Sunday, 2018-08-12

echo_salik[m]hey guys, how can i solve this error:05:33
echo_salik[m]error: in `sync --fetch-submodules`: revision refs/tags/android-8.1.0_r26 in platform/external/libpcap not found05:33
echo_salik[m]im using hybris-15.105:33
jjjhhhpppIs there anyone that is currently working on porting Sailfish for the Nextbit Robin? I happened to come across a Halium port for this device recently, and I'm curious as to if this can be useful in terms of getting Sailfish to run on this device.08:21
r0kk3rzsure, a lot of the fixes should be relevant08:24
YardanicoI'm so stupid :DDDDDD So I thought that the issue is in _my_ kernel changes, but in fact it was in some of 9 commits merged by original maintainer of this kernel for my device08:44
YardanicoI reverted 9 commits back and kernel worked :DDDD omg08:44
Yardanicoalso is it known that sometimes sailfish os doesn't boot and you need to reboot once more?08:46
r0kk3rzis it known? that will be a problem with your adaptation somewhere08:50
Yardanicor0kk3rz, oh, ok08:50
ghosalmartinmorning :)09:03
ghosalmartinif a repo is patternised, will all it takes to trigger a pattern update is a push on git?09:05
r0kk3rzand update on OBS yeah09:06
ghosalmartinahh yeah the update on obs bit i forgot :P09:06
ghosalmartinabout to shoot for a cheeky OTA from 2.1.x to 2.2.02909:07
Yardanicohow can I debug why sometimes GUI doesn't boot? what in the logs should I grep? :)09:21
r0kk3rzjust read them09:22
Yardanicooh, and also about my kernel spam - it only happens if USB is plugged in09:34
ghosalmartindo you no longer need the pulseaudio glue?09:46
r0kk3rzghosalmartin: you probably still need it10:00
ghosalmartinahh fairs fairs, i was looking at patterns in the repo and noticed it wasnt there10:03
ghosalmartinbut guess we need to add it in hadk10:03
r0kk3rzit depends on the device10:15
r0kk3rzit may not need it10:16
ghosalmartinhmm rmt_storage is now failing10:25
ghosalmartinwonder why10:25
Yardanicook, I hope I can solve wifi, bt and video playback today and if they all work I'll probably request store access :)10:26
ghosalmartinhmm ive masked ofono but this still causes a reboot [  +0.000059] modem subsystem failure reason: dog.c:1484:Watchdog detects stalled initialization.10:30
ghosalmartinhmm /dev/block/platform is missing10:31
r0kk3rzghosalmartin: which device are you breaking today?10:32
ghosalmartinr0kk3rz, bullhead, i updated my dcd and now /dev/block/platform no longer exists10:32
ghosalmartinwas there some removal?10:32
malnew udev rules were added a while ago10:34
ghosalmartinmal: i have those in there, actually i have them in my sparse and in dcd sparse10:34
ghosalmartinhmm /lib/udev/platform-device doesn't exist10:35
malyes, that was removed also because new rules don't need it anymore10:38
malghosalmartin: maybe try removing the custom rule you had is sparse and just use the submodule rules10:38
ghosalmartinmal: will do, is masking ofono just systemctl mask ofono?10:40
malyes10:44
Yardanicohmm, libav fails to build for me - "libavcodec/libavcodec.so: undefined reference to `ff_put_vc1_chroma_mc8_neon" and a bunch of other errors10:44
malghosalmartin: ofono is probably only failing because of the failing udev rules10:44
Yardanicoah I see mb2 can solve it probably10:45
ghosalmartinthat will be it, but i cant stop the device reboots long enough to update the device :P10:45
ghosalmartini guess i can mask dhi10:45
ghosalmartinmal, r0kk3rz bosh working again, thanks :D10:48
Yardanicobtw, for store access I should just ask testers of my build to capture screenshots of 'ssu s' and send them to p*eto ?10:49
r0kk3rzyou dont need to do that10:50
ghosalmartinisnt it just in the patterns?10:51
r0kk3rzjust make sure the ids match the imei10:51
Yardanicor0kk3rz, ah ok10:51
ghosalmartinmal: did you ever have any more luck with the ROME bluetooth chipset?10:52
malghosalmartin: not yet, I have been doing some android 8 stuff recently, got gps working10:57
ghosalmartinmal: noicee, i think for my mi mix issue am just going to take guhls config and try and get it to build, boot, then minimize down till i can find hte flags10:58
malghosalmartin: about rome, quite odd why I don't get the accknowledgement from the chip after uploading the firmware from hciattach11:05
ghosalmartinmal: it is, i wonder if that codes been tested at all?11:07
malnot sure11:08
malghosalmartin: btw, there was a new driver in mainline kernel for some of the chips, maybe we could compare that https://www.spinics.net/lists/linux-bluetooth/msg76794.html11:09
ghosalmartini couldnt find anyone to confirm or deny it ever working. looking through the code itself it seems to be right since its roughly the same as https://android.googlesource.com/platform/hardware/qcom/bt/+/refs/heads/master/msm8996/libbt-vendor/src/hw_rome.c11:11
ghosalmartinhmm one of my chargers doesnt work in SailfishOS but it works fine at boot11:12
ghosalmartinis that normal?11:12
malghosalmartin: for android 8 based ports there is the option to make a wrapper to communicate from bluez5 via android side binder interface11:12
ghosalmartinmal: wouldnt that make things much easier for us tbh?11:13
malghosalmartin: it might11:13
malghosalmartin: the interface is quite simple, it's just about sending an receiving hci messages11:14
Yardanicowhen I'm doing "build_packages.sh" should I answer "no" when it asks me to build packages like gstreamer-omx and libav (which I've built manually)?11:14
Yardanicofor example after building GMS-OMX as described in faq11:15
malyou shouldn't build that11:16
malthat is only used as the fallback in case gst-droid doesn't work11:16
r0kk3rzmal: that would be fairly magic tbh, using bluedroid drivers would solve a lot of headaches with bt11:17
Yardanicowell it seems like it doesn't for me - with gst-droid built I can't watch videos (if I try to watch a video in some app it hangs and SFOS asks me to keep waiting or kill the app)11:17
malr0kk3rz: ghosalmartin I already used the binder library for gps so it shouldn't be too complicated for me to try it11:17
malYardanico: logs logs logs11:18
malas usual11:18
r0kk3rzalso running things with GST_DEBUG can be useful too11:19
r0kk3rzbut make sure minimediaservice is running and happy first11:19
Yardanicor0kk3rz, "minimediaservice.service" ?11:19
malno, a process named minimediaservice11:20
Yardanico"media     2765  0.0  0.4  69400 11920 ?        Sl   12:47   0:02 /usr/libexec/droid-hybris/system/bin/minimediaservice" it seems it's fine11:20
Yardanicowhile trying to play a video (GST_DEBUG=true jolla-gallery) it prints out three times "library "libavenhancements.so" not found"11:22
malthat's not how GST_DEBUG is used11:22
r0kk3rzgst_debug takes a number, so GST_DEBUG=5 to get the full firehose11:22
malGST_DEBUG=droid*:511:22
maland additionally you can limit to certain parts like I did11:23
YardanicoI got this " WARN               droidvdec gstdroidvdec.c:669:gst_droidvdec_error:<droidvdec0> error: error 0x80000000 from android codec"11:23
Yardanicoother ones are info or debug11:23
malso logcat11:23
malalso check ls -l /dev/video*11:24
Yardanicomal, https://gist.github.com/Yardanico/f37b694b806fc4aed02f01699a75d87b11:25
malYardanico: also give ls -l /etc/ and ls -l /system/etc/11:27
Yardanicomal, https://gist.github.com/Yardanico/6a72ef29b8253d1fd76a835b1a252b3311:28
Yardanicook, another error (about wifi as a module): modprobe can't load wlan module: "modprobe: ERROR: could not insert 'wlan': Exec format error"13:15
Yardanicowlan.ko is 64-bit13:16
malis the wlan.ko a normal file or a symlink, some device have it as a symlink while the actual module has different name13:19
Yardanicomal, it seems to be a normal file13:20
Yardanicothis exec format error happens for all compiled modules13:21
malhmm, very odd13:21
mal64-bit kernels haven't been a problem before13:22
Yardanicosomeone had the same problem - http://www.merproject.org/logs/%23sailfishos-porters/%23sailfishos-porters.2017-01-13.log.html#t2017-01-13T21:36:32 but he solved it somehow13:24
Yardanico"file wlan.ko" -> "wlan.ko: ELF 64-bit LSB  relocatable, ARM aarch64, version 1 (SYSV), BuildID[sha1]=0x4f34cec97eab51e0be52d1ce427af50657702bab, not stripped"13:27
malYardanico: based on that log it was a mismatch in kernel and module somehow, he didn't flash the new kernel13:28
YardanicoI'll try to rebuilt kernel then13:29
vknechthello13:49
Yardanicohi :)13:50
T4<adampigg> Sitting by pool,.drinking beer,  scanning the logs for anything interesting,.and i read about using bluedroid drivers with bliez514:38
T4<adampigg> :what is this magic????14:38
vknecht:)14:49
* vknecht grabs a beer and opens logs14:50
Yardanicoso current big bugs for me are wlan (I need to figure out why modprobe doesn't want to load modules) and video playback15:13
mal@adampigg using the new hidl/treble API via libgbinder15:28
mal@adampigg I used that for gnss (i.e. gps and related stuff), it's also used for RIL in android 8 based ports, in the future even more subsystems will need that as the legacy APIs are not used anymore15:29
malsensors probably need it at some point also15:30
vknechtYardanico, one wlan module problem : in logs I had "Required key not available", worked around with CONFIG_MODULE_SIG_FORCE=n15:31
Yardanicoohh maybe that's the issue, thanks15:32
mal@adampigg using gbinder approach might also help with missing 32-bit libs on some devices, not 100% sure15:34
vknechtYardanico, another trap : at one point I found I had to boot LOS at least once before flashing SFOS, otherwise WLAN wouldn't work15:42
vknecht(rather, a tester found that, thanks to him :-)15:43
Yardanicovknecht, about wlan - did you have same error Exec format error?15:54
vknechtYardanico, iirc it was just "Required key not available"15:55
malYardanico: did you rebuild kernel and run build_packages.sh after that and then installed the droid-hal rpms (and flashed kernel) or rebuild the image16:01
Yardanicomal, well I rebuilt the kernel but now I'll run build_packages.sh too (with -d option right?)16:01
Yardanicooh wait I got this from dmesg: "module wlan: overflow in relocation type 261 val ffffffbffc960984"16:06
Yardanicoand googling brings up this https://lkml.org/lkml/2016/5/30/57416:06
YardanicoO16:10
YardanicoI'll try backporting two changes which fixed that16:10
Yardanicohttps://android-review.googlesource.com/c/kernel/common/+/234335 https://android-review.googlesource.com/c/kernel/common/+/23433616:10
vknechtany insights on CSD "suspend test" results interpretation, especially the iteration number ? Is it bad when it's 41 instead of 20, for example ?16:16
vknechtI've been fiddling with kernel updates a lot, and this test seems a good indicator of overall stability ; do you agree, or know a better way ?16:19
vknechteg. erroneus updates/merges made it drop to ~3mn suspend time, and upstreaming kernel seems to made it go from 12-13mn to 16mn16:20
Yardanicovknecht, how do I run suspend test?16:22
Yardanicoah I see16:22
Yardanico(found it)16:22
* vknecht expects the discussion to proceed now :-)16:44
T4<adampigg> Mal, ah cool,.so similar.approach for a few subsystems? Is it relatively straight forward to implement? And does bluedroid work better?16:44
mal@adampigg we haven't tried bluetooth with that yet, need to check how bluez handles plugins16:46
Yardanicovknecht, I have 14:27 suspend time, is that ok? :P and 20 iterations17:01
vknechtseems good, but I don't have other reference17:02
* vknecht realizes he didn't even tried it on X17:02
vknechtabout iterations, it's a mystery ; looked for source but couldn't find it17:08
YardanicoLOOOOOL IT WORKED AHAH17:08
Yardanico(I backported these two commits and wlan module loaded!!11)17:09
vknechtcongratulations :-)17:09
kimmolithen, when you get wifi working, check that you get correct mac address17:12
Yardanicokimmoli, how do I check that it's correct?17:14
Yardanicoreboot and see mac address again?17:14
Yardanicoor check in LOS?17:14
kimmolisomething like that17:15
kimmolimaybe check against LOS, or atleast check that OUI matches vendor17:16
Yardanicook17:17
Yardanicowell OUI search is ok: "78:02:F8 XiaomiCoXiaomi Communications Co Ltd"17:18
Yardanicokimmoli, yes it's the same in LOS17:26
kimmoligood17:26
Yardanicohow can I debug why GUI doesn't bootup (it happens sometimes?)18:27
Yardanicosometimes it does and sometimes doesn't18:27
kimmolifaq, logs, systemd-analyze if it is a service timing issue18:30
vknechthmmm, csd suspend test fails on my sailfishX (00:00)18:39
Yardanicobtw, can I already request access to jolla store or not yet? my device has two IMEIs (because it has two sim card slots), and `ssu s` correctly shows my first IMEI as a unique ID18:49
Yardanicoand wlan/bt mac adresses are working too18:49
Yardanicoand is it ok to have small binary files (3 files less than 100kb) in droid-configs repo?18:54
r0kk3rzwhat binaries?18:55
Yardanicor0kk3rz, binaries for sensor to work18:55
YardanicoI thought that I can put them to vendor repo but it seems that SFOS doesn't use vendor repo at all18:56
r0kk3rzthose are proprietary18:56
Yardanico*sensors18:56
Yardanicoyes :(18:56
r0kk3rzthe reason why we flash over an existing image is so we dont ship proprietary stuff in the sfos image18:56
Yardanicooh well18:57
YardanicoI will need to rebuild LOS 14.1 zip myself then :)18:57
Yardanicothat's not a big deal, I'm gonna do it tomorrow18:57
r0kk3rzdo sensors work in LOS?18:59
Yardanicor0kk3rz, yes they do, but xiaomi provides only 64-bit sensor libs for my device, that's why they don't work in SFOS18:59
Yardanicoand it's the same for redmi 4x/redmi note 4 ports18:59
r0kk3rzah ok19:04
Yardanicowell, idk what to do about video playback, so the error can be *maybe* fixed by using updated venus firmware, but that can only be done if system is using treble (vendor partition can contain firmware)19:28
Yardanicohttps://github.com/freemarkw/android_device_xiaomi_markw/commit/148c92f1583fca1174a1a305cb8fc32cd21613f819:28
Yardanicoor wait, maybe I can still provide my own19:28
malYardanico: pretty sure the issue is something else19:44
malassuming video playback works in android19:45
Yardanicomal, yes it works in LOS 14.1 which I use as a base19:45
ghosalmartinmal: have you hit any devices yet that break under using CONFIG_VT because I think its now impossible for me to enable :P19:55
ghosalmartinive used guhls config and managed to get it to build but it still corrupt the shit outta the build to the point where it doesnt hit the init script. the kernels own init main or even console-ramoops19:56
ghosalmartinliterally outta ideas now on how to get this device into telnet :P19:59
malghosalmartin: nope, it's quite odd how they have managed to break that20:00
malghosalmartin: what device was that?20:00
ghosalmartinmal: mi mix 2. it is very odd. i read somewhere on google they were abandoning CONFIG_VT but we'll see20:02
ghosalmartinseems something wayy too major to break20:02
malisn't that quite basic thing, wondering what kind of hacks they have do to break it20:03
ghosalmartinam wondering how they've managed to break it that the entire kernel becomes corrupt tbh20:03
ghosalmartinand it seems the files havent really changed for it20:03
ghosalmartinmal, do you have a link handy to the configfs devices kernel please?20:06
r0kk3rzgrep for CONFIG_VT to check for anything else that requires it20:07
ghosalmartinis CONFIG_VT all that important/?20:09
malghosalmartin: xa2 uses configfs https://github.com/mer-hybris/android_kernel_sony_msm/tree/hybris-sony-aosp-8.1.0_r20_2018040220:10
ghosalmartinta20:10
malghosalmartin: there is a branch of usb-moded with configfs support in mer-core, not yet merged20:12
ghosalmartinmal, id be amazed if i get that far :P20:13
malI was just testing phone calls in android 8, no audio but otherwise seemed to work20:14
ghosalmartinnoicee :P the weird thing is the CONFIG_VT breaks the kernel so badly i need to do a clean build otherwise the kernel is tainted for good20:16
malwtf20:21
ghosalmartinexactly my thoughts on the matter20:23
ghosalmartin...think it could just be my build machine?20:24
ghosalmartinits the only factor left20:24
ghosalmartinwell enabling any of these breaks it too CONFIG_RD_BZIP2=y20:28
ghosalmartinCONFIG_RD_LZMA=y20:28
ghosalmartinCONFIG_RD_XZ=y20:28
ghosalmartinCONFIG_RD_LZO=y20:28
ghosalmartinCONFIG_RD_LZ4=y20:28
ghosalmartinCONFIG_PROC_DEVICETREE=y20:28
ghosalmartinCONFIG_TMPFS_XATTR=y20:28
ghosalmartinCONFIG_MEMCG_KMEM=y20:28
malghosalmartin: that was 15.1?20:28
ghosalmartinyes20:28
malghosalmartin: maybe point me to the needed stuff and I can try to build a kernel for you20:29
ghosalmartinthis https://github.com/ghosalmartin/android_kernel_xiaomi_msm8998 branch hybris-15.1?20:29
malI probably need the other repos too so I can build it the usual way20:30
ghosalmartinthis repo https://github.com/ghosalmartin/android_device_xiaomi_chiron20:30
ghosalmartinnot sure what else is required20:30
malthat has some dependencies20:31
ghosalmartinhttps://github.com/LineageOS/android_device_xiaomi_msm8998-common20:31
ghosalmartinthink thats all three required repos20:32
ghosalmartintbf if you have a 15.1 codebase you can just breakfast chiron and itll pull all the appropriate repos in20:34
malhmm, why is repo sync failing20:42
malrepo command is not very good with error messages20:44
malfatal: Couldn't find remote ref refs/heads/master20:44
malwhy not tell which repo is causing that20:44
malit seems I need to do repo init again, not sure why20:48
ghosalmartinmal: none of them have a master, or are you repo syncing the entire manifest?20:55
malghosalmartin: I was just trying to sync the devices repos, maybe it tried to check something else also21:02
malghosalmartin: it seems github is slow so I won't build today21:02
ghosalmartinmal: fair enough, thanks :)21:02
malghosalmartin: kernel is syncing 50 KiB/s21:05
malghosalmartin: so it will take a while21:05
ghosalmartinmal: yeah ive noticed github have weird moments21:05
ghosalmartinam going to try and get a build that gets me atleast some kind of console-ramoops and then got from there21:05

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