.gitignore: only ignore meta-* at top level

meta-tizen contains a directory called meta-common which should
not be ignored.

.gitignore: do not ignore the new components

bitbake and oe-core paths are not ignored either. Neither should
be the components that we are adding.

tizen-distro: be more careful when importing patches

The combo-layer hooks for bitbake and meta-tizen patch the Subject
line of patches. This change must be limited to the Subject of
the main patch itself, in particular patches embedded in the patch
must not be changed.

Achieved by limiting the replacement to the lines until the first
Subject in the patch.

Found when importing meta-tizen and diffing against a direct copy
of the final revision.

tizen-distro: improve layerhook for meta-tizen

Not all commits in meta-tizen have a Signed-of-by tag. If missing,
just append the rev number at the end of the commit message.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>

tizen-distro: configure combo-layer

Up to this point, the tizen-distro repository is identical to the poky
master branch minus the "yocto" and "documentation" parts. It was created
by checking out 081fddd3e464935e5f438a7686eb8f8856da6281 (the commit where
"dizzy" was branched off from master) and stripping out these undesired files
with: git filter-branch --prune-empty --tree-filter 'rm -rf *yocto* documentation' HEAD

Then the last_revision properties for bitbake and openembedded-core were set
based on the commit messages generated in poky by the combo-layer commit hook.
Because these extra comments are useful, the same commit hooks are also
used in tizen-distro.

Now the repository is ready for importing the other components.

combo-layer: let user choose where properties get updated

When updating a property (at the moment, only 'last_revision'),
first check whether the user already added it to the main config.
If so, update there even if there is a local config.

This way, 'last_revision' can be shared between developers
as part of the repository while still configuring per-developer
repo paths outside of the git repository in a local config.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>

build-appliance-image: Update to dizzy head revision

(From OE-Core rev: 97756472d3a69eaca95d105494ffea78c6b077e0)

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

gst-ffmpeg: Security Advisory - ffmpeg - CVE-2013-0869

The field_end function in libavcodec/h264.c in FFmpeg before 1.1.2
allows remote attackers to have an unspecified impact via crafted H.264
data, related to an SPS and slice mismatch and an out-of-bounds array
access.

http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2013-0869

(From OE-Core rev: 464123172c92d92fa08e2125c04653590ad654c7)

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

gst-ffmpeg: Security Advisory - ffmpeg - CVE-2013-4358

libavcodec/h264.c in FFmpeg before 0.11.4 allows remote attackers to
cause a denial of service (crash) via vectors related to alternating bit
depths in H.264 data.

http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2013-4358

(From OE-Core rev: 6a029bee3da097b9ed8c421fcd5ea61cd1a31b34)

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

mtools: fix broken /usr/bin/lz

When build fs with mtools-3.9.9, has file /usr/bin/lz in rootfs,
it is the symlink to uz:

root@qemu3:~# /usr/bin/lz
-sh: /usr/bin/lz: No such file or directory
$root@qemu3:~# ls -l /usr/bin/lz
lrwxrwxrwx 1 root root 2 Jul 18 18:07 /usr/bin/lz -> uz
root@qemu3:~# uz
-sh: uz: command not found

But the uz isn't actually exist, so the result is that lz is a
broken symlink.

The root cause is that uz hasn't been installed when install-scripts.

(From OE-Core rev: 7308da9ccd4b8b9b5077aacd0442be28a6c73c61)

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

gnupg_1.4.7: add package config libusb

If build gnupg 1.4.7 after libusb-compat, it shows warning:

WARNING: QA Issue: gnupg rdepends on libusb-compat, but it isn't a build dependency? [build-deps]

Add package config libusb to fix it.

(From OE-Core rev: e1c77392143e538a2305f63ee3add611ec66e877)

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

openssl: upgrade to 1.0.1j

(From OE-Core rev: 390916b8400a46088c71183aef6e17b947cf4b74)

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

readline: Security Advisory - readline - CVE-2014-2524

The _rl_tropen function in util.c in GNU readline before 6.3 patch 3
allows local users to create or overwrite arbitrary files via a symlink
attack on a /var/tmp/rltrace.[PID] file.

http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-2524

(From OE-Core rev: 0e95eef8817f51504dcc50d855dcbef172cfc897)

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

gnupg: CVE-2013-4242

GnuPG before 1.4.14, and Libgcrypt before 1.5.3 as used in GnuPG 2.0.x
and possibly other products, allows local users to obtain private RSA
keys via a cache side-channel attack involving the L3 cache, aka
Flush+Reload.

Patch from commit e2202ff2b704623efc6277fb5256e4e15bac5676 in
git://git.gnupg.org/libgcrypt.git

(From OE-Core rev: d1e0f3e71ce9978ff0fc94d71e67b528dad84c5c)

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

sstate.bbclass: Fix up white space lost in last commit.

Commit e9672387 split one long line into a multi-line string, but in
the process white space between words was lost. This results in badly
formatted output when this message is printed.

(From OE-Core rev: b145374c0a498de0160a9b81f50ce0066ab14862)

Signed-off-by: Peter Urbanec <openembedded-devel@urbanec.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sstate.bbclass: specify func dirs for sstate_hardcode_path

For some recipes that inhrient cmake, the ${B} may be removed by
cmake_do_configure() while sstate_hardcode_path() running, this
causes build errors:

  Exception: OSError: [Errno 2] No such file or directory: \
  '/path/to/build'

The function sstate_hardcode_path() called command:

  $SSTATE_SCAN_CMD which extended as "find ${SSTATE_BUILDDIR} ..."

So the proper function dirs could be ${SSTATE_BUILDDIR}.

(From OE-Core rev: a949943e622b08485fc7632a0a743bc009079c67)

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

file: add wrapper to nativesdk-file

"file" command in exported SDK doesn't work:
...
$ file sysroots/
file: could not find any valid magic files!
...

In oe-core commit 68d548cbae729eaea8ce1403dc95ff63c4a7375c,
it added wrapper to file-native. Do the same thing for
nativesdk-file.

(From OE-Core rev: 69a3ab38d1c725c2b575065739cae4a3b45015a2)

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

openssh: avoid screen sessions being killed on disconnect with systemd

Tell systemd just to kill the sshd process when the ssh connection drops
instead of the entire cgroup for sshd, so that any screen sessions (and
more to the point, processes within them) do not get killed.

(This is what the Fedora sshd service file does, and what we're already
doing in the dropbear service file).

(From OE-Core rev: 3c238dff41fbd3687457989c7b17d22b2cc844be)

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

python: force off_t size to 8 to enable large file support

If DISTRO_FEATURES contains "largefile", force the size of off_t to 8 as
a workaround for having ac_cv_sizeof_off_t=4 on 32-bit systems. In
future we will likely drop the value from the site file, but for now
this is a slightly safer fix.

Fixes [YOCTO #6813].

(From OE-Core rev: a8216030ee6c65531de8fbf3eed878a345a94edc)

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

wic/bootimg-pcbios.py: checking the existance of syslinux

While syslinux not existed in $bootimg_dir, there was a error:

$ wic create directdisk -e core-image-minimal
...
|Creating image(s)...
|Error: exec_cmd: install -m 444 /home/jiahongxu/yocto/
build-20141010-yocto/tmp/deploy/images/qemux86-64/syslinux/
ldlinux.sys /var/tmp/wic/build/hdd/boot/ldlinux.sys
returned '1' instead of 0
...

Add checking for the existance of syslinux to fix this issue.
If syslinux didn't exist in anywhere, prompt user to build it.

[YOCTO #6826]

(From OE-Core rev: c2d4364d9fbbda64598f0a3eb0b0890932bb7742)

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

build-appliance-image: Update to dizzy head revision

(From OE-Core rev: fcbd1b19383daf5e160a864df1418130da01be28)

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

bitbake: fetcher: fix getVar call due to incorrect argument datatype

(Bitbake rev: 2ac33aac3446cb12227f1b8daa5f27f417c9bb9e)

Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc: backport patch for gcc bug 61144

This fixes gcc bug 6144, which in my case exhibited itself as a kernel
module that failed to load. This was because static platform_data
structures were being corrupted with the optimiser being set to any
value other than -O0.

Originally-submitted-by: Peter Urbanec <openembedded-devel@urbanec.net>
(From OE-Core rev: 365221f7285c0e392f573deaab3b1e00b12bc293)

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

package_deb: skip pre/postrm scripts on upgrade, write only one shebang

Trying to upgrade busybox removing symlinks but update-alternatives
need these links (sed, cut, tail, etc) in order to work.

Adding test to avoid this scripts on upgrade fix the problem, same
solution are found in package_rpm class.

[YOCTO #6768]

(From OE-Core rev: 7b9161dd0c475cca6ea7eb507f7c3c51869eb493)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mesa_git.bb: Fix fetch and license errors.

Without these changes mesa_git.bb can't be enabled as the
PREFERRED_VERSION due to build breakage.

(From OE-Core rev: 04f48ad3fab4e21a23c46b90f6a62269a1cf1ee7)

Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

build-appliance-image: Update to dizzy head revision

(From OE-Core rev: 1d2a7e0d4ab91b5ad7d0a1c300ab1da7db49520d)

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

package_do_shlibs: Look for provider in the path thats in shlib_provider dictionary

shlib2 code puts the information about path where a provider
will be found. e.g.

{'/usr/lib/llvm3.3': ('libllvm3.3-llvm-3.3', '3.3')}

This is obtained from new shlib2 pkgdata from

llvm3.3/3.3-r0/pkgdata/shlibs2/libllvm3.3-llvm-3.3.list

However when we search for NEEDED libraries we ignore the
key above which is the path where the provider library is installed
and instead just seach in libdir and base_libdir and hence
libraries which are not in above standard search paths gets
ignored even if they appear in DT_NEEDED sections

and a note is emitted

NOTE: Couldn't find shared library provider for libLLVM-3.3.so, used by
files: ....

IMO this note should actually become an error since if we do
not have all DT_NEEDED libraries in image the system is dysfunctional.

This patch extracts this libpath from key and add it to seach paths
when looing for a provider of a shared library

[YOCTO #6798]

Change-Id: Ie5f08632e37ba8d3439c8aaae33bc68b8996792f
(From OE-Core rev: bf8472274ac1e9a35f8cbc82357da6c95b396759)

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

build-appliance-image: Update to dizzy head revision

(From OE-Core rev: 5d0aac8d85d927667c4c26cab53132ff4fa465ff)

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

build-appliance-image: Update to dizzy head revision

(From OE-Core rev: 1216427655dd67d49100cd7bfa50563e2f5965f7)

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

ncurses: enable-pc-files requires PKG_CONFIG_LIBDIR existed

Fixed ncurses.do_configure:
configure: WARNING: did not find library /path/to/tmp/sysroots/qemuarm/usr/lib/pkgconfig

And then anyone requires ncurses.pc will fail.

The configure.in checks:
[snip]
if test -n "$PKG_CONFIG_LIBDIR" && test -d "$PKG_CONFIG_LIBDIR" ; then
[snip]

Create PKG_CONFIG_LIBDIR in do_configure will fix the problem.

We can reproduce the problem by:
Set SSTATE_DIR=/path/to/sstate-cache
1) In build1, make sure everything is ready in SSTATE_DIR
$ bitbake ncurses
2) In build2, rebuild ncurses only:
$ bitbake ncurses -ccleansstate && bitbake ncurses

Then we will see the warning in log.do_configure.

(From OE-Core rev: ea4010c8398745736b0c22ed037604db8fc43212)

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

glibc: remove bad patch snippet that eglibc forward ported

The s_sin.c patch undoes some code changes in glibc itself, these changes have nothing to
do with the option groups and I suspect crept in as part of the initial conversion.  Undoing
this patch also fixes a test failure in test-double and test-idouble.

[YOCTO #6808]

(From OE-Core rev: 5f225067dcb7244065f857d34e1041171844b243)

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

toolchains-scripts: Add support for target environment scripts

In a similar way to the previous script which adds support for native
environment scripts, this adds support for target environment scripts
too.

(From OE-Core rev: 2d9466734f0c0c90724820bc36992b2800ffa4d0)

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

sdk.py: fix write target sdk manifest failed based on ipk

bitbake meta-toolchain
ls tmp/deploy/sdk/poky-glibc-x86_64-meta-toolchain-i586-toolchain-1.7.manifest -al
...
|-rw-rw-r-- 1 root root 0 Oct 10 15:05 tmp/deploy/sdk/poky-glibc-x86_64-
meta-toolchain-i586-toolchain-1.7.manifest
...

The manifest is empty, the reason is target's ipk config path is
d.getVar('IPKGCONF_TARGET') rather than d.getVar('IPKGCONF_Target')

(From OE-Core rev: 81b3cc448f040dcb4c2f2b05983231ac53270663)

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

wpa-supplicant: fix for rebuild

Fixed when rebuild:
make: *** No rule to make target `/path/to/old//sysroots/qemux86-64/usr/lib/dbus-1.0/include/dbus/dbus-arch-deps.h', needed by `dbus/dbus_old.o'.  Stop.

The .d files save the path of the dependencies files which may not exist
when rebuild, we can remove them to make the rebuild work.

(From OE-Core rev: e336102e59dbbd01fe67121738203563476f9456)

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

git: fix do_install error

Fixed when MACHINE = qemux86-64 and libdir = /usr/lib64:
mv: cannot stat `/path/to/image/usr/lib64/perl-native/perl': No such file or directory

The perl-native files are always installed to /usr/lib on both 32/64
bits targets.

(From OE-Core rev: fad6d25e548cb82c2106eb30ccdc0b8f3408de0a)

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

avahi-ui: fix for building systemd with multilib

Fixed:
MACHINE = "qemux86-64"
VIRTUAL-RUNTIME_init_manager = " systemd "
DISTRO_FEATURES_append = " systemd"
DISTRO_FEATURES_BACKFILL_CONSIDERED += " sysvinit "

require conf/multilib.conf
MULTILIBS = "multilib:lib32"
DEFAULTTUNE_virtclass-multilib-lib32 = "x86"

$ bitbake avahi avahi-ui
ERROR: QA Issue: avahi-ui: Files/directories were installed but not shipped
  /lib
  /lib/systemd
  /lib/systemd/system
  /lib/systemd/system/avahi-daemon.socket
  /lib/systemd/system/avahi-dnsconfd.service
  /lib/systemd/system/avahi-daemon.service [installed-vs-shipped]
ERROR: The recipe avahi-ui is trying to install files into a sharedarea when those files already exist. Those files and their manifestlocation are:
   /path/to/sysroots/qemux86-64/lib/systemd/system/avahi-daemon.socket
 Matched in manifest-qemux86-64-avahi.populate_sysroot
 /path/to/sysroots/qemux86-64/lib/systemd/system/avahi-dnsconfd.service
 Matched in manifest-qemux86-64-avahi.populate_sysroot
 /path/to/sysroots/qemux86-64/lib/systemd/system/avahi-daemon.service
 Matched in manifest-qemux86-64-avahi.populate_sysroot
Please verify which recipe should provide theabove files.

And remove the duplicated line:
rm ${D}${base_libdir} -rf

(From OE-Core rev: 1e6ec39d4996d2812b01c4452c579e476e70e85d)

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

nss-myhostname: skip it when systemd

Fixed error when systemd is in DISTRO_FEATURES:
ERROR: The recipe systemd is trying to install files into a sharedarea when those files already exist. Those files and their manifestlocation are:
   /path/to/sysroots/qemux86/usr/lib/libnss_myhostname.so.2
 Matched in manifest-qemux86-nss-myhostname.populate_sysroot
Please verify which recipe should provide theabove files.

(From OE-Core rev: da70a62d456f7efac379381c2c431cc4ded877e9)

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

rpcbind: make service socket activated

Instead of setting 'Restart=always' in the service file, we should
make the service socket activated, just like what Fedora does.

(From OE-Core rev: 176e91ef28800adb6295b29c455b2efb91a01876)

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

sstate.bbclass: split the too long line

The too long line would cause "git send-email" report errors:
patch contains a line longer than 998 characters

Though we can use "--no-validate" to force the send.

(From OE-Core rev: e96723879eb3352a5bdea7b3e1a576edf9550e5a)

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

sstate.bbclass: the second bb.fatal not work

The code:
bb.fatal("foo1")
bb.fatal("foo2")

Would make the second one not work, use bb.error for first one to fix
the problem.

(From OE-Core rev: 33a87187a8520e190bcade76cc965aa58faaa85a)

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

valgrind: Enforce 30 seconds limit for each test

(From OE-Core rev: ae9a89fc1beac13f30395d191954fd70c3a9f85e)

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

boost: fix build when ${PARALLEL_MAKE} contains '-l'

The '-l' option which is valid for GNU make (--> limit by load) has a
different meaning in bjam (--> limit maximum execution time) and will
break very likely the build.

Keep only the the '-l' option when passing PARALLEL_MAKE options to
bjam.

(From OE-Core rev: 1ff36aaec25a7ee89514366fe484345e8d1d7b64)

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

dosfstools: place files in /sbin, not in /usr/sbin

systemd expects fsck programs to be located in /sbin and fails for
dosfstools with

| Checking was requested for ..., but /sbin/fsck.vfat cannot be used: No such file or directory

Other fsck programs (e.g. ext4) are located in /sbin already so move
vfat programs into this directory too.

(From OE-Core rev: fa579cb243b8441d95e6c129e07d9e141f808539)

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

python-pygobject: add libffi to DEPENDS

libffi is generally present as it's also a build dependency on glib-2.0, but
explicitly declare it for determinism.

(From OE-Core rev: 85232b154dbaf3fc2ed3fa9291e3cbeaa7f318ab)

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

neard: update service file

Update the service file to more closely match the service file that has been
committed upstream.

In particular we don't want to restart neard on failure (this results in it
restarting repeatedly if no NFC hardware is found), redirecting stdout to
/dev/null means that any messages are lost instead of being sent to the journal,
and the DBus alias is required for bus activation to work correctly.

(From OE-Core rev: b6afd8e5abcd412c17e14c59379b9583b95fd517)

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

xserver-xorg: update patch to match upstream

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

oe-git-proxy: use SOCKS4a instead of SOCKS4

In some situations where a proxy is required the client can't even do DNS
lookups, so instead of using SOCKS4 use SOCKS4a which moves the name resolution
from the client to the proxy.

(From OE-Core rev: 984455a95b4302d527ff54e019d8ed00611f3664)

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

gnome-desktop-testing: Fix build on uclibc

[YOCTO #6577]

Change-Id: Ia29fb963a7df4e77504993172f85851e0b540c7e
(From OE-Core rev: 1fc544811582ed68b6e0a61fd22c169e1825b725)

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

curl: add PACKAGECONFIG option to use libssh2

The user can enable libssh2 via conf/local.conf or custom distro
configuration, this will pull in libssh2, which is not used by default.

For example, a curl_x.y.z.bbappend file containing the following line:
PACKAGECONFIG += "libssh2"

(From OE-Core rev: d425e005d274cac0ef7160f53c41bda175444f69)

Signed-off-by: Fabrice Coulon <fabrice.coulon@axis.com>
Signed-off-by: Olof Johansson <olof.johansson@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xserver-xorg: report DRI3 and Present modules as present

The DRI3 and Present modules are built-in but some drivers (such as
xf86-video-intel) want to query their presence.  Backport a patch from upstream
to stop this causing an error.

[ YOCTO #6583 ]

(From OE-Core rev: d7bb1d8b1a7cd11895037b7984b9aa916efa9733)

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

latencytop: Creates x11 PACKAGECONFIG.

* Creates x11 packageconfig and update DEPENDS and EXTRA_OEMAKE_X
  for better handling.

(From OE-Core rev: b853cead2e0965b9797b40d6b59bed67804f459f)

Signed-off-by: Noor Ahsan <noor_ahsan@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libassuan: fix for rebuild

Fixed when rebuild:
rm: cannot remove `/path/to/2.1.2-r0/libassuan-2.1.2/m4/*.m4': No such file or directory

The files may not exist when rebuild.

(From OE-Core rev: f04576c761c568083be1143f421e29fc2365846a)

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

meta-toolchain-qt: Fix environment population

The generation of the environment has change since the change to use a
meta-environment canadian package in the OE-Core, the SDK environment
setting has been broken. This uses the new subscript environment to
fix it.

(From OE-Core rev: e7b9e1df19062cfbcd72c90295829424cae6fbed)

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

toolchain-scripts.bbclass: Allow sourcing of subscript for environment

Sometimes we require extra environment settings to be available on the
environment for proper SDK work. This were done, in past, using
'_append' tasks however with the split of the environment in a
canadian package this has been broken.

The easier and more flexible solution is to use environment subscripts
which are sources by the main script. These are now looked at:

 $OECORE_NATIVE_SYSROOT/environment-setup.d/*.sh

and sourced.

(From OE-Core rev: 457291f2ca084d1f43c0cca2175b448a22761887)

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

bitbake: fetcher: fix BB_STRICT_CHECKSUM datatype check

Forcing strict to be a string, to avoid problems when performing comparisons

[YOCTO #6762]

(Bitbake rev: b8ed2098bdea2afd93ab4e3e1b834f3a31cb60de)

Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package_ipk.bbclass: Fix SRC_URI whitespace handling

The SRC_URI may contain whitespace, but be otherwise empty.  This can happen
in the case:

MYSRC = ""
MYSRC_arm = "file://myarm.patch"
SRC_URI += "${MYSRC}"

Unless we strip blank spaces, to determine if it is empty, we can end up
generating a "Source: " line which in invalid.  This leads to the error:

  invalid Source: field is speified in the generated CONTROL file

(From OE-Core rev: 16cedc3bce6fc37543e9ef053cd7c589e523ca1c)

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

bash: update and CVE-2014-6278 fix

Update both bash 3.2.48 (to 57), and bash 4.3 (to 30) to fix the remaining
'shellshock' security issues, CVE-2014-6278.

(From OE-Core rev: a2709547644ae417fbd5435e1372068c7cd5db4c)

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

linuxdoc-tools-native: Makedoc.sh uses /tmp and fails w/ noexec mount

The Makedoc.sh script uses the following line to set TMPDIR

export TMPDIR=`mktemp -d ${TMPDIR:-/tmp}/ldt.XXXXXXXXXX`;

and then later in the script:

chmod u+x $TMPDIR/linuxdoc

Since TMPDIR is not set the script will default to /tmp and if /tmp
is set to noexec (which is becoming more common), the chmod call fails.

(From OE-Core rev: 3a8b8812ac1b9a724f11b2011f8ee3416ac3d4df)

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

package.bbclass: Reverse runtime symlinks should be tied to package generation

In case a package is not generated (is empty and does not has allow
empty flag set) the package data regarding reverse runtime dependency
shouldn't be done.

This were causing a false-positive in the meta-fsl-arm layer, when
building mesa, as:

,----[ Error during build of MX53 in meta-fsl-arm ]
| ERROR: The recipe mesa is trying to install files into a shared area
|  when those files already exist. Those files and their manifest
|  location are:
|    /.../build/build/tmp/sysroots/imx53qsb/pkgdata/runtime-reverse/libopenvg-dev
|    Matched in manifest-imx53qsb-amd-gpu-x11-bin-mx51.packagedata
| Please verify which recipe should provide the above files.
`----

Fixes [YOCTO: #6795]

(From OE-Core rev: 9ef8728514b02dd2e18e87645298d9ec2e8a785a)

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

LTP - realtime tests - fix bad robust mutex conditionals

The tests for robust mutexes contained conditional clauses which failed in
autoconf and/or used nonexistent variable names.  Modified these
conditional clauses to use only the variables actually created by
LTP autoconf for this purpose.

(From OE-Core rev: c0189ef8b58b1e63e227c5040cb1c9e915f225c2)

Signed-off-by: Gary S. Robertson <gary.robertson@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

LTP - realtime tests - fix bad PI mutex conditionals

The priority inheritance tests for mutexes used conditional variables
which were non-existent.  Changed the conditional clauses to use the
variables which were actually generated by LTP autoconf for that purpose.

(From OE-Core rev: 812114ad23def92306fbf9f7afb03cee4cbd10d4)

Signed-off-by: Gary S. Robertson <gary.robertson@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bash: Upgrade bash to latest patch level to fix CVEs

We upgrade bash_4.3 to patch revision 29, and bash_3.2.48 to 56.

There are numerous community bug fixes included with this set, but the key
items are:

bash32-052      CVE-2014-6271                           9/24/2014
bash32-053      CVE-2014-7169                           9/26/2014
bash32-054      exported function namespace change      9/27/2014
bash32-055      CVE-2014-7186/CVE-2014-7187             10/1/2014
bash32-056      CVE-2014-6277                           10/2/2014

bash43-025      CVE-2014-6271                           9/24/2014
bash43-026      CVE-2014-7169                           9/26/2014
bash43-027      exported function namespace change      9/27/2014
bash43-028      CVE-2014-7186/CVE-2014-7187             10/1/2014
bash43-029      CVE-2014-6277                           10/2/2014

(From OE-Core rev: 43deeff0c6b0ea7729d3e5f1887dfd1647dea1da)

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

pseudo*.bb: update to pseudo 1.6.2

pseudo 1.6.2 fixes problems with 64-bit inodes and some underlying issues
involving file renames that could occasionally cause very strange behaviors
files being deleted, linked, or renamed, mostly observed as strange
recovery if an inode got reused.

(From OE-Core rev: b2c6a032d6e5deb07e76ed75fcd0931fad6a748c)

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

cross-canadian: Disable the packagedata stamp-extra-info

Similarly to native/cross disable this since otherwise the packagedata
can be marked as machine specific and if you switch machines
which share an architecture, you'll get toolchain overlapping files
errors.

(From OE-Core rev: 96d557be3dedd6aea6199b3d28fbb7f5549fad69)

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

meta-environment: Deal with machines which change TARGET_OS

Some machines change TARGET_OS, cross-canadian resets this which
is not what we want in this specific case. This fixes spe toolchains
for example.

(From OE-Core rev: 0038634ee6e2b6035c023a2702547f20f67c103a)

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

gcc-runtime: Add linux-gnuspe symlink to fix c++ headers

Some architectures can mix different TARGET_OS values, in most cases
we just use one but in the ppc case, can use two different values. In this
case, to use one toolchain with both, we need to ensure the symlinks exist.

This isn't ideal but does fix the ppc toolchains for the release, after
which better ways of handling this can be investiaged. Without this, failures
in the C++ toolchain are seen.

(From OE-Core rev: 112641117f1152bad8a806f1aa872a67575d5316)

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

kernel.bbclass: enable a link for external module building

Even though the kernel-dev package provides the required support for
building external kernel modules on the target, some commonly used scripts
and utilities fail as they are not finding the kernel module build support
files at the desired location.

Create the /lib/modules/<kernel-version>/build link on target pointing to
the sources provided by the kernel-dev package, to fix the issue.

Fixes Bug:
[YOCTO #2968]

(From OE-Core rev: aafa4bc896eb944aa4fc406807dd7e02f4b9b7ba)

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

kernel: Added bc-native as DEPENDS

The makefile checks for bc during for compilation

[YOCTO #6781]

(From OE-Core rev: c067e52cffe002de3b39aa1bced308dd532859c1)

Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

adt-installer: Fix to work with meta-environment changes

There is a proper sysroot specified in the meta-environment script now,
which isn't a bad thing. We adjust the sed expression to cope with this.

That means the SDK installations in non-default paths start to work again.

(From OE-Core rev: 3be7b59893ed77f63eeca35b686df06a1dafb53d)

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

uninative-tarball: Update eglibc -> glibc

(From OE-Core rev: 2b85b3f33af5157cd4b6f8a6dc737015c85018c3)

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

oeqa/dmesg: Whitelist usbhid failures

We fixed these in parselog but not here. This test really can just be
deleted now really.

(From OE-Core rev: c598dbdbaebf95cc26e95138b4c3fcb15af67a88)

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

adt-installer: Set INHIBIT_DEFAULT_DEPS

We've been seeing an issue on the autobuilder due to the way it builds
package feed indexes. Packages get copied into the nightly build which
then creates indexes however this is done without the knowledge of
sstate. When adt-installer is built, it has dependencies on the toolchain
and when ipk files already exist, the build throws errors.

Since this recipe doesn't need a toolchain, we can remove the dependencies
to work around this for now. I'm not particularly happy with what the
autobuilder is doing but that is a post release issue.

(From OE-Core rev: 6055263b28698a2c79c1605aca2f3810d2aa140d)

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

build-appliance-image: Update to dizzy head revision

(From OE-Core rev: 8e3e36bcafcb7a97006f8cc49e51c3a5b8e51a65)

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

oeqa/parselogs: Ignore qemu usbhid errors

These are harmless from the USB pointer device we install, ignore them.

(From OE-Core rev: f5bdf41e78ff378fe23d8ba1543917bc64def62f)

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

gtk+3: Add missing MLPREFIX

Without this each tries to provide libgtk-3.0 which directly conflict when
using multilibs.

(From OE-Core rev: afc4412f9ff14dd45aba6a47bfc87e703f3c4763)

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

meta-environment: Set libdir correctly to fix PKGCONFIG entries

Currently libdir is coming from cross-canadian which is incorrect. We
need to reset this to target_libdir so that the toolchains contain the
correct value.

(From OE-Core rev: 41f8f32c8da705ead464ee69bc5a1e120b137693)

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

apt: apt-key binary was not being installed, including it in the installation

[YOCTO #6476]

(From OE-Core rev: d31fc181005734953c70ba9338d6a5eb88b6fe8b)

Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

perf: fix undefined pr_* routines

When cross compiling libunwind support for ARM a missing debug include means
that pr* macros are not expanded, and hence link failures on the undefined
functions.

Since we must be compatible with many versions of the kernel and perf, we
sed the proper include into the files, while the permanent fix goes upstream
to the mainline kernel.

(From OE-Core rev: 81bb7a163d7e3c0bdcc72894ef731521d58edf90)

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

linux-yocto/3.17: update to v3.17-rc7

Updating the 3.17 tree to the rc7 release.

(From OE-Core rev: 6957ce3e726de5dbdcf3e94c5919916d88736e8d)

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

linux-yocto/3.14/3.17: allmodconfig/allyesconfig build fixes

It was found that some of the recent feature merges for 3.14/3.17 are
not allmodconfig and allyesconfig safe.

Since this is a basic test before kernel patches are submitted, we've
fixed the features to meet this standard.

Integrating the following fixes from Paul Gortmaker:

  b4213d81ea3f fat: don't use obsolete random32 call in namei_vfat
  2cc7eba15c1f cryptodev: stomp dynamic version numbering for in tree builds
  5d1dda7aae4b Target/dif: Introduce protection-passthough-only mode
  3d9772d8facf vhost: fix compile fail due to reallocated acked_features field.
  efad59d3a174 virtio-scsi.h: Add virtio_scsi_cmd_req_pi + VIRTIO_SCSI_F_T10_PI bits
  902f34d36102 aufs: apply aufs3-mmap.patch from 3.14 branch
  30efc2e9484e aufs: import core files from aufs3.14 20140915
  e42f87adef10 Revert "aufs: aufs3-mmap.patch"
  a818774bd338 Revert "aufs: core aufs filesystem"

(From OE-Core rev: d1c40ccb522b5c5a61d5faab7e0f65491e201a27)

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

linux-yocto/3.14: revert BFP feature, and fix intel-common -rt

The BPF feature is causing build failures, so we are reverting it
for now.

In this update we also have the following fix for -rt:

  fb6271a942b5 intel: Remove the standard ktype nesting

(From OE-Core rev: 57a82d3bf0bf34bba9d0801057a4b8a6aa230228)

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

linux-yocto/3.14/3.17: menuconfig and cryptodev

Two issues were reported with the 3.14 kernel, cryptodev was not properly
building and working on all devices, and menuconfig was not working on some
hosts.

To fix this, we pull in the latest cryptodev updates, and restore an old
ncurses patch for menuconfig.

(From OE-Core rev: 35f932314541067b16b60ed5bc054a80f973dd35)

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

kern-tools: fix overly greedy path relocations

During patch processing a consolidated set of configs, patches and directives
is created under the kernel source tree being modified. During that processing,
absolutely paths are converted to relative. It has been found that if directories
are sufficiently similar, like so:

  /path/to/my-linux
  /path/to/my-linux-3.16

The processing will chop to much of some paths, resulting in invalid relative
directories (like -3.16 in the above example).

Importing the following two kern tools fixes for the issue:

  23345b8846fe kgit: retain trailing / in directory processing
  a8cf93a3bc94 kgit-s2q: move subject and diffstat mismatch to 'fuzzy' matching

[YOCTO: #6753]

(From OE-Core rev: 660c90458e8b4114e4a8deb920e44263e03a1ec6)

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

build-appliance-image: Update to dizzy head revision

(From OE-Core rev: ae8319fb460b5d588044f6b00bf73255773f4ff0)

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

perl: Improve sysroot regexp

When rebuilding libxml-parser-perl with a change to libdir, you see strange
build failures due to MakerMake looking in strange library paths. The error
is obtuse and hard to track down. I'm therefore proposing we change the regexp
once and for all to resolve the issue. Currently it only does a replacement
once, this change ensures it always gets set the correct value upon rebuilds.

(From OE-Core rev: 2c1c70eef4df66a0208f60ee51bd36d8f794144e)

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

gdb-cross-canadian: Add to sstate whitelist

The various gdb variants overwrite in datadir so whitelist this for now, they
are the same files and this is not an issue.

(From OE-Core rev: 472efca12108d26201d2236ff436a08845313f48)

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

gtk-doc-stub: update to latest commit

* fix build for packages with AC_CONFIG_MACRO_DIR set but not shipping macros
  causing:
  | ln: target 'm4/' is not a directory: No such file or directory
  | cp: cannot create regular file 'm4/': Not a directory
* In 2012 version was increased to 1.1

(From OE-Core rev: 748056672b2dc8d65107dde84c83171d9ba53091)

Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mx: remove version from patch directory

(From OE-Core rev: 11bc6489d534ecf904b04ab9e9b6415d4415492a)

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

rpm: add version 4.11.2

add patch to remove db3 from configure.ac
add inherit pkgconfig

(From OE-Core rev: 9a17f938e738a16a1ef9a00be6a8317d71f92573)

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

python-smartpm: Add patches for rpm4

Add DEPENDS for python-rpm package from either rpm4 or rpm5

Extend the smart-dflags patch to catch an exception if the api does exist

(From OE-Core rev: 756e499a95cc928688684cc52bde8e31306e6bbc)

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

rpm5: add python-rpm PROVIDES

Since python-smartpm depends on python-rpm, we should provide here as appropriate.

(From OE-Core rev: a3598f87bd22354cce2f2be06f09c2b24f2adb63)

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

nss: nss.pc is not target specific

RPM4 requires an nss-native component

(From OE-Core rev: f70efca58e9411feb251c9d00066f8631b167004)

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

tzdata: update to 2014h

(From OE-Core rev: 1bfefe647911b46efa2f7aaf84dc02c25a9a158d)

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

tzcode-native: update to 2014h

(From OE-Core rev: c50e5f1bc8c597ad055a243bef4591a7cfee6355)

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

deb_packaging: Added support for multilib

The Package Manager implementation for deb didn't
take a look about multilib enabled variants.

Changes are made for generate apt.conf, sources.list and
debian repo index Release and Packages files.

[YOCTO #1502]

(From OE-Core rev: b5fb879b351cc23977f3e441f758101551297566)

Signed-off-by: Anibal Limon <anibal.ezau.limon.belmares@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ltp: Add additional m4 path to autoconf.

The ltp code has some m4 macros that are deeper than the default depth
that the directory depth the autotools.bbclass checks. This causes some
macros to not be found and for supported features to not be enabled.

This patch adds the extra m4 path to the autoconf arguments.

(From OE-Core rev: 4d8fa4b7a02d1d53f75943607df7e8126e6dbeb8)

Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sstate: Add rpm allarch to overwrite whitelist

The packagegroup allarch rpm files for multilib can overwrite each other since
they are in theory indentical (in contrast to the other backends). We therefore
need to whitelist this to avoid build failures now this overwrite failure
is fatal.

(From OE-Core rev: d59ade0ca2cf629937434fa423dfbf35ce1209fc)

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

ltp: add PACKAGECONFIG for numactrl

(From OE-Core rev: 4c7873552e13dfdba96afca7562c398d2966ca71)

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

squashfs-tools_4.3: modify SPDX_S variable

Modify SPDX_S to the source tree of squashfs instead of ${S}.

(From OE-Core rev: 1d79bea9dadd7e78fd558046497cb48b7d9b46e9)

Signed-off-by: leimaohui <leimaohui@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

icu.inc: modify SPDX_S variable

Modify SPDX_S to the source tree of icu instead of ${S}.

(From OE-Core rev: 69c8f17eb35ef42dcf538e39b80802c0a70548f9)

Signed-off-by: leimaohui <leimaohui@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>