Re: [PR debug/47106] account used vars only once

[Jan Hubicka <hubicka at ucw dot cz>]

> Yes, we can have cycles and no referenced vars in the callee.  But
> we won't inline that callee (as it isn't in SSA form), so we can just
> as well skip looking at its estimated stack frame size.

Checking in_ssa_p tests in ipa-inline, I think we check them early enough
(before we check only cgraph properties - whether edge is inlined or inlining
is recursive etc.).

Functions checking inline parameters (check_inline_limits etc.) are done only
after in_ssa_p check is done, so we should be safe here.  (with exception of
disregard inline limits, you can move that check later but any value of that
flag is OK since we won't do anything with the call)

Honza
> 
> >> The cgraphunit.c change looks ok in this regard, can you explain
> >> the predict.c change?
> >
> > We split a function after profile-guessing, and initialize_cfun copied
> > the PROFILE_GUESSED status from the original function to the split.
> > Then, when we got to running the various passes over the newly-created
> > version, we failed the assertion that profile_status != PROFILE_GUESSED
> > in gimple_predict_edge. ?I didn't try to find out why this didn't happen
> > before the pass rearrangement. ?I figured it made sense that the pass
> > rearrangement enabled some more inlining/versioning, exposing this kind
> > of situation just like it did expose the need for cleanups after early
> > inlining to avoid getting SSA verify failures because formerly-throwing
> > call stmts ended up known not to throw.
> 
> I think you are now feeding more functions into re-optimization
> (that's another reason I want to avoid the pass reorg for stage4).
> 
> Richard.
> 
> >
> > --
> > 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
> >