Wednesday, 2020-05-20

T42MG %lastname% was added by: MG %lastname%00:32
T42Paul Fishback %lastname% was added by: Paul Fishback %lastname%01:20
*** ahjolinna_ is now known as ahjolinna06:41
T42<edp_17> @elros34 : Thank you for thinking about that! I have tried some other things but none of them really worked. 😔 I'll check /init*rc and /proc/deferred_initcalls.  Thanks!07:06
T42姚桦敦 %lastname% was added by: 姚桦敦 %lastname%07:08
T42<edp_17> There is something else which is weird. The device doesn't recognize when I plug the cable in. It charges the phone but to be able to get into it with ssh, it requires a reboot while the cable is plugged in. If I just plug in without a reboot, cannot be accessed via ssh. Is it something to fix on the android side, or something in Sailf07:09
r0kk3rzusb doesnt touch any android things07:10
T42<edp_17> r0kk3rz: Thanks. Is it a service?07:11
r0kk3rzusb moded07:12
r0kk3rzis the thing that handles it, and kernel drivers07:12
r0kk3rzso check dmesg07:12
T42<edp_17> Thanks!07:13
T42<edp_17> r0kk3rz: The dmesg is full of 'usb_init_communication: com_state=0, open fail' and it looks like somehow connected to the rild issue (the SIM card is not detected):
r0kk3rzit shouldnt have anything to do with rild07:25
T42<edp_17> Okay, it is just red herring then.07:27
r0kk3rzthe modem could be attached to an internal usb07:27
T42<edp_17> Oh, I see.07:31
T42<edp_17> @elros34 : I have checked all init*.rc for deferred_initcalls but didn't find anywhere. However I have a init.wifi.rc :
T42<edp_17> There is a wifiloader and macloader services. Should I mask bot or any of them and try to start only when the partitions are mounted?07:35
T42<edp_17> r0kk3rz : I am a little bit lost in dmesg as am not sure what I need to look for to see the problem. When I log in via wifi and issue the command usb_moded, I get this:
T42<Cornelius %lastname%> (Photo, 851x477) 💥Telegram群,微信群,potato拉人,增粉,10万个群群发广告,私发广告.09:37
T42<Cornelius %lastname%> 💥出售微信号,查不到IP,无需实名手机卡,水军群内聊天吹捧.09:37
T42<Cornelius %lastname%> 💥非小号,Mytoken关注,排名,四件套,支付宝出售.短信营销.09:37
T42<Cornelius %lastname%> 💥各大APP交易所代实名认证,国外实名认证,KYC出售.09:37
T42<Cornelius %lastname%> 💥币圈朋友圈广告,币圈推广增粉,上币投票.币圈各种业务.09:37
T42<Cornelius %lastname%> 💥电报自动聊天软件,更多业务 请咨询  @uuana09:37
T42<Odelette %lastname%> (Photo, 851x477) 💥Telegram群,微信群,potato拉人,增粉,10万个群群发广告,私发广告.09:37
T42<Odelette %lastname%> 💥出售微信号,查不到IP,无需实名手机卡,水军群内聊天吹捧.09:37
T42<Odelette %lastname%> 💥非小号,Mytoken关注,排名,四件套,支付宝出售.短信营销.09:37
T42<Odelette %lastname%> 💥各大APP交易所代实名认证,国外实名认证,KYC出售.09:37
T42<Odelette %lastname%> 💥币圈朋友圈广告,币圈推广增粉,上币投票.币圈各种业务.09:37
T42<Odelette %lastname%> 💥电报自动聊天软件,更多业务 请咨询  @uuana09:37
T42<elros34> You never start services just like that. First stop previous instance then check what arguments it use in usb_moded.service and add some debug args.09:41
T42<elros34> About wifi: based on sources wifiloader basically just cat /proc/deferred_initcall. First read from it should return 0. You can check whether it returns 1 ( it means wifiloader started) but not sure it will be useful as it's probably already started when partitions are ready09:43
T42<elros34> So maybe the easiest way would be to set mac via ifconfig (or something) from /efs/wifi/ This is what macloader use generate /data/.cid.info09:45
T42<elros34> maybe is created by kernel driver, not sure09:49
T42<edp_17> @elros34 : Thanks. The 'cat /proc/deferred_initcall' returns with 'can't open '/proc/deferred_initcall': No such file or directory'. This is probably because of all partitions are mounted but I am not sure.09:49
T42<elros34> But file exist, right? What if you get rid of wifiloader09:51
T42<edp_17> I think the mac is always the same. I noted the mac a few days ago from /efs/wifi/ and now it has the same value. So the mac is not changing after reboot.09:51
T42<edp_17> Nope, the file deferred_initcall doesn't exist: ls: /proc/deferred_initcall: No such file or directory09:52
T42<elros34> hm, as far as I understand that file should be there, there even config in your kernel which enable it10:01
T42<edp_17> Well, it seems I was a bit khm. careless. I do not have '/proc/deferred_initcall' but I do have '/proc/deferred_initcalls' and it has a value 1.10:06
T42<edp_17> Sorry.10:06
T42<elros34> my bad, but anyway you should use tab completion10:08
T42<edp_17> No worries. 😊10:09
T42<edp_17> However it seems the issue is not the mac address changes.10:10
T42<edp_17> Do you how how and where the system stores the wifi networks and passwords?10:11
T42<edp_17> It might be that for a reason that is cleared out on reboot.10:11
T42<elros34> so does triggering initcalls manually (it should return 0) instead in wifiloader helps in anyway?10:18
T42<edp_17> I don't know how to trigger initcalls manually.10:20
T42<elros34> reading from that file trigger them, return value it is just indication whehter it's first call10:21
T42<edp_17> So should I do a 'cat /proc/deferred_initcalls' before the wifiloader in init.wifi.rc?10:24
T42<elros34> no, like I said get rid of wifiloader, reboot, wait for a while, wifi should be unavailable, then trigger deferred calls (it should return 0)10:30
T42<edp_17> Okay, I got it now.10:30
T42<edp_17> I have commented out line 44, 45 and 46 here:
T42<edp_17> After reboot the wifi was off. When I did a 'cat /proc/deferred_initcalls' the device rebooted. I have tried twice and it happened every time.10:37
T42<elros34> so maybe setting mac via ifconfig is the best way10:42
T42<edp_17> Now I rebooted again and did a 'file /proc/deferred_initcalls' and it gave me '/proc/deferred_initcalls: empty'10:43
T42<edp_17> when I tried to echo 0 into it, I got: 'write error: Input/output error'10:43
T42<elros34> I guess empty means 0 so it's was correct but if it doesn't works everytime then forget about it10:45
T42<edp_17> How do you mean 'doesn't work everytime'?10:47
T42<elros34> you said cat caused reboot twice10:48
T42<edp_17> Oh, I see. Yes it rebooted. Then I tried to run '/system/bin/wifiloader' but the result was also a reboot.10:56
T42<edp_17> I un-comment those lines as this doesn't seems to be working.10:56
T42<elros34> something similar to: ifconfig wlan0 hw ether $(cat /efs/wifi/ could work if that file doesn't change11:00
T42<edp_17> Unfortunately this doesn't work ether. I switched wifi off, then rebooted.11:06
T42<edp_17> Then I issued the command 'ifconfig wlan0 hw ether $(cat /efs/wifi/' but it didn't bring up wifi. I turned wifi on, but it still required to enter the pasword.11:07
T42<edp_17> I have tried when the wifi was on, then rebooted.11:08
T42<edp_17> The result was the same.11:08
T42<elros34> so how about disable wifi first, check mac in ifconfig, set it manually and check it again11:09
T42<edp_17> When the wifi is disabled (off) the wlan0 doesn't exist in ifconfig, hence cannot see the mac.11:12
T42<elros34> ifconfig -a11:13
T42<edp_17> Oh. 😳11:14
T42<edp_17> Wifi off: 90:B6:86:4C:CA:B611:14
T42<edp_17> I issued command: 'ifconfig wlan0 hw ether 90:B6:86:4C:CA:B1'  (change the last bit)11:15
T42<edp_17> ifconfig -a again: 90:B6:86:4C:CA:B611:15
T42<edp_17> It didn't set what I gave (B1).11:16
T42<edp_17> In the 'init.wlan.rc' there is: '-iwlan0 -Dnl80211 -c/data/misc/wifi/wpa_supplicant.conf'11:29
T42<edp_17> However the '/data/misc/wifi/wpa_supplicant.conf' doesn't exist.11:30
T42<elros34> I doubt this is enabled and started in sailfish. We use connman11:30
T42<edp_17> I see. I am trying to find out how/where Sailfish stores the wifi password but I am not sure that would take me anything closer to the solution.11:32
T42<edp_17> I am not sure whether we can get closer to this so can I ask you to help me with something else, please? With this:
T42<elros34> You can read your wifi kernel sources. About usb as always get some logs, Can't you ssh via wifi?11:41
T42<edp_17> I can but it's very slow. Plus if I logged in via usb I cannot do via wifi and vice verse.11:45
T42<edp_17> Okay, how can I stop the usb_moded.service? It is set to always restart.11:54
T42<edp_17> Interesting. Anything I set as Default USB mode only takes account after reboot. However if I set 'Always ask' and I issue the 'killall usb_moded' command, the options to select a USB mode pops up.12:18
T42<elros34> last idea about wlan is to add -DREAD_MACADDR to EXTRA_CFLAGS in drivers/net/wireless/bcmdhd4358/Makefile12:30
T42<edp_17> Added. I'll try once the new kernel is done and report.12:33
T42<edp_17> How can I check what event gets fired when I change the Default USB mode? (The udevadm monitor doesn't show that.)12:33
T42<elros34> maybe also in drivers/net/wireless/bcmdhd/Makefile12:34
T42<elros34> verbose log from usb-moded could be useful12:35
piggzrinigus: step 1 - build, check13:01
T42<edp_17> @elros34 : Adding the '-DREAD_MACADDR' into both Makefiles (bcmdhd4358 and bcmdhd) unfortunately didn't solve the wifi issue. It still requires to enter the password after every reboot. Thanks for you help anyway. Maybe later I'll resume to it but now I think should step forward to solve some other stuff.13:13
T42<edp_17> @elros34 : How can I get verbose log from usb_moded? If I run 'usb_moded -sT' I get :
T42<edp_17> If I killall usb_moded, it restarts automatically.13:19
T42<elros34> don't kill it use systemctl, use all default arguments from systemd service, you are missing at least 113:22
riniguspiggz: sounds good! we will probably have to look into guts of the authentication system. there was a list of open/closed components somewhere. as far as I remember, bunch of auth  stack was closed source13:28
rinigusmal: is there a plan to review sensorfwd and qtsensors merge requests related to compass?13:28
riniguswho should I ping and ask for review?13:29
piggzrinigus: it has a test feature wher you can enroll etc on the command line13:42
piggzbut its crashing13:42
riniguspiggz: maybe tries to hook to some specific fp reader?13:43
piggzrinigus: i dont think so, there is an android side that you build which is a hidl hooks into that..13:47
malrinigus: I will review the changes when I have time13:47
rinigusmal: thanks!13:48
riniguspiggz: I must say I know too little 'bout FP stack and how its hooked. will have to read about it. few things have to be finished though. maybe towards next week, assuming that I find current bugs fast13:50
piggzsure, you be good to bounce ideas!13:51
piggz[root@Pro1 nemo]# biometryd enroll --device=android13:51
piggzsetActiveGroup failed: SYS_EINVAL13:51
piggzStarting template enrollment for User[0] to Android HAL Bridge13:51
riniguspiggz: have you asked about it on ubports channel? I am sure they will be happy to get more devs working on the same stack13:52
piggzi havnt, good idea13:52
T42<edp_17> @elros34 : The systemctl stop usb_moded gives me : 'Failed to stop usb_moded.service: Unit usb_moded.service not loaded.' Probably this is the reason why changing the Default USB mode doesn't make any affect.13:56
T42<elros34> ls /lib/systemd/system/ and find correct name for that service13:57
T42<edp_17> There is only one with usb in its name: usb_moded.service13:59
T42<elros34> not really, it's usb-moded.service14:00
T42<edp_17> Yeah, with a dash rather than underscore. 👍14:01
T42<edp_17> This is what I got:
T42<edp_17> When I ran the exact same command that is in the usb-moded.service, I didn't get any message to the console.14:09
T42<edp_17> However running it with only -sTl parameters did pop up the selection and then did set the selected usb mode.14:10
piggzrinigus: atleast it does "something" ... wheni set the enroll command, logcat says
riniguspiggz: indeed, there is a reaction... I wonder how it will work with sony's
piggzrinigus: feel free to try!15:11
riniguspiggz: later. :)15:11
T42<erfanoabdi> @piggz [rinigus: atleast it does "something" ... wheni …], don't use enroll command it's broken15:14
T42<erfanoabdi> `biometryd test --config=test.json`15:14
T42<erfanoabdi> test.json:15:14
T42<erfanoabdi> `{"device": {"id": "android"}}`15:14
rinigusspiiroin: I ended up writing tiny daemon that listens to pickup sensor and asks mce, politely, via dbus, to turn on the screen on trigger. then stops sensor on screen on and switches it back on off.15:18
riniguson sensorfwd side, had to enable wakelock as for proximity at
rinigusthe daemon code at
T42<adampigg> @erfanoabdi that got a little further than the last time i tried the test command
T42<adampigg> @erfanoabdi oh, and youre the dev, right person to be asking!15:22
T42<erfanoabdi> @adampigg [@erfanoabdi that got a little further than the …], Please send me logcat15:22
T42<adampigg> @erfanoabdi this is the logcate from the test command
T42<adampigg> im popping out for 1.5 hours, speak to you later15:24
T42<erfanoabdi> @adampigg [im popping out for 1.5 hours, speak to you lat …], Oh error isn't on fp hal15:25
T42<erfanoabdi> `GatekeeperHalDevice: resp->status: -24`15:25
T42<adampigg> i note that biometryd run always prints setActiveGroup failed: SYS_EINVAL15:25
T42<erfanoabdi> @adampigg [i note that biometryd run always prints setAct …], that will be problem later in post enroll15:26
T42<erfanoabdi> fix is:
T42<adampigg> @erfanoabdi also, pls check this commit in case i messed something up
T42<erfanoabdi> @adampigg [@erfanoabdi also, pls check this commit in cas …], it's fine15:28
T42<erfanoabdi> check dmesg to see if qseecomd is giving error?15:28
T42<erfanoabdi> whats your device keymaster hal version? 3 or 415:28
T42<adampigg> i guess 4 /vendor/bin/hw/android.hardware.keymaster@4.0-service-qti15:28
T42<erfanoabdi> keymaster4 needs this hack:
T42<erfanoabdi> build fake_crypt, it will run on boot it self15:29
T42<adampigg> (this isnt helping me go for a bike ride :D )15:29
T42<adampigg> uk wil try that later15:29
T42<erfanoabdi> kk ;)15:29
spiiroinrinigus: nice, I'll try to find time to take a closer look at it16:24
rinigus... and a bunch of sensors landed in released tama port: compass fixed, step counter & pressure; + pickup gesture turns on the screen (not 100%, usually with a delay, but mostly works)17:36
rinigusin particular, step counter with kimmoli's stpcntrd is true mobile linux experience17:37
rinigusyou have to use dbus (visual dbus app) to check the number of steps. what can be better?17:38
T42<ankaos> ninja: error: 'libhwc2_compat_layer', needed by 'hybris-hal', missing and no known rule to make it18:16
T42<ankaos> where is this hwc?18:16
T42<elros34> abranson: This commit was reverted but this is still issue in I can't ssh from sailfish to sailfish with chacha20-poly1305@openssh.com18:40
T42<adampigg> @erfanoabdi ive install fake_crpyt and the filesystem changes .......19:15
T42<adampigg> i no longer get an error on startup19:15
T42<adampigg> test still fails19:15
T42<erfanoabdi> gatekeeper error -24?19:15
T42<adampigg> 05-20 20:14:54.069  4472  4671 D btl_algo: btl_fingerprint 82:HAL notify, msg:1, ext1:2, ext2:0.19:16
T42<adampigg> 05-20 20:14:54.069  4472  4671 D android.hardware.biometrics.fingerprint@2.1-service: onError(2)19:16
T42<adampigg> that ?19:16
T42<erfanoabdi> @adampigg [that ?], `E GatekeeperHalDevice: resp->status: -24`19:17
T42<adampigg> there is no -24 in logcat19:17
T42<adampigg> looks like something else, will psot og19:17
T42<adampigg> s/post log19:17
T42<erfanoabdi> thing is the preenroll token should match the enroll challenge, if it's not gatekeeper hal doesn't do the job properly19:17
T42<erfanoabdi> `QSEECOMAPI: Error::send command ioctl failed. ret = -1, errno = 22`19:19
T42<erfanoabdi> check loaded hals with `lshal` and make sure gatekeeper and keymaster are on the list19:19
T42<erfanoabdi> after that make sure qseecomd service is loaded19:20
T42<adampigg> all seem to be19:22
T42<adampigg> system    3504  0.0  0.0  32076  4000 ?        Sl   20:11   0:00 /vendor/bin/qseecomd19:23
T42<erfanoabdi> it doesn't crash at beginning of boot in logcat?19:25
T42<adampigg> what, qseecomd?19:27
T42<erfanoabdi> yeah any trustzone service19:27
T42<adampigg> first think it fails on is loading a firmware19:27
T42<adampigg> s/thing19:28
T42<erfanoabdi> ah so that should be fixed19:28
T42<erfanoabdi> idk how firmwares work in sailfish, but in ubports it just loads fine19:28
T42<adampigg> ill see if i can find the file its looking for19:29
T42<adampigg> hmmm, its not on my filesystem!19:31
T42<erfanoabdi> not in modem partition too?19:33
T42<erfanoabdi> what file is it looking for?19:34
abransonelros34: I think the new toolchain was supposed to have fixed that19:37
T42<adampigg> @erfanoabdi haventkn.mdb ... im told by the devs that its optional anyway19:47
T42<erfanoabdi> yeah it's optional19:48
T42<adampigg> so, thats likely not the issue?19:49
T42<elros34> abranson: with chacha I have: 'message authentication code19:49
T42<elros34> incorrect'. I fixed it by adding 'Ciphers' to sshd_config19:49
T42<adampigg> @erfanoabdi after the failure to load haventkn, there are no messages from qseecom until that ioctl19:51
T42<erfanoabdi> @erfanoabdi [QSEECOMAPI: Error::send command ioctl failed. …], problem is this, error 22 is invalid arg, it can't ioctl to kernel driver19:51
T42<erfanoabdi> not sure what's broken19:51
T42<adampigg> kernel prints this...19:52
T42<adampigg> [ 2234.586522] scm_call failed: func id 0x70000006, ret: -2, syscall returns: 0xffffffffffffffed, 0x0, 0x019:52
T42<adampigg> [ 2234.586558] QSEECOM: __qseecom_send_cmd: scm_call() failed with err: -22 (app_id = 196611)19:52
T42<adampigg> [ 2234.586574] QSEECOM: qseecom_ioctl: failed qseecom_send_cmd: -2219:52
T42<adampigg> @erfanoabdi does the code work on android 7 ?19:58
T42<erfanoabdi> biomertyd yeah, but hybris part should be ported to load non hidl hals19:58
T42<adampigg> thought that might be an issue ... i also have an android 7 port with a fp sensor19:59
T42<erfanoabdi> shouldn't be hard to backport..20:00
T42<adampigg> if i could get it tor work, i could look at writing a proxy between this and sailfish-fpd20:03
T42<erfanoabdi> TEE fingerprints are really problem for us, as u can see we have connected to FP hal and works fine, but gatekeeper and qseecom doesn't cooperate20:05
T42<erfanoabdi> some older android devices doesn't have TEE fp20:05
T42<adampigg> maye you can spot something in a full logcat?
T42<erfanoabdi> weird thing is on your older log it was working :)20:10
T42<adampigg> @erfanoabdi is the keystore service needed?20:10
T42<erfanoabdi> @adampigg [@erfanoabdi this is the logcate from the test …], here20:10
T42<erfanoabdi> @adampigg [@erfanoabdi is the keystore service needed?], nope20:10
T42<adampigg> im just looking all the things i have disabled20:11
T42<adampigg> @erfanoabdi fake_crypt, does that just start and exit?20:15
T42<erfanoabdi> @adampigg [@erfanoabdi fake_crypt, does that just start a …], yeah20:15
T42<erfanoabdi> need to run after keymaster and before fp hal20:15
T42<adampigg> hang on, keymaster isnt running20:21
T42<adampigg> @erfanoabdi something odd about keymaster
T42<erfanoabdi> @adampigg [@erfanoabdi something odd about keymaster http …], it's correct20:29
T42<adampigg> ok20:29

Generated by 2.17.1 by Marius Gedminas - find it at!