contrib/masmx86/inffas32.asm

;/* inffas32.asm is a hand tuned assembler version of inffast.c -- fast decoding
; *
; * inffas32.asm is derivated from inffas86.c, with translation of assembly code
; *
; * Copyright (C) 1995-2003 Mark Adler
; * For conditions of distribution and use, see copyright notice in zlib.h
; *
; * Copyright (C) 2003 Chris Anderson <christop@charm.net>
; * Please use the copyright conditions above.
; *
; * Mar-13-2003 -- Most of this is derived from inffast.S which is derived from
; * the gcc -S output of zlib-1.2.0/inffast.c.  Zlib-1.2.0 is in beta release at
; * the moment.  I have successfully compiled and tested this code with gcc2.96,
; * gcc3.2, icc5.0, msvc6.0.  It is very close to the speed of inffast.S
; * compiled with gcc -DNO_MMX, but inffast.S is still faster on the P3 with MMX
; * enabled.  I will attempt to merge the MMX code into this version.  Newer
; * versions of this and inffast.S can be found at
; * http://www.eetbeetee.com/zlib/ and http://www.charm.net/~christop/zlib/
; *
; * 2005 : modification by Gilles Vollant
; */
; For Visual C++ 4.x and higher and ML 6.x and higher
;   ml.exe is in directory \MASM611C of Win95 DDK
;   ml.exe is also distributed in http://www.masm32.com/masmdl.htm
;    and in VC++2003 toolkit at http://msdn.microsoft.com/visualc/vctoolkit2003/
;
;
;   compile with command line option
;   ml  /coff /Zi /c /Flinffas32.lst inffas32.asm

;   if you define NO_GZIP (see inflate.h), compile with
;   ml  /coff /Zi /c /Flinffas32.lst /DNO_GUNZIP inffas32.asm


; zlib122sup is 0 fort zlib 1.2.2.1 and lower
; zlib122sup is 8 fort zlib 1.2.2.2 and more (with addition of dmax and head
;        in inflate_state in inflate.h)
zlib1222sup      equ    8


IFDEF GUNZIP
  INFLATE_MODE_TYPE    equ 11
  INFLATE_MODE_BAD     equ 26
ELSE
  IFNDEF NO_GUNZIP
    INFLATE_MODE_TYPE    equ 11
    INFLATE_MODE_BAD     equ 26
  ELSE
    INFLATE_MODE_TYPE    equ 3
    INFLATE_MODE_BAD     equ 17
  ENDIF
ENDIF


; 75 "inffast.S"
;FILE "inffast.S"

;;;GLOBAL _inflate_fast

;;;SECTION .text


        .586p
        .mmx

        name    inflate_fast_x86
        .MODEL  FLAT

_DATA                   segment
inflate_fast_use_mmx:
        dd      1


_TEXT                   segment


ALIGN 4
        db      'Fast decoding Code from Chris Anderson'
        db      0

ALIGN 4
invalid_literal_length_code_msg:
        db      'invalid literal/length code'
        db      0

ALIGN 4
invalid_distance_code_msg:
        db      'invalid distance code'
        db      0

ALIGN 4
invalid_distance_too_far_msg:
        db      'invalid distance too far back'
        db      0


ALIGN 4
inflate_fast_mask:
dd      0
dd      1
dd      3
dd      7
dd      15
dd      31
dd      63
dd      127
dd      255
dd      511
dd      1023
dd      2047
dd      4095
dd      8191
dd      16383
dd      32767
dd      65535
dd      131071
dd      262143
dd      524287
dd      1048575
dd      2097151
dd      4194303
dd      8388607
dd      16777215
dd      33554431
dd      67108863
dd      134217727
dd      268435455
dd      536870911
dd      1073741823
dd      2147483647
dd      4294967295


mode_state       equ    0       ;/* state->mode */
wsize_state      equ    (32+zlib1222sup)        ;/* state->wsize */
write_state      equ    (36+4+zlib1222sup)      ;/* state->write */
window_state     equ    (40+4+zlib1222sup)      ;/* state->window */
hold_state       equ    (44+4+zlib1222sup)      ;/* state->hold */
bits_state       equ    (48+4+zlib1222sup)      ;/* state->bits */
lencode_state    equ    (64+4+zlib1222sup)      ;/* state->lencode */
distcode_state   equ    (68+4+zlib1222sup)      ;/* state->distcode */
lenbits_state    equ    (72+4+zlib1222sup)      ;/* state->lenbits */
distbits_state   equ    (76+4+zlib1222sup)      ;/* state->distbits */


;;SECTION .text
; 205 "inffast.S"
;GLOBAL inflate_fast_use_mmx

;SECTION .data


; GLOBAL inflate_fast_use_mmx:object
;.size inflate_fast_use_mmx, 4
; 226 "inffast.S"
;SECTION .text

ALIGN 4
_inflate_fast proc near
.FPO (16, 4, 0, 0, 1, 0)
        push  edi
        push  esi
        push  ebp
        push  ebx
        pushfd
        sub  esp,64
        cld


        mov  esi, [esp+88]
        mov  edi, [esi+28]


        mov  edx, [esi+4]
        mov  eax, [esi+0]

        add  edx,eax
        sub  edx,11

        mov  [esp+44],eax
        mov  [esp+20],edx

        mov  ebp, [esp+92]
        mov  ecx, [esi+16]
        mov  ebx, [esi+12]

        sub  ebp,ecx
        neg  ebp
        add  ebp,ebx

        sub  ecx,257
        add  ecx,ebx

        mov  [esp+60],ebx
        mov  [esp+40],ebp
        mov  [esp+16],ecx
; 285 "inffast.S"
        mov  eax, [edi+lencode_state]
        mov  ecx, [edi+distcode_state]

        mov  [esp+8],eax
        mov  [esp+12],ecx

        mov  eax,1
        mov  ecx, [edi+lenbits_state]
        shl  eax,cl
        dec  eax
        mov  [esp+0],eax

        mov  eax,1
        mov  ecx, [edi+distbits_state]
        shl  eax,cl
        dec  eax
        mov  [esp+4],eax

        mov  eax, [edi+wsize_state]
        mov  ecx, [edi+write_state]
        mov  edx, [edi+window_state]

        mov  [esp+52],eax
        mov  [esp+48],ecx
        mov  [esp+56],edx

        mov  ebp, [edi+hold_state]
        mov  ebx, [edi+bits_state]
; 321 "inffast.S"
        mov  esi, [esp+44]
        mov  ecx, [esp+20]
        cmp  ecx,esi
        ja   L_align_long

        add  ecx,11
        sub  ecx,esi
        mov  eax,12
        sub  eax,ecx
        lea  edi, [esp+28]
        rep movsb
        mov  ecx,eax
        xor  eax,eax
        rep stosb
        lea  esi, [esp+28]
        mov  [esp+20],esi
        jmp  L_is_aligned


L_align_long:
        test  esi,3
        jz   L_is_aligned
        xor  eax,eax
        mov  al, [esi]
        inc  esi
        mov  ecx,ebx
        add  ebx,8
        shl  eax,cl
        or  ebp,eax
        jmp L_align_long

L_is_aligned:
        mov  edi, [esp+60]
; 366 "inffast.S"
L_check_mmx:
        cmp  dword ptr [inflate_fast_use_mmx],2
        je   L_init_mmx
        ja   L_do_loop

        push  eax
        push  ebx
        push  ecx
        push  edx
        pushfd
        mov  eax, [esp]
        xor  dword ptr [esp],0200000h


        popfd
        pushfd
        pop  edx
        xor  edx,eax
        jz   L_dont_use_mmx
        xor  eax,eax
        cpuid
        cmp  ebx,0756e6547h
        jne  L_dont_use_mmx
        cmp  ecx,06c65746eh
        jne  L_dont_use_mmx
        cmp  edx,049656e69h
        jne  L_dont_use_mmx
        mov  eax,1
        cpuid
        shr  eax,8
        and  eax,15
        cmp  eax,6
        jne  L_dont_use_mmx
        test  edx,0800000h
        jnz  L_use_mmx
        jmp  L_dont_use_mmx
L_use_mmx:
        mov  dword ptr [inflate_fast_use_mmx],2
        jmp  L_check_mmx_pop
L_dont_use_mmx:
        mov  dword ptr [inflate_fast_use_mmx],3
L_check_mmx_pop:
        pop  edx
        pop  ecx
        pop  ebx
        pop  eax
        jmp  L_check_mmx
; 426 "inffast.S"
ALIGN 4
L_do_loop:
; 437 "inffast.S"
        cmp  bl,15
        ja   L_get_length_code

        xor  eax,eax
        lodsw
        mov  cl,bl
        add  bl,16
        shl  eax,cl
        or  ebp,eax

L_get_length_code:
        mov  edx, [esp+0]
        mov  ecx, [esp+8]
        and  edx,ebp
        mov  eax, [ecx+edx*4]

L_dolen:


        mov  cl,ah
        sub  bl,ah
        shr  ebp,cl


        test  al,al
        jnz   L_test_for_length_base

        shr  eax,16
        stosb

L_while_test:


        cmp  [esp+16],edi
        jbe  L_break_loop

        cmp  [esp+20],esi
        ja   L_do_loop
        jmp  L_break_loop

L_test_for_length_base:
; 502 "inffast.S"
        mov  edx,eax
        shr  edx,16
        mov  cl,al

        test  al,16
        jz   L_test_for_second_level_length
        and  cl,15
        jz   L_save_len
        cmp  bl,cl
        jae  L_add_bits_to_len

        mov  ch,cl
        xor  eax,eax
        lodsw
        mov  cl,bl
        add  bl,16
        shl  eax,cl
        or  ebp,eax
        mov  cl,ch

L_add_bits_to_len:
        mov  eax,1
        shl  eax,cl
        dec  eax
        sub  bl,cl
        and  eax,ebp
        shr  ebp,cl
        add  edx,eax

L_save_len:
        mov  [esp+24],edx


L_decode_distance:
; 549 "inffast.S"
        cmp  bl,15
        ja   L_get_distance_code

        xor  eax,eax
        lodsw
        mov  cl,bl
        add  bl,16
        shl  eax,cl
        or  ebp,eax

L_get_distance_code:
        mov  edx, [esp+4]
        mov  ecx, [esp+12]
        and  edx,ebp
        mov  eax, [ecx+edx*4]


L_dodist:
        mov  edx,eax
        shr  edx,16
        mov  cl,ah
        sub  bl,ah
        shr  ebp,cl
; 584 "inffast.S"
        mov  cl,al

        test  al,16
        jz  L_test_for_second_level_dist
        and  cl,15
        jz  L_check_dist_one
        cmp  bl,cl
        jae  L_add_bits_to_dist

        mov  ch,cl
        xor  eax,eax
        lodsw
        mov  cl,bl
        add  bl,16
        shl  eax,cl
        or  ebp,eax
        mov  cl,ch

L_add_bits_to_dist:
        mov  eax,1
        shl  eax,cl
        dec  eax
        sub  bl,cl
        and  eax,ebp
        shr  ebp,cl
        add  edx,eax
        jmp  L_check_window

L_check_window:
; 625 "inffast.S"
        mov  [esp+44],esi
        mov  eax,edi
        sub  eax, [esp+40]

        cmp  eax,edx
        jb   L_clip_window

        mov  ecx, [esp+24]
        mov  esi,edi
        sub  esi,edx

        sub  ecx,3
        mov  al, [esi]
        mov  [edi],al
        mov  al, [esi+1]
        mov  dl, [esi+2]
        add  esi,3
        mov  [edi+1],al
        mov  [edi+2],dl
        add  edi,3
        rep movsb

        mov  esi, [esp+44]
        jmp  L_while_test

ALIGN 4
L_check_dist_one:
        cmp  edx,1
        jne  L_check_window
        cmp  [esp+40],edi
        je  L_check_window

        dec  edi
        mov  ecx, [esp+24]
        mov  al, [edi]
        sub  ecx,3

        mov  [edi+1],al
        mov  [edi+2],al
        mov  [edi+3],al
        add  edi,4
        rep stosb

        jmp  L_while_test

ALIGN 4
L_test_for_second_level_length:


        test  al,64
        jnz   L_test_for_end_of_block

        mov  eax,1
        shl  eax,cl
        dec  eax
        and  eax,ebp
        add  eax,edx
        mov  edx, [esp+8]
        mov  eax, [edx+eax*4]
        jmp  L_dolen

ALIGN 4
L_test_for_second_level_dist:


        test  al,64
        jnz   L_invalid_distance_code

        mov  eax,1
        shl  eax,cl
        dec  eax
        and  eax,ebp
        add  eax,edx
        mov  edx, [esp+12]
        mov  eax, [edx+eax*4]
        jmp  L_dodist

ALIGN 4
L_clip_window:
; 721 "inffast.S"
        mov  ecx,eax
        mov  eax, [esp+52]
        neg  ecx
        mov  esi, [esp+56]

        cmp  eax,edx
        jb   L_invalid_distance_too_far

        add  ecx,edx
        cmp  dword ptr [esp+48],0
        jne  L_wrap_around_window

        sub  eax,ecx
        add  esi,eax
; 749 "inffast.S"
        mov  eax, [esp+24]
        cmp  eax,ecx
        jbe  L_do_copy1

        sub  eax,ecx
        rep movsb
        mov  esi,edi
        sub  esi,edx
        jmp  L_do_copy1

        cmp  eax,ecx
        jbe  L_do_copy1

        sub  eax,ecx
        rep movsb
        mov  esi,edi
        sub  esi,edx
        jmp  L_do_copy1

L_wrap_around_window:
; 793 "inffast.S"
        mov  eax, [esp+48]
        cmp  ecx,eax
        jbe  L_contiguous_in_window

        add  esi, [esp+52]
        add  esi,eax
        sub  esi,ecx
        sub  ecx,eax


        mov  eax, [esp+24]
        cmp  eax,ecx
        jbe  L_do_copy1

        sub  eax,ecx
        rep movsb
        mov  esi, [esp+56]
        mov  ecx, [esp+48]
        cmp  eax,ecx
        jbe  L_do_copy1

        sub  eax,ecx
        rep movsb
        mov  esi,edi
        sub  esi,edx
        jmp  L_do_copy1

L_contiguous_in_window:
; 836 "inffast.S"
        add  esi,eax
        sub  esi,ecx


        mov  eax, [esp+24]
        cmp  eax,ecx
        jbe  L_do_copy1

        sub  eax,ecx
        rep movsb
        mov  esi,edi
        sub  esi,edx

L_do_copy1:
; 862 "inffast.S"
        mov  ecx,eax
        rep movsb

        mov  esi, [esp+44]
        jmp  L_while_test
; 878 "inffast.S"
ALIGN 4
L_init_mmx:
        emms


        movd mm0,ebp
        mov  ebp,ebx
; 896 "inffast.S"
        movd mm4,dword ptr [esp+0]
        movq mm3,mm4
        movd mm5,dword ptr [esp+4]
        movq mm2,mm5
        pxor mm1,mm1
        mov  ebx, [esp+8]
        jmp  L_do_loop_mmx

ALIGN 4
L_do_loop_mmx:
        psrlq mm0,mm1

        cmp  ebp,32
        ja  L_get_length_code_mmx

        movd mm6,ebp
        movd mm7,dword ptr [esi]
        add  esi,4
        psllq mm7,mm6
        add  ebp,32
        por mm0,mm7

L_get_length_code_mmx:
        pand mm4,mm0
        movd eax,mm4
        movq mm4,mm3
        mov  eax, [ebx+eax*4]

L_dolen_mmx:
        movzx  ecx,ah
        movd mm1,ecx
        sub  ebp,ecx

        test  al,al
        jnz L_test_for_length_base_mmx

        shr  eax,16
        stosb

L_while_test_mmx:


        cmp  [esp+16],edi
        jbe L_break_loop

        cmp  [esp+20],esi
        ja L_do_loop_mmx
        jmp L_break_loop

L_test_for_length_base_mmx:

        mov  edx,eax
        shr  edx,16

        test  al,16
        jz  L_test_for_second_level_length_mmx
        and  eax,15
        jz L_decode_distance_mmx

        psrlq mm0,mm1
        movd mm1,eax
        movd ecx,mm0
        sub  ebp,eax
        and  ecx, [inflate_fast_mask+eax*4]
        add  edx,ecx

L_decode_distance_mmx:
        psrlq mm0,mm1

        cmp  ebp,32
        ja L_get_dist_code_mmx

        movd mm6,ebp
        movd mm7,dword ptr [esi]
        add  esi,4
        psllq mm7,mm6
        add  ebp,32
        por mm0,mm7

L_get_dist_code_mmx:
        mov  ebx, [esp+12]
        pand mm5,mm0
        movd eax,mm5
        movq mm5,mm2
        mov  eax, [ebx+eax*4]

L_dodist_mmx:

        movzx  ecx,ah
        mov  ebx,eax
        shr  ebx,16
        sub  ebp,ecx
        movd mm1,ecx

        test  al,16
        jz L_test_for_second_level_dist_mmx
        and  eax,15
        jz L_check_dist_one_mmx

L_add_bits_to_dist_mmx:
        psrlq mm0,mm1
        movd mm1,eax
        movd ecx,mm0
        sub  ebp,eax
        and  ecx, [inflate_fast_mask+eax*4]
        add  ebx,ecx

L_check_window_mmx:
        mov  [esp+44],esi
        mov  eax,edi
        sub  eax, [esp+40]

        cmp  eax,ebx
        jb L_clip_window_mmx

        mov  ecx,edx
        mov  esi,edi
        sub  esi,ebx

        sub  ecx,3
        mov  al, [esi]
        mov  [edi],al
        mov  al, [esi+1]
        mov  dl, [esi+2]
        add  esi,3
        mov  [edi+1],al
        mov  [edi+2],dl
        add  edi,3
        rep movsb

        mov  esi, [esp+44]
        mov  ebx, [esp+8]
        jmp  L_while_test_mmx

ALIGN 4
L_check_dist_one_mmx:
        cmp  ebx,1
        jne  L_check_window_mmx
        cmp  [esp+40],edi
        je   L_check_window_mmx

        dec  edi
        mov  ecx,edx
        mov  al, [edi]
        sub  ecx,3

        mov  [edi+1],al
        mov  [edi+2],al
        mov  [edi+3],al
        add  edi,4
        rep stosb

        mov  ebx, [esp+8]
        jmp  L_while_test_mmx

ALIGN 4
L_test_for_second_level_length_mmx:
        test  al,64
        jnz L_test_for_end_of_block

        and  eax,15
        psrlq mm0,mm1
        movd ecx,mm0
        and  ecx, [inflate_fast_mask+eax*4]
        add  ecx,edx
        mov  eax, [ebx+ecx*4]
        jmp L_dolen_mmx

ALIGN 4
L_test_for_second_level_dist_mmx:
        test  al,64
        jnz L_invalid_distance_code

        and  eax,15
        psrlq mm0,mm1
        movd ecx,mm0
        and  ecx, [inflate_fast_mask+eax*4]
        mov  eax, [esp+12]
        add  ecx,ebx
        mov  eax, [eax+ecx*4]
        jmp  L_dodist_mmx

ALIGN 4
L_clip_window_mmx:

        mov  ecx,eax
        mov  eax, [esp+52]
        neg  ecx
        mov  esi, [esp+56]

        cmp  eax,ebx
        jb  L_invalid_distance_too_far

        add  ecx,ebx
        cmp  dword ptr [esp+48],0
        jne  L_wrap_around_window_mmx

        sub  eax,ecx
        add  esi,eax

        cmp  edx,ecx
        jbe  L_do_copy1_mmx

        sub  edx,ecx
        rep movsb
        mov  esi,edi
        sub  esi,ebx
        jmp  L_do_copy1_mmx

        cmp  edx,ecx
        jbe  L_do_copy1_mmx

        sub  edx,ecx
        rep movsb
        mov  esi,edi
        sub  esi,ebx
        jmp  L_do_copy1_mmx

L_wrap_around_window_mmx:

        mov  eax, [esp+48]
        cmp  ecx,eax
        jbe  L_contiguous_in_window_mmx

        add  esi, [esp+52]
        add  esi,eax
        sub  esi,ecx
        sub  ecx,eax


        cmp  edx,ecx
        jbe  L_do_copy1_mmx

        sub  edx,ecx
        rep movsb
        mov  esi, [esp+56]
        mov  ecx, [esp+48]
        cmp  edx,ecx
        jbe  L_do_copy1_mmx

        sub  edx,ecx
        rep movsb
        mov  esi,edi
        sub  esi,ebx
        jmp  L_do_copy1_mmx

L_contiguous_in_window_mmx:

        add  esi,eax
        sub  esi,ecx


        cmp  edx,ecx
        jbe  L_do_copy1_mmx

        sub  edx,ecx
        rep movsb
        mov  esi,edi
        sub  esi,ebx

L_do_copy1_mmx:


        mov  ecx,edx
        rep movsb

        mov  esi, [esp+44]
        mov  ebx, [esp+8]
        jmp  L_while_test_mmx
; 1174 "inffast.S"
L_invalid_distance_code:


        mov  ecx, invalid_distance_code_msg
        mov  edx,INFLATE_MODE_BAD
        jmp  L_update_stream_state

L_test_for_end_of_block:


        test  al,32
        jz  L_invalid_literal_length_code

        mov  ecx,0
        mov  edx,INFLATE_MODE_TYPE
        jmp  L_update_stream_state

L_invalid_literal_length_code:


        mov  ecx, invalid_literal_length_code_msg
        mov  edx,INFLATE_MODE_BAD
        jmp  L_update_stream_state

L_invalid_distance_too_far:


        mov  esi, [esp+44]
        mov  ecx, invalid_distance_too_far_msg
        mov  edx,INFLATE_MODE_BAD
        jmp  L_update_stream_state

L_update_stream_state:

        mov  eax, [esp+88]
        test  ecx,ecx
        jz  L_skip_msg
        mov  [eax+24],ecx
L_skip_msg:
        mov  eax, [eax+28]
        mov  [eax+mode_state],edx
        jmp  L_break_loop

ALIGN 4
L_break_loop:
; 1243 "inffast.S"
        cmp  dword ptr [inflate_fast_use_mmx],2
        jne  L_update_next_in


        mov  ebx,ebp

L_update_next_in:
; 1266 "inffast.S"
        mov  eax, [esp+88]
        mov  ecx,ebx
        mov  edx, [eax+28]
        shr  ecx,3
        sub  esi,ecx
        shl  ecx,3
        sub  ebx,ecx
        mov  [eax+12],edi
        mov  [edx+bits_state],ebx
        mov  ecx,ebx

        lea  ebx, [esp+28]
        cmp  [esp+20],ebx
        jne  L_buf_not_used

        sub  esi,ebx
        mov  ebx, [eax+0]
        mov  [esp+20],ebx
        add  esi,ebx
        mov  ebx, [eax+4]
        sub  ebx,11
        add  [esp+20],ebx

L_buf_not_used:
        mov  [eax+0],esi

        mov  ebx,1
        shl  ebx,cl
        dec  ebx


        cmp  dword ptr [inflate_fast_use_mmx],2
        jne  L_update_hold


        psrlq mm0,mm1
        movd ebp,mm0

        emms

L_update_hold:


        and  ebp,ebx
        mov  [edx+hold_state],ebp


        mov  ebx, [esp+20]
        cmp  ebx,esi
        jbe  L_last_is_smaller

        sub  ebx,esi
        add  ebx,11
        mov  [eax+4],ebx
        jmp  L_fixup_out
L_last_is_smaller:
        sub  esi,ebx
        neg  esi
        add  esi,11
        mov  [eax+4],esi


L_fixup_out:

        mov  ebx, [esp+16]
        cmp  ebx,edi
        jbe  L_end_is_smaller

        sub  ebx,edi
        add  ebx,257
        mov  [eax+16],ebx
        jmp  L_done
L_end_is_smaller:
        sub  edi,ebx
        neg  edi
        add  edi,257
        mov  [eax+16],edi


L_done:
        add  esp,64
        popfd
        pop  ebx
        pop  ebp
        pop  esi
        pop  edi
        ret
_inflate_fast endp

_TEXT   ends
end
Revision:	1096
Committed:	Sat Dec 30 14:40:33 2017 UTC (7 years, 10 months ago) by s10k
File size:	16392 byte(s)
Log Message:	Added zlib, quazip, basicxmlsyntaxhighlighter, conditionalsemaphore and linenumberdisplay libraries. zlib and quazip are pre-compiled, but you can compile them yourself, just delete the dll files (or equivalent binary files to your OS)
#	Content
1	;/* inffas32.asm is a hand tuned assembler version of inffast.c -- fast decoding
2	; *
3	; * inffas32.asm is derivated from inffas86.c, with translation of assembly code
4	; *
5	; * Copyright (C) 1995-2003 Mark Adler
6	; * For conditions of distribution and use, see copyright notice in zlib.h
7	; *
8	; * Copyright (C) 2003 Chris Anderson <christop@charm.net>
9	; * Please use the copyright conditions above.
10	; *
11	; * Mar-13-2003 -- Most of this is derived from inffast.S which is derived from
12	; * the gcc -S output of zlib-1.2.0/inffast.c. Zlib-1.2.0 is in beta release at
13	; * the moment. I have successfully compiled and tested this code with gcc2.96,
14	; * gcc3.2, icc5.0, msvc6.0. It is very close to the speed of inffast.S
15	; * compiled with gcc -DNO_MMX, but inffast.S is still faster on the P3 with MMX
16	; * enabled. I will attempt to merge the MMX code into this version. Newer
17	; * versions of this and inffast.S can be found at
18	; * http://www.eetbeetee.com/zlib/ and http://www.charm.net/~christop/zlib/
19	; *
20	; * 2005 : modification by Gilles Vollant
21	; */
22	; For Visual C++ 4.x and higher and ML 6.x and higher
23	; ml.exe is in directory \MASM611C of Win95 DDK
24	; ml.exe is also distributed in http://www.masm32.com/masmdl.htm
25	; and in VC++2003 toolkit at http://msdn.microsoft.com/visualc/vctoolkit2003/
26	;
27	;
28	; compile with command line option
29	; ml /coff /Zi /c /Flinffas32.lst inffas32.asm
30
31	; if you define NO_GZIP (see inflate.h), compile with
32	; ml /coff /Zi /c /Flinffas32.lst /DNO_GUNZIP inffas32.asm
33
34
35	; zlib122sup is 0 fort zlib 1.2.2.1 and lower
36	; zlib122sup is 8 fort zlib 1.2.2.2 and more (with addition of dmax and head
37	; in inflate_state in inflate.h)
38	zlib1222sup equ 8
39
40
41	IFDEF GUNZIP
42	INFLATE_MODE_TYPE equ 11
43	INFLATE_MODE_BAD equ 26
44	ELSE
45	IFNDEF NO_GUNZIP
46	INFLATE_MODE_TYPE equ 11
47	INFLATE_MODE_BAD equ 26
48	ELSE
49	INFLATE_MODE_TYPE equ 3
50	INFLATE_MODE_BAD equ 17
51	ENDIF
52	ENDIF
53
54
55	; 75 "inffast.S"
56	;FILE "inffast.S"
57
58	;;;GLOBAL _inflate_fast
59
60	;;;SECTION .text
61
62
63
64	.586p
65	.mmx
66
67	name inflate_fast_x86
68	.MODEL FLAT
69
70	_DATA segment
71	inflate_fast_use_mmx:
72	dd 1
73
74
75	_TEXT segment
76
77
78
79	ALIGN 4
80	db 'Fast decoding Code from Chris Anderson'
81	db 0
82
83	ALIGN 4
84	invalid_literal_length_code_msg:
85	db 'invalid literal/length code'
86	db 0
87
88	ALIGN 4
89	invalid_distance_code_msg:
90	db 'invalid distance code'
91	db 0
92
93	ALIGN 4
94	invalid_distance_too_far_msg:
95	db 'invalid distance too far back'
96	db 0
97
98
99	ALIGN 4
100	inflate_fast_mask:
101	dd 0
102	dd 1
103	dd 3
104	dd 7
105	dd 15
106	dd 31
107	dd 63
108	dd 127
109	dd 255
110	dd 511
111	dd 1023
112	dd 2047
113	dd 4095
114	dd 8191
115	dd 16383
116	dd 32767
117	dd 65535
118	dd 131071
119	dd 262143
120	dd 524287
121	dd 1048575
122	dd 2097151
123	dd 4194303
124	dd 8388607
125	dd 16777215
126	dd 33554431
127	dd 67108863
128	dd 134217727
129	dd 268435455
130	dd 536870911
131	dd 1073741823
132	dd 2147483647
133	dd 4294967295
134
135
136	mode_state equ 0 ;/* state->mode */
137	wsize_state equ (32+zlib1222sup) ;/* state->wsize */
138	write_state equ (36+4+zlib1222sup) ;/* state->write */
139	window_state equ (40+4+zlib1222sup) ;/* state->window */
140	hold_state equ (44+4+zlib1222sup) ;/* state->hold */
141	bits_state equ (48+4+zlib1222sup) ;/* state->bits */
142	lencode_state equ (64+4+zlib1222sup) ;/* state->lencode */
143	distcode_state equ (68+4+zlib1222sup) ;/* state->distcode */
144	lenbits_state equ (72+4+zlib1222sup) ;/* state->lenbits */
145	distbits_state equ (76+4+zlib1222sup) ;/* state->distbits */
146
147
148	;;SECTION .text
149	; 205 "inffast.S"
150	;GLOBAL inflate_fast_use_mmx
151
152	;SECTION .data
153
154
155	; GLOBAL inflate_fast_use_mmx:object
156	;.size inflate_fast_use_mmx, 4
157	; 226 "inffast.S"
158	;SECTION .text
159
160	ALIGN 4
161	_inflate_fast proc near
162	.FPO (16, 4, 0, 0, 1, 0)
163	push edi
164	push esi
165	push ebp
166	push ebx
167	pushfd
168	sub esp,64
169	cld
170
171
172
173
174	mov esi, [esp+88]
175	mov edi, [esi+28]
176
177
178
179
180
181
182
183	mov edx, [esi+4]
184	mov eax, [esi+0]
185
186	add edx,eax
187	sub edx,11
188
189	mov [esp+44],eax
190	mov [esp+20],edx
191
192	mov ebp, [esp+92]
193	mov ecx, [esi+16]
194	mov ebx, [esi+12]
195
196	sub ebp,ecx
197	neg ebp
198	add ebp,ebx
199
200	sub ecx,257
201	add ecx,ebx
202
203	mov [esp+60],ebx
204	mov [esp+40],ebp
205	mov [esp+16],ecx
206	; 285 "inffast.S"
207	mov eax, [edi+lencode_state]
208	mov ecx, [edi+distcode_state]
209
210	mov [esp+8],eax
211	mov [esp+12],ecx
212
213	mov eax,1
214	mov ecx, [edi+lenbits_state]
215	shl eax,cl
216	dec eax
217	mov [esp+0],eax
218
219	mov eax,1
220	mov ecx, [edi+distbits_state]
221	shl eax,cl
222	dec eax
223	mov [esp+4],eax
224
225	mov eax, [edi+wsize_state]
226	mov ecx, [edi+write_state]
227	mov edx, [edi+window_state]
228
229	mov [esp+52],eax
230	mov [esp+48],ecx
231	mov [esp+56],edx
232
233	mov ebp, [edi+hold_state]
234	mov ebx, [edi+bits_state]
235	; 321 "inffast.S"
236	mov esi, [esp+44]
237	mov ecx, [esp+20]
238	cmp ecx,esi
239	ja L_align_long
240
241	add ecx,11
242	sub ecx,esi
243	mov eax,12
244	sub eax,ecx
245	lea edi, [esp+28]
246	rep movsb
247	mov ecx,eax
248	xor eax,eax
249	rep stosb
250	lea esi, [esp+28]
251	mov [esp+20],esi
252	jmp L_is_aligned
253
254
255	L_align_long:
256	test esi,3
257	jz L_is_aligned
258	xor eax,eax
259	mov al, [esi]
260	inc esi
261	mov ecx,ebx
262	add ebx,8
263	shl eax,cl
264	or ebp,eax
265	jmp L_align_long
266
267	L_is_aligned:
268	mov edi, [esp+60]
269	; 366 "inffast.S"
270	L_check_mmx:
271	cmp dword ptr [inflate_fast_use_mmx],2
272	je L_init_mmx
273	ja L_do_loop
274
275	push eax
276	push ebx
277	push ecx
278	push edx
279	pushfd
280	mov eax, [esp]
281	xor dword ptr [esp],0200000h
282
283
284
285
286	popfd
287	pushfd
288	pop edx
289	xor edx,eax
290	jz L_dont_use_mmx
291	xor eax,eax
292	cpuid
293	cmp ebx,0756e6547h
294	jne L_dont_use_mmx
295	cmp ecx,06c65746eh
296	jne L_dont_use_mmx
297	cmp edx,049656e69h
298	jne L_dont_use_mmx
299	mov eax,1
300	cpuid
301	shr eax,8
302	and eax,15
303	cmp eax,6
304	jne L_dont_use_mmx
305	test edx,0800000h
306	jnz L_use_mmx
307	jmp L_dont_use_mmx
308	L_use_mmx:
309	mov dword ptr [inflate_fast_use_mmx],2
310	jmp L_check_mmx_pop
311	L_dont_use_mmx:
312	mov dword ptr [inflate_fast_use_mmx],3
313	L_check_mmx_pop:
314	pop edx
315	pop ecx
316	pop ebx
317	pop eax
318	jmp L_check_mmx
319	; 426 "inffast.S"
320	ALIGN 4
321	L_do_loop:
322	; 437 "inffast.S"
323	cmp bl,15
324	ja L_get_length_code
325
326	xor eax,eax
327	lodsw
328	mov cl,bl
329	add bl,16
330	shl eax,cl
331	or ebp,eax
332
333	L_get_length_code:
334	mov edx, [esp+0]
335	mov ecx, [esp+8]
336	and edx,ebp
337	mov eax, [ecx+edx*4]
338
339	L_dolen:
340
341
342
343
344
345
346	mov cl,ah
347	sub bl,ah
348	shr ebp,cl
349
350
351
352
353
354
355	test al,al
356	jnz L_test_for_length_base
357
358	shr eax,16
359	stosb
360
361	L_while_test:
362
363
364	cmp [esp+16],edi
365	jbe L_break_loop
366
367	cmp [esp+20],esi
368	ja L_do_loop
369	jmp L_break_loop
370
371	L_test_for_length_base:
372	; 502 "inffast.S"
373	mov edx,eax
374	shr edx,16
375	mov cl,al
376
377	test al,16
378	jz L_test_for_second_level_length
379	and cl,15
380	jz L_save_len
381	cmp bl,cl
382	jae L_add_bits_to_len
383
384	mov ch,cl
385	xor eax,eax
386	lodsw
387	mov cl,bl
388	add bl,16
389	shl eax,cl
390	or ebp,eax
391	mov cl,ch
392
393	L_add_bits_to_len:
394	mov eax,1
395	shl eax,cl
396	dec eax
397	sub bl,cl
398	and eax,ebp
399	shr ebp,cl
400	add edx,eax
401
402	L_save_len:
403	mov [esp+24],edx
404
405
406	L_decode_distance:
407	; 549 "inffast.S"
408	cmp bl,15
409	ja L_get_distance_code
410
411	xor eax,eax
412	lodsw
413	mov cl,bl
414	add bl,16
415	shl eax,cl
416	or ebp,eax
417
418	L_get_distance_code:
419	mov edx, [esp+4]
420	mov ecx, [esp+12]
421	and edx,ebp
422	mov eax, [ecx+edx*4]
423
424
425	L_dodist:
426	mov edx,eax
427	shr edx,16
428	mov cl,ah
429	sub bl,ah
430	shr ebp,cl
431	; 584 "inffast.S"
432	mov cl,al
433
434	test al,16
435	jz L_test_for_second_level_dist
436	and cl,15
437	jz L_check_dist_one
438	cmp bl,cl
439	jae L_add_bits_to_dist
440
441	mov ch,cl
442	xor eax,eax
443	lodsw
444	mov cl,bl
445	add bl,16
446	shl eax,cl
447	or ebp,eax
448	mov cl,ch
449
450	L_add_bits_to_dist:
451	mov eax,1
452	shl eax,cl
453	dec eax
454	sub bl,cl
455	and eax,ebp
456	shr ebp,cl
457	add edx,eax
458	jmp L_check_window
459
460	L_check_window:
461	; 625 "inffast.S"
462	mov [esp+44],esi
463	mov eax,edi
464	sub eax, [esp+40]
465
466	cmp eax,edx
467	jb L_clip_window
468
469	mov ecx, [esp+24]
470	mov esi,edi
471	sub esi,edx
472
473	sub ecx,3
474	mov al, [esi]
475	mov [edi],al
476	mov al, [esi+1]
477	mov dl, [esi+2]
478	add esi,3
479	mov [edi+1],al
480	mov [edi+2],dl
481	add edi,3
482	rep movsb
483
484	mov esi, [esp+44]
485	jmp L_while_test
486
487	ALIGN 4
488	L_check_dist_one:
489	cmp edx,1
490	jne L_check_window
491	cmp [esp+40],edi
492	je L_check_window
493
494	dec edi
495	mov ecx, [esp+24]
496	mov al, [edi]
497	sub ecx,3
498
499	mov [edi+1],al
500	mov [edi+2],al
501	mov [edi+3],al
502	add edi,4
503	rep stosb
504
505	jmp L_while_test
506
507	ALIGN 4
508	L_test_for_second_level_length:
509
510
511
512
513	test al,64
514	jnz L_test_for_end_of_block
515
516	mov eax,1
517	shl eax,cl
518	dec eax
519	and eax,ebp
520	add eax,edx
521	mov edx, [esp+8]
522	mov eax, [edx+eax*4]
523	jmp L_dolen
524
525	ALIGN 4
526	L_test_for_second_level_dist:
527
528
529
530
531	test al,64
532	jnz L_invalid_distance_code
533
534	mov eax,1
535	shl eax,cl
536	dec eax
537	and eax,ebp
538	add eax,edx
539	mov edx, [esp+12]
540	mov eax, [edx+eax*4]
541	jmp L_dodist
542
543	ALIGN 4
544	L_clip_window:
545	; 721 "inffast.S"
546	mov ecx,eax
547	mov eax, [esp+52]
548	neg ecx
549	mov esi, [esp+56]
550
551	cmp eax,edx
552	jb L_invalid_distance_too_far
553
554	add ecx,edx
555	cmp dword ptr [esp+48],0
556	jne L_wrap_around_window
557
558	sub eax,ecx
559	add esi,eax
560	; 749 "inffast.S"
561	mov eax, [esp+24]
562	cmp eax,ecx
563	jbe L_do_copy1
564
565	sub eax,ecx
566	rep movsb
567	mov esi,edi
568	sub esi,edx
569	jmp L_do_copy1
570
571	cmp eax,ecx
572	jbe L_do_copy1
573
574	sub eax,ecx
575	rep movsb
576	mov esi,edi
577	sub esi,edx
578	jmp L_do_copy1
579
580	L_wrap_around_window:
581	; 793 "inffast.S"
582	mov eax, [esp+48]
583	cmp ecx,eax
584	jbe L_contiguous_in_window
585
586	add esi, [esp+52]
587	add esi,eax
588	sub esi,ecx
589	sub ecx,eax
590
591
592	mov eax, [esp+24]
593	cmp eax,ecx
594	jbe L_do_copy1
595
596	sub eax,ecx
597	rep movsb
598	mov esi, [esp+56]
599	mov ecx, [esp+48]
600	cmp eax,ecx
601	jbe L_do_copy1
602
603	sub eax,ecx
604	rep movsb
605	mov esi,edi
606	sub esi,edx
607	jmp L_do_copy1
608
609	L_contiguous_in_window:
610	; 836 "inffast.S"
611	add esi,eax
612	sub esi,ecx
613
614
615	mov eax, [esp+24]
616	cmp eax,ecx
617	jbe L_do_copy1
618
619	sub eax,ecx
620	rep movsb
621	mov esi,edi
622	sub esi,edx
623
624	L_do_copy1:
625	; 862 "inffast.S"
626	mov ecx,eax
627	rep movsb
628
629	mov esi, [esp+44]
630	jmp L_while_test
631	; 878 "inffast.S"
632	ALIGN 4
633	L_init_mmx:
634	emms
635
636
637
638
639
640	movd mm0,ebp
641	mov ebp,ebx
642	; 896 "inffast.S"
643	movd mm4,dword ptr [esp+0]
644	movq mm3,mm4
645	movd mm5,dword ptr [esp+4]
646	movq mm2,mm5
647	pxor mm1,mm1
648	mov ebx, [esp+8]
649	jmp L_do_loop_mmx
650
651	ALIGN 4
652	L_do_loop_mmx:
653	psrlq mm0,mm1
654
655	cmp ebp,32
656	ja L_get_length_code_mmx
657
658	movd mm6,ebp
659	movd mm7,dword ptr [esi]
660	add esi,4
661	psllq mm7,mm6
662	add ebp,32
663	por mm0,mm7
664
665	L_get_length_code_mmx:
666	pand mm4,mm0
667	movd eax,mm4
668	movq mm4,mm3
669	mov eax, [ebx+eax*4]
670
671	L_dolen_mmx:
672	movzx ecx,ah
673	movd mm1,ecx
674	sub ebp,ecx
675
676	test al,al
677	jnz L_test_for_length_base_mmx
678
679	shr eax,16
680	stosb
681
682	L_while_test_mmx:
683
684
685	cmp [esp+16],edi
686	jbe L_break_loop
687
688	cmp [esp+20],esi
689	ja L_do_loop_mmx
690	jmp L_break_loop
691
692	L_test_for_length_base_mmx:
693
694	mov edx,eax
695	shr edx,16
696
697	test al,16
698	jz L_test_for_second_level_length_mmx
699	and eax,15
700	jz L_decode_distance_mmx
701
702	psrlq mm0,mm1
703	movd mm1,eax
704	movd ecx,mm0
705	sub ebp,eax
706	and ecx, [inflate_fast_mask+eax*4]
707	add edx,ecx
708
709	L_decode_distance_mmx:
710	psrlq mm0,mm1
711
712	cmp ebp,32
713	ja L_get_dist_code_mmx
714
715	movd mm6,ebp
716	movd mm7,dword ptr [esi]
717	add esi,4
718	psllq mm7,mm6
719	add ebp,32
720	por mm0,mm7
721
722	L_get_dist_code_mmx:
723	mov ebx, [esp+12]
724	pand mm5,mm0
725	movd eax,mm5
726	movq mm5,mm2
727	mov eax, [ebx+eax*4]
728
729	L_dodist_mmx:
730
731	movzx ecx,ah
732	mov ebx,eax
733	shr ebx,16
734	sub ebp,ecx
735	movd mm1,ecx
736
737	test al,16
738	jz L_test_for_second_level_dist_mmx
739	and eax,15
740	jz L_check_dist_one_mmx
741
742	L_add_bits_to_dist_mmx:
743	psrlq mm0,mm1
744	movd mm1,eax
745	movd ecx,mm0
746	sub ebp,eax
747	and ecx, [inflate_fast_mask+eax*4]
748	add ebx,ecx
749
750	L_check_window_mmx:
751	mov [esp+44],esi
752	mov eax,edi
753	sub eax, [esp+40]
754
755	cmp eax,ebx
756	jb L_clip_window_mmx
757
758	mov ecx,edx
759	mov esi,edi
760	sub esi,ebx
761
762	sub ecx,3
763	mov al, [esi]
764	mov [edi],al
765	mov al, [esi+1]
766	mov dl, [esi+2]
767	add esi,3
768	mov [edi+1],al
769	mov [edi+2],dl
770	add edi,3
771	rep movsb
772
773	mov esi, [esp+44]
774	mov ebx, [esp+8]
775	jmp L_while_test_mmx
776
777	ALIGN 4
778	L_check_dist_one_mmx:
779	cmp ebx,1
780	jne L_check_window_mmx
781	cmp [esp+40],edi
782	je L_check_window_mmx
783
784	dec edi
785	mov ecx,edx
786	mov al, [edi]
787	sub ecx,3
788
789	mov [edi+1],al
790	mov [edi+2],al
791	mov [edi+3],al
792	add edi,4
793	rep stosb
794
795	mov ebx, [esp+8]
796	jmp L_while_test_mmx
797
798	ALIGN 4
799	L_test_for_second_level_length_mmx:
800	test al,64
801	jnz L_test_for_end_of_block
802
803	and eax,15
804	psrlq mm0,mm1
805	movd ecx,mm0
806	and ecx, [inflate_fast_mask+eax*4]
807	add ecx,edx
808	mov eax, [ebx+ecx*4]
809	jmp L_dolen_mmx
810
811	ALIGN 4
812	L_test_for_second_level_dist_mmx:
813	test al,64
814	jnz L_invalid_distance_code
815
816	and eax,15
817	psrlq mm0,mm1
818	movd ecx,mm0
819	and ecx, [inflate_fast_mask+eax*4]
820	mov eax, [esp+12]
821	add ecx,ebx
822	mov eax, [eax+ecx*4]
823	jmp L_dodist_mmx
824
825	ALIGN 4
826	L_clip_window_mmx:
827
828	mov ecx,eax
829	mov eax, [esp+52]
830	neg ecx
831	mov esi, [esp+56]
832
833	cmp eax,ebx
834	jb L_invalid_distance_too_far
835
836	add ecx,ebx
837	cmp dword ptr [esp+48],0
838	jne L_wrap_around_window_mmx
839
840	sub eax,ecx
841	add esi,eax
842
843	cmp edx,ecx
844	jbe L_do_copy1_mmx
845
846	sub edx,ecx
847	rep movsb
848	mov esi,edi
849	sub esi,ebx
850	jmp L_do_copy1_mmx
851
852	cmp edx,ecx
853	jbe L_do_copy1_mmx
854
855	sub edx,ecx
856	rep movsb
857	mov esi,edi
858	sub esi,ebx
859	jmp L_do_copy1_mmx
860
861	L_wrap_around_window_mmx:
862
863	mov eax, [esp+48]
864	cmp ecx,eax
865	jbe L_contiguous_in_window_mmx
866
867	add esi, [esp+52]
868	add esi,eax
869	sub esi,ecx
870	sub ecx,eax
871
872
873	cmp edx,ecx
874	jbe L_do_copy1_mmx
875
876	sub edx,ecx
877	rep movsb
878	mov esi, [esp+56]
879	mov ecx, [esp+48]
880	cmp edx,ecx
881	jbe L_do_copy1_mmx
882
883	sub edx,ecx
884	rep movsb
885	mov esi,edi
886	sub esi,ebx
887	jmp L_do_copy1_mmx
888
889	L_contiguous_in_window_mmx:
890
891	add esi,eax
892	sub esi,ecx
893
894
895	cmp edx,ecx
896	jbe L_do_copy1_mmx
897
898	sub edx,ecx
899	rep movsb
900	mov esi,edi
901	sub esi,ebx
902
903	L_do_copy1_mmx:
904
905
906	mov ecx,edx
907	rep movsb
908
909	mov esi, [esp+44]
910	mov ebx, [esp+8]
911	jmp L_while_test_mmx
912	; 1174 "inffast.S"
913	L_invalid_distance_code:
914
915
916
917
918
919	mov ecx, invalid_distance_code_msg
920	mov edx,INFLATE_MODE_BAD
921	jmp L_update_stream_state
922
923	L_test_for_end_of_block:
924
925
926
927
928
929	test al,32
930	jz L_invalid_literal_length_code
931
932	mov ecx,0
933	mov edx,INFLATE_MODE_TYPE
934	jmp L_update_stream_state
935
936	L_invalid_literal_length_code:
937
938
939
940
941
942	mov ecx, invalid_literal_length_code_msg
943	mov edx,INFLATE_MODE_BAD
944	jmp L_update_stream_state
945
946	L_invalid_distance_too_far:
947
948
949
950	mov esi, [esp+44]
951	mov ecx, invalid_distance_too_far_msg
952	mov edx,INFLATE_MODE_BAD
953	jmp L_update_stream_state
954
955	L_update_stream_state:
956
957	mov eax, [esp+88]
958	test ecx,ecx
959	jz L_skip_msg
960	mov [eax+24],ecx
961	L_skip_msg:
962	mov eax, [eax+28]
963	mov [eax+mode_state],edx
964	jmp L_break_loop
965
966	ALIGN 4
967	L_break_loop:
968	; 1243 "inffast.S"
969	cmp dword ptr [inflate_fast_use_mmx],2
970	jne L_update_next_in
971
972
973
974	mov ebx,ebp
975
976	L_update_next_in:
977	; 1266 "inffast.S"
978	mov eax, [esp+88]
979	mov ecx,ebx
980	mov edx, [eax+28]
981	shr ecx,3
982	sub esi,ecx
983	shl ecx,3
984	sub ebx,ecx
985	mov [eax+12],edi
986	mov [edx+bits_state],ebx
987	mov ecx,ebx
988
989	lea ebx, [esp+28]
990	cmp [esp+20],ebx
991	jne L_buf_not_used
992
993	sub esi,ebx
994	mov ebx, [eax+0]
995	mov [esp+20],ebx
996	add esi,ebx
997	mov ebx, [eax+4]
998	sub ebx,11
999	add [esp+20],ebx
1000
1001	L_buf_not_used:
1002	mov [eax+0],esi
1003
1004	mov ebx,1
1005	shl ebx,cl
1006	dec ebx
1007
1008
1009
1010
1011
1012	cmp dword ptr [inflate_fast_use_mmx],2
1013	jne L_update_hold
1014
1015
1016
1017	psrlq mm0,mm1
1018	movd ebp,mm0
1019
1020	emms
1021
1022	L_update_hold:
1023
1024
1025
1026	and ebp,ebx
1027	mov [edx+hold_state],ebp
1028
1029
1030
1031
1032	mov ebx, [esp+20]
1033	cmp ebx,esi
1034	jbe L_last_is_smaller
1035
1036	sub ebx,esi
1037	add ebx,11
1038	mov [eax+4],ebx
1039	jmp L_fixup_out
1040	L_last_is_smaller:
1041	sub esi,ebx
1042	neg esi
1043	add esi,11
1044	mov [eax+4],esi
1045
1046
1047
1048
1049	L_fixup_out:
1050
1051	mov ebx, [esp+16]
1052	cmp ebx,edi
1053	jbe L_end_is_smaller
1054
1055	sub ebx,edi
1056	add ebx,257
1057	mov [eax+16],ebx
1058	jmp L_done
1059	L_end_is_smaller:
1060	sub edi,ebx
1061	neg edi
1062	add edi,257
1063	mov [eax+16],edi
1064
1065
1066
1067
1068
1069	L_done:
1070	add esp,64
1071	popfd
1072	pop ebx
1073	pop ebp
1074	pop esi
1075	pop edi
1076	ret
1077	_inflate_fast endp
1078
1079	_TEXT ends
1080	end