secnet robustness: tolerate SLIP decoding errors

[Simon Tatham]

As discussed with Ian on Sunday, I attach a patch to secnet which
makes it stop treating SLIP decoding failures as fatal error
conditions worthy of assertion failure or secnet termination, and
instead logs them and resynchronises at the next packet boundary.

(I've recently been running secnet with gdb attached in order to track
down a behavioural oddity, and have noticed that gdb suspending the
secnet process tends to have the side effect that corruptions can
appear in the SLIP stream after it's restarted; presumably some data
got lost when the kernel got impatient. Therefore, Ian suggested that
such corruption should no longer be an assertion-failure condition.)

Cheers,
Simon
-- 
Simon Tatham         "A cynic is a person who smells flowers and
<anakin at pobox.com>    immediately looks around for a coffin."
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-slip-tolerate-SLIP-decoding-errors.patch
Type: text/x-diff
Size: 3732 bytes
Desc: not available
URL: <http://www.chiark.greenend.org.uk/pipermail/sgo-software-discuss/attachments/20130625/090a8966/attachment.patch>