ARM Cortex-M4 Programming Model
ARM = Advanced RISC Machines, Ltd.
ARM licenses IP to other companies (ARM does not fabricate chips)
2005: ARM had 75% of embedded RISC market, with 2.5 billion processors
ARM available as microcontrollers, IP cores, etc.
www.arm.com
1
ARM instruction set architecture
ARM versions.
ARM programming model.
ARM memory organization.
ARM assembly language.
ARM data operations.
ARM flow of control.
2
ARM Architecture versions
(From arm.com)
3
Instruction Sets
4
Arm Processor Families
Cortex-A75 Cortex-A55
Cortex-A73 Cortex-A53
Cortex-A series (advanced application) Cortex-A72 Cortex-A35
High-performance processors for open OSs Cortex-A57 Cortex-A32
Cortex-A
App’s: smartphones, digital TV, server solutions, and home Cortex-A17
Cortex-A15
Cortex-A8
Cortex-A7
gateways. Cortex-A9 Cortex-A5
Cortex-R series (real-time) Cortex-R8 Cortex-R52
Exceptional performance for real-time applications Cortex-R7 Cortex-R5 Cortex-R
App’s: automotive braking systems and powertrains. Cortex-R4
Cortex-M7 Cortex-M0
Cortex-M series (microcontroller) Cortex-M4 Cortex-M23
Cost-sensitive solutions for deterministic microcontroller
Cortex-M3 Cortex-M33
Cortex-M1
Cortex-M
applications Cortex-M0+
App’s: microcontrollers, smart sensors, automotive body
electronics, and airbags. SC000
SC100 SecurCore
SecurCore series SC300
High-security applications such as smartcards and e- Arm11
Arm9 Classic
government Arm7
Classic processors As of Nov 2017
Include Arm7, Arm9, and Arm11 families
ARM Cortex-M instruction sets
6
Programmer’s model of a CPU
What information is specified in an “instruction” to accomplish a
task?
Operations: add, subtract, move, jump
Operands: data manipulated by operations
# of operands per instruction (1-2-3)
Data sizes & types
# bits (1, 8, 16, 32, …)
signed/unsigned integer, floating-point, character …
Locations of operands
Memory – specify location by a memory “address”
CPU Registers – specify register name/number
Immediate – data embedded in the instruction code
Input/output device “ports”/interfaces
7
RISC vs. CISC architectures
CISC = “Complex Instruction Set Computer”
Rich set of instructions and options to minimize #operations
required to perform a given task
Example: Intel x86 instruction set architecture
RISC = “Reduced Instruction Set Computer”
Fixed instruction length
Fewer/simpler instructions than CISC CPU 32-bit load/store
architecture
Limited addressing modes, operand types
Simple design easier to speed up, pipeline & scale
Example: ARM architecture
Program execution time =
(# instructions) x (# clock cycles/instruction) x (clock period)
8
ARM instruction format
Add instruction: ADD R1, R2, R3 ;2nd source operand = register
ADD R1, R2, #5 ;2nd source operand = constant
1 2 3 4
1. operation: binary addition (compute R1 = R2 + 5)
2. destination: register R1 (replaces original contents of R1)
3. left-hand operand: register R2
4. right-hand operand:
Option 1: register R3
Option 2: constant 5 (# indicates constant)
operand size: 32 bits (all arithmetic/logical instructions)
operand type: signed or unsigned integer
9
ARM assembly language
Fairly standard assembly language format:
memory address/pointer
LDR r0,[r8] ;a comment
label ADD r4,r0,r1 ;r4=r0+r1
destination source/left source/right
label (optional) refers to the location of this instruction
10
Processor core registers
• All registers are 32 bits wide
• 13 general purpose registers
• Registers r0 – r7 (Low registers)
• Registers r8 – r12 (High registers)
• Use to hold data, addresses, etc.
• 3 registers with special meaning/usage
• Stack Pointer (SP) – r13
• Link Register (LR) – r14
• Program Counter (PC) – r15
• xPSR – Program Status Register
• Composite of three PSRs
• Includes ALU flags (N,Z,C,V)
11
Program status register (PSR)
Flags
Program Status Register xPSR is a composite of 3 PSRs:
APSR - Application Program Status Register – ALU condition flags
N (negative), Z (zero), C (carry/borrow), V (2’s complement overflow)
Flags set by ALU operations; tested by conditional jumps/execution
IPSR - Interrupt Program Status Register
Interrupt/Exception No.
EPSR - Execution Program Status Register
T bit = 1 if CPU in “Thumb mode” (always for Cortex-M4), 0 in “ARM mode”
IT field – If/Then block information
ICI field – Interruptible-Continuable Instruction information
xPSR stored on the stack on exception entry
12
Data types supported in ARM
Integer ALU operations are performed only on 32-bit data
Signed or unsigned integers
Data sizes in memory:
Byte (8-bit), Half-word (16-bit), Word (32-bit), Double Word (64-bit)
Bytes/half-words are converted to 32-bits when moved into a register
Signed numbers – extend sign bit to upper bits of a 32-bit register
Unsigned numbers –fill upper bits of a 32-bit register with 0’s
Examples:
255 (unsigned byte) 0xFF=>0x000000FF (fill upper 24 bits with 0)
-1 (signed byte) 0xFF=>0xFFFFFFFF (fill upper 24 bits with sign bit 1)
+1 (signed byte) 0x01=>0x00000001 (fill upper 24 bits with sign bit 0)
-32768 (signed half-word) 0x8000=>0xFFFF8000 (sign bit = 1)
32768 (unsigned half-word) 0x8000=>0x00008000
+32767 (signed half-word) 0x7FFF=>0x00007FFF (sign bit = 0)
Cortex-M4F supports single and double-precision IEEE floating-point data
(Floating-point ALU is optional in Cortex-M4 implementations)
13
C/C++ language data types
Type Size Range of values
(bits)
char 8 [-27 .. +27–1] = [-128 .. +127]
signed char Compiler-specific (not specified in C standard)
ARM compiler default is signed
unsigned char 8 [0 .. 28–1] = [0..255]
short 16 [-215 .. +215–1]
signed short
unsigned short 16 [0 .. 216–1]
int 32 [-231 .. +231–1] (natural size of host CPU)
signed int int specified as signed in the C standard
unsigned int 32 [0 .. 232–1]
long 32 [-231 .. +231–1]
long long 64 [-263 .. +263–1]
float 32 IEEE single-precision floating-point format
double 64 IEEE double-precision floating-point format
14
Directive: Data Allocation
Directive Description Memory Space
DCB Define Constant Byte Reserve 8-bit values
DCW Define Constant Half-word Reserve 16-bit values
DCD Define ConstantWord Reserve 32-bit values
DCQ Define Constant Reserve 64-bit values
SPACE Defined Zeroed Bytes Reserve a number of zeroed bytes
FILL Defined Initialized Bytes Reserve and fill each byte with a value
DCx : reserve space and initialize value(s) for ROM
(initial values ignored for RAM)
SPACE : reserve space without assigning initial values
(especially useful for RAM)
15
Directive: Data Allocation
AREA myData, DATA, READWRITE
hello DCB "Hello World!",0 ; Allocate a string that is null-terminated
dollar DCB 2,10,0,200 ; Allocate integers ranging from -128 to 255
scores DCD 2,3,-8,4 ; Allocate 4 words containing decimal values
miles DCW 100,200,50,0 ; Allocate integers between –32768 and 65535
p SPACE 255 ; Allocate 255 bytes of zeroed memory space
f FILL 20,0xFF,1 ; Allocate 20 bytes and set each byte to 0xFF
binary DCB 2_01010101 ; Allocate a byte in binary
octal DCB 8_73 ; Allocate a byte in octal
char DCB ‘A’ ; Allocate a byte initialized to ASCII of ‘A’
16
Memory usage
Code memory (normally read-only memory)
Program instructions
Constant data
Data memory (normally read/write memory – RAM)
Variable data/operands
Stack (located in data memory)
Special Last-In/First-Out (LIFO) data structure
Save information temporarily and retrieve it later
Return addresses for subroutines and interrupt/exception handlers
Data to be passed to/from a subroutine/function
Stack Pointer register (r13/sp) points to last item placed on the stack
Peripheral addresses
Used to access registers in “peripheral functions” (timers, ADCs,
communication modules, etc.) outside the CPU
17
Cortex-M4 processor
memory map
Cortex peripheral function registers
(NVIC, tick timer, etc.)
(off-chip)
STM32F407 microcontroller: All ARM
addresses
Peripheral function registers (off-chip) are 32 bits
SRAM1 (128Kbyte):
[0x2000_0000 .. 0x2001_FFFF]
SRAM2 (64Kbyte):
[0x1000_0000 .. 0x1000_FFFF]
Flash memory (1MByte):
[0x0800_0000 .. 0x0800F_FFFF]
18
We will use Flash for code, SRAM1 for data.
Endianness
Relationship between bit and byte/word ordering defines
“endianness”:
little-endian (default) big-endian (option)
bit 31 bit 0 bit 0 bit 31
byte 3 byte 2 byte 1 byte 0 byte 0 byte 1 byte 2 byte 3
Address: 100 0x78 Address: 100 0x12
Example:
101 0x56 101 0x34
32-bit data =
102 0x34 0x12345678 102 0x56
103 0x12 103 0x78
103 102 101 100 100 101 102 103
19 12 34 56 78 12 34 56 78
Physical memory organization
Physical memory may be organized as N bytes per addressable word
ARM memories normally 4-bytes wide
“Align” 32-bit data to a Word boundary (address that is a multiple of 4)
All bytes of a word must be accessible with one memory read/write
Byte 3 Byte 2 Byte 1 Byte 0
Byte 103 102 101 100 Word 100
addresses
107 106 105 104 Word 104
10B 10A 109 108 Word 108
10F 10E 10D 10C Word 10C
20 ARM instructions can read/write 8/16/32-bit data values
First Assembly
21
Directive: AREA
AREA myData, DATA, READWRITE ; Define a data section
Array DCD 1, 2, 3, 4, 5 ; Define an array with five integers
AREA myCode, CODE, READONLY ; Define a code section
EXPORT __main ; Make __main visible to the linker
ENTRY ; Mark the entrance to the entire program
__main PROC ; PROC marks the begin of a subroutine
... ; Assembly program starts here.
ENDP ; Mark the end of a subroutine
END ; Mark the end of a program
The AREA directive indicates to the assembler the start of a new data or code section.
Areas are the basic independent and indivisible unit processed by the linker.
Each area is identified by a name and areas within the same source file cannot share the same name.
An assembly program must have at least one code area.
By default, a code area can only be read and a data area may be read from and written to.
22
Directive: END
AREA myData, DATA, READWRITE ; Define a data section
Array DCD 1, 2, 3, 4, 5 ; Define an array with five integers
AREA myCode, CODE, READONLY ; Define a code section
EXPORT __main ; Make __main visible to the linker
ENTRY ; Mark the entrance to the entire program
__main PROC ; PROC marks the begin of a subroutine
... ; Assembly program starts here.
ENDP ; Mark the end of a subroutine
END ; Mark the end of a program
The END directive indicates the end of a source file.
Each assembly program must end with this directive.
23
Directive: ENTRY
AREA myData, DATA, READWRITE ; Define a data section
Array DCD 1, 2, 3, 4, 5 ; Define an array with five integers
AREA myCode, CODE, READONLY ; Define a code section
EXPORT __main ; Make __main visible to the linker
ENTRY ; Mark the entrance to the entire program
__main PROC ; PROC marks the begin of a subroutine
... ; Assembly program starts here.
ENDP ; Mark the end of a subroutine
END ; Mark the end of a program
The ENTRY directive marks the first instruction to be executed within an application.
There must be one and only one entry directive in an application, no matter how many source
files the application has.
24
Directive: PROC and ENDP
AREA myData, DATA, READWRITE ; Define a data section
Array DCD 1, 2, 3, 4, 5 ; Define an array with five integers
AREA myCode, CODE, READONLY ; Define a code section
EXPORT __main ; Make __main visible to the linker
ENTRY ; Mark the entrance to the entire program
__main PROC ; PROC marks the begin of a subroutine
... ; Assembly program starts here.
ENDP ; Mark the end of a subroutine
END ; Mark the end of a program
PROC and ENDP are to mark the start and end of a function (also called subroutine or procedure).
A single source file can contain multiple subroutines, with each of them defined by a pair of PROC
and ENDP.
PROC and ENDP cannot be nested. We cannot define a subroutine within another subroutine.
25
Directive: EXPORT and IMPORT
AREA myData, DATA, READWRITE ; Define a data section
Array DCD 1, 2, 3, 4, 5 ; Define an array with five integers
AREA myCode, CODE, READONLY ; Define a code section
EXPORT __main ; Make __main visible to the linker
IMPORT sinx ; Function sinx defined in another file
ENTRY ; Mark the entrance to the entire program
__main PROC ; PROC marks the begin of a subroutine
... ; Assembly program starts here.
BL sinx ; Call the sinx function
ENDP ; Mark the end of a subroutine
END ; Mark the end of a program
The EXPORT declares a symbol and makes this symbol visible to the linker.
The IMPORT gives the assembler a symbol that is not defined locally in the current assembly file.
The IMPORT is similar to the “extern” keyword in C.
26
Directive: EQU
; Interrupt Number Definition (IRQn)
BusFault_IRQn EQU -11 ; Cortex-M3 Bus Fault Interrupt
SVCall_IRQn EQU -5 ; Cortex-M3 SV Call Interrupt
PendSV_IRQn EQU -2 ; Cortex-M3 Pend SV Interrupt
SysTick_IRQn EQU -1 ; Cortex-M3 System Tick Interrupt
MyConstant EQU 1234 ; Constant 1234 to use later
The EQU directive associates a symbolic name to a numeric constant. Similar to the use of
#define in a C program, the EQU can be used to define a constant in an assembly code.
Example:
MOV R0, #MyConstant ; Constant 1234 placed in R0
27
Directive: ALIGN
AREA example, CODE, ALIGN = 3 ; Memory address begins at a multiple of 8
ADD r0, r1, r2 ; Instructions start at a multiple of 8
AREA myData, DATA, ALIGN = 2 ; Address starts at a multiple of four
a DCB 0xFF ; The first byte of a 4-byte word
ALIGN 4, 3 ; Align to the last byte of a word
b DCB 0x33 ; Set the fourth byte of a 4-byte word
c DCB 0x44 ; Add a byte to make next data misaligned
ALIGN ; Force the next data to be aligned
d DCD 12345 ; Skip three bytes and store the word
ALIGN generally used as in this example,
to align a variable to its data type.
28
Directive: INCLUDE or GET
INCLUDE constants.s ; Load Constant Definitions
AREA main, CODE, READONLY
EXPORT __main
ENTRY
__main PROC
...
ENDP
END
The INCLUDE or GET directive is to include an assembly source file within another source
file.
It is useful to include constant symbols defined by using EQU and stored in a separate source
file.
29