Re: BPF RNG

[Daniel Borkmann]

On 04/16/2019 03:55 PM, Kees Cook wrote:
> [correcting Alexei's email address and resending...]
> 
> On Tue, Apr 16, 2019 at 8:54 AM Kees Cook <keescook@xxxxxxxxxx> wrote:
>>
>> Hi,
>>
>> In looking at prandom_u32() users, I noticed that BPF uses its own
>> state variable with bpf_user_rnd_u32(). It appears that this state is

Correct, we split that off from the main one.

>> never reseeded like regular prandom_u32(). (See __prandom_timer().) Is
>> this intentional, or should reseeding be happening?

The prng is not to be used in any security-critical context. That said,
we optionally could perform reseeding from time to time, though that
should probably be done generically, so that all "external"
prandom_u32_state() users could register themselves upon init whether
they opt-in for periodic reseeding or not.

Thanks,
Daniel