Thursday, 2019-04-11

osum4estyea keep getting "Skipping build of hybris-updater-script since HYBRIS_BOOT_PART is not specified" do i need to fix this by changing one of the fstabs or is there a different way to fix it?00:16
r0kk3rzyou need to pick one00:50
r0kk3rzor specify both and flash boot.img to both00:50
osum4estgot it. just add one of them to the fstab then? is there an example fstab somewhere that i can look at? i'm not sure what type, mnt_flags, options, and fs_mgr_flags should be00:52
r0kk3rzcheck out the xa2 adaptation, that does a and b boot stuff00:57
T4<Hound77> … Is there something wrong with this manifest ?02:11
r0kk3rzwhat makes you think there is?02:15
T4<birdzhang> you may need vendor source02:15
T4<Hound77> When I ran repo sync I see an error to fetch device kernel.02:16
r0kk3rzso why didnt you say that :P02:16
T4<austenite> @Hound77 see device revision02:16
T4<austenite> its 1ineage02:16
T4<austenite> XD02:17
T4<birdzhang> @austenite 👍02:17
T4<Hound77> I have mentioned lineage-15.102:17
r0kk3rzseems correct -
r0kk3rzwhats the error?02:21
r0kk3rzgenerally you want to fork those repos though02:21
r0kk3rzso you should do that, and point to your fork02:22
T4<Hound77> Do I want to do that and rerun repo sync ?02:22
T4<Hound77> Do02:22
r0kk3rzgo for it02:22
T4<Hound77> Do I need to add this ?
r0kk3rzyou will need that yeah02:23
r0kk3rzand possibly others02:23
r0kk3rzlike this one
T4<Hound77> What exactly do I need to add ?02:35
T4<Hound77> Just the vendor blobs.02:36
r0kk3rzthey are a set of repos, you need the whole set02:40
r0kk3rzits different for practically every device02:40
T4<Hound77> I am being a idiot.02:40
r0kk3rzcommon ones are always device,kernel,vendor02:40
r0kk3rzand then there might be assorted others that are also needed02:41
r0kk3rzyou'll soon know if something is missing, because hybris-hal wont build02:41
T4<Hound77> I'm for to build my manifest.02:42
T4<Hound77> refering for onyx02:43
T4<Hound77> refering from onyx02:43
r0kk3rzonyx has nothing to do with your device02:44
T4<Hound77> I'm using it as a example to create local-manifest for my device.02:45
T4<Hound77> So what repos do I have to mention on my local-manifest ?02:46
r0kk3rzfigure it out :P02:47
T4<Hound77> I think I have fustrated you with my lame chat.02:48
r0kk3rzslightly, but im not interested in telling you which keys to press, and which buttons to click02:50
r0kk3rzbut i will tell you generally what you need to do and look for02:51
r0kk3rzall the things you need are part of the lineage os build process for that device02:51
r0kk3rzall of that is open source02:51
r0kk3rzso start digging02:51
osum4esti finally got my zip file built and flashed, but it is boot-looping. it shows the vendors splash screen for ~30 seconds then reboots. how can i troubleshoot this? i'm not sure where to look for logs02:52
r0kk3rzosum4est: plug in usb, check dmesg in host pc, maybe try getting on to telnet if it shows02:52
osum4estr0kk3rz: no changes in dmesg during the boot loops02:55
r0kk3rznothing at all about usb devices? must be looping quite early then02:57
osum4estnope. i'm honestly not 100% it installed correctly and copied all the files where they are supposed to go to the correct partitions. i booted into twrp, adb shell, and looked at the system partition. theres stuff in there, but there is nothing in the data partition. are there supposed to be files there?02:59
r0kk3rzsounds wrong03:05
r0kk3rzdata is like, the whole rootfs03:06
osum4estoh. well uh, then thats probably my problem03:06
r0kk3rzits certainly not going to help03:07
r0kk3rzbut should should normally see stuff in usb / telnet before then03:07
r0kk3rzyou did flash the base image right?03:10
r0kk3rzso you flash lineageos, dont wipe anything, flash sfos.zip03:10
osum4estyep. the a/b slots are also confusing things. gimme a sec, i want to make sure i can flash just lineage and see my data partition03:12
r0kk3rzgood idea03:12
osum4estsince i have a/b, do i need to install lineage, reboot, then install sfos? or all at once?03:13
r0kk3rzyou shouldnt need to reboot, but you can boot lineage first to check03:14
r0kk3rzyou should also double check which boot partition is active03:16
osum4estwhich boot partition is active? and do you know, does the lineage/sfos installer write to the current slot or the other slot?03:16
r0kk3rznot sure, ive not worked with one of these devices03:18
r0kk3rzbut since there is two, it must choose one to boot from03:19
osum4estright, right. i need to make sure that both zips install to a and that i'm selected on a, since i setup the boot partition to boot_a03:20
r0kk3rzyou can also flash the boot.img manually03:25
osum4estr0kk3rz: thanks! alright, reset everything, installed lineage (data appeared after reboot, and booted into lineage), then installed sfos. checked the data partition, and the stuff from lineage was there but didnt notice anything new (is there an sfos specific file i can look for?). i also flashed the boot img to boot_a manually but i'm still getting the same boot loop and no output in dmesg03:37
r0kk3rzthe sfos stuff is in .stowaways under the home folder03:45
osum4estperfect. well i have a /data/.stowaways/sailfishos folder with a bunch of folders, including the nemo home folder, bin with a bunch of programs, etc. so it looks l like that part worked. is there a way to mount my boot partition in twrp to make sure it got installed correctly?03:48
r0kk3rzit flashed ok didnt it?03:55
osum4estit should have. just rewrote the hybris-boot.img to both boot_a and boot_b but still no dice.  i also have aboot_a, and aboot_b partitions if i should try using those instead03:58
T4<birdzhang> do not touch aboot03:59
r0kk3rztry doing a fastboot boot boot.img03:59
T4<birdzhang> it's bootloader partitions03:59
osum4estoh shoot, thanks for the warning. definitely wont touch those then. and good idea! didnt realize that fastboot boot just boots a regular boot image04:00
osum4estalright tried booting it with fastboot and the exact same thing is happening. must be something wrong with my boot image04:04
osum4estnot sure what it could be though since it doesnt connect over usb at all04:06
osum4estis it most likely going to be a bad setting in the kernel config?04:10
r0kk3rzpossibly, its not always that easy to tell04:16
r0kk3rzcheck for a kernel dump log04:16
r0kk3rzi think some newer devices needed some new things to get the usb up and running04:17
osum4estwhere is that?04:17
r0kk3rzif pstore is enabled in your kernel config, there might be a pstore-ramoops file04:20
osum4estfound it, thanks. is there anything specific i should look for? i'm not sure how to differentiate between hybris logs and twrp logs (unless its all hybris)04:33
T4<m_aurel> Our potential ODM offered us an alternative model with a Mediatek chipset MTK MT6761 Quad Core 2.0 GHz. The PCBA board vendor adapted Android Pie. So it's still challenging, because there is no experience with Android Pie for Sailfish OS. Maybe Hardware Adaption is easier for Mediatek than the Octa-Core SC9863 of Spreadtrum? Jolla confirm09:02
T4ed a high risk an more efford of HA for the last one.09:02
r0kk3rzyeah definitely stay away from that first one09:04
r0kk3rzmediatek based ports have been successful before09:05
r0kk3rzjust to put it in perspective, the hisilicon based phones that huawei makes, nobody has successfully made a port for those phones09:06
T4<m_aurel> Thanks for your recommendation. We won't work with Huawai.09:18
T4<NotKit> I would prefer MT6761 to SC9863 for sure09:19
T4<NotKit> both due to performance and MediaTek chips being more explored09:20
T4<m_aurel> Really? It has only 4 instead of 8 cxores. I was hesitating to choose that option becuase of this fact.09:21
r0kk3rz8 cores wont do you any good if you cant get it to work09:26
T4<NotKit> I was wrong actually, SC9863 might be more performant in multicore. See vs
T4<NotKit> but Redmi 6A Antutu score is higher than of Doogee with SC9863A: vs
T4<m_aurel> Thanks for the links. HA estimation for Spreadtrum is 25 MM. For any Qualcomm chipset, that is listed at Code Aurora, about 10 MM. What do you think about the effort for Mediatek? Something in between?09:31
T4<NotKit> Likely. There is Gemini which is Helio X25 and Cosmo will have Helio P7009:33
masha11mal: I get the following error in telnet dmesg: "F2FS-fs (sda20): Can't find valid F2FS filesystem in 1th superblock"
masha11And I get the error in /init.log: mount: mounting /dev/sda20 on /data failed: Invalid argument
masha11My fstab.qcom has two lines with /userdata (sda20). It is normal?
masha11How can I solve these problems?10:14
NeoChapaymasha11: you must format you data to ext2/ext310:33
masha11@NeoChapay I have to do it, being in telnet or twrp?10:35
NeoChapaymasha11: all ways is possible :)10:35
masha11@NeoChapay Why format not in ext4, but ext2/ext3?10:36
NeoChapaymasha11: if you can - format to ext4 not in all recovery we have mkfs tools for ext410:37
masha11@NeoChapay From telnet:  # mkfs -t ext3 /dev/sda20 Error: /bin/sh: mkfs: not found10:42
malext4 is the one to use, nothing else10:44
malfastboot can be often used to format also10:44
masha11@mal @NeoChapay thanks10:46
T4<austenite> @masha11 is your data encrypted12:43
masha11@austenite I don't now. How to check it?12:45
T4<austenite> in android, most likely it sia not encrypted by default in lineage..12:47
masha11I am confused by the mounting partition /data : /dev/block/dm-0 /data ext4.  why dm-0 and not sda20? TWRP Recovery  adb shell $mount:
Nokiusnew chapter to start soon native port :D19:23
osum4esthello, i still haven't been able to figure out why sfos keeps bootlooping early enough that it doesn't connect over usb. i found the kernel log at /sys/fs/pstore/console-ramoops but im not sure how to differentiate between hybris logs and twrp logs (unless its all hybris)
malhow quickly does it bootloop?19:29
osum4estit shows the vendors splash screen for about 30 seconds, then shows a black screen for about 5-10 then shows the vendor screen again. checking dmesg shows no devices connecting during the reboots.19:30
malwhich device and android base?19:31
osum4estmarlin, lineage-15.1 base19:31
osum4esthere's my defconfig if that helps: i've disabled selinux, enabled the selinux_bootparam and added selinux=0 to my kernel parameters20:06
osum4esti got rid of all the errors using the tool, and tried to get rid of all the warnings (even after fixing them the tools that they are still the old value?)20:07
T4<elros34> it's recomended to disable selinux via bootparam (read comment in CONFIG_AUDIT in mer-kernel-checker)20:12
vknechtand if that's not enough, try with disabling Yama20:13
osum4estgot it, i had CONFIG_AUDIT as no. should i leave CONFIG_SECURITY_SELINUX as yes then, and just make sure to have those kernel params?20:16
T4<elros34> yes20:17
osum4estthanks, ill give that a shot20:18
malosum4est: another hint
malif the device has secure boot enabled sfos might need it off20:29
mallike that20:29
osum4estmal: thanks! ill try that as well20:32
osum4estis there a sailfish specific directory in system that i can check for? dealing with the a/b slots is a pain and im still trying to make sure that both lineage and sfos are writing to the correct slot20:57
malwhat do you mean sailfish specific directory, all sailfish stuff is in userdata partition in .stowaways/sailfishos folder21:00
osum4estah, so all sailfish writes is data and the boot image21:00
osum4estso it shouldnt matter what slot im on as long i specified the correct boot slot partition21:00
malwell it matters that you mount correct partitions in sailfish and that kernel goes to correct slow21:02
malalso some a/b devices need a service to tell the system that boot was successful21:03
osum4estma: alright, added the verity config settings (and fixed the selinux ones, added selinux=0 and audit=0 to my kernel params) but still no dice. running make hybris-boot and then fastbooting to it should be enough to test changes right? i shouldn't have to reinstall the data partition?21:13
maldid you use fastboot boot or flash it?21:14
osum4estfastboot boot21:14
malsome device are picky about fastboot boot and require signed boot images21:14
osum4estweird, in that case, adb pushing it, then manually writing to the boot partition should work21:15
malor just fastboot flash boot hybris-boot.img21:16
maland then fastboot reboot21:16
osum4est... oh thats much easier. the guide used dd, didnt realize you could flash like that21:17
osum4estwell still same thing happens :/ would the log at /sys/fs/pstore/console-ramoops be able to tell me whats wrong? i have pstore logging enabled in my config21:20
T4<elros34> you could dump logs to sdcard or just write something to kmsg (echo " " > /dev/kmsg) somewhere here: to make sure it even reach that state21:54
malassuming it even reaches that far21:54
malit could even fail to run the init21:55
malPSA: hybris-15.1 is now rebased, I didn't have time to test it properly but it builds at least22:00
osum4estthanks for the idea, ill try that in a bit. and nice! although i might have done something wrong since i was able to build everything on 15.1... not sure what was broken before22:02
malsome people had issues building it, not sure if it was device specific22:02
osum4estelros32: mal: tried echoing there, but didnt find my message in kmsg. is there a way to view kmsg from fastboot? i've boot booting into twrp, adb shell in, and view it from there. at the beginning of the script i see that it logs some info to init.log. where would that be stored? the boot partition?22:31
T4<elros34> if you write something to /dev/kmsg you should see msg in console-ramoops provided kernel crash/reboots straight to twrp. You can't view kernel logs in fastboot you need serial port (uart) but that is usualy unavailable.22:41
T4<elros34> of course you can manually try to save init.log from the first stage (initramfs) to sdcard22:45
osum4estit does not reboot straight to twrp, i've gotta hold down power+voldown to get to bootloader again. save the init.log from inside the init-script?22:48
T4<elros34> if you turn off device it doesn't make sense to read console-ramops because kernel log from last boot is saved in RAM. Like I said earlier first just make sure script is even run22:51
osum4estelros34: sorry for all the dumb questions... but how do i actually dump those logs to /sdcard? i tried changing the init.log to be on the sd card, (exec > /sdcard/init.log 2>&1) but i doubt the init script even has that mounted.23:28
r0kk3rzyoud need to tell it to mount it23:30
T4<elros34> yes something like that should be enough (line 345): mkdir /sd; mount /dev/mmcblk0p1 /sd; touch /sd/iamalive23:32
T4<elros34> or mmcblk1p1 it depends on the device23:36
osum4estcan i just mount my data partition and write there instead? i don't have that partiton/an sd card (physically, at least)23:37
r0kk3rzyou can do whatever you like in the script23:38
r0kk3rzone of my favourite tricks is turning on the notification led at certain stages23:39
osum4estoo that'd be pretty nice! should probably just write a file first :P but then i might find the right file for doing that23:39
r0kk3rzthese were the paths for my sony
r0kk3rzi was trying to get the sony fotakernel recovery working23:42
osum4estok well it looks like boot script isn't being run, mounted my data directory and created a file there, and confirmed it didnt exist...23:45
r0kk3rzor it cant mount the data partition23:48
r0kk3rzwhich is a common enough issue in early stage boot failures23:49
osum4estoh, is there somewhere more reliable? i dont see any sdcard partitions23:51
r0kk3rzit still bootloops?23:57
r0kk3rztime it, and then slam in like a sleep 120 in the script at the start23:58
osum4estman that is a much better idea that trying to mount stuff lol23:59
r0kk3rzgetting the actual log would be quite useful23:59

Generated by 2.17.1 by Marius Gedminas - find it at!