Messages in this thread | ![/](/images/icornerl.gif) | | Date | Thu, 6 Aug 1998 10:29:31 -0400 (EDT) | From | "Richard B. Johnson" <> | Subject | Re: floating-point abuse in 2.1.113 |
| |
On Thu, 6 Aug 1998, Andi Kleen wrote:
> On Thu, Aug 06, 1998 at 03:23:01PM +0200, Richard B. Johnson wrote: > > It seems that gcc does not treat a short procedure like a large one. > > This may have something to do with resources available during a kernel > > compile??? > > If it is all contained in a single expression that is very unlikely. > fold-const operates on "tree"s build by the frontend for single expressions, > and it does not depend much on context. > > For constant propagation it could make a difference, because the gcc backend > works with the RTL abstraction then and depends on context (e.g. various > optimizations are not done when the function contains gotos etc.) > > -Andi >
In any event, the way to prevent unexplained compiler real-number quirks from happening is don't use real-numbers in the kernel. That was the whole point --and I see that a lot of the (int)(float)x.x... is being removed and replaced with integer math, which is good(tm).
Cheers, Dick Johnson ***** FILE SYSTEM MODIFIED ***** Penguin : Linux version 2.1.113 on an i586 machine (66.15 BogoMips). Warning : It's hard to remain at the trailing edge of technology.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html
| ![\](/images/icornerr.gif) |