enchant: fix unrecognised option

Fix the unrecognised option --disable-binreloc, the configure now uses:

AM_CONDITIONAL(WITH_BINRELOC, test "x$br_cv_binreloc" = "xyes")

We can use br_cv_binreloc=no to fix it.

(From OE-Core rev: 00c893c0fe524abb4ae4f1aa6d007ff543fb6afb)

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

bluez-hcidump: remove unrecognised options

Remove the following two options since they are not needed any more:
* --with-bluez-libs
* --with-bluez-includes

  commit 78631c5f66f0580695ff3aac4dc8831e3795b735
  Author: Marcel Holtmann <marcel@holtmann.org>
  Date:   Wed Aug 31 00:57:30 2011 -0700

      Change includes to use local copy of Bluetooth library

(From OE-Core rev: 42877cedf24629fe61855ce3de5831e55d53a9bc)

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

gstreamer1.0: remove unrecognised options

Remove the following two unrecognised options:
* --disable-oggtest
* --disable-vorbistest

They are defined in m4/ogg.m4 and m4/vorbistest.m4, but now configure
doesn't use them, configure.ac uses another way to define the
--enable-ogg and --enable-vorbis, and there is no test options, so
remove them.

(From OE-Core rev: 246119c1ec216c66395c26292938168c27b73f69)

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

gstreamer1.0: remove unrecognised options

Fix the following options:

--disable-docs-build: replace with --disable-docbook --disable-gtk-doc
--with-check=no: --disable-check

(From OE-Core rev: 035f0e282ced1d1a2ebaf6900aa870a88e0f9450)

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

gstreamer: fix unrecognised options

Fix the following options:

--disable-docs-build: replace with --disable-docbook --disable-gtk-doc
--with-check=no: replace with --disable-check

(From OE-Core rev: 72a75bb0fbe7caddf00710a61015f18a96e40cbf)

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

libid3tag: remove unrecognised option

There is no --enable-speed in libid3tag's configure, but in libmad, and
it's already there, so remove it.

(From OE-Core rev: bc64977ac02fc71de620254eb9f65998ae842629)

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

dpkg: fix/remove unrecognised options

Fix/remove the following unrecognised options:
* --without-static-progs: can't find anything about this in the history,
  so remove it.
* --without-dselect -> --disable-dselect
* --with-start-stop-daemon -> --enable-start-stop-daemon
* --with-bz2lib, typo, should be --with-bz2
* --without-sgml-doc: the sgml doc had been removed from dpkg, so
  remove it.

(From OE-Core rev: 8d30a464cdb8c6bf0b9d2757ff7f8fc3445f51ba)

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

tremor: remove unrecognised option

Remove the --disable-rpath, it seems taht there is never such an option
in tremor's configure, and it uses our libtool, so we don't have to
worry about the rpath.

(From OE-Core rev: 7803d98c7d72bc2313b965dfdf198bac6f31e6b0)

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

alsa-lib: remove unrecognised options

Remove --with-cards and --with-oss, they are never in alsa-lib, but in
alsa-driver, and the --with-cards had also been dropped from
alsa-driver.

(From OE-Core rev: ba2901ca9de118d92abe63ac78fa3652ffa66ee7)

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

flac: remove unrecognised options

Remove --disable-id3libtest and --without-id3lib which had been removed
from flac for many years, here is the log from flac:

  commit e31d9eb05cb82bd5f9cab1e6ad5295ddad22bcab
  Author: Josh Coalson <jcoalson@users.sourceforce.net>
  Date:   Tue Sep 28 00:23:57 2004 +0000

      remove id3 support from the plugins

(From OE-Core rev: 4c75b4f973b156bfeabedbe79a7bcfaaf9fe5bfa)

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

bitbake: fetch2: avoid printing "no checksum" error message twice

Because of the way we were handling this error, it was printed twice -
once via logger.error() (to avoid the log being printed) and a second
time when the exception gets wrapped in a FuncFailed at a higher level.
Call logger.error() earlier and change the text we send in the
exception to be more brief, so it more closely resembles the behaviour
when there is an invalid checksum.

(Bitbake rev: 46765369d7f76ec7f67b90430131a79eb6a66235)

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

bitbake: fetch2: fail checksum validation if SRC_URI checksums set to ""

We were checking SRC_URI md5sum/sha256sum values against None here, so
if they were set to "" then no error was produced. Since the value is
still effectively unset in this case, this is not the right behaviour;
just check if the value doesn't evaluate to False instead.

(Bitbake rev: 040943a718795c64dc4e604abfcf08b26b7d00e6)

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

openssl: use PACKAGECONFIG to disable perl bits

Adding perl to the RDEPENDS caused a performance hit to the overall build time since this was
the only package that depended on perl.  The openssl-misc package is not installed by default
so use a PACKAGECONFIG which can be overridden to allow the perl scripts along with  perl to
 be installed.

(From OE-Core rev: 421e927bd453259f4b3cdbd1676f6e12f97bf34f)

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

toaster.bbclass: fix crash on buildstats collection

Toaster needs buildstats to be enabled in order to
collect task statistics. The toaster.bbclass didn't
have a guard to stop task data collection if the buildstats
weren't enabled, leading to a crash.

This patch verifies that the task variables are defined before
trying to use them.

(From OE-Core rev: 7f2637ec8d1b8840c31255572b5cb058777382e7)

Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

valgrind: upgrade to 3.9.0

Changed patches directory to "valgrind";
Removed 4 patches because they were merged upstream;
Updated 2 patches because now it has configure.ac instead of .in;
Changed license md5 for 2 files because of a small change for copyright
year

(From OE-Core rev: 57794c201aa732392a135e03e04faf18929f645f)

Signed-off-by: Cristiana Voicu <cristiana.voicu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/insane: restore printing LIC_FILES_CHKSUM checksum if not specified

OE-Core rev ec8590aa81e201e28e500935d31cd7266114471f (by me)
unintentionally disabled printing the actual checksum value if no
checksum was specified, i.e.:

LIC_FILES_CHKSUM = "file://COPYING;md5="

Printing the actual checksum in this case is really useful when writing
a new recipe, so put this back in.

(From OE-Core rev: e58cdd385584d7449236e6b1dc1bce0cbc8f2a0a)

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

libav: add libpostproc to PROVIDES (for 0.8.x version only)

There is a separate libpostproc recipe in meta-oe for use with 9.x and
later versions of libav for those few that need libpostproc; however if
you just add meta-oe and try to build libpostproc without selecting the
libav 9.x version recipe, you'll be building the libpostproc recipe
together with libav 0.8.x, which provides its own libpostproc; this
leads to confusing errors at packaging time. In order to flag up that
these conflict more appropriately, add libpostproc to PROVIDES
explicitly so that you at least get a multiple providers error at the
start of the build.

Fixes [YOCTO #5335].

(From OE-Core rev: e8f9420fe901675fc1a8d4e41302c2faa4a7dc4a)

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

sstate: Use datastore copies due to data changes persisting

The way the "all arch" PKGSPEC is enabled, it causes corruption
of the datastore of sstate operations against other tasks.

Data store copies are cheap and allow us to use that trick, resetting
to a clean copy of the data afterwards.

(From OE-Core rev: afaf16100efa79a275a2f4b9f2caa80decfdeb81)

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

externalsrc: Fix deltask usage

An incorrect version of the patch merged which entirely removed the covered
variable. This corrects the patch so the code works.

(From OE-Core rev: 33874f9af895757c4b0fd2f4d350ed308edac03c)

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

terminal: Exclude BB_ORIGENV from the signatures

devshell was printing a traceback when exiting due to the use of dump_sigs()
being called on the task. This is turn was since this function referenced
BB_ORIGENV. We might as well globally exclude this for now since its a
data store object and cannot be pickled, not would it make sense to do so.

[YOCTO #5683]

(From OE-Core rev: 84b549afb46fce7b5cdaa977286aeb2e90d3bfdb)

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

yocto-bsp: Add missing interfaces file for qemu BSP templates

The qemu BSPs were picking up the wrong interfaces file, resulting in
some erratic interface behavior seen with qemu-based BSPs - this fixes
the problem.

[YOCTO #5636]

(From meta-yocto rev: bbc3d56d6ec28b4cd92874fe4f98e1cd499415be)

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

sstate/sstatesig: Add populate_lic to list of arch invariant sstate tasks

Like fetch, unpack and patch, populate_lic doesn't vary between different
archs so we should mark it as such. This means better sstate cache reuse
with fewer duplicate files as well as less confusing sstate debugging.

sstatesig also needs to account for the fact BPN is used for sstate files
in these cases.

(From OE-Core rev: 3d59d0bed756f64d0092caa3892239c779c4a341)

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

bitbake.conf: Exclude WORKDIR changes from sstate checksums

The layout of stamp files ensures that changes to WORKDIR mean recipes get rebuilt correctly.
Since WORKDIR usually contains MULTIMACH_TARGET_SYS and that depends on tune variables,
including WORKDIR in sstate checksums adds a lot of noise to the system for what amounts to
no gain.

On the other hand, removing it reduces noise, reduces the size of the siginfo files and
reduces the amount of processing bitbake has to do. It therefore seems like dropping it
from the checksums is an all around win.

(From OE-Core rev: 453353e05d027c6a505d1e13a7982718a13bca8b)

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

sstate: Convert to use ':' as a filename sperator and use SSTATE_SWSPEC globally

Currently the code has problems differentiating between "gcc-cross" and "gcc-cross-initial"
sstate files. We could add in a ton of special casing but tests show this isn't scaling
well. Using a more unique separator resolves the issue.

The choice of which separator to use is a hard one. We need something which isn't commonly
used in PN, PV, PR, *_OS and *_ARCH which rules out '-', '_' and it needs to work ok with
webservers/http which makes ';' and '%' harder.

The change also sets SSTATE_SWSPEC globally since writing out differently named siginfo
files for the fetch/unpack/patch tasks is a waste of diskspace, the hashes match for
all PN in the majority of cases and if they don't, its not a big issue as the hash is
different. This makes the results from sstate debugging more understandable.

(From OE-Core rev: 6f823a23c5f1d0ffa0a27db1c1bc1907de788505)

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

libcap: Allow rebuilds to work correctly

Currently rerunning configure for libcap causes a build failure. This fixes
the sed expressions to work correctly for a rebuild.

(From OE-Core rev: 2a7b96549aab7c6dd854edf3c18e2bb6bb48479d)

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

beecrypt: Allow PACKAGECONFIG to be overridden

(From OE-Core rev: ec480d385dd61798dfa224d0c940a36f053f2942)

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

harfbuzz: Allow PACKAGECONFIG to be overridden

(From OE-Core rev: ca35307b988367f18c3039c1d6c2bc400965a23f)

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

nfs-utils: modify the boot sequence of the init script

Move the nfscommon out of rcS and let it start just before
nfsserver. Modify the useradd parameter with long parameter
to make it more readable.

(From OE-Core rev: 73f94bed2c7603e1b3ded6ebd1a72ba237ff6e3d)

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

runqemu: Allow user to set -vga option with qemuparams

At the moment, the user cannot to set -vga other then vmware
(because "vmware" is set by default); and the first argument
in qemuparams has higher precedence.

(From OE-Core rev: 54a43397c48c974570e3eade55163eb766994a55)

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

testimage: include traceback when loading a test fails

Makes it much easier to figure out where a syntax error is.

(From OE-Core rev: 0c30a25c3d5f7fb1087dff45e01595329620235f)

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

lib/oeqa/runtime: rewrite the systemd test module

They are basically the same tests but:
- they look cleaner, using one single method / assert
- output from unittest will be cleaner (and includes a verbose status when needed)
- they are better grouped and use a real, active, enabled service
  (machineid will be dropped and hostnamed was a static service)

(From OE-Core rev: 0aa5b97f9e0d8941ce342e9f162dd3b19d4b49bb)

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

oe-selftest: added buildhistory buildtime tests in module buildoptions.py

(From OE-Core rev: 0e462295d9db0201fb1ba2b6247ecb05fa7fcee1)

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

oe-selftest: New test module for OE scripts.

Added a new module meta/lib/oeqa/selftest/oescripts.py containing tests
for scripts from ${COREBASE}/scripts

(From OE-Core rev: 685879739017317f234824689cbd89388d236816)

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

oe-selftest: New BuildhistoryBase object for buildhistory testing.

(From OE-Core rev: 8c77911b0924dfb5b966bc40a541a02a29568603)

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

conf-notes: remove meta-toolchain-sdk references

[YOCTO #5676]

(From OE-Core rev: 4c81f743eb15604eb389f3ceafe7af7567a02e0d)

(From OE-Core rev: 987802907ef702bfec65507b276e166f0aee59ec)

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

libtirpc: upgrade to 0.2.4

obsolete_automake_macros.patch removed,
no longer needed.
Disabled GSS APIs, as those APIs are
not present in our distro.
Patches directory is now generic,
as patches are not created for a specific version.

(From OE-Core rev: 3d42e107379d2e79b14a2dbcd35820c63adbfdb0)

(From OE-Core rev: 59c847b8cda4d70077c89f5761e000d089f8eb4b)

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

sstatesig: Correctly handle matches spanning stamps and sstatedir

By resetting filesdates at this point, we lose matches from stamps which
may not have been in sstatedir. When we don't have hashes specicifed,
its better to return all matches and have the caller decide which are
relavent and which are not since this function has no ability to
decide. There will almost always be one match from stamps we need
to keep and refer to.

(From OE-Core rev: f4c1c9ad2c7e944d4926d0629611da97f9df6a9a)

(From OE-Core rev: 8629844640a5a4c6a9d9375e7b637bcbf3451c4c)

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

sstatesig: Update for the removal of sstate-name

We've dropped sstate-name so we can remove this code. The fallback was
incorrect since we use taskname without the do_ prefix so this patch
updates to account for that too.

(From OE-Core rev: 72ff58124081333d46d37f31f2d1bf40d715e3bd)

(From OE-Core rev: dbc1426ee75bcf2f6d8b18312522a9b823e23173)

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

sstatesig: Ensure we return all matches for find_sigdata

When the hashes to find isn't specified we need to return matches from both
the sstate cache and the local stamps directory regardless of how many we've
found so far. If we don't do this, we can miss stamps and the comparison is
less accurate/incorrect.

(From OE-Core rev: 08a074e11e2d517b81ca71fd9bda65297bb015a7)

(From OE-Core rev: 04fc682af7c1657b1c9f37b66a78c4ffc8066e24)

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

ptest: Disable for native recipes

We don't use or otherwise care about ptest for native recipes. Its therefore
pointless to take the performance hit for them and we can disable them.

(From OE-Core rev: f3d35bb4719d0b8f8e6fc5976e9dbfc0e2019c2e)

(From OE-Core rev: 16dd4cd5564a19b0a221661205430e01c35673af)

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

ptest: Fix dependency/ordering issue

We've seen occasional issues on the autobuilder where files appear during
do_populate_sysroot and this is due to the ptest installation happening
in parallel. This fixes the order to be deterministic.

(From OE-Core rev: 0ed1641d870f4bb5735aeeb7bd29cc196e61b7cc)

(From OE-Core rev: 96aa9bf8880088c9a710c86cc8ee106138152a98)

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

externalsrc: Use deltask API

Now we have deltask API, stop poking around bitbake internal variables.

(From OE-Core rev: 98637df0c04fd14b506d9eec7da8ec6ae441221b)

(From OE-Core rev: 3aa30ca04752c6366a81db6ff99209fa0b46861c)

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

native: Use deltask to remove tasks which have no purpose

core-image-sato has 47703 inter task dependencies before this patch
and 29883 afterwards which is a significant worthwhile task graph
simplification.

(From OE-Core rev: ac4b1518ec549b5a4d7a79c60dbf96b90797e1d8)

(From OE-Core rev: 8f4f229ba8117ba411a858bf47f88a370bb70491)

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

sanity.conf: Require bitbake 1.21.1 as a minumum version for deltask functionality

(From OE-Core rev: cdc6dfd497c7a96f8ccd666fe0b8970d5b906786)

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

bitbake: Update to version 1.21.1 for master

(Bitbake rev: 4cc6e61fe11eb233bdba7c1bdc110b8cdafa56f8)

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

bitbake: siggen: Fix reversed difference output

The output when comparing siginfo files for dict_diff is reversed and shows
additions when things were removed and vice versa. This patch reverses the operation
so the changes are shown correctly and makes the output less confusing.

(Bitbake rev: 9b4142df36619099670740a5d3bc94e404ab2b56)

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

bitbake: runqueue: Further extend bitbake -S output to view signature differences

Based upon the list of difference starting points, we can use the siggen.find_siginfo()
function call and the difference printing code to provide a list of differences
between the current build target and whatever can be obtained from the sstate cache.

(Bitbake rev: 7a77861feb62750ef166d2d1e89ed1f444ca8dc7)

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

bitbake: siggen: When printing signatures recursively, limit the output

Currently the code prints all differences. If the task dependencies have changed hash,
we recurse into those and print those differences as well. This leads to a lot
of output. The reality is if the parents changed signature, we might as well just
say that and recurse with no other output since we're much more interested in how
the parents changed in nearly all cases. The changes in the parent are probably
the same ones we'd have printed at each level anyway.

By doing this we focus the output more carefully on the thing the user wants/needs
to see.

(Bitbake rev: 7a17fd6e51a76d3582c357b79f5ef86e1969650c)

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

bitbake: runqueue: Fix data being written into siginfo/sigdata files

The way hash_deps was being generated was different to the way siggen generated
the data internally which lead to seemingly different sigdata/siginfo files
for the same checksum. The -S output correct but the files written during
builds contained superflous data which would look like a difference.

This patch removes the badly duplicated data and uses it from the source
which ensures its consistent.

(Bitbake rev: e6d5e925c402cd2cc7ee034e9de4cc6df8944a34)

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

security_flags: more relocation issues

These are similar relocation R_X86_64_PC32 issues that are solved by
removing the -pie flags.

[YOCTO #5515]

(From OE-Core rev: cd94dd3d9bba32c3fd55959586128b236d1d4e34)

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

dbus-ptest: Fix SRC_URI checksums

The upgrade changed the checksums for dbus but not dbus-ptest. This
patch corrects that.

(From OE-Core rev: 9e56856a15d80bf57b8a4fe54508141da46581f2)

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

bitbake: runqueue: Add output for -S option for listing the changepoints compared with an sstate cache

Its useful to understand where the delta starts against an existing sstate cache
for a given target. Adding this to the output of the -S option seems like a
natural fit.

We use the hashvalidate function to figure this out and assume it can find siginfo
files for more than just the setscene tasks.

(Bitbake rev: c18b8450640ebfd55a2b35b112959f9ea3e0a700)

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

bitbake: bb.fetch2.git: reuse basecmd attribute

The basecmd is initialized in urldata_init; there's no need redoing that
work.

(Bitbake rev: f8df6f746fb2e27f029a5449cee6c891b1f36f4f)

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

bitbake: fetch2/git: Improve handling of unresolved names verses branches

Currently the fetcher doesn't distinguish between names that the fetcher
needs to resolve verses branch names that the user specified.

This meant that if you specify a tag and a branch, the fetcher broke. This
separates the two so that the branch name is preserved and can be used in
appropriate places.

(Bitbake rev: e85f39fe9d1b224414b5da0780da514f75c5df92)

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

bitbake: build/ast: Create strong task add/del API in bb.build

Currently its near impossible to control task addition/deletion from
metadata context. This adds stong add/deltask API to bb.build
which is traditionally where it resided. The rather broken
remove_tasks function was removed, it didn't appear to do anything
useful or have any users.

This allows us to clean up hacks currently in use in metadata and use
standard API for it instead.

(Bitbake rev: bf7138dd38fc1f8efca80891198e3422fef64093)

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

bitbake: ast/BBHandler/build: Add support for removing tasks (deltask)

Back in the depths of time we did support task removal. In the pre
AST days it was nearly impossible to continue supporting it, it wasn't
used so it was dropped. With the modern codebase we can easily now support
deltask and it would be very useful within the metadata since it can
massively simplify dependency trees.

As an example, a core-image-sato had 47703 inter task dependencies before
this patch and a patch to native.bbclass, afterwards with the noexec tasks
deleted, we had 29883. Such a significant simplification is worthwhile
and justifies adding a deltask operation to the system.

(Bitbake rev: acecbde6fb70ff3c96deab3cdf819d8442e87ed4)

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

Minicom depends on libiconv

(From OE-Core rev: 1d6caef7222d0c1086a08a109ea4135a388c88e6)

Signed-off-by: Krzysztof Sywula <krzysztof.m.sywula@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd-compat-units: drop machineid.service

systemd will create the machine ID automatically if it isn't set, so this unit
doesn't serve any purpose.

(From OE-Core rev: 8e6da3d8973777d9b651d085e06ee20a761ed62c)

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

systemd: delete files that end up in /var/volatile

${localstatedir}/log/journal is created by tmpfiles.d on boot, and the README
just causes warnings that /var/volatile isn't empty when mounting on top of it.

(From OE-Core rev: dfcd8f2a6f5847d735755c1644c4c93be4fbb3e8)

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

systemd: pass --root in prerm if running on host

If a systemd image is built without a package manager then packages will be
removed from an image during rootfs generation, but without passing --root the
systemctl will look on the *host* system.

(From OE-Core rev: d01da862d10d9544f8da846b577cf955041d4c0c)

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

dbus: upgrade to 1.6.18

Drop PR and INC_PR.

Move patches into a non-versioned directory, and update dbus-ptest to match.

(From OE-Core rev: 0289a1c325eea881ce3d68def5262f9e763d4fd2)

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

buildhistory: be quiet when garbage collecting the git repo

(From OE-Core rev: b117403423f700eb4cc6967b9fb8ff3e2858e279)

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

useradd.bbclass: add dependency on base-files

Packages that use useradd.bbclass should have a dependency on base-files so that
the /etc/skel directory is populated.  Without this dependency base-files may or
may not be installed when the postinst runs, and the skel content may or may not
be copied.

(From OE-Core rev: 556368ba8a1f933a86b69be024bd0711d4bfe0a3)

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

sstate: Get rid of crazy name mapping

When originally developed, it was thought a task may have more than one associated
sstate archive. The way the code has grown that idea is now not possible or needed.
We can therefore assume one sstate archive per task and drop the crazy name
mapping code. Simpler is better in this case.

The downside is that various sstate archives will change name so this forces a cache
rebuild. Given the other sstate changes going in at this time, this isn't really
a bad thing as things would rebuild anyway.

(From OE-Core rev: 5afe86a6854b21692fd97c5fc7fab50dbc068acb)

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

sstate/gcc: Fix shared workdir handling for siginfo files

For a shared workdir, any one of the fetch/unpack/patch tasks may run yet the
PN and architecture fields in SSTATE_PKGSPEC may differ. This makes looking up
the appropriate siginfo file near impossible.

I've tried several different ways of resolving this and this is the neatest
solution I could find, its still rather ugly. I believe the usefulness of
better sstate debugging outweighs the ugliness of the code.

This patch also changes the sstate_checkhashes() code to look for siginfo
files rather than the actual sstate packages themselves. This means the
function can be used in other contexts to find info files for tasks that
may not have sstate data. It is assumed that sstate mirrors will have both
files available. This is done to allow bitbake to query whether tasks have
matching signatures in sstate directories or not.

(From OE-Core rev: 068e4289b597699cbff2dfde44ba833af4535281)

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

sstate: Write out siginfo files for non-sstate tasks

Currently siginfo files are only written for sstate tasks. In order to be truly
debuggable, its helpful to have the siginfo for intermediate tasks. This
adds that functionality so the extra siginfo files are written out too.

This will be used to add better sstate debugging in future changes.

(From OE-Core rev: 04d108cd16f5ad8f92a62ea537d1330fee712470)

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

openssl.inc: Install c_rehash utility with openssl

c_rehash utility is not being installed with openssl.It conveniently
generates hash and symbolic links based on it for CA certificates
stored locally for SSL based server authentication

(From OE-Core rev: 3c2f9cf615c964e8303fd3e225ea7dd7b5485155)

Signed-off-by: Yasir-Khan <yasir_khan@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

pointercal-xinput: Mark as SIGGEN_EXCLUDERECIPES_ABISAFE

When including xinput-calibrator, in commit "xinput-calibrator: move
it from meta-oe to oe-core" the pointercal-xinput has not been added
to the SIGGEN_EXCLUDERECIPES_ABISAFE. This changes adds it to the
meta/conf/layer.conf's file list.

(From OE-Core rev: 9582ed663d12e635c02c59bff8665c929fa35656)

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

gcc-4.7/gcc: disable sdt from configure.ac to keep compatibility with configure

We had disabled the sdt from configure, let's also disable it from
confgure.ac to keep them compatible.

BTW, the libstdc++-v3 of gcc-4.7 doesn't use the sdt, so we don't need
to edit libstdc++-v3/configure as gcc-4.8.

NOTE, this commit edit the patch gcc-4.7/disablesdt.patch directly.

[YOCTO #5657]

(From OE-Core rev: 452c6afc3883b0304e06eb1764614348fb6599c8)

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

gcc-4.8/libstdc++-v3: disable sdt

We may meet such an error when building gcc/libstdc++-v3:

gcc-4.8.1/libstdc++-v3/libsupc++/unwind-cxx.h:41:21: fatal error:
sys/sdt.h: No such file or directory

We already have a patch to disable the sdt for gcc, we also need disable
it for libstdc++-v3.

BTW, we need edit both configure.ac and configure to make them keep
compatible.

NOTE, this commit edit the patch gcc-4.8/0031-Disable-sdt.patch directly.

[YOCTO #5657]

(From OE-Core rev: 32854af3cc6c0626620e827dc1915f61c51250b8)

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

libsoup-2.4: add intltool-native to DEPENDS

The configure script looks for this; most of the time dependency chains
ensure this is present but we need to be explicit or failures can
occur.

Reported by Nicolas Dechesne <nicolas.dechesne@linaro.org>

(From OE-Core rev: 22e45ed7d74ceb4a719e7b5889400c20ed4a0783)

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

grub: move xz to DEPENDS list from RDEPENDS list

liblzma5 is really requiring by grub, setting RDEPENDS to xz would pull
unneeded xz binaries into rootfs.

(From OE-Core rev: 78526905999fa38047ae8f3491127cc03de3e3f6)

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

scripts/contrib: Add graph-tool

A simple script I put together for getting the paths from one node to
another in a dot graph. This is useful for example in working out why
a particular recipe is getting built in conjunction with dot graph files
produced by bitbake -g.

For example:

$ bitbake -g core-image-minimal
...
$ graph-tool find-paths pn-depends.dot core-image-minimal util-linux
core-image-minimal -> packagegroup-core-boot -> udev -> glib-2.0 -> python-dbus -> dbus-glib -> dbus -> libsm -> e2fsprogs -> util-linux
core-image-minimal -> packagegroup-core-boot -> udev -> glib-2.0 -> python-dbus -> dbus -> libsm -> e2fsprogs -> util-linux
core-image-minimal -> packagegroup-core-boot -> udev -> util-linux

Partially addresses [YOCTO #3362].

(From OE-Core rev: 0b76f034dd0320ec545229872be8095c44ddee73)

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>

bluez4: use legacy pygobject instead of gobject-introspection

pygobject-3.0 use GObject from gi.repository instead of
gobject modules. Since oe-core use pygobject-2.x, change
all of "from gi.repository import GObject" to
"import gobject" for bluez4.

(From OE-Core rev: 9976a860b495256d94ec44c5084b13f214351547)

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

bluez4: Install the test script

(From OE-Core rev: c55918d73318fb86af122529bf1034826268356b)

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

libpcap: upgrade to 1.5.2

aclocal.patch was updated, still needed.

(From OE-Core rev: 89f9d07d8e291c994469d015d39ef90fb39a4ec4)

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

man-pages: upgrade to 3.55

(From OE-Core rev: 2d9de951d3271c18e721957575450eff81d85952)

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

lsbtest: add ppc64 support

LSB supports to test on PPC64 target, so update lsbtest recipe and test
scripts to enable PPC64 support.

(From OE-Core rev: 59905859064d4c7a09e04115daa4a93939a6dafc)

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

LSB_Test.sh: update locales

Add missing localedef statements to the LSB_Test.sh which are required
by the libstdc++ LSB test.

(From OE-Core rev: 827fc325f0462b93f360b31ac27ef15491034da3)

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

LSB_Test.sh: check ldconfig before update cache

LSB_Test.sh run ldconfig to update library cache. If command ldconfig
doesn't exists, test will fail.

Check whether ldconfig exists. If not, don't update library cache.

(From OE-Core rev: e3e5ebd6d05263bd7878537df93e9f1572f1172a)

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

LSB_Test.sh: update rpm platform file when enable multilib

When enable multilib for x86_64, lsb rpm packages for archs x86_64 and
i486 will be installed. But i486 target packages fail to be installed
that no such pattern in rpm platform file. Update platform file to
enable install i486 rpm packages as well.

(From OE-Core rev: 02d3233b57cda03dfef08983c7f9a03285444c83)

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

lsb: update directory of install_initd and remove_initd

According to LSB specification:

http://refspecs.linuxfoundation.org/LSB_4.1.0/LSB-Core-generic/LSB-Core-generic/installinitd.html
http://refspecs.linuxfoundation.org/LSB_4.1.0/LSB-Core-generic/LSB-Core-generic/removeinitd.html

commands install_initd and remove_initd should under directory /usr/lib/lsb/.

lsb recipe creates these commands under directory ${libdir} which may
expand to /usr/lib64 when multilib is enabled on qemux86-64. That will
cause LSB command check for install_initd and remove_initd fail. So
correct it.

(From OE-Core rev: f9c37768caf7edf9343f76f16fa5fd4e7cd772c1)

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

insane: handle recursive configures when checking for unknown configure options

Some recipes have configure scripts that recursively call other configure
scripts (e.g. dropbear).  These multiple-line matches were not being handled
correctly, so iterate over every matching line instead of assuming only one line
was found.

[ YOCTO #5646 ]

(From OE-Core rev: b226ab4cf7779f4dfaa78210cb6249766ed564c1)

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

busybox: get rid of nested functions in find

This allows us to compile busybox using clang. Nested functions is
a gcc extension not supported by clang.

(From OE-Core rev: d2bf599f36ed1a04c661fc0a71e664e219532cbc)

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

pulseaudio: only package consolekit module when x11 is enabled

As requested by Martin Jansa <martin.jansa@gmail.com>.

(From OE-Core rev: 3e148f863d55728bbfa2d94b602b03dc56b70d4c)

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

git: leverage upstream relocation support

(From OE-Core rev: 4fe4b432b7323a73144f317f8bbbd09c9d9fcb77)

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

libdrm: use RRECOMMENDS for libdrm-drivers

As libdrm enables modules automatically, based on what it needs / can build,
it's unlikely we'll *ever* get a situation where all the modules are emitted,
so libdrm-drivers, which depends on all of them, ends up being useless and
non-installable. Make it use rrecommends, so it only pulls in what we have.

(From OE-Core rev: faa5fc8c4d6cf236cd87d3308a2828d37da97742)

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

xz: make the LICENSE info more accurate

(From OE-Core rev: fb9b12121f97f59d92ec2b8fdbe0e68f336f0576)

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

nativesdk: Set LIBCOVERRIDE to a value

nativesdk was just unsetting LIBCOVERRIDE however that causes some build failures
for xorg-libs which used a libc override. This adds in a mechanism to force
nativesdk to glibc and give the option of allowing another selection like
uclibc if anyone ever does the work to make it operational.

(From OE-Core rev: 154f5782d95b4bca8e574d3d3fde170ce3d196ae)

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

gcc-crosssdk.inc: Fix missing dependencies (such as libmpc-native)

Without this sstate builds can fail with missing dependencies.

(From OE-Core rev: f92ebf78d94cb8f4010f8d444d1d0336c1fb1341)

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

useradd-example: add example for setting clear text password

Use '-P' option for user3 to set a clear text password. This is supposed
to serve as an example of how to set clear text password in recipes.

(From OE-Core rev: 6f47698e1ef8103cade8b954825848bb8986158c)

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>

image.bbclass: fix for zap_root_password

Previously, this function replaces the root password with '*' if
'debug-tweaks' is not in IMAGE_FEATURES. It not only zaps empty root
password, but also zaps non-empty root password. That means, if the
user uses a bbappend file for base-passwd to set the root password, he
would not be able to login as root; if the user uses 'EXTRA_USERS_PARAMS'
to set the root password, he would still not be able to login as root.

What we really want from this function is to disallow empty root password
if 'debug-tweaks' is not in IMAGE_FEATURES. This function should not remove
non-empty root password because that password is usually deliberately set
by the user.

This patch renames zap_root_password to zap_empty_root_password to
better reflect the intent of this function. It also modifies the code
to make this function work correctly.

(From OE-Core rev: c1037a74f934966a0df8c85138b09d672b9f8b36)

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>

shadow-native: allow for setting password in clear text

Allow user to set password in clear text. This is convenient when
we're building out an image.

This feature is mainly used by useradd.bbclass and extrausers.bbclass.

This patch adds a new option '-P' to useradd, usermod, groupadd and groupmod
commands provided by shadow-native. The shadow package on target and in SDK
will not be affected.

[YOCTO #5365]

(From OE-Core rev: 31dee7946340bf0f1e94e4e714191d3d6ca3bf6a)

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>

image.bbclass: default USE_DEVFS to '1'

Default USE_DEVFS to "1" so that the `makedevs' command is not run
at rootfs time by default. There are mainly two reasons to do so.

1. This will fix a build failure with initramfs-kexecboot-klibc-image.
   "makedevs: No entry for root in search list "

2. Most of our images use a filesystem over /dev. Most of the time, it's
   just devtmpfs. So we actually are using a filesystem over /dev.

(From OE-Core rev: f54fdd6673a136ee1cee1f3263a8a7820de43ca3)

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>

btrfs-tools: Update to 3.12

(From OE-Core rev: cd3df9fd31c206a7222bc227f545021241514eec)

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

lib/oeqa/runtime: systemd failure case improvements

If the hostnamed service can't be started or stopped, show the output from
systemctl status to assist debugging.

(From OE-Core rev: 024cc1cee3a9954272ecbbff1ba7153725c56dce)

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>

init-ifupdown: adjust priority of networking to fix system freeze problem

On mpc8315e, a system freeze is encountered at system boot time if
connman and init-ifupdown are installed.

The error message before the freeze is:
    "ip: RTNETLINK answers: File exists"

This problem is introduced by the following commit.
    dc80eea sysvinit: fix problem in switching runlevels

Part of the above commit is to make the networking init script run
at runlevel 2 3 4 5 instead of runlevel S. However, after the change,
networking is run after connman. And this causes the problem stated
above.

Make networking run first when entering runlevel 2 3 4 5, so that we
don't break things.

[YOCTO #5651]

(From OE-Core rev: d518892d38ac399c091ff509a9fd90fc00d71224)

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>

ltp: Use patch generated without -M

* in my builds both versions worked, but Saul reported that it fails to
  apply with:
  Applying patch
  0001-Rename-runtests_noltp.sh-script-so-have-unique-name.patch
  patch: **** Only garbage was found in the patch input.

  Now I've see the same issue on different builder (with Ubuntu 12.04).

(From OE-Core rev: ebc4c7384c7a355165838ee0935b986840f950e5)

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>

gstreamer1.0-plugins-base: fix missing rprovides

split_gstreamer10_packages creates libgst* packages which are missing in PACKAGES list.

(From OE-Core rev: 6432037e3089b6f92483fc343aac459a05ae2500)

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

glib-2.0: upgrade to 2.38.2

(From OE-Core rev: b445008484639d0d3e0e397019241466f4361b9e)

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