Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH] memory: samsung: exynos5422-dmc: propagate error from
 exynos5_counters_get()
To:     Marek Szyprowski <m.szyprowski@samsung.com>,
        linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org
Cc:     Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>,
        Krzysztof Kozlowski <krzk@kernel.org>,
        linux-samsung-soc@vger.kernel.org,
        Chanwoo Choi <cw00.choi@samsung.com>
References: <CGME20200804061225eucas1p283c1e0dc404bc420a2184480fdfd2b0d@eucas1p2.samsung.com>
 <20200804061210.5415-1-m.szyprowski@samsung.com>
From:   Lukasz Luba <lukasz.luba@arm.com>
Message-ID: <24675559-b807-5b80-1318-805c1530ffb3@arm.com>
Date:   Tue, 4 Aug 2020 10:11:03 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101
 Thunderbird/60.9.0
MIME-Version: 1.0
In-Reply-To: <20200804061210.5415-1-m.szyprowski@samsung.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Hi Marek,

On 8/4/20 7:12 AM, Marek Szyprowski wrote:
> exynos5_counters_get() might fail with -EPROBE_DEFER if the driver for
> devfreq event counter is not yet probed. Propagate that error value to
> the caller to ensure that the exynos5422-dmc driver will be probed again
> when devfreq event contuner is available.
> 
> This fixes boot hang if both exynos5422-dmc and exynos-ppmu drivers are
> compiled as modules.
> 
> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
> ---
>   drivers/memory/samsung/exynos5422-dmc.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/memory/samsung/exynos5422-dmc.c b/drivers/memory/samsung/exynos5422-dmc.c
> index b9c7956e5031..639811a3eecb 100644
> --- a/drivers/memory/samsung/exynos5422-dmc.c
> +++ b/drivers/memory/samsung/exynos5422-dmc.c
> @@ -914,7 +914,7 @@ static int exynos5_dmc_get_status(struct device *dev,
>   	} else {
>   		ret = exynos5_counters_get(dmc, &load, &total);
>   		if (ret < 0)
> -			return -EINVAL;
> +			return ret;
>   
>   		/* To protect from overflow, divide by 1024 */
>   		stat->busy_time = load >> 10;
> 

Thank you for the patch, LGTM.
Some questions are still there, though. The function
exynos5_performance_counters_init() should capture that the counters
couldn't be enabled or set. So the functions:
exynos5_counters_enable_edev() and exynos5_counters_set_event()
must pass gently because devfreq device is registered.
Then devfreq checks device status, and reaches the state when
counters 'get' function returns that they are not ready...

If that is a kind of 2-stage initialization, maybe we should add
another 'check' in the exynos5_performance_counters_init() and call
the devfreq_event_get_event() to make sure that we are ready to go,
otherwise return ret from that function (which is probably EPROBE_DEFER)
and not register the devfreq device.

Marek do want to submit such patch or I should bake it and submit on top
of this patch?

Could you tell me how I can reproduce this? Do you simply load one
module after another (exynos-ppmu than exynos5422-dmc) or in parallel?

Regards,
Lukasz