ipfw2 net.inet.ip.fw.verbose_limit broken

Maxim Konovalov maxim at macomnet.ru
Sun Aug 29 01:41:50 PDT 2004


> On Mon, 16 Aug 2004, 23:05-0700, Jeremy Chadwick wrote:
> > Just wanted to toss this one up here.  Also, apologies for not
> > cross-posting this to freebsd-ipfw, but I'm not on the list; although
> > they seem to be aware of it:
> >
> > http://lists.freebsd.org/mailman/htdig/freebsd-ipfw/2004-July/001239.html
> >
> > Seems that ipfw2's support for net.inet.ip.fw.verbose_limit is, to
> > put it bluntly, broken.  This applies to both -STABLE and -CURRENT.
> > The following PR has been sitting around for quite some time...
> >
> > http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/46080

I have fixed this bug in -CURRENT and will MFC it in a week to
RELENG_4 and RELENG_5 with the re@ approval.

> > I've managed to confirm this still exists even as of an August 5th build
> > of -CURRENT.  Using `logamount' directives per rule works properly as
> > a workaround.
> >
> > I've also looked at the patch, although I'm not sure about the performance
> > implications of looking up a sysctl value per packet with a matching
> > ipfw2 `log' directive.

No, it works in a different way.  There are max_log and log_left
fields in ipfw_insn_log sructure associated with each ipfw rule.  They
are checked on every packet matches the rule.  That is why a changing
verbose_limit value does not affect the existent rules.

Maxim Konovalov

More information about the freebsd-current mailing list