RARE user and assistance email list

[mc36 <>]

no need to do any testing on this particular question, our cpuheader is open 
source:
https://bitbucket.software.geant.org/projects/RARE/repos/rare/browse/p4src/include/hdr_cpu.p4

the sonic builds use the switch.p4 from the bf-sde which i'm attaching you 
here for you convenience:

mc36@noti:~/Downloads/bf-sde-9.9.0/packages/switch-p4-16-9.9.0/submodules/SAI/bm/p4-sai/p4src$
 cat headers.p4 | tail -n8
header_type cpu_header_t {
    fields {
        bit<16> trap_id;
        bit<16> bridge_id;
        bit<8>  ingress_port;
        bit<8>  bridge_port;
    }
}
mc36@noti:~/Downloads/bf-sde-9.9.0/packages/switch-p4-16-9.9.0/submodules/SAI/bm/p4-sai/p4src$

as you can see, our portid overlaps with their trapid so it wont fly...

br,
cs


On 5/3/22 08:27,  wrote:
> Thanks mc36 for the insight. Actually this is what Marcos wanted to test.
>
> @Marcos, the quickest way to confirm your aim would be to directly post a 
> question to ICRP.
> I did not test it personally, INTEL released also a "folded pipe    mode that 
> might related to your objective.
>
> > Le 3 mai 2022    06:36, mc36 <> a   crit :
> >
> > hi,
> >
> > On 5/2/22 19:58, Fr  d  ric LOUI wrote:
> > >      Run SONIC and RARE/freeRtr AT_THE_SAME_TIME
> >
> > i have concerns on this particular topic:
> > we already had a small chat with alex (the author of an other p4 program 
> > packaged into the rare builds called packet-broker)
> > and finally we found out that, that regardless freerouter already have the 
> > (test covered) capability to multiplex up/downstream
> > some front-panel ports, it most probably will interwork only with other rare 
> > profiles of ours, because this capability depends
> > both in the control plane and in the dataplane code on that the cpu-port 
> > format is common between the pipes, so a single
> > freerouter can understand (and multiplex up and downstream) the packets from 
> > both the tofino pipes.
> >
> > moreover, some features like the hairpin interfaces or the tunnel 
> > decapsulation depends on recirculation, which also uses a
> > rare specific header to pass all the extra information between the packet 
> > processing rounds, and, recirculation happens
> > between the pipes, that is, this particular case will most probably simply 
> > won't work at all because the pipes will diverge
> > in the forwarding info, and in case of different p4 programs, even the extra 
> > header will not match.
> >
> > br,
> > cs
>
>
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Groups.io Links: You receive all messages sent to this group.
> View/Reply Online (#282): https://groups.io/g/freertr/message/282
> Mute This Topic: https://groups.io/mt/90840395/6006518
> Group Owner: 
> Unsubscribe: https://groups.io/g/freertr/unsub []
> -=-=-=-=-=-=-=-=-=-=-=-