Date: Mon, 20 Jul 2015 20:04:00 +0800 From: arthurwang <wangzheng@jrq.com> To: net@freebsd.org, rizzo@iet.unipi.it Subject: Two netmap issues : 1. "protocol 0000 is buggy" , 2: network applications stuck Message-ID: <B0D3961A-B30B-426D-BB1C-2C2D0EEB5E4B@jrq.com> In-Reply-To: <8714904D-22E4-4738-ABF8-C42ACACFEF3B@jrq.com> References: <8714904D-22E4-4738-ABF8-C42ACACFEF3B@jrq.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi ,
I=E2=80=99m new to netmap , =E2=80=9Ca novel framework for fast =
packet I/O=E2=80=9D accroding to =
http://info.iet.unipi.it/~luigi/netmap/#85cb =
<http://info.iet.unipi.it/~luigi/netmap/#85cb>; , :). Im doing a test =
for its bridge.c example program , which do a bridging(namely proxying , =
I suppose ) job between NIC and host stack ,by running =E2=80=9C./bridge=
-i net map:ethic=E2=80=9D from command line , and I have two problems =
:
Problem 1 :=20
It issues =E2=80=9Cprotocol 0000 is buggy=E2=80=9D once per =
second , seems not running correctly, by contrast to the normal state , =
under which print =E2=80=9Cpoll timeout =E2=80=9C once per 2.5 seconds . =
This issue not always exist.=20
Here is the procedure to reproduce:
1> install a centOS 6.5 on vmware fusion 7( host: mbp )
=20
2> download kernel 2.6.32.67 from kernel.org =
<http://kernel.org/>, compile & install it, reboot the guest OS.=20
3> clone netmap from https://github.com/ppLorins/netmap =
<https://github.com/ppLorins/netmap>; , unpack & compile it , insert the =
netma.ko to kernel.
4> compile the bridge.c example=20
5> run example with =E2=80=9C./bridge -i netmap:eth0=E2=80=9D
Thus , it will print the error message =E2=80=9Cprotocol 0000 is =
buggy=E2=80=9D , I explore it by google for quite a long time, found =
nothing .=20
Environment info about the guestOS centOS6.5 , from a =E2=80=9Cn=
ame -a=E2=80=9D call : =20
Linux localhost.localdomain 2.6.32.67 #1 SMP Fri Jul 17 08:41:59 =
CEST 2015 i686 i686 i386 GNU/Linux
As for the NIC driver , I have no idea about it but it was =
simulated by vmware, sorry for this ,but if you need extra information =
about it , tell me how to view it by command line , and I will reply as =
soon as possible .=20
Problem 2:=20
Even though Problem 1 not exists , the program also appears to be =
something wrong , Here are what I found :
1> ping request from another machine can return normally ( same as I =
supposed ) =20
=20
2> new tcp connection(requested by telnet, destination port is 22 ) =
can be successfully established from other machine ( same as I =
supposed ) =20
3> existing session (pts/0 , pts/1 ) founded by a precede ssh call =
, are stuck , cannot enter any character or get any response into/from =
screen. (confusing)
4> new ssh request (represents all tcp applications ), cannot =
get response and timed out. (confusing )
In my assumption, there should be no difference with or without the =
bridge program running , for the upper layer network applications =
.Since bridge just acts as a proxy between NIC and host stack ,and do =
nothing except that . Thus all ip(ping) or tcp(ssh) layer =
applications should act as normal , but unfortunately just part of them =
are okay .
Can anyone help ? I know there are few people using netmap , it=E2=80=
=99s quite a novel technique for network application=E2=80=A6.
Thanks a lot.
=20
=E2=80=94 ArthurWang
=E2=80=94 Best Regards
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?B0D3961A-B30B-426D-BB1C-2C2D0EEB5E4B>