The waitAfterCommandFinishesExecution failPoint is activated via a call to executeIf. If the predicate provided to executeIf, the action provided is activated and the activation count is incremented. In this specific case, CurOpFailpointHelpers::waitWhileFailPointEnabled is called.

Within CurOpFailpointHelpers::waitWhileFailPointEnabled, a second executeIf is called on the failPoint, incrementing the activation count again. Within that call, shouldFail() is called, incrementing the activation count yet again.

This is responsible for BF-38502, in which the failPoint is expected to be entered exactly one time per thread, but because a single hit of the failPoint can register multiple activations, there is a race condition where a single thread can register enough fail point activations to unblock the test before enough requests are queued leading to the failure seen in BF-38502.