Re: Automated report: NetBSD-current/i386 test failure

[Martin Husemann <martin%netbsd.org@localhost>]

On Sun, Sep 30, 2018 at 06:13:41PM +0000, NetBSD Test Fixture wrote:
> This is an automatically generated notice of new failures of the
> NetBSD test suite.
> 
> The newly failing test cases are:
> 
>     net/ipsec/t_ipsec_natt:ipsec_natt_transport_null
>     net/ipsec/t_ipsec_natt:ipsec_natt_transport_rijndaelcbc

This is due to npfctl not being "hijacked" properly:

tc-so:Executing command [ rump.ping -c 1 -n -w 3 20.0.0.2 ]
tc-so:
tc-so:00:00:02.290008 b2:a0:24:e0:87:88 > ff:ff:ff:ff:ff:ff, ethertype ARP (0x08
06), length 42: Request who-has 20.0.0.2 tell 20.0.0.1, length 28
tc-so:00:00:02.010008 b2:a0:f7:f3:4e:44 > b2:a0:24:e0:87:88, ethertype ARP (0x08
06), length 42: Reply 20.0.0.2 is-at b2:a0:f7:f3:4e:44, length 28
tc-so:00:00:02.290008 b2:a0:24:e0:87:88 > b2:a0:f7:f3:4e:44, ethertype IPv4 (0x0
800), length 98: 10.0.1.2 > 20.0.0.2: ICMP echo request, id 33511, seq 0, length
 64
tc-so:00:00:02.010008 b2:a0:f7:f3:4e:44 > b2:a0:24:e0:87:88, ethertype IPv4 (0x0
800), length 98: 20.0.0.2 > 10.0.1.2: ICMP echo reply, id 33511, seq 0, length 6
4
tc-so:Executing command [ cat ./out ]
tc-so:Executing command [ cat ./out ]
tc-so:Executing command [ env LD_PRELOAD=/usr/lib/librumphijack.so RUMPHIJACK=path=/rump,socket=all:nolocal,sysctl=yes,blanket=/dev/npf npfctl reload ./npf.conf ]
tc-se:Fail: incorrect exit status: 1, expected: 0


Ktraceing it shows the failing call is sendmsg() - which should be properly
intercepted by librumphijack, but talks to the host kernel and gets errno 32.

Martin