From: Flavio Leitner
The netdev receiving packets is supposed to provide the flags
indicating if the IP csum was verified and it is GOOD or BAD,
otherwise the stack will check when appropriate by software.
If the packet comes with good checksum, then postpone the
checksum calculation to the
From: Flavio Leitner
The netdev receiving packets is supposed to provide the flags
indicating if the L4 csum was verified and it is OK or BAD,
otherwise the stack will check when appropriate by software.
If the packet comes with good checksum, then postpone the
checksum calculation to the
This is a subset of the larger TSO patchset with various checksumming
improvements. This set includes additional documentation, new appctl
command "dpif-netdev/offload-show" to display interface offload
support, and improvements to tracking when an updated checksum is
required.
Flavio Leitner
From: Flavio Leitner
This patch introduces a command to display the current checksum offload
status by port, allowing the user to gain insight into where checksum
offloading is active.
Signed-off-by: Flavio Leitner
Co-authored-by: Mike Pattrick
Signed-off-by: Mike Pattrick
Reviewed-by:
From: Flavio Leitner
Document the implementation of netdev hardware offloading
in userspace datapath.
Signed-off-by: Flavio Leitner
Co-authored-by: Mike Pattrick
Reviewed-by: David Marchand
Signed-off-by: Mike Pattrick
---
Documentation/automake.mk| 1 +
Hello:
This patch was applied to netdev/net-next.git (master)
by Jakub Kicinski :
On Tue, 18 Oct 2022 02:06:33 -0700 you wrote:
> Round up allocations with kmalloc_size_roundup() so that openvswitch's
> use of ksize() is always accurate and no special handling of the memory
> is needed by KASAN,
On Wed, Oct 19, 2022 at 12:27 PM Ilya Maximets wrote:
>
> On 10/19/22 15:48, Mike Pattrick wrote:
> > On Wed, Oct 19, 2022 at 9:30 AM Flavio Leitner wrote:
> >>
> >>
> >> Hi Mike,
> >>
> >> Thanks for the patch.
> >>
> >> Does this patch need to change this line too?
> >>
> On 10/19/22 23:00, Lorenzo Bianconi wrote:
> >> On 9/14/22 14:42, Lorenzo Bianconi wrote:
> >>> Similar to single load balancers, add the capability to automatically
> >>> deploy a load-balancer group on each logical-switch connected to a
> >>> logical router where the load-balancer group has
On 10/19/22 23:10, Dumitru Ceara wrote:
> On 10/19/22 23:00, Lorenzo Bianconi wrote:
>>> On 9/14/22 14:42, Lorenzo Bianconi wrote:
Similar to single load balancers, add the capability to automatically
deploy a load-balancer group on each logical-switch connected to a
logical router
On 10/19/22 23:00, Lorenzo Bianconi wrote:
>> On 9/14/22 14:42, Lorenzo Bianconi wrote:
>>> Similar to single load balancers, add the capability to automatically
>>> deploy a load-balancer group on each logical-switch connected to a
>>> logical router where the load-balancer group has been
> On 9/14/22 14:42, Lorenzo Bianconi wrote:
> > Similar to single load balancers, add the capability to automatically
> > deploy a load-balancer group on each logical-switch connected to a
> > logical router where the load-balancer group has been installed by the
> > CMS.
> >
> > Reported-at:
On 10/14/22 16:04, Lorenzo Bianconi wrote:
> Rely on the following new actions in order to introduce affinity timeout
> support to load-balancer sessions:
> - commit_lb_aff
> - chk_lb_aff
>
> Introduce the following tables in switch and router pipelines
> respectively:
> -
On 10/14/22 16:04, Lorenzo Bianconi wrote:
> commit_lb_aff action translates to an openflow "learn" action that
> inserts a new flow in the OFTABLE_CHK_LB_AFFINITY table. The new flow is
> used to match on the the 5-tuple and set REGBIT_KNOWN_LB_SESSION bit.
> Moreover the new flow stores backend
On Wednesday, October 19, 2022 10:39:55 AM EDT Eelco Chaudron wrote:
> On 19 Oct 2022, at 12:14, Adrian Moreno wrote:
> > Hi Kevin.
> >
> > Thanks for following up with this patch!
> >
> > On 10/3/22 22:07, Kevin Sprague wrote:
> >> During normal operations, it is useful to understand when a
Previous commit resolves a WARN splat that can be difficult to reproduce,
but with the ovs-dpctl.py utility, it can be trivial. Introduce a test
case which creates a DP, and then downgrades the feature set. This will
include a utility 'ovs-dpctl.py' that can be extended to do additional
work.
As noted by Paolo Abeni, pr_warn doesn't generate any splat and can still
preserve the warning to the user that feature downgrade occurred. We
likely cannot introduce other kinds of checks / enforcement here because
syzbot can generate different genl versions to the datapath.
Reported-by:
Syzbot recently caught a splat when dropping features from
openvswitch datapaths that are in-use. The WARN() call is
definitely too large a hammer for the situation, so change
to pr_warn.
Second patch in the series introduces a new selftest suite which
can help show that an issue is fixed. This
On 10/19/22 15:48, Mike Pattrick wrote:
> On Wed, Oct 19, 2022 at 9:30 AM Flavio Leitner wrote:
>>
>>
>> Hi Mike,
>>
>> Thanks for the patch.
>>
>> Does this patch need to change this line too?
>>
On 9/14/22 14:42, Lorenzo Bianconi wrote:
> Similar to single load balancers, add the capability to automatically
> deploy a load-balancer group on each logical-switch connected to a
> logical router where the load-balancer group has been installed by the
> CMS.
>
> Reported-at:
On 19 Oct 2022, at 15:01, Mike Pattrick wrote:
> Previously the minimum thread stack size was always set to 512 kB to
> help accomidate smaller OpenWRT based systems. Often these devices
> don't have a lot of total system memory, so such a limit makes sense.
>
> The default under x86-64 linux
OVS now just allow the ARP Relpy whitch the destination address is matched
against the known xbridge addresses to udpate tunnel neighbor. So when ovs
receive the gratuitous ARP from underlay gateway which the source address
and destination address are all gateway IP, tunnel neighbor will not be
On Wed, Oct 19, 2022 at 09:48:18AM -0400, Mike Pattrick wrote:
> On Wed, Oct 19, 2022 at 9:30 AM Flavio Leitner wrote:
> >
> >
> > Hi Mike,
> >
> > Thanks for the patch.
> >
> > Does this patch need to change this line too?
> >
On 19 Oct 2022, at 12:14, Adrian Moreno wrote:
> Hi Kevin.
>
> Thanks for following up with this patch!
>
> On 10/3/22 22:07, Kevin Sprague wrote:
>> During normal operations, it is useful to understand when a particular flow
>> gets removed from the system. This can be useful when debugging
On 10/19/22 11:31, Fengqi Li wrote:
> When segmentation fault occured in ovn-northd, the ovn-northd
> daemon process will try to restart for every 10s.
> Remove the segmentation fault condation,the ovn-northd daemon
> process becomes active again,but fatal_signal fds are not recreated
> resulting
On Wed, Oct 19, 2022 at 9:30 AM Flavio Leitner wrote:
>
>
> Hi Mike,
>
> Thanks for the patch.
>
> Does this patch need to change this line too?
> https://github.com/openvswitch/ovs/blob/31db0e043119cf597d720d94f70ec19cf5b8b7d4/rhel/usr_lib_systemd_system_ovs-vswitchd.service.in#L18
>
>
>
Hi Mike,
Thanks for the patch.
Does this patch need to change this line too?
https://github.com/openvswitch/ovs/blob/31db0e043119cf597d720d94f70ec19cf5b8b7d4/rhel/usr_lib_systemd_system_ovs-vswitchd.service.in#L18
Wouldn't it be better to have a config option that we
can change at runtime?
sorry to miss that part,
will take some time to think if we need it or not.
BTW, looks like the kernel handle_upcalls code path has also this issue.
the code is here:
for (i = 0; i < n_ops; i++) {
struct udpif_key *ukey = ops[i].ukey;
if (ukey) {
I'll take a look at that!
thanks.
Eelco Chaudron 于2022年10月19日周三 18:21写道:
>
>
> On 3 Oct 2022, at 6:02, Peng He wrote:
>
> > The following comments (brought in at 0de8783a9):
> >
> > /* XXX: There's a race window where a flow covering this packet
> > * could have already been installed since we
Hi Abhiram,
Thanks for the patch! I only skimmed the changes so this is not a full
review but more of a discussion starter.
On 10/18/22 17:33, Abhiram Sangana wrote:
> To identify connections dropped by ACLs, users can enable logging for ACLs
> but this approach does not scale. ACL logging uses
Previously the minimum thread stack size was always set to 512 kB to
help accomidate smaller OpenWRT based systems. Often these devices
don't have a lot of total system memory, so such a limit makes sense.
The default under x86-64 linux is 2MB, this limit is not always enough
to reach the
On Wed, Oct 19, 2022 at 02:17:42PM +0200, Ilya Maximets wrote:
> On 10/19/22 10:12, Simon Horman wrote:
> > On Fri, Oct 14, 2022 at 10:40:30AM -0400, Jamal Hadi Salim wrote:
> >> On Fri, Oct 14, 2022 at 9:00 AM Ilya Maximets wrote:
> >>>
> >>
> >> [..]
> I thought it was pipe but maybe it is
On 10/19/22 10:12, Simon Horman wrote:
> On Fri, Oct 14, 2022 at 10:40:30AM -0400, Jamal Hadi Salim wrote:
>> On Fri, Oct 14, 2022 at 9:00 AM Ilya Maximets wrote:
>>>
>>
>> [..]
I thought it was pipe but maybe it is OK(in my opinion that is a bad code
for just "count"). We have some (at
On 29 Jul 2022, at 13:16, mit...@outlook.com wrote:
> From: Lin Huang
>
> After user add a static fdb entry, the get_lru() function will always return
> the
> static fdb entry. That's normal fdb entries will not age out through
> mac_learning_run().
>
> Fix the issue by modify the get_lru()
On 10/3/22 22:07, Kevin Sprague wrote:
During normal operations, it is useful to understand when a particular flow
gets removed from the system. This can be useful when debugging performance
issues tied to ofproto flow changes, trying to determine deployed traffic
patterns, or while debugging
On 10 Oct 2022, at 9:12, Eelco Chaudron wrote:
> On 8 Oct 2022, at 5:27, Peng He wrote:
>
>> Hi,Eelco
>>
>> after a second thought, I think this patch is not needed neither,
>> the code here is trying to find a rule which cover the packet,
>> it does not mean the match and action of rule equals
On 3 Oct 2022, at 6:02, Peng He wrote:
> The userspace datapath mananges all the magaflows by a cmap. The cmap
> data structrue will grow/shrink during the datapath processing and it
> will re-position megaflows. This might result in two revalidator threads
> might process a same megaflow
On 3 Oct 2022, at 6:02, Peng He wrote:
> push_dp_ops only handles delete ops errors but ignores the modify
> ops results. It's better to handle all the dp operation errors in
> a consistent way.
>
> We observe in the production environment that sometimes a megaflow
> with wrong actions keep
On 3 Oct 2022, at 6:02, Peng He wrote:
> The following comments (brought in at 0de8783a9):
>
> /* XXX: There's a race window where a flow covering this packet
> * could have already been installed since we last did the flow
> * lookup before upcall. This could be solved by moving the
> *
Hi Kevin.
Thanks for following up with this patch!
On 10/3/22 22:07, Kevin Sprague wrote:
During normal operations, it is useful to understand when a particular flow
gets removed from the system. This can be useful when debugging performance
issues tied to ofproto flow changes, trying to
When segmentation fault occured in ovn-northd, the ovn-northd
daemon process will try to restart for every 10s.
Remove the segmentation fault condation,the ovn-northd daemon
process becomes active again,but fatal_signal fds are not recreated
resulting in ovn-northd 100% cpu utilization.
On 19 Oct 2022, at 1:08, Ilya Maximets wrote:
> On 10/19/22 00:07, Jan Scheurich wrote:
>> Hi guys,
>>
>> I am afraid that commit is too long ago that would remember any details that
>> caused us to change the code in beb75a40fdc2 ("userspace: Switching of L3
>> packets in L2 pipeline"). What
Hi All,
For people interested in how the OVS revalidator process works, here is a nice
article trying to explain it.
https://developers.redhat.com/articles/2022/10/19/open-vswitch-revalidator-process-explained
Cheers,
Eelco
___
dev mailing list
On Tue, Oct 11, 2022 at 11:35:27AM +0200, Simon Horman wrote:
> From: Baowen Zheng
>
> Add tc action flags when adding police action to offload meter table.
>
> There is a restriction that the flag of skip_sw/skip_hw should be same for
> filter rule and the independent created tc actions the
On Fri, Oct 14, 2022 at 10:40:30AM -0400, Jamal Hadi Salim wrote:
> On Fri, Oct 14, 2022 at 9:00 AM Ilya Maximets wrote:
> >
>
> [..]
> > > I thought it was pipe but maybe it is OK(in my opinion that is a bad code
> > > for just "count"). We have some (at least NIC) hardware folks on the list.
>
44 matches
Mail list logo