Wednesday, 2019-08-14

ghosalmartinkrnlyng, have you seen any isses with lipstick memory corruption during certain actions like closing apps?09:07
krnlyngghosalmartin, yeah i have seen it09:14
krnlyngghosalmartin, usually it was due to gralloc bugs09:14
ghosalmartinkrnlyng, :O any idea how to fix? :P09:17
krnlyngghosalmartin, this fixed it in my case:
krnlyngghosalmartin, can you paste some logs?09:53
ghosalmartinlots of this in dmesg
ghosalmartindmesg of crash
ghosalmartinkrnlyng, can this bug occur when using surfaceflinger to drive teh UI?09:59
krnlyngghosalmartin, hm that looks a bit bad, what android version etc.?10:07
ghosalmartinkrnlyng, hybris-15.110:19
ghosalmartinnexus 5x10:19
ghosalmartinhappens for ngfd as well :(10:20
T42<birdzhang> i have a similar log
T42<birdzhang> UI crashes randomly, hybris 15.1, hwc2, vince10:25
ghosalmartinbirdzhang, wow not just me then10:25
Mister_Magisterghosalmartin: well there was one problem with huawei p8lite… quite similar but different10:26
ghosalmartinMister_Magister, was there a fix in the end/10:29
ghosalmartinit feels like a issue with permissions10:29
Mister_Magisterghosalmartin: kindof but i never made it10:31
Mister_Magisteri have different issue from you but just only similar10:32
ghosalmartinMister_Magister, whats your iss10:33
Mister_Magisterghosalmartin: in huawei after reentering running app it would crash because stupid huawei devs didn't implement 32bit functions in gpu driver in kernel10:34
Mister_Magisterkrnlyng: told me to do it myself but i'm too stupid for that10:34
ghosalmartinits not always ngfd or maliit-server, sometimes qmllthread10:34
Mister_Magisterin my case it was comming form kernel10:34
ghosalmartinMister_Magister, i remember you told me how to debug those statements10:34
ghosalmartinMister_Magister, wanna tell me again :P10:35
Mister_Magistertell what10:35
Mister_Magisterbtw i don't think anybody would be interested in running sfos on p8lite anymore10:35
ghosalmartinhow to debug those statements10:36
Mister_Magisterwhat statements10:36
Mister_Magisternot sure if its kernel or ngfd binary tho10:38
ghosalmartinhmmm okay10:40
krnlyngghosalmartin, it might be like Mister_Magister said, broken COMPAT layer for framebuffer in kernel10:41
Mister_Magisterghosalmartin: gib kernal sauces10:41
mal@adampigg one less custom package for your native port
Mister_Magisterrudi36: o710:43
ghosalmartinkrnlyng, would that lie in the drivers/video/msm part of the kernel?10:44
Mister_Magisterghosalmartin: gib kernal sauce but yes probably10:44
ghosalmartinplus if it was why would ngfd break10:44
Mister_Magisterghosalmartin: can you give the source?10:45
ghosalmartinhybris-15.1 branch10:46
krnlyngghosalmartin, but actually if you're using hwc2 then it shouldn't matter10:49
Mister_Magisterghosalmartin: iirc this w as the line in my source fb_ioctl = hi3620fb_ioctl; fb_compat_ioctl = hi3620fb_ioctl10:50
Mister_Magistersame function10:50
ghosalmartinkrnlyng, probably not no, pulseaudio also fails in a similar manor10:51
Mister_Magisterwait no10:51
Mister_Magisterlet's call 64bit function in compat function this totally gonna work10:52
malghosalmartin: how new config repo you have, just thinking if all firmware for the additional things is found10:53
Mister_Magistermal: you broke my gui ;-;10:54
ghosalmartinmal: pretty new? lemme check10:55
ghosalmartinwell that might be a lie its 4 months out of date :P10:55
Mister_Magisterghosalmartin: mal breaks guis ;-;10:57
malMister_Magister: well what do logs say11:01
Mister_Magisterposted them yesterday11:04
malnot all11:05
malthe logs looked quite bad with so many failing android side services11:06
Mister_Magistersure when i get home will post everythin11:06
Mister_Magistermal: only one more failing service11:06
Mister_Magisterbut servicemanger is waiting for it so i think it's the reason11:06
malMister_Magister: at least two are failing, loc_launcher and vendor.dataqti and a lot of qti errors which are never a good thing11:09
Mister_Magisteryeah but gui was up with them failing11:09
malalso qmi errors11:10
malMister_Magister: all could be related11:10
Mister_Magisterneed to debug that somehow but it's super hard to debug blobs11:10
malis some very important android side service is failing then nothing will work11:10
ghosalmartinwonder if this is related11:12
ghosalmartinshared memfd open() failed: Function not implemented11:12
Mister_Magisteriirc nope11:13
ghosalmartinno luck with dcd update either11:18
ghosalmartinall crashing when attempting to use syscall 38511:18
malghosalmartin: that is not a problem11:21
ghosalmartinwell outside of that i have nothing11:21
ghosalmartinmaybe some weird selinux issue?11:22
ghosalmartinwell minimer runs like a champ11:27
Mister_Magisterghosalmartin: addr2line11:39
ghosalmartinMister_Magister, do I just run on the vmlinuz?11:43
ghosalmartinI assume ??11:49
ghosalmartin??:0 means its not in the kenrel?11:49
Mister_Magisterprably not11:49
Mister_Magisterif its same kernel you are running then it's not correct place11:50
Mister_Magisterthen its ngfd or smth11:50
ghosalmartinokays ta11:50
Mister_Magistermine too11:52
jusaghosalmartin: memfd from pulseaudio is not related, it's just too verbose log, you can get rid of that by disabling memfd in pulseaudio config12:58
Mister_Magistermal:  logs
* Mister_Magister is reading them himself now14:56
Mister_Magistermal: keystore was disabled in 15.115:05
Mister_Magisterwait it's keymaster15:10
Mister_Magisterkeymaster was disabled aswell15:11
T42<elros34> about keystore: I posted it yesterday, shouldn't it have _HYBRIS_DISABLED added to binary not to argument:
Mister_Magisterno no15:11
Mister_Magisteri'm talking about hybris1615:11
Mister_Magister it doesn't disable keymaster15:12
Mister_Magisterand i'm having issues with it15:13
malMister_Magister: maybe build a new image using latest sfos?15:14
Mister_Magisteri could try but is it going to fix android services?15:14
malwell you said those worked before doing the changes I suggested15:15
maldid you do anything else than change .rc files?15:15
maland updating config repo?15:15
Mister_Magistermal: yep it worked before changing .rc files15:16
malso does it work if you put the old versions back?15:16
Mister_Magisteri removed old ones15:16
malyou should have local copies?15:16
malin your droid-configs repo15:16
malso you didn't add those there15:16
malwhy not15:17
Mister_Magisterthere was no point15:17
malwell you should know where you got the previous ones?15:17
Mister_Magistersurely form 15.1 xperia port15:17
Mister_Magistermentioned in faq15:17
Mister_Magister16 is not even disabling surfaceflinger15:18
Mister_Magisterwait it is15:19
Mister_Magisteri think i did something more to the .rc files because after adding them bacc gui isn't getting up15:21
malMister_Magister: it's not ?15:22
Mister_Magister[17:19] <Mister_Magister> wait it is15:23
Mister_Magistermal: getting this spam from hwc
malMister_Magister: strange that your logs have very few messages from droid-hal-init15:23
Mister_Magisteryeah i'm wondering about that too15:23
Mister_Magistergreat back to debugging hwcomposer when it was working15:31
Mister_Magistermal: i'm getting cannot remove mesa to get libhybris16:38
malso manually install libhybris16:38
Mister_Magisterm'kay i just read that in logs16:39
T42<adampigg> (Photo, 2560x1440)
T42<adampigg> Oh dear, what have i done17:29
T42<adampigg> Kimmoli^^ :D17:30
*** Konsieur is now known as kabouik17:30
r0kk3rzindeed, what have you done?17:40
kimmolidoghouse? soapbox?17:45
kimmoliah, rellu17:45
T42<adampigg> r0zz3rz: cool eh!17:57
T42<adampigg> my first 'classic car'17:58
mal@adampigg are the latter configs ok for new sensorfw? the iio scaling18:03
T42<adampigg> mal: not testted.....what should i be looking for?18:04
T42<adampigg> <goes get tablet>18:04
T42<eugenio_g7> They aren't yet, I need to adjust the configs first!18:06
mal@adampigg the removal of multiplying by 1000 was not there anymore, so sensorfw alwats multiplies the values by 1000 first, so you need to make sure the custom scaling factors for latte are correct18:06
T42<adampigg> @eugenio_g7 ^^18:07
T42<eugenio_g7> Will do that tonight :)18:07
mal@eugenio_g7 ok, I just happened to see that OBS project was using the custom sensorfw and thought that maybe ti could be changed to upstream18:07
malso we'll wait for the config changes first18:08
T42<adampigg> hmm, my latte is flashing blue led and not doing much else ... wht is a blue flash?18:08
T42<adampigg> turned white now18:08
T42<adampigg> oh, it came to life18:08
kimmoli@adampigg, how much if i can ask? they sell those (late 80s) for 2 to 6ke here18:14
r0kk3rzprolly £50018:23
T42<adampigg> r0kk3rz, kimmoli, 1k gbp18:30
T42<adampigg> so, approx the same euro18:30
T42Thelkishere was added by: Thelkishere18:38
T42<adampigg> (Photo, 1536x2048)
T42<adampigg> mal ^^ those value are probably a little large18:47
T42<adampigg> :D18:47
Mister_Magister@adampigg not if you are playing star wars xd18:49
T42<adampigg> mal: all our scale values are 0.0001 .... what should they be?18:51
T42<adampigg> 0.0000001 ??18:52
T42<adampigg> thats a v small number18:52
T42<adampigg> @eugenio_g7 ^^ ?18:52
T42<adampigg> maybe we need a few more zeros ... those will still be large numbers18:55
T42<adampigg> i never really paid attendtion before18:55
mal@adampigg just wondering why you didn't fix the kernel driver to provide correct scale?19:03
T42<adampigg> mal: iirc, @eugenio_g7 did the kernel changes for iio :D19:03
mal@adampigg kernel documentation says "Units after application of scale and offset are m/s^2."19:04
mal@adampigg heh, the of that iioadaptor in sensorfw has "data is wrong"19:08
T42<adampigg> mal: we also need iio for pinephone, so nned to understand all this19:09
T42<adampigg> @DylanVanAssche ^^19:09
mal@adampigg for reference
mal@adampigg also check that sensorfw should send out milli-G values so the * 1000 is needed because of that19:14
T42<adampigg> mal: @eugenio_g7 oh, we were using a custom sensorfw-qt5, with patched scaling .... its not being installed now to being out of date, need to review the situation
eugenio@adampigg, my fault... forgot to push the new tags on my own copy of the sensorfw repository20:38
eugeniowhich is now pointless as it got merged (and also version bumped), thanks spiiroin!20:38
T42<adampigg> so, its in 3.1 and we can drop the package?20:38
eugeniono, not yet, got merged this morning :D20:39
T42<adampigg> ah, ok, push the tag and ill rebuild on devel:20:39
eugenioI guess I'll simply switch repositories as it's now on mer-core/sensorfw20:40
malyes, no need for custom branch20:43
eugenio...aaaaand it's building! :)20:43
T42<adampigg> fab20:44
T42<adampigg> explains why my scaling wasnt working20:48
malhopefully it works now20:52
eugenio@adampigg, merging this: as it's already on obs anyways :P20:55
T42<adampigg> eugenio: good ... i was wondering where the maxpath changes had gone about 2 hours ago!20:56
T42<adampigg> wondered if id ever comitted it!20:56
eugenioheh :D20:57
eugenioalso we should modify that, it should be -maxpath 220:57
T42<adampigg> yes, thats what im using atm20:57
T42<adampigg> ill let you sort it as youre on the ball atm20:57
T42<adampigg> is dvd updated to master?20:57
T42<adampigg> dcd20:57
eugeniothe upstream submodule?20:58
eugenioyup, seems so20:58
T42<adampigg> good, i think we can update it now 3.1 is out21:02
T42<adampigg> eugenio: u tested those scale changes?21:12
eugeniodoing that in a few seconds21:13
T42<adampigg> eugenio: any thoughts on the system not going to sleep/screen off after inactivity?21:17
eugenio@adampigg, it does here... are you sure mce is working correctly? Or perhaps you have some misbehaving app using keepalive?21:23
T42<adampigg> hmmm, i have all apps closed. time set to 15 sec, and it doesnt dim21:28
T42<adampigg> egenio: can you post output of mcetool ?21:31
T42<adampigg> hmmm, it just dimmed and slept!21:31
eugeniomcetool output:
T42<adampigg> i wonder if task switcher could be keeping it awake21:34
T42<adampigg> probably a likely candidate21:34
T42<adampigg> nope,....21:37
T42<adampigg> that will ahve to wait for another time, gnight21:37
eugeniogood night21:37

Generated by 2.17.1 by Marius Gedminas - find it at!