[En-Nut-Discussion] (In)stability of NutOS

Bernd Walter enut at cicely.de
Tue Feb 21 01:41:24 CET 2006


On Mon, Feb 20, 2006 at 05:36:55PM +0100, Harald Kipp wrote:
> The obvious difference to the systems, for which failures
> had been reported here, is that HTTP is not used and TCP
> connections are seldom closed and re-opened.

I recently noticed a bug with the rs232d example.
Nothing bad seemed to happen under 3.9, but the system regulary
stopped when a TCP connect was dropped and reconnected under
4.0.2.
The problem was the receiver-thread, which tried writing to
an outdated filedescriptor.
The device on the rs232 was a server which did not send anything
unless something was typed, so the receiver thread blocked forever
until you reconnect and type something.
Maybe something similar can happen within the HTTP codepath.
I don't know the HTTP code in detail, so it's just a wild guess,
it could also be impossible.
Otherwise I can say, that in gerenal there is much that can be
done wrong while implementing your own TCP threads.

-- 
B.Walter                   BWCT                http://www.bwct.de
bernd at bwct.de                                  info at bwct.de




More information about the En-Nut-Discussion mailing list