Re: [PATCH v4 2/6] random: Async and sync API for accessing kernel_pool

On Sun, May 3, 2015 at 11:33 AM, Stephan Mueller [off-list ref] wrote:

The kernel_pool is intended to be the in-kernel equivalent to the
blocking_pool, i.e. requests for random data may be blocked if
insufficient entropy is present.

I cannot see any reason this would be useful, let alone necessary.

Of course /dev/random should block and it seems to me there is a good
argument for making both /dev/urandom and get_random_bytes() block
until there is emough entropy to seed them well. For everything else,
though, a properly seeded PRNG seems adequate so there is no reason to
block.