I think after move 31 Black is threatening Bc2 e.g. 32...Bc2 33.Bxc2 Nxc2 and d4 falls. Hence 32.Nxf5.

Both players start the endgame with potential weaknesses, but Smyslov defends his more efficiently.

In retrospect, 29.Ng5 might be worth considering e.g. 29...Ke7 30.Nxh7 Nc2 but maybe this just decentralises White's knight.

30...Ke7 defends the b6 pawn because if 31.Nc8+ Kd7 32.Nxb6 Kc7 the knight has no good retreat.

34.d5 would also be an interesting try. Then White can play e6 and if Black ever plays Nxd5 Bxd5 Kxd5 the e pawn queens. Black, however, has 34...Nd3 35.e6+ Kd6 36.Ke2 Nxb2 37. Kd2 c4 38.Kc3 (Bc2 may be better when Nd3 leads to a similar ending but with White's king more central) cxb3 39.Kxb2. Black can only create distant passed pawns here by playing b5 at some stage, when White will get unstoppable passed pawns as well and it's a race e.g. 39...g5 40.Kxb3 h5 41.Kc4 f4 42.Kd4 g4 43.hg hg 44.Ke4 f3 44.gf g3 45.Ke3 b5 44.ab a4 45.b6 a3 46.e6+ Kxe6 47.b7 a2 48 b8=Q a1=Q 49 Qxg3 looks as though it might be good for White. Black may be able to draw by sitting tight, though, as it's difficult for the White king to approach the pawns on f5 and g5 and if the king goes to b5 the Black king's side pawns will break through.

As the game goes, the e pawn is obviously weak but Black has to choose the right circumstances to cash the advantage in. 41...Nd4 42.Kc4 Nxe6 43.Bxe6 Kxe6 produces exactly the same pawn endgame as the game, but I guess White can do better with 43.Kb5 as the bishop should be worth more than the knight here.