DE1-SoC Computer ARM
DE1-SoC Computer ARM
DE1-SoC Computer ARM
1 Introduction
This document describes a computer system that can be implemented on the Intel® DE1-SoC development and
education board. This system, called the DE1-SoC Computer, is intended for use in experiments on computer
organization and embedded systems.
The hard processor system (HPS), as shown in Figure 1, includes an ARM Cortex A9 dual-core processor. The A9
dual-core processor features two 32-bit CPUs and associated subsystems that are implemented as hardware circuits
in the Intel Cyclone V SoC chip. An overview of the ARM A9 processor can be found in the document Introduction
to the ARM Processor, which is provided in Intel’s FPGA University Program web site. All of the I/O peripherals in
the DE1-SoC Computer are accessible by the processor as memory mapped devices, using the address ranges that
are given in this document. A summary of the address map can be found in Section 7.
A good way to begin working with the DE1-SoC Computer and the ARM A9 processor is to make use of a utility
called the Intel® FPGA Monitor Program. It provides an easy way to assemble/compile ARM A9 programs written
in either assembly language or the C language. The Monitor Program, which can be downloaded from Intel’s web
site, is an application program that runs on the host computer connected to the DE1-SoC board. The Monitor
Program can be used to control the execution of code on the ARM A9, list (and edit) the contents of processor
registers, display/edit the contents of memory on the DE1-SoC board, and similar operations. The Monitor Program
includes the DE1-SoC Computer as a pre-designed system that can be downloaded onto the DE1-SoC board, as well
as several sample programs in assembly language and C that show how to use the DE1-SoC Computer’s peripherals.
Section 8 describes how the DE1-SoC Computer is integrated with the Monitor Program. An overview of the
Monitor Program is available in the document Intel® FPGA Monitor Program Tutorial, which is provided in the
University Program web site.
Host computer
... USB
DE1-SoC
Blaster Audio VGA TV
Peripherals Board CODEC DAC Decoder
ARM
Cortex A9 JTAG Nios II Audio Video-out Video-in PS/2
Ports
ports (2) port port port ports
MPCore
FPGA Bridges
Cyclone V Cyclone V
Timers Timers On-chip
HPS FPGA
memory
2.2 Memory
The HPS includes a memory port that connects the ARM MPCORE* to a 1 GB DDR3 memory. This memory is
normally used as the storage location of programs and data used by the ARM processors. The memory is organized
as 256M x 32-bits, and is accessible using word accesses (32 bits), halfwords, and bytes. The DDR3 memory is
mapped to the address space 0x00000000 to 0x3FFFFFFF. There is also a 64 KB on-chip memory available inside
each ARM A9 processor. This small memory is organized as 16K x 32-bits, and is mapped to the address space
0xFFFF0000 to 0xFFFFFFFF.
The HPS includes a general purpose I/O port, called GPIO1, that is accessible by the ARM A9 processor. As
illustrated in Figure 2, this parallel port is assigned the Base address 0xFF709000, and includes several 32-bit
registers. These registers can be read or written using word accesses. Only two bit locations in GPIO1 are used for
the DE1-SoC Computer. Bit 24 of the data register (DR) is connected to a green light, LEDG, and bit 25 is connected
to a pushbutton switch, KEY. To use these devices, the data direction register (DDR) shown in the figure has to be
configured such that bit 24 is an output and bit 25 is an input. Writing a 1 into a corresponding bit position in the
DDR sets this bit as an output, while writing a 0 sets the bit as an input. After the direction bits have been set, the
green light LEDG can be turned on/off by writing to bit 24 in the data register. The value of the pushbutton switch
KEY can be obtained by reading the external port register and checking the value of bit 25. An example program
for the ARM A9 processor that uses GPIO1 is given in Section 2.4.
As indicated in Figure 2, the GPIO1 port includes several other registers in addition to the DR and DDR registers.
These other registers are mostly used for setting characteristics of input pins, which affects only the KEY input in
our system. Detailed information about these registers can be found in the Intel Cyclone V Hard Processor System
documentation, which is available on Intel’s website.
Address 31 . . . 25 24 23 ... 0
The HPS includes several hardware timer modules that can be used to keep track of time intervals. The ARM A9
MPCore includes one private timer module for each A9 core, and the HPS provides four other timer modules that
can be used by either A9 core. The timers are described in more detail below.
Figure 3 shows the registers in the programmer’s interface for each A9 core private timer. These registers have the
base address 0xFFFEC600, as shown in the figure, and can be read or written using word accesses. To use the
timer, it is necessary to first write an initial count value into the Load register. The timer can then be started by
setting the enable bit E in the Control register to 1, and it can be stopped by setting E back to 0. Once enabled the
timer decrements its count value until reaching 0. When it reaches 0, the timer sets the F bit in the Interrupt status
register. The F bit can be checked by software using polled-I/O to determine when the timer period has expired. The
F bit can be reset to 0 by writing a 1 into it. Also, if bit I in the Control register is set to 1, then a processor interrupt
can be generated when the timer reaches 0. Using interrupts with the timer is discussed in Section 3.
When it reaches 0, the timer will stop if the auto bit (A) in the control register is set to 0. But if bit A is set to 1, then
the timer will automatically reload the value in the Load register and continue decrementing. The current value of
the timer is available to software in the Counter register shown in Figure 3. The timer uses a clock frequency of 200
MHz. The Prescaler field in the Control register can be used to slow down the counting rate, as follows. The timer
decrements each Prescaler +1 clock cycle. Therefore, if Prescaler = 0, then the timer decrements every clock cycle,
if Prescaler = 1, the timer decrements every second clock cycle, and so on.
Figure 4 shows the registers in the programmer’s interface for one of the HPS timers. These registers have the base
address 0xFFC08000, as shown in the figure, and can be read or written using word accesses. To configure the
timer, it is necessary to ensure that it is stopped by setting the enable bit E in the Control register to 0. A starting
count value for the timer can then be written into the Load register. To instruct the timer to use the specified starting
count value, the M in the Control register should be set to 1, and the timer can be started by setting E = 1. The timer
counts down to 0, and then sets both bit F in the End-of-interrupt register and bit S in the Interrupt status register to
1. Software can poll the value of S to determine when the timer period has expired. The S bit, and the F bit can be
reset to 0 by reading the contents of the End-of-Interrupt register. Also, if bit I , the interrupt mask bit, in the Control
register is set to 0, then an interrupt can be generated when the timer reaches 0 (note that bit I in the ARM A9 private
timer shown in Figure 3 has the opposite polarity). The use of interrupts with the timer is discussed in Section 3.
The current value of the timer is available to software in the Counter register shown in Figure 4. The timer uses a
clock frequency of 100 MHz.
There are three other identical timers in the HPS, with the following base addresses: 0xFFC09000, 0xFFD00000,
and 0xFFD01000. The first of these timers uses a 100 MHz clock, and the last two timers use a 25 MHz clock.
We should mention that other timer modules also exist in the HPS. The ARM A9 MPCore has a global timer that is
shared by both A9 cores, as well as a watchdog timer for each processor. Also, the HPS has two additional watchdog
timers. Documentation about the global timer and watchdog timers is available in the ARM Cortex A9 MPCore
Technical Reference Manual, and in the Intel Cyclone V Hard Processor System Technical Reference Manual.
An example of ARM A9 assembly language code is included in the Appendix in Listing 1. The code configures the
private timer for the A9 core so that it produces one-second timeouts. An infinite loop is used to flash the green light
connected to GPIO1, discussed in Section 2.3. The light is turned on for one second, then off, and so on.
An example of C code is also included in Listing 2. This code performs the same actions as the assembly language
program in Listing 1—it flashes on/off the green light connected to GPIO1 at one-second intervals.
The source code files shown in Listings 2 and 1 are distributed as part of the Intel® FPGA Monitor Program. The
files can be found under the heading sample programs, and are identified by the name Timer Lights.
The FPGA bridges depicted in Figure 1 provide connections between the HPS and FPGA in the Cyclone V SoC
device. The bridges are enabled, or disabled, by using the Bridge reset register, which is illustrated in Figure 5 and
has the address 0xFFD0501C. Three distinct bridges exist, called HPS-to-FPGA, lightweight HPS-to-FPGA, and
FPGA-to-HPS. In the DE1-SoC Computer the first two of these bridges are used to connect the ARM A9 processor
to the FPGA. As indicated in Figure 5 the bridges are enabled/disabled by bits 0 − 2 of the Bridge reset register. To
use the memory-mapped peripherals in the FPGA, software running on the ARM A9 must enable the HPS-to-FPGA
and lightweight HPS-to-FPGA bridges by setting bits #0 and #1 of the Bridge reset register to 0. We should note that
if a user program is downloaded and run on the ARM A9 by using the Intel® FPGA Monitor Program, described in
Section 8, then these bridges are automatically enabled before the user program is started.
In addition to the components described above, the HPS also provides a number of other peripheral devices, such
as USB, Ethernet, and a 3-D accelerometer (G-sensor). The G-sensor is described in the tutorial Using the DE1-
SoC Accelerometer with ARM, available from Intel’s FPGA University Program website. Documentation about the
other devices connected to the HPS can be found in the Intel Cyclone V Hard Processor System Technical Reference
Manual, as well as in the DE1-SoC Board User Manual.
Address 31 ... 2 1 0
HPS2FPGA
Lightweight HPS2FPGA
FPGA2HPS
As shown in Figure 1 a number of components in the DE1-SoC Computer are implemented inside the FPGA in the
Cyclone® V SoC chip. Several of these components are described in this section, and the others are presented in
Section 4.
The Intel Nios II processor is a 32-bit CPU that can be implemented in an Intel FPGA device. Two versions of
the Nios II processor are available, designated economy (/e) and fast (/f). The DE1-SoC Computer includes two
instances of the Nios II/f version, configured with floating-point hardware support. Instructions for using the Nios
II processors in the DE1-SoC Computer are provided in a separate document, called DE1-SoC Computer with Nios
II.
The DE1-SoC Computer has an SDRAM port, as well as two memory modules implemented using the on-chip
memory inside the FPGA. These memories are described below.
2.8.1 SDRAM
An SDRAM Controller in the FPGA provides an interface to the 64 MB synchronous dynamic RAM (SDRAM) on
the DE1-SoC board, which is organized as 32M x 16 bits. It is accessible by the ARM A9 processor using word
(32-bit), halfword (16-bit), or byte operations, and is mapped to the address space 0xC0000000 to 0xC3FFFFFF.
The DE1-SoC Computer includes a 256 KB memory that is implemented inside the FPGA. This memory is organized
as 64K x 32 bits, and spans addresses in the range 0xC8000000 to 0xC803FFFF. The memory is used as a pixel
buffer for the video-out and video-in ports.
The DE1-SoC Computer includes an 8 KB memory implemented inside the FPGA that is used as a character buffer
for the video-out port, which is described in Section 4.2. The character buffer memory is organized as 8K x 8 bits,
and spans the address range 0xC9000000 to 0xC9001FFF.
There are several parallel ports implemented in the FPGA that support input, output, and bidirectional transfers of
data between the ARM A9 processor and I/O peripherals. As illustrated in Figure 6, each parallel port is assigned
a Base address and contains up to four 32-bit registers. Ports that have output capability include a writable Data
register, and ports with input capability have a readable Data register. Bidirectional parallel ports also include a
Direction register that has the same bit-width as the Data register. Each bit in the Data register can be configured
as an input by setting the corresponding bit in the Direction register to 0, or as an output by setting this bit position
to 1. The Direction register is assigned the address Base + 4.
Address 31 30 ... 4 3 2 1 0
Some of the parallel ports in the DE1-SoC Computer have registers at addresses Base + 8 and Base + C, as indicated
in Figure 6. These registers are discussed in Section 3.
The red lights LEDR9−0 on the DE1-SoC board are driven by an output parallel port, as illustrated in Figure 7. The
port contains a 10-bit Data register, which has the address 0xFF200000. This register can be written or read by the
processor using word accesses, and the upper bits not used in the registers are ignored.
Address
LEDR9 LEDR0
There are two parallel ports connected to the 7-segment displays on the DE1-SoC board, each of which comprises
a 32-bit write-only Data register. As indicated in Figure 8, the register at address 0xFF200020 drives digits HEX3
to HEX0, and the register at address 0xFF200030 drives digits HEX5 and HEX4. Data can be written into these
two registers, and read back, by using word operations. This data directly controls the segments of each display,
according to the bit locations given in Figure 8. The locations of segments 6 to 0 in each seven-segment display on
the DE1-SoC board is illustrated on the right side of the figure.
Address
4 2
HEX36-0 HEX26-0 HEX16-0 HEX06-0
3
Segments
The SW9−0 slider switches on the DE1-SoC board are connected to an input parallel port. As illustrated in Figure 9,
this port comprises a 10-bit read-only Data register, which is mapped to address 0xFF200040.
Address
SW9 SW0
The parallel port connected to the KEY3−0 pushbutton switches on the DE1-SoC board comprises three 4-bit regis-
ters, as shown in Figure 10. These registers have the base address 0xFF200050 and can be accessed using word
operations. The read-only Data register provides the values of the switches KEY3−0 . The other two registers shown
Address 31 30 ... 4 3 2 1 0
Unused Unused
The DE1-SoC Computer includes two bidirectional parallel ports that are connected to the JP1 and JP2 40-pin
headers on the DE1-SoC board. These parallel ports include the four 32-bit registers that were described previously
for Figure 6. The base address of the port for JP1 is 0xFF200060, and for JP2 is 0xFF200070. Figure 11 gives a
diagram of the 40-pin connectors on the DE1-SoC board, and shows how the respective parallel port Data register
bits, D 31−0 , are assigned to the pins on the connector. The figure shows that bit D 0 of the parallel port is assigned
to the pin at the top right corner of the connector, bit D 1 is assigned below this, and so on. Note that some of the
pins on the 40-pin header are not usable as input/output connections, and are therefore not used by the parallel ports.
Also, only 32 of the 36 data pins that appear on each connector can be used.
2.9.6 Using the Parallel Ports with Assembly Language Code and C Code
The DE1-SoC Computer provides a convenient platform for experimenting with ARM A9 assembly language code,
or C code. A simple example of such code is provided in the Appendix in Listings 3 and 4. Both programs perform
the same operations, and illustrate the use of parallel ports by using either assembly language or C code.
The code in the figures displays the values of the SW switches on the LED lights. A rotating pattern is displayed
on the LEDs. This pattern is rotated to the left by using an ARM A9 rotate instruction, and a delay loop is used to
make the shifting slow enough to observe. The pattern can be changed to the values of the SW switches by pressing
a pushbutton KEY. When a pushbutton key is pressed, the program waits in a loop until the key is released.
The source code files shown in Listings 3 and 4 are distributed as part of the Intel® FPGA Monitor Program. The
files can be found under the heading sample programs, and are identified by the name Getting Started.
The JTAG* port implements a communication link between the DE1-SoC board and its host computer. This link can
be used by the Intel Quartus® Prime software to transfer FPGA programming files into the DE1-SoC board, and by
the Intel® FPGA Monitor Program, discussed in Section 8. The JTAG port also includes a UART, which can be used
JP1 JP2
Pin 1 D0 Pin 1 D0
D1 D1
D2 D3 D2 D3
D4 D5 D4 D5
D6 D7 D6 D7
Unused Unused
D8 D9 D8 D9
D10 D11 D10 D11
D12 D13 D12 D13
D14 D14
D15 D15
D16 D17 D16 D17
D18 D19 D18 D19
D20 D21 D20 D21
Unused Unused
D22 D23 D22 D23
D24 D25 D24 D25
D26 D27 D26 D27
D28 D29 D28 D29
D30 D31 Pin 40 D30 D31 Pin 40
Figure 11. Assignment of parallel port bits to pins on JP1 and JP2.
to transfer character data between the host computer and programs that are executing on the ARM A9 processor. If
the Intel® FPGA Monitor Program is used on the host computer, then this character data is sent and received through
its Terminal Window. The programming interface of the JTAG UART consists of two 32-bit registers, as shown
in Figure 12. The register mapped to address 0xFF201000 is called the Data register and the register mapped to
address 0xFF201004 is called the Control register.
When character data from the host computer is received by the JTAG UART it is stored in a 64-character FIFO.
The number of characters currently stored in this FIFO is indicated in the field RAVAIL, which are bits 31−16 of the
Data register. If the receive FIFO overflows, then additional data is lost. When data is present in the receive FIFO,
then the value of RAVAIL will be greater than 0 and the value of bit 15, RVALID, will be 1. Reading the character
at the head of the FIFO, which is provided in bits 7 − 0, decrements the value of RAVAIL by one and returns this
decremented value as part of the read operation. If no data is present in the receive FIFO, then RVALID will be set
to 0 and the data in bits 7 − 0 is undefined.
The JTAG UART also includes a 64-character FIFO that stores data waiting to be transmitted to the host computer.
Character data is loaded into this FIFO by performing a write to bits 7−0 of the Data register in Figure 12. Note
that writing into this register has no effect on received data. The amount of space, WSPACE, currently available in
the transmit FIFO is provided in bits 31−16 of the Control register. If the transmit FIFO is full, then any characters
written to the Data register will be lost.
Bit 10 in the Control register, called AC, has the value 1 if the JTAG UART has been accessed by the host computer.
This bit can be used to check if a working connection to the host computer has been established. The AC bit can be
cleared to 0 by writing a 1 into it.
The Control register bits RE, WE, RI, and WI are described in Section 3.
2.10.1 Using the JTAG* UART with Assembly Language Code and C Code
Listings 5 and 6 give simple examples of assembly language and C code, respectively, that use the JTAG UART.
Both versions of the code perform the same function, which is to first send an ASCII string to the JTAG UART, and
then enter an endless loop. In the loop, the code reads character data that has been received by the JTAG UART, and
echoes this data back to the UART for transmission. If the program is executed by using the Intel® FPGA Monitor
Program, then any keyboard character that is typed into the Terminal Window of the Monitor Program will be echoed
back, causing the character to appear in the Terminal Window.
The source code files shown in Listings 5 and 6 are made available as part of the Intel® FPGA Monitor Program.
The files can be found under the heading sample programs, and are identified by the name JTAG UART.
The DE1-SoC Computer includes a second JTAG UART that is accessible by the ARM A9 MPCORE. This second
UART is mapped to the base address 0xFF201008, and operates as described above. The reason that two JTAG
UARTs are provided is to allow each processor in the ARM A9 MPCORE to have access to a separate UART.
The DE1-SoC Computer includes a timer module implemented in the FPGA that can be used by the ARM A9 pro-
cessor. This timer can be loaded with a preset value, and then counts down to zero using a 100-MHz clock. The
programming interface for the timer includes six 16-bit registers, as illustrated in Figure 13. The 16-bit register at
address 0xFF202000 provides status information about the timer, and the register at address 0xFF202004 allows
control settings to be made. The bit fields in these registers are described below:
• TO provides a timeout signal which is set to 1 by the timer when it has reached a count value of zero. The TO
bit can be reset by writing a 0 into it.
• RUN is set to 1 by the timer whenever it is currently counting. Write operations to the status halfword do not
affect the value of the RUN bit.
Address 31 . . . 17 16 15 ... 3 2 1 0
• CONT affects the continuous operation of the timer. When the timer reaches a count value of zero it auto-
matically reloads the specified starting count value. If CONT is set to 1, then the timer will continue counting
down automatically. But if CONT = 0, then the timer will stop after it has reached a count value of 0.
• (START/STOP) is used to commence/suspend the operation of the timer by writing a 1 into the respective bit.
The two 16-bit registers at addresses 0xFF202008 and 0xFF20200C allow the period of the timer to be changed
by setting the starting count value. The default setting provided in the DE1-SoC Computer gives a timer period
of 125 msec. To achieve this period, the starting value of the count is 100 MHz × 125 msec = 12.5 × 106 . It is
possible to capture a snapshot of the counter value at any time by performing a write to address 0xFF202010. This
write operation causes the current 32-bit counter value to be stored into the two 16-bit timer registers at addresses
0xFF202010 and 0xFF202014. These registers can then be read to obtain the count value.
A second interval timer, which has an identical interface to the one described above, is also available in the FPGA,
starting at the base address 0xFF202020.
2.12 System ID
The system ID module provides a unique value that identifies the DE1-SoC Computer system. The host computer
connected to the DE1-SoC board can query the system ID module by performing a read operation through the
JTAG port. The host computer can then check the value of the returned identifier to confirm that the DE1-SoC
Computer has been properly downloaded onto the DE1-SoC board. This process allows debugging tools on the
host computer, such as the Intel® FPGA Monitor Program, to verify that the DE1-SoC board contains the required
computer system before attempting to execute code that has been compiled for this system.
The IRQ exception allows I/O peripherals to generate interrupts for the A9 processor. All interrupt signals from the
peripherals are connected to a module in the processor called the generic interrupt controller (GIC). The GIC allows
individual interrupts for each peripheral to be either enabled or disabled. When an enabled interrupt happens, the GIC
causes an IRQ exception in the A9 processor. Since the same vector table entry is used for all interrupts, the software
for the interrupt service routine must determine the source of the interrupt by querying the GIC. Each peripheral is
identified in the GIC by an interrupt identification (ID) number. Table 1 gives the assignment of interrupt IDs for
each of the I/O peripherals in the DE1-SoC Computer. The rest of this section describes the interrupt behavior
associated with the timers and parallel ports.
Figure 3, in Section 2.4.1, shows four registers that are associated with the A9 private timer. As we said in Section
2.4.1, bit F in the Interrupt status register is set to 1 when the timer reaches a count value of 0. It is possible to
generate an A9 interrupt when this occurs, by using bit I of the Control register. Setting bit I to 1 causes the timer to
send an interrupt signal to the GIC whenever the timer reaches a count value of 0. The F bit can be cleared to 0 by
writing writing a 1 into the Interrupt status register.
Figure 4, in Section 2.4.2, shows five registers that are associated with each HPS timer. As we said in Section 2.4.2,
when the timer reaches a count value of zero, bit F in the End-of-Interrupt register is set to 1. The value of the F bit
is also reflected in the S bit in the Interrupt status register. It is possible to generate an A9 interrupt when the F bit
becomes 1, by using the I bit of the Control register. Setting bit I to 0 unmasks the interrupt signal, and causes the
timer to send an interrupt signal to the GIC whenever the F bit is 1. After an interrupt occurs, it can be cleared by
reading the End-of-Interrupt register.
Figure 13, in Section 2.11, shows six registers that are associated with the interval timer. As we said in Section 2.11,
the TO bit in the Status register is set to 1 when the timer reaches a count value of 0. It is possible to generate an
interrupt when this occurs, by using the ITO bit in the Control register. Setting the ITO bit to 1 causes an interrupt
request to be sent to the GIC whenever TO becomes 1. After an interrupt occurs, it can be cleared by writing any
value into the Status register.
Parallel ports implemented in the FPGA in the DE1-SoC Computer were illustrated in Figure 6, which is reproduced
as Figure 14. As the figure shows, parallel ports that support interrupts include two related registers at the addresses
Base + 8 and Base + C. The Interruptmask register, which has the address Base + 8, specifies whether or not an
interrupt signal should be sent to the GIC when the data present at an input port changes value. Setting a bit location
in this register to 1 allows interrupts to be generated, while setting the bit to 0 prevents interrupts. Finally, the
parallel port may contain an Edgecapture register at address Base + C. Each bit in this register has the value 1 if the
corresponding bit location in the parallel port has changed its value from 0 to 1 since it was last read. Performing a
write operation to the Edgecapture register sets all bits in the register to 0, and clears any associated interrupts.
Figure 10, reproduced as Figure 15, shows the registers associated with the pushbutton parallel port. The Interrupt-
mask register allows interrupts to be generated when a key is pressed. Each bit in the Edgecapture register is set to
1 by the parallel port when the corresponding key is pressed. An interrupt service routine can read this register to
determine which key has been pressed. Writing any value to the Edgecapture register deasserts the interrupt signal
being sent to the GIC and sets all bits of the Edgecapture register to zero.
Address 31 30 ... 4 3 2 1 0
Figure 14. Registers used for interrupts from the parallel ports.
Address 31 30 ... 4 3 2 1 0
Unused Unused
Figure 15. Registers used for interrupts from the pushbutton parallel port.
Figure 12, reproduced as Figure 16, shows the data and Control registers of the JTAG UART. As we said in Section
2.10, RAVAIL in the Data register gives the number of characters that are stored in the receive FIFO, and WSPACE
gives the amount of unused space that is available in the transmit FIFO. The RE and WE bits in Figure 16 are used to
enable processor interrupts associated with the receive and transmit FIFOs. When enabled, interrupts are generated
when RAVAIL for the receive FIFO, or WSPACE for the transmit FIFO, exceeds 7. Pending interrupts are indicated
in the Control register’s RI and WI bits, and can be cleared by writing or reading data to/from the JTAG UART.
An example of assembly language code for the DE1-SoC Computer that uses interrupts is shown in Listing 7, which
has three main parts. The beginning part of the code sets up the exception vector table. This code must be in a special
assembler section called .section, as shown. The entries in the table provide branches to the various exception service
routines; they are discussed later in this section.
When this code is executed on the DE1-SoC board it displays a rotating pattern on the LEDs. The pattern’s rotation
can be toggled through pressing the pushbutton KEYs. Different types of interrupts are used in the code. The LEDs
are controlled by interrupts from the FPGA interval timer, and the KEYs are also handled through interrupts.
The main program initializes the A9 banked stack pointer (sp) registers for interrupt (IRQ) mode and supervisor
(SVC) mode, because these are the processor modes that are used in the program. The code then calls subroutines
to initialize the HPS timer, FPGA interval timer, and FPGA pushbutton KEYs. Finally, the code initializes the HPS
GPIO1 port, enables IRQ interrupts in the A9 processor, and then enters an infinite loop. The loop code turns on and
off a green light whenever the global variable named tick is set to 1. This variable is set to 1 by the exception service
routine for the HPS timer, which is described later in this section.
Following are the subroutines used to initialize the timers and pushbutton KEYs. The CONFIG_HPS_TIMER
routine sets up the HPS timer 0 so that it will produce an interrupt every one second. Since this timer uses a 100
MHz clock, the timer load register is initialized to the value 100 × 106 . The CONFIG_INTERVAL_TIMER routine
configures the FPGA interval timer to produce interrupts every 50 msec. Since this timer uses a 100 MHz clock,
the required starting count value is 5 × 106 . The CONFIG_KEYS routine sets up the FPGA KEYs parallel port to
produce an interrupt when any KEY is pressed.
The last portion of the code shows the global data used by the program. It includes the tick variable that was discussed
for the code earlier, and other variables. The pattern variable holds the bit-pattern that is written, the key_pressed
variable indicates which FPGA KEY has been recently pressed, and the shift_dir variable specifies the direction of
shifting for the HEX displays.
Also included in part c of Listing 7 is the subroutine that initializes the GIC. This code performs the minimum-
required steps needed to configure the three interrupts used in the program, by writing to the processor targets
(ICDIPTRn) registers in the GIC, and the set enable (ICDISERn) registers. For the HPS timer, the registers used
have addresses 0xFFFED8C4 and 0xFFFED118, as shown in the listing. For the FPGA interval timer and KEYs,
the register addresses are 0xFFFED848 and 0xFFFED108. Instructions for calculating these addresses, and de-
termining the bit patterns to write into them can be found in the tutorial Using the Generic Interrupt Controller,
available in Intel’s FPGA University Program website. The last part of the code in this section enables the CPU
Interface and Distributor in the GIC.
The exception service routines for the main program in Listing 7 are given in Listing 8. The first part of the listing
gives the IRQ exception handler. This routine first reads from the interrupt acknowledge register in the GIC to
determine the interrupt ID of the peripheral that caused the interrupt. The code then checks which of the three
possible sources of interrupt has occurred, and calls the corresponding interrupt service routine for the HPS timer,
FPGA interval timer, or FPGA KEY parallel port. These interrupt service routine are shown in Listings 9 to 10.
Finally, the exception handler in Listing 8 writes to the end-of-interrupt register in the GIC to clear the interrupt, and
then returns to the main program by using the instruction “SUBS PC, LR, #4”.
The latter part of Listing 8 shows handlers for exceptions that correspond to the reset exception, various types
of error conditions, and the FIQ interrupt. The reset handler shows a branch to the start of the main program in
Listing 7. This handler is just an indicator of the result of performing a reset of the A9 processor—the actual reset
process involves executing code from a special boot ROM on the processor, and then executing a program called
the pre-loader before actually starting the main program. More information about the reset process for the A9
processor can be found in the document “Using the Pre-loader Software for the A9 Processor,” which is available
from Intel’s FPGA University Program website. The other handlers in the latter part of Listing 8, which are just
loops that branch to themselves, are intended to serve as placeholders for code that would handle the corresponding
exceptions. More information about each of these types of exceptions can be found in the document Introduction to
the ARM Processor, also available in Intel’s FPGA University Program web site.
An example of C code for the DE1-SoC Computer that uses interrupts is shown in Figure 12. This code performs
exactly the same operations as the code described in Listing 7.
Before it call subroutines to configure the generic interrupt controller (GIC), timers, and pushbutton KEY port, the
main program first initializes the IRQ mode stack pointer by calling the routine set_A9_IRQ_stack(). The code for
this routine uses in-line assembly language instructions, as shown in Part b of the listing. This step is necessary
because the C compiler generates code to set only the supervisor mode stack, which is used for running the main
program, but the compiler does not produce code for setting the IRQ mode stack. To enable IRQ interrupts in the A9
processor the main program uses the in-line assembly code shown in the subroutine called enable_A9_interrupts().
The exception handlers for the main program in Listing 12 are given in Listing 13. These routines have unique names
that are meaningful to the C compiler and linker tools, and they are declared with the special type of __attribute__
called interrupt. These mechanisms cause the C compiler and linker to use the addresses of these routines as the
contents of the exception vector table.
The function with the name __cs3_isr_irq is the IRQ exception handler. As discussed for the assembly language
code in Listing 8 this routine first reads from the interrupt acknowledge register in the GIC to determine the interrupt
ID of the peripheral that caused the interrupt, and then calls the corresponding interrupt service routine for either
the HPS timer, FPGA interval timer, or FPGA KEY parallel port. These interrupt service routines are shown in
Listings 14 to 15.
Listing 13 also shows handlers for exceptions that correspond to the various types of error conditions and the FIQ
interrupt. These handlers are just loops that are meant to serve as place-holders for code that would handle the
corresponding exceptions.
The source code files shown in Listing 7 to Listing 16 are distributed as part of the Intel® FPGA Monitor Program.
The files can be found under the heading sample programs, and are identified by the name Interrupt Example.
4 Media Components
This section describes the audio in/out, video-out, video-in, PS/2, IrDA*, and ADC ports.
The DE1-SoC Computer includes an audio port that is connected to the audio CODEC (COder/DECoder) chip on
the DE1-SoC board. The default setting for the sample rate provided by the audio CODEC is 8K samples/sec. The
audio port provides audio-input capability via the microphone jack on the DE1-SoC board, as well as audio output
functionality via the line-out jack. The audio port includes four FIFOs that are used to hold incoming and outgoing
data. Incoming data is stored in the left- and right-channel Read FIFOs, and outgoing data is held in the left- and
right-channel Write FIFOs. All FIFOs have a maximum depth of 128 32-bit words.
The audio port’s programming interface consists of four 32-bit registers, as illustrated in Figure 17. The Control
register, which has the address 0xFF203040, is readable to provide status information and writable to make control
settings. Bit RE of this register provides an interrupt enable capability for incoming data. Setting this bit to 1 allows
the audio core to generate a ARM A9 interrupt when either of the Read FIFOs are filled 75% or more. The bit RI
will then be set to 1 to indicate that the interrupt is pending. The interrupt can be cleared by removing data from the
Read FIFOs until both are less than 75% full. Bit WE gives an interrupt enable capability for outgoing data. Setting
this bit to 1 allows the audio core to generate an interrupt when either of the Write FIFOs are less that 25% full. The
bit WI will be set to 1 to indicate that the interrupt is pending, and it can be cleared by filling the Write FIFOs until
both are more than 25% full. The bits CR and CW in Figure 17 can be set to 1 to clear the Read and Write FIFOs,
respectively. The clear function remains active until the corresponding bit is set back to 0.
Address 31 . . . 24 23 . . . 16 15 . . . 10 9 8 7 ...4 3 2 1 0
0xFF203040 Unused WI RI CW CR WE RE Control
The read-only Fifospace register in Figure 17 contains four 8-bit fields. The fields RARC and RALC give the number
of words currently stored in the right and left audio-input FIFOs, respectively. The fields WSRC and WSLC give the
number of words currently available (that is, unused) for storing data in the right and left audio-out FIFOs. When all
FIFOs in the audio port are cleared, the values provided in the Fifospace register are RARC = RALC = 0 and WSRC
= WSLC = 128.
The Leftdata and Rightdata registers are readable for audio in, and writable for audio out. When data is read from
these registers, it is provided from the head of the Read FIFOs, and when data is written into these registers it is
loaded into the Write FIFOs.
A fragment of C code that uses the audio port is shown in Listing 17. The code checks to see when the depth of
either the left or right Read FIFO has exceeded 75% full, and then moves the data from these FIFOs into a memory
buffer. This code is part of a program that is distributed as part of the Intel® FPGA Monitor Program. The source
code can be found under the heading sample programs, and is identified by the name Audio.
The DE1-SoC Computer includes a video-out port connected to the on-board VGA controller that can be connected
to a standard VGA monitor. The video-out port support a screen resolution of 640 × 480. The image that is displayed
by the video-out port is derived from two sources: a pixel buffer, and a character buffer.
The pixel buffer for the video-out port holds the data (color) for each pixel that will be displayed. As illustrated in
Figure 18, the pixel buffer provides an image resolution of 320 × 240 pixels, with the coordinate 0,0 being at the
top-left corner of the image. Since the video-out port supports the screen resolution of 640 × 480, each of the pixel
values in the pixel buffer is replicated in both the x and y dimensions when it is being displayed on the screen.
0 1 2 3 ... 319
0
1
2
...
..
.
.. ..
. .
...
239
Figure 19a shows that each pixel color is represented as a 16-bit halfword, with five bits for the blue and red
components, and six bits for green. As depicted in part b of Figure 19, pixels are addressed in the pixel buffer by
using the combination of a base address and an x,y offset. In the DE1-SoC Computer the default address of the pixel
buffer is 0xC8000000, which corresponds to the starting address of the FPGA on-chip memory. Using this scheme,
the pixel at location 0,0 has the address 0xC8000000, the pixel 1,0 has the address base + (00000000 000000001 0)2
= 0xC8000002, the pixel 0,1 has the address base + (00000001 000000000 0)2 = 0xC8000400, and the pixel at
location 319,239 has the address base + (11101111 100111111 0)2 = 0xC803BE7E.
You can create an image by writing color values into the pixel addresses as described above. A dedicated pixel buffer
controller continuously reads this pixel data from sequential addresses in the corresponding memory for display on
15 11 10 5 4 0
red green blue
the screen. You can modify the pixel data at any time, simply by writing to the pixel addresses. Thus, an image can
be changed even when it is in the process of being displayed. However, it is also possible to avoid making changes
to the pixel buffer while it is being displayed, by using the concept of double-buffering. In this scheme, two pixel
buffers are involved, called the front and back buffers, described below.
The DE1-SoC Computer contains an RGB Resampler for converting video streams between RGB color formats.
Reading from the 32-bit Status register at address 0xFF203010 provides information about alpha/no alpha, col-
or/grayscale, and mode for the incoming and outgoing formats. The incoming format for the DE1-SoC Com-
puter video stream is 0x14, which corresponds to no alpha, color, 16-bit RGB (5-bit Red, 6-bit Green, 5-bit Blue).
For more information, the reader should refer to the video module’s online documentation, Video IP Cores for Intel
DE-Series Boards, which is available from Intel’s FPGA University Program web site.
As mentioned above, a pixel buffer controller reads data out of the pixel buffer so that it can be displayed on the
screen. This pixel buffer controller includes a programming interface in the form of a set of registers, as illustrated in
Table 2. The register at address 0xFF203020 is called the Buffer register, and the register at address 0xFF203024
is the Backbuffer register. Each of these registers stores the starting address of a pixel buffer. The Buffer register
holds the address of the pixel buffer that is displayed on the screen. As mentioned above, in the default configuration
of the DE1-SoC Computer this Buffer register is set to the address 0xC8000000, which points to the start of the
FPGA on-chip memory. The default value of the Backbuffer register is also 0xC8000000, which means that there
is only one pixel buffer. But software can modify the address stored in the Backbuffer register, thereby creating a
second pixel buffer. The pixel buffer can be located in the SDRAM memory in the DE1-SoC Computer, which has
the base address 0xC0000000. Note that the pixel buffer cannot be located in the DDR3 memory in the DE1-SoC
Computer, because the pixel buffer controller is not connected to the DDR3 memory. An image can be drawn into
the second buffer by writing to its pixel addresses. This image is not displayed on the screen until a pixel buffer
swap is performed, as explained below.
A pixel buffer swap is caused by writing the value 1 to the Buffer register. This write operation does not directly
modify the content of the Buffer register, but instead causes the contents of the Buffer and Backbuffer registers
to be swapped. The swap operation does not happen right away; it occurs at the end of a screen-drawing cycle,
after the last pixel in the bottom-right corner has been displayed. This time instance is referred to as the vertical
synchronization time, and occurs every 1/60 seconds. Software can poll the value of the S bit in the Status register,
at address 0xFF20302C, to see when the vertical synchronization has happened. Writing the value 1 into the Buffer
register causes S to be set to 1. Then, when the swap of the Buffer and Backbuffer registers has been completed S is
reset back to 0.
Notes:
(1) Reserved. Read values are undefined. Write zero.
In a typical application the pixel buffer controller is used as follows. While the image contained in the pixel buffer
that is pointed to by the Buffer register is being displayed, a new image is drawn into the pixel buffer pointed to by
the Backbuffer register. When this new image is ready to be displayed, a pixel buffer swap is performed. Then, the
pixel buffer that is now pointed to by the Backbuffer register, which was already displayed, is cleared and the next
new image is drawn. In this way, the next image to be displayed is always drawn in the “back” pixel buffer, and the
two pixel buffer pointers are swapped when the new image is ready to be displayed. Each time a swap is performed
software has to synchronize with the video-out port by waiting until the S bit in the Status register becomes 0.
As shown in Table 2 the Status register contains additional information other than the S bit. The fields n and m give
the number of address bits used for the X and Y pixel coordinates, respectively. The B S field specifies the number of
data bits per symbol minus one. The SB field specifies the number of symbols per beat minus one. The A field allows
the selection of two different ways of forming pixel addresses. If configured with A = 0, then the pixel controller
expects addresses to contain X and Y fields, as we have used in this section. But if A = 1, then the controller expects
addresses to be consecutive values starting from 0 and ending at the total number of pixels−1. The E N field is used
to enable or disable the DMA controller. If this bit is set to 0, the DMA controller will be turned off.
In Table 2 the default values of the status register fields in the DE1-SoC Computer are used when forming pixel
addresses. The defaults are n = 9, m = 8, and A = 0. If the pixel buffer controller is changed to provide different
values of these fields, then the way in which pixel addresses are formed has to be modified accordingly. The
programming interface also includes a Resolution register, shown in Table 2, that contains the X and Y resolution of
the pixel buffer(s).
The character buffer for the video-out port is stored in on-chip memory in the FPGA on the DE1-SoC board. As
illustrated in Figure 20a , the buffer provides a resolution of 80 × 60 characters, where each character occupies an
8 × 8 block of pixels on the screen. Characters are stored in each of the locations shown in Figure 20a using their
ASCII codes; when these character codes are displayed on the monitor, the character buffer automatically generates
the corresponding pattern of pixels for each character using a built-in font. Part b of Figure 20 shows that characters
are addressed in the memory by using the combination of a base address, which has the value 0xC9000000, and an
x,y offset. Using this scheme, the character at location 0,0 has the address 0xC9000000, the character 1,0 has the
address base + (000000 0000001)2 = 0xC9000001, the character 0,1 has the address base + (000001 0000000)2 =
0xC9000080, and the character at location 79,59 has the address base + (111011 1001111)2 = 0xC9001DCF.
0 1 2 3 ... 79
0
1
2
...
...
...
...
59 ...
31 ... 13 12 . . . 7 6 ... 0
1100100100000000000 y x
A fragment of C code that uses the pixel and character buffers is shown in Listing 18. The first for loop in the figure
draws a rectangle in the pixel buffer using the color pixel_color. The rectangle is drawn using the coordinates x 1 , y 1
and x 2 , y 2 . The second while loop in the figure writes a null-terminated character string pointed to by the variable
text_ptr into the character buffer at the coordinates x, y. The code in Listing 18 is included in the sample program
called Video that is distributed with the Intel® FPGA Monitor Program.
The DE1-SoC Computer includes a video-in port for use with the composite video-in connector on the DE1-
SoC board. The video analog-to-digital converter (ADC) connected to this port is configured to support an NTSC
video source. The video-in port provides frames of video at a resolution of 320 x 240 pixels. These video frames
can be displayed on a monitor by using the video-out port described in Section 4.2. The video-in port writes each
frame of the video-in data into the pixel buffer described in Section 4.2.1. The video-in port can be configured to
provide two types of images: either the “raw” image provided by the video ADC, or a version of this image in which
only “edges” that are detected in the image are drawn.
The video-in port has a programming interface that consists of two registers, as illustrated in Figure 21. The Control
register at the address 0xFF20306C is used to enable or disable the video input. If the EN bit in this register is set
to 0, then the video-in core does not store any data into the pixel buffer. Setting EN to 1 and then changing EN to 0
can be used to capture a still picture from the video-in port.
The register at address 0xFF203070 is used to enable or disable edge detection. Setting the E bit in this register
to 1 causes the input video to passed through hardware circuits that detect edges in the images. The image stored
in the pixel buffer will then consist of dark areas that are puctuated by lighter lines along the edges that have been
detected. Setting E = 0 causes a normal image to be stored into the pixel buffer.
Address
2 1 0
0xFF20306C Unused EN Control
0
0xFF203070 Unused E Edge-detection
The data provided by the Video-In core is stored into memory using a DMA Controller for Video. When operating
in Stream to Memory mode, the DMA stores the incoming frames to memory. Table 3 describes the registers used
in the DMA Controller.
Notes:
(1) Reserved. Read values are undefined. Write zero.
The incoming video is stored to memory, starting at the address specified in the Buffer register. The BackBuffer
register is used to store an alternate memory location. To change where the video is stored, the new location should
first be written into the BackBuffer. Then the value in the BackBuffer and Buffer registers can be switched by
performing a write to the Buffer register.
Bit 2 of the Status/Control register, EN, is used to enable or disable the Video DMA controller. In the DE1-SoC
Computer, the DMA controller is disabled by default. To enable the DMA controller, write a 1 into this location.
The Video DMA Controller will then begin storing the video into the location specified in the Buffer register.
The default value stored in the Buffer register is 0x08000000. This address is also used as the source for the Video-
Out port, as described in Section 4.2, allowing the Video In stream to be displayed on the VGA. If the Video-Out is
intended to display a different signal, than the address stored in the Video DMA Controller’s Buffer register should
be changed.
The audio/video configuration module controls settings that affect the operation of both the audio port and the video-
out port. The audio/video configuration module automatically configures and initializes both of these ports whenever
the DE1-SoC Computer is reset. For typical use of the DE1-SoC Computer it is not necessary to modify any of these
default settings. In the case that changes to these settings are needed, the reader should refer to the audio/video
configuration module’s online documentation, which is available from Intel’s FPGA University Program web site.
The DE1-SoC Computer includes two PS/2 ports that can be connected to a standard PS/2 keyboard or mouse. The
port includes a 256-byte FIFO that stores data received from a PS/2 device. The programming interface for the PS/2
port consists of two registers, as illustrated in Figure 22. The PS2_Data register is both readable and writable. When
bit 15, RVALID, is 1, reading from this register provides the data at the head of the FIFO in the Data field, and
the number of entries in the FIFO (including this read) in the RAVAIL field. When RVALID is 1, reading from the
PS2_Data register decrements this field by 1. Writing to the PS2_Data register can be used to send a command in
the Data field to the PS/2 device.
The PS2_Control register can be used to enable interrupts from the PS/2 port by setting the RE field to the value 1.
When this field is set, then the PS/2 port generates an interrupt when RAVAIL > 0. While the interrupt is pending
the field RI will be set to 1, and it can be cleared by emptying the PS/2 port FIFO. The CE field in the PS2_Control
register is used to indicate that an error occurred when sending a command to a PS/2 device.
0xFF200104 CE RI RE PS2_Control
A fragment of C code that uses the PS/2 port is given in Listing 19. This code reads the content of the Data register,
and saves data when it is available. If the code is used continually in a loop, then it stores the last three bytes of
data received from the PS/2 port in the variables byte1 , byte2 , and byte3 . This code is included as part of a sample
program called PS2 that is distributed with the Intel® FPGA Monitor Program.
The DE1-SoC Computer includes a second PS/2 port that allows both a keyboard and mouse to be used at the same
time. To use the dual port a Y-splitter cable must be used and the keyboard and mouse must be connected to the
PS/2 connector on the DE1-SoC board through this cable. The PS/2 port dual has the same registers as the PS/2 port
shown in Listing 19, except that the base address of its PS2_Data register is 0xFF200108 and the base address of
its PS2_Control register is 0xFF20010C.
The DE1-SoC Computer includes an IrDA UART for communicating wirelessly with peripherals over the infrared
spectrum. It is configured for 8-bit data and one stop bit, and operates at a baud rate of 155,200. The default
configuration does not use a parity bit. The programming interface consists of two 32-bit registers, as shown in
Figure 23. The register at address 0xFF201020 is referred to as the Data register, and the register at address
0xFF201024 is the Control register.
The operation of the IrDA UART is similar to the Serial Port UART described above. Data recieved through the IrDA
is stored in a 128-character FIFO in the UART. As shown in Figure 23, the number of characters, RAVAIL, currently
stored in this FIFO is provided in bits 23−16 of the Data register. If the FIFO overflows, then any additional data is
lost. When a read of the Data register is performed, the character at the head of the FIFO is provided in bits 7−0.
If the character read is valid, the the value of bit 15, RVALID will be one. If no data is available to be read from the
receive FIFO, then RVALID will be set to 0 and the data in bits 7 − 0 is undefined.
Address 31 . . . 24 23 . . . 16 15 14 . . . 10 9 8 7 ... 1 0
The Analog-to-Digital Conversion (ADC) Port provides access to the eight-channel, 12-bit analog-to-digital con-
verter on the DE1-SoC board. As illustrated in Figure 24, the ADC port comprises eight 12-bit registers starting at
the base address 0xFF204000. The first two registers have dual purposes, acting as both data and control registers.
By default, the ADC port updates the A-to-D conversion results for all ports only when instructed to do so. Writing
to the control register at address 0xFF204000 causes this update to occur. Reading from the register at address
0xFF204000 provides the conversion data for channel 0. Reading from the other seven registers provides the con-
version data for the corresponding channels. It is also possible to have the ADC port continually request A-to-D
conversion data for all channels. This is done by writing the value 1 to the control register at address 0xFF204004.
The R bit of each channel register in Figure 24 is used in Auto-update mode. R is set to 1 when its corresponding
channel is refreshed and set to 0 when the channel is read.
Address 31 ... 16 15 14 12 11 ... 0
Figure 25 shows the connector on the DE1-SoC board that is used with the ADC port. Analog signals in the range
of 0 V to the VCC 5 power-supply voltage can be connected to the pins for channels 0 to 7.
Gnd Ch7
Ch6 Ch5
Ch4 Ch3
Ch2 Ch1
Ch0 Vcc5
JP15
Table 4 lists the names of the Qsys IP cores that are used in this system. When the DE1-SoC Computer design files
are opened in the Quartus Prime software, these cores can be examined using the Qsys System Integration tool. Each
core has a number of settings that are selectable in the Qsys System Integration tool, and includes a datasheet that
provides detailed documentation.
1. Install the University Program IP Cores from Intel’s FPGA University Program web site
2. Copy the design source files for the DE1-SoC Computer from the University Program web site. These files
can be found in the Design Examples section of the web site
4. Open the Qsys System Integration tool in the Quartus Prime software, and modify the system as desired
5. Generate the modified system by using the Qsys System Integration tool
6. It may be necessary to modify the Verilog or VHDL code in the top-level module, DE1-SoC_Computer.v/vhd,
if any I/O peripherals have been added or removed from the system
The DE1-SoC Computer includes a Nios II/f processor. When using the Quartus Prime Web Edition, compiling a
design with a Nios II/s or Nios II/f processor will produce a time-limited SOF file. As a result, the board must remain
connected to the host computer, and the design cannot be set as the default configuration, as discussed in Section 6.
Designs using only Nios II/e processors and designs compiled using the Quartus Prime Subscription Edition do not
have this restriction.
7 Memory Layout
Table 5 summarizes the memory map used in the DE1-SoC Computer.
Pressing Next again opens the window in Figure 28. Here the user selects the type of program that will be used,
such as Assembly language, or C. Then, the check box shown in the figure can be used to display the list of sample
programs for the DE1-SoC Computer that are described in this document. When a sample program is selected in
this list, its source files, and other settings, can be copied into the project folder in subsequent screens of the Monitor
Program.
Figure 29 gives the final screen that is used to create a new project in the Monitor Program. This screen shows the
default addresses of compiler and linker sections that will be used for the assembly language or C program associ-
ated with the Monitor Program project. In the figure, the drop-down menu called Linker Section Presets has been
set to Exceptions. With this setting the Monitor Program uses specific compiler and linker sections for the selected
processor. For the Nios II processor, these sections are for reset and exceptions code, and another section for the
main program, called .text. For the A9 processor, it has a section for the exception table, called .vectors, and another
section for the main program, called .text. It also shows the initial value used to set the main stack pointer for C
programs, which is the starting address of the .stack section.
9 Appendix
This section contains all of the source code files mentioned in the document.
9.1 Timers
.include "address_map_arm.s"
.equ bit_24_pattern, 0x01000000
/* This program provides a simple example of code for the ARM A9. The program
* performs the following:
* 1. starts the ARM A9 private timer
* 2. loops forever, toggling the HPS green light LEDG when the timer expires
*/
.text
.global _start
_start:
LDR R0, =HPS_GPIO1_BASE // GPIO1 base address
LDR R1, =MPCORE_PRIV_TIMER // MPCore private timer base address
LDR R2, =bit_24_pattern // value to turn on the HPS green light LEDG
STR R2, [R0, #0x4] // write to the data direction register to
// set bit 24 (LEDG) of GPIO1 to be an
output
LDR R3, =200000000 // timeout = 1/(200 MHz) x 200x10^6 = 1 sec
STR R3, [R1] // write to timer load register
MOV R3, #0b011 // set bits: mode = 1 (auto), enable = 1
STR R3, [R1, #0x8] // write to timer control register
LOOP:
STR R2, [R0] // turn on/off LEDG
WAIT:
LDR R3, [R1, #0xC] // read timer status
CMP R3, #0
BEQ WAIT // wait for timer to expire
.end
Listing 1. An example of assembly language code that uses a timer.
#include "address_map_arm.h"
/* This program provides a simple example of code for the ARM A9. The
* program performs the following:
* 1. starts the ARM A9 private timer
* 2. loops indefinitely, toggling the green light LEDG when the timer expires
*/
int main(void)
{
/* Declare volatile pointers to I/O registers (volatile means that the
* locations will not be cached, even in registers) */
volatile int * HPS_GPIO1_ptr = (int *)HPS_GPIO1_BASE;
volatile int * MPcore_private_timer_ptr = (int *)MPCORE_PRIV_TIMER;
int HPS_LEDG = bit_24_pattern; // value to turn on the HPS green light LEDG
int counter = 200000000; // timeout = 1/(200 MHz) x 200x10^6 = 1 sec
*(HPS_GPIO1_ptr + 1) =
bit_24_pattern; // write to the data direction register to set
// bit 24 (LEDG) of GPIO1 to be an output
*(MPcore_private_timer_ptr) = counter; // write to timer load register
*(MPcore_private_timer_ptr + 2) = 0b011; // mode = 1 (auto), enable = 1
while (1)
{
*HPS_GPIO1_ptr = HPS_LEDG; // turn on/off LEDG
while (*(MPcore_private_timer_ptr + 3) == 0)
; // wait for timer to expire
* (MPcore_private_timer_ptr + 3) = 1; // reset timer flag bit
HPS_LEDG ^= bit_24_pattern; // toggle bit that controls LEDG
}
}
Listing 2. An example of C code that uses a timer.
/********************************************************************************
* This program demonstrates use of parallel ports in the Computer System
*
* It performs the following:
* 1. displays a rotating pattern on the LEDs
* 2. if a KEY is pressed, uses SW switches as the pattern
********************************************************************************/
.text /* executable code follows */
.global _start
_start:
DO_DISPLAY:
LDR R5, [R2] // load SW switches
NO_BUTTON:
STR R4, [R1] // store pattern to the LED displays
ROR R4, R0 // rotate the displayed pattern to the left
B DO_DISPLAY
LED_bits:
.word 0x0F0F0F0F
.end
Listing 3. An example of ARM A9 assembly language code that uses parallel ports.
#include "address_map_arm.h"
/* This program demonstrates use of parallel ports in the Computer System
*
* It performs the following:
* 1. displays a rotating pattern on the green LED
* 2. if a KEY is pressed, uses the SW switches as the pattern
*/
int main(void) {
/* Declare volatile pointers to I/O registers (volatile means that IO load
* and store instructions will be used to access these pointer locations,
* instead of regular memory loads and stores)
*/
volatile int * LED_ptr = (int *)LED_BASE; // LED address
volatile int * SW_switch_ptr = (int *)SW_BASE; // SW slider switch address
volatile int * KEY_ptr = (int *)KEY_BASE; // pushbutton KEY address
while (1) {
SW_value = *(SW_switch_ptr); // read the SW slider (DIP) switch values
/********************************************************************************
* Subroutine to send a character to the JTAG UART
* R0 = character to send
********************************************************************************/
.global PUT_JTAG
PUT_JTAG:
LDR R1, =JTAG_UART_BASE // JTAG UART base address
LDR R2, [R1, #4] // read the JTAG UART control register
LDR R3, =0xFFFF0000
ANDS R2, R2, R3 // check for write space
BEQ END_PUT // if no space, ignore the character
STR R0, [R1] // send the character
END_PUT:
BX LR
Listing 5. An example of assembly language code that uses the JTAG UART (Part a).
/********************************************************************************
* Subroutine to get a character from the JTAG UART
* returns the character read in R0
********************************************************************************/
.global GET_JTAG
GET_JTAG:
LDR R1, =JTAG_UART_BASE // JTAG UART base address
LDR R0, [R1] // read the JTAG UART data register
ANDS R2, R0, #0x8000 // check if there is new data
BEQ RET_NULL // if no data, return 0
AND R0, R0, #0x00FF // return the character
B END_GET
RET_NULL:
MOV R0, #0
END_GET:
BX LR
Listing 5. An example of assembly language code that uses the JTAG UART (Part b).
#include "JTAG_UART.h"
/*******************************************************************************
* Subroutine to send a character to the JTAG UART
******************************************************************************/
void put_jtag(volatile int * JTAG_UART_ptr, char c) {
int control;
control = *(JTAG_UART_ptr + 1); // read the JTAG_UART control register
if (control & 0xFFFF0000) // if space, echo character, else ignore
* (JTAG_UART_ptr) = c;
}
/*******************************************************************************
* Subroutine to read a character from the JTAG UART
* Returns \0 if no character, otherwise returns the character
******************************************************************************/
char get_jtag(volatile int * JTAG_UART_ptr) {
int data;
data = *(JTAG_UART_ptr); // read the JTAG_UART data register
if (data & 0x00008000) // check RVALID to see if there is new data
return ((char)data & 0xFF);
else
return (’\0’);
}
Listing 6. An example of C code that uses the JTAG UART (Part a).
#include "JTAG_UART.h"
#include "address_map_arm.h"
/*******************************************************************************
* This program demonstrates use of the JTAG UART port
*
* It performs the following:
* 1. sends a text string to the JTAG UART
* 2. reads character data from the JTAG UART
* 3. echos the character data back to the JTAG UART
******************************************************************************/
int main(void) {
/* Declare volatile pointers to I/O registers (volatile means that IO load
and store instructions will be used to access these pointer locations,
instead of regular memory loads and stores) */
volatile int * JTAG_UART_ptr = (int *)JTAG_UART_BASE; // JTAG UART address
Listing 6. An example of C code that uses the JTAG UART (Part b).
9.4 Interrupts
/*********************************************************************************
* Initialize the exception vector table
********************************************************************************/
.section .vectors, "ax"
/*********************************************************************************
* This program demonstrates use of interrupts with assembly code. It first starts
* two timers: an HPS timer, and the Altera interval timer (in the FPGA). The
* program responds to interrupts from these timers in addition to the pushbutton
* KEYs in the FPGA.
*
* The interrupt service routine for the HPS timer causes the main program to flash
* the green light connected to the HPS GPIO1 port.
*
* The interrupt service routine for the interval timer displays a pattern on
* the LED lights, and shifts this pattern either left or right. The shifting
* direction is set in the pushbutton interrupt service routine; it is reversed
* each time a KEY is pressed
********************************************************************************/
.text
.global _start
_start:
/* Set up stack pointers for IRQ and SVC processor modes */
MOV R1, #INT_DISABLE | IRQ_MODE
MSR CPSR_c, R1 // change to IRQ mode
LDR SP, =A9_ONCHIP_END - 3 // set IRQ stack to top of A9 onchip
// memory
/* Global variables */
.global TICK
TICK:
.word 0x0 // used by HPS timer
.global PATTERN
PATTERN:
.word 0x0F0F0F0F // pattern to show on the LED lights
.global KEY_DIR
KEY_DIR:
.word 0
.end
Listing 7. An example of assembly language code that uses interrupts (Part b).
/*
* Configure the Generic Interrupt Controller (GIC)
*/
.global CONFIG_GIC
CONFIG_GIC:
/* configure the HPS timer interrupt */
LDR R0, =0xFFFED8C4 // ICDIPTRn: processor targets register
LDR R1, =0x01000000 // set target to cpu0
STR R1, [R0]
/* configure the FPGA IRQ0 (interval timer) and IRQ1 (KEYs) interrupts */
LDR R0, =0xFFFED848 // ICDIPTRn: processor targets register
LDR R1, =0x00000101 // set targets to cpu0
STR R1, [R0]
/* Set the enable bit in the Distributor Control Register (ICDDCR). This bit
* allows the distributor to forward interrupts to the CPU interface(s) */
LDR R0, =MPCORE_GIC_DIST
STR R1, [R0, #ICDDCR]
BX LR
Listing 7. An example of assembly language code that uses interrupts (Part c).
HPS_TIMER_CHECK:
CMP R5, #HPS_TIMER0_IRQ // check for HPS timer interrupt
BNE INTERVAL_TIMER_CHECK
BL HPS_TIMER_ISR
B EXIT_IRQ
INTERVAL_TIMER_CHECK:
CMP R5, #INTERVAL_TIMER_IRQ // check for FPGA timer interrupt
BNE KEYS_CHECK
BL TIMER_ISR
B EXIT_IRQ
KEYS_CHECK:
CMP R5, #KEYS_IRQ
UNEXPECTED:
BNE UNEXPECTED // if not recognized, stop here
BL KEY_ISR
EXIT_IRQ:
/* Write to the End of Interrupt Register (ICCEOIR) */
STR R5, [R4, #ICCEOIR]
.include "address_map_arm.s"
.extern TICK
/******************************************************************************
* HPS timer0 interrupt service routine
*
* This code increments the TICK global variable, and clears the interrupt
*****************************************************************************/
.global HPS_TIMER_ISR
HPS_TIMER_ISR:
.include "address_map_arm.s"
.include "defines.s"
/* externally defined variables */
.extern KEY_DIR
.extern PATTERN
/*****************************************************************************
* Interval timer interrupt service routine
*
* Shifts a PATTERN being displayed on the LED lights. The shift direction
* is determined by the external variable KEY_PRESSED.
*
******************************************************************************/
.global TIMER_ISR
TIMER_ISR:
PUSH {R4-R7}
LDR R1, =TIMER_BASE // interval timer base address
MOVS R0, #0
STR R0, [R1] // clear the interrupt
SHIFT:
LDR R5, [R7] // get shift direction
CMP R5, #RIGHT
BNE SHIFT_L
MOVS R5, #1 // used to rotate right by 1 position
RORS R6, R5 // rotate right for KEY1
B END_TIMER_ISR
SHIFT_L:
MOVS R5, #31 // used to rotate left by 1 position
RORS R6, R5
END_TIMER_ISR:
STR R6, [R2] // store LED display pattern
POP {R4-R7}
BX LR
.end
Listing 10. Interrupt service routine for the interval timer.
.include "address_map_arm.s"
.include "defines.s"
.extern KEY_DIR /* externally defined variable */
/***************************************************************************************
* Pushbutton KEY - Interrupt Service Routine
*
* This routine toggles the KEY_DIR variable from 0 <-> 1
****************************************************************************************/
.global KEY_ISR
KEY_ISR:
LDR R0, =KEY_BASE // base address of pushbutton KEY parallel port
/* KEY[1] is the only key configured for interrupts, so just clear it. */
LDR R1, [R0, #0xC] // read edge capture register
STR R1, [R0, #0xC] // clear the interrupt
END_KEY_ISR:
BX LR
.end
Listing 11. Interrupt service routine for the pushbutton KEYs.
void set_A9_IRQ_stack(void);
void config_GIC(void);
void config_HPS_timer(void);
void config_HPS_GPIO1(void);
void config_interval_timer(void);
void config_KEYs(void);
void enable_A9_interrupts(void);
/* key_dir and pattern are written by interrupt service routines; we have to
* declare these as volatile to avoid the compiler caching their values in
* registers */
volatile int tick = 0; // set to 1 every time the HPS timer expires
volatile int key_dir = 0;
volatile int pattern = 0x0F0F0F0F; // pattern for LED lights
/* ********************************************************************************
* This program demonstrates use of interrupts with C code. It first starts
* two timers: an HPS timer, and the FPGA interval timer. The program responds
* to interrupts from these timers in addition to the pushbutton KEYs in the
* FPGA.
*
* The interrupt service routine for the HPS timer causes the main program to
* flash the green light connected to the HPS GPIO1 port.
*
* The interrupt service routine for the interval timer displays a pattern on
* the LED lights, and shifts this pattern either left or right. The shifting
* direction is reversed when KEY[1] is pressed
********************************************************************************/
int main(void)
{
volatile int * HPS_GPIO1_ptr = (int *)HPS_GPIO1_BASE; // GPIO1 base address
volatile int HPS_timer_LEDG =
0x01000000; // value to turn on the HPS green light LEDG
while (1)
{
if (tick)
{
tick = 0;
*HPS_GPIO1_ptr = HPS_timer_LEDG; // turn on/off the green light LEDG
HPS_timer_LEDG ^= 0x01000000; // toggle the bit that controls LEDG
}
}
}
/* setup HPS GPIO1. The GPIO1 port has one green light (LEDG) and one pushbutton
* KEY connected for the DE1-SoC Computer. The KEY is connected to GPIO1[25],
* and is not used here. The green LED is connected to GPIO1[24]. */
void config_HPS_GPIO1()
{
volatile int * HPS_GPIO1_ptr = (int *)HPS_GPIO1_BASE; // GPIO1 base address
*(HPS_GPIO1_ptr + 0x1) =
0x01000000; // write to the data direction register to set
// bit 24 (LEDG) to be an output
// Other possible actions include setting up GPIO1 to use the KEY, including
// setting the debounce option and causing the KEY to generate an interrupt.
// We do not use the KEY in this example.
}
/* set the interval timer period for scrolling the HEX displays */
int counter = 5000000; // 1/(100 MHz) x 5x10^6 = 50 msec
*(interval_timer_ptr + 0x2) = (counter & 0xFFFF);
*(interval_timer_ptr + 0x3) = (counter >> 16) & 0xFFFF;
/*
* Initialize the banked stack pointer register for IRQ mode
*/
void set_A9_IRQ_stack(void)
{
int stack, mode;
stack = A9_ONCHIP_END - 7; // top of A9 onchip memory, aligned to 8 bytes
/* change processor to IRQ mode with interrupts disabled */
mode = INT_DISABLE | IRQ_MODE;
asm("msr cpsr, %[ps]" : : [ps] "r"(mode));
/* set banked stack pointer */
asm("mov sp, %[ps]" : : [ps] "r"(stack));
/*
* Turn on interrupts in the ARM processor
*/
void enable_A9_interrupts(void)
{
int status = SVC_MODE | INT_ENABLE;
asm("msr cpsr, %[ps]" : : [ps] "r"(status));
}
/*
* Configure the Generic Interrupt Controller (GIC)
*/
void config_GIC(void)
{
int address; // used to calculate register addresses
void HPS_timer_ISR(void);
void interval_timer_ISR(void);
void pushbutton_ISR(void);
return;
}
;
}
#include "address_map_arm.h"
/******************************************************************************
* HPS timer0 interrupt service routine
*
* This code increments the tick variable, and clears the interrupt
*****************************************************************************/
void HPS_timer_ISR()
{
volatile int * HPS_timer_ptr = (int *)HPS_TIMER0_BASE; // HPS timer address
#include "address_map_arm.h"
return;
}
Listing 15. Interrupt service routine for the interval timer.
#include "address_map_arm.h"
return;
}
Listing 16. Interrupt service routine for the pushbutton KEYs.
9.5 Audio
#include "address_map_arm.h"
/* globals */
#define BUF_SIZE 80000 // about 10 seconds of buffer (@ 8K samples/sec)
#define BUF_THRESHOLD 96 // 75% of 128 word buffer
/* function prototypes */
void check_KEYs(int *, int *, int *);
/*******************************************************************************
* This program performs the following:
* 1. records audio for 10 seconds when KEY[0] is pressed. LEDR[0] is lit
* while recording.
* 2. plays the recorded audio when KEY[1] is pressed. LEDR[1] is lit while
* playing.
******************************************************************************/
int main(void) {
/* Declare volatile pointers to I/O registers (volatile means that IO load
and store instructions will be used to access these pointer locations,
instead of regular memory loads and stores) */
volatile int * red_LED_ptr = (int *)LED_BASE;
volatile int * audio_ptr = (int *)AUDIO_BASE;
while (1) {
check_KEYs(&record, &play, &buffer_index);
if (record) {
*(red_LED_ptr) = 0x1; // turn on LEDR[0]
fifospace =
*(audio_ptr + 1); // read the audio port fifospace register
if ((fifospace & 0x000000FF) > BUF_THRESHOLD) // check RARC
{
// store data until the the audio-in FIFO is empty or the buffer
// is full
while ((fifospace & 0x000000FF) && (buffer_index < BUF_SIZE)) {
left_buffer[buffer_index] = *(audio_ptr + 2);
right_buffer[buffer_index] = *(audio_ptr + 3);
++buffer_index;
if (buffer_index == BUF_SIZE) {
// done recording
record = 0;
*(red_LED_ptr) = 0x0; // turn off LEDR
}
fifospace = *(audio_ptr +
1); // read the audio port fifospace register
}
}
} else if (play) {
*(red_LED_ptr) = 0x2; // turn on LEDR_1
fifospace =
*(audio_ptr + 1); // read the audio port fifospace register
if ((fifospace & 0x00FF0000) > BUF_THRESHOLD) // check WSRC
{
// output data until the buffer is empty or the audio-out FIFO
// is full
while ((fifospace & 0x00FF0000) && (buffer_index < BUF_SIZE)) {
*(audio_ptr + 2) = left_buffer[buffer_index];
*(audio_ptr + 3) = right_buffer[buffer_index];
++buffer_index;
if (buffer_index == BUF_SIZE) {
// done playback
play = 0;
*(red_LED_ptr) = 0x0; // turn off LEDR
}
fifospace = *(audio_ptr +
1); // read the audio port fifospace register
}
}
}
}
}
/****************************************************************************************
* Subroutine to read KEYs
****************************************************************************************/
void check_KEYs(int * KEY0, int * KEY1, int * counter) {
volatile int * KEY_ptr = (int *)KEY_BASE;
volatile int * audio_ptr = (int *)AUDIO_BASE;
int KEY_value;
*(audio_ptr) = 0x4;
*(audio_ptr) = 0x0;
*KEY0 = 1;
} else if (KEY_value == 0x2) // check KEY1
{
// reset counter to start playback
*counter = 0;
// clear audio-out FIFO
*(audio_ptr) = 0x8;
*(audio_ptr) = 0x0;
*KEY1 = 1;
}
}
Listing 17. An example of code that uses the audio port.
/* function prototypes */
void video_text(int, int, char *);
void video_box(int, int, int, int, short);
int resample_rgb(int, int);
int get_data_bits(int);
/*******************************************************************************
* This program demonstrates use of the video in the computer system.
* Draws a blue box on the video display, and places a text string inside the
* box
******************************************************************************/
int main(void) {
volatile int * video_resolution = (int *)(PIXEL_BUF_CTRL_BASE + 0x8);
screen_x = *video_resolution & 0xFFFF;
screen_y = (*video_resolution >> 16) & 0xFFFF;
/* update color */
short background_color = resample_rgb(db, INTEL_BLUE);
/*******************************************************************************
* Subroutine to send a string of text to the video monitor
******************************************************************************/
void video_text(int x, int y, char * text_ptr) {
int offset;
volatile char * character_buffer =
(char *)FPGA_CHAR_BASE; // video character buffer
/*******************************************************************************
* Draw a filled rectangle on the video monitor
* Takes in points assuming 320x240 resolution and adjusts based on differences
* in resolution and color bits.
******************************************************************************/
void video_box(int x1, int y1, int x2, int y2, short pixel_color) {
int pixel_buf_ptr = *(int *)PIXEL_BUF_CTRL_BASE;
int pixel_ptr, row, col;
int x_factor = 0x1 << (res_offset + col_offset);
int y_factor = 0x1 << (res_offset);
x1 = x1 / x_factor;
x2 = x2 / x_factor;
y1 = y1 / y_factor;
y2 = y2 / y_factor;
/********************************************************************************
* Resamples 24-bit color to 16-bit or 8-bit color
*******************************************************************************/
int resample_rgb(int num_bits, int color) {
if (num_bits == 8) {
color = (((color >> 16) & 0x000000E0) | ((color >> 11) & 0x0000001C) |
((color >> 6) & 0x00000003));
color = (color << 8) | color;
} else if (num_bits == 16) {
/********************************************************************************
* Finds the number of data bits from the mode
*******************************************************************************/
int get_data_bits(int mode) {
switch (mode) {
case 0x0:
return 1;
case 0x7:
return 8;
case 0x11:
return 8;
case 0x12:
return 9;
case 0x14:
return 16;
case 0x17:
return 24;
case 0x19:
return 30;
case 0x31:
return 8;
case 0x32:
return 12;
case 0x33:
return 16;
case 0x37:
return 32;
case 0x39:
return 40;
}
}
Listing 18. An example of code that uses the video-out port.
9.7 PS/2
#include "address_map_arm.h"
/* function prototypes */
void HEX_PS2(char, char, char);
/*******************************************************************************
* This program demonstrates use of the PS/2 port by displaying the last three
* bytes of data received from the PS/2 port on the HEX displays.
******************************************************************************/
int main(void) {
/* Declare volatile pointers to I/O registers (volatile means that IO load
and store instructions will be used to access these pointer locations,
instead of regular memory loads and stores) */
volatile int * PS2_ptr = (int *)PS2_BASE;
while (1) {
PS2_data = *(PS2_ptr); // read the Data register in the PS/2 port
RVALID = PS2_data & 0x8000; // extract the RVALID field
if (RVALID) {
/* shift the next data byte into the display */
byte1 = byte2;
byte2 = byte3;
byte3 = PS2_data & 0xFF;
HEX_PS2(byte1, byte2, byte3);
/****************************************************************************************
* Subroutine to show a string of HEX data on the HEX displays
****************************************************************************************/
void HEX_PS2(char b1, char b2, char b3) {
volatile int * HEX3_HEX0_ptr = (int *)HEX3_HEX0_BASE;
volatile int * HEX5_HEX4_ptr = (int *)HEX5_HEX4_BASE;
Copyright © Intel Corporation. All rights reserved. Intel, the Intel logo, Altera, Arria, Avalon, Cyclone, Enpirion,
MAX, Nios, Quartus and Stratix words and logos are trademarks of Intel Corporation or its subsidiaries in the U.S.
and/or other countries. Intel warrants performance of its FPGA and semiconductor products to current specifications
in accordance with Intel’s standard warranty, but reserves the right to make changes to any products and services
at any time without notice. Intel assumes no responsibility or liability arising out of the application or use of any
information, product, or service described herein except as expressly agreed to in writing by Intel. Intel customers
are advised to obtain the latest version of device specifications before relying on any published information and
before placing orders for products or services.