siteinfo: add SITECONFIG_SYSROOTCACHE definition in place of hard-coded path
siteconfig: use SITECONFIG_SYSROOTCACHE and use sstate-interceptfuncs

Use the new sstate-interceptfuncs functionality to interpose do_siteconfig
between the sstate_install and package.

Signed-off-by: Jeff Polk <jeff.polk@windriver.com>

sstate: Add sstate-interceptfuncs called between install and package

Adds the ability to impose intercept function call(s) after
sstate_install and before sstate_package in sstate_task_postfunc.

Signed-off-by: Jeff Polk <jeff.polk@windriver.com>

[e]glibc-package.inc: fix providers for libsegfault and other

[e]glibc-nativesdk should not be providing libsegfault.

This commit fixes the following error while doing this:
bitbake poky-image-minimal-initramfs

DEBUG: providers for libsegfault are: ['glibc-nativesdk', 'glibc',
'eglibc-nativesdk', 'external-csl-toolchain', 'eglibc',
'external-poky-toolchain']
DEBUG: selecting
virtual:nativesdk:/rphome/poky/meta/recipes-core/glibc/glibc_2.10.1.bb
as PREFERRED_VERSION 2.10.1 of package glibc-nativesdk (for item
libsegfault)
DEBUG: selecting /rphome/poky/meta/recipes-core/glibc/glibc_2.10.1.bb as
PREFERRED_VERSION 2.10.1 of package glibc (for item libsegfault)
NOTE: checking PREFERRED_PROVIDER_glibc-nativesdk
NOTE: checking PREFERRED_PROVIDER_glibc-nativesdk-2.10.1
NOTE: checking PREFERRED_PROVIDER_glibc-nativesdk-2.10.1-r3
NOTE: checking PREFERRED_PROVIDER_virtual/libc-nativesdk
NOTE: selecting glibc-nativesdk to satisfy runtime libsegfault due to
PREFERRED_PROVIDER_virtual/libc-nativesdk = glibc-nativesdk

And bumped PR.

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

libc-package.bbclass: fix qemu issue with pseudo

This solution is what Mark Hatle recommended.

To disable pseudo, while already running, you need to set:
PSEUDO_RELOADED=YES, and then exec something...

This causes pseudo to disable itself from LD_PRELOAD, and thus fall
out of memory on the exec.

This Fixes [BUGID #226]

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>

pseudo: Update pseudo to latest revision

Fix bug 231 by changeing to the latest revision of pseudo.

The latest revision of pseudo includes the following changes:

2010-09-08:
  * (seebs) handle mkfifo without guaranteeing an EINVAL response.

2010-09-02:
  * (seebs) fix errno for getcwd() with insufficient size
  * (seebs) Add an RPATH entry to the pseudo binary to find the sqlite3
    library.

2010-09-01:
  * (seebs) add missing casts to even more printf arguments

2010-08-31:
  * (seebs) add missing casts to printf arguments, after being warned
    about them a mere twenty or thirty thousand times.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

rootfs_rpm: Fix workaround

The exit 0 in the workaround seems to have causes the filesystem creation
process to stop.  Instead change it to an if that can never succeed.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

handbook: Drop USB Networking section, its no longer appropriate

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

handbook: Various minor fixes

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

handbook: review and update CH1(introduction) and CH2(using Poky)

Signed-off-by: Yu Ke <ke.yu@intel.com>
Signed-off-by: Kevin Tian <kevin.tian@intel.com>

handbook: review and modify Appendix A, D and E

Major change are about the structure of Poky source tree, which has been
changed a lot recently

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Edwin Zhai <edwin.zhai@intel.com>
Signed-off-by: Kevin Tian <kevin.tian@intel.com>

handbook: review and modify CH4 (BSP) and Appendix B

Besides basic corrections, also add .bbappend to bsp introduction
and update bitbake help to match latest output

Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>
Signed-off-by: Kevin Tian <kevin.tian@intel.com>

handbook: Fix some errors in Appendix F/G

CVS_TARBALL_STASH and SHELLRCCMD is not found in current poky.
Fix some grammar issues.

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Qing He <qing.he@intel.com>

handbook: review and correct CH5(development)

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Liping Ke <liping.ke@intel.com>

handbook: update appendix C/H/I

The following is suggested but not changed:
 - adding documentation of cmake.bbclass
 - remove src_distribute_local.bbclass

Signed-off-by: Qing He <qing.he@intel.com>
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>

handbook: review and correct CH3(extend poky) in handbook

Fix inaccurate descriptions.
Update recipe examples to make sure they do build
Add some examples for better guidance

Signed-off-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Ke Yu <ke.yu@intel.com>

linux-wrs/meta: fix qemuarm boot issues with 2.6.34.6

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

qemu: add qemu-mipsel to the list of simulators

Add qemu-mipsel to the list of generated simulators. Obviously
this is needed to build a MIPS little endian system.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

connman-gnome: Add some functionality.

Add offline mode.
Add ethernet/wifi enable/disable functionality.
Classify each service (ethernet, wifi) into certain group.
Do some code clean up.

Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>

bitbake: Add proxy variables to standard export list

With "no_proxy" exported to bitbake, both internal and externel file mirror can
be used. "https_proxy" enable fetching "https://" file through proxy.

Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>

Fix the matchbox-panel seg fault on netbook & emenlow

netbook & emenlow use "-march=core2"(in tune-atom.inc) instead of "=i586"(for
qemux86), plus the "-fomit-frame-pointer" in default FULL_OPTIMIZATION, this
cause strange seg fault when starting matchbox-panel.

seg fault happened @ tmp_reset_bg (gtk+-2.20.1/gdk/x11/gdkwindow-x11.c), which
is inline function extended inside _gdk_x11_window_tmp_unset_bg. When expanding
GDK_DRAWABLE_XID(obj->bg_pixmap) with gcc 4.5.0 optimization, a function call
is missing. Saving a local var to stack also disappeared, but still tried to
restore it from stack, which cause seg fault after getting a worng value.

This fix avoid "-fomit-frame-pointer" in gtk+ on netbook & emenlow.

[BUGID #224] fixed by this.

Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>

netbase: Don't hardcode the qemu IP config

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

scripts/runqemu: Drop broken toolchain symlink code

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

scripts/poky-qemu-ifup: Ensure the host has a route to it

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

scrtips/poky-qemu: Fix network interface device numbering

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

poky-qemu-internal: Fix ifdown script to use the correct variable

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

gnu-config: Fix target version to contain all the substitutions and not overwrite the first half with the second set of changes

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

libzypp: change /usr/lib64 to /usr/lib

similar to previous sat-solver fix, we use /usr/lib for x86-64 target

Signed-off-by: Kevin Tian <kevin.tian@intel.com>

linux-wrs: Update now SRCREV variable order is fixed

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

linux-wrs: validate branches against known revisions

It is possible for the git repository to have commits on branches
that are ahead of the SRCREVs for those branches. Since the
kern tools will simply checkout and build, we can end up building
a different kernel than what is expected.

It is up to the recipe to ensure that the branches are at the
correct commits. To do this, we simply check the current machine
branch against the SRCREV and if they don't match we move the
existing branch and do a checkout at the desired commit.

Note: This probably needs to be disabled with AUTOREV.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

poky-default-revisions.inc: Fix variable order for linux-wrs recipes

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake/__init__.py: Fix named SRCREVS variable name ordering with backwards compatibility

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

trace-cmd: add to profiling recommends

Make trace-cmd available by adding it to RECOMMENDS_task-poky-tools-profile

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

task-poky-lsb: Remove man-pages due to build failures

Signed-off-by: Saul Wold <Saul.Wold@intel.com>

task-poky-lsb: add man-pages, which will pull man and groff

Signed-off-by: Saul Wold <Saul.Wold@intel.com>

image.bbclass: change IMAGE_EXTRA_SPACE to IMAGE_OVERHEAD_FACTOR

This change allows the IMAGE_OVERHEAD_FACTOR (default to 1.2) to
adjust dynamicly how much space to add for each image, thus there
is 20% overhead space by default.

Signed-off-by: Saul Wold <Saul.Wold@intel.com>

man-pages:Add new man-pages package

The Linux man-pages project documents the Linux kernel and C library interfaces that are employed by user programs

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>

groff:Add new groff package

groff allows for a short reference for the GNU roff language

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>

man: Add new man package

A set of documentation tools: man, apropos and whatis

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>

Distro tracking: add package aliases

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>

at: Disable PARALLEL_MAKE for at and add back to task-basic

Signed-off-by: Saul Wold <Saul.Wold@intel.com>

gcc_4.5.0: add --enable-poison-system-directories option to cross
recipes

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>

binutils: add binutils_poison.patch

and also correct the license field in the recipe

This fixes [BUGID #245]

Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>

bitbake: save the initial PATH for use when starting a subprocess

It was possible for bitbake-runtime to be run against a semi-installed
python-native resulting in tracebacks with ImportError's.

To prevent this we stash the initial PATH in the BBConfiguration when bitbake
is started and then set this in the env when launching bitbake-runtask through
subprocesses Popen() call.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

kernel.bbclass/linux-wrs: move definition of perf package to kernel class

By defining the perf package in the shared kernel class bitbake will be able to
find the package provider even when the preferred kernel doesn't build perf,
preventing dependency resolution failures.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

package.bbclass: Ensure all packages package when running a build

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake/fetch/git: fix try_premirror() definition .Method takes 4 parameters.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

bitbake/fetch: fix by moving try_premirror() to the right place

The previous commit didn't define the method as part of the Fetch object

Signed-off-by: Joshua Lock <josh@linux.intel.com>

powertop: fix a typo: REPENDS ==> RDEPENDS

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>

libzypp: workaround to pass mips build

libzypp failed due to GCC "internal compiler error: segmentation fault" error on
MIPS target. The cause is about boost tribool type usage. If having assignement
within conditional check as below:

if ( (a = b) )
...
<a is a tribool type, b is a normal bool type>

then gcc4.3.3 throws internal error. Then the workaround is to move assignement
out of the conditional check.

However I didn't find same case from web.

We can come back to recheck this issue after upgrading to gcc4.5.0 for MIPS. If
this issue is still there, we'll need more analysis to decide whether to report
to gcc upstream or to libzypp upstream.

This fixes [BUGID #277]

Signed-off-by: Kevin Tian <kevin.tian@intel.com>

sat-solver: change /usr/lib64 to /usr/lib

we use /usr/lib for x86-64 target, but sat-solver will use /usr/lib64 if 64bit
target arch is detected. This finally causes libzypp failed to find libsatsolver.a.
use "-DLIB=lib" to explicitly enforce the requirement

fix [BUGID #286]

Signed-off-by: Kevin Tian <kevin.tian@intel.com>

bitbake/fetch: add try_premirror method and special case git fetcher

Add a new method, try_premirror, to the Fetch object which checks to see
whether the file needs to be fetched from a premirror.
Override this in the Git fetcher to only require a pre-mirror fetch when the
clone directory does not exist.

Fixes [BUGID 290]

Signed-off-by: Joshua Lock <josh@linux.intel.com>

bitbake/bitbake-runtask: Ensure signals to the parent don't pass to the children directly

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

qemu: Apply fix for armv6 locale generation using TLS registers

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

ldconfig-native: Set LICENSE

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

poky-image: Set LICENSE field for the image recipes (note this doesn't apply to the image contents, just the recipe and code used

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

poky-image: Set LICENSE field for the image recipes (note this doesn't apply to the image contents, just the recipe and code used

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

tasks: Set LICENSE field

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

QT4 mips: disable the LDFLAGS "-Bsymbolic-functions"

"-Bsymbolic-functions" flag makes ld to bind the global function symbol of shared library at compiling time, thus reduce the runtime relocation cost.

However, this flag in binutil 2.20 cause QT4 mips building failure, which is "final link failed: Bad value" error in libQtWebKit.so linking stage. If downgrad to binutil 2.19, this error disappears.

so before binutil new release is out, it is better to disable this flag for QT4 mips.

Fix [BUGID #271]

Signed-off-by: Yu Ke <ke.yu@intel.com>

linux: Fix the compulab image post deploy function by making it a postfunc

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

sstate: Make sure the sstate function is the last to run, not the first

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

pseudo: Honour the NO32LIBS variable

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake/codeparser: Deal with functions with trailing whitespace

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

udev: Drop old versions

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

poky-default.inc: Use standard udev version for PPC, seems to work fine

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

scripts: use the exported POKY_NATIVE_SYSROOT variable

Rather than trying to determine things through guess-work use the newly
exported variables to determine where the native binaries reside and
whether we are running in a build directory or not.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

toolchain-scripts: export extra variables for use by scripts

Export the location of the native sysroot for use in the scripts to
determine the location of native binaries.

Also add a POKY_DISTRO_VERSION to the external toolchain environment as a
method of determining if we're running under a build directory or not
i.e. not if the variable is empty

Signed-off-by: Joshua Lock <josh@linux.intel.com>

runqemu-nfs: check for required programs

The unfs server requires either rpcbind or portmap to be installed and
running to start so check for their presence in the script.

Signed-off-by: Joshua Lock <josh@linux.intel.com>

linux-wrs: Ensure S directory exists else task can fail

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

gcc-crosssdk-initial: Ensure native dependencies are built, these can differ accross different gcc versions causing build failures

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

kernel/wrs: update SRCREV for v2.6.34.6 + changes

Updating the SRCREVs for the recent port of bug fixes for all boards
and the integration of qemux86-64 config and connector changes.

Now that the qemux86-64 changes are in tree, we can drop them from
the linux-wrs recipe.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>

task-poky-tools.bb: avoid multiple RRECOMMENDS to override each other

there're two RRECOMMENDS_task-poky-tools-profile both using direct assignment,
which makes 'perf' overriden by "kernel-module-oprofile"

Fix [BUGID #97]

Signed-off-by: Kevin Tian <kevin.tian@intel.com>

poky-default-revisions.inc: Move SRCREVs to correct locations, fix whitespace

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

fetch: fix setting of localpath in SRC_URI parameters

When setting localpath in the SRC_URI parameters the basename should be set
explicitly as the localpath() method is not called.

Fixes [BUGID #81]

Signed-off-by: Joshua Lock <josh@linux.intel.com>

poky-default.inc: Make sure sdk version of gcc is fixed and doesn't change

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

tasks-poky-basic and tasks-poky-qt: bump PR and remove at

Remove at since it's failing build
Add Zypper to lsb image

Signed-off-by: Saul Wold <Saul.Wold@intel.com>

task-poky-qt.bb: Exclude mips from building

Signed-off-by: Saul Wold <Saul.Wold@intel.com>

clutter-mozembed: move patches back into meta-moblin

These patches were mistakenly moved into meta/recipes-graphics/clutter
during the metadata reorg. This moves them back where they belong.

Signed-off-by: Scott Garman <scott.a.garman@intel.com>

rpm: Fix autoconf/libtool usage

In order to resolve a host-contamination problem, we re-work the way that
autoconf and friends are invoked during the compilation of RPM.

This has a side effect of fixing another bug where RPM was being renamed
HOST_ARCH-HOST_OS-rpm.  So we remove the "fixes" for that behavior as well.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

rootfs_rpm: Workaround for checksum changes

With the new checksum changes, the system is no longer able to determine
if the function resolve_package is used.  So it doesn't get emitted.  The
workaround resolves this by triggering an exit 0, followed by a bogus usage
of the function.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

zypper: add new package

Signed-off-by: Saul Wold <saul.wold@intel.com>
Signed-off-by: Qing He <qing.he@intel.com>

libzypp: add new package

Signed-off-by: Saul Wold <saul.wold@intel.com>
Signed-off-by: Qing He <qing.he@intel.com>

sat-solver: added new package

Signed-off-by: Saul Wold <saul.wold@intel.com>
Signed-off-by: Qing He <qing.he@intel.com>

augeas: add new package

version 0.7.3

Signed-off-by: Qing He <qing.he@intel.com>

Fix broken references to libcheck

* check has been renamed to libcheck

Signed-off-by: Scott Garman <scott.a.garman@intel.com>

rpm: Change the way the python version is determined so its sane and works with checksumming

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

task-poky-qt: create a task and add to poky-image-sdk

Fixes [BUGID #260]

Signed-off-by: Saul Wold <Saul.Wold@intel.com>

bootmisc.h: use "date -s" to set time

without "-s", 201009031653 would be interpreted incorrectly by date
and then we saw below warning:
date: invalid date 165320100903

Fix [BUGID #265]

Signed-off-by: Kevin Tian <kevin.tian@intel.com>

distro-tracking: update lttng-ust, lttng-viewer

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>

lttng-viewer: upgrade to the latest version 0.12.35

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>

lttng-ust: upgrade to the latest version 0.7

Signed-off-by: Dexuan Cui <dexuan.cui@intel.com>

poky.conf: Set hash policy

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: Add missing signature files

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake: Implement signatures

Includes functionality to find out what changes between two different singature data dumps.

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake/codeparser: Ensure cached sheel entries return the correct dependencies

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

bitbake/ast: Ensure function definitions are always added to the dictonary

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

sstate: Handle symlink directory entries correctly

Signed-off-by: Richard Purdie <rpurdie@linux.intel.com>

rpm: Raise the number of potential rescans before aborting with a dep loop

Within Poky we have a large number of self-referencing packages.  So we
need to raise the number of rescans before triggering a dependency loop
error.  100 seems like a reasonable number.

Signed-off-by: Mark Hatle <mark.hatle@windriver.com>

eglibc: Add siteconfig cache generation

Signed-off-by: Jeff Polk <jeff.polk@windriver.com>

uclibc: Add siteconfig cache generation

Signed-off-by: Jeff Polk <jeff.polk@windriver.com>

ncurses: Add siteconfig cache generation

Signed-off-by: Jeff Polk <jeff.polk@windriver.com>