off by one bounds

[Skip Ford]

Maxim Konovalov wrote:
> On Sat, 21 Aug 2004, 13:19+0400, Maxim Konovalov wrote:
> > On Sat, 21 Aug 2004, 05:00-0400, Skip Ford wrote:
> > > Maxim Konovalov wrote:
> > > > On Fri, 20 Aug 2004, 12:36-0700, Ted Unangst wrote:
> > > >
> > > > > errors in freebsd 4.10 found by Coverity's analysis.
> > > >
> > > > > ip_icmp.c:ip_next_mtu, i == sizeof, dir >= 0
> > > >
> > > > If i == sizeof then mtutab[i] == 0
> > >
> > > If "i == sizeof" then mtutab[i] is out of bounds, off by one.
> > > There is no mtutab[sizeof mtutab / sizeof mtutab[0]].
> > >
> > > This isn't specific to RELENG_4
> 
> After the second thought I still think it is not a error.  mtu is
> always >= than the minimal value in mtutab[] that is why i is always
> less than (sizeof mtutab) / sizeof mtutab[0]).  What do you think?

I have no idea if it can actually be triggered.

Callers may never invoke it with the necessary parameters but if
they do, the function doesn't handle it.  If mtu is < 0 and dir >= 0
then mtutab is accessed out of bounds.

-- 
Skip