In this change:
1. we refactor the code that fills up information about the DP into
a seprate function.
2. use the netlink set APIs to fill up the netlink attributes.
Signed-off-by: Nithin Raju
---
datapath-windows/ovsext/Datapath.c | 123 +++-
1 files changed, 6
We should be returning value of tail before the increment
and not after.
Signed-off-by: Nithin Raju
---
datapath-windows/ovsext/Netlink/NetlinkBuf.c |6 ++
1 files changed, 2 insertions(+), 4 deletions(-)
diff --git a/datapath-windows/ovsext/Netlink/NetlinkBuf.c
b/datapath-windows/ovse
> On Sep 12, 2014, at 11:35 AM, Ben Pfaff wrote:
>
>> On Fri, Sep 12, 2014 at 11:24:23AM -0700, Jarno Rajahalme wrote:
>> Use the new OVS_FLOW_ATTR_PROBE flag when probing for datapath feature
>> support. Suppress also dpif error logging when requested, as probe
>> failures are already logged a
There are couple of reasons to remove this support:
* This is used in very old OVS use-case. It is much better
to read stats directly from OVS.
* Forthcoming commit will remove support for setting stats
for vport. The stats update depends on stats-set.
Signed-off-by: Pravin B Shelar
-
With this commit, ovs by default will create one pmd thread
for each numa node and pin the pmd thread to available cpu
core on the numa node.
NON_PMD_CORE_ID (currently 0) is used to reserve a particular
cpu core for the I/O of all non-pmd threads. No pmd thread
can be pinned to this reserved cor
The previous commit makes OVS create one tx queue for each
cpu core, each pmd thread will use a separate tx queue.
Also, tx of non-pmd threads on dpdk interface is all through
'NON_PMD_THREAD_TX_QUEUE', protected by the 'nonpmd_mempool_mutex'.
Therefore, the spinlock is no longer needed. And this
Previous commit makes OVS create one tx queue for each cpu
core. An upcoming patch will allow multiple pmd threads be
created and pinned to cpu cores. So each pmd thread will use
the tx queue corresponding to its core id.
Moreover, the pmd threads running on different numa node than
the dpdk int
Before this commit, ovs creates one tx and one rx queue for
each dpdk interface and uses only one poll thread for handling
I/O of all dpdk interfaces. An upcoming patch will allow multiple
poll threads be created. As a preparation, this commit changes
the dpif-netdev to create multiple tx/rx queu
This commit adds a new API to the 'struct netdev_class' which
allows user to configure the number of tx queues and rx queues
of 'netdev'. Upcoming patches will use this function to set
multiple tx/rx queues when adding the netdev to dpif-netdev.
Currently, only netdev-dpdk module implements this
On Fri, Sep 12, 2014 at 12:18 PM, Andy Zhou wrote:
> Yes, older distribution kernels may not follow the upstream kernel
> pattern. For example, centos-6.5-linux-2.6.32-431.
>
ok.
Acked-by: Pravin B Shelar
> On Thu, Sep 11, 2014 at 6:23 PM, Pravin Shelar wrote:
>> On Wed, Sep 10, 2014 at 3:58 PM
Hi Guru, I tried to use the /TP compilation flag. This resolved the issue of
the decltype operator but the __cplusplus reserved compilation variable was
automatically defined which caused different code segments of the OVS code to
be comiled.
I don't know of any other way to enable decltype w
My understating is the Cleanup callback is called when the file handle is
closed by user mode. However, at that time you can still have outstanding I/O
requests in the driver or even synchronous I/O requests sent to the driver from
a context of a different user mode thread. We do complete the ou
On Sep 12, 2014, at 12:46 PM, Eitan Eliahu
mailto:elia...@vmware.com>>
wrote:
Sorry for coming late on this one.
We should free the dump state when the system calls the driver on cleanup as
you did. But, the cleanup IOCTL can be (actually will be) executed from a
different thread context. This
Hi Nithin,
Sorry for coming late on this one.
We should free the dump state when the system calls the driver on cleanup as
you did. But, the cleanup IOCTL can be (actually will be) executed from a
different thread context. This means that we need to protect dumpState.ovsMsg.
Eitan
-Origina
Yes, older distribution kernels may not follow the upstream kernel
pattern. For example, centos-6.5-linux-2.6.32-431.
On Thu, Sep 11, 2014 at 6:23 PM, Pravin Shelar wrote:
> On Wed, Sep 10, 2014 at 3:58 PM, Andy Zhou wrote:
>> Current autoconfig detection logic for HAVE_PER_CPU_PTR is not robust
On Fri, Sep 12, 2014 at 11:22 AM, Eitan Eliahu wrote:
> /TP compilation flag can be used on these specific files which uses these
> macros. There should be no problem to mix standard C file with files that are
> compiled with the /TP flag. Unfortunately, I am not so familiar with
> controlling
Applied the first two patches to master.
Will rebase the following ones based on the discussion and comments.
On Thu, Sep 11, 2014 at 10:38 PM, Pravin Shelar wrote:
> On Tue, Sep 9, 2014 at 5:00 PM, Alex Wang wrote:
> > This commit adds new variable n_txq to 'struct netdev' for recording
> > t
On Fri, Sep 12, 2014 at 11:24:23AM -0700, Jarno Rajahalme wrote:
> Use the new OVS_FLOW_ATTR_PROBE flag when probing for datapath feature
> support. Suppress also dpif error logging when requested, as probe
> failures are already logged at ofproto-dpif.
>
> Signed-off-by: Jarno Rajahalme
> Signe
On Thu, Sep 11, 2014 at 11:12:29AM -0700, Jarno Rajahalme wrote:
> This could fix using ovs-dpctl to add userspace datapath flows.
>
> Signed-off-by: Jarno Rajahalme
s/prerequisities/prerequisites/ in a comment somewhere.
It looks like we need a new exception for the xregs in
odp_flow_key_to_fl
/TP compilation flag can be used on these specific files which uses these
macros. There should be no problem to mix standard C file with files that are
compiled with the /TP flag. Unfortunately, I am not so familiar with
controlling the build under mingw.
Thanks,
Eitan
-Original Message
This new flag is useful for suppressing error logging while probing
for datapath features using flow commands. For backwards
compatibility reasons the commands are executed normally, but error
logging is suppressed.
Signed-off-by: Jarno Rajahalme
---
v2: renamed SILENT as PROBE, added comments,
Use the new OVS_FLOW_ATTR_PROBE flag when probing for datapath feature
support. Suppress also dpif error logging when requested, as probe
failures are already logged at ofproto-dpif.
Signed-off-by: Jarno Rajahalme
Signed-off-by: Ben Pfaff
---
v2: renamed SILENT as PROBE.
lib/dpif-linux.c
If 'configure' with clang fails (this can happen for example if it doesn't
support all the cflags), simply skip it
Signed-off-by: Daniele Di Proietto
---
utilities/ovs-dev.py | 9 ++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/utilities/ovs-dev.py b/utilities/ovs-dev.py
--enable-dummy was useless anyway for ovsdb-server. Now it is an error to pass
it.
Signed-off-by: Daniele Di Proietto
---
utilities/ovs-dev.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/utilities/ovs-dev.py b/utilities/ovs-dev.py
index 5cf4771..869d89e 100755
--- a/u
Acked-by: Daniele Di Proietto
On 9/5/14, 2:20 PM, "Ethan Jackson" wrote:
>They have slightly different support characteristics, so it's nice to
>easily switch between them for testing.
>
>Signed-off-by: Ethan Jackson
>---
> utilities/ovs-dev.py | 18 ++
> 1 file changed, 14 inse
There¹s a minor issue with this patch: not every option is supported by
every compiler (for example, on my system, clang doesn¹t support -Og).
In this case ovs-dev.py fails to generate a Makefile.
I think we can address this in a separate commit. In the meantime:
Acked-by: Daniele Di Proietto
O
Applied to master with the changes mentioned in previous email.
Thanks,
On Fri, Sep 12, 2014 at 10:18 AM, Alex Wang wrote:
> Thx for pointing it out!
>
> Actually, I did 'perthread = pthread_getspecific(perthread_key);' instead
> of
> using 'ovsrcu_perthread_get()', which is wrong. Will fix it
Thx for pointing it out!
Actually, I did 'perthread = pthread_getspecific(perthread_key);' instead of
using 'ovsrcu_perthread_get()', which is wrong. Will fix it and remote the
call to ovsrcu_init_module().
On Fri, Sep 12, 2014 at 9:47 AM, Ben Pfaff wrote:
> On Tue, Sep 09, 2014 at 11:23:34A
Yes, it is C++ type specifier ( part "C++11" extensions) but we could compile
our C code to use it.
I came across a custom implementation of "typeof" (but C++ too):
http://www.boost.org/doc/libs/1_55_0/boost/typeof/msvc/typeof_impl.hpp
-Original Message-
From: Ben Pfaff [mailto:b...@nici
On Fri, Sep 12, 2014 at 04:54:52PM +, Eitan Eliahu wrote:
> Yes, it is C++ type specifier ( part "C++11" extensions) but we
> could compile our C code to use it.
I doubt OVS compiles cleanly as C++.
___
dev mailing list
dev@openvswitch.org
http://ope
On Thu, Sep 11, 2014 at 10:01:02AM -0700, Eitan Eliahu wrote:
> Use a specific out of band device control to subscribe/unsubscribe a socket
> to the driver event queue for notification.
>
> Signed-off-by: Eitan Eliahu
> Acked-by: Nithin Raju
> Acked-by: Saurabh Shah
> Acked-by: Ankur Sharma
>
On Wed, Sep 10, 2014 at 05:36:22PM -0700, Ankur Sharma wrote:
> Added APIs for creating and parsing nested netlink attributes.
> APIs are on similar lines as userspace netlink code.
>
> Signed-off-by: Ankur Sharma
> Acked-by: Nithin Raju
Applied, thanks!
On Wed, Sep 10, 2014 at 04:20:16PM -0700, Ankur Sharma wrote:
> Added an inline function to return used size in the buffer.
>
> Signed-off-by: Ankur Sharma
> Acked-by: Nithin Raju
Applied, thanks.
___
dev mailing list
dev@openvswitch.org
http://openvs
Ben,
Are you happy with this? I really see this as something we could have done at
the time we introduced datapath flow key masks (megaflows), but did not. The
basic change is that we allow datapath flow setup without requiring key fields
that are fully wildcarded to be present.
Jarno
On Se
On Tue, Sep 09, 2014 at 11:23:34AM -0700, Alex Wang wrote:
> On current master, the per-thread callback event set is flushed
> when ovsrcu_quiesce_start() is called or when the callback
> event set is full. For threads that only call 'ovsrcu_quiesce()'
> to indicate quiescient state, their callbac
On Thu, Sep 11, 2014 at 11:29 PM, Alex Wang wrote:
> You mean netdev_open(), or rxq_open() or some function else?
>
I was referring to netdev_open().
> netdev_open() seems not a good place, it can be called at multiple
> places. and we need to keep record of the n_rxq config at some high
> level
On Sep 11, 2014, at 11:20 PM, Pravin Shelar wrote:
> On Thu, Sep 11, 2014 at 10:43 PM, Alex Wang wrote:
>> One example is when STP config pkts are received from the dpdk port,
>>
>> when processing the pkts in xlate's process_special(), if the stp state
>> machine need to send a config immedia
On Fri, Sep 12, 2014 at 1:22 AM, Philipp S. Schmidt wrote:
>
> The patch below fixes the problem for me - Thanks
I applied the patch to master branch and 2.3 branch of OVS repo found here:
https://github.com/openvswitch/ovs
I suppose Ben will eventually upload it to debian.
_
> ((LONG)(LONG_PTR)&((decltype(OBJECT))0)->MEMBER)
> I think it is the right way to implement it. But, this keyword "decltype" is
> ignored by the compiler when above macro is used in the OVS source base. I
> don't know the reason for it but it might be related to custom compilation
> opt
On Fri, Sep 12, 2014 at 02:11:00PM +, Eitan Eliahu wrote:
> [2] The usage of the decltype does work with of a built in VS project I
> created:
> #define OBJECT_OFFSETOF(OBJECT, MEMBER) \
> ((LONG)(LONG_PTR)&((decltype(OBJECT))0)->MEMBER)
> I think it is the right way to implement it. Bu
[1] The local warning suppression added to the macro does not work as the
warning suppression is effected only after the complier encounter the first
curly brace "{" (which does not appear in this specific MACRO)
#define OBJECT_OFFSETOF(OBJECT, MEMBER) __pragma(warning(push)
__pragma(warning(d
Hi,
Would you be interested in *Rackspace *users list for your Sale/Marketing
campaigns?
Information Fields – Name, Title, Email, Company Name, and Company Details
like, Physical Address, Web Address, Revenue Size, Employee Size and
industry.
We also have other Software/Hardware users like
Hi All,
We are working on the implementation of eviction on the basis of importance in
accordance to the OpenFlow Specs 1.4.Also we are using Open vSwitch 2.3.0
stable version downloaded from
"http://openvswitch.org/releases/openvswitch-2.3.0.tar.gz"; for development.
The following changes h
Hi Ben/Team,
> I would enable eviction if it is configured through the database or
through the OpenFlow connection.
In current implementation, to enable eviction using create table ...
overflow-policy=evict, oftable_enable_eviction() in ofproto.c is called
which sets table->eviction fields.
I
As the first packet of a new flow goes to ovs-vswitchd, we are tracing where
the packet is parsed in the source code of ovs-vswitchd. We found the packet is
copied to user_skb in datapath.c:queue_userspace_packet, and then passed
to the user space by genlmsg_unicast. In the user space, we guess
dp
The patch below fixes the problem for me - Thanks
On 11.09.2014, at 18:43, Gurucharan Shetty wrote:
> Commit b2a0daa5bd (debian: Don't recreate bridges during manual restart.)
> added a check on $RUNLEVEL to only create bridges and ports when the
> system starts up. This fix does not work with
depends on this patch:
http://openvswitch.org/pipermail/dev/2014-September/045617.html
On Fri, Sep 12, 2014 at 12:14 AM, Alex Wang wrote:
> This patch is a RFC for the bash command-line completion script
> for ovs-appctl command. Right now, the script can do the following:
>
> - accept user sp
This patch is a RFC for the bash command-line completion script
for ovs-appctl command. Right now, the script can do the following:
- accept user specified '--target' and query available completions
from the corresponding deamons.
- once the subcommand (e.g. ofproto/trace) has been given, the
48 matches
Mail list logo
