Re: [PATCH 39/41] clocksource: vf_pit: Migrate to new 'set-state' interface

From: Stefan Agner <stefan@agner.ch>
To: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Thomas Gleixner <tglx@linutronix.de>,
	Daniel Lezcano <daniel.lezcano@linaro.org>,
	linaro-kernel@lists.linaro.org, linux-kernel@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org,
	Jingchang Lu <b35083@freescale.com>,
	Shawn Guo <shawn.guo@linaro.org>
Subject: Re: [PATCH 39/41] clocksource: vf_pit: Migrate to new 'set-state' interface
Date: Fri, 03 Jul 2015 13:11:04 +0200	[thread overview]
Message-ID: <6a6d4ec3156cb00537ed632f60430a37@agner.ch> (raw)
In-Reply-To: <20150703085749.GE23297@linux>

On 2015-07-03 10:57, Viresh Kumar wrote:
> On 03-07-15, 10:10, Stefan Agner wrote:
>> >  	.features	= CLOCK_EVT_FEAT_PERIODIC | CLOCK_EVT_FEAT_ONESHOT,
>> > -	.set_mode	= pit_set_mode,
>> > +	.set_state_shutdown = pit_shutdown,
>> > +	.set_state_periodic = pit_set_periodic,
>>
>> I'm not really familiar with the interface, but given that we announce
>> the feature CLOCK_EVT_FEAT_ONESHOT shouldn't we add a set_state_oneshot
>> callback here?
> 
> We weren't doing anything in pit_set_mode(ONESHOT) and so that
> callback is not implemented. In case you need to do something in
> set_state_oneshot(), we can add it back.

True, weren't doing anything. I wonder if that is right. Afaik, we
should set the same timer for oneshot too, hence call
pit_set_next_event. With your change we can just reuse the same function
(pit_set_periodic) for set_state_oneshot.

To maintain the atomicity of the changes, this would need to be fixed in
a separate patch anyway. So this change looks good to me:

Acked-by: Stefan Agner <stefan@agner.ch>

I guess "clockevents: Allow set-state callbacks to be optional" makes it
before this patch? Otherwise we would call a null pointer...

--
Stefan