nspr: Also update nspr.pc to 4.9.5

Earlier commits bumped the upstream version, but we need to update
our copy of the pkg-config file too.

(It'd probably be better to generate this at build time, otherwise
 this will be a trap people continually fall into)

(From OE-Core rev: 14878f34645f5e9a63d3c3be6d6fe558bbda9900)

Signed-off-by: Colin Walters <walters@verbum.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

uclibc-git: Fix build on x86 and move to latest master

debugedit from rpm has unearthed a bug in uclibc
where it was mixing stabs with elf/dwarf

(From OE-Core rev: be9f1c63aff93f9cdcb69d6cf5b4639690602af6)

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

udev: move /run volatile entry to udev instead of initscripts

initscripts is generally installed on systemd-using images, but because it
specifies that /run is a symlink to /var/run managed by volatiles it totally
breaks systemd by copying/deleting /run from underneath systemd.  Deleting
sockets mid-boot doesn't leave systemd in a happy place.

As this volatile reference of /run was introduced by udev 182, move it's
reference to the udev recipe.  This way it will never be present on systemd
images, as systemd manages /run as a tmpfs itself.

(From OE-Core rev: 5b0257e318340c2d6c8d3b0c3fa32272d6e9526b)

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

busybox: order and group initscript variables logically

(From OE-Core rev: 94acb39385a14d54503db08351a717449e2d4b50)

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

dbus: explicitly disable systemd when no systemd

If systemd isn't a distro feature, explicitly disable the systemd unit path
check as otherwise it will search the sysroot.

(From OE-Core rev: 7c39f21cbde23ad678ddf54cb54b7f01e971a325)

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

systemd: recommend systemd-compat-units

These are more than useful as they ensure some services are not started twice,
and cause the first-boot postinstalls to run.

(From OE-Core rev: c254ab4e3bdc4a3ba18e89219594fffa7895184d)

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

systemd-compat-units: disable dbus-1

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

Revert "base-files: add fstab for systemd based systems"

For hybrid systemd/sysvinit builds, only one fstab can be used.
The default fstab used by sysvinit should work fine with systemd.

Since virtually every machine will ship its own fstab in its bsp
layer, the bsp layer may decide how to override the fstab based
on distro features.

This reverts commit 77bbb839ba25b974a538b90d346b454ccd5deefd.

(From OE-Core rev: e9352e8a43639564af0a97f5e8a642e0989b0256)

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

packagegroup-core-boot: revert to specifying sysvinit as default init manager

Don't follow DISTRO_FEATURES_INITMAN as that makes the packagegroups rebuild if
you switch init manager.

As in hybrid situations there's generally a clear primary and minimal init
manager choice, so change VIRTUAL-RUNTIME_init_manager to set the primary init
manager, and roll your own groups/images for the secondary.

(From OE-Core rev: 7480814753bacbb6363125fce0738a93a602bcc9)

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

default-distrovars: remove obsolete DISTRO_FEATURES_INITMAN reference

(From OE-Core rev: d75dbfc34dcefb5b37b2e7e79a3d4e1a7903883d)

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

bitbake.conf: explicitly backfill sysvinit, not DISTRO_FEATURES_INITMAN

Reflect reality by backfilling sysvinit support, instead of whatever value was
in DISTRO_FEATURES_INITMAN.

(From OE-Core rev: 0b6559cd93a64498646d18a121746c6816382407)

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

default-distrovars: don't add INITMAN to DISTRO_FEATURES and DISTRO_FEATURES_BACKFILL

DISTRO_FEATURES_INITMAN is going away as it's not useful in a hybrid init script
environment.

(From OE-Core rev: 7afd57993277ae7aa30e56edda327bb5f28ad153)

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

update-rc.d/systemd: change communication variable name

Rename SYSTEMD_BBCLASS_ENABLED to INHIBIT_UPDATERCD_BBCLASS to reflect the
action, for clarity.

(From OE-Core rev: cf43320c343437659aee94acd005bf7712f273cd)

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

systemd: add udev init script for hybrid sysvinit/systemd usage

With both sysvinit and systemd features it's possible to use systemd's udev with
sysvinit, so add the required init script.

(From OE-Core rev: b58a176936740e8e291f1e82229a8ca044bdb044)

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

systemd: check for systemctl first, and don't force systemd to be installed.

With both sysvinit and systemd features enabled these postinsts may actually run
on a target without systemd, so check that systemctl is present before using it.

(From OE-Core rev: ac00e56cb9daacef17a6fdebe7b8ca1667b7e1c4)

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

update-rcd: drop depends to recommends, check for update-rcd in scripts

(From OE-Core rev: 2c403979c03898c679c5a1e1092aec784dbeb77c)

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

util-linux: split uuidd into it's own package, and enable for systemd

(From OE-Core rev: cd3c8c9cea560a584178ed831bfc3c014b6663e6)

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

update-rcd.bbclass: handle both sysvinit and systemd features being present

Run the helper if the sysvinit feature is present, or if the systemd feature is
present but the systemd class hasn't been inherited.  We want to run in the
latter case as systemd has sysvinit compatibility, but we don't want to always
run so that pure systemd images don't have redundant sysvinit files.

(From OE-Core rev: a3856ab19f1d4ae312f559521785ad4384700729)

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

default-providers: change udev selection logic

Change the logic so that the udev provider is the standalone udev, unless the
systemd DISTRO_FEATURE is set.  The previous logic was designed to fail if both
sysvinit and systemd were enabled, which we're supporting now.

(From OE-Core rev: f5d018a769fa297efa629cbbf6e42a49173faa8b)

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

systemd: split out the hwdb data

The hardware databases are not essential and also quite large, so split them out
into udev-hwdb.

(From OE-Core rev: 3e8da06c1faeb7884689a8af959cd9fa5bdf4e4f)

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

systemd: don't depend on the PCI/USB databases

systemd ships its own databases (hwdb), so we don't need another copy.
--with-pci-ids isn't recognised by configure, so remove it.

(From OE-Core rev: 69abfae6c81c8d7e7920817a55c3bea84615446d)

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

core-image-minimal-initramfs: don't install busybox-syslog

This uses BAD_RECOMMENDATIONS which isn't supported by all package backends, but
it's a start.

(From OE-Core rev: 996b3e344838db40ef8ef17efd719b830c23fa40)

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

systemd: merge udev-systemd into udev

Merge the contents of udev-systemd, which is just the service files, into udev
itself.  This split wasn't intended to ever happen in oe-core.

(From OE-Core rev: c54970c5ce85a6155ed00cbb4044e1830f9538bc)

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

systemd: make xz support (compressed journal) optional, defaulting to on.

Compressed journals means using liblzma, sf the journal isn't going to be used
this can be disabled.

(From OE-Core rev: 5dcfe269c844673102beaacc6007fbd49f6b6d90)

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

busybox: add strictatime support to mount

systemd uses strictatime when mounting tmpfs.  Luckily this is already supported
upstream, so backport the patch from git.

(From OE-Core rev: 7379a5a2035ef670329551783c372d9310ddd983)

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

busybox: enable systemd integration for syslogd

(From OE-Core rev: cf3618f9a57e46fb78d5be35d473e2dd5290e961)

Signed-off-by: Radu Moisan <radu.moisan@intel.com>
Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/rm_work: rename RM_WORK_WHITELIST to RM_WORK_EXCLUDE

This really functions as a blacklist, not a whitelist, since we are
listing recipes to exclude. To avoid any possibility of confusion, since
this was a recent addition, rename the variable.

(From OE-Core rev: 4d56de8018d550b3d181d451900cbfb698d64141)

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

binutils: Don't generate .PPC.EMB.apuinfo sections for 64-bit output

This can cause issues building ppc64/eabi kernel. For details
see the patch header

(From OE-Core rev: 3c6622e22b0e1b2a8ceea6465ea84c6fb8299518)

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

rpm: Fix debugedit buildid processing

[ YOCTO #4089 ]

When constructing a new buildid, the items being hashed need to be
returned to their native endian.  In the process we were munging
the sh_type field that we relied on to determine if a section was
loadable or not.  The patch avoids this behavior by only modifying
a copy of the local endian data.

(From OE-Core rev: ac4d2d44c88cace8dbce0c8e7df3fd1f2ed244b4)

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

grub-efi-native_2.00.bb: fix a build issue

Due to recent change in the oecore layer grub-efi recipe uses separate
builddir and the paths used in the do_deply need to change accordingly.

This change avoids this build issue:
install: cannot stat `/build-fri2/tmp/work/x86_64-linux/grub-efi-i586-native/2.00-r1/grub-2.00/bootia32.efi': No such file or directory
ERROR: Function failed: do_deploy (see /build-fri2/tmp/work/x86_64-linux/grub-efi-i586-native/2.00-r1/temp/log.do_deploy.5875 for further information)

(From OE-Core rev: 11ba0231244b8a27939969353e3aa668ce78f7c6)

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.8: qemuarm boot fix

The preempt-rt kernel was wasn't inheriting the standard kernel
configuration and hence was not able to boot on qemuarm due to
missing ABI options.

Changing the main preempt-rt kernel include to inherit standard
configuration again fixes this boot issue.

(From OE-Core rev: d569f44c296f6de1f7bb6ad9f581c237d5947321)

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

linux-yocto/3.8: fix perf host contamination

Updating the 3.8 SRCREVs with the following fix:

    libtraceevent: Remove hard coded include to /usr/local/include in Makefile

    commit b9e8c37220c80e78289a1e87b50c09418eb59a7e upstream

    having /usr/local/include hardcoded into the makefile is not necessary
    as this is automatically included by GCC. It also infects cross-compile
    builds with the host systems includes.

Signed-off-by: Jack Mitchell
Acked-by: Namhyung Kim
Cc: Ingo Molnar
Cc: Paul Mackerras
Cc: Peter Zijlstra
Link: http://lkml.kernel.org/r/1362741712-21308-1-git-send-email-ml@communistcode.co.uk
Signed-off-by: Arnaldo Carvalho de Melo
    This implements the fix I described in Yocto [BUG #3993][perf using
    host includes], now upstream.

Integrated-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
[YOCTO #3993]

(From OE-Core rev: 90d9147068a6e2c766976b2092d5e188c45a9040)

Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.4: import v3.4.36 -stable fixes

Updating the 3.4 tree to import the korg -stable fixes for v3.4.36.

(From OE-Core rev: 19d77ddf6c5d7822b757342a4e41ea68403889c3)

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

kern-tools: initialization and meta branch fixes

Bumping the kern-tools SRCREV to pickup the following fixes and documentation
updates:

  d484e3f kgit-meta: remove hardcoded meta directory name
  affad20 yocto-kernel-tools: Typoes, "fragement", "depreciated"
  142ed49 kgit-init: update tools list

(From OE-Core rev: 65113af811afcf53d3056d372861cd4d1a6bff07)

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

linux-yocto/3.8: introduce v3.8.4-rt1

The -rt series has been updated for the 3.8 kernel. Updating the
SRCREVs to reflet its import into the tree.

(From OE-Core rev: 1daa242e78a5fddb1ae60b79f990811c85058943)

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

linux/yocto-3.4: update to v3.8.4

Updating the base v3.8 kernel to the 3.8.4 -stable version.

(From OE-Core rev: 81e599c9f24c0a8937694c66ae349fcb8f618f38)

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

linux-yocto/3.8: allow kernel feature _appends to be overriden

It was pointed out that the current way the KERNEL_FEATURES variable
is appended in the base linux-yocto recipe doesn't allow the appended
features to be prevented in a layer without using python code and
a recipe finalize hook.

To allow easier overriding of 'extra' or 'optional' features that are
defined in the linux-yocto recipe, we create a KERNEL_EXTRA_FEATURES
variable. This variable can be set in a layer to define extra features
or cleared to prevent the recipe's extra features from being appended
to the core functionality.

(From OE-Core rev: 779cb6e51fab87e00784bd97ec4771e69a79cf82)

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

qemuimagetest: collect and print runqemu output

If runqemu (or qemu itself) fails we need to know why, so tee out to a
log file and print it when we can't find the qemu process or determine
its IP address.

(From OE-Core rev: 827106a57ca88760a19f9309d859b500c5c4fe97)

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

qemuimagetest: fix erroneous ps errors when qemu couldn't be started

The helper script was printing an error to stdout when it couldn't find
any qemu child processes; output this error to stderr instead and
redirect stderr to /dev/null when running from qemuimage-testlib so that
QEMUPID is actually blank if there are no qemu instances found.

(From OE-Core rev: 7c2137a07cca8e1d40d3c8b4b2c6321d80f2b1de)

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

eglibc-locale: Ensure files have correct ownership

The copy operation leaves the files owned by the person running the
build which results in warnings in do_package_write_ipk like:

*** Warning: The following files have a UID greater than 99

and incorrect ownership in the packges. This patch addresses this
ownership problem.

(From OE-Core rev: 90fc8efb88f82f111523e6584e66ae4ff5851426)

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

font-alias: Add depends on font-util-native

Without this, the font-util.m4 macro can be missing leading to an undefined
macro. The recipe will still build but generate an empty font-alias package
since the files are installed into "${D}@XORGROOTFONTDIR@".

(From OE-Core rev: 12dc7ae22ddaae0d79e0f86b66c1f5a9b18329f9)

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

opkg-utils: Update to newer git rev for bashism fix

On systems with dash as /bin/sh there were failures in do_package_write_ipk
due to bashisms in opkg-build. This newer revision contains a fix for this.

(From OE-Core rev: 0e185b86e80989a7904292f5866540a049cc4daa)

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

package.bbclass: Correctly handle /usr/src/debug file ownership

After the recent cached path operation improvements, we need to tell
the cache about directory changes. /usr/src/debug was was cached as
not created and this was leading to bugs in file ownership. This
change updates the cache and avoids this issue.

(From OE-Core rev: 553a8010e81448629a94ab04fd989e6e49066f2b)

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

docbook-utils-native: Unbreak the build after source and build dir split

docbook-utils-native started to install frontends and backends in
wrong locations after this change. The absolute build and source
directories were being created in install locations and the recipes
using docbook-utils-native were complaining like dbus-java is saying

| jw: There is no frontend called
"/b/kraj/jlinux-next/poky/build/tmp-eglibc/sysroots/x86_64-linux/usr/share/sgml/docbook/utils-0.6.14/frontends/docbook".
| make: *** [CreateInterface.1] Error 2
| ERROR: oe_runmake failed

Rightly so because now they are installed under
/b/kraj/jlinux-next/poky/build/tmp-eglibc/sysroots/x86_64-linux/usr/share/sgml/docbook/utils-0.6.14/b/kraj/jlinux-next/poky/build/tmp-eglibc/work/x86_64-linux/docbook-utils-native/0.6.14-r3/

This patch fixes the install to be like it was but consider
the build and sourcedir split.

(From OE-Core rev: d60aec3f5b319f4583fa72a8357b9ddd3be62b1a)

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

at: Fix --with/without pam options

The --with/--without pam support in configure appears to have been dropped
leading to builds which can detect host headers, then fail or result in non-
determinstic builds. Add an explicit option to fix this detection.

(From OE-Core rev: 4e1ea0e22cf5310dbe18b5427a91d86cc4a136fe)

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

package.bbclass: Ensure debug source file is generated correctly

The command for generating debug sources appends to the file, therefore we need
to ensure it does not exist before we run the command else we end up doing
more work that we should have to.

(From OE-Core rev: 48dfd61901ce07491ef913cf04dbd9de9a52759f)

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

conf/machine: Clean up MACHINEOVERRIDES handling

OVERRIDES reads from left to right, least to most specific. We were
appending to MACHINEOVERRIDES when we should have been prepending so
the ordering of qemuall verses qemuxxx was incorrect, as was the x86
override and several of the arm overrides. This patch is a batch cleanup
of the various issues to correct the order from least to most specific.
The include order does matter and we needed to tweak some of that in this
patch too.

[YOCTO #4090]

(From OE-Core rev: bdc1b214431c9c93a929b547b9a61e7b87fbd366)

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

package.bbclass: Handle subprocess errors correctly

If an error occurs in subprocess.call() we currently don't catch it. In particular
we have issues where debugedit is segfaulting unnoticed. This fixes up
various code paths to catch the errors.

[YOCTO #4089]

(From OE-Core rev: 262a69ffd33e9d001a7a15fc73671a015e3b5dd1)

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

gnupg: Add symlink for gpgv

Gnupg 2.0.19 installs 'gpgv2' but apt-get and possibly other utilities expect
to find this as 'gpgv'. A symlink is created in the same way as the link for
'gpg'.

(From OE-Core rev: 32f3596cefc034398803032785714b289766bb66)

Signed-off-by: Paul Barker <paul@paulbarker.me.uk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

rpm: Add workaround for debugedit-segv

[ YOCTO #4089 ]

On PPC and MIPS, there appears to be a condition that causes
debugedit to segfault.  The segfault is related to a call into
the md5hash algorithm, an address of '0', and a size > 0 is passed
causing the access of the address to segv.

This workaround may prove to be the final fix, but it's currently
unclear what the actual cause of the 0 address is.

(From OE-Core rev: a046029eb96cd9307253937ceeadafaaa6d06dce)

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

nspr: remove unnecessary files from the main package

This patch does the following:
 * removes compile-et.pl and prerr.properties from the main package
   because these files are needed only at compile time;
 * moves nspr-config script to nspr-dev package where it belongs;
 * adds 'perl' RDEPENDS for nspr-dev since the package contains a perl
   script for running some tests;

(From OE-Core rev: d70df34bcd6b345f600624f289ce688e5dd90f7e)

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

bitbake: bitbake/fetch: Add git submodules fetcher

This adds very basic git submodule support to the fetcher. It can be
used by replacing a git:// url prefix with a gitsm:// prefix, otherwise
behaviour is the same as the git fetcher. Whilst this code should be
functional, its not as efficient as the usual git fetcher due to the
need to checkout the tree to fetch/update the submodule information. git
doesn't support submodule operations on the bare clones the standard git
fetcher uses which is also problematic.

This code does however give a starting point to people wanting to use
submodules.

(Bitbake rev: 25e0b0bc50114f1fbf955de23cc0c96f5f7a41e3)

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

bitbake: doc/bitbake.1: Update with missing parameters

Patch from Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com> to
add missing parameters to bitbake man page. Added hob to ui list too.

[YCOTO #4049]

(Bitbake rev: ca7cd6c1318e0ef066f9e12e7516a47b2af3a7d1)

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

packagegroup-core-lsb: add nspr

Add nspr for LSB library check.

[YOCTO 4015]

(From OE-Core rev: dd0c9e755d47cd86cec3a3bae2482969460f2a3b)

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

xf86-input-evdev: add mtdev dependency

This is needed in order to have multitouch protocol enabled.

[YOCTO #4087]

(From OE-Core rev: be7e4da2a402ddea196b8b25f8ed9e9647b50563)

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

lttng: babeltrace: Update to upstream version 1.1.0

Also updated the BUGTRACKER

(From OE-Core rev: 851ebb58b6d7585e1861f4d16340d748767545e4)

Signed-off-by: Maxin B. John <maxin.john@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libxcb: fix repeated configures

If configure is re-ran on an existing build tree the string substitutions we
need are done twice, resulting in invalid paths.  Anchor the expressions so they
only match a pristine configure.ac.

(From OE-Core rev: a167176c3e41e4eb2a1931df566367e3da2b3b86)

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

separatebuilddir.inc: mxsldr should never have been added to this list, remove

(From OE-Core rev: 0d3a5282bd84edd4100fc8855f5c4ab7d6016803)

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

tcl: Fix the location of the installed headers

Having '${S}' in the for loop was causing the headers to be installed
into the wrong location.  Move the 'S' to the install line.

(From OE-Core rev: 41c0241a810f0a97ddc98a834e717645e0047958)

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

valgrind: Fix out of tree builds

(From OE-Core rev: 1944d362866fef1af406ed50955f1ed7cd9c29a8)

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

package/populate_sdk: Move functions from package_* to populate_sdk_*

This fixes build failures introduced with "classes/buildhistory: implement history
collection for SDK" by moving the functions to files where only the specific image
type which is enabled is inherited. The failures occured when multiple PACKAGE_CLASSES
were enabled.

(From OE-Core rev: 9a414a57ad69a426a8d8a2970c42ca7427240615)

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

libpcre: Fix BUILD_CFLAGS for out of tree support

The BUILD_CFLAGS were broken since /include no longer exists in this package. We
do need to include ${B} in the include path for pcre.h though.

(From OE-Core rev: d9130e5113c8f93f327fbe19dbfe39036c1c3995)

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

sudo: Fix out of tree builds ${B} != ${S}

The last change to sudo broke out of tree builds, fix this.

(From OE-Core rev: a1226175c405db6eacb6e490e8e635b9dd156126)

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

Perl packages shouldn't be allarch

Perl packages should know about arch so that they could install to the
correct libdir. Or else they always install to the default libdir in
multilib builds.

(From OE-Core rev: a6b186a1a5cfc6ae9c099af0ed4217d91f5d4ef3)

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

packagegroup-core-tools-debug: split out Eclipse packages

Split out the packages added for Eclipse remote debugging to a separate
package group so that we can avoid pulling them in if not using Eclipse.

Fixes [YOCTO #3251].

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

packagegroup-core-tools-debug: remove rsync

This was added for supporting Anjuta, which is no longer actively
supported; in any case the anjuta-remote-run package RDEPENDS on rsync
so it will be brought in by that if that is added.

(From OE-Core rev: 21375469094e58c6ee860d656f0c997ff1e3d79c)

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

sudo: Compile mksigname and mksiglist for build host

cross compiling sudo doesnt work well since it uses
mksigname and mksiglist to generate C sources which
are then used in sudo build itself. With this patch
now we make sure those hosttools are compiled for
build machine. It fixes the build failures like

./mksigname > signame.c
/bin/sh: ./mksigname: cannot execute binary file
make[1]: *** [signame.c] Error 126

(From OE-Core rev: 2ce5b1364ef6eb276352f7170d23dc3492c779f6)

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

polkit: remove

Previous commits have meant that PolicyKit isn't used by default anymore, so
remove this now that it's been integrated into meta-gnome.

Rationales for this move inludes that PolicyKit isn't "core" for embedded
systems, and future versions require the SpiderMonkey JavaScript runtime.

(From OE-Core rev: 8c9fad86d8bd86f13d61a1a5cd65d12b16a9e0de)

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

linux-firmware: Package Broadcom firmwares

Adds specific packages for bcm4329, bcm4330 and bcm4334, including
handling of symbolic link for the needed firmware filename.

(From OE-Core rev: a8b8fb0447e8fb72050d1d9775ccb0e2b894a477)

Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cairo: explicity disable lzo

By explicitly disabling the LZO check, we ensure a deterministic build

[YOCTO #4076]

(From OE-Core rev: 7d855cb624f9330e185ff0f03b5b317d805eda09)

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

classes/buildhistory: trim trailing spaces in file listings

These are mildly annoying when viewing git diffs of the buildhistory
repository, so let's just get rid of them.

(From OE-Core rev: c257f292bf75061647e380889487c7e2625592de)

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

classes/buildhistory: implement history collection for SDKs

SDKs are constructed in a similar manner to images, and the contents can
be influenced by a number of different factors, thus tracking the
contents of produced SDKs when buildhistory is enabled can help detect
the same kinds of issues as with images.

This required adding POPULATE_SDK_POST_HOST_COMMAND and
SDK_POSTPROCESS_COMMAND variables so that data collection functions can
be injected at the appropriate points in the SDK construction process,
as well as moving the list_installed_packages and
rootfs_list_installed_depends functions from the rootfs_{rpm,ipk,deb} to
the package_{rpm,ipk,deb} classes so they can also be called during
do_populate_sdk as well as do_rootfs.

Implements [YOCTO #3964].

(From OE-Core rev: c3736064483d4840e38cb1b8c13d2dd3a26b36aa)

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

libuser: Use abs_srcdir instead of srcdir

srcdir is still relative to pwd. We need to use
abs_srcdir to get absolute source directory names

(From OE-Core rev: 9297e01b2095d90b5365d23c10de699a7cf411ea)

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

scripts/lib/bsp: replace mesa-dri with mesa in machine.conf files

mesa-dri has been renamed to mesa.

[YOCTO #3385]

(From meta-yocto rev: ba8d5b6dcb6fa4721e85b62f15713072cc0fa23f)

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

vte: Fix conflict between FILES_${PN}-dbg and FILES_vte-dbg

Due to quarks with the way bitbake handles variable key values, the
hard coded FILES_vte-dbg or automatic FILES_${PN}-dbg could replace
each other, leaving the system in an inconsistent state.

(From OE-Core rev: 7846f68537a942d340d5931e23a4fceb84b6edcb)

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

mesa: rename mesa-dri recipe to just mesa

Rename mesa-dri recipes to just mesa. Also, replace all references to
mesa-dri in all recipes/configs.

The reason for this renaming (quote from bugzilla):

"mesa-dri is a artefact of mesa-xlib existing, which doesn't anymore.
mesa-dri should be renamed to mesa."

[YOCTO #3385]

(From OE-Core rev: c8bbb9983bcc7cfc5332e89c3e8148505b4ca83f)

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

fix march sanity check issue

1. check if gcc_test is really generate before os.remove("gcc_test") to avoid
   following error:
      ERROR: Execution of event handler 'check_sanity_eventhandler' failed
      Traceback (most recent call last):
      File "check_sanity_eventhandler(e)", line 4, in check_sanity_eventhandler(e=<bb.event.ConfigParsed object at 0x3151450>)
      File "sanity.bbclass", line 107, in check_sanity(sanity_data=<bb.data_smart.DataSmart object at 0x11ba110>)
      File "sanity.bbclass", line 22, in check_gcc_march(sanity_data=<bb.data_smart.DataSmart object at 0x11ba110>)
      OSError: [Errno 2] No such file or directory: 'gcc_test'
2. set result to False when build failed with -march=native to ensure
   -march=native is appended to BUILD_CFLAGS  when host gcc really supports this flag,
   otherwise following error appears when build native packages.
      | cap_text.c:1: error: bad value (native) for -march= switch
      | cap_text.c:1: error: bad value (native) for -mtune= switch

(From OE-Core rev: 4a4228fe250c8b23a5deeb25825d61c6e84a47a2)

Signed-off-by: Zhenhua Luo <zhenhua.luo@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

perl: remove the duplicated RDEPENDS

Remove a duplicated line:
RDEPENDS_${PN}-module-cpanplus += "${PN}-module-load"

[YOCTO #4064]

(From OE-Core rev: 5660aac1cf3f71ce14b58f2c0e69460732f09d7c)

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

perl-native: remove duplicated LIC_FILES_CHKSUM

Remove duplicated LIC_FILES_CHKSUM:
LIC_FILES_CHKSUM = "file://Copying;md5=2b4c6ffbcfcbdee469f02565f253d81a \
                   file://Artistic;md5=f921793d03cc6d63ec4b15e9be8fd3f8"

[YOCTO #4064]

(From OE-Core rev: 9b298098401c61761bd65fc1d9efff2197cc45a5)

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

pm-utils: remove duplicated RDEPENDS

Remove duplicated RDEPENDS_${PN} = "grep".

[YOCTO #4064]

(From OE-Core rev: 770c846ec5b9b6c555afbb7ce0ff116dbbfb519e)

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

bitbake: Adjust spacing on information windows

[Hob #4070]

- tweaked the spacing on info windows

- fixed text bolding in simple settings dialog

(Bitbake rev: 2ecb102968cdbbdbbfa91e1dcccf45bcd0b59a89)

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

bitbake: hob: implementation of search functionality in Hob

Implemented the search functionality for recipes and packages using
filters on the listmodel. I have used the design which can be found in
bugzilla.

[YOCTO #3529]
(Bitbake rev: b77166ad7b8571895f73a84f7789d93fbd4f6d04)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob: Hob custom image build error

For a custom image from scratch, the image version is not set and
it gives an error.

[YOCTO #4065]
(Bitbake rev: 43703f3eb5f71a117e6315fa5df91711653636a3)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mkefidisk: Use msdos partition tables

While GPT works fine when writing to actual media, it cannot be reliably
used for distributing disk images as it requires the backup table to be
on the last block on the device, which of course varies from device to
device. Use MSDOS tables instead.

Use mkfs to label the filesystems as msdos tables do not support
partition labeling.

(From OE-Core rev: 049ea1e0a6a1017a5020de38bd7ce93515bd62f4)

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

mkefidisk: Remove startup.nsh

Most firmware implementations use the EFI specified
EFI/BOOT/bootia32.efi (and similar) boot paths. Only broken firmware
uses different paths for removable media. In those cases, the user can
add their own startup.nsh.

For the compliant case, selecting "Shell" from the EFI boot menu should
go to the shell.

(From OE-Core rev: d031cdbf40231b8c103d78c69252bf9d584d0605)

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

mkefidisk: Always use rootwait and document kernel parameters used

Without a reliable way of knowing if the target device with be an
asyncronous block device on the target (MMC or USB), err on the side of
caution of always specifcy "rootwait", ensuring the kernel will wait for
the device to appear and not abort if it hasn't appeared in time for
mount.

Document the remaining kernel parameters added by this script on the
same line as rootwait.

(From OE-Core rev: 5b6a6a3872fd341cf978be40c69707223e3c29df)

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

mkefidisk: Boot with ro kernel parameter instead of rw

There is no need to boot with "rw". Booting with "ro" will allow for
fsck to be run during boot, and a proper /etc/fstab will still ensure
the rootfs is "rw" by the time the user can interact with the system.

Change the "rw" to "ro" in the kernel parameters specified in the
generated grub.cfg file.

Fixes [YOCTO 4036] mkefidisk.sh hardcodes 'rw' as root mount option

(From OE-Core rev: 960f0cbf85a4124adbc74d8b2ceb09a7d39ecc04)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Cc: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mkefidisk: Cleanup comment length

Keep comments under 80 characters in length.

(From OE-Core rev: 1fbab279edd0057193900646cf9f66323698e774)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Cc: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mkefidisk: Always specify a root= kernel parameter

The current script only replaces an existing root= kernel parameter
which can result images created without a root= paremeter, even though
the script expects a target rootfs parameter.

Rather than replacing the root= parameter, delete the root= parameter if
it exists, then append an appropriate root= parameter.

Fixes [YOCTO 4035] mkefidisk.sh forgets to add root= parameter

(From OE-Core rev: e5dbec7e7d3bb29676280823b0337ad429c75120)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Cc: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

consolekit: don't inherit gnome

By inheriting gnome this package was build-depending on
gtk-update-icon-cache-native, gconf, gconf-native, shared-mime-info and
hicolor-icon-theme for no reason.  Change this to inherit autotools pkgconfig to
reduce the build-dependencies.

Also, add a missing dbus-glib build-dependency that was previously pulled in
through something in the gnome class.

(From OE-Core rev: bde2c0ab1500976e2cb50f1526df04fb8b2cd408)

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

package_{ipk, deb, rpm}: drop the TARGET_OS conditional

The tclibc file for uclibc already empties IMAGE_LINGUAS, so there's no point
to this conditional as far as I can tell, and it can cause issues for certain
values of TARGET_OS.

(From OE-Core rev: 0fdda5840452c7715ed775d5d18d9f850396f483)

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

package_deb: don't install glibc-localedata-i18n separately

This was removed from package_rpm and package_ipk years ago, and shouldn't be
needed, installing the packages from package_linguas should be sufficient.

(From OE-Core rev: e9485c7077e5a9508371bbbbebdaaf1e9685da61)

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

coreutils: disable ACL as we do not depend on it

Coreutils does not depend on ACL but also does not disable it. Effect is
that from time to time I have a copy in sstate-cache which got built
with ACL and then I have problem while building images or sdk:

| Collected errors:
|  * satisfy_dependencies_for: Cannot satisfy the following dependencies for packagegroup-core-standalone-hhvm-sdk-target:
|  *    libacl1 (>= 2.2.51) *
|  * opkg_install_cmd: Cannot install package packagegroup-core-standalone-hhvm-sdk-target.

(From OE-Core rev: d54141130475b091358807dce17ba8d244d33ea1)

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

perl: avoid splitting out .debug directories as packages

The recursive parameter is set to True, the regex for the first call to
do_split_packages matches any path under ${libdir}/perl/${PV}/auto/, and
the .debug directories contain .so files, so each one was getting picked
up as a package. Change the regex to disallow dots in the path beneath
auto/ and thus avoid the .debug directories.

Fixes [YOCTO #4048].

(From OE-Core rev: f8f6992fe0f29db1cc4df15b7449e06188052041)

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

connman-conf: Avoid appending IPv4 address

If run more than once, the IP addresses would be appended
to IPv4 variable. Avoid that by rewriting the IPv4 always.

(From OE-Core rev: da682e752e351e5bb7636ee187db3eabc208c177)

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

connman-conf: Move the setup script into /usr/lib/connman

The /etc/connman is not suitable for the setup script.
There are other connman related scripts in /usr/lib/connman
so moving the wired setup script there.

(From OE-Core rev: 858ad53aee9ef5848c09d76c0e6bf464bc31d30e)

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

distrodata.bbclass: added some more git processing

* Truncated all git MD5 sums to 7 digits
* Added regex checking for git packages as well

(From OE-Core rev: bc830ab3f6e9704c830e934c6f39c85ef11f867d)

Signed-off-by: Emilia Ciobanu <emilia.maria.silvia.ciobanu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

distro: Add separatebuilddir.inc

Going forward its going to be useful to separate build data from source data
in those autotooled projects which support it. Unfortunately there is a lot
of breakage so for now, this starts the creation of an opt in list which
we can iterate over enable more recipes over time.

(From OE-Core rev: 9e64079063fc4748b48eee0e2592caf8ba9de10e)

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

apt: Fix case where ${B} != ${S}

Need to reference objects from ${B}.

(From OE-Core rev: 0610cf3ef864acc9ca98498b9810bcce0fbb935a)

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

base-passwd: Fix case where ${B} != ${S}

Need to reference build objects from ${B}.

(From OE-Core rev: 2b58618406195aa53df2335c354d7e45ad08f4d5)

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