Gerald Combs wrote:
Why does the dissector have its own MAC_to_str() function? Wouldn't
ether_to_str() or get_ether_name() work?
As far as I can tell, it would, so I got rid of MAC_to_str() in favor of
ether_to_str().
The code beginning at line 619 seems odd:
strcpy(tempStr,MAC_to_str(mac_addr, 6, ':'));
proto_tree_add_text(chassis_tree, tvb, (offset+3), 6,
"Chassis Id: %s", tempStr);
proto_tree_add_ether_hidden(chassis_tree, hf_chassis_id_mac,
tvb, (offset+3), 6, mac_addr);
Couldn't this just be
proto_tree_add_ether(chassis_tree, hf_chassis_id_mac, tvb,
(offset+3), 6, mac_addr);
It probably should be, so I made it do so, and also got did so in some
other cases.
The code uses strcpy() a LOT. Can these be removed?
I've removed most of them, and am building a change to get rid of the
last of them. That should get rid of most if not all of the on-stack
buffers for strings, so that there are fewer places where we have to
worry about whether the buffer can overflow or not.
I also change it to use tvb_format_text() and tvb_format_stringzpad(),
so that non-printable characters in strings don't cause problems.
In dissect_media_tlv(), shouldn't LCI_Length be unsigned in order to
avoid excessive looping?
It already is unsigned.
However, more stringent checks should be done, to make sure it never
goes "negative"; I've added those.
