tmlind | freemangordon: ok i'll take a look at rebasing ngsm serdev patches against v6.0-rc series when i can, then test with restarting ofono and looking for the unsol messages | 06:51 |
---|---|---|
tmlind | freemangordon: that's after no double open issuses :) if there's some other test you're using, lmk | 06:52 |
freemangordon | tmlind: not sure I understand fully, but ok :) | 07:58 |
freemangordon | OTOH, if you give me a couple of hints, I can spend some more time on it (serdev driver) | 07:58 |
freemangordon | like, I don't understand the layering, it seems here is so many abstraction layers that at the end of the day n_gsm/tty code is like spaghetti to me :) | 08:00 |
freemangordon | *there are | 08:00 |
freemangordon | also, I was not able to find who is sending the bytes over the wire - is it some OMAP uart? or usb->serial convertor? who it is? | 08:01 |
tmlind | freemangordon: yeah the traffic goes over a 8250_omap uart | 08:05 |
freemangordon | ah, and we do power management on the opposite side | 08:06 |
tmlind | n_gsm sets up the channels, serdev takes care of io | 08:06 |
tmlind | then each channel can have a real kernel driver, like the gnss driver for mdm6600 | 08:06 |
freemangordon | I see | 08:07 |
tmlind | earlier all that was done over userspace via whatever attach function to init n_gsm, but did not work with pm at all | 08:07 |
freemangordon | yeah, I was wondering why kernel driver is needed | 08:08 |
freemangordon | can;t we just have some small one to control modem gpios? | 08:08 |
tmlind | heh ttried that :) | 08:08 |
freemangordon | why it would not work? | 08:09 |
freemangordon | could it be because n_gsm was too buggy back then? | 08:09 |
freemangordon | not that I know how kernel driver (gnss for example) can open tty port :) | 08:10 |
tmlind | the traffic works with userspace tools too, but the userspace has no clue about the pm | 08:10 |
tmlind | some more info here, that's what i initially tried: https://github.com/tmlind/droid4-ngsm/blob/master/droid4-ngsm.c | 08:11 |
tmlind | so with serdev-gsm, we can have proper voice modem audio controls via alsa, gnss driver and so on | 08:13 |
freemangordon | ok, what userspace has to do is more or less clear | 08:13 |
freemangordon | ah, I see | 08:13 |
freemangordon | makes sense | 08:13 |
tmlind | also android has kernel ts27010 driver because of this | 08:13 |
freemangordon | yeah | 08:13 |
tmlind | android has no serdev, so all the modem features are done in user space libraries | 08:14 |
freemangordon | ok, so serdev driver shall do what userspace does, right? | 08:14 |
freemangordon | more or less | 08:15 |
tmlind | serdev replaces pretty much all the old hciattach, irattach whatever attach glue stuff | 08:15 |
freemangordon | not that I know those :) | 08:15 |
tmlind | well it sucked to use hciattach or irattach for a long time :) | 08:16 |
freemangordon | IIUC for porting to 6.x some of the init code in serdev shall be removed | 08:16 |
freemangordon | as this is already taken care of in n_gsm | 08:16 |
tmlind | ok | 08:16 |
freemangordon | like explicit control channel opening | 08:16 |
tmlind | ok | 08:17 |
freemangordon | ok, I'll spend some more time on that, now I have a better idea what happens | 08:17 |
tmlind | the serdev-gsm has some issues to fix before upstreaming, johan wanted it to behave more like serdev and less like a mdm driver | 08:17 |
tmlind | just fyi | 08:17 |
freemangordon | well, I am almost sure I won;t be able to make it work, but ok | 08:18 |
tmlind | heh | 08:18 |
tmlind | the recent n_gsm fixes are good though, earlier the whole n_gsm layer was based on some earlier version of the protocol | 08:19 |
freemangordon | yes, they say they put that in production | 08:19 |
tmlind | yeah | 08:20 |
freemangordon | but, they break serdev :) | 08:20 |
tmlind | so ideally the voice call audio mixer and gnss drivers just would use plain serdev read write functions | 08:20 |
freemangordon | without opening tty? | 08:21 |
tmlind | no just plain serdev_read instead of serdev_ngsm_read in drivers/gnss/motmdm.c for example | 08:22 |
freemangordon | ah | 08:23 |
freemangordon | now I got it | 08:23 |
tmlind | so each dlci would show up as a serial port for the serdev drivers | 08:23 |
tmlind | we already just pass the n_gsm fifo around, should be totally doable | 08:23 |
tmlind | not sure what needs to be done about the new double init issue, but sounds like you just need to leave out some n_gsm related call from the serdev init in n_gsm.c | 08:25 |
freemangordon | yeah | 08:26 |
freemangordon | I already removed tty port register | 08:26 |
tmlind | ok | 08:26 |
freemangordon | but, have to figure out how to do the control port init correctly | 08:26 |
tmlind | ok | 08:26 |
sicelo | tmlind: so you're quite conversant with serdev? i may need some handholding working with N900's bluetooth. uses serdev too | 08:40 |
tmlind | sicelo: not much to it, see for example what sre did for drivers/bluetooth/btwilink.c | 08:44 |
tmlind | sicelo: hmm but drivers/bluetooth/hci_nokia.c is already using serdev? | 08:45 |
tmlind | it should just work | 08:45 |
sicelo | yes. he already ported the driver. except it never worked for N900, but only N950. can't remember about N9 | 08:46 |
sicelo | the firmware doesn't load. he did troubleshoot back then and it seemed that there was nothing obviously wrong. i don't think i'm even going to fix it either, but i still have interest in N900, and wouldn't mind 'wasting' some cycles on these | 08:48 |
tmlind | ok interesting | 08:51 |
sicelo | i don't see drivers/bluetooth/btwilink.c anywhere (looking at linux 6.0-rc3) | 09:01 |
tmlind | sicelo: oh sorry, that's the old driver that got dropped in favor of the generic driver | 09:17 |
freemangordon | hmm, I think I know how to make that (serdev) work | 09:18 |
tmlind | sicelo: on d4, you can just do modprobe hci-uart; hciconfig hci0 up; hcitool scan | 09:18 |
tmlind | freemangordon: ok great | 09:19 |
freemangordon | lemme recompile | 09:19 |
freemangordon | takes time :( | 09:19 |
freemangordon | yep, it does work :) | 09:40 |
freemangordon | ok, going to the gym, will continue laters | 09:44 |
sicelo | freemangordon: nice | 10:53 |
sicelo | tmlind: yes I've tested bt on d4 successfully before | 10:54 |
uvos | i use bt on d4 all the time | 11:11 |
uvos | its fine, except you have to rmmod the module again for it to hit ret | 11:12 |
uvos | so the driver needs some pm work | 11:12 |
freemangordon | tmlind: 2 questions: | 11:14 |
freemangordon | 1. shall I restore tty register behaviour or leave it to n_gsm driver? Now it unconditionally registers 63 tty devices | 11:14 |
freemangordon | 2. shall I squash my change to your commit or you prefer separate ("port to upstream") commit? | 11:14 |
freemangordon | *chnages | 11:17 |
freemangordon | argh... changes | 11:17 |
tmlind | freemangordon: 63 tty devices should be ok.. not ideal but who cares | 11:39 |
freemangordon | ok | 11:39 |
tmlind | freemangordon: maybe squash it to the earlier commit and add your s-o-b, the patches need to be rebased anyways for upstream | 11:40 |
freemangordon | right | 11:40 |
freemangordon | ok, will do | 11:40 |
tmlind | ok great, so you got it working? | 11:40 |
freemangordon | yes | 11:40 |
tmlind | cool, i'll set up v6.0 patches when i get a chance | 11:40 |
freemangordon | ok, I'll do some more cleanup and experinmenting and will push to leste tree | 11:41 |
tmlind | ok | 11:56 |
freemangordon | tmlind: do we really need init_retry_quirk? | 12:10 |
tmlind | freemangordon: not sure, maybe the need for that has gone now with all the n_gsm fixes | 12:16 |
freemangordon | mhm, I will remove it | 12:16 |
freemangordon | tmlind: shall I use wait_event_timeout or wait_event_timeout_interruptable in gsm_serdev_register_dlci while waiting for control channel to be opened? | 12:29 |
freemangordon | or simply wait_event? | 12:33 |
buZz | sooooo, any updates on the 2/3G IAPs? :D | 12:41 |
buZz | oh hey, now i'm getting the 'simcard not found' icon | 12:42 |
buZz | maybe my sim just came loose again | 12:43 |
tmlind | freemangordon: whatever works :) | 12:49 |
tmlind | bbl | 12:49 |
freemangordon | heh | 12:49 |
freemangordon | ok :) | 12:49 |
buZz | hmm dangit, seems i cant get sim registered in ofono? | 16:31 |
buZz | at least, reseating million times, powercycling, even swapped it all over to second d4, keep giving 'sim unregistered' icon in statusbar | 16:32 |
Wizzup | buZz: what ofono version is this? | 16:34 |
buZz | 1.34.3-1+2m7 , from -devel afaik | 16:35 |
Wizzup | there shouldbe 1.34.4 | 16:37 |
Wizzup | is there not? | 16:37 |
Wizzup | hm there is not | 16:37 |
Wizzup | building new one | 16:39 |
Wizzup | buZz: update and upgrade in ~5 mins | 16:41 |
Wizzup | then it should be fixed | 16:41 |
buZz | alrighty, lets see | 16:47 |
buZz | alright, getting 2G icon again | 17:04 |
buZz | still no GPRS IAP visible, i deleted the one through gconftool-2 -u --recursive-unset | 17:05 |
buZz | lets see | 17:05 |
Wizzup | well, whatever ofono fixes freemangordon did likely do not relate to your provisioning problem(s) | 17:05 |
buZz | it didnt reprovision it either! (on powercycle) | 17:12 |
Wizzup | not surprised, that never worked for you right? | 17:13 |
Wizzup | is this still youfone? | 17:13 |
buZz | yeah | 17:14 |
buZz | i do see it in debug_netreg() calls in iap_conndlg.log | 17:14 |
Wizzup | aren't those just rssi stuff? | 17:16 |
buZz | grep -a OFONO /var/log/maemo/icd2.log <- shows me it always said ' | 17:17 |
buZz | 'sim seen for the first time, provisioning' after a 'status 2' msg | 17:18 |
buZz | but since yesterdays updates no more | 17:18 |
buZz | no more Status 2 | 17:18 |
buZz | no more 'search finished' either, not sure what that means | 17:18 |
buZz | Wizzup: it did provision every time -again- when starting ofono (or icd2? not sure) | 17:49 |
buZz | just overwrote the old settings | 17:49 |
buZz | but now its not even doing the initial provisioning | 17:49 |
norayr | Uvos, i think there is important bug in new backend of vkbd | 17:56 |
norayr | Shouldn't i be able to write armenian/etc with it (i. e. by pressing search button on droid4) in gtk2 apps? | 17:58 |
norayr | I write by tapping only, then i trgger other backend by search button to press enter | 18:01 |
norayr | Bcs enter doesnt work in maemo old backend | 18:01 |
norayr | But what i say new backend should also work with pidgin and other gtk2 apps, no? | 18:02 |
uvos | norayr: not sure what you are saying | 18:25 |
uvos | norayr: you should be able to write armenian in gtk apps by tapping on a field | 18:26 |
uvos | you should not be able to write armenian into gtk or any apps when using the xcb backend | 18:27 |
uvos | unless you have armenian keysims available via xkb keycodes | 18:27 |
uvos | (ie have armeinan xkb layout active) | 18:27 |
uvos | enter handling via the old backend is a bit wierd in gtk2 when using the hildon backend, as there is logic when to enter should advance fields, when enter should give the app "enter" and some other stuff | 18:30 |
uvos | this is in gtk tho not in him itself | 18:30 |
uvos | this works ok for maemo apps, but yeah some desktop gtk2 apps that do something special with enter dont like it (i gues thats pdigin) | 18:30 |
uvos | so i dont see the bug here | 18:32 |
norayr | I mean | 18:53 |
norayr | Shuldnt i be able to write armenian in gtk2 pidgin with new backend (search button) and without typing setxkbmap am in terminal. | 18:55 |
uvos | no | 18:55 |
norayr | Hmmmm | 18:55 |
uvos | the target window widget set is immaterial to the xcb backend | 18:55 |
norayr | Okay | 18:55 |
freemangordon | tmlind: what is "etsi,3gpp-ts27010-adaption1" ? | 19:15 |
buZz | freemangordon: any pointers for me to regain provisioning my gprs sim? :) | 19:38 |
freemangordon | nope | 19:38 |
buZz | (no rush) | 19:38 |
freemangordon | I am still at trying to make kernel driver stable | 19:38 |
buZz | any ideas then? :) | 19:38 |
buZz | ah alright | 19:38 |
buZz | np | 19:38 |
freemangordon | restart ofono | 19:38 |
buZz | yeah i've done all the dances thusfar, nothing seems to kick it to report 'status 2' to icd2.log | 19:39 |
buZz | (which seems required to have it provision itself) | 19:39 |
freemangordon | sorry, no idea | 19:40 |
buZz | aw, alright tnx | 19:40 |
buZz | woooot, restarting ofono and icd2 a bunch of times DID help! | 19:55 |
buZz | Sep 1 19:53:14 localhost icd2 0.98[5884]: [OFONO] Status 2 | 19:55 |
buZz | Sep 1 19:53:14 localhost icd2 0.98[5884]: [OFONO] SIM 204086485211992 seen for the very first time, provisioning. | 19:55 |
buZz | Sep 1 19:53:14 localhost icd2 0.98[5884]: [OFONO] Search finished | 19:55 |
buZz | still no APN in gconf, but, it works | 19:57 |
freemangordon | Wizzup: is it ok to push the new kernel in a separate branch? | 23:30 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!