[Click] Basic question about IP header fields.
Josh Butler
tuna_armadillo at hotmail.com
Wed Nov 7 19:44:10 EST 2007
Hi all,
Ok, I think I've got my head around using the hashcode function, but now
I'm just wondering how to get hold of the source and destination port
numbers for the packet. This is the core of my element as it stands,
just a modified version of the Tee element:
void
MyTee::push(int, Packet *p)
{
const click_ip *iph = p->ip_header(); // Get pointer to IP Header
if (!iph) // Make sure it's a valid header
{
p->kill(); // Kill if invalid to avoid memory leaks
return;
}
SrcIP = iph->ip_src;
DstIP = iph->ip_dst;
SrcPort = iph->ip_len; //Having to use length as don't know how to get
DstPort = iph->ip_len; //Having to use length as don't know how to get
IPFlowID *ipfid = NULL;
//Create the container for the instances of FlowIDs.
ipfid = new IPFlowID(SrcIP, SrcPort, DstIP, DstPort);
//Create an instance of an IPFlowID containing data from the packet.
NewFlowIDVar = hashcode(*ipfid);
//Pass the hashcode function the pointer to instance of the IPFlowID.
if (NewFlowIDVar != PrevFlowIDVar)
{
output(0).push(p);
//If flow is brand new, push through port 0.
PrevFlowIDVar = NewFlowIDVar;
//Record brand new flow for comparison next time.
}
else
{
output(1).push(p);
//If flow is same as previously recorded flow, push through port 1.
}
delete ipfid;
}
Seems to work in a simple configuration when passed identical
InfiniteSource data, pushing the first packet through port 0 and all
subsequent packets through port 1.
Is this code actually doing what I'm intending it to - creating a hashed
ID for a flow, recording it and then pushing any subsequent packets with
the same hashed ID out of the other port?
Or is it a load of rubbish? - I'm not too sure what I'm doing, so that
wouldn't surprise me.
And how do I get the Source and Dest Port numbers from the packet - I've
not stripped the transport header. Do I need to create a pointer to the
transport header, like I do for the IP header?
Your help is massively appreciated.
Thanks,
Josh
On Thu, 2007-10-25 at 23:19 +0200, Beyers Cronje wrote:
> Hi,
>
> See [click]/include/click/ipflowid.hh , the IPFlowID class has a
> hashcode() function. Then use Click's HashMap
> [click]/include/click/hashmap.hh to keep state of the flows. Grep for
> HashMap and/or IPFlowID under [click]/elements to find examples on how
> it's used.
>
> Beyers
>
>
> On 10/25/07, Josh Butler <J.R.Butler-05 at student.lboro.ac.uk> wrote:
> Hi,
>
> Thanks again, another problem solved. But another one begins.
> Those elements I've created are dead simple, I want to do
> something a bit more complicated. I'm planning on creating an
> element that creates a flow id for packets based on their
> source and destination addresses and ports. Apparently I need
> to take each of those fields and do an XOR (^) on them to get
> a hash value that I can use as a flow ID - does that sound
> right? My first issue is that I can't manipulate the IP
> addresses as I would the other header fields. So doing this is
> not as straight forward as I though it would be. How can I do
> this?
>
> I then need to store this flow ID in an array, so that any
> subsequent packets with the same ID are identifiable. But I
> can't really approach this until I know how to create the hash
> value first, so any help on that would be much appreciated.
>
> Regards
> Josh
> ----- Original Message -----
> From: Beyers Cronje
> To: tuna_armadillo at hotmail.com
> Cc: click at amsterdam.lcs.mit.edu
> Sent: Tuesday, October 23, 2007 3:50 AM
> Subject: Re: [Click] Basic question about IP header
> fields.
>
>
> Hi Josh,
>
> The ip_len field is in network byte order, assuming
> you are running on an Intel platform you need to
> convert it to host byte order using ntohs(iph->ip_len)
>
> Google on these terms: ntohs, ntohl, endianness,
> network byte order, host byte order
>
> See http://en.wikipedia.org/wiki/Endianness for an
> overview.
>
> Beyers
>
>
> On 10/23/07, Nikki <tuna_armadillo at hotmail.com>
> wrote:
> Hi guys,
>
> Firstly, many many thanks for the help, its
> been great - I've only just
> got round to putting it into practice now I've
> got most of the theory
> out of the way, but the code you gave me has
> worked a treat. I think I
> need to read up on pointers a bit more to
> fully understand it, but I
> think I see what its doing.
>
> So anyway, on to my new question. I tried
> messing around with the code,
> in my new element and it now looks like this:
>
> const click_ip *iph = p->ip_header(); // Get
> pointer to IP Header
>
> if (!iph){ // Make sure it's a valid
> header
> p->kill();
> return;
> }
> else if (iph->ip_len <= 1000){ //
> Check Total Length field
> output(0).push(p);
> }
> else{
> output(1).push(p);
> }
>
> I want to split packets through different
> outputs depending on packet
> size, but the code above just pushes
> everything through output 1. I'm
> using FromDump in my config script, and am
> using a dump with packets of
> varying sizes, from 40 to 1420 bytes - but
> they all come out of output
> 1, not output 0. Where am I going wrong?
>
> Thanks,
> Josh
>
>
> On Tue, 2007-09-11 at 01:02 +0200, Beyers
> Cronje wrote:
> > Josh,
> >
> > And so I inadvertently demonstrated one of
> the classic beginner
> > mistakes to watch out for :) Always either
> push/pull, kill or store
> > the packet for later use...
> >
> >
> > On 9/11/07, Eddie Kohler
> <kohler at cs.ucla.edu> wrote:
> > Thanks, Beyers -- but I'd either
> "assert(iph)" or "if (!iph) {
> > p->kill(); return; }", to avoid
> leaking memory.
> > E
> >
> >
> > Beyers Cronje wrote:
> > > Hi Josh,
> > >
> > > You need to get a pointer to the
> IP Header. A simple
> > example:
> > >
> > > const click_ip *iph =
> p->ip_header(); // Get pointer to IP
> > Header
> > > if (!iph) // Make sure it's a
> valid header
> > > return;
> > >
> > > if (iph->ip_tos <= n) // Check
> TOS field
> > > output(0).push(p);
> > > else
> > > output(1).push(p)
> > >
> > > The above example expects IP
> packets. So you'll have to
> > filter on IP packets
> > > and also set the IP header
> annotations i.e.:
> > >
> > > FromDevice(eth0) ->
> Classifier(12/0800) -> MarkIPHeader(14)
> > -> YourElement;
> > >
> > > Classifier to filter IP packets,
> and MarkIPHeader to set
> > annotations. You
> > > can also use CheckIPHeader instead
> of MarkIPHeader.
> > >
> > > Hope this helps.
> > >
> > > Beyers
> > >
> > > On 9/10/07, Josh Butler
> <tuna_armadillo at hotmail.com> wrote:
> > >> Hi,
> > >>
> > >> This is my first question and
> probably won't be my last.
> > I've been doing a
> > >> project involving Click for a
> while, but owing to my very
> > basic programming
> > >> skills have struggled quite a
> lot, though I've enjoyed
> > messing about with
> > >> click in the process - I've been
> writing a bit of an
> > idiot's guide as I've
> > >> gone along!
> > >>
> > >> I've got my head round creating
> the router configuration
> > scripts by
> > >> combining elements and am now
> looking to go inside the
> > elements and make
> > >> some changes. I've chosen the tee
> element as one of the
> > easier elements to
> > >> look at and modify, and have a
> good understanding of what
> > it does in push
> > >> mode.
> > >>
> > >> I now want to modify this element
> in a very basic way,
> > using a simple if
> > >> statement if possible so that I
> can push packets through
> > either the
> > >> element's first or second output
> port depending on some
> > criteria such as the
> > >> type of service field within the
> IP header. Here's what I
> > mean in
> > >> pseudocode, if that's the correct
> term;
> > >>
> > >> if (type_of_service_field <= n){
> > >> output(0).push(p);
> > >> }
> > >> else {
> > >> output(1).push(p);
> > >> }
> > >>
> > >> I've looked through all the
> material on the click site and
> > can't find a
> > >> simple answer to the question of
> how to access the
> > different fields within
> > >> the IP header. Maybe I'm being
> stupid, but a bit of help
> > would be awesome.
> > >>
> > >> Thanks,
> > >> Josh
> > >>
> > >>
> >
> _________________________________________________________________
> > >> 100's of Music vouchers to be won
> with MSN Music
> > >> https://www.musicmashup.co.uk
> > >>
> _______________________________________________
> > >> click mailing list
> > >> click at amsterdam.lcs.mit.edu
> > >>
> https://amsterdam.lcs.mit.edu/mailman/listinfo/click
> > >>
> > >
> _______________________________________________
> > > click mailing list
> > > click at amsterdam.lcs.mit.edu
> > >
> https://amsterdam.lcs.mit.edu/mailman/listinfo/click
> >
>
>
>
>
More information about the click
mailing list