ranges: use signed type for negative ints less than an unsigned parameter

ranges: use signed type for negative ints less than an unsigned parameter

Say you can't divide the returns up into success and fail.  So you have
some negative values mixed with positive values less than size_t size.
We could either lose the comparison or just use signed values.  Using
signed values is a bit of a hack but I feel like it's the right thing.

In the old code, the negative values just silently disappeared without a
trace.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>