On Tue, Nov 26, 2019 at 12:46 AM Loftus, Ciara wrote:
>
> >
> > Hi Ciara and Ian,
> >
> > I'm checking the zero-copy support on both rx/tx side when
> > using ovs-dpdk with vhostuser and dpdkport.
> > Assuming PVP, IIUC[1,2], the rx (P to V) does not have zero-copy and
> > the V to P has zero copy
Hi Ciara and Ian,
I'm checking the zero-copy support on both rx/tx side when
using ovs-dpdk with vhostuser and dpdkport.
Assuming PVP, IIUC[1,2], the rx (P to V) does not have zero-copy and
the V to P has zero copy support by enabling dq-zero-copy.
Am I understanding this correctly?
Thank you!
W
be processed in kernel.
The patch enables loading the user-provided XDP program by
$ovs-vsctl -- set int afxdp-p0 options:xdp-obj=
So users can implement their filtering logic or traffic steering idea
in their XDP program, and rest of the traffic passes to AF_XDP socket
handled by OVS.
Signed-off
Hi Ilya and Eelco,
Yiyang reports very poor TCP performance on his setup and I can
also reproduce it on my machine. Somehow I think this might be a
kernel issue, but I don't know where to debug this. Need your suggestion
about how to debug.
So the setup is like the system-traffic, creating 2 name
Make the code simpler by removing the use of
xasprintf and free, and use smap_add_format.
Cc: Ben Pfaff
Signed-off-by: William Tu
---
ofproto/ofproto-dpif.c | 18 --
1 file changed, 4 insertions(+), 14 deletions(-)
diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c
On Wed, Nov 20, 2019 at 05:20:35PM -0800, William Tu wrote:
> On Wed, Nov 20, 2019 at 11:25:12AM -0800, Ben Pfaff wrote:
> > On Wed, Nov 06, 2019 at 12:35:49PM -0800, William Tu wrote:
> > > On Tue, Nov 05, 2019 at 04:49:10PM -0800, Ben Pfaff wrote:
> > > > On Tue, No
On Thu, Nov 21, 2019 at 11:37:48AM +0100, Eelco Chaudron wrote:
>
> On 20 Nov 2019, at 21:25, William Tu wrote:
>
> > libbpf has pr_warn, pr_info, and pr_debug. The patch registers
> > these print functions, integrating the libbpf logs to OVS log.
> >
> > Sign
On Wed, Nov 20, 2019 at 06:00:44PM -0800, Ben Pfaff wrote:
> On Wed, Nov 20, 2019 at 05:20:35PM -0800, William Tu wrote:
> > I think the above is not related. I will remove the manpages.mk diff.
> > Rest below looks good to me.
>
> Yes, thanks!
I fixed a couple minor issue
On Wed, Nov 20, 2019 at 11:25:12AM -0800, Ben Pfaff wrote:
> On Wed, Nov 06, 2019 at 12:35:49PM -0800, William Tu wrote:
> > On Tue, Nov 05, 2019 at 04:49:10PM -0800, Ben Pfaff wrote:
> > > On Tue, Nov 05, 2019 at 02:57:46PM -0800, William Tu wrote:
> > > &
libbpf has pr_warn, pr_info, and pr_debug. The patch registers
these print functions, integrating the libbpf logs to OVS log.
Signed-off-by: William Tu
---
lib/netdev-afxdp.c | 20
lib/netdev-afxdp.h | 1 +
lib/netdev-linux.c | 1 +
3 files changed, 22 insertions(+)
diff
On Tue, Nov 19, 2019 at 10:51:01AM -0800, William Tu wrote:
> Hi Eelco,
>
> Thanks for your testing.
>
> On Tue, Nov 12, 2019 at 11:25:55AM +0100, Eelco Chaudron wrote:
> > See one remark below, however when I did a quick test with a program that
> > would not load it
On Wed, Nov 20, 2019 at 05:09:10PM +0100, Eelco Chaudron wrote:
>
>
> On 20 Nov 2019, at 17:06, Ilya Maximets wrote:
>
> >On 31.10.2019 20:21, William Tu wrote:
> >>The patch detects the numa node id from the name of the netdev,
> >>by reading the '/sys
On Wed, Nov 20, 2019 at 05:43:48PM +0100, Ilya Maximets wrote:
> On 20.11.2019 8:35, Eelco Chaudron wrote:
> >
> >
> > On 19 Nov 2019, at 17:52, Ilya Maximets wrote:
> >
> >> On 19.11.2019 17:16, Eelco Chaudron wrote:
> >>>
> >>>
> >>> On 7 Nov 2019, at 12:36, Ilya Maximets wrote:
> >>>
> U
netdev-afxdp supports different configuration modes, with different
limitations. The patch adds configuration and end-to-end tests using
veth/tap devices.
Signed-off-by: William Tu
---
This patch depends on
[PATCHv2] netdev-afxdp: Best-effort configuration of XDP mode.
---
tests/automake.mk
On Fri, Nov 08, 2019 at 03:23:44PM +, Emma Finn wrote:
> Add an ovs-appctl command to iterate through the dpcls
> and for each subtable output the miniflow bits for any
> existing table.
>
> $ ovs-appctl dpif-netdev/subatable-show
> pmd thread numa_id 0
> dpcls port 2:
> subtable:
>
Hi,
Any feedback on this patch?
Thanks
William
On Thu, Oct 31, 2019 at 12:21:21PM -0700, William Tu wrote:
> The patch detects the numa node id from the name of the netdev,
> by reading the '/sys/class/net//device/numa_node'.
> If not available, ex: virtual device, or any erro
On Tue, Nov 19, 2019 at 09:26:40PM +0100, Ilya Maximets wrote:
> On 19.11.2019 21:09, William Tu wrote:
> > On Tue, Nov 19, 2019 at 08:51:19PM +0100, Ilya Maximets wrote:
> >> On 19.11.2019 20:45, William Tu wrote:
> >>> On Tue, Nov 19, 2019 at 05:52:22PM +0100,
On Tue, Nov 19, 2019 at 12:09:02PM -0800, William Tu wrote:
> On Tue, Nov 19, 2019 at 08:51:19PM +0100, Ilya Maximets wrote:
> > On 19.11.2019 20:45, William Tu wrote:
> > > On Tue, Nov 19, 2019 at 05:52:22PM +0100, Ilya Maximets wrote:
> > >> On 19.11.20
On Tue, Nov 19, 2019 at 08:51:19PM +0100, Ilya Maximets wrote:
> On 19.11.2019 20:45, William Tu wrote:
> > On Tue, Nov 19, 2019 at 05:52:22PM +0100, Ilya Maximets wrote:
> >> On 19.11.2019 17:16, Eelco Chaudron wrote:
> >>>
> >>>
> >
On Tue, Nov 19, 2019 at 05:52:22PM +0100, Ilya Maximets wrote:
> On 19.11.2019 17:16, Eelco Chaudron wrote:
> >
> >
> > On 7 Nov 2019, at 12:36, Ilya Maximets wrote:
> >
> >> Until now there was only two options for XDP mode in OVS: SKB or DRV.
> >> i.e. 'generic XDP' or 'native XDP with zero-co
On Tue, Nov 12, 2019 at 04:46:09AM -0500, Eelco Chaudron wrote:
> Drivers natively supporting AF_XDP will check that a configured MTU size
> will not exceed the allowed size for AF_XDP. However, when the skb
> compatibility mode is used there is no check and any value is accepted.
> This, for examp
Hi Eelco,
Thanks for your testing.
On Tue, Nov 12, 2019 at 11:25:55AM +0100, Eelco Chaudron wrote:
> See one remark below, however when I did a quick test with a program that
> would not load it goes into some re-try loop:
>
> 2019-11-12T10:13:21.658Z|01609|netdev_afxdp|INFO|eno1: Removing xdp p
On Mon, Nov 18, 2019 at 12:32:30PM -0800, William Tu wrote:
> On Mon, Nov 18, 2019 at 12:19:26PM -0800, Yi-Hung Wei wrote:
> > On Mon, Nov 18, 2019 at 12:06 PM Yifeng Sun wrote:
> > >
> > > Current code iterates every kernel under '/lib/modules' for a mat
On Mon, Nov 18, 2019 at 12:19:26PM -0800, Yi-Hung Wei wrote:
> On Mon, Nov 18, 2019 at 12:06 PM Yifeng Sun wrote:
> >
> > Current code iterates every kernel under '/lib/modules' for a matched
> > version. As a result, this script may create invalid soft links if the
> > matched kernel doesn't have
On Thu, Nov 7, 2019 at 3:36 AM Ilya Maximets wrote:
>
> Until now there was only two options for XDP mode in OVS: SKB or DRV.
> i.e. 'generic XDP' or 'native XDP with zero-copy enabled'.
>
> Devices like 'veth' interfaces in Linux supports native XDP, but
> doesn't support zero-copy mode. This ca
On Thu, Nov 07, 2019 at 01:39:28PM +0100, Eelco Chaudron wrote:
>
>
> On 7 Nov 2019, at 12:36, Ilya Maximets wrote:
>
> >Until now there was only two options for XDP mode in OVS: SKB or DRV.
> >i.e. 'generic XDP' or 'native XDP with zero-copy enabled'.
> >
> >Devices like 'veth' interfaces in Li
l build time at an acceptable level only a single build
> job is included in the matrix for ppc64le.
>
> A build report example can be found here [1]
> [0] http://travis-ci.org/
> [1] https://travis-ci.org/djlwilder/ovs/builds/607851729
>
> Signed-off-by: David W
On Fri, Nov 08, 2019 at 05:12:55PM +0100, Matteo Croce wrote:
> Hi,
>
> I need to add a field to enum ovs_action_attr, but I see that the
> definition between the upstream header[1] and the one in compat[2]
> differs.
> Upstream enum stops at OVS_ACTION_ATTR_CHECK_PKT_LEN, with an extra
> "hidden"
Signed-off-by: William Tu
---
v4:
Feedbacks from Eelco.
- First load the program, then configure xsk.
Let API take care of xdp prog and map loading, don't set
XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD.
- When loading custom xdp, need to close(prog_fd) and close(map_fd)
to r
w generic mode in OVS = generic SKB in kernel,
native mode in OVS = native mode without zc...
>
> Since 'best-effort' is a default mode, users will not need to
> explicitely set 'xdp-mode' in most cases.
>
> TCP related tests enabled back in system afxdp t
On Thu, Nov 07, 2019 at 02:28:18PM +0100, Eelco Chaudron wrote:
>
>
> On 7 Nov 2019, at 14:21, Ilya Maximets wrote:
>
> >On 07.11.2019 13:39, Eelco Chaudron wrote:
> >>
> >>
> >>On 7 Nov 2019, at 12:36, Ilya Maximets wrote:
> >>
> >>>Until now there was only two options for XDP mode in OVS: SKB
On Thu, Nov 07, 2019 at 03:01:18PM +0100, Eelco Chaudron wrote:
> Any feedback on this?
>
>
> On 1 Oct 2019, at 11:55, Eelco Chaudron wrote:
>
> >Drivers natively supporting AF_XDP will check that a configured MTU size
> >will not exceed the allowed size for AF_XDP. However, when the skb
> >comp
On Wed, Nov 06, 2019 at 01:47:26PM -0500, Aaron Conole wrote:
> Ilya Maximets writes:
>
> > While running tests under valgrind, it could take more than 10 seconds
> > for process to disappear after successful 'ovs-appctl exit' command.
> >
> > Same applies to some other events that tests are wait
On Tue, Nov 05, 2019 at 05:06:10PM +0100, Eelco Chaudron wrote:
> Hi William,
>
> See some comments inline.
>
> //Eelco
>
>
Hi Eelco,
Thanks for your feedback.
> On 1 Nov 2019, at 21:14, William Tu wrote:
>
> >Now netdev-afxdp always forwards all packets to
On Tue, Nov 05, 2019 at 04:49:10PM -0800, Ben Pfaff wrote:
> On Tue, Nov 05, 2019 at 02:57:46PM -0800, William Tu wrote:
> > +/* ODP_SUPPORT_FIELDS */
> > +str_value = xasprintf("%"PRIuSIZE, odp.max_vlan_headers);
> > +smap_add(cap, "max_vlan
On Wed, Nov 6, 2019 at 10:20 AM Ilya Maximets wrote:
>
> On 06.11.2019 18:56, William Tu wrote:
> > On Wed, Nov 06, 2019 at 01:50:58PM +0100, Ilya Maximets wrote:
> >> On 05.11.2019 22:20, Ben Pfaff wrote:
> >>> On Tue, Nov 05, 2019 at 09:54:08PM +0100, Ilya Ma
On Wed, Nov 06, 2019 at 08:49:53AM +0100, Eelco Chaudron wrote:
>
>
> On 5 Nov 2019, at 23:16, William Tu wrote:
>
> >The RFC patch enables shared umem support. It requires kernel change and
> >libbpf change, I will post it in another thread. I tested with multiple
>
On Wed, Nov 06, 2019 at 01:50:58PM +0100, Ilya Maximets wrote:
> On 05.11.2019 22:20, Ben Pfaff wrote:
> >On Tue, Nov 05, 2019 at 09:54:08PM +0100, Ilya Maximets wrote:
> >>'xdpmode' is part of 'options', not the 'other_config'.
> >>
> &
On Wed, Nov 06, 2019 at 05:53:09PM +0100, Magnus Karlsson wrote:
> On Wed, Nov 6, 2019 at 12:41 AM William Tu wrote:
> >
> > Currently the shared umem feature is not supported in libbpf.
> > The patch removes the refcount check in libbpf to enable use of
> > shared
-November/364392.html
Signed-off-by: William Tu
---
net/xdp/xsk.c | 5 -
tools/lib/bpf/xsk.c | 10 +-
2 files changed, 5 insertions(+), 10 deletions(-)
diff --git a/net/xdp/xsk.c b/net/xdp/xsk.c
index 6040bc2b0088..0f2b16e275e3 100644
--- a/net/xdp/xsk.c
+++ b/net/xdp/xsk.c
On Tue, Nov 05, 2019 at 01:20:35PM -0800, Ben Pfaff wrote:
> On Tue, Nov 05, 2019 at 09:54:08PM +0100, Ilya Maximets wrote:
> > 'xdpmode' is part of 'options', not the 'other_config'.
> >
> > CC: William Tu
> > Fixes: 0de1b425962d ("
e \
ct_orig_tuple=true ct_eventmask=true ct_state=true \
ct_clear=true max_vlan_headers=1 recirc=true ct_label=true \
max_hash_alg=1 ct_state_nat=true ct_timeout=true \
ct_mark=true ct_orig_tuple6=true check_pkt_len=true \
masked_set_action=true max_mpls_depth=3 trunc=true ct_zone=true
Signe
e shared umem for all skb-mode netdevs, and for the rest drv-mode
netdevs, keep using dedicated umem. Or create one umem per mode? So the
drv-mode netdevs also share one umem.
Any comments are welcome.
Suggested-by: Eelco Chaudron
Signed-off-by: William Tu
---
On Mon, Nov 4, 2019 at 2:10 PM Pravin Shelar wrote:
>
> On Mon, Nov 4, 2019 at 6:00 AM William Tu wrote:
> >
> > On Sat, Nov 2, 2019 at 11:50 PM Pravin Shelar wrote:
> > >
> > > On Fri, Nov 1, 2019 at 7:24 AM wrote:
> > > >
> > > >
On Mon, Nov 4, 2019 at 9:05 AM Eelco Chaudron wrote:
>
>
>
> On 4 Nov 2019, at 17:58, Ilya Maximets wrote:
>
> > On 04.11.2019 17:17, William Tu wrote:
> >> On Mon, Nov 4, 2019 at 2:56 AM Eelco Chaudron
> >> wrote:
> >>>
> >>>
On Mon, Nov 4, 2019 at 2:56 AM Eelco Chaudron wrote:
>
> William,
>
> I was wondering if you could add some more details to the documentation
> why it’s not working/supported.
> Currently, all that is in the documentation is this:
>
> “Due to limitations of current upstream kernel, TCP and various
On Sat, Nov 2, 2019 at 11:50 PM Pravin Shelar wrote:
>
> On Fri, Nov 1, 2019 at 7:24 AM wrote:
> >
> > From: Tonghao Zhang
> >
> > The full looking up on flow table traverses all mask array.
> > If mask-array is too large, the number of invalid flow-mask
> > increase, performance will be drop.
>
gt; | affect packet processing performance.
>
> Link:
> https://github.com/openvswitch/ovs/commit/5604935e4e1cbc16611d2d97f50b717aa31e8ec5
> Signed-off-by: Tonghao Zhang
> Tested-by: Greg Rose
> Acked-by: William Tu
> ---
Do you co
AF_XDP veth does not support TCP with namespaces.
This patch skips two cases that use it.
118: conntrack - floating IP
119: conntrack - negative test for recirculation optimization
Signed-off-by: William Tu
---
v2:
- skip only 118 and 119.
v1:
- Tested-at: https://travis-ci.org/williamtu
On Fri, Nov 1, 2019 at 3:10 PM Ilya Maximets wrote:
>
>
>
> On 01.11.2019 22:51, William Tu wrote:
> > AF_XDP veth does not support TCP with namespaces. This patch skips them.
> >
> > Signed-off-by: William Tu
> > ---
> > v1:
> >- Tested-at: h
AF_XDP veth does not support TCP with namespaces. This patch skips them.
Signed-off-by: William Tu
---
v1:
- Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/606194550
- make check-afxdp all pass on my machine, but cirrus, it is still unstable,
observe 1 failed at:
https
Signed-off-by: William Tu
---
v3:
Feedbacks from Eelco.
- keep using xdpobj not xdp-obj (because we alread use xdpmode)
or we change both to xdp-obj and xdp-mode?
- log a info message when using external program for better debugging
- combine some failure messages
- upd
o use case.
Signed-off-by: William Tu
Acked-by: Eelco Chaudron
---
v6:
Feedbacks from Ilya
- add thread safety check at netdev_linux_get_numa_id__, and
pass netdev_linux
- preserve numa cache on netlink updates
- Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/605
On Wed, Oct 30, 2019 at 10:16:58AM -0700, Ben Pfaff wrote:
> On Mon, Oct 28, 2019 at 09:39:55PM -0700, Zhiwei Cen wrote:
> > It seems I can change the limit for Openflow flows.
> > This works:
> > $ovs-vsctl --no-wait set Open_vSwitch . other_config:flow-limit=21
> >
> > $ovs-vsctl --no-wait g
On Wed, Oct 30, 2019 at 09:18:29AM -0700, Gregory Rose wrote:
>
> On 10/28/2019 9:39 PM, Zhiwei Cen wrote:
> >Dear ovs-dev,
> >
> >I'm trying to see if I can increase the number of flows supported in
> >kernel cache. I found two relevant config options
> >(other_config:flow-limit and upcall/flow l
> > The root cause: Every change in the state of the network interface
> > of a linux kernel device generates if-notifier event and if-notifier
> > event triggers the OVS code to re-apply the configuration of ports,
> > i.e. add broken ports back. The most obvious part is that dpif-netdev
> > chang
On Tue, Oct 29, 2019 at 9:15 AM Ilya Maximets wrote:
>
> On 23.10.2019 23:08, William Tu wrote:
> > The patch detects the numa node id from the name of the netdev,
> > by reading the '/sys/class/net//device/numa_node'.
> > If not available, ex: virtual device
k cirrus to gcp.
https://cirrus-ci.org/guide/supported-computing-services/#google-cloud
Then replaces the gcp_credentials in this patch with your own.
Example results:
https://cirrus-ci.com/task/4880415994675200
Signed-off-by: William Tu
---
.cirrus
On Tue, Oct 29, 2019 at 4:20 AM Noa Levy wrote:
>
>
>
> > -Original Message-
> > From: William Tu [mailto:u9012...@gmail.com]
> > Sent: Monday, October 28, 2019 10:46 PM
> > To: Noa Levy
> > Cc: ovs-dev@openvswitch.org; Oz Shlomo ; Majd
> >
Hi Eelco,
Thanks for your review.
On Tue, Oct 29, 2019 at 8:28 AM Eelco Chaudron wrote:
>
> See comments inline…
>
> On 5 Oct 2019, at 2:12, William Tu wrote:
>
> > Now netdev-afxdp always forwards all packets to userspace because
> > it is using libbpf
On Mon, Oct 28, 2019 at 12:11 PM Ilya Maximets wrote:
>
> On 28.10.2019 11:46, Eelco Chaudron wrote:
> >
> >
> > On 23 Oct 2019, at 23:06, William Tu wrote:
> >
> >> The patch adds support for using need_wakeup flag in AF_XDP rings.
> >> A new opti
Hi Noa,
Thanks for your reply.
> > > > Hi Noa,
> > > >
> > > > I have a couple more questions. I'm still at the learning stage of
> > > > this new feature, thanks in advance for your patience.
> > > >
> > > > On Thu, Oct 17, 2019 at 02:16:56PM +0300, Noa Ezra wrote:
> > > > > dpdkvdpa netdev work
eptember/363126.html
> Signed-off-by: Ilya Maximets
> ---
LGTM.
Thanks for the fix
Acked-by: William Tu
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
Hi Noa,
I have a couple more questions. I'm still at the learning stage of this
new feature, thanks in advance for your patience.
On Thu, Oct 17, 2019 at 02:16:56PM +0300, Noa Ezra wrote:
> dpdkvdpa netdev works with 3 components:
> vhost-user socket, vdpa device: real vdpa device or a VF and
> r
On Wed, Oct 16, 2019 at 04:13:14PM +, Loftus, Ciara wrote:
> >
> > The patch adds support for using need_wakeup flag in AF_XDP rings.
> > A new option, use_need_wakeup, is added. When this option is used,
> > it means that OVS has to explicitly wake up the kernel RX, using poll()
> > syscall a
o use case.
Signed-off-by: William Tu
---
v5:
Feedbacks from Ilya
- reafactor the error handling
- add mutex lock
- Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/601947245
v4:
Feedbacks from Eelco
- Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/599
nce
the virtual device's AF_XDP xmit is synchronous: the sendto syscall
enters kernel and process the TX packet on tx queue directly.
On Intel Xeon E5-2620 v3 2.4GHz system, performance of physical port
to physical port improves from 6.1Mpps to 7.3Mpps.
Suggested-by: Ilya Maximets
Signed-off-by
On Mon, Oct 21, 2019 at 03:18:51PM +0200, Ilya Maximets wrote:
> On 17.10.2019 22:08, William Tu wrote:
> >The patch detects the numa node id from the name of the netdev,
> >by reading the '/sys/class/net//device/numa_node'.
> >If not available, ex: virtual device,
On Tue, Oct 22, 2019 at 12:17 AM Noa Ezra wrote:
>
> Hi,
> Please see the answer below.
>
> Thanks,
> Noa.
>
> > -Original Message-
> > From: William Tu [mailto:u9012...@gmail.com]
> > Sent: Friday, October 18, 2019 12:34 AM
> > To: Noa Ezr
On Mon, Oct 21, 2019 at 6:16 PM Tonghao Zhang wrote:
>
> On Tue, Oct 22, 2019 at 1:14 AM William Tu wrote:
> >
> > On Wed, Oct 16, 2019 at 5:50 AM wrote:
> > >
> > > From: Tonghao Zhang
> > >
> > > This series patch optimi
> > Hi Tonghao,
> >
> > Does this improve performance? After all, the original code simply
> > check whether the mask is NULL, then goto next mask.
> I tested the performance, but I disable the mask cache, and use the
> dpdk-pktgen to generate packets:
> The test ovs flow:
> ovs-dpctl add-dp system
On Wed, Oct 16, 2019 at 5:50 AM wrote:
>
> From: Tonghao Zhang
>
> This series patch optimize openvswitch for performance or simplify
> codes.
>
> Patch 1, 2, 4: Port Pravin B Shelar patches to
> linux upstream with little changes.
btw, should we keep Pravin as the author of the above three patc
On Fri, Oct 18, 2019 at 05:14:46PM +0200, Ilya Maximets wrote:
> On 26.09.2019 21:29, William Tu wrote:
> >The patch adds support for using need_wakeup flag in AF_XDP rings.
> >A new option, use_need_wakeup, is added. When this option is used,
> >it means that OVS has to e
On Thu, Oct 17, 2019 at 01:38:08PM -0700, Yi-Hung Wei wrote:
> On Thu, Oct 17, 2019 at 12:56 PM William Tu wrote:
> >
> > The patch add libunwind dev package to debian and rhel.
> >
> > Signed-off-by: William Tu
> > ---
>
> Looks good to me.
>
&
> Tested-by: Greg Rose
> ---
LGTM
Acked-by: William Tu
On the other hand, maybe we should have an upper limit on the
mask cash size?
Regards,
William`
> net/openvswitch/flow_table.c | 33 +++--
> 1 file changed, 23 insertions(+), 10 deletions(-)
>
> d
g
> Tested-by: Greg Rose
> ---
LGTM
Acked-by: William Tu
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
masks can
> | affect packet processing performance.
>
> Link:
> https://github.com/openvswitch/ovs/commit/5604935e4e1cbc16611d2d97f50b717aa31e8ec5
> Signed-off-by: Tonghao Zhang
> Tested-by: Greg Rose
> ---
LGTM
Acked-by: William Tu
___
On Wed, Oct 16, 2019 at 5:56 AM wrote:
>
> From: Tonghao Zhang
>
> use the specified functions to init resource.
>
> Signed-off-by: Tonghao Zhang
> Tested-by: Greg Rose
> ---
Looks like this is simply moving code around.
I don't have any opinion.
> net/openvswitch/datapath.c | 60
>
t OvS recirc_id from tc chain index")
> Cc: Paul Blakey
> Signed-off-by: Tonghao Zhang
> Tested-by: Greg Rose
> ---
LGTM
Acked-by: William Tu
> net/openvswitch/datapath.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/net/openvswit
On Wed, Oct 16, 2019 at 5:55 AM wrote:
>
> From: Tonghao Zhang
>
> The most case *index < ma->max, and flow-mask is not NULL.
> We add un/likely for performance.
>
> Signed-off-by: Tonghao Zhang
> Tested-by: Greg Rose
> ---
LGTM
Acked-by: William Tu
&g
On Wed, Oct 16, 2019 at 5:54 AM wrote:
>
> From: Tonghao Zhang
>
> Simplify the code and remove the unnecessary BUILD_BUG_ON.
>
> Signed-off-by: Tonghao Zhang
> Tested-by: Greg Rose
> ---
LGTM
Acked-by: William Tu
> net/openvswitch/flow_table.c | 8 ++--
>
On Wed, Oct 16, 2019 at 5:54 AM wrote:
>
> From: Tonghao Zhang
>
> The full looking up on flow table traverses all mask array.
> If mask-array is too large, the number of invalid flow-mask
> increase, performance will be drop.
>
> One bad case, for example: M means flow-mask is valid and NULL
> o
On Thu, Oct 17, 2019 at 02:16:56PM +0300, Noa Ezra wrote:
Hi Noa,
Thanks for the patch. I'm new to this and have a question below.
> dpdkvdpa netdev works with 3 components:
> vhost-user socket, vdpa device: real vdpa device or a VF and
> representor of "vdpa device".
>
> In order to add a new
o use case.
Signed-off-by: William Tu
---
v4:
Feedbacks from Eelco
- Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/599308893
v3:
Feedbacks from Ilya and Eelco
- update doc, afxdp.rst
- fix coding style
- fix limit of numa node max, by using ovs_numa_numa_id_is_valid
e \
ct_orig_tuple=true ct_eventmask=true ct_state=true \
ct_clear=true max_vlan_headers=1 recirc=true ct_label=true \
max_hash_alg=1 ct_state_nat=true ct_timeout=true \
ct_mark=true ct_orig_tuple6=true check_pkt_len=true \
masked_set_action=true max_mpls_depth=3 trunc=true ct_zone=true
Signed-off
The patch add libunwind dev package to debian and rhel.
Signed-off-by: William Tu
---
debian/control | 3 ++-
rhel/openvswitch.spec.in | 1 +
2 files changed, 3 insertions(+), 1 deletion(-)
diff --git a/debian/control b/debian/control
index ad4b187860f3..de092b898cd7 100644
--- a
On Tue, Oct 15, 2019 at 01:45:19PM +0200, Eelco Chaudron wrote:
> See some comment below…
Thank you. Sorry for my late response.
>
> On 2 Oct 2019, at 22:26, William Tu wrote:
>
> >The patch detects the numa node id from the name of the netdev,
> >by reading the
On Wed, Oct 16, 2019 at 01:26:39PM -0700, Yifeng Sun wrote:
> LGTM, thanks Greg.
>
> Reviewed-by: Yifeng Sun
>
Thanks! Applied to master.
Sorry Yifeng, I forgot to add your Reviewed-by tag.
William
> On Wed, Oct 16, 2019 at 1:21 PM Greg Rose wrote:
> >
> > In commit 057772cf2477 the functio
On Wed, Oct 16, 2019 at 9:36 AM Gregory Rose wrote:
>
>
> On 10/15/2019 2:06 PM, Gregory Rose wrote:
>
>
> On 10/4/2019 3:32 PM, William Tu wrote:
>
> The patch adds support for fetching the datapath's capabilities
> from the result of 'check_support()',
On Wed, Oct 16, 2019 at 09:35:55PM -0700, Yi-Hung Wei wrote:
> It is possible that user install libunwind but not libunwind-devel,
> and it will run into a compilation error. So we need to check the
> existence of the library and the header file.
>
> Fixes: e2ed6fbeb18c ("fatal-signal: Catch SIGS
On Wed, Oct 16, 2019 at 9:13 AM Loftus, Ciara wrote:
>
> >
> > The patch adds support for using need_wakeup flag in AF_XDP rings.
> > A new option, use_need_wakeup, is added. When this option is used,
> > it means that OVS has to explicitly wake up the kernel RX, using poll()
> > syscall and wake
On Fri, Oct 11, 2019 at 01:16:55PM +0200, Ilya Maximets wrote:
> On 10.10.2019 20:21, William Tu wrote:
> >On Wed, Oct 09, 2019 at 04:17:58PM +0200, Ilya Maximets wrote:
> >>Later version of 'struct xsk_umem_config' contains additional field
> >>'f
On Fri, Oct 11, 2019 at 01:17:59PM +0200, Ilya Maximets wrote:
> On 10.10.2019 20:29, William Tu wrote:
> >On Wed, Oct 09, 2019 at 04:23:31PM +0200, Ilya Maximets wrote:
> >>Any type of AF_XDP socket in all modes implies creation of BPF map of
> >>type BPF_MAP_
On Wed, Oct 09, 2019 at 04:18:38PM -0700, William Tu wrote:
> On Wed, Oct 09, 2019 at 11:20:00AM -0700, Ben Pfaff wrote:
> > On Wed, Oct 09, 2019 at 04:36:49PM +0200, Ilya Maximets wrote:
> > > New 'make' target 'check-afxdp-valgrind'.
> > >
>
On Wed, Oct 09, 2019 at 04:23:31PM +0200, Ilya Maximets wrote:
> Any type of AF_XDP socket in all modes implies creation of BPF map of
> type BPF_MAP_TYPE_XSKMAP. This leads to BPF_MAP_CREATE syscall and
> subsequently 'xsk_map_alloc()' function that will charge required
> memory from the memlock
On Wed, Oct 09, 2019 at 04:17:58PM +0200, Ilya Maximets wrote:
> Later version of 'struct xsk_umem_config' contains additional field
> 'flags'. OVS doesn't use that field passing uninitialized stack
> memory to the 'xsk_umem__create()' call that could fail with
> 'Invalid argument' if 'flags' are
On Wed, Oct 09, 2019 at 09:44:14AM -0700, Ben Pfaff wrote:
> On Wed, Oct 09, 2019 at 05:10:08PM +0300, Alin Gabriel Serdean wrote:
> > This patch fixes the appveyor build by adding the python version 3 to path
> > as per:
> > https://www.appveyor.com/docs/windows-images-software/#python
> >
> > We
On Wed, Oct 09, 2019 at 11:20:00AM -0700, Ben Pfaff wrote:
> On Wed, Oct 09, 2019 at 04:36:49PM +0200, Ilya Maximets wrote:
> > New 'make' target 'check-afxdp-valgrind'.
> >
> > Signed-off-by: Ilya Maximets
>
> Makes sense to me. I did not test it.
>
> Acked-by: Ben Pfaff
Thanks, I tested th
On Tue, Oct 08, 2019 at 10:04:17AM -0700, David Wilder wrote:
> Since commit e2ed6fbeb1, Ci on ppc64le with Ubuntu 16.04.6 LTS throws
> this error:
>
> lib/fatal-signal.c: In function 'send_backtrace_to_monitor':
> lib/fatal-signal.c:168:9: error: variable 'dep' might be clobbered by
> 'longjmp' o
is new structure.
> Removed redundant conversions between network and host representations.
> Now fields are always in be.
>
> This also, probably, makes flow_compose_l4 more robust by avoiding
> possible unaligned accesses to 32 bit value.
>
> Fixes: 9b2b84973db7 (&quo
601 - 700 of 1254 matches
Mail list logo
