About the format of floating point...

[Stephen L Moshier]

>    .float 0f1.00000000000000000000e0

To generate this output, change the ASM_OUTPUT_FLOAT macro
to look something like this:

#define ASM_OUTPUT_FLOAT(FILE,VALUE)  \
do { char dstr[30];  \
     REAL_VALUE_TO_DECIMAL (VALUE, "%.20g", dstr);  \
     fprintf (FILE, "\t.float 0f%s\n", dstr);  \
   } while (0)

If you want NaNs and infinity to work this way, you might adapt
the macro found in m68k/sun3.h.  But you would need to find out
what your assembler wants to receive for these special values.