grub: add xz RDEPENDS

grub_2.0.0 requires xz to run or an error may occur.

(From OE-Core rev: ffa2877c06c587d4ea56c55bfd0f67a88e42a772)

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

lib32-qt4-examples: fix QA warnings and INSANE_SKIP not work

Fix the following warnings:
...
lib32-qt4-examples: found library in wrong location: /usr/bin/qt4/examples/tools/echoplugin/plugin/libechoplugin.so
lib32-qt4-examples: found library in wrong location: /usr/bin/qt4/examples/tools/plugandpaint/plugins/libpnp_extrafilters.so
...

In oe-core f119566477243ce43b727492dc78b9cb3dd76de4, it added INSANE_SKIP
libdir for examples packages as it includes plugin shared libraries outside
of libdir.

But it hardcoded ${QT_BASE_NAME} as INSANE_SKIP's package name
and it didn't work while multilib enabled.

Add ${MLPREFIX} as package name's prefix in INSANE_SKIP

[YOCTO #5516]

(From OE-Core rev: d63588ddd5a1f08fff90f5bb3f7278490e359720)

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

flex: fix m4 issue on target

Flex needs m4 to run (see below) and, since the create_wrapper
introduces a bash dependency on target, give the path to m4 binary in
the configure command line.

Snippet from the flex documentation:
"The macro processor m4 must be installed wherever flex is installed.
<...>
m4 is only required at the time you run flex."

[YOCTO #5329]

(From OE-Core rev: 64030f37b34f75144f53eef42d5822ede79e08bd)

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

expect: Add recipe

Nedeed for gcc-runtime tests.

Fixed build on multilib and add patch to remove
!/depot/path/expect -f
which caused rpm to puke on rfs generation

(From OE-Core rev: 8f06d2975ed7d5db3828dc116c0df72d1161e8e5)

Signed-off-by: Mihaela Sendrea <mihaela.sendrea@enea.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libnl: Fix random segfaults due to memory corruption

This is a backport from upstream fixes a severe problem
w.r.t memory management, where it would result in random
segfaults in applications depending on libnl

(From OE-Core rev: 3c58ea10f90c657d34133d7244a550456bc93cf9)

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

tcl: Install header into 8.6 instead of PN-PV in user/include

This helps in compiling other programs like expect which
depend on private headers but 8.5, 8.6 and so on is enough
granularity and currently we had 8.6.x and so on which
means that expect recipe will need to be touched whenever there
is minor update of tcl.

Additionally the encode creating symlink to shared object in
patch and remove it from recipe

Refresh patches after making changes to Configure.in we
propertly generate configure and not patch is directly as
was the case.

(From OE-Core rev: 67f44193135c789e478410347ff58ed110ed9484)

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

mdadm: flag __SANE_USERSPACE_TYPES__ to include int-ll64.h for powerpc64

      *PPC64 uses long long for u64 in the kernel, but powerpc's asm/types.h
       prevents 64-bit userland from seeing this definition, instead defaulting
       to u64 == long in userspace.

      *fix the below error
       |super-ddf.c:4542:5: error: format '%llu' expects argument of type 'long long unsigned int',
       |but argument 5 has type '__u64' [-Werror=format=]
       |dprintf("BVD %u has %08x at %llu\n", 0,

(From OE-Core rev: d3caab6eb03264b4f4d744f914598022299011ba)

Signed-off-by: Chunrong Guo <B40290@freescale.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kernel-grub.bbclass: support /boot area within root partition

Previously, it supported the situation that /boot area with separate boot
partition:
...
menuentry "Update bzImage-3.10.10-WR6.0.0.0_standard-3.10" {
    set root=(hd0,1)
    linux /bzImage-3.10.10-WR6.0.0.0_standard root=/dev/sdb1 rw ip=dhcp
}
...

But didn't consider the situation that /boot within root partition:
...
menuentry "Update bzImage-3.10.10-WR6.0.0.0_standard-3.10" {
    set root=(hd0,1)
    linux /boot/bzImage-3.10.10-WR6.0.0.0_standard root=/dev/sdb1 rw ip=dhcp
}
...

This fix supported them both.
[YOCTO #5514]

(From OE-Core rev: 53d342db9f9995564573715f28c3e4c9c8c68bf9)

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

sysvinit: fix problem in switching runlevels

Previously, if we switch to runlevel 1 and then switch back to runlevel
5, the network interface will be brought down and the NFS service will
not be restarted correctly.

The problem is that the networking and rpcbind services are brought down
in runlevel 1 but not brought up in runlevel 5.

This patch fixes the above problem. It's based on the assumption that
in sysvinit-based system, runlevel 1 does not have networking support.

This patch adjusts some init script parameters used by update-rc.d. It
makes sure that networking starts before rpcbind which in turn starts
before mountnfs.sh. When switching to runlevel 0, 1 and 6, the umountnfs.sh
is run first before stopping rpcbind service, and the network is brought
down afterwards.

[YOCTO #5513]

(From OE-Core rev: f12e1291b51ba6692b6809570d6a9b73c70a7fe9)

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

libcheck: upgrade to 0.9.11

Both patches have been merged upstream.

(From OE-Core rev: 9982f9836f06b1a9282d657ee249eb08261518cb)

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

sysstat: upgrade to 10.2.0

Cleanup by moving files to $PN

(From OE-Core rev: b9d63e318c2d8edb9b360544cca4c76be2904552)

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

bitbake: runqueue: Fix hole in setsceneverify skipped task logic

We have do_bundle_initramfs which is a task inserted after compile and
before build. It is not covered by sstate.

If we run a build with a valid sstate cache present, the setsceneverify
function realises it will rerun the do_compile step (due to the
bundle_initramfs task) and hence marks do_populate_sysroot to rerun.
do_install, a dependency of do_populate_sysroot is left as marked as
covered by sstate.

What we need to do is traverse the dependency tree for any setsceneverify
invalided task and ensure any dependencies are also invalidated. We can
stop at any point we reach another setscene task though.

This means the do_populate_sysroot task has the data from do_install
available and doesn't crash.

(Bitbake rev: f21910157d873c030b149c4cdc5b57c5062ab5a6)

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

bitbake: fetch2: Fix handling of SCM mirrors in MIRRORS

If an SCM mirror is in PREMIRRORS, the tarball is downloaded and then found
by the "upstream" check and handled correctly.

If an SCM mirror is in MIRRORS, the tarball is downloaded but not used
since there is no "upstream" run after MIRRORS completes. It therefore
sits there useless and unused. This code change forces the upstream to
run after a mirror tarball is found and fixes the usage of SCM mirrors
in MIRRORS.

(Bitbake rev: a66ee0994645aa5658b2f5ea134ed17d89f8751a)

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

bitbake: git: Use git branch for ancestor checks

Using git merge-base for checking for ancestors is nice but required git 1.8.0
which is not in many distrbutions yet. We therefore revert to a more ugly
check using git branch --contains until such times as we can upgrade.

(Bitbake rev: 31467c0afe0346502fcd18bd376f23ea76a27d61)

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

bitbake: serv.py: Give pr-server up to 5 seconds to commit data

The default value of 0.5 seconds before sending the pr-server a
SIGTERM is not enough to guarantee that sqlite has committed all
the pr data to the database. By polling the pid to see if it is
still running, this allows the pr-server process to shutdown
cleanly and finish the final pr data commit.

(Bitbake rev: 22eec978e70794923c85689928c6be0cfe71cdcd)

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

bitbake: git: Use merge-base instead of log for testing if a commit is present

The current use of git log to check if a given revision is present can be
a little fragile.

For example if revision X was on branch A, and then later added to branch
B, the update checks would not notice this since they just check for X
being in the repository.

We also had some autobuilder corruption where an older packed-refs file
was copied over a new repository containing newer pack files. There
was no update to the refs file since the revision was present but
not accessible in any branch.

The correct fix is to check that the required revisions are present
on the specific branches. This patch does this using merge-base.

(Bitbake rev: 89abfbc1953e3711d6c90aff793ee622c22609b1)

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

python: Fix ptest not working problems

The run-ptest contained in python uses a Makefile. The path of the
Makefile is changed to ${libdir}/python/ptest in do_install_ptest.
However, the directory is wrong when the project is configured with
"--enable-multilib=lib32"

In addition, do_install_ptest is defined before "inherit ptest", so
it is overriden by the one in ptest.bbclass. do_install_ptest is
moved down.

(From OE-Core rev: 0eb947454e1c92467283e6f1adeca67c7c57698b)

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

sysvinit: unmount the psplash lazyily

There is an race condition where psplash is not quite exited before the unmount occurs
causing a umount: /mnt/.psplash: target is busy message to appear, it's ok to lazyily
unmount and not get this message

[YOCTO #5244]

(From OE-Core rev: 9ded366084f22f48ef72aa22acf6a38982d16d97)

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

pybootchartgui: Add option --minutes to show time in minutes

(From OE-Core rev: b93eefd039a956b7d1d184592dd4342eb43f9341)

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

pybootchartgui: Add a color for the package_write_* tasks

(From OE-Core rev: 2cec29b47a1de5da712cf4e2c6e25daf45d9f265)

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

pybootchartgui: Reorder the legend to match the task execution order

(From OE-Core rev: a12cf87471de40ce432d0176eabfb111de5310c6)

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

pybootchartgui: Make bars without a specified color white

Previously they were transparent.

(From OE-Core rev: 4ebdd8672cc5589a3e2f8d1b75cde7fae9fd6c99)

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

pybootchartgui: Make "Show more" show all processes

While "Show more" is enabled, all processes are shown, regardless of
--mintime.

This also has the added benefit of making the first shown bar start at
its correct offset from the start time, rather than always starting at
0.

(From OE-Core rev: 5f1b8730f90099c0f73a6b08599990ee71e831b5)

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

pybootchartgui: Draw a lot less ticks

With this, one second ticks are only enabled if the width of a second is
five pixels or more. It is also possible to distinguish 1, 5 and 30
second ticks.

(From OE-Core rev: bd0bde6d04fd6cd9f8e7773d68da127144afa7de)

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

pybootchartgui: Disable options that do not make sense

(From OE-Core rev: e1cde0c8b65a56657a5a5669890dad442223fef4)

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

pybootchartgui: No need to do a double list comprehension over files

(From OE-Core rev: bd00a768d88c22eabee90407684f1fb84139acfb)

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

pybootchartgui: Avoid having overlapping process bars

(From OE-Core rev: 901afb35da814661ed20b2895f1d4055bf73fae2)

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

pybootchartgui: Use correct header height

(From OE-Core rev: f7d1f49aa46a46ecb881386f0399aa7da288d805)

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

pybootchartgui: Correct the X offset for the chart

This will make the first bar actually start within the graph. It will
also move the graph to the right so the names of the first tasks are
more likely to be visible.

(From OE-Core rev: 388daa9a8ce7f2216fb55ce65cab1d4060f6c41d)

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

pybootchartgui: Set the initial state of "Show more" correctly

(From OE-Core rev: 55fa7f768bb7618f2daaf43f147609c76e077b8b)

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

pybootchartgui: Make the horizontal scaling stay within bounds

(From OE-Core rev: 50c2c3435915ef1ecbde395c71c5c9581c83fb2e)

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

pybootchartgui: Import pybootchartgui 0.14.5

This update the pybootchartgui code to the latest release from its new
location at "https://github.com/mmeeks/bootchart". This only imports
the relevant parts, and not all of bootchart2.

(From OE-Core rev: 6f1568e54a7808b2ab568618fc5bb244249579f1)

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

mtdev: upgrade to 1.1.4

Refresh fixsepbuild.patch to apply cleanly, and clean up the description.

(From OE-Core rev: 3cfce710faf82024c5d8973fe83a125d0375309b)

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

libxrandr: upgrade to 1.4.1

License checksum changes as COPYING was updated upstream with additional
copyright notices.

(From OE-Core rev: 1261cf6fdc14529d10d61e2f21d675555ed6bec6)

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

gdk-pixbuf: upgrade to 2.30.1

(From OE-Core rev: e5b861c5ff2e9e0e20ffcc12acb193db677227d7)

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

nfs-utils: Add new option to disable nfsidmap and PACKAGECONFIG for that

* it's autodetecting keyutils from meta-ivi

(From OE-Core rev: 64d04c7449f47cc49468b54cd6529700e0c7a608)

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

lttng-tools: Fixes incorrect path of ptest cases

In ptest cases, some paths are still the source code path.
Continue to fix the paths of executable files in ptest cases
to make them work on target.

(From OE-Core rev: 95a0eb99b9fd88288b6f03c7d0173c392d25de28)

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

packagegroup-core-boot: Drop build-time dependency on virtual/kernel

* because kernel isn't needed to build packagegroup and building
  it doesn't influence if it's included in image or not.

(From OE-Core rev: 2935e7b2fc3c1a35e810eb4043638ffcd6682d0c)

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

ofono: upgrade to 1.13

obsolete_automake_macros.patch and
missing-ssize_t.patch no longer needed,
included in upstream.
Better support for BlueZ5.

(From OE-Core rev: 50ea6d79fbbece279cda908ea768673de02a3d82)

Signed-off-by: Cristian Iorga <cristian.iorga@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

buildhistory_analysis: fix error when comparing image contents

OE-Core commit b7de1eaac9eed559b2d68058f5de67de74a6cb58 added an extra
argument to the compare_dict_blobs() function but missed adding the
argument to one call to compare two versions of the image-info.txt file.

(From OE-Core rev: 24a45d752c3e3d0d8b59c040355e4fe7de22b041)

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

kernel.bbclass: fix hardlink test

The commit [kernel.bbclass: test for hardlinks before installing] doesn't
work on all build machines without a space between the ! and the test for
hardlink equivalance. The build continues, while the test fails and no
copy is made into the sysroot. Eventually tasks like build_hdimg will fail
with the missing kernel with the root cause being less than obvious.

(From OE-Core rev: 0704648e9dfd2897f6f54773a0e7d2762117970c)

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

kernel.bbclass: test for hardlinks before installing

commit df564c4 [kernel: Use hardlinks for do_populate_sysroot for speed]
does indeed speed up the build, but it also means that the KERNEL_OUPUT
and KERNEL_IMAGETYPE may be hardlinks.

If they are hardlinks, install complains that they are actually the same
file and stops the build. We can easily test and avoid the copy if the
file is already in place.

[YOCTO #5527]

(From OE-Core rev: 6a3b3e501c66733216ac43d3f020cfdb512dd640)

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

bitbake: cooker, toaster: variable definition tracking

In order to track the file where a configuration
variable was defined, this patch bring these changes:

* a new feature is defined in CookerFeatures, named
BASEDATASTORE_TRACKING. When a UI requests BASEDATASTORE_TRACKING,
the base variable definition are tracked when configuration
is parsed.

* getAllKeysWithFlags now includes variable history in the
data dump

* toaster_ui.py will record the operation, file path
and line number where the variable was changes

* toaster Simple UI will display the file path
and line number for Configuration page

There is a change in the models to accomodate the recording
of variable change history.

[YOCTO #5227]

(Bitbake rev: 78e58fed82f2a71f052485de0052d7b9cca53ffd)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: task with outcome 2 (sstate), should have sstate_result!=0

0 (not applicable) is not a valid sstate_result for tasks with
outcome 2 (sstate), which should return 3 (restored), 2
(failed) or 1 (missed).
Sstate_result for tasks with outcome 2 is equal to the outcome
of _setscene corespondent task.

[YOCTO #5220]
(Bitbake rev: 8ff8d75318ea88ba80c744b471e486901ef6749a)

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

bitbake: hob/hoblistmodel: check if vals of packages/recipes names are not None

[YOCTO #5053]
(Bitbake rev: ba9fe77e37be31e8246431578902e871dd94515e)

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

bitbake: cooker: warn if user specifies a target listed in ASSUME_PROVIDED

If the user explicitly asks to build a target that is listed in the
value of ASSUME_PROVIDED, show a warning mentioning that it will be
ignored.

(Bitbake rev: 65be09f5e2a4a41e65c9232e208d8154b822fc4e)

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

bitbake: hob: refresh tables when the user is redirected to Edit packages

When the build fails and I click "Edit packages", we should be returned
to the tab we launched the build from. Hob should remember the tab,
not the search state.

[YOCTO #5257]
(Bitbake rev: afdea16516084ef6c046d80e12bf244f25a9da50)

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

bitbake: toaster: fix path to buildstats file

The buildstats file path changes based on the
optional PE variable that may be defined for a
recipe.

The toasterui simply ignored the PE value, and
as such it didn't correctly reach buildstats files
for some of the tasks.

This patch fixes the issue.

    [YOCTO #5073]

(Bitbake rev: 97b8ab88edc7c8dfb26b4cf305701ec96e52cc4f)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toasterui: mark failed sceneQueue tasks as failed

This patch addresses an issue where a failed sceneQueue task
entry was not updated on the Fail event. As a result, it
always showed the task as not-available.

    [YOCTO #5216]

(Bitbake rev: 9b99a417f58381bac4bda412bcfd11de50403318)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: remove author field

The AUTHOR field in most recipes is not defined,
or it's not really consistently set in the metadata,
Also does it seem particularly useful.

This patch removes the AUTHOR variable from the
toaster system

    [YOCTO #5449]

(Bitbake rev: da3ac049300be84defab7b32b0b99ab07c7d0a27)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: toaster: fix tasks showing as NoExec

Tasks without script type information showed by default
as NoExec; this happens for all Prebuild or Covered
tasks, as script type information comes only on TaskStarted
event. Such a default value may drive confusion, as NoExec value
should be reserved for the NoExec-flagged tasks.

This patch adds a new default value named Unknown that will be
used for all tasks that don't have script type information
available.

    [YOCTO #5327]

(Bitbake rev: ec6cac74290f0d4f5b60222019c23416b4b8e1ef)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.10: meta: ARM: OMAP3: Add USB PHY driver for Beagleboard

Updating the meta branch SRCREV to update the USB configuration:

    The Beagleboard needs the USB PHY drivers in the kernel in order to enable
    USB and Ethernet functionality. This fix ensures that they are built in
    by tweaking the kernel config.

    Tested on Beagleboard xM Rev. C2.

(From OE-Core rev: 89a372840a957e540bed954e629aa68335b3dfe0)

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

linux-yocto-rt/3.10: fix ntp merge issue

Updating the -rt SRCREVs to pick up the following fix:

    ntp: fix ntp_notify_cmos_timer merge issue

    PREEMPT_RT_FULL has a stubbed ntp_notify_cmos_timer due to a bad merge.
    Renaming and restoring the full -rt functionality to this routine.

(From OE-Core rev: 41d4f0feca69bf1b41f16f5f7d21bf7540e6c47a)

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

linux-yocto/3.10: meta: ARM: OMAP3: Add USB PHY driver for Beagleboard

Updating the meta branch SRCREV to update the USB configuration:

    The Beagleboard needs the USB PHY drivers in the kernel in order to enable
    USB and Ethernet functionality. This fix ensures that they are built in
    by tweaking the kernel config.

    Tested on Beagleboard xM Rev. C2.

(From OE-Core rev: 2a9944514362445ee891f6e77c4ae62950e247b3)

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

classes/license: use "1" to set boolean variables

The expected usage of COPY_LIC_MANIFEST and COPY_LIC_DIRS appears to be
to set them to "1" to enable; however the test here is just testing
whether they have a value at all, so setting them to "0" would also
enable them which is somewhat disingenuous. Actually check if they are
set to "1" instead in order to fix this.

(From OE-Core rev: 2d8e124adcf27af524eeeae61daf1b21a1c2f27c)

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

classes/pkg_metainfo: remove

This is very old, and we now have pkgdata which is much more complete.
Nobody appears to be using this class, and even OE-Classic had no
current references to it.

(From OE-Core rev: e67dbb638044b804738bdd589d64d45963a3297a)

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

classes/pkg_distribute: remove

This appears to be a very old way of archiving recipes and associated
files. We have better ways of doing this now, and nobody appears to be
using this class - even OE-Classic had no current references to it.

(From OE-Core rev: e3141844ef84d8a48efa81e0c9c85821ced16f7c)

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

systemd: fix comments

It's DISTRO_FEATURES; DISTRO_FEATURE is invalid.

(From OE-Core rev: d2380c683d5411e7a85c3ce80115aea329187534)

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

classes/license: fix comments

It's LICENSE_FLAGS, LICENSE_FLAG is invalid.

(From OE-Core rev: 70bd620b58ddd90477d9f438922ee3bdc7e8c5a5)

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

gtk-immodules-cache: fix error message to use correct names

It's gtk-immodules-cache and GTKIMMODULES_PACKAGES.

(From OE-Core rev: 10cc4ceb813128676e78d04068ad089f6b4a697b)

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

bitbake.conf: remove CPU_FEATURES defaults

This variable has been unused since the tune file overhaul two years
ago.

(From OE-Core rev: a1d9f2374ede768057fd364da6c0e1eeeb10499f)

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

classes/sanity: validate SDKMACHINE value

If SDKMACHINE is set then check that a configuration file matching it
actually exists, otherwise the user won't know that they've set it
incorrectly.

(From OE-Core rev: 8c984f92af821a4048c93f8e308c5f4a3fa39ca4)

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

bitbake.conf: remove BOOTSTRAP_EXTRA_* variable defaults

These were for task-bootstrap in OE-Classic and have never been used in
OE-Core.

(From OE-Core rev: f4692afb518f07e17fbd35a2023877b7041abef9)

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

classes/insane: improve LIC_FILES_CHKSUM messages

* If the md5 parameter is specified with no value, report that the
  checksum is not specified instead of reporting that it has changed
* If the md5 checksum has changed, point directly to the license file in
  a way that is easy to copy and paste and give the line numbers in an
  easy to read form, as well as asking the user to verify that the new
  contents matches the current LICENSE value.

(From OE-Core rev: ec8590aa81e201e28e500935d31cd7266114471f)

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

scripts/create-recipe: fix handling of --help

If --help is specified as the first argument, show the standard help
text instead of trying to process it as a URL.

(From OE-Core rev: abb139b10c3f431bcebb1847621f97d7ec6249ce)

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

perl: perl-ptest.inc polutes package dependencies when ptest not enabled

When ptest is not enabled, the populate_packages_prepend function runs
wheter ptest is enabled or not.  This causes ptest packages to get in the
dependencies list when ptest is not enabled.

(From OE-Core rev: 826f4e4057a221127ac4c1d0658d975032fc7d90)

Signed-off-by: Jeffrey C Honig <jeffrey.honig@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/image-empty: remove

We don't actually need this dummy class; "inherit" can be used with an
expression that evaluates to nothing with current BitBake.

(From OE-Core rev: f312eb2d2a2715e772ed9e8afc7aea326d8079ec)

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

staging: Use hardlinking for sysroot_stage_dir

This saves about 2GB on a core-image-sato build so is worth doing and is
consistent with our efforts to try and decrease our build footprint.

Build time in my local test seemed unaffected but on more IO bound
machines it should help.

(From OE-Core rev: 7136cbc64e5efb09f3fae3e2e35a181ca3d66dd4)

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

bitbake: cooker: replace "w" file opening mode with "a" mode

Also removed some redundant file manipulation code
Based on patch sent by Stefan Stanacar <stefanx.stanacar@intel.com>

(Bitbake rev: e054c1e7c8581f66082fcdfb89769401ca6e78a3)

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

bitbake: toaster: convert build_package size to bytes to keep consistence

[YOCTO #5503]
(Bitbake rev: 19eb6e01b675c439ff0a817be6fa5e34ad42ba37)

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

bitbake: newbb.vim: fix spelling

(Bitbake rev: 0bd1a3026a26c2c01abd31e265668541f6201bae)

Signed-off-by: Trevor Woerner <trevor.woerner@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

binutils: Add gnu-config-native to DEPENDS

do_configure() in binutils.inc includes an explicit call to
gnu-configize so we need to make sure that gnu-config-native is
present.  Previously this was being dragged in with the rest of the
autotools stuff, but commit 54a3e2ee37003fc56af0339f857b0b6442790c26
disabled that for binutils-cross on the grounds that "we don't
autoreconf" the toolchain components.  Fix this by adding
gnu-config-native itself explicitly to DEPENDS.

(From OE-Core rev: 616354f13732d13c17434d5b60b166f691c25761)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

insane, license: Trap MalformedUrl exceptions when parsing LIC_FILES_CHKSUM

bb.fetch.decodeurl() will throw if it doesn't like the look of the URL that
it's given.  (Bitbake's idea of what constitutes a valid URL is somewhat
idiosyncratic so it is fairly easy to trip over this by mistake when writing
a recipe.)

If these exceptions are allowed to propagate all the way up to better_exec()
then we will get a large amount of python stack trace spew when they are
finally caught.  Avoid that by catching them locally and throwing
bb.build.FuncFailed() with a suitable explanation instead.

(From OE-Core rev: ef35e164c62d89806367b822e3baeff482ec237f)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

autogen-native: remove the unnecessary RDEPENDS

The recipe only provides native packge, the RDEPENDS is not necessary.
Remove it and also two unnecessary comment lines.

(From OE-Core rev: 915c1a35ed3ac7eb650d68c079d6b7cda781b106)

Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Revert "glib-2.0: Specify libiconv when building on uclibc"

This reverts commit 38b6c4df7c215ed7fd6be107fbc2527e66791e2e.

Its not needed anymore after upgrade to 2.38

(From OE-Core rev: 3efd8530053cf54e3f0b0fc6a96272fdb2ee27ba)

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

bitbake.conf: Default DISTRO to nodistro

An empty distro value leads to OVERRIDES and FILESOVERRIDES containing
"::" entries which causes odd issues such as files being included when
they shouldn't be. We could put in anonymous python to guard against
empty entries but its messy and setting a default value for DISTRO to
something harmless is much easier.

This patch adds a weak default and ensures the sanity test doesn't
complain about it.

DISTRO_VERSION and SDK_VERSION are also updated to match.

(From OE-Core rev: b7279f99639774674da806d37d252f388f33055f)

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

e2fsprogs: fix tests/f_extent_oobounds

tests/f_extent_oobounds runs debugfs from the system, not from the
source tree, and if the system's debugfs doesn't have the extent_open
command it fails silently.

Use $DEBUGFS and $MKE2FS to get the in-tree executables for this test,
just like other test scripts do.

(Build machines which run make check shouldn't need to have e2fsprogs
installed, and we should be testing just-built versions of the tools
anyway)

This patch is from:
http://www.spinics.net/lists/linux-ext4/msg38880.html

Eric Sandeen had sent it to the linux-ext4 mailing list, but haven't
been merge by now.

[YOCTO #5511]

(From OE-Core rev: 22465cef87b4b1685b3b131751eaf528503b264c)

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

core-image-weston: add SSH and hwcodecs to the image

core-image-sato has these by default so add them here too.

(From OE-Core rev: 1ddcdc9cbc94188a7ca564ca4cb783a356fe7ab6)

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

icecc: Fix race condition when packaging toolchain

In the current implementation there can be a race condition while
creating the toolchain archive causing the build to break.

This is fixed by locking the toolchain archiving step using flock.

(From OE-Core rev: 8a30be803e91e66688cfc27ca4c21f26fb22eed8)

Signed-off-by: Tobias Henkel <tobias.henkel@bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

icecc: Add package whitelist

There are some recipes which parse the PARALLEL_MAKE variable by their
own and set them to an empty string afterwards. This disables icecc
for this recipe.

Adding a whitelist for forcing icecc makes it possible to use icecc
also with these recipes.

(From OE-Core rev: d2735ac44887c7e01134d6870a4875a786501eba)

Signed-off-by: Tobias Henkel <tobias.henkel@bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

icecc: Support shell evaluation of KERNEL_CC

In the current implementation a KERNEL_CC variable containing shell
evaluation breaks the build process. Shell expansion is not happening
before general expansion in get_cross_kernel_cc which results in a
syntax error and an aborted parse process.

Before expanding the KERNEL_CC variable get_cross_kernel_cc now checks
for backticks or '$(' in the KERNEL_CC variable and performs a shell
evaluation using a call to echo if it finds one.

(From OE-Core rev: b28bae30fc5d8d1d7cc675ddb4159c39fb9bc3fd)

Signed-off-by: Tobias Henkel <tobias.henkel@bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

icecc: Enable icecc also for install step

Some packages are calling the compiler in the install step. In this
case either the build breaks or icecc is not used for building. The
proper environment has to be set to enable icecc based building.

(From OE-Core rev: bb1366cebb60593fc21fd7a9a678a159da8ec81c)

Signed-off-by: Tobias Henkel <tobias.henkel@bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

icecc: Add dummy python version of set_icecc_env

Bitbakes prepend mechanism for the tasks disregards the type of the
function. Thus bitbaking recipes using python functions for configure,
compile or install steps fail due to the missing python version of
set_icecc_env.

Assuming that icecc doesn't need to be used in such situations adding
a dummy python version of set_icecc_env fixes this.

(From OE-Core rev: 365dec500166b8eb9d64c573dd7139d3a26ae445)

Signed-off-by: Tobias Henkel <tobias.henkel@bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

icecc: Add blacklist vars to BB_HASHBASE_WHITELIST

Changing any of the ICECC blacklist variables should not change the
sstate checksum as this doesn't influence the build result.

(From OE-Core rev: 0f7dc969e64bc7563e208a4fb7d3ebe7c4e69e79)

Signed-off-by: Tobias Henkel <tobias.henkel@bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

icecc: Reduce verbosity with empty PARALLEL_MAKE

Currently the icecc class prints a note for every package which
disables parallel make at parse time. This is unneccessary as many
packages don't support parallel building. Changing the log level from
info to debug hides these messages in normal builds without removing
the information when needed.

(From OE-Core rev: 6d2735629c20560a9406964195726b1a6e2d7d99)

Signed-off-by: Tobias Henkel <tobias.henkel@bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

icecc: Log reason in error cases

The current implementation doesn't give a hint about the cause in case
something went wrong in set_icecc_env. This makes it harder to find
out why a package is not being built using icecc. Therefore warnings
are inserted in the various error cases.

(From OE-Core rev: 0adea170c7019304471d442784fa0eb9bfcdf94b)

Signed-off-by: Tobias Henkel <tobias.henkel@bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemu: use PACKAGECONFIG to address xfsprogs dependency

To avoid a implicit build result.

(From OE-Core rev: 3e302e94ba5bcbba2736f37c0f67cfaf7fa45c0c)

Signed-off-by: Ming Liu <ming.liu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

vala: cleanup DEPENDS

Patch cleans up and fixes dependencies of 'vala' in the following way:

* remove deps on 'gtk+' + 'dbus'; these packages are not required and
  'gkt+' is high barrier preventing using 'vala' on headless systems

* 'bison-native' is required both on native and target builds; 'DEPENDS'
  and 'DEPENDS_virtclass-native' and 'DEPENDS' can be merged hence

* 'flex-native' is required

* although 'xsltproc' (from libxslt-native) is optional, a dependency
  on it was added to make builds deterministic.  An alternative method
  might be 'EXTRA_OECONF += "ac_cv_path_XSLTPROC=:"' but this is error
  prone as it depends on internals of the build system.

(From OE-Core rev: 8834e6b2aa70a2b3d43550d3912e16b8f5c12d7e)

Signed-off-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wpa-supplicant: enable CONFIG_CTRL_IFACE_DBUS_NEW

Without this option wifi support in connman will fail:

src/technology.c:technology_get() No matching drivers found for wifi

(From OE-Core rev: 403e365e433c54633bcc843b32487a766282226e)

Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

packagegroup-core-basic: define VIRTUAL-RUNTIME_syslog

We have rsyslog and syslog-ng in our layer, and oe-core
has sysklogd, define VIRTUAL-RUNTIME_syslog so that we
can switch them easily. Set sysklogd as default here,
we will have setting in distro confs to override it.

(From OE-Core rev: 532e3efe76b62a4454193af9da59ced24e05bd33)

Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc-common: Enable building multiarch mips gcc

MIPS gcc is not configured with multiarch enabled.  This causes
compiler generates local label with $ prefix, which is specified
in default o32 abi.  It is not recognized as local symbol by n64
assembler, so we get a lot of unexpected external symbols.  We
should configure MIPS gcc with --enable-targets=all, as for other
archs.

(From OE-Core rev: c210393b3e0c54606493e311d7c2040efebc4dcd)

Signed-off-by: Lei Liu <lei.liu2@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd: Upgrade 206 -> 208

Remove patches accepted upstream
Forward-port uclibc patches
Create additional packages to bundle zsh support files

(From OE-Core rev: 9131cd014443243dc41c828efdcc8d87b179940d)

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

elfutils: remove eu-objdump if it does not work

[YOCTO #5283]

eu-objdump does not work on mips, arm and ppc, so remove it from these
unsupported architectures

(From OE-Core rev: abf01e2f36943b5a5c5d2d827186054a51902130)

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

mklibs: add dependency on dpkg-native

mklibs requires the "dpkg-architecture" utility to work.
Add dependency on dpkg-native.

(From OE-Core rev: 9811641e95dd7e1514eb41900e033a0548bd13d8)

Signed-off-by: Lei Liu <lei.liu2@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image-mklibs: Fix grep pattern when mklibs collects executables in rootfs

File command in some version could print extra space between
"LSB" and "executable" - it causes mklibs can't find any executables
using grep "LSB executable".  Fix the grep pattern to catch
multiple spaces.

(From OE-Core rev: a52ef8c5dcd71f39bb48c71fb868cc0db662560e)

Signed-off-by: Lei Liu <lei.liu2@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

weston-init: use /run instead of /var/run

/var/run is just a symlink to /run now, so use /run directly.

(From OE-Core rev: b6201d2a27639f9d384ba1137096a05f6795797a)

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

openssh: upgrade to 6.4p1

(From OE-Core rev: 19a483c157e783ded40e8dacac338a4d770a2d1a)

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

ethtool: upgrade to 3.12.1

(From OE-Core rev: 3241f767e08df95f3d839fd50f53e6b88e39e094)

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

nfs-utils: upgrade to 1.2.9

(From OE-Core rev: 6ded53ea2bfa80bf0a7b4edd460e739123b96c86)

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

cmake: upgrade to 2.8.12.1

(From OE-Core rev: c38fe8e31d5c84e56f18a7c1b4709cc239e579d7)

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