Jackie Huang [Sat, 4 Aug 2012 13:14:08 +0000 (21:14 +0800)]
python: Fix to support the python module _bsddb built with db 5.3
_bsddb module in python 2.7 could be built only with db version
between 4.1 and 4.7. A patch was added to avoid build warning
about this for [YOCTO #1937] but not actually fixed it.
This patch enable _bsddb module be built with db 5.3, and remove
--disable-statistics from the DB5_CONFIG to fix segmentation fault
when using _bsddb module in python.
Khem Raj [Sat, 4 Aug 2012 04:58:20 +0000 (21:58 -0700)]
glib-2.0-nativesdk: Add dependency on glib-2.0-native
when we build for nativesdk it needs tools from installed
glib-2.0-dev package. We zeroes out compile_resourses but
then genmarshal came up so instead of going about disabling
them one by one lets depend on glib-2.0-native since
thats the right thing to do.
Khem Raj [Sat, 4 Aug 2012 04:58:21 +0000 (21:58 -0700)]
populate_sdk_rpm.bbclass: Depend do_populate_sdk task on createrepo-native
Now that we have moved the rpm-createsolvedb.py
script to createrepo-native for good reason of
python-native dependency, we have to make sure
that it is staged before its used, unlike before
where it was under scripts/ dir and always existed
outside the realm of package management.
Fixes build error when doing meta-toolchain builds from
scratch
|
/home/kraj/yocto/poky/build/tmp/work/ppc64e5500-poky-linux/meta-toolchain-1.0-r7/temp/run.populate_sdk_image.26712:
line 610: rpm-createsolvedb.py: command not found
| DEBUG: Python function do_populate_sdk finished
| ERROR: Function failed: populate_sdk_image (see
/home/kraj/yocto/poky/build/tmp/work/ppc64e5500-poky-linux/meta-toolchain-1.0-r7/temp/log.do_populate_sdk.26712
for further information)
Martin Jansa [Mon, 23 Jul 2012 14:25:09 +0000 (16:25 +0200)]
bitbake.conf: exclude whole MACHINEOVERRIDES from OVERRIDES vardeps
* whole MACHINEOVERRIDES can change e.g. between MACHINES with different arm architecture, causing allarch packages to reexecute do_package
bitbake-diffsigs ../shr-core/tmp-eglibc/stamps/all-oe-linux/xserver-nodm-init-2.0-r16.do_package.sigdata.90e760a8f6cecbd87cb2e95f1237e3cc ../shr-core/tmp-eglibc/stamps/all-oe-linux/xserver-nodm-init-2.0-r16.do_package.sigdata.9eeccfd15f25032b3b6b132534660fff
basehash changed from 7618e17d3fda05d1f15246e6800ca0f0 to 97bc4dc8c1521c535bd96b2aa62d8a03
Variable MACHINEOVERRIDES value changed from ${MACHINE}${@bb.utils.contains("TUNE_FEATURES", "armv5", ":armv5", "" ,d)}${@bb.utils.contains("TUNE_FEATURES", "armv4", ":armv4", "" ,d)}:${MACHINE_CLASS} to ${MACHINE}${@bb.utils.contains("TUNE_FEATURES", "armv7a", ":armv7a", "" ,d)}${@bb.utils.contains("TUNE_FEATURES", "armv6", ":armv6", "" ,d)}${@bb.utils.contains("TUNE_FEATURES", "armv5", ":armv5", "" ,d)}${@bb.utils.contains("TUNE_FEATURES", "armv4", ":armv4", "" ,d)}:${MACHINE_CLASS}
Jackie Huang [Mon, 6 Aug 2012 07:05:44 +0000 (15:05 +0800)]
tcl: Add ${bindir_crossscripts}/tclConfig.sh to sysroot stage
tclConfig.sh is changed in do_install for cross compile and
is installed to STAGING_BINDIR_CROSS, but if SSTATE_DIR is set
and tcl is from sstage, tclConfig.sh can't be found in
STAGING_BINDIR_CROSS, add ${bindir_crossscripts}/tclConfig.sh
to sysroot stage can fix it.
Wenzong Fan [Fri, 3 Aug 2012 03:30:21 +0000 (11:30 +0800)]
gdk-pixbuf: fix parallel install issue
This patch fixes parallel install issue that lib libpixbufloader-png.la
depends on libgdk_pixbuf-2.0.la which will be regenerated during insta-
llation, if libgdk_pixbuf-2.0.la is regenerating and at the same time
libpixbufloader-png.la links it, the error will happen.
Error message is:
* usr/bin/ld: cannot find -lgdk_pixbuf-2.0
* collect2: ld returned 1 exit status
Make an explicit dependency to the libs install targets would fix this
issue.
Roy.Li [Mon, 6 Aug 2012 02:50:01 +0000 (10:50 +0800)]
busybox: fix du-k-works test case
[YOCTO #2896]
The du-k-works test seems to assume that du -k on a directory with two files
totaling 80 kilobytes should report either 80k or 88k. In ext4 filesystm however,
both busybox' du -k and coreutils' du -k reports a size of 84k which causes the
test to fail.
Richard Purdie [Mon, 6 Aug 2012 11:13:06 +0000 (11:13 +0000)]
gcc-cross-initial: Ensure it uses an isolated sysroot
If we don't do this, a stale limits.h may be detected in STAGING_DIR_TARGET
which would result in a different limits.h getting generated by gcc-cross-initial
that references it. The referenced limits.h will then not get found by eglibc-initial
causing rather strange build failures.
The simplest solution is to create a temporary sysroot containing only the things
gcc-cross-initial should care about and this results in a correct limits.h file
regardless of what else may have been built.
and treat them like they were do_install_append. Clearly this isn't desirable.
Only one instance of this type of issue was found in OE-Core and has been fixed
so correcting the regexp should be safe to do.
Martin Jansa [Thu, 2 Aug 2012 16:52:15 +0000 (18:52 +0200)]
bitbake: cooker: fix OVERRIDES in BB_SIGNATURE_HANDLER (to override value from DISTRO.conf from local.conf)
We really need to pass the finalised data store into the parser init function
(and hence the siggen init function). This ensures any value changes get
passed into the correct code.
RP: The list of recipes in multilib.conf needs to go away and we need
to just be able to extend all recipes with the multilib class.
Tested by building and running lib32-core-image-sato-sdk.
Signed-off-by: Shane Wang <shane.wang@intel.com> Signed-off-by: Liming An <limingx.l.an@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Martin Jansa [Tue, 31 Jul 2012 23:33:41 +0000 (01:33 +0200)]
bitbake: fetch2: print checksums when they are different then expected
* in form which can be copied into a recipe without modification
* like oe-classic did since:
http://git.openembedded.org/openembedded/commit/?id=68abc465559a68e9201c9525be3c4acc6297eaed
* it shows them in right form when they are missing completely, but in
more verbose form when different
* it needs to print that only when checksums were requested, e.g.
fetching from sstate mirror sets both md5mismatch and sha256mismatch,
but your checksums shouldn't be shown
bitbake: fetch2: handle broken symlinks in local mirror handling
If a file:// mirror is being used, the fetcher will create a symlink to the
local file. However, if the local file gets removed, that link will be dead,
and os.path.exists() returns False in that case, so it tries and fails to
recreate the link. Now we unlink such a dead link if it exists.
Without the new exception, when the system is configured to use premirrors,
but not allow network access (via BB_NO_NETWORK), when a recipe was lacking a
checksum the wrong error message(s) were being generated.
Instead of complaining about trying to perform network access, if the system
was able to find the item in the premirror, it should inform the user of the
lack of checksums, and the two SRC_URI fields they should use to update their
recipe.
Some patches (disable_goops_optimizations, fix_cross_compilation) don't
seem to be needed anymore. Others (change-install-data-hook, mark-unused-modules)
were updated to work on 2.0.6. Tested by building and running guile under
QEMU with core-image-sato-sdk and also builing guile on MIPS.
udhcpc is invoked from the networking service, so it shouldn't exist
standalone. If /etc/init.d/busybox-udhcpc is present, it will cause
udhcpc to be started two times.
And since the script doesn't support the stop action, it will cause
udhcpc to be started even when system is being shutdown.
Remove /etc/init.d/busybox-udhcpc to resolve the issues.
Most of the code in relocatable.bbclass will be used for relocating the
SDK binaries. So, create another class chrpath.bbclass that will contain
the core of the relocatable.bbclass, so we can reuse it.
In order for the SDK to be relocatable, the user would need to call a
setup script to change the binaries acordingly. Having an auto-extracting
archive has the advantage of being more user friendly and the user does
not have to call the setup script separately after extracting the SDK.
It is called automatically.
This script will be embedded in the SDK tarball and will be called by
the SDK installer. It replaces the interpreter path in all binaries and
it also changes the ld.so.cache and SYSDIRS in dynamic loader.
This patch is needed so that all SDK binaries have the PT_INTERP section
size set to 4096 (max path size in Linux) in order to be able to parse
the binaries later and change the interpreter to the path we want.
Modifying the dynamic loader path in all binaries by the SDK installer
would not be possible because we cannot know in advance the SDK target
location. Hence, the PT_INTERP section size has been set to 4096 (which
is the maximum path lengh in Linux).
Also, for the dynamic loader SYSDIRS and ld.so.cache paths, special
4096 bytes sections were allocated so that we can easily parse the ELF
binary at install time and modify the paths.
Martin Jansa [Thu, 2 Aug 2012 12:09:50 +0000 (14:09 +0200)]
u-boot.inc: fix regexp used when ld-is-gold
* with old regexp it wasn't reentrant, causing stuff like this:
-LD = $(CROSS_COMPILE)ld
+LD = $(CROSS_COMPILE)ld.bfd.bfd
-LDR = $(CROSS_COMPILE)ldr
+LDR = $(CROSS_COMPILE)ld.bfd.bfdr
when do_compile was reexecuted, also breaking LDR variable and
http://git.yoctoproject.org/cgit/cgit.cgi/meta-ti/tree/recipes-bsp/u-boot/u-boot/0001-config-Always-use-GNU-ld.patch
when it was used together with u-boot.inc from oe-core (meta-ti is using own u-boot.inc)
* This patch is also better solution to ld-is-gold problem then regexp in metadata.
Richard Purdie [Wed, 1 Aug 2012 08:01:28 +0000 (08:01 +0000)]
gcc-configure-target: Set native-system-header-dir for target gcc
Without this set, in the case host=target (which it does for on-target
gcc), the native header directory is set to the target sysroot with
no prefix. This means it would look for sdt.h on the build system
instead of in the target headers and this can lead to build failures
and is host contamination of the build.
The correct fix is to explicitly set the native header directory to
the correct location and then the headers get detected correctly.
package_rpm.bbclass: Accomodate dash when using arrays
we were assigning local variable to an array coming through
positional arguments. local is a non posix contruct thats
also supported by dash luckily but operates differently in this
case it exapnds the array before assignment. so
local pkgs="$@"
turns into pkgs=locale-base-en-us locale-base-en-gb
ant we see errors
run.do_rootfs.25593: 932: local: locale-base-en-gb: bad variable name
So lets not use defining and assigning local in one go
first define a local and then the assignment
Paul Eggleton [Mon, 30 Jul 2012 08:52:01 +0000 (09:52 +0100)]
scripts/oe-pkgdata-util: exclude unpackaged in glob output
Check for .packaged file and exclude packages if this file does not
exist - this avoids attempting to install empty packages during
complementary package installation within do_rootfs that didn't end up
being created (and failing as a result).
Paul Eggleton [Tue, 31 Jul 2012 10:22:05 +0000 (11:22 +0100)]
combo-layer: ensure init works with split local config
If the local configuration is already split out, ensure the init action
works properly and does not error in the case that last_revision is not
set. Additionally, if the local configuration is within the repository,
prevent it from being committed and add it to .gitignore.
When you have a distro which has older version of
glib-2.0 installed then it does not have compile_resources
installed which essentially glib-2.0 nativesdk is
looking for on build host. So like in a cross build
we disable the configure check which will force it
to build an internal copy and use it.
This problem may not be seen on newer distros like
ubuntu 12.04 since they have newer version of glib-2.0
installed
Martin Jansa [Tue, 31 Jul 2012 02:00:01 +0000 (04:00 +0200)]
package.bbclass: fix TypeError in runstrip
* some packages have .ko files which are not elf, without this change
it fails with TypeError, with this change only runstip fails and
reports where:
ERROR: runstrip: ''arm-oe-linux-gnueabi-strip' '/OE/shr-core/tmp-eglibc/work/armv4t-oe-linux-gnueabi/emacs-23.4-r0/package/usr/share/emacs/23.4/etc/tutorials/TUTORIAL.ko'' strip command failed
The patch of converting char type to signed char is only applied to arm,
in fact it should be applied to powerpc too since powerpc has the same
assumption for char type. To x86 and mips, this patch has not impact.
scripts/bitbake: unbreak the git version comparison
With the current code, we're calling awk to do a floating point comparison
between '1.7.0.4' and '1.7.5' (on an ubuntu 10.04 LTS machine). These clearly
aren't proper floating point numbers, and the comparison is incorrect. It's
returning true for 1.7.0.4 >= 1.7.5.
Instead of using a floating point comparison for this, call out to python and
let it do it.
Paul Eggleton [Tue, 31 Jul 2012 00:06:26 +0000 (01:06 +0100)]
combo-layer: allow splitting out local config
Allow splitting the local parts of the configuration (mostly
local_repo_dir and last_revision, although there is no limitation) to
a side-by-side -local.conf file, with component sections optionally
tagged with the combo layer branch name. This effectively allows you to:
* avoid polluting the history by committing the updated last revision
to the combo repository for every update
* avoid putting local repo paths into the combo repository
* manage multiple branches of the combo repository whilst avoiding the
possibility of mixing the configuration for one branch with another.
An example split configuration (note, values may be artificial):
Paul Eggleton [Tue, 31 Jul 2012 00:06:24 +0000 (01:06 +0100)]
combo-layer: improve patch list handling and output
* Ignore blank lines in patch list
* Don't fail in interactive mode if patch list is deleted
* Show patch counter
* Show relative path for patches
* Print headings before applying patch list for each component
Also change to using a "with" block to read the patch list so it gets
closed properly when we're finished.
Paul Eggleton [Tue, 31 Jul 2012 00:06:23 +0000 (01:06 +0100)]
combo-layer: drop to a shell when apply fails during update
If applying a patch fails during the update process, drop to a shell
instead of exiting; at that point the user can manually apply the patch,
do nothing and "exit" to skip it, or "exit 1" to abort the process.
Paul Eggleton [Mon, 30 Jul 2012 23:31:00 +0000 (00:31 +0100)]
qt4: allow recipes building commercial edition
To enable building the commercial edition of Qt (through additional
recipes that are *not* provided by OE-Core) we need to tweak a few
things:
* Don't make recipes that inherit qt4x11.bbclass depend on qt4-x11-free
- instead add qt4-x11 to DEPENDS and then have qt4-x11-free include
this in its PROVIDES. A commercial equivalent recipe should do the
same.
* Add a QT_LICENSE_FILE variable that can be used to specify the license
file required by the commercial edition.
* Add a QT_LICENSE_FLAGS variable which the recipe can set to select the
license option being used. The default of "-opensource" retains the
current behaviour; a commercial recipe should set it to "-commercial".
We set do_populate_sdk[depends] in populate_sdk_*.bbclass, but since
these are inherited at the top of populate_sdk_base.bbclass, those
values are wiped out by using = at the end of the latter class. Use +=
to avoid this.
guile:meta/Makefile change install-data-hook to install-exec-hook
In guile meta/Makefile, what install-data-hook do
must be run after install-binSCRIPTS.so it's a exec hook, not a data hook.
If keep these exec-hook in data-hook,when we run make with multi-jobbing ,
install-data-hook and install-binSCRIPTS will be separately
processed in different thread, no any dependence.
that will cause error sometimes.
Paul Eggleton [Sun, 29 Jul 2012 05:41:24 +0000 (06:41 +0100)]
classes/package: ensure pkgdata is cleaned out first
If do_package needs to be re-run and packages are removed/renamed as a
result, we want the corresponding stale pkgdata files to no longer be
present; for that we need to ensure all of the old pkgdata files are
cleaned out prior to creating the new ones. This is particularly
important now we rely on pkgdata more heavily (for complementary
packages, license manifest, etc.).
Peter Seebach [Fri, 27 Jul 2012 21:54:48 +0000 (16:54 -0500)]
pseudo.inc/pseudo_1.4.bb: update pseudo to 1.4
This update replaces the half-baked --arch logic with the use
of $CFLAGS to pick compiler flags, on the grounds that it makes
a lot more sense for the build system to pick flags than for
pseudo to try to guess what they should be; this should allow
pseudo to at least compile for targets, and possibly run on
them.
This doesn't solve the problem of guessing how to forcibly
build the 32-bit variant on hosts, because we really don't
have a general solution for that. There's no idiom for "given
this set of compiler flags and this architecture, what flags
would you use to request a 32-bit compile instead?" So we
basically ignore that for now. If someone comes along trying
to use the build system to build pseudo-native on a 64-bit
host that also supports 32-bit binaries and isn't x86, we
will revisit this.
Paul Eggleton [Mon, 23 Jul 2012 06:59:11 +0000 (07:59 +0100)]
buildhistory: improve performance of image info collection
Reduce the number of calls to the packaging tool, especially in the case
of rpm, using helper utilities to gather the required information more
efficiently where possible.
projects / thirdparty / openembedded / openembedded-core-contrib.git / log
