devices, otherwise they will be deleted and the two layers will have an
inconsistent view.
The problem was found while developing new code.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
v2:
* Translate ENOENT into ENODEV in dpif_port_query_by_name(), instead of
handlin
2017-01-05 2:42 GMT-08:00 Ciara Loftus :
> Prior to this commit, the 'dpdk' port type could only be used for
> physical DPDK devices. Now, virtual devices (or 'vdevs') are supported.
> 'vdev' devices are those which use virtual DPDK Poll Mode Drivers eg.
> null, pcap. To
plug is now
> automatically invoked when a valid PCI address is set in the
> dpdk-devargs. The format for ovs-appctl netdev-dpdk/detach command
> has changed in that the user now must specify the relevant PCI address
> as input instead of the port name.
>
> Signed-off-by: Ciara Loftus
) at
../vswitchd/ovs-vswitchd.c:111
Fixes: 8d8ab6c2d574 ("tun-metadata: Manage tunnel TLV mapping table on a
per-bridge basis.")
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
include/openvswitch/ofp-util.h | 5 +++--
lib/ofp-print.c| 4 +---
On 22/12/2016 21:20, "Darrell Ball" <db...@vmware.com> wrote:
>Some comments inline
Thanks for the review, I've sent a v3
>
>On 12/22/16, 6:36 PM, "Daniele Di Proietto" <diproiet...@vmware.com> wrote:
>
>The userspace connection tracker tr
On 22/12/2016 18:55, "Darrell Ball" <db...@vmware.com> wrote:
>
>
>On 12/22/16, 6:36 PM, "Daniele Di Proietto" <diproiet...@vmware.com> wrote:
>
>ICMP error packets (e.g. destination unreachable messages) are
>considered 'rela
that explicitly checked that neighbor
discovery was treated as invalid.
Reported-by: Sridhar Gaddam <sgad...@redhat.com>
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/conntrack.c | 134
tests/ofproto-dpi
This is just a naming change. When we extract the key of an ICMP error
message we suspect that it might be related, but we're not sure until we
perform a lookup in the connection table.
Suggested-by: Darrell Ball <db...@vmware.com>
Signed-off-by: Daniele Di Proietto <diproiet...@v
2016-12-22 3:05 GMT-08:00 Kevin Traynor <ktray...@redhat.com>:
> On 12/22/2016 10:02 AM, Kevin Traynor wrote:
>> On 12/21/2016 07:35 PM, Daniele Di Proietto wrote:
>>> 2016-12-21 10:18 GMT-08:00 Kevin Traynor <ktray...@redhat.com>:
>>>> On 12/21/2016 03:02
This is just a naming change. When we extract the key of an ICMP error
message we suspect that it might be related, but we're not sure until we
perform a lookup in the connection table.
Suggested-by: Darrell Ball <db...@vmware.com>
Signed-off-by: Daniele Di Proietto <diproiet...@v
that neighbor
discovery was treated as invalid.
Reported-by: Sridhar Gaddam <sgad...@redhat.com>
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
v2: Update comment to reflect that we do not do special validation with
the packet.
---
lib/conntrack.c
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
v2: Handle ICMP error for non existing connection in else branch without
restructuring the whole code flow.
---
lib/conntrack.c | 6 +-
tests/system-traffic.at | 27 ---
2 files changed, 21 i
e should take the new patch.
Great, I pushed the new patch to master.
Thanks!
Daniele
>
> Best regards,
> Zoltan
>
> -Original Message-
> From: Daniele Di Proietto [mailto:diproiet...@ovn.org]
> Sent: Thursday, December 22, 2016 4:06 AM
> To: Zoltá
2016-12-13 9:27 GMT-08:00 Zoltán Balogh :
>
> Hi Daniele,
>
>> Have you tried avoiding also computing the new field if the mask is 0?
>> For example, if
>> mask->ipv4_src is 0, there's not reason to compute new_ip_src, or to
>> extract ip_src_nh.
>
> Yes, I have
On 21/12/2016 12:06, "Aaron Conole" wrote:
>Aaron Conole writes:
>
>> DPDK documentation is recently updated to reflect that DPDK does not
>> hold any references to, nor take ownership of, the argv/argc elements.
>> With that understanding, let's
2b3f("ofp-actions: Check that 'alg=ftp' matches on TCP.")
I have a few more comments inline, otherwise:
Acked-by: Daniele Di Proietto <diproiet...@vmware.com>
> ---
> Documentation/intro/install/general.rst | 2 +
> NEWS| 1 +
> Vagran
2016-12-20 13:28 GMT-08:00 Joe Stringer <j...@ovn.org>:
> Signed-off-by: Joe Stringer <j...@ovn.org>
Acked-by: Daniele Di Proietto <diproiet...@vmware.com>
> ---
> tests/system-ovn.at | 2 ++
> tests/system-traffic.at | 3 +++
> 2 files changed, 5 insertions
2016-12-21 10:18 GMT-08:00 Kevin Traynor <ktray...@redhat.com>:
> On 12/21/2016 03:02 AM, Daniele Di Proietto wrote:
>> 2016-12-20 14:08 GMT-08:00 Kevin Traynor <ktray...@redhat.com>:
>>> On 12/15/2016 11:54 AM, Ciara Loftus wrote:
>>>> 'dpdk' ports no lon
2016-12-20 16:49 GMT-08:00 Joo Kim :
> (changed the subj)
>
> Could anybody help confirm about pmd->flow_table below? -Thanks-
>
> -- Forwarded message --
> From: Joo Kim
> Date: Mon, Dec 19, 2016 at 4:35 PM
> Subject: Re: [ovs-dev]
On 20/12/2016 17:14, "Joe Stringer" <j...@ovn.org> wrote:
>On 20 December 2016 at 14:46, Darrell Ball <db...@vmware.com> wrote:
>>
>>
>> On 12/20/16, 12:25 PM, "ovs-dev-boun...@openvswitch.org on behalf of Daniele
>> Di Proietto"
ovn.org<mailto:ja...@ovn.org>> wrote:
>
>
>
> On Dec 16, 2016, at 9:12 AM, Sridhar Gaddam
> <sgad...@redhat.com<mailto:sgad...@redhat.com>> wrote:
>
>
>
>
> On Tue, Dec 13, 2016 at 1:12 PM, Daniele Di Proietto
> <diproiet...@ovn.org<ma
that neighbor
discovery was treated as invalid.
Reported-by: Sridhar Gaddam <sgad...@redhat.com>
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/conntrack.c | 134
tests/ofproto-dpif.at | 32
On 14/12/2016 06:38, "Balazs Nemeth" <balazs.nem...@ericsson.com> wrote:
>>On 12/12/2016 13:14, "Ben Pfaff" <b...@ovn.org> wrote:
>>
>>>On Thu, Dec 08, 2016 at 06:50:32PM -0800, Daniele Di Proietto wrote:
>>>> The order of t
Not super important, but can we support detaching vdevs?
I'd like to see an example of usage with af_packet devices on the
documentation, but I can add that later.
Thanks,
Daniele
2016-12-13 11:22 GMT-08:00 Ciara Loftus :
> Prior to this commit, the 'dpdk' port type
I've been playing with this a little and I've come up with an
incremental with these changes:
* Detaching fails if port_id is in use by netdev
* I suggested before to process the arguments in reconfigure(). I
realize now that it's probably better to populate 'port_id' in
set_config() and start
2016-12-13 11:22 GMT-08:00 Ciara Loftus :
> In order to use dpdk ports in ovs they have to be bound to a DPDK
> compatible driver before ovs is started.
>
> This patch adds the possibility to hotplug (or hot-unplug) a device
> after ovs has been started. The implementation
2016-12-12 3:21 GMT-08:00 Sridhar Gaddam :
> Hello,
>
> In our setup, we are using OVS 2.6+dpdk and are noticing that conntrack for
> IPv6 traffic is NOT working.
> However, the same use-case works fine for Ipv4 traffic.
>
> We had a look at OVS 2.6 release notes[1], and it
On 12/12/2016 16:24, "Ben Pfaff" <b...@ovn.org> wrote:
>On Mon, Dec 12, 2016 at 04:12:30PM -0800, Daniele Di Proietto wrote:
>> The order of the options in the packet generated by ovs-controller
>> depends on the hash function. I believe that murmur hash (our
/bugreport.cgi?bug=840770
Suggested-by: Ben Pfaff <b...@ovn.org>
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
ovn/northd/ovn-northd.c | 25 +
tests/ovn.at| 4 ++--
2 files changed, 19 insertions(+), 10 deletions(-)
diff --git a/ov
On 12/12/2016 13:14, "Ben Pfaff" <b...@ovn.org> wrote:
>On Thu, Dec 08, 2016 at 06:50:32PM -0800, Daniele Di Proietto wrote:
>> The order of the options in the packet generated by ovs-controller
>> depends on the hash function. I believe that murmur hash (our
On 09/12/2016 05:44, "Numan Siddique" <nusid...@redhat.com> wrote:
>
>
>
>On Fri, Dec 9, 2016 at 8:20 AM, Daniele Di Proietto
><diproiet...@vmware.com> wrote:
>
>The packet in userdata generated by ovn-controller when translating the
>
2016-12-09 8:22 GMT-08:00 Aaron Conole :
> The DPDK EAL library intents that all argc/argv arguments passed on the
> command line will be in the form:
>
> progname dpdk arguments program arguments
>
> This means the argv array will look something like:
>argv[0] =
On 08/12/2016 12:18, "Stokes, Ian" wrote:
>> Expected behavior for attribute removal from the database is resetting it
>> to default value. Currently this doesn't work for n_rxq/n_txq options of
>> pmd netdevs (last requested value used):
>>
>> # ovs-vsctl set
Thanks for the patch, it looks good to me and I can confirm the
throughput improvements.
Have you tried avoiding also computing the new field if the mask is 0?
For example, if
mask->ipv4_src is 0, there's not reason to compute new_ip_src, or to
extract ip_src_nh.
2016-12-02 3:46 GMT-08:00 Zoltán
ency: 14.1309 usec
> Max. Latency: 28.9880 usec
>
>Device 1->0:
> Throughput: 6.3941 Mpps
> Min. Latency: 11.9780 usec
> Avg. Latency: 18.0692 usec
> Max. Latency: 29.5200
>
>Signed-off-by: Ilya Maximets <i.maxim...@samsung.com>
On 09/12/2016 11:04, "Jarno Rajahalme" <ja...@ovn.org> wrote:
>Acked-by: Jarno Rajahalme <ja...@ovn.org>
Thanks Jarno, I pushed this to master and up to branch-2.3
>
>> On Dec 8, 2016, at 6:50 PM, Daniele Di Proietto <diproiet...@vmware.com>
>&
s fixes a couple of connection tracking tests in userspace for big
endian platforms.
I guess RFC1071 4.1 (implementation example of the checksum in C), would
manifest the same problem on big endian.
Reported-at: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=840770
Signed-off-by: Daniele Di Pr
>
Interesting, I pushed a related fix a few days ago, and on my machine
it builds fine even without your patch.
Just curious, which version of pycodestyle do you have?
In any case, I tested it and it builds on my machine
Acked-by: Daniele Di Proietto <diproiet...@vmware.com>
> -
On 06/12/2016 04:48, "Ilya Maximets" <i.maxim...@samsung.com> wrote:
>On 03.12.2016 05:14, Daniele Di Proietto wrote:
>> +/* Assign pmds to queues. If 'pinned' is true, assign pmds to pinned
>> + * queues and marks the pmds as isolated. Otherwise,
he expressions that follow them. (Not only in that patch)
Good point, I fixed that in every commit of the series.
>
>Few more comments inline.
>
>Best regards, Ilya Maximets.
>
>On 03.12.2016 05:14, Daniele Di Proietto wrote:
>> Currently we have three different code path
On 02/12/2016 18:21, "Daniele Di Proietto" <diproiet...@vmware.com> wrote:
>
>
>
>
>
>On 02/12/2016 06:48, "Ilya Maximets" <i.maxim...@samsung.com> wrote:
>
>>This reduces code duplication. Additionally dp_netdev_free
>>wi
On 02/12/2016 06:48, "Ilya Maximets" wrote:
>'need_reconfigure' is actually local variable for the
>'reconfigure_datapath()'. It's better to keep common
>structures clean.
>
>Signed-off-by: Ilya Maximets
That's true, I actually coded it
On 02/12/2016 06:48, "Ilya Maximets" wrote:
>This reduces code duplication. Additionally dp_netdev_free
>will not reconfigure datapath on each removed port.
>
>Signed-off-by: Ilya Maximets
I like reducing code duplication, but I'm not sure I
On 02/12/2016 06:48, "Ilya Maximets" wrote:
>Hi, Daniele.
>I've prepared few more changes to your patch-set
>"[PATCH 00/17] DPDK/pmd reconfiguration refactor and bugfixes".
Thanks a lot for these
>
>Some of this changes can be squashed to your patches, or, maybe,
>some
ovs-numa doesn't need to keep the state of the pmd threads, it is an
implementation detail of dpif-netdev.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/ovs-numa.c | 175 -
lib/ovs-numa.h | 7 ---
2 files c
it's blocked) and set 'reload' to true.
This is useful to avoid wasting CPU cycles and is also necessary for a
future commit.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/dpif-netdev.c | 15 +++
1 file changed, 15 insertions(+)
diff --git a/lib/dpif-netd
I think it's better to iterate the hmap than to use
hmap_first_with_hash(), because it handles hash collisions.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/ovs-numa.c | 26 ++
1 file changed, 14 insertions(+), 12 deletions(-)
diff --git a/l
As documented in dp_netdev_pmd_thread, we must take non_pmd_mutex to
access the tx port caches for the non pmd thread.
Found by inspection.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/dpif-netdev.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/dpif-netd
It will be used by a future commit. struct ovs_numa_dump now uses an
hmap instead of a list to make ovs_numa_dump_contains_core() more
efficient.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/ovs-numa.c | 26 +++---
lib/ovs-numa.h | 10 ++-
will not consume
any CPU, so it seems easier to create all the threads at once.
This will also make future commits easier.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/dpif-netdev.c | 206 ++
tests/pmd.at | 2 +-
2
We keep all the per-port classifiers around, since they can be reused,
but when a pmd thread is destroyed we should free them.
Found using valgrind.
Fixes: 3453b4d62a98("dpif-netdev: dpcls per in_port with sorted
subtables")
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.
-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/dpif-netdev.c | 19 +++
1 file changed, 7 insertions(+), 12 deletions(-)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
index 12f11e4..5be3acf 100644
--- a/lib/dpif-netdev.c
+++ b/lib/dpif-netdev.c
@@ -485,7 +485,7 @@
come from OpenFlow.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/netdev-dpdk.c | 28
1 file changed, 12 insertions(+), 16 deletions(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index 6e5cd43..73860fe 100644
--- a/lib/netdev-
2016-11-29 5:13 GMT-08:00 Kevin Traynor :
> On 11/29/2016 10:24 AM, billy.o.mah...@intel.com wrote:
>> From: billyom
>>
>> pmd-cpu-mask is interpreted as a hex bit mask. So it should be written
>> with a leading 0x to indicate this. But if this is
On 02/12/2016 11:02, "Ben Pfaff" <b...@ovn.org> wrote:
>On Fri, Dec 02, 2016 at 10:53:04AM -0800, Daniele Di Proietto wrote:
>> pycodestyle >= 2.1.0 reports E305 otherwise, and the flake8-check step
>> fails.
>>
>> Signed-off-by: Daniele Di Proi
On 29/11/2016 05:26, "Loftus, Ciara" wrote:
>>
>> On 29.11.2016 15:20, Ilya Maximets wrote:
>> > If 'vhost-server-path' not provided for vhostuserclient port,
>> > 'netdev_dpdk_vhost_destruct()' will try to unregister an empty string.
>> > This leads to error
On 25/11/2016 06:22, "Ilya Maximets" <i.maxim...@samsung.com> wrote:
>On 16.11.2016 03:45, Daniele Di Proietto wrote:
>> We keep all the per-port classifiers around, since they can be reused,
>> but when a pmd thread is destroyed we should free them.
>>
2016-11-22 12:22 GMT-08:00 Bodireddy, Bhanuprakash <
bhanuprakash.bodire...@intel.com>:
>
> >-Original Message-
> >From: ovs-dev-boun...@openvswitch.org [mailto:ovs-dev-
> >boun...@openvswitch.org] On Behalf Of Daniele Di Proietto
> >Sent: Wednesday, No
2016-11-11 5:44 GMT-08:00 Binbin Xu :
> If the eth device starts failed or vhost port is not running,
> the flags of dpdk netdev should be restored.
>
> Signed-off-by: Binbin Xu
> ---
> lib/netdev-dpdk.c | 8 +++-
> 1 file changed, 7
2016-10-28 4:32 GMT-07:00 Patrik Andersson
:
> The non-pmd thread static_tx_qid is assumed to be equal to the highest
> core ID + 1. The function dp_netdev_del_pmds_on_numa() invalidates
> this assumption by re-distributing the static_tx_qid:s on all pmd and
>
2016-11-11 5:44 GMT-08:00 Binbin Xu :
> When we set physical port's admin state via ovs-appctl, the application
> seems to work and returns "OK". But the application doesn't work perfectly,
> the state stored in database doesn't change.
>
> Signed-off-by: Binbin Xu
2016-11-11 5:44 GMT-08:00 Binbin Xu :
> When we set a vhost port's admin state via ovs-appctl, the
> application doesn't work and returns the information
> "Not a DPDK Interface".
>
> Signed-off-by: Binbin Xu
>
Thanks for the patch
I'm thinking
ovs-numa doesn't need to keep the state of the pmd threads, it is an
implementation detail of dpif-netdev.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/ovs-numa.c | 168 -
lib/ovs-numa.h | 7 ---
2 files c
more
important than optimizing port addition or removal (a port is added and
removed only once, but can be reconfigured many times)
Lastly, this commit moves the pmd threads state away from ovs-numa. Now
the pmd threads state is kept only in dpif-netdev.
Signed-off-by: Daniele Di Proietto <dipro
is an implementation
detail of dpif-netdev, dpctl shouldn't know anything about it.
This commit changes the dpif_flow_put() and dpif_flow_del() calls to
iterate over all the pmd threads, if pmd_id is PMD_ID_NULL.
A simple test is added.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.
Since previous commit, 'static_tx_qid' doesn't need to be atomic and is
actually never touched (except for initialization), so it can be made
const.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/dpif-netdev.c | 11 +--
1 file changed, 5 insertions(+), 6 del
-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/dpif-netdev.c | 19 +++
1 file changed, 7 insertions(+), 12 deletions(-)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
index 81366b2..23546b9 100644
--- a/lib/dpif-netdev.c
+++ b/lib/dpif-netdev.c
@@ -485,7 +485,7 @@
it's blocked) and set 'reload' to true.
This is useful to avoid wasting CPU cycles and is also necessary for a
future commit.
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/dpif-netdev.c | 14 ++
1 file changed, 14 insertions(+)
diff --git a/lib/dpif-netd
We keep all the per-port classifiers around, since they can be reused,
but when a pmd thread is destroyed we should free them.
Found using valgrind.
Fixes: 3453b4d62a98("dpif-netdev: dpcls per in_port with sorted
subtables")
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.
but it makes the performance better with bigger cmaps.
What do you think? Have you tried this with bigger flow tables?
Thanks,
Daniele
>
> >-Original Message-
> >From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Bodireddy,
> >Bhanuprakash
> >Sent: T
:
> Throughput: 6.3941 Mpps
> Min. Latency: 11.9780 usec
> Avg. Latency: 18.0692 usec
> Max. Latency: 29.5200
>
> Signed-off-by: Thadeu Lima de Souza Cascardo <casca...@redhat.com>
> Cc: Ilya Maximets <i.maxim...@samsung.com>
> Cc: Daniele Di Proietto &l
cf44a5 ("netdev-dpdk: Use RCU for egress QoS.")
Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com>
---
lib/netdev-dpdk.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index 7c1523e..49750c3 100644
--- a/lib/netdev-d
Hi Flavio,
I was thinking that instead of having a separate appctl we could integrate
the attach into netdev_dpdk_construct() while changing the naming
convention, as discussed here:
http://openvswitch.org/pipermail/dev/2016-August/078113.html
What do you think?
Thanks,
Daniele
2016-10-26
On 24/10/2016 00:31, "Michael Qiu" <08005...@163.com> wrote:
>Hi, all
>
>
>When I self-compiled the OVS-2.6 from commit:
>
>commit 7a0f907b2393626dac1387617355990eab69aef7
>Author: Justin Pettit
>Date: Tue Sep 27 10:13:42 2016 -0700
>
> Set release date for 2.6.0.
>
>
Thanks, pushed to master
On 19/10/2016 14:51, "Sairam Venugopal" <vsai...@vmware.com> wrote:
>Acked-by: Sairam Venugopal <vsai...@vmware.com>
>
>
>On 10/19/16, 1:50 PM, "Daniele Di Proietto" <diproiet...@vmware.com> wrote:
>
>>OVS_
OVS_ALIGNED_VAR(...) should be at the beginning of a definition, as
the example in include/openvswitch/compiler.h shows.
Fixes: 38ee0814978c ("dpif-netdev: Cache align netdev_flow_keys.")
Reported-by: Joe Stringer <j...@ovn.org>
Signed-off-by: Daniele Di Proietto <diproiet...@v
;> Aligning the 'keys' array seems to have positive performance impact.
> >>
> >> Signed-off-by: Bhanuprakash Bodireddy <bhanuprakash.bodire...@intel.com
> >
> >> Co-authored-by: Antonio Fischetti <antonio.fische...@intel.com>
> >> Signed-off-by: A
2016-10-18 16:59 GMT-07:00 Jarno Rajahalme <ja...@ovn.org>:
> Compile fails otherwise.
>
> Signed-off-by: Jarno Rajahalme <ja...@ovn.org>
>
Acked-by: Daniele Di Proietto <diproiet...@vmware.com>
Thanks!
> ---
> Makefile.am | 1 +
> 1 file changed, 1 in
2016-10-14 7:37 GMT-07:00 Bhanuprakash Bodireddy <
bhanuprakash.bodire...@intel.com>:
> prefetching the data in to the caches isn't improving the performance in
> cmap_find_batch(). Moreover its found that there is slight improvement
> in performance with out prefetching.
>
> This patch removes
2016-10-14 7:37 GMT-07:00 Bhanuprakash Bodireddy <
bhanuprakash.bodire...@intel.com>:
> This patch adds comments to mf_get_next_in_map() to make it more
> comprehensible.
>
> Signed-off-by: Jarno Rajahalme
> Acked-by: Bhanuprakash Bodireddy
>
Thanks for the series, I applied most of it to master.
I sent some comments on the few remaining patches.
Thanks again,
Daniele
2016-10-14 7:37 GMT-07:00 Bhanuprakash Bodireddy <
bhanuprakash.bodire...@intel.com>:
> This patch series is aimed at improving the performance of OVS-DPDK
> dpcls.
2016-10-14 7:37 GMT-07:00 Bhanuprakash Bodireddy <
bhanuprakash.bodire...@intel.com>:
> This patch increases the number of packets processed in a batch during a
> lookup from 16 to 32. Processing batches of 32 packets improves
> performance and also one of the internal loops can be avoided here.
Thanks for v2.
I had a few extra comments on patch one
Sorry for nitpicking on this and sorry I wasn't clear enough on the signoff
requirements.
From CONTRIBUTING.md:
Co-authored-by: Author Name
Git can only record a single person as the author of a
Acked-by: Daniele Di Proietto <diproiet...@vmware.com>
2016-10-13 2:58 GMT-07:00 Bhanuprakash Bodireddy <
bhanuprakash.bodire...@intel.com>:
> Add comments in dp_netdev_input__() to explain the reason behind
> clearing the flow batches before packet_batch_execute
Acked-by: Daniele Di Proietto <diproiet...@vmware.com>
2016-10-13 2:58 GMT-07:00 Bhanuprakash Bodireddy <
bhanuprakash.bodire...@intel.com>:
> Aligning the 'keys' array seems to have positive performance impact.
>
> Signed-off-by: Bhanuprakash Bodireddy <bhanuprakash.b
I think this looks good, thanks.
I have a few comments inline to simplify the code a little bit more.
2016-10-13 2:58 GMT-07:00 Bhanuprakash Bodireddy <
bhanuprakash.bodire...@intel.com>:
> This patch increases the number of packets processed in a batch during a
> lookup from 16 to 32.
2016-10-06 3:25 GMT-07:00 Mark Kavanagh :
> The DPDK pdump sample app was renamed from 'dpdk_pdump' to
> 'dpdk-pdump'. Update references to same within
> INSTALL.DPDK-ADVANCED.md.
>
> Add an additional sample command line that shows how to capture all
> traffic
2016-10-13 10:27 GMT-07:00 Ciara Loftus :
> The rte_pdump header file was not included in the file that requires it.
> Fix this.
>
> Fixes: 01961bbdd34a ("dpdk: New module with some code from netdev-dpdk.")
> Signed-off-by: Ciara Loftus
>
Oops,
for (i = 0; i < n_batches; i++) {
> batches[i].flow->batch = NULL;
> }
>
>
> From: Daniele Di Proietto [mailto:diproiet...@ovn.org]
> Sent: Monday, October 10, 2016 1:31 AM
> To: Fischetti, Antonio <antonio.fische...@intel.com>
> Cc: Jarno Rajahalme <
On 12/10/2016 13:59, "Aaron Conole" <acon...@redhat.com> wrote:
>Daniele Di Proietto <diproiet...@vmware.com> writes:
>
>> There's a lot of code in netdev-dpdk which is not at all related to the
>> netdev interface, mostly the library initialization c
On 12/10/2016 16:23, "Jarno Rajahalme" <ja...@ovn.org> wrote:
>
>> On Oct 4, 2016, at 6:22 PM, Daniele Di Proietto <diproiet...@vmware.com>
>> wrote:
>>
>> We can run out of hugepage memory coming from rte_*alloc() more easily
&g
On 12/10/2016 13:27, "Ben Pfaff" <b...@ovn.org> wrote:
>On Tue, Oct 04, 2016 at 06:22:22PM -0700, Daniele Di Proietto wrote:
>> It is customary to have the vlog module name similar to the filename.
>> Plus a following commit will introduce a 'dpdk' module.
&
On 12/10/2016 12:45, "Ben Pfaff" <b...@ovn.org> wrote:
>On Tue, Oct 04, 2016 at 06:22:14PM -0700, Daniele Di Proietto wrote:
>> The error handling path in dpdk_mp_get() is getting complicated, it
>> even requires a boolean variable.
>>
>
On 12/10/2016 13:04, "Ben Pfaff" <b...@ovn.org> wrote:
>On Tue, Oct 04, 2016 at 06:22:15PM -0700, Daniele Di Proietto wrote:
>> I think it's clearer to use RCU than to check for a pointer twice in the
>> fast path (before and after taking the spinlock). No
On 12/10/2016 12:37, "Ben Pfaff" <b...@ovn.org> wrote:
>On Tue, Oct 04, 2016 at 04:25:35PM -0700, Daniele Di Proietto wrote:
>> dp_netdev_get_pmd() is allowed to return NULL (even if we call it with
>> NON_PMD_CORE_ID) for different reasons:
>>
>>
On 12/10/2016 12:43, "Ben Pfaff" <b...@ovn.org> wrote:
>On Tue, Oct 04, 2016 at 06:22:13PM -0700, Daniele Di Proietto wrote:
>> Looks like we forgot to add the copyright headers to netdev-dpdk.h.
>> Looking at the contribution history of the file, this commit
On 12/10/2016 12:42, "Ben Pfaff" <b...@ovn.org> wrote:
>On Tue, Oct 04, 2016 at 06:22:12PM -0700, Daniele Di Proietto wrote:
>> Looks like we forgot to add the copyright headers to netdev-dpdk.h.
>> Looking at the contribution history of the file, this commit
aking wrong
assumptions.
Thanks,
Daniele
> Antonio
>
> > -Original Message-
> > From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Daniele Di
> > Proietto
> > Sent: Friday, October 7, 2016 11:46 PM
> > To: Jarno Rajahalme <ja...@ovn.org>
>
On 04/10/2016 12:02, "Ben Pfaff" wrote:
>On Mon, Sep 19, 2016 at 04:31:04PM -0300, Thadeu Lima de Souza Cascardo wrote:
>> Use RECHECK=yes for both kernel and userspace datapath tests.
>>
>> Signed-off-by: Thadeu Lima de Souza Cascardo
>
>Looks good to
2016-10-07 14:10 GMT-07:00 Jarno Rajahalme :
>
> > On Oct 7, 2016, at 9:17 AM, Bhanuprakash Bodireddy <
> bhanuprakash.bodire...@intel.com> wrote:
> >
> > Aligning the 'keys' array seems to positively impact performance.
> >
> > Signed-off-by: Bhanuprakash Bodireddy
101 - 200 of 1432 matches
Mail list logo