Paul Eggleton [Tue, 12 Jan 2016 21:39:03 +0000 (10:39 +1300)]
classes/populate_sdk_ext: avoid unnecessary sstate being brought in
Create a separate task where we can just have the recursive dependencies
for the tasks corresponding to constructing the SDK content (i.e. from
the image contents). This avoids us recursing into dependencies from
buildtools and getting a bunch of nativesdk stuff, for example.
(This isn't an ideal way to have to implement it, but without
overcomplicating things on the BitBake side just for this use-case I
can't see a better way.)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Tue, 12 Jan 2016 18:00:13 +0000 (18:00 +0000)]
insane/package: Fix cases where QA errors aren't fatal
When using package_qa_handle_error(), we need to check QA_SANE and
give a fatal error if issues were raised. The main insane checks do
this, the anonymous python in insane does not, nor does the code in
package.bbclass. This was likely missed when the function was
introduced to package.bbclass.
The impact is that errors can be shown but don't stop the build
(do_package succeeds), the initial build fails due to the errors
shown but subsequent builds succeed. This is clearly broken.
The solution is to check QA_SANE in places we use the handle_error
function.
[YOCTO #8907]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Tue, 12 Jan 2016 18:47:47 +0000 (07:47 +1300)]
classes/populate_sdk_ext: check that extensible SDK prepared correctly
After the change to use --setscene-only when running bitbake to prepare
the SDK at the end of installation, add a check that the SDK got
prepared correctly by doing a dry-run and looking at the output for any
real tasks that we don't expect. In order to make this easier, the
preparation shell script was rewritten in python.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Tue, 12 Jan 2016 18:47:45 +0000 (07:47 +1300)]
classes/populate_sdk_ext: support auto.conf
If auto.conf exists in the user's configuration we need to also run it
through the same filter and write the result into the ext SDK, or we
risk missing configuration applied on an autobuilder.
Fixes [YOCTO #8904].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Tue, 12 Jan 2016 18:47:44 +0000 (07:47 +1300)]
classes/populate_sdk_ext.bbclass: handle if local.conf doesn't end with a newline
If there is no newline at the end of local.conf, appending
INHERIT = "uninative" won't work, it will corrupt the line and the
installed eSDK will build things, making the "Preparing build system..."
step take an age.
Fixes [YOCTO #8897].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Hongxu Jia [Mon, 11 Jan 2016 00:30:39 +0000 (08:30 +0800)]
gnupg: 2.1.7 -> 2.1.10
Add PACKAGECONFIG for sqlite3 to avoid QA Warning
...
WARNING: QA Issue: gnupg rdepends on libsqlite3, but it isn't a build dependency? [build-deps]
...
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Mark Hatle [Mon, 11 Jan 2016 21:12:05 +0000 (15:12 -0600)]
image.bbclass: Unconditional includes of populate_sdk_ext fails
populate_sdk_ext requires uninative support, which is only available on
glibc based SDKMACHINES. For instance, when using mingw32 a dependency
error will occur:
NOTE: Runtime target 'nativesdk-glibc' is unbuildable, removing...
ERROR: Required build target 'core-image-minimal' has no buildable providers.
Missing or unbuildable dependency chain was: ['core-image-minimal', 'uninative-tarball', 'nativesdk-glibc']
This is dues to populate_sdk_ext.bbclass having:
do_populate_sdk_ext[depends] += "buildtools-tarball:do_populate_sdk uninative-tarball:do_populate_sdk"
addtask populate_sdk_ext
Since bitbake can't determine for dependency resolution if the task is going
to be run yet, it blows up and says it simply can't be resolved.
Workaround this problem by making the inherit conditional on the SDK_OS
containing 'linux'.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Similar to the libc-for-gcc preferred provider, we also need a libc-initial
version. Layers such as meta-mingw need the ability to override these
values in order to generate an SDK that works on non Linux environments.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Awais Belal [Tue, 12 Jan 2016 09:55:03 +0000 (14:55 +0500)]
dhcp: search libxml2 for bind
libdns requires libxml2 if bind was built with libxml2 support
enabled. Compilation will fail for omapip/test.c in case
-lxml2 isn't used during the build. So, we add losely coupled
search path which will pick up the lib if it is present.
Signed-off-by: Awais Belal <awais_belal@mentor.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Andre McCurdy [Tue, 12 Jan 2016 07:47:16 +0000 (23:47 -0800)]
harfbuzz: update 1.1.2 -> 1.1.3
Misc recipe formatting cleanup, plus version update as described below.
http://cgit.freedesktop.org/harfbuzz/tree/NEWS
Overview of changes leading to 1.1.3
Monday, January 11, 2016
====================================
- Ported Indic shaper to Unicode 8.0 data.
- Universal Shaping Engine fixes.
- Speed up CoreText shaper when font fallback happens in CoreText.
- Documentation improvements, thanks to Khaled Hosny.
- Very rough directwrite shaper for testing, thanks to Ebrahim Byagowi.
- Misc bug fixes.
- New API:
Ross Burton [Fri, 8 Jan 2016 20:44:15 +0000 (20:44 +0000)]
update_font_cache: only scan system font directories
By default fc-cache will scan both system and user directories, which means it
attempts to scan $HOME/.fonts. As this is the build host's idea of $HOME this
generally doesn't exist, and causes fc-cache to exit with a failure.
Solve this by passing --system-only so that fc-cache will only scan system
directories, as is appropriate for a rootfs-time invocation.
Signed-off-by: Ross Burton <ross.burton@intel.com>
* at least with git 2.6.3 I see git request-pull failing
when there is only :{BRANCH} as ending commit
* $ git request-pull origin/dizzy git://git.openembedded.org/openembedded-core-contrib jansa/dizzy-backports:jansa/dizzy-backports
The following changes since commit 7bb182bdd130266100fc541fd09b82d09c51cd80:
build-appliance-image: Update to dizzy head revision (2015-09-29 14:56:04 +0100)
...
And finds correct 7 changes there
* $ git request-pull origin/dizzy git://git.openembedded.org/openembedded-core-contrib :jansa/dizzy-backports
warn: No match for commit 6068d1c90336ddc1fb32856efd1d9ccf07733896 found at git://git.openembedded.org/openembedded-core-contrib
warn: Are you sure you pushed 'jansa/dizzy-backports' there?
The following changes since commit 97756472d3a69eaca95d105494ffea78c6b077e0:
build-appliance-image: Update to dizzy head revision (2014-10-18 16:16:27 +0200)
...
and lists all commits in _current_ branch since origin/dizzy, then it refuses
to continue, because there are too many changes.
* 6068d1c90336ddc1fb32856efd1d9ccf07733896 is this commit in jansa/master-submitted
branch so it really shouldn't be included in pull request from jansa/dizzy branch.
* git help says:
<end>
Commit to end at (defaults to HEAD). This names the commit at the tip of the history you are asking to be pulled.
When the repository named by <url> has the commit at a tip of a ref that is different from the ref you have locally, you can use
the <local>:<remote> syntax, to have its local name, a colon :, and its remote name.
* maybe the syntax got changes since git 2.1.0 when Saul added :${BRANCH}
* I haven't found how to respect ${COMMIT_ID in the new syntax
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Split runuser into its own package (previously provided by util-linux).
runuser is compiled only when DISTRO_FEATURES includes pam and also the
package is created only when runuser exists.
Signed-off-by: Ioan-Adrian Ratiu <adrian.ratiu@ni.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
LICENSE checksums are changed due to a change in FSF address, or copyright
years update.
Added patches:
0001-Remove-tests-that-fail-to-build-on-some-PPC32-config.patch replaces remove-ppc-tests-failing-build.patch
and removes only those tests that are known to break the build on ppc32 configurations tested by
poky autobuilders
Rebased patches:
sepbuildfix.patch rebased to 0004-Fix-out-of-tree-builds.patch
add-ptest.patch rebased to 0005-Modify-vg_test-wrapper-to-support-PTEST-formats.patch
Removed patches:
force-nostabs.patch removed because it's patching lines that have been removed upstream
remove-ppc-tests-failing-build.patch replaced with 0001-Remove-tests-that-fail-to-build-on-some-PPC32-config.patch
enable.building.on.4.x.kernel.patch removed because the problem is fixed upstream
glibc.patch removed for the same reason
AnÃbal Limón [Wed, 6 Jan 2016 12:21:36 +0000 (04:21 -0800)]
DpkgRootfs: Fix logcheck_error false-positive when use multilib
Rootfs with dpkg was failing due to false-positive in logcheck_error
because current logic of DpkgPM handles missing dependencies failure
using apt-get -f install [1][2].
This support was broken due to addition of logcheck and don't take into
account dpkgpm cases, in order to fix add an attr for specify expected
errors regex'es by package manager.
Signed-off-by: AnÃbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Matt Madison <matt@madison.systems> Signed-off-by: Ross Burton <ross.burton@intel.com>
Matt Madison [Wed, 6 Jan 2016 12:21:35 +0000 (04:21 -0800)]
package_deb.bbclass: add 'Multi-Arch: foreign' tag to allarch packages
This tells APT that it can use such packages to resolve dependencies
from packages of any architecture in a multilib build.
Signed-off-by: Matt Madison <matt@madison.systems> Signed-off-by: AnÃbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Matt Madison [Wed, 6 Jan 2016 12:21:34 +0000 (04:21 -0800)]
package_manager.py: fixes for multilib deb packaging builds
* tmp/deploy/deb subdirectories do not get hyphens replaced
with underscores, so don't do that translation when building
the sources list.
* Fix MULTILIB_VARIANTS handling to be more general and
work for all architectures
* Also include a fix for a warning generated by apt
due to missing apt/preferences.d directory.
Signed-off-by: Matt Madison <matt@madison.systems> Signed-off-by: AnÃbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Matt Madison [Wed, 6 Jan 2016 12:21:33 +0000 (04:21 -0800)]
package_deb.bbclass, cross-canadian.bbclass: DPKG_ARCH mapping function
Have DPKG_ARCH set by directly invoking a mapping function, rather
than using an anonymous Python function modify the variable under
the hood, so we can have proper handling of overrides.
Also bring in some additional mappings to Debian architecture names
that weren't being handled.
Signed-off-by: Matt Madison <matt@madison.systems> Signed-off-by: AnÃbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Ed Bartosh [Thu, 7 Jan 2016 11:39:39 +0000 (13:39 +0200)]
meta: rename perl-native-runtime
The code in native.bbclass adds -native suffix to the package
names that don't have it. perl-native-runtime becomes
perl-native-runtime-native because of this.
Renamed perl-native-runtime -> hostperl-runtime-native to avoid
mangling it and to conform with the naming convetion for native
packages.
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Hongxu Jia [Mon, 4 Jan 2016 02:08:42 +0000 (10:08 +0800)]
dbus: support large-file for stat64
While starting dbus-daemon on a 32-bit linux host and it invokes
fstat to load /etc/dbus-1/system.conf through NFS. If system.conf
was created with a large indoe number on 64-bit host. The above
fstat invoking failed. Here is the log of strace:
............
$ ls -i /etc/dbus-1/system.conf 53778558109 /etc/dbus-1/system.conf
$ strace /usr/bin/dbus-daemon --system --address=systemd: --nofork --nopidfile --systemd-activation
|open("/etc/dbus-1/system.conf", O_RDONLY) = 4
|fstat64(4, {st_mode=S_IFREG|0644, st_size=3340, ...}) = 0
|close(4) = 0
|close(3) = 0
|write(2, "Failed to start message bus: Fai"..., 109Failed to start message bus:
Failed to stat "/etc/dbus-1/system.conf": Value too large for defined data type
|) = 109
|exit_group(1) = ?
|+++ exited with 1 +++
............
In this situation, we should support large-file for stat64. Add marco
AC_SYS_LARGEFILE to do the detection at configure time. It can be disabled
by configuring with the `--disable-largefile' option.
[YOCTO: #8863]
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Ross Burton [Thu, 7 Jan 2016 21:13:07 +0000 (21:13 +0000)]
freetype: enable out-of-tree builds, and use host zlib
Add a few ${S} and ${B} to make out of tree builds work, and stop using
autotools-brokensep. Annoyingly we still need to use a custom do_configure so
add a comment explaining why so someone else doesn't spend 30 minutes trying to
make it work.
Whilst here add a small patch so we don't need to tell the build where libtool
is, and remove class-native do_configure as it doesn't appear to be required
anymore.
At this point I started to get carried away.
The do_compile_prepend is redundant now that configure is being told what
compiler to use for build tools, so remove that.
Instead of using the integrated zlib fork, add a PACKAGECONFIG to use the zlib
we build and enable that by default. Also add a disabled PACKAGECONFIG for
bzip2 support.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Robert Yang [Thu, 7 Jan 2016 07:45:22 +0000 (23:45 -0800)]
cogl-1.0: fix may be used uninitialized error
Fixed when DEBUG_BUILD = "1":
test-backface-culling.c:206:7: error: 'cull_front' may be used uninitialized in this function [-Werror=maybe-uninitialized]
| validate_part (framebuffer,
| ^
| cc1: some warnings being treated as errors
Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Lukas Bulwahn [Wed, 6 Jan 2016 17:31:01 +0000 (18:31 +0100)]
boost: update to 1.60.0
Due to the version update to 1.59.0, the two patches previously
backported in this recipe are now dropped. The actual status of
the arm-intrinsics.patch (reported with Upstream-Status: Backport)
is unknown to the committer, and hence, that patch is kept in the
recipe, even if it is possibly not required anymore. A deeper
analysis is required to determine need and status of that patch.
Second, boost 1.59.0 provides a new library, called timer,
which is packaged as further library in the boost recipe. The
update to boost 1.60.0 required no further changes beyond
changing the checksum hashes.
Third, on PPC architectures, qemuppc & mpc8315e-rdb [1] (tested
by Ross Burton), the boost test library with the vanilla
version 1.60.0 fails with:
| gcc.compile.c++ bin.v2/libs/test/build/aca09349fdb84d131321425f6c3a38ed/execution_monitor.o
| In file included from [...]/tmp/sysroots/mpc8315e-rdb/usr/include/fenv.h:114:0,
| from [...]/tmp/sysroots/mpc8315e-rdb/usr/include/c++/5.3.0/fenv.h:36,
| from ./boost/detail/fenv.hpp:97,
| from ./boost/test/execution_monitor.hpp:64,
| from ./boost/test/impl/execution_monitor.ipp:31,
| from libs/test/src/execution_monitor.cpp:16:
| ./boost/test/impl/execution_monitor.ipp: In function 'unsigned int boost::fpe::enable(unsigned int)':
| ./boost/test/impl/execution_monitor.ipp:1383:7: error: expected id-expression before '(' token
| ::feclearexcept(BOOST_FPE_ALL);
| ^
| ./boost/test/impl/execution_monitor.ipp: In function 'unsigned int boost::fpe::disable(unsigned int)':
| ./boost/test/impl/execution_monitor.ipp:1420:7: error: expected id-expression before '(' token
| ::feclearexcept(BOOST_FPE_ALL);
| ^
The commit f50e7bc9fed323c5705c0ed992c03e80eeec1dd6 from the
boostorg/test github repository (branch: develop) [2] addresses
that issue, but it was not merged in the 1.60.0 release. This
commit adds this upstream-accepted patch, which was created by:
Marek Vasut [Wed, 6 Jan 2016 02:13:29 +0000 (03:13 +0100)]
gcc5: Fix build on NIOS2
The gcc 5.3 does not build on NIOS2 due to a missing MUSL_DYNAMIC_LINKER
definition in it's config file. Add the definition to fix the build issue.
The output produced during the failing build is as follows:
g++ -isystem/b/tmp/sysroots/x86_64-linux/usr/include -O2 -pipe -DIN_GCC -DCROSS_DIRECTORY_STRUCTURE -fno-exceptions -fno-rtti -fasynchronous-unwind-tables -W -Wall -Wno-narrowing -Wwrite-strings -Wcast-qual -Wmissing-format-attribute -Woverloaded-virtual -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings -DHAVE_CONFIG_H -static-libstdc++ -static-libgcc -L/b/tmp/sysroots/x86_64-linux/usr/lib -L/b/tmp/sysroots/x86_64-linux/lib -Wl,-rpath-link,/b/tmp/sysroots/x86_64-linux/usr/lib -Wl,-rpath-link,/b/tmp/sysroots/x86_64-linux/lib -Wl,-rpath,/b/tmp/sysroots/x86_64-linux/usr/lib -Wl,-rpath,/b/tmp/sysroots/x86_64-linux/lib -Wl,-O1 gcc-ar.o -o gcc-ar \
file-find.o libcommon.a ../libcpp/libcpp.a ../libbacktrace/.libs/libbacktrace.a ../libiberty/libiberty.a ../libdecnumber/libdecnumber.a
In file included from ./tm.h:27:0,
from /b/tmp/work-shared/gcc-5.3.0-r0/gcc-5.3.0/gcc/gcc.c:34:
./config/linux.h:92:28: error: expected ',' or ';' before 'MUSL_DYNAMIC_LINKER'
BIONIC_DYNAMIC_LINKER, MUSL_DYNAMIC_LINKER)
^
./config/linux.h:59:60: note: in definition of macro 'CHOOSE_DYNAMIC_LINKER1'
"%{" LIBC2 ":" LD2 ";:%{" LIBC3 ":" LD3 ";:%{" LIBC4 ":" LD4 ";:" LD1 "}}}"
^
./config/linux.h:91:3: note: in expansion of macro 'CHOOSE_DYNAMIC_LINKER'
CHOOSE_DYNAMIC_LINKER (GLIBC_DYNAMIC_LINKER, UCLIBC_DYNAMIC_LINKER, \
^
./config/nios2/linux.h:40:25: note: in expansion of macro 'GNU_USER_DYNAMIC_LINKER'
-dynamic-linker " GNU_USER_DYNAMIC_LINKER "} \
^
/b/tmp/work-shared/gcc-5.3.0-r0/gcc-5.3.0/gcc/gcc.c:884:32: note: in expansion of macro 'LINK_SPEC'
static const char *link_spec = LINK_SPEC;
^
<command-line>:0:27: warning: deprecated conversion from string constant to 'char*' [-Wwrite-strings]
/b/tmp/work-shared/gcc-5.3.0-r0/gcc-5.3.0/gcc/gcc.c:1295:48: note: in expansion of macro 'STANDARD_STARTFILE_PREFIX'
static char *const standard_startfile_prefix = STANDARD_STARTFILE_PREFIX;
Signed-off-by: Marek Vasut <marex@denx.de> Cc: Khem Raj <raj.khem@gmail.com> Cc: Ross Burton <ross.burton@intel.com> Cc: Richard Purdie <richard.purdie@linuxfoundation.org> Signed-off-by: Ross Burton <ross.burton@intel.com>
Haris Okanovic [Tue, 5 Jan 2016 19:04:29 +0000 (13:04 -0600)]
tzdata: Make /etc/timezone optional
Add INSTALL_TIMEZONE_FILE config variable to enable/disable installation
of ${sysconfdir}/timezone (/etc/timezone) by tzdata packages. Defaults
to "1" to maintain previous behavior.
Most libc implementations can be configured to retrieve system's
defaults timezone from /etc/localtime, and don't need a second file
(/etc/timezone) to express this configuration. Maintaining this file is
an unnecessary burden on sysadmins unless there's software using
/etc/timezone directly (I.e. outside of libc). Some distributions may
choose not to provide it.
Testing: Built tzdata under default config and verified CONFFILES_tzdata
still has both /etc/timezone and /etc/localtime and both are in the
image. Built with INSTALL_TIMEZONE_FILE = "0" and verified /etc/timezone
is removed from CONFFILES_tzdata and the image. Successfully installed
package to an x64 target.
Signed-off-by: Haris Okanovic <haris.okanovic@ni.com> Reviewed-by: Brad Mouring <brad.mouring@ni.com> Reviewed-by: Rich Tollerton <rich.tollerton@ni.com> Reviewed-by: Ken Sharp <ken.sharp@ni.com> Reviewed-by: Ross Burton <ross.burton@intel.com>
Natinst-ReviewBoard-ID: 121628 Signed-off-by: Ross Burton <ross.burton@intel.com>
On sysvinit systems, volatiles is configured to make /etc/resolv.conf symlink
to a file in a volatile path, which lets us write to /etc/resolv.conf for
read-only-rootfs. For systemd, this isn't set up unless we enable
systemd-resolved, which we don't by default. When it's not enabled, create the
/etc/resolv.conf symlink and ensure the volatile path is created on boot with
tmpfiles.d.
Signed-off-by: Christopher Larson <chris_larson@mentor.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
We don't need nss-myhostname on systemd systems, because systemd already
provides myhostname, but we weren't configuring nsswitch to use it. Being able
to resolve the hostname is useful for a number of different applications, so
enable it using the same postinst/prerm bits which are in nss-myhostname.
Signed-off-by: Christopher Larson <chris_larson@mentor.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
This lets someone use a different update-alternatives-native provider. Without
this available, they'll step on one another in the sysroot unconditionally,
since we need to build opkg-utils-native for ipk based builds regardless.
Signed-off-by: Christopher Larson <chris_larson@mentor.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Richard Purdie [Fri, 8 Jan 2016 17:36:46 +0000 (17:36 +0000)]
populate_sdk_ext: Change to include siginfo and non sstate task sigs
Right now, the locked task hashes list for the extensible SDK locks
down only the sstate tasks.
Whilst asthetically pleasing, this gives two problems:
* Half the task are left floating meaning checksum mismatches
are a pain to debug
* The later code which copies relavent data files out the sstate
cache can't use any of this data.
This patch modifies things so all the checksums are listed in the locked
file. An exclusion of tasks probably makes more sense for the library
function rather than an allowed list.
The only sstate task being deliberaly excluded here was do_package
so add in a function to explictly exclude those sstate object files.
The net result of this that siginfo files for all tasks are included in
the SDK, which means commands like "bitbake -S printdiff" now function.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Wed, 6 Jan 2016 22:57:46 +0000 (22:57 +0000)]
image: Create separate tasks for rootfs construction
This patch splits the code in lib/oe/image into separate tasks, one per
image type. This removes the need for the simple task graph code and defers
to the bitbake task management code to handle this instead.
This is a good step forward in splitting up the monolithic code and starting
to make it more accessible to people.
It should also make it easier for people to hook in other tasks and processes
into the rootfs code.
Incidentally, the reason this code was all combined originally was due to
limitations of fakeroot where if you exited the session, you lost permissions
data. With pseudo this constraint was removed.
We did start to rework the rootfs/image code previously and got so far with
untangling it however we did prioritise some performance tweaks over splitting
into separate tasks and in hindsight, this was a mistake and should have been done
the other way around. That work was suspended due to changes in the people working
on the project but this split has always been intended, now is the time to finish
it IMO.
There were some side effects of doing this:
* The symlink for the manifest moves to the rootfs-postcommands class and into
the manifest function.
* There is no seperate "symlink removal" and "symlink creation", they are merged
* The date/time stamps of the manifest and the built images can now be different since
the tasks can be run separately and the datetime stamp will then be different
between do_rootfs and the do_image_* tasks.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Wed, 6 Jan 2016 22:57:40 +0000 (22:57 +0000)]
image: Move pre/post process commands to bbclass
As the next step in splitting up do_image, move the pre and post processing
commands to separate tasks. This also creates the do_image_complete task
which acts as the end marker task for image generation.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Wed, 6 Jan 2016 22:57:34 +0000 (22:57 +0000)]
image.bbclass: Separate out image generation into a new task, do_image
I've heard complaints from people trying to create more interesting image
types about how hard it is to understand the rootfs/image generation code
and that its a pain to develop/test/debug.
Having looked at it myself, the internal construction of shell functions which
then gets passed into a multiprocessing pool is rather convoluted and it places
rather odd constraints on when variables are expanded. Its therefore no wonder
people find it confusing/complex.
This patch starts the process of splitting this up by separating out image
generation from the do_rootfs task into a new do_image task.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Fri, 8 Jan 2016 18:22:11 +0000 (18:22 +0000)]
sstatesig: Handle special case of gcc-source shared-workdir for printdiff
Often, bitbake -S printdiff would show that there was a checksum not found
which would turn out to be from gcc-source. This is due to it being a
shared-workdir recipe.
For now, hardcode the special case into the sstatesig code to stop people
(including me) puzzling over this.
If/as/when we add any other shared workdir recipes, we'll need to rethink
this.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ross Burton [Fri, 8 Jan 2016 16:25:04 +0000 (16:25 +0000)]
bitbake.conf: add virtual/libiconv-native to ASSUME_PROVIDED
It's possible for a native recipe to have virtual/libiconv-native as a build
dependency, but as we expect that the host provides that add it to
ASSUME_PROVIDED.
Signed-off-by: Ross Burton <ross.burton@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:55 +0000 (00:15 +1300)]
devtool: build: support using BBCLASSEXTENDed names
It's logical that you would want to build BBCLASSEXTENDed items
separately through devtool build, so simply allow that - we're just
passing the name verbatim to bitbake, so all it means is adjusting the
validation.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:54 +0000 (00:15 +1300)]
devtool: reset: support recipes with BBCLASSEXTEND
If the recipe file itself was created in the workspace, and it uses
BBCLASSEXTEND (e.g. through devtool add --also-native), then we need to
clean the other variants as well.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:53 +0000 (00:15 +1300)]
devtool: refactor code for getting local recipe file
We're doing this in a couple of places, let's just find the recipe file
if it exists within the workspace (which it will if it's been added
through "devtool add") when we read in the workspace.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:52 +0000 (00:15 +1300)]
devtool: add: support adding a native variant
Sometimes you need to build a variant of a recipe for the build
host as well as for the target (i.e. BBCLASSEXTEND = "native"); add a
--also-native command line option to "recipetool create" that enables
this and plumb it through from an identical option for "devtool add".
(We could conceivably do the same for nativesdk, but I felt it might be
confusing within the context of the extensible SDK, where nativesdk
isn't really relevant to the user.)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:51 +0000 (00:15 +1300)]
devtool: reset: do clean for multiple recipes at once with -a
We need to run the clean for all recipes that are being reset before we
start deleting things from the workspace; if we don't, recipes providing
dependencies may be missing when we come to clean a recipe later (since
we don't and couldn't practically reset them in dependency order). This
also improves performance since we have the startup startup time for the
clean just once rather than for every recipe.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:50 +0000 (00:15 +1300)]
recipetool: create: support creating standalone native/nativesdk recipes
If the recipe name ends with -native then we should inherit native;
likewise if it starts with nativesdk- then inherit nativesdk.
(Note that the recipe name must actually be specified by the user in
order to trigger this - we won't do it based on any name auto-detected
from e.g. the tarball name.)
Since we're doing this based on the name, "devtool add" will also gain
this functionality automatically.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:49 +0000 (00:15 +1300)]
recipetool: create: lower case name when determining from filename
As a matter of general convention we expect recipe names to be lower
case; in fact some of the packaging backends insist upon it. Since in
this part of the code we're auto-determining the name, we should convert
the name to lowercase if it's not already so that we're following
convention and avoiding any validation issues later on.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:47 +0000 (00:15 +1300)]
devtool: sdk-update: fix error checking
Running "raise" with no arguments here is invalid, we're not in
exception handling context. Rather than also adding code to catch the
exception I just moved the check out to the parent function from which
we can just exit.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:46 +0000 (00:15 +1300)]
devtool: sdk-update: fix metadata update step
* Clone the correct path - we need .git on the end
* Pull from the specified path instead of expecting a remote to be set
* up in the repo already (it isn't by default)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:45 +0000 (00:15 +1300)]
devtool: sdk-update: fix not using updateserver config file option
We read the updateserver setting from the config file but we never
actually used that value - the code then went on to use only the value
supplied on the command line.
Fix courtesy of Dmitry Rozhkov <dmitry.rozhkov@intel.com>
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Wed, 6 Jan 2016 11:15:43 +0000 (00:15 +1300)]
classes/populate_sdk_ext: fix cascading from preparation failure
During extensible SDK installtion, if the build system preparation step
fails we try to put something at the end of the environment setup script
to show an error when it is sourced, in case the user doesn't realise
that the partially-installed SDK is broken. However, an apostrophe in
the message (actually a single quote) appears to terminate the string
and therefore breaks the command. Drop it to avoid that.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
