git: add Git perl module to perltools package

Git perl tools such as add--interactive load the Git module at runtime.
A previous patch to eliminate a QA error by deleting it instead of
packaging it was incorrect.

    beaglebone[62]$ git add -i
    Can't locate Git.pm in @INC (you may need to install the Git module) (@INC contains: /usr/lib/perl/5.20.0 /prj/pab/Utils/lib/perl5/linux-arm/5.020000 /prj/pab/Utils/lib/perl5/ /prj/pab/Utils/lib/perl5/site_perl/linux-arm /prj/pab/Utils/lib/perl5/site_perl /etc/perl /usr/lib/perl/site_perl/5.20.0/ /usr/lib/perl/site_perl/5.20.0 /usr/lib/perl/vendor_perl/5.20.0/ /usr/lib/perl/vendor_perl/5.20.0 /usr/lib/perl/5.20.0/ /usr/local/lib/site_perl .) at /usr/lib/git/git-core/git-add--interactive line 7.
    BEGIN failed--compilation aborted at /usr/lib/git/git-core/git-add--interactive line 7.

[YOCTO#3780]

(From OE-Core rev: 804f8e650f433d00907ec04282c22aaff2e5c044)

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

oeqa/utils: Added filter to LogResults decorator to enforce custom log level.

(From OE-Core rev: 65c97d4abd26f38917cd89b5f50f7299221b2123)

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

sstate.bbclass: update the timestamps after install

Update the sstate file's timestamps after it is installed, it will be
very useful for removing the old sstate file, especially, it's not easy
to remove when use the shared SSTATE_DIR, we can easily remove them with
this change, for example:

$ find state-cache -type f -ctime +10 -exec rm -f {} \;

Will remove the sstate file which isn't used by recent 10 days.

We can use the -atime, but it is not always available, for example,
when mounted with "-o noatime".

The touch is a very light weight action, and the
scripts/sstate-cache-management.sh also requires this.

(From OE-Core rev: bbee747466a6947319cff2ffd676abf9432c16ae)

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

gnupg: add pinentry into RRECOMMENDS

Pinentry is needed for most function of GnuPG, so add it into RRECOMMENDS

(From OE-Core rev: dc274b6325cfc6b78d51c97445b1af445ccfed93)

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

pinentry: add recipes

gnupg 2.x.x needs pinentry to work

(From OE-Core rev: af893d747f3ee7894b2b444cf75024757f389742)

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

dhcp: use ${PN} for SYSTEMD_SERVICES

We should use ${PN} instead of hardcoding 'dhcp' for SYSTEMD_SERVICES,
otherwise we would have 'installed-not-shipped' QA error if we are building
lib32-dhcp.

(From OE-Core rev: c3a152f946f7cb1666384fd7a214f883cbaecb56)

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

volatile-binds: use ${PN} for SYSTEMD_SERVICE

As this recipe inherits allarch, it makes no real difference whether
we are using ${PN} or 'volatile-binds'. But using ${PN} would keep
the same style with the other recipes in OE.

(From OE-Core rev: 708cc039b6cc891e466e89d2b10fcdea6c19287c)

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

acpid: use ${PN} for SYSTEMD_SERVICE

We should use ${PN} instead of 'acpid' for SYSTEMD_SERVICE, otherwise
we would have the 'installed-not-shipped' QA error if multilib is enabled
and we run `bitbake lib32-acpid'.

(From OE-Core rev: f8217853c69cf06c92b2f3885f7d49851213188f)

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

packagegroup-core-standalone-sdk-target: Add libgcov-dev to on device SDK

When a user takes coverage in gcc of SDK, it becomes the link error in SDK
because there is not libgcov.

----
ld: cannot find -lgcov
collect2: error: ld returned 1 exit status
----

(From OE-Core rev: 99b26dba6e4f5a64579f183883265498000e3104)

Signed-off-by: Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gnomebase: fix indentation

(From OE-Core rev: 103ded50380916d210e293913eebb173f014063e)

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

syslog: Change unused variable declaration LOCAL=0 to LOG_LOCAL=1

This fix allows the correct appending of -L to syslogd arguments when both file and remote logging are selected.

(From OE-Core rev: ca68019760de6c5d5401be8eae7e65e7e6ca9021)

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

qtdemo-init: Make qtdemo startup correctly

The qtdemo can't launch via qtdemo-init initscript. Fix it.

(From OE-Core rev: bdd7a07db6b41c6c87c272d410af63c2d0251fc8)

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

perf: fix issue about package splitting

Currently, perf can't split to perf-archive, perf-tests, perf-python and
perf-perl. All files are included in perf package. Change the perfexecdir
variable to make split successfull. Add python to RDEPENDS_perf-tests.

(From OE-Core rev: 32fcc621401e7761d9b96bc5b7bef143c1c29695)

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

libpam / xtests: remove bash dependency

There's not bash specific syntax in the xtests scripts:

  $ cd Linux-PAM-1.1.6/xtests
  # replace /bin/bash to /bin/sh and check the bashisms:
  $ checkbashisms *.sh
  No output

So the runtime dependency to bash could be removed.

(From OE-Core rev: 1917bf7aa74aa1b86756c73c56537db2591115e5)

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

gst-plugins-base: fix build failure for x86

On x86, EMMINTRIN is defined but not usable without SSE so check for
__SSE__ and __SSE2__ as well.

(From OE-Core rev: 556a19423d15c7c13f60d57528a3b880f95750b9)

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

coreutils: selinux/flask.h should respect to with_selinux

Fixed when build with meta-selinux even with --without-selinux:
runcon.c:49:28: fatal error: selinux/flask.h: No such file or directory
 # include <selinux/flask.h>
                            ^
compilation terminated.

(From OE-Core rev: d52a606c8a75496f3b7239adc19fdb66e3ae576a)

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

systemd: disable resolv.conf symlink unless resolved is enabled

The tmpfiles configuration in systemd unconditionally creates a symlink
from /etc/resolv.conf to the location where systemd's resolved service
will place the real file.  This link is only appropriate when resolved
is enabled and running: its presence prevents connman or other systems
from providing a working resolv.conf when systemd is not assigned that
responsibility.  OE has not yet enabled systemd's networkd or resolved
by default.

There is a TODO in the systemd source to fix this, but it has not been
addressed upstream.  This patch comments out the corresponding line when
resolved is not enabled in the package configuration.

(From OE-Core rev: 7e2b05a6f59209687829225878a19a1dad8143b9)

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

sstate: Change overlapping files warning to a fatal error

When files overlap in the sysroot, something bad usually happened. We've had
two independent cases recently where a couple of months after one of these
warnings was shown, builds failed due to corruption.

This change moves the warning to become a fatal error. The complaint I've had
about this is that we need to tell the user what happened and more importantly
how to recover from it. If we could recover from it, great but the trouble is
we simply don't know what happened.

As a compromise, we can document several of the possible scenarios in the error
message. We don't normally go to this level of detail however in this case, I'm
lacking other viable alternatives.

I do believe it is important to stop as corruption occurs rather than letting the
build contunue into territory that is not deterministic amongst other things.

The complex message is followed by a simpler one in case the long message is too
much for the user.

(From OE-Core rev: 179ac7de03977b6e440409eddb2166819e07286a)

(From OE-Core rev: 4b503f25f1ef8f554d3c76d88399db379dc818cc)

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

bitbake: prserv/serv: Improve error message when prserver cannot bind to supplied host address

If localhost resolves to a remote address (due to a misconfigured network),
starting the pr server will fail without useful information.

To reproduce, add '<bogus ip> localhost' to /etc/hosts and run
'bitbake -p'. The error message will be:

ERROR: Timeout while attempting to communicate with bitbake server
ERROR: Could not connect to server False:

Running 'bitbake-prserv --host=localhost --port=0 --start' will fail with:

error: [Errno 99] Cannot assign requested address

Since these errors does not show the IP address of the attempted socket
binding, this results in a lot of wasted time looking at firewall rules, etc.

This patch results in the following error message if the socket binding fails:

PR Server unable to bind to <bogus ip>:0

(Bitbake rev: fae5914030bcf4c061c22fc61034c40c87b7121a)

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

bitbake: hob: Fix sstate mirrors mangling

hob was adding the redundant characters "\1" in SSTATE_MIRRORS variable. If
needed it is expected the user will add this instead so remove the code
that was doing this.

[YOCTO #6600]

(Bitbake rev: 73bf120062fc00c7e26dc4e77a7d140658d89daf)

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

bitbake: fetch: Extend testing of subdir unpack parameter and fix

This fixes urls of the form file://some/path/file;subdir=b. It also
adds in a couple of tests so we now tests these corner cases.

(Bitbake rev: 46306912a96444790efa9418d934dfdd36773ba1)

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

bitbake: fetch: SRC_URI parameter "subdir" does not work for local files

Check if the 'subdir' parameter exists and assign it to 'destdir' so that
files are copied in ${WORKDIR}/destdir. This fixes urls that are of the form
file://a;subdir=b.

(Bitbake rev: 836a986b365eb9798563ec08d90b346596de7791)

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

bitbake: monitordisk: don't log when not monitoring a filesystem for inodes

Writing a log that the filesystem isn't being monitored for inode usage just
confuses users who are not aware about the nature of inodes in their filesystem,
so don't say anything, just silently disable the monitor.  In general this only
happens on filesystems which don't have a limit on inodes.

(Bitbake rev: ca93bc84ee5fb94a50c11c47e4d212d7da649e24)

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

build-appliance-image: Update to dizzy head revision

(From OE-Core rev: e6c6d3fcfd2faf867e8145d25c1ba197fb9ee6b5)

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

bitbake: bin/bitbake: Update to version 1.24.0

(Bitbake rev: 637ce8df2658e4905fab8a0600a45505596bf472)

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

sanity.conf: Update minimum bitbake version to 1.23.2 due to event changes

(From OE-Core rev: b4e2a769e8def2e78ffca3f006a0cc88407aaeda)

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

layer.conf: Mark opkg-utils as ABISAFE for update-alternatives usage

Currently linux-firmware rebuilds for each machine due to its usage of
update-alternatives which in turn means a dependency on opkg-utils.
Marking opkg-utils as ABISAFE is the only option we have right now
to avoid this.

(From OE-Core rev: 6829c0e161c4a8cde6624f211865922fce62d4fa)

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

update-rc.d/systemd: Remove OVERRIDES dependency

Taking run-postinsts and building for two machines which have different
OVERRIDES leads to two different sets of stamps for an allarch package.

We don't need to depend on OVERRIDES in these classes, the end resulting
variables are good enough. We can therefore exclude the dependency
and allow a single package to be generated for run-postinsts.

(From OE-Core rev: fdc949154e64afb41dd4db3a97be74a15963128d)

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

layer.conf: Add in useradd dependencies as ABISAFE dependencies

Currently allarch recipes using useradd rebuild each time MACHINE
changes which is not desireable. Adding the useradd dependencies
to this list ensures they do not change under these circumstances.

(From OE-Core rev: 7743a309017f0fb9286f00b1f6f546ee95c05303)

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

man: fix not support xz/bz2 compression

In oe-core, bunzip and unzx located in /usr/bin/
rather than /usr, so tweak man's config.

[YOCTO #6750]

(From OE-Core rev: a8f07626d627b49913835778cc8039accd8b9896)

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

ltp: make setregid02 be able to pass

[YOCTO #6748]

replace "nobody" group with "nogroup", since the user "nobody"
belongs to "nogroup" group and no "nobody" group in oe-core

(From OE-Core rev: 3fc1b5acb50aa864d386f69a1c0d0db0425f970c)

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

systemtap: disable libvirt

libvirt is automatically linked to if present, this undetermined
dependency may cause build errors like:

  ../lib/libvirt.so: undefined reference to `libssh2_channel_open_ex'
  ../lib/libvirt.so: undefined reference to `libssh2_session_hostkey'
  ...

Both libvirt and libssh2 are not oe-core recipes for now, just disable
libvirt to fix this issue.

(From OE-Core rev: 815b44914eb30b86dfce0e81ffda7884e0efcc34)

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

wic: add sdimage-bootpart kickstart file

Add kickstart for generating a SD card image that should cover most use
case scenarios. The layout is as follows:
- 16MB vfat partition that IMAGE_BOOT_FILES will be copied to, 4k
  alignment
- ext4 rootfs, 4k alignment

(From OE-Core rev: bb01a6be7b32aa675f5003a6012a60a081212e8c)

Signed-off-by: Maciej Borzecki <maciej.borzecki@open-rnd.pl>
Signed-off-by: Maciek Borzecki <maciek.borzecki@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: add new bootimg-partition plugin

This patch implements 'bootimg-partition source plugin class for 'wic'. The
plugin creates an image of boot partition, copying over files listed in
IMAGE_BOOT_FILES bitbake variable.

(From OE-Core rev: 4a3200d710d953956064c28188577fbd461d093d)

Signed-off-by: Maciej Borzecki <maciej.borzecki@open-rnd.pl>
Signed-off-by: Maciek Borzecki <maciek.borzecki@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

documentation.conf: document IMAGE_BOOT_FILES

Document IMAGE_BOOT_FILES variable.

(From OE-Core rev: 8007f41aa824db2ee681d9dd98a0b06f44fc9d6d)

Signed-off-by: Maciej Borzecki <maciej.borzecki@open-rnd.pl>
Signed-off-by: Maciek Borzecki <maciek.borzecki@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: set bootimg_dir when using image-name artifacts

Running wic with -e to use artifacts from a named image, bootimg_dir was
always passed as empty string to partition source plugins. The patch
sets bootimg_dir to current value of DEPLOY_DIR_IMAGE, as bootloader
artifacts end up in that location as well.

(From OE-Core rev: d7f69e6f0932a927b6ce289fb47ba575d7aaa1c8)

Signed-off-by: Maciej Borzecki <maciej.borzecki@open-rnd.pl>
Signed-off-by: Maciek Borzecki <maciek.borzecki@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: fix vfat partition sector count only if needed

VFAT rootfs partitions sector count would get updated always even if
there is no need. Since parition size in wic is expressed in MB, any
sub MB change will cause the generated partition image to be larger
than allocated space within the disk image. Fortunately, partitions
sized in MB will most of the time have a proper sector count.

(From OE-Core rev: 99bee4cb489800d74dea4d2158ff834413685f04)

Signed-off-by: Maciej Borzecki <maciej.borzecki@open-rnd.pl>
Signed-off-by: Maciek Borzecki <maciek.borzecki@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: use IMAGE_EXTRA_SPACE for vfat rootfs

Functions for generating rootfs use IMAGE_EXTRA_SPACE rather than
BOOTDD_EXTRA_SPACE. The latter is used in boot image source plugins.

(From OE-Core rev: 44abf447f8e4ed11cbbe53a4fb0ecf10a20f0f9e)

Signed-off-by: Maciej Borzecki <maciej.borzecki@open-rnd.pl>
Signed-off-by: Maciek Borzecki <maciek.borzecki@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: minor comment update

Update comment about types of generated partition images.

(From OE-Core rev: ba1503f77e0c03fe6c715b61983008b7a534ae20)

Signed-off-by: Maciej Borzecki <maciej.borzecki@open-rnd.pl>
Signed-off-by: Maciek Borzecki <maciek.borzecki@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

uninative: Add uninative - a way of reusing native/cross over multiple distros

These patches are the start of a new idea, a way of allowing a single set of
cross/native sstate to work over mutliple distros, even old ones.

The assumption is that our own C library is basically up to date. We build
and share a small tarball (~2MB) of a prebuilt copy of this along with a
patchelf binary (which sadly is C++ based so libstdc++ is in there). This
tarball can be generated from our usual SDK generation process through
the supplied recipe, uninative-tarball.

At the start of the build, if its not been extracted into the sysroot, this
tarball is extracted there and configured for the specified path.

When we install binaries from a "uninative" sstate feed, we change the
dynamic loader to point at this dynamic loader and C librbary. This works
exactly the same way as our relocatable SDK does. The only real difference
is a switch to use patchelf, so even if the interpreter section is too small,
it can still adjust the binary.

Right now this implements a working proof of concept. If you build the tarball
and place it at the head of the tree (in COREBASE), you can run a build from
sstate and successfully build packages and construct images.

There is some improvement needed, its hardcoded for x86_64 right now, its trivial
to add 32 bit support too. The tarball isn't fetched right now, there is just a
harcoded path assumption and there is no error handling. I haven't figured
out the best delivery mechanism for that yet. BuildStarted is probably not
the right event to hook on either.

I've merged this to illustrate how with a small change, we might make the
native/cross sstate much more reusable and hence improve the accessibility
of lower overhead builds. With this change, its possible the Yocto Project may
be able to support a configured sstate mirror out the box. This also has
positive implications for our developer workflow/SDK improvements.

(From OE-Core rev: e66c96ae9c7ba21ebd04a4807390f0031238a85a)

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

bitbake: fetch2/perforce: Use replace (1 line) instead of find (3 lines)

(Bitbake rev: 5bf5a937b26896bedbfea78dd1d62bce5a26ac2a)

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

bitbake: fetch2/perforce: Fix localfile name if labels are used

I could apply the label "release_1.0" to a super project that contains
many sub projects.  If my recipes have SRC_URI's that use that label but
grab different sub-folders, than there's a bug where the cached localfile
(tar.gz) will not be unique and reused at the wrong times.

SRC_URI = "p4://perforce::1234@//depot/SuperProject/MiniProjectAAA/...;label=release_1.0 \
                     p4://perforce::1234@//depot/SuperProject/MiniProjectBBB/...;label=release_1.0"

(Bitbake rev: 3b5b1703b77490116dda895b29737cea51a3d6a0)

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

bitbake: knotty: Ensure commandline parameters are updated in memres server

When using options like -k, -f, -v and so on with the memory resident
server, they'd currently only be set on the initial values passed to
the original command. This ensures they now match those specified
on the commandline for the options where this makes sense.

To make this work, a command to update the options on the server side
is required so this is added.

[YOCTO #5292]

(Bitbake rev: 1c75cc4d0c8b606c1fe76e6bf60bf6a32298b105)

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

irda-utils: remove inherit autotools

irda-utils doesn't use autotools, so don't inherit it.

(From OE-Core rev: d5e2a59ab59e3d67d09c5f25b8623186af855e17)

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

oe-init-build-env-memres: Fix automatic port usage

The use of an automatic port wasn't working correctly since the server
was never getting started when port == -1. This fixes things so the
server is started when port is not specified (i.e. automatic) ensuring
this happens before BBSERVER is set.

[YOCTO #6563]

(From OE-Core rev: 982553b6d56ca4bfd095c1bcb736ae3b77deefa7)

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

linux-yocto/3.14: update to v3.14.19

Updating to the latest korg -stable update.

(From OE-Core rev: 7212dcb3a67b9a9b844b74e997d2e3ea7902555f)

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

linux-yocto/3.10: update to v3.10.55

Updating to the latest korg -stable update.

(From OE-Core rev: 9745ffc0ae25be980d92f195937cef6d1f406ab2)

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

linux-yocto/3.17: bump to v3.17-rc6

Updating the 3.17 recipe to -rc6. This is nearly the release kernel, and should
have very few changes aftert this point.

(From OE-Core rev: 845e01d906982c2147828b97129e95e8a79dce7c)

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

linux-yocto/3.17: switch to dedicated 3.17 repository

The 3.17 repository is ready, so we no longer need to reuse the
linux-yocto-dev tree.

(From OE-Core rev: c304674833360e0e2dceca3ebeb535025597e46f)

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

linux-yocto/3.17: update to v3.17-rc5

Bumping the SRCREVs to import the latest korg -rc.

(From OE-Core rev: 8ae153f7da3244c3dd24cc5dbd722af26624201b)

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

linux-yocto/3.14: update to v3.14.18 and -rt9

Refreshing the 3.14 kernel to a new korg stable and -rt release.

(From OE-Core rev: 91204afeb508ae21f8f2a32c340ec85efbf33fd0)

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

linux-yocto/3.10: update to v3.10.54 and -rt55

The 3.10 kernel needed a refresh to the latest -stable and -rt releases.

(From OE-Core rev: c7360e9e72f06dab2617e16ae546f4d8e5262fa1)

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

bitbake: bitbake-worker: Fix bitbake -n

Without this you see:

File "bitbake/bin/bitbake-worker", line 201, in fork_off_task
    os._exit(child())
TypeError: an integer is required

(Bitbake rev: cd477b5e77ab0373248b8a8fa30e1c7b8ea984fd)

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

oeqa/selftest: Added decorators to buildoptions.py

(From OE-Core rev: ea393e90055ea13b0c5a6950dbd388e826a66623)

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

kernel.bbclass: use one package split for all firmware filename extensions

(From OE-Core rev: dd5313228879487ca2b11fc5d38fb821a2a810a5)

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

u-boot: cleanup indentation and consolidate .inc file

(From OE-Core rev: 71525643909ac765e6b6a4b419cddae7d5812a8d)

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

populate_sdk_base/meta-environment: Remove overlap from the two

Currently we have the horrible situation where meta-environment packages
the toolchain environment files and they get included in the SDK but are
broken, then, the SDK code overwrites them with good versions. This is
suboptimal.

This change fixes the code in meta-environment to create working
files and adds in the multilib support from populate_sdk_base, then
we remove the code in that base bbclass and rely on the packages
being installed if/as/when needed. This removes the duplication
and the broken versions of the files, hopefully making all well.

[YOCTO #6608]
[YOCTO #6613]

(From OE-Core rev: 063355e5965439c7b3253d692d7ab0ed1189d123)

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

meta-environment: Fix TARGET_OS problem for ppc

Due to the use of the cross-canadian inherit, TARGET_OS can be reset to
values other than the correct one, leading to generation of the wrong
files.

By expanding REAL_MULTIMACH_TARGET_SYS before the inherit, the correct
OS value is preserved and we generate the right environment files.

[YOCTO #6608]
[YOCTO #6613]

(From OE-Core rev: ae265973c96ff4b8b5580436647fc361490e134c)

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

gstreamer1.0-plugins-bad: disable libssh2

libssh2 is automatically linked to if present, this undetermined
dependency may cause build errors like:

  .../x86_64-poky-linux/4.9.0/ld: cannot find -lssh2

libssh2 isn't an oe-core recipe, disable it for now.

(From OE-Core rev: 0cc59247ece1ea134d060d3ff064b5561972a92b)

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

perl: remove unneeded patches in ptest directory

The perl patches have been copied to ptest directory and these patches in
target rootfs are linked to build dir. The ptest of perl doesn't need these
patches, so remove them from target.

(From OE-Core rev: 1982095255917befd93ed14f9abc1f9fc4149f99)

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

python: obey LDFLAGS

(From OE-Core rev: 7191b6b7503a5a17f93bd61283f22d409c5cb17b)

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

irda-utils: obey LDFLAGS

While we're at it, also switch to explicit variable passing rather than
relying on make -e, and pass V=1 so the actual commands hit the logs.

(From OE-Core rev: 7106f840b7d2a71171131c3c3e5fc311718ca718)

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

blktrace: obey LDFLAGS for btrecord

(From OE-Core rev: 43be08bbfc3fd7ae2a8143ccd5d51d48394f4e77)

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

hostap-utils: obey LDFLAGS

(From OE-Core rev: 6b2395210e8c7fd7eb91fa78e043a811c3c0af3d)

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

setserial: obey LDFLAGS

(From OE-Core rev: aaebe0f814a031b06ba72bc9de8b5ec4dbf80f0a)

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

ossp-uuid: obey LDFLAGS

(From OE-Core rev: 3a006550fcc99af58fa5d933160f169e97c3a6de)

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

gdbm-1.8.3: obey LDFLAGS

(From OE-Core rev: efdc4bff8dff4bf065a7c65e7d1c9f5460839a24)

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

zlib: obey LDFLAGS for tests

(From OE-Core rev: c91d9153d5dc6750d1f4c7b3be58da0a1248245b)

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

ltp: Added zip-native as a DEPENDS

The Makefile checks for zip during installation

[YOCTO #6699]

(From OE-Core rev: a6e8ced3fa8e8e2aa3df0798b80eb26e5ebc4b15)

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

flex: fix the deps for ptest builds

Building the tests for flex requires flex-native and bison-native, but the
attempt to add this dependency was done incorrectly. Use an inline python
conditional based on PTEST_ENABLED instead.

(From OE-Core rev: fce2be6dcf8e7320b7b9df9947745e78fd160815)

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

cmake.bbclass : Add support for cmake projects that use .S files.

UHD and GNU radio use the cmake build system. The toolchain file made
from cmake.bbclass does not set the variable needs by cmake projects
that use .S files. UHD added some .S files and these changes are required
to build recent UHD.

(From OE-Core rev: 43ce4b804d433662fe77c6f5298060ba74a0e639)

Signed-off-by: Philip Balister <philip@balister.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sstatesig: fix overrides behaviour to remove SIGGEN_LOCKEDSIGS_i586

Require a sig file which SIGGEN_LOCKEDSIGS_i586 is not null, but
the actual SIGGEN_LOCKEDSIGS_i586 is null.
Invoking 'bitbake -e' and we got:
...
 3935 # $SIGGEN_LOCKEDSIGS_i586 [2 operations]
 3936 #   set /path/to/locked-sigs.inc:8576
 3938 #   del data_smart.py:406 [finalize]
 3939 #     ""
...

It was caused by the following commit:
...
(Bitbake rev: 899d45b90061eb3cf3e71029072eee42cd80930c)
Author: Richard Purdie <richard.purdie@linuxfoundation.org>
Date:   Tue May 31 23:52:50 2011 +0100

    bitbake/data_smart: Change overrides behaviour to remove expanded
      variables from the datastore
...

We add prefix 't-' to type to workaround the overrides behaviour.

(From OE-Core rev: f6a39cc957bf85ff43513f0b76afc3b2c9c906b6)

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

sstate: set SIGGEN_LOCKEDSIGS_CHECK_LEVEL default to error

(From OE-Core rev: 65020364bd089afbb83cd216e7ae2f837077bfc5)

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

sstatesig.py: Replace '_'s with '-'s in SSTATE_LOCKEDSIGS* names.

Using underscores in the "types" parts of the variable names can cause
unexpected issues with overrides. For example, if you have both
SSTATE_LOCKEDSIGS_x86_64 and SSTATE_LOCKEDSIGS_x86_64_i586, and i586
is in OVERRIDES, then you lose all of the contents in
SSTATE_LOCKEDSIGS_x86_64 and thus don't get some of the locked sstate.

Using '-'s in the variable names instead, eliminates these issues.

(From OE-Core rev: 6662c412a949a9f6b602c848e6303b19db7e5272)

(From OE-Core rev: 65f558a6f762fb13224091dc22903b58eeb9b392)

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

sstatesig: Improve the support for locked down sstate cache usage

Add code in the sstate hash validation code to ensure it really did
install these from sstate since if it didn't should to warn/abort
the build. The judgment condition is:
1) If a build is replaced by locked sstate-cache, it will triger a
   warn/error;
2) If objects are not used from the locked cache, it will triger a
   warn/error;
3) Use SIGGEN_LOCKEDSIGS_CHECK_LEVEL variable controls whether this
   is just a warning or a fatal error or nothing to report.

[YOCTO #6639]

(From OE-Core rev: 305912dce61c4fed0cbf631aa98a9e6f29db88e4)

(From OE-Core rev: 1683815695f39d4bad352348913f927ac8a1bcf5)

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

sstatesig: Improve to handle locking of multiple machines

Instead of a single monolithic SIGGEN_LOCKEDSIGS, split this into
separate variables, one per sstate package architecture. Add in
a new SIGGEN_LOCKEDSIGS_TYPES variable which lists the package
architectures to load in.

SIGGEN_LOCKEDSIGS_TYPES is made machine specific using overrides.

Also sort the hashes in the lists by PN to make diffing them easier.

(From OE-Core rev: d8b0ce35981931a39e7db9d8e78de6e009b34688)

(From OE-Core rev: b42f305ce38b9e0f1a2b7cb9586bbabcd2d27429)

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

sstatesig/sstate: Add support for locked down sstate cache usage

I've been giving things some thought, specifically why sstate doesn't
get used more and why we have people requesting external toolchains. I'm
guessing the issue is that people don't like how often sstate can change
and the lack of an easy way to lock it down.

Locking it down is actually quite easy so patch implements some basics
of how you can do this (for example to a specific toolchain). With an
addition like this to local.conf (or wherever):

SIGGEN_LOCKEDSIGS = "\
gcc-cross:do_populate_sysroot:a8d91b35b98e1494957a2ddaf4598956 \
eglibc:do_populate_sysroot:13e8c68553dc61f9d67564f13b9b2d67 \
eglibc:do_packagedata:bfca0db1782c719d373f8636282596ee \
gcc-cross:do_packagedata:4b601ff4f67601395ee49c46701122f6 \
"

the code at the end of the email will force the hashes to those values
for the recipes mentioned. The system would then find and use those
specific objects from the sstate cache instead of trying to build
anything.

Obviously this is a little simplistic, you might need to put an override
against this to only apply those revisions for a specific architecture
for example. You'd also probably want to put code in the sstate hash
validation code to ensure it really did install these from sstate since
if it didn't you'd want to abort the build.

This patch also implements support to add to bitbake -S which dumps the
locked sstate checksums for each task into a ready prepared include file
locked-sigs.inc (currently placed into cwd). There is a function,
bb.parse.siggen.dump_lockedsigs() which can be called to trigger the
same functionality from task space.

A warning is added to sstate.bbclass through a call back into the siggen
class to warn if objects are not used from the locked cache. The
SIGGEN_ENFORCE_LOCKEDSIGS variable controls whether this is just a warning
or a fatal error.

A script is provided to generate sstate directory from a locked-sigs file.

(From OE-Core rev: 7e14784f2493a19c6bfe3ec3f05a5cf9797a2f22)

(From OE-Core rev: 884d4fa3e77cf32836f14a113c11489076f4a84d)

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

bitbake: data_smart: Clarify what 'computed' means in the data store history context

(Bitbake rev: a2ca038dd1d0be4e0a0b20ae16a467d5a0075514)

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

gcc-configure/gcc-common: Move preconfigure definition to common include

There is a race where:

NOTE: recipe libgcc-initial-4.9.1-r0: task do_configure: Started
NOTE: recipe gcc-runtime-4.9.1-r0: task do_preconfigure: Started

| checking build system type... /home/pokybuild/yocto-autobuilder/yocto-worker/nightly-deb/build/build/tmp/work-shared/gcc-4.9.1-r0/gcc-4.9.1/libgcc/../config.sub: line 1711: syntax error near unexpected token `;;'
| /home/pokybuild/yocto-autobuilder/yocto-worker/nightly-deb/build/build/tmp/work-shared/gcc-4.9.1-r0/gcc-4.9.1/libgcc/../config.sub: line 1711: ` ;;'
| configure: error: /bin/bash /home/pokybuild/yocto-autobuilder/yocto-worker/nightly-deb/build/build/tmp/work-shared/gcc-4.9.1-r0/gcc-4.9.1/libgcc/../config.sub x86_64-linux failed
| WARNING: exit code 2 from a shell command.

so we need to make sure the preconfigure task executes in all shared
work contexts.

(From OE-Core rev: 3c30331d6eaf804b83a6d27189a12efc94310e91)

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

libgcrypt: Fix ARM assembly when building __PIC__

libgcrypt.so.20 contains .text relocations, backport a patch
to fix it.

(From OE-Core rev: a81d59bc8d13402725f0f5b42af92332570484d3)

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

build-appliance-image: remove hardwired path

Remove wrong absolute hardwired path
for the VM disk image.

Fixes [YOCTO #6728].

(From OE-Core rev: f8a9b81192f8846937d85c7189018e35c860f7bd)

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

rsync: remove trailing whitespace

(From OE-Core rev: 75d9bdd9c2942992064808baa83a8c8499daff55)

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

rsync: Add PACKAGECONFIG for acl/attr

Add PACKAGECONFIG for acl/attr
and setting the default value to enable them

[ RB: fix PACKAGECONFIG assignment ]

(From OE-Core rev: d84262d34920dd827b6d7f76c01ce0787a70f649)

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

systemd: fix to use ${libdir} for libraries

We need to use ${libdir} instead of ${exec_prefix}/lib for libraries.
Otherwise, we would meet do_install errors if multilib is enabled.

(From OE-Core rev: 6beeccb81416b0a7af6a4b19cb641f0d66a7198a)

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

lttng-tools: add PACKAGECONFIG to support --enable-python-bindings and lttng-ust

Add PACKAGECONFIG to support --enable-python-bindings
Add PACKAGECONFIG to support lttng-ust

And python-bindings requires swig-native as DEPENDS, and
swig-native is not in oe-core, so disable python-bindings
by default

(From OE-Core rev: 9557bf88287216ff8cb98005cbc85b6928f4495c)

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

i2c-tools: create i2c-tools-misc package for perl scripts

The patch is to split those perl scripts from the main package.
Add perl to RDEPENDS_i2c-tools-misc.

(From OE-Core rev: 9a8b35ec689ca583bb1e117ca4998215da7fcac6)

Signed-off-by: Michel Thebeau <michel.thebeau@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Chong Lu <Chong.Lu@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

portmap: add systemd service file

Add systemd service file for systemd support.

(From OE-Core rev: cc44cb2888f1ddecdd01d7bc414f7ac3043d1372)

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

portmap: fix INITSCRIPT_PARAMS

Fix portmap init to make it run at runlevel 2,3,4,5.

(From OE-Core rev: 93845fe60d89c0592a5c1ed631fc737c724cbc5b)

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

dropbear: add pam modules dependencies

If pam distro feature enabled, dropbear will need below pam rpms
to work:

  * libpam-runtime
  * pam-plugin-deny
  * pam-plugin-permit
  * pam-plugin-unix

Just add the runtime dependencies explicitly.

(From OE-Core rev: dfbeb663e99f3280d055ec04454353f2082ced03)

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

at-spi2-core: fix DBIND_CHECK_ALIGNOF m4 macro

Modify the macro so it will work when cross-compiling.
Note that the values checked are normally in a site file
so the macro rarely has to discover them.

(From OE-Core rev: 09610d6cdda95e964a7565b13eb35a26ae652aa9)

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

grub-efi: Replace _BSD_SOURCE macro with _DEFAULT_SOURCE

This is needed with glibc 2.20 since _BSD_SOURCE
macro is gone

Khem fixed grub, but the grub-efi also needs it, the patch can't apply
to grub-efi, so made a new patch for it.

(From OE-Core rev: da120dec96f8f019c0fd50576fb7490689234257)

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gstreamer1.0-plugins-bad.inc: Add rtmp support in gstreamer.

RTMP feature depends on rtmpdump.

(From OE-Core rev: 8d3dc353c8857d27957c9db4ced8b2009f2e9b2a)

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

sstate.bbclass: fix sstate_hardcode_path()

The "grep -e (x|y)" doesn't work, for example:
$ echo xy | grep -e '(x|y)'
No output

We can use "grep -E" (extended regexp) or "grep -e x -e y" to fix it.
It only affected the cross recipes.

(From OE-Core rev: 62722de6d0ec00608eacc2cb0396362aced00047)

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

license: Improve disk usage

Currently copies of the license files are made which wastes disk space
and adversely affects performance. We can link these instead in most
cases for small performance gains.

(From OE-Core rev: 0b0f3631fd22f731b6aeedb73965e367b695028b)

(From OE-Core rev: fe00d4f479c4fb5e4be5dda616a4de0a257ef6c3)

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

sstate: Fix incorrect return value handling

The use of [ and && here means $? is reset and the exit 1 error
interception wasn't working, leading to "file changed as we read it"
errors from sstate_create_package when heavily using hardlinks.

Fix this by placing $? into a variable.

(From OE-Core rev: 6e51f900b76b06c09a3d6927f8db7398e2c035ed)

(From OE-Core rev: 75040a098e11927e6872e3a2a6286fe3ed0c7f47)

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

shadow: Make useradd work correctly with --root again

Even if useradd --root <root> is used it would still read login.defs
before doing the chroot() and thus use the one provided by the host
rather than the sysroot.

(From OE-Core rev: b85917a4ebe636316fa7305017cd32a47b392039)

(From OE-Core rev: 0af59a04135f067f0e01883defa77c6f714eab2e)

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

bitbake: data_smart: Fix remove operator and its interaction with data expansion

If you have:

FOO = "${bindir}/X Y"
FOO_remove = "${bindir}/X"

the expected result is "Y". Currently this doesn't work since the removed
expressions are not expanded first. This patch adjusts things so the
expressions are expanded before being processed for removal.

Also add a test to ensure this case continues to work.

[YOCTO #6624]

(Bitbake rev: 72a1ca4a104ccab73d6abcbd44db9c2636a58572)

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

classes/populate_sdk_base: enable adding custom commands to SDK install script

Add an SDK_POST_INSTALL_COMMAND variable which allows additional
commands to be added towards the end of the SDK install script, for e.g.
additional processing that needs to be done as part of installing the
SDK.

(From OE-Core rev: 10df0718d6a626d99beb68cde8d914ee0820d7eb)

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

bitbake: bin/bitbake: Update to version 1.23.2

(Bitbake rev: e24095f54c52a547c0462836586a5d716249036e)

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

bitbake: siggen/runqueue/bitbake-worker: Improve siggen data transfer interface

We need to transfer some of the siggen data from the core/cooker into
the worker instances. There was a partial API created for this but
its ugly and its not possible to extend it from the siggen class.

This patch completes the interface/abstraction for the data and
means the class can extend/customise it in any siggen class.

(Bitbake rev: cf2d642052979d236185c5b8ca2c5478c06e62ae)

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