FA-89605 / Instruction set emulation / Member archive

Operand-size prefix treated as address-size · case 05

Immediates after a 0x66 prefix are decoded as 4 bytes.

Member previewVariant 5 · 3 implementations · 8 checks per implementation

Case contract

Input [code]: a byte list decoded as a toy x86-like ISA. Prefixes 0x66 (operand size 16), 0x67 (address size 16), 0xF0/0xF2/0xF3 in any order. 0x0F escapes to a two-byte opcode. ModRM follows 01/03/89/8B/81/83/0F AF. Immediates: 81, B8-BF, E8, 0F 84 take the operand size (2 or 4 bytes); 83 and EB take 1. 32-bit ModRM: mod 3 none; rm 4 adds a SIB byte and a SIB base of 5 with mod 0 adds disp32; mod 0 rm 5 adds disp32; mod 1 disp8; mod 2 disp32. 16-bit ModRM: mod 0 rm 6 or mod 2 add disp16, mod 1 disp8, no SIB. Return the length of each instruction.

Why this case matters

Emulators and disassemblers must find instruction boundaries exactly; one wrong length desynchronises all following decoding.

One recorded failure

Sample boundary fixture

This sample comes from the broken implementation of a controlled reproducer.

Boundary fixture	Actual	Expected	Outcome
operand size prefix on immediates	`[6, 1, 2, 5]`	`[4, 5, 5]`	Failed

MEMBER ARCHIVE

The complete case is available to members.

This record includes three runnable implementations, regression fixtures, execution results, and source hashes.

Member access is invitation-based. Sign in with your invited account to inspect the sources.