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

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

--- Comment #9 from Richard Scheffenegger <rscheff@freebsd.org> ---
I can see a number of retransmission timeouts (so the theory around EPERM on
the ipsec may be valid).

There were 3 reordering/loss episodes in that short flow (169 logging events), 
and during the last one, the ACK has a SACK block, where the right edge moved 
left compared to an ACK received prior to that...

Which is peculiar (maybe reordering induced by the packet held up in
processing?),
but obviously the tcp stack shouldn't loose it's bearings accounting for such
events.

I'll update this once I have decoded more of the timing and events.

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