ella-0_ has quit [Read error: Connection reset by peer]
bluetail has joined #asahi-dev
chadmed has quit [Quit: Konversation terminated!]
chadmed has joined #asahi-dev
<chadmed>
povik: been doing some testing on the userspace side, JACK also opens the secondary FE as a 32 channel device and complains about being unable to buffer stuff properly
thelounge7571340 has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
<chadmed>
the latest ucm changes also break pipewire on the laptops, it prevents PW from being able to set profiles so it forces the device off
<chadmed>
ucm stuff notwithstanding, this really doesnt look like a userspace problem to me anymore
dgb has joined #asahi-dev
dgb has quit [Quit: dgb]
chadmed has quit [Quit: Konversation terminated!]
dgb has joined #asahi-dev
duban6 has quit []
duban6 has joined #asahi-dev
thelounge7571340 has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
chip_x has joined #asahi-dev
chipxxx has quit [Ping timeout: 480 seconds]
psykose has joined #asahi-dev
gyronaut has joined #asahi-dev
gyronaut has quit []
thelounge7571340 has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
thepigguy has quit [Quit: thepigguy]
<jannau>
aahhhhhhh! j493 CS42L84 Jack DAC wraps around from min to max gain, asahi-wip and alsa-ucm-conf-asahi 645f6c7
dgb has quit [Quit: dgb]
___nick___ has joined #asahi-dev
___nick___ has quit []
jn has quit [Ping timeout: 484 seconds]
___nick___ has joined #asahi-dev
seb4nihel has joined #asahi-dev
chadmed has joined #asahi-dev
blazra has joined #asahi-dev
jn has joined #asahi-dev
<jannau>
I guess the converted integer values get negative and due to the overflow in the signed to unsigned conversion we write 511 (the max) to the registers, probably fixed by using SOC_DOUBLE_R_S_TLV which allows us to specify a minimum
thelounge7571340 has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
<jannau>
no, the problem seems to be in alsa code before the volume write function, maybe the issue is the alsa uses the register values directly, the control has just the LSB regs
chadmed has quit [Quit: Konversation terminated!]
<jannau>
progression of volume values is 283, 200, 511. so it gets bad after LSB increases despite decreasing the volume
chadmed has joined #asahi-dev
<chadmed>
ok my last ditch hail marry attempt to fix the 32 channel thing failed so im officially out of ideas on that front
<chadmed>
s/marry/mary
povik has joined #asahi-dev
<jannau>
no, it's the mute flag in DECLARE_TLV_DB_SCALE. single character fix
<povik>
chadmed: noted
<povik>
jannau: i think we still want the mute flag, but that then means the minimum is off, no?
<povik>
(i was just looking up this thread on register/control value confusion in ASoC, but that does not seem relevant in the end vvv )
<povik>
chadmed: it's as if it doesn't read anything besides the headset mic from the profiles. only idea i have at the moment, is that it fails if the no. of channels isn't specified for a device
<povik>
because that's how the mic differs from the others
<povik>
you can try inserting PlaybackChannels here
<povik>
there's no formal requirement to specify the no. of channels AFAIK, and pulseaudio assumes stereo if unspecified
<povik>
it sucks we have to probe the behavior of different sound servers like this
<jannau>
povik: known and fixed in the version merged for 6.1
<povik>
:)
<jannau>
povik: headset detection didn't work on boot, I had to replug the headset
<jannau>
behavior in alsamixer is now good, muting works
<povik>
ayy!
<povik>
will see about the replug
<jannau>
the kde balance behavior might be intentional. it seems to adjust both channels and the resets unmodified channel. with the fixed volume scale it's much better
<jannau>
user experience seems to still bad due to slow volume updates of the driver
<jannau>
did you check if the i2c irq support improves this?
<povik>
ah, there's two causes for that reinforce each other: the slow i2c driver and no regcache in the codec driver
<povik>
it for sure does improve it, because otherwise all transactions take some number of milliseconds
<povik>
s/that/that that/
<povik>
we should pick it up, no? last version had merely formal issues i think
thelounge7571340 has joined #asahi-dev
chadmed has joined #asahi-dev
<jannau>
yes, I was thinking the same
<marcan>
yeah, we should
thelounge7571340 has quit [Remote host closed the connection]
<marcan>
I'm thinking tomorrow I'm going to rewrite cpufreq one more time and then do another kernel rebase/update dance
<marcan>
anything else I should pick up?
r0ni has joined #asahi-dev
<jannau>
povik: plug/mic detection is logged during boot so the driver might just lose the state?
<povik>
marcan: i will prepare a batch. i can throw the amarioguy's patch in
<marcan>
sure :)
<povik>
jannau: it does report it detected the mic?
<marcan>
on my side I have the m1 max studio hub thing
<jannau>
marcan: bluetooth v3 to get rid of the revert which broke bluetooth
<jannau>
there is the keyboard backlight PR
<marcan>
noted
<marcan>
(bbl, dinner; will catch up on anything else when I get back)
<jannau>
povik: please include the cs42l42 compatible removal, I will close my PR as the m1 ultra cpufreq dt fix won't make sense if marcan rewrites cpufreq
<sven>
hm, Bluetooth v3 is still in patchwork as “new” and with a few more other series in front of it :(
kov has joined #asahi-dev
<jannau>
povik: yes, plug and mic are detected on initial driver load
chadmed has quit [Quit: Konversation terminated!]
chadmed has joined #asahi-dev
<jannau>
yes, the i2c irq patch makes the volume control much faster. thanks amarioguy
<chadmed>
interesting, that HSBIAS pin disable results in a null pointer deref on j314
Core2080 has joined #asahi-dev
<jannau>
povik: do you know if it is expected that alsamixer uses 2 write ioctls to update the DAC volume?
adryzz has quit [Ping timeout: 480 seconds]
<jannau>
ChaosPrincess: I'm looking and will test on j493
<ChaosPrincess>
thats the m2, right?
<jannau>
yes, m2 macbook pro 13"
Core9066 has joined #asahi-dev
chadmed has quit [Quit: Konversation terminated!]
chadmed has joined #asahi-dev
<chadmed>
nvm that was a dtb issue, the jack works with ucm and pipewire with the branch you linked povik
chadmed has quit []
chadmed has joined #asahi-dev
chadmed has quit []
chadmed has joined #asahi-dev
<chadmed>
ok no nvm kernel null pointer if speakers are enabled???
<chadmed>
how fragile is asoc...
chadmed has quit []
<povik>
chadmed: that will be a bug in my driver
<povik>
if speakers are enabled, it takes longer to probe the sound card, hence the jack detection happens before the codec is bound to a sound card
<povik>
but if it doesn't have a sound card, it doesn't have a dapm reference
<povik>
that will probably be it
chadmed has joined #asahi-dev
<povik>
jannau: no idea, makes sense to me if the ioctl is limited to updating one channel at a time
chadmed has quit []
<jannau>
povik: the ioctl argument is struct snd_ctl_elem_value so it should be able to set channels values atomically
<povik>
yeah, that's weird then
<povik>
maybe it does a read after the write?
<povik>
it probably does that, otherwise it wouldn't have the wraparound from earlier before the fix
<amarioguy>
povik: alright lemme send you an updated version of my patch today, need to add the reg_imask write in the IRQ handler as pointed out in the v3 mailing list thread
<amarioguy>
jannau: no problem :)
<jannau>
it does a read after a write but for both channels. I added a debug printk in the generic ioctl handler
thelounge7571340 has quit [Remote host closed the connection]
<amarioguy>
i'll send over the revised patch when i can, do you need the email commit description too or just the diff?
amarioguy has quit [Quit: Leaving]
<povik>
amarioguy: i need a patch i can give to git am, as long as it has a sensible patch title it's good
amarioguy has joined #asahi-dev
adryzz has joined #asahi-dev
Core2080 has quit [Read error: Connection reset by peer]
<jannau>
ChaosPrincess: done
<amarioguy>
povik: alright i'll send the .patch then, gimme a sec
<ChaosPrincess>
ty
<sven>
jannau: huh, so devm_clk_* helpers finally got merged?
<sven>
in may already. fun!
<sven>
i know that there have been various patch series trying to add devm_clk_* since a few years ago already that all didn't go anywhere for various reason
<sven>
urgh, looks like I'll have to implement a stripped down version of OSSerialize for dcp :<