> 1. What uController and uProcessor using harvard architecture?
> 2. What is the advantages and disadvantages about Von and harvard?

Nearly all microcontrollers use an architecture functionally equivalent to
a Harvard architecture (from the programmer's viewpoint); generally there
are a few features which differ between Harvard and Von Neuman machines:

<1> Ability to read from code space: guaranteed on Von Neuman architectures;
    possible, via kludge, on some Harvard architecture machines.

<2> Ability to write to code space: only possible on Von Neuman machines,
    though some Harvard Architecture machines may be modified to allow writes
    to code space (often not elegantly, but nonetheless a useful feature
    during code development).

<3> Simultaneous transactions on code/data busses: only possible on the
    Harvard architectures; can result in big speed improvement.

Personally, I think #3 is the best distinction to be made between Harvard
and Von architectures; let's see how some common uC's hold up:

[for first two columns, Y=do-able with no time penalty, k=do-able via special-
ized instructions and/or a time penalty, K=do-able via hardware mods; -=no]

123  uC

K-Y  Microchip PIC 16Cxx series
kkY  Microchip PIC 17Cxx series
kKY  Intel 8x51 series [data space assumed to be the 128/256 bytes internal]
YYN  Motorola HCxx.
k-Y  NatSemi COP8

Note that nearly all microprocessors use a single code/data bus, at least
externally; while many have seperate code/data caches, the issues raised
by this go beyond what can be dealt with in this post.