<uvos>
yeah xorg got support for these via extension recently
<MartijnBraam[m]>
Wizzup: are you still unlocking the droids?
<uvos>
MartijnBraam[m]: he can unlock xt860 and sholes based devices (< droid 2 and friends)
<Wizzup>
MartijnBraam[m]: yes, I made that tweet, happy to provide any info you'd need
<uvos>
xt860 = droid 3
<Wizzup>
well, we all can, I can easily provide remote access to a VM with the dongle attached if you're looking to RE
<MartijnBraam[m]>
I guess it would be nice to have a t least usb dumps of the unlocking
<MartijnBraam[m]>
dongle?
<Wizzup>
MartijnBraam[m]: don't worry I have a lot of droid3's here locked still
<Wizzup>
MartijnBraam[m]: yes, the sigmakey sw requires you to pay for it, get a physical smartcard, and buy extension packs for features
<MartijnBraam[m]>
heh, and they don't support droid4? :D
<Wizzup>
well, the droid4 comes as unlocked as it can, with the prime exception being the US
<Wizzup>
but iirc (per tmlind) they filter the operator in the modem firmware, which is a whole different story
<Wizzup>
'as unlocked as it can' is perhaps an overstatement, I mostly meant that you don't need to network unlock it for global (sans US) usage
<Wizzup>
as far as I can tell sigmakey just needs a rooted device and then it uploads a bunch of payloads over adb
<Wizzup>
so if there is a tool to trace adb access/protocol we could dump that rather than dumping the usb protocol, but perhaps there's more going on
<uvos>
sure
<Wizzup>
I opened sigmakey in ida before and it's packed and not easy to RE (I spent a day or two trying to use it without paying)
<uvos>
you can do adb over network btw
<Wizzup>
uvos: that might be better than usbip I guess
<uvos>
if it only uses adb, yeah
<Wizzup>
right
<MartijnBraam[m]>
dumping it one layer higher is a lot nicer for RE yes
<MartijnBraam[m]>
as long as it doesn't do any magic
<MartijnBraam[m]>
over the usb part
<Wizzup>
as longh as that's everything, but it's something we could start with I guess
<Wizzup>
MartijnBraam[m]: so if you have any specific thoughts or ideas on how to make an open tool for it I am all ears, just let me know how I can help, I dreamed about doing it but don't have the time / resources
<Wizzup>
I could also do some usbredir hack with the sigmakey from my usb hub to some other remote computer to allow others to mess around with it
<MartijnBraam[m]>
I can take a look at figuring it out, I'm afraid it's just pulling a file, doing magic adjustments to it and writing it back, which would be very hard
<uvos>
dose sigmakey implement the adb protcoll itself
xmn has quit [Ping timeout: 260 seconds]
<uvos>
or dose it come with some version of the offical foss adb server
<Wizzup>
so the windows vm requires the motorola kernel drivers to be installed, but when I select a device I select it using adb in their gui, whatever that means
<uvos>
the moto drivers dont include adb iirc
<Wizzup>
MartijnBraam[m]: even if the binaries they upload are not foss, that's already a big step forward, I bet those binaries are not packed/obfuscated
<Wizzup>
but there is probably also some clever code in sigmakey itself to massage whatever data it gets back from the modem
<Wizzup>
it downloads some (iirc) security area from the modem and writes back changed values later
<Wizzup>
so we'd still have to trace that too
<uvos>
so if it uses the real adb
<uvos>
there is ADB_TRACE=1
<uvos>
envvar
<uvos>
that prints everything it dose via the bridge
<Wizzup>
it also has a 'root my device' mode, so that could be used to test this path perhaps
<Wizzup>
since that's more easily repeatable
<Wizzup>
I can set up usbredirect(1) with the dongle or provide some form of vm access
<Wizzup>
but my head is too full with other stuff we have to do for leste to dedicated a lot more of my brain/time
<MartijnBraam[m]>
yeah I think usb dumps are fine for now for at least figuring out what's happening, if that's easy to do
<tmlind>
oh looks like ifup wwan1 in 3g also works, but for some reason it takes a really long time.. i've seen that before now that i think of it
<tmlind>
ifup wwan1 just sits there at "Starting network with 'qmicli -d /dev/cdc-wdm1 --wds-start-network=..."
<tmlind>
then a few minutes later it continues while in gsm mode it happens right away
<tmlind>
Wizzup, uvos: you guys do rmmod phy-cpcap-usb on screen blank or when closing the slider? that's about 20 mW savings
<tmlind>
the uart mode in cpcap is a power hog
System_Error has quit [Remote host closed the connection]
System_Error has joined #maemo-leste
<uvos>
blacklist phy_cpcap_usb
<uvos>
yeah
<uvos>
tmlind: could you sudo omapconf audit os_idle full_log
<uvos>
just to prove its not some omap internal device
<uvos>
(unlikely but still)
<uvos>
also how are we mesureing power here
<uvos>
my numbers are as reported by cpcap loged every 5 minutes by crond
<uvos>
mesureing on the battery terminals with external equptent results in slightly lower numbers for me
<uvos>
maybe 10mW or so
<tmlind>
uvos: ok my numbers are from cpcap logged every 1 minutes
<bencoh>
206,1 ms/s 10,7 Interrupt [19] IPI means it's working 20% of the sampling time (?)
<bencoh>
the first line (delayed_work_timer_fn) is odd considering it reports 0 events/s
<uvos>
no idea but those numbers seam screwy in several ways
<bencoh>
well, that's not the only odd one, but I'd love to understand what it means
<uvos>
yeah
<tmlind>
yup i've not had much luck catching any pollers based on the powertop data..
<uvos>
i also have no idea how it comes up with the 52,7% CPU use number
<bencoh>
52,7% CPU use doesn't sound too idle to me, so ... yeah
<uvos>
top reports mutch less
<bencoh>
maybe it's 52% CPU when outside of idle
<uvos>
i gues this is of the time not spent in ret or something
<bencoh>
(get out of my head :D)
<uvos>
better type fast :P
<bencoh>
still, if the first column is right, it's a lot
<bencoh>
(first column of the first line)
<uvos>
might also be the time spent outside of ret
<uvos>
really ret confusing it its the only way this makes sense
<uvos>
ie its 206,1 ms/s of seconds running
<bencoh>
delayed_work_timer_fn comes from kernel/workqueue.c, so you should enable /sys/kernel/debug/tracing/events/workqueue and check resulting traces
<uvos>
and its 0,00 Events/s because the powertop sample time started at the tail end of an event
<bencoh>
enabling it might affect the result, but at least we might be able to know which driver keeps scheduling work
<uvos>
i dont have that file
<uvos>
(also no idea how the tracing interface works atm)
<uvos>
ah
<uvos>
nvm
<bencoh>
mount -t debugfs debugfs /sys/kernel/debug/
<freemangordon>
so somehow ix should be wrong there
<freemangordon>
wrong or zero
<uvos>
the difference is ABE
<uvos>
tmlinds ABE is gated
<uvos>
mine isent
<uvos>
let me kill pulse
<tmlind>
ok maybe that's it
<uvos>
maybe
<uvos>
but killing pulse dosent make it idle
<uvos>
ABE also dosent idle on Wizzup's device
<uvos>
hmm
<Wizzup>
hi, sorry, I was away
<Wizzup>
uvos: I'm planning to spend some time first on the n900 pm stuff and responding to the mailing list stuff (bandgap, panel), but can look at the droid4 stuff later this week or this weekend, let me know if you need anything from me
<Wizzup>
I am not sure what ABE does, but if it is related to audio, could it have something to do with hdmi audio not working perhaps?
<freemangordon>
uvos: hildon_pannable_area_motion_notify_cb doesn't get called at all
<freemangordon>
uvos: p, li { white-space: pre-wrap; } g_dbus_connection_call_sync
<freemangordon>
'If the parameters GVariant is floating, it is consumed. This allows convenient 'inline' use of g_variant_new(), e.g.: "
<freemangordon>
we get "(controlpanel:21262): GLib-CRITICAL **: 22:18:20.201: g_variant_unref: assertion 'value->ref_count > 0' failed" currently
<freemangordon>
just FYI, going to fix that
linmob has quit [Ping timeout: 252 seconds]
linmob has joined #maemo-leste
<Wizzup>
neat the hubs I have at home already are supported by uhubctl
mardy has quit [Quit: WeeChat 2.8]
Daanct12 has joined #maemo-leste
Danct12 has quit [Ping timeout: 268 seconds]
Danct12 has joined #maemo-leste
<Wizzup>
freemangordon: btw did you do some write up for xorg server or should I just try to figure it out
<Wizzup>
(wrt pp)
Daanct12 has quit [Ping timeout: 256 seconds]
<freemangordon>
Wizzup: better figure it out, it is obvious iirc
<Wizzup>
k
xmn has joined #maemo-leste
cockroach has quit [Quit: leaving]
<freemangordon>
hmm, dns resolution stopped working in leste VM
<freemangordon>
ping: abv.bg: Temporary failure in name resolution
<sicelo>
resolvconf weirdness - at least on my d4, i 'fixed' it with something like `rm -r /etc/resolvconf/run , rc-service stop resolvconf , rc-service start resolvconf`
<freemangordon>
rebooting VM, lets see
<sicelo>
i thought it was because my d4 install was partly broken
<freemangordon>
doesn't help :(
<freemangordon>
hmm:
<freemangordon>
Host abv.bg not found: 5(REFUSED)
<freemangordon>
maybe my host OS is behaving
<Wizzup>
freemangordon: not likely
<Wizzup>
(re: host 0s
<Wizzup>
(re: host os)
<Wizzup>
freemangordon: so we started to use resolvconf and it reads specific files for dns
<Wizzup>
or rather, dnsmasq reads specific resolvconf files
<Wizzup>
it's possible that the vm, which doesn't go through icd2, doesn't make any of these fiels
<freemangordon>
ah
<freemangordon>
bu, I have /etc/resolvconf/resolv.conf.d/original
<Wizzup>
I think maybe resolvconf hook never gets called for our /etc/network/interfaces eth0 setup in vm
<Wizzup>
this would make it work (one off): # cat /etc/resolvconf/resolv.conf.d/original | resolvconf -a eth0
<Wizzup>
so either we need to modify /etc/network/interfaces to call resolvconf, or figure out how to do it automatically, or finally add libicd-network-wired or something
<freemangordon>
ok, that made it work, more or less
<freemangordon>
it seems I've lost default gateway too
<freemangordon>
but, let me reboot to be sure
<freemangordon>
after reboot it s ok
<freemangordon>
how hard would be to add libicd-network-wired?
<Wizzup>
phone src
<Wizzup>
sry*
<Wizzup>
if it's gconf with values we could do it, but still, I'd rather not rely on icd2
<Wizzup>
like what if it crashes
<Wizzup>
then we cannot ssh in
<freemangordon>
maybe we shall ln /run/dnsmasq/resolv.conf to /etc/resolv.conf ?
<freemangordon>
though I have no idea how dnsmasq works
<freemangordon>
Wizzup: hmm, see /etc/default/dnsmasq
<Wizzup>
hm
mepy has quit [Ping timeout: 260 seconds]
Wikiwide_ has joined #maemo-leste
<Wizzup>
freemangordon: what about it, I don't see what you'd like me to check per se