Hi Elin,

Looks like this is out-of-sync, unfortunately. Could you update it to master?

Reviewed the patch - looks clean. I couldn't see any major issues with the code. Made a few optimization comments to elin offline.

Can you rebase onto master please?

I think my merging of the two-factor auth patch threw you out of sync again, the merge button is greyed out.

There was some comments somewhere about Using SHA512 vs SHA256 for fall back.

Even though SHA512 is faster then SHA256, we're not doing it often enough to make it really required.

Second, this is a fall back, not a "we need faster speed". If they are worried about speed, use bcrypt.

Third, IANACE (I am not a crypto expert), but given that SHA512 uses a block size of 64 bits. Thus, As I understand it, Anything less then 8 character passwords would require padding, this could yield to other attacks. SHA256 uses a block size of 32 bits, so at 8 character passwords, you'd have at least two runs instead of the one.

(EDIT)
Duh, we should be using a salt > 8 chars, so you would have at least 2 rounds on a 64bit block size. But, the lower block size would be better in the long run as there would be more rounds.