Wednesday, 2021-05-12

T42<elros34> Umeaboy I am not sure which droid config you have now: some new generated or that in your github?00:13
Umeaboy@elros34: I don't think I have one in my github.00:14
T42<elros34> ...00:14
UmeaboyApparently I do.00:14
UmeaboyIt's kind of late so therfor my brain is a bit sluggish.00:15
T42<elros34> yes is good to make breaks some time and rethink what you did00:16
UmeaboyI did what Thaodan told me to do to create a new droid-config-beyond2lte.spec00:16
T42<elros34> so make sure your local droid-config match that one in github. I am not sure about error you got but I bet you did install outdated target so maybe that is the reason. You can check your target release with: sb2 -t $VENDOR-$DEVICE-$PORT_ARCH -m sdk-install -R ssu re00:22
T42<elros34> for sdk just: ssu re00:24
UmeaboyDevice release is currently:
T42<elros34> yeah should be
Umeaboyssu re gave me:
UmeaboyI just checked to compare and the droid-config-beyond2lte.spec files are identical.00:30
UmeaboySo, the problem seems to lie somewhere else.00:31
UmeaboyGotta go to sleep.00:42
riniguscongratulations with 4.1.0 release and release for 10II !11:46
Mister_Magisterrinigus: the links handling is the most important shit12:01
T42<A_T_R> Which python version should use for building hybris-hal (make hybris-hal) for hybris-17.1?12:21
Mister_Magisterwhy not both12:22
Mister_Magisteractually idk if its 312:23
Mister_Magisterits jolla after all12:23
rinigusYou should use python 3, probably on your host pc12:25
Thaodanmake hybris-hal will use python from $ANDROID_ROOT not host12:44
rinigusThaodan: nice - didn't know that :)12:47
T42<A_T_R> HABUILD_SDK [x00td] atr@atr-DL-G41MXE:~/hadk$ python --version12:47
T42<A_T_R> Python 2.7.612:47
Thaodanthat's python from path..12:48
Thaodanplease read12:48
T42<A_T_R> my python version is 2.7.6 (ubuntu - 20.04 )12:48
T42<A_T_R> any soluion to this ?
Thaodantry to build aosp without patches first12:52
T42<A_T_R> sorry , i didnt understand that12:55
T42<A_T_R> should i update and upgrade ubuntu chroot (habuild_sdk) - sudo apt upgrade and sudo apt upgrade )???14:10
T42<Алексей %lastname%> @elros34 [you use _unpackaged_files_terminate_build, did …], Hello, this is my fixed git but the problem is not resolved, do you have any other going?14:33
T42<Алексей %lastname%> @elros34 [you use _unpackaged_files_terminate_build, did …], Hello, this is my fixed git, but the problem is not resolved, do you have any other idea?14:34
Mister_Magisterm01: have you updated to 4.0 yet? any special caveats?15:10
Mister_Magistermal: *15:10
T42<elros34> Алексей set your nickname in telegram so you can be pinged. Your droid config looks empty. To avoid repeating same thing all the time, here you can find some notes: Focus especially on hybris-17 point. Then rebuild hybris-hal, droid-hal and droid-config. Your error looks strange, can you ls -al /sbin/droid15:12
T42-hal-init? Also in future please either enable persistent logging (hadk-faq) or try to be faster and catch also early boot logs.15:12
T42<elros34> I just find out that your droid-config is still no up to date. Your sparse is empty but in logs you have some stuff copied from cheeseburger repo15:20
T42<leha155> Yes, maybe I tried using droid-config from cheeseburger, but this is not enough results.15:32
T42<leha155> This is last Journal:
T42<elros34> there is no point in providing new logs until you fix your droid-config and kernel (selinux)15:35
malMister_Magister: I haven't updated fp2 yet, fp3 works well with 4.015:48
Mister_Magistermal: oh… any special things i need to watch out for?15:53
Mister_Magisteror i can just add target on obs and voila15:53
malMister_Magister: older kernels will have issues with 3.415:56
Mister_Magisterwas waiting for this15:56
malfor example devices with 3.4 kernel15:56
Mister_Magistergot 3.415:56
Mister_Magisterknew it15:56
Mister_Magisterany solution?15:56
malit depends a bit on what kind of kernel version you have on those device15:57
malthere were some patches mentioned here before which most of the time make it work15:57
Mister_Magisternot updated since years15:58
Mister_Magisterso i prably have to get hadk…15:58
malI probably need to again try to get fp2 kernel fixed16:05
malMister_Magister: you also need to check that you have the needed kernel namespace option enabled in defconfig16:06
Mister_Magisterwhich one?16:10
malkernel check script will tell16:10
Mister_Magisterthere were some namespaces required for 3.0 already16:11
Mister_Magisterlike these16:11
malsame ones16:12
Mister_Magisterthey were needed for 3.016:12
Mister_Magistersuddenly everyone talking about them being needed again16:12
Mister_Magisterwhen oyu need them for 3.x in first place16:12
malI think those were not really needed before16:13
Mister_Magisteros wasn't getting up without them16:13
malat least not all16:13
Mister_Magisterbut ye i got them16:13
malhopefully those work as they should, my device has a bug in network namespaces which causes kernel panic when dsme shuts down16:14
Mister_Magisterso other than those i need some patches still huh16:14
Mister_Magisterain't nobody got space for this16:14
malI need to find where those patches were in logs16:24
Mister_Magisteri'll find them16:42
Mister_Magisterdon't worry16:42
Mister_Magisterbacc to hybris1617:09
riniguscontinuing discussion from last night - is there a way to force droidmedia to 32 bit?19:07
rinigusI can see that DROIDMEDIA_32 should be true, but would be nice to check it somehow in build files19:08
rinigusThaodan: let's see if we don't need to bind mount 64bit libs over 32 ones19:08
rinigusforce switching to droidmedia 32 bit made no difference.20:29
rinigusstill videos don't resume after pause and browser switching. but well, maybe should bump build to 4.1. and take it from there20:30
Thaodanrinigus: lets hope so bind mount was just an idea20:40
Thaodanit can avoid the /system symlinks20:40
rinigusThaodan: not sure if it is even relevant, tbh. to fix the video bug20:41
rinigusas for symlinks, we have /vendor/lib and /vendor/lib64 files with the same name. so, I would have to move lib one away...20:43
Thaodanrinigus: just tried to trow my wild guess in the room. I could at least avoid the symlinks for camera (see droid-config-sony-seine) and in case of community fpd20:59
T42<b100dian> Is hwcomposer patching a current thing? I'm struggling to find the hwcomposer module in my repo - the closest I could get is
Umeaboy@b100dian: You can add it to the defconfig and build.22:26
UmeaboyHave you confirmed that the mer kernel check script gives out a warning or error about it?22:27
T42<b100dian> No, I don't  think the faq reference to hwc.c patching is for a kernel module, but for a vendor one22:28
deathmist_@b100dian is there a specific reason you'd need to patch hwcomposer?22:30
T42<b100dian> I have crashes in the android graphics composer service, and test_hwcomposer to, and minimer doesnt do nothing22:32
Umeaboy@elros34: Are really ALL of these defconfigs needed or can I remove the cuttlefish, defconfig and ranchu64 one?22:32
UmeaboyI like to keep my repos clean.22:32
deathmist_Umeaboy: at that point you might as well go start deleting most of the things in the kernel tree because it is full of unused things and configs for various test platforms that Qualcomm themselves used for example ;)22:35
Umeaboydeathmist_: Mind helping me to fix that? :)22:36
deathmist_mainlining is the only real fix :D22:37
UmeaboyI didn't say that you'd do EVERYTHING. Just help out. :)22:37
UmeaboyYou mean to upload the stock kernel and build with that?22:37
deathmist_but no only one defconfig is needed unless your tree does some weird things or has some different models of the device for example22:37
deathmist_Umeaboy: mainlining as in getting the phone booted using as a base22:38
UmeaboyCool! That'd be neat.22:38
deathmist_my device (OnePlus 5/5T) is actually now officially supported by 5.13-rc1 fwiw :p22:38
UmeaboyOK. :)22:39
calebccffyoooooo i forgot that landed22:39
calebccffvery pog22:39
deathmist_@b100dian which hybris version? also link your droid-config repo for example so someone with knowledge can help22:45
Umeaboydeathmist_: I think @b100dian is using hybris-17.1 like me.22:50
T42<b100dian> android base is lineage-17.122:50
T42<b100dian> is the repo22:50
T42<b100dian> mal and elros already looked through it, maybe you shouldn't spend a lot of time cause I'd hate wasting three people's time on the same thing22:51
deathmist_ah, I never got graphics with hybris-17.1, some early stuff was causing droid-hal-init to die iirc and I never bothered to get past that as there wasn't much notes on anything (not sure if the situation is still the same); I currently have a fine working hybris-16.0/aarch64 port with SFOS 422:51
T42<b100dian> the current status is telnet + logcat (and droid-init-hal) working22:51
T42<b100dian> but 2/3 services generate stack corruption at launch22:52
T42<b100dian> and are display replated (2) or camera (1)22:52
T42<b100dian> strace or gdb doesnt' show much :(22:53
T42<b100dian> niice.. mine will also be aarch64:D22:53
deathmist_ ah yes I couldn't even run logcat; what happens if you run /usr/libexec/droid-hybris/system/bin/logcat?22:54
T42<leha155> Hallow. I found a lot of loops in /system. How can this be fixed?22:55
T42<leha155> / # ls -al /system22:55
T42<leha155> total 16422:55
T42<leha155> drwxr-xr-x   20 root     root          4096 Jan  1  2009 .22:55
T42<leha155> drwxr-xr-x   28 root     root          4096 Feb 10 00:33 ..22:55
deathmist_oh ffs22:55
T42<b100dian> for logcat, i've enabled selinux=1 in cmdline, copied some files in etc/selinux/ that I can point to (or grep a couple of days before of logs) and fixed a permission on /sys/fs/selinux/enforce. All in only one week:))22:56
T42<b100dian> the key point was the fact that I had 20 seconds where logcat was running before droid-hal-init would be brought down with it all22:56
Umeaboydeathmist_: If you're interested I'm looking to make a program similar to Odin or Heimdall based of the HADK so that actions can be done using buttons instead. Of course, having people learning to do the underlying work is best, but Heimdall was made to make flashing easier. So is my idea with the HADK.22:57
UmeaboyI'm not saying that you ARE interested. Just wanted to show you.22:57
T42<b100dian> Alexei: what about ls -al /22:58
deathmist_@b100dian no. @b100dian read the pinned message, no paste spam22:59
deathmist_uh, @leha155 I mean22:59
T42<b100dian> Yeah, a pastebin.ubuntu,com link @leha15523:00
T42<b100dian> deathmist_ luckily I didnt tag him correctly first:)23:01
deathmist_@Umeaboy at some point I was thinking about it, but the brittle nature and other random issues with especially kind of make it unfeasible imo. there's just too many variables to make it stable imho23:02
Umeaboydeathmist_: If I make my own repo for the Linux kernel from Torvalds do I still call it android_kernel_exynos9820 or do I call it something better?23:03
Umeaboylinux_kernel_exynos9820 ?23:03
T42<leha155> @b100dian
T42<leha155> @deathmist_ [uh, @leha155 I mean], normal view
deathmist_@Umeaboy mainlining is a massive undertaking, fork naming doesn't really matter. I called mine linux-mainline-oneplus5 fwiw. the biggest issue is your Exynos 9820 SoC which doesn't have even basebones support there, which means you have to start from scratch23:10
T42<leha155> lrw-r--r--    1 root     root            11 Jan  1  2009 bin -> /system/bin refers to itself23:12
T42<b100dian> yeah, normal view - but some of those links are created by rc. scripts so..23:12
deathmist_@leha155 that's a system-as-root issue, which you shouldn't be hitting assuming everything is fine on droid-hal side, what is output of "mount"?23:13
T42<b100dian> deathmist_ here's a useless example of crash + strace + gdb that made me think maybe I should rebuild,
T42<leha155> mount
T42<leha155> my fstab
T42<leha155> @deathmist_ [@leha155 that's a system-as-root issue, which …], mount
T42<leha155> my fstab
T42<leha155> i dont have system-as-root, or i am wrong?23:23
deathmist_@leha155 the broken ls /system indicates you have SAR, and the fstab is your issue here. fork it and change system mountpoint to /23:27
deathmist_optionally you can also remove /vendor in favor of letting systemd handle the mount with guaranteed dependency ordering: (see also symbolic link in my sparse/usr/lib/systemd/system/
T42<leha155> ok, after change mountpoint i need rebuild hybris-hal?23:31
deathmist_no need, only droid-hal and new rootfs ( -di), but obviously if you do the second thing (which I recommend) you'd have to change the What= on your copy of the file to match which vendor partition you're mounting23:33
deathmist_actually I think you need to rebuild hybris-hal, the fstab may only get copied to out then23:33
T42<leha155> Ok thank. unsubscribe after23:42
Umeaboydeathmist_: Rome wasn't built in a day. :)23:47

Generated by 2.17.1 by Marius Gedminas - find it at!