linux-kernel - RE: [PATCHv1 5/5] ASoC: da7213: add default clock handling

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-prev] [day] [month] [year] [list]
Message-ID: <AM5PR1001MB0994D91E64529B8B3C13977F80760@AM5PR1001MB0994.EURPRD10.PROD.OUTLOOK.COM>
Date:   Wed, 13 Nov 2019 11:25:33 +0000
From:   Adam Thomson <Adam.Thomson.Opensource@...semi.com>
To:     Sebastian Reichel <sebastian.reichel@...labora.com>,
        Adam Thomson <Adam.Thomson.Opensource@...semi.com>
CC:     Support Opensource <Support.Opensource@...semi.com>,
        Liam Girdwood <lgirdwood@...il.com>,
        Mark Brown <broonie@...nel.org>,
        Jaroslav Kysela <perex@...ex.cz>,
        Takashi Iwai <tiwai@...e.com>,
        "alsa-devel@...a-project.org" <alsa-devel@...a-project.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "kernel@...labora.com" <kernel@...labora.com>
Subject: RE: [PATCHv1 5/5] ASoC: da7213: add default clock handling

On 12 November 2019 16:30, Sebastian Reichel wrote:

> > > @@ -1188,38 +1190,54 @@ static int da7213_hw_params(struct
> > > snd_pcm_substream *substream,
> > >  	switch (params_rate(params)) {
> > >  	case 8000:
> > >  		fs = DA7213_SR_8000;
> > > +		freq = DA7213_PLL_FREQ_OUT_98304000;
> > >  		break;
> > >  	case 11025:
> > >  		fs = DA7213_SR_11025;
> > > +		freq = DA7213_PLL_FREQ_OUT_90316800;
> > >  		break;
> > >  	case 12000:
> > >  		fs = DA7213_SR_12000;
> > > +		freq = DA7213_PLL_FREQ_OUT_98304000;
> > >  		break;
> > >  	case 16000:
> > >  		fs = DA7213_SR_16000;
> > > +		freq = DA7213_PLL_FREQ_OUT_98304000;
> > >  		break;
> > >  	case 22050:
> > >  		fs = DA7213_SR_22050;
> > > +		freq = DA7213_PLL_FREQ_OUT_90316800;
> > >  		break;
> > >  	case 32000:
> > >  		fs = DA7213_SR_32000;
> > > +		freq = DA7213_PLL_FREQ_OUT_98304000;
> > >  		break;
> > >  	case 44100:
> > >  		fs = DA7213_SR_44100;
> > > +		freq = DA7213_PLL_FREQ_OUT_90316800;
> > >  		break;
> > >  	case 48000:
> > >  		fs = DA7213_SR_48000;
> > > +		freq = DA7213_PLL_FREQ_OUT_98304000;
> > >  		break;
> > >  	case 88200:
> > >  		fs = DA7213_SR_88200;
> > > +		freq = DA7213_PLL_FREQ_OUT_90316800;
> > >  		break;
> > >  	case 96000:
> > >  		fs = DA7213_SR_96000;
> > > +		freq = DA7213_PLL_FREQ_OUT_98304000;
> > >  		break;
> > >  	default:
> > >  		return -EINVAL;
> > >  	}
> > >
> > > +	/* setup PLL */
> > > +	if (da7213->fixed_clk_auto) {
> > > +		snd_soc_component_set_pll(component, 0,
> > > DA7213_SYSCLK_PLL,
> > > +					  da7213->mclk_rate, freq);
> > > +	}
> > > +
> >
> > Are we happy with the PLL being always enabled? Seems like a power drain,
> > especially if you're using an MCLK which is a natural harmonic for the SR in
> > question in which case the PLL can be bypassed. Also the bias level function in
> > this driver will enable and disable the MCLK, if it has been provided, so it's a
> > bit strange to have the PLL enabled but it's clock source taken away.
> 
> So you are suggesting adding something like this to
> da7213_set_bias_level()?
> 
> if (freq % da7213->mclk_rate == 0)
>     source = DA7213_SYSCLK_MCLK
> else
>     source = DA7213_SYSCLK_PLL
> snd_soc_component_set_pll(component, 0, source, da7213->mclk_rate, freq);

Yes it would make more sense to control the PLL there as for MCLK. Also for the
transition back to SND_SOC_BIAS_STANDBY you would want to configure the PLL as
DA7213_SYSCLK_MCLK so it's in bypass mode (i.e. disabled).

Selecting bypass mode for natural harmonic MCLK frequencies (11.2896/12.288,
22.5792/24.576 and 45.1584/49.152 as stated in the datasheet) would be ideal. I
think the check you suggest above though might not be enough as it will pick out
valid MCLK rates of 5.6448/6.144 as allowing PLL bypass but the datasheet
doesn't state those as valid options for bypass.

> > > @@ -1836,6 +1854,20 @@ static int da7213_probe(struct
> snd_soc_component
> > > *component)
> > >  			return PTR_ERR(da7213->mclk);
> > >  		else
> > >  			da7213->mclk = NULL;
> > > +	} else {
> > > +		/* Store clock rate for fixed clocks for automatic PLL setup */
> > > +		ret = clk_prepare_enable(da7213->mclk);
> > > +		if (ret) {
> > > +			dev_err(component->dev, "Failed to enable mclk\n");
> > > +			return ret;
> > > +		}
> >
> > I've not gone through the clk framework code but surprised to see the need to
> > enable a clock to retrieve it's rate.
> 
> /**
>  * clk_get_rate - obtain the current clock rate (in Hz) for a clock source.
>  *                This is only valid once the clock source has been enabled.
>  * @clk: clock source
>  */
> unsigned long clk_get_rate(struct clk *clk);
> 
> Which makes sense for a non-fixed clock, which uses the same API.

Hmmm. Fair enough. Just seems odd to me that you would need to enable a clock
to retrieve it's configured rate. Surely it would have been configured prior to
enabling. Reading into that code, the clk rate value is always taken from cache
so until it's enabled then the cache isn't updated. Actually with the suggestion
to move PLL control to the bias_level() function, you could just get the clk
rate there instead.

> 
> > > +		da7213->mclk_rate = clk_get_rate(da7213->mclk);
> > > +
> > > +		clk_disable_unprepare(da7213->mclk);
> > > +
> > > +		/* assume fixed clock until set_sysclk() is being called */
> > > +		da7213->fixed_clk_auto = true;
> >
> > I don't see any code where 'fixed_clk_auto' is set to false so it seems that
> > previous operational usage is being broken here.
> 
> oops.

:)