ERROR: Table '_____' is invoked multiple times.

[Raymondma-public]

I face to a problem that when I am going to do the same thing in different switch role I defined.
I have read the following and know that applying a table multiple times is normally not allowed.
http://lists.p4.org/pipermail/p4-dev_lists.p4.org/2016-June/000388.html

But in my case I will only invoke a table for once for different switch role.
Would there be any solution for my case to reduce duplication of P4 code and switch commands?

P4 Code:

    if(user_metadata.switch_role==0){ //ECMP
	if(valid(ipv4) and ipv4.ttl > 0) {

	    apply(ecmp_group);
	    apply(forward);

	}
    }else if(user_metadata.switch_role==1){//Expeditus-ToR
    	
        //the counter show P4 recgonized SYN
        if(tcp.ctrl==TCP_SYN){ 
    	    apply(increment_SYN_table);
            apply(expeditus_header_table);
            apply(ecmp_group);
            apply(forward);
    	}

    }

Error:
ERROR: Table 'ecmp_group' is invoked multiple times.

[dhankook]

You have at least three options: 1) [BEST] Restructure your code. Looks like "ecmp_group" and "forward" are both applied in both cases so just move them so that they are invoked after the conditional blocks. 2) [OK] Make functionally equivalent copies of ecmp_group and forward, with new names. This is not so great because you'll have more table entries to manage but performance-wise this should work fine. 3) [NOT RECOMMENDED] Make use of the 'resubmit' primitive to sidestep the 'invoke each table only once' limitation. Throughput will take a hit, with the performance penalty depending on how often the resubmit is triggered.

…

--David

On 11/25/2016 08:54 PM, Raymondma-public wrote: I face to a problem that when I am going to do the same thing in different switch role I defined. I have read the following and know that applying a table multiple times is normally not allowed. http://lists.p4.org/pipermail/p4-dev_lists.p4.org/2016-June/000388.html But in my case I will only invoke a table for once for different switch role. Would there be any solution for my case to reduce duplication of P4 code and switch commands? P4 Code: |if(user_metadata.switch_role==0){ //ECMP if(valid(ipv4) and ipv4.ttl > 0) { apply(ecmp_group); apply(forward); } }else if(user_metadata.switch_role==1){//Expeditus-ToR //the counter show P4 recgonized SYN if(tcp.ctrl==TCP_SYN){ apply(increment_SYN_table); apply(expeditus_header_table); apply(ecmp_group); apply(forward); } } | Error: |ERROR: Table 'ecmp_group' is invoked multiple times.| — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#231>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AO6YA4K0PliVXg06M7n5y7d-i7fWYn5Jks5rB61tgaJpZM4K8zLg>.

[Raymondma-public]

For the first option.
Indeed I will have some more roles that do not contain those "ecmp_group" and "forward". So does it mean I should use the second option that copy the tables and commands to new one with different name?

[dhankook]

Maybe. But perhaps you can write sufficiently complex boolean logic that identifies all of the roles that *do* contain ecmp_group and forward so that you only apply those tables for those roles. In other words, you have not convinced me that your code can't be restructured so that each table is only invoked once in your program. But you know your project best, so if restructuring isn't possible, go for the second option.

…

--David

On 11/28/2016 06:00 AM, Raymondma-public wrote: For the first option. Indeed I will have some more roles that do not contain those "ecmp_group" and "forward". So does it mean I should use the second option that copy the tables and commands to new one with different name? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#231 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AO6YAzDoJtw9LUXmT_gRbGLJpratOxoLks5rCtBsgaJpZM4K8zLg>.

[antoninbas]

Just adding some more info to @dhankook answer. In theory, the compiler and software switch should allow you to invoke the same table from 2 different places in the control flow, providing the 2 invocations are mutually exclusive (i.e. the same packet cannot go through the same table twice). However, because of the way we process the control flow in the current P4 compiler (we transform it into a table graph), it is actually pretty difficult to allow this, so we just throw an error if the same table is "applied" twice. While there is no immediate plan to fix this in the current compiler, we are hoping the next compiler (for P4_16) will not have this limitation.

Regarding your specific use case, I'd like to point out that most people will probably achieve different roles by writing different P4 programs and swapping the programs on the target as needed. This can be done in bmv2 through the runtime CLI. When you include all your roles in a single P4 program, you may limit the amount of physical resources available for each role. While in bmv2 your programs can be arbitrarily complex, this won't be the case for hardware P4 targets.