Re: Autoincrement ideas

[Jeffrey A Law <law at cygnus dot com>]

  In message <199801292146.KAA24270@ongaonga.chch.cri.nz>you write:
  > Jeffrey A Law writes:
  >  > There was a contribution *years* ago which introduced PRE_MODIFY
  >  > and POST_MODIFY to do basically the same thing.  I don't know why
  >  > they never went into the tree.
  > 
  >  > I did manage to save a copy of the submission as well as some of
  >  > RMS's comments.
  > 
  > Can you contact the person who submitted it?  They might know why it
  > never got included.  
Their address is included in some of the stuff I've got (basically
an rmail folder).  So you should be able to get in contact with them.

I don't see a copyright assignment on file for the person that wrote
the changes, so you might not be able to use them directly.  But they
might give us ideas about how to approach the problem of supporting
these address modes.

  > I imagine its utility was limited without mods to the loop pass.  In
  > the followinq, for example, the increment 4 needs to be preloaded into
  > a register and hoisted out of the loop, if a POST_MODIFY with a
  > register argument is going to be noticed by the flow pass.
Quite possibly.  But I think the basic concept of modeling this stuff
with PRE_MODIFY and POST_MODIFY rtl is the right approach.

We don't have to solve the whole problem at once; let's figure how we
want to represent these addressing modes first.

I'll send the rmail folder separately, it's a little large to send
to the list -- if anyone else is interested let me know and I'll
pass it along to all interested parties.

jeff