lkml.org 
[lkml]   [1998]   [Jul]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Process time measurement too optimistic
On Wed, 8 Jul 1998, Pavel Machek wrote:

> Linux currently charges processes for time at end of timeslice. That

s/timeslice/jiffie/

> means that if process wakes because of timer and finishes its
> processing before end of jiffie, no time is charged to him.
>
> It is possible to exploit this: I have code which eats ~50% of system
> time, still, no time is charged to him. (This is partly dangerous
> because scheduling strategy depends on same times.)

The patch looks good, but I propose we try first what
impact it will have on interactive and multimedia
performance.

These tasks depend on the fact that their CPU time is
being charged to background simulations...
Besides, I wonder what would happen if your exploit is
being run on an already loaded machine? It'll probably
be charged for it's CPU time like all other processes.

In short: a good patch, but we should test it
thoroughly for it's impact on responsiveness.

Rik.
+-------------------------------------------------------------------+
| Linux memory management tour guide. H.H.vanRiel@phys.uu.nl |
| Scouting Vries cubscout leader. http://www.phys.uu.nl/~riel/ |
+-------------------------------------------------------------------+

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu

\
 
 \ /
  Last update: 2005-03-22 13:43    [from the cache]
©2003-2011 Jasper Spaans