From d54b975506c392cfb59dbb3caf313ac061b6fa01 Mon Sep 17 00:00:00 2001
From: Jani Uusi-Rantala <jani.uusi-rantala@nokia.com>
Date: Thu, 11 Aug 2011 18:37:56 +0300
Subject: [PATCH] Magic file path should be given for rpmbuild in
_rpmfc_magic_path define so that build system default file
is not used by accident. Not doing this caused many
packages to fail building in several systems.
Fixes [YOCTO #1358]
Signed-off-by: Jani Uusi-Rantala <jani.uusi-rantala@nokia.com>
Xiaofeng Yan [Wed, 10 Aug 2011 08:57:32 +0000 (16:57 +0800)]
chkconfig: remove link for update-alternatives to disambiguate
[YOCTO #936]
"update-alternatives" installed: one in /usr/sbin from chkconfig \
(symlinked to "alternatives"), and the other in /usr/bin \
from update-alternatives-cworth.
It appears for whatever reason that the one from chkconfig is run \
during postinst processing but if you run the script from the command \
line it gets the other one.
this is due to differences in the PATH environment variable.
The following is the sequence to call scripts after kernel booting
inittab
-->rcS
-->./S40networking(PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin)
-->S98configure(call rpm-postinstall)
#"/usr/sbin" is found prior to "/usr/bin", so update-alternatives from chkconfig is run in this script
-->rc5.d
-->profile(profile:4:PATH="/usr/local/bin:/usr/bin:/bin" profile:15: PATH=$PATH:/usr/local/sbin:/usr/sbin:/sbin)
#"/usr/bin/" is found prior to "/usr/sbin", so update-alternatives from update-alternatives-cworth is run in this script
So I remove the symlink (update-alternatives linked to chkconfig). The one from update-alternatives-cworth is left alone.
Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Leandro Dorileo [Wed, 10 Aug 2011 05:09:05 +0000 (01:09 -0400)]
scripts/combo-layer: a simple way to script the combo-layer conf
This small patch introduces a a very simple and basic way to script
the combo-layer conf file. With that a combo can be shared with no
need to change its config - associated to the use of environment
variables for example.
*Similar* to bitbake it considers every value starting with @ to be
a python script. So local_repo could be easily configured as:
Richard Purdie [Wed, 10 Aug 2011 18:39:31 +0000 (19:39 +0100)]
sstate: Add level 2 debug so its possible to see what sstate is doing
Currently its hard to figure out if/when sstate is checking for possible
packages to speed up builds. This patch adds level 2 debug output which
better indicates what files are being searched for an why.
[YOCTO #1259]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Tue, 9 Aug 2011 17:51:44 +0000 (18:51 +0100)]
gcc: Various fixups to ensure consistent gcc builds
We ensure that:
* the shared work directory contains PR and ensure PR values are consistent across gcc builds
* the regexp to handle library directories is in a specific task and run once
This avoids breakage that was seen in incremental builds after commit be1f70d68b6b75772ebab8bdff683ddd7c42b0cd where the interpretor could
become corrupted. This was due to the sed expression corrupting
the source directory.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Kumar Gala [Tue, 9 Aug 2011 03:52:11 +0000 (22:52 -0500)]
eglibc: force GLIBC_EXTRA_OECONF to not be uset for nativesdk
Don't use GLIBC_EXTRA_OECONF for the nativesdk builds. Easiest solution
is to just override it settings. Otherwise we might pickup configure
options ment for the target build.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Chris Larson [Tue, 5 Apr 2011 19:01:04 +0000 (12:01 -0700)]
oe.terminal: improve how we spawn screen
- Name the screen session 'devshell', to avoid confusion if running bitbake
itself under a screen session.
- Display a warning message when spawning screen, so it's clear to the user
that screen has been run (otherwise do_devshell just appears to hang).
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Chris Larson [Tue, 29 Mar 2011 19:53:19 +0000 (12:53 -0700)]
Rework how the devshell functions
In the new implementation, each known terminal is defined as a class in
oe.terminal, as a subclass of bb.process.Popen. terminal.bbclass wraps this
functionality, providing the metadata pieces. It obeys the OE_TERMINAL
variable, which is a 'choice' typed variable. This variable may be 'auto',
'none', or any of the names of the defined terminals.
When using 'auto', or requesting an unsupported terminal, we attempt to spawn
them in priority order until we get one that's available on this system (and
in the case of the X terminals, has DISPLAY defined). The 'none' value is
used when we're doing things like automated builds, and want to ensure that no
terminal is *ever* spawned, under any circumstances.
Current available terminals:
gnome
konsole
xterm
rxvt
screen
Signed-off-by: Chris Larson <chris_larson@mentor.com>
Chris Larson [Mon, 8 Aug 2011 23:09:37 +0000 (16:09 -0700)]
image: implement IMAGE_FEATURES
IMAGE_FEATURES is analagous to DISTRO_FEATURES and MACHINE_FEATURES, for
root filesystem construction. Currently, the only supported features are
any defined package groups, as used by the oe.packagegroup python
module.
Dexuan Cui [Thu, 4 Aug 2011 06:53:20 +0000 (14:53 +0800)]
scripts/oe-buildenv-internal: improve the error detecting for $BDIR
The previous fix for a bug in Ubuntu 10.04 readlink, be2a2764d8ceb398d81714661e6f199c8b11946c, notified the user when a trailing
slash was used. As there is no semantic difference, simply remove any
trailing slashes and proceed without nagging the user.
See [YOCTO #671] for more details.
Signed-off-by: Dexuan Cui <dexuan.cui@intel.com> Acked-by: Darren Hart <dvhart@linux.intel.com> Cc: Paul Eggleton <paul.eggleton@linux.intel.com>
Richard Purdie [Tue, 9 Aug 2011 14:04:27 +0000 (15:04 +0100)]
sqlite3: Ensure nativesdk package extension is handled correctly
This avoids the warning:
NOTE: multiple providers are available for runtime libsqlite3-dev (sqlite3, sqlite3-nativesdk)
NOTE: consider defining a PREFERRED_PROVIDER entry to match libsqlite3-dev
until such times as we convert nativesdk to use a prefix like multilib.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Tue, 9 Aug 2011 11:01:04 +0000 (12:01 +0100)]
default-distrovars.inc: Allow world builds to be successful
Currently we exclude some packages with license issues from world builds
but we don't exclude packages that depend on them leading to errors
when trying a "bitbake world". This patch also blacklists the dependees
so that a world build doesn't show horrible errors and requires the -k
option.
[YOCTO #1262]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Daniel Lazzari [Fri, 5 Aug 2011 19:56:17 +0000 (12:56 -0700)]
defaultsetup: Use .= to add TCLIBCAPPEND to TMPDIR
Lazily appending causes a bug where wrong cache is
cleared when BB_SRCREV_POLICY = "clear".
Tested with qemuarm on uclibc/eglibc in same build dir
Signed-off-by: Daniel Lazzari Jr <dlazzari@leapfrog.com> Signed-off-by: Khem Raj <raj.khem@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Dongxiao Xu [Mon, 8 Aug 2011 07:35:38 +0000 (15:35 +0800)]
module-init-tools-cross: Clean the RDEPENDS value
cross recipes should not RDEPENDS on any package. Cross recipe has no
"PACKAGES" define, thus clean RDEPENDS_${PN} value to ensure correctness
in multilib.
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Saul Wold [Sat, 6 Aug 2011 20:39:21 +0000 (13:39 -0700)]
kernel-yocto: add CCACHE_DIR dependency to do_kernel_configme
[YOCTO #1350]
Since do_kernel_configme is added before the standard do_configure task
we needed to add CCACHE_DIR so when the kernel builds it's host configure
tools the CCACHE_DIR exists.
Bruce Ashfield [Fri, 5 Aug 2011 20:22:18 +0000 (16:22 -0400)]
linux-yocto: merge v3.0.1
3.0.1 -stable has been released. This now becomes the baseline for
the 3.0 linux-yocto tree. As was the policy in the 2.6.34 and
2.6.37 kernels, the version stays at 3.0 in the recipe.
Build and boot tested on qemu* targets.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Bruce Ashfield [Tue, 2 Aug 2011 19:09:50 +0000 (15:09 -0400)]
linux-yocto: allow configuration of arbitrary branches
When building an external tree or bootstrapping a BSP the
external branch may not have been checked out. The tools now ensure
that the tree is ready for configuration, so we no longer need to
force the checkout of the external branch.
This change is coupled with some kern tools tweaks as follows:
40d9bab updateme: allow the location of board descriptions based on defines 59859ca createme: use branch name when creating meta data 91b4275 configme: determine meta branch based on directories, not branch naming f5a915c kgit-meta: make branch creation and renaming more robust
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Bruce Ashfield [Fri, 29 Jul 2011 14:19:00 +0000 (10:19 -0400)]
linux-yocto: pass KMACHINE to updateme, not MACHINE
To support the mapping of any oe/yocto MACHINE to a kernel
branch that may not share that naming structure we have
KMACHINE and KBRANCH. To allow the mapping to work, we
actually have to pass KMACHINE into updateme and not MACHINE.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Mei Lei [Fri, 5 Aug 2011 01:52:15 +0000 (09:52 +0800)]
apr-util: disable pqsql support to avoid configure error
The pqsql config script will check host path and add host paths to compiler and linker options:
adding "-I/usr/include/postgresql" to CPPFLAGS
adding "-L/usr/lib" to LDFLAGS
Disable pqsql support since we didn't use this feature in other recipes.
Mei Lei [Thu, 28 Jul 2011 07:20:36 +0000 (15:20 +0800)]
python-native: Fix a compiler finding issue
The CC variable sometimes add option information after compiler name, but python can't get the real compiler name if those information added.
Fix this issue by dropping the option information when finding compiler name.
Kumar Gala [Thu, 4 Aug 2011 18:54:59 +0000 (13:54 -0500)]
gcc: use ${base_lib} to match gcc default configuration
Rather than tweaking MULTILIB_DIRNAMES & MULTILIB_OSDIRNAMES like is
done for x86-64 via 64bithack.patch. We can just go with gcc defaults
and utilize ${base_lib} for where to find gcc libs.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Dongxiao Xu [Fri, 5 Aug 2011 05:48:25 +0000 (13:48 +0800)]
task-core-boot.bb: Add PACKAGE definition to facilitate multilib
If there is no PACKAGES defined in bb file, its default definition would
be "${PN}-dbg ${PN} ${PN}-doc ${PN}-dev ${PN}-staticdev ${PN}-locale".
In multilib case, ${PN} is extended with ${MLPREFIX}, so there is no
rename occurred for the PACKAGES. Therefore for task-core-boot,
RDEPENDS_lib32-task-core-boot will be empty, resulting busybox,
tinylogin, etc. are not bulit out.
This fixed the issue that /bin/sh is not provided while do_rootfs.
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Lianhao Lu [Fri, 5 Aug 2011 01:56:42 +0000 (09:56 +0800)]
tcp-wrappers: Using ${PN} in PACKAGES.
[YOCTO #1334]
Using ${PN} instead of tcp-wrappers in PACKAGES and FILES_*, since the
FILES_tcp-wrappers would be overwritten by FILES_${PN} when the variable
name gets expansioned.
Richard Purdie [Fri, 5 Aug 2011 16:22:16 +0000 (17:22 +0100)]
machine/include/arm/feature-arm-thumb: Allow thumb to be disabled
The previous commit to this file meant thumb was always being turned on
even when TUNE_FEATURES did not contain "thumb". This is clearly wrong
and this patch corrects this so thumb options are no longer specified
in that case.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
KERNEL_LD was using ${LD} in it's definition, which is not correct for
different ABIs such as x32 or i386 on x86_64 machine. This brings it
into sync with the corresponding gcc settings, likewise the same with
the KERNEL_AR variable.
[RP: Updated commit message] Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Kumar Gala [Thu, 4 Aug 2011 07:23:47 +0000 (02:23 -0500)]
tune/arch-powerpc64: include arch-powerpc.inc to keep things in sync
Added a DEFAULTTUNE setting and included arch-powerpc.inc. This way we
pick up the changes to TUNE_PKGARCH and things should be kept more in
sync going forward.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Khem Raj [Wed, 3 Aug 2011 21:12:30 +0000 (14:12 -0700)]
coreutils: Fix build on uclibc
coreutils has getloadavg.m4 of its own. So we need
to make sure that we remove the portions which look
into hardcoded /usr/local/lib paths. These tests are
only for AIX so we dont lose much
Khem Raj [Wed, 3 Aug 2011 21:12:29 +0000 (14:12 -0700)]
autoconf: Remove code checking for hardcoded /usr/local
This code is only relevant for AIX and causes
problems with OE QA sniffer which detects -L/usr/local/lib
on linker commandline during cross compile and rightly barfs
This only happens when getloadavg() is not found in libc
so we do not reach that code with eglibc but only with uclibc
Khem Raj [Wed, 3 Aug 2011 21:12:28 +0000 (14:12 -0700)]
autoconf: Remove patch=1 from SRC_URI
Edit summary field s/produced/produce/
Fix the license related warning
WARNING: autoconf-native: No generic license file exists for:
GPLv2|GPLv3 at /b/openembedded-core/meta/files/common-licenses
WARNING: autoconf-native: There is also no SPDXLICENSEMAP for this
license type: GPLv2|GPLv3 at
/b/openembedded-core/meta/files/common-licenses
Beth Flanagan [Wed, 3 Aug 2011 19:31:08 +0000 (12:31 -0700)]
buildstats.bbclass: Adding disk io measurement
This commit adds disk-io statistics functionality on a build
and per-task basis.
It pulls measurements for whatever partition TMPDIR exists on.
This data could be off if SSTATE_DIR and DL_DIR exist on
a different partition/volume.
Notes on what this pulls:
ReadsComp: Total number of reads complete
ReadsMerged: Total number of adjacent reads merged
SectRead: Total number of sectors read
TimeReads: Total number of m/s spent reading
WritesComp: Total number of writes completed
SectWrite: Total number of sectors written
TimeWrite: Total number of m/s spent writing
IOinProgress: Total amount of IO in progress at the time of
we look at /proc/diskstats.
TimeIO: Total number of m/s spent doing IO
WTimeIO: Weighted time doing I/O. From iostats.txt:
"This field is incremented at each I/O start, I/O completion, I/O
merge, or read of these stats by the number of I/Os in progress
(field 9) times the number of milliseconds spent doing I/O since the
last update of this field. This can provide an easy measure of both
I/O completion time and the backlog that may be accumulating."
Tom Rini [Wed, 27 Jul 2011 18:08:15 +0000 (11:08 -0700)]
Various siteinfo files: Consolidate va_copy/__va_copy/va_val_copy
Providing va_copy / __va_copy come down to the libc. va_val_copy
comes down to the architecture. Unfortunately it's assumed true
if not set, so we need to make sure to set this to false for
x86_64 where it is not true.
Tom Rini [Tue, 26 Jul 2011 23:23:01 +0000 (16:23 -0700)]
Various siteinfo files: Consolidate ac_cv_func_getaddrinfo
This is a libc feature, move there. Drop the ipsec-tools specific
test as ipsec-tools (a) assumes fine on cross-compile and (b) doesn't
support buggy getaddrinfo now.
eglibc builds for arm-eabi unconditionally enables the libc-do-syscall.S
file, which contains thumb assembly. It's unused in arm-eabi-nothumb case,
so just ifdef the actual code.
V2: use the patch from oe.dev.
The patch is submitted to sourceware bugzilla by Khem Raj back in October
of 2010, but the status is still UNCONFIRMED and the patch isn't commited.
Tom Rini [Wed, 27 Jul 2011 22:01:35 +0000 (15:01 -0700)]
connman_test.sh: Rework for busybox 'ps'
This script has two problems today. First, it does 'ps -ef cmd'
in failure which real ps doesn't grok and busybox ps just ignores
the argument on. Switch that to 'ps -ef'. Second, busybox ps -o
doesn't understand cmd but does understand comm. Using comm lets
us simplify the test logic as well, so switch to that.
Zhai Edwin [Tue, 2 Aug 2011 07:24:00 +0000 (15:24 +0800)]
xserver-nodm-init: Fix X start failure on some platform
sudo is used to run rootless X in xserver-nodm, and start to fail with
"xf86OpenConsole: Cannot open /dev/tty0" error after upgrade to 1.8.1. sudo
seems wait for some resource that is unavailable in early phase of booting.
This patch swith to "su" for rootless X starting as a fix.
Noor, Ahsan [Mon, 1 Aug 2011 10:16:18 +0000 (15:16 +0500)]
bison: Add dependency on flex-native
* This is 0479b70418ef553859029911c57c63a7aaebe299 from OE. flex-native is needed to build bison. The dependency was being satisfied indirectly but we need to add it explicitly.
