[En-Nut-Discussion] (no subject)

[Michael Svob]

Hello,

I have an application where I am connecting to the Ethernut from a PC based
client using a TCP/IP socket connection (I used the 'tcps' sample code as my
"template").  Everything works fine, so long as I shut down the PC client
gracefully, which in turn sends a message to the Ethernut telling it to
terminate the session.

However, if I abruptly lose communication between the PC and the Ethernut
(e.g. PC crashes or reboots), then I cannot reconnect to the Ethernut
without resetting it.  I added a "keep alive" mechanism, whereby if the
Ethernut thinks it's connected, it looks for a keep-alive message sent by
the client on a regular basis.  If it fails to see this message for some
time, it attempts to tear down the connection on its side so it can wait for
a new socket request.

The problem is that in my ProcessRequests() function, fread() sits there and
happily blocks forever for data that never comes.  I would have expected
that when the socket stream is closed (by my keep-alive watchdog thread),
that fread would return due either to a read error, or end-of-file.
However, this is not the case; fread() just waits indefinitely.

Is there a straightforward way to get fread() to return from blocking so
that program flow can continue?  Is there a better way to not only detect
the loss of connection with the PC client, but more importantly close the
connection on the Ethernut side?

Thanks!!!

Michael Svob
LFI International