oeqa/sdkext/devtool: Ensure dependencies for test_extend_autotools_recipe_creation are present

test_extend_autotools_recipe_creation needs libxml2 so ensure this is
installed/present as it may not be in the minimal eSDK case.

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

sanity: Allow whitespace only mirror entries

Forcing the use of "\n" in mirror variables is pointless, we can just require that
there are pairs of values. With the bitbake restriction relaxed, we can relax
the sanity check too.

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

testsdk: Use auto.conf instead of local.conf to set SSTATE_MIRROR

The local.conf file can get overritten and changed by the SDK so (ab)use
auto.conf instead to ensure our configuration tweaks get applied.

Also ensure we append to any other configured SSTATE_MIRROR variable
that already exists.

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

meta-estsdk-toolchain: use WORKDIR instead of D to hold locked-sigs

do_locked_sigs() uses ${D} to hold a file as input to the sstate staging code.
Generally this isn't a problem as it happens after packaging, but sometimes (if
the recipe is rebuild, perhaps) do_package() will see the file inside ${D} and
fail because there is an uninstalled file.

Solve this by using a directory under ${WORKDIR} so it can't be packaged.

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

core-image-lsb-sdk: Lower IMAGE_OVERHEAD_FACTOR

We're on the margin of the 4GB size limit for certain image types with
this configuration. This is a temporary workaround to avoid total
failures whilst we figure out better ways to save space. On such a
large image, the standard 1.3 is large and 1.2 is reasonable.

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

oeqa/sdkext: don't skip tests if there isn't a toolchain

Skipping the tests if a toolchain wasn't installed out of the box (for example,
a minimal eSDK) doesn't make sense as the first thing the tests should do is
install a toolchain.

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

oeqa/sdkext/devtool: use finally instead of repeating cleanup

Use the finally: block to always to cleanup.

Now that the test harness in testsdk.bbclass has monkey-patched
CalledProcessException to display the output we don't need to do that in the
test case.

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

oeqa: add output to subprocess exceptions

Out of the box subprocess.CalledProcessError.__str__() just displays the command
and exit code, which isn't very useful for debugging.

Add a function to oeqa.utils.subprocesstweak to monkey-patch __str__() so that
it can also display the value of stdout and stderr.

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

oeqa/esdk/devtool: clean setUpClass/tearDownClass

These methods are class not instance methods, so the argument should be cls not
self.

Also don't put variables into cls that we don't need there.

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

oeqa/selftest: Drop http sstate sharing

Using httpServer from python for sharing sstate is known to be buggy, it can't
cope with the number/type of requests coming from bitbake and quietly fails
to share files.

This causes intermittent build failures which are hard to debug. We can
use a file:// url for the sstate mirror instead, removing the need for
the http server.

The sdk-update test is simply dropped since the SDK is never published
to this location and hence it would never have any update. Its equiavalent
to pointing at an empty web server. There is a better eSDK update test in
testsdk so rather than improve this one, lets drop it and concentrate on
the one there.

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

populate_sdk_ext: Allow generation of meta-extsdk-toolchain even for minimal SDKs

If you build a minimal eSDK currently, you don't build meta-extesdk-toolchain
even if you will have built most of its dependencies. This means when you try
and install a toolchain into the eSDK, it fails, breaking our automated testing
of the eSDK.

Therefore add the dependency unconditionally even when a minimal eSDK is being
built and allow the automated testing to work.

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

oeqa/sdkext/context: Work around broken dependency checks to get sdk tests running

This is admitted a bit of a hack but it does allow a number of significant sdk
tests to run successfully and hence improves testing of eSDK which is good.

I'm therefore proposing we do this until we come up with a better solution
since the current lack of testing is worrying and would have caught other issues
had it been present.

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

oeqs/sdk*/case: Use universal_newlines for subprocess calls

This removes the need for some of the ugly decode calls with hardcoded
locales.

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

oeqa/sdkext: Ensure we run a deterministic set of tests

The directory list of sdk tests to run can vary so this code effectively selects
a random set of SDK tests to run in the eSDK. We want to attemp all the SDK tests
so remove the element selection.

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

devtool/sdk: Run build-sysroots after installing new things

After running sdk-install we need to ensure that the standalone sysroots are
updated as done when the eSDK is originally built. Add such a call so this
happens automatically and the envrionment scripts in the SDK work correctly
after updates.

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

testsdk: Handle minimal eSDK and avoid download costs

When using a minimal eSDK, testing currently fails as the sdk isn't
populated. We therefore setup the eSDK under test to point at local
sstate and execute a command to ensure the toolchain is populated
since most of the tests depend on this being present.

At the same time, add in a link to DL_DIR through own-mirrors so
that tests which fetch source (e.g. the kernel module one) can
use the local stash. This cuts test execution of the kernel module
test from 2000s to 120s.

We did try using DL_DIR directly but that causes uninative issues
requiring other workarounds so own-mirrors is neater.

Together these fixes unbreak eSDK testing on the autobuilder.

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

oeqa/sdkext/devtool: rename recipe name to match upstream

Call the generated recipe librdfa instead of bb-example to make it clearer what
is happening.

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

weston: Upgrade 1.11.1 -> 2.0.0, separate libweston

* Drop two patches that are upstream. Rebase other patches.
* Separate libweston into its own package, modify the recipe
  as needed because files have changed location.
* Remove "--disable-rpi-compositor": the backend does not exist
  anymore.

Libweston is already at version 2 and is likely to have new major
versions. The versions should be parallel installable (but weston
itself will not be).

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

wayland: Upgrade 1.12.0 -> 1.13.0

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

ruby: upgrade 2.3.3 -> 2.4.0

Two LIC_FILES_CHKSUM checksums changed (COPYING and LEGAL) but LICENSE remains
the same.

Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libusb1: upgrade 1.0.20 -> 1.0.21

Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

ffmpeg: use static package list

Dynamic packaging isn't useful if every library needs to be
listed manually.

This also merges the -dev packages into a single ffmpeg-dev, as is typical.

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>

insane.bbclass: Add missing dependencies

package_qa task requires some tools installed in sysroot; with
the introduction of recipe specific sysroot this task won't
have such tools installed if it's forced to run.

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

generate-manifest-2.7.py: fix python-tests rdepends

A potential flaw was introduced by commit a8dc4bc0:
[ python: fix python-tests rdepends ]

It made python-tests RDEPENDS on python-modules which is correct, but
this should be done in generate-manifest-2.7.py, and re-generate the
manifest.

Fixed by adding this dependency in generate-manifest-2.7.py as well.

Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

generate-manifest-3.5.py: add logic to generate native manifest

python3-native supposes to RPROVIDE all native packages as added in
generate-manifest-3.5.py, but it does not so far, this leads a problem
that sometimes bitbake cant find a runtime provider for a python3-*-native
when a new runtime dependency on it being required, this usualy happens
after a new native python3-* recipe is created or the old native python3-*
recipes are upgraded.

To avoid manually extending RPROVIDE every time when a new runtime
dependency is introduced, an argument '-n/--native' is added to the
manifest generator, allowing it create a native python3 manifest, with a
RPROVIDE line only, the RPROVIDE should contain all the sub-packages.

The generated python-native-3.5-manifest.inc is also added which is
included by python3-native recipe.

Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

generate-manifest-2.7.py: add logic to generate native manifest

python-native supposes to RPROVIDE all native packages as added in
generate-manifest-2.7.py, but it does not so far, this leads a problem
that sometimes bitbake cant find a runtime provider for a python-*-native
when a new runtime dependency on it being required, this usualy happens
after a new native python-* recipe is created or the old native python-*
recipes are upgraded.

To give a example, the following commit is trying to address such a issue:
commit 4583cd1bb15306e8f0ab7bcd80732e6f35aa4533:
[ python-native: Make python-native also RPROVIDE python-unittest-native ]

To avoid manually extending RPROVIDE every time when a new runtime
dependency is introduced, an argument '-n/--native' is added to the
manifest generator, allowing it create a native python manifest, with a
RPROVIDE line only, the RPROVIDE should contain all the sub-packages.

The generated python-native-2.7-manifest.inc is also added which is
included by python-native recipe.

Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

base.bbclass: Fix OE_IMPORTS

Udpate OE_IMPORTS to include oe.license because it is being used.

Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

oelib/buildhistory.py: Add unittest for buildhistory_analysis

The buildhistory_analysis module (in which buildhistory-diff is
based) was lacking unittest for its functions. Created selftest
module for this and a few testcases to cover basic cases.

[YOCTO #10727]

Signed-off-by: Humberto Ibarra <humberto.ibarra.lopez@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

oeqa/core/loader.py: Avoid importing tests with built-ins name

If importing a test with the same name as a built-in module,
it will silently import the built-in and check for tests in
built-in module. This happened with syslog module in debian
based machines, so add a raise to avoid this behavior.

[YOCTO #10978]

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

oeqa/core/loader.py: Give meaningful error when failed to load classes

With this we get the class that is actually having the problem,
not just a TypeError with an unknown class causing the error.

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

wireless-tools: Update URLs

wireless-tools is now hosted on
https://hewlettpackard.github.io/wireless-tools/Tools.html

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

selftest: Optimize get_bb_var use

get_bb_var calls bitbake every time it is used and every call
would take about 7 seconds. There are tests that calls get_bb_var
several times when they can use get_bb_vars. Also there are tests
that calls it to fetch the same variable over and over again.

This will optimize the use of get_bb_var and get_bb_vars for a
little speed up in the tests.

[YOCTO #11037]

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

perl-native: Remove usage of -fstack-protector=strong

Some distributions (like opensuse421) supported by the project
comes with older gcc releases, -fstack-protector=strong is supported
by GCC>=4.9.

This causes a build failure when install perl-native from a sstate that
comes from a machine supporting -fstack-protector=strong [1].

So disable usage of this flag in perl-native builds, this patch could
be removed when all supported distros comes with GCC>=4.9.

[YOCTO #10338]

[1] http://errors.yoctoproject.org/Errors/Details/109589/

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

gstreamer: update to 1.10.4

bug-fix release:
 *) Various fixes for crashes, assertions, deadlocks and memory leaks on
fuzzed input files and in other situations (CVE-2017-5847, CVE-2017-5848)
 *) gst-libav was updated to ffmpeg 3.2.4, fixing a couple of CVEs

Signed-off-by: Maxin B. John <maxin.john@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

iproute2: upgrade to 4.10.0

4.9.0 -> 4.10.0

added the following patch to fix build with musl libc
 1) 0001-libc-compat.h-add-musl-workaround.patch

Signed-off-by: Maxin B. John <maxin.john@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

python3-pip: RDEPEND on python3-html.

Without this pip3 fails with "ImportError: No module named 'html'"

Signed-off-by: Ismo Puustinen <ismo.puustinen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

tar: CVE-2016-6321

Skip members whose names contain "..".

Reference:
https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2016-6321

Upstream patch:
http://git.savannah.gnu.org/cgit/tar.git/commit/?id=7340f67b9860ea0531c1450e5aa261c50f671

Signed-off-by: Sona Sarmadi <sona.sarmadi@enea.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

buildstats.bbclass: Avoid index exception in /proc/PID/io parsing

There is some probability (depends on system load) to get empty
or line containing "0" as the last line while reading /proc/PID/io.
Avoid build failure by checking if line contains separator
":" before split.

Signed-off-by: Pavel Modilaynen <pavelmn@axis.com>
Signed-off-by: Daniel Lublin <daniel@lublin.se>
Signed-off-by: Ross Burton <ross.burton@intel.com>

webkitgtk: update to 2.14.5

Remove a reference to WebKit/LICENSE, as the directory is not there anymore
(and the LICENSE file in it hasn't been moved somewhere else) - it was
a relic of webkit 1.x era.

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

systemtap: update to 3.1

Removed patches are in upstream.

Rebase monitor-option.patch and no-msgfmt-check.patch

Add support for python 3 probes
(this necessitates the 0001-Do-not-let-configure-write-a-python-location-into-th.patch
and the Add 0001-Install-python-modules-to-correct-library-dir.patch to
install python modules to correct destination).

Add perl to RDEPENDS to fix a QA warning about one of the installed files.

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

ltp: upgrade to 20170116

1. Upgrade ltp from 20160126 to 20170116.

2. Delete some patches because these have been integrated in upstream.

0001-ltp-Don-t-link-against-libfl.patch
0006-sendfile-Use-off64_t-instead-of-__off64_t.patch
0007-replace-SIGCLD-with-SIGCHLD.patch
0009-Guard-error.h-with-__GLIBC__.patch
0012-fsstress.c-Replace-__int64_t-with-int64_t.patch
0013-include-fcntl.h-for-getting-O_-definitions.patch
0014-hyperthreading-Include-sys-types.h-for-pid_t-definit.patch
0015-mincore01-Rename-PAGESIZE-to-pagesize.patch
0016-ustat-Change-header-from-ustat.h-to-sys-ustat.h.patch
0017-replace-sigval_t-with-union-sigval.patch
0019-tomoyo-Replace-canonicalize_file_name-with-realpath.patch
0022-include-sys-types.h.patch
0029-trace_shed-Fix-build-with-musl.patch
0031-vma03-fix-page-size-offset-as-per-page-size-alignmen.patch
0032-regen.sh-Include-asm-unistd.h-explicitly.patch
0037-containers-netns_netlink-Avoid-segmentation-fault.patch
0038-run-posix-option-group-test-replace-CWD-qith-PWD.patch
0040-containers-userns05-use-unsigned-int-for-ns-id.patch

3. Add new patch based on Khem Raj's 0007-replace-SIGCLD-with-SIGCHLD.patch

0001-add-_GNU_SOURCE-to-pec_listener.c.patch

4. Add new patches fix building errors with musl:

0006-fix-PATH_MAX-undeclared-when-building-with-musl.patch
0007-fix-__WORDSIZE-undeclared-when-building-with-musl.patch
0009-fix-redefinition-of-struct-msgbuf-error-building-wit.patch

   Add new patch fix building errors, when the distro is poky-lsb

0012-fix-faccessat01.c-build-fails-with-security-flags.patch

5. Modify some patches because the upstream contains changes, those patches
   maybe failed when they apply to the new version of ltp.

0011-Rename-sigset-variable-to-sigset1.patch
0027-sysconf01-Use-_SC_2_C_VERSION-conditionally.patch
0030-lib-Use-PTHREAD_MUTEX_RECURSIVE-in-place-of-PTHREAD_.patch
0035-fix-test_proc_kill-hang.patch

6. Rename and modify this patch.

fcntl-fix-the-time-def-to-use-time_t.patch to
0039-fcntl-fix-the-time-def-to-use-time_t.patch

7. The new version of ltp add the checking for ksh and csh, we doesn't support
   ksh and csh in oe-core, so remove the checking files when the system executes
   do_install function.

Signed-off-by: Dengke Du <dengke.du@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

ed: update to 1.14.1

Revert to the upstream server, now that we can handle .lz files.
License checksum changes are due to copyright years.

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

lzip: add a recipe from meta-oe

ed tarballs are now available only in .lz format, which saves them
an amazing 30 kilobytes over gz.

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

glib-2.0/glib.inc: If 'charset.alias' exists drop it

Delete the 'charset.alias' file during the do_install task if it exists.
This file is not shipped in any packages and on the targets that it is
generated (musl and mingw32) it is undesired.

Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

glib-2.0/glib.inc: Only apply MLPREFIX renaming to class-target

When renaming the 'gio-querymodules' binary to prefix with the current
MLPREFIX, only do this for target packages where the MLPREFIX makes
sense. This skips renaming for nativesdk targets which have MLPREFIX set
to 'nativesdk-'.

Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

btrfs-tools: update to 4.9.1

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

slang: 2.3.1 -> 2.3.1a

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

license.bbclass: run chown only under pseudo

The copy_license_files() function in license.bbclass is called in two
different contexts. First, it is run as part of the do_populate_lic
task. In addition, it is called from do_package task when
LICENSE_CREATE_PACKAGE is enabled. The function has code that changes
the owner of license files to root which is meant only to happend in the
latter case - i.e. under do_package which is run under pseudo.
Previously, the code was blindly always running chown and just ignored
errors that happened when running from do_populate_lic. This patch
changes it to be more intelligent, only doing chown when running under
pseudo.

[YOCTO #10897]

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

boost: update to 1.63.0

Drop backported 0002-boost-test-execution_monitor.hpp-fix-mips-soft-float.patch
Rebase consider-hardfp.patch to 0001-When-using-soft-float-on-ARM-we-should-not-expect-th.patch

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

kconfig-frontends: fetch source from git

This is where development happens; tarballs are no longer produced.

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

glib-2.0: Upgrade 2.50.2 -> 2.50.3

Tiny bug fix release.

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

unzip: CVE-2014-9913 CVE-2016-9844

Backport the patches for CVE-2014-9913 CVE-2016-9844

CVE-2016-9844:
Buffer overflow in the zi_short function in zipinfo.c in Info-Zip
UnZip 6.0 allows remote attackers to cause a denial of service
(crash) via a large compression method value in the central
directory file header.
CVE-2014-9913:
Buffer overflow in the list_files function in list.c in Info-Zip
UnZip 6.0 allows remote attackers to cause a denial of service
(crash) via vectors related to the compression method.

Patches come from:
https://github.com/NixOS/nixpkgs/blob/master/pkgs/tools/archivers/unzip/  or
https://release.debian.org/proposed-updates/stable_diffs/unzip_6.0-16+deb8u3.debdiff

Bug-Debian: https://bugs.debian.org/847486
Bug-Ubuntu: https://launchpad.net/bugs/1643750

(LOCAL REV: NOT UPSTREAM) --send to oe-core on 20170222

Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

nss: update to 3.28.1

Rebase nss-fix-support-cross-compiling.patch

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

libwebp: update to 0.6.0

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

libnl: update to 3.2.29

Dropped patches are all backports.

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

kbd: update to 2.0.4

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

iso-codes: update to 3.74

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

gnutls: update to 3.5.9

Drop 0001-Do-not-add-cli-args.h-to-cli-args.stamp-Makefile-tar.patch, it's merged upstream.
Rebase 0001-configure.ac-fix-sed-command.patch.

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

ffmpeg: update to 3.2.4

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

bash-completion: update to 2.5

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

babeltrace: update to 1.5.2

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

libva: fix upstream version check

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

libepoxy: fix upstream version check

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

sysprof: inherit upstream-version-is-even

It is a gnome project, and so is subject to same versioning convention.

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

zlib: fix upstream version check

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

ruby: fix upstream version check

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

initscripts/sysfs.sh: mount configfs if present

configfs is another kernel virtual file system that should be mounted
if configured, so if it's configured into the kernel, mount it. It is
used to configure e.g. USB gadget mode and devicetree overlays.

Signed-off-by: Mike Looijmans <mike.looijmans@topic.nl>
Signed-off-by: Ross Burton <ross.burton@intel.com>

bitbake.conf: fix ineffective include conf/target/${TARGET_SYS}.conf

TARGET_SYS is defined in terms of TARGET_ARCH, so it's not valid
until after TUNE_ARCH has been set by the machine config. The
original order of includes resulted in an attempt to include
non-existent files such as:

  conf/target/INVALID-oe-linux.conf

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

python-pexpect: BBCLASSEXTEND to native

Some developers might need it.

Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

python-ptyprocess: BBCLASSEXTEND to native

Some developers might need it.

Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

selftest: Avoid sstate corruption by calling cleansstate

Currently selftest doesn't use sstates because some tests
clean sstate cache; using sstates would give a performance
boost instead of building everything from scratch.

With this sstates are not corrupted using different methods
depending on tests:

devtool: These tests needed to delete the cache so SSTATE_DIR
as SSTATE_MIRRORS and set a temporal SSTATE_DIR.

sstatetests: This module already used a temporal SSTATE_DIR, so
just set up the SSTATE_MIRRORS.

Rest: Removed cleansstate, some of them required to force a
certain task, others were just removed or changed for another
task.

[YOCTO #10929]

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

curl: upgrade to 7.53.0

Signed-off-by: Oleksandr Kravchuk <oleksandr.kravchuk@pelagicore.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

lib/oe/gpg_sign: fix rpm signing with gpg > 2.1

We need to check the gpg version and alter its command line options
accordingly.

[YOCTO #11054]

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

lib/oe/gpg_sign: make gpg version a property of the signer

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

rpm: support customizing gpg command line

Add a new %_gpg_sign_cmd_extra_args macro that allows customizing the
gpg options used when signing rpm packages. This is needed to be able to
sign packages with gpg 2.1 which requires "--pinentry-mode loopback" to
allow non-interactive signing.

[YOCTO #11054]

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

lib/oe/gpg_sign: sign rpm packages in chunks of 100

Split the file list into chunks in order to avoid
"OSError: [Errno 7] Argument list too long"

This would happend when a package has huge amount of subpackages, e.g.
glibc-locale.

[YOCTO #11069]

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

qemu: fix build with glibc-2.25

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

e2fsprogs: Fix build with glibc-2.25

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

syslinux: fix build with glibc-2.25

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

parted: fix build with glibc-2.25

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

x264: Upgrade to stable branch head

Small number of bug fixes and x86 optimizations.

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

gst-player: Upgrade to recent git revision

The update contains a small number of bug fixes. Removed one
upstreamed patch.

gst-player does not have releases: this is the current git master.

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

xf86-video-intel: Upgrade to recent git revision

This includes a number of small fixes and some new PCI IDs, no
major changes.

For background, xf86-video-intel does not get releases so
we have to follow git.

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

epiphany: update to 3.22.6

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

gdk-pixbuf: Upgrade 2.36.1 -> 2.36.5

Mostly bug fix releases.

Generating the thumbnailer metadata now requires running yet another
tool at build time. This is broken for cross-compiling, add a
work-around.

Add gdk-pixbuf-native to DEPENDS to make the above workaround possible:
We already build gdk-pixbuf-native anyway so this is not a huge deal.

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

menu-cache: Upgrade 1.0.1 -> 1.0.2

Small bug fix release.

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libdrm: Upgrade 2.4.74 -> 2.4.75

* Remove udev from depends (it's not actually used).
* Rebase a patch

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libgudev: Upgrade 230 -> 231

Very small bug fix release.

Use --disable-umockdev: Testing is great but a single regression test
isn't really worth a new build dependency (that isn't in oe-core yet).

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

gtk+3: Upgrade 3.22.7 -> 3.22.8

Bug fix release.

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

sudo: upgrade to 1.8.19p2

The license checksum for doc/LICENSE is changed. It's a small change.
'2015' is changed to '2017'. Nothing else is changed. So the licenses
remain the same.

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

file: 5.29 -> 5.30

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

guile: 2.0.13 -> 2.0.14

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

strace: 4.15 -> 4.16

* The license cheksum is changed becuase a new line is added:
  Copyright (C) 2001-2017 The strace developers.

* Remove use-asm-sgidefs.h.patch, it doesn't check sgidefs.h any more, it was
  use for building on mips, I checked it built well.

* Update Makefile-ptest.patch and disable-git-version-gen.patch.

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

dpkg: Upgrade to 1.18.10

Add a patch to don't use --clamp-time when call tar because
isn't supported in tar hosts versions. See
0007-dpkg-deb-build.c-Remove-usage-of-clamp-mtime-in-tar.patch
patch for details.

Rebased patch:

-  0003-Our-pre-postinsts-expect-D-to-be-set-when-running-in.patch

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

perl: Upgrade to 5.24.1

Upgrade config.sh to match new version.

Removed CVE patches already in upstream:

- perl-fix-CVE-2016-1238.patch
- perl-fix-CVE-2016-6185.patch

Update customized.dat patch to match new hashes.

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

Revert "flex: upgrade to 2.6.2"

This reverts commit 3632abd01abb8dfff230e18f828af705da488f97.

Multiple people have expressed issues with flex-2.6.2; personally I had
problems compiling libsepol from meta-selinux (for libselinux). I tried
upgrading to flex-2.6.3, but that caused binutils-cross_2.27 to fail.

The simplest for now is to downgrade to flex-2.6.0.

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

sqlite3: upgrade to 3.17.0

3.16.2 -> 3.17.0

* Approximately 25% better performance from the R-Tree extension.
* Other performance improvements. Uses about 6.5% fewer CPU cycles.

Signed-off-by: Maxin B. John <maxin.john@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

python: Upgrade both python and python-native to 2.7.13

Rebased:
- python-native/multilib.patch
- python/multilib.patch
- python/01-use-proper-tools-for-cross-build.patch

Upstream:
- CVE-2016-1000110

Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

kernel-devsrc: remove symlink source

Fixed:
$ rpm -qplv tmp/deploy/rpm/qemuppc/kernel-devsrc-1.0-r0.qemuppc.rpm | grep kernel/source
/usr/src/kernel/source -> /buildarea/lyang1/test_yocto/tmp/work/qemuppc-poky-linux/kernel-devsrc/1.0-r0/image/usr/src/kernel

It is generated by kernel's "make clean _mrproper_scripts", the
kernel-devsrc includeds full sources, this symlink is not needed, and
the path is invalid on target, so remove it.

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

scripts/lib/devtool/deploy.py: add --port/-P argument for target connection

Enable using, e.g. host port 2222 for connection to qemu target.
Defaults to 22 for standard ssh/scp port.

[YOCTO #11079]

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>