Age | Commit message (Collapse) | Author |
|
The SendOePerfEmail was taking controller hostname instead of
worker one when the performance script was run.
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
|
|
We need to store results by buildhost to make comparisions against
previous data and find regressions or improvements on performance.
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
|
|
The commit results functionality was remove [1] from the
oe-build-perf-test script and new script was create called
oe-git-archive.
A buildstep needs to be created in order to publish results
into git futher.
[1] http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=1440b3e9aa98c4b5f846226c4a5b433325313907
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
|
|
No need to ensure layerversion carries a value and cast it to an int as we
both convert to an int and ensure the value is not-None at property get
time.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
core-image-sato-sdk only builds on musl for master/rocko or newer, older
releases will build core-image-sato-sdk. Ensure we're testing the same
image we built.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
core-image-sato-sdk only builds with master (eventually rocko) and should
be replaced with core-image-sato for older releases. master/rocko has
LAYERVERSION = "11" so replace for older layerversion values.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
We replaced buildbot's standard copy function with our _srccopy
function when method=barecopy or movecopy. The srccopy will copy the
contents of the "source" directory into the build dir. We cannot
assume that submodules have not been added since the buildslave last
populated the source dir. Adding --init to the git submodule update
resolves this issue.
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com>
|
|
>From some quick tests the string translate function is comparable in
efficiency and certainly more legible than regex. We should note that
if we ever move this functionality to an environment running python3,
the translate function takes one dictionary argument, but functions
in much the same way.
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com>
|
|
Replace the core-image-sato build/test with core-image-sat-sdk to provide
a slightly larger test-case for musl builds.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
These aren't used by YP QA or published for a YP release, therefore stop
publishing them for builds and save a bit of time and storage space.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Bitbake has many assignment operators, around which there may or may not be
space characters. Try to be more flexible in the format to ensure the target
variable is correctly scraped.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Using awk, sed, or grep to pull a shell variable out of stdio proved
complex. Instead, simply cat the entire "inc" file to stdio and use
python/regex to find the variable.
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com>
|
|
The way we source refkit's oe-init-build-env from a different path to the
scripts location isn't POSIX shell compliant, workaround this by using the
forcebash option to RunPreable and BuildImages.
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com>
|
|
Certain (Debian-derived) distros use a less desirable shell for /bin/sh
(dash). In some edge cases, if we source oe-init-build-env from
outside the OEROOT folder, the script will fail (as stated in the
oe-init-build-env script). By adding a variable to force bash as the
shell, we can call "bash -c" to ensure that a compatible shell is
used.
Currently, refkit is the only build that suffers from this edge case.
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com>
|
|
The new variable SSTATE_RELEASE_DIR, when set, will be used as the
SSTATE_DIR for release builds. The usual SSTATE_DIR will be configured as a
file SSTATE_MIRRORS entry.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
If the IMAGE_FSTYPES variable in conf/autobuilder.conf is set to the empty
string this should not result in an empty append assignment.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
|
|
nightly-qa-* are small buildsets which have overheads; scheduling to
a distinct worker, building an image with full tmpdir population, etc.
By folding these into the existing nightly-qa-extras buildset we can benefit
from build artefact reuse and therefore reduce the amount of time this test
build takes.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
This build isn't currently triggered and therefore is of little value.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
When scraping the build targets from refkit-ci.inc, grep would only
return the first line of the search result. By replacing grep with
awk, we are now searching for multi-line variables, and should now
scrape build targets even if they have line breaks between them.
[YOCTO #11720]
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com>
|
|
Since we are parsing and building all the images in refkit-ci.inc,
there are some layers that are now required. This includes flatpak,
industrial, and ROS layers. Since we are building a dynamic set of
images based on this inc file, we should be adding these layers
dynamically as well at some point.
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com>
|
|
Adapt to the recent change in shared state configuration for releases:
* fix source path (sstate_dir) to match new scheme
* use --dereference to perform a "deep-copy" of symlinks
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
When making a release we create a separate SSTATE_DIR (with the major
release number in the SSTATE_DIR's name) and use the shared SSTATE_DIR as
an SSTATE_MIRROR.
This simplifies things because we no longer rely on SSTATE_MIRROR having
been set in order to create the directory of sstate artefacts for a release
and further we no longer need to traverse the network to fetch sstate
objects when making a release build.
The downside is that the publishing of the artefacts now requires a "deep"
copy where we detect symlinks and follow them to ensure the file is copied,
rather than the link (i.e. pass -L or --dereference to cp).
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
The mkdir calls for a release managed to include a typo in a variable name
(pipleline→pipeline) and use the wrong directory variable name
(pub_dir→rel_dir).
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
We'd like the nightly-refkit buildset to be automatically triggered by the
nightly buildset to smoke-test refkit with pre-merge OE-Core and Bitbake
changes.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
intel-iot-refkit is new enough that it's very unlikely to build with older
versions of OE-Core.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
We perform various tests on the build targets and layer versions to prevent
attempts to run builds which are untested and likely to fail. These tests
should also be run on any scraped targets.
We therefore move the processing of scraped targets earlier in the code and
allow the usual logic which checks targets and layers to also process the
scraped targets.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Check whether GitPoller for this repository already is in yocto-sources and
only add it if not. This prevents an assertion error occuring in buildbot
when two different buildset configurations register a GitPoller for the same
repository.
Signed-off-by: Johannes Schrimpf <johannes.schrimpf@blueye.no>
|
|
Add buildsteps to test shared state reuse and that the shared state
continues to woerk when the original build directory is no longer present.
[YOCTO #11500, #11497]
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
intel-iot-refkit is a distro and set of layers geared towards IoT development.
This buildset-configuration will perform a build of an image for the
intel-iot-refkit distribution as well as the same image for the poky
distribution, thereby testing both that intel-iot-refkit continues to work
against changes in OE-Core/Bitbake and that the intel-iot-refkit layers and
profiles are distribution agnostic.
[YOCTO #11428]
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
We currently default to minimal eSDK builds when building Pyro or newer,
a scenario which leaves full eSDK builds untested on the YP AB.
Add an eSDK build to nightly-qa-extras which builds a full eSDK for
core-image-minimal and core-image-sato in an attempt to detect bugs in the
full eSDK build.
[YOCTO #11366]
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
atextappend specifies text to be appended at the end of generated the
auto.conf, yet some recent changes have resulted in entries being written
to auto.conf *after* the atextappend. This resolves that so that the
atextappend value is the final thing to be written to the generated
auto.conf file.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
This buildstep enables a buildset to remove files as a step within the
build. With zero parameters this buildset will delete the tmp subdirectory
of the build directory.
If the buildstep is passed a files parameter the value of that parameter
will be joined with the build directories path to form a path argument to
the rm command.
The buildstep attempts to ensure the resulting path TMPDIR, or a child of
TMPDIR, skipping otherwise.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Fix the tmpdir argument so that the value is include after the variable
name and assignment operator in the output. Further, add logic so that if
the argument doesn't begin with a / the provided value is assumed to be
relative to the build directory, rather than a full path.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Initialisation scripts can set environment variables which we'd rather unset
or override before invoking bitbake, i.e. a custom init script may set
OEROOT but for the YP autobuilder's purposes pointing to the OE bundled with
the poky repository is a much more frequent workflow (in order to test the
layer with custom init script against pending changes in OE-Core and
Bitbake).
Add overrideenv variable which takes a list of VARIABLE=ASSIGNMENT pairs.
The assignment value may include the term #YPDIR which will be replaced
with the path to the top-level directory layers are fetched to, the parent
of the meta (OE-Core) layer.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
The bitbake parser doesn't like spurious whitespace between an assignment
operator and the value to be assigned. In order to prevent issues with the
generated bblayers.conf use tr to strip any extra spaces from the value we
cut from the skeleton bblayers.conf
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Add the paths and pattern to search for DISTRO_VERSION when building
intel-iot-refkit.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
We already have support for running a pre-defined list of tests, this
change enables use of the ScrapeTargets buildstep to pull the list of tests
to run from the layers being built. If the tests property of the buildstep
is set to #SCRAPEDTARGETS we try and read the scraped_targets property and
set its value as the list of tests to run.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Support building targets which have been saved in a property by the
ScrapeTargets step.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Simple buildstep which can scrape a given variable from a file in a checked
out layers contents and save the value of the variable to a property for use
in future buildsteps.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Some layers provide their own environment initialisation script, add a the
configuration item to enable use of a non-standard environemtn
initialisation script and default to oe-init-build-env.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Remove unused ast module importn and remove python shebang.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Simplify code a bit by using dict.get() to supply a default, empty, value
when accessing a key, rather than detecting whether a buildset definitio
in the buildset-config has a 'repos' key or not and calling the BuildSet
constructor differently we can just pass the dict returned by get() and know
that the default empty dict will not cause issues.
Note: BuildSet doesn't appear to handle a None value for the layers variable
of its constructor so the old code path was likely to cause problems were it
not for the fact that a buildset definition without layers is useless.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Prettify the links to error logs by creating a named link of the step name,
rather than using bare links.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
|
|
Update yoctoAB.conf to reflect what is currently used in production and
add nightly-musl and nightly-x32 builds to the configuration.
Signed-off-by: Graydon, Tracy <tracy.graydon@intel.com>
|
|
Add the new nightly-musl and nightly-x32 builds to nightly-meta-intel
build scheduler triggers.
Signed-off-by: Graydon, Tracy <tracy.graydon@intel.com>
|
|
Add nightly-x32 buildset with world build for meta-intel autobuilder.
Signed-off-by: Graydon, Tracy <tracy.graydon@intel.com>
|
|
Add nightly-musl buildset for meta-intel autobuilder.
Signed-off-by: Graydon, Tracy <tracy.graydon@intel.com>
|