vpp-dev@lists.fd.io
> Subject: Re: [vpp-dev] help with review
>
> Hi Benoit, A new one also need your help to review
> https://gerrit.fd.io/r/c/vpp/+/38135 , thanks a lot.
>
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Benoit Ganne
> (bganne
Hi, thanks for the head's up.
> Please help to review below patches again, all your comments
> were addressed. Thanks a lot.
> https://gerrit.fd.io/r/c/vpp/+/38008
Merged!
> https://gerrit.fd.io/r/c/vpp/+/38009
A few minor comments left on that one, but then we should be good to merge.
Thanks,
Hi Laszlo,
> Could you please point me to some description of how to report security
> vulnerability on vpp? This time I do not have any, just for documentation
> purposes.
The process is documented here:
https://wiki.fd.io/view/TSC:Vulnerability_Management
Best
ben
-=-=-=-=-=-=-=-=-=-=-=-
Lin
I rebased it to make sure it still passes CI, and once done we'll merge it as I
see Hongjun Ni did +1 it and is listed as maintainer.
Thanks for the heads up.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Nobuhiro MIKI
> Sent: Wednesday, January 18, 2023 10:52
>
> I could get LACP bonding working with RDMA driver now. I was going through
> the RDMA plugin code and found there is a concept of "mode" in RDMA
> interfaces. When I modified the interface command like below, I could see
> the LACP bonding working:-
> vpp# create int rdma host-if eth1 mode ibv
>
Do you use SRIOV VFs? If so make spoof-checking etc. is off. See
https://s3-docs.fd.io/vpp/23.02/developer/devicedrivers/rdma.html for more
details.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of r...@gmx.net
> Sent: Friday, January 6, 2023 15:38
> To: vpp-dev@l
Interesting! Thanks Fan for bringing that up.
So if I understand correctly, with the previous DPDK behavior we could have say
128 packets in the rxq, VPP would request 256, get 32, and the request 224
(256-32) again, etc.
While VPP request more packets, the NIC has the opportunity to add packets
> Hi, I am looking for veth pair alternative in VPP. I want to pass traffics
> between two VRFs and I want them to be processed as input packets to an
> interface( to use NAT or other ip features), so route leakage is not gonna
> help. Also, since I'm using one instance of VPP with multiple VRFs, u
The issue seems to be that the main thread wants to reply to an ARP, but it
timeouts on the worker barrier:
1) why the worker does not yield to the main thread in a timely manner:
workers should always complete processing in less than 1s. You can try to use
elog to identify which nodes takes to
> I am using GCC version 8.5.0 20210514 (Red Hat 8.5.0-17) (GCC) which is
> upgraded and has the above errors.
> How it can be fixed?
My point was to use either a newer version of GCC or LLVM - AddressSanitizer is
quite demanding for the compiler, the newest the better. I think CentOS
maintains
I see you run the VPP worker on core 6, is it on the same NUMA node as the NIC?
You can check with 'sh pci' and 'sh thr'.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of r...@gmx.net
> Sent: Friday, December 30, 2022 20:05
> To: vpp-dev@lists.fd.io
> Subject: [vpp
Which compiler do you use? Try upgrading to a newer toolchain or switch between
GCC and LLVM.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of
> 2201kamalj...@gmail.com
> Sent: Tuesday, January 3, 2023 8:09
> To: vpp-dev@lists.fd.io
> Subject: [vpp-dev] Unable to r
> We want to push some monitoring information from vpp to govpp based on
> some condition. Please help me to know if there is any procedure through
> which we can meet this requirement as there is no request from agent in
> this case.
You can register for events, see src/vnet/ip-neighbor/ip_neighb
> Basically it looks like it is lcp problem and routing - dont know how this
> tests are done for lcp but it looks like those tests are like
> 1. load 900k routes
> 2. connect to traffic generator - and push 10Mpps (and dont check if You
> have any reply) or more from one single ip src to one singl
Hi,
> So the hardware is:
> Intel 6246R
> 96GB ram
> Mellanox Connect-X 5 2x 100GB Ethernet NIC
> And simple configuration with vpp/frr where one vlan interface all
> traffix is RX-ed and second vlan interface where this traffic is TX-ed -
> it is normal internet traffic - about 20Gbit/s with
API and CLI are executed under the worker barrier by default in VPP, only the
VPP main thread runs, and workers are stopped.
For an API/CLI to be executed concurrently with workers, it must be explicitly
marked as such (mp_safe attribute).
It does not seem to be the case for ACL, so it should be
I'm no expert in LACP so maybe someone can jump in to confirm whether it looks
good on VPP side, but as we're receiving LACP packets and VPP seems to be
happy, I wonder if the issue could be VPP failing to send LACP packets back to
the switch? Eg. because the NIC eat them up for some reason (sec
Hi,
I think the 1st thing to try would be whether you can reproduce it with a debug
build. The backtrace you have is not usable unfortunately.
The easiest should be to build debug binaries:
~# make rebuild
~# ./build-root/install-vpp_debug-native/vpp/bin/vpp -c /etc/vpp/startup.conf
Best
Ben
>
Hi everyone,
I'd like to get the discussion started on the topic of bound checking in VPP:
some of us recently discussed a patch that added explicit bound checking within
VPP core dataplane infrastructure to prevent a crash when the function is
misused from a plugin. The bug is clearly in the p
Hi Zhang,
I'd expect the interface deletion to happen under the worker barrier. VPP
workers should drain all their in-flight packets before entering the barrier,
so it should not be possible for the interface to disappear between your node
and ip4-lookup. Or am I missing something?
What I have
/device/device.c:453
> #9 0x7f2e0c243a15 in dispatch_pending_node
> (vm=,
> pending_frame_index=,
> last_time_stamp=)
> at /usr/src/debug/vpp-1.0.0/src/vlib/main.c:1024
> #10 0x7f2e0c2454ff in vlib_worker_loop
> (vm=v
Did you installed vpp from packages? If so, you'll need the vpp-dbg package.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Venkataraju,
> Raghu via lists.fd.io
> Sent: Monday, November 28, 2022 14:03
> To: vpp-dev@lists.fd.io
> Subject: [vpp-dev] VPP - 'No debugg
> Neale's suggestion is very usefull, the sa deleting crash was solved by
> deleting policy and sa in two barrier calls.
But is it a bug you see in the open source VPP ikev2 plugin? I'm asking because
if so, we need to fix it...
Best
ben
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages s
I agree with Neale.
I looked at the ike plugin and we seem to always manipulate SA under the worker
barrier, which should guarantee no workers are active. If not, we need to fix
the bug in ike.
You mentioned you rewrote the ike plugin, which changes did you made?
Best
ben
> -Original Messag
+1
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Andrew
> Yourtchenko
> Sent: Tuesday, November 22, 2022 17:29
> To: vpp-dev
> Subject: [vpp-dev] RFC: move the “1”st instance of the community meeting 3
> hours earlier to reduce impact for earlier time zone
>
> Hi all,
>
Hi Elias,
Sorry, this slipped through my mind!
I do not have any Cx6 to test (I think we should receive some in CSIT at some
point), but as it seems to complain about the QP type 8 which is supposed to be
the ethernet queue type, can you check if your adapter supports Ethernet and if
so, if it
Hi,
Note we also asked for sh tr, sh hard, sh err, sh log and sh lacp but if I'm
not mistaken, we did not saw anything.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Chinmaya
> Aggarwal
> Sent: Monday, November 21, 2022 20:04
> To: vpp-dev@lists.fd.io
> Subject
> As suggested we tried putting the rdma interfaces in promiscuous mode, but
> still facing the same issue:-
> vpp# set interface promiscuous on eth1/0 vpp# set interface promiscuous on
> eth2/1
> What could be the possible reason for this issue?
Can you share a packet trace?
vpp# cle tr
vpp# tr a
Hello,
> In our intentions:
> Our process --> Memif --> VPP (TCP to eth) --> ??? custom plug-in -->
> Our radio stack (including driver)
> I walked through the VPP docs, but it is not clear to me if there is
> already something that can let us interface with the bottom of the stack.
From what I u
Hello,
Hard to tell without info about the various nodes etc.
The best should be to look at the "show run" output:
https://s3-docs.fd.io/vpp/23.02/gettingstarted/troubleshooting/cpuusage.html#vpp-cpu-load
When traffic run through, do "clear run" and then "show run" in the VPP cli to
see where cy
Hi,
I'm not very familiar with bonding but maybe it's an issue with mac filtering?
Can you try putting the rdma interfaces in promiscuous mode, to see if it is
any better?
vpp# set interface promiscuous on eth1
vpp# set interface promiscuous on eth2
best
ben
> -Original Message-
> From
>>> Issue #1: VPP crashes when scales to multiple worker cores/threads.
[...]
>>> We guess VPP buffer metadata has been corrupted in the case of multiple
>>> worker threads using VPP release version
>> BG: do you reproduce the issue on both x86 and ARM? Or only on ARM? I’m
>> especially thinking a
Hi Jieqiang,
Thanks a lot for your report! CC’ing vpp-dev because that may be of interest to
others, and also you can get better support.
My comments/questions inline prefixed with BG: in red
Best
ben
From: Jieqiang Wang
Sent: Monday, November 7, 2022 9:25
To: Benoit Ganne (bganne) ; moham...@
Hi Petr,
Unfortunately I can't confirm the issue, but I do not think it is a very
commonly tested scenario so it is not really surprising...
What could help understand the issue would be if you could capture the RA
traffic in both scenarios (VPP vs radvd) so we can compare.
Best
ben
> -Ori
Oct 2022, 18:26 Benoit Ganne (bganne) via lists.fd.io
> <http://lists.fd.io> , <mailto:cisco@lists.fd.io> > wrote:
>
>
> I did not heard anything like this.
> Can you try to reproduce with latest master? Do you have some
> proprietary
As you'd have to poll SA timestamp counters anyway to check the timestamp, why
not just poll existing SA bytes/packets counters instead to see if some traffic
is going through?
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Guangming
> Sent: Friday, October 14,
I did not heard anything like this.
Can you try to reproduce with latest master? Do you have some proprietary
plugins loaded also?
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Prashant
> Upadhyaya
> Sent: Wednesday, October 12, 2022 11:32
> To: vpp-dev
> Subje
Hi Jeff,
> Hello, I tried to join the VPP meeting this morning on Zoom but it
> required login. I tried creating an account using my company email, but
> this was somehow restricted.
> Can the Zoom meeting be updated to not require login?
You should be able to connect to https://zoom.us/my/fastda
Hi,
Yes, you can look at src/plugins/pppoe/ which does exactly that for example,
esp. src/plugins/pppoe/pppoe_decap.c:pppoe_input_node()
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Burcu YUKSEL
> Sent: Tuesday, October 11, 2022 8:17
> To: vpp-dev@lists.fd.io
Hi Pim,
Looking at https://doc.dpdk.org/guides/nics/dpaa2.html, it looks like it does
not use the PCIe bus but something specific called the "FSL MC bus".
Normally, VPP DPDK plugin should build a list of eligible devices at startup,
by scanning PCIe or VMBus (HyperV virtualized bus) but it does
> From a old message, you saidthat set no-multi-seg will get good
> performance . https://lists.fd.io/g/vpp-dev/message/18489
> My question is when i do not neeed Jumbo MTU in my scenario, is set
> no-multi-seg and no-tx-checksum-offload in startup.conf a good option
It depends from the d
You should be able to use ethernet_register_input_type() to add support for
your custom L2.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Burcu YUKSEL
> Sent: Friday, October 7, 2022 18:15
> To: vpp-dev@lists.fd.io
> Subject: [vpp-dev] How to forward masked IP p
> So there is no way to police per IPs.
> In other words to incorporate separate limiter (1.0.0.1, 1.0.0.2, 1.0.0.3
> - 60in/20out, 1.0.0.10, 1.0.0.20, 1.0.0.30 - 90in/80out etc.) to all users
> behind interface?
I do not think so.
Best
ben
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages
AFAICT, output policer is only supported per interface, not per flow.
You should be able to apply an output policer on an interface with the command
'policer output name '.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of ??
> Sent: Monday, October 3, 202
Hi Wentian, Xiaodong,
When testing VM-to-VM iperf (ie TCP) throughput like Wentian does, the most
important factor is whether GSO is turned on: when using Linux as a router, it
is by default whereas when using VPP as a router it is not.
GSO means Linux is going to use 64K bytes TCP packets where
Looks like it would be a good time to unify slow-path/fastpath and
inbound/outbound policy matching code 😊
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Zachary Leaf
> Sent: Wednesday, September 28, 2022 10:10
> To: vpp-dev@lists.fd.io
> Cc: Zhang, Roy Fan ; Bro
I suspect the issue is that you let the IKEv2 plugin create the ipip0 interface
automatically for you: by default it will use table 0 to reach its peer.
When you set Radio-0 in table 1, ipip0 can no longer reach its peer by looking
up the next hop in table 0.
Instead, try to create the ipip inter
> Currently just playing with some settings and checking what I can show for
> debug when I will test it again - because now it is a little hard to debug
> when there is no info how much load traffic is doing to cpu's (for example
> cpu cycles etc... something like perf top)
It is the purpose of t
5 QSFP28
> part number: MCX516A-CCAT
> revision: AC
> serial number: MT2024K11216
> flags: admin-up link-up mlx5dv
> rss: ipv4-tcp ipv6-tcp
> mlx5: version 0
> device flags: cqe-v1 mpw-allowed enhanced-mpw cqe-128b-comp cqe-128b-
> pad
> vpp#
>
>
This is what we measure on CSIT with Intel Cascade Lake for 1 CPU core:
http://csit.fd.io/trending/#eNrVlMFuwyAMhp8mvVSWwC3JduihXd6jYuAu0UjKgFbtnn6kyuRE2ibtUm0HjNBn-8e_EDEdA-0juU2hdkW1K7BqbQ7FarvM29l7wB6Mu4AU4oXQS3NRqtQQbKeh9Wso188gDVBqhlNe0WhHKF4h9Ba6EIZu-DR0s6c0a83EN1cm3wtygQ6kuSKrMkoUJzJf34uzD0F
> It's just a notice, that I reworked a minor fix for build on openSUSE.
> Excuse me if I confused the reviewer with my updates on this commit.
> The link for gerrit review: https://gerrit.nordix.org/c/fdio/vpp/+/14826
Hi Laszlo,
Thanks but the link points to gerrit.nordix.org which is not fd.io
> I did poke around hash.h but didn’t find the api I need. Would vpp
> consider adding crc32 and crc16? ECMP and Network Packet Broker use
> hashes.
We typically use crc32c if supported by HW or xxhash, see
clib_bihash_hash_16_8_32() for example.
For ECMP we also have ip4_compute_flow_hash() and
vlib_buffer_validate_alloc_free() should only be called with debug builds to
debug buffer allocation issues.
Debug builds are slow, because there is no optimizations and lots of additional
checks.
Please make sure you are using a release build before benchmarking (eg. make
build-release).
Best
Merged, thanks Atzm!
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of
> atzm...@gmail.com
> Sent: Monday, August 8, 2022 10:18
> To: vpp-dev@lists.fd.io
> Subject: [vpp-dev] ikev2 rekeying with multiple notify payloads
>
> Hi,
>
> In RFC 7296, Child SA rekey using
Can you share the output of 'cat /proc//stack'?
This should tell us where the worker is stuck in the kernel.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Satish Singh
> Sent: Saturday, August 6, 2022 20:36
> To: vpp-dev@lists.fd.io; vpp-dev+h...@lists.fd.io
> Su
Hi Saravana,
As far as I know, it was lack of maintenance (by lack of interest from the
community). There was no maintainer, and the code did not keep up and was
broken.
For it to re-instantiated, it would need to be updated for current master, and
someone should volunteer to maintain it.
Best
> I am wondering if VPP could allow
> VLIB_BUFFER_POOL_PER_THREAD_CACHE_SZ 512 to be changed to a build time
> value as done in DPDK ?
It is a define, you can change it - but I do not think it has been tested 😊.
Maybe we can make it configurable through cmake also.
> Since in this specific use c
t; buffer，
> can vpp use dpdk's rte_mbuf directly?
>
> best
> wanghe
>
> Benoit Ganne (bganne) via lists.fd.io <http://lists.fd.io>
> mailto:cisco@lists.fd.io> > 于2022年8月
> 1日周一 15:08写道：
>
>
> You probably want to use vlib_buffe
You probably want to use vlib_buffer_clone() instead.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of NUAA??
> Sent: Monday, August 1, 2022 4:53
> To: vpp-dev@lists.fd.io
> Subject: [vpp-dev] Copy packet forwarding performance issues
>
> Hi，vpp experts
> I have a
> Thank you very much for your message. Is there anyway figure out the
> buffer allocation cross cpu spin-lock contention?
I'd expect it to appear during profiling if it becomes an issue.
> Are there any test results that you could point us to?
Not that I know of.
ben
-=-=-=-=-=-=-=-=-=-=-=-
L
I agree that having all allocations on 1 thread (rx) and all de-allocations on
other threads (tx) is not ideal.
However, if you can reserve enough memory, I suspect it might be easier to just
*increase* the per-thread buffer cache.
The per-thread buffer cache basically batches spinlock operations
The memory traces track the memory allocated by backtrace, ie the functions
call chain that leads to the allocation.
Traceback is the backtrace similar to what you get in a debugger.
Bytes is the total memory currently allocated by this specific backtrace.
Count is the number of active allocations
Not sure if that is what you are looking for, but in that case, I just use
plain telnet to connect to VPP, as "telnet localhost 5002".
No need for vppctl in that case.
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Amaresh
> Parida
> Sent: Saturday, July 16, 2022 4:32
> Thanks for the info. I am not too familiar with rdma interfaces, well at
> least the way vpp is utilizing them. It seems its polling the interfaces
> just like dpdk.
Polling is the default for both dpdk and rdma interfaces. You can set them to
interrupt mode or adaptive mode (automatically swit
> Could you provide something about your environment if that is convenient,
> such as gcc or clang version, linux release or version,
> and I will try it later,
I'm using clang-14 on ubuntu 22.04.
Best
ben
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Onl
This works fine for me (see output excerpt below)...
This might be an issue with your toolchain, try to upgrade clang/gcc.
bganne@vm:~/src/vpp$ ASAN_OPTIONS=handle_segv=2 make run
WARNING: STARTUP_CONF not defined or file doesn't exist.
Running with minimal startup config: unix { interac
Hi Chetan,
> You mean we don't support centos anymore and we only support ubuntu.
I mean it is not run in CI anymore so things will break w/o us realizing, but
the packaging scripts etc still exists.
So it is mostly a matter of someone using VPP on CentOS volunteering to
maintain it and proposi
> Hello vpp-dev, Is there any plan to enable fuzzing for VPP ? e. g. to
> use https://llvm.org/docs/LibFuzzer.html in VPP.
We experimented profile-guided fuzzing of VPP with AFL++ in the past and it was
promising, but we never had the time to "productize" it.
I'd be interested!
Best
ben
-=-=-
We do not run CentOS/opensource RHEL-equivalent anymore and I do not think
anybody is looking at it.
Chances are it is quietly bit-rotting...
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of chetan bhasin
> Sent: Wednesday, June 22, 2022 8:00
> To: vpp-dev
> Subject: [
Are you sure you're downloading the right packages for your distro?
Check the content of /etc/apt/sources.list.d/fdio_release.list and
/etc/os-release.
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Lokesh Chakka
> Sent: Tuesday, June 21, 2022 9:36
> To: vpp-dev@lists
Looks like the crash is happening in the host stack...
Can you try to reproduce it in debug mode, and maybe even with Address
Sanitizer:
https://s3-docs.fd.io/vpp/22.06/gettingstarted/troubleshooting/sanitizer.html#id2
best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf
Looks like a simple and worthwhile optimization, merged.
Thanks!
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Velichkin
> Viktor
> Sent: Tuesday, May 31, 2022 22:47
> To: vpp-dev@lists.fd.io
> Subject: [vpp-dev] papi performance improvement patch
>
> Hi!
>
> Can an
This is weird - when you step-by-step in gdb and it works, is msg_size still
110190? Ie in the failing case, do we fail because we decode a bad msg_size or
because we are missing some data?
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Kevenvy Smith
> Sent: Thursday
That's not how a layer 3 network works: the dest mac is the mac address of the
L3 nexthop.
If you want to use the dest mac of the switch as nexthop, it should act as a
router (with an IP address you can use as nexthop).
Best
Ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behal
It all depends on what you want to do. Usually, we try to avoid sharing buffers
read/write between multiple process, it makes debugging much harder -
especially buffer leaks or use-after-free...
Here is an example to share VPP buffers read-only with snort so that snort can
inspect traffic and gi
I think you should run step-by-step with gdb to see exactly what is failing. As
it works on a recent system but not on an old one, I suspect an issue with an
unsupported feature (eg. special flag in syscall or similar)...
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Beh
Hmm good catch, I wonder why we did not catch it - maybe the unit tests use
only non-standard port...
Can you try to patch it accordingly and if it solves the issue, push it on
gerrit for review?
If you can update the unit tests to reproduce the issue and verify the fix
fixes it it would be idea
Hi Sindhu,
> When we try to build with : /build-root/vagrant/build.sh, we see below
> issues:
I'd not use this script; it is not really maintained anymore. Maybe we should
remove it entirely.
Anyway, follow these steps instead and see if it works better
https://fd.io/gettingstarted/building/
Hi Petr, thanks for sharing, this looks great!
Out of curiosity, how would you compare it to Pim vppcfg
https://github.com/pimvanpelt/vppcfg ?
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Petr Boltík
> Sent: Monday, May 2, 2022 11:12
> To: vpp-dev
> Subject: [
Hi Petr,
Thanks for sharing the solution for future readers.
Regarding doc improvements, we'd love more contributions 😊
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of Petr Boltík
> Sent: Saturday, April 30, 2022 17:54
> To: Petr Boltík
> Cc: vpp-dev
> Subject:
It is an error returned by the DPDK driver, you'll need to check the DPDK
driver code to be sure.
As an alternative, did you try to use VPP native AVF driver?
https://s3-docs.fd.io/vpp/22.06/developer/devicedrivers/avf.html
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On B
I am all for a consistent formatter for make test, the current situation is
painfull .
Regarding defaults, I think just using the defaults from the tools is fine as
long it is not completely crazy.
I'll just use make fixstyle before pushing anyway.
ben
> -Original Message-
> From: vpp-d
Hi Matt, John,
This is a great summary!
>> The only reference I could find for VPP in Azure was in this link
>> but it is very outdated:
>> https://fd.io/docs/vpp/v2101/usecases/vppinazure.html
> Yes, that doc is way outdated. I advise against following the instructions
> in it.
Matt or John, w
Hi Sonia,
Could you try to apply this diff and see if that fixes your problem:
diff --git a/src/plugins/ip_session_redirect/redirect.c
b/src/plugins/ip_session_redirect/redirect.c
index 5c0123571..bb832c8a0 100644
--- a/src/plugins/ip_session_redirect/redirect.c
+++ b/src/plugins/ip_session_redi
Hmm sounds like the virtio driver does not return an error to userspace when
failing to create txq?
With af_xdp we try to create 1 txq per VPP worker, up to the number available
txq advertised by the kernel for the interface. If the txq creation failed, we
just use the txq we got. Unless there i
as well. Thanks!
>
>
>
> On 23 Mar 2022, at 20:03, Catalin Iordache
><mailto:catalinn.iorda...@gmail.com> <mailto:catalinn.iorda...@gmail.com>
> > wrote:
>
> Hi Benoit, Dave,
>
> Thanks for gett
tremely useful for me if someone would add a
> multi-threaded example and documentation in that merge request as well.
>
> Thanks for helping me so far.
>
> Best regards,
> Catalin
>
>
>
>
> Let me know if a similar problem was reported in vnet/ipsec.
I am not aware of anything specific to that. This really looks like your ike
implementation failed to program the right SA from time to time: once it
failed, it remains wrong until the next re-negotiation fixes it.
Did you check the
Hi,
> * I'm working on a plugin development,does anyone know how to share a
> global variable between two plugins?Thank you very much.
You can use vlib_get_plugin_symbol() to load an exported symbol from another
plugin.
Best
ben
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent
Hi,
Thanks for the patch but please note that the dpdk ipsec backend is deprecated
since 21.06 in favor of the vpp async crypto infra (which does support dpdk
cryptodev).
You might want to update your VPP instead.
Best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of
Is your system AVX-512 enabled? We just fixed 2 bugs in AES CBC with VAES:
https://gerrit.fd.io/r/c/vpp/+/35746
https://gerrit.fd.io/r/c/vpp/+/35767
best
ben
> -Original Message-
> From: vpp-dev@lists.fd.io On Behalf Of
> hustu...@gmail.com
> Sent: Tuesday, March 29, 2022 21:24
> To: vpp
Yes, you can call 'gdb_show_traces()' from gdb - but you'll need to have
tracing enable beforehand of course.
If that's not enough, you can enable trajectory tracing:
https://git.fd.io/vpp/tree/src/vlib/buffer.h#n101 and inspect them via
'gdb_dump_trajectory_trace()'.
ben
> -Original Messa
Hi,
> I found that while running show event-logger all, resident memory of VPP
> is suddenly jumped, still, main-heap is not growing, Does anyone has an
> idea about this behavior? Thanks in advance.
You can try to track allocations with memory traces:
https://s3-docs.fd.io/vpp/22.06/gettingstar
load the doc
> verify results to an AWS S3 bucket with a 7 day retention policy to
> enhance the documentation review process.
>
> Thanks,
> -daw-
>
>
> On 3/22/22 5:21 AM, Benoit Ganne (bganne) via lists.fd.io wrote:
>
>
> Hi Catalin,
>
>
Hi Catalin,
CC'ed vpp-dev, but examples and doc can be found here
https://gerrit.fd.io/r/c/vpp/+/30573
It would be good to have this patch merged I guess but it looks like there are
some issues to fix in the doc.
Best
Ben
> From: Catalin Iordache
> Date: Mar 20, 2022 15:36
> Subject: In regar
Hi Sudhir,
Yes Asan is pretty picky about compiler version (probably issue 1), how you
link your plugins (issue 3 is ASan complaining about your private plugin
librtbvpp.so redefining an already defined global symbol, issue 2 might be
linked to that too).
Anyway, using GCC, LD_PRELOAD and ASAN_
Ok, then I guess you could give a shot at adding checksum offload support in
the AF_XDP plugin.
I think that should roughly mean:
- add a flag at interface creation such as "csum-enabled"
- when this flag is set, in the rx function add something akin to
https://git.fd.io/vpp/tree/src/vnet/devic
Hi Carl,
Indeed, there is no way in AF_XDP to detect whether the checksum is offloaded
or not.
We could always add a flag but I am not sure it would worth it.
Can you try host-interface (AF_PACKET) and see if it solves your issue:
https://s3-docs.fd.io/vpp/22.06/cli-reference/clis/clicmd_src_vne
> Could you please suggest How to debug if a vlib buffer leak is happening
> inside some plugin.
> Is there a way to print the content of the allocated vlib buffer ?
Ha sorry I misunderstood. You can try
https://git.fd.io/vpp/tree/src/plugins/bufmon/bufmon_doc.rst
There are a few issues I still n
> You decrement/increment n_left_to_next/to_next twice in one iteration of
> the loop, so you are writing out of bounds.
>> if (is_send_syn && (1 < n_left_to_next))
I think this should be taken care of by this test though.
>> vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next,
>> n_
1 - 100 of 478 matches
Mail list logo
