The crash on purpose

ud2 is an x86 instruction guaranteed to raise an invalid opcode exception. Compilers drop it into unreachable code so a wrong turn crashes loudly instead of running whatever bytes come next.

“an architecturally undefined instruction, guaranteed to raise an "invalid opcode" exception”

Two rival byte sequences

Before Intel blessed anything, programmers hunted for bytes that reliably faulted. One camp found 0F FF, another found 0F B9, and both worked well enough that neither side won.

“you now had two factions, the 0F FF believers and the 0F B9 adherents”

The next chip broke it

Intel changed something on a later processor and 0F FF stopped faulting. Software started failing, and the investigation traced it back to programs depending on an opcode nobody had promised anything about.

Hyrum's Law in silicon

With enough users, every observable behaviour becomes someone's dependency. An accidental crash became a contract Intel had to honour, and 0F B9 turned out to have the same problem.

“With a sufficient number of users, all observable behaviors will be depended upon by somebody.”

Why the name starts at 2

Intel made the behaviour official and retroactively named the old sequences ud0 and ud1. That left ud2 as the recommended one, a clean two-byte instruction with no operands to decode.

Phantom operands bite back

ud0 and ud1 decode a register and a memory operand they never use. If that decoding runs off into a missing page, you get an access violation instead of the invalid opcode you wanted.

“even though the processor doesn't use it, it still decodes it”