Thursday, 2018-12-20

*** ChanServ sets mode: +v T402:25
T4Prashantpp was added by: Prashantpp03:27
jusapiggz: what's with BT? :)08:00
T4<adampigg> Jusa, bt keyboard presses are doiubled08:05
T4<adampigg> Ive got fw lpading fixed, and can pair08:06
T4<adampigg> But struggling with pairing my headphones...but, keyboard first08:06
*** ChanServ sets mode: +v T408:46
T4<adampigg> Jusa, i guess no ideas? :(08:59
jusawhat happens when trying to pair, logs from bluetoothd, pulseaudio?09:00
jusafor keyboard no ideas (other than are there any logs :))09:01
T4<adampigg> I had a btmon log, but i posted it in #bluez!09:51
T4<adampigg> (Document)
T4<adampigg> Jusa, can you see that? It shows keyboard connecting, and pressing qwerty09:55
r0kk3rz@adampigg so it doubles each keypress?09:58
T4<adampigg> Yes09:58
r0kk3rzi'll have to take a btmon log of mine for you09:58
T4<adampigg> Which makes typing a pain :D09:58
r0kk3rzi'm wondering whether these messages should be there `a1 01 00 00 00 00 00 00 00 00`09:59
r0kk3rzbut i guess maybe thats keyrelease09:59
T4<adampigg> Trying evdev trace...10:00
T4<adampigg> I dont think evdev trace shows 2 presses10:01
T4<adampigg> Hang on10:01
T4<adampigg> (Document)
T4<adampigg> What u make of that?10:02
r0kk3rzyeah its definitely a kernel driver issue10:05
r0kk3rzi think the bt side is fine10:05
T4<adampigg> If i had my usb otg,, id try a real keyboard10:07
r0kk3rzthe issue seems to be whatever piece of code translates the bt HID protocol into keyboard signals10:09
T4<adampigg> What did evdev look like to you?10:09
T4<adampigg> Did you see doubles?10:10
r0kk3rzbut there isnt doubles in the btmon log10:10
r0kk3rzhmm maybe they arent doubles actually10:10
r0kk3rz0x010/KEY_Q - 1 = press, 0x010/KEY_Q - 0 = release?10:11
T4<adampigg> Yes, thats what i thought10:15
T4<adampigg> No evdev, so, wft!!!!10:16
r0kk3rzmust be higher up then10:33
T4<adampigg> That would be where...qt? Wayland?10:34
r0kk3rzlibinput im guessing10:34
r0kk3rzmaybe spiiroin knows10:39
spiiroinr0kk3rz: 1st though ... both scancode (or stte?) and key id are sent - could it be possible that both of those get mapped into separate key presses?10:48
spiiroinI do not recall ever seeing MSC_SCAN events, so no idea how those might get handled ...10:49
T4<adampigg> I cant seen any libinput packages ... There iz a trace tool.for that apparntly10:50
r0kk3rzyeah i'll grab some logs from my keyboard later to compare against10:51
T4<adampigg> the same later with a real kb10:52
T4<adampigg> Doing some reading, msc events should maybe only be  generated for unknown keys11:30
r0kk3rzwhat i was just reading says they're the raw scan codes11:32
r0kk3rzso idk, see what my keyboard does i guess11:32
T4<adampigg> This worked on tbj11:32
T4<adampigg> But on mido, i cant get it to connect!11:33
T4<adampigg> What kernel did tbj have?11:33
*** OhYash1 is now known as ohyash11:33
T4<Prashantpp> Hi.11:34
r0kk3rz@adampigg which kernel are you using?11:41
r0kk3rzan idea is to steal the hid driver from the sony kernel -
T4<adampigg> 3.14.7911:53
T4<adampigg> Good plan11:53
UmeaboyMerry Christmas to everyone.13:42
UmeaboyIf I take this guide made for the Xperia X model and adapt it to my XZ Premium, I need a hybris-sony-aosp-7.1.2_r36-20181220 branch on
UmeaboyWho can add it?13:45
r0kk3rzyou cant use aosp8?13:46
UmeaboyNope. Memory to low.13:47
Umeaboy8 GB.13:47
UmeaboyNeed 16 GB.13:47
UmeaboyYeah, the build fails due to lack of memory if I use AOSP 8.1..13:48
r0kk3rzgeez, thats a bit rough13:48
UmeaboyPlus there is no Oreo branch for my device.13:48
T4<adampigg> R0kk3rz, there is a subtle difference between the 2 versions of hidinput.c regarding creating the ev_msc event13:55
T4<adampigg> Def worth trying13:55
T4<adampigg> Actually, im looking at a newer kernel13:55
T4<adampigg> I might just comment out those events!13:55
T4<adampigg> Yours is the same13:55
malUmeaboy: what? I assume that XZ premium is codenamed maple which has android 8 branch in sony repos13:56
r0kk3rz@adampigg: yeah maybe grab from torvalds is a cleaner idea13:56
Umeaboymal: It is, but the problem is that refuses to initialize even with the --no-clone-bundle flag.14:03
UmeaboySo, unless you have an alternative location to initialize the repo from I can't use the 8.1.0 branch for mer-hybris.14:04
malwhich branch?14:05
*** ChanServ sets mode: +v T414:05
Umeaboyrepo init -u -b android-7.1.2_r36 --no-clone-bundle and many more.14:06
malUmeaboy: you did the usual repo init and then normal repo sync or some extra for repo sync?14:06
malUmeaboy: I still think it would much easier to use 8.1 since you would have the repo ready and things like bluetooth would be easier14:06
UmeaboyIt's the repo init that fails. I wanted to try to build Oreo for maple, but no luck. Initializing the Lineage cm-14.1 branch works fine though.14:07
malUmeaboy: does the hybris build fail with memory issue or the aosp build?14:07
UmeaboyHaven't tried the hybris build yet, but I'll do that later tonight.14:07
malI have aosp sources for 8.1 so I can build an image for any sony aosp 8.1 device14:08
UmeaboyIf I do the guide to initialize android-7.1.2_r36 and above it halts before it's finished.14:10
malI still think doing 7.1 is a bad idea14:10
malhow about you try the hybris build for 8.1 for your device and I'll build you the aosp images needed?14:11
UmeaboyI even used the --no-clone-bundle command and it seemed to work at the start, but it doesn't finish.14:11
UmeaboyI'll do that when I get home later tonight though.14:12
UmeaboyTake care.14:12
piggzr0kk3rz: that confirms it ... pluggin in my usb-otg hub + keyboard ..... each press is doubled!18:11
piggzwhich begs the question .... WTF??18:12
r0kk3rzpiggz: what did you use to trace input event for the log?18:14
r0kk3rzok, i get lots of msc_scan messages as well18:16
r0kk3rzthe output seems identical to yours18:16
r0kk3rzbut im pretty sure a usb keyboard would go through the same hid-input driver18:16
piggzr0kk3rz: so, i say again, wtf! :D18:38
piggz@eugenio_g7 can you try a usb keyboard on your tab?18:38
piggzhow do we have similar evdev traces, but I get double keypresses18:39
r0kk3rzgood question18:41
piggzr0kk3rz: afaict we dont use libinput18:41
piggzatleast i couldnt see any packages18:42
r0kk3rzi always though that was ike *the* wayland thing18:44
r0kk3rzbut perhaps not18:44
piggzr0kk3rz: i have a module loaded called ....hid.....ko18:45
piggzinfact lots of hid....ko modules18:55
piggzr0kk3rz: for various sensors18:56
piggzand a sensor hub18:56
piggzpotentially, a lot going through the hd system18:56
piggzr0kk3rz: not modules .... i renamed the module dir, rebooted, no modules loaded, and its the same19:09
piggzr0kk3rz: an idea ... i only traced one event device ... i wonder if i trace them all, i will get inputs from more than one....19:13
r0kk3rzworth a shot at least19:13
r0kk3rzi just did evdev_tract -t19:13
piggzi did the bt device specifically19:14
piggznope, not that19:15
piggz/dev/input/event7: 1545333332.830 - 0x04/EV_MSC - 0x004/MSC_SCAN - 45877219:15
piggz/dev/input/event7: 1545333332.830 - 0x01/EV_KEY - 0x010/KEY_Q - 119:15
piggz/dev/input/event7: 1545333332.830 - 0x00/EV_SYN - 0x000/SYN_REPORT - 019:16
piggz/dev/input/event7: 1545333332.933 - 0x04/EV_MSC - 0x004/MSC_SCAN - 45877219:16
piggz/dev/input/event7: 1545333332.933 - 0x01/EV_KEY - 0x010/KEY_Q - 019:16
piggz/dev/input/event7: 1545333332.933 - 0x00/EV_SYN - 0x000/SYN_REPORT - 019:16
piggzone keypress of Q19:16
r0kk3rzseems reasonable19:16
piggzbut still the same 2 presses in apps!19:17
r0kk3rzi think we need to know what the higher parts of the stack are19:26
piggzr0kk3rz: im thinking the lipstick options evdev* plugins ?19:27
piggzr0kk3rz: yup, ive fixed it19:27
piggzbut, i need to figure out how/why!19:28
piggzyay, ive got my BT keyboard setup again19:28
piggzr0kk3rz: so, i commentoued out the lipstick options a-la19:29
piggz#LIPSTICK_OPTIONS="-plugin evdev:/dev/input/event3 -plugin evdevkeyboard:keymap=/usr/share/qt5/keymaps/droid.qmap"19:29
piggzand it works....19:29
piggzstrangely, the touchscreen still works19:30
piggzi did try the evdevtouch plugin again, and that gave the same result with double-keys and bad TS19:30
piggzi guess I should try it with the evdev plugin, and without the evdevkeyboard plugin19:31
piggzand videversa19:31
r0kk3rzwell its a lead anyway19:32
r0kk3rzit must be listening on the event twice, which sounds weird19:33
r0kk3rzi wonder why its not doing that for me19:33
piggzr0kk3rz: what re your options?19:33
piggzr0kk3rz: ok, so it also works without the evdevkeyboard plugin loaded19:34
r0kk3rzhang on19:34
r0kk3rzmine says evdevtouch, yours says evdev19:34
piggzyes, that is becuase of my screen19:35
r0kk3rzoh ok19:35
piggzdo you have the keyboard one?19:35
piggzr0kk3rz: right, its the evdevkeyboard plugin19:37
piggzwith just that plugin loaded, it fails19:38
piggzso, i guess for the time being, i'll go with no options set as it all seems to work19:38
piggzwho would have a view on whats going on there?19:39
piggzjusa: BT audio problem
piggzthis shows my scanning, trying to pair with "Plattan ADV Wireless" ... the first 2 times fail, the third pairs, but then it fails to connect19:47
piggzr0kk3rz: and harbour-taskswitcher is working20:15
r0kk3rzthats good!20:18
*** ChanServ sets mode: +v T420:54
T4<adampigg> Jusa, mybe headphone connection problem is cos we dont have PA running yet20:56
r0kk3rzyeah you need pulse21:13

Generated by 2.17.1 by Marius Gedminas - find it at!