On Mon, Feb 08, 2016 at 05:35:31PM +0000, Nick Lowe wrote: > Because it is a relatively insignificant factor for the quantity of > random bits that are being consumed in insensitive areas, where > os_random suffices, compared to everything else. > > /dev/urandom isn't so CPU heavy that it is tangible for the use cases > and it stops a future changes using what can be a dangerous function. > > I agree there are places where os_random could remain, but it just > isn't worth it for no tangible benefit. I'd prefer to do this type of changes one-by-one should there be any stronger justification for some of the changes to the point of someone potentially wanting to merge in such a change to a release branch. In other words, about nine patches to convert the existing users and a separate patch to remove the function in the end. Each of the patches should have a commit message that explains why the particular change is done (e.g., identify that this can be somewhat stronger design like the EAP-pwd server case or state that it is just for the sake of getting rid of os_random() callers where no such justification can be described). -- Jouni Malinen PGP id EFC895FA _______________________________________________ Hostap mailing list Hostap@xxxxxxxxxxxxxxxxxxx http://lists.infradead.org/mailman/listinfo/hostap
