Re: [ANNOUNCE] v5.2.9-rt3

From: Alexandre Belloni <alexandre.belloni@bootlin.com>
To: Alexander Dahl <ada@thorsis.com>
Cc: linux-rt-users@vger.kernel.org,
	Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
	Thomas Gleixner <tglx@linutronix.de>,
	LKML <linux-kernel@vger.kernel.org>,
	Steven Rostedt <rostedt@goodmis.org>
Subject: Re: [ANNOUNCE] v5.2.9-rt3
Date: Tue, 20 Aug 2019 17:44:18 +0200	[thread overview]
Message-ID: <20190820154418.GM3545@piout.net> (raw)
In-Reply-To: <2182739.9IRgZpf3R8@ada>

Hi,

On 19/08/2019 13:03:51+0200, Alexander Dahl wrote:
> Hei hei,
> 
> just tried to compile this v5.2.9-rt3 for SAMA5D27-SOM1-EK1 based on 
> arch/arm/configs/sama5_defconfig and with running oldconfig and selecting 
> defaults, but that fails if CONFIG_ATMEL_TCB_CLKSRC_USE_SLOW_CLOCK is not set. 
> 
> I think this is due to changes for Atmel TCLIB in v5.2 and the not yet adapted 
> RT patch "clocksource: TCLIB: Allow higher clock rates for clock events", 
> right?

Patch clocksource-tclib-allow-higher-clockrates.patch needs to be
changed so:

ret = setup_clkevents(tc, best_divisor_idx);

becomes

ret = setup_clkevents(&tc, best_divisor_idx);

Also, I would think clocksource-tclib-add-proper-depend.patch could be
dropped. Instead, setup_clkevents should use atmel_tcb_divisors. It
would then be necessary to move its declaration before the function.

Sebastian, can you take care of that or do you expect a patch? In the
latter case, do you want a patch for the patch?

> 
> What's the recommended setting of this option for RT?
> 

Using the slow clock, will make the platform wakeup less frequently,
having a higher clock rate will give a better clockevent resolution.

-- 
Alexandre Belloni, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com