subsystems/win32/win32k/dib/i386/dib24bpp_hline.s

   1 /*
   2  * PROJECT:         Win32 subsystem
   3  * LICENSE:         See COPYING in the top level directory
   4  * FILE:            subsystems/win32/win32k/dib/i386/dib24bpp_hline.c
   5  * PURPOSE:         ASM optimised 24bpp HLine
   6  * PROGRAMMERS:     Magnus Olsen
   7  */
   8
   9 .globl _DIB_24BPP_HLine
  10 .intel_syntax noprefix
  11
  12 .def   _DIB_24BPP_HLine;
  13 .scl    2;
  14 .type   32;
  15 .endef
  16       _DIB_24BPP_HLine:
  17                          push    edi
  18                          push    esi
  19                          push    ebx
  20                          sub     esp, 24
  21                          mov     ebx, [esp+40]
  22                          mov     edi, [esp+52]
  23                          mov     ecx, [esp+44]
  24                          mov     eax, [ebx+36]
  25                          mov     esi, [ebx+32]
  26                          mov     edx, [esp+48]
  27                          imul    eax, edi
  28                          sub     edx, ecx
  29                          mov     [esp], edx
  30                          add     eax, esi
  31                          lea     eax, [eax+ecx*2]
  32                          add     eax, ecx
  33                          cmp     edx, 7
  34                          mov     esi, edx
  35                          mov     [esp+4], eax
  36                          ja      Align4byte
  37                          lea     eax, [edx-1]
  38                          mov     [esp], eax
  39                          inc     eax
  40                          jnz     small_fill
  41                          add     esp, 24
  42                          pop     ebx
  43                          pop     esi
  44                          pop     edi
  45                          ret
  46
  47                         /* For small fills, don't bother doing anything fancy */
  48                small_fill:
  49                          movzx   ecx, word ptr [esp+58]
  50                          mov     edx, [esp+4]
  51                          mov     esi, [esp+56]
  52                          lea     eax, [edx+2]
  53                          mov     [esp+4], eax
  54                          mov     [edx+2], cl
  55                          mov     eax, [esp]
  56                          inc     dword ptr [esp+4]
  57                          mov     [edx], si
  58                          dec     eax
  59                          mov     [esp], eax
  60                          inc     eax
  61                          jnz     small_fill
  62                          add     esp, 24
  63                          pop     ebx
  64                          pop     esi
  65                          pop     edi
  66                          ret
  67
  68                Align4byte:
  69                          /* Align to 4-byte address */
  70                          test    al, 3
  71                          mov     ecx, eax
  72                          jz      loop1
  73                          lea     esi, [esi+0]
  74                          lea     edi, [edi+0]
  75
  76            loopasmversion:
  77                         /* This is about 30% faster than the generic C code below */
  78                          movzx   edx, word ptr [esp+58]
  79                          lea     edi, [ecx+2]
  80                          mov     eax, [esp+56]
  81                          mov     [esp+4], edi
  82                          mov     [ecx+2], dl
  83                          mov     ebx, [esp+4]
  84                          mov     [ecx], ax
  85                          mov     edx, [esp]
  86                          inc     ebx
  87                          mov     [esp+4], ebx
  88                          dec     edx
  89                          test    bl, 3
  90                          mov     [esp], edx
  91                          mov     ecx, ebx
  92                          jnz     loopasmversion
  93                          mov     esi, edx
  94
  95                    loop1:
  96                          mov     ecx, [esp+56]
  97                          and     ecx, 16777215
  98                          mov     ebx, ecx
  99                          shr     ebx, 8
 100                          mov     eax, ecx
 101                          shl     eax, 16
 102                          or      ebx, eax
 103                          mov     edx, ecx
 104                          shl     edx, 8
 105                          mov     eax, ecx
 106                          shr     eax, 16
 107                          or      edx, eax
 108                          mov     eax, ecx
 109                          shl     eax, 24
 110                          or      eax, ecx
 111                          mov     ecx, [esp]
 112                          shr     ecx, 2
 113                          mov     edi, [esp+4]
 114                     loop2:
 115                          mov     [edi], eax
 116                          mov     [edi+4], ebx
 117                          mov     [edi+8], edx
 118                          add     edi, 12
 119                          dec     ecx
 120                          jnz     loop2
 121                          mov     [esp+4], edi
 122                          and     esi, 3
 123                          lea     eax, [esi-1]
 124                          mov     [esp], eax
 125                          inc     eax
 126                          jnz     leftoverfromthemainloop
 127                          add     esp, 24
 128                          pop     ebx
 129                          pop     esi
 130                          pop     edi
 131                          ret
 132
 133   leftoverfromthemainloop:
 134
 135                         /*  Count = Count & 0x03; */
 136                          mov     ecx, [esp+4]
 137                          mov     ebx, [esp+56]
 138                          lea     esi, [ecx+2]
 139                          mov     [ecx], bx
 140                          shr     ebx, 16
 141                          mov     [esp+4], esi
 142                          mov     [ecx+2], bl
 143                          mov     eax, [esp]
 144                          inc     dword ptr [esp+4]
 145                          dec     eax
 146                          mov     [esp], eax
 147                          inc     eax
 148                          jnz     leftoverfromthemainloop
 149                          add     esp, 24
 150                          pop     ebx
 151                          pop     esi
 152                          pop     edi
 153                          ret