The problem seemed to be a pipelining problem of write and read instructions when swapping the context. An isync instruction when returning from a context switch solves the potential pipelining problem.
Reason for the problem was that tast_exit function in thread_arch.c tried to release the thread a second time although it was already released in sched_task_exit. A simple check whether the thread is already released (sched_active_thread == NULL) solved the problem.
Xtensa newlib version requires pthread_setcancelstate as symbol. Therefore, the module pthread was always used, which in turn requires the module xtimer. The xtimer module, however, uses TIMER_DEV(0). Therefore, tests/timers failed for TIMER_DEV(0).
The `periph_flash_common` feature was only defined here to trigger
inclusion of a source file with common functions.
It even only defines private symbols `_lock` and `_unlock` so no reason
to expose it to the build system.
And in practice, all stm cpus providing `periph_flashpage` or
`periph_eeprom` were required to provide `periph_flash_common` to allow
including it.
The previous implementation was only parsing in the modules were in
`FEATURES_REQUIRED` wich did not take cases of `FEATURES_OPTIONAL` into
account.
And also, in the same time, as the dependencies was declared in
`Makefile.include` it was processed before `Makefile.dep` so never handled
cases where a module could depend on `periph_flashpage` or
`periph_eeprom` feature.
It is replaced by selecting the common source file when module using it
are included.
The now useless feature `periph_flash_common` is removed from
`FEATURES_PROVIDED`.
`cpu/stm32_common/Makefile.dep` was never included by the global
`Makefile.dep`, so declaring this `USEMODULE +=` here was never shown to
the build system and never exported as `MODULE_PERIPH_I2C_X` variables.
In fact, `USEMODULE` was only used to trigger the `periph.mk`/`SUBMODULES`
mechanism to add matching source files names to `SRC` and so select
the implementation for each CPU type.
It is replaced by just explicitly selecting the right source file.
The commit
- improves the timing of the SDA and SCL signals that fixes communication problems with some slaves (#10115),
- introduces the internal function _i2c_clear which clears the bus when SDA line is locked at LOW, and
- renames internal _i2c_*_sda and _i2c_*_scl functions to function names that are more clear, e.g., _i2c_clear_sda to _i2c_sda_low.
