It's common to provide a non-machine-suffixed link in DEPLOY_DIR_IMAGE, so
let's be consistent and do so here as well.
(From OE-Core rev: c1c8d2f3cffc540380c0a5fcdda48d64cbec333a)
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If you point to snapshot.debian.net/archive/pool then it will fetch
you a html page which will end up in corrupt download. The locations
have changed for archives and here we point the mirror to right
location.
(From OE-Core rev: 15e8aa4cd773022b36a66f70e4f6436289f3d1bc)
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The fixmepath file that is generated contains a list of all of the files
that need their paths fixed. In the previous version the fixmepath was
generated to include all of the files that sed may have changed. In the
new version, we first grep the files to see if they contain a path that
needs to be changed, only then do we perform the sed operation on those files.
This results in a modest performance increate in the creation of the sstate
file. The following numbers include the do_package and do_populate_sysroot
tasks on the perl recipe.
Before the change:
real 4m23.018s
user 1m57.067s
sys 1m33.327s
After the change:
real 4m13.083s
user 1m54.062s
sys 1m26.064s
However, a more significnt performance gain is felt during the
extraction/install of sstate cache files, as the fixmepaths file now has a
significantly smaller list of files to modify.
Before the change:
real 0m39.798s
user 0m11.158s
sys 0m12.642s
After the change:
real 0m25.511s
user 0m8.408s
sys 0m5.077s
(All numbers above were recorded with a cold filesystem cache on a machine
with 12 GB of ram.)
(From OE-Core rev: 46067264bedeff8248a2b2441733420fe6651f84)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If a + character appears in a version specification within the list of
package dependencies, the version will not be removed from the list in
list_package_depends/recommends leading to garbage appearing in the
dependency graphs generated by buildhistory. To avoid any future
problems due to unusual characters appearing in versions, change the
regex to match almost any character.
Fixes [YOCTO #2451].
(From OE-Core rev: 36d1717e2ad4ca1620ee9f01b524b5ff2f499b26)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
As the comment says, using packages[0] is rather broken and can be incorrect,
as demonstrated by the recent gdk-pixbuf change. Replacing it with PN is
a bit more correct and more likely to do what was originally intended
by this function.
(From OE-Core rev: 5d0144fe6bd1febc2d1e651bbcff891c80f20b88)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
One of the patch backends to linux-yocto is guilt, which normally
tracks patches under .git. But .git isn't something that can be
checked into a SCM and repeated. So it has been moved under meta/patches
and committed to the meta branch.
If devshell is used, GUILT_BASE isn't set, so patch manipulations will
fail. We export GUILT_BASE and point it at the meta directory when
devshell is invoked for linux-yocto.
(From OE-Core rev: 10546e42a0db6a9585488536e29759239ef6e73b)
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The usage of this function renaming and it being called using
bb.build.exec_func() causes needless indirection loops, confusing log
files and seems generally pointless.
This simplification makes the process much simpler and faster. I can't
come up with a good reason why the export_functions functionality is
needed for this function.
(From OE-Core rev: 74daad03ca29a03b0005f7d2b90a0347d5b583a5)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In a pathological case, lots of files to process, the sstate_installpkg
performance was very poor. It interated over each file and ran 3
individual sed commands per file. Changing this to keep iterating
but running only a single command took about 1/3 time time.
However, when looking at the corresponding sstate_hardcode_path
function, it was clear we could optimize this further.
Using the same encoding logic to specify only the minimumal sed
operation necessary, and using xargs to avoid the os.system call the
install step was able to be performed in 13% of the original time.
Example timing numbers for perl:
3m7s original code
1m20s single sed, but interating
0m26s using xargs and limited sed
(From OE-Core rev: d9f655753fbdc8cbd8e705577430fed4f23732b3)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Empty components in $PATH have the same effect as a . in $PATH,
and are a common side-effect of inserting a misspelled or unset
shell variable in $PATH.
(From OE-Core rev: ac4201b714c83c614113bfa735d0a2fc6f64db99)
Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Revise the handling from ANGSTROM_BLACKLIST to PNBLACKLIST[pn].
Refactor the code to eliminate references to the distribution and recipe
name in the message.
Change the skipPackage message message from:
ERROR: <recipe> was skipped: <distro> DOES NOT support <recipe> because <reason>
to:
ERROR: <recipe> was skipped: Recipe is blacklisted: <reason>
(From OE-Core rev: 0893457465e184550906844fbab2dec397201835)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We need to add MLPREFIX when set RDEPENDS by setVar() function.
(From OE-Core rev: 9407895c88e46490d0ef9bb185f73466e7204c15)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This introduces a sanity check for the toolchain, which verifies
each tuning (including any multilibs), producing meaningful diagnostics
for problems, and also provides some higher-level tuning features.
The TUNEVALID and TUNECONFLICT/TUNECONFLICTS settings were not
implemented. Listed one or two missing features in TUNEVALID,
also (in a previous patch) fixed the references to
features which didn't exist.
This patch also provides a whitelisting mechanism (which is completely
unused) to allow vendors providing prebuilt toolchain components to
restrict tunings to those based on or compatible with a particular ABI.
(From OE-Core rev: 2a91ff0ba0d587c516a5a972553280364853faa4)
Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In order to support repositories of various types (with or without
meta data, branched, pristine, custom, etc) information about the
type of processing that is required was passed to the processing
phases via variables.
The combination of variables involved in coordinating the processing
creates a learning curve and overly complicates recipe extensions.
With minor tweaks to the kern-tools, adding flexibility and keying
off the existence of the meta branch it is possible to remove all
of the variables that were added to support different repository
types.
(From OE-Core rev: 06e5f45c8f38925cd5902a3a3f436f5e9451dd16)
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Some versions of the screen utility provided from the host OS vendor
write the socket directory to $HOME/.screen. When using a shared home
directory across many servers, one sets the SCREENDIR environment
variable to avoid collisions in the shared home directory. This
results in problems launching a devshell where it is not entirely
obvious what happened because the SCREENDIR environment variable
got stripped from the environment prior to setting up the screen
in detached mode.
Example:
% bitbake -c devshell busybox
# ...Please connect in another terminal with "screen -r devshell"
% screen -r devshell
There is no screen to be resumed matching devshell.
The temporary work around was to do something like:
sh -c "unset SCREENDIR; screen -r devshell"
This patch adds SCREENDIR to the white list to ensure screen
works properly on systems where a developer needs to use
the SCREENDIR with shared home directories.
(From OE-Core rev: 5568a8f5a1c65bae021b2e36d735d3153acc6d72)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The echo line here was merging multiple lines into one, and the result
was that if both image and package changes had to be comitted then only
the image changes were being committed and the package changes could
potentially be merged into the next package change. Quoting the variable
reference fixes this.
Fixes [YOCTO #2411]
(From OE-Core rev: 2086bb86885951d0a74342e30ff205c24cb93f0d)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If KERNEL_IMAGETYPE is vmlinux, the expectation is most likely that there will
be no kernel image package, but we still want a vmlinux package for debugging,
so move kernel-vmlinux in front of kernel-image in PACKAGES.
(From OE-Core rev: b62a44a9bf7b268296aeed64469390aa7efc2e73)
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There were a couple problems with the handling of precompiled locales.
- it gathered the list of locales from the directories - this breaks due to
the naming mismatch, e.g. en_US.UTF-8 vs en_US.utf8.
- it retained its hardcoded assumption that the non-suffixed locale (en_US, as
opposed to en_US.*) is UTF-8, while the others are otherwise. Hardcoding
this is both inflexible and just plain wrong for some toolchains. It's most
common in desktop distros for 'en_US' to be non-utf8, and ''en_US.UTF-8' is
utf8, and this is the case in some external toolchains as well.
The code now uses the SUPPORTED file to hold the knowledge it needs. This file
not only holds the list of locales to generate, but also maps the locale names
to the charsets they correspond to. The code now uses this to assemble its
charset map, falling back to the '.' suffix as charset when the locale is not
in the map. For precompiled, it now uses the locale->charset knowledge it has,
thereby allowing non-utf8 non-suffixed locale names, whereas for
non-precompiled, it reverts to the previous assumption, renaming the utf8
locale and forcibly suffixing the others.
So, a person maintaining an external toolchain recipe is responsible for
ensuring that the SUPPORTED file they provide matches up with the compiled
locales in the toolchain, if they want to utilize precompiled locales.
I believe in the long term the compiled case should do the same thing
precompiled does, and use SUPPORTED or a similar mechanism to encode the
knowledge, and if people want all the non-suffixed names to be utf8, they can
change that file to do so. This would avoid the hardcoded assumption in the
code, as well as consolidating the behavior between the compiled and
precompiled cases.
(From OE-Core rev: 3f36058923ccda25a3dd85046542e65b6034c09e)
Signed-off-by: Christopher Larson <kergoth@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This change farms the solvedb creation out to a separate script which
handles creation of the index, only if mtime of any of the packages
has changed.
For a core-image-minimal set of rpm's this saves ~20s of a 45s rootfs
build. For core-image-sato it saves 1 minute of a 5 minute rootfs build.
The more packages in the system, the bigger the saving will be.
(From OE-Core rev: 3021136e7b42ab64ca16f30c88467c4b00d51ee0)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The existing shell script is a fork bomb and forks off hundreds of
grep/cur/wc calls as it reads from its input stream and iterates over
the file data table for each line of input. This patch replaces the
shell code with python code which doesn't exec anything and hence runs
much faster without the exec() overhead. This speeds up rpm packaging
considerably, as can be measured simply by timing it, or watching the
processor utilisation.
(From OE-Core rev: 2fb5f72b47840ed7f540df181595878e3f5ba4c6)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We have currently no override to detect a recipe being build cross, crosssdk
or for target at times we can use virtclass-native and virtclass-nativesdk to
override stuff in recipes but we dont have way to modify a variables
based on recipe type always.
This patch adds in such an override and in particular makes a target override
class available.
With this change now we can say:
EXTRA_OECONF_class-target = "...."
EXTRA_OECONF_class-native = "..."
EXTRA_OECONF_class-nativesdk = "..."
EXTRA_OECONF_class-crosssdk= "..."
Based of an original patch by Khem Raj
(From OE-Core rev: cf332fd9bf685f6d42b11c1f0c37b934c7f5bcbe)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* depmod already gets executed by pkg_postinst_kernel-image.
* If you build a module using module.bbclass, pkg_postinst returns 1 in
do_rootfs, causing pkg_postinst to run again on first boot. To improve
this situation, I copied pkg_postinst from kernel.bbclass to module.bbclass.
This was rejected by Koen, because he doesn't like the code from
kernel.bblcass, which uses ${STAGING_DIR_KERNEL}. Richard then suggested
that calling depmod during do_rootfs wasn't necessary at all, because
it already gets done by kernel-image.
(From OE-Core rev: c7809c03080925b5e9171df5c9175c7c6420b376)
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* indentation was with spaces and tabs, unify to use tabs instead of
spaces, for shell code and populate_packages_preppend,
because "python populate_packages" expects tabs (or 8 spaces)
* and use 4 spaces for anonymous python
(From OE-Core rev: 24472a794e1787ebcd700dde66018a31069aa50b)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* allows to detect ENTRYPOINT from kernel binary marked with UBOOT_ENTRYSYMBOL
used e.g. by ben-nanonote
(From OE-Core rev: 9ce6bd1d58b4418d3a53be417c18a0e57314462a)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For example, pseudo-native wants to set LD_LIBRBARY_PATH but setting this
into the environment here causes the existing pseudo (running during do_install)
to poke into paths in /opt and this breaks builds.
The simplest fix is simply not to do this. Comments tweaks to match the code.
(From OE-Core rev: 1efa999ed02186c9c58cb48418e0c62ecdc82ac7)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This allows iso images to be written to usb keys and booted on systems which
have a BIOS which support this. There is no real down side to tweaking the iso
images in this way.
[YOCTO #1763 partial]
(From OE-Core rev: 9a946554e4843841c2ec3cbb2575e80d0ccdc29f)
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Remove /etc since it is empty, when creating a machine that does not
deliver any module config files, the /etc is empty and is then warned
about not being shipped, so we remove it.
This occurs in the routerstationpro with the following warning:
WARNING: For recipe linux-yocto, the following files/directories were installed but not shipped in any package:
WARNING: /etc
(From OE-Core rev: 963742f5a9bce2dfedd58350af754d256e2211cb)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
patch.bbclass orignally pointed at /usr/bin/quiltrc for an empty
version to ensure that no user setting were picked up, change this
to /etc/quiltrc in the Native sysroot since we now have a native
sysconfdir.
Make sure that the quiltrc is actually installed in the Native
sysconfdir, not the target, so fix this after the recipe split.
(From OE-Core rev: 1af73900cea82e63fb0f94e6f057144f723146ec)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Judge if "SanityCheck" event is received, it will issue the
sanity_check() and send "SanityCheckPassed" back if succeeded.
(From OE-Core rev: 19704f9e69ecf09531687385b478b47f49fe372d)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It's useful for Hob to be able to disable the sanity checks completely
without marking them as passed so that the user can get into the GUI to
configure their settings, etc.
Add a variable, DISABLE_SANITY_CHECKS, to do so.
(From OE-Core rev: b022641f939bcfcdaddddc4db3af4d2dc70de832)
Signed-off-by: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Kernel modules are not marked as executable but we do expect to strip them.
This patch adds in missing code to ensure we do this. Without this images
are getting sigificantly bloated in size.
(From OE-Core rev: 00b0a5f2f51bb3f88bbb9ae558c2859e3c1c406c)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In particular, this overwrites the value from cross-canadian.bbclass in
some cases which isn't the desired behaviour and unnecessarily
complicates/breaks the dependency chain.
(From OE-Core rev: 751ead4fa7d4120de906a1d9cb1d5a29357bebad)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For beagleboard platform, the PACKAGE_ARCH for certain recipes is
"armv7a-vfp-neon", however, the architecture label in RPM file name is
"armv7a" due to a potential bug in RPM backend.
This commit is a workaround to make Hob work in this case.
(From OE-Core rev: 00bd4589b21fe3a716ff2732ea55651c4abd77e4)
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We call re.sub here, so we need to "import re" or an error occurs
(tested on OpenSuSE 12.1).
(From OE-Core rev: cb1f7cffc171e6b182f33a69ff688d76b7f1baed)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Any trailing whitespace needs to be stripped before comparing it to the
SANITY_TESTED_DISTROS list.
(From OE-Core rev: bea0311a6478febe8ab4884fb1c479f610856534)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In fixing an existing patch migration bug, the list of valid extensions
got out of sync from the core patch class. As a result, valid patches
were not being applied to the tree.
Updating the tools to migrate .diff files fixes the issue.
Also in this fix is the removal of .patch in the find_sccs() routine, since
it will never be returned by patch.bbclass when all non-patches are
requested, it is simply confusing.
(From OE-Core rev: 0ade3f26f40b67d7296725b1e956c46be9a86089)
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If the sanity tests fail, we still were writing out the stamp which means
they'd get skipped the next time we run bitbake. This is clearly wrong
and we should only write out the stamp file if the sanity tests complete
successfully.
(From OE-Core rev: b99acaa8aa06dcbca46c3f7048124d6f8d643a8b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
MLPREFIX is needed in RDEPENDS for multilib build
(From OE-Core rev: 380efadd5640e5b57e710549b1ae761d0d3b3792)
Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The value of ALL_MULTILIB_PACKAGE_ARCHS needs to be consistent both
in multilib extended recipes and in normal context. If this isn't the
case it can lead to inconsistent configuration files at a minimum.
This patch ensures the value is preserved during the class extension code
since computing it after that point is hard.
[YOCTO #2290]
(From OE-Core rev: 529bc145152bb36d9696226b93729377a3b6a240)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The PN == BPN test was designed to exclude native/nativesdk but also unfortunately
excludes multilib unintentionally. This change fixes this and allows multilib images
with alternatives to funciton correctly.
[YOCTO #2214]
(From OE-Core rev: 2878d19c4e19d76cb8be15825c40ff03f25f35f5)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
During testing/extension of the linux-yocto-tiny kernel it was found that
defconfigs were not always properly applied. This was due to two issues:
- not being able to fully control the order of objects applied to the
git tree on the SRC_URI
- defconfigs triggering --allnoconfig before being applied
To fix this, the recipe space code that previously detected and generated
automatic features moves back to the kernel tools (where it was before) and
is updated to also process .cfg and defconfigs. Moving this back to the
tools allow other recipes to automatically benefit from the additional
support.
The second issue is addressed by allowing configme to take --alldefconfig
when a recipe wishes to pass a defconfig and override the default
behaviour.
Fixes [YOCTO: 2250]
(From OE-Core rev: 08c368b9980716e459d846dd7183940a5bf8bea4)
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch fixes up the issues that were being seen where BBCLASSEXTEND and
PACKAGECONFIG were interacting badly. It also ensures PACKAGECONFIG interacts
properly with multilib builds.
Ideally some of this code will be abstracted into lib/oe/classextend.py but
at this point in release more invasive changes like this are inappropriate.
This patch also removed empty strings from expressions rather than
passing them around as this was complicating the additional code
unnecessarily.
The patch was verified against the OE-Core metadata where the return values of
expandFilter() were sanity checked by hand for native/nativesdk and
multilib combinations.
[YOCTO #2225]
(From OE-Core rev: 46db11c4a789034b7040faf127ab865148bedad8)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If the rootfs has a specific layout in mind for /var/tmp, it should be
free to create this. Preinst scripts trigger rpm to create its tmppath
directory which could potentially conflict with this.
We're seeing issues if the provider of the /var/tmp wants to use a symlink
and a preinst script executes before it has been installed. This change
places tmppath as used by rpm somewhere safe which is already automatically
cleaned up too.
(From OE-Core rev: 71e95c744eaa4dda1b3237db2e13f666f121c92b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
