Thursday, 2019-08-22

T42<Zosenko> @adampigg [<reply to media>], Formentera?04:51
T42<Zosenko> @adampigg [<reply to media>], Cala Saona?04:53
T42<meierrom> @Zosenko: paradise rather :))06:21
electro575hi mal07:49
*** ChanServ sets mode: +o sledges07:52
*** sledges sets mode: -n 07:52
*** ChanServ sets mode: -o sledges07:52
sledgesr0kk3rz: done ^07:53
Mister_Magistersledges: hii07:53
sledgesNokius: have you gotten your obs target for pinephone?07:53
sledgesPSA: SFOS collab discussion at #mer-meeting channel starting in 5 (minutes)07:54
sledgesNokius: ok, dontbeevil is there08:00
*** vknecht_ is now known as vknecht08:01
r0kk3rzmoo sledges08:03
electro575it's a little bit strange ->
electro575why this ?08:04
electro575Clean step: rm -rf /home/sailfish-dev/hadk/out/target/product/t0lte/obj/STATIC_LIBRARIES/libxml2_intermediates08:04
electro575Clean step: rm -rf /home/sailfish-dev/hadk/out/host/linux-x86/obj/STATIC_LIBRARIES/libxml2_intermediates08:04
electro575i have done repo sync external/libxml208:05
r0kk3rzsledges: was thinking of organising a meetup mid september if you're in08:21
r0kk3rzlike 14th or such08:21
sledgesis 21st also on the cards or will you have swapped the island for a bigger one already? :)08:33
T42<adampigg> @sledges [London?], Or.cumbria08:33
r0kk3rzyou need to come down south piggz!08:34
r0kk3rzcould maybe do 21st sledges08:36
sledgescool, i'll tell ya mine availability in couple of days' time, too many events on the weekend the 14th, but ya never know:)08:37
T42<adampigg> Waiting patiently.for meeting log...08:43
sledges@adampigg: this one?
T42<adampigg> Ta08:46
sledgesit's still rolling, curltail it:)08:46
T42<adampigg> Just wanted to read about.qt update08:50
sledgeswe want newer qt as much as you do if not more08:51
T42<adampigg> Jfdi ;)08:53
r0kk3rzaye, just push it08:59
sledgesgit push jollas-partners..09:03
electro575is it a problem if my SDK version is rathen than i have cm 14.1 repo ?09:14
rinigussledges: it would make sense now to move to 5.12 then. unless that will give extra postponedTM bonus09:15
T42<DylanVanAssche> Indeed... 5.12 would be awesome. 5.9 is already out of date in terms of new APIs and stuff09:16
sledges5.9 already exists in feature branches, as veskuh mentioned in the meeting09:19
T42<DylanVanAssche> That's true, existing in terms of: merging and done or still some work ?09:22
sledgesit worked when it was done;)09:24
malelectro575: android side build has nothing to do with sdk09:27
T42<DylanVanAssche> @sledges awesome :D09:28
r0kk3rzthat was done aaaaaages ago now09:32
electro575i remove all my project and redo09:45
malelectro575: NO10:07
malthere is almostt never need to remove things10:08
maldon't expect any help from if you do that10:09
electro575i have done all backup to remake all things10:11
T42<DylanVanAssche> @mal I build droid-hal-version, added these files:
T42<DylanVanAssche> Since the vibrator is a native input device, I added `%define native_build 1` to this spec:
T42<DylanVanAssche> However, when typing on the keyboard, nothing happens.10:37
T42<DylanVanAssche> Any tips?10:37
r0kk3rzsome devices need custom configs to increase the vibeness10:38
r0kk3rzwhat was wrong with the ffmemless vibes?10:40
T42<DylanVanAssche> Is there some usefull test program to find the best values?10:40
T42<DylanVanAssche> We use a native adaption without hybris, `native_build 1` is needed for this10:40
T42<DylanVanAssche> This is my spec file:
r0kk3rzooooh, you arent building onyx?10:42
T42<DylanVanAssche> Nope :P but I started from the onyx adapation for the vibrator (as suggested by mal :D )10:43
r0kk3rzi got confused because you linked onyx repos10:43
r0kk3rzso you want to use ffmemless for pinephone?10:43
T42<DylanVanAssche> Oh yeah that was the link mal gave to me a couple days ago.10:43
T42<DylanVanAssche> Well I wanted to get the GPIO vibrator working (input event device). From the HADK, I read that ffmemless was the way to go here10:44
r0kk3rzjust making sure i understand the situation :)10:47
T42<DylanVanAssche> :D sure ;)10:47
T42<elros34> not sure it make difference but there is new format of ffmemless.ini, check /usr/share/ngfd/plugins.d/50-ffmemless.ini10:47
kimmoliyour kernel needs to support memles10:48
malkimmoli: he said he sees the input event for vibrator10:53
kimmolitehn ok.10:53
T42<DylanVanAssche> @r0kk3rz These things are there :) they point to the right files.10:54
T42<DylanVanAssche> @kimmoli indeed, I see the vibrator already in the input devices/10:54
T42<DylanVanAssche> @elros34 I will try to use that config instead, maybe that makes a difference10:55
kimmolii'm just getting highlights on ONYX10:55
mal@DylanVanAssche show output of evdev_trace -i, from mce-tools package10:56
kimmoliand have this, no android vibrathings
kimmoliso this must be removed: pattern:jolla-hw-adaptation-hybris-droid-vibrator10:57
T42<DylanVanAssche> @mal
T42<DylanVanAssche> @kimmoli qt5-feedback-haptics-ffmemless is missing in our droid-config. But I installed the package manually.11:03
mal@DylanVanAssche I told you previously that config repo doesn't need that qt5-feedback-haptics-ffmemless, it comes from droid-hal-version if you have ffmemless defined there11:08
mal@DylanVanAssche noticed a problem, you can't have you need to remove the whole line11:09
malat least previously defining something as 0 didn't work11:09
Mister_Magisterhow tf does pro1 and xperia 10 keymasters work11:17
Mister_Magisteri can't understand11:17
Mister_Magistermal: spl version in kernel has to patch system and vendor in order to get keymaster working and sfos bootimg doesn't have those and xperia 10 and fxtec keymaster's are working probably because this requrement was removed in android9 and these 2 are 9 from beginning but this requirement was in android8 which my zenfone had before i updated that's why it could be still there so i need to add spl version to bootimg o smth11:22
malmaybe you should show what error you get11:23
Mister_Magisteroh i did already but sure11:23
Mister_Magisteri need this probably
Mister_Magisteryep that solved the issue and that is the issue11:51
malnot sure what that even does11:55
Mister_Magister[13:22] <Mister_Magister> mal: spl version in kernel has to patch system and vendor in order to get keymaster working and sfos bootimg doesn't have those and xperia 10 and fxtec keymaster's are working probably because this requrement was removed in android9 and these 2 are 9 from beginning but this requirement was in android8 which my zenfone had before i updated that's why it could be still there so i need to add spl version to11:56
Mister_Magisterbootimg o smth11:56
Mister_Magisterhere is everything you need to know :P11:56
electro575why this error ?
Mister_Magistermal: what could block ioctl?12:55
Mister_Magisteron /dev/dri/card012:55
Mister_Magistermaybe you know case like this12:55
T42<DylanVanAssche> @mal Okay, I must have missed that.13:05
T42<DylanVanAssche> Rebuilding now :)13:05
T42<DylanVanAssche> @mal hmmm installed the fixed package, no difference13:10
mallogs logs logs13:15
T42<DylanVanAssche> I would love to give them, but I'm not sure from which daemon13:16
T42<DylanVanAssche> ngfd?13:16
T42<DylanVanAssche> `journal-ctl --since=today --no-pager | grep ngfd` ?13:17
malfull journal and dmesg13:19
T42<DylanVanAssche> @mal Journal:
T42<DylanVanAssche> dmesg:
malonly one message from ngfd13:26
T42<DylanVanAssche> Yeah that's my concern too13:27
malshow the actual changes you did to config repo13:28
T42<DylanVanAssche> Running manually:13:28
T42<DylanVanAssche> ```13:28
T42<DylanVanAssche> [root@Sailfish ~]# ngfd13:28
T42<DylanVanAssche> [0.029] ERROR: core: unable to init required plugin 'profile'13:28
T42<DylanVanAssche> [0.029] ERROR: daemon: Initialization failed.13:28
T42<DylanVanAssche> ``13:28
jusa@DylanVanAssche ngfd should be run as nemo13:29
T42<DylanVanAssche> @jusa hmmm you're right :P13:31
T42<DylanVanAssche> ```[nemo@Sailfish root]$ ngfd13:31
T42<DylanVanAssche> Vibra upload effect: No space left on device13:31
T42<DylanVanAssche> [0.221] ERROR: ffmemless: Could not load default fall-back effect13:31
T42<DylanVanAssche> [0.237] ERROR: core: sink 'ffmemless' failed to initialize13:31
T42<DylanVanAssche> Segmentation fault```13:31
jusathat segmentation fault is when initialization fails, it's fixed in upstream but not in release yet13:32
jusathat ffmemless error is something with kernel, check kernel logs.. why are you running ffmemless btw?13:32
r0kk3rzno droids13:33
T42<DylanVanAssche> @jusa native vibrator as an input device13:33
jusa@DylanVanAssche ok. any logs from kernel driver?13:33
T42<DylanVanAssche> @jusa13:34
T42<DylanVanAssche> ```[    8.821441] gpio-vibrator vibrator: vibrator supply vcc not found, using dummy regulator13:34
T42<DylanVanAssche> [    8.846974] input: gpio-vibrator as /devices/platform/vibrator/input/input3```13:34
T42<DylanVanAssche> but nothing more13:34
jusamaybe enable verbose logs for the driver?13:35
jusaffmemless was implemented for jolla1, ported to jolla C, and afaik never used with anything else so it very well may be there are some issues13:36
T42<DylanVanAssche> Euhm will check how to do enable those verbose logs :)13:38
maljusa: @DylanVanAssche is pulseaudio needed for that, there seems to be a dependency in ngfd?13:44
r0kk3rzjusa: a few ported devices used ffmemless too13:44
r0kk3rzas well as asteroid smartwatches :)13:44
T42<DylanVanAssche> @jusa I increased the loglevel to 7 (KERNEL_DEBUG), but still no additional information from the driver.13:58
T42<DylanVanAssche> I'm wondering if it's not possible to trigger the vibrator from the command line (to make sure that the driver is doing its job)13:59
albertus1any new ideas about my conflicts?
T42<elros34> dylan: did you increase maximum number of effects in memless driver in kernel (HADK)?14:30
T42<DylanVanAssche> @elros34 hmmm no I didn't ... :S14:34
T42<elros34> @albertus1: you could try to run mic in more verbose level like: —verbose —debug14:46
kimmolijusa: fyi, at least onyx uses ffmemless14:56
sledgesr0kk3rz: it'll have to be the 21th for me to make it to sfos meetup in London15:46
T42<DylanVanAssche> @elros34 I recompiled the kernel with the patch. No difference :(15:50
r0kk3rzsledges: aye aye captain15:59
T42<elros34> @DylanVanAssche: cyou could at least try whether device works with fftest16:05
vknechtalbertus1, maybe try this, then rebuild geoclue-provider-hybris-binder:
Mister_Magistermal: hwc cannot use ioctl on /dev/dri/card0 because of permission denied. Any idea what could cause that?16:12
Mister_Magisteri'm outta options16:12
vknecht(just a guess: narrowing provided files to actual files, not specifying upper dirs)16:12
malMister_Magister: what are the permissions of that device?16:18
malMister_Magister: those look normal16:21
Mister_Magisteri'm fully aware of that16:21
malso the issue is something else, more logs16:21
Mister_Magisterthough my composer is full of EACCESS16:21
Mister_Magisterno i'm sure hat's the issue but strace of hwc incomming16:22
* Mister_Magister where do i upload 8MB16:22
T42<DylanVanAssche> @elros34 Is that available in some package already? I found on the web that you need the package `joystick` or `linuxconsole`. Non of them are found by zypper16:22
malMister_Magister: the issue can be caused by some other error16:22
malvknecht: that doesn't explain why nobody else has had that issue16:22
Mister_Magistermal: strace
Mister_Magisterlast couple lines are repeated endlessly16:24
Mister_Magisterioctl(19, 0xc03864bc, 0x7706800270) = -1 EACCES (Permission denied)16:24
T42<elros34> @DylanVanAssche: long time ago I compiled it but not sure whether it is clean version (without my modifications) or it works at all:!AuDqiTFly4jxgxwkSC3gy05e1y9d?e=aAZKOt16:25
T42<DylanVanAssche> I can recompile it if needed :P but if it's already available somewhere :D16:27
T42<elros34> sure it is:)16:28
Mister_Magistermal: logcat
Mister_Magistererror -13 is EACCES16:29
Mister_Magisterare you sure problem is somewhere else?16:29
malyou have other failures there so maybe try to fix those and see if it helps16:33
Mister_Magisterwhat failures? don't see anything important16:34
Mister_Magisterthose in logcat are normal they doesn't matter16:35
Mister_Magisterioctl is the problem16:35
Mister_Magisterand it was working goddamnit16:36
Mister_Magistermal: interesting fact after booting phone i have failed to create composer client but after restarting it i have what you see there16:36
malqti failures16:37
Mister_Magisterthey don't matter16:37
Mister_Magisterthat's normal16:38
malno it's not16:38
Mister_Magisteri have same thing on fxtec and it has gui16:38
Mister_Magisterit's totally normal16:38
malqti and qmi are quite low level things that shouldn't usually fail16:38
Mister_Magisterthey fail on fxtec too16:38
T42<DylanVanAssche> @elros34 fftest needs to be modified (event3 instead of event0), running a sine wave effect makes the vibrator turning on!16:38
T42<DylanVanAssche> Wow finally something that does something :D16:39
malMister_Magister: well they don't on other devices so there is probably something the failures cause16:39
Mister_Magistersurely not blocking ioctl16:39
malMister_Magister: those are interprocess communication things which can cause bad things16:40
Mister_Magisteri have test_hwc2 working16:40
Mister_Magistermal: like i said they don't matter got test_hwc2 working16:40
Mister_Magisterthe difference is16:40
Mister_Magisteri masked droid-hal-init and started it after booting16:40
Mister_Magisterbut it runs only once…16:41
Mister_Magistermaybe that's why second tutorial was failing!16:41
Mister_Magisteryeajh same thing happening!16:41
malare you really not going to fix the errors ever?16:42
Mister_Magisterlater syre16:42
Mister_Magisterhmm surfaceflinger is fine though16:42
Mister_Magistertest_hwc2 isn't16:42
Mister_Magisteryeah test_hwc2 is not even displaying anything anymore16:43
Mister_Magisterwhat in the name of jesus is happening16:45
* Mister_Magister i don't need answers i need vodka16:47
Mister_Magistersaad that mccreary is deadish16:52
Mister_Magisterhmmm i ran out of andorid devs to help me with qti17:05
electro575how can i resolve this problem ? Warning[08/22 17:08:59] : repo problem: nothing provides needed by qt5-qpa-hwcomposer-plugin-,17:09
*** psachin is now known as psachin|away17:12
electro575sorry, it's good17:13
T42<DylanVanAssche> Any suggestions how to further debug the ngfd? The driver is functional (fftest sine wave), so it's definitely higher up the food chain.17:19
Mister_Magisterstrace your life whole way through17:24
Mister_Magisterstrace everything even working things17:24
r0kk3rzstrace till yoy cant strace no more17:37
Mister_Magistermal: any idea what creates rmnet_ctrl?17:37
Mister_Magisterwhat if i strace strace17:37
Mister_Magisterdouble power17:37
T42<DylanVanAssche> :P recursive strace?17:38
Mister_Magisterlike… if you do strace strace strace strace -p 1 you get 4 times the power of strace17:38
Mister_Magisterit's like more -vvvvv17:39
T42<elros34> better run with more verbose output:
mal@DylanVanAssche did you show the exact configs you have on device for ffmemless?17:44
T42<DylanVanAssche> @elros34 [better run with more verbose output: https://g …], ```[0.225] WARNING: ffmemless: ffm_setup_effects: invalid parameters17:45
T42<DylanVanAssche> [2.675] WARNING: stream-restore: failed to open connection to pulseaudio: Failed to connect to socket /run/user/100000/pulse/dbus-socket: No such file or directory```17:45
T42<DylanVanAssche> @mal I will post a pastebin with the configs17:45
mal@DylanVanAssche maybe you need to fix pulseaudio?17:45
T42<DylanVanAssche> Maybe... vibration is almost always connected to sound.17:46
T42<DylanVanAssche> @mal17:53
T42<DylanVanAssche> Config file pointer:
T42<DylanVanAssche> ngfd config file:
T42<DylanVanAssche> qtfeedback config file:
T42<DylanVanAssche> Pulseaudio hangs on registering with ofono:
electro575@elros34 : have you this repo to debug your device on ?18:19
T42<gischpelino> Where can I find the kernel sources Jolla uses for sailfish x for the geminipda?18:25
T42<elros34> electro575: I this repo has nothing to debugging, glibc dropped support for kernel < 3.2 so I patched it. I told you this already18:34
malelectro575: if you are doing the newer device then you won't need that18:36
electro575it's not a new device, it's the same device18:36
electro575i don't arrived to established connection with my samsung galaxy note II18:40
electro575my device enter in bootloop with enp0s29u1u2 not ready18:41
malso what are you trying to do now?18:41
electro575just established a telent connection18:41
electro575no telnet18:41
malso how did you break that?18:42
electro575this yes18:42
electro575ip address show dev enp0s29u1u1 && sudo ip address add dev enp0s29u1u1 && ip address show dev enp0s29u1u1 && sudo ip route add dev enp0s29u1u1 && ping -c 2 && telnet 232318:42
electro575i have remove all of my project, to restart all my debug18:42
electro575maybe i need this @elros3418:43
T42<elros34> what do you think?18:44
electro575i don't arrived to established telnet connection cause enp0s29u1u1 is not ready18:45
electro575i have tried this as my htc but, no way18:45
electro575activate usb0 for samsung18:46
deathmistheyy, haven't been on here for a while :)  just checked some logs and that PA droid hidl module sounds really promising for fixing my call audio issues once and for all (I hope)! Danct12: all you had to do was compile, include and load it in pa cfg right? how has it worked so far for you?19:01
T42<DylanVanAssche> I saw that the volume keys are ignored by MCE.19:06
T42<DylanVanAssche> After some debugging, I found out that this is the cause:
T42<DylanVanAssche> With mcetool I changed it the volume-key-policy to default and know tklock seems to be emitting the events. However, Lipstick isn't showing the bar to indicate a volume change19:06
T42<DylanVanAssche> *now19:06
T42<adampigg> Dylan,.nice work...if you are in tklock, try and figure out the lockscreen problem!19:27
T42<DylanVanAssche> Thanks :) maybe I get there :P19:29
T42<DylanVanAssche> Trying again to get the vibrator up and running atm19:29
electro575@elros34 : <elros34> @electro: I cheked your device repo and looks like your device use unusual way for enabling rndis20:00
T42<elros34> ? what you are trying to fix now?20:01
electro575no for telnet20:02
electro575<elros34> add this: write /sys/devices/platform/android_usb/usb_function_switch 4 after
electro575<elros34> also: echo "hybris-boot: initramfs" > /dev/kmsg" here and if it still reboots without any mer messages in host dmesg then grab last_kmsg20:02
electro575that's all20:02
T42<elros34> htc again?20:03
electro575no, samsung galaxy note II20:03
electro575but not sdk
electro575i need glibc ? or not20:04
electro575to do a telent connection20:04
T42<elros34> What do you think? is obvious20:05
electro575i have delete my project to do a new project for this device20:05
electro575it's not very obvious for me but20:06
electro575say it to me if you know20:06
electro575the sdk is too newer ?20:06
T42<elros34> ok if you really dont know, YES you need patched glibc because you have 3.0 kernel and glibc dropped support for it.20:07
electro575okey, sorry20:07
electro575@elros34 : this ?
T42<elros34> yes this one20:08
deathmistjusa: if you're on, can I completely remove audioflingerglue from my builds (didn't seem to do much anyways for me) when using PA droid-hidl module?20:17
maldeathmist: if you have the droid-hidl module you don'tt use audioflingerglue20:21
deathmistmal: alright, thanks for the clarification! this seems to be a recent development and I haven't been on here for a couple of weeks, I'll be trying to bring everything up-to-date again and possibly finally fix call audio via that PA module20:23
maldeathmist: which device is that20:23
deathmistOnePlus 5 (cheeseburger) on LOS 15.1 base20:23
maldeathmist: did hidl module help with call audio for you?20:25
deathmistmal: I have no clue, seemed to fix it for Danct12 tho so I have high hopes as he had seemingly the exact same issues I do currently. haven't built anything myself for almost a month soon, so I'll be getting everything back set up20:27
albertus1hmm. vknecht: git says "fatal: corrupt patch at line 13" (upon `git apply` with your diff), which is end of that file... am I missing anything obvious?22:05
albertus1CRLF line terminators from that "download" button on the paste site...22:08
albertus1okay, that's merely preventing `git apply` from complaining about trailing whitespace, but not about calling it a corrupt patch, which it still does. I'll apply the changes manually, so...22:10
albertus1okay, seems I have different line numbers22:12
albertus1oh dear :(
albertus1grepping logs found me this but no avail to `sb2 -t $VENDOR-$DEVICE-$PORT_ARCH -m sdk-install -R zypper rm geoclue-providers-hybris` (or -binder, or -hal in the end), it always says "'geoclue-providers-hybris' not found in package names. Trying capabilities." and then "No provider of 'geoclue-providers-hybris' found."22:25
vknechtalbertus1, afaik now it 'provider', not 'providers'22:26
albertus1okay. problem between keyboard and chair, thanks22:27
vknechtjust a tricky name change some time ago :-)22:28
albertus1I think I just typed that, and got it wrong22:28
albertus1the log grepping was with a different thing22:28
albertus1different package22:29
albertus1still getting me the old error22:29
albertus1what's this rpm cache thing all about?22:31
albertus1and why do -hal and -binder conflict with each other now?22:32
vknechtmight have to remove the non-binder package from local repo, not sure clean cache really useful22:32
albertus1oh, the spec file of -binder says "Conflicts"22:34
vknechtthe conflit is because both do the same thing, but with different API22:34
albertus1hmm. why is this now a problem and wasn't before?22:35
vknechtcould be the start of the problem is you built both: selecting a spec file is preferable I think ; the binder one22:35
albertus1hmm, now that returned immediately22:37
albertus1i.e., `rpm/dhd/helpers/ --spec=hybris/mw/geoclue-providers-hybris/rpm/geoclue-providers-hybris-binder.spec`22:37
albertus1maybe nothing to do because it has already been built22:38
albertus1so just regenerate patterns and retry mic now?22:38
vknechtguess so, especially if a SB2 zypper se geoclue-providers-hybris returns nothing now22:42
albertus1hmm, only one conflict remaining in mic: file /lib/systemd/system conflicts between attempted installs of droid-hal-pioneer-0.0.6-201908191403.armv7hl and systemd-225+git16-1.6.1.jolla.armv7hl22:48
albertus1it seems your approach has "solved" the other conflicts (by not including the conflicting files in one of the conflicting packages)22:48
albertus1(if I understand it correctly)22:49
albertus1so likewise editing the specfile (or its include) for droid-hal-pioneer?22:49
vknechtcan't see where similar treatment could be done in hal, might be editing systemd package instead is required, dunno ; and perhaps that should be solved some other way22:54
albertus1but also the other thing only looks like a workaround to me22:55
albertus1where do these conflicts come from in the first place?22:55
vknechtcould be that newer version of RPM is more touchy ?22:56
albertus1what does "touchy" mean?22:56
vknechtmore strict about what it accepts22:57
albertus1RPM version 4.14.1 on that PlatformSDK22:57
albertus1which is not exactly new22:58
vknechtmaybe forking for local build and editing (and probably add the files in there, checking with rpm -qpl systemd.rpm)23:07
vknechthmm, no this line is about sysconfdir, not lib (like in /lib/systemd/system)23:10
* vknecht gives up for now, good night :-)23:10
albertus1isn't there a possibility to just remove that one particular conflicting file from either of the rpms?23:10
albertus1me too, good night and thanks23:10

Generated by 2.17.1 by Marius Gedminas - find it at!