you're right... I forgot that there are 3 knights... you're both right. i'm sorry.
possible or impossible?
It's like two weeks ago that I've been posting anything but I made another one, there are still problems open, because I do want an analyses/game. Here's my new one:

Black's last move seems to be Kg3-h2, and white's last move seems to be Bxf2+.
Black's b pawn promoted on a1 by capturing the only white piece available for capture - a rook, and h pawn promoted on h1.
In order for white to have two dark-square bishop promotions, his e pawn must have promoted on d8/f8, and h pawn on h8 (more efficient than d8/f8) for a total of 3 captures. Black's e pawn in this scenario seems to be useless to me. That leaves black with 1 extra piece (dark-square bishop, rook or queen) to be captured later on f2.
Other than that I cannot see any difficulties at the moment - alas I do not have time to make a proof game. Perhaps someone else can, or find a flaw in my analysis.
You're right Shoopi! maybe someone else can post a proof game on this position (or my other positions).
Here's one I made today, I kind of like the answer, but I don't think it's hard, after you've thought a little...

Why this is "the right answer"?
Black might as well have gobbled a white piece at e2 with his last move.

Here is mine: can this pawn formation occur when castling queenside. And have it ever happened in one serious game before?

knights on either e2, b3, e7, b6 for black. But seriously i highly doubt such position have arived during tournament playing. Online blitz and bullet it's a complete different story...
Still while i think about it, it can't have been many cases of enemy knights been allowed to infiltrate the camp and just take the bishop and retreat the same way out. Also the King can not have been touched earlier.
White has a knight for his h pawn. The h pawn couldn't have promoted with the other pawns in that position if there was a capture on the c file.