Problems with control (TCP) connections

[borg at uu3.net]

Okey, I found out with commit introduced this behaviour:
commit d3297fbd3b8c8c8a4661f5bbf89aca5cacba8b5a
Prevent oracle attacks (CVE-2018-16737, CVE-2018-16738)

Can someone explain this in more technical details?
I tried to find more data about it but there are more articles
about Padding Oracle Attack that with (I belive) is not the case here.

Why this was done that way? Can it be done better?
I am not really sure what lingering TCP connection would even
achieve here.

---------- Original message ----------

From: borg at uu3.net
To: tinc-devel at tinc-vpn.org
Subject: Problems with control (TCP) connections
Date: Thu, 28 Feb 2019 21:59:37 +0100 (CET)

Hello.

Im running tinc from long time, but recently I noticed that
tinc TCP connections are not closed correctly (timeout).
They seem to linger around like this:

Proto Recv-Q Send-Q Local Address           Foreign Address         State
tcp       80      0 xxx.x.x.xx:58801        xx.xxx.xxx.xx:***       ESTABLISHED
tcp      515      0 xxx.x.x.xx:45422        xx.xxx.xxx.xxx:***      ESTABLISHED
tcp        0      0 xxx.x.x.xx:34066        xx.xxx.xxx.xxx:***      CLOSE_WAIT
tcp        0      0 xxx.x.x.xx:37178        xx.xxx.xxx.xxx:***      CLOSE_WAIT
tcp        0      0 xxx.x.x.xx:58106        xx.xxx.xxx.xxx:***      CLOSE_WAIT
tcp      515      0 xxx.x.x.xx:43048        xx.xxx.xxx.xx:***       ESTABLISHED
tcp        0      0 xxx.x.x.xx:57942        xx.xxx.xxx.xxx:***      CLOSE_WAIT

Anyone else noticed this? Im running tinc version 1.0.35

Regards,
Borg
_______________________________________________
tinc-devel mailing list
tinc-devel at tinc-vpn.org
https://www.tinc-vpn.org/cgi-bin/mailman/listinfo/tinc-devel