;*--------------------------------------------------------------------------- ; :Program. World of Wonders ; :Contents. "California Games" Cracktro ; :Author. Stefano "Max Headroom/TEX" Pucino ; :Requires. WoW-Cracktro (Bytekiller packed) ; :Copyright. Open Source (like GNU) ; :Language. 68000 Assembler ; :Translator. ASM-One v1.21 / Barfly v2.00 ; :To Do. ; ; :History. ; 24.09.2001 - Version 0.1 ; NEW: Completely rewritten code. ; Easier to maintain now ;) ; ; 21.10.2001 - Version 0.2 ; NEW: Removed empty DBRA from ST-replayer ; ; 03.10.2001 - Version 1.0 ; NEW: Fixed Copperlist Modulo of scroller ; NEW: First AmiNET Release ; ; 06.10.2001 - Version 1.1 ; UPD: Source cleaned up (again) ; UPD: Uses the AmigaScene-executable now. ;---------------------------------------------------------------------------* ;====================================================================== ; Load all includes and macros ;====================================================================== INCDIR INCLUDE: INCLUDE whdload.i INCLUDE whdmacros.i INCLUDE lvo/exec_lib.i INCLUDE lvo/dos_lib.i ;====================================================================== ; Special BarFly options and optimisations ;====================================================================== IFD BARFLY OUTPUT "WorldOfWonders.slave" ; Name of the slave BOPT O+ OG+ ; Enable optimizing BOPT ODd- ODe- ; Disable mul optimizing BOPT w4- ; Disable 64k warnings SUPER ; Disable supervisor warnings ENDC ;====================================================================== _base SLAVE_HEADER ; ws_Security + ws_ID dc.w 14 ; WHDLoad version needed dc.w WHDLF_NoError ; Flags dc.l $80000 ; BaseMem Size (512 KB) dc.l $0 ; Exec Install dc.w _Start-_base ; Introloader dc.w 0 ; Current Dir dc.w 0 ; Don't Cache _keydebug dc.b $5f ; DebugKey = HELP _keyexit dc.b $5d ; Exit Key = PrtScr _expmem dc.l 0 ; ExpMem (No Fast-Mem) dc.w _name-_base ; Name of file dc.w _copy-_base ; Copyright dc.w _info-_base ; Additional informations ;====================================================================== ; The description part ; Strings are zero terminated. 10 means Carricage Return (CR) ;====================================================================== _name ; Full name of the game dc.b "California Games Cracktro",0 _copy ; Copyright information dc.b "World of Wonders",0 _info ; Who am I ? ;) dc.b "--------Installed by:--------",10 dc.b "Max Headroom",10 dc.b "of",10 dc.b "The Exterminators",10 dc.b "-----------------------------",10 dc.b "Version 1.1 " ; Installer-version IFD BARFLY IFND .passchk DOSCMD "WDate >T:date" INCBIN "T:date" ; Include current date of slave .passchk ENDC ENDC dc.b 0 ; End this string even ;====================================================================== ; The magic part. Now we start the slave ;) ;====================================================================== _Start ; A0 = resident loader ; This routine simply loads the empty variable '_resload' to a1 and ; puts the resident-loader (location=A0) to it. lea _resload(pc),a1 ; Get Slave-base move.l a0,(a1) ; Save for later use move.l a0,a2 ; A2 = resload, too ; Now we enable the Instruction-Cache but disable the Data-Cache at the ; same time. This gives us a very huge compatibility. move.l #CACRF_EnableI,d0 ; Enable CPU Instruction-Cache move.l d0,d1 ; Mask jsr (resload_SetCACR,a0) ; WHD SetCache() ;====================================================================== ; Load the OSEmu module ;====================================================================== lea _OSEmu(pc),a0 ; file name lea $400.w,a1 ; A3 = OSEmu base address jsr (resload_LoadFileDecrunch,a2) ; Load and decrunch OSEmu file ; Initialize OSEmu move.l a2,a0 ; resload lea _base(pc),a1 ; slave structure jsr $400.w ; Start it move.w #0,sr ; switch to user mode ; Open dos.library to provide file loading function moveq.l #0,d0 ; Clear d0 (=lib-version not important) lea _dosname(pc),a1 ; load dos.library move.l $4.w,a6 ; Get execbase to a6 jsr _LVOOpenLibrary(a6) ; Open the library lea.l _dospointer(pc),a4 ; Load dos-pointer address to a4 move.l d0,(a4) ; Put dos-pointer to a4 move.l d0,a6 ; Put it also to a6 ;====================================================================== ; Clear Memory ;====================================================================== ; This isn't important, so you can also leave this out. But for ; a nice and real clean memory we will "simulate" a software reset ; by clearing the memory from $10000 to $80000. lea.l $10000,a0 ; Get Start into a0 move.l #$6fffff,d0 ; How many bytes to clear ? .clearloop clr.l (a0)+ ; Clear a0 ($10000) dbra d0,.clearloop ; Continue until d0 is -1 ;====================================================================== ; Load the intro via dos.library LoadSeg() function ;====================================================================== lea _exe(pc),A0 ; Get executable-filename move.l a0,d1 ; And copy it to d0 also jsr _LVOLoadSeg(A6) ; Load the file now. lsl.l #2,d0 ; Left-scroll result by 2 move.l d0,a1 ; Copy result to a1 addq.l #4,a1 ; Add 4 to it sub.l a0,a0 ; Clear a0 now moveq.l #0,d0 ; No pointer on argumentline lea _introstart(pc),a2 ; Get variable move.l a1,(a2) ; Save start-address of intro ;====================================================================== ; The now loaded and relocated executable will be fixed here. ;====================================================================== jsr _patchdecruncher(pc) ; Jump to the fixing-routine move.l _introstart(pc),a1 ; Source-Address lea.l _arg(pc),a0 ; Argument line moveq.l #1,d0 ; One argument jsr (a1) ; Start the decruncher jmp _exitintro(pc) ; Jump to EXIT-OF-INTRO _patchdecruncher: JMPADR = $aa ; Relative address of Intro-JUMP move.l _introstart(pc),a1 ; Source-Address move.w #$4ef9,JMPADR(a1) ; Poke JMP to destination pea (_fix,pc) ; Put Fix-List on Stack move.l (a7)+,(JMPADR+$2,a1) ; Write Stack to destination rts _fix ; First fix the soundtracker-bug INTROSTART = $39e20 ; Start-address of the intro BUGADR = INTROSTART+$39b3e ; Address of the SoundTracker-loop move.w #$4eb9,BUGADR ; Poke JSR to destination pea (_dbra,pc) ; Put SoundTracker-fix on Stack move.l (a7)+,BUGADR+$2 ; Write Stack to destination move.w #$4e71,BUGADR+$6 ; Add NOP-word after the call ; If you will start the intro now, you will notice that the scroller and ; the lower background-bitplane are misplaced. It looks like they have been ; shifted. What ? You don't understand wha I'm writing =??= ; ; Imagine a vertical line that you draw on a piece of paper. ; You start from the bottom to the top and at the end you get a beatiful ; line ;) ; Now rotate the paper in front of you by about 45 degrees to the right. The ; line changes it's start/end position. It starts now at the lower left ; corner and ends at the upper right one. ; ; correct modulo: modulo too high: ; ; ######### ######### ; # # # # ; # | # # | # ; # | # # | # ; # | # # | # ; # | # # | # ; # # # # ; ######### ######### ; ; The "degree" of the bitplanes are controlled by the MODULO. It's done by ; setting $dff108 (BPL1MOD) and $dff10a (BPL2MOD). ; $dff108 controls the odd planes (bitplane 1,3,5) and the $dff10a the even ; ones (2,4,6). ; These registers are usually set in the copperlist. You can disassemble ; the source and look for the CList or you can use a monitor (AR3, HRTMon) ; to directly look at it. ; I took the AR3-method and typed "c" to display the copperlist. ; Look out for $108 and $10a and you will find the values inside them. ; Now it's the easy part: Change it and exit the monitor. ; If you've found the right modulo of the bitplane, write down the value. ; Take the disassembled source and look out for the copperlist. ; Find lines that contains datas like $e0, $e2, $e4 (Bitplane pointers) ; and/or $180, $182, $184 (colors) and you will probably find them fast. ; If the source won't set the values automatically, you can directly poke ; the new values inside the copperlist. The intro will take it and ; initialize them correctly. ; Else you have to find the routine that sets the desired registers and ; change them by yourself. ; In this case I found...: ; ; 03903E dc.w $E6,0,$E8,0,$EA,0,$EC,0,$EE,0,$F0,0,$F2,0,$108,0 ; 03905E dc.w $10A,0,$FFFF,$FFFF ; ; The wanted addresses are.... ; ; 03905A dc.w $108,$0000 ; 03905E dc.w $10A,$0000 ; ; As you see, the original values are "0". I will rotate the bitplane to ; the left by substracting the value by 2. The result is a readable ; scrolltext. ; ; I directly poke the addresses with the following commands... move.w #$0000-$2,INTROSTART+$3905c ; Substract 2 from the old value move.w #$0000-$2,INTROSTART+$39060 ; Substract 2 from the old value jmp $71b6c ; Start the Intro now ;====================================================================== ; Exit slave ;====================================================================== _exitintro ; We put the "OK" reason to the stack and load the resident-base to a0. ; Then we abort the whole show giving WHD the reason. pea TDREASON_OK ; Everything went O.K. move.l (_resload,pc),a0 ; Put base to a0 for use jmp (resload_Abort,a0) ; Exit the slave ;====================================================================== ; FIX Routines ;====================================================================== ;====================================================================== ; We have to fix an empty DBRA-loop at $2e27e ! ; It's the well-known SoundTracker bug ;) ; But for a better understanding of WHAT a _dbra_ loop is, we will step ; back in time... ; ; Remember old BASIC days ? ... ; ; 10 FOR a = 0 TO 10000 ; 20 NEXT a ; 30 PRINT "Some time has passed" ; ; This will count up the variable "a" from 0 to 10.000 and print out a text. ; On the C64 you could be sure that this count-up would take a specific ; time. And this time is the same on ALL C64 computers sold. ; This also happened on the Amiga. On the old 7.14 MHz CPU you could be ; sure that such a loop will do the work on every Amiga sold. ; But there is a "small" problem... How about a faster processor ? >7 MHz ? ; The loop will reach the end MUCH faster than previsted and the routine ; will be f+cked up faster than you can think of. So we have to replace ; this good-ol' bad habit with a more real one. ; ; This is the ASM-pendant of the BASIC example: ; ; wait_now: ; move.l #$12c,d0 ; Ammount of time to wait ; dbra_loop: ; dbra d0,dbra_loop ; Decrease d0 and branch again to loop ; if not zero ! ; ; On FAST processors (>7,14 MHz) this will surely cause problems. In this ; case, the SoundTracker-player will hick and skip some samples from time ; to time. ;) ; ; This will be replaced by a more nice vertical-beam-wait routine... ;====================================================================== _dbra: DBRALOOPS = 10 movem.l d0-d7/a0-a6,-(a7) ; Save registers move.w #DBRALOOPS,d1 ; How many loop-runs to run ? .1 move.b ($dff006),d0 ; Get beam-position from VHPOSR .2 cmp.b ($dff006),d0 ; Compare actual position with d0 beq.b .2 ; If found, test again (old bug) dbf d1,.1 ; Repeat the loop movem.l (a7)+,d0-d7/a0-a6 ; Restore registers rts ; And return to old code _OSEmu dc.b 'OSEmu.400',0 ; Name of the OSEmu module to emulate DOS functions _exe dc.b "wow-californiagames",0 ; Name of the intro-executable _resload dc.l 0 ; Address of resident loader _dospointer dc.l 0 ; DOS.library base pointer _introstart dc.l 0 ; Start-address after LoadSeg() _arg dc.b $a ; Argument-line (= Empty) _dosname dc.b 'dos.library',0 ; DOS.library name even