|
Bugzilla – Full Text Bug Listing |
|
Description
Gabriele Tettamanzi
2020-05-12 07:46:20 UTC
Created attachment 837654 [details]
alsa info report kernel vanilla k5.6.11-3.g85c1d1e
Created attachment 837655 [details]
alsa info report kernel vanilla k5.6.11-3.g85c1d1e with /usr/share/alsa/ucm/chtmax98090
Created attachment 837656 [details]
/usr/share/alsa/ucm/chtmax98090/
Could you try to remove /usr/share/alsa/ucm2/* files and check whether the problem persists? The alsa-lib should fall back to ucm1 if ucm2 master config isn't found. (In reply to Takashi Iwai from comment #4) > Could you try to remove /usr/share/alsa/ucm2/* files and check whether the > problem persists? The alsa-lib should fall back to ucm1 if ucm2 master > config isn't found. I removed ucm2 and added ucm/chtmax98090/: it is not working with kernel-default and it is not working with kernel-vanilla, in both cases pulseaudio does not start. (In reply to Gabriele Tettamanzi from comment #5) > (In reply to Takashi Iwai from comment #4) > > Could you try to remove /usr/share/alsa/ucm2/* files and check whether the > > problem persists? The alsa-lib should fall back to ucm1 if ucm2 master > > config isn't found. > > I removed ucm2 and added ucm/chtmax98090/: it is not working with > kernel-default and it is not working with kernel-vanilla, in both cases > pulseaudio does not start. Do you mean that kernel-default 5.6.x from TW doesn't work while kernel-vanilla 5.6.x works? That's weird. About UCM: you can check with alsaucm whether it recognized properly. (In reply to Takashi Iwai from comment #6) > (In reply to Gabriele Tettamanzi from comment #5) > > (In reply to Takashi Iwai from comment #4) > > > Could you try to remove /usr/share/alsa/ucm2/* files and check whether the > > > problem persists? The alsa-lib should fall back to ucm1 if ucm2 master > > > config isn't found. > > > > I removed ucm2 and added ucm/chtmax98090/: it is not working with > > kernel-default and it is not working with kernel-vanilla, in both cases > > pulseaudio does not start. > > Do you mean that kernel-default 5.6.x from TW doesn't work while > kernel-vanilla 5.6.x works? That's weird. No. Neither kernel vanilla nor kernel default make sound work. > > About UCM: you can check with alsaucm whether it recognized properly. alsaucm does NOT detect any card (In reply to Gabriele Tettamanzi from comment #7) > (In reply to Takashi Iwai from comment #6) > > (In reply to Gabriele Tettamanzi from comment #5) > > > (In reply to Takashi Iwai from comment #4) > > > > Could you try to remove /usr/share/alsa/ucm2/* files and check whether the > > > > problem persists? The alsa-lib should fall back to ucm1 if ucm2 master > > > > config isn't found. > > > > > > I removed ucm2 and added ucm/chtmax98090/: it is not working with > > > kernel-default and it is not working with kernel-vanilla, in both cases > > > pulseaudio does not start. > > > > Do you mean that kernel-default 5.6.x from TW doesn't work while > > kernel-vanilla 5.6.x works? That's weird. > > No. Neither kernel vanilla nor kernel default make sound work. > > > > > About UCM: you can check with alsaucm whether it recognized properly. > > alsaucm does NOT detect any card Even after restoring ucm2/* profiles? In anyway, below is a patch to add chtmax98090 profile to ucm2 directory. Totally untested. Created attachment 837679 [details]
Test patch to add chtmax98090 profile
(In reply to Takashi Iwai from comment #9) > In anyway, below is a patch to add chtmax98090 profile to ucm2 directory. > Totally untested. I applied the patch and rebooted on kernel-default Good news: I see the card with its profiles in pavucontrol Bad news: it is not emitting any sound, alsaucm listcards reports "the list is empty" I attach below the relevant dmesg and alsa-info report. Created attachment 837683 [details]
alsa-info kernel-default with patch 0001-Add-profile-for-chtmax98090 applied
Created attachment 837684 [details]
dmesg kernel-default with patch 0001-Add-profile-for-chtmax98090 applied
OK, then instead of my patch, just install your own profiles into /usr/share/alsa/ucm2/chtmax98090/*. It's likely that my configs are somehow incorrect. (In reply to Takashi Iwai from comment #14) > OK, then instead of my patch, just install your own profiles into > /usr/share/alsa/ucm2/chtmax98090/*. It's likely that my configs are somehow > incorrect. Tested my old style profile: there's no difference, I see the sound card in piseaudio, but when testing the speakers it is not emitting any sound. (In reply to Takashi Iwai from comment #14) > OK, then instead of my patch, just install your own profiles into > /usr/share/alsa/ucm2/chtmax98090/*. It's likely that my configs are somehow > incorrect. I removed every ucm2 profile for chtmax98090: nothing changes, apparently neither your patch nor my profile are not used. It's interesting the error message I received from alsa-info.sh with no profile in place: "cat: '/sys/module/snd_soc_sst_cht_bsw_max98090_ti/parameters/*' : no such file or directory." This means that it selects cht_bsw_max98090 but a something is missing to make it work. FYI when running Debian testing, it used alsa ucm2, it selected cht_bsw_max98090, last kernel version was 5.6 and the sound works out of the box. Hrm, and yet Leap 15.2 + Kernel:stable worked? Interesting. It implies that the problem is in user-space. You can try to install the latest alsa-lib from git manually: % git clone https://github.com/alsa-project/alsa-lib.git % cd alsa-lib % ./gitcompile % sudo make install The requirement is relatively small, automake, libtool, gcc and glibc-devel. (In reply to Takashi Iwai from comment #17) > Hrm, and yet Leap 15.2 + Kernel:stable worked? Interesting. Yes, but it is using UCM v1 > > It implies that the problem is in user-space. You can tngesry to install the > latest alsa-lib from git manually: > % git clone https://github.com/alsa-project/alsa-lib.git > % cd alsa-lib > % ./gitcompile > % sudo make install > The requirement is relatively small, automake, libtool, gcc and glibc-devel. Cloned, compiled, installed, re booted, tested: I see the sound card in the volume widget, but no sound is emitted. According to alsa-info report, both snd_soc_sst_cht_bsw_max98090_ti and snd_soc_max98090 are loaded, does it make sense? Furthermore alsa-info gives the usual error: "cat: '/sys/module/snd_soc_sst_cht_bsw_max98090_ti/parameters/*' : no such file or directory." (In reply to Gabriele Tettamanzi from comment #18) > (In reply to Takashi Iwai from comment #17) > Cloned, compiled, installed, re booted, tested: I see the sound card in the > volume widget, but no sound is emitted. So it's either not fixed in the latest alsa-lib or something else. To be sure, you can go back to v1.2.2 release point (or whatever Debian package uses), rebuild, re-install and retest: % cd alsa-lib % git reset --hard v1.2.2 % make % sudo make install If this doesn't help, you may try my latest pulseaudio package that is being built in OBS home:tiwai:branches:multimedia:libs/pulseaudio repo. > According to alsa-info report, both snd_soc_sst_cht_bsw_max98090_ti and > snd_soc_max98090 are loaded, does it make sense? Yes. The former is the ASoC machine driver and the latter is the codec driver. ASoC consists of several different components. > Furthermore alsa-info gives the usual error: > "cat: '/sys/module/snd_soc_sst_cht_bsw_max98090_ti/parameters/*' : no such > file or directory." This module really gives no module option, hence there is no corresponding sysfs directories. alsa-info.sh assumes blindly the existence of the option, so you can ignore this error safely. (In reply to Takashi Iwai from comment #19) > (In reply to Gabriele Tettamanzi from comment #18) > > (In reply to Takashi Iwai from comment #17) > So it's either not fixed in the latest alsa-lib or something else. > To be sure, you can go back to v1.2.2 release point (or whatever Debian > package uses), rebuild, re-install and retest: > % cd alsa-lib > % git reset --hard v1.2.2 > % make > % sudo make install Debian Testing is 1.2.2. I did it, it does not help. > If this doesn't help, you may try my latest pulseaudio package that is being > built in OBS home:tiwai:branches:multimedia:libs/pulseaudio repo. > Pulseaudio in your repo on OBS is i586 architecture only, I cannot install it. It's for all archs. It's only that the build didn't finish or wasn't published yet at the moment you tested. Please check it later. And if this doesn't work yet, make sure that your config is right: e.g. clean up ~/.config/pulse/*. If the problem persists, it has to be handled in the upstream. (In reply to Takashi Iwai from comment #21) > It's for all archs. It's only that the build didn't finish or wasn't > published yet at the moment you tested. Please check it later. > > And if this doesn't work yet, make sure that your config is right: e.g. > clean up ~/.config/pulse/*. If the problem persists, it has to be handled > in the upstream. The status of pulseaudio x86_64 13.0.9-1 for Tumbleweed is "blocked" since 12+ hours ago, what if I test the Leap 15.2 version? Also, do I test pulseaudio only or do I change vendor to your OBS repo for all pulseaudio packages in it? The build in OBS went south since yesterday. I re-triggered the builds again, so let's cross fingers. About the update of pulseaudio: you can update the packages selectively manually, too. Download and install pulseaudio.rpm, pulseaudio-module-* and libpulse*.rpm. (In reply to Takashi Iwai from comment #23) > The build in OBS went south since yesterday. I re-triggered the builds > again, so let's cross fingers. > > About the update of pulseaudio: you can update the packages selectively > manually, too. Download and install pulseaudio.rpm, pulseaudio-module-* and > libpulse*.rpm. I deleted all content in .config/pulse/ I added the report and made a zypper dup --allow-vendor-change. I rebooted: no sound emitted yet. I installed the last vanilla kernel from obs HEAD and rebooted on it: no sound emitted. If it makes sense and helps, I'm available to reinstall Debian testing and you guide me to make a comparison within it and TBW. I've tested the internal mic: it works with kernel-default and alsa/pulseaudio from TBW official repos. There's a something that is preventing the hardware from emitting sound. Furthermore the headset mic is working, while the headphone is not. Aha, so something missing in the mixer setup? You're testing with your single-file UCM profile on ucm2 directory, right? It'd be great if you can run Debian system and get alsa-info.sh output from the working environment. (In reply to Takashi Iwai from comment #27) > Aha, so something missing in the mixer setup? Yes, I think it's something like that. The mics (heaadset + internal) work, the hdmi sound too. Only the internal speakers do not work. > You're testing with your single-file UCM profile on ucm2 directory, right? No. At best of my knowledge, I removed all the pieces of software I added for testing purpose: it is an up to date official Tumbleweed now. > It'd be great if you can run Debian system and get alsa-info.sh output from > the working environment. I'm going to install Debian and post back. I installed Debian Bullseye: I have to correct myself, sound is NOT working out of the box but I needed to add my UCMv1 profile in /usr/share/alsa/ucm/ and reboot once and after a positive sound test, I removed my UCMv1 profile and the sound keeps working over reboots. I attach the alsa-info reports from Debian with sound working and from Tumbleweed. Created attachment 837824 [details]
Debian testing alsa-info report, sound working
Created attachment 837825 [details]
Tumbleweed alsa-info report, sound NOT working
(In reply to Gabriele Tettamanzi from comment #29) > I installed Debian Bullseye: I have to correct myself, sound is NOT working > out of the box but I needed to add my UCMv1 profile in /usr/share/alsa/ucm/ > and reboot once and after a positive sound test, I removed my UCMv1 profile > and the sound keeps working over reboots. It means that the mixer setup saved in /var/lib/alsa/asound.state is applied after the reboot and it isn't overridden again by UCM, so it keeps working (at least until the change is needed). It's still interesting how ucmv1 got in place although alsa-lib 1.2.2 is used... UCM profile directories were empty beforehand? > I attach the alsa-info reports from Debian with sound working and from > Tumbleweed. Could you run "alsactl -f alsactl.out store" on the working Debian system, and copy the generated alsactl.out file to TW, restore it via "alsactl -f alsactl.out restore"? At best, copy user UCMv1 profiles into ucm2 directory as-is, restart the session, then run alsactl restore above. (In reply to Takashi Iwai from comment #32) > It means that the mixer setup saved in /var/lib/alsa/asound.state is applied > after the reboot and it isn't overridden again by UCM, so it keeps working > (at least until the change is needed). > > It's still interesting how ucmv1 got in place although alsa-lib 1.2.2 is > used... UCM profile directories were empty beforehand? Correct: on Debian testing UCM is in place, but empty - it should be fallback > > Could you run "alsactl -f alsactl.out store" on the working Debian system, > and copy the generated alsactl.out file to TW, restore it via "alsactl -f > alsactl.out restore"? At best, copy user UCMv1 profiles into ucm2 directory > as-is, restart the session, then run alsactl restore above. Done, copied my ucmv1 profile in ucm2. Rebooted. Now the internal speaker is working. It was not working the automatic switch headphone - speaker. I unmuted left and right channels in alsamixer and stored the value by alsactl store. (In reply to Gabriele Tettamanzi from comment #33) > (In reply to Takashi Iwai from comment #32) > > > > It means that the mixer setup saved in /var/lib/alsa/asound.state is applied > > after the reboot and it isn't overridden again by UCM, so it keeps working > > (at least until the change is needed). > > > > It's still interesting how ucmv1 got in place although alsa-lib 1.2.2 is > > used... UCM profile directories were empty beforehand? > > Correct: on Debian testing UCM is in place, but empty - it should be fallback Thanks, that explains. > > Could you run "alsactl -f alsactl.out store" on the working Debian system, > > and copy the generated alsactl.out file to TW, restore it via "alsactl -f > > alsactl.out restore"? At best, copy user UCMv1 profiles into ucm2 directory > > as-is, restart the session, then run alsactl restore above. > > Done, copied my ucmv1 profile in ucm2. Rebooted. > > Now the internal speaker is working. Good to hear :) Now the remaining piece is to brush up / fix the UCM profile. > It was not working the automatic switch headphone - speaker. > I unmuted left and right channels in alsamixer and stored the value by > alsactl store. Which controls are unchanged while switching between Speaker and Headphone profiles, exactly?
> Which controls are unchanged while switching between Speaker and Headphone
> profiles, exactly?
Headphone left and Headphone right were muted, therefore when the headset was plugged in no sound was emitted by the headphone.
Now both Headphone left and Headphone are always unmuted.
The unmute doesn't happen even if you change the output profile to "Headphone"? The enable sequence there has the toggle of those mixer switches, so it should have influence... Does alsaucm invocation to switch the profile show any error? (In reply to Takashi Iwai from comment #36) > The unmute doesn't happen even if you change the output profile to > "Headphone"? > The enable sequence there has the toggle of those mixer switches, so it > should have influence... > > Does alsaucm invocation to switch the profile show any error? If there's no chtmax98090 folder/profile in ucm2, alsaucm is complaining about its absence. If the folder is in place, this is the error: #~> alsaucm -i alsaucm: Interacive mode - 'q' to quit alsaucm>> open chtmax98090 ALSA lib parser.c:1592:(parse_master_file) Syntax field not found in chtmax98090 ALSA lib main.c:962:(snd_use_case_mgr_open) error: failed to import chtmax98090 use case configuration -22 alsaucm: error failed to open sound card chtmax98090: Invalid argument You need to put "Syntax 2" line at the beginning of chtmax90909.conf. Take a look at other profiles in /usr/share/alsa/ucm2/* as examples. (In reply to Takashi Iwai from comment #38) > You need to put "Syntax 2" line at the beginning of chtmax90909.conf. > Take a look at other profiles in /usr/share/alsa/ucm2/* as examples. Done. No errors, as you can see below: # alsaucm -i alsaucm: Interacive mode - 'q' to quit alsaucm>> open chtmax98090 alsaucm>> reset alsaucm>> set _verb HiFi alsaucm>> set _enadev Headphone alsaucm>> After reboot with the modified chtmax98090 (Syntax 2 added), pulseaudio does not start, therefore sound is not working. And after chtmax98090 removal and reboot, pulseaudio starts, but sound is not emitted again. So we seem to still have some issues with PA. Are you using the pulseaudio stuff from my OBS home:tiwai:branches:multimedia:libs repo? In anyway, let's check alsaucm behavior at first. After you switching the profile via alsaucm, you should be able to check the difference of the mixer setup from the output file of "alsactl -f somefile store". Check that each profile change really does what's expected. (In reply to Takashi Iwai from comment #42) > So we seem to still have some issues with PA. Are you using the pulseaudio > stuff from my OBS home:tiwai:branches:multimedia:libs repo? > > In anyway, let's check alsaucm behavior at first. After you switching the > profile via alsaucm, you should be able to check the difference of the mixer > setup from the output file of "alsactl -f somefile store". Check that each > profile change really does what's expected. We already know it is not working from previous tests (see .2 below); I already attached a few alsa-info reports if you need to compare different settings. I sum up for you the story. 1. **My ucmv1 profile in /usr/share/alsa/ucm/ (not ucm2) folder**: it prevents pulseaudio to start, while as far as I understood, it should work as a fallback mode. 2. **alsaucm** thing: if there's no profile in ucm2, pulseaudio works but alsaucm complains about missing profile; if my original ucmv1 profile is in place, pulseaudio works but alsaucm complains about errors; if my ucmv1 profile corrected with the "Syntax 2" line is in place, pulseaudio does not start 3. **Pulseaudio and the likes from multilib**: they do not make any difference when compared to the official packages 4. **Your ucmv2 profile**: it does not make the sound work 5. **Debian comparison**: Debian Bullseye uses my ucmv1 profile in /usr/share/alsa/ucm/; we know from my tests that using on Tumbleweed the alsa settings from Debian makes sound work, but it keeps working only till the next reset. **My conclusion from the above story:** alsa ucmv1 fallback mode is not working on Tumbleweed: this is the bug in my opinion. In addition to that there's neither a proper ucmv2 profile nor a proper ucmv1 profile installed for my specific sound card. Side note: on Debian there's a specific non-free firmware-intel-sound package installed, I removed it and the soundcard stopped working; an old version of the same package is on OBS, but there's no openSUSE version - here is the link to it on Debian repo: https://packages.debian.org/en/bullseye/firmware-intel-sound. Gabriele, we're tracking rather the issue in the latest upstream. Debian is way behind the latest upstream code, after all. The behavior of alsaucm I asked was the combination of the latest alsa-lib git, the latest patched pulseaudio and your modified UCMv2 profile. I know that you wrote that PA didn't work, so the question isn't about whether the speaker works or not: instead, we need to check whether alsaucm behaves as expected or not. That's why I asked to check to perform alsaucm with different profile switches and confirm the alsactl output at each profile switch. It's not a question of one-off shot at some profile change, but it needs your manual inspection of alsactl outputs at each profile switch. You wrote your UCM profile, so you must know what's executed there. Once after we confirm that alsaucm works, then we can go ahead and look into pulseaudio problem -- at best with the involvement of upstream devs. (In reply to Takashi Iwai from comment #44) > Gabriele, we're tracking rather the issue in the latest upstream. Debian is > way behind the latest upstream code, after all. You're by far more expert than I am, just keep in mind that I'm running the comparison with Debian testing (and not stable). > The behavior of alsaucm I asked was the combination of the latest alsa-lib > git, the latest patched pulseaudio and your modified UCMv2 profile. I know > that you wrote that PA didn't work, so the question isn't about whether the > speaker works or not: instead, we need to check whether alsaucm behaves as > expected or not. What you are asking me to do is: - to re install alsa-lib from git - to add again the multilib repo and update with vendor change allowed Is this correct? Also, you're asking me to add a ucm profile, but which one and where? Which one in: my original ucm V1 rofile, my original ucm V1 with the line "Syntax "2 added, the ucm V2 profile you posted as patch? Where: either ucm or ucm2 folder? > > That's why I asked to check to perform alsaucm with different profile > switches and confirm the alsactl output at each profile switch. It's not a > question of one-off shot at some profile change, but it needs your manual > inspection of alsactl outputs at each profile switch. You wrote your UCM > profile, so you must know what's executed there. Unfortunately I have not even a clue of what I should check manually that cannot be checked by a diff in a by far faster and more accurate way. Here is the story of my ucmv1 profile: I was running GalliumOS (at present Ubuntu 18.04 based, focus on Chromebooks), they ship the profiles of the git ucm project by PL Bossart, however the mics were not working and while inspecting the HiFi.conf I discovered two #FIXME comments exactly related to the mics: I fixed them by a trivial and boring trial and error process. > Once after we confirm that alsaucm works, then we can go ahead and look into > pulseaudio problem -- at best with the involvement of upstream devs. I did the following test: 1. to place my ucmV1 profile in /usr/share/alsa/ucm/ (that's the correct folder for ucmV1 profiles) 2. to install pulseaudio from home_tiwai_branches_multimedia_libs 3. clone / compile / install alsa-lib from github/alsa-project 4. reboot After reboot pulseaudio was up and running, the sound card was recognized by pulseaudio but it was not emitting sound. I issued one by one the following alsaucm instructions, mentioned in the comment section at the beginning of the chtmax98090/HiFi.conf file: a. alsaucm -c chtmax98090 set _verb HiFi set _enadev Headphone b. alsaucm -c chtmax98090 set _verb HiFi set _enadev Speakers c. alsaucm -c chtmax98090 set _verb HiFi set _enadev HeadsetMic d. alsaucm -c chtmax98090 set _verb HiFi set _enadev InternalMic After each one of the above commands, I did a manual inspection of the alsa switches by alsamixer. To the best of my understanding and experience, it turned out the switches are muted and unmuted the correct way. Furthermore after this test sound card started to work and to plug/unplug the headphone works as it should. It looks promising, what are the next steps? Sorry for the late reply, as I was stuck at other (unwilling) things. And thanks for the detailed testing. The result sounds good, so far. The fact that alsaucm works indicates that basically the driver and the alsa-lib are OK. Let's move on to PA now. I built pulseaudio package from the latest git repo, based on 13.3-rc1 release. The packages are found in OBS home:tiwai:test:pa-git/pulseaudio repo. Could you upgrade to those ones and try with your UCMv2 profile? (In reply to Takashi Iwai from comment #47) > Sorry for the late reply, as I was stuck at other (unwilling) things. And > thanks for the detailed testing. The result sounds good, so far. No worries mate, hope you're doing well now. > The fact that alsaucm works indicates that basically the driver and the > alsa-lib are OK. Let's move on to PA now. This makes sense to me - the behavior is reproducible: after each and every boot with alsa from git and pulseaudio from tiwai multilib the sound from speakers starts to work only if I issue in a terminal: alsaucm -c chtmax98090 set _verb HiFi set _enadev Speakers > I built pulseaudio package from the latest git repo, based on 13.3-rc1 > release. The packages are found in OBS home:tiwai:test:pa-git/pulseaudio > repo. > Could you upgrade to those ones and try with your UCMv2 profile? First tentative: I installed pulseaudio 13.99.1-rebootstrapped from tiwai:test:pa-git, my profile ucmv1 in /usr/share/alsa/ucm/chtmax98090/: no changes, the sound from speakers starts to work only if I issue in a terminal: alsaucm -c chtmax98090 set _verb HiFi set _enadev Speakers. sent t I'll do two more tests with the ucmV2 patch you sent and with my ucmV1 placed in ucm2 and report back. Second tentative: moved my ucmV1 profile to /usr/share/alsa/ucm2/ and added the Syntax 2 line at the top of chtmax98090.conf, rebooted: same results as first tentative. Third tentative: removed ucmv1 profile and added the ucmV2 profile from your patch in /usr/share/alsa/ucm2/, rebooted: - speakers working - headphone working - internal mic working - headset mic not working, the execution of: alsaucm -c chtmax98090 set _verb HiFi set _enadev HeadsetMic gives the following error: ALSA lib main.c:392:(execute_sequence) cdev is not equal for playback and capture! alsaucm: error failed to set _enadev=HeadsetMic: Invalid argument Great job Takashi, it looks very close to be fixed! (In reply to Gabriele Tettamanzi from comment #49) > - headset mic not working, the execution of: > > alsaucm -c chtmax98090 set _verb HiFi set _enadev HeadsetMic > > gives the following error: > > ALSA lib main.c:392:(execute_sequence) cdev is not equal for playback and > capture! > alsaucm: error failed to set _enadev=HeadsetMic: Invalid argument Not clear yet, but could you try to just remove the line defining "CaptureCTL" in HeadsetMic.conf? This was carried from your modified config, but it might be superfluous. ... and now I re-read the UCM code and noticed that the CaptureCTL is utterly wrong. I guess it did work for your old UCMv1 casually because it wasn't used at all :) (In reply to Takashi Iwai from comment #51) > ... and now I re-read the UCM code and noticed that the CaptureCTL is > utterly wrong. I guess it did work for your old UCMv1 casually because it > wasn't used at all :) As I wrote previously, I'm far from being an expert... Success! I removed the CaptureCTL line from /usr/share/alsa/ucm2/codecs/max98090/HeadsetMic.conf and 1. alsaucm -c chtmax98090 set _verb HiFi set _enadev HeadsetMic works without errors 2. after reboot, the sound system is working as expected, including: - to switch from speakers to headphone and to internal mic to headset mic when the headset jack is plugged in - to work properly with hdmi external speakers and usb external mic Does it make sense to test bluetooth sound too? I did the bluetooth test too with a headset: both mic and headphone work. It sounds like this bug has a fix! Sorry for the trivial question: I'm now running alsa from github and pulseaudio from a test repo of yours, are these packages going to be part of official Tumbleweed? if yes, what is your time estimation for it to happen? Thank you so much for all. I'm going to update alsa package for TW, and that's an easy part. And, which pulseaudio package did you test and succeed? Is it OBS home:tiwai:branches:multimedia:libs, or OBS home:tiwai:test:pa-git? The latter is a bit hard to submit to TW, as it's a complete upgrade to the development version. The former is a result with selective backports, so we can take easier. One more question: since the UCM profile is with some non-standard values, it might be better to put it specific for Chromebook. Below is another patch for alsa-ucm-conf. Could you try this one instead of the previous generic chtmax98090? Created attachment 838155 [details]
alsa-ucm-conf patch for Chromebook profile
(In reply to Takashi Iwai from comment #54) > I'm going to update alsa package for TW, and that's an easy part. > > And, which pulseaudio package did you test and succeed? Is it OBS > home:tiwai:branches:multimedia:libs, or OBS home:tiwai:test:pa-git? > The latter is a bit hard to submit to TW, as it's a complete upgrade to the > development version. The former is a result with selective backports, so we > can take easier. I'm running pulseaudio from OBS home:tiwai:test:pa-git. I'll repeat the test with home:tiwai:branches:multimedia:libs (In reply to Takashi Iwai from comment #56) > Created attachment 838155 [details] > alsa-ucm-conf patch for Chromebook profile Sure, I'll test it with both pulsaudio versions from OBS home:tiwai:branches:multimedia:libs and OBS home:tiwai:test:pa-git Good news! First, the new ucmV2 profile from 0001-Add-profile-for-Chromebook-Asus-C300.patch is working. Second, I tested the new profile as follows: a. added OBS home:tiwai:test:pa-git, zypper ref, zypper dup --allow-vendor change, reboot b. removed previous repo, added OBS home:tiwai:branches:multimedia:libs , zypper ref, zypper dup --allow-vendor change, reboot c. removed previous repo, official repos only in place, zypper ref, zypper dup --allow-vendor change, reboot Sound is working in the three above described conditions. At the moment my Asus C300 is running with sound working with: pulseaudio --version pulseaudio 13.0-rebootstrapped ALSA Version Driver version: k5.6.12-1-default Library version: 1.2.2 Utilities version: 1.2.2 What are the next steps? Then let's go for submitting the UCM profile to upstream. Meanwhile I'll submit the pulseaudio changes to TW; this should be needed in some corner cases with UCMv2. Just to be sure, could you give the alsa-info.sh output from the working state? Created attachment 838208 [details]
alsa-info working state
(In reply to Takashi Iwai from comment #60) > Then let's go for submitting the UCM profile to upstream. Great! > Meanwhile I'll submit the pulseaudio changes to TW; this should be needed in > some corner cases with UCMv2. There's a large number of variations in sound hardware indeed. > Just to be sure, could you give the alsa-info.sh output from the working > state? Just added the alsa-info report Thanks. I tested ucm-validator with the attached alsa-info.sh and it worked. (Actually it exposed a bug in ucm-validator, and now it's fixed.) The ucm-validator spotted a few issues in my previous UCM profile. The revised version is attached below. It should work like before. Created attachment 838210 [details]
Revised patch to alsa-ucm-conf
Created attachment 838217 [details] alsa-info report working state pulseaudio 13.0-9.1 from home_tiwai_branches_multimedia_libs alsa 1.2.2-3.1 from Main Repository (OSS) - Driver version: k5.6.12-1-default - Library version: 1.2.2 - Utilities version: 1.2.2 ucmV2 profile: attachment #838210 [details] (In reply to Takashi Iwai from comment #63) > Thanks. > > I tested ucm-validator with the attached alsa-info.sh and it worked. > (Actually it exposed a bug in ucm-validator, and now it's fixed.) > > The ucm-validator spotted a few issues in my previous UCM profile. The > revised version is attached below. It should work like before. I realized that I did an incomplete test with the previous ucmV2 profile: I didn't try to plug and unplug the headset when testing the different versions of pulseaudio. With the last ucmV2 profile (#83210) and pulseaudio 13.0-6.1 from Main Repository (OSS), when headset jack is plugged and unplugged the switch internal / headset are not working. With the last ucmv2 profile (#83210) and pulseaudio 13.0-9.1 from OBS home:tiwai:branches:multimedia:libs everything is working. I attached the alsa-info report for the working condition. Thanks for the updates. My OBS branch of pulseaudio was now merged and submitted to TW. Also the update of alsa-ucm-conf is on its way to TW. So, all done for TW. Let's wait until the needed updates reach to TW snapshot. (In reply to Takashi Iwai from comment #67) > Thanks for the updates. My OBS branch of pulseaudio was now merged and > submitted to TW. Also the update of alsa-ucm-conf is on its way to TW. > > So, all done for TW. Let's wait until the needed updates reach to TW > snapshot. Good to read it, thank you so much. The upstream merged the change for UCM now. The next release will include the support officially. Let's close this bug. |