Fixing linkage error. First successful euser build.
// Copyright (c) 1995-2009 Nokia Corporation and/or its subsidiary(-ies).
// All rights reserved.
// This component and the accompanying materials are made available
// under the terms of the License "Eclipse Public License v1.0"
// which accompanies this distribution, and is available
// at the URL "http://www.eclipse.org/legal/epl-v10.html".
//
// Initial Contributors:
// Nokia Corporation - initial contribution.
//
// Contributors:
//
// Description:
// e32\include\arm_vfp.h
//
// WARNING: This file contains some APIs which are internal and are subject
// to change without notice. Such APIs should therefore not be used
// outside the Kernel and Hardware Services package.
//
#ifndef __ARM_VFP_H__
#define __ARM_VFP_H__
#include <cpudefs.h>
#if defined(__CPU_ARM) && defined(__CPU_HAS_VFP)
// MRCcccc Ppppp, iii, Rdddd, Cnnnn, Cmmmm, ttt
// cccc 1110 iii1 nnnn dddd pppp ttt1 mmmm
// MCRcccc Ppppp, iii, Rdddd, Cnnnn, Cmmmm, ttt
// cccc 1110 iii0 nnnn dddd pppp ttt1 mmmm
// CDPcccc pppp, iiii, Cdddd, Cnnnn, Cmmmm, ttt
// cccc 1110 iiii nnnn dddd pppp ttt0 mmmm
// CDP2 pppp, iiii, Cdddd, Cnnnn, Cmmmm, ttt
// 1111 1110 iiii nnnn dddd pppp ttt0 mmmm
// LDCcccc
// cccc 110P UNW1 nnnn dddd pppp oooo oooo
// STCcccc
// cccc 110P UNW0 nnnn dddd pppp oooo oooo
#define _MRC(cc,p,i,r,c,c2,t) asm("mrc"#cc" p"#p", "#i", r"#r", c"#c", c"#c2", "#t )
#define _MCR(cc,p,i,r,c,c2,t) asm("mcr"#cc" p"#p", "#i", r"#r", c"#c", c"#c2", "#t )
#define _CDP(cc,p,i,d,n,m,t) asm("cdp"#cc" p"#p", "#i", c"#d", c"#n", c"#m", "#t )
#define _MRC(cc,p,i,r,c,c2,t) asm("mrc"#cc" p"#p", "#i", r"#r", c"#c", c"#c2", "#t )
#define _MCR(cc,p,i,r,c,c2,t) asm("mcr"#cc" p"#p", "#i", r"#r", c"#c", c"#c2", "#t )
#define _CDPS(cc,p,i,d,n,m,t) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((p)<<8)|((i)<<20)|((t)<<5)| \
(((d)>>1)<<12)|(((n)>>1)<<16)|((m)>>1)| \
(((d)&1)<<22)|(((n)&1)<<7)|(((m)&1)<<5)| \
0x0e000000 ))
#define VFP_CPID_S 10 // coprocessor ID for single precision
#define VFP_CPID_D 11 // coprocessor ID for double precision
#define VFP_XREG_FPSID 0
#define VFP_XREG_FPSCR 1
#define VFP_XREG_MVFR0 7
#define VFP_XREG_FPEXC 8
#define VFP_XREG_FPINST 9
#define VFP_XREG_FPINST2 10
#define VFP_FPSID_IMP_SHIFT 24
#define VFP_FPSID_IMP_MASK (255u<<VFP_FPSID_IMP_SHIFT)
#define VFP_FPSID_SW 0x00800000 // software emulation
#define VFP_FPSID_FMT_SHIFT 21
#define VFP_FPSID_FMT_MASK (3<<VFP_FPSID_FMT_SHIFT)
#define VFP_FPSID_FMT1 0x00000000 // FLDMX/FSTMX format 1
#define VFP_FPSID_FMT2 0x00200000 // FLDMX/FSTMX format 2
#define VFP_FPSID_SNG 0x00100000 // single precision only
#define VFP_FPSID_ARCH_SHIFT 16
#define VFP_FPSID_ARCH_MASK (15<<VFP_FPSID_ARCH_SHIFT)
#define VFP_FPSID_PART_SHIFT 8
#define VFP_FPSID_PART_MASK (255<<VFP_FPSID_PART_SHIFT)
#define VFP_FPSID_VAR_SHIFT 4
#define VFP_FPSID_VAR_MASK (15<<VFP_FPSID_VAR_SHIFT)
#define VFP_FPSID_REV_MASK 15
#define VFP_FPSCR_N 0x80000000 // less than
#define VFP_FPSCR_Z 0x40000000 // equal
#define VFP_FPSCR_C 0x20000000 // equal greater or unordered
#define VFP_FPSCR_V 0x10000000 // unordered
#define VFP_FPSCR_DN 0x02000000 // enable default NAN mode
#define VFP_FPSCR_FZ 0x01000000 // enable flush to zero mode
#define VFP_FPSCR_RMODE_SHIFT 22
#define VFP_FPSCR_RMODE_MASK (3<<VFP_FPSCR_RMODE_SHIFT)
#define VFP_FPSCR_RMODE_NEAR 0x00000000 // round to nearest
#define VFP_FPSCR_RMODE_PLUS 0x00400000 // round up
#define VFP_FPSCR_RMODE_MINUS 0x00800000 // round down
#define VFP_FPSCR_RMODE_ZERO 0x00C00000 // round towards zero
#define VFP_FPSCR_STRIDE_SHIFT 20
#define VFP_FPSCR_STRIDE_MASK (3<<VFP_FPSCR_STRIDE_SHIFT)
#define VFP_FPSCR_LEN_SHIFT 16
#define VFP_FPSCR_LEN_MASK (7<<VFP_FPSCR_LEN_SHIFT)
#define VFP_FPSCR_IDE 0x00008000 // enable input subnormal exception
#define VFP_FPSCR_IXE 0x00001000 // enable inexact exception
#define VFP_FPSCR_UFE 0x00000800 // enable underflow exception
#define VFP_FPSCR_OFE 0x00000400 // enable overflow exception
#define VFP_FPSCR_DZE 0x00000200 // enable division by zero exception
#define VFP_FPSCR_IOE 0x00000100 // enable invalid operation exception
#define VFP_FPSCR_IDC 0x00000080 // input subnormal cumulative flag
#define VFP_FPSCR_IXC 0x00000010 // inexact cumulative flag
#define VFP_FPSCR_UFC 0x00000008 // underflow cumulative flag
#define VFP_FPSCR_OFC 0x00000004 // overflow cumulative flag
#define VFP_FPSCR_DZC 0x00000002 // division by zero cumulative flag
#define VFP_FPSCR_IOC 0x00000001 // invalid operation cumulative flag
#define VFP_FPSCR_RUNFAST (VFP_FPSCR_DN|VFP_FPSCR_FZ)
#define VFP_FPSCR_IEEE_NO_EXC 0
#define VFP_FPSCR_EXCEPTIONS (VFP_FPSCR_IDE|VFP_FPSCR_IXE|VFP_FPSCR_UFE|VFP_FPSCR_OFE|VFP_FPSCR_DZE|VFP_FPSCR_IOE)
#define VFP_FPSCR_MODE_MASK (VFP_FPSCR_EXCEPTIONS|VFP_FPSCR_RUNFAST|VFP_FPSCR_RMODE_MASK)
#define VFP_FPEXC_EX 0x80000000 // exceptional state
#define VFP_FPEXC_EN 0x40000000 // enable VFP
#define VFP_FPEXC_FP2V 0x10000000 // FPINST2 register valid
#define VFP_FPEXC_VECITR_SHIFT 8
#define VFP_FPEXC_VECITR_MASK (7<<VFP_FPEXC_VECITR_SHIFT) // (remaining iterations - 1) mod 7
#define VFP_FPEXC_INV 0x00000080 // input exception flag (subnormal or NaN)
#define VFP_FPEXC_UFC 0x00000008 // underflow cumulative flag
#define VFP_FPEXC_OFC 0x00000004 // overflow cumulative flag
#define VFP_FPEXC_IOC 0x00000001 // invalid operation cumulative flag
#define VFP_FPEXC_INIT (VFP_FPEXC_EN|VFP_FPEXC_VECITR_MASK)
#define VFP_FPEXC_THRD_INIT (VFP_FPEXC_VECITR_MASK)
#define VFP_MVFR0_ASIMD32 0x00000002 // Full 32 x 64-bit registers are supported for Advanced SIMD
#define VFP_CPACR_ASEDIS 0x80000000 // Access to the NEON unit is disabled
#define VFP_CPACR_D32DIS 0x40000000 // Access to the upper 16 64-bit registers is disabled
#define VFP_FMRX(cc,Rd,reg) _MRC(cc,10,7,Rd,reg,0,0)
#define VFP_FMXR(cc,reg,Rd) _MCR(cc,10,7,Rd,reg,0,0)
// VFPv3 adds D16-D31 extra double precision registers in previously UNDEF opcodes
#define _VFP_DN(Dn) (((Dn)>=16) ? ((Dn)-16) : (Dn))
#define _VFP_U(Dn) (((Dn)>=16) ? 1 : 0)
#define _VFP_D(Dn) (_VFP_U(Dn)<<22)
#define _VFP_N(Dn) (_VFP_U(Dn)<<7)
#define VFP_FLDMIAX(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N)+1)|_VFP_D(Dd)|0x0c900b00 )) )
#define VFP_FSTMIAX(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N)+1)|_VFP_D(Dd)|0x0c800b00 )) )
#define VFP_FLDMIAXW(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N)+1)|_VFP_D(Dd)|0x0cb00b00 )) )
#define VFP_FSTMIAXW(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N)+1)|_VFP_D(Dd)|0x0ca00b00 )) )
#define VFP_FLDMDBXW(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N)+1)|_VFP_D(Dd)|0x0d300b00 )) )
#define VFP_FSTMDBXW(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N)+1)|_VFP_D(Dd)|0x0d200b00 )) )
#define VFP_FMDLR(cc,Dn,Rd) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rd)<<12)|(_VFP_DN(Dn)<<16)|_VFP_N(Dn)|0x0e000b10 )) )
#define VFP_FMDHR(cc,Dn,Rd) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rd)<<12)|(_VFP_DN(Dn)<<16)|_VFP_N(Dn)|0x0e200b10 )) )
#define VFP_FMRDL(cc,Rd,Dn) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rd)<<12)|(_VFP_DN(Dn)<<16)|_VFP_N(Dn)|0x0e100b10 )) )
#define VFP_FMRDH(cc,Rd,Dn) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rd)<<12)|(_VFP_DN(Dn)<<16)|_VFP_N(Dn)|0x0e300b10 )) )
#ifdef __CPU_ARM_HAS_MCRR
#define VFP_FMRRD(cc,Rd,Rn,Dm) MRRCcc(cc,11,1,Rd,Rn,Dm)
#define VFP_FMDRR(cc,Dm,Rd,Rn) MCRRcc(cc,11,1,Rd,Rn,Dm)
#endif
// MRCcccc Ppppp, iii, Rdddd, Cnnnn, Cmmmm, ttt
// cccc 1110 iii1 nnnn dddd pppp ttt1 mmmm
// MCRcccc Ppppp, iii, Rdddd, Cnnnn, Cmmmm, ttt
// cccc 1110 iii0 nnnn dddd pppp ttt1 mmmm
// CDPcccc pppp, iiii, Cdddd, Cnnnn, Cmmmm, ttt
// cccc 1110 iiii nnnn dddd pppp ttt0 mmmm
#define VFP_FABSD(cc,Dd,Dm) _CDP(cc,11,11,Dd,0,Dm,6)
#define VFP_FADDD(cc,Dd,Dn,Dm) _CDP(cc,11,3,Dd,Dn,Dm,0)
#define VFP_FCMPD(cc,Dd,Dm) _CDP(cc,11,11,Dd,4,Dm,2)
#define VFP_FCMPED(cc,Dd,Dm) _CDP(cc,11,11,Dd,4,Dm,6)
#define VFP_FCMPEZD(cc,Dd) _CDP(cc,11,11,Dd,5,0,6)
#define VFP_FCMPZD(cc,Dd) _CDP(cc,11,11,Dd,5,0,2)
#define VFP_FCPYD(cc,Dd,Dm) _CDP(cc,11,11,Dd,0,Dm,2)
#define VFP_FDIVD(cc,Dd,Dn,Dm) _CDP(cc,11,8,Dd,Dn,Dm,0)
#define VFP_FMACD(cc,Dd,Dn,Dm) _CDP(cc,11,0,Dd,Dn,Dm,0)
#define VFP_FMSCD(cc,Dd,Dn,Dm) _CDP(cc,11,1,Dd,Dn,Dm,0)
#define VFP_FMULD(cc,Dd,Dn,Dm) _CDP(cc,11,2,Dd,Dn,Dm,0)
#define VFP_FNEGD(cc,Dd,Dm) _CDP(cc,11,11,Dd,1,Dm,2)
#define VFP_FNMACD(cc,Dd,Dn,Dm) _CDP(cc,11,0,Dd,Dn,Dm,2)
#define VFP_FNMSCD(cc,Dd,Dn,Dm) _CDP(cc,11,1,Dd,Dn,Dm,2)
#define VFP_FNMULD(cc,Dd,Dn,Dm) _CDP(cc,11,2,Dd,Dn,Dm,2)
#define VFP_FSQRTD(cc,Dd,Dm) _CDP(cc,11,11,Dd,1,Dm,6)
#define VFP_FSUBD(cc,Dd,Dn,Dm) _CDP(cc,11,3,Dd,Dn,Dm,2)
#define VFP_FMSTAT(cc) _MRC(cc,10,7,15,1,0,0)
#define _VFP_ADDR_U(off) ((off)>=0 ? 1 : 0)
#define _VFP_ADDR_O(off) ((off)>=0 ? (off) : -(off))
#define VFP_FLDD(cc,Dd,Rn,off) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|(_VFP_D(Dd))|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(_VFP_ADDR_U(off)<<23)|_VFP_ADDR_O(off)|0x0d100b00)))
#define VFP_FSTD(cc,Dd,Rn,off) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|(_VFP_D(Dd))|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(_VFP_ADDR_U(off)<<23)|_VFP_ADDR_O(off)|0x0d000b00)))
#define VFP_FLDMIAD(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|(_VFP_D(Dd))|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N))|0x0c900b00 )) )
#define VFP_FSTMIAD(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|(_VFP_D(Dd))|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N))|0x0c800b00 )) )
#define VFP_FLDMIADW(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|(_VFP_D(Dd))|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N))|0x0cb00b00 )) )
#define VFP_FSTMIADW(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|(_VFP_D(Dd))|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N))|0x0ca00b00 )) )
#define VFP_FLDMDBDW(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|(_VFP_D(Dd))|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N))|0x0d300b00 )) )
#define VFP_FSTMDBDW(cc,Rn,Dd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|(_VFP_D(Dd))|((Rn)<<16)|(_VFP_DN(Dd)<<12)|(2*(N))|0x0d200b00 )) )
#define VFP_FABSS(cc,Sd,Sm) _CDPS(cc,10,11,Sd,0,Sm,6)
#define VFP_FADDS(cc,Sd,Sn,Sm) _CDPS(cc,10,3,Sd,Sn,Sm,0)
#define VFP_FCMPS(cc,Sd,Sm) _CDPS(cc,10,11,Sd,8,Sm,2)
#define VFP_FCMPES(cc,Sd,Sm) _CDPS(cc,10,11,Sd,8,Sm,6)
#define VFP_FCMPEZS(cc,Sd) _CDPS(cc,10,11,Sd,11,0,6)
#define VFP_FCMPZS(cc,Sd) _CDPS(cc,10,11,Sd,10,0,2)
#define VFP_FCPYS(cc,Sd,Sm) _CDPS(cc,10,11,Sd,0,Sm,2)
#define VFP_FDIVS(cc,Sd,Sn,Sm) _CDPS(cc,10,8,Sd,Sn,Sm,0)
#define VFP_FMACS(cc,Sd,Sn,Sm) _CDPS(cc,10,0,Sd,Sn,Sm,0)
#define VFP_FMSCS(cc,Sd,Sn,Sm) _CDPS(cc,10,1,Sd,Sn,Sm,0)
#define VFP_FMULS(cc,Sd,Sn,Sm) _CDPS(cc,10,2,Sd,Sn,Sm,0)
#define VFP_FNEGS(cc,Sd,Sm) _CDPS(cc,10,11,Sd,2,Sm,2)
#define VFP_FNMACS(cc,Sd,Sn,Sm) _CDPS(cc,10,0,Sd,Sn,Sm,2)
#define VFP_FNMSCS(cc,Sd,Sn,Sm) _CDPS(cc,10,1,Sd,Sn,Sm,2)
#define VFP_FNMULS(cc,Sd,Sn,Sm) _CDPS(cc,10,2,Sd,Sn,Sm,2)
#define VFP_FSQRTS(cc,Sd,Sm) _CDPS(cc,10,11,Sd,3,Sm,6)
#define VFP_FSUBS(cc,Sd,Sn,Sm) _CDPS(cc,10,3,Sd,Sn,Sm,2)
#define VFP_FLDS(cc,Sd,Rn,off) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|((Rn)<<16)|(_VFP_ADDR_U(off)<<23)|_VFP_ADDR_O(off)|0x0d100a00))
#define VFP_FSTS(cc,Sd,Rn,off) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|((Rn)<<16)|(_VFP_ADDR_U(off)<<23)|_VFP_ADDR_O(off)|0x0d000a00))
#define VFP_FLDMIAS(cc,Rn,Sd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(N)|0x0c900a00 )) )
#define VFP_FSTMIAS(cc,Rn,Sd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(N)|0x0c800a00 )) )
#define VFP_FLDMIASW(cc,Rn,Sd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(N)|0x0cb00a00 )) )
#define VFP_FSTMIASW(cc,Rn,Sd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(N)|0x0ca00a00 )) )
#define VFP_FLDMDBSW(cc,Rn,Sd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(N)|0x0d300a00 )) )
#define VFP_FSTMDBSW(cc,Rn,Sd,N) asm(".word %a0" : : "i" ((TInt)( ((cc)<<28)|((Rn)<<16)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(N)|0x0d200a00 )) )
#define VFP_FMSR(cc,Sn,Rd) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((Sn)>>1)<<16)|(((Sn)&1)<<7)|((Rd)<<12)|0x0e000a10))
#define VFP_FMRS(cc,Rd,Sn) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((Sn)>>1)<<16)|(((Sn)&1)<<7)|((Rd)<<12)|0x0e100a10))
#define VFP_FCVTDS(cc,Dd,Sm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((Sm)>>1)|(((Sm)&1)<<5)|((Dd)<<12)|0x0eb70ac0))
#define VFP_FCVTSD(cc,Sd,Dm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(Dm)|0x0eb70bc0))
#define VFP_FSITOD(cc,Dd,Sm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((Sm)>>1)|(((Sm)&1)<<5)|((Dd)<<12)|0x0eb80bc0))
#define VFP_FSITOS(cc,Sd,Sm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((Sm)>>1)|(((Sm)&1)<<5)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|0x0eb80ac0))
#define VFP_FTOSID(cc,Sd,Dm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(Dm)|0x0ebd0b40))
#define VFP_FTOSIZD(cc,Sd,Dm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(Dm)|0x0ebd0bc0))
#define VFP_FTOSIS(cc,Sd,Sm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((Sm)>>1)|(((Sm)&1)<<5)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|0x0ebd0a40))
#define VFP_FTOSIZS(cc,Sd,Sm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((Sm)>>1)|(((Sm)&1)<<5)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|0x0ebd0ac0))
#define VFP_FUITOD(cc,Dd,Sm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((Sm)>>1)|(((Sm)&1)<<5)|((Dd)<<12)|0x0eb80b40))
#define VFP_FUITOS(cc,Sd,Sm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((Sm)>>1)|(((Sm)&1)<<5)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|0x0eb80a40))
#define VFP_FTOUID(cc,Sd,Dm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(Dm)|0x0ebc0b40))
#define VFP_FTOUIZD(cc,Sd,Dm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|(Dm)|0x0ebc0bc0))
#define VFP_FTOUIS(cc,Sd,Sm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((Sm)>>1)|(((Sm)&1)<<5)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|0x0ebc0a40))
#define VFP_FTOUIZS(cc,Sd,Sm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((Sm)>>1)|(((Sm)&1)<<5)|(((Sd)>>1)<<12)|(((Sd)&1)<<22)|0x0ebc0ac0))
// VFPv3
// conversion between floating point and fixed point
#define _VFP_VCVT_D(dp,d) ( (dp) ? ( (((d)>>4)<<22)|(((d)&0xf)<<12) ) : ( (((d)&1)<<22)|(((d)>>1)<<12) ) )
#define _VFP_I_IMM4(sx,fbits) (((sx)==0 ? 16 : 32) - (fbits))
#define _VFP_VCVT_FBITS(sx,fbits) ( ((_VFP_I_IMM4(sx,fbits)&1) <<5) | (_VFP_I_IMM4(sx,fbits)>>1) )
#define VFP_VCT(cc,op,sf,U,sx,d,fbits) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|((op)<<18)|((U)<<16)|((sf)<<8)|((sx)<<7)|_VFP_VCVT_D(sf,d)|_VFP_VCVT_FBITS(sx,fbits)|0x0eba0a40))
// from fixed to floating point
// S32=>F64
#define VFP_VCT_F64_S32(cc,Dd,fbits) VFP_VCT((cc),0,1,0,1,(Dd),(fbits))
// S32=>F32
#define VFP_VCT_F32_S32(cc,Sd,fbits) VFP_VCT((cc),0,0,0,1,(Sd),(fbits))
// from floating point to fix
// F64=>S32
#define VFP_VCT_S32_F64(cc,Dd,fbits) VFP_VCT((cc),1,1,0,1,(Dd),(fbits))
// F32=>S32
#define VFP_VCT_S32_F32(cc,Sd,fbits) VFP_VCT((cc),1,0,0,1,(Sd),(fbits))
// put immediate value to the register
// single_register=(sz==0)
// imm (abcdefgh)
/*
bcd 000 001 010 011 100 101 110 111
efgh
0000 2.0 4.0 8.0 16.0 0.125 0.25 0.5 1.0
0001 2.125 4.25 8.5 17.0 0.1328125 0.265625 0.53125 1.0625
0010 2.25 4.5 9.0 18.0 0.140625 0.28125 0.5625 1.125
0011 2.375 4.75 9.5 19.0 0.1484375 0.296875 0.59375 1.1875
0100 2.5 5.0 10.0 20.0 0.15625 0.3125 0.625 1.25
0101 2.625 5.25 10.5 21.0 0.1640625 0.328125 0.65625 1.3125
0110 2.75 5.5 11.0 22.0 0.171875 0.34375 0.6875 1.375
0111 2.875 5.75 11.5 23.0 0.1796875 0.359375 0.71875 1.4375
1000 3.0 6.0 12.0 24.0 0.1875 0.375 0.75 1.5
1001 3.125 6.25 12.5 25.0 0.1953125 0.390625 0.78125 1.5625
1010 3.25 6.5 13.0 26.0 0.203125 0.40625 0.8125 1.625
1011 3.375 6.75 13.5 27.0 0.2109375 0.421875 0.84375 1.6875
1100 3.5 7.0 14.0 28.0 0.21875 0.4375 0.875 1.75
1101 3.625 7.25 14.5 29.0 0.2265625 0.453125 0.90625 1.8125
1110 3.75 7.5 15.0 30.0 0.234375 0.46875 0.9375 1.875
1111 3.875 7.75 15.5 31.0 0.2421875 0.484375 0.96875 1.9375
*/
#define VFP_VMOV_IMM(cc,sz,d,imm) asm(".word %a0" : : "i" ((TInt) ((cc)<<28)|(((imm)>>4)<<16)|((imm)&0xf)|(sz)<<8|((sz)?((((d)>>4)<<22)|(((d)&0xf)<<12)):((((d)&1)<<22)|(((d)>>1)<<12)))|0x0eb00a00))
#endif
#endif