<makefu>
anybody managed to build the rockchip kernel from scratch? i am having a hard time here with a bazillion of compile errors. i am not even sure how anybody can actually build the kernel with the rockchip patches at all.... i am really unsure if i am missing something obvious. i've also tested out a lot if different branches/tags and repos . I am starting to doubt my own setup here.
<samueldr>
I'm in a different kind of hell, trying to figure out if it's too much investment fixing the `fastboot boot` support in u-boot to support the newer android boot images
<samueldr>
though I gave ti say, while it looks like there is some legacy in the codebase, it's generally easy to figure out what's doing what in u-boot
<samueldr>
I have to say*
<lopsided98>
makefu: I haven't tried that particular kernel, but I could never get the Rock64 4.4 kernel to build
<lopsided98>
I think they are both based on the rockchip BSP
<makefu>
sheesh that is not very promising
<makefu>
:D
<lopsided98>
I gave up on it when I found a good mainline based kernel
<lopsided98>
maybe try disabling autoModules
<lopsided98>
Nix automatically enables a lot of options that are not in the defconfig, and this can sometimes break vendor kernels
<makefu>
okay, will try that
<makefu>
(that is some unexpected behavior i must say)
<lopsided98>
yeah, it has bitten me a couple of times
<samueldr>
100% can confirm, ODM kernels from android devices are often wildly broken outside their defconfig
<makefu>
lopsided98: well the build now takes longer than before, i am crossing my finger that all what was needed was ( autoModule = false )!
<makefu>
lopsided98: wooo it seems i managed to actually build the rockhip stable kernel for rk3399, autoModule = false was the trick!
<lopsided98>
great!, glad I could help
petersjt014 has joined #nixos-aarch64
<samueldr>
hmm, the hacked-together code doesn't boot while the same kernel+fdt when loaded from disk and using `booti` works :/
<samueldr>
two things to explore, either the initramfs *somehow* breaks the bootm boot, as it was detected as invalid when used separately with booti, or bootm has other semantics conflicting
<samueldr>
and on the hdmi output, the mobile nixos splash
<samueldr>
there are things I need to understand better to know what exactly was at fault, but it looks like the location the kernel is loaded is more important than I thought
<samueldr>
which makes sense, if I think about it a couple seconds
<samueldr>
I'm curious as to how LOADADDRESS(?) works and how it can be changed, if even possible
<samueldr>
hmm, I'll need to read about all that when I'm less tired
samrose has quit [Ping timeout: 245 seconds]
<makefu>
samueldr: wow you really are in a different layer of hell :D
orivej has joined #nixos-aarch64
alex_giusi_tiri has joined #nixos-aarch64
samrose has joined #nixos-aarch64
<samueldr>
I'm excited for the next step: making an sd image look like a partitioned android device :) in theory fastboot from u-boot should be able to flash directly to one, and assuming it all boots, I should have a safe platform to make progress with mobile-nixos
samrose_ has joined #nixos-aarch64
samrose has quit [Ping timeout: 240 seconds]
samrose has joined #nixos-aarch64
samrose_ has quit [Ping timeout: 244 seconds]
<samrose>
so, it looks like nixos-install may actually be working on my bananapi-m64 and I assume I will need to copy over u-boot to actually see it boot from the emmc as well
<samueldr>
most likely
<samrose>
my current "build" seems to be installing everything under the sun, but I am going to let it roll, and then will try the u-boot. I will start a wiki page on this too today
<samrose>
oh no I see a "CC: blah/lib.o" and it's building the old 4.14.89 kernel so I'll see if I can figure out why
<samrose>
I guess it may be better for me to do the
<samueldr>
configuration.nix refers to the right kernel? channel/nixpkgs checkout/overlays are right?
<samrose>
yes. i guess I will need to be specifying the right kernel in configuration.nix
<samrose>
if the kernel matches what is on my current sd card, I take it that nixos will just use that vs a total recompile?
<samueldr>
not 100% confident, but IIRC it uses the installation media's store first
<samrose>
it looked like it wa doing that in the case of some packages
<samrose>
this time trying with boot.kernelPackages = pkgs.linuxPackages_latest; and hopefully it will use what I had specified in testing
<samrose>
I had to speicfy nixpkgs.config.allowUnfree = true; in my configuration.nix here too, I will try to figure that out sometime
<samrose>
ho! it's installing rhe right kernel and just copying over what I need this time :-D
<samrose>
no kernel compilingthis time
<samrose>
damn, she died on downlaod of thi sjdk package
<samrose>
I wonder if there is a way in configuration.nix to tell nixos not to try to install this jdk package?
<samrose>
hmmm. it may be nestled into something I am trying to insall in my expressions
Thra11 has joined #nixos-aarch64
samrose has quit [Ping timeout: 240 seconds]
samrose_ has joined #nixos-aarch64
orivej has quit [Ping timeout: 244 seconds]
samrose_ has quit [Ping timeout: 244 seconds]
samrose_ has joined #nixos-aarch64
samrose_ has quit [Ping timeout: 250 seconds]
orivej has joined #nixos-aarch64
samrose_ has joined #nixos-aarch64
petersjt014 has quit [Remote host closed the connection]
<samrose_>
I think my jdk issue stemmed from my workign from a branch that merged in upstream instead of staying pegged at release-18.09
<samrose_>
weird, when I try to use udisks2, nixos is trying to install oraclejdk, when I eliminate it, the problem disappears.