On qemu_cortex_a53 with icount enabled, the qemu process will spin
hard at the end of its run (common with icount), but will also ignore
the SIGTERM twister is sending to terminate the (successful!) test
run. The resulting zombie spinning processes end up accumulating and
poisoning the test run.
Just use a hard KILL instead of trying to be a good citizen (and in
lieu of fixing the upstream qemu bug). I can't see why this should be
a problem.
Signed-off-by: Andy Ross <andyross@google.com>
Fix the missing logic of setting the flash_error flag when
caught the exceptions. And we should let the serial thread
can be run first if flash error. Otherwise the twister will
hang up while an exception happens on closing serial, after
the #47820 got merged. This mostly happens on frdm_k64f.
Fixes#49086
Signed-off-by: Enjia Mai <enjia.mai@intel.com>
Add parameter for possible extending in the future, --tool-opt
passes arguments to execution of the run service.
Ex.
--tool-opt=--arg='white space' --tool-opt=-r --tool-opt=12345
will be parsed as ['--arg=white space', '-r', '12345']
Signed-off-by: Enjia Mai <enjia.mai@intel.com>
Because we use INADDR_ANY by default now, specifying a net interface
is no longer needed. Remove it and remove the dependency of the
netifaces python package.
Fixes. #48584
Signed-off-by: Enjia Mai <enjia.mai@intel.com>
Add specifying the port for the remote cavstool remote service.
Ex.
west flash --remote-host {host}:{port} \
--pty {host}:{port}
Specify the port is optional when running west.
And another major change is now we have to specify the remote log
service by --pty to make it output the log message immediately.
Previously it will output directly. Why we make this change is
because this is more close to the behavior we use west flash.
Fixes. #46865
Signed-off-by: Enjia Mai <enjia.mai@intel.com>
This patch introduces the basic infrastructure to list and fetch binary
blobs. This includes:
- The new 'blobs' extension command
- An implementation of the `west blobs list` command
with custom formatting
- A very simple mechanism for loading fetchers
- A basic implementation of an HTTP fetcher
In order to ensure consistency among the west extension commands in the
main zephyr tree, we reuse a similar class factory pattern that is present
for ZephyrBinaryRunner instances in the ZephyrBlobFetcher case. This
could be achieved with a simpler mechanism, but opted for consistency
before simplicity.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Instead of using strings, use pathlib's functionality to handle the
fact that it can have two different extensions.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
prevent `Wundef` warnings from occurring due to
missing CONFIG_ symbols and __cplusplus.
Signed-off-by: Christoph A Schnetzler <Christoph.Schnetzler@husqvarnagroup.com>
Currently, twister keeps on monitoring serial output until timeout
even if we have a flash error. And the error reason will be reported
as 'No Console Output(Timeout)', which is not accurate.
So add a flash_error flag to control if we need to monitor the output.
Signed-off-by: Guo Lixin <lixinx.guo@intel.com>
This is this initial sysbuild guide on what sysbuild is and how to build
a sample using sysbuild.
It provides an architectural overview of sysbuild.
Descriptions on how to use `west build` or `cmake` + `ninja` to build
projects with the sysbuild infrastructure.
Flashing is described through the use of `west flash`.
Extending sysbuild with additional Zephyr based applications are
described, and reference to CMake documentation for including non-Zephyr
based applications are provided.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Update to Sphinx 5.x, also update breathe since version 4.34 is the
first to officially support Sphinx 5.x.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
This reverts commit 87c6789355, restoring
commit 3b341085a2.
Restore the original change now that the underlying issue has been fixed
in 6cfb18686e.
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
Add the hearing aid sample to the footprint tracking for the
nrf52840dk_nrf52840 and nrf5340dk_nrf5340_cpuapp.
The sample should represent a simple hearing aid implementation,
and is thus a very good candidate for tracking the footprint
of LE Audio in Zephyr.
Signed-off-by: Emil Gydesen <emil.gydesen@nordicsemi.no>
The boards command was not properly using the zephyr_module
functionality to obtain the board roots of all modules. Fix that by
moving the functionality required to the core zephyr_module file and
reuse it from external scripts.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
support to use board@revision as platform filter when running
twister, like "twister -p nucleo_f030r8@1 ...".
Signed-off-by: Chen Peng1 <peng1.chen@intel.com>
This file is no longer used by device.h, so let's avoid spending time
generating the content formerly in it.
In order to leave a pointer in place for users who are expecting to
see it or are pulling it into their own builds, however, replace its
contents with an #error directive that tells them what happened. This
can be removed later on when we expect people are used to the
transition.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Add two helper macros:
- DT_FOREACH_HELPER
- DT_FOREACH_OKAY_HELPER
These are internal generated macros which will be used to define
devicetree.h APIs for iterating over every node in the tree, and every
node in the tree with status "okay", respectively.
We can use these new APIs to implement the functionality in
device_extern.h natively from device.h without requiring a generated
header. It will also be useful for other purposes later on.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Now that we've removed the label property from most devicetree nodes
the dt_chosen_label will end up returning "" in most cases. For the
small handful of cases that the function is used, return the node.name
instead as this matches what DEVICE_DT_NAME will do.
Signed-off-by: Kumar Gala <galak@kernel.org>
Replicate the devicetree dependencies into a sorted list. This ensures
that the structures added to the .__device_handles_pass2 section are
reproducible from build to build.
Tested with: west build -b native_posix tests/drivers/build_all/sensor
Without this change, two consecutive builds do not compare.
Signed-off-by: Keith Short <keithshort@google.com>
pylink 0.14.0 changed the class variable where JLink DLL library name
(libjlinkarm) is stored. This patch adds support for new pylink
libraries while keeping backwards compatibility.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
The elf_parser library now generates a dot file with device dependencies
that can be later rendered using Graphviz. Each node in the diagram
contains the device label (taken from DT node). In some cases the label
property can be None, leading to build failures like:
```
line 273, in device_dependency_graph
text = '{:s}\\nOrdinal: {:d} | Handle: {:d}\\n{:s}'.format(
TypeError: unsupported format string passed to NoneType.__format__
```
This patch switches to node name instead, which will always be set to
some value. This value is actually what devices get now as a name if
they do not have a label set.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
There is a breaking change in Junitparser 'TestCase.result' between
v1.x and v2.x.
Update check_compliance.py minimally to take into account this change.
This avoids relying on an outdated package.
Signed-off-by: Henri Xavier <datacomos@huawei.com>
This commit extends the west commands build, flash, and debug to support
--domain when having multiple domains (images) defined in a domains.yaml
build file.
The domains.yaml uses the following yaml format to specify the
build directory of each domain in the multi image build:
> default: <domain-n>
> domains:
> <domain-1>:
> build_dir: <build_dir-domain-1>
> <domain-2>:
> build_dir: <build_dir-domain-2>
> ...
`west <build|flash|debug>` has been extended to support
`--domain <domain>`.
`west build` calls CMake to create the build system, and if `--domain`
is given, then the build tool will be invoked afterwards for the
specified domain.
`west flash` will default flash all domains, but `--domain <domain>`
argument can be used to select a specific domain to flash, for example:
> west flash --domain mcuboot
`west debug` only a single domain can be debugged at any given time.
If `--domain` is not specified, then the default domain specified in the
domains.yml file will be used.
Users can still select a different domain, for example with:
> west debug --domain mcuboot
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
west build now support the sysbuild CMake project which allows users
to easily enable and build a bootloader together with any sample.
This will allow for cleaning up samples and boards which already does
custom multi image in a non-generic way, for example:
- mps2_an521 which includes a build for remote board
- samples/ipc/ which includes a samples for remote board
- esp32 which includes custom bootloaders
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
The --tool-opt runner option is the recommended practice for allowing
runners to take additional arguments that are passed on to the
underlying tool. It exists because we don't want to add one runner
option for every single tool option that users might want to tweak --
that would be a nightmare.
Enough runners are using this option that it's time to promote it to a
common runner capability with consistent behavior, the same way we did
for the --dev-id option in the past. This removes boilerplate from
individual runner files and ensures consistent argument handling for
this option when it is supported.
Since --tool-opt is a bit long to type, and we've had some complaints
about that, take this as an opportunity to standardize on -O as a
short option equivalent for it.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Even when BreakBeforeBraces is set to Linux, implying BraceWrapping for
AfterEnum to be true, clang-format version 12 puts the opening brace of
an enum declaration on a new line. [1]
One possible workaround would be to change AllowShortEnumsOnASingleLine
from false to true, but that would yield a different kind of unwanted
formatting.
clang-format version 13 and newer have this issue fixed, therefore
bumping the minimal version.
[1] https://github.com/llvm/llvm-project/issues/48983
Signed-off-by: Reto Schneider <reto.schneider@husqvarnagroup.com>
If a device manually specifies that it depends on a second DT device,
add the first device to the second devices list of supported devices.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Output the final dependency graph as a `.dot` file, which when rendered
by graphviz can be easier to comprehend than the text descriptions.
This output is optional in that it will not be generated if `graphviz`
is not installed.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Use the new elf_parser module to simplify the process of generating the
final device handle arrays.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Adds a python module intended to simplify other scripts that need to
work with device information compiled into the first pass `.elf` file.
Scripts utilizing this module can focus on iterating over dependency
graphs to generate the desired output, instead of also needing to
extract and build the graphs in the first place.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Rename the placeholder variable generated for PM slots so that the
prefix doesn't colide with the PM structs declared by devices. This
simplifies the process of searching for symbols in `.elf` files.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Escape quotes provided in extra arguments on the command line.
example: ... -x "CONFIG_COMPILER_OPT=\"-fanalyzer\""
Fixes#46382
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
When doing a split build and test with coverage enabled, (i.e.
`twister --coverage --build-only` followed by `twister --coverage
--test-only`), Twister attempts to run coverage reports immediately
after building, which fails because tests haven't actually run yet
and causes several error messages to print in to the log.
This change causes twister to skip the call to `run_coverage()` if
`--build-only` is set and print an info message. This suppresses the
error messages from the coverage report tool complaining about missing
coverage files, but still instruments the built code for coverage data
collection. When twister is invoked again, but with `--test-only`
instead, the coverage files will be emitted and the reports
successfully generated.
#### Testing
I ran...
```
$ zephyr/scripts/twister -T zephyr/tests/lib/cmsis_dsp/basicmath/
--coverage -p native_posix --build-only
$ zephyr/scripts/twister -T zephyr/tests/lib/cmsis_dsp/basicmath/
--coverage -p native_posix --test-only
```
... followed by a combined build and test ...
```
$ zephyr/scripts/twister -T zephyr/tests/lib/cmsis_dsp/basicmath/
--coverage -p native_posix
```
... as a control and diffed the HTML coverage reports, which were
identical other than a timestamp.
Signed-off-by: Tristan Honscheid <honscheid@google.com>
Fix bug in verify_platforms_existence method - make it possible to
verify all platform names from list - not only first.
Fixes: #48321
Signed-off-by: Piotr Golyzniak <piotr.golyzniak@nordicsemi.no>
This commit makes the `--ninja` argument default to `true` when neither
`-k` nor `--make` is specified (i.e. when Make generator is not
selected).
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
When `--short-build-path` argument is specified to the twister, the
following error message is displayed because the Python argparse module
does not allow specifying a mutually exclusive argument as required:
ValueError: mutually exclusive arguments must be optional
This commit removes the `required` parameter when adding the `--ninja`
argument, which is a mutually exclusive argument, and adds a manual
check to validate this condition instead.
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>
Given binding dirs the script will generate a Kconfig.dts of the
form:
DT_COMPAT_ADI_ADT7420 := adi,adt7420
config DT_HAS_ADI_ADT7420_ENABLED
depends on DTS_HAS_ADI_ADT7420
def_bool $(dt_compat_enabled,$(DT_COMPAT_ADI_ADT7420))
Than a driver Kconfig can use these Kconfig symbols as follows:
menuconfig ADT7420
bool "ADT7420 Temperature Sensor"
default y
depends on DT_HAS_ADI_ADT7420_ENABLED
...
Signed-off-by: Kumar Gala <galak@kernel.org>
Following changes has been made by this commit:
- fix issue when a binary is not able to be found by relative path
- implement "west debug" command
- implement --dt-flash option
Signed-off-by: Alex Kolosov <rikorsev@gmail.com>
This fixes commit
3d5cc38cf6.
That commit only works if removing the quotes from e.g. a string in an
array actually results in a token. If the string's value is something
like
"foo,bar"
though, it will result in the "token"
foo,bar
in the generated output.
This is wrong; fix it using the new edtlib.str_as_token() API
introduced to allow callers to perform the same procedure as that
library for fixing up strings.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
