Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [PATCH 7/7] clocksource/arm_arch_timer: Use
 arch_timer_read_counter to access stable counters
To:     Marc Zyngier <marc.zyngier@arm.com>,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org
Cc:     Russell King <linux@arm.linux.org.uk>,
        Will Deacon <will.deacon@arm.com>,
        Catalin Marinas <catalin.marinas@arm.com>,
        Mark Rutland <mark.rutland@arm.com>,
        Wim Van Sebroeck <wim@linux-watchdog.org>,
        Guenter Roeck <linux@roeck-us.net>,
        Valentin Schneider <valentin.schneider@arm.com>
References: <20190408154907.223536-1-marc.zyngier@arm.com>
 <20190408154907.223536-8-marc.zyngier@arm.com>
 <2a60a031-1eab-2d5e-89ff-b5d516545eeb@linaro.org>
 <bbe9b8c1-132f-bbfa-e3d0-ad10c4165ad7@arm.com>
From:   Daniel Lezcano <daniel.lezcano@linaro.org>
Message-ID: <2f3417f6-95e5-f27a-693d-5aa460fb152d@linaro.org>
Date:   Fri, 3 May 2019 22:31:42 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101
 Thunderbird/60.6.1
MIME-Version: 1.0
In-Reply-To: <bbe9b8c1-132f-bbfa-e3d0-ad10c4165ad7@arm.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


Hi Marc,

On 30/04/2019 17:27, Marc Zyngier wrote:
> On 15/04/2019 13:16, Daniel Lezcano wrote:
>> On 08/04/2019 17:49, Marc Zyngier wrote:
>>> Instead of always going via arch_counter_get_cntvct_stable to
>>> access the counter workaround, let's have arch_timer_read_counter
>>> to point to the right method.
>>>
>>> For that, we need to track whether any CPU in the system has a
>>> workaround for the counter. This is done by having an atomic
>>> variable tracking this.
>>>
>>> Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
>>> ---
>>
>> [ ... ]
>>
>>> +
>>>  /*
>>>   * Default to cp15 based access because arm64 uses this function for
>>>   * sched_clock() before DT is probed and the cp15 method is guaranteed
>>> @@ -372,6 +392,7 @@ static u32 notrace sun50i_a64_read_cntv_tval_el0(void)
>>>  DEFINE_PER_CPU(const struct arch_timer_erratum_workaround *, timer_unstable_counter_workaround);
>>>  EXPORT_SYMBOL_GPL(timer_unstable_counter_workaround);
>>>  
>>> +static atomic_t timer_unstable_counter_workaround_in_use = ATOMIC_INIT(0);
>>
>> Wouldn't make sense to READ_ONCE / WRITE_ONCE instead of using an atomic?
> 
> I don't think *_ONCE says anything about the atomicity of the access. It
> only instruct the compiler that this should only be accessed once, and
> not reloaded/rewritten. In this case, WRITE_ONCE() would work just as
> well, but I feel that setting the expectations do matter.
> 
> I also had a vague idea to use this as a refcount to drop the
> workarounds as CPUs get hotplugged off, in which case we'd need the RMW
> operations to be atomic.
> 
> Anyway, I'm not hell bent on this. If you fundamentally disagree with me
> I'll change it.

As you are planning to extend its usage for refcounting in the hotplug
path, I think it is fine.

Thanks

  -- Daniel


-- 
 <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog