<Entitlement>
tmlind - [ [PATCH v2] power: supply: cpcap-battery: fix invalid usage of list cursor ]
inky has quit [Ping timeout: 252 seconds]
<uvos>
tmlind: sounds likey, ill test it later. For sure if(!d) is wrong as he correctly states a fully traversed list will point to head. but if the list is expected to possibly be empty list_enry_is_head is also wrong an we must guard before list_for_each_entry. Ill check if that can occure and then write to the ml
<uvos>
ok no it cant be empty
<tmlind>
yeah does not seem to help
<tmlind>
based on one boot
<uvos>
sad :(
<tmlind>
uvos: so did you already try not loading phy-cpcap-usb, cpcap-battery and cpcap-charger on boot, then manually loading them with charger connected?
<tmlind>
tried with those modules renamed to not load, still won't boot to console properly
<tmlind>
maybe mbm bootloader passes somehow a slightly different kernel cmdline when usb charger is connected that somehow causes trouble?
<uvos>
tmlind: yes i did
<uvos>
tmlind: it never hanged across several boots when doing that. however i never happens to me when the state of charge is high and soc was high at the time. i wanted to try again whith soc low but forgot
inky has joined #maemo-leste
<uvos>
*it never happens...
<tmlind>
ok, i'll try with watchdog built-in and nowayout enabled
<uvos>
tmlind: it just did it with cpcap-battery and cpcap-charger blacklisted
<uvos>
also phy-cpcap-usb
<uvos>
but i allways have that blackliste
<uvos>
d
<uvos>
also it really is soc related somehow i think
<uvos>
it happens 100% of the time when booting the devce with usb when it wont boot otherwise because the battery is to low for mbm to start
<uvos>
and more like 20% 80% when full
<uvos>
another thing i noticed is that really wierd shit can hapen like this
<uvos>
the hanged device will sometimes have random corruped files in /
<tmlind>
oh
<uvos>
but its not memory corruption
<uvos>
i did try stress and freinds by booting to init=/bin/bash and then executing them immidatly
<uvos>
and it dosent flip any bits
<uvos>
this is why i went on tangent about the nosiy ness of vbus
<uvos>
but its not realted i think, suppressing the noise with a filter dosent help any
<uvos>
another thing btw
<uvos>
that makes me think its not directly related to cpcap-charger or battery is
<uvos>
that in the hanged state
<uvos>
often the charge light will continue to operate if you plug usb in and out
<uvos>
suggesting that the cpcap modules are responsive
<uvos>
to a degree at least
<tmlind>
yeah it's like init is not properly running
<tmlind>
yup charge light works ok in that state
inky has quit [Ping timeout: 246 seconds]
<tmlind>
hmm wdt reset is not working so userspace is somewhat active.. but acpi power button works, so maybe i can trigger sysrq reset with that
<uvos>
tmlind:
<uvos>
i got this once in that state
<uvos>
uvos.xyz/maserati/bootlog.txt
<uvos>
userspace was freaked out something fierce
<uvos>
with bash somehow being passed dmseg as commands
<tmlind>
weird
<uvos>
this is also maybe where the "corruped files" originate
<uvos>
if you check that trace sometimes bash gets a command that it ends up doing with some redirect >
<tmlind>
i have not seen corruption with this issue afaik
<tmlind>
so i added echo c > /proc/sysrq-trigger to my /etc/acpi power button script
<tmlind>
heh just keeps blinking after pressing the power button.. will wait and see a bit
inky has joined #maemo-leste
<tmlind>
weird had to add mount -t proc - /proc to the script before echo c > /proc/sysrq-trigger, it's like /proc was not mounted
<tmlind>
i can now see the crash gets triggered properly, but device is not restarting after wdt reset..
<tmlind>
annoying
<siceloa>
besides facilitating hildon-desktop rotation, do we currently have another use for iio-proxy?
<uvos>
siceloa: mce uses it for everything
<uvos>
siceloa: als, proximitry, what have you
<uvos>
tmlind: yeah i tryed the sysrq path too and never got it work to get me a trace
<uvos>
siceloa: why?
jonsger has joined #maemo-leste
<tmlind>
uvos: i got a trace now, but it looks all normal based on a quick glance.. need to debug more later
<uvos>
tmlind: well thats something ;)
<tmlind>
so far the only weird things are no console shows up on the lcd, and /proc somehow is not mounted, pstore files show wlan gets associated etc, but no ssh for example
<tmlind>
no indication of sysrq doing anything in the files
<tmlind>
there is a line about ttyS2 buffer overruns with usb charger cable connected on boot, might be worth trying with uart3 set to status = "disabled"
<tmlind>
anyways, that's it for tonight's debugging, ttyl