Now that xserver provides driver ABI names, all drivers should depend on the ABI
version that they have been built against.
All drivers that include xorg-driver-input.inc or xorg-driver-video.inc will get
these automatically, so this should only impact binary drivers.
(From OE-Core rev: 800b256390b22c3d3d8d6a69f6fb668376a5030b)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The xserver driver ABIs can and do change in a way that is unrelated to the
version of xserver, so it's entirely possible to build an image that has a
mismatch between the server ABI version and the version that the drivers were
built against. xserver detects this and refuses to load the modules.
By adding RPROVIDEs to the xserver package that describe the ABI versions it has
(such as xorg-abi-video-13, xorg-abi-input-11), drivers can RDEPEND on the
version that they were built against. This means that when the ABIs change,
there will be package dependency errors at image time instead of images that
build fine but don't work.
(From OE-Core rev: 8ef5f205aec04140198d5ba0f5c405ae6e977dbe)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We seem to be mostly installing modules into vendor_perl nowadays.
Make sure that the .debug data from there is captured appropriately.
Also, expand ${PERLLIBDIRS} at the point of assignment so we don't
call the python again and again.
[RP: Fixup to whitespace]
(From OE-Core rev: ed7690bf5cc964b5cee55f5ef13c10c75d8e3463)
Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When copying the source files needed for the -dbg package, use hardlinks
where possible. This saves some disk space and hence helps performance
of the builds.
(From OE-Core rev: 6775feb9fe935ab01fd9cae2b2d3fce5824a9a72)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This has been disabled by default upstream since 2007, nothing uses it.
(From OE-Core rev: 06d27cf0fbcc4004e6f456880eca49893c9290bf)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The symlink from uImage-3... to uImage is not created at image creation
time and not properly update on kernel upgrades. This is fixed by removing
the chdir. The other users of update-alternative do not change the directory
before calling it.
(From OE-Core rev: c77ca9ee901468c93570b5264b226f7d17a41c16)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Adapt the recipes to fetch a tarball.
Tarball is generated from latest 2.16 branch
which has e500-math_private.patch already applied
hence we remove that patch.
(From OE-Core rev: 77ee4d7d88976c7bb2bb25b57e06b83edaacbd4c)
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This removes the dependency of eglibc.bb itself on perl and bash
which, in turn, eliminates the need to build those two recipes if the
scripts which need them are not going to be installed.
Also provide dummy do_evacuate_scripts() for all variants of eglibc-initial
otherwise the nativesdk and multilib variants might crash trying to
copy a non-existent mtrace script.
(From OE-Core rev: 74b5f8943b2b29c7b3b62be7d81fb2b3a86b9584)
Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Without this, we'd rerun packaging for every machine since this variable contains
a machine specific component.
(From OE-Core rev: 61131828c59178c923b3d5b5fcacf0dbcba275a5)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Now that PACKAGES_DYNAMIC is more standardised, starting with ^ anchors,
the variable manipulations performed by clsextend for multilib don't work.
This patch at least improves it to hack around the problem and enable
mulitlib builds to work again. If this code doesn't do the right thing, the
recipe is free to override the variable with the correct multilib case.
(From OE-Core rev: 593faec6e0155bdd7a43ee84c24de8ee20287681)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Now do_package isn't machine specific, we're only left with do_populate_sysroot as a
machine specific task. This change marks only the machine specific manifests as machine
specific, defaulting to PACKAGE_ARCH for everything else.
This means we do less work where there are multiple machines using the same
core package architecture and we can start to clean up the sstate duplicate files
whitelist.
(From OE-Core rev: febeaf3d1b8917b660c7279b008d8b03337568e9)
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>
These variables correspond to the PACKAGE_ARCH list combined with the TARGET_VENDOR
and TARGET_OS values. These can be used to traverse the pkgdata structure.
Setting these once in base.bbclass stops pkgdata needing to recalculate the values
and is also useful for the reworked shlibs code in a patch that will follow this.
(From OE-Core rev: f91322edc8b9f2a5906f3908bde2508ae97f2816)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
With the move of the strict/vars/config/warnings modules to the main perl
recipe, we need to RPROVIDE those modules to ensure that package dependencies
on those modules continue to work correctly.
(From OE-Core rev: fe88ae8605f22d9075e4200159aa66605ec36587)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* kernel-modules is always added to PACKAGES later in python code and
needed to be defined as PACKAGES_DYNAMIC
* add it to PACKAGES directly and set
ALLOW_EMPTY_kernel-modules
FILES_kernel-modules
DESCRIPTION_kernel-modules
outside populate_packages_prepend like for other packages and set only
RDEPENDS_kernel-modules from python code
(From OE-Core rev: 0884bdbbf39f2b3a8a342918812f29ddcd3b1e6f)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* to keep ${PN}-locale from
bitbake.conf:PACKAGES_DYNAMIC = "^${PN}-locale-.*"
(From OE-Core rev: 73252b16b501c0986b0ca0895e4534895a9ba3db)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* bitbake uses PACKAGES_DYNAMIC as regexp
^ could make matching faster (and it will be more clear that we're expecting regexp not glob)
* made all those last '-' optional, use .* (or nothing)
(From OE-Core rev: 2f3ebdfa5f42dae51063b043cc4b0fbe20b40064)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The intl directory is part of older gettext and has macros which no
longer get expanded with recent gettext versions. This simply removes
the intl directory from the equation since we'd never need it.
(From OE-Core rev: 199298b6b114db09fd8ff100642ae101050a2e9a)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The compilation routine for the kernel has an explicit call to
build version.h, which works fine for most kernels, but the
location of it has recently changes.
commit d183e6f5 [UAPI: Move linux/version.h]
commit 10b63956 [UAPI: Plumb the UAPI Kbuilds into the user
header installation and checking]
moves the file to include/generated/linux/version.h and then to
include/generated/uapi/linux/version.h.
As a result kernel builds of 3.7 or bisection builds of intermediate
kernel commits will fail with:
make[2]: *** No rule to make target `include/linux/version.h'. Stop.
Making the explicit version.h build conditional on the version, or
via a file test would fix the problem, but it introduces some complexity
to the build.
Even without an explicit call to build version.h, it is always produced
by the kernel build, so it can simply be removed.
This extra make line was originally so that the kernel version could be
determined, so that then different instructions could be executed depending
on whether it was a 2.4 or 2.6 kernel. Since we no longer support 2.4, this
code is no longer needed.
[YOCTO: #3293]
(From OE-Core rev: 4cb20fa89e571ffbc448c579a758db0b9074acf4)
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch fixes 2 problems.
The first one is that when run "perl -V" on target, it fails with lack
of some .pm files. So add these perl module files to package perl itself
to fix this failure.
The second problem is that package nativesdk-perl-modules doesn't depends
on the single perl modules.
In the .bb file, dependencies of perl-modules are set by:
RRECOMMENDS_perl-modules = "${@d.getVar('PACKAGES', True)...}"
The PACKAGES would be reset by do_split_packages since:
PACKAGES_DYNAMIC = "perl-module-*"
PACKAGES_DYNAMIC_virtclass-nativesdk = "nativesdk-perl-module-*"
Then:
1) The target perl-modules RRECOMMENDS on perl-module-*, this is what
we expect.
2) But the nativesdk-perl-modules doesn't RRECOMMENDS on
nativesdk-perl-module-*, this is not what we expect.
The value of PACKAGES after do_split_packages has been set correctly (it
contains the nativesdk-perl-module-* packages)
But the:
RRECOMMENDS_perl-modules = "${@d.getVar('PACKAGES', True)...}"
doesn't work correctly for nativesdk, the
d.getVar('RRECOMMENDS_perl-modules', True)
doesn't get the new value of the PACKAGES, it gets the value of PACKAGES
before the do_split_packages.
This patch will fix the problem.
(From OE-Core rev: d50be1876f7a41822ef7e73207fdf8cccd39e400)
Signed-off-by: Kang Kai <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For builds that don't use gettext, config.rpath may not exist in the target
datadir. This change uses the native directory where it will always
be present due to gettext-minimal-native (which allows us to autoreconf
recipes using gettext even if we don't have gettext built).
(From OE-Core rev: 0ea24447842e6b76ccfee0881f557e1a82e89ef1)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Use "rm -f" in do_install_append() so we don't fail if the files we're
trying to delete have already been removed. This can happen if the
distro policy suppresses both static libs and .la files.
(From OE-Core rev: 2ee9601ae590b2964d1c44b131188a80cc5198a9)
Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This avoids propagating a failure if we encounter an ELF file
that objdump can't parse for any reason. Some versions and/or
configurations of objdump will refuse to read files for "the
wrong" architecture.
(From OE-Core rev: 11f5998e539f7b884ae1387252f8995b2dc7437f)
Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If the link is absolute then we might end up reading from a host binary
or a nonexistent path, neither of which will produce useful results and
may result in objdump failure and python backtrace spew. If the link
does point to a binary within the installation root then we will scan the
pointed-to file at some point anyway so there is no need to do it again.
(From OE-Core rev: 91769af1c1175ac9bb43d16d05fb1c8736dd9287)
Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Various different QA checks are based on essentially the same data from
the ELF program headers. Calling objdump to extract it repeatedly is
inefficient, particularly if the shell is involved. Instead, let's
cache the output from objdump inside the qa.elf object and allow it to
be reused by multiple tests.
Also, using objdump instead of scanelf to check for bad RPATHs (in the
same way that the useless-rpaths check was doing already) allows the
dependency on pax-utils-native to be dropped.
(From OE-Core rev: bf19eeb9f65e91bf2b5d89e7c0b099c55d7c15ff)
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>
When we changed the /usr/libexec default to be /usr/lib/<pn>,
the icon name mapping needed to be updated also.
(From OE-Core rev: d05e63c6c936a9ff209efd1341401ef7cd2ec0c1)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When we changed the /usr/libexec default to be
/usr/lib/<pn>, the icon name mapping needed to
be updated also.
(From OE-Core rev: 58b20de32f4b6ca684120d1b87a7aece2df6f0a6)
(From OE-Core rev: b3ff57c7a15c8509397d4b8e84410aab08ea4d1c)
(From OE-Core rev: 49d8f5d2257f00f02437b95e624960ba4aadd9d4)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This allow the relocation of libexecdir to be done correctly
for the qemu-brigde-helper.
(From OE-Core rev: 945c8f5c687ec61c312209e075edc402f6272186)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This allows the files installed into /usr/libexec to be
relocated to ${libexecdir}. removed unneded prefix=/usr,
which would prevent ${prefix} relocation.
(From OE-Core rev: 10d28438c1e7d793bc398a0ad484782e5baa4877)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
mesa-dri is an empty package, so depending on it doesn't achieve anything.
(From OE-Core rev: a41f8341971a958cf55c07f3c91e1742570053cd)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The sysklogd recipe had a cut-n-paste version of the
update-rc.d.bbclass code which didn't work, but this was hidden
because all images contain the busybox version which does. Building a
busybox-free image unmasked the issue and syslogd wouldn't start on
first boot.
The comments seem to be wrong/stale. AFAICT update-rc.d and
update-alternatives work fine with each other, though there is an
ordering constraint (alternatives must be specified last, so it
"wraps" update-rc.d). This version builds and works both with and
without busybox.
(From OE-Core rev: 644673631bf57bd8d0e152b5fe7621344b5ad24f)
Signed-off-by: Andy Ross <andy.ross@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The PulseAudio server recently added a dependency by default on the
switch-if-port-available module, but this was not enforced by the package
dependencies so the server won't start.
(From OE-Core rev: b0fd984dd132e332056863dcc11b312141d443fd)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
THis change extends COMAPTIBLE_HOST matchings, which allows to include more
hosts with TARGET_OS like linux-gnuspe or linux-gnueabi etc.
(From OE-Core rev: 76138d4b183eff28c678ab13cb1a6da358be2340)
Signed-off-by: Vladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* qt4-embedded was forcing -DQT_KEYPAD_NAVIGATION which depends on feature-completer
* separate variable makes it easier to not enable QT_KEYPAD_NAVIGATION in some upper layer where we have disabled feature-completer
(From OE-Core rev: 0479242a18661cb7fc3d76d208c82fe6ae4378ce)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Otherwise it will fail if these two directories are not the same.
(From OE-Core rev: 491823fdc65d124093f1fed5a56173917443e1d6)
Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This makes the symbolic link portable with the dated zip file, otherwise
the link still points to the original deploy directory.
(From OE-Core rev: 0fc83102eeb48b85027c5b1202d8a584f51679a7)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
gettext m4 macros don't use the usual versioning/serial mechanism used by
aclocal. It therefore won't update them over and above any local version of
the macro. Equally, we don't run gettextize due to it doing slightly crazy
things to the build.
When we put the aclocal directory as a -I option to aclocal, if this was
found first compared to any recipe provided macros, the correct version
of the gettext macro would still "win". With the switch so correctly override
the system directory, older recipe provided macros may get used.
This patch manually removes the problematic m4 macros in the case we're using
gettext and need to use the correct m4 macros.
This patch also always ensures the gettext manipulations happen, even in the
-native case since missing or stale gettext files could cause build failures.
(From OE-Core rev: e9645d2bbeabaa5251d49edd659ab320fd66d0ee)
(From OE-Core rev: 841ea3c1c18e50e77fccbd5f44d6a79a50913b67)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The previous steps taken to address races over the aclocal macro directory and the removal
of files hasn't been sufficient since aclocal still looks at that directory as part of its
default search path. This patch passes the aclocal-copy directory into aclocal as its system
directory, removing any chance of it accessing the original aclocal directory.
Hopefully this should therefore fix the race issues once and for all.
In order to do this, cp-noerror needs to not error if the directory already exists.
Its also been noticed that aclocal defaults to using STAGING_DATADIR_NATIVE even when
building for the target. Only using the target directory would cause errors such as
missing pkgconfig macros (since we only depend on pkgconfig-native, not pkgconfig).
This patch processes both sets of macros maintaining existing behaviour. At a future
date we could look into potentially optimsing this.
[YOCTO #3216]
(From OE-Core rev: ad29b331e0d61708e68ef772cdb19154956fa67e)
(From OE-Core rev: f362cc419e5a480acd16c71c802636dbedc932d9)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Without this, if configure fails, it won't be able to run again as the directory
already exists.
(From OE-Core rev: 71a3ba536d022eea3a199cf4d6c5c791d91603a0)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
diffutils has a rather confused set of getext macros with different names and
strange conflicting version requirements. This patch removes the problematic
macros allowing it to 'gettextize' to the latest standard gettext code without
issue.
(From OE-Core rev: a40b89333652ca22a6e6957ab8a2a4e41b87b4c0)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
kexec-tools optionally looks for the lzma code provided by xz. Since this
is generally useful for lzma compressed kernels, add the dependency and
make builds determinstic.
(From OE-Core rev: accea64234124f25345a9288c0739c433de671f8)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
