tmap.nas

The source code of Doom legacy for windows

;;-----------------------------------------------------------------------------
;;
;; $Id: tmap.nas,v 1.10 2001/02/24 13:35:21 bpereira Exp $
;;
;; Copyright (C) 1998-2000 by DooM Legacy Team.
;;
;; This program is free software; you can redistribute it and/or
;; modify it under the terms of the GNU General Public License
;; as published by the Free Software Foundation; either version 2
;; of the License, or (at your option) any later version.
;;
;; This program is distributed in the hope that it will be useful,
;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
;; GNU General Public License for more details.
;;
;;
;; $Log: tmap.nas,v $
;; Revision 1.10  2001/02/24 13:35:21  bpereira
;; no message
;;
;; Revision 1.9  2001/02/10 15:24:19  hurdler
;; Apply Rob's patch for Linux version
;;
;; Revision 1.8  2000/11/12 09:48:15  bpereira
;; no message
;;
;; Revision 1.7  2000/11/06 20:52:16  bpereira
;; no message
;;
;; Revision 1.6  2000/11/03 11:48:40  hurdler
;; Fix compiling problem under win32 with 3D-Floors and FragglScript (to verify!)
;;
;; Revision 1.5  2000/11/03 03:27:17  stroggonmeth
;; Again with the bug fixing...
;;
;; Revision 1.4  2000/11/02 17:50:10  stroggonmeth
;; Big 3Dfloors & FraggleScript commit!!
;;
;; Revision 1.3  2000/04/24 20:24:38  bpereira
;; no message
;;
;; Revision 1.2  2000/02/27 00:42:11  hurdler
;; fix CR+LF problem
;;
;; Revision 1.1.1.1  2000/02/22 20:32:32  hurdler
;; Initial import into CVS (v1.29 pr3)
;;
;;
;; DESCRIPTION:
;;      assembler optimised rendering code for software mode
;;      draw floor spans, and wall columns.
;;
;;-----------------------------------------------------------------------------


[BITS 32]

%ifdef LINUX
%macro cextern 1
[extern %1]
%endmacro

%macro cglobal 1
[global %1]
%endmacro

%define CODE_SEG .data
%else
%macro cextern 1
%define %1 _%1
[extern %1]
%endmacro

%macro cglobal 1
%define %1 _%1
[global %1]
%endmacro

%define CODE_SEG .text                         
%endif


;; externs
;; columns
cextern dc_x
cextern dc_yl
cextern dc_yh
cextern ylookup
cextern columnofs
cextern dc_source
cextern dc_texturemid
cextern dc_iscale
cextern centery
cextern dc_colormap
cextern dc_transmap
cextern colormaps

;; spans
cextern ds_x1
cextern ds_x2
cextern ds_y
cextern ds_xfrac
cextern ds_yfrac
cextern ds_xstep
cextern ds_ystep
cextern ds_source
cextern ds_colormap
;cextern ds_textureheight

; polygon edge rasterizer
cextern prastertab


;;----------------------------------------------------------------------
;;
;; R_DrawColumn
;;
;; New  optimised version 10-01-1998 by D.Fabrice and P.Boris
;; TO DO: optimise it much farther... should take at most 3 cycles/pix
;;      once it's fixed, add code to patch the offsets so that it
;;      works in every screen width.
;;
;;----------------------------------------------------------------------

[SECTION .data]

;;.align        4
loopcount       dd      0
pixelcount      dd      0
tystep          dd      0

[SECTION CODE_SEG write]

;----------------------------------------------------------------------------
;fixed_t FixedMul (fixed_t a, fixed_t b)
;----------------------------------------------------------------------------
cglobal FixedMul
;       align   16
FixedMul:
        mov     eax,[esp+4]
        imul    dword [esp+8]
        shrd    eax,edx,16
        ret

;----------------------------------------------------------------------------
;fixed_t FixedDiv2 (fixed_t a, fixed_t b);
;----------------------------------------------------------------------------
cglobal FixedDiv2
;       align   16
FixedDiv2:
        mov     eax,[esp+4]
        mov     edx,eax                 ;; these two instructions allow the next
        sar     edx,31                  ;; two to pair, on the Pentium processor.
        shld    edx,eax,16
        sal     eax,16
        idiv    dword [esp+8]
        ret

;----------------------------------------------------------------------------
; void  ASM_PatchRowBytes (int rowbytes);
;----------------------------------------------------------------------------
cglobal ASM_PatchRowBytes
;       align   16
ASM_PatchRowBytes:
        mov     eax,[esp+4]
        mov     [p1+2],eax
        mov     [p2+2],eax
        mov     [p3+2],eax
        mov     [p4+2],eax
        mov     [p5+2],eax
        mov     [p6+2],eax
        mov     [p7+2],eax
        mov     [p8+2],eax
        mov     [p9+2],eax
        mov     [pa+2],eax
        mov     [pb+2],eax
        mov     [pc+2],eax
        mov     [pd+2],eax
        mov     [pe+2],eax
        mov     [pf+2],eax
        mov     [pg+2],eax
        mov     [ph+2],eax
        mov     [pi+2],eax
        mov     [pj+2],eax
        mov     [pk+2],eax
        mov     [pl+2],eax
        mov     [pm+2],eax
        mov     [pn+2],eax
        mov     [po+2],eax
        mov     [pp+2],eax
        mov     [pq+2],eax
        add     eax,eax
        mov     [q1+2],eax
        mov     [q2+2],eax
        mov     [q3+2],eax
        mov     [q4+2],eax
        mov     [q5+2],eax
        mov     [q6+2],eax
        mov     [q7+2],eax
        mov     [q8+2],eax
        ret


;----------------------------------------------------------------------------
; 8bpp column drawer
;----------------------------------------------------------------------------

cglobal R_DrawColumn_8
;       align   16
R_DrawColumn_8:
        push    ebp                     ;; preserve caller's stack frame pointer
        push    esi                     ;; preserve register variables
        push    edi
        push    ebx
;;
;; dest = ylookup[dc_yl] + columnofs[dc_x];
;;
        mov     ebp,[dc_yl]
        mov     ebx,ebp
        mov     edi,[ylookup+ebx*4]
        mov     ebx,[dc_x]
        add     edi,[columnofs+ebx*4]  ;; edi = dest
;;
;; pixelcount = yh - yl + 1
;;
        mov     eax,[dc_yh]
        inc     eax
        sub     eax,ebp                 ;; pixel count
        mov     [pixelcount],eax        ;; save for final pixel
        jle     near vdone                   ;; nothing to scale
;;
;; frac = dc_texturemid - (centery-dc_yl)*fracstep;
;;
        mov     ecx,[dc_iscale]        ;; fracstep
        mov     eax,[centery]
        sub     eax,ebp
        imul    eax,ecx
        mov     edx,[dc_texturemid]
        sub     edx,eax
        mov     ebx,edx
        shr     ebx,16                  ;; frac int.
        and     ebx,0x7f
        shl     edx,16                  ;; y frac up

        mov     ebp,ecx
        shl     ebp,16                  ;; fracstep f. up
        shr     ecx,16                  ;; fracstep i. ->cl
        and     cl,0x7f
        mov     esi,[dc_source]
;;
;; lets rock :) !
;;
        mov     eax,[pixelcount]
        mov     dh,al
        shr     eax,2
        mov     ch,al                   ;; quad count
        mov     eax,[dc_colormap]
        test    dh,0x3
        je      near v4quadloop
;;
;;  do un-even pixel
;;
        test    dh,0x1
        je      two_uneven

        mov     al,[esi+ebx]            ;; prep un-even loops
        add     edx,ebp                 ;; ypos f += ystep f
        adc     bl,cl                   ;; ypos i += ystep i
        mov     dl,[eax]                ;; colormap texel
        and     bl,0x7f                 ;; mask 0-127 texture index
        mov     [edi],dl                ;; output pixel
p1:     add     edi,0x12345678
;;
;;  do two non-quad-aligned pixels
;;
two_uneven:
        test    dh,0x2
        je      f3

        mov     al,[esi+ebx]            ;; fetch source texel
        add     edx,ebp                 ;; ypos f += ystep f
        adc     bl,cl                   ;; ypos i += ystep i
        mov     dl,[eax]                ;; colormap texel
        and     bl,0x7f                 ;; mask 0-127 texture index
        mov     [edi],dl                ;; output pixel
        mov     al,[esi+ebx]
        add     edx,ebp                 ;; fetch source texel
        adc     bl,cl                   ;; ypos f += ystep f
        mov     dl,[eax]                ;; ypos i += ystep i
        and     bl,0x7f                 ;; colormap texel
p2:     add     edi,0x12345678          ;; mask 0-127 texture index
        mov     [edi],dl
p3:     add     edi,0x12345678          ;; output pixel
;;
;;  test if there was at least 4 pixels
;;
f3:
        test    ch,0xff                 ;; test quad count
        je      near vdone
;;
;; ebp : ystep frac. upper 16 bits
;; edx : y     frac. upper 16 bits
;; ebx : y     i.    lower 7 bits,  masked for index
;; ecx : ch = counter, cl = y step i.
;; eax : colormap aligned 256
;; esi : source texture column
;; edi : dest screen
;;
v4quadloop:
        mov     dh,0x7f                 ;; prep mask
align 4
vquadloop:
        mov     al,[esi+ebx]            ;; prep loop
        add     edx,ebp                 ;; ypos f += ystep f
        adc     bl,cl                   ;; ypos i += ystep i
        mov     dl,[eax]                ;; colormap texel
        mov     [edi],dl                ;; output pixel
        and     bl,0x7f                 ;; mask 0-127 texture index

        mov     al,[esi+ebx]            ;; fetch source texel
        add     edx,ebp
        adc     bl,cl
p4:     add     edi,0x12345678
        mov     dl,[eax]
        and     bl,0x7f
        mov     [edi],dl

        mov     al,[esi+ebx]            ;; fetch source texel
        add     edx,ebp
        adc     bl,cl
p5:     add     edi,0x12345678
        mov     dl,[eax]
        and     bl,0x7f
        mov     [edi],dl

        mov     al,[esi+ebx]            ;; fetch source texel
        add     edx,ebp
        adc     bl,cl
p6:     add     edi,0x12345678
        mov     dl,[eax]
        and     bl,0x7f
        mov     [edi],dl

p7:     add     edi,0x12345678

        dec     ch
        jne     vquadloop

vdone:
        pop     ebx                     ;; restore register variables
        pop     edi
        pop     esi
        pop     ebp                     ;; restore caller's stack frame pointer
        ret

;;----------------------------------------------------------------------
;;13-02-98:
;;      R_DrawSkyColumn : same as R_DrawColumn but:
;;
;;      - wrap around 256 instead of 127.
;;      this is needed because we have a higher texture for mouselook,