rootfs_rpm.bbclass: save rpmlib rather than remove it

The rpmlib was removed when images that add
"remove_packaging_data_files" to ROOTFS_POSTPROCESS_COMMAND, which would
make the increment rpm image generation doesn't work in the second
build, since list_installed_packages would get incorrect value in the
second build, move the rpmlib to ${T} rather than remove it, and move it
back when INC_RPM_IMAGE_GEN =1.

[YOCTO #2440]

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

apt 0.7.14: do_compile failed with gcc 4.7

apt do_compile failed with gcc 4.7:

deb/deblistparser.cc: In member function 'virtual short unsigned int debListParser::VersionHash()':
deb/deblistparser.cc:212:13: error: redeclaration of 'char* I'
deb/deblistparser.cc:202:22: error: 'const char** I' previously declared here

Backport the patch from the upstream would fix the problem, both target and
native apt need it.

[YOCTO #2488]

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

apt-get 0.7.14: hangs at runtime

This patch had a little problem, it made a infinite loop:

for (K = CmdL.FileList + 1; *K != 0; J++)

But the value of "K" will nerver change, so the apt-get hangs, revert
this patch first, and will fix the problem in another patch.

Note: Increment the PR from 11 to 12

Revert "apt: Fix a latent bug exposed by gcc 4.7"

This reverts commit 9d526462fe97564652771812ad1d363e9962b1e3.

[YOCTO #2488]

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

archiver.bbclass: Reduce some duplication for function get_licenses

The content to modify this bbclass is as follow:
- Use the existing functions to get license as a directory instead of
  rewriting it for avoiding code duplication.
- Use SPDXLICENSEMAP to map licenses

[YOCTO #2473]

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

connman: Document the ACL-breaking sed

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

connman: respect the 3g feature and enable/disable ofono support

[YOCTO #2357]

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

insane.bbclass: Make ldd exec_prefix error reporting easier to read

Remove the long obtuse sysroot path from the ldd output.

Make the error message significantly easier to read and understand.

Old Style:
WARNING: QA Issue: keyutils: /home/mhatle/git/oss/oe-core/build-i386/tmp-eglibc/work/i586-oe-linux/keyutils-1.5.5-r1/packages-split/keyutils/sbin/request-key links to something under exec_prefix
WARNING: QA Issue: ldd reports:         libkeyutils.so.1 => /home/mhatle/git/oss/oe-core/build-i386/tmp-eglibc/sysroots/qemux86/usr/lib/libkeyutils.so.1 (0xdead1000)
        libc.so.6 => /home/mhatle/git/oss/oe-core/build-i386/tmp-eglibc/sysroots/qemux86/lib/libc.so.6 (0xdead2000)
        /lib/ld-linux.so.2 => /home/mhatle/git/oss/oe-core/build-i386/tmp-eglibc/sysroots/qemux86/lib/ld-linux.so.2 (0xdead3000)

New style:
WARNING: QA Issue: keyutils: /sbin/request-key, installed in the base_prefix, requires a shared library under exec_prefix (/usr): libkeyutils.so.1 => /usr/lib/libkeyutils.so.1 (0xdead1000)

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

gettext: install libgettextlib.a before removing it

In a multiple job build, Makefile can simultaneously
be installing and removing libgettextlib.a.  We serialize
the operations.

Signed-off-by: Joe Slater <jslater@windriver.com>
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

shared-mime-info: fix build race condition

The definition of install-data-hook in Makefile.am leads
to multiple, overlapping, executions of the install-binPROGRAMS
target.  We modify the definition to avoid that.

Signed-off-by: Joe Slater <jslater@windriver.com>
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

shadow-native: Disable nscd support

Disabling nscd support prevents error messages when shadow utilities attempt
to trigger nscd to reload.  This does nothing unless the user is root, and even
if they are root, it's the wrong action to perform when creating a target fs.

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

groff: Fix build on Fedora 17

Generally distros keep perl at /usr/bin/perl
But Fedora 17 also has /bin/perl,
this causes groff_1.20.1 build to put perl
interpreter path as /bin/perl
But we set perl location for target as /usr/bin/perl

This mismatch of perl path causes failure of rootfs image creation
like this:

| error: Failed dependencies:
|       bin/perl is needed by groff-1.20.1-r1.ppc603e

Signed-off-by: Song.Li <song.li@windriver.com>
Sync up with the do_install_append_virtclass-native chunk.

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

beecrypt: disable java

If java is installed on host, beecrypt will attempt to use it.

Signed-off-by: Jesse Zhang <sen.zhang@windriver.com>
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu-ifup: enable arp proxying

This allows core-image-sato to access the WAN.

Thanks to Dexuan Cui for proposing this fix.

Fixes [YOCTO #2329]

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

gnupg: disable CCID driver

The CCID driver driver is apparently unnecessary, so disable it.

Also remove the associated libusb dependency, since that won't be
needed either.

According to Scott Garman <scott.a.garman@intel.com>:

I'd just note that the CCID smartcard reader is a specific piece of
hardware that is unlikely to be used in a majority of our use cases.

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

gnupg: add libusb to DEPENDS

gnupg apparently depends on libusb:

| error: Failed dependencies:
|   libusb-0.1-4 >= 0.1.3 is needed by gnupg-2.0.18-r1.core2

So add libusb to gnupg DEPENDS.

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

telepathy-glib: update to 0.19.0

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

task-core-basic: drop portmap

Both rpcbind and portmap are RPC port mappers. Having both is redundant. Chose
rpcbind over portmap as rpcbind supports ipv6, nfs4, and builds without the
glibc rpc headers, which have been obsoleted in glibc 2.14.

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

rpcbind: add and use startup script, move binaries

- Add init script from debian, tweaked for us
- Move binaries from bindir to sbindir, as debian does

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

systemtap: Unbreak build for systems with nss headers installed

* discussed on [1]
* patch sent to systemtap [2]

[1] http://lists.linuxtogo.org/pipermail/openembedded-core/2012-June/023377.html
[2] http://sourceware.org/ml/systemtap/2012-q2/msg00202.html

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

perl: Allow perl to cross build and native build in a directory named "t"

If any directory in leading up to your tmp directory has the name "t"
perl will fail to build with a very cryptic error shown below:

pod/buildtoc: no pods at pod/buildtoc line 305.
make[1]: *** [pod/perltoc.pod] Error 255

This is a result of the perl file checking making an assumption
that it is only looking at files and directories with in the perl
source directory.  This assumption fails with the way bitbake
sets up perl to properly cross compile.

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

ltp: Add patch to correct failing build

ltp makefile does not interpret correctly the result of git describe
command and assumes that it is working with a git repo, while in fact
working with a source code archive.
Added a patch to corect makefile system.

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

pulseaudio: fix typo in the patch name, pulseaudo -> pulseaudio

No PR bump is needed.

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

lsbinitscripts: Missing popt DEPENDS

Missing DEPENDS causes build to fail.

Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>

license.bbclass: optimize pkg runtime data

This fixes an ugly way I was trying to find pkg runtime
data for package and license manifest creation.

rootfs generation times for core-image-minimal:

Prior to patch
real 0m41.570s
user 1m40.466s
sys 0m6.768s

With patch
real 0m27.527s
user 0m9.833s
sys 0m3.496s

Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Conflicts:

meta/classes/license.bbclass

license.bbclass: Adding PV to package.manifest

Per request, adding the package version to the package manifest
file.

Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>

dbus: update to version 1.4.20

Unstable version 1.5.12 also supported

Signed-off-by: Radu Moisan <radu.moisan@intel.com>

connman: just give xuser the extra rights it needs

archiver.bbclass: Improve the usability for the archiver classes

The usability of the archiver classes can be improved, beyond the
simple addition of default values for the variables. A user could
well inherit just archiver rather than the individual useful classes,
and not realize it will do nothing.

[YOCTO #2472]

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>

generate-manifest-2.7.py: replace os.popen with os.unlink

The os.popen function would fail (more or less) silently if the executed
program cannot be found, and here what we need is os.system not os.popen
since it doesn't use the return value, use os.unlink() and ignore
exceptions from it would be better as Chris suggested.

[YOCTO #2454]

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

external-sourcery-toolchain: forcibly create usr/lib

If the usr/lib directory doesn't exist, the toolchain can fail to even try to
find crti.o in a completely different directory. This causes a failure for the
case where baselib is lib64.

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

tcmode-external-sourcery: pass -msgxx-glibc for x86

This is needed to work around an issue with the toolchain search paths. It can
pick up the wrong features.h without it, it seems, even with the system32
symlink in the oe sysroot. Investigate this further in the future.

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

external-sourcery: extract src and dest sysroot paths from gcc

Rather than hardcoding the multilib path in a map, and hardcoding dest sysroot
symlink creation in a hook, now we just use -print-sysroot for both, and pass
the appropriate multilib args to the toolchain for particular tunes.

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

task-core-lsb: Add another initscripts to an lsb image

Initscripts with stronger functions will replace the simple one,
which will avoid error when some packages need functions which could
be absent in the simple initscripts.

[YOCTO #2133]

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>

lsbinitscripts: Patch file functions for confirming to an lsb image

Add the condition judgment to functions for avoiding to print error
information when system start up at first.

[YOCTO #2133]

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>

lsbinitscripts: Add initscripts to enhance init functions in an lsb image

The version of initscripts has more functions than the simple.
There could be some errors for current initscripts when running
some programe because of absent some functions provided by initscripts.

[YOCTO #2133]

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>

directfb-examples: upgrade to 1.5.3

Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>

directfb: upgrade to 1.5.3

Also, backported a couple of patches from upstream that fix some
compilations issues for ARM and PPC.

Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>

cpan.bbclass: Fix config error while patches to Makefile.PL

If there is a patch to Makefile.PL, a Makefile.PL but no Makefile
will be placed in ${B}/.pc/xxx.patch/ after do_patch.
And no Makefile will be generated for *this* Makefile.PL.
While do_configure, the original code tries to sed Makefiles
matching with each Makefile.PL in {B}, so this would fail.

Signed-off-by: Xin Ouyang <Xin.Ouyang@windriver.com>

webkit-gtk: Use glib as unicode backend to avoid browser crash

webkit-gtk depends on ICU for the unicode, but ICU is not safe when build and
target system owns different endian. ICU's community is not responsive to make
a patch for this, so glib is used as work around here.

[YOCTO #1570] got fixed

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

bitbake.conf: set an exclusion list for varflags in checksums

Set BB_SIGNATURE_EXCLUDE_FLAGS, which is used by the new varflag
checksum code in BitBake. The list is intended to cut out most of the
varflags that are generated internally, don't influence the output or
are already included in the checksum in some way. If need be this list
can be extended in the future. The existing vardepsexclude mechanism
can also be used to exclude undesired varflags, but they must be fully
specified, e.g.:

do_patch[vardepsexclude] += "do_patch[someflag]"

Implements [YOCTO #2517].

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

bash: Update to new update-alternatives syntax

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

bzip2: Update to new update-alternatives syntax

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

kmod: Remove bashism from recipe

Without this, the installed files get mangled when a dash shell is used.

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

update-alternatives: Fix absolute symlink handling

After the class changes, absolute symlinks are not being handled correctly
by the class file. This adds handling for absolute symlinks to account for the
pkgdest directory, removing dangling symlink messages from recipes like bzip2.

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

package.bbclass: Add warning about FILES containing '//'

'//' in a FILES variable causes hard to track down issues with
packaging. This adds a warning and attempts to auto-correct the issue to
try and make the problem more user friendly.

[YOCTO #2448]

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

layer.conf/bitbake.conf: Tweak PATH

We want the help2man script in scripts/ to be found in preference to any from
the host system and from the native sysroot. It turns out to be tricky to
get the order right from layer.conf so we move the addition of the scripts
directory to bitbake.conf.

Without this, "bitbake libtasn1 -c cleansstate; bitbake help2man-native; bitbake libtasn1"
will fail due to finding the host system help2man before ours.

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

Revert "meta: replace os.popen with subprocess.Popen"

This reverts commit e83d8e58a6b107eea87df0ec233a1bc932b2c6e as the conversion
is not correct. Its replacing readlines() calls which generate an array with
what are effectively strings. There are split("\n") calls missing in many
cases so this needs to be reverted until it gets fixed.

classes/sanity: fix a couple of grammatical errors in messages

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

ccache: Separate out into its own class

Currently, ccache is used if it is present. When building from scratch it gives
no performance improvement and creates a ton of empty directories even when its
not in use.

This change moves ccache support to a bbclass file which the user can choose to
enable. This should make builds more determinstic and make it easier/clearer
to the end user when its being used and when it is not.

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

scripts/cp-noerror: Add a special copy function to fix autotools issues

Currently we copy the aclocal directory to the build so that autotools
doesn't see .m4 files disappear when its processing them. This can happen
if for example, package X is being rebuilt at the same time as Y and it
gets uninstalled from sstate (assuming there are no dependencies between
X and Y). This code making the copy was added to avoid races but introduces
a race of its own, namely that the files can disappear during the copy.

This patch adds a cp-noerror script which silently ignores such errors
and gives the behaviour we need in this case. It hence fixes issues which
crop up for users and the autobuilder occasionally.

[YOCTO #2485]

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

xz: updated to version 5.1.1alpha

The licenses are the same, only some white spaces added/removed.

Signed-off-by: Valentin Popa <valentin.popa@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemu: Add an option to remove host sdl/gl checking

Add an PACKAGECONFIG in qemu to disable GL acceleration:
* By default configure try best to enable GL acceleration and fail when missing
  host dependency(libSDL and libGL).
* End user can also choose to turn off GL capability, thus remove the host
  dependence in building.

[YOCTO #2407] got fixed.

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

classes/sanity: send sanity check failure as a separate event for Hob

In order to show a friendlier error message within Hob that does not
bury the actual sanity error in our typical preamble about disabling
sanity checks, use a separate event to indicate that sanity checks
failed.

This change is intended to work together with the related change to
BitBake, however it has a check to ensure that it does not fail with
older versions that do not include that change.

Fixes [YOCTO #2336].

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

foomatic-filters: update to version 4.0.15

Signed-off-by: Radu Moisan <radu.moisan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mtd-utils: Update to version 1.5.0

Signed-off-by: Radu Moisan <radu.moisan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

man-pages: update to version 3.41

Signed-off-by: Radu Moisan <radu.moisan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/sanity: remove broken TARGET_ARCH check

The variable name has been typo'd as TARGE_ARCH since it was introduced
some time ago, so the check has never worked. Fixing the typo shows that
the test is not quite functional, so let's just remove it:

http://lists.linuxtogo.org/pipermail/openembedded-core/2011-October/010613.html

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

libpciaccess: upgrade to 0.13.1

Removed also the fix_deletion_of_last_handle.patch because the new
version includes it.

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

fontsproto: upgrade to 2.1.2

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

fontconfig: upgrade to 2.9.0

Other changes:
- licence checksums because the licence headers changed slightly:
  several other authors were added and Keith Packard's name has been
  replaced with "author(s)". That's it.
- rebased the fix-pkgconfig.patch to apply on 2.9.0
- removed all sed instructions in do_configure_append() that were
  changing fc-arch/Makefile. In the new version the fc-arch/ directory
  has been removed in order to make the package cross-compiling safe

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

scripts/buildhistory-diff: add GitPython version check

Display an error if the user does not have at least version 0.3.1 of
GitPython installed.

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

buildhistory_analysis: fix error when version specifier missing

Passing None to split_versions() will raise an exception, so check that
the version is specified before passing it in.

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

meta: replace os.popen with subprocess.Popen

Replace os.popen with subprocess.Popen since the older function would
fail (more or less) silently if the executed program cannot be found

There are both bb.process.run() and bb.process.Popen() which wraps the
subprocess module, use it for simplifying the code.

Note: We don't need the "2>/dev/null" or "2>&1" since bb.process.run()
can handle it, it will raise exception when error occurs, we should
handle the exception ourselves if we want to ignore the error.

More info:
http://docs.python.org/library/subprocess.html#subprocess-replacements

[YOCTO #2454]

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

scripts: replace os.system with subprocess.call

Replace os.system with subprocess.call since the older function would
fail (more or less) silently if the executed program cannot be found

More info:
http://docs.python.org/library/subprocess.html#subprocess-replacements

[YOCTO #2454]

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

meta: replace os.system with subprocess.call

Replace os.system with subprocess.call since the older function would
fail (more or less) silently if the executed program cannot be found

More info:
http://docs.python.org/library/subprocess.html#subprocess-replacements

[YOCTO #2454]

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

shadow-utils-native: remove unnecessary alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

kmod: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

sysklogd: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

psmisc: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

iputils: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

grep: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

cpio: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

opkg: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

tar: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

binutils: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

util-linux: use new update-alternatives

In addition to switching to the new update-alternatives, it was discovered
that the sln, shutdown, halt, reboot and related commands were never generated.
Remove these from the list.  (These appear to have been removed from util-linux
some time ago.)

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

psplash: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

ncurses: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

dropbear: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

coreutils: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

busybox: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

openssh: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

lrzsz: use new update-alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

sysvinit: use new update-alternatives

Also remove postinst and prerm which were not being used.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

tinylogin: use BPN for alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

patch: use BPN for alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

less: use BPN for alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

mktemp: use BPN for alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

sed: use BPN for alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

shadow: use BPN for alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

time: use BPN for alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

wget: use BPN for alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

which: use BPN for alternatives

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

update-alternatives.bbclass: Change from using PN to BPN

In order to better support multilib processing, switch from PN
to BPN.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

update-alternatives.bbclass: Refactor the implementation

Refactor in order to:
  * Deprecate the old interfaces, but keep them for compatibility
  * Provide a new, interface -- capable of working with split packages
  * Each update-alternative will now set proper "per-file" provides

Note: this adds a warning message when the older deprecated behavior is
used.  The older behavior has been fully tested using oe-core.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

image.bbclass: Ensure ${S} is cleaned at the start of rootfs generation

Some image classes such as bootimg save files into ${S} as part of rootfs
generation. For correctness we should therefore clean this at the start of
image generation to ensure reproducibility.

I found this issue when some files I thought should disappear from my rootfs
would not disappear.

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

glib-2.0: Clean up OECONF flags into one variable

Without this change, libelf was "floating" for linuxstdbase for example,
leading to build failures. This patch moves all the common options
into a common variable.

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

libxml2/libxslt: Don't depend on ansidecl.h header

We don't DEPEND on binutils for ansidecl.h so ensure we should never
use the header. This makes builds determinstic and means something like:

bitbake binutils
bitbake libxml2 -c configure
bitbake binutils -c clean
bitbake libxml2

doen't fail to build.

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

ltp: Add attr to DEPENDS list

Signed-off-by: Saul Wold <sgw@linux.intel.com>

binutils: Drop old/stale/broken csl recipe

The patches this recipe refers to don't exist so nobody can have
been using this so lets just remove it.

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