oe-init-build-env*: Make them actually return failures

If either of the internal scripts (oe-buildenv-internal and
oe-setup-builddir) failed, oe-init-build-env (and
oe-init-build-env-memres) would still return success.

Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

oe-init-build-env*: Remove unnecessary differences between the scripts

While at it, also fix:
* consistent indentation (four spaces)
* unset temporary variables
* use $(...) instead of `...`

Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

oe-init-build-env*: Update/correct comment about specifying arguments

Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

oe-init-build-env*: Allow $OEROOT to be predefined

The current implementation of oe-init-build-env and
oe-init-build-env-memres requires that they are sourced from the
directory that will be known as $OEROOT. This makes it hard to write a
wrapper script with the same name as the original OE script which,
e.g., sources the original OE script from a sub-directory.

With this change, $OEROOT can be predefined when oe-init-build-env or
oe-init-build-env-memres is sourced, allowing the original OE scripts
to be anywhere.

Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

bluez5: allow D-Bus to spawn obexd in systems without systemd

This includes a proper D-Bus service file for obexd in systems that do
not support systemd.

Signed-off-by: Javier Viguera <javier.viguera@digi.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

oeqa: remove RPM 4 self test

We're dropping RPM 4 as it's not well maintained, therefore remove
a selftest which tests RPM 4 availability.

Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

lib/package_manager: remove RPM4 support code

Simplify the RPM code by removing support for RPM 4 now that we've
dropped the RPM 4 recipe.

Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

smartpm: remove rpm4 patch

The RPM4 support we added to SMART doesn't appear to work, remove
it as part of the removal of RPM4 from OE-Core.

Refresh the smart-add-for-rpm-ignoresize-check.patch which was
applied after smart-rpm4-fixes.patch and doesn't apply cleanly once
that patch is removed.

Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

rpm: remove RPM 4

RPM4 support is buggy and incomplete. As we don't have the
resources or interest to maintain it this patch removes it.

Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

grub: fix documentation rebuilds

If the documentation needs to rebuild then it will fail as the syntax isn't
valid with modern texinfo.  Backport a patch from git to fix the syntax.

[ YOCTO #9306 ]

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

oe-selftest: Fixed --list-tests-by tag option

Commit 35be67951305950ba797dc2efddbc7d88fc0556a broke the
--list-tests-by tag option.
This patch fixes that.

Having a module in lib/oeqa/selftest named testmodule:
class TestClass(oeSelfTest):
    @tag(feature='tag1')
    def test_func1(self):
        pass
    @tag(feature=('tag1', 'tag2'))
    def test_func2(self):
        pass
    @tag(feature=('tag2', 'tag3'))
    def test_func3(self):
        pass
    @tag(feature=('tag1', 'tag2', 'tag3'))
    def test_func4(self):
        pass

$ oe-selftest --list-tests-by tag tag1
 ID  TAG(s)            NAME        CLASS      MODULE
----  ----------------  ----------  ---------  --------
      tag1              test_func1  TestClass  testmodule
      tag1, tag2        test_func2  TestClass  testmodule
      tag1, tag2, tag3  test_func4  TestClass  testmodule
______________________________
Filtering by:  tag
Looking for:  tag1
Total found:  3

$  oe-selftest --list-tests-by tag tag1 tag2
  ID  TAG(s)            NAME        CLASS      MODULE
----  ----------------  ----------  ---------  --------
      tag1              test_func1  TestClass  testmodule
      tag1, tag2        test_func2  TestClass  testmodule
      tag1, tag2, tag3  test_func4  TestClass  testmodule
      tag2, tag3        test_func3  TestClass  testmodule
______________________________
Filtering by:  tag
Looking for:  tag1, tag2
Total found:  4

$ oe-selftest --list-tests-by tag tag*
  ID  TAG(s)            NAME        CLASS      MODULE
----  ----------------  ----------  ---------  --------
      tag1              test_func1  TestClass  testmodule
      tag1, tag2        test_func2  TestClass  testmodule
      tag1, tag2, tag3  test_func4  TestClass  testmodule
      tag2, tag3        test_func3  TestClass  testmodule
______________________________
Filtering by:  tag
Looking for:  tag*
Total found:  4

Signed-off-by: Daniel Istrate <daniel.alexandrux.istrate@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

gcc-runtime.inc: set LICENSE for all gcc-runtime packages

LICENSE_${PN} doesn't apply to all gcc-runtime packages. Set LICENSE
instead. Without this fix, gcc-runtime packages such as libstdc++ are
excluded from rootfs for builds which blacklist GPLv3.

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

ParaTypeFFL-1.3: Add license file

The ParaType Free Font License is used by various fonts, out of which a
recipe will be added to meta-openembedded for ttf-pt-sans.

Signed-off-by: Ioan-Adrian Ratiu <adrian.ratiu@ni.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

externalsrc: use shared stamp directory if B=S

We have a common build directory for all target architectures if
externalsrc is used and ${B}=${S}. In this case we also need to have a
common stamp directory. The reason is that e.g. changing MACHINE will
basically "invalidate" the artefacts in ${B} (wrt. to the new MACHINE)
but old stamp files with matching hashes from an earlier build might be
available in the machine-specific stamps directory and tasks would not
be correctly re-run.

Previous attempt in f44f12af346888bdeb3ae01a275cb5dd4396b505 to handle
this correctly was faulty as it threw some spurious warnings and caused
do_configure to be re-run in every build if that would not be necessary.

[YOCTO #8950]
[YOCTO #9237]

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

rpm: fix error when 'lua' is enabled

Fix the following compilation error when 'lua' is enabled in PACKAGECONFIG.

  | gcc: error: lib21.c: No such file or directory
  | gcc: fatal error: no input files

Modify FILES for dev package to avoid installed-not-shipped error.

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

matchbox-keyboard: Update to latest HEAD to fix 64bit issue

This brings us to the current HEAD and adds the following patches:

ebc330e Saul Wold matchbox-keyboard-ui.c: Fix int -> long to address 32bit vs 64 bit addressing
8063096 Khem Raj config-parser: Use matching printf format

[YOCTO #9254]

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

oeqa/selftest/buildoptions: test read-only-rootfs

Add a test to build core-image-sato with read-only-rootfs enabled.

[ YOCTO #9214 ]

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

oeqa/selftest/sstatetests: verify more variables don't impact the hash

Add DL_DIR and TIME to the hash tests.  We can't add DATE for some reason.

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

gobject-introspection.bbclass: wrap comments at 80 columns

Whitespace and comment updates only, no functional changes.

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

qemuarm64.conf: don't clear MACHINE_FEATURES

Accept the default MACHINE_FEATURES from qemu.inc (qemuarm64
shouldn't need to be a special case).

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

sanity.bbclass: allow customizing config file update error messages

The default file pattern may be amiguous and "meld" might not always
be the preferred tool, so allow distros to override those parts of the
error messages.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

sanity.bbclass: fix success message when config file was updated

The code now iterates over different config files, but always printed a
message about conf/bblayers.conf for each file.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

sanity.bbclass: expand error messages for version checks

The ${WORKDIR} variable was not getting expanded (anymore?), leading to
less helpful error messages like:

   Exception: NotImplementedError: Your version of bblayers.conf has the wrong LCONF_VERSION (has 7, expecting 8).
   Please compare your file against bblayers.conf.sample and merge any changes before continuing.
   "meld conf/bblayers.conf ${COREBASE}/meta*/conf/bblayers.conf.sample"

   is a good way to visualise the changes.

   ERROR: Execution of event handler 'check_sanity_eventhandler' failed

After adding expansion, embedding ${LCONF_VERSION} and ${LAYER_CONF_VERSION}
in the error message seems a bit more readable and consistent.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

lighttpd: fix /usr/lib/mod_cgi.so: undefined symbol: chunkqueue_written

lighttpd fails to load when mod_cgi is enabled at run time, with the
message "dlopen() failed for: /usr/lib/mod_cgi.so /usr/lib/mod_cgi.so:
undefined symbol: chunkqueue_written".

This is caused by a patch intended to prevent memory exhaustion by
naively streaming CGIs, aimed at upstream issue
http://redmine.lighttpd.net/issues/1264 . The patch uses internal API
functions from older versions of lighttpd which don't exist in this
version.  Remove the patch, pending a better fix.

[ YOCTO #9289 ]

Signed-off-by: Nick Leverton <nick@leverton.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>

valgrind: Disable nios2 support

valgrind does not support nios2 yet, so disable it.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Ley Foon Tan <lftan@altera.com>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Ross Burton <ross.burton@intel.com>
Cc: Thomas Chou <thomas@wytron.com.tw>
Cc: Walter Goossens <waltergoossens@home.nl>
Signed-off-by: Ross Burton <ross.burton@intel.com>

systemtap: Disable nios2 support

systemtap does not support nios2 yet, so disable it.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Ley Foon Tan <lftan@altera.com>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Ross Burton <ross.burton@intel.com>
Cc: Thomas Chou <thomas@wytron.com.tw>
Cc: Walter Goossens <waltergoossens@home.nl>
Signed-off-by: Ross Burton <ross.burton@intel.com>

lttng-modules: Add nios2 support

The lttng-modules are compatible with nios2, so just amend
the compatibility string.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Ley Foon Tan <lftan@altera.com>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Ross Burton <ross.burton@intel.com>
Cc: Thomas Chou <thomas@wytron.com.tw>
Cc: Walter Goossens <waltergoossens@home.nl>
Signed-off-by: Ross Burton <ross.burton@intel.com>

kexec: Disable on nios2

kexec is not supported on nios2, so disable it.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Ley Foon Tan <lftan@altera.com>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Ross Burton <ross.burton@intel.com>
Cc: Thomas Chou <thomas@wytron.com.tw>
Cc: Walter Goossens <waltergoossens@home.nl>
Signed-off-by: Ross Burton <ross.burton@intel.com>

packagegroup-core-sdk: Disable sanitizers for nios2

These are not available on nios2 yet, so disable them.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Ley Foon Tan <lftan@altera.com>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Ross Burton <ross.burton@intel.com>
Cc: Thomas Chou <thomas@wytron.com.tw>
Cc: Walter Goossens <waltergoossens@home.nl>
Signed-off-by: Ross Burton <ross.burton@intel.com>

bdgwc: Backport nios2 support

Backport patch from bdwgc mainline which adds initial
support for nios2 architecture.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Ley Foon Tan <lftan@altera.com>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Ross Burton <ross.burton@intel.com>
Cc: Thomas Chou <thomas@wytron.com.tw>
Cc: Walter Goossens <waltergoossens@home.nl>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libatomic-ops: Backport nios2 support

Backport a patch adding nios2 support into libatomic-ops

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Ley Foon Tan <lftan@altera.com>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Ross Burton <ross.burton@intel.com>
Cc: Thomas Chou <thomas@wytron.com.tw>
Cc: Walter Goossens <waltergoossens@home.nl>
Signed-off-by: Ross Burton <ross.burton@intel.com>

selftest/buildoptions: Renamed one test case

'test_layer_git_revisions_are_displayed_and_do_not_fail_without_git_repo'
was renamed to 'test_layer_without_git_dir' which is shorter.

fix for [YOCTO #9243]

Signed-off-by: Daniel Istrate <daniel.alexandrux.istrate@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

python-numpy: Fix build on musl

Match API version in cached _numpyconfig.h to the
one mandated by numpy version in use.

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

socat: Access c_ispeed and c_ospeed via APIs

make it more portable across libc implementations

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

watchdog: Disable nfs on musl targets

Fixes build issue on musl where it needs to link to sunrpc
when enabling nfs

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

bdwgc: Check for getcontext() API during configure

This helps in compiling dependent components like
guile where it shows the problem of missing getcontext
API when using libc which dont implement it e.g. musl

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

devtool: change config symlink name to .config.new

Otherwise (if the symlink is named .config) kernel build considers
source tree as dirty and fails.

[YOCTO #9270]

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

systemd: Fix and expand ptests

systemd-ptest only runs a couple of tests currently.
* Install all test binaries and required data files
* Add missing runtime dependencies for -ptest
* Fix paths required for on-target testing in do_configure_prepend()
* cleanup do_install_ptest()

For reference, results on current core-image-sato:
  PASS: 109
  SKIP:   5
  FAIL:   2

test-execute failure:
https://github.com/systemd/systemd/issues/2852

test-acl-util failure:
https://bugzilla.yoctoproject.org/show_bug.cgi?id=9269

Fixes [YOCTO #8767].

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

oeqa/utils/testexport.py: add functionality for exporting binaries

This new file is encapsulating functionality for both
running tests with binaries support via TestNeedsBin() decorator
and exporting these binaries via testimage.bbclass file.
Addresses [YOCTO #7850], [YOCTO #8478], [YOCTO #8481],
[YOCTO #8536], [YOCTO #8694].

Signed-off-by: Costin Constantin <costin.c.constantin@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

init-live : make it easier to add custom boot targets

When booting from the live image, the label from the bootloader is passed
to init.sh. init.sh uses the label to either boot a live image or call a
script to take over and install the system.

It is possible to add new labels to the bootloader via the LABELS family of
variables, but the names in init.sh were hardcoded to install and
install-efi

this patch checks if a shell script with the same name as the label is
available instead of using a hardcoded list. Any recipe can add such file
and this provide a new boot target to the live image

Signed-off-by: Jérémy Rosen <jeremy.rosen@openwide.fr>
Signed-off-by: Ross Burton <ross.burton@intel.com>

useradd_base.bbclass: replace retry logic with flock

When perform useradd during populate sysroot, it locks files passwd.lock
and group.lock at same time. And then it meets a dead lock issue
randomly.

Use flock to reslove it by using an universal lock directory for all the
user and group related operations.

[YOCTO #9022]

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

image.bbclass: track ROOTFS_POSTUNINSTALL_COMMAND in do_rootfs vardeps

The list of variables influencing do_rootfs was not updated when
introducing ROOTFS_POSTUNINSTALL_COMMAND. As a result, making changes
in commands listed there or the variables they depend on did not trigger
a re-run of do_rootfs.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

eudev: split eudev-hwdb from eudev

The eudev recipe included the hwdb.d directory by default,
which ended up making the eudev package a lot heavier than
the old udev, this patch splits the eudev-hwdb package to
avoid using unecessary space when required

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

openssl: don't move libcrypto to base_libdir

For now, if 'openssl' is enabled for ntp, ntp would still be built
without openssl & libcrypto. This is because that ntp thinks openssl
and libcrypto locates under the same directory.

This patch removes the codes of moving libcrypto to base_libdir.

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

xcb-util-image: Fix build with clang

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

musl: Update to get mips64 port

mips64 port is now available on top of 1.1.14
this updates brings it to OE

Copyright file changed due to mips64 attribution being added

http://git.musl-libc.org/cgit/musl/commit/COPYRIGHT?id=589aefa5b061647e8b9ad9bca3aaa8dc6222460a

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

dhcp: enable gentle shutdown

For now, `systemctl stop dhcpd' cannot stop dhcpd correctly, the SIGTERM
signal would time out, causing a SIGKILL signal sent to dhcpd.

Patch site.h to enable gentle shutdown to so that dhcpd could be stopped
by SIGTERM.

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

coreutils: fix reporting 'unknown' by `uname -p' and `uname -i'

This patch make `uname -p' and `uname -i' not reporting 'unknown'.
It refers a to Fedora's way to do this.

The coreutils upstream rejects to accept this patch, blaming the
'unknown' result to the kernel not providing enough information.
As on normal distros, `uname -p' and `uname -i' do not report 'unknown',
we refer to Fedora's patch to fix this problem.

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

ncurses_6: Improve installation

As "install.libs" also installs header files, it is
redundant to also call "install.includes".
In fact, doing so can lead to a race, as both targets could
try to install the header files at the same time if running
parallel make. Obviously, with only calling "install.libs",
there is no race with "install.includes".
If there is no race, then the patch fix-include-files-race.patch
is no longer needed.

Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

Revert "selftest: Added MACHINE = "qemux86" to tests that use runqemu"

This is not the correct way to handle this - it significantly increases
the time taken to run oe-selftest anywhere MACHINE is set to some other
value (for example "qemux86-64"), because all of the artifacts for
qemux86 need to then be built as well when running the test. If we need
to skip these tests on non-QEMU machines, the devtool test already
demonstrates how to do that.

This reverts commit 169e1eaa4fc5ed03e2307b68686a7f5b1db37a36.

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

busybox: Drop -r passthrough patch

This was added to help clang compile busybox but since then
the option has been added to llvm and now we have upgraded clang
to 3.8 which has this option available

Fix git recipe to build with mdev feature and update to tip when here

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

linux-yocto/4.1: usb: add usb_otg_caps to usb_gadget structure.

We backported a series of usb OTG commits for BSPs that required
updated support, but this commit was missed. We backport it as
well to fix build issues in some configurations.

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

linux-yocto/4.1: Intel Broxton and Sunrisepoint-H: pinctrl and drm

Integrating the following mainline commits:

  9073fd86a770 pinctrl: intel: Add Intel Broxton pin controller support
  e27702b5f5bd pinctrl: intel: Allow requesting pins which are in ACPI mode as GPIOs
  3466e9acff01 pinctrl: intel: Add support for multiple GPIO chips sharing the interrupt
  5577b683671b pinctrl: intel: Fix compilation warning when !CONFIG_PM_SLEEP
  bd4a97523251 pinctrl: make pinctrl_register() return proper error code
  08422ac780d5 pinctrl: intel: sunrisepoint: Add Intel Sunrisepoint-H support
  e6556d6f8924 drm/i915/audio: add codec wakeup override enabled/disable callback
  4a4d5aab97cc ALSA: hda - Add hduadio support to DEVTABLE

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

linux-yocto/4.1: powercap/RAPL: Backport powercap/RAPL

backporting the following powercap commits:

  24e2e14aa446 powercap / RAPL: Enable Broxton RAPL support
  6be05b47d625 powercap / RAPL: Add support for Skylake H/S
  879e8d02e07d powercap / RAPL: Support Knights Landing
  ec583656ad93 powercap / RAPL: Floor frequency setting in Atom SoC

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

linux-yocto/4.1: Thermal: Enable Broxton SoC thermal reporting device

Mainline backport for thermal reporting.

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

linux-yocto/4.1: usb backports for Apollo Lake/Broxton

Importing a series of mainline backports to support USB on a couple of
Intel platforms:

   b5f6aa9957dd usb: otg: add usb_otg_caps structure for otg capabilities
   bbce8fe2cc42 usb: dwc3: pci: add support for Intel Broxton SOC
   e734e1d9f827 usb: dwc3: pci: Set enblslpm quirk for Synopsys platforms
   1c6bb6694d50 usb: dwc3: Add dis_enblslpm_quirk
   1c6be99e56b8 usb: dwc3: pci: trivial: Formatting
   2f2b89764b97 usb: dwc3: pci: passing forward the ACPI companion
   ea4b3c72d976 usb: dwc3: core: convert to unified device property interface
   dc670b52c69a usb: common: of_usb_get_dr_mode to usb_get_dr_mode
   586fc5174649 usb: dwc3: st: prepare the driver for generic usb_get_dr_mode function
   0624bd9af7ef usb: common: of_usb_get_maximum_speed to usb_get_maximum_speed
   e65bc5467e07 usb: dwc3: Add frame length adjustment quirk
   a90954c5d267 usb: musb: dsps: control musb speed based on dts setting
   b48ff160a993 usb: renesas_usbhs: Allow an OTG PHY driver to provide VBUS
   d1c59752195e usb: chipidea: set usb otg capabilities
   733eada2cdec usb: common: add API to update usb otg capabilities by device tree
   7ab2108dd82b usb: dwc3: core: avoid NULL pointer dereference
   1aedb48b7dc9 usb: dwc3: add ULPI interface support
   07e42a29fb7e usb: dwc3: pci: add quirk for Baytrails
   065917252622 usb: dwc3: add hsphy_interface property
   b2bb32a363a3 usb: dwc3: setup phys earlier
   bf6bb0a6ebb5 usb: dwc3: soft reset to it's own function
   d481da949476 usb: dwc3: cache hwparams earlier
   9ac66262a201 usb: dwc3: store driver data earlier
   5f940588938c usb: dwc3: ULPI or UTMI+ select
   04fdce097f83 usb: dwc3: USB2 PHY register access bits
   b7209213cc05 usb: add bus type for USB ULPI

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

recipetool: create: don't create extra files directory unconditionally

The extra directory next to the recipe should only be created if there
are files to put into it; currently only the npm plugin does this. I
didn't notice the issue earlier because the test was actually able to
succeed under these circumstances if the recipe file came first in the
directory listing, which was a fault in my original oe-selftest test;
apparently on some YP autobuilder machines the order came out reversed.

With this change we can put the oe-selftest test that highlighted the
issue back to the way it was, with an extra check to reinforce that only
a single file should be created.

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

local.conf.sample: Disable prelink by default

prelink is having issues with IFUNC symbols such as tha for fork() in
libpthread. Disable prelink for now until we can get to the bottom of
the issue.

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

oeqa/selftest/recipetool: Fix test_recipetool_create_simple

Correctly handle the any other files/directories that may exist
during the test.

======================================================================
FAIL: test_recipetool_create_simple (oeqa.selftest.recipetool.RecipetoolTests)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/pokybuild/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/meta/lib/oeqa/utils/decorators.py", line 106, in wrapped_f
    return func(*args, **kwargs)
  File "/home/pokybuild/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/meta/lib/oeqa/selftest/recipetool.py", line 414, in test_recipetool_create_simple
    self.fail('recipetool did not create recipe file; output:\n%s\ndirlist:\n%s' % (result.output, str(dirlist)))
AssertionError: recipetool did not create recipe file; output:
NOTE: Fetching http://www.dest-unreach.org/socat/download/socat-1.7.3.0.tar.bz2...
NOTE: Unpacking /srv/www/vhosts/autobuilder.yoctoproject.org/current_sources/socat-1.7.3.0.tar.bz2 to /tmp/recipetool-Uj7MIh/
NOTE: Recipe /tmp/recipetoolqaebTo9s/recipe/socat_1.7.3.0.bb has been created; further editing may be required to make it fully functional
dirlist:
['socat', 'socat_1.7.3.0.bb']

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

Revert "packagegroup-core-x11-sato: add python-pygobject and gtk+3"

This reverts commit 7b60964d87698572f2f13d8dfa966a85f169b698.

OE-Core defaults to no opengl in DISTRO_FEATURES and this would break
sato with the default DISTRO settings. Need to revisit as Sato moves
to gtk+3 but unbreak the builds for now as this change isn't essential.

oeqa/recipetool: Fix syntax error

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

oeqa/recipetool: Improve debugging output by adding dirlist

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

uninative: Add a fix for icu-native to use the correct ABI

If no -std= option is passed to icu's configure, it defaults to CXX11.
This isn't what we want for uninative, so pass an explicit option
which selects an older ABI on newer versions of g++.

This avoids the __cxa_bad_array_new_length@CXXABI_1.3.8 symbol
being used.

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

scripts/oe-selftest: Add short names to most common options

Add short names to most common options in oe-selftest. The options
changed were --run-tests, --run-all-tests, --list-tests and
--list-modules.

[Yocto #9079]

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

gcc: Fix the license on GNU OpenMP

Poky jethro has libgomp ( GNU OpenMP ) license marked as GPL-3.0,
where's in fact the correct is GPL-3.0 with GCC Library Runtime Exception

As stated on https://github.com/gcc-mirror/gcc/blob/master/libgomp/libgomp.h
header license:
 ...
   Under Section 7 of GPL version 3, you are granted additional
   permissions described in the GCC Runtime Library Exception, version
   3.1, as published by the Free Software Foundation.
 ...

Signed-off-by: Helio Chissini de Castro <helio.castro@bmw-carit.de>
Signed-off-by: Ross Burton <ross.burton@intel.com>

Revert "gcc: Fix the license on GNU OpenMP"

This reverts commit 892fbe373c5cff7b2f28b58aa2508b47e53d3e63.

perl: fix missing dependency for perl-misc

"perl-misc" package is adding tens of perl utilities which have
depencies on perl-modules.

For example, in order for "prove" to work correctly following modules
are needed:

RDEPENDS_perl-misc += "perl perl-module-app-prove
perl-module-overloading perl-module-tap-base perl-module-file-glob
perl-module-tap-formatter-console perl-module-tap-formatter-base
perl-module-tap-formatter-file perl-module-tap-formatter-session
perl-module-tap-parser perl-module-tap-parser-aggregator
perl-module-tap-parser-scheduler"

If we compile a list of modules needed by the utilities added by
"perl-misc" we may end up having to add hundreds of modules to
RDEPENDS_perl-misc.

Rather than adding hundreds of dependencies only perl-modules was added.

Signed-off-by: Catalin Enache <catalin.enache@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

classes/buildhistory: record a few more variables for extensible SDK

We've recently added SDK_EXT_TYPE and SDK_RECRDEP_TASKS and these will
have an effect on the size and functionality of the SDK, so record them
in sdk-info.txt.

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

package-deb: Ignore circular dependencies

If a package depends on itself apt-get fails to install it with the
error attached to this patch.

This patch checks for this conditions and notifies the user so the
recipe maintainer can fix his RDEPENDS variable.

root@qt5022:~# apt-get install perl-module-cpan
Reading package lists... Done
Building dependency tree
Reading state information... Done
The following NEW packages will be installed:
  perl-module-cpan
0 upgraded, 640 newly installed, 0 to remove and 0 not upgraded.
Need to get 0 B/5964 kB of archives.
After this operation, 0 B of additional disk space will be used.
Do you want to continue? [Y/n]
E: Internal error: MaxLoopCount reached in SmartUnPack (2) for
perl-module-cpan:amd64, aborting
E: Internal error, packages left unconfigured. perl-module-cpan:amd64
root@qt5022:~# apt-get install perl-modules

Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

package_deb: Fix python runtime error

The dictionary cannot be used for iteration if its size may change
during the loop.

ERROR: Error executing a python function in
/home/ricardo/curro/qt5022/build-qt5022/repo/yocto/meta/recipes-devtools/perl/perl_5.22.0.bb:
Exception: RuntimeError: dictionary changed size during iteration

Acked-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

python-numpy: fix buildpaths QA issue

Remove build path in comments:

Without the fix:
...
|This file is generated by /buildarea/raid0/hjia/build-20160225-yocto-
buildpath/tmp/work/core2-64-poky-linux/python-numpy/1.10.4-r0/numpy
-1.10.4/setup.py
...

With the fix:
...
|This file is generated by /numpy-1.10.4/setup.py
...

[YOCTO #7058]

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

python: move ast module into python-core

The compile() method returns objects that inherit from ast.AST so it's best that
python-core contains this class.

[YOCTO #8684]

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

xserver: require sufficiently new libdrm

This fixes the check of the libdrm version in the configure script.

Signed-off-by: Johannes Pointner <johannes.pointner@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

package_manager.py: Fix race condition in OpkgIndexer.write_index()

When writing the index using ipk packages there could be a race condition
when populate the index. This happens because the architectures
are repeated (specially all) and the commands generated to write the index
run in parallel.

This change avoid the duplication of commands using a set instead of a list.

[YOCTO #8924]

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

scripts/oe-selftest: Add search expression matching to run/list options

The oe-selftest script required an exact matching for the parameters
passed to its run-tests-by and list-tests-by options. Many tests
can be retrieved here and filtering is a must.

This patch add this filtering functionality by enabling the use
of wildcards such as "*".

[Yocto #8916]

Signed-off-by: Humberto Ibarra <humberto.ibarra.lopez@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glib-2.0: relocate the GIO module directory for native builds

Glib hard-codes the install path in search path for GIO modules, which causes
problems when glib-2.0-native is restored from sstate with a different build
directory.

In the future we should relocate symbols directly using the same system that the
eSDK uses, but for now use dladdr() to look up where the library was loaded from
to build the search path.

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

image-buildinfo.bbclass: fix performance problems

Inheriting image-buildinfo.bbclass primarily slowed down image
building for two reasons:
1. The content of the shell command "buildinfo" gets expanded
   multiple times, each time again checking the state of all
   layers.
2. When expanded as part of the actual image creation, git
   is invoked under pseudo, which makes the check quite a bit
   slower (from a few seconds to a minute with many layers).

To fix this, buildinfo now is a Python method which calls the checks
only when really executed. Pseudo is told to unload itself when
starting git.

In addition, "git diff" is invoked with "--quiet", which avoids
producing output that is just getting thrown away. As before, any kind
of problem or output causes the layer to be marked as "modified".

[Revision 2 of the change with some dead code removed]

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.4: gpio-pca953x: add "drive" property

Integrating the following commit for the 4.4 kernel:

[
    gpio-pca953x: add "drive" property

    Galileo gen 2 has support for setting GPIO modes. Expose these
    properties through the GPIO sysfs interface. This approach is
    bit hacky, since it changes the interface semantics.

    The original patch was by Josef Ahmad <josef.ahmad@linux.intel.com>
    and  made on top of kernel 3.8.
]

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

python3: fix do_configure check platform triplet error

For p1022ds bsp, the MULTIARCH is powerpc-linux-gnuspev1 and
python3 did not recognize the extra 'v1' which caused python3
configure error for the platform triplet.

Q:Why Python3 check platform triplet?

A:Under Linux, GNU/KFreeBSD and the Hurd, C extensions now include
the architecture triplet in the extension name, to make it easy to
test builds for different ABIs in the same working tree.

vim config.log
...
EXT_SUFFIX='.cpython-35m-powerpc-linux-gnuspe.so'
...

Here is the generated C extensions which take platform triplet into account.
...
|image/usr/lib/python3.5/lib-dynload/_datetime.cpython-35m-powerpc-linux-gnuspe.so
...

https://bugs.python.org/issue22980
https://www.python.org/dev/peps/pep-3149/
https://bugs.python.org/review/22980/patch/14593/54808

[YOCTO #9226]

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ncurses_6: Fix an install race condition

Both targets install.libs and install.includes install the same
files, resulting in a race condition when running parallel make.
This race is addressed in a patch file, making sure only one
of the targets (install.includes) installes the include files.
This will work properly (i.e.ncurses will install as intended
by the recipe) as long as we always install both targets.

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

build-appliance: make the inclusion of downloaded sources optional

Including the entirety of DL_DIR in the generated build appliance
image adds a significant amount of space and makes the build
appliance image more awkward to distribute. Add a configuration
option to make the inclusion of sources option and default to
disabling this functionality.

Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

builder: remove hob from autostart

hob has been removed so don't try and autostart it with the mini-x
session in the build appliance.

Remove the please_wait_dialog program which informed the user to
wait for Hob to start.

Rename the mini-x autostart file to reflect the removal of hob, we
now just start a matchbox-terminal with the environment configured.

Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Revert "gstreamer1.0-plugins-XXX: move inherit gettext into common .inc file"

Do not assume that the recipes in OE-Core are the only ones using
gstreamer1.0-plugins.inc or that all GStreamer plugins support
localization. Without this reversal, all our GStreamer plugin recipes
started spewing out:

configure: WARNING: unrecognized options: --disable-nls

This reverts commit 282fa5ff059db6f6daa108d5ec4ea3d0bfbe0fe2.

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

musl: disable building of gobject introspection data

When musl is in use, GLib binaries crash on assertion failures under qemu

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

machine/include/arch-x86: Make x32 ABI not supporting gobject-introspection-data

x32 isn't supported by user mode qemu so we can't build
gobject-introspection-data, so disable it in this case.

(From OE-Core rev: 4ee1eb8ddd3fbe144fbaeb32e07b66e191aa7548)

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

bitbake.conf: add 'gobject-introspection-data' to DISTRO/MACHINE_FEATURES_BACKFILL

This means that introspection data will be generated using qemu by default.

If this causes trouble (qemu does not always emulate target architectures
perfectly), then add
MACHINE_FEATURES_BACKFILL_CONSIDERED = "gobject-introspection-data"
to machine configuration or
DISTRO_FEATURES_BACKFILL_CONSIDERED = "gobject-introspection-data"
to distro configuration.

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

packagegroup-core-x11-sato: add python-pygobject and gtk+3

This allows easy testing of introspection for example using
http://python-gtk-3-tutorial.readthedocs.org/en/latest/introduction.html

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

webkitgtk: enable gobject introspection

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

recipes-gnome: fix introspection support

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

python-pygobject: update to 3.18.2

The new version of pygobject relies in gobject introspection;
this has prevented the update previously.

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

gnomebase.bbclass: do not disable gobject introspection

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

gstreamer: enable gobject introspection

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

libsoup-2.4: enable gobject introspection

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

clutter: enable gobject introspection

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

gtk+3: enable gobject-introspection

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

gtk+: enable gobject introspection

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

avahi: enable gobject-introspection

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

python-pygtk: remove the recipe

python-pygtk hasn't been updated in several years, is incompatible
with the current version of pygobject, and is generally obsolete as
all modern python gtk apps use introspection directly.

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

avahi-ui: remove the dependency on python-pygtk by disabling avahi-discover

python-pygtk is removed in a separate commit; the reasons for
that are explained in that commit's message.

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

vala.bbclass: remove pre-packaged vapigen.m4 from tarballs

We need to make sure that our fixed vapigen.m4 is used instead.

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