Sunday, 2022-08-21

T42<kreatowastaken> droid-hal-init is spitting out a lot of "exited 4 times in 4 minutes" errors05:19
T42<KernelPanix> This Can use ?09:46
T42<KernelPanix> This Will work ? (edited)09:46
malwhat do you mean? is that config file from some other device?10:51
T42<11> did you have any problems with not working wifi? (re @SailfishFreenodeIRCBridgeBot: <mal>what do you mea...)10:52
T42<11> [    2.540236] Register interface [wlan0]  MAC: 00:d3:e9 [    2.540258] CFG80211-INFO2) wl_event_handler : was terminated10:56
malnot that I remember11:01
T42<kreatowastaken> mine has working wifi even tho i couldnt even get gui working11:06
T42<elros34> so did you managed to get full journal/dmesg/logcat log?11:07
T42<elros34> not 10s of boot, at least 60s11:07
T42<kreatowastaken> uh i could11:08
T42<kreatowastaken> tho the phone is powered on for like 30 mins so i should reboot it to get logs (re @elros34: so did you managed t...)11:08
T42<kreatowastaken> what do you mean by this (re @elros34: not 10s of boot, at ...)11:08
T42<kreatowastaken> do you mean i should wait 1m before trying to get logs?11:09
T42<elros34> no you should start collecting them as soon as possible and at least 60s (usual boot time)11:10
T42<kreatowastaken> thats what i usually do anyway ok11:10
T42<elros34> if you didn't see lipstick in logs then unlikely11:11
T42<kreatowastaken> `Warning: Journal has been rotated since unit was started. Log output is incomplete or unavailable.`11:13
T42<kreatowastaken> i literally booted this thing up 5 seconds ago...11:13
T42<kreatowastaken> ok now i do11:14
T42<kreatowastaken> i have to get the logs like11:15
T42<kreatowastaken> in 2s11:15
T42<kreatowastaken> then it gets rotated11:15
T42<kreatowastaken> ` Aug 21 14:14:24 GalaxyS10 lipstick[5500]: couldn't load system EGL wrapper libraries`11:15
T42<kreatowastaken> okay11:18
T42<kreatowastaken> now i did get journalctl output11:18
T42<kreatowastaken> finally11:18
T42<elros34> 5s of journal11:35
T42<kreatowastaken> i did this as soon as possible tho11:36
T42<elros34> you hide your command so I do not know what you really executing11:37
T42<kreatowastaken> oh11:37
T42<kreatowastaken> ok now i see what you mean11:37
T42<kreatowastaken> you mean like this?
T42<elros34> I still do not know how do you generate this log: journalctl --nopage of live log with -f --no-tail, how about you read 'generating logs' section in notes?11:43
T42<kreatowastaken> i did journalctl > file11:43
T42<kreatowastaken> ok11:46
T42<kreatowastaken> 5 secs of logs11:46
T42<kreatowastaken> with journalctl --no-pager --no-tail -f > file11:46
T42<kreatowastaken> 1sec11:46
T42<elros34> did you read instruction? How about reboot then as soon as possible journalctl -f --no-tail, dmseg -w in other window, and /usr/libexec/droid-hybris/system/bin/logcat when droid-hal-init-starts11:48
T42<kreatowastaken> all at the same time11:49
T42<kreatowastaken> hm11:49
T42<kreatowastaken> ok11:49
T42<kreatowastaken> ill try11:49
T42<kreatowastaken> i couldnt do it at the same time11:50
T42<kreatowastaken> i hope thats ok11:50
T42<kreatowastaken> i got 5s of each11:50
T42<elros34> then do not even post it if it's just 5s. what is so hard to keep it running for 1 minute?11:51
T42<kreatowastaken> i thought u wanted it for 5s11:51
T42<kreatowastaken> ok11:51
T42<kreatowastaken> sending them as gists now11:56
T42<kreatowastaken> journalctl
T42<kreatowastaken> dmesg
T42<kreatowastaken> logcat
T42<elros34> these kind of errors looks strange: dlopen("/")12:12
T42<elros34> this one is probaly important:' servicemanager: Could not self register servicemanager Missing permissions to add IBase'12:14
T42<elros34> you said something about sparse/etc/selinux and symlinks there instead real files? If yes then do it properly and copy files there12:16
T42<kreatowastaken> alright12:32
T42<kreatowastaken> ok12:44
T42<kreatowastaken> doing that made wait_for_keymaster.service work12:45
T42<kreatowastaken> and theres a camera click now (so camera hal now starts)12:45
T42<kreatowastaken> still no lipstick and same output on that though12:45
T42<kreatowastaken> i can do the same 1min journal + dmesg + logcat if u want12:45
T42<elros34> just tell whether *servicemanager starts correctly now12:46
T42<kreatowastaken> it does but it gives this output every 20 secs12:47
T42<kreatowastaken> `Aug 21 15:41:46 GalaxyS10 ServiceManager: Waiting for service 'vold' on '/dev/binder'...`12:47
T42<elros34> I guess it's not critical if there are no other servicemanager errors.  What about this strange dlopen error?12:48
T42<kreatowastaken> need to reboot to check that onesec12:49
T42<kreatowastaken> it does12:50
T42<elros34> I see common libandroidicu error too12:51
T42<kreatowastaken> oh yeah i forgot to fix that12:51
T42<elros34> I think you could also get rid of failing adb_root android service12:55
T42<elros34> I guess android.hardware.wifi@1.0-service too12:57
T42<kreatowastaken> wifi does work btw12:57
T42<kreatowastaken> i connected thru connmanctl12:57
T42<kreatowastaken> it does work so idk why that error exists12:57
T42<elros34> wifi is handled in sfos/kernel side12:58
T42<elros34> this is android service12:58
T42<kreatowastaken> i see12:58
T42<kreatowastaken> "then create sparse/odm/lib64/ (or sparse/odm/lib for 32bit minimediaservice) "12:59
T42<kreatowastaken> should i make both? I did a search for the lib and i have both lib and lib64 of it12:59
T42<elros34> try and report, just a  symlink to test12:59
T42<kreatowastaken> `08-21 13:10:29.319  5578  5578 F linker  : CANNOT LINK EXECUTABLE "/usr/libexec/droid-hybris/system/bin/minimediaservice": library "" not found:13:11
T42<kreatowastaken> needed by /apex/ in namespace (default)`13:11
T42<kreatowastaken> getting this on logcat after doing the fix13:11
T42<elros34> so what you would do with this kind of error?13:15
T42<kreatowastaken> i tried doing the same thing but for but it didnt change anything13:15
T42<elros34> so how your /odm looks like now: ls -lR /odm13:17
T42<kreatowastaken> oh wait13:20
T42<kreatowastaken> now i see the issue lol13:20
T42<kreatowastaken> ok that was just me being dumb13:22
T42<kreatowastaken> now it complains bout ""13:22
T42<kreatowastaken> i guess ill do this until everything is set ig13:22
T42<elros34> mind sharing what was wrong then?13:22
T42<kreatowastaken> the dot (re @elros34: mind sharing what wa...)13:22
T42<kreatowastaken> on ls -lR13:22
T42<kreatowastaken> it should be /apex/ instead of ./apex/ so me being dumb13:23
T42<elros34> ah13:23
T42<kreatowastaken> okay13:24
T42<kreatowastaken> now thats gone13:24
T42<Spidey24Z> Hello13:25
T42<Spidey24Z> any fix for this error?13:26
T42<kreatowastaken> dlopen still exists though13:26
T42<elros34> @Spidey24Z yeah first entry in hadk-hot13:28
T42<KernelPanix> i want this may work, or not ? (re @SailfishFreenodeIRCBridgeBot: <mal>what do you mea...)13:36
malyou can't just take configs from other devices if that is what you mean13:55
T42<Spidey24Z> hybris 14.1 ? (re @elros34: @Spidey24Z yeah firs...)13:57
T42<elros34> what? I am talking about listAudioSessions error13:58
T42<Spidey24Z> i know but the fix in hadk-hot is revert a commit "" this commit not in cm-14.114:01
T42<Spidey24Z> i know but the fix in hadk-hot is revert a commit "" this commit not in cm-14.1 branch (edited)14:02
T42<elros34> it is:
T42<Spidey24Z> thanks bro14:06
T42<Spidey24Z> @elros34 fixed thanks14:16
T42<KernelPanix> i make it. follow mer check config. (re @SailfishFreenodeIRCBridgeBot: <mal>you can't just ...)14:25
T42<kreatowastaken> any ideas about the dlopen issue?16:07
T42<kreatowastaken> already exists on `/system/lib` and `/system/lib64`16:09
Mister_Magisterwhats the latest hybris?16:15
T42<kreatowastaken> @e16:26
T42<kreatowastaken> @elros34 it seems like the vold not working is the problem16:26
T42<kreatowastaken> `08-21 16:22:30.248  4489  4489 E SELinux : avc:  denied  { find } for pid=4775 uid=0 name=vold scontext=u:r:kernel:s0 tcontext=u:object_r:vold_service:s0 tclass=service_manager permissive=1`16:26
T42<kreatowastaken> and this seems like the cause of it16:27
T42<kreatowastaken> this is the only log i found on logcat that mentioned vold16:27
voidanix[m]@Thaodan any changes when applying 4ab3f8675342ed88127c92b862dde95d40b08aaa?16:51
Thaodanvoidanix[m]: I did apply that however I'm not sure if that made a difference16:54
ThaodanI found out that triggers the issue16:55
Thaodanhowever I'm not sure if reverthing is the right fix16:55
voidanix[m]so even with the above commit applied it does not suspend...17:16
voidanix[m]you can try your luck with copying an updated version of the function from
voidanix[m]notice how it doesn't check the PCIe link status and return ETIMEDOUT unlike LA.UM.9.12...17:18
T42<Spidey24Z> need help with radio,RIL this log from logcat and journalctl
maldoes your device use legacy ril or the newer IRadio interface?17:34
T42<Spidey24Z> Legacy (re @SailfishFreenodeIRCBridgeBot: <mal>does your devic...)17:34
malcan you check if that /dev/CHNPTYW3 mentioned in log exists and if it does what permissions/user/group it has17:35
T42<Spidey24Z> It's not exist even on android there are /dev/CHNPTYW1~1117:37
T42<kreatowastaken> @elros34 i found the reason for those libEGL errors17:45
T42<kreatowastaken> it seems to be because it checks /vendor/lib, not /system/lib17:45
T42<kreatowastaken> so i copied stuff that it needed from /system/lib to /vendor/lib17:46
T42<kreatowastaken> and i think libegl works now17:46
T42<kreatowastaken> now test_hwcomposer and lipstick doesn't segfault but they hang it seems17:47
T42<elros34> how about making symlink from /system/vendor /vendor?17:53
T42<kreatowastaken> ln -s /system/vendor /vendor17:54
T42<kreatowastaken> like this?17:54
T42<Spidey24Z> it exist actually from 0 to 14 (re @SailfishFreenodeIRCBridgeBot: <mal>can you check i...)18:06
malwhat happens you change the group of the numbers 0 and 3 to radio? I mean those target ones18:08
T42<Spidey24Z> root     radio           10 Aug 21 20:53 /dev/CHNPTYW0 -> /dev/pts/0 it's radio already18:14
malI meant the target that symlink points to18:21
malwhich is crw-rw----    1 root     tty       136,   0 Aug 21 20:53 018:21
T42<kreatowastaken> why would that help18:22
T42<kreatowastaken> and its already there (re @elros34: how about making sym...)18:22
T42<Spidey24Z> I/use-Rlog/RLOG-RIL: [w] AT channel [0] open successfully, ttyName:/dev/CHNPTYW318:23
T42<Spidey24Z> I/use-Rlog/RLOG-RIL: [w] AT channel [1] open fail, ttyName:/dev/CHNPTYW4, errno = 13 (Permission denied)18:23
T42<Spidey24Z> I/use-Rlog/RLOG-RIL: [w] AT channel [0] open successfully, ttyName:/dev/CHNPTYW018:23
T42<Spidey24Z> I/use-Rlog/RLOG-RIL: [w] AT channel [1] open fail, ttyName:/dev/CHNPTYW1, errno = 13 (Permission denied) (re @SailfishFreenodeIRCBridgeBot: <mal>which is crw-rw...)18:23
T42<Spidey24Z> ah sorry (re @SailfishFreenodeIRCBridgeBot: <mal>I meant the tar...)18:23
Thaodanvoidanix[m]: hm I can look. Yeah I noticed that too even in the tree for sm8350 there wasn't such a thing18:26
ThaodanWhat might be that 16517fc0bdf1a2810865a9d86b2043212fd15223 was added from another tree/branch but isn't handled.18:27
ThaodanRight now with it the resume is done wrong and breaks all subsequent suspends and resume.18:28
ThaodanReally these whole of branching and then copying of code into the new branch doesn't do good, context is hard to follow and fixes are stuck in newer versions18:29
ThaodanLike I found this one
T42<Spidey24Z> log now i changed all /dev/pts/ group to radio (re @SailfishFreenodeIRCBridgeBot: <mal>I meant the tar...)18:30
ThaodanAlready used in LA.UM.9.14.r1 but not in LA.UM.9.12.r118:30
mal@Spidey24Z that looks much better now, does ofono print any errors anymore or work?18:31
T42<Spidey24Z> still cannot connect to rild (re @SailfishFreenodeIRCBridgeBot: <mal>@Spidey24Z that...)18:36
T42<Spidey24Z> still cannot connect to ril (edited) (re @SailfishFreenodeIRCBridgeBot: <mal>@Spidey24Z that...)18:36
T42<Spidey24Z> lol18:43
T42<Spidey24Z> it works18:43
T42<Spidey24Z> is any way to add this to udev rules ? (re @Spidey24Z: log now i changed al...)18:44
malprobably some way, other option is to check why the permissions are wrong, does one of the android side .rc files set those or ueventd?18:53
malso grep device repos etc for those and hopefully there is some simple fix18:56
T42<Spidey24Z> couldn't found it on .rc files (re @SailfishFreenodeIRCBridgeBot: <mal>probably some w...)19:06
T42<Spidey24Z> this all device repo (re @SailfishFreenodeIRCBridgeBot: <mal>so grep device ...)19:06
T42<11> hello everyone, such a problem that wlan is not looking for networks19:10
T42<11> hybris 14.119:13
T42<elros34> @Spidey24Z check also /system and /vendor19:13
T42<kreatowastaken> btw now i still get the libEGL errors19:15
T42<kreatowastaken> even tho i put them to the place i did before to fix it19:15
T42<kreatowastaken> idk why19:15
Thaodanvoidanix[m]: mhi support was rewritten against mainline from that tree19:36
Thaodanit's like 80% mainline code19:36
Thaodan(from commit history pov)19:37
voidanix[m]by "that tree" you mean K.P.1.0?19:48
Thaodanvoidanix[m]: yes19:50
ThaodanLA.UM.9.14.r1 has the downstream version19:50
ThaodanK.P.1.0 has mainline plus patches19:50
voidanix[m]even more justified to copy `mhi_pm_fast_resume()` from there :D19:53
ThaodanIf it was just that easy:D19:53
Thaodanfor example there's no fast resume anymore19:53
ThaodanOther variants of return -EIO instead of -ETIMEDOUT no idea if that makes a difference19:54
Thaodanvoidanix[m]: the mainline version is a rewrite theirs no fast suspend or resume anymore just suspend and resume19:55
voidanix[m]ah yes not mainline, i sent you the link from where you can get K.P.1.0's (there's both fast resume and suspend there)19:56
voidanix[m]-EIO indeed makes more sense there though...19:56
ThaodanAh yes they moved that out to the misc file where they put all there downstream stuff there19:57
ThaodanBut ahain it's a totally different driver19:58
Thaodanthat version of mhi_fast_resume doesn't have MHI_ERR("Unable to access EP Config space\n");    either (K.P.1.0)19:59
voidanix[m]about being different: the driver (including that function) is the same but the order of calls is different in K.P.1.020:00
voidanix[m]yes because that's the bogus commit that doesn't exist outside LA.UM.9.1220:00
voidanix[m]have you noticed there are 2 missing commits compared to upstream msm-4.19?20:10
voidanix[m]9b4eddb8 and 020b270b20:10
ThaodanYeah the order of calls is about the same in the version based on mainline20:13
Thaodansome cleanup and etc20:13
ThaodanUpstream upgrades qualcomm code20:13
ThaodanI've seen the two commits that you mentioned however I don't know why they are there20:15
Thaodanshould ask kholk about that20:16
Thaodanvoidanix[m]:  changing the bogus commit to return -EIO didn't help besides creating a kernel panic20:52
ThaodanRemoving the bogus commit is better20:52
ThaodanAdding those two commits that you mentioned that were missing didn't help.20:53
ThaodanBut I guess they are still useful20:53
ThaodanNow suspend works but pci_pm_suspend_noirq(): cnss_pci_suspend_noirq+0x0/0x48 returns -1120:53
Thaodan bothers me20:53
voidanix[m]congrats! you on 4.19.188 correct?20:57
voidanix[m]would like a log of that `pci_pm_suspend_noirq()` just for the record :)20:58
ThaodanStock has a similar error:
Thaodan(expand the second error)21:01
voidanix[m]`smcinvoke: process_accept_req: accept thread returning with ret: -11` wut21:12
voidanix[m]this is wlan though21:12
Thaodanin stock?21:17
voidanix[m]that error msg is from stock yes21:18
voidanix[m]but in this case the error stems from wlan on both SODP and stock21:19
voidanix[m]two things: try
voidanix[m]maybe adding these 3 lines can also help:
Thaodanvoidanix[m]:  ah thanks21:24
Thaodangood ideas21:24
ThaodanDo you found any docs on the qcom branch naming?21:24
voidanix[m]lol nope21:25
voidanix[m]for 5.10 i just look for waipio (sd 8 gen 1) tags21:25
Thaodancode linaros gitlab is finally working21:25
Thaodanvoidanix[m]: I guess there's some idea behind them. some things make sense such as s s = snapdragon , m = mobile, a = auto21:26
Thaodanwhy is this not already there...21:28
voidanix[m]yep, got buried in those "CNSS2 snapshot" commits21:29
voidanix[m]about branch names:
voidanix[m]K.P might just mean mainline+something extra but no weird out of tree drivers21:32
ThaodanThose weird out of tree drivers are now in techpacks, sigh:/21:33
ThaodanBut those branches kinda make sense in these scheme21:33
Thaodanwant to see the stuff that runs on linux without android by them.21:34
Thaodanwant to know what this is thermal thermal_zone91: failed to read out thermal zone (-19)21:36
Thaodan stock has the same21:36
Thaodanthere's a thermal zone that can't be fully read21:37
Thaodanbut don't know which driver created it21:37
voidanix[m]at least techpacks do not bleed inside core kernel files like in the past21:39
voidanix[m]maybe you can use the board as well without them, gonna miss features like a display though21:39
ThaodanIdk there's benefit in both since this way the drivers can reuse in kernel code21:40
Thaodanas external modules I'm not sure21:40
Thaodanbut it's not just for android as their still autotools support for Car Linux21:41
voidanix[m]i've heard something about caf now just being codelinaro...21:45
voidanix[m]not being android specific like caf benefits everyone :)21:45
ThaodanThere's native linux support from some of the qcom blobs21:48
Thaodanbut I only saw that kind of thing mentioned in commit messages but not in any public docs besides their linux sdk21:49
Thaodanthe two commits from codelinaro apply cleanly on sony aosp21:51
voidanix[m]nice, also have you checked `/sys/class/thermal/thermal_zone91/type`?21:54
Thaodanits wireless21:56
Thaodanthe two commits don't fix the issue22:00
Thaodanvoidanix[m]: how do you know that those noirq suspend errors are from wifi?22:27
voidanix[m]it comes from `struct cnss_wlan_driver *driver_ops`22:30
voidanix[m]iirc your logs mention that it was wlan as well22:31
Thaodanmakes sense22:32
Thaodanwell wlan is running while use adb over tcp22:33
ThaodanI can check if it happens if I don't22:33
ThaodanIt doesn't seam to happen if wlan is off22:36
voidanix[m]yeah as expected22:36
ThaodanI wonder what's wrong, the driver should just don't do that action22:37
ThaodanMaybe that's just not fixed upstream22:37
Thaodanpci-msm-rc 0002:00:00.0: Refused to change power state, currently in D3 is the only thing that's left besides the thermalzone messages in that case22:38
voidanix[m]you still have these even after you said it suspends?22:38
Thaodanyes but stock has those too22:39
voidanix[m]oh, for thermal zone, is modem-wifi-usr mentioned anywhere there?22:39
Thaodansee stock dmesg here: for the22:40
Thaodanyou mean in the sysfs entry?22:40
Thaodanno that's disabled22:42
voidanix[m]check kona-thermal.dtsi22:43
Thaodanthere it's defined22:44
Thaodanthe disable comes from stock22:44
voidanix[m]then what does `cat type` exactly give?22:45
voidanix[m]i get it's wireless but looking for a string22:45
voidanix[m]oh, that's it?22:45
Thaodan1|pdx203:/ $ cat /sys/class/thermal/thermal_zone91/type22:46
ThaodanI wonder why it's disabled in stock and deleted on the sdxprairie22:46
Thaodaneh deleted on  kona-v2.1-iot-rb5.dtsi22:47
ThaodanAny opinions on adding those two cnss commits you meantioned earlier anyway?22:51
voidanix[m]i don't think they hurt...22:59

Generated by 2.17.1 by Marius Gedminas - find it at!