Opencores.org



SV Directed Test Bench User Guide

by

Ken Campbell

Version: 1.1

January 2019

Copyright (c) 2019 Ken Campbell.

Licensed under the Apache License, Version 2.0 (the "License");

you may not use this file except in compliance with the License.

You may obtain a copy of the License at



Unless required by applicable law or agreed to in writing, software

distributed under the License is distributed on an "AS IS" BASIS,

WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.

See the License for the specific language governing permissions and

limitations under the License

Revision History:

|Version |Revised By |Description |Date |

| | | | |

|1.0 |Ken Campbell |Initial release |August 15, 2014 |

|1.1 |Ken Campbell |Updates after support of user. |April 2019 |

| | | | |

| | | | |

| | | | |

| | | | |

| | | | |

| | | | |

Table of Contents:

1. Introduction 6

1.1 HISTORY 6

1.2 OVER VIEW 6

1.3 SCOPE 6

1.4 INTERESTED PARTIES 6

1.5 WHY? 6

1.6 DOCUMENT ACRONYMS 7

2. TEST ENVIRONMENT USAGE FLOW 8

2.1 THE DUT 9

2.2 CHOOSE THE METHODOLOGY 9

2.3 GENERATING THE TEST BENCH 9

2.4 CREATING THE INITIAL INSTRUCTIONS 9

2.5 WRITING TEST CASES (STM FILES) 10

2.6 THE REGRESSION SET 11

3. THE SV DIRECTED TEST BENCH 11

3.1 RECOMMENDED DIRECTORY AND FILE STRUCTURE 11

3.2 DEFAULT TEST BENCH STRUCTURE 12

3.3 IMPLEMENTATION VARIATIONS 12

3.3.1 INTERNAL TEST BENCH VARIANT 12

3.3.2 MULTI SCRIPT IMPLEMENTATIONS 13

3.4 SCRIPT PARSING CONVENTIONS 13

3.4.1 CASE 13

3.4.2 WHITE SPACE 13

3.4.3 COMMENTS 13

3.4.4 VARIABLES 14

3.4.5 SPECIAL VARIABLES 14

3.4.6 CONDITION VARIABLES 14

3.4.7 NUMBER NOTATION: 15

3.4.8 DYNAMIC TEXT STRINGS 15

3.5 SEARCH ORDER 16

4. TEST ENVIRONMENT INSTRUCTIONS 16

4.1 DEFAULT INSTRUCTIONS 16

4.2 USER DEFINED INSTRUCTIONS 19

4.2.1 TASKS 20

4.2.2 CONCURRENCY 20

5. TEST BENCH WORKING DETAILS 20

5.1 SYSTEMVERILOG CLASSES 21

5.2 USER FUNCTIONS 21

6. TEST BENCH GENERATOR TOOL 21

6.1 TB_GEN USAGE 22

6.2 TB_GEN STEP BY STEP 23

7. RELEASES AND UPDATES. 23

7.1 UPDATE #1 APRIL 2019 23

APPENDIX A: EXAMPLES 24

APPENDIX A: EXAMPLES 24

TABLE OF FIGURES

Figure 1 - Default Test Bench Structure 12

Figure 2: Script Driven Processor Implementation 13

Index of Tables

Introduction

1 History

The VHDL test bench system was released in 2007. This version is the same implementation but is implemented using SystemVerilog (SV). Using some of the facilities of SV this implementation duplicates the scripting syntax of the VHDL version.

2 Over View

The SV Directed test bench is a collection of SystemVerilog classes and functions which enable the user to create their own scripting instructions for directed test stimulus. The stimulus script or test case contains the instructions in a regular ASCII text file. The functionality of the instructions are coded in SystemVerilog as part of the test bench. The SV Directed package contains classes and functions to read, parse and execute the test script (stimulus file, test case, script). The script is evaluated in two passes. The first pass reads the instructions from the stimulus file, checks the validity of the instructions, adds valid instructions to the cmd_lst object and creates the variable list. The first pass leaves everything needed in memory and happens at time zero of the simulation. The second pass is the execution pass. Instructions are referenced by their index numbers and return the instruction text, up to 6 parameters in integer form and possibly one text string. This is then fed down an “elsif” chain where the instruction text is used to choose the correct SV instruction sequence. At this point each instruction could be controlling the timing of the test case.

3 Scope

This document provides the usage recommendations and detailed functionality of the test bench environment. It is expected that once this document has been read, the user will have the knowledge to use the environment.

4 Interested Parties

All Verilog designers can benefit from the use of a simple but flexible foundation. The SV Directed test bench package provides a very good starting point for any effort requiring verification using the SV language. This document should provide you all you need to use and implement the package.

VHDL test bench package users may find this interesting they are told they MUST use SystemVerilog. In fact if one has used the VHDL system, this SV system will be VERY familure.

5 Why?

There are several reasons why you may want to use this test bench package

Used the VHDL test bench package before and liked it.

Don’t want a heavy verification environment. (Don’t want to use UVM)

As a designer, want an environment that is quick and easy to use.

You do not need a fully randomized test environment to do your verification.

Have access to SV but not UVM (FPGA tools)

6 Document Acronyms

|Term |Definition |

| | |

|DUT |Design Under Test |

| | |

|FPGA |Field Programmable Gate Array |

| | |

| | |

| | |

|SV |SystemVerilog |

| | |

| | |

|UVM |(ultimate, unbelievable, Ultra) Universal Verification Methodology |

|VHDL |Very High Level Design Language |

| | |

| | |

Test Environment Usage Flow

[pic]

Illustration 1: Typical Test Bench Usage Flow

Consider the flow illustration above, Illustration 1: Typical Test Bench Usage Flow. This flow diagram presents a basic flow of how the test bench package can be used to create a verification environment, to test a DUT.

The following assumes that the DUT is a synchronous design that has a reset input.

1 The DUT

To start using the environment it is required that the Verilog module of the DUT be available. With the DUT module file available, the test bench can be generated using the tb_gen tool. The tb_gen tool uses the pin definitions within the module definition to generate several parts of the test bench structure.

2 Choose The Methodology

Depending on how you like to do your verification, the test bench will be coded differently. If you are going to create dynamic self-checking test cases or you are going to record vectors and compare to some good vector set, the test bench will be implemented in different ways. The commands that need to be created will be different.

As a preference, dynamic self-checking test cases should be the first choice. This will suggest that you will have to create commands to “read” various outputs, and “verify” the outputs are as expected.

Vector collection and compare should be the last choice of method. But this implementation may include creating instructions that enable the test environment to collect different outputs at different times.

3 Generating the Test Bench

Using the tb_gen tool, generate the test bench. Once the initial files have been generated, the test bench writer will then edit them for specifics. The tb_mod.sv file is the users play ground. It is recommended that all edits, instructions and tasks be placed in this file, or a file that is included by this file. The other files that make up the test bench should not be edited because they can be regenerated if the module pinout changes significantly. The tb_mod.sv file is optionally generated, as you would not want to over write the code that has been created for this test environment.

4 Creating the Initial Instructions

The test bench package file cmd_lst.sv contains several default instructions. Those are presented later in this document. It would be beneficial to review these instructions and be familiar with their implementation before creating your own instructions.

As a starting point, the first instruction that would be included in all test environments is the “reset” type instruction. This instruction is the one that applies the reset to the DUT and test bench elements. It is expected that all items need to be triggered to get into a default starting state. There may be several reset type instructions in one environment, one to reset everything, one to reset just the DUT and could be one to reset test bench models. These instructions would be created if the elements of the test environment need to be reset independent of each other.

Another very common instruction to consider is the “wait” type instructions. These instructions are used to time activities in a test case. For instance, an instruction to wait so many clock cycles is useful for just waiting for some know amount of time. A wait on interrupt is also useful to enable the test case to wait till some indicator triggers the test case to continue. As an addition to the wait instruction there should be a wait_max type instruction that enables the test writer to set a time out for the wait. This is so that if the single event you are waiting for never happens, the environment can terminate the simulation.

To facilitate control of the DUT, write type instructions will be needed. This type of instruction will access the DUT possibly through an address and data buss or just the placing of values on input pins. Whatever the write type instruction(s) do, they will be the instructions used to setup and control the DUT.

To enable the collection of data, some kind of “read” instruction will be created. This instruction reads a target item, data bus, register, output pins, and puts the value into a common place. As in, all read instructions put the data in the same place, this could be a variable.

To confirm the data values are as expected, some kind of “verify” instructions will be created. The verify instruction(s) enable the test writer to check that the value of the last read item is as expected. Several types of verify instructions can be created, verify (whole word), verify_slice, verify_bit and they would all look to the same place for the data to compare.

The four basic instruction groups described above will enable the test bench developer to get started at the test bench / test case creation. As new instructions are needed, they are added.

Depending on the complexity of the test bench environment, there may be some instructions that are needed to initialize, access and or control test bench models.

5 Writing Test Cases (stm files)

Once the test bench initial instructions have been created, test cases may start to be created. The test case is created by a person or persons through the use of a simple text editor. The test case writer will create the file containing instructions defined in the tb_mod.sv file. As an example, please examine the following:

-- This is an example test case

-- using instructions defined in previous sections. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download