Hello Stanislav,
Sorry for late answer, I was temporary deprived of my testbed :-)
So, applied patches:
1. Last patch attached to this thread
2. Patch from Gerrit from above.
Looks everything working now. OSI/IS-IS packets are passed to CP, routing
protocol working.
Issue with VPP crash after .
Hello Gennady,
Could you try again with this patch[0]?
[0] - https://gerrit.fd.io/r/c/vpp/+/38118
On Wed, 1 Feb 2023 at 19:38, Gennady Abramov wrote:
> Hello Stanislav,
>
> Here is it!
>
> Thread 1 "vpp_main" received signal SIGABRT, Aborted.
> __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/s
Hello Stanislav,
Here is it!
Thread 1 "vpp_main" received signal SIGABRT, Aborted.
__GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:50
50 ../sysdeps/unix/sysv/linux/raise.c: No such file or directory.
(gdb)
(gdb) f 9
#9 0x7fffaf404b6a in lcp_router_link_add (rl=0x5a38e
Hi Gennady,
Could you execute the following commands in gdb and show the output?
f 9
info locals
p lipi
p *lip
On Tue, 31 Jan 2023 at 11:16, Gennady Abramov wrote:
> Hello Stanislav,
>
> The is-is itself is working, thank you!
> tn3# show isis neighbor
> Area myisis:
> System Id Int
Hello Stanislav,
The is-is itself is working, thank you!
tn3# show isis neighbor
Area myisis:
System Id Interface L State Holdtime SNPA
tn1 Ten0.1914 3 Up 28 2020.2020.2020
Unfortunately, both lcp lcp-auto-subint and lcp lcp-sync still looks
I really hoped that there's a better and shorter solution, but I couldn't
find it.
You can try with this ugly patch with a (lot of) copy-paste from
lip_punt_node :)
On Thu, 26 Jan 2023 at 14:14, wrote:
> Hi Stanislav,
>
> Situation is better now, but still only half of the problem solved :-)
> O
Hi Stanislav,
Situation is better now, but still only half of the problem solved :-)
OSI IS-IS packets passed from network to tap, but not passed from tap to
network.
These are on TAP interface, where:78 is VPP-based router, :7a is non-VPP peer,
both directions are seen:
13:03:22.911887 3c:ec:ef
Could you check again with the following patch? The previous didn't work
because the linker couldn't find a symbol with the node (apparently, linux
cp plugin divided into 2 parts: a library and a plugin).
I replayed a pcap containing some OSI frames and got this trace:
00:00:18:025096: virtio-inp
Hi Stanislav!
Here is it!
00:31:48:504910: dpdk-input
TenGigabitEthernet1c/0/1 rx queue 0
buffer 0x9ad69: current data 0, length 1518, buffer-pool 0, ref-count 1, trace
handle 0xb
ext-hdr-valid
PKT MBUF: port 0, nb_segs 1, pkt_len 1518
buf_len 2176, data_len 1518, ol_flags 0x180, data_off 128, p
That's not surprising, could you also show me a trace? (trace add
dpdk-input 10 and then show trace with ISIS packet)
On Tue, 24 Jan 2023 at 16:25, wrote:
> Hi Stanislav,
>
> Unfortunately, your patch didn't help. VPP builds, but IS-IS packets still
> cannot be passed between the CP and the wire
Hi Stanislav,
Unfortunately, your patch didn't help. VPP builds, but IS-IS packets still
cannot be passed between the CP and the wire.
Furthermore, it looks like LCP lcp-auto-subint feature was broken:
root@tn3:/home/abramov/vpp# vppctl
___ _ _ _ ___
__/ __/ _ \ (_)__ |
Hello folks,
In old router plugin it was done with the following snippet:
```
#include
...
osi_register_input_protocol (OSI_PROTOCOL_isis, im->tx_node_index);
```
So, I'd give it a try to register this protocol for `lip_punt_node`
(`lcp_router_init` seems quite suitable). Something like this:
di
Hoi,
I would suggest not matching (only) MAC but (foremost) the ethertype, and
then punting those packets into the TAP, take a look at VLIB_REGISTER_NODE
(lip_punt_node) in src/plugins/linux-cp/lcp_node.c, contributions are
welcome.
groet,
Pim
On Mon, Jan 23, 2023 at 6:06 PM wrote:
> Hoi Pim,
Hoi Pim,
As for distinguishing IS-IS packets, I think that should not be really
difficult, it's just all the packets with specific DST MACs:
09:00:2b:00:00:05, 09:00:2b:00:00:14,09:00:2b:00:00:15.
It's hard to imagine situation when they are needed to be processed by
DataPlane.
-=-=-=-=-=-=-
Hoi,
Linux CP supports ARP, IPv4 and IPv6. ISIS uses its its own ethertype, as
do other protocols (like LLDP for example). Those will not be punted into
the TAP by the plugin (and it's difficult to uniquely identify the ethernet
frames that should be punted as compared to being handled entirely i
15 matches
Mail list logo
