dstzd has quit [Quit: ZNC - https://znc.in]
orivej has joined #nixos-aarch64
<Gaelan> Are the ids of the aarch64 NixOS AMIs documented anywhere? I can find them in Amazon's search but I'd like to make sure I'm using the official builds
dstzd has joined #nixos-aarch64
<samueldr> AFAIK there is no official aarch64 AMI, but I could be wrong
<angerman> I haven't seen any in years, always had to build my own ones.
<thefloweringash> there was a comment relating to that semi-recently: https://github.com/NixOS/nixpkgs/pull/104740#issuecomment-732820378
<angerman> Nice!
h0m1 has quit [Ping timeout: 258 seconds]
h0m1 has joined #nixos-aarch64
<Gaelan> can AWS graviton2 instances not run armv6l binaries? I'm trying to use one to build some stuff for armv6l, and it's failing with illegal instruction
orivej has quit [Ping timeout: 264 seconds]
rajivr has joined #nixos-aarch64
<samueldr> nope, it's stritcly AArch64
<samueldr> and trying to build armv6 on AArch64 AFAIK is problematic when it supports 32 bit
<samueldr> because of how it's armv8 32 bit
<samueldr> though building armv7l could work... but there's issues with "personality" in the kernel
<samueldr> but, you can run a KVM VM of armv7l to do so with not much loss
<samueldr> (when the CPU supports 32 bit)
<samueldr> not sure if the same could be done for armv6l
orivej has joined #nixos-aarch64
<Gaelan> darn, guess I'm back to building on my pi
<Gaelan> would be really nice to have cross-compiling to armv6l working again…
orivej has quit [Ping timeout: 240 seconds]
<noneucat> my rockpro64 has bit the dust :( was nice while it lasted
h0m1 has quit [Quit: WeeChat 3.0]
h0m1 has joined #nixos-aarch64
<sphalerite> noneucat: RIP
lgcl has joined #nixos-aarch64
<sphalerite> What happened?
<noneucat> not quite sure :( found it with the fan dead and now it won't boot past the first few steps of u-boot
<noneucat> wondering what to replace it with..
<sphalerite> m, maybe power supply issues? Can you power the fan directly?
<sphalerite> Because if so maybe you can "fix" this with an ATX power supply
<noneucat> i haven't considered the power supply now that i think about it
<noneucat> pci is not being powered
<noneucat> it did seem to be weird power wise
<sphalerite> yeah maybe the pins are exposed enough that you can just shove your own voltage in there to get it running again
<Ke> did you let it rest for sufficient amount of time
<Ke> some devices do recover
<Ke> though simpler things like minimal sbcs have lower footprint to store the bad energies
<Ke> like drain all the capacitors, let all the extra heat out
<noneucat> that is a good point also!
<noneucat> i will see what happens tomorrow
<Ke> if you have a multimeter, you can check the psu independently
<Ke> though multimeter does not show, whether the voltage drops on load
<noneucat> the psu seems to be holding it's voltage fine, but yeah it does not show how it does under load
<noneucat> the board seems to be powered fine... just not doing anything haha
<clever> sphalerite: the gcc shipped with raspi-os has some default flags, to force armv6, even on an armv8 cpu
<clever> samueldr: oops, ^
<clever> in theory, the pkgsCross in nixpkgs could do the same
<Ke> wait what?
<Ke> you mean to make everything rpi1 compatible?
<clever> Ke: a dedicated pkgsCross target that is rpi1 compatible
<clever> i believe pkgsCross.raspberrypi already is that
<Ke> ah, ure
<clever> the armv7l is named in a more generic way
<samueldr> clever: but this won't be "native" compilation like i686 on x86_64
<samueldr> which is what most people want when they want to build something on the wrong arch in backward compatible manners
<samueldr> if you're going to cross-compile, you might just as well do it from x86_64
lgcl has quit [Read error: Connection reset by peer]
ib07 has quit [Ping timeout: 272 seconds]
Darkmatter66 has joined #nixos-aarch64
nbp has quit [Ping timeout: 265 seconds]
nbp has joined #nixos-aarch64
<clever> samueldr: in the case of rpios, the "native" gcc forces armv6 by default, so it such a target could be made in nixpkgs, then import <nixpkgs> { system = "armv6l-linux"; } selects it...
<clever> if the aarch64 cpu is also 32bit capable, it should work
<clever> since v6 is a subset of v7
cole-h has quit [Ping timeout: 260 seconds]
justan0theruser has quit [Ping timeout: 268 seconds]
orivej has joined #nixos-aarch64
justan0theruser has joined #nixos-aarch64
Darkmatter66 has quit [Ping timeout: 240 seconds]
orivej has quit [Ping timeout: 240 seconds]
lgcl has joined #nixos-aarch64
orivej has joined #nixos-aarch64
justan0theruser has quit [Ping timeout: 240 seconds]
justanotheruser has joined #nixos-aarch64
alpernebbi has joined #nixos-aarch64
orivej has quit [Quit: orivej]
orivej has joined #nixos-aarch64
orivej_ has joined #nixos-aarch64
orivej has quit [Ping timeout: 246 seconds]
justanotheruser has quit [Read error: Connection reset by peer]
ryantrinkle has quit [Ping timeout: 240 seconds]
orivej_ has quit [Ping timeout: 264 seconds]
orivej has joined #nixos-aarch64
ryantrinkle has joined #nixos-aarch64
justanotheruser has joined #nixos-aarch64
bridge[evilred] has quit [Remote host closed the connection]
bridge[evilred] has joined #nixos-aarch64
orivej has quit [Ping timeout: 240 seconds]
justanotheruser has quit [Ping timeout: 240 seconds]
lgcl has quit [Read error: Connection reset by peer]
alpernebbi has quit [Quit: alpernebbi]
cole-h has joined #nixos-aarch64
rajivr has quit [Quit: Connection closed for inactivity]
Darkmatter66 has joined #nixos-aarch64
v0|d has joined #nixos-aarch64
danielrf has quit [Quit: danielrf]
<samueldr> sure, the compiler targets the right arch, but we're back at the problem of impurities during builds, with detection
veleiro has joined #nixos-aarch64
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nixos-aarch64
<veleiro> https://paste.rs/QIt does anyone know why libaom is not compiling on 32bit arm?
ajs124 has quit [Quit: killed]
ajs124 has joined #nixos-aarch64
monk has joined #nixos-aarch64
<andi-> samueldr: you once posted a dd statement to "transform" the u-boot output into an SPI compatible way. I searched the log but didn't really find a thing.. Mind posting that again?
<samueldr> it's in my pinebook pro repo
<samueldr> obv. not everything in there is needed
<andi-> thanks <3
<samueldr> except if you want to add a spiffy boot splash to your board
<andi-> ha, headless with boot splash.. that would be nice. Does it support truecolor terminals?
<samueldr> that's what I had in mind
<andi-> Does it support animations? A booting Nix flake..
<samueldr> nah, only having a centered image
<samueldr> but that's u-boot's bit
<samueldr> like your computer bios
<samueldr> the plan was at some point to figure out a good way with u-boot to do seamless boot splashes
<samueldr> like it's possible with intel i915 driver
<samueldr> and also a menu-driven interface to do everything you might want to do configuration-wise
<samueldr> because it's nicer that way
<samueldr> but the menu stuff in u-boot has... issues
<andi-> If I get to that point I am super happy already :D
v0|d has quit [Ping timeout: 256 seconds]
Darkmatter66 has quit [Ping timeout: 260 seconds]
ryantrinkle has quit [Ping timeout: 246 seconds]
orivej has joined #nixos-aarch64
justanotheruser has joined #nixos-aarch64
<flokli> samueldr: got some time for a bit of device tree overlay fun?
<samueldr> just ask
<samueldr> I don't know if I'll be able to help
<samueldr> but I'll try
<flokli> Error: /nix/store/fhv2n8ddgb10alz6jc7xapb9im7ws9zy-pine64-wifi.dts:34.31-32 syntax error
<flokli> FATAL ERROR: Unable to parse input tree
<flokli> but I seem to be getting somethign wrong with how to create that new wifi_pwrseq node
<flokli> the error output is pointing to the reset-gpios line, but I have no idea why
<flokli> the dts I'm extending is in <linux/arch/arm64/boot/dts/allwinner/sun50i-a64-pine64-plus.dts>
<flokli> I'm also reading https://source.android.com/devices/architecture/dto/optimize#root , wonder if that's my problem , but miss the part "as described below" is referring to
<samueldr> flokli: your patch please :)
<samueldr> is it a missing semicolon after a closing brace?
<flokli> which patch?
<flokli> there's a gist up there
<samueldr> the one you're applying
<samueldr> ah, missed it
<flokli> :-)
<samueldr> note that by default device trees from the kernel are built with the labels(?) stripped
<samueldr> I don't know if the NixOS module knows and works around that
<flokli> so the fun part is
<flokli> If I comment out line 15 and all the wifi_pwrseq stuff
<flokli> it compiles
<samueldr> does mainline have wifi_pwrseq defined?
<flokli> as soon as I add it, either as fragment@1, or further down (like now), it breaks
<flokli> not in that tree
<flokli> but sun50i-a64-pinebook.dts does in a very similar way
<samueldr> I don't think it would be ordering, but maybe it is?
<samueldr> >> reset-gpios = <&r_pio 0 2 GPIO_ACTIVE_LOW>; /* PL2 */
<samueldr> you're missing #include I think
<flokli> including what?
<samueldr> GPIO_ACTIVE_LOW would come from a .h
* flokli drops the glass on the floor
<flokli> #include <dt-bindings/gpio/gpio.h>
<samueldr> #include <dt-bindings/input/input.h>
<samueldr> uh
<samueldr> maybe yeah
<samueldr> transitively
<samueldr> I just picked a random .dts
<samueldr> dtc's errors are pretty bad :(
<flokli> no, doesn't work with any of the two includes
<flokli> does the dtc have the right "search path"?
<samueldr> no idea
<samueldr> no idea about how we're compiling them either
<samueldr> it might be that it needs CPP to be ran beforehand?
<samueldr> note that I'm not an expert, I've just been exposed enough to kind of see stuff
<flokli> hui
<flokli> so including isn't possible for some reason
<flokli> but I poked at the #define
<flokli> replaced it with its value
<flokli> boom
<flokli> it builds
<flokli> :-)
<flokli> Thanks a ton :-D
<samueldr> yeah, I was about to suggest to just place the value to test
<samueldr> n/p
<samueldr> at some point we'll have a collective understanding
<flokli> I'll try to see if we can teach our device tree machinery to find these include paths
<samueldr> it would be really helpful I think
<flokli> (and whether this crazy stuff works at all :-D )
<flokli> this were some very hard last hours. thanks for making me feel a bit saner again <3