Some TARGET_* flags are being exported in bitbake.conf currently, so
they are impacting all the tasks of a cross-canadian recipe even they
are not in use at all.
This can lead a lot of churn when the cross-canadian sysroot are shared
by machines while they have defined different TARGET_* flags.
Fix it by overriding with BUILDSDK_* flags.
(From OE-Core rev: 767335c92b7cc657a008722a908380a3c89c3c66)
Signed-off-by: Ming Liu <peter.x.liu@external.atlascopco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch is comparatively large and invasive. It does only do one thing, switching the
system to build using recipe specific sysroots and where changes could be isolated from it,
that has been done.
With the current single sysroot approach, its possible for software to find things which
aren't in their dependencies. This leads to a determinism problem and is a growing issue in
several of the market segments where OE makes sense. The way to solve this problem for OE is
to have seperate sysroots for each recipe and these will only contain the dependencies for
that recipe.
Its worth noting that this is not task specific sysroots and that OE's dependencies do vary
enormously by task. This did result in some implementation challenges. There is nothing stopping
the implementation of task specific sysroots at some later point based on this work but
that as deemed a bridge too far right now.
Implementation details:
* Rather than installing the sysroot artefacts into a combined sysroots, they are now placed in
TMPDIR/sysroot-components/PACKAGE_ARCH/PN.
* WORKDIR/recipe-sysroot and WORKDIR/recipe-sysroot-native are built by hardlinking in files
from the sysroot-component trees. These new directories are known as RECIPE_SYSROOT and
RECIPE_SYSROOT_NATIVE.
* This construction is primarily done by a new do_prepare_recipe_sysroot task which runs
before do_configure and consists of a call to the extend_recipe_sysroot function.
* Other tasks need things in the sysroot before/after this, e.g. do_patch needs quilt-native
and do_package_write_deb needs dpkg-native. The code therefore inspects the dependencies
for each task and adds extend_recipe_sysroot as a prefunc if it has populate_sysroot
dependencies.
* We have to do a search/replace 'fixme' operation on the files installed into the sysroot to
change hardcoded paths into the correct ones. We create a fixmepath file in the component
directory which lists the files which need this operation.
* Some files have "postinstall" commands which need to run against them, e.g. gdk-pixbuf each
time a new loader is added. These are handled by adding files in bindir with the name
prefixed by "postinst-" and are run in each sysroot as its created if they're present.
This did mean most sstate postinstalls have to be rewritten but there shouldn't be many of them.
* Since a recipe can have multiple tasks and these tasks can run against each other at the same
time we have to have a lock when we perform write operations against the sysroot. We also have
to maintain manifests of what we install against a task checksum of the dependency. If the
checksum changes, we remove its files and then add the new ones.
* The autotools logic for filtering the view of m4 files is no longer needed (and was the model
for the way extend_recipe_sysroot works).
* For autotools, we used to build a combined m4 macros directory which had both the native and
target m4 files. We can no longer do this so we use the target sysroot as the default and add
the native sysroot as an extra backup include path. If we don't do this, we'd have to build
target pkg-config before we could built anything using pkg-config for example (ditto gettext).
Such dependencies would be painful so we haven't required that.
* PKDDATA_DIR was moved out the sysroot and works as before using sstate to build a hybrid copy
for each machine. The paths therefore changed, the behaviour did not.
* The ccache class had to be reworked to function with rss.
* The TCBOOTSTRAP sysroot for compiler bootstrap is no longer needed but the -initial data
does have to be filtered out from the main recipe sysroots. Putting "-initial" in a normal
recipe name therefore remains a bad idea.
* The logic in insane needed tweaks to deal with the new path layout, as did the debug source
file extraction code in package.bbclass.
* The logic in sstate.bbclass had to be rewritten since it previously only performed search and
replace on extracted sstate and we now need this to happen even if the compiled path was
"correct". This in theory could cause a mild performance issue but since the sysroot data
was the main data that needed this and we'd have to do it there regardless with rss, I've opted
just to change the way the class for everything. The built output used to build the sstate output
is now retained and installed rather than deleted.
* The search and replace logic used in sstate objects also seemed weak/incorrect and didn't hold
up against testing. This has been rewritten too. There are some assumptions made about paths, we
save the 'proper' search and replace operations to fixmepath.cmd but then ignore this. What is
here works but is a little hardcoded and an area for future improvement.
* In order to work with eSDK we need a way to build something that looks like the old style sysroot.
"bitbake build-sysroots" will construct such a sysroot based on everything in the components
directory that matches the current MACHINE. It will allow transition of external tools and can
built target or native variants or both. It also supports a clean task. I'd suggest not relying on
this for anything other than transitional purposes though. To see XXX in that sysroot, you'd have
to have built that in a previous bitbake invocation.
* pseudo is run out of its components directory. This is fine as its statically linked.
* The hacks for wayland to see allarch dependencies in the multilib case are no longer needed
and can be dropped.
* wic needed more extensive changes to work with rss and the fixes are in a separate commit series
* Various oe-selftest tweaks were needed since tests did assume the location to binaries and the
combined sysroot in several cases.
* Most missing dependencies this work found have been sent out as separate patches as they were found
but a few tweaks are still included here.
* A late addition is that extend_recipe_sysroot became multilib aware and able to populate multilib
sysroots. I had hoped not to have to add that complexity but the meta-environment recipe forced my
hand. That implementation can probably be neater but this is on the list of things to cleanup later
at this point.
In summary, the impact people will likely see after this change:
* Recipes may fail with missing dependencies, particularly native tools like gettext-native,
glib-2.0-native and libxml2.0-native. Some hosts have these installed and will mask these errors
* Any recipe/class using SSTATEPOSTINSTFUNCS will need that code rewriting into a postinst
* There was a separate patch series dealing with roots postinst native dependency issues. Any postinst
which expects native tools at rootfs time will need to mark that dependency with PACKAGE_WRITE_DEPS.
There could well be other issues. This has been tested repeatedly against our autobuilders and oe-selftest
and issues found have been fixed. We believe at least OE-Core is in good shape but that doesn't mean
we've found all the issues.
Also, the logging is a bit chatty at the moment. It does help if something goes wrong and goes to the
task logfiles, not the console so I've intentionally left this like that for now. We can turn it down
easily enough in due course.
(From OE-Core rev: 809746f56df4b91af014bf6a3f28997d6698ac78)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
getVar() now defaults to expanding by default, thus remove the True
option from getVar() calls with a regex search and replace.
Search made with the following regex: getVar ?\(( ?[^,()]*), True\)
(From OE-Core rev: 7c552996597faaee2fbee185b250c0ee30ea3b5f)
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Arm is unusual in that we force it to "linux-gnueabi" and "linux" doesn't
build. This was causing problems for multilib configurations which were assuming
"linux" was the default compiler rather than linux-gnueabi.
This change does two things, ensures symlinks are generated for linux-gnueabi
and also adapts the libgcc code to account for the difference on arm.
It still needs to immediately expand/save TARGET_VENDOR but we defer
deciding what TARGET_OS should be until we know TARGET_ARCH (which the
multilib code may change).
[YOCTO #8642]
Note that sanity tests of a 32 bit arm multilib still break due to issues
with the kernel headers on a mixed bit system. This looks to be a general
headers issue for the platform though and a different type of bug.
(From OE-Core rev: bcddc3e7eff138add031bc9c9728be5a42fa62ef)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
By default the system will expand the extra os entries for uclibc and musl
even if they are not enabled in the build. There was no way to prevent this
behavior while still getting the expansion for things like x32 or spe.
The change adds a new setting which a distribution creator can override
easily, setting the base set of canadianextraos components. The other
expansions are then based on this setting.
(From OE-Core rev: ea24d69fdf7ebbd7f2d9811cff8a77bffc19a75c)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We build SDKs such that gcc-cross-candian is built for only one
target *-*-linux and then use -muclibc or -mmusl to let it compile
code for other libc variants. This works fine when libc = glibc
however it does not work for c++ programs when libc != glibc since
there are c++ headers installed under ${includedir}/c++/${BINV}/${TARGET_SYS}
which is fine when gcc-runtime and gcc-cross-candian uses same --target options
gxx includedir searches in right triplet, but it fails with musl/uclibc
since gcc will look for glibc based triplet but gcc-runtime will install
them under musl/uclibc triplet.
This patch symlinks the musl/uclibc triplet to glibc triplet when libc != glibc
This fixes SDKs for musl/uclibc
(From OE-Core rev: fcaaabb401fffcda4db9a7d1f927a2a404e4776d)
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Have DPKG_ARCH set by directly invoking a mapping function, rather
than using an anonymous Python function modify the variable under
the hood, so we can have proper handling of overrides.
Also bring in some additional mappings to Debian architecture names
that weren't being handled.
(From OE-Core rev: 8d042ea4e755cb0bb28b88333e10e04ec4e86a36)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
With the autodebug package generation logic, specifically setting FILES_${PN}-dbg
isn't needed in most cases, we can remove them.
(From OE-Core rev: 3ab59d49dd7c18e194b58d1248b4b87709b5a738)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If building for a BE8 ARM target, arch is "armeb" rather than
"arm", but ABI should still be "gnueabi". Otherwise gcc won't
build.
(From OE-Core rev: d2e1cf176b2a705f3e4bd4ab7c35bb1de5dc6985)
Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
"n32" is a mips64 variant we need to consider when processing the TARGET_OS
extensions. Also add the multilib extensions for mips64.
(From OE-Core rev: fe26f809aaad5d5d608e841c99b817316c5a59a0)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In the same way we map various TARGET_OS options back to the single
cross-canadian compiler, add mappings for the TARGET_VENDOR cases we
know about in the multilib case.
(From OE-Core rev: 753c98324ae82a67104eaf36e7ebf3553ee1dad7)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
gcc-cross-canadian-<arch> is only built once. It needs to target all the
different libcs, not just the currently selected one. This change
ensures that if another libc is used, symlinks are present such that the
compiler can be found. The base version is always assumed to be "glibc"
with symlinks from musl and uclibc compiler names.
This means gcc-cross-canadian is consistent regardless of which libc is
selected when its build in multimachine builds.
[YOCTO #8025]
(From OE-Core rev: 83ead626c0da75edec2833ffb1a29011ec7b83d2)
(From OE-Core rev: ee16cb0a3f40384f509083f0ef3f36b68f73b7cf)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There are some cases we want the manipulation cross-canadian performance
on TARGET_OS, there are also cases like meta-environment where we do not
want this manipulation.
We did try and use immediate expansion to avoid this problem and it
works in the non multilib case. If we have a multilib that used an
extension, like for example:
require conf/multilib.conf
MULTILIBS = "multilib:lib32 multilib:lib64"
DEFAULTTUNE = "mips32r2"
DEFAULTTUNE_virtclass-multilib-lib32 = "mips64-n32"
DEFAULTTUNE_virtclass-multilib-lib64 = "mips64"
then the n32 extension case will be misconfigured.
It turns out saving an unexpanded variable is hard. The best I could
come up with was:
SAVEDTOS := "${@d.getVar('TARGET_OS', False).replace("{", "*")}"
and then
localdata.setVar("TARGET_OS", d.getVar("SAVEDOS", False).replace('*','{'))
which is rather evil, I'd challenge someone to come up with a nicer way
of making it work though!
Rather than the above madness, we modify cross-canadian to make the
problamtic code conditional.
This fixes the original issue (where a linux-gnuspe target was seeing
'linux') of
http://cgit.openembedded.org/openembedded-core/commit/?id=0038634ee6e2b6035c023a2702547f20f67c103a
but also fixes the multilib one.
(From OE-Core rev: 85ff3d6491c54aa712ed238c561742cda4f4ba07)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Similarly to native/cross disable this since otherwise the packagedata
can be marked as machine specific and if you switch machines
which share an architecture, you'll get toolchain overlapping files
errors.
(From OE-Core rev: 96d557be3dedd6aea6199b3d28fbb7f5549fad69)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
A while back we fixed the cross definitions to work better in multilib
configurations, apply the same fixes to cross-candian.bbclass
(From OE-Core rev: 4544b7f1d0abd1b1efd74da430f1ddedf3fdbd1d)
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 previous cross-canadian change was missing some tweaks
to the comments. This clarifies them slightly.
(From OE-Core rev: 154ecc40c289b15fe9cbb33befb20dd10112e788)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
PowerPC toolchains can use the OS "linux" or "linux-gnuspe". This
patch links them together so the one cross-canadian toolchain can support
both.
GCC_FOR_TARGET is set for the GCC recipe as otherwise configure
can pick up an incorrect value.
[YOCTO #5354]
(From OE-Core rev: a1d6331238982b0c5d39b0a18794f6654b00d46a)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Both nativesdk and multilib use MLPREFIX for their partciular purposes. When
we have both set, cross-canadian can confuse SHLIBSDIR. This forces the
variable to the correct value for cross-canadian, fixing toolchains in
multilib builds.
[YOCTO #5333]
(From OE-Core rev: 0633b93086a7de7226f4dc6ca403ee116bc58669)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The cross-canadian compilers are now build once per architecture but were being
installed into tune specific locations which is incorrect. This adjusts things
so they are make TARGET_ARCH specific. We gain the tune specific parts from the
target sysroot which remains tune specific, the compiler and tools are independent
ot that.
binutils/gcc require sysroot options but since we reset at runtime, these shouldn't
have dependencies in the sstate checksums. They are therefore also excluded.
With these patches, switching machines does not result in a rebuild of *-cross-canadian
and the compiler is correctly located and referenced in the target images.
(From OE-Core rev: f58acab6414fe96d9e07ebbe86b348d2ac2bed5f)
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>
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>
The gettext handling of USE_NLS has become a bit tricky to understand, or
alter from the SDK context. This patch introduces a SDKUSE_NLS which can
be set to configure a given SDK/ADT to use NLS or not. This is independent
of the target system NLS usage.
The code in gettext.bbclass is therefore simplified and the classes
themselves now set USE_NLS to appropriate values. No NLS is used
for native, cross and crosssdk since it is never used there and
would just increase build time.
(From OE-Core rev: fe634d47449899f7424adb77ff5bc7ddf8a07a47)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This was needed when do_package for target recipes was target specific
however since it now isn't we can remove these stale references.
(From OE-Core rev: 1c54d8c3639659bc8cf8fa2786a1aa9ed785b723)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently, do_package is machine specific since the shlibs data is installed
into each machine specific sysroot. This change moves the shlibs data to the
pkgdata structure, at the expense of having to iterate over a set of shlibs
directories instead of a single one.
It turns out this isn't any particular hardship for the code and as a result,
do_package stops being machine specific leading to optimisations for builds
that use a common PACKAGE_ARCH.
(From OE-Core rev: cc088489d70fb27d460c3dbe35d6ea382123c134)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In order for the RPATHs in 32bit toolchain binaries to be relocated
properly, chrpath >=0.14 is needed.
[YOCTO #3161]
[YOCTO #3201]
(From OE-Core rev: 71c71b972100803d33fbb062a237e8a15167387b)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
As discussed on the mailing lists, using a suffix to package names is
hard and has lead to many recipes having to do PKGSUFFIX games. Its
looking extremely hard to scale nativesdk much further without hacking
many recipes.
By comparison, using a prefix like multilib does works much better and
doesn't involve "hacking" as many recipes. This change converts nativesdk
to use a prefix using the existing multilib infrastructure.
(From OE-Core rev: 81813c0e322dc04ce4b069117188d8a54dfddb8c)
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>
As reported by Martin Jansa, the path to nativesdk sysroot was changing between
nativesdk and cross-canadian recipes. The problem was the incorrect deinfition of
STAGING_DIR_HOST in cross-canadian.bbclass.
Since nothing really uses the cross-canadian output in the sysroot, only the
packages, its not surprising this bug has gone un-noticed.
(From OE-Core rev: 8c6966cb8e353dc28819419ea7e395fb0d5f2536)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Complete the bb.data.getVar/setVar replacements with accesses
directly to the data store object.
(From OE-Core rev: 2864ff6a4b3c3f9b3bbb6d2597243cc5d3715939)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is comming from x32 need to pass special parameters to ld & as.
(From OE-Core rev: 96931af89f9cc3056e413cff437a85eca85b3b75)
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch removes the variables BASE_PACKAGE_ARCH, BASEPKG_HOST_SYS,
BASEPKG_TARGET_SYS and also removes the immediate assignments in
several core classes as these are no longer required.
This should make it clearer what some of the core variables do and
simplfy some overly complex and confusing class code.
(From OE-Core rev: d5521be2dcbaf213c140b0d12a4176380874426b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This commit changes the sysroots path to be machine specific.
Changes includes:
1) STAGING_DIR_TARGET and STRAGING_DIR_HOST points to machine specific
paths.
2) task stamp files. Adding ${MACHINE} info into stamp files for
do_populate_sysroots and do_package tasks. Add a BB_STAMPTASK_BLACKLIST
to keep native, nativesdk, crosssdk, and cross-canadian stamp unchanged.
3) siteconfig path. Separate the site config path for different machines
to avoid one machine adopting the cache file of another machine.
4) sstate. Add machine name to sstate manifest file.
Change relocation code for sstate paths since sysroot is machine.
Keep native, nativesdk, crosssdk, and cross-canadian unchanged.
5) toolchain scripts. Change the environment path to point to machine
specific sysroots in toolchain scripts bbclass.
6) Relocate la files when populating to a different machine of the same
architecture.
7) Exclude STAGING_DIR_TARGET and STAGING_DIR_HOST parameter from sstate
siginfo since they contain ${MACHINE} information.
Signed-off-by: Dongxiao Xu <dongxiao.xu@intel.com>
environment files.
[BUGID #565] Fixing bug #565, added package of
meta-environment-${TARGET_ARCH} for environment files used by
cross-canadian toolchain.
Also corrected the situation of empty config site file for target.
Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
This patch massively simplifies the canadian packaging and allows
multiple toolchain targets to be parallel installed into the same
nativesdk sysroot without package name conflits. Since we now do
this, we can simplify cross-canadian to become more like nativesdk.
This is a first pass over this task, similar changes would be
desireable to cross and the whole MULTIMACH_ARCH mess can then
probably be similified much further.
Signed-off-by: Richgard Purdie <rpurdie@linux.intel.com>
Using TRANSLATED_TARGET_ARCH instead of TARGET_ARCH for
cross-canadian packages. This is due to the TARGET_ARCH of x86_64
would results incorrect packaging in cross-canadian packages.
The pacakge name appendix of x86_64 target in cross-canadian
packages is x86-64.
Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Ming Liu
Richard Purdie
Joshua Lock
Mark Hatle
Khem Raj
Matt Madison
Peter Seebach
Mario Domenech Goulart
Juro Bystricky
Laurentiu Palcu
Nitin A Kamble
Lianhao Lu
Dongxiao Xu
Richard Purdie
Koen Kooi