[Bug rtl-optimization/26232] [4.2 Regression] cc0 targets broken; loop-invariants-move code doesn't handle cc0.

[rakdver at atrey dot karlin dot mff dot cuni dot cz]

------- Comment #11 from rakdver at atrey dot karlin dot mff dot cuni dot cz  2006-02-12 23:42 -------
Subject: Re:  [4.2 Regression] cc0 targets broken; loop-invariants-move code
doesn't handle cc0.

> What kind of invariant insns can we miss if we don't move any cc0 setters?  Are
> there things like cc0-based conditional moves/sets?
> 
> If someone has a test case with a missed optimization due to not moving any cc0
> setters at all, I'm willing to spend some time figuring out a suitable fix.

at the moment, we will not consider anything that uses CC0 invariant.
Loop.c will not, either, so it is not a regression.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26232