summaryrefslogtreecommitdiffstatshomepage
path: root/src/devices/cpu/m6502/xavix2000.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'src/devices/cpu/m6502/xavix2000.cpp')
-rw-r--r--src/devices/cpu/m6502/xavix2000.cpp93
1 files changed, 54 insertions, 39 deletions
diff --git a/src/devices/cpu/m6502/xavix2000.cpp b/src/devices/cpu/m6502/xavix2000.cpp
index 34f41a483b6..17c20b3376d 100644
--- a/src/devices/cpu/m6502/xavix2000.cpp
+++ b/src/devices/cpu/m6502/xavix2000.cpp
@@ -25,50 +25,65 @@
some of the documented but 'undocumented' opcodes on the 6502 to have additonal
meaning for one specific function on startup
- 01BC37: A0 3F ldy #$3f
- 01BC39: B2 ??
- 01BC3A: 1B ??
- 9B ??
- -- outer loop
- 01BC3C: 98 tya
- 01BC3D: 5B (has to be 1 byte?)
- -- inner loop
- 01BC3E: A3 (has to be 1 byte)
- 73 (could be 3 byte?)
- 01BC40: 83 FB presumably fb is the address
- 01BC42: A7 (presumably 1 byte)
- 73 (could be 3 byte?)
- 01BC44: 87 FB presumably fb is the address
- 01BC46: D0 F6 bne $1bc3e
- 01BC48: 88 dey
- 01BC49: 10 F1 bpl $1bc3c
- 01BC4B: 8D FB 00 sta $00fb
- 01BC4E: A3 ??
- 01BC4F: 8D FA 00 sta $00fa
- 01BC52: 07 ??
- --
- 01BC53: D0 03 bne $1bc58
- 01BC55: CE fa 00 dec $00fa
- 01BC58: 80 retf
+ this just seems to be some very buggy checksum code where the games don't even care about the result...
+ (actually it looks like some games might be using it as a PRNG seed?)
+
+ a is 80 when entering here?
+
+ 01BC37: A0 3F ldy #$3f
+ 01BC39: B2 clrl // clear 32-bit 'long' register (part of it might be the accumulator?) (or clear accumulator?)
+ 01BC3A: 1B spa0 a // store 'accumulator' into byte 0 of PA 'address' register
+ 01BC3B: 9B spa2 a // store 'accumulator' into byte 2 of PA 'address' register
+
+ -- loop point 2
+ 01BC3C: 98 tya // y -> a (3f on first run of loop)
+ 01BC3D: 5B spa1 a // store 'accumulator' into byte 1 of PA 'address' register (803f80 on first loop? 003f00 if we clear accumulator with clrl instead) maybe should be 3f0000 ? (would make more sense for a ram test)
+
+ -- loop point 1
+ 01BC3E: A3 ldal0 a // read byte 0 of 32-bit 'long' register into accumulator
+ 01BC3F: 73 adcpa // adc ($Address PA), y
+ 01BC40: 83 stal0 a // store accumulator back in byte 0 of 32-bit 'long' register (even byte checksum?)
+ 01BC41: FB incpa // increase 'address' register PA
+ 01BC42: A7 ldal1 a // read byte 1 of 32-bit 'long' register into accumulator
+ 01BC43: 73 adcpa // adc ($Address PA), y
+ 01BC44: 87 stal1 a // store accumulator back in byte 0 of 32-bit 'long' register (odd byte checksum?)
+ 01BC45: FB incpa // increase 'address' register PA
+ 01BC46: D0 F6 bne $1bc3e // (branch based on PA increase, so PA must set flags?, probably overflows after 0xffff if upper byte is 'bank'? or at 0xff if this really is a mirror of the function below
+
+ 01BC48: 88 dey // decrease y, which contained 3f at the start
+ 01BC49: 10 F1 bpl $1bc3c // branch back to loop point 2 to reload counter
+
+ // contains the odd byte checksum once we drop out the loop
+ 01BC4B: 8D FB 00 sta $00fb // store it in zero page memory
+ 01BC4E: A3 ldal0 a // get the even byte checksum from byte 0 of 32-bit 'long' register
+ 01BC4F: 8D FA 00 sta $00fa // store it in zero page memory
+ 01BC52: 07 oral1 a // why do we want to do this? (routine below does it too)
+ 01BC53: D0 03 bne $1bc58
+ 01BC55: CE FA 00 dec $00fa
+ 01BC58: 80 retf
this is presumably meant to be similar to the function found in Namco
Nostalgia 2
+ // A is 80 on entry
09FFD8: A9 3F lda #$3f
- 09FFDA: 85 01 sta $01
- 09FFDC: A0 00 ldy #$00
- 09FFDE: AD FA 00 lda $00fa
- 09FFE1: 71 00 adc ($00), y
- 09FFE3: 8D FA 00 sta $00fa
- 09FFE6: C8 iny
- 09FFE7: AD FB 00 lda $00fb
- 09FFEA: 71 00 adc ($00), y
- 09FFEC: 8D FB 00 sta $00fb
- 09FFEF: C8 iny
- 09FFF0: D0 EC bne $c
- 09FFF2: C6 01 dec $01
- 09FFF4: 10 E8 bpl $9ffde
- 09FFF6: 0D FA 00 ora $00fa
+ 09FFDA: 85 01 sta $01 // set upper bit of pointer at 0x0000 to 3f (it completely fails to initialize the ram at 00, we currently init it to ff, but should probably be 0?)
+ 09FFDC: A0 00 ldy #$00 // clear inner loop counter
+ -- loop point 1 and 2
+ 09FFDE: AD FA 00 lda $00fa // read current even byte checksum value from zero page ram fa (game also completely fails to initialize this)
+ 09FFE1: 71 00 adc ($00), y // so 3f00 + y first outer loop, 3e00 + y second outer loop with y increasing in inner loop (add byte read to even checksum byte?)
+ 09FFE3: 8D FA 00 sta $00fa // store even checksum at 0xfa in zero page ram
+ 09FFE6: C8 iny // increase counter y
+ 09FFE7: AD FB 00 lda $00fb // read current odd byte checksum value from zero page ram fb (likewise never initialized!)
+ 09FFEA: 71 00 adc ($00), y // 3f00 + y (add byte read to odd checksum byte)
+ 09FFEC: 8D FB 00 sta $00fb // store odd checksum byte at 0xfb in zero page ram
+ 09FFEF: C8 iny // increase y
+ 09FFF0: D0 EC bne $9ffde // branch back if y hasn't overflowed (so work on 0x100 bytes, looping back 0x7f times)
+ 09FFF2: C6 01 dec $01 // decrease accumulator (started at 3f) (inner loop counter already 0 because it overflowed)
+ 09FFF4: 10 E8 bpl $9ffde // branch back
+
+ // checksums are already in place after loop so no need to copy them like in above
+ 09FFF6: 0D FA 00 ora $00fa // same weird 'or' call
--
09FFF9: D0 03 bne $9fffe
09FFFB: CE FA 00 dec $00fa