Re: [Qemu-devel] [PATCH 3/4] sockets: add AF_VSOCK support

[Michael Roth]

Quoting Stefan Hajnoczi (2016-10-12 10:06:10)
> On Fri, Oct 07, 2016 at 11:42:35AM -0500, Michael Roth wrote:
> > Quoting Stefan Hajnoczi (2016-10-06 11:40:17)
> > > Add the AF_VSOCK address family so that qemu-ga will be able to use
> > > virtio-vsock.
> > > 
> > > The AF_VSOCK address family uses <cid, port> address tuples.  The cid is
> > > the unique identifier comparable to an IP address.  AF_VSOCK does not
> > > use name resolution so it's seasy to convert between struct sockaddr_vm
> > > and strings.
> > > 
> > > This patch defines a VsockSocketAddress instead of trying to piggy-back
> > > on InetSocketAddress.  This is cleaner in the long run since it avoids
> > > lots of IPv4 vs IPv6 vs vsock special casing.
> > > 
> > > Signed-off-by: Stefan Hajnoczi <address@hidden>
> > > ---
> > >  qapi-schema.json    |  23 +++++-
> > >  util/qemu-sockets.c | 222 
> > > ++++++++++++++++++++++++++++++++++++++++++++++++++++
> > >  2 files changed, 244 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/qapi-schema.json b/qapi-schema.json
> > > index c3dcf11..8864a96 100644
> > > --- a/qapi-schema.json
> > > +++ b/qapi-schema.json
> > > @@ -987,12 +987,14 @@
> > >  #
> > >  # @unix: unix socket
> > >  #
> > > +# @vsock: vsock family (since 2.8)
> > > +#
> > >  # @unknown: otherwise
> > >  #
> > >  # Since: 2.1
> > >  ##
> > >  { 'enum': 'NetworkAddressFamily',
> > > -  'data': [ 'ipv4', 'ipv6', 'unix', 'unknown' ] }
> > > +  'data': [ 'ipv4', 'ipv6', 'unix', 'vsock', 'unknown' ] }
> > > 
> > >  ##
> > >  # @VncBasicInfo
> > > @@ -3017,6 +3019,24 @@
> > >      'path': 'str' } }
> > > 
> > >  ##
> > > +# @VsockSocketAddress
> > > +#
> > > +# Captures a socket address in the vsock namespace.
> > > +#
> > > +# @cid: unique host identifier
> > > +# @port: port
> > > +#
> > > +# Note that string types are used to allow for possible future hostname 
> > > or
> > > +# service resolution support.
> > > +#
> > > +# Since 2.8
> > > +##
> > > +{ 'struct': 'VsockSocketAddress',
> > > +  'data': {
> > > +    'cid': 'str',
> > > +    'port': 'str' } }
> > 
> > Is there any reason to not define these as uint32_t? Not sure if there
> > are other reasons for this, but if it's just for consistency with how
> > Inet is handled, the code seems to do straight atoi()<->printf("%d") to
> > covert between numerical and string representation so it doesn't seem
> > like we need to account for any differences between command-line and
> > internal representation in sockaddr_vm.
> 
> Just in case AF_VSOCK ever supports name and service resolution like
> TCP/IP.  In that case cid could be a host name and port could be a
> service name.
> 
> (I mentioned this in the doc comment.)

Ahh, sorry I missed that completely. Makes perfect sense then.

> 
> Stefan