;TEX_X = 512
;TEX_Y = 512
;ROUND equ 8
;SIZE_X = 512
;SIZE_Y = 512
;TEX_SHIFT = 9
CATMULL_SHIFT equ 8

;------------------------------------------------------------------------
;- Procedure drawing textured triangle using Catmull Z-buffer algorithm -
;------------------------------------------------------------------------
tex_triangle_z:
;----------in - eax - x1 shl 16 + y1
;-------------- ebx - x2 shl 16 + y2
;---------------ecx - x3 shl 16 + y3
;---------------edx - pointer to Z-buffer
;---------------esi - pointer to texture buffer
;---------------edi - pointer to screen buffer
;-------------stack - texture coordinates
;------------------ - z coordinates
.tex_x1 equ  ebp+4
.tex_y1 equ  ebp+6
.tex_x2 equ  ebp+8
.tex_y2 equ  ebp+10
.tex_x3 equ  ebp+12
.tex_y3 equ  ebp+14
.z1     equ  word[ebp+16]
.z2     equ  word[ebp+18]
.z3     equ  word[ebp+20]

.tex_ptr equ dword[ebp-4]        ; pointer to texture
.z_ptr   equ dword[ebp-8]        ; pointer to z-buffer
.x1      equ word[ebp-10]
.y1      equ word[ebp-12]
.x2      equ word[ebp-14]
.y2      equ word[ebp-16]
.x3      equ word[ebp-18]
.y3      equ word[ebp-20]

.dx12     equ dword[ebp-24]
.tex_dx12 equ dword[ebp-28]
.tex_dy12 equ dword[ebp-32]
.dz12     equ dword[ebp-36]

.dx13     equ dword[ebp-40]
.tex_dx13 equ dword[ebp-44]
.tex_dy13 equ dword[ebp-48]
.dz13     equ dword[ebp-52]

.dx23     equ dword[ebp-56]
.tex_dx23 equ dword[ebp-60]
.tex_dy23 equ dword[ebp-64]
.dz23     equ dword[ebp-68]

.scan_x1  equ dword[ebp-72]
.scan_x2  equ dword[ebp-76]
.scan_y1  equ dword[ebp-80]
.scan_y2  equ dword[ebp-84]
.cz1      equ dword[ebp-88]
.cz2      equ dword[ebp-92]

        mov     ebp,esp
        push    esi               ; store memory pointers
        push    edx
.tt_sort3:
        cmp     ax,bx                     ;sort all parameters
        jle     .tt_sort1
        xchg    eax,ebx
        mov     edx,dword [.tex_x1]
        xchg    edx,dword [.tex_x2]
        mov     dword[.tex_x1],edx
        mov     dx,.z1
        xchg    dx,.z2
        mov     .z1,dx
.tt_sort1:
        cmp     bx,cx
        jle     .tt_sort2
        xchg    ebx,ecx
        mov     edx,dword [.tex_x2]
        xchg    edx,dword [.tex_x3]
        mov     dword [.tex_x2],edx
        mov     dx,.z2
        xchg    dx,.z3
        mov     .z2,dx
        jmp     .tt_sort3
.tt_sort2:

        push    eax     ; and store to user friendly variables
        push    ebx
        push    ecx

        mov      edx,80008000h  ; eax,ebx,ecx are ANDd together into edx which means that
        and      edx,ebx        ; if *all* of them are negative a sign flag is raised
        and      edx,ecx
        and      edx,eax
        test     edx,80008000h  ; Check both X&Y at once
        jne      .tt_loop2_end
    ;    cmp      ax,SIZE_Y
    ;    jl       @f
    ;    cmp      bx,SIZE_Y
    ;    jl       @f
    ;    cmp      cx,SIZE_Y
    ;    jl       @f
        ror      eax,16
        ror      ebx,16
        ror      ecx,16
    ;    cmp      ax,SIZE_X
    ;    jl       @f
    ;    cmp      bx,SIZE_X
    ;    jl       @f
    ;    cmp      cx,SIZE_X
    ;    jl       @f
    ;    jmp      .tt_loop2_end
     @@:
        mov      eax,dword[.tex_x1] ; texture coords must be in [0..TEX_X(Y)]
        mov      ebx,dword[.tex_x2]
        mov      ecx,dword[.tex_x3]
        mov      edx,eax
        or       edx,ebx
        or       edx,ecx
        test     edx,80008000h
        jne      .tt_loop2_end
        cmp      ax,TEX_X
        jge      .tt_loop2_end
        cmp      bx,TEX_X
        jge      .tt_loop2_end
        cmp      cx,TEX_X
        jge      .tt_loop2_end
        ror      eax,16
        ror      ebx,16
        ror      ecx,16
        cmp      ax,TEX_Y
        jge      .tt_loop2_end
        cmp      bx,TEX_Y
        jge      .tt_loop2_end
        cmp      cx,TEX_Y
        jge      .tt_loop2_end


        movsx    ebx,.y2    ; calc delta
        sub      bx,.y1
        jnz      .tt_dx12_make
        xor      edx,edx
        mov      ecx,4
    @@:
        push     edx
        loop     @b
        jmp      .tt_dx12_done
    .tt_dx12_make:
        mov      ax,.x2
        sub      ax,.x1
        cwde
        shl      eax,ROUND
        cdq
        idiv     ebx
       ; mov      .dx12,eax                      ; dx12 = (x2-x1)/(y2-y1)
        push     eax

        mov      ax,word[.tex_x2]
        sub      ax,word[.tex_x1]
        cwde
        shl      eax,ROUND
        cdq
        idiv     ebx
       ; mov     [.tex_dx12],eax               ; tex_dx12 = (tex_x2-tex_x1)/(y2-y1)
        push     eax

        mov      ax,word[.tex_y2]
        sub      ax,word[.tex_y1]
        cwde
        shl      eax,ROUND
        cdq
        idiv     ebx
     ;   mov    [.tex_dy12],eax              ; tex_dy12 = (tex_y2-tex_y1)/(y2-y1)
        push     eax

        mov      ax,.z2
        sub      ax,.z1
        cwde
        shl      eax,CATMULL_SHIFT
        cdq
        idiv     ebx
        push     eax
   .tt_dx12_done:

        movsx    ebx,.y3    ; calc delta
        sub      bx,.y1
        jnz      .tt_dx13_make
        xor      edx,edx
        mov      ecx,4
    @@:
        push     edx
        loop     @b
        jmp      .tt_dx13_done
    .tt_dx13_make:
        mov      ax,.x3
        sub      ax,.x1
        cwde
        shl      eax,ROUND
        cdq
        idiv     ebx
       ; mov      .dx12,eax                      ; dx13 = (x3-x1)/(y3-y1)
        push      eax

        mov      ax,word[.tex_x3]
        sub      ax,word[.tex_x1]
        cwde
        shl      eax,ROUND
        cdq
        idiv     ebx
       ; mov     [.tex_dx12],eax               ; tex_dx13 = (tex_x3-tex_x1)/(y3-y1)
        push     eax

        mov      ax,word[.tex_y3]
        sub      ax,word[.tex_y1]
        cwde
        shl      eax,ROUND
        cdq
        idiv     ebx
     ;   mov    [.tex_dy12],eax              ; tex_dy13 = (tex_y3-tex_y1)/(y3-y1)
        push     eax

        mov      ax,.z3
        sub      ax,.z1
        cwde
        shl      eax,CATMULL_SHIFT
        cdq
        idiv     ebx
        push     eax
   .tt_dx13_done:

        mov      bx,.y3    ; calc delta
        sub      bx,.y2
        jnz      .tt_dx23_make
        xor      edx,edx
        mov      ecx,4
    @@:
        push     edx
        loop     @b
        jmp      .tt_dx23_done
    .tt_dx23_make:
        mov      ax,.x3
        sub      ax,.x2
        cwde
        shl      eax,ROUND
        cdq
        movzx    ebx,bx
        idiv     ebx
       ; mov      .dx23,eax                      ; dx23 = (x3-x2)/(y3-y2)
        push      eax

        mov      ax,word[.tex_x3]
        sub      ax,word[.tex_x2]
        cwde
        shl      eax,ROUND
        cdq
        idiv     ebx
       ; mov     [.tex_dx23],eax               ; tex_dx23 = (tex_x3-tex_x2)/(y3-y2)
        push     eax

        mov      ax,word[.tex_y3]
        sub      ax,word[.tex_y2]
        cwde
        shl      eax,ROUND
        cdq
        idiv     ebx
     ;   mov    [.tex_dy23],eax              ; tex_dy23 = (tex_y3-tex_y2)/(y3-y2)
        push     eax

        mov      ax,.z3
        sub      ax,.z2
        cwde
        shl      eax,CATMULL_SHIFT
        cdq
        idiv     ebx
        push     eax
   .tt_dx23_done:

        movsx    eax,.x1     ;eax - cur x1
        shl      eax,ROUND   ;ebx - cur x2
        mov      ebx,eax

        movsx    edx, word[.tex_x1]
        shl      edx,ROUND
      ;  mov [.scan_x1],edx
      ;  mov [.scan_x2],edx
        push     edx
        push     edx
        movsx    edx, word[.tex_y1]
        shl      edx,ROUND
       ; mov [.scan_y1],edx
       ; mov [.scan_y2],edx
        push     edx
        push     edx
        movsx    edx,.z1
        shl      edx,CATMULL_SHIFT
        push     edx
        push     edx
        mov      cx,.y1
        cmp      cx,.y2
        jge      .tt_loop1_end

   .tt_loop1:
        pushad

        push    .z_ptr
        push    .cz1      ; z coords shifted shl catmull_shift
        push    .cz2
        push    .scan_y2
        push    .scan_x2
        push    .scan_y1
        push    .scan_x1
        push    esi               ;[.tex_ptr]

        push    cx
        sar     ebx,ROUND
        push    bx
        sar     eax,ROUND
        push    ax
        call    textured_line_z

        popad
        mov     edx,.dz13
        add     .cz1,edx
        mov     edx,.dz12
        add     .cz2,edx

        mov     edx, .tex_dx13
        add     .scan_x1, edx
        mov     edx, .tex_dx12
        add     .scan_x2, edx
        mov     edx, .tex_dy13
        add     .scan_y1, edx
        mov     edx, .tex_dy12
        add     .scan_y2, edx

        add     eax, .dx13
        add     ebx, .dx12
        inc     cx
        cmp     cx,.y2
        jl      .tt_loop1

  .tt_loop1_end:


        mov     cx,.y2
        cmp     cx,.y3
        jge     .tt_loop2_end

        movsx   ebx,.x2
        shl     ebx,ROUND
        movsx   edx,.z2
        shl     edx,CATMULL_SHIFT
        mov     .cz2,edx
        movzx   edx, word [.tex_x2]
        shl     edx,ROUND
        mov     .scan_x2,edx
        movzx   edx, word[.tex_y2]
        shl     edx,ROUND
        mov     .scan_y2,edx

.tt_loop2:

        pushad

        push    .z_ptr
        push    .cz1      ; z coords shifted shl catmull_shift
        push    .cz2

        push    .scan_y2
        push    .scan_x2
        push    .scan_y1
        push    .scan_x1
        push    esi               ;[.tex_ptr]

        push    cx
        sar     ebx,ROUND
        push    bx
        sar     eax,ROUND
        push    ax
        call    textured_line_z

        popad


        mov     edx,.dz13
        add     .cz1,edx
        mov     edx,.dz23
        add     .cz2,edx

        mov     edx, .tex_dx13
        add     .scan_x1, edx
        mov     edx, .tex_dx23
        add     .scan_x2, edx
        mov     edx, .tex_dy13
        add     .scan_y1, edx
        mov     edx, .tex_dy23
        add     .scan_y2, edx

        add     eax, .dx13
        add     ebx, .dx23
        inc     cx
        cmp     cx,.y3
        jl      .tt_loop2

.tt_loop2_end:

.tt_end:
        mov esp,ebp
ret 18

textured_line_z:
;-----in -edi screen buffer pointer
;------------ stack:
  .x1 equ word [ebp+4]
  .x2 equ word [ebp+6]
  .y  equ word [ebp+8]

  .tex_ptr equ dword [ebp+10]
  .tex_x1  equ ebp+14
  .tex_y1  equ ebp+18
  .tex_x2  equ ebp+22
  .tex_y2  equ ebp+26
  .z2      equ dword [ebp+30]     ;z1, z2  coords shifted shl CATMULL_SHIFT
  .z1      equ dword [ebp+34]
  .z_ptr   equ dword [ebp+38]

  .tex_dy  equ dword [ebp-4]
  .tex_dx  equ dword [ebp-8]
  .dz      equ dword [ebp-12]
  .cz      equ dword [ebp-16]
  .c_tex_x equ dword [ebp-20]  ; current tex x
  .m_sft1 equ ebp-28
  .m_sft2 equ ebp-32
;  .c_tex_xM equ ebp+14
  .tex_dxM  equ ebp-8

        mov     ebp,esp

        mov     ax,.y
        or      ax,ax
        jl      .tl_quit
        mov     bx,[size_y_var]
        dec     bx
        cmp     ax,bx  ;SIZE_Y
        jge     .tl_quit

        mov     ax,.x1
        cmp     ax,.x2
        je      .tl_quit
        jl      .tl_ok

        xchg    ax,.x2    ; sort params
        mov     .x1,ax
if Ext >= SSE2
        movdqu  xmm0,[.tex_x1]
        pshufd  xmm0,xmm0,01001110b
        movdqu  [.tex_x1],xmm0
else if Ext >= MMX
        movq    mm0,[.tex_x1]
        movq    mm1,[.tex_x2]
        movq    [.tex_x2],mm0
        movq    [.tex_x1],mm1

else
        mov     eax,dword[.tex_x1]
        xchg    eax,dword[.tex_x2]
        mov     dword[.tex_x1],eax

        mov     eax,dword[.tex_y1]
        xchg    eax,dword[.tex_y2]
        mov     dword[.tex_y1],eax

end if

        mov     eax,.z1
        xchg    eax,.z2
        mov     .z1,eax

     .tl_ok:
        mov     cx,[size_x_var]
        dec     cx
        cmp     .x1,cx   ;SIZE_X
        jge     .tl_quit
        cmp     .x2,0
        jle     .tl_quit

        mov     bx,.x2
        sub     bx,.x1
        movsx   ebx,bx

        mov     eax,dword[.tex_y2]       ; calc .dty
        sub     eax,dword[.tex_y1]
        cdq
        idiv    ebx
        push    eax

        mov     eax,dword[.tex_x2]       ; calc .dtx
        sub     eax,dword[.tex_x1]
        cdq
        idiv    ebx
        push    eax

        mov     eax,.z2       ; calc .dz
        sub     eax,.z1
        cdq
        idiv    ebx
        push    eax

        cmp    .x1,0          ; clipping
        jg     @f

        movsx   ebx,.x1
        neg     ebx
        imul    ebx           ; eax = .dz * abs(.x1)
        add     .z1,eax
        mov     .x1,0

        mov     eax,.tex_dy
        imul    ebx
        add     dword[.tex_y1],eax

        mov     eax,.tex_dx
        imul    ebx
        add     dword[.tex_x1],eax

      @@:
        cmp     .x2,cx ;SIZE_X
        jl      @f
        mov     .x2,cx  ;SIZE_X
      @@:

        movsx   ebx,.y        ; calc mem begin in buffers
        movzx   eax,word[size_x_var]  ;SIZE_X
        mul     ebx
        movsx   ebx,.x1
        add     eax,ebx
        mov     ebx,eax

        lea     eax,[eax*3]
        add     edi,eax               ; edi - scr buff
        shl     ebx,2
        add     .z_ptr,ebx            ; z buffer pointer

        mov     cx,.x2
        sub     cx,.x1
        movzx   ecx,cx

;if Ext >= MMX
;        movq    mm0,[.tex_x1]
;        movq    mm4,mm0
;        movq    mm1,qword[.tex_dxM]
;        mov     ebx,.z1
;        mov     eax,.dz
;else
        mov     eax,dword[.tex_x1]
        mov     ebx,dword[.tex_y1]
        push    .z1           ; .cz
        push    eax           ;.c_tex_x
;end if
        mov     edx,.z_ptr

     .tl_loop:

;if Ext >= MMX
;        cmp     ebx,[edx]                       ;  ebx - current z
;        jge     @f
;        movq    mm2,mm0
;        psrad   mm2,ROUND
;        movq    mm3,mm2
;        psrlq   mm2,32-TEX_SHIFT
;        paddd   mm3,mm2
;        movd    esi,mm3
;        mov     dword[edx],ebx    ; renew z buffer
;else
                                                        ;  eax - temp
        mov     eax,.cz                         ;  ebx - cur tex y shl ROUND
        cmp     eax,[edx]                       ;  ecx - l.lenght
        jge     @f         ; ebx - cur tex_y ;  edx - temp
        mov     esi,ebx                 ;  edi - scr buff
        sar     esi,ROUND               ;  esi - tex_ptr temp
        shl     esi,TEX_SHIFT           ;  .z_ptr - cur pointer to z buff
        mov     eax,.c_tex_x            ;  .cz - cur z coord shl CATMULL_SHIFT
        sar     eax,ROUND
        add     esi,eax
        mov     eax,.cz
        mov     dword[edx],eax    ; renew z buffer
;end if
        and     esi,TEXTURE_SIZE
        lea     esi,[esi*3]
        add     esi,.tex_ptr
        movsd
        dec     edi
        jmp     .no_skip
      @@:
        add     edi,3
     .no_skip:
        add     edx,4
;if Ext >= MMX
;        add     ebx,eax
;        paddd   mm0,mm1
;else
        mov     eax,.dz
        add     .cz,eax
        mov     eax,.tex_dx
        add     .c_tex_x,eax
        add     ebx,.tex_dy
;end if
        loop    .tl_loop
 .tl_quit:

        mov     esp,ebp

ret 30+8