kernel.bbclass: update KERNEL_IMAGE_MAXSIZE

* Make KERNEL_IMAGE_MAXSIZE and IMAGE_ROOTFS_SIZE have the same algorithm:
  - Use Kbytes as the unit since we use this in other codes.
  - Use "du" rather than "ls" to figure out the size since we use this in
    image.py.

[YOCTO #2610]

(From OE-Core rev: 7384d2831c713ac5999aca83c312154dc15cec56)

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

gzip: move uncompress bin to avoid cross-fs link

(From OE-Core rev: 16aa94e4543fc6e6540b203bc716e6f275f514c7)

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

package.bbclass: Generate runtime-rprovides data

Currently, given a list of rdepends its near impossible to figure out
which ones are potentially invalid and which might be an RPROVIDES of a
package.

This problem is simple to solve, we can write lookup data into the
pkgdata. This patch does that, accounting for the fact that multiple
packages can RPROVIDE the same thing (particularly with locales). Its
done with symlinks so the performance overhead is minimal.

(From OE-Core rev: 30b4bff944c4625696e9be97ce4912a2710d13e9)

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

insane: Split do_package_qa into a separate task (from do_package)

Its possible to run the package QA checks as a separate task rather than
as part of the do_package task. This offers more parallelism but the
fact that made me propose this is that ideally we'd like to access
pkgdata to help add new tests and to do that, we need to run later in
the task list. We also need to add in RDEPENDS to the task which apply
to do_package_write_* but not do_package. See the subsequent patches
for why this is desireable.

If we split into a separate task, we need to add in calls to read
the sub package data, build the cache structure used by do_package and
cover the task with sstate (which is empty and just acts as a stamp
saying it passed package QA). We also need to handle our own
dependencies.

(From OE-Core rev: d31f75a72cb2322f753d945da86d5f3287eef06b)

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

package.bbclass: Rewrite shlib_provider handling to include RPATH

Change the do_package shlibs code to account for RPATHS. This means that
for library dependency purposes, only libraries in system paths or in
any declared RPATH will be seen. This is important to resolve problems
people have been having where similarly named libraries in "private"
paths were conflicting, e.g. with gstreamer.

For now this code assumes the default search path is libdir and
base_libdir and places ASSUME_SHLIBS in libdir so they are searched by
default.

(From OE-Core rev: 48b9ecd5335f63fb2bcc47c4b23d067122734fe9)

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

package.bbclass: Improve shlibs pkgdata file format

Instead of having a .list file and a .ver file, place the version
information into the .list file in a ":" delimited string.

Also place the path to the library here, this can then be used to
evaluate RPATHs in the shlib dependency code.

Since the disk format has changed, the easiest way to avoid build
failures in the same TMPDIR is to change the shlibs directory to
shlibs2. sstate dependency code with ensure everything rebuilds.

(From OE-Core rev: 8dd09b3ddc572a338b4a6f1d859b68b2b4482d00)

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

package.bbclass: Rewrite sonames data structure to include library path

In order to do more advanced processing of the shared libraries, we need
to know where a given library is located on disk so we can know whether
its a system path or a private directory for example.

This patch adds this information into the 'sonames' data structure.

(From OE-Core rev: 5f48d53c5ed94b5c6aad9f73ff177aa35af90397)

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

package.bbclass: Improve shlibs needed data structure

Improve the shlibs 'needed' data structure to include the file and
any rpath information. This allows various cleanups to the data structure
and moves us closer to being able to resolve shlibs providers issues
based on path in due course.

This commit doesn't change any stored data, just cleans up internal data
structures (for example dropping the needed_from dict).

(From OE-Core rev: d3aa7668a9f001044d0a0f1ba2de425a36056102)

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

sanity.bbclass: Update minimum git version to 1.7.8

To quote "Maxin B. John" <maxin.john@enea.com>:

git version 1.7.8 added the --list option to git-branch. Since we depend on this
option in git.py, the minimum requiremnt for git should be updated to Git
1.7.8+

(From OE-Core rev: a5776ae532d6af1e880ab1a712dc768c900f88db)

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

adt_installer: not download repo page

Add "--spider" argument to wget command, since we don't need to download
repo page, just check that it is there.

(From OE-Core rev: 72f24557bceace990e8e5d2d3b91586fc394cc3b)

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

adt_installer: check the result of updating opkg

We should check the result of updating opkg rather than echo_info.
So we should call check_result function after updating opkg.

(From OE-Core rev: d55df5c6a707a3cf5e8d95896b1e43193482d95f)

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

adt_installer: fix syntax error

We need to check YOCTOADT_QEMU variable whether is equal to "y".
So we should use "==" rather than "=".

(From OE-Core rev: 656a784a7c43b619d51a47aab926d7c519cc3b4b)

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

adt_installer: remove unnecessary variable

YOCTOADT_VERSION is not included in adt_installer.conf file.
Remove this variable, because we don't need it.

(From OE-Core rev: 9b7720f4efc578f0673935ab9ad98ad3f392716e)

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

binconfig-disabled: add a default value and comment for BINCONFIG

To avoid syntax errors when inheriting this class without setting BINCONFIG, add
a default value and a comment.

(From OE-Core rev: a7e808e9a0ff45f14235b2de3c53d89d98fde714)

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

xserver-xorg: fix xshmfence autodetection

Add a PACKAGECONFIG for xshmfence.  If DRI3 is enabled, xshmfence also needs to
be enabled.

[ YOCTO #6507 ]

(From OE-Core rev: ff2e96cacc116cf362d4c5dc36f1b846672f98e0)

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

xserver-xorg: add PACKAGECONFIG for libunwind

The automatic detection of libunwind causes non-deterministic builds, so add a
PACKAGECONFIG option for it.

(From OE-Core rev: 6b1418a2ba1544ea481fd4a89b5aa25111ca20e8)

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

bluez5: upgrade to 5.21

Bug fixes;
Features parity with Bluedroid and additional
support for Bluetooth 4.1 core technology and
upgraded profiles including HFP 1.6 + Wideband speech,
A2DP 1.3, AVRCP 1.5 and MPS;
It also provides full integration for aptX(R)
high-quality low-latency codec from CSR (codec license not included).

(From OE-Core rev: 1aeda16c353d855dc7e58aef5cbfebe2b462b1d8)

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

libcroco: disable /usr/bin/croco-6.0-config

librsvg is the only recipe depending on libcroco, which doesn't refer to
croco-6.0-config any more, so we inherit binconfig-disabled here to
disable it.

(From OE-Core rev: 6776b9210deab868557e7871ad5f6ade4c79376c)

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

libcap: use PACKAGECONFIG for attr and pam configurations

Instead of hard-coding PAM/xattr on for target and off for native, slightly
abuse PACKAGECONFIG to add options (and pass them to make in do_compile).

(From OE-Core rev: 51540b64f62234c145fc32cfa3fbbaaebbeece08)

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

groff: fix bindir transaction error

bindir is transfered to Makefile as ${D}${bindir}, hence it should not
use DESTDIR as the prefix directory any more during the install.

(From OE-Core rev: c050e5c9490e9e8e07db7fbd5e6c1e1dbef3bf8c)

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

rpm_5.4+cvs: Add RPM community tree for future development

Enable the rpm-5_4 branch via a specific recipe to help track the latest
community development.  This should allow us to more quickly move to the next
release when it is available.

(From OE-Core rev: c461454d7f3a20ccf9ca20c5c41c80c9becd985c)

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

package_rpm.bbclass: Should be using HOST_* not TARGET_*

When building target packages, HOST_OS and TARGET_OS are the same, as is the
VENDOR field.  However, when building an SDK this is not true.  The patch
corrects the oversight and switches to using the 'HOST' version and resolves
the issue of meta-mingw not working w/ the rpm packaging.

(From OE-Core rev: 0fa1eb5c1296a38b332b63d539608da7ef56c3c0)

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

rpm: Replace patch with backport from SCM version of RPM5

(From OE-Core rev: 17419e4577eb4b5786d4c0120f070e1660fa28ec)

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

rpm: Upgrade to 5.4.14

Update various patches.  A few corrections to the patch descriptions,
otherwise simple quilt refresh or conflict resolution.

Remove rpm-solvedb.patch and rpm-respect-arch.patch.  These are both related
to the old solvedb package dependency solver.  This is no longer used since
we moved to smartpm.

rpm-stub-out-git_strerror was a backport and is no longer needed.

RPM 5.4.12 and newer normally requires Berkley DB 6.0 or newer.  A small
patch to configure allows RPM to dynamically select DB 5.3 or DB 6.0 based
on what is available at configure time.

(From OE-Core rev: 0c7b4a5e23836889196f85f472f081d51529e94e)

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

db: Add version 6.0.30

The latest version of RPM5 requires Berkley DB 6.0.20 or newer.

The license is now AGPL-3.0 in BDB 6.0.  This may not be acceptable to some, so
the previous version is retained as an alternative.

(From OE-Core rev: 04bb223110da2f92725c341bc3ec0cf26325f675)

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

perl, perl-native, perl-ptest: upgrade from 5.14.3 to 5.20.0

Changed:

- The Copying has no change, except the company address.

- pick patches from debian
  http://ftp.de.debian.org/debian/pool/main/p/perl/perl_5.20.0-1.debian.tar.xz
  - Not used by oe:
    deprecate-with-apt.diff
    patchlevel.diff
    fakeroot.diff

- Create/Update perl-rdepends_${PV}.inc by the hardcode script;

- Update config.sh by:
  1) Copy the Perl 5.20.0 source code onto your TARGET machine
    linux qemuarm 3.14.5-yocto-standard from OE-Core rev:
    f506d0660c9949485268a92724ac770b5457b0ca
  2) Execute sh Configure as normal and configure as required,
     do not "make";
  3) Compare with the old config.sh files, and update;

- perl-ptest.inc
  1) Copy the souce code to ptest since almost 112 test cases
     failed with the reason that no souce code found;
  2) Add two patches to fix test case issue;

- perl-native
  Reference perl (5.20.0-1) in debian to update perl shared library headers
  https://packages.debian.org/experimental/i386/perl/filelist

Obsolete:

- 09_fix_installperl.patch
  The dead code was removed from installperl
  http://perl5.git.perl.org/perl.git/commit/236818e0b9d9fe874831086b4d0b94dc6f245dfd

- perl-build-in-t-dir.patch
  The upstream has fix it. The issue description:
  Perl cannot cross build in a path containing a directory that has the
  name of "t".  As an example, you can make the perl build fail with
  "mkdir -p /tmp/build/t", go to the directory, unpack the sources,
  configure and cross build.

- 0001-Fix-misparsing-of-maketext-strings.patch
  as they are part of the upstream code now:
  http://perl5.git.perl.org/perl.git/commit/1735f6f53ca19f99c6e9e39496c486af323ba6a8

- 0001-Prevent-premature-hsplit-calls-and-only-trigger-REHA.patch
  the hash function changed:
  http://perl5.git.perl.org/perl.git/commit/7dc8663964c66a698d31bbdc8e8abed69bddeec3

(From OE-Core rev: c7ac82415efc42ff7a93c6df163f88f2dde00d26)

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

kexec-tools: upgrade to 2.0.7

Kexec-tools v2.0.7 is out, upgrade accordingly.

(From OE-Core rev: 318a8e65c8a9cc016fd9e0b7093443a27ade79a2)

Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
CC: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libtasn1: update to 4.0

Bump version to 4.0

(From OE-Core rev: b4b52af3781b2c32b78b8a897a027bc51e056016)

Signed-off-by: Maxin B. John <maxin.john@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gnutls: update to 3.3.5

Bump version to 3.3.5

(From OE-Core rev: daa99947cb03d9e1a7d6c2c2fa71ab0399209689)

Signed-off-by: Maxin B. John <maxin.john@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

db_5.3.*: Upgrade to 5.3.28

Simple upgrade, the license checksum change is related to a copyright date
change.

(From OE-Core rev: cc6223d24bfd85240399272ebf5f1dddb6150c79)

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

bitbake: fetch2/svn: Add transportuser parameter

There may be a need to set the user for the transport rather than the subversion
command itself. Add a parameter to the url to allow this.

[YOCTO #6475]

(Bitbake rev: 56c294dc30b6c2575b1cf904e26b8b8bef7677c2)

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

bitbake: lib: fix no newline at end of file

Add a '\n' to the last line of the file to fix:

No newline at end of file

(Bitbake rev: 54f1359ed2e9d47980cd221b7b43ef56543fe06d)

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

bitbake: toaster: fix no newline at end of file

Add a '\n' to the last line of the file to fix:

No newline at end of file

(Bitbake rev: 5eb65d92c35264087e5d82c35638f3b8805b1b3e)

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

yocto-bsp: Drop GLIBC_ADDONS setting

This has been copied and pasted around and is not needed, simply drop it.
The defaults in the recipe work just fine.

(From meta-yocto rev: 037bce9a5f196df3cded9dfe6200188c1dd1d84a)

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

yocto-bsp: Update qemu inclusion lists

Update qemu tune definitions to match changes in main qemu machines.

[YOCTO #6482]

(From meta-yocto rev: 0d78ffd509c6caba6c74c6e75c485fb8d923cd31)

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

build-appliance-image: Update to master head revision

(From OE-Core rev: ee70d0cd42a280fd230abf35036ad237c96b3065)

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

libxml-parser-perl: fix do_compile failed with unrecognized option '-Wl, -O1'

Use '${CCLD}' as '${LD}' which the cpan was doing.
...
i586-poky-linux-ld: unrecognized option '-Wl,-O1'
i586-poky-linux-ld: use the --help option for usage information
...

Also fix do_configure warnings
...
'--SYSROOT' is not a known MakeMaker parameter name.
'-MARCH' is not a known MakeMaker parameter name.
...

(From OE-Core rev: c67dc89179977b2df80f25ebf66b7e983819a833)

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

openssh: avoid checking empty passwords to mess up with PAM modules

Previously, even if PAM is enabled for ssh, the daemon still tries to
authenticate an empty password. This leads to authentication failure
which would mess up with PAM modules.

As a result, if 'UsePAM', 'PermitEmptyPasswords' and 'PasswordAuthentication'
are enabled, no user can login correctly. We would meet the following error
message at the client side.

Write failed: Broken Pipe

This patch fixes the above problem by checking whether PAM is enabled before
authenticating an empty password. After all, if PAM is enabled, the task of
authenticating passwords should be handled to PAM modules.

[YOCTO #6466]

(From OE-Core rev: e017ae71dad4837b0d22f291b0b0e0949075f822)

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

autotools.bbclass: remove autotools_set_crosscompiling

The only reference to this function is a commented-out assignment, and nothing
in oe-core nor meta-oe uses autotools_set_crosscompiling directly.  As it's
unused, remove it.

(From OE-Core rev: adaa0568390d0dfef1c4d87809601aab85299e97)

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

scripts/contrib/verify-homepage.py: add a script to verify HOMEPAGE

Add a script to verify all the recipes' HOMEPAGE.

[YOCTO #5689]

(From OE-Core rev: 2dfb2ef06763cb6044dc1630875024e7310b3df4)

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

bitbake.conf: automatically add libexecdir/BPN/.debug to -dbg

pkglibexecdir is a fairly common location for package-specific binaries (in
automake this is $libexecdir/$PACKAGE), and binaries in there are already
installed to FILES_PN, so add the corresponding .debug directory to
FILES_PN-dbg.

(From OE-Core rev: 4d3ffde4649ed116a1c21afef41f71bfe1d471de)

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

glib-2.0: silence QA warnings about libraries outside ${libdir}

The test suite is installed into ${libexecdir} and as that may be outside of
${libdir} the QA test needs to be silenced.

(From OE-Core rev: d354d339a38d22f1a5bca14341229d2282647fbb)

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

Added testcase decorator to use in logging. Added class decorator LogResults that outputs test results in separate log file.

(From OE-Core rev: 7e2b73f1ccfe2968ef780fef2edfaa31c3dae853)

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

elfutils/elfutils-native: Fix patching generated files

 Patch redhat-portability.diff was patching generated files.
 This patch removes every hunk related to Makefile.in and configure files.

 [YOCTO #6491]

(From OE-Core rev: 60efcf67ba2b556bcd5ffea0056884189d5bed52)

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

lib/oe/package_manager.py: Fix _pkg_translate_smart_to_oe function

File: '/home/mhatle/git/oss/oe-core/meta/lib/oe/package_manager.py', lineno: 278, function: _pkg_translate_smart_to_oe
     0274:
     0275:            if found == 1 and fixed_arch == fixed_cmp_arch:
     0276:                break
     0277:        #bb.note('%s, %s -> %s, %s' % (pkg, arch, new_pkg, new_arch))
 *** 0278:        return new_pkg, new_arch
     0279:
     0280:    def _list_pkg_deps(self):
     0281:        cmd = [bb.utils.which(os.getenv('PATH'), "rpmresolve"),
     0282:               "-t", self.image_rpmlib]
Exception: UnboundLocalError: local variable 'new_arch' referenced before assignment

(From OE-Core rev: acd8bfc9378df0a2e1d6ea3858675b9fe350946d)

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

libav: add PACKAGECONFIG for theora, libvorbis, speex and openssl

Dropped libvorvis dependency because there is already a (better) builtin vorbis codec.
Don't confuse codec names: "vorbis" (builtin) and "libvorbis" (using external library).

Changes:
- add --enable-nonfree when faac or openssl are used
- add DESCRIPTION
- sort PACKAGECONFIG entries

Tested with libav-0.8.11 & libav-9.13.

Note: Be sure to have a recent version of bitbake (2014-06-11 or better) including this:
http://cgit.openembedded.org/bitbake/commit/?id=2e742c03e8dfdfa67899e7f5d579ed14bd87e139
It affects behavior of bb.utils.contains_any

(From OE-Core rev: 62eede9de6fa1c76fa97ca5c6ba2d31309759b10)

Signed-off-by: Matthieu Crapet <Matthieu.Crapet@ingenico.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gtk+: Add PACKAGECONFIG for directfb

* building without x11 doesn't work, because it sets default
  gdkbackend to x11 and then requires cairo-xlib to be available
* checking for CAIRO_BACKEND... no
  configure: error: Package requirements (cairo-xlib >= 1.6) were not met:

  No package 'cairo-xlib' found

(From OE-Core rev: a9f36bee8ef79ce7c9e3545ab5c854a45c232661)

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

adt_installer: Fix real envrionment script issues

The previous change to adt_installer was incorrect, this reverts it and
fixes the real issue. The real issue is that if a previous installation was
present, it may have been moved and therefore the paths returned by opkg
could be a little odd.

To address this, when we move files to a different location, ensure we
update the *.list files with the revised paths. This means later installs
over the top return consistent values from opkg and the script works
as intended.

(From OE-Core rev: 8a206f88e5396d7d136480e0efdb5071d5aafeb6)

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

toolchain-scripts: Add handling for CANADIANEXTRAOS

When we build a general toolchain script we should add all the OS
variants to PATH, not just the current one. This is because some can
cross reference each other and if the triplet prefixed ld can't be found
for example, you recieve strange errors.

Doing this resolves bugs exposed during SDK testing.

[YOCTO #6490]

(From OE-Core rev: 5f8bbb15c2b79a94ad547347306b59bd67dba3e8)

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

meta: fix no newline at end of file

Add a '\n' to the last line of the file to fix:

No newline at end of file

(From OE-Core rev: b3090263ba31702631270643c7a7d7af8f4d9234)

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

runqemu-extract-sdk: normalize paths to improve output

Previously, we would have output like below from runqemu-extract-sdk.

    Creating directory /home/chenqi/poky/build-systemd/./nfs-root

Normalize paths for runqemu-extract-sdk so that we have output like below.

    Creating directory /home/chenqi/poky/build-systemd/nfs-root

(From OE-Core rev: c85f5ec502fecb1ff63f8e795a0da5fc92eca0c1)

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

qemu: fix qemu-native pkg-config paths

For qemu-native, the pkg-config paths do not
include build host paths.
This is an issue for libsdl for example, where SDL is
used by qemu, but for qemu-native libsdl-native is not
built, but assumed to be provided by the build host.
Because pkg-config do not search for libsdl config files
on the build host sysroot, the configure stage of qemu-native
will fail because it will not find SDL as being installed.
Usually, the isssue is masked by a functional sdl-config that
will be interogated instead of pkg-config. However, on Build
Appliance, sdl-config is deliberately made non-functional,
so the issue manifests itself.

The fix will create an extended PKG_CONFIG_PATH, which does
include the build host sysroot paths for pkg-config.

Fix for [YOCTO #6495].

(From OE-Core rev: 68a5ed337f8f7ee8e5bf55542ec82d786eb754db)

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

init-install-testfs.sh: add '--hotkey x' to 'test' menuentry

Add a hotkey for the GRUB 'test' menuentry. This can be used by expect scripts to boot into 'test' when doing runtime hardware tests.

(From OE-Core rev: 17b97fd6c724ba6e506cbadb18facdfd9c472e79)

Signed-off-by: Corneliu Stoicescu <corneliux.stoicescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: templates: Added newline character to end of script.

Prevents errors when editing in vi and writing and quiting.

(Bitbake rev: 01692654cd792eaf6362bbac9c34314db4a7ffea)

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

bitbake: bitbake-user-manual-ref-variables.xml: Edits to OVERRIDES variable.

More detail.

(Bitbake rev: 0ac8a504de1798dae02eb2dade2776e7e25d034d)

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

bitbake: bitbake-user-manual-execution.xml: Minor fixes.

Collection of typo and grammar fixes from "Execution"
chapter.

(Bitbake rev: ebca1b7847ff34213e6da71c79f00cb4b6b0757b)

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

bitbake: bitbake-user-manual-intro.xml: Minor fixes mostly grammar adjustments.

(Bitbake rev: fedc5470e9f4f5438d3de957f47985f76a484533)

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

bitbake: bitbake-user-manual-fetching.xml: Small fix in Git Fetcher

Fixed how we refer to the "rev" paramater.

(Bitbake rev: 90b1aec42cd29716f1cbd058757941cbfda235b0)

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

bitbake: bitbake-user-manual-execution.xml: Typo and grammar fixes.

(Bitbake rev: c469ceb86c1f76e9fa867ffb8a5c9b120ed7942b)

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

bitbake: bitbake-user-manual-execution.xml: Edits to "Providers and Preferences"

I split this section into two sections and tried to simplify what
was being said about PROVIDES.

(Bitbake rev: 394f64123d5ef1e9b4628d083488b9432eccbded)

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

bitbake: bitbake-user-manual-fetching.xml: Small wording change.

(Bitbake rev: 66529a17d50bbce211361d98ace012b9424459c9)

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

bitbake: bitbake-user-manual-ref-variables.xml: Edits to PROVIDES variable.

Added some more detail to help describe the variable.

(Bitbake rev: 4b2019e67324f0f86e1a49d256eddb4f9ca597f4)

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

bitbake: bitbake-user-manual-style.css: Added permalink style for display

Fixes [YOCTO #5772]

Added a couple styles to the CSS style sheet so that the displayed
permalink symbol is barely visible unless you hover over it with
the mouse.

(Bitbake rev: ee8db0cda1c8716ab0151a4859bfff84b2f3bd58)

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

bitbake: bitbake-user-manual-customization.xsl: Added support for variable permalinks.

Fixes [YOCTO #5772]

Including the gloss-permalink.xsl file in the customization layer
so that making the manual now generates permalinks for each variable
definition in the glossary.

(Bitbake rev: 04467594a599472c367365dffe205bf727c4d592)

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

bitbake: template: Added XSL template to support variable glossary permalinks.

Fixes [YOCTO #5772]

This file enables creation of permalinks for variables defined
in the glossary.

(Bitbake rev: 772f2bdc10fa242646e4d09f6bdbaf8558f06b22)

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

bitbake: bitbake-user-manual-intro.xml: Edits to "Obtaining BitBake"

Updated the last bullet item that describes a final way to
obtain a copy of BitBake.  The example was too specific regarding
Poky and OE.  So, I made it quite generic and kept with the
convention of not having any specific links from the BitBake
manual into YP docs or such.

(Bitbake rev: a2a3b534871aa23a5d442b93dd9660d5f9e27536)

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

bitbake: bitbake-user-manual-intro.xml: Basic proofreading, grammar fixes.

(Bitbake rev: 266a0b5ef41dcc4048e2a4d1c43567568d7449d7)

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

scripts: fix no newline at end of file

Add a '\n' to the last line of the file to fix:

No newline at end of file

(From meta-yocto rev: 58cc5c502bd0309feed65ab7836ed8fbaab6ced9)

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

groff: Fix build on Fedora 17 for 1.18.1.4

The commit 79fc5576:
[ groff: Fix build on Fedora 17 ]

was applied to 1.22.2, but it's also needed by 1.18.1.4 to solve a same
issue.

(From OE-Core rev: b019f2cdacfb64f149483df8a5b6b7b3f7943222)

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

gawk: fix optional dependency mpfr

 Gawk was failing to compile because of the optional feature mpfr if it found
 libmpfr on build host and linked against that or mpfr existed in sysroot but
 not in deploy, or if gawk was pulled from sstate and someone had mpfr built.

 [YOCTO# 6473]

(From OE-Core rev: ec0def7cc204f402fba017264ea201956add342d)

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

groff: remove redundant code snippet

The exact same code snippet exists both in do_install_append and
do_install_append_class-native which is not necessary.

Remove it from do_install_append_class-native.

(From OE-Core rev: 0f107a76713cebb49004d48fd410d9f014dda4b3)

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

python3: fix builtins imports

The intent of this code is to generate things like -L=/usr/lib/foo so
for paths which start with "/" we should appent the "=". We should not do this
for ".". On some recent versions of binutils or patchsets on some recent
distros this appears to work differently and causes build failures, for example
on Ubuntu 14.04 LTS.

The simple fix is to check for "." as the path prefix.

[YOCTO #6467].

(From OE-Core rev: c36d459f0d40bdbd3ba809835e0475e8992bc778)

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

adt_installer: Fix sed file not found error

We need to look for the installed file within the sysroot we just installed
so add in the missing prefix. This avoids errors like:

Installing cross canadian packages for qemux86 ...
sed: no input files

[YOCTO #6443]

(From OE-Core rev: 3042dcaecadee266b1e4dcae102030c7e4802b89)

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

oeqa/selftest/_toaster.py: fix 'IndentationError' issues

There were some indentation errors when the _toaster.py was merged into master.
This patch fix those errors.

(From OE-Core rev: 173194493294574e49874858ee7003000f41b4d6)

Signed-off-by: Ionut Chisanovici <ionutx.chisanovici@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/toaster.py: renamed to _toaster.py

In order not to run the toaster selftests by default, we renamed to _toaster.py

(From OE-Core rev: 8da2ebba10e0128938919b39c29be40b7c1d80aa)

Signed-off-by: Ionut Chisanovici <ionutx.chisanovici@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Revert "libusb: add missing build dependency on udev"

This reverts commit 6cebce6e355c4c7a87f4088bfa849d0a1c3439da.

Leads to build failures as nativesdk-udev doesn't exist (nor should it).

It also leads to circular dependencies and build failures from these.

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

classes/kernel: Use full path for symlink in update-alternatives.

Use a fully qualified path for the <path> parameter in calls
to update-alternatives.  The chkconfig-alternatives version
requires a full path and without it, the symlink is not
properly created.

(From OE-Core rev: 78ee4d8b1782445caecce8331e68efe83fc32044)

Signed-off-by: Drew Moseley <drew_moseley@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

populate_sdk_base: Remove confusing echo

Remove a confusing echo after the previous commit, thanks for a suggestion
from Laurentiu.

(From OE-Core rev: cd1fdd05cc7457706d1e40042854c154cbb3d9e9)

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

populate_sdk_base: add auto-completion in setup

(From OE-Core rev: d6c55f6910270e395f668c9114ec374b44c57538)

Signed-off-by: Dennis Meier <meier.dennis@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/report-error: tweak summary message

* We don't want everyone to remove their identifying info, just if they
  feel the need to
* Split lines for clarity
* A couple of grammar/spelling tweaks

(From OE-Core rev: b5c7538416e4c7a9e594edf930fa7ee844a347e6)

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

libusb: add missing build dependency on udev

This fixes the ./configure error in case libusb is built before udev/systemd.

(From OE-Core rev: 6cebce6e355c4c7a87f4088bfa849d0a1c3439da)

Signed-off-by: Fabien Chereau <fabien.chereau@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

test-dependencies.sh: Return non-zero return code when there was some failure detected

* this is useful when using from jenkins job where you don't want to read
  output just to dectect how bad it was
* add .log suffix to all files, so they can be easily downloaded from
  http servers without default mimetype set to something useful
* add recipes failed in step 1 to steps 2 and 3 to generate standalone
  logs for them

(From OE-Core rev: cef1d6deb5437edae56740436d8e77b8d941945a)

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

sstate-diff-machines.sh: Return non-zero return code when there was some failure detected

* add --analyze option, which is useful when using from jenkins job
  where you don't want to read output just to dectect how bad it was
* I was always using something like this inside jenkins job, but better
  to share it in original script

(From OE-Core rev: e73e1261879d9154d89cec35669ba22b499d8331)

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

patch.bbclass: unset TMPDIR after use

GNU Patch < 2.6.1 has a race condition so we create a per-instance TMPDIR to
avoid this.  This was implemented by setting os.environ[TMPDIR] but at the end
of do_patch the temporary directory is deleted but TMPDIR is not unset.

In general this doesn't cause a problem but if do_patch is embedded in a larger
function then TMPDIR is set to a directory that doesn't exist.  Avoid this by
removing TMPDIR from os.environ when the directory is deleted.

(From OE-Core rev: 51ea4378864f1468df2ca282a84f78a17d6861aa)

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

recipes-devtools: fix segfault in lib32-gcc with "." multilib_dir

When enabling a lib32-gcc in a 64 bit build, without doing any
other configuration, the mutilib dir is unspecified, which is
represented internally in gcc as "." and as such uncovers an
invalid free on a non-malloc'd pointer.

As suggested by the gcc folks, simply make sure the "." case
is also stored in a malloc'd pointer, so that the intended
runtime behaviour of the code remains unchanged.

Patch has been accepted by upstream maintainers of gcc.

(From OE-Core rev: bf1473d0c1b099b8d919835cc430b99606134aab)

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

initrdscripts: Try to boot if boot label is unknown

The init-live.sh scripts assumes that the boot label set by
the LABELS variable is either "boot", "install", or
"install-efi".  If that variable is overridden to something else
we fall off the end of the case statement and the system locks
up.  If the boot label is unknown, at least attempt to boot.

(From OE-Core rev: 98353862c08be2f1724aaad7aa4ed0521e3621f2)

Signed-off-by: Drew Moseley <drew_moseley@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

distutils/distutils3: Fix bashism

read -d is a bashism. Replace with a direct exec to avoid the problem
in this case. This fixes silent build failures in do_install of
tasks on systems with dash as /bin/sh.

Also merge the fix to distutils for only changing necessary files
to disutils3 as well.

(From OE-Core rev: 7d61661348cf48cbe379ae600565840ea08664b1)

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

u-boot-fw-utils: install config file

The on-target fw_printenv and fw_setenv needs configuration file
(fw_env.config) to be present in the target.

(From OE-Core rev: b4cd268cf7506a521d7800112cb775bd49e1ef09)

Signed-off-by: Maxin B. John <maxin.john@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

createrepo: RDEPENDS on libxml2-python

Fixed the error when run createrepo on target:
  Traceback (most recent call last):
    File "/usr/share/createrepo/genpkgmetadata.py", line 27, in <module>
      import libxml2
    File "/usr/lib64/python2.7/site-packages/libxml2.py", line 1, in <module>
  ImportError: No module named libxml2mod

(From OE-Core rev: 3ec0f411985cf6891b34e1f9624df9df38c9fc6f)

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

sudo: upgrade from 1.8.10p2 to 1.8.10p3

Upgrade sudo from 1.8.10p2 to 1.8.10p3.

[YOCTO #6457]

(From OE-Core rev: acd711f7faa98990f32b11f6235e739723ec2743)

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

initramfs-framework: get rid of udev-utils

Now that udev-utils package has been removed and 'udevadm' has been
moved to udev package, the initramfs-framework should also get rid
of the udev-utils package.

(From OE-Core rev: f0b8f08e13f395820172d16c9d4c016a0fbdbfe9)

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

systemd: remove udev-utils, move 'udevadm' to udev package

The udev package RDEPENDS on the udev-utils package, and the actually
udev-utils package relies on the udev package because if there's no
udev daemon, 'udevadm' command doesn't make much sense.

Now that we have removed udev-utils package and moved 'udevadm' to
the udev package in the older udev recipe, we should do the same
for systemd.

(From OE-Core rev: 393867e8818b54255271bb70bcd7b6ffea137c0a)

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

udev: remove udev-utils, add 'udevadm' to udev package

Previously, the udev package only RRECOMMENDS udev-utils package,
but its init script in the udev package really relies on the
'udevadm' command to work properly. As a result, if 'NO_RECOMMENDATIONS'
is set to '1', errors occur at system start-up.

The udev-utils package only contains one file, that is /bin/udevadm.
So it's better that we remove the udev-utils package and move the
'udevadm' command to the udev package.

[YOCTO #6388]

(From OE-Core rev: 4814a7a85ccff74923a554bdc11820fceff84e2e)

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

udev-cache.default: set PROBE_PLATFORM_BUS to "yes" by default

Force probing of "platform" bus by default.

(From OE-Core rev: 6aae37c66fb6e7153b829ad860b7e7f94e804bd4)

Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev: update init script for conditional probing of platform bus

Make probing of "platform" bus conditional based on PROBE_PLATFORM_BUS
variable from /etc/default/udev-cache on subsequent boots when udev
cache is used. PROBE_PLATFORM_BUS has to be set to "yes" in order to
force probing on every boot, otherwise it uses the old default behaviour
of probing it just the first time.

This is helpful on modern SoCs where most of the low level peripheral
drivers are registered on the platform bus and need to be probed to load
the necessary modules and enable the connected buses and subsystems.

(From OE-Core rev: 70a695735e0a7d14448f2f5a9986bfe105210a91)

Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

own-mirrors: Add gitsm:// mirror support

(From OE-Core rev: 2abc6923956a3ebf8c0a00122ce605b8ec10ceb7)

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

gtk-doc: Add pkgconfig inherit

Anything using gtk-doc will be using pkgconfig as well so
add in the dependency rather than doing it for each and every recipe.

(From OE-Core rev: 7ee05554c55607d9aa02f9a93762e2024bcd4bb0)

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

libsdl: Convert to use pkg-config in internal m4 macros

Rather than code which doesn't even work properly when cross compiling,
lets just use pkg-config instead. Its a little simpler.

(From OE-Core rev: b550572bdad318aed02230496721430eec89c937)

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

gst-openmax: Add missing pkgconfig inherit

(From OE-Core rev: 0f11090e566ae13fe76c4273655db320a87ba7ae)

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

lib/oe/package_manager.py: Add processing for alternative SDK_OS

For the meta-mingw layer, we need to process alternative SDK_OS, since this
is not a Linux based OS.

(From OE-Core rev: b31e015d2e379c24610948d345c5970545887468)

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