Sammalvuo gives 9...b5 (treatening 10...b4 with 11...Nxe4 with threat to queen) and references the following two games:
(a) 10 O-O-O Qc7 11 Be2 Rc8 12 Kb1 Be7 13 Rhe1 O-O 14 a3 Rfd8 15 g4 Nc5! 16 Nd5 Bxd5 17 exd5 Nxb3 18 cxb3 Ne4 ("complicated but approx. level") Ye Jiangchuan-Zhang Zhong, HeiBei Z 2001
(b) 10 f5 Bc4 11 O-O-O Rb8! 12 h3 Bxf1 13 Rhxf1 b4 14 Nd5 Nxe4 15 Qd3 Nef6 16 Qxa6 Nxd5 17 Rxd6 Be7 18 Rfd1 O-O ("unclear") Byrne-Tarjan, USA Ch (Berkeley) 1984.

There's some other comments/analysis, but I'll leave those off the public board.

OK, don't know the latest theory about this, but this is my gut feeling:

a) 9. f4 exf4?! does not look good with the Knight already at d7: 10. Bxf4 Ne5 11. Nd4 +=,

b) 9. f4 Qc7 10. f5 Bc4 11. Bxc4 Qxc4 looks adequate to my eyes (11. O-O-O may be better?!?) due to the threat on the e4-pawn. Of course you are a tempo down in comparison to similar 6. Be2 lines (where Bf1 has been already moved once).

c) 9. f4 Rc8 10. f5 Bc4 11. Bxc4 Rxc4 looks to me better than 10. ... BxNb3 11. axBb3 d5 12. exd5 (this looks alien to me for both sides  ),

d) 9. f4 Ng4 10. f5 Bxb3 11. axb3 Nxe3 12. Qxe3 Nf6 13. Bc4 you end up badly placed for the fight over the d5-hole. 

All in all the Bc4 plan (probably supported by Qc7, i.e. option (a) above) looks best to me. And I wouldn't go for Be6xNb3 as it looks like a poor exchange for Black in this position (unless it is supported by concrete lines).