[PATCH] ofp-util: Clean up cookie handling.

Commit e72e793 (Add ability to restrict flow mods and flow stats
requests to cookies.) modified cookie handling.  Some of its behavior
was unintuitive and there was at least one bug (described below).
Commit f66b87d (DESIGN: Document uses for flow cookies.) attempted to
document a clean design for cookie handling.  This commit updates the
DESIGN document and brings the implementation in line with it.

In commit e72e793, the code that handled processing OpenFlow flow
modification requests set the cookie mask to exact-match.  This seems
reasonable for adding flows, but is not correct for matching, since
OpenFlow 1.0 doesn't support matching based on the cookie.  This commit
changes to cookie mask to fully wildcarded, which is the correct
behavior for modifications and deletions.  It doesn't cause any problems
for flow additions, since the mask is ignored for that operation.

Bug #9742

Reported-by: Luca Giraudo <lgiraudo@...>
Reported-by: Paul Ingram <paul@...>
Signed-off-by: Justin Pettit <jpettit@...>
---
 DESIGN                        |   31 +++++++++--
 include/openflow/nicira-ext.h |    6 +-
 lib/learn.c                   |    5 +-
 lib/ofp-parse.c               |   24 +++++++-
 lib/ofp-print.c               |    4 +-
 lib/ofp-util.c                |   36 +++++-------
 lib/ofp-util.h                |   24 +++++++-
 ofproto/ofproto-dpif.c        |    1 +
 ofproto/ofproto.c             |   31 +++++++---

[loss-report 1/4] poll-loop: More strictly rate-limit high CPU use.

120 messages per minute just isn't helpful.

Signed-off-by: Ben Pfaff <blp@...>
---
 lib/poll-loop.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/poll-loop.c b/lib/poll-loop.c
index ba6c3a1..516cf13 100644
--- a/lib/poll-loop.c
+++ b/lib/poll-loop.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2008, 2009, 2010, 2011 Nicira, Inc.
+ * Copyright (c) 2008, 2009, 2010, 2011, 2012 Nicira, Inc.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -157,7 +157,7 @@ poll_immediate_wake(const char *where)
 static void
 log_wakeup(const char *where, const struct pollfd *pollfd, int timeout)
 {
-    static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(120, 120);
+    static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(10, 10);
     enum vlog_level level;
     int cpu_usage;
     struct ds s;
--

-- 
1.7.2.5

Re: [ovs-discuss] kernel crash - openvswitch_mod

On 5/25/12 11:05 AM, Jesse Gross wrote:
> No, the caller will perform that check. rpl_skb_gso_segment() is a 
> compatibility replacement for skb_gso_segment() on older kernels. It 
> needs to return a pointer, not an error code. 

Hmmm... Yea, Im trying to do some debug on this.

I did the objdump, I think I found the line but I got stuck.

<0>EIP: [<f1f0cd29>] queue_userspace_packet+0x19/0x310 [openvswitch_mod] 
SS:ESP 0069:ef935af4

00001d10 <queue_userspace_packet>:
     1d10:   55                      push   %ebp
     1d11:   89 e5                   mov    %esp,%ebp
     1d13:   57                      push   %edi
     1d14:   56                      push   %esi
     1d15:   53                      push   %ebx
     1d16:   83 ec 30                sub    $0x30,%esp
     1d19:   89 45 d4                mov    %eax,0xffffffd4(%ebp)
     1d1c:   89 55 d0                mov    %edx,0xffffffd0(%ebp)
     1d1f:   89 4d cc                mov    %ecx,0xffffffcc(%ebp)
     1d22:   c7 45 d8 00 00 00 00    movl   $0x0,0xffffffd8(%ebp)
     1d29:   66 83 ba 8c 00 00 00    cmpw   $0x0,0x8c(%edx) <========= 
This line
     1d30:   00
     1d31:   0f 85 e7 01 00 00       jne    1f1e 
<queue_userspace_packet+0x20e>
     1d37:   8b 75 d0                mov    0xffffffd0(%ebp),%esi
     1d3a:   bb e5 ff ff ff          mov    $0xffffffe5,%ebx

Re: [ovs-discuss] kernel crash - openvswitch_mod

On 5/16/12 11:47 PM, Jesse Gross wrote:
> On Wed, May 16, 2012 at 7:45 PM, Ben Pfaff<blp@...>  wrote:
>> I would expect that only a host kernel bug (inside or outside the OVS
>> kernel module) would cause a host kernel panic and reboot, so I am
>> surprised at the idea that a bug in Xen tools (in the guest) would cause
>> a host reboot.  Is that really what you meant?
> Well, it's also a bug in the host kernel if it allows it but I think I
> remember a problem along these lines.
>
Changed to Dev Mailing...

We might have found the problem in the kernel function 
"skb_gso_segment"... Doing some test to check.

Anyways, looking at the OVS code I see this skb_gso_segment called in 
other places and handled in different ways:
./datapath/vport-netdev.c
./datapath/tunnel.c
./datapath/datapath.c

./datapath/linux/compat/netdevice.c

Doesn't rpl_skb_gso_segment need to check IS_ERR and return the PTR_ERR 
if it's true ?

Does it make sense ?

Something like this:

diff --git a/datapath/linux/compat/netdevice.c 

87% Off TESOL Teaching Certification Course

$79 for a 150-Hour Specialist TESOL/TEFL Course from Global Leadership  
College ($599 Value)
Please go to http://www.teacherbuys.com/nationwide/ 
for more details.

Certificate awarded upon completion of course

-150 hours of training from the comfort of your own home or office
-Information including TESOL/TEFL theory and practice
-100% comprehensive
-Easy to follow classes
-Get your certificate in as little as 8 weeks
-Learn and study at your own pace
-A solid base for a career
-Makes a great gift!

Expand your mind, and your future, with today's TeacherBuy!
If you prefer not to receive future Daily Deal emails, you can always  
send email to unsubscribe@... to unsubscribe.

Hello my dear,

[PATCH] memory: Memory leak when generating reports.

Signed-off-by: Ethan Jackson <ethan@...>
---

Mildly ironic.

---
 lib/memory.c |    1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/memory.c b/lib/memory.c
index 779860e..1babfe9 100644
--- a/lib/memory.c
+++ b/lib/memory.c
@@ -119,6 +119,7 @@ compose_report(const struct simap *usage, struct ds *s)
         ds_put_format(s, "%s:%u ", node->name, node->data);
     }
     ds_chomp(s, ' ');
+    free(nodes);
 }

 /* Logs the contents of 'usage', as a collection of name-count pairs.
--

-- 
1.7.10.2

XII Congreso Ciencia y Espiritu en Madrid

[PATCH] ovs-dpctl: Don't include include/openvswitch/tunnel.h

The inclusion include/openvswitch/tunnel.h does not seem to be needed any
more.

Signed-off-by: Simon Horman <horms@...>
---
 lib/odp-util.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/lib/odp-util.c b/lib/odp-util.c
index 13cdf8e..8693d3c 100644
--- a/lib/odp-util.c
+++ b/lib/odp-util.c
@@ -30,7 +30,6 @@
 #include "flow.h"
 #include "netlink.h"
 #include "ofpbuf.h"
-#include "openvswitch/tunnel.h"
 #include "packets.h"
 #include "simap.h"
 #include "timeval.h"
--

-- 
1.7.10.2.484.gcd07cc5

MSS Clamp in User-Space

Hi Jesse,

as part of our discussion on flow-based tunneling a few weeks ago
you mentioned that you would like to see an implementation of MSS Clamping
in user-space in order to eliminate the need for gre_err().

I made a start on implementing this, and I have a non-configurable
version working. Its pretty crude and just adjust the MSS on all TCP
SYN packets with no payload that are seen in handle_flow_miss_common().

That much seems to work. But I am unsure about a few things.
* Should it be configurable per-output port?
  If so that may be problematic if there are multiple outputs as
  the packet data is mangled.

* I am unsure how to automatically calculate the desired
  Clamp size. In particular, in for GRE it would be nice
  to allow it to be automatically clamped to the underlying MTU less
  the header size. But I'm unsure of a sane way to obtain the MTU.

Perhaps I am on the wrong track?

It does seem to me that this might be easier implemented in the datapath
as ovs_tnl_send() already has access to skb, the MTU and could leverage
the existing netfilter MMS clamp code.

Multiple Controllers in OpenFlow network?

Hi all,

I'm doing a project concerning OpenFlow and I have some questions  
about deploying OpenFlow network.

-  Can we deploy an OpenFlow network with multiple Controllers over  
multiple network Domains, each Controller manage a subset of OF  
Switches? (Can we deploy an OpenFlow network in which some OF Switches  
of this network connect to a Controller ( called Controller 1), some  
other Switches connect to Controller 2, some others connect to  
Controller 3, ... ?)

-  Can we deploy this OpenFlow network using OpenvSwitch and how to do  
that ?

Thanks in advance for your answer.

Best regards,

Thien.