In the previous fix for a passed max duration, the assumption was made
that at maximum one second will pass between the duration assignment and
the std::condition_variable::sleep_for() call.
This implementation makes the behavior more predictable by using
sleep_until() instead to emulate the sleep_for() behavior.
The standard library has the implicit requirement that for
std::condition_variable::sleep_for() the duration must not cause an
overflow if added to the current time.
This commit will reduce the duration accordingly to fit into the
duration type.
This keeps the function consistent with sleep_until() and sleep_for()
which both can be interrupted.
This is relevant to allow an update via a "signal" without an "interval"
in a custom module.
The changes in GCC 11.x made `std::condition_variable` implementation
internals `noexcept`. `noexcept` is known to interact particularly bad
with `pthread_cancel`, i.e. `__cxxabiv1::__force_unwind` passing through
the `noexcept` call stack frame causes a `std::terminate` call and
immediate termination of the program
Digging through the GCC ML archives[1] lead me to the idea of patching
this with a few pthread_setcancelstate's. As bad as the solution is, it
seems to be the best we can do within C++17 limits and without major
rework.
[1]: https://gcc.gnu.org/legacy-ml/gcc/2017-08/msg00156.html
Tamino Bauknecht
Evan Overman
Patrick Nicolas
Alex
Aleksei Bavshin
hoellen