Kernel panic when trying to ping6

[Aleksej Lebedev]

Sorry, I guess I should have simply create an issue on bugs.dragonflybsd.org.
Just did so.

But while here, is it a correct way to add a route the way I did? I mean this:

route add -inet6 fe80::201:71ff:fe0c:f607 -iface vke0

(The address is a link-local address of the second vkernel.)

Thanks.

--
Aleksej Lebedev

On Sun, Mar 8, 2020, at 16:44, Aleksej Lebedev wrote:
> Hi,
> 
> I got a kernel panic on my vkernel trying to ping6 another vkernel on 
> the same host.
> I might have done an incorrect network configuration, but I guess it 
> shouldn't panic anyway.
> 
> Here is what I've done.
> 
> Two fresh vkernels with ipv6_enable=yes in rc.conf started with:
> 
> /boot/kernel/kernel -m 1g -r /vhost/vm0.img -I /var/run/vknet 
> 
> and
> 
> /boot/kernel/kernel -m 1g -r /vhost/vm1.img -I /var/run/vknet
> 
> vknetd is executed as "/usr/sbin/vknetd -U -t tap0".
> 
> I run the following version of DragonFly (on all three machines: host 
> and both vkernels):
> 
> # uname -a
> DragonFly  5.8-RELEASE DragonFly v5.8.0rc1.28.g3e3e37-RELEASE #2: Sun 
> Mar  1 17:46:44 CET 2020     
> root at fi.zta.lk:/usr/obj/usr/src/sys/VKERNEL64  x86_64
> 
> On one of the vkernels I first check that I can ping the other one:
> 
> # ping6 fe80::201:71ff:fe0c:f607%vke0
> PING6(56=40+8+8 bytes) fe80::201:68ff:fef8:f5f8%vke0 --> 
> fe80::201:71ff:fe0c:f607%vke0
> 16 bytes from fe80::201:71ff:fe0c:f607%vke0, icmp_seq=1 hlim=64 
> time=0.371 ms
> 16 bytes from fe80::201:71ff:fe0c:f607%vke0, icmp_seq=2 hlim=64 
> time=0.262 ms
> ^C
> --- fe80::201:71ff:fe0c:f607%vke0 ping6 statistics ---
> 3 packets transmitted, 2 packets received, 33.3% packet loss
> round-trip min/avg/max/std-dev = 0.262/0.317/0.371/0.054 ms
> 
> Then do the following to get the panic:
> 
> # route add -inet6 fe80::201:71ff:fe0c:f607 -iface vke0
> add host fe80::201:71ff:fe0c:f607: gateway vke0
> 
> # ping6 fe80::201:71ff:fe0c:f607
> PING6(56=40+8+8 bytes) fe80::201:68ff:fef8:f5f8%vke0 --> 
> fe80::201:71ff:fe0c:f607
> panic: assertion "sdl->sdl_family == AF_LINK && sdl->sdl_alen != 0" 
> failed in nd6_resolve at /usr/src/sys/netinet6/nd6.c:2031
> cpuid = 0
> Trace beginning at frame 0x8026783680
> nd6_resolve() at 0x60ab55
> nd6_resolve() at 0x60ab55
> ether_output_frame() at 0x59427c
> ip6_output() at 0x6026a1
> rip6_output() at 0x610e9f
> rip6_output() at 0x611446
> Debugger("panic")
> 
> CPU0 stopping CPUs: 0x0000000000000002
>  stopped
> Stopped at      0x70bde9:       movb    $0,0xa0c83c(%rip)
> db> trace
> Debugger() at 0x70bde9
> panic() at 0x4c7710
> nd6_resolve() at 0x60ab55
> ether_output_frame() at 0x59427c
> ip6_output() at 0x6026a1
> rip6_output() at 0x610e9f
> rip6_output() at 0x611446
> netmsg_sync_handler() at 0x5a1739
> 
> --
> Aleksej Lebedev
>