Messages in this thread | | | Date | Fri, 26 Oct 2012 11:24:21 +0200 | From | Ingo Molnar <> | Subject | Re: [PATCH 01/16] math128: Introduce various 128bit primitives |
| |
* Peter Zijlstra <peterz@infradead.org> wrote:
> On Thu, 2012-10-25 at 15:26 -0700, Linus Torvalds wrote: > > it's the *rest* of the "u128" math I really object to. I also wonder > > about the u64xu64 math case for SCHED_DEADLINE, because I assume that > > it doesn't actually end up using the 128-bit result in that form, but > > scales it down again some way? > > No, it does a compare on two u128, so it doesn't loose any > precision. If it were to scale down again and loose precision > I'd agree with you that introducing the u128 stuff is > pointless. > > The point is (as mentioned in the comments below) overflowing > an actual u64 is rare, however since some of this > (specifically the dl_{runtime,deadline} parameters) is user > specified, we have to assume we will overflow.
So can we control this by restricting the users and avoiding the overflow?
A 2^64 result should be a *huge* amount of space already for just about anything.
Thanks,
Ingo
| |