Respond For More Details.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Hi Folks,
I set up a VM to access its network via a tap connected to an OVS datapath
(the tap is added as a 'system' port). In my system the VM's network is
virtual - isolated from other VMs and from the host OS. Yet, the VM still
receives some packets from the host OS (e.g. ICMP6) and this breaks
On Wed, Dec 14, 2011 at 06:21:13PM -0800, Ethan Jackson wrote:
> Since SLB bonds are fairly risky (see vswitchd/INTERNALS), it seems
> prudent to require users to explicitly configure them. This patch
> defaults to active-backup bonds when no other bond_mode is
> explicitly specified.
>
> Signed-
On Wed, Dec 14, 2011 at 05:50:39PM -0800, Ansis Atteka wrote:
> On Thu, Dec 8, 2011 at 2:01 PM, Ben Pfaff wrote:
> > +const char *
> > +eth_from_hex(const char *hex, struct ofpbuf **packetp)
> > +{
> > +struct ofpbuf *packet;
> > +
> > +packet = *packetp = ofpbuf_new(strlen(hex) / 2);
> >
On Wed, Dec 14, 2011 at 02:50:43PM -0800, Ethan Jackson wrote:
> > + ? ? ? ? ? ? ? ? ? ? ? ? ? ?tag_set_add(&o->revalidate_set, e->tag);
> > + ? ? ? ? ? ? ? ? ? ? ? ? ? ?mac_learning_expire (o->ml, e);
>
> Redundant space in this function call.
Thanks, removed.
> As I read this, I was wondering
On Wed, Dec 14, 2011 at 06:34:04PM -0800, Jesse Gross wrote:
> On Wed, Dec 14, 2011 at 4:12 PM, Ben Pfaff wrote:
> > Why do we care? ??Current userspace isn't really affected. ??At most, we
> > get flow statistics wrong by one packet (I guess multiple packets is
> > theoretically possible?), if an
Hi,
please don’t get offended at my email because i send it for a good cause. I
have an intention of helping the needy so my mail here may seem unusual to you,
but please read it and will find out what i am trying to do.
My name’s Mrs Aurora Romero, i am 72 years old, i live in England. I am
Ben and I talked this over offline and came to the following conclusion.
This could potentially be a disruptive change, but we do want to
encourage users to use active-backup instead of SLB in general. For
now, we will update NEWS indicating that we intend to switch to
active-backup as a default.
On Thu, Dec 15, 2011 at 10:48 AM, Ben Pfaff wrote:
> On Wed, Dec 14, 2011 at 06:34:04PM -0800, Jesse Gross wrote:
>> On Wed, Dec 14, 2011 at 4:12 PM, Ben Pfaff wrote:
>> > Why do we care? ??Current userspace isn't really affected. ??At most, we
>> > get flow statistics wrong by one packet (I gues
On Thu, Dec 15, 2011 at 09:35:14AM -0800, Ben Pfaff wrote:
> I think I'm going to try to write some unit tests for this before I
> push it.
I wrote some, but the unit test revealed a bug in the "dummy" dpif.
Please review this, and then I'll push both patches:
--8<--cut he
On Thu, Dec 15, 2011 at 11:42:21AM -0800, Jesse Gross wrote:
> >> > ?? ?? ?? ??3. Somehow actually eliminate the problem with deleting flows,
> >> > ?? ?? ?? ?? ?? so that when userspace receives the response to the flow
> >> > ?? ?? ?? ?? ?? deletion we know that no more packets can go through the
You need a Signed-off-by. Otherwise looks good.
Ethan
On Thu, Dec 15, 2011 at 11:49, Ben Pfaff wrote:
> On Thu, Dec 15, 2011 at 09:35:14AM -0800, Ben Pfaff wrote:
>> I think I'm going to try to write some unit tests for this before I
>> push it.
>
> I wrote some, but the unit test revealed a bu
On Thu, Dec 15, 2011 at 12:34 PM, Ben Pfaff wrote:
> On Thu, Dec 15, 2011 at 11:42:21AM -0800, Jesse Gross wrote:
>> >> > ?? ?? ?? ??3. Somehow actually eliminate the problem with deleting
>> >> > flows,
>> >> > ?? ?? ?? ?? ?? so that when userspace receives the response to the flow
>> >> > ?? ??
On Thu, Dec 15, 2011 at 01:00:00PM -0800, Jesse Gross wrote:
> On Thu, Dec 15, 2011 at 12:34 PM, Ben Pfaff wrote:
> That's true; I don't really think that the whole worker thread concept
> is really all that great overall anyways.
>
> > A workaround would be to call synchronize_rcu() and send the
Post version 1.4.*, we are planning to change the default bond_mode
from balance-slb to active-backup. This commit warns users of the
change so that they can prepare.
---
NEWS |7 +++
vswitchd/bridge.c | 20 +++-
2 files changed, 22 insertions(+), 5 deletion
This needs a signed-off-by.
Ethan
On Thu, Dec 15, 2011 at 13:42, Ethan Jackson wrote:
> Post version 1.4.*, we are planning to change the default bond_mode
> from balance-slb to active-backup. This commit warns users of the
> change so that they can prepare.
> ---
> NEWS | 7 ++
On Thu, Dec 15, 2011 at 01:42:09PM -0800, Ethan Jackson wrote:
> Post version 1.4.*, we are planning to change the default bond_mode
> from balance-slb to active-backup. This commit warns users of the
> change so that they can prepare.
Looks good, thank you.
__
On Thu, Dec 15, 2011 at 1:34 PM, Ben Pfaff wrote:
> On Thu, Dec 15, 2011 at 01:00:00PM -0800, Jesse Gross wrote:
>> On Thu, Dec 15, 2011 at 12:34 PM, Ben Pfaff wrote:
>> > A workaround would be to call synchronize_rcu() and send the genl
>> > reply from some context that doesn't hold genl_lock, b
On Thu, Dec 15, 2011 at 03:24:38PM -0800, Jesse Gross wrote:
> On Thu, Dec 15, 2011 at 1:34 PM, Ben Pfaff wrote:
> > On Thu, Dec 15, 2011 at 01:00:00PM -0800, Jesse Gross wrote:
> >> On Thu, Dec 15, 2011 at 12:34 PM, Ben Pfaff wrote:
> >> > A workaround would be to call synchronize_rcu() and send
When a flow consists solely of an output to OFPP_CONTROLLER, we avoid a
round trip to the kernel and back by calling execute_controller_action()
from handle_flow_miss(). However, execute_controller_action() frees the
packet passed in. This is dangerous, because the packet and the upcall
key are i
On Thu, Dec 15, 2011 at 03:46:11PM -0800, Ben Pfaff wrote:
> On Thu, Dec 15, 2011 at 03:24:38PM -0800, Jesse Gross wrote:
> > On Thu, Dec 15, 2011 at 1:34 PM, Ben Pfaff wrote:
> > > On Thu, Dec 15, 2011 at 01:00:00PM -0800, Jesse Gross wrote:
> > >> On Thu, Dec 15, 2011 at 12:34 PM, Ben Pfaff wro
On Thu, Dec 15, 2011 at 3:46 PM, Ben Pfaff wrote:
> On Thu, Dec 15, 2011 at 03:24:38PM -0800, Jesse Gross wrote:
>> On Thu, Dec 15, 2011 at 1:34 PM, Ben Pfaff wrote:
>> > On Thu, Dec 15, 2011 at 01:00:00PM -0800, Jesse Gross wrote:
>> >> On Thu, Dec 15, 2011 at 12:34 PM, Ben Pfaff wrote:
>> >> >
On Thu, Dec 15, 2011 at 04:49:57PM -0800, Jesse Gross wrote:
> >> > Could we just use the existing spinlock in sw_flow plus the 'dead'
> >> > variable to ensure that no packets go through a deleted flow after
> >> > it's deleted? ??On the delete side, take the spinlock and set 'dead'.
> >> > On the
On Thu, Dec 15, 2011 at 4:56 PM, Ben Pfaff wrote:
> On Thu, Dec 15, 2011 at 04:49:57PM -0800, Jesse Gross wrote:
>> >> > Could we just use the existing spinlock in sw_flow plus the 'dead'
>> >> > variable to ensure that no packets go through a deleted flow after
>> >> > it's deleted? ??On the dele
Looks good.
I'm planning to get rid of this function entirely in the fairly near future.
Ethan
On Thu, Dec 15, 2011 at 16:24, Ben Pfaff wrote:
> When a flow consists solely of an output to OFPP_CONTROLLER, we avoid a
> round trip to the kernel and back by calling execute_controller_action()
> f
On Wed, Dec 14, 2011 at 6:09 PM, Pravin B Shelar wrote:
> diff --git a/datapath/linux/compat/genetlink.inc
> b/datapath/linux/compat/genetlink.inc
> index bf96980..43c3227 100644
> --- a/datapath/linux/compat/genetlink.inc
> +++ b/datapath/linux/compat/genetlink.inc
> +static int genl_exec_cmd(st
On Wed, Dec 14, 2011 at 2:09 PM, Pravin B Shelar wrote:
> Signed-off-by: Pravin B Shelar
Thanks, this is definitely an improvement.
Acked-by: Jesse Gross
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Wed, Dec 14, 2011 at 2:09 PM, Pravin B Shelar wrote:
> diff --git a/datapath/datapath.c b/datapath/datapath.c
> index 72cf9f6..2eedbf5 100644
> --- a/datapath/datapath.c
> +++ b/datapath/datapath.c
> +static int __rehash_flow_table(void *dummy)
> +{
> + struct datapath *dp;
> +
> +
28 matches
Mail list logo
