Wizzup | arno11: great | 00:09 |
---|---|---|
freemangordon | tmlind: heh, I think I found the missing events issue :) . here https://github.com/maemo-leste-upstream-forks/ofono/blob/maemo/chimaera-devel/plugins/udevng.c#L410 | 08:04 |
duuude | U-Boot with kernel 2.6.28-omap1 won't get installed | 08:05 |
duuude | it violates "third party package policy" | 08:05 |
duuude | what to exactly do now? | 08:05 |
freemangordon | it takes the last wdm device (cdc-wdm3). if I change that to use first device (cdc-wdm0), at least NAS and SMS notifications are comming | 08:06 |
tmlind | freemangordon: oh nice :) | 08:07 |
freemangordon | this is for device in BP mode | 08:07 |
freemangordon | but I assume it will be the same if we boot leste | 08:07 |
tmlind | should be yeah :) | 08:07 |
freemangordon | will test that later on | 08:08 |
freemangordon | also, I have upstream ofono working with it, without gsmmux | 08:08 |
freemangordon | is it needed for voica calls, right? | 08:08 |
freemangordon | or, is it? | 08:08 |
freemangordon | if qmimodem in ofono support voice calls over qmi, do we need gsmmux? | 08:09 |
tmlind | for voice call routing and mixer we need something for alsa | 08:09 |
freemangordon | but do we need gsmmux? | 08:09 |
tmlind | gnss should be doable over qmi too | 08:10 |
freemangordon | cool | 08:10 |
tmlind | you can use the tool i did to dump and diff the modem nvram and attempt to replicate the same with qmi commands | 08:10 |
freemangordon | wait, WTYM? qmi interface is standard | 08:11 |
tmlind | alsa needs some notification, but that might be doable with the audiocalld or whatever the newish daemon was called | 08:11 |
tmlind | motorola implemented some custom calls that likely just tweak some nvram options and the modem firmware then sets things as requested | 08:11 |
tmlind | the qmi calls also mostly just set nvram values fyi | 08:12 |
freemangordon | are you sure motorola implementd custom qmi calls? what for? | 08:12 |
tmlind | for their audio setup i think | 08:13 |
freemangordon | I just doubt that motorola had to do it besides perhaps some FW download | 08:13 |
freemangordon | hmm | 08:13 |
freemangordon | ok, will see | 08:13 |
tmlind | anyways, you can diff the nvram values with https://github.com/tmlind/tcmdrw | 08:13 |
tmlind | maybe you can dump the nvram over qmi too, don't know | 08:14 |
freemangordon | me neither | 08:14 |
freemangordon | qmicli is far from being complete | 08:14 |
freemangordon | but I found qmi specs over the inet | 08:14 |
tmlind | anyways, if you got the qmi notifications working for incoming sms, we can get pretty far with not gsmmux stuff | 08:15 |
freemangordon | yes, I sent sms and saw the notification | 08:15 |
tmlind | great | 08:15 |
freemangordon | tmlind: https://pastebin.com/AAHkamJQ | 08:17 |
tmlind | anyways, if we end up adding the gsmmux stuff to ofono later on, it should be implemented with raw read and write using ell | 08:17 |
freemangordon | but we still need the kernel driver, right? | 08:17 |
tmlind | no need to try to use atmodem stuff for it | 08:17 |
tmlind | right, kernel driver or the userspace tool that won't work with pm | 08:18 |
tmlind | whatever attach it was called | 08:18 |
freemangordon | ok I lack teh details here | 08:18 |
tmlind | cool nice to hear you figured out the missing notifications | 08:18 |
tmlind | anyways, i agree with you, let's try to use mainline ofono with qmi where possible and then patch in the gsmmux if and when needed | 08:19 |
freemangordon | :nod: | 08:19 |
freemangordon | I'll try to prepare and send patches for upstreaming of motmdm | 08:21 |
tmlind | for the voice call alsa driver, don't recall if we need to do some gsmmux modem commands beyond mixer and dtmf | 08:21 |
freemangordon | I have no clue | 08:21 |
freemangordon | but anyways we need qmi voice calls working in upstream first | 08:22 |
tmlind | it could be we're only configuring the mcbsp and pmic for the tdm use for voice call | 08:22 |
freemangordon | mhm | 08:22 |
freemangordon | but you do that in the driver | 08:22 |
tmlind | yeah | 08:22 |
freemangordon | which looks like a hack as well, but that's another story :) | 08:23 |
tmlind | right, it should be the whatever voicecallaudiod doing it nowadays and that likely already has qmi support | 08:23 |
freemangordon | ok, thanks! will keep you informed on the progress. | 08:23 |
freemangordon | bbl | 08:24 |
tmlind | ok great, ttyl | 08:24 |
sicelo | note that voicecallaudiod might not have a future ... there's talk to drop it once pipewire can handle some of those things by itself | 08:25 |
tmlind | oh ok good to know, anyways nice to hear there's some kind of userspace path for dealing with these things | 08:26 |
duuude | U-Boot with kernel 2.6.28-omap1 won't get installed | 09:25 |
duuude | it violates "third party package policy" | 09:25 |
duuude | what to exactly do now? | 09:25 |
duuude | Wizzup freemangordon | 09:29 |
duuude | it wasn't like this the year before or at least I don't remember such a thing. | 09:30 |
sicelo | i said install it with apt. did you try that? | 09:33 |
duuude | sicelo: I probably wasn't online when you said that | 09:37 |
duuude | though, I found an apt command from postmarketos' gitlab issues which also didn't work | 09:37 |
sicelo | what command was it? | 09:39 |
duuude | apt install uboot-menu? was something like this | 09:39 |
sicelo | maybe share the link :-) ... there's no 'apt install' in maemo fremantle | 09:51 |
duuude | I don't know what was the problem the first time (I had entered "apt-get") but it worked this time: apt-get install u-boot-flasher u-boot-tools u-boot-bootimg | 09:56 |
duuude | sicelo: how to enable on-screen keyboard? I can't enter ">" | 09:59 |
duuude | these weren't problems the first time | 09:59 |
sicelo | what device? | 10:00 |
duuude | n900 | 10:00 |
duuude | on the terminal | 10:00 |
sicelo | Fn+Ctrl, i think ... the standard way in Fremantle | 10:00 |
freemangordon | tmlind: why not using AT interface for voice calls? | 10:13 |
freemangordon | I tried and got NO CARRIER, but this could be because I didn;t set-up the modem properly | 10:13 |
duuude | I got stuck in "Welcome to Leste" | 10:24 |
duuude | it doesn't seem to take input | 10:24 |
duuude | it somehow went away, but looking frozen isn't good ux | 10:26 |
sicelo | i think on first boot, N900 just can't handle the amount of stuff happening in the background. using a fast SD card helps somewhat | 10:41 |
Wizzup | freemangordon: great @ ofono | 11:03 |
sarah_ | duuude: see leste n900 wiki: https://leste.maemo.org/Nokia_N900#Tips_.2F_Tweaks (SD card and first boot) | 11:59 |
arno11 | btw "Welcome to Leste" seems to be a problem on N900 and a bit annoying for users | 12:08 |
arno11 | if you click nowhere for at least 3 min once H-D is visible, everything is fine then. Otherwise N900 is slowing down a lot and users think Leste is really slow and unusable on N900... | 12:11 |
Wizzup | we could have a splash screen, or ry to emit the h-d is ready signal later | 12:12 |
arno11 | a splash screen could be cool | 12:25 |
arno11 | btw there is quite a similar issue with 'pin entry code' applet: N900 is unusable until the applet appears on the screen. Not a major concern but also annoying for new users. | 12:43 |
arno11 | bbl | 12:43 |
duuude | sicelo, Wizzup: addition of a text "(the freeze is normal)" could be nice. | 14:23 |
duuude | would leste dry up the battery to the last second like before? | 14:39 |
uvos__ | the greeting text is common to all devies | 16:10 |
uvos__ | only the n900 has issues so its not so easy | 16:10 |
uvos__ | if it really is unresponsive for 3 minutes i dont think a splash screen is enough really | 16:11 |
uvos__ | that calls for at least a thobber otherwise people will think its frozen | 16:11 |
tmlind | uvos__: fyi https://lore.kernel.org/dri-devel/20240219142001.19727-1-tony@atomide.com/T/#t | 16:12 |
tmlind | freemangordon: well the at interface is buggy from what i recall.. | 16:13 |
tmlind | freemangordon: also, if you're booting to bp mode with m-l kernel, the gsmtty kick might cause the qmi notifications to happen for you on your x86 vm.. | 16:13 |
uvos__ | tmlind: sweet :) | 16:15 |
tmlind | uvos__: also another series fyi https://lore.kernel.org/dri-devel/20240215123222.42609-11-tony@atomide.com/ | 16:16 |
sicelo | for the first boot on N900, maybe let's introduce a lysti LED pattern, in a fast heartbeart, to signal to user that *something* is happening | 16:16 |
tmlind | i've moved to mostly automated git email scripts, currently missing option to add a manual cc list.. | 16:17 |
bencoh | sicelo: why not a "please wait while bootstrapping" message on first boot then? | 16:17 |
sicelo | because, ... 17:11 < uvos__> that calls for at least a thobber otherwise people will think its frozen | 16:20 |
bencoh | ah, so you mean both, nevermind | 16:22 |
uvos__ | a pattern would work yeah | 16:26 |
sicelo | freemangordon: "I want to discuss with | 16:40 |
sicelo | the author of QMI voice calls support in PMOS" ... who's that? | 16:40 |
freemangordon | sicelo: https://codeberg.org/305a385/pmaports/src/branch/master/temp/ofono/0005-qmimodem-implement-voice-calls.patch | 16:45 |
freemangordon | tmlind: I am booting from power-off and d4 is attached to my ubuntu laptop :) | 16:46 |
freemangordon | I am not sure what you meant | 16:46 |
freemangordon | qmi notifications didn't happen in that setup until I changed to /dev/cdc-wdm0 | 16:47 |
freemangordon | this https://pastebin.com/qwnjSXmG is how things look like with upstream ofono and d4 modem attached in BP only mode | 17:00 |
freemangordon | only qmi interface is used, not AT, no MUX (gsmtty) | 17:00 |
freemangordon | not sure how to test gps though | 17:01 |
freemangordon | BTW, what is the 'official' name of the modem? m6600? | 17:04 |
uvos__ | freemangordon: if you use bp mode from the bootloader the deivce routes out the usb interface of the modem | 17:11 |
uvos__ | freemangordon: if you then boot the leste kernel it might affect the result because it uses the gsmtty interface | 17:11 |
uvos__ | is what he is saying | 17:12 |
freemangordon | uvos__: I understand that, but I think the plan is to try to not use gsmtty. Also, I doubt I will lose notifications in qmi because of gsmtty. However, will test on the device in couple of minutes and will see | 17:13 |
freemangordon | also, we'll have issue with notifications as currently I am not aware of any way to disable them (on qmi interface that is) outide of ofono | 17:14 |
freemangordon | and ofono does not expose any interface for doing that | 17:14 |
uvos__ | sure @plan, no idea at if gsmtty use by ofono on device affects qmi connected to the host | 17:14 |
freemangordon | it should not, but... who knows :) | 17:14 |
freemangordon | oh, but ofono *does not* use gsmtty | 17:15 |
freemangordon | well, the upstream one | 17:15 |
freemangordon | not what we have currently in leste | 17:15 |
uvos__ | sure but what we have in leste ist running in this configuration | 17:15 |
uvos__ | also mce kicks gsmtty | 17:15 |
freemangordon | but I will replace ofono on my d4 with upstream one | 17:16 |
uvos__ | right ok | 17:16 |
uvos__ | mce still kicks gsmtty iirc | 17:16 |
freemangordon | yeah | 17:16 |
freemangordon | do disable RSSI on display off, right? | 17:16 |
uvos__ | yeah and it just kicks it too | 17:16 |
freemangordon | why? | 17:16 |
uvos__ | for some reason doing open() and then close() on the if gets the modem "unstuck" | 17:17 |
uvos__ | the modem gets stuck in some wierd state where it uses more power | 17:17 |
uvos__ | for reasons unkown | 17:17 |
uvos__ | as a workaround we do that every 10 min or so | 17:17 |
freemangordon | uvos__: by 'modem' you mean AT modem, right? | 17:28 |
tmlind | freemangordon: so best to run your tests with no ofono running on d4 while testing :) | 17:28 |
tmlind | freemangordon: the modem name is mdm6600 | 17:29 |
freemangordon | tmlind: why no ofono (upstream)? | 17:29 |
freemangordon | it does not touch either AT modem (/dev/ttyUSBn) or gsmmux | 17:30 |
tmlind | freemangordon: only run ofono on your x86 during the test, not on d4 to avoid d4 patched ofono kicking the usb interface | 17:30 |
freemangordon | nothing more to test on x86 | 17:30 |
freemangordon | and I will replace ofono on d4 with upstream one | 17:30 |
tmlind | ok yeah interesting to see what happens :) | 17:31 |
freemangordon | mhm | 17:31 |
freemangordon | besides missing voice call it should work, in theory | 17:31 |
tmlind | Wizzup: i think your usb hang may be reproducable with something like this: mosh or ssh over usb doing dmesg -w, then unblank the lcd | 17:52 |
tmlind | so usb traffic while tiler doing something maybe? | 17:52 |
tmlind | at least i can hang the device pretty easily with that, no oops logs so far though | 17:53 |
tmlind | hehe yup, i've had two of these now: | 18:06 |
tmlind | dmm 4e000000.dmm: timed out waiting for done | 18:06 |
tmlind | omapdrm omapdrm.0: could not unpin pages: -110 | 18:06 |
tmlind | dmm 4e000000.dmm: timed out waiting for done | 18:06 |
tmlind | dmm 4e000000.dmm: timed out waiting for done | 18:06 |
tmlind | omapdrm omapdrm.0: could not pin: -110 | 18:06 |
tmlind | PVR_K:(Error): DmaBufImportAndAcquirePhysAddr: dma_buf_map_attachment failed: -110 | 18:06 |
tmlind | PVR_K:(Error): PVRSRVMapDmaBufKM: Failed to get dma-buf phys addr | 18:06 |
tmlind | PVR_K:(Error): PVRSRVMapDmaBufBW: Failed to map dma-buf handle | 18:06 |
tmlind | ... | 18:06 |
tmlind | and then bad things happen and device will reboot quit soon | 18:06 |
tmlind | i'll try enabling i878 workaround in omap_dmm_tiler.c, let's see if that helps | 18:19 |
freemangordon | ok, upstream ofono (with motmdm ported) works in leste, besides there is some issue with reporting RAT | 18:20 |
tmlind | cool | 18:23 |
tmlind | enabling i878 workaround does not help with the "timed out waiting for done" errors | 18:24 |
tmlind | freemangordon: so you got sms receiving working with just qmi? | 18:24 |
freemangordon | yes | 18:24 |
tmlind | freemangordon: and you get a qmi notification on disconnecting from a network with tin foil? :) | 18:25 |
freemangordon | tmlind: seems only RSSI and SMS notifications work properly so far | 18:25 |
tmlind | freemangordon: ok | 18:25 |
freemangordon | but this is a matter of fixing here and there, not some general issue | 18:27 |
tmlind | ok great | 18:28 |
freemangordon | hmm... actually | 18:29 |
freemangordon | I had to kick it | 18:30 |
freemangordon | maybe gsmtty behaves | 18:30 |
freemangordon | anyway, have to run now, bbl | 18:30 |
tmlind | yeah me too later | 18:31 |
freemangordon | tmlind: if I rmmod n_gsm, would that release gsmtty? | 22:12 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!