It turns out that changing WARN_QA and ERROR_QA results in do_configure's QA
postfunc re-executing, so changing a QA test results in a complete rebuild.
This is just too much and the lesser evil of needing to do a full rebuild to
verify changed QA flags is preferable to an enforced full rebuild.
This reverts commit daecfc3438122b5d146a59a5053e57006d55ccc4.
(From OE-Core rev: 4c5895da16de6f00148a0755b421f07223083d09)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changing WARN_QA and ERROR_QA should cause do_package_qa to re-execute, so
removing them from the sstate hashes is harmful.
They were added back when sanity testing was part of packaging and this was the
lesser evil, compared to changing sanity tests causing a re-package of
everything.
(From OE-Core rev: daecfc3438122b5d146a59a5053e57006d55ccc4)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In meta/conf/bitbake.conf, PKGDATA_DIR is default to:
PKGDATA_DIR = "${STAGING_DIR_HOST}/pkgdata"
But in meta/conf/multilib.conf, PKGDATA_DIR is set as:
PKGDATA_DIR = "${STAGING_DIR}/${MACHINE}/pkgdata"
When multilib enabled, linux-libc-headers cache will be machine
specific:
$ bitbake-diffsigs sstate-cache/1a/sstate:linux-libc-headers:ppce6500-poky-linux:3.17.7:r0:ppce6500:3:1a0c3934d91479fd7242a5b1d407d155_package.tgz.siginfo sstate-cache/28/sstate:linux-libc-headers:ppce6500-poky-linux:3.17.7:r0:ppce6500:3:28c918e8f9f4a4cfceb3a38b258f7501_package.tgz.siginfo
basehash changed from 8d3158bbddcee612fa30badd05f47b8e to 68ac258fc6c8e489f360fde3123a5894
Variable MACHINE value changed from 'b4420qds' to 'b4860qds'
(From OE-Core rev: c511f65a3ccfcbaabd2ba1d1c89be81498240a2b)
Signed-off-by: Ting Liu <ting.liu@freescale.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
[YOCTO #7230]
In certain system configurations TRANSLATED_TARGET_ARCH will not
expand in the right order for gcc-cross-candian-mips64n32 to be
generated properly.
This will cause SDKs to fail to generate properly.
Changing the global definition of TRANSLATED_TARGET_ARCH always
expands the ABIEXTENSION, which causes the OVERRIDES to pick it up
as well. This effectively defines a new class of overrides for the 'n32'.
The side effect is that we need to duplicate some mips64 overrides, and
redefine others that were previously 'n32' or 'mips64' exclusive to have
the correct semantics.
(From OE-Core rev: 4b3a2b703b20583bd107f00a297d972e9bfb514a)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
commit 3b3f7e785e279 [kernel: Rearrange for 1.8] began the process of
moving the kernel source and build artefacts out of sstate control and
into a shared location.
This changed triggered some workflow issues, as well as bugs related
to the kernel source containing build output, and hence being dirty and
breaking kernel rebuilds.
To solve these issues, and to make it clear that the kernel is not under
sstate control, we move the source and build outputs to:
work-shared/MACHINE/kernel-source
work-shared/MACHINE/kernel-build-artifacts
Where kernel-build-artifacts is the kernel build output and
kernel-source is kept "pristine". The build-artifacts contain everything
that is required to build external modules against the kernel source,
and includes the defconfig, the kernel-abiversion, System.map files and
output from "make scripts".
External module builds should either pass O= on the command line, or
set KBUILD_OUTPUT to point to the build-artifacts. module-base.bbclass
takes care of setting KBUILD_OUTPUT, so most existing external module
recipes are transparently adapted to the new source/build layout.
recipes that depend on the kernel source must have a depedency on the
do_shared_workdir task:
do_configure[depends] += "virtual/kernel:do_shared_workdir"
With this dependency added, the STAGING_KERNEL_DIR will be populated and
available to the rest of the build.
(From OE-Core rev: 6a1ff0e7eacef595738f2fed086986fd622ec32a)
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
As the code that uses BB_SIGNATURE_EXCLUDE_FLAGS uses d.getVarFlags() so doesn't
get to see the internal flags, remove _append and _prepend. Also defaultval is
now _defaultval and thus internal, so remove that too.
(From OE-Core rev: b53e06c8fc4a8183a2f8232c13931a39b1ca0e23)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Recipe packaging for the target requires permissions that are consistent
with meta/files/fs-perms.txt which specifies certain user and group
names. In the early parts of a target build base-passwd is not yet
available to provide the target /etc files used for user/group lookup.
Allow pseudo to fall-back to the last-resort files it installs if the
target ones aren't there yet.
(From OE-Core rev: 071d364b7a758ba5e546bb18c5816ac4c2e6747c)
Signed-off-by: Peter A. Bigot <pab@pabigot.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Instead of replicating the logic for the host compiler naming from bitbake.conf,
use the BUILD_* variables directly.
Also change BUILD_CPP to use gcc -E (which native.bbclass previously used), as
some recipes (e.g. grub-efi) use ${CPP} with multiple input files, which gcc -E
can handle but cpp can't.
(From OE-Core rev: 9237d18964ff0bf76f5c37fca21ab3974d81d0d2)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Previously, when building core-image-minimal, the rootfs size would
default to 64M because we use '?=' in bitbake.conf and also '?=' in
core-image-minimal.bb.
The thing is, we'd like to have a default value for all images set
in bitbake.conf but still allow each image recipe to set its own default
value which could be overridden by users in local.conf.
(From OE-Core rev: 18f499df6bcbf79d7bd0a99c4c8693268683485f)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
glibc 2.20+ wont support any older than that
(From OE-Core rev: 32b3a9ca554d9ff8f3b9c2ff62cc66ee865c61bf)
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Might as well move this default to the class which uses it allowing
for easier reading/understanding of the class.
(From OE-Core rev: 177aec177306e68bcd822dee6b29a7efbd558a91)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently its near impossible for other classes to sanely override
this value with their own default. By setting a weak default we can
allow other classes to change the default and allow end recipes to
again override this.
As far as I can tell, there shouldn't be any regressions from this
change.
(From OE-Core rev: 12b2a73d336d66596939eae5c9947d4054c0316e)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
A few native scipts requires bash-native, and we don't build
bash-native, so add it to ASSUME_PROVIDED.
(From OE-Core rev: 283a418a838ef285988a5ffc3888501ca7de63f1)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When adding patches or config files from bbappend files, it requires
the use of FILESEXTRAPATHS, which has been an issue and failure point
for people starting to work with bitbake and oe-core.
We add checking to standardize how to use FILESEXTRAPATHS. Only the
format of:
FILESEXTRAPATHS_append := ":${THISDIR}/Your_Files_Path" or
FILESEXTRAPATHS_prepend := "${THISDIR}/Your_Files_Path:"
is acceptable.
[YOCTO #5412]
(From OE-Core rev: 69e083237e632f7d84a7b218dd12d1a5ad95a229)
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently, BB_NUMBER_THREADS and PARALLEL_MAKE default to unset and
are set in local.conf. Now that we have the automatic probing,
the default values can be set in bitbake.conf and an example of
explicitly defining how many tasks to run can be moved to
local.conf.sample.extended.
[YOCTO #6217]
Signed-off-by: Roxana Ciobanu <roxana.ciobanu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* Check the rootfs size against IMAGE_ROOTFS_MAXSIZE (if set)
* Add comments for IMAGE_ROOTFS_SIZE to not confuse with IMAGE_ROOTFS_MAXSIZE
[YOCTO #2610]
(From OE-Core rev: 6acd4fc8d5e642b5c6c75fcc40dd8f37caf7ddcf)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
pkglibexecdir is a fairly common location for package-specific binaries (in
automake this is $libexecdir/$PACKAGE), and binaries in there are already
installed to FILES_PN, so add the corresponding .debug directory to
FILES_PN-dbg.
(From OE-Core rev: 4d3ffde4649ed116a1c21afef41f71bfe1d471de)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For most users this commit will have no effect. But if you come across the idea of giving
different names for paths, you'll get some troubles.
When a recipe inherit native, properly define bindir, sbindir, includedir, sysconfdir, datadir
(using xxxdir_native definitions from meta/conf/bitbake.conf).
For example, edit "${BASE_WORKDIR}/${MULTIMACH_TARGET_SYS}/quilt-native/temp/log.do_configure"
and see what are the arguments given by oe_runconf.
Notice that ${docdir}, ${mandir}, ${infodir}, ${localstatedir} have no associated _native definition.
(From OE-Core rev: 15345ddd4be6a0b041b3d6caaad48d46b22142e9)
Signed-off-by: Matthieu Crapet <Matthieu.Crapet@ingenico.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The QEMU_OPTIONS variables belong in qemu.bbclass so move them there. The
only users of them inherit qemu.bbclass. There is no point in pushing
these into every recipe.
(From OE-Core rev: 5824293de37919e89f60192836997281933e23d6)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently, qemuppc prints warnings about gdk-pixbuf postinstalls
not working due to illegal instructions. This is due to qemu
running with the wrong cpu type. Add an option for ppc7400 so
that qemuppc works correctly.
(From OE-Core rev: 5995fdbe81799f1ecf5de722cb2eb95ccb2aa860)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
To unpack that to more than a single line: -native and -cross recipes are made
to use the dummy Texinfo utilities provided by texinfo-dummy-native if they
invoke those utilities at build time. The target-architecture (cross-compiled)
recipes still use the genuine Texinfo utilites. Right now, they still use
the host system's Texinfo utilities, but could be made to use the
texinfo-native recipe we already ship with some config file changes.
(From OE-Core rev: 160087f754eabf5da90fb51997e19d2e585aac4a)
Signed-off-by: Max Eliaser <max.eliaser@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If we don't do this, we see an exception:
ERROR: Failure expanding variable MACHINE_ARCH, expression was ${@[d.getVar('TUNE_PKGARCH', True),
d.getVar('MACHINE', True)][bool(d.getVar('MACHINE', True))].replace('-', '_')} which triggered
exception AttributeError: 'NoneType' object has no attribute 'replace'
Setting a default value avoids this error and allows the sanity checker
to trigger instead.
(From OE-Core rev: 106e9a3f594658b6a207f1f29bd4007616cc31d6)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* The SRC_URI is not accessible.
So need to add mirror site referred by the original site.
* The problem is that
http://download.savannah.gnu.org/releases redirects to closest mirror
and few mirrors (e.g. .jp) weren't working correctly while
http://download-mirror.savannah.gnu.org/releases/ seems to be reliable.
* Add SAVANNAH_GNU_MIRROR and SAVANNAH_NONGNU_MIRROR variable in bitbake.conf.
* Change the SRC_URI using the new variable.
(From OE-Core rev: af00b6544f60e4d7581f9d9767f9d3f574392359)
Signed-off-by: Changhyeok Bae <changhyeok.bae@lge.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This allows them to co-exist together in the native sysroot, with one
set of cross tools per target architecture.
(From OE-Core rev: a2c5509520d5c3e082f55844e6545d0309565f8f)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The problem is that do_configure.sigdata depends on STAMPS_DIR because:
do_configure -> STAMPCLEAN -> STAMPS_DIR
this will make the sigdata generated by "STAMPS_DIR=/tmp/stps bitbake -S
recipe" doesn't match the ones in our build dir, but it should. We can
add STAMPS_DIR or STAMPCLEAN to BB_HASHBASE_WHITELIST to fix the
problem, but we can't add STAMPS_DIR since once it is in
BB_HASHBASE_WHITELIST, the "STAMPS_DIR=/tmp/stps bitbake -S recipe"
would not run again.
[YOCTO $6031]
(From OE-Core rev: faf3e74d5c488a66fdabd485eb916f555d7353fd)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We don't want the value of this varflag itself entering any signatures,
ever.
Part of the fix for [YOCTO #5897].
(From OE-Core rev: 1497d3d4b10844aa19ce6dcceed25aa36454160f)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Drop the -fpermissive C++ compiler flag. We've had this around for years, most
code should have been fixed long ago. Its possible some recipes may fail
however we can (and should) just use the flag where needed.
An OE-Core world build seems to work just fine with this change.
(From OE-Core rev: 24dd8e129447013ee98609f3892ec414b1b21340)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
These variables should not influence the config hash, i.e. changing them
shouldn't trigger a reparse of the metadata, so whitelist them.
(From OE-Core rev: 8feb51267647d0760f5bec3a8b6f95f4481d9b0d)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The default value for HOMEPAGE of "unknown" has been in place since the
early OE-Classic days, but it doesn't really make sense - "unknown" is
not a valid URL and it just means we have to explicitly check for this
hardcoded string if we're displaying the value in some form of UI, such
as Toaster.
This has required some changes to the packaging classes as they
previously did not expect the value to be blank.
(From OE-Core rev: 244e1d73ef58e92d73c098044c66bd784644b933)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
SUMMARY should not end with a full stop; however if DESCRIPTION is not
set in a recipe and thus defaulted from SUMMARY, the additional
DESCRIPTION values for other standard packages e.g. ${PN}-dev look a bit
odd without a full stop separating the SUMMARY value and the rest of the
text. Add a full stop to avoid this.
(From OE-Core rev: 4b022399815f32166c402d458a40afa6470fc776)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ensure that if MACHINE_FEATURES is not set by the machine config that we
don't end up with expansion errors during parsing. Technically since the
introduction of MACHINE_FEATURES_BACKFILL = "rtc" this is unlikely to be
a problem unless "rtc" is also added to
MACHINE_FEATURES_BACKFILL_CONSIDERED, however we should be consistent
with DISTRO_FEATURES which is defaulted in bitbake.conf.
(From OE-Core rev: bf2c8946d96524aaa91ab43762c963ea38ccc342)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The layout of stamp files ensures that changes to WORKDIR mean recipes get rebuilt correctly.
Since WORKDIR usually contains MULTIMACH_TARGET_SYS and that depends on tune variables,
including WORKDIR in sstate checksums adds a lot of noise to the system for what amounts to
no gain.
On the other hand, removing it reduces noise, reduces the size of the siginfo files and
reduces the amount of processing bitbake has to do. It therefore seems like dropping it
from the checksums is an all around win.
(From OE-Core rev: 453353e05d027c6a505d1e13a7982718a13bca8b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
gcc tooling appears to be standardising around the FC variable naming.
This patch changes the F77 namespace to FC instead and use the default
gfortran compiler. If anyone needs the F77 variables or tools, those
can still be made on a case by case basis.
Also updates local.conf.sample.extended accordingly.
(From OE-Core rev: ae8c17be2845eff2be8394a5d9a45e6aa321c33d)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
ADOBE_MIRROR, HANDHELDS_CVS and E_SVN were broken links and not used by
any recipe in oe-core.
FREEDESKTOP_CVS is no longer useful because all the source code that
matters is in git; no recipe in oe-core still uses the CVS repository.
E_MIRROR, FREEBSD_MIRROR, FREESMARTPHONE_GIT still point to valid-seeming
locations but there are no recipes in oe-core that use them. Any layers
which need these variables can define them for themselves.
GPE_SVN, GPE_EXTRA_SVN, GPEPHONE_MIRROR and GPEPHONE_SVN are not used by
any recipe in oe-core and the corresponding projects seem to be mostly
dead upstream. Again, any layers which still wish to use these variables
can define them locally.
All the above are just wasting space in bitbake's datastore and would be
better deleted.
(From OE-Core rev: 3b333896c71689c664475d53daed52404bf6b21b)
Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This variable has been unused since the tune file overhaul two years
ago.
(From OE-Core rev: a1d9f2374ede768057fd364da6c0e1eeeb10499f)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
These were for task-bootstrap in OE-Classic and have never been used in
OE-Core.
(From OE-Core rev: f4692afb518f07e17fbd35a2023877b7041abef9)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
An empty distro value leads to OVERRIDES and FILESOVERRIDES containing
"::" entries which causes odd issues such as files being included when
they shouldn't be. We could put in anonymous python to guard against
empty entries but its messy and setting a default value for DISTRO to
something harmless is much easier.
This patch adds a weak default and ensures the sanity test doesn't
complain about it.
DISTRO_VERSION and SDK_VERSION are also updated to match.
(From OE-Core rev: b7279f99639774674da806d37d252f388f33055f)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
I discovered bitbake rebuilding packages because WARN_QA had changed. These
variables don't influence the output, so add them to the whitelist.
(From OE-Core rev: 96204ae6e1b19783d6a3f8c590890714eaa9e2d9)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* we correctly have
${STAGING_DIR_NATIVE}${base_sbindir_native}
and then double slash in
${STAGING_DIR_NATIVE}/${base_bindir_native}
* similar in PKG_CONFIG_DIR where libdir also starts with slash
${STAGING_DIR_HOST}/${libdir}/pkgconfig
* also fix double slash in insane.bbclass and staging.bbclass
* I was a bit nervous about staging change (in case the / was important
in some weird use-case, but the extra slash is there since following
commit where other extra slashes were removed only the one before
libdir was kept:
commit 6ea78d6489
Author: Richard Purdie <rpurdie@linux.intel.com>
Date: Mon Nov 2 17:10:51 2009 +0000
autotools.bbclass: Separate out useful staging functions into
base.bbclass and call from autotools classes
* this isn't fixing any real-world issue AFAIK, I was just trying to
debug one weird case where debugedit fails with
canonicalization unexpectedly shrank by one character
and it's easier to grep for '//' without many harmless instances
already in run* scripts etc
(From OE-Core rev: 0ddaf52e9e344986ae2b016cc068d9eee71b4347)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
To make it easier to move WORKDIR, define it using the new variable
BASE_WORKDIR, which is the root of the work directory.
(From OE-Core rev: 1eee097f2e29b9d6934711c0b1d32e59e9542f53)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently we have a hierarchy of pkgdata directories and the code has to put together
a search path and look through each in turn until it finds the data it needs.
This has lead to a number of hardcoded paths and file globing which
is unpredictable and undesirable. Worse, certain tricks that should be
easy like a GL specific package architecture become problematic with the
curretn search paths.
With the modern sstate code, we can do better and construct a single pkgdata
directory for each machine in just the same way as we do for the sysroot. This
is already tried and well tested. With such a single directory, all the code that
iterated through multiple pkgdata directories and simply be removed and give
a significant simplification of the code. Even existing build directories adapt
to the change well since the package contents doesn't change, just the location
they're installed to and the stamp for them.
The only complication is the we need a different shlibs directory for each
multilib. These are only used by package.bbclass and the simple fix is to
add MLPREFIX to the shlib directory name. This means the multilib packages will
repackage and the sstate checksum will change but an existing build directory
will adapt to the changes safely.
It is close to release however I believe the benefits this patch give us
are worth consideration for inclusion and give us more options for dealing
with problems like the GL one. It also sets the ground work well for
shlibs improvements in 1.6.
(From OE-Core rev: 1b8e4abd2d9c0901d38d89d0f944fe1ffd019379)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This allows a clean seperation between image outputs from different
machines, and makes it possible to have convenience symlinks to make
the output ready to deploy.
This did require some surgery in runqemu; if explicit paths to the image
and kernel are not supplied then DEPLOY_DIR_IMAGE needs to be determined
from bitbake or set in the environment. However the script does try to
avoid requiring it unless it really is needed. Corresponding changes
were made in the automated testing code as well.
Based on an RFC patch by Koen Kooi <koen@dominion.thruhere.net>
(From OE-Core rev: 7e90261aec61f79680b5eaeaf5b18c7b795412a4)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For a long time we've provided PN-PV and PN-PV-PR by tweaking PROVIDES. This looks
nice at first glance however it turns out to be a bit problematic. Taking make as an
example where there are two versions, 3.81 and 3.82, what should "bitbake make-3.81" do?
Currently it builds make-3.81 and make-3.82 and breaks in interesting ways. Is that
a bitbake bug? Well, it certainly shouldn't try and run the build. Why is it building
3.82 though? Its due to finding a dependency on "make-dev" and then trying to figure
out what provides it? The answer is "make" and the default version of "make" is 3.82.
So arguably, finding "make-3.81" should infer PREFERRED_VERSION_make = "3.81". Doing
so resolved the above problem since now "make" resolves to "make-3.81".
So what about if we have Recipe A:
DEPENDS = "make-3.81"
and Recipe B:
DEPENDS = "make-3.82"
That is clearly an error, easy. So finally what about if we have Recipe A:
DEPENDS = "make-3.81"
and Recipe B:
DEPENDS = "make"
The first recipe infers the PREFERRED_VERSION_make = "3.81" and then forces that
version on everything else. Is that desired? Probably not in most cases, at least not
silently.
As mitigation, we could print a WARNING about this happening. The final part of the problem
is that we can ony figure this out within bitbake itself. That means we'd have to teach bitbake
about the PN-PV format of PROVIDES which is breaking the separation between bitbake and the
metadata. We can't win :(.
Nobody that I know of is using or relying on this functionality so perhaps we should
just remove it instead which is what this patch does. Opinions?
(From OE-Core rev: a87c205bb6cefd5e1a41b8e7ef02b5bfa380e3b6)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The gcc recipes reference this however we account for it in the work
directory paths and we don't want recipes depending on the value changing.
This avoids unecessary rebuilds when switching SDKs.
(From OE-Core rev: 6cdcc543ce8f532a4f66246114241b43821a111e)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There were darwin8/darwin9 overrides spinkled in the code from times gone
by. Lets settle on the darwin override and remove the others since its pointless
duplication. We always inject darwin into OVERRIDES if needed in the darwin8/9
cases.
(From OE-Core rev: 8d5e6eed7802a6056f9eaa50a85e3eee00fe2742)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
On darwin, we have:
libxxx.dylib -> libxxx.Y.dylib
compared to Linux which has:
libxxx.so -> libxxx.so.Y
Our ordering of PACKAGES with -dev first and then ${PN} makes it impossible to
match the files correctly using simple globbing. This makes darwin targets
completely broken since both the libs and the dev symlinks end up in ${PN}-dev.
Whilst this commit is a hack, it at least puts the files into ${PN} and allows the
builds to be used. Symlinks don't take up much space so this isn't the end of
the world. I'm open to better solutions to this.
(From OE-Core rev: 51c3dbe2df45096bbd7866adabb08e114952ff13)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Hardcoding -nativesdk as the sdk package architecture is inflexible. We may have
multiple different target OS and we need a way to be able to separate them. Turning
this into a configurable value allows the flexibility we need to build different
SDKMACHINEs with different OS targets.
The commit should have no behaviour change, just makes things more configurable.
(From OE-Core rev: a2110e86b98d646e136de9ec6b8e668079b0d4f4)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
