While it is optimal to do hall measurements on a "Hall Bar" geometry, it is not required. In particular, if you want to do good quick-and-dirty hall measurements on a semiconductor film, you don't need to make a hall bar out of your sample: just four contacts will do!
The above figure shows sample geometries for hall measurements from NIST's hall effect page.
You can correct for the geometry by making measurements for the resistivity in both directions and for the hall coefficient in a crossed arrangement:
The sheet resistivity, which yeilds the mobility, can be calculated from the van der Pauw equation:
exp(-piRA/RS) + exp(-piRB/RS) = 1
This is solved numerically.
I've got a copy of van der Pauw's original paper here.