To my understanding too. We will try to look closer at this, maybe there
are smarter alternatives at least on some platforms (PF_PACKET under
Linux...?).
Klaus explained to me that there is some switch in winpcap to support
this, but it caused troubles due to weird packet loops. I don't have a
Windows platform to test, but I will see if we can clarify this in some
other way.
That's a blocker IMHO because it will invariably lead to many folks
asking why ping to the host doesn't work (just like slirp). Since it
still requires root privileges, I don't think it improves a lot on tap.
Well, in case we do not find a workaround for host<->guest
communication, would the patch remain unacceptable even with proper,
prominent documentation of this shortcoming?