[Bug debug/42244] [4.5 Regression] var-tracking ICE for 300.twolf

[aoliva at gcc dot gnu dot org]

------- Comment #3 from aoliva at gcc dot gnu dot org  2009-12-03 15:34 -------
@richi it would, if we didn't have code to deal with that in sched-deps.  now,
depl_on_debug_p() are accepted, ignored for purposes of scheduling decisions,
and used only to reset debug stmts that were rendered invalid when insns that
anti-dependent on them were scheduled ahead of them

@jakub thanks, the patch is good.  making the change at the assert rather than
at the caller would work, but then it wouldn't help catch situations that don't
deal with debug insns but that might have to.  explicitly dealing with them is
kind of a way of saying “yeah, I thought this through”, even if that's all it
takes.  Now, this used to be far more important, before we dealt with
dep-on-debug the way we do now.  Maybe we could make this simplification in 4.6
or so?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42244