<craige[m]>
After working through the night without sleep samueldr , I believe I now have the phone firmly under control as well as all the resources to build it with everything working... when I get back to it.
<gchristensen>
:O
<samueldr>
oof
<samueldr>
that twrp post looks terribad for playing around
<samueldr>
(and porting)
<craige[m]>
Well, we just need to dodge Android 10 for now. Should be good.
<samueldr>
good luck on devices shipping with 10
<samueldr>
then there's the dynamic partition stuff
<samueldr>
the silver lining of this is that we may have a bit more freedom with partitions
<craige[m]>
Definitely an upside and some breathing room.
<samueldr>
I think I'll need to figure out a good source for documentation about that kind of stuff, and how to identify
FRidh has joined #nixos-aarch64
<craige[m]>
I've got a mountain of resources to share now. (currently on a bus to Vienna though)
<sphalerite>
samueldr: twrp post?
<sphalerite>
oh silly me, I see it
<samueldr>
hah
snicket has quit [Read error: Connection reset by peer]
snicket has joined #nixos-aarch64
makefu has quit [Quit: WeeChat 2.4]
makefu has joined #nixos-aarch64
snicket has quit [Ping timeout: 250 seconds]
snicket has joined #nixos-aarch64
snicket has quit [Read error: Connection reset by peer]
ryantrinkle has quit [Ping timeout: 240 seconds]
snicket has joined #nixos-aarch64
snicket has quit [Ping timeout: 240 seconds]
snicket has joined #nixos-aarch64
snicket has quit [Ping timeout: 240 seconds]
<sphalerite>
samueldr: do you think it's possible that the 4.9 kernel needs its dtb attached in a different way? I've seen some mentions of dt.img e.g. https://github.com/xiaolu/mkbootimg_tools
<sphalerite>
(the 3.18 kernel booted fine, the 4.9 one failed with "dtb not found"
<samueldr>
but I didn't have an intuition of stuff to try
<samueldr>
so to get you started more quickly, obviously, it was easier to use the older one
<sphalerite>
yeah
<sphalerite>
absolutely fair, I was just looking into what might help
<sphalerite>
I've opened a PR for my device, with some cleanups that you said I'd need to do for you to accept it :)
<sphalerite>
and the fbterm one
<craige>
heya clever (if you've found WiFi in Brno) did you resolve which package we needed ncurses.h from?
* craige
is back on the game after a lengthy sleep in a quiet corner of VIE :-)
Thra11 has quit [Ping timeout: 240 seconds]
<craige>
Nevermind clever - evntually found the correct doco.
Thra11 has joined #nixos-aarch64
<sphalerite>
samueldr: what was the org that ported modern kernels to devices again? (or something like that)
<craige>
Patches coming thick and fast from you sphalerite :-D
<sphalerite>
:)
<sphalerite>
you can talk ;)
<craige>
I have no cables with me at the airport to flash to my build target but I have just started a build of on my Hydra that is based on the kernel configs from Google. Fingers croosed that this builds and if it does...looking forward to flashing it when I'm eventually home.
<DigitalKiwi>
amazon is barely the 2nd most expensive
* DigitalKiwi
never heard of EMC
<gchristensen>
yeah but it is really good
<DigitalKiwi>
i don't have enough experience with it to say much other than their interface is extremely anti-user and pro pay someone to do this for you
<gchristensen>
it is definitely built for experts
<craige>
EMC have been around since the dawn of time. Early proteins used EMC magnetic media to back up their evolutionary plans.
<DigitalKiwi>
i was unaware evolution was planned
<DigitalKiwi>
gchristensen: can everyone with an arm machine setup theirs as remote builders ? >.>>
<DigitalKiwi>
crowd sourced build farm
<gchristensen>
no
<DigitalKiwi>
:(
<DigitalKiwi>
security issues or technology issues?
<gchristensen>
security
<DigitalKiwi>
technically possible though?
<gchristensen>
it *could* be done, yes
<DigitalKiwi>
how come there is the community arm server but not community x86-64
<gchristensen>
arm servers are hard to get :)
<DigitalKiwi>
well, now this might be shocking, for me at least and I am sure others, both of those are equally out of reach :P
<gchristensen>
nah
<gchristensen>
well, that may be true
<DigitalKiwi>
$$$
<gchristensen>
but an x86 server can be had for $2-$4/mo
<gchristensen>
an arm server is much harder to find at the low or high end
<DigitalKiwi>
i have $10 DO and it's not really all that good for building
t184256 has left #nixos-aarch64 ["Error from remote client"]
<craige>
I have a Linode 50 or some such that my Hydra runs on DigitalKiwi and it builds, one job at a time, quit nicely - faster than my laptop :-D
t184256 has joined #nixos-aarch64
<samueldr>
there _is_ a community x86_64 box
<samueldr>
since like a month ago
<gchristensen>
whaaat?
<DigitalKiwi>
and something with 64 or even like a fraction of that would be extremely cost prohibitive
<DigitalKiwi>
:0
<samueldr>
I think zimbatm and ad*sbladis forgot to advertise that fact
<DigitalKiwi>
64cpu*
<samueldr>
not sure what the requirements to get access are though
<DigitalKiwi>
that would have been so nice when i was trying to package ocrmypdf
<DigitalKiwi>
tests took 20 minutes
<DigitalKiwi>
20+
<samueldr>
I... don't know either where to go to ask, if anywhere
<DigitalKiwi>
and...i was working on it for months
<DigitalKiwi>
probably built it hundreds of times :|
<samueldr>
craige: haven't looked at your links, though do remember I was pretty sure the kernel was running, framebuffer not working was the likely issue... which I don't think "an official google kernel config" would fix, just in case
<gchristensen>
oh yay I think I managed to bring up some capcity finally
<samueldr>
a couple things to try to identify if your kernel does control the device is to set like a 2 minutes delay to reboot the device
<samueldr>
that way you time it
* gchristensen
shouldn't say anything until it is true
<samueldr>
it should be 2 minutes give or take some leeway from the bootloader
<samueldr>
meanwhile a failing kernel should reboot quicker
<samueldr>
and a hanging kernel... wouldn't reboot
<samueldr>
THOUGH, if you do try this, busybox's `reboot` doesn't work
WilliButz has joined #nixos-aarch64
WilliButz has quit [Changing host]
<samueldr>
you would need to use `hard-reboot`, which basically echoes the right value in /proc/sysrq
<DigitalKiwi>
gchristensen: i'm just wondering for the sake of wondering but like couldn't you do something like what the F@H etc. do and have multiple people build it and if they don't match then somethings wrong
<craige>
Will do that when I get a kernel building again and am in a place I can flash it, samueldr
<samueldr>
I believe something like that is being worked on, DigitalKiwi
<DigitalKiwi>
big if true
<samueldr>
won't steal the thunder of the project owner, not sure if the project is public
<DigitalKiwi>
now i'm more curious :|
<samueldr>
:) sorry
<DigitalKiwi>
can you message them and see if they can message me
<DigitalKiwi>
s/can/will/
<craige[m]>
just read back, I agree with your assertion samueldr - I can always revert back to that one but the diff between Google's and Lineage was fairly extensive and *lot's* of related drivers enabled (I generated it with `make menuconfig` against their source) - so I'd like to get it building and see how it flashes.
<craige[m]>
Interestingly(?) three of the 5 patches are not needed against this tree.
<samueldr>
interesting
<samueldr>
not saying that what you've been doing is wrong, just something to keep in mind
<samueldr>
hanging "Google" is good
<samueldr>
rebooting device (unless you wrote it) is bad
<craige>
Yeah, I just thought I'd try another tact while I was unable to flash to the device.
<samueldr>
then, the easiest, but 1-bit of data, is to get the device rebooting
<craige>
It will be interesting to see if that build when I land next.
<samueldr>
the easiest way to get knowledge about what's going on*
<craige>
:+1:
<craige>
Time for me to exit this continent and get on plane #1.
<craige>
o/
<samueldr>
safe trip!
ris has joined #nixos-aarch64
<samueldr>
then, if you can get twrp working in parallel with our kernel flashed, in theory you may be able to crash the kernel (through /proc/sysrq) and get a console ramoops
<{^_^}>
#72159 (by FRidh, 7 hours ago, open): Lack of aarch64 builders is blocking nixos-unstable channel
<gchristensen>
with graphs!
<samueldr>
but... with the pixel 2 I did not have much success with console ramoops... they wouldn't show up in twrp most of the time, but spuriously I got some
<FRidh>
I was just reading it, thank you gchristensen!
<gchristensen>
yep!
<FRidh>
can see already more running builds :)
<gchristensen>
great
<samueldr>
wow, gchristensen++, I wouldn't have the (brain) bandwidth to deal with that these last few days
<{^_^}>
gchristensen's karma got increased to 168
<gchristensen>
I think once the deploy is done, about 30 more jobs will be building
<DigitalKiwi>
<3 gchristensen
<{^_^}>
gchristensen's karma got increased to 169
<DigitalKiwi>
nice
<gchristensen>
samueldr: gotta take care of things or else FRidh might disable aarch:)
<FRidh>
samueldr: maybe we're actually talking to grahamcofborg...he keeps going like a machine ;)
<samueldr>
the real graham is currently on a Train or a Plane to somewhere cozy, gotcha
<DigitalKiwi>
yeah how are you here aren't you on a plane
* DigitalKiwi
has been informed airplanes have internet access now. wonders how they keep the cable from breaking
<gchristensen>
I'm in Paris :)
<DigitalKiwi>
how long are you there
<gchristensen>
until like November 8 or so
<DigitalKiwi>
working or vacation
<gchristensen>
work :)
<gchristensen>
but also very regular vacations to the cafe for a croissant
<DigitalKiwi>
i don't know how i am overweight, i get tons of exercise, i walk to the fridge every hour!
v0|d has joined #nixos-aarch64
orivej has joined #nixos-aarch64
<gchristensen>
ok back in a while ... time to find some food somewhere.