I've realized that often in my code (and I've seen it in HmH too) I assume that adding 1 to a valid memory position won't do an overflow.

For example, to iterate over things contained in a memory block, you can do:

int *limit = (int *)(my_memory_block + memory_block_size);
int *it = (int *)my_memory_block;
while(it < limit){
    ...
    it++;
}

But if your allocator had given you a block including the last addressable byte, then this would not work as expected, because 'limit' would overflow into 0. Is this a real concern?

EDIT: I found online that on Windows and Linux the upper and lower addresses are used by the OS so your application won't get that. I guess all other 32-bit or 64-bit platforms will do the same, so I guess it's not really a concern, but if someone knows more about other platforms, that can still help clear things out.

This video might interest you, but it's only about windows: Mysteries of Memory Management Revealed,with Mark Russinovich.

And not all 64-bit addresses are valid on x86 platform. Currently only 48-bit address space is available to use - 128TB for userspace, and 128TB for kernel space. So for your user space pointers upper 17 bits are always zero.

Thanks as always guys. The video linked seems very interesting.