Re: [PATCH] Input: pwm-beeper - Support volume setting via sysfs

From: Marek Vasut <marex@denx.de>
To: John Watts <contact@jookia.org>,
	Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: "Jeff LaBundy" <jeff@labundy.com>, "Takashi Iwai" <tiwai@suse.de>,
	linux-input@vger.kernel.org,
	"Uwe Kleine-König" <u.kleine-koenig@pengutronix.de>,
	"Frieder Schrempf" <frieder.schrempf@kontron.de>,
	"Manuel Traut" <manuel.traut@mt.com>,
	"Thierry Reding" <thierry.reding@gmail.com>,
	linux-pwm@vger.kernel.org, alsa-devel@alsa-project.org,
	"Jaroslav Kysela" <perex@perex.cz>,
	"Takashi Iwai" <tiwai@suse.com>
Subject: Re: [PATCH] Input: pwm-beeper - Support volume setting via sysfs
Date: Mon, 14 Aug 2023 04:26:29 +0200	[thread overview]
Message-ID: <545844bb-c16e-4675-0f13-d76646d866fd@denx.de> (raw)
In-Reply-To: <ZNYsGr6yBeVTtNMK@titan>

On 8/11/23 14:39, John Watts wrote:
> On Tue, Aug 01, 2023 at 12:28:29AM -0700, Dmitry Torokhov wrote:
>> If we want to extend the API we will need to define exactly how it will
>> all work. I.e. what happens if userspace mixes the old SND_TONE and
>> SND_BELL with the new SND_BELL_VOL or whatever. Does it play with
>> previously set volume? The default one? How to set the default one? How
>> to figure out what the current volume is if we decide to make volume
>> "sticky"?
>>
>> As far as userspace I expect it is more common to have one program (or
>> component of a program) to set volume and then something else requests
>> sound, so having one-shot API is of dubious value to me.
>>
>> I hope we can go with Takashi's proposal downthread, but if not I wonder
>> if the sysfs approach is not the simplest one. Do we expect more beepers
>> that can control volume besides pwm-beeper?
>>
>> Thanks.
>>
>> -- 
>> Dmitry
> 
> (Just to duck in as someone that has written a little program to play beeps and
> tones using the EV_TONE API)
> 
> It might be worth distinguishing between the goals of having some beeps with
> different volumes compared to all beeps with different volumes.
> 
> Sound card mixers generally control some sort of global volume while I would
> imagine the tone API would control per-tone volume. I don't know too much about
> safety guarantees but writing an input then sysfs or mixer then input again
> seems like it could get jumbled up.
> 
> In that speicfic case I think it would make more sense to send volume and tone
> from whatever beep API is being used, with the volume being a multiplier of the
> loudest volume. This is similar to how audio works with PCM output. Existing
> beeps would have the volume set to 100%.

I agree binding tone frequency and volume together would be better.
The API would be nicer and easier to use in my opinion too.