| sicelo | add it. | 08:28 |
|---|---|---|
| Wizzup | yeah | 11:20 |
| sicelo | btw how to passthrough a modem into qemu? last night i wanted to use my laptop's modem in leste vm | 11:56 |
| sicelo | i added `-device usb-ehci,id=ehci -device usb-host,id=modem0,vendorid=0x0bdb,productid=0x193e` in my qemu command line, but that didn't seem to have any effect | 11:57 |
| Wizzup | what I do: | 12:10 |
| Wizzup | https://pastebin.com/raw/6QY57HxV | 12:10 |
| Wizzup | I think that should be it | 12:11 |
| Wizzup | you might awnt either ehci.0 or usb-bus.0 depending on the speed | 12:11 |
| sicelo | no luck. | 12:15 |
| sicelo | wonder what's wrong :-/ | 12:15 |
| Wizzup | maybe modemmanager grabs the modem on your host | 12:15 |
| Wizzup | you should see usb resets in dmesg by qemu | 12:16 |
| Wizzup | also some modems change usb ids once they are in modem mode | 12:16 |
| Wizzup | https://www.aptly.info/ this seems nice, nicer than reprepro | 12:17 |
| Wizzup | maybe we can use it at some point | 12:17 |
| sicelo | ModemManager is disabled, and nothing happens on dmesg | 12:17 |
| Wizzup | are you sure you got the right usb ids? | 12:18 |
| sicelo | yeah :-) | 12:19 |
| sicelo | $ lsusb | grep Ericsson | 12:19 |
| sicelo | Bus 003 Device 005: ID 0bdb:193e Ericsson Business Mobile Networks BV N5321 gw | 12:19 |
| Wizzup | I don't know if the id= matters, but try what I did in my paste for the id ust in case | 12:19 |
| Wizzup | you can also start qemu with monitor mode (as in my paste) and then manually add usb devices and see if that does anything different | 12:20 |
| sicelo | maybe the vm's kernel doesn't support my modem | 12:25 |
| sicelo | at least qemu itself appears to see the modem, https://paste.debian.net/1340239/ | 12:26 |
| Wizzup | You should see it in lsusb regardless | 12:37 |
| sicelo | really weird. it's a built in modem | 12:46 |
| sicelo | anyway, will look at it some other time | 12:54 |
| Wizzup | sicelo: maybe it's already bound to some kernel driver | 13:27 |
| Wizzup | you might have to rmmod those | 13:27 |
| sicelo | that might be it indeed. will do | 13:42 |
| mkf | hello there. | 18:07 |
| freemangordon | hi | 18:09 |
| mkf | i've got a n900! :D | 18:10 |
| mkf | sadly it's locked. | 18:10 |
| mkf | what can i do for it? | 18:11 |
| freemangordon | better ask on #maemo | 18:12 |
| mkf | ah. | 18:12 |
| mkf | hanks | 18:12 |
| dsc__ | vkb progress https://plak.infrapuin.nl/selif/tdhuoq5x.mp4 | 18:13 |
| freemangordon | dsc__: :) | 18:13 |
| freemangordon | nice | 18:13 |
| sicelo | nice. that's in a Qt application? | 18:14 |
| dsc__ | cheated a bit, I'm not "properly" communicating yet, needs some more work etc ;p | 18:14 |
| dsc__ | yes | 18:14 |
| dsc__ | via a Qt input module (plugin) | 18:14 |
| dsc__ | https://github.com/kroketio/qt5-platform-him (heavy WIP) | 18:15 |
| inky | > vkb progress https://plak.infrapuin.nl/selif/tdhuoq5x.mp4 | 18:17 |
| inky | what are you trying to accomplish? | 18:17 |
| freemangordon | inky: hmm? | 18:17 |
| freemangordon | vkb in qt | 18:17 |
| inky | any qt? | 18:17 |
| inky | non patched? | 18:18 |
| freemangordon | what is 'patched qt'? | 18:18 |
| inky | i thought maemo uses own forks of gtk2, gtk3 and qt for him to work. | 18:18 |
| inky | no qt? | 18:18 |
| freemangordon | qt is not forked | 18:18 |
| inky | oh | 18:18 |
| freemangordon | it has maemo platform plugin | 18:19 |
| freemangordon | gtk3 is not forked as well | 18:19 |
| inky | what did you do with gtk3? | 18:19 |
| freemangordon | rephrase, I don't understand the question | 18:20 |
| inky | and if qt has platform plugin, what dsc is trying to improve? | 18:20 |
| inky | i remember you was working on gtk3 and | 18:20 |
| freemangordon | we don;t have qt plugin for maemo vkb | 18:20 |
| inky | since that vkb works in dino let's say. | 18:20 |
| freemangordon | this is what dsc__ is doing | 18:20 |
| freemangordon | I made gtk3 input method plugin | 18:21 |
| freemangordon | well, ported gtk2 one | 18:21 |
| freemangordon | but that does not mean qt applications have vkb because gtk3 ones have | 18:21 |
| dsc__ | <inky> and if qt has platform plugin, what dsc is trying to improve? <== platform plugin != input handling module, there are different types of plugins for Qt, they are under "/usr/lib/x86_64-linux-gnu/qt5/plugins", O | 18:22 |
| freemangordon | :nod: | 18:22 |
| dsc__ | I'm working on a "platforminputcontext" for qt5, which works with hildon/him for its vkb | 18:22 |
| freemangordon | ther is also maemo5 style plugin | 18:22 |
| inky | okay i didn't know vkb doesnt work with qt. | 18:22 |
| dsc__ | freemangordon: indeed | 18:22 |
| freemangordon | inky: it does not | 18:23 |
| inky | i probably forgot since for months i use only my friend's coolkbd, which is abre to enter characters everywhere. | 18:23 |
| freemangordon | maybe we shall ask your friend to work on HIM vkb and make it working everywhere :) | 18:24 |
| dsc__ | yes please :) | 18:25 |
| arno11 | dsc__: nice @vkb :) | 18:51 |
| inky | > maybe we shall ask your friend to work on HIM vkb and make it working everywhere :) | 19:06 |
| inky | he won't. he likes to write own things. but | 19:06 |
| inky | * i can package coolkbd and write a wiki page on how to use it. | 19:06 |
| inky | * i can port all vkb language configurations to coolkbd | 19:07 |
| inky | or | 19:07 |
| inky | * i know what did he do. i can work with someone who understands vkb and i think if they eliminate that window which opens to write to buffer, and catch what vkb buttons show, then there's a function in xorg which adds that button to the current layout. and it can be used. | 19:08 |
| Wizzup | mkf: if you get stuck on unlocking or #maemo is not response, do ask again here and we can try to help | 19:17 |
| Wizzup | s/response/responsive/ | 19:17 |
| tmlind | freemangordon: starting uart gsm mode from the phy driver makes sense to me, but not so sure about the audio driver accessing /dev/gsmtty instances though.. | 19:25 |
| freemangordon | tmlind: yeah, for 2 days I am thinking of it, not a good idea | 19:28 |
| freemangordon | serdev_ngsm seems the proper way, but with doing dlci links serdev too | 19:29 |
| freemangordon | not sure how easy would be to achieve that | 19:29 |
| freemangordon | arno11: do we have caltool installed on leste n900? | 19:30 |
| tmlind | freemangordon: yup dlci links serdev too would be ideal | 19:34 |
| freemangordon | yeah, buty I am not sure when (and if) I will have time and will to do it | 19:34 |
| freemangordon | *but | 19:34 |
| freemangordon | given that so far motmdm is the only user | 19:35 |
| arno11 | freemangordon: @caltool, apparently not | 19:36 |
| freemangordon | devlocktool? | 19:36 |
| freemangordon | arno11: ^^^ | 19:36 |
| freemangordon | I have it in the VM | 19:37 |
| freemangordon | on d4 too | 19:37 |
| arno11 | i have libdevlock-bin | 19:37 |
| arno11 | i don t have devlocktool | 19:39 |
| freemangordon | it is in libdevlock-bin | 19:40 |
| freemangordon | tmlind: not sure what to do :( | 19:41 |
| freemangordon | sicelo: tmlind said you mentioned some usermode daemon to control audio instead of alsa, could you elaborate? | 19:42 |
| arno11 | freemangordon: ah ok | 19:43 |
| freemangordon | tmlind: so, do you know - in order dlci link to become serdev device, shall I create uart device for each link? or, tty device is enough? | 19:46 |
| freemangordon | if latter, then creating new tty driver is the proper way, or? | 19:47 |
| sicelo | freemangordon: tmlind: i don't think it's useful/related to the d4 issue(s). although i don't fully remember what it was, i guess it was either callaudiod or wys, or even eg25-manager. | 20:06 |
| tmlind | freemangordon: each dlci is really packet data at that point.. a tty should not be needed | 20:14 |
| freemangordon | so just register serdev? but then, how would it appear in /dev? | 20:15 |
| tmlind | there is no /dev entry, it's all kernel except for the ones that need a tty and a /dev entry for userspace access | 20:15 |
| freemangordon | sorry for maybe stupid questions, but I am not really into guts of uart/tty/serial bus | 20:16 |
| freemangordon | so, you say we shall not expose any dlci to userland? | 20:16 |
| tmlind | there is a mask in the dts files with the ngsm patches for the dlci that get the /dev entry created | 20:17 |
| freemangordon | I mean - volume control dlci stays in kernel (so no /dev), but what about the others? | 20:17 |
| freemangordon | yes, I saw that | 20:17 |
| tmlind | well gnss allows writes over the /dev/gnss so no dlci needed there either | 20:17 |
| freemangordon | but, if I am to implement each dlci as serdev, wouldn;t we want unused (by kernel) links to be userspace exposed? | 20:18 |
| tmlind | the unused ones could have n_gsm created /dev/gsmtty | 20:18 |
| tmlind | no, serdev is for kernel drivers | 20:18 |
| freemangordon | ah, I see now | 20:18 |
| tmlind | then the protocol layer whatever bluetooth/gnss/audio exposes stuff to userspace | 20:19 |
| freemangordon | ok | 20:19 |
| freemangordon | got it | 20:19 |
| freemangordon | so, do you have a link to the latest serdev_ngsm version on lore, I want to see the commets/requirements | 20:20 |
| tmlind | i guess rather the kernel device framework like bluetooth/gnss/audio (instead of protocol layer..) | 20:20 |
| freemangordon | yeah, got it | 20:20 |
| tmlind | just search for gsm serdev in lore and they should all come up | 20:20 |
| freemangordon | or, what is the lates version setn? | 20:20 |
| freemangordon | ok | 20:20 |
| tmlind | so maybe the only thing that is really needed is.. make the child device read/write use serdev | 20:21 |
| tmlind | instead of the custom read/write functions | 20:21 |
| freemangordon | ok, will have a look during the holidays | 20:21 |
| tmlind | ok | 20:21 |
| tmlind | freemangordon: i think serdev gsm may need to just call serdev_controller_add() for each dlci? | 20:23 |
| tmlind | and i guess serdev_device_add() | 20:24 |
| freemangordon | ok, will have a look at how it's done for normal serdev devices, thanks for the hint | 20:25 |
| tmlind | freemangordon: hmm i guess ideally one serdev controller (which we might already have) that calls serdev_device_add() for each dlci? this just based on memory though.. | 20:36 |
| freemangordon | ok, lemme have a look | 20:41 |
| freemangordon | serdev_tty_port_register already calls serdev_controller_alloc | 20:43 |
| freemangordon | oh, wait, that's not a new code | 20:44 |
| freemangordon | tmlind: do I get it right, that gnss shall be changed to serdev_device_driver from platform_driver? | 20:47 |
| freemangordon | the same for motmdmd audio | 20:47 |
| freemangordon | *motmdm | 20:47 |
| arno11 | sicelo: i got image generation working on n900 :P | 22:44 |
| arno11 | what is working atm: chat, vision, img generation, audio, model and api key selection, saving-loading chat, temperature, uploading files | 22:56 |
| arno11 | using libcurl | 22:58 |
| Wizzup | cool :) | 23:21 |
| arno11 | :) the ui is quite fast btw but i need to clean up lot of things before sharing | 23:29 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!