Hi,
On some of my systems the commit 0c47331 ofproto-dpif.at: Wait for the
monitor's pidfile disappears where necessary consistently breaks test
700: ofproto-dpif - controller. Reverting the commit makes the test
pass again. This happens on a physical machine (Sandy Bridge laptop
with
I've been looking at some races in the megaflow testcases recently as well,
with a slightly different approach, could you a look at my patch below?
http://openvswitch.org/pipermail/dev/2014-April/038536.html
my patch inserts sleeps between two netdev-dummy/receive calls,
where it's expected
Uniformly wait for calls to ovs-ofctl exit in the test-suite.
Signed-off-by: Simon Horman ho...@verge.net.au
---
tests/ofproto-dpif.at | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/tests/ofproto-dpif.at b/tests/ofproto-dpif.at
index f2ddda9..6f70ba0 100644
---
This corrects an inconsistency in the MPLS tests and
avoids a race by waiting for sufficient output before
checking its contents
Signed-off-by: Simon Horman ho...@verge.net.au
---
tests/ofproto-dpif.at | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/tests/ofproto-dpif.at
A call to ofpbuf_put() may cause the data of the passed to be resized and
its base address to change. Thus the address returned by ofpbuf_put()
should be used as the base address rather than relying on the base address
prior to ofpbuf_put().
This avoids the following assertion in
Thanks for catching this, Simon.
Pushed to master,
Jarno
Acked-by: Jarno Rajahalme jrajaha...@nicira.com
On Apr 7, 2014, at 1:43 AM, Simon Horman ho...@verge.net.au wrote:
A call to ofpbuf_put() may cause the data of the passed to be resized and
its base address to change. Thus the
On Sun, Apr 6, 2014 at 6:11 PM, sowmini varadhan sowmin...@gmail.com wrote:
I've built and booted up a kernel built from a recent-ish kernel.git.
Uname -r reports: 3.14.0-rc3+
Then I try to set up openvswitch 1.10.2 thus:
./configure --with-linux=/lib/modules/`uname -r`/build
make
but
On Mon, Apr 7, 2014 at 9:36 AM, sowmini varadhan sowmin...@gmail.com wrote:
All of these versions of OVS long predate the kernel that you are
trying to use, so they don't support it. However, you should just be
able to use the version of the OVS kernel module packaged with the
upstream kernel
Acked-by: Alex Wang al...@nicira.com
On Tue, Mar 25, 2014 at 3:51 PM, Ben Pfaff b...@nicira.com wrote:
Otherwise you get a fairly obscure message about a lockfile already being
locked.
Signed-off-by: Ben Pfaff b...@nicira.com
---
ovsdb/ovsdb-server.c | 40
Acked-by: Alex Wang al...@nicira.com
On Tue, Mar 25, 2014 at 3:51 PM, Ben Pfaff b...@nicira.com wrote:
Resource deadlock avoided is a pretty obscure message.
Signed-off-by: Ben Pfaff b...@nicira.com
---
lib/lockfile.c|6 +-
tests/lockfile.at | 20 ++--
2
On Sun, Apr 06, 2014 at 11:08:18AM +0900, YAMAMOTO Takashi wrote:
25 kB is about 25000 bytes, not 2500.
Signed-off-by: Ben Pfaff b...@nicira.com
Acked-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Thanks for the acks, I applied both of these to master.
Thanks for the reviews. I applied both of these to master.
On Mon, Apr 07, 2014 at 09:53:32AM -0700, Alex Wang wrote:
Acked-by: Alex Wang al...@nicira.com
On Tue, Mar 25, 2014 at 3:51 PM, Ben Pfaff b...@nicira.com wrote:
Otherwise you get a fairly obscure message about a lockfile
Pravin, this patch includes a change to linux/openvswitch.h, will
you look at that part?
On Thu, Mar 27, 2014 at 03:38:57PM +0900, YAMAMOTO Takashi wrote:
Wrap long lines, fix whitespaces, and fix a typo in a comment.
No functional changes are intended.
Cc: Andy Zhou az...@nicira.com
On Thu, Mar 27, 2014 at 03:38:57PM +0900, YAMAMOTO Takashi wrote:
Wrap long lines, fix whitespaces, and fix a typo in a comment.
No functional changes are intended.
Cc: Andy Zhou az...@nicira.com
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
For files other than
On Sun, Apr 06, 2014 at 11:30:03AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sun, Apr 06, 2014 at 11:30:05AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sun, Apr 06, 2014 at 11:30:07AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sun, Apr 06, 2014 at 11:30:04AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sun, Apr 06, 2014 at 11:30:12AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sun, Apr 06, 2014 at 11:30:08AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sun, Apr 06, 2014 at 11:30:09AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sun, Apr 06, 2014 at 11:30:11AM +0900, YAMAMOTO Takashi wrote:
Also, add a comment to explain the expected sequences.
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
On Sun, Apr 06, 2014 at 11:30:10AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sun, Apr 06, 2014 at 11:30:13AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
Thanks a lot for fixing all these tests! Most of them do not
regularly fail for me, but I can believe that they do for you in a
different
On Mon, Apr 07, 2014 at 05:39:38PM +0900, Simon Horman wrote:
Uniformly wait for calls to ovs-ofctl exit in the test-suite.
Signed-off-by: Simon Horman ho...@verge.net.au
This inserts two OVS_APP_EXIT_AND_WAITs and two OVS_WAIT_UNTILs. The
latter do not actually wait for ovs-ofctl to exit.
On Wed, Mar 26, 2014 at 11:38 PM, YAMAMOTO Takashi
yamam...@valinux.co.jp wrote:
Wrap long lines, fix whitespaces, and fix a typo in a comment.
No functional changes are intended.
Cc: Andy Zhou az...@nicira.com
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
for openvswitch.h.
On Mon, Apr 07, 2014 at 05:41:50PM +0900, Simon Horman wrote:
This corrects an inconsistency in the MPLS tests and
avoids a race by waiting for sufficient output before
checking its contents
Signed-off-by: Simon Horman ho...@verge.net.au
Yamamoto-san, this seems to be close to some of your
On Thu, Apr 03, 2014 at 05:52:01PM -0700, Pravin Shelar wrote:
On Thu, Apr 3, 2014 at 3:09 PM, Ben Pfaff b...@nicira.com wrote:
I don't think that ofproto_flow_lookup() modifies its 'flow' argument,
but dp_handle_miss() makes a writable copy and passes in that copy. I
think that we could
On Thu, Apr 03, 2014 at 05:47:04PM -0700, Pravin Shelar wrote:
On Thu, Apr 3, 2014 at 2:41 PM, Ben Pfaff b...@nicira.com wrote:
I don't like seeing this as a global variable. Can it be per-datapath
(probably set using a new provider callback function)?
There can be only one definition of
On 7 April 2014 20:12, YAMAMOTO Takashi yamam...@valinux.co.jp wrote:
I've been looking at some races in the megaflow testcases recently as
well,
with a slightly different approach, could you a look at my patch below?
http://openvswitch.org/pipermail/dev/2014-April/038536.html
my patch
From: Pravin Shelar pshe...@nicira.com
Ths simplifies flow-table-destroy API. This change is required
for following patches.
Signed-off-by: Pravin B Shelar pshe...@nicira.com
---
datapath/datapath.c |5 ++---
datapath/flow_table.c |6 +++---
datapath/flow_table.h |2 +-
3 files
From: Pravin Shelar pshe...@nicira.com
Along with flow-table rehashing OVS can compact masks array. This
allows us to calculate highest index for mask array.
Signed-off-by: Pravin B Shelar pshe...@nicira.com
---
datapath/flow_table.c | 24
datapath/flow_table.h |
From: Pravin Shelar pshe...@nicira.com
mask caches index of mask in mask_list. On packet recv OVS
need to traverse mask-list to get cached mask. Therefore array
is better for retrieving cached mask. This also allows better
cache replacement algorithm by directly checking mask's existence.
From: Pravin Shelar pshe...@nicira.com
On every packet OVS needs to lookup flow-table with every mask.
the packet flow-key is first masked with mask in the list and
then the masked key is looked up in flow-table. Therefore number
of masks can affect packet processing performance.
Following
Flow statistics need to take into account the TCP flags from the packet
currently being processed (in 'key'), not the TCP flags matched by the
flow found in the kernel flow table (in 'flow').
This bug made the Open vSwitch userspace fin_timeout action have no effect
in many cases.
Bug #1219516.
Hi All,
We are facing issues in deleting bridge after creating group in OVS 2.1.0
release.
It appears like the issue is in ofproto_destroy function:
=
static void
ofproto_destroy__(struct ofproto *ofproto)
OVS_EXCLUDED(ofproto_mutex)
{
struct
On Mon, Apr 07, 2014 at 11:19:17AM -0700, Ben Pfaff wrote:
On Mon, Apr 07, 2014 at 05:39:38PM +0900, Simon Horman wrote:
Uniformly wait for calls to ovs-ofctl exit in the test-suite.
Signed-off-by: Simon Horman ho...@verge.net.au
This inserts two OVS_APP_EXIT_AND_WAITs and two
On Fri, Apr 04, 2014 at 04:03:02PM -0700, Andy Zhou wrote:
Infrastructure to enable megaflow support for bond ports using
recirculation. This patch adds the following features:
* Generate RECIRC action when bond can benefit from recirculation.
* Populate post recirculation rules in a hidden
On Thu, Apr 03, 2014 at 09:02:48PM -0700, Alex Wang wrote:
Commit 6b59b543 (ovs-thread: Use fair (but nonrecursive)
rwlocks on glibc.) changed the rwlocks to nonrecursive,
writer-biased lock. It also made the following deadlock
situation possible.
Assume BFD is used on both end of a link.
Yes, I can reproduce it, for example I have 2 vms, and I configure 2K bfd
monitored tunnel together at the
same time. And then, I can see one end will be deadlocked.
On Mon, Apr 7, 2014 at 5:20 PM, Ben Pfaff b...@nicira.com wrote:
On Thu, Apr 03, 2014 at 09:02:48PM -0700, Alex Wang wrote:
[ Added CC to dev@openvswitch.org which was strangely missing
even though I recived the post via the list. Is the list
configuration broken somehow? ]
On Mon, Apr 07, 2014 at 10:50:57AM +0300, Lori Jakab wrote:
Hi,
On some of my systems the commit 0c47331 ofproto-dpif.at: Wait for
the
Uniformly wait for calls to ovs-ofctl exit using
OVS_APP_EXIT_AND_WAIT() in the test-suite.
Signed-off-by: Simon Horman ho...@verge.net.au
v2
* Always use OVS_APP_EXIT_AND_WAIT(ovs-ofctl) instead of
OVS_WAIT_UNTIL([ovs-appctl -t ovs-ofctl exit])
---
tests/ofproto-dpif.at | 36
On 7 April 2014 20:12, YAMAMOTO Takashi yamam...@valinux.co.jp wrote:
I've been looking at some races in the megaflow testcases recently as
well,
with a slightly different approach, could you a look at my patch below?
http://openvswitch.org/pipermail/dev/2014-April/038536.html
my
This corrects an inconsistency in the MPLS tests and
avoids a race by waiting for sufficient output before
checking its contents
this one looks identical to: http://patchwork.openvswitch.org/patch/3579/
YAMAMOTO Takashi
Signed-off-by: Simon Horman ho...@verge.net.au
---
Pravin, this patch includes a change to linux/openvswitch.h, will
you look at that part?
thanks for review.
does that file need special treatments?
YAMAMOTO Takashi
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Tue, Apr 08, 2014 at 09:39:57AM +0900, YAMAMOTO Takashi wrote:
This corrects an inconsistency in the MPLS tests and
avoids a race by waiting for sufficient output before
checking its contents
this one looks identical to: http://patchwork.openvswitch.org/patch/3579/
Agreed.
Uniformly wait for calls to ovs-ofctl exit using
OVS_APP_EXIT_AND_WAIT() in the test-suite.
are they all necessary to avoid some races?
in commit 0c473314, i intended to make it wait only when necessary.
ie. only tests which have multiple ovs-ofctl --pidfile.
YAMAMOTO Takashi
On Tue, Apr 08, 2014 at 10:00:47AM +0900, YAMAMOTO Takashi wrote:
Uniformly wait for calls to ovs-ofctl exit using
OVS_APP_EXIT_AND_WAIT() in the test-suite.
are they all necessary to avoid some races?
in commit 0c473314, i intended to make it wait only when necessary.
ie. only tests
On Tue, Apr 08, 2014 at 10:00:47AM +0900, YAMAMOTO Takashi wrote:
Uniformly wait for calls to ovs-ofctl exit using
OVS_APP_EXIT_AND_WAIT() in the test-suite.
are they all necessary to avoid some races?
in commit 0c473314, i intended to make it wait only when necessary.
ie. only tests
On Sun, Apr 06, 2014 at 11:30:13AM +0900, YAMAMOTO Takashi wrote:
Signed-off-by: YAMAMOTO Takashi yamam...@valinux.co.jp
Acked-by: Ben Pfaff b...@nicira.com
Thanks a lot for fixing all these tests! Most of them do not
regularly fail for me, but I can believe that they do for you in a
Von Dr. Andrea Orcel
Ich bin Dr. Andrea Orcel von Harlesden , North West London, hier in England.
Ich arbeite für UBS Bank , London . Ich schreibe Ihnen über ein Unternehmen
Vorschlag, dass einer immensen Nutzen für beide von uns sein wird . In meiner
Abteilung , wobei der Leiter Investment
On Tue, Apr 08, 2014 at 10:20:54AM +0900, YAMAMOTO Takashi wrote:
On Tue, Apr 08, 2014 at 10:00:47AM +0900, YAMAMOTO Takashi wrote:
Uniformly wait for calls to ovs-ofctl exit using
OVS_APP_EXIT_AND_WAIT() in the test-suite.
are they all necessary to avoid some races?
in commit
To clarify, this patch is aimed at the following type of situation:
* netdev-dummy puts the packet on the queue.
* Upcall handler receives the packet, installs a flow.
* Revalidator decides to delete the flow (examples are the flow fails
revalidation, or flow is deemed unimportant, other such
On 8 Apr 2014 12:56, YAMAMOTO Takashi yamam...@valinux.co.jp wrote:
On 7 April 2014 20:12, YAMAMOTO Takashi yamam...@valinux.co.jp wrote:
I've been looking at some races in the megaflow testcases recently as
well,
with a slightly different approach, could you a look at my patch
On 8 April 2014 13:53, YAMAMOTO Takashi yamam...@valinux.co.jp wrote:
To clarify, this patch is aimed at the following type of situation:
* netdev-dummy puts the packet on the queue.
* Upcall handler receives the packet, installs a flow.
* Revalidator decides to delete the flow (examples
Thanks Ben for many review comments for this patch series. Pushed to master.
On Mon, Apr 7, 2014 at 5:10 PM, Ben Pfaff b...@nicira.com wrote:
On Fri, Apr 04, 2014 at 04:03:02PM -0700, Andy Zhou wrote:
Infrastructure to enable megaflow support for bond ports using
recirculation. This patch adds
Recirculation implementation for Linux kernel data path.
Signed-off-by: Andy Zhou az...@nicira.com
---
datapath/actions.c | 55 ++-
datapath/datapath.c | 35 ++
datapath/datapath.h |3 ++-
@@ -3859,7 +4001,7 @@ ovs-appctl time/stop
ovs-appctl time/warp 5000
AT_CHECK([ovs-appctl netdev-dummy/receive p7
'in_port(1),eth(src=50:54:00:00:00:09,dst=50:54:00:00:00:0a),eth_type(0x0800),ipv4(src=10.0.0.2,dst=10.0.0.1,proto=1,tos=0,ttl=64,frag=no),icmp(type=8,code=0)'])
From: Ethan Jackson et...@nicira.com
Previously, we had a separate flow_dumper thread that fetched flows from
the datapath to distribute to revalidator threads. This patch takes the
logic for dumping and pushes it into the revalidator threads, resulting
in simpler code with similar performance to
This patch took advantage of the recirculation infrastructure
introduced in commit adcf00ba35a0, allowing megaflows to be generated
when the flow output to bond ports.
Without recirculation, it is necessary flows output to Bond ports
in balance_tcp mode to unmask all hash fields. With
You are right, it is not related.
The sleep 1 is not needed to the machines I have tried it on. But
since you are working on testing racing recently, I will leave it up
to you. But it is not related to recirculation at any rate, as you
pointed out.
On Mon, Apr 7, 2014 at 8:48 PM, YAMAMOTO
61 matches
Mail list logo
