Fix problem on real HW where the player will start the next level shooting if they died while shooting, regardless of the state of the mouse. It now ignores the first poll of the mouse at the start of a level to avoid stale poll information. Add some debug in the code for jumping to the head and body segment draw routines to confirm that the table offsets are sane. I have seen a crash which indicates that sometimes the offset is incorrect.

BUGS.md

@@ -4,11 +4,12 @@ BUGS
 This is a list of the software bugs (as opposed to the bugs in the game that you shoot) that still need attention:
 
 * On the latest build (0136579125034f41f8415b496d5ba706e86d65d9) on real HW, on startup I head the spider loop once and then I can play a game and finish that game.  But when I try to start a second game, I crash.  The stack looks corrupted and execution is way off in the weeds somewhere.  This isn't happening on either emulator.
+   * I am thinking this is the same crash as the one which follows.  I tried to reproduce this on real HW again with that debug in place but could not.
+* I reproduced a crash on GSPlus after dying and starting a new game.  From the stack, it was clear it was trying to draw a body segment but the jump instruction was set to garbage.  I am hoping this is the same crash as the one saw on real HW.
+   * I have added some debug code to detect this.  I am validating that the offset into the draw table for head and body segments is "sane" and if not brk.
 * A spider moving left to right went off screen and left garbage on the RHS as it exited.  I have only seen this once.  I think it coincided with the player dying.
 * Sometimes when the player dies, the "you can shoot" indicator is left behind as garbage on-screen.
-* If you die holding the mouse button down, your next game will start shooting without pressing the mouse button.
-   * I was reproducing this reliably on real HW but I don't see it from either emulator.  Need to try to reproduce it again on real HW.
-   * I also don't see how this happens given how the code is structured.
+
 
 
 
@@ -24,3 +25,6 @@ FIXED
    * Caused by both gamePlayer.s and level.s calling start level when the last segment is the one which kills the player.  The player code now defers to the level code to handle starting the level after the player dies that way.
 * Sometimes centipede segments seem to be stacked one on top of another.  You think there is just one left but you shoot it and there is one underneath it.  This should not happen.
     * Caused by the lack of head segment collision detection in the slow segment code paths.  I had thought that slow head segments would never collide.  But once you split a slow centipede into two separate segments, you now have slow heads that can collide.
+* If you die holding the mouse button down, your next game will start shooting without pressing the mouse button.
+   * I was reproducing this reliably on real HW but I don't see it from either emulator.  Need to try to reproduce it again on real HW.
+   * The problem is that the game should ignore the first poll from the mouse when starting a new level.  The mouse HW/ADB has put some info in about the state of the mouse and we don't poll it for quite some time.  Even if the mouse changes state, the last state is still there and becomes stale.  The solution I implemented was to throw away the first mouse poll at the start of every level.

BuGS/gamePlayer.s

@@ -22,6 +22,7 @@ PLAYER_RESTART_LEVEL_FRAME_COUNT	equ 20
 initPlayer entry
 		ldy #STARTING_NUM_LIVES
 		sty numLives
+		stz playerIgnoreFirstPoll
 		ldx #P1_LIVES_OFFSET
 initPlayer_loop anop
 		lda #TILE_PLAYER
@@ -37,6 +38,7 @@ initPlayer_loop anop
 		
 		
 playerLevelStart entry
+		stz playerIgnoreFirstPoll
 		lda #PLAYER_STATE_ONSCREEN
 		sta playerState
 		lda #STARTING_MOUSE_X
@@ -194,7 +196,11 @@ updatePlayer_mouseYOnly anop
 updatePlayer_doneMousePoll anop
 		plb
 		long i,m
-		bra updatePlayer_handleDeltas
+		lda playerIgnoreFirstPoll
+		bne updatePlayer_handleDeltas
+		lda #1
+		sta playerIgnoreFirstPoll
+		jmp updatePlayer_skipDeltas
 
 updatePlayer_noMousePoll anop
 		plb
@@ -528,6 +534,7 @@ updatePlayer_done anop
 
 playerFrameCount 		dc i2'0'
 playerExplosionOffset	dc i2'0'
+playerIgnoreFirstPoll   dc i2'0'
 mouseDown	dc i2'0'
 mouseTemp 	dc i2'0'
 

BuGS/gameSegments.s

@@ -131,6 +131,18 @@ segmentHeadJump entry
         beq segmentHeadJump_noShift
         
         tay
+		
+; DEBUG - Validate the offset into the table and crash if not valid
+		and #3
+		beq segmentHeadJump_isMultipleOf4
+		brk $5
+segmentHeadJump_isMultipleOf4 anop
+		cpy #157
+		blt segmentHeadJump_isInsideTable
+		brk $6
+segmentHeadJump_isInsideTable anop
+; DEBUG - End of validation
+
         lda headShiftJumpTable,y
         sta segmentHeadJump_jumpInst+1
         lda headShiftJumpTable+2,y
@@ -141,6 +153,18 @@ segmentHeadJump entry
         
 segmentHeadJump_noShift anop
         tay
+		
+; DEBUG - Validate the offset into the table and crash if not valid
+		and #3
+		beq segmentHeadJumpShift_isMultipleOf4
+		brk $7
+segmentHeadJumpShift_isMultipleOf4 anop
+		cpy #157
+		blt segmentHeadJumpShift_isInsideTable
+		brk $8
+segmentHeadJumpShift_isInsideTable anop
+; DEBUG - End of validation
+
         lda headJumpTable,y
         sta segmentHeadJump_jumpInst+1
         lda headJumpTable+2,y
@@ -168,6 +192,17 @@ segmentBodyJump entry
         beq segmentBodyJump_noShift
         
         tay
+; DEBUG - Validate the offset into the table and crash if not valid
+		and #3
+		beq segmentBodyJump_isMultipleOf4
+		brk $1
+segmentBodyJump_isMultipleOf4 anop
+		cpy #157
+		blt segmentBodyJump_isInsideTable
+		brk $2
+segmentBodyJump_isInsideTable anop
+; DEBUG - End of validation
+		
         lda bodyShiftJumpTable,y
         sta segmentBodyJump_jumpInst+1
         lda bodyShiftJumpTable+2,y
@@ -178,7 +213,18 @@ segmentBodyJump entry
         
 segmentBodyJump_noShift anop
         tay
-        lda bodyJumpTable,y
+; DEBUG - Validate the offset into the table and crash if not valid
+		and #3
+		beq segmentBodyJumpShift_isMultipleOf4
+		brk $3
+segmentBodyJumpShift_isMultipleOf4 anop
+		cpy #157
+		blt segmentBodyJumpShift_isInsideTable
+		brk $4
+segmentBodyJumpShift_isInsideTable anop
+; DEBUG - End of validation
+
+		lda bodyJumpTable,y
         sta segmentBodyJump_jumpInst+1
         lda bodyJumpTable+2,y
         sta segmentBodyJump_jumpInst+3