|
Posted by Sandman on 02/28/06 21:11
In article <BuJMf.19835$bw1.4310@newsfe2-win.ntli.net>,
Oli Filth <catch@olifilth.co.uk> wrote:
> Sandman said the following on 27/02/2006 20:27:
> > In article <mYIMf.60881$mf2.26167@newsfe6-win.ntli.net>,
> > Oli Filth <catch@olifilth.co.uk> wrote:
> >
> >> Sandman said the following on 27/02/2006 19:59:
> >>> So, I have this list of valid IP scopes, in the form below. How do I
> >>> match if $_SERVER[REMOTE_ADDR] is covered in any of these scopes?
> >>>
> >>> 193.11.120.0/21
> >>> 193.11.128.0/24
> >>> 193.11.129.0/24
> >>> 193.11.130.0/24
> >>> 193.11.131.0/24
> >>>
> >>>
> >>> First, I don't really know how to interprete the "/24" ending. I am
> >>> guessing that "193.11.131.0/24" means "193.11.131.X to 193.11.131.Y"
> >>> or something, but what? When that's translated to something useful,
> >>> how do I match IP numbers reliably?
> >> The /XX represents the length of the subnet mask in bits, so your hunch
> >> is pretty much correct.
> >>
> >> An IP address matches a given network IP address if:
> >>
> >> (Address ^ Mask) == (NetAddress ^ Mask)
> >>
> >> where Mask = 11111...0000, the number of ones given by the /XX.
> >
> > Ok, but how do I calculate it? Bits you say, but how do I translate 24
> > bits to addresses? For instance, the first line above, what addresses
> > does it cover?
> >
>
> An IP address is a 32-bit quantity, which can be represented as A.B.C.D,
> where the actual 32-bit value is given by:
>
> Y = (A * 2^24) + (B * 2^16) + (C * 2^8) + D
>
> [I'm using ^ to represent "to the power of" in this case.]
>
> So convert your test address and network address to this form, either
> directly or by using the ip2long() function.
>
> Then form your subnet mask as a 32-bit value.
> [HINT: (2^X - 1) = (1000....000 - 1) = 111....111]
>
> Then test the equality of the expression I originally posted, noting
> that I got it wrong, and it should be:
>
> (Address & Mask) == (NetAddress & Mask)
I get it now, I think I have it working. Thanks!
--
Sandman[.net]
Navigation:
[Reply to this message]
|