Re: [mpls-linux-devel] About mpls-packet to ip-packet process

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Thank you first.
I was follow the example "IPv4 over MPLS: two LER example for mpls-linux-1.95x".
And had follow config.
________________________________________________________________________________________________
                                                    172.16.0.0/24
10.0.0.1/32                                             |                                  10.0.0.2/32
dummy0                                                |                                    dummy0
    |                                                        |                                         |
-----------eth1        eth1------eth2            v       eth2------eth3    eth3--------
| HOST A |------------| LER1 |--------------------------| LER2 |---------| HOST B |
-----------.1    ^       2------.2                               3------.3   ^     4-----------
                     |                                                                  |
                     |                                                                  |
               192.168.0.0/24                                            192.168.1.0/24
__________________________________________________________________________________________________
HOST A
------
ip route add 10.0.0.2/32 via 192.168.0.2 src 10.0.0.1

LER1
-----
mpls nhlfe add key 0 instructions push gen 1000 nexthop eth2 ipv4 172.16.0.3
(returns key 0x2)
ip route add 10.0.0.2/32 via 172.16.0.3 mpls 0x2
echo 1> /proc/sys/net/ipv4/ip_forward
mpls labelspace set dev eth2 labelspace 0
mpls ilm add label gen 2000 labelspace 0
mpls nhlfe add key 0 instructions nexthop eth1 ipv4 192.168.0.1
(returns key 0x3)
mpls xc add ilm_label gen 2000 ilm_labelspace 0 nhlfe_key 0x3

LER2
-----
mpls labelspace set dev eth2 labelspace 0
mpls ilm add label gen 1000 labelspace 0                               <-------------------Had set this
mpls nhlfe add key 0 instructions nexthop eth3 ipv4 192.168.1.4
(returns key 0x2)
mpls xc add ilm_label gen 1000 ilm_labelspace 0 nhlfe_key 0x2
mpls nhlfe add key 0 instructions push gen 2000 nexthop eth2 ipv4 172.16.0.2
(returns key 0x3)
ip route add 10.0.0.1/32 via 172.16.0.2 mpls 0x3
echo 1> /proc/sys/net/ipv4/ip_forward

HOST B
------
ip route add 10.0.0.1/32 via 192.168.1.3 src 10.0.0.2
_________________________________________________________________________________________________________________
BUT, the packet from HOST A to HOST B does not through the IP stack on LER2.

Is't my config error?

And another problem for quagga, that the patch file quagga-mpls.diff can not patch to quagga-0.99.9 (latest unstable release) or quagga-0.98.6 (latest stable release).
Do you have the new patch file?

Thank you for help.
----------------------------------------
> Date: Thu, 1 May 2008 12:35:52 -0500
> From: jl...@mi...
> To: @mindspring.com
> CC: mpl...@li...
> Subject: Re: [mpls-linux-devel] About mpls-packet to ip-packet process
> 
> On Thu, May 01, 2008 at 03:55:33PM +0000, =?GB2312?B?wfXA2iA8b3JwaGVuX2xlaWxpdUBtc24uY29tPg==?= wrote:
>> 
>> I'm sorry for my poor english.
>> Just like the follow instance.
>> 
>> < Client Router>-------| LER |--------{ MPLS NETWORK }
>> 
>> The packet from Client got into MPLS NETWORK, will through 'ip_rcv, ip_rcv_finish (ip_route_input ip_route_input_slow ip_mkroute_input __mkroute_input rt_set_nexthop mpls_set_nexthop mpl_set_nexthop2) ip_forward ip_forward_finish ip_output ip_finish_output ip_finish_output2 mpls_output mpls_output_shim mpls_output2 mpls_send'. Am I right?
>> And the packet from MPLS NETWORK to client, will through 'mpls_skb_recv mpls_input mpls_switch mpls_output2 mpls_send neighbour->output'.
>> If I want to the packet out from mpls_send got into ip_rcv ......, how to config it?
>> "mpls nhlfe add key 0 instruction pop nexthop ethX ipv4 (Client Router's IP)"? Right?
> 
> This is a PHP config, IE no IPv4 route lookup is being done on the LER
> when the packet goes from MPLS to IPv4.
> 
> For the packet to go through the IPv4 stack you will want your
> ILM to be something like:
> 
>     mpls ilm add label gen 200 labelspace 0
> 
> That is it.  The default instruction on a ILM are "pop peek".
> Which results in mpls_input() returning MPLS_RESULT_DLV which
> ends up with the packet being handle via the IPv4 protocol driver's
> (net/ipv4/mpls4.c) local_deliver function.
> 
>> ----------------------------------------
>>> Date: Thu, 1 May 2008 06:45:41 -0500
>>> From: jl...@mi...
>>> To: @mindspring.com
>>> CC: mpl...@li...
>>> Subject: Re: [mpls-linux-devel] About mpls-packet to ip-packet process
>>> 
>>> Just to make sure I understand what you're trying to do:
>>> 
>>> You want to get hooked into the PREROUTE netfilter chain for packets
>>> coming to a LER from MPLS.  To do this the normal 'pop peek' instructions
>>> should work just.  If you want to 'force' the trasistion from MPLS to
>>> IPv4 for packets coming to a LER from MPLS you can use the 'deliver'
>>> instruction on the associated ILM.
>>> 
>>> On Thu, May 01, 2008 at 10:44:42AM +0000, =?GB2312?B?wfXA2iA8b3JwaGVuX2xlaWxpdUBtc24uY29tPg==?= wrote:
>>>> 
>>>> I was follow the example of "IPv4 over MPLS: two LER example for mpls-linux-1.95x".
>>>> Do you means I have to change the "mpls nhlfe add key 0 instructions nexthop eth3 ipv4 192.168.1.4
>>>> (returns key 0x2)" to "mpls nhlfe add key 0 instructions deliver nexthop eth3 ipv4 192.168.1.4
>>>> (returns key 0x2)" ?
>>>> 
>>>> ----------------------------------------
>>>>> Date: Wed, 30 Apr 2008 21:16:22 -0500
>>>>> From: jl...@mi...
>>>>> To: @mindspring.com
>>>>> CC: mpl...@li...
>>>>> Subject: Re: [mpls-linux-devel] About mpls-packet to ip-packet process
>>>>>
>>>>> Hmm. I think you are looking at the PHP code. Take a look at
>>>>> the 'deliver' instruction. It will show you that packets destined from
>>>>> IPv4 may be configured to go through 'ip_rcv' which will result in
>>>>> the packet travesing all of the IPv4 netfilter hooks.
>>>>>
>>>>> On Wed, Apr 30, 2008 at 03:56:51AM +0000, =?GB2312?B?wfXA2iA8b3JwaGVuX2xlaWxpdUBtc24uY29tPg==?= wrote:
>>>>>>
>>>>>> Hi all:
>>>>>> I found that the packet which from ip stack to mpls stack in the ip_finish_output2 had passed POSTROUTING hook in netfilter.
>>>>>> BUT the packet from mpls stack to ip stack just use neighbour->output, which NOT hook in PREROUTING in netfilter.
>>>>>>
>>>>>> In this way, if I do SNAT for ip packet and send into mpls stack, will NOT do Un-SNAT for the reply packet.
>>>>>>
>>>>>> Is my understand right? How to solved it?
>>>>>> _________________________________________________________________
>>>>>> 用手机MSN聊天写邮件看空间，无限沟通，分享精彩！
>>>>>> http://mobile.msn.com.cn/
>>>>>>
>>>>>
>>>>>> -------------------------------------------------------------------------
>>>>>> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
>>>>>> Don't miss this year's exciting event. There's still time to save $100.
>>>>>> Use priority code J8TL2D2.
>>>>>> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
>>>>>> _______________________________________________
>>>>>> mpls-linux-devel mailing list
>>>>>> mpl...@li...
>>>>>> https://lists.sourceforge.net/lists/listinfo/mpls-linux-devel
>>>>>
>>>>>
>>>>> --
>>>>> James R. Leu
>>>>> jl...@mi...
>>>> 
>>>> _________________________________________________________________
>>>> MSN 中文网，最新时尚生活资讯，白领聚集门户。
>>>> http://cn.msn.com
>>> 
>>> -- 
>>> James R. Leu
>>> jl...@mi...
>> 
>> _________________________________________________________________
>> MSN 中文网，最新时尚生活资讯，白领聚集门户。
>> http://cn.msn.com
> 
> -- 
> James R. Leu
> jl...@mi...

_________________________________________________________________
MSN 中文网，最新时尚生活资讯，白领聚集门户。
http://cn.msn.com