[PATCH] Fix DEBUG stmt handling in the vectorizer (PR debug/42604)

[Alexandre Oliva]

On Jan  6, 2010, Richard Guenther <richard.guenther@gmail.com> wrote:

> Hm, I fail to see where there can be such a case w/o a definition
> which means a different SSA name ...

Remember that we're propagating the PHI def when the arg we're replacing
is the one used in a debug var.

  <pre-clone of the loop that sets x_5, to be renumbered to x_31>
<bb>:
  x_1 = PHI <x_5>;

  <loop that sets x_5>
<bb>:
  x_3 = PHI <x_5>;

  # DEBUG x => x_5;  

Surely you agree we don't want x_1 in the debug stmt.  We want x_3.

The same applies if we insert a duplicate of the loop *after* it rather
than before, or even both before and after.

> Btw, do you now insert PHIs for the case where the only uses outside
> of the loop are debug-uses?

No, that would most certainly wreak havoc in -fcompare-debug.

We use PHIs that would have been added anyway, or we reset the debug
stmt that would require one.  I have some thoughts on how to avoid
resetting, but I'll save them for another day.

-- 
Alexandre Oliva, freedom fighter    http://FSFLA.org/~lxoliva/
You must be the change you wish to see in the world. -- Gandhi
Be Free! -- http://FSFLA.org/   FSF Latin America board member
Free Software Evangelist      Red Hat Brazil Compiler Engineer