Subject: magic mtu limit of 8192 bytes (one VM page)
To: None <>
From: Ignatios Souvatzis <>
List: tech-net
Date: 09/25/1999 19:10:13
Is there a magic 8192 bytes or one VM page limit in the IPv6 sending code,
or maybe in ping6, although it claims it was the send call?

jocelyn AMIGA40 !% ping6 -s 8184 fe80:1::4d

PING6(8232=40+8+8184 bytes) fe80::49 --> fe80:1::4d
8192 bytes from fe80::4d, icmp_seq=0 hlim=64 time=88.092 ms
--- fe80:1::4d ping6 statistics ---
1 packets transmitted, 1 packets received, 0% packet loss
round-trip min/avg/max = 88.092/88.092/88.092 ms

jocelyn AMIGA40 !% ping6 -s 8185 fe80:1::4d

PING6(8233=40+8+8185 bytes) fe80::49 --> fe80:1::4d
ping6: sendmsg: Message too long
ping6: wrote fe80:1::4d 8193 chars, ret=-1
--- fe80:1::4d ping6 statistics ---
1 packets transmitted, 0 packets received, 100% packet loss

Note that the interface in question should has a MTU of 60480 bytes (default
should be smaller, but isn't currently implemented).

It works fine with IPv4:

jocelyn AMIGA40 103% ping -s 9000 beverly

PING ( 9000 data bytes
9008 bytes from icmp_seq=0 ttl=0 time=96.945 ms
9008 bytes from icmp_seq=1 ttl=0 time=96.460 ms
^C9008 bytes from icmp_seq=2 ttl=0 time=96.505 ms PING Statistics----
3 packets transmitted, 3 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 96.460/96.637/96.945/0.268 ms