Message-ID: <41632BB2.6000202@yahoo.com.au>
Date: Wed, 06 Oct 2004 09:18:10 +1000
From: Nick Piggin <nickpiggin@yahoo.com.au>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.2) Gecko/20040820 Debian/1.7.2-4
MIME-Version: 1.0
To: Roland McGrath <roland@redhat.com>
CC: Linus Torvalds <torvalds@osdl.org>, Andrew Morton <akpm@osdl.org>,
       Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
       Ulrich Drepper <drepper@redhat.com>,
       Christoph Lameter <clameter@sgi.com>
Subject: Re: [PATCH] CPU time clock support in clock_* syscalls
References: <200410051828.i95ISVoc006842@magilla.sf.frob.com>
In-Reply-To: <200410051828.i95ISVoc006842@magilla.sf.frob.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1562
Lines: 47

Roland McGrath wrote:

>>CPU. In which case you don't need to worry about timestamp_last_tick.
> 
> 
> I don't really understand this comment.  update_cpu_clock is called from
> schedule and from scheduler_tick.  When it was last called by schedule,
> p->timestamp will mark this time.  When it was last called by
> p->scheduler_tick, rq->timestamp_last_tick will mark this time.
> Hence the max of the two is the last time update_cpu_clock was called.
> 

OK I see what its doing - ignore my comments then :P

> 
> 
>>It also seems to conveniently ignore locking when reading those values
>>off another CPU. Not a big deal for dynamic load calculations, but I'm
>>not so sure about your usage...?
> 
> 
> Here again I don't know what you are talking about.  Nothing is ever read
> "off another CPU".  A thread maintains its own sched_time counter while it
> is running on a CPU.
> 

It seemed like a syscall could read the values from a task currently
running on another CPU. If not, great.

> 
>>Lastly, even when using timestamp_last_tick correctly, I think sched_clock
>>will still drift around slightly, especially if a task switches CPUs a lot
>>(but not restricted to moving CPUs). 
> 
> 
> Please explain.
> 

As you pointed out, you are only measuring on-cpu time so this shouldn't
be a problem either.

Nick
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/