Subject: Re: [PATCH 23/23] clocksource tsc: add verify routine
From: Daniel Walker <dwalker@mvista.com>
To: Ingo Molnar <mingo@elte.hu>
Cc: akpm@osdl.org, linux-kernel@vger.kernel.org, johnstul@us.ibm.com,
       Thomas Gleixner <tglx@linutronix.de>
In-Reply-To: <20070131124300.GG1847@elte.hu>
References: <20070131033710.420168478@mvista.com>
	 <20070131033810.319576375@mvista.com>  <20070131124300.GG1847@elte.hu>
Content-Type: text/plain
Date: Wed, 31 Jan 2007 09:02:35 -0800
Message-Id: <1170262956.9781.117.camel@imap.mvista.com>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1265
Lines: 33

On Wed, 2007-01-31 at 13:43 +0100, Ingo Molnar wrote:
> * Daniel Walker <dwalker@mvista.com> wrote:
> 
> > I've included this as another user of the clocksource interface. I 
> > don't see a usage for this across all achitectures. So a fully generic 
> > version isn't needed.
> 
> well, this implementation is buggy in at least two ways:
> 
> firstly, it allows a circular verification dependency in highres+dyntick 
> mode between the jiffies and tsc clocksources.

In the current implementation, it's only happens if the only clocks that
exist are "tsc" and "jiffies" which makes verification impossible
anyway .. It's unlikely, but it is possible , the fix is not that
complex.

> secondly, verification is not done while the system is booting up:

Do you know that this is needed ? The current implementation
(clocksource watchdog) doesn't fully settle until after
device_initcall() .. So boot-up is about 99% complete ..

It's trivial to make my version do verification during most of
device_initcall() ..

Daniel

-
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/