[Graeme Hewson <ghewson@xxxxxxxxxxxxxxxxxxx>]

Graeme Hewson wrote:

> No replies to this, so here's a series of patches to disallow negative
> slice lengths and fix a few other things.
>
> scanner.l
>     Use gint32 instead of guint32 and strtol() instead of strtoul()
>     for signed integers.
>
>     Pathological slice specifications could cause Flex default rule
>     to be invoked, echoing characters to stdout.
>     Example: frame[0foo]==1
>
> syntax-tree.h
> syntax-tree.c
>     Use gint32 instead of guint32 for node data.
>
> semcheck.c
>     Check slice lengths as well as offsets.  Disallow negative/zero
>     lengths.
>
>     Range on RHS of display filter expression wasn't being checked in
>     every case.
>
> ftypes.c
>     Fix assertion failure when absolute value of negative slice
>     offset > field length.
>
>     Remove code for handling negative slice lengths.
>
> ethereal-filter.pod.template
>     Remove reference to negative slice lengths.
>
>     Clean and polish.

This set of patches seems to have been overlooked.  Please could it be 
checked in.

If anyone wants to defend negative slice lengths, I can put them back 
in, but more importantly there's a fix for an assertion failure, and the 
 patch to the documentation is well worthwhile, IMHO, amounting to 
changes on over 100 lines.


Graeme Hewson

Attachment: dfilter.tar.gz
Description: application/gzip