]> code.ossystems Code Review - openembedded-core.git/log
openembedded-core.git
9 years agorpm: Sync CVS to regular version
Mark Hatle [Wed, 9 Mar 2016 19:58:42 +0000 (13:58 -0600)]
rpm: Sync CVS to regular version

A previous fix for a python dep issue was not merged to the CVS version.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorpm: Fix musl integration with RPM5
Mark Hatle [Wed, 9 Mar 2016 19:58:41 +0000 (13:58 -0600)]
rpm: Fix musl integration with RPM5

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agogcc: Disable libitm for nios2
Marek Vasut [Wed, 9 Mar 2016 11:30:48 +0000 (12:30 +0100)]
gcc: Disable libitm for nios2

The libitm is not supported on nios2, so disable it.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Ley Foon Tan <lftan@altera.com>
Cc: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Ross Burton <ross.burton@intel.com>
Cc: Thomas Chou <thomas@wytron.com.tw>
Cc: Walter Goossens <waltergoossens@home.nl>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agopopulate_sdk_ext: Correct commit 8b81bb56c69aabdea984352f8e267a9783c0bdbc
Richard Purdie [Wed, 9 Mar 2016 22:31:15 +0000 (22:31 +0000)]
populate_sdk_ext: Correct commit 8b81bb56c69aabdea984352f8e267a9783c0bdbc

Commit 8b81bb56c69aabdea984352f8e267a9783c0bdbc was accidentally merged.
The DL_DIR piece was simply incorrect and should be removed.

The patch commit message should have mentioned that the changes were
to update populate_sdk_ext after the changes to uninative now the
download is placed into a specific directory in DL_DIR. We also
need to specify the uninative tarball checksum.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorecipetool: create: shrinkwrap and lockdown npm modules
Paul Eggleton [Wed, 9 Mar 2016 04:48:55 +0000 (17:48 +1300)]
recipetool: create: shrinkwrap and lockdown npm modules

"npm shrinkwrap" creates a file that ensures that the exact same
versions get fetched the next time the recipe is built. lockdown is
similar but also includes sha1sums of the modules thus validating they
haven't changed between builds. These ensure that the build is
reproducible.

Fixes [YOCTO #9225].

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorecipetool: create: support creation of additional files by plugins
Paul Eggleton [Wed, 9 Mar 2016 04:48:54 +0000 (17:48 +1300)]
recipetool: create: support creation of additional files by plugins

Allow plugins to create additional files to go alongside the recipe. The
plugins don't know what the output filename is going to be, so they need
to put the files in a temporary location and add them to an "extrafiles"
dict within extravalues where the destination filename is the key and
the temporary path is the value.

devtool add was also extended to ensure these files get moved in and
preserved upon reset if they've been edited by the user.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorecipetool: create: check if npm available if npm:// URL specified
Paul Eggleton [Wed, 9 Mar 2016 04:48:53 +0000 (17:48 +1300)]
recipetool: create: check if npm available if npm:// URL specified

If the user specifies an npm:// URL then the fetcher needs npm to be
available to run, so check if it's available early rather than failing
later.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorecipetool: create: split npm module dependencies into packages
Paul Eggleton [Wed, 9 Mar 2016 04:48:52 +0000 (17:48 +1300)]
recipetool: create: split npm module dependencies into packages

Rather than rolling all of an npm module's dependencies into the same
package, split them into one module per package, setting the SUMMARY and
PKGV values from the package.json file for each package. Additionally,
mark each package with the appropriate license using the license
scanning we already do, falling back to the license stated in the
package.json file for the module if unknown. All of this is mostly in
aid of ensuring all modules and their licenses now show up in the
manifests for the image.

Additionally we set the main LICENSE value more concretely once we've
calculated the per-package licenses, since we have more information at
that point.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorecipetool: create: add license file crunching
Paul Eggleton [Wed, 9 Mar 2016 04:48:51 +0000 (17:48 +1300)]
recipetool: create: add license file crunching

Matching license texts directly to md5sums only goes so far. Some
licenses make the copyright statement an intrinsic part of the license
statement (e.g. MIT) which of course varies between projects. Also,
people often seem to take standard license texts such as GPLv2 and
reformat them cosmetically - re-wrapping lines at a different width or
changing quoting styles are seemingly popular examples. In order to
match license files to their actual licenses more effectively, "crunch"
out these elements before comparing to an md5sum. (The existing plain
md5sum matching has been left in since it's a shortcut, and our list of
crunched md5sums isn't a complete replacement for it.)

As always, this code isn't providing any guarantees (legal or otherwise)
that it will always get the license correct - as indicated by the
accompanying comments the LICENSE values it writes out to the recipe are
indicative and you should verify them yourself by looking at the
documentation supplied from upstream for the software being built if you
have any concerns.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorecipetool: create: match *LICENSE* as a license file
Paul Eggleton [Wed, 9 Mar 2016 04:48:50 +0000 (17:48 +1300)]
recipetool: create: match *LICENSE* as a license file

For example, this picks up a file named MIT-LICENSE.txt.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorecipetool: create: improve mapping for autotools program macros
Paul Eggleton [Wed, 9 Mar 2016 04:48:49 +0000 (17:48 +1300)]
recipetool: create: improve mapping for autotools program macros

Make the following improvements to mapping items specified in
AC_CHECK_PROG, AC_PATH_PROG and AX_WITH_PROG to recipes/classes:

* Produce a map of native recipe -> binary for all binaries currently in
  STAGING_BINDIR_NATIVE and use this when mapping items
* Add some more entries to the class map
* Ignore autotools binaries since they are covered by the inherit of
  autotools
* Ignore coreutils-native since that would almost always be a bogus
  dependency

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorecipetool: create: be more tolerant of spacing in configure.ac
Paul Eggleton [Wed, 9 Mar 2016 04:48:48 +0000 (17:48 +1300)]
recipetool: create: be more tolerant of spacing in configure.ac

Allow for whitespace in appropriate places, and ensure we match all
whitespace chars not just the space character.

(This fixes extracting dependencies from tmux's configure.ac, for
example.)

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agolib/sstatesig: skip shared_workdir when checking locked sigs
Joshua Lock [Wed, 9 Mar 2016 13:15:25 +0000 (13:15 +0000)]
lib/sstatesig: skip shared_workdir when checking locked sigs

do_shared_workdir is not a proper sstate task, it always reruns if
needed, so special-case it in warnings when checking locked sigs.

Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agopython3: fix patching get_python_lib() in distutils/sysconfig.py
Alexander Kanavin [Wed, 9 Mar 2016 15:01:49 +0000 (17:01 +0200)]
python3: fix patching get_python_lib() in distutils/sysconfig.py

Previous, two things were wrong:
1) lib_basename was set from STAGING_LIBDIR only if prefix parameter was empty or missing
2) if prefix was not empty, lib_basename reverted to sys.lib, even if STAGING_LIBDIR
should've overriden it

Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agopython3-native: use the previous version of python-config script
Alexander Kanavin [Wed, 9 Mar 2016 15:01:48 +0000 (17:01 +0200)]
python3-native: use the previous version of python-config script

In python 3.4 python-config was rewritten in shell, ironically
to support cross-compilation:
https://bugs.python.org/issue16235

This new shell version is broken in several ways, and doesn't
have our oe-specific tweaks. Let's revert to the old script,
which is still provided.

Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoqemu.bbclass: add qemu_wrapper_cmdline()
Alexander Kanavin [Wed, 9 Mar 2016 15:01:47 +0000 (17:01 +0200)]
qemu.bbclass: add qemu_wrapper_cmdline()

The class had qemu_run_binary() which was not suitable for gobject-introspection,
as it required the name of the binary to run.

qemu_wrapper_cmdline() returns just the command line string needed to run
binaries, and does not require the binary name.

Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agodb: remove the NO_UPDATE_REASON and replace it a comment about RPM
Mark Hatle [Mon, 29 Feb 2016 15:37:43 +0000 (09:37 -0600)]
db: remove the NO_UPDATE_REASON and replace it a comment about RPM

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
9 years agorpmresolve: It is not necessary to manually specify -lpopt
Mark Hatle [Fri, 4 Mar 2016 19:02:50 +0000 (13:02 -0600)]
rpmresolve: It is not necessary to manually specify -lpopt

Popt may be internal or external to rpm.  Either way the rpm libraries
will link properly with or without -lpopt.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
9 years agorpm: A number of the patches have been submitted upstream
Mark Hatle [Thu, 25 Feb 2016 16:54:37 +0000 (10:54 -0600)]
rpm: A number of the patches have been submitted upstream

Note the upstream submission in the patches.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
9 years agorpm: Enable specific crypto and digest settings via variables
Mark Hatle [Thu, 25 Feb 2016 15:49:57 +0000 (09:49 -0600)]
rpm: Enable specific crypto and digest settings via variables

Allow the user to set the specific digest and non-repudiable signature
algorithms.  This should be done on a distribution wide basis.

See recipe for exact instructions, but values are now set using:
RPM_FILE_DIGEST_ALGO (default 1 - md5)
RPM_SELF_SIGN_ALGO   (default DSA)

Also, change the PACKAGECONFIG to define the default crypto engine for
RPM5.  Not just the available crypto engines.  If a crypto engine is not
selected, the system will default to the internal beecrypt version.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
9 years agosecurity_flags.inc: Special flags are needed for RPM
Mark Hatle [Fri, 4 Mar 2016 19:03:52 +0000 (13:03 -0600)]
security_flags.inc: Special flags are needed for RPM

RPM interally has support to build and work with the stack protector.
This is disabled by default in the RPM package, and the proper settings
should be specified in the security_flags.  Using the default setting of
stack-protector-strong causes linking problems due to issues with libtool
selecting the wrong GCC objections to link against.

Falling back to the RPM values of stack-protector will permit linking to
work properly, and some level of protection.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
9 years agorpm: Uprev to rpm-5.4.16 (pre) and rpm-5.4+cvs to current CVS head
Mark Hatle [Tue, 23 Feb 2016 17:28:21 +0000 (11:28 -0600)]
rpm: Uprev to rpm-5.4.16 (pre) and rpm-5.4+cvs to current CVS head

meta/lib/oe/package_manager.py was also updated.  This ensures that any
diagnostic messages are ignored from the output of rpmresolve.

The patches have been split into bug fixes (things that belong upstream)
and local changes that are OE specific.

The following patches are obsolete and have been removed:

rpm-remove-sykcparse-decl.patch
fstack-protector-configure-check.patch
rpm-disable-Wno-override-init.patch
rpm-lua-fix-print.patch
rpm-rpmpgp-fix.patch
verify-fix-broken-logic-for-ghost-avoidance-Mark-Hat.patch

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
9 years agolinux-yocto/4.4: explicitly enable ftrace in tracing fragment
Bruce Ashfield [Tue, 8 Mar 2016 04:22:22 +0000 (23:22 -0500)]
linux-yocto/4.4: explicitly enable ftrace in tracing fragment

The recent split/factoring between production and development
kernel configurations, EXPERT, EMBEDDED and DEBUG_KERNEL are no longer
selected for all kernel types. This means that ftrace is no longer
selected by default in in standard BSPs, causing breakage in things
that relied on it.

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agolinux-yocto/4.4: iwlwifi: mvm: don't allow sched scans without matches to be started
Bruce Ashfield [Mon, 7 Mar 2016 16:45:41 +0000 (11:45 -0500)]
linux-yocto/4.4: iwlwifi: mvm: don't allow sched scans without matches to be started

Integrating the following commit for improved iwlwifi support:

    iwlwifi: mvm: don't allow sched scans without matches to be started

    commit 5e56276e7555b34550d51459a801ff75eca8b907 upstream.

    The firmware can perform a scheduled scan with not matchsets passed,
    but it can't send notification that results were found.  Since the
    userspace then cannot know when we got new results and the firmware
    wouldn't trigger a wake in case we are sleeping, it's better not to
    allow scans without matchsets.

    This fixes https://bugzilla.kernel.org/show_bug.cgi?id=110831

Signed-off-by: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Wu Zheng <wu.zheng@intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agolinux-yocto/kernel-meta: ktype refactoring: move DEBUG_KERNEL, EXPERT and EMBEDDED
Bruce Ashfield [Fri, 4 Mar 2016 18:21:16 +0000 (13:21 -0500)]
linux-yocto/kernel-meta: ktype refactoring: move DEBUG_KERNEL, EXPERT and EMBEDDED

Integrating the folliwing patch series from Cal:

  This patch series refactors the ktypes so that base and standard ktypes
  do not enable EMBEDDED, EXPERT, or DEBUG_KERNEL. The reason this
  decision was made is because production platforms likely do not want
  DEBUG_KERNEL enabled, and EMBEDDED selects EXPERT which selects
  DEBUG_KERNEL.

  A new ktype called "developer" was also created. This ktype
  enables the options now missing from standard and base, making it easy
  to maintain the functionality of a BSP through simply swapping the ktype
  from standard to developer. The preempt-rt ktype is now based off of
  developer in order to maintain its functionality.

  The new standard ktype does not include EMBEDDED, EXPERT, or
  DEBUG_KERNEL. Without DEBUG_KERNEL it loses a number of debug features
  that are selected by default. Without EXPERT it gains RFKILL_INPUT and
  DEBUG_MEMORY_INIT, while losing VMSPLIT_3G. These are only available to
  configure with CONFIG_EXPERT=y and default to EXPERT or !EXPERT. Not
  selecting EMBEDDED has no apparent impacts.

  Some coordination is required for this change, as existing BSPs WILL be
  affected, and will either need to accept the changes in the standard
  ktype or move to the developer ktype.

  California Sullivan (12):
    features/debug: add debug-kernel feature
    ktypes: add developer ktype
    ktypes/base: Disable EMBEDDED and DEBUG_KERNEL
    CONFIG_PROCESSOR_SELECT: do not enable
    intel-common-drivers.scc: move profiling and latencytop to a new file
    romley.scc remove profiling and latencytop features
    bsp/intel-common: add intel-core* developer BSPs
    preempt-rt.scc: include developer ktype instead of standard
    intel-common: add intel-developer-drivers.scc to preempt-rt BSPs
    CONFIG_I2C_I801: set option to yes in intel-core* BSPs
    bsp: add developer common-pc BSPs
    bsp: remove profiling and latencytop from non-developer common-pc BSPs

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoxmlto: tell xmlto where cp is
Ross Burton [Tue, 8 Mar 2016 11:27:35 +0000 (11:27 +0000)]
xmlto: tell xmlto where cp is

xmlto looks for a cp binary, and on e.g. Fedora 23 will find it at /usr/bin/cp
but most other distros have it at /bin/cp.  This causing problems with sharing
sstate between distributions, but as /bin is a symlink on F23 we can safely
force the path to /bin/cp.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agotoaster.bbclass: improve how we gather buildstats for Toaster
Elliot Smith [Tue, 8 Mar 2016 11:27:24 +0000 (11:27 +0000)]
toaster.bbclass: improve how we gather buildstats for Toaster

Clean up the code which gathers buildstats for Toaster,
and modify the field names so that the correct parts of the
buildstats files are used to derive the CPU usage values.

Also derive elapsed time for the build here, rather than in
Toaster, as we have ready access to the data in the correct
format.

[YOCTO #8842]

Signed-off-by: Elliot Smith <elliot.smith@intel.com>
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoimage-prelink: use STAGING_*_NATIVE variables
Ross Burton [Tue, 1 Mar 2016 08:54:38 +0000 (08:54 +0000)]
image-prelink: use STAGING_*_NATIVE variables

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agostrace: Backport fixes for compiling with clang
Khem Raj [Mon, 29 Feb 2016 19:41:43 +0000 (19:41 +0000)]
strace: Backport fixes for compiling with clang

Backport fixes needed to avoid use of VLAs which is not available
on clang/llvm

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoghostscript: 9.16 -> 9.18
Hongxu Jia [Mon, 29 Feb 2016 19:41:42 +0000 (19:41 +0000)]
ghostscript: 9.16 -> 9.18

- Backport patches to fix build failure caused by
  '--without-jbig2dec' and '--without-jbig2dec'.
  ...
  |make[1]: *** No rule to make target `obj/sjbig2_.dev',
  needed by `obj/sjbig2.dev'.  Stop.
  ...
  http://bugs.ghostscript.com/show_bug.cgi?id=696497

- Previously, it did not build the whole local libpng
  source in ghostscript, only picked up specific files
  and compile them. But on ghostscript 9.18, when the
  arm's FPU has been set to NEON (-mfpu=neon * with GCC),
  the selected file "libpng/pngrutil.c" needs to link
  'png_init_filter_functions_neon' which should be
  compiled by a non-selected file "libpng/arm/arm_init.c".
  ...
  |./obj/pngrutil.o: In function `png_init_filter_functions':
  |armv7a-neon-poky-linux-gnueabi/ghostscript/9.18-r0/build/../
  ghostscript-9.18/libpng/pngrutil.c:3921: undefined reference to
  `png_init_filter_functions_neon'
  ...
  So do not compile local libpng source in ghostscript,
  use shared libpng to instead.

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agofontconfig: Revert changes made to FcConfigAppFontAddDir() recently
Li Zhou [Tue, 1 Mar 2016 05:23:33 +0000 (13:23 +0800)]
fontconfig: Revert changes made to FcConfigAppFontAddDir() recently

Backport <commit 46ec6a52d4cc447cc3ff4a13b2067ecb76c9db2e> from
fontconfig upstream <http://cgit.freedesktop.org/fontconfig/> to solve
issue: fontconfig changes break pybootchartgui tool generating bootchart
png file.

Signed-off-by: Li Zhou <li.zhou@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agopopulate_sdk_ext: Make populate_sdk_ext nostamp
Randy Witt [Tue, 1 Mar 2016 18:08:21 +0000 (10:08 -0800)]
populate_sdk_ext: Make populate_sdk_ext nostamp

Since the metadata for multiple layers will be put into the sdk, anytime
those layers change the sdk needs to be rebuilt. Ideally, this would use
checksums to only rebuild when necessary.

However, since this feature needs to be in the release, opt for a less
intrusive change by setting the task to nostamp. Unnecessary rebuilds
may occur, but it's better than a user wondering why their changes
didn't get added.

Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agosystemd: binfmt should be added to SYSTEMD_PACKAGES only if binfmt is enabled
Herve Jourdain [Mon, 7 Mar 2016 15:59:32 +0000 (16:59 +0100)]
systemd: binfmt should be added to SYSTEMD_PACKAGES only if binfmt is enabled

Signed-off-by: Herve Jourdain <herve.jourdain@neuf.fr>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agolicense.bbclass: fix host contamination warnings for license files
Jagadeesh Krishnanjanappa [Sat, 28 Nov 2015 08:09:54 +0000 (13:39 +0530)]
license.bbclass: fix host contamination warnings for license files

We get below host contamination warnings of license files for
each recipe, when we try to create a separate ${PN}-lic package (which
contains license files), by setting LICENSE_CREATE_PACKAGE equal to "1"
in local.conf.

-- snip --
WARNING: QA Issue: libcgroup: /libcgroup-lic/usr/share/licenses/libcgroup/generic_LGPLv2.1 is owned by uid 5001, which is the same as the user running bitbake. This may be due to host contamination [host-user-contaminated]
WARNING: QA Issue: attr: /attr-lic/usr/share/licenses/attr/libattr.c is owned by uid 5001, which is the same as the user running bitbake. This may be due to host contamination [host-user-contaminated]
WARNING: QA Issue: bash: /bash-lic/usr/share/licenses/bash/COPYING is owned by uid 5001, which is the same as the user running bitbake. This may be due to host contamination [host-user-contaminated]
-- CUT --

Since the license files from source and OE-core, are populated in a normal
shell environment rather in pseudo environment (fakeroot); the ownership of
these files will be same as host user running bitbake. During the do_package
task (which runs in pseudo environment (fakeroot)), os.link preserves the
ownership of these license files as host user instead of root user.
This causes license files to have UID same as host user id and resulting in
above warnings during do_package_qa task.

Changing ownership of license files to root user (which has UID and GID as 0)
under pseudo environment will solve above warnings, and on exiting pseudo
environment the license files will continue to be owned by host user. Perform
this manipulation within try/except statements, as tasks which are not exected
under pseudo (such as do_populate_lic) result in OSError when trying to
change ownership of license files.

Signed-off-by: Jagadeesh Krishnanjanappa <jkrishnanjanappa@mvista.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agooeqa/selftest/buildoptions: Test build does not fail without git rev
Daniel Istrate [Fri, 4 Mar 2016 10:55:33 +0000 (12:55 +0200)]
oeqa/selftest/buildoptions: Test build does not fail without git rev

Test that layer git revisions are displayed and
do not fail without git repository.

fix for [YOCTO #8852]

Signed-off-by: Daniel Istrate <daniel.alexandrux.istrate@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agobusybox.inc: add tail symlink so busybox can commit suicide cleanly
Alejandro del Castillo [Tue, 1 Mar 2016 23:16:17 +0000 (17:16 -0600)]
busybox.inc: add tail symlink so busybox can commit suicide cleanly

update-alternatives-opkg uses tail, which requires a temporary symlink
on tmpdir during removal, to avoid errors of the type:

/usr/bin/update-alternatives: line 113: tail: command not found

Signed-off-by: Alejandro del Castillo <alejandro.delcastillo@ni.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoavahi-ui: add dbus to PACKAGECONFIG
Ross Burton [Tue, 8 Mar 2016 11:05:49 +0000 (11:05 +0000)]
avahi-ui: add dbus to PACKAGECONFIG

Now that avahi has a dbus PACKAGECONFIG we need to ensure it's enabled as
otherwise the avahi-ui module won't build.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoavahi: add missing intltool-native build dependency
Ross Burton [Tue, 8 Mar 2016 11:05:48 +0000 (11:05 +0000)]
avahi: add missing intltool-native build dependency

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoavahi: make dbus optional but default
Jens Rehsack [Tue, 1 Mar 2016 12:49:44 +0000 (13:49 +0100)]
avahi: make dbus optional but default

Since do_install fails when dbus is removed by .bbappend, add packageconfig
to allow users to get rid of desktop ipc helper dbus.

Signed-off-by: Jens Rehsack <sno@netbsd.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agooe-setup-builddir: tidy up local.conf and bblayers.conf commentary
Andre McCurdy [Fri, 4 Mar 2016 04:18:21 +0000 (20:18 -0800)]
oe-setup-builddir: tidy up local.conf and bblayers.conf commentary

Remove obsolete reference to enabling parallel build via local.conf.

Unify wording used to explain that local.conf and bblayers.conf have
been created based on defaults.

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agonet-tools: Add SCTP option support
Li Zhou [Tue, 1 Mar 2016 08:35:19 +0000 (16:35 +0800)]
net-tools: Add SCTP option support

Porting three net-tools SCTP related patches from
<https://archive.fedoraproject.org/pub/archive/fedora/linux/releases/
12/Fedora/source/SRPMS/net-tools-1.60-95.fc12.src.rpm> to add support
for SCTP option.

Signed-off-by: Li Zhou <li.zhou@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agotune-corei7.inc: Fix PACKAGE_EXTRA_ARCHS for corei7-32
Chang Rebecca Swee Fun [Wed, 2 Mar 2016 09:55:41 +0000 (17:55 +0800)]
tune-corei7.inc: Fix PACKAGE_EXTRA_ARCHS for corei7-32

Change the name to core2-32 from core2.

There's no AVAILTUNES with the name core2. Make sure that we specify
the correct TUNE name so PACKAGE_EXTRA_ARCHS is expanded correctly.

[ YOCTO #9197 ]

Signed-off-by: Chang Rebecca Swee Fun <rebecca.swee.fun.chang@intel.com>
Signed-off-by: Anuj Mittal <anujx.mittal@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoeudev: remove redundant udev_run assignment
Ross Burton [Mon, 7 Mar 2016 13:49:37 +0000 (13:49 +0000)]
eudev: remove redundant udev_run assignment

The path isn't correct as we use /run/udev now, also /run/udev is the default
value, and eudev doesn't let you override it like this anyway.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoxcursor-transparent-theme: use a version glob in the selftest bbappend
Ross Burton [Fri, 4 Mar 2016 09:46:46 +0000 (09:46 +0000)]
xcursor-transparent-theme: use a version glob in the selftest bbappend

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agopopulate_sdk_ext: Update after uninative changes
Richard Purdie [Mon, 7 Mar 2016 11:52:30 +0000 (11:52 +0000)]
populate_sdk_ext: Update after uninative changes

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoimage.bbclass: support chaining compression (aka conversion) commands
Patrick Ohly [Mon, 7 Mar 2016 14:51:14 +0000 (15:51 +0100)]
image.bbclass: support chaining compression (aka conversion) commands

It makes sense to use the compression mechanism also for conversion,
for example of a whole-disk image into .vdi (VirtualBox). That part
already works, like this:

   COMPRESSIONTYPES_append = " vdi"
   COMPRESS_CMD_vdi = "qemu-img convert -O vdi ${IMAGE_NAME}${IMAGE_NAME_SUFFIX}.${type} ${IMAGE_NAME}${IMAGE_NAME_SUFFIX}.${type}.vdi"
   IMAGE_DEPENDS_vdi = "qemu-native"

But then it also makes sense to allow compressing the resulting image,
which only works after enhancing the image.bbclass.

For example, suppose a custom image command produces "dsk" images. Then
it becomes possible to set
   IMAGE_FSTYPES = " dsk.xz dsk.vdi.xz"
and do_image_dsk will automatically produce the intermediate images,
convert to dsk.xz resp. dsk.vdi -> dsk.vdi.xz and delete all
intermediate images. Symlinks are also set correctly.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoimage.bbclass: fix incomplete .rootfs customization
Patrick Ohly [Mon, 7 Mar 2016 21:13:48 +0000 (22:13 +0100)]
image.bbclass: fix incomplete .rootfs customization

The patch for making the .rootfs configurable was incomplete: in the
python create_symlinks() method the new variable must be expanded
explicitly.

Not doing so broke the symlink creation and that led to hard build
failures in image types depending on the boot-directdisk.bbclass (like
qcow2) because the build_boot_dd() method relied on the symlink.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoimage creation: allow overriding .rootfs suffix
Patrick Ohly [Mon, 7 Mar 2016 17:07:52 +0000 (18:07 +0100)]
image creation: allow overriding .rootfs suffix

By default, the image file name contains ".rootfs" to distinguish the
image file from other files created during image building. However,
for certain image types (for example, .hddimg) the ".rootfs" suffix is
redundant because the type suffix alone already uniquely identifies
the main image file (core-image-minimal-intel-corei7-64.hddimg instead
of core-image-minimal-intel-corei7-64.rootfs.hddimg).

With this change, distros that prefer the shorter image name can
override the .rootfs suffix unconditionally with
   IMAGE_NAME_SUFFIX ?= '' in their distro configuration
or with some condition check like this:
   python () {
       if <whole-disk image format active>:
           d.setVar('IMAGE_NAME_SUFFIX', '')
   }

The exact logic when to remove the extra suffix depends on the distro
and how it enables its own image type.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoscripts/hob: drop
Paul Eggleton [Fri, 4 Mar 2016 03:22:53 +0000 (16:22 +1300)]
scripts/hob: drop

The Hob UI is gone from BitBake, so we should remove this shortcut
script.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoclasses/packageinfo: remove
Paul Eggleton [Fri, 4 Mar 2016 03:22:52 +0000 (16:22 +1300)]
classes/packageinfo: remove

This class was only used by Hob, and since Hob has now been removed we
can drop it as well.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoconf/documentation.conf: remove BBLAYERS_NON_REMOVABLE
Paul Eggleton [Fri, 4 Mar 2016 03:22:51 +0000 (16:22 +1300)]
conf/documentation.conf: remove BBLAYERS_NON_REMOVABLE

Hob was the only thing paying attention to this, and now Hob itself has
been removed we can remove this as well.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agobblayers.conf.sample: remove BBLAYERS_NON_REMOVABLE
Paul Eggleton [Fri, 4 Mar 2016 03:22:50 +0000 (16:22 +1300)]
bblayers.conf.sample: remove BBLAYERS_NON_REMOVABLE

Hob was the only thing paying attention to this, and now Hob itself has
been removed we can remove this as well.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoyocto-uninative: Add common include for uninative
Richard Purdie [Fri, 4 Mar 2016 16:30:57 +0000 (16:30 +0000)]
yocto-uninative: Add common include for uninative

uninative has some specific setup requirements. Rather than have everyone
doing this themselves, do this centrally and allow people to opt into it
based on some Yocto Project builds of the uninative tarballs.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agomtools: Drop GCONV_PATH manipulation
Richard Purdie [Mon, 7 Mar 2016 12:04:57 +0000 (12:04 +0000)]
mtools: Drop GCONV_PATH manipulation

Now that nativesdk-glibc handles GCONV_PATH itself we don't need to do
this here. This unbreaks mtools for the native case without uninative
since the existing patch wasn't nativesdk specific.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agouninative: Handle relocate of GCONV_PATH in libc
Richard Purdie [Sun, 6 Mar 2016 10:17:20 +0000 (10:17 +0000)]
uninative: Handle relocate of GCONV_PATH in libc

uninative hand codes the list of files which need relocation, add the libc
to that list to ensure GCONF_PATH is updated.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agouninative: Add checksum support
Richard Purdie [Mon, 7 Mar 2016 12:02:51 +0000 (12:02 +0000)]
uninative: Add checksum support

We need to be able to update uninative if the version changes. To do this,
stash a checksum of the installed uninative tarball into a file. If this
changes, we update uninative.

For cleaner download messages, we place the tarballs into directories
based on the checksum.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agouninative: Refactor common code
Richard Purdie [Mon, 7 Mar 2016 12:06:53 +0000 (12:06 +0000)]
uninative: Refactor common code

Move duplicate code into a common function

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agouninative: Use CXX11 ABI for interoperation between gcc4 and gcc5
Richard Purdie [Sat, 5 Mar 2016 08:22:33 +0000 (08:22 +0000)]
uninative: Use CXX11 ABI for interoperation between gcc4 and gcc5

We may see binaries built with gcc5 run or linked into gcc4 environment
so use the older libstdc++ standard for now until we don't support gcc4
on the host system. https://wiki.debian.org/GCC5 has more details about this.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agouninative: correctly enable uninative
Ross Burton [Fri, 4 Mar 2016 16:48:37 +0000 (16:48 +0000)]
uninative: correctly enable uninative

The previous attempt at soft-failing when uninative was enabled didn't actually
work, because the workers didn't evaluate the function that actually enabled
uninative.

In a BuildStarted handler we can check if we need to download or extract the
uninative tarball.

In a ConfigParsed handler on the workers we can check if the uninative loader is
present, and if so enable it.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoglibc: Add relocation of GCONV_PATH
Richard Purdie [Mon, 7 Mar 2016 10:46:11 +0000 (10:46 +0000)]
glibc: Add relocation of GCONV_PATH

We're seeing various failures where nativesdk glibc can't find the gconv
modules. We've tried various workaround but this needs fixing 'properly'.
One significant problem is knowing when a binary would use this glibc
and hence when to set the path and when not to.

Add the default path to its own special section which the sdk
relocation script already knows how to handle and remap.

This significantly improves the usability of uninative.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agouninative-tarball: Add glibc-gconv-iso8859-1 for guile
Richard Purdie [Sun, 6 Mar 2016 18:33:58 +0000 (18:33 +0000)]
uninative-tarball: Add glibc-gconv-iso8859-1 for guile

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agodkpg: Use tar everywhere (not gtar)
Richard Purdie [Mon, 7 Mar 2016 00:13:48 +0000 (00:13 +0000)]
dkpg: Use tar everywhere (not gtar)

All our systems usually have tar, if we want the native sstate to work
universally, we need to prefer this. This avoids a system with gtar
causing dpkg-native to use it and it not being present on some systems.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agogtk3+: Add missing DEPENDS on wayland-native
Richard Purdie [Mon, 7 Mar 2016 00:13:09 +0000 (00:13 +0000)]
gtk3+: Add missing DEPENDS on wayland-native

configure needs wayland-scanner which comes from wayland-native, add missing
DEPENDS.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agotune-cortexa17.inc: apply changes similar to a15
Trevor Woerner [Sun, 6 Mar 2016 00:07:35 +0000 (19:07 -0500)]
tune-cortexa17.inc: apply changes similar to a15

Apply the same sort of changes to the Cortex-A17 tune as were done in commit
35392025f3236f5e5393f9cf0857732da9a2e503.

Signed-off-by: Trevor Woerner <twoerner@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agosstate: Allow late expansion of NATIVELSBSTRING
Richard Purdie [Fri, 4 Mar 2016 17:10:48 +0000 (17:10 +0000)]
sstate: Allow late expansion of NATIVELSBSTRING

uninative needs to adjust NATIVELSBSTRING fairly late in the
configuration parsing process but the sstate code encodes it into
variables. Since this string doesn't vary on a per recipe basis, we
defer its expansion until usage time.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agolinuxloader/image-prelink/image-mklibs: Fix non-standard path prelinking
Richard Purdie [Fri, 4 Mar 2016 16:28:40 +0000 (16:28 +0000)]
linuxloader/image-prelink/image-mklibs: Fix non-standard path prelinking

Prelinking on x86-64 wasn't working out the box as it uses /lib and
not /lib64 for libs. Prelink was refusing to link as the dynamic loader
didn't match its idea of the right path. Passing in the --dyanmic-linker
option avoids this.

We can share code from image-mklibs so abstract that into a new class,
linuxloader.bbclass.

This does break prelinking of multilib images, I've opened a bug so we
can loop back and fix that problem, the code would need to iterate the
dynamic loaders (and setup ld.so.conf files for it).

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoinsane/prelink: Handle nonstandard library paths
Richard Purdie [Fri, 4 Mar 2016 16:27:11 +0000 (16:27 +0000)]
insane/prelink: Handle nonstandard library paths

Prelink contains some hardcoded assumptions about the path layout of
the target system. Unfortunately if the system doesn't match, prelink
doesn't work. This breaks:

a) prelink of those images
b) the unsafe-references-in-binaries QA test (which uses prelink-rtld)

One way to work around this is to construct an ld.so.conf file which
lists the library paths in question. We do this in sanity QA check and
in the rootfs prelink code, being careful not to trample any existing
target ld.so.conf.

There is an additional problem that $LIB references in RPATHs won't be
handled correctly, I've not see any system use these in reality though
so this change at least improves things.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoext-sdk-prepare: Catch setscene tasks which should have run but didn't
Richard Purdie [Fri, 4 Mar 2016 16:22:26 +0000 (16:22 +0000)]
ext-sdk-prepare: Catch setscene tasks which should have run but didn't

When installing the eSDK, if setscene task fail for some reason, the tests
would ignore this. This is bad since we assume they're working.

This adds some sanity test code which detects if setscene tasks are
needing to run and errors if there are any.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agocreaterepo: Fix stat floating timestamps
Richard Purdie [Thu, 3 Mar 2016 17:18:25 +0000 (17:18 +0000)]
createrepo: Fix stat floating timestamps

When reading the mtime from disk, the system can get a floating point
value. Convert this to an int for comparision purposes, else some
packages always get reindexed as the value in the index is an int.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoxmlto: ensure /bin/bash is used as bash
Ross Burton [Sat, 5 Mar 2016 00:12:03 +0000 (00:12 +0000)]
xmlto: ensure /bin/bash is used as bash

The xmlto script uses bashisms and checks at configure time to find a bash
binary.  If the build host has /bin/sh as bash then this gets detected, which
causes problems in native builds if the sstate is then shared to a machine with
/bin/sh as dash.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoopenssl: add a patch to fix parallel builds
Ross Burton [Sat, 5 Mar 2016 00:12:02 +0000 (00:12 +0000)]
openssl: add a patch to fix parallel builds

Apply a patch taken from Gentoo to hopefully fix the remaining parallel make
races.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoxdg-utils: remove trailing whitespace in multiline string
Ross Burton [Sat, 5 Mar 2016 00:12:01 +0000 (00:12 +0000)]
xdg-utils: remove trailing whitespace in multiline string

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agobtrfs-tools: Add libgcc to RDEPENDS
Robert Joslyn [Sat, 5 Mar 2016 00:54:19 +0000 (16:54 -0800)]
btrfs-tools: Add libgcc to RDEPENDS

libgcc_s.so.1 is required by btrfs-tools at runtime for certain
operations, such as scrub due to the use of pthread_cancel.

Signed-off-by: Robert Joslyn <robert_joslyn@selinc.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agobitbake.conf: Add libgcc-native to ASSUME_PROVIDED
Richard Purdie [Sat, 5 Mar 2016 08:29:17 +0000 (08:29 +0000)]
bitbake.conf: Add libgcc-native to ASSUME_PROVIDED

Changes to the btrfs-tools recipe means we need to add libgcc-native
to ASSUME_PROVIDED.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agonet-tools: Override CFLAGS/LDFLAGS in do_install too
Khem Raj [Sat, 5 Mar 2016 00:05:11 +0000 (00:05 +0000)]
net-tools: Override CFLAGS/LDFLAGS in do_install too

do_compile does this but do_install needs same env as well

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agonspr: Fix build regression on musl from last upgrade
Khem Raj [Sat, 5 Mar 2016 00:03:22 +0000 (00:03 +0000)]
nspr: Fix build regression on musl from last upgrade

This patch is no longer required as upstreamed has fixed
the problem in more comprehensive way

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agogdb: fix builds with internal readline and no static libraries
Ross Burton [Thu, 3 Mar 2016 21:01:13 +0000 (21:01 +0000)]
gdb: fix builds with internal readline and no static libraries

If gdb was configured to use the internal readline but static libraries were
disabled, gdb wouldn't dutifully not build libreadline.a which was a problem
when it tried to link with that library.

Solve this by ensuring --enable-static is passed to the readline configure.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agofeature-arm-thumb.inc: Fix thumb tune override warning
Nathan Rossi [Thu, 3 Mar 2016 16:01:00 +0000 (02:01 +1000)]
feature-arm-thumb.inc: Fix thumb tune override warning

Fix the quotes in the bb.utils.contains feature check so that the call
results in a boolean value instead of a string, which allows the warning
check to occur.

Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agorecipetool: create: fix support for AX_CHECK_LIBRARY
Paul Eggleton [Thu, 3 Mar 2016 17:44:56 +0000 (06:44 +1300)]
recipetool: create: fix support for AX_CHECK_LIBRARY

Clearly I didn't test this part of the code - lists don't have an "add"
method. Needless to say I have tested it now.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoformfactor: assume a keyboard is plugged in
Ross Burton [Thu, 3 Mar 2016 16:56:43 +0000 (16:56 +0000)]
formfactor: assume a keyboard is plugged in

A sensible assumption is that BSPs have a USB keyboard and mouse connected
unless told otherwise, so flip the logic in the formfactor config script that
previously assumed that a keyboard was not connected by default.

[ YOCTO #9174 ]

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoacl: Fix re pattern in test cases
He Zhe [Thu, 3 Mar 2016 07:16:43 +0000 (02:16 -0500)]
acl: Fix re pattern in test cases

ls adds a '.' at the end of the permission field list on SELinux
machines, filter this out so root tests work on SELinux machines.
And backport one patch for other tests.

Signed-off-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agogcc-runtime.inc: disable libitm for little endian MIPS too
Andre McCurdy [Thu, 3 Mar 2016 05:07:13 +0000 (21:07 -0800)]
gcc-runtime.inc: disable libitm for little endian MIPS too

libitm is already disabled for big endian MIPS, but needs to be
disabled for little endian MIPS targets too.

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agodevtool: add build-sdk subcommand
Paul Eggleton [Wed, 2 Mar 2016 10:44:21 +0000 (23:44 +1300)]
devtool: add build-sdk subcommand

Add a build-sdk command which is only available within the extensible
SDK that builds a derivative extensible SDK. The idea is recipes in the
workspace become a part of the new SDK - for example, this allows taking
a vendor provided SDK, adding a few libs and then producing a new SDK
with those included.

When normally building the extensible SDK, the workspace is excluded;
here we need to copy into the new SDK (renaming it in the process); the
recipes' task signatures become locked and thus the sources are no
longer needed, so they are removed along with the workspace bbappends
which would interfere with the locked signatures. Additionally we need
to just copy the configuration files (i.e. local.conf and auto.conf)
rather than filtering and appending to them since that work has already
been done when constructing the original SDK. The extra sstate artifacts
from workspace recipes are also determined and copied into the new SDK
in minimal mode (on the assumption that you won't set up a new sstate
mirror).

This reuses some code from build-image, so that needed to be
generalised to allow that.

Implements [YOCTO #8892].

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agodevtool: build-image: rename module
Paul Eggleton [Wed, 2 Mar 2016 10:44:20 +0000 (23:44 +1300)]
devtool: build-image: rename module

Hyphens aren't allowed in python identifiers, so you shouldn't use them
in module names or they are more difficult to import.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agooeqa/buildoptions: Improve unsafe references tests
Richard Purdie [Mon, 29 Feb 2016 15:36:48 +0000 (15:36 +0000)]
oeqa/buildoptions: Improve unsafe references tests

Fixing one of the recipes not to contain unsafe references to paths
results in this QA test failing. Improve the test so that we test
the recipe works, then if we intentionally break the recipe, the
issue is detected.

Also split out the binaries test from the scripts test. The binaries
issue may also get 'fixed' in future and need the same fix.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoinsane.bbclass: make the checking stricter for unsafe references in scripts
Chen Qi [Fri, 19 Feb 2016 02:38:39 +0000 (10:38 +0800)]
insane.bbclass: make the checking stricter for unsafe references in scripts

Previously, the checking for unsafe references is not strict enough. It
only checks whether '/usr/' is in the script. As a result, any script
containing statements like below will match this check.

   PATH="/bin:/sbin:/usr/bin:/usr/sbin"

However, as we can see, this is actually not an unsafe reference. What
we really want to check is something like '/usr/bin/tail', so we should
make the checking stricter.

This patch solves the QA warning in gzip and nfs-utils.

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
9 years agoe2fsprogs: do not enable non-stable features by default
Jonathan Liu [Tue, 1 Mar 2016 03:51:50 +0000 (14:51 +1100)]
e2fsprogs: do not enable non-stable features by default

Signed-off-by: Jonathan Liu <net147@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agosdk_update.py: Enable local sdk-update tests
Juro Bystricky [Wed, 2 Mar 2016 19:23:25 +0000 (11:23 -0800)]
sdk_update.py: Enable local sdk-update tests

Testing of local sdk updates was commented out.
Local sdk updates are functional now, so the tests should be re-enabled.

Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agosdk.py: Fix undefined variable
Juro Bystricky [Wed, 2 Mar 2016 19:23:24 +0000 (11:23 -0800)]
sdk.py: Fix undefined variable

"sdk_update" uses a variable newsdk_path, which was never declared.
This would cause the command:

    devtool sdk-update <poky-sdk-latest>

to fail with an error:
NameError: global name 'newsdk_path' is not defined

The remedy is to declare newsdk_path as it was no doubt intended,
corresponding to the argument specifying <poky-sdk-latest>.

[YOCTO#9042]

Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoeudev: recipe formatting improvements
Andre McCurdy [Wed, 2 Mar 2016 22:45:49 +0000 (14:45 -0800)]
eudev: recipe formatting improvements

Formatting improvements only, no functional changes.

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agoopenssl: Security fix Drown via 1.0.2g update
Armin Kuster [Wed, 2 Mar 2016 07:38:41 +0000 (23:38 -0800)]
openssl: Security fix Drown via 1.0.2g update

CVE-2016-0800
CVE-2016-0705
CVE-2016-0798
CVE-2016-0797
CVE-2016-0799
CVE-2016-0702
CVE-2016-0703
CVE-2016-0704

https://www.openssl.org/news/secadv/20160301.txt

Updated 2 debian patches to match changes in 1.0.2g

Signed-off-by: Armin Kuster <akuster@mvista.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agolayer.conf: Update after replacement of udev with eudev
Richard Purdie [Wed, 2 Mar 2016 23:07:09 +0000 (23:07 +0000)]
layer.conf: Update after replacement of udev with eudev

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agobootimg: set default value for LABELS variable
Ed Bartosh [Tue, 1 Mar 2016 09:00:56 +0000 (11:00 +0200)]
bootimg: set default value for LABELS variable

With empty LABELS variable build_efi_cfg skips most of its
functionality producing warning message:
    'LABELS not defined, nothing to do'
This causes build failure for efi images.

Setting default value for LABELS to 'boot install' should fix
this issue.

Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
9 years agosanity: Do not mistake meta-yocto-bsp for meta-yocto
Peter Kjellerstedt [Mon, 29 Feb 2016 16:28:56 +0000 (17:28 +0100)]
sanity: Do not mistake meta-yocto-bsp for meta-yocto

The code in oecore_update_bblayers() handling the transition from
meta-yocto to meta-poky was not very resilient. It would, e.g.,
mistake meta-yocto-bsp for meta-yocto if the former occurred before
the latter in BBLAYERS.

The code also failed to update multiple existences of meta-yocto in
the bblayers.conf file, e.g., if it was present in
BBLAYERS_NON_REMOVABLE in addition to BBLAYERS (which it is by
default).

Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
9 years agosanity.bbclass: remove conflict checking for image vm and live
Robert Yang [Wed, 24 Feb 2016 09:23:59 +0000 (01:23 -0800)]
sanity.bbclass: remove conflict checking for image vm and live

[YOCTO #9161]

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
9 years agosyslinux.bbclass: make vm and live can be built together
Robert Yang [Wed, 24 Feb 2016 09:23:58 +0000 (01:23 -0800)]
syslinux.bbclass: make vm and live can be built together

* The vm image(hdddirect, vmdk, qcow2, vdi) and live image (hddimg, iso)
  couldn't be built together because the following vars settings are
  conflicted:
  - SYSLINUX_ROOT (/dev/sda2 vs /dev/ram0)
  - LABELS (boot vs boot install)
  - INITRD (None vs live install)
  - SYSLINUX_CFG (see above)
  Introduce new vars (SYSLINUX_ROOT_VM/_LIVE, the samilar to others) to
  make them can work together, now we can build all of them together:

  IMAGE_FSTYPES += "live iso hddimg hdddirect vmdk qcow2 vdi"

* Use SYSLINUX_CFG rather than SYSLINUXCFG to keep align with others
  SYSLINUX vars.

* The SYSLINUX_TIMEOUT had been set, but it didn't work since
  AUTO_SYSLINUXMENU wasn't set, this would cause confusions, so also set
  AUTO_SYSLINUXMENU.

* Move SYSLINUX_PROMPT and SYSLINUX_TIMEOUT to syslinux.bbclass rather
  than in separate classes since they are the same.

* Set SYSLINUX_TIMEOUT to 50 to have a unique timeout for syslinux.

[YOCTO #9161]

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
9 years agorecipetool: create: add basic support for new npm fetcher/class
Paul Eggleton [Mon, 29 Feb 2016 11:48:26 +0000 (00:48 +1300)]
recipetool: create: add basic support for new npm fetcher/class

Add detection for npm modules and support for extracting the name and
version from package.json as is usually part of an npm module contents.

Note: this will likely only produce a buildable recipe if you use an
npm:// URL; simply pointing to a node.js source repository isn't going
to fetch the module's dependencies. It also doesn't set up the
shrinkwrap/lockdown automatically, so there is some room for improvement
later.

Implements [YOCTO #8690].

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
9 years agorecipetool: create: add basic support for generating linux kernel recipes
Paul Eggleton [Mon, 29 Feb 2016 11:48:25 +0000 (00:48 +1300)]
recipetool: create: add basic support for generating linux kernel recipes

Add support for detecting a Linux kernel source tree and generating a
basic kernel recipe using meta-skeleton's linux-yocto-custom recipe as a
base.

Implements [YOCTO #8981].

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
9 years agorecipetool: create: add support for out-of-tree kernel modules
Paul Eggleton [Mon, 29 Feb 2016 11:48:24 +0000 (00:48 +1300)]
recipetool: create: add support for out-of-tree kernel modules

Detect kernel modules by looking for #include <linux/module.h>, and
handle the various styles of Makefile that appear to be used. I was able
to use this code to successfully build a number of external kernel
modules I found.

Implements [YOCTO #8982].

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
9 years agogcc: Add support for atomic opertions (libitm) where available
Mark Hatle [Tue, 1 Mar 2016 04:18:41 +0000 (22:18 -0600)]
gcc: Add support for atomic opertions (libitm) where available

GCC 4.7 and newer have supported various automic operation directives,
however these have not been previously enabled.

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