Add an API to search for the frequency supported by a timer that
is closest to the given target frequency.
This is in fact non-trivial to get right, as pre-scaler registers can
be 16 bit or even 32 bit in size, making a naive loop over all possible
pre-scalers too expensive (computationally).
Co-authored-by: mguetschow <mikolai.guetschow@tu-dresden.de>
Both SPI and I2C peripheral drivers *MUST NOT* be initialized by the
application code / (non-peripheral) device driver, as this is done
automatically be default. Some peripheral drivers have a non-idempotent
initialization and initializing them twice will break things.
Sadly, the doc states the exact opposite.
This updates the documentation to match the implementation. In addition
the special case is pointed out of disabling the automatic
initialization during boot, in which case the app indeed has to do the
initialization.
During the data phase of a FDCAN transmission only one node is
transmitting, all others are receivers. The length of the bus line has
no impact.
When transmitting via pin FDCAN_TX the protocol controller receives the
transmitted data from its local CAN transceiver via pin FDCAN_RX. The
received data is delayed by the CAN transceiver loop delay.
If this delay is greater than TSEG1 (time segment before sample point),
a bit error is detected. Without transceiver delay compensation, the bit
rate in the data phase of a FDCAN frame is limited by the transceiver's
loop delay.
Since this parameter is related to the transceiver used, there cannot be
a default value, and it must be explicitly defined with the
configuration variable CONFIG_FDCAN_DEVICE_TRANSCEIVER_LOOP_DELAY.
Signed-off-by: Gilles DOFFE <gilles.doffe@rtone.fr>
Add CAN FD specifities to CAN system library in RIOT:
* 64 bytes payload
* Bit rate switching
* Error State Indicator
Signed-off-by: Gilles DOFFE <gilles.doffe@rtone.fr>
Whole CAN code in RIOT is using 'struct can_frame' to represent a CAN
frame.
However incoming CAN FD support will bring 'struct canfd_frame' to
represent CAN FD frames.
Even if the 'struct canfd_frame' has additional flags and a bigger
payload, it is aligned on 'struct can_frame' and thus they can be
referenced by the same pointers in the code.
As it is impossible to predict which one will be used in RIOT, just
define a new type 'can_frame_t' which will map to the right struct
according to the MCU CAN supported format.
Signed-off-by: Gilles DOFFE <gilles.doffe@rtone.fr>
The UART API has not spelled out what happens when `uart_init()` is
called twice. This adds precise language that states that
acquire/release semantic is to be expected from the caller. Hence,
a caller needs to call `uart_poweroff()` before reconfiguring the UART
with a second call `uart_init()` for the same UART interface.
In practise, few apps will ever reconfigure the symbol rate. So the
impact is rather low.
However: This API now allows drivers to implement sharing of a serial
peripheral that can provide multiple interfaces (e.g. UART, SPI, I2C,
etc.). It would require some cooperation from the code that does use
the UART to actually release the UART again after each transaction;
something that will only work when RX data is only expected at
known points in time (e.g. in response to a request, or never in case
of TX only stdio). But still, this can mean the difference between
a use case becoming feasible on an MCU with a low number of serial
peripherals or not.
This changes the API of xfa from
XFA(array_name, prio) type element_name = INITIALIZER;
to
XFA(type, array_name, prio) element_name = INITIALIZER;
this allows forcing natural alignment of the type, fixing failing tests
on `native64`.
Since https://github.com/RIOT-OS/RIOT/pull/20935 gpio_write()
uses a `bool` instead of an `int`. This does the same treatment for
`gpio_read()`.
This does indeed add an instruction to `gpio_read()` implementations.
However, users caring about an instruction more are better served with
`gpio_ll_read()` anyway. And `gpio_read() == 1` is often seen in
newcomer's code, which would now work as expected.
It turns out that the feature to switch the GPIO direction quickly
is not only a way to emulate open drain / open source mode for less
sophisticated GPIO peripherals that do not natively support it.
It also enables tri-state output (push-pull high, push-pull low,
high impedance), which is useful e.g. for driven charlieplexed LEDs
quickly.
This changes the API by introducing a `gpio_ll_prepare_switch_dir()`
function that prepares the value used to identify which pins should
be switched to input or to output mode. This is useful for GPIO
peripherals in which the GPIO mode register does not allocate one bit
per pin (so that only the direction is given there), such as the one
for STM32. This allows an STM32 implementation in which preparing the
bitmask needed to modify the direction of pins is not trivial.
The assumption that every MCU has this feature turned out wrong. Hence,
add a feature to allow testing for support of edge triggered IRQs on
both flanks.
The API was based on the assumption that GPIO ports are mapped in memory
sanely, so that a `GPIO_PORT(num)` macro would work allow for constant
folding when `num` is known and still be efficient when it is not.
Some MCUs, however, will need a look up tables to efficiently translate
GPIO port numbers to the port's base address. This will prevent the use
of such a `GPIO_PORT(num)` macro in constant initializers.
As a result, we rather provide `GPIO_PORT_0`, `GPIO_PORT_1`, etc. macros
for each GPIO port present (regardless of MCU naming scheme), as well as
`GPIO_PORT_A`, `GPIO_PORT_B`, etc. macros if (and only if) the MCU port
naming scheme uses letters rather than numbers.
These can be defined as macros to the peripheral base address even when
those are randomly mapped into the address space. In addition, a C
function `gpio_port()` replaces the role of the `GPIO_PORT()` and
`gpio_port_num()` the `GPIO_PORT_NUM()` macro. Those functions will
still be implemented as efficient as possible and will allow constant
folding where it was formerly possible. Hence, there is no downside for
MCUs with sane peripheral memory mapping, but it is highly beneficial
for the crazy ones.
There are also two benefits for the non-crazy MCUs:
1. We can now test for valid port numbers with `#ifdef GPIO_PORT_<NUM>`
- This directly benefits the test in `tests/periph/gpio_ll`, which
can now provide a valid GPIO port for each and every board
- Writing to invalid memory mapped I/O addresses was treated as
triggering undefined behavior by the compiler and used as a
optimization opportunity
2. We can now detect at compile time if the naming scheme of the MCU
uses letters or numbers, and produce more user friendly output.
- This is directly applied in the test app
We occasionally have some public `foo.h` header that includes a private
`foo_arch.h` header. Users are expected to include the `foo.h` header
and not the `foo_arch.h`. However, clangd will claim that the `#include`
of `foo.h` is unused if only functions / macros/ types / ... from
`foor_arch.h` is used and nothing from `foo.h`.
This adds the `IWYU pragma: export` comment to the include of
`foo_arch.h` in `foo.h`, so that clangd treats functions / macros /
types provided by `foo_arch.h` as if they were instead provided by
`foo.h`, which fixes the false positives.
If not included before, IS_USED macro from modules.h is undefined,
leading to such errors:
error: missing binary operator before token "("
Signed-off-by: Gilles DOFFE <g.doffe@gmail.com>
