Re: ktimers in RT causing bad bogomips and more.

[Steven Rostedt <rostedt@goodmis.org>]

On Wed, 2005-10-26 at 00:26 +0200, Thomas Gleixner wrote:
> On Tue, 2005-10-25 at 18:15 -0400, Steven Rostedt wrote:
> > 	if (!hres->active)
> > 		return CLOCK_EVT_RUN_CYCLIC;
> 
> I'm searching for the brown paperbag. 
> 
> Thats the startup code before switching over to high resolution, so it
> should return 1, nothing else.
> 
> CLOCK_EVT_RUN_CYCLIC is a leftover of the initial implementation, which
> did not take tick overruns into account. 

Here, I'll make it easy for you. ;-)

-- Steve

Index: rt_linux_ernie/kernel/ktimers.c
===================================================================
--- rt_linux_ernie.orig/kernel/ktimers.c	2005-10-25 18:31:51.000000000 -0400
+++ rt_linux_ernie/kernel/ktimers.c	2005-10-25 18:32:38.000000000 -0400
@@ -338,10 +338,9 @@
 	 * we expect, that the event source is running in periodic
 	 * mode when it is a source serving other (tick based)
 	 * functionality than next event
-	 *
 	 */
 	if (!hres->active)
-		return CLOCK_EVT_RUN_CYCLIC;
+		return 1;
 
 	now = do_ktime_get();