[3429] in linux-net channel archive
Re: MTU problems
daemon@ATHENA.MIT.EDU (Michael Lausch)
Sun Jun 23 15:58:23 1996
To: inr-linux-net@ms2.inr.ac.ru (really kuznet@ms2.inr.ac.ru)
cc: linux-net@vger.rutgers.edu
In-reply-to: Your message of "23 Jun 1996 21:59:54 +0400."
<4qk0mq$76p@flint.inr.ac.ru>
Date: Sun, 23 Jun 1996 20:56:43 +0200
From: Michael Lausch <mla@gams.co.at>
>>>>> "kuznet" == really kuznet@ms2 inr ac ru <inr-linux-net@ms2.inr.ac.ru>
writes:
kuznet> Michael Lausch (mla@gams.co.AT) wrote:
kuznet> : 16:46:19.305143 ppp-dialin2.gams.co.at.1022 >
loki.gams.co.at.22: S
kuznet> : 1298278750:1298278750(0) win 512 <mss 1460>
kuznet> : 16:46:19.305143 loki.gams.co.at.22 > ppp-dialin2.gams.co.at.1022:
S
kuznet> : 1034541773:1034541773(0) ack 1298278751 win 31744 <mss 236>
kuznet> How loki guessed that mss=236 ?
That's one thing that bothered me. I tried this setup with different
MRU values for the PPP client.
kuznet> Besides that, this value is wrong, should be 196.
I know ;-)
kuznet> : 16:46:22.295810 loki.gams.co.at.22 > ppp-dialin2.gams.co.at.1022:
S
kuznet> : 1034541773:1034541773(0) ack 1298278751 win 31744 <mss 236>
kuznet> : 16:46:22.335819 ppp-dialin2.gams.co.at.1022 >
loki.gams.co.at.22: . ack 1
kuznet> : win 31744 [tos 0x10]
kuznet> : 16:46:22.335819 loki.gams.co.at.22 > ppp-dialin2.gams.co.at.1022:
P
kuznet> : 1:16(15) ack 1 win 31744 (DF) [tos 0x10]
kuznet> : 16:46:22.375828 ppp-dialin2.gams.co.at.1022 >
loki.gams.co.at.22: P
kuznet> : 1:16(15) ack 16 win 31744 (DF) [tos 0x10]
kuznet> : 16:46:22.375828 loki.gams.co.at.22 > ppp-dialin2.gams.co.at.1022:
P
kuznet> : 252:292(40) ack 16 win 31744 (DF) [tos 0x10]
kuznet> : 16:46:22.415836 ppp-dialin2.gams.co.at.1022 >
loki.gams.co.at.22: . ack 16
kuznet> : win 31744 [tos 0x10]
kuznet> Until now we did not see any packet with size > ppp mtu on
ethernet.
kuznet> PMTU discovery has not been yet started.
kuznet> : 16:46:22.575872 truncated-ip - 44 bytes missing!193.46.122.142 >
kuznet> : ppp-dialin2.gams.co.at: (frag 10725:256@1824) [tos 0x10]
kuznet> : ......
kuznet> loki sends truncated&garbled datagrams, loki believes that mtu=276
---> loki is broken or strongly misconfigured.
Well loki has a frech compiled linux 2.0.0 kernel. I don't think user
programs play any role in this, because PMTU is done at kernel level.
kuznet> BTW loki is the only host on your picture that needs PMTU
discovery.
I know. It seems that the IP fragmentation is broken when used by the
PMTU discovery. Maybe i'm going to insert some printk's in the kernel
to find out something more. What i've found out is that the MTU
discovery does work. So i think the fault may be the fragmentation
code.
kuznet> Alexey Kuznetsov.
---
g.a.m.s. edv dienstleistungen gmbh
finger mla@spirit.luga.or.at for PGP key
<http://www.gams.at/~mla>
