The E/B button selects between end of release or begin of cycle for the last trigger (ER) output for each of the ADSRs. I think this was implemented to make sure that this trigger will fire whenever (suddenly mid patch) triggers or gates at the inputs of the ADSR are coming in too fast for the envelopes to reach the end of the cycle, and something else in your patch depends on the ER triggers to fire.
It's the loop button that actually causes the envelope to retrigger when the end is reached, thus creating a looping envelope. The one shot together with the loop button makes it an ADR envelope (to work with trigger inputs instead of gates). The legato or reset button only makes sense when the envelopes are looping as well.