Re: help spot the error in this floating point code

"Bin Xin" wrote:
[...]
> So I try to follow the stack state after each call to my_pow,  it
> looks ok for a few calls (9), then I see a weird thing: when "fld1" is
> executed at some point, it loads a "nan" instead of "1" into st0:
>
> (gdb) ni
> 0x08048397      270
> 8: $st7 = 15
> 7: $st6 = 14
> 6: $st5 = 13
> 5: $st4 = 12
> 4: $st3 = 2
> 3: $st2 = 0
> 2: $st1 = 2
> 1: $st0 = 0
> (gdb) ni
> 0x08048399      270
> 8: $st7 = 14
> 7: $st6 = 13
> 6: $st5 = 12
> 5: $st4 = 2
> 4: $st3 = 0
> 3: $st2 = 2
> 2: $st1 = 0
> 1: $st0 = -nan(0xc000000000000000)
>
> So what am I missing here?

A classical stack-overrun ?
I'd try to insert a FFREE %st(7) before the FLD1
__
wolfgang