This document outlines the bugs discovered in the code during the disassembly process.
When animating the player falling into water, a comparison was missed which
causes sprite_18
to not be used for the animation sequence. This bug also
exists in the Japanese version as well as Probotector. It's clear from the
code that there was an attempt to show sprite_18
(frame 3 in the table below)
as part of the animation sequence. sprite_18
is the same sprite as when the
player is in water and presses the d-pad down button.
Frame 1 | Frame 2 | Frame 3 (missing) | Frame 4 |
---|---|---|---|
@set_enter_water_sprite:
...
lda PLAYER_WATER_TIMER,x ; load water animation timer
beq ... ; if timer has elapsed, branch
cmp #$0c ; see if timer is greater than or equal to #$0c
bcs ... ; branch if timer >= #$0c (keep current sprite)
lda #$73 ; PLAYER_WATER_TIMER less than #$0c, update sprite
sta PLAYER_SPRITE_CODE,x ; set player sprite (sprite_73) water splash
cmp #$08 ; !(BUG?) always branch, doesn't compare to PLAYER_WATER_TIMER
; but instead compares against #$73
; no lda PLAYER_WATER_TIMER,x before this line, so part of splash animation is missing
bcs ... ; branch using sprite_73
lda #$18 ; dead code - a = #$18 (sprite_18) water splash/puddle
sta PLAYER_SPRITE_CODE,x ; dead code - set player animation frame to water splash/puddle
After defeating the boss UFO (Guldaf), before calling the method
level_boss_defeated
, the developers forgot to load into the accumulator the
appropriate sound code, usually #$57 for sound_57
. So bytes in bank 7 are
interpreted as an audio sample incorrectly and a short DMC channel audio clip is
played.
level_boss_defeated
uses whatever is in the accumulator as the sound code to
play. Since the accumulator wasn't set, it has the previous value of #$ff (see
boss_ufo_routine_0b
). This is interpreted by the audio engine as a DMC sound
code that doesn't exist.
The audio engine will incorrectly read bytes from sound_table_00
as the dmc
data below.
- sampling rate #$03 (5593.04 Hz), no loop
- counter length #$77
- offset #$97 --> $c000 + (#$97 * #$40) --> $e5c0
- sample length #$19
The data that is interpreted as a DPCM-encoded DMC sample is $e5c0, which is
in bank 7's collision_box_codes_03
data. This bug does not exist in the
Japanese version of the game because the level_boss_defeated
is not
responsible for playing the boss defeated sound. This bug does exist in
Probotector.
Extracted sound sample: sound_ff.mp3
Note that the boss defeated audio (sound_55
) is still played because the enemy
defeated routine is set to boss_ufo_routine_09
(see
enemy_destroyed_routine_05
). boss_ufo_routine_09
plays sound_55
.
Under normal circumstances, when the level 4 indoor/base boss gemini helmet
(enemy type = #$1c) splits into 'phantoms', then they don't take damage. Only
when the helmet re-combines is it vulnerable to damage. However, Mr. K
researched a glitch to find that
if a player's bullet collides with the helmet at just the right frame, then when
the helmet separates into two, it can still take damage.
The reason this happens is because the boss gemini uses ENEMY_ANIMATION_DELAY
to know when to be vulnerable or when to be invulnerable.
ENEMY_ANIMATION_DELAY
specifies how long for the helmet to stay still when
merged, or when at the farthest distance apart. The timer is set to #$20 when
farthest apart, and #$30 when merged. If the helmets are moving (either toward
each other or away), the value will be #$00.
When merged and ENEMY_ANIMATION_DELAY
is 2, the helmet is not moving, but
about to start separating. The value will be decremented to 1. This logic
happens in boss_gemini_routine_02
. After boss_gemini_routine_02
runs,
bullet_enemy_collision_test
is executed to check for a bullet to enemy
collision. If during this frame, a bullet hits the gemini, then the
ENEMY_ROUTINE
is updated to boss_gemini_routine_03
. This is known as the
enemy destroyed routine and it will be executed in the next frame. However,
boss gemini is special, in that it isn't automatically destroyed in the
destroyed routine. Instead, unless boss gemini doesn't have any more health
(ENEMY_VAR_4
), the routine will decrement ENEMY_ANIMATION_DELAY
to 0 and
set back to boss_gemini_routine_02
to be called the next frame.
Now when the next frame executes and the boss_gemini_routine_02
routine is
run, ENEMY_ANIMATION_DELAY
is 0 and game thinks that the code that makes the
helmet invulnerable has already been executed when it hasn't!
In short, the bug happens because for the special time when
ENEMY_ANIMATION_DELAY
is decremented from 1 to 0, the code should make the
helmet invincible by calling disable_enemy_collision
. However, if you time
the bullet collision to happen on the frame when ENEMY_ANIMATION_DELAY
is 2,
then the regular game code will decrement the timer to 1, and then next frame
will have a different routine (boss_gemini_routine_03
) set the timer to
0, but that routine doesn't call disable_enemy_collision
, leaving the helmet
in a vulnerable state.
Interestingly, if you take advantage of this bug, then you can exploit the same
logic mistake when the helmet is not moving at the edge of the screen, before it
starts merging. If a bullet collides with the helmet right when
ENEMY_ANIMATION_DELAY
is 2, then the helmets will remain vulnerable while
merging.
; ----- Frame 1 -----
boss_gemini_routine_02:
...
lda ENEMY_ANIMATION_DELAY,x ; ENEMY_ANIMATION_DELAY = 2
beq @calc_offset_set_pos ; branch doesn't occur
dec ENEMY_ANIMATION_DELAY,x ; ENEMY_ANIMATION_DELAY = 1
bne @set_x_pos ; helmet still not moving, branch
...
rts
...
bullet_enemy_collision_test:
...
jsr bullet_collision_logic ; set boss gemini routine `boss_gemini_routine_03`
; ----- Frame 2 -----
boss_gemini_routine_03:
lda ENEMY_ANIMATION_DELAY,x ; ENEMY_ANIMATION_DELAY = 1
beq @continue ; branch doesn't occur
dec ENEMY_ANIMATION_DELAY,x ; ENEMY_ANIMATION_DELAY = 0
...
lda #$03 ; a = #$03
jmp set_enemy_routine_to_a ; set enemy routine index to boss_gemini_routine_02
; ----- Frame 3 -----
boss_gemini_routine_02:
lda ENEMY_ANIMATION_DELAY,x ; ENEMY_ANIMATION_DELAY = 0
beq @calc_offset_set_pos ; branch skipping disabling of collision!!
dec ENEMY_ANIMATION_DELAY,x ; skipped!
bne @set_x_pos ; skipped!
jsr disable_enemy_collision ; skipped!
@calc_offset_set_pos:
...
On 2020-09-08, user aiqiyou had
posted
a .fm2 file glitch.fm2
on the TASVideos forums. This showed a 2 player play through where the game
freezes on the level 3 - waterfall dragon boss. The next day
feos posted a
video of the run on YouTube for
easier viewing. Sand did some
initial investigation as to the cause of the freeze and noticed the game was in
a forever loop inside the @enemy_orb_loop
code and it was looping forever due
to an invalid doubly linked list (ENEMY_VAR_3
and ENEMY_VAR_4
values).
The reason this freeze happens is due to a race condition where the left 'hand' (red dragon arm orb) is destroyed, but before the next frame happens where the 'orb destroyed' routine is executed, another orb on the left arm changes the routine of the left 'hand' to be a different routine. Since the expected 'orb destroyed' routine wasn't run, the rest of the arm didn't get the notice to self-destruct. Then, a few frames later, the left shoulder creates a projectile, which takes over the same slot where the left 'hand' was. Finally, one frame later, when the left shoulder tries to animate the arm, the left 'hand' not having correct data (because it is now a projectile), causes the game to get stuck in an infinite loop.
Below is a diagram of the dragon boss and its arm orbs. Each number below is the enemy slot, i.e. the enemy number. #$06 and #$05 are the left and right 'hands' respectively, and are red. #$0d and #$0a are the left and right 'shoulders' respectively. () represents the dragon's mouth and is uninvolved in this bug. In fact, only the left arm is involved in this bug.
06 08 0c 0f 0d () 0a 0e 0b 07 05
- Frame #$aa - Enemy #$06 (the left 'hand') is destroyed, the memory address
specifying which routine to execute is updated to point to
dragon_arm_orb_routine_04
. - Frame #$ab - Enemy #$0f has a timer elapse in
dragon_arm_orb_routine_02
. Enemy #$0d updates the enemy routine for all orbs on the left arm. It does this by incrementing a pointer. Usually, this updates the routine fromdragon_arm_orb_routine_02
todragon_arm_orb_routine_03
. However, since arm orb #$06 (the left 'hand') was no longer pointing todragon_arm_orb_routine_02
, but instead todragon_arm_orb_routine_04
, incrementing this pointer, set #$06's routine toenemy_routine_init_explosion
. - Frames #$ac-#$d1 - The animation for the left 'hand' explosion completes and
the 'hand' is removed from memory (
enemy_routine_remove_enemy
) - Frame #$d2 - The #$0d (left shoulder) decides that it should create a
projectile. The game logic finds an empty enemy slot where the left 'hand'
originally was (slot #$06). A bullet is created and initialized. This
initialization clears the data that linked the hand to the rest of the arm, in
particular
ENEMY_VAR_3
andENEMY_VAR_4
. - Frame #$d3 - When #$0d (left shoulder) executes, it animates the rest of the
orbs to make an attack pattern. It loops down to the hand by following the
links among the orbs. When it gets to the hand, it expects that the hand's
will have its
ENEMY_VAR_3
set to#$ff
indicating there aren't any more orbs to process. However, since the enemy at slot #$06 is no longer a hand, but instead a projectile, the value atENEMY_VAR_3
has been cleared and is #$00. This causes the logic to get stuck in@arm_orb_loop
as an infinite loop.
Step (2) caused dragon_arm_orb_routine_04
to be skipped. Since this routine
was not executed as expected, the rest of the arm didn't get updated to know
that the 'hand' was destroyed. dragon_arm_orb_routine_04
is responsible for
updating each orb on the arm to be begin its self destruct routine. However,
that never happens. So, the shoulder doesn't know to destroy itself. Instead
the shoulder operates as if it wasn't destroyed and when it decides that a
projectile should be created, that overwrites the hand with a different enemy
type, and clears all the links between the hand and the arm.