Why is a UINT_MAX 64 bits on 32 bit linux OS?
 

Linux debianvm 2.6.32-5-686 #1 SMP Sun May 6 04:01:19 UTC 2012 i686 GNU/Linux

This is true:

#if UINT_MAX == 0xFFFFFFFFFFFFFFFFULL
#error UINT_MAX
#endif


?

I have a feeling it's the preprocessor since they define it as ~0U

it fails this:

#if UINT_MAX == 0xFFFFFFFFU

but it should be the same.

Yes, that's what it is, gcc / g++ doesn't handle ~0U as expected since U should make it an unsigned int in the preprocessor (ULL for larger values). That's the way all the other compilers I've used over the years work.