Problem

The Failpoint API does not provide a non-blocking way to access the underlying _hitCount, so users must do extra work to poll a failpoint while doing another operation simultaneously.

For example, the following code must use two threads to check a failpoint was entered while calling requestFlush():

Atomic<bool> stop{false};
FailPointEnableBlock fp("failDuringFlush");
stdx::thread driver([&] {
    while (!stop.load()) {
        requestFlush();
    }
});
fp->waitForTimesEntered(fp.initialTimesEntered() + 1);
stop.store(true);
driver.join();

Request

Ideally, this code is much simpler:

FailPointEnableBlock fp("failDuringFlush");
while (fp->timesEntered() < fp.initialTimesEntered() + 1) {
        requestFlush();
}

My proposal is to add a Failpoint::timesEntered() member function that just returns _hitCount, but I am open to other solutions that address the underlying problem.

There does exist a waitForTimesEntered(Interruptible* interruptible, EntryCountT targetTimesEntered) that can be interrupted, but this requires additional boilerplate in the form of an OperationContext deadline + a try-catch block.