Problems of reception with USRP2 at nonwhole MHz frequencies

I am trying to get GSM data at band 1800 (about 1800Mhz). and I
discovered that something goes wrong when I use the USRP2 to
down-convert at non whole frequencies. (eg 1810.2Mhz). It works only in
whole ones (eg 1810Mhz)
does any body else know this?

I also want to ask if the complex samples are linear in range. because I
saw that when you increase the gain, they don’t go more than about
20000, and reach there non linearly. If it’s not linear, does anyone
know the formula?