💾 Archived View for spam.works › mirrors › textfiles › apple › CRACKING › ysdepro.txt captured on 2023-06-16 at 21:13:32.

View Raw

More Information

-=-=-=-=-=-=-


Crack for Ancient Land of Y's (GS) - Broderbund/Kyodai
by Jim Maricondo (The Mouse2 - AppleLink PE) - January 1990.

After playing (cheating) with the game for awhile, the need for a crack became evident, so I set out to find one.  The type of protection used is the now almost 'standard' nibble count on tracks $20/21 of side one.  The actual nibble count code is on the Y's Data disk, but the nibble count is done on the Y's Program disk.  I began by searching for the equivalent hexidecimal code, A2 20 A0 01 on the Y's Data disk.  I found it on block $4A6.  It looks like this:

06: 6B          RTL                     end of routine before disk check
07: A9 00       LDA #$0000              start of this routine
0A: 8D CA 65    STA $65CA
0D: 8D C6 65    STA $65C6
10: 8D C8 65    STA $65C8
13: E2 30       SEP #$30                make accumulator/index regs 8-bit
15: A2 20       LDX #$20                track 20
17: A0 01       LDY #$01                of side 1
19: 20 4E 65    JSR $654E               go do disk check
1C: B0 13       BCS $31
1E: 8E C6 65    STX $65C6               store half of returned value
21: 8C C7 65    STY $65C7               store other half of returned value
24: A2 21       LDX $21                 track 21
26: A0 01       LDY $1                  of side 1
28: 20 4E 65    JSR $654E               go do disk check
2B: 8E C8 65    STX $65C8               store half of returned value
2E: 8C C9 65    STY $65C9               store other half
31: C2 30       REP #$30                make accumulator/index regs 16-bit
33: 90 0C       BCC $41
35: AD CA 65    LDA $65CA               load A and X regs depending on check
38: 49 FF FF    EOR #$FFFF
3B: 1A          INC
3C: A2 FF FF    LDX #$FFFF
3F: 80 0C       BRA $47
41: A9 C6 65    LDA #$65C6
44: A2 00 00    LDX #$0000
47: 60          RTS                     return to sender

At first I tried just placing a RTS at $07, but that didn't work so I decided to find the routine that called this one by using the PLA PLX BRK technique.  I found out that the disk check routine was called by the routine right before it!  Here's the routine right before it (of which you see the end at $06 of the above example).  The line numbers here have no relation to the ones in the above example.

bank 01-
6D77: 6B         RTL		end of routine before this one
6D78: 8B         PHB
6D79: 5A         PHY
6D7A: 4B         PHK
6D7B: AB         PLB		data bank = program bank
6D7C: AF E1 1E 05  LDA 051EE1
6D80: 48         PHA
6D81: 22 53 6E 01  JSL 016E53 prepare for nibble count (_NewHandle)
6D85: 68         PLA
6D86: 20 94 6D   JSR 6D94     do disk check routine listed above
6D89: 48         PHA          get results from routine
6D8A: DA         PHX          get results from routine
6D8B: 22 E4 6E 01  JSL 016EE4 restore things back to normal (_DisposeAll & _UnloadSeg)
6D8F: FA        PLX
6D90: 68        PLA
6D91: 7A        PLY
6D92: AB        PLB
6D93: 6B        RTL			return to calling routine

--disk check routine described first comes here--

I traced this routine to blocks $4A5-4A6 on Y's Data.  Once again, I attempted to just place a RTL at the beginning of this routine, but that didn't work either so I did Brian's PLA PLX BRK trick again.  This time I was rewarded with the following code.  (This time, these line numbers are relative to the ones in the previous example.)

bank 01-
6DED: 60        RTS			end of subroutine before this one
6DEE: 50 52 4F 47 52 41 4D 00 "PROGRAM" in hex (C-String)
6DF6-6E13: miscellaneous code part of this routine, but not worthy of listing
6E14: 22 78 6D 01  JSL 016D78 go do routine transcribed above

--more code--

I traced this routine to block $4A6 on Y's Data.  This time, I attempted to NOP out the JSL at 6E14.  Again, it didn't work, so there must be another important routine calling this one.  So I used the PLA PLX BRK trick again and was rewarded another time!  I found that the above routine was called from 0E/1363.  Once again, these line numbers are relative to the subroutine above.  Because of the length of this segment, I'm leaving out the hex opcodes.  Here is the final code:

bank 0E-
1363: jsl 016df6		do routine transcribed directly above
1367: stx d2
1369: sta d0
136B: stz e0
136D: lda e0
136F: beq 1374			go process results
1371: brl 1428			ONLY way to get out of this check loop (original disk)
1374: sec				process results
1375: lda d0
1377: sbc #0000
137A: bne 1383
137C: lda d2
137E: sbc #0000
1381: bra 138a
1383: lda d2
1385: sbc #0000
1388: rep #02
138A: bvs 138f
138C: eor #8000
138F: bmi 13b3
1391: pea 0001
1394: pea 0060
1397: pea 0010
139A: pea 0002
139D: pea 33b2
13A0: jsl 0e10fb
13A4: ply
13A5: ply
13A6: ply
13A7: ply
13A8: ply
13A9: jsl 016df6	not original.  prompt for original and check again
13AD: stx d2
13AF: sta d0
13B1: bra 1374
13B3: ldx d2		further process results
13B5: lda d0
13B7: stx de
13B9: sta dc
13BB: ldx de
13BD: lda dc
13BF: phx
13C0: pha
13C1: clc
13C2: adc #0002
13C5: sta dc
13C7: lda de
13C9: bcc 13cc
13CB: inc
13CC: sta de
13CE: pla
13CF: plx
13D0: sta f6
13D2: sts f8
13D4: lda [f6]
13D6: sta da
13D8: lda [dc]
13DA: sta d8
13DC: lda da
13DE: sec
13DF: sbc #206B
13E2: beq 141b
13E4: bvs 13e9
13E6: eor #8000
13E9: bpl 141b
13EB: lda da
13ED: sec
13EE: sbc #2103
13F1: bvs 13f6
13F3: eor #8000
13F6: bmi 141b
13F8: lda d8
13FA: sec
13FB: sbc #1daf
13FE: beq 141b
1400: bvs 1405
1402: eor #8000
1405: bpl 141b
1407: lda d8
1409: sec
140A: sbc #1e79
140D: bvs 1412
140F: eor #8000
1412: bmi 141b
1414: lda #0001
1417: sta e0
1419: bra 1425
141B: ldx #ffff
141E: lda #ffff
1421: stx d2
1423: sta d0
1425: brl 136d		results processed further pass/fail value stored at $E0
1428: pea 0002		Original inserted, so continue loading...
142B: pea 33d7
142E: jsl 106e1c
 
By careful examination, you can tell that there is only one way to get out of the check loop, and that is to have a non-zero result stored at $E0.  The code checks to see if $E0 is zero or not at $136F, and brances to continue loading or continue disk checking depending on the result.  So with this information, I just NOPed out the BEQ 1374 at $136F making the program always continue to load without doing the copy protection at all.  Next I traced this code to block $3BF on Y's Data.  So, after many hours of work, here is the patch!

Block  Bytes   From    To
----------
3BF    70-71   F0 03   EA EA

Or, if using Block Warden, go to relative block 9 of YSGS, relative bytes $1070-$1071.

Or, if that doesn't do it for ya, try scanning for 85 D0 64 E0 A5 E0 F0 03 and changing it to 85 D0 64 E0 A5 E0 EA EA.

You now have a deprotected copy!  Enjoy!  <<Whew!>>  Not bad for a first crack!