09:20 <DylanVanAssche> srinik: I heard you are the author of https://github.com/alsa-project/alsa-ucm-conf/pull/88. I'm trying this on an SDM845 device where the speaker output is connected to an earpiece, but I get no sound. Any tips to debug this?

09:43 <Mis012[m]> Dylan Van Assche: ideally you would have schematics so you can be absolutely sure there are not three other amplifiers between the speaker output and the earpiece :P

09:44 <DylanVanAssche> Mis012: Well yeah, but we all know that these things are not easy to get or impossible to get 😉

09:45 <Mis012[m]> then 3d xray of the PCB?

09:46 <Mis012[m]> we've seen analog ampliers no less...

09:46 <Mis012[m]> *amplifiers

09:47 <Mis012[m]> I guess scour downstream for suspicious i2c devices

09:50 <Mis012[m]> actually, isn't downstream Windows here :P

09:51 <Mis012[m]> yeah, would probably be better to find the primary sources...

09:52 <Mis012[m]> though I guess Windows+ACPI isn't nowhere near as flexible as device trees

09:53 <Mis012[m]> in the sense of stuff from different vendors working together

09:53 <Mis012[m]> so maybe they don't have a way to add on random amplifiers?

09:54 <DylanVanAssche> AFAIK from somebody who knows a bit about the schematic, the earpiece is connected directly to the codec.

10:02 <Mis012[m]> then I don't see why it wouldn't work

10:03 <Mis012[m]> speaker or earpiece... basically the same thing

10:03 <Mis012[m]> maybe alsamixer decided to put some important thing to 0?

10:25 <DylanVanAssche> Well that's what I'm trying to find out 🙂 I'm just missing the secret sauce to properly debug this

10:25 <DylanVanAssche> `amixer` works, but you can quickly see over a control, there are like hundreds of them

11:35 <srinik> DylanVanAssche: Speakers are SoundWire devices these are Smart Speaker Amplifiers so when you say speaker out is connected to earpiece do you mean the actual WSA881X output connected to a 8/16 ohm earpiece?

11:35 <Mis012[m]> uh huh, WSA881X sounds like board specific choice

11:35 <srinik> DylanVanAssche: or do you know if these are connected to codec Analog output of EAR_OUT_P and OUT_M ?

11:37 <srinik> Mis012[m]: Yes, that is correct, but I want to understand what was the reference to speaker output actually ment

11:39 <Mis012[m]> srinik: well, he has a Windows (*shivers*) device without schematics

11:39 <DylanVanAssche> srinik: Ah I'm not sure about that, let me ask some people 🙂

11:39 <Mis012[m]> so not sure how you would even find out if there is WSA881X

11:39 <srinik> DylanVanAssche: its normal to connect earpiece to codec analog out, but connecting to speaker out sounds unusual

11:40 <DylanVanAssche> Mis012: Sorry missed your message about Windows, it is an Android device with a SDM845 SoC chi

11:40 <Mis012[m]> oh, right, srinik has Windows

11:40 <Mis012[m]> my bad

11:40 <srinik> Mis012[m]: finding out if WSA is connected is easy, if you are able to find the reset line for WSA which is normally same pin recommended pin, then SoundWire should be able to enumerate it.

11:41 <srinik> Mis012[m]: you could see them in /sys/bus/soundwire/devices

11:41 <Mis012[m]> srinik: is there even only one place where it can be connected?

11:41 <srinik> Mis012[m]: you mean WSA or earpiece?

11:41 <Mis012[m]> WSA

11:42 <Mis012[m]> earpiece can be connected to GPCLK if you're adventurous enough xD

11:42 <srinik> Mis012[m]: Yes, WSA has to be connected to SWR_CLK and SWR_DATA line of codec

11:42 <Mis012[m]> is soundwire some new fun replacement for slimbus?

11:43 <srinik> Mis012[m]: SDM845 is very weird.. we use SoundWire over SLIMBus :-)

11:43 <Mis012[m]> srinik: btw, would you happen to know if there is ANY benefit to having the LPASS hexagon be involved?

11:43 <srinik> Mis012[m]: most of the new SoCs are fully moved to SounWire now, except for BT audio

11:43 <Mis012[m]> since the codecs have xtensa

11:43 <Mis012[m]> *are xtensa I guess

11:43 <Mis012[m]> sounds like one too many DSPs

11:44 <aka_[m]> Is there any sheet of which family uses sound wire and which is legacy?

11:44 <aka_[m]> Like since idk sd835

11:45 <srinik> Mis012[m]: there are many advantages of using Hexagon... like offloading, codecs, and pre and post process algos like speaker protection, Noise cancellation, Hot word detection.. and many more

11:45 <Mis012[m]> srinik: well, the wcs codecs are xtensa, so you don't need the DSP for *that*

11:46 <Mis012[m]> Sound Open Firmware could directly run on the wcs

11:46 <srinik> aka_[m]: from SDM845 we started using SoundWire specially for Speakers, but since sm8250 its moved totally to SoundWire and all the digital parts of codec are moved to LPASS

11:47 <Mis012[m]> when you say offloading, I imagine having the hexagon fw coordinate with Linux to read songs from RAM while arm cores sleep, but there's no way qcom has the same idea of offloading :P

11:47 <srinik> Mis012[m]: I meant compress offloading

11:48 <srinik> Mis012[m]: codecs, like mp3, aac...

11:48 <Mis012[m]> srinik: what does the xtensa core on the wcs do then

11:49 <Mis012[m]> SOF currently targets those xtensa cores

11:49 <Mis012[m]> well, not the wcs ones, presumably

11:49 <srinik> Mis012[m]: sorry I have no Idea :-)

11:52 <DylanVanAssche> srinik: while waiting for a response, do you maybe have any insights in the errors in dmesg? They are cryptic for me. https://bin.disroot.org/?b05b45517f74cd4e#8pzEsmb6xBZqtpbLnFyRzE8TX5fRxfM6pipCotXUQjUt

11:55 <Mis012[m]> srinik: well, SOF can run on various xtensa codecs, and it seems it has hot word detection at least

11:55 <Mis012[m]> looks like for AAC, it only has drivers for a hardware codec in some... codec xD

12:07 <DylanVanAssche> srinik: Yes, the earpiece is connected to EAR_OUT_P and EAR_OUT_M.

12:07 <Mis012[m]> doesn't sound like speaker out to me

12:07 <srinik> DylanVanAssche: looks like headset is detected and I also see overflow errors this can be when we try to send SLIMBus stream without setting up channels for ports(as result of incorrect mixer settings) let me share my mixer settings

12:07 <DylanVanAssche> No amplifiers or other ICs are between the earpiece speaker and these connections

12:08 <DylanVanAssche> srinik: Yes, I got the headset mic working with your PR, but not the speakers of the headset

12:09 <srinik> DylanVanAssche: its pretty much from the ucm, headset speakers should also work.. Let dig in my amixer settings

12:11 <srinik> DylanVanAssche: try these https://termbin.com/rd2q then aplay on plughw:0,0

12:16 <srinik> DylanVanAssche: for earpiece try this https://termbin.com/x0om

12:20 <DylanVanAssche> srinik: CDC_IF RX0 MUX, CDC_IF RX1 MUX, and SLIM_0_RX Channels cannot be found, do we miss a config?

12:27 <srinik> DylanVanAssche: pl ingore them I think these are part of downstream mixers..

12:27 <srinik> DylanVanAssche: do you hear anything on headset?

12:29 <Mis012[m]> srinik: do you think it would make sense to port SOF to hexagon then?

12:29 <Mis012[m]> and add sw aac etcdecoders

12:34 <srinik> Mis012[m]: That would be Nice :-) did you know that Qualcomm upgraded Audio firmwares on Hexagon to AudioReach framework which is data driven model, that means you can use ASoC topology to load graphs on the DSP. All the new releases are based on it.

12:35 <DylanVanAssche> srinik: No, I get the same error I had before (tried all devices of the card): https://termbin.com/1b55

12:35 <DylanVanAssche> 0,1 --> Multimedia2, 0,2 --> Multimedia3

12:35 <srinik> DylanVanAssche: you should use MultiMedia1 not MUltiMedia2

12:36 <srinik> DylanVanAssche: aplay -D plughw:0,0 *.wav

12:36 <DylanVanAssche> dmesg

12:36 <Mis012[m]> srinik: ASoC support sure sounds nice

12:36 <DylanVanAssche> srinik: That doesn'

12:37 <DylanVanAssche> doesn't give any sound

12:37 <Mis012[m]> srinik: can it be retrofitted to the old drivers or no luck?

12:38 <srinik> DylanVanAssche: that should have worked.. do you get any overflow errors?

12:39 <DylanVanAssche> srinik: Yes: https://termbin.com/r118 I have to go for a few hours, I will see the messages appear later on

12:40 <srinik> DylanVanAssche: is it that same situation with earpiece?? you can also try to set 'RX1 Digital Volume' mixer

12:40 <srinik> DylanVanAssche: that overflow message should be the last one i guess.

12:41 <srinik> DylanVanAssche: tbh, vkoul tested these mixers few days back am guessing there is something else that we are missing..

12:44 <vkoul> DylanVanAssche: if you trying to set mixers thru amixer, pls try amixer -C0 (if card 0 is your sound card)

16:01 <calebccff> srinik: DylanVanAssche: the SHIFT6mq in question has the speaker codec connected via QUATERNARY_MI2S, I don't think the WCD codec is involved at all. For the op6 and poco F1 we route it to MultiMedia1

16:02 <calebccff> downstream seems to use AIF1_PB on wcd934x for headphones and AIF4_PB for earpiece, I'm not sure how to reflect that properly on mainline, it would require some specific kernel configuration?

16:03 <calebccff> the way I have it sort-of working on op6 is a big mess

16:50 <DylanVanAssche> calebccff: vkoul srinik Ah the speaker output of the WCD codec is not the earpiece? I wasn't aware.

16:51 <DylanVanAssche> Indeed, the main speaker is connected to QUATERNARY_MI2S on the SHIFT 6mq.

16:53 <Mis012[m]> it would be a lot easier if we could all just look at the schematics together ;)

16:54 <DylanVanAssche> Mis012: Blame NDAs 😉 even I don't have the schematics

16:55 <Mis012[m]> SHIFT6mq sounds familliar, I almost feel like I might have seen the schematics at some point... that would be embarassing wouldn't it xD

16:55 <Mis012[m]> could be a false memory though

16:57 <calebccff> Mis012[m] I deem that unlikely

16:58 <Mis012[m]> calebccff: we defintitely discussed *something* about it

16:59 <Mis012[m]> or was it just me getting it confused with other devices that we discussed :P

17:01 <Mis012[m]> I should probably consider writing things down...