On Wed, Feb 05, 2003 at 09:55:36AM +1100, Shaun wrote: > I've been perusing the archives and noticed the (relatively) recent > discussion and patch applied to the main line regarding AIX BPF reads > returning EFAULT. The conclusion was the the AIX BPF driver returns an > EFAULT on the next read after it was forced to drop packets (as an > indication that packets have been dropped since the last read). I'm not > sure of this conclusion, we've lately been developing a pcap based > application that has been receiving EFAULTs but those errors are not > accompanied by any change to the dropped packet counter. If I > deliberately slow the program sufficiently to insure dropped packets I get > the EFAULTs and the dropped packet counter is incremented as expected.
So, so far, it appears that the rule is "an EFAULT is returned if a packet is dropped", not "an EFAULT is returned iff a packet is dropped" - i.e., it appears to be true that an EFAULT is returned if a packet is dropped, but it does not appear to be true that a packet was dropped if an EFAULT is returned. Given that we're currently just ignoring EFAULT and driving on - we're not doing anything with packet drop statistics if that happens - it probably doesn't matter why EFAULT is returned, unless there's some situation where the application (or library) *does* need to do something with an EFAULT. - This is the TCPDUMP workers list. It is archived at http://www.tcpdump.org/lists/workers/index.html To unsubscribe use mailto:[EMAIL PROTECTED]?body=unsubscribe
