init-install-efi.sh: Fix root= specification

Fixes [YOCTO #5237]

The current grub.cfg manipulation depends on an existing root=
parameter. If this doesn't exist, the correct root= parameter will not
be added.

Instead, remove any existing root= parameters and add the correct one
explicitly.

(From OE-Core rev: 14b124122c1b7d10b9a3a96fe4617c6fc1c661c5)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Cc: mihaix.lindner@linux.intel.com
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package_deb, apt.conf: fix apt failure in do_rootfs()

The changes in this commit, dc5f6c3898555b59f16bf809ae4c5418656e6ac9,
moved apt config directory from native SYSROOT to WORKDIR.
Unfortunately, Dir::Etc in apt.conf was not changed accordingly and
sources.list file could not be found during do_rootfs().

This commit fixes this issue.

[YOCTO #5241]

(From OE-Core rev: 65a2a1e1d0b0e3dcc34d47f46185f24d753b53dd)

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>

elfutils: remove configure bits from patch

That is a generated file and patching it might fail, the patch already patches configure.ac
which will have the configure file regenerated.

(From OE-Core rev: daa56093de3a518b79480c90ec5124a79189bf7b)

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

libpam: Avoid host contamination issue w. libprelude

Since we dont use prelude in OE, we just disable autodetection of
prelude in the libpam configuration.

Seems like an old bug:
http://lists.openembedded.org/pipermail/openembedded-devel/2012-March/083804.html

(From OE-Core rev: 9096c6a46cf2467c90873c235b4533faf97d6175)

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

makedevs: add nativesdk to BBCLASSEXTEND

This is a first in a series of patches to add items
to the nativesdk toolchain tarball.
Many of which are Yocto specific and needed to create a
rootfs with pre and post install hooks successfully run,
using only the toolchain tarball.

End goal is to create a sandbox where _users_ can customize a rootfs from a
package feed with their package manager of choice.

(From OE-Core rev: e70019bcdce83ae2a202338518f8725775abaa32)

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

autoconf: Add fix for sh verses bash issues

Libtool scripts were finding bash was /bin/sh and then using bashisms which
then got into sstate and used on machines where /bin/sh might be dash.

This changes things to search for bash first since its preferred. We then hardcode
bash into the scripts which is more correct.

This does mean we have a dependency on bash but many of our scripts have
that anyway.

(From OE-Core rev: 367a19aaf31bcf997f10d045e7954cc800189052)

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

qemu: don't claim support for IrDA and PCMCIA

QEMU machines don't have virtual IrDA or PCMCIA hardware, so don't claim to
support them.

(From OE-Core rev: 694ca965eea971077e135cda4e54fa1cb0243233)

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

boost: set up PACKAGECONFIG for boost-python library

Until now, the comment in the boost.inc file suggested to enable
the boost-python library by uncommenting some lines in that file.
Using the new PACKAGECONFIG feature, boost-python can now be added
optionally without need to modify the file or copying those lines
into a bbappend file.
Furthermore, we obtain the python version by inheriting python-dir
instead of fixing the python version in this file.

This commit is motivated by the need in the meta-ros layer, as
discussed in the issue #145 of the meta-ros issue tracker [1].

[1] https://github.com/bmwcarit/meta-ros/pull/145

(From OE-Core rev: 7568bfdd114597956a1da68746f207ec7f93a48d)

Signed-off-by: Lukas Bulwahn <lukas.bulwahn@oss.bmw-carit.de>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kern-tools: fix multi patch application without headers

Updating the kern-tools SRCREV to import the following change:

    kgit-s2q: always update ORIG_HEAD after applying changes

    In situations where git am fails to apply patches, and git apply is used,
    we must update ORIG_HEAD as well as HEAD. This is required, since if the
    next patch in the queue also fails git am application, it will reset to
    ORIG_HEAD before using git apply. If we haven't updated ORIG_HEAD, we'll
    end up warping back to the top of the branch each time.

This problem can only be seen in very specific situations, in particular if
a generated BSP branches from qemuppc, and has a series of non git "am able"
patches. We fail, since all of the qemuppc patches are not applied due to
the branch head constantly being reset.

(From OE-Core rev: 5126ac0aeb3154d31769dc20a46b6b1a6b2e3d9b)

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>

telepathy-mission-control: do_compile failed (race issue)

There might be an error when parallel build:

[snip]
Traceback (most recent call last):
  File "/path/to/tools/glib-gtypes-generator.py", line 304, in <module>
    GTypesGenerator(dom, argv[1], argv[2])()
  File "/path/to/tools/glib-gtypes-generator.py", line 295, in __call__
    file_set_contents(self.output + '.h', ''.join(self.header))
  File "/path/to/tools/libtpcodegen.py", line 42, in file_set_contents
    os.rename(filename + '.tmp', filename)
OSError: [Errno 2] No such file or directory
[snip]

This is a race issue, the _gen/gtypes.h and _gen/gtypes-body.h may
write(remove/rename) _gen/gtypes.tmp at the same time, then there would
be the error.

There was a similar bug in telepathy-glib which was already fixed, we use the
similar patch to fix it here.

[YOCTO #5184]

(From OE-Core rev: b0f81f460cf96798d79d72da7a3246c321caf654)

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

libav: make X11 dependencies dependent on DISTRO_FEATURES

This enables building libav when x11 is not in DISTRO_FEATURES.

(From OE-Core rev: 9b7131e855293064e044cd79259eb350a14f319b)

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

classes/testimage: add informational messages about tests and do some cleanup/formatting

Adds info about tests run even if they passed (in case of errors log will
be printed anyway). Also some style change for qemu object.

(From OE-Core rev: 4adcf960be9c7477e314788d42b80753c2aaf572)

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

image.bbclass: uninstall package only if it's already installed

Remove a package from rootfs only if it's already installed. Also,
if a package is uninstalled, remove it from installed_pkgs.txt.

[YOCTO #5169]

(From OE-Core rev: b86dc63bc87763119fce8286f37f44361da824d0)

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>

sstate.bbclass: ignore the tar failure.

sstate_package creates hardlink from sysroot to SSTATE_BUILDDIR, and
sstate_create_package stores SSTATE_BUILDDIR into a archive file by
tar, these two task can be run simultaneously for different packages,
and make a hardlink for a file will lead to the change of the links
number of file, and if tar is reading this file, it will fail with exit
code 1, and report "file changed as we read it":

    DEBUG: Executing shell function sstate_create_package
    tar: x86_64-linux/usr/share/aclocal/xorg-macros.m4: file changed as we read it

4b3e353a5[sstate.bbclass: fix parallel building issue] tries to use the
tar parameter --ignore-failed-read to fix, but it does not work, and
tar parameter --warning=no-file-changed can close the warning, but can
not change the exit code. so close shell immediate exit, only fail
if tar returns not 1 and 0.

Exit codes of tar:
    http://www.gnu.org/software/tar/manual/html_section/Synopsis.html

(From OE-Core rev: fad604b719e00b03e09da5fdb485e72332275b4a)

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

yocto-bsp: add 3.10/remove 3.8 kernel from templates

For Yocto 1.5, 3.10 is the preferred kernel and 3.8 is obsolete.

This also removes any mention of emgd from the templates - we want to
discourage users from using it - it will be obsolete soon in any case.

Fixes [YOCTO #5107]

(From meta-yocto rev: 4dd4bf6ac2dcc7652ec8f807df02298546bdb41b)

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

glib-2.0: fix broken python script header on machines using buildtools

With buildtools (which contains Python) installed on a build machine,
glib-2.0's gtester-report script was ending up with the full path to
the installed python binary in the shebang, which when rpm packaging
was used led to this being added as a per-file dependency by rpmdeps for
the libglib-2.0-utils package in which it ends up. This of course broke
do_rootfs when the package was included in the rootfs and had been
restored from sstate from another machine, as happened on the Yocto
Project autobuilder.

We were already trying to sed this script apparently only for the
shebang (since it appears that there are no other paths in the script)
so let's just sed the shebang properly; it also seems sensible to do
this for native as well instead of explicitly trying to exclude that
case.

Fixes [YOCTO #5205].

(From OE-Core rev: 1d16e8035dda062041394b1e51839a9a7d077cf5)

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

avahi: bump INC_PR to avoid do_configure failure in existing workdir

When we split S and B for avahi in OE-Core commit
6112a07f4e9865f7ae0e5a953669c1adf789f9f0, files left over in the workdir
from a previous build seem to break re-execution of do_configure. Bump
PR to give a fresh workdir and avoid this problem.

(From OE-Core rev: bdcddb4fa7ceb3408d687d4c39b0f631d3b31f96)

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

bitbake: tests/data: Whitespace in key names is a really bad idea

The parser never has supported it, the datastore API did happen to work
but whitespace in key names is a really bad idea and not something I think we
should encourage or support.

Fix test case failures after excplitly ignoring it for variable expansion
purposes.

(Bitbake rev: a2074ddaba6f53962d6caf34dbd27bdbc259935b)

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

bitbake: data_smart: Variable references don't contain newlines, spaces or tabs

The code is happily trying to expand variable names containing newlines,
spaces and tabs which are illegal characters in variable names. This
patch stops it doing this. This will change dependency checksums
since some rather weird dependencies were being attempted to be expanded.

(Bitbake rev: 37e13b852b33d98fa40f49dc1e815b3bbe912ff0)

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

bitbake: hob: populate "Save image" dialog when saving changes to a custom image recipe

If my build starts from a YP standard image recipe, the 'Save image
recipe' dialog should be empty.

If my build starts from a custom image recipe, the 'Save image recipe'
dialog should populate the 'Name' and 'Description' fields with the values
set for the custom image recipe.

[YOCTO #5004]
(Bitbake rev: 10757c529fe0b4b9a39740d269831347a3aab4a0)

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

bitbake: bin/bitbake: Improve --help text

The --help text was rather inconsistent in style and plain incorrect in places,
using confusing terminology in others. I guess most people know what the options
do and don't read this but its confusing to new users.

This updates it to use the terms recipe and task consistently, remove
the references to stage, bbread and generally try and make the output
more useful.

[YOCTO #4856]

(Bitbake rev: 516311946c7bd14c84947dc44c3bb0563e5a9667)

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

bitbake: build: Add BB_TASK_NICE_LEVEL to task code

On Linux its not possible for processes to regain a previous nice level after
it has changed. Its therefore not possible to have a core low priority and
then raise the priorities of individual tasks.

This variable allows us to do something like:

BB_TASK_NICE_LEVEL = "5"
BB_TASK_NICE_LEVEL_task-testimage = "0"

to give priority to specific tasks which the BB_NICE_LEVEL functionality
doesn't give us the option of.

(Bitbake rev: 94d82997220c6cfc7028f76719df028ba8254a5c)

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

bitbake: bitbake: runqueue: add task hash to Queue events

Adding the sstate-related hash for all runqueue and
scenequeue tasks, as it's needed in the WebHob data.

(Bitbake rev: b6e2ce1cf7a0ede890f08fabf536a556dc4263c5)

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

bitbake: bitbake: event: adding generic event for metadata usage

Adding the generic bb.event.MetadataEvent that is
targeted specifically at metadata usage. This is
needed in order to let the metadata code send and receive
events during asynchrous execution without having
to define each event specifically in Bitbake.

Metadata code should subscribe to and fire the MetadataEvent
in order to communicate asynchronously, and identify
the object using event.type field, and parse the
data in the event.data field.

Knotty UI will ignore these event by default.

This deprecates RequestPackageInfo/PackageInfo, and that
event pair will be removed in the future.

(Bitbake rev: ae1ea51aaab73e010d1c3db39df058bebebc11dd)

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

bitbake: bitbake: cooker, command: add a command to return global data

Adding the 'getAllKeysWithFlags' read-only command that will
return a dump of the global data state, together with specified
flags for each key. The flag list is passed in as the first
parameter to the command.

This will be used by UI clients to get the build configuration.

(Bitbake rev: 3e094da513e1220319288806cb76ddf804772afd)

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

bitbake: bitbake: cooker: get extra information from recipe cache

The loaded cache modules may add extra attributes to
the recipecache, that will be populated by the cache
classes required by the UI. These attributes
will be used by the UI to display relevant information.

Adds cachefields cache class field to specify
for each cache class which attributes will be set
in the recipecache.

Adds code to automatically expand depends tree with the
fields exported by the extra cache class.

Fixes a cache field name in the HOB UI.

(Bitbake rev: 47c171005fb3803d936e65fcd4436c643883ae16)

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

bitbake: bitbake: build, runqueue: adds info to the *runQueue* events

This patch adds task identifying information for all
runQueue and sceneQueue events, and for bb.build.Task* events.
This will allow matching event to specific tasks in the UI
handlers processing these events.

Adds RunQueueData functions to get the task name and task
file for usage with the runQueue* events.

Adds taskfile and taskname properties to bb.build.TaskBase.

Adds taskfile and taskname properties to the *runQueue* events

(Bitbake rev: b4a5e4be50d871a80dbe0993117d73f5ad82e38f)

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

bitbake: bitbake: cooker,runqueue: send the task dependency tree

Adding a CookerFeature that allows UIs to enable
receving a dependency tree once the task data has been
computed and the runQueue is ready to start.

This will allow the clients to display dependency
data in an efficient manner, and not recompute the runqueue
specifically to get the dependency data.

(Bitbake rev: 75466a53b6eece5173a9bfe483414148e4c06517)

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

bitbake: bitbake: cooker,xmlrpc,servers: implement CookerFeatures

Implementing feature set selection that allows a client
to enable specific features in the server at connection time.

Only enabling of features is supported, as there is
no way to safely remove data loaded into the cooker.
Once enabled, a feature will remain enabled for the
life of the cooker.

Client-server connection now supports specifying the feature
set required by the client. This is implemented in the Process
server using a managed proxy list, so the server cooker
will now load dynamically needed features based on what client
connects to it.

In the XMLRPC server the feature set is requested by
using a parameter for registerUIHandler function.
This allows observer-only clients to also specify features
for the server.

The server code configuration now is completly separated
from the client code. All hardcoding of client knowledge is
removed from the server.

The extra_caches is removed as the client can now specify
the caches it needs using the feature. The UI modules
now need to specify the desired featureSet. HOB is modified
to conform to the featureSet specification.

The only feature available is CookerFeatures.HOB_EXTRA_CACHES
which forces loading the bb.cache_extra:HobRecipeInfo class.

(Bitbake rev: 98e594837aab89ea042cfa9f3740d20a661b14e2)

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

gcc-target: Fix libatomic dependency tracking issues

The --enable-dependency-tracking option was added to workaround build
issues in libatomic. This fixes that build problem properly and removes
the flag since the dependency tracking code appears to be full of races
which are much deeper and harder to fix.

As per the automake manual, dependency tracking is only useful and worth
the build performance cost if you are doing more than one compile of the same
source code which in most cases we are not so this is a good thing anyway.

(From OE-Core rev: a3b665a80abed4c0659925e1cceb1568af023711)

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

gzipnative: Ensure dependencies apply to do_unpack and do_populate_lic

| DEBUG: Executing python function sstate_task_postfunc| DEBUG: Staging files from /srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/work/qemux86_64-poky-linux-gnux32/core-image-minimal/1.0-r0/license-destdir to /srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/deploy/licenses| NOTE: Using umask 002 (not 22) for sstate packaging| DEBUG: Preparing tree /srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/work/qemux86_64-poky-linux-gnux32/core-image-minimal/1.0-r0/license-destdir for packaging at /srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/work/qemux86_64-poky-linux-gnux32/core-image-minimal/1.0-r0/sstate-build-populate-lic/license-destdir| NOTE: Removing hardcoded paths from sstate package: 'find /srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/work/qemux86_64-poky-linux-gnux32/core-image-minimal/1.0-r0/sstate-build
 -populate-lic/ \( -name "*.la" -o -name "*-config" -o -name "*_config" \) -type f | xargs grep -l -e '/srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/sysroots/qemux86-64' | tee /srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/work/qemux86_64-poky-linux-gnux32/core-image-minimal/1.0-r0/sstate-build-populate-lic/fixmepath | xargs --no-run-if-empty sed -i -e 's:/srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/sysroots/qemux86-64:FIXMESTAGINGDIRHOST:g''
| DEBUG: Executing shell function sstate_create_package
| gzip: /lib64/libz.so.1: version `ZLIB_1.2.5.1' not found (required by gzip)
| tar: Child returned status 1
| tar: Error is not recoverable: exiting now
| WARNING: /srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/work/qemux86_64-poky-linux-gnux32/core-image-minimal/1.0-r0/temp/run.sstate_create_package.20384:1 exit 2 from
|   tar --ignore-failed-read -czf $TFILE license-destdir
| DEBUG: Python function sstate_task_postfunc finished
| ERROR: Function failed: sstate_create_package (log file is located at /srv/home/pokybuild/yocto-autobuilder-new/yocto-slave/nightly-x32/build/build/tmp/work/qemux86_64-poky-linux-gnux32/core-image-minimal/1.0-r0/temp/log.do_populate_lic.20384)
NOTE: recipe core-image-minimal-1.0-r0: task do_populate_lic: Failed

Imagine:

pigz-native is used from sstate.
zlib-native is getting rebuilt.

pigz-native has some special handling to ensure its not used until the
system is ready. This is through a class and installing into a subdir of
PATH which only gets added in when we believe its available. We use
pigz-native in the image generation code and its in DEPENDS.

DEPENDS are guaranteed available for do_configure. do_populate_lic can
run before do_configure so the DEPENDS isn't met and I think this is our
corner case.I suspect ways of fixing this are to either:

a) force do_populate_lic after do_configure everywhere
b) statically link pigz-native
c) add in an explicit dependency to gzipnative.bbclass forcing
do_populate_lic after do_configure. If do_unpack handled a tar file in
an image, it would also be at risk of course.

Looking at each, a) is overkill and our dependency tree is nasty enough
already. b) sounds nice but is also risky since what happens if the gzip
binary is half copied when we run it. Our hardlink copying should deal
with that but I'm still nervous. This leaves us with c) so we could do:

do_unpack[depends] += "gzip-native:do_populate_sysroot"

The reproducer is:

bitbake pigz-native
bitbake zlib-native -c clean
bitbake core-image-minimal -c populate_lic --no-setscene -f

however your system needs to have an old version of zlib on it which
pigz-native can't run against. The line above fixes it.

(From OE-Core rev: 16cb83d13bdb7e1abc88c18ec224efcd9ceef4f6)

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

btrfs-tools: Fix parallel make issue

btrfs-tools was failing occasionally due to version.h being missing. This
fixes the problems, thanks to several people on #yocto for helping out
why my lack of make knowledge of old fashioned suffix rules :)

(From OE-Core rev: 7be8010186889cece97829025d97d94f21c1f2e5)

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

uboot-config.bbclass: Handle UBOOT_MACHINE setting

The class now consolidate the handle of UBOOT_MACHINE and UBOOT_CONFIG
variables and handle possible mistakes done by user when using these
variables.

(From OE-Core rev: 87d9b585b2784bec04e9e244dc52c6e929484fd0)

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

u-boot, u-boot-fw-utils, u-boot-fw-utils-cross: Use uboot-config class

The U-Boot configuration has been consolidates into a single class to
avoid code duplication. This is now done by uboot-config class, so we
now use it.

(From OE-Core rev: d0bc7a53b1e61283fb155b4dcb67bc2ab3b1d7f0)

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

uboot-config.bbclass: Use an annonymous python function

The processing needs to happen per recipe and thus it ought to use
annonymous python function instead to be triggered at event.

(From OE-Core rev: 75bde3ee02262cb3c6b91279ca277e3e5324ee5e)

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

oprofileui-server: use two digit initscript sequene

initscript sequence numbers are 00-99, so using 999 resulted in systemd warning
that it couldn't find "9oprofileui-server".

(From OE-Core rev: dac7dd70eb7a55ada5fdfc224aba5bac7c49f63f)

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

libpam: only use pam_systemd.so if systemd is enabled

So that sysvinit images don't warn on every login only add it to common-session
if systemd is a DISTRO_FEATURE.

[ YOCTO #3805 ]

(From OE-Core rev: 3ccb0855a7a6b147e5025855c6376747ba72986a)

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

x11vnc: respect zeroconf DISTRO_FEATURE to enable Avahi support

There's a standard way of announcing VNC services over mDNS that x11vnc
supports, so respect the feature and enable/disable it.

Also re-order the statements and drop the redundant PR.

(From OE-Core rev: e5443a0b5a70bf054cbeb6ff1fd6b5ef9d2347f8)

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

pseudo: force NO32LIB to true for nativesdk

The exported SDK only needs simulated root privileges for specific
tasks, such as the user-mode NFS server or rootfs extraction, and
oe-core does not support multilib builds in the generated SDK, so
it is neither necessary nor possible to build a 32-bit libpseudo.so
for a 64-bit SDK.

[YOCTO #5135]

(From OE-Core rev: 908b179b798704db74c886ec4c70c0942b09cb00)

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

runqemu: set qemuarm memory size back to 128MB

The following commit, 6ccd4d6, increased the RAM size for qemu machines
to 256MB due to some smart sanity tests failing on autobuilder because
more memory was needed.

Unfortunately this leads to various, potentially dangerous, issues like
the one observed during sudoku-savant project compilation:

collect: relinking
collect2: error: '_ZNK6sudoku5ClearINS_6SquareEEclERS1_' was assigned to
'board.rpo', but was not defined during recompilation, or vice versa
board.o:(.rodata+0x8): undefined reference to
`sudoku::Clear<sudoku::Square>::operator()(sudoku::Square&) const'
board.o:(.rodata+0x20): undefined reference to
`sudoku::Clear<sudoku::Sequence>::operator()(sudoku::Sequence&) const'
board.o:(.rodata+0x34): undefined reference to `typeinfo for
sudoku::Action<sudoku::Sequence>'
...AND THE LIST CONTINUES...
collect2: error: ld returned 1 exit status
make: *** [sudoku-savant] Error 1

After some tests, I found that the maximum amount of memory needed for
sudoku to compile properly is 146MB(!?!).

My attempts to create a simpler test case (using templates), in order to
replicate and isolate the issue failed. All the tests compiled just
fine.

So, my guess is that this problem is certainly memory related but the
cause might be hidden in any of the following: qemu versatile hw model,
in the kernel or, highly unlikely but not impossible, the toolchain
itself. The reason I don't really think the cause is in the toolchain is
the fact that the compilation completes just fine for 128MB on qemuarm but
also on other qemu machines (with 256MB of memory).

Since this issue might need lots of time to have a proper fix, I'll revert back
to using 128MB for qemuarm for the time being.

[YOCTO #5133]

(From OE-Core rev: 06605bd6ddd4d6a788e1a107dcf15dde1027c094)

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

classes/imagetest-qemu: remove old image testing class

This has now been superseded by testimage.

(From OE-Core rev: d469c92394a1a95ae7a45b8b80dc4c2918e0e9a6)

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

classes/sanity: test for DISPLAY being set with testimage class

Update the sanity test for DISPLAY being set to handle the new testimage
class rather than the old imagetest-qemu class.

(From OE-Core rev: d1297c2c3ae71de0e9e5cab36e582f5df8666391)

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

classes/testimage-auto: add class to allow automatically running image tests

Setting TEST_IMAGE = "1" alone will now automatically run tests on the
image immediately after the image is built instead of having to add
INHERIT += "testimage" and run bitbake -c testimage <image> manually
(but that will still work). This restores functionality that was
present in the older imagetest-qemu class with IMAGETEST.

(From OE-Core rev: 72269a8fbec35c39af30fbabb1fa9ca7c5ee8d69)

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

classes/testimage: remove odd characters in comments

(From OE-Core rev: 5d46c3df578f44597d8ce70add423e0c15982705)

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

findutils_4.2.31: backport fixes for doc build errors

[YOCTO #5212]

(From OE-Core rev: cb41aff1ac343b4fa8b4c9d43cc38652f7593ee6)

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

libunistring: remove the test to convert euc-jp in configure

Remove the test "Test against HP-UX 11.11 bug:
No converter from EUC-JP to UTF-8 is provided"
since we don't support HP-UX and if the euc-jp is not
installed on the host, libunistring will be built without
iconv support and will cause guild-native configure fail.

(From OE-Core rev: 0470bd7a9658d3d8aa10e9d081f42b61b9b7a6f4)

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

pciutils: remove pcimodules

pcimodules can not work due to no modules.pcimap file, and has been
replaced by "lspci -k", so we can remove it.

Update lib-build-fix.patch since remove pcimodules-pciutils.diff.

[YOCTO# 5210]

(From OE-Core rev: 57bec61b05c0fea42f988dd52942c7d5ce8833fd)

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

lib/oeqa/runtime: cleanup and improve output readability for some tests

   - move everything in the same test. setUp/tearDown aren't quite
the right thing here, everything it's part of the same test. (and
it get's confusing when ssh fails)

ldd:
   - change test name and add output to error message

vnc:
   - remove unnecessary check as there is no point in doing
both ps and netstat. Also improve error output a bit.

(From OE-Core rev: 2f91bb438cfcdd0a40daed5902b6e98fc0aee67f)

Signed-off-by: Stefan Stanacar <stefanx.stanacar@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/oeqa/utils: sshcontrol: log how long the last command take

It might be useful for debugging to have in the ssh log
the number of seconds a command has run.

(From OE-Core rev: 48b9e45b9716130b015ae2ab7398d6aa243933dc)

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

qt-mobility: fix build without X11

else we get :
arm-oe-linux-gnueabi-g++  -march=armv7-a -mthumb-interwork -mfloat-abi=softfp -mfpu=neon .../...
ibQtSystemInfoE.so.1 -o libQtSystemInfoE.so.1.2.0 .../... -lblkid -ludev -lX11 -lXrandr -lQtDBusE -lQtXmlE -lQtGuiE -lEGL -lQtNetworkE -lQtCoreE -lpthread
.../build/tmp-defaultsetup-eglibc-eglibc/sysroots/x86_64-linux/usr/libexec/armv7a-vfp-neon-oe-linux-gnueabi/gcc/arm-oe-linux-gnueabi/4.8.1/ld: cannot find -lXrandr
make[2]: *** [../../lib/libQtSystemInfoE.so.1.2.0] Error 1

(From OE-Core rev: 21857f0842501b0c72c77e5fb290aec85b1c28ff)

Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wipe-sysroot: delete pkgdata stamps

The pkgdata stamps now need to be wiped away if the sysroot is destroyed.

(From OE-Core rev: e6ef8399cd8e97b9cd59855e11f1792445f0e65b)

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

insane: don't abort if workdir is not TMPDIR/work

The BASE_WORKDIR variable can be used instead of enforcing WORKDIR being
TMPDIR/work (and aborting the build if it isn't).

(From OE-Core rev: 176a36ace1624f3bbe498307aeabbd7935de14e6)

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

bitbake.conf: define WORKDIR in terms of BASE_WORKDIR

To make it easier to move WORKDIR, define it using the new variable
BASE_WORKDIR, which is the root of the work directory.

(From OE-Core rev: 1eee097f2e29b9d6934711c0b1d32e59e9542f53)

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

gcc-4.8: temporary disable thumb

* temporary work around for build issue on armv4t:
  | cp/decl.o: In function `bad_specifiers':
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7171:(.text.unlikely+0x24): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7173:(.text.unlikely+0x32): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7176:(.text.unlikely+0x3e): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7180:(.text.unlikely+0x4c): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7182:(.text.unlikely+0x5a): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7185:(.text.unlikely+0x66): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7189:(.text.unlikely+0x74): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7191:(.text.unlikely+0x82): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7194:(.text.unlikely+0x8e): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7198:(.text.unlikely+0x9c): relocation truncated to fit: R_ARM_THM_CALL against symbol `error(char const*, ...)' defined in .glue_7 section in linker stubs
  | gcc-4.8.0-r0/gcc-4.8.0/gcc/cp/decl.c:7200:(.text.unlikely+0xaa): additional relocation overflows omitted from the output
  | collect2: error: ld returned 1 exit status
  | make[2]: *** [cc1plus] Error 1

(From OE-Core rev: 148c7e990831ed708f7fb064ec8f05657a323850)

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

core-image-x11: Ensure x11 distro feature is enabled

(From OE-Core rev: 4cdf1601e0c5e46988f84a5abc38a201bbefec62)

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

qemu: Fix broken patch

(From OE-Core rev: 9934141d5df9d334f7f7f76e1990234af8a25a63)

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

linux-yocto/3.10: fix long perf compile times

perf's builtin-sched.c triggers extremly long build times on some
architectures due to gcc 4.7+ var-tracking functionality.

To fix this, we can cherry pick the 3.12 commit:

  f36f83f94 [perf sched: Move struct perf_sched definition out of cmd_sched()]

With this change build times are reduced from 15 to 20 minutes for qemuarm to:

  real    2m19.940s
  user    1m35.438s
  sys     0m11.165s

For kernel's that are not carrying this patch, the following can be added
to the perf recipe to also fix the issue:

    +++ b/meta/recipes-kernel/perf/perf.bb
    @@ -68,6 +68,7 @@ EXTRA_OEMAKE = \
    CC="${CC}" \
    AR="${AR}" \
    perfexecdir=${libexecdir} \
    +               EXTRA_CFLAGS="-fno-var-tracking" \
    NO_GTK2=1 ${TUI_DEFINES} NO_DWARF=1 ${SCRIPTING_DEFINES} \

(From OE-Core rev: 82ad5305381c2f541ef051a8fc28243cd91776fe)

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

linux-yocto/3.10: mips configuration changes

Updating the 3.10 SRCREV to import the following MIPS configuration changes

   4f689aa meta: remove ftrace/ftrace-disable feature
   3058d81 mips: have the mips BSPs disable function tracing instead of ftrace
   935f43f meta: add ftrace/ftrace-function-tracer-disable feature
   0d72a03 mti-malta64: Default to support o32 and n32 userspace binaries

The first three changes improve the ftrace disabling fragments, to allow tracepoints
and ftrace to be enabled, while only disabling dynamic ftrace. This allows tools
that required tracepoints (like lttng) to be built against MIPS.

The mti-malta64 change adds n32 and o32 support to the default configuration to
support a broader range of userspace binaries.

[YOCTO #5215]

(From OE-Core rev: b2cadab5c7a5d3f772c34c04b759823ed6a122bf)

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

linux-yocto/3.8: add haswell-wc board config and branch

Bumping the meta branch SRCREV to import the following board support:

  meta: add haswell-wc bsp for Intel Haswell Platform (Walnut Canyon CRB) scc and config files

Signed-off-by Ong Boon Leong <boon.leong.ong@intel.com>
(From OE-Core rev: fcca6317543e8be0bd2da1f45ac99448c24b4e48)

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

perf: Add LDFLAGS to allow build of old kernels without patching

The LDFLAGS is required or some old kernels fails due missing
symbols and this is preferred than requiring patches to every old
supported kernel.

Fixes [YOCTO: #5221]

(From OE-Core rev: 0eccbf2016e89e6f1c3796f138b02a508d2edbcf)

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

qemurunner: disable grabs in automated testing

Use the new QEMU_DONT_GRAB environment variable to disable grabs,
finally/hopefully solving the random hangs that the autobuilder has been hitting
for a while.

[ YOCTO #5131 ]

(From OE-Core rev: 32f9575a565f350649264c11eceba8311584b0fd)

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

qemu: add option to disable X grabs

When the mouse pointer enters the qemu window it takes a pointer grab. This
doesn't sound too dangerous at first but it turns out that SDL will infinitely
busy-loop if it can't get the grab (e.g. if the screen is locked) and the
average autobuilder setup's X server will have locked the screen a few minutes
after boot.

The result is that on many autobuilders apparently random qemu instances (the
top-most one under the pointer) will hang during boot.

To resolve this add an option (via an environment variable) to never attempt a
grab.  The default behaviour remains to grab so that everyone else doesn't see
any change.

(From OE-Core rev: a60b1ebbb8f81245f3ccf25b3f9d63677de75b85)

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

u-boot-fw-utils: Skip package if UBOOT_MACHINE is unset

The UBOOT_MACHINE variable needs to be set so the firmware utils can
know about some configuration settings which are board dependent. This
patch ensures the package is skipped in case UBOOT_MACHINE is unset
thus avoid its build in 'bitbake world' builds for incompatible
machines.

Fixes [YOCTO: #5223]

(From OE-Core rev: d9abcc0a2a691ca60cc1cb2f48f1748b0de73ac8)

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

bitbake: data_smart: Fix variable reference issues

The change to use the expansion cache in VariableParse was incorrect as
it was adding in references it shouldn't have been. This patch corrects
the codepaths and ensures the references are correct.

The cache version is bumped since the previous bug could have leave
to invalid checksum calculations and a clean cache is therefore desireable.

The impact of the bug was that sstate was not getting reused when it should
and some tasks were also being rerun when they should not have been.

(Bitbake rev: 8a42d082315bd6ce091d006bf83476db257fa48b)

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

bitbake: cooker: return a copy checkPackages

The syntax used for checkPackages implies a copy is returned but it
did not do so. Make it so.

This is fixes universe builds where error messaages were being shown
but there should have only been warnings.

[YOCTO #5222]

(Bitbake rev: 97db2a0792d605f27d434bc6a4acce52857deee1)

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

bitbake: usermanual.xml: Two new sections added to BitBake "Description"

1. Added a new section "Appending and Prepending (override style
   syntax)".  This section shows how the append and prepend operators
   work using the override style syntax.

2. Added a new section "Removing (override style syntax)".  This
   section describes the new "_remove" operator.

(Bitbake rev: 6983afab0ce8d82d102142636d5a570f7d86a844)

Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

perf: Ensure we use CFLAGS and LDFLAGS settings from kernel build system

The kernel build system does the right thing here and we should stop
overriding it. This code has been added based on a change from
'meta-metro' layer, revision 9d698004137c1a888d40d6a4808d94afa22387e7,
without any information about what problem it fixes so I am reverting
it.

Using the CFLAGS and LDFLAGS makes it impossible for kernel build
system to append to it, thus making the build fail in various ways as:

|     CC /.../perf/1.0-r8/perf-1.0/perf.o
| In file included from builtin.h:4:0,
|                  from perf.c:9:
| util/util.h:74:24: fatal error: lk/debugfs.h: No such file or directory
|  #include <lk/debugfs.h>
|                         ^
| compilation terminated.

The unset is done in do_compile and do_install otherwise it /rebuild/
perf as it detects the compiler options has change.

(From OE-Core rev: 2e90f8846db0f3ed99a175befff9ec67fe12bc4e)

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

cmake-native: fix dependencies

When building cmake for native, we don't use the system libraries and
thus cmake builds its own internal version of libarchive; this requires
zlib, bzip2, and e2fsprogs. We can add to DEPENDS for the two former
libraries and patch out the latter.

(From OE-Core rev: fe6ca47a256b775e1aa5750b9dd31e27230cf781)

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

mesa: add virtual/mesa provider

As there are two alternative mesa recipes (mesa and mesa-gl), there needs to be
a virtual provider that recipes that explicitly need Mesa (such as xserver-xorg)
can depend on.

(From OE-Core rev: 4a407568472d3c87cd2ce11baf199568249640b6)

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

libtool: fix resolve of lt_sysroot

This patch updates libtool.m4 (and its output) to resolve a problem
with variable 'lt_sysroot' not being properly updated if the option
'--with[-libtool]-sysroot' is not provided when running the 'configure'
script for a package.

According to the help text ouput from 'configure':
--with-libtool-sysroot=DIR Search for dependent libraries within DIR
                        (or the compiler's sysrooot if not specified).

Due to mixed up cases in a switch statement, when checking if the option
was specified or not, wrong actions were taken resulting in an incorrect
sysroot and failures to properly locate e.g. .la files when using the
populated SDK toolchain.

For current upstream status see:
http://lists.gnu.org/archive/html/bug-libtool/2013-09/msg00005.html

(From OE-Core rev: f5cf7e1a5c85fb320faa9cbeef24f491706b4c1d)

Signed-off-by: Hans Beckerus <hans.beckerus at gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

avahi: fix and enable out-of-tree builds

(From OE-Core rev: 6112a07f4e9865f7ae0e5a953669c1adf789f9f0)

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

python-smartpm: truncate the filename to meet NAME_MAX

There is a "File name too long" error when len(TMPDIR) = 410, this is
because the function getLocalPath() converts the filepath into the
filename, so there would be the error when len(filename)
> NAME_MAX, truncate the filename to meet NAME_MAX will fix the problem.

[YOCTO #5201]

(From OE-Core rev: 9f0427edee6bf62d3fe7cdceb07f59a5776c8c4f)

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

u-boot-fw-utils: reinstate target recipe for fw_printenv

The latest upgrade changed u-boot-fw-utils to u-boot-fw-utils-cross, which
removed the on-target fw_printenv (and fw_setenv).

Re-create the updated version of the u-boot-fw-utils recipe.

U-Boot does try to strip the fw_printenv binary by default now. To avoid
this, without patching the Makefile, we add HOSTSTRIP=true to EXTRA_OEMAKE.

The new U-Boot do also require us to configure the build for a specific
machine, thus we change the package arch.§

(From OE-Core rev: 83cb07047658e2e2b2d721e1b01553d4df3d8636)

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

oeqa/runtime/smart: Increase timeout to 1500 test

(From OE-Core rev: a99edb5552839fd50326dd3aa4ee2f36f6026882)

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

weston: change mesa dependency to virtual/mesa

(From OE-Core rev: 65667415ecc8abdf1f15a48c5520ed78140afb60)

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

xserver-xorg: change mesa dependency to virtual/mesa

(From OE-Core rev: 57197949a980e8578cb5d758feb0e595e1f721e4)

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

mesa-gl: add GL-only Mesa recipe

Some machines have hardware-specific GL drivers that do EGL and GLES (many ARM
boards).  Others have their own EGL/GLES drivers and provide a Mesa DRI driver
(EMGD).  Previously adding Mesa, for software GL/GLX rendering in the first case
and hardware GLX in the second, involved bbappends and changing Mesa to be
machine-specific.

By adding a just-GL Mesa the machine definition can combine it with the hardware
drivers cleanly.

(From OE-Core rev: f5a3a4bc33109181c741a2e66c13d0b45566e8fa)

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

linux-libc-headers: do_install: fix "Argument list too long" error

There would be an "Argument list too long" error when the TMPDIR is in a
deep dir, for example, when "len(readlink -f TMPDIR) >= 350 (our
supported value is 410)". Use "$(foreach ,,$(shell echo))" to fix it.

There was already a patch which tried to fix this issue, so squash the
current change into the previous one as Bruce suggested.

[YOCTO #5138]

(From OE-Core rev: 4e10c45d7c61ac1ccf8b53ef525ca03d3d458bba)

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
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>

util-linux: move su to /bin for lsb command check test

In busybox, util-linux and shadow, su has been moved to /usr/bin/,
but lsb cmdchk needs su in /bin.

Move su to /bin could fix this issue.

[YOCTO#5175]

(From OE-Core rev: ba84662bd9fb7575e68c87449e986535a2216b30)

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>

shadow: move su to /bin for lsb command check test

In busybox, util-linux and shadow, su has been moved to /usr/bin/,
but lsb cmdchk needs su in /bin.

Move su to /bin could fix this issue.

[YOCTO#5175]

(From OE-Core rev: 464d274e45bb94d9990577d85aa5ad02eb5da99c)

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>

busybox: move su to /bin for lsb command check test

In busybox, util-linux and shadow, su has been moved to /usr/bin/,
but lsb cmdchk needs su in /bin.

Move su to /bin could fix this issue.

[YOCTO#5175]

(From OE-Core rev: daff19fe6f0490dc7036602e8b0ca40a23b55556)

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>

bitbake: Revert "bb.fatal: Raise a BBHandledException instead of exiting"

Sanity test failures are no longer fatal with this change so whilst
its the right idea, the code paths need more work.

This reverts commit a50017ba71250e1710a6425b60ac7e3f03d88295.

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

openssh: use volatiles for managing /var/run/sshd

It fixes the following failure:
"fatal: Missing privilege separation directory: /var/run/sshd"

when sshd is started through xinetd.

(From OE-Core rev: a343c32891aa46a7f7d5f0cc6d1266a387900dad)

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>

Revert "initrdscripts: look for new systemd-udevd location"

systemd-udevd is back in /lib, so revert this change.

This reverts commit 27bb516be433ed9dcde6fe4a17d2429951a010a0.

(From OE-Core rev: 3f6324a86cb8c1c253af06a1033ac71fa61c58d3)

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

imake: remove

Remove both imake and the associated xorg-cf-files, as nothing has used imake to
build for a long time.

(From OE-Core rev: aa6d9296e7c7e5b96d868d9c17d639b4f33ca18e)

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

transfig-native: remove, nothing depends on it

(From OE-Core rev: 3ba57642ceca73c601e4297e1b2abd8a4ccd671b)

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

rpm: fix typo in PACKAGECONFIG

Fix typo, change PACAKGECONFIG to PACKAGECONFIG.

(From OE-Core rev: 77363d06121ceec264e06165ddda7b829c963301)

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>

python: fix failures of LSB python-runtime tests

It has been fixed in:
http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=e261c1760d8a660c13cd00039a4812d51f47f3df

But there is a typo in:
http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=78ac027f2ac6c6663fff7080eabbd3d09c1241bb
It missed to remove an extra "--with-wctype-functions" in EXTRA_OECONF

[YOCTO #5154]

(From OE-Core rev: 4672b10ff9b6fcbfc446cabc4323387be21053cb)

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>

kbd: Allow resizecons on x86_64

(From OE-Core rev: cec49d5083f749d9886207e88648ddd4c12fbc87)

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

kbd: Fix coding sytle

(From OE-Core rev: 2ababaf74255e7ad43cc951f2f2bc569ef955837)

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

LSB 4.1 Library Check: fix unable to find library libqt-mt.so.3

The libqt-mt is tested by lsb-dist-checker and lsb-test-desktop,
and it locates in meta-qt3 layer.
So if meta-qt3 is not added, there should be a warning to call
attention; if added, it will add libqt-mt to RDEPENDS.

[YOCTO #5153]

(From OE-Core rev: f646f96015b408f4c6d56998f08178a69ed9f2a6)

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>

cdrtools-native: Update from 3.00 to 3.01a17

The update is needed to support generation of EFI boot images that
work with optical media.  Specifically the "-eltorito-platform efi"
capability for mkisofs is needed.

[YOCTO #4100]

(From OE-Core rev: aaa85b0706197423786dfeae386dbb402ee15684)

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

bootimage.bbclass: Move fat image creation into a function

In order to call the fat image creation multiple times it needs to be
in its own function.  A future commit will make use of the new
function to additionally create EFI image files for use with an ISO.

[YOCTO #4100]
[YOCTO #1913]

(From OE-Core rev: 6d5181dc68766f42416a41f4988e8400d37fd7fa)

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

coreutils: set acpaths to avoid "Argument list too long" error

There would be an error when the TMPDIR is long/deep, for example when
len(TMPDIR) = 350 while our supported longest value is 410:

[snip]
aclocal: error: cannot open xxx
autoreconf: aclocal failed with exit status: 1
ERROR: autoreconf execution failed.
[snip]

Let aclocal use the relative path for the m4 file rather than the
absolute would fix the problem.

Another fix is that we can modify autotools.bbclass to let it use the
relative path rather than the absolute, but I don't think that we have
to do that based on the following 2 thoughts:

* The coreutils is the only recipe which has this issue as far as we
  know when len(TMPDIR) <= 410, because it has the most amount of m4
  files (more than 400 ones).

* That would impact all the recipes which use autotools.bbclass, and we
  are not sure about the side effect, for example, it would break the
  build there is a sub-configure.

[YOCTO #2766]

(From OE-Core rev: 22ac874512c2c1213aae8e1644bd59050b37a63c)

Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

vala: add bison-native depends for vala-native

vala-native builds were working because most systems have
bison installed on the host. Add an explicit dependency on
bison-native since bison isn't a sanity checked host tool.

(From OE-Core rev: 4bcc087290661544dd5f6466d2d6ab74488f34ec)

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

udev: avoid autosuspend of USB input devices

Some of the external USB devices has internal USB hub, which
make them look like "fixed" rather than "removable". And USB
autosuspend does not work with some of these devices resulting
in inoperable pointing device.

Now the code detect these false "fixed" devices by looking at their
parents.  If any of their parent is "removable", then USB autosuspend
is not enabled for that device, which keeps the pointing device
functional.

Fixes bug:
[YOCTO #5166]

(From OE-Core rev: d74a0ecdbc85a482cab6e7eae8dcb48185d44d84)

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-libc-headers: Fix comment typo spotted by Mark Hatle

(From OE-Core rev: 27e7162840482c7790480be5f52c7651d5b71794)

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

webkit-gtk: remove compile-three-times hack

Remove the hack to handle Make 3.82 which ran make three times, as we sanity
check Make and refuse to build with the broken release.

(From OE-Core rev: dccd55eaeaee123238372c02c34d476d68816232)

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

connman: Try to figure out NFS root interface if using DHCP

The commit c73487302270a7c2d3bbbb191ce1dd63ee2016dc missed the case
where the NFS is using DHCP.

[YOCTO #5176]

(From OE-Core rev: 29be8e79a200d33555d2887578975e33b8417795)

Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libxml2: remove patch for CVE-2012-2871

This CVE patch is actually against Chromium as they ship an internal fork of
libxml2 and breaks ABI.  The real issue has been resolved in libxslt 1.1.27, and
we're shipping 1.1.28.

(From OE-Core rev: e6c60252ab4ba6842f63c6b8a519a85f2ff238fb)

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