linux-kernel - Re: [PATCH -next] ASoC: tegra: Fix missing clk_disable

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: <fedff22c-2a25-0def-4980-8ba8ae75757b@gmail.com>
Date:   Thu, 17 Jun 2021 15:46:57 +0300
From:   Dmitry Osipenko <digetx@...il.com>
To:     Zou Wei <zou_wei@...wei.com>, lgirdwood@...il.com,
        broonie@...nel.org, perex@...ex.cz, tiwai@...e.com,
        thierry.reding@...il.com, jonathanh@...dia.com
Cc:     linux-tegra@...r.kernel.org, alsa-devel@...a-project.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH -next] ASoC: tegra: Fix missing clk_disable_unprepare() on
 error path

17.06.2021 14:30, Zou Wei пишет:
> Fix the missing clk_disable_unprepare() before return
> from tegra_machine_hw_params() in the error handling case.
> 
> Reported-by: Hulk Robot <hulkci@...wei.com>
> Signed-off-by: Zou Wei <zou_wei@...wei.com>
> ---
>  sound/soc/tegra/tegra_asoc_machine.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/sound/soc/tegra/tegra_asoc_machine.c b/sound/soc/tegra/tegra_asoc_machine.c
> index a53aec3..397f326 100644
> --- a/sound/soc/tegra/tegra_asoc_machine.c
> +++ b/sound/soc/tegra/tegra_asoc_machine.c
> @@ -306,6 +306,7 @@ static int tegra_machine_hw_params(struct snd_pcm_substream *substream,
>  
>  	err = snd_soc_dai_set_sysclk(codec_dai, clk_id, mclk, SND_SOC_CLOCK_IN);
>  	if (err < 0) {
> +		clk_disable_unprepare(machine->clk_cdev1);

It should be the opposite, the clock must be enabled here and it could
be re-enabled in the other error cases. In practice the chance of those
errors is close to zero, so we don't care about them since we will have
much bigger problems in that case.

>  		dev_err(card->dev, "codec_dai clock not set: %d\n", err);
>  		return err;
>  	}
> @@ -523,8 +524,10 @@ int tegra_asoc_machine_probe(struct platform_device *pdev)
>  	}
>  
>  	err = devm_snd_soc_register_card(dev, card);
> -	if (err)
> +	if (err) {
> +		clk_disable_unprepare(machine->clk_cdev1);

We also don't disable clock after removing driver module. This is
intentional because that clock is enabled to workaround obscure bug on
one of devboards that I can't reproduce. Ideally we should should fix
the bug and remove the clk-enable hack. So I'm not sure whether this is
a worthwhile "fix".