BUG in h8300.md (cc peepholes)

[Jeffrey A Law]

  In message < 199912150805.JAA05778@server5.aranea.net >you write:
  > I discovered a bug in h8300.md !!
  > ---------------------------------
  > 
  > The very last definition installs a peephole for
  > predecremented, register indirect memory accesses:
  > 
  > But due to a typo this optimization isn't in effect.
  > The line
  >    (plus:HI (match_dup 1) (const_int -1))) ;; <<wrong decrement!
  > 
  > should read
  >    (plus:HI (match_dup 1) (const_int -2))) ;; int is 2 bytes !!
  > 
  > In normal production code, the impact of this bug is not
  > very serous: the optimization is simply not done.
Thanks.  I have fixed this bug in the development sources.

jeff