RFC: Thread abort handler in unified kernel
Any comment on this?
Zephyr microkernel tasks allow user to set the task abort handler by
invoking task_abort_handler_set(). The abort handler is executed by
microkernel server fiber when it aborts the task.
As moving to unified kernel, it may be a good time to decide if this
functionality is needed in the future (for now it is needed for
compatibility, of course).
Dmitriy Korovkin wrote:
Zephyr microkernel tasks allow user to set the task abort handler byWhat's the use case? We don't really have a working exception/longjmp
framework that could clean up things after a stack unwind, so what
value is an "abort handler" going to provide anyway? A quick grep
says that we don't even have a test case for this (it gets linked in
explicitly by the footprint test, but never called).
I agree we should junk it on general principle.