libera/#maemo-leste/ Wednesday, 2024-12-11

tmlindfreemangordon: abe_mcbsp1_dr.gpio_115 offset 0x100, omap4_pmx_core padconf offset 0x40, 0x100 - 0x40 = 0xc005:27
tmlindso this should work for gpio_115 padconf interrupt, untested: interrupts-extended = <&omap4_pmx_core 0xc0>;05:28
tmlindif that interrupt is common for modem uart and usb, then it could be set up as a generic wakeirq that ends up calling the runtime pm resume function05:29
tmlindif it's specific to the usb only, then it should have a separate threaded irq handler like you already have05:29
tmlindnot sure if the powerup sequence with the gpio pins needs to be redone, i think the modem can watchdog reboot and return online, not sure. so maybe just detecting a modem reboot can be used to rescan the usb bus for devices05:31
tmlindmaybe there's some way to do a modem reset or reboot over qmi?05:32
freemangordontmlind: so, I should not add another gpio, but just interrupt, right?07:07
freemangordonlets try it07:08
freemangordon253:          5   pinctrl 192 Edge      mdm6600-usb-wake07:43
freemangordonlemme disable PM runtime resume for the UART from the irq handler07:45
freemangordonotherwise seems to work07:46
freemangordonpm resuming uart seems not needed07:59
freemangordonYAY!08:03
dsc_freemangordon: https://github.com/maemo-leste/conversations/commit/8d5c9f65ef6283edcb5b09abfdde63222f6477a0.diff09:11
dsc_I think this HoldTimer was there so it doesnt trigger when scrolling09:12
dsc_else you cant really scroll without it triggering a popup09:12
dsc_iirc09:13
freemangordonnot sure I understand the question, but otherwise the patch seems reasonable09:14
freemangordonalso, I thunk tapandhold is always synthesized09:15
freemangordon*think09:15
dsc_you patched the holdTimer out, im just wondering if scrolling is a bit awkward now09:15
dsc_anyway09:16
freemangordonah09:16
freemangordonthis is path is mine?09:16
dsc_yeah :D09:16
freemangordonwell, it is good that I find my patches reasonable :D09:17
dsc_(y)09:17
freemangordonI don't see issues with scrolling09:17
dsc_alright09:17
freemangordonscrolls just fine here09:17
freemangordonat least on d409:17
dsc_i put presence and avatar into master09:19
dsc_an avatar is 8kb09:19
dsc_i dont suspect any drastic changes in memory09:19
dsc_expect*09:19
freemangordoncool09:20
freemangordoneven if you made it so images are loaded in background :)09:20
dsc_they are yes, sorry09:20
freemangordongreat09:21
dsc_wait, what do you mean?09:21
dsc_as in seperate thread?09:21
dsc_no09:21
freemangordonno09:22
freemangordonanyway, lets see how it behaves09:22
dsc_yes09:22
dsc_we can do on-demand if its neccesary09:22
freemangordonwill optimize further if needed09:22
freemangordonyeah, that's what I meant09:22
freemangordonsorry, still less coffee than needed09:23
freemangordonnot to say I am dealing with kernel code ATM and is a bit hard to switch ;)09:23
dsc_im going to try and get a virtual keyboard going with Qt apps09:39
freemangordonoh09:39
dsc_on a OS level09:39
freemangordonwhich keyboard?09:39
dsc_wizzup did some work, but it was for qt4 iirc09:40
freemangordonhim?09:40
dsc_will have to see what keyboard09:40
dsc_but just like the one we already have I suppose09:40
dsc_for gtk apps09:40
freemangordonperhaps see how it was done in fremantle qt409:40
Wizzuphi09:41
dsc_yeah09:41
dsc_im just wondering, why cant we use the gtk widget we already have?09:41
Wizzupyeah I spoke with dsc about this briefly09:41
freemangordonhttps://github.com/community-ssu/qt-x11-maemo09:41
dsc_we just need to spawn it when focus is on a QLineEdit09:41
Wizzupthe code is probably here https://github.com/maemo-leste/qt-platform-maemo/tree/him09:41
Wizzupbut it wasn't finished and a bit of a mess09:41
Wizzupbut I wanted to take more time to ensure that this was actually the latest code09:41
freemangordonWizzup: I have d4 working with upstream ofono over qmi :)09:42
freemangordonincluding notifications09:42
sicelothis is awesome news!!!09:42
freemangordonsicelo: yeah09:42
freemangordondoing some code cleanup09:42
Wizzupfreemangordon: wooo :)09:42
sicelowe really appreciate your work, and tmlind ;-)09:42
Wizzupwe can build experimental kernel and ofono then09:42
freemangordonwill push later on09:42
freemangordonyes09:42
dsc_i have no idea what this means but goodjob09:43
freemangordonyeah, tmlind's hint about padconf was the key09:43
* dsc_ gets coffee09:43
freemangordondsc_: this means we will have fully-functional modem on d409:43
dsc_nice! :)09:43
Wizzupwe've had a bunch of issues with notifications and some other stuff, so this is really great09:43
freemangordonand will remove (hopefully) the need for upsreaming serdev_gsm09:43
WizzupI wonder if that will also make things easier eventually over the spi mapphone devices09:44
freemangordonor whatever the name was09:44
freemangordonperhaps09:44
Wizzupmight make sense to check on the pm too, I suppose pm is ok?09:44
freemangordonif qmi_wwan works over spi09:44
freemangordonyes, it is09:44
Wizzup it does not, but the android kernel did stuff to make it work over spi iirc :D09:44
freemangordonwe also get AT interface working09:44
freemangordonthe only missing piece is voice contro;l09:45
freemangordonbut I'll look into it09:45
sicelonew ofono helps Librem5, but I believe should also help PP. AFAIK it's currently using some weird hacks?09:47
freemangordonyes, PP too09:48
freemangordonand basically all QMI09:49
freemangordonwhich means all devices after 2012, no?09:49
siceloalthough 2.13 is somewhat broken for PP, based on piggz report. Hopefully will be fixed soon.09:49
freemangordonI haven't had any issues with 2 quectel modems I have here and d409:50
Wizzupyes, pp is also qmi09:50
siceloyes it's not clear yet what broke specifically for pp.09:51
sicelohttps://lore.kernel.org/ofono/2292663.Vs0nDVz2aQ@adam-laptop-hp/T/#t for details09:54
freemangordonyeah, this also make it incompatible with older kernels (at least cannot compile on even 5.10)09:56
siceloheh, at least we're mainline09:57
freemangordonyeah09:57
freemangordonbut had to upgrade VM09:57
freemangordonand my PC09:57
dsc_so the keyboard we have right now (that works with GTK apps)10:01
dsc_its via him (hildon input method)10:01
dsc_correct?10:01
freemangordonyes10:03
Wizzupsorry, yes10:04
dsc_and it needs to be build into the maemo qt plugin10:05
Wizzupdsc_: basically what needs to happen in the qt side is to implement all the dbus calls, so even when sending a single key you need to send that over dbus and him will send the input10:05
dsc_right10:05
Wizzupand if you want to provide 'context' - eg. line editing that also has to go over dbus10:05
Wizzupit's not ideal really, but it's what it is currently10:06
Wizzupthere's also the at-spi2 stuff, but I don't know much about it10:06
Wizzupuvos has said at some point he thinks that might be a better way to go about this, but I don't know enough10:06
Wizzupthis is hooking into the accessibility frameworks10:07
freemangordonIIUC that's the proper way, however, uvos said there are some issues10:07
freemangordondon;t remember the details10:07
Wizzupyeah10:08
dsc_ok cool10:18
freemangordonhmm modem uart is waking-up device hundred times more than usb11:51
freemangordon254:         14   pinctrl 192 Edge      mdm6600-usb-wake11:51
freemangordon263:        203 4805b000.gpio  21 Edge      mdm6600-wake11:52
dsc_(hello:9849): sapwood-engine-WARNING **: 11:25:25.732: sapwood-theme: Failed to load pixmap file /usr/share/themes/alpha/images/TextInput.png: Failed to connect to sapwood server using `/var/tmp/sapwood-:0.0': Connection refused12:25
dsc_is there an env. var to supress sapwood-engine messages?12:26
freemangordondsc_: export DISPLAY=:012:38
freemangordontmlind: https://github.com/maemo-leste/droid4-linux/commit/c9acd3b9c606274f5cc421b199cff2f58eb60b1c12:41
dsc_ty :)12:41
freemangordonthis seems to work12:41
freemangordontmlind: regarding modem disconnects - never seen them here, no idea how to test12:42
freemangordonbut, I guess this can be handled in usb notifier12:42
freemangordonplease have a look at the commit, if you see no obvious issues I'll send for upstreaming12:44
freemangordonWizzup: could you make a new kernel release?12:44
dsc_https://akademy2008.kde.org/conference/slides/maemo_qt.pdf14:04
dsc_slide 2114:04
dsc_is also what I saw in the code yes14:05
dsc_but its funny to read a presentation from when I was still a teenager :P14:05
dsc_hmm14:06
dsc_these slides were handy *before* diving into the code :D14:06
dsc_would have been*14:06
freemangordonh,, aren;t we supposed to install event filter in polish()14:15
freemangordon?14:15
dsc_and yes, I can see how this is hacky re: at-spi214:15
freemangordonthe way we do for scrollbasr14:15
freemangordondsc_: which one is hacky? at-spi or him?14:16
freemangordonor qt14:16
freemangordoncan;t parse :)14:16
dsc_uhm I mean the way it communicates, qt + xcb14:16
dsc_it stops some events, based on atoms, etc.14:16
freemangordonwhy hacky?14:16
freemangordonbtw, qt5 should have standard way for input methods14:17
dsc_idk what polish() is14:17
dsc_basically what we should do is (and what Wizzup was working on)14:18
dsc_register a QPlatformInputContext plugin14:18
freemangordonmhm14:18
dsc_it hooks all widgets/inputs and installs a filter14:18
freemangordonright14:18
dsc_it then creates a xcb conenction to him14:18
freemangordonmakes sense14:18
freemangordonbtw, that might be useful:14:19
freemangordonhttps://github.com/maemo-leste/hildon-input-method-framework/tree/master/src14:19
dsc_yes I found this already14:20
freemangordonok14:20
dsc_i was using it to create a standalone program14:20
dsc_but it really needs to be a plugin I think14:20
freemangordoncool14:20
freemangordonyeah14:20
freemangordonbut at least gives you an idea what to do14:20
dsc_yeah14:20
freemangordonthere is also documentation14:20
freemangordonhttp://maemo.org/api_refs/5.0/5.0-final/hildon-input-method/14:21
freemangordonnot sure how useful it is though14:21
dsc_wizzup was quite far already I think14:22
freemangordoncool14:22
dsc_some technicalities with key combos to sort out14:22
dsc_modifiers14:22
dsc_im just trying to create a standalone application so its easier to test14:23
Wizzupfreemangordon: yes I can make a kernel release  today14:55
dsc_hmm14:57
dsc_might be more difficult, sending HILDON_IM_SHOW does not show the keyboard14:57
freemangordonit should, iirc14:57
Wizzupyes, there's bugs in the qt code for sure :)14:57
dsc_btw14:58
dsc_xcb_connection14:58
dsc_this is a x11 connection right14:58
dsc_does it use unix sockets for that?14:58
dsc_well doesnt matter14:59
Wizzupdepends on DISPLAY I think14:59
dsc_i have not come across any dbus stuff yet Wizzup14:59
Wizzupsorry, bbiab14:59
dsc_freemangordon: (hello:11016): hildon-1-WARNING **: 14:03:14.533: Tried to initialized Hildon more than once.15:03
dsc_can I ignore this?15:03
dsc_omg, I had "Enable Virtualkeyboard" disabled @ settings15:06
freemangordonyes15:06
freemangordon:)15:06
dsc_k it works now ;p15:06
tmlindfreemangordon: nice looks good to me, not sure if you need the wakeup-source though, maybe check with suspend/resume if it keeps some stuff enabled unnecessarily16:43
tmlindfreemangordon: are you seeing events now also for incoming sms on the usb qmi?16:44
freemangordonyes16:47
freemangordontmlind: without wakeup-source it does not work16:47
freemangordonwhat to check if properly suspended?16:48
tmlindi think the wakeup-source might have an effect of keeping things enabled, but maybe usb needs it16:50
freemangordonI did't see increase of power usage16:51
tmlindhmm maybe it's needed to keep the interrupt active though16:51
tmlindok16:52
freemangordonyeah, no interrupt without it16:52
tmlindok16:52
tmlindso incoming sms working?16:52
freemangordonyes16:52
freemangordonand call16:52
tmlindgreat!16:52
tmlindalso seeing it on the qmi and not just ttyUSB4?16:52
freemangordonit is using qmi only :)16:53
tmlindamazing16:53
freemangordonbut RING is seen on AT as well16:53
tmlindok16:53
tmlindso the audio mixer, any clues?16:53
freemangordondidn't have time16:54
freemangordonthis https://github.com/maemo-leste-upstream-forks/ofono/tree/upstream-wip16:54
tmlindok16:54
freemangordonif you want to test it16:54
freemangordonthis is latest ofono + d4 patches16:54
tmlindmaybe over the holidays i get a chance16:54
freemangordoncool16:54
freemangordonplease verify what you have concers for16:54
freemangordonsuspend/resume I mean16:55
freemangordonbecause I verified that usb device and phy are suspended16:55
freemangordonbut maybe more things have to be checked, dunno16:55
freemangordonI am trying to upstream d4 support in ofono though16:56
freemangordonwell d4 qmi support16:56
tmlindno concerns :)16:57
freemangordon"not sure if you need the wakeup-source though, maybe check with suspend/resume if it keeps some stuff enabled unnecessarily"16:57
freemangordonthats what I mean :)16:57
freemangordonor, give me some hints what to check16:57
tmlindright sounds like that's needed to keep the interrupt enabled though16:57
tmlindnothing to check imo, the usb suspend should clear the modem if it reboots..16:58
tmlindsorry i mean the usb notifier16:58
freemangordonyou mean udev16:58
freemangordonyeah16:58
freemangordonBTW, I think if modem falls off the bus, maybe it can be rebooted in the notifier16:59
tmlindyup16:59
tmlindyeah that would be cool16:59
freemangordonbut, I have never seen that happened here16:59
freemangordonso can;t test16:59
tmlindi need to reload the modem modules about every two weeks17:00
tmlindotherwise modem stops working17:00
freemangordonnever have any issues here, but I am not sure I keep my d4 without reboot for so long17:00
tmlindlet's hope that gets fixed too17:00
freemangordondo you use leste already?17:01
tmlindonly on a test device still, should update it now :)17:01
freemangordon:)17:01
freemangordonWizzup: is prting to current debian stable17:02
freemangordon*porting17:02
freemangordonshould be ready soon(tm)17:02
tmlindcool17:02
freemangordonok, lets see if I can re-send one more ofono patch17:02
freemangordonbbl17:02
tmlindlater17:03
freemangordonsicelo: https://git.kernel.org/pub/scm/network/ofono/ofono.git/19:41
freemangordon:)19:41
siceloyeah i'm subscribed to the ML ;-)19:42
freemangordonthose 4 patches are d4 related19:42
sicelothis is all that's needed for it? any still pending?19:42
freemangordonno I only have to send the big one19:43
freemangordonlemme push first19:44
WizzupI didn't get to the kernel yet, I hope to do it later tonight, but I need a break first19:44
freemangordonsure, no hurry19:45
siceloget a break, don't break :-)19:45
freemangordonsicelo: that one https://github.com/maemo-leste-upstream-forks/ofono/commit/b44ce9c07bfd7371b44e15849db7c2b268ac92b419:50
freemangordonther is at least one more I will have to make, location related19:50
freemangordonalso, back then I promised to implement better API for call forwarding19:50
freemangordonbut those are details19:51
sicelogreat, i understand19:52
freemangordonDec 11 21:03:38 localhost ofonod[2291]: drivers/qmimodem/voicecall.c:all_call_status_ind() Call 1 in state QMI_VOICE_CALL_STATE_SETUP(10)20:04
freemangordonDec 11 21:04:38 localhost ofonod[2291]: drivers/qmimodem/sms.c:event_notify() msg format 6 PDU length 3520:05
freemangordon:)20:05
freemangordontmlind: if anything, at least we can drop dtmf support from motmdm codec22:23
Wizzupdoes qmi support dtmf?22:34
freemangordonlooks like22:36
freemangordonby the looks of  eQMI_VOICE_START_CONT_DTMF and friends22:37
Wizzupok22:40
freemangordond4 modem also supports +VTS command22:42
freemangordonAT+VTS=?22:43
freemangordon+VTS: (0-9,A-D,*,#)22:43
freemangordonso that's not an issue22:43
Wizzupcool22:48
freemangordonvolume control is the issue :)22:49
arno11do you know guys why we can't stop (and restart) tone-generator ? it fails to find the correct pid23:20
arno11and pgrep returns no pid23:21
arno11i ask because i lost alerting tone again23:22
Wizzupis tone generator used for this?23:22
Wizzup(it might be, I don't know)23:22
arno11iirc yes23:23
arno11but that s a bit confusing23:23
arno11Wizzup: i got troubles with 2g calls as well at the same time (after dist-upgrade). Is new debian rules in libcmtspeechdata possibly the root cause ?23:33
arno11(with port to dh 10+)23:34
Wizzupcould be, but I don't think I built any of my changes for chimaera23:34
arno11ok23:35
arno11it s late, i ll try tomorrow a local cmt build i have, to see if it works better23:36
arno11indeed you didn t build it to chimaera so the problem is elsewhere23:41
arno11thx for details23:41
Wizzupnp23:44

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!