Xilinx IP core drivers for RTEMS- diff attached

[gregory.menke at gsfc.nasa.gov]

Chris Caudle writes:
 > > I like the new bsp idea as well.
 > 
 > I would like to add a vote for new BSP as well.
 > There are six models of PowerPC 405 devices with IBM designed peripherals
 > which are likely to not be the same as the peripherals you end up with in
 > a Virtex.
 > http://www.amcc.com/Embedded/
 > 
 > Unless a "BSP" could be designed in such a way that you could call in
 > instances of either the IBM peripherals in the appropriate addresses to
 > use them on a 405GP, 405CR, 405EP, etc. I think it would be confusing to
 > use the same name for the processor products integrated with peripherals,
 > and the processor core embedded in an FPGA.
 > 

I think the way to approach the bsp is to set up a "virtex4_shared" bsp;
sort of like how the motorola_shared bsp works.  So all Virtex 4
variants get the same base bsp, presumably with some defines to handle
idiosyncracies.

I suppose libcpu would need a ppc_405, perhaps also set up with defines
to handle some of the architectural variations within the 405 product
line.  At present, ppc_403 contributes some code to gen405 which made
things confusing sometimes.

I'm of two minds with respect to the soft core mac though.  On one hand,
its probably logical to put it out in libchip where my diffs place it,
though its unlikely to be used by anything other than a Virtex 4.
However I'd like to defer that question to people more experienced with
the Xilinx product line to answer that for sure.  If its only used by
Virtex 4, then it would probably make sense to move it back there.  If
the firmware has small variations between the different Virtex designs,
then I think it ought to go into libchip and be renamed to something
like "virtex_softmac".

Greg