Commit 72f416f382 added a horizontal
scan for .yaml files when a modification was made in "common" folder.
If yamls were found in such way, the loop ended. However, the
implementation didn't address what happens if such yamls are not found.
This made the script going into an infinite loop. If yamls are not
found next to "common", the script should proceed as before, i.e. go
to the directory above an start looking there.
Signed-off-by: Maciej Perkowski <Maciej.Perkowski@nordicsemi.no>
Some tests use a "common" directory to store pieces of code which
are reused by different scenarios. In those cases, no test yaml is
found within such director nor within its parents.
If no test yaml is found in a directory, and the directory is called
common, also look in collocated directories if they have test yamls.
If so, add all those locations to the scope.
E.g. tests/bluetooth/controller/common
Signed-off-by: Maciej Perkowski <Maciej.Perkowski@nordicsemi.no>
The script was not resolving all detected changes uniformly:
find_excludes() could skip or not certain patterns based on
required testing scope. The idea was to not include files
that were already handled by find_test() and find_boards() workflows.
However, only boards and tests folders could be removed but
not samples. This also led to blind spots: changes in some files
were not triggering any tests. E.g. change in a test/common, where
no corresponding yaml can be found by find_tests() which is also
ignored by find_excludes(). In the new workflow a list of resolved
files (for which find_arch(), find_tests() or find_boards()
found scope) is created. Instead of using skip in find_excludes,
files are excluded only if they were resolved.
Signed-off-by: Maciej Perkowski <Maciej.Perkowski@nordicsemi.no>
The arg --testsuite-root was copied from twister. When it is used
for test_plan.py it will be propagated to twister calls. This allows
to make alternative test locations (e.g. from another repo) to work
with test_plan.py
Signed-off-by: Maciej Perkowski <Maciej.Perkowski@nordicsemi.no>
Add --alt-tag arg for test_plan.py script. User can use it and point
to an alternative file with tag-directories relations. If so, such
file will be used instead of the default one.
Signed-off-by: Maciej Perkowski <Maciej.Perkowski@nordicsemi.no>
Expand test_plan.py args with --ignore-path. This allows to provide an
alternative lists of patterns for the script.
Signed-off-by: Maciej Perkowski <Maciej.Perkowski@nordicsemi.no>
The test_plan.py script has a path to repository to be scanned for
changes hard coded to zephyr. This patch separates zephyr path from
such repository's path and adds an arg to pass repo to scan
Signed-off-by: Maciej Perkowski <Maciej.Perkowski@nordicsemi.no>
An option --no-detailed-test-id was added to twister to help align
names for test outside of zephyr tree. This commit add this arg
to test_plan.py script which is then propagated to twister.
Signed-off-by: Maciej Perkowski <Maciej.Perkowski@nordicsemi.no>
Add platform filtering when generating tests for manifest modules.
In the clang workflow, we do select to run only on one platform, so this
needs to be applied as well or we end up building unwanted tests using
the wrong toolchain.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
We have been launching 10 runners up a certain number of tests, although
we only neeed half of that for some scenarios.
Too many runners started that have to execute just a small number of
tests wastes times on setup and blocks the queue. Just start the number
of nodes needed based on initial calculation.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Disables allowing the python argparse library from automatically
shortening command line arguments, this prevents issues whereby
a new command is added and code that wrongly uses the shortened
command of an existing argument which is the same as the new
command being added will silently change script behaviour.
Signed-off-by: Jamie McCrae <jamie.mccrae@nordicsemi.no>
path for posix was not being matched correctly due to a missing /. So
some posix tests were excluded and not excercised.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
This commit calculate memory footprint from build.log and
proposes an alternative approach to #2465.
Signed-off-by: Katarzyna Giądła <katarzyna.giadla@nordicsemi.no>
The commit c3620c8a8e changed the
`list_boards` script such that it no longer searches the boards from
the `ZEPHYR_BASE` path by default.
This commit updates the `test_plan` script to invoke the `list_boards`
script with the `ZEPHYR_BASE` as a search path.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
In cases of global changes where 100s of nodes are launched, i.e. on
samples and tests (more than 20 tests/samples changed), do a full
covrage run.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Now that we have partially oerhauled twister in the tree, modify actions
with new options:
- not reporting filtered tests is now default
- output is json, not csv
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
twister now exports json instead of csv, so we make the testplan read
the json data and generate the consolidated plan for the workflow.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
When doing global changes, like typo fixes or header changes, keep
default scope and do not build each test for each platform.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Use the common code in scripts/list_boards.py for finding matching
boards in scripts/ci/test_plan.py. This has the benefit of not trying to
use board revisions files as board names.
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
If we are invoking testplan for only one board, do not bother with other
board related changes, just generate plan for the specified platform.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
We were missing smaller sets and skipping tests when the set is smaller
than the allowed number of tests per node.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Improve calculation of matrix and move calculations from workflow to the
testplan script. We now generate a file that can be parsed by the action
with the data needed to start twister with the right number of nodes.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Do not invoke --integration when dealing with one platform only and
generate testplan only for the needed platforms.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Merged 3 files used to generate the test plan per PR based on the
changes. 2 python scripts and a shell script are now all merged into 1
python script that generates the input file for twister based on a list
of changed files by the PR.
This remove lots of old and obsolete code and simplifies things a bit,
no need anymore for an intermediate script to call twister, we call it
directly in the workflow and use the new test_plan script to generate
the test plan.
This also reenables the recently disabled tag based filtering which had
a bug, bug is resolved in this new implementation.
On push events, we now run twister without the --integration option to
catch any issues in the main branch that were not caught in PRs. PRs
continue to run with --integration enabled. This event (push) is now run
on 15 builders due to the increased size.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
