Tuesday, 2021-03-02

T42<T %lastname%> @hacker12455 does Cedric boot hybris 16?03:41
riniguspiggz: saw your post regarding OBS at forum. Jolla is very quiet regarding it ...06:48
rinigusstill no target for
T42<hacker12455> @T %lastname% [@hacker12455 does Cedric boot hybris 16?], did not check06:58
T42<hacker12455> @elros34 [@hacker12455 by 'no sign' you mean there is no …], nothing at all from usb06:58
T42<hacker12455> hybris-boot was flashed06:58
T42<hacker12455> otherwise the device would boot to Lineage06:58
T42<edp_17> @swanux [I can get telnet (with some forced connect, an …], Have you checked mount moints?07:35
T42<edp_17> @swanux : On one of my port with older kernel I also stuck on the boot logo. I needed to apply one or both of these:07:37
T42<edp_17> fs/statfs.c - https://github.com/shr-distribution/linux/commit/5187f535beeb3731677c6f6d05402f9adec6e03407:37
T42<edp_17> fs/stat.c - https://github.com/shr-distribution/linux/commit/01141a353241d545a2aaf9addaa32afe39a9856207:37
T42<edp_17> Then the device booted up.07:37
T42<swanux> @edp_17 [Have you checked mount moints?], Yeah, mount points should be fine (added device in fixup-mountpoints)07:53
T42<swanux> @edp_17 Thank you for the advice. You used it with 16.0 base too?07:54
T42<edp_17> @swanux : No, I only used cm12.1 and los14.1. But I think worth to check with 16.0 too.08:48
T42<hacker12455> hm, how would I check if hybris-boot boots properly09:14
T42<hacker12455> if I get no usb09:14
T42<edp_17> @hacker12455 : You can manually bring up usb. Example here: https://paste.ubuntu.com/p/vdM9CKgQzT/09:24
T42<edp_17> More info here: https://docs.halium.org/en/latest/porting/debug-build/early-init.html09:24
T42<hacker12455> the problem is that no interface appears09:48
T42<hacker12455> the device just stays on the OEM logo09:48
T42<hacker12455> I did check if config_g_android is enabled09:48
T42<hacker12455> it is09:48
T42<edp_17> Plug the device into your PC and on you host, what the output of: dmesg | tail09:49
T42<Verevka86> @hacker12455 [I did check if config_g_android is enabled], what is your kernel version?  can I see the source?09:53
T42<hacker12455> https://github.com/LineageOS/android_kernel_motorola_msm895309:54
T42<hacker12455> my kernel version is 3.18.12409:55
T42<hacker12455> @edp_17 there is no output09:57
T42<hacker12455> after the twrp disconnect message there is no other output09:58
T42<hacker12455> granted I am not testing it on my own phone09:58
T42<hacker12455> just sending the images to testers09:58
T42<edp_17> This way, it is very hard to support.10:05
T42<Verevka86> @hacker12455 [my kernel version is 3.18.124], no, unfortunately this is not the same case as mine 😔10:05
T42<hacker12455> @edp_17 yes10:05
T42<edp_17> When I asked for the out put for that command I meant, boot the phone into sfos leave it on the boot logo and connect to the pc and issue the command there. (If the phone is in twrp that won't help.)10:06
T42<elros34> @hacker12455 hybris boot initramfs will reboot device after 1 minute if it can't setup usb connectios. Did you disable skip_initramfs handling in your kernel?10:54
T42<hacker12455> Don't recall disabling anything like that10:54
T42<elros34> btw hybris17 is not polished yet so it wil be very hard to finish it without having access to device10:55
T42<elros34> @hacker12455 in kernel init/initramfs find commit which addes skip_Initramfs support and revert it10:56
T42<hacker12455> will do that later and report results10:57
T42<hacker12455> this?11:11
T42<hacker12455> https://github.com/LineageOS/android_kernel_motorola_msm8953/commit/f73aa0eab7d955d782cf623b8d358af4dbaf5b7111:11
T42<elros34> yes but there is also usually one more commit which free memory related to that which you need revert first11:25
T42<hacker12455> btw11:36
T42<hacker12455> does anyone else have a problem with 2GB RAM devices and extracting the rootfs tar archive?11:37
T42<hacker12455> on cedric 3GB RAM versions extract it fine11:37
T42<hacker12455> on 2GB RAM tar dies before it fully extracts11:37
T42<hacker12455> https://github.com/LineageOS/android_kernel_motorola_msm8953/commit/c2b30b43e0182209f197d8b597f08551a2d6db4d11:38
T42<hacker12455> that commit?11:38
T42<elros34> yes. OOM issue in twrp it's known bug in tar. You can use older twrp or busybox-static tar from sailfish out/11:41
T42<hacker12455> where exactly is that built tar?11:44
T42<elros34> I don't have access to that now but busybox used in initramfs should be somewhere in $ANDROID_ROOT/out/*/hybris-boot*. I think you can also download busybox-static in platform sdk target. Be aware I did not try it11:48
T42<elros34> or unpack it from sfos rootfs/usr/bin/busybox-static11:49
T42<hacker12455> think I reverted everything I needed now12:02
T42<hacker12455> thing is we don't even get the LED as a sign of life12:06
T42<hacker12455> it just stays on splash12:06
T42<hacker12455> also that tar from busybox-static worked12:06
T42<hacker12455> think I might switch to hybris 16.0 if that does not work12:07
T42<elros34> you will not get LED on early boot only somwhere when gui starts12:09
T42<hacker12455> what would be the most stable base atm?12:10
T42<elros34> the one for which android works best on your device except hybris 17 aarch64 which is not peace of cake for now12:11
T42<elros34> piece*12:11
T42<hacker12455> then I might switch to 16.012:12
T42<hacker12455> how would I do it?12:13
T42<hacker12455> preferably without nuking everything12:14
T42<elros34> different android base, different sdk target arch.. I do not think is possible without starting from scratch. Maybe except platform sdk and droid repos12:16
T42<hacker12455> so I need to set armv7 on 16.0?12:22
T42<elros34> not sure if anybody have tried aarch64 on hybris-16.0, maybe mal12:25
mal@elros34 I had almost everything working on aarch64 hybris-16.012:30
malsome small device specific issues with fingerprint and graphics (some strange mode was enabled on the device)12:31
malpeople seem to give up way too quickly12:31
T42<hacker12455> would it be better to use armv7hl or aarch6412:34
piggzlbt: wee reminder to bring 4.0.1 to obs :)12:42
malarmv7hl vs aarch64 should not matter much during porting13:06
swanuxSince yesterday I have droid-hal working, but still no GUI (stuck at oem logo). It seems that lipstick fails.16:03
swanuxHere's a journal from the boot, trying to start lipstick: https://paste.ee/p/f3nzv16:03
swanuxHere's the strace of hwcomposer test: https://paste.ee/p/N21Wa16:03
swanuxAnd strace of running lipstick manually as root: https://paste.ee/p/wuywN16:03
swanuxTried minimer, ouputs this: (same as lipstick) https://paste.ee/p/oErup16:03
swanuxmal, @elros34, or anyone, any ideas?16:03
piggzswanux: id look, but it seems past.ee is blocked by opendns, and im too lazy to change my dns settings :D16:05
swanuxpiggz: Same order then:16:07
maljust to be sure, did you export the needed environmental variables before starting lipstick and other things?16:14
swanuxYes, I did16:15
piggzswanux: what about logcat logs?16:18
maldoes surfaceflinger work?16:20
swanuxpiggz: Here's logcat (from boot to running lipstick and core dumped): https://pastebin.ubuntu.com/p/ntpGbj6FfQ/16:24
swanuxmal: How do I check that?16:24
malrun /system/bin/surfaceflinger16:25
swanuxmal It shows lineage boot animation, nothing else changed, no errors.16:26
malok, now try again other ones16:32
swanuxOther ones?16:34
maltest_hwcomposer and lipstick16:36
malof course16:36
swanuxOk. Is it okay with logcat?16:37
swanuxmal Here's the logcat log after boot > run lipstick > run hwcomposer: https://pastebin.ubuntu.com/p/JMwnXp3Mfw/ (The last few lines are repeating infinitely)16:49
malswanux: can you check if you have /dev/mdss_rotator and what user and group it has?16:55
swanuxmal: Here's the ouput of mount and trying to acces to it. As I can see I don't have it. https://pastebin.ubuntu.com/p/zDDTpn5kRN/16:57
T42<elros34> could be also msm_rotator16:58
malnow a terrible hack, open /var/lib/environment/compositor/droid-hal-device.conf , do you have the width and height set there?16:59
malswanux: ^17:00
swanux@elros34 I neither have that17:01
swanuxmal Checking now17:01
swanuxNope, I don't have17:01
malok, add these lines there17:02
malor you can just export those values before starting lipstick manually17:03
malswanux: maybe export those manually first for testing17:03
swanuxmal Ok, so export them an run lipstick manually again?17:04
swanuxNothing changed17:05
maltry changing height to 224817:06
swanuxStill seg fault17:06
mal@elros34 the messages in logcat "SDM     : DRMAtomicReq::Validate: drmModeAtomicCommit failed with error 22" suggest a similar issue I had on fp3 with hybris-16.0 base, there was some strange graphics thing happening which caused a buffer size mismatch in kernel17:08
malswanux: do you see anything in dmesg?17:08
swanuxmal: Here's my dmesg: https://pastebin.ubuntu.com/p/t4phRHhgTD/ I can only see some write errors17:10
malswanux: can you show output of "ls -l /dev/"17:11
swanuxmal: Sure : https://pastebin.ubuntu.com/p/QhRNWqMZhx/17:12
T42<elros34> so you have droid-hal-init messages in dmesg only, that is what we were missing previously when debugging17:15
T42<elros34> maybe not only...17:17
T42<elros34> @swanux so if that is same issue as mentioned by mal on fp3, maybe you could try to compare logcat output when you run surfacefliner and minimer.17:28
mal@elros34 my hack for the issue was to guess some values for the width and height17:33
T42<elros34> I was wondering whether logcat output when starting surfaceflinger could help getting/guessing those values17:46
swanuxFrom the logs it seems to be 1080*2248, but it didn't work out17:48
T42<elros34> that is probably  default value used by hwcomposer because this is your native resolution17:51
swanuxSo it has nothing to do with my native resolution, and I should just try random numbers to make it work?17:57
malswanux: on my device I had to increase the sizes a bit so there is space for some metadata buffer layers17:59
malI don't have the same issue on hybris-17.117:59
swanuxOkay then, I guess I'll try it. If no success I'll try to go with 15.1, and if still nothing I'm giving 17.1 a second shot I guess. Thanks.18:01
malchanging android base is usually not the solution18:02
swanuxI see. Then my device is simply a "black sheep"?18:07
malor maybe it's some strangeness on aarch64 hybris-16.0 devices18:13
rinigusTheDayAfter: don't think so. OBS targets are not updated => no updates for community devices.19:34
T42<hacker12455> so well19:48
T42<hacker12455> still no usb on 17.119:48
T42<hacker12455> I have no errors in the kernel config checker19:52
T42<wiktorek140> @hacker12455 [still no usb on 17.1], Early telnet available?19:58
T42<hacker12455> nope19:59
T42<hacker12455> we have nothing19:59
T42<hacker12455> nothing in dmesg19:59
T42<hacker12455> the phone just stays on the logo19:59
T42<wiktorek140> @hacker12455 [nothing in dmesg], Something in journal?19:59
T42<hacker12455> would need to check20:00
T42<wiktorek140> @hacker12455 [the phone just stays on the logo], U sure u have installed in proper place? Maybe u have encrypted data?20:00
T42<hacker12455> data is decrypted20:00
T42<wiktorek140> @hacker12455 [data is decrypted], Formatted before flashing?20:01
T42<hacker12455> yes20:01
T42<wiktorek140> @hacker12455 [yes], Mapped correctly? I remember that sometimes partition doesn't mapped correct in sfos initram20:02
T42<wiktorek140> @hacker12455 [yes], Can u send boot.img?20:03
T42<hacker12455> you want boot.img or hybris-boot.img20:03
T42<wiktorek140> @hacker12455 [you want boot.img or hybris-boot.img], Right, hybris-boot20:03
T42<hacker12455> (Document) https://irc.thaodan.de/.imgstore/Rr6zGYrgPI.null20:03
malit's certain the kernel isn't trying to skip initrd?20:04
T42<hacker12455> if you want I can send the kernel config too20:04
T42<hacker12455> I reverted these commits20:05
T42<hacker12455> so it should not skip initrd20:05
T42<wiktorek140> @hacker12455 [so it should not skip initrd], Send kernel link if have one20:05
T42<hacker12455> https://github.com/LineageOS/android_kernel_motorola_msm895320:05
T42<hacker12455> just reverted this20:06
T42<hacker12455> Fwd from hacker12455: https://github.com/LineageOS/android_kernel_motorola_msm8953/commit/f73aa0eab7d955d782cf623b8d358af4dbaf5b7120:06
T42<hacker12455> Fwd from hacker12455: https://github.com/LineageOS/android_kernel_motorola_msm8953/commit/c2b30b43e0182209f197d8b597f08551a2d6db4d20:06
T42<elros34> @hacker12455 lineage-17.1 branch? If yes then it's quite unlikely that it stays on logo. I might be wrong but based on source it has breaking commit which should bootloop your device now20:11
T42<hacker12455> yes 17.1 branch20:12
T42<hacker12455> well20:12
T42<hacker12455> it surely does not bootloop20:12
T42<elros34> Disable CONFIG_VT also show full output of make hybris-hal20:13
T42<wiktorek140> mal: @hacker12455 did fstab in intramfs should contain some entry about what to mount from block devices?20:17
T42<elros34> his kernel certainly doesn't reach initramfs20:20
T42<wiktorek140> @hacker12455 [https://github.com/LineageOS/android_kernel_mo …], Not sure if need to revert this20:20
T42<wiktorek140> @elros34 [his kernel certainly doesn't reach initramfs], @hacker12455 can u grab pstore? And did u enable pstore in hybris-boot?20:21
T42<hacker12455> pstore is enabled20:22
T42<hacker12455> hybris-hal is still building20:22
T42<wiktorek140> @hacker12455 [pstore is enabled], So grab it after force reboot from hybris-boot and send here20:23
T42<hacker12455> when you have to grab testers cuz usb in your phone is broken20:32
T42<hacker12455> 90% on hybris-hal20:32
T42<hacker12455> almost done20:36
T42<hacker12455> done20:40
T42<elros34> did you clean sources or why it takes so long?20:40
T42<hacker12455> https://paste.ubuntu.com/p/xCsv5p4bty/20:40
T42<elros34> anyway line 52 and 53 in your paste. I guess you don't exported variable mal told you to20:42
T42<hacker12455> I did20:42
T42<hacker12455> except for some reason it does not show up20:42
T42<hacker12455> https://bpa.st/LZMA20:43
T42<hacker12455> see here20:43
T42tbages was added by: tbages20:47
T42<elros34> did you export it before that first make?20:49
T42<hacker12455> iirc yes20:50
T42<elros34> so check in bash history20:54
T42<hacker12455> well right now my PC is off so yea20:56
T42<elros34> so next time change something in kernel or remove kernel from out/ and make sure you export that variable before running make20:57
T42<hacker12455> hey21:12
T42<hacker12455> after disabling vt21:12
T42<hacker12455> we now have an IP assigned21:12
T42<hacker12455> bootlooping though21:12
T42<hacker12455> guessing it's from disabled vt21:13
T42<elros34> how fast does it reboot21:13
T42<elros34> Find in kernel/drivers/tty history, commit: 'move tty_port workqueue to be a kthread' and few related to that and revert all of them21:14
T42<elros34> After that you can enable config_vt back. Looks like same breaking commit is applied on some kernels since lineage-1521:15
T42<hacker12455> reboots in around 15 seconds21:16
T42<hacker12455> will revert the commits tomorrow21:16

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!