[Bug 275798] panic: sackhint bytes rtx >= 0

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=275798

--- Comment #10 from Richard Scheffenegger <rscheff@freebsd.org> ---
It appears that the issue may have to do with some LRO going on as well on the
receiver side - which doesn't ACK each individual packet, but typically 2
consecutive one, even while the receiver knows loss/reordering is on-going. At
that stage the receiver normally would ACK every packet received...

The ultimate issue appears to happen after a RTO timeout, when the last
outstanding (2-packet) hole is filled, and the cumulative ACK covering it all
also
contains a DSACK block at the beginning of the former hole...

Normally, such a DSACK block should be discareded and ignored during the SACK
update process - not really sure why it's apparently processed erraneously,
leading to the panic...

-- 
You are receiving this mail because:
You are on the CC list for the bug.