linux-kernel - Re: [PATCH] [MT6660] Mediatek Smart Amplifier Driver

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20190904115630.GA4348@sirena.co.uk>
Date:   Wed, 4 Sep 2019 12:56:30 +0100
From:   Mark Brown <broonie@...nel.org>
To:     Richtek Jeff Chang <richtek.jeff.chang@...il.com>
Cc:     lgirdwood@...il.com, perex@...ex.cz, tiwai@...e.com,
        matthias.bgg@...il.com, alsa-devel@...a-project.org,
        linux-arm-kernel@...ts.infradead.org,
        linux-mediatek@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] [MT6660] Mediatek Smart Amplifier Driver

On Wed, Sep 04, 2019 at 03:07:06PM +0800, Richtek Jeff Chang wrote:

> > > +static int32_t mt6660_i2c_update_bits(struct mt6660_chip *chip,
> > > +	uint32_t addr, uint32_t mask, uint32_t data)
> > > +{

> > It would be good to implement a regmap rather than open coding
> > *everything* - it'd give you things like this without needing to open
> > code them.  Providing you don't use the cache code it should cope fine
> > with variable register sizes.

> Due to our hardware design, it is hard to implement regmap for MT6660.

You definitely can't use all the functionality due to the variable
register sizes but using reg_write() and reg_read() should get you most
of it.

> > > +static int mt6660_i2c_init_setting(struct mt6660_chip *chip)
> > > +{
> > > +	int i, len, ret;
> > > +	const struct codec_reg_val *init_table;
> > > +
> > > +	init_table = e4_reg_inits;
> > > +	len = ARRAY_SIZE(e4_reg_inits);
> > > +
> > > +	for (i = 0; i < len; i++) {
> > > +		ret = mt6660_i2c_update_bits(chip, init_table[i].addr,
> > > +				init_table[i].mask, init_table[i].data);
> > > +		if (ret < 0)
> > > +			return ret;

> > Why are we not using the chip defaults here?

> Because MT6660 needs this initial setting for working well.

What are these settings?  Are you sure they are generic settings and
not board specific?

> > > +	if (on_off) {
> > > +		if (chip->pwr_cnt == 0) {
> > > +			ret = mt6660_i2c_update_bits(chip,
> > > +				MT6660_REG_SYSTEM_CTRL, 0x01, 0x00);
> > > +			val = mt6660_i2c_read(chip, MT6660_REG_IRQ_STATUS1);
> > > +			dev_info(chip->dev,
> > > +				"%s reg0x05 = 0x%x\n", __func__, val);
> > > +		}
> > > +		chip->pwr_cnt++;

> > This looks like you're open coding runtime PM stuff?  AFAICT the issue
> > is that you need to write to this register to do any I/O.  Just
> > implement this via the standard runtime PM framework, you'll need to do
> > something about the register I/O in the controls (ideally in the
> > framework, it'd be a lot easier if you did have a cache) but you could
> > cut out this bit.

> In our experience, some Customer platform doesn't support runtime PM.

Tell your customers to turn it on, it's a standard kernel framework and
there's really no excuse for open coding it.  If there's some reason why
runtime PM can't work for them then we should get that fixed but it
really is *very* widely deployed.

Download attachment "signature.asc" of type "application/pgp-signature" (489 bytes)