Wireshark · Ethereal-dev: Re: [Ethereal-dev] Validating IPs?

[Guy Harris <gharris@xxxxxxxxx>]

Alex Kirk wrote:
> Already had, actually. There's nothing that extracts IP addresses directly (at
> least not that I can see), and getting the bytes individually with
> tvb_get_guint8(), converting them to a string with dots, and then calling
> inet_aton() seemed wasteful.

It's especially wasteful because it won't tell you waht you want to know!

If you convert the 4 bytes to a string with dots, then "inet_aton()" 
will *ALWAYS* be able to convert that string back to an IP address - it 
will *never* fail on a string of the form

	{integer from 0 to 255}.{integer from 0 to 255}.{integer from 0 to 
255}.{integer from 0 to 255}

and converting 4 bytes to a string with dots involves treating each byte 
as an unsigned integer in the range 0 to 255 and putting dots between them.

I.e., as implied by Michael Tuexen's mail, all 2^32 4-byte integral 
values are, in some sense, syntactically valid IP addresses, although 
there are conventions that render some of them not valid as host 
addresses - but "inet_aton()" doesn't know that it's being asked to 
convert a valid host IP address, and doesn't check for an address being 
valid (doing so would require that it know what the network mask for the 
address is - and it might well be used in applications that are 
*intentionally* being supplied addresses with zeroes in the host part 
and that are *expected* to convert them to IP addresses).