NOTES:

* STILL BROKEN: 

1.  The format code can't initialize GCR disks (800k).  When the
    track goes by for re-read, only sectors 1-10 or 1-11 are seen,
    which means that sector 0 and sometimes sector 11 are lost or
    corrupted.  The most likely cause would be that the code to
    initialize the track in the cache is broken and is writing a
    track that's too long.  More info when I find the bug.
2.  Formatting nit: you should not be allowed to try to format a
    SSDD disk as double-sided.  This should be done in a way that
    doesn't prevent you from taking a DSDD disk that wass previously
    formatted single-sided and reformatting it as double-sided.
    This requires actually checking for index marks on the back
    side of the disk and doing the right thing.  :-)

Cleanup: do SWIMIIICopyRegs in EnterHardwareLockSection.

* WARNINGS (that should be cleaned up):

1.  Need to include headers for kvtophys, delay, flush_dcache, invalidate_dcache,
    IOMalloc, sleep, wakeup in order to verify usage....
2.  Verify use of invalidate_dcache in place of invalidate_cache_for_io.
3.  dbdmasupport.c: implicit declaration of IOMallocContiguous and IOFreeContiguous.
4.  extern.c: implicit declarations of delay, invalidate_cache_for_io, and
    flush_dcache.
5.  floppy_iocache.c: implicit declarations of flush_dcache and invalidate_dcache.
6.  grcswimiiihal.c: implicit declaration of PrintDMA and kvtophys.
    incompatible pointer type for arg 2 of dbdma_start.
7.  swimiiicommonhal.c: implicit declaration of delay.