tmlind | Wizzup: thanks for the report, i'll be posting some rfc 8250_omap fixes today and might have a fix for that one too now, still have not seen it here somehow | 05:40 |
---|---|---|
tmlind | seems similar to the one reported by freemangordon a few days ago | 05:41 |
tmlind | that driver clearly has other serious issues that have gone unnoticed.. | 05:41 |
freemangordon | Wizzup: ugh: | 08:31 |
freemangordon | Sep 27 22:28:50 localhost icd2 0.99[3483]: calling module 'libicd_network_wpasupplicant.so' link_up | 08:31 |
freemangordon | Sep 27 22:28:50 localhost icd2 0.99[3483]: libicd-network-wpasupplicant: wlan_bring_up: A1_23EC | 08:31 |
freemangordon | Sep 27 22:28:50 localhost icd2 0.99[3483]: libicd-network-wpasupplicant: Got network: (null), (null) | 08:31 |
freemangordon | Sep 27 22:28:50 localhost icd2 0.99[4388]: icd2 version 0.99 starting | 08:31 |
freemangordon | Wizzup: I think it is wrong to assume that network_id == iap_id without checking for ICD_NW_ATTR_IAPNAME here: https://github.com/maemo-leste/libicd-network-wpasupplicant/blob/master/src/wlan.c#L262 | 08:59 |
freemangordon | hmm, maybe there is another issue | 09:33 |
Wizzup | freemangordon: right | 11:08 |
Wizzup | tmlind: ok, ty | 11:08 |
freemangordon | tmlind: oh, those fixes really look as what we need | 14:32 |
freemangordon | will test as soon as I have some spare time | 14:32 |
tmlind | freemangordon: ok thanks, the first one hopefully fixes what you're seeing | 15:15 |
uvos | tmlind: Wizzup: freemangordon: applied and building on ci @uart patches | 15:16 |
tmlind | uvos: do you have it reproducable too? | 15:17 |
uvos | i can repoduce it | 15:17 |
uvos | sometimes | 15:17 |
uvos | its not reliable | 15:17 |
uvos | im not sure what makes the difference | 15:17 |
tmlind | hmm and are the messages appearing only in /var/log/messages on next reboot? | 15:17 |
freemangordon | I see it every time I issue sudo reboot/sudo poweroff from ssh session | 15:17 |
tmlind | or /var/log/syslog? | 15:17 |
uvos | also console | 15:17 |
tmlind | weird | 15:18 |
tmlind | no luck seeing it here with poweroff from ssh | 15:18 |
freemangordon | running leste? | 15:18 |
tmlind | nope, not seeing why that would make a difference though | 15:19 |
uvos | i clearly makes some kind of difference | 15:19 |
freemangordon | yes, it does | 15:19 |
uvos | the debian droid never sees it either | 15:19 |
tmlind | weird | 15:19 |
freemangordon | maybe timing issue | 15:19 |
tmlind | well let's hope the patch fixes it.. | 15:20 |
freemangordon | uvos: does it happen if you issue sudo reboot from ssh session? | 15:21 |
freemangordon | here it is 100% reproducible | 15:21 |
tmlind | freemangordon: and you see it on the serial console as it happens, or on next boot in logs? | 15:21 |
uvos | freemangordon: not allways | 15:21 |
freemangordon | I see it on ssh console | 15:21 |
freemangordon | I have no serial here | 15:22 |
freemangordon | serial console that is | 15:22 |
uvos | tmlind: enableing serial makes it go away | 15:22 |
tmlind | ah ok | 15:22 |
uvos | tmlind: so dose enableing more traceing features in kconfig | 15:22 |
tmlind | heh | 15:22 |
uvos | not sure exactly which but my full debug options kernel dosent show it | 15:22 |
tmlind | ok | 15:22 |
tmlind | so i've seen flakey reboot occasionally where the device and does not reboot, maybe that's related as i don't think i've seen that with the pending 8250 fixes | 15:24 |
uvos | thats pretty mutch this | 15:24 |
tmlind | anybody else seen that one? | 15:24 |
uvos | when it happens the getty hangs | 15:24 |
uvos | and shutdown hangs | 15:24 |
tmlind | ok, and reboot hangs too? | 15:24 |
uvos | yes same thing | 15:24 |
tmlind | ok fingers crossed then | 15:25 |
freemangordon | uvos: you can't do 5.18.19.2-1 IIUC | 16:23 |
freemangordon | what you should do is 5.18.19-2 and force-push 5.18.19 tag | 16:26 |
freemangordon | Wizzup: correct? ^^^ | 16:26 |
freemangordon | uvos: do you want me to release? | 16:29 |
uvos | freemangordon: i think it should work | 16:30 |
uvos | i just tagged it wrong sec | 16:30 |
freemangordon | ok | 16:31 |
uvos | yeah typo | 16:31 |
freemangordon | wait a bit | 16:31 |
freemangordon | I want to push battery low fix | 16:31 |
freemangordon | ok? | 16:31 |
buZz | omg yes please | 16:32 |
uvos | freemangordon: sure | 16:32 |
uvos | buZz: it wont solve your issue | 16:32 |
freemangordon | ok, will let you know when I am ready | 16:32 |
buZz | yesterday i pulled d4 from charger with 'fully charged' displayed, enabled wifi, started chromium, before page was loaded 'Battery empty at xxxxmV' | 16:32 |
buZz | and white led and powerdown \o | 16:32 |
freemangordon | buZz: yes, it will not fix your issue | 16:32 |
buZz | aw, sad | 16:32 |
uvos | buZz: that sounds like a bad battery | 16:32 |
freemangordon | your issue is bad battery | 16:32 |
buZz | ¯\_(ツ)_/¯ | 16:33 |
buZz | it works fine if i make it ignore the 'battery empty' | 16:33 |
freemangordon | maybe try to reset the calibration | 16:33 |
uvos | callibration has nothing to do with it | 16:34 |
buZz | calibration wont change resistance | 16:34 |
uvos | itbuZz: maybe, but your symptoms suggests the battery spikeing down to 3.35v while at 4 ish volts | 16:34 |
uvos | thats bad | 16:34 |
uvos | that means its spikeing down below 2.8V when your battery is lower | 16:34 |
uvos | the omap/dram start missbehaveing at that voltage | 16:35 |
buZz | ¯\_(ツ)_/¯ | 16:37 |
buZz | i'll fiddle around, maybe just a moving average over battery voltage | 16:37 |
freemangordon | uvos: pushed https://github.com/maemo-leste/droid4-linux/commit/d67cc2eba98cc3dca4739f9775303e46b1401870, please pull and tag as appropriate | 16:41 |
buZz | 'once every 30 seconds' might -totally- fix my issue :P | 16:44 |
uvos | no | 16:44 |
uvos | this isent whats powering off the device in your case | 16:44 |
buZz | its what is reporting the voltage | 16:44 |
uvos | upower polls voltage | 16:45 |
freemangordon | lets see | 16:45 |
buZz | then why is cpcap doing this at all? | 16:45 |
uvos | if its a whilte led during shutdown | 16:45 |
uvos | it was not the kernel | 16:45 |
buZz | if we arent using the interrupt, wouldnt it be better to not create it? | 16:45 |
uvos | its complicated | 16:45 |
freemangordon | uvos: actually it might help | 16:45 |
buZz | thats why i'm asking to clarify it ;) | 16:46 |
uvos | freemangordon: how so? | 16:46 |
freemangordon | as we may have the first 'low' irq at relatively ok voltage | 16:46 |
uvos | that dosent matter | 16:46 |
uvos | its not the kernel shutting down buzz device | 16:46 |
uvos | its mce | 16:46 |
uvos | and thats based on upower polling voltage_Now | 16:46 |
uvos | totaly unreladed to the machinery here | 16:46 |
freemangordon | yes, but it shits down on 'low' event and critical voltage | 16:46 |
buZz | where does the voltage come from? not cpcap? | 16:46 |
freemangordon | *shuts | 16:46 |
uvos | sure but if its spikeing below the kernel thresh of 3.1 or what was it v | 16:47 |
uvos | it will spike below mces of 3.35 even more | 16:47 |
buZz | 3.5v | 16:47 |
uvos | 3.35 | 16:47 |
buZz | which is quite high imho | 16:47 |
freemangordon | buZz: it is the same here | 16:47 |
freemangordon | it is simply that those batteries are amost dead | 16:48 |
uvos | https://github.com/maemo-leste/leste-config/blob/25160ce6596a88f790129e4419fcc280adec22aa/leste-config-droid4/etc/mce/mce.ini.d/70-droid4.ini.leste#L40 | 16:48 |
uvos | 3.35v | 16:48 |
buZz | 'cpcap_battery cpcap_battery.0: Battery low at 3461mV! | 16:48 |
uvos | cpcap fires the irq at 3.3v | 16:48 |
uvos | the kernel then mesures the voltage | 16:49 |
uvos | the 3.46 comes from that messurement | 16:49 |
buZz | but it -always- reports the voltage as 'low' ? | 16:49 |
freemangordon | it was triggered because a spike | 16:49 |
uvos | beacuse your battery is dead | 16:49 |
freemangordon | :nod: | 16:49 |
buZz | even without signalling with IRQ that its low? | 16:49 |
uvos | as in its internal resistance is really high | 16:49 |
buZz | whats the logic in that | 16:49 |
buZz | 3.46V isnt low | 16:49 |
freemangordon | buZz: wait | 16:49 |
freemangordon | there is a period between irq being triggered and voltage measured | 16:50 |
freemangordon | it is triggered @ < 3.3 | 16:50 |
buZz | ah, and it -always- outputs 'cpcap_battery cpcap_battery.0: Battery low at 3461mV!' -regardless- if its low or not? | 16:50 |
freemangordon | but this is a spike caused by a temporal high load | 16:50 |
freemangordon | it *was* | 16:50 |
uvos | battery low here means that the low battery irq fired | 16:50 |
buZz | ah :) | 16:50 |
buZz | uvos: maybe put that in the text then? :D | 16:51 |
uvos | complain to tmlind | 16:51 |
uvos | its his text :P | 16:51 |
freemangordon | heh :) | 16:51 |
buZz | he's the only person with a texteditor eh | 16:51 |
freemangordon | yeah, this is misleading | 16:51 |
freemangordon | buZz: feel free to send a patch | 16:51 |
freemangordon | like "batter low irq triggered, current voltage xxxx" | 16:52 |
freemangordon | or somesuch | 16:52 |
buZz | where does the IRQ originate? | 16:52 |
freemangordon | cpcap | 16:52 |
buZz | so 'triggering low battery IRQ, voltage i see is xxxx' ? | 16:52 |
freemangordon | no, it is triggering the irq | 16:52 |
freemangordon | and irq handler measures the voltage | 16:53 |
buZz | ahh | 16:53 |
freemangordon | but it takes time for the handler to be called | 16:53 |
buZz | its only triggering to -itself- | 16:53 |
buZz | i wonder if IRQ would even be the best method for that | 16:53 |
freemangordon | and by the time of the measurement, the voltage is already different to the one that caused the irq to be triggered | 16:53 |
freemangordon | I can't explain it any better than that | 16:53 |
buZz | i follow | 16:54 |
freemangordon | uvos: it is still missing the tag | 17:04 |
uvos | freemangordon: sigh ok | 17:05 |
uvos | the tag its looking for is def there | 17:09 |
freemangordon | it is not | 17:09 |
freemangordon | I just pulled and there is no such tag | 17:09 |
uvos | git clone --depth 1 --branch maemo-kernel-5.18.19.2 git@github.com:maemo-leste/droid4-linux.git | 17:09 |
uvos | works here | 17:09 |
freemangordon | you should do "git push --tags" | 17:09 |
freemangordon | not, it searches for tag, not branch | 17:10 |
uvos | i cant do that, to many tags | 17:10 |
uvos | git push maemo tag maemo-kernel-5.18.19.2 | 17:10 |
uvos | should have done the trick | 17:10 |
uvos | --branch takes tags here | 17:10 |
uvos | i know thats counter intuative | 17:10 |
freemangordon | no idea, lemme pull, to see if it has appeared | 17:11 |
freemangordon | ok, now it appeared | 17:11 |
uvos | i dident change anything | 17:11 |
uvos | or push | 17:11 |
uvos | so wierd | 17:11 |
uvos | ill just kick it again and see | 17:11 |
freemangordon | I have no idea too | 17:11 |
freemangordon | ok | 17:11 |
freemangordon | maybe github was too slow, dunno | 17:12 |
freemangordon | yay, it seems the issue is fixed :D | 17:19 |
freemangordon | and device reboots/powers down way faster | 17:22 |
tmlind | freemangordon: ok maybe reply to the list with a tested-by then? | 17:26 |
freemangordon | sure | 17:27 |
freemangordon | I want to have it running for a while | 17:27 |
tmlind | yup | 17:50 |
tmlind | thanks | 17:50 |
uvos | freemangordon: github must have been to slow then | 17:57 |
uvos | wierd | 17:57 |
freemangordon | yeah | 17:58 |
freemangordon | tmlind: how to check if serial idles correctly? | 18:27 |
freemangordon | as I think I see increased idle current | 18:27 |
uvos | omapconf | 18:30 |
freemangordon | I guess I have to build that,right? | 18:32 |
uvos | i think we put it into the repos | 18:32 |
uvos | not sure tho | 18:32 |
uvos | you can also read the register by hand ofc | 18:32 |
freemangordon | I'll just build the tool | 18:33 |
uvos | i dont know what it is of the top of my head | 18:33 |
uvos | if we dont have a omapconf package | 18:34 |
uvos | we should | 18:34 |
freemangordon | we don;t | 18:34 |
Wizzup | uvos: yeah I can make a pkg, but let's also have doc for it | 18:36 |
uvos | Wizzup: its fairly well documented by its help | 18:38 |
uvos | not sure what you want | 18:38 |
uvos | for details it quickly becomes "read the register manual" | 18:38 |
uvos | as its a tool that shows you registers | 18:38 |
freemangordon | ok, I compiled it, now what do I want from it? | 18:39 |
uvos | omapconf audio os_idle --log-all | 18:39 |
uvos | or so | 18:39 |
uvos | aproximate command | 18:39 |
uvos | check help | 18:39 |
freemangordon | I checked but there are too many options | 18:39 |
uvos | s/audio/audit | 18:39 |
uvos | my d4 is empty sorry | 18:39 |
uvos | its something very close to that | 18:40 |
uvos | also do sleep 20; omapconf ..... | 18:41 |
uvos | so that its idle when it fires | 18:41 |
Wizzup | uvos: for our purposes maybe | 18:42 |
uvos | its "omapconf audit os_idle full_log" | 18:44 |
freemangordon | yep | 18:46 |
freemangordon | I have full og but not sure how to read that | 18:46 |
freemangordon | all I can see is that ABE clocks are running but they should be stopped | 18:47 |
freemangordon | and I see abe_cm:clk:0030:0: failed to disable in dmesg | 18:49 |
freemangordon | anyway, have to run, ttyl | 18:50 |
tmlind | ok abe is mostly for audio etc, ttty | 18:50 |
tmlind | ttyl | 18:50 |
freemangordon | tmlind: do you want the log? | 18:50 |
freemangordon | tmlind: https://pastebin.com/65PkMSVp | 18:51 |
uvos | the abe issue | 19:05 |
uvos | we have allways had | 19:05 |
uvos | so thats not a regression | 19:05 |
uvos | otherwise log looks like your system is ideling ok | 19:05 |
tmlind | yeah it looks pretty good | 19:07 |
Wizzup | ret is also increasing for me | 19:29 |
freemangordon | 60 mA @ idle, is that ok? | 21:10 |
freemangordon | iirc it was 40 | 21:10 |
buZz | quite sure 60mA @ idle is lower than many modern phones | 21:12 |
buZz | i was seeing 50-60 with screen off, gprs connected, my modified mosh open | 21:12 |
freemangordon | POWER_SUPPLY_CURRENT_AVG=32458 | 21:19 |
freemangordon | oh, ok | 21:19 |
freemangordon | uvos: do we know the cause for that abe clock | 21:21 |
freemangordon | issue? | 21:21 |
buZz | 32! nice | 21:22 |
buZz | any idea how i should get rid of > Unknown media type in type 'font/ttf' | 21:24 |
buZz | shouldnt that mime type be defined by shared-mime-info ? | 21:24 |
* buZz tried dpkg-reconfigure shared-mime-info | 21:30 | |
buZz | tries* | 21:31 |
freemangordon | buZz: this is hildon thing | 21:38 |
freemangordon | buZz: https://github.com/maemo-leste/libhildonmime/blob/master/libhildonmime/update-category-database.c#L178 | 21:39 |
buZz | but still, font/ttf shouldnt be a weird mime-type for apt, should it? | 21:39 |
freemangordon | https://github.com/maemo-leste/libhildonmime/blob/master/libhildonmime/update-category-database.c#L41 | 21:39 |
freemangordon | I guess adding "font" there will fix it | 21:40 |
freemangordon | but still, this is just a warning anyways | 21:40 |
buZz | yeah i know, just annoying | 21:40 |
buZz | shall i PR 'font' into that? | 21:40 |
freemangordon | yep | 21:40 |
freemangordon | if it fixes the issue | 21:40 |
buZz | hmmm, is the src package of this in leste? then i could easily do it on my own system | 21:41 |
freemangordon | did you look ath the link ^^^ | 21:41 |
buZz | yes | 21:41 |
freemangordon | this is libhildinmome package | 21:41 |
freemangordon | and yes, this is in leste | 21:42 |
buZz | i'll check myself then :P | 21:42 |
buZz | i ment the source package ;) | 21:42 |
freemangordon | yes, it is | 21:42 |
buZz | not all leste packages (that we have source of) have source in leste | 21:42 |
freemangordon | how's that? | 21:42 |
freemangordon | wtym 'leste package' | 21:42 |
buZz | like the kernel for instance | 21:42 |
freemangordon | sure we have source | 21:43 |
buZz | we -have- it , but 'apt-get source linux-image-omap' doesnt grab it | 21:43 |
freemangordon | weird] | 21:43 |
freemangordon | anyway, ttyl | 21:43 |
buZz | \o | 21:44 |
buZz | libhildonmime -does- have a src package , nice | 21:46 |
freemangordon | buZz: just clone the git repo | 21:46 |
freemangordon | it is exactly the same source | 21:46 |
freemangordon | but you'll be able to do PR | 21:46 |
buZz | right, but installing it is annoying from source | 21:46 |
buZz | i'll just try editting this source, building a pkg for my phone, and see if it removes the warnings | 21:47 |
buZz | and -then- do a PR if it does ;) | 21:47 |
freemangordon | buZz: cloning from git is 100% identical to source package in terms of building | 21:49 |
freemangordon | but ok, up to you | 21:49 |
buZz | freemangordon: sure, but a -mess- to install imho , apt-get source, edit, build, install updated pkg, makes it just superconvenient :) | 21:50 |
uvos | freemangordon: no i never investigated it | 22:15 |
uvos | freemangordon: i have no idea | 22:15 |
uvos | d4 should get around 30 ish mA unter load | 22:16 |
uvos | or around 100mW | 22:16 |
uvos | *under idle ofc | 22:16 |
uvos | it its more, something regessed | 22:16 |
uvos | but its not visible in omapconf | 22:17 |
uvos | (could be more wakeups/s without idle being outright broken) | 22:17 |
buZz | sadly, that warning is not fixed by just adding "font" on https://github.com/maemo-leste/libhildonmime/blob/master/libhildonmime/update-category-database.c#L41 (tested with dpkg-reconfigure hildon-update-category-database) | 23:08 |
buZz | oh maybe i should do the mime one first | 23:08 |
buZz | oh, no that actually is the mime one | 23:08 |
buZz | oh, thats from a seperate package! lol | 23:09 |
freemangordon | so, does https://github.com/maemo-leste/libhildonmime/pull/1/commits/995726e24394f2211c7c03485666d7e5b8be7b46 fix it or not? | 23:23 |
uvos | freemangordon: tmlind: indeed the new kernel has higher power consumption | 23:28 |
uvos | looks like ~30mW | 23:28 |
freemangordon | uvos: I am getting POWER_SUPPLY_CURRENT_AVG=32458 | 23:29 |
freemangordon | which is near those 30mA you mentioned | 23:30 |
uvos | mW | 23:30 |
uvos | i have a cron job that logs power_avg every minute | 23:30 |
uvos | im geting around 30mW (thus around 10mA) more | 23:31 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!