Wizzup | looks like syncevo broke in the chimaera upgrade | 00:39 |
---|---|---|
Wizzup | at least with our current instructions | 00:39 |
tmlind | Wizzup: ok yup that pretty much confirms that modem shuts down around 3.5v | 06:26 |
tmlind | Wizzup: also seems like we have some large cumulative error with the coulomb counter capacity after several charge cycles? | 06:26 |
tmlind | Wizzup: i guess the next step would be to check how fast the coulomb counter capacity gets out of line with the voltage based capacity estimate user space app | 06:27 |
tmlind | Wizzup: and the modem should singnal something before shutting down the usb phy, that's a bug | 06:28 |
tmlind | anyways, that might explain why we only get a few weeks of uptime | 06:29 |
tmlind | Wizzup: hmm that log of yours is not for invalid register access though like earlier | 06:31 |
rafael2k | freemangordon, right, but as far as I can check, it is all good for hitting master | 07:14 |
Wizzup | tmlind: it might also be different, since this was my bionic just doing gps, the other cases were my d4 with bt | 10:03 |
Wizzup | but in both cases the modem seemingly went away | 10:03 |
Wizzup | uvos__: fyi I will try to pick up the sphone work today | 10:31 |
Wizzup | I also received the xt910 | 10:31 |
Wizzup | SuperMarioSF: my impression seems to be that the phones look to be in a great state, with the box containing all the original stuff, I plugged in one of my batteries and am charging it now | 10:36 |
uvos__ | Wizzup: ok | 10:40 |
Wizzup | SuperMarioSF: phone booted fine :) | 13:12 |
SuperMarioSF | Wizzup: nice to hear that, happy hacking! | 13:58 |
buZz | Wizzup: ooo charger patches? exciting! | 14:09 |
Wizzup | buZz: I think you already knew that it was being worked on no? what I mean is, it's nothing new | 14:10 |
Wizzup | but yeah | 14:10 |
Wizzup | freemangordon has some stuff lined up | 14:10 |
bencoh | "ok yup that pretty much confirms that modem shuts down around 3.5v" that's super high :/ | 14:11 |
Wizzup | bencoh: right but it sounds like (I think) this is probably due to some other issue | 14:12 |
Wizzup | it is also possible that this can be changed and maybe android does that | 14:12 |
uvos__ | for sure android dosent have this issue | 14:23 |
uvos__ | then again it has never happend to me on leste either | 14:23 |
Wizzup | well it happens to me on several phones | 14:23 |
Wizzup | so it's not likely a hw issue | 14:23 |
buZz | Wizzup: the only modem-panics i got were during USB connected btw | 14:24 |
buZz | seemingly USB modem fighting with the USB gadget mode, or something | 14:24 |
bencoh | on bionic only? | 14:24 |
Wizzup | well then there are three different ones | 14:24 |
buZz | (and failing) | 14:24 |
Wizzup | bencoh: no, also on droid 4 for me | 14:24 |
bencoh | oh | 14:25 |
buZz | Wizzup: maybe the gadgetmode should be shutdown below ~x voltage? to prevent USB modem from disconnecting | 14:25 |
buZz | just a guess | 14:25 |
Wizzup | I don't think so, that sounds like a hack around the problem, and again, it's not the problem I am seeing | 14:25 |
Wizzup | it's also possible we maybe just draw (a lot) more power, and we also don't gracefully deal with the modem being gone for a bit in any case | 14:26 |
buZz | yeah its kinda flaky , ofono stop/start doesnt always recover it | 14:26 |
buZz | but sometimes it does | 14:26 |
Wizzup | that is for sure not the right solution :p | 14:29 |
buZz | right, but how should i restart the modem then? :D | 14:32 |
buZz | (in failure cases, ofono stop doesnt even 'kill' the 2G icon from statusbar) | 14:32 |
Wizzup | probably we first need to see if the modem even re-appears on usb | 14:33 |
Wizzup | and then we need to ensure things are good on the kernel level | 14:33 |
Wizzup | and then we need to ensure ofono picks it up | 14:33 |
Wizzup | and then we need to ensure userspace deals with it well | 14:33 |
buZz | i ment for a user | 14:33 |
Wizzup | some of these things might already work, but those are the steps I think | 14:33 |
peetah | uvos__: I'm ready to test your kernels, given that your explain how to do it | 14:36 |
uvos__ | peetah: uname -a, verfiy that kernel version is 6.0-rc8 | 14:59 |
uvos__ | peetah: edit /etc/apt/sources.list copy the final deb https://maedevu.maemo.org/leste beowulf main contrib non-free... line | 15:00 |
uvos__ | and edit it so it sais deb https://maedevu.maemo.org/leste beowulf-experimental main contrib... | 15:00 |
uvos__ | and apt update && apt upgrade | 15:01 |
uvos__ | reboot | 15:01 |
uvos__ | verfiy that uname -a is now 6.0.8 | 15:01 |
sicelo | not 6.1.8? | 15:06 |
peetah | my current kernel is 5.18.9 | 15:10 |
peetah | uvos__: ^ | 15:11 |
peetah | Linux devuan-n900 5.18.9 #1 SMP PREEMPT Sun Sep 11 16:52:18 UTC 2022 armv7l GNU/Linux | 15:12 |
uvos__ | peetah: yeah thats fine too | 15:15 |
uvos__ | sicelo: right 6.1 | 15:16 |
peetah | Linux devuan-n900 6.1.9 #1 SMP PREEMPT Wed Feb 1 13:59:50 UTC 2023 armv7l GNU/Linux | 15:25 |
Wizzup | SuperMarioSF: other one boots too :) | 15:27 |
peetah | uvos__: done ^ | 15:32 |
uvos__ | peetah: great thanks | 15:37 |
sicelo | meantime, 6.2-rc7 boots fine too :-) | 15:47 |
sicelo | fmg and sre really helped us figuring out that panel bug. since then we've been reliably booting into new kernels with less headache than before | 15:47 |
uvos__ | sicelo: sure this dosent free us from the agreed policy of testing every release that hits the -devel repo | 16:01 |
sicelo | i wasn't referring to Leste | 16:01 |
peetah | uvos__: you're welcome, that was an easy first one | 16:06 |
peetah | feel free to ask whenever you need to test something on the N900 | 16:06 |
peetah | Is there a way to follow battery discharge more closely, like bq2700.sh script was doing for maemo ? is this script safe to use with maemo-leste or did things evolve to far from mameo for this script to be still of any use ? | 17:49 |
bencoh | afaik I'd say using that script should be safe on n900/leste (since it just reads registers) | 17:51 |
bencoh | but droid4 relies on a different charger/fuelgauge | 17:51 |
bencoh | the droid4 battery management hw is part of "cpcap" | 17:51 |
bencoh | on droid4/leste users can rely on upower (upower -d iirc?) to monitor battery status | 17:53 |
bencoh | (although that wont give you raw readings from the hw) | 17:54 |
peetah | thanks bencoh | 18:14 |
sicelo | if you want to use the script in Leste, must first unload the bq27xxx_i2c module | 19:07 |
sicelo | peetah: also, what do you specifically need from that script which isn't already available in /sys../power_supply? asking because I've contributed a few patches to the kernel driver for bq27200 | 19:08 |
rafael2k | https://ftp.heanet.ie/mirrors/fosdem-video/2023/UD2.120%20(Chavanne)/linux_camera_apps.webm | 19:18 |
rafael2k | there is screenshot of Maemo running pinphone ^ | 19:18 |
rafael2k | camera app on Maemo | 19:18 |
Wizzup | timestamp? :D | 19:32 |
Wizzup | 19:30 | 19:33 |
Wizzup | rafael2k_: nice | 19:45 |
rafael2k_ | : ) | 19:52 |
peetah | sicelo: it's more about being used to have all the data available on one line in loop mode, than needing something that is not available elsewhere | 21:42 |
sicelo | ok. i guess `upower -d` is one way to expose them in one place :-) | 21:44 |
sicelo | but also `cat /sys/class/power_supply/bq27200-0/uevent` :-) | 21:44 |
sicelo | loop mode: watch -n 17 cat ... | 21:45 |
sicelo | but yeah, i agree the script is mode detailed. unloading the module will work just fine. see also https://github.com/maemo-leste/bugtracker/issues/170#issue-355242327 | 21:47 |
sicelo | s/mode/more/ | 21:47 |
peetah | but bq27200.sh work out of the box and is hassle free :) | 21:47 |
sicelo | actually even in fremantle, bq27200.sh didn't work ootb on power kernel ;-) | 21:48 |
Guest224 | have anyone asked officially from Nokia, how much cost code for N900 to get voice calls code and cpu power management code? | 23:40 |
Guest224 | back to lurking logs -> | 23:41 |
Wizzup | hi lurker ;-) | 23:53 |
Wizzup | not sure what exactly the cpu power management code would do for us, I think the kernel is all open, and most of the code is there, we just need to fix some of it up | 23:54 |
Wizzup | for the voice calls code, we have it mostly in the form of nemo pulseaudio modules and libcmtspeech | 23:54 |
Wizzup | I spoke with Pavel at FOSDEM and he had both directions working, but ran into latency issues, which we can try to tackle, and perhaps use the pa cmtspeech module as well, and we might get pretty far | 23:55 |
Wizzup | in this sense there is probably not much that nokia could give us, but there is a lot of other stuff they could open source that would help | 23:55 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!