in_cksum_skip: out of data by 40290 ...

Hi ALL!
in syslog is - :


Is this normal? And if not - where the search cause of the problem?

No, this is not normal.
Can you post some more details on your network cards?

Some chips say they support hardware checksumming, but in reality do a poor job at it (as you can see in your log). This might be either a hardware or a driver issue.

To disable hardware checksumming:
# ifconfig interface -TXCSUM -RXCSUM -VLAN_HWCSUM
If this helps, you'd want to add those options to your interface configuration in /etc/rc.conf in order to make them permanent.

Because the length is too big for normal mbuf, I guess it was caused by missing host byte order conversion before doing checksum computation.
Are you using pf/ipfw/ipf?

Savagedlight said:

No, this is not normal.
Can you post some more details on your network cards?

Click to expand...

I use vmware esxi 4.1 freebsd FreeBSD 8.1 (cvs updated as guest os and netcard in esxi intel (em driver)

Savagedlight said:

Some chips say they support hardware checksumming, but in reality do a poor job at it (as you can see in your log). This might be either a hardware or a driver issue.

To disable hardware checksumming:
# ifconfig interface -TXCSUM -RXCSUM -VLAN_HWCSUM
If this helps, you'd want to add those options to your interface configuration in /etc/rc.conf in order to make them permanent.

Click to expand...

I tried, but the situation has not changed :-(

yongari@ said:

Are you using pf/ipfw/ipf?

Click to expand...

yes Yes - i I use mpd5 as pppoe server + radiusd + ipfw (for nat).
i I tried embedded em driver, em driver from intel site, em driver from yandex, but w/o without success..