Skip to content

Latest commit

 

History

History
201 lines (133 loc) · 7.59 KB

list.md

File metadata and controls

201 lines (133 loc) · 7.59 KB

UAL Instruction List

Instruction Conditions

All UAL op-codes except directives and pseudo-instructions are can have an execution condition attached, see Instruction Conditions.

Data Processing Instructions

These instructions alter register values and/or status flags based on the values in registers. They all use an op2 operand which can be a registers, like op1 and dest, or various other options. See Flexible Operand 2 for details.

All DP instructions except Compare have an optional S suffix that controls if flags are changed (S) or preserved (no S). Flags are written from the value computed. The value computed is written to dest except in compare instructions.

Flag Value written (if changed)
N dest < 0
Z dest = 0
C Carry out from addition or subtraction
Last bit shifted out from shift or rotate
V Signed overflow from addition or subtraction

Arithmetic Instructions

OpCode dest, op1, op2

Examples

SUB R10, R5, R4 ; set R10 equal to R5 - R4
ADD R1, R1, #1  ; add 1 to R1
RSBS R1, R1, #0 ; negate R1 setting NZCV (R1 = 0 - R1)
OpCode Function Flags changed
if S Suffix
ADD dest := op1 + op2 NZCV
SUB dest := op1 - op2 NZCV
RSB dest := op2 - op1 NZCV
ADC dest := op1 + op2 + C NZCV
SBC dest := op1 - op2 + C - 1 NZCV
RSC dest := op2 - op1 + C - 1 NZCV

Logical Instructions

Opcode dest, op1, op2

Examples

ORR R1, R1, #16 ; sets bit 4 in R1
BICS R10, R10, 0x80000; clear bit 19 of R10, setting falgs NZ
OpCode Function Flags changed
if S Suffix
AND dest := op1 Bitwise AND op2 NZ
ORR dest := op1 Bitwise OR op2 NZ
EOR dest := op1 Bitwise exclusive OR op2 NZ
BIC dest := op1 Bitwise AND NOT op2 NZ

Move Instructions

OpCode dest, op2

Examples

MVN R1, R1   ; inverts bits in R1
MOVS R10, R3 ; load R10 with copy of value in R3, write flags `NZ`.
OpCode Function Flags changed
if S suffix
MOV dest := op2 NZ
MVN dest := Bitwise invert op2 NZ

Shift/Rotate instructions

OpCode dest, op2

Examples

LSLS R0, R1, #4     ; same as MOVS R0, R1, LSL #4
ROR  R1, R1, #2     ; same as MOV R1, R1, ROR #2
RRX  R1, R2         ; same as MOV R1, R2, RRX

These are shorthand for the corresponding MOV with shifted op2.

OpCode Function
LSR Logical Shift Right (shift 0s in)
ASR Arithmetic Shift Right (shift sign bit in)
LSL Logical Shift Left
ROR Rotate Right (shift RHS bit in)
RRX Rotate Right eXtended - always 1 bit

Compare Instructions

OpCode op1, op2

Examples

CMP R0, R1 ; set `NZCV` flags on subtraction R0 - R1
TEQ R3, R4 ; set `NZ` flags on R3 XOR R4
OpCode Function Flags changed
CMP set flags on op1 - op2 NZCV
CMN set flags on op1 + op2 NZCV
TST set flags on op1 Bitwise AND op2 NZ
TEQ set flags on op1 Bitwise XOR op2 NZ

Single Register Memory Transfer Instructions

Examples

LDR R0, [R1,#8] ; load into R0 mem32[R1+8]
STRB R11, [R2,#3]! ; store LS byte R11 in mem8[R2+3], set R2 := R2+3
LDR R0, [R1],#8 ; load into R0 mem32[R1]. then set R1 := R1+8.
LDR R5, =DATA ; load 32 bit immediate value of symbol DATA into R5

See Address Modes for details of EA modes.

OpCode Instruction Function Notes
LDR LDR Rd, EA Rd := mem32[EA] 1
LDRB LDRB Rd, EA Rd := mem8[EA] 2
STR STR Rs, EA mem32[EA] := Rs 1
STRB STRB Rs, EA mem8[EA] := Rs 3
LDR = LDR Rd, =Literal
LDR Rd, =DATA+4
LDR Rd, =1571
Rd := Literal 4
  1. EA must be divisible by 4.
  2. This loads one byte from memory into the LS 8 bits of Rd.
    The MS 24 bits are zeroed.
  3. This writes the LS 8 bits of Rd into the specified memory byte.
    Other bytes in the same memory word are unaffected.
  4. The literal can be any numeric expression of constants and symbols, and does not have a #. Unlike MOV there is no restriction on value. Technically this instruction is a pseudo-instruction, translated by assembler into an LDR with EA offset from PC, and a DCD that defines the the 32 bit constant. Programmers can use it exactly as MOV with literal op2, noting that it is slower than MOV.
Effective Address EA Address Mode
[Rb] Register
[Rb,#N] Register Offset
[Rb,#N]! Register Pre-increment
Rb, #N Register Post-increment
other Other Modes

Details

Multiple Register Memory Transfer Instructions

Examples

LDMED R0, {R1-R3,R8,R10} ; pop from ED stack pointer R0 into R1,R2,R3,R8,R10
STMFA R11,{R1,R12}  ; push R1,R12 onto FA stack pointer R11

The register list (in {}) can contain any distinct set of individual registers or ranges, for 1 - 15 data words transferred. The pointer register must not be in the register list.

The instruction suffix (FD here) indicating stack or tranfer type can be any of FD,FA,ED,EA,IA,IB,DA,DB`. See suffixes page for details of suffixes and specifcation of which memory addresses are used in the transfer.

A ! after the first (stack pointer) register means that this is updated as is needed for a stack operation.

OpCode Instruction Function Notes
LDM LDMFD R10!, {R1,R4-R6} Load registers R1,R4,R5,R6
Stack Pointer = R10
update R10
LDMFD R13, {R1,R10} Load registers R1,R10
Stack pointer = R13
Do not change R13
STM STMFD R8!, {R1,R4-R6} Store registers R1,R4,R5,R5
Stack pointer = R8
Update R8
STMFD R3, {R1,R10,R14} Store registers R1,R10,R11,R12,R13
Stack pointer = R3
Do not change R3

Pseudo-instructions and Directives

OpCode Example Notes
DCD DAT DCD d1, d2,...,dn Set next n data words to numeric values
given by operands d1,...,dn (no #).
Label DAT is usual but not compulsory
DCB DAT1 DCB d1, d2,...,dn Set next n data bytes to numeric values
given by operands d1,...,dn (no #).
n must be divisible by 4.
Bytes are written in little-endian address order.
Label DAT1 is usual but not compulsory.
FILL DAT2 FILL 40 40 can be any literal numeric expression (without #).
Fill next 40 words of data memory with 0
Label DAT2 is usual but not compulsory.
ADR ADR Ra, LABEL Set Ra to LABEL. Same as MOV Ra, #LABEL but
literal values close to PC are allowed.
Useful to load data area label values.
EQU LAB1 EQU LABEL + 4 EQU sets its label, in this case LAB1,
equal to the given numeric expression.
Forward references are allowed.

See also LDR = under Single Register Memory Transfer