base.bbclass: Fix PACKAGECONFIG handling when no flags are set

When the main PACKAGECONFIG variable was empty with no flags set,
the options were not being added to explicitly disable features.
This patch corrects that problem and ensures the disable fields
are correctly parsed and added to variables.

(From OE-Core rev: d7b8c247227f3cc82f92292407f548927e9fde78)

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

polkit: Default to shadow, not none to fix build failures

Without this builds would fail due to an invalid confiuration being
selected by the configure script.

(From OE-Core rev: d62b083fe1ee33cbd3ea3929e592c403000fb3b0)

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

bitbake: Drop fetch v1, v2 provides a much better codebase to build from

(Bitbake rev: 292e3430e5140b602cad86f55b5453e8cebb28a1)

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

meta/classes: Convert to use appendVar and appendVarFlags

(From OE-Core rev: 3b57de68e70e77dbc03c0616a83a29a2e99e40b4)

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

libx11: reinstate PE=1 that got lost in the big meta-oe Xorg merge

This prevents package managers 'upgrading' 1.4.4 to 1:1.4.3

(From OE-Core rev: 906641853e8ac6fd296dfb91f8bb647c7da8e23e)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gtk+ 2.24.8: bump PR to force rebuild and feed update

This should errors similar to this one:

/usr/libexec/gdm-simple-greeter: symbol lookup error: /usr/lib/libgdk-x11-2.0.so.0: undefined symbol: _XGetRequest

(From OE-Core rev: 98b95330b66406625c8fef9b58f2ac7983585fcb)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

polkit 0.104: add 2 backports from upstream to fix crashes during auth

(From OE-Core rev: bcb8e65ab3f7d7f0f884f05c758f8a779c260306)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

polkit 0.104: convert pam code to PACKAGECONFIG and adhere to OE recipe styleguide

(From OE-Core rev: 7a6fcc7f79445e5f290dd499c223619a2a426f78)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta: Replace bb.data.expand(xxx, d) -> d.expand(xxx)

sed \
 -e 's:bb.data.\(expand([^,()]*\), *\([^) ]*\) *):\2.\1):g' \
 -i `grep -ril bb.data.expand *`

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

sstate: Remove unecessary bb.data.expand()

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

meta: Convert getVar/getVarFlag(xxx, 1) -> (xxx, True)

Using "1" with getVar is bad coding style and "True" is preferred.
This patch is a sed over the meta directory of the form:

sed \
 -e 's:\(\.getVar([^,()]*, \)1 *):\1True):g' \
 -e 's:\(\.getVarFlag([^,()]*, [^,()]*, \)1 *):\1True):g' \
 -i `grep -ril getVar *`

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

meta/classes: Convert to use appendVar and appendVarFlags

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

soc-family.inc: to be included in machine.conf to add SOC_FAMILY to MACHINEOVERRIDE

Add a soc-family.inc file that can be included in a machine.conf to enable
the use of SOC_FAMILY in MACHINEOVERRIDE, which could be useful to group
multiple machines with the same common base. Some examples can be seen in
meta-ti BSP layer.

(From OE-Core rev: 641cdbc7ee0186053dd541e0dd5fb7b03b1c10d1)

Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libx11-trim: Add error information from LSB Test Suite to makekeys_crosscompile.patch

Append error information from LSB Test Suite to this patch for an illustrative purpose \
why this patches should be needed.

(From OE-Core rev: 0e6dfa5dcf324a1d32b634bbcd9487106a6f5413)

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libx11-1.4.4: Add patch "makekeys_crosscompile.patch"

Add patch "makekeys_crosscompile.patch" to libx11-1.4.4 for fixing
bug 1970.

[YOCTO #1970]

(From OE-Core rev: 21499be467ce14f165b56c1ce67507f7c41c5a1c)

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libx11-1.4.4: Add patch makekeys_crosscompile.patch

LSB 4.1 complain a host contamination error from libx11 because of absent
patch "makekeys_crosscompile.patch" from libx11-trim-1.4.4.

The error from LSB Test Suite is as follow:
VSW5TESTSUITE PURPOSE 7
Assertion XStringToKeysym-7.(A)
When the string argument is the name of a KeySym in the
table with the prefix XK_ removed, then a call to
XStringToKeysym returns that KeySym.
METH: For each KeySym name in table with code G:
METH: Call XStringToKeysym to obtain the KeySym defined for that string.
METH: Verify that XStringToKeysym did not return NoSymbol.
METH: Verify that the returned string is correct.
CHECK: XStringToKeysym-7 1, line 130
CHECK: XStringToKeysym-7 2, line 140
CHECK: XStringToKeysym-7 3, line 150
CHECK: XStringToKeysym-7 4, line 160
CHECK: XStringToKeysym-7 5, line 170
CHECK: XStringToKeysym-7 6, line 180
CHECK: XStringToKeysym-7 7, line 190
CHECK: XStringToKeysym-7 8, line 200
CHECK: XStringToKeysym-7 9, line 210
CHECK: XStringToKeysym-7 10, line 220
CHECK: XStringToKeysym-7 11, line 230
CHECK: XStringToKeysym-7 12, line 240
CHECK: XStringToKeysym-7 13, line 250
CHECK: XStringToKeysym-7 14, line 260
CHECK: XStringToKeysym-7 15, line 270
CHECK: XStringToKeysym-7 16, line 280
CHECK: XStringToKeysym-7 17, line 290
CHECK: XStringToKeysym-7 18, line 300
CHECK: XStringToKeysym-7 19, line 310
CHECK: XStringToKeysym-7 20, line 320

[YOCTO #1970]

(From OE-Core rev: c9824083d4ab5bfd37fa1efdfbc3e93288cb47b6)

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc-cross.inc: Fix cross testing scripts to work with dash

dash does not like >& so be explicit and say 2>&1

(From OE-Core rev: 1d262630853e65be9167d904b934b581acf64182)

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

eglibc-testing: Fix testing script to work well with dash

Dash did not like >& so we do 2 >& 1 > /dev/null

(From OE-Core rev: 06c783e2d6f4e3b316b230565b28d9e4c535c31b)

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

wpa-supplicant: fix breakage during rebuild

Install phase currently edits files in $S with sed. This causes breakage
if install is run a second time (due to sstate hash rebuild for example)

The result is hidden build breakage, in particular
/usr/share/dbus-1/system-services/fi.w1.wpa_supplicant1.service now contains:

Exec=/usr/usr/sbin/wpa_supplicant -u

rather than:

Exec=/usr/sbin/wpa_supplicant -u

This patch does the sed edit after the files are copied to $D, which should be safe.

(From OE-Core rev: 857840472705af1c0fbb8db917b4bb6809b929a6)

Signed-off-by: Steve Sakoman <steve@sakoman.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image_types: Ensure duplicate compressed types aren't created

This allows IMAGE_FSTYPES = "tar.bz2 tar.bz2" to work (and now
is faster since it will only do it once).

(From OE-Core rev: b24d4bcde21b013e48eaffea5f3a70d8b1df1047)

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

image_types.bbclass: We need to preserve order in the types variable and avoid set()

(From OE-Core rev: c547169f7dfe043affc2ddc5ae0a9091379df6de)

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

base.bbclass: If unpacking again, wipe out ${S}/patches

If we unpack again, its assumed the data in any patches directory is invalid
since do_patch will run again. This ensures old patch data doesn't get
reused in a confused way.

Ideally we should probably wipe out ${S} here but that is probably a change
for another time.

[YOCTO #2043 partially]

(From OE-Core rev: 5fe5e6a15f26f23f0c5b863fafad7a0d382a55e2)

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

lib/oe/patch.py: Fix and improve PatchTree() resolver logic

Currently, if PATCHRESOLVE is user and and PatchTree() is being used, you can
get backtraces if patch application fails. This is because even in the failure
case, self._current is incremented, meaning second time around, there are array
range issues.

This patch changes the code so _current is only incremented upon successful
patch application, thereby resolving this failure.

Secondly, if you bitbake -c patch -f a recipe using PatchTree(), the
clean method was unimplemented leading to patch failures.

The other part of this patch changes the logic so a series file and
set of applied patches are maintained in a quilt like fashion. This
means a the Clean method can be implemented correctly and rerunning
the patch task of an existing patches source now works reliably.

[YOCTO #2043 partially]

(From OE-Core rev: f0fc47aea37793a62c43f10eea27ca014c420924)

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

qt4: make pulseaudio support conditional upon DISTRO_FEATURES

Add a pulseaudio DISTRO_FEATURES item and disable pulseaudio support in
phonon within Qt4 if it is not present; otherwise it is a race condition
as to whether it is enabled or disabled or breaks the build.

Note that this adds pulseaudio to DISTRO_FEATURES_BACKFILL thus leaving
existing distro configurations unaffected.

(From OE-Core rev: b82c216c1ee8e2a009e87856b7adad08f7f50482)

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

stderr would previously be appended to stdout, corrupting the result when something was outputed to stderr but exit code was still 0 (non-fatal warning messages). This commit makes the code parse only stdout, but output stderr if an error happened.

(Bitbake rev: 4a480a052f450c4ee061ab0e60a495a45f140cf9)

Signed-off-by: Adrien Bustany <adrien.bustany@nokia.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake/bitbake-prserv: Set file mode bits "x".

[YOCTO #2051] Set 'x' bit to make bitbake-prserv executable.

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

initscripts: Create the volatiles cache atomically.

If a machine loses power while building the volatiles cache, it will
continue to operate with an incomplete set of volatiles. Fix this by
updating atomically.

(From OE-Core rev: a60432ff4588390f5f955a6da234298e958546e6)

Signed-off-by: James Limbouris <james@digitalmatter.com.au>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto: update to v3.2.8

3.2.8 was released and contains a floating point fix for x86 32
bit and some x86-64 fixes. So it's worth picking up right away.

(From OE-Core rev: 8bd3038b62eaf46f9ac09fcdc1ca5e6f6e53a126)

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

linux-yocto/meta: add pvr feature description

Updating the 3.0 and 3.2 kernel SRCREVS to make the PVR feature
description available to machines with the appropriate hardware.

(From OE-Core rev: 0027bb45a40d209f43088e2c2b38191446ea1190)

Signed-off-by: Kishore Bodke <kishore.k.bodke@intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto: make kernel_checkout reentrant

The steps in do_kernel_checkout modify the source tree in WORKDIR.
If it is called multiple times, or interrupted, the tree is left
in an inconsistent state.

This change adds protections around branch names, and around the
manipulations of directories to ensure that it is safe to call
at any point.

(From OE-Core rev: f937977f241e786c5a7438449ed4c9da4c55829b)

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

sudo: upgrade to 1.8.4

During the upgrade to 1.8.4, the UCB-licensed fnmatch.c was replaced
with a non-recursive BSD-licensed version, hence the removal of UCB
and addition of BSD in the LICENSE field. This led to checksum changes
in the doc/LICENSE file, and we now additionally track the comment
headers in redblack.c.

These changes were confirmed on the sudo mailing list:

http://www.sudo.ws/pipermail/sudo-workers/2012-February/000736.html

This upgrade also fixes CVE-2012-0809.

(From OE-Core rev: 7147a569758414467c9d022e4c11fbc303e050aa)

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

image_types: Ensure dependencies for compression types containing multiple dots are handled correctly

This ensures dependencies for image types like ext2.gz.u-boot are handled correctly.

(From OE-Core rev: d94cb18f7f17a492c7a948337d7d2d1d0785573f)

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

V2 Fix libpam's chmod

The libpam's has an error when generating the rootfs:

chmod: cannot access `/usr/sbin/unix_chkpwd': No such file or directory

This is because the following code in libpam_1.1.5.bb:

pkg_postinst_pam-plugin-unix () {
    # below is necessary to allow unix_chkpwd get user info from shadow file
    # on lsb images
    chmod 4755 ${sbindir}/unix_chkpwd
}

This is to set the setuid permission for unix_chkpwd (the lsb test
requires this), but it lacks a "${D}", and we can do this in the install
stage.

[YOCTO #2049]

(From OE-Core rev: 0725c7f01b173b1cc2090f4a03a274c7017b8a1a)

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

scripts/oe-git-proxy-socks-command: Add fallback to use nc

If our own proxy command isn't available for some reason and nc is available,
fall back to use it.

(From OE-Core rev: aaab6ae94258277dcd2f29823a64689e6245fa7a)

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

scripts/oe-git-proxy-socks-command: Improve error fallback/handling

If oe-git-proxy-socks isn't available, try and create it.

If that fails, tell the user there is a problem, don't just fail
to find the command.

[YOCTO #2007]

(From OE-Core rev: 46c675d85603d7a573bc59638ce615aba9fd7df2)

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

busybox: Restore 'date' compatability

Restore CONFIG_FEATURE_DATE_COMPAT so that all versions of 'date',
whether from busybox or coreutils, agree on the format when
setting the date from the command line.

(From OE-Core rev: 5506cc481b7afd9a30af53e8966bcbe2ff67b6d0)

Signed-off-by: Gary Thomas <gary@mlbassoc.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

initscripts: Properly format date when set from timestamp

Reformat date, as stored in /etc/timestamp, to match CLI format.

(From OE-Core rev: 7aa4c51089cc4a2811bd11842647839d1e4b1e95)

Signed-off-by: Gary Thomas <gary@mlbassoc.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libproxy: Fix build errors due to missing prototypes

g++ really does not like the missing prototypes
here we were missing close() and read() so include
unistd.h to get them

(From OE-Core rev: f32b30ef4711fe602a11549f77be06a0eef2fc0d)

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

site.conf.sample: Fix broken SOCKS proxy setup and configuration

SOCKS proxy specification with git was using conflicting methods and
thus was failing when mixed SOCKS needs were in place (requiring no
proxy for some hosts and proxy for the rest)

- GIT_PROXY_COMMAND is an environment variable GIT uses to OVERRIDE
  all proxy configuration in ~/.gitconfig or any other gitconfig. By
  using it to configure, it was breaking havoc on site git
  configuration or the one generated by bitbake in tmp/.

  Renamed to OE_GIT_PROXY_COMMAND in meta/conf/site.conf.sample
   (with a doc tidbit on the name chosen), meta/classes/base.bbclass.

- The gitconfig generated by bitbake was wrong. There was a typo error
  (gitproxy vs gitProxy), thus all lines were being ignored. Fixed in
  meta/classes/base.bbclass.

- The gitconfig generated was being placed in
  ${STAGING_DIR_NATIVE}/usr/etc/gitconfig; git was looking for it in
  ${STAGING_DIR_NATIVE}/etc/gitconfig. Fixed that in
  meta/classes/base.bbclass, at the same time creating a
  GIT_CONFIG_PATH variable, since it is also referenced in
  generate_git_config() and have all instances refer to that.

(From OE-Core rev: e579eb7f33462258c8e82a0936d970593614840d)

Signed-off-by: Inaky Perez-Gonzalez <inaky.perez-gonzalez@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

externalsrc.bbclass: Add class for handling external source trees

This is loosly based upon srctree.bbclass from OE-Classic but with some
changes appropriate to OE-Core.

(From OE-Core rev: bdb341953ba7d8299cba4d49d857107fb7b01e5b)

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

toolchain-scripts: Exclude variables causing the recipe to become unecessarily machine specific

   basehash changed from 30e97f9eeed1df8488b62b4fb47a3a0c to 72ea9a277e6599ae9052b169b9a94f1b
   Variable TOOLCHAIN_CONFIGSITE_SYSROOTCACHE value changed from
   /OE/oe-core/tmp-eglibc/sysroots/qemux86-64/usr/share/x86_64-oe-linux_config_site.d
   to
   /OE/oe-core/tmp-eglibc/sysroots/qemux86-64copy/usr/share/x86_64-oe-linux_config_site.d
   Variable TIME value changed from 084543 to 085638

(From OE-Core rev: 9a7895b24c09e347e3d708107fb24b9e42b1d9c5)

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

util-linux: Fix build when NLS is disabled

AC_CHECK_DECLS generated defines should be
checked with #if not #ifdef

(From OE-Core rev: 14b86a350d07b0da550a7a39886a660210918fbf)

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

distro_tracking: Upgrade and Manual Checks

(From OE-Core rev: 983c1264e93d19bc9e605633f224c73e961912ae)

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

util-linux: Fix removal of lscpu for non-GPLv3 Build

(From OE-Core rev: 998abea2748c6c7e0873ffd382b2bb2fd233c3f1)

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

pixman: updat to 0.24.4

(From OE-Core rev: de046e430cf2954ae83dea7e27cb1da5d5ab59d3)

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

glproto: Update to 1.4.15

(From OE-Core rev: 90e72f89757e0fcb9dafa0a87220704e8a4d32f2)

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

hdparm: update to 9.39

(From OE-Core rev: c81b2241181e373cf07126b80537f2e4d13217df)

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

netbase: fix QA Warnings

WARNING: For recipe netbase, the following files/directories were installed but not shipped in any package:
WARNING:   /usr/sbin
(From OE-Core rev: 64f3a58026909b4462e8e1a3e2ccabdf9bdb4468)

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

perl: remove empty directories to clean QA Warnings

WARNING: For recipe perl, the following files/directories were installed but not shipped in any package:
WARNING:   /usr/lib/perl/site_perl
WARNING:   /usr/lib/perl/site_perl/5.14.2
(From OE-Core rev: 2bda13df6feee87ea6fcecdce96fb5234cfa5674)

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

libx11-trim: Drop obsolete 1.3.4 version

(From OE-Core rev: 6cb72d5f008c6cbf4960d9dc260b82a056042b0f)

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

self-hosted-image: Create a VMDK image with correct SYSLINUX_* settings

(From OE-Core rev: 1a3bd2ae195d515fe2c93bc178b45cce8505d831)

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

image-vmdk: Create image_vmdk class and setup image and image_types to use it

This creates a new image_vmdk class similar to live. The image_vmdk
class needs to have a hddimg created by the image-live class, so it
inherits it directly.

The changes to image_types is to ensure that both live and vmdk images
get the ext3 tools and dependencies.

(From OE-Core rev: aa961e112b07d42c272e01f2d69f3c139e9ae70f)

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

libx11-trim/diet: Add RPROVIDES for libx11-dev

We have things that depend on libx11-dev, this patch ensures the -trim
and -diet versions provide it. This resolves some multiple providers
warnings.

(From OE-Core rev: c051d6c59c71a5f90c2d545491facd2d131592fd)

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

image-live: Allow LABELS to be modified

This is to allow other image types to set the syslinux labels

(From OE-Core rev: 825e5a552bbaa215c55da4425e78df3c2f1cddaf)

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

linux-libc-headers-yocto: bump to v3.2.8

Making two changes for BSPs that prefer linux-libc-headers-yocto:

  - bumping to v3.2.8 headers
  - stubbing out unecessary linux-yocto functions when headers
    are being built

[YOCTO #2032]

Acked-by: Darren Hart <dvhart@linux.intel.com>
(From OE-Core rev: 0889f693916bb8f77003af419f71212201b4d9c9)

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

bitbake/build: report TaskFailedSilent events as "Failed"

The change for setscene events to fire a TaskFailedSilent event instead
of TaskFailed resulted in "FailedSilent" being reported in the task
finish note log entry, which is not really desirable, so change it back
to reporting "Failed" again.

(Bitbake rev: 224bc74d4e901b7886b845fbb3b5fe7564a2f6cb)

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

bitbake/uihelper: restore line to record failed normal tasks

Save failed real (non-setscene) tasks to uihelper's failed task list.
as before commit e8a3499c95a6d4f2b8fed002fb9504733c5be3c6. Currently
this list is only used by the ncurses UI.

(Bitbake rev: cc74cad0742ea0d4e09e843883cdc55bad39b22e)

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

bitbake/runqueue: fix python error on scenequeue task failure

Fixes a regression introduced in commit
e8a3499c95a6d4f2b8fed002fb9504733c5be3c6 which resulted in a
backtrace on setscene task failure due to trying to dereference
the setscene task ID twice.

(Bitbake rev: 8b846a92a58b5c20d7cfd2efd32b763e95c3c2fd)

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

runqueue.py: initialize rqexe at RunQueue's init function

Sometimes user stops the build before runqueue is established,
for example, at the stage of running add_unresolved() function. This
will cause RunQueue to use rqexe field in finish_runqeue() before
initialized. This will cause endless print of "Running idle function"
if use process server.

This commit initialize rqexe variable in RunQueue's init function,
and add a judgement in finish_runqueue().

(Bitbake rev: 59f817723172092a87738c79f555e605f55ea375)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: Fix pressing "stop" build

This commit fixes the behavior of clicking stop build button, it will
stay in the build detail screen and show build is failed.

(Bitbake rev: 80291865fa15012a3734e8724eb73c62b4ddc62f)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob(crumbs/builder.py): adjust the main window's position and ignore the case lower than 1024x768.

* don't set the position of the main window to the center always
* don't show the message to users if their screen dimensions are lower than 1024x768

(Bitbake rev: 1afa500cb1cb5c10fc0a3ea0f65c7ecc8887efa8)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: fix a bug that the image size is shown incorrectly in the image details page.

Originally, the image size shows the last item in the image tree view in the image details page.
That is not correct. We need to show the size of the image which the user chooses.

(Bitbake rev: 01c18a24252b35959a4cc01088678f93cb2f95e5)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob(crumbs/hoblistmodel.py): Add mapping for rprovides and pkg

When checking the package's RDEPENDS information, we will see some case like
A RDEPENDS virtual-b, and B RPROVIDES virtual-b, we need to reflect this relationship
in packaging selection.

(Bitbake rev: 44562593556e67d7976a124d5a420938aff95e0c)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob(crumbs/hoblistmodel.py): Fix recipe's populate() function

According to the cooker's change, this commit fixes the way to getting
recipe's build dependency.

(Bitbake rev: da64e59f2e738d6103605139ba2d3e2cdaa35b11)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cooker: fix calculating of depends and rdepends

Remove part of the original logic that stores pkg dependency and pkg
recomends, which will not be used in Hob.

Add the judgement for preferred fn provider of a certain package,
which maps package name to its providing recipe name.

The above approaches correct the build dependency calculation,
and they also reduced the total depend_tree size, which speed up the
tree data generation time.

(Bitbake rev: d668eb1300b3b3115964e98127b1bef554caae17)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cooker.py: Fix a bug due to variable name

fnid is an outer variable and shouldn't be overrided, change the inner variable
name from 'fnid' to 'id'.

(Bitbake rev: c6dc8d6310af22c95adb06c64339e7ec9eaeb315)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

command.py: remove the resolve parameter in generateTargetsTree

Remove the "resolve" parameter since the original resolve=False
option is no longer be used.

(Bitbake rev: dadce609149cfb09ecdc53bfe1f416a3f57a5033)

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob(crumbs/builder.py): remove the code commented out in the builder.py

(Bitbake rev: 7af747778121aa399f69aff58425ede523b2495c)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

crumbs/hig: tweak UI and layout of LayerSelectionDialog

Several minor tweaks for appearance:
* Try to rework the labels so that the English flows better
* Fix spacing
* Remove the separator - it's not needed

(Bitbake rev: 53c5807c38e97d2e44a6f5c48449178b8b6e6261)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

hig: try to avoid setting explicit dialogue sizes

Gtk+ does a good job of creating dialogues which fit all of the packed
widgets and respect the spacing. Let it do its job.

(Bitbake rev: 2469784b38f21716c09df89323c78cd20d3a4c14)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

crumbs: fix button order in several dialogues

GNOME HIG and Gtk+ convention is for the buttons to be ordered
<secondary action> <primary action> so that the primary action can be
selected easily by navigating to the bottom right of the dialogue.

We should try and match the convention and the HIG standard so that we
aren't contrasting the rest of the users applications.

(Bitbake rev: e90828680e3bc655db54ed2797cad4587e8796d0)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

crumbs: move towards more standard dialogue spacing

Inherit from the base CrumbsDialog class which implements standard border and spacing.

Switch all explicitly set dialogue spacings to 6 to converge towards GNOME HIG.

(Bitbake rev: bf938987a007c94fc4bbacb2b4741b7c18cb62ec)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

crumbs/builder: use the name Hob consistently

We have mentions of HOB and Hob, the approved name is Hob.

"the name's Hob, BitBake Hob"
   - Surly BitBake GUI

(Bitbake rev: 06adabbb36472625c1e47991e418346ef7438577)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

crumbs: Factor common dialogue configuration out

Factor out the base dialogue configuration into CrumbsDialog and create a new CrumbsMessageDialog which acts as the CrumbsDialog did to handle dialogues with buttons.

Adopt CrumbsMessageDialog wherever CrumbsDialog was used.

(Bitbake rev: d8c62f54d9f2421cfe7f1083b9d0acfe0373d38a)

Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: enable indicators on the "Included" tab in the recipe selection page and the package selection page

(Bitbake rev: 504d480b208cacad7a5595312890f49fe19b80be)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: make the image configuration page not to flash the progress bar

Sometimes the image configuration page flashes the progress bar, that is because we show_all() and then hide() the progress bar. For this case, the patch doesn't add the progress bar onto the gtable. Then, it will not be flashed any more.

(Bitbake rev: ef472710589580b9bfc64d2c02fa42f3ecbdfeb1)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: fix a bug in builder.configuration.curr_mach

When the user changes the machine to be "--select a machine--", the builder.configuration.curr_mach should be set to empty.
Otherwise, if the user adds more layers after the above, the action will trigger recipe parsing, which is not correct.

(Bitbake rev: d5c7c9471b8e101ebcb91d707415b9c820b1419f)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: image configuration page doesn't need dialogs any more

The dialog references are all in builder.py. We remove the useless "import" in image configuration page.

(Bitbake rev: 804f35e6864aeae2bb02550d9eb34120bbb60fa0)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: include hddimg and iso into image types.

For image types, so far we don't include hddimg and iso. Then those files can not be deployed because they are filtered out in the image selection dialog.
This patch is to include hddimg and iso (which are "live" in image types).

Again, we have a TODO in the code for the future, that is to retrieve image types from the bitbake server instead of to use the walkaround.

(Bitbake rev: d565507940be73fb5ea3ae7048d8d143c44c2a95)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: cleanup those class methods in HobWidget

Some dialogs like advanced settings dialog, and layer selection dialog are using the class methods in HobWidget to create widgets for themselves, which is not a good design for OO.

Clean up the code, and split the functions into the separate classes which use them actually.
Finally, remove the class HobWidget.

(Bitbake rev: f9cccea4d1c52ae2173fd94d5b07ceba7e5c0851)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: fixed the issue in the brought-in-by dialog.

When typing any thing in the search entry, the brough-in-by dialog will be shown. That is because we call back "selection-changed" signal to pop up the dialog, which is not correct.
This patch is to fix the problem by using "row-activated" signal.

(Bitbake rev: ea56ae787153460166697bbcae92f51a77ca1571)

Signed-off-by: Liming An <limingx.l.an@intel.com>
Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: avoid the image selection dialog to walk through all directories and its sub-directories, when users click "My images".

The current implementation of the image selection dialog walks through all directories and its sub-directories, when users click "My images" to choose a directory. If the directory is /, the system becomes slow. This patch is to avoid walking through all directories but the child directories only, given a directory.

(Bitbake rev: 536fa633b442ff37d43f45cf346ba281d69de496)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Hob: make HobViewTable more general in hob and make the image selection dialog and the image details page reuse it.

This patch is to make the class HobViewTable more general as a tree view in Hob.
Now the recipe selection page and the package selection page are using it.
And we have tree views in the image selection dialog and the image details page, which used the class methods in HobWidget to create the tree views. That is not good in OO.

So, make them reuse HobViewTable to create its instances.

(Bitbake rev: 3c900211e8bc0311542873480d79b347d7449f59)

Signed-off-by: Shane Wang <shane.wang@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image_types_uboot: Update to work after recent image_types changes

(From OE-Core rev: 8054497b393d40a2ed8c802c74a02e92a3001297)

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

image_types: Correctness fixes

* Add a newline to improve the output formatting
* Use set() to turn the list into a set of unique items to prevnt
  the same image code running twice (for e.g. IMAGE_FSTYPES = "tar.gz tar.bz2")
* Support multiple compression extensions such as ".gz.u-boot"
* Fix basetype/type typo and fix multiple image generation

(From OE-Core rev: eacedb4f2afa98dbd2f5ea7a9f52e6ea952a72d2)

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

More quoting fixes

(From OE-Core rev: 17f2d5d70ad323cea4fb149154bc38844661fa3a)

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

parse/ConfHandler: Fix enthusiatic export regexp matching

The export regexp was only meant to catch values like:

export VARIABLENAME

however after the stricter quoting patch was applied, it was also matching
variables like:

export BAR=foo

and setting the export flag on a variable called "BAR=foo". The = character
is an invalid variable name character. This patch tightens up the regexp
match so it only matches the intended character set and only matches variable
names.

(Bitbake rev: 6d1765c2eac8c1958ceb9c81d55d04a9bc961cb1)

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

image_types.bbclass: fix bzip2 and xz compression commands

We need to use -f (force) or the command fails in the image file
already exists.

(From OE-Core rev: 419ddab8266ecfd6da1841d38a451a9fc5be49b0)

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

image_types.bbclass: properly support IMAGE_LINK_NAME as empty

[RP: Remove unneeded len()]
(From OE-Core rev: 45a094372ea9e68888efee45d8e21cf2b7fa2df6)

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

image.bbclass: do not create image manifest link if IMAGE_LINK_NAME is empty

(From OE-Core rev: d0f1ae1f8cf8ef4e5adc24cc6246d3849e51aa98)

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

syslinux: fix packaging of ${PN} and ${PN}-staticdev

 - ${PN} where not being build due ordering;
 - ${PN}-staticdev lacked the ${libdir} files.

(From OE-Core rev: e83b2e25ebf0169371035483095a06b3ca5eb479)

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

busybox: Enable 64 bit shell tests, and disable non-standard date format interpretation.

Without 64 bit shell tests, the timestamp comparison in initscripts' bootmisc.sh throws 'out of range' errors.
With CONFIG_FEATURE_DATE_COMPAT, date reads the YYYYMMddhhmm string as hhmmYYYYMMdd when setting the time from the timestamp in bootmisc.sh.

(From OE-Core rev: e32e23649f4eab0bdc71a854f99b11fee19125ac)

Signed-off-by: James Limbouris <james@digitalmatter.com.au>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

licenses.conf: fix quotting of SRC_DISTRIBUTE_LICENSES

(From OE-Core rev: 61216d2743502ef38955054aad6a4f2ed63e6d43)

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

tcl: Ensure native verison has correct dependencies and builds

(From OE-Core rev: 9b63fbb5f41d8a7167bbb534da03908c996693ad)

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

bootimg: Fix Quoting

(From OE-Core rev: 18ad7e003e36510ff0097d71bad0378a77fabbdd)

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

syslinux: bump PR

The change done in 868a81e869a6193aada2073ae533d937a1c0baf4 has
changed the packaging however it haven't bump the PR making
auto-building fail.

(From OE-Core rev: eaa8c7d7c047c44067c8931edc81fe476b9a36db)

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

bitbake: implement BB_VERBOSE_LOGS

Enable configuring whether "set +x" is added to all shell tasks rather
than forcing it; this is enabled by setting BB_VERBOSE_LOGS to 1.

(Bitbake rev: 659411b6bb30e1a8355afc1c29b8170a8f2b55ac)

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

tcl 8.5.11: fix packaging for native version

Fix an issue in the previous commit:

ERROR: Nothing RPROVIDES 'tcl-native-lib-native' (but virtual:native:/srv/home/pokybuild/yocto-autobuilder/yocto-slave/nightly/build/meta/recipes-devtools/tcltk/tcl_8.5.11.bb RDEPENDS on or otherwise requires it)
NOTE: Runtime target 'tcl-native-lib-native' is unbuildable, removing.

by replacing ${PN}-lib with tcl-lib

(From OE-Core rev: 3cb519598d9ce044ccccb2d6a52a3c0933433501)

Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

More quoting fixes

(From OE-Core rev: 749d252475df090d51313cfbbe3f159db9f0566d)

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

bitbake/ConfHandler: Be more strict about variable quoting

Currently, bitbake will accept variables in the forms:

X = 1
X = '1 \

X = "1"
X = '1'

which will all set X=1. This patch removes the first two possibilities
and makes quoting mandatory. There is little metadata out there which
doesn't quote properly and bitbake will exit with an error about the
exact line number and file with any problem so users can easily identify
and fix issues. OE-Core has already been checked/fixed.

The motivation for this is being able to give sane errors if a user
does something like:

IMAGE_INSTALL += # tslib mtd-utils"

which currently gives a really nasty failure.

(Bitbake rev: a8ae80741fea5e0ec0fb9a52a963a4baa38d2564)

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