bitbake: fetch2/wget: Check downloaded file isn't zero size

I can't think of a reason we'd download zero sized files however there are
reasons zero length files can accidently make it onto source mirrors.

This check allows us to ignore the broken files and switch to another
mirror rather than fail with odd checksum failures.

(Bitbake rev: 300cba2e1a720dba4b83b0c76208ea93c608c1de)

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

libtool-cross/native: Force usage of bash due to sstate inconsistencies

Scenario:
a) libtool script is built on system with bash as /bin/sh
b) machine B installs sstate from build a)
c) machine B has dash as /bin/sh

In this scenario, the script fails to work properly since its expecting
/bin/sh to have bash like syntax and it no longer does have it.

This patch forces the configure process to use /bin/bash, not /bin/sh
and hence allows the scripts to work correctly when used from sstate.

(From OE-Core rev: 24d5b449e5f4d91119f0d8e13c457618811aadfc)

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

Revert "e2fsprogs/populate-extfs.sh: fix a problem on dash"

This reverts commit 22f90c5aec4f0b0360d1d960226f9965d83d589b.

This causes build failures with:

| dirname: missing operand
| Try 'dirname --help' for more information.

under some circumstances.

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

boot-directdisk: fix the support of vmdk

Previous change (086ce22b88f5ef5f75a83119a32c8b3fdcfa296d) broke
the creating of vmdk images. This protects shell expansion variables
and let dd generate the image to be transformed to vmdk by image-vmdk.class.

(From OE-Core rev: 2fe667afbdc6880c377657b5ff27e6db3b6cbe77)

Signed-off-by: Joao Henrique Ferreira de Freitas <joaohf@gmail.com>
[edit to change the usage of IMAGE_FSTYPE to IS_VMDK]

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

pybootchartgui: Add option -T to allways use the full time

When --full-time (or -T) is used, the graph allways shows the full
time regardless of which processes are currently shown. This is
especially useful in combinationm with the -s flag when outputting to
multiple files.

(From OE-Core rev: c6e88199ddf2c4ae243d42afc403d28ab56f00f0)

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

pybootchartgui: Adopt the width of the index in split output files

Add minimum width zero-padding to the index used in split output files
with -s and -o. I.e., if -s 200 is used, then the index will be
zero-padded to three digits width.

(From OE-Core rev: 45565b24651ab502ae49dc49261dc3ad5634191f)

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

pybootchartgui: Simplify adding processes to the trace

(From OE-Core rev: 5fa869007b5ba762bf5679197cf98b1d14a34a22)

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

pybootchartgui: Correct the legend

(From OE-Core rev: 1ca2e1a2ae3dc4d1e62a9daf25df588ec27a195a)

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

pybootchartgui: Make the -s option work again

[YOCTO #5588]

(From OE-Core rev: 8245ceab3acd02618f24665ff5dc203c1e5cce1d)

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

libtool: remove the unrecognized configure option

Remove the unrecognized configure option '--with-sysroot' to avoid
build time warnings.

(From OE-Core rev: 6f6a10372b6a318be7695b6b50275a8a3e9ed033)

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

cmake.bbclass: fix note when warning about deprecated variables

The note issues when OECMAKE_BUILDPATH and OECMAKE_SOURCEPATH were being used
stated that an in-tree build would be done, but the default is in fact an
out-of-tree build.

(From OE-Core rev: 0dafb9f78e9ab9ec1a1483efc37902c2e8de3623)

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

adt_installer: do not install rootfs if target is not selected

Currently, if YOCTOADT_TARGETS does not contain an architecture but the
rootfs/machine settings are uncommented, then the rootfs is installed
and adt will throw an error because is not able to find the toolchain
environment script.

This patch will:
 * not allow to install a target rootfs if the toolchain for the
   target architecture is not selected;
 * uncomment the target rootfs/machine settings for the other
   architectures since it's easier for the user to just add a new
   architecture in YOCTOADT_TARGETS and have the target rootfs
   installed;

[YOCTO #5727]

(From OE-Core rev: 22351d27de76b39cb71904396e73497e36d9e1d6)

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

lib/oeqa: sshcontrol: fix false timeout failures

Ocasionally AB shows odd false fails like:
http://autobuilder.yoctoproject.org/main/builders/nightly-arm/builds/1/steps/Running%20Sanity%20Tests/logs/stdio
This should fix that by checking for eof instead of
polling the return code of the ssh process, because the process
might still be there.

(From OE-Core rev: 3a22b5df5aa38a98b35bc2931d646a2b7702fbec)

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

gdk-pixbuf: use PACKAGECONFIG to control loaders with external dependencies

(From OE-Core rev: aaa3644a75a7698604102b3b68d40b4dcc02df1d)

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

gdk-pixbuf: don't forcibly disable GIO sniffing, use PACKAGECONFIG.

There's a configure option for GIO sniffing so don't use a patch to disable it.
Instead use a PACKAGECONFIG for this and default to off, as using GIO for
sniffing means a hard dependency on shared-mime-info.

(From OE-Core rev: 624e79deb58c08bcc32053e792df140f527d20b1)

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

gdk-pixbuf: upgrade to 2.30.3

(From OE-Core rev: 992c054c0656efce50d3902b3f0b101ab9ea602a)

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

libxcb: upgrade to 1.10

Remove spurious libxcb-xinerama addition to PACKAGES, this is handled by the
dynamic split_packages() now.

(From OE-Core rev: cfb0b1124d38c332536abb5d3e2726a3b8993140)

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

xcb-proto: upgrade to 1.10

(From OE-Core rev: d31d1a3e6cfada74bfe5e288e6af2de409fb7a6a)

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

glproto: upgrade to 1.4.17

(From OE-Core rev: aad0b0495ebdc3d0757ea664daa4053f91a8dbe4)

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

sstate.bbclass: remove previous version's stamp

There is a potential problem if we don't remove the previous version's
stamp, for example:

The depend chain is:
libtool-native -> autoconf-native -> m4-native
We have two m4-native: 1.4.9 and 1.4.7

1) Clean all of them to make a fresh build so that we can reproduce the
   problem
$ bitbake m4-native autoconf-native libtool-native -ccleansstate

2) Build libtool-native so that the m4-native_1.4.17 will be built
$ bitbake libtool-native

3) Set PREFERRED_VERSION_m4-native = "1.4.9" and build again
$ bitbake libtool-native

4) Set PREFERRED_VERSION_m4-native = "1.4.17" and build again
$ bitbake libtool-native -ccleansstate && bitbake libtool-native

Then the build will fail:
[snip]
| m4: unrecognized option '--gnu'
| Try `m4 --help' for more information.
| autom4te: m4 failed with exit status: 1
[snip]

The is because when we change m4-native to 1.4.17 and build
libtool-native again:
5) libtool-native depends on autoconf-native, and autoconf-native's
   version isn't change, so it can remove the current stamp and mirror
   the sstate (the one depends on m4-native_1.4.9) from the SSTATE_DIR
   correctly.

6) The mirrored autoconf-native depends on m4-native_1.4.17's
   do_populate_sysroot, and the stamp is already there (which is made
   by step 2), so it would do nothing, but this is incorrect, since
   the one that really in the sysroot is m4-native_1.4.9, then the
   error happens.

Remove previous version's stamp in sstate_clean() will fix the problem.

[YOCTO #5422]

(From OE-Core rev: 4659d29b1040349116549644e45035a5b37d9311)

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

e2fsprogs/populate-extfs.sh: fix a problem on dash

The dash can't handle the or [[ in parameter expansion, for example:

A=/usr/bin/[[
B=[[
C="${A%$B}"

The C should be "/usr/bin" in common, but it will be /usr/bin/[[ on
dash, use dirname to fix it.

NOTE:
There are 3 lines about parameter expansion, only fix the
DIR="${DIR%$TGT}" since the other 2 works will and are very useful in
this case.

[YOCTO #5712]

(From OE-Core rev: 22f90c5aec4f0b0360d1d960226f9965d83d589b)

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

linux-firmware: package Marvell SD8797 firmware

(From OE-Core rev: 8fcd5accbe8ffca9e0c5233419c1571741c76c03)

Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

guile: don't use the identifier 'noreturn'

Fix the build error of autogen-native which depends on guile-native:

ysroots/x86_64-linux/usr/include/guile/2.0/libguile/error.h:40:24: error: expected ')' before '__attribute__'
sysroots/x86_64-linux/usr/include/guile/2.0/libguile/error.h:40:24: error: expected ',' or ';' before ')' token
sysroots/x86_64-linux/usr/include/guile/2.0/libguile/error.h:42:27: error: expected ')' before '__attribute__'

[YOCTO #5743]

(From OE-Core rev: 05d226bee199e9d45f0bb6143d3a78f3f2e93186)

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

bitbake: fetch2/git: Dereference unresolved tags with ls-remote

We need to deference tags when trying to map them to commit IDs with
ls-remote. If we don't do this, a given commit might not show up
later in a specific branch. There appears to be no good reason not
to do this.

(Bitbake rev: 8ef24f4c834298348172b96ec0b855bf09552b09)

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

bitbake: fetch2/git: Anchor names when using ls-remote

When specifying tags, they're searched for unanchored so foo/bar could
match:

refs/heads/abc/foo/bar
refs/heads/xyz/foo/bar
refs/heads/foo/bar

This change anchors the expressions so they are based against heads
or tags (or any other base level tree that has been created).

(Bitbake rev: df2e0972cd1db7abd5ec8b7cb295fb0c42e284a4)

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

bitbake: fetch2: Improve invalid SRCREV error message

The current message can be ambiguous, improve it (and also rename a
variable to clean up the rest of the function).

(Bitbake rev: 0c1bb7c0fce7b0f334311a2893ccb00385fa8d55)

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

bitbake: fetch2: Sanity check SRCREV matches rev/tag parameter

Add a sanity check so that if some SRCREV is set and a rev parameter is given
to the url, the revision given should match.

Any tag parameter behaves the same as rev. If both are specified, error to
tell the user we're confused rather than do something which may or may not
be what they intended.

Also add some unittests for this.

(Bitbake rev: e82a4ab48991035866da9914c8b75a9bfbc9a7fc)

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

bitbake: fetch2: Clean up srcrev_internal_helper

Currently INVALID and None are checked as incorrect values under different
circumstances. This code standardises those checks to be consistent. We
should phase out the use of "INVALID".

(Bitbake rev: 86ef4e65ce18b71dc69643586bd2aa8f48703171)

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

package.bbclass: show warning when package is providing already provided shlib

* move read_shlib_providers before registering package as provider
  and show warning when different package tries to provide something
  already provided.

[YOCTO #4628]

(From OE-Core rev: 8141e3f61f12c8901b990496bcf6b76a9db95a57)

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

package.bbclass: move reading shlibs providers to separate function

* prepare for reading shlibs providers only from dependency tree of
  current recipe

[YOCTO #4628]

(From OE-Core rev: c5076f33ac27c0c2b0743bf6dc4edc983254c467)

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

icecc: use exact match in blacklists, re-start with empty system_package_blacklist

* unify debug messages a bit
* old implementation allowed partial match in blacklist, it's safer
  to explicitly list exact matches
* I was able to build all entries from system_package_blacklist with
  icecc enabled, lets assume that they were already resolved by newer
  versions (we've fixed a lot of parallel issues in recipes which were
  detected even without icecc and this list is very old).

(From OE-Core rev: 5a5319d2e6f41bb0e290d6a1decbd996e9572690)

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

icecc: Fix allarch and native recipes having different signatures

* for different MACHINES
* is there more elegant way to have "overridable" function so that
  signature handler properly uses only the branch without
  STAGING_BINDIR_TOOLCHAIN?

(From OE-Core rev: 418a353a011ca8f04ecc3e2d29f2d1a415492081)

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

icecc: use bb.utils.which also for 'as'

* it was introduced in
  commit 3a842ec52e7d010767b13bdcb5629ac07b3ee9e7
  Author: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
  Date:   Fri Sep 16 10:55:16 2011 +0400
  Subject: icecc.bbclass: replace with updated version

  without any explanation in which case
    ${ICECC_CC} -print-prog-name=as
  is returning as in current working directory, but will keep old
  behavior just in case

(From OE-Core rev: 6092da20fc3ceb1bc6b4872ad16df565f05723b7)

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

icecc: Don't replace non-empty PARALLEL_MAKE with empty ICECC_PARALLEL_MAKE

* it's needed for use-case like this:
  # Inherit icecc here, so that all builders have the same sstate signatures
  INHERIT_DISTRO += "icecc"
  # and then disable its function by default (so that people still need to explicity
  # enable it in local.conf if they have configured icecc and want to use it.
  # You need to set _empty_ value in local.conf to enable icecc function:
  # ICECC_DISABLED = ""
  ICECC_DISABLED ??= "1"
* so default ICECC_PARALLEL_MAKE is still empty, but we want build
  to respect our PARALLEL_MAKE, unfortunately we cannot do something
  like ICECC_PARALLEL_MAKE ??= "${PARALLEL_MAKE}", because that would
  cause PARALLEL_MAKE to reference itself.

(From OE-Core rev: 7e586d5b7c8c7f20eafc32624200f60a8ed9a582)

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

allarch: Set empty TARGET_PREFIX and TARGET_FPU

* set empty TARGET_PREFIX
  This has a bit weird reason caused by unsupported setup where
  external-toolchain is used in some DISTRO only for some MACHINEs
  and internal is used for other MACHINEs.
  Because external-toolchain usually comes with different TARGET_PREFIX
  it was causing allarch recipes to have different signatures even
  when they don't use toolchain at all.
  Empty TARGET_PREFIX also helps to find allarch recipes which still
  have default dependency on e.g. virtual/${TARGET_PREFIX}gcc.
* add TARGET_FPU just for completeness (it was used in icecc.bbclass
  but now it's vardepexcluded there as well)

(From OE-Core rev: 180ba7f6603b35eb66946649d8860022d2c329b7)

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

sstatesig: include native/cross/nativesdk deps in target signatures

* I don't have any real evidence or good statistics for this, but when
  comparing signature dumps from my big bitbake world builds I usually
  see a lot of rebuilds caused by changes in .bbclasses and only very
  rare would be the case where oe-core upgrade brings changes in -native
  recipes and no change in .bbclasses used from target recipes
* changing the default to include them shouldn't cause significant
  increase in rebuilds and sstate reuse a bit safer
* people working on toolchain (e.g. using gcc from AUTOREV) can easily
  extend sstate_rundepfilter to ignore them again (it's easier than
  removing existing filter), example how add own signature handler in
  your layer is here:
  https://github.com/openwebos/meta-webos/commit/9ac3a7c803e7793b3274e4998f167b6278db8042

(From OE-Core rev: 336a7897e39b9e42dcfcba9e2520ea96b0c6a8d6)

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

opkg-utils: Remove strange PACKAGES native override

This code makes no sense, native.bbclass clears PACKAGES anyway. Drop
it.

(From OE-Core rev: 9c8d9781794ed0886a79c8ce4544ba98be0ff858)

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

opkg-utils: Add basic PACKAGECONFIG for python dependencies

In small configurations its useful not to have python dependencies. This
patch adds code to disable those using PACKAGECONFIG. This allows us to
fix poky-tiny after the recent move of update-alternatives to opkg-utils.

(From OE-Core rev: 0c0039048fd97cdd3cf28f9e1c829a716dfd68ea)

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

glib-2.0: add explicit build dependency on DBus when ptest is enabled

If the tests are enabled then configure will check for the presence DBus.  It's
generally present through the runtime dependencies so this often succeeds but as
it isn't a build dependency it's possible for DBus to be present at configure
time but removed at compile time, resulting in build failures.

(From OE-Core rev: 0150e86a7609579cf26f5ef5c6c69b521340218e)

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

oe-selftest: Patch sstate tests to work with new sstate-cache naming and content

Made modifications to account for:
- .siginfo files present in sstate-cache from non sstate-enabled tasks
- new naming format for sstate files

(From OE-Core rev: d23ad4255ad7465383286fa7805392aa70845ace)

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

tcl: fix the TCL_INCLUDE_SPEC

We have moved the header files to ${includedir}/tcl${VERSION}, but we
didn't fix the TCL_INCLUDE_SPEC which is still ${includedir}, it should
also be ${includedir}/tcl${VERSION}

Note: this commit modifiey alter-includedir.patch, so it doesn't look
very clear, I only fixed one line in both configure and configure.in:

-eval "TCL_INCLUDE_SPEC=\"-I${includedir}\""
+eval "TCL_INCLUDE_SPEC=\"-I${includedir}/tcl${VERSION}\""

The other changes are because I use git to create the patch while the
previous one uses svn.

[YOCTO #5732]

(From OE-Core rev: a263281d9c774a03cce76caa3cc477cc85a9c2a9)

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

opkg/opkg-utils/chkconfig: Clean up u-a-cworth references

Catch some u-a-cworth references that slipped through the move of u-a
to opkg-utils and its rename to -opkg.

(From OE-Core rev: a9ff0bbac5ae0688525c71f0a358f0750a277269)

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

update-alternatives: We now use u-a in nativesdk so make sure the dependency is present

When update-alternatives was part of opkg which got built in most nativesdk
scenarios, this missing dependency wasn't an issue. We now need nativesdk-opkg-utils
so we need to ensure the dependency is present in nativesdk cases.

This avoids build failures with the recent u-a move to opkg-utils.

(From OE-Core rev: 7e0adf676da45e49287b7ce6478a6dbfd8fa117f)

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

linux-yocto-dev: bump version to 3.13+

(From OE-Core rev: 0e0307a23d50b2a0efd228e7e99baabc980ab9de)

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

lib/oeqa: allow a layer to provide it's own TEST_TARGET class

Allows a layer to define new classes in <layer>/lib/oeqa/utils/controllers.py
and completely control or extend deployment of a target. (core currently
has QemuTarget and SimpleRemoteTarget).
The value of TEST_TARGET must be the name of the new class.

(From OE-Core rev: 9b81aff0aca42353d448b1e9522f89842e23c7b2)

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

x11vnc: fix CAPS_LOCK issues

Currently, pressing CAPS_LOCK on the viewer changes the lock state on
the server and the key will not change the case.

To fix this, use -skip_lockkeys option to ignore all Caps_Lock,
Shift_Lock, Num_Lock, Scroll_Lock keysyms received from viewers, in
order to leave the lock state on the server side unchanged. However, the
keys will appear correctly on the remote side.

[YOCTO #4149]

(From OE-Core rev: 1e06d5ce83439b5bd75a958f305e6a880d40333d)

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

weston: depend on drm for 'launch' packageconfig enabled

configure fails with:

| configure:15654: checking for WESTON_LAUNCH
| configure:15661: $PKG_CONFIG --exists --print-errors "libdrm"
| Package libdrm was not found in the pkg-config search path.
| Perhaps you should add the directory containing `libdrm.pc'
| to the PKG_CONFIG_PATH environment variable
| No package 'libdrm' found
| configure:15664: $? = 1
| configure:15678: $PKG_CONFIG --exists --print-errors "libdrm"
| Package libdrm was not found in the pkg-config search path.
| Perhaps you should add the directory containing `libdrm.pc'
| to the PKG_CONFIG_PATH environment variable
| No package 'libdrm' found
| configure:15681: $? = 1
| configure:15695: result: no
| No package 'libdrm' found
| configure:15711: error: Package requirements (libdrm) were not met:
|
| No package 'libdrm' found

and configure.ac says:

| if test x$enable_weston_launch == xyes; then
|  PKG_CHECK_MODULES(WESTON_LAUNCH, [libdrm])

(From OE-Core rev: 6d0625a5a2ddd34760b2f85366810f20ad4ea15e)

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

opkg: No longer PROVIDES update-alternatives

The new provider is the opkg-utils recipe.

(From OE-Core rev: c63d241d15608ac431a1d716ec84ef568928cf85)

Signed-off-by: Paul Barker <paul@paulbarker.me.uk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

default-providers: Change update-alternatives provider to opkg-utils

This allows dependencies to be added to the opkg recipe without causing circular
dependency loops. As opkg-utils has minimal dependencies it is the best recipe
to provide update-alternatives.

This partially solves Yocto Project issue 4836. More work is still needed for a
complete solution.

(From OE-Core rev: 2f18289493f9c2c67ba343fb8e16743bf5dfee24)

Signed-off-by: Paul Barker <paul@paulbarker.me.uk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

opkg-utils: Upgrade to latest git HEAD

The latest version of opkg-utils PROVIDES virtual/update-alternatives via a
script copied from opkg, packaged as update-alternatives-opkg.

The nativesdk class is added to the recipe to ensure that
nativesdk-update-alternatives-opkg is built.

(From OE-Core rev: 1e2c38ce13f8e4b25d8656d237343380cbc970aa)

Signed-off-by: Paul Barker <paul@paulbarker.me.uk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wpa-supplicant-2.0: don't exit in pkg_postinst

Exiting explicitly in pkg_postinst makes it impossible to use the
update-rc.d class in a .bbappend because the link creation is appended
to the pkg_postinst script.

(From OE-Core rev: 758d53d3044f29f3c33ffee3ada88c9edc9f864f)

Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glib-2.0: fix mapped file ptest

The mappedfile ptest was attempting to write into directories that may not exist
on a non-Sato image.  Instead, write into TMPDIR.

(From OE-Core rev: 93b4ecc263947826421d5b4fb1f99fe506b24287)

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

glib-2.0: fix the dbus-appinfo test when running outside of X

This test was attempting to connect to a session bus and hanging if it couldn't.
Take a patch from upstream to correctly use a private bus.

[ YOCTO #5696 ]

(From OE-Core rev: 12e30b653899567312a7ec29b6e972c85ba8e25e)

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

bitbake: fetch2/__init__.py: let try_mirror_url return correct value

The fetcher will try:

1) PREMIRROR
2) Upstream
3) MIRROR

If it fails to download from the Upstream, but succeeds from the MIRROR,
and ud.localpath != origud.localpath (for example, the git tarball),
then we will get the error (e.g.: xf86-video-omapfb):

ERROR: Function failed: Fetcher failure for URL: 'xxx'. Unable to fetch URL from any source.
ERROR: Logfile of failure stored in: /path/to/log.do_fetch.28024

It should not show the error and let the build go on since it succeeds.
(e.g.: xf86-video-omapfb)

[YOCTO #5686]

(Bitbake rev: c08ca1e4eeb04f78e1354780cf5a4c3855e49572)

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

bitbake: SignatureGenerator: Add empty implementation for dump_sigs

* bitbake -S throws exception when 'noop' signature handler is used

[YOCTO #5738]

(Bitbake rev: 53352e8d388b7fc4da73f95b93dcc087e76d0426)

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

bitbake: fetch2/gitsm: Use ud.basecmd instead of hardcoding git

This allows FETCHCMD_git to override the fetcher command as the git fetcher does.

[YOCTO #5717]

(Bitbake rev: 23ab943be3a33077d6ad8be68bba53cd1e2270b4)

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

bitbake: fetch2: Don't allow '/' in user:pass, fix branch containing '@'

* currently decode_url regexp parses branch=@foo as username so it ends like this:
  - ('git', '', 'foo', 'git.openembedded.org/bitbake;branch=', '', {})
  + ('git', 'git.openembedded.org', '/bitbake', '', '', {'branch': '@foo'})
* http://hg.python.org/cpython/file/2.7/Lib/urlparse.py also assumes
  that there is at least one '/' as separator between netloc and path,
  params, so it looks reasonable to prevent including '/' in username

(Bitbake rev: 2c82742114091cb55055328b54223686816582f2)

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

bitbake: siggen: Remove fsync that is causing performance issues

This fsync was added for belt and braces protection for things like sstate
on NFS. To be honest, it probably doesn't buy much, if the rename isn't
atomic, all bets are off anyway and there are bigger issues to worry about.

The issue is that at the end of every task, the dump_sig() code is triggered
to save out information about the task and this was triggering an fsync(fd).

Whilst it may select the file descriptor, on file systems like ext4, it will
require large parts of the journal to be written out so it can have
significant impact. latencytop showed an average fsync() call overhead of about
2s and if that happens for 5000 tasks, the time mounts up. This blocks the next
task execution by that time.

We therefore drop the fsync since in reality its causing problems and is unlikely
to buy much.

(Bitbake rev: 46fd841319479f6079d850b3813e64bd8c2680a3)

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

oe-selftest: separated the SStateBase and SStateTests in different modules

- SStateBase now has its own module to be imported by itself by other modules like sstatetests.py

(From OE-Core rev: 8163854adf87ac42a8f08ee25685d0ce1efb4724)

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

oe-selftest: renamed sstate.py module to sstatetests.py

(From OE-Core rev: 91115cf06b009427a444abfd4cf0317ae5215c6b)

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

oe-selftest: New object SStateBase cut off from SStateTests.

- SStateBase object contains basic methods used to run sstate related tests
- SStateTests now contains only sstate-related tests

(From OE-Core rev: 78e929bbcdce1f9a544b230433b93f4fc1f841e2)

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

util-linux: add reset.1 manpage in update-alternatives

Adding "doc-pkgs" to IMAGE_FEATURES (in an image recipe) reports a conflict with reset.1 file, present both in util-linux and ncurses-doc packages.

<log>
| Collected errors:
|  * check_data_file_clashes: Package util-linux-doc wants to install file /home/matt/tmp/oe-p/build/tmp-eglibc/work/qemux86_64-tiny-linux/test-image/1.0-r3/rootfs/usr/share/man/man1/reset.1
|       But that file is already provided by package  * ncurses-doc
|  * opkg_install_cmd: Cannot install package util-linux-doc.
| WARNING: /home/matt/tmp/oe-p/build/tmp-eglibc/work/qemux86_64-tiny-linux/test-image/1.0-r3/temp/run.do_rootfs.13877:1 exit 255 from
|   opkg-cl -f $INSTALL_CONF_IPK -o $INSTALL_ROOTFS_IPK --force_postinstall --prefer-arch-to-version install `cat $1`
| ERROR: Function failed: do_rootfs (log file is located at /home/matt/tmp/oe-p/build/tmp-eglibc/work/qemux86_64-tiny-linux/test-image/1.0-r3/temp/log.do_rootfs.13877)
</log>

Same issue already occured few months ago:
Re: [OE-core] Clashing man pages
https://www.mail-archive.com/openembedded-core@lists.openembedded.org/msg38590.html

(From OE-Core rev: 67499dbffd0f7241fd199b7fb94edfe2cebe8a9b)

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

initrdscripts: add $CMDLINE to init-live switch_root

init-live.sh: $CMDLINE variable should be provided to switch_root
to let user specify runlevel on grub command line.
Feeding with -c /dev/console as well as busybox switch_root enables that option.

(From OE-Core rev: 5a2f4feeaac4f9278fa0cf808c2f495f0c19324f)

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

package_rpm: Fix a race with package_deb/package_ipk

We have the odd situation where the CONTROL/DEBIAN directory can be removed
in the middle of the walk, the isdir() test would then fail and the walk code
would assume its a file hence we check for the names in files too.

This resolves the autobuilder failure:

error: File not found: /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-fsl-arm/build/build/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/xinit/1_1.3.3-r0/package/DEBIAN

RPM build errors:
     File not found: /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-fsl-arm/build/build/tmp/work/armv7a-vfp-neon-poky-linux-gnueabi/xinit/1_1.3.3-r0/package/DEBIAN

(From OE-Core rev: e38d7702be279d6d6d4c79b3f2379e689a7473d2)

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

alsa-lib: remove hard-coded /usr/include in dev package

Specify the location of the alsa headers using ${includedir}. Without
this no headers get added to the image when the alsa-lib-dev package is
included.

(From OE-Core rev: 720d30aa5f0745185fc948376ce284cc17f0edb0)

Signed-off-by: Andrew McDermott <andrew.mcdermott@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-firmware: add firmware for additional iwlwifi devices

(From OE-Core rev: 3654d1383cc71389d84e063baf2ecf505cd8b9a3)

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

nativesdk-packagegroup-sdk-host: reverse inherit order

Currently, tow successive toolchain builds for different hosts, will
issue a warning:

WARNING: The recipe nativesdk-packagegroup-sdk-host is trying to install
files into a shared area when those files already exist. Those files and
their manifest location are:
   /ssd/work/yp1/build/tmp/deploy/ipk/all/nativesdk-packagegroup-sdk-host_1.0-r11_all.ipk
   Matched in manifest-x86_64-nativesdk-packagegroup-sdk-host.package_write_ipk
Please verify which package should provide the above files.

That's because packagegroup is manually inherited after nativesdk which
is usually a BBCLASSEXTEND operation, done last.

[YOCTO #5396]

(From OE-Core rev: 7621a40d24f3e53b373f233e070894e4afc2b4b2)

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

allarch.bbclass: set SDK_ARCH and SDK_CC_ARCH to "none"

allarch packages shouldn't use these variables.

[YOCTO #5396]

(From OE-Core rev: c7253ad1b00fcef8fe0aeaab06d2fe76c82ba0dc)

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

glib: add missing dependencies for ptest

Some of the GLib tests that we run under ptest are missing runtime dependencies.

iochannel-test needs the EUC-JP encoding.

convert needs the ISO-8859-1 and ISO-8859-15 encodings.

contenttype needs shared-mime-info.

Add these to the RDEPENDS so that these tests pass.

[ YOCTO #5696 ]

(From OE-Core rev: 7b6316093ab28782edd45084d43dbd5c309be7c8)

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

weston: cleanup mesa depends

* do not depend by default
* depend on virtual/mesa instead of mesa

(From OE-Core rev: d8d02cb15f9d64f0e773223965e3ff6b6ae0874f)

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

linux-yocto/3.10: mohonpeak bsp config and scc files

(From OE-Core rev: 9f086987f995729235601281161858d851b8df72)

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

linux-yocto/3.10: update to 3.10.25

Updating the 3.10 tree to the 3.10.25 korg -stable release.

(From OE-Core rev: 85ec7e0bb1306b3ac3506627a3aa04a42c0ec180)

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

linux-yocto/3.10: update meta data for media fragments

Updating the meta SRCREV to import the following changes:

   d9cd83c0292b remove old MEDIA config fragments
   06b76256d7e2 common-pc-standard.scc: Enable USB webcam support
   acb8b43837d8 common-pc-64.scc: update as per changes in the media config fragments
   5513fd2ad72a minnow-standard.scc: Enable media features
   172ba799bedc media-all.scc: A feature including all the media features
   1a7e1d3a292e media-platform: A feature for platform media devices
   03c48dacbb9a media-dvb-frontends : A feature for Digital Video Broadcast Devices
   59b92b9d6c72 media-usb-tv: A feature for USB TV media adapters
   5ec0709b1fed media-tuners: A feature for media tuner devices
   356dc83e39f3 media-rc: A feature for remote control media devices
   89d96cf9d574 media-radio: A feature for AM/FM radio devices
   3dd2ebeaf49b media-pci-capture: A feature for PCI media capture devices
   9ee0e95bfc52 media-i2c: A feature for I2C media devices
   28976f4a3e27 media-usb-webcams: A feature for USB media devices
   f8206f4e00b1 media: A feature for media infrastructure
   08bfb248a17f standard.scc: Add firmware loading feature
   e567a3d53593 firmware: A feature for firmware loading support

(From OE-Core rev: f4b00c4ddf1efb2a728b7a08b542c9c7254c787f)

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

ltp: fix makefile race

There is a Makefile dependency race causing occasional build failures:

*** No rule to make target `.../work/core2-poky-linux/ltp/20130904-r0/git/testcases/kernel/include/linux_syscall_numbers.h.23161.sh', needed by `linux_syscall_numbers.h'.  Stop.

This is due to a bad dependency wildcard that is matching more than the one file
it should match, so replace it with a concrete filename.

(From OE-Core rev: 639ecfb578ff926f5deddb984f4f8600e161e22a)

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

dbus-ptest: fix incorrect option name

(From OE-Core rev: e27aee379c5f254f47daeb2fcb0a4a177740ea01)

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

systemd-compat-units: remove dnsmasq from sysv disabled list

The dnsmasq recipe in meta-networking ships a dnsmasq.service file, that will
correctly override the SysV init script. Thus, as pointed out by Ross Burton,
we should remove dnsmasq from the list.

(From OE-Core rev: 69a8153c55bf74fa6178d6bd65ac8e761c041a0e)

Signed-off-by: Anders Darander <anders@chargestorm.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

shutdown-desktop: use poweroff for the Exec field in the shutdown.desktop

Use 'poweroff' instead of 'halt' for the Exec field in the shutdown.desktop.

The purpose of this patch is to make the 'shutdown' icon on on our sato
images work as expected for both sysvinit and systemd images. Previously,
the Exec field was 'halt'. The `halt' command could poweroff the system
in sysvinit images but it only could halt the system in systemd images.
The difference is due to the different implementations of the `halt' command.

In sysvinit, the `halt' command will effective execute `/sbin/init 0'. This
is for the compatibility with sysvinit 2.4, as stated in the comments of the
source code. In systemd, the `halt' command will effectively execute the
reboot(RB_HALT_SYSTEM).

As the 'shutdown' icon is expected to actually shutdown the system, we should
use poweroff instead of halt for its Exec field.

[YOCTO #4347]

(From OE-Core rev: c87d44a079c13a98734f9f726ec1b55d86793f8c)

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

dbus-ptest: disable systemd

Explicitly disable systemd support so that it isn't a floating dependency and
libsystemd-login doesn't become a runtime dependency on non-systemd images.

Also don't bother setting the systemd unit directory as we're not installing
anything.

(From OE-Core rev: 4f5d00b551e2ee6056ebfc8365a05a5ca8e5d651)

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

dropbear: Refresh pam patch so it applies against recent version

Patch application failed on the autobuilder for pam, this refresh of the
patch should resolve the build failure.

(From OE-Core rev: c4c5ec52effc2ff97ac17270c1aa7884c808f5a9)

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

eglibc-locale: Fix multilib builds to depend on the correct binutils

(From OE-Core rev: 858c60adbcc5e21c585383fe90f6803d52f0807f)

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

python-nose: don't install as an Egg

Nose is installed as an Egg by default which we don't want, so change the
inherit to setuptools from distutils to disable the eggification.

(From OE-Core rev: 47a07c42f5f67dcf69f844531f7a527029e6280e)

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

eglibc-locale: Fix previous dependency change to properly work in nativesdk case

(From OE-Core rev: 7a4f3ee1b137cc8465f0cd9b1e461d3643182a81)

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

python-nose: Add BBCLASSEXTEND for native/nativesdk

Without this, nativesdk-python-numpy doesn't build.

(From OE-Core rev: 78711c887520c88d733182a80edac49650a7e94f)

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

harfbuzz: Move ancillary binaries to ${PN}-bin

The binaries:

-rwxr-xr-x root/root     21956 2014-01-09 19:39 ./bin/hb-shape
-rwxr-xr-x root/root     21504 2014-01-09 19:39 ./bin/hb-ot-shape-closure
-rwxr-xr-x root/root     36520 2014-01-09 19:39 ./bin/hb-view

are not generally useful on target systems.  Also, their presence in the
same binary package as libharfbuzz.so defeats the debian package autonamer.
Inherit lib_package to move these to ${PN}-bin.

(From OE-Core rev: d716f63d5785fa9239497d25bffc5035a9e58d28)

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

security_flags: db can't use pie flags from gcc for security build

[YOCTO #5721]

(From OE-Core rev: 0cfe254e7eafed27f512216cccfb7fee76fc0be7)

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

packagegroup-core-tools-testapps: add Piglit

Respect the OpenGL distro feature and if it's enabled, pull in piglit and
mesa-tools.

(From OE-Core rev: cd61c62ed30694eeb82e88e058a3a3e1bfb06f78)

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

piglit: add (from meta-oe)

Piglit is an OpenGL testing tool.  This recipe is taken directly from meta-oe.

(From OE-Core rev: cab8dad6f3b14f21115a765a4e55d38a0f73f337)

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

waffle: add (from meta-oe)

Waffle is needed for Piglit.  This recipe is based on the recipe in meta-oe,
upgraded to the latest upstream 1.3.0.

(From OE-Core rev: 67228f682cb8acc71dd5f8b44af67f529727f780)

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

python-numpy: add (from meta-oe)

python-numpy is needed for Piglit.  This recipe is taken directly from meta-oe.

(From OE-Core rev: 9bf355cceaec6ebacdcbcc35f9713ff73e1c85da)

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

python-nose: add (from meta-oe)

python-nose is a build dependecy for python-numpy, needed for Piglit.  This
recipe is taken directly from meta-oe.

(From OE-Core rev: 6b0c3e9955292d2ea13c6871c686824cd04150f1)

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

python-mako: add (from meta-oe)

python-mako is needed for Piglit.  This recipe is taken from meta-oe, and
upgraded to the latest upstream version.

Checksums updated as the upstream LICENSE file contains the copyright dates.

(From OE-Core rev: 1ad759508d834f3bcb3bb8542ae2012a2b6ef2e0)

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

cmake: specify all install paths

Specify the full set of install paths (bindir, libdir, etc) for packages that
use the GNUInstallDirs module, instead of just the prefix and leaving the rest
as default (which breaks with multilib).

(From OE-Core rev: d3995ac14ba05c0420f15f264f7e9d0c3af71f74)

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

separatebuilddir: Drop cmake lines since cmake class has this as default now

The changes to cmake make this unneeded now.

(From OE-Core rev: 92472980b816ee9ada502c1965976cb6eedc0a27)

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

grub-efi: remove configure patch, simply seed the cache

grub-2.00-disable-help2man patches configure.ac to disable the help2man check
because we don't need the man pages.  It also then patches configure itself so
that autogen.sh doesn't have to be called for performance reasons.  However,
do_configure causes a full autoreconf so this optimisation is moot, and can
cause patch failures when an existing build tree is re-used.

Instead, simply use CACHED_CONFIGUREVARS to tell configure that it can't find
help2man.

(From OE-Core rev: 34b9b2931da1fbf4c68a3e35f4ca56553058574b)

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

alsa-tools: fix the AUTOMAKE_DIR

There should be only one automake under the STAGING_DATADIR_NATIVE in
theory, but der_steffi@gmx.de has reported an odd problem which seemed
like that there are more than one. However, the "automake
--print-libdir" is the regular way to locate automake libdir.

[YOCTO #5706]

Reported-by: der_steffi@gmx.de
(From OE-Core rev: 59f46f9c9f8a36d8829b9b33291249a7b92ebcac)

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

nativesdk-packagegroup-sdk-host: Cleanup strange darwin override

The cross-canadian packages were split into their own package group
some time ago. These two are leftovers and can be safely removed.

(From OE-Core rev: 0f05393bad8ac84627f339aa1619c96c22d2dc37)

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

ptest: d.keys() is slow, use a list instead

Unfortunately d.keys is extremely slow. Using a list in this case should be
fine since the addtask lines are immediately above the code and aren't
going to change often.

(From OE-Core rev: 5901189f0442cefc4d17623547f7358ffc2d37a6)

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

eglibc-locale: Fix depends on binutils

The dependency here needs to apply for nativesdk as well as target packages
as the autobuilder just tripped over that. We'd never want a native version
so I'm not sure why the target class override was even present. The dependency
also applies to do_package so lets be explicit about that in case sstate
decides to get clever.

(From OE-Core rev: b7ec21ac8ebac9d7fba34d6f11d93ecb8f561ca8)

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

make: Disable floating guile dependency

This makes builds of make deterministic.

(From OE-Core rev: 9ab5e5f16cb485f8838ba886216912a78cee01e2)

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

image_types bbclass: add support for tar.lz4 and cpio.lz4

(From OE-Core rev: 0a541a0497902ee9993449b47da853b6c6a306d9)

Signed-off-by: Koen Kooi <koen.kooi@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>