Hi Ben,
I do not understand why add the dynamic arp entry to ovs table 65, not the
same table as static arp.
Though the packet match the flows in table 65, there is no output port, so
the packet should be drop, in this way, the dynamic arp is useless, right?
BR
Juno
2016-03-15 23:27 GMT+08:00
Change installing documentation to suggest to use
jemalloc memory allocator.
This memory allocator showed great performance gains
when used at ovsdb-server and other components.
Signed-off-by: Esteban Rodriguez Betancourt
---
.travis.yml | 3 +++
INSTALL.md | 7 +++
2
This commit change reconfigure_pmd_threads() to interact with the ports
cmap using RCU semantics (the content of the port structure is not
altered while concurrent readers might access it) and to fail more
gracefully in case of a set_multiq fail (now we remove the port from the
datapath, instead
Only the main thread will delete ports after pausing every other
thread. There's no need to keep count.
Signed-off-by: Daniele Di Proietto
---
lib/dpif-netdev.c | 50 ++
1 file changed, 14 insertions(+), 36 deletions(-)
It is only used in the testsuite and it can be replaced by a dpctl
command.
Signed-off-by: Daniele Di Proietto
---
lib/dpif-netdev.c | 33 -
tests/bridge.at | 4 ++--
2 files changed, 2 insertions(+), 35 deletions(-)
diff --git
This introduces in dpif-netdev and netdev-dpdk the first use for the
newly introduce reconfigure netdev call.
When a request to change the number of queues comes, netdev-dpdk will
remember this and notify the upper layer via
netdev_request_reconfigure().
The datapath, instead of periodically
This will ease deleting a port with no open rxqs.
Signed-off-by: Daniele Di Proietto
---
lib/dpif-netdev.c | 24 +---
1 file changed, 13 insertions(+), 11 deletions(-)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
index 9c30dad..a2281b8 100644
Signed-off-by: Daniele Di Proietto
---
lib/dpif-netdev.c | 140 ++--
lib/dpif-provider.h | 3 +-
2 files changed, 83 insertions(+), 60 deletions(-)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
index
This just documents what the current code already assumes.
Signed-off-by: Daniele Di Proietto
---
lib/dpif-netdev.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
index 0268b9b..597e1d7 100644
--- a/lib/dpif-netdev.c
+++
A netdev provider, especially a PMD provider (like netdev DPDK) might
not be able to change some of its parameters (such as MTU, or number of
queues) without stopping everything and restarting.
This commit introduces a mechanism that allows a netdev provider to
request a restart
This fixes multiple error path mistakes in do_add_port, none of which
has been a problem in practice so far. This change will make it easier
for a following commit to return in case of error.
Also, this removes an unneeded special case for tunnel ports.
Signed-off-by: Daniele Di Proietto
An upcoming commit will make dpif_poll_threads_set() record the
requested configuration and dpif_run() apply it, so it makes sense to
change the order.
Signed-off-by: Daniele Di Proietto
---
ofproto/ofproto-dpif.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
The dpif-netdev datapath keeps ports in a cmap which is written only by
the main thread (holding port_mutex), but which is read concurrently by
many threads (most notably the pmd threads).
When removing ports from the datapath we should postpone the deletion,
otherwise another thread might access
Currently we treat set_multiq() calls specially in netdev and dpif-netdev:
every pmd thread must be stopped and set_multiq() is allowed to destroy and
recreate the device.
I think we can improve this by:
* Generalizing the mechanism to allow changing other parameters at runtime
(such as MTU).
*
Hi Ilya,
thanks for your comments, replies inline
I'll post a v3 soon
On 15/03/2016 06:43, "Ilya Maximets" wrote:
>Few comments inline.
>
>Best regards, Ilya Maximets.
>
>On 03.03.2016 04:33, Daniele Di Proietto wrote:
>> This introduces in dpif-netdev and netdev-dpdk
Thank for spotting this Mark, I've updated the comment and resent
On 07/03/2016 03:26, "Kavanagh, Mark B" wrote:
>Hi Daniele,
>
>Thanks for making the changes to this patch and also 1/11 and 7/11.
>
>Apart from one minor comment below, all LGTM.
>
>Thanks,
>Mark
>
>>
Hi Ilya,
On 15/03/2016 06:43, "Ilya Maximets" wrote:
>comment inline.
>
>Best regards, Ilya Maximets.
>
>On 03.03.2016 04:33, Daniele Di Proietto wrote:
>> This commit change reconfigure_pmd_threads() to interact with the ports
>> cmap using RCU semantics (the content of
I see.
I think one option could be do extra validations depending on the error
returned by ofproto_check_ofpacts like checking if previous entries in the
bundle would make the validation successful. But since group and meter mods
are not implemented yet this isn't feasible.
I need to focus on my
On 15 March 2016 at 11:32, Ben Pfaff wrote:
> On Sun, Feb 28, 2016 at 10:33:24PM -0800, Gurucharan Shetty wrote:
> > Signed-off-by: Gurucharan Shetty
>
Thank you for looking through the series. I agree with all your comments so
far in the series and I will get them
Currently the datapath reports the tick counter to the userspace.
The userspace uses KeQueryPerformanceCounter as a monotonic clock.
This patch changes the flow stats to be reported in a monotonic format, while
also decaying the time between the flow actual usage and the flow report usage.
This
On 2016-03-15 11:16, Andy Zhou wrote:
On Mon, Mar 14, 2016 at 12:11 PM, Ramu Ramamurthy <
srama...@linux.vnet.ibm.com> wrote:
When a logical switch (localnet) has lots of ports on a hypervisor,
I find that broadcast packets from one of the ports is only forwarded
to
a subset of the other
Hi Lance,
Lance Richardson writes:
> Add minimimal support for building under OS X. With this series,
> "make check" reports:
>
> ERROR: 1831 tests were run, 30 failed (1 expected failure). 199 tests were
> skipped.
> testsuite: 232 368 394 395 396 397 398 740 910 920
"dev" wrote on 03/15/2016 11:21:16 AM:
> From: Ryan Moats/Omaha/IBM@IBMUS
> To: "ovs dev"
> Date: 03/15/2016 11:22 AM
> Subject: Re: [ovs-dev] [ovs-dev, RESEND] openvswitch: add support
> for kernel 4.4
> Sent by: "dev"
On Sun, Feb 28, 2016 at 10:33:24PM -0800, Gurucharan Shetty wrote:
> Signed-off-by: Gurucharan Shetty
GCC complains about strtok_r() in parse_ct_lb_action():
../ovn/lib/actions.c: In function 'actions_parse':
../ovn/lib/actions.c:251:5: error: 'save' may be used
On Mon, Mar 14, 2016 at 12:11 PM, Ramu Ramamurthy <
srama...@linux.vnet.ibm.com> wrote:
> When a logical switch (localnet) has lots of ports on a hypervisor,
> I find that broadcast packets from one of the ports is only forwarded to
> a subset of the other ports, and the kernel module shows the
Current openvswitch implementation allows up to 10 recirculation actions
for each packet. This limit was sufficient for most use cases in the
past, but with more new features, such as supporting connection
tracking, and testing in larger scale network environment,
This limit may be too
On Tue, Mar 15, 2016 at 10:20 AM, Jesse Gross wrote:
> On Tue, Mar 15, 2016 at 9:57 AM, Pravin B Shelar wrote:
>> diff --git a/vswitchd/vswitch.xml b/vswitchd/vswitch.xml
>> index f7b7b9c..98377b0 100644
>> --- a/vswitchd/vswitch.xml
>> +++
On Tue, Mar 15, 2016 at 10:10 AM, Thadeu Lima de Souza Cascardo
wrote:
> Fix subject: s/cashe/cache/.
>
> On Tue, Mar 15, 2016 at 09:57:29AM -0700, Pravin B Shelar wrote:
>> Currently arp and nd snooping is pretty loose. That causes
>> unnecessary entries in neighbour cache.
On Sun, Feb 28, 2016 at 10:33:23PM -0800, Gurucharan Shetty wrote:
> Invalid packets get dropped. Established connectections
> need to be sent through conntrack with just "nat" so
> that the VIP is changed to a loadbalanced IP address (by
> setting reg1 = 2). New connections need to be
On Thu, Mar 10, 2016 at 7:43 PM, Alexandru Ardelean
wrote:
> A bit cleaner than my previous patch.
> http://patchwork.ozlabs.org/patch/595969/
>
> Though I couldn't figure out a clean solution for ip6_local_out(),
> genl_notify(), and vport-vxlan
>
> Signed-off-by:
I think I’ll worry about that when I get to adding group and meter support.
Jarno
> On Mar 14, 2016, at 8:47 PM, Ben Pfaff wrote:
>
> Jarno, what do you think?
>
> On Mon, Mar 14, 2016 at 11:16:27PM +, André Mantas wrote:
>> Ok. And what about Packet_Out validation
On Sun, Feb 28, 2016 at 10:33:21PM -0800, Gurucharan Shetty wrote:
> Also add the ability to run database commands on the
> new schema using ovn-nbctl.
>
> Signed-off-by: Gurucharan Shetty
Some of the indentation in ovn/ovn-nb.xml looked a bit funny. I think
that it might use
On Tue, Mar 15, 2016 at 9:57 AM, Pravin B Shelar wrote:
> diff --git a/vswitchd/vswitch.xml b/vswitchd/vswitch.xml
> index f7b7b9c..98377b0 100644
> --- a/vswitchd/vswitch.xml
> +++ b/vswitchd/vswitch.xml
> @@ -1951,7 +1951,7 @@
>
>
>
> -An IPv4
Fix subject: s/cashe/cache/.
On Tue, Mar 15, 2016 at 09:57:29AM -0700, Pravin B Shelar wrote:
> Currently arp and nd snooping is pretty loose. That causes
> unnecessary entries in neighbour cache. Following patch
> adds required checks.
>
> Signed-off-by: Pravin B Shelar
>
On Mon, Mar 14, 2016 at 8:00 PM, Simon Horman
wrote:
> Currently output of MPLS packets on tunnel vports is not allowed by Open
> vSwitch. This is because historically encapsulation was done in such a way
> that the inner_protocol field of the skb needed to hold the
OK.
I'm probably not going to take this patch. It's just too specific.
I'd accept a patch that adds a note to suggest using jemalloc, via LIBS,
in INSTALL.md.
On Tue, Mar 15, 2016 at 04:56:22PM +, Rodriguez Betancourt, Esteban wrote:
> We opted for the flag option because we wanted to
On 3/15/16, 7:10 AM, "dev on behalf of Ryan Moats" wrote:
>
>-Original Message
>
>> @@ -256,8 +258,17 @@ physical_run(struct controller_ctx *ctx, enum
>mf_field_id mff_ovn_geneve,
>> } else {
>>
Signed-off-by: Pravin B Shelar
---
lib/ovs-router.c | 33 +++--
lib/ovs-router.h | 1 -
lib/route-table-bsd.c| 15 +++
lib/route-table-stub.c | 8
lib/route-table.c| 8
tnl-port monitors netdev change using netdev sequence number.
So to propogate ip address change we need to change netdev seq-no.
Signed-off-by: Pravin B Shelar
---
lib/netdev-dummy.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/netdev-dummy.c
From: Pravin B Shelar
There is check to disable IPv6 tunneling. Following patch
removes it and reintroduces the tunneling automake tests.
This reverts mostly commit 250bd94d1e500a89c76cac944e660bd9c07ac364.
Signed-off-by: Pravin B Shelar
---
NEWS
Since netdev can have multiple IP address use
generic api netdev_get_addr_list(). This also make it
easier to handle IPv4 and IPv6 address across vswitchd
layers.
Signed-off-by: Pravin B Shelar
---
lib/netdev-bsd.c | 61
Currently arp and nd snooping is pretty loose. That causes
unnecessary entries in neighbour cache. Following patch
adds required checks.
Signed-off-by: Pravin B Shelar
---
lib/tnl-neigh-cache.c| 7 +--
tests/ofproto-dpif.at| 2 +-
tests/tunnel-push-pop.at | 4
Following patch fixes number of issues with compose nd, like
setting ip packet header, set ICMP opt-len, checksum.
Signed-off-by: Pravin B Shelar
---
lib/packets.c | 59 +--
1 file changed, 41 insertions(+), 18
From: Pravin B Shelar
Device can have multiple IP address but netdev_get_in4/6()
returns only one configured IPv6 address. Following
patch fixes it.
OVS router is also updated to return source ip address for
given destination, This is required when interface has multiple
IP
We opted for the flag option because we wanted to affect just OVSDB-Server, and
not other programs. We don't have any specific issue with linking the other
programs with jemalloc, but we don't have tests for the other programs so we
are not sure if performance gain would be the same for them.
Found a pair of places where the patch wasn't clean...
original message
> diff --git a/datapath/linux/compat/include/net/ip.h
b/datapath/linux/compat/include/net/ip.h
> index cd87bcc..b749301 100644
> --- a/datapath/linux/compat/include/net/ip.h
> +++
OS X does not implement clock_gettime(), implement replacement.
Signed-off-by: Lance Richardson
---
Changes from v1: None
lib/timeval.c | 33 -
1 file changed, 32 insertions(+), 1 deletion(-)
diff --git a/lib/timeval.c b/lib/timeval.c
index
Avoid collision with system headers when building under OS X.
Signed-off-by: Lance Richardson
---
Changes from v1: this patch is new in v2.
lib/byte-order.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/byte-order.h b/lib/byte-order.h
index
Conditional compilation to account for:
- OS X does not implement RTM_IFANNOUNCE.
- OS X does not implement tap netdeivces.
- OS X does not implement RT_ROUNDUP().
Signed-off-by: Lance Richardson
---
Changes from v1:
- Combined patches 4 and 5 into patch number 4
-
Handle OS X like other BSDs (date -r is not supported, use sha1
instead of sha1sum).
Signed-off-by: Lance Richardson
---
Changes from v1: None
utilities/ovs-pki.in | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/utilities/ovs-pki.in
Add minimimal support for building under OS X. With this series,
"make check" reports:
ERROR: 1831 tests were run, 30 failed (1 expected failure). 199 tests were
skipped.
testsuite: 232 368 394 395 396 397 398 740 910 920 1018 1019 1020 1021 1022
1023 1084 1343 1951 2013 2014 2015 2016 2025
Rename ofproto-dpif wait() to avoid collision with wait(2) under OS X.
Signed-off-by: Lance Richardson
---
Changes from v1: None
ofproto/ofproto-dpif.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c
On Tue, Mar 15, 2016 at 06:04:04PM +0800, Na Zhu wrote:
> Hi Ben,
>
> I have one question about ARP support for logical router, i notice you add
> table MAC_Binding and ovn-controller translates the entries in MAC_Binding
> to Openflow flow table 65. The flow in table 65 is in this format,
>
>
Please don't drop the mailing list.
OK, so you'll have to add code to support the new vendor actions, some
in build-aux/extract-ofp-msgs, some in ofp-actions.c.
If it's your vendor code then you can use any types you want.
On Tue, Mar 15, 2016 at 11:36:20AM +0530, Ajmer Singh wrote:
> Hi Ben,
>
Original Message
> Proposal: add an option to ovs-ofctl utility so as to obtain colorized
output
> in tty, for easier reading. Currently, only the dump-flows command
supports
> colors.
[Snipping for BW]
Gave this a test spin this morning and I like it!
Acked-by: Ryan Moats
Original Message
> A (physical port + vlan) or (physical port + no tag) can be bound to a
> logical switch. When one unbinds (physical port + no tag) from a logical
> switch, the emulator inadvertantly deletes flows for (physical port +
vlan)
> too. This commit fixes it.
>
> VMware-BZ:
-Original Message
> @@ -256,8 +258,17 @@ physical_run(struct controller_ctx *ctx, enum
mf_field_id mff_ovn_geneve,
> } else {
> ofport = u16_to_ofp(simap_get(_to_ofport,
>binding->logical_port));
> -if
Ping.
Bets regards, Ilya Maximets.
On 24.02.2016 15:47, Ilya Maximets wrote:
> Since 5f2ccb1c0d3b ("dpif: Allow adding ukeys for same flow by
> different pmds.") there is the possibility to reassign queues among
> pmd threads without restarting them and deleting the megaflow cache.
>
> So,
comment inline.
Best regards, Ilya Maximets.
On 03.03.2016 04:33, Daniele Di Proietto wrote:
> This commit change reconfigure_pmd_threads() to interact with the ports
> cmap using RCU semantics (the content of the port structure is not
> altered while concurrent readers might access it) and to
Few comments inline.
Best regards, Ilya Maximets.
On 03.03.2016 04:33, Daniele Di Proietto wrote:
> This introduces in dpif-netdev and netdev-dpdk the first use for the
> newly introduce reconfigure netdev call.
>
> When a request to change the number of queues comes, netdev-dpdk will
>
> -Original Message-
> From: Ilya Maximets [mailto:i.maxim...@samsung.com]
> Sent: Wednesday, March 9, 2016 5:30 PM
> To: Wang, Zhihong ; dev@openvswitch.org
> Cc: Flavio Leitner ; Traynor, Kevin
> ;
> Dyasly Sergey
Hi Ben,
I have one question about ARP support for logical router, i notice you add
table MAC_Binding and ovn-controller translates the entries in MAC_Binding
to Openflow flow table 65. The flow in table 65 is in this format,
table=65, .., priority=100,reg0=0xa64,reg7=0x2,metadata=0x4
62 matches
Mail list logo