21 messages in com.xensource.lists.xen-ia64-develRe: [Xen-ia64-devel] [PATCH][3/3] Ste...
FromSent OnAttachments
Atsushi SAKAI07 Mar 2007 02:03.patch
Alex Williamson07 Mar 2007 15:59 
Atsushi SAKAI08 Mar 2007 03:51 
Alex Williamson08 Mar 2007 09:41 
Atsushi SAKAI08 Mar 2007 23:58.patch
Alex Williamson12 Mar 2007 11:43 
Aron Griffis05 May 2008 20:33 
Isaku Yamahata06 May 2008 20:45 
Atsushi SAKAI08 May 2008 04:02 
Isaku Yamahata08 May 2008 04:48 
Atsushi SAKAI08 May 2008 23:48 
Isaku Yamahata09 May 2008 00:02 
Atsushi SAKAI09 May 2008 00:30 
Isaku Yamahata11 May 2008 20:39 
Atsushi SAKAI13 May 2008 04:24 
Isaku Yamahata13 May 2008 19:07 
Atsushi SAKAI14 May 2008 02:23 
Isaku Yamahata14 May 2008 23:55 
Atsushi SAKAI15 May 2008 02:01 
Isaku Yamahata15 May 2008 02:10 
Atsushi SAKAI15 May 2008 03:46 
Subject:Re: [Xen-ia64-devel] [PATCH][3/3] Steal time accounting forPVdomain/IA64 TAKE2
From:Atsushi SAKAI (sak@jp.fujitsu.com)
Date:05/09/2008 12:30:43 AM
List:com.xensource.lists.xen-ia64-devel

Hi, Isaku

Thank you for commenting this.

1)about new_itm value. "new_itm" is set from local_cpu_data->itm_next (later I use this as itm_next) at header part of timer_interrupt.

So it does not effect itm_next changes in consider_steal_time().

2)The difference of following time

ia64_get_itc() - (the itc of the last time the timer interrupt handler was invoked)

Every time should set next ITM like follows. local_cpu_data->itm_next(itm_next)+local_cpu_data->itm_delta(itm_delta).

So "guessed last itc" should be itm_next - itm_delta This itm_delta effect is already considered on stolentick++;

Thanks Atsushi SAKAI

Isaku Yamahata <yama@valinux.co.jp> wrote:

On Fri, May 09, 2008 at 03:48:24PM +0900, Atsushi SAKAI wrote:

Isaku Yamahata <yama@valinux.co.jp> wrote:

do_div(stolentick, NS_PER_TICK); stolentick++;

do_div(stolen, NS_PER_TICK);

if (stolen > stolentick) stolen = stolentick;

stolentick -= stolen; do_div(blocked, NS_PER_TICK);

if (blocked > stolentick) blocked = stolentick;

Could you please explain the above logic? I guess that stolentick should be ia64_get_itc() - (the itc of the last time the timer interrupt handler was invoked) or something like that.

your suggested value is new_itm. That variable keeps as "local_cpu_data->itm_next" in the ia64 time code.

No. local_cpu_data->itm_next doesn't hold such value because the valuable is updated by consider_steal_time() so that the wanted value is lost.

-- yamahata

_______________________________________________ Xen-ia64-devel mailing list Xen-@lists.xensource.com http://lists.xensource.com/xen-ia64-devel