| Age | Commit message (Collapse) | Author |
|---|
|
Found via verify-bashisms.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.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.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
wic-tools has tasks which would always rerun and not come from sstate
to ensure we have a correctly populated sysroot. This is low overhead
and can be ignored from an eSDK perspective.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Since Paul reverted the sstate.bbclass change which was checking the sstate
mirror test results, this change should also not be needed anymore.
This reverts commit e30f5002c4f216757ace27ad8d06164716ca46b5.
|
|
If TMPDIR is configured to be somewhere outside of TOPDIR (a not
uncommon configuration where you have multiple disks and space on /home
is at a premium) then our attempt to find out the location of paths
under TMPDIR by using a relative path led to horribly broken paths
ending up in the eSDK. To save pain, just force a known value for TMPDIR
(i.e. ${TOPDIR}/tmp) and then we can assume that everywhere else.
Fixes [YOCTO #10797].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
There were a few straggling expansion parameter removals left for
getVar/getVarFlag where the odd whitespace meant they were missed
on previous passes. There were also some plain broken ussages such
as:
d.getVar('ALTERNATIVE_TARGET', old_name, True)
path = d.getVar('PATH', d, True)
d.getVar('IMAGE_ROOTFS', 'True')
which I've corrected (they happend to work by luck).
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
With enabled SSTATE_MIRRORS sstate code expects mirrors to
contain entries for all tasks, which is not the case for ext
installer as it uses reduced sstate cache.
Added do_package tasks to BB_SETSCENE_ENFORCE_WHITELIST to prevent
installer failing with ERROR: Sstate artifact unavailable
[YOCTO #10832]
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Mapped uninative sstate directories to make ext SDK installer to
use them when it's run on systems with gcc version different from
gcc version used to build installer.
[YOCTO #10832]
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
getVarFlag() now defaults to expanding by default, thus remove the
True option from getVarFlag() calls with a regex search and
replace.
Search made with the following regex:
getVarFlag ?\(( ?[^,()]*, ?[^,()]*), True\)
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
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 ?\(( ?[^,()]*), True\)
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Some c++ libraries fail to build if uninative is built
with gcc 5.x and host gcc version is either 4.8 or 4.9.
The issue should be solved by making separate uninative sstate
directory structure sstate-cache/universal-<gcc version> for host gcc
versions 4.8 and 4.9. This causes rebuilds of uninative if host gcc
is either 4.8 or 4.9 and it doesn't match gcc version used to build
uninative.
[YOCTO #10441]
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
The change to move C++ ABI tweaks to bitbake.conf should make this redundant, so
revert it.
This reverts commit c56cd49a12645e82d0a16bb94be16ac509f8813c.
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
The TOOLCHAINEXT_OUTPUTNAME is different from TOOLCHAIN_OUTPUTNAME, it
is used for eSDK only, so that it doesn't mix with SDK, use "?=" for it
so that other conf file can define it.
If we don't use "?=" here, then we need use forcevariable to redfine it:
TOOLCHAINEXT_OUTPUTNAME_forcevariable = "foo"
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Fixed when nativesdk-unfs3 is installed:
$ bitbake <image> -c populate_sdk_ext
| Traceback (most recent call last):
| File "/path/to/oe-core/scripts/lnr", line 21, in <module>
| os.symlink(target, linkname)
| FileExistsError: [Errno 17] File exists: '../../../../tmp/sysroots/x86_64-linux/usr/bin/unfsd' -> '/path/to/9.0/sysroots/x86_64-wrlinuxsdk-linux/usr/bin/unfsd'
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
It seems that possibly due to OE-Core commit
ac59063bee0e32d0737340974f657341717a6abe, binaries produced without
uninative aren't compatible with the uninative glibc. I did try earlier
to ensure that the eSDK could work without uninative since the default
configuration in OE-Core does not enable it, but it seems like I didn't
go far enough. Given the practical considerations, just give up and
require uninative to be enabled in order to build the eSDK. I'm not
particularly happy about this, but I don't seem much of an alternative.
Fixes [YOCTO #10566].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
If you are using a repository which contains a .templateconf file that
sets TEMPLATECONF to point into a layer it contains, but you aren't
using that layer in your bblayers.conf, the eSDK would produce an error
during the preparation step of the installation. An example would be
using the poky repository but setting DISTRO to your own custom distro
and removing meta-poky from your bblayers.conf. The eSDK doesn't
support creating new build directories, so we don't care about the
templates and can thus force a known good value to prevent this from
happening.
Fixes [YOCTO #10568].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
The eSDK generation assumes that DL_DIR is downloads/ under the build directory,
and puts files such as a freshly buily uninative tarball in there expecting
bitbake will find it later.
Whilst ${TOPDIR}/downloads/ is in fact the default value for DL_DIR in
bitbake.conf, and any instances of DL_DIR are removed from the original
local.conf, there is still the possibility that other layers could contain a
site.conf that assigns DL_DIR.
If this happens the errors are quite mysterious as it fails to find the
uninative tarball and so the hashes all change, and eSDK building fails.
Ensure that this cannot happen by explicitly assigning the DL_DIR that we
require, instead of assuming that the default value will be used.
[ YOCTO #10439 ]
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
The Yocto Project Eclipse plugin requires that runqemu and unfsd are
accessible within the SDK, and indeed the standard SDK has these. This
turns out to be fairly easy to do - we just need to add unfsd and symlink
it, runqemu and a few other scripts into the SDK's bin directory.
Fixes [YOCTO #10214].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
After we run the build system within the eSDK internally as part of the
sstate filtering that happens during do_populate_sdk_ext, we need to
ensure that the TMPDIR created during that process gets deleted. However
we were using the TMPDIR path for the build producing the eSDK which may
not be the same (since that value would typically be filtered out) thus
if the user had set TMPDIR to something other than the default, the
temporary TMPDIR would not be deleted which not only led to extraneous
junk entering the SDK but also failures during install because the
TMPDIR path was different. In order to fix this, force TMPDIR to a known
value during the sstate filtering run so we know what to delete
afterwards.
Fixes [YOCTO #10210].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Tasks for image recipes cannot be locked and should be excluded from eSDK
generated locked-sigs.inc. get_sdk_install_targets() was not returning right
image targets to be excluded incase of 'minimal' sdk. This change fixes the issue.
Signed-off-by: Amarnath Valluri <amarnath.valluri@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Adding populate_sdk task to SSTATE_TASKS should make sstate machinery
to generate manifest for deployed ext sdk artifacts and do final deployment
to SDK_DEPLOY.
This is done in a similar way to do_populate_sdk in a previous patch.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Adding populate_sdk task to SSTATE_TASKS should make sstate machinery
to generate manifest for deployed sdk artifacts and do final deployment
to SDK_DEPLOY.
Set stamp-extra-info flag for do_populate_sdk task. This flag is used
in the name of sstate manifest. Setting it to predetermined value for
populate_sdk task should help to get correct manifest filenames when
processing runQueueTask events.
The do_populate_sdk function is also executed by do_populate_sdk_ext
so in order to avoid conflicts with the sstate postfuncs, split
the main code into a separate function.
We also need to set SDKDEPLOYDIR as do_populate_sdk_ext expects
it in order not to break ESDK generation.
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The preparation script itself prints out an error on failure, and we
aren't redirecting its output anymore, so we no longer need to print out
a message here when it fails. At the same time, make the message printed
out by the script a little clearer - we're just writing the log out to
the file, we shouldn't give the user an expectation that there will be
extra details in there (other than the output produced by
oe-init-build-env there won't be).
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Check a number of things as early as possible in the eSDK installer
script so that the user gets an error up front rather than waiting for
the build system to be extracted and then have the error produced:
* Check for missing utilities specified in SANITY_REQUIRED_UTILITIES
(along with gcc and g++), taking into account that some of these are
satisfied by buildtools which ships as part of the SDK. We use the
newly added capability to list an SDK's contents to allow us to see
exactly which binaries are inside the buildtools installer.
* Check that Python is available (since the buildtools installer's
relocate script is written in Python).
* Check that locale value set by the script is actually available
* Check that the install path is not on NFS
This does duplicate some of the checks in sanity.bbclass but it's
difficult to avoid that given that here they have to be written in shell
and there they are written in Python, as well as the fact that we only
need to run some of the checks here and not all (i.e. the ones that
relate to the host system or install path, and not those that check the
configuration or metadata). Given those issues and the fact that the
amount of code is fairly small I elected to just re-implement the checks
here.
Fixes [YOCTO #8657].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Determine the name of the current buildtools installer ahead of time,
set it in a variable and use that variable rather than the wildcarded
version everywhere, since it's much tidier.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
If the buildtools installation failed, we were using a subshell instead
of a compound command and thus the subshell exited but the script
continued on, which is really not what we want to happen. Additionally
log the buildtools installer output to a file and cat it if it fails so
that you can actually see what went wrong, as well as amending the
environment setup script to print a warning as we do when the
preparation fails.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
A couple of fixes for the recent sstate filtering implemented in OE-Core
revision 4b7b48fcb9b39fccf8222650c2608325df2a4507:
* We shouldn't be deleting the downloads directory here, since it
contains the uninative tarball that we will need
* TMPDIR might not be named "tmp" - in OE-Core the default is tmp-glibc
so use the actual name of TMPDIR here instead.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
If the build in which the eSDK is being built isn't using uninative,
this will have an effect on NATIVELSBSTRING, which will mean that the
eSDK installer won't be able to find any of the native sstate packages.
To keep things simple, under this scenario just disable uninative
temporarily while we run the SDK installer to help us check the presence
of the sstate artifacts we need. Ideally I'd rather not have things like
this that are artificial in this verification step, but on the other
hand this was the least ugly way to solve the problem.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
We were relying on uninative being enabled in the build in which the
eSDK was being produced, which is not the case for example for OE-Core's
default configuration. Move the code that copies the uninative tarball
and writes the checksum to copy_buildsystem so that it happens early
enough for that part of the configuration to be set up when we do the
filtering (which requires running bitbake).
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
If SDK_EXT_TYPE is set to "full" then we really ought to be shipping
everything that is expected to be in the SDK, and that includes gdb
(it's already referred to by the environment setup script if nothing
else). This is implemented by using the SDK_INCLUDE_TOOLCHAIN
functionality I just added, since the only material thing that adds on
top of a full SDK is gdb and we should always have the rest of it in a
full SDK anyway.
Fixes [YOCTO #9850].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Use the new oe-check-sstate to filter the sstate artifacts shipped with
the extensible SDK by effectively running bitbake within the produced
eSDK and and getting it to tell us which tasks it will restore from
sstate. This has several benefits:
1) We drop the *-initial artifacts from the minimal + toolchain eSDK.
This still leaves us with a reasonably large SDK for this
configuration, however it does pave the way for future reductions
since we are actually filtering by what will be expected to be there
on install rather than hoping that whatever cuts we make will match.
2) We verify bitbake's basic operation within the eSDK, i.e. that
we haven't messed up the configuration
3) We verify that the sstate artifacts we expect to be present are
present (at least in the sstate cache for the build producing the
eSDK). Outside deletion of sstate artifacts has been a problem up to
now, and this should at least catch that earlier i.e. during the
build rather than when someone tries to install the eSDK.
This does add a couple of minutes to the do_populate_sdk_ext time, but
it seems like the most appropriate way to handle this.
Should mostly address [YOCTO #9083] and [YOCTO #9626].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If we're to completely replace the standard SDK with the extensible SDK,
we need to be able to provide the standard toolchain on install without
doing anything other than installing it, so that you can install the SDK
and then point your IDE at it. This is particularly applicable to the
minimal SDK which normally installs nothing by default.
NOTE: enabling this option currently adds ~280MB to the size of the
minimal eSDK installer. If we need to reduce this further we would have
to look at adjusting the dependencies and/or the sstate_depvalid()
function in sstate.bbclass which eliminates dependencies, or look at
reducing the size of the artifacts themselves.
Implements [YOCTO #9751].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We don't absolutely need this - it doesn't change the default
behaviour, but it seems to me we have a convention to set default values
so we should add one here.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
During the extensible SDK installation process the final step is to
prepare the internal copy of the build system. This can take some time,
especially if you have SDK_EXT_TYPE set to "minimal" (downloading
sstate artifacts) and SDK_INCLUDE_PKGDATA set to "1" (restoring
pkgdata for world). To make this a bit less painful, use BitBake's new
quiet mode to display status during this operation so you have some idea
of how it's progressing; instead of redirecting the output to
preparing_build_system.log we grab the last console log and append it
instead.
One result of this change is that you get the errors printed on the
console during normal output rather than this going to the
preparing_build_system.log file first. In OE-Core revision
227d2cbf9e0b8c35fa6644e3d72e0699db9607fa, we changed to always print the
contents of preparing_build_system.log on failure, but now at least the
error contents of that log is duplicated. Besides, I intentionally
didn't print out the contents of that log during normal usage because
it's quite verbose - the bug that we were attempting to fix was about
not getting this information when seeing failures in the automated
tests, thus I've moved printing the log to the test handling code
instead.
Part of the implementation for [YOCTO #9613].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Tasks for image recipes cannot be locked - there's nothing to restore
from shared state to cover them and as a result, if you had "live" in
IMAGE_FSTYPES the build would fail with "taskhash mismatch" errors for
do_rootfs and do_image_complete for the initramfs image recipe, since it
had to try to run those. We should probably catch that issue earlier in
the build and produce a proper error, but for now at least exclude these
signatures from the locked-sigs.inc file so that extensible SDK
installers built when IMAGE_FSTYPES includes "live". (It turned out we
already had code to find other image tasks in the task list in order to
generate the list of install targets.)
Follow-up fix for [YOCTO #9826].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
If you build an extensible SDK for an image and IMAGE_FSTYPES includes
"live" then the extensible SDK will fail to install with a bunch of
unexpected task execution errors, matching the missing items required to
build the live image. The issue was we were still depending on do_rootfs
rather than do_image_complete. The fix was slightly more complicated
than just changing the task name as do_image_complete's dependencies are
in the form of dependencies on tasks within the same recipe (represented
in the "deps" varflag rather than the "depends" varflag).
Fixes [YOCTO #9826].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Use the new BB_SETSCENE_ENFORCE functionality to avoid having to run
bitbake twice on installing the extensible SDK - we can now do it all in
one invocation which not only takes less time, but we should also get
more meaningful errors for some types of failure, in particular where
downloading from an sstate mirror fails.
One result of this change is that you get the errors printed on the
console during normal output rather than this going to the
preparing_build_system.log file first. In OE-Core revision
227d2cbf9e0b8c35fa6644e3d72e0699db9607fa, we changed to always print the
contents of preparing_build_system.log on failure, but now at least the
error contents of that log is duplicated. Besides, I intentionally
didn't print out the contents of that log during normal usage because
it's quite verbose - the bug that we were attempting to fix was about
not getting this information when seeing failures in the automated
tests, thus I've moved printing the log to the test handling code
instead.
Part of the implementation of [YOCTO #9367].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This patch contains all the other misc pieces of the transition to
python3 which didn't make sense to be broken into individual patches.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
python3 standardises its use of iteration operations. Update
the code to match the for python3 requires.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
* Blacklist SSTATE_DIR, DL_DIR and TMPDIR by default to avoid problems
such as the one mentioned in [YOCTO #9605].
* Blacklist BB_NUMBER_PARSE_THREADS since we already blacklist
BB_NUMBER_THREADS and they may be set separately.
Fixes [YOCTO #9605].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
It has been determined that it is highly likely that users might get
signatures that don't match in an extensible sdk. This doesn't
necessarily happen with oe-core, so we can set the mismatch to an error
during testing if we like.
However, for the case where users are creating their own sdks, we don't
need an error halting their progress. locked-sigs will still function as
it should.
Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
To avoid lots of output in the SDK EXT installation phase, system redirects
it to a logfile ($target_sdk_dir/preparing_build_system.log) but in case of error,
the contents should be shown so debugging could be faster.
[YOCTO #9576]
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
When a recipe is added to the workspace, the signatures for the tasks
will change. This means that bitbake must be told to allow the
signatures to be different if they are in locked-sigs.inc.
This is done by creating an unlocked-sigs.inc file which contains all
the recipes in the workspace each time devtool reads the workspace.
So not only will necessary things get added, previously added items will
be removed by virtue of them no longer being in the workspace.
This also makes sure that the extensible sdk picks up unlocked-sigs.inc
as part of the configuration.
[YOCTO #9195]
Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
With the extensible sdk we want there to be an error if a task tries to
run without signatures that match locked-sigs.inc. This patch enables
that error.
[YOCTO #9195]
Signed-off-by: Randy Witt <randy.e.witt@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Otherwise the task hash changes between server and worker context
leading to changing task checksums. The dependency data here is
tracked by other pieces of the signature.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If SDK_EXT_TYPE is set to "minimal" then the SDK won't contain many
sstate artifacts, and you're required to set up an sstate mirror in this
case anyway so there's no point publishing the "stub" sstate-cache
directory from within the SDK since it won't be useful for update
purposes and may be confused with the real sstate-cache.
There is however a possibility that people might publish the real
sstate-cache directory under the same output directory provided to
oe-publish-sdk, thus deleting it after extracting (as we were doing with
other files we wanted to clean up at the end) would be problematic,
besides which extracting it and then deleting it is wasteful. Thus,
introduce a "-p" command line option to the SDK installer that we can
use to tell tar not to extract the items we don't want when publishing.
This has the added benefit of mostly keeping references to these in the
place they belong i.e. in populate_sdk_ext.bbclass.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When running bitbake you may pass in values of variables from the
external environment (making use of BB_ENV_EXTRAWHITE), and you may
choose to do this when building the extensible SDK, for example:
MACHINE=qemuarm bitbake -c populate_sdk_ext core-image-minimal
You would naturally expect those settings to be reflected in the
extensible SDK itself; however they were not, since we were only
considering local.conf and auto.conf. Check the variables mentioned in
BB_ENV_EXTRAWHITE to see if any are different than the values set in
local.conf/auto.conf and add lines setting them in the SDK's local.conf
if so.
Fixes [YOCTO #9339].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
There are various concerns about md5 so use sha256 instead.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Instead of skipping the build system preparation step within the
extensible SDK install process when SDK_EXT_TYPE is "minimal", run
bitbake -p so that the cache is populated ready for the first time
devtool is run.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Commit 8b81bb56c69aabdea984352f8e267a9783c0bdbc was accidentally merged.
The DL_DIR piece was simply incorrect and should be removed.
The patch commit message should have mentioned that the changes were
to update populate_sdk_ext after the changes to uninative now the
download is placed into a specific directory in DL_DIR. We also
need to specify the uninative tarball checksum.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
Patrick Ohly
Richard Purdie
Paul Eggleton
Ed Bartosh
Joshua Lock
Ross Burton
Robert Yang
Amarnath Valluri
Randy Witt
Leonardo Sandoval