In section B.1 (Parsing) it said that BBFILES variable by default
specified the direcotyr 'meta/packages/' as the place to look for .bb
files. This directory is invalid and needed to be changed to
'meta/packages/'.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Due to some changes in the file structure for 1.0 there is a new directory
for the shared state. The variable SSTATE_DIR can be used to point to
the directory. I added this variable to the list of documented variables.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Scott Rifenbark [Mon, 4 Apr 2011 22:49:11 +0000 (17:49 -0500)]
documentation/poky-ref-manual/ref-structure.xml: Edits for Rel 1.0
These edits reflect changes in the directory structure from the 0.9
version of the software to the 1.0 version. This set of changes still
is missing a few items. Changes were based on Saul Wold's input.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Scott Rifenbark [Mon, 4 Apr 2011 21:20:58 +0000 (16:20 -0500)]
documentation/poky-ref-manual/ref-classes.xml: re-write of autotooled packages
Section C.2 (Autotooled Packages) was re-written. I removed a bunch
of <variable> tags and replaced them with <filename>. Also removed
some Britishisms.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Section A.1.1 (bitbake/) contained two URLs. One supposedly went to
a BitBake site and the other to the BitBake online manual. In reality,
they both went to the online manual. I removed the one referencing
the site.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Section 5.3.2 (Using OProfileUI) was out of date. Several of the URLs
would not resolve. They were pointing to openedhand links that had
not been maintained. I updated the entire section.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
In section 5.2.2. (Building the Cross-GDB Package) there was a resulting
directory listed where you could find the binary. The directory had
a couple of variables for 'host-arch' and 'target-abi'. There was
a mis-guided angle bracket wrapping the 'host-arch' variable. This
was fixed.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Mei Lei [Fri, 25 Mar 2011 08:18:21 +0000 (16:18 +0800)]
distrodata.bbclass: Merge the get_pkg_info.log into checkpkg.csv
For easy view the check package result, merge the two files into checkpkg.csv, after that, the package report system will use checkpkg.csv instead of get_pkg_info.log.
When building one of the native, nativesdk or crosssdk packages TARGET_*
variables' values are no longer related to the target we set via MACHINE
variable, they are now related to the BUILD (native) or SDK (nativesdk,
crosssdk) targets instead. We need to change TARGET_FPU variable
accordingly or some of the recipes (the ones that check for TARGET_FPU
value, most notably gcc and eglibc) might be confused.
It's probably cleaner not to reset TARGET_FPU but to change it to
something like ${BUILD_FPU} (for native) or ${SDK_FPU} (for crosssdk and
nativesdk) but as long as BUILD and SDK are x86 it's safe to just reset
TARGET_FPU.
Signed-off-by: Ilya Yanok <yanok@emcraft.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Nitin A Kamble [Tue, 29 Mar 2011 17:46:45 +0000 (10:46 -0700)]
libxcb: fix for broken library link in the image
log.do_package:
NOTE: the following files were installed but not shipped in any package:
NOTE: /usr/lib/libxcb-dri2.so.0
NOTE: /usr/lib/libxcb-dri2.so.0.0.0
NOTE: libxcb-dev contains dangling symlink to
/usr/lib/libxcb-dri2.so.0.0.0
Then because of the dangling symlink, ldconfig fails at the time of
rootfs creation of image.
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Scott Rifenbark [Fri, 1 Apr 2011 01:33:18 +0000 (20:33 -0500)]
documentation/poky-ref-manual/development.xml: Edits for Eclipse and Anjuta
I commented out a large section of the chapter, which went into detail
on how to locate, install, configure, and use the Yocto Ecliple
plug-in. This information is redundant in this book and is better
explained in the ADT Manual. I am referencing the information from
this chapter now.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Bruce Ashfield [Sat, 2 Apr 2011 03:12:45 +0000 (23:12 -0400)]
linux-yocto/meta: cleanup and streamline kernel meta data
Fixes [YOCTO #945]
As part of the update to 2.6.37 existing configuration and
patches from the -stable kernel were left in place as a reminder
of features and configuration carry forward. A lot of these
reminders are no longer necessary and the kernel meta data
needs to be cleaned up to prepare for activities related to
newer kernels.
Also in this meta update are configuration changes to allow
common_pc derived BSPs to have a clean baseline configuration.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Tom Rini [Thu, 31 Mar 2011 17:45:57 +0000 (10:45 -0700)]
patch.bbclass: allow importing patches into other dirs
This is taken from openembedded.master:
commit 9d0d70da60ca8ef278916f442125d41161846f38
Author: Chris Larson <chris_larson@mentor.com>
Date: Mon Aug 23 13:03:55 2010 -0400
patch: allow importing patches into other dirs
Use the 'patchdir' flag. If relative, it's assumed to be relative to ${S}.
It creates a patchset class per patch directory, so for quilt this results i
multiple .pc/patches directories to manage each of the patchsets.
Signed-off-by: Chris Larson <chris_larson@mentor.com> Signed-off-by: Tom Rini <tom_rini@mentor.com>
Joshua Lock [Thu, 31 Mar 2011 20:46:42 +0000 (21:46 +0100)]
openjade: fix build with GCC 4.6
In GCC 4.6 the compiler no longer allows objects of const-qualified type to
be default initialized unless the type has a user-declared default
constructor.
Patch from Gentoo bugzilla: http://bugs.gentoo.org/show_bug.cgi?id=358021
Joshua Lock [Thu, 31 Mar 2011 12:43:41 +0000 (13:43 +0100)]
xorg-[lib-common|proto-common]: disable use of fop document generation
On Fedora 15 I see a huge Java backtrace when document generation runs for
some xorg libs. As fop is automatically detected, with the possibility of
detecting fop on the host whilst doing target builds, the safest bet is to
explicitly disable fop for document generation.
Joshua Lock [Wed, 30 Mar 2011 22:21:50 +0000 (23:21 +0100)]
elfutils: fix builds with gcc 4.6
gcc 4.6 (as used in Fedora 15) adds some extra warnings which are
included with Werror. The new unused-but-set variable warning causes
an error in libasm of elfutils. Work around this by removing
unused-but-set from Werror.
Scott Rifenbark [Thu, 31 Mar 2011 18:31:27 +0000 (13:31 -0500)]
documentation/yocto-project-qs/yocto-project-qs.xml: Edits plus Matt Madison note regarding older host systems
I made a few small edits and I added a reference to the
wiki page 'https://wiki.yoctoproject.org/wiki/BuildingOnRHEL4'
that has entries for older development hosts. Right now all that
is there is the RHEL4 notes but the wiki page can be expanded as needed.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Scott Rifenbark [Wed, 30 Mar 2011 14:17:33 +0000 (09:17 -0500)]
documentation/poky-ref-manual/usingpoky.xml: [BUGID# 929] - Note added warning about switching up GPL versions
In chapter 2 where we talk about building images I added a new
note indicating that the user should not switch around using different
GPL versions when trying to rebuild an image as it can cause dependency
failures.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Khem Raj [Thu, 17 Mar 2011 23:33:43 +0000 (16:33 -0700)]
scripts/poky-qemu-internal: call stty sane before exit
When qemu is booted into console with -nographics
then after exiting the terminal line settings are messed
up. This patch calls stty sane to restore the terminal
settings to default.
stty is part of coreutils which is installed on all
host distros hence there is no need to warn about it
being available or not
Paul Eggleton [Thu, 31 Mar 2011 13:33:16 +0000 (14:33 +0100)]
README.hardware: remove u-boot flashing instructions for mpc8315e-rdb
Upgrading u-boot is apparently not necessary with current board revisions,
and these instructions may not work properly anyway (our toolchain seems not
to be able to compile u-boot in the way described), and given that they are
potentially risky they should be removed.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Mark Hatle [Wed, 30 Mar 2011 02:16:16 +0000 (21:16 -0500)]
Workaround for Global C++ Constructor problem on ARM
[YOCTO #938]
Workaround for a problem with the order of the global C++ constructors on ARM.
The workaround is simply to avoid defining the ID numbers outside of the
usage of the ID's.
This also has the effect of fixing a problem on MIPS, where "_mips" is a
defined symbol and unavailable on the system for a variable name.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
