REG_INC notes going AWOL during reload

[Jeffrey A Law]

  In message <14035.11111.883403.300067@ongaonga.elec.canterbury.ac.nz>you writ
  > Michael Hayes writes:
  >  > Jeffrey A Law writes:
  >  >  > Handling it in reload is also a little cleaner -- as far as the rest 
  > of
  >  >  > the compiler is concerned reload kept the REG_INC notes accurate.
  >  > 
  >  > In the interim, here's a patch to flow.c so that it is not dependent
  >  > upon the accuracy of the REG_INC notes.  This fixes the dead store bug.
  > 
  > Please ignore this previous patch I sent for the meantime since I have
  > rewritten it to take advantage of the following patch which correctly
  > restores the REG_INC notes in reload.  Note, I don't have access to a
  > machine with autoincrement addressing modes that I can bootstrap with;
  > however, it appears to work correctly.
  > 
  > Wed Feb 24 11:07:27 1999  Michael Hayes  <m.hayes@elec.canterbury.ac.nz>
  > 
  > 	* rtlanal.c (auto_inc_p): New function.
  >         * rtl.h (auto_inc_p): Prototype it.
  > 	* reload1.c (add_auto_inc_notes): New function.
  > 	(reload): Strip REG_INC notes and call add_auto_inc_notes
  > 	for each insn to restore them correctly.
You need a comment before the definition of add_auto_inc_notes.  Otherwise
it looks pretty good.  I've got a bootstrap of an auto-inc machine running
overnight.  Hold off on the install until I've got the bootstrap results
though please.

Thanks!

jeff