ChanServ changed the topic of #linux-msm to:
Caterpillar has quit [Quit: Konversation terminated!]
Daanct12 has joined #linux-msm
whiskey9__ has joined #linux-msm
whiskey9_ has quit [Ping timeout: 480 seconds]
niej_ has joined #linux-msm
niej_ has quit [Quit: Leaving]
vedm has joined #linux-msm
vedm has left #linux-msm [#linux-msm]
pespin has joined #linux-msm
BhushanShah[m] has joined #linux-msm
<BhushanShah[m]> aklimov : Hi! We are trying to debug the audio on FairPhone 5, which is using sm8250 machine driver, and when trying to figure out why amplifier aw88261 is not starting, we had to revert one of your commit. https://gitlab.postmarketos.org/postmarketOS/pmaports/-/issues/3793#note_513819
<krzk> BhushanShah[m]: please refer to mainline commits. We have no clue and no guarantees of what is in pmOS repo or any other fork.
<BhushanShah[m]> z3ntu says that he had to revert same commit in the FP4 case too. So while reverting this makes audio "work", I still need to play around with bclk rate for MI2S as well as available formats. I am wondering if we are missing something wrt clock rate?
<krzk> Above link is some 3rd party gitlab - do you have actual commit?
<krzk> That is a bit odd, setting 16bit is pretty typical for every qcom ADSP platform
<krzk> Even the older ones like sdm845
<BhushanShah[m]> So reverting this, we still have issue with some formats, and it refuses to bring up the amplifier when playing I think 8k sample formats.
<BhushanShah[m]> and audio is still quiet, which may be possibly issue with aw88261 driver, but still just a datapoint
<adrian> awinic amplifiers iirc want S24LE
<adrian> the resourcemanager config on android uses both 24 and 32 bit audio on awinic amps and 16 on basically everything else
<adrian> regarding audio volume, this may be the same problem as https://review.lineageos.org/c/LineageOS/android_vendor_qcom_opensource_arpal-lx/+/436944
<adrian> not 100% sure where this should be configured / applied outside of Android though
Daanct12 has quit [Quit: WeeChat 4.7.0]
<BhushanShah[m]> adrian: interestingly
<BhushanShah[m]> sorry, pressed enter too early
<BhushanShah[m]> regarding, s24le instead of s16le, would that be handled through compatible options I guess?
<aklimov> BhushanShah[m]: does setting it to 24bit result in working audio?
<BhushanShah[m]> I am trying it right now :)
<aklimov> iirc correctly downstream sources about that supported only 16bit and 24bits
<BhushanShah[m]> at least w/o your patch and 24le file, it works (but quietly)
<aklimov> btw, does aw88261 codec driver supports 16bit?
<BhushanShah[m]> I think not
<BhushanShah[m]> but I maybe wrong?
<adrian> nominally the driver includes SNDRV_PCM_FMTBIT_S16_LE
<aklimov> extending qcom audio to support 24bit (not only default 16bit+48kHz) and make it all work with i2s and compressed audio would be a big fun activity
<BhushanShah[m]> aklimov : it seems to advertise support for 16, 24 and 32
<BhushanShah[m]> and rates, of 8k to 48k, and 96k
<aklimov> maybe we should a dapm widget/switch to toggle between {16,24} bits, no idea
<BhushanShah[m]> downstream lahaina.c seems to have widget/switch for formats too xD
<BhushanShah[m]> (is this something typical in other codecs?)
<aklimov> BhushanShah[m]: can you please report it on mailist?
<aklimov> Yep, iirc there was a switch for that in downstream
<aklimov> BhushanShah[m]: you can also add Patrick from Qcom to c/c when reporting
<BhushanShah[m]> aklimov : setting 24bit works
bharadwaj-raju[m] has joined #linux-msm
eugenhristev has quit [Ping timeout: 480 seconds]
eugenhristev has joined #linux-msm
eugenhristev is now known as Guest31057
eugenhristev has joined #linux-msm
Guest31057 has quit [Ping timeout: 480 seconds]
eugenhristev has quit [Read error: Connection reset by peer]
whiskey9__ has quit [Remote host closed the connection]
whiskey9__ has joined #linux-msm
pespin has quit [Remote host closed the connection]
<aklimov> adrian: the link you shared, is it an issue for audioreach devices?
<adrian> i think so? as far as i know, this is because of acdb which also exists on audioreach devices
<adrian> i've been told vendors set the master gain super low in acdb to avoid a "plop" sound at boot
<adrian> qcom apparently fixed this later by setting the ramp period to 0
<adrian> but some devices have acdb entries with the master gain workaround that just sets the master gain super low, presumably because the fix from qcom came later
eugenhristev has joined #linux-msm