From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1752562AbcG2Kbu (ORCPT <rfc822;w@1wt.eu>);
	Fri, 29 Jul 2016 06:31:50 -0400
Received: from foss.arm.com ([217.140.101.70]:53062 "EHLO foss.arm.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1751504AbcG2Kbs (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
	Fri, 29 Jul 2016 06:31:48 -0400
Subject: Re: [PATCH] clocksource: arm_arch_timer: Support reading clock rate
 from a driver
To: Mark Rutland <mark.rutland@arm.com>,
        =?UTF-8?B?UmFmYcWCIE1pxYJlY2tp?= <zajec5@gmail.com>
References: <1469784191-3998-1-git-send-email-zajec5@gmail.com>
 <20160729101811.GB21580@leverpostej>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        =?UTF-8?B?UmFmYcWCIE1pxYJlY2tp?= <rafal@milecki.pl>,
        "open list:CLOCKSOURCE, CLOCKEVENT DRIVERS" 
	<linux-kernel@vger.kernel.org>,
        linux-arm-kernel@lists.infradead.org
From: Marc Zyngier <marc.zyngier@arm.com>
X-Enigmail-Draft-Status: N1110
Organization: ARM Ltd
Message-ID: <579B3090.1070709@arm.com>
Date: Fri, 29 Jul 2016 11:31:44 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101
 Icedove/38.7.0
MIME-Version: 1.0
In-Reply-To: <20160729101811.GB21580@leverpostej>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 29/07/16 11:18, Mark Rutland wrote:
> [adding Marc to Cc]
> 
> On Fri, Jul 29, 2016 at 11:23:11AM +0200, Rafał Miłecki wrote:
>> From: Rafał Miłecki <rafal@milecki.pl>
>>
>> On some devices using arch code for reading clock rate doesn't work. So
>> far the only option was to specify clock-frequency in a DT. This works
>> only if a clock frequency doesn't have to be calculated on runtime.
>>
>> On BCM53573 SoC (with Cortex-A7) there is ILP clock that needs its own
>> driver. With this change we can query such clocks by using a standard:
>> clocks = <&foo>;
> 
> The clock for the architected timer(s) are supposed to be configured
> before entering Linux. The clock-frequency property is at best a dodgy
> workaround, and this is even worse.
> 
> Please fix your firmware to configure and enabled this clock before
> entering Linux, and to program CNTFRQ appropriately.

I'll add that failure to do so breaks other subsystems which do rely on
CNTFRQ to be correctly programmed *on each CPU* (like KVM and Xen).

Please follow the requirements of the architecture and don't just paper
over such a bug.

> As this path stands, NAK.

Seconded.

	M.
-- 
Jazz is not dead. It just smells funny...