linux-kernel - Re: Odroid-XU4 sound issue after suspend-resume

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <24165241-1f65-fafa-0c59-b85cf89bc5bb@samsung.com>
Date:   Wed, 7 Aug 2019 10:22:14 +0200
From:   Sylwester Nawrocki <s.nawrocki@...sung.com>
To:     Jaafar Ali <jaafarkhalaf@...il.com>
Cc:     linux-samsung-soc@...r.kernel.org, linux-kernel@...r.kernel.org,
        kgene@...nel.org, Krzysztof Kozlowski <krzk@...nel.org>,
        sam@...nborg.org, linux-clk@...r.kernel.org
Subject: Re: Odroid-XU4 sound issue after suspend-resume

Hi Jaafar, 

On 8/5/19 15:27, Jaafar Ali wrote:
> Dear All,
> Kernel 5.3-rc1
> OS: ubuntu 18.04
> Hardware: Odroid-XU4
> The sound of Odroid-XU4 after suspend/resume cycle is choppy and slow. 
> I have found a workaround, the I2SMOD register value should be set to 
> zero after resume to force using internal codec clock (cdclkcon bit = 0),
> also the rclk_srcrate which is obtained from the function 
> *clk_get_rate(rclksrc) *inside *hw_params* function is not correct and 
> must be divided by 2 to obtain proper value, i2s_resume function 
> is modified to:
> 
> static int i2s_resume(struct snd_soc_dai *dai)
> {
>         struct samsung_i2s_priv *priv = dev_get_drvdata(dai->dev);
>         priv->suspend_i2smod = 0;//workaround-1 ,
>         return pm_runtime_force_resume(dai->dev);
> 
> }
> 
> inside hw_params function, the rclk_srcrate must be halved to solve 
> unknown problem of clock shift, so before return from hw_params we 
> must insert:
> if(mod == 0){
> 	priv->rclk_srcrate = priv->rclk_srcrate / 2; //workaround-2, 
> }
> 
> With these two workaround sound issue was solved, but I hope we can 
> get concrete fix.
Thank you for the bug report. I spent some time on debugging this and
it turned out that there is a clock mux between EPLL and the audio 
subsystem which looses its configuration during suspend/resume cycle.
So we end up with the I2S controller clocked from the main oscillator
clock (24 MHz) rather than the EPLL (196.608 MHz) after system suspend/
resume. I will post a patch for clk-exynos5420 driver shortly.

-- 
Regards,
Sylwester