[alsa-devel] [PATCH 1/4] ALSA: control: return payload length for TLV operation

[Takashi Iwai]

On Fri, 02 Sep 2016 13:18:04 +0200,
Takashi Sakamoto wrote:
> 
> Clemens,
> 
> On Aug 31 2016 20:54, Clemens Ladisch wrote:
> > Takashi Iwai wrote:
> >> TLV has never been and (will never be) an API to handle a
> >> generic binary stream.
> > 
> > It would be possible to define something like SNDRV_CTL_TLVT_HWDEP_BLOB
> > or _COEFFICIENTS, or to reserve a range of TLVT values for driver-
> > defined types.  (But we cannot change soc-wm-adsp to support only proper
> > TLV data, because this would introduce a regression.)
> 
> For our information, I wrote a patch including your idea.
> 
> 
> Thanks for your bright comment ;)
> 
> Takashi Sakamoto
> 
> ---- 8< ----
> 
> >From d1a0eabf1b33fae5de7cedf7aee23778f5dcf46c Mon Sep 17 00:00:00 2001
> From: Takashi Sakamoto <o-takashi at sakamocchi.jp>
> Date: Fri, 2 Sep 2016 00:12:20 +0900
> Subject: [PATCH] ALSA: control: coefficient support
> 
> ---
>  include/uapi/sound/asound.h |  5 +++++
>  include/uapi/sound/tlv.h    |  1 +
>  sound/soc/codecs/wm_adsp.c  |  3 ++-
>  sound/soc/soc-ops.c         | 37 ++++++++++++++++++++++++++++++++++---
>  sound/soc/soc-topology.c    |  3 ++-
>  5 files changed, 44 insertions(+), 5 deletions(-)
> 
> diff --git a/include/uapi/sound/asound.h b/include/uapi/sound/asound.h
> index 609cadb..69c0585 100644
> --- a/include/uapi/sound/asound.h
> +++ b/include/uapi/sound/asound.h
> @@ -848,6 +848,11 @@ typedef int __bitwise snd_ctl_elem_iface_t;
>  #define SNDRV_CTL_ELEM_ACCESS_TLV_WRITE		(1<<5)	/* TLV write is possible */
>  #define SNDRV_CTL_ELEM_ACCESS_TLV_READWRITE
> (SNDRV_CTL_ELEM_ACCESS_TLV_READ|SNDRV_CTL_ELEM_ACCESS_TLV_WRITE)
>  #define SNDRV_CTL_ELEM_ACCESS_TLV_COMMAND	(1<<6)	/* TLV command is
> possible */
> +/*
> + * Arbitrary data is accepted for coefficients, instead of pure
> threshold level
> + * information.
> + */
> +#define SNDRV_CTL_ELEM_ACCESS_TLV_COEFF		(1<<7)
>  #define SNDRV_CTL_ELEM_ACCESS_INACTIVE		(1<<8)	/* control does actually
> nothing, but may be updated */
>  #define SNDRV_CTL_ELEM_ACCESS_LOCK		(1<<9)	/* write lock */
>  #define SNDRV_CTL_ELEM_ACCESS_OWNER		(1<<10)	/* write lock owner */

The introduction of a new flag is a good idea (although it's better to
be named not specifically to coef).  Then we can skip such elements to
be accessed from alsactl or amixer.

However...

> @@ -773,19 +774,49 @@ int snd_soc_bytes_tlv_callback(struct snd_kcontrol
> *kcontrol, int op_flag,
>  				unsigned int size, unsigned int __user *tlv)
>  {
>  	struct soc_bytes_ext *params = (void *)kcontrol->private_value;
> -	unsigned int count = size < params->max ? size : params->max;
> +	unsigned int count;
> +	unsigned int type;
>  	int ret = -ENXIO;
> 
> +	/*
> +	 * The TLV packet can transfer numerical ID for one control element.
> +	 * But ALSA control core don't tell it to each implementation of
> +	 * TLV callback. Here, instead, use the first volatile data to
> +	 * check access information.
> +	 */
> +	if (!(kcontrol->vd[0].access & SNDRV_CTL_ELEM_ACCESS_TLV_COEFF))
> +		return -ENXIO;
> +
> +	/* The data should be constructed according to TLV protocol. */
> +	if (copy_from_user(&type, tlv, sizeof(unsigned int)))
> +		return -EFAULT;
> +
> +	/* The type should be an arbitrary data. */
> +	if (type != SNDRV_CTL_TLVT_COEFF)
> +		return -ENXIO;

This breaks the already existing application and is a regression, as
Clemens mentioned.  This should have been forced from the beginning,
but it's too late to put now.  So, it's likely no-go.

But, the addition of the flag is helpful alone.  Could you concentrate
on that and resubmit?


thanks,

Takashi