<laduke-13>
I was thinking about trying to boot this espressobin. Looking at the u-boot env, there's all this gak (from the vendor?) about booting from a tftp server and stuff like that... I don't know much about u-boot; I guess I need the env to be configured like "u-boot's generic distro configuration", but it looks nothing like it. even if i do env default -a. There must be some way to load u-boot
<laduke-13>
env vars from a file?
<samueldr>
what's the vendor's u-boot version?
<samueldr>
IIRC that was added late 2017 or early 2018
<laduke-13>
next step would be build u-boot or get it out of the image and flash it?
<samueldr>
building it, finding the fine manual for mainline u-boot, and adding it to the u-boot derivation of nixos
<samueldr>
the nix-build -A pkgsCross.aarch64-multiplatform.uBootEspressobin (if that's the name the attribute you add in all--packages is)
<laduke-13>
ok thanks for answering!
<samueldr>
I have no experience with the marvell family of hardware, so I can't help much further :)
<thefloweringash>
I think I'm hitting more purity problems, but just in case anyone's seen it before: my perl Net::DBus connection seems to fail with `Can't locate object method "_send_with_reply_and_block" via package "Net::DBus::Binding::C::Connection"`
mla has joined #nixos-aarch64
vika_nezrimaya has joined #nixos-aarch64
<clever>
samueldr: its not currently setup, but it is on the desk
<clever>
samueldr: oh, you mean cache, yeah it hasnt been building for a while
pbb has joined #nixos-aarch64
vika_nezrimaya has quit [Ping timeout: 240 seconds]
bennofs has quit [Remote host closed the connection]
bennofs has joined #nixos-aarch64
<thefloweringash>
which targets did the cache build?
<clever>
thefloweringash: its still configured to build some things for armv7, but the build machines arent online
<clever>
and qemu-user has issues
<thefloweringash>
I have my little arvm7 vm, I'm slightly tempted to add it to my hydra, but I'm not sure which targets are most useful to build. Maybe nixos/release-small.nix?
<clever>
thefloweringash: i think the biggest pain is just the stdenv, so even just building something basic like pkgs.hello would get that out of the way
<thefloweringash>
now that you've all seen behind the curtain of how hacky my hydra setup is, is there a nicer way to selectively cache things than <runcommand> hooks?
<gchristensen>
on an unhelpful note, I'm wondering if you can help samueldr and I get armv7 working in a VM on a Packet server so hydra.nixos.org can build it
<thefloweringash>
are there more challenges there than what I hacked up with the above?
<gchristensen>
it didn't boot iirc
<thefloweringash>
hmm
<thefloweringash>
can I get access to an equivalent machine to test with?
<gchristensen>
sure
<gchristensen>
but not until tomorrow after 5PM CET
rajivr___ has quit []
rajivr___ has joined #nixos-aarch64
ryantrinkle has quit [Ping timeout: 258 seconds]
ryantrinkle has joined #nixos-aarch64
<samueldr>
thefloweringash: some gicv3 issue with qemu iirc
<samueldr>
I may have found some info about that scrolling aimlessly at twitter lately
<gchristensen>
oh cool
tilpner has quit [Quit: tilpner]
<thefloweringash>
I remember `-machine gic-version=3` being critical but never looked into why
<samueldr>
the thing is on the unobtainium arm server it works fine without anything special
<samueldr>
(and on my raspberry pie when I tested)
<samueldr>
but not on that new obtainable server with 32 bit support
tilpner has joined #nixos-aarch64
<gchristensen>
I'll get y'all one tomorrow, 5:01PM CET :P
<thefloweringash>
That’s 22 hours from now? I’ll be sure to be around
<gchristensen>
(maybe not literally then, but if you're available to hack on it, I'll prioritize it highly)
<thefloweringash>
Probably only for a couple of hours, it’s 1am local time.
<thefloweringash>
What’s the target machine time in packet terms?
<samueldr>
Just a heads up, if you are using qemu-system-aarch64 with large numbers of SMP cores with KVM, add the "gic-version=3,kernel_irqchip=on" option to your machine configuration.
<samueldr>
This will allow you to scale beyond 8 cores as GICv2 (the default) does not support more.
<gchristensen>
nice
<samueldr>
and interesting to me is in the comment the note about using "nodefault"
<samueldr>
I don't know if our tests do, and if they don't, maybe they should, rather than rely on the default of qemu
jackdk has quit [Ping timeout: 250 seconds]
pkral has quit [Ping timeout: 260 seconds]
alunduil has quit [Ping timeout: 252 seconds]
cstrahan____ has quit [Ping timeout: 272 seconds]
c00w has quit [Ping timeout: 264 seconds]
feepo has quit [Ping timeout: 252 seconds]
davidtwco has quit [Ping timeout: 246 seconds]
lirzhv has quit [Ping timeout: 260 seconds]
jackdk has joined #nixos-aarch64
feepo has joined #nixos-aarch64
alunduil has joined #nixos-aarch64
pkral has joined #nixos-aarch64
cstrahan____ has joined #nixos-aarch64
davidtwco has joined #nixos-aarch64
lirzhv has joined #nixos-aarch64
jackdk has quit [Ping timeout: 260 seconds]
mDuff has joined #nixos-aarch64
mDuff has quit [Ping timeout: 272 seconds]
CrystalGamma[m] has joined #nixos-aarch64
jackdk has joined #nixos-aarch64
pkral has quit [Remote host closed the connection]
mDuff has joined #nixos-aarch64
mDuff has quit [Ping timeout: 268 seconds]
pbb has quit [Remote host closed the connection]
pbb has joined #nixos-aarch64
<mla>
is the pinebook pro bootable with any sd image produced on hydra currently?
<samueldr>
speaking of, you had troubles with android_usb on that new init?
<samueldr>
did adb/rndis work on current master?
<samueldr>
(I haven't had the chance to read more, simply saw your comment)
<danielrf[m]>
I don't think so--but I haven't tested in a while
zarel has joined #nixos-aarch64
<danielrf[m]>
tested against master I mean
<samueldr>
well, I said master, but when you contributed would have been good too
<samueldr>
at any point since you contributed to now
<danielrf[m]>
I remember it wasn't working when I originally made the contribution
<samueldr>
if adb or rndis never worked, then it's "good", it means that there's nothing wrong in what I ported
<samueldr>
but it's "bad" since it means we may have more quirks to deal with
<danielrf[m]>
for sure--nothing has been broken for me with the new stage-1 PR
<danielrf[m]>
at least, nothing became broken
<samueldr>
didn't android_usb break?
* samueldr
re-reads your comment
<danielrf[m]>
it wasn't working before or now
<samueldr>
ah, so when you don't enable it, the new stage-1 works, that's it?
<danielrf[m]>
yes
<samueldr>
great
<danielrf[m]>
sorry if my comments were confusing!
<samueldr>
no worries, it's that you said that it worked, but I clearly saw your comment saying it doesn't [when enabling usb gadget with android_usb]!
<samueldr>
I seem to remember that motorola-addison was confusing
<samueldr>
it had the gadgetfs stuff in the kernel, enabled, but it's still an android_usb using device
orivej has joined #nixos-aarch64
<samueldr>
oooof, if we need to do something before mounting the functionfs it needs a small rework of that bit of code
<danielrf[m]>
yeah I wasm
<samueldr>
that means we can't just dump it as a `boot.specialFilesystem`
<danielrf[m]>
yeah I wasn't sure how to do something before mounting functionfs....
<samueldr>
oh
<samueldr>
no, we have a solution maybe, but not something that great
<samueldr>
it relies too much on the order of execution to my taste, but that wouldn't be the first time
<samueldr>
ah, maybe a first thing to try, costs almost nothing
<samueldr>
disable adbd
<samueldr>
but make sure usb networking is enabled for rndis
<danielrf[m]>
ok i'll look into that
<samueldr>
if it works, and rndis works (do you know how to check?) that means only adb requires that, and it's likely easier to fixup
<danielrf[m]>
nope I've never used rndis
<samueldr>
make sure your stage-1 doesn't go to stage-2, and plug your device via usb to your computer, rndis will add a usb network interface to your computer
<samueldr>
the phone will serve an IP via dhcp
<samueldr>
there's bin/ssh-initrd as a helper to connect easier
<danielrf[m]>
that worked!
<samueldr>
right, so only adb
<danielrf[m]>
is there much benefit to having adb? given that we already have networking via usb and/or wifi?
<danielrf[m]>
not saying that we shouldn't try to fix it...
<samueldr>
there's the bragging rights :3
<danielrf[m]>
hehe true!
rajivr___ has joined #nixos-aarch64
<samueldr>
in actuality it allows fixing stuff when wifi or rndis doesn't work yet
<samueldr>
(once adb works with gadgetfs, that is)
<samueldr>
it may also have other uses on a running system
<samueldr>
like adbfs
<samueldr>
adb push/pull
<samueldr>
and mtp is part of the same subsystems
<samueldr>
I think using the phone as a sound card or webcam *could* be part of this too, but haven't looked closely yet