Saturday, 2023-03-25

malI think gsi needs to have android side running for it to work00:10
malthere was some userspace service for it00:10
malthat is why we have these enabled in official devices like x10 III because otherwise sailfish recovery won't have working usb00:11
malsame goes for early init telnet00:12
T42<b100dian> But that's 4.19. Maybe on 5.4 already has "(GSI, which replaced BAM)" ... I have RMNET_BAM in defconfig but not in .config00:27
malyes, but x10 III uses gsi in android00:36
T42<TheVancedGamer> so I got farther than last time, now it wants root/ folder in out directory which I don't have06:28
T42<TheVancedGamer> instead I have ramdisk/ folder and TARGET_COPY_OUT_ROOT points to it as well06:29
T42<TheVancedGamer> making it point to ramdisk/ got it this far, now it wants init*.rc files in there08:15
T42<k1gen> mal: I have a question about lib & lib64 symlinks: what if I have odm in /vendor, and not /, how do I create symlinks in sparse then? if I try this on-device: mkdir: can't create directory '/vendor/odm/lib': Read-only file system08:22
T42<k1gen> wait, I also have odm in system_root08:48
T42<k1gen> nvm, it's a symlink to read-only /vendor/odm08:52
T42<k1gen> mal: please ping me if you answer10:26
T42<adampigg> mal: Updated UCM and Pulse config on the PP too.....11:11
T42<adampigg> Tested calls on both the PP and PPP11:11
T42<adampigg> On PP - audio quality seems ok to me11:11
T42<adampigg> But no ringtone/system sounds11:11
T42<adampigg> On PPP, mic is a bit quiet, but calls were working, and ringtone worked11:11
T42<adampigg> But there seemed to be some stutter/lag when the call came in11:12
T42<elros34> @TheVancedGamer paste whole logs not just few lines. So what android version (that would be clear from whole paste)? did you build hybris-hal without errors?11:22
T42<TheVancedGamer> building for hybris-18.1, hybris-hal built without issues11:23
T42<elros34> so run 'allmod' and search for fstab uevent11:23
T42<TheVancedGamer> hmrph, I've deleted hadk now11:24
T42<elros34> ?11:24
T42<TheVancedGamer> ill try again, I saved device trees and rpm repos11:24
T42<TheVancedGamer> I got as far as droid-hal wanting init.zygote*11:25
T42<elros34> I don't get why did you delete hadk?11:25
T42<TheVancedGamer> I was getting impatient probably, and thought starting fresh and using some other device trees might fix this11:27
mal@k1gen you just add /odm folder and add the symlinks there12:11
T42<k1gen> so, sparse/odm/lib{,64}?12:12
malcan you show what you have in that /vendor/odm12:12
mal@k1gen first test on device to get things working, then add to droid-config repo12:13
T42<k1gen> can't test on-device because of read-only /vendor12:13
malI SAID /odm12:14
malyou DO NOT do anything to /vendor/odm12:14
T42<k1gen> there is no odm folder in root, should I create it?12:14
malwhat did I just say few minutes ago12:15
T42<k1gen> got it12:15
T42<k1gen> sorry12:15
malI still would like know what the content of /vendor/odm is12:17
malbecause I had this kind of symlink also
T42<k1gen> mal: phone was booting, couldn't send earlier12:19
T42<k1gen> also contents of permissions:
malI think that etc symlink could be useful, some devices have useful things in the build.prop file12:21
T42<k1gen> not really:
malok, then it's not needed12:23
malI have one device which has a lot of audio related properties in that file, you seem to have just unused file12:24
mal@k1gen if you try to start /usr/libexec/droid-hybris/system/bin/minimediaservice manually if it complains about libraries it doesn't find then you need to add symlinks for those to /odm12:27
T42<k1gen> mal: I have library "" needed or dlopened by "/usr/libexec/droid-hybris/system/lib64/" is not accessible for the namespace "(default)" trying to run gst-inspect-1.0 | grep droid12:27
T42<k1gen> thanks for minimediaservice suggestion, I'll do that12:28
malas you can see that is one of the libraries symlinked in my repo12:29
T42<k1gen> with this /usr/libexec/droid-hybris/system/bin/minimediaservice doesn't output anything12:30
maldoes it start?12:30
malif so then things should be ok12:31
T42<k1gen> let me reboot and check, because after I first tried to launch it manually, device wouldn't turn on12:32
T42<k1gen> mal: sorry, I'm back. when I launch minimediaservice, it doesn't output anything, and camera still doesn't work14:06
T42<k1gen> maybe there's some env that could make it output debug information?14:07
malminimediaservice should be starting automatically14:11
malcheck if it's running14:11
malalso just in case remove /home/defaultuser/.cache/gstreamer*14:12
malgstreamer blocks plugins that fail so gst-droid could have been blocked from your earlier tests14:13
T42<k1gen> it's running, and I deleted gst cache14:14
maldoes gst-inspect show droid things now?14:15
T42<k1gen> it returns library "" needed or dlopened by "/usr/libexec/droid-hybris/system/lib64/" is not accessible for the namespace "(default)" again, though I have in both/odm/lib and /odm/lib6414:18
malrecheck that the symlinks are pointing to correct files, I mean that the target exists14:20
malyou could try to do something like: strace -Ff -s1024 gst-inspect-1.0 --gst-disable-segtrap --gst-disable-registry-fork droidcamsrc14:24
malhopefully you have strace installed14:24
T42<elros34> add '-e file' to simplify output14:25
T42<k1gen> thanks, saw the message while writing > file :D14:26
malI need to go out for a while, back in an hour or so14:27
T42<k1gen> @elros34 full strace:, part:
T42<k1gen> what should I do now?14:33
T42<elros34> so you still have same error?14:39
T42<elros34> you again remove prompt to make it harder and I bet you start it as root14:40
T42<k1gen> wait for mal I guess14:44
T42<k1gen> sorry, messages loaded right after I hit send14:45
T42<k1gen> I'll start it as defaultuser and send log with prompt in a minute14:46
T42<k1gen> prompt: strace -Ff -s1024 -e file -o gst-inspect.strace gst-inspect-1.0 --gst-disable-segtrap --gst-disable-registry-fork droidcamsrc14:48
T42<k1gen> file:
T42<elros34> you are stubborn14:49
T42<k1gen> am I? I'm sorry. I did as you asked, should I remove -o gst-inspect.strace?14:50
T42<elros34> again same question do you see same error in output? Have you checked logcat at the same time?14:51
T42<k1gen> I see the same error in output, yes. I didn't check logcat, I'll send it now14:53
T42<elros34> then this might be linker related. What if you create same symlink but int that user/libexec path14:54
T42<k1gen> wait, pager opened then segfault, let me get all this in pastebin14:57
T42<elros34> maybe simlink will not change anything and it would better to bind mount library14:58
T42<k1gen> logcat:, stdout:, strace file:
T42<k1gen> main camera works after a reboot15:01
T42<k1gen> after I close and open camera app, black screen again15:01
T42<elros34> so what was this debuging about if camera works at least once?15:02
T42<k1gen> I worked just now and just once, and it doesn't work anymore after app relauch15:02
T42<elros34> for that you must see what happens excatly in logcat when you start camera first time, close and second time15:04
T42<elros34> making sure camera app doesn't hang or crash15:05
T42<elros34> do you have some dirs in ls -alR /linkerconfig/ and different files?15:09
T42<k1gen> yes:
T42<elros34> is ld.config in default dir different than in other? This part is not handled correcty because of hybris patches but maybe your device need it. First check logcat, it should gives you better overview15:20
T42<elros34> ah based on size it is different15:21
T42<k1gen> logcat: camera app freezed on ioctl c0044901 failed with code -1: Inappropriate ioctl for device15:36
T42<k1gen> camera works only once, right after reboot, then black screen in camera app15:37
T42<k1gen> @edp_17 no matter what I do, I can't resize the icons. I follow you script, but there are 7 icons in a row. any ideas? how did you resize the icons on your last device?15:39
rostam98[m]Hello Is there a docker image for saiilfishos sdk  that I can use for my cross development of arm64 and x86 ?15:43
T42<k1gen> can I apply changes from droid-config-$DEVICE.spec without reflashing rootfs?15:49
T42<elros34> @k1gen btw you should run logcat as root16:14
T42<k1gen> any ideas about icons? I already edited values in /etc/dconf/db/vendor.d/silica-configs.txt, and after reboot nothing changed16:18
T42<k1gen> I also defined icon_res in droid-config-$DEVICE.spec16:19
T42<elros34> you want to resize icons or change number of icons in row? If first wouldn't be easier to copy droid-config packages and install them on device?16:24
T42<elros34> did you run dconf update after manually changing dconf config?16:24
T42<elros34> have you check they relly changed by dconf read?16:25
T42<k1gen> thank you for suggestions. answering first question: both preferably, second and third: I'll check this soon16:31
mal@k1gen you can always copy the droid-config rpms to the device and install the ones of those you need, note that not all of those are installed16:42
T42<elros34> there was patch for changing number of icons but I am not sure it still works16:59
T42<k1gen> 4 icons in a row now, thanks for helping17:21
T42<k1gen> why is there no battery status icon? there's a lot of files in /sys/class/power_supply/battery, I can cat them17:28
T42<elros34> because you mess with icons or you switch to IVI mode?:)17:29
T42<k1gen> no, there was no battery status in ui before I messed with icons. what's IVI mode?17:30
T42<b100dian> heh, wasnt' expecting In-Vehicule-Infotainment reference so soon17:34
T42<k1gen> ah, the built-in-car tablet. why would I be in IVI mode?17:35
T42<elros34> just joking, jolla is involved int that area17:35
T42<elros34> don't you have battery percentage on the left top side of homescreen?17:36
T42<elros34> even with usb connected?17:36
T42<k1gen> nah, all I have is Wi-Fi icon in the right side17:36
T42<k1gen> even with usb17:37
T42<elros34> maybe something change in code but so far batter icon was always visible. You need to see if mce works and detects everything correctly, maybe start in more verbose mode17:38
T42<b100dian> step 1: does your phone have a battery17:38
T42<k1gen> :D17:39
T42<b100dian> (sorry, couldnt' resist)17:39
T42<k1gen> @elros34 how do I "start in more verbose mode" and what's the name of mce responsible for battery indicator stuff?17:39
T42<elros34> @b100dian you got it:
T42<k1gen> lol17:41
T42<elros34> did you add hw-settings.ini file?17:42
T42<k1gen> if it wasn't added by-default then no, I didn't17:43
T42<k1gen> /usr/share/csd/settings.d exists, but is empty, if this directory is relevant17:45
T42<b100dian> Well, of course that line would exist, in hindsight17:45
T42<b100dian> Btw, hw-settings.ini is also missing from hadk17:46
T42<k1gen> first hearing about it17:46
T42<elros34> use magic ctrl+f on hadk docs17:46
T42<k1gen> magic ctrl+f :P17:47
T42<k1gen> hw-settings.ini Phrase not found17:47
T42<k1gen> in hadk.pdf17:47
T42<k1gen> also in hadk-hot17:48
T42<k1gen> ah, didn't search in hadk-faq17:50
T42<k1gen> what's VoltageTest for?17:52
T42<b100dian> Yes, I think mal was asking about stuff that's missing from hadk pdf, and I remember I was surprised to find out so late about the hw-settings.ini myself back in the day17:53
T42<k1gen> after I get this device being usable as a daily-driver I'll commit everything and contribute to hadk with mal's help17:55
T42<b100dian> Unfortunately I don't think the source text for hadk is in a git repo, is it?17:55
T42<k1gen> we'll worry about it later :)17:56
malsuggestions for improving hadk pdf are always welcome17:57
T42<k1gen> mal: do you know what VoltageTest does?17:57
malwhere is that?17:59
T42<b100dian> probably a csd tool thing?17:59
T42<k1gen> /usr/share/csd/settings.d/hw-settings.ini17:59
malnot sure really what it does, it seems to be disabled for example on xperia 10 III18:00
malmaybe I should check the sources18:00
T42<k1gen> there's no VoltageTest in hadk folder, used ripgrep to find out18:02
T42<b100dian> @k1gen some packages aren't compiled on your device and I don't think ripgrep knows rpm/zstd/cpio18:03
T42<k1gen> thanks, I guess it's somewhere on github then18:04
T42<b100dian> @elros34 if you can still follow, I added gsi.rndis along with rndis_bam.rndis and next I disabled `CONFIG_USB_CONFIGFS_F_GSI=n` and `CONFIG_USB_CONFIGFS_ECM=n` Both logs add depmod -a in the hope that rmnet_core/ctl .ko modules would help (they don't?)18:08
T42<k1gen> how can I reset ambience icons in top drawer? most of them are grey somehow18:08
T42<b100dian> I see that mkdir rndis_bam.rndis fails in both and mkdir gsi.rndis fails in the second log (expected). Neither show usb life on my host. Also dmesg shows IPA framework init but no corresponding "IPA driver is now in ready state" (ipa_fmwk.c)18:11
T42<elros34> you disabled ecm? IIRC it was disabled according to your .config so how about enabling it and adding ecm.rndis to script? Maybe you should first remove rndis.usb0 when testing both ecm and gsi. This could be wrong approach and problem maybe with this dummy controller18:25
T42<elros34> hmm according to first log ipa is not ready so gsi fails18:28
T42<elros34> second log looks more promissing. I wonder if you could also disable IPA?18:32
T42<b100dian> Well, I added ECM=n but you're right it was not set before.18:33
T42<b100dian> I tried that, it was in a techpack config, but first try things didn't compile. I will try to see more info about that18:33
T42<elros34> what about dummy_hcd? is it config18:34
T42<b100dian> about ipa not being ready - that is made ready in a post_init which seems to need a firmware named ipa_fws but I did not grok where it looks for it (it is not in firmware_mnt)18:34
T42<b100dian> CONFIG_USB_DUMMY_HCD=y in .config18:35
T42<b100dian> >  looping all data transfer requests back to a USB "gadget driver" in the same host.18:35
T42<elros34> worth to try getting rid of it18:37
T42<b100dian> Thanks, plently of stuff to try next now18:42
T42<k1gen> should battery icon in status bar always be with a lightning, even if the device isn't charging?18:42
T42<edp_17> Yes. When the device is charging at the left side of that icon, there is a dash (line/cable).18:43
T42<k1gen> got it, thanks18:43
T42<elros34> @b100dian I think only that dummy controller with disabled gsi is worth to try18:51
T42<b100dian> Yes, that's what I tried first, but no real difference from the second log19:07
T42<b100dian> Next I try to disable these since just disabling IPA doesn't work
T42<k1gen> I guess all that's left are cameras and sound, with them I can comfortably use the device19:19
T42<b100dian> @k1gen what device do you have?19:20
T42<k1gen> Google Pixel 419:20
T42<k1gen> oh, and bluetooth would be nice, as device has no 3.5mm jack19:20
malhow does bluetooth fail? do you have bluebinder installed?19:24
T42<elros34> @b100dian just to be sure do you confirm your changes in .config?19:25
T42<b100dian> @elros34 I set CONFIG_USB_DUMMY_HCD=n and in .config is showing as `# CONFIG_USB_DUMMY_HCD is not set`19:27
T42<k1gen> mal: just installed it. bluetooth doesn't "fail", it just doesn't find any devices in settings19:28
malhmm, check journal log?19:28
malalso check that you have these the BNEP ones are not critical though19:29
T42<k1gen> nothing suspicious in journal19:29
T42<k1gen> I have neither CONFIG_BT_HIDP nor CONFIG_BT_HCIVHCI in defconfig, no BNEP too19:31
malCONFIG_BT_HCIVHCI is required for bluebinder to work, the other are not critical I think but you can enable those while at it19:33
malI think I should add that to mer kernel checker19:33
T42<b100dian> +1 for that19:34
T42<k1gen> mal: in tg history I found you helping edp_17 with bluetooth, should I add all keys you suggested to him?19:34
T42<elros34> @b100dian would be good to get dmesg after few seconds delay so there is chance to catch potential errors19:34
T42<b100dian> Good idea indeed19:36
mal@k1gen well what did I tell him to add? :)19:37
T42<k1gen> mal: CONFIG_BT_BREDR=y19:38
T42<k1gen> CONFIG_BT_BNEP=y19:38
T42<k1gen> CONFIG_BT_BNEP_MC_FILTER=y19:38
T42<k1gen> CONFIG_BT_HCIVHCI=y19:38
T42<k1gen> CONFIG_BT_HIDP=y19:38
T42<k1gen> CONFIG_BT_RFCOMM=y19:38
T42<k1gen> any more .config values I should add before building hybris-hal?19:38
T42<k1gen> sound-related, for example19:42
T42<b100dian> @elros34 unfortunately nothing more than [sda] Synchronizing SCSI cache after 15 seconds (which is the `sync` after the dmesg I added)19:44
T42<elros34> have you tried connecting usb later, maybe there is no usb detection for some reason19:48
T42<elros34> like boot without usb connected and connect it after few seconds19:48
malthose should be all fine19:50
T42<elros34> @b100dian  you have quite recent kernel but do you have /sys/class/android_usb19:51
mal@k1gen that list looks ok for bluetooth, can't think of anything else to enable now19:52
T42<k1gen> mal: got it, thanks. building19:52
T42<elros34> @b100dian I assume you grep for anything usb related in init*rc files for unusual initialization19:54
T42<b100dian> I do have /sys/class/android_usb in recovery, but for some reason the -d test in init fails.19:56
T42<b100dian> I'll try to grep more rc files, maybe gadget or android_usb @elros3419:57
T42<b100dian> Just looking for /vendor *.rc files, apart from the init.qcom.usb I shared earlier, grepping for usb|gadget|android_usb shows almost nothing - other than setprops :(20:10
T42<b100dian> >     chmod 0444 /sys/devices/platform/msm_hsusb/gadget/usb_state20:10
T42<b100dian> Ouch, I accidentally booted Lineage:) That wiped my /data. For some fskcing reason that is also using dm-style partition. I wiped Lineage's, it wipe mine's:)20:26
T42<b100dian> This recovery-on-boot is actually part of the problem. To maintain access to recovery, I kept the lineage boot20:27
malI made a PR to add those missing checks to mer-kernel-check
T42<k1gen> mal: yeah, about that: on kernel with those .config values defined - still no devices found in settings22:17
T42<k1gen> nvm, just had to reboot one more time. thanks for help!22:20
T42<k1gen> now I can connect my headphones, but there's still no sound, for example Fernschreiber freezes if I try to play a video or listen to voice message22:27
malis pulseaudio running?22:27
T42<k1gen> inactive (dead) and can't start22:29
malso check journal log for error22:29
malnote that pulseaudio is in user session i.e. check with systemctl --user22:30
T42<k1gen> debugging with fingerterm is painful, I'll work on this in the morning. thank you for helping with bluetooth, I hope we'll fix pulseaudio soon22:41
T42<b100dian> didn't you have wireless to use ssh? or are you debuggin from the bar;)22:45
malfixing wlan is a good idea if usb is not working22:46
maland a good idea anyway22:46
T42<b100dian> Well that hit home for me:) Do you know if anyone trained wpa-supplicant offline to connect automatically to a wifi at boot, with ssh enabled?22:47
T42<k1gen> I have wlan, it's just a local blackout and since my laptop battery is dead all I can use to ssh into pixel is either iphone I use until I get pixel functional enough and pixel itself with fingerterm22:49
T42<k1gen> *or pixel22:49
T42<k1gen> it's a pain either way :)22:53
T42<k1gen> @b100dian I have this by default if you mean autoconnect to wifi with sshd running22:55
T42<b100dian> on first boot:)?22:55
T42<k1gen> oh22:55

Generated by 2.17.1 by Marius Gedminas - find it at!