linux-dummy: bundle_initramfs should not be nostamp any more

The same change has already been applied to kernel.bbclass in commit 55989cb5:
[ kernel.bbclass: Stop bundle_initramfs thwarting sstate cache and fix race ]

The dummy kernel should comply with it.

(From OE-Core rev: 0897fd6feb19b545af2ebc148a2f6f99341841a9)

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

qemu: add PACKAGECONFIG for numa

Add PACKAGECONFIG for numa to avoid build error:
backends/hostmem.c:21:20: fatal error: numaif.h: No such file or directory

Package numctl is from meta-oe.

(From OE-Core rev: 9661ce6ed5e01dd21360946c561b3c8a1ce9fc2c)

Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libtiff: fix CVE-2013-1961

Integrate community fix for the issue CVE-2013-1961
and migrated to version 4.0.3.

Stack-based buffer overflow in the t2p_write_pdf_page function
in tiff2pdf in libtiff before 4.0.3 allows remote attackers to
cause a denial of service (application crash) via a crafted
image length and resolution in a TIFF image file.

(From OE-Core rev: f24e3456c60951d2985d7c23bdcc1f8c15d6c167)

Signed-off-by: Priyanka Shobhan <priyanka_shobhan@mentor.com>
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Muzaffar Mahmood <muzaffar_mahmood@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd: Support building on uclibc

include missing.h to get MAX_HANDLE_SZ which
is not defined in uclibc.

(From OE-Core rev: fa2a90ca632d10fe1a14098c3f4fcacc7cea6ac5)

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

uclibc: Support systemd builds

Systemd needs some extra defines e.g.
ADJ_NANO and IPTOS_CLASS_CS6

(From OE-Core rev: 7562021eef5b7585122c92db8b686808ebe7d85e)

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

uclibc: Upgrade to tip of master

Adjust mips to cater for new mips32r2 defaulttune
Remove upstreamed patches

(From OE-Core rev: 8003ce0df3f6297391a709cc9720adffbd8fbd0f)

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

openssl: Re-add linux-uclibc tuple

With last restructuring for musl, some of uclibc targets got ignored
fsl/ppc and ARM worked ok since they use special target triplets which
were already considered but other like mips, x86 and so on failed

(From OE-Core rev: 63ab0ce2103bcf3a42ce5812a22409779126e114)

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

IBM power7 v2: Add new tune file for PPC power7

v2: rename file

(From OE-Core rev: 14c773f61a6380f76b58ea0c1cca6e6010d581f8)

Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

IBM power6 v2: Add new tune file for PPC power6

v2: rename file

(From OE-Core rev: b41be209514c2cb69359ee5e26f87beb078f01b2)

Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

IBM Power5 v2: Add new tune file for PPC power5 cpu

V2: rename file
(From OE-Core rev: fd46da6f37acbbac8a8b14d5991f75947688a9c2)

Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nss: CVE-2014-1544

the patch comes from:
https://bugzilla.redhat.com/show_bug.cgi?id=CVE-2014-1544
https://hg.mozilla.org/projects/nss/rev/204f22c527f8

author  Robert Relyea <rrelyea@redhat.com>
https://bugzilla.mozilla.org/show_bug.cgi?id=963150
Bug 963150: Add nssCertificate_AddRef and nssCertificate_Destroy calls
to PK11_ImportCert to prevent nssTrustDomain_AddCertsToCache from
freeing the CERTCertificate associated with the NSSCertificate. r=wtc.

(From OE-Core rev: 7ef613c7f4b9e4ff153766f31dae81fc4810c0df)

Signed-off-by: Li Wang <li.wang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

logrotate: add packageconfigs

add packageconfigs for acl and selinux.

(From OE-Core rev: 885cec11c978975d5c51ded92882a3b00d5ccf8c)

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

logrotate: obey our flags

Needed to quiet GNU_HASH warnings, and some minor fixes.

(From OE-Core rev: a46d3646a3e1781be4423b508ea63996b3cfca8a)

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

oeqa/sdk/buildsudoku.py: add setUpModule method to run only when gtk+ in installed.

Adding setUpModule in order to skip the module when gtk+ is not installed in the toolchain.

(From OE-Core rev: 97ac0fc33b9277825089ac36f9037d472d397b71)

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

oeqa/oetest.py: enable sdk tests to use hasFeature and hasPackage methods.

In order to use the hasFeature and hasPackage methods, we need to make oeSDKTest extend oeTest and also set the test context (tc) attribute in the oeTest class when loading the tests.

(From OE-Core rev: 54436aeed5ac5e0da0779919f8524a0603e19c6a)

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

classes/testimage.bbclass: add more fields to the sdk TestContext

In order to use hasFeature and hasPackage methods in sdk test modules, we need specific fields to be set in the TestContext object.
Adding pkgmanifest, imagefeatures and distrofeatures to the TestContext.

(From OE-Core rev: cd342b399b2d78724032cdd7042968d3238cd548)

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

classes/populate_sdk_base.bbclass: add a manifest for target sdk

Similar to the way BSP images have rootfs a manifest, the toolchain now also has a manifest file created alongside the sdk image.

(From OE-Core rev: 3d42fd1f050a1382b15c3c4d59fd02d0ed7091b2)

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

qemux86-64: support X11 when QEMU is emulating a different GPU than vmware

QEMU is capable of emulating four different VGA adapters: cirrus, std, vmware,
and QXL. By adding the cirrus and fbdev X.Org drivers to the qemux86-64 image,
the image can be made to launch an X server on when cirrus and std are chosen,
in addition to just vmware. (The build of QEMU in OE-Core appears to have QXL
disabled, meaning a driver for it is unnecessary.)

The runqemu script now allows the choice of emulated VGA adapter to be
specified manually, so it's important that qemux86-64 supports any configuration
the user might choose without requiring the image to be rebuilt.

(From OE-Core rev: 1216de77a7f23fa10e34aee1ebe27fcc6a6589c0)

Signed-off-by: Max Eliaser <max.eliaser@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemux86-64: xorg.conf: allow X.Org driver to be selected automatically.

A section specifying the driver to use is deleted from the image's xorg.conf,
allowing the X server to automatically select the most appropriate driver.
Testing shows that it does always pick the correct one.

(From OE-Core rev: 5dd70b1917bc60e4ea7bded8f348e46b4551dfb6)

Signed-off-by: Max Eliaser <max.eliaser@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemux86: support X11 when QEMU is emulating a different GPU than vmware

QEMU is capable of emulating four different VGA adapters: cirrus, std, vmware,
and QXL. By adding the cirrus and fbdev X.Org drivers to the qemux86 image,
the image can be made to launch an X server on when cirrus and std are chosen,
in addition to just vmware. (The build of QEMU in OE-Core appears to have QXL
disabled, meaning a driver for it is unnecessary.)

The runqemu script now allows the choice of emulated VGA adapter to be
specified manually, so it's important that qemux86 supports any configuration
the user might choose without requiring the image to be rebuilt.

(From OE-Core rev: 9e4ca6739d65716fcb0a1b7d635749083da98c52)

Signed-off-by: Max Eliaser <max.eliaser@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemux86: xorg.conf: allow X.Org driver to be selected automatically.

A section specifying the driver to use is deleted from the image's xorg.conf,
allowing the X server to automatically select the most appropriate driver.
Testing shows that it does always pick the correct one.

(From OE-Core rev: 1354fdea6e04070fde4cb5a48ef824d0b22f0870)

Signed-off-by: Max Eliaser <max.eliaser@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xorg-drivers: xf86-video-cirrus: add a recipe for the Cirrus Logic X.Org driver

This driver allows X.Org to use the Cirrus Logic family of VGA adapters. Since
QEMU can emulate a Cirrus VGA adapter, this driver will be of use for images
that are expected to run under QEMU, if QEMU's other VGA adapters are
unavailable.

(From OE-Core rev: d407734e3d25b66a0281cc287a3e929945dc52ed)

Signed-off-by: Max Eliaser <max.eliaser@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

archiver.bbclass: add revision to git tarfile name

When archiving original source, insert the first 10
characters of srcrev, after stripping off any
AUTOINC+, into the filename for tarfiles created
from directories.

(From OE-Core rev: 36726049a6a452f85d7cc849d57ad6af05d6d3ea)

Signed-off-by: Joe Slater <jslater@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hicolor-icon-theme: Upgrade to 0.13

(From OE-Core rev: b6cc93f1008e2313eb9a17a11bfbc9980736878d)

Signed-off-by: Roxana Ciobanu <roxana.ciobanu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xkeyboard-config: upgrade to 2.12

(From OE-Core rev: a277a80f788082fa01d3794443e454d1a6a6c1da)

Signed-off-by: Roxana Ciobanu <roxana.ciobanu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xf86-input-mouse: upgrade to 1.9.1

(From OE-Core rev: 67c8388f2c4b40d29273c55c8eebd85448e85de0)

Signed-off-by: Roxana Ciobanu <roxana.ciobanu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

xf86-input-synaptics: upgrade to 1.8.0

    The two main changes in this release:
    - the eventcomm backend uses libevdev, we recommend libevdev 1.2 or newer
    - support for a secondary software button area for the Lenovo *40 series
    (T440, T540, X240, Helix, Yoga, ...).

(From OE-Core rev: a445d82fd4e7b3849a3708061368efb541f64e66)

Signed-off-by: Ciobanu Roxana <roxana.ciobanu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

psmisc: Update patch status

The patch "Include <limits.h> for PATH_MAX" was accepted into the master branch
of psmisc by the maintainer. It should be included in the next release.

(From OE-Core rev: 2da97e3638e6fd7730df55963ad8392f7054c7dd)

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

native/nativesdk: Clear MACHINEOVERRIDES

In cases where we override variables e.g PACKAGE_ARCH
conditionally and use a MACHINEOVERRIDE to qualify the
override like

PACKAGE_ARCH_<MACHINE> = "some target PACKAGE_ARCH"

This would also be effective if this recipe is
BBEXTENDED to native or nativesdk which is undesired
This particular issue will cause wrong sysroot to be
used for nativesdk case since PACKAGE_ARCH would have
been modified and eventually changing value of
STAGING_DIR_TARGET which we use as --sysroot option
to cross and cross-canadian built SDK gcc

(From OE-Core rev: 6907da7eb44dca812dbb9e214357ee96e9ff738a)

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

sanity: refactor mirrors checks to be more pythonic

- Use clearer variable names
- Use variable unpacking to reference elements by name rather than index
- Sacrifice a small amount of time (iterate over protocols twice per entry
  rather than once) for clarity: use readable generator expressions with any()
  rather than maintaining state.

(From OE-Core rev: 9d31e1e6ce07991fe360e67295311e62a55603af)

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

sanity: fix support for regex schemes in mirrors check

(From OE-Core rev: c8afcb9cab9d610892db9c41b29685583f3b5773)

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

sanity: handle both \n and \\n in mirror vars

(From OE-Core rev: 3e203e91afa48557eb754dd554944012f7f0c0a2)

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

autogen-native: inherit pkgconfig to fix a build failure

This patch is to inherit pkgconfig to fix a build
failure.

(From OE-Core rev: 217aa4d9802609d2c3628c8751a27a5d25900898)

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

libxml2: Explicitly enable zlib support

The zlib support is a must if you are using RPM backend. So this
explicitly enable it and adds a comment in the recipe to avoid its
removal by mistake.

(From OE-Core rev: 7d056397ab9912316064db850aae05aacabc726c)

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

buildhistory: Restore parsing time optimalization

* it was added in:
  http://git.openembedded.org/openembedded-core/commit/meta/classes/buildhistory.bbclass?id=540a2a30be21c3eca4323efbe91e7dcfc31a4c97
* then accidentally removed in:
  http://git.openembedded.org/openembedded-core/commit/meta/classes/buildhistory.bbclass?id=6db6c74f2876df7a183990426bb4262972966441

(From OE-Core rev: cbd868c34718e8fd2a41f5382da71ee959ea4e71)

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

bitbake: lib/bb/*.py: Typo fixes/grammar/comment fixes, nothing functional.

(Bitbake rev: 587b144ee409d444494d8d7f2d1c53ede8f7c953)

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: depexp: don't use undefined variable 'logging'

This fixes an issue when trying to use depexp without specifying a
recipe. Before change:

  $ bitbake -g -udepexp
  Traceback (most recent call last):
    File ".../bitbake/bin/bitbake", line 382, in <module>
      ret = main()
    File ".../bitbake/bin/bitbake", line 370, in main
      return ui_module.main(server_connection.connection, server_connection.events, configParams)
    File ".../bitbake/lib/bb/ui/depexp.py", line 201, in main
      logger.error(cmdline['msg'])
  NameError: global name 'logger' is not defined

After change:

  $ bitbake -g -udepexp
  Please specify a package name for dependency graph generation.

(Bitbake rev: 984ad90b2f1e29634dc79803a4a0404ab0534039)

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

bitbake.conf: Drop unused MKTEMP* variables

(From OE-Core rev: 2bfe071d141117ddf41eade5404a0d27c349bbe8)

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

package: Disable runtime mapping of RPROVIDES/RCONFLICTS/RREPLACES

It doesn't really make sense to run the remapping code on these
variables. If specific renaming is needed, it should be applied
manually. This means that the debian RPROVIDES of the original package
name can be preserved. There was also a bug report about this on the
OE-Core mailing list recently where someone else ran into this problem
too.

(From OE-Core rev: a0cd18ff5405b63b8296b8ef1fc2a488b906f576)

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

package: Don't remap names in packagegroup recipes

For package groups, use the original package names and don't use the
remapped (e.g. debian) naming.

(From OE-Core rev: afde71e8784144f05cebb0bdeeff6b3184819362)

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

debian: Set RPROVIDES to include the original packagename when renaming

Currently its hard for a packagegroup recipe to depend on something who's name
can change unpredictably. We therefore add in RPROVIDES for the original
package name so that these are also available to use as dependencies.

(From OE-Core rev: 3409c4379559afbb1d1d29045582995147a33bbc)

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

sstatesig: Only squash dependencies for allarch packagegroups

The idea of squashing packagegroup dependencies was to avoid allarch
packages rebuilding upon tune/arch changes. Now that the allarch class
inclusion is conditional, we can narrow down the packagegroup squashing
to be specifically applied to allarch recipes.

(From OE-Core rev: bd8018792155ae842952432900c8431feeaffe30)

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

packagegroup: Move inherit packagegroup to after PACKAGE_ARCH

This means these classes now avoid the allarch inherit entirely.

(From OE-Core rev: a56aed2e6e1b73bffd2bf82145fe48b82fde9292)

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

allarch: Add warning about packagegroup

Since we want to start including this class conditionally, detect cases
where packagegroup files use the old ordering and inform the user they
need to update this.

(From OE-Core rev: 30aaada4b79fc1acad5fdaba98071b63c52b9e05)

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

packagegroup: Make allarch inherit conditional

Make the allarch inherit conditional on PACKAGE_ARCH being left as
all. The one downside to this approach is that recipes need to set
PACKAGE_ARCH *before* inheriting the class in order to avoid the inherit.

The advantage is that we could start to detect the use of the
allarch inherit in the sstatesig code for improved task checksums.

(From OE-Core rev: 9c826962ec8fa45c2b035427442b90a41517144e)

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

bitbake: cooker: tweak CookerCollectFiles::find_bbfiles

since python-2.5 string.endswith() takes a tuple

(Bitbake rev: 86a67a1fd4244da9343dbf14deed1ad0d3003f32)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

guile: workaround ice ssa corruption while DEBUG_BUILD

Add:
MACHINE = "qemux86-64"
DEBUG_BUILD = "1"
to local.conf and there was a ssa corruption to build guile:
..
Unable to coalesce ssa_names 48 and 3476 which are marked as MUST COALESCE.
sp_48(ab) and  sp_3476(ab)
guile-2.0.11/libguile/vm-engine.c: In function 'vm_debug_engine':
guile-2.0.11/libguile/vm.c:673:19: internal compiler error: SSA corruption
 #define VM_NAME   vm_debug_engine
                   ^
guile-2.0.11/libguile/vm-engine.c:39:1: note: in expansion of macro 'VM_NAME'
 VM_NAME (SCM vm, SCM program, SCM *argv, int nargs)
 ^
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
...

Tweak libguile/vm-i-system.c to check the value of nshuf to workaround it.
The tweak didn't change previous logic.

[YOCTO #6411]

(From OE-Core rev: 3470bbf4ae178479bd090d7191d6f0469f154b0d)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

puzzles: fix do_compile failed when DEBUG_BUILD

There were compiling failure with option -g -O
...
././gtk.c: In function 'configure_area':
././gtk.c:397:2: error: 'cr' may be used uninitialized in this function [-Werror=maybe-uninitialized]
  cairo_set_source_rgb(cr,
  ^
././gtk.c:384:14: note: 'cr' was declared here
     cairo_t *cr;
              ^
././gtk.c: In function 'main':
././gtk.c:2911:6: error: 'error' may be used uninitialized in this function [-Werror=maybe-uninitialized]
      fprintf(stderr, "%s: %s\n", pname, error);
      ^
cc1: all warnings being treated as errors
...

Initialized the pointers 'cr' and 'error' with NULL

[YOCTO #6410]

(From OE-Core rev: 2281180499cf3bec5c7e75cb596725dab190f72c)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

fixup

(From OE-Core rev: 20992e8a00c585ab6b0611ca23ccd57e2b86b727)

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

busybox : fix do_compile failed on qemumips when DEBUG_BUILD (ICE)

Build busybox on mips with -g -O, there is a ice failure:
...
shell/ash.c:8758:1: internal compiler error: in dwarf2out_var_location, at dwarf2out.c:21264
...
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
{standard input}: Assembler messages:
{standard input}: Warning: missing .end at end of assembly
{standard input}: Error: open CFI at the end of file; missing .cfi_endproc directive
...

The following line caused the ice failure on mips:
...
void evaltreenr(union node *, int) __attribute__ ((alias("evaltree"),__noreturn__));
...
So remove evaltreenr which was a alias of evaltree, and use evaltree
to instead.

[YOCTO #6625]

(From OE-Core rev: e9e2884a29e95703c5b1a2fffe5e6febfa77b5cd)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

opkg: Add runtime dep on gnupg if 'gpg' is in PACAKGECONFIG

The 'opkg-key' utility requires gpg to manage the list of trusted keys. The
ability to verify package signatures is not much use without the ability to add
keys to the trusted list...

(From OE-Core rev: ebd7dce320dfdbac449b60e23cfe41125b658319)

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

gpgme: Add pkg-config file

A basic gpgme.pc file is installed with gpgme which should allow the library to
be found and used using pkg-config rather than gpgme-config.

(From OE-Core rev: 066eb8fbfd90f1a5af01a8f63adbe0cd2ae8cb5c)

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

bind: refer ubuntu/redhat to add bind user/group

We refer what ubuntu/redhat did, gave bind a user/group

Here is the example in ubuntu 14.04:

$ ps -eo user,group,cmd | grep "named"
...
bind     bind     /usr/sbin/named -u bind
...

$vim bind9_1%3a9.9.5.dfsg-3_amd64.deb/postinst
...
    # lets give them a bind user/group in all cases.
    getent group bind >/dev/null 2>&1 || addgroup --system bind
    getent passwd bind >/dev/null 2>&1 ||
    adduser --system --home /var/cache/bind --no-create-home \
        --disabled-password --ingroup bind bind
...

(From OE-Core rev: e37841faf746895f41627130623196c0bebe0740)

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ptest-runner: return 1 instead of zero if any test fails

It is better to know this information from the exit code rather than parsing
test specific outputs.

(From OE-Core rev: d6b1c9e7e29aaa8d71ccb485b7a88863a117a8b1)

Signed-off-by: Laszlo Papp <lpapp@kde.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev-cache: parametrize sysconf file paths

The udev-cache facility uses files that represent system states, to
ensure that the cache tarball is valid to apply. These paths were
hardcoded in several places; collect them into SYSCONF_CACHED and
SYSCONF_TMP.

(From OE-Core rev: f4fcb237da0d1013005e9a0bb2381cfeb4c5316c)

Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev-cache: Don't ignore error messages from cache extract

Previous changes should obviate all known spurious errors coming out of
tar. Since real extraction failures can and will occur, stop redirecting
stdout/stderr to /dev/null.

Take this opportunity to also remove an unnecessary subshell.

(From OE-Core rev: 09089962be353280201ba3899fd5ef9cc3c0ba32)

Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev-cache: omit sockets and filesystems mounted under /dev

Archiving sockets causes tar to report an error and return a nonzero
exit status. Archiving a mounted filesystem is harmless, but may greatly
bloat the size of the cache tarball, and wastes time on boot.

To fix these issues, use `find` to only include the files we want, which
are the file types that udev will create (block/char devices and
symlinks) that are on the same filesystem as /dev.

While we're at it, remove a subshell by archiving /dev as an absolute
path. However, `tar` will complain about stripping the leading slash on
stderr. To inhibit this, `cut` out the leading slash.

An alternative solution is to use `tar --exclude`, but that is modestly
more brittle, since we'd need to explicitly list every socket and
filesystem to exclude. Note that `tar --one-file-system` is
GNU-specific, and tar implementations generally have nothing equivalent
to `find -type`.

If using busybox `find`, this change requires CONFIG_FEATURE_FIND_TYPE=y
and CONFIG_FEATURE_FIND_XDEV=y. If using busybox `tar`, this change
requires CONFIG_FEATURE_TAR_FROM=y.

(From OE-Core rev: e89df123e2ec516ae61763eab3c9e78e067e28d5)

Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev-cache: strip timestamps on extract

Under normal udev operation, device nodes are obviously timestamped
based on the system time at current boot. However, when using
udev-cache, they are timestamped from a previous boot.

The existence of machines lacking RTCs makes this more than a cosmetic
issue: if the current time is set further on in the boot, so that the
system time is still 1970 by the time the cache is extracted, tar will
print a timestamp warning for every extracted file (potentially hundreds
of them).

To fix, use -m on extract.

If using busybox `tar`, this commit requires
CONFIG_FEATURE_TAR_NOPRESERVE_TIME=y.

(From OE-Core rev: b31f8f1f053cdfa9428e3f667c05e7e2c600061e)

Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: enable `tar -m`

CONFIG_FEATURE_TAR_NOPRESERVE_TIME=y is needed for the commit
"udev-cache: strip timestamps on extract". Enabling this flag increases
the size of busybox by 0 bytes on x86.

(From OE-Core rev: 11a772b3297ef36f3ebc4b67c9477022352d1c55)

Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev-cache: choose a more descriptive cache filename

"/etc/dev.tar.gz" doesn't adequately imply that udev-cache maintains it.
Instead, call it "/etc/udev-cache.tar.gz".

(From OE-Core rev: b33c678e2a5d42138a15b69802c5ddd109cb3b55)

Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Ben Shelton <ben.shelton@ni.com>
Acked-by: Gratian Crisan <gratian.crisan@ni.com>
Acked-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev-cache: Compress the cache

$DEVCACHE is observed to be 100k uncompressed; compressing it reduces
its size to ~5k. But compress it outside of `tar` so that archival
operation takes as little time as possible, to minimize the risk of
devices being created/removed during execution.

(From OE-Core rev: 571df6ddba1caa6805f7c96cd592eea399c2aee2)

Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

udev-cache: Update cache tarball atomically

Overwriting the tarball in-place could cause a partial write, if the
system stops at an inopportune time. This is mitigated by first writing
to a temporary file, then moving that file on top of the final location.

(From OE-Core rev: 5dbf43fe32a1cf259f9379a7d2c008260eabf3ac)

Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Ben Shelton <ben.shelton@ni.com>
Acked-by: Gratian Crisan <gratian.crisan@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.14: feature merges and configuration updates

As part of LTSI efforts, the 3.14 kernel is receiving feature backports
from newer kernels.

With this change, we update the 3.14 SRCREVs to make the following
changes available:

 - full nohz backports (v3.16)
 - loopback NFS (v3.16)
 - numa updates (v3.16)
 - cgroups: vm_cgroup: basic infrastructure
 - cgroups: Resource controller for open files

And the following configuration change:

 - meta: features/ieee802154: add configuration fragment for IEEE 802.15.4
 - meta: features/bluetooth: add configuration fragment for Bluetooth support
 - meta: intel-common-standard: add ericson-3g to intel-common-standard.scc
 - cgroups: enable virtual memory resource controller
 - full nohz: Enable full dynticks system for nohz

(From OE-Core rev: fe2f51571f97e2cceae974f6a18782025e080c32)

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

linux-yocto/3.14: cgroups and hugetlb updates

Update SRCREVs for the following feature ports:

    d61940e2aaee fs/hugetlbfs/inode.c: remove null test before kfree
    01a9d1b96a67 fs/hugetlbfs/inode.c: use static const for dentry_operations
    b58086c38038 fs/hugetlbfs/inode.c: add static to hugetlbfs_i_mmap_mutex_key
    182a45b35b74 fs/hugetlbfs/inode.c: complete conversion to pr_foo()
    a3c4b02bbb05 mm: hugetlb: fix copy_hugetlb_page_range()
    3757fbab4c89 hugetlb: rename hugepage_migration_support() to ..._supported()
    4d91ab2d26a0 mm, hugetlb: move the error handle logic out of normal code path
    49ea68f6eb8b hugetlb: add support for gigantic page allocation at runtime
    78dcff916fdf hugetlb: move helpers up in the file
    07b911e0a765 hugetlb: update_and_free_page(): don't clear PG_reserved bit
    d38ffe086659 hugetlb: add hstate_is_gigantic()
    46b2dd4acbaa hugetlb: prep_compound_gigantic_page(): drop __init marker
    8b4da9338c22 hugetlb: ensure hugepage access is denied if hugepages are not supported
    450e7d23d98c mm/hugetlb.c: add NULL check of return value of huge_pte_offset
    370cd4423fc4 mm, hugetlb: mark some bootstrap functions as __init
    31f0f8869bc9 mm, hugetlb: improve page-fault scalability
    f57f488c939b mm, hugetlb: use vma_resv_map() map types
    3263ee955d10 mm, hugetlb: remove resv_map_put
    92067ee5ebfd mm, hugetlb: fix race in region tracking
    30e1ab0b7502 mm, hugetlb: improve, cleanup resv_map parameters
    2d5919397a7b mm, hugetlb: unify region structure handling
    a36af1327d41 Fix [RFC] cgroups: Resource controller for open files.
    7dc531576e41 cgroups: Resource controller for open files.

(From OE-Core rev: 37fb3b84f701b88e95d6a3e2efccc46e8c01c1cf)

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

linux-yocto/3.14: fix common clock framework support in PCI glue layer

The initial version of "spi/pxa2xx: Add common clock framework support in PCI glue layer"
enables the baytrail series of processors, but breaks some existing users of
the framework.

There's a new version of the patch out for review, so we revert the broken one and
udpate to the latest.

(From OE-Core rev: beae066d5af96e78d3bbaa52884922cb39bb48fd)

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

linux-yocto/3.14: KASLR and VDSO backports

Updating the 3.14 linux-yocto SRCREVs for feature backports:

   6cfe9c448741 x86, kaslr: boot-time selectable with hibernation
   881a5f5812fd x86, kaslr: fix module lock ordering problem
   121ec40d905d x86, kaslr: randomize module base load address
   6efd2d21273c x86, vdso: Add 32 bit VDSO time support for 64 bit kernel
   642441e17072 x86, vdso: Add 32 bit VDSO time support for 32 bit kernel
   1d8dd2e617d4 x86, vdso: Patch alternatives in the 32-bit VDSO
   353f3eadb16d x86, vdso: Introduce VVAR marco for vdso32
   5e6e6ec6eb5c x86, vdso: Cleanup __vdso_gettimeofday()
   91efa61ed6c2 x86, vdso: Replace VVAR(vsyscall_gtod_data) by gtod macro
   f9c259ee1b2f x86, vdso: __vdso_clock_gettime() cleanup
   63d9b1c4ec72 x86, vdso: Revamp vclock_gettime.c
   e3a8ed44c964 mm: Add new func _install_special_mapping() to mmap.c
   02a67e32e58d x86, vdso: Make vsyscall_gtod_data handling x86 generic
   d076cfb68504 x86, vdso, xen: Remove stray reference to FIX_VDSO
   949153696c99 x86_32, mm: Remove user bit from identity map PDE
   e5cf7dfec74d x86, vdso: Remove compat vdso support

(From OE-Core rev: 00dcb084682b7599f2a3e69dcb4978cbb38f5c58)

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

linux-yocto/3.14: update to v3.14.13

Updating from 3.14.5 -> 3.14.13 to pick up the latest series of korg
stable updates.

(From OE-Core rev: bb1b6b7e580d5dfebf8aa424bff83f9ccadfba19)

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

linux-yocto/3.14: add ARM kprobes support

Adding the following commits to the 3.14 tree for ARM kprobe support:

   c1750e55eb75 ARM: add uprobes support
   33e8a1334afe ARM: Make arch_specific_insn a define for new arch_probes_insn structure
   b1d3feead083 ARM: Add an emulate flag to the kprobes/uprobes instruction decode functions
   84b5b70a01a2 ARM: Change the remaining shared kprobes/uprobes symbols to something generic
   624e5e9aab5a ARM: Rename the shared kprobes/uprobe return value enum
   7741eac39c2e ARM: Change more ARM kprobes symbol names to something more generic
   47f51bac6a2d ARM: Make the kprobes condition_check symbol names more generic
   3af2f454bb8e ARM: Remove use of struct kprobe from generic probes code
   a31964d3143a ARM: use a function table for determining instruction interpreter action
   6de77767abfc ARM: move generic thumb instruction parsing code to new files for use by other feature
   eb2db188b856 ARM: Move generic arm instruction parsing code to new files for sharing between features
   bbc0a43ff46e ARM: move shared uprobe/kprobe definitions into new include file
   f5f2896020e6 uprobes: allow ignoring of probe hits
   cb133db6bd60 ARM: Fix missing includes in kprobes sources

(From OE-Core rev: 835bb10d07d1a64119379e9837dd853dacfe79bc)

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

linux-yocto: intel-common: Enable Industrial IO

Updating the META branch SRCREV to import iio configuration settings.

(From OE-Core rev: bb8cada68e70b1246a76046a37ba57a3ca7aedde)

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

sanity.bbclass: check the format of MIRRORS

Check the format of MIRRORS, PREMIRRORS and SSTATE_MIRRORS:
* Each mirror shoudl contain two memebers.
* The local "file://" url must use absolute path (file:///).
* The protocol must in protocols list.

(From OE-Core rev: c8c213bb25b137cf70ba8ce9a45e60065d926735)

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

sanity.bbclass/kernel.bbclass: minor fixes

Fixed:
- Remove an unused line
- Remove unneeded code after d.getVar(), we don't need the "or ''" after
  d.getVar() unless we need a string.
- typo: PREMIRROS -> PREMIRRORS.

(From OE-Core rev: 7849e50107a27b0ff2aaac47480ac1a0a79533dc)

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

gmp: ppc64 build issue

tmp-gcd_1.s: Assembler messages:
| tmp-gcd_1.s:94: Error: unsupported relocation against BMOD_1_TO_MOD_1_THRESHOLD
| make[2]: *** [gcd_1.lo] Error 1

V2: fixed PN name

(From OE-Core rev: add84a15023d6c277340d6ab4810c605fab01e00)

Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mc: Fix musl build failure

The CTRL macro just needs to be defined.

(From OE-Core rev: 27b63e39413b1bc5209270ffcd4888e9a4795ad1)

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

libc-common.bbclass: rename ALL the packages

The DEBIAN_NAMES feature renames some of the libc packages to
"libc6*" names --but only some. A previous patch added the -dbg
package. However, this doesn't cover other packages (such as
the -doc package), and it didn't take multilibs into account.

(From OE-Core rev: 64362a8f7ad7534515a9448445606581f3c572f6)

Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Jeff Polk <jeff.polk@windriver.com>
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

expect: fix shared object file cannot be opened

When we require Expect package in tcl script, we will get following error:

couldn't load file "/usr/lib/expect5.45/libexpect5.45.so": /usr/lib/expect5.45/libexpect5.45.so: cannot open shared object file: No such file or directory
    while executing
"load /usr/lib/expect5.45/libexpect5.45.so"
    ("package ifneeded Expect 5.45" script)
    invoked from within
"package require Expect"
    (file "hello.tcl" line 3)

This patch fixes this issue.

(From OE-Core rev: 43813ddfa8cade9ea1c5de24dcd4e275b1486cff)

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

sudo: remove volatile configuration file

The new version of sudo has fixed the problem and will create the
directory if it doesn't exist. So the configuration file is no longer
needed.

(From OE-Core rev: 7c962ca5d01ae4b9f511bb2b3b7700f970051727)

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

bitbake: siggen: Fix a subtle bug in hash calculation for shared work tasks

With a shared work task like gcc, the task can be run from a variety of
different recipes which may have different virtual extensions in place.

Depending on whether gcc-runtime or nativesdk-gcc-runtime's do_preconfigure
task is called for example will change the sorting of the task hashes due
to the way clean_basename currently works.

The correct thing to do here is sort on the base filename first, then any
extension when ordering the hashes. This means we do account for things
like recipes with both a native and non-native dependency but we also fix
the shared work case where we don't care whether it was a virtual version
or not.

(Bitbake rev: 2e80b5d10a5037ed6f0bc227a1f9b42529c87086)

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

bitbake: process: Further improve robustness against server shutdown

Currently, if an exception occurs in an event handler, the server shuts
down but the UI simply hangs. This happens in two places, firstly waiting
for events and secondly, sending events to a server which no longer exists.

The latter does time out, the former does not. These patches improve
both code sections to check if the main server process is alive and if not,
trigger things to shut down gracefully. This avoids the timeout in the
command sending case too.

This resolves various cases where the UI would simply hang indefintely.

(Bitbake rev: ac418e1112ff5f9c3157569316902f7a27fba4b4)

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

bitbake: utils: Add workaround for multiprocessing bug

Our usage of multitprocessing is problematic. In particular, there is a bug
in python 2.7 multiprocessing where signals are not handled until command
completion instead of immediately.

This adds a workaround into our wrapper function to deal with the issue.

(Bitbake rev: a16185e602b39b71475aa7e9ee80ad2b1f28d0f7)

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

bitbake: command: Trigger updateCache to shut down any active parser threads

If we trigger a shutdown whilst parsing for whatever reason, in some
cases we were not closing down the parser threads. This change
ensures we do so. The function names are not entirely intuitive
but the behaviour is more correct (and commented). The previous bug
with the stdout failure would trigger this one, if there was a cold
cache and parsing was required (but not otherwise).

(Bitbake rev: 25bfa2478f1c3a8eb695e1e5760e06db5be8f2fc)

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

bitbake: bitbake-worker: Improve sigterm handler

When processes terminate, we really want all of the child processes to
terminate too. This was not happening for worker processes which spawned their
own multiprocessing pools, leading to build hangs. This change ensures any
sigterm gets passed to the whole process group. In local tests, this resolved
some hanging process workloads I could generate. It does rely on signals
being delivered in a timely fashion and there is a multiprocessing bug we have
to work around there.

(Bitbake rev: 96f8ea07ace1379380fab2d78eb592fa40c867d4)

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

bitbake: knotty: Improve exception handling

Currently, IOErrors are just passed over due to the broken Exception
clause. A command like "bitbake X | <invalid command>" would break stdout
triggering a traceback. With these changes we print the exceptions, shut down
the server gracefully and exit which is a much nicer behaviour and is less
confusion to the user.

(Bitbake rev: 9544108f7b413038d871ce6ca88232de2f2434d9)

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

bitbake: process: Deal with infinite looping of the server

Currently if an exception occurs, we just run the idle handler again and
again, usually looping indefintely. Chances are the exception that occurred
will keep occuring and this is not a good place to be.

This was breaking the autobuilders with gigabytes of logs.

At least improve things so the cooker shuts down gracefully when this happens.
Some trace of the original problem may still be present on the console too!

(Bitbake rev: 1f28d8d3311262427938180435b68f0a35c2b330)

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

bitbake.conf/debian.bbclass: Move AUTO_LIBNAME_PKGS definition to class file

Might as well move this default to the class which uses it allowing
for easier reading/understanding of the class.

(From OE-Core rev: 177aec177306e68bcd822dee6b29a7efbd558a91)

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

packagegroup-core-tools-debug: Drop MACHINE_ARCH as PACKAGE_ARCH

There is nothing machine specific about this recipe.

(From OE-Core rev: 07848cee072e62969d19b2fc9627783bcb016bbe)

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

package.bbclass: Fixup for using common function

We may as well use the common function for this rather than
duplicating the code.

(From OE-Core rev: 20979854c52607212c16e0b01cf610c80d8b524e)

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

lib/oe/utils: Add utils function for multiprocess execution

Our usage of multitprocessing is problematic. In particular, there is a bug
in python 2.7 multiprocessing where signals are not handled until command
completion instead of immediately.

This factors the multiprocess code into a function which is enhanced with
a workaround to ensure immediate signal handling and also better SIGINT
handling which should happen in the parent, not the children to ensure
clean exits. The workaround for the signals is being added to the core
bb.utils function so it can benefit all users.

package_manager is then converted to use the new code.

(From OE-Core rev: 72d153a3a90d31d9f4e41d77da24e44ccb33c56e)

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

bitbake.conf: Set PACKAGE_ARCH with ??=

Currently its near impossible for other classes to sanely override
this value with their own default. By setting a weak default we can
allow other classes to change the default and allow end recipes to
again override this.

As far as I can tell, there shouldn't be any regressions from this
change.

(From OE-Core rev: 12b2a73d336d66596939eae5c9947d4054c0316e)

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

oeqa/utils/httpserver: Reset SIGTERM handler

With bitbake-worker installing a SIGTERM handler, we now need to reset the one here
to ensure that when this process shuts down, it doesn't take the rest of the task
with it. This does appear to be the only place in OE that we have this problem.

(From OE-Core rev: e1ac67d6470dde70239ca0430b18ca0bffbc0295)

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

archiver: delete the tail slash in directory name

If directory names in SRC_URI ended with "/", function
do_ar_original() in layers/oe-core/meta/classes/archiver.bbclass
will generate a tar file whose name is ".tar.gz".
So delete the "/" at the tail of the directory names before use the names.

(From OE-Core rev: a539e823d002fefe129e3045f893d1237fadb87f)

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

grub_2.00: strip the sparc64 files from building

Avoid building and packaging of sparc64 specific files in the x86 builds.

Fixes Bug:
[YOCTO #6427]

(From OE-Core rev: e9d1264a63f47965c2f10e18e532b117bbda26be)

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

mesa: fix Multilib QA Issue

This patch fixed warning:
WARNING: Multilib QA Issue: lib32-mesa package lib32-mesa-megadriver - suspicious values
'mesa-driver-swrast mesa-driver-nouveau-vieux mesa-driver-r200 mesa-driver-i915 mesa-driver-radeon mesa-driver-i965' in RPROVIDES
WARNING: Multilib QA Issue: lib32-mesa package lib32-mesa-megadriver - suspicious values
'mesa-driver-swrast mesa-driver-nouveau-vieux mesa-driver-r200 mesa-driver-i915 mesa-driver-radeon mesa-driver-i965' in RREPLACES
WARNING: Multilib QA Issue: lib32-mesa package lib32-mesa-megadriver - suspicious values
'mesa-driver-swrast mesa-driver-nouveau-vieux mesa-driver-r200 mesa-driver-i915 mesa-driver-radeon mesa-driver-i965' in RCONFLICTS

(From OE-Core rev: 95f16f4bba660f0a6108fbbc98ad9caf69e6036b)

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

udev: fix /var/volatile/tmp permissions

Mariadb postinstall will fail with:

ERROR: 1  Can't create/write to file '/tmp/#sql_a2a_0.MAI' (Errcode: 13)
140814  8:14:06 [ERROR] Aborting

/tmp is a link to /var/volatile/tmp, set as:

drwxr-xr-x 3 root root 60 Aug 14 08:07 .

populate-volatiles.sh does not create the directory or set
permissions, because it already exists:

Checking for -/var/volatile/tmp-.
Creating directory -/var/volatile/tmp-.
Target already exists. Skipping.

Traced the creation to the initfile in udev. Create with -m 1755
to be sure. With this patch applied, mysql postinst succeeds and
creates the neccesary user tables.

(From OE-Core rev: 5743c7632ac9c33051f6fca60f5507d3808e8e3a)

Signed-off-by: Riku Voipio <riku.voipio@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grep: set CONFIG_SHELL to /bin/sh

It would be bash, sh, ksh or sh5 according to the host if we don't set
this, and its scripts don't have bashism as the checkbashisms shows.

(From OE-Core rev: 6fe82a9ed6c7967aaf25c861400de4a9379e92c6)

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

gzip: set CONFIG_SHELL to /bin/sh

It would be bash, sh, ksh or sh5 according to the host if we don't set
this, and its scripts don't have bashism as the checkbashisms shows.

(From OE-Core rev: 49ee0da024dcb80026b5dc23367e79e0ed80e6aa)

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

xz: set CONFIG_SHELL to /bin/sh

It would be bash, sh, ksh or sh5 according to the host if we don't set
this, and its scripts don't have bashism as the checkbashisms shows.

(From OE-Core rev: 786530de31553b3e0a93874bdc425bfca50a5866)

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

qemumips.conf: Default to (and support) mips32r2

The MIPS emulation for qemumips actually supports
mips32r2:
isa                     : mips1 mips2 mips32r1 mips32r2

We should probably use that tuning file.

This implicitly changes the default value of DEFAULTTUNE to
mips32r2.

(From OE-Core rev: 5d64516d81750e4e0d65792a3215568d652bec6c)

Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

fix up eglibc-utils description to match what is actually included

iconf must have been a typo for iconv
rpcinfo is found elsewhere (currently rpcbind) and not eglibc-utils - replace its mention with
rpcgen

(From OE-Core rev: ae96e831e62e37a018cc924d38a338ffc05ab4da)

Signed-off-by: Bob Cochran <openembedded@mindchasers.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

qemu: Explicitly disable libiscsi, its not in DEPENDS

(From OE-Core rev: 71fd367f6a460257b6a5866b4932762b041e9563)

Signed-off-by: Jonathan Liu <net147@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

strace: fix ptest execution failure

ptest needs runtest-TESTS target.
serial-tests is required to generate this target.

(From OE-Core rev: d8e65c07ee8a4c6e29922ec37af82ed31ff814cc)

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