Andrew Lunn wrote:
On Tue, Jul 06, 2010 at 07:39:56PM +0200, Antonio Quartulli wrote:
Hi all, Hi Andreas,
+#define FRAG_TIMEOUT 60000 /* purge frag list entrys after time in ms */
Why did you choose a so large timeout? Is there a particular reason for that?
I think that in 60 seconds an host could receive a very large amount of "orphan" packets in case of a bad link causing memory problems.
It will not cause memory problems. The linked list is limited to FRAG_BUFFER_SIZE (6) segments.
He didn't say that it will take all available system memory. There is still the problem that the 6 slots per originator (so the memory per originator) could be full when we have a bad link and a too big waiting time for missing fragments.
Best regards, Sven