Override in UVM

Use the below eda link as a reference :
https://www.edaplayground.com/x/5fXJ


Why to override ??????
For example: In your environment, you have a driver component. You would like the extend the driver component for error injection scenario. After defining the extended driver class with error injection, how will you replace the base driver component which is deep in the hierarchy of your environment ? Using hierarchical path, you could replace the driver object with the extended driver. This could not be easy if there are many driver objects. Then you should also take care of its connections with the other components of testbenchs like scoreboard etc. 

Using the uvm fatroy, it is very easy to solve the above requirements. Only class extended from uvm_object and uvm_component are supported for this.

There are three basic steps to be followed for using uvm factory.
1) Registration   
  • While defining a class , its type has to be registered with the uvm factory. To do this job easier, uvm has predefined macros.
    `uvm_component_utils(class_type_name)
2) Construction 
  • To construct a uvm based component or uvm based objects, static method create() should be used. This function constructs the appropriate object based on the overrides and constructs the object and returns it. So while constructing the uvm based components or uvm based objects , do not use new() constructor.
3) Overriding 
  • If required, user could override the registered classes or objects. User can override based of name string or class-type.

    There are 4 methods defined for overriding: 

The factory makes it is possible to override the type of uvm component /object or instance of a uvm component/object in 2 ways. 
  • uvm component/object type 
  • uvm compoenent/object  name.  
inst_override_by_type :

Syntax : 
a)   a1 = A::type_id::create("a1", this); (In environment)
      factory.set_inst_override_by_type(A::get_type(), A_ovr::get_type(),  {get_full_name, ".","env.a1"}); (In test)
or 
b)  a1 = A::type_id::create("a1", , "path1"); (In environment)
      factory.set_inst_override_by_type(A::get_type(), A_ovr::get_type(),  "path1.a1");  (In test)


inst_override_by_name :

Syntax : 
a)   a1 = A::type_id::create("a1", this); (In environment)
      factory.set_inst_override_by_type(A, A_ovr,  {get_full_name, ".","env.a1"}); (In test)
or 
b)  a1 = A::type_id::create("a1", , "path1"); (In environment)
      factory.set_inst_override_by_type(A, A_ovr,  "path1.a1");  (In test)


Description : 
  • Here we just need to provide three arguments first whom to override , by which to override and path of particular instance that you want to override. 
  • If you have create particular component using this method then you should use first method for overriding and if you have created particular comp using path method then you should use second method for overriding. 



type_override_by_type :

Syntax : 

a)    factory.set_type_override_by_type(A::get_type(), A_ovr::get_type());  (In test)
                                                 or
b)   A::type_id::set_type_override(A_ovr::get_type() , 0);  (In test)
type_override_by_name :

Syntax : 

     factory.set_type_override_by_name("A", "A_ovr"); (In test)
Description : 
  • In type_override_by_type there are two methods
    • a) Here you do not require to pass exact path as we were giving in inst_override
    • b) Same but with this syntax you can control override operation in hierarchy. If the last argument is 0 then once its override in up hierarchy then down hierarchy comp will not be able to override it.So you can override it just once. While if its value is 1 then you can override it in down hierarchy.
  • In type_override_by_name one can simply override the component just by using its name. 



override_through_command_line :
By using below run time option one can use over ride through command line.

+uvm_set_inst_override=A,A_override,uvm_test_top.env.a1
+uvm_set_inst_override=A,A_ovr,path1.a1  
+uvm_set_type_override=A,A_ovr  
+uvm_set_type_override=A,A_override





Reference : http://www.testbench.in/UT_06_UVM_FACTORY.html

Comments

Post a Comment

Popular posts from this blog

'WITH' Operator in SV Coverage

Hello guys, Today we are going to see how we can use 'with' operator which is there in sv coverage. Many of us are writing coverage for our projects. I think most of us are under using the use of 'with' operator  in sv coverage. So lets explore one rarely explored feature which is already available for us in sv coverage. SV LRM DESCRIPTION : The with clause specifies that only those values in the covergroup_range_list that satisfy the given expression (i.e., for which the expression evaluates to true, as described in 12.4) are included in the bin. In the expression, the name item shall be used to represent the candidate value. The candidate value is of the same type as the coverpoint. The name of the coverpoint itself may be used in place of the covergroup_range_list to denote all values of the coverpoint. Only the name of the coverpoint containing the bin being defined shall be allowed; no other coverpoint names shall be permitted. Consider the following exa...
Read more

Difference Between m_sequencer and p_sequencer

Image
Hello guys, Today we will see what is the difference between m_sequencer and p_sequencer, may be this is the most frequently asked question in interview from uvm section and may be this is the most confusing topic too. Let's today solve this puzzle forever. Here I am assuming that you are aware about uvm components and uvm object. Basically sequencer is used to route the random transactions to driver which is generated by sequence class. If you are not specifying any thing by default uvm is using m_sequencer.      WHY WE NEED P_SEQUENCER ?????? Say you want to design your sequence class such that based on RTL response it should generate further transactions. In this scenario sequence needs to access  the user defined properties or methods of the sequencer as per requirement of the project which are unavailable in m_sequencer. So in this scenario you need to use p_sequencer. Basically it overrides the actual sequencer and now you can use p_sequencer...
Read more

Cross Auto Bin Max

Hello friend, In this blog we are going to discuss about the cross auto bin max. ex. bit [3:0] a; bit [3:0] b; A_C : coverpoint a_c {    bin1 = bins[7:0];    bin2 = bins[7:15];  } B_C : coverpoint b_c {    bin1 = bins[7:0];    bin2 = bins[7:15];  } i)  A_B_CROSS : cross a_c , b_c; ii) A_B_CROSS : cross a , b; In first case we are doing cross coverage of to coverpoints where in second case two variables. In first case four cross bins will be generated as each cover point is having two two bins where in second case (2^4 * 2^4) = 256 bins will be generated. This happens because for simple coverage auto_bin_max value is 64 where cross_auto_bin_max is infinite for cross coverage. So say if you have declared two vars as int and then you are directly performing their cross coverage then (2^32 * 2^32) bins will be generated because cross_auto_bin_max value is infinite. If you are just performing simple covera...
Read more