This is only the communication part of the bundles functionality.
The actual message pre-validation and commits are not implemented.
We also enable OF1.4 for all the tests.
Signed-off-by: Alexandru Copot
Cc: Daniel Baluta
---
v6:
* return error code for invalid inner message size
v5:
* fold
The initial bond distribution indeed improved significantly with
e58f91a1. Thanks for solving one of the mysteries.
I have already pushed Simon's patch assuming test failure was still
observed with this patch. Using 256 is not a bad idea at any rate --
there is no particular good reason to pick 9
On Thu, May 1, 2014 at 3:50 PM, Pritesh Kothari
wrote:
> Signed-off-by: Pritesh Kothari
> ---
> v7: change a specific version to HAVE_RXHASH,
> fix skb_clear_hash to include l4_rxhash.
> v6: rebase for changes in datapath/datapath.c
> v5: move skb_clear_rxhash() from compat.h to skbuff.h,
>
> On Thu, May 01, 2014 at 03:54:22PM +0900, YAMAMOTO Takashi wrote:
>> > On Wed, Apr 30, 2014 at 10:24:46AM +0900, YAMAMOTO Takashi wrote:
>> >> The problem mentioned by Simon Horman in the following mail.
>> >> http://openvswitch.org/pipermail/dev/2014-April/039492.html
>> >>
>> >> Cc: Simon Horm
> netdev_rxq_open() open-codes much of netdev_ref(), so re-use that
> function instead.
>
> Signed-off-by: Joe Stringer
Reviewed-by: YAMAMOTO Takashi
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
> netdev_open() would previously increment a netdev's refcount without
> holding a lock for it. This commit shifts the locking to protect it.
>
> Found by inspection.
>
> Signed-off-by: Joe Stringer
Reviewed-by: YAMAMOTO Takashi
___
dev mailing list
Thanks you for clarifying the need. Since there is no urgent need to
introduce a stop gap measure, I am happy to drop this patch for now.
On Thu, May 1, 2014 at 10:10 PM, Simon Horman wrote:
> On Thu, May 01, 2014 at 07:46:21AM -0700, Ben Pfaff wrote:
>> On Thu, May 01, 2014 at 02:03:05AM -0700
On Thu, May 01, 2014 at 07:46:21AM -0700, Ben Pfaff wrote:
> On Thu, May 01, 2014 at 02:03:05AM -0700, Andy Zhou wrote:
> > On Thu, May 1, 2014 at 1:20 AM, Simon Horman wrote:
> > > On Thu, May 01, 2014 at 03:59:20PM +0900, YAMAMOTO Takashi wrote:
> > >> >> Something like this (I have not tested e
On Thu, May 01, 2014 at 06:46:08PM -0700, Andy Zhou wrote:
> On Thu, May 1, 2014 at 6:23 PM, YAMAMOTO Takashi
> wrote:
> >> It is my observation that when sending 91 packets to a mode=balance
> >> bond interface with three underlying ports in some cases not all ports
> >> receive at least 7 packe
> On Wed, Apr 30, 2014 at 11:59 PM, YAMAMOTO Takashi
> wrote:
Something like this (I have not tested either scenario):
I think this will fail to match but that may not be obvious to users:
packet_out: in_port=CONTROLLER actions=goto_table:1
table 1: match=in_port=CON
When controller sends OFPT_PACKET_OUT message with the in_port set
to a patch port or as CONTROLLER, and the message execution requires
recirculation, those packets will be dropped in the datapath.
This is because the post recirculation flow will not be set up by
Xlate layer that rejects up call wi
On 30 April 2014 12:46, Alex Wang wrote:
>
> +dnl set the stats update interval to 100K ms, the following 'recv' should
> not be updated.
> +AT_CHECK([ovs-vsctl set O . other_config:stats-update-interval=10])
> +for i in `seq 0 50`; do ovs-appctl time/warp 1000; done
> +for i in `seq 1 5`; do
On Thu, May 1, 2014 at 6:23 PM, YAMAMOTO Takashi wrote:
>> It is my observation that when sending 91 packets to a mode=balance
>> bond interface with three underlying ports in some cases not all ports
>> receive at least 7 packets. This causes the test to fail.
>
> was it with or without commit e5
Signed-off-by: Bruce Davie
---
vtep/vtep.ovsschema | 104 -
vtep/vtep.xml | 218 ++--
2 files changed, 161 insertions(+), 161 deletions(-)
diff --git a/vtep/vtep.ovsschema b/vtep/vtep.ovsschema
index 4a25f2f..94494ab 1
> It is my observation that when sending 91 packets to a mode=balance
> bond interface with three underlying ports in some cases not all ports
> receive at least 7 packets. This causes the test to fail.
was it with or without commit e58f91a1?
YAMAMOTO Takashi
>
> It may be that failure indicate
netdev_open() would previously increment a netdev's refcount without
holding a lock for it. This commit shifts the locking to protect it.
Found by inspection.
Signed-off-by: Joe Stringer
---
lib/netdev.c |6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/lib/netdev.c b
Previously, it was possible to open a netdevice as one type, then
proceed to open it as a different type without first closing it. The
bridge code would expect it to be opened as the latter type and try to
apply configuration to it. This patch catches the problem earlier by
detecting the case in ne
netdev_rxq_open() open-codes much of netdev_ref(), so re-use that
function instead.
Signed-off-by: Joe Stringer
---
lib/netdev.c |5 +
1 file changed, 1 insertion(+), 4 deletions(-)
diff --git a/lib/netdev.c b/lib/netdev.c
index 2fc1834..6f997fd 100644
--- a/lib/netdev.c
+++ b/lib/netde
On Thu, May 01, 2014 at 08:44:29AM -0700, Ben Pfaff wrote:
> On Tue, Apr 22, 2014 at 05:54:52PM +0900, Simon Horman wrote:
> > Rather then prohibiting hash actions dpif_execute_helper_cb()
> > execute them by passing them on to aux->dpif->dpif_class->execute()
> > and save the updated packet metada
Signed-off-by: Ben Pfaff
---
lib/dpif-netdev.c | 153 +
1 file changed, 73 insertions(+), 80 deletions(-)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
index 55712dd..2dddad5 100644
--- a/lib/dpif-netdev.c
+++ b/lib/dpif-netdev.c
@@ -32,6
This implements an "optimistic concurrent cuckoo hash", a single-writer,
multiple-reader hash table data structure. The point of this data
structure is performance, so this commit message focuses on performance.
I tested the performance of cmap with the test-cmap utility included in
this commit.
Commit 3e912ffcbb (netdev: Add 'change_seq' back to netdev.) added per-
netdev change number for indicating status change. Future commits used
this change number to optimize the netdev status update to database.
However, the work also introduced the bug in the following scenario:
- assume interfa
Signed-off-by: Pritesh Kothari
---
acinclude.m4 | 1 +
datapath/linux/compat/include/linux/skbuff.h | 3 +++
2 files changed, 4 insertions(+)
diff --git a/acinclude.m4 b/acinclude.m4
index 518a66b..e8a8a2e 100644
--- a/acinclude.m4
+++ b/acinclude.m4
@@ -295,6 +29
Signed-off-by: Pritesh Kothari
---
v1: backporting commit from master to fix skb_clear_rxhash.
---
acinclude.m4 | 1 +
datapath/compat.h | 3 +++
2 files changed, 4 insertions(+)
diff --git a/acinclude.m4 b/acinclude.m4
index cb563c6..d5f5830 100644
--- a/acinclude.m4
+++ b/acinclude.m4
@@
Signed-off-by: Pritesh Kothari
---
v1: backporting commit from master to fix skb_clear_rxhash.
---
acinclude.m4 | 1 +
datapath/compat.h | 5 -
2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/acinclude.m4 b/acinclude.m4
index b675e41..f51d66b 100644
--- a/acinclude.m4
+++ b
Signed-off-by: Pritesh Kothari
---
v1: backporting commit from master to fix skb_clear_rxhash.
---
acinclude.m4 | 1 +
datapath/compat.h | 3 +++
2 files changed, 4 insertions(+)
diff --git a/acinclude.m4 b/acinclude.m4
index d1271a4..3b7a5eb 100644
--- a/acinclude.m4
+++ b/acinclude.m4
@@
Signed-off-by: Pritesh Kothari
---
v7: change a specific version to HAVE_RXHASH,
fix skb_clear_hash to include l4_rxhash.
v6: rebase for changes in datapath/datapath.c
v5: move skb_clear_rxhash() from compat.h to skbuff.h,
simplify skb_get_hash() integration,
fix random.h order in acin
Signed-off-by: Pritesh Kothari
---
v1: backporting commit from master to fix skb_clear_rxhash.
---
acinclude.m4 | 1 +
datapath/compat.h | 5 -
2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/acinclude.m4 b/acinclude.m4
index fdcdf44..4c569ce 100644
--- a/acinclude.m4
+++ b
On Apr 30, 2014, at 3:42 PM, Jesse Gross wrote:
> On Tue, Apr 29, 2014 at 3:24 PM, Pritesh Kothari
> wrote:
>> diff --git a/datapath/linux/compat/include/linux/skbuff.h
>> b/datapath/linux/compat/include/linux/skbuff.h
>> index 714c955..de9b29d 100644
>> --- a/datapath/linux/compat/include/lin
Signed-off-by: Justin Pettit
---
NEWS |4
configure.ac |2 +-
debian/changelog |6 ++
3 files changed, 11 insertions(+), 1 deletions(-)
diff --git a/NEWS b/NEWS
index e1fb093..6790fd0 100644
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,7 @@
+v2.1.3 - xx xxx
+---
Thanks Jesse for the review. Push to master and branch-2.2 with the
fix and some other clean ups.
On Thu, May 1, 2014 at 1:21 PM, Jesse Gross wrote:
> On Wed, Apr 30, 2014 at 4:46 PM, Andy Zhou wrote:
>> diff --git a/datapath/actions.c b/datapath/actions.c
>> index 5871d82..5556a0c 100644
>> ---
On Wed, Apr 30, 2014 at 4:46 PM, Andy Zhou wrote:
> diff --git a/datapath/actions.c b/datapath/actions.c
> index 5871d82..5556a0c 100644
> --- a/datapath/actions.c
> +++ b/datapath/actions.c
> @@ -644,22 +649,24 @@ static DEFINE_PER_CPU(struct loop_counter,
> loop_counters);
> static int loop_su
Thanks for the patch. I will apply this in a few minutes.
I also find the hash distribution trouble some, but was not able to
pin it down. I will look at this more.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
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. Instructions for use are in
INSTA
On Thu, May 1, 2014 at 1:54 AM, Simon Horman wrote:
> On Wed, Apr 30, 2014 at 03:56:44PM -0700, Jesse Gross wrote:
>> On Tue, Apr 29, 2014 at 10:58 PM, Simon Horman wrote:
>> > On Tue, Apr 29, 2014 at 11:41:57AM -0700, Jesse Gross wrote:
>> >> On Mon, Apr 28, 2014 at 5:13 PM, Simon Horman wrote:
On Thu, May 1, 2014 at 2:00 AM, Suresh Kumar Reddy Reddygari
wrote:
> Hi,
>
> Is MacVTap support there in OVS. If not, is there any plan to support it.
No.
What are you trying to do?
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailm
On 29/04/14 17:36, Thomas Graf wrote:
On Tue, Apr 29, 2014 at 05:17:07PM +0100, Zoltan Kiss wrote:
On 23/04/14 22:56, Thomas Graf wrote:
On 04/23/2014 10:12 PM, Ethan Jackson wrote:
The problem has actually gotten worse since we've gotten rid of the
dispatcher thread. Now each thread has it's
Thanks, applied.
On Thu, May 01, 2014 at 08:44:10AM -0700, Jarno Rajahalme wrote:
> Acked-by: Jarno Rajahalme
>
> > On May 1, 2014, at 8:26 AM, Ben Pfaff wrote:
> >
> > Commit 05bf6d3c62e1d (ovs-thread: Add checking for mutex and rwlock
> > initialization.) updated mutex and rwlock initializer
On Tue, Apr 22, 2014 at 05:54:52PM +0900, Simon Horman wrote:
> Rather then prohibiting hash actions dpif_execute_helper_cb()
> execute them by passing them on to aux->dpif->dpif_class->execute()
> and save the updated packet metadata afterwards. It is assumed
> that if a hash is calculated it will
Acked-by: Jarno Rajahalme
> On May 1, 2014, at 8:26 AM, Ben Pfaff wrote:
>
> Commit 05bf6d3c62e1d (ovs-thread: Add checking for mutex and rwlock
> initialization.) updated mutex and rwlock initializers to set the "where"
> member to a nonnull value, but missed this one. This commit fixes the
>
On Tue, Apr 29, 2014 at 11:06:29AM -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
> possible.
>
> Assume BFD is used on both end of a link. Cons
Commit 05bf6d3c62e1d (ovs-thread: Add checking for mutex and rwlock
initialization.) updated mutex and rwlock initializers to set the "where"
member to a nonnull value, but missed this one. This commit fixes the
problem.
This does not cause real problems in practice because
OVS_ADAPTIVE_MUTEX_INI
On Thu, May 01, 2014 at 08:21:36AM -0700, Jarno Rajahalme wrote:
>
> > On May 1, 2014, at 7:53 AM, Ben Pfaff wrote:
> >
> >> On Mon, Apr 28, 2014 at 03:57:58PM -0700, Jarno Rajahalme wrote:
> >>
> >>> On Mar 20, 2014, at 10:05 AM, Ben Pfaff wrote:
> >>>
> On Fri, Mar 14, 2014 at 04:19:52
> On May 1, 2014, at 7:53 AM, Ben Pfaff wrote:
>
>> On Mon, Apr 28, 2014 at 03:57:58PM -0700, Jarno Rajahalme wrote:
>>
>>> On Mar 20, 2014, at 10:05 AM, Ben Pfaff wrote:
>>>
On Fri, Mar 14, 2014 at 04:19:52PM +0900, Simon Horman wrote:
When creating a flow in the datapath as the re
On Thu, Apr 24, 2014 at 04:03:46PM -0700, Gurucharan Shetty wrote:
> On Fri, Apr 11, 2014 at 11:18 AM, Ben Pfaff wrote:
> > It seems that it is useful to admins after all.
> >
> > Reported-by: Brian Candler
> > Signed-off-by: Ben Pfaff
> > ---
> > debian/openvswitch-switch.manpages |1 +
> >
On Thu, May 01, 2014 at 05:58:00PM +0900, Simon Horman wrote:
> On Wed, Apr 30, 2014 at 09:54:30AM +0900, YAMAMOTO Takashi wrote:
> > > On Tue, Apr 29, 2014 at 06:34:22PM +0900, Simon Horman wrote:
> > >> Add per-table counters. This resolves some short-comings
> > >> in the data provided in a tabl
On Mon, Apr 28, 2014 at 10:25:21PM -0700, Arun Sharma wrote:
> For each field, if the mask is all zeros, don't print it at all,
> if it is all ones, print it without the mask trailer.
> If it is neither of those two, print the value and mask in the current
> form.
>
> Removed inconsistency in form
On Mon, Apr 28, 2014 at 03:57:58PM -0700, Jarno Rajahalme wrote:
>
> On Mar 20, 2014, at 10:05 AM, Ben Pfaff wrote:
>
> > On Fri, Mar 14, 2014 at 04:19:52PM +0900, Simon Horman wrote:
> >> When creating a flow in the datapath as the result of an upcall
> >> the match itself is the match supplied
On Wed, Apr 30, 2014 at 04:27:47PM +0200, Daniele Venturino wrote:
> In lib/mac-learning.h there is a function (mac_learning_flush()) to
> flush the whole table. Is there a function to flush the entries
> pertaining to a single port?
There's bundle_flush_macs() in ofproto/ofproto-dpif.c.
_
On Thu, May 01, 2014 at 02:03:05AM -0700, Andy Zhou wrote:
> On Thu, May 1, 2014 at 1:20 AM, Simon Horman wrote:
> > On Thu, May 01, 2014 at 03:59:20PM +0900, YAMAMOTO Takashi wrote:
> >> >> Something like this (I have not tested either scenario):
> >> >>
> >> >> I think this will fail to match
On Thu, May 01, 2014 at 05:40:17PM +0900, Simon Horman wrote:
> It is my observation that when sending 91 packets to a mode=balance
> bond interface with three underlying ports in some cases not all ports
> receive at least 7 packets. This causes the test to fail.
>
> It may be that failure indica
On Thu, May 01, 2014 at 05:34:20PM +0900, Simon Horman wrote:
> This is a proposed correction for what appears to be
> an editing error.
>
> Signed-off-by: Simon Horman
I applied this with a few additional editorial improvements.
___
dev mailing list
d
On Wed, Apr 30, 2014 at 05:29:01PM -0700, Ethan Jackson wrote:
> If one tries to install a rule that's identical to another rule in
> another OpenFlow table which is being deleted, it's possible that
> is_flow_deletion_pending() might confuse them and block the
> installation. This is such an edge
On Thu, May 01, 2014 at 01:22:20PM +0300, Alexandru Copot wrote:
> Also fix some tests that can now properly print packets
> with the new protocol version.
>
> Signed-off-by: Alexandru Copot
> Cc: Daniel Baluta
Applied.
___
dev mailing list
dev@openvs
On Thu, May 01, 2014 at 01:22:19PM +0300, Alexandru Copot wrote:
> This is only the communication part of the bundles functionality.
> The actual message pre-validation and commits are not implemented.
>
> We also enable OF1.4 for all the tests.
>
> Signed-off-by: Alexandru Copot
> Cc: Daniel Ba
On Thu, May 01, 2014 at 05:29:59PM +0900, Simon Horman wrote:
> There does not seem to be a reason to expose
> recirc members in struct xlate_out to move them
> to struct xlate_ctx.
>
> Signed-off-by: Simon Horman
Applied.
___
dev mailing list
dev@open
On Thu, May 01, 2014 at 03:54:22PM +0900, YAMAMOTO Takashi wrote:
> > On Wed, Apr 30, 2014 at 10:24:46AM +0900, YAMAMOTO Takashi wrote:
> >> The problem mentioned by Simon Horman in the following mail.
> >> http://openvswitch.org/pipermail/dev/2014-April/039492.html
> >>
> >> Cc: Simon Horman
> >
This is only the communication part of the bundles functionality.
The actual message pre-validation and commits are not implemented.
We also enable OF1.4 for all the tests.
Signed-off-by: Alexandru Copot
Cc: Daniel Baluta
---
v5:
* fold tests patch
* add ofputil_encode_bundle_add()
* remo
Also fix some tests that can now properly print packets
with the new protocol version.
Signed-off-by: Alexandru Copot
Cc: Daniel Baluta
---
lib/ofp-print.c| 3 +++
tests/ofp-print.at | 6 +++---
tests/ofp-util.at | 2 +-
3 files changed, 7 insertions(+), 4 deletions(-)
diff --git a/lib/of
On Thu, May 1, 2014 at 1:20 AM, Simon Horman wrote:
> On Thu, May 01, 2014 at 03:59:20PM +0900, YAMAMOTO Takashi wrote:
>> >> Something like this (I have not tested either scenario):
>> >>
>> >> I think this will fail to match but that may not be obvious to users:
>> >> packet_out: in_port=CON
Hi,
Is MacVTap support there in OVS. If not, is there any plan to support it.
Thanks,
Suresh.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Wed, Apr 30, 2014 at 09:54:30AM +0900, YAMAMOTO Takashi wrote:
> > On Tue, Apr 29, 2014 at 06:34:22PM +0900, Simon Horman wrote:
> >> Add per-table counters. This resolves some short-comings
> >> in the data provided in a table stats reply message.
> >>
> >> * Lookups and matches are calculated
On Wed, Apr 30, 2014 at 03:56:44PM -0700, Jesse Gross wrote:
> On Tue, Apr 29, 2014 at 10:58 PM, Simon Horman wrote:
> > On Tue, Apr 29, 2014 at 11:41:57AM -0700, Jesse Gross wrote:
> >> On Mon, Apr 28, 2014 at 5:13 PM, Simon Horman wrote:
> >> > On Mon, Apr 28, 2014 at 02:37:47PM -0700, Jesse Gr
It is my observation that when sending 91 packets to a mode=balance
bond interface with three underlying ports in some cases not all ports
receive at least 7 packets. This causes the test to fail.
It may be that failure indicates that the bonding code
needs improvement. But it seems to me that its
On Wed, Apr 30, 2014 at 11:59 PM, YAMAMOTO Takashi
wrote:
>>> Something like this (I have not tested either scenario):
>>>
>>> I think this will fail to match but that may not be obvious to users:
>>> packet_out: in_port=CONTROLLER actions=goto_table:1
>>> table 1: match=in_port=CONTROLLER a
This is a proposed correction for what appears to be
an editing error.
Signed-off-by: Simon Horman
---
ofproto/ofproto-dpif.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c
index 5eea5ab..ce73afe 100644
--- a/ofproto/ofprot
There does not seem to be a reason to expose
recirc members in struct xlate_out to move them
to struct xlate_ctx.
Signed-off-by: Simon Horman
---
ofproto/ofproto-dpif-xlate.c | 20
ofproto/ofproto-dpif-xlate.h | 3 ---
2 files changed, 12 insertions(+), 11 deletions(-)
Thi
On Thu, May 01, 2014 at 03:59:20PM +0900, YAMAMOTO Takashi wrote:
> >> Something like this (I have not tested either scenario):
> >>
> >> I think this will fail to match but that may not be obvious to users:
> >> packet_out: in_port=CONTROLLER actions=goto_table:1
> >> table 1: match=in_port=
>> Something like this (I have not tested either scenario):
>>
>> I think this will fail to match but that may not be obvious to users:
>> packet_out: in_port=CONTROLLER actions=goto_table:1
>> table 1: match=in_port=CONTROLLER actions=normal
>>
>> I think this will match but that may not b
69 matches
Mail list logo
