Hi, I?m porting gcc for a 16-bit target with no hardware fp capabilities. Playing with floats/doubles, gcc complains about an undefined reference to __extendsfdf2. This function should be defined in libgcc.a from fp-bit.c. Looking a bit deeper, I found that _sf_to_df.o obj file where __extendsfdf2 should be defined has an is empty text section? So I guess this is a matter of Macro definition which inhibits this code. In my t-target, the only macro definitions are #define FLOAT #define CMPtype HItype #define SMALL_MACHINE And the xgcc command line is : xgcc ?B? -nostdinc -isystem ? -L? -I? -DIN_GCC -DCROSS_DIRECTORY_STRUCTURE -Os -Dinhibit_libc -fno-inline -g -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED -Dinhibit_libc -DHAVE_CC_TLS -o _sf_to_df.o -MT _sf_to_df.o -MD -MP -MF _sf_to_df.dep -DFINE_GRAINED_LIBRARIES -DL_sf_to_df -c ../.././gcc/fp-bit.c Any hint? Aurelien
Attachment:
smime.p7s
Description: S/MIME cryptographic signature
