https://research.swtch.com/openssl provides more context: openssl was asked about the change, and seemingly approved it (whether everyone understood what was being approved is a different question). It's not clear why openssl never adopted the patch (was everyone else just lucky?), but I wonder what the reaction would have been if the patch had been applied (or the lines hidden away by a build switch).
In one of two cases I believe? I wonder what would have happened if both calls ended up being wrapped, if the bug would have taken even longer than it did to be found...