Overview of ARM Microprocessors
Overview of ARM Microprocessors
University of Belgrade
School of Electrical Engineering
Department of Computer Science
Authors:
Goran Timotic gogi@[Link]
Gvozden Marinkovic mgvozden@[Link]
Prof. Dr. Veljko Milutinovic vm@[Link]
ARM - Introduction
Advances RISC Machines (now known as ARM) was established
as a joint venture between Acorn, Apple and VLSI between Acorn,
Apple and VLSI in November 1990
ARM is the industry's leading provider of 16/32-bit embedded
RISC microprocessor solutions
The company licenses its high-performance, low-cost, power-
efficient RISC processors, peripherals, and system-chip designs
to leading international electronics companies
ARM provides comprehensive support required in
developing a complete system
2/70
ARM - Introduction
Currently available processors:
Arm 1
– This was the very first ARM processor and
was superseded by the ARM2 fairly quickly
Arm 2
– The ARM2 chip, and processor cell, features 27 registers of which 16
are accessible at any one time. Four processor modes are available:
• USR : user mode
• IRQ : interrupt mode (with a private copy of R13 and R14)
• FIQ : fast interrupt mode (private copies of R8 to R14)
3/70
ARM - Introduction
Arm 3
– This is an upgraded ARM2 cell, with a cache
and dedicated coprocessor interface added
Arm 6
– This processor cell is the first of the commercially available ARMs to
have a full 32bit addressing capability
– Additionally the processor now has 31 registers in it along with six new
processor modes:
• User32 - 32 bit USR mode
• Supervisor32 - 32 bit SVC mode (private SPSR register)
• IRQ32 - 32 bit IRQ mode (private SPSR register)
• FIQ32 - 32 bit FIQ mode. (private SPSR register)
• Abort32 - Memory fetch abort mode (private SPSR register)
4/70
ARM - Introduction
Thumb
– This isn't a processor per se, but more a 16 bit compressed
instruction set that is decoded by the Thumb core into full 32 bit
instructions that are then fed to an ARM core for execution
– This offers two important advantages for
the embedded processor market:
• Increased code density
• Simplified system design
5/70
ARM - Introduction
Arm 7
– The ARM7 cell is functionally identical to the ARM6 cell in capabilities
but may be clocked faster than the ARM6
– A variant of the ARM7 cell offers an improved hardware multiply,
suitable for DSP work
Arm 8
– Includes a five stage pipeline, a speculative instruction fetcher
and internal tweaks to the processor to allow a higher clock speed
StrongARM
– This is the high speed variant of the ARM chip family
– Architecturally it is similar to the ARM8 core, sharing the five stage
pipeline with that processor
– A further difference is change from a unified data and instruction
cache to a split, Harvard architecture, instruction and data cache 6/70
ARM - Introduction
ARM9
– An incremental improvement over the ARM8 this chip features the
same five stage pipeline but is now a Harvard Architecture chip,
like the StrongARM
ARM 10
– 300 MHz
– 400 MIPS
– 600 mWatts
7/70
ARM Powered Products
Consumer
Automotive Digital Imaging
Entertainment
8/70
ARM7 - Features
32-bit RISC processor (32-bit data & address bus)
Big and Little Endian operating modes
High performance RISC (17 MIPS sustained @ 25 MHz (25 MIPS
peak) @ 3V)
Low power consumption (0.6mA/MHz @ 3V fabricated in .8µm
CMOS)
Fully static operation (ideal for power-sensitive applications)
Fast interrupt response (for real-time applications)
Virtual Memory System Support
Excellent high-level language support
Simple but powerful instruction set
9/70
ARM7 - Applications
The ARM7 is ideally suited to those applications requiring RISC
performance from a compact, power-efficient processor
– Telecomms - GSM terminal controller
– Datacomms - Protocol conversion
– Portable Computing - Palmtop computer
– Portable Instrument - Hendheld data acquisition unit
– Automotive - Engine management unit
– Information systems - Smart cards
– Imaging - JPEG controller
10/70
ARM7 - Block Diagram
A[31:0]
ALE nEXEC
DATA32
I BIGEND
Address Register n PROG32
c
r MCLK
e nWAIT
P m
e nRW
C n
A Address nBW
L B t
U u Incrementer e Instruction nIRQ
s r nFIQ
B Decoder &
u B nRESET
s u Control Logic
s ABORT
Register Bank nOPC
(31 x 32 bit registers) nTRANS
(6 status registers) nMREQ
SEQ
LOCK
nCPI
B CPA
A CPB
Booth's b
b u nM[4:0]
u Multiplier s
s
Barrel Shifter
PROG32 DATA[31:0]
Configuration DATA32
BIGEND DOUT[31:0]
nENOUT Memory
nEXEC Interface
nMREQ
nIRQ ARM7 SEQ
Interrupts nFIQ nRW
nBW
nRESET LOCK
13/70
ARM7 - Signal Description
Memory Interface Memory Management Interface
– A[31:0] Addresses – nTRANS Not memory translate
– DATA[31:0] Data bus in – ABORT Memory abort
– DOUT[31:0] Data bus out Coprocessor Interface
– nENOUT Not enable data – nOPC Not op-code fetch
outputs – nCPI Not coprocessor
– nMREQ Not memory request instruction
– SEQ Sequential address – CPA Coprocessor absent
– nRW Not read/write – CPB Coprocessor busy
– nBW Not byte/word Other
– LOCK Locked operation – nEXEC ******************
– nRESET Not reset
14/70
ARM7 - Programmer’s Model
ARM7 supports a variety of operating configurations
– hardware configurations (controlled by inputs)
– operating modes (controlled by software)
15/70
ARM7 - Hardware Configuration Signals
3 hardware configuration signals
– Big Endian
– Little Endian
– Configuration Bits for Backward Compatibility
Hardware configuration signals may be changed while the
processor is running
16/70
ARM7 - Hardware Configuration Signals
Big Endian
17/70
ARM7 - Hardware Configuration Signals
Little Endian
18/70
ARM7 - Hardware Configuration Signals
Configuration Bits for Backward Compatibility
19/70
ARM7 - Operating Modes
20/70
ARM7 - Registers
37 registers
– 31 general 32 bit registers
– 6 status registers
16 general registers and one or two status registers are visible at
any time
The visible registers depend on the processor mode
The other registers (the banked registers) are switched in to
support IRQ, FIQ, Supervisor, Abort and Undefined mode
processing
21/70
ARM7 - Registers
R0 to R15 are directly accessible
– R0 to R14 are general purpose
– R15 holds the Program Counter (PC)
CPSR - Current Program Status Register contains condition code
flags and the current mode bits
5 SPSRs (Saved Program Status Registers) which are loaded with
CPSR when an exceptions occurs
22/70
ARM7 - Registers
General Registers and Program Counter Modes
User32 Fiq32 Supervisor32 Abort32 IRQ32 Undefined32
R0 R0 R0 R0 R0 R0
R1 R1 R1 R1 R1 R1
R2 R2 R2 R2 R2 R2
R3 R3 R3 R3 R3 R3
R4 R4 R4 R4 R4 R4
R5 R5 R5 R5 R5 R5
R6 R6 R6 R6 R6 R6
R7 R7 R7 R7 R7 R7
R8 R8_fiq R8 R8 R8 R8
R9 R9_fiq R9 R9 R9 R9
R10 R10_fiq R10 R10 R10 R10
R11 R11_fiq R11 R11 R11 R11
R12 R12_fiq R12 R12 R12 R12
R13 R13_fiq R13_svc R13_abt R13_irq R13_und
R14 R14_fiq R14_svc R14_abt R14_irq R14_und
R15(PC) R15(PC) R15(PC) R15(PC) R15(PC) R15(PC)
23/70
ARM7 - Registers
R14 is used as the subroutine link register and receives a copy of
R15 when a Branch and Link instruction is executed
R14_svc, R14_irq, R14_fiq, R14_abt and R14_und are used to
– hold the return values of R15 when interrupts and exceptions arise,
– when Branch and Link instructions are executed within interrupt or
exception routines
FIQ mode has seven banked registers mapped to R8-14 (R8_fiq-
R14_fiq)
User mode, IRQ mode, Supervisor mode, Abort mode and
Undefined mode each have two banked registers mapped to R13
and R14
The two banked registers allow these modes to each have a
private stack pointer and link register
24/70
ARM7 - Registers
The N, Z, C and V are condition code flags
– may be changed as a result of arithmetic and logical operations in the
processor
– may be tested by all instructions to determine if the instruction is to
be executed
The I and F bits are the interrupt disable bits
The M0, M1, M2, M3 and M4 bits are the mode bits
31 30 29 28 27 8 7 6 5 4 3 2 1 0
N Z C V F I M4 M3 M2 M1 M0
25/70
ARM7 - Registers
The Mode Bits
26/70
ARM7 - Exceptions
Exceptions arise whenever there is a need for the normal flow of
program execution to be broken, so that the processor can be
diverted to handle an interrupt from a peripheral
Many exceptions may arise at the same time
When multiple exceptions arise simultaneously, a fixed priority
determines the order in which they are handled
ARM7 handles exceptions by making use of the banked registers
to save state
27/70
ARM7 - Exceptions
Types of Exceptions
FIQ (Fast Interrupt reQuest)
– The FIQ exception is externally generated by taking the nFIQ input LOW.
– This input can accept asynchronous transitions, and is delayed by one
clock cycle for synchronisation before it can affect the processor
execution flow
IRQ (Interrupt ReQuest)
– The IRQ exception is a normal interrupt caused by a LOW level on the
nIRQ input
ABORT
– An ABORT can be signalled by the external ABORT input
– ABORT indicates that the current memory access cannot be completed
28/70
ARM7 - Exceptions
Types of Exceptions
Software interrupt
– The software interrupt instruction (SWI) is used for getting into
Supervisor mode, usually to request a particular supervisor function
29/70
ARM7 - Exceptions
Exception Priorities
30/70
ARM7 - Reset
When the nRESET signal goes LOW, ARM7 abandons the
executing instruction and then continues to fetch instructions
from incrementing word addresses
31/70
ARM7 - Instruction Set
Instruction Set Summary
31 28 19 15 11 7 3 0
32/70
ARM7 - Instruction Set
The Condition Field
31 27 0
Cond
33/70
ARM7 - Instruction Set
31 27 0
Cond
Condition field
– 0000 = EQ - Z set (equal)
– 0001 = NE - Z clear (not equal)
– 0010 = CS - C set (unsigned higher or same)
– 0011 = CC - C clear (unsigned lower)
– 0100 = MI - N set (negative)
– 0101 = PL - N clear (positive or zero)
– 0110 = VS - V set (overflow)
– 0111 = VC - V clear (no overflow)
– 1000 = HI - C set and Z clear (unsigned higher)
– 1001 = LS - C clear or Z set (unsigned lower or same)
– 1010 = GE - N set and V set, or N clear and V clear (greater or equal)
– 1011 = LT - N set and V clear, or N clear and V set (less than)
– 1100 = GT - Z clear, and either N set and V set, or N clear and V clear (greater than)
– 1101 = LE - Z set, or N set and V clear, or N clear and V set (less than or equal)
– 1110 = AL - always
– 1111 = NV - never
34/70
ARM7 - Instruction Set
MUL - Multiply Only
MLA - Multiply and Accumulate
B - Branch
BL - Branch with Link
35/70
ARM7 - Instruction Set
Data Processing
Logical Operations
Assembler
OpCode Action
Mnemonic
AND 0000 operand1 AND operand2
EOR 0001 operand1 EOR operand2
TST 1000 as AND, but result is not written
TEQ 1001 as EOR, but result is not written
ORR 1100 operand1 OR operand2
MOV 1101 operand2 (operand1 is ignored)
BIC 1110 operand1 AND NOT operand2 Bit clear)
MVN 1111 NOT operand2 (operand1 is ignored)
36/70
ARM7 - Instruction Set
Data Processing
Arithmetic Operations
Assembler
OpCode Action
Mnemonic
SUB 0010 operand1-operand2
RSB 0011 operand2-operand1
ADD 0100 operand1+operand2
ADC 0101 operand1+operand2+carry
SBC 0110 operand1-operand2+carry-1
RSC 0111 operand2-operand1+carry-1
CMP 1010 as SUB, but result is not written
CMN 1011 as ADD, but result is not written
37/70
ARM7 - Instruction Set
PSR Transfer (MRS, MSR)
The MRS and MSR instructions are formed from a subset of the
Data Processing operations
They are implemented using the TEQ, TST, CMN and CMP
instructions without the S flag set
These instructions allow access to the CPSR and SPSR registers:
– The MRS instruction allows the contents of the CPSR or
SPSR_<mode> to be moved to a general register
– The MSR instruction allows the contents of a general register to be
moved to the CPSR or SPSR_<mode> register
38/70
ARM7 - Instruction Set
Single data transfer (LDR, STR)
39/70
ARM7 - Instruction Set
Block data transfer (LDM, STM)
Block data transfer instructions are used to load (LDM) or store
(STM) any subset of the currently visible registers
They support all possible stacking modes, maintaining full or
empty stacks which can grow up or down memory, and are very
efficient instructions for saving or restoring context, or for moving
large blocks of data around main memory
40/70
ARM7 - Instruction Set
Single data swap (SWP)
41/70
ARM7 - Instruction Set
Software interrupt (SWI)
42/70
ARM7 - Instruction Set
Coprocessor data operations (CDP)
This class of instruction is used to tell a coprocessor to perform
some internal operation
No result is communicated back to ARM7, and it will not wait for
the operation to complete
43/70
ARM7 - Instruction Set
Coprocessor register transfers (MRC, MCR)
Undefined instruction
44/70
ARM7 - Memory Interface
32 bit bidirectional data bus D[31:0]
32 bit address bus
The nRW signal gives the direction of transfer
Control signals give additional information about the transfer
cycle
DRAM page mode
45/70
ARM7 - Memory Interface
Four categories of transfer cycles
– Non-sequential cycle
• ARM7 requests a transfer to or from an address which is unrelated
to the address used in the preceding cycle
– Sequential cycle
• ARM7 requests a transfer to or from an address which is either the
same as the address in the preceding cycle, or is one word after
the preceding address
– Internal cycle
• ARM7 does not require a transfer, as it is performing an internal
function and no useful prefetching can be performed at the same
time
– Coprocessor register transfer
• ARM7 wishes to use the data bus to communicate with a
coprocessor, but does not require any action by the memory
system 46/70
ARM7 - Memory Interface
nMREQ and SEQ lines define cycle type
These control lines are generated during phase 1 of the cycle
before the cycle whose characteristics they forecast
This pipelining of the control information gives the memory
system sufficient time to decide whether or not it can use a page
mode access
47/70
ARM7 - Memory Interface
ARM Memory Cycle Timing
When an S-cycle follows N-cycle S-cycle I-cycle C-cycle
48/70
ARM7 - Memory Interface
Memory Cycle Optimization
I-cycle S-cycle
When an S-cycle follows an I- MCLK
or C-cycle, the address will
be the same as that used in A[31:0]
the I- or C-cycle
nMREQ
This be used to start the
DRAM access during the SEQ
49/70
ARM7 - Memory Interface
Byte Addressing
The processor address bus gives byte addresses,
Instructions are always words (4 bytes)
The nBW control line is used to request a byte from the memory
system
– normally it is HIGH, signifying a request for a word quantity
– it goes LOW during phase 2 of the preceding cycle to request a byte
transfer
When the processor is fetching an instruction from memory, the
state of the bottom two address lines A[1:0] is undefined
ARM7 will perform the byte extraction internally. Alternatively, the
memory system may activate only the addressed byte of the
memory
50/70
ARM7 - Memory Interface
Decoding Byte Access to Memory
One way of implementing the byte decode in a DRAM system is to
separate the 32 bit wide block of DRAM into four byte wide banks,
and generate the column address strobes independently
A[0] A[1] nBW MCLK CAS
NCAS0
NCAS1
D Q
NCAS2
Quad
Latch NCAS3
51/70
ARM7 - Memory Interface
Little Endian
– operation byte 0 of the memory system should be connected to data
lines 7 through 0 (D[7:0]) and strobed by nCAS0
– nCAS1 drives the bank connected to data lines 15 though 8
Big Endian
– byte 0 of the memory system should be connected to data lines 31
through 24
52/70
ARM7 - Memory Interface
Address Timing
Static RAMs and ROMs require the address to be stable until after
the access has completed
The address transition must be delayed until after the end of
phase 2
An on-chip address latch, controlled by ALE, allows the address
timing to be modified in this way
53/70
ARM7 - Memory Interface
Memory Management
The ARM7 address bus may be processed by an address
translation unit before being presented to the memory
ARM7 is capable of running a virtual memory system
Various signals enable different page protection levels to be
supported:
– nRW can be used by the memory manager to protect pages from being
written to
– nTRANS indicates whether the processor is in user or a privileged
mode, and may be used to protect system pages from the user, or to
support completely separate mappings for the system and the user
– nM[4:0] can give the memory manager full information on the
processor mode.
54/70
ARM7 - Memory Interface
Locked Operations
ARM7 includes a data swap (SWP) instruction that allows the
contents of a memory location to be swapped with the contents of
a processor register
This instruction is implemented as an uninterruptable pair of
accesses:
– the first access reads the contents of the memory
– the second writes the register data to the memory
These accesses must be treated as a contiguous operation by the
memory controller to prevent another device from changing the
affected memory location before the swap is completed
ARM7 drives the LOCK signal HIGH for the duration of the swap
operation to warn the memory controller not to give the memory
to another device
55/70
ARM7 - Memory Interface
Streching Access Times
Long access times can be accommodated by stretching MCLK
It is usual to stretch the LOW period of MCLK
Taking nWAIT LOW has the same effect as stretching the LOW
period of MCLK
nWAIT must only change when MCLK is LOW
There is no limit upon the maximum period for which MCLK may
be stretched, or nWAIT held LOW
56/70
ARM7 - Coprocessor Interface
The ARM7 instruction set may be extended by the addition of up
to 16 external coprocessors
Adding the coprocessor will then increase the system
performance in a software compatible way
When the coprocessor is not present, instructions intended for it
will trap, and suitable software may be installed to emulate its
functions
57/70
ARM7 - Coprocessor Interface
Interface Signals
Three dedicated signals control the coprocessor interface:
– nCPI
– nCPA
– nCPB
The CPA and CPB inputs should be driven high except when they
are being used for handshaking
58/70
ARM7 - Coprocessor Interface
Coprocessor present/absent
ARM7 takes nCPI LOW whenever it starts to execute a coprocessor
(or undefined) instruction
Each coprocessor will have a copy of the instruction
Each coprocessor can inspect the CP# field to see which
coprocessor it is for
Every coprocessor in a system must have a unique number
If that number matches the contents of the CP# field the
coprocessor should drive the CPA (coprocessor absent) line LOW
If no coprocessor has a number which matches the CP# field, CPA
and CPB will remain HIGH, and ARM7 will take the undefined
instruction trap
59/70
ARM7 - Coprocessor Interface
Busy-waiting
If CPA goes LOW, ARM7 will watch the CPB (coprocessor busy)
line
ARM7 will busy-wait while CPB is HIGH, unless an enabled
interrupt occurs, in which case it will break off from the
coprocessor handshake to process the interrupt
When CPB goes LOW, the instruction continues to completion
60/70
ARM7 - Coprocessor Interface
Pipeline Following
In order to respond correctly when a coprocessor instruction
arises, each coprocessor must have a copy of the instruction
All ARM7 instructions are fetched from memory via the main data
bus, and coprocessors are connected to this bus
The nOPC signal indicates when an instruction fetch is taking
place
MCLK gives the timing of the transfer
61/70
ARM7 - Coprocessor Interface
Data Transfer Cycles
Coprocessor must supply or accept data at ARM7 bus rate
(defined by MCLK)
The L bit in the instruction defines direction of transfer
The coprocessor is responsible for determining the number of
words to be transferred
ARM7 will continue to increment the address by one word per
transfer until the coprocessor tells it to stop
The termination condition is indicated by the coprocessor driving
CPA and CPB HIGH
There is no limit to the number of words which one coprocessor
data transfer can move, but no coprocessor should allow more
than 16 words in one instruction
62/70
ARM7 - Coprocessor Interface
Register Transfer Cycle
ARM7 requires the data bus without requiring the memory to be
active
The memory system is informed that the bus is required by ARM7
taking both nMREQ and SEQ HIGH
When the bus is free, DBE should be taken HIGH to allow ARM7 or
the coprocessor to drive the bus, and an MCLK cycle times the
transfer
63/70
ARM7 - Coprocessor Interface
Privileged Instructions
The coprocessor may restrict certain instructions for use in
privileged modes only
64/70
ARM7 - Coprocessor Interface
Idempotency
All instructions may be interrupted at any point up to the time
when the coprocessor goes not-busy
The instruction will normally be restarted from the beginning after
the interrupt has been processed
Any action taken by the coprocessor before it goes not-busy must
be idempotent, ie must be repeatable with identical results
65/70
ARM7 - Coprocessor Interface
Undefined Instructions
66/70
ARM7 - DC Parameters
Absolute Maximum Ratings
67/70
ARM7 - DC Parameters
DC Operating Conditions
68/70
ARM7 - AC Parameters
AC Test Loads
69/70
ARM7 - Backward Compatibility
Two inputs, PROG32 and DATA32, allow one of three
processor configurations to be selected as follows:
– 26 bit program and data space - (PROG32 LOW, DATA32 LOW)
– 26 bit program space and 32 bit data space - (PROG32 LOW,
DATA32 HIGH)
– 32 bit program and data space - (PROG32 HIGH, DATA32 HIGH)
70/70