Status Update
Zach Riggle
zachriggle at gmail.com
Mon Jun 29 22:48:27 UTC 2009
Looks like I might've skipped a status update, but I'll try to fill in
the gaps.
The TCP regression testing framework is currently at a state where
using it to actually *do* testing should be viable. Packets are
constructed, chained, injected, and processed properly. The TCP state
machine is partially functioning (i.e. if its state is 'closed' and
you tell it to send a packet with syn=1, it should move to SYN-SENT,
etc.), and will continue development as pertinent tests are being
written that require the functionality.
Yesterday I ran into some issues where, upon injecting a packet (for
example, a SYN packet to initiate a connection) the OS would see the
incoming response (SYN-ACK) and send a RST packet to terminate the
connection because, as far as the OS's TCP stack is concerned, there
is no open socket that the data would be going to. Turning on the
'stealth' feature, which should disable these packets, does not work
on loopback. Spoofing a source address also does not work the way it
should, for some reason. I'd really like the TCP tests to *not*
require a second host to function, so this is something I'm going to
have to nail down.
Aside from that, and a few typos here and there, things are coming
along swimmingly!
More information about the soc-status
mailing list