<samueldr> trying with the dance and with the hs400 re-enabled
<samueldr> I don't think it's gonna happen, looking at the logs
orivej has quit [Ping timeout: 268 seconds]
wavirc22 has quit [Read error: Connection reset by peer]
wavirc22 has joined #nixos-aarch64
<thefloweringash> Xeon nas club member, checking in! hw.model: Intel(R) Xeon(R) CPU E3-1230 V2 @ 3.30GHz
<thefloweringash> It was the only reasonable way to get ecc ram.
h0m1 has quit [Ping timeout: 246 seconds]
h0m1 has joined #nixos-aarch64
<DigitalKiwi> an i3 wasn't enough for you?
<DigitalKiwi> or like...any AMD :)
<thefloweringash> IIUC intel's segmentation means that if you want ecc, you gotta pay for xeon
<thefloweringash> did amd exist in 2012?
<DigitalKiwi> ...
<DigitalKiwi> there are a few i3 that have ecc
<DigitalKiwi> and i have an AMD from 2004 with ecc
<thefloweringash> oh neat, when did that come to be? a quick google is inconclusive but suggests maybe 8th gen (ie, a lot newer)?
<DigitalKiwi> idk
<DigitalKiwi> a long time afaik
<thefloweringash> the more I google the weirder the answers are
<thefloweringash> the best evidence for support for an ie-2100 is a random forum thread saying "it totally works, I called a guy to confirm"
<thefloweringash> ie, no first hand evidence of a system actually running
<DigitalKiwi> AMD: Founded May 1, 1969; 50 years ago
<samueldr> I never have heard of a consumer-market intel chip with ecc support
<samueldr> but I never looked either
<thefloweringash> and then wikipedia says for some specific models in each generation there's ecc support in i3 cpus
<samueldr> in my case the xeon cpu was more about upgrading that machine while it was still my main machine in like 2015-2017
<samueldr> yes, 2009 era hardware at that time!
<samueldr> in 2017 I upgraded to 2012 era hardware with a used off-lease workstation that haad a xeon + ecc
<thefloweringash> I don't think AMD has had competitive performance at a time when I've been in the market for buying a computer. Though I hear good things now.
<samueldr> what you're saying is likely
<samueldr> I wonder in what year I'll be using this decade's hardware (though this decade is quite fresh still)
<DigitalKiwi> you'll find a lot of discussions on nas4free and freenas forums etc about the consumer intels that support ecc
<samueldr> and you can come clean: it's for the thrill of owning xeon hardware a bit, right?
<thefloweringash> I’m still quite fond of that aging hardware. Slightly smaller mATX case, a bunch of ram, remote management. Though that remote management is probably more of a security vulnerability than anything now.
<samueldr> just don't tell anyone
<DigitalKiwi> too late
<thefloweringash> Is 2020 the year of ~linux~ arm on the desktop?
<samueldr> hopefully!
<samueldr> I now have the laptop sorted out, finally
<thefloweringash> I would have bought ecc ram for my arm if I had found any
<DigitalKiwi> " All AMD desktop CPUs since the Newcastle Athlon 64, including all Semprons, and all APUs since Trinity have supported ECC as far as I know.
<DigitalKiwi> "
<DigitalKiwi> so like...since 2003
<DigitalKiwi> <3 AMD
<thefloweringash> Yeah. Intel aren’t nice, but they’re fast. Come on amd, we need you.
<DigitalKiwi> it's amazing they have any employees
orivej has joined #nixos-aarch64
andi- has quit [Ping timeout: 245 seconds]
orivej has quit [Ping timeout: 265 seconds]
andi- has joined #nixos-aarch64
wavirc22 has quit [Ping timeout: 240 seconds]
lopsided98 has quit [Quit: Disconnected]
lopsided98 has joined #nixos-aarch64
lopsided98 has quit [Quit: Disconnected]
lopsided98 has joined #nixos-aarch64
lopsided98 has quit [Client Quit]
lopsided98 has joined #nixos-aarch64
zarel_ has joined #nixos-aarch64
zarel has quit [Ping timeout: 268 seconds]
zarel has joined #nixos-aarch64
zarel_ has quit [Ping timeout: 265 seconds]
orivej has joined #nixos-aarch64
orivej has quit [Ping timeout: 260 seconds]
orivej has joined #nixos-aarch64
wavirc22 has joined #nixos-aarch64
wavirc22 has quit [Quit: ZNC 1.6.6+deb1ubuntu0.2 - http://znc.in]
wavirc22 has joined #nixos-aarch64
t184256 has left #nixos-aarch64 [#nixos-aarch64]
t184256 has joined #nixos-aarch64
wavirc22 has quit [Read error: Connection reset by peer]
wavirc22 has joined #nixos-aarch64
wavirc22 has quit [Read error: Connection reset by peer]
wavirc22 has joined #nixos-aarch64
orivej has quit [Ping timeout: 268 seconds]
t184256 has left #nixos-aarch64 [#nixos-aarch64]
t184256 has joined #nixos-aarch64
wavirc22_ has joined #nixos-aarch64
wavirc22 has quit [Read error: Connection reset by peer]
m15k has joined #nixos-aarch64
<m15k> Hej! Anyone is willing to share his configuration.nix for his/her raspberry pi 3?
ar has joined #nixos-aarch64
<evils> m15k: far from perfect, it doesn't boot unless hdmi is connected and i don't get a wm xD, but it works headless once booted... https://hastebin.com/kanawetaco.bash
<m15k> evils: are you currently working on getting it run with a wm?
<evils> not really, my main goal is replacing raspbian for my server pi, and even that's low priority xD (and the not booting thing has higher priority)
<m15k> Have you installed NixOS with the sd-image. I just wonder where the hardware-condiguration.nix is coming from.
<m15k> I just installed the sd image to my sd card. And don't have such a file.
<evils> i think this started with the sd image
<m15k> Thanks evils! Good reference for me to get into the topic.
<evils> well, not sure it's a good reference, but it may be better than none
<m15k> I guess it helps for for reflecting what I need to do :)
orivej has joined #nixos-aarch64
m15k has quit [Remote host closed the connection]
craige has joined #nixos-aarch64
<tilpner> ,stateVersion evils
<{^_^}> evils: Setting stateVersion to the latest release doesn't upgrade anything and can only break your setup at best. To actually upgrade NixOS see https://nixos.org/nixos/manual/#sec-upgrading. If you want to update the stateVersion option regardless, Ctrl-F for "stateVersion" in https://nixos.org/nixos/manual/release-notes.html to see things that need to be manually migrated with the new value.
<tilpner> Setting it like that only brings disadvantages, and you gain nothing
ryantrinkle has joined #nixos-aarch64
t184256 has left #nixos-aarch64 ["Disconnected: Replaced by new connection"]
t184256 has joined #nixos-aarch64
<evils> tilpner: i set that before i understood what that did, haven't changed it since, and am still unaware of the harm of setting it to unstable
<evils> though i would set it to a specific version if i were to deploy this
<tilpner> evils: This is how the value of stateVersion is used, and "nixos-unstable" will not work with versionOlder
<tilpner> > lib.versionOlder "nixos-unstable" "19.09"
<{^_^}> true
<tilpner> So this will result in using the behaviour of the oldest possible version
<tilpner> And you get postgres 9.5 instead of 11
<evils> huh, would have thought it'd get you the latest
<tilpner> It might also break in unexpected ways because the assumption "this system was installed when XX.YY was current NixOS" doesn't hold
<evils> wouldn't it make sense to just not accept anything but a valid version string for that then?
<tilpner> Actually no, you won't get 9.5
<qyliss> possibly, but there's only so much harm prevention you can do
<qyliss> regardless of what we've tried, people keep changing it
<qyliss> and if we limit what it can be changed to they'll just change it to other things that fit within those limitations
<qyliss> it's like we're giving people a big red button that says "DO NOT PRESS"
<tilpner> It will just not work and throw an error
<tilpner> So don't do that, evils
<evils> is 20.03 already a valid value?
<tilpner> See link above
<tilpner> It's valid, yes
<tilpner> 20.04 is valid too, as is 99.99
<tilpner> (And larger)
<evils> well, i saw it used there, but presumably there could still be breaking changes before 20.03 is actually released xD
<tilpner> Sure, but those are expected on unstable
<evils> huh, why would it break with "nixos-unstable" if versionOlder accepts that as an argument?
<tilpner> Because "nixos-unstable" is not a version number
<evils> versionAtLeast also accepts it as well, what would make it throw an error?
<tilpner> Nothing would
<tilpner> compareVersions just lexically compares strings, it appears
<tilpner> > builtins.compareVersions "a" "b"
<{^_^}> -1
<tilpner> > builtins.compareVersions "b" "b"
<{^_^}> 0
<tilpner> > builtins.compareVersions "c" "b"
<{^_^}> 1
<tilpner> But "nixos" and "unstable" don't compare meaningfully to numbers
<tilpner> > builtins.compareVersions "nixos" "1"
<{^_^}> -1
<tilpner> So the version "nixos" will be smaller (older) than every actual version
zupo has joined #nixos-aarch64
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
orivej has quit [Ping timeout: 268 seconds]
fadenb has quit [Remote host closed the connection]
fadenb has joined #nixos-aarch64
m15k has joined #nixos-aarch64
<m15k> Do I get it right: https://github.com/raspberrypi/linux/wiki/Upstreaming#downstream-drivers - Cpufreq for raspberry pi would be supported with kernel 5.3?
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
<m15k> Do I need `boot.kernelPackages = pkgs.linuxPackages_rpi3;` in my `configuration.nix` for RPI3?
<samueldr> if you want to use the raspberry pi foundation's kernel
<samueldr> otherwise no, it will use mainline
<m15k> samueldr: Is there anything I can read about the prod/cons?
<samueldr> I don't know of reading material about that
<samueldr> I think the main thing is that the whole "ecosystem" around the raspberry pi assumes the raspberry pi foundation's stuff is used
<samueldr> e.g. if you need to use device tree overlays to make some hardware work
<clever> you can also put a dto onto an i2c eeprom, and the firmware will auto-load it when found on the i2c bus
<clever> that lets the hat boards auto-configure things
<samueldr> clever: I don't think that'll help when u-boot then loads anoter device tree
<clever> yeah
<clever> u-boot breaks all the things :P
<samueldr> u-boot making things sane
<m15k> pkgs.linuxPackages_latest is referenced in the manual for pi 2/3. Is this mainline or some rpi adjustments? Oo
<samueldr> m15k: manual or nixos on arm wiki page?
<samueldr> (anyway it needs to be fixed wherever it is)
<samueldr> linuxPackages_latest is the mainline kernel, it should work just fine
<samueldr> at one point in the past the linuxPackages_rpi package only worked for the original raspberry pi
<samueldr> yeah, those instructions are from all the way back in 2017, I started re-doing this page last week to refresh it all and remove outdated or "weird" info
<m15k> But _latest is something different than the default value? (https://nixos.org/nixos/options.html#boot.kernelpackages)
<clever> m15k: _latest is just the latest version, likely ahead of what mainline recomends
<m15k> Like latest kernel version?
<clever> yeah
<samueldr> nah
<samueldr> it's latest released mainline linux
<samueldr> so not ahead of what mainline recommends
<samueldr> so yeah, _latest is generally better for ARM boards since many are wildly getting better :)
<m15k> Sorry if it is stupid but mainline kernel is 5.4?
<samueldr> it should be yeah
<samueldr> clever: _testing is probably what you had in mind, which is the RC for the next release
<clever> ah, probably
<m15k> Ahhh and without any configuration it woul be 4.19
<clever> Program terminated with signal SIGSEGV, Segmentation fault.
<clever> (gdb) bt
<clever> #0 0x76e25814 in greedy_realloc () from /nix/store/mx3lwmxapp47b9m4my81njwg0xn2175n-extra-utils/lib/libsystemd-shared-243.so
<clever> #1 0x76e4060c in read_line_full () from /nix/store/mx3lwmxapp47b9m4my81njwg0xn2175n-extra-utils/lib/libsystemd-shared-243.so
<clever> samueldr: you got any clues on how this could happen? heh
<samueldr> not at all
<clever> samueldr: it fails in basically the same way, with both cross-compiled and native-compiled
<samueldr> m15k: exactly
<samueldr> nixos stable releases with whatever LTS was released at that moment
<m15k> Okay. Got it!
<m15k> Thanks
<samueldr> 5.4 is the "next" LTS, which released after 19.09
<samueldr> 20.03 will ship with 5.4
<samueldr> (and once 5.5 releases, it will be the _latest, while 5.4 is kept as the default kernel)
<bennofs> clever: do you have more info on the segfault?
<clever> bennofs: `set -- foo bar` hangs ash sometimes (only at certain points in the stage-1-init.sh script)
<clever> bennofs: i can bypass that by just removing the root= from the cmdline
<clever> bennofs: then, `systemd-udev --daemon` hangs, despite the source saying it should immediately fork and return...
<clever> if i bypass that with `systemd-dev --daemon & sleep 10`, then `udevadm trigger --action=add` causes several segfaults in parallel
<clever> i do have an interactive shell, to debug things further, and now have the ability to make coredumps
<bennofs> the backtrace suggests some kind of heap corruption
<clever> #2 0x004e5900 in ?? ()
<clever> Backtrace stopped: previous frame identical to this frame (corrupt stack?)
<clever> bennofs: backtrace ends like that
<bennofs> and which executable is segfaulting? init?
orivej has joined #nixos-aarch64
<clever> pi@raspberrypi:~ $ gdb /nix/store/mx3lwmxapp47b9m4my81njwg0xn2175n-extra-utils/bin/udevadm core.udevadm.84
<clever> # echo '/mnt/core.%e.%p' > /proc/sys/kernel/core_pattern
<clever> bennofs: that line, lets the system write all cores to /mnt, so i can retreive them later on
zupo has joined #nixos-aarch64
<samueldr> clever: and that's on your custom firmware, right?
<clever> samueldr: yeah
<bennofs> no idea what that is. i guess you have to gdb some more :)
<clever> samueldr: but i dont see how anything the firmware is doing could impact it
<clever> [ 0.000000] OF: fdt: Command line is: print-fatal-signals=1 console=ttyAMA0,115200 earlyprintk loglevel=7 printk.devkmsg=on boot.trace boot.shell_on_fail memtest=17
<clever> i had linux run its own internal memtest, and it found zero problems
<clever> so all ram linux is touching, is functioning properly
<clever> and the firmware isnt doing anything to the system at the time of the fault
<clever> bennofs: gdb currently fails in the initrd itself, copy_bin_and_libs broke it, but i can try for a full closure...
<bennofs> clever: just the core would be enough I think
<bennofs> to try and find out where it fails inside greedy_realloc, to see if it is an error in the arguments that get passed to it or if the heap state is corrupt
<clever> bennofs: oh, and things dont fail if running under strace, not sure why yet
<bennofs> clever: you could try perf trace if you want a less-invasive syscall trace
ryantrinkle has quit [Remote host closed the connection]
ryantrinkle has joined #nixos-aarch64
<m15k> Can somebody tell me why sometime dmesg output is leaking into the login mask?
<samueldr> the login mask?
<m15k> terminal login prompt
<samueldr> I think it's simply because the kernel writes to the console whenever it feels it needs to
<samueldr> it just happens it's also the console you login into
<samueldr> it would also happen over "your session", if you know what I mean
<m15k> mh
<samueldr> adapting loglevel and other boot arguments can help
<m15k> I see.
lovesegfault has joined #nixos-aarch64
<m15k> samueldr: Do I get it right that you assume the problem is fixed? (https://github.com/NixOS/nixpkgs/issues/66960)
<{^_^}> #66960 (by samueldr, 22 weeks ago, open): Tracking issue for Raspberry Pi 3B+ and the latest kernel
<samueldr> I personally can't verify
<samueldr> I believe it should though
<m15k> I guess it's not. At least I'm facing the problem with the latest image.
<samueldr> which image exactly, and what is the issue?
<samueldr> I mean, state what you're observing on your device
<samueldr> oh, and which image was used?
<m15k> At least there are boots without the problem...
<m15k> Kernel: Linux rspbpi3b-hydrogen 5.4.13 #1-NixOS SMP Fri Jan 17 18:49:08 UTC 2020 aarch64 GNU/Linux
<samueldr> it could be another issue; if I remember right that issue was an all or nothing situation, and caused by using an older u-boot that didn't know about the B+/A+ raspberry pi 3, and a newer kernel
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<m15k> 19.09.1936
<samueldr> I didn't personally face the issue, but a couple users chimed in and there was no tracking issues for it
<samueldr> okay, so 99% sure if there is an issue it's not *that* one
<samueldr> first thing to note is that HDMI would never work, at all
<m15k> Basically it's not very consistent. I had several boots with problem. Now after disconnecting power I hadn't.
<samueldr> the pi 3 already was stretching its limits power-wise, not sure about the 3B+ if it's worse or better
<samueldr> but one of the issues with the pi 3 was issues with video output
<m15k> So you mean it's a hardware bug?
<samueldr> possibly something that's hard to actually fix, though unverified
<samueldr> *if* it's power issues
<samueldr> you could edit `config.txt` in the FIRMWARE partition, and remove or comment out avoid_warnings=1 https://github.com/NixOS/nixpkgs/blob/ea87d5fc8ade9e102032fd4dc04e9ae220c51415/nixos/modules/installer/cd-dvd/sd-image-aarch64.nix#L48-L50
<samueldr> that would make the pi print a lightning bolt if it's underpowered
<m15k> But if this would be the problem shouldn't I see such a warning?
<samueldr> we're actively removing the warnings in the config.txt since they step, or stepped, on the toes of the mainline linux graphics driver and caused bad behaviour
<samueldr> for the 3B (not +) it would wreck the display when it tried to print the warning
<m15k> Mh but that might explain why it's not happening all the time
orivej has quit [Ping timeout: 268 seconds]
<samueldr> that's my hunch, but I could be off base
zupo has joined #nixos-aarch64
zupo has quit [Ping timeout: 260 seconds]
orivej has joined #nixos-aarch64
m15k has quit [Ping timeout: 260 seconds]
orivej has quit [Ping timeout: 265 seconds]
craige has quit [Ping timeout: 265 seconds]
craige has joined #nixos-aarch64