On Thu, Feb 21, 2013 at 07:19:10PM +0100, Andi Kleen wrote: > > So I am fine with the cutoff. We may need to add more overflow guards (we > > already have quite few for time) that makes me wonder if all this should > > not be > > done all in saturating arithmetic now when it can be done theoretically > > with one > > C++ class? > > Sounds like a good idea, although I don't know how intrusive that is. But can > try.
I looked at this now and it would be very intrusive. Not only in tree-inline, but also various callers elsewhere. And there's no saturating class either, so that would need to be added too. I think I prefer the simple solution. BTW this is a regression, old gccs compiled this. -Andi