Eugene,
here is my take on it. Let's get ICMP fragmented and see what fields we
should match.

R4(config-cmap)#do sh run class-map
Building configuration...
Current configuration : 222 bytes
!
class-map type stack match-all CM_FPM_IP_ICMP
 match field IP protocol eq 1 next ICMP
class-map type access-control match-any CM_FPM_ICMP_FRAGMENTS
 match field IP flags eq 1
 match field IP fragment-offset gt 0
!
end

R4(config-cmap)#do sh run policy-map
Building configuration...
Current configuration : 206 bytes
!
policy-map type access-control PM_FPM_ICMP_FRAGMENTS
 class CM_FPM_ICMP_FRAGMENTS
   log
policy-map type access-control PM_FPM_IP_ICMP
 class CM_FPM_IP_ICMP
  service-policy PM_FPM_ICMP_FRAGMENTS
!
end

R3 fa01 - 180.6.34.0/24 -- fa0/0 R4

PM_FPM_IP_ICMP is applied inbound on Fa 0/0 of R4.

First, as you can see, there is no IP length field in my class-map, because
it is always less or equal than 1500.

The most generic scenario is that we have 3 type of fragments and those IP
header fields should be set as per below:
- initial fragment -> IP Length = 1500 MF flag = 1 Fragment Offset = 0
- non-initial fragment -> IP Length = 1500 MF flag = 1 Fragment Offset > 0
- last fragment -> IP Length < 1500 MF flag = 0 Fragment Offset > 0
In some solution they match on IP length too, which IMO achieves nothing.

Am I missing sonething?

Cheers
A.





On 4 June 2012 15:37, Eugene Pefti <[email protected]> wrote:

>  Sorry, didn't mean to send it yet. Starting it all over:
>
>  If you don't mind, guys, I'd start some sort of compilation of FPM
> related knowledge here, specifically that we need to know and which is not
> easily available as a reference in Cisco docs.
>
> 1. Protocol numbers. So far we have to remember only three numbers
> (mostly) - 1 (ICMP), 6 (TCP), 17 (UDP)
> An accompanying question, is it good to specify them in decimal when
> defining the stack type class-map, i.e. I want to match for UDP protocol in
> the IP header
>    class-map type stack IP-UDP-CLASS
>      match field IP protocol eq 17 next UDP
>
> And another one, is mask a loose parameter that can always be left off
> here ?
>
> 2. I found that I can't retain in my head what goes first in the Flags
> field of IP header, DF or MF. Is there any reference in Cisco docs that
> could be looked up during the exam?
> Let's take Mike's example with fragmented packets. King's solution is as
> follows (I quote it):
>
>  You should match packets with FO> 0 and Flags = 1.
>
> For the first packet F0=0 and Flags=1
> For other packets F0>0
> For the last packet FO>0 and flags with be 1.
>
> Hence the following is the solution.
>
> rtr(config)# class-map type access-control match-any fragudp
> rtr(config-cmap)# match field ip flags eq 1 mask 6
> rtr(config-cmap)# match field ip fragment-offset gt 0
>
>    2.1 When you say "Flags=1" in the first line what flag do you mean?
> I'd assume MF (More Fragments).
>    2.2 When you mention flags in the third line as should be set to 1
> again, what flags did you mean, MF or DF? If MF, why would IP stack care
> about MF in the last packet?
>    2.3 You say that we have to match for packets with FO>0 and Flags=1,
> why does the class-map have "match-any" option?
>    2.4 If I were to match on DF flag then I'd need to set mask to 5 ? Are
> there any conditions that I'd need to care and be prepared to match for the
> DF flag?
>
> Enough for now ;)
>
> Eugene
>
> From: Eugene Pefti <[email protected]>
> Date: Sunday, June 3, 2012 10:09 PM
> To: Mike Rojas <[email protected]>, "[email protected]" <
> [email protected]>, "[email protected]" <
> [email protected]>
>
> Subject: Re: [OSL | CCIE_Security] FPM ICMP large Packets
>
>  I feel sorry for you, Mike, but I know that you like all others here
> will definitely do it next time.
> I start feeling that FPM is the most intimidating subject on the exam ;)
>
> If you don't mind, guys, I'd start some kind of compilation of FPM related
> knowledge here, specifically that we need to know and which is not easily
> available as a reference in Cisco docs.
>
>
>    1. Protocol numbers. So far we have to remember only three numbers
>    (mostly) - 1 (ICMP), 6 (TCP), 17 (UDP)
>
>
> From: Mike Rojas <[email protected]>
> Date: Sunday, June 3, 2012 2:25 PM
> To: "[email protected]" <[email protected]>, "
> [email protected]" <[email protected]>
> Subject: Re: [OSL | CCIE_Security] FPM ICMP large Packets
>
>  Oh no Kings, I failed it because I suck at it... a got an FPM question
> where you had to do something about the ICMP packet size... I was looking a
> question more like finding something inside of the payload... some you win
> some you loose...
>
> Cheers!
>
> Mike
>
>  ------------------------------
> Date: Sun, 3 Jun 2012 10:56:59 +0530
> Subject: Re: [OSL | CCIE_Security] FPM ICMP large Packets
> From: [email protected]
> To: [email protected]
>
> Mike, did you fail in the CCIE lab? And is it due to the wrong solution of
> FPM?
>
> With regards
> Kings
>
> On Sun, Jun 3, 2012 at 3:08 AM, Mike Rojas <[email protected]> wrote:
>
>  I just want to recall one of the Replies from Kingsley... BTW I failed
> the test....
>
> http://onlinestudylist.com/archives/ccie_security/2012-February/029078.html
>
> Mike
>
> _______________________________________________
> For more information regarding industry leading CCIE Lab training, please
> visit www.ipexpert.com
>
> Are you a CCNP or CCIE and looking for a job? Check out
> www.PlatinumPlacement.com <http://www.platinumplacement.com/>
>
>
>
> _______________________________________________
> For more information regarding industry leading CCIE Lab training, please
> visit www.ipexpert.com
>
> Are you a CCNP or CCIE and looking for a job? Check out
> www.PlatinumPlacement.com <http://www.platinumplacement.com/>
>
_______________________________________________
For more information regarding industry leading CCIE Lab training, please visit 
www.ipexpert.com

Are you a CCNP or CCIE and looking for a job? Check out 
www.PlatinumPlacement.com

Reply via email to