Hello, everyone.
I am modifying the datapath of ovs to queue incoming
packets and delay the time of process of them.
In datapath.c, I added codes creating a thread
that inputs incoming packets to a queue, and process them later
(in regular).
Once ovs_dp_process_received_packet() is
Sorry for long delay here.
I wanted to point out something about:
"At a high level, the fraction of packets that make it to userspace should be
very small. So it is not worth the optimization"
If we have the port NORMAl (and perhaps, FLOOD) set, then the queuing of
packets to userspace happens
Hi All,
I have gone the research paper : P4: Programming protocol independent
packet processors. Following is its link
http://arxiv.org/pdf/1312.1719.pdf
This paper takes about implementing Openflow2.0 .
I am interested in contributing in making open vSwitch Openflow2.0 compliant.
Do we have a pla
On Thu, Aug 21, 2014 at 03:08:59PM +, Samuel Ghinet wrote:
> I wanted to point out something about:
> "At a high level, the fraction of packets that make it to userspace should be
> very small. So it is not worth the optimization"
>
> If we have the port NORMAl (and perhaps, FLOOD) set, then
The goal of this is to provide a possibility to suport various switch
chips. Drivers should implement relevant ndos to do so. Now there is a
couple of ndos defines:
- for getting physical switch id is in place.
- for work with flows.
Note that user can use random port netdevice to access the switc
The netdevice represents a port in a switch, it will expose
IFLA_PHYS_SWITCH_ID value via rtnl. Two netdevices with the same value
belong to one physical switch.
Signed-off-by: Jiri Pirko
---
include/uapi/linux/if_link.h | 1 +
net/core/rtnetlink.c | 26 +-
2 fil
This patchset can be divided into 3 main sections:
- introduce switchdev api for implementing switch drivers
- add hardware acceleration bits into openvswitch datapath, This uses
previously mentioned switchdev api
- introduce rocker switch driver which implenets switchdev api
More info in separa
After this, flow related structures can be used in other code.
Signed-off-by: Jiri Pirko
---
include/linux/sw_flow.h| 98 ++
net/openvswitch/actions.c | 3 +-
net/openvswitch/datapath.c | 74 +-
net/openvswitch/datapath
Signed-off-by: Jiri Pirko
---
net/dsa/Kconfig | 2 +-
net/dsa/slave.c | 16
2 files changed, 17 insertions(+), 1 deletion(-)
diff --git a/net/dsa/Kconfig b/net/dsa/Kconfig
index f5eede1..66c445a 100644
--- a/net/dsa/Kconfig
+++ b/net/dsa/Kconfig
@@ -1,6 +1,6 @@
config HAVE_NET
Signed-off-by: Jiri Pirko
---
net/core/net-sysfs.c | 24
1 file changed, 24 insertions(+)
diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
index 55dc4da..69e3d64 100644
--- a/net/core/net-sysfs.c
+++ b/net/core/net-sysfs.c
@@ -12,6 +12,7 @@
#include
#include
Signed-off-by: Jiri Pirko
Acked-by: Scott Feldman
---
include/linux/netdevice.h | 7 +++
1 file changed, 7 insertions(+)
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 8b5d14c..b48028d 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -751,6
Dummy switch implementation using switchdev interface
Signed-off-by: Jiri Pirko
---
drivers/net/Kconfig | 7 +++
drivers/net/Makefile | 1 +
drivers/net/dummyswitch.c| 131 +++
include/uapi/linux/if_link.h | 9 +++
4 files change
Signed-off-by: Jiri Pirko
---
include/linux/sw_flow.h | 3 +++
net/core/switchdev.c | 10 ++
net/openvswitch/hw_offload.c | 23 +++
3 files changed, 36 insertions(+)
diff --git a/include/linux/sw_flow.h b/include/linux/sw_flow.h
index 079d065..e2ee54a 10
This will allow to query easily if the vport has netdev. Also it allows
to unexpose netdev_vport_priv and struct netdev_vport.
Signed-off-by: Jiri Pirko
---
net/openvswitch/datapath.c | 2 +-
net/openvswitch/dp_notify.c | 7 ++---
net/openvswitch/vport-internal_dev.c | 56 ++
Benefit from the possibility to work with flows in switch devices and
use the swdev api to offload flow datapath.
Signed-off-by: Jiri Pirko
---
include/linux/sw_flow.h| 14 +++
net/openvswitch/Makefile | 3 +-
net/openvswitch/datapath.c | 33 ++
net/openvswitch/datapath
This patch introduces the first driver to benefit from the switchdev
infrastructure and to implement newly introduced switch ndos. This is a
driver for emulated switch chip implemented in qemu:
https://github.com/sfeldma/qemu-rocker/
This patch is a result of joint work with Scott Feldman.
Signed
Ankur,
Thanks for working on this. I had the following overall comments. I have
annotated the patch with specific comments as appropriate.
I didn't realize initially that the Netlink.c, Netlink.h that you are
introducing have major modifications compared to the userspace netlink.c and
netlink.h
hi Ankur,
Thanks for working on this. Looks good overall, but for the minor comments I
had.
On Aug 20, 2014, at 9:46 AM, Ankur Sharma wrote:
> In this patch we incorporate the usage of netlink message
> and the parsing APIs that were added in previous commit.
>
> Change-Id: If627ba5a10d78b9366
hi Ankur,
Looks good, but for a few minor comments.
On Aug 20, 2014, at 9:46 AM, Ankur Sharma wrote:
> In this patch we remove reference to OvsNetlink.h.
> Since we do not refer to lib/netlink-protocol.h anymore,
> hence removed the WIN_DP based check.
>
> Change-Id: I281a0c6478e3de2e9b04c988be
Hi,
Would you be interested in *Virtualization Software Users *contact
information in excel sheet for unlimited marketing usage? We provide Top
Decision Makers information like Name, Email, Title, Phone and company
details of *companies using Virtualization.*
*Cloud*
*Docker*
*Big Data*
*Had
hi Alin/Sam,
Like I said earlier, this is a critical fix, and thanks for working on this.
I started looking at OvsPacketIO.c, and you are right in using the
OvsPartialCopyToMultipleNBLs() API. However, is it required to update many of
the functions to process an NB instead of an NBL? From my und
2014-08-21 9:19 GMT-07:00 Jiri Pirko :
> Signed-off-by: Jiri Pirko
> ---
> net/dsa/Kconfig | 2 +-
> net/dsa/slave.c | 16
> 2 files changed, 17 insertions(+), 1 deletion(-)
>
> diff --git a/net/dsa/Kconfig b/net/dsa/Kconfig
> index f5eede1..66c445a 100644
> --- a/net/dsa/Kconfi
Thu, Aug 21, 2014 at 06:41:23PM CEST, b...@decadent.org.uk wrote:
>On Thu, 2014-08-21 at 18:18 +0200, Jiri Pirko wrote:
>> The goal of this is to provide a possibility to suport various switch
>> chips. Drivers should implement relevant ndos to do so. Now there is a
>> couple of ndos defines:
>> -
2014-08-21 9:18 GMT-07:00 Jiri Pirko :
> The goal of this is to provide a possibility to suport various switch
> chips. Drivers should implement relevant ndos to do so. Now there is a
> couple of ndos defines:
> - for getting physical switch id is in place.
> - for work with flows.
>
> Note that us
Thu, Aug 21, 2014 at 06:56:13PM CEST, f.faine...@gmail.com wrote:
>2014-08-21 9:19 GMT-07:00 Jiri Pirko :
>> Signed-off-by: Jiri Pirko
>> ---
>> net/dsa/Kconfig | 2 +-
>> net/dsa/slave.c | 16
>> 2 files changed, 17 insertions(+), 1 deletion(-)
>>
>> diff --git a/net/dsa/Kconfi
Hello Saurabh,
Thanks for the clarifications!
The param name "copySize" was very ambiguous.
I did not debug this exact scenario, but you must be right:
- Cloning an original NB that has DataOffset > 0 causes the clone to have
DataOffset = 0
- NdisRetreatNetBufferDataStart doc says
"
DataOffsetDe
2014-08-21 10:06 GMT-07:00 Jiri Pirko :
> Thu, Aug 21, 2014 at 06:56:13PM CEST, f.faine...@gmail.com wrote:
>>2014-08-21 9:19 GMT-07:00 Jiri Pirko :
>>> Signed-off-by: Jiri Pirko
>>> ---
>>> net/dsa/Kconfig | 2 +-
>>> net/dsa/slave.c | 16
>>> 2 files changed, 17 insertions(+),
2014-08-21 9:19 GMT-07:00 Jiri Pirko :
> This patch introduces the first driver to benefit from the switchdev
> infrastructure and to implement newly introduced switch ndos. This is a
> driver for emulated switch chip implemented in qemu:
> https://github.com/sfeldma/qemu-rocker/
>
> This patch is
On 8/6/14 4:37 AM, Jesse Gross wrote:
On 8/2/14, 4:11 AM, Jesse Gross wrote:
On Fri, Aug 1, 2014 at 3:19 PM, Lori Jakab wrote:
On 8/1/14, 3:25 AM, Jesse Gross wrote:
On Thu, Jul 31, 2014 at 5:45 AM, Lori Jakab wrote:
On 7/31/14, 10:09 AM, Jesse Gross wrote:
Going further, what happens if p
struct ovs_skb_cb is full on kernels < 3.11 due to compatibility code.
This patch removes the 'flow' member in order to make room for data
needed by layer 3 flow/port support that will be added in an upcoming
patch. The 'flow' memeber was chosen for removal because it's only used
in ovs_execute_ac
On 8/2/14 4:11 AM, Jesse Gross wrote:
On Fri, Aug 1, 2014 at 3:19 PM, Lori Jakab wrote:
On 8/1/14, 3:25 AM, Jesse Gross wrote:
On Thu, Jul 31, 2014 at 5:45 AM, Lori Jakab wrote:
On 7/31/14, 10:09 AM, Jesse Gross wrote:
On Fri, Jun 27, 2014 at 6:21 AM, Lorand Jakab wrote:
diff --git a/dat
OpenFlow 1.2 and later have "experimenter errors". The OVS implementation
was buggy in a few ways. First, a bug in extract-ofp-errors prevented
OF1.2+ experimenter errors from being properly decoded. Second,
OF1.2+ experimenter errors have only a type, not a code, whereas all other
types of erro
From: Jean Tourrilhes
Signed-off-by: Jean Tourrilhes
[b...@nicira.com removed the definitions of these errors in OF1.1 and OF1.2]
Signed-off-by: Ben Pfaff
---
lib/ofp-errors.h | 12 ++--
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/lib/ofp-errors.h b/lib/ofp-errors.h
Thanks Nithin for your opinions on coding style!
[QUOTE]
The guideline is fine. It seems to be a rewording of the existing guideline,
but it is ok to do it.
[/QUOTE]
Except, the current coding style says:
"For types, use all upper case for all letters with words separated by '_'. If
camel casin
Hello Nithin,
The part where I am trying to save a lock is in
"4. Process the packet using the flow"
Also, I believe that, when ref count = 0, it should mean that the data is
"unsafe" (i.e. you need to lock to access / write to it).
When ref count = 0, it shouldn't mean that the flow must be del
Sorry, I'll remake the commit.
Sam
From: Nithin Raju [nit...@vmware.com]
Sent: Saturday, August 16, 2014 9:34 AM
To: Samuel Ghinet
Cc: dev@openvswitch.org
Subject: Re: [ovs-dev] [PATCH 10/15] datapath-windows: Compile as C by default
hi Sam,
I don't unders
Okay, got it!
Thanks!
Sam
From: Nithin Raju [nit...@vmware.com]
Sent: Saturday, August 16, 2014 9:38 AM
To: Samuel Ghinet
Cc: dev@openvswitch.org
Subject: Re: [ovs-dev] [PATCH 07/15] datapath-windows: Replace Types.h
On Aug 6, 2014, at 9:11 AM, Samuel Ghin
Hello Nithin,
[QUOTE]
I don't think we need two separate definitions. If ASSERT() is defined debug
only, then it would be a no-op in 'OVS_USE_ASSERTS'.
[/QUOTE]
The more complex assert-based macros would be useful if we would want to do a
more 'proper' cleanup in release mode, for a bug.
e.g.
i
When there are multiple vlan headers present in a received frame, the first
one is put into vlan_tci and protocol is set to ETH_P_8021Q. Anything in the
skb beyond the VLAN TPID may be still non-linear, including the inner TCI
and ethertype. While ovs_flow_extract takes care of IP and IPv6 headers,
Also, I think it would be cleaner & would improve code readability if we would
use a max 1 goto Label, e.g. a "Cleanup". Rather than using multiple labels per
func.
A "Cleanup" would be something like cleanup in case something went wrong - the
all-ok part means reaching the end of the function n
Hello Nithin,
[QUOTE]
Is this a pre-cursor for implementing the persistent ports feature that the
cloudbase repo had implemented.
[/QUOTE]
Yes.
I think it is better to drop the "persistent port" term. When I had designed it
long ago, I chose the name "persistent" because it "persists" during hy
I have just asked on osronline.
Even though it is not critical now, it will be useful to know for sure :)
[QUOTE]
We an add an additional API like OvsAllocateMemory() which also zeros the
memory.
[/QUOTE]
I would sincerely prefer something shorter than an OvsAllocateZeroedMemory :)
Memory alloca
Joe,
What was the reason to go from 64-bit hash to 128-bit hash?
Jarno
> On Aug 20, 2014, at 10:41 PM, Joe Stringer wrote:
>
> One of the current bottlenecks for revalidation performance is the cost of
> assembling flow keys, masks and actions in the (linux) datapath. This series
> seeks to
Hello Nithin,
Thanks for the clarifications!
Some notes:
- AFAIK, RARP is long since obsolete, so I don't know if we actually need to
take it into account.
- When you say GARP, you mean the Gratuitous Address Resolution Protocol,
right? not the Generic Attribute Registration Protocol.
If it's t
Port "NORMAL" in userspace means:
- when a packet is coming from dp port1, its eth address, ethAddr1 is saved for
the dp port.
- when another packet is coming from dp port2, and has dest eth address =
ethAddr1, a kernel flow must be created that says "if packet has all frames
like this, send to
I have also looked upon the patch.
It's good to see the netlink protocol being implemented!
There would be a few minor issues:
Netlink.h:
[PATCH]
/* Netlink attribute types. */
typedef enum
{
NL_A_NO_ATTR = 0,
NL_A_UNSPEC,
NL_A_U8,
NL_A_U16,
NL_A_BE16 = NL_A_U16,
NL_A_U32,
Hi Sam,
Thanks a lot for the review.
1. Rename NL_ATTR_TYPE:
[Ankur]: Agreed that it should be renamed. Will it be fine if i do it in a
followup checkin (will create an issue on github to make sure that this
suggestion is not missed out)?
2. Use MAXUINT64
[Ankur]: My bad i missed this macro.
On Fri, Aug 22, 2014 at 01:55:13AM +, Samuel Ghinet wrote:
> port NORMAL and port FLOOD, unlike physical (i.e. ports corresponding to
> specific VMs or to external) or logical (GRE / VXLAN / etc.) are a userspace
> concept only.
> This means that, when a "ovs-ofctl add-flow br0 actions=normal
48 matches
Mail list logo
