Re: VKernel progress update - 10 Jan 2006

["Sepherosa Ziehau" <sepherosa@xxxxxxxxx>]

On 1/11/07, Matthew Dillon <dillon@apollo.backplane.com> wrote:
>     Buildworld blows past the point where it failed before.  Hopefully
>     it will be done when I wake up in a few hours :-).  I've definitely
>     squashed the last major VM bug.
>
>     Real (non virtual) kernels will be slightly nerfed until I am able to
>     commit a followup later on thursday.  The virtual kernel support
>     requires the real kernel to force a fault on (initial) page writes in
>     order to be able to set the Modify bit in the virtual page table entries,
>     but the fault is currently being forced on ALL user pages.  Needless
>     to say, that is not entirely desireable.  I will fix this on thursday to
>     make it apply only to virtual page tables.  Doing it right is somewhat
>     non-trivial.

Multi network interface support for vkernel:
sephe-test1:VKERNEL# ./kernel.debug -I /dev/tap0 -I /dev/tap1 -m 128m
-r /opt/vkernel/rootimg.01

on another terminal:
sephe-test1:~# ifconfig tap1 up
sephe-test1:~# ifconfig tap0 up
sephe-test1:~# ifconfig bridge0 addm nfe0 addm tap1
sephe-test1:~# ifconfig bridge1 addm re0 addm tap0
sephe-test1:~# ifconfig bridge1 up
sephe-test1:~# ifconfig bridge0 up

in vkernel:
# ifconfig -a
vke0: flags=8802<BROADCAST,SIMPLEX,MULTICAST> mtu 1500
       ether 00:be:63:e0:15:00
vke1: flags=8802<BROADCAST,SIMPLEX,MULTICAST> mtu 1500
       ether 00:be:66:e0:15:01
lo0: flags=8008<LOOPBACK,MULTICAST> mtu 16384
ppp0: flags=8010<POINTOPOINT,MULTICAST> mtu 1500
sl0: flags=c010<POINTOPOINT,LINK2,MULTICAST> mtu 552
faith0: flags=8002<BROADCAST,MULTICAST> mtu 1500
# dhclient vke0
# ifconfig -a
vke0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
       inet6 fe80::2be:63ff:fee0:1500%vke0 prefixlen 64 scopeid 0x1
       inet 192.168.3.122 netmask 0xffffff00 broadcast 192.168.3.255
       ether 00:be:63:e0:15:00
vke1: flags=8802<BROADCAST,SIMPLEX,MULTICAST> mtu 1500
       ether 00:be:66:e0:15:01
lo0: flags=8008<LOOPBACK,MULTICAST> mtu 16384
ppp0: flags=8010<POINTOPOINT,MULTICAST> mtu 1500
sl0: flags=c010<POINTOPOINT,LINK2,MULTICAST> mtu 552
faith0: flags=8002<BROADCAST,MULTICAST> mtu 1500
# ifconfig vke1 inet 192.168.2.100
# ifconfig -a
vke0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
       inet6 fe80::2be:63ff:fee0:1500%vke0 prefixlen 64 scopeid 0x1
       inet 192.168.3.122 netmask 0xffffff00 broadcast 192.168.3.255
       ether 00:be:63:e0:15:00
vke1: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
       inet 192.168.2.100 netmask 0xffffff00 broadcast 192.168.2.255
       inet6 fe80::2be:66ff:fee0:1501%vke1 prefixlen 64 scopeid 0x2
       ether 00:be:66:e0:15:01
lo0: flags=8008<LOOPBACK,MULTICAST> mtu 16384
ppp0: flags=8010<POINTOPOINT,MULTICAST> mtu 1500
sl0: flags=c010<POINTOPOINT,LINK2,MULTICAST> mtu 552
faith0: flags=8002<BROADCAST,MULTICAST> mtu 1500
# ping 192.168.3.100
PING 192.168.3.100 (192.168.3.100): 56 data bytes
64 bytes from 192.168.3.100: icmp_seq=0 ttl=64 time=111.181 ms
64 bytes from 192.168.3.100: icmp_seq=1 ttl=64 time=99.948 ms
64 bytes from 192.168.3.100: icmp_seq=2 ttl=64 time=79.950 ms
^C
--- 192.168.3.100 ping statistics ---
3 packets transmitted, 3 packets received, 0% packet loss
round-trip min/avg/max/stddev = 79.950/97.026/111.181/12.916 ms
# ping 192.168.2.1
PING 192.168.2.1 (192.168.2.1): 56 data bytes
64 bytes from 192.168.2.1: icmp_seq=0 ttl=64 time=177.495 ms
64 bytes from 192.168.2.1: icmp_seq=1 ttl=64 time=19.974 ms
64 bytes from 192.168.2.1: icmp_seq=2 ttl=64 time=19.976 ms
64 bytes from 192.168.2.1: icmp_seq=3 ttl=64 time=19.962 ms
^C
--- 192.168.2.1 ping statistics ---
4 packets transmitted, 4 packets received, 0% packet loss
round-trip min/avg/max/stddev = 19.962/59.352/177.495/68.210 ms


Now vke(4) will have same unit as the backend tap(4), i.e. if you
specify only -I /dev/tap1 then vke1 will appear, instead of vke0.
This can make bridge configuration in real kernel less confusion.

Patch is at:
http://leaf.dragonflybsd.org/~sephe/vke_multi.diff

If no objection comes, it will be committed tomorrow.

Best Regards,
sephe

--
Live Free or Die