I've implemented a patch (attached to the bug, but in a gist as well
here https://gist.github.com/1641725 ) that implements a hash_pbkdf2()
function. The patch also slightly refactors hash_hmac() to split out
a little bit of common functionality (since pbkdf2 relies upon hmac).
I've targeted the patch towards trunk for now, but I'd like to see if
it could get into 5.3 and 5.4 as well (in a point release).
Let me know your thoughts,