Alex, could you review this series?
I think that the correct solution is to perform retroactive side-effects,
that is, perform side effects as though they had happened at the time that
the flow was hit. Anything that may hide these effects (like mac table
flush) would need to be taken into
This reverts commit a48c85b2d672505b89e488d28066538705b94942. The commit
was causing intermittent testsuite failures and unexpected re-install of
stale mac-learning entries.
VMware-BZ: 1268574
Reported-by: Len Gao l...@vmware.com
Signed-off-by: Joe Stringer joestrin...@nicira.com
---
I'd like to
On second thought, I noticed that this doesn't fix odd behaviour in the
netflow expiration test.
I sent out a patch to revert the original patch:
http://openvswitch.org/pipermail/dev/2014-June/041868.html
On 23 June 2014 18:29, Joe Stringer joestrin...@nicira.com wrote:
Alex, could you review
DO YOU NEED LOANS? AT 3% IF YES EMAIL WITH AMOUNT NEEDED. Note all
reply should be copy and send to this email
tracyharriso...@gmail.commailto:tracyharriso...@gmail.com
tracyharriso...@gmail.commailto:tracyharriso...@gmail.com
___
dev mailing list
Thanks.
-Anoob
On 09/06/14 19:10, Ben Pfaff wrote:
Sorry about the delay. I took a few minutes to properly understand
the situation this morning. I applied the following form of your
patch to master and branch-2.[123]:
--8--cut here--8--
From:
Signed-off-by: Lorand Jakab loja...@cisco.com
---
datapath/linux/.gitignore | 1 +
1 file changed, 1 insertion(+)
diff --git a/datapath/linux/.gitignore b/datapath/linux/.gitignore
index d10d028..be233fc 100644
--- a/datapath/linux/.gitignore
+++ b/datapath/linux/.gitignore
@@ -41,6 +41,7 @@
On Thu, Jun 19, 2014 at 09:20:57PM -0700, Alex Wang wrote:
This patch adds a new configuration option, max-idle to the
Bridge other-config column. This sets how long datapath flows,
for the configured bridge, are cached in the datapath before
ovs-vswitchd thread expires them.
This commit is
GCC 4.9.0 triggers a warning (array-bounds) while compiling test-classifier.c
This commit introduces an assertion that suppresses the warning.
Signed-off-by: Daniele Di Proietto ddiproie...@vmware.com
---
While I feel that this is a GCC problem, this simple (IMHO harmful) assertion
fixes the
Hey Joe,
Sorry for the delayed reply,
Please see my comments inline,
On Sun, Jun 22, 2014 at 11:29 PM, Joe Stringer joestrin...@nicira.com
wrote:
Alex, could you review this series?
I think that the correct solution is to perform retroactive
side-effects, that is, perform side effects as
Hey Ben,
Thanks for pointing it out, I actually do not have a clear judge on this.
I think it is necessary in this case,
For branch-2.0, we are before the revalidator-as-a-thread era. So, it is
pretty
different from the implementation in commit 72310b04.
For branch-1.11, 1.10, we need to
OK. I understand now. Do you still need any reviews?
On Mon, Jun 23, 2014 at 10:30 AM, Alex Wang al...@nicira.com wrote:
Hey Ben,
Thanks for pointing it out, I actually do not have a clear judge on this.
I think it is necessary in this case,
For branch-2.0, we are before the
Acked-by: Alex Wang al...@nicira.com
On Mon, Jun 23, 2014 at 12:52 AM, Joe Stringer joestrin...@nicira.com
wrote:
This reverts commit a48c85b2d672505b89e488d28066538705b94942. The commit
was causing intermittent testsuite failures and unexpected re-install of
stale mac-learning entries.
On Sun, Jun 22, 2014 at 5:30 PM, Daniele Di Proietto
ddiproie...@vmware.com wrote:
The netdev_send function has been modified to accept multiple packets, to
allow netdev providers to amortize locking and queuing costs.
This is especially true for netdev-dpdk.
Later commits exploit the new
On Sun, Jun 22, 2014 at 5:30 PM, Daniele Di Proietto
ddiproie...@vmware.com wrote:
This commit introduces a new data structure used for receiving packets from
netdevs and passing them to dpifs.
The purpose of this change is to allow storing some private data for each
packet. The subsequent
Thx, I already pushed them,
Forgot to mention in this thread.
Thanks,
Alex Wang,
On Mon, Jun 23, 2014 at 10:33 AM, Ben Pfaff b...@nicira.com wrote:
OK. I understand now. Do you still need any reviews?
On Mon, Jun 23, 2014 at 10:30 AM, Alex Wang al...@nicira.com wrote:
Hey Ben,
Thanks
On Sun, Jun 22, 2014 at 5:30 PM, Daniele Di Proietto
ddiproie...@vmware.com wrote:
This change in dpif-netdev allows faster packet processing for devices which
implement batching (netdev-dpdk currently).
Signed-off-by: Daniele Di Proietto ddiproie...@vmware.com
You have defined struct
On Fri, Jun 13, 2014 at 05:00:40PM +1200, Joe Stringer wrote:
This allows us to remove some of the sleeps from the testsuite.
Signed-off-by: Joe Stringer joestrin...@nicira.com
---
There are still two other common cases that require sleeps:
* Waiting for handler threads to create flows.
*
On Jun 23, 2014, at 10:35 AM, Pravin Shelar pshe...@nicira.com wrote:
On Sun, Jun 22, 2014 at 5:30 PM, Daniele Di Proietto
ddiproie...@vmware.com wrote:
This commit introduces a new data structure used for receiving packets from
netdevs and passing them to dpifs.
The purpose of this change
On Fri, Jun 13, 2014 at 07:40:00AM -0700, Gurucharan Shetty wrote:
WSAStartup() needs to be called before using winsock2 related
functions. We need this for almost all the utilities. So call
it through OVS_CONSTRUCTOR.
Signed-off-by: Gurucharan Shetty gshe...@nicira.com
Did you make sure
I get the following sparse warnings on current master:
datapath/linux/flow_table.c:289:26: error: incompatible types in comparison
expression (different address spaces)
datapath/linux/flow_table.c:300:26: error: incompatible types in comparison
expression (different address spaces)
On Jun 23, 2014, at 10:36 AM, Pravin Shelar pshe...@nicira.com wrote:
On Sun, Jun 22, 2014 at 5:30 PM, Daniele Di Proietto
ddiproie...@vmware.com wrote:
This change in dpif-netdev allows faster packet processing for devices which
implement batching (netdev-dpdk currently).
Signed-off-by:
On Fri, Jun 13, 2014 at 07:40:01AM -0700, Gurucharan Shetty wrote:
After the change, both of them compile. test-netflow related
unit tests pass.
test-sflow related tests do not pass because
of LOOPBACK_INTERFACE constraints for 'agent'.
(It should be revisited later.)
Signed-off-by:
On Fri, Jun 13, 2014 at 07:40:02AM -0700, Gurucharan Shetty wrote:
There is no 'kill -l' type functionality available on Windows.
So instead of looking for the string 'ABRT', check for the exit
code in both platforms. On msys (unit test environment), it is 9
and on Linux, it is 134 (SIGABRT +
The netdev_send function has been modified to accept multiple packets, to
allow netdev providers to amortize locking and queuing costs.
This is especially true for netdev-dpdk.
Later commits exploit the new API.
Signed-off-by: Daniele Di Proietto ddiproie...@vmware.com
---
lib/dpif-netdev.c
This commit introduces a new data structure used for receiving packets from
netdevs and passing them to dpifs.
The purpose of this change is to allow storing some private data for each
packet. The subsequent commits make use of it.
Signed-off-by: Daniele Di Proietto ddiproie...@vmware.com
---
Changes from v7:
renamed ofp to buf in dp_netdev_input()
define struct netdev_flow_key globally in lib/dpif-netdev.c
Changes from v6:
free_dpdk_buf should accept dpif_packet also if DPDK_NETDEV is not defined
Changes from v5:
Pravin's suggestions:
reintroduced dpif_packet_delete()
This change in dpif-netdev allows faster packet processing for devices which
implement batching (netdev-dpdk currently).
Signed-off-by: Daniele Di Proietto ddiproie...@vmware.com
---
lib/dpif-netdev.c| 340 +--
lib/dpif.c |
On Mon, Jun 23, 2014 at 3:48 AM, Lorand Jakab loja...@cisco.com wrote:
Signed-off-by: Lorand Jakab loja...@cisco.com
---
datapath/linux/.gitignore | 1 +
1 file changed, 1 insertion(+)
Applied, thanks.
___
dev mailing list
dev@openvswitch.org
Shared memory ring patch
This patch enables the client dpdk rings within the netdev-dpdk. It adds
a new dpdk device called dpdkr (other naming suggestions?). This allows
for the use of shared memory to communicate with other dpdk applications,
on the host or within a virtual machine.
With this series of patches, ovs will create multiple rx queues for
each dpdk iface. Each of rx queue will be handled by the pmd thread
on the same cpu socket. For tx, each dpdk iface will have a tx queue
for each cpu core. So, different pmd threads will not contend with
each other over the tx
Signed-off-by: Alex Wang al...@nicira.com
---
lib/automake.mk |2 +
lib/ovs-numa.c | 208 +++
lib/ovs-numa.h | 33
tests/ofproto-macros.at |1 +
vswitchd/bridge.c |2 +
5 files changed, 246
The pmd threads are pinned to available cpu cores on the
corresponding cpu socket.
Signed-off-by: Alex Wang al...@nicira.com
---
lib/dpif-netdev.c | 172 ++---
lib/netdev-dpdk.c | 27 -
lib/netdev-dpdk.h | 14 +
3 files changed,
This commit makes the memory pool name contain the socket id.
Since dpdk library do not allow creation of memory pool with
same name, this commit serves as a simple way of making each
name unique.
Signed-off-by: Alex Wang al...@nicira.com
---
lib/netdev-dpdk.c |3 ++-
1 file changed, 2
Signed-off-by: Alex Wang al...@nicira.com
---
lib/netdev-dpdk.c | 14 ++
lib/netdev-dpdk.h | 11 +++
2 files changed, 25 insertions(+)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index c43049a..0e91b43 100644
--- a/lib/netdev-dpdk.c
+++ b/lib/netdev-dpdk.c
@@
Previously, flows were retrieved one by one when dumping flows for
datapaths of type 'netdev'. This increased contention for the dump's
mutex, negatively affecting revalidator performance.
This patch retrieves batches of flows when dumping flows for datapaths
of type 'netdev'.
Signed-off-by:
This will be the default setting for dpdk iface.
Signed-off-by: Alex Wang al...@nicira.com
---
lib/netdev-dpdk.c | 47 ---
1 file changed, 40 insertions(+), 7 deletions(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index e2ce1ea..6268c10
___Excuse to talk about something. Since we can understand what.
1Hxy6IXi8GTHP-S4Q6dU35AFjcL26zIMeTwYK wSMcEXxD6NIgDC1AGT6UIlOyOdNR9S1 N7PFRq9OER7 TCH8qE0 53fBL5E1yQS9xTSq jPVvR7YIQ8YC21E0!Shouted abby
On Fri, Jun 13, 2014 at 07:40:03AM -0700, Gurucharan Shetty wrote:
When a OVS daemon is configured to run as a Windows service,
when the service is stopped by calling service_stop(), the
windows services manager does not give enough time to do
everything in the atexit handler. So call the exit
On Fri, Jun 13, 2014 at 07:40:04AM -0700, Gurucharan Shetty wrote:
basename() and dir_name() are not used for Windows and won't work well if
used. So put a '#ifndef _WIN32' around them to prevent future calls.
test-file_name.c tests the above 2 functions. It makes sense to merge
this single
On Fri, Jun 13, 2014 at 07:40:05AM -0700, Gurucharan Shetty wrote:
As of now, when a process tries to reacquire a lockfile, we
return EEXIST on Windows and print a different error message.
This means that the unit tests need to look for different error
messages too. Linux uses EDEADLK for the
On Fri, Jun 13, 2014 at 07:40:06AM -0700, Gurucharan Shetty wrote:
Most of the Python code in Open vSwitch repo does not work
on Windows. Till that porting is done, simply skip all
Python tests.
Signed-off-by: Gurucharan Shetty gshe...@nicira.com
---
v1-v2:
* Abandoned commits in v1 that
On Fri, Jun 13, 2014 at 12:01:50PM -0700, Neil McKee wrote:
Standard LACP counters are added to the LACP module, and
the sFlow library and test modules are extended to support the
export of those LACP counters as well as tunnel and OpenFlow
related structures. None of these structures are
I like this better, If you send me signed off line I will merge it.
On Thu, Jun 19, 2014 at 5:27 PM, Ryan Wilson 76511 wr...@vmware.com wrote:
This patch will fix the naming issue, too. But yea ideally I'd like to merge
something soon so we can have this fixed.
diff --git a/lib/netdev-dpdk.c
From: Daniele Di Proietto ddiproie...@vmware.com
rte_eal_init() returns the number of parsed dpdk arguments to skip.
dpdk_init() should add 1 to that number, because it has already skipped
the --dpdk argument itself.
This patch also makes sure the program name is ovs-vswitchd in
rte_eal_init()
Sent out a new version of the patch with the correct commit message and
signed-off-bys.
Ryan
On 6/23/14 1:57 PM, Pravin Shelar pshe...@nicira.com wrote:
I like this better, If you send me signed off line I will merge it.
On Thu, Jun 19, 2014 at 5:27 PM, Ryan Wilson 76511 wr...@vmware.com
On Fri, Jun 13, 2014 at 02:58:32PM -0700, Daniele Di Proietto wrote:
This commit intruduces a new appctl command: dpif/dpctl
s/intruduces/introduces/
It's needed to interact with userspace datapaths (dpif-netdev), because the
ovs-dpctl command runs in a separate process and cannot see the
On Sat, Jun 14, 2014 at 08:33:13AM +0900, Simon Horman wrote:
Previously the out_port of a flow monitor was
checked in ofmonitor_report() using ofoperation_has_out_port().
When ofoperation_has_out_port() was removed so was the call to
it in ofmonitor_report() thus flow monitor updates are
On Mon, Jun 23, 2014 at 2:11 PM, Ryan Wilson wr...@nicira.com wrote:
From: Daniele Di Proietto ddiproie...@vmware.com
rte_eal_init() returns the number of parsed dpdk arguments to skip.
dpdk_init() should add 1 to that number, because it has already skipped
the --dpdk argument itself.
This
Thanks for fixing all issues. I pushed this series to master.
On Mon, Jun 23, 2014 at 11:43 AM, Daniele Di Proietto
ddiproie...@vmware.com wrote:
Changes from v7:
renamed ofp to buf in dp_netdev_input()
define struct netdev_flow_key globally in lib/dpif-netdev.c
Changes from v6:
On Mon, Jun 16, 2014 at 10:09:38AM +0900, Simon Horman wrote:
In some cases an pop MPLS action changes a packet to be a non-mpls packet.
In this case subsequent any L3+ actions require access to portions
of the packet which were not decoded as they were opaque when the
packet was MPLS. Allow
On Mon, Jun 16, 2014 at 10:09:39AM +0900, Simon Horman wrote:
With recirculation in place this should be safe.
Signed-off-by: Simon Horman ho...@verge.net.au
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
On Mon, Jun 16, 2014 at 10:09:40AM +0900, Simon Horman wrote:
Co-Authored-by: Joe Stringer j...@wand.net.nz
Signed-off-by: Simon Horman ho...@verge.net.au
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
On Mon, Jun 16, 2014 at 11:29:21AM +0900, Simon Horman wrote:
Use enum ofp14_flow_monitor_flags in place of enum nx_flow_monitor_flags in
struct ofputil_flow_monitor_request and struct rule, translating between
the two enumerations as necessary.
The motivation for this patch is to move
On Mon, Jun 16, 2014 at 11:29:22AM +0900, Simon Horman wrote:
If the ONLY_OWN flag of a flow monitor is set then ofmonitor_report should
skip sending updates to other controllers.
This is in preparation for supporting OpenFlow1.4 flow monitors.
Signed-off-by: Simon Horman
On Mon, Jun 16, 2014 at 11:29:23AM +0900, Simon Horman wrote:
This is in preparation for supporting OpenFlow flow monitor request
commands other than OFPFMC_ADD.
Signed-off-by: Simon Horman ho...@verge.net.au
---
v2
* No change
Acked-by: Ben Pfaff b...@nicira.com
On Mon, Jun 16, 2014 at 11:29:24AM +0900, Simon Horman wrote:
This is in preparation for supporting OpenFlow1.4 flow monitors
Signed-off-by: Simon Horman ho...@verge.net.au
I'd squash that with the previous patch.
___
dev mailing list
Thanks, I pushed this to master. I'll follow up on the original discussion
in the other thread.
On 24 June 2014 05:35, Alex Wang al...@nicira.com wrote:
Acked-by: Alex Wang al...@nicira.com
On Mon, Jun 23, 2014 at 12:52 AM, Joe Stringer joestrin...@nicira.com
wrote:
This reverts commit
On Mon, Jun 16, 2014 at 11:29:25AM +0900, Simon Horman wrote:
This is in preparation for adding out_group support to flow monitors
which is in turn in preparation for supporting OpenFlow1.4 flow monitors.
Signed-off-by: Simon Horman ho...@verge.net.au
Acked-by: Ben Pfaff b...@nicira.com
On Mon, Jun 16, 2014 at 11:29:26AM +0900, Simon Horman wrote:
Filter updates based on the out_group of a flow monitor.
This is analogous to the existing treatment of out_port.
This is in preparation for supporting OpenFlow1.4 flow monitors.
Signed-off-by: Simon Horman ho...@verge.net.au
I
On Mon, Jun 16, 2014 at 11:29:27AM +0900, Simon Horman wrote:
Filter collection of rules for flow monitor updates
based on the monitor's out_group as well as its out_port.
Currently the flow monitor request decoder only allows the out_group to be
OFPG_ANY which means that this logic change
On 06/23/14 at 12:37pm, Alex Wang wrote:
Signed-off-by: Alex Wang al...@nicira.com
Some comments below
---
lib/automake.mk |2 +
lib/ovs-numa.c | 208
+++
lib/ovs-numa.h | 33
tests/ofproto-macros.at |
On Mon, Jun 23, 2014 at 02:38:50PM -0700, Ben Pfaff wrote:
On Sat, Jun 14, 2014 at 08:33:13AM +0900, Simon Horman wrote:
Previously the out_port of a flow monitor was
checked in ofmonitor_report() using ofoperation_has_out_port().
When ofoperation_has_out_port() was removed so was the
On Mon, Jun 23, 2014 at 02:50:20PM -0700, Ben Pfaff wrote:
On Mon, Jun 16, 2014 at 10:09:38AM +0900, Simon Horman wrote:
In some cases an pop MPLS action changes a packet to be a non-mpls packet.
In this case subsequent any L3+ actions require access to portions
of the packet which were not
On Mon, Jun 23, 2014 at 3:55 PM, Simon Horman ho...@verge.net.au wrote:
On Mon, Jun 23, 2014 at 02:50:20PM -0700, Ben Pfaff wrote:
On Mon, Jun 16, 2014 at 10:09:38AM +0900, Simon Horman wrote:
In some cases an pop MPLS action changes a packet to be a non-mpls packet.
In this case subsequent
On 06/23/14 at 12:37pm, Alex Wang wrote:
This will be the default setting for dpdk iface.
Signed-off-by: Alex Wang al...@nicira.com
I think you can remove the NR_QUEUE #define in lib/dpif-netdev.h
with this patch.
---
lib/netdev-dpdk.c | 47 ---
On 06/23/14 at 12:37pm, Alex Wang wrote:
The pmd threads are pinned to available cpu cores on the
corresponding cpu socket.
Signed-off-by: Alex Wang al...@nicira.com
---
lib/dpif-netdev.c | 172
++---
lib/netdev-dpdk.c | 27 -
The motivation of this series is to allow some sequences of actions
that include MPLS actions to be performed using recirculation.
Sequences of actions that could not previously be handled.
For example pop_mpls:0x0800,dec_ttl.
This uses recirculation after the pop_mpls action to allow access to
With recirculation in place this should be safe.
Signed-off-by: Simon Horman ho...@verge.net.au
Acked-by: Ben Pfaff b...@nicira.com
---
v8
* Add Ack from Ben Pfaff
v2 - v7
* No change
---
lib/ofp-actions.c | 6 --
1 file changed, 6 deletions(-)
diff --git a/lib/ofp-actions.c
In some cases an pop MPLS action changes a packet to be a non-mpls packet.
In this case subsequent any L3+ actions require access to portions
of the packet which were not decoded as they were opaque when the
packet was MPLS. Allow such actions to be translated by
first recirculating the packet.
On Mon, Jun 23, 2014 at 03:58:28PM -0700, Ben Pfaff wrote:
On Mon, Jun 23, 2014 at 3:55 PM, Simon Horman ho...@verge.net.au wrote:
On Mon, Jun 23, 2014 at 02:50:20PM -0700, Ben Pfaff wrote:
On Mon, Jun 16, 2014 at 10:09:38AM +0900, Simon Horman wrote:
In some cases an pop MPLS action
On Mon, Jun 16, 2014 at 11:29:28AM +0900, Simon Horman wrote:
Use a list rather than an array to track monitor requests
in handle_flow_monitor_request().
This is in preparation for supporting OpenFlow1.4 flow monitor requests
with delete and modify commands.
Signed-off-by: Simon Horman
Found by sparse.
Signed-off-by: Ben Pfaff b...@nicira.com
---
lib/dpif-netdev.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
index 33f1c15..4dcc268 100644
--- a/lib/dpif-netdev.c
+++ b/lib/dpif-netdev.c
@@ -215,7 +215,7 @@ struct
On Mon, Jun 16, 2014 at 11:29:29AM +0900, Simon Horman wrote:
Break out monitor deletion code into a new function
flow_monitor_delete which is paramatised over the id of
the monitor to delete.
This is in preparation for supporting OpenFlow1.4 flow monitor requests
with delete and modify
On Mon, Jun 16, 2014 at 11:29:30AM +0900, Simon Horman wrote:
Handle modify and delete commands in OpenFlow1.4 flow monitor requests.
These commands are not yet allowed by the decoder which
will be updated by a subsequent patch.
Signed-off-by: Simon Horman ho...@verge.net.au
---
v2
* No
On Sun, Jun 22, 2014 at 5:42 PM, Alessandro Pilotti
apilo...@cloudbasesolutions.com wrote:
In addition to the code repositories made previously available, here’s a
wiki
document describing our OVS Hyper-V porting architecture and
implementation:
On 24 June 2014 05:26, Alex Wang al...@nicira.com wrote:
(snip)
Totally agree,
So, I'm thinking this solution,
- in revalidate_ukey(), if need_revalidate is set, we always:
1. save the current xcache to old_xcache pointers.
2. conduct full revalidation (but never push stats during
On Mon, Jun 23, 2014 at 6:10 PM, Joe Stringer joestrin...@nicira.com
wrote:
On 24 June 2014 05:26, Alex Wang al...@nicira.com wrote:
(snip)
Totally agree,
So, I'm thinking this solution,
- in revalidate_ukey(), if need_revalidate is set, we always:
1. save the current xcache to
77 matches
Mail list logo