Wizzup | freemangordon: yes, please, sorry | 01:17 |
---|---|---|
buZz | i wish the calibration of battery was a bit simpler to hit :P | 11:46 |
buZz | i went 1100 > 900 > 1200 > 400 > now 750 | 11:47 |
buZz | lol | 11:47 |
buZz | no clue why it went from 1200 to 400 in a single calibration cycle, maybe i was burning a lot of cpu | 11:48 |
sicelo | bad battery :-p | 11:55 |
buZz | still, i'd love to be able to do 'calibrate now' and some interface tells me its status / if its done | 12:10 |
buZz | but no doubt these batteries are bad, sure :D | 12:11 |
Wizzup | maybe discharge it almost entirely and then go to charge mode? | 12:11 |
Wizzup | btw, my / filled up again with kernel oopses | 12:12 |
Wizzup | 6GB kern.log, 6GB syslog, 6GB messages | 12:12 |
buZz | gee | 12:13 |
buZz | maybe we should install logrotate by default? :) | 12:13 |
r3boot | ^_^ | 12:13 |
buZz | Wizzup: i keep trying to hit the 'charge now!' weep-weep-weep msg, but so often i just does that while i'm not near the phone before going full-off, which (afaik?) prevents calibration loop to run | 12:14 |
sicelo | huh? what's 'charge now'? | 12:14 |
freemangordon | missing, most-probably | 12:15 |
buZz | there's a notification between 10% and 5% battery charge, with 'almost empty' | 12:15 |
buZz | and a different one between 5 and 0% | 12:15 |
sicelo | ok, i get you | 12:15 |
buZz | afaik you need to hit the 5-0% one to start a calibration, while -not- having rebooted since phone knew about charge stats? | 12:16 |
freemangordon | buZz: not really | 12:17 |
freemangordon | unfortunately those are independent | 12:17 |
sicelo | i'm not sure about that ... the script doesn't care | 12:17 |
buZz | aw | 12:17 |
freemangordon | also, I think it works on the ooposite | 12:17 |
sicelo | https://github.com/maemo-leste/droid4-battery-calibration/blob/master/scripts/openrc/droid4-battery-calibration | 12:17 |
freemangordon | you have to charge full | 12:17 |
buZz | afaik uvos said the script no longer is functional, and calibration is now by kernel? | 12:17 |
freemangordon | and then wait for "battery low at xxx" kernel message | 12:18 |
freemangordon | afaik | 12:18 |
buZz | and thats the design_low_voltage now? | 12:18 |
freemangordon | but, because mce is too conservative, on bad battery it is really hard to hit that | 12:18 |
freemangordon | no | 12:18 |
buZz | ah | 12:18 |
freemangordon | it is still 3.35 (or somesuch) | 12:18 |
buZz | so calibration only runs between ~3.35 and 4.2? | 12:19 |
sicelo | btw who shuts down the device on low battery in general? mce? maybe for d4, it should first run that calibration script ... or it does? | 12:19 |
freemangordon | or 3.3, can't remember, look at driver sources, there is a explanation when 'high' low gets triggered | 12:19 |
freemangordon | sicelo: mce | 12:19 |
freemangordon | calibration it done by the driver | 12:20 |
freemangordon | scripts just saves what has been calibrated for the next boot | 12:20 |
freemangordon | IIUC | 12:20 |
sicelo | the driver just counts coulombs, yes | 12:21 |
freemangordon | well, not really | 12:21 |
freemangordon | it comes up with 'charge_full' when calibration is done | 12:22 |
sicelo | ok. | 12:24 |
Wizzup | Pali: I will hook up serial in the next hour and try u-boot | 12:26 |
sicelo | awesome :-) | 12:26 |
sicelo | in the meantime, tmlind has provided some nice patches that fix the musb problem for N900. | 12:27 |
buZz | freemangordon: so, charge_full to 'low at xxx' , would that be full calibration? | 12:31 |
buZz | or would it need to hit charge_full twice | 12:31 |
freemangordon | buZz: status 'Full' and then 'low at xxx' | 12:32 |
buZz | ok, so even if it poweroff after 'low at xxx' ? | 12:32 |
freemangordon | yes | 12:32 |
Wizzup | sicelo: great, what was the fix? | 12:33 |
freemangordon | that's the theory, at least | 12:33 |
buZz | nice, ok | 12:34 |
buZz | i'm at full now, and will run it till shutting down , to see if i can get it back to ~1000 | 12:34 |
buZz | :) | 12:34 |
sicelo | Wizzup: https://marc.info/?l=linux-omap&m=166876674609260&w=2 ... so it's a1a2b7125e10 that was causing my issues. anyway he has further patches to convert omap3 to work similar to omap4 regarding hwmods/ti-sysc, | 12:35 |
Pali | Wizzup: ok, thanks! | 12:36 |
Wizzup | sicelo: great | 12:42 |
uvos__ | buZz: one wrinkle about callibration on d4 is that after boot | 13:08 |
uvos__ | you still need to hit full OR empty even if you hit full AND before and have charge_full saved | 13:09 |
uvos__ | *full AND empty | 13:09 |
uvos__ | since the driver has no way of knowing what the coulomb counter should be at after boot | 13:10 |
uvos__ | probubly we should just improve soc estimation (ie do what android on d4 dose) and forget about callibration alltogether | 13:11 |
uvos__ | without usable nvram in the battery its not that viable | 13:11 |
buZz | uvos__: yeah thats the wrinkle that prevents battery percentage, i guess? | 13:11 |
uvos__ | buZz: the battery is totaly uncallibrated untill it starts showing the percentage | 13:12 |
buZz | right, not uncalibrated, but unknown | 13:12 |
uvos__ | its functionally uncallibrated | 13:12 |
buZz | it can be calibrated -and- unknown, is what you mean | 13:12 |
uvos__ | the state of charge display comes from the same place | 13:12 |
buZz | and only after hitting 0% or 100% will it be calibrated -and- known | 13:12 |
uvos__ | regardless of if charge_full exitsts or not | 13:12 |
buZz | hmhm | 13:13 |
uvos__ | if the counter state is not known | 13:13 |
uvos__ | ie calibrated -and- unknown == uncalibrated -and- unknown | 13:13 |
uvos__ | makes no difference to the accuracy of the state of charge display | 13:13 |
buZz | right, but it can become known only if calibrated before | 13:15 |
buZz | -or- if doing calibration and finishing it | 13:15 |
uvos__ | sure | 13:15 |
uvos__ | btw the reason the capacity jumps around is | 13:15 |
uvos__ | that the empty value is taken from an irq that is suseptable to noise | 13:16 |
uvos__ | and load | 13:16 |
buZz | sometimes i hit some mode where my d4 suddenly sees 'no battery' , i think that might have caused 1200 > 400 | 13:16 |
uvos__ | so if the device is heavly loaded or voltage noisy | 13:16 |
uvos__ | it will take the battery as empty for callibration purposes | 13:16 |
uvos__ | way to early | 13:17 |
buZz | the battery icon will go from normal green 'growing bar' on charging, to a red bar and a lot of notification sounds play very fast | 13:17 |
sicelo | uh ... no battery? you're sure your contacts are fine? | 13:17 |
buZz | sicelo: yeah, phone remains functional the whole time really | 13:17 |
uvos__ | there is no way d4 displays no battery hw wise | 13:17 |
buZz | unless kernel thinks its really empty , leading to mce to powerdown | 13:17 |
uvos__ | oh nvm thats only true on my device | 13:17 |
sicelo | oh, red bar means empty, not no battery :-) | 13:17 |
buZz | right, but even with charger connected, it shouldnt be showing red + not charging | 13:18 |
uvos__ | on stock d4s the kernel driver checks if the thermistor is present | 13:18 |
Wizzup | buZz: getting the sounds very fast sounds bad | 13:18 |
buZz | hmhm, there's cpu spikes around it aswell, but not so noticeable | 13:18 |
buZz | if i find some reliable way to trigger it i'll open a issue | 13:18 |
uvos__ | check kernel events | 13:19 |
uvos__ | when it happens | 13:19 |
sicelo | tmlind: i tried the 'Update musb to probe with ti-sysc for omap3' series, without applying 'usb: musb: omap2430: Fix probe regression for missing resources' - unfortunately musb breaks again for N900 ... missing interrupt numbers again | 19:16 |
sicelo | did i jump the gun here, i.e. 'usb: musb: omap2430: Fix probe regression for missing resources' remains necessary? | 19:17 |
sicelo | Wizzup: which kernel config is used when building d4/n900 kernel? | 20:27 |
sicelo | i think there are missing options | 20:30 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!