I looked into what it would take to have the kernel pass up the actions
when it samples a packet. The changes seemed straightforward (see
attached diff) and worked well. I prefer this approach because:
(1) These really are the actions that the kernel is applying to the sampled
packet. No assu
On Thu, 23 Apr 2015 20:58:51 +
"Kavanagh, Mark B" wrote:
> Hi all,
>
> Just a quick poll: are the resolutions to review comments in this
> patch acceptable to everyone?
>
> If I've missed anything, or are there any additional opens that need
> to be addressed before it can be merged, just l
On 4/23/15 4:58 PM, Kavanagh, Mark B wrote:
Hi all,
Just a quick poll: are the resolutions to review comments in this patch
acceptable to everyone?
If I've missed anything, or are there any additional opens that need to be
addressed before it can be merged, just let me know.
Thanks in advanc
They ship it quickly to get you on the horse quickly
Your astoundingly erect penis will make the ladies drool
On our site you will find everything you need !
Free pills for every customer!
http://zzb.bz/yuUAg
Order best erection pharmaceuticals here
__
On Thu, Apr 23, 2015 at 1:27 PM, Sorin Vinturis
wrote:
> There is no need to use gOvsCtrlLock spinlock to guard the switch
> context, as there is now the switch context's reference count used
> for this purpose.
>
> Now the gOvsCtrlLock spinlock guards only one shared resource, the
> OVS_OPEN_INST
On Thu, Apr 23, 2015 at 11:46 AM, Alin Serdean
wrote:
> When linking executables on windows the following argument is passed
> to the linker -Qunused-arguments.
> This results in the following warning:
> Command line warning D9002 : ignoring unknown option '-Qunused-arguments'
>
> This patch remov
Currently the OVS driver code does not handle NBLs with multiple NBs.
All the packets in the latter case are dropped. This patch series
address this issue by creating multiple NBLs with only one NB for each
existing NB from a NBL with multiple NBs.
Sorin Vinturis (5):
[PATCH 1/5] datapath-window
Added support for handling multiple NBLs when creating and forwarding
packets to userspace.
Signed-off-by: Sorin Vinturis
Reported-by: Alessandro Pilotti
Reported-at: https://github.com/openvswitch/ovs-issues/issues/2
---
datapath-windows/ovsext/User.c | 15 +--
1 file changed, 9 i
Added support for handling multiple NBLs with only one NB when
encapsulating VXLAN packets.
Signed-off-by: Sorin Vinturis
Reported-by: Alessandro Pilotti
Reported-at: https://github.com/openvswitch/ovs-issues/issues/2
---
datapath-windows/ovsext/Actions.c | 21 -
datapath-w
Added support for creating and handling multiple NBLs with only one NB
for ingress data path.
Signed-off-by: Sorin Vinturis
Reported-by: Alessandro Pilotti
Reported-at: https://github.com/openvswitch/ovs-issues/issues/2
---
datapath-windows/ovsext/PacketIO.c | 91 ++
OvsPartialCopyToMultipleNBLs function failed to correctly link the newly
created NBL with single NB to the multiple NBLs list.
Signed-off-by: Sorin Vinturis
Co-authored-by: Alin Gabriel Serdean
---
datapath-windows/ovsext/BufferMgmt.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff
Signed-off-by: Sorin Vinturis
---
datapath-windows/ovsext/PacketIO.c | 99 ++
1 file changed, 47 insertions(+), 52 deletions(-)
diff --git a/datapath-windows/ovsext/PacketIO.c
b/datapath-windows/ovsext/PacketIO.c
index 1f5c8b0..69d7110 100644
--- a/datapath-w
Hi all,
Just a quick poll: are the resolutions to review comments in this patch
acceptable to everyone?
If I've missed anything, or are there any additional opens that need to be
addressed before it can be merged, just let me know.
Thanks in advance,
Mark
>
>Update relevant artifacts to add s
Instance PID is already deleted in the OvsCleanupPacketQueue function.
v2: Rebased the patch.
Signed-off-by: Sorin Vinturis
---
datapath-windows/ovsext/User.c | 6 --
1 file changed, 6 deletions(-)
diff --git a/datapath-windows/ovsext/User.c b/datapath-windows/ovsext/User.c
index 03f0377..
There is no need to use gOvsCtrlLock spinlock to guard the switch
context, as there is now the switch context's reference count used
for this purpose.
Now the gOvsCtrlLock spinlock guards only one shared resource, the
OVS_OPEN_INSTANCE global instance array.
v2: Atomically test and set if the ins
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Thu, Apr 23, 2015 at 11:55 AM, Jesse Gross wrote:
> On Wed, Apr 22, 2015 at 8:55 PM, Pravin Shelar wrote:
>> On Wed, Apr 22, 2015 at 8:50 PM, Jesse Gross wrote:
>>> On Wed, Apr 22, 2015 at 8:37 PM, Pravin Shelar wrote:
On Wed, Apr 22, 2015 at 8:29 PM, Jesse Gross wrote:
> On Wed, A
On Wed, Apr 22, 2015 at 8:55 PM, Pravin Shelar wrote:
> On Wed, Apr 22, 2015 at 8:50 PM, Jesse Gross wrote:
>> On Wed, Apr 22, 2015 at 8:37 PM, Pravin Shelar wrote:
>>> On Wed, Apr 22, 2015 at 8:29 PM, Jesse Gross wrote:
On Wed, Apr 22, 2015 at 8:22 PM, Pravin Shelar wrote:
> On Wed,
Until now the exact match cache processing was able to handle only four
megaflow. The rest of the packets was passed to the megaflow
classifier.
The limit was arbitraly set to four also because the algorithm used to
group packets in output batches didn't perform well with a lot of
megaflows.
Aft
Initializing the dp_packet's metadata can be a hot spot, especially
for very simple pipelines. Therefore improving the code here can
sometimes make a difference.
Using memcpy instead of a plain assignment helps GCC and clang generate
faster code. Here's a comparison of the compiler generated code
The userspace datapath
1. receives a batch of packets.
2. finds a 'netdev_flow' (megaflow) for each packet.
3. groups the packets in output batches based on the 'netdev_flow'.
Until now the grouping (2) was done using a simple algorithm with a
O(N^2) runtime, where N is the number of distinct meg
Also, removes an unused function
Signed-off-by: Daniele Di Proietto
---
lib/dp-packet.h | 230
1 file changed, 131 insertions(+), 99 deletions(-)
diff --git a/lib/dp-packet.h b/lib/dp-packet.h
index 1d10d99..e4c2593 100644
--- a/lib/dp-pa
DPDK buf_len is only 16-bit wide ('allocated' was 32-bit), but it should
be enough to store the number of allocated bytes.
This will reduce 'struct dp_packet' size.
Signed-off-by: Daniele Di Proietto
---
lib/dp-packet.c | 6 +++---
lib/dp-packet.h | 26 +++---
2 files chang
In 'struct ofpbuf' the 'frame' pointer was used to parse different kinds of
data (Ethernet, OpenFlow, Netlink attributes). For Ethernet packets the
'frame' pointer was supposed to have the same value as the 'data'
pointer.
Since 'struct dp_packet' is only used for Ethernet packets, there's no
nee
This series contains different tweaks to improve the performance of the
userspace datapath with DPDK ports.
The first commits reduce the size of struct dp_packet to three cachelines
(two used by DPDK and one for our metadata). I've put in also some style
fixes for lib/dp-packet.h
Then, a microopt
The 'list' member is only used (two users) in the slow path.
This commit removes it to reduce the struct size
Signed-off-by: Daniele Di Proietto
---
lib/dp-packet.c| 13 -
lib/dp-packet.h| 8
lib/netdev-dummy.c | 50 ++
On Thu, Apr 23, 2015 at 08:46:40AM +, Sorin Vinturis wrote:
> There is no need to use gOvsCtrlLock spinlock to guard the switch
> context, as there is now the switch context's reference count used
> for this purpose.
>
> Now the gOvsCtrlLock spinlock guards only one shared resource, the
> OVS_
Applied to master, thanks!
On Thu, Apr 23, 2015 at 12:49:39PM +, Sorin Vinturis wrote:
> That's a good catch. Thanks Nithin!
>
> Acked-by: Sorin Vinturis
>
> -Original Message-
> From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Nithin Raju
> Sent: Thursday, 23 April, 2015
On Thu, 23 Apr 2015 10:52:18 -0700
Gurucharan Shetty wrote:
> >
> > The patch is still needed on Windows, so I suggest to keep it with
> > a reference for "Windows-only" or move the requirement to
> > INSTALL.Windows.md.
> In INSTALL.Windows.md we get patch as a bundle in MinGW/Msys and we
> ask
>
> The patch is still needed on Windows, so I suggest to keep it with
> a reference for "Windows-only" or move the requirement to
> INSTALL.Windows.md.
In INSTALL.Windows.md we get patch as a bundle in MinGW/Msys and we
ask it as a requirement to be installed. It will be a little odd to
list the '
Thanks! I pushed it.
--Justin
> On Apr 22, 2015, at 1:15 PM, Russell Bryant wrote:
>
> The show command takes an optional LSWITCH argument. In that case, it
> only prints the info for a single switch instead of all of them.
> Update the usage text to reflect that option.
>
> Signed-off-by:
It will be like youre a new man in the bedroom
You wont be able to keep your boner down with these pills
On out site you will find the cure!
Free pills only this day!
http://zzb.bz/cJdml
World Web-store by FDA
___
dev mailing list
dev@openvswit
On Thu, 23 Apr 2015 08:42:47 -0700
Gurucharan Shetty wrote:
> On CentOS machines which use autoconf version 2.63, the patch
> application would fail.
>
> Reported-by: Ian Stokes
> Tested-by: Ian Stokes
> Signed-off-by: Gurucharan Shetty
> ---
> AUTHORS |1 +
> INSTALL.md
>
>
> This reverts the previous fix to build in another directory.
> s/ -o testsuite.tmp/ -o $@.tmp/
>
> Since it's now only used on Windows I don't care, but still
> I think 'patch' will break on Windows too in the same scenario.
You are right. Though we don't use a different build directory in
Wi
On CentOS machines which use autoconf version 2.63, the patch
application would fail.
Reported-by: Ian Stokes
Tested-by: Ian Stokes
Signed-off-by: Gurucharan Shetty
---
AUTHORS |1 +
INSTALL.md|2 --
tests/automake.mk |7 +++
3 files changed, 8 insertions(+),
On Thu, 23 Apr 2015 07:22:12 -0700
Gurucharan Shetty wrote:
> On CentOS machines which use autoconf version 2.63, the patch
> application would fail.
>
> Reported-by: Ian Stokes
> Signed-off-by: Gurucharan Shetty
> ---
> AUTHORS |1 +
> INSTALL.md|2 --
> tests/autom
> -Original Message-
> From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Stokes, Ian
> Sent: Thursday, April 23, 2015 4:51 PM
> To: Gurucharan Shetty; dev@openvswitch.org
> Cc: Gurucharan Shetty
> Subject: Re: [ovs-dev] [PATCH] testsuite: Don't apply the
> testsuite.patch on non
> -Original Message-
> From: Gurucharan Shetty [mailto:shet...@nicira.com]
> Sent: Thursday, April 23, 2015 3:22 PM
> To: dev@openvswitch.org
> Cc: Stokes, Ian; Gurucharan Shetty
> Subject: [PATCH] testsuite: Don't apply the testsuite.patch on non-
> Windows platforms.
>
> On CentOS machin
On CentOS machines which use autoconf version 2.63, the patch
application would fail.
Reported-by: Ian Stokes
Signed-off-by: Gurucharan Shetty
---
AUTHORS |1 +
INSTALL.md|2 --
tests/automake.mk | 13 ++---
3 files changed, 11 insertions(+), 5 deletions(-)
This patch simply forwards unicast netpoll packets via one of physical
interface in datapath depending on source mac address from the skb.
It seems possible to use common net flow classification for netpoll but
there is no way to guarantee presence of right flow in kernel cache.
Signed-off-by: Ko
That's a good catch. Thanks Nithin!
Acked-by: Sorin Vinturis
-Original Message-
From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Nithin Raju
Sent: Thursday, 23 April, 2015 03:10
To: dev@openvswitch.org
Subject: [ovs-dev] [PATCH] datapath-windows: don't free switch cxt until re
Instance PID is already deleted in the OvsCleanupPacketQueue function.
Signed-off-by: Sorin Vinturis
---
datapath-windows/ovsext/User.c | 6 --
1 file changed, 6 deletions(-)
diff --git a/datapath-windows/ovsext/User.c b/datapath-windows/ovsext/User.c
index 4fc3497..efe52b1 100644
--- a/dat
Signed-off-by: yinpeijun
---
ofproto/ofproto-dpif-xlate.c | 12
1 file changed, 12 insertions(+)
diff --git a/ofproto/ofproto-dpif-xlate.c b/ofproto/ofproto-dpif-xlate.c
index f737877..396d261 100644
--- a/ofproto/ofproto-dpif-xlate.c
+++ b/ofproto/ofproto-dpif-xlate.c
@@ -2749,6 +2
You're in luck! Get ready to earn
Dear dev@openvswitch.org,
It looks like your email software does not support HTML.
Please visit the webpage below in order to read this message in your webbrowser:
http://newsletter.eachgame.com/x/?S7Y1.J9ra2Jpafm.yNbE0NLS2Oh.jq2ZsZkxAAA44
__
The hottest news of our thin growing world! The way out was invented in 1980-s!
Give yourself the best chance to decrease your size.
Our stock list is so extensive you are guaranteed to find what suits you!
http://zzb.bz/JdmKP
___
dev mailing list
dev@o
There is no need to use gOvsCtrlLock spinlock to guard the switch
context, as there is now the switch context's reference count used
for this purpose.
Now the gOvsCtrlLock spinlock guards only one shared resource, the
OVS_OPEN_INSTANCE global instance array.
v2: Atomically test and set if the ins
The message could not be delivered
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 04/23/15 09:19, Mark D. Gray wrote:
On 04/22/15 22:18, Jesse Gross wrote:
On Wed, Apr 22, 2015 at 1:15 AM, mark.d.gray
wrote:
Pravin Shelar wrote:
On Tue, Apr 21, 2015 at 7:30 PM, Jesse Gross wrote:
On Tue, Apr 21, 2015 at 7:23 PM, Pravin Shelar
wrote:
On Tue, Apr 21, 2015 at 6:33 PM,
On 04/22/15 22:18, Jesse Gross wrote:
On Wed, Apr 22, 2015 at 1:15 AM, mark.d.gray wrote:
Pravin Shelar wrote:
On Tue, Apr 21, 2015 at 7:30 PM, Jesse Gross wrote:
On Tue, Apr 21, 2015 at 7:23 PM, Pravin Shelar
wrote:
On Tue, Apr 21, 2015 at 6:33 PM, Ethan Jackson wrote:
I really don't
49 matches
Mail list logo
