CPULATOR Functionalities

F28HS Hardware-Software Interface:
Systems Programming
Hans-Wolfgang Loidl
School of Mathematical and Computer Sciences,

Heriot-Watt University, Edinburgh
Semester 2 — 2020/21
0
No proprietary software has been used in producing these slides
Hans-Wolfgang Loidl (Heriot-Watt Univ) F28HS Hardware-Software Interface 2020/21 1 / 39
Outline
1 Tutorial 1: ARM Assembler Exercise: Caesar Cipher (ROT13)
2 Tutorial 2: Programming an LED
3 Tutorial 3: Programming a Button
4 Tutorial 4: Programming a HEX display
5 Tutorial 5: Cache-friendly Programming
Hans-Wolfgang Loidl (Heriot-Watt Univ) F28HS Hardware-Software Interface 2020/21 2 / 39

Overview of the Caesar Cipher (ROT13) Exercises
This sequence of screencasts is intended to give you a deeper

understanding of ARM Assembler programming
We will develop one non-trivial program step-by-step in ARM
Assembler
Follow the online screencasts, and do the matching lab exercises
A Caesar cipher is a very simple cipher, that encodes a string by
rotating each char by a certain amount
For example, with rotation value of 1: a → b, b → c, ..., z → a
A Caesar cipher with rotation value of 13 (ROT13) is sometimes
used for low-security encryption.
Background on Caesar cipher:
I see slides (15–20) in Introduction to Cryptography
I exercise of breaking (optional, search for “Exercise 1: Caesar
cipher”)
I see also this wikipedia page on ROT13
Hans-Wolfgang Loidl (Heriot-Watt Univ) F28HS Hardware-Software Interface Tutorial 1: Prging an LED 3 / 39
Structure of the Sequence of Screencasts
TASK: Implement a Caesar cipher (rotation cipher) in ARM

Assembler
Background: you should have watched the ARM Assembler
programming leactures (F28HS lectures Assembler 1-5) and done
the matching tutorials and lab exercises
Structure:
I Version 0: C version
I Version 1: plain (lower case letters only)
I Version 2: printing the result (using printf libc function)
I Version 3: using subroutines
I Version 4: covering upper case letters as well
I Version 5: using word load and store (for fewer memory operations)
I Version 6: printing the result (in Assembler as well)
Tutorial 2: Programming an LED
This tutorial will deal with programming an LED output device.

This is the “hello world” program for external devices.
It will deal with programming techniques common to other output
devices.
The learning objective of this exercise is to learn how to directly
control an external device through C and Assembler programs.
Note: This year we will be using the online CPUlator (an ARM
simulator) for programming (virtual) external devices. This uses a
different chipset (De1-SoC by Altera) from the chip-set that’s on a RPi
(BCM2836 SoC by Broadcom). The RPi code that you might find
online, will NOT work on the CPUlator.
Documentation for the De1-SoC (Altera) chip is available here.
Tutorial 2: Programming an LED
This tutorial will deal with programming an LED output device.

This is the “hello world” program for external devices.
It will deal with programming techniques common to other output
devices.
The learning objective of this exercise is to learn how to directly
control an external device through C and Assembler programs.
Note: This year we will be using the online CPUlator (an ARM
simulator) for programming (virtual) external devices. This uses a
different chipset (De1-SoC by Altera) from the chip-set that’s on a RPi
(BCM2836 SoC by Broadcom). The RPi code that you might find
online, will NOT work on the CPUlator.
Documentation for the De1-SoC (Altera) chip is available here.
The high-level picture
From the main chip we want to control an (external) device, here

an LED.
We use one of the LEDs that is connected to the chip on the
CPUlator.
Logically we want to send 1 bit to this device to turn it on/off.
The low-level picture
Programmatically we need to go through these steps:

Identify the peripheral registers, in charge of LEDs, from the Altera
De1-SoC manual
Identify the bits in the registers that control a specific LED
Then, directly access the device via memory read/writes into
these registers
Note: In the program, turning on an LED is typically just one write
operation
Finding the registers in the manual
The relevant information can be found the hardware manual for

the relevant chip (for us, the DE1-SoC chip by Altera)
Such manuals are very detailed, since they need to cover all
issues about programming and usage (for CS as well as EE
applications)
To find the info in these documents, look for sections on
components / peripherals (for us, “2. DE1-SoC Computer
Contents”)
Specifically, we need the information on the LEDs, in Section
“2.5.7. Red LED Parallel Port”.
Data explained
The picture (and discussion in the manual) tells us

The lowest 10 bits in the register at address 0xFF200020 are in
charge of 10 LEDs hard-wired to the chip
The remaining bits in the register are unused
Note that the processor has a word-size of 32-bits
Each of the lower bits controls one LED, and turns it on (bit: 1) or
off (bit: 0)
Controlling an LED in C
Here is a simple program to turn on the first LED (from the right):
#define GPIOBASE 0xFF200000
static unsigned long *gpio = GPIOBASE;
void main () {
*gpio = 0b00000001;
}
In essence, this program performs one write operation: it writes the

value 1 into the memory address 0xFF200000, which we know is in
charge of the LEDs.
The value 1 represents the bitmask 0b00000001, meaning turn on the
first (rightmost) LED.
CPUlator after running the program
Controlling an LED in ARM Assembler
We can achieve the same behaviour in ARM Assembler:

.text
LDR R0, =LEDBASE @ physical address of LEDs
MOV R2, #0b00000001 @ bitmask to turn on first LED
STR R2, [R0, #0] @ turn it on
.data
.equ LEDBASE, 0xFF200000
First we load the bitmask 0b00000001 (first LED on, all other
LEDs off) into register R2.
Then we write this value into the LED register.
The behaviour is the same as in the C program before.
Next Task: Blinking LED in ARM Assembler
We now want to implement a blinking LED

This should be coded in ARM Assembler
We use the same principles as before, but need an (infinite) loop
We also need some kind of delay / timer functionality
Blinking LED in ARM Assembler
.global _start
_start:
start: LDR R2, =LEDBASE @ physical address of the LED

MOV R1, #10 @ counter
loop: NOP
EOR R3, #0b1 @ flip current value: 1->0, 0->1
STR R3, [R2] @ store value into LED register
BL delay
SUBS R1, #1 @ decrement counter & set flags
BNE loop @ R1 != 0 => continue loop
.data
.equ LEDBASE, 0xFF200000
One loop (with counter in R1), flipping a value (R3) in each iteration.
The delay function in ARM Assembler
delay: PUSH {R4-R7,LR}

MOV R4, #0x0f
loop1: LDR R5, =#0xffff
body: NOP
SUBS R5, #1
BNE body
SUBS R4, #1
BNE loop1
POP {R4-R7,LR}
BX LR
In essence, two nested loops that do nothing.

The choice of the counters depends on the chip / simulator speed.
Summary
Controlling a simple external device means logically sending 1 bit

of information (on/off)
Realising this control means physically writing into special
registers which have special meaning
The information on the special meaning is usually in bulky
hardware-description documentation
Once uncovered, the code for direct device control is fairly short
Tutorial 3: Programming a Button input device
In this tutorial we want to use a button, as simulated on the

CPUlator
Usually, these devices would be connected through a breadboard.
A button is the simplest input device that we will cover.
The CPUlator simulates 4 push-buttons
The code needed to control is typical for such devices.
This tutorial deals with programming a button as input device.
Hans-Wolfgang Loidl (Heriot-Watt Univ) F28HS Hardware-Software Interface Tutorial 3: Prging a Button 17 / 39
Accessing Push-Buttons on the CPUlator
Core Techniques
In the LED tutorial, we have seen that we first need to identify the
registers that give control to the device.
For that we will again look into the De1-SoC Manual
We need to identify the documentation for peripheral devices:
“2.5.10 Pushbutton Key Parallel Port” on page 10.
The low-level picture
We need to read from address 0xFF200050 to determine the

state of the buttons
Again there is a one-to-one relationship between the lowest 4 bits
in the register and the buttons:
if button 0 is pressed, bit 0 will be set to 1
if button 1 is pressed, bit 1 will be set to 1 etc
Thus, reading from the register and masking the right bit is all we
need to do.
Reading from a Button in ARM Assembler
.global _start
_start:
LDR R0, =BUTTONBASE @ physical address of the buttons
LDR R3, [R0] @ read value from BUTTON register
@ this will be a binary
representation of buttons
pressed
@ i.e. 1st and 2nd button pressed
=> 0b11 == 3
.data
.equ BUTTONBASE, 0xFF200050
We read the contents of the button register, containing the status of all
buttons, into register R3.
A button Press Counter in ARM Assembler
.global _start
_start:
start: LDR R0, =BUTTONBASE @ physical address of buttons
MOV R1, #10 @ loop counter
MOV R2, #0 @ button press counter
MOV R5, #1
LSL R5, #BUTTON_NO @ bitmask for testing BUTTON_NO
loop0: NOP
LDR R3, [R0] @ read value from BUTTON
register
CMP R3, R5
BNE nocount
ADD R2, R2, #1 @ increment press counter
nocount: BL delay
SUBS R1, #1
BNE loop0
.data
.equ BUTTON_NO, 0
(Discussion)
This code continuously reads the button register

We prepare a bit-mask in R5 to identify the bit in charge of button
number BUTTON_NO.
We first read the contents of the button register into register R3.
We use the bit-mask in R5 to extract the bit in charge of button
number BUTTON_NO.
If this bit is on, we increment the counter in R2.
We finish the program after a fixed number of iterations (could
also use an infinite loop here)
When running this in the CPUlator do check that R2 acts as a
counter of how often a press has been registered
Note: In practice, you want to count distinct button presses, i.e.
don’t count a continuous button press several times
.global _start
_start:
MOV R1, #10 @ loop counter
MOV R2, #0 @ button press counter
MOV R5, #1
LSL R5, #BUTTON_NO @ bitmask for testing BUTTON_NO
loop0: NOP
LDR R3, [R0] @ read value from register
AND R3, R5 @ extract the relevant bit
CMP R3, #0 @ check whether it is on
BEQ nocount
ADD R2, R2, #1 @ increment press counter
nocount: BL delay
SUBS R1, #1
BNE loop0
.data
.equ BUTTON_NO, 0
This version checks only for the specific button BUTTON_NO.
The delay function is the same as before
delay: PUSH {R4-R7,LR}

MOV R4, #0x0f
loop1: LDR R5, =#0xffff
body: NOP
SUBS R5, #1
BNE body
SUBS R4, #1
BNE loop1
POP {R4-R7,LR}
BX LR
These are in essence two nested loops, doing nothing in the body.
Needed to provide a delay between button reads.
Summary
Reading input from a button works in the same way as writing to

the LED:
I We need to read from a certain bit in a register, in charge of buttons.
With the button you have a basic input device to communicate
with the system
Other devices use the same principle, but communicate more
data, e.g. an 8-bit wide parallel port for I/O communication
With the CPUlator’s restrictions, we don’t discuss these devices in
any more detail this year
Tutorial 4: Programming a HEX display
The De1-SoC chip (by Altera) on the CPUlator, provides several

7-segment displays, or HEX displays
These displays can be used to display numbers, and some letters
Therefore, the output is more meaningful than just one LED
The principle to program the display is the same as for the LED
But first we need to find the relevant information in the manual:
“2.5.8 7-Segment Displays Parallel Port” (page 10)
Program the 7-Segment (HEX) Displays in Assembler
Note:
1 segment controlled by 1 bit
The address for the data register of the HEX displays is

0xFF200020
The 7 segments in the HEX display are each controlled by 1 bit in
the HEX data register
The mapping of the bit positions to the segments is shown in the
picture on the right, e.g. bit 0 controls the upper bar in display 1
We need 7 bits per HEX display in the register (8th bit unused).
The next 7 bits are in charge of the next HEX display (HEX1) etc.
Program the 7-Segment (HEX) Displays in Assembler
Note:
1 segment controlled by 1 bit
The address for the data register of the HEX displays is

0xFF200020
The 7 segments in the HEX display are each controlled by 1 bit in
the HEX data register
The mapping of the bit positions to the segments is shown in the
picture on the right, e.g. bit 0 controls the upper bar in display 1
We need 7 bits per HEX display in the register (8th bit unused).
The next 7 bits are in charge of the next HEX display (HEX1) etc.
Controlling a HEX display in ARM Assembler
.global _start
_start: LDR R0, =HEXBASE @ physical address of the HEX
MOV R2, #0b1011011 @ bitmask for pattern of a ’2’
MOV R3, R2, LSL #24 @ display: 2
MOV R2, #0x3f @ bitmask using hex-repres.
ORR R3, R2, LSL #16 @ display: 0
MOV R2, #91 @ bitmask using a decimal no.
MOV R2, #6 @ again usin a decimal value
STR R3, [R0]
.data
.equ HEXBASE, 0xFF200020
This code displays the number 2021 on the four rightmost displays of
the CPUlator
HEX Display Example
Exercise: Button read and HEX display
As next exercise we want to continuously read from a button

device
and display on the HEX display whether we detected a press
(“on”) or not (“off”)
This combines the previous 2 exercises
The core structure is a loop (either infinite or with a fixed counter)
A delay is used in the loop to control the frequency of checks
Exercise: Button read and HEX display (structure)
_start:
________________________ @ (1) construct bitmask

________________________ @ (1) construct bitmask
loop0: NOP
LDR R3, [R2] @ read value from BUT register
CMP R3, ___ @ (2) check against bitmask
______________________ @ (3) not pressed => OFF
nocount:
______________________ @ (3) pressed => ON
BL delay
B loop0 @ NB: infinite loop
We need to construct a bitmask for selecting the bit matching button

BUTTON_NO in (1).
Exercise: Button read and HEX display (structure)
_start:
MOV R5, #1
LSL R5, #BUTTON_NO @ (1) bitmask testing BUTTON_NO
loop0: NOP
CMP R3, ___ @ (2) check against bitmask
______________________ @ (3) not pressed => OFF
nocount:
______________________ @ (3) pressed => ON
BL delay
In (2) we need to check R3 against the bitmask constructed above.

In (3) we need to call sub-routines to show either “on” or “off”.
Exercise: Button read and HEX display
_start:
MOV R5, #1
LSL R5, #BUTTON_NO @ (1) bitmask for testing
BUTTON_NO
loop0: NOP
CMP R3, R5
BLNE showoff @ (3) not pressed => OFF
nocount:
BLEQ showon @ (3) pressed => ON
BL delay
Now we still need to define the sub-routines showon and showff.
Exercise: Button read and HEX display (show)
showon: PUSH {R4-R7, LR}

LDR R4, =HEXBASE
LDR R5, =char_o
LDRB R6, [R5]
LSL R6, #8
LDR R5, =char_n
LDRB R7, [R5]
ORR R6, R7
MOV R0, R6
STR R0, [R4]
POP {R4-R7, LR}
BX LR
This displays “on” on the HEX displays, without disrupting registers.
Exercise: Button read and HEX display (show)
showoff: PUSH {R4-R7, LR}

LDR R4, =HEXBASE
LDR R5, =char_o
LDRB R6, [R5]
LSL R6, #8
LDR R5, =char_f
LDRB R7, [R5]
ORR R6, R7
LSL R6, #8
LDR R5, =char_f
LDRB R7, [R5]
ORR R6, R7
MOV R0, R6
STR R0, [R4]
POP {R4-R7, LR}
BX LR
This displays “off” on the HEX displays, without disrupting registers.

Constants for the letters to display
Below we define the bit-patterns for the letters that we need.

.align 4
char_o: .byte 0b1011100 @ o
char_n: .byte 0b1010100 @ n
char_f: .byte 0b1110001 @ f
See the picture about the 7-segement display at the beginning of the
slide set to match bit-patterns to segments on the display.
Summary
Reading from a button, amounts to reading from a register and

selecting the right bit (1 bit)
Writing to a display, amounts to writing to a register and defining
the right bit pattern (7 bits)
All other input/output operations for external devices can be built
on top of this basic functionality
Details of registers and how to access them vary from chip to chip
For example, the Raspberry Pi 2 also uses an ARM core, but a
BCM2835 chip
The handling of LEDs and buttons is therefore different, and more
complex on the Raspberry Pi 2.
Optional: For those interested, complete tutorials for programming the
Raspberry Pi 2 are available on the course’s web page for 2019/20.
Summary
Reading from a button, amounts to reading from a register and

selecting the right bit (1 bit)
Writing to a display, amounts to writing to a register and defining
the right bit pattern (7 bits)
All other input/output operations for external devices can be built
on top of this basic functionality
Details of registers and how to access them vary from chip to chip
For example, the Raspberry Pi 2 also uses an ARM core, but a
BCM2835 chip
The handling of LEDs and buttons is therefore different, and more
complex on the Raspberry Pi 2.
Optional: For those interested, complete tutorials for programming the
Raspberry Pi 2 are available on the course’s web page for 2019/20.
Tutorial 5: Cache-friendly Programming
Hans-Wolfgang Loidl (Heriot-Watt Univ) F28HS Hardware-Software Interface Tutorial 5: Cache-friendly Programming 39

CPULATOR Functionalities

Uploaded by

Copyright:

CPULATOR Functionalities

Uploaded by

Document Information

Copyright

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

CPULATOR Functionalities

Uploaded by

Copyright:

F28HS Hardware-Software Interface:

School of Mathematical and Computer Sciences,

1 Tutorial 1: ARM Assembler Exercise: Caesar Cipher (ROT13)

2 Tutorial 2: Programming an LED

3 Tutorial 3: Programming a Button

4 Tutorial 4: Programming a HEX display

5 Tutorial 5: Cache-friendly Programming

Hans-Wolfgang Loidl (Heriot-Watt Univ) F28HS Hardware-Software Interface 2020/21 2 / 39

This sequence of screencasts is intended to give you a deeper

TASK: Implement a Caesar cipher (rotation cipher) in ARM

This tutorial will deal with programming an LED output device.

This tutorial will deal with programming an LED output device.

From the main chip we want to control an (external) device, here

Programmatically we need to go through these steps:

The relevant information can be found the hardware manual for

The picture (and discussion in the manual) tells us

In essence, this program performs one write operation: it writes the

We can achieve the same behaviour in ARM Assembler:

We now want to implement a blinking LED

start: LDR R2, =LEDBASE @ physical address of the LED

delay: PUSH {R4-R7,LR}

In essence, two nested loops that do nothing.

Controlling a simple external device means logically sending 1 bit

In this tutorial we want to use a button, as simulated on the

We need to read from address 0xFF200050 to determine the

This code continuously reads the button register

delay: PUSH {R4-R7,LR}

Reading input from a button works in the same way as writing to

The De1-SoC chip (by Altera) on the CPUlator, provides several

The address for the data register of the HEX displays is

The address for the data register of the HEX displays is

As next exercise we want to continuously read from a button

________________________ @ (1) construct bitmask

We need to construct a bitmask for selecting the bit matching button

In (2) we need to check R3 against the bitmask constructed above.

Now we still need to define the sub-routines showon and showff.

showon: PUSH {R4-R7, LR}

This displays “on” on the HEX displays, without disrupting registers.

showoff: PUSH {R4-R7, LR}

This displays “off” on the HEX displays, without disrupting registers.

Below we define the bit-patterns for the letters that we need.

Reading from a button, amounts to reading from a register and

Reading from a button, amounts to reading from a register and

You might also like