As you wish.
Alin.
From: Ben Pfaff [b...@nicira.com]
Sent: Saturday, November 23, 2013 4:27 AM
To: Alin Serdean
Cc: Alessandro Pilotti; dev@openvswitch.org
Subject: Re: [ovs-dev] Windows port status
On Sat, Nov 23, 2013 at 12:23:41AM +, Alin Serdean wr
That sounds great, thanks a lot.
On Fri, Nov 22, 2013 at 05:06:11PM -0800, Andy Zhou wrote:
> Sure, I will test it more and wait for a bit before pushing, just in case
> Guolin has additional feedback. Thanks for the review and redactor
> suggestion.
>
>
> On Fri, Nov 22, 2013 at 4:58 PM, Ben P
On Sat, Nov 23, 2013 at 12:23:41AM +, Alin Serdean wrote:
> > That reads like a non sequitur, I don't see how it responds to my
> > comments.
>
> My problem is where to insert the macro exactly.
util.h perhaps?
___
dev mailing list
dev@openvswitch.o
Sure, I will test it more and wait for a bit before pushing, just in case
Guolin has additional feedback. Thanks for the review and redactor
suggestion.
On Fri, Nov 22, 2013 at 4:58 PM, Ben Pfaff wrote:
> Do you want to check it in then?
>
> On Fri, Nov 22, 2013 at 04:53:09PM -0800, Andy Zhou
Do you want to check it in then?
On Fri, Nov 22, 2013 at 04:53:09PM -0800, Andy Zhou wrote:
> Yes, this patch would eliminate code duplications. The logic looks right
> to me.
>
>
> On Fri, Nov 22, 2013 at 3:57 PM, Ben Pfaff wrote:
>
> > On Thu, Nov 21, 2013 at 08:45:22PM -0800, Andy Zhou wro
Yes, this patch would eliminate code duplications. The logic looks right
to me.
On Fri, Nov 22, 2013 at 3:57 PM, Ben Pfaff wrote:
> On Thu, Nov 21, 2013 at 08:45:22PM -0800, Andy Zhou wrote:
> > Noting updating slow path subfacet's time stamp can cause their datapath
> > flows deleted periodic
> That reads like a non sequitur, I don't see how it responds to my
> comments.
My problem is where to insert the macro exactly.
Alin.
From: Ben Pfaff [b...@nicira.com]
Sent: Saturday, November 23, 2013 2:04 AM
To: Alin Serdean
Cc: Alessandro Pilotti; dev@
Yes, exactly.
On Fri, Nov 22, 2013 at 04:08:00PM -0500, Ethan Jackson wrote:
> I think what Ben is saying is that we want some way to verify that no
> one uses the old "%zu" macro accidentally. Basically, make should do
> a grep for it, and fail if it's used.
>
> Ethan
>
> On Fri, Nov 22, 2013
On Fri, Nov 22, 2013 at 07:25:14PM +, Alin Serdean wrote:
>
> > PRI_ZU is a bad name because it doesn't follow the existing pattern.
> > The name PRIuSIZE fits the pattern, so that is what I would suggest.
>
> > The existing PRI* macros only include a type modifier, not the format
> > specifi
Thank you. I will apply this to master soon.
On Thu, Nov 21, 2013 at 04:31:49PM -0800, Jarno Rajahalme wrote:
> Acked-by: Jarno Rajahalme
>
> On Oct 8, 2013, at 13:42 PM, Ben Pfaff wrote:
> > The ONF header file shows this as a 3-byte field.
> >
> > Reported-by: YAMAMOTO Takashi
> > Signed-o
On Thu, Nov 21, 2013 at 08:45:22PM -0800, Andy Zhou wrote:
> Noting updating slow path subfacet's time stamp can cause their datapath
> flows deleted periodically. For example, CFM datapath flow have usespace
> actions that are handled in dpif slow path. They are deleted and
> recreated periodicall
On Fri, Nov 22, 2013 at 2:57 PM, Thomas Graf wrote:
> On 11/22/2013 11:47 PM, Pravin Shelar wrote:
>>
>> This patch actually fixes race in case of device is moved from one
>> bridge to another.
>>
>> e.g.
>> ovs-vsctl -- remove Bridge br0 ports -- add Bridge br1
>> ports
>>
>> events in this cas
On 11/22/2013 11:47 PM, Pravin Shelar wrote:
This patch actually fixes race in case of device is moved from one
bridge to another.
e.g.
ovs-vsctl -- remove Bridge br0 ports -- add Bridge br1
ports
events in this case (This bug only effects RHEL6 kernel).
1. netdev destroy (vport1, dev1) : sch
pushed to master.
Thanks.
On Thu, Nov 21, 2013 at 7:06 PM, Jesse Gross wrote:
> On Wed, Nov 20, 2013 at 9:50 AM, Pravin B Shelar wrote:
>> On 3.11 kernel, gre module exports IP_GRE demux API, ovs needs
>> to use it to register for GRE protocol handler.
>>
>> Reported-by: Nithin Nayak Sujir
>>
This patch actually fixes race in case of device is moved from one
bridge to another.
e.g.
ovs-vsctl -- remove Bridge br0 ports -- add Bridge br1
ports
events in this case (This bug only effects RHEL6 kernel).
1. netdev destroy (vport1, dev1) : schedule rcu callback (vport1).
2. netdev create (
pcap_read() can use it whether we're big-endian or not.
Signed-off-by: Ben Pfaff
---
lib/byte-order.h |4 +---
lib/pcap-file.c |6 ++
2 files changed, 3 insertions(+), 7 deletions(-)
diff --git a/lib/byte-order.h b/lib/byte-order.h
index f9be78b..352f91e 100644
--- a/lib/byte-order
Based on the number of people who ask about Wireshark support for OpenFlow,
this is likely to be widely useful.
Signed-off-by: Ben Pfaff
---
lib/pcap-file.c | 138 ++
lib/pcap-file.h |9 +++
utilities/ovs-ofctl.8.in | 18 +-
Only the write support is initially useful, but an upcoming commit will add
a user for the read support.
Signed-off-by: Ben Pfaff
---
lib/pcap-file.c | 31 +--
lib/pcap-file.h |2 +-
tests/test-flows.c|4 ++--
utilities/ovs-ofctl.c |2 +-
An upcoming commit will add the first user.
Signed-off-by: Ben Pfaff
---
lib/ofpbuf.c | 18 ++
lib/ofpbuf.h |1 +
2 files changed, 19 insertions(+)
diff --git a/lib/ofpbuf.c b/lib/ofpbuf.c
index d578ab5..27b6f17 100644
--- a/lib/ofpbuf.c
+++ b/lib/ofpbuf.c
@@ -359,6 +359,2
Occasionally I get asked for the ability to capture all OpenFlow traffic
in a raw format, to allow for analysis and debugging. OVS has a few
existing tools for this but none of them really fit the bill. For example,
turning up the log level for the "vconn" module will dump all traffic but
only in
Use percpu allocator for stats due to objection to stats array.
But percpu allocator is not designed for high churn allocation/
deallcation. so we need to avoid allocating percpu flow for
short lived flows. One cheaper way to detect flow is by checking
if 5-tuple used in RSS are masked or not. if a
I think what Ben is saying is that we want some way to verify that no
one uses the old "%zu" macro accidentally. Basically, make should do
a grep for it, and fail if it's used.
Ethan
On Fri, Nov 22, 2013 at 2:25 PM, Alin Serdean
wrote:
>
>> PRI_ZU is a bad name because it doesn't follow the exi
On Nov 22, 2013, at 12:32 PM, Ben Pfaff wrote:
> Is it OK for me to wait for a v4 to continue review?
Yes,
Jarno
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Fri, Nov 22, 2013 at 11:56:19AM -0800, Jarno Rajahalme wrote:
> On Nov 21, 2013, at 5:10 PM, Ben Pfaff wrote:
> > I am not done studying this code, but I have some preliminary
> > comments.
> >
> > Some of the fields that allow prefix length are in network byte order,
> > and others (the regis
Ben,
Thank you very much for the review so far! See responses below:
On Nov 21, 2013, at 5:10 PM, Ben Pfaff wrote:
> I am not done studying this code, but I have some preliminary
> comments.
>
> Some of the fields that allow prefix length are in network byte order,
> and others (the registers)
> PRI_ZU is a bad name because it doesn't follow the existing pattern.
> The name PRIuSIZE fits the pattern, so that is what I would suggest.
> The existing PRI* macros only include a type modifier, not the format
> specifier, so PRIuSIZE should not include the format specifier either.
Is this w
On Fri, Nov 22, 2013 at 10:41:18AM -0800, Joe Stringer wrote:
> On 20 November 2013 11:15, Ben Pfaff wrote:
> >> NB: I'm tracking STP changes in ofproto_port_set_state(), as the testsuite
> >> would fail when I tracked them in stp_set_port_state(). The latter makes
> >> more
> >> sense to me, but
On 20 November 2013 11:15, Ben Pfaff wrote:
>> NB: I'm tracking STP changes in ofproto_port_set_state(), as the testsuite
>> would fail when I tracked them in stp_set_port_state(). The latter makes more
>> sense to me, but I clearly don't understand some interaction there because it
>> makes STP s
Use of skb_zerocopy() can avoids the expensive call to memcpy()
when copying the packet data into the Netlink skb. Completes
checksum through skb_checksum_help() if needed.
Zerocopy is only performed if user space supported unaligned
Netlink messages. memory mapped netlink i/o is preferred over
ze
Use memory mapped Netlink i/o for all unicast openvswitch
communication if a ring has been set up.
Benchmark
* pktgen -> ovs internal port
* 5M pkts, 5M flows
* 4 threads, 8 cores
Before:
Result: OK: 67418743(c67108212+d310530) usec, 500 (9000byte,0frags)
74163pps 5339Mb/sec (53397360
Drop user features if an outdated user space instance that does not
understand the concept of user_features attempted to create a new
datapath.
Signed-off-by: Thomas Graf
Reviewed-by: Daniel Borkmann
---
include/uapi/linux/openvswitch.h | 10 +-
net/openvswitch/datapath.c | 9 +++
Consolidates ovs_dp_cmd_new() and ovs_dp_cmd_set() to simplify
handling and avoid code duplication.
Allows user space to specify NLM_F_REPLACE with OVS_DP_CMD_NEW and
overwrite the settings such as the user features of an existing
datapath.
Signed-off-by: Thomas Graf
Reviewed-by: Daniel Borkmann
Signed-off-by: Thomas Graf
Reviewed-by: Daniel Borkmann
---
include/uapi/linux/openvswitch.h | 4
net/openvswitch/datapath.c | 14 ++
net/openvswitch/datapath.h | 2 ++
3 files changed, 20 insertions(+)
diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/
An insufficent ring frame size configuration can lead to an
unnecessary skb allocation for every Netlink message. Check frame
size before taking the queue lock and allocating the skb and
re-check with lock to be safe.
Signed-off-by: Thomas Graf
Reviewed-by: Daniel Borkmann
---
net/netlink/af_ne
Reposting this before the merge window as it will go through Jesse's
tree.
Given jumbo frames, the capacity of the slow path is improved by
a factor of 2.5x.
V7: - removed unintential kernel-doc comment
- WARN_ONCE() -> WARN(), message on single line, added \n
V6: - Added memory mapped netli
Allocates a new sk_buff large enough to cover the specified payload
plus required Netlink headers. Will check receiving socket for
memory mapped i/o capability and use it if enabled. Will fall back
to non-mapped skb if message size exceeds the frame size of the ring.
Signed-of-by: Thomas Graf
Rev
Make the skb zerocopy logic written for nfnetlink queue available for
use by other modules.
Signed-off-by: Thomas Graf
Reviewed-by: Daniel Borkmann
---
include/linux/skbuff.h | 3 ++
net/core/skbuff.c| 85
net/netfilter/nfn
Hello.
On 22-11-2013 0:52, Ben Pfaff wrote:
Following commit (''netlink: Do not enforce alignment of last Netlink
attribute''), signal the ability to receive unaligned Netlink messages
to the datapath to enable utilization of zerocopy optimizations.
Signed-off-by: Thomas Graf
[...]
diff
Title: FORMACIÓN SUPERIOR EN REDES SOCIALES Y MEDIOS ONLINE - UNIVERSIDAD DE ALCALÁ
FORMACIÓN SUPERIOR EN REDES SOCIALES Y MEDIOS ONLINE UNIVERSIDAD DE ALCALÁ Pulse AQUÍ si no visualiza correctamente las imágenes.
39 matches
Mail list logo
