The standard is 9 years old now, it is well supported in all mayor compilers.
In fact, features of the 'new' standard are already used in RIOT (std_atomic).
Let's make it the default and adapt the Makefiles accordingly.
clang/LLVM is a bit more finicky with what target triples it accepts compared
to GCC. However, the GCC target triple is needed to properly locate the C
libraries. In case a locally installed GCC toolchain uses a technically not
fully correct target triple, it might be impossible to build with LLVM.
The solution here is to allow specifying a TARGET_ARCH_LLVM target triple
(which defaults to TARGET_ARCH), which is passed to clang. For locating the
C lib, still TARGET_ARCH is used.
FEATURES_CONFLICTING previously was declared prior to the function it is
calling, resulting in empty output during the first dependency resolution
iteration. This fixes the order so that the conflicting features are detected
right from the first recursion.
Previously, FEATURES_REQUIRED_ANY didn't honor the order of the alternatives
provided, if none of the features were already in used and multiple options
are provided. This fixes this.
RISC-V support semihosting in very similar way as the cortex-m
microcontrollers. The code calls a breakpoint instruction and the
attached debugger reads/writes registers and memory for stdio.
The RISC-V architecture doesn't support a call number with the EBREAK
instruction, to allow the debugger to detect a semihosting break point,
the EBREAK instruction is wrapped in a SLLI and SRAI instruction. These
use x0 as output register, making them NOP instructions.
One caveat when using this is that the RISC-V core traps the EBREAK
instruction with trap code 3 when no debugger is attached. Restarting
the application with the debugger attached avoids this.
These targets cannot be used in an automated testing workflow without
complex configuration or extend rights.
- Add new 'test-as-root' target for tests that require to be root or
start an external daemon as root
- Add new 'test-with-config' target for tests that require a specific
configuration to succeed (module configuration or hardware
configuration)
This commit makes overflow of signed integers to behave as expected by at 90%
of the C developers, even though overflow of signed integers are strictly
undefined behavior.
Note: Please do not add code relying on a specific behavior for the overflow of
signed integers, even though `-fwrpav` will make that code work. This is
intended to mitigate the risk of bugs in overflow checks being exploited,
not to encourage adding new bugs.
See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=30475 for details and see
http://c-faq.com/misc/intovf.html on how to implement overflow checks properly.
The driver uses the netdev interface. Due to the limited
capabilities of the transceiver (32 byte FIFO and no source address in the layer2 frame),
it relies on 6LowPAN compression and adds the source address to the frame for that.
Split out Gunar Schorcht's clever approach to provide thread safe malloc for
AVR into a system module and make AVR depend on this. This allows other
platforms to also use this.
In Engineering mode (BOOT0 off and BOOT2 on), only the Cortex-M4
core is running. It means that all clocks have to be setup
by the Cortex-M4 core.
In other modes, the clocks are setup by the Cortex-A7 and then should
not be setup by Cortex-M4.
stm32mp1_eng_mode pseudomodule have to be used in Engineering mode
to ensure clocks configuration with IS_USED(MODULE_STM32MP1_ENG_MODE)
macro.
This macro can also be used in periph_conf.h to define clock source
for each peripheral.
Signed-off-by: Gilles DOFFE <gilles.doffe@savoirfairelinux.com>
In case of muticore CPU, openocd opens one debug port by core for gdb.
Thus add a GDB_PORT_CORE_OFFSET port offset to select the right port
for debugging.
Signed-off-by: Gilles DOFFE <gilles.doffe@savoirfairelinux.com>
For multi-arch SoC like STM32MP1, the right target core has
to be selected to avoid debugging the wrong default cpu.
This is done using openocd command 'targets ${OPENOCD_CORE}'.
OPENOCD_CORE has to be set in board Makefile.include file.
In case it is not set, the command just display available targets, thus it
has no effect on already existing boards using openocd.
Signed-off-by: Gilles DOFFE <gilles.doffe@savoirfairelinux.com>
Expose the auto-negotiation feature of the Ethernet device via the
pseudo-module stm32_eth_auto. With this enabled, the static speed configuration
set in the boards periph_conf.h will only be used if the PHY lacks
auto-negotiation capabilities - which is unlikely to ever happen.
In analogy to the existing GPIO mappings, this provides (write-only)
SAUL entries for PWM'd LEDs in a single-LED (as SAUL_ACT_DIMMER) and an
RGB (as SAUL_ACT_RGB_LED) mode.
Co-authored-by: Marian Buschsieweke <marian.buschsieweke@ovgu.de>
This feature is only used to blacklist stdio via CDC ACM.
Introduce the `highlevel_stdio` feature instead to indicate
that stdio functionality requires a running thread to print
anything
This adds a check in the build process to verify that no ERROR_ Kconfig
symbols are set. These symbols indicate invalid conditions in the
Kconfig configuration.
This requires -nostartfiles to be only passed to the linker, not the
compiler, as it is a linker flag and passing it to the compiler causes a
clang warning to be emitted.
Additionally, clang does not seem to support `-mcmodel=medlow` and
`-msmall-data-limit=8` but these options do not seem strictly necessary
to me anyhow thus they are deactivated conditionally when using clang.
Busybox grep does not support long options. In fact, the utilized long
options are not mandated by POSIX.1-2008. Using the short options allows
building RIOT on Alpine Linux which utilizes Busybox instead of GNU
coreutils by default.
This commit introduces a common storage backend for SUIT manifest
payloads. Different backends can be compiled into a single firmware.
Degending on the component name in the SUIT manifest, a storage backend
is selected by the parser.
- Set XTIMER_HZ to something that is actually possible to generate with one
of the available clock dividers from the core frequency
- Use xtimer_on_ztimer if xtimer is used and not ztimer_xtimer_compat is used
- This is needed because xtimer is simply not compatible with any of the
possible clock frequencies of this board
Kconfig.dep depends on FORCE, so it is always generated when compiling
with Kconfig under normal conditions. Whan TEST_KCONFIG=1 is set, this
file is no longer a dependency for out.config. So when cleaning the
'bin' directory, out.config has no direct dependencies that force its
rebuilding (the generated directory is order only). This causes the file
not to be produced when calling `TEST_KCONFIG=1 make clean all`.
This PR changes the dependency on the 'generated' directory to a direct
dependency when `CLEAN` is set and leaves it as order-only when not.
This allows to generate out.config only when needed by not depending on
FORCE.
Use riscv-none-elf instead of legacy riscv-none-embed as target triplet for
RISC-V development. However, if ricsv-none-elf is not present, try
riscv64-unknown-elf and riscv-none-embed instead. If the legacy riscv-none-embed
is used, a warning is printed.
Add `TARGET_ARCH_<ARCH>` for each architecture (e.g. `TARGET_ARCH_CORTEX` for
Cortex M) to allow users to overwrite the target triple for a specific arch
from ~/.profile or ~/.bashrc (or the like) without overwriting it for all others
as well.
It has no real purpose other than pulling in `gnrc_ipv6_router` as a
dependency, which is already done in other places (or by pulling in its
dependencies).
CDC ACM, ethos, Semihosting and SLIP all benefit from sending data out
in larger chunks and will benefit from stdout buffering.
`stdio_rtt` does have an internal TX buffer, so we don't need to do any
buffering on top.
With plain UART there was a slight advantage *without* buffering when
testing with `tests/periph_uart_nonblocking` (with the non-blocking feature
disabled).
Since the removal of the buffering saves us some RAM and ROM, disable it by
default there.
This will be different with DMA enabled UARTs.
This changes the prefixes of the symbols generated from USEMODULE and
USEPKG variables. The changes are as follow:
KCONFIG_MODULE_ => KCONFIG_USEMODULE_
KCONFIG_PKG_ => KCONFIG_USEPKG_
MODULE_ => USEMODULE_
PKG_ => USEPKG_
This makes RIOT use the integer-only printf/scanf code by default and
includes a new make parameter to select the full floating point
version. This saves about 6kB of text space when building hello-world
for the microbit board.
Signed-off-by: Keith Packard <keithp@keithp.com>
----
v2:
Use USEMODULE=printf_float instead of separate parameter
Support for picolibc as alternative libc implementation is added with
this commit. For now only cortex-m CPU's are supported.
Enable via PICOLIBC=1
---
v2:
squash fixes in
v3:
Remove picolibc integer printf/scanf stuff from sys/Makefile.include,
it gets set in makefiles/libc/picolibc.mk
fixup for dependency
It is desireable to have a way to identify network devices.
This should be independent from the type of netdev, so a common identifier is needed.
Base this on the driver ID and the index in the configuration struct.
This way we achive unique IDs that stay consistent for any firmware flashed on a board.
