Icarus Verilog GTKWave Guide

Icarus_Verilog_GTKWave_guide

Icarus_Verilog_GTKWave_guide

User Manual:

Open the PDF directly: View PDF PDF.
Page Count: 9

DownloadIcarus Verilog GTKWave Guide
Open PDF In BrowserView PDF
Icarus Verilog + GTKWave Guide
with support for MIPS architecture implementation
BY IOANNIS KONSTADELIAS 

Introduction
Here is a guide for those who want to develop and test hardware on Linux OS. The positives of
working/developing on Linux is that you don't have to work on Windows and thus in most cases
you can have everything for free. The most important thing is that you have everything open source.
Noone restricts you from using something that already exists, in the way you want. That's the main
reason I prefer solutions in Linux like that than examples like ModelSim. They are ok, but they
restict me.
This guide isn't supposed to include every little detail of either Icarus Verilog or GTKWave, but the
things you have to do to use some extra functionality that probably you haven't in mind to use or
you couldn't know how to do it because of lack of support on the Internet. The main goal of this
guide is to show you how to view in GTKWave the values of the signals of your program in a more
human way. In a way that will help you a lot with debugging. In a way such the image below:

Let's get it started!

Icarus Verilog
About
Icarus Verilog is a free Verilog simulation and synthesis tool. It compiles source code written in
Verilog (IEEE-1364) into some target format. It also generates netlists for the synthesis part. It's by
far the best free tool and many people work on that making it more and more complete day by day.
Well, because I use Ubuntu (and specifically 13.10, which doesn't mean anything for the whole
guide), I'll show you everything with commands for that OS.
Installation
Open a terminal and type (or copy-paste) the below commands.
Firstly, add a new ppa:
$ sudo add­apt­repository ppa:team­electronics/ppa
Then, update the local repository cache:
$ sudo apt­get update
Finally, install (update) the Icarus Verilog package
$ sudo apt­get install iverilog
Usage
Let's say you have the source files one.v and two.v to compile. A quick way to compile them is
just by typing
$ iverilog one.v two.v
In that way, a.out is created and that's our binary:
$ ls
$ a.out one.v two.v
Now we can execute a.out with vvp command:
$ vvp a.out
If you want to specify another name for the binary, you can use the ­o option just like in gcc:
$ iverilog ­o my_out.out one.v two.v
$ ls
$ one.v two.v my_out.v

Well, all these are ok, but if you want to write on big, or complex verilog code, you should also use
during the compilation the ­Wall option:
$ iverilog ­Wall ­o my_out.out one.v two.v
­Wall option works in the same logic as in gcc. It enables more warnings during the compilation,
that can help you find that little thing that destroys the functionality of your hardware. And yes,
warnings in our case can affect the functionality of our design more often than programming in C.
So ­Wall is a must. You should also consider on ­Winfloop. You can imagine its usage.
Well, that's all from me about Icarus Verilog. If you want to learn more about it, you should visit it's
official website at http://iverilog.icarus.com/ and for further How Tos and support its official wiki at
http://iverilog.wikia.com/wiki/Main_Page. I also recommend you to subscribe to the mailing list
https://lists.sourceforge.net/lists/listinfo/iverilog-devel.

GTKWave
About
GTKWave is a fully featured GTK+ wave viewer for Unix, Win32, and Mac OSX which reads
LXT, LXT2, VZT, FST, and GHW files as well as standard Verilog VCD/EVCD files and allows
their viewing. Its official website is at http://gtkwave.sourceforge.net/.
GTKWave is the best free wave viewer and is the recommended viewer by Icarus Verilog
developer.
Installation
Open a terminal and type (or copy-paste) the below commands.
Firstly, update the local repository cache:
$ sudo apt­get update
Finally, install (update) the GTKWave package
$ sudo apt­get install gtkwave
Usage
It's usage it's really easy. In order to open GTKWave you can either type gtkwave in terminal, or by
clicking on its icon:
As you can imagine, there's nothing you can do by simply opening it alone. In order to see some
waveform, you must open through GTKWave some saved dumpfiles. In our case we will use .vcd
files, which are produced for our Icarus Verilog tool.
Creation of .vcd files
In our verilog code in an initial block we should include these two lines:
module Test;
...
initial begin
$dumpfile("my_dumpfile.vcd");
$dumpvars(0, my_module_name);
end
...
endmodule

NOTE: In case you have defined arrays in our design. E.g.:
module Mem (bla, bla, bla, …);
...
reg [M:0] data [N:0];
…
endmodule
its important that you include also these lines in your initial block (for every array you
defined), in order to be able to see the array signals in GTKWave:
module Test;
integer i;
...
initial begin
$dumpfile("my_dumpfile.vcd");
$dumpvars(0, my_module_name);
for(i = 0; i < M; i = i + 1)
$dumpvars(1, full.path.to.array.data[i]);
end
...
endmodule
It's probable that the last action will result a Warning type like that below during the compilation.
You don't need to worry:
VCD warning: array word cpu_tb.cpu0.cpu_dp.data_mem.data[39] will
conflict with an escaped identifier.

Opening VCD files with GTKWave
You can open o VCD file either from terminal or by the GUI. In this guide I'll use as much as I can
the terminal way. It's more catchy and for sure it'll save as useful time, especially when we test a
hardware design. So here we go:
$ gtkwave my_dumpfile.vcd

As you can see, there aren't any waves in the Wave window. That's because the user has to choose
which signals want to see. In order to choose the signal you want to see, you should go in the left
window with the SST name. Here you can see your hardware hierarchy. By clicking the + of every
instance, you can see the signals that are related with that instance. Then you can drag&drop them,
or copy&paste them in the Signals window. Voilà! Here are the waveforms.
NOTE: When I close and re-open GTKWave with the same VCD file, the signals that I have chosen
before are lost. Should I always spend all this time by only inserting signals to be shown?
The answer is “of course not!”. After you have decided on the signals you want to be shown you
can click File → Write Save File As and save a file with a .gtkw ending. Let's say for this example
config.gtkw. From now and then you can open it with your .vcd file and get instantly to the point,
which is debugging ;) :
$ gtkwave my_dumpfile.vcd config.gtkw
A better view on things
Now you are able to see the values of the signals in a faster way. But how about the interpretation?
The values of the signals are in hexadecimal format and all waves are colored green. Yes! You can
change these properties.
You can change them by right-clicking on the signals and choosing Data Format or Color Format.
You are free to experiment with these options. When you have the desired optical result you can
save your configuration by going File → Write Save File.
All crystal clear until now...
Take GTKWave by your side
When you right-click on a signal and especially in the Data Format option you can see the choice
Translate Filter File. We'll stick with that option for the rest of this guide. It's a very powerful
option, but it goes by the most times as there's not there.
With that option you can translate the values of the signals in a more human way. You can do this
by writing a .txt translation filter. It isn't as complex as it sounds. Let's see an example:
You write a MIPS processor in Verilog. You have a Register File with the register of this
architecture. You don't want to see the names of them as hex numbers, even as decimal. So you can
simply write a .txt file like this:
00

$0 ­ $zero

01

$1 ­ $at

02

$2 ­ $v0
...

The first column represents the hexadecimal values that a signal that represents registers can have.
The second more complex column is what we want to see instead of the first column. And that's it!.
Now we only have to go to the signal we want to change, right-click on it, choose Data Format →
Translate Filter File → Enable and Select. Then you have to browse to the filter you wrote,
choose it and hit OK. Voilà! (Make sure that the data format is in hex before you choose the filter).
Now you can save again your configuration. You're ready to go!
In that way you can filter whatever you want. From registers' names and opcodes to full
instructions! Full instructions???
Supposing that we talk about a MIPS implementation, we initialize the Instruction Memory array
with the verilog command $readmemb() from a file. This file must be of such format:
@0

000000_01000_00100_01001_00000_100000

// add

$t1, $t0, $a0

@4

101011_01011_01001_0000000000000101

// sw

$t1, 5($t3)

@8

100011_01011_10010_0000000000000101

// lw

$s2, 5($t3)

@c

000000_01000_10011_01001_00000_101010

// slt

$t1, $t0, $s3

@10

000100_01001_10010_1111111111111011

// beq

$t1, $s2, ­5

where the first column is the address in hex of the command (in MIPS every address takes 32 bits)
and the second column is the binary representation of the MIPS assembly instruction. You can
include some C++ style comments if you like to.
This special command is used like that in an initial block:
$readmemb("program.mbin", full.path.to.data);

Before we continue with my favorite part, let's make our life a lot easier.

Makefile: Our best friend
Let's create a simple make file with what we saw up to here. The scenario is that we have a design
in Verilog and we want to run it and see the waveforms instantly, without touching the mouse at all:

CC = iverilog
FLAGS = ­Wall ­Winfloop
library_input: one.v two.v testbench.v
$(CC) $(FLAGS) ­o test one.v two.v testbench.v
vvp test
gtkwave my_dumpfile.vcd config.gtkw
The makefile should be in the same directory with the other .v, .vcd and .gtkw files in our example.
You simply execute this kind of script by just typing make in the terminal.
$ make
Now, when you want to run your verilog code you can just type make and everything comes up for
you.
No pain No gain
Now comes the last but the best part. How can you represent the instructions in GTKWave from the
moment you have to change them from test to test? It's time consuming. It's also time consuming to
convert an instruction to its binary representation in order to test the processor. There are some
online converters, but they restrict you in the way that you cannot insert a whole program and get its
binary. Furthermore, its boring to fill the file with addresses (@0, @4, …). There are times you
want to run a big program, and you spend the ¾ time in converting it in a format that
$readmemb() can read it.
For this reason there's no solution but to develop your own parser. You'll spend a few hours to finish
it, but the results will save you a lot more hours in the testing of your code. For me I developed a
MIPS assembly to MIPS binary parser. I named it MASMBIN and works like this: It takes as first
argument the .masm assembly file (input), and as a second argument the .mbin binary file (output):
$ ./masmbin input.masm output.mbin
In this stage, it can take 2 options:
­c
Shows in .mbin file in comments the parsed instruction
­f filter.txt
Outputs a .txt filter file for GTKWave

So as you see, this parser does another job with the right option: It creates the filter that we wanted
to represent the values of signals in GTKWave as higher level instructions.
Now we're going to put everything together and have a nice MIPS assembly simulator.
Let's start from the conventions you should take:
•

In $readmemb(), the file must be the same (for our example, program.mbin)

•

The filter for our instructions in GTKWave should have a stable name (for our example,
translate_instruction.txt)

Ok, now lets write our final makefile:
CC = iverilog
FLAGS = ­Wall ­Winfloop
library_input: one.v two.v testbench.v
$(CC) $(FLAGS) ­o test one.v two.v testbench.v
./masmbin ­c ­f translate_instruction.txt program.masm
program.mbin
vvp test
gtkwave my_dumpfile.vcd config.gtkw
Now, everything is in place. You can edit the program.masm file as you want, save it, and execute
make. Everything will come up automatically.
If you want to download the new version of MASMBIN you can do so at
https://dl.dropboxusercontent.com/u/63037297/MIPS_asm_to_bin_parser_v2.0.tar.gz.
For the previous version refer to the github repository:
https://github.com/gon1332/MIPS-asm-to-bin-parser
Source Exif Data: 
File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.4
Linearized                      : No
Page Count                      : 9
Language                        : en-US
Author                          : gon1332
Creator                         : Writer
Producer                        : LibreOffice 4.1
Create Date                     : 2013:12:13 04:38:52+02:00
EXIF Metadata provided by EXIF.tools

Navigation menu