Apple-2gs-SW/ORCA-C
1
0
Fork 0
mirror of https://github.com/byteworksinc/ORCA-C.git synced 2024-11-17 05:06:10 +00:00
Code Issues Projects Releases Wiki Activity
67ffeac7d4
ORCA-C/CGC.asm

276 lines
6.7 KiB
NASM
Raw Normal View History

Expand all tabs in assembly files to spaces. This allows the code to be displayed properly on GitHub and in modern text editors, which typically do not support the irregularly-spaced tab stops used for ORCA/M code. It also avoids any possibility of problems building the code if the SysTabs file is missing or has been customized with non-standard tab stops.
2018-02-11 03:55:24 +00:00
mcopy cgc.macros
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
****************************************************************
*
* CnvSX - Convert floating point to SANE extended
*
* Inputs:
* rec - pointer to a record
*
****************************************************************
*
Move some more code out of the blank segment to make space for static data.
2017-07-15 00:45:40 +00:00
CnvSX start cg
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
rec equ 4 record containing values
Process floating-point values within the compiler using the extended type. This means that floating-point constants can now have the range and precision of the extended type (aka long double), and floating-point constant expressions evaluated within the compiler also have that same range and precision (matching expressions evaluated at run time). This new behavior is intended to match the behavior specified in the C99 and later standards for FLT_EVAL_METHOD 2. This fixes the previous problem where long double constants and constant expressions of type long double were not represented and evaluated with the full range and precision that they should be. It also gives extra range and precision to constants and constant expressions of type double or float. This may have pluses and minuses, but at any rate it is consistent with the existing behavior for expressions evaluated at run time, and with one of the possible models of floating point evaluation specified in the C standards.
2021-03-05 05:52:41 +00:00
rec_real equ 0 disp to real (extended) value
rec_ext equ 10 disp to extended (SANE) value
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
tsc set up DP
phd
tcd
ph4 rec push addr of real number
clc push addr of SANE number
lda rec
adc #rec_ext
tax
lda rec+2
adc #0
pha
phx
Process floating-point values within the compiler using the extended type. This means that floating-point constants can now have the range and precision of the extended type (aka long double), and floating-point constant expressions evaluated within the compiler also have that same range and precision (matching expressions evaluated at run time). This new behavior is intended to match the behavior specified in the C99 and later standards for FLT_EVAL_METHOD 2. This fixes the previous problem where long double constants and constant expressions of type long double were not represented and evaluated with the full range and precision that they should be. It also gives extra range and precision to constants and constant expressions of type double or float. This may have pluses and minuses, but at any rate it is consistent with the existing behavior for expressions evaluated at run time, and with one of the possible models of floating point evaluation specified in the C standards.
2021-03-05 05:52:41 +00:00
fx2x convert TOS to extended
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
move4 0,4 return
pld
pla
pla
rtl
end
****************************************************************
*
* CnvSC - Convert floating point to SANE comp
*
* Inputs:
* rec - pointer to a record
*
Work around SANE bug in FX2C. This could give incorrect results for extended-to-comp conversions of certain negative integers like -2147483648 and -53021371269120. To get a fix for the same problem with regard to long long, ORCA/C should be linked with the latest version of ORCALib (which also works around some instances of the problem at run time). There are still other cases involving code in SysFloat that has not yet been patched.
2021-11-28 20:28:42 +00:00
* Note: This avoids calling FX2C on negative numbers,
* because it is buggy for certain values.
*
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
****************************************************************
*
Move some more code out of the blank segment to make space for static data.
2017-07-15 00:45:40 +00:00
CnvSC start cg
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
rec equ 4 record containing values
Process floating-point values within the compiler using the extended type. This means that floating-point constants can now have the range and precision of the extended type (aka long double), and floating-point constant expressions evaluated within the compiler also have that same range and precision (matching expressions evaluated at run time). This new behavior is intended to match the behavior specified in the C99 and later standards for FLT_EVAL_METHOD 2. This fixes the previous problem where long double constants and constant expressions of type long double were not represented and evaluated with the full range and precision that they should be. It also gives extra range and precision to constants and constant expressions of type double or float. This may have pluses and minuses, but at any rate it is consistent with the existing behavior for expressions evaluated at run time, and with one of the possible models of floating point evaluation specified in the C standards.
2021-03-05 05:52:41 +00:00
rec_real equ 0 disp to real (extended) value
rec_ext equ 10 disp to extended (SANE) value
rec_cmp equ 20 disp to comp (SANE) value
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
tsc set up DP
phd
tcd
Work around SANE bug in FX2C. This could give incorrect results for extended-to-comp conversions of certain negative integers like -2147483648 and -53021371269120. To get a fix for the same problem with regard to long long, ORCA/C should be linked with the latest version of ORCALib (which also works around some instances of the problem at run time). There are still other cases involving code in SysFloat that has not yet been patched.
2021-11-28 20:28:42 +00:00
ldy #rec_real+8
lda [rec],y
pha save sign of real number
and #$7fff
sta [rec],y set sign of real number to positive
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
ph4 rec push addr of real number
Process floating-point values within the compiler using the extended type. This means that floating-point constants can now have the range and precision of the extended type (aka long double), and floating-point constant expressions evaluated within the compiler also have that same range and precision (matching expressions evaluated at run time). This new behavior is intended to match the behavior specified in the C99 and later standards for FLT_EVAL_METHOD 2. This fixes the previous problem where long double constants and constant expressions of type long double were not represented and evaluated with the full range and precision that they should be. It also gives extra range and precision to constants and constant expressions of type double or float. This may have pluses and minuses, but at any rate it is consistent with the existing behavior for expressions evaluated at run time, and with one of the possible models of floating point evaluation specified in the C standards.
2021-03-05 05:52:41 +00:00
clc push addr of SANE comp number
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
lda rec
adc #rec_cmp
tax
lda rec+2
adc #0
pha
phx
Process floating-point values within the compiler using the extended type. This means that floating-point constants can now have the range and precision of the extended type (aka long double), and floating-point constant expressions evaluated within the compiler also have that same range and precision (matching expressions evaluated at run time). This new behavior is intended to match the behavior specified in the C99 and later standards for FLT_EVAL_METHOD 2. This fixes the previous problem where long double constants and constant expressions of type long double were not represented and evaluated with the full range and precision that they should be. It also gives extra range and precision to constants and constant expressions of type double or float. This may have pluses and minuses, but at any rate it is consistent with the existing behavior for expressions evaluated at run time, and with one of the possible models of floating point evaluation specified in the C standards.
2021-03-05 05:52:41 +00:00
fx2c convert TOS to SANE comp number
Work around SANE bug in FX2C. This could give incorrect results for extended-to-comp conversions of certain negative integers like -2147483648 and -53021371269120. To get a fix for the same problem with regard to long long, ORCA/C should be linked with the latest version of ORCALib (which also works around some instances of the problem at run time). There are still other cases involving code in SysFloat that has not yet been patched.
2021-11-28 20:28:42 +00:00
pla
bpl ret if real number was negative
ldy #rec_real+8 restore original sign of real number
sta [rec],y
sec negate the comp value
ldy #rec_cmp
ldx #0
txa
sbc [rec],y
sta [rec],y
iny
iny
txa
sbc [rec],y
sta [rec],y
iny
iny
txa
sbc [rec],y
sta [rec],y
iny
iny
txa
sbc [rec],y
sta [rec],y
ret move4 0,4 return
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
pld
pla
pla
rtl
end
Add various intermediate code peephole optimizations. These mainly cover 64-bit arithmetic and shifts, but also include a few optimizations for 16-bit and 32-bit shifts.
2021-02-19 01:17:39 +00:00
****************************************************************
*
* procedure CnvXLL (var result: longlong; val: extended);
*
* Convert floating point to long long
*
* Inputs:
* result - longlong to hold the converted value
* val - the real value
*
****************************************************************
CnvXLL start cg
subroutine (4:result,10:val),0
pei (val+8)
pei (val+6)
pei (val+4)
pei (val+2)
pei (val)
jsl ~CnvRealLongLong
pl8 [result]
return
end
****************************************************************
*
* procedure CnvXULL (var result: longlong; val: extended);
*
* Convert floating point to unsigned long long
*
* Inputs:
* result - longlong to hold the converted value
* val - the real value
*
****************************************************************
CnvXULL start cg
subroutine (4:result,10:val),0
pei (val+8)
pei (val+6)
pei (val+4)
pei (val+2)
pei (val)
jsl ~CnvRealULongLong
pl8 [result]
return
end
Implement conversions from long long to other types in the optimizer. The code of PeepHoleOptimization is now big enough that it triggers bogus "Relative address out of range" range errors from the linker. This is a linker bug and should be fixed there.
2021-02-19 05:27:18 +00:00
****************************************************************
*
* function CnvLLX (val: longlong): extended;
*
* convert a long long to a real number
*
* Inputs:
* val - the long long value
*
****************************************************************
CnvLLX start cg
subroutine (4:val),0
ph8 [val]
jsl ~CnvLongLongReal
pla
sta >rval
pla
sta >rval+2
pla
sta >rval+4
pla
sta >rval+6
pla
sta >rval+8
lla val,rval
return 4:val
rval ds 10
end
****************************************************************
*
* function CnvULLX (val: longlong): extended;
*
* convert an unsigned long long to a real number
*
* Inputs:
* val - the unsigned long long value
*
****************************************************************
CnvULLX start cg
subroutine (4:val),0
ph8 [val]
jsl ~CnvULongLongReal
pla
sta >rval
pla
sta >rval+2
pla
sta >rval+4
pla
sta >rval+6
pla
sta >rval+8
lla val,rval
return 4:val
rval ds 10
end
Move some more code out of the blank segment to make space for static data.
2017-07-15 00:45:40 +00:00
datachk off
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
****************************************************************
*
* InitLabels - initialize the labels array
*
* Outputs:
* labelTab - initialized
* intLabel - initialized
*
****************************************************************
*
Move some more code out of the blank segment to make space for static data.
2017-07-15 00:45:40 +00:00
InitLabels start cg
Increase the limit on the number of intermediate code labels in a function from 2400 to 3200. This is necessary to compile some very large functions, such as the main interpreter loop in Git. This consumes about 8K of extra memory for the additional label records.
2016-11-24 03:03:04 +00:00
maxLabel equ 3200
Change all text/source files to LF line endings.
2017-10-21 23:40:19 +00:00
! with labelTab[0] do begin
lda #-1 val := -1;
sta labelTab+6
sta labelTab+8
stz labelTab defined := false;
stz labelTab+2 chain := nil;
stz labelTab+4
! end; {with}
ldx #labelTab for i := 1 to maxLabel do
ldy #labelTab+10 labelTab[i] := labelTab[0];
lda #maxLabel*10-1
mvn labelTab,labelTab
stz intLabel intLabel := 0;
rtl
end
Add a new optimize flag for FP math optimizations that break IEEE rules. There were several existing optimizations that could change behavior in ways that violated the IEEE standard with regard to infinities, NaNs, or signed zeros. They are now gated behind a new #pragma optimize flag. This change allows intermediate code peephole optimization and common subexpression elimination to be used while maintaining IEEE conformance, but also keeps the rule-breaking optimizations available if desired. See section F.9.2 of recent C standards for a discussion of how these optimizations violate IEEE rules.
2021-11-30 02:31:15 +00:00
datachk on
****************************************************************
*
* function SignBit (val: extended): integer;
*
* returns the sign bit of a floating-point number
* (0 for positive, 1 for negative)
*
****************************************************************
*
SignBit start cg
subroutine (10:val),0
asl val+8
stz val
rol val
return 2:val
end
Reference in New Issue Copy Permalink