From esj at cs.fiu.edu  Tue May  1 07:47:07 2012
From: esj at cs.fiu.edu (Eric S. Johnson)
Date: Tue, 01 May 2012 10:47:07 -0400
Subject: [Xorp-hackers] ospfv3 for IPv6 ospf - oddity
In-Reply-To: <20120427224212.6432F3680006@cheetah.cs.fiu.edu>
References: <20120427224212.6432F3680006@cheetah.cs.fiu.edu>
Message-ID: <20120501144707.C4B473680003@cheetah.cs.fiu.edu>


Answering myself. 

I kinda decided it is a bug. Everybody else doing V6 ospf seems
to send metrics with Intra-Area prefix LSA, and everybody else
seems to want to see them to give a cost to a link.

(everybody else == cisco and quagga :)

Here is a patch a worked up that seems to let xorp
"do the right thing". 

I left in the sending of /128 interface prefix along with 
the /64 network prefix becuase I have no "virtual links" nor 
experience with them. I kinda think that those /128's should be 
advertised ONLY on virtual links and certainly not on broadcast 
networks... But haven't worked that out yet.

Patch is against a git pull on 4-29-2012 but applies clean to 
1.8.5 also.

C++ is not my native language, and  I am not certain on some OSPF3 
things still, so I would appreciate some feedback.

Thoughts?

Anyone out there acutally trying to use xorp ospf6 besides me? :)

E






esj at cs.fiu.edu said:
> I started playing vith ospf for v6 recently. I have noticed an
> oddity. I am curious if I have something miss-configured  or I have
> found a bug.

> OSPFv3 for ipv6 has a couple new link-state adv types.  The one I am
> running into interoperability issues with is  the Intra-area prefix
> type. 

> It appears that xorp is sending these out with a cost of zero. No
> matter what the config file cost setting is. 

> This causes all the other ospf3 impelmentations it talks to (well,
> ok, cisco IOS and quagga are the only ones I have tested) to count
> the cost to that network as zero. 

> xorp sends the ole fashioned router lsa's with the appropriate
> costs.  But not the new fangled intra-area prefix costs.

> xorp also seems to send a /128 IA prefix LSA with the address of
> the interface also with metric of 0. This doesn't impact things as
> much as the network prefixs being 0, but seems odd too.

> I have verified with wireshark that it is sending the IA prefix
> lsa's with cost 0.

> Below are relavant snips of my configs.

> Xorp showing this behavior is 1.8.5 and git pulled on 2012-04-19
> Running on centos 5.6 (RHEL 5).  Linux test-router 2.6.18-164.6.1.el5
>  #1 SMP Tue Nov 3 16:18:27 EST 2009 i686 i686 i386 GNU/Linux

> Am I missing something? Or is this a bug?

> E 





diff -r -U5 /opt/router/src/xorp.ct.20120429.orig/xorp/ospf/area_router.cc ./area_router.cc
--- /opt/router/src/xorp.ct.20120429.orig/xorp/ospf/area_router.cc      2012-04-29 08:34:47.000000000 -0400
+++ ./area_router.cc    2012-04-29 09:01:10.000000000 -0400
@@ -3444,11 +3444,10 @@
                        if (prefix.get_nu_bit() /*|| prefix.get_la_bit()*/)
                            continue;
                        if (prefix.get_network().masked_addr().
                            is_linklocal_unicast())
                            continue;
-                       prefix.set_metric(0);
                        prefixes.push_back(prefix);
                    }
                }
            } else if (configured_virtual_link()) {
                uint32_t interface_id = pm.get_interface_id(i->first);
diff -r -U5 /opt/router/src/xorp.ct.20120429.orig/xorp/ospf/peer.cc ./peer.cc
--- /opt/router/src/xorp.ct.20120429.orig/xorp/ospf/peer.cc     2012-04-29 08:34:47.000000000 -0400
+++ ./peer.cc   2012-04-29 08:50:33.000000000 -0400
@@ -706,18 +706,18 @@
     return _areas[area]->get_attached_routers(routers);
 }
 
 template <typename A>
 bool
-PeerOut<A>::add_advertise_net(OspfTypes::AreaID area, A addr, uint32_t prefix)
+PeerOut<A>::add_advertise_net(OspfTypes::AreaID area, A addr, uint32_t prefix, uint16_t interface_cost)
 {
     if (0 == _areas.count(area)) {
        XLOG_ERROR("Unknown Area %s", pr_id(area).c_str());
        return false;
     }
 
-    return _areas[area]->add_advertise_net(addr, prefix);
+    return _areas[area]->add_advertise_net(addr, prefix, interface_cost);
 }
 
 template <typename A>
 bool
 PeerOut<A>::remove_all_nets(OspfTypes::AreaID area)
@@ -3220,37 +3220,39 @@
     return _hello_packet.get_interface_id();
 }
 
 template <>
 bool
-Peer<IPv4>::add_advertise_net(IPv4 /*addr*/, uint32_t /*prefix_length*/)
+Peer<IPv4>::add_advertise_net(IPv4 /*addr*/, uint32_t /*prefix_length*/, uint16_t /*cost*/ )
 {
     XLOG_FATAL("Only IPv6 not IPv4");
 
     return true;
 }
 
 template <>
 bool
-Peer<IPv6>::add_advertise_net(IPv6 addr, uint32_t prefix_length)
+Peer<IPv6>::add_advertise_net(IPv6 addr, uint32_t prefix_length, uint16_t interface_cost)
 {
     XLOG_ASSERT(OspfTypes::VirtualLink != get_linktype());
 
     LinkLsa *llsa = dynamic_cast<LinkLsa *>(_link_lsa.get());
     XLOG_ASSERT(llsa);
 
     if (addr.is_linklocal_unicast())
        return false;
 
-    IPv6Prefix prefix(_ospf.get_version());
+    IPv6Prefix prefix(_ospf.get_version(), true);
     prefix.set_network(IPNet<IPv6>(addr, prefix_length));
+    prefix.set_metric(interface_cost);
     llsa->get_prefixes().push_back(prefix);
 
     // Add a host route that can be used if necessary to advertise a
     // virtual link endpoint.
-    IPv6Prefix host_prefix(_ospf.get_version());
+    IPv6Prefix host_prefix(_ospf.get_version(), true);
     host_prefix.set_network(IPNet<IPv6>(addr, IPv6::ADDR_BITLEN));
+    host_prefix.set_metric(interface_cost);
     host_prefix.set_la_bit(true);
     llsa->get_prefixes().push_back(host_prefix);
 
     return true;
 }
diff -r -U5 /opt/router/src/xorp.ct.20120429.orig/xorp/ospf/peer.hh ./peer.hh
--- /opt/router/src/xorp.ct.20120429.orig/xorp/ospf/peer.hh     2012-04-29 08:34:47.000000000 -0400
+++ ./peer.hh   2012-04-29 08:43:09.000000000 -0400
@@ -327,11 +327,11 @@
                              list<RouterInfo>& routes);
 
     /**
      * Set a network to advertise OSPFv3 only.
      */
-    bool add_advertise_net(OspfTypes::AreaID area, A addr, uint32_t prefix);
+    bool add_advertise_net(OspfTypes::AreaID area, A addr, uint32_t prefix, uint16_t interface_cost);
 
     /**
      * Remove all the networks that are being advertised OSPFv3 only.
      */
     bool remove_all_nets(OspfTypes::AreaID area);
@@ -1110,11 +1110,11 @@
     uint32_t get_interface_id() const;
 
     /**
      * Set a network to advertise OSPFv3 only.
      */
-    bool add_advertise_net(A addr, uint32_t prefix);
+    bool add_advertise_net(A addr, uint32_t prefix, uint16_t interface_cost);
 
     /**
      * Remove all the networks that are being advertised OSPFv3 only.
      */
     bool remove_all_nets();
diff -r -U5 /opt/router/src/xorp.ct.20120429.orig/xorp/ospf/peer_manager.cc ./peer_manager.cc
--- /opt/router/src/xorp.ct.20120429.orig/xorp/ospf/peer_manager.cc     2012-04-29 08:34:47.000000000 -0400
+++ ./peer_manager.cc   2012-04-29 08:57:46.000000000 -0400
@@ -704,10 +704,11 @@
        XLOG_ERROR("Unknown PeerID %u", peerid);
        return false;
     }
 
     set<AddressInfo<A> >& info = _peers[peerid]->get_address_info(area);
+    uint16_t interface_cost = _peers[peerid]->get_interface_cost() ;
     
     // Unconditionally remove all the global addresses that are being
     // advertised.
     _peers[peerid]->remove_all_nets(area);
 
@@ -739,21 +740,21 @@
                                         interface_prefix_length)) {
                XLOG_ERROR("Unable to get prefix length for %s", cstring(*i));
                continue;
            }
            if (!_peers[peerid]->add_advertise_net(area, (*i),
-                                                  interface_prefix_length)) {
+                                                  interface_prefix_length,interface_cost)) {
                XLOG_WARNING("Unable to advertise %s in Link-LSA\n",
                             cstring(*i));
            }
        }
     } else {
        typename set<AddressInfo<A> >::iterator i;
        for (i = info.begin(); i != info.end(); i++) {
            if ((*i)._enabled) {
                if (!_peers[peerid]->add_advertise_net(area, (*i)._address,
-                                                     (*i)._prefix)) {
+                                                     (*i)._prefix,interface_cost)) {
                    XLOG_WARNING("Unable to advertise %s in Link-LSA\n",
                                 cstring((*i)._address));
                }
            }
        }






From greearb at candelatech.com  Tue May  1 09:07:51 2012
From: greearb at candelatech.com (Ben Greear)
Date: Tue, 01 May 2012 09:07:51 -0700
Subject: [Xorp-hackers] ospfv3 for IPv6 ospf - oddity
In-Reply-To: <20120501144707.C4B473680003@cheetah.cs.fiu.edu>
References: <20120427224212.6432F3680006@cheetah.cs.fiu.edu>
	<20120501144707.C4B473680003@cheetah.cs.fiu.edu>
Message-ID: <4FA00A57.1060607@candelatech.com>

On 05/01/2012 07:47 AM, Eric S. Johnson wrote:
>
> Answering myself.
>
> I kinda decided it is a bug. Everybody else doing V6 ospf seems
> to send metrics with Intra-Area prefix LSA, and everybody else
> seems to want to see them to give a cost to a link.
>
> (everybody else == cisco and quagga :)
>
> Here is a patch a worked up that seems to let xorp
> "do the right thing".
>
> I left in the sending of /128 interface prefix along with
> the /64 network prefix becuase I have no "virtual links" nor
> experience with them. I kinda think that those /128's should be
> advertised ONLY on virtual links and certainly not on broadcast
> networks... But haven't worked that out yet.
>
> Patch is against a git pull on 4-29-2012 but applies clean to
> 1.8.5 also.
>
> C++ is not my native language, and  I am not certain on some OSPF3
> things still, so I would appreciate some feedback.
>
> Thoughts?
>
> Anyone out there acutally trying to use xorp ospf6 besides me? :)

We do a bit of testing with it, but not extensively, and only
xorp to xorp so far.

I'm about to do a release of our LANforge product, but as soon
as we kick that out the door I'll add apply this patch unless
someone suggests otherwise.  It looks sane enough to me.

Thanks,
Ben

-- 
Ben Greear <greearb at candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


From noreply at github.com  Tue May  1 14:13:40 2012
From: noreply at github.com (GitHub)
Date: Tue, 01 May 2012 14:13:40 -0700
Subject: [Xorp-hackers] [greearb/xorp.ct] 0e2a1f: rtrmgr: Fix typo in log
	message.
Message-ID: <4fa052048fda2_33fc16acaec134166@sh2.rs.github.com.mail>

  Branch: refs/heads/master
  Home:   https://github.com/greearb/xorp.ct
  Commit: 0e2a1ff8b6f3249addf6e306ba5a6f8b2fd097d2
      https://github.com/greearb/xorp.ct/commit/0e2a1ff8b6f3249addf6e306ba5a6f8b2fd097d2
  Author: Ben Greear <greearb at candelatech.com>
  Date:   2012-05-01 (Tue, 01 May 2012)

  Changed paths:
    M xorp/rtrmgr/userdb.cc

  Log Message:
  -----------
  rtrmgr: Fix typo in log message.


  Commit: 6bb7a27cab5376df9ca5066da73a6987b8a8e2a9
      https://github.com/greearb/xorp.ct/commit/6bb7a27cab5376df9ca5066da73a6987b8a8e2a9
  Author: Ben Greear <greearb at candelatech.com>
  Date:   2012-05-01 (Tue, 01 May 2012)

  Changed paths:
    M xorp/Makefile.deb

  Log Message:
  -----------
  Merge branch 'master' of github.com:greearb/xorp.ct


Compare: https://github.com/greearb/xorp.ct/compare/3939675...6bb7a27

From dambaa.bb at gmail.com  Fri May  4 08:12:43 2012
From: dambaa.bb at gmail.com (Dambaa)
Date: Fri, 4 May 2012 08:12:43 -0700
Subject: [Xorp-hackers] xorp developing
Message-ID: <CACY8cSnLP9WKdwMM0sALCmATg5LdARqwUpZvkL6XL9z5SM4GNQ@mail.gmail.com>

Hello from Mongolia
I liked xorp. I have a question?
Is it possible to add dynamic route in the xorp? How to do it? is it
difficult?
Could you give me some advice, please?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ICSI.Berkeley.EDU/pipermail/xorp-hackers/attachments/20120504/761e3bbf/attachment.html 

From terri at unm.edu  Mon May 14 15:57:05 2012
From: terri at unm.edu (Terri Oda)
Date: Mon, 14 May 2012 16:57:05 -0600
Subject: [Xorp-hackers] Reproducible xorp bugs?
Message-ID: <4FB18DC1.8040607@unm.edu>

I'm doing some research on routers and wanted to try an experiment that 
requires a reproducible bug in xorp.  I'm working my way through the bug 
queue, but I figure I should ask as well: does anyone have a bug they 
recommend I try?  I need something that can be reproduced in a fully 
automated, repeatable way, so anything that requires me to unplug cables 
is out, but surely there's some other known bugs that I could use?

  Terri


From noreply at github.com  Mon May 14 16:06:38 2012
From: noreply at github.com (GitHub)
Date: Mon, 14 May 2012 16:06:38 -0700
Subject: [Xorp-hackers] [greearb/xorp.ct]
Message-ID: <4fb18ffe9536b_132187aaf4528545@sh3.rs.github.com.mail>

  Branch: refs/tags/lf-5.2.4
  Home:   https://github.com/greearb/xorp.ct


From greearb at candelatech.com  Mon May 14 17:12:58 2012
From: greearb at candelatech.com (Ben Greear)
Date: Mon, 14 May 2012 17:12:58 -0700
Subject: [Xorp-hackers] xorp developing
In-Reply-To: <CACY8cSnLP9WKdwMM0sALCmATg5LdARqwUpZvkL6XL9z5SM4GNQ@mail.gmail.com>
References: <CACY8cSnLP9WKdwMM0sALCmATg5LdARqwUpZvkL6XL9z5SM4GNQ@mail.gmail.com>
Message-ID: <4FB19F8A.8050808@candelatech.com>

On 05/04/2012 08:12 AM, Dambaa wrote:
> Hello from Mongolia
> I liked xorp. I have a question?
> Is it possible to add dynamic route in the xorp? How to do it? is it difficult?
> Could you give me some advice, please?

You should be able to use xorpsh to add one I think.

Thanks,
Ben

>
>
> _______________________________________________
> Xorp-hackers mailing list
> Xorp-hackers at icir.org
> http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-hackers


-- 
Ben Greear <greearb at candelatech.com>
Candela Technologies Inc  http://www.candelatech.com


From noreply at github.com  Mon May 14 17:29:52 2012
From: noreply at github.com (GitHub)
Date: Mon, 14 May 2012 17:29:52 -0700
Subject: [Xorp-hackers] [greearb/xorp.ct] 639cd2: Fix OSPFv6 route cost.
Message-ID: <4fb1a380c0b25_7eb41282ae81614e2@sh2.rs.github.com.mail>

  Branch: refs/heads/master
  Home:   https://github.com/greearb/xorp.ct
  Commit: 639cd2bdcb4f80af3e5ce82e36771ba84b4a3d69
      https://github.com/greearb/xorp.ct/commit/639cd2bdcb4f80af3e5ce82e36771ba84b4a3d69
  Author: Ben Greear <greearb at candelatech.com>
  Date:   2012-05-14 (Mon, 14 May 2012)

  Changed paths:
    M xorp/ospf/area_router.cc
    M xorp/ospf/peer.cc
    M xorp/ospf/peer.hh
    M xorp/ospf/peer_manager.cc

  Log Message:
  -----------
  Fix OSPFv6 route cost.

Original patch by: Eric S. Johnson <esj at cs.fiu.edu>
I had to apply it by hand due to conflicts, hopefully
didn't miss anything.

>From the email description:

I kinda decided it is a bug. Everybody else doing V6 ospf seems
to send metrics with Intra-Area prefix LSA, and everybody else
seems to want to see them to give a cost to a link.

(everybody else == cisco and quagga

Here is a patch a worked up that seems to let xorp
"do the right thing".

I left in the sending of /128 interface prefix along with
the /64 network prefix becuase I have no "virtual links" nor
experience with them. I kinda think that those /128's should be
advertised ONLY on virtual links and certainly not on broadcast
networks... But haven't worked that out yet.

Patch is against a git pull on 4-29-2012 but applies clean to
1.8.5 also.

C++ is not my native language, and  I am not certain on some OSPF3
things still, so I would appreciate some feedback.

Thoughts?

Anyone out there acutally trying to use xorp ospf6 besides me?

E

esj at cs.fiu.edu said:
> I started playing vith ospf for v6 recently. I have noticed an
> oddity. I am curious if I have something miss-configured  or I have
> found a bug.
> OSPFv3 for ipv6 has a couple new link-state adv types.  The one I am
> running into interoperability issues with is  the Intra-area prefix
> type.
> It appears that xorp is sending these out with a cost of zero. No
> matter what the config file cost setting is.
> This causes all the other ospf3 impelmentations it talks to (well,
> ok, cisco IOS and quagga are the only ones I have tested) to count
> the cost to that network as zero.
> xorp sends the ole fashioned router lsa's with the appropriate
> costs.  But not the new fangled intra-area prefix costs.
> xorp also seems to send a /128 IA prefix LSA with the address of
> the interface also with metric of 0. This doesn't impact things as
> much as the network prefixs being 0, but seems odd too.
> I have verified with wireshark that it is sending the IA prefix
> lsa's with cost 0.

Signed-off-by: Ben Greear <greearb at candelatech.com>




From jiangxin.hu at crc.gc.ca  Tue May 15 07:09:56 2012
From: jiangxin.hu at crc.gc.ca (Jiangxin Hu)
Date: Tue, 15 May 2012 10:09:56 -0400
Subject: [Xorp-hackers] XORP to Linux kernel
Message-ID: <20120515141033.584066B274C@mailhub.crc.ca>

I am going to do some experiements about XORP. Any one can tell me where is
the part (in source code) which XORP talks to kernel. 
 
Thanks,
Jiangxin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ICSI.Berkeley.EDU/pipermail/xorp-hackers/attachments/20120515/d273f260/attachment.html 

From igorm at etf.rs  Tue May 15 07:44:22 2012
From: igorm at etf.rs (=?ISO-8859-2?Q?Igor_Maravi=E6?=)
Date: Tue, 15 May 2012 16:44:22 +0200
Subject: [Xorp-hackers] XORP to Linux kernel
In-Reply-To: <20120515141033.584066B274C@mailhub.crc.ca>
References: <20120515141033.584066B274C@mailhub.crc.ca>
Message-ID: <CAFdo_mVk1+hfJqo78b16pOAXTBtD8ifa9+F6XLO39+pJoDwSLQ@mail.gmail.com>

It's located in xorp/fea folder.
BR
Igor

2012/5/15 Jiangxin Hu <jiangxin.hu at crc.gc.ca>

> **
> I am going to do some experiements about XORP. Any one can tell me where
> is the part (in source code) which XORP talks to kernel.
>
> Thanks,
> Jiangxin
>
> _______________________________________________
> Xorp-hackers mailing list
> Xorp-hackers at icir.org
> http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-hackers
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ICSI.Berkeley.EDU/pipermail/xorp-hackers/attachments/20120515/feb83ae1/attachment.html 

From noreply at github.com  Thu May 17 16:49:28 2012
From: noreply at github.com (GitHub)
Date: Thu, 17 May 2012 16:49:28 -0700
Subject: [Xorp-hackers] [greearb/xorp.ct] 135910: Update LF version.
Message-ID: <4fb58e88296f1_6c9e1449ae860821@sh2.rs.github.com.mail>

  Branch: refs/heads/master
  Home:   https://github.com/greearb/xorp.ct
  Commit: 13591030197b90f5c838a74fade9de6ccbeab108
      https://github.com/greearb/xorp.ct/commit/13591030197b90f5c838a74fade9de6ccbeab108
  Author: Ben Greear <greearb at candelatech.com>
  Date:   2012-05-17 (Thu, 17 May 2012)

  Changed paths:
    M xorp/Makefile.deb

  Log Message:
  -----------
  Update LF version.