Saturday, 2022-11-19

xmnAnyone get stuck after updating to the latest Can't open any apps and the background are all gone. Can't even get to the terminal to clean up after with post_sfos_update06:04
T42<elros34> testing or devel sailfish image/repos?08:41
*** Ischwitch is now known as Ingvix09:01
T42<Ayman_rgab> @edp_17 samsung also ? (re @edp_17: I have the same issu...)16:27
T42<Ayman_rgab> hello17:51
T42<Ayman_rgab> on my device changing usb mode didn't work unless i restart usb-moded manually17:53
T42<Ayman_rgab> this log from usb-moded
malwhich device is that?18:25
T42<Ayman_rgab> mal:
T42<edp_17> Yep. :] (re @Ayman_rgab: @edp_17 samsung also...)18:44
T42<Ayman_rgab> is anyway to restart usb-moded when usb connected or disconnected 0_018:46
malprobably we should find a proper fix instead of such a hack19:04
T42<Ayman_rgab> any suggestion19:07
mal@Ayman_rgab could you try to check what kind of power supply type it reports in /sys/class/power_supply/*/type for all of the devices there19:08
T42<Ayman_rgab> okay19:08
T42<Ayman_rgab> Mains19:09
T42<Ayman_rgab> Battery19:09
T42<Ayman_rgab> OTG19:09
T42<Ayman_rgab> POWER_SHARING19:09
T42<Ayman_rgab> Unknown19:09
T42<Ayman_rgab> Unknown19:09
T42<Ayman_rgab> USB19:09
T42<Ayman_rgab> Wireless19:09
maldoes any of those change when you disconnect the cable19:10
T42<Ayman_rgab> no they are the same19:12
malwell probably the present value in the USB one changes19:19
T42<Ayman_rgab> root@SamsungA52017 ~# cat /sys/class/power_supply/*/online19:25
T42<Ayman_rgab> 019:25
T42<Ayman_rgab> 419:25
T42<Ayman_rgab> 019:25
T42<Ayman_rgab> 019:25
T42<Ayman_rgab> 119:25
T42<Ayman_rgab> 019:25
T42<Ayman_rgab> usb connected19:25
T42<Ayman_rgab> root@SamsungA52017 ~# cat /sys/class/power_supply/*/online19:25
T42<Ayman_rgab> 019:25
T42<Ayman_rgab> 119:25
T42<Ayman_rgab> 019:25
T42<Ayman_rgab> 019:25
T42<Ayman_rgab> 019:25
T42<Ayman_rgab> 019:25
T42<Ayman_rgab> usb disconnected19:25
malso it's the one that says USB as type if I read those correctly, also please don't paste that many lines here, use pastebin or similar service19:26
T42<Ayman_rgab> type USB changed to 0 from 1 when cable disconnected and type BATTERY changed to 1 from 4 (re @SailfishFreenodeIRCBridgeBot: <mal>so it's the one...)19:33
T42<Ayman_rgab> this's all types with full path
malwhat could be causing it to not start charging19:35
malcan you check for any usb_moded related messages in journal log after reboot and also show those when you restart usb_moded19:37
piggz_mal: spiiroin: another fun weeirdness in the PP port ... sometimes pressing the power button shuts the system down....19:38
piggz_[32731.785667] DSME[2167]: pwrkeymonitor: Timer triggered, initiating shutdown19:39
piggz_[32731.620753] sun4i-drm display-engine: [drm:drm_atomic_check_only] checking 00000000fd30c25219:39
piggz_[32731.887976] DSME[2167]: state: shutdown request received from dsme19:39
piggz_[32731.908613] D[32731.645005] sun4i-drm display-engine: [drm:drm_atomic_nonblocking_commit] committing 00000000fd30c252 nonblocking19:39
piggz_SME[2167]: state: Normal shutdown19:39
piggz_[32731.935577] DSME[2167]: state: new state: SHUTDOWN19:39
piggz_[32731.952958] DSME[2167]: state: Shutdown or reboot in 2 seconds19:39
piggz_wonder why dsme is doing that19:39
T42<Ayman_rgab> maybe because of this /sys/class/power_supply/s2mu004-charger/ it's of type Unknown and it's online value always 0 (re @SailfishFreenodeIRCBridgeBot: <mal>what could be c...)19:39
T42<Ayman_rgab> okay (re @SailfishFreenodeIRCBridgeBot: <mal>can you check f...)19:39
piggz_@Ayman_rgab you can specify the exact device to monitor in the usb_moded config, i did this recently19:40
T42<Ayman_rgab> how ? (re @SailfishFreenodeIRCBridgeBot: <piggz_>@Ayman_rgab ...)19:40
piggz_grep the logs for udevadmin commands which monitor power_supply devices19:41
malpiggz_: but that seems to detect usb just fine, if I understood correct, just won't start charging19:41
T42<Ayman_rgab> charging work when i connect usb cable19:42
malso what doesn't work?19:42
malI'm a bit confused how that behaves19:42
T42<Ayman_rgab> when i change usb mode i had to restart usb-moded to take effects19:43
piggz_i would add -d to usb-moded command line to see what its doing19:44
malwait so changing usb mode manually is not working but otherwise it works?19:44
T42<Ayman_rgab> i did that (re @SailfishFreenodeIRCBridgeBot: <piggz_>i would add ...)19:45
piggz_it might be my problem, beause charging is done from other events19:45
piggz_on the pinephonepro, plugging in usb would not trigger the dialog, but would detect charging, so, had to manually specify the udev device to monitor19:46
T42<Ayman_rgab> for example i change mode  from developer mode to mtp mode and disconnect the usb cable then connect it again it still on developer mode only if i restart usb-moded.service mtp works (re @SailfishFreenodeIRCBridgeBot: <mal>wait so changin...)19:48
T42<Ayman_rgab> exactly the same issue with my device (re @SailfishFreenodeIRCBridgeBot: <piggz_>on the pinep...)19:50
spiiroinpiggz_: mce & display on/off - in theory it can be mce, compositor, or both. if mce does it, then it is in mce via fbdev ioctl, or in mce-plugin-libhybris using android hal19:55
spiiroinit really should be only in compositor, but both doing fbdev ioctl:s does not hurt. if it goes to that hybris plugin... usually it should not19:56
piggz_spiiroin: its a native port, so not hybris19:56
piggz_ive been trying to collect logs for the kernel dev, however, last few times ive tried, dsme has decided to shut the phone down!19:56
spiiroinpiggz_: dsme & power key shutdown -> the idea is that if ui goes bonkers you can still do an orderly shutdown by pressing power key for five seconds19:57
piggz_spiiroin: this isnt 5 seconds though19:57
piggz_just a quick press19:57
piggz_maybe ill disable dsme until i get the other logs19:58
spiiroin1st thought: missing key up event? or dsme being seriously late with processing it?19:58
spiiroinpiggz_: you can also just move/rename the dsme power key plugin19:58
spiiroinhmm... "sun4i-*drm* display-engine" -> drm/kms? -> display power control should be in compositor20:00
* spiiroin is afk for a while, but... you might need to explicitly disable/hack mce side if it finds fbdev nodes20:01
piggz_spiiroin: i enabled drm logs to see if display is being commanded to power up...20:02
piggz_spiiroin: ill paste some logs showing mce getting power button then dsme shutting down...20:02
piggz_spiiroin:,143 those lines are mce and dsme20:03
T42<Ayman_rgab> mal: all related usb-moded messages after reboot, when changing mode and after restart usb-moded
T42<Ayman_rgab> udevadm monitor -p -s power_supply
piggz_is that while plug /unplug cable?20:22
piggz_doesnt look like it20:22
T42<Ayman_rgab> yes (re @SailfishFreenodeIRCBridgeBot: <piggz_>is that whil...)20:22
T42<Ayman_rgab> POWER_SUPPLY_STATUS=Charging and POWER_SUPPLY_STATUS=Discharging20:23
T42<adampigg> but, that only lists the battery device, no others .. no wonder usb-moded doesnt react20:24
T42<Ayman_rgab> what could cause this20:25
piggz_try without the -s power_supply and see what event happen when you plug in a cable20:26
T42<Ayman_rgab> okay20:30
T42<Ayman_rgab> udevadm monitor -p
piggz_interesting, there is typec add/remove events only20:36
piggz_best ask an expert now :)20:37
T42<Ayman_rgab> waiting for mal20:38
mal@Ayman_rgab I see something like "droid-hal-init: starting service 'usbd'.." try disabling that usbd if it's running20:53
malif android side usb process is running it can confuse usb when usb_moded is also controlling it20:54
spiiroinpiggz_: there is ~4.5 seconds from mce logging power key down and dsme deciding to shut down because it is still down... is there key up event? does mce/dsme see it? is it coming from a input device that could be grabbed by mce? which should no longer be necessary, but I can't recall if it is still default behavior21:09
malso evdev_trace -i and check also -t21:09
spiiroin(if kernel sends it, it could be/seem lost by a process X due to buffer overrun, input device grabbing, scheduling delay, whatnot)21:11
spiiroinbut testing will be simpler if you disable/mask whole dsme, or just the power key handler i.e. "mv /usr/lib/dsme/ /usr/lib/dsme/"21:12
spiirointhere have been some devices where weird things can happen to input devices around suspend, so kernel bugs are possible too21:14
T42<Ayman_rgab> mal usbd disabled
malany difference?21:20
malin how it works now21:21
T42<Ayman_rgab> no difference21:21
malhow did you test it21:22
T42<Ayman_rgab> . (re @Ayman_rgab: for example i change...)21:22
maltry some other mode than mtp just in case mtp has some issue21:24
mallike charging only mode or something21:24
T42<Ayman_rgab> same also every mode change requires restarting usb-moded21:27
malit's quite difficult to check those logs when I have no idea at what timestamp you did what21:28
T42<Ayman_rgab> and with always ask mode when plugging cable it just charging not show a menu of modes21:29
malshow log from such attempt21:30
T42<Ayman_rgab> setting mode to ask then detached usb and attach again didn't trigger usb moded21:37
malcan you check what usb-moded packages you have installed?21:41
malI think you should have usb-moded-defaults-android and usb-moded-developer-mode-android instead the non -android ones21:46
T42<Ayman_rgab> # USB mode controller21:51
T42<Ayman_rgab> Requires: usb-moded21:51
T42<Ayman_rgab> Requires: usb-moded-developer-mode-android21:51
T42<Ayman_rgab> Requires: usb-moded-defaults-android21:51
T42<Ayman_rgab> Requires: usb-moded-mtp-mode-android21:51
malyou have those?22:03
T42<Ayman_rgab> now yes22:10
T42<Ayman_rgab> but no difference22:11
malno difference in dmesg either?22:13
maljournal I mean22:13
T42<Ayman_rgab> yes same22:13
malthe udevadmin output you showed is odd since it only showed battery messages22:19
malyou mentioned you tried hardcoding path in usb moded config, have you done anything else there? and what was the change you did22:20
T42<Ayman_rgab> this one shows android_usb and typec also
malyou didn't answer if you had changed something in usb moded configs22:30
T42<Ayman_rgab> ah didn't see that, no i did not change anythings22:31
T42<Ayman_rgab> "you mentioned you tried hardcoding path in usb moded config" didn't help at all22:33
malwhich path did you use?22:34
T42<Ayman_rgab> this one /sys/class/power_supply/battery/22:35
T42<Ayman_rgab> is this ok22:36
T42<Ayman_rgab> rpm -qe -s usb-moded-defaults-android22:36
T42<Ayman_rgab> (contains no files)22:36
T42<Ayman_rgab> what all this
T42<Ayman_rgab> is it normal ?22:39
T42<Ayman_rgab> what if i run udevadmin while changing mode and restarting usb-moded22:41
T42<Ayman_rgab> (re @Ayman_rgab: what if i run udevad...)22:44
T42<edp_17> I've tried exactly that and it worked well before Then this update broke something, so I had to remove that hack. (re @Ayman_rgab:
T42<Ayman_rgab> which hack ? (re @edp_17: I've tried exactly t...)22:47
T42<edp_17> This hack:
T42<edp_17> Basically, with these two used rule I force-restarted usb-moded when I plugged the cable in.22:50
T42<elros34> mask usb-moded, restart device and then get logs from udevadm. I think usb-moded can mess with gadget so your kernel stops sending power_supply present/online events. IIRC there wer some options to handle it but it could be removed22:51
malone option is -a parameter for usb-moded22:52
malwhich is "keep gadget active on broken android kernels"22:52
T42<elros34> but is it even handled or it is just dummy option. I think long time ago when I have tried it didn't help but I could be wrong22:53
malseems that is not used anymore22:54
T42<Ayman_rgab> why you remove it (re @edp_17: Basically, with thes...)22:55
T42<edp_17> Because the usb menu popped up when the cable wasn't attached.22:56
T42<Ayman_rgab> when i set -a or -i it said that these are deprecated (re @SailfishFreenodeIRCBridgeBot: <mal>one option is -...)22:56
T42<Ayman_rgab> i'll try this (re @elros34: mask usb-moded, rest...)22:57
T42<Ayman_rgab> that's Mess (re @edp_17: Because the usb menu...)22:59
T42<Ayman_rgab> (re @elros34: mask usb-moded, rest...)23:07
T42<elros34> can't check it, this paste service blocked me now23:08
T42<edp_17> I've added -a to usb-moded.service (into usb-moded-args.conf) and applied the hack I described above. Are the "Fallback since cable detection might not be accurate." and "unexpected change" messages expected?
T42<elros34> but support for -a was removed 4 years ago23:10
T42<edp_17> mal has mentioned it a few minutes ago. ;)23:11
T42<Ayman_rgab> yes same here (re @edp_17: I've added -a to usb...)23:11
T42<Ayman_rgab> if you add -d you will see a message telling that -a are deprecated (re @edp_17: mal has mentioned it...)23:12
T42<edp_17> Ah, okay. Thanks. I've removed it as didn't help anyway.23:13
T42<elros34> just a hint, you should be adding -D for debugging purposes23:13
T42<edp_17> Thanks!23:14
T42<Ayman_rgab> (re @elros34: can't check it, this...)23:14
T42<Ayman_rgab> i have a wired behavior if usb-moded running when i remove usb cable battery icon take 5 seconds to changed but if usb-moded stop it changed immediately23:17
T42<edp_17> Interesting, when I plug the cable in the device plays a sound so it recognises that a cable is attached. But the usb  mode selection window isn't popping up.23:17
T42<Ayman_rgab> same sound, vibration and showing charging message (re @edp_17: Interesting, when I ...)23:18
T42<Ayman_rgab> * only if mode set to ask23:19
T42<edp_17> or it set to Charging only23:20
T42<Ayman_rgab> yeah23:21
T42<elros34> so you never get any udev event from power_suppply device other then battery when connecting/disconnecting cable?23:22
T42<Ayman_rgab> only battery23:22
T42<elros34> but parameters change if you do cat /sys/class/power_supply/<something different than battery>/* right? If yes then I think it would be the best to fix it in kernel and call power_supply_changed() method23:24
T42<edp_17> @elros34 : I have these. udevadm monitor -p -s power_supply:
T42<edp_17> grep . /sys/class/power_supply/*/uevent:
T42<elros34> looks sane, just no power_supply_cahnged in kernel23:26
T42<Ayman_rgab> yes it's changed (re @elros34: but parameters chang...)23:26
T42<elros34> I wonder if usb-moded how some poll method23:27
T42<edp_17> Yes. I've checked /sys/class/power_supply/max77843-charger/*: (re @elros34: but parameters chang...)23:28
T42<edp_17> Do you know what to fix in kernel?23:29
T42<elros34> I was messing with this but it was long time ago23:33
T42<Ayman_rgab> @elros34
T42<Ayman_rgab> or
T42<elros34> yeah that function must be called whenever usb connection change but not with otg but some different device23:34
T42<edp_17> Thanks, I'll try this.23:35
T42<elros34> but maybe there is easier way, because IIRC mainline drivers also do not use this23:37
T42<edp_17> How should I know what charger the device has? I only have drivers/battery in kernel source. Do not have battery_v2. And within I have these:
T42<Ayman_rgab> this one max77843-charger23:42
T42<Ayman_rgab> not v223:42
T42<elros34> I think I linked you obj2pro whhich should help you figure out which drivers your device use23:42
T42<edp_17> Oh, yes. I remember that.23:44
T42<elros34> this is the ugly patch I  have created once for usb issues: vbus_id_hw_scan changed on every connection so I call power_supply_changed(&axp_charger->usb); based on some params23:48

Generated by 2.17.1 by Marius Gedminas - find it at!