This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Add word-sized conversion functions to libgcc2
- From: Richard Sandiford <richard at codesourcery dot com>
- To: Adam Nemet <anemet at caviumnetworks dot com>
- Cc: gcc-patches at gcc dot gnu dot org, roger at eyesopen dot com, mark at codesourcery dot com
- Date: Wed, 17 May 2006 18:54:22 +0100
- Subject: Re: [PATCH] Add word-sized conversion functions to libgcc2
- References: <17512.10364.797535.170801@localhost.localdomain>
First of all, sorry for the delay in reviewing the patch, and thanks
for spending so much effort on this issue.
This message is probably going to annoy you, sorry, because I'm going
to propose an alternative patch. However, even if we go with that
alternative patch, your description of the problem has surely helped
me get this far. I'll leave it up to the reviewer (hi Mark!) to decide
which approach to go for, or whether we should do something else entirely.
Adam Nemet <anemet@caviumnetworks.com> writes:
> To recap first, for soft-float, libgcc2.c provides one set of
> functions for each floating-point mode to convert to or from
> double-word integer mode. The implementations are built up from
> simpler conversion functions from fp-bit or libgcc2.c itself. One
> exception is fixuns where in addition to double-word single-word
> conversion functions are also provided by libgcc2.c.
Right. To recap (in case it's useful to anyone reading), the conversion
functions implemented by libgcc are:
_fixunsMMWW
_fixMMDD
_fixunsMMDD
_floatDDMM
_floatunDDMM
for each MM in {SF,DF,XF,TF}, _except that_ there's no _fixunstfWW.
WW is a word mode picked by libgcc and DD is a double-word mode.
In function names, WW and DD are of course the names of the mode
that is actually being used: "qi", "hi", "si", "di", or "ti".
However -- and very confusingly -- WW is always "si" in both the
name of the libgcc2.c "L*" macro and the name of the object file.
Likewise DD and "di". This means, for example, that _fixsfti()
will be selected by L_fixsfdi rather than L_fixdfti on 64-bit
targets, and will be defined in a file called _fixsfdi.o rather
than _fixsfti.o.
In an attempt to avoid this confusion in the rest of the message,
I'll use foo() to refer to functions and Lfoo to refer to the
libgcc2 selector macros. For example, *si*() refers to a literal
SImode routine whereas L*si* refers to what I've called a "WW" routine.
> Another detail about libgcc2 and fp-bit is that the former defines
> functions for double-word integer mode whereas the latter does it for
> SImode. That means that on 64-bit we end up missing the
> DImode/single-word conversion functions. 64-bit ports supporting
> soft-float usually create port-specific routines based on expanding
> libgcc2.c to fix around this problem, see ppc64-fp.c for an example.
Right. And again to recap, "64-bit" here means "MIN_UNITS_PER_WORD == 8".
Before Roger's patch, MIN_UNITS_PER_WORD was 4 for 64-bit MIPS targets
and we got a working set of *si*() and *di*() routines. The problem that
Roger's patch fixed was that we needed some TI routines too.
> Just like ppc64-fp.c I also had to find a solution for the recursion
> between fixsfdi and fixunsfdi. To describe the problem: for 32-bit
> this how fix* and fixuns* functions call each other:
>
> fix fixuns
>
> SF<->SI fp-bit <----- libgcc2 <-.
> |
> SF<->DI libgcc2 -----> libgcc2 --'
>
>
> For 64-bit, libgcc2 functions a "pushed down" to the next widest mode
> or:
>
> fix fixuns
>
> SF<->SI fp-bit <----- <-.
> |
> SF<->DI <----- libgcc2 |
> |
> SF<->TI libgcc2 -----> libgcc2 --'
>
> If I simply provide fixsfdi and fixunssfsi using the usual 32-bit
> definitions (i.e. merge the two figures above) I get an infinite loop
> between fixsfdi and fixunsfdi:
>
> fix fixuns
>
> SF<->SI fp-bit <----- libgcc2 <-.
> |
> SF<->DI libgcc2 <----> libgcc2 |
> |
> SF<->TI libgcc2 -----> libgcc2 --'
>
> PowerPC resolves this problem by calling a local version of fixunssfdi
> from fixsfdi instead of the one in libgcc.a which in turn will call
> fixunsfsi. My solution is to arrange libgcc2.h so that we generate
> fixunssfdi using the fixunssfdi rather than the fixunssfsi template
> which will has the same effect as the PowerPC hack (see comment in the
> patch for some more details):
>
> fix fixuns
>
> SF<->SI fp-bit <----- libgcc2 <-.-.
> | |
> SF<->DI libgcc2 -----> libgcc2 --' |
> |
> SF<->TI libgcc2 -----> libgcc2 ----'
As you say, using the old MIN_UNITS_PER_WORD==4 definitions for
_fixuns*di is fine. That's what we had on MIPS before Roger's patch,
and the functions it provided worked fine.
My main concern with your patch is that it adds to, rather than steps
away from, this traditional silliness of using "si" as WW and "di" as DD
in the names of object files, regardless of the mode being used. It also
pushes the same silliness in the names of L* macros into target makefiles
(at the moment, it's usually local to the libgcc files themselves).
For example:
> +# ifdef L_fixunsdfhi
> +# undef L_fixunsdfhi
> +# define L_fixunsdfsi
> +# endif
We're having to define L_fixunsdfhi in order to get a function called
_fixunsdfsi() ("hi" -> "half a WW" -> "si").
Since you're not inventing this situation, what I said above shouldn't
be a blocker for your patch. However, your patch does make the new
behaviour conditional on a new target makefile variable, in order to
avoid breaking other targets. I agree that's a good thing. But if
we have the freedom to do what we want when that variable is defined,
I think it would be better for it to select a mode in which the names
of the object files match the names of the functions they define.
Also, as you rightly pointed out -- and I missed when I reviewed his
patch -- the functions that Roger added could be implemented equally
well by libgcc. Indeed, two of them were taken from there, just like
the ones in your first patch were.
With your patch as things stand, we'd use one technique for TFmode
conversions and another for SFmode and DFmode conversions. I think
it would be better to use libgcc for all of them.
The way I was intending to fix the problem was:
(1) to use the MIN_UNITS_PER_WORD==4 definitions for *si*() and *di*()
conversion routines, as we did earlier. (I think this is better
than treating _fixunsdfdi() as a special case.)
(2) to have a mode in which libgcc can set the word size itself,
and in which object files are named after the function they
implement. Like you, I made this conditional on a makefile
variable.
Specifically, the patch below:
- Adds a LIBGCC2_UNITS_PER_WORD macro. This macro defaults to the
word size that libgcc2 currently picks, but can be overridden
on the command line.
- Adds a macro, LIB2LITERAL_CONVERSION_MODES, in which we try
to compile:
- the L*si* templates to generate *si*() functions, with
LIBGCC2_UNITS_PER_WORD set to 4.
- the L*di* templates to generate *di*() functions, with
LIBGCC2_UNITS_PER_WORD set to 4. (As I said above,
this word size is effectively what 64-bit MIPS targets
used before; as Adam says, a 4-byte word is necessary
for _fixuns*di at the very least.)
- the L*di* templates to generate *ti*() functions, with
LIBGCC2_UNITS_PER_WORD set to 8.
I've done this allowing entries in lib2funcs to specify an object
file name, L* selector name, and word size.
- Make compiling libgcc2.c a no-op if LIBGCC2_UNITS_PER_WORD
> MIN_UNITS_PER_WORD.
Longer term, it would be nicer to:
(1) Name all object files after the functions they implement,
not just the conversion functions we're dealing with here.
(2) Replace "si" and "di" in the L* names with something
more neutral (perhaps "WW" and "DD", like I used as above.)
I haven't fully tested this patch yet. I just wanted to send it
out for comments, especially given the length of time it has taken
me to respond.
Richard
* libgcc2.c (MIN_UNITS_PER_WORD): Move default definition from
libgcc2.h.
(LIBGCC2_UNITS_PER_WORD): Provide default definition, using old
MIN_UNITS_PER_WORD logic from libgcc2.h. Do nothing if
LIBGCC2_UNITS_PER_WORD > MIN_UNITS_PER_WORD.
* libgcc2.h (MIN_UNITS_PER_WORD): Remove definition from here.
Use LIBGCC2_UNITS_PER_WORD rather than MIN_UNITS_PER_WORD to
determine the size of Wtype, etc.
* mklibgcc.in (LIB2LITERAL_CONVERSION_MODES): New argument.
(swfloatfuncs): New variable.
(dwfloatfuncs): Likewise.
(lib2funcs): Remove floating-point conversion functions from
initial assignment. Use LIB2LITERAL_CONVERSION_MODES to determine
the set of conversion routines needed. Allow entries to specify
a object name, filename and word size. Update users accordingly.
* Makefile.in (libgcc.mk): Pass LIB2LITERAL_CONVERSION_MODES.
* config/mips/t-mips (LIB2LITERAL_CONVERSION_MODES): Define.
Revert:
2006-02-08 Roger Sayle <roger@eyesopen.com>
PR target/22209
* config/fixtfdi.c: New libgcc source file.
* config/fixunstfdi.c: New source file.
* config/floatditf.c: New source file.
* config/floatunditf.c: New souce file.
* config/mips/t-iris6 (LIB2FUNCS_EXTRA): Include the new source
files above instead of config/mips/_tilib.c.
* config/mips/t-linux64 (LIB2FUNCS_EXTRA): Likewise.
Index: gcc/libgcc2.c
===================================================================
--- gcc/libgcc2.c (revision 113842)
+++ gcc/libgcc2.c (working copy)
@@ -40,6 +40,23 @@ #define ATTRIBUTE_HIDDEN __attribute__
#define ATTRIBUTE_HIDDEN
#endif
+#ifndef MIN_UNITS_PER_WORD
+#define MIN_UNITS_PER_WORD UNITS_PER_WORD
+#endif
+
+#ifndef LIBGCC2_UNITS_PER_WORD
+# if MIN_UNITS_PER_WORD > 4
+# define LIBGCC2_UNITS_PER_WORD 8
+# elif (MIN_UNITS_PER_WORD > 2 \
+ || (MIN_UNITS_PER_WORD > 1 && LONG_LONG_TYPE_SIZE > 32))
+# define LIBGCC2_UNITS_PER_WORD 4
+# else
+# define LIBGCC2_UNITS_PER_WORD MIN_UNITS_PER_WORD
+# endif
+#endif
+
+#if LIBGCC2_UNITS_PER_WORD <= MIN_UNITS_PER_WORD
+
#include "libgcc2.h"
#ifdef DECLARE_LIBRARY_RENAMES
@@ -2162,3 +2179,4 @@ func_ptr __DTOR_LIST__[2] = {0, 0};
#endif
#endif /* no INIT_SECTION_ASM_OP and not CTOR_LISTS_DEFINED_EXTERNALLY */
#endif /* L_ctors */
+#endif /* LIBGCC2_UNITS_PER_WORD <= MIN_UNITS_PER_WORD */
Index: gcc/libgcc2.h
===================================================================
--- gcc/libgcc2.h (revision 113842)
+++ gcc/libgcc2.h (working copy)
@@ -125,10 +125,6 @@ #define IS_IBM_EXTENDED(SIZE) (SIZE == 1
#define IS_IBM_EXTENDED(SIZE) 0
#endif
-#ifndef MIN_UNITS_PER_WORD
-#define MIN_UNITS_PER_WORD UNITS_PER_WORD
-#endif
-
/* In the first part of this file, we are interfacing to calls generated
by the compiler itself. These calls pass values into these routines
which have very specific modes (rather than very specific types), and
@@ -201,7 +197,7 @@ #define double bogus_type
turns out that no platform would define COMPAT_DIMODE_TRAPPING_ARITHMETIC
if it existed. */
-#if MIN_UNITS_PER_WORD > 4
+#if LIBGCC2_UNITS_PER_WORD == 8
#define W_TYPE_SIZE (8 * BITS_PER_UNIT)
#define Wtype DItype
#define UWtype UDItype
@@ -212,8 +208,7 @@ #define UDWtype UTItype
#define __NW(a,b) __ ## a ## di ## b
#define __NDW(a,b) __ ## a ## ti ## b
#define COMPAT_SIMODE_TRAPPING_ARITHMETIC
-#elif MIN_UNITS_PER_WORD > 2 \
- || (MIN_UNITS_PER_WORD > 1 && LONG_LONG_TYPE_SIZE > 32)
+#elif LIBGCC2_UNITS_PER_WORD == 4
#define W_TYPE_SIZE (4 * BITS_PER_UNIT)
#define Wtype SItype
#define UWtype USItype
@@ -223,7 +218,7 @@ #define DWtype DItype
#define UDWtype UDItype
#define __NW(a,b) __ ## a ## si ## b
#define __NDW(a,b) __ ## a ## di ## b
-#elif MIN_UNITS_PER_WORD > 1
+#elif LIBGCC2_UNITS_PER_WORD == 2
#define W_TYPE_SIZE (2 * BITS_PER_UNIT)
#define Wtype HItype
#define UWtype UHItype
Index: gcc/mklibgcc.in
===================================================================
--- gcc/mklibgcc.in (revision 113842)
+++ gcc/mklibgcc.in (working copy)
@@ -19,6 +19,7 @@
# LIB2ADDEHSTATIC
# LIB2ADDEHSHARED
# LIB2ADDEHDEP
+# LIB2LITERAL_CONVERSION_MODES
# LIBUNWIND
# LIBUNWINDDEP
# SHLIBUNWIND_LINK
@@ -64,16 +65,48 @@ echo 'dirs = libgcc'
echo
# Library members defined in libgcc2.c.
+
+# The floating-point conversion routines that involve a single-word integer.
+# XX stands for the integer mode.
+swfloatfuncs=
+for mode in sf df xf; do
+ swfloatfuncs="$swfloatfuncs _fixuns${mode}XX"
+done
+
+# Likewise double-word routines.
+dwfloatfuncs=
+for mode in sf df xf tf; do
+ dwfloatfuncs="$dwfloatfuncs _fix${mode}XX _fixuns${mode}XX"
+ dwfloatfuncs="$dwfloatfuncs _floatXX${mode} _floatunXX${mode}"
+done
+
+# Entries of the form <objfile>:<func>:<wordsize> indicate that libgcc2.c
+# should be compiled with L<func> defined and with LIBGCC2_UNITS_PER_WORD
+# set to <wordsize>. <objfile> is the name of the associated object file
+
lib2funcs='_muldi3 _negdi2 _lshrdi3 _ashldi3 _ashrdi3
- _cmpdi2 _ucmpdi2 _floatdidf _floatdisf _fixunsdfsi _fixunssfsi
- _fixunsdfdi _fixdfdi _fixunssfdi _fixsfdi _fixxfdi _fixunsxfdi
- _floatdixf _fixunsxfsi _fixtfdi _fixunstfdi _floatditf _clear_cache
+ _cmpdi2 _ucmpdi2 _floatdidf _clear_cache
_enable_execute_stack _trampoline __main _absvsi2 _absvdi2 _addvsi3
_addvdi3 _subvsi3 _subvdi3 _mulvsi3 _mulvdi3 _negvsi2 _negvdi2 _ctors
_ffssi2 _ffsdi2 _clz _clzsi2 _clzdi2 _ctzsi2 _ctzdi2 _popcount_tab
_popcountsi2 _popcountdi2 _paritysi2 _paritydi2 _powisf2 _powidf2
_powixf2 _powitf2 _mulsc3 _muldc3 _mulxc3 _multc3 _divsc3 _divdc3
- _divxc3 _divtc3 _floatundidf _floatundisf _floatundixf _floatunditf'
+ _divxc3 _divtc3'
+
+if [ "$LIB2LITERAL_CONVERSION_MODES" ]; then
+ for func in $swfloatfuncs; do
+ sifunc=`echo $func | sed -e 's/XX/si/'`
+ lib2funcs="$lib2funcs $sifunc:$sifunc:4"
+ done
+ for func in $dwfloatfuncs; do
+ difunc=`echo $func | sed -e 's/XX/di/'`
+ tifunc=`echo $func | sed -e 's/XX/ti/'`
+ lib2funcs="$lib2funcs $difunc:$difunc:4 $tifunc:$difunc:8"
+ done
+else
+ lib2funcs="$lib2funcs `echo $swfloatfuncs | sed -e 's/XX/si/g'`"
+ lib2funcs="$lib2funcs `echo $dwfloatfuncs | sed -e 's/XX/di/g'`"
+fi
# Disable SHLIB_LINK if shared libgcc not enabled.
if [ "@enable_shared@" = "no" ]; then
@@ -162,8 +195,8 @@ fi
# defined as optimized assembly code in LIB1ASMFUNCS or as C code
# in LIB2FUNCS_EXCLUDE.
for name in $LIB1ASMFUNCS $LIB2FUNCS_EXCLUDE; do
- lib2funcs=`echo $lib2funcs | sed -e 's/^'$name' //' \
- -e 's/ '$name' / /' \
+ lib2funcs=`echo $lib2funcs | sed -e 's/^'$name'[ :]//' \
+ -e 's/ '$name'[ :]/ /' \
-e 's/ '$name'$//'`
LIB2_DIVMOD_FUNCS=`echo $LIB2_DIVMOD_FUNCS | sed -e 's/^'$name' //' \
-e 's/ '$name' / /' \
@@ -265,16 +298,25 @@ for ml in $MULTILIBS; do
#
for name in $lib2funcs; do
+ case $name in
+ *:*:*)
+ defines=`echo $name | sed -e 's/.*:\(.*\):\(.*\)/-DL\1 -DLIBGCC2_UNITS_PER_WORD=\2/'`
+ name=`echo $name | sed -e 's/\(.*\):.*:.*/\1/'`
+ ;;
+ *)
+ defines="-DL$name"
+ ;;
+ esac
if [ "$libgcc_s_so" ]; then
out="libgcc/${dir}/${name}${objext}"
outS="libgcc/${dir}/${name}_s${objext}"
echo $outS: $libgcc2_c_dep
- echo " $gcc_s_compile" $flags -DL$name -c '$(srcdir)/libgcc2.c' \
+ echo " $gcc_s_compile" $flags $defines -c '$(srcdir)/libgcc2.c' \
-o $outS
echo $out: $libgcc2_c_dep
- echo " $gcc_compile" $flags -DL$name '$(vis_hide)' \
+ echo " $gcc_compile" $flags $defines '$(vis_hide)' \
-c '$(srcdir)/libgcc2.c' -o $out
echo $libgcc_a: $out
@@ -285,7 +327,7 @@ for ml in $MULTILIBS; do
else
out="libgcc/${dir}/${name}${objext}"
echo ${out}: stmp-dirs '$(srcdir)/config/$(LIB1ASMSRC)'
- echo " $gcc_compile" $flags -DL$name -c '$(srcdir)/libgcc2.c' -o $out
+ echo " $gcc_compile" $flags $defines -c '$(srcdir)/libgcc2.c' -o $out
echo $libgcc_a: $out
fi
done
Index: gcc/Makefile.in
===================================================================
--- gcc/Makefile.in (revision 113842)
+++ gcc/Makefile.in (working copy)
@@ -1415,6 +1415,7 @@ libgcc.mk: config.status Makefile mklibg
LIB2ADDEHSTATIC='$(LIB2ADDEHSTATIC)' \
LIB2ADDEHSHARED='$(LIB2ADDEHSHARED)' \
LIB2ADDEHDEP='$(LIB2ADDEHDEP)' \
+ LIB2LITERAL_CONVERSION_MODES='$(LIB2LITERAL_CONVERSION_MODES)' \
LIBUNWIND='$(LIBUNWIND)' \
LIBUNWINDDEP='$(LIBUNWINDDEP)' \
SHLIBUNWIND_LINK='$(SHLIBUNWIND_LINK)' \
Index: gcc/config/floatunditf.c
===================================================================
--- gcc/config/floatunditf.c (revision 113842)
+++ gcc/config/floatunditf.c (working copy)
@@ -1,25 +0,0 @@
-/* Public domain. */
-#if __LDBL_MANT_DIG__ == 106 || __LDBL_MANT_DIG__ == 113
-typedef int DItype __attribute__ ((mode (DI)));
-typedef int SItype __attribute__ ((mode (SI)));
-typedef unsigned int UDItype __attribute__ ((mode (DI)));
-typedef unsigned int USItype __attribute__ ((mode (SI)));
-typedef float DFtype __attribute__ ((mode (DF)));
-typedef float TFtype __attribute__ ((mode (TF)));
-
-TFtype __floatunditf (UDItype);
-
-TFtype
-__floatunditf (UDItype u)
-{
- DFtype dh, dl;
-
- dh = (USItype) (u >> (sizeof (SItype) * 8));
- dh *= 2.0 * (((UDItype) 1) << ((sizeof (SItype) * 8) - 1));
- dl = (USItype) (u & ((((UDItype) 1) << (sizeof (SItype) * 8)) - 1));
-
- return (TFtype) dh + (TFtype) dl;
-}
-
-#endif
-
Index: gcc/config/fixtfdi.c
===================================================================
--- gcc/config/fixtfdi.c (revision 113842)
+++ gcc/config/fixtfdi.c (working copy)
@@ -1,18 +0,0 @@
-/* Public domain. */
-#if __LDBL_MANT_DIG__ == 106 || __LDBL_MANT_DIG__ == 113
-typedef int DItype __attribute__ ((mode (DI)));
-typedef float TFtype __attribute__ ((mode (TF)));
-
-DItype __fixtfdi (TFtype);
-DItype __fixunstfdi (TFtype);
-
-
-DItype
-__fixtfdi (TFtype x)
-{
- if (x < 0)
- return - __fixunstfdi (-x);
- return __fixunstfdi (x);
-}
-
-#endif
Index: gcc/config/fixunstfdi.c
===================================================================
--- gcc/config/fixunstfdi.c (revision 113842)
+++ gcc/config/fixunstfdi.c (working copy)
@@ -1,35 +0,0 @@
-/* Public domain. */
-#if __LDBL_MANT_DIG__ == 106 || __LDBL_MANT_DIG__ == 113
-typedef int DItype __attribute__ ((mode (DI)));
-typedef int SItype __attribute__ ((mode (SI)));
-typedef unsigned int UDItype __attribute__ ((mode (DI)));
-typedef unsigned int USItype __attribute__ ((mode (SI)));
-typedef float TFtype __attribute__ ((mode (TF)));
-
-DItype __fixunstfdi (TFtype);
-
-DItype
-__fixunstfdi (TFtype a)
-{
- if (a < 0)
- return 0;
-
- /* Compute high word of result, as a flonum. */
- const TFtype b = (a / (((UDItype) 1) << (sizeof (SItype) * 8)));
- /* Convert that to fixed (but not to DItype!),
- and shift it into the high word. */
- UDItype v = (USItype) b;
- v <<= (sizeof (SItype) * 8);
- /* Remove high part from the TFtype, leaving the low part as flonum. */
- a -= (TFtype) v;
- /* Convert that to fixed (but not to DItype!) and add it in.
- Sometimes A comes out negative. This is significant, since
- A has more bits than a long int does. */
- if (a < 0)
- v -= (USItype) (-a);
- else
- v += (USItype) a;
- return v;
-}
-
-#endif
Index: gcc/config/mips/t-iris6
===================================================================
--- gcc/config/mips/t-iris6 (revision 113842)
+++ gcc/config/mips/t-iris6 (working copy)
@@ -6,8 +6,6 @@ MULTILIB_OSDIRNAMES=../lib32 ../lib ../l
LIBGCC = stmp-multilib
INSTALL_LIBGCC = install-multilib
-LIB2FUNCS_EXTRA = $(srcdir)/config/fixtfdi.c $(srcdir)/config/fixunstfdi.c $(srcdir)/config/floatditf.c $(srcdir)/config/floatunditf.c
-
TPBIT = tp-bit.c
tp-bit.c: $(srcdir)/config/fp-bit.c
Index: gcc/config/mips/t-linux64
===================================================================
--- gcc/config/mips/t-linux64 (revision 113842)
+++ gcc/config/mips/t-linux64 (working copy)
@@ -4,8 +4,6 @@ MULTILIB_OSDIRNAMES = ../lib32 ../lib ..
EXTRA_MULTILIB_PARTS=crtbegin.o crtend.o crtbeginS.o crtendS.o crtbeginT.o
-LIB2FUNCS_EXTRA = $(srcdir)/config/fixtfdi.c $(srcdir)/config/fixunstfdi.c $(srcdir)/config/floatditf.c $(srcdir)/config/floatunditf.c
-
TPBIT = tp-bit.c
tp-bit.c: $(srcdir)/config/fp-bit.c
Index: gcc/config/mips/t-mips
===================================================================
--- gcc/config/mips/t-mips (revision 113842)
+++ gcc/config/mips/t-mips (working copy)
@@ -19,3 +19,5 @@ fp-bit.c: $(srcdir)/config/fp-bit.c
echo '#endif' >> fp-bit.c
echo '#define QUIET_NAN_NEGATED' >> fp-bit.c
cat $(srcdir)/config/fp-bit.c >> fp-bit.c
+
+LIB2LITERAL_CONVERSION_MODES=yes
Index: gcc/config/floatditf.c
===================================================================
--- gcc/config/floatditf.c (revision 113842)
+++ gcc/config/floatditf.c (working copy)
@@ -1,25 +0,0 @@
-/* Public domain. */
-#if __LDBL_MANT_DIG__ == 106 || __LDBL_MANT_DIG__ == 113
-typedef int DItype __attribute__ ((mode (DI)));
-typedef int SItype __attribute__ ((mode (SI)));
-typedef unsigned int UDItype __attribute__ ((mode (DI)));
-typedef unsigned int USItype __attribute__ ((mode (SI)));
-typedef float DFtype __attribute__ ((mode (DF)));
-typedef float TFtype __attribute__ ((mode (TF)));
-
-TFtype __floatditf (UDItype);
-
-TFtype
-__floatditf (UDItype u)
-{
- DFtype dh, dl;
-
- dh = (SItype) (u >> (sizeof (SItype) * 8));
- dh *= 2.0 * (((UDItype) 1) << ((sizeof (SItype) * 8) - 1));
- dl = (USItype) (u & ((((UDItype) 1) << (sizeof (SItype) * 8)) - 1));
-
- return (TFtype) dh + (TFtype) dl;
-}
-
-#endif
-