Currently PSA backends cannot use other backends because the necessary
definitions are defined in a single big header file. This prevents us
from creating a generic HMAC backend based on the available hash
backends, as the hash context struct is not available when defining the
HMAC context struct.
Fix this by spliting the headers into separate files. This makes it
possible us use the hash context definitions without pulling in the
remaining context definitions.
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
In Arduino IDE the on-board LED pin is defined by macro `LED_BUILTIN`. It is used whenever the LED is controlled. To make it easier to use Arduino sketches as they are, `LED_BUILTIN` is defined by using @ref ARDUINO_LED as defined by the board.
When using `OUTPUT=COLORTEST` the return value of the unit test's
`main()` on failure was `0`, while it should be `1`. As a result,
running the unit tests from a script using
$ OUTPUT=COLORTEXT make BOARD=native64 RIOT_TERMINAL=native -j32 flash term -C tests/unittests
would not catch a test failure.
This commit changes the behavior so that `0` is returned on success
and `1` on failure for both regular and colored text output.
This allows using and compiling the URL handling functions of sock_util
without a network stack used.
Co-authored-by: benpicco <benpicco@googlemail.com>
This changes the documentation of how to configure stdio, especially
with regard on how to configure the stdio frontends with
`printf_float`, `printf_long_long`, and `stdin`.
Co-authored-by: crasbe <crasbe@gmail.com>
This adapts to an API change for providing stdin/stdout/stderr: The
macro to test whether globals are to be used is now prefixed with two
underscores.
Co-authored-by: crasbe <crasbe@gmail.com>
This makes sure that newlib's printf is correctly overridden by:
1. Adding wrappers for some more obscure printf() variants
2. Forcing a compilation failure when newlib's stdio is still linked in
while mpaland-printf is used
3. Not adding `-u _printf_float` to the linker flags when mpaland-printf
is used.
So far, the atomic utils only support known width integer types. This
adds `unsigned int` to the supported list by mapping it to the
corresponding fixed width integer type.
This also sneaks in a few minor style fixes that clang-tidy frowned
upon.
Co-authored-by: crasbe <crasbe@gmail.com>
