From mboxrd@z Thu Jan  1 00:00:00 1970
Subject: Re: [Xenomai-core] [RFC, PATCH] per-thread exec-time stats
From: Philippe Gerum <rpm@xenomai.org>
In-Reply-To: <17582.13798.272437.979081@domain.hid>
References: <44ACF5FA.2050205@domain.hid> <1152196379.4978.74.camel@domain.hid>
	<17581.41221.527736.416354@domain.hid>
	<1152258866.5017.4.camel@domain.hid>
	<17582.13798.272437.979081@domain.hid>
Content-Type: text/plain
Date: Fri, 07 Jul 2006 15:23:30 +0200
Message-Id: <1152278610.5035.12.camel@domain.hid>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Reply-To: rpm@xenomai.org
List-Id: "Xenomai life and development \(bug reports, patches,
	discussions\)" <xenomai.xenomai.org>
List-Unsubscribe: <https://mail.gna.org/listinfo/xenomai-core>,
	<mailto:xenomai-core-request@domain.hid>
List-Archive: </public/xenomai-core>
List-Post: <mailto:xenomai@xenomai.org>
List-Help: <mailto:xenomai-core-request@domain.hid>
List-Subscribe: <https://mail.gna.org/listinfo/xenomai-core>,
	<mailto:xenomai-core-request@domain.hid>
To: Gilles Chanteperdrix <gilles.chanteperdrix@xenomai.org>
Cc: Jan Kiszka <jan.kiszka@domain.hid>, xenomai-core <xenomai@xenomai.org>

On Fri, 2006-07-07 at 12:22 +0200, Gilles Chanteperdrix wrote:
> Philippe Gerum wrote:
>  > On Fri, 2006-07-07 at 01:47 +0200, Gilles Chanteperdrix wrote:
>  > > Philippe Gerum wrote:
>  > >  > > +#ifdef CONFIG_XENO_OPT_STATS
>  > >  > > +static inline void xnpod_acc_exec_time(xnsched_t *sched,
>  > >  > > +                                       xnthread_t *threadout)
>  > >  > > +{
>  > >  > > +    xnticks_t now = xntimer_get_rawclock();
>  > >  > > +    threadout->stat.exec_time += now - sched->last_csw;
>  > >  > > +    sched->last_csw = now;
>  > >  > > +}
>  > >  > 
>  > >  > It would be better to only pass the thread pointer, then use the
>  > >  > thread->sched member. This would clearly explain the relationship
>  > >  > between both, and prevent any bugous attempt at mixing things.
>  > > 
>  > > Beware, when xnpod_schedule is called from within
>  > > xnpod_migrate_thread(), the sched pointer of the switched out thread is
>  > > the one of the destination cpu. So, passing the sched and threadout
>  > > pointers to xnpod_acc_exec_time is safer.
>  > 
>  > Nope, the nklock is held by the migrating thread all along until the
>  > switch has actually occured.
> 
> xnpod_acc_exec_time() is called from within xnpod_schedule() that is
> called from within xnpod_migrate_thread() at a time where
> threadout->sched is the wrong pointer.

Gasp. Unfortunately, we are trapped by an exception case, and I don't
see any better approach.

> 
-- 
Philippe.