linux-kernel - Re: [PATCH 1/2] ALSA: cs35l41: Add shared boost feature

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

Date:   Tue, 7 Feb 2023 11:48:55 +0000
From:   Charles Keepax <ckeepax@...nsource.cirrus.com>
To:     Lucas Tanure <lucas.tanure@...labora.com>
CC:     David Rhodes <david.rhodes@...rus.com>,
        Liam Girdwood <lgirdwood@...il.com>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Mark Brown <broonie@...nel.org>,
        "Rob Herring" <robh+dt@...nel.org>,
        Jaroslav Kysela <perex@...ex.cz>,
        Takashi Iwai <tiwai@...e.com>, <alsa-devel@...a-project.org>,
        <devicetree@...r.kernel.org>, <patches@...nsource.cirrus.com>,
        <linux-kernel@...r.kernel.org>, <kernel@...labora.com>
Subject: Re: [PATCH 1/2] ALSA: cs35l41: Add shared boost feature

On Tue, Feb 07, 2023 at 10:40:20AM +0000, Lucas Tanure wrote:
> Shared boost allows two amplifiers to share a single boost
> circuit by communicating on the MDSYNC bus.
> The passive amplifier does not control the boost and receives
> data from the active amplifier.
> 
> Shared Boost is not supported in HDA Systems.
> 

Probably would be nice to put at least a note to say based on
David's patches.

> +static const struct reg_sequence cs35l41_shd_boost_seq[] = {
> +	{CS35L41_PWR_CTRL3,	0x01000110},

This will blat whatever the user set in the DRE switch.
Technically blats the CLASS H enable from the DAPM widget too,
but as that always turns on should be a no-op. Probably should
either not register the DRE switch or have setting it return an
error for these boost modes.

> +int cs35l41_global_enable(struct regmap *regmap, enum cs35l41_boost_type b_type, int enable,
> +			  struct completion *pll_lock)
>  {
>  	int ret;
> +	unsigned int gpio1;
>  
>  	switch (b_type) {
> +	case CS35L41_SHD_BOOST_ACTV:
> +	case CS35L41_SHD_BOOST_PASS:
> +		regmap_update_bits(regmap, CS35L41_PWR_CTRL3, CS35L41_SYNC_EN_MASK, 0);
> +
> +		gpio1 = enable ? CS35L41_GPIO1_MDSYNC : CS35L41_GPIO1_HIZ;
> +		regmap_update_bits(regmap, CS35L41_GPIO_PAD_CONTROL, CS35L41_GPIO1_CTRL_MASK,
> +				   gpio1 << CS35L41_GPIO1_CTRL_SHIFT);
> +
> +		ret = regmap_update_bits(regmap, CS35L41_PWR_CTRL1, CS35L41_GLOBAL_EN_MASK,
> +					 enable << CS35L41_GLOBAL_EN_SHIFT);
> +		usleep_range(3000, 3100);
> +		if (!enable)
> +			break;
> +
> +		if (!pll_lock)
> +			return -EINVAL;
> +
> +		ret = wait_for_completion_timeout(pll_lock, msecs_to_jiffies(1000));
> +		if (ret == 0)
> +			ret = -ETIMEDOUT;

This feels kinda scary, in that you are relying on a 1 to 1
correspondence between this code running and getting a PLL lock
signal. The datasheet is helpfully completely vague on when PLL
locks are triggered.

The PLL enable seems to be set through set_sysclk, which could
be called multiple times, per DAPM power up.  Does the PLL
lock only go once global enable has been set? Can't help
but wonder if a reinit_completion should probably go somewhere
to ensure we are getting this lock of the PLL not a past one.

> @@ -483,6 +483,11 @@ static irqreturn_t cs35l41_irq(int irq, void *data)
>  		ret = IRQ_HANDLED;
>  	}
>  
> +	if (status[2] & CS35L41_PLL_LOCK) {
> +		regmap_write(cs35l41->regmap, CS35L41_IRQ1_STATUS3, CS35L41_PLL_LOCK);
> +		complete(&cs35l41->pll_lock);
> +	}
> +

If you fall into any of the error cases in this IRQ handler above
this, it will blat values you don't want into BST_EN although, to
be fair that does look currently broken for external boost as
well.

Thanks,
Charles