On Fri, Oct 28, 2011 at 09:00:59AM +0200, Stig Bj�rlykke wrote:
> On Thu, Oct 27, 2011 at 9:12 PM, Stephen Fisher
> <steve@xxxxxxxxxxxxxxxxxx> wrote:
> > Is there a problem with accepting -1 in that filter?
>
> It's not a problem, but it's a bug in the logic because the filter
> does not do what it's supposed to.
I understand now: instead of the filter showing tcp.len that is less
than -1 (so -2 and so on), it's showing less than the underflow value of
MAXINT.
> > If so, should the filter be checked against possible values of the
> > value, i.e. tcp.len is a FT_UINT32 so only accept unsigned 32-bit
> > values and mark the background as red / bad filter if not?
>
> The previously attached patch does check for signed/unsigned issues,
> and will mark the filter as bad/red. I think it would be nice to check
> all values if they are valid for the given field.
Good idea. I wonder how much work that would be... never thought of
that.
