From: Jeff Dike <jdike@ka...> - 2002-10-20 17:55:59
> *all* non-even sized packets are getting tossed on mcast transports,
> and possibly other transports as well.
I just set up an mcast network between two UMLs, did the ol ping, and got:
usermode:~# ping -s 221 192.168.0.253
PING 192.168.0.253 (192.168.0.253): 221 data bytes
229 bytes from 192.168.0.253: icmp_seq=0 ttl=64 time=53.2 ms
229 bytes from 192.168.0.253: icmp_seq=1 ttl=64 time=2.0 ms
229 bytes from 192.168.0.253: icmp_seq=2 ttl=64 time=1.9 ms
This looks fine here. Can you reproduce it with the latest UML?
I have a hard believing that there's a generic odd-size packet bug because
that would be screwing up all kinds of things. I'd guess that ~50% of all
packets are odd-sized, and we definitely would have noticed 50% packet loss
From: Steve Schmidtke <steve_schmidtke@ho...> - 2002-10-20 21:33:16
>This looks fine here. Can you reproduce it with the latest UML?
Yes, I can, but now that I've explored more, I found I can reproduce the
odd-sized-packet-loss problem if I use the same ping binary on the host. :(
It looks like the ping binary from slackware 8 is broken, not UML.
Surf the Web without missing calls! Get MSN Broadband.