011. Lda absolute

File to update

emulator/cpu/cpu.py

Location

CPU.step, new $AD branch

Why this step exists

Immediate LDA loads its operand directly. Absolute LDA instead consumes a two-byte little-endian address, reads the value stored there, and applies the same Zero and Negative flag behavior learned in Test 010.

Complete example implementation

ZERO_FLAG = 1 << 1
NEGATIVE_FLAG = 1 << 7


class CPU:
    def step(self) -> None:
        opcode = self.fetch_byte()

        if opcode == 0xA9:
            self.a = self.fetch_byte()
            if self.a == 0:
                self.p |= ZERO_FLAG
            else:
                self.p &= ~ZERO_FLAG

            if self.a & NEGATIVE_FLAG:
                self.p |= NEGATIVE_FLAG
            else:
                self.p &= ~NEGATIVE_FLAG
            return

        if opcode == 0xAD:
            address = self.fetch_word()
            self.a = self.bus.read(address)

            # In this step, flag behavior is intentionally
            # duplicated before Test 012 extracts the common helper.
            if self.a == 0:
                self.p |= ZERO_FLAG
            else:
                self.p &= ~ZERO_FLAG

            if self.a & NEGATIVE_FLAG:
                self.p |= NEGATIVE_FLAG
            else:
                self.p &= ~NEGATIVE_FLAG
            return

        raise NotImplementedError(
            f"Opcode {opcode:02X} not implemented"
        )

Important invariants

• $AD consumes three instruction bytes: opcode, low address, high address
• the bus read uses the assembled 16-bit address
• both setting and clearing Z/N work across consecutive instructions

Common misconception

The bytes 34 12 are not the value $1234 loaded into A. They form address $1234; LDA then loads the byte stored at that address.

Out of scope

• extracting shared flag logic into a helper
• extracting addressing and instruction functions
• other LDA addressing modes