Does Newlib support printf for float/double numbers?

[Qihang Huang]

Hi everyone,

On a 64 bit AMD machine with Linux 2.6.30, I installed newlib by
filling the system call stubs with Linux system calls. Everything
works fine, except when a C program uses printf for floating point
numbers, it gets a segmentation fault. But printf works fine with
integer. So does Newlib support it? Do I need to define some macro for
supporting floating point in the initial build of newlib?

The specific place where the seg fault happened (by tracing with gdb)
is (in newlib/libc/stdio/printf.c:45)
 0x0000000000400b6c <+92>:	movaps %xmm0,-0x7f(%rdx)

Please help. What possible reason could cause that? I really run out
of ideas. Thank you very much!


Cheers,
Tim