Rare project developers

[Alexander Gall <>]

On Thu, 2 Jun 2022 12:07:52 +0200, mc36 <> said:

> hi,
> taking a closer look on what's going on, i just realized that this pre_emit 
> stuff is basically just before the very core tocpu check:
> https://github.com/rare-freertr/freeRtr/blob/7a417d4537db9cbae8fdcb9f618ef008f62f3913/misc/p4bf/include/ig_ctl.p4#L187
> so for now, i strongly believe that we can get rid of all the ternary 
> matches here if we split the table... the cpuhdr
> insertion could go directly to the inner part of the if, and the rest could 
> go to the else part... for the bridging ternary,
> i can easily list all the possible values for the sap_en/de_cap so this 
> table could easily become an all-exact.... let's see.... :)

Ok, great. In the meantime I found that 9.7.2 fixes the compiler bug
in 9.7.1 that shows up with the BNG profile. I'll move RARE to 9.7.2
and tigether with your changes we should have a complete RARE on
Tofino2 :)

BTW, I think we now need an additional optimizer run for Tofino2.

-- 
Alex

> br,
> cs


> On 6/2/22 11:02, mc36 wrote:
>> hi,
>> thanks for playing with it and finding a workaround...
>> that field is not exactly ternary, it was just written this way to shrink 
>> that table...
>> so the values are very well defined and could be listed explicitly... imho 
>> i'll do that,
>> i'll validate on the test cases and then we can replace ternary to exact 
>> for that field...
>> br,
>> cs
>> 
>> 
>> 
>> On 6/2/22 10:27, Alexander Gall wrote:
>>> 
>>> Interesting. The problem goes away when I add a redundant entry with a
>>> fixed value of mpls_encap_decap_sap_type in the global part of
>>> include/ig_ctl_pkt_pre_emit.p4:
>>> 
>>> diff --git a/p4src/include/ig_ctl_pkt_pre_emit.p4
>>> b/p4src/include/ig_ctl_pkt_pre_emit.p4
>>> index c53597f..0575f34 100644
>>> --- a/p4src/include/ig_ctl_pkt_pre_emit.p4
>>> +++ b/p4src/include/ig_ctl_pkt_pre_emit.p4
>>> @@ -309,6 +309,7 @@ ig_md.nexthop_id:
>>> (1, 0, 0, 0, 0, 0, 0, 0, 4, _):act_pkt_srv_l3vpn_ipv4_encap();
>>> (1, 0, 0, 0, 0, 0, 0, 0, 6, _):act_pkt_srv_l3vpn_ipv6_encap();
>>> #endif
>>> +                       (0, 0, 0, 0, 0, 0, 0, 0, 4, 
>>> CPU_PORT):act_pkt_send_to_cpu();
>>> (0, 0, 0, 0, 0, 0, 0, 0, _, CPU_PORT):act_pkt_send_to_cpu();
>>> (0, 0, 0, 0, 0, 0, 0, 1, _, CPU_PORT):act_pkt_send_to_cpu();
>>> (0, 0, 0, 0, 0, 0, 1, 0, _, CPU_PORT):act_pkt_send_to_cpu();
>>> 
>>> That seems like an easy workaround :)
>>>