coreutils_6.9: Update the older coreutils to no longer provide 'su'

'su' was removed from newer versions of coreutils, and in OE-Core
we pick up 'su' from shadow instead. So simply remove the binary
from the install image to match expected behavior and avoid failures
due to conflicting 'su' binaries during do_rootfs when building
non-GPLv3 images.

(From OE-Core rev: b4c5fc7c025ff864280be9678e6dd856971c71d7)

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

rootfs_rpm.bbclass: Revert to default shell logging behavior

Remove the set +x, and revert to the normal logging behavior.

(From OE-Core rev: e799d771c9ed9e41a900e2e96c23d4ae35d5fe08)

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

bootimg: Use FAT 32 for images larger than 512MB

Fixes [YOCTO #2138]

Commit 217584211625b1c496fe5b78aa4765ccf605d2b9 dropped the forced use
of FAT32 for the hddimg generation as it broke with very small images
(< 32MB). Unfortunately, left to its own devices, mkdosfs appears to select
FAT16 even for very large images, resulting in 2.2GB images being
generated as FAT16:

$ ls -lah core-image-lsb-sdk-atom-pc-20121010233936.hddimg
-rw-rw-r-- 1 dvhart dvhart 2.2G 2012-10-17 08:00 core-image-lsb-sdk-atom-pc-20121010233936.hddimg

$ file !$
file core-image-lsb-sdk-atom-pc-20121010233936.hddimg
core-image-lsb-sdk-atom-pc-20121010233936.hddimg: x86 boot sector, code offset 0x58, OEM-ID "SYSLINUX", sectors/cluster 128, root entries 512, Media descriptor 0xf8, sectors/FAT 138, heads 64, sectors 4502496 (volumes > 32 MB) , serial number 0x50761926, label: "boot       ", FAT (16 bit)

The result was a runtime boot error from SYSLINUX and a failure to boot
live images greater than 1GB in size.

While strictly speaking it is the cluster count that determines which
FAT size is used, that calculation requires more information than we
have readily available (such as sectors per cluster). If we let mkdosfs
determine sectors per cluster and just set a sane threshold above which
FAT32 is used, we get correct bootable images. With this patch the 2.2GB
core-image-lsb-sdk uses FAT32 and the 21 MB core-image-minimal uses
FAT16, and both boot in qemu successfully:

$ ls -lah tmp/deploy/images/core-image-lsb-sdk-atom-pc-20121212220835.hddimg
-rw-r--r-- 1 dvhart dvhart 2.2G 2012-12-12 14:18 tmp/deploy/images/core-image-lsb-sdk-atom-pc-20121212220835.hddimg

$ file !$
file tmp/deploy/images/core-image-lsb-sdk-atom-pc-20121212220835.hddimg
tmp/deploy/images/core-image-lsb-sdk-atom-pc-20121212220835.hddimg: x86 boot sector, code offset 0x58, OEM-ID "SYSLINUX", sectors/cluster 8, Media descriptor 0xf8, heads 64, sectors 4470304 (volumes > 32 MB) , FAT (32 bit), sectors/FAT 4357, reserved3 0x800000, serial number 0x50c902b7, label: "boot       "

$ ls -lah tmp/deploy/images/core-image-minimal-atom-pc-20121212220600.hddimg
-rw-r--r-- 1 dvhart dvhart 21M 2012-12-12 14:06 tmp/deploy/images/core-image-minimal-atom-pc-20121212220600.hddimg

$ file !$
file tmp/deploy/images/core-image-minimal-atom-pc-20121212220600.hddimg
tmp/deploy/images/core-image-minimal-atom-pc-20121212220600.hddimg: x86 boot sector, code offset 0x58, OEM-ID "SYSLINUX", sectors/cluster 4, root entries 512, sectors 41408 (volumes <=32 MB) , Media descriptor 0xf8, sectors/FAT 41, heads 64, serial number 0x50c8ffec, label: "boot       ", FAT (16 bit)

I have tested and booted core-image-minimal and core-image-lsb-sdk for
atom-pc with qemu-system-i386 using this patch.

(From OE-Core rev: de808c552d445502bd99f78bb8159d21149f87c1)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Cc: Steve Sakoman <steve@sakoman.com>
Cc: Joshua Immanuel <josh@hipro.co.in>
Cc: Przemek Czesnowicz <przemyslawx.czesnowicz@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

eglibc: remove dependency of eglibc-utils on bash

Adapt a patch from Peter Seebach <peter.seebach@windriver.com> to remove
the non-POSIX elements from the tzselect script, and add a separate
patch to work around a bug in the current version of busybox's awk
command. This replaces the /bin/bash reference in the script header with
/bin/sh and thus eliminates the dependency on bash picked up during
packaging.

Fixes [YOCTO #3551].

(From OE-Core rev: fe19d0e01cb1563cf4735ef250f80af20059103b)

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

tar: apply fedora patch for fortify sources sigabort

See https://qa.mandriva.com/show_bug.cgi?id=61419 for details.

(From OE-Core rev: 648cda68def82fa12b3af599c06caecdc7568668)

Signed-off-by: Christopher Larson <chris_larson@mentor.com>
the patch was imported from meta-mentor layer on yoctoproject git server
http://git.yoctoproject.org/cgit/cgit.cgi/meta-mentor as of commit id
333d2e0510a1e052cb83a6f8beed6d8bcea59b2c

Signed-off-by: Fahad Usman <fahad_usman@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

matchbox-keyboard: fix the build with libpng15

(From OE-Core rev: d49facde4fffc95cd153195fc0114dfbc1bf512c)

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

libmatchbox: upgrade to 1.10, drop git

All patches have been upstreamed so drop them, and as upstream isn't heavily
developed drop the git recipe.

(From OE-Core rev: 3465570601480d2e476e82b8b7254e94f87d2682)

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

libpng: enable nativesdk variant

(From OE-Core rev: bdc9e932869cc5d08761c762c003b42de21e5c1f)

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

create-pull-request: detect trailing white space

Add logic in the create-pull-request to detect and warn about the
trailing white space inserted by patches.

(From OE-Core rev: 3e35310db3cd6d265092724a0e542b9616aca9c5)

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

eglibc: add AArch64 support

(From OE-Core rev: 72087d2cbdcc0da66b38720769d8ad52dbfb4abc)

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

gdk-pixbuf: handle postinstall errors differently

Apparently gdk-pixbuf-query-loaders always returns 0 but writes any
errors to stderr. So, the conditional exit at the end of the command was
useless. Write the errors to a file and exit only if the file size is
greater than 0.

(From OE-Core rev: 901fbfac48918db7bf809ded3df533504141c5ca)

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

populate_sdk_(deb|ipk|rpm): export NATIVE_ROOT and INTERCEPT_DIR variables

In order for the postinstall scripts to have access to the recent
improvements when generating the target SDK sysroot, export these
variables in populate_sdk.

(From OE-Core rev: 06151c34f4f4cb669b2f93ebb2f78cfecf698355)

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

telepathy-glib: remove pkgconfig dependency in pkgconfig

A pkg-config file depending on pkg-config doesn't really make sense, and this is
causing unexpected dependencies on pkg-config at image contruction time (see

(From OE-Core rev: 1024f59c78ebdec132fde6a2d7eccde4b03ba468)

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

libpcre: Improve the recipe

- Improve the cross patch to determine *_FOR_BUILD in configure

- Leverage the fact that autotools.bbclass handles *_FOR_BUILD now

- Drop the now unnecessary do_compile override

- Drop the do_split_packages

- Split out packages for pcregrep, pcretest, in addition to libpcrecpp and
libpcreposix

- add nativesdk to BBCLASSEXTEND as well as native

(From OE-Core rev: 9867a9ad7a80a26654326650752dbbe89c8de3c5)

Signed-off-by: Christopher Larson <chris_larson@mentor.com>
- the patch was imported from meta-mentor layer on yoctoproject git server
http://git.yoctoproject.org/cgit/cgit.cgi/meta-mentor as of commit id
62f819b97e356c1d7468fbccbf3a3ceed78b09a8

- broken up the original patch in two parts, this part contains some
improvements to the recipe and the cross patch

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

libpcre: upgrade to 8.32

- Also improve description
- Add summary and homepage

(From OE-Core rev: 57907458fe4b6c7037f35862adabf29b8ac1f029)

Signed-off-by: Christopher Larson <chris_larson@mentor.com>
- the patch was imported from meta-mentor layer on yoctoproject git server
http://git.yoctoproject.org/cgit/cgit.cgi/meta-mentor as of commit id
62f819b97e356c1d7468fbccbf3a3ceed78b09a8

- broken up the original patch in two parts, this first part upgrades the
version to 8.32

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

libuser: enable python support

Compile libuser with python support and add python to DEPENDS.

(From OE-Core rev: 57114b3f631b66edf136bcc7ba23eec7c7bc9970)

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

qemu: upgrade to 1.3.0

remove-hardcoded-libexec.patch: removed
    - included in the new version

qemu-vmware-vga-depth.patch: removed
    - doesn't apply anymore
    - the problem addressed by the patch is fixed by
reverting commit 1f202568e0553b416483e5993f1bde219c22cf72

Revert-vmware_vga-Add-back-some-info-in-local-state-.patch:
    - VMware VGA requires that the depth presented to the guest
to be the same as the Display Surface depth in order to do not
corrupt the display
    - do not cache the DS depth (the depth might change)
    - revert commit 1f202568

- QEMU now uses pixman (DEPENDS += "pixman")
- rearrange the recipe in order to mimic the bitbake flow
- update both variants (.tar.bz2, git)

(From OE-Core rev: 7622c4f6c050f26f252066a0fcaacdbf340dcefa)

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

pixman: upgrade to 0.28.0

* Generic-C-implementation-of-pixman_blt-with-overlapp.patch:
    - adapted to the new version
* enable nativesdk variant

(From OE-Core rev: b41e55a7ee226a0ae4efdd633cab94e1cc846525)

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

documentation-audit.sh: Add/fix checks for skipped lines.

Have the script skip:

  * "Recipe" lines.
  * Lines with 11 "=", not 12.

(From OE-Core rev: b9f14425a3f975a40a881b868b7bddcbd3c22580)

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: BBhandler/data: Fix __inherit_cache duplication

The inherits cache contains duplicate entries, some with the full patch, some
just starting classes/xxx. This is a waste of parse time and potentially
error prone. This patch fixes various pieces of code so the absolute paths are
always preferred and work correctly. The inherits_class function did not work
with full paths so the patch fixes this.

(Bitbake rev: f3a71e509af196391ec126d079cf1bd178e62ad5)

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

bitbake: BBHandler/ast: Simplify/fix EXPORT_FUNCTIONS usage

The current usage of EXPORT_FUNCTIONS is rather problematic since a class
list (classes) is passed into the ast statement and cached as it was
when first parsed. This class list may be different in other cases but
is locked once in the cache.

Worse, the construction of classes can be broken by exceptions during parsing
at the wrong moments since the state of the parser is not always reset
correctly. This can lead to leakage of other classes into the classes list.

The current EXPORT_FUNCTIONS implementation looks at the last two currently
inherited classes and sets up an indirect function call view the second last
class inherited, e.g.:

 do_configure calls gnomebase_do_configure
 gnomebase_do_configure calls autotools_do_configure

This intermediary doesn't seem to serve a useful purpose.

This patch therefore makes builds deterministic and fixes various cache problems
and indirection by removing the intermediaries and simply performing
directly mapping for the cases where its needed.

(Bitbake rev: 9fc98f96f0e0320beda0ce9546275a99336732c1)

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

bitbake.conf/utils: Drop some OVERRIDES from FILESPATH

There are several overrides that don't make sense as part of FILESPATH. This
introduces FILESOVERRIDES and allows us to drop some of the pointless ones,
simplifying the files search path further and improving the user experience.

If needed by specific recipes, other overrides can be added back in for
specific cases.

(From OE-Core rev: b8b1b39961332c99d62ee466f7859bd62a0f806f)

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

x11vnc: bind vnc server to port 5900

At the moment, if one clicks on X11VNC Server icon, a new vnc server is
started on an automatically assigned port (starting with 5900). The
problem is that the user can click multiple times on the icon and a new
vnc server is started on another port. So, we'll end up with multiple
vnc servers, wasting memory.

This patch will restrict the x11vnc server port to 5900 (default VNC
port) and another process will not be started (because the port is
already used), unless the user chooses to start the server manually,
from the command line, in which case the port can be changed as the
user wishes.

[YOCTO #3369]

(From OE-Core rev: 8f7bd81f4769984a5acdb40f3a76e290615c3020)

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

package.bbclass: prepend MLPREFIX to LOCALEBASEPN

We need to prepend MLPREFIX to LOCALEBASEPN in order to
fully enable multilib

[YOCTO #3440]

(From OE-Core rev: 91fe09a9a8b317d27145efff4201f1138d6a45d4)

Signed-off-by: Constantin Musca <constantinx.musca@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

multilib.bbclass: fix do_package_qa_multilib

The packages which start with "rtld" are ok

[YOCTO #3440]

(From OE-Core rev: 1bb3f44065d0470dd2f6950e267ef991c2ce6fd5)

Signed-off-by: Constantin Musca <constantinx.musca@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

build-appliance: Updated to include bblayers.conf.sample fix

In Build Appliance, Hob didn't start because it crashed
checking bblayers.conf

(From OE-Core rev: fb04d8e6404c7ee35b636cf2cb886acfba22b8d7)

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

build-appliance: provides git proxified acces

Inside Build Appliance, external repos can be
accesed using the git protocol through a proxy

Fixes [YOCTO #3175]

(From OE-Core rev: 938bfe375b6fca3c9281f3c906fba1aca6677aca)

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

mesa-dri: upgrade -git version to current master

license.html was reformated so the checksum changed.

The GLSL compiler can be built using CC_FOR_BUILD, so we don't need
mesa-dri-glsl-native.

Update common packaging to include libdricore and the skeleton libgles3, and
remove the driver .la files.

mesa-git/uclibc.patch isn't applied, remove.

x32 support is integrated into mklib, drop 0003-fix-for-x32.patch.

uclibc is checked for upstream, drop 0001-Compile-with-uclibc.patch.

(From OE-Core rev: 8060a24c679ba17aea48bdeb4b8cfd460885f65f)

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

mesa: upgrade to 8.0.5

Licence checksum change, use an upstream source file instead of a generated
file: we were previously checksumming the Bison license.

(From OE-Core rev: b85388ad0b6c3e7176295949d88c504abf0ba5cc)

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

mesa: common packaging cleanup

mesa-dri ships the configuration file in mesa 9, so ship that and don't claim
it's empty.   Clean up libegl and libgbm packaging.

(From OE-Core rev: 22549ed65f0e413d7f4375e5eba7c86302a46acd)

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

mtd-utils:fix corrupt cleanmarker with flash_erase -j command

     *Flash_erase -j should fill discrete freeoob areas with required bytes
     of JFFS2 cleanmarker in jffs2_check_nand_cleanmarker(). Not just fill
     the first freeoob area.

(From OE-Core rev: c8e3fe46df84e782cc1becc4c5a8cd954ef85f34)

Signed-off-by: Chunrong Guo <b40290@freescale.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nativesdk-qt4-tools: fix do_configure to be reentrant

* without = it can add g++ prefix multiple time when you reexecute do_configure (e.g. after failure)

(From OE-Core rev: 4f891901bcf7be95cebca1a7296e7ea656adee74)

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

qt4: upgrade to 4.8.4

* changes in licenses are only s/Nokia/Digia/

(From OE-Core rev: 5d898695ab3ce368a1d1f2066bb980ee829ae077)

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

package.bbclass: make do_split_packages handle non-existent root directories

This function has different behaviour if the split directory doesn't exist
depending on the recursive argument: non-recursive uses os.listdirs which throws
an exception, recursive uses os.walk which doesn't.

do_split_packages should silently handle non-existent directories because it's
mainly used for plugin directories, which may end up being empty though changing
the distro configuration (for example, connman without wifi distro feature).

So, add an early exit if the split root doesn't exist.

(From OE-Core rev: 937101e3fdd6afd00f6f8a8be411a67110c4ae78)

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

base/license.bbclass: handle multiple INCOMPATIBLE_LICENSEs

Allow INCOMPATIBLE_LICENSE to be a whitespace-separated list of
incompatible license strings and/or glob patterns.

Also fix wildcarding: the string in INCOMPATIBLE_LICENSE was clearly
intended to match with wildcards (e.g. "*GPLv3" to match both GPLv3
and LGPLv3), but this was broken because of a bug in return_spdx()
which would die with a runtime error when there was no SPDXLICENSEMAP
entry for the string.

(From OE-Core rev: 8a8d00f4c9f7fe5f9f173b43b86cd18a6c75435c)

Signed-off-by: Andy Ross <andy.ross@windriver.com>
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

populate_sdk_deb.bbclass: put back INSTALL_PACKAGES_NORMAL_DEB

This commit (0b0674ebcd8a51783e4bb38e3ead3e419dbba376) was also changing
the populate_sdk_deb.bbclass but the deb backend does not support
multilib just yet. So, when compiling the standalone toolchain tarball,
we ended up with an empty tarball if the deb package manager was chosen.

Reverting the deb part until the multilib support is added to deb.

[YOCTO #3532]

(From OE-Core rev: 34baee37caccced13fb26f446b865c9f25edf339)

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

hob-toolchain: Update for removal of dummy classes

Without this hob will fail to parse.

(From OE-Core rev: b63ba9104584c5aa15792fcfe706d59d39235ad0)

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

bitbake: Revert "BBHandler: Ensure parser state engine is correctly reset for new parsing"

This reverts commit 3dc5e896a97154914cee6c47900e3bb2a2627cdb.

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

base.bbclass: Remove implicit dependency on license.bbclass

(From OE-Core rev: 28c06a32201351894546b3095f80daa84acf7b39)

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

rt-tests: added missing dependencies in Makefile

The following targets missed dependency on librttest.a:
pi_stress
rt-migrate-test
hackbench

[YOCTO #3549]

(From OE-Core rev: 44bdd58494aeb7cef14b7931ba6f6628849f2c19)

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

bitbake: BBHandler: Ensure parser state engine is correctly reset for new parsing

The classes variable in particular could be in an odd state if parsing of a
previous recipe had abruptly ended.

(Bitbake rev: 3dc5e896a97154914cee6c47900e3bb2a2627cdb)

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

rootfs_ipk/package_ipk: Simplify opkg commandline option variables

The current maze of IPKG_ARGS variables is hard to understand and worse,
not even used in some places. This simplifies the code and specifies the
options in one place.

(From OE-Core rev: 5e482b278c3775b0acecc621cf2d3b90ee516f26)

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

image/populate_sdk: Simplify imagetest inclusion and drop dummy class file

Having to include some dummy class is suboptimal and we no longer need to
do this. Also move this check to populate_sdk_base since we then don't
need to include it in toolchains specifically.

(From OE-Core rev: 2ee979ff71d31e497be856ea1443667c6d799d34)

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

bootimg: Remove now unnecessary dummy inherit usage

bitbake now supports empty expansions for inherit usage so we can simplify
these statements.

(From OE-Core rev: 77cd2ef06bdf701b047c4f8c817b364b8b4b8837)

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

license.bbclass: Including locale packages

As locale packages are installed on the image, we should
be including them in the package/license manifest.

This ensures that the manifests are accurate and complete.

[ YOCTO #2461 ]

(From OE-Core rev: 4578b708ada1ffb99ab542da793977ffb90a9b50)

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

update-alternatives.bbclass: Add missing runtime dependency

When using update-alternatives, there should be a runtime dependency on
${VIRTUAL-RUNTIME_update-alternatives}.  Without this, it's possible to
get into a situation where the package is not installable.

(If VIRTUAL-RUNTIME_update-alternatives is not defined, no requirement
is added.)

(From OE-Core rev: 8a4c5765fb56d17c3d82ff6db47863e934c9fcbe)

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

sstate: fix typo

(From OE-Core rev: 17d27fd6e2750e6455f656159da56c3350c4b3cd)

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

sstate: fix typo

(From OE-Core rev: 10cceb3854acd5c619d81011bac02da11ed580d2)

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

git: fix unpackaged python files

* EXTRA_OEMAKE is needed after inheriting autotools which sets it to empty
* fixes:
git-1.7.7: git: Files/directories were installed but not shipped
  /usr/lib64
  /usr/lib64/python2.7
  /usr/lib64/python2.7/site-packages
  /usr/lib64/python2.7/site-packages/git_remote_helpers-0.1.0-py2.7.egg-info
  /usr/lib64/python2.7/site-packages/git_remote_helpers
  /usr/lib64/python2.7/site-packages/git_remote_helpers/__init__.py
  /usr/lib64/python2.7/site-packages/git_remote_helpers/util.pyc
  /usr/lib64/python2.7/site-packages/git_remote_helpers/__init__.pyc
  /usr/lib64/python2.7/site-packages/git_remote_helpers/util.py
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/__init__.py
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/importer.pyc
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/non_local.pyc
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/exporter.py
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/git.py
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/repo.py
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/__init__.pyc
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/importer.py
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/exporter.pyc
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/repo.pyc
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/git.pyc
  /usr/lib64/python2.7/site-packages/git_remote_helpers/git/non_local.py

* fixes [YOCTO #3344]

(From OE-Core rev: 8231e9d455c71fe1bd3c4776985ff775c6af3f6d)

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

populate_sdk_base, adt_installer_internal: fix issue on older distributions

This patch addresses an SDK installer issue on older distributions if
the installation takes place in another directory than the default one. In
fact, is all about the 'file' utility version used for determining if a
file is a text file. For example, for a perl script, newer 'file' versions
return:
"...script, ASCII text executable"
while older versions return:
"...script text executable"
Hence the regex pattern didn't match the scripts.

Also the patch contains two unrelated minor fixes:
 * return an exit code of 1 instead of -1 when installation machine is
   not supported. That because on an older distribution we also get this
   error message: "exit: 9: Illegal number: -1";
 * remove unnecessary $SUDO_EXEC prefix to grep;

[YOCTO #3538]

(From OE-Core rev: e7e23a1e5aae4028e21e37ec09e9d431a9adfbcb)

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

buildhistory_analysis: fix broken list length checks

Fix erroneous use of .count instead of len(), which unfortunately is not
reported by Python as an error in a numeric comparison.

(From OE-Core rev: 63fd76190f503660119dcc8efdcfc6fbff406c26)

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

sstate: fetch .siginfo files from SSTATE_MIRROR

This would be useful for doing siginfo compares to understand why a build
is not reusing something when using SSTATE_MIRROR. No error will be reported
if it fails to find the .siginfo file

[YOCTO #2898]

[RP: Small tweaks]
(From OE-Core rev: 6d86690330f0d43839b904fced4b4b02cb27b8c6)

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

util-linux: Package systemd files correctly

2.21 has some linking issues with gold on uclibc where mount
program does not work. Its some sort of underlinking.

2.22 did not solve that problem completely either

(From OE-Core rev: 61e7de12a8e735c36ca3285f6a77d7b7cb863207)

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

Added regex functionality to distrodata.bbclass

In order to make the PRS use regexes for the versioning
of packages who have the distros on sourceforge and
for those who have faulty reports.

(From OE-Core rev: 2fe52b23c643d2125867dc9fcc9c01a184a9e238)

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

util-linux: rebase remove-lscpu patch

(From OE-Core rev: b7c343fc4ae8c65cdc3e12e5b75257e3780bd2a0)

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

qt4-native & ghostscript: update upstream status for their patches

Update the upstream status for patches of long building path.

(From OE-Core rev: 58b83f7dad8f0d252f8028349d734987081b7507)

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

zip: Remove QA warnings: No GNU_HASH in the elf binary

Update the LFLAGS1 in the Makefile to remove warnings:
QA Issue: No GNU_HASH in the elf binary

(From OE-Core rev: 3b41513a63df6636b617eff09ff1ea9574c5e6f0)

Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Muhammad Shakeel <muhammad_shakeel@mentor.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

babeltrace: upgrade to 1.0.0

License checksum change due to diff:
15a16,22
>
> * LGPLv2.1
>
> The file include/babeltrace/list.h is licensed under LGPLv2.1. It only
> contains trivial static inline functions and macros, and, therefore,
> including it does not make babeltrace a derivative work on this header.
> Please refer to the LGPLv2.1 license for details.

(From OE-Core rev: 275ea500809e70ba2e432b950311678b2ff4aef6)

Signed-off-by: Constantin Musca <constantinx.musca@intel.com>
Signed-off-by: Radu Moisan <radu.moisan@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

liburcu: upgrade to 0.7.4

License checksum change due to diff:
3c3
< May 10, 2009
> May 15, 2012
59a60,65
>
> * GPLv3 (or later)
>
> The following build-related macro is under GPLv3 (or later):
>
> m4/ax_tls.m4

(From OE-Core rev: dfff77b22880a5078dd1b7fe0d3b4d0145f28545)

Signed-off-by: Constantin Musca <constantinx.musca@intel.com>
Signed-off-by: Radu Moisan <radu.moisan@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

prs: use the PRServer to replace the BB_URI_LOCALCOUNT functionality

- adapt the PRServer to generate an incremental number for every revision set
associated with a (PN, PV)
- use one incrementing value rather than several
- PV becomes 0.1+gitAUTOINC+deadbeefdecafbad_decafbaddeadbeef
- if PV contains 'AUTOINC' set PKGV to the same thing with AUTOINC replaced
with the "auto PR" value
- if the PRServer isn't running replace AUTOINC with 0 (no package feeds, but usual
image creation would work)
- use the PR table with "AUTOINC-${PV-without-SRCREVs}" as key
- modify prexport/primport to address the generated srcrev entries
- this patch addresses the following proposal:
http://lists.linuxtogo.org/pipermail/bitbake-devel/2012-November/003878.html

(From OE-Core rev: b27b438221e16ac3df6ac66d761b77e3bd43db67)

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

nasm: upgrade to 2.10.05

(From OE-Core rev: eebd134989b32577cb1e59897de25c320e61f03c)

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

libconvert-asn1-perl: upgrade to 0.26

License md5sum change due to:
- README is removed
- README.md contains license info

(From OE-Core rev: 3c93065cfb53edea920469c8bdfde1dcbf61ff2e)

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

libaio: Remove QA warning: No GNU_HASH in the elf binary

Update the LINK_FLAGS in the Makefile to remove warning:
QA Issue: No GNU_HASH in the elf binary

(From OE-Core rev: 07501f14121a1882f26de66b4ca991392ab45dff)

Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Muhammad Shakeel <muhammad_shakeel@mentor.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

quicky: tidy up inherit

* inherit qt4x11 which inherits qmake2 and sets DEPENDS already.
* Remove pkgconfig from inherit since quicky doesn't actually use
  pkg-config nor does it produce a .pc file

(From OE-Core rev: 62db8745c1b907ebec222751cd9cb174dd5f2f96)

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

python-smartpm: don't try to remove /usr/share/share twice

* distutils.bbclass now removes that directory

(From OE-Core rev: 5990bda7e0c21629ac5ecaa2ba9783af5a77a125)

Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Conflicts:
meta/recipes-devtools/python/python-smartpm_1.4.1.bb
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

at: change initscript dependency from build time to runtime

* doesn't look like build dependency

(From OE-Core rev: de02e36ff17a44ce8cf33c59a78360cb2313f016)

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

classes/binconfig.bbclass: Allow packages to add sed expressions

packages/apr/apr-util,apr use this, but it not been in the class

patch the config files to be 'installed' and not point to the build
and source-directory.

From OE-Classic:
http://git.openembedded.org/openembedded/commit/?id=030992bdd415c8892cf2925cf6fbe888226be046

[YOCTO #3267]

Orignal Author: Holger Freyther <zecke@selfish.org>
(From OE-Core rev: ab56f1063a666df5faddd97f44758c8ae378e51c)

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

libpcap: PR bump because of PU for libnl

(From OE-Core rev: ebcd909451ae6a2d00b83f6aa7f28794355a6935)

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

wpa-supplicant: configured to work with libnl 3.2.14

* Updated the configuration file for libnl-3
* Added python include dir path to configuration file

(From OE-Core rev: e328400ec02300be1ce3c8a9f63d8b30f9fe2136)

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

libnl: upgrade to 3.2.14

* Based on the 3.2.9 recipe from meta-oe
* Modified the patches to work with the new name of the library file (libnl-3.so)

(From OE-Core rev: a343ced372ba659b3388a09ce615ee9a9127d921)

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

qa.py: Toolchain path should not be hardcoded to execute objdump

Currently objdump command is invoked from the STAGING_BINDIR_TOOLCHAIN
directory. In case of external toolchain if this directory doesn't
exist then objdump fails to execute. Instead of hardcoding the path
it should search PATH to find it.

(From OE-Core rev: a3a0e29d5d796fdf6755503cc31e45199b79d756)

Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Muhammad Shakeel <muhammad_shakeel@mentor.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libassuan: add pkgconfig support

Add pkgconfig support to libassuan.

[YOCTO #2453]

(From OE-Core rev: ef2951d052f18be9c5391dce438f5c3fafda3cd5)

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

pth: add pkgconfig support

Add pkgconfig support to pth.

[YOCTO #2453]

(From OE-Core rev: c5f14ef7797394bcee1b0e295cc2dac4b62e55e2)

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

ICU: add pkgconfig support

Add pkgconfig support to ICU.

[YOCTO #2453]

(From OE-Core rev: a0b553bf00f819c3f0d6744461189179f578a75e)

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

libksba: add pkgconfig support

Add pkgconfig support to libksba.

[YOCTO #2453]

(From OE-Core rev: d155ab2277c9ff2778f650a2acefee9b717e3d8f)

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

linux-yocto/3.4: inform the fetcher if the meta branch changes

The meta data branch for the kernel can be varied by changing the
KMETA variable, but that change wasn't reflected in the SRC_URI,
which means that the fetcher won't check for changes on that branch
and is a potential source of errors.

Making the meta branch specification KMETA vs meta solves the problem.

(From OE-Core rev: 5bfc678aeb65b50710384e93b64add3a62146908)

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

linux-yocto/3.4: emenlow: switch from psb to emgd graphics

Updating the meta branch SRCREVs for the following commit:

[
   meta/emenlow: use emgd instead of psb graphics driver

    This commit essentially does just this:
      Remove all the references of psb patches for emenlow BSP.
      Change the emenlow scc files to use emgd instead of psb.
]

(From OE-Core rev: 06a2b9eb087d1d90213d67dc8fe14c53201afe37)

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.4: merge v3.4.19, v3.4.20

Updating the linux-yocto-3.4 trees to incorporate both 3.4.19 and
3.4.20 from the kernel.org -stable team.

(From OE-Core rev: 8ac0c82bb8fe4fe17b6a32b18e57546d21b4d5ef)

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

puzzles: upgrade to r9712

License checksum change due to:
< Schmidt and Steffen Bauer.
> Schmidt, Steffen Bauer and Lennard Sprong.

(From OE-Core rev: 699b3b4451a7fe88e0dab46c5a781998226e377e)

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

lighttpd: upgrade to 1.4.32

lighttpd_fix_for_automake-1.12.patch: removed
- included in the new version

(From OE-Core rev: 6b87ac303c1f7357c9630359614b373082a59506)

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

libxslt: upgrade to 1.1.28

(From OE-Core rev: b3f32cd46d8ee8a17037365b9c2ab19bbf542c72)

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

libxcb: upgrade to 1.9

(From OE-Core rev: 39b96435a8199a09c586ca0002ad447bf42058b8)

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

ethtool: upgrade to 3.6

(From OE-Core rev: 22d2ca62df4e1b5edf2881864eb602c1fba83f13)

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

gconf.bbclass: run postinstalls at rootfs time

(From OE-Core rev: 50b7f625cc2c4b9224726b11b6d4bd192209cc6e)

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

gdk-pixbuf: generate the pixbuf loader's cache ar rootfs time

This will generate the loaders.cache file for pixbuf, at rootfs time.

(From OE-Core rev: 5ba420bc0e282a1cbf000ce32034ad7ab15c01d5)

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

gtk-icon-cache: run the icon generation at rootfs time

This change will allow for the icon cache generation at rootfs time and
only once, at the end. So, even though there will be many packages
depending on gtk+, the icon cache generation will be done once.
Hopefully, this will lower the target's first boot time significantly
by using the power of the host machine to generate the cache.

(From OE-Core rev: 74e34f6828771095a25d2c2de89f468b33e359f7)

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

rootfs generation: export two new variables to postinst scriptlets

In order for the postinst scriptlets to be able to run once we need to
export the location of the intercept scripts and also the location of
native sysrootfs. The gdk-pixbuf binaries will need the latter because
in order to generate the loaders.cache it will need to scan some shared
libraries that must be native. Even though the output is a text file.

(From OE-Core rev: 927fabf549bb79cc179d2cb0a953dcd515acf464)

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

image.bbclass: run postinst scripts once, at rootfs time

This patch will allow the repeating postinst scriptlets to be run
only once, on host, at do_rootfs time. This will lower the time for
rootfs generation and, also, instead of running some time consuming
scriptlets at target's first boot, we will do on the host.

(From OE-Core rev: 0d239328b1c6b9e4afadad75972d359181fbe1a5)

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

license.bbclass: Avoid grep error message

Touch a file that is later greped to make sure it exists.

(From OE-Core rev: a17bf5d840b5777918f324d26e085498dec86b76)

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

pkgconfig: Add a manual dependency of pkgconfig(pkg-config)

The libtelepathy-glib-dev package contains a .pc file that has a requires
of pkgconfig(pkg-config) > 0.21 in it.  In order to satisfy this requirement
we add a manual RPROVIDE to the pkgconfig package.

(From OE-Core rev: f9ddae16ef31b2c1cd988e1b416a39d516c986c4)

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

package_deb: Dependencies with a ( or ) in them are invalid in debs

Replace ( or ) with __.  This allows RPM style dependencies to be satisfied
in deb style packages.

(From OE-Core rev: 12acb6bb00ba665571fcfbdb0ef4668ca0a98d5b)

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

shadow: add runtime dependency on base-passwd

This fixes an issue where the shadow package was sometimes getting
installed before base-passwd during rootfs construction with Smart,
which meant the postinstall scripts could not find /etc/passwd and
/etc/group and failed as a result.

(From OE-Core rev: 6a660d39250cc635f7c1b6ae437f8fb40b7aec12)

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

python-smartpm: Fix incorrect comparison arguments

Built-in provides were not being compared properly within smart.
This was caused by an incorrect argument to the match function.

(From OE-Core rev: 81f8972323a0a89e67f5cc4be247bc8e1b637cc4)

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

python-smartpm: allow setting arbitrary macros in smart config

Add support for an rpm-extra-macros config option which can be
used to define some extra macros that need to be set for rootfs
construction to work on the host and that don't really warrant
their own specific named config options to be created.

(From OE-Core rev: 0564f7bbe8e06868e107b70d5abbf3dd6c113bad)

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

python-smartpm: support nolinktos and noparentdirs rpm options

Allowing us to turn on these options enables installation of base-files
which contains a symlink to /proc/mounts which is auto-detected as a
dependency and isn't provided by any other package.

(From OE-Core rev: 1f0ad78b8fa739f9a68ed4c7bfae2b6d8586fda2)

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

python-smartpm: don't create /var/tmp on every rpm database open

This directory if present in the rootfs interferes with proper
installation of base-files and doesn't seem to need to be there
for smart/rpm to work in any case.

(From OE-Core rev: 54d227e97ff3498fc2149f96f1ca40452ad21a32)

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

python-smartpm: fix package metadata not being read from rpm-md feeds

The XML indexes in rpm-md were being parsed incorrectly leading to the
package metadata (e.g. dependencies) not being read.

(From OE-Core rev: bb3abe241ae9d7e62e084cde383ebe1c6b95d825)

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

python-smartpm: style fixes

* Use ${PN} instead of python-smartpm
* Use multi-line strings instead of multiple single line += (also
  avoiding the confusing "_append +=")
* Use class-native instead of virtclass-native - no difference in
  behaviour, but this keeps things consistent if we have to add
  any target overrides in future.

(From OE-Core rev: 7dbe5004380b97949a7910706364ba57c6c52a33)

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

python-smartpm: add support for recommends

Implement support within Smart for handling RRECOMMENDS relationships
between RPM packages as used by OE. This includes support within the
base system for caching and resolving these relationships as well as
specific support in the RPM backend for reading the information from
packages, and reading the "missingok" flag added to createrepo for
rpm-md feeds.

(From OE-Core rev: 35e3bc90ec3bae824804bd176b3128efdb5b4e2b)

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

createrepo: implement support for recommends

Adds a flag into the output metadata to note recommends relationships
in a way that should not break compatibility with clients that don't
understand this flag.

(From OE-Core rev: e7df818182e9e68b2e0fdede20a41d492b776a5b)

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