| Age | Commit message (Collapse) | Author |
|---|
|
One of the uses of externalsrc is to enable iterative editing and
rebuilding of source files during development. In such situations,
disabling Automake dependency tracking can lead to sources not being
rebuilt even though files they depend on have been modified.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
It's possible that a trailing or extra slash somewhere in the external
source path could result in the directory not being removed from
cleandirs; it's also possible that a cleandirs entry is somewhere
underneath the source tree and that tree should never have parts of it
deleted by the build system. Use oe.path.is_path_parent() (which makes
paths absolute before checking them) to find out if any path in
cleandirs is anywhere underneath the external source path, and drop it
if it is.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Use the existing oe.recipeutils.split_var_value() function to split the
unexpanded value of the cleandirs varflag, in case it contains python
expressions - we don't want to split the expression itself as the chunks
will not expand properly individually and we can miss something that
expands to the source tree (and thus it can get deleted, the avoidance
of which is the whole point of this code).
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
ERROR: distrho-ports-0.0.0+git999-r0 do_buildclean: oe_runmake failed
ERROR: distrho-ports-0.0.0+git999-r0 do_buildclean: Function failed: do_buildclean
ERROR: Logfile of failure stored in: <...>/temp/log.do_buildclean.17285
Log data follows:
| DEBUG: Executing shell function do_buildclean
| NOTE: make clean
| make clean -C libs/drowaudio
| make[1]: Entering directory '/home/a.mueller/data/oe-core/workspace/sources/distrho-ports/libs/drowaudio'
| make clean -C build-drowaudio
| make[2]: Entering directory '/home/a.mueller/data/oe-core/workspace/sources/distrho-ports/libs/drowaudio/build-drowaudio'
| make[2]: *** No rule to make target 'clean'. Stop.
| make[2]: Leaving directory '/home/a.mueller/data/oe-core/workspace/sources/distrho-ports/libs/drowaudio/build-drowaudio'
| make[1]: *** [Makefile:7: clean] Error 2
| make[1]: Leaving directory '/home/a.mueller/data/oe-core/workspace/sources/distrho-ports/libs/drowaudio'
| make: *** [Makefile:73: clean] Error 2
| ERROR: oe_runmake failed
Signed-off-by: Andreas Müller <schnitzeltony@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Suppress any warnings git might generate when searching for a valid git
directory, as there are use cases where the directory is expected to not
exist and the warning is superfluous
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The externalsrc class code assumes that the source directory
(EXTERNALSRC) exists before bitbake is called. Otherwise do_configure
will fail obviously since externalsrc does not fetch anything.
Commit 3ca6085729d9 ("externalsrc: Handle .git not being a directory")
changed this behaviour. Now on a missing EXTERNALSRC directory we get
a bb.data_smart.ExpansionError during _parsing_, way before
do_configure can be run.
This new behaviour creates two problems:
* First, there error message is very cryptic (and it's hard to
provide a better message since no task is ever run):
ERROR: ExpansionError during parsing /<...>/<...>.bb
Traceback (most recent call last):
bb.data_smart.ExpansionError: Failure expanding variable do_compile[file-checksums], expression was ${@srctree_hash_files(d)} which triggered exception FileNotFoundError: [Errno 2] No such file or directory: '<...>'
* Second, this prevents creating a class based on externalsrc that
automatically fetches the code in EXTERNALSRC before do_compile
runs.
Fix both problems by simply calling git with '-C ${EXTERNALSRC}'
instead of calling git inside the non-existing directory. This changes
from a bb.data_smart.ExpansionError to a
subprocess.CalledProcessError, which is in line with what's actually
going on: git is telling us it can't find the git dir.
Also remove a comment that does not apply anymore.
Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Cc: Joshua Watt <jpewhacker@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
To help users find the work and log directories (especially within the
eSDK) we create symlinks to these from the source tree. However during
do_buildclean we call "make clean", and some project Makefiles which
delete something like "*/*/lib" which will match files underneath the
oe-workdir symlink and fail. do_buildclean is called from do_clean which
is in turn called by devtool reset by default, and thus devtool reset is
blocked. An example of a recipe where this is visible is the openssl-qat
recipe in meta-intel.
In order to fix this, delete the symlinks at the start of do_buildclean
since we shouldn't need them at that point anyway, and thus make clean
won't be able to trip over them.
Fixes [YOCTO #11036].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Use git rev-parse to determine the location of the .git directory, in
case it is not an immediate child of EXTERNALSRC (e.g. when using
submodules). In the event git can't resolve the .git directory, fall
back to the non-git method for hashing.
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Using shutil.copy2() to copy .git/index to a temporary file tries to
copy SELinux attributes which might fail for confined users in SELinux
environments.
E.g. our builders are running in docker containers and modification of
sources (inclusive updated of .git/index) is done outside. Trying to
copy .git/index fails with
| $ python3 -c 'import shutil; shutil.copy2("index", "a")'
| ...
| PermissionError: [Errno 13] Permission denied: 'a'
and an AVC like
| denied { relabelto } for pid=18043 comm="python3" name="a" dev="dm-29" ino=1067553 scontext=system_u:system_r:container_t:s0:c39,c558 tcontext=unconfined_u:object_r:build_file_t:s0 tclass=file permissive=0
is created. This can not be solved by adapting the SELinux policy because
this is a very deep constraint violation:
| constrain file { create relabelfrom relabelto } ((u1 == u2 -Fail-) or (t1 == can_change_object_identity -Fail-) ); Constraint DENIED
|
| Possible cause is the source user (system_u) and target user (unconfined_u) are different.
I do not see much sense in using 'shutil.copy2()' here; 'shutil.copyfile()'
seems to be a better choice (target file is created in a secure way by
tempfile.NamedTemporaryFile()).
By placing the tempfile into /tmp we avoid potential problems related to
git's 'core.sharedRepository'. As a (positive) side effect, the source
tree will not be modified anymore (at least by this part of code) which
prevented to mount it read-only from somewhere else.
Signed-off-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If these are set to URLs then the errors produced are not helpful.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
I've been debugging a selftest failure on Centos7. The problem turns out
to be the elderly git version (1.8.3.1) on those systems. It means that
the system doesn't correctly checksum changed files in the source tree,
which in turn means do_compile fails to run and this leads to the following
selftest failure:
======================================================================
FAIL [141.373s]: test_devtool_buildclean (oeqa.selftest.devtool.DevtoolTests)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/pokybuild/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/meta/lib/oeqa/selftest/devtool.py", line 530, in test_devtool_buildclean
assertFile(tempdir_mdadm, 'mdadm')
File "/home/pokybuild/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/meta/lib/oeqa/selftest/devtool.py", line 497, in assertFile
self.assertTrue(os.path.exists(f), "%r does not exist" % f)
AssertionError: False is not true : '/tmp/devtoolqag88s39z8/mdadm' does not exist
The solution is to use -A on the git add commandline which matches the behaviour
in git 2.0+ versions and resolves the problem.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Add created symlinks to the exclude file. This will both make them
less distracting and hide them from the srctree_hash_files function.
Signed-off-by: Ola x Nilsson <olani@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
The buildclean task should call the package build system clean
command, just implemented for Make for now.
This is meant for recipes where S == B, but can be useful as a
standalone task for other recipes too.
When S == B, set it to run before do_clean which will do what most
developers expect when calling bitbake -c clean. For S != B, do not
add it before clean as it is not needed and may take some time.
Signed-off-by: Ola x Nilsson <olani@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Make it easier to reuse the function for other dirs than EXTERNALSRC.
Signed-off-by: Ola x Nilsson <ola.x.nilsson@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Only the last stamp file should be kept, but unless STAMPCLEAN matches
files generated using STAMP old stamp files may linger. This may
cause false positives for skipping tasks.
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
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>
|
|
If the user modifies files such as CMakeLists.txt in the case of cmake,
we want do_configure to re-run so that those changes can take effect. In
order to accomplish that, have a variable CONFIGURE_FILES which
specifies a list of files that will be put into do_configure's checksum
(either full paths, or just filenames which will be searched for in the
entire source tree). CONFIGURE_FILES then just needs to be set
appropriately depending on what do_configure is doing; for now I've set
this for autotools and cmake which are the most common cases.
Fixes [YOCTO #7617].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
To make sure changes to any source files are detected when externalsrc
is used, it sets BB_DONT_CACHE to force the recipe to be reparsed
every time. Previously, this was done conditionally based on whether
EXTERNALSRC was set. This worked fine for building the base recipe.
But if one tried to build, e.g., a native version of it (provided via
BBCLASSEXTEND), the recipe would not be reparsed as expected.
To solve the above problem, BB_DONT_CACHE is now set for the base
recipe if EXTERNALSRC is set for it or any of it derivatives.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
In python3, strings are unicode by default. We need to encode/decode
from command pipelines and other places where we interface with the
real world using the correct locales. This patch updates various
call sites to use the correct encoding/decodings.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Use a unique tempfile as the temporary git index file when determining
the git hash of the source tree. A fixed filename was obviously causing
races (with the git index.lock file) under oe-selftest where multiple
bitbake instances were run against the same source tree at the same
time.
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Extend the srctree_hash_files() function in externalsrc.bbclass to
handle non-Git source trees. If the source tree is not a git
repository, srctree_hash_files() now simply adds the whole source tree
as a dependency, causing bitbake to basically hash every file in it.
Hidden files and directories in the source tree root are ignored by the
glob currently used. This has the advantage of automatically ignoring
.git directory, for example.
During the first bitbake run preparing of the task runqueue may take
much longer if the source tree is not a git repository. The reason is
that all the files in the source tree are hashed. Subsequent builds are
not significantly slower because (most) file hashes are found from the
cache.
[YOCTO #8853]
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Be a bit more intelligent than mindlessly re-compiling every time.
Instead of always using 'nostamp' flag for do_compile run a python
function to get a list of files to add as 'file-checksums' flag. The
intention is to only re-run do_compile if something in the source tree
content changes.
This python function, srctree_hash_files(), works differently, depending
if the source tree is a git repository clone or not. If the source tree
is a git repository, the function runs 'git add .' and 'git write tree'
to get a hash of the working tree and writes this hash into a file under
the .git directory. This file containing the hash is then returned as
the file for the task to depend on. Hash is changed if any changes are
made in the source tree causing the task to be re-run. A trick is used
to parse the recipe every time so that the hash file gets updated.
If the source tree is not a git repository behaviour remains the same.
In this case srctree_hash_files() currently sets the 'nostamp' flag for
do_compile causing it to be re-run every time.
This method of tracking changes source tree changes to determine if
re-build is needed does not work perofectly, though. Many packages are
built under ${S} which effectively changes the source tree causing some
unwanted re-compilations. However, if do_compile of the recipe does not
produce new/different artefacts on every run (as commonly is and should
be the case) the re-compilation loop stops. Thus, you should usually see
only one re-compilation (if any) after which the source tree is
"stabilized" and no more re-compilations happen.
[YOCTO #8853]
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We have a common build directory for all target architectures if
externalsrc is used and ${B}=${S}. In this case we also need to have a
common stamp directory. The reason is that e.g. changing MACHINE will
basically "invalidate" the artefacts in ${B} (wrt. to the new MACHINE)
but old stamp files with matching hashes from an earlier build might be
available in the machine-specific stamps directory and tasks would not
be correctly re-run.
Previous attempt in f44f12af346888bdeb3ae01a275cb5dd4396b505 to handle
this correctly was faulty as it threw some spurious warnings and caused
do_configure to be re-run in every build if that would not be necessary.
[YOCTO #8950]
[YOCTO #9237]
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
If the oe-workdir / oe-logs symlink exists and points to a different
path then the symlink needs to be removed before calling os.symlink() or
it'll fail.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Auto-create symlinks in the source directory to the work directory
(${WORKDIR}) and logs directory (${T}) so that they are easier for the
user to find. This is particularly useful within the extensible SDK
where the user is less likely to be familiar enough with the structure
of the build system to know where to find these things, but otherwise
they are a useful shortcut for anyone.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If you're using externalsrc, it's very likely that you're going to want
to examine the intermediate build results even if the recipe builds
successfully; therefore you won't want rm_work to delete those.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This sets the scene for removing the default False for expansion
from getVarFlag. This would later allow True to become the default.
On the most part this is an automatic translation with:
sed -e 's:\(\.getVarFlag([^,()]*, [^,()]*\)):\1, True):g' -i `grep -ril getVar *`
In this case, the default was False, but True was used since in most
cases here expansion would be expected.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
External source tree is used as the builddir even for different
architectures in case ${B}=${S}. This may cause problems if MACHINE is
changed as do_configure is not being re-run (which would run make
clean). This patches changes externalsrc to use a common (per-recipe)
CONFIGURESTAMPFILE under 'work-shared' if ${B}=${S}. In addition,
do_configure will depend on changes of this stamp file. As a result,
do_configure is re-run and the build dir is cleaned correctly if a
rebuild for different MACHINE is done.
[YOCTO #8950]
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The value of the deps varflag for tasks is (unusually) expected to be a
list object rather than a space-separated string, hence we cannot use
appendVarFlag() here. This fixes a traceback when parsing the gcc recipe
with externalsrc enabled, for example.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
It turns out that there are folks out there who use externalsrc in
normal builds and don't really need to be warned; additionally within
the extensible SDK or when using devtool, it shouldn't be a warning
situation. Thus, scale it back to a note (we can't use bb.note() here
since that wouldn't actually be piped through to the bitbake UI). Also
touch up the message a little bit.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We really do not want our (external) source tree to be removed. There
might be multiple values in the 'cleandirs' varflag pointing to our
source tree - causing it to be wiped out. This patch improves the
filtering of 'cleandirs' by examining the expanded values inside it. Any
(expanded) values pointing to our source tree get removed.
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Kernel metadata repository needs to be fetched/unpacked into the work
directory in order to build the kernel. Sources pointing to a remote
location are skipped by externalsrc by default which caused kernel build
to fail (because of remote kernel-meta was not made availeble). This
patch will make kernel-meta always available.
[YOCTO #6658]
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We need to ensure we add a leading space to the value we are prepending
here in case lockfiles already has a value.
Fixes [YOCTO #7813].
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
If SRC_URI contains local files (file:// references) these will almost
certainly be required at some point during the build process, so we need
to actually fetch these to ${WORKDIR} as we would normally.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Make sure there's no chance of anyone forgetting they have a recipe set
up for externalsrc; otherwise you could get confused about what is going
on. (With our default logging setup we can't make it a note because the
UI doesn't forward those; otherwise I would have used bb.note().)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Here we set SRC_URI to blank, however doing so means that the function
that is called when you expand the default value of SRCPV
(i.e. bb.fetch2.get_srcrev()) will fail, so any recipe that references
SRCPV in PV couldn't previously be used with externalsrc.
(At some point we may fix the function to work in the externalsrc case,
but then we would also need to ensure that ${B} did not change as a
result of PV changing any time the HEAD revision changes in the external
source tree, or you'll lose any intermediate build artifacts.)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
kernel.bbclass adds ${S} do do_clean[cleandirs], but this means if you
run bitbake -c clean <kernelrecipe> then your external source tree will
be trashed, which could be a disaster. For safety, remove ${S} from
cleandirs for every task. We also have to do the same for ${B} in the
case where EXTERNALSRC_BUILD is set to the same value as EXTERNALSRC.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Most of the time what you want when using this class is for do_compile
to execute more than just once - every time the source changes would be
ideal, but that's a little tricky to accomplish. Thus, set do_compile as
nostamp to get something close. Note that in order to be effective this
also requires the change to bitbake that causes nostamp task signatures
to change on each execution.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
An incorrect version of the patch merged which entirely removed the covered
variable. This corrects the patch so the code works.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Now we have deltask API, stop poking around bitbake internal variables.
(From OE-Core rev: 98637df0c04fd14b506d9eec7da8ec6ae441221b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This class can now be inherited globally using INHERIT += rather than
needing to inherit it in the recipe itself. Additionally, instead of
setting S (and optionally B), set EXTERNALSRC (and optionally
EXTERNALSRC_BUILD) to point to the external source and build locations
respectively.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This is loosly based upon srctree.bbclass from OE-Classic but with some
changes appropriate to OE-Core.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
Andre McCurdy
Paul Eggleton
Andreas Müller
Joshua Watt
Luca Ceresoli
Enrico Scholz
Ross Burton
Richard Purdie
Ola x Nilsson
Joshua Lock
Peter Kjellerstedt
Markus Lehtonen