Re: [PATCH v4 2/4] util/qemu-sockets.c: Split host:port parsing out of inet_parse

[Doug Evans <dje@google.com>]

[-- Attachment #1: Type: text/plain, Size: 2843 bytes --]

On Sat, Mar 6, 2021 at 11:29 AM Samuel Thibault <samuel.thibault@gnu.org>
wrote:

> Hello,
>
> Doug Evans, le ven. 05 mars 2021 17:00:13 -0800, a ecrit:
> > Is it possible for QEMU to lazily determine the guest's IPv6
> > address? I.e., postpone the ""->guest address mapping until it's
> > needed and then, say, take the first entry in the NDP table?
>
> That would probably be possible, yes, by moving the
>
> if (!guest_addr.s_addr) {
>     guest_addr = slirp->vdhcp_startaddr;
> }
>
> from slirp_add_hostfwd() and alike to tcp_connect() and sorecvfrom()
> (along the other sotranslate call).
>
> > That feels a bit fragile: what if someone else gets the first entry in
> > the NDP table? But is that any more fragile than assuming the first
> > handed out DHCP address is to the guest?
>
> I don't think it's really more fragile.
>


Good to know, thanks.


> > [<<-- Honest question, can we assume the first handed out DHCP address
> > will necessarily be the guest?]
>
> It "cannot" be anything else. What could happen is a PXE loader that
> uses DHCP/NDP, and then the OS that does it again.
>
> > But that would mean the defaults for the guest would have to be
> > different than for the host. E.g.,
> > host: ",ipv4" means both,
>
> Why would it mean both? I don't follow you here.
>
> > whereas guest: ",ipv4" (ideally) means ipv4 (since both is meaningless)
>


I guess one has to define:
- how these flags work,
- do they have defaults,
- and if they do have defaults what is the default value?

For the host, neither flag present means "both are on", which could mean,
effectively, that the defaults for
ipv4[={off,on}] and ipv6[={off,on}] are both "on".
[Assuming they have defaults. See above: Do they?
For the guest ipv4=on,ipv6=on is an error.]
But does that then mean that the presence of only "ipv4=on" or "ipv6=on" is
a no-op?
After all, why specify "ipv4=on" at all if it's the default?
I think a reader would get confused if they saw only one of "ipv4=on" or
"ipv6=on"
specified and learned that both were on.
It also means that to specify only one of ipv4 or ipv6 you have to turn the
other off.
It's a bit awkward, but it is consistent and easy to explain (if awkward to
use and read).

On the other hand, for the host, one could, for example,
make these flags tri-state (or call it whatever).
Is specifying only "ipv4=off" the equivalent of specifying only "ipv6=on"?
Presumably it must (it makes the most sense).
There is also the invalid setting of ipv4=off,ipv6=off.
One also needs to specify the order in which the flags are processed,
to define what ipv6=on,ipv6=off means.
Either that or document that specifying them multiple times is undefined.

This is getting a bit verbose to have to explain in documentation,
but it is what it is.
I don't have a say in the decision. I just need to know what to implement.

[-- Attachment #2: Type: text/html, Size: 5508 bytes --]