Просмотр исходного текста страницы UVM/UVM Tutorial for Candy Lovers/Register Access Methods

{{UVM TOC}}

The register abstraction layer (RAL) of UVM provides several methods to access registers. This post will explain how the register-access methods work. In Register Abstraction, we introduced the overview of RAL and explained how to define registers. In this post, we will cover how to access the registers.

Регистр уровень абстракции (RAL) из УВМ предоставляет несколько методов для доступа к регистрам. Это сообщение будет объяснить, как методы регистров доступа работать. Регистрация Абстракция, мы ввели обзор RAL и объяснил, как определить регистры. В этой статье мы рассмотрим, как получить доступ к регистрам.

== Properties of uvm_reg_field ==

<!-- 
Before diving into the register-access methods, let’s look at how a register value is stored. As seen in [http://cluelogic.com/2012/10/uvm-tutorial-for-candy-lovers-register-abstraction/ Register Abstraction], <code>uvm_reg_field</code> is the lowest register-abstraction layer which represents the bits of a register. The <code>uvm_reg_field</code> uses several properties to store a variety of register-field values:
-->
Перед рассмотрением методов доступа к регистрам (регистровой модели), необходимо рассмотреть как значение регистра хранится. Как рассмотрено в статье [http://cluelogic.com/2012/10/uvm-tutorial-for-candy-lovers-register-abstraction/ Register Abstraction], Класс поля <code>uvm_reg_field</code> является наименьшим (в иерархии) слоем в регистровой модели, который представляет биты регистра. Класс <code>uvm_reg_field</code> использует некоторые свойства для сохранения множества значений поля регистра:

<!--
* <code>m_reset["HARD"]</code> stores a hard reset value. Note that the <code>m_reset</code> is an associative array with a kind of reset as the key.
* <code>m_mirrored</code> stores the value of what we ''think'' in our design under test (DUT).
* <code>m_desired</code> stores the value of what we ''want'' to set to the DUT.
* <code>value</code> stores the value to be sampled in a functional coverage, or the value to be constrained when the field is randomized.
-->
* <code>m_reset["HARD"]</code> хранит значение HARD (системного) сброса. Следует заметить, что <code>m_reset</code> является ассоциативным массивом с типом сброса в качестве ключа.
* <code>m_mirrored</code> хранит значение того, что мы «думаем» в нашем DUT`е.
* <code>m_desired</code> хранит значение того, что мы «хотим» установить в DUT`е.
* <code>value</code> хранит значение для учета его в функциональном покрытии, или значение законстрейненное (to be constrained) когда поле рандомизируется.

<!--
Note that among these properties, only the <code>value</code> property is public. The other properties are local, thus we cannot access them directly from the out side of the class. We will show you how to access these local properties using register-access methods later.
-->
Заметим что среди этих свойств только свойство <code>value</code> является public. Остальные свойства являются локальными, The other properties are local, таким образом мы не можем к ним обратиться напрямую вне данного класса. Далее будет рассмотрено как обратиться к этим локальным свойствам используя методы доступа к регистровой модели.


{| align=center
!<html><img alt="Properties of uvm_reg_field" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_reg_field_properties.png" width="800" />
</html>
|-
! Свойства класса <code>uvm_reg_field</code>
|}


=== Метод configure() ===

<!--
The first thing we do after creating a <code>uvm_reg_field</code> is configuring it. In [http://cluelogic.com/2012/10/uvm-tutorial-for-candy-lovers-register-abstraction/ Register Abstraction], we configured the <code>flavor</code> field as follows. Note that in [http://cluelogic.com/2012/10/uvm-tutorial-for-candy-lovers-register-abstraction/ Register Abstraction], we defined the <code>flavor</code> field as <code>"WO"</code> (write-only), but we defined it as <code>"RW"</code> (read/write) here to make the field more generic.
-->
Первое что делается после создания класса <code>uvm_reg_field</code> это его настройка. В статье [http://cluelogic.com/2012/10/uvm-tutorial-for-candy-lovers-register-abstraction/ Register Abstraction] настаивалось/конфигурировалось поле <code>flavor</code> как отмечено ниже. Заметим, что в статье [http://cluelogic.com/2012/10/uvm-tutorial-for-candy-lovers-register-abstraction/ Register Abstraction], задавалось поле  <code>flavor</code> как <code>"WO"</code> (write-only), но здесь зададим его <code>"RW"</code> (read/write), чтобы сделать поле более типовым.

<big><big><source lang="verilog" line=1>flavor = uvm_reg_field::type_id::create( "flavor" );
flavor.configure( .parent                 ( this ),
                  .size                   ( 3    ),
                  .lsb_pos                ( 0    ),
                  .access                 ( "RW" ),
                  .volatile               ( 0    ),
                  .reset                  ( 0    ),
                  .has_reset              ( 1    ),
                  .is_rand                ( 1    ),
                  .individually_accessible( 0    ) );</source></big></big>

<!-- 
If the <code>has_reset</code> argument is <code>1</code>, the value of <code>reset</code> argument is taken as the <code>"HARD"</code> reset value. If the <code>has_reset</code> value is <code>0</code>, the value of <code>reset</code> is ignored. The value of <code>reset</code> should match the reset state of the DUT. If you want to modify the reset value after the configuration, you can use <code>set_reset()</code> method.
-->
Если аргумент <code>has_reset</code> равен <code>1</code>, то значение аргумента  <code>reset</code> задаётся в как значение <code>"HARD"</code> сброса. Если же <code>has_reset</code> равень <code>0</code>, то значение аргумента <code>reset</code> игнорируется. Значение  <code>reset</code> должно совпадать с состоянием DUT. Если же необходимо изменить значение сброса после конфигурации, то можно использовать метод  <code>set_reset()</code>.

<big><big><source lang="verilog">
flavor.set_reset( .value( 0 ), .kind( "HARD" ) ); // kind == "HARD" by default
</source></big></big>


{| align=center
!<html><img src="http://cluelogic.com/wp-content/uploads/2013/02/jb16_configure.png" width="800" /></html>
|-
! Как действуют методы configure() и set_reset()
|}


=== Метод reset() ===

The <code>reset()</code> method resets the properties of a register field, if the <code>m_reset[kind]</code> exists. The default <code>kind</code> is <code>"HARD"</code>. If the <code>m_reset[kind]</code> does not exist, the <code>reset()</code> method does nothing. Note that the <code>reset()</code> method '''does not''' reset a register in the DUT. It only resets the properties of a register-field object.

<big><big><source lang="verilog">flavor.reset( .kind( "HARD" ) ); // kind == "HARD" by default</source></big></big>


{| align=center
!<html><img src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_reset.png" width="800" /></html>
|-
! How the reset() method works
|}


=== Метод set() ===

The <code>set()</code> method sets the desired value of a register field. The <code>set()</code> method '''does not''' set the value to a register in the DUT. It only sets the value to the <code>m_desired</code> and the <code>value</code> properties of a register-field object. To actually set the value to the register in the DUT, use <code>write()</code> or <code>update()</code> method. These methods will be explained later.

<big><big><source lang="verilog">flavor.set( .value( 1 ) );</source></big></big>


{| align=center
!<html><img alt="How the set() method works" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_set.png" width="800" /></html>
|-
! Как действует метод set()
|}


=== Метод get() ===

The <code>get()</code> method gets the desired value of a register field. The <code>get()</code> method '''does not''' get the value from a register in the DUT. It only gets the value of the <code>m_desired</code> property. To actually get the value from the DUT, use <code>read()</code> or <code>mirror()</code> methods. These methods will be explained later. Similarly to the <code>get()</code> method, there are two more ''getters'' to access the local properties. The <code>get_reset()</code> retrieves the value of the <code>m_reset[kind]</code> property, while the <code>get_mirrored_value()</code> method retrieves the value of the <code>m_mirrored</code> property.

<big><big><source lang="verilog">
uvm_reg_data_t desired_value  = flavor.get();
uvm_reg_data_t reset_value    = flavor.get_reset( .kind( "HARD" ) ); // kind == "HARD" by default
uvm_reg_data_t mirrored_value = flavor.get_mirrored_value();
</source></big></big>


{| align=center
!<html><img alt="How the get(), get_reset(), and get_mirrored_value() methods work" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_get.png" width="800" /></html>
|-
! How the get(), get_reset(), and get_mirrored_value() methods work
|}


=== Метод randomize() ===

The <code>randomize()</code> method is a SystemVerilog method. It randomizes the <code>value</code> property of a register-field object. After the randomization, the <code>post_randomize()</code> method copies the value of the <code>value</code> property to the <code>m_desired</code> property. Note that the <code>pre_randomize()</code> method copies the value of the <code>m_desired</code> to the <code>value</code> property if the <code>rand_mode</code> of the <code>value</code> property is <code>OFF</code>.

<big><big><source lang="verilog">assert( flavor.randomize() );</source></big></big>


{| align=center
!<html><img alt="How the randomize() method works" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_randomize.png" width="800" /></html>
|-
! How the randomize() method works
|}


=== Метод write() ===

The <code>write()</code> method actually writes a value to the DUT.

<big><big><source lang="verilog">
uvm_status_e status;
flavor.write( .status( status ), .value( 1 ) );</pre>
</source></big></big>

The <code>write()</code> method involves multiple steps.

# A <code>uvm_reg_item</code> object corresponding to the write operation is created.
# The <code>uvm_reg_adapter</code> converts the write operation to a corresponding bus transaction.
# The <code>uvm_driver</code> executes the bus transaction to the DUT.
# The <code>uvm_monitor</code> captures the bus transaction.
# The <code>uvm_reg_predictor</code> asks the <code>uvm_reg_adapter</code> to convert the bus transaction to a corresponding register operation.
# The register operation is converted to a <code>uvm_reg_item</code>.
# The <code>uvm_reg_item</code> is used to update the <code>value</code>, <code>m_mirrored</code>, and <code>m_desired</code> properties.

Note that if the <code>individually_accessible</code> argument was <code>0</code> when the register field was configured, the entire register containing the field is written, because the field is not individually accessible. In this case, the <code>m_mirrored</code> values are used as the write values for the other fields.


{|  align=center
!<html><img alt="How the write() method works" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_write.png" width="800" /></html>
|-
! How the write() method works
|}


=== Метод read() ===

The read() method actually reads a register value from the DUT.

<big><big><source lang="verilog">
uvm_status_e   status;
uvm_reg_data_t value;

flavor.read( .status( status ), .value( value ) );
</source></big></big>

Similarly to the <code>write()</code> method, the <code>read()</code> method involves multiple steps.

# A <code>uvm_reg_item</code> object corresponding to the read operation is created.
# The <code>uvm_reg_adapter</code> converts the read operation to a corresponding bus transaction.
# The <code>uvm_driver</code> executes the bus transaction to the DUT.
# The <code>uvm_reg_apapter</code> converts the bus transaction with read data to a register operation.
# The <code>read()</code> method returns the read value to the caller.
# In the mean time, the <code>uvm_monitor</code> captures the bus transaction.
# The <code>uvm_reg_predictor</code> asks the <code>uvm_reg_adapter</code> to convert the bus transaction to a corresponding register operation.
# The register operation is converted to a <code>uvm_reg_item</code>.
# The <code>uvm_reg_item</code> is used to update the <code>value</code>, <code>m_mirrored</code>, and <code>m_desired</code> properties.

Note that if the <code>individually_accessible</code> argument was <code>0</code> when the register field was configured, the entire register containing the field is read. In this case, the <code>m_mirrored</code> values are updated for the other fields as well.


{| align=center
!<html><img alt="How the read() method works" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_read.png" width="800" /></html>
|-
! How the read() method works
|}


=== Метод update() ===

The <code>update()</code> method actually writes a register value to the DUT. The <code>update()</code> method belongs to the <code>uvm_reg</code> class. The <code>uvm_reg_field</code> class does not have the <code>update()</code> method.

<big><big><source lang="verilog">
uvm_status_e status;
 
jb_recipe_reg.update( .status( status ) );
</source></big></big>

The differences between the <code>write()</code> method and the <code>update()</code> method are:

* The <code>write()</code> method takes a value as its argument, while the <code>update()</code> method uses the value of the <code>m_desired</code> property as the value to write.
* The <code>update()</code> method writes the value only if the <code>m_mirrored</code> and the <code>m_desired</code> are not equal.


{| align=center
!<html><img alt="Before the update() is executed" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_before_update.png" width="800" /></html>
|-
! Before the update() is executed
|}


The <code>update()</code> method internally calls the <code>write( .value( m_desired ) )</code>. Because of this, the value of the <code>m_mirrored</code> will be updated as well, after the update.


{| align=center
!<html><img alt="After the update() is executed" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_after_update.png" width="800" /></html>
|-
! After the update() is executed
|}


=== Метод mirror() ===

The <code>mirror()</code> method actually reads a register from the DUT.

<big><big><source lang="verilog">
uvm_status_e status;
 
flavor.mirror( .status( status ), .check( UVM_CHECK ) );
</source></big></big>

The differences between the <code>read()</code> method and the <code>mirror()</code> method are:

* The <code>read()</code> method returns the register value to the caller, while the <code>mirror()</code> method does not return the register value. The <code>mirror()</code> method only updates the value of the <code>m_mirrored</code> property.
* The <code>mirror()</code> method compares the read value against the <code>m_desired</code> if the value of the <code>check</code> argument is <code>UVM_CHECK</code>. <del>Note that the UVM Class Library document states that it compares the read value against the mirrored value, but if you look at the line 2,944 of <code>uvm_reg.svh</code> of uvm-1.1c code base, it actually compares against the desired value, not against the mirrored value.</del> '''(Update 4/11/2014: uvm-1.1d code base has corrected this issue. The <code>mirror()</code> compares the read value against the mirrored value now. Please see the line 2,951 of <code>uvm_reg.svh</code> if you are curious about this fix.)''' Another caveat about the check is that if you set the <code>volatile</code> argument to be <code>1</code> when you configured the register field, the register field won’t be checked even though you set the <code>check</code> argument to be <code>UVM_CHECK</code>. This is because we cannot predict the value of the register field deterministically as it might have been changed (volatile) in the DUT.

The <code>mirror()</code> method internally calls <code>do_read()</code> method. This is the same method the <code>read()</code> method internally calls. Because of this, the <code>mirror()</code> method will update the <code>value</code> and the <code>m_desired</code> properties, in addition to the <code>m_mirrored</code> property.


{| align=center
!<html><img alt="How the mirror() method works" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_mirror.png" width="800" /></html>
|-
! How the mirror() method works
|}


=== Метод predict() ===

The <code>predict()</code> method updates the mirrored value.

<big><big><source lang="verilog">
flavor.predict( .value( 1 ) );
</source></big></big>

The <code>predict()</code> method also updates the <code>value</code> and the <code>m_desired</code> properties.


{| align=center
!<html><img alt="How the predict() method works" src="http://cluelogic.com/wp-content/uploads/2013/01/jb16_predict.png" width="800" /></html>
|-
! How the predict() method works
|}


== Summary ==

The table below summarizes how each method updates the properties of the register-field object.

<div>
{| border="1"
! Method
! <code>m_reset
["HARD"]</code>
! <code>value</code>
! <code>m_desired</code>
! <code>m_mirrored</code>
! DUT
|-
| <code>configure(
.reset(val),
.has_reset(1))</code>
| set the value of <code>val</code>
|
|
|
|
|-
| <code>set_reset(val)</code>
| set the value of <code>val</code>
|
|
|
|
|-
| <code>reset()</code>
|
| copy the value of <code>m_reset
["HARD"]</code>
| copy the value of <code>m_reset
["HARD"]</code>
| copy the value of <code>m_reset
["HARD"]</code>
|
|-
| <code>set(val)</code>
|
| set the value of <code>val</code>
| set the value of <code>val</code>
|
|
|-
| <code>get_reset()</code>
| return the value of <code>m_reset
["HARD"]</code>
|
|
|
|
|-
| <code>get()</code>
|
|
| return the value of <code>m_desired</code>
|
|
|-
| <code>get_mirrored_value()</code>
|
|
|
| return the value of <code>m_mirrored</code>
|
|-
| <code>randomize()</code>
|
| randomize
| copy the value of <code>value</code>
|
|
|-
| <code>write(.value(val))</code>
|
| set the value of <code>val</code>
| set the value of <code>val</code>
| set the value of <code>val</code>
| write the value of <code>val</code>
|-
| <code>read(.value(val))</code>
|
| set the read value
| set the read value
| set the read value
| read the register
|-
| <code>update()</code>
|
| set the value of <code>m_desired</code>
| set the value of <code>m_desired</code>
| set the value of <code>m_desired</code>
| write the value of <code>m_desired</code>
|-
| <code>mirror()</code>
|
| set the read value
| set the read value
| set the read value
| read the register
|-
| <code>predict(
.value(val))</code>
|
| set the value of <code>val</code>
| set the value of <code>val</code>
| set the value of <code>val</code>
|
|}
</div>

In this post, we only covered so-called front-door access. We will cover back-door access in a separate post. I hope this tutorial helped you to understand the register access methods.