sanity.bbclass: Detect empty $PATH components too

Empty components in $PATH have the same effect as a . in $PATH,
and are a common side-effect of inserting a misspelled or unset
shell variable in $PATH.

(From OE-Core rev: ac4201b714c83c614113bfa735d0a2fc6f64db99)

Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

blacklist.bbclass: Refactor, use PNBLACKLIST[pn]

Revise the handling from ANGSTROM_BLACKLIST to PNBLACKLIST[pn].

Refactor the code to eliminate references to the distribution and recipe
name in the message.

Change the skipPackage message message from:

ERROR: <recipe> was skipped: <distro> DOES NOT support <recipe> because <reason>

to:

ERROR: <recipe> was skipped: Recipe is blacklisted: <reason>

(From OE-Core rev: 0893457465e184550906844fbab2dec397201835)

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

blacklist: fix typo in name

(From OE-Core rev: 02a5fe21cad2ec29268de7f3a556bb827f726998)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Import directly from meta-openembedded commit:  a63c374cdc785ade69d2998978d08280e671dc1f

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

h264_qpel_mmx.patch: fix Upstream-Status tag

(From OE-Core rev: 1410f735bca5d1ad2ad1c75fa17288d223d078a0)

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

bitbake/runqueue: Drop check_stamp_fn, it's obsolete and should not be used

This function was used by old code such as packaged staging but is thankfully
obsolete now and replaced with better mechanisms. Its time to remove it and
the horrible internal only variables associated with it.

(Bitbake rev: 2995b8d551e0532eca20f8862730acd062c608ad)

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

bitbake/runqueue: Drop check_stamps function, it is obsolete and unused

This was originally used to check the consistency of the stamps in one function
call. This turns out to be inefficient, unnecessary and if it were necessary,
check_stamp_task() could be called in a loop. The function has been unmodified
for a while and likely contains bugs. Its best simply removed.

(Bitbake rev: 728ffde1bd69b880d48fe8523b1616956d13616e)

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

bitbake/runqueue: Fix 'full' stamp checking to be more efficient and cache results

This should fix issues where bitbake would seemingly lock up when checking
certain configurations of stampfiles.

The cache is kept within the runqueue since that feels like the right
place to associate this cache data.

(Bitbake rev: e95755c4931b26d9f8102ed3652dff969145cfc9)

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

runqueue: drop unnecessary keys() in runqueue_process_waitpid

(Bitbake rev: d03dc07dea2f4e594fdbe4abe618670fe628a7c9)

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

runqueue: handle task exit due to signal/stop

- for a normal exit, use WEXITSTATUS, rather than manually shifting
- for exit via signal, set the exit code to 128+N, per shell convention
- if a process was stopped, return and don't handle it, as the process can yet
  be continued

This should fix the case where bitbake says a task failed with an exit code of
0 (we assumed failure based on the overall status, but didn't pass all the
information along to task_fail).

(Bitbake rev: 84ea614bc56d35a414eb5bf5658891b340bfc569)

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

Hob: change the workflow about click 'Run Image' to run directly after builded a qemu image

[YOCTO #2155]

(Bitbake rev: d8a1fc4613ce1d6c1d0e3dad53d7af15b2011fd5)

Signed-off-by: Liming An <limingx.l.an@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: add group number show and 'bold' visible effect for Package view

as ui design, add number of packages for each group, and make the
text font to 'bold' when row be selected

[YOCTO #2195]

(Bitbake rev: 5812176a679a39a4d096134c871c3d24a7e505ea)

Signed-off-by: Liming An <limingx.l.an@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: add original url show function with the tooltip hyperlink for user

When case about No browser, such as running in 'Build Appliance', user can't open
the hyper link, so add this work around for user. (Checking the browser is avaiable
or not is hard by different system and browser type)

[YOCTO #2340]

(Bitbake rev: 02cc701869bceb2d0e11fe3cf51fb0582cda01b0)

Signed-off-by: Liming An <limingx.l.an@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: change the refresh icon speed to make it view clear

Because the arrow icon refresh so fast as the go backward by illusion, so adjust it slow.

[YOCTO #2335]

(Bitbake rev: ac4a8885fafdc0d1e79831334ead9a8ddb6e2472)

Signed-off-by: Liming An <limingx.l.an@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

eglibc: Add patch to fix /var installation location

This address the packaging issues since it's installed in the wrong place
ERROR: For recipe eglibc-nativesdk, the following files/directories were installed but not shipped in any package:
ERROR:   /var
ERROR:   /var/db
ERROR:   /var/db/Makefile

(From OE-Core rev: 910452727fc277c1caec7612b36c37b58d845350)

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

libc-package.bbclass: Add MLPREFIX while dynamically set RDEPENDS

We need to add MLPREFIX when set RDEPENDS by setVar() function.

(From OE-Core rev: 9407895c88e46490d0ef9bb185f73466e7204c15)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image_types.bbclass: redefine EXTRA_IMAGECMD_jffs2 to leverage siteinfo

(From OE-Core rev: 7c803cea58737ea0abc62cd21c3813d955f07224)

Signed-off-by: Ting Liu <b28495@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

zlib: upgrade to 1.2.7

1.2.6 is no more fetchable and 1.2.7 fix some bugs, full changelog is
available here : http://zlib.net/ChangeLog.txt

(From OE-Core rev: 2f071e32128d8c022277c8af72dae47f60cf212b)

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

guile: upgrade from 2.0.3 to 2.0.5

(From OE-Core rev: 29afcb7ad976db62f9a46abf305a47a24a99dbda)

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

openssl: Fix build for mips64(el)

(From OE-Core rev: b2f89bd7bf3ae86b7db8cbef4b103d61cb68a218)

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

sanity.bbclass: Implement initial toolchain sanity checks

This introduces a sanity check for the toolchain, which verifies
each tuning (including any multilibs), producing meaningful diagnostics
for problems, and also provides some higher-level tuning features.

The TUNEVALID and TUNECONFLICT/TUNECONFLICTS settings were not
implemented.  Listed one or two missing features in TUNEVALID,
also (in a previous patch) fixed the references to
features which didn't exist.

This patch also provides a whitelisting mechanism (which is completely
unused) to allow vendors providing prebuilt toolchain components to
restrict tunings to those based on or compatible with a particular ABI.

(From OE-Core rev: 2a91ff0ba0d587c516a5a972553280364853faa4)

Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qt4-graphics-system: set graphic system in profile instead of X11 session

To allow the use of Qt applications started manually to use the
configured graphics system the setting needs to be available in user
environment. This moves the setting to /etc/profile.d ensuring it is
set and available. This allow the removal of x11-common as dependency.

(From OE-Core rev: 3804784b6200f82f5d8d6f533ce5e1a36ee2aeac)

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

libatomics-ops: Make it build for SH4

(From OE-Core rev: 0023aa21687287754c972c5e8edd527822e4e7c0)

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

kmod: fix upgrade path from module-init-tools

* PROVIDES does not need _${PN}
* CONFLICTS should be RCONFLICTS
* opkg needs RREPLACES to automatically remove older module-init-tools
  instead of just reporting it conflicts with kmod

SHR root@gjama ~ $ opkg install kmod
Installing kmod (7+git02629fa02e96763db7460a930239cc93649a52f8-r0.0) to root...
Downloading http://jama.dyndns-home.com/org.openembedded.shr-core//armv4t/kmod_7+git02629fa02e96763db7460a930239cc93649a52f8-r0.0_armv4t.ipk.
Removing package module-init-tools-depmod from root...
update-alternatives: removing //sbin/depmod as no more alternatives exist for it
Removing package module-init-tools from root...
update-alternatives: Linking //sbin/insmod to ../bin/busybox
update-alternatives: Linking //sbin/modprobe to ../bin/busybox
update-alternatives: Linking //sbin/rmmod to ../bin/busybox
update-alternatives: removing //sbin/modinfo as no more alternatives exist for it
update-alternatives: removing //bin/lsmod as no more alternatives exist for it
update-alternatives: Linking //sbin/lsmod to ../bin/busybox
Configuring kmod.
update-alternatives: Linking //sbin/insmod to /sbin/insmod.kmod
update-alternatives: Linking //sbin/modprobe to /sbin/modprobe.kmod
update-alternatives: Linking //sbin/rmmod to /sbin/rmmod.kmod
update-alternatives: Linking //sbin/modinfo to /sbin/modinfo.kmod
update-alternatives: Linking //bin/lsmod to /bin/lsmod.kmod
update-alternatives: Linking //sbin/lsmod to /bin/lsmod.kmod
update-alternatives: Linking //sbin/depmod to /sbin/depmod.kmod

(From OE-Core rev: 8de7f8045f28aecfe796afcdb013c557ce9d1372)

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

kmod: Use base_libdir for installing libkmod

This also helps avoid QA errors about binaries
accessing contents from /usr/lib

Also fixes emptry libkmod problem since now
the files are installed in expected place

(From OE-Core rev: 6b74f2461735272bd950a4f060dab6e778a36f92)

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

kmod: Update to .8+ release

(From OE-Core rev: 8dd0e28809dba8ce6d42c127041c591664c81c59)

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

kern-tools: fix do_patch errors

The linux-yocto repository and scripts can support a mode of
meta data management that merges a base meta branch to every
BSP branch. In this case, the scripts don't have to restore
a checkpoint for the meta data to be globally accessible.

The decision to restore or not is made based on whether or
not the meta branch is part of all branches or not.

The linux-yocto recipes have a sanity check to determine if
the requested SRCREV for meta data matches the head of the
meta branch (via do_validate_branches). If the wrong commit
is at the head, the meta branch is moved aside and the branch
reset to the right commit. This creates two meta branches that
contain the base meta data.

The test for integrated meta data mistakes this for a globally
merged set of meta data and doesn't restore the checkpoint, which
leads to build failures.

The immediate fix is to allow two branches to have the meta
data. The long term fix is to make the detection only consider
if the build branch contains the meta data.

(From OE-Core rev: bd794b92d12ceda2728520701e980b7a3cabd23d)

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

kmod: include libkmod also from libdir

* until it's decided it should be installed in base_libdir or libdir
* fixes libkmod packaging

(From OE-Core rev: 87249372b4a9951f9e4dc6acece950718382a164)

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

kern-tools: integrate minor fixes

Updating the SRCREV to pick up two minor fixes:

1/2:
    kgit-init: correct spelling of createme

    kgit-init copies the kern-tools scripts and intends to copy createme.

    The typo is in the usage() of updateme as well.

    (From OE-Core rev: 043871d7e5d2d19c2ff43e54d2ff180c09e8903e)

Signed-off-by: Michel Thebeau <michel.thebeau@windriver.com>
2/2:
    kconf_check: fix bad quoting around missing_required.cfg

    missing_required.cfg won't have it's path truncated (if applicable), since
    the quoting it wrong.

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

linux-yocto: streamline support for multiple upstream repo types

In order to support repositories of various types (with or without
meta data, branched, pristine, custom, etc) information about the
type of processing that is required was passed to the processing
phases via variables.

The combination of variables involved in coordinating the processing
creates a learning curve and overly complicates recipe extensions.

With minor tweaks to the kern-tools, adding flexibility and keying
off the existence of the meta branch it is possible to remove all
of the variables that were added to support different repository
types.

(From OE-Core rev: 06e5f45c8f38925cd5902a3a3f436f5e9451dd16)

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

linux-yocto/3.2: configuration and pch merge

Updating the 3.2 SRCREVs to import the following meta/config
changes:

   6b3d4e0 meta: add mei feature
   519abac meta: add usb/uhci-hcd feature
   a67c5a3 meta/crownbay: use usb features
   0855066 meta: add usb/ohci-hcd feature
   15f1a99 meta: add usb/ehci-hcd feature
   8fa6408 meta: add usb/xhci-hcd feature
   c724a55 meta: add usb/base feature
   b55b3a1 sys940x: Cleanup sys940x.scc
   93f2e97 sys940x: Use PHYSICAL_START of 0x200000 to boot
   aaa034b sys940x: Add common standard and preempt-rt features
   e2b1286 sys940x: Add efi-ext to standard and preempt-rt configs
   d188c21 sys940x: Move emgd-1.10 data to the standard scc file
   72d9369 fri2: Cleanup fri2-$KTYPE.scc files re efi-ext.scc
   dbcb120 fri2: Use emgd-1.10 feature and branch

And the following driver fix:

   f39a0a9 pch_gbe: Do not abort probe on bad MAC

(From OE-Core rev: 612047a20ca400b932ccc634ddb759460d69fdce)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
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>

pango: Fix modules load failure in multilib environment

Multi-libs of Pango need different modules, thus different config files and
utils. This patch separate config file and utils with different MLPREFIX to
avoid conflict.

[YOCTO #2356] got fixed.

(From OE-Core rev: 80c431989eef64d2c34a9cddb7fb95b7b9768706)

Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

openssl: fix incorrect INC_PR

Restore INC_PR to r15 to prevent breakage with out of tree openssl
recipes (e.g, meta-oe).

(From OE-Core rev: 370b186b7c39897b868a5e3798a11a285277f145)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xcb-util-image: Add missing dependency on xcb-util

(From OE-Core rev: 616552506708d6dbb35a716be5bdff345edf2be0)

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

sed: Explicitly disable acl for deterministic builds

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

libgdbus: delete from oe-core

* as suggested here:
  http://lists.linuxtogo.org/pipermail/openembedded-core/2012-May/022071.html

(From OE-Core rev: 7da4a24fc5a781602aff20d25f4b9162e60810fc)

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

libsoup: delete from oe-core

* as suggested here:
  http://lists.linuxtogo.org/pipermail/openembedded-core/2012-May/022060.html

(From OE-Core rev: 95849df48d1f3802a593d57d7917bc420cbf811a)

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

xeyes/pong-clock: Update recipes

(From OE-Core rev: a1bdb86c6fcb925cef10e613dd553c766e359188)

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

meta-demoapps: Remove except for xeyes and pong-clock

http://lists.linuxtogo.org/pipermail/openembedded-core/2012-May/022059.html

As proposed by Martin Jansa with a couple of changes from me.

(From OE-Core rev: 204ee7391c19f9b2fa7f8b2ef46d04c7ccf5bd42)

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

evolution-data-server: delete old 2.20.0 version

* evolution-data-server_git.bb provides 2.30.2 from 2010-06-20 and is default for a while

(From OE-Core rev: 3138acfe6bcd86e41900a2d5e90851e30ea44bec)

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

evolution-data-server: delete unused patch parallelmake.patch

(From OE-Core rev: 557d81e4c323fb0f919b5ea466555b13d95b7584)

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

dbus-glib: bump PR to rebuild after glib upgrade

* some apps had issues in runtime when dbus-glib wasn't rebuilt and
  reinstalled after glib upgrade, see:
  http://www.shr-project.org/trac/ticket/1980

(From OE-Core rev: af04af955b1443861a9f7e5fbf87a93546f9acd9)

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

distro_tracking: updates for upgraded recipes

libpng, openssl, openssh, and opensp

(From OE-Core rev: d9e67aade8dbfd6ebd6836c703f65a043510043b)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

opensp: upgrade to 1.5.2

Removed all patches, they've been integrated upstream. :)

Added --disable-doc-build to prevent creation of docs, which
otherwise fails with the following configure error:

could not find xmlto; set XMLTO or consider --disable-doc-build

Removed configure_prepend step that was deleting the m4/ directory,
since some macros needed for the build are defined there, and I've
not encountered any problems with keeping them there in my testing.

(From OE-Core rev: cbd055917b04bf8fc640eee289e443230de139a2)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

openssh: upgrade to 6.0p1

LICENSE checksum changed due to a trivial difference in the credits
list.

(From OE-Core rev: ce1fc60b0044bccf6cf8432700d6c0b6f042dae2)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qt4: move functions from python to shell style

In qt4's do_configure operation, it will refer to some variables that
are derived from 'd', however these variable values may be not correct
in multilib case since the extraction of these variables happens before
the multilib handler.

The fix is to move these python style functions back to shell style.

This fixes [YOCTO #2355]

[RP: Fix whitepace]
(From OE-Core rev: 977ba301a1063a84b865ddf7367c35827fbffc86)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glib-2.0: Fix issues with libelf dependency creeping in

(From OE-Core rev: 2929e7d590862d9649458c90804e79a1dce40423)

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

connman: Disable tist plugin on powerpc

This plugin doesn't build on powerpc due to powerpc's terminal ioctl defintions
being incompatible with assumptions being made by this module. Until someone has
need and can test this on powerpc, disabling is the safest option.

(From OE-Core rev: 672846b0262add037daa2c9e434fe09388b58da8)

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

image.bbclass,kernel.bbclass: Use kmod-native instead of module-init-tools-cross

(From OE-Core rev: 74fcf61224dc3e12e753c5de793591677275dffd)

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

module-init-tools: Delete

Since its provided by kmod

(From OE-Core rev: 27bc204010b9dccb30847d7b3216f8b1e106f64e)

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

kmod: disable man pages generation because we don't have working xsltproc

(From OE-Core rev: 7d782a6ef77072d7f0fe50d8208c24c9e9ea9507)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kmod: Add recipes

kmod is replacement for module-init-tools

(From OE-Core rev: 716606f5446534e48b45fcc017e8bbdfe7f15e26)

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

libsoup: Fix for glib-2.0 2.32.2

(From OE-Core rev: d3efcf15cee6734b4bf2277346ceba9a991fbd0f)

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

libgdbus: Fix for glib-2.0 2.32.2

(From OE-Core rev: 0e33eade3fec1d9df052332d37d266405467a354)

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

evolution-data-server: enable deprecated glib API

* we have very old version and deprecated g_atexit is used even in current master
  http://git.gnome.org/browse/evolution-data-server/log/libedataserver/e-categories.c

(From OE-Core rev: 155ea975f5e0287f9ed49413785d1948c56601a5)

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

libsoup-2.4: upgrade to 2.38.1

* newer webkit-efl needs 2.37.4 or newer

(From OE-Core rev: 271c3afd8b72e321981b193be6deb968027e6149)

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

gobject-introspection: upgrade to 1.32.1

* simplify SRC_URI
* add COPYING to LIC_FILES_CHKSUM

(From OE-Core rev: 117cb418dedd35afce35598fd68a18721db6e9c9)

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

glib-2.0: upgrade to 2.32.2

(From OE-Core rev: 82551379ddd5440f3cdbbe2f7028e2af050e5788)

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

glib-2.0: upgrade to 2.32.1

* g_once_init_enter.patch shouldn't be needed after upstream commit
  794c1a30bc27b4c8d77537813acb1213d5ac80f2

* LIC_FILES_CHKSUM change is only because PCRE was upgraded and now
  pcre.h says "Copyright (c) 1997-2012 University of Cambridge" instead
  of 2010 and COPYING file
  "Please see the file LICENCE in the PCRE distribution for licensing
  details."  but LICENSE file is not part of glib distribution (but still
  BSD in standalone PCRE)

(From OE-Core rev: 231a3e046900f8fae2e0aa4fe5f2bee4c290b0b0)

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

site/arm-common: alignment values for guin32, guin64 and unsigned long

These are required to build recent versions of glib-2.0

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

site/powerpc-common: alignment values for guin32, guin64 and unsigned long

    These are required to build recent versions of glib-2.0

(From OE-Core rev: e9fd8e3d01091aebbc4b3584d9febc695836e4c3)

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

site/mips-common: alignment values for guin32, guin64 and unsigned long

    These are required to build recent versions of glib-2.0

(From OE-Core rev: 8fa607c1430674bee1f1f80d33f8939d7f0b1100)

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

Hob: Clear the building status if command failed

We may meet certain command failure during build time, for example,
out of memory. In this case, we need to clear the "building" status.

This fixes [YOCTO #2371]

(Bitbake rev: 283dbbbf5d34adb4c9e3aa87e3925fdebe21ff42)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake/tests/fetch: Add simple http fetcher tests

(Bitbake rev: 27b89561781f19fd95308433bec06fa8b7b354e8)

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

bitbake/fetch2/wget: Add fallback/default wget commands

(Bitbake rev: 9586808572d06de4127f6a524e87e9ad75232423)

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

bitbake/fetch2/wget: Remove unneeded function indirection and reindent

(Bitbake rev: 769b92b14a1dfbbf697b8f4bf9a5a828807fd885)

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

bitbake/fetch2/wget: Remove unnecessary use of OVERRIDES

This allows some simplifications of the function and is slightly faster.

(Bitbake rev: 5999dc9985ad087c036611bfaa59b090a08781a6)

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

bitbake/fetch2: When unpacking, only use PATH variable if its set

(Bitbake rev: ffec38675c0d78ee9fcd1d8f5a746d162145554e)

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

bitbake/fetch2: Remove WORKDIR reference, should use passed parameter

(Bitbake rev: 9384dfaf644365c731be26572443287ce8f190b2)

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

yocto-bsp: clarify help with reference to meta-intel

The current yocto-bsp help assumes knowledge that the meta-intel layer
needs to be cloned before it's put into the BBLAYERS.  Avoid the
guesswork and state the details explicitly in the help.

Also, the shorter 'usage' string doesn't mention it at all; it would
help to at minimum mention it and refer the user to the detailed help.

Fixes [YOCTO #2330].

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

yocto-kernel: use BUILDDIR to find bblayers.conf

The current code assumes that builddir == srcdir/build, which it
obviously isn't sometimes.  Use BUILDDIR to get the actual builddir
being used.

Fixes [YOCTO #2219].

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

initrdscripts: Update install.sh to work with mmc devices

Fixes [YOCTO #2385]

The installer only searches for hd[ab] sd[ab]. Some newer BSPs have mmcblk
devices that should be used as the install target. These devices also have a
partition prefix (mmcblk0p1 instead of mmcblk01). As they are detected
asynchronously, it is necessary to add the rootwait kernel parameter to avoid
a race condition trying to mount the root device.

As BSPs like the FRI2 and the sys940x have mmc devices and will have a 1.2
release, we should push this to 1.2.1. The changes are perfectly contained and
easily verified.

Test for an mmcblk device and add the p partition prefix if necessary. Add the
rootwait kernel parameter when an mmcblk device is detected.  Replace the series
of explicit umount commands with a single umount using a wildcard. This will
find all the partitions and will not try to unmount non-existant devices. Avoid
copy and paste errors by replacing /dev/${device}${pX} references with the
previously assigned rootfs, bootfs, and swap variables.

These changes have been tested on the FRI2 Sato image which installed to
/dev/mmcblk0 as well as the N450 Sato image which installed to /dev/sda. Both
were successful.

(From OE-Core rev: bf403680d72e360c7382f540ea25cfdcbe77b4e5)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
CC: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu: be sh neutral

Now runs with dash and busybox' ash as well as with bash

(From OE-Core rev: 2b93ed90736ed08c25561e24343a5ef2b8f7dbef)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu: minor tweaks

(From OE-Core rev: cda565317eefbac1b7fb268d3d8720ebae8057fa)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu: add and use error()

(From OE-Core rev: d77186606efdbb03fd92e7ee9e9ee2f9be601ba5)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu: simplify process_filename()

(From OE-Core rev: 042efbe653b699bd33175117e1363d87e4602e4f)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu: use modern, single-char name of test(1)

I do not have "[[", just "[". Be gentle to users of legacy-free setups,
also by using '=' instead of the double notation.

(From OE-Core rev: e96ba42a977f4c07aa196ce379ecd73e4ddc23c5)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu: Use OE_TMPDIR

The error message erroneously talked about TMPDIR.
Just use OE_TMPDIR everywhere to make the name of the variable obvious.

(From OE-Core rev: 7b633d0a4cf9aa05e6243974bab2b780c246f8ba)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc-4.6: disable cloog, change syntax to disable ppl

* Backported from 523df15bcc21e1749771cca757613f3520243b8e
  as suggested by Khem.

* Original commit message:

  If build system has those libraries installed
  gcc configure will pick them up. We want
  consistent builds so we disable them since we
  do not (yet) support them

(From OE-Core rev: 9de1de7c2c4e9f5accb4dd082fe819a711defbfb)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

uclibc: Return correct endianness for mips64

Regular expression did not check for mips64

(From OE-Core rev: 945b817d381bdb68202982536daaaaff78ad5761)

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

terminal.bbclass, oe-buildenv-internal: pass SCREENDIR environment variable

Some versions of the screen utility provided from the host OS vendor
write the socket directory to $HOME/.screen.  When using a shared home
directory across many servers, one sets the SCREENDIR environment
variable to avoid collisions in the shared home directory.  This
results in problems launching a devshell where it is not entirely
obvious what happened because the SCREENDIR environment variable
got stripped from the environment prior to setting up the screen
in detached mode.

Example:
   % bitbake -c devshell busybox
   # ...Please connect in another terminal with "screen -r devshell"

   % screen -r devshell
   There is no screen to be resumed matching devshell.

The temporary work around was to do something like:
   sh -c "unset SCREENDIR; screen -r devshell"

This patch adds SCREENDIR to the white list to ensure screen
works properly on systems where a developer needs to use
the SCREENDIR with shared home directories.

(From OE-Core rev: 5568a8f5a1c65bae021b2e36d735d3153acc6d72)

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

conf/machine: Clean up configuration values.

This cleans up and/or corrects a few values from machine includes
for consistency with future toolchain sanity checks, and also adds
the TUNEVALID and TUNECONFLICTS to documentation.conf.

(From OE-Core rev: 6ffe53c721a80cf156b44f59b564f2e899c6af50)

Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: Fixup bitbake-selftest

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

bitbake: Add start of bitbake regression/self testing

This adds some basic unit testing for the codeparser and data store code. Many of
the actual test cases were taken from work by Chris Larson's OE-Signatures work but with
changes to adapt to the current bitbake APIs we need to test.

I also imported CoW tests written by Holger Freyther from the original bitbake-test
codebase: http://svn.berlios.de/wsvn/bitbake/trunk/bitbake-tests/tests/ and
some tests from the doctests that were removed in commit:
http://git.openembedded.org/bitbake/commit?id=3a11c2807972bbbddffde2fa67fc380d159da467

(Bitbake rev: ae4a95780e3e08cf73c854efa8cd93379e00c4e5)

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

connman: enable TI shared transport plugin when wifi or bluetooth are requested

This plugin manages the firmware loading for the TI wl12xx series of bluetooth/wifi/fm chips.

The double --enable-tist when both bluetooth and wireless are enabled is ugly, but harmless

The background on the plugin: https://bugs.meego.com/show_bug.cgi?id=8418

(From OE-Core rev: 22254a32f05e9212411216bf1a4e48fd86b9c3ef)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: port improvements to simple.script from Debian (udhcpc)

* Support resolvconf
* Avoid bashism $((metric++))
* Use 'domain' instead of 'search' for domain parameter

(From OE-Core rev: 3d4ccf98b83fe662f375cd5b029fdd602824c0e8)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: remove unused patches for simple.script (udhcpc)

* OE-Core provides its own copy of simple.script.

(From OE-Core rev: 3a36efc17c8504b00a7a277a67a7b10caa4d1cdc)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc-4.6: explicitly disable ppl

* Fixes the following error after a system library upgrade
| .../mipsel-oe-linux/4.6.4/cc1: error while loading shared libraries: libppl.so.10: cannot open shared object file: No such file or directory

(From OE-Core rev: 2c0fb8265d4876db537850f4b20c898dde3a0f4c)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

buildhistory: fix multiple commit of images and packages at the same time

The echo line here was merging multiple lines into one, and the result
was that if both image and package changes had to be comitted then only
the image changes were being committed and the package changes could
potentially be merged into the next package change. Quoting the variable
reference fixes this.

Fixes [YOCTO #2411]

(From OE-Core rev: 2086bb86885951d0a74342e30ff205c24cb93f0d)

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

bitbake/utils: Convert vercmp_string() to use vercmp internally

Having two different version comparision algorithms in bitbake has never seemed
like a sensible idea. Worryingly, they also return different results to each other.

The vercmp_string API is relatively unused with no users in OE-Core or BitBake
itself for example. This patch converts it to use vercmp internalls, bringing
consitency to the comparisions which is easy now we have other recently added
functions. Yes, this changes behaviour but in this case I'd prefer we were
consistent than having two different comparisions.

(Bitbake rev: a569c816e016447d60624c59a750709d59a0f455)

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

bitbake/fetch2: Fix spelling error in network access error message

(Bitbake rev: b766630ed7e099fffe817928a2811272677a1b26)

Signed-off-by: Jeff Polk <jeff.polk@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

fetch2/git: Allow file:// URLs without checking for network access

Don't check for network access before grabbing the the current head,
cloning, or updating a clone when the protocol is 'file'.

(Bitbake rev: d5847bc5254b9d2f28a6b574f6157d1286add27c)

Signed-off-by: Jeff Polk <jeff.polk@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc-package-target: add libexec for plugin

Fixes:

ERROR: For recipe gcc, the following files/directories were installed but not shipped in any package:
ERROR:   /usr/libexec/gcc/i586-poky-linux/4.7.1/plugin/gengtype
(From OE-Core rev: 0be64cda0a20ca3aab75ee0cc360236d2dc40c95)

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

kernel.bbclass: move kernel-vmlinux up in PACKAGES

If KERNEL_IMAGETYPE is vmlinux, the expectation is most likely that there will
be no kernel image package, but we still want a vmlinux package for debugging,
so move kernel-vmlinux in front of kernel-image in PACKAGES.

(From OE-Core rev: b62a44a9bf7b268296aeed64469390aa7efc2e73)

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

runqemu: Fix TAP='TUNSETGROUP: Invalid argument' by falling back to tunctl -u

By default the runqemu script tries to set the group permissions on any
tap device it creates.  The TUNSETGROUP ioctl is not implemented on some
popular host enterprise linux distributions.

Internally the script will exit as follows:

++ /opt/qemux86/bitbake_build/tmp/sysroots/x86_64-linux/usr/bin/tunctl -b -g 100
+ TAP='TUNSETGROUP: Invalid argument'
+ STATUS=1
+ '[' 1 -ne 0 ']'
+ echo 'tunctl failed:'
tunctl failed:
+ echo TUNSETGROUP: Invalid argument

This patch implements a fallback to using the userid as the owner of
the tap device which is supported by all 2.6 kernels, the default remains
to try and use the groupid first.

(From OE-Core rev: 3af2bc59776fb738bd795160512a2f3f49ce6d32)

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

qt-4.8.1: fix creating /include directory in build host root

(From OE-Core rev: 7d3fd68480d54a8ef0d91a772df3506cd45ef2c2)

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

qt4: fix compilation issue

gcc 4.7 uncovers a type mismatch in the code. And this commit fixes the issue:

| /srv/home/nitin/builds/build-gcc47/tmp/sysroots/qemux86/usr/include/glib-2.0/glib/gthread.h:58:7: error: 'union' tag used in naming 'struct _GMutex' [-Werror=permissive]
| In file included from ./wtf/Platform.h:1217:0,
|                  from ./config.h:30,
|                  from wtf/gobject/GOwnPtr.cpp:19:
| wtf/gobject/GTypedefs.h:55:16: note: 'struct _GMutex' was previously declared here
NOTE: package qt4-x11-free-4.8.1-r41.1: task do_compile: Failed

(From OE-Core rev: 1923455340fca5e4b68a2c3ac3566a99baa93cd9)

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

site/ix86-common, site/x86_64-linux: alignment values for guin32, guin64 and unsigned long

These are required to build recent versions of glib-2.0

(From OE-Core rev: bb996003547f9b93b734ddac11407b4eb42ecfac)

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

gtk-engines: fix build with glib-2.32

(From OE-Core rev: 93a4f4a8e7c298c5a19e757ae797328369da104e)

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

nfs-utils: don't try to unload nfsd module

* When nfsserver get's stopped, "modprobe -r nfsd" may
  hang indefinitely. As there's no need to unload the
  module, just remove the call to modprobe.

(From OE-Core rev: aba81b64679ec0b1809946cb480e261af47f7625)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

pulseaudio: disable tcpwrap by default

This ensures that tcpwrapper usage is always disabled, this was
inconsistent because it would test for libwrap and sometimes enable
and sometimes not.

This ensures consistent build reproducibility.

(From OE-Core rev: afa0efc3473fb5ba88d95594e382da2e03145c3b)

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