Erik:
The location of the bits for the RW, RS, and E are not specified but if they are located in the low 4-bits of P1 then I do not like how the code that outputs the two nibbles to P1 causes these to clear to 0000b each time a nibble goes out.

There is also a bugaboo in the implementation wherein the busy bit is being polled after every nibble is sent out. This is incorrect and should be done only after the pair of nibbles are sent out. As a matter of fact I would lift the busy bit poll logic out of the low level routine completely and make it its own routine that is called when necessary. I do not even like to have the busy poll down in the lowest routine because then it is more difficult to support a busy wait timeout.

Michael Karas