Message-ID: <305eccfb-4202-4694-9dd3-2769429b915c@maciej.szmigiero.name>
Date: Mon, 17 Feb 2025 14:18:57 +0100
From: "Maciej S. Szmigiero" <mail@...iej.szmigiero.name>
To: Takashi Iwai <tiwai@...e.de>
Cc: Jaroslav Kysela <perex@...ex.cz>, Takashi Iwai <tiwai@...e.com>,
 Kailang Yang <kailang@...ltek.com>, Oder Chiou <oder_chiou@...ltek.com>,
 Shuming Fan <shumingf@...ltek.com>, Qiu Wenbo <qiuwenbo@...inos.com.cn>,
 linux-sound@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] ALSA: hda/realtek: Enable PC beep passthrough for HP
 EliteBook 855 G7

On 17.02.2025 11:52, Takashi Iwai wrote:
> On Mon, 17 Feb 2025 11:17:31 +0100,
> Maciej S. Szmigiero wrote:
>>
>> On 17.02.2025 11:02, Takashi Iwai wrote:
>>> On Sun, 16 Feb 2025 22:31:03 +0100,
>>> Maciej S. Szmigiero wrote:
>>>>
>>>> PC speaker works well on this platform in BIOS and in Linux until sound
>>>> card drivers are loaded. Then it stops working.
>>>>
>>>> There seems to be a beep generator node at 0x1a in this CODEC
>>>> (ALC269_TYPE_ALC215) but it seems to be only connected to capture mixers
>>>> at nodes 0x22 and 0x23.
>>>> If I unmute the mixer input for 0x1a at node 0x23 and start recording
>>>> from its "ALC285 Analog" capture device I can clearly hear beeps in that
>>>> recording.
>>>>
>>>> So the beep generator is indeed working properly, however I wasn't able to
>>>> figure out any way to connect it to speakers.
>>>>
>>>> However, the bits in the "Passthrough Control" register (0x36) seems to
>>>> work at least partially: by zeroing "B" and "h" and setting "S" I can at
>>>> least make the PIT PC speaker output appear either in this laptop speakers
>>>> or headphones (depending on whether they are connected or not).
>>>>
>>>>
>>>> There are some caveats, however:
>>>> * If the CODEC gets runtime-suspended the beeps stop so it needs HDA beep
>>>> device for keeping it awake during beeping.
>>>>
>>>> * If the beep generator node is generating any beep the PC beep passthrough
>>>> seems to be temporarily inhibited, so the HDA beep device has to be
>>>> prevented from using the actual beep generator node - but the beep device
>>>> is still necessary due to the previous point.
>>>>
>>>> * In contrast with other platforms here beep amplification has to be
>>>> disabled otherwise the beeps output are WAY louder than they were on pure
>>>> BIOS setup.
>>>>
>>>>
>>>> Unless someone (from Realtek probably) knows how to make the beep generator
>>>> node output appear in speakers / headphones using PC beep passthrough seems
>>>> to be the only way to make PC speaker beeping actually work on this
>>>> platform.
>>>>
>>>> Signed-off-by: Maciej S. Szmigiero <mail@...iej.szmigiero.name>
>>>> ---
>>>>
>>>> Since more than 6 weeks has now passed since the previous version of this
>>>> patch was posted, yet no better or other solution was provided I'm
>>>> re-submitting an updated version of the original patch.
>>>>       This solution has been working fine for me on this platform
>>>> all that time,
>>>> without any obvious issues.
>>>>       Changes from v1:
>>>> * Correct the typo in !IS_ENABLED(CONFIG_INPUT_PCSPKR) code that the
>>>> kernel test robot found.
>>>>       * Change codec_warn() into dev_warn_once(hda_codec_dev(codec))
>>>> to avoid spamming the kernel log at runtime PM CODEC re-init.
>>>
>>> This is really a thing to be checked by Realtek people at first, as
>>> it's very vendor-specific thing.
>>>
>>> Kailang, please check this.
>>
>> Realtek people has been asked to comment/provide alternative solution
>> 3 times in last 6 weeks:
>> On the original v1 submission, by your message from Jan 12, by my
>> message on Feb 2.
>>
>> Looking at https://lore.kernel.org/linux-sound/?q=f%3Arealtek
>> it seems Kailang sent two e-mails about unrelated cases to linux-sound
>> during that time.
>>
>> To be honest, I don't understand why Realtek people don't comment
>> on this since I would think that's a rather simple matter without any
>> truly confidential aspects but on the other hand this fact should *not*
>> permanently block fixing PC beep on this platform.
>>
>> So I think there should be some deadline here for the vendor response -
>> like 1 month more or so?
> 
> Sorry, I don't like that kind of attitude.
> 
> If the patch were perfectly fine, I'd have already taken.  But there
> is a thing that can't be validated without the confirmation from the
> vendor, and that's not what we can accept because it's supposed to
> work on your machine -- that's only one special use case, and it
> doesn't qualify to prove the safety.

I don't want to pressure you as the maintainer to take this patch,
sorry if it sounded that way.

Just wanted to make sure that if the vendor does not respond within
a reasonable time (I think 2.5 months total would be such time) then
there's some alternate way.

But let's wait then.

> In general, Kailang (and Realtek) has been helpful over decades for
> HD-audio stuff development, but they might be busy sometimes.  Let's
> keep asking until catching their attention, at first.

Nice, I hope Realtek will finally provide a good answer at some point.

Also, that's just PC beep passthrough, not CPU voltage adjustment :)
So the risk is rather limited here.

>>> And, except for that, I'm still concerned by the behavior change.
>>
>> AFAIK most sound card drivers in ALSA were developed without any docs,
>> and the register that's being changed is unofficially documented in ALSA:
>> https://docs.kernel.org/sound/hd-audio/realtek-pc-beep.html
>>
>> Also, the behavior change is clearly limited to this single laptop
>> platform (HP EliteBook 855 G7) so the "blast radius" is very limited.
> 
> If you were the only user of this laptop in the world, I wouldn't be
> concerned, sure :)  But certainly that's not the case.

I don't think my particular laptop is a special specimen - it's running
totally stock HP firmware, and AFAIK nothing has been changed there.

>>> Also the caveat you mentioned about the runtime PM raises some doubt,
>>> too.
>>
>> I think it's simply because the CODEC get re-initialized when it comes
>> out of runtime PM sleep so if we print a message there then it would
>> be printed each time the CODEC resumed from runtime PM sleep.
>>
>> That's why I changed to print this hint about CONFIG_INPUT_PCSPKR
>> just once per CODEC device.
> 
> Well, but this runtime PM has to be turned off manually, otherwise the
> beep gets broken, right?  This is already some trade-off, so it's not
> super trivial to apply the suggested change blindly.

No, runtime PM does *not* have to be turned off manually - in fact,
I'm running with CONFIG_SND_HDA_POWER_SAVE_DEFAULT=10 and when testing
that patch I could see the CODEC being runtime suspended and resumed
automatically.

But for the beep passthrough or speakers to be working the CODEC has to
be awake during that beep duration.
That's already the case for the existing hda_beep.c code - that's
why its runtime PM wakeup code was reused in this patch.

> I thought that the input beep infrastructure may work with multiple
> input beep devices, and it usually triggers the all beep devices?

That's true and in fact *necessary* for this patch to work since
it needs *both* the temporary runtime PM wakeup provided by hda_beep.c AND
the actual beep provided by PIT (or CONFIG_INPUT_PCSPKR).

> If so, you can still keep the HD-audio beep (even though it doesn't do
> actually output) so that it can manage the runtime PM of HD-audio
> device at beeping, too.

The beep widget is *not* just not generating beep, it seems to also
inhibit the passthrough path, that's why I wrote in the patch message that:
> If the beep generator node is generating any beep the PC beep passthrough
> seems to be temporarily inhibited, so the HDA beep device has to be
> prevented from using the actual beep generator node - but the beep device
> is still necessary due to the previous point.

> 
> thanks,
> 
> Takashi

Thanks,
Maciej

Powered by blists - more mailing lists