cpan_build: Unify directory layout for native and target builds

There seems to be no benefit in having them be different and this
appears to cause a certain amount of confusion about paths for
the native modules.

(From OE-Core rev: 3926a45a26c2d23d9c2fc9c3956f780f607ec7a4)

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

Add packagegroup-core-buildessential

[Not heavily tested, but sent for discussion]

task-core-sdk is too big - for example, I simply don't want to ship tcl, ever =)
Historically distcc caused a large dependency graph explosion because it has
a utility which uses gtk+, although that was fixed.

packagegroup-core-sdk also includes coreutils, which is a bit
confusing; conceptually things dependent on coreutils should pull it
in explicitly, or possibly we just declare coreutils to be in the
minimal build set.

So packagegroup-core-buildessential is intended to be similar to
Debian's "build-essential" package.  It's the stuff needed by say 80+%
of components, not worth repeating over and over.

(From OE-Core rev: 7d6cc169c95fecf6388a275281eb8b8f5d8eb4a2)

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

libtool: Ensure the paths to sed are not hardcoded

If you:

bitbake sed-native
bitbake libtool-cross

then libtool-cross has SED="/path/to/sysroot/sed" which is incorrect. If that
is reused from sstate or sed-native is cleaned, the build will fail.

This patch simply sets sed to be "sed" since we're not on systems where
the sed from PATH is broken.

(From OE-Core rev: 86c6fa8175482283268dfa8782c6643a3510e0fd)

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

pkgconfig: Drop automatic pkgconfig RDEPENDS

Just because a recipe uses pkgconfig, it doesn't mean that it's -dev
package should RDEPENDS on pkgconfig. I can understand the thinking
that lead to this but it makes sense to be able to install the package
when pkgconfig hasn't been built.

Currently you can also get failures where pkgconfig wasn't built yet
a -dev package is included that depends on it leading to rootfs failures.

I considered making this a RRECOMMENDS but it should probably be an
RSUGGESTS at best given the tenuous pkgconfig requirement any given
-dev package has.

(From OE-Core rev: 8f41b69578eef5ea750e8f93dcd9d37375ce7d88)

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

package: Recommend virtual-locale-*, don't depend on it

The virtual-locale-* packages are provided by libc which may or
may not have a matching locale for any given recipes's provided locales.
Certainly, we shouldn't get a failure if the locale package isn't
available.

This patch therefore makes the dependency optional rather than
required, allowing the locale packages in question to install. This
resolves errors like:

virtual-locale-eo is needed by bash-locale-eo-4.2-r5.i586
virtual-locale-en+boldquot is needed by bash-locale-en+boldquot-4.2-r5.i586
virtual-locale-en+quot is needed by bash-locale-en+quot-4.2-r5.i586

(From OE-Core rev: 2be67f95abaa7e8655a1ca8f4ffe66b3e099a650)

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

libc-package: Drop bogus replacement operation

The names used to generate the binary-localdata packages need to match the location
the dependencies are added. In one case the dash replacement is made, in the other it
is not leading to packages which cannot be installed:

eglibc-binary-localedata-af-za.iso88591 is needed by locale-base-af-za.iso-8859-1-2.16-r22.i586
eglibc-binary-localedata-cs-cz.iso88592 is needed by locale-base-cs-cz.iso-8859-2-2.16-r22.i586
eglibc-binary-localedata-ru-ru.koi8r is needed by locale-base-ru-ru.koi8-r-2.16-r22.i586
eglibc-binary-localedata-pl-pl.iso88592 is needed by locale-base-pl-pl.iso-8859-2-2.16-r22.i586
eglibc-binary-localedata-hu-hu.iso88592 is needed by locale-base-hu-hu.iso-8859-2-2.16-r22.i586
eglibc-binary-localedata-de-at+euro.iso885915 is needed by locale-base-de-at+euro.iso-8859-15-2.16-r22.i586
eglibc-binary-localedata-sv-fi.iso88591 is needed by locale-base-sv-fi.iso-8859-1-2.16-r22.i586

This fixes things so the names are consistent.

(From OE-Core rev: 17e1bfe2b1260add9749b5ff73c72d57c0215fdc)

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

gitignore: Fix for poky repository

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

gdbm: Resolve host contamination issue

The autoconf macros detect the presence of -ldbm or -lndbm on the host
system and add the library to link against, for now good reason I can
explain.

This patch makes the build behave determinstically whether they're
present or not. Other than the extra linkage, there doesn't appear to be
any other change in behaviour from these options and they look like
dead code.

The extra linkage can cause problems where sstate is used on a machine
where the extra librbary isn't present causing build failures.

(From OE-Core rev: f609bf5525450bfdb8e0864d44c41cce7f9319c9)

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

pkgconfig: Drop the RREPLACES for pkgconfig-dev

This line causes pkgconfig-dev to replace pkgconfig so the package with all the files
in is replaced by one with no files. This makes no sense and hence we should just
remove this broken line.

At this point in the release, this is the safest way to fix this even if an empty -dev
package is left available.

[YOCTO #2878]

(From OE-Core rev: 5bed2bb831b379a8fbf2f725435af4b7c934359e)

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

eglibc: force make to use /bin/bash

The eglibc core build still has a number of issues with /bin/dash.
Recently found is both sysdeps/unix/make-syscalls.sh and it's output,
which make calls via SHELL do not play well with /bin/dash. By force
make to use /bin/bash via SHELL setting it works correctly.

Currenly known issues are: make-syscalls.sh line with a bad substitution,
which can be corrected by:
     vdso_symver="$(printf '%s\n' ${vdso_symver} | sed 's/\./_/')"

Following that there is an issue with emiting a '\n' through a second
echo and then to the compiler. There maybe more issues beyond that.

[YOCTO #3080]

(From OE-Core rev: 9d002f7cdc5309c4d850a76e4fd73ff04c980a07)

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

perl: add archlib_exp variable used to generate ARCHLIB_EXP in config.h

perl.c uses an ARCHLIB_EXP define to generate compile-time code that
adds the archlibexp path to @INC during run-time initialization of a
new perl interpreter.

Because we've changed this value in a temporary way to make it
possible to use ExtUtils::Embed in the target build (the temporary
value in config.sh gets re-stripped out during packaging), the
ARCHLIB_EXP value that gets generated still uses the temporary version
instead of the original expected version (i.e. becauses it's in the
generated config.h, it doesn't get stripped out during packaging like
the others in config.sh).

This creates an unmodified version called archlib_exp that gets used
by a modified config_h.SH to get the correct value into config.h

This patch uses an unmodified version of archlibexp called
archlib_exp, introduced to config.sh, which is used to generate the
correct value of ARCHLIB_EXP into config.h

Fixes [YOCTO #3099].

(From OE-Core rev: cbcfdeb1d55e2e76f199750bda401bad126ae234)

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

sstate-cache-management: hide error message when one of possible layer location doesn't exist

* fixes [YOCTO #3116]

(From OE-Core rev: bde88116d9d7e86ca7ecac4cf990689f972b0b1c)

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

bitbake: hob-toolchain: task-core-standalone-sdk-target renamed to packagegroup-core-standalone-sdk-target

This change also applies to task-core-standalone-sdk-target-dbg and resolves
build failures caused by the missing packages.

(Bitbake rev: 4cd0200e96fb282980a945b80af641a6e022e0b4)

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

bzip license: bzip2 not bzip.

The name of the license for bzip2 is wrong causing warnings
to be thrown.

(From OE-Core rev: 566c6101cc7a8d90973eb22478ffc77eac23f81c)

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

gcc-cross-canadian: Fix gettext dependency

There was a problem in commit ad23395cd1f6c0cbfcc6980b6071d589f9f6cde0 since
gettext-nativesdk was translated to gettext instead of nativesdk-gettext.

This fixes to use the correct dependency.

(From OE-Core rev: a6e325342cb489e05927d6cb2bb0a24fa6c20ef8)

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

bitbake: hob/imageconfigurationpage: a 'hob-image' appears listed in the base image combo box

-remove this image from image combo box

[YOCTO #3230]

(Bitbake rev: 90fd57ee3cb2856c10bda1f5af4879d2b7cf2668)

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

bitbake: hob/hobeventhandler: Describe the runCommand failure exception

[YOCTO #1245]

(Bitbake rev: 17f28f09452f70dfb67fce9a397a99deec84dfe5)

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

linux-yocto/3.4: perf: parallel build and tools fixes

korg commit 42dcd1f4e [perf tools: Fix parallel build] fixes parallel
build issues that are being seen in the autobuilder.

We also have a fix from Tom:

[
    perf annotate: replace 'expand' with equivalent sed expression

    We don't have 'expand' in our userspace so we need to accomplish the
    same thing using 'sed', which we do have.
]

So we apply it to all BSP branches and kernel types.

(From OE-Core rev: f06e7d38db35c56c71a42264361ec45fb3777a14)

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

sstate: Also add datadir/sgl to sstate whitelist to avoid openjade warning

(From OE-Core rev: e0ff54db5a5ab171ee1d0dbcf7f267235c21e601)

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

qemu: When applying qemu-mips workaround, check the file exists first

If qemu-mips was disabled as done in some distros, this wrapper would fail.
Therefore check if the file exists before wrapping it.

(From OE-Core rev: 9ec1c06915b10d142bf5646396c4e91bb61a40a5)

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

linux-yocto-tiny: Add tiny recipe for the 3.4 linux-yocto kernel

This recipe builds the "tiny" kernel type defined by the linux-yocto
meta-data. Support is defined for the qemux86 machine via
common-pc-tiny.scc in the linux-yocto meta branch. The resulting
kernel is 1.8 MB and boots to a serial console with with qemux86 and
core-image-minimal using the following command:

qemu -kernel tmp/deploy/images/bzImage-qemux86.bin -initrd tmp/deploy/images/core-image-minimal-qemux86.cpio.gz -append "root=/dev/ram0 console=ttyS0" -nographic

(From OE-Core rev: cf25f211ec420e1e8dd48c8e62f60deefe2c6d53)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nfs-export-root: add explict no_subtree-check to suppress warning

exportfs: /etc/exports [1]: Neither 'subtree_check' or 'no_subtree_check' specified for export "*:/".
  Assuming default behaviour ('no_subtree_check').
  NOTE: this default has changed since nfs-utils version 1.0.x

(From OE-Core rev: 1438228d3b54dfdcf8c36154c927c80fcecf688e)

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

linux-yocto/3.4: tiny: Add qemu KMACHINE to common-pc-tiny.scc

Updating the meta SRCREV to pickup the following change:

  Ensure the qemux86 machine is defined in common-pc-tiny as it is
  for -standard and -rt.

(From OE-Core rev: 1076910ac3cd55a3f87b5ca7a1db1e38c623480a)

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

linux-yocto/3.4: fix kconfig warnings and unnecessary options

Updating the kernel configuration fragments to fix the following
issues:

  - remove options that are no longer in the 3.4 kernel
  - disable unused, but large kernel modules
  - fix kconfig audit warnings for x86 BSPs
  - make uprobes reusable by multiple fragments

The following meta branch commits are represented by this update:

  3da1172 uprobes: split into enable and patch fragments
  17ec51a meta: cleanup invalid/obselete 3.4 CONFIG options
  b5cee42 meta: disable OCFS2 by default
  efe937e meta: drm: tag DRM options as 'hardware'
  10b5155 meta: emenlow: clean emenlow configuration warnings
  a907b82 meta: add CONFIG_SHMEM to standard kernel config

(From OE-Core rev: a01bb3ec72c375c0f06006769969f63fed3ef566)

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

oe-buildenv-internal: Add BB_NO_NETWORK to BB_ENV_EXTRAWHITE

This allows for use of bitbake in offline mode, but override it in
command line.

(From OE-Core rev: bcefd015fb163d9c382ae05a86569dbcfd3d736a)

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

pulseaudio: ensure X11 and consolekit are disabled

When DISTRO_FEATURES does not include X11 ensure that both x11
and gtk are diabled correctly.  ConsoleKit also has x11, so ensure
that any RDEPENDS is also excluded.

The flags for x11 changed at somepoint to use enable/disable, but
this recipe was not updated.

(From OE-Core rev: 0730d3449aa28600488e73de883240ba2bd60aec)

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

package_rpm.bbclass: Fix an issue where complementary installs fail

Also ensure that we always cleanup the temporary install manifest files,
some of them will cause problems if they exist in multiple install
attempts.

Finally verify that the lists remain uniquely sorted otherwise the
complementary install may install the same files numerous times,
triggering a failure.

(From OE-Core rev: 4f2a290cbcc6c21afbb2a6e6148efdef4d135b41)

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

sstate: Add extra entries to the sstate duplicate files whitelist

This avoids errors where gcc/binutils get installed to the native sysroot
in the same location for multiple package architectures. Ultimately making
these native recipes with ${PACKAGE_ARCH} appended to PN will resolve this
but hide the warnings until this gets sorted out.

Also hide the python and docbook catalog warnings since they're known about,
nothing to worry about and we'll aim to clean them up properly in the 1.4 cycle.

(From OE-Core rev: 5bae58a5b59c04d8947f4842f19837a914c29b52)

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

sstate: Normalise paths before comparing with the whitelist

Without this, path components like // could break comparisions with the whitelist leading
to warnings being displayed to the user unintentionally.

(From OE-Core rev: d3c46ca56fab2f07bf16b61514f30765543a8747)

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

libdrm: Explicitly disable the cairo dependency

We don't want the cairo dependency. Unfortunately simply checking whether its present
isn't good enough. If its not in DEPENDS, it can disappear half way through building.
We therefore need to explictly disable it.

(From OE-Core rev: 51df11c5747f69b4112121df78fc1e10644d390a)

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

toolchain-scripts.bbclass: Export M4

some packages use M4 variable from environment and sometimes
its hardcoded to /usr/bin/m4 if not found in environment. Lets
define it such that it is picked from path

(From OE-Core rev: 06c5593d15f206458b9a5b45ed1229abfee16e95)

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

cross-canadian.bbclass: add native chrpath dependency

In order for the RPATHs in 32bit toolchain binaries to be relocated
properly, chrpath >=0.14 is needed.

[YOCTO #3161]
[YOCTO #3201]

(From OE-Core rev: 71c71b972100803d33fbb062a237e8a15167387b)

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

runqemu: allow multiple unfs instances to run simultaneously

A miscalculation in the way the port numbers of mountd and nfsd
are created was causing conflicts when starting multiple instances
of qemu using userspace nfs.

Thanks to Rudolf Streif for proposing this fix!

Fixes [YOCTO #1969]

(From OE-Core rev: 94eef772c283170d19ba92c8de0054cd093fc487)

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

runqemu-export-rootfs: improve rpcbind error detection

mountd requires rpcbind or portmap. Check that one of these
services is running before doing anything else, and report
a user-friendly error when they are not found.

(From OE-Core rev: 16d6ec51f4b976c9b86a8b6bf6251089df2d2732)

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

runqemu-export-rootfs: use consistent whitespace

(From OE-Core rev: b05185240669e0ae811a23620913b35ca99493fb)

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

common-licenses: Adding bzip

bzip requires it's own specific license.

(From OE-Core rev: df2b756436b90f8f9ff229ba64d0af30d9d4f923)

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

bitbake: hob/hobeventhandler: Throw an exception if runCommand fails

- throw a Hob exception if runCommand returns 'Busy' or
'No such command'

[YOCTO #1245]

(Bitbake rev: 5a8e3baa66f845599a616f080a7efce81ecda631)

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

build-appliance-image: Fix spacing in DESCRIPTION

Fixes [YOCTO #2636]

(From OE-Core rev: 831b88806a3cb0125003aa6d3348716bcfd662a1)

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

shutdown-desktop: ensure the postinst script succeeds

When the hostname isn't qemuarm the grep fails so the postinst fails. Stop this
happening by explicitly evaluating true.

[YOCTO #3224]

(From OE-Core rev: 8848ea6793ddaab61c9dad250ec578d68d7d087d)

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

sato-icon-theme: use gtk-icon-cache helper class

Instead of explicitly updating the icon cache use the helper class that also
forces a loader update at the same time.  This eliminates the possibility of
updating the icon cache without any gdk-pixbuf loaders.

Also check that the Sato icon theme isn't already set to avoid appending to the
file every time the postinst runs.

[YOCTO #2399]

(From OE-Core rev: 9d98dbdae4c05fcf50d546f554a04dc3f0bd66c3)

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

distrodata: Update distrocheck functions

Fix the distro check functions for the change of nativesdk
being a suffix to a prefix. Also added crosssdk as another
case for converting to PN for matching in the distro_tracking

(From OE-Core rev: ae9dbd0e1e26ba2b35cbd08ec731aee62adedc23)

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

bitbake: hob/imagedetailspage: disable the deploy tool for qemu machines

- qemu images cannot be deployed to hardware, even if live
images (hddimg and iso) files are created

[YOCTO #3196]

(Bitbake rev: 001b1c439ffa450cb8728b0fa9469fed63ae9bed)

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

bitbake: hob/settings: alignment and spacing fixes on "Shared state" tab

-alignment and separation on vertical and horizontal axis
-change controls width
[YOCTO #3188]

(Bitbake rev: a84c466eae7c2616c041faf930163f23834f0685)

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

bitbake: hob/image_details/run_image: Kernel must be set

- when displaying image details, even if the kernel information is not shown, the kernel value must be set (if dealing with qemu) otherwise running the image will throw an error
(Bitbake rev: 334125a98ecb8a938aee4bc530205ad75099361c)

Signed-off-by: Ioana Grigoropol <ioanax.grigoropol@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: fetch2: raise an exception if user specifies protocol=git with 

It is a common mistake to use http:// and protocol=git when attempting
to fetch from a git repository using the http protocol; if this is
detected then throw an error explaining that you need to use git:// with
protocol=http instead.

(Bitbake rev: 5bc4930c1638db16bcd5f9c8cfc4081f9ffc192b)

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

rpmresolve: fix reporting of multiple matches error

We were mistakenly writing what was meant to go to stderr into the
output file, so when the "Multiple matches" error showed we weren't
actually seeing the matches printed.

Also change the wording of the "Unable to find package..." to "Unable
to resolve package..." instead so that it makes more sense if it is
printed after the "Multiple matches" error.

(From OE-Core rev: c26542eb4e8707b9639ec309a44809a728839db8)

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

buildhistory_analysis: update to use explode_dep_versions2()

Handle where multiple version specifications are present for the same
dependency.

(From OE-Core rev: 1600c916ae410c57a783a5aa35abe07a3f673311)

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

qemu: Fixed running QEMU with virtio error reporting

If vhost_net module is not properly installed,
runqemu script will report the error and
provide the user with a link to the guide.
Also corrected small cosmetic issues in
runqemu script messages.
Also removed <> (read/write) check.

Fixes [YOCTO #3184]

(From OE-Core rev: f7365f62325189b0f9a9a1d440f11f2356c8f01d)

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

wpa-supplicant: don't break the DBus service file

The recipe exports $BINDIR as ${sbindir} and the build system uses this when
writing the DBus service file, so sedding it and replacing $base_sbindir with
$sbindir (/sbin and /usr/sbin) isn't useful when it ends up as
/usr/usr/sbin/wpa_supplicant.

[YOCTO: #3202]

(From OE-Core rev: 41388c3ae0f4d9cd07e1599fbe125123c20820f8)

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

bitbake: providers.py: Fix PREFERRED_VERSION containing epochs

For some reason the code calls int() on the epoch component of any
PREFERRED_VERSION. Since this is compared against strings, the comparison
would always fail. This removes the stray cast and allows epochs
in preferred_version to work correctly.

[YOCTO #3187]

(Bitbake rev: 117b47553970fc5307374cbf500744b7c302efb4)

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

xf86-video-intel: drop libxvmc dependency

xvmc is explicitly disabled, so remove the dependency.

(From OE-Core rev: eb96be4db46039752c44dc37ef676eaac04e3dba)

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

qemu: add wrapper for qemu-mips binary

qemu-mips user emulation binary segfaults when running any kind of
binary. This is due to a MMU access fault in the virtual CPU. This
problem has been introduced in qemu when 4GB of vmem were reserved for
32-on-64 bit.

This workaround will need to be reverted once the proper fix is found.

[YOCTO #3143]

(From OE-Core rev: 53b3103abdf21123b1c7be49b05cfe97a7cd9ed7)

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

e2fsprogs: Don't install findfs

This binary is provided by util-linux nowadays.  Fixes:

WARNING: The recipe is trying to install files into a shared area when those files already exist. Those files are:
     /fast/jenkins/workspace/.../tmp-eglibc/sysroots/x86_64-linux/sbin/findfs

(From OE-Core rev: e71c6bb75239926aceebbb53d158cbf8de6112a4)

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

bzip2 and busybox: Incorrect LICENSE

The license for bzip2 is not quite BSD. I have an email out to the
maintainer to see if we can utilize a common BSD license (or something
else) however, for now, we should revert bzip2 back to a special
license.

As busybox also utilizes a lightly modified bzip2, this also
effects busybox.

(From OE-Core rev: a0b132798d2c1adf79414787b8317327a554f852)

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

linux-yocto-custom: Clarify defconfig usage

It is necessary to supply file://defconfig to the SRC_URI when using
a defconfig (it is not implicitly understood as the commentary might
currently suggest).

(From OE-Core rev: 3e244e0e9c10438c2236e56b3de664d4560791f0)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
CC: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

package.bbclass: Fix regression in -dbg packages introduced by explode_dep_versions change

We need to iterate over the dictionary pulling out the values, not take the top
level keys. If we don't do this, we end up with dependencies on the values of
PACKAGES, not library dependencies.

(From OE-Core rev: 7219bca11f554fbe2ed30f1537491987d65e9316)

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

qt4: Avoid circular dependencies with multilib

Without this, circular dependencies are found when attempting to build
multilib versions of qt4 (or bitbake world in a multilib enabled build).

(From OE-Core rev: b2e8cc5ae227656211fb7f32260e7dc4e2fb556e)

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

bitbake.conf: Add CCACHE_DISABLE to BS_HASHBASE_WHITELIST

If CCACHE is in the whitelist then CCACHE_DISABLE probably should be too.

(From OE-Core rev: c03f76160e3cc3cb4fbf3cee114665c34bff06e6)

Signed-off-by: Mike Crowe <mac@mcrowe.com>
Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: Add missing INITSCRIPT_NAME_${PN}-hwclock

Fix failures like:

Configuring busybox-hwclock.
usage: update-rc.d [-n] [-f] [-r <root>] <basename> remove
       update-rc.d [-n] [-r <root>] [-s] <basename> defaults [NN | sNN kNN]
       update-rc.d [-n] [-r <root>] [-s] <basename> start|stop NN runlvl [runlvl] [...] .
                -n: not really
                -f: force
                -v: verbose
                -r: alternate root path (default is /)
                -s: invoke start methods if appropriate to current runlevel
Collected errors:
 * pkg_run_script: package "busybox-hwclock" postinst script returned status 1.
 * opkg_configure: busybox-hwclock.postinst returned 1.

(From OE-Core rev: 43b4ffc11874803db37c43b521ce27c51c677c8b)

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

bitbake: hob/hobpages: Generate the title label every time

- the title label is destroyed at page switching (that's why we need
to generate it every time)

[YOCTO #3195]

(Bitbake rev: d6d991c08f66cf9ab27c53075109212ea9129380)

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

qt4: add 4.8.3 version with negative D_P

* rebased patches, 3 patches are resolved upstream:
  0019-javascriptcore-Fix-compile-error-when-using-toolchai.patch
    resolved in upstream commit 7ac8d8597db1f58b11338f91fb27f6ad8696b34b
  0022-webkit-fix-conflicting-types.patch
    resolved in upstream commit929b4443d53fcf3a7ad1cb9f3af5569e41ef56f1

(From OE-Core rev: b9fc4928bb93ad720c47920db3869d860c531d0a)

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

qt4: PR bumps

* sofar only formal changes, but to test that everything still builds the same

(From OE-Core rev: 054a0e6c850f92c03fbb6314702de4e6318ccd25)

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

qt4: replace all local patches with git patches with headers

* in preparation for upgrade to 4.8.3

(From OE-Core rev: 24bf06bbcda4c0dfdad33cdff6394faa52657bb9)

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

qt4: drop patches not used in any recipe

(From OE-Core rev: 5f0684c1d23a3520095a4d450a0c1fa95fa1c7b2)

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

qt4: move patches from files to qt4-4.8.1

* faster lookup in FILESPATH as small bonus

(From OE-Core rev: 8cc54d2d154b2ed9f931da39d75dc9c135f5e26d)

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

qt-mobility: move qt-mobility patches to separate dir

(From OE-Core rev: 8dccc55a623f0c5f3469c7cdf63aa788683aa186)

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

qt4.inc: move more options to separate variables

* make it easier to override them in bbappend
* convert pulseaudio to more common -pulseaudio/-no-pulseaudio form

(From OE-Core rev: 34e3687394c6fa18ef0443d63b8d7d0a68c441e0)

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

qt4: rename qt-4.8.1 to qt4-4.8.1 to match other .inc and .bb

(From OE-Core rev: dcda03d3f6ec442740e3683a1971103dc639689d)

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

qt4: use releases.qt-project.org instead of get.qt.nokia.com

* releases.qt-project.org has 4.8.1 as well as 4.8.3

(From OE-Core rev: f12df439b893c70a8cd271ff8b8e6d760b78a2b3)

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

qt4-tools-nativesdk.inc: rename to nativesdk-qt4-tools.inc

(From OE-Core rev: 11970283424f213f870ba7e96d70bd507b10bc63)

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

libdrm: Remove Cairo dependency

This causes a build loop, when DRM depends on Cairo depends on Mesa
depends on DRM.  We can safely remove it as it's only one libdrm example
program which uses Cairo, which we won't be needing.  At least it's not
worth the build loop.

(From OE-Core rev: a6d305261dc925210185d8b70fb1a923e012153b)

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

binutils.inc: add vardep on multiarch DISTRO_FEATURE

binutils will build differently if this feature is enabled, so
make the do_configure step depend on it

(From OE-Core rev: 0788cf349fe37ef4a36c626dbc396c97d1ab14d7)

Signed-off-by: Matthew McClintock <msm@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libx11.inc: fix build issues for older CentOS distros

Fixes these sorts of issues present on older gcc (CentOS 5.x in this case)

| cc1: error: unrecognized command line option "-Werror=implicit"
| cc1: error: unrecognized command line option "-Werror=nonnull"
| cc1: error: unrecognized command line option "-Werror=init-self"
| cc1: error: unrecognized command line option "-Werror=main"
| cc1: error: unrecognized command line option "-Werror=missing-braces"
| cc1: error: unrecognized command line option "-Werror=sequence-point"
| cc1: error: unrecognized command line option "-Werror=return-type"
| cc1: error: unrecognized command line option "-Werror=trigraphs"
| cc1: error: unrecognized command line option "-Werror=array-bounds"
| cc1: error: unrecognized command line option "-Werror=write-strings"
| cc1: error: unrecognized command line option "-Werror=address"
| cc1: error: unrecognized command line option "-Werror=int-to-pointer-cast"
| cc1: error: unrecognized command line option "-Werror=pointer-to-int-cast"

Also fixes:

makekeys-makekeys.o: In function `main':
makekeys.c:(.text+0x85): undefined reference to `__isoc99_sscanf'
makekeys.c:(.text+0xa7): undefined reference to `__isoc99_sscanf'
collect2: ld returned 1 exit status
make: *** [makekeys] Error 1

Older libc do not have this defined, we can use the -D_GNU_SOURCE
to the compiler to prevent generating calls to this function and
make linking work

(From OE-Core rev: 83c560ae282c1a28fd2c311c66debd02a69f1678)

Signed-off-by: Matthew McClintock <msm@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Revert "initrd: Spawn an emergency shell when something goes wrong"

This had nowhere near enough testing...

This reverts commit ffb6928f5783e5202d9849c3a185e29be1d41c63.

(From OE-Core rev: f162f0ecc96fdfb564aad968e5b8bc670640ea68)

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

nativesdk.bbclass: Ensure we have chrpath >=0.14

Versions earlier than 0.14 can't cope with 32 bit binaries on a 64 bit
system and vice versa. This results in problems for certain SDKMACHINE
combinations on certain hosts. By ensuring we build
chrpath-replacement-native we avoid this problems and the binaries work
correctly.

[YOCTO #3161]
[YOCTO #3201]

(From OE-Core rev: f89bced26de055817100d0b0e03094b031fcfd48)

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

chrpath: We should provide chrpath-replacement-native and install into a native specific directory

chrpath is assumed to be provided by the build host system. This means
we need to provide a replacement version and install into a specific directory
to avoid races.

(From OE-Core rev: 147c44c882a3dc667c079165d3ddc9d78b702286)

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

gzip: The native version should provide gzip-replacement-native

(From OE-Core rev: 9c2e5b04e40102fc7276fd3a6467725617dc33ce)

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

bitbake.conf: Add chrpath-native to ASSUME_PROVIDED

We assume chrpath is provided natively so it should be listed in ASSUME_PROVIDED.

(From OE-Core rev: 97a3ea712003e8d48dc68c282e656591f39d2d1a)

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

scripts/oe-buildenv-internal: Ensure we detect the SDK/ADT and error out

The SDK/ADT may ship with a python installed which may not have all the modules
need for a bitbake build. We should therefore detect if its already present in the
environment and error out in this case, asking the user to use a clean environment.

This also removes the potential for any other conflict between the two.

[YOCTO #2979]

(From OE-Core rev: 9496d4cd77ae632251b4262b63be857fc4fcb31e)

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

opkg: Convert select-higher-version option to prefer-arch-to-version

This converts the option to maintain the existing behaviour unless the option is
specified. We do specify the option during the builds themselves to ensure what
the users expects is built.

(From OE-Core rev: 0cc479699fe885049625d54c712b500c1b719e75)

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

bitbake: hob: Improved behavior for error reporting window

Scrollbars have now an automatic behavior, depending on
the error's text size and error window size.

Fixes [YOCTO #2983]

(Bitbake rev: 0c0a25672498520fb2c46164f08959dda83c61e0)

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

libart-lgpl: add art_config.h for mipsel

Fixes:
WARNING: Unable to get checksum for libart-lgpl SRC_URI entry art_config.h: file could not be found

which otherwise happens during parsing, even if libart-lgpl isn't being built.

(From OE-Core rev: a9245e0d1dd1bee4ac01fe0c73d95179033ba979)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

opkg: Drop nogpg version since the main version now has nogpg too

(From OE-Core rev: c8197c993c5848576a9c63660ff342571d31a4ba)

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

opkg-native: remove spurious dependency on curl-native

All variants of opkg are currently configured --disable-curl so there
seems no point in depending on it.

(From OE-Core rev: 875f1eb876c17c038a77bc7b7a5fed775d9fd3ea)

Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libpcap: add dependency on libnl

libpcap uses libnl on Linux to support sniffing mac80211 devices, which could be
useful.

(From OE-Core rev: 052a8406e66c9dcccc1fc506a32cc1706b93467b)

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

galago: remove

Galago has been replaced with Telepathy and Folks, and has been unmaintained for
years.

(From OE-Core rev: b878a9b1af63603779850f420fc0feb2ef78da9f)

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

sanity.bbclass: Fix invalid test for network error

The test for network error in sanity.bbclass was negated.

(From OE-Core rev: 9fcd0866f0e30a50182434f6bcae13bf9575807f)

Signed-off-by: Bogdan Marinescu <bogdan.a.marinescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: Update version to 1.16.0

(Bitbake rev: a579754a04bdcf450e6957dde614a15c11df39e2)

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

bitbake: hob: Further improvements to hob dialogs

1. Replace some labels in the "Build environment" tab
2. 'defaultsetup' changed to 'Default' in the "Image types" tab
3. Fixed the moving icon in the "Output" tab

For more details: https://bugzilla.yoctoproject.org/show_bug.cgi?id=2162

[Yocto #2162]

(Bitbake rev: db7d98569117b7a75262eb555e1c7ae9a421bdf8)

Signed-off-by: Bogdan Marinescu <bogdan.a.marinescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hob: Fix 'User selected' tag functionality

- the 'User selected' tag is only used when the user
selects a package
- fix hob to remember which packages are 'User selected'
- if the package is already brought in by some other package,
it should not appear as 'User selected'

[YOCTO #3108]

(Bitbake rev: 2391e9ba7034d4f90bafa5732d8efa8166f69950)

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

bitbake: hob/settings: Change the 'Delete' button behaviour in the shared state settings tab

-the tab shows an empty mirror row when no mirror is configured
-able to delete the mirror row even if it's not empty(if it's not
the first mirror)

[YOCTO #3189]
(Bitbake rev: d6472608112b8af2e98f247e6f89a7f948b2d020)

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

bitbake: hob/image_details: Remove kernel info from details

- removed kernel information from image details

[Yocto #3002]

(Bitbake rev: 7fc33f0a8a38d9b8984bf884e47e505791536d16)

Signed-off-by: Ioana Grigoropol <ioanax.grigoropol@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: tests/utils: Add test for explode_dep_versions2

(Bitbake rev: b1b0aabfab3c94c3b515070d0fb4d7819e2548bc)

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

bitbake: utils: Add explode_dep_versions2 to replace explode_dep_versions

The API for explode_dep_versions is flawed since there can only be one version
constraint against any given dependency. This adds a new function with an API
without this limitation. explode_dep_versions() is maintained with a warning
printed when its used in a situation where information is lost.

This should allow a simple transition to the new API to fix the lost dependency
information.

join_deps() is updated to deal with data in either format.

(Bitbake rev: babeeded21827d8d3e7c7b785a62332ee9d45d4f)

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

bitbake: utils.py: Allow explode_dep_versions comparisons to have arbitrary whitespace

Refactor the explode_dep_versions to be more lenient on whitespace values.

The required format is:
   foo (= 1.10)
   foo (=1.10)
   foo ( = 1.10)
   foo ( =1.10)
   foo ( = 1.10 )
   foo ( =1.10 )

(Bitbake rev: 39c1c12c58fadd854098cf14ebe92f4d307a36dd)

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

bitbake: utils.py: Check for duplicate dependency entries

explode_dep_versions is not able to have duplicate entries.  Previously
duplicate entries ended up with the last item being the one returned to
the caller.

We now detect a collision.  We do allow an empty item to have a comparison
added to it, or a duplicate with the same comparison without error.

When a collision is detected a ValueError exception is thrown.

Allowed:
   foo foo (= 1.12) foo

Invalid:
   foo (= 1.12) foo (= 1.13)

(Bitbake rev: d40448f0483a2959e9dcaac9b6dd35839f396a6e)

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

classes: Update to use corrected bb.utils.explode_dep_versions2 API

The bb.utils.explode_dep_versions function has issues where dependency information
can be lost. The API doesn't support maintaining the correct information so this
changes to use a new function which correctly handles the data.

This patch also fixes various points in the code to ensure that we do not have any
duplicates in things that use explode_dep_versions.

A new sanity test to test the contents of the R* variables is also added.

[Some changes from Mark Hatle <mark.hatle@windriver.com>]

(From OE-Core rev: 16a892431d0c0d03f8b561b92909cf2f11af4918)

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

insane.bbclass: Remove copy and paste confusion when using OVERRIDES

People keep copying this code and its confusing and unnecessary. Remove the
bad examples to try and stop this happening.

(From OE-Core rev: 48aa4b00cfb7f01195c6d20b7ba660715fe792ef)

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

sanity.conf: Require bitbake 1.16.0 (stable series) prior to release (and for explode_dep_versions changes)

(From OE-Core rev: bf334f01cb90483ba304b4a830d53e705637b87a)

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

gdk-pixbuf: Ensure gdk-pixbuf-native dependencies are correct with linuxstdbase

Without this change, anything using linuxstdbase would incorrectly
try and pull in X dependencies.

(From OE-Core rev: 664fce61c9c49276039b18606e1e88585b47f724)

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