The macros CONCAT(), MIN(), and MAX() are defined over and over again in
RIOT's code base. This de-duplicates the code by moving the macros to a
common place.
The mbox code contains a race condition in `mbox_put()`: When it
waits for a slot in the queue to become available, it is woken up with
IRQs enabled. It disables IRQs again as first thing, but by then
another thread may already have preempted the running thread and filled
the queue back up. In this case, a message in the queue would be
silently overwritten.
This restores a pre-existing design decision to implement both
blocking and non-blocking mutex locking with the same code. Those
implementations have been split prior to the introduction of
the `core_mutex_priority_inheritance` module when `mutex_trylock()`
indeed was trivial. This decision didn't age well, so undo it.
Presently, RIOT just emits a warning when a stack overflow is
encountered but still resumes execution. In my view, execution should be
aborted as the detection of a stack overflows via the heuristic provided
by the scheduler is an unrecoverable error.
I ran into this while performing automated tests of a RIOT application
where a stack overflow occurred but I only noticed this after inspecting
the application output more closely.
Similar to SSP failures, I added crash_code for stack overflows.
This fixes https://github.com/RIOT-OS/RIOT/issues/18545 as the code
previously relied on `sched_change_priority()` not directly scheduling
a new thread while IRQs are disabled, but rather later when IRQs are
restored. This is true for Cortex-M MCUs (where the PendSV IRQ is used
to trigger the scheduler), but not e.g. for AVR.
An `1 << x` with `x >= 15` is undefined behavior on 8-bit / 16-bit
machines (which typically have `sizeof(int) == 2`).
Using `1UL << x` is safe for `x <= 31`, which is large enough to make
use of the full 32 bits in `runqueue_bitcache`.
In addition, a `static_assert()` is added to enforce that
`SCHED_PRIO_LEVELS` is never set to anything larger than 32.
This is intended for the bootloader module where we don't enter thread
mode, so mutex must never attempt to switch context.
Instead use a simple busy wait that is enough to make the possible mutex
users (e.g. interrupt based SPI) in bootloader mode work.
Use `uintptr_t` for arithmetic rather than `const char *` to fix the
following warning:
> comparePointers: Subtracting pointers that point to different objects
- activate THREAD_CREATE_STACKTEST also if test_utils_print_stack_usage
is used
- make thread_measure_stack_free() available unconditionally
- if DEVELHELP is active, call test_utils_print_stack_usage() on any
thread exit
- if DEVELHELP is active, call test_utils_print_stack_usage() after main
for the idle thread, if that is used
