HexRays generates the following code:
void __cdecl Psx_gte_ncs_446930()
{
double vx_scaled; // st7
double vy_scaled; // st6
double vz_scaled; // st5
double light_matrix_2; // st7
double light_colour_matrix_3; // st7
char v5; // cl
char never_written_v7; // c0
char v9; // c0
char never_written_v11; // c0
double light_matrix_0; // [esp+0h] [ebp-24h]
double b_value; // [esp+0h] [ebp-24h]
double light_matrix_1; // [esp+8h] [ebp-1Ch]
double light_colour_matrix_2; // [esp+10h] [ebp-14h]
double g_value; // [esp+10h] [ebp-14h]
double light_colour_matrix_1; // [esp+18h] [ebp-Ch]
double r_value; // [esp+18h] [ebp-Ch]
++gGteData_722688.gte_ncs_count_72268C; // Normal color single
vx_scaled = (double)gGte_VXY0_993EC0.regs.VX * 0.000244140625;
vy_scaled = (double)gGte_VXY0_993EC0.regs.VY * 0.000244140625;
vz_scaled = (double)gGte_VXY0_993EC0.regs.VZ * 0.000244140625;
light_matrix_0 = ((double)gGte_light_source_matrix_993E60.m[0][2] * vz_scaled
+ (double)gGte_light_source_matrix_993E60.m[0][1] * vy_scaled
+ (double)gGte_light_source_matrix_993E60.m[0][0] * vx_scaled)
* 0.000244140625;
light_matrix_1 = ((double)gGte_light_source_matrix_993E60.m[1][2] * vz_scaled
+ (double)gGte_light_source_matrix_993E60.m[1][1] * vy_scaled
+ (double)gGte_light_source_matrix_993E60.m[1][0] * vx_scaled)
* 0.000244140625;
light_matrix_2 = ((double)gGte_light_source_matrix_993E60.m[2][2] * vz_scaled
+ (double)gGte_light_source_matrix_993E60.m[2][1] * vy_scaled
+ (double)gGte_light_source_matrix_993E60.m[2][0] * vx_scaled)
* 0.000244140625;
if ( light_matrix_0 < 0.0 )
light_matrix_0 = 0.0;
if ( light_matrix_1 < 0.0 )
light_matrix_1 = 0.0;
if ( light_matrix_2 < 0.0 )
light_matrix_2 = 0.0;
if ( light_matrix_0 > 7.999 )
light_matrix_0 = 7.999;
if ( light_matrix_1 > 7.999 )
light_matrix_1 = 7.999;
if ( light_matrix_2 > 7.999 )
light_matrix_2 = 7.999;
light_colour_matrix_1 = ((double)gGte_light_colour_matrix_source_993E80.m[0][2] * light_matrix_2
+ (double)gGte_light_colour_matrix_source_993E80.m[0][1] * light_matrix_1
+ (double)gGte_light_colour_matrix_source_993E80.m[0][0] * light_matrix_0
+ (double)gGte_background_colour_993E74.x)
* 0.000244140625;
light_colour_matrix_2 = ((double)gGte_light_colour_matrix_source_993E80.m[1][2] * light_matrix_2
+ (double)gGte_light_colour_matrix_source_993E80.m[1][1] * light_matrix_1
+ (double)gGte_light_colour_matrix_source_993E80.m[1][0] * light_matrix_0
+ (double)gGte_background_colour_993E74.y)
* 0.000244140625;
light_colour_matrix_3 = (light_matrix_2 * (double)gGte_light_colour_matrix_source_993E80.m[2][2]
+ (double)gGte_light_colour_matrix_source_993E80.m[2][1] * light_matrix_1
+ (double)gGte_light_colour_matrix_source_993E80.m[2][0] * light_matrix_0
+ (double)gGte_background_colour_993E74.z)
* 0.000244140625;
if ( light_colour_matrix_1 < 0.0 )
light_colour_matrix_1 = 0.0;
if ( light_colour_matrix_2 < 0.0 )
light_colour_matrix_2 = 0.0;
if ( light_colour_matrix_3 < 0.0 )
light_colour_matrix_3 = 0.0;
if ( light_colour_matrix_1 > 7.999 )
light_colour_matrix_1 = 7.999;
if ( light_colour_matrix_2 > 7.999 )
light_colour_matrix_2 = 7.999;
if ( light_colour_matrix_3 > 7.999 )
light_colour_matrix_3 = 7.999;
gGte_RGB0_993F10.cd = gGte_RGB1_993F14.cd;
r_value = (double)gGte_r_993ED8 * light_colour_matrix_1 * 0.00390625;
gGte_RGB1_993F14.cd = gGte_RGB2_993F18.cd;
gGte_RGB2_993F18.cd = gGte_OTZ_993EDB;
gGte_RGB0_993F10.r = gGte_RGB1_993F14.r;
gGte_RGB1_993F14.r = gGte_RGB2_993F18.r;
v5 = 0;
g_value = (double)gGte_g_993ED9 * light_colour_matrix_2 * 0.00390625;
b_value = (double)gGte_b_993EDA * light_colour_matrix_3 * 0.00390625;
if ( never_written_v7 )
{
gGte_RGB2_993F18.r = 0;
}
else if ( r_value * 256.0 <= 255.0 )
{
gGte_RGB2_993F18.r = (signed int)(r_value * 256.0);
}
else
{
gGte_RGB2_993F18.r = -1;
}
gGte_RGB0_993F10.g = gGte_RGB1_993F14.g;
gGte_RGB1_993F14.g = gGte_RGB2_993F18.g;
if ( v9 )
{
gGte_RGB2_993F18.g = 0;
}
else if ( g_value * 256.0 <= 255.0 )
{
gGte_RGB2_993F18.g = (signed int)(g_value * 256.0);
}
else
{
gGte_RGB2_993F18.g = -1;
}
gGte_RGB0_993F10.b = gGte_RGB1_993F14.b;
gGte_RGB1_993F14.b = gGte_RGB2_993F18.b;
if ( never_written_v11 )
goto LABEL_39;
if ( b_value * 256.0 <= 255.0 )
{
v5 = (signed int)(b_value * 256.0);
LABEL_39:
gGte_RGB2_993F18.b = v5;
goto LABEL_40;
}
gGte_RGB2_993F18.b = -1;
LABEL_40:
gGte_MAC1_993F24.MAC_32 = (signed int)(r_value * 4096.0);
gGte_MAC2_993F28.MAC_32 = (signed int)(g_value * 4096.0);
gGte_MAC3_993F2C.MAC_32 = (signed int)(b_value * 4096.0);
}
Which comes from the following asm:
.text:00446930
.text:00446930 ; Attributes: bp-based frame
.text:00446930
.text:00446930 ; void __cdecl Psx_gte_ncs_446930()
.text:00446930 Psx_gte_ncs_446930 proc near ; CODE XREF: sub_443D4F+238↑p
.text:00446930
.text:00446930 light_matrix_0 = qword ptr -24h
.text:00446930 light_matrix_1 = qword ptr -1Ch
.text:00446930 light_colour_matrix_2= qword ptr -14h
.text:00446930 light_colour_matrix_1= qword ptr -0Ch
.text:00446930 var_4 = dword ptr -4
.text:00446930
.text:00446930 push ebp
.text:00446931 mov ebp, esp
.text:00446933 sub esp, 24h
.text:00446936 inc gGteData_722688.gte_ncs_count_72268C
.text:0044693C movsx eax, word ptr gGte_VXY0_993EC0
.text:00446943 movsx ecx, word ptr gGte_VXY0_993EC0+2
.text:0044694A mov [ebp+var_4], eax
.text:0044694D fild [ebp+var_4]
.text:00446950 movsx edx, word ptr gGte_VXY0_993EC0+4
.text:00446957 fmul ds:dbl_64BF78
.text:0044695D mov [ebp+var_4], ecx
.text:00446960 fild [ebp+var_4]
.text:00446963 movsx eax, gGte_light_source_matrix_993E60.m+4
.text:0044696A fmul ds:dbl_64BF78
.text:00446970 mov [ebp+var_4], edx
.text:00446973 fild [ebp+var_4]
.text:00446976 movsx ecx, gGte_light_source_matrix_993E60.m+2
.text:0044697D fmul ds:dbl_64BF78
.text:00446983 mov [ebp+var_4], eax
.text:00446986 fild [ebp+var_4]
.text:00446989 movsx edx, gGte_light_source_matrix_993E60.m
.text:00446990 fmul st, st(1)
.text:00446992 mov [ebp+var_4], ecx
.text:00446995 fild [ebp+var_4]
.text:00446998 mov [ebp+var_4], edx
.text:0044699B movsx eax, gGte_light_source_matrix_993E60.m+0Ah
.text:004469A2 fmul st, st(3)
.text:004469A4 movsx ecx, gGte_light_source_matrix_993E60.m+8
.text:004469AB faddp st(1), st
.text:004469AD fild [ebp+var_4]
.text:004469B0 mov [ebp+var_4], eax
.text:004469B3 movsx edx, gGte_light_source_matrix_993E60.m+6
.text:004469BA fmul st, st(4)
.text:004469BC movsx eax, gGte_light_source_matrix_993E60.m+10h
.text:004469C3 faddp st(1), st
.text:004469C5 fmul ds:dbl_64BF78
.text:004469CB fstp [ebp+light_matrix_0]
.text:004469CE fild [ebp+var_4]
.text:004469D1 mov [ebp+var_4], ecx
.text:004469D4 movsx ecx, gGte_light_source_matrix_993E60.m+0Eh
.text:004469DB fmul st, st(1)
.text:004469DD fild [ebp+var_4]
.text:004469E0 mov [ebp+var_4], edx
.text:004469E3 movsx edx, gGte_light_source_matrix_993E60.m+0Ch
.text:004469EA fmul st, st(3)
.text:004469EC faddp st(1), st
.text:004469EE fild [ebp+var_4]
.text:004469F1 mov [ebp+var_4], eax
.text:004469F4 fmul st, st(4)
.text:004469F6 faddp st(1), st
.text:004469F8 fmul ds:dbl_64BF78
.text:004469FE fstp [ebp+light_matrix_1]
.text:00446A01 fild [ebp+var_4]
.text:00446A04 mov [ebp+var_4], ecx
.text:00446A07 fmul st, st(1)
.text:00446A09 fild [ebp+var_4]
.text:00446A0C mov [ebp+var_4], edx
.text:00446A0F fmul st, st(3)
.text:00446A11 faddp st(1), st
.text:00446A13 fild [ebp+var_4]
.text:00446A16 fmul st, st(4)
.text:00446A18 faddp st(1), st
.text:00446A1A fmul ds:dbl_64BF78
.text:00446A20 fstp st(3)
.text:00446A22 fstp st
.text:00446A24 fstp st
.text:00446A26 fld [ebp+light_matrix_0]
.text:00446A29 fcomp ds:dbl_64A328
.text:00446A2F fnstsw ax
.text:00446A31 test ah, 1
.text:00446A34 jz short loc_446A44
.text:00446A36 mov dword ptr [ebp+light_matrix_0], 0
.text:00446A3D mov dword ptr [ebp+light_matrix_0+4], 0
.text:00446A44
.text:00446A44 loc_446A44: ; CODE XREF: Psx_gte_ncs_446930+104↑j
.text:00446A44 fld [ebp+light_matrix_1]
.text:00446A47 fcomp ds:dbl_64A328
.text:00446A4D fnstsw ax
.text:00446A4F test ah, 1
.text:00446A52 jz short loc_446A62
.text:00446A54 mov dword ptr [ebp+light_matrix_1], 0
.text:00446A5B mov dword ptr [ebp+light_matrix_1+4], 0
.text:00446A62
.text:00446A62 loc_446A62: ; CODE XREF: Psx_gte_ncs_446930+122↑j
.text:00446A62 fcom ds:dbl_64A328
.text:00446A68 fnstsw ax
.text:00446A6A test ah, 1
.text:00446A6D jz short loc_446A77
.text:00446A6F fstp st
.text:00446A71 fld ds:dbl_64A328
.text:00446A77
.text:00446A77 loc_446A77: ; CODE XREF: Psx_gte_ncs_446930+13D↑j
.text:00446A77 fld [ebp+light_matrix_0]
.text:00446A7A fcomp ds:dbl_64BFB0
.text:00446A80 fnstsw ax
.text:00446A82 test ah, 41h
.text:00446A85 jnz short loc_446A95
.text:00446A87 mov dword ptr [ebp+light_matrix_0], 0DB22D0E5h
.text:00446A8E mov dword ptr [ebp+light_matrix_0+4], 401FFEF9h
.text:00446A95
.text:00446A95 loc_446A95: ; CODE XREF: Psx_gte_ncs_446930+155↑j
.text:00446A95 fld [ebp+light_matrix_1]
.text:00446A98 fcomp ds:dbl_64BFB0
.text:00446A9E fnstsw ax
.text:00446AA0 test ah, 41h
.text:00446AA3 jnz short loc_446AB3
.text:00446AA5 mov dword ptr [ebp+light_matrix_1], 0DB22D0E5h
.text:00446AAC mov dword ptr [ebp+light_matrix_1+4], 401FFEF9h
.text:00446AB3
.text:00446AB3 loc_446AB3: ; CODE XREF: Psx_gte_ncs_446930+173↑j
.text:00446AB3 fcom ds:dbl_64BFB0
.text:00446AB9 fnstsw ax
.text:00446ABB test ah, 41h
.text:00446ABE jnz short loc_446AC8
.text:00446AC0 fstp st
.text:00446AC2 fld ds:dbl_64BFB0
.text:00446AC8
.text:00446AC8 loc_446AC8: ; CODE XREF: Psx_gte_ncs_446930+18E↑j
.text:00446AC8 movsx eax, gGte_light_colour_matrix_source_993E80.m+4
.text:00446ACF movsx ecx, gGte_light_colour_matrix_source_993E80.m+2
.text:00446AD6 mov [ebp+var_4], eax
.text:00446AD9 fild [ebp+var_4]
.text:00446ADC movsx edx, gGte_light_colour_matrix_source_993E80.m
.text:00446AE3 fmul st, st(1)
.text:00446AE5 mov [ebp+var_4], ecx
.text:00446AE8 fild [ebp+var_4]
.text:00446AEB mov [ebp+var_4], edx
.text:00446AEE movsx eax, gGte_light_colour_matrix_source_993E80.m+0Ah
.text:00446AF5 fmul [ebp+light_matrix_1]
.text:00446AF8 movsx ecx, gGte_light_colour_matrix_source_993E80.m+8
.text:00446AFF faddp st(1), st
.text:00446B01 fild [ebp+var_4]
.text:00446B04 mov [ebp+var_4], eax
.text:00446B07 movsx edx, gGte_light_colour_matrix_source_993E80.m+6
.text:00446B0E fmul [ebp+light_matrix_0]
.text:00446B11 movsx eax, gGte_light_colour_matrix_source_993E80.m+10h
.text:00446B18 faddp st(1), st
.text:00446B1A fild gGte_background_colour_993E74.x
.text:00446B20 faddp st(1), st
.text:00446B22 fmul ds:dbl_64BF78
.text:00446B28 fstp [ebp+light_colour_matrix_1]
.text:00446B2B fild [ebp+var_4]
.text:00446B2E mov [ebp+var_4], ecx
.text:00446B31 movsx ecx, gGte_light_colour_matrix_source_993E80.m+0Eh
.text:00446B38 fmul st, st(1)
.text:00446B3A fild [ebp+var_4]
.text:00446B3D mov [ebp+var_4], edx
.text:00446B40 movsx edx, gGte_light_colour_matrix_source_993E80.m+0Ch
.text:00446B47 fmul [ebp+light_matrix_1]
.text:00446B4A faddp st(1), st
.text:00446B4C fild [ebp+var_4]
.text:00446B4F mov [ebp+var_4], eax
.text:00446B52 fmul [ebp+light_matrix_0]
.text:00446B55 faddp st(1), st
.text:00446B57 fild gGte_background_colour_993E74.y
.text:00446B5D faddp st(1), st
.text:00446B5F fmul ds:dbl_64BF78
.text:00446B65 fstp [ebp+light_colour_matrix_2]
.text:00446B68 fild [ebp+var_4]
.text:00446B6B mov [ebp+var_4], ecx
.text:00446B6E fmulp st(1), st
.text:00446B70 fild [ebp+var_4]
.text:00446B73 mov [ebp+var_4], edx
.text:00446B76 fmul [ebp+light_matrix_1]
.text:00446B79 faddp st(1), st
.text:00446B7B fild [ebp+var_4]
.text:00446B7E fmul [ebp+light_matrix_0]
.text:00446B81 faddp st(1), st
.text:00446B83 fild gGte_background_colour_993E74.z
.text:00446B89 faddp st(1), st
.text:00446B8B fmul ds:dbl_64BF78
.text:00446B91 fld [ebp+light_colour_matrix_1]
.text:00446B94 fcomp ds:dbl_64A328
.text:00446B9A fnstsw ax
.text:00446B9C test ah, 1
.text:00446B9F jz short loc_446BAF
.text:00446BA1 mov dword ptr [ebp+light_colour_matrix_1], 0
.text:00446BA8 mov dword ptr [ebp+light_colour_matrix_1+4], 0
.text:00446BAF
.text:00446BAF loc_446BAF: ; CODE XREF: Psx_gte_ncs_446930+26F↑j
.text:00446BAF fld [ebp+light_colour_matrix_2]
.text:00446BB2 fcomp ds:dbl_64A328
.text:00446BB8 fnstsw ax
.text:00446BBA test ah, 1
.text:00446BBD jz short loc_446BCD
.text:00446BBF mov dword ptr [ebp+light_colour_matrix_2], 0
.text:00446BC6 mov dword ptr [ebp+light_colour_matrix_2+4], 0
.text:00446BCD
.text:00446BCD loc_446BCD: ; CODE XREF: Psx_gte_ncs_446930+28D↑j
.text:00446BCD fcom ds:dbl_64A328
.text:00446BD3 fnstsw ax
.text:00446BD5 test ah, 1
.text:00446BD8 jz short loc_446BE2
.text:00446BDA fstp st
.text:00446BDC fld ds:dbl_64A328
.text:00446BE2
.text:00446BE2 loc_446BE2: ; CODE XREF: Psx_gte_ncs_446930+2A8↑j
.text:00446BE2 fld [ebp+light_colour_matrix_1]
.text:00446BE5 fcomp ds:dbl_64BFB0
.text:00446BEB fnstsw ax
.text:00446BED test ah, 41h
.text:00446BF0 jnz short loc_446C00
.text:00446BF2 mov dword ptr [ebp+light_colour_matrix_1], 0DB22D0E5h
.text:00446BF9 mov dword ptr [ebp+light_colour_matrix_1+4], 401FFEF9h
.text:00446C00
.text:00446C00 loc_446C00: ; CODE XREF: Psx_gte_ncs_446930+2C0↑j
.text:00446C00 fld [ebp+light_colour_matrix_2]
.text:00446C03 fcomp ds:dbl_64BFB0
.text:00446C09 fnstsw ax
.text:00446C0B test ah, 41h
.text:00446C0E jnz short loc_446C1E
.text:00446C10 mov dword ptr [ebp+light_colour_matrix_2], 0DB22D0E5h
.text:00446C17 mov dword ptr [ebp+light_colour_matrix_2+4], 401FFEF9h
.text:00446C1E
.text:00446C1E loc_446C1E: ; CODE XREF: Psx_gte_ncs_446930+2DE↑j
.text:00446C1E fcom ds:dbl_64BFB0
.text:00446C24 fnstsw ax
.text:00446C26 test ah, 41h
.text:00446C29 jnz short loc_446C33
.text:00446C2B fstp st
.text:00446C2D fld ds:dbl_64BFB0
.text:00446C33
.text:00446C33 loc_446C33: ; CODE XREF: Psx_gte_ncs_446930+2F9↑j
.text:00446C33 mov eax, dword ptr gGte_r_993ED8
.text:00446C38 mov edx, 0FFh
.text:00446C3D and eax, edx
.text:00446C3F xor ecx, ecx
.text:00446C41 mov [ebp+var_4], eax
.text:00446C44 mov cl, gGte_g_993ED9
.text:00446C4A fild [ebp+var_4]
.text:00446C4D mov [ebp+var_4], ecx
.text:00446C50 xor eax, eax
.text:00446C52 mov al, gGte_b_993EDA
.text:00446C57 mov cl, gGte_RGB1_993F14.cd
.text:00446C5D fmul [ebp+light_colour_matrix_1]
.text:00446C60 mov gGte_RGB0_993F10.cd, cl
.text:00446C66 mov cl, byte ptr gGte_OTZ_993EDB
.text:00446C6C fmul ds:dbl_64BFA8
.text:00446C72 fstp [ebp+light_colour_matrix_1]
.text:00446C75 fild [ebp+var_4]
.text:00446C78 mov [ebp+var_4], eax
.text:00446C7B mov al, gGte_RGB2_993F18.cd
.text:00446C80 mov gGte_RGB1_993F14.cd, al
.text:00446C85 mov al, gGte_RGB1_993F14.r
.text:00446C8A fmul [ebp+light_colour_matrix_2]
.text:00446C8D mov gGte_RGB2_993F18.cd, cl
.text:00446C93 mov cl, gGte_RGB2_993F18.r
.text:00446C99 mov gGte_RGB0_993F10.r, al
.text:00446C9E mov gGte_RGB1_993F14.r, cl
.text:00446CA4 fmul ds:dbl_64BFA8
.text:00446CAA xor cl, cl
.text:00446CAC fstp [ebp+light_colour_matrix_2]
.text:00446CAF fild [ebp+var_4]
.text:00446CB2 fmul st, st(1)
.text:00446CB4 fmul ds:dbl_64BFA8
.text:00446CBA fstp [ebp+light_matrix_0]
.text:00446CBD fstp st
.text:00446CBF fld [ebp+light_colour_matrix_1]
.text:00446CC2 fmul ds:dbl_64BF90
.text:00446CC8 fcom ds:dbl_64A328
.text:00446CCE fst [ebp+light_matrix_1]
.text:00446CD1 fnstsw ax
.text:00446CD3 test ah, 1
.text:00446CD6 jz short loc_446CE2
.text:00446CD8 fstp st
.text:00446CDA mov gGte_RGB2_993F18.r, cl
.text:00446CE0 jmp short loc_446D05
.text:00446CE2
.text:00446CE2 loc_446CE2: ; CODE XREF: Psx_gte_ncs_446930+3A6↑j
.text:00446CE2 fcomp ds:dbl_64BF88
.text:00446CE8 fnstsw ax
.text:00446CEA test ah, 41h
.text:00446CED jnz short loc_446CF7
.text:00446CEF mov gGte_RGB2_993F18.r, dl
.text:00446CF5 jmp short loc_446D05
.text:00446CF7
.text:00446CF7 loc_446CF7: ; CODE XREF: Psx_gte_ncs_446930+3BD↑j
.text:00446CF7 fld [ebp+light_matrix_1]
.text:00446CFA fistp [ebp+var_4]
.text:00446CFD mov al, byte ptr [ebp+var_4]
.text:00446D00 mov gGte_RGB2_993F18.r, al
.text:00446D05
.text:00446D05 loc_446D05: ; CODE XREF: Psx_gte_ncs_446930+3B0↑j
.text:00446D05 ; Psx_gte_ncs_446930+3C5↑j
.text:00446D05 fld [ebp+light_colour_matrix_2]
.text:00446D08 mov al, gGte_RGB1_993F14.g
.text:00446D0D fmul ds:dbl_64BF90
.text:00446D13 mov gGte_RGB0_993F10.g, al
.text:00446D18 mov al, gGte_RGB2_993F18.g
.text:00446D1D mov gGte_RGB1_993F14.g, al
.text:00446D22 fcom ds:dbl_64A328
.text:00446D28 fst [ebp+light_matrix_1]
.text:00446D2B fnstsw ax
.text:00446D2D test ah, 1
.text:00446D30 jz short loc_446D3C
.text:00446D32 fstp st
.text:00446D34 mov gGte_RGB2_993F18.g, cl
.text:00446D3A jmp short loc_446D5F
.text:00446D3C
.text:00446D3C loc_446D3C: ; CODE XREF: Psx_gte_ncs_446930+400↑j
.text:00446D3C fcomp ds:dbl_64BF88
.text:00446D42 fnstsw ax
.text:00446D44 test ah, 41h
.text:00446D47 jnz short loc_446D51
.text:00446D49 mov gGte_RGB2_993F18.g, dl
.text:00446D4F jmp short loc_446D5F
.text:00446D51
.text:00446D51 loc_446D51: ; CODE XREF: Psx_gte_ncs_446930+417↑j
.text:00446D51 fld [ebp+light_matrix_1]
.text:00446D54 fistp [ebp+var_4]
.text:00446D57 mov al, byte ptr [ebp+var_4]
.text:00446D5A mov gGte_RGB2_993F18.g, al
.text:00446D5F
.text:00446D5F loc_446D5F: ; CODE XREF: Psx_gte_ncs_446930+40A↑j
.text:00446D5F ; Psx_gte_ncs_446930+41F↑j
.text:00446D5F fld [ebp+light_matrix_0]
.text:00446D62 mov al, gGte_RGB1_993F14.b
.text:00446D67 fmul ds:dbl_64BF90
.text:00446D6D mov gGte_RGB0_993F10.b, al
.text:00446D72 mov al, gGte_RGB2_993F18.b
.text:00446D77 mov gGte_RGB1_993F14.b, al
.text:00446D7C fcom ds:dbl_64A328
.text:00446D82 fst [ebp+light_matrix_1]
.text:00446D85 fnstsw ax
.text:00446D87 test ah, 1
.text:00446D8A jz short loc_446D90
.text:00446D8C fstp st
.text:00446D8E jmp short loc_446DAE
.text:00446D90
.text:00446D90 loc_446D90: ; CODE XREF: Psx_gte_ncs_446930+45A↑j
.text:00446D90 fcomp ds:dbl_64BF88
.text:00446D96 fnstsw ax
.text:00446D98 test ah, 41h
.text:00446D9B jnz short loc_446DA5
.text:00446D9D mov gGte_RGB2_993F18.b, dl
.text:00446DA3 jmp short loc_446DB4
.text:00446DA5
.text:00446DA5 loc_446DA5: ; CODE XREF: Psx_gte_ncs_446930+46B↑j
.text:00446DA5 fld [ebp+light_matrix_1]
.text:00446DA8 fistp [ebp+var_4]
.text:00446DAB mov cl, byte ptr [ebp+var_4]
.text:00446DAE
.text:00446DAE loc_446DAE: ; CODE XREF: Psx_gte_ncs_446930+45E↑j
.text:00446DAE mov gGte_RGB2_993F18.b, cl
.text:00446DB4
.text:00446DB4 loc_446DB4: ; CODE XREF: Psx_gte_ncs_446930+473↑j
.text:00446DB4 fld [ebp+light_colour_matrix_1]
.text:00446DB7 fmul ds:dbl_64BF70
.text:00446DBD fstp [ebp+light_matrix_1]
.text:00446DC0 fld [ebp+light_matrix_1]
.text:00446DC3 fistp [ebp+var_4]
.text:00446DC6 fld [ebp+light_colour_matrix_2]
.text:00446DC9 fmul ds:dbl_64BF70
.text:00446DCF mov edx, [ebp+var_4]
.text:00446DD2 mov dword ptr gGte_MAC1_993F24, edx
.text:00446DD8 fstp [ebp+light_matrix_1]
.text:00446DDB fld [ebp+light_matrix_1]
.text:00446DDE fistp [ebp+var_4]
.text:00446DE1 fld [ebp+light_matrix_0]
.text:00446DE4 fmul ds:dbl_64BF70
.text:00446DEA mov eax, [ebp+var_4]
.text:00446DED mov dword ptr gGte_MAC2_993F28, eax
.text:00446DF2 fstp [ebp+light_matrix_0]
.text:00446DF5 fld [ebp+light_matrix_0]
.text:00446DF8 fistp [ebp+var_4]
.text:00446DFB mov ecx, [ebp+var_4]
.text:00446DFE mov dword ptr gGte_MAC3_993F2C, ecx
.text:00446E04 mov esp, ebp
.text:00446E06 pop ebp
.text:00446E07 retn
.text:00446E07 Psx_gte_ncs_446930 endp
Now what I can't figure out is how should the code around:
.text:00446CC2
fmul ds:dbl_64BF90
fcom ds:dbl_64A328
fst [ebp+light_matrix_1]
fnstsw ax
test ah, 1
jz short loc_446CE2
fstp st
mov gGte_RGB2_993F18.r, cl
jmp short loc_446D05
Should be decompiled. Since
b_value = (double)gGte_b_993EDA * light_colour_matrix_3 * 0.00390625;
if ( never_written_v7 )
{
gGte_RGB2_993F18.r = 0;
}
Can't be correct if never_written_v7 is never written.
After much digging I managed to scale down the issue to this:
Original source code:
#include <stdio.h>
struct Reg_RGB
{
char r, g, b, cd;
};
Reg_RGB gGte_RGB2_993F18;
short int field_0_v = 123;
unsigned char gGte_r_993ED8 = 99;
unsigned char gGte_g_993ED9 = 15;
void Test2()
{
double g_value = (double)gGte_g_993ED9 * field_0_v;
double r_value = (double)gGte_r_993ED8 * field_0_v;
if ( r_value * 256.0 < 0.0)
{
gGte_RGB2_993F18.r = 0;
}
else if ( r_value * 256.0 <= 255.0 )
{
gGte_RGB2_993F18.r = (signed int)(r_value * 256.0);
}
if (g_value * 256.0 < 255.0)
{
printf("Hello1\n");
}
}
Which VC5 generates the following asm for:
.text:00401000 ; =============== S U B R O U T I N E =======================================
.text:00401000
.text:00401000
.text:00401000 ; void __cdecl Test2()
.text:00401000 public ?Test2@@YAXXZ
.text:00401000 ?Test2@@YAXXZ proc near ; DATA XREF: .rdata:off_408488↓o
.text:00401000
.text:00401000 var_C = dword ptr -0Ch
.text:00401000 g_value = qword ptr -8
.text:00401000
.text:00401000 sub esp, 0Ch
.text:00401003 movsx eax, field_0_v ; 123
.text:0040100A mov ecx, gGte_r_993ED8 ; 99
.text:00401010 mov edx, gGte_g_993ED9 ; 15
.text:00401016 mov [esp+0Ch+var_C], eax
.text:0040101A and ecx, 0FFh
.text:00401020 fild [esp+0Ch+var_C]
.text:00401024 mov [esp+0Ch+var_C], ecx
.text:00401028 and edx, 0FFh
.text:0040102E fild [esp+0Ch+var_C]
.text:00401032 fxch st(1)
.text:00401034 fstp [esp+0Ch+g_value]
.text:00401038 fmul [esp+0Ch+g_value]
.text:0040103C mov [esp+0Ch+var_C], edx
.text:00401040 fild [esp+0Ch+var_C]
.text:00401044 fxch st(1)
.text:00401046 fmul ds:dbl_408000 ; 256.0
.text:0040104C fxch st(1)
.text:0040104E fmul [esp+0Ch+g_value]
.text:00401052 fld st(1)
.text:00401054 fcomp ds:dbl_408008 ; 0.0
.text:0040105A fstp [esp+0Ch+g_value] ; any other instruction between fcom/fcomp and fnstsw will break HexRays decompiler
.text:0040105E fnstsw ax
.text:00401060 test ah, 1
.text:00401063 jz short loc_401070
.text:00401065 fstp st
.text:00401067 mov gGte_RGB2_993F18, 0
.text:0040106E jmp short loc_40108B
.text:00401070 ; ---------------------------------------------------------------------------
.text:00401070
.text:00401070 loc_401070: ; CODE XREF: Test2(void)+63↑j
.text:00401070 fcom ds:dbl_408010
.text:00401076 fnstsw ax
.text:00401078 test ah, 41h
.text:0040107B jz short loc_401089
.text:0040107D call __ftol
.text:00401082 mov gGte_RGB2_993F18, al
.text:00401087 jmp short loc_40108B
.text:00401089 ; ---------------------------------------------------------------------------
.text:00401089
.text:00401089 loc_401089: ; CODE XREF: Test2(void)+7B↑j
.text:00401089 fstp st
.text:0040108B
.text:0040108B loc_40108B: ; CODE XREF: Test2(void)+6E↑j
.text:0040108B ; Test2(void)+87↑j
.text:0040108B fld [esp+0Ch+g_value]
.text:0040108F fmul ds:dbl_408000
.text:00401095 fcomp ds:dbl_408010
.text:0040109B fnstsw ax
.text:0040109D test ah, 1
.text:004010A0 jz short loc_4010AF
.text:004010A2 push offset aHello1 ; "Hello1\n"
.text:004010A7 call _printf
.text:004010AC add esp, 4
.text:004010AF
.text:004010AF loc_4010AF: ; CODE XREF: Test2(void)+A0↑j
.text:004010AF add esp, 0Ch
.text:004010B2 retn
.text:004010B2 ?Test2@@YAXXZ endp
And HexRays decompiler generates:
void __cdecl Test2()
{
double temp; // ST08_8
double r_value; // st7
char never_written; // c0
double g_value; // [esp+4h] [ebp-8h]
temp = (double)field_0_v;
r_value = (double)(unsigned __int8)gGte_r_993ED8 * temp * 256.0;
g_value = (double)(unsigned __int8)gGte_g_993ED9 * temp;
if ( never_written )
{
gGte_RGB2_993F18 = 0;
}
else if ( r_value <= 255.0 )
{
gGte_RGB2_993F18 = (signed __int64)r_value;
}
if ( g_value * 256.0 < 255.0 )
printf("Hello1\n");
}
Here we can see that the decompiler has again generated incorrect output because char never_written; // c0 should never exist and it should actually be using r_value instead.
This limitation or bug in the decompiler seems to trigger when other floating point instructions are executed after fcom fcomp and before fnstsw. In this case it appears that fstp is what causes the broken output.
Re-arranging the code so that the sequnce is fcom followed by fnstsw resolves the issue.