run it (fails)

mandel.s

@@ -16,6 +16,13 @@ dist  = $9c     ; fixed6.26: z_x^2 + z_y^2
 iter  = $a0     ; u8: iteration count
 zoom  = $a1     ; u8: zoom shift level
 temp  = $a2     ; u16
+temp2 = $a4     ; u16
+
+pixel_ptr    = $b0 ; u16
+pixel_color  = $b2 ; u8
+pixel_mask   = $b3 ; u8
+pixel_shift  = $b4 ; u8
+pixel_offset = $b5 ; u8
 
 ; FP registers in zero page
 FR0 = $d4
@@ -33,7 +40,8 @@ framebuffer_end    = $a000
 height = 184
 half_height = height >> 1
 width = 160
-half_width = 160 >> 1
+half_width = width >> 1
+stride = width >> 2
 width_ratio_3_13 = (5 << 11) ; 5/4
 height_ratio_3_13 = (3 << 11) ; 5/4
 
@@ -106,8 +114,8 @@ bit_masks:
     .byte 3 << 6
 
 display_list_start:
-    ; 48 lines overscan
-    .repeat 5
+    ; 24 lines overscan
+    .repeat 3
         .byte $70 ; 8 blank lines
     .endrep
 
@@ -133,6 +141,14 @@ display_list_start:
 display_list_end:
 display_list_len = display_list_end - display_list_start
 
+color_map:
+    .byte 0
+    .repeat 85
+        .byte 1
+        .byte 2
+        .byte 3
+    .endrepeat
+
 .code
 
 .export start
@@ -175,8 +191,8 @@ display_list_len = display_list_end - display_list_start
 
 .macro shl bytes, arg
     asl arg
-    .repeat bytes-1
-        rol arg
+    .repeat bytes-1, i
+        rol arg + 1 + i
     .endrepeat
 .endmacro
 
@@ -228,6 +244,21 @@ display_list_len = display_list_end - display_list_start
     neg 4, arg
 .endmacro
 
+.macro extend_8_16 dest, src
+    ; clobbers A, X
+    ; 13-15 cycles
+    .local positive
+    .local negative
+    ldx #0       ; 2 cyc
+    lda src      ; 3 cyc
+    sta dest     ; 3 cyc
+    bpl positive ; 2 cyc
+negative:
+    dex          ; 2 cyc
+positive:
+    stx dest + 1 ; 3 cyc
+.endmacro
+
 ; inner loop for imul16
 ; bitnum < 8: 25 or 41 cycles
 ; bitnum >= 8: 30 or 46 cycles
@@ -499,6 +530,86 @@ enough:
     ; screen coords in signed sx,sy
     ; iter holds the target to use
     ; @todo implement
+
+    ; iter -> color
+    ldx iter
+    lda color_map,x
+    sta pixel_color
+    lda #(255 - 3)
+    sta pixel_mask
+
+    ; sy -> line base address in temp
+    lda sy
+    bpl positive
+
+negative:
+    ; temp1 = top half
+    lda #.lobyte(framebuffer_top + stride * half_height)
+    sta pixel_ptr
+    lda #.hibyte(framebuffer_top + stride * half_height)
+    sta pixel_ptr + 1
+    jmp point
+
+positive:
+
+    lda #.lobyte(framebuffer_bottom)
+    sta pixel_ptr
+    lda #.hibyte(framebuffer_bottom)
+    sta pixel_ptr + 1
+
+point:
+
+    ; pixel_ptr += sy * stride
+    ;    temp * 40
+    ; =  temp * 32  +  temp * 8
+    ; = (temp << 5) + (temp << 3)
+    copy16 temp, sy
+    shl16 temp
+    shl16 temp
+    shl16 temp
+    add16 pixel_ptr, pixel_ptr, temp
+    shl16 temp
+    shl16 temp
+    add16 pixel_ptr, pixel_ptr, temp
+
+    ; Ok so temp1 points to the start of the line, which is 40 bytes.
+    ; Get the byte and bit offsets
+    lda sx
+    clc
+    adc #half_width
+    sta temp
+
+    ; pixel_shift = temp & 3
+    ; pixel_color <<= pixel_shift (shifting in zeros)
+    ; pixel_mask <<= pixel_shift (shifting in ones)
+    and #3
+    sta pixel_shift
+    tax
+shift_loop:
+    beq shift_done
+    asl pixel_color
+    asl pixel_color
+    sec
+    rol pixel_mask
+    sec
+    rol pixel_mask
+    dex
+    jmp shift_loop
+shift_done:
+
+    ; pixel_offset = temp >> 2
+    lda temp
+    lsr a
+    lsr a
+    sta pixel_offset
+    tay
+
+    ; read, mask, or, write
+    lda (pixel_ptr),y
+    and pixel_mask
+    ora pixel_color
+    sta (pixel_ptr),y
+
     rts
 .endproc
 
@@ -596,33 +707,32 @@ loop_sx:
     jsr mandelbrot
     jsr pset
 
-
-    lda sx
     clc
+    lda sx
     adc #1
     sta sx
-
-    cmp #half_width
-    beq loop_sx_done
-
     lda sx + 1
     adc #0
     sta sx + 1
+
+    lda sx
+    cmp #half_width
+    beq loop_sx_done
     jmp loop_sx
 
 loop_sx_done:
 
-    lda sy
     clc
+    lda sy
     adc #1
     sta sy
-
-    cmp #half_height
-    beq loop_sy_done
-
     lda sy + 1
     adc #0
     sta sy + 1
+
+    lda sy
+    cmp #half_height
+    beq loop_sy_done
     jmp loop_sy
 
 loop_sy_done: