Re: Timestamp counters on SMP

From: Boris Okun (bokun@home.com)
Date: Tue Mar 28 2000 - 12:44:53 EST


Mark Hahn wrote:
>
> > see comments there. In general, I don't think you should count on TSC's
> > being syncronized at all - some people have computers with TSC's of
> > different rates (me included).
>
> in this case, TSC is completely and utterly useless to user-space.
> there are only three sensible alternatives:
> - TSC's are incoherent and TSC's are turned off in user-space.
> - TSC's can be relied upon, and the kernel syncs them.
> - the kernel saves/restores TSC contents as necessary to make
> user-space see coherent time. ouch!
>
Why would you want to use TSC in the user space? Coherent time and TSC
are two different things. I think relying on TSC content in the user
space is like relying on content of any other register, we don't expect
eax to have any particular value ;-)
 
> the sanity of SMP with different clocks is ambiguously in any case:
> the MPS standard says that OS's should merely try to function,
> but doesn't mandate any particular or sensible user-space behavior.
>
> regards, mark hahn.

Boris

-
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.tux.org/lkml/



This archive was generated by hypermail 2b29 : Fri Mar 31 2000 - 21:00:22 EST