z_pm_save_idle_exit() is only declared if CONFIG_PM is enabled but
unconditionally defined. Just move it around in the header to always
be declared.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Create stubs for when the subsystem is not compiled. Some drivers need
to use constraints when are in the middle of a transaction to avoid
the system to sleep. As the same driver may be used in a target that
does not support power management, they have to conditionally (#if
CONFIG_PM) call these functions, these stubs avoid this regular problem.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
The priority of the LPTim interrupt on each stm32 soc
is put at the same level as the SysTick. This priority
level is given by the system Handler priority reg
(SHPR3) when the PM is not configured.
Signed-off-by: Francois Ramu <francois.ramu@st.com>
Fixes: #37637
toolchain_parse_make_rule() parses depfiles and converted `\\` to the
CMake list separator `;`.
However, gcc -M might create depfiles with windows path separator `\`
causing this conversion to fail, as a path like:
c:\...\zephyr\samples\drivers\led_ws2812\nrf52-bindings.h
resulting in a file list as:
c:;...;zephyr;samples;drivers;led_ws2812;nrf52-bindings.h
which results in a CMake configure dependency to be added to `C:`.
As C: is always newer than the build.ninja file, this resulted in
continues CMake re-invocation.
As a small side-note, the `\` in file name did only occur in situations
where a relative past had been used elsewhere in the build system, such
as here:
https://github.com/zephyrproject-rtos/zephyr/blob/\
c3050a5aab1e86d02642eb4c6a027419e6a3a096/samples/drivers/led_ws2812/\
boards/nrf52dk_nrf52832.overlay#L9
To ensure proper handling of files, then all files are converted to
CMake paths, that is with forward slashes: `/`
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
This commit removes the `z_` prefix from the stdio syscall functions
(`z_zephyr_write_stdout` and `z_zephyr_read_stdin`) since it is
redundant and does not align with the convention used by the equivalent
minimal libc syscall functions (e.g. `zephyr_fputc` and
`zephyr_fwrite`).
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
The newlib `write()` and `read()` functions must call the
`z_zephyr_write_stdout()` and `z_zephyr_read_stdin()` syscall functions
in order to function properly in a user mode context.
The existing incorrect implementation was copied off the newlib hooks
implementation, which was corrected in the previous commit.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
The commit 4344e27c26 changed the syscall
function invocation in the `write()` and `read()` functions to the
direct syscall implementation function invocation by mistake.
The newlib `write()` and `read()` functions must call the
`z_zephyr_write_stdout()` and `z_zephyr_read_stdin()` syscall functions
in order to function properly in a user mode context.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This patch is testing the test_sleep_abs with a longer
real time slot value. The reason is that for platforms
like stm32wb55rg with PM, the real time slot must be adjusted
because of the LPTIM ticker.
Signed-off-by: Francois Ramu <francois.ramu@st.com>
In the case of native_posix (and other "host" toolchain based
platforms) Twister shouldn't filter tests by ZEPHYR_TOOLCHAIN_VARIANT.
The PR replaces `- zephyr` with `- host` in native_posix(_64).yaml
and modify the filter in twisterlib.py to not block building
if "host" is under the platform supported toolchains.
Fixes: #38418
Signed-off-by: Maciej Perkowski <Maciej.Perkowski@nordicsemi.no>
Add CONFIG_XTENSA_SMALL_VECTOR_TABLE_ENTRY
for NXP_ADSP_IMX8 board to bypass the size
constraint of the interrupt vector table entry.
Fixes#38216
Signed-off-by: Iuliana Prodan <iuliana.prodan@nxp.com>
For some platforms, like NXP's IMX8 or Mediatek's MT8195,
the size of an interrupt vector table entry is 0x1C bytes,
less than usual (0x30 for Intel's platforms).
So, the interrupt handlers don't fit in the vector table
entries.
I've added a small indirection to bypass this size
constraint and moved the default handlers to the end
of vector table, renaming them to
_Level\LVL\()VectorHelper.
For this, I've added a generic configuration -
XTENSA_SMALL_VECTOR_TABLE_ENTRY.
Signed-off-by: Iuliana Prodan <iuliana.prodan@nxp.com>
When CONFIG_USERSPACE is turned off, the POOL_SECTION will be located in
.data section. This will increase the target binary size. Since the
memory pool is for malloc() use and it doesn't need for initial values,
locate it in the .bss section to reduce binary size.
Signed-off-by: Dylan Hung <dylan_hung@aspeedtech.com>
Change-Id: Iee52ac06a48414c083518c79775fe31334eab674
Since it's not possible to encode full range of 64-bit unsigned integer,
remove this type from the LwM2M implementation, and replace its uses
with 64-bit signed integer.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
As the resource values represented by the unsigned integers were casted
to integers of a corresponding size in the read handler, they were not
ecoded properly if the unsigned value was larger than the maximum
integer value of the corresponding size (i.e. they were encoded as
negative values).
Fix this by casting the unsinged value to a wider integer type, to
prevent incorrect interpratetion of the data provided. The TLV encoding
functions take care of the optimization (i. e. encoding integers on the
minimum number of bytes needed), so it should prevent bandwith waste if
the unsigned value would actually fit into the integer of the
corresponding size.
Similar case is for the write hander, where unsigned integers encoded at
8 bytes were not processed correctly. Fix this by using wider decoder as
well.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Listing a neighbour table with "net nbr" command, when a neighbour w/o
assigned link address was present, resulted in an assert condition. Add
additional check to prevent this.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
If the BOARD_RUNNER_ARGS_${runner_id} variable contains anything, add
it to the global property of the same name with all the final runner
args from board_runner_args() and friends.
This allows use cases like this:
west build ... -- -DBOARD_RUNNER_ARGS_jlink="--some-jlink-arg"
The ${runner_id} part is the runner's name converted to a C
identifier, i.e. with special characters changed to underscores.
So e.g. the 'dfu-util' runner would use BOARD_RUNNER_ARGS_dfu_util.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
The test assumed initial equality between CLOCK_MONOTONIC and
CLOCK_REALTIME and also assumed coarse granularity for each clock.
Neither of those assumptions are solid.
Furthermore, the test failed on multiple vendor boards which caused
some concern.
Remove the poorly conditioned tests and remove some comments
/ printks.
Fixes#35703
Signed-off-by: Christopher Friedt <chrisfriedt@gmail.com>
TCP state machine gets stuck in TCP_FIN_WAIT_2 state
when server responds with [ FIN, PSH, ACK ]
Fixes#37842
Signed-off-by: Nicolas Marty <nicolas.marty@zuehlke.com>
Using zsock_ in http_client instead of the POSIX API versions of the
functions allows the usage of http_client in combination with
CONFIG_POSIX_API.
Signed-off-by: Benedikt Schmidt <benedikt.schmidt@embedded-solutions.at>
When enable thread awareness feature for OpenOCD the search path was
converted to a list. In some environments OPENOCD_DEFAULT_PATH may
not be defined. That create an empty search path list system fails.
This add a test to skips fill search_args with openocd_search values
when list is empty.
Fixes#38272.
Signed-off-by: Gerson Fernando Budke <nandojve@gmail.com>
When using the stack fork type in the philosophers sample,
there were a few pointer type mismatch since the k_stack_*()
functions are using stack_data_t but the parameters are uint32_t.
So cast them to void pointer to suppress compiler warnings.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Xtensa does not store the stack pointers in thread objects but
pushing the registers into the stack. There is no fixed location
to retrieve the stack pointer so mark it as unimplemented to
avoid the #warning.
Fixes#38405
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
When the application sets more than 251 bytes of advertising data,
the data is sent over multiple HCI commands.
This is only allowed if the advertiser is not running
as defined by the HCI specification.
The data is sent to the controller one AD-field at the time.
If an AD-field is larger than BT_HCI_LE_EXT_ADV_FRAG_MAX_LEN,
the data is split over two commands.
This introduces some additional complexity.
Signed-off-by: Rubin Gerritsen <rubin.gerritsen@nordicsemi.no>
add a bitarray into struct osThreadDef_t to indicate whether the
thread is used or not, then we can get the first available thread
by searching this array when creating a new thread, and update this
array to add a free thread when terminating a thread.
Signed-off-by: Chen Peng1 <peng1.chen@intel.com>
To test fallback to remaining bearers PTS might request IUT to
disconnect one of the connected EATT channels, while the others remain
intact. Test function must be added, because we cannot create L2CAP
server on EATT PSM and manage this server as normal and have EATT
enabled at same time.
This is affecting GATT/SR/GAW/BV-14-C
Signed-off-by: Krzysztof Kopyściński <krzysztof.kopyscinski@codecoup.pl>
BTP change extended L2CAP_CONNECT command by ECFC flag, which determines
which connection procedure is used (non-enhanced or ecred). Now, only
this flag determines the procedure used, not number of requested
channels.
This was affecting L2CAP/ECFC/BV-25-C
Signed-off-by: Krzysztof Kopyściński <krzysztof.kopyscinski@codecoup.pl>
The overview at the top of the page only listed ble and serial as
available transports. This commit adds udp to this list.
Signed-off-by: Christian Taedcke <christian.taedcke@lemonbeat.com>
This commit disables the "minimum required newlib heap size" check for
the newlib thread safety tests since they do not make use of any newlib
functions that require large heap.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
If the count is zero in the heap_malloc_free test, a div by zero
would happen, so add a condition to handle this potential error.
Signed-off-by: Chen Peng1 <peng1.chen@intel.com>
Add some tracing functions to fix that the samples/tracing code
cannot generate coverage by twister.
Fixed#38323
Signed-off-by: Lixin Guo <lixinx.guo@intel.com>
stm32f411 includes stm32f401 which do already correctly define spi4
The definition in stm32f411 was missing a clock attribute
Signed-off-by: Kim Bøndergaard <kim.boendergaard@escoglobal.com>
stm32f410, stm32f411 and stm32f412 enable the SPI5 device with
bit 20 on the APB2 clock
Signed-off-by: Kim Bøndergaard <kim.boendergaard@escoglobal.com>
In commit "pm: Fix weak linkage symbols" (PR #35274),
PM SoC hooks were converted to __weak to avoid clash with
new definition of these symbols in subsys/pm/power.c.
G0 power implementation was implemented in parallel
with this change and missed the update.
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
Cadence XCC is based off of a very old 4.2 gcc compiler, which didn't
perfectly support C99 "inline" semantics with respect to
cross-translation-unit inline linkage (which Zephyr does not use, our
inlines are static only) and declaration order.
Fix the one spot where we were calling an inline before its
ALWAYS_INLINE definition, and add a flag to suppress the warning so
CI's trying to build with XCC and -Werror don't flip out.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Older binutils, like the (nine-year-old!) 2.23 version that powers
many Cadence XCC toolchains, happen not to support the "~" operator to
perform bitwise negation. And they generate an absolutely hilarious
series of inscrutable error messages when they try to tell you this
fairly simple fact.
Just fold it into the constant.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Due to a historical error introduced during an hci_uart refactor, the
get_len() function was using the wrong HCI H4 packet identifier for
commands. Instead of using H4_CMD (0x01) it was trying to match
BT_BUF_CMD (0x00).
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
This function wasn't being defined when SMP_BOOT_DELAY was set or when
SMP wasn't enabled. There's no reason for either, then function
doesn't depend on any kconfig-dependent build-time state, and (given
that we use -ffunction-sections) it won't appear in output binaries
unless called.
And there are use cases (e.g. z_smp_start_cpu()) where we need that
function even when BOOT_DELAY is enabled.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Th Configuration Server should respond with and Invalid AppKey Index
status code when the AppKey identified by AppKeyIndex is not known to
the node or is not bound to the model identified by the ModelIdentifier.
Signed-off-by: Michał Narajowski <michal.narajowski@codecoup.pl>
When the MSI clock is selected as source on the stm32wbx device,
the MSI has a range to choose the MSI input frequency.
Signed-off-by: Francois Ramu <francois.ramu@st.com>
