perf.bb: fix multilib build

Fixed:
$ bitbake perf lib32-perf
ERROR: QA Issue: lib32-perf: Files/directories were installed but not shipped
  /usr/lib64
  /usr/lib64/traceevent
  /usr/lib64/traceevent/plugins
  /usr/lib64/traceevent/plugins/plugin_function.so
  /usr/lib64/traceevent/plugins/plugin_scsi.so
  /usr/lib64/traceevent/plugins/plugin_hrtimer.so
  /usr/lib64/traceevent/plugins/plugin_kmem.so
  /usr/lib64/traceevent/plugins/plugin_jbd2.so
  /usr/lib64/traceevent/plugins/plugin_xen.so
  /usr/lib64/traceevent/plugins/plugin_mac80211.so
  /usr/lib64/traceevent/plugins/plugin_cfg80211.so
  /usr/lib64/traceevent/plugins/plugin_sched_switch.so
  /usr/lib64/traceevent/plugins/plugin_kvm.so [installed-vs-shipped]

The perf.do_configure edits kernel's
${STAGING_KERNEL_DIR}/tools/perf/config/Makefile, there would be
problems since kernel doesn't have multilib, and the build result is
undetermined.

Previously, the sed command changed libdir to /usr/lib64 (or 32) in the
Makefile, so the build result was different if we build perf (64) first
or lib32-perf first.

Use the weak assignment "libdir ?=" to instead of "libdir =" will fix
the problem since the multilib builds are in different processes, and
they won't affect each other any more.

The sed command will match both $(prefix)/$(lib) and $(prefix)/lib since
the Makefile may has been modified before this patch.

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

pango / ptest: clean CFLAGS for host binary

The binary gen-all-unicode needs to be compiled for the host
architecture, the CFLAGS passed to target system could cause
build issues for it.

Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/create-recipe: minor fixes

Fixed:
- Remove PR. We don't need the "PR = r0" in the bb file.
- Function length can't get the array's size correctly.
  Should use function scalar.

Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package_manager: Add rpm v4 support

Currently the package manager code only supports rpm v5. To be useful outside
of OE or with OE layers using v4, it makes sense to add in rpm v4 support. This
takes a patch from "Bartosh, Eduard" <eduard.bartosh@intel.com> and enhances
it to also include versions of the workarounds from poky-eurogiciel to
allow rpm v4 usage with the class for image construction.

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

package_rpm: Add builddir macro to define source directory

This allows its usage in other RPM macros so files in ${S} can be found.

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

package_rpm: Add custom extension support for spec generation.

Add hooks to allow customisation of the rpm spec files. Since python functions
aren't visible in the data store, one variable is used to trigger the call to
a separately named function. A dummy function is not provided since this then
triggers various class ordering complexities which are best avoided.

Ultimately this will be replaced by a refactor of the code to generate the
spec file using a python class.

This allows the tizen layer to add hooks for the security manifests for
example.

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

lz4: update to version 122

Makefile md5sum changed its is where they bump the version.

r122 fixes:
Fix : AIX & AIX64 support (SamG)
Fix : mips 64-bits support (lew van)
Added : Examples directory, using code examples from Takayuki Matsuoka
Updated : Framing specification, to v1.4.1
Updated : xxHash, to r36

r121 fixes:
Fix : make install for OS-X and BSD, thanks to Takayuki Matsuoka
Added : make install for kFreeBSD and Hurd (Nobuhiro Iwamatsu)
Fix : LZ4 HC streaming bug

Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python-numpy: Fix build for mips64

Add missing mips64 specific config files

Signed-off-by: Khem Raj <raj.khem@gmail.com>

grub: Replace _BSD_SOURCE macro with _DEFAULT_SOURCE

This is needed with glibc 2.20 since _BSD_SOURCE
macro is gone

Signed-off-by: Khem Raj <raj.khem@gmail.com>

xf86-video-intel/xf86-video-omapfb: Fix build with glibc 2.20

include xorg-xserver.h

Signed-off-by: Khem Raj <raj.khem@gmail.com>

xf86-input-synaptics/xf86-input-vmmouse: Fix build with glibc 2.20

Backport the upstream patches needed for them to work with glibc 2.20

Signed-off-by: Khem Raj <raj.khem@gmail.com>

bitbake.conf: Use 2.6.32 for oldest supported kernel

glibc 2.20+ wont support any older than that

Signed-off-by: Khem Raj <raj.khem@gmail.com>

recipes: Remove references to eglibc

change use of eglibc related variabled to glibc equivalents

Signed-off-by: Khem Raj <raj.khem@gmail.com>

oeqa: sstatetests should now look for glibc-initial

Signed-off-by: Khem Raj <raj.khem@gmail.com>

image-swab.bbclass: Account for eglibc -> glibc move

Signed-off-by: Khem Raj <raj.khem@gmail.com>

conf: Account for eglibc->glibc move

Change references in documentation appropriately

Signed-off-by: Khem Raj <raj.khem@gmail.com>

distro: TCLIBC now defines glibc instead of eglibc

Adjust naming conventions to reflect eglibc->glibc move

Signed-off-by: Khem Raj <raj.khem@gmail.com>

glibc: Migrate eglibc 2.19 -> glibc 2.20

- This is a big swoop change where we switch to using glibc
- option-groups are forward ported
- cross-localedef is extracted out from eglibc and hosted
  at github.com/kraj/localedef, its used for cross-localedef
  recipe
- Other non ported patches from eglibc are forward ported
  ppc8xx cache line workaround
  SH fpcr values
  dynamic resolver
  installing PIC archives is there but is not applied
  libc header bootstrap

- Delete eglibc recipes we moved back to using glibc now
- Fix ppc/e500 build

- Fix crypt module build when options are used
- Fix fnmatch build when options OPTION_EGLIBC_LOCALE_CODE is unset
  HAVE_MBSTATE_T and HAVE_MBSRTOWCS should be defined conditionally
  based upon OPTION_EGLIBC_LOCALE_CODE being set/unset
- Move the ports/ patches to relevant files now that ports is gone

Signed-off-by: Khem Raj <raj.khem@gmail.com>

perl: add explicit configs for Time-HiRes

The compilation of Time-HiRes tries to compile and run the testing
program but definitely fails since we are cross compiling:
Looking for clock_gettime()... ./tmp26733: ./tmp26733: cannot execute binary file
Looking for clock_getres()... ./tmp26733: ./tmp26733: cannot execute binary file

so add explicit configs to avoid the auto detecting.

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

ghostscript: Don't build-depend on libgcrypt for the local cups

Backported for the ghostscript supplied cups code from
http://www.cups.org/strfiles.php/3308/cups-no-gcrypt.patch

This addresses the cryto dependency seen during build.

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

native.bbclass: deltask package_qa

The native recipe doesn't need package_qa, and it has a race issue with
do_unpack which requires the clean dir of ${S}, then causes error:

NOTE: recipe opkg-utils-native-0.1.8+gitAUTOINC+eae0d8fa44-r0: task do_package_qa: Failed
ERROR: Task 283 (virtual:native:/path/to/opkg-utils_git.bb, do_package_qa) failed with exit code '1'

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

linux-yocto: ensure that recipe specific defaults take precedence

In order to make KBRANCH assignments more flexible and easy to override the
default in each recipe as changed to a conditional assignment. But the
common include file, sets its own branch default to master.

This results in the recipes not overriding KBRANCH and master always being
used.

By putting the include after the default assignment, we now get the proper
default branch.

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

kernel-yocto: replace --is-ancestor with basic git porcelain commands

--is-ancestor is a relatively new git option [commit 5907cda1, Aug 30 2012].
To support build machines with older versions of git installed, we can use
the basic porcelain commands to acheive the same check.

merge-base: "--is-ancestor A B" can be replaced with:

    if test "$(git rev-parse --verify A)" = "$(git merge-base A B)"

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

linux-yocto/3.4: remove 3.4 name recipes

Removing the 3.4 recipes, since support has not shifted to 3.10 and
3.14 for LTSI kernel versions, with 3.16+ as the development version.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

linux-yocto/3.14: update to v3.14.17

Updating the 3.14 repository to the latest korg stable udpate.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

linux-yocto/3.14: vhost, vxland, openvswitch and block/bfq updates

Updating the linux-yocto/3.14 SRCREVs to integrate more feature additions
and backports to the LTSI kernel base:

   42477caf6bfd block, bfq: add Early Queue Merge (EQM) to BFQ-v7r5 for 3.14.0
   349aa3f0848d block: introduce the BFQ-v7r5 I/O sched for 3.14
   121ed2738c47 block: cgroups, kconfig, build bits for BFQ-v7r5-3.14
   9925795a3e54 vhost-scsi: Include prot_bytes into expected data transfer length
   68fe340cb836 vhost: move memory pointer to VQs
   72fa27bcf526 vhost: move acked_features to VQs
   757680234478 vhost: replace rcu with mutex
   050c1440ac73 vhost-net: extend device allocation to vmalloc
   d68641313156 vhost/scsi: Enable T10 PI IOV -> SGL memory mapping
   fb89a8df7f1c vhost/scsi: Add T10 PI IOV -> SGL memory mapping logic
   5caf8475ee8a vhost/scsi: Add preallocation of protection SGLs
   c0785b9490a8 vhost/scsi: Move sanity check into vhost_scsi_map_iov_to_sgl
   bd947327cdf6 vhost: don't open-code sockfd_put()
   83fab2df94c6 openvswitch: Use exact lookup for flow_get and flow_del.
   9c185c40f7a8 openvswitch: Fix tracking of flags seen in TCP flows.
   0d2455332847 openvswitch: supply a dummy err_handler of gre_cisco_protocol to prevent kernel crash
   4513a2fd2703 openvswitch: Fix a double free bug for the sample action
   1acf10deff18 openvswitch: Simplify genetlink code.
   9ea6a4dea3f5 openvswitch: Minimize ovs_flow_cmd_new|set critical sections.
   9e76764432cf openvswitch: Split ovs_flow_cmd_new_or_set().
   7bafcd59a838 openvswitch: Minimize ovs_flow_cmd_del critical section.
   67980f929444 openvswitch: Reduce locking requirements.
   347f9442b559 openvswitch: Fix ovs_flow_stats_get/clear RCU dereference.
   c7aa7c522ec0 openvswitch: Fix typo.
   b244c7b19a55 openvswitch: Minimize dp and vport critical sections.
   21883a3ffdb6 openvswitch: Make flow mask removal symmetric.
   ffa173197cf5 openvswitch: Build flow cmd netlink reply only if needed.
   f3e8e5c07505 openvswitch: Clarify locking.
   df42a8f505c2 openvswitch: Avoid assigning a NULL pointer to flow actions.
   91b07542da22 openvswitch: Compact sw_flow_key.
   24bb1a576332 net/openvswitch: Use with RCU_INIT_POINTER(x, NULL) in vport-gre.c
   808868318939 openvswitch: Use TCP flags in the flow key for stats.
   a57851bae1cf openvswitch: Fix output of SCTP mask.
   8b70125106ee openvswitch: Per NUMA node flow stats.
   606497442f52 openvswitch: Remove 5-tuple optimization.
   01e74b175909 openvswitch: Use ether_addr_copy
   b5d02cfdc985 openvswitch: flow_netlink: Use pr_fmt to OVS_NLERR output
   4da9e8d176f9 openvswitch: Use net_ratelimit in OVS_NLERR
   ee8f673dd798 openvswitch: Added (unsigned long long) cast in printf
   b67f35f8b0aa openvswitch: avoid cast-qual warning in vport_priv
   3e01428f6e0c openvswitch: avoid warnings in vport_from_priv
   069ee359a487 openvswitch: use const in some local vars and casts
   cbec86356a89 vxlan: add x-netns support
   a51970560923 vxlan: ensure to advertise the right fdb remote
   696068dca072 vxlan: remove unused port variable in vxlan_udp_encap_recv()
   e94003f3b58d sched/deadline: Fix sched_yield() behavior

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kern-tools: allow meta branch and meta data directory to differ

From the kern-tools commit:

    tools: allow meta directories that are not the same as the branch name

    With this change it is now possible to have a meta branch with meta data
    in a directory that is not the same name as the branch.

    The changes to three parts of the build are required to discover the name
    of the meta directory by relying on the fact that in a clean/proper build
    the meta directory is the only untracked, top level directory in the build.

    As such, we can restore a checkpoint and then examine the build directory
    to determine the meta directory name .. avoiding any new variables to
    indicate this to the scripts and build system.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: convert echo statements to bb* equivalents

Use the bbinfo, bberror, bbfatal equivalents to the existing echo statements
within the kernel-yocto processing. This makes us consistent with the other
messages from the build system.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: remove KBRANCH_DEFAULT

KBRANCH_DEFAULT is no longer used, so we can remove it from all
recipes (and it won't be missed).

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: allow custom non-meta, SRCREV format builds

When custom respositories are built (like a pure kernel.org
repo), the machine_meta SRCREV format is not applicable. As
such, we shouldn't check for the meta branch and we shouldn't
only check SRCREV_machine based revisions.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: clean overly complex branch checkout

Since the git fetcher ensures that branches exist, we no longer need to
validate the branch and have a conditional checkout of the source.

We can remove some checks and ensure that whenever we exit the
do_kernel_checkout routine that a branch is always checked out.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: use show-ref instead of branch -a

It's better to check a branches existence via show-ref versus the end
user branch commands. So we make the switch.

Also as part of this change, we move the conversion of remote branches
to local branches above the meta branch checking. This is required to
ensure that the branch is local for the show-ref check.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: simplify branch SRCREV validation

The checking of machine and meta branch SRCREVs was inconsistent and
didn't allow a mixed AUTOREV machine/meta branch combination. By
simplifying the checks and changing the logic, we can now allow this
combination.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: remove KBRANCH_DEFAULT

KBRANCH_DEFAULT was introduced as a way to trigger the enforced build
of a particular branch of the tree. With the fetcher now enforcing
SRCREVs existing on a branch, we can simply validate that the SRCREV
is reachable from the final branch and no longer care about enforcing
a given branch.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: remove branch existence checking in do_validate_branches

Now that the fetcher will enforce branch existence, we no longer need to
confirm that a branch exists, and that it was the branch requested to
be built.

We know the branch exists and we'll confirm that the specified SRCREV
is going to be built after we've patched the tree.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: remove SRC_URI kbranch validation

We no longer need to check if the KBRANCH matches the branch specified
in the SRC_URI. This is taken care of by the fetcher at the beginning
and SRCREV ancestor validation after patching.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: remove containing branch check

The bitbake fetcher now enforces that a commit is contained by a branch,
so this code can be dropped from do_validate_branches.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: move SRCREV validation to patching phase

Rather than attempting to condition the entire tree to machine SRCREV (since
we don't know what branch will be built), we can instead wait until patching
has completed and then confirm that we are indeed building a decendant of the
specified SRCREV. The result is a much simpler check, and no mangling of the
tree.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: use cat-file instead of git-show

Parsing the output of git show is error prone, since it changes based on
the type of issue with bad comit IDs. Since the output is no longer used
in the case of a valid ref, we can switch to git-cat-file and  simply
check the return code.

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

kernel-yocto: remove redundant SRCREV check

do_validate_branches checks to ensure that a valid machine SRCREV was
set. A test against an empty SRCREV is done in two separate locations,
we only need one, since the first check immediately returns and the
second check never hits.

At the same time, we can stop referring to the same commit hash by
3 different names. Instead we assign to a local variable at the
top of the routine, and refer to it at all times.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

tzcode: update to 2014g

The readme md5sum changed do you wording changes.

Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tzdata: update to 2014g

For a full description of the changes see:

http://mm.icann.org/pipermail/tz-announce/2014-August/000024.html

Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libpam: remove MLPREFIX from PN

The commit df3038768f59f7a0c814974ff674d4e59cbdfca4 changed 'libpam' to
'pn', then we don't need the "MLPREFIX + pn" any more, otherwise we
would get the name like: "lib32-lib32-libpam-x", and the warn:

WARNING: QA Issue: lib32-pam-plugin-access rdepends on
lib32-lib32-libpam-suffix, but it isn't a build dependency? [build-deps]

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

linux-libc-headers: update to 3.16

Updating the linux-libc-headers to the 3.16 release. This matches the
current -dev kernel version, and is compatible with the existing
named kernel versions (3.10, 3.14).

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

libpam: Fix multilib packaging

The plugin runtime dependencies were not including the multilib
prefix, fix it.

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

libpam: Fix runtime providers

The runtime providers were commented out. Removing the comment brings
up a issue with the native renaming which has been workarounded
disabling the runtime recommendation. This is indeed a workaround so a
FIXME comment has been added to remind about it in case we someday
move to native prefix.

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

libical: upgrade to 1.0.0

pthread-fix.patch no longer needed, included upstream;
Switched to new website, new bug tracker, new source repo.

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

libpcap: upgrade to 1.6.1

- Still BSD licensed, some changes into
pcap triggered a checksum change;
- ieee80215-arphrd.patch removed, obsolete.
- switched to ${BP} variable;

Changelog from previous version:
Saturday  Jul. 19, 2014 mcr@sandelman.ca
  Summary for 1.6.1 libpcap release
some fixes for the any device
changes for how --enable-XXX works

Wednesday Jul. 2, 2014 mcr@sandelman.ca
  Summary for 1.6.0 libpcap release
        Don't support D-Bus sniffing on OS X
        fixes for byte order issues with NFLOG captures
        Handle using cooked mode for DLT_NETLINK in activate_new().
        on platforms where you can not capture on down interfaces, do not list them
        but: do list interfaces which are down, if you can capture on them!

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

iproute2: upgrade to 3.16.0

Bugfixes.

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

at: Upgrade to 3.1.15

Rebase pam.conf.patch
Move to generic files -> at for patches directory

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

libassuan: Upgrade to 2.1.2

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

libxkbcommon: Upgrade to 0.4.3

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

man-pages: Upgrade to 3.71

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

less: Upgrade to 466

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

alsa-utils: upgrade to 1.0.28

Changelog:
http://www.alsa-project.org/main/index.php/Changes_v1.0.27.2_v1.0.28

0001-alsactl-don-t-let-systemd-unit-restore-the-volume-wh.patch updated.

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

alsa-tools: upgrade to 1.0.28

mips_has_no_io_h.patch removed, no longer necessary.
Changelog:
http://www.alsa-project.org/main/index.php/Changes_v1.0.27.2_v1.0.28

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

alsa-lib: upgrade to 1.0.28

Changelog:
http://www.alsa-project.org/main/index.php/Changes_v1.0.27.2_v1.0.28

Update-iatomic.h-functions-definitions-for-mips.patch removed,
integrated upstream.
Check-if-wordexp-function-is-supported.patch updated.

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

package_manager.py: enable smart non-interactive mode

Added --quiet option to smart command line.
Without this option smart 1.4.1 turns into interactive mode, i.e.
start asking questions and expecting answers.

Internally within smart, this changes the default UI to one which
just prints to stderr, the naming of the parameter is a little
odd but does what we need.

Signed-off-by: Ed Bartosh <eduard.bartosh@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libtirpc: upgrade to 0.2.5

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

gstreamer1.0-plugins-ugly: upgrade to 1.4.1

Bugfix release.

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

gstreamer1.0-rtsp-server: upgrade to 1.4.1

Bugfix release.

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

gstreamer1.0-plugins-bad: upgrade to 1.4.1

Bugfix release.

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

gstreamer1.0-plugins-good: upgrade to 1.4.1

Bugfix release.

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

gstreamer1.0-plugins-base: upgrade to 1.4.1

Bugfix release.

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

gstreamer1.0-libav: upgrade to 1.4.1

Bugfix release.

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

gstreamer1.0: upgrade to 1.4.1

bugfix release.

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

systemd: add PACKAGECONFIG for 'elfutils'

Add PACKAGECONFIG for 'elfutils'.
This also fixes a build-deps QA warning about libdw.

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

useradd: Add setscene dep on pseudo-native

A recipe will sometimes be rebuilt unnecessarily if it adds users via
useradd and the pseudo-native task has not been populated.

This patch adds the correct dependency.

Signed-off-by: Jate Sujjavanich <jatedev@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemu-native: depends on libxext-native when enable sdl

Fixed do_configure error:

ERROR: User requested feature sdl
       configure was not able to find it.
       Install SDL devel

The error message from log.do_configure, and if we check config.log, the
error is:

sysroots/x86_64-linux/usr/lib/libXext.so.6: undefined reference to `_XEatDataWords'

The _XEatDataWords is provided by libX11.so, the dependencies are:

qemu-native -> sdl(from the host) -> libXext.so(maybe host or native) -> libX11.so (maybe host or native)

For example on Ubuntu 12.04 x86_64, the link route is:

qemu-native's do_configure sdl checking:
/usr/lib/x86_64-linux-gnu/libSDL.so --> <sysroots>/x86_64-linux/usr/lib/libXext.so.6 --> /usr/lib/x86_64-linux-gnu/libX11.so

So it is:
host sdl -> sysroot libXext.so.6 -> host X11.so

qemu-native doesn't depend on libxext-native, so the error would happen
if the libx11-native is being rebuilding and the host's libX11.so
doesn't have _XEatDataWords.

Manually reproduce it on Ubuntu 12.04 x86_64:

$ bitbake qemu-native libx11-native libxext-native
$ bitbake qemu-native libx11-native -ccleansstate && bitbake qemu-native

Note, only qemu-native has this problem since libxext depends on libX11,
and SDL would depend on libxext when x11 is enabled.

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

package: Correct two typos in a comment

This quite coincidentally invalidates the sstate for do_package which
is needed due to the correction of oe.utils.multiprocess_exec().

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

lib/oe/utils: Make multiprocess_exec() return anything

The variable "results" was accidentally used for multiple different
things at the same time, which unintentionally discarded anything that
was supposed to be returned from the function...

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

oeqa: xorg log test is being replaced by parselogs so remove

Now we have the parselogs test, this one can be removed.

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

testimage: Add parselogs to the default tests for sato images

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

oeqa/pasrselogs: Improve the machine/string whitelist

Currently the whitelist is imcomplete, inaccurate and suffers duplication.

These changes:
 * Add common groups of errors
 * Change to make the default whitelist clear
 * Correctly (or at least better) escape the regexp expressions
 * Add in missing machines to allow builds on the autobuilder to suceed

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

oeqa/runtime: Automatic test for parsing the logs on a machine and search for certain error keywords.

This adds a common new qa test for general processing of log files. One
significant improvement is machine dependent ignore filters.

This can be used to replace several weaker individual QA tests
that are currently used.

Signed-off-by: Lucian Musat <georgex.l.musat@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/cpan_build.bbclass: add EXTRA_CPAN_BUILD_FLAGS

Some packages have additional configuration options that are currently
not accessible through cpan_build.bbclass. Similar to cpan.bbclass, add
EXTRA_CPAN_BUILD_FLAGS to pass the optional parameters to perl Build.PL.

Signed-off-by: Tim Orling <TicoTimo@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

buildtools-tarball: export GIT_SSL_CAINFO

export GIT_SSL_CAINFO so git can find the certs

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

buildtools-tarball: include nativesdk-ca-certificates

nativesdk-ca-certificates is needed in order to support oe/yocto builds with
buildtools-tarball on old hosts, as we provide libcurl, and that needs to be
able to find the certs, and there's no standard path or bundle path, so we
can't rely on the host.

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

lttng-tools: fix ptest execution failure

Ptest will hang when execute "./run.sh fast_regression" and lttng background
daemon can't be killed normally. This patch fixes this issue and use ${PTEST_PATH}
instead of long path.

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

populate_sdk_base: Remap TOOLCHAIN_HOST_TASK variable

Currently this variable isn't passed through the magic rename
mangling. This means that if you try adding "nativesdk-eglibc"
to an image directly, you can't since the package is renamed
by debian.bbclass and nothing sees the renaming.

This is annoying since I wanted to exactly that. This code
change passes it through the standard renaming function, the
tricky part is that we have to set PKGDATA_DIR to point to
the correct sysroot during the call. We create a copy of the
 datastore for the purposes of the call to do this.

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

python3-distribute: fix interpreter

It should use "/usr/bin/env python3" rather than python.

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

bootchart2: no bashism in bootchartd.in

So use /bin/sh

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

sed: add sed to RDEPENDS sed-ptest

Its test scripts require /bin/sed.

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

lsbtest: no bashism in LSB_Test.sh

So use /bin/sh

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

run-ptest: fix bashism

These script use /bin/sh as the interpreter, but contains bashism:
    recipes-devtools/insserv/files/run-ptest
    recipes-devtools/quilt/quilt/run-ptest
    recipes-devtools/tcltk/tcl/run-ptest
    recipes-extended/gawk/gawk-4.1.1/run-ptest
    recipes-support/beecrypt/beecrypt/run-ptest

Fixed:
    "==" -> "=" (should be -eq when integer)
    "&>log" -> ">log 2>&1"

And quilt's test scripts requires bash, add bash to RDEPENDS_quilt-ptest

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

valgrind/oprofile/systemd: no bashism in run-ptest

There is no bashism in the scripts, so use /bin/sh, and add /bin/sh
interpreter for systemd's run-ptest.

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

meta: fix RDEPNEDS for the test related pkgs

Add bash, python or perl to the ptest pkgs to fix the RDEPENDS issues.

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

ltp: fix RDEPENDS

Fixed:
/opt/ltp/testcases/bin/aio01 -> libaio
/opt/ltp/testcases/bin/*.sh -> bash
/opt/ltp/testcases/bin/*.py -> python
/opt/ltp/bin/*.awk -> awk

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

ethtool: upgrade to 3.15

Signed-off-by: Roxana Ciobanu <roxana.ciobanu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gstreamer1.0-plugins-bad: Add DEPENDS on jpeg

libjpeg is automatically linked to if present. This cannot be turned off.
Adding DEPENDS on jpeg ensures that a proper dependency is established.

Signed-off-by: Carlos Rafael Giani <dv@pseudoterminal.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gstreamer1.0-plugins-bad: Fix GL/GLES configuration

* Replace --disable/enable-gl with the individual EGL, GLES, GL switches
  (--disable-gl turns off GL support entirely)
* Put the default opengl packageconfig into its own variable to make it
  easier for distros and BSP layers to define what to use (GL, GLES,
  or neither)
* Add libglu as a dependency for desktop GL
* Patch configure.ac to ensure libraries are only searched and linked to
  if the corresponding API isn't disabled (this prevents cases where
  libgstgl is linked to GLU even though desktop GL is dis- and GLES
  enabled)

Signed-off-by: Carlos Rafael Giani <dv@pseudoterminal.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gstreamer1.0-plugins-bad: Add DEPENDS on libpng

libpng is automatically linked to if present. This cannot be turned off.
Adding DEPENDS on libpng ensures that a proper dependency is established.

Signed-off-by: Carlos Rafael Giani <dv@pseudoterminal.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gstreamer1.0-omx: Add DEPENDS on gstreamer1.0-plugins-bad

gst-omx links to libgstgl if it is present, and libgstgl is built by
gstreamer1.0-plugins-bad. Adding it as a DEPENDS ensures a proper
dependency is established.

Signed-off-by: Carlos Rafael Giani <dv@pseudoterminal.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gstreamer1.0-plugins-bad: add opencv haar cascade XML files to package

Signed-off-by: Carlos Rafael Giani <dv@pseudoterminal.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cpan.bbclass: matches more lines

Fixed:
- There might be a space between "#!" and "/pat/to/usr/bin/perl", e.g.:
  "#! /usr/bin/perl", now also matches it.

- There might be the lines like the following in the body:
  eval 'exec /path/to/usr/bin/perl-native/perl -S $0 ${1+"$@"}'

  Now we only check "#! */path/to/usr/bin/perl" to make sure it is a
  perl script, but match and fix the lines in both header and body.

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

perl-native: fixed bad interpreter error

We may get the error on some hosts when build in deeper dir:
/bin/sh: /path/to/tmp/sysroots/i686-linux/usr/bin/perl-native/pod2man: /path/to/tmp/sysroots/i686-li:
bad interpreter: No such file or directory

Note the "i686-li", it should be "i686-linux", but is truncated by the
host.

We can use "/usr/bin/env nativeperl" as we have done in cpan.bbclass for
other recipe's perl script to fix the problem.

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

packagegroup-core-directfb: Set PACKAGE_ARCH to MACHINE_ARCH

The contents change depends on the MACHINE_FEATURES so this is clearly
a machine specific package. Set it accordingly.

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

ofono: upgrade to 1.15

- Bug fixes;
Add support for Handsfree subscriber number feature;
Add support for Handsfree multiple DTMF characters;
Add support for PAP authentication.
- Added patch to revert tests to work with Python2.x.

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

bluez5: upgrade to 5.22

Several fixes to HID over GATT (HoG) and for AVRCP.
Notable additions with the Linux kernel 3.17:
- BR/EDR whitelist support;
- Proper LE passive scanning support.

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

harfbuzz: upgrade to 0.9.35

Bug fixes and minor changes/cleanups.

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