The code assumes that PKG_DATADIR exists and will fail if an image has not been
generated which creates it. This occurs when something like buildtools-tarball
is built which doesn't have target packages, only nativesdk ones.
Since this shouldn't be fatal, workaround this by creating the missing
directory.
(From OE-Core rev: 319c5d55bb0c7e429766f46dd42a15e16a43c4dd)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When looking at and trying to build a slightly customised SDK, I
realised the code could be simplified and written in a way which was
more customisable.
This patch moves various function calls into the SDK_POSTPROCESS_COMMAND
which was intended for this kind of use.
(From OE-Core rev: baa4e43a29e45df17eaa3456acc179b08d571db6)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
aarch64 sets OLDEST_KERNEL to 3.14. This stops the aarch64 SDK installing on
anything with an older kernel which is clearly incorrect.
I attempted to extract the correct non-overridden version from the data store
but it proved problematic and I was running into data store issues. Those
are a separate problem but there isn't time to fix this right now.
Instead just code the SDK kernel version separately to work around this for
now (and fix the autobuilder tests and SDK usage).
(From OE-Core rev: 522ba4c51fff53566678b2689d0d63c393e417b3)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Extensible SDK cannot be installed as root so by default offer to install it in
user's home directory under distro/distro_version replacing the normal SDK
version '+' char with a '_' as that's a restricted character for bitbake
(From OE-Core rev: 5486e76cd8abb946b81cff78719d67cfb87cddc6)
Signed-off-by: Brendan Le Foll <brendan.le.foll@intel.com>
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since the extensible sdk uses bitbake, which can't run as root, the sdk
shouldn't be installed as root.
Previously it would error out late into setup when bitbake errored
saying not to run bitbake as root.
Now the script errors with a message saying the extensible sdk can't be
installed as root.
[Yocto #7545]
(From OE-Core rev: 309e8f4e536148056223f50637ed291c48d148ca)
Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fixed when enable multilib:
require conf/multilib.conf
MULTILIBS = "multilib:lib32"
DEFAULTTUNE_virtclass-multilib-lib32 = "x86"
$ bitbake core-image-minimal
No lib32 package should be built since we don't build
lib32-core-image-minimal, but a lot them are built:
lib32-opkg-utils
lib32-cryptodev-linux
lib32-sqlite3
lib32-libtool-cross
lib32-ncurses
lib32-db
lib32-diffutils
lib32-zlib
lib32-gcc-cross-i686
lib32-gcc-cross-initial-i686
lib32-flex
lib32-libgcc-initial
lib32-libffi
lib32-linux-libc-headers
lib32-expat
lib32-openssl
lib32-glibc
lib32-binutils-cross-i686
lib32-gcc-runtime
The are introduced by toolchain-scripts.bbclass' anonymous function (the
MULTILIB_VARIANTS), but it doesn't need toolchain-scripts.bbclass, in
fact, so remove it to fix the problem.
[YOCTO #7961]
(From OE-Core rev: 000ff33729854b1a41206a087cde1e50f49017e0)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For now, we can create manifest for the target part for SDK.
I think it's nice to have a place for users to look into to examine contents
of the host part of SDK.
This also affects uninative-tarball and buildtools-tarball as they
inherit populate_sdk.bbclass. After this change, we could have a manifest file
created in the deploy directory containing a list of packages used to
construct them.
[YOCTO #7604]
(From OE-Core rev: 3069641589942f52b11e88cfea564ebba4beb3c2)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
`bitbake uninative-tarball' raises the following warning.
WARNING: Function doesn't exist
This is because SDK_PACKAGING_FUNC is set to "" in its recipe.
Anyway, we need to check this variable to avoid executing empty function.
[YOCTO #7598]
(From OE-Core rev: 4c0ae7cce06de15f6881654ecec7f8bb743ff389)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Show a friendly title when running the SDK installer, so the user knows
what SDK they are installing. The title is controlled by the
SDK_INSTALLER_TITLE variable and includes the distro name and SDK
version by default.
(From OE-Core rev: 0af913887f4c0a79c6b950bd5d57c06333520a14)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If the buildsystem is copied into the sdk and its toolchain is to
be used, then the relocation provided in toolchain-shar-template.sh
isn't needed and will actually fail.
So break the relocation aspect out and essentially make it another
SDK_POST_INSTALL_COMMAND script.
(From OE-Core rev: 9721378688a05cd8d8443c6ee4be823e5c0688f6)
Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add a check to the SDK so that it only runs on systems with kernel versions
it supports.
[YOCTO #6856]
(From OE-Core rev: b4caa8085aa15674162ff5135b13409998db9510)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This was added back in the depths of history (around 2006 in OE-Classic)
when apparently the host sed couldn't always be relied upon. We now call
the host sed all over the place without this dependency and don't have
any problems. On the other hand, having it around can lead to races
where we're calling sed in one task and staging it to the sysroot in
another, the result being nasty failures compiling binutils for example.
Since it isn't needed, let's just drop it completely.
Fixes [YOCTO #7264].
(From OE-Core rev: 863a3966e1a200e7a15fbdbb8e04dce6e3f6c1a2)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently we have the horrible situation where meta-environment packages
the toolchain environment files and they get included in the SDK but are
broken, then, the SDK code overwrites them with good versions. This is
suboptimal.
This change fixes the code in meta-environment to create working
files and adds in the multilib support from populate_sdk_base, then
we remove the code in that base bbclass and rely on the packages
being installed if/as/when needed. This removes the duplication
and the broken versions of the files, hopefully making all well.
[YOCTO #6608]
[YOCTO #6613]
(From OE-Core rev: 063355e5965439c7b3253d692d7ab0ed1189d123)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add an SDK_POST_INSTALL_COMMAND variable which allows additional
commands to be added towards the end of the SDK install script, for e.g.
additional processing that needs to be done as part of installing the
SDK.
(From OE-Core rev: 10df0718d6a626d99beb68cde8d914ee0820d7eb)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently this variable isn't passed through the magic rename
mangling. This means that if you try adding "nativesdk-eglibc"
to an image directly, you can't since the package is renamed
by debian.bbclass and nothing sees the renaming.
This is annoying since I wanted to exactly that. This code
change passes it through the standard renaming function, the
tricky part is that we have to set PKGDATA_DIR to point to
the correct sysroot during the call. We create a copy of the
datastore for the purposes of the call to do this.
(From OE-Core rev: d2fbc55d6863a767e69092bac686c02c3ec34650)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Similar to the way BSP images have rootfs a manifest, the toolchain now also has a manifest file created alongside the sdk image.
(From OE-Core rev: 3d42fd1f050a1382b15c3c4d59fd02d0ed7091b2)
Signed-off-by: Corneliu Stoicescu <corneliux.stoicescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It makes sense to move the script to a separate file, making the bbclass
clearer and allowing the end user to more easily customise or replace
it. There are no functionality changes, just the addition of some
substitution variables.
(From OE-Core rev: f99732a29689c65083ad09abb302f372042c8cd4)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There is no real reason to have these separate files any longer. It does
mean in meta-toolchain type recipes some extra variables are defined but
it also means the common code can be grouped and maintained together
which I believe is more beneficial. We therefore merge the classes.
(From OE-Core rev: 2cf42b49003494e1b10775523c9a2547eaf16ea4)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
"man grep" on centos:
-R, -r, --recursive
Read all files under each directory, recursively; this is equivalent to the -d recurse option.
"man grep" on a more recent ubuntu system:
-r, --recursive
Read all files under each directory, recursively, following symbolic links only if they are on the command line. This
is equivalent to the -d recurse option.
So we have an issue when the SDK installer (even with
buildtools-tarball) is used on old hosts since it may try and
dereference paths which it should not. This is caused by differences in
the behaviour of grep -r on older systems.
The fix is to wrap this in find so that only real files are found (as
elsewhere in the script.
[YOCTO #6577]
(From OE-Core rev: 7986adeac16550b33f65fded39a55f668e0e543f)
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>
"bitbake core-image-minimal -c populate_sdk; rm tmp -rf; bitbake core-image-minimal -c populate_sdk"
breaks due to missing dependencies on things like update-rc.d. Since
we're effecitvely building a rootfs, we need all the same depends
as do_rootfs has.
The easiest was to achieve this is to simply load them. This resolves
the failure when targeting the task from sstate.
(From OE-Core rev: cdb0bcef28493edd4cad2a882fb8ba402485a2f1)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Most people expect the toolchain from a multilib build to contain multilib
components. This change makes that happen and is easy for users to override
should they want something different.
(From OE-Core rev: 396371588c7fd2d691ca9c39cd02287e43cb665b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When extracting toolchain, if the list $executable_files has filenames
with empty space character, the list will created but relocate_sdk.sh
will not handle it well. This will lead to the below erro:
./tmp/deploy/sdk/buildtools-mytools-x86_64-nativesdk-standalone-1.6.1.0.sh
Enter target directory for SDK (default: /opt/mydistro/mytoolset/1.6.1.0):
You are about to install the SDK to "/opt/mydistro/mytoolset/1.6.1.0". Proceed[Y/n]?
Extracting SDK...done
Setting it up.../opt/mydistro/mytoolset/1.6.1.0/relocate_sdk.sh: line 2: sintaxe error `token' `('
/opt/mydistro/mytoolset/1.6.1.0/relocate_sdk.sh: line 2: `e
The same occurs with replacement of ${SDKPATH} in configs/scripts/etc files.
We should ensure that full path is protected before relocate_sdk.sh
and ${SDKPATH} replacement calls.
(From OE-Core rev: d7adf8349a65da6f0fdd0f00e606e8bc27ce3f28)
Signed-off-by: João Henrique Ferreira de Freitas <joaohf@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Remove a confusing echo after the previous commit, thanks for a suggestion
from Laurentiu.
(From OE-Core rev: cd1fdd05cc7457706d1e40042854c154cbb3d9e9)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
A
The variable was only partially implemented, and the part that was there
was named incorrectly to, missing the 'TASK' piece.
(From OE-Core rev: d0cb34cfe9a51fd8bc1e6e28c8eda60a25adc1ec)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It is introduced a bug, since The return of "file ld-linux-x86-64.so.*"
does not include "dynamically linked" in redhat 5.9/6.0(32 bit), and lead
to that ld-linux-x86-64.so.* is not in executable file list.
(From OE-Core rev: fc9603d7d7042efe8941172091cca8578bdde15b)
Signed-off-by: Roy Li <rongqing.li@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If the user changes the DEPLOY_DIR variable to set up a custom deploy
directory for images, packages, SDKs as explained in the documentation,
the variable SDK_DEPLOY does not take it into account and fallback to
TMPDIR as default. Therefore, SDKs were not found in the correct
location.
(From OE-Core rev: 0f043e9a3fdc4b489b55e0605fee01927854205a)
Signed-off-by: David Vincent <freesilicon@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
nativesdk packages were created only for the first backend listed in
PACKAGE_CLASSES. Hence, if one had it set to "package_rpm package_ipk"
and did a 'bitbake -c populate_sdk core-image-something', the nativesdk
packages were created only for rpm.
This is particularily bad for adt-installer which is based on opkg
repos.
Credits go to richard.purdie@linuxfoundation.org who suggested me this
fix.
[YOCTO #5900]
(From OE-Core rev: 85c3238ee713bc27e99a2e393e3bf8438ed4d91f)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We don't need to depend on do_package anymore after the split out of
do_packagedata (which happened a while back in OE-Core commit
6107ee294afde395e39d084c33e8e94013c625a9). We do need dependencies on
do_packagedata though since the SDK construction needs to make use of
the pkgdata files.
(From OE-Core rev: 8491b998d290f9717d537ca84bc2beb475cf3ced)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
do_rootfs has ${S} in cleandirs, and during do_populate_sdk we call
exec_func() several times, which by default uses ${B} as the working
directory. If do_populate_sysroot and do_rootfs race against eachother,
the directory may not exist at the exact instant that the setup code
for do_populate_sdk tries to cd into it. We don't actually use ${B}
for do_populate_sysroot so we can set it to something else just for that
task to avoid the race.
NOTE: because this task name contains an underscore, the override will
not work; the BitBake patch that changes these to hyphens for the
task override is required for this patch to work (but won't break things
without it.)
(From OE-Core rev: 53578cef2cbc59dcc637d1cc561f63b3c448425a)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The sdk complementary install operation was defined in image.bbclass,
but the sdk recipe (such as meta-toolchain.bb) didn't inherit this
bbclass but populate_sdk, and both of image and populate_sdk bbclass
inherited populate_sdk_base bbclass, so move the sdk complementary
install definition to populate_sdk_base bbclass fixed this issue.
(From OE-Core rev: 907458935bab391f5bfba8f581ea9835078548d9)
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Lose the old shell populate_sdk_image() function and use the new python
implementation.
(From OE-Core rev: 5a81b8a0820ceac972a68af2caebdc8d2f3945a1)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We've removed nostamp from do_rootfs in image.bbclass in OE-Core commit
8505008c115efb54d18e5f25441c7a938a32ffaf, and we should do the same for
do_populate_sdk here for the same reason - we can now rely on task
signatures so if nothing has changed, we don't need to re-run it.
(From OE-Core rev: de9b693f4ff311f1310a1c6005e0d5c225aabef6)
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>
When toolchain directory is changed to execute mode, some non-executable
files or empty files are sorted. This will result in some errors. Thus when
sorting executable files or dynamically linked library, additional conditions
are to exclude non-executable files or empty files.
(From OE-Core rev: c9d56308bfa9ee7f4a9b22eae86390626ddc1c35)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This has now been superseded by testimage.
(From OE-Core rev: d469c92394a1a95ae7a45b8b80dc4c2918e0e9a6)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since packagegroup-cross-canadian package name has MACHINE in it, make
the necessary changes here.
[YOCTO #4783]
(From OE-Core rev: 23531590d40acde2775b7c3b90682edc501d68f4)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It can be useful to override or append options to the SDK tarball creation command
so add a variable to allow this.
(From OE-Core rev: ae86a46c1b255e7c2833eb6d48ed46eba440e95a)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Old way find -perm +mode is no longer supported in newer
versions of find (Fedora 19). Man page says:
-perm +mode
This is no longer supported (and has been deprecated since
2005). Use -perm /mode instead.
[YOCTO #4853]
(From OE-Core rev: 21b079e01873e2fb4d8674541e8c5818ba73554e)
Signed-off-by: Stefan Stanacar <stefanx.stanacar@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Only grep for 'OECORE_NATIVE_SYSROOT=' otherwise things like
toolchain_create_sdk_env_script_append() {
echo 'export MY_DIR_FOO=$OECORE_NATIVE_SYSROOT/my/dir/foo' >> $script
}
trigger the following error while executing the install script:
find: `$OECORE_NATIVE_SYSROOT/my/dir/foo': No such file or directory
(From OE-Core rev: d084c31720f9c13a71c5981f4eda21e18ba2350f)
Signed-off-by: Bernhard Guillon <Bernhard.Guillon@hale.at>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Only the printf bash builtin knows about %q format option for escaping
spaces. The coreutils version doesn't. Unfortunately, neither dash nor
sh have a printf builtin. So, escape the spaces using sed.
[YOCTO #4811]
(From OE-Core rev: 6ac06a65ce52d4c123da53f115c84cb0a98bc18f)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Spaces are not handled properly in some parts of oe-core and it's safer
to abort toolchain installation if path contains spaces. Even though
we fix space handling in the toolchain installation script, there are
various other parts in the toolchain (perl scripts, sysroot path passed to
toolchain binaries, shebang lines) that would need special handling. So,
for now, just bail out if path contains spaces.
The checking for spaces in the path is done after expanding relative
paths to absolute and tilde conversion.
[YOCTO #4488]
(From OE-Core rev: 8c35ba2d3048ce69f74f72cb2676e4bc162cfb63)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
allarch multilib recipes are meant to provide a list of different multilib variants.
Unfortunately since the pkgdata also has mappings for these, they get mapped back to
the original package name which means the effect is undone at package creation time
when the remapping code is called.
This patch adds in a conditional to break that chain meaning the packages get
the correct RPROVIDES and image builds work correctly with opkg.
[YOCTO #3453]
(From OE-Core rev: 1a1927f8a04fe0a2b3b853ebdd33ccb807f00b59)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
SDKs are constructed in a similar manner to images, and the contents can
be influenced by a number of different factors, thus tracking the
contents of produced SDKs when buildhistory is enabled can help detect
the same kinds of issues as with images.
This required adding POPULATE_SDK_POST_HOST_COMMAND and
SDK_POSTPROCESS_COMMAND variables so that data collection functions can
be injected at the appropriate points in the SDK construction process,
as well as moving the list_installed_packages and
rootfs_list_installed_depends functions from the rootfs_{rpm,ipk,deb} to
the package_{rpm,ipk,deb} classes so they can also be called during
do_populate_sdk as well as do_rootfs.
Implements [YOCTO #3964].
(From OE-Core rev: c3736064483d4840e38cb1b8c13d2dd3a26b36aa)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
grep will throw "No such file or directory" errors for all invalid
symlinks. To overcome this, move the search after the symlinks have been
changed to point to the new location.
(From OE-Core rev: 239a43fc4037bfe941bb60dd077ee477de177b7c)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When using the -c populate_sdk option, images are not generated quite as
they should be under certain circumstances. For example the dropbear
feature may not get replaced with openssh, leading to both being installed
with an appropriate rootfs failure.
This patch moves the remapping logic to later points in the code, ensuring
there is no conflict. The result is slightly simpler too as an added bonus.
[YOCTO #3749]
(From OE-Core rev: 90cfa16bd4a02ada84ef94f6ae6f182beb8bdc01)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When run "autoreconf" in toolchain, there is an error if the host's perl's
version is not the same as the one in the SDK, the error says that the
executable perl mismatches the perl lib's version.
This is because most of the autotools' scripts use the "#!/usr/bin/perl -w"
which is host perl, but the gnu-configize uses "#! /usr/bin/env perl" which
invokes the perl wrapper in the SDK, and the wrapper will set the PERL5LIB to
the SDK which causes the mismatch. We can make all the perl scripts to use the
host perl or the SDK perl to fix this problem.
[YOCTO #3338]
(From OE-Core rev: 365aa1412a3d31128cb3df02959acdc7df5f2ebc)
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently, SDKTARGETSYSROOT points to PACKAGE_ARCH which, sometimes, can
be set to MACHINE_ARCH. When this happens, the default target sysroot
passed to the cross-canadian toolchain, which points to TUNE_PKGARCH,
will be different from the directory where the target sysroot has been
deployed.
In order to fix this, use REAL_MULTIMACH_TARGET_SYS variable instead of
MULTIMACH_TARGET_SYS.
[YOCTO #3784]
(From OE-Core rev: 41437aaac0cfc6f931d3b2974d380f20ec01f6e8)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
After having to debug the SDK installer a few times in
addition to the relocation code the following patch was created
to improve the capabilities around debugging the SDK installer.
1) Add a verbose mode -D which set a set -x to see what
the SDK installer is doing.
2) Add a mode -S to save the relocation scripts for the purpose
of debugging them in conjunction with -D
3) Add a mode -R to not execute the relocation scripts for the
purpose of debugging the relocations.
(From OE-Core rev: 0e6dd19b9736d2a8ae7c0f0ab124337d579b8f06)
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
[ CQID: WIND00392947 ]
It is not good user experience that the self-extracting archive (.sh file)
has no execution permission by default.
(From OE-Core rev: 3d6537f6ab1ce98075461b9a5d49885c2454417f)
Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Commit c04f5435 "populate_sdk_base.bbclass: use SDK_ARCH instead
of SDKMACHINE" prevents not only the installation of 64 bit SDK
configurations on 32 bit hosts (which indeed cannot work), but also
the legitimate installation of a 32 bit SDK on a 64 bit host.
Fix this.
While there, also make sure we use the same patterns ("i[3-6]86" resp.
"x86[-_]64" to get unified strings for both INST_ARCH and SDK_ARCH.
(From OE-Core rev: 7ddd97b9b09fe7a327916ea88908a63375556ae6)
Signed-off-by: Wolfgang Denk <wd@denx.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The gnu tar versions less than 1.24 pose a problem with the SDK
extraction. A typical SDK will have hard links. The SDK is created
with a newer version of tar but extracted with the host version of tar
which may be much older. The specific defect is that older version of
gnu tar do not properly handle the "--strip-components=" argument when
processing hard links and the paths do not get properly truncated and
hooked up with the corresponding real file payloads. This leads to
errors like the following during the SDK install:
tar: sysroots/qemux86-yocto-linux/usr/bin/getconf: Cannot hard link to `./opt/yocto/SDK-qemux86/sysroots/qemux86-yocto-linux/usr/bin/POSIX_V6_ILP32_OFFBIG': No such file or directory
The simple way to work around the problem is to just not save the path
information that is going to get chopped off anyway. This has the
added benefit that it saves a small amount of space in the tar archive
as well.
(From OE-Core rev: 0f4112016bac07f0294f2a4d0fde4659c0fed2d2)
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The perl module for automake has an embedded path in it, this needs
to be relocated.
(From OE-Core rev: ad79360c1d992830d4f0e06a3bbf0622658c0540)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Having to include some dummy class is suboptimal and we no longer need to
do this. Also move this check to populate_sdk_base since we then don't
need to include it in toolchains specifically.
(From OE-Core rev: 2ee979ff71d31e497be856ea1443667c6d799d34)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch addresses an SDK installer issue on older distributions if
the installation takes place in another directory than the default one. In
fact, is all about the 'file' utility version used for determining if a
file is a text file. For example, for a perl script, newer 'file' versions
return:
"...script, ASCII text executable"
while older versions return:
"...script text executable"
Hence the regex pattern didn't match the scripts.
Also the patch contains two unrelated minor fixes:
* return an exit code of 1 instead of -1 when installation machine is
not supported. That because on an older distribution we also get this
error message: "exit: 9: Illegal number: -1";
* remove unnecessary $SUDO_EXEC prefix to grep;
[YOCTO #3538]
(From OE-Core rev: e7e23a1e5aae4028e21e37ec09e9d431a9adfbcb)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
My previous patch removed the mkdir and added it at the end of the
"gaining SUDO rights" block in order to fix directory creation when
installing in a location without proper rights. Unfortunately this
messed up the directory creation as normal user as it will ask for
sudo right in order to create it...
Hopefully, this will fix both cases.
(From OE-Core rev: 86286287d0134ade73f6a282158dde86bf0159e8)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
mkdir was in the wrong place and missing sudo rights. Hence, the
installation to default location (or any other for which the user didn't
have rights) would fail. Unless the installer itself is run with sudo.
(From OE-Core rev: 555d03c466490ab12b2b1d049736593da2334e97)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The shell archive that populates the external SDK should fail if it
cannot find the ld-linux.so else it will corrupt all binaries because
a random path will be used from the list of executables when dl_path
is empty.
(From OE-Core rev: 206aa794933d41dfb037e9c654c818cd4d4f956d)
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
1.Adds a sudo passwd prompt during installation if the
user couldn't install on the dir.
2.Adds option -d <dir> to enter dir without prompt.
3.Adds option -y for automatic yes to all prompts, a non-
interactive method.
[YOCTO# 3153]
[YOCTO# 3309]
(From OE-Core rev: e85e1aace604626914a06dab4663094d6dd9b035)
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch contains two fixes:
* if the user wants to install the SDK in a directory that already
contains a SDK for the same architecture, a warning will be shown;
* when the symbolic links are relocated use -n option. Otherwise,
symbolic links to existing directories will be created in the
directory itself;
[YOCTO #3401]
(From OE-Core rev: cd035f170a2448b9ebcf0cfac5d4c0fa472b7049)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If SDKMACHINE is not set, the toolchain will be built but the tarball
installer will not run. A better choice is to use SDK_ARCH because, even
if SDKMACHINE is not set, SDK_ARCH is set, by default, to BUILD_HOST.
(From OE-Core rev: b6d391903ae8baf902fa142a58533857ade6afd3)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Do not allow installer to continue if the installation machine architecture
does not match the intended SDK machine architecture.
[YOCTO: #3269]
(From OE-Core rev: 1f78e2c97f978f0f02e884870e7c495751f0802c)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In the previous version, tar extraction use the --strip-component
option with "4" hard coded value.
If we set another SDKPATH, with a different depth, the sdk installation
fails.
This patch computes the level from the SDKPATH value.
(From OE-Core rev: 7aee4e9438755c230e1399bd5226d6c8e7fcca31)
Signed-off-by: Fabien Proriol <fabien.proriol@jdsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Update the host toolchain list, for cross-canadian toolchains, to ensure
that all of the supported multilibs are built and installed. This
dynamically generates the dependnecy set based on the current multilib
configuration.
(From OE-Core rev: 54bc658416ea5679bbfdc76e3ef8767c0a15211c)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When multilibs are enabled, there will be more then one environment
file created. We need to be sure to process each environment file.
The next function can simply use the last environment file processed
to get the magic value(s) that it requires.
(From OE-Core rev: 6f0537c835c35dcff5154de0bec066ec3e71a4f8)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If you wish to change the install/unpack method of the sdk, this can now be done by making
your own create_shar method, and setting a SDK_PACKAGING_FUNC variable to your
new create_shar function.
(From OE-Core rev: 3955c8eced352226bb4c9ceb710dbe02474b9024)
Signed-off-by: Thomas Kristensen <thkriste@cisco.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This patch will allow one to run the installer from another directory
than the one where it's actually located.
Suppose the installer is in /home/user/test/my/sdk and the current
directory is in a different place. With this patch, one can run the
installer like this:
$ sh ~/test/my/sdk/poky-eglibc-x86_64-arm-toolchain-gmae-1.2+snapshot-20120920.sh
[YOCTO #3135]
(From OE-Core rev: 3c7aac33cb63dc63b989db4e9d7389a7f4d3c18d)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The directory usr/libexec/ in the SDK sysroot contains the default
symlinks to the toolchain binaries and these, too, need to point to the
correct toolchain path.
[YOCTO #3090]
(From OE-Core rev: 6e4923c0c9b218271fd44d78df9987b5cabb1c03)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The SDK installation scripts should not canonicalize symlinked
directories because the entire relocation would be done to the directory
to which the symlink points. Instead, if the installation is a symlink,
use that path to relocate the binaries.
For example, if we have the following symlink: /opt/sdk -> ~/my/test/sdk
the binaries will be relocated to /opt/sdk not ~/my/test/sdk as it is
done now.
[YOCTO #3102]
(From OE-Core rev: 9e6a25e2e9a7f37c3baa0b2949a43ac4127868da)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
oe-core removed the prerequisite to have sh as bash. POSIX doesn't define
any options and furthermore allows 'echo -e' to be the default behavior.
This means that in dash 'echo -e' will actually print '-e' and interpret
backslashes by default. We use instead 'printf' builtin command with or
without '\n' to simulate 'echo -e' or 'echo -n'.
'printf' needs format while 'echo' can be used without any arguments. So
'echo >' was replaced by 'printf "" >'.
'echo' without '-n' flag adds a new line by default so to keep the same
behavior of two new lines while using 'echo "\n"', 'printf "\n\n"' is
used.
[YOCTO #3138]
(From OE-Core rev: a19880ad10ccb5d7d909dcf9de5c3dc58a0ebcd3)
Signed-off-by: Andrei Gherzan <andrei@gherzan.ro>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
"Package group" is a much more appropriate name for these than task,
since we use the word task to describe units of work executed by
BitBake.
(From OE-Core rev: 424dcf7046e4ad09dcc664eb1992201195247fcf)
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>
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>
The problem appears if multiple setup environment scripts are found. In
order to find only the script we're interested in, I removed globbing in
matching pattern with ${REAL_MULTIMACH_TARGET_SYS} that will be expanded
to the correct string.
Also, fix a problem when changing the scripts/configs. The grep pattern
matched also files that contained "text" in their name.
(From OE-Core rev: d81768560676da201fd730ae3930b080ab8c5c75)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In order for the SDK to be relocatable, the user would need to call a
setup script to change the binaries acordingly. Having an auto-extracting
archive has the advantage of being more user friendly and the user does
not have to call the setup script separately after extracting the SDK.
It is called automatically.
(From OE-Core rev: d4e976849da845d595db7029d8089946ae16d982)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We set do_populate_sdk[depends] in populate_sdk_*.bbclass, but since
these are inherited at the top of populate_sdk_base.bbclass, those
values are wiped out by using = at the end of the latter class. Use +=
to avoid this.
(From OE-Core rev: a87b5d6cab34579cf3abe03f4663d9a72e218397)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When building an image recipe, you can now build a companion SDK by
calling the populate_sdk task:
bitbake -c populate_sdk core-image-minimal
Note: there are still issues w/ the SDK not working completely with
multilibs.
A lock is required between rootfs and populate_sdk activities to prevent
configuration file clashes and similar package management problems in ipk
and deb based systems. (RPM already had a lock for a different reason.)
(From OE-Core rev: a0de2a56f19ae4d8cd88e46e96917a7a019fe1ab)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In order to enable basic multilib support for the export of an SDK
image, a number of minor changes had to be made:
The value of MULTIMACH_TARGET_SYS needs to be variable. This way we
can define the value to the appropriate multilib. (Also in some cases
the default PACKAGE_ARCH is set to MACHINE_ARCH which is incorrect for
the SDK.) Add a companion REAL_MULTIMACH_TARGET_SYS, based on code
from meta-environment, to allow for this.
We have to convert the do_populate_sdk into a python call, and then
break up the previous items into three parts.
* Image construction
* Setup of environment files
* Generation of the tarball
Then we can iterate over the multilibs to populate the environment files.
Finally, matching changes were needed in the toolchain-scripts file. And
what I presume is an optimization of immediate evaluation for
TOOLCHAIN_CONFIGSITE_NOCACHE and ..._SYSROOTCACHE needed to be done at
runtime, otherwise the wrong values may be used.
(From OE-Core rev: 42545ffbb37f2646a2a8c20999c21d3300e24f59)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If we combine the do_populate_sdk with the image generation, we want
to avoid the dependency processing unless do_populate_sdk is run.
This requires the bitbake change to implement task based rdepends.
(From OE-Core rev: 5fa9f980248e7813ce74f48a29c4b7d94e308cf9)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We want to be able to supply attempt only packages in the SDK in order to
support more advanced SDK images that more closely match specific image
recipes.
(From OE-Core rev: 9b9efa96537f4977b158c29151e53d02600d2294)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
populate_sdk was renamed to populate_sdk_base in order to allow for changes
that may break existing SDK recipes. Any such changes need an analog
in populate_sdk (new version) to restore previous desired behavior.
In addition to the rename, one minor change was made. The _base version
only had the do_populate_sdk as an added task, but no before or after defined.
For compatibility, populate_sdk has do_populate_sdk defined as "after"
do_install and before do_build, this is identical to the original behavior.
(From OE-Core rev: bde224ba44c16edc1892cea1b33ab973ae971115)
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
