**** BEGIN LOGGING AT Sun Jan 10 03:00:21 2021
Jan 10 09:08:36 <Tofe> Morning!
Jan 10 09:09:05 <Tofe> Herrie: was just thinking, did you check kernel cmdline ? does it match the one for 16.0 kernel ?
Jan 10 09:15:01 <Herrie> Tofe: Ehm yeah I might need to update that
Jan 10 09:18:47 <Herrie> Tofe: Seems it added: "firmware_class.path=/vendor/firmware_mnt/image"
Jan 10 09:18:51 <Herrie> Which I don't have
Jan 10 09:18:55 <Herrie> Let me recompile kernel
Jan 10 09:19:25 <Tofe> also, please make sure wlan is a module, so that we only load it when /vendor is ready
Jan 10 09:20:01 <Herrie> Well it's 7.1 defconfig so should be
Jan 10 09:21:30 <Tofe> yes, should it :)
Jan 10 09:21:42 <Herrie> Yeah should be OK: https://github.com/Herrie82/linux-1/blob/mido/3.18/halium-7.1/arch/arm64/configs/mido_defconfig
Jan 10 09:21:50 <Tofe> ok
Jan 10 09:22:00 <Herrie> I played a bit with some of the panic flags in defconfig so I was able to pull a later log ;)
Jan 10 09:22:54 <Tofe> that will help for sure
Jan 10 09:23:23 <Tofe> what I wonder also, is if a newer version of the wlan driver (like we may have in lineage 16's kernel)
Jan 10 09:23:26 <Tofe> would help
Jan 10 09:23:38 <Herrie> Probably
Jan 10 09:23:47 <Herrie> But the other kernel would fail early on
Jan 10 09:23:51 <Herrie> So not very helpful
Jan 10 09:25:45 <Herrie> But I kept the different variants 3.18 and 4.9 locally
Jan 10 09:28:18 <Herrie> So can easily switch if needed
Jan 10 09:28:30 <Tofe> on the other end, I'm using the newer kernel, and it also crashes, so...
Jan 10 09:28:39 <Tofe> (on rosy)
Jan 10 09:29:14 <Tofe> maybe we should just disable wlan for the moment, to be able to look at the rest
Jan 10 09:29:54 <Tofe> i.e. just remove wlan.ko from the flashed luneos
Jan 10 09:35:01 <Herrie> Also true let me try this one first
Jan 10 09:35:07 <Herrie> With updated cmdline
Jan 10 09:35:29 <Herrie> Might actually solve the issue
Jan 10 09:45:53 <Herrie> Still cannot find the firmware it seems
Jan 10 09:45:56 <Herrie> Even though it's there
Jan 10 09:46:04 <Herrie> I'll disable the WLAN completely
Jan 10 09:47:44 <Tofe> I had to add firmware_class.path=/vendor/firmware_mnt/image on my side
Jan 10 09:48:10 <Tofe> (was in BoardConfig.mk)
Jan 10 09:48:39 <Herrie> That's what I did
Jan 10 09:48:41 <Herrie> Still doesn't work
Jan 10 09:49:03 <Herrie> I'll now try with WLAN completely disabled
Jan 10 09:49:06 <Herrie> See what it does
Jan 10 09:50:19 <Tofe> ok, let's hope it helps
Jan 10 09:52:33 <Tofe> I'm really curious about this multiple loading of "ueventd: firmware: loading 'adsp.b03' "; I don't think it should happen
Jan 10 09:57:21 <Herrie> No it's probably not helping
Jan 10 09:57:23 <Herrie> You have the same?
Jan 10 09:57:29 <Herrie> Or is it my mido only?
Jan 10 09:57:42 <Tofe> same on rosy
Jan 10 09:58:12 <Tofe> and I suspect it causes strong issue, loading conflicting firmwares in parallel
Jan 10 10:08:22 <Herrie> Seems it doesn't reboot now anymore
Jan 10 10:08:27 <Herrie> But lots of funny WLAN stuff in logs now
Jan 10 10:10:49 <Herrie> Seems luna-next crashes with "Sep 20 10:17:40 mido luna-next[2331]: failed to find/load gralloc module"
Jan 10 10:11:11 <Herrie> sh-5.0# lxc-info android
Jan 10 10:11:11 <Herrie> Name:           android
Jan 10 10:11:11 <Herrie> State:          RUNNING
Jan 10 10:11:15 <Herrie> So that looks good
Jan 10 10:13:36 <Herrie> https://paste.ubuntu.com/p/JBfhrd8h7W/
Jan 10 10:23:22 <Herrie> Well seems it cannot find any hwcomposer binaries :P https://paste.ubuntu.com/p/RJYC4BXjkQ/
Jan 10 10:30:43 <Herrie>  Should be there in vendor: https://paste.ubuntu.com/p/tbnxrd4HfX/
Jan 10 10:33:37 <Herrie> But cannot find it... So vendor.img not mounted somehow?
Jan 10 10:43:05 <Herrie> Seems that it's in the device-common where it should build it
Jan 10 10:55:04 <Tofe> well, vendor.img should be mounted on /android/vendor
Jan 10 10:56:17 <Tofe> ah but you don't have any vendor.img, it seems
Jan 10 10:56:32 <Tofe> no sorry, misread your logs
Jan 10 10:56:43 <Tofe> (and of course you have one :p )
Jan 10 10:57:40 <Tofe> you can see the content of the container if you go to /proc/$(pidof init)/root/
Jan 10 11:05:30 <Herrie> Seems my init has 5 pid's...
Jan 10 11:05:32 <Herrie> sh-5.0# pidof init
Jan 10 11:05:32 <Herrie> 1730 1729 1728 1726 1
Jan 10 11:08:18 <Tofe> ah, mmh, maybe take ueventd's pid :)
Jan 10 11:09:25 <Herrie> Seems it might just be missing from the vendor common makefile
Jan 10 11:21:11 <Herrie> I always get confused with this Android build stuff
Jan 10 11:21:15 <Herrie> It's such a mess :P
Jan 10 11:22:01 <Tofe> Agreed
Jan 10 11:22:27 <Tofe> on my side, I think the issue I have is related to "init: Could not find service hosting interface android.hardware.graphics.composer@2.1::IComposer/default"
Jan 10 11:22:44 <Tofe> but it seems by device.mk and manifest.xml (for services) are both including it
Jan 10 11:23:19 <Tofe> maybe it just can't load the .so because of a missing dependency, but there's nothing about that in the logs
Jan 10 11:25:56 <Herrie> Well I simply seem to be missing a bunch of .so's for Display it seems
Jan 10 11:26:03 <Tofe> though the service *is* actually started, I have its pid...
Jan 10 11:26:21 <Tofe> ok, that's easier to fix :)
Jan 10 11:26:33 <Tofe>  -- will be away on a walk for a little while
Jan 10 11:27:26 <Herrie> Wich should be built by https://github.com/Herrie82/android_device_xiaomi_msm8953-common/blob/halium-9.0/msm8953.mk#L122
Jan 10 11:27:35 <Herrie> Just somehow it doesn't seem to end up in my final image
Jan 10 11:28:14 <Herrie> It actually exists in my vendor.img
Jan 10 13:51:28 <Herrie> Tofe: Seems correct vendor.img is installed, but we don't do anything with it
Jan 10 13:51:42 <Herrie> Shouldn't something be added in: https://github.com/Tofee/meta-smartphone-1/commit/86c83745880548a26abf662d74bad923d7fafe75
Jan 10 13:51:51 <Herrie> I guess other Halium 9.0 for UBPorts doesn't do it because of GSI
Jan 10 13:52:00 <Herrie> So it doesn't need to do anything with Vendor?
Jan 10 13:54:12 <Herrie> But we would need it for non-gsi devices?
Jan 10 13:54:24 <Herrie> Non-GSI/non-Treble that is
Jan 10 14:16:00 <Tofe> Herrie: vendor is mounted by halium's init script
Jan 10 14:16:06 <Tofe> into /android/vendor
Jan 10 14:16:43 <Tofe> then, we have /vendor symlink to /android/vendor, and we have pre-start.sh that will bind-mount /android/vendor onto /var/lib/lxc/android/rootfs/vendor
Jan 10 14:17:25 <Tofe> therefore we don't need anything in lxc's config
Jan 10 14:21:28 <Herrie> Tofe: Well but we don't do anything with vendor.img
Jan 10 14:21:36 <Herrie> At least I cannot find what we do with it
Jan 10 14:21:39 <Herrie> We only seem to copy it
Jan 10 14:22:10 <Herrie> Ì.e. we install it with https://github.com/Tofee/meta-smartphone-1/commit/2e5101852910d3652dfdf4f6876aa64851339509
Jan 10 14:22:14 <Herrie> But then nothing is done with it
Jan 10 14:22:18 <Tofe> Herrie: we mount it onto /android/vendor ?
Jan 10 14:22:25 <Herrie> Seems not
Jan 10 14:22:35 <Tofe> yes, with your patch of halium's init :)
Jan 10 14:22:39 <Herrie> My contents of /android/vendor != vendor.img
Jan 10 14:22:47 <Tofe> ah, that's an issue then
Jan 10 14:24:50 <Tofe> normally, vendor.img is mounted in /android-vendor, and later on we do mount --move of /android-vendor onto /android/vendor
Jan 10 14:30:30 <Herrie> I.e. my vendor.img has 38 files in /lib64/hw, my android/vendor/lib64/hw only has 11
Jan 10 14:31:01 <Herrie> So I think it takes the existing vendor partition on device instead of mounting the vendor.img we provide
Jan 10 14:33:29 <Herrie> So I think that in case we provide vendor.img we should mount it. In case we don't, the current scripts should be OK
Jan 10 14:47:09 <Herrie> I don't find any references to android-vendor when I grep -r
Jan 10 14:47:24 <Herrie> Ehm it's still grepping... fat fingered that one
Jan 10 14:49:17 <Tofe> oh, you have a vendor partition ?
Jan 10 14:49:45 <Herrie> Tofe: I guess so
Jan 10 14:49:47 <Herrie> Not sure
Jan 10 14:50:01 <Herrie> It does find references to android-vendor btw
Jan 10 14:50:04 <Herrie> When I grep
Jan 10 14:50:18 <Tofe> can you look at "ls /dev/disk/by-partlabel/" ?
Jan 10 14:50:57 <Tofe> well, if you do have such partition, i'll be mounted *over* the previous mount
Jan 10 14:51:03 <Tofe> it'll*
Jan 10 14:51:21 <Herrie> Ehm no, not there
Jan 10 14:51:41 <Tofe> ah, ok... could you paste your dmesg, or at least the part corresponding to the halium scripts ?
Jan 10 14:51:46 <Herrie> https://paste.ubuntu.com/p/QrsxySfq3x/
Jan 10 14:52:18 <Tofe> usually the treble roms use the "cust" partition as a placeholder, but our halium script doesn't know that anyway
Jan 10 14:52:46 <Tofe> or maybe via fstab ?...
Jan 10 14:53:10 <Herrie> https://paste.ubuntu.com/p/FKg9BV7tkg/
Jan 10 14:53:13 <Herrie> That's dmesg
Jan 10 14:53:18 <Herrie> This might be a clue? Sep 20 10:16:04 mido initrd: mounting android vendor image from system rootfs
Jan 10 14:53:43 <Tofe> well that's android-vendor
Jan 10 14:53:49 <Tofe> so it's good news
Jan 10 14:54:47 <Tofe> there even is "initrd: moving Android vendor to /android/vendor"
Jan 10 14:57:19 <Tofe> I don't see anything wrong in there
Jan 10 14:59:37 <Tofe> if you manually mount vendor.img from within adb (mkdir /tmp/vendor ; mount -o loop /userdata/luneos-data/luneos-lxc-android/vendor.img /tmp/vendor ) it's still different from /vendor content ? just in case
Jan 10 15:00:09 <Tofe> I'm running out of ideas
Jan 10 15:00:35 <Herrie> Content is the same there
Jan 10 15:00:40 <Herrie> But still not the 38 files :S
Jan 10 15:00:44 <Herrie> I'm also lost
Jan 10 15:00:55 <Herrie> Let me pull the vendor.img from device and compare with the one I have here
Jan 10 15:01:53 <Tofe> would it just be that your .bb pulls from build.webos-ports.org and you didn't re-upload it ?...
Jan 10 15:01:58 <Herrie> The vendor.img I have here is when I manually created it with mka vendorimage
Jan 10 15:02:10 <Herrie> Tofe: Well it should be the latest
Jan 10 15:02:22 <Herrie> But my build script does the same really :S
Jan 10 15:02:48 <Tofe> at least it solves a part of the mystery: the vendor.img on device is outdated
Jan 10 15:03:26 <Herrie> `Yeah seems so
Jan 10 15:03:33 <Tofe> usually I push it brutally with adb
Jan 10 15:03:34 <Herrie> Could it be that when I recreated the system-as-root
Jan 10 15:03:40 <Herrie> It didn't update vendor.img somehow
Jan 10 15:03:47 <Herrie> That's the only thing I can think off
Jan 10 15:03:55 <Herrie> How to push it?
Jan 10 15:04:08 <Tofe> just adb push vendor.img  /userdata/luneos-data/luneos-lxc-android/vendor.img
Jan 10 15:04:32 <Tofe> and then reboot
Jan 10 15:04:38 <Herrie> OK
Jan 10 15:04:40 <Herrie> Easy enough
Jan 10 15:04:55 <Herrie> Seems there's 17mb difference in the ext4 vendor images that I have
Jan 10 15:05:55 <Herrie> OK rebooting
Jan 10 15:08:17 <Herrie> Helps if I put the ext4 file :p
Jan 10 15:08:22 <Herrie> And not the sparse I guess
Jan 10 15:21:31 <Herrie> Tofe: OK that works
Jan 10 15:21:38 <Herrie> test_hwcomposer doesn't work but well
Jan 10 15:26:29 <Herrie> All binaries are there
Jan 10 15:26:33 <Herrie> luna-next seems fairly happy
Jan 10 15:26:36 <Herrie> But nothing on screen
Jan 10 15:28:26 <Tofe> starts and waits, like me ? or does more ?
Jan 10 15:29:04 <Herrie> Tofe: Well seems so: Start to relevant parts: https://paste.ubuntu.com/p/KxW5YhrKrN/
Jan 10 15:29:15 <Herrie> Other bits is mainly the spam for audio & wlan
Jan 10 15:29:56 <Herrie> Sep 20 10:27:16 mido systemd[1]: luna-next.service: start operation timed out. Terminating.
Jan 10 15:31:03 <Tofe> Herrie: ah, looks like mine :p
Jan 10 15:31:49 <Tofe> Herrie: can you add this line to pre-start.sh:
Jan 10 15:31:51 <Tofe> mkdir -p /mnt/vendor/persist && mount /dev/disk/by-partlabel/persist /mnt/vendor/persist
Jan 10 15:32:05 <Tofe> just after the mkdir of __properties__
Jan 10 15:32:40 <Tofe> I thought it was rosy specific, but looking at your dmesg log it seems you have the same mount in fstab
Jan 10 15:36:08 <Herrie> Done rebooting
Jan 10 15:40:24 <Herrie> Hmmz seems to crash earlier now :P
Jan 10 15:40:35 <Herrie> I.e. Android container stops running
Jan 10 15:41:10 <Herrie> https://paste.ubuntu.com/p/Bnbf8d7qfM/
Jan 10 15:44:15 <Tofe> if you have adb, can you do "lxc-start -n android -F"
Jan 10 15:44:27 <Tofe> it should output a bit more details on what failed
Jan 10 15:44:38 <Tofe> but I guess it's that new line in pre-start.sh, what else
Jan 10 15:47:26 <Herrie> https://paste.ubuntu.com/p/qr2mYSXY2R/
Jan 10 15:47:28 <Herrie> I guess o?
Jan 10 15:49:07 <Tofe> where is that log file again...
Jan 10 15:50:06 <Herrie> Well I edited and pushed from Windows but that shouldn't matter?
Jan 10 15:50:29 <Tofe> ah, yes, it does
Jan 10 15:50:35 <Tofe> chmox +x pre-start.sh
Jan 10 15:50:38 <Tofe> chmod*
Jan 10 15:54:06 <Herrie> Yup that works better for tests
Jan 10 15:54:17 <Herrie> Still not showing anything on screen but well
Jan 10 15:56:26 <Herrie> https://paste.ubuntu.com/p/5hw5xFJFnJ/
Jan 10 15:57:05 <Herrie> Might be worth pinging NotKit about it?
Jan 10 15:57:17 <Herrie> Seems Android container runs, but there's nothing on display?
Jan 10 15:59:57 <Herrie> Seems something on Android side might be the issue still?
Jan 10 16:00:43 <Herrie> I mean I see: Sep 20 10:25:46 mido init: Received control message 'interface_start' for 'android.hardware.graphics.composer@2.1::IComposer/default' from pid: 17 (/system/bin/hwservicemanager)
Jan 10 16:00:43 <Herrie> Sep 20 10:25:46 mido init: Could not find service hosting interface android.hardware.graphics.composer@2.1::IComposer/default
Jan 10 16:06:13 <Tofe> ah-AH
Jan 10 16:06:19 <Tofe> same as me then
Jan 10 16:06:35 <Tofe> I already pinged NotKit about that one... but he didn't seem to have any clue about it
Jan 10 16:14:35 <Herrie> Ah ok
Jan 10 16:54:22 <Herrie> Tofe: I came across this: https://review.lineageos.org/c/LineageOS/android_device_nextbit_ether/+/233470
Jan 10 16:54:34 <Herrie> Specifically: https://review.lineageos.org/c/LineageOS/android_device_nextbit_ether/+/233470/6/manifest.xml
Jan 10 16:54:38 <Herrie> Might that work?
Jan 10 17:02:43 <Tofe> already tried, but it's already there
Jan 10 18:17:03 <Herrie> I guess I could check logcat to see what's in there
Jan 10 18:17:17 <Herrie> Maybe something obvious is missing
Jan 10 18:21:22 <Herrie> Seems I need some same binaries as you :P https://paste.ubuntu.com/p/WM58gnNjmT/
Jan 10 18:23:47 <Herrie> libandroid, libnetutils, libwpa_client etc
Jan 10 18:53:22 <Herrie> Tofe: We probably need most from https://github.com/Halium/android_device_halium_halium_arm64/blob/7315445edc7b046138241560e1b72e36371d269b/device.mk
Jan 10 18:57:36 <Herrie> I'll build it with your additions and the things from there
Jan 10 18:57:41 <Herrie> Better to have too much vs too little
Jan 10 18:57:45 <Herrie> Can always trim it down when needed
Jan 10 18:58:31 <Herrie> Could be missing drm or displayconfig causes the issues, who knows
Jan 10 19:02:25 <Tofe> Herrie: my latest version: https://bpa.st/WJOQ
Jan 10 19:02:44 <Tofe> Though it didn't solve the main issue...
Jan 10 19:17:20 <Herrie> OK
Jan 10 19:17:23 <Herrie> well I'll try the same
Jan 10 19:19:57 <Herrie> Halium build done, now putting together the image
Jan 10 19:20:02 <Herrie> Will paint some stairs in the meanwhile :P
Jan 10 20:42:06 <Herrie> Tofe: This stuff is weird with Vendor
Jan 10 20:42:13 <Herrie> Have more things there now, but missing others...
Jan 10 20:42:20 <Herrie> like hwcomposer.msm8953.so :S
Jan 10 20:43:15 <Herrie> I'll push my vendor.img that worked before just to be sure
Jan 10 20:43:31 <Herrie> Really weird because I cleaned up my halium dir before this build
Jan 10 20:45:36 <Tofe> weird indeed
Jan 10 20:51:37 <Herrie> I do have more compared to previous image though
Jan 10 20:54:46 <Herrie> OK now behaving slightly better, but still not booting
Jan 10 20:57:08 <Herrie> Tofe: 09-20 10:19:26.240  2854  2854 E Adreno-GSL: <ioctl_kgsl_driver_entry:777>: open(/dev/kgsl-3d0) failed: errno 11. Resource temporarily unavailable
Jan 10 20:57:11 <Herrie> Is that hurting?
Jan 10 20:57:20 <Herrie> We might want to update udev rules as well?
Jan 10 20:57:25 <Herrie> I don't think I updated mine
Jan 10 20:57:27 <Herrie> Just thinking now
Jan 10 21:01:55 <Herrie> Seems there are quite a few less in this device tree
Jan 10 21:25:16 <Tofe> ooh what did you add? I never got this far
Jan 10 21:26:21 <Herrie> Tofe: well the vendor.img the line in pre-start.sh
Jan 10 21:26:26 <Herrie> That's it
Jan 10 21:26:35 <Herrie> Doesn't bootloop anymore
Jan 10 21:26:56 <Herrie> But still stuck @ luna-next/graphics
Jan 10 21:27:13 <Tofe> ok
Jan 10 22:29:37 <Herrie> Tofe; This might be what we're facing... https://irclogs.sailfishos.org/logs/%23sailfishos-porters/%23sailfishos-porters.2019-06-25.log.html
Jan 10 22:29:52 <Herrie> I'm seeing open(/dev/kgsl-3d0) failed: errno 11
Jan 10 22:29:58 <Herrie> In logcat
Jan 10 22:30:05 <Herrie> Adreno-GSL: <ioctl_kgsl_driver_entry:777>: open(/dev/kgsl-3d0) failed: errno 11. Try again
Jan 10 22:31:36 <Tofe> Herrie: looks like it's a missing udev rule, yes
Jan 10 22:34:54 <Herrie> Seems we might need to extend the Halium script we use to generate the udev rules: https://github.com/mer-hybris/droid-hal-device/blob/master/droid-hal-device.inc#L462
Jan 10 22:35:20 <Herrie> I was only doing "cat out/target/product/[codename]/root/ueventd*.rc | grep ^/dev | sed -e 's/^\/dev\///' | awk '{printf "ACTION==\"add\", KERNEL==\"%s\", OWNER=\"%s\", GROUP=\"%s\", MODE=\"%s\"\n",$1,$3,$4,$2}' | sed -e 's/\r//' > 70-[codename].rules" as per Halium docs
Jan 10 22:35:39 <Tofe> Herrie: you mean we're missing the vendor stuff
Jan 10 22:35:47 <Tofe> and system/
Jan 10 22:36:55 <Herrie> Yes
Jan 10 22:40:34 <Herrie> 200 more udev rules :P
Jan 10 22:45:37 <Herrie> Still the same issue though it seems
Jan 10 22:46:19 <Tofe> me too
Jan 10 22:47:12 <Tofe> "ServiceManagement: Waited one second for android.hardware.graphics.composer@2.1::IComposer/default. Waiting another..."
Jan 10 22:47:41 <Tofe> anyway, time for bed
Jan 10 22:47:43 <Tofe> gn8 !
Jan 10 22:54:24 <Herrie> Yeah same here, gn8 too!
Jan 10 23:08:38 <Herrie> Tofe: Seems that a530_pm4.fw doesn't get copied at my end
Jan 10 23:18:13 <Herrie> https://github.com/Herrie82/android_device_xiaomi_msm8953-common/commit/0cc9c5bb843d9726c13167efffb97a70b83cdb4b
**** ENDING LOGGING AT Mon Jan 11 03:00:17 2021