If spi_sam_configure() is called with more than 4 chip selects the
current check fails even if the pins are used as gpios. The variable
spi_csr_idx already has the correct index to be checked and must be
used in this check.
Signed-off-by: Munawar Ali <ali@sofiha.de>
Signed-off-by: Axel Utech <utech@sofiha.de>
If the event timer or free-run timer is close to expiration when
system enters idle with I2C target DMA mode enabled, the memory
and CPU clocks may become unsynchronized after wakeup. This
causes CPU to fetch incorrect data and eventually trigger SoC
watchdog timeout.
Due to this hardware limitation, SoC should skip entering idle
mode if the remaining timer value is less than 150µs(safe margin).
Signed-off-by: Ren Chen <Ren.Chen@ite.com.tw>
Returning `-ENOTSUP` (and logging a `LOG_ERR`) is not the correct
behaviour if there is no enable GPIO. The fact there was no work to do
does not make the result an error.
From the docs for `pm_device_runtime_get`
```
* @retval 0 If it succeeds. In case device runtime PM is not enabled or
* not available this function will be a no-op and will also return 0.
```
Signed-off-by: Jordan Yates <jordan@embeint.com>
Minimal changes to OV760 drive to use generic ov767x structures.
- This includes changes to function names.
- Includes changes to defines as necessary.
- Deleted camera_id as no longer required
Signed-off-by: Michael Smorto <CyberMerln@gmail.com>
This commit introduces a new comparison for the ARM (AArch32) architecture
in the GICv3 interrupt controller.
It specifically compares the lower 24 bits of the affinity values,
because there is no aff3 in AArch32 or Arm-v7.
Signed-off-by: Guowei Li <15035660024@163.com>
Allows "egis_et171_spi" to use basic clock information
provided by the clock driver to calculate the runtime
SPI frequency.
Signed-off-by: Jacky Lee <jacky.lee@egistec.com>
The Cadence SPI controller has a hardware issue (Xilinx AR#65885
contains more details) where the RX_NOT_EMPTY (RNE) status bit
reports inconsistent state. This causes the driver to unreliably
detect when RX FIFO contains valid data, leading to transfer
failures and test hangs.
Signed-off-by: Srikanth Boyapally <srikanth.boyapally@amd.com>
Microchip MEC parts have a similar GPIO peripheral block. We
create a unified driver for all parts. NOTE: MEC GPIO interrupt
detection sets active status when changed from interrupt detect
disabled to any enabled mode. Driver ISR and interrupt configuration
implementation includes work-arounds for this issue.
Signed-off-by: Scott Worley <scott.worley@microchip.com>
Setting a prescaler of 256 would overflow the uint8_t.
Also get cycles calculation was wrong.
Signed-off-by: Peter van der Perk <peter.vanderperk@nxp.com>
add driver and dts-binding for kinetis_scg driver
used on ke1xz
Signed-off-by: Hake Huang <hake.huang@nxp.com>
Signed-off-by: Michael Galda <michael.galda@nxp.com>
In install timeout api, the cfg parameter does not make sense to cop, we
can ignore it rather than return an error code to keep forward
compatibility.
Signed-off-by: Yves Wang <zhengjia.wang@nxp.com>
The mcux_adc16 driver's read function fails to validate the user-provided
buffer size when `adc_sequence.options` are used (extra samplings).
The calculation erroneously considered only the size required for one
channel, neglecting the total number of channels in the sequence.
This leads to a **buffer overflow** when reading multiple channels.
Fix this by using `POPCOUNT(sequence->channels)` to determine the
correct channel count and ensure the buffer size is sufficient.
Signed-off-by: Badr Bacem KAABIA <badrbacemkaabia@gmail.com>
1. Fix wrong backlight pin in driver overlay
2. Remove the power-on pin configuration in code and binding, and
add mipi display panel regulator in panel overlay instead. Set
regulator-boot-on' to true means the power-on pin will be enabled
uring system boot.
3. Remove 'last_known_framebuffer' from panel data structure. It is
not used anywhere
4. Fix bug in 'co5300_set_pixel_format' function.
5. Fix the issue that the panel does not support start coordinates
and the width/height of the updated area being odd value.
Solution: In panel driver, maintain a full screen-sized buffer,
its address and pitch alignment is configurable in device tree
and shall be compliant with the display controller's requirements.
It can be placed in RAM or if the RAM space is not enough it can
also be placed in other memory resion. When there is a frame
update request, the updated area will be first filled to the
buffer, if the area's size or coordinate is odd, adjust the value
so the real updated area covers the requested updated area, then
use this buffer to send pixel to panel, this can ensure the
updated area's size and coordinate are always even.
Signed-off-by: Kate Wang <yumeng.wang@nxp.com>
The irq_lock area including net_pkt_rx_alloc_with_buffer() was causing
error "Context switching while holding lock!".
Actually we didn't have any hardware access to protect in netc_eth_rx,
so just removed irq_lock.
Fixes#99694
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
For stm32h7 the order matters to get the pwm to output the values as
expected. As reported for the the H730 doesn't produce PWM output when
the LL_TIM_OC_SetDeadTime is called after LL_TIM_EnableAllOutputs. So,
switch the order of the calls and this doesn't impact the stm32f413zh
boards.
Signed-off-by: Charles Hardin <ckhardin@gmail.com>
Extend the number of SPI data sizes supported for series that allow it.
The driver uses the new property st,stm32-data-width to create a table
with all compatible data widths and to check if the SPI instance
support the required width (some instances may have a limited set
of widths compared to others on the same SoC).
Signed-off-by: Guillaume Gautier <guillaume.gautier-ext@st.com>
Inverts ll_func_spi_dma_busy logic so that the function returns true if
the SPI DMA is busy (as suggested by the name) instead of the other way
around.
Also completes the check for H7-compatible with the LL_SPI_IsActiveFlag_EOT
function when transfer size is set.
Signed-off-by: Guillaume Gautier <guillaume.gautier-ext@st.com>
The xmc47_relax_kit i2c driver is incorrectly using the
I2C_SPEED_STANDARD macro to set the clock frequency. The correct macro
to use is XMC4_I2C_SPEED_STANDARD which is defined by the XMC4 driver.
Signed-off-by: Yassine Safraoui <yassine.safraoui@grenoble-inp.org>
Guard callback registration with CONFIG_SPI_ASYNC to avoid
build error for blocking SPI transfers using DMA
Signed-off-by: Sai Santhosh Malae <Santhosh.Malae@silabs.com>
Swap the write and read data paths within the GSPI controller to ensure
correct data is seen on the SPI lines during 16-bit transfers.
Signed-off-by: Sai Santhosh Malae <Santhosh.Malae@silabs.com>
Updated mosi_overrun alignment from 4 bytes to 32 bytes to
ensure compatibility with the GPDMA driver. This prevents
alignment-related errors during DMA transfers.
Signed-off-by: Sai Santhosh Malae <Santhosh.Malae@silabs.com>
Enable higher SPI burst sizes when using GPDMA to improve throughput.
Bursts are only enabled when both TX and RX channels use GPDMA and
the SPI clock frequency is at least 10 MHz, as DMA flow control is
unreliable at lower SPI rates.
Signed-off-by: Sai Santhosh Malae <Santhosh.Malae@silabs.com>
Add gspi_siwx91x_pick_lower_freq() to select the nearest lower
supported GSPI clock frequency based on the requested value.
The GSPI clock can operate only at discrete frequencies such as
80 MHz, 40 MHz, 26.6 MHz, 20 MHz, and so on (80 MHz divided by
integer factors). If the requested frequency does not match one
of these valid steps, it is rounded down to the nearest lower
supported frequency.
The driver now logs both the requested and the actual programmed
frequency along with the divider value, helping users verify the
effective SPI clock configuration.
Signed-off-by: Sai Santhosh Malae <Santhosh.Malae@silabs.com>
The GSPI and QSPI peripherals run on the interface PLL clock.
To ensure correct operation, the interface PLL frequency should
be set to 160 MHz. This provides a base clock of 80 MHz to the
QSPI peripheral, which matches its required operating frequency.
The GSPI peripheral can continue to operate at higher frequencies
as it receives the full interface PLL clock.
Signed-off-by: Sai Santhosh Malae <Santhosh.Malae@silabs.com>
The existing device runtime PM model requires applications to call
pm_device_runtime_get()/put() on the GPIO controller device directly,
which is not ideal when GPIO ports are exposed as child nodes.
This update enables runtime PM at the GPIO port (child node) level,
allowing applications to manage power through the port devices instead
of the top-level controller.
Signed-off-by: Sai Santhosh Malae <Santhosh.Malae@silabs.com>
Add `get_calibration` and `set_calibration` subcommands to allow users
to read and modify the RTC calibration settings from the shell.
Signed-off-by: Tim Pambor <tim.pambor@codewrights.de>
The previous delay could cause missed URCs because the pipe and chat
layer were not yet attached when the modem started sending its first
messages (+KSUP). The modem boot takes ~1.8s, while the driver only
needs ~120ms to set up the interfaces, so reducing the delay ensures
the receiver is ready before the modem outputs any URCs.
Increase hl78xx_init_chat_script command limit from 10 to 100 to allow
longer initialization sequences.
Signed-off-by: Zafer SEN <zafersn93@gmail.com>
This update refactors the HL78xx initialization and runtime state
machine to provide robust handling of unexpected modem restarts
and improve PDP and boot status management.
Key changes:
- Add KSUP unsolicited response to the global chat match table
so modem boot status is monitored independently from the
initialization script.
- Introduce a boot-status structure tracking previous boot state
to detect unexpected restarts.
- Dispatch a new MODEM_HL78XX_EVENT_MDM_RESTART event when the
modem reports a READY status after having been previously
booted.
- Add restart-handling transitions across relevant states
(await-power-on, RAT-config, await-registered, carrier-on) to
ensure the driver cleanly re-enters the initialization
sequence.
- Rework init and post-restart script entry points to run
asynchronously via hl78xx_run_init_script_async() and
hl78xx_run_post_restart_script_async().
- Move bus-open and chat-attach operations out of the
init-script state and into the await-power-on state for
clearer separation of responsibilities.
- Expose a getter for the AT-ready match structure.
These changes improve resilience during modem restarts, clarify
the state machine flow, and prepare the driver for multi-PDP-
context configurations.
Signed-off-by: Zafer SEN <zafersn93@gmail.com>
Fix several issues in GSM mode for the HL78xx modem driver to
ensure correct registration reporting, RAT handling, and PDP
context activation.
Changes include:
- Skip band configuration when RAT=GSM to prevent +CME ERROR
responses.
- Explicit PDP activation (AT+CGACT=1,1) for GSM data sessions.
- Parse +CGACT URCs and track PDP activation state internally.
- Introduce gprs status struct and update initialization
scripts to handle GSM reliably.
- Add helper functions and chat scripts to manage GSM
registration and PDP flow.
- Correct GSM registration-status handling: enable CREG=3 and
disable unused CEREG.
- Dynamically switch between CREG (GSM) and CEREG (LTE/NB-IoT)
based on active RAT.
These changes improve driver stability and ensure reliable data
and registration behavior on GSM networks.
Signed-off-by: Zafer SEN <zafersn93@gmail.com>
This patch applies multiple small cleanups across the HL78xx modem
driver, improving readability, consistency, and diagnostics without
changing logic.
Key changes:
- Added missing blank lines in Kconfig sections for readability.
- Improved several log messages for clarity and guidance to users.
- Added a debug log when RAT/Band configuration requires a modem restart.
- Replaced an incorrect SET_RAT_GMS_CMD macro name with SET_RAT_GSM_CMD.
- Added TODO marker for unhandled script failure case.
- Corrected log format string in event handler to include event ID.
- Added an informational log on modem initialization.
- Added clarifying comments in header structures and APN lookup path.
- Normalized #endif comment style throughout the driver.
- Fixed use of CME_ERROR_STRING in chat match definitions.
- Applied formatting hints (clang-format off/on) around multi-line
MODEM_CHAT_MATCHES blocks.
- Improved enum documentation in public API header.
These changes improve maintainability and developer usability without
modifying driver behavior.
Signed-off-by: Zafer SEN <zafersn93@gmail.com>
Preprocessor evaluates CYC_PER_TICK as zero in condition
`(COUNTER_MAX / CYC_PER_TICK) == 1` when
CONFIG_TIMER_READS_ITS_FREQUENCY_AT_RUNTIME is defined.
This commit skips the check when
CONFIG_TIMER_READS_ITS_FREQUENCY_AT_RUNTIME is defined.
Fixes: #100040
Signed-off-by: Ioannis Damigos <ioannis.damigos.uj@renesas.com>
moves the fixed link functionality of the
generic ethernet phy into its own driver.
This makes both drivers more readable and
efficient.
Also makes it easier to use the normal generic
phy as a base for a vendor specific driver, as
the fixed link functionality is not needed on
them.
Signed-off-by: Fin Maaß <f.maass@vogl-electronic.com>
sockaddr_un.sun_path is limited to 108 in Linux
(and to similarly small numbers in other systems).
Let's check that it fits before copying it, and error out otherwise
(the user won't be able to connect to that socket if the path
is truncated, and if we overflow we will just corrupt memory).
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
