Although very unlikely for this value to ever be negative in practice,
this would make Coverity happy.
Fixes#67969
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Refactors modem chat module to use TRANSMIT_IDLE event instead
of polling when transmitting, cleaning up the transmit handler
by combining duplicate code and using an internal state instead
of multiple iterators to track send state.
Signed-off-by: Bjarki Arge Andreasen <bjarki@arge-andreasen.me>
Remove receive and transmit timeouts which are no
longer useful as the RECEIVE_READY and
TRANSMIT_IDLE events will be used to efficiently
manage timeouts between transmit/receive calls.
Then update the the in-tree drivers using the
modem_chat module to omit the process timeout
parameter.
Signed-off-by: Bjarki Arge Andreasen <bjarki@arge-andreasen.me>
Avoid unexpected memory access in cases where the IPSR has an odd
register index. In the scenario where an odd register index is present
in IPSR, the LSB of the register index is utilized as MSB of the bank
number. Observe how we pack reg/bank:
'IPSR(((reg) << 4U) | (bank), shift, func)' (macro IPnSR)
and how it is read from the device tree source:
'(RCAR_IPSR(node_id) >> 10U) & 0x1FU' (macro RCAR_PIN_FUNC).
Finally, this bank is used to obtain the required IPSR offset:
'PFC_RCAR_IPSR + rcar_func->bank * sizeof(uint32_t)'
in the 'pfc_rcar_set_ipsr' function. For example, if we have the 1 as
a reg index and the 3-rd bank, the resulting offset would be 19,
which is beyond the IPSR range.
Align the IPSR comment with the definition of the 'rcar_pin_func'
structure.
Note: we can omit the usage of the register index entirely since this
information is obtained from the pin number inside the
'pfc_rcar_get_reg_index' function.
Signed-off-by: Mykola Kvach <mykola_kvach@epam.com>
This allows to link code and data blocks, e.g. the vector table, into
tightly coupled memory using `zephyr_linker_sources`.
Signed-off-by: Greter Raffael <rgreter@baumer.com>
Use `DT_CAT*` macros to paste simple things together instead
of `UTIL_CAT` wherever possible. This should reduce the amount
of error messages from the compiler when something goes wrong.
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
This commit adds source and header files required for bmp581 I2C driver.
I have used bmp581_user.h to add more usage related definitions
but bmp581.h to add hardware related definitions.
Signed-off-by: Talha Can Havadar <havadartalha@gmail.com>
Signed-off-by: Gerhard Jörges <joerges@metratec.com>
In usb_hid.h `struct usb_setup_packet` is used. This is defined in
usb_ch9.h but not included in usb_hid.h which leads to build errors when
you include the usb header files in the wrong order. This commit fixes
this behaviour.
Signed-off-by: Gerhard Jörges <joerges@metratec.com>
Add a pair of functions to translate from input events to hid codes,
mapping most of the current hid codes defined in zephyr/usb/class/hid.h.
Use a sparse table for the mapping, which takes advantage of the fact
that code 0 is reserved. Inspired by the linux equivalent hid to input
map:
https://elixir.bootlin.com/linux/latest/source/drivers/hid/hid-input.c#L27
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
1. add function sensing_iodev_submit and defined struct
sensing_submit_config
2. fix the issue of phy_3d_sensor
Signed-off-by: Zhang Lixu <lixu.zhang@intel.com>
Initializers must be compile time constants (not expressions!) for clang
to be happy. Clang rightfully pointed out that the callback_list member
of sensing_connection was being initialized using an expression that was
not compile-time constant.
The macros passed along a pointer created by taking the address of a
global constant and then at initialization time attempting to
dereference it using the * operator. So in the end the compiler has
identifier its trying to first take the address of and then later
dereference in an initializer. Clang didn't appreciate this, though gcc
seemed to be just fine.
Actually clang 17 might work just fine with this as well, but clang 16
(the clang I tried) didn't like it at all.
Instead store the pointer to the callback list rather than copying the
struct. This could be done the other way and not passing a
pointer through the macros perhaps.
Signed-off-by: Tom Burdick <thomas.burdick@intel.com>
Some doc comments were making doxygen unhappy and the doc build fail.
The description of the sensor connection was also updated slightly to
simplify the description a bit.
Signed-off-by: Tom Burdick <thomas.burdick@intel.com>
Many sensors have multiple functions, for example, icm42688 supports
accel, gyro and temperature, and the sensor streaming api always mixes
the multiple functions in one function call. So we need add a layer in
sensing subsystem to dispatch the result returned from sensor streaming
api for each function.
I changed the sensor-type(int) to sensor-types(array) in sensing sensor
device bindings, so that one device can map to multiple instances of
sensing sensor.
Signed-off-by: Zhang Lixu <lixu.zhang@intel.com>
1. use the sensor_driver_api instead of the sensing_sensor_api.
2. clean up the sensing_sensor.h after using the unified api.
3. select RTIO while CONFIG_SENSING is enabled.
4. remove the sensing_sensor_post_data and sensing_sensor_xxx_data_ready by
introducing the RTIO
Signed-off-by: Zhang Lixu <lixu.zhang@intel.com>
This set of changes brings in raw packet tx injection mode
feature to ethernet and removes it from being a Wi-Fi only feature.
It was earlier envisaged as Wi-Fi net management feature only.
Signed-off-by: Vivekananda Uppunda <vivekananda.uppunda@nordicsemi.no>
Now that we have a very easy way to get the interrupt
controller and the interrupt level of a node, we can
simplify the multilevel interrupt encoding macros.
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
It looks like all SoCs in tree check if an exception comes from an IRQ
the same way, so let's provide a common logic by default, still
customizable if the SoC selects RISCV_SOC_ISR_CHECK.
Signed-off-by: Gerard Marull-Paretas <gerard@teslabs.com>
Add callbacks to the stream objects that reflects the
state of the isochronous channel. The connected callback
is called when the isochronous channel is connected, and
similarly the disconnected callback is called when it is
disconnected.
There is a special case for unicast, where if the ACL
disconnects first, then we won't get a ISO disconnect
callback. It should be assumed that the isochronous channel
is no longer valid when the BAP stream enters the idle state,
i.e. when the "released" callback is called.
The purpose of the new callbacks is to provide additional
information to the application. Especially the unicast client
can use this to determine when the stream_start function
can be called again, as there can only ever be 1 outstanding
CIS connection request at a time, but there can be multiple
GATT requests.
Signed-off-by: Emil Gydesen <emil.gydesen@nordicsemi.no>
The iterable section should be located in ROM and not RAM, this caused
crashes on multiple platforms running socket services.
Fixes#67762
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Nothing critical, but it just looks better if the service name is
separated from the prefix, i.e:
_z_net_socket_service_idx_udp_service_async
vs current:
_z_net_socket_service_idxudp_service_async
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
The sigevent struct and sigval union members were previously not
ordered in a way that produces optimal alignment / reduces
padding on 64-bit systems.
Reorder members so that pointers come first.
Signed-off-by: Christopher Friedt <cfriedt@meta.com>
The previous behavior of the CONFIG_SHELL_AUTOSTART option, where setting
it to 'n' disabled shell interaction at startup but kept the logger
active as a shell backend, was inconsistent.
Now, when CONFIG_SHELL_AUTOSTART is set to 'n', both the shell and the
logger are inactive at startup. Calling the shell_start function will
activate them and print any pending logs. This change ensures a more
consistent and logical behavior regarding shell and logger activity at
startup.
Additionally, now when the shell_stop function is called, both the shell
and the logger are halted. This enhancement ensures that stopping
the shell also effectively stops the associated logger.
Signed-off-by: Jakub Rzeszutko <jakub.rzeszutko@verkada.com>
A growing number of CAN controllers do not have support for individual RX
hardware filters based on the Remote Transmission Request (RTR) bit. This
leads to various work-arounds on the driver level mixing hardware and
software filtering.
As the use of RTR frames is discouraged by CAN in Automation (CiA) - and
not even supported by newer standards, e.g. CAN FD - this often leads to
unnecessary overhead, added complexity, and worst-case to non-portable
behavior between various CAN controller drivers.
Instead, move to a simpler approach where the ability to accept/reject RTR
frames is globally configured via Kconfig. By default, all incoming RTR
frames are rejected at the driver level, a setting which can be supported
in hardware by most in-tree CAN controllers drivers.
Legacy applications or protocol implementations, where RTR reception is
required, can now select CONFIG_CAN_ACCEPT_RTR to accept incoming RTR
frames matching added CAN filters. These applications or protocols will
need to distinguish between RTR and data frames in their respective CAN RX
frame handling routines.
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
Added the GATT authorization callback API that allows the user to
define application-specific authorization logic for GATT operations.
Signed-off-by: Kamil Piszczek <Kamil.Piszczek@nordicsemi.no>
Bluetooth Mesh Protocol 1.1 is backward compatible with Bluetooth Mesh
Profile 1.0.1, therefore the stack can still be qualified for 1.0.1 if
needed. But explicit support for both versions requires additional
maintenance efforts and doubles the CI time. To make the stack
qualifiable for 1.0.1, the one needs to remove Private Beacons reception
and compile out SHA256 algorithm support.
What is changed:
- Removed `CONFIG_BT_MESH_V1d1` option.
- Removed `transport_legacy.c` which was using 1.0.1 implementation. The
new transport layer has new SaR logic that should still be possible to
qualify for 1.0.1.
- Removed the legacy transport Kconfig options. They are superseded by
the new transport Kconfig options.
- Tester app: `overlay-mesh-v1d1.conf` is merged into
`overlay-mesh.conf`.
- Removed BabbleSim tests for 1.0.1.
- Updated documentation.
Signed-off-by: Pavel Vasilyev <pavel.vasilyev@nordicsemi.no>
When the -Werror=missing-field-initializers is enabled, the compiler
complains about missing initialization of dev pointer in the
init_entry struct when using the SYS_INIT[_NAMED] macro.
This commit adds explicit assignment of NULL to it.
Signed-off-by: Michał Barnaś <barnas@google.com>
Remove the CAN_FILTER_FDF flag for filtering on classic CAN/CAN FD frames
as it is not supported natively by any known CAN controller.
Applications can still filter on classic CAN/CAN FD frames in their receive
callback functions as needed.
Fixes: #64554
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
L2 PPP is not able to handle the carrier being lost gracefully,
nor is it able to gracefully close the PPP connection when
net_if_down() is called.
This patch refactors the L2 PPP module to use the carrier state
to either properly close or simply terminate the PPP connection.
Additionally, it ensures that the PPP session is closed properly
before calling ppp->stop().
Signed-off-by: Bjarki Arge Andreasen <bjarki@arge-andreasen.me>
When using the auto discover CCC, and the function is called
more than once with the same parameters before the previous
discovery has completed, then that may cause issues when we
reset the parameters when we get the response.
This commit adds a small check for the callback function
which is only set to the specified function when the
discovery is in progress. This way we can return an error
if the application calls bt_gatt_subscribe multiple times
before the previous discovery has completed, rather
than asserting.
Signed-off-by: Emil Gydesen <emil.gydesen@nordicsemi.no>
Since a85ffa8, `struct sd_card.card_buffer` has naturally fallen at an
offset that is unaligned. This meant that on systems without support for
unaligned access (e.g: Cortex-M0), a hard fault would present when
executing code that casts the buffer to `uint32_t` (such as
`sdmmc_spi_read_cxd()`, `card_query_written()`, etc...)
Historically, it appears that the alignment of the `card_buffer` member
was good and operational only by chance.
Altering the default value of `CONFIG_SDHC_BUFFER_ALIGNMENT` was
rejected, as this has wider implications.
Fixes#62619
Signed-off-by: Attie Grande <attie.grande@argentum-systems.co.uk>
