Kang Kai [Tue, 7 Jun 2011 08:16:43 +0000 (16:16 +0800)]
eglibc: migrate configurability from oe
Migrate configurability from oe, try to shrink minimal image size
All eglibc configuration options are included in conf/distro/include/default-distrovars.inc.
Define DISTRO_FEATURES_LIBC and DISTRO_FEATURES in local.conf with the same way
in local.conf.sample.extended to enable those options.
Anders Darander [Mon, 13 Jun 2011 17:26:16 +0000 (19:26 +0200)]
runqemu: take TMPDIR from bitbake
The old standard of using /home/anders/openembedded-core/build/tmp as TMPDIR do not work any longer,
as - is appended to the real TMPDIR. Thus, always take TMPDIR from
bitbake, unless TMPDIR is set in the environment.
Khem Raj [Mon, 13 Jun 2011 06:05:16 +0000 (23:05 -0700)]
eglibc 2.12/2.13: Upgrade SRCREV
We need http://www.eglibc.org/archives/patches/msg00923.html
to get through the problem we see when OPTION_EGLIBC_RTLD_DEBUG
is disabled. This patch has been backported to both 2.12 and 2.13
branches. We bump SRCREVs to have that change.
Kang Kai [Tue, 7 Jun 2011 08:29:41 +0000 (16:29 +0800)]
ghostscript: update SRC_URI
Build ghostscript-native fails on a i686 machine because it can't get
the source objarch.h and soobjarch.h, and .h files are not needed for
native package, so update the SRC_URI to fix it.
Signed-off-by: Kang Kai <kai.kang@windriver.com>
Removed the do_configure_prepend and folded into the target do_configure
since the native was picking it up for the do_configure_virtclass/native()
and then failing on i686.
Signed-off-by: Saul Wold <sgw@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Dexuan Cui [Fri, 10 Jun 2011 07:25:19 +0000 (15:25 +0800)]
git: restore the dependency on perl-native
[YOCTO #1155]
I thought git-native could depend on perl-native-runtime and tests on
Ubuntu 9.04/10.10 and Fedora 13 show it could buid fine (looks these distros
install perl-ExtUtils-MakeMaker by default).
However Joshua reported on Fedora 15 i686 host, git-native can't build unless
he manually installed perl-ExtUtils-MakeMaker to the host.
This makes me think we may as well make git-native depend on perl-native.
Bruce Ashfield [Wed, 8 Jun 2011 19:33:02 +0000 (15:33 -0400)]
linux-yocto: rename recipes to explicitly indicate version
The existing recipe names for the linux-yocto kernel builds
worked well when there was a single, or two versions of the
kernel available. But with the impending kernel updates and
retirement of older kernels, the re-use of the same recipes
for different kernel versions violates the principle of least
surprise.
To address this, the recipes are being renamed as follows:
Phil Blundell [Fri, 3 Jun 2011 19:55:33 +0000 (19:55 +0000)]
uclibc: remove PACKAGE_ARCH
There is no good reason for uclibc to be machine specific. Remove local
assignment to PACKAGE_ARCH so that it gets the default target
architecture and bump PR for that change.
See http://lists.linuxtogo.org/pipermail/openembedded-core/2011-May/003064.html
Phil Blundell [Fri, 10 Jun 2011 22:01:40 +0000 (22:01 +0000)]
uclibc: remove redundant python code
This chunk of python code has been around for a while (witness the
comment about gcc 3.4.0) and predates the availability of
COMPATIBLE_HOST. Rewrite it using a more modern idiom.
Dexuan Cui [Mon, 13 Jun 2011 13:24:14 +0000 (21:24 +0800)]
pixman: upgrade from 0.20.2 to the latest stable 0.22.0
Update LIC_FILES_CHKSUM:
1) COPYING: only format change; no actual license change.
2) 0.22.0 doesn't have pixman-x64-mmx-emulation.h, so let's remove it.
For some test I need to back to use gcc-4.5, so I changed GCCVERSION and
SDKGCCVERSION to "4.5.1", but I found gcc-cross-canadian's 4.6.0, rather
than 4.5.1, would be built! Finally I made this patch to fix the issue.
Richard Purdie [Mon, 13 Jun 2011 13:36:12 +0000 (14:36 +0100)]
nativesdk.bbclass: Correct ordering of manipulations
OVERDIDES is the key variable that needs to be set prior to data finalisation.
The other variables should be manipulated after finalisation so any
DEPENDS_prepend and _append are accounted for. This patch ensures this is
the case.
The PACKAGES maniupulations are not enabled at this time as they
don't function 100% correctly yet.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Removed several instances to the 5.0 Bernard release. I replaced
these with the 5.0.1 release. I also re-wrote a paragraph that
instructs the user on where to find and download the most recent
YP tarball. It used to point to the Index of /Downloads into the
"yocto-1.0" folder. I now instruct the user to go to the
Yocto Project website and download the desired release from there.
Scott Rifenbark [Tue, 7 Jun 2011 14:08:50 +0000 (07:08 -0700)]
documentation: updated manual history
I added an entry to represent the Yocto Project 1.0.1 Release into
the manual history table for all of the manuals except the Yocto
Project Quick Start, which does not have a table.
Scott Rifenbark [Fri, 3 Jun 2011 15:44:36 +0000 (08:44 -0700)]
- documentation/Makefile: Cleaned up Makefile
[BUGID#_1025] - I added some conditionals to handle the
Yocto Project Quick Start case. This manual does not have
a PDF version. I put in tests for publishing and for a case
where a user might attempt to specifically generate a PDF
using 'make pdf'. I also converted the version variable into
a command-line argument so we don't have to edit the make file
when a new release comes out.
Beth Flanagan [Thu, 2 Jun 2011 17:18:35 +0000 (10:18 -0700)]
documentation: [YOCTO #1025] build system for docs
These changes fix the following issues:
1. Multiple Makefiles. There really is no need for this.
2. Unable to maintain more than one version of the docs on the webserver.
This is a quick fix to enable the above.
In order to build the documentation, at the top level, issue a:
make all DOC=<doc directory name>
make all DOC=kernel-manual
Also, some changes need to occur on the webserver to fully
incorporate these fixes.
http://www.yoctoproject.org/documentation/
The docs are now published to:
http://www.yoctoproject.org/docs/<Release MM.mm>/<doc name>
The main page should be changed to point not only to the current doc release,
but also to the prior releases. This will enable us to maintain prior release
documentation without stomping over it when we publish new docs. Also, we'll
need to repoint the yocto-quick-start link to yocto-project-qs. Or rename
documentation/yocto-project-qs/* to support the website naming.
Richard Purdie [Fri, 10 Jun 2011 08:13:24 +0000 (09:13 +0100)]
bitbake/fetch2: When replacing URLs in mirror handling mask out empty entries
The symptom of this problem is something like a cvs url which specifies
a username where the username is then passed through to something like
an http mirror.
This patch fixes things by ensuring empty entries are preserved in the
new URL.
This change clearly breaks builds since subsequent patches
depend on changes this patch removes. Not sure what's going
on here but this change isn't correct.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Thu, 9 Jun 2011 16:30:17 +0000 (17:30 +0100)]
bitbake/event/ast: Add RecipePreFinalise event
One of the implications is we need to register the event handlers before
executing the anonymous python functions. I can't find any issue with making
that change in any existing metadata use cases.
This patch changes the behaviour such that at the time bitbake expands the DEPENDS_z
override, it removes "DEPENDS_z" from the data store. In the above example this would
mean that it wouldn't matter how often you call d.update_data(), you'd always get
"a c" back.
See the bitbake-devel mailing list for further discussion and analysis of the
potential impact of this change.
Mark Hatle [Thu, 9 Jun 2011 17:06:44 +0000 (12:06 -0500)]
pseudo: Fix problem related to realpath
When pseudo is disabled, certain programs that call realpath may not
work properly. This was discovered when using the Qt MOC tool when
certain qmake project features are used.
Paul Eggleton [Wed, 1 Jun 2011 14:37:07 +0000 (15:37 +0100)]
track skipped packages
Add skiplist to cooker that allows access to the list of packages skipped
via SkipPackage (this includes COMPATIBLE_MACHINE, INCOMPATIBLE_LICENSE,
etc.) This can be used to enhance error reporting.
The default list of EXTRA_IMAGE_FEATURES adds alot to the generic non-dev and non-sdk image
so ensure that the sato-sdk and lsb-sdk includes this list by default.
Phil Blundell [Thu, 9 Jun 2011 11:58:43 +0000 (12:58 +0100)]
kexec-tools: don't depend on virtual/kernel
There doesn't appear to be any terribly good reason for kexec-tools to
depend on the kernel. (I verified that kexec-tools is buildable in a
clean TMPDIR without having previously built virtual/kernel.)
Having this dependency in place is a nuisance because it makes it
awkward to put kexec into an initramfs. So, it seems like we would be
better off without.
Paul Eggleton [Wed, 8 Jun 2011 13:20:25 +0000 (14:20 +0100)]
iproute2: update to 2.6.38
Fixes "ip route get" not producing any output (a regression in 2.6.35).
See http://marc.info/?l=linux-netdev&m=129442470405398&w=2 and
http://marc.info/?l=linux-netdev&m=130038222321440&w=2 for a list of
other changes since 2.6.35.
projects / thirdparty / openembedded / openembedded-core-contrib.git / log
