Friday, 2023-03-24

T42<k1gen> mal: systemd says dmsetup.service doesn't exist, but I can find unit file at /usr/lib/systemd/system/sparse/usr/lib/systemd/system/dmsetup.service07:06
T42<k1gen> I also fixed permissions, nut it was in the right place07:09
T42<k1gen> *but07:09
T42<k1gen> should I move it from /usr/lib/systemd/system/sparse/usr/lib/systemd/system/ to /usr/lib/systemd/system/ and then enable?07:23
T42<k1gen> anyone?08:05
T42<parmjotsinghrobot> @k1gen I think for this you should just follow systemd guidelines, which I'd assume would be the second one08:47
T42<parmjotsinghrobot> Considering my limited experience with systemd08:47
T42<k1gen> "second one"? move to /usr/lib/systemd/system and enable?08:51
T42<parmjotsinghrobot> Yea09:00
T42<k1gen> @parmjotsinghrobot thanks for suggesting, I think it worked. can someone check my jounalctl: ?09:06
T42<elros34> it's unreadable with not disabled audit. Did you check whether service you created is started and partitions are mounted: findmnt command?09:11
T42<k1gen> @elros34 I ran ls -l /dev/mapper, all dynamic partitions are mounted automatically09:13
T42<elros34> /dev/mapper dosn't tell you whether they are mounted09:14
kcrootthis step -> " echo 'mount_sdk() { sudo mount -o rbind "$sdkroot/srv" "$sdkroot/parentroot/srv"; }' >> ~/.mersdkrc "09:14
kcrootis needed for SDK v ?09:14
kcrooti need to setup SDK in other place than /srv/sailfish09:15
T42<k1gen> @elros34 for audit: should I add BOARD_KERNEL_CMDLINE += audit=0 to and rebuild hybris-hal in ubu-chroot and flash hybris-boot.img? is that all or I forgot something?09:18
T42<k1gen> findmnt output:
mal@k1gen /usr/lib/systemd/system/ is the correct location of course09:19
T42<k1gen> mal: seeing my findmnt, do you think we're done with dynparts?09:22
T42<elros34> compare against your fstab09:23
T42<k1gen> by the way, I still need a way to turn off the display over ssh, until I get gui working09:23
T42<elros34> and you need it because?09:23
malseems ok09:23
malhmm, is there system_root.mount?09:24
T42<k1gen> @elros34 because device has a four-year-old POLED panel09:24
malI mean does /system contain something now? is there /system_root ?09:24
T42<k1gen> mal: /system is empty09:25
T42<k1gen> also, no mention of system_root anywhere for my device09:25
malso you didn't copy the mount for that from my repo?09:26
T42<k1gen> no, not system_root. I copied only {system.system_ext,vendor?product}.mount and dmsetup.*09:27
T42<k1gen> *vendor,product09:27
T42<elros34> check whether you have brightness in /sys/class/backlight/09:30
T42<k1gen> @elros34 yes, panel0-backlight09:31
mal@k1gen replace the system.mount on your device with except use /system in there instead of /system_root, looking at your fstab it mounts /system normally not as root
T42<elros34> see what value it has and try to write 1 or 0 to it09:31
T42<k1gen> echo 0/1 > /sys/class/backlight/panel0-backlight/brightness changes nothing09:31
T42<elros34> any other related files?09:32
T42<elros34> maybe check bl_power09:33
T42<k1gen> nope, doesn't work09:33
malsometimes backlight can be in the leds subsystem also09:34
T42<k1gen> led:flash_0/   led:flash_1/   led:switch_0/  led:switch_1/  led:switch_2/  led:torch_0/   led:torch_1/   vibrator/09:34
maldoesn't seem like it's there09:35
T42<elros34> grep for brightness in /vendor/etc os some imilar path  init*rc files09:35
T42<k1gen> /vendor/etc/init/hw/init.sm8150.rc:265:    chown system system /sys/class/backlight/panel0-backlight/brightness09:36
T42<k1gen> mal: so, just rename my system.mount to system_root.mount and replace Before= in
T42<elros34> are you sure you echo it correctly? maybe it's dummy for oled09:37
T42<k1gen> I don't understand what do you mean. I do echo 0 > /sys/class/backlight/panel0-backlight/brightness, cat returns 0, display stays on09:39
T42<k1gen> same with 1 instead of 009:39
T42<k1gen> mal: there's my system.mount: , and there's system_root.mount: which one should I leave?09:42
T42<elros34> you need system.mount but with system_root.content and mountpoint (Where=) set to /system09:47
T42<k1gen> thanks, let me try09:48
T42<k1gen> like that: ?09:49
T42<elros34> if that is system.mount then yes (don't remove command prompt)09:51
T42<k1gen> I'm sorry, I usually just do ix (pastebin cli tool) <name of file>09:52
T42<k1gen> I will use more often to include prompt09:52
T42<k1gen> /system is populated now:
T42<k1gen> now I need to disable audit so we can read journalctl. I have to repeat my recent message: for audit: should I add BOARD_KERNEL_CMDLINE += audit=0 to and rebuild hybris-hal in ubu-chroot and flash hybris-boot.img? is that all or I forgot something?09:59
T42<elros34> yes. What is in your /system/system?09:59
T42<elros34> so that system_root change was all wrong10:01
T42<elros34> you have system-as-root device10:01
T42<k1gen> what's that?10:01
T42<elros34> you have / in /system and /system in system/system10:02
T42<k1gen> what can I do about it? can I ask for detailed instructions?10:03
T42<elros34> intruction is same just you latest mofications are wrong. It' didn't work for the first time probably because some other mistake like missing symlink10:04
T42<k1gen> can you please explain like that: you need system.mount but with system_root.content and mountpoint (Where=) set to /system ? it is both short and very helpful10:05
T42<elros34> mal latest advice and my explanations how to change system.mount where wrong just follow initial instruction: system_root<=>system10:07
T42<k1gen> @elros34 like that?10:13
T42<TheVancedGamer> does anybody know why I get this error during `mka hybris-hal droidmedia`?10:28
T42<TheVancedGamer> ```10:28
T42<TheVancedGamer> build/make/core/ error: RSUnbundledTests: Unknown installed file for module 'RSTestBackward'```10:28
T42<TheVancedGamer> oh, and am I supposed to apply hybris-patches at this point?10:30
T42<elros34> before first make hybris-hal10:30
T42<TheVancedGamer> gotcha10:31
T42<TheVancedGamer> still getting the same error after applying patches10:32
T42<k1gen> elros, is everything fine with my system{,_root}.mount units?10:34
T42<elros34> clean up /out. If this will not help then search whether this Test* are defined in your device/*10:34
T42<TheVancedGamer> I checked out the defines in device trees, nothing related there10:34
T42<TheVancedGamer> i'll clean out10:35
T42<elros34> @k1gen isn't it copy from fp4 repo? Then it should be fine. You are not newbe anymore to gives you step by step confirmation, try yourself10:36
T42<k1gen> alright. what am I looking for after reboot?10:36
T42<elros34> findmnt and content in /system10:37
T42<k1gen> ok, thanks10:37
T42<elros34> if that is correct then droid-hal-init in logs, maybe mce will work and your display should dimm10:38
mal@k1gen I said replace the content, no renaming of system.mount11:10
T42<k1gen> mounts are fine now, what was in /system before is now in /, and /system/system is in /system now11:11
mal@elros34 odd why fstab had system as /system11:12
T42<k1gen> gotta go for groceries while hybris-hal with audit=0 is rebuiding11:12
T42<TheVancedGamer> @elros34 didn't work, still same issue11:13
T42<elros34> mal: it's not that uncommon even for regular non dynamic partitions11:23
malok, I didn't know that11:26
T42<TheVancedGamer> still getting this, grep reveals nothing, no idea what this even does (re @TheVancedGamer: does anybody know wh...)11:45
T42<TheVancedGamer> disabling renderscript to me doesn't seem like the smartest choice11:45
malwhich android base?11:46
T42<TheVancedGamer> 18.111:46
T42<TheVancedGamer> I did search chat history, hadk-faq and hadk-hot11:46
T42<elros34> I think I saw this error but no idea what and where was this. Can you show your device repo?11:55
T42<elros34> be aware is case sensitive search engine11:56
T42<TheVancedGamer> kernel is at (branch is halium-10.0-test)11:56
T42<TheVancedGamer> i've already ported UT so kernel isn't really necessary, but just in case11:56
T42<TheVancedGamer> I don't know what I did but i'm already farther :P12:03
T42<TheVancedGamer> ah, adding `android.hardware.renderscript@1.0-impl` fixes it12:03
T42<k1gen> read father first time, lmao12:25
T42<k1gen> @elros34 I flashed kernel with audit=0, now I have droid-hal-init: Unable to move mount at '/data': Invalid argument in jounalctl:
T42<adampigg> I can take a PR ;) (re @elros34: be aware https://pig...)12:44
T42<elros34> it's there already for quite a time but not sure if correct12:44
mal@k1gen can you show kernel commandline?12:45
T42<k1gen> sure, a minute12:45
T42<k1gen> mal:
T42<elros34> with permissive selinux on hybris-18 you may need custom hw/vnd/servicemanager.rc with selinux stubs12:48
malis seems that androidboot.force_normal_boot=1 is causing odd things, wondering if this is the first time this happens12:49
T42<k1gen> @elros34 which one of these rc's and what are selinux stubs and where do I get them?12:51
T42<elros34> it's in hadk-hot hybris-18 section but it's guess. First you need to fix issue mal is talking about:  droid-hal-init12:53
mal@elros34 maybe we should comment out that code in init, I think bootloader might be adding the parameter to commandline12:53
T42<k1gen> what should I do to make droid-hal-init work?12:55
T42<k1gen> I sent journalctl earlier, you said it's unreadable without audit=0, and now that I've booted with audit=0, I don't know where to look in "readable" journalctl12:56
T42<elros34> like I said in droid-hal-init, you see it failed right after starting12:57
T42<k1gen> yeah, with status=6/NOTCONFIGURED12:57
T42<elros34> yeah but error is earlier and mal gives you tip immediately: force_normal_boot. If you search it in google then you can see this part android source code12:58
T42<k1gen> can I reboot without androidboot.force_normal_boot=1?13:00
T42<k1gen> Devices that use recovery as a ramdisk must use the kernel command line parameter androidboot.force_normal_boot=1 to decide whether to boot into Android or continue booting into recovery.13:00
T42<k1gen> source:
T42<elros34> only if it's not added by bootloader. Other/better solution would be commenting out code which check that probably in system/core13:02
malI didn't find that in device repo, maybe grepping is needed to be sure13:06
T42<k1gen> there is a couple lines of code in system/core/init/first_stage_init.cpp: bool ForceNormalBoot(const std::string& cmdline) {return cmdline.find("androidboot.force_normal_boot=1") != std::string::npos;}13:06
T42<k1gen> on lines 99-101:
T42<elros34> do you know c or some other launguages little?13:07
T42<k1gen> @elros34 little. what do I have to do?13:07
T42<TheVancedGamer> just make it return 1 ig?13:08
T42<b100dian> I think return false13:08
T42<b100dian> since next it will chroot13:08
T42<TheVancedGamer> yeah false is better, I keep forgetting whether 0 or 1 is true or false13:08
T42<b100dian> I was thinking - isn't that parameter an "opportunity" to package the lineage recovery and exec that from out init when force_normal_boot is missing13:09
T42<b100dian> *our13:09
malI think it would be better to just comment out that whole "if (ForceNormalBoot(cmdline)) {" part13:09
malwe won't be needing it anyway13:10
T42<k1gen> mal: just comment out this ?13:13
malyes, the whole if13:14
T42<b100dian> Yes, that ForceNormalBoot is also used above, for example to make GetModuleLoadList look at another module file. Better to comment 279-28713:14
T42<k1gen> what should I build to check?13:15
malhybris-hal, you can copy just the init to droid-hal-init on device for testing13:17
T42<k1gen> mal: replace /sbin/droid-hal-init on device with what from out/?13:24
T42<elros34> I think it should be clear what after make hybris-hal13:25
T42<k1gen> scp out/target/product/flame/system/bin/init device:/sbin/droid-hal-init ?13:41
malyes after you have built the new version13:43
T42<k1gen> thank you!13:43
T42<k1gen> what should I do about missing /usr/bin/droid/
malthat is optional, not a problem13:44
malthat file should be added only if device needs some special init, most don't need13:44
T42<k1gen> ~20min till hybris-hal builds. can we solve anything else while you are here? :)13:45
T42<elros34> really that long? did you add -j4 or more to make?13:46
T42<k1gen> well usually make -j$(nproc --all) hybris-hal takes about 45min and up to 30-35Gb of ram13:47
T42<elros34> but you are not building everything now13:50
malare you using ccache?13:50
T42<k1gen> sure. 99% of that time and ram is for linking kernel13:51
T42<k1gen> #### build completed successfully (39:47 (mm:ss)) ####13:58
T42<k1gen> oh my god, I can turn off the screen14:01
T42<TheVancedGamer> how 35GB? i'm building hybris-hal with 1614:01
T42<k1gen> that's a gorgeous charging icon!14:02
T42<k1gen> it stays on screen even if I stop charging the device though14:02
T42<k1gen> no rndis, so no ssh/telnet14:04
T42<k1gen> gui and touchscreen works!14:06
T42<k1gen> everything is really small though, that's because of low pixel_ratio, right?14:06
T42<k1gen> setup detects two instances of each AP and stays in Connecting loop14:09
T42<k1gen> thank you @elros34 and mal for helping me all this time! all I have to do now is debug a few small issues, but I have gui! I never thought I'd get to it knowing myself :)14:13
malsize issue is pixel_ratio, check the instructions for it, usually on modern devices it's about 1.75 or 2, but depends on device14:59
T42<k1gen> after some testing I found out that cameras, gyroscope (screen orientation sensor) and bluetooth are not working. also, I don't have pixel_ratio formula for 5.7" 1080x2280 19:9 (~444 ppi) display, so all icons and text is too little15:00
T42<k1gen> mal: I started texting a bit earlier :)15:00
T42<k1gen> can I test pixel_ratio without reflashing rootfs?15:02
T42<k1gen> dconf write 1.75 /desktop/sailfish/silica/theme_pixel_ratio ?15:04
malcheck faq15:28
mallooks like that part of faq needs some formatting fixes but I think you can find out where commands start and end15:29
T42<k1gen> I got it on my own :)15:30
T42<k1gen> how do I calculate scale for resizing icons in mogrify -resize <scale> /usr/share/themes/sailfish-default/meegotouch/z1.75/icons/*.png ?16:09
T42<k1gen> because now I have normal scaling and pixel_ratio, but 7 icons in a row16:12
T42<edp_17> What's your device resolution?16:12
T42<k1gen> @edp_17 1080x2280. I took that mogrify part from your script btw (, thanks for it16:13
T42<edp_17> My Note 4 (1440x2560) uses pixel_ratio 216:17
T42<edp_17> and my motoG7 (720x1570) uses pixel_ratio 1.2516:17
T42<edp_17> I guess you should be fine with pixel_ratio 2.16:17
T42<XAP2P> i think pixel_ratio depends on your resolution and display size16:18
T42<edp_17> Yeah, you can modify that script and hard-code values to play with that.16:18
T42<XAP2P> It seems to me that in the ratio of examples it is possible to calculate how it will be better and it will turn out perfect16:18
T42<k1gen> nah, I'm fine with 1.75, I don't want stuff to get bigger16:19
T42<edp_17> Okay, it's up to you. :)16:20
T42<k1gen> between two of your examples, my 1080x2280 is made for 1.7516:20
T42<k1gen> anyway, can you answer my initial question?16:20
T42<k1gen> about mogrify scale16:20
T42<edp_17> I can't. On my old S2, I just played with different values, then stick with the one looked the best.16:22
T42<k1gen> well there's awk "BEGIN {print $1*100/2.0}" in the script, but it produces value of 87.516:23
T42<k1gen> I don't think that's what I need16:24
T42<k1gen> I ran mogrify -resize 175 /usr/share/themes/sailfish-default/meegotouch/z1.75/icons/*.png and after reboot nothing changed, same 7 icons in a row16:30
T42<TheVancedGamer> getting this while building droid-hal16:34
T42<TheVancedGamer> I don't think .mb2/spec missing is fatal?16:35
T42<elros34> error is probably fatal16:35
T42<TheVancedGamer> is it because I installed sdk tooling and targets in home dir?16:36
T42<elros34> could be, where did you install sdk itself?16:37
T42<TheVancedGamer> /srv/sailfishos16:40
T42<elros34> I do not remember exactly whether its sdk or sdk target requirement but did you remove nosuid nodev from home mount?16:41
T42<TheVancedGamer> from where?16:45
T42<TheVancedGamer> it wasn't in hadk16:45
T42<TheVancedGamer> I installed sdk target and tooling with sdk-assistant if that changes anything16:46
T42<elros34> I am not sure I follow so what exactly you changed?16:46
T42<elros34> do you follow hadk or make some modyfication16:46
T42<TheVancedGamer> follow hadk16:52
T42<TheVancedGamer> I didn't change anything at all16:52
T42<elros34> so this was related to "is it because I installed sdk tooling and targets in home dir?"?16:53
T42<elros34> check whether you have env set up: 'env' command16:54
T42<TheVancedGamer> ok17:05
T42<TheVancedGamer> @elros34 my env
T42<TheVancedGamer> also, I can access the target sdk, I don't think its that17:10
T42<elros34> so you say no modification and still you use /hdd/hadk  instead home17:12
T42<TheVancedGamer> /hdd/hadk is android root17:12
T42<elros34> outside of home17:13
T42<TheVancedGamer> I mean, is it necessary for android root to be in home?17:13
T42<elros34> IIRC if it's in another partition then you must bind mount it but I do not remember details17:15
T42<TheVancedGamer> I brought over that dir from /parentroot if that affects anything17:16
T42<elros34> not sure what do you mean show some files. I use android root on another disk which is mounted inside home and with bind mounts for sdks that works17:26
T42<k1gen> alright, today we got pixel to gui, now that I have a headache I'll leave the device till tomorrow. I will need help with getting icons to resize, sound, bluetooth, gyroscope and battery sensor to work17:29
T42<TheVancedGamer> @elros34
T42<elros34> so how do you handle bind mount? Maybe try to mount that dnroid sources partition inside home17:43
T42<TheVancedGamer> `sudo mount --bind /parentroot/hdd /hdd`17:43
T42<TheVancedGamer> ok17:43
T42<TheVancedGamer> now build_packages wants /hdd/hadk, huh17:45
T42<elros34> did you change ANDROID_ROOT?17:45
T42<TheVancedGamer> right, I didn't17:46
T42<TheVancedGamer> thanks17:46
T42<TheVancedGamer> definitely farther than before17:46
T42<TheVancedGamer> getting a config not set error while it is set in .config17:51
T42<TheVancedGamer> log is filled with config not set errors, wtf17:52
mal@k1gen something to consider for camera issues you noticed, I suspect you don't have the mini* services running, first of course make sure you have droidmedia and gst-droid installed, if you still have failing mini* services then have a look at lib and lib64 stuff here those symlinks inside those folder might need adjusting so18:51
malfind where the files are on your device and add needed symlinks18:51
T42<k1gen> thanks for helping, I'll do it tomorrow in the morning18:52
T42<k1gen> mal: can I eventually get fully functional device, how do you think?18:53
piggzmal: whats this rude file which broke lipstick on my pinephon :) /usr/lib/systemd/user/lipstick.service.d/50-compositor.conf19:12
piggzsome new file in config packge19:12
malpiggz: so use delete_file.list19:39
piggzmal: i have :)19:39
malpiggz: another reason to finally make the mainline port submodule19:40
malwhich we talked about long time ago19:40
piggzmal: currently trying to merge in PA config changes into native version ....19:41
piggzwhat i was was preventing pa loading becuase the bluez5 stream restore file is no longer present19:41
mal@k1gen it's doable usually19:41
T42<k1gen> nice to hear :D19:42
malsometimes some things take time but difficult to say how easy some device is19:42
T42<b100dian> Hello, I got my first init.log in real rootfs with no telnet running:
T42<b100dian> (Let me know if you prefer another paste service)20:41
T42<b100dian> I did try to change c.1 to b.1 and I tried with minimal kernel changes like in but also with the hybris-18.1 branch which is the one passing the kernel checker20:45
T42<elros34> change 'echo "$(ls /sys/class/udc)" > $GADGET_DIR/g1/UDC' to 'echo a600000.dwc3 > $GADGET_DIR/g1/UDC'20:53
T42<elros34> btw did you completely disable selinux for some reason?20:57
T42<b100dian> I appended androidboot.selinux=permissive. I just noticed where the cmdline is in the log - first look tells me it's concatenated (boot, vendor_boot)21:05
T42<b100dian> thanks for noticing that "ls" output! I do see usb0 and telenetd in the log now. I may need to switch cables I don't see them on the host yet21:06
T42<elros34> I think upstream fix could be simple like 'ls /sys/class/udc | grep -v dummy'21:08
T42<b100dian> Yes, that's the change I'm doing locally, maybe it will end up in a PR21:20
T42<elros34> you should check kernel config whether you have selinux enabled there.21:24
T42<b100dian> what did you see suspicious about selinux?21:41
T42<elros34> the point is I do not see selinux initialization21:42
T42<elros34> usual one21:43
T42<b100dian> I see.. should be around the 0th second "SELinux:  Initializing"..21:48
T42<elros34> yeah, maybe zcat /proc/config.gz shows some points21:49
T42<b100dian> After Yama: becoming mindful :) I have "AppArmor: AppArmor initialized"21:50
T42<b100dian> I don't have yet telnet access. I may use extract-ikconfig.21:53
T42<b100dian> I see /init-debug: line 281: can't create /etc/udhcpd.conf: nonexistent directory and in rootfs there's a link /etc/udhcpd.conf -> /run/usb-moded/udhcpd.conf21:53
T42<elros34> did you create /init_enter_debug2?21:54
T42<b100dian> did that now. Is "vomitbin" better for the eyes?
T42<elros34> sure. So rndisk still doesn't work, did you enable all these usb configs linked here few days ago?22:10
T42<elros34> hmm but there is usb022:12
T42<b100dian> yes, that's weird. and that appeared after I fixed the dummy udc writing22:14
T42<elros34> so nothing in host dmesg?22:14
T42<b100dian> no. I just checked with my other phone and it is detected as rndis_host22:25
T42<b100dian> I do have the three usb configs linked a few days ago. I extracted this init.qcom.usb.rc from the vendor partition, where is "gsi":
T42<elros34> can you show your .config23:38
T42<b100dian> Of course. .config is
T42<elros34> try changing rndis_bam.rndis to gsi.rndis in init script23:46
T42<b100dian> and c.1 to b.1 right?23:46
T42<elros34> I am not sure this is relevant. If that doesn't change anything maybe disable  CONFIG_USB_CONFIGFS_F_GSI. You should also include dmesg, maybe there will be something and of course always check host dmesg for any message23:48
T42<elros34> I think CONFIG_USB_CONFIGFS_ECM  could be also an option23:53
T42<b100dian> I also see a lot of modules in the config, I just remembered they probably can't load if they are searched for in vendor_boot cause I dont' yet build that. But nothing rndis related though23:56

Generated by 2.17.1 by Marius Gedminas - find it at!