Tuesday, 2020-06-02

rinigusspiiroin: morning! as I am going to test updated fingerprint driver to check if they fixed remove(0), I have a chance to inject response to double tap on FP here https://github.com/sonyxperiadev/vendor-sony-oss-fingerprint/blob/master/fpc_imp_yoshino_nile_tama.c#L57508:05
rinigusspiiroin: if I want it to be hooked to doubletap at mce level, seems like KEY_POWER would be OK choice?08:06
riniguslooks like that and MENU, BACK & HOMEPAGE are checked. or is there something better to consider?08:06
spiiroinrinigus: power key events do work, but going via gesture events allows immediate configurability...  e.g. mce assumes EV_MSC:MSC_GESTURE:4 is display double tap; then there is 16 = fingeprint wakeup; config info via: mcetool set-display-on-single-powerkey-press-actions set-touchscreen-gesture-actions09:03
spiiroinif that low level fp tap detection uses less power than keeping sensor in identify -> I'd perhaps make it use that gesture(16)09:04
spiiroinwhich would make it work pretty much similarly as fpwakeup already does09:05
rinigusspiiroin: I don't know much about power use. by default, when not in identify, it does pull gestures though09:05
rinigusso that's probably assumed to be rather small in terms of power consumption09:07
rinigusso, to verify: EV_MSC:MSC_GESTURE:16 is recommended in this case?09:08
spiiroinrinigus: diff between gestures and power key: latter differentiates behavior / display state -> can be used for unblank and blank; gestures are assumed to be wakeup signals09:08
spiiroinso depends a bit how you want to use it09:09
spiiroinEV_MSC:MSC_GESTURE:16 = if you want to make it compatible with mce fp wakeup; but you can as well take some other gesture code (<=20 iirc) and configure it according to your needs09:10
rinigusspiiroin: mainly to unblank the screen when pickup gesture did not fire. although, it maybe useful to switch the screen off (latter probably requires power key). I'll have to experiment and will start with 16.09:11
spiiroinrinigus: in case you've not seen it - there is some more info at https://together.jolla.com/question/212934/great-wake-up-with-fingerprint-but/?answer=214202#post-id-21420209:12
rinigusspiiroin: I saw it a while ago - thanks for the reference!09:14
rinigusspiiroin: but the current fpwake does it bring fp reader into identify?09:15
spiiroinrinigus: yes, but it is not used for authentication. and you probably need to enable it separately - iirc it is enabled only on select devices09:16
spiiroini.e. performs very similar function as what can be accomplished if driver would emit gesture events in any device that supports fp sensor09:18
rinigusspiiroin: so, in my case, I would have to emit EV_MSC:MSC_GESTURE:16 on android side on double tap and then not enable indentify pulling by mce. or just emit :4 and then will not confuse users09:18
T42Theresa %lastname% was added by: Theresa %lastname%09:50
T42<edp_17> @elros34: You advised me to remove TvoutService_C from droid-hal-init as it is crashing. The /sbin/droid-hal-init is an elf executable, so I am not sure how I can remove this TvoutService_C. Can you let me know, please?09:58
T42<edp_17> mal: You wrote the reason of 'ERROR! Can't connect to RILD: No such file or directory' error means "the android side service is not running properly" and the reason of this could be "either missing or broken mounts or not found firmware files". Can you tell me how I can check whether the mounts are missing/broken and the missing firmwa10:22
T42re files, please?10:22
T42<elros34> in one of *.rc file not in init binary10:35
T42<edp_17> @elros34: Oh, I see. Thanks. It was in init.universal5433.rc10:43
T42<edp_17> I have tried to add into defconfig: CONFIG_MEMCG_SWAP=y CONFIG_MEMCG=y CONFIG_MEMCG_KMEM=y CONFIG_AUTOFS4_FS=y10:46
T42<edp_17> but got an error: https://paste.ubuntu.com/p/cz7QvvGD7b/10:46
T42<edp_17> I have checked all dependecies of those flags and all are fine. They also fine (the same) in .config file too.10:47
T42<edp_17> I am trying to fix the crashing rild by the way.10:48
riniguspiggz: thanks for merging - still away from sources till later tonight12:10
T42<ankaos> wlan fiexd12:45
T42<ankaos> @edp_17 @elros34 @Verevka8612:45
T42<ankaos> thank youuu12:45
T42<KernelPanix> @ankaos [@edp_17 @elros34 @Verevka86], hello12:45
T42<KernelPanix> I Can Disable12:46
T42<KernelPanix> CONFIG_BT_HCIUART12:46
T42<KernelPanix> ?12:46
T42<ankaos> CONFIG_BT_HCIUART=y ı use12:52
T42<KernelPanix> @ankaos [CONFIG_BT_HCIUART=y ı use], errors build12:53
T42<KernelPanix> :(12:53
T42<ankaos> error log please12:54
T42<ankaos> browser app not work. :(13:20
T42<edp_17> @KernelPanix: To disable that just comment it out in defconfig like: # CONFIG_BT_HCIUART is not set13:57
T42<edp_17> However this probably will need for bt.13:57
T42<KernelPanix> @edp_17 [@KernelPanix: To disable that just comment it …], i can Add To Default Defconfig ( or Only Out Defconfig )13:57
T42<edp_17> @KernelPanix: Please don't use reply! You should comment it out in the default defconfig.13:59
T42<KernelPanix> Thanks U @edp_1714:00
T42<edp_17> @KernelPanix : Np. You should use a much shorter nick by the way. 😊14:01
T42<edp_17> or less complex at least.14:01
T42<KernelPanix> @edp_17 [or less complex at least.], Yeah ( Like Name Kernel Panic )14:01
T42<neochapay> https://www.youtube.com/watch?v=GXVeLhWmsY814:40
T42<neochapay> https://www.youtube.com/watch?v=2ag7KHinMSM14:40
riniguspiggz: I can continue. thanks for moving Clear out - wasn't aware its not a part of the API15:14
T42<birdzhang> rinigus: it's working on vince, lel15:15
T42<birdzhang> https://imgur.com/gallery/5WexGpG15:15
T42<birdzhang> btw, i use rpms from your obs, too lazy to build them, xD15:16
riniguspiggz: anything outstanding?15:18
T42<KernelPanix> @edp_17 hello15:18
T42<KernelPanix> @edp_17 /home/Nobita/hadk/kernel/xiaomi/whyred/drivers/bluetooth/hci_ldisc.c:483:20: error: 'struct hci_uart' has no member named 'rx_lock'15:18
T42<KernelPanix> @KernelPanix [@edp_17 /home/Nobita/hadk/kernel/xiaomi/whyred …], this log15:18
piggzbirdzhang: great15:19
T42<elros34> please stop, use pastebin15:19
piggzrinigus: im about to start on 1715:20
T42<edp_17> @KernelPanix: Please use pastebin or paste.ubuntu.com and provide full log, not only pieces.15:21
T42<KernelPanix> okay15:22
T42<KernelPanix> @edp_17 https://paste.ubuntu.com/p/ttyZJSWPCP/15:23
T42<KernelPanix> this log15:23
piggz rinigus: correct, i _was_ about to start, now need to go and collect car15:24
T42<edp_17> @KernelPanix: Lol, have you checked what you pasted? There's one line only. 😊15:25
T42<KernelPanix> @edp_17 [@KernelPanix: Lol, have you checked what you p …], :(15:25
T42<KernelPanix> @edp_17 [@KernelPanix: Lol, have you checked what you p …], Idk Copy And Past Only15:25
T42<KernelPanix> i can give screenshot to web upload ?15:26
T42<KernelPanix> @KernelPanix [i can give screenshot to web upload ?], @edp_1715:26
T42<edp_17> @KernelPanix: Just select the text from the console, then right click and copy then paste into paste.ubuntu.com.15:27
T42<KernelPanix> @KernelPanix [@edp_17 https://paste.ubuntu.com/p/ttyZJSWPCP/], like this15:28
T42<KernelPanix> @edp_17 https://ibb.co/vV6xPMt15:30
T42<KernelPanix> :( i can't Copy And paste15:31
T42<ankaos> https://paste.ubuntu.com/p/xkNzxKHZ8b/ ofono :(15:34
T42<KernelPanix> @elros34 [that is your error: ../../../../../../kernel/m …], yes. yes15:42
T42<KernelPanix> @elros34 [that is your error: ../../../../../../kernel/m …], this errors15:42
T42<KernelPanix> that is your error: ../../../../../../kernel/motorola/sdm660/drivers/bluetooth/hci_ldisc.c:483:22: error: no member named 'rx_lock' in 'struct hci_uart'15:43
T42<elros34> why do you even care about bluetooth at this stage? Do you build it on device or what? I see no point in porting sailfish without pc with usb to debug things15:45
T42<KernelPanix> @elros34 [why do you even care about bluetooth at this s …], my defconfig15:46
T42<KernelPanix> https://github.com/Linux-On-Whyred/mer-kernel-check/blob/hybris-16.0-kernel-4.4/check-kernel-config15:46
T42<KernelPanix> i will try remove BT HCI*15:47
T42<elros34> I suggest you to follow hadk and hadk-faq and not use some random scripts15:49
T42<KernelPanix> @elros34 [I suggest you to follow hadk and hadk-faq and …], Okay15:50
T42<KernelPanix> I Make This By HADK . Auto Config Defconfig ( Like Halium )15:50
zinstackHello, am trying to make it display something, at the moment all I can see is vendor image and a lot of binder transaction failed messages in dmesg and journalctl. Also there is failure mounting cpuctl, but that should only affect deep sleep. My device has QC sdm632, I believe it's caf/msm8996. test_hwcomposer just hangs, strace reports openat /dev/binder. I guess there is something wrong with binder, but what16:06
riniguspiggz: trying to figure out where some error messages disappeared ... something is wrong, will look into it - hopefully my setup16:10
rinigus@birdzhang: that's why you have so old test app :). congratulations!16:11
riniguspiggz: all is fine over here, just had to complain. probably compiled older version.16:14
rinigus@birdzhang: rebuilding at OBS, let's see if it will go through16:17
riniguspiggz: I will take #13. some really low hanging fruit :)16:27
T42<adampigg> (Photo, 2560x1920) https://irc.thaodan.de/.imgstore/lYoYSInbfx.png16:57
T42<adampigg> One road legal renault 5 :D16:57
piggzrinigus: will fixing 17 also mean 5 can be closed, as, we have some docs now, and the folder handling will be moved into the daemon, and nothing will be needed on the droid-init side?17:04
piggzrinigus: when you get reboot, what is the cause of it?17:05
piggzsailfish or android?17:05
riniguspiggz: regarding 5 - yes, sure. 17 will close 5 as well17:05
riniguspiggz: regarding reboot - no idea. its all freezes, no journal, no logcat17:06
rinigusbut maybe we should ignore it for now... let's see if other devices will get similar17:07
rinigusI'll try to see if sony's changes in FP module would clear #217:08
piggzrinigus: have you tried tap to wake yet?17:11
riniguspiggz: no. but if I can make new sony module to work, I will be able to inject that as well. should be very easy with their source17:12
piggzrinigus: i was looking at how fp is implemented on mido, which is los14 .... different api, so would need a new android library and new implementation of androidfp17:13
rinigusproblem is that my last build was for other device, so most of aosp has to be rebuilt. so, will probably know tomorrow.17:13
piggzi may look into that later17:13
piggzrinigus: for the fingerprint group id, we should use the userid ?17:20
riniguspiggz: I think so. That should be unique17:25
rinigusooo, with `mmm` fp module was ready in 2 minutes. let's see whether it works...17:30
T42<erfanoabdi> @piggz [rinigus: for the fingerprint group id, we shou …], Group id should be always 017:38
riniguspiggz: see above. I probably missed the question - I thought it was regarding userID in the file path for stored FPs17:41
piggz@erfanoabdi why is that?17:41
T42<erfanoabdi> piggz: fp hal crashes for me on 3 devices when i used non 0 value for gid17:42
T42<erfanoabdi> Same as Android17:42
piggzi was  thinking that if you used different group id;s for each user, then their data would get sttored seperatley17:42
piggzbut, i guess we have to rely just on differnt paths17:43
T42<erfanoabdi> Yeah exactly i wanted to do that and realized not accepting non 017:47
riniguspiggz: new sony module clears data correctly. and our daemon does not clear it correctly. will fix it for our daemon as well18:00
piggzrinigus: gr8 ... should we go with the gid=0 limitation then?18:01
rinigus(our module map stays bit off)18:01
riniguspiggz: yes, let it be 018:01
rinigusand if we run into trouble, it will be future piggz and rinigus problem. let them deal with it18:02
T42<adampigg> @erfanoabdi if we choose to use completely differnent paths for setActiveGroup, can we ignore that checking of api level and using the android folders?18:44
T42<adampigg> that isnt know to cause problems is it?18:44
T42<erfanoabdi> @adampigg [@erfanoabdi if we choose to use completely dif …], First i tried to use my own path for active group but some phones qseecomd was looking at that specific path and results in enroll failure18:46
T42<erfanoabdi> (Op5 and zenfone6)18:46
T42<adampigg> rinigus: ugh ^^18:46
rinigus@adampigg: I think its an issue of that phone, not ours. let's see if it is more widespread and use our own path. unless android api demands the path18:48
rinigusI would expect that whatever we do, something will be wrong on some devices.18:49
rinigustry to see if it works for your device.18:50
riniguspiggz: will test on mine18:53
T42<erfanoabdi> Custom path worked for example on moto z2 force19:00
T42<erfanoabdi> But i don't think it will work on newer devices19:00
riniguspiggz and @erfanoabdi: works on sony xz2 (aosp9)19:02
T42<erfanoabdi> https://github.com/LineageOS/android_hardware_interfaces/blob/lineage-17.1/biometrics/fingerprint/2.1/vts/functional/VtsHalBiometricsFingerprintV2_1TargetTest.cpp#L21719:02
T42<erfanoabdi> Google does the same check in VTS19:02
rinigusso I could use /var/lib/sailfishos-fpd-community/100000/fpdata/user.db to enroll, identify and so on19:03
rinigusmaybe devices not supporting custom path would be ok with changing gid?19:04
T42<erfanoabdi> Nope i don't think so19:04
rinigusworked after reboot as well, btw19:04
T42<erfanoabdi> Are u doing this just to bring up multiuser support?19:05
riniguserfanoabdi: yes.19:05
rinigusits not there yet, but there was a warning to expect it19:05
T42<erfanoabdi> there's other thing we can do but dirty19:07
T42<erfanoabdi> Creating a file in fpdata and write down fingerid with enrolled user id19:07
T42<erfanoabdi> And on auth we check for that fp id and uid19:07
rinigus@erfanoabdi: yes, that could be done. or keeping separate maps for different users. but we are then still limited to 5 fingers in total19:08
T42<erfanoabdi> @rinigus [@erfanoabdi: yes, that could be done. or keepi …], yeah19:09
riniguspiggz: does the custom path work for you?19:17
piggzrinigus: so, what do we choose as the least worst option?19:21
piggzstandard android folder, with limit of 5 fingers, and map per user?19:22
riniguspiggz: obviously, we don't know when multiuser is going to come. so, in theory, we can keep 'standard' folder as well and have maps that are separate - when the multiuser will come.19:28
rinigusbut I would suggest to make that 'standard' folder in our daemon.19:28
rinigusmaybe we could even just have a config option which switches to the android standard folder if needed by that device19:29
rinigusor command line option, as it easier.19:29
riniguspiggz: hmm, in this respect, we can go with the original plan and then have cmd option to set folder to `/data/system/users/0/fpdata` or `/data/vendor_de/0/fpdata/` that porter will be able to specify in systemd unit19:31
piggzhmm, ok19:32
piggzsounds like a plan19:32
riniguspiggz: would that be ok?19:33
T42<erfanoabdi> how do u want to check uid if fpdata isn't in one user dir19:36
T42<erfanoabdi> For example if data is on ../users/100/..  and ../users/101/..19:36
T42<erfanoabdi> Set active dir is only set once and Auth function and others will act based on that dir19:36
T42<erfanoabdi> We can't call auth 2 times19:36
T42<erfanoabdi> Sorry my English is bad i hope explained correctly19:36
T42<erfanoabdi> but systemd plan is good imo19:37
piggz@erfanoabdi can we not call setActiveGroup > 1 per run of the daemon?19:40
T42<erfanoabdi> Piggz: U can call it, but how are u gonna call the auth function if you don't know the user id yet19:43
T42<erfanoabdi> I'm comparing with biomertyd bot sure how Sailfish fpd works, but there we have identifier function that recognizes user id based on authorized finger19:44
piggzwe should be ok .... we would call stActiveGroup on a (currently theoretical) user change .... that user would enroll, and we would store the userid against the fingerid .... or something like that!19:47
T42<erfanoabdi> great19:48
riniguspiggz, @erfanoabdi: don't forget we are talking about something that will happen after soon^TM. don't think they were suggesting it will happen in the near future19:50
rinigusso, we should make it work well for one user and just have an eye for multiuser storage. so, that in X-XX years we can switch19:51
T42<erfanoabdi> Yeah cool19:51
T42<erfanoabdi> I wanted to know how to do it correctly on biomertyd too19:51
T42<erfanoabdi> Because currently I just hardcoded userid.. :(19:51
riniguspiggz: just sent PR ensuring that Clear works as expected19:52
rinigus@erfanoabdi: depends on how you switch the user. if user first selects login name and then auth, it should be relatively simple to switch the databases19:52
T42<erfanoabdi> Yeah but when i checked 'user.current()' on where auth function called wasn't same as the enrolled userid19:56
T42<erfanoabdi> For me only way is another map file to store uid, fid19:57
T42<adampigg> @erfanoabdi  thats what im doing! :)20:14
T42<erfanoabdi> Nice thanks20:14
T42<ankaos> Hı21:24
T42<ankaos> ✋21:24
T42<KernelPanix> @erfanoabdi [Nice thanks], u make GSI SailfishOS ?21:47

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!