Hi Nithin,
I'd also add at least:
1) Design for mapping hyper-v ports to OVS ports
2) CI testing plans
3) Roadmap
Thanks,
Alessandro
On 29.07.2014, at 05:01, "Nithin Raju" wrote:
> 10.00 AM Pacific - Sounds good to me.
>
> hi Alessandro,
> We'll be meeting tomorrow. In the light of that
On Tue, Jul 29, 2014 at 1:01 AM, Ben Pfaff wrote:
> Needs a sign-off.
>
Oh.. I will not forget it in the next version.
>
> I'm pretty sure that bitmap_equal() is wrong: doesn't it access one
> past the end of the bitmap arrays?
You are right. I'll fix it.
> I think that rightmost_1bit_idx
Heavy Netlink users such as Open vSwitch spend a considerable amount of
time in netlink_lookup() due to the read-lock on nl_table_lock. Use of
RCU relieves the lock contention.
Makes use of the new resizable hash table to avoid locking on the
lookup.
The hash table will grow if entries exceeds 75
Netlink sockets are maintained in a hash table to allow efficient lookup
via the port ID for unicast messages. However, lookups currently require
a read lock to be taken. This series adds a new generic, resizable,
scalable, concurrent hash table based on the paper referenced in the first
patch. It
Generic implementation of a resizable, scalable, concurrent hash table
based on [0]. The implementation supports both, fixed size keys specified
via an offset and length, or arbitrary keys via own hash and compare
functions.
Lookups are lockless and protected as RCU read side critical sections.
Au
Add MAX_LINKS define needed for nl_pool.
Add NLM_F_CREATE and NLM_F_EXCL defines also.
Signed-off-by: Alin Gabriel Serdean
---
lib/netlink-protocol.h | 4
1 file changed, 4 insertions(+)
diff --git a/lib/netlink-protocol.h b/lib/netlink-protocol.h
index 88b7abf..8938055 100644
--- a/lib/n
This is needed to create, get, set records in the Tunnel table.
Signed-off-by: Gurucharan Shetty
---
vtep/vtep-ctl.c |4
1 file changed, 4 insertions(+)
diff --git a/vtep/vtep-ctl.c b/vtep/vtep-ctl.c
index 0b9463a..129fa03 100644
--- a/vtep/vtep-ctl.c
+++ b/vtep/vtep-ctl.c
@@ -2283,6 +
Bypass the error compilation when compiling under MSVC.
Signed-off-by: Alin Gabriel Serdean
---
lib/netlink-socket.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/netlink-socket.h b/lib/netlink-socket.h
index d53db4e..2b9ec52 100644
--- a/lib/netlink-socket.h
+++ b/lib/netlink-socket
Add two functions set_sock_pid_in_kernel and portid_next. This will allow
the channel identification for the kernel extension to send back messages.
Replace send with WriteFile equivalent and ignore nl_sock_drain for the moment
under MSVC.
Replace sendmsg and recvmsg with ReadFile and WriteFile e
Add netlink related files to the windows build.
Signed-off-by: Alin Gabriel Serdean
---
lib/automake.mk | 9 +
1 file changed, 9 insertions(+)
diff --git a/lib/automake.mk b/lib/automake.mk
index 0997df5..87a8faa 100644
--- a/lib/automake.mk
+++ b/lib/automake.mk
@@ -319,6 +319,15 @@ l
Add defines needed to compile netlink-socket.c and netlink.c.
Add a wrapper and the functionality behind it for syconf.
Add the newly created files to the noinst_HEADERS in windows/automake.mk
Signed-off-by: Alin Gabriel Serdean
---
include/windows/automake.mk| 1 +
include/windows/ne
On Tue, Jul 29, 2014 at 01:41:32PM +0200, Thomas Graf wrote:
> Generic implementation of a resizable, scalable, concurrent hash table
> based on [0]. The implementation supports both, fixed size keys specified
> via an offset and length, or arbitrary keys via own hash and compare
> functions.
>
>
This is excellent. Thanks, Ben.
> -Original Message-
> From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Ben Pfaff
> Sent: Monday, July 28, 2014 11:35 PM
> To: dev@openvswitch.org
> Cc: Ben Pfaff
> Subject: [ovs-dev] [PATCH] netlink-socket: Add conceptual documentation.
>
> Base
On 2014-07-29 at 13:41:33 +0200, Thomas Graf wrote:
> Heavy Netlink users such as Open vSwitch spend a considerable amount of
> time in netlink_lookup() due to the read-lock on nl_table_lock. Use of
> RCU relieves the lock contention.
>
> Makes use of the new resizable hash table to avoid locking
Thanks for the review.
I realized that I left off a couple of important points, so I'm
folding in the following and will apply this to master soon.
diff --git a/lib/netlink-socket.h b/lib/netlink-socket.h
index 1450862..cdf32d2 100644
--- a/lib/netlink-socket.h
+++ b/lib/netlink-socket.h
@@ -149,
On Mon, Jul 28, 2014 at 02:03:49PM -0700, Jarno Rajahalme wrote:
> Ben,
>
> Thanks for reminding, somehow I lost track of this.
>
> This is actually a nice cleanup :-)
>
> Acked-by: Jarno Rajahalme
Thanks, applied to master.
___
dev mailing list
dev@
ONF-JIRA: EXT-244
Suggested-by: Jean Tourrilhes
Signed-off-by: Ben Pfaff
---
tests/ofproto-dpif.at | 21 +
1 file changed, 21 insertions(+)
diff --git a/tests/ofproto-dpif.at b/tests/ofproto-dpif.at
index 4386639..9ee1698 100644
--- a/tests/ofproto-dpif.at
+++ b/tests/ofpr
On Tue, Jul 29, 2014 at 03:21:16PM +, Alin Serdean wrote:
> Add defines needed to compile netlink-socket.c and netlink.c.
>
> Add a wrapper and the functionality behind it for syconf.
>
> Add the newly created files to the noinst_HEADERS in windows/automake.mk
>
> Signed-off-by: Alin Gabriel
We have pushed in the changes we made upon our repository:
https://github.com/cloudbase/openvswitch-hyperv-kernel.git
This conspicuous update includes:
* Rebased on the the tip of master.
* Functionality to take into account the NETLINK padding.
* Hashing (SpookyHash) to improve performance over
On Tue, Jul 29, 2014 at 03:22:37PM +, Alin Serdean wrote:
> Bypass the error compilation when compiling under MSVC.
>
> Signed-off-by: Alin Gabriel Serdean
Applied to master, thanks!
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/m
On 07/29/14 at 05:58pm, Tobias Klauser wrote:
> On 2014-07-29 at 13:41:33 +0200, Thomas Graf wrote:
> > Heavy Netlink users such as Open vSwitch spend a considerable amount of
> > time in netlink_lookup() due to the read-lock on nl_table_lock. Use of
> > RCU relieves the lock contention.
> >
> >
Hi,
Unfortunately, I don't really have time to test the code, but
from here it looks good and make sense.
Have fun...
Jean
Acked-by: Jean Tourrilhes
On Tue, Jul 29, 2014 at 09:27:23AM -0700, Ben Pfaff wrote:
> ONF-JIRA: EXT-244
> Suggested-by: Jean Tourrilhes
>
On Tue, Jul 29, 2014 at 03:23:28PM +, Alin Serdean wrote:
> Add two functions set_sock_pid_in_kernel and portid_next. This will allow
> the channel identification for the kernel extension to send back messages.
>
> Replace send with WriteFile equivalent and ignore nl_sock_drain for the moment
On Tue, Jul 29, 2014 at 03:24:08PM +, Alin Serdean wrote:
> Add netlink related files to the windows build.
>
>
> Signed-off-by: Alin Gabriel Serdean
Applied to master, thanks!
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailma
That's good enough for me, for a review of a test case. Thanks! I've
applied this to master.
On Tue, Jul 29, 2014 at 09:35:57AM -0700, Jean Tourrilhes wrote:
> Hi,
>
> Unfortunately, I don't really have time to test the code, but
> from here it looks good and make sense.
> Hav
Thank you for review and the changes.
Kind Regards,
Alin.
From: Ben Pfaff [b...@nicira.com]
Sent: Tuesday, July 29, 2014 7:38 PM
To: Alin Serdean
Cc: dev@openvswitch.org
Subject: Re: [ovs-dev] [PATCH v2 4/5] Changes needed to netlink-socket for MSVC
On Tu
One of the VMware internal autobuilder builds failed due to extraneous
logging in these tests of the form:
2014-07-28T21:11:07Z|1|poll_loop|INFO|wakeup due to [POLLIN] on fd 3
(...) at lib/stream-fd-unix.c:124 (93% CPU usage)
I think this must be because these tests have tons of output
On 07/29/14 at 08:30am, Josh Triplett wrote:
> On Tue, Jul 29, 2014 at 01:41:32PM +0200, Thomas Graf wrote:
> > Generic implementation of a resizable, scalable, concurrent hash table
> > based on [0]. The implementation supports both, fixed size keys specified
> > via an offset and length, or arbit
struct dp_upcall_info has pointer to pkt_key which is already
available in OVS_CB. This also simplifies upcall handling
for gso packet.
Signed-off-by: Pravin B Shelar
---
datapath/actions.c | 3 ---
datapath/datapath.c | 34 +-
datapath/datapath.h | 4 +---
3
Signed-off-by: Alex Wang
---
ofproto/connmgr.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/ofproto/connmgr.c b/ofproto/connmgr.c
index b7d5d3b..c6432bf 100644
--- a/ofproto/connmgr.c
+++ b/ofproto/connmgr.c
@@ -1496,7 +1496,7 @@ ofconn_wants_packet_in_on_miss(struct ofc
connmgr_wants_packet_in_on_miss() is called by multiple threads
and thusly should be protected by the mutex.
Signed-off-by: Alex Wang
---
ofproto/connmgr.c |6 +-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/ofproto/connmgr.c b/ofproto/connmgr.c
index c6432bf..89af6b6 100
The connmgr_wants_packet_in_on_miss() should only check 'ofconn's
of type OFCONN_PRIMARY (i.e. controller connections). This commit
makes it happen.
Signed-off-by: Alex Wang
---
ofproto/connmgr.c | 16 +---
1 file changed, 9 insertions(+), 7 deletions(-)
diff --git a/ofproto/conn
Hey Simon,
I think we should not check 'ofconn's for OFCONN_SERVICE... This
change broke several tests:
*Please send `tests/testsuite.log' and all information you think might
help:*
* To: >*
* Subject: [openvswitch 2.3.0] testsuite: 737 739 747 817 failed*
Wherein we use the ofctl monitor
Acked-by: Andy Zhou
On Tue, Jul 29, 2014 at 11:20 AM, Alex Wang wrote:
> Signed-off-by: Alex Wang
> ---
> ofproto/connmgr.c |2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/ofproto/connmgr.c b/ofproto/connmgr.c
> index b7d5d3b..c6432bf 100644
> --- a/ofproto/connmgr.
Should we also consider lock annotate the ofproto->connmgr variable?
Also, it seems to me the lock can be per ofproto, instead of a global
lock, but I am not sure this use case will benefit from a finer grain
lock.
On Tue, Jul 29, 2014 at 11:20 AM, Alex Wang wrote:
> connmgr_wants_packet_in_on
Thx for the suggestion, I think it will make the connmgr module safer
if we annotate the ofproto->connmgr. With a quick look, I think it will
take more checking and efforts, (for functions we know that are only called
by the main thread, there is no lock protection)
For branch-2.3, I want to use
On Tue, Jul 29, 2014 at 2:09 PM, Alex Wang wrote:
> Thx for the suggestion, I think it will make the connmgr module safer
> if we annotate the ofproto->connmgr. With a quick look, I think it will
> take more checking and efforts, (for functions we know that are only called
> by the main thread, t
On Tue, Jul 29, 2014 at 2:16 PM, Andy Zhou wrote:
> On Tue, Jul 29, 2014 at 2:09 PM, Alex Wang wrote:
> > Thx for the suggestion, I think it will make the connmgr module safer
> > if we annotate the ofproto->connmgr. With a quick look, I think it will
> > take more checking and efforts, (for fu
Thx, pushed both patches to master and branch-2.3.
I'll try improving the thread-safety of connmgr module later,
On Tue, Jul 29, 2014 at 2:32 PM, Alex Wang wrote:
>
>
>
> On Tue, Jul 29, 2014 at 2:16 PM, Andy Zhou wrote:
>
>> On Tue, Jul 29, 2014 at 2:09 PM, Alex Wang wrote:
>> > Thx for the
On Tue, Jul 29, 2014 at 11:30:12AM -0700, Alex Wang wrote:
> I think we should not check 'ofconn's for OFCONN_SERVICE... This
> change broke several tests:
>
> *Please send `tests/testsuite.log' and all information you think might
> help:*
>
> * To: >*
>
> * Subject: [openvswitch 2.3.0] test
Hey Ben,
Sorry for the confusion,
On Tue, Jul 29, 2014 at 2:52 PM, Ben Pfaff wrote:
> On Tue, Jul 29, 2014 at 11:30:12AM -0700, Alex Wang wrote:
> > I think we should not check 'ofconn's for OFCONN_SERVICE... This
> > change broke several tests:
> >
> > *Please send `tests/testsuite.log' and al
OVS flow extract is called on packet receive or packet
execute code path. Following patch defines separate API
for these two cases to simplify code.
Signed-off-by: Pravin B Shelar
---
datapath/actions.c | 3 +-
datapath/datapath.c | 8 ++---
datapath/flow.c | 82 +
From: Himangi Saraogi
Date: Sun, 27 Jul 2014 12:37:46 +0530
> This patch introduces the use of the macro IS_ERR_OR_NULL in place of
> tests for NULL and IS_ERR.
>
> The following Coccinelle semantic patch was used for making the change:
>
> @@
> expression e;
> @@
>
> - e == NULL || IS_ERR(e)
DPDK rings must have a power-of-two size.
Signed-off-by: Daniele Di Proietto
---
lib/netdev-dpdk.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index b45e367..c1b644a 100644
--- a/lib/netdev-dpdk.c
+++ b/lib/netdev-dpdk.c
@@ -125,
This patch fixes following kernel crash that could happen, if geneve
vport was not added yet, but revalidator thread attempted to dump flows.
To reproduce:
1. switch tunnel type between geneve and gre in a loop; and
2. run ping.
BUG: unable to handle kernel NULL pointer dereference at
On Mon, Jul 28, 2014 at 7:54 PM, Jesse Gross wrote:
> On Tue, Jul 22, 2014 at 9:34 AM, Jesse Gross wrote:
>> On Mon, Jul 21, 2014 at 5:11 PM, Ansis Atteka wrote:
>>> On Thu, Jul 17, 2014 at 4:08 PM, Jesse Gross wrote:
One thing that I worry about is that this has the possibility to
ch
No need for the unlikely(), WARN_ON() and BUG_ON() internally use
unlikely() on the condition.
Signed-off-by: Thomas Graf
---
net/core/dev.c | 2 +-
net/openvswitch/datapath.h | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/core/dev.c b/net/core/dev.c
index
On Tue, Jul 29, 2014 at 03:01:03PM -0700, Alex Wang wrote:
> Hey Ben,
>
> Sorry for the confusion,
>
> On Tue, Jul 29, 2014 at 2:52 PM, Ben Pfaff wrote:
>
> > On Tue, Jul 29, 2014 at 11:30:12AM -0700, Alex Wang wrote:
> > > I think we should not check 'ofconn's for OFCONN_SERVICE... This
> > >
On Tue, Jul 29, 2014 at 2:46 PM, Alex Wang wrote:
> Thx, pushed both patches to master and branch-2.3.
>
> I'll try improving the thread-safety of connmgr module later,
Sounds good. Thanks.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org
>
>
> I don't wish to add to any confusion but I will none the less
> make a comment on this change which you may choose to ignore if you like.
>
> The motivation for adding connmgr_wants_packet_in_on_miss() is to help
> implement per-OpenFlow version behaviour of packet_in messages. So if
> monit
From: Chunhe Li
If the internal device is not up, it should drop received packets. Sometimes
it receive the broadcast or multicast packets, and the ip protocol stack will
casue more cpu usage wasted.
Signed-off-by: Chunhe Li
---
datapath/vport-internal_dev.c | 5 +
1 file changed, 5 insert
From: Chunhe Li
Only check wheather is daemon pid exist is not enough, becasue the pid which
store in pidfile maybe assign to another process by OS. So it will checking
failed for pid exist, but the starting process which own the pid is not the
ovs daemon.
Signed-off-by: Chunhe Li
Signed-off-b
52 matches
Mail list logo
