Commit e503cc199 (ofproto: Optimise OpenFlow flow expiry) optimized
OpenFlow flow expiration by putting expirable flows on a list. However,
the list is only configured at rule creation time. If the rule is
created without a timeout, but is later set by the fin_timeout action,
it will never expire
Woops! Just sent a review.
On Wed, Jul 17, 2013 at 2:44 PM, Ben Pfaff wrote:
> On Tue, Jul 16, 2013 at 10:16 PM, Joe Stringer wrote:
>> Commit fab52e16f7d8ccdbad53c5037cd04d5863a2cb4a introduces a comparison
>> between mismatching types, shown by the following compiler warnings:-
>>
>> lib/bfd.c
Looks good.
On Wed, Jul 17, 2013 at 2:06 AM, Ben Pfaff wrote:
> This is not a real bug, since htonl(0) and htonll(0) have the same value
> although not the same type.
>
> Found by sparse.
>
> CC: Pavithra Ramesh
> CC: Ethan Jackson
> Signed-off-by: Ben Pfaff
> ---
> lib/bfd.c |3 ++-
> 1
On Tue, Jul 16, 2013 at 10:16 PM, Joe Stringer wrote:
> Commit fab52e16f7d8ccdbad53c5037cd04d5863a2cb4a introduces a comparison
> between mismatching types, shown by the following compiler warnings:-
>
> lib/bfd.c:461:52: warning: restricted __be64 degrades to integer
> lib/bfd.c:461:68: warning:
Commit fab52e16f7d8ccdbad53c5037cd04d5863a2cb4a introduces a comparison
between mismatching types, shown by the following compiler warnings:-
lib/bfd.c:461:52: warning: restricted __be64 degrades to integer
lib/bfd.c:461:68: warning: restricted __be32 degrades to integer
This patch changes the ht
set the type of skb to outgoing when encapsulating a packet,
since it changes from the original type to an outgoing packet.
When two tunnel ports created on two OVS bridges on same host and
two ports are the end point of the tunnel, packets are dropped as below.
If a packet has pkt_type of PACKET
> On Tue, Jul 16, 2013 at 01:43:15PM +0900, YAMAMOTO Takashi wrote:
>> it seems that openflow implementations disagree about OFPMBT_ and
>> encoding of ofp_meter_features.band_types. (see below)
>> can anyone clarify which is spec-wise correct? i peeked the spec
>> but it was unclear to me. i'm
Review of 11-13 of this patch series.
Got a bunch of spare error: warning: Using plain integer as NULL pointer,.
Most seem to be caused VLOG_RATE_LIMIT_INIT.
also:
lib/bfd.c:461:52: warning: restricted __be64 degrades to integer
lib/bfd.c:461:68: warning: restricted __be32 degrades to integer
May
On Tue, Jul 16, 2013 at 4:54 PM, Ben Pfaff wrote:
> On Tue, Jul 16, 2013 at 12:32:46PM -0700, Gurucharan Shetty wrote:
> > On Fri, Jul 12, 2013 at 2:54 PM, Ben Pfaff wrote:
> >
> > > Signed-off-by: Ben Pfaff
> > >
> > Looks good to me.
>
> Thanks, I applied patches 4 and 5.
>
Just an FYI, patch
---
lib/cfm.c|4 +++-
lib/compiler.h | 10 ++
lib/lacp.c |6 --
lib/packets.h| 18 --
lib/stp.c|9 ++---
ofproto/ofproto-dpif-ipfix.c | 30 -
On Tue, Jul 16, 2013 at 12:32:46PM -0700, Gurucharan Shetty wrote:
> On Fri, Jul 12, 2013 at 2:54 PM, Ben Pfaff wrote:
>
> > Signed-off-by: Ben Pfaff
> >
> Looks good to me.
Thanks, I applied patches 4 and 5.
___
dev mailing list
dev@openvswitch.org
h
On Mon, Jul 15, 2013 at 06:55:45PM -0700, Justin Pettit wrote:
> Don't install a flow if it's the result of the "userspace" action for an
> already installed facet. This can occur when a datapath flow with
> wildcards has a "userspace" action and flows sent to userspace result in
> a different sub
Right, I think it should work because in order to have a non-zero TCP
port the flow setup code will require that we have an exact match on
the IP protocol.
I added a log message and applied this patch.
On Tue, Jul 16, 2013 at 3:09 PM, Andy Zhou wrote:
> Never mind. It is also covered.
>
>
> On T
All of this sounds reasonable.
> I don't really like how udpif_revalidate() works, but I don't have a
> better suggestion.
Agreed. I've had a lot of trouble thinking up something better.
Perhaps it will come to me as I rework this patch.
Ethan
X-CudaMail-Whitelist-To: dev@openvswitch.org
__
Never mind. It is also covered.
On Tue, Jul 16, 2013 at 3:07 PM, Andy Zhou wrote:
> What if the IP protocol field is partially masked?
>
>
> On Tue, Jul 16, 2013 at 1:54 PM, Jesse Gross wrote:
>
>> I think it actually is safe even in the presence of partial masks
>> because of the way that we
What if the IP protocol field is partially masked?
On Tue, Jul 16, 2013 at 1:54 PM, Jesse Gross wrote:
> I think it actually is safe even in the presence of partial masks
> because of the way that we are enforcing the checks. For example, if
> we have a set TCP action then it isn't safe to just
On Fri, Jul 05, 2013 at 03:21:28PM -0700, Ethan Jackson wrote:
> This patch pulls flow miss handling into its own module,
> ofproto-dpif-upcall. In the short term, this new design imposes
> some additional overhead, but this is necessary to enable
> multithreading in the near future.
>
> Signed-o
> Does that make sense?
Yep this is pretty much what I was looking for -- a reasonable
explanation of why we should include the feature. In the patch you've
submitted, would you please add a condensed version of this to the
LACP section of vswitchd/INTERNALS so we have it to reference in
future?
I think it actually is safe even in the presence of partial masks
because of the way that we are enforcing the checks. For example, if
we have a set TCP action then it isn't safe to just look at the IP
protocol because that could correspond to a packet that is truncated
after the IP header. Instead
This change includes the output of ovs-appctl bfd/show to
ovs-bugtool output.
Signed-off-by: Pavithra Ramesh
---
utilities/bugtool/automake.mk |1 +
utilities/bugtool/ovs-bugtool-bfd-show | 19 +++
.../bugtool/plugins/network-status/openvswi
Bonjour Que Dieu vous bénisse.
Je m’appelle VERNHET Marie-Louise. Bon si je vous adresse ce mail,sachiez que
c'est l'œuvre de DIEU qui m'a amené à vous choisir parmi tant de mails archives
sur le net pour vous confier cette grande responsabilité dont je suis convaincu
que vous allez prendre co
On Fri, Jul 12, 2013 at 2:54 PM, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
>
Looks good to me.
> ---
> lib/poll-loop.c | 107
> +++
> lib/poll-loop.h | 14 +--
> 2 files changed, 78 insertions(+), 43 deletions(-)
>
> diff --git a/lib
Using masked key for validate_and_copy_actions() won't be safe against
partially masked fields.
It may not be a problem in practice, but it is technically possible with
the netlink messages.
Would you please add a error log before returning EINVAL.
I like the API clean ups. It makes the code easi
The dpif interface supports 65536 ports, but OpenFlow 1.0 supports fewer,
so ofproto-dpif needs to filter out port numbers larger than OF1.0 supports
and report an error to the caller.
Reported-by: Hiroshi Tanaka
Signed-off-by: Ben Pfaff
---
ofproto/ofproto-dpif.c |5 +
1 files changed,
On Tue, Jul 16, 2013 at 10:41:25AM -0700, Gurucharan Shetty wrote:
> On Fri, Jul 12, 2013 at 2:54 PM, Ben Pfaff wrote:
>
> > The simplification comes from dropping support for canceling a
> > poll_waiter, which was a feature that was never used. The speedup
> > comes from avoiding a malloc() for
On Fri, Jul 12, 2013 at 2:54 PM, Ben Pfaff wrote:
> The simplification comes from dropping support for canceling a
> poll_waiter, which was a feature that was never used. The speedup
> comes from avoiding a malloc() for every call to poll_fd_wait().
> (I doubt that this significantly improves pe
On Mon, Jul 15, 2013 at 05:25:34PM -0700, Andy Zhou wrote:
> Fix a corner case bug where ARP packet with ARP opcode value of 1 would cause
> tp_src and tp_dst to appear in the output string.
>
> This bug caused some output from 'ovs-appctl -t ovs-l3d l3d/recorder
> show' not be accepted by 'ovs-ap
On Mon, Jul 15, 2013 at 03:04:02PM -0700, Gurucharan Shetty wrote:
> On Fri, Jul 12, 2013 at 9:39 AM, Ben Pfaff wrote:
>
> > I don't see anything that guaranteed that ovs-ofctl would receive and print
> > the packets before it exited. This commit inserts explicit waits, to avoid
> > the problem.
On Tue, Jul 16, 2013 at 06:46:22PM +0800, ZhengLingyun wrote:
> As you have said in hindex.h, hindex need a high-quality hash function to
> work appropriately. So maybe these two kinds of hindex_head_node() are
> similar in performance.
No. The quality of the hash function ensures that each bucke
This is not a real bug, since htonl(0) and htonll(0) have the same value
although not the same type.
Found by sparse.
CC: Pavithra Ramesh
CC: Ethan Jackson
Signed-off-by: Ben Pfaff
---
lib/bfd.c |3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/lib/bfd.c b/lib/bfd.c
i
On Tue, Jul 16, 2013 at 01:43:15PM +0900, YAMAMOTO Takashi wrote:
> it seems that openflow implementations disagree about OFPMBT_ and
> encoding of ofp_meter_features.band_types. (see below)
> can anyone clarify which is spec-wise correct? i peeked the spec
> but it was unclear to me. i'm afraid
On Tue, Jul 16, 2013 at 1:18 AM, Thomas Graf wrote:
> On 07/15/2013 09:54 PM, Jesse Gross wrote:
>>
>> On Mon, Jul 15, 2013 at 12:45 PM, Pravin B Shelar
>> wrote:
>>>
>>> This reverts commit 752378e1cd1f133a8366fbacec3b281a45ff8268
>>> (datapath: rhel: Account for RHEL specific backports).
>>> Ch
On Tue, Jul 16, 2013 at 1:18 AM, Thomas Graf wrote:
> On 07/15/2013 09:54 PM, Jesse Gross wrote:
>>
>> On Mon, Jul 15, 2013 at 12:45 PM, Pravin B Shelar
>> wrote:
>>>
>>> This reverts commit 752378e1cd1f133a8366fbacec3b281a45ff8268
>>> (datapath: rhel: Account for RHEL specific backports).
>>> Ch
Following patch series restructures kernel vxlan tunneling so that
most of dataplane code is pushed to compat code and vport code looks
more like upstream openvswitch vxlan code.
Upstream vxlan tunneling code is not in kernel tree yet but patch
are sent out to netdev mailing list.
Pravin B Shelar
Following patch restructures vxlan tunneling so that it is more
in sync with upstream vxlan tunneling code.
Signed-off-by: Pravin Shelar
---
v1-v2:
- Added create flag to vxlan-port add.
- Moved rxhash functions to flow_dissector.c
---
datapath/compat.h |6 +
da
Following patch improves compat net-namespace code which is mostly
targeted for 2.6.32 kernel. upcoming vxlan restructuring patch
needs this patch.
Signed-off-by: Pravin Shelar
---
v1-v2:
Fixed code for compat net init and exit.
---
datapath/datapath.c |2 +
dat
When there is no incoming data traffic at the interface for a period,
BFD decay allows the bfd session to increase the min_rx. This is
helpful in that some interfaces usually idle for long time. And cpu
consumption can be reduced by processing fewer bfd control packets.
Signed-off-by: Alex Wang
-
Hi Ethan,
> I really don't want to go down the rabbit hole of retaining optional
> backwards compatibility for every little feature we remove just because
> there might possibly me some badly configured user who is affected on
> upgrade. I'd prefer not to accept this patch until there's a more sp
As you have said in hindex.h, hindex need a high-quality hash function to
work appropriately. So maybe these two kinds of hindex_head_node() are
similar in performance.
If unfortunately hindex meets a low-quality hash function, i think it will
depands on the hash function, which will more likely to
On 07/15/2013 09:54 PM, Jesse Gross wrote:
On Mon, Jul 15, 2013 at 12:45 PM, Pravin B Shelar wrote:
This reverts commit 752378e1cd1f133a8366fbacec3b281a45ff8268
(datapath: rhel: Account for RHEL specific backports).
Change related to netif_needs_gso() is cuasing panic on RHEL
and Xen platforms.
40 matches
Mail list logo
