PMP V
jz_mxu.h
程式碼如下：
/* Jz47xx Ingenic Media Extension Instruction Set
These are ~60 SIMD instructions for the Jz47xx MIPS core.
To compile assembly files using these instructions, they
must be piped through a bash script called mxu_as.
*/
#ifndef JZ_MXU_H_
#define JZ_MXU_H_
#define ptn0 0
#define ptn1 1
#define ptn2 2
#define ptn3 3
#ifdef C_VERSION
/* MXU registers */
#ifndef MXU_REGS_USE_ARRAY
#define xr0 0
static int xr1, xr2, xr3, xr4, xr5, xr6, xr7, xr8, xr9;
static int xr10, xr11, xr12, xr13, xr14, xr15, xr16;
#else
static int mxu_xr[17] = {0};
#define xr0 mxu_xr[ 0]
#define xr1 mxu_xr[ 1]
#define xr2 mxu_xr[ 2]
#define xr3 mxu_xr[ 3]
#define xr4 mxu_xr[ 4]
#define xr5 mxu_xr[ 5]
#define xr6 mxu_xr[ 6]
#define xr7 mxu_xr[ 7]
#define xr8 mxu_xr[ 8]
#define xr9 mxu_xr[ 9]
#define xr10 mxu_xr[10]
#define xr11 mxu_xr[11]
#define xr12 mxu_xr[12]
#define xr13 mxu_xr[13]
#define xr14 mxu_xr[14]
#define xr15 mxu_xr[15]
#define xr16 mxu_xr[16]
#endif
#else /* C_VERSION */
#define xr0 0
#define xr1 1
#define xr2 2
#define xr3 3
#define xr4 4
#define xr5 5
#define xr6 6
#define xr7 7
#define xr8 8
#define xr9 9
#define xr10 10
#define xr11 11
#define xr12 12
#define xr13 13
#define xr14 14
#define xr15 15
#define xr16 16
#endif /* C_VERSION */
#ifdef C_VERSION
#define S32I2M(xr, r) if (&xr != mxu_xr) xr = r
#define S32M2I(xr) xr
#define S32LDD(xr, p, o) if (&xr != mxu_xr) xr = *(long*)((unsigned long)p + o)
#define S32STD(xr, p, o) *(long*)((unsigned long)p + o) = xr
#define S32LDDV(xr, p, o, s) if (&xr != mxu_xr) xr = *(long*)((unsigned long)p + ((o) << s))
#define S32STDV(xr, p, o, s) *(long*)((unsigned long)p + ((o) << s)) = xr
#define S32LDIV(xra, rb, rc, strd2) \
{\
if (&xra != mxu_xr) xra = *(long*)((unsigned long)rb + ((rc) << strd2));\
rb = (char*)rb + ((rc) << strd2);\
}
#define S32SDIV(xra, rb, rc, strd2) \
{\
*(long*)((unsigned long)rb + ((rc) << strd2)) = xra;\
rb = (char*)rb + ((rc) << strd2);\
}
#define S32LDI(xra, rb, o) \
{\
if (&xra != mxu_xr) xra = *(long*)((unsigned long)rb + o);\
rb = (char*)rb + o;\
}
#define S32SDI(xra, rb, o) \
{\
*(long*)((unsigned long)rb + o) = xra;\
rb = (char*)rb + o;\
}
#define S32LDIV(xra, rb, rc, strd2) \
{\
if (&xra != mxu_xr) xra = *(long*)((unsigned long)rb + ((rc) << strd2));\
rb = (char*)rb + ((rc) << strd2);\
}
#define S32SDIV(xra, rb, rc, strd2) \
{\
*(long*)((unsigned long)rb + ((rc) << strd2)) = xra;\
rb = (char*)rb + ((rc) << strd2);\
}
#define Q16ADD_AS_WW(a, b, c, d) \
{\
short bh = b >> 16;\
short bl = b & 0xFFFF;\
short ch = c >> 16;\
short cl = c & 0xFFFF;\
int ah = bh + ch;\
int al = bl + cl;\
int dh = bh - ch;\
int dl = bl - cl;\
if (&a != mxu_xr) a = (ah << 16) | (al & 0xFFFF);\
if (&d != mxu_xr) d = (dh << 16) | (dl & 0xFFFF);\
}
#define Q16ADD_AS_XW(a, b, c, d) \
{\
short bh = b >> 16;\
short bl = b & 0xFFFF;\
short ch = c >> 16;\
short cl = c & 0xFFFF;\
int ah = bl + ch;\
int al = bh + cl;\
int dh = bl - ch;\
int dl = bh - cl;\
if (&a != mxu_xr) a = (ah << 16) | (al & 0xFFFF);\
if (&d != mxu_xr) d = (dh << 16) | (dl & 0xFFFF);\
}
#define Q16ADD_AA_WW(a, b, c, d) \
{\
short bh = b >> 16;\
short bl = b & 0xFFFF;\
short ch = c >> 16;\
short cl = c & 0xFFFF;\
int ah = bh + ch;\
int al = bl + cl;\
if (&a != mxu_xr) a = (ah << 16) | (al & 0xFFFF);\
if (&d != mxu_xr) d = (ah << 16) | (al & 0xFFFF);\
}
#define D16MUL_LW(a, b, c, d)\
{\
short bl = b & 0xFFFF;\
short cl = c & 0xFFFF;\
short ch = c >> 16;\
if (&a != mxu_xr) a = ch * bl;\
if (&d != mxu_xr) d = cl * bl;\
}
#define D16MUL_WW(a, b, c, d)\
{\
short bh = b >> 16;\
short bl = b & 0xFFFF;\
short ch = c >> 16;\
short cl = c & 0xFFFF;\
if (&a != mxu_xr) a = ch * bh;\
if (&d != mxu_xr) d = cl * bl;\
}
#define D16MAC_AA_LW(a, b, c, d)\
{\
short bl = b & 0xFFFF;\
short cl = c & 0xFFFF;\
short ch = c >> 16;\
if (&a != mxu_xr) a += ch * bl;\
if (&d != mxu_xr) d += cl * bl;\
}
#define D16MUL_HW(a, b, c, d)\
{\
short bh = b >> 16;\
short cl = c & 0xFFFF;\
short ch = c >> 16;\
if (&a != mxu_xr) a = ch * bh;\
if (&d != mxu_xr) d = cl * bh;\
}
#define D16MAC_AA_HW(a, b, c, d)\
{\
short bh = b >> 16;\
short cl = c & 0xFFFF;\
short ch = c >> 16;\
if (&a != mxu_xr) a += ch * bh;\
if (&d != mxu_xr) d += cl * bh;\
}
#define D32SLL(a, b, c, d, sft)\
{\
if (&a != mxu_xr) a = b << sft;\
if (&d != mxu_xr) d = c << sft;\
}
#define D32SARL(a, b, c, sft) if (&a != mxu_xr) a = (((long)b >> sft) << 16) | (((long)c >> sft) & 0xFFFF)
#define S32SFL(a, b, c, d, ptn) \
{\
unsigned char b3 = (unsigned char)((unsigned long)b >> 24);\
unsigned char b2 = (unsigned char)((unsigned long)b >> 16);\
unsigned char b1 = (unsigned char)((unsigned long)b >> 8);\
unsigned char b0 = (unsigned char)((unsigned long)b >> 0);\
unsigned char c3 = (unsigned char)((unsigned long)c >> 24);\
unsigned char c2 = (unsigned char)((unsigned long)c >> 16);\
unsigned char c1 = (unsigned char)((unsigned long)c >> 8);\
unsigned char c0 = (unsigned char)((unsigned long)c >> 0);\
unsigned char a3, a2, a1, a0, d3, d2, d1, d0;\
if (ptn0 == ptn) \
{\
a3 = b3;\
a2 = c3;\
a1 = b2;\
a0 = c2;\
d3 = b1;\
d2 = c1;\
d1 = b0;\
d0 = c0;\
}\
else if (ptn1 == ptn)\
{\
a3 = b3;\
a2 = b1;\
a1 = c3;\
a0 = c1;\
d3 = b2;\
d2 = b0;\
d1 = c2;\
d0 = c0;\
}\
else if (ptn2 == ptn)\
{\
a3 = b3;\
a2 = c3;\
a1 = b1;\
a0 = c1;\
d3 = b2;\
d2 = c2;\
d1 = b0;\
d0 = c0;\
}\
else if (ptn3 == ptn)\
{\
a3 = b3;\
a2 = b2;\
a1 = c3;\
a0 = c2;\
d3 = b1;\
d2 = b0;\
d1 = c1;\
d0 = c0;\
}\
if (&a != mxu_xr) a = ((unsigned long)a3 << 24) | ((unsigned long)a2 << 16) | ((unsigned long)a1 << 8) | (unsigned long)a0;\
if (&d != mxu_xr) d = ((unsigned long)d3 << 24) | ((unsigned long)d2 << 16) | ((unsigned long)d1 << 8) | (unsigned long)d0;\
}
#define D32SAR(a, b, c, d, sft)\
{\
if (&a != mxu_xr) a = (long)b >> sft;\
if (&d != mxu_xr) d = (long)c >> sft;\
}
#define D32SLR(a, b, c, d, sft)\
{\
if (&a != mxu_xr) a = (unsigned long)b >> sft;\
if (&d != mxu_xr) d = (unsigned long)c >> sft;\
}
#define Q16SLL(a,b,c,d,sft)\
{\
short bh=b>>16;\
short bl=b&0xffff;\
short ch=c>>16;\
short cl=c&0xffff;\
if(&a!=mxu_xr) a=((bh<<sft)<<16)|(((long)bl<<sft) & 0xffff);\
if(&d!=mxu_xr) d=((dh<<sft)<<16)|(((long)bl<<sft) & 0xffff);\
}
#define Q16SAR(a,b,c,d,sft)\
{\
short bh = b >> 16;\
short bl = b & 0xffff;\
short ch = c >> 16;\
short cl = c & 0xffff;\
if(&a!=mxu_xr) a=(((short)bh>>sft)<<16)|((long)((short)b1>>sft) & 0xffff);\
if(&d!=mxu_xr) d=(((short)ch>>sft)<<16)|((long)((short)cl>>sft) & 0xffff);\
}
#define D32ACC_AA(a, b, c, d)\
{\
int _b = b;\
int _c = c;\
int _a = a;\
int _d = d;\
if (&a != mxu_xr) a = _a + _b + _c;\
if (&d != mxu_xr) d = _d + _b + _c;\
}
#define D32ACC_AS(a, b, c, d)\
{\
int _b = b;\
int _c = c;\
int _a = a;\
int _d = d;\
if (&a != mxu_xr) a = _a + _b + _c;\
if (&d != mxu_xr) d = _d + _b - _c;\
}
#define D32ADD_AS(a, b, c, d)\
{\
int _b = b;\
int _c = c;\
if (&a != mxu_xr) a = _b + _c;\
if (&d != mxu_xr) d = _b - _c;\
}
#define D32ADD_SS(a, b, c, d)\
{\
int _b = b;\
int _c = c;\
if (&a != mxu_xr) a = _b - _c;\
if (&d != mxu_xr) d = _b - _c;\
}
#define D32ADD_AA(a, b, c, d)\
{\
int _b = b;\
int _c = c;\
if (&a != mxu_xr) a = _b + _c;\
if (&d != mxu_xr) d = _b + _c;\
}
#define D16MADL_AA_WW(a, b, c, d) \
do { \
short _ah = a >> 16;\
short _al = (a << 16) >> 16;\
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = _bh * _ch;\
R32 = _bl * _cl; \
_ah += (L32 << 16) >> 16; \
_al += (R32 << 16) >> 16; \
if (&d != mxu_xr) d = (_ah << 16) + (_al & 0xffff);\
} while (0)
#define D16MACF_AA_WW(a, b, c, d) \
do { \
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = (_bh * _ch) << 1;\
R32 = (_bl * _cl) << 1; \
L32 = a + L32; \
R32 = d + R32; \
if (&a != mxu_xr) a = ((((L32 >> 15) + 1) >> 1) << 16) + ((((R32 >> 15) + 1) >> 1) & 0xffff);\
} while (0)
#define D16MAC_AA_WW(a, b, c, d) \
do { \
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = (_bh * _ch);\
R32 = (_bl * _cl); \
if (&a != mxu_xr) a = a + L32;\
if (&d != mxu_xr) d = d + R32;\
} while (0)
#define D16MAC_SS_WW(a, b, c, d) \
do { \
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = (_bh * _ch);\
R32 = (_bl * _cl); \
if (&a != mxu_xr) a = a - L32;\
if (&d != mxu_xr) d = d - R32;\
} while (0)
#define D16MAC_SA_HW(a, b, c, d) \
do { \
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = (_bh * _ch);\
R32 = (_bh * _cl); \
if (&a != mxu_xr) a = a - L32;\
if (&d != mxu_xr) d = d + R32;\
} while (0)
#define D16MAC_SS_HW(a, b, c, d) \
do { \
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = (_bh * _ch);\
R32 = (_bh * _cl); \
if (&a != mxu_xr) a = a - L32;\
if (&d != mxu_xr) d = d - R32;\
} while (0)
#define D16MAC_AS_HW(a, b, c, d) \
do { \
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = (_bh * _ch);\
R32 = (_bh * _cl); \
if (&a != mxu_xr) a = a + L32;\
if (&d != mxu_xr) d = d - R32;\
} while (0)
#define D16MAC_AS_LW(a, b, c, d) \
do { \
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = (_bl * _ch);\
R32 = (_bl * _cl); \
if (&a != mxu_xr) a = a + L32;\
if (&d != mxu_xr) d = d - R32;\
} while (0)
#define D16MAC_SA_LW(a, b, c, d) \
do { \
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = (_bl * _ch);\
R32 = (_bl * _cl); \
if (&a != mxu_xr) a = a - L32;\
if (&d != mxu_xr) d = d + R32;\
} while (0)
#define D16MAC_SS_LW(a, b, c, d) \
do { \
short _bh = b >> 16;\
short _bl = (b << 16) >> 16;\
short _ch = c >> 16;\
short _cl = (c << 16) >> 16;\
int L32, R32; \
L32 = (_bl * _ch);\
R32 = (_bl * _cl); \
if (&a != mxu_xr) a = a - L32;\
if (&d != mxu_xr) d = d - R32;\
} while (0)
#define Q8ADDE_AA(xra, xrb, xrc, xrd) \
{\
unsigned char b3 = (unsigned char)((unsigned long)xrb >> 24);\
unsigned char b2 = (unsigned char)((unsigned long)xrb >> 16);\
unsigned char b1 = (unsigned char)((unsigned long)xrb >> 8);\
unsigned char b0 = (unsigned char)((unsigned long)xrb >> 0);\
unsigned char c3 = (unsigned char)((unsigned long)xrc >> 24);\
unsigned char c2 = (unsigned char)((unsigned long)xrc >> 16);\
unsigned char c1 = (unsigned char)((unsigned long)xrc >> 8);\
unsigned char c0 = (unsigned char)((unsigned long)xrc >> 0);\
short ah, al, dh, dl;\
ah = b3 + c3;\
al = b2 + c2;\
dh = b1 + c1;\
dl = b0 + c0;\
if (&xra != mxu_xr) xra = ((unsigned long)ah << 16) | (unsigned short)al;\
if (&xrd != mxu_xr) xrd = ((unsigned long)dh << 16) | (unsigned short)dl;\
}
#define Q16SAT(xra, xrb, xrc) \
{\
short bh = xrb >> 16;\
short bl = xrb & 0xFFFF;\
short ch = xrc >> 16;\
short cl = xrc & 0xFFFF;\
if (bh > 255) bh = 255;\
if (bh < 0) bh = 0;\
if (bl > 255) bl = 255;\
if (bl < 0) bl = 0;\
if (ch > 255) ch = 255;\
if (ch < 0) ch = 0;\
if (cl > 255) cl = 255;\
if (cl < 0) cl = 0;\
if (&xra != mxu_xr) xra = ((unsigned)bh << 24) | ((unsigned)bl << 16) | ((unsigned)ch << 8) | (unsigned)cl;\
}
#define Q8SAD(xra, xrb, xrc, xrd) \
{\
short b3 = (unsigned char)((unsigned long)xrb >> 24);\
short b2 = (unsigned char)((unsigned long)xrb >> 16);\
short b1 = (unsigned char)((unsigned long)xrb >> 8);\
short b0 = (unsigned char)((unsigned long)xrb >> 0);\
short c3 = (unsigned char)((unsigned long)xrc >> 24);\
short c2 = (unsigned char)((unsigned long)xrc >> 16);\
short c1 = (unsigned char)((unsigned long)xrc >> 8);\
short c0 = (unsigned char)((unsigned long)xrc >> 0);\
int int0, int1, int2, int3;\
int3 = labs(b3 - c3);\
int2 = labs(b2 - c2);\
int1 = labs(b1 - c1);\
int0 = labs(b0 - c0);\
if (&xra != mxu_xr) xra = int0 + int1 + int2 + int3;\
if (&xrd != mxu_xr) xrd += int0 + int1 + int2 + int3;\
}
#define Q8AVGR(xra, xrb, xrc) \
{\
short b3 = (unsigned char)((unsigned long)xrb >> 24);\
short b2 = (unsigned char)((unsigned long)xrb >> 16);\
short b1 = (unsigned char)((unsigned long)xrb >> 8);\
short b0 = (unsigned char)((unsigned long)xrb >> 0);\
short c3 = (unsigned char)((unsigned long)xrc >> 24);\
short c2 = (unsigned char)((unsigned long)xrc >> 16);\
short c1 = (unsigned char)((unsigned long)xrc >> 8);\
short c0 = (unsigned char)((unsigned long)xrc >> 0);\
unsigned char a3, a2, a1, a0;\
a3 = (unsigned char)((b3 + c3 + 1) >> 1);\
a2 = (unsigned char)((b2 + c2 + 1) >> 1);\
a1 = (unsigned char)((b1 + c1 + 1) >> 1);\
a0 = (unsigned char)((b0 + c0 + 1) >> 1);\
if (&xra != mxu_xr) xra = ((unsigned long)a3 << 24) | ((unsigned long)a2 << 16) | ((unsigned long)a1 << 8) | (unsigned long)a0;\
}
#define S32ALN(xra, xrb, xrc, rs) \
{\
if (0 == rs)\
{\
if (&xra != mxu_xr) xra = xrb;\
}\
else if (1 == rs)\
{\
if (&xra != mxu_xr) xra = (xrb << 8) | ((unsigned long)xrc >> 24);\
}\
else if (2 == rs)\
{\
if (&xra != mxu_xr) xra = (xrb << 16) | ((unsigned long)xrc >> 16);\
}\
else if (3 == rs)\
{\
if (&xra != mxu_xr) xra = (xrb << 24) | ((unsigned long)xrc >> 8);\
}\
else if (4 == rs)\
{\
if (&xra != mxu_xr) xra = xrc;\
}\
}
#else /* C_VERSION */
/***********************************LD/SD***********************************/
#define S32LDD(xra,rb,s12) \
do { \
__asm__ __volatile ("S32LDD xr%0,%z1,%2" \
: \
:"K"(xra),"d" (rb),"I"(s12)); \
} while (0)
#define S32STD(xra,rb,s12) \
do { \
__asm__ __volatile ("S32STD xr%0,%z1,%2" \
: \
:"K"(xra),"d" (rb),"I"(s12):"memory"); \
} while (0)
#define S32LDDV(xra,rb,rc,strd2) \
do { \
__asm__ __volatile ("S32LDDV xr%0,%z1,%z2,%3" \
: \
:"K"(xra),"d" (rb),"d"(rc),"K"(strd2)); \
} while (0)
#define S32STDV(xra,rb,rc,strd2) \
do { \
__asm__ __volatile ("S32STDV xr%0,%z1,%z2,%3" \
: \
:"K"(xra),"d" (rb),"d"(rc),"K"(strd2):"memory"); \
} while (0)
#define S32LDI(xra,rb,s12) \
do { \
__asm__ __volatile ("S32LDI xr%1,%z0,%2" \
:"+d" (rb) \
:"K"(xra),"I"(s12)); \
} while (0)
#define S32SDI(xra,rb,s12) \
do { \
__asm__ __volatile ("S32SDI xr%1,%z0,%2" \
:"+d" (rb) \
:"K"(xra),"I"(s12):"memory"); \
} while (0)
#define S32LDIV(xra,rb,rc,strd2) \
do { \
__asm__ __volatile ("S32LDIV xr%1,%z0,%z2,%3" \
:"+d" (rb) \
:"K"(xra),"d"(rc),"K"(strd2)); \
} while (0)
#define S32SDIV(xra,rb,rc,strd2) \
do { \
__asm__ __volatile ("S32SDIV xr%1,%z0,%z2,%3" \
:"+d" (rb) \
:"K"(xra),"d"(rc),"K"(strd2):"memory"); \
} while (0)
/***********************************D16MUL***********************************/
#define D16MUL_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MUL xr%0,xr%1,xr%2,xr%3,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MUL_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MUL xr%0,xr%1,xr%2,xr%3,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MUL_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MUL xr%0,xr%1,xr%2,xr%3,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MUL_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MUL xr%0,xr%1,xr%2,xr%3,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/**********************************D16MULF*******************************/
#define D16MULF_WW(xra,xrb,xrc) \
do { \
__asm__ __volatile ("D16MULF xr%0,xr%1,xr%2,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define D16MULF_LW(xra,xrb,xrc) \
do { \
__asm__ __volatile ("D16MULF xr%0,xr%1,xr%2,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define D16MULF_HW(xra,xrb,xrc) \
do { \
__asm__ __volatile ("D16MULF xr%0,xr%1,xr%2,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define D16MULF_XW(xra,xrb,xrc) \
do { \
__asm__ __volatile ("D16MULF xr%0,xr%1,xr%2,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
/***********************************D16MAC********************************/
#define D16MAC_AA_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,AA,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_AA_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,AA,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_AA_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,AA,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_AA_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,AA,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_AS_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,AS,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_AS_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,AS,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_AS_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,AS,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_AS_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,AS,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_SA_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,SA,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_SA_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,SA,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_SA_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,SA,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_SA_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,SA,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_SS_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,SS,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_SS_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,SS,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_SS_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,SS,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MAC_SS_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MAC xr%0,xr%1,xr%2,xr%3,SS,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/**********************************D16MACF*******************************/
#define D16MACF_AA_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,AA,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_AA_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,AA,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_AA_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,AA,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_AA_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,AA,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_AS_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,AS,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_AS_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,AS,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_AS_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,AS,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_AS_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,AS,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_SA_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,SA,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_SA_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,SA,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_SA_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,SA,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_SA_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,SA,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_SS_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,SS,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_SS_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,SS,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_SS_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,SS,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MACF_SS_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MACF xr%0,xr%1,xr%2,xr%3,SS,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/**********************************D16MADL*******************************/
#define D16MADL_AA_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,AA,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_AA_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,AA,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_AA_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,AA,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_AA_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,AA,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_AS_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,AS,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_AS_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,AS,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_AS_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,AS,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_AS_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,AS,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_SA_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,SA,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_SA_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,SA,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_SA_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,SA,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_SA_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,SA,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_SS_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,SS,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_SS_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,SS,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_SS_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,SS,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D16MADL_SS_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D16MADL xr%0,xr%1,xr%2,xr%3,SS,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/***********************************S16MAD*******************************/
#define S16MAD_A_HH(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("S16MAD xr%0,xr%1,xr%2,xr%3,A,0" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define S16MAD_A_LL(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("S16MAD xr%0,xr%1,xr%2,xr%3,A,1" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define S16MAD_A_HL(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("S16MAD xr%0,xr%1,xr%2,xr%3,A,2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define S16MAD_A_LH(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("S16MAD xr%0,xr%1,xr%2,xr%3,A,3" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define S16MAD_S_HH(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("S16MAD xr%0,xr%1,xr%2,xr%3,S,0" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define S16MAD_S_LL(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("S16MAD xr%0,xr%1,xr%2,xr%3,S,1" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define S16MAD_S_HL(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("S16MAD xr%0,xr%1,xr%2,xr%3,S,2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define S16MAD_S_LH(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("S16MAD xr%0,xr%1,xr%2,xr%3,S,3" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/***********************************Q8MUL********************************/
#define Q8MUL(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8MUL xr%0,xr%1,xr%2,xr%3" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/***********************************Q8MAC********************************/
#define Q8MAC_AA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8MAC xr%0,xr%1,xr%2,xr%3,AA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8MAC_AS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8MAC xr%0,xr%1,xr%2,xr%3,AS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8MAC_SA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8MAC xr%0,xr%1,xr%2,xr%3,SA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8MAC_SS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8MAC xr%0,xr%1,xr%2,xr%3,SS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/***********************************Q8MADL********************************/
#define Q8MADL_AA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8MADL xr%0,xr%1,xr%2,xr%3,AA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8MADL_AS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8MADL xr%0,xr%1,xr%2,xr%3,AS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8MADL_SA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8MADL xr%0,xr%1,xr%2,xr%3,SA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8MADL_SS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8MADL xr%0,xr%1,xr%2,xr%3,SS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/***********************************D32ADD********************************/
#define D32ADD_AA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D32ADD xr%0,xr%1,xr%2,xr%3,AA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D32ADD_AS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D32ADD xr%0,xr%1,xr%2,xr%3,AS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D32ADD_SA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D32ADD xr%0,xr%1,xr%2,xr%3,SA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D32ADD_SS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D32ADD xr%0,xr%1,xr%2,xr%3,SS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/***********************************D32ACC********************************/
#define D32ACC_AA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D32ACC xr%0,xr%1,xr%2,xr%3,AA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D32ACC_AS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D32ACC xr%0,xr%1,xr%2,xr%3,AS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D32ACC_SA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D32ACC xr%0,xr%1,xr%2,xr%3,SA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define D32ACC_SS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("D32ACC xr%0,xr%1,xr%2,xr%3,SS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/***********************************S32CPS********************************/
#define S32CPS(xra,xrb,xrc) \
do { \
__asm__ __volatile ("S32CPS xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define S32ABS(xra,xrb) \
do { \
__asm__ __volatile ("S32CPS xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrb)); \
} while (0)
/***********************************Q16ADD********************************/
#define Q16ADD_AA_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,AA,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_AA_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,AA,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_AA_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,AA,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_AA_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,AA,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_AS_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,AS,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_AS_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,AS,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_AS_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,AS,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_AS_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,AS,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_SA_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,SA,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_SA_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,SA,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_SA_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,SA,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_SA_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,SA,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_SS_WW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,SS,WW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_SS_LW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,SS,LW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_SS_HW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,SS,HW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ADD_SS_XW(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ADD xr%0,xr%1,xr%2,xr%3,SS,XW" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/***********************************Q16ACC********************************/
#define Q16ACC_AA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ACC xr%0,xr%1,xr%2,xr%3,AA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ACC_AS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ACC xr%0,xr%1,xr%2,xr%3,AS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ACC_SA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ACC xr%0,xr%1,xr%2,xr%3,SA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q16ACC_SS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q16ACC xr%0,xr%1,xr%2,xr%3,SS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/***********************************D16CPS********************************/
#define D16CPS(xra,xrb,xrc) \
do { \
__asm__ __volatile ("D16CPS xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define D16ABS(xra,xrb) \
do { \
__asm__ __volatile ("D16CPS xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrb)); \
} while (0)
/*******************************D16AVG/D16AVGR*****************************/
#define D16AVG(xra,xrb,xrc) \
do { \
__asm__ __volatile ("D16AVG xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define D16AVGR(xra,xrb,xrc) \
do { \
__asm__ __volatile ("D16AVGR xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
/************************************Q8ADD********************************/
#define Q8ADD_AA(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8ADD xr%0,xr%1,xr%2,AA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define Q8ADD_AS(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8ADD xr%0,xr%1,xr%2,AS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define Q8ADD_SA(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8ADD xr%0,xr%1,xr%2,SA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define Q8ADD_SS(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8ADD xr%0,xr%1,xr%2,SS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
/************************************Q8ADDE********************************/
#define Q8ADDE_AA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8ADDE xr%0,xr%1,xr%2,xr%3,AA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8ADDE_AS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8ADDE xr%0,xr%1,xr%2,xr%3,AS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8ADDE_SA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8ADDE xr%0,xr%1,xr%2,xr%3,SA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8ADDE_SS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8ADDE xr%0,xr%1,xr%2,xr%3,SS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/************************************Q8ACCE********************************/
#define Q8ACCE_AA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8ACCE xr%0,xr%1,xr%2,xr%3,AA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8ACCE_AS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8ACCE xr%0,xr%1,xr%2,xr%3,AS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8ACCE_SA(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8ACCE xr%0,xr%1,xr%2,xr%3,SA" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
#define Q8ACCE_SS(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8ACCE xr%0,xr%1,xr%2,xr%3,SS" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/************************************Q8ABD********************************/
#define Q8ABD(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8ABD xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
/************************************Q8SLT********************************/
#define Q8SLT(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8SLT xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
/************************************Q8SAD********************************/
#define Q8SAD(xra,xrb,xrc,xrd) \
do { \
__asm__ __volatile ("Q8SAD xr%0,xr%1,xr%2,xr%3" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd)); \
} while (0)
/********************************Q8AVG/Q8AVGR*****************************/
#define Q8AVG(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8AVG xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define Q8AVGR(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8AVGR xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
/**********************************D32SHIFT******************************/
#define D32SLL(xra,xrb,xrc,xrd,SFT4) \
do { \
__asm__ __volatile ("D32SLL xr%0,xr%1,xr%2,xr%3,%4" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd),"K"(SFT4)); \
} while (0)
#define D32SLR(xra,xrb,xrc,xrd,SFT4) \
do { \
__asm__ __volatile ("D32SLR xr%0,xr%1,xr%2,xr%3,%4" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd),"K"(SFT4)); \
} while (0)
#define D32SAR(xra,xrb,xrc,xrd,SFT4) \
do { \
__asm__ __volatile ("D32SAR xr%0,xr%1,xr%2,xr%3,%4" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd),"K"(SFT4)); \
} while (0)
#define D32SARL(xra,xrb,xrc,SFT4) \
do { \
__asm__ __volatile ("D32SARL xr%0,xr%1,xr%2,%3" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(SFT4)); \
} while (0)
#define D32SLLV(xra,xrd,rb) \
do { \
__asm__ __volatile ("D32SLLV xr%0,xr%1,%z2" \
: \
:"K"(xra),"K"(xrd),"d"(rb)); \
} while (0)
#define D32SLRV(xra,xrd,rb) \
do { \
__asm__ __volatile ("D32SLRV xr%0,xr%1,%z2" \
: \
:"K"(xra),"K"(xrd),"d"(rb)); \
} while (0)
#define D32SARV(xra,xrd,rb) \
do { \
__asm__ __volatile ("D32SARV xr%0,xr%1,%z2" \
: \
:"K"(xra),"K"(xrd),"d"(rb)); \
} while (0)
#define D32SARW(xra,xrb,xrc,rb) \
do { \
__asm__ __volatile ("D32SARW xr%0,xr%1,xr%2,%3" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"d"(rb)); \
} while (0)
/**********************************Q16SHIFT******************************/
#define Q16SLL(xra,xrb,xrc,xrd,SFT4) \
do { \
__asm__ __volatile ("Q16SLL xr%0,xr%1,xr%2,xr%3,%4" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd),"K"(SFT4)); \
} while (0)
#define Q16SLR(xra,xrb,xrc,xrd,SFT4) \
do { \
__asm__ __volatile ("Q16SLR xr%0,xr%1,xr%2,xr%3,%4" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd),"K"(SFT4)); \
} while (0)
#define Q16SAR(xra,xrb,xrc,xrd,SFT4) \
do { \
__asm__ __volatile ("Q16SAR xr%0,xr%1,xr%2,xr%3,%4" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd),"K"(SFT4)); \
} while (0)
#define Q16SLLV(xra,xrd,rb) \
do { \
__asm__ __volatile ("Q16SLLV xr%0,xr%1,%z2" \
: \
:"K"(xra),"K"(xrd),"d"(rb)); \
} while (0)
#define Q16SLRV(xra,xrd,rb) \
do { \
__asm__ __volatile ("Q16SLRV xr%0,xr%1,%z2" \
: \
:"K"(xra),"K"(xrd),"d"(rb)); \
} while (0)
#define Q16SARV(xra,xrd,rb) \
do { \
__asm__ __volatile ("Q16SARV xr%0,xr%1,%z2" \
: \
:"K"(xra),"K"(xrd),"d"(rb)); \
} while (0)
/*********************************MAX/MIN*********************************/
#define S32MAX(xra,xrb,xrc) \
do { \
__asm__ __volatile ("S32MAX xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define S32MIN(xra,xrb,xrc) \
do { \
__asm__ __volatile ("S32MIN xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define D16MAX(xra,xrb,xrc) \
do { \
__asm__ __volatile ("D16MAX xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define D16MIN(xra,xrb,xrc) \
do { \
__asm__ __volatile ("D16MIN xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define Q8MAX(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8MAX xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
#define Q8MIN(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q8MIN xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
/*************************************MOVE********************************/
#define S32I2M(xra,rb) \
do { \
__asm__ __volatile ("S32I2M xr%0,%z1" \
: \
:"K"(xra),"d"(rb)); \
} while (0)
#define S32M2I(xra) \
__extension__ ({ \
int __d; \
__asm__ __volatile ("S32M2I xr%1, %0" \
:"=d"(__d) \
:"K"(xra)); \
__d; \
})
/*********************************S32SFL**********************************/
#define S32SFL(xra,xrb,xrc,xrd,optn2) \
do { \
__asm__ __volatile ("S32SFL xr%0,xr%1,xr%2,xr%3,ptn%4" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"K"(xrd),"K"(optn2)); \
} while (0)
/*********************************S32ALN**********************************/
#define S32ALN(xra,xrb,xrc,rs) \
do { \
__asm__ __volatile ("S32ALN xr%0,xr%1,xr%2,%z3" \
: \
:"K"(xra),"K"(xrb),"K"(xrc),"d"(rs)); \
} while (0)
/*********************************Q16SAT**********************************/
#define Q16SAT(xra,xrb,xrc) \
do { \
__asm__ __volatile ("Q16SAT xr%0,xr%1,xr%2" \
: \
:"K"(xra),"K"(xrb),"K"(xrc)); \
} while (0)
// cache ops
// cache
#define Index_Invalidate_I 0x00
#define Index_Writeback_Inv_D 0x01
#define Index_Load_Tag_I 0x04
#define Index_Load_Tag_D 0x05
#define Index_Store_Tag_I 0x08
#define Index_Store_Tag_D 0x09
#define Hit_Invalidate_I 0x10
#define Hit_Invalidate_D 0x11
#define Hit_Writeback_Inv_D 0x15
#define Hit_Writeback_I 0x18
#define Hit_Writeback_D 0x19
// pref
#define PrefLoad 0
#define PrefStore 1
#define PrefLoadStreamed 4
#define PrefStoreStreamed 5
#define PrefLoadRetained 6
#define PrefStoreRetained 7
#define PrefWBInval 25
#define PrefNudge 25
#define PrefPreForStore 30
#define mips_pref(base, offset, op) \
__asm__ __volatile__( \
" .set noreorder \n" \
" pref %1, %2(%0) \n" \
" .set reorder" \
: \
: "r" (base), "i" (op), "i" (offset))
#define cache_op(op, addr) \
__asm__ __volatile__( \
" .set noreorder \n" \
" cache %0, %1 \n" \
" .set reorder" \
: \
: "i" (op), "m" (*(unsigned char *)(addr)))
#define i_pref(hint,base,offset) \
({ __asm__ __volatile__("pref %0,%2(%1)"::"i"(hint),"r"(base),"i"(offset):"memory");})
struct unaligned_32 { unsigned int l; } __attribute__((packed));
#define LD32(a) (((const struct unaligned_32 *) (a))->l)
#define ST32(a, b) (((struct unaligned_32 *) (a))->l) = (b)
#define REVERSE_LD32(xra, xrb, rb, s12) \
__extension__ ({ \
int __d; \
__asm__ __volatile ("S32LDD xr%1,%z3,%4\n\t" \
"S32SFL xr%1,xr%1, xr%1, xr%2, ptn0\n\t" \
"S32SFL xr%1,xr%2, xr%1, xr%2, ptn3\n\t" \
"S32SFL xr%1,xr%2, xr%1, xr%2, ptn2\n\t" \
"S32M2I xr%1,%0" \
:"=d"(__d) \
:"K"(xra), "K"(xrb), "d"(rb), "I"(s12)); \
__d; \
})
#define IU_CLZ(rb) \
__extension__ ({ \
int __d; \
__asm__ __volatile ("clz %0, %1" \
:"=d"(__d) \
:"d"(rb)); \
__d; \
})
#endif /* C_VERSION */
#endif /* JZ_MXU_H_ */