The principle of the current blackhole detection is to reduce the MSS in two steps and try each candidate two times. However, if for example the two candidates are identical (which is the case for TCP/IPv6), the candidate was tested four times. This patch ensures that each possible candidate is tested two times. This reduces the time window of misclassifying a temporary outage as an MTU issue. While there, also make sure the correct statistic counter is incremented.
t_blackhole_enter is currently only set to 2 and one could argue that this doesn't need to be stored. In a followup commit t_blackhole_enter will be computed and possibly will have different values, too.