linux-kernel - Re: [PATCH v2 01/10] ASoC: tegra: Fix kcontrol put callback in ADMAIF

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] [thread-next>] [day] [month] [year] [list]
Message-ID: <d27bf513-6f16-5ad6-59cb-79fad5cc951c@nvidia.com>
Date:   Mon, 8 Nov 2021 21:33:25 +0530
From:   Sameer Pujar <spujar@...dia.com>
To:     Jaroslav Kysela <perex@...ex.cz>, Takashi Iwai <tiwai@...e.de>
Cc:     broonie@...nel.org, lgirdwood@...il.com, tiwai@...e.com,
        thierry.reding@...il.com, jonathanh@...dia.com,
        alsa-devel@...a-project.org, linux-tegra@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 01/10] ASoC: tegra: Fix kcontrol put callback in ADMAIF



On 11/3/2021 10:55 PM, Jaroslav Kysela wrote:
> On 03. 11. 21 15:16, Takashi Iwai wrote:
>> On Wed, 03 Nov 2021 14:52:17 +0100,
>> Sameer Pujar wrote:
>>>
>>> The kcontrol put callback is expected to return 1 when there is change
>>> in HW or when the update is acknowledged by driver. This would ensure
>>> that change notifications are sent to subscribed applications. Update
>>> the ADMAIF driver accordingly
>>>
>>> Fixes: f74028e159bb ("ASoC: tegra: Add Tegra210 based ADMAIF driver")
>>> Suggested-by: Jaroslav Kysela <perex@...ex.cz>
>>> Suggested-by: Mark Brown <broonie@...nel.org>
>>> Signed-off-by: Sameer Pujar <spujar@...dia.com>
>>> ---
>>>   sound/soc/tegra/tegra210_admaif.c | 23 ++++++++++++++++++-----
>>>   1 file changed, 18 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/sound/soc/tegra/tegra210_admaif.c 
>>> b/sound/soc/tegra/tegra210_admaif.c
>>> index bcccdf3..dc71075 100644
>>> --- a/sound/soc/tegra/tegra210_admaif.c
>>> +++ b/sound/soc/tegra/tegra210_admaif.c
>>> @@ -452,16 +452,29 @@ static int tegra_admaif_put_control(struct 
>>> snd_kcontrol *kcontrol,
>>>      struct tegra_admaif *admaif = 
>>> snd_soc_component_get_drvdata(cmpnt);
>>>      int value = ucontrol->value.integer.value[0];
>>>
>>> -    if (strstr(kcontrol->id.name, "Playback Mono To Stereo"))
>>> +    if (strstr(kcontrol->id.name, "Playback Mono To Stereo")) {
>>> +            if (admaif->mono_to_stereo[ADMAIF_TX_PATH][ec->reg] == 
>>> value)
>>> +                    return 0;
>>> +
>>> admaif->mono_to_stereo[ADMAIF_TX_PATH][ec->reg] = value;
>>> -    else if (strstr(kcontrol->id.name, "Capture Mono To Stereo"))
>>> +    } else if (strstr(kcontrol->id.name, "Capture Mono To Stereo")) {
>>> +            if (admaif->mono_to_stereo[ADMAIF_RX_PATH][ec->reg] == 
>>> value)
>>> +                    return 0;
>>> +
>>> admaif->mono_to_stereo[ADMAIF_RX_PATH][ec->reg] = value;
>>> -    else if (strstr(kcontrol->id.name, "Playback Stereo To Mono"))
>>> +    } else if (strstr(kcontrol->id.name, "Playback Stereo To Mono")) {
>>> +            if (admaif->stereo_to_mono[ADMAIF_TX_PATH][ec->reg] == 
>>> value)
>>> +                    return 0;
>>> +
>>> admaif->stereo_to_mono[ADMAIF_TX_PATH][ec->reg] = value;
>>> -    else if (strstr(kcontrol->id.name, "Capture Stereo To Mono"))
>>> +    } else if (strstr(kcontrol->id.name, "Capture Stereo To Mono")) {
>>> +            if (admaif->stereo_to_mono[ADMAIF_RX_PATH][ec->reg] == 
>>> value)
>>> +                    return 0;
>>> +
>>> admaif->stereo_to_mono[ADMAIF_RX_PATH][ec->reg] = value;
>>> +    }
>>>
>>> -    return 0;
>>> +    return 1;
>>
>> Hrm, that looks too redundant.  The similar checks are seen in the get
>> part, so we may have a better helper function to reduce the string
>> checks, something like below.
>

Thanks Takashi for your inputs. This would make the get/put callbacks 
simpler. But in some cases, for few controls additional handling is 
required (tegra210_i2s.c driver for example). In such cases additional 
checks would be required if the callback is common.

> While proposing such cleanups, I would create separate get/put 
> callbacks for
> all four ops instead using strstr(). The callbacks may put the common 
> code to
> one function. It may reduce the code size (and the text segment size).

With separate callbacks, the string checks can be removed. However for 
most of the controls, the common part is minimal. So there would be 
multiple independent small functions depending on the number of controls 
and the local variables are duplicated that many times. Would there be 
any concern on the space these local variables take? One pair of 
callbacks for a control may look like this.

static int kctl_pget_mono_to_stereo(struct snd_kcontrol *kcontrol,
                                     struct snd_ctl_elem_value *ucontrol)
{
         struct snd_soc_component *cmpnt = 
snd_soc_kcontrol_component(kcontrol);
         struct soc_enum *ec = (struct soc_enum *)kcontrol->private_value;
         struct tegra_admaif *admaif = snd_soc_component_get_drvdata(cmpnt);

         ucontrol->value.integer.value[0] =
admaif->mono_to_stereo[ADMAIF_TX_PATH][ec->reg];

         return 0;
}

static int kctl_pput_mono_to_stereo(struct snd_kcontrol *kcontrol,
                                     struct snd_ctl_elem_value *ucontrol)
{
         struct snd_soc_component *cmpnt = 
snd_soc_kcontrol_component(kcontrol);
         struct soc_enum *ec = (struct soc_enum *)kcontrol->private_value;
         struct tegra_admaif *admaif = snd_soc_component_get_drvdata(cmpnt);
         int value = ucontrol->value.integer.value[0];

         if (value == admaif->mono_to_stereo[ADMAIF_TX_PATH][ec->reg])
                 return 0;

         admaif->mono_to_stereo[ADMAIF_TX_PATH][ec->reg] = value;

         return 1;
}


Looks like having separate callbacks make it look more cleaner. If this 
appears fine, I can send next revision.