Re: [PATCH] pasta: make it possible to disable socket splicing

[Stefano Brivio <sbrivio@redhat.com>]

On Thu, 28 Nov 2024 20:21:23 -0500
Jon Maloy <jmaloy@redhat.com> wrote:

> I am even questioning if it is necessary: If the port is bound on the host,
> the client only needs to use some of the non-loopback addresses on the
> host to reach it via the tap interface.

The same applies to inbound traffic by the way, as I mentioned: just use
a non-loopback address in the namespace/guest to reach it via the tap
interface:

$ { sleep 1; : | nc -4 -N passt.top 9999; } & ./pasta -q --config-net -p nc.pcap -t 9999 -- nc -q1 -l 9999
[1] 2052460
[1]+  Done                    { sleep 1; : | nc -4 -N passt.top 9999; }
$ tshark -r nc.pcap ip
    8   0.968929 88.198.0.161 → 88.198.0.164 TCP 62 55898 → 9999 [SYN] Seq=0 Win=65535 Len=0 MSS=61440 WS=256
   11   0.968967 88.198.0.164 → 88.198.0.161 TCP 62 9999 → 55898 [SYN, ACK] Seq=0 Ack=1 Win=65480 Len=0 MSS=65480 WS=4096
   12   0.969000 88.198.0.161 → 88.198.0.164 TCP 54 55898 → 9999 [ACK] Seq=1 Ack=1 Win=65536 Len=0
   13   0.969017 88.198.0.161 → 88.198.0.164 TCP 54 55898 → 9999 [FIN, ACK] Seq=1 Ack=1 Win=65536 Len=0
   14   0.969063 88.198.0.164 → 88.198.0.161 TCP 54 9999 → 55898 [FIN, ACK] Seq=1 Ack=2 Win=65536 Len=0
   15   0.969118 88.198.0.161 → 88.198.0.164 TCP 54 55898 → 9999 [ACK] Seq=2 Ack=2 Win=65536 Len=0

...that is, I guess that you or David find this convenient, but strictly
speaking, it's also unnecessary (actually, I won't use it myself, because
it's more typing).

-- 
Stefano