On 10 April 2011 20:13, Aurelien Jarno <aurel...@aurel32.net> wrote: > Fixing the definition of CPU86_LDoubleU allow building cpu-exec.o with > softfloat. > > Cc: Peter Maydell <peter.mayd...@linaro.org> > Signed-off-by: Aurelien Jarno <aurel...@aurel32.net> > --- > target-i386/exec.h | 4 ++-- > 1 files changed, 2 insertions(+), 2 deletions(-) > > diff --git a/target-i386/exec.h b/target-i386/exec.h > index 6f9f709..d6922f3 100644 > --- a/target-i386/exec.h > +++ b/target-i386/exec.h > @@ -145,7 +145,7 @@ static inline void svm_check_intercept(uint32_t type) > > /* only for x86 */ > typedef union { > - long double d; > + floatx80 d; > struct { > unsigned long long lower; > unsigned short upper; > @@ -164,7 +164,7 @@ typedef union { > > /* NOTE: arm is horrible as double 32 bit words are stored in big endian ! */ > typedef union { > - double d; > + float64 d; > #if !defined(HOST_WORDS_BIGENDIAN) && !defined(__arm__) > struct { > uint32_t lower;
I note that the !USE_X86LDOUBLE typedef for CPU86_LDoubleU is an out-of-date version of cpu-all.h's CPU_DoubleU (it doesn't get the ARM VFP endianness right). Given patch 1 in this series, is it possible now just to have #ifdef USE_X86LDOUBLE typedef CPU_LDoubleU CPU86_LDoubleU; /* other stuff */ #else typedef CPU_DoubleU CPU86_LDoubleU; /* ... */ #endif ? -- PMM