#!/usr/bin/env python3 """Assemble a test program, run it on an independent 6502 emulator, run a2c on it, compile the C, and compare final memory states.""" import subprocess, sys, os # ---- tiny assembler (labels + subset) ---- M = [ ("LDX","#",0xA0 if False else 0xA2),("LDY","#",0xA0),("LDA","#",0xA9), ("LDA","abs",0xAD),("LDA","absx",0xBD),("LDA","zp",0xA5), ("STA","abs",0x8D),("STA","absx",0x9D),("STA","zp",0x85), ("ADC","absx",0x7D),("ADC","#",0x69),("SBC","#",0xE9), ("CPX","#",0xE0),("CPY","#",0xC0),("CMP","#",0xC9), ("BEQ","rel",0xF0),("BNE","rel",0xD0),("BCC","rel",0x90),("BCS","rel",0xB0), ("JMP","abs",0x4C),("JSR","abs",0x20),("RTS","",0x60), ("INX","",0xE8),("INY","",0xC8),("DEX","",0xCA),("DEY","",0x88), ("CLC","",0x18),("SEC","",0x38),("TAX","",0xAA),("TXA","",0x8A), ("LSR","a",0x4A),("ASL","a",0x0A),("ROL","a",0x2A), ("NOP","",0xEA), ] def opcodes(): d={} for mn,mode,op in M: d[(mn,mode)]=op return d OPC=opcodes() def parsev(s): s=s.strip() if s.startswith('$'): return int(s[1:],16) if s.startswith('%'): return int(s[1:],2) return int(s,0) def asm(src, org=0x0600): lines=[l.split(';')[0].strip() for l in src.splitlines()] lines=[l for l in lines if l] labels={}; pc=org def size(l): parts=l.split(None,1) mn=parts[0]; rest=parts[1].strip() if len(parts)>1 else '' if rest=='': return 1 if rest=='A': return 1 if mn in ('BEQ','BNE','BCC','BCS'): return 2 if rest.startswith('#'): return 2 if rest.endswith(',X') or rest.endswith(',Y'): try: int(rest[:-2],0); return 3 except: return 3 try: v=parsev(rest); return 2 if v<0x100 else 3 except: return 3 # label pass pc=org; ordered=[] for l in lines: if l.endswith(':'): labels[l[:-1]]=pc else: ordered.append((pc,l)); pc+=size(l) out=bytearray() for addr,l in ordered: parts=l.split(None,1) mn=parts[0]; rest=parts[1].strip() if len(parts)>1 else '' if rest=='': out.append(OPC[(mn,'')]) elif rest=='A': out.append(OPC[(mn,'a')]) elif rest.startswith('#'): out.append(OPC[(mn,'#')]); out.append(parsev(rest[1:])&0xFF) elif mn=='JMP' or mn=='JSR': t=labels[rest] if rest in labels else parsev(rest) out.append(OPC[(mn,'abs')]); out+=bytes([t&0xFF,(t>>8)&0xFF]) elif rest.endswith(',X') or rest.endswith(',Y'): base=rest[:-2] t=labels[base] if base in labels else parsev(base) suffix='absx' if rest.endswith(',X') else 'abs' if mn=='LDA': out.append(OPC[(mn,'absx')]) elif mn=='STA': out.append(OPC[('STA','absx')]) elif mn=='ADC': out.append(OPC[('ADC','absx')]) else: raise Exception('bad '+l) out+=bytes([t&0xFF,(t>>8)&0xFF]) elif mn in ('BEQ','BNE','BCC','BCS'): t=labels[rest] if rest in labels else parsev(rest) off=t-(addr+2) assert -128<=off<127, (l,off) out.append(OPC[(mn,'rel')]); out.append(off&0xFF) else: t=labels[rest] if rest in labels else parsev(rest) if t<0x100: out.append(OPC[(mn,'zp')]); out.append(t&0xFF) else: out.append(OPC[(mn,'abs')]); out+=bytes([t&0xFF,(t>>8)&0xFF]) return bytes(out), labels # ---- independent emulator ---- def emulate(blob, org, entry, steps=200000): m=[0]*65536 for i,b in enumerate(blob): m[org+i]=b a=x=y=0; sp=0xFD; c=z=n=v=0 pc=entry; st=[] def setzn(r): return (0 if r==0 else 1), ((r>>7)&1) for _ in range(steps): op=m[pc] def rd(k=1): return [m[(pc+i)&0xFFFF] for i in range(0,k+1)] if op==0xA9: a=rd()[1]; z,n=setzn(a); pc+=2 elif op==0xA2: x=rd()[1]; z,n=setzn(x); pc+=2 elif op==0xA0: y=rd()[1]; z,n=setzn(y); pc+=2 elif op==0xAD: ad=rd(3)[1]|(rd(3)[2]<<8); a=m[ad]; z,n=setzn(a); pc+=3 elif op==0xBD: b2=rd(3); ad=(b2[1]|(b2[2]<<8)); ad=(ad+x)&0xFFFF; a=m[ad]; z,n=setzn(a); pc+=3 elif op==0x8D: b2=rd(3); ad=b2[1]|(b2[2]<<8); m[ad]=a; pc+=3 elif op==0x9D: b2=rd(3); ad=((b2[1]|(b2[2]<<8))+x)&0xFFFF; m[ad]=a; pc+=3 elif op==0x85: m[rd()[1]]=a; pc+=2 elif op==0x7D: b2=rd(3); ad=(b2[1]|(b2[2]<<8)+x)&0xFFFF; r=m[ad] t=a+r+c; c=1 if t>0xFF else 0 nv=1 if ((a^t)&(r^t)&0x80) else 0; a=t&0xFF; z,n=setzn(a); v=nv; pc+=3 elif op==0x69: r=rd()[1]; t=a+r+c; c=1 if t>0xFF else 0 nv=1 if ((a^t)&(r^t)&0x80) else 0; a=t&0xFF; z,n=setzn(a); v=nv; pc+=2 elif op==0xE9: r=rd()[1]; t=a-(r+(1-c)); c=1 if t<0x100 else 0 nv=1 if ((a^r)&(a^t)&0x80) else 0; a=t&0xFF; z,n=setzn(a); v=nv; pc+=2 elif op==0xE0: r=rd()[1]; t=x-r; c=1 if t<0x100 else 0; z,n=setzn(t&0xFF); pc+=2 elif op==0xC9: r=rd()[1]; t=a-r; c=1 if t<0x100 else 0; z,n=setzn(t&0xFF); pc+=2 elif op==0xF0: off=rd()[1]; off=off-256 if off>127 else off pc = pc+2+off if z==1 else pc+2 elif op==0xD0: off=rd()[1]; off=off-256 if off>127 else off pc = pc+2+off if z==0 else pc+2 elif op==0x90: off=rd()[1]; off=off-256 if off>127 else off pc = pc+2+off if c==0 else pc+2 elif op==0xB0: off=rd()[1]; off=off-256 if off>127 else off pc = pc+2+off if c==1 else pc+2 elif op==0x4C: pc = rd(3)[1]|(rd(3)[2]<<8) elif op==0x20: b2=rd(3); st.append(pc+2); pc=b2[1]|(b2[2]<<8) elif op==0x60: if not st: break pc=(st.pop()+1)&0xFFFF elif op==0xE8: x=(x+1)&0xFF; z,n=setzn(x); pc+=1 elif op==0xC8: y=(y+1)&0xFF; z,n=setzn(y); pc+=1 elif op==0xCA: x=(x-1)&0xFF; z,n=setzn(x); pc+=1 elif op==0x18: c=0; pc+=1 elif op==0x38: c=1; pc+=1 elif op==0xAA: x=a; z,n=setzn(x); pc+=1 elif op==0x8A: a=x; z,n=setzn(a); pc+=1 elif op==0x4A: c=a&1; a=a>>1; z,n=setzn(a); pc+=1 elif op==0x0A: c=1 if a&0x80 else 0; a=(a<<1)&0xFF; z,n=setzn(a); pc+=1 elif op==0xEA: pc+=1 else: raise Exception("emu: unknown opcode %02X at %04X"%(op,pc)) print("TRACE op=%02X -> pc=%04X a=%02X x=%02X z=%d c=%d"%(op,pc,a,x,z,c)) return m SRC = """ start: LDX #$00 LDY #$00 LDA #$00 STA $0200 LDA #$01 STA $0201 loop: CPX #$0A BEQ done CLC LDA $0200,X ADC $0201,X STA $0202,X INX JMP loop done: JSR checksum RTS checksum: LDA #$00 LDX #$00 loop2: CPX #$0C BEQ end2 CLC ADC $0200,X INX JMP loop2 end2: STA $00 RTS """ def main(): blob,labels=asm(SRC,0x0600) open("test.bin","wb").write(blob) print("assembled %d bytes, labels=%s"%(len(blob),{k:hex(v) for k,v in labels.items()})) m=emulate(blob,0x0600,labels['start']) ref={i:m[i] for i in range(65536) if m[i] and not (0x0600<=i<0x0600+len(blob))} open("ref.txt","w").write("".join("%04X %02X\n"%(i,ref[i]) for i in sorted(ref))) print("ref nonzero bytes: %d, checksum cell $00 = %d"%(len(ref),ref.get(0,0))) r=subprocess.run([sys.executable,"a2c.py","test.bin","0600","0600","gen.c"],capture_output=True,text=True) print(r.stdout,r.stderr) r=subprocess.run(["gcc","-O2","-o","gen","gen.c"],capture_output=True,text=True) if r.returncode: print("GCC FAIL:",r.stderr[:4000]); sys.exit(1) got={} for line in subprocess.run(["./gen"],capture_output=True,text=True).stdout.splitlines(): p=line.split(); got[int(p[0],16)]=int(p[1],16) if got==ref: print("MATCH: transpiled C is memory-state equivalent (%d nonzero bytes)"%len(ref)) else: only_ref={k:ref[k] for k in ref if k not in got or got[k]!=ref[k]} only_got={k:got[k] for k in got if k not in ref or ref[k]!=got[k]} print("MISMATCH\n ref-only/diff:",{hex(k):v for k,v in list(only_ref.items())[:20]}, "\n gen-only/diff:",{hex(k):v for k,v in list(only_got.items())[:20]}) sys.exit(2) blob,labels=asm(SRC,0x0600) emulate(blob,0x600,labels["start"],steps=14)