

The fix is in the file line.c. We derive this patch from:

https://blog.fuzzing-project.org/3-less-out-of-bounds-read-access-TFPA-0022014.html

Thank Claire Robinson for validating it on Mageia 4 i586. Refer to:

https://bugs.mageia.org/show_bug.cgi?id=15567

Signed-off-by: Junling Zheng <zhengjunling@huawei.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

rsync: backport a patch to fix CVE-2014-9512

rsync 3.1.1 allows remote attackers to write to arbitrary files via a symlink
attack on a file in the synchronization path.

Backport Complain-if-an-inc-recursive-path-is-not-right-for-i.patch to fix it

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

kernel-devsrc: depends on virtual/kernel:do_install

The linux-yocto.inc may remove the meta dir:
do_install_append(){
if [ -n "${KMETA}" ]; then
rm -rf ${STAGING_KERNEL_DIR}/${KMETA}
fi
}

Which may cause the error:
[snip]
find: `./meta/cfg/kernel-cache/bsp/altera-socfpga/0073-FogBugz-116676-Align-clk.c-with-kernel.org.patch': No such file or directory
find: `./meta/cfg/kernel-cache/bsp/altera-socfpga/0047-FogBugz-90657-Fix-SD-MMC-driver-for-VT.patch': No such file or directory
find: `./meta/cfg/kernel-cache/bsp/altera-socfpga/0006-spi-qspi-cadence-Add-spi-and-qspi-driver.patch': No such file or directory
[snip]
cpio: ./meta/scripts/kgit-config-cleaner: Cannot stat: No such file or directory
cpio: ./meta/scripts/kgit-s2q: Cannot stat: No such file or directory
cpio: ./meta/scripts/kgit-clean: Cannot stat: No such file or directory
[snip]

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

gnu-efi: fix parallel issue

Fixed:
Assembler messages:
Fatal error: can't create runtime/rtlock.o: No such file or directory
Assembler messages:
Fatal error: can't create runtime/rtdata.o: No such file or directory
Assembler messages:
Fatal error: can't create runtime/vm.o: No such file or directory
Assembler messages:
Fatal error: can't create runtime/efirtlib.o: No such file or directory

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

icecc.bbclass: blacklist libgcc-initial

Building libgcc-initial with IceCC enabled can fail with the CPP sanity
check error in the following case (using ARM for example):
* sysroot contains cross gcc built for another ARM variant
* sysroot contains initial cross gcc built to suit the target machine
* bitbake tries to configure libgcc-initial
* libgcc-initial calls icecc wrapper
* icecc wrapper calls non-initial cross gcc via the full path
* non-initial cross gcc looks for the headers in the wrong place
* BOOM

Signed-off-by: Dmitry Eremin-Solenikov <dmitry_eremin@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libunwind: don't use gold for mips

The gold doesn't work for mips or mips64:
configure:3867: checking whether the C compiler works
configure:3889: mips-poky-linux-gcc  -meb -mabi=32 -mhard-float -march=mips32r2 --sysroot=/buildarea/lyang1/test_uni/tmp/sysroots/qemumips  -O2 -pipe -g -feliminate-unused-debug-types   -Wl,-O1  -Wl,--as-needed  -fuse-ld=gold conftest.c  >&5
collect2: fatal error: cannot find 'ld'

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

mesa: respect x11 in PACKAGECONFIG not DISTRO_FEATURES

* default value of PACKAGECONFIG is set from DISTRO_FEATURES,
  but in the end we should respect what user sets in PACKAGECONFIG

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

udev-cache: improve error handling

If an error occurs while the udev cache is being populated, the system
is left in a state where udev is stopped.  Remedy this with a clean up
function to restart udev and remove any intermediate files.

Signed-off-by: Ken Sharp <ken.sharp@ni.com>
Reviewed-by: Ben Shelton <ben.shelton@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev-cache: Remove unnecessary tar read from stdin

xargs already formats the inputs to tar correctly, so the
'-T -' argument to tar is unnecessary.

Signed-off-by: Ken Sharp <ken.sharp@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

alsa-utils: Remove double dependency on udev

"udev" is already added when PACKAGECONFIG includes it, so it should
not be in the DEPENDS list on itself.

This caused udev to be built for systems that don't use it.

Signed-off-by: Mike Looijmans <mike.looijmans@topic.nl>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

subversion: upgrade it from 1.8.11 to 1.8.13

upgrade to fix two CVE defects: CVE-2015-0248 and CVE-2015-0251

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

populate_sdk_base: also record the host part into manifest

For now, we can create manifest for the target part for SDK.
I think it's nice to have a place for users to look into to examine contents
of the host part of SDK.

This also affects uninative-tarball and buildtools-tarball as they
inherit populate_sdk.bbclass. After this change, we could have a manifest file
created in the deploy directory containing a list of packages used to
construct them.

[YOCTO #7604]

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

ptest: add recommends to ptest-runner

If you're installing ptest packages into an image there's a good chance
you want to run the tests.  Assist with this by recommending
ptest-runner in the -ptest packages.

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

mkefidisk.sh: be more explicit with device error

* Specify whether the device can't be found or can't be written to
* Give a hint to use sudo

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

mkefidisk.sh: fix hanging on non-writeable device

If cleanup() is called early on, as happens when the device isn't
writeable, then none of the mount point variables are set; thus the
script was calling grep with only one argument and appeared to hang
since it was waiting for input on stdin.

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

mkefidisk.sh: use script mode when running parted

This avoids parted showing prompts and thus effectively hanging the
script in in the case of initially malformed disks.

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

devtool: update-recipe: check if source tree is a git repository

If you've done "devtool add" (or "devtool modify" without -x) then it's
possible that the external source tree is not a git repository, so we
should handle that case here instead of printing a traceback.

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

devtool: update-recipe: handle unversioned bbappends

Use the proper bbappend file name, don't just assume it will have a
version suffix (because it won't if the original recipe doesn't).

Fixes [YOCTO #7651].

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

devtool: reset: avoid errors in case file no longer exists

If you manually delete files in the workspace layer (which you really
shouldn't) it was possible to get yourself into the situation where you
couldn't reset because we were attempting to check if the file had been
modified and erroring out if it couldn't be opened. If the file's not
there anymore there's not much point checking if it needs to be
preserved, just skip it.

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

devtool: add: use correct bbappend file name with -V option

We weren't adding the version into the bbappend file name when -V was
specified which meant that building or resetting failed.

Also adjust one of the tests so that we're testing devtool add both with
and without this option.

Fixes [YOCTO #7647].

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

devtool: handle . in recipe name

Names such as glib-2.0 are valid (and used) recipe names, so we need to
support them.

Fixes [YOCTO #7643].

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

devtool: include bbappends in recipe parsing

In order to get correct metadata, SRCREV for example.

Fixes [YOCTO #7648].

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oe-selftest: devtool: fix test_devtool_update_recipe_git

Make this test work after recent changes to the mtd-utils recipe, and
hopefully make it robust against any future changes.

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

devtool: modify: implement --no-same-dir

This option can be used to have a separate build directory, in order to
keep the srctree directory clean for packages that do not need to be
built in the source directory.

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

devtool: modify: use B=S if that is the default for the recipe

Makes the build succeed for packages which do not support separate build
directory, e.g. zlib. The same outcome could be achieved with the
--same-dir option, but, it's generally hard to tell if a random package
would need that option. The negative side effect of this patch is that
dev srctree (of some packages that build fine without this modification)
gets dirtied by build artefacts.

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

xkeyboard-config: modified the license

Because usr/share/X11/xkb/symbols/eu and usr/share/X11/xkb/symbols/ph are GPLv3.
So add GPLv3 in the license.

Signed-off-by: Lei Maohui <leimaohui@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sysstat: upgrade to 11.1.4

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

pciutils: upgrade to 3.3.1

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

linux-yocto/3.19: configuration updates and stable integration

Bumping the linux-yocto 3.19 SRCREVs to integrate the 3.19.5 korg
stable updates, as well as the following meta data changes:

  205aca0c1241 meta: intel-common: Enable USB-based Bluetooth hardware
  b6a810e8e808 meta: features/bluetooth: add support for USB Bluetooth hardware
  767f3fa34680 common-pc-drivers: Add CONFIG_EEPROM_AT24
  e308b2c52519 intel-core*: Add Braswell soc support
  8c1c74d5052b braswell: Add features/soc/braswell

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

tzdata: update to 2015d

Changes affecting future time stamps

Egypt will not observe DST in 2015 and will consider canceling it
permanently.  For now, assume no DST indefinitely.
(Thanks to Ahmed Nazmy and Tim Parenti.)

Changes affecting past time stamps
America/Whitehorse switched from UTC-9 to UTC-8 on 1967-05-28, not
1966-07-01.  Also, Yukon's time zone history is documented better.
(Thanks to Brian Inglis and Dennis Ferguson.)

Change affecting past and future time zone abbreviations
The abbreviations for Hawaii-Aleutian standard and daylight times
have been changed from HAST/HADT to HST/HDT, as per US Government
Printing Office style.  This affects only America/Adak since 1983,
as America/Honolulu was already using the new style.

Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tzcode: update to 2015d

Changes affecting code

    zic has some minor performance improvements.

Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

insserv: Fix build with gcc5 and clang

Kill 2 bugs with one patch

| insserv.c:997:5: note: each undeclared identifier is reported only
once for each function it appears in
| insserv.c:997:15: error: expected ';' before 'char'
|      extension char buf[strlen(myname)+2+strlen(fmt)+1];
|                ^
| insserv.c: In function 'main':
| insserv.c:2379:5: error: 'extension' undeclared (first use in this
function)
|      extension char * argr[argc];
|      ^
| insserv.c:2379:15: error: expected ';' before 'char'
|      extension char * argr[argc];
|                ^
| insserv.c:2401:2: error: 'argr' undeclared (first use in this
function)
|   argr[c] = (char*)0;
|   ^

Change-Id: I36b7fb9e8baeda5a7cc252da10c0527401248226
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

fontconfig: Fix pkgconfig file generation to not have absolute paths

Fixes cross building dependent packages with gcc-5
e.g. matchbox-panel-2

Change-Id: I6290cfa4aac5d03a7200472d0c713b7ea84b99ac
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

fontconfig: Upgrade to 2.11.93

LIC_FILES_CHKSUM change is due to shifting of
the hunk to different line numbers, checksum remains same still

Change-Id: I0ce0cc96d39c801e57394f7a839a01d11a6f3979
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

pkgconfig: Upgrade to latest git tip

0.28 was released 2 years ago and bunch of patches
have gone in since. The commit rate is quite low
so its not a lot of churn to use. We have backports
for few bugs that are removed. Git version of recipe is removed too
since its no longer needed

Change-Id: I4b57db15320c76b1de5d26a733e60436663ff34a
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gummiboot: upgrade 43 -> 48+git

43 does not compile with gcc-5
Refresh objcopy patch to incorporate latest changes

Change-Id: Ied41ed1335bb1db3438b98121e05c22242a8a13e
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub: Backport const qualifier fix for gcc-5

gcc-5 is stricter and complains about const to non-const
conversions, we backport the patch from upstream into 2.00

Change-Id: I17db365fdd253daaa1ab726e2a70ecad0ac7b2ae
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gtk+: Correct function prototype

This is turned on by default in gcc-5 so far
we have been getting by this issue but time to fix it
has now come

Change-Id: I012248731de0be7a9b7e232073746360fe6c83dd
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>