As-is I think this is scary. I'd be happier about it if it shifted the deterministic behaviour to a separate function rather than having the rand_deterministic global switching all users of rand() in the process into insecure mode. But that will be a bunch more work than the previous sweep changing srand to srand_deterministic.
- Further unification of random number generators Theo de Raadt
- Re: Further unification of random number generators Ted Unangst
- Re: Further unification of random number generators Stuart Henderson
- Re: Further unification of random number generators Bryan Steele