In some email I received from [EMAIL PROTECTED], sie wrote:
> OpenBSD 2.9 i386 custom kernel
> IP Filter v.3.4.23
> PIII-833MHz
> 
> Every once in a while, I have cron run "ipfstat -ios" into a text file
> so I can awk it for statistics.  A couple of times a day (87 times since
> Mar 13, 2002), ipfstat throws the error "pos=0xdeadbeef kmemcpy:read: Bad
> address" to my cron account's mailbox.  Any idea why?  

Enumeration of the state list happens in user space and is always "out of
sync with the kernel".  That is, between printing it out and attempting to
get the next one, the list contents may change or the memory to where it
is going to go next change.  Almost calls for a "snapshot" type call where
the list is copied back, atomically to ipfilter, to user space.  This
could take quite some time, however, for "busy" systems.

Darren

Reply via email to