Binary arm-thumb-1 cheat sheet

Warning

#

This is a work in progress which I publish mostly for my own reference and there may be errors and typos.

Please refer to the source of truth if you want to use thumb-1 and pardon my english.

Context

#

These days when I have time I’m playing ScratchOS and I’m currently at the nibble stage where I can only write code via GPIOs and have the firmware execute it.

To move forward I need to write a few simple programs to start taking over from here.

The challenge is that I can’t use a computer to translate text assembly to hex assembly, and that requires me to get a better grip of the thumb assembly encoding.

This page is intended as a reference for future me or anyone playing masochist-mode-bringup.

Source of truth

#

All the encodings listed here can be found in the ARM Thumb 1 spec.

Instructions.

#

PC-relative branch.

#

Branch syntax :

branch behavior : takes PC of instruction, adds 4, sign extends 2 * X to 32 bits, adds it, and sets PC to the result.

branch forward :

branch backward :

deadloop : branch with offset of -4 -> X = -2 = (~0b10 + 1 = 0b111 1111 1101 + 1 = 0b111 1111 1110)

PC-relative load :

#

Generic case :

Operands :

• d : dest register.
• o : offset

Behavior :

• take v = (PC of instruction + 4) & ~(3);
• take o and multiply it by 4.
• load at v + o, write result in dest reg.

Particular cases :

Load at N words after PC (N > 0) after PC into r0 :

Load four bytes after PC into r0 :

Load eight bytes after PC into r0 :

And so on.

Operations on high registers.

#

Encoding :

Operands :

• o : op : 00 = add, 01 = cmp, 10 = mov, 11 = bx.
• S : high flag for source.
• D : high flag for dest.
• sss : source register
• ddd : dest register

high register move syntax

special registers ;

• sp = r13, 1 101
• lr = r14, 1 110
• pc = r15, 1 111

mov r0, pc = mov 0000 1111 :

mov sp, r0 = mov 1101 0000 :

bx <s>

bx r0

Move/compare/add/subtract immediate syntax :

#

Operands :

• o : operation : 00 mov 01 cmp 10 add 11 sub
• d : dest low reg.
• i : offset.

addition syntax :

which makes a nice lil 3 [reg] offset mnemonic.

Add offset to r0 : 3 0 offset.

push pop

#

Operands :

• m : 0 store 1 load
• r : 0 don’t store LR / load PC, 1 store LR / load PC
• l : register mask. bit : 0 not pushed / popped, 1 pushed/popped.

push (regular) :

pop (regular)

pop {r0, r1}

load-store (immediate offset) syntax :

#

Operands :

• s : 0 word, 1 byte
• l : 0 store, 1 load
• o : base offset
• b : base register
• d : data register (source or dest).

word store with no offset :

word load with no offset :

str r1, [r0] :

generally, store any register at address of r0 :

generally, load any register at address contained in r0 :