From 3a51b862659071a4dab8296f1c9cfbe38d287887 Mon Sep 17 00:00:00 2001 From: Made Date: Fri, 1 May 2020 21:57:41 +0200 Subject: finish decompiling FX_vec.s --- arm9/lib/src/FX_vec.c | 125 ++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 125 insertions(+) create mode 100644 arm9/lib/src/FX_vec.c (limited to 'arm9/lib/src/FX_vec.c') diff --git a/arm9/lib/src/FX_vec.c b/arm9/lib/src/FX_vec.c new file mode 100644 index 00000000..dc28bedf --- /dev/null +++ b/arm9/lib/src/FX_vec.c @@ -0,0 +1,125 @@ +#include "global.h" +#include "main.h" +#include "fx.h" + +void VEC_Add(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ + dst->x = a->x + b->x; + dst->y = a->y + b->y; + dst->z = a->z + b->z; +} + +void VEC_Subtract(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ + dst->x = a->x - b->x; + dst->y = a->y - b->y; + dst->z = a->z - b->z; +} + +void VEC_Fx16Add(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ + dst->x = a->x + b->x; + dst->y = a->y + b->y; + dst->z = a->z + b->z; +} + +s32 VEC_DotProduct(struct Vecx32 *a, struct Vecx32 *b){ + return ((s64)a->x * b->x + (s64)a->y * b->y + (s64)a->z * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT; +} + +s32 VEC_Fx16DotProduct(struct Vecx16 *a, struct Vecx16 *b){ + s32 temp1, temp2; + temp1 = (a->x * b->x) + (a->y * b->y); + temp2 = (a->z * b->z) + (1 << (FX64_INT_SHIFT - 1)); + return (s32)(((s64)temp1 + temp2) >> FX64_INT_SHIFT); +} + +void VEC_CrossProduct(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ + s32 x, y, z; + x = (s32)(((s64)a->y * b->z - (s64)a->z * b->y + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + y = (s32)(((s64)a->z * b->x - (s64)a->x * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + z = (s32)(((s64)a->x * b->y - (s64)a->y * b->x + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + dst->x = x; + dst->y = y; + dst->z = z; +} + +void VEC_Fx16CrossProduct(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ + s32 x, y, z; + x = ((a->y * b->z - a->z * b->y + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + y = ((a->z * b->x - a->x * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + z = ((a->x * b->y - a->y * b->x + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + dst->x = x; + dst->y = y; + dst->z = z; +} + +#define HW_REG_DIVCNT 0x04000280 +#define HW_REG_DIV_NUMER 0x04000290 +#define HW_REG_DIV_DENOM 0x04000298 +#define HW_REG_DIV_RESULT 0x040002A0 +#define HW_REG_DIVREM_RESULT 0x040002A8 + +#define HW_REG_SQRTCNT 0x040002B0 +#define HW_REG_SQRT_RESULT 0x040002B4 +#define HW_REG_SQRT_PARAM 0x040002B8 + +#define SETREG16(x, y) ((*(vu16 *)x) = y) +#define SETREG32(x, y) ((*(vu32 *)x) = y) +#define SETREG64(x, y) ((*(vu64 *)x) = y) +#define READREG16(x) (*(vu16 *)x) +#define READREG32(x) (*(vu32 *)x) +#define READREG64(x) (*(vu64 *)x) + +s32 VEC_Mag(struct Vecx32 *a){ + s64 l2 = (s64)a->x * a->x; + l2 += (s64)a->y * a->y; + l2 += (s64)a->z * a->z; + SETREG16(HW_REG_SQRTCNT, 0x1); + SETREG64(HW_REG_SQRT_PARAM, l2 * 4); + while (READREG16(HW_REG_SQRTCNT) & 0x8000); //wait for coprocessor to finish + return ((s32)READREG32(HW_REG_SQRT_RESULT) + 1) >> 1; +} + +void VEC_Normalize(struct Vecx32 *a, struct Vecx32 *dst){ + s64 l2 = (s64)a->x * a->x; + l2 += (s64)a->y * a->y; + l2 += (s64)a->z * a->z; + //1/sqrt(l) is computed by calculating sqrt(l)*(1/l) + SETREG16(HW_REG_DIVCNT, 0x2); + SETREG64(HW_REG_DIV_NUMER, 0x0100000000000000); + SETREG64(HW_REG_DIV_DENOM, l2); + SETREG16(HW_REG_SQRTCNT, 0x1); + SETREG64(HW_REG_SQRT_PARAM, l2 * 4); + while (READREG16(HW_REG_SQRTCNT) & 0x8000); //wait for sqrt to finish + s32 sqrtresult = READREG32(HW_REG_SQRT_RESULT); + while (READREG16(HW_REG_DIVCNT) & 0x8000); //wait for division to finish + l2 = READREG64(HW_REG_DIV_RESULT); + l2 = sqrtresult * l2; + dst->x = (l2 * a->x + (1LL << (0x2D - 1))) >> 0x2D; + dst->y = (l2 * a->y + (1LL << (0x2D - 1))) >> 0x2D; + dst->z = (l2 * a->z + (1LL << (0x2D - 1))) >> 0x2D; +} + +void VEC_Fx16Normalize(struct Vecx16 *a, struct Vecx16 *dst){ + s64 l2 = a->x * a->x; + l2 += a->y * a->y; + l2 += a->z * a->z; + //1/sqrt(l) is computed by calculating sqrt(l)*(1/l) + SETREG16(HW_REG_DIVCNT, 0x2); + SETREG64(HW_REG_DIV_NUMER, 0x0100000000000000); + SETREG64(HW_REG_DIV_DENOM, l2); + SETREG16(HW_REG_SQRTCNT, 0x1); + SETREG64(HW_REG_SQRT_PARAM, l2 * 4); + while (READREG16(HW_REG_SQRTCNT) & 0x8000); //wait for sqrt to finish + s32 sqrtresult = READREG32(HW_REG_SQRT_RESULT); + while (READREG16(HW_REG_DIVCNT) & 0x8000); //wait for division to finish + l2 = READREG64(HW_REG_DIV_RESULT); + l2 = sqrtresult * l2; + dst->x = (l2 * a->x + (1LL << (0x2D - 1))) >> 0x2D; + dst->y = (l2 * a->y + (1LL << (0x2D - 1))) >> 0x2D; + dst->z = (l2 * a->z + (1LL << (0x2D - 1))) >> 0x2D; +} + +void VEC_MultAdd(s32 factor, struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ + dst->x = (s32)(((s64)factor * a->x) >> FX32_INT_SHIFT) + b->x; + dst->y = (s32)(((s64)factor * a->y) >> FX32_INT_SHIFT) + b->y; + dst->z = (s32)(((s64)factor * a->z) >> FX32_INT_SHIFT) + b->z; +} -- cgit v1.2.3 From dba3fd7f6eb8d0895bec720db75a6b37fc7259a0 Mon Sep 17 00:00:00 2001 From: Made Date: Sat, 2 May 2020 17:39:44 +0200 Subject: Decompile FX_cp.s --- arm9/lib/src/FX_vec.c | 17 ----------------- 1 file changed, 17 deletions(-) (limited to 'arm9/lib/src/FX_vec.c') diff --git a/arm9/lib/src/FX_vec.c b/arm9/lib/src/FX_vec.c index dc28bedf..7b838829 100644 --- a/arm9/lib/src/FX_vec.c +++ b/arm9/lib/src/FX_vec.c @@ -51,23 +51,6 @@ void VEC_Fx16CrossProduct(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst dst->z = z; } -#define HW_REG_DIVCNT 0x04000280 -#define HW_REG_DIV_NUMER 0x04000290 -#define HW_REG_DIV_DENOM 0x04000298 -#define HW_REG_DIV_RESULT 0x040002A0 -#define HW_REG_DIVREM_RESULT 0x040002A8 - -#define HW_REG_SQRTCNT 0x040002B0 -#define HW_REG_SQRT_RESULT 0x040002B4 -#define HW_REG_SQRT_PARAM 0x040002B8 - -#define SETREG16(x, y) ((*(vu16 *)x) = y) -#define SETREG32(x, y) ((*(vu32 *)x) = y) -#define SETREG64(x, y) ((*(vu64 *)x) = y) -#define READREG16(x) (*(vu16 *)x) -#define READREG32(x) (*(vu32 *)x) -#define READREG64(x) (*(vu64 *)x) - s32 VEC_Mag(struct Vecx32 *a){ s64 l2 = (s64)a->x * a->x; l2 += (s64)a->y * a->y; -- cgit v1.2.3 From ab28b0b37f2fd9fba6010e48e15bd31170d678d7 Mon Sep 17 00:00:00 2001 From: Made Date: Mon, 4 May 2020 17:07:22 +0200 Subject: Add typedefs for fixed point types --- arm9/lib/src/FX_vec.c | 50 +++++++++++++++++++++++++------------------------- 1 file changed, 25 insertions(+), 25 deletions(-) (limited to 'arm9/lib/src/FX_vec.c') diff --git a/arm9/lib/src/FX_vec.c b/arm9/lib/src/FX_vec.c index 7b838829..10a792be 100644 --- a/arm9/lib/src/FX_vec.c +++ b/arm9/lib/src/FX_vec.c @@ -20,29 +20,29 @@ void VEC_Fx16Add(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ dst->z = a->z + b->z; } -s32 VEC_DotProduct(struct Vecx32 *a, struct Vecx32 *b){ - return ((s64)a->x * b->x + (s64)a->y * b->y + (s64)a->z * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT; +fx32 VEC_DotProduct(struct Vecx32 *a, struct Vecx32 *b){ + return ((fx64)a->x * b->x + (fx64)a->y * b->y + (fx64)a->z * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT; } -s32 VEC_Fx16DotProduct(struct Vecx16 *a, struct Vecx16 *b){ - s32 temp1, temp2; +fx32 VEC_Fx16DotProduct(struct Vecx16 *a, struct Vecx16 *b){ + fx32 temp1, temp2; temp1 = (a->x * b->x) + (a->y * b->y); temp2 = (a->z * b->z) + (1 << (FX64_INT_SHIFT - 1)); - return (s32)(((s64)temp1 + temp2) >> FX64_INT_SHIFT); + return (fx32)(((fx64)temp1 + temp2) >> FX64_INT_SHIFT); } void VEC_CrossProduct(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ - s32 x, y, z; - x = (s32)(((s64)a->y * b->z - (s64)a->z * b->y + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); - y = (s32)(((s64)a->z * b->x - (s64)a->x * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); - z = (s32)(((s64)a->x * b->y - (s64)a->y * b->x + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + fx32 x, y, z; + x = (fx32)(((fx64)a->y * b->z - (fx64)a->z * b->y + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + y = (fx32)(((fx64)a->z * b->x - (fx64)a->x * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); + z = (fx32)(((fx64)a->x * b->y - (fx64)a->y * b->x + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); dst->x = x; dst->y = y; dst->z = z; } void VEC_Fx16CrossProduct(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ - s32 x, y, z; + fx32 x, y, z; x = ((a->y * b->z - a->z * b->y + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); y = ((a->z * b->x - a->x * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); z = ((a->x * b->y - a->y * b->x + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); @@ -51,20 +51,20 @@ void VEC_Fx16CrossProduct(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst dst->z = z; } -s32 VEC_Mag(struct Vecx32 *a){ - s64 l2 = (s64)a->x * a->x; - l2 += (s64)a->y * a->y; - l2 += (s64)a->z * a->z; +fx32 VEC_Mag(struct Vecx32 *a){ + fx64 l2 = (fx64)a->x * a->x; + l2 += (fx64)a->y * a->y; + l2 += (fx64)a->z * a->z; SETREG16(HW_REG_SQRTCNT, 0x1); SETREG64(HW_REG_SQRT_PARAM, l2 * 4); while (READREG16(HW_REG_SQRTCNT) & 0x8000); //wait for coprocessor to finish - return ((s32)READREG32(HW_REG_SQRT_RESULT) + 1) >> 1; + return ((fx32)READREG32(HW_REG_SQRT_RESULT) + 1) >> 1; } void VEC_Normalize(struct Vecx32 *a, struct Vecx32 *dst){ - s64 l2 = (s64)a->x * a->x; - l2 += (s64)a->y * a->y; - l2 += (s64)a->z * a->z; + fx64 l2 = (fx64)a->x * a->x; + l2 += (fx64)a->y * a->y; + l2 += (fx64)a->z * a->z; //1/sqrt(l) is computed by calculating sqrt(l)*(1/l) SETREG16(HW_REG_DIVCNT, 0x2); SETREG64(HW_REG_DIV_NUMER, 0x0100000000000000); @@ -72,7 +72,7 @@ void VEC_Normalize(struct Vecx32 *a, struct Vecx32 *dst){ SETREG16(HW_REG_SQRTCNT, 0x1); SETREG64(HW_REG_SQRT_PARAM, l2 * 4); while (READREG16(HW_REG_SQRTCNT) & 0x8000); //wait for sqrt to finish - s32 sqrtresult = READREG32(HW_REG_SQRT_RESULT); + fx32 sqrtresult = READREG32(HW_REG_SQRT_RESULT); while (READREG16(HW_REG_DIVCNT) & 0x8000); //wait for division to finish l2 = READREG64(HW_REG_DIV_RESULT); l2 = sqrtresult * l2; @@ -82,7 +82,7 @@ void VEC_Normalize(struct Vecx32 *a, struct Vecx32 *dst){ } void VEC_Fx16Normalize(struct Vecx16 *a, struct Vecx16 *dst){ - s64 l2 = a->x * a->x; + fx64 l2 = a->x * a->x; l2 += a->y * a->y; l2 += a->z * a->z; //1/sqrt(l) is computed by calculating sqrt(l)*(1/l) @@ -92,7 +92,7 @@ void VEC_Fx16Normalize(struct Vecx16 *a, struct Vecx16 *dst){ SETREG16(HW_REG_SQRTCNT, 0x1); SETREG64(HW_REG_SQRT_PARAM, l2 * 4); while (READREG16(HW_REG_SQRTCNT) & 0x8000); //wait for sqrt to finish - s32 sqrtresult = READREG32(HW_REG_SQRT_RESULT); + fx32 sqrtresult = READREG32(HW_REG_SQRT_RESULT); while (READREG16(HW_REG_DIVCNT) & 0x8000); //wait for division to finish l2 = READREG64(HW_REG_DIV_RESULT); l2 = sqrtresult * l2; @@ -101,8 +101,8 @@ void VEC_Fx16Normalize(struct Vecx16 *a, struct Vecx16 *dst){ dst->z = (l2 * a->z + (1LL << (0x2D - 1))) >> 0x2D; } -void VEC_MultAdd(s32 factor, struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ - dst->x = (s32)(((s64)factor * a->x) >> FX32_INT_SHIFT) + b->x; - dst->y = (s32)(((s64)factor * a->y) >> FX32_INT_SHIFT) + b->y; - dst->z = (s32)(((s64)factor * a->z) >> FX32_INT_SHIFT) + b->z; +void VEC_MultAdd(fx32 factor, struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ + dst->x = (fx32)(((fx64)factor * a->x) >> FX32_INT_SHIFT) + b->x; + dst->y = (fx32)(((fx64)factor * a->y) >> FX32_INT_SHIFT) + b->y; + dst->z = (fx32)(((fx64)factor * a->z) >> FX32_INT_SHIFT) + b->z; } -- cgit v1.2.3 From f4c61411673207da9d6be013674f95eeb07408b5 Mon Sep 17 00:00:00 2001 From: Made Date: Mon, 4 May 2020 17:48:38 +0200 Subject: Add function attributes and remove thumb pragmas --- arm9/lib/src/FX_vec.c | 22 +++++++++++----------- 1 file changed, 11 insertions(+), 11 deletions(-) (limited to 'arm9/lib/src/FX_vec.c') diff --git a/arm9/lib/src/FX_vec.c b/arm9/lib/src/FX_vec.c index 10a792be..af36fe89 100644 --- a/arm9/lib/src/FX_vec.c +++ b/arm9/lib/src/FX_vec.c @@ -2,36 +2,36 @@ #include "main.h" #include "fx.h" -void VEC_Add(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ +ARM_FUNC void VEC_Add(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ dst->x = a->x + b->x; dst->y = a->y + b->y; dst->z = a->z + b->z; } -void VEC_Subtract(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ +ARM_FUNC void VEC_Subtract(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ dst->x = a->x - b->x; dst->y = a->y - b->y; dst->z = a->z - b->z; } -void VEC_Fx16Add(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ +ARM_FUNC void VEC_Fx16Add(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ dst->x = a->x + b->x; dst->y = a->y + b->y; dst->z = a->z + b->z; } -fx32 VEC_DotProduct(struct Vecx32 *a, struct Vecx32 *b){ +ARM_FUNC fx32 VEC_DotProduct(struct Vecx32 *a, struct Vecx32 *b){ return ((fx64)a->x * b->x + (fx64)a->y * b->y + (fx64)a->z * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT; } -fx32 VEC_Fx16DotProduct(struct Vecx16 *a, struct Vecx16 *b){ +ARM_FUNC fx32 VEC_Fx16DotProduct(struct Vecx16 *a, struct Vecx16 *b){ fx32 temp1, temp2; temp1 = (a->x * b->x) + (a->y * b->y); temp2 = (a->z * b->z) + (1 << (FX64_INT_SHIFT - 1)); return (fx32)(((fx64)temp1 + temp2) >> FX64_INT_SHIFT); } -void VEC_CrossProduct(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ +ARM_FUNC void VEC_CrossProduct(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ fx32 x, y, z; x = (fx32)(((fx64)a->y * b->z - (fx64)a->z * b->y + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); y = (fx32)(((fx64)a->z * b->x - (fx64)a->x * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); @@ -41,7 +41,7 @@ void VEC_CrossProduct(struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ dst->z = z; } -void VEC_Fx16CrossProduct(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ +ARM_FUNC void VEC_Fx16CrossProduct(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst){ fx32 x, y, z; x = ((a->y * b->z - a->z * b->y + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); y = ((a->z * b->x - a->x * b->z + (1 << (FX64_INT_SHIFT - 1))) >> FX64_INT_SHIFT); @@ -51,7 +51,7 @@ void VEC_Fx16CrossProduct(struct Vecx16 *a, struct Vecx16 *b, struct Vecx16 *dst dst->z = z; } -fx32 VEC_Mag(struct Vecx32 *a){ +ARM_FUNC fx32 VEC_Mag(struct Vecx32 *a){ fx64 l2 = (fx64)a->x * a->x; l2 += (fx64)a->y * a->y; l2 += (fx64)a->z * a->z; @@ -61,7 +61,7 @@ fx32 VEC_Mag(struct Vecx32 *a){ return ((fx32)READREG32(HW_REG_SQRT_RESULT) + 1) >> 1; } -void VEC_Normalize(struct Vecx32 *a, struct Vecx32 *dst){ +ARM_FUNC void VEC_Normalize(struct Vecx32 *a, struct Vecx32 *dst){ fx64 l2 = (fx64)a->x * a->x; l2 += (fx64)a->y * a->y; l2 += (fx64)a->z * a->z; @@ -81,7 +81,7 @@ void VEC_Normalize(struct Vecx32 *a, struct Vecx32 *dst){ dst->z = (l2 * a->z + (1LL << (0x2D - 1))) >> 0x2D; } -void VEC_Fx16Normalize(struct Vecx16 *a, struct Vecx16 *dst){ +ARM_FUNC void VEC_Fx16Normalize(struct Vecx16 *a, struct Vecx16 *dst){ fx64 l2 = a->x * a->x; l2 += a->y * a->y; l2 += a->z * a->z; @@ -101,7 +101,7 @@ void VEC_Fx16Normalize(struct Vecx16 *a, struct Vecx16 *dst){ dst->z = (l2 * a->z + (1LL << (0x2D - 1))) >> 0x2D; } -void VEC_MultAdd(fx32 factor, struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ +ARM_FUNC void VEC_MultAdd(fx32 factor, struct Vecx32 *a, struct Vecx32 *b, struct Vecx32 *dst){ dst->x = (fx32)(((fx64)factor * a->x) >> FX32_INT_SHIFT) + b->x; dst->y = (fx32)(((fx64)factor * a->y) >> FX32_INT_SHIFT) + b->y; dst->z = (fx32)(((fx64)factor * a->z) >> FX32_INT_SHIFT) + b->z; -- cgit v1.2.3