linux-kernel - Re: Subject: [PATCH v2] ASoC: soc-pcm: fix BE dai not hw

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]

Date:   Tue, 2 Jun 2020 11:19:56 -0500
From:   Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
To:     朱灿灿 <zhucancan@...o.com>, lgirdwood@...il.com,
        broonie@...nel.org, perex@...ex.cz, tiwai@...e.com,
        alsa-devel@...a-project.org, linux-kernel@...r.kernel.org
Cc:     kernel <kernel@...o.com>,
        王文虎 <wenhu.wang@...o.com>,
        trivial@...nel.org
Subject: Re: Subject: [PATCH v2] ASoC: soc-pcm: fix BE dai not hw_free and
 shutdown during mixer update



On 5/29/20 5:12 AM, 朱灿灿 wrote:
> FE state is SND_SOC_DPCM_STATE_PREPARE now, BE1 is
> used by FE.
> 
> Later when new BE2 is added to FE by mixer update,
> it will call dpcm_run_update_startup() to update
> BE2's state, but unfortunately BE2 .prepare() meets
> error, it will disconnect all non started BE.
> 
> This make BE1 dai skip .hw_free() and .shutdown(),
> and the BE1 users will never decrease to zero.
> 
> Signed-off-by: zhucancan <zhucancan@...o.com>
> ---
> re-format patch content v2
> ---
>   sound/soc/soc-pcm.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/sound/soc/soc-pcm.c b/sound/soc/soc-pcm.c
> index 1f302de44052..df34422bd0dd 100644
> --- a/sound/soc/soc-pcm.c
> +++ b/sound/soc/soc-pcm.c
> @@ -2730,12 +2730,12 @@ static int dpcm_run_update_startup(struct snd_soc_pcm_runtime *fe, int stream)
>   close:
>   	dpcm_be_dai_shutdown(fe, stream);
>   disconnect:
> -	/* disconnect any non started BEs */
> +	/* disconnect any closed BEs */
>   	spin_lock_irqsave(&fe->card->dpcm_lock, flags);
>   	for_each_dpcm_be(fe, stream, dpcm) {
>   		struct snd_soc_pcm_runtime *be = dpcm->be;
> -		if (be->dpcm[stream].state != SND_SOC_DPCM_STATE_START)
> -				dpcm->state = SND_SOC_DPCM_LINK_STATE_FREE;
> +		if (be->dpcm[stream].state == SND_SOC_DPCM_STATE_CLOSE)
> +			dpcm->state = SND_SOC_DPCM_LINK_STATE_FREE;
>   	}
>   	spin_unlock_irqrestore(&fe->card->dpcm_lock, flags);

This change is quite hard to review, this error handling can be called 
from multiple places.

I *think* it's correct because in all cases where the 
disconnect/close/hw_free labels are reached, the non-shared BEs either 
remain or are put in the DPCM_STATE_CLOSE state before doing this test.

Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>

It really wouldn't hurt though if others double-checked this change...