There is a major issue with opkg images at the moment as preinst
functions are not being executed before their dependencies are installed
and this is leading to corruption of images containing avahi/dbus in
particular.
There are various changes in upstream opkg in the last 8 revisions which
make changes in this area but sadly these aren't enough to get things
working for us. I've updated to the latest svn revision with this patch
since it makes sense to pull in those changes first and then supplement
them with the attached patches.
There is a full description of the patches in the patch headers but in
summary they:
a) Ensure preinst functions execute with their dependencies installed.
This is a pretty invasive change as it changes the package install
ordering in general.
b) Ensure opkg sets $D, not $PKG_ROOT which we don't use
c) Change opkg to allow execution of postinstall functions which fail
resulting in execution on the target device as rootfs_ipk.bbclass
currently does manually.
The remaining changes interface this with the rest of the OE build
infrastructure, adding in the option to tell opkg to run the preinst and
postinst functions, ensure the correct environment is present for the
postinst scripts and removing the now unneeded rootfs_ipk class code
which opkg now does itself.
[YOCTO #1711]
(From OE-Core rev: 2feba313c991170747381c7cf821a45c2cd04632)
(From OE-Core rev: b7e2eff8c18bc59605fb711ac4540985c71f155a)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Occasionally we keep seeing "unable to open temp file" messages during
do_package_write_rpm tasks. This appears to happen when multiple
processes are writing rpm files and is likely due to using the
shared system temp directory. This patch changes the tmp path
to the package work directory meaning conflicts should become
a non-issue.
(From OE-Core rev: b2ef543284c8c8d0d3badb2e1bcadad1106982d2)
(From OE-Core rev: 018442ed2cd251f85212dfa1d03c0b24a0750bfa)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If PACKAGE_CLASSES does not set package_rpm as the first item, the root filesystem
will not be generated based on rpm. We need remove rpm/zypper tests against
non-rpm filesystem.
[YOCTO #1757]
(From OE-Core rev: 43adb8dcf4461b68a7ce0ba9d8acdb2012a70416)
(From OE-Core rev: f541517be97e27951157e1dd10256e132c31ab1f)
Signed-off-by: Jiajun Xu <jiajun.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If rpm-native was built in an alternative location, it may not relocate correctly
unless the rpmpopt and macros paths are explicitly specified.
This fixes errors seen on the Yocto autobuilder where pkgconfig
"provides" entries could disappear leading to image dependency failures.
(From OE-Core rev: fb01bd81197057e62106aac966f9ebc4c5054f97)
(From OE-Core rev: 15f50ab3ee454dc3510801d61bb09bf37d78d1af)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Without this, if rpmddeps came from a sstate package which was relocated
it might not find its magic file and if that happens, requires/provides
in packages could get corrupted. This leads to failures at rootfs time
during builds with messages like:
libdbus-1.so.3 is needed by libdbus-glib-1-2-0.92-r1.armv5te
since the provides would be missing in the dbus package.
(From OE-Core rev: abe2a948905a997314c61a8fcd35e2b42a3f4408)
(From OE-Core rev: 5ee145ebbb32824e3870b6dd689ce2b3f9bf3f17)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The method of resolving the patch should not effect the sstate-cache
signature.
(From OE-Core rev: b64cbe0b511de8d8943ce34cbb4901239d9f0cb0)
(From OE-Core rev: 896bd7d1442dcd3f080dc741a72f50ab95d7c38f)
Signed-off-by: Matthew McClintock <msm@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Without this change we can end up looking for <stamp>.${MACHINE}
instead of the expected expanded value.
(From OE-Core rev: 9f743b5033177216fe0e1d3e43ba831f356df08e)
(From OE-Core rev: de9f47b09d5434642ba925182ae21a8e77e7e429)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently bad recommendations are added to the status file with status
"ok". After a single opkg command, whilst it will ignore the recommendation,
the status changes to "installed" even if the recommended package was not
installed. Whilst this is likely a glitch in opkg's logic, the correct
way to persist the information in the status file is to set the status
to "hold" as deinstall packages with that status remain. With this change
the bad recommendations persist accross multiple opkg runs and the system
behaves as expected.
[YOCTO #1758]
(From OE-Core rev: 215ff6b2e9676c8c7dd8acfd696151bcd0f1490f)
(From OE-Core rev: 525743f5513feff67fb8fd2e4c7a1a05ae22ddc9)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Disable script debugging, as the log files become huge and take a
long time to process during the log check step. This results in a
performance improvement.
(From OE-Core rev: a7e70227bac72c4f7d3419f94f6915da4c7e3f43)
(From OE-Core rev: 9b6ecd1fd2f6870ace033362e3bb86fd98935bc9)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
[YOCTO #1755]
We change the want the RPM rootfs install works to install pre and post install
scripts. The new method uses a script helper that is invoked by RPM outside
of the normal chroot.
The wrapper is dynamically generated prior to the install starting. It will
check the return code of the script. If the script fails, it will store a copy
to be executed on the first system boot. This is similar to the previous
mechanism.
In addition, a line of debug was added to the scripts as written by package_rpm
to list which package and which script for later debugging, if necessary.
(From OE-Core rev: 3e7120d6a9fd5e46214673d0a6e1085a7314ff42)
(From OE-Core rev: 5d74a2bbe036cf586b76aef0d9907ecb3d4a5f1d)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Without this rootfs generation fails as an RDEPENDS is added
but the package might not have bneen built.
(From OE-Core rev: bfe70c6446e6686f826f01040ba74c7d7d28bf42)
(From OE-Core rev: 30a1f1d3ec763b4929b052ab3388499dfb40b1fa)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Exclude the addition of user/group code and RDEPENDS changes for
-nativesdk packages.
(From OE-Core rev: 2f057dd905ccb497890ce73ac4e4c256edcf0351)
(From OE-Core rev: 9acbe80fea3dbd5405030c95d8b6d411689c4911)
Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Previously we injected the user/group preinstall script into all
output packages. This fixes that so that only packages listed in
USERADD_PACKAGES get modified.
It also removes the USERADDPN variable, which is no longer needed.
(From OE-Core rev: 2f73466eb5018040a123ccb0e2af8c519525f958)
(From OE-Core rev: 424b6447ebce761c9027ffdaf68ecbcd6f28e4ec)
Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Our current assumption (based on the behaviour of opkg) when writing
recipes is that prerm and postrm do not get called during an upgrade.
When using rpm however, these are mapped to the rpm "preun" and "postun"
events which occur after postinst for upgrades, and when these contain
removal type operations (such as update-alternatives --remove) this
causes problems.
This patch wraps each preun and postun script for rpm in a check that
determines whether or not the script is being called during an upgrade,
and skips the entire script if it is, which mimics the behaviour of opkg
under the same conditions.
Fixes [YOCTO #1760]
(From OE-Core rev: 1d3f37dc9a43ba6d6beb7b4530c077f239032b99)
(From OE-Core rev: 6e66ecd201760fe418a9884e3605b88a68208776)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* without this patch, building dbus-nativesdk leads to a missing
dependency on 'base-passwd-nativesdk'
This was added by commit 46e6c3fa8034b12d178d605f3f5d7efe69671a13
* this patch handle the nativesdk case in the class useradd
* close bug 1702 http://bugzilla.pokylinux.org/show_bug.cgi?id=1702
* v2 from Scott Garman with Richard Purdie's tricks
(From OE-Core rev: 140a3507fb5c14cd9bcebe4304f491aa1c5c47a2)
(From OE-Core rev: 79d5ce46b4d73e5ed39c509ce872e99e6bcb94ee)
Signed-off-by: Eric Bénard <eric@eukrea.com>
Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There should just be a single IMAGE_INSTALL variable. If the package
backends need this split into different multilib components they should
be responsible for doing this, not the user.
This commit removes the MULTILIB_IMAGE_INSTALL variable.
[YOCTO #1564]
(From OE-Core rev: 7736862a74c92fe1afe42e170822be13117575c2)
(From OE-Core rev: 4889865934d590bf18d9f8f8ec3b63ce992cd4c5)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fix bug where only packages named PN included base-passwd in
RDEPENDS.
This fixes [YOCTO #1727]
(From OE-Core rev: 2c55d51afd71d708a54afc8377e10c4f80f810e3)
(From OE-Core rev: 213d31f24d911a10132ddcd75f50363a80c4dc2e)
Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
genext2fs only creates the minimum number of inodes, after this patch it will scale with the rootfs size
(From OE-Core rev: c31cb0bdc5a61d2d9f21a2cea34c3d8ac3b47cb9)
(From OE-Core rev: 41d1091e6b821404eeb73a7c363537c2835558d3)
Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
tmpfs/encryptfs/(and most likely, but not confirmed)ramfs TMPDIRs
cause diskstats to choke. No device entry ends up in /proc/diskstats
for these fs types, which ends up causing the failure.
The short term solution is to exclude these fs types from diskstat
collection. Longer term we will want to see if we can collect
meaningful diskio for each of these, and other, use cases, but for
this cleans up Bug 1700.
[YOCTO #1700]
(From OE-Core rev: 2b14046c12855b6f484ba5bd6bc0a8022de6873e)
(From OE-Core rev: e0d26e0e1dfb9d35d71f20488c16f3cea3da862e)
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Corrected YOCTO bug location and format
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Preinst are run alphabetically which breaks when e.g. avahi-daemon needs /etc/passwd present.
(From OE-Core rev: d6793165feb26c51b5f19ad1e6d1a4099878e879)
(From OE-Core rev: 0485c362ac6ee0a3e310de078d7a202b961fed11)
Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Before this change:
a) Ownership and permissions of files copied from packages to
package-split could get lost during the copy process. This change
ensures they are preserved.
b) Ownership and permissions of directories could also get lost.
Most of the complexity in this patch is addressing this problem
ensuring newly created directories match the source ones being
copied.
c) There was no warning about directories being created but not
shipped by any package.
This patch fixes all of the above issues.
(From OE-Core rev: 6021e309e69d823e1467648aee12a32182945569)
(From OE-Core rev: 5f9228b32c243ae499398763ce7c90b776dc9d24)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The use of groupadd -f makes much more difficult to figure when a
group is not add. This was the case of the class not working for our
usage and this being caused by the lack of '/etc/group' file but
unnoticed as groupadd wasn't failing according.
(From OE-Core rev: 82933a1ff921fd0836f03e6f379fd8536cdc0a30)
(From OE-Core rev: e3e8f15176107fa26248e878af548835692d3068)
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When external modules are built, files in $STAGING_KERNEL_DIR/scripts/basic will/can get
rebuilt.
This raises a potential race condition. Prevent this by adding a lock around the
do_make_scripts() function. Further, make sure that the kernel has been installed
to the sysroot, prior to executing this new task.
(From OE-Core rev: 8681b82e8b466929205edde7ba479f3ac1a6143e)
(From OE-Core rev: 694e3016e25dff3f573291830d79982c8b8793a2)
Signed-off-by: Anders Darander <anders@chargestorm.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fix broken rootfs image link when ubifs is used.
Function runimagecmd is using image name "${IMAGE_NAME}.rootfs.${type}".
Let's use the same name in IMAGE_CMD_ubifs.
(From OE-Core rev: 766f6165471691f651584ebda004e1abb4ea9eb6)
(From OE-Core rev: 6c4276ee968bed7a5b3e74637183414a428facb8)
Signed-off-by: Lauri Hintsala <lauri.hintsala@bluegiga.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is needed to better support things like the following (with a
multi-word BUILD_CC):
EXTRA_OECONF += '"ac_cv_prog_CC_FOR_BUILD=${BUILD_CC}"'
(From OE-Core rev: 38a394e7ffedccfabda085c97add8944718943c2)
(From OE-Core rev: 5c26de72b97a670a263428ef3a1846385683feeb)
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changes to baselib by specific machine configuration were resulting
in sstate cache invalidation, particularly in multilib configurations.
This patch ensures this doesn't happen and native sstate cache files
are reusable.
(From OE-Core rev: d0915fb0a2cc80ad45b3fd526d3b29a91d99572c)
(From OE-Core rev: 4fe88a2a3c7cec3ad9ea13d39d71d317405c910a)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
sstate was being a little too ethusiastic about removing stamp files and
was removing stamp files for other machines when it shouldn't have been.
This patch teaches sstate about machine specific stamp extensions and
allows it to only remove the current task's stampfiles.
Based on a patch from Phil Blundell <philb@gnu.org> with some tweaks
from me.
(From OE-Core rev: 5e9488495401399d39fcb5012b86c313b6caca73)
(From OE-Core rev: e8efeedbc2ec1587b1c4d938c25cacd4e8611053)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
patch depends on unpack
configure depends on patch
We simply don't need a configure dependency on unpack. This simplifies
the dependencies of every recipe slightly and should make bitbake
slightly faster at resovling dependency graphs.
It also makes the .dot dependency graphs slightly more readable by
removing noise.
(From OE-Core rev: c54c1280fc0d06a53e23339c3913ec88eead13d9)
(From OE-Core rev: a5b205090d3244cd578d611fd45f2e2f4818b284)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
On OpenSUSE within an X session, TMPDIR is set to the system temporary
directory (/tmp) which is incorrect for these scripts. Thus, change
runqemu and oe-setup-rpmrepo to use OE_TMPDIR from the external
environment rather than TMPDIR.
Fixes [YOCTO #1530]
(From OE-Core rev: 4e24c10952c7a52af7f2447595fd484692d35534)
(From OE-Core rev: 354497bf3e3bf68374875caa97d9b4fdcad74789)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
libtool obtains the search path from /etc/ld.so.conf and hardcodes /usr/lib
and /lib. This results in host contamination and variable sets of RPATH
values ending up in binaries.
By exporting the correct values for all autotools recipes we avoid this.
(From OE-Core rev: 93e595d5c89ebacdb8d1e6fcfe6f58fe2d30de28)
(From OE-Core rev: 5e41e0973a9be890ac310e1bbf465fcd08b0add5)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This fixes some error messages in the do_rootfs logs of non-multilib
builds.
(From OE-Core rev: fb554596e031cf92b62a19cabdd10e8e23ab4453)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The target SDK packages don't need to fulfil a shell dependency
so add /bin/sh to the list of packages we don't need to resolve.
(From OE-Core rev: 9283255da08f45a368fa9355dbafd3840dfd5056)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The help2man script is pretty useless to us. It requires to run the target
binary to extract help information which is not possible for any of our
cross compiled target binaries.
We're not interested in man pages for -cross/-native tools.
It therefore makes no sense to have this as a core build dependency.
This patch removes the dependeny and replaces it with a script
returning false. This will trigger autotool's missing utility
to use the copy of the man page included with the sources which
is what would already happen when we tried to run cross compiled
binaries anyway.
(From OE-Core rev: c6e0f23363f24ae9f02cd753621ce45470285b16)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add MLPREFIX to multilib deploy forlder to avoid the confliction between
multilib and normal package deploy directory.
(From OE-Core rev: b5e8cad5a782015f2216325203847c287c778cac)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There are some places that PACKAGES are dynamically set. To support
multilib, we need to add MLPREFIX before the package name in those
settings.
(From OE-Core rev: 98d356a9f788291c849be7b51fcd8ad07a8a066e)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When RPM does the real install, if the first manifest file is empty, the
installation will stop without handling the second manifest file.
Merge the two manifest files together to fix this issue.
(From OE-Core rev: 20e6f166858751c6305cd8a52f87cdf78c1a8126)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
currently we have allarch type of recipes, which may still have
architecture dependency, like x11-common. So we need to drop the
handling to allarch in multilib case.
Also remove the PV postfix in python-pygobject DEPENDS, since multilib
code will treat a native package multilib capable.
[YOCTO #1497]
[YOCTO #1498]
(From OE-Core rev: d9dc64a251bc66f16a0c5d12aa872152d43c4776)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
RDEPENDS of image type recipe needs to be mapped to make sure that the
packages included in the image should be multilib version.
Also add LINGUAS_INSTALL into MULTILIB_PACKAGE_INSTALL list.
[YOCTO #1496]
[YOCTO #1527]
(From OE-Core rev: ad52cf921b2e08f2a99f494b229d5b7099b33990)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This fixes a problem when building meta-toolchain-gmae, by adding items that
will be provided by the host system, such as /bin/bash, /usr/bin/env and
libGL.so
(From OE-Core rev: 01361f9d25b0a0027bbbe713b93051a4663b14fc)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
RP: libGL.so() -> libGL.so
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch partially fixes problems when building multilib extended
images such as libXX-core-image-minimal. Its not a perfect/complete
solution but works much better than any previous code did.
[YOCTO #1496] (partial)
[YOCTO #1497] (partial)
[YOCTO #1498] (partial)
(From OE-Core rev: 00c38774ef0232cc2be924ed8e59220e7c452096)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There is no point in warning against .so symlinks in -nativesdk
packages. They have to contain such links, so shut up the warning.
(From OE-Core rev: 287d02a731e45ce5d843db1fd4d78ba17e9ed487)
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
icecc.bbclass assumed that nativesdk and cross-canadian packages should
use local icecc version. Instead those packages should use target-like
scheme for toolchain tarball generation.
(From OE-Core rev: 394c7c8cf262efeaafce40dc5eb480a1e041ac50)
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
icecc-create-env(-native) is a native package. But if it doesn't have
the -native suffix, some dependency tracking stuff in oe-core (in
nativesdk.bbclass) get's crazy about it, trying to generate dependencies
to icecc-create-env-native packages. To fix that simply add -native
suffix to this package.
(From OE-Core rev: 83f9ebe9afb8bec4531899f8001a585c974e1db4)
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie
Jiajun Xu
Matthew McClintock
Martin Jansa
Mark Hatle
Scott Garman
Paul Eggleton
Eric Bénard
Dongxiao Xu
Koen Kooi
Elizabeth Flanagan
Otavio Salvador
Anders Darander
Lauri Hintsala
Christopher Larson
Jessica Zhang
Saul Wold
Dmitry Eremin-Solenikov