[csmith-bugs] Issue with --float and strict-aliasing on AMD64

[Ole Christensen <ole.christensen@web.de>]

* Generator: csmith 2.2.0
* Git version: dcef523
* Options:   --seed 011 --no-math64 --no-unions --no-bitfields --float 
-o test.c
* Seed:      11

The generated func_24() should return a (int32_t *):

static int32_t * func_24(uint8_t  p_25, int32_t * const  p_26);

But actually it returns a (float *):

float *l_744 = &g_165;
...
return l_744;

This seems to lead to a aliasing issue with -O2 on AMD64 gcc-4.8.4. When 
I fix the returned value, the issue disappears.

gcc -DNO_LONGLONG -O2 test.c -o test

Cheers, Ole