That comment applies to all posix arch based boards,
not just native_posix.
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
It is not just native_posix that supports the sqrt test,
but all posix arch based ones.
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
c++2B standard is not supported for xt clang compiler so
exclude cpp2B testcase for xt-clang toolchain.
Signed-off-by: Aastha Grover <aastha.grover@intel.com>
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
On some SMP platforms (such as intel_adsp_ace15_mtpm) it is not
safe to access the stack of a thread that is currently executing
on another CPU. This requires that tests be written such that
objects and data that are shared between threads that may execute
concurrently on different CPUs be placed into memory that is not
a thread's stack.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
Enable this test which runs in native_posix_64 in
native_sim(_64), and set default test platform
to native_sim.
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
Enable these tests which run in native_posix in native_sim,
Switch from native_posix to native_sim as default test platform
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
Enable this test which runs in native_posix in native_sim,
Switch from native_posix to native_sim as default test platform
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
C library testing is mainly there to support what is
necessary to support Zephyr. We do test a variety of libcs
currently, which is where YAML comes in handy.
However, the main libc testsuite can be overkill for testing
some things, and might not be suitable for testing optional
features.
Create a 'common' subdirectory for common libc tests.
Signed-off-by: Christopher Friedt <cfriedt@meta.com>
The acpi_get_irq_routing_table() takes a pointer to an array of
ACPI_PCI_ROUTING_TABLE elements rather than a generic buffer pointer
(e.g. void *).
Because of the above, it makes sense to specify the array size as an
actual ARRAY_SIZE() value, since it makes no sense to accept buffers
which are not a multiple of sizeof(ACPI_PCI_ROUTING_TABLE) long.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Add basic ACPI test application. Initially there are only tests for one
ACPI table (MCFG) and fetching the ACPI PRT, but this is already a good
start for caching regressions on the supported boards.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Add tests for new DT APIs added in the previous commit:
- `DT_IRQN_BY_IDX`
- `DT_INST_IRQN_BY_IDX`
Added additional tests for the following existing DT APIs when
`CONFIG_MULTI_LEVEL_INTERRUPTS` is enabled:
- `DT_IRQN`
- `DT_INST_IRQN`
Added `qemu_riscv32` for the multi-level interrupt tests.
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
Passing an invalid errno value to strerror is undefined behavior in
POSIX. While all Zephyr-specific C libraries may be required to return a
specific value, we can't hold an external C library to that standard.
Signed-off-by: Keith Packard <keithp@keithp.com>
The native library may use a 32-bit time_t, so we need to avoid
having the compiler generate an error during compilation.
Signed-off-by: Keith Packard <keithp@keithp.com>
Merging 565c9376f1
exposed an issue in this test, which causes an assert in the
mps3_an547.
For more information check
https://github.com/zephyrproject-rtos/zephyr/issues/64387
The issue needs further analysis.
With this platform being an integration
platform for this test, it gets triggered in CI by unrelated
PRs, causing CI failures and blocking development.
As an interim measure, to unblock development in the main branch,
let's exclude this platform from this test.
This commit should be reverted once the underlaying issue is
indentified and addressed.
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
* Select always PICOLIBC. This is anyhow the default for all
but the posix architecture. And for the posix architecture
we need it as POSIX_CLOCK does not work with the host C
library.
* Fix the testcase filtering. This sample works fine
in native_sim (when not using the host libC),
but does not work for native_posix(_64), as those can
only be built with the host libC, and in that case,
the POSIX_CLOCK provided functions will not be properly
linked to.
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
Instead of forcing use of NEWLIB_LIBC, select any available complete C
library implementation. Add CONFIG_REQUIRES_FLOAT_PRINTF where needed.
Signed-off-by: Keith Packard <keithp@keithp.com>
These tests cannot be run in this architecture as it does not
support this toolchain.
Today they are filtered by kconfig, which works but spends
time running cmake.
As native_posix is a default test platform it is better
to filter it alltogether by arch, which saves quite a lot
of time.
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
The test delegates putting the server socket into listening (i. e.
calling listen() on the socket) to a separate thread, which did have a
chance to run before client attempted to establish TCP connection.
This was not visible before, as we did not reply with RST to a
connection attempt on a closed port, so the connection was eventually
establish after SYN retransmission. But as we do reject such a
connection now with RST, the connection attempt failed. Therefore, a
small delay was added after spawning the server thread, to give it a
chance to configure the server socket.
Additionally, lower the CONFIG_NET_TCP_TIME_WAIT_DELAY value so that TCP
contexts are released earlier, and add a respective delay in the test
teardown function. Not doing so also triggered unneeded SYN
retransmissions, as there were no enough TCP context to accept the
incoming connection, before freeing the resources allocated for the
previous one.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
The malloc arena/heap size setting can be adjusted using different
Kconfig options, depending on the libc implementation. This means
prj.conf can't be used to set this value on projects that can be built
for multiple libcs without generating a Kconfig warning.
Note: similar fix was applied for the hash_map sample, see
7ef8911e8c
Signed-off-by: Gerard Marull-Paretas <gerard@teslabs.com>
rand32.h does not make much sense, since the random subsystem
provides more APIs than just getting a random 32 bits value.
Rename it to random.h and get consistently with other
subsystems.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
This was already excluded from
libraries.libc.newlib_nano.thread_safety.userspace.stress and it's now
failing in libraries.libc.newlib.thread_safety.userspace.stress as well.
The issue is being rootcaused to a toolchain issue but a proper fix has
not been identified yet, exclude that test as well in the meantime.
Link: https://github.com/zephyrproject-rtos/zephyr/issues/61129
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
Rearranges the sys_mem_blocks fields so that information that describes
how much of the memory block is used is co-located. This will allow
easier of its statistics into the object core statistics reporting
framework.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
Update the descriptions for the various CAN devicetree timing properties
specified in Time Quanta (TQ) to make it clear that these, if present, are
only used for the initial timing parameters.
Deprecate the (Re-)Synchronization Jump Width (SJW) devicetree properties
for both arbitration and data phase timing as these are now only used in
combination with the other TQ-based CAN timing properties, which are all
deprecated.
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
Changes the Kconfig symbols for the sys_heap based memory management option
and removes the notion of `BLOCKS` with a more concise single
`LV_Z_MEM_POOL_SIZE` option. Also adds `LV_MEM_CUSTOM` to the lv_conf.h,
since in any option the memory management is considered to be custom.
Signed-off-by: Fabian Blatz <fabianblatz@gmail.com>
This adds the cmsis-nn module since it moved to its own realm.
This also adds a kconfig for the long short-term memory.
Signed-off-by: Ryan McClelland <ryanmcclelland@meta.com>
Use CMSIS-DSP from its new realm. This also changes change how you
initialize FFT tables as well to use arm_cfft_init_64_f32 if you
know the FFT size in advance rather than the generic initialization
arm_cfft_init_f32.
Signed-off-by: Ryan McClelland <ryanmcclelland@meta.com>
