| T42 | <linusdan> @elros34 [@linusdan line 21-23, do you have device repo …], Yes, i have :( | 00:56 |
|---|---|---|
| T42 | <linusdan> fstab path is correct | 00:56 |
| T42 | <linusdan> I compiled the cm-11 with the same device tree | 00:57 |
| T42 | <elros34> but for some reason all of those "find" can not find correct fstab: https://github.com/mer-hybris/hybris-boot/blob/master/Android.mk#L47 | 01:02 |
| T42 | <elros34> If I get it right according to your log, you don't even have $ANDROID_ROOT/device/motorola directory | 01:05 |
| T42 | <kabouik> So we can get the cursor to render in LXC (and probably chroot too) when using `Xephyr` @elros34, but sadly the cursor coordinates are only updated when left- or right clicking. Confirmed visually and in `xev`. I think it's the same without `xephyr`, with just the evdevmouse option on SFOS side, but more obvious when the cursor is pai | 01:19 |
| T42 | nted by `xephyr`. | 01:19 |
| T42 | <kabouik> Also, with `xephyr` we lose touch support in Firefox with xinput2, and multitouch. :( | 01:19 |
| T42 | <elros34> so not very usable:) | 01:20 |
| T42 | <kabouik> So, some critical regressions, and yet some progress on other fronts. I didn't mention also that key changes made with `xmodmap` are no longer ignored when using `xephyr`, while they are with `xwayland` | 01:20 |
| T42 | <kabouik> Sadly no, progress and regressions at the same time. :( I hope there are other ways | 01:21 |
| T42 | <kabouik> remote-desktop/xrdp shows a cursor on the client, probably painted too, but it's moving in real time | 01:22 |
| T42 | <kabouik> But I'm afraid it's just local on the client | 01:22 |
| T42 | <kabouik> Perhaps it's possible also to make a permanent event with the mouse so coordinates are updated all the time. That's what it looks like when holding right click: https://youtu.be/6mzt5FMFcUw | 01:26 |
| T42 | <kabouik> It's choppy because of OBS, but was relatively smooth otherwise. | 01:28 |
| T42 | <elros34> I still think that some mouse tracking item in qml would the best. It could be even added as patch to lipstick. | 01:29 |
| T42 | <kabouik> But that would make qt-5.12 a requirement, this was all in the container so no change on SFOS | 01:30 |
| T42 | <kabouik> But it doesn't work as expected anyway so qt surely is worth a try | 01:30 |
| T42 | <elros34> for sure it can be done on qt5.6 also: event filter to track mouse position. When I will sort out some issue in my rootless compositor I will try if I will be able to do this because qtwayland mouse tracker doesn't works correctly for me. | 01:33 |
| T42 | <kabouik> What are you trying to do with your rootless compositor? | 01:34 |
| T42 | <kabouik> Please do ping me if you find something interesting on mouse tracker | 01:35 |
| T42 | <elros34> I want to start apps in chroot in separate windows with rotation support. So far it works it works partially | 01:36 |
| T42 | <kabouik> Oh nice, using Xephyr/xnest for that too? | 01:37 |
| T42 | <elros34> no, qtwayland + xwayland compositor | 01:37 |
| T42 | <kabouik> I remember running Tilix in its own window with chroot when it was still sailfish_ubu_chroot after you spoonfed me to guide me with the corresponding script | 01:38 |
| T42 | <kabouik> That was cool. Not rotating of course | 01:39 |
| T42 | hassanh92 was added by: hassanh92 | 02:32 |
| T42 | <hassanh92> 0248 | 02:32 |
| Mister_Magister | lbt_: asking for that patternization of testing:asus:5z | 10:13 |
| Mister_Magister | mal: looks like people can access wide angle (third) camera on lineage. Would it be possible to access it on sfos (customizing piggz'o'camera maybe? winkwink) | 11:08 |
| vknecht | there's this MAX_CAMERAS = 2 in gstdroidcamsrc... https://github.com/sailfishos/gst-droid/blob/c21a53eb0a58b359b8a8d8dd2f77f27ad135c05a/gst/droidcamsrc/gstdroidcamsrc.h#L51 ... had to change the check against it to get cam app to even work when there's more than two... | 11:28 |
| vknecht | don't know if POV depends on it | 11:28 |
| Mister_Magister | vknecht: what will happen if i do this https://github.com/Michal-Szczepaniak/gst-droid/commit/e0eb1cf164ae1828c646fa0b65af8eeee3b3d855 | 11:29 |
| vknecht | dunno, if you're lucky it will use 3rd cam as if it's second one ? | 11:31 |
| Mister_Magister | yeah XD | 11:31 |
| Mister_Magister | entire sfos camera thing is based around primary and secondary, changing that won't be easy but first check if its even working | 11:33 |
| Mister_Magister | vknecht: sadly it just gives standard camera :/ | 11:41 |
| mal | Mister_Magister: vknecht: it depends on how the camera is exposed in android | 12:17 |
| mal | is it like other cameras or some special way | 12:17 |
| Mister_Magister | mal: it sounds like it's exposed like other 2 cameras are | 12:17 |
| vknecht | I guess for POV it's qtmultimedia support which matters ? and obviously add support for a "ternary" (or tertiary ?) cam in qml/pages/CameraUI.qml ? | 12:20 |
| mal | first someone should test if and how camera is enumerated in sfos if any hardcoded 2 camera limitation from middleware is removed | 12:23 |
| Mister_Magister | mal: any idea how i can check this? https://github.com/sailfishos/droidmedia/blob/master/droidmediacamera.cpp#L271 | 12:32 |
| Mister_Magister | mal: sfos has hardcoded id 'primary' and 'secondary' and it's based around them so adding more will be pita | 12:33 |
| Mister_Magister | noone in 2013 thought about more than 2 cameras | 12:34 |
| mal | Mister_Magister: what does dr8id-camres show, afaik it asks for number of cameras, unless gst-droid hardcodes it | 12:36 |
| piggz | mal: spiiroin: is it possible to configre mce so that it doesnt use a notification LED? this device doesnt have one. I probably dont need to do anything, just wondered if there was a way to tell the system no led exists | 12:37 |
| mal | piggz: have you checked what led configs mce has | 12:37 |
| Mister_Magister | mal: umm i don't think so | 12:37 |
| Mister_Magister | or… it does | 12:38 |
| Mister_Magister | https://paste.opensuse.org/c3c8c090 mal do you notice "resolution for :" | 12:38 |
| Mister_Magister | missing name | 12:38 |
| mal | if it doesn't print it then add some debug printing | 12:38 |
| Mister_Magister | but it seems to show resolutions for third camera | 12:39 |
| mal | so it sees it | 12:39 |
| piggz | mal: its got to configs as part of mce package, 20hybris-led and 20debug-led | 12:39 |
| Mister_Magister | yeah you can see resolution for primary resolution for secondary and resolutoin for : missing label means third | 12:40 |
| piggz | mal: i found how to override it | 12:40 |
| piggz | [featurehybris] notificationled -= false | 12:41 |
| mal | piggz: is there some null backend for led or something, I remember there was some other device without led, check channel logs | 12:41 |
| mal | ah | 12:41 |
| piggz | mal: i think that + hw-settings.ini should be enough | 12:44 |
| Mister_Magister | mal: if it sees third camera why setting id to 2 didn't work | 12:45 |
| mal | not sure | 12:46 |
| Mister_Magister | https://github.com/Michal-Szczepaniak/gst-droid/blob/testing/gst/droidcamsrc/gstdroidcamsrcenums.c#L32 mhm mhm no name i bet if i added name here, droid-camres would display it | 12:50 |
| piggz | what are these 3rd cameras really for? on the volla, the rear cameras are 16mpx + 2mpx ... why would you take a pic with the 2mpx version? is it not for augmenting quality info for software processing on the main camera? | 12:50 |
| Mister_Magister | piggz: here it's wide angle camera, pretty useful | 12:50 |
| Mister_Magister | mal: it's even better, 0:00:03.099102684 11838 0xb30c1000 INFO droidcamsrc gstdroidcamsrc.c:558:gst_droidcamsrc_get_hw:<droidcamsrc2> Found 4 cameras | 12:57 |
| mal | hmm | 12:58 |
| Mister_Magister | i'm checking id 3 | 12:59 |
| Mister_Magister | mal: behaviour is weird, first it shows normal back camera then i switched to front camera and when i switch back it shows front camera lmao maybe 3 is wrong but 2 is'nt right either | 13:06 |
| Tofe | If it's of any interest, I also get 4 cameras on tissot (xiaomi A1) | 13:06 |
| Mister_Magister | its just plain weird it won't go into the third camera | 13:06 |
| Mister_Magister | https://paste.opensuse.org/0611270a | 13:10 |
| mal | maybe some limitation in some middleware | 13:13 |
| Mister_Magister | mal: maybe i could try to add >2 camera support in gst-droid? | 13:20 |
| Mister_Magister | like for example here https://github.com/Michal-Szczepaniak/gst-droid/blob/testing/gst/droidcamsrc/gstdroidcamsrc.c#L571 | 13:20 |
| Mister_Magister | because it looks like you won't get any info for camera id 2 cause it's all only 0 and 1 | 13:21 |
| Mister_Magister | but idk if it makes sense, what do you think mal | 13:22 |
| Mister_Magister | i am dumb after all so it might be dumb idea | 13:22 |
| piggz | on a side note, i dislike when line lengths are artificially limited like that, even if those 3 lines were on 1 it would barely be over 80, and its not like we use 80column displays anymore :D | 13:25 |
| Mister_Magister | piggz: it looks prettier but ye it would fit | 13:27 |
| abranson | I don't think it would be too bad to extend gst-droid to use all the cameras. it already fetches some metadata for them so it knows which ones are back and front mounted. but I think the tricky bits would be qtmultimedia and above. everything's geared up to one back, one front and a simple switch between them | 13:48 |
| Mister_Magister | abranson: i know, jolla camera or piggzs camera just use primary and secondary but i was thinking of adding support to 3 cameras in gst-droid and just setting primary camera as third camera just for now to check if its working. do you think it makes sense? | 13:49 |
| abranson | maybe that would be quite simple if you just override the camera setting to droid | 13:51 |
| Mister_Magister | abranson: wdym? | 13:52 |
| abranson | well, right now it uses 0 and 1, so you just change that to use 0 and 2 or whatever. nothing above needs to know. | 13:52 |
| Mister_Magister | no that doesn't work | 13:52 |
| Mister_Magister | i already tested it | 13:52 |
| abranson | oh really? | 13:53 |
| Mister_Magister | yeah | 13:53 |
| Mister_Magister | it was my firrst dumb idea | 13:53 |
| abranson | :D | 13:53 |
| Mister_Magister | [13:29] <Mister_Magister> vknecht: what will happen if i do this https://github.com/Michal-Szczepaniak/gst-droid/commit/e0eb1cf164ae1828c646fa0b65af8eeee3b3d855 | 13:53 |
| Mister_Magister | but id 2 is problematic cause gst-droid doesn't get any info for id 2 | 13:53 |
| Mister_Magister | [15:20] <Mister_Magister> like for example here https://github.com/Michal-Szczepaniak/gst-droid/blob/testing/gst/droidcamsrc/gstdroidcamsrc.c#L571 | 13:54 |
| abranson | i don't know if that's enough. there are some hardcoded 1s in there | 13:55 |
| Mister_Magister | abranson: dont you think we should add support for id 2 first like above? | 13:55 |
| abranson | see gst_droidcamsrc_get_hw | 13:55 |
| Mister_Magister | abranson: that's not the only place ofc | 13:55 |
| Mister_Magister | it was just example | 13:55 |
| abranson | ok | 13:56 |
| Mister_Magister | but if we added supoprt for id 2 and then changed 0 to 2 like the original dumb idea + some debugging it could just work? | 13:56 |
| Mister_Magister | am i saying something sane? | 13:56 |
| abranson | i think it should work as far as gst-droid is concerned yet | 13:57 |
| abranson | the higher ups might not like it. it's very hardcoded | 13:57 |
| abranson | but I think it needs to be done at some point. it's not the first time it's been a problem. some devices only have one | 13:58 |
| abranson | and don't always count from 0 either :/ | 13:58 |
| Mister_Magister | abranson: but we would jsut fool higher ups into thinking its main camera | 13:58 |
| Mister_Magister | i jsut want to test if camera 2 works at all | 13:59 |
| abranson | maybe do it lower down than that. e.g. switch the number in droid_media_camera_connect? | 14:00 |
| abranson | or gst_droidcamsrc_dev_open to avoid rebuilding droidmedia | 14:00 |
| abranson | so even the rest of gst-droid thinks it's using camera 1 | 14:01 |
| Mister_Magister | abranson: but info is gotten from gst_droidcamsrc_get_hw so i need to add 2 anyway | 14:02 |
| abranson | you could fake it in gst_droidcamsrc_get_hw too | 14:04 |
| abranson | i.e. in gst_droidcamsrc_fill_info | 14:04 |
| abranson | see what I mean? do it much lower down | 14:04 |
| Mister_Magister | abranson: i could just go through entire file and add support for id 2 as in load info and increase max camera count and stuff | 14:04 |
| Mister_Magister | it would be same work | 14:04 |
| abranson | yeah. i think i tried that once, but didn't get very far. might be a branch somewhere | 14:05 |
| Mister_Magister | with some GST debugs i could get somewhere | 14:05 |
| Mister_Magister | looks simple enough | 14:05 |
| Mister_Magister | abranson: what's worse is that when there are 3 cameras it reports 4 cameras | 14:05 |
| abranson | yeah that's not very nice :) | 14:06 |
| Mister_Magister | rild is more rude tho | 14:07 |
| Mister_Magister | abranson: so what i'm saying makes sense? | 14:09 |
| Mister_Magister | should i get to coding? | 14:10 |
| abranson | yeah, but I don't think you'll really know until it works | 14:10 |
| Mister_Magister | what do you mean | 14:10 |
| Mister_Magister | higher up just use 'primary' for id | 14:10 |
| Mister_Magister | and i can replace id of primary to 2 and just add support for 2 | 14:11 |
| Mister_Magister | so should work for higher levels | 14:11 |
| abranson | yeah, makes sense. but I couldn't say if there's some wall you might hit. you'll never know until you try. | 14:11 |
| Mister_Magister | sure, if gst-droid won't be enough i'll get through qtmultimedia and any wall i need | 14:11 |
| Mister_Magister | i want my wide angle if its possible | 14:12 |
| Mister_Magister | it gonna be revolutionary i'm telling ya | 14:12 |
| Mister_Magister | i'm balls deep in it already anyway | 14:12 |
| Mister_Magister | and looks liek interesting challenge | 14:12 |
| abranson | it needs doing tbh | 14:14 |
| Mister_Magister | i'll try to do it. technically i should do it not with aliases as primary and secondary but we should assume unlimited amount of cameras | 14:18 |
| Mister_Magister | because we will be updating into ethernity | 14:18 |
| eduardas | hello, I would like to ask an oFono specific question. One particular line in the docs is slightly unclear. | 14:20 |
| eduardas | https://git.kernel.org/pub/scm/network/ofono/ofono.git/tree/doc/features.txt#n465 | 14:20 |
| eduardas | 'silent' modem reset is mentioned | 14:20 |
| eduardas | what does this really mean from a technical standpoint. Is this related to power cycling the modem via an SoC GPIO pin? | 14:21 |
| eduardas | or is this something else | 14:21 |
| eduardas | any connman or ofono contributors here? | 14:22 |
| monich | eduardas: that doesn't ring a bell, really | 14:25 |
| monich | besides, the ril driver used by sfos doesn't talk directly to the modem, god only knows what happens on the android side of the ril interface | 14:25 |
| eduardas | monich: AFAIK Sailfish uses ofono + connman as the userspace stack to get cellular internet connectivity | 14:26 |
| monich | yep, that's right | 14:26 |
| monich | but we are insulated from gpio-level stuff by several layers of abstractions | 14:27 |
| eduardas | monich: I am currently working with connman + ofono in a Yocto/OE context with a simcom modem | 14:27 |
| monich | with atmodem driver? | 14:27 |
| eduardas | monich: the guy who does the modem integration pushes for a handcrafter C binary to do power cycling of the modem via GPIO when it supposedly hangs | 14:28 |
| Mister_Magister | monich: stupid rild even on 1.2 doesn't return registration state :< | 14:28 |
| eduardas | monich: QMI | 14:28 |
| eduardas | monich: I am totally new to this, first day I am seriously reading up on ofono and this cellular stuff | 14:29 |
| eduardas | though I had experience with connman for regular Wi-fi before | 14:30 |
| monich | eduardas: afaict ofono itself won't be much of a help here - whatever hw-specific recovery you need to do, most likely needs to be done in your hw specific code | 14:30 |
| monich | Mister_Magister: are you working with mal's code or did you write your own? | 14:31 |
| eduardas | monich: can the kernel networking subsystem be made aware of the gpio pin that enables power to the modem? | 14:33 |
| eduardas | monich: controlling the power pin from userspace via GPIO sysfs sounds kind of lame to me | 14:33 |
| eduardas | monich: but that is how the guy working on it implemented it for the project I'm working on | 14:33 |
| eduardas | monich: so I am looking for a better solution | 14:34 |
| eduardas | so that if the modem does need a hard reset for some reason, it can be done by whatever the standard cellular stack for Linux is considered to be | 14:35 |
| eduardas | i.e. in a non-custom fashion | 14:35 |
| eduardas | using FOSS userspace or kernel code | 14:35 |
| monich | eduardas: sounds too hw specific to be done in a generic way, honestly | 14:36 |
| eduardas | since modem is connected via USB, there aren't any devicetree entries for it | 14:36 |
| eduardas | so I can not pass the info about the power on pin to the kernel that way | 14:37 |
| eduardas | monich: do you know who is considered an expert in the community on connman + ofono and modem integration under embedded Linux? | 14:38 |
| eduardas | monich: where should I go if I need an expert opinion on this? | 14:38 |
| monich | eduardas: I maintain sfos fork of ofono and have written most of the ril intergration code | 14:39 |
| monich | but there's also upstream's chat | 14:40 |
| monich | or at least there was | 14:40 |
| eduardas | monich: have not found an ofono channel on freenode or elsewhere | 14:40 |
| eduardas | monich: there is only connman and ell | 14:40 |
| monich | hmm | 14:40 |
| Mister_Magister | monich: mal's | 14:41 |
| eduardas | monich: oh, there is one... just not listed in Konversation | 14:41 |
| eduardas | for some reason | 14:41 |
| eduardas | when I tell it to list channels on server | 14:42 |
| monich | eduardas: try #ofono | 14:42 |
| monich | but I haven't been there for quite some time, dunno who's there | 14:44 |
| eduardas | monich: just curious.. what is this "RIL integration code" you speak of | 14:45 |
| eduardas | is this some kind of middleware on top of ofono? | 14:45 |
| eduardas | or the other way around? | 14:45 |
| monich | https://git.sailfishos.org/mer-core/ofono/tree/master/ofono/drivers/ril | 14:45 |
| monich | not to me confused with upstream's rilmodem :D | 14:46 |
| monich | historically we've had our own even before upstream had one | 14:46 |
| eduardas | monich: so this is an ofono backend? why is this one in particular needed? | 14:47 |
| eduardas | monich: why did you need a new backend? | 14:47 |
| monich | we talk to android's ril service rather than directly to the modem | 14:49 |
| monich | that makes it easier to write an adaptation for a new platform | 14:49 |
| eduardas | monich: isn't SFOS built more like a regular Linux distro, systemd and all, how do you still have an Android service running? | 14:50 |
| monich | that would require a new adaptation, currently we are not 100% android-free | 14:51 |
| monich | the first official android-free port of sfos is still to be made | 14:52 |
| eduardas | monich: this is Android code running as a systemd service or something else entirely? | 14:52 |
| monich | it's typically android init and then systemd and the rest of the system on top of it | 14:52 |
| eduardas | monich: oh boy, that sounds nasty | 14:53 |
| monich | as I said that's because it's easier | 14:53 |
| monich | not because we like it that way | 14:53 |
| Mister_Magister | monich: u gotta hit that stupid rild :< | 14:53 |
| eduardas | monich: is there some all-encompassing SFOS architecture doc I could take a look at that shows how this works? with schematics or something | 14:54 |
| T42 | <adampigg> <cough> pinephone </cough> | 14:54 |
| monich | yeah, <cough> pinephone </cough> | 14:55 |
| monich | we've heard of it :D | 14:55 |
| T42 | <adampigg> also, latte tablet is 99.9999999& android free | 14:56 |
| T42 | <eugenio_g7> Working towards 100%! | 14:56 |
| monich | that's nice | 14:56 |
| eduardas | kinda still waiting for my Librem 5, but I have asked for a later batch since I know how electronics get made in real life ;) | 14:57 |
| eduardas | would not really risk it with early revisions | 14:58 |
| eduardas | that is also supposed to be Android-free | 14:58 |
| eduardas | monich: who is the current ofono maintainer and are they responsive to technical questions? | 15:00 |
| monich | Mister_Magister: I would wait for mal to finish and merge his ril 1.2 stuff and then see what's still broken | 15:00 |
| Mister_Magister | monich: but its faily working | 15:01 |
| monich | eduardas: that would be me, I guess | 15:01 |
| Mister_Magister | monich: the getting request is handled so i don't know what you are talking about | 15:01 |
| eduardas | monich: not just the SFOS ofono fork maintainer, but the ofono maintainer in general? | 15:03 |
| monich | Mister_Magister: I haven't seen any patches from mal yet, and I don't really know what he's done so far | 15:03 |
| eduardas | monich: the one hosted by kernel.org I mean | 15:03 |
| Mister_Magister | monich: i've seen that _1_2 registration status is handled so… | 15:03 |
| Mister_Magister | where is the code i'll show you | 15:03 |
| monich | eduardas: https://lists.ofono.org/hyperkitty/list/ofono@ofono.org/ | 15:04 |
| monich | eduardas: that should give you an idea ^ | 15:04 |
| Mister_Magister | monich: https://github.com/mer-hybris/ofono-ril-binder-plugin/commit/f9383630a6643740356c18485b10d6e99d1b39d8#diff-2b7ad7bbdb95daee77049466092ecf5fR3413 | 15:05 |
| monich | on what's happening with upstream | 15:05 |
| eduardas | monich: if so I feel you could really use a MAINTAINERS file on kernel.org or t least a note in the source README about who the active maintainers are | 15:06 |
| eduardas | monich: not everyone would have the idea to get on sailfish-posrters on freenode to ask about ofono like me | 15:07 |
| monich | eduardas: with open source you quite often have to do some research just to figure out who's doing what and what's going on :) | 15:07 |
| monich | although in ofono's case it could've been done better, there's intel behind it | 15:08 |
| eduardas | monich: I know, but just because it has become the norm, I do not feel that is very good... there are people who have been doing modem integration for years and are not aware of ofono or modemmanager | 15:09 |
| monich | and they still do it like it's something they're doing in their spare time | 15:09 |
| eduardas | still doing handcrafted C and such | 15:09 |
| eduardas | monich: I know corporations don't really want to fund this | 15:10 |
| eduardas | monich: which is a huge shame | 15:10 |
| eduardas | monich: since lots of people do in fact benefit from this work | 15:10 |
| eduardas | monich: I am still thankful for the fact that there are people like you doing this work | 15:11 |
| monich | intel does, most of bluez, connman and ofono code are paid for by intel | 15:11 |
| eduardas | monich: and willing to give me some kind of support | 15:11 |
| eduardas | monich: perhaps it is just me being spoiled, but there is no real reference distro + hardware platform from Intel or someone else where all of this stuff comes together | 15:13 |
| eduardas | monich: and then OEMs could use this as a starting point and example | 15:14 |
| eduardas | monich: for various 4G related features, correct power management, etc. | 15:14 |
| monich | eduardas: which reminds me that we still don't support volte :| | 15:15 |
| eduardas | monich: when I needed to work on a Linux handheld device in the past and use connman, SailfishOS and Mer were the closest examples on how to use this component in real life | 15:16 |
| eduardas | monich: what is that? | 15:16 |
| monich | voice over lte | 15:16 |
| eduardas | monich: well, I am merely concerned about M2M and am thankful for it | 15:17 |
| T42 | <ippokratis> @elros34 [I want to start apps in chroot in separate win …], that would be great. | 15:19 |
| T42 | <ippokratis> but for you f(x)tec pro1 guys keyboard is no problem . | 15:19 |
| T42 | <ippokratis> we in ports need external keybord if chromium is not open from inside xfce4 | 15:19 |
| eduardas | monich: so again to my initial problem, ofono has nothing to do with modem power management? | 15:19 |
| monich | eduardas: not much, other that providing more or less standard interfaces to power it on and off | 15:21 |
| eduardas | monich: oh, but that IS important... perhaps I can bind GPIO to that? | 15:21 |
| eduardas | monich: can one write a plugin/backend for that? | 15:22 |
| eduardas | monich: if ofono has a d-bus call to power on the modem, it makes sense to bind the GPIO control that actually DOES power it on, doesn't it? | 15:24 |
| eduardas | monich: or am I just being naive? | 15:24 |
| monich | in our ofono ril driver we have code which is tracking display state, charging etc | 15:24 |
| monich | https://git.sailfishos.org/mer-core/ofono/blob/master/ofono/drivers/ril/ril_devmon_ds.c | 15:24 |
| monich | like this one ^ | 15:24 |
| monich | but that gets translated into appropriate ril calls | 15:25 |
| monich | and whatever happens there, is up to the ril service | 15:25 |
| monich | that's speaking of power management | 15:26 |
| eduardas | monich: the problem in my case that qmimodem is the one that should actually get used | 15:26 |
| eduardas | monich: so if a USB modem disappers from the USB bus and then re-appears again because I toggle its power outside the connman + ofono framework, that is a-ok as far as the connman/ofono stack is concerned? | 15:28 |
| eduardas | monich: or is it bad practice? | 15:28 |
| eduardas | monich: I am kinda skeptical whether modems really hang that often in practice to require a hard reset | 15:29 |
| Mister_Magister | abranson: gst_droidcamsrc_fill_info is so retarded | 15:29 |
| eduardas | monich: bus some people are telling me that and I do not know how to respond | 15:29 |
| eduardas | monich: do you ever see modems hang in a way that your stack can not really do anything about it without cycling power? | 15:30 |
| monich | eduardas: in ril setup it's often hard to say what exactly hangs - the modem itself or something in between us and the modem | 15:38 |
| monich | eduardas: denkenz is the upstream guy in the ofono channel | 15:39 |
| eduardas | monich: what can userspace do when the stack hangs? | 15:40 |
| eduardas | monich: what do you think is the reasonable response from the business logicc point of view? | 15:40 |
| eduardas | monich: what do you do in your case? reboot? print error message in UI? | 15:41 |
| monich | eduardas: we actually do have a sfos specific dbus interface for reporting unrecoverable errors and then the ui can show a warning like "some kind of shit happened, you better restart your device if you want cellular connectivity to work reliably" | 15:43 |
| eduardas | monich: any idea how often that happens in real life? though that might be a loaded question... :D | 15:44 |
| monich | it depends on the phone of course | 15:45 |
| eduardas | I do not own a Sailfish device, sadly. | 15:45 |
| monich | with l500d/jollac that was happening often enough for us to implement that ui warning mechanism | 15:45 |
| monich | maybe one person out of 100 was getting that once a month | 15:46 |
| monich | roughly | 15:46 |
| eduardas | monich: do you attempt to do some kind of restart for connman or ofono or rild in userspace before you print that message or just give up right away? | 15:47 |
| eduardas | monich: i.e. is there some kind of retry sequence? | 15:48 |
| monich | rild crash was the trigger for it | 15:48 |
| monich | we reconnect to it but... | 15:48 |
| monich | the question is whether it always comes back up successfully | 15:49 |
| monich | basically, at startup it's a cold boot essentially | 15:49 |
| monich | and after crash rild comes up but the modem is in an unpredictable state | 15:50 |
| monich | we also have an option to do a power-cycle sequence after connecting to rild | 15:50 |
| monich | sometimes it's necessary, sometimes rild doesn't like it | 15:51 |
| monich | on some multi-slot devices power state of both modems are tied to each other - either both are on or both off | 15:52 |
| monich | it's really case-by-case | 15:52 |
| monich | impossible to do it in a way that would work everywhere | 15:53 |
| eduardas | monich: but you did mention that ofono does in fact have a d-bus API call to cycle power? | 15:54 |
| eduardas | monich: or did I misunderstand? | 15:54 |
| eduardas | monich: what about my idea of adapting this to specific hardware? | 15:54 |
| eduardas | monich: i.e. patching ofono | 15:55 |
| monich | https://git.kernel.org/pub/scm/network/ofono/ofono.git/tree/doc/modem-api.txt | 15:55 |
| monich | Powered and Online properties ^ | 15:55 |
| monich | eduardas: for a specific hardware you may need to write a plugin | 15:56 |
| monich | although there is already one for qmi | 15:56 |
| monich | or do the call it a driver | 15:57 |
| monich | https://git.kernel.org/pub/scm/network/ofono/ofono.git/tree/drivers/qmimodem | 15:57 |
| eduardas | monich: that is kinda my problem, qmi does everything I need except this... I wonder what qmi backend does for Powered property | 15:57 |
| monich | https://git.kernel.org/pub/scm/network/ofono/ofono.git/tree/plugins/ | 15:59 |
| monich | eduardas: grep for qmi there ^ | 16:00 |
| monich | modem power up/down should be there | 16:00 |
| monich | if qmimodem driver does the right thing for you, you may only need a plugin which implements the modem abstraction | 16:02 |
| monich | https://git.kernel.org/pub/scm/network/ofono/ofono.git/tree/plugins/gobi.c | 16:03 |
| monich | that ^ seems to be the qmi modem plugin | 16:03 |
| eduardas | monich: yes, at first glance this seems to be what I wanted... guess I need a plugin for my specific hardware, though I am not sure about the relationship/difference between plugin and backend? | 16:03 |
| Mister_Magister | abranson: now looking at (retarded) code its obvious, even if you set it to 2 it takes id 0 | 16:05 |
| eduardas | monich: I know you've already helped me a lot, but just one last question if I may: how does ofono know which plugins to load? | 16:07 |
| monich | eduardas: I think it's the plugin module which is responsible for detecting the hardware and providing the modem implementation which then tells the code which driver to use for each internal interface, see gobi_pre_sim and gobi_post_sim for example | 16:07 |
| monich | in that sense our ril thing isn't quite following that pattern | 16:08 |
| monich | it's all in one place, isn't broken into plugin and driver parts | 16:09 |
| monich | and in our case the modem configuration it fetched from the config file | 16:09 |
| monich | our case it generally mode complex, on some phones we are talking to rild via sockets, on some ohones via binder (android specific kernel ipc) | 16:11 |
| monich | so I ended up having plugins (socket vs binder) into our ril plugin | 16:12 |
| monich | and even a component that can be described as a plugin into a plugin into a plugin :D | 16:12 |
| monich | it's not something that I would recommend for studying as a reference implementation | 16:13 |
| eduardas | monich: that is really interesting to know, but it still does not answer my question - how does ofono know which plugin it should use in a specific case? There is a config file or is this done automatically? | 16:15 |
| Mister_Magister | abranson: I THINK ITS WORKING | 16:16 |
| monich | eduardas: https://git.kernel.org/pub/scm/network/ofono/ofono.git/tree/plugins/gobi.c#n462 | 16:17 |
| monich | see those ofono_netreg_create(modem, 0, "qmimodem", data->device); and so on& | 16:17 |
| monich | that "qmimodem" string tells ofono core which driver modules to use | 16:17 |
| Mister_Magister | let me double cehck | 16:18 |
| eduardas | monich: thank you for the useful answer, though I still do not understand how this relates to USB hotplugging | 16:25 |
| eduardas | and what calls what | 16:25 |
| Mister_Magister | abranson: https://twitter.com/Mister1Magister/status/1290325207668981760 | 16:34 |
| Mister_Magister | am i first in history? i think i am | 16:35 |
| Mister_Magister | also mal ^ | 16:36 |
| eduardas | monich: anyway, thank you a lot for your time. Your help is really appreciated. I think you've put me on the right track. Got to leave now. Will probably drop in here some time later. | 16:41 |
| Mister_Magister | abranson: u there | 16:49 |
| T42 | <ippokratis> @elros34 ubu-exec.sh is also start.sh now? | 16:52 |
| Mister_Magister | mal: abranson: when adding support for unlimited cameras… i think it should be per device config cause 5z is reporting 4 cameras but somehow droid-camres can get only 3 and some cameras are like 2mpix for bokmeh effect which won't be usable as camera either so we should pick which one to use | 16:52 |
| Mister_Magister | dang this is retarded on so many levels | 19:53 |
| Mister_Magister | mal: abranson: i present you this https://twitter.com/Mister1Magister/status/1290383374994022409 | 20:26 |
| Mister_Magister | https://github.com/sailfishos/gst-droid/pull/56 https://git.sailfishos.org/mer-core/qtmultimedia/merge_requests/34 my job is done here abranson mal | 20:47 |
| Mister_Magister | i'm selling rpms if there is someone with 2+ cameras and if they want to try, just pm me | 20:57 |
| mal | Mister_Magister: the PR is just terrible, huge amount of unnecessary formatting changes like indentation etc | 21:08 |
| mal | at least the gst-droid one | 21:08 |
| Mister_Magister | mal: i never said it isn't | 21:08 |
| Mister_Magister | it's just proof of concept to be done properly | 21:08 |
| Mister_Magister | it's not to be accepted, more like base to do it properly | 21:08 |
| mal | well reviewing that would be a pain with all the useless changes | 21:08 |
| * Mister_Magister shrug | 21:09 | |
| Mister_Magister | mal: well i'm sorry i cann't really roll it back | 21:09 |
| mal | why not? | 21:09 |
| Mister_Magister | ehhh | 21:09 |
| mal | well don't expect anyone to look at such a mess | 21:09 |
| Mister_Magister | ehhh | 21:10 |
| mal | do you really think for example I would want to go through every single line in the file manually because you changed almost every line | 21:12 |
| mal | trying to find any actual changes would take way too long | 21:12 |
| Mister_Magister | fine… | 21:12 |
| Mister_Magister | mal: what about now, lot better right? | 21:21 |
| Mister_Magister | mal: now i've removed your changes so it's only my changes now. super clean right? :) | 21:24 |
| Mister_Magister | i was just tired | 21:24 |
| * Mister_Magister lets see if it even compiles | 21:25 | |
| mal | at least it's readable now | 21:25 |
| Mister_Magister | :) | 21:26 |
| Mister_Magister | i did good job then :3 | 21:26 |
| mal | I think we should actually send to direction of the camera from droidmedia to gst-droid | 21:26 |
| mal | *the direction | 21:26 |
| Mister_Magister | but it's being done | 21:27 |
| mal | hmm | 21:27 |
| Mister_Magister | mal: https://github.com/sailfishos/droidmedia/blob/master/droidmediacamera.cpp#L286 | 21:27 |
| Mister_Magister | info.facing is from android and from droidmedia | 21:27 |
| mal | ok, I thought it wasn't because of the comment in qtmm PR | 21:28 |
| Mister_Magister | https://github.com/sailfishos/gst-droid/pull/56/files#diff-342a1c4301107bde6e204fb5f3cda8d8R450 and i grabbed it | 21:28 |
| Mister_Magister | mal: problem is here https://git.sailfishos.org/mer-core/qtmultimedia/merge_requests/34/diffs#32f812d7401e19e3d1b8c46b6eeec4a4b3f343e5_551_542 | 21:28 |
| Mister_Magister | for testing i set it to back camera. There is no property with camera orientation (or there is and i'm unaware) but such property needs to be added, array or something | 21:29 |
| Mister_Magister | mal: some changes are missing i'll fix it as long as i'm councious | 21:32 |
| mal | Mister_Magister: also fix the remaining formatting issues in gst-droid PR if you are changing it at some point | 21:33 |
| Mister_Magister | what formatting isssues? | 21:33 |
| mal | space messing between function name and ( | 21:33 |
| mal | if you look at the existing code there is always a space there | 21:33 |
| Mister_Magister | ah ok i'll handle it | 21:34 |
| Mister_Magister | note for future: don't fix formatting when you aren't asked to | 21:37 |
| Mister_Magister | mal: oki fixes done, and it's compiling | 21:41 |
| Mister_Magister | mal: all good now? | 21:42 |
| Mister_Magister | i've also added sfos-meeting topic about that | 21:51 |
| Mister_Magister | also if somebody wants to test hit me up cause i'm interested if it's working on other devices | 21:52 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!