[PATCH v2] arc4random: simplify design for better safety

[Adhemerval Zanella Netto]

On 26/07/22 08:20, Jason A. Donenfeld wrote:
> Hey Florian,
> 
> On Tue, Jul 26, 2022 at 01:12:28PM +0200, Florian Weimer wrote:
>>>> What happens if /dev/random is actually /dev/urandom?  Will the poll
>>>> call fail?
>>>
>>> Yes. I'm unsure if you're asking this because it'd be a nice
>>> simplification to only have to open one fd, or because you're worried
>>> about confusion. I don't think the confusion problem is one we should
>>> take too seriously, but if you're concerned, we can always fstat and
>>> check the maj/min. Seems a bit much, though.
>>
>> Turning /dev/random into /dev/urandom (e.g. with a symbolic link) used
>> to be the only way to get some applications working because they tried
>> to read from /dev/random at a higher rate than the system was estimating
>> entropy coming in.  We may have to do something differently here if the
>> failing poll causes too much breakage.
> 
> The "backup plan" would be to sleep-loop-read /proc/sys/kernel/random/entropy_avail
> until it passes a certain threshold one time. This might also work on even older
> kernels than the poll() trick. But that's pretty darn ugly, so it's not
> obvious to me where the cut-off in frustration is, when we throw our
> hands up and decide the ugliness is worth it compared to whatever
> problems we happen to be facing at the time with the poll() technique.
> But at least there is an alternative, should we need it.

I think the poll trick is way better, although I also think it is very Linux
specific.  Should we move it to Linux sysdeps?

The /proc/sys/kernel/random/entropy_avail would require to open another file
descriptor, which I think we avoid for arc4random if possible.