Re: [Qemu-devel] [PATCH 2/2] VirtIO RNG

From: Jamie Lokier <jamie@shareable.org>
To: Gerd Hoffmann <kraxel@redhat.com>
Cc: Ian Molton <ian.molton@collabora.co.uk>,
	Paul Brook <paul@codesourcery.com>,
	qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH 2/2] VirtIO RNG
Date: Wed, 21 Apr 2010 10:40:07 +0100	[thread overview]
Message-ID: <20100421094007.GC13114@shareable.org> (raw)
In-Reply-To: <4BCEAC99.8000206@redhat.com>

Gerd Hoffmann wrote:
> On 04/20/10 23:31, Ian Molton wrote:
> 
> >Using virtio-rng means that the data is going into the guest
> >kernels hwrng subsystem.
> 
> Which is *the* major advantage of the virtio-rng driver.  In case the 
> guest kernel is recent enougth to have support for it, it will 
> JustWork[tm].  No need for guest configuration, no need for some 
> userspace tool.  I'd like to see this driver being merged.
> 
> With any kind of serial port (be it a emulated 16550 or virtio-serial) 
> you'll need some daemon running inside the guest grabbing entropy data 
> from the port and feeding it back into the kernel.

That's a bunch of false assumptions.

There's no reason a hwrng connector to virtio-serial could not be
automatic in a similar way to the console.

But enough of that: It's history now; the guest virtio-rng has existed
for more than a year.  It is also amazingly short and simple.  Yay for Rusty!

I don't object to virtio-rng; I think it's fine in principle and would
be happy to see the existing guests which have a virtio-rng driver
make use of it.

A bit of overlapping functionality is rife in emulators anyway :-)

-- Jamie