kolibrios/programs/emulator/PrMK/trunk/MASMFpuLib.asm

789 lines
21 KiB
NASM
Raw Normal View History

;FpuAtoFL PROTO :DWORD,:DWORD,:DWORD
;FpuFLtoA PROTO :DWORD,:DWORD,:DWORD,:DWORD
;FpuAdd PROTO :DWORD,:DWORD,:DWORD,:DWORD
;FpuSub PROTO :DWORD,:DWORD,:DWORD,:DWORD
;FpuMul PROTO :DWORD,:DWORD,:DWORD,:DWORD
;FpuDiv PROTO :DWORD,:DWORD,:DWORD,:DWORD
;FpuSqrt PROTO :DWORD,:DWORD,:DWORD
;FpuXexpY PROTO :DWORD,:DWORD,:DWORD,:DWORD
;FpuAbs PROTO :DWORD,:DWORD,:DWORD
;FpuTrunc PROTO :DWORD,:DWORD,:DWORD
;FpuRound PROTO :DWORD,:DWORD,:DWORD
;FpuChs PROTO :DWORD,:DWORD,:DWORD
;FpuLnx PROTO :DWORD,:DWORD,:DWORD
;FpuLogx PROTO :DWORD,:DWORD,:DWORD
;FpuEexpX PROTO :DWORD,:DWORD,:DWORD
;FpuTexpX PROTO :DWORD,:DWORD,:DWORD
;FpuSin PROTO :DWORD,:DWORD,:DWORD
;FpuCos PROTO :DWORD,:DWORD,:DWORD
;FpuTan PROTO :DWORD,:DWORD,:DWORD
;FpuArcsin PROTO :DWORD,:DWORD,:DWORD
;FpuArccos PROTO :DWORD,:DWORD,:DWORD
;FpuArctan PROTO :DWORD,:DWORD,:DWORD
;FpuSinh PROTO :DWORD,:DWORD,:DWORD
;FpuCosh PROTO :DWORD,:DWORD,:DWORD
;FpuTanh PROTO :DWORD,:DWORD,:DWORD
;FpuArcsinh PROTO :DWORD,:DWORD,:DWORD
;FpuArccosh PROTO :DWORD,:DWORD,:DWORD
;FpuArctanh PROTO :DWORD,:DWORD,:DWORD
;FpuSize PROTO :DWORD,:DWORD,:DWORD
;FpuComp PROTO :DWORD,:DWORD,:DWORD
;FpuExam PROTO :DWORD,:DWORD
;FpuState PROTO :DWORD
SRC1_FPU EQU 1
SRC1_REAL EQU 2
SRC1_DMEM EQU 4
SRC1_DIMM EQU 8
SRC1_CONST EQU 16
ANG_DEG EQU 0
ANG_RAD EQU 32
DEST_MEM EQU 0
DEST_IMEM EQU 64
DEST_FPU EQU 128
SRC2_FPU EQU 256
SRC2_REAL EQU 512
SRC2_DMEM EQU 1024
SRC2_DIMM EQU 2048
SRC2_CONST EQU 4096
STR_REG EQU 0
STR_SCI EQU 32768
FPU_PI EQU 1
FPU_NAPIER EQU 2
XAM_VALID EQU 1
XAM_ZERO EQU 2
XAM_NEG EQU 4
XAM_SMALL EQU 8
XAM_INFINIT EQU 16
CMP_EQU EQU 1
CMP_GREATER EQU 2
CMP_LOWER EQU 4
; #########################################################################
;
; FpuFLtoA
;
;##########################################################################
; -----------------------------------------------------------------------
; This procedure was written by Raymond Filiatreault, December 2002
; and modified April 2003. A minor flaw was corrected in July 2003.
; Modified March 2004 to avoid any potential data loss from the FPU
;
; This FpuFLtoA function converts an 80-bit REAL number (Src) to its
; decimal representation as a zero terminated alphanumeric string which
; is returned at the specified memory destination unless an invalid
; operation is reported by the FPU or the definition of the parameters
; (with uID) is invalid.
;
; The format of the string can be specified as regular (default) or
; scientific notation. The number of decimal places returned must also be
; specified but the total number of significant digits must not exceed 16.
; When the regular format is specified, the integer portion can also be
; padded with preceding spaces to position the decimal point at a
; specified location from the start of the string.
;
; The source can be an 80-bit REAL number from the FPU itself or from
; memory.
;
; The source is not checked for validity. This is the programmer's
; responsibility.
;
; This procedure is based on using an FPU instruction to convert the
; REAL number into a specific packed decimal format. After unpacking,
; the decimal point is positioned as required.
;
; Only EAX is used to return error or success. All other CPU registers
; are preserved. All FPU registers are preserved.
;
; -----------------------------------------------------------------------
; #########################################################################
align 4
proc FpuFLtoA stdcall, lpSrc:DWORD, lpDecimal:DWORD, lpDest:DWORD, uID:DWORD
locals
tempdw dd ?
esize dd ?
Padding dd ?
Decimal dd ?
content: times 108 db ?
tempst dt ?
bcdstr dt ?
oldcw dw ?
truncw dw ?
unpacked: times 20 db ?
endl
;get the specified number of decimals for result
;and make corrections if necessary
mov eax,[lpDecimal]
test [uID],SRC2_DMEM
jz @F
mov eax,[eax] ;get the decimals from memory
@@:
push eax
movzx eax,al ;low byte - number of decimal digits
cmp eax,15
jbe @F
mov eax,15 ;a maximum of 15 decimals is allowed
@@:
mov [Decimal],eax
pop eax
movzx eax,ah ;2nd byte - number of char before decimal point
cmp eax,17
jbe @F
mov eax,17 ;a maximum of 17 characters is allowed
@@:
mov [Padding],eax
test [uID],SRC1_FPU ;is data taken from FPU?
jz @F ;continue if not
;-------------------------------
;check if top register is empty
;-------------------------------
fxam ;examine its content
fstsw ax ;store results in AX
fwait ;for precaution
sahf ;transfer result bits to CPU flag
jnc @F ;not empty if Carry flag not set
jpe @F ;not empty if Parity flag set
jz srcerr1 ;empty if Zero flag set
@@:
fsave [content]
;----------------------------------------
;check source for Src and load it to FPU
;----------------------------------------
test [uID],SRC1_FPU
jz @F
lea eax,[content]
fld tbyte [eax+28]
jmp dest0
@@:
test [uID],SRC1_REAL ;is Src an 80-bit REAL in memory?
jz srcerr ;no proper source identificaiton
mov eax,[lpSrc]
fld tbyte [eax]
jmp dest0 ;go complete process
srcerr:
frstor [content]
srcerr1:
push edi
mov edi,[lpDest]
mov eax,"ERRO";ORRE"
stosd
mov ax,"R"
stosw
pop edi
xor eax,eax
ret
dest0:
;-------------------------------------------
;first examine the value on FPU for validity
;-------------------------------------------
fxam ;examine value on FPU
fstsw ax ;get result
fwait
sahf ;transfer to CPU flags
jz maybezero
jpo srcerr ;C3=0 and C2=0 would be NAN or unsupported
jnc getnumsize ;continue if normal finite number
;--------------------------------
;value to be converted = INFINITY
;--------------------------------
push ecx
push esi
push edi
mov edi,[lpDest]
mov al,"+"
test ah,2 ;C1 field = sign
jz @F
mov al,"-"
@@:
stosb
mov eax,"INFI";IFNI"
stosd
mov eax,"NITY";YTIN"
stosd
jmp finish
;-------------------------
;value to be converted = 0
;-------------------------
maybezero:
jpe getnumsize ;would be denormalized number
fstp st0 ;flush that 0 value off the FPU
push ecx
push esi
push edi
mov edi,[lpDest]
test [uID],STR_SCI ;scientific notation?
jnz @F ;no padding
mov ecx,[Padding]
sub ecx,2
jle @F ;no padding specified or necessary
mov al," "
rep stosb
@@:
mov ax,3020h ;" 0" szstring
stosw ;write it
jmp finish
;---------------------------
; get the size of the number
;---------------------------
getnumsize:
fldlg2 ;log10(2)
fld st1 ;copy Src
fabs ;insures a positive value
fyl2x ;->[log2(Src)]*[log10(2)] = log10(Src)
fstcw [oldcw] ;get current control word
fwait
mov ax,[oldcw]
or ax,0c00h ;code it for truncating
mov [truncw],ax
fldcw [truncw] ;insure rounding code of FPU to truncating
fist [esize] ;store characteristic of logarithm
fldcw [oldcw] ;load back the former control word
ftst ;test logarithm for its sign
fstsw ax ;get result
fwait
sahf ;transfer to CPU flags
sbb [esize],0 ;decrement esize if log is negative
fstp st0 ;get rid of the logarithm
;-----------------------------------------------------------------------
; get the power of 10 required to generate an integer with the specified
; number of significant digits
;-----------------------------------------------------------------------
mov eax,[uID]
and eax,STR_SCI
jnz .els0 ;regular decimal notation
mov eax,[esize]
or eax,eax ;check if number is < 1
js @F
; .if eax > 15 ;if number is >= 10^16
cmp eax,15
jbe .endif0
or [uID],STR_SCI ;switch to scientific notation
mov [Decimal],15 ;insures 15 decimal places in result
jmp scific
.endif0:
add eax,[Decimal]
; .if eax > 15 ;if integer + decimal digits > 16
cmp eax,15
jbe .endif1
sub eax,15
sub [Decimal],eax ;reduce decimal digits as required
.endif1:
@@:
push [Decimal]
pop [tempdw]
jmp @f
.els0: ;scientific notation
scific:
mov eax,[Decimal]
sub eax,[esize]
mov [tempdw],eax
; .endif
@@:
;----------------------------------------------------------------------------------------
; multiply the number by the power of 10 to generate required integer and store it as BCD
;----------------------------------------------------------------------------------------
;.if tempdw != 0
cmp [tempdw],0
je @f
fild [tempdw]
fldl2t
fmulp st1,st0 ;->log2(10)*exponent
fld st0
frndint ;get the characteristic of the log
fxch
fsub st,st1 ;get only the fractional part but keep the characteristic
f2xm1 ;->2^(fractional part)-1
fld1
faddp st1,st0 ;add 1 back
fscale ;re-adjust the exponent part of the REAL number
fstp st1 ;get rid of the characteristic of the log
fmulp st1,st0 ;->16-digit integer
;.endif
@@:
fbstp [bcdstr] ;->TBYTE containing the packed digits
fstsw ax ;retrieve exception flags from FPU
fwait
shr eax,1 ;test for invalid operation
jc srcerr ;clean-up and return error
;------------------------------------------------------------------------------
; unpack BCD, the 10 bytes returned by the FPU being in the little-endian style
;------------------------------------------------------------------------------
push ecx
push esi
push edi
lea esi,[bcdstr+9] ;go to the most significant byte (sign byte)
lea edi,[unpacked]
mov eax,3020h
mov cl,byte [esi] ;sign byte
; .if cl == 80h
cmp cl,0x80
jne @f
mov al,"-" ;insert sign if negative number
; .endif
@@:
stosw
mov ecx,9
@@:
dec esi
movzx eax,byte [esi]
ror ax,4
ror ah,4
add ax,3030h
stosw
dec ecx
jnz @B
mov edi,[lpDest]
lea esi,[unpacked]
test [uID],STR_SCI ;scientific notation?
jnz scientific
;************************
; REGULAR STRING NOTATION
;************************
;------------------------------
; check if padding is specified
;------------------------------
mov ecx,[Padding]
or ecx,ecx ;check if padding is specified
jz nopadding
mov edx,2 ;at least 1 integer + sign
mov eax,[esize]
or eax,eax
js @F ;only 1 integer digit if size is < 1
add edx,eax ;->number of integer digits
@@:
sub ecx,edx
jle nopadding
mov al," "
rep stosb
nopadding:
pushfd ;save padding flags
movsb ;insert sign
mov ecx,1 ;at least 1 integer digit
mov eax,[esize]
or eax,eax ;is size negative (i.e. number smaller than 1)
js @F
add ecx,eax
@@:
mov eax,[Decimal]
add eax,ecx ;->total number of digits to be displayed
sub eax,19
sub esi,eax ;address of 1st digit to be displayed
pop eax ;retrieve padding flags in EAX
; .if byte ptr[esi-1] == "1"
cmp byte [esi-1],"1"
jne @f
dec esi
inc ecx
push eax ;transfer padding flags through stack
popfd ;retrieve padding flags
jle @F ;no padding was necessary
dec edi ;adjust for one less padding byte
; .endif
@@:
@@:
rep movsb ;copy required integer digits
mov ecx,[Decimal]
or ecx,ecx
jz @F
mov al,"."
stosb
rep movsb ;copy required decimal digits
@@:
jmp finish
;********************
; SCIENTIFIC NOTATION
;********************
scientific:
movsb ;insert sign
mov ecx,[Decimal]
mov eax,18
sub eax,ecx
add esi,eax
cmp byte [esi-1],"1"
pushfd ;save flags for extra "1"
jnz @F
dec esi
@@:
movsb ;copy the integer
mov al,"."
stosb
rep movsb ;copy the decimal digits
mov al,"E"
stosb
mov al,"+"
mov ecx,[esize]
popfd ;retrieve flags for extra "1"
jnz @F ;no extra "1"
inc ecx ;adjust exponent
@@:
or ecx,ecx
jns @F
mov al,"-"
neg ecx ;make number positive
@@:
stosb ;insert proper sign
;Note: the absolute value of the size could not exceed 4931
mov eax,ecx
mov cl,100
div cl ;->thousands & hundreds in AL, tens & units in AH
push eax
and eax,0ffh ;keep only the thousands & hundreds
mov cl,10
div cl ;->thousands in AL, hundreds in AH
add ax,3030h ;convert to characters
stosw ;insert them
pop eax
shr eax,8 ;get the tens & units in AL
div cl ;tens in AL, units in AH
add ax,3030h ;convert to characters
stosw ;insert them
finish:
xor eax,eax
stosb ;string terminating 0
pop edi
pop esi
pop ecx
frstor [content]
or al,1 ;to insure EAX!=0
ret
endp
; #########################################################################
align 4
; #########################################################################
; FpuCos
;##########################################################################
;
; cos(Src) -> Dest
FpuCos:
test [flags],(1 shl 30)
jz @f ;jump if angle already in radians
fldpi ;load pi (3.14159...) on FPU
fmulp
test [flags],(1 shl 31)
jnz .1
pushd 200
jmp .2
.1: pushd 180
.2: fidiv dword [esp] ;value now in radians
fwait
add esp,4 ;clean the stack
@@: fldpi
fadd st,st ;->2pi
fxch
@@: fprem ;reduce the angle
fcos
fstsw ax ;retrieve exception flags from FPU
fwait
shr al,1 ;test for invalid operation
sahf ;transfer to the CPU flags
jpe @B ;reduce angle again if necessary
fstp st1 ;get rid of the 2pi
ret
align 4
; #########################################################################
; FpuSin
;##########################################################################
;
; sin(Src) -> Dest
FpuSin:
test [flags],(1 shl 30)
jz @f ;jump if angle already in radians
fldpi ;load pi (3.14159...) on FPU
fmulp
test [flags],(1 shl 31)
jnz .1
pushd 200
jmp .2
.1: pushd 180
.2: fidiv dword [esp] ;value now in radians
fwait
add esp,4 ;clean the stack
@@:
fldpi
fadd st,st ;->2pi
fxch
fprem ;reduce the angle
fsin
fstp st1 ;get rid of the 2pi
ret
align 4
; #########################################################################
; FpuArctan
;##########################################################################
;
; atan(Src) -> Dest
FpuArctan:
fld1
fpatan
test [flags],(1 shl 30)
jz @F ;jump if angle is required in radians
test [flags],(1 shl 31)
jnz .1
pushd 200
jmp .2
.1: pushd 180
.2: fimul dword [esp] ;*180 degrees
fldpi ;load pi (3.14159...) on FPU
fdivp ;*180/pi, angle now in degrees
add esp,4 ;clean CPU stack
ftst ;check for negative angle
fstsw ax ;retrieve status word from FPU
fwait
sahf
jnc @F ;jump if positive number
test [flags],(1 shl 31)
jnz .3
pushd 400
jmp .4
.3: pushd 360
.4: fiadd dword [esp] ;angle now 0-360
fwait
add esp,4 ;clean CPU stack
@@:
ret
align 4
; #########################################################################
; FpuTan
;##########################################################################
; a = tan(x)
FpuTan:
fldpi ;load pi (3.14159...) on FPU
fadd st,st ;->2pi
fxch
test [flags],(1 shl 30)
jz @F ;jump if angle already in radians
test [flags],(1 shl 31)
jnz .1
pushd 400
jmp .2
.1: pushd 360
.2: fmul st,st1
fidiv dword [esp] ;value now in radians
pop eax ;clean the stack
@@: fprem ;reduce the angle
fptan
fstp st ;get rid of the 1
fstp st1 ;get rid of the 2pi
ret
align 4
; #########################################################################
; FpuArccos
;##########################################################################
; sqrt(1-Src^2)
; acos(Src) = atan ------------- -> Dest
; Src
FpuArccos:
fld st ;copy cosine value
fmul st0,st1 ;cos^2
fld1
fsubrp ;1-cos^2 = sin^2
fsqrt ;->sin
fxch
fpatan ;i.e. arctan(sin/cos)
test [flags],(1 shl 30)
jz @F ;jump if angle is required in radians
test [flags],(1 shl 31)
jnz .1
pushd 200
jmp .2
.1: pushd 180
.2: fimul dword [esp] ;*180 degrees
fldpi ;load pi (3.14159...) on FPU
fdivp ;*180/pi, angle now in degrees
pop eax ;clean CPU stack
@@:
ret
align 4
; #########################################################################
; FpuArcsin
;##########################################################################
; Src
; asin(Src) = atan ------------- -> Dest
; sqrt(1-Src^2)
FpuArcsin:
fld st0 ;copy sine value
fmul st0,st0 ;sin^2
fld1
fsubrp ;1-sin^2 = cos^2
fsqrt ;->cos
fpatan ;i.e. arctan(sin/cos) = arcsin
test [flags],(1 shl 30)
jz @F ;jump if angle is required in radians
test [flags],(1 shl 31)
jnz .1
pushd 200
jmp .2
.1: pushd 180
.2: fimul dword [esp] ;*180 degrees
fldpi ;load pi (3.14159...) on FPU
fdivrp ;*180/pi, angle now in degrees
add esp,4 ;clean CPU stack
@@:
ret
align 4
; #########################################################################
; FpuEexpX
;##########################################################################
; e^(Src) = antilog2[ log2(e) * Src ] -> Dest
FpuEexpX:
fldl2e ;->log2(e)
fmulp ;->log2(e)*Src
fld st0 ;copy the logarithm
frndint ;keep only the characteristic
fsub st1,st ;keeps only the mantissa
fxch ;get the mantissa on top
f2xm1 ;->2^(mantissa)-1
fld1
faddp ;add 1 back
fscale ;scale it with the characteristic
fstp st1 ;get rid of the characteristic
ret
align 4
; #########################################################################
; FpuXexpY
;##########################################################################
; Src1^Src2 = antilog2[ log2(Src1) * Src2 ] -> Dest
FpuXexpY:
fxch ;set up FPU registers for next operation
fyl2x ;->log2(Src1)*exponent
fld st0 ;copy the logarithm
frndint ;keep only the characteristic
fsub st1,st ;keeps only the mantissa
fxch ;get the mantissa on top
f2xm1 ;->2^(mantissa)-1
fld1
faddp ;add 1 back
fscale ;scale it with the characteristic
fstp st1 ;overwrite the characteristic
ret
if 0
exp:
.N equ 10
xor eax,eax
mov ecx,.N
fld1
.0: dec ecx
jz .end
mov eax,.N
sub eax,ecx
mov [.d],eax
fild [.d]
@@: dec eax
jz @f
mov [.d],eax
fild [.d]
fmulp
jmp @b
@@: mov eax,.N
sub eax,ecx
fld st2
@@: dec eax
jz @f
fld st3
fmulp
jmp @b
@@: fdivrp
faddp
jmp .0
.end: fxch
fstp st0
ret
.d dd 0
endf