code.ossystems Code Review - openembedded-core.git/log

libsolv: update to version 0.6.33

0.6.32 -> 0.6.33

* new Selection.clone() method in the bindings
* new pool.parserpmrichdep() method in the bindings
* fix bad assignment in solution refinement that led to a memory leak
* use license tag instead of doc in the spec file [bnc#1082318]

Signed-off-by: Maxin B. John <maxin.john@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libunistring: update version to 0.9.9

License-Update: checksum change is due to bump in copyright year
to 2018.

Signed-off-by: Maxin B. John <maxin.john@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

flex: create separate package for libfl

Target binaries linked with libfl currently generate a runtime
dependency on the entire flex package (and therefore m4 and bison
too). Copy Debian's approach and create a separate package for libfl.

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

gdb: Add signed-off-by tag to patch

A patch went in (in 4aaf747) without a proper signed-off-by
because the project (in its upstream repository) does not use
Git.

This will take care of that before spreading the patch to
other branches.

Signed-off-by: Daniel Díaz <daniel.diaz@linaro.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>

gtk-doc: inherit classes only if gtk-doc is enabled

Respect GTKDOC_ENABLED when inheriting python3native and DEPENDing on
qemu-native, as they're not needed when disabled.

python3native is required as otherwise the host Python is most likely used which
may or may not have python3-six installed (a requirement of gtk-doc).

Signed-off-by: Ross Burton <ross.burton@intel.com>

libfm: fix dependencies

libfm uses glib-gettextize so explicitly depend on glib-2.0-native.

Instead of depending on gettext-native, inherit gettext.

Signed-off-by: Ross Burton <ross.burton@intel.com>

valgrind: Mask CPUID support in HWCAP on aarch64

valgrind currently does not know anything about the CPUID flag added to
the HWCAP auxv entry in kernel 4.11+

At runtime it will fails like this:

ARM64 front end: branch_etc
disInstr(arm64): unhandled instruction 0xD5380001
disInstr(arm64): 1101'0101 0011'1000 0000'0000 0000'0001 ==2082==
valgrind: Unrecognised instruction at address 0x4014e64.

This patch is a workaround by masking all HWCAP. This patch is dervied
from https://bugzilla.redhat.com/show_bug.cgi?id=1464211

Signed-off-by: Manjukumar Matha <manjukumar.harthikote-matha@xilinx.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

package_manager.py: Print offending package instead of non-sense trace

If you have a package that does not generate a manifest due to using a
noexec rule, the package name should be printed so the problem can be
tracked down.  With out the patch you get an error that makes it look
more like the package_manager is broken as shown below.

oe-core/meta/lib/oe/package_manager.py', lineno: 534, function: create_packages_dir
     0530:
     0531:    for dep in rpmdeps:
     0532:        c = taskdepdata[dep][0]
     0533:        manifest, d2 = oe.sstatesig.find_sstate_manifest(c, taskdepdata[dep][2], taskname, d, multilibs)
*** 0534:        if not os.path.exists(manifest):
     0535:            continue
     0536:        with open(manifest, "r") as f:
     0537:            for l in f:
     0538:                l = l.strip()
File: '/usr/lib/python3.5/genericpath.py', lineno: 19, function: exists
     0015:# This is false for dangling symbolic links on systems that support them.
     0016:def exists(path):
     0017:    """Test whether a path exists.  Returns False for broken symbolic links"""
     0018:    try:
*** 0019:        os.stat(path)
     0020:    except OSError:
     0021:        return False
     0022:    return True
     0023:
Exception: TypeError: stat: can't specify None for path argument

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

usbutils: drop upstreamed patch

This has been fixed upstream since 008, albeit slightly differently so the patch
continued to apply.

Signed-off-by: Ross Burton <ross.burton@intel.com>

libpcre: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

vulkan: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

xproto: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

libxcb: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

libaio: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

lsb: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

screen: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

sysstat: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

unzip: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

watchdog: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

sysklogd: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

btrfs-tools: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

elfutils: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

ccache: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

flex: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

mtools: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

squashfs-tools: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

iproute2: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

neard: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

nfs-utils: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

dropbear: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

kbd: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

libxml: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

ovmf: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

gtk+: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

gobject-introspection: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

dbus-glib: refresh patches

The patch tool will apply patches by default with "fuzz", which is where if the
hunk context isn't present but what is there is close enough, it will force the
patch in.

Whilst this is useful when there's just whitespace changes, when applied to
source it is possible for a patch applied with fuzz to produce broken code which
still compiles (see #10450). This is obviously bad.

We'd like to eventually have do_patch() rejecting any fuzz on these grounds. For
that to be realistic the existing patches with fuzz need to be rebased and
reviewed.

Signed-off-by: Ross Burton <ross.burton@intel.com>

dbus: remove upstreamed patch

Signed-off-by: Ross Burton <ross.burton@intel.com>

util-linux: add taskset to alternatives list

The taskset command is provided by both busybox and util-linux.

Signed-off-by: Lars Persson <larper@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

gdb: fix header ordering for TRAP_HWBKPT

This error can appear in gdb/nat/linux-ptrace.c because of
the order in which some headers are processed:
| In file included from ../../gdb-7.11.1/gdb/nat/linux-ptrace.c:20:0:
| ../../gdb-7.11.1/gdb/nat/linux-ptrace.h:175:22: error: expected identifier before numeric constant
| # define TRAP_HWBKPT 4
| ^
| Makefile:2357: recipe for target 'linux-ptrace.o' failed
| make[2]: *** [linux-ptrace.o] Error 1
| make[2]: *** Waiting for unfinished jobs....
| make[2]: Leaving directory '/oe/build/tmp-rpb-glibc/work/aarch64-linaro-linux/gdb/7.11.1-r0/build-aarch64-linaro-linux/gdb'
| Makefile:8822: recipe for target 'all-gdb' failed
| make[1]: *** [all-gdb] Error 2
| make[1]: Leaving directory '/oe/build/tmp-rpb-glibc/work/aarch64-linaro-linux/gdb/7.11.1-r0/build-aarch64-linaro-linux'
| Makefile:846: recipe for target 'all' failed
| make: *** [all] Error 2

A patch from GDB's current master solves the issue.

Signed-off-by: Daniel Díaz <daniel.diaz@linaro.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>

glibc: add missing TRAP_BRANCH/TRAP_HWBKPT definitions

Patch submitted upstream, pending to be merged:
https://sourceware.org/bugzilla/show_bug.cgi?id=21286

Signed-off-by: Fathi Boudra <fathi.boudra@linaro.org>
Signed-off-by: Daniel Díaz <daniel.diaz@linaro.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>

e2fsprogs_1.43.8.bb: improve reproducibility

Various builds of e2fsprogs 1.43.7 package locales which may or may
not have POT-Creation-Date removed. There is no obvious pattern, it
affects different locales each time, the build being non-deterministic.

The root cause was tracked to non-deterministic time stamps (as GIT does
not preserve file mktime), so some "make" rules sometimes fired, sometimes
did not.

The remedy is to explicitly "touch" files that cause non-deterministic build.

[YOCTO #12516]

Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

kmscube: Add patch for missing GBM function

Some implementations of GBM, like the one included with
libMali, do not have gbm_bo_map() nor gbm_bo_unmap().
This patch enables kmscube to work with those implementations
even if it doesn't work as great.

Signed-off-by: Daniel Díaz <daniel.diaz@linaro.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>

meson.bbclass: fix to build for more projects

We should use the value of CC for the c compiler setting in cross
compilation configuration file for meson. For example, if we only
use ${HOST_PREFIX}gcc instead of ${CC}, we would meet the following
do_compile failure for systemd.

cc1: fatal error: linux/capability.h: No such file or directory

Do the same change for LD, AR, NM, STRIP and READELF.

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libconvert-asn1-perl: inherit ptest-perl

* Enable ptest using new ptest-perl.bbclass

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

dtc: add rdepends_${PN}-misc = diffutils

Needed by dtdiff which calls `diff` to display its result.

Signed-off-by: Ioan-Adrian Ratiu <adi@adirat.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

diffutils: allow native & nativesdk builds

Required by the new dtc rdepends to avoid errors like this:

ERROR: Required build target 'ionel-rpi-image' has no buildable providers.
Missing or unbuildable dependency chain was: ['ionel-rpi-image', 'nativesdk-packagegroup-sdk-host', 'nativesdk-qemu', 'nativesdk-dtc', 'nativesdk-diffutils']

Signed-off-by: Ioan-Adrian Ratiu <adi@adirat.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

godep.bbclass: Add helper class to enable go-dep tool

Many go packages can take advantage of dep tool since
they manage their own dependencies, this class helps
in using go dep tool for such packages

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

tcmode-default.inc: add settings for go

to allow GOVERSION to be set for using an older
go toolchain.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go-runtime: remove unneeded nativesdk override, rename variable

since GO_LDFLAGS is also used by the dist tool, and it's confusing
to use a variable with the same name (but not exported, so unused
by make.bash/dist).

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

packagegroup-go-sdk-target: add go-runtime-staticdev

to allow go programs to be linked either statically or
dynamically when cross-compiling with the SDK.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go: disable PIE CFLAGS for nativesdk and cross-canadian builds

The statically-linked Go code in the toolchain is not compatible
with PIE, so disable its use in the C compiler during the
toolchain build.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go.bbclass, goarch.bbclass: update SECURITY_CFLAGS

With go1.10 the NOPIE flags are only required for
MIPS target builds, and are now incompatible for
the other architectures.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go: move common settings to go-common.inc

Eliminate some redundancy in the recipes by moving
some commonly-used variable settings to the common
include file. Also removed a duplicate inherit
from go-target.inc that was already in go-common.inc.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

goarch.bbclass: disable shared runtime for nativesdk builds

While useful on embedded devices for saving disk space, use
of shared runtime in Go is not the usual practice, so disable
it for nativesdk builds. We don't use it for native builds,
either, so this makes the SDK match the native environment
more closely.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go.bbclass: ptest cleanup and improvements

* Don't enable verbose test output (-test.v)
  by default, as it generates too much noise
  for automated results parsing

* Override do_install_ptest_base in the bbclass,
  so recipes can provide their own modifications
  with do_install_ptest.

* Improve the generated run-ptest script to better
  handle large numbers of tests, and to generate
  'status: test name' output similar to Automake
  tests.

* Install all non-vendored 'testdata' directories
  from the source into the ptest package, as some
  packages share test data among multiple tests.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go.bbclass: don't stage test data with sources

Any directory in a Go package's source tree called
'testdata' contains test data, and isn't necessary
for building.

Some packages include ELF files and other binaries
as test data, and staging them in the sysroot and
-dev package leads to unnecessary QA warnings.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go.bbclass: remove debug-related commands

The 'go env' in the do_compile function and
the set -x/+x in the do_install_ptest function
were used for debugging the bbclass, and aren't
really needed.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go.bbclass: rename GO_TMPDIR -> GOTMPDIR

and export it. Go 1.10 now supports using this
separate variable locating its temporary files.

TMPDIR is still set, for compatibility with go1.9;
that can be dropped once 1.9 is retired.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go: set GOMIPS envrionment variable

Go 1.10 adds support for selecting hard/soft float
object code through the GOMIPS environment variable.

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

go: update go 1.9 -> go 1.10

* Patches and recipes reworked for go 1.10's significant
  changes to its bootstrap and build steps.

* Update go1.4 source tarball used for go-native
  bootstrapping to the version recommended
  in the current go documentation

* Remove test data from installed sources to eliminate
  some packaging QA warnings

* Set GOCACHE to 'off' to disable 1.10's build caching
  in the go recipes and bbclass

* Update go_do_compile to compile both static and
  dynamic objects dynamic linking is in use, since
  go1.10's build tool is pickier about this

Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Ross Burton <ross.burton@intel.com>

expect: upgrade 5.45.3 -> 5.45.4

Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

maintainers.inc: add myself as -bootconf recipe maintainer

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd-boot-cfg.bbclass: Don't reference or set OVERRIDES

There's no need to add to the local copy of overrides and then not do
anything with it.

Now that this function is being used in package creation it was causing
sstate issues as well, as MACHINE is always in OVERRIDES, so something
trivial such as the name of the MACHINE would cause the hash to change.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub-efi-cfg.bbclass: Don't reference or set OVERRIDES

There's no need to add to the local copy of overrides and then not do
anything with it.

Now that this function is being used in package creation it was causing
sstate issues as well, as MACHINE is always in OVERRIDES, so something
trivial such as the name of the MACHINE would cause the hash to change.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

selftests: Add test case for booting a generic EFI boot partition image

Simple test case that adds 'efi' to MACHINE_FEATURES, sets WKS_FILE to
"efi-bootdisk.wks.in", installed required boot items, and attempts to
boot the wic image.

Quick check to make sure that the feature actually works.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

layer.conf: add -bootconf recipes to SIGGEN_EXCLUDE_SAFE_RECIPE_DEPS

The bootloaders depend on these to be functional in the non-deploy case,
but changes in them don't require rebuilding of the packages.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

init-install.sh: support multiple kernels and don't assume vmlinuz

Since kernels will not necessarily be installed as vmlinuz anymore,
don't assume that's its name for either the bootloader config or the
copy of the kernel.

Also, allow installing multiple kernels by searching for common kernel
names.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

packagegroup-core-boot: add kernel for EFI systems

Previously this wasn't needed because the kernel was added at image
creation time to the boot partition. Now that the boot partition is
created from the /boot/ partition of the rootfs, it needs to be
installed there.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd-boot: add package that installs to boot

If the EFI_PROVIDER is systemd-boot, install as boot(x64|ia32) as per
convention. If its not the EFI_PROVIDER, install as
systemd-boot(x64|ia32), as to not collide with other possible
bootloaders.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd-boot*.bbclass: Don't use vmlinuz

We can't guarantee vmlinuz anymore. Use KERNEL_IMAGETYPE instead.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd: add systemd-bootconf recipe

Reuses our systemd-boot-cfg bbclass to generate systemd-boot
configuration files.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd-boot.bbclass: break out configuration creation

This class is useful on its own and can be used to create configuration
recipes.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

syslinux.bbclass: don't use vmlinuz

We can't guarantee the kernel will be named vmlinuz anymore. Use
KERNEL_IMAGETYPE instead.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

init-install-efi.sh: Update to support installing multiple kernels

We can no longer rely on the kernel having a static name of "vmlinuz".
This means we can't use it as a sentinel value in our sed commands, and
we can't just copy vmlinuz to the boot directory.

Instead, we'll use "root=" as the sentinel value for our sed commands,
and we'll search for common kernel names to copy into our boot
directory.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic: add wks file to make use of exclude-path functionality

Using --exclude-path and a wks.in file we can create an image that takes
the /boot/ directory for the boot partition, has an empty /boot/
directory in the rootfs partition. The boot partition gets mounted to
/boot/ after startup.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub-efi*.bbclass: don't reference vmlinuz

Rather than renaming the kernel to vmlinuz and assuming the name is
vmlinuz in the grub.cfg, copy to ${KERNEL_IMAGETYPE} and also use that
value in the grub.cfg file.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

live-vm-common.bbclass: Don't use vmlinuz or VM_DEFAULT_KERNEL

I can't find VM_DEFAULT_KERNEL used anywhere else, and we should not be
statically installing the kernel as vmlinuz.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

packagegroup-core-boot: add bootloader to EFI systems

Since the bootloader is now installed instead of added at image creation
time, this is necessary for booting.

Also set the default to grub-efi. This is done because a default of
grub-efi is already used by live-vm-common.bbclass, and in the event
that EFI_PROVIDER isn't defined errors would occur.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub-efi: if installed and EFI_PROVIDER, install as bootx64 or bootia32

This way we could theoretically support multiple bootloaders, and we
keep the convention of boot(x64|ia32).

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub-efi: install to /boot/

Since /boot/ will be recipe/package controlled now we can't just deploy.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub: create recipe for configuration

This makes use of the grub-efi-cfg bbclass that was split out to create
a grub.cfg file just like the old one.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub-efi.bbclass: split out configuration portion

This part is useful on its own, whereas the whole class together is
specific for image-live.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glibc: Compile using thumb2 for arch > armv5 if user chooses

For arm we enforce ARM mode regardless of ARM_INSTRUCTION_SET
choice from config metadata, glibc works fine with thumb2 for
armv7+ so limit the restriction to armv5 and lower, tested on
rpi3 works equally well as arm mode glibc and sheds about 0.5MB
in size for main package alone. Other glibc build packages also
gets smaller

ARM: 2696 KiB libc6
Thumb2: 2132 KiB libc6

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

maintainers.inc: self-assign recipes from recipes-extended/perl

In prior round, the perl module recipes in recipes-extended/perl
were inadvertantly overlooked.

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

maintainers: self-assign perl module maintainership

Also, take over the other recipes that were just enabled for
ptest:
  - liberror-perl
  - liburi-perl
  - libxml-parser-perl
  - libxml-perl
  - libxml-simple-perl

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libxml-sax-perl: inherit ptest-perl

* Enable ptest by inheriting new ptest-perl.bbclass
* Install testfiles/ into PTEST_PATH

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libxml-sax-base-perl: inherit ptest-perl

* Enable ptest by inheriting new ptest-perl.bbclass

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libtimedate-perl: inherit ptest-perl

* Enable pteset with new ptest-perl.bbclass

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

libsdl2: Upgrade 2.0.7 -> 2.0.8

License checksum changed due to copyright year changes see commit

https://github.com/SDL-mirror/SDL/commit/a9072159b2afff5a338804781312067f0a174c3c#diff-21c55fa400e4d25aed3a755371e32151

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

libxml-simple-perl: inherit ptest-perl

* Enable ptest by inheriting new ptest-perl.bbclass

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libxml-perl: inherit ptest-perl

* Enable ptest by inheriting new ptest-perl.bbclass

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libxml-parser-perl: inherit ptest-perl

* Enable ptest by inheriting new ptest-perl.bbclass

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

liburi-perl: inherit ptest-perl

* Enable ptest by inheriting new ptest-perl.bbclass
* Remove t/cwd.t and t/file.t which require "-T"
(taint) command line option as they will fail.

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

liberror-perl: inherit ptest-perl

* Enable ptest by inheriting new ptest-perl.bbclass

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libtest-needs-perl: add 0.002005

Skip tests when modules not available

* Dependency for ptest of liburi-perl

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta/classes: add ptest-perl.bbclass

* Enable easier testing of perl modules
- Installs t/* to PTEST_PATH
- Uses common run-ptest script

Signed-off-by: Tim Orling <timothy.t.orling@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

populate_sdk_base: depend on nativesdk-glibc-locale

If we're building a SDK and we're using glibc so may be installing locales,
add a build-dependency on natiesdk-glibc-locale so the locales we need will
exist.

Signed-off-by: Ross Burton <ross.burton@intel.com>

sdk: only install locales if we're using glibc

Using glibc-locale to install locales only makes sense if we're using glibc.

Signed-off-by: Ross Burton <ross.burton@intel.com>

testexport-tarball: sync with buildtools-tarball

This is basically a copy/paste of buildtools-tarball so copy some of the
recent changes to buildtools across.

Signed-off-by: Ross Burton <ross.burton@intel.com>