Skip to Content.

rare-users - Re: [RARE-users] 1x400G to 4x100G breakout on Tofino2

Subject: RARE user and assistance email list

List archive


Re: [RARE-users] 1x400G to 4x100G breakout on Tofino2


Chronological Thread 
  • From: Alexander Gall <>
  • To: Marcos Felipe Schwarz <>
  • Cc: <>, "" <>
  • Subject: Re: [RARE-users] 1x400G to 4x100G breakout on Tofino2
  • Date: Wed, 15 Mar 2023 08:27:08 +0100


Hi Marcos

Just a quick follow-up. Can you please show me the output of
"module-show" for your plugins in bf_pltfm.qsfp context of bfshell?
At the bottom of the output it will say something similar to

-------------- Host interface --------------
Standard supported Lns
XLPPI 4 X 10.3125Gbd NRZ 4
100GBASE-CR4 4 x 25.78125Gbd NRZ 4
25GBASE-CR CA-L 1 x 25.78125 Gbd NRZ 1

This will show which of the speed settings (on the electrical/host
side) can be applied to that particular plugin.

And after a bit more thought, please ignore my comment about something
missing in the SDE. The $PORT table also has a field "$N_LANES" which
conveys the number of lanes requested for a particular speed setting,
so that's ok. Then we will only have to come up with a way to
configure this in freertr and pass it on to bf-forwarder.

--
Alex


On Wed, 15 Mar 2023 07:44:25 +0100, Alexander Gall <> said:

> Hi Marcos
> Yes, I'm afraid this is not yet supported. The issue is that with
> Tofino2 there are now multiple configurations for a 100G port. With
> Tofino1, it could only mean 4 lanes with 25G, but now we can have 100G
> on 1 or two lanes as well. If you look at the command help in bfshell
> pm context, you can see the additional speed designators

> 100G(100G/100G-R4, 100G-R2, 100G-R1)

> 100G-R1 and 100G-R2 are the 1 and two lane variants. The problem is
> that freertr currently only accepts an integer as the speed and only
> "100" is accepted for a 100G port, so we can't pass the additional
> information on to the bf-forwarder. So we'll need that and the
> additional logic to translate this into the correct setting for the
> port configuration.

> However, there is an additional problem in the SDE :( It seems to me
> that these new speed settings are not yet supported by the BFRT
> either. Looking at the "string_choises" for the $PORT pseudo table in
> bfrt_python via bfshell:

> 'BF_SPEED_1G',
> 'BF_SPEED_10G',
> 'BF_SPEED_25G',
> 'BF_SPEED_40G',
> 'BF_SPEED_40G_NB',
> 'BF_SPEED_40G_NON_BREAKABLE',
> 'BF_SPEED_50G',
> 'BF_SPEED_100G',
> 'BF_SPEED_200G',
> 'BF_SPEED_400G'],

> It seems like we cannot implement the equivalent of "port-add <foo>
> 100G-R1" at the moment. I'll have to look into that and maybe ask in
> the Intel forum.

> --
> Alex

> On Tue, 14 Mar 2023 17:47:42 +0000, Marcos Felipe Schwarz
> <> said:

>> Hi all,
>> We are trying to configure a 1x400G to 4x100G breakout on Tofino2, but it
>> seems to not be supported.

>> Follows my configuration:

>> +------------------------------------------------------------------------------------+
>> |server p4lang p4
>> |
>> |
>> |
>> |export-port sdn13000 288 100 3 1 0
>> |
>> |
>> |
>> |export-port sdn13004 292 100 3 1 0
>> |
>> |
>> |
>> |export-port sdn12000 296 100 3 1 0
>> |
>> |
>> |
>> |export-port sdn12004 300 100 3 1 0
>> |
>> |
>> |
>> |export-port sdn11000 304 100 3 1 0
>> |
>> |
>> |
>> |export-port sdn11002 306 100 3 1 0
>> |
>> |
>> |
>> |export-port sdn11004 308 100 3 1 0
>> |
>> |
>> |
>> |export-port sdn11006 310 100 3 1 0
>> |
>> |
>> |
>> |interconnect ethernet0
>> |
>> |
>> |
>> |vrf p4
>> |
>> |
>> |
>> |exit
>> |
>> +------------------------------------------------------------------------------------+

>> sdn12XXX and sdn13XXX are 1x400G to 2x200G breakout, and are working
>> correctly. But for sdn11XXX it didn’t work, at bfshell, we can verify that
>> it didn’t configured the lanes correctly:

>> +------------------------------------------------------------------------------------+
>> |11/0 |22/0|304|2/48|100G | RS |Ds|Au|NO |ENB|DWN| NONE |
>> 0| |
>> | 0|
>> |
>> |
>> |
>> |11/1
>> |22/1|305|2/49|-------|----|--|--|---|---|---|--------|----------------|
>> |
>> |----------------|-
>> |
>> |
>> |
>> |11/2
>> |22/2|306|2/50|-------|----|--|--|---|---|---|--------|----------------|
>> |
>> |----------------|-
>> |
>> |
>> |
>> |11/3
>> |22/3|307|2/51|-------|----|--|--|---|---|---|--------|----------------|
>> |
>> |----------------|-
>> |
>> |
>> |
>> |11/4 |22/4|308|2/52|100G | RS |Ds|Au|NO |ENB|DWN| NONE |
>> 0| |
>> | 0|
>> |
>> |
>> |
>> |11/5
>> |22/5|309|2/53|-------|----|--|--|---|---|---|--------|----------------|
>> |
>> |----------------|-
>> |
>> |
>> |
>> |11/6
>> |22/6|310|2/54|-------|----|--|--|---|---|---|--------|----------------|
>> |
>> |----------------|-
>> |
>> |
>> |
>> |11/7
>> |22/7|311|2/55|-------|----|--|--|---|---|---|--------|----------------|
>> |
>> |----------------|-
>> |
>> +------------------------------------------------------------------------------------+

>> Regards,

>> Marcos Schwarz | Gerente de P&D | R&D Manager

>> Gerência de P&D em Ciberinfraestrutura (GCI) | Cyberinfrastructure R&D
>> Management

>> Tel.: +55 (19) 98303-3010 | Teams:

>> www.rnp.br

>> cid:fefa97ac-2f79-4bc6-842e-23dbdb886128

> --
> Alexander Gall, Network

> SWITCH
> Werdstrasse 2, P.O. Box, 8021 Zurich, Switzerland
> phone +41 44 268 15 15, direct +41 44 268 15 22
> https://switch.ch https://swit.ch/linkedin https://swit.ch/twitter



Archive powered by MHonArc 2.6.24.

Top of Page