On Tue, Apr 5, 2022 at 7:10 AM Jason A. Donenfeld <Jason@xxxxxxxxx> wrote:
>
> Practically speaking, this means that at least on x86, /dev/urandom
> becomes safe. Probably other architectures with working cycle counters
> will also become safe. And architectures with slow or broken cycle
> counters at least won't be affected at all by this change.
I think this is a good change, as it's a bit pointless to warn about
uninitialized random data if we can just initialize it.
I do wonder if it wouldn't be better to perhaps move this all into
wait_for_random_bytes(), though, and add an argument to that function
for "no delay".
Because I think we should at the same time also add a warning to
wait_for_random_bytes() for the "uhhhuh, it timed out".
Right now wait_for_random_bytes() returns an error that most people
then just ignore. Including drivers/net/wireguard/cookie.c.
So instead of returning an error that nobody can do much about, how
about we move the warning code into wait_for_random_bytes()?
And make that urandom_read() call the same wait_for_random_bytes()
that random_read() calls, just with GRND_NONBLOCK as an argument?
Not a big deal. Your patch is fine by me too.
Linus
