ris has quit [Ping timeout: 245 seconds]
THFKA4 has joined #nixos-aarch64
THFKA4 has quit [Changing host]
ryantrinkle has quit [Ping timeout: 246 seconds]
chiefgoat has quit [Ping timeout: 272 seconds]
ryantrinkle has joined #nixos-aarch64
chiefgoat has joined #nixos-aarch64
orivej has joined #nixos-aarch64
orivej has quit [Ping timeout: 248 seconds]
t184256 has left #nixos-aarch64 ["Disconnected: Replaced by new connection"]
t184256 has joined #nixos-aarch64
sphalerite_ has joined #nixos-aarch64
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-aarch64
sphalerite has quit [Quit: WeeChat 2.4]
sphalerite_ is now known as sphalerite
orivej has joined #nixos-aarch64
orivej has quit [Ping timeout: 272 seconds]
orivej has joined #nixos-aarch64
orivej has quit [Ping timeout: 248 seconds]
orivej has joined #nixos-aarch64
ryantrinkle has quit [Ping timeout: 245 seconds]
jtojnar has joined #nixos-aarch64
orivej has quit [Ping timeout: 272 seconds]
orivej has joined #nixos-aarch64
ryantrinkle has joined #nixos-aarch64
orivej has quit [Ping timeout: 268 seconds]
orivej has joined #nixos-aarch64
wildtrees has joined #nixos-aarch64
wildtrees has quit [Max SendQ exceeded]
wildtrees has joined #nixos-aarch64
orivej has quit [Ping timeout: 268 seconds]
wildtrees has quit [Quit: Leaving]
Thra11 has joined #nixos-aarch64
<Thra11> After a not-quite-successful boot from an SD-card, where on the card can I find the journal or other useful logs, once I've mounted the SD card under a working system?
<samueldr> journalctl with -D I guess
<samueldr> likely under /var/log/journal/
<samueldr> so something like journalctl `-D /mnt/var/log/journal/ -b0` -b0 for the last boot
<samueldr> that is, if it booted far enough to log and the journal flushed
<samueldr> else, it's probably wise to hook to serial
<Thra11> var/log only contains two files, "lastlog" and "wtmp"
ris has joined #nixos-aarch64
<Thra11> both empty
<samueldr> I guess it hasn't booted far enough to log to the journal and flush
<samueldr> is it from sd_image?
<samueldr> if so, is the main partition resized to fill the sd_card?
<samueldr> if so, it means that it at least went far enough in stage-1 to do filesystems manipulation
<Thra11> Watching the uart and hdmi output, it *looks* like it's reaching the login prompt
<samueldr> oh
<Thra11> Although oddly enough, it's not doing the partition resize
<samueldr> any other storage on the device?
<samueldr> it looks for partitions by label
<samueldr> if you have a usb drive or an eMMC with the right partitions labels it might end up booting that instead of the usb drive
<samueldr> instead of the sd card*
<samueldr> (I don't know what's the target board here :))
<Thra11> Oddity #2: It couldn't find the root partition by LABEL=NIXOS_SD, but it can if I put root=/dev/mmcblk1p2
<Thra11> I'll try disconnecting the eMMC just in case it's confusing it.
<samueldr> werid
<samueldr> weirD*
<samueldr> ugh, typing
<Thra11> Board is a RockPro64. Trying to boot downstream 5.3 kernel
<samueldr> ah!
<Thra11> Is that the "ah!" of a person who has seen such weirdness and found all the solutions?
<samueldr> not sure if related
<samueldr> maybe not since you said "it looks like it booted to the login prompt"
<samueldr> I think it's more likely that the eMMC is confusing the initrd
<samueldr> (from the few bits of infos I have here)
<Thra11> w.r.t login prompt, it appears on the hdmi output, then flickers on and off for a while, eventually ending up with just a flashing "_". No login prompt appears on the uart, not sure if I need to tweak "console=<something>" on the kernel command line
orivej has joined #nixos-aarch64
<Thra11> "[FAILED] Failed to start Login Service. See 'systemctl status systemd-logind.service' for details." <- Well that sucks.
<samueldr> hmmm
<samueldr> you say the partition didn't resize, right?
<samueldr> is it... too full?
<Thra11> Looks like there are lots of other failures. I'll have a read through the uart output.
<samueldr> I *think* I've had this issue beforehand
<samueldr> though I thought I added enough slack space for this not to happen
koral has joined #nixos-aarch64
<koral> Hello
<koral> I am trying to install NixOS on a Raspberry Pi 3 B+ ; the nixos image for SD card boots fine, but as soon as I `nixos-rebuild switch` with a custom configuration.nix, HDMI stops working after reboot
<Thra11> samueldr: Presumably it's simple enough to test by building an image with *plenty* of space
<samueldr> koral: if you're using linuxPackages_latest, for now don't; there's an annoyance with how our configuration is made to use the dtbs
<samueldr> (unverified, though I suspect this is the case)
<samueldr> I don't have a B+/A+ pair to confirm
<samueldr> Thra11: or you could resize (with e.g. gparted) the partition offline in your case
<samueldr> Thra11: that would at least let you know whether it is that or not
<koral> samueldr: which one should I use then ? according to the wiki, linuxPackages_rpi is for RPi 1 only
<samueldr> don't configure it, it'll use the same default as it did for the initial build; which is the lastest LTS; linuxPackages
<samueldr> latest LTS*
<koral> oh, okay, let me try that
<samueldr> the issue is (from my flawed memory) the DTB filename changed from the mainline kernel, after the release of the A+, but our u-boot build doesn't know that
<koral> FYI, the HDMI issue I am having is exactly the one described in the wiki: the display hangs on "Starting kernel..."
<samueldr> it's the same symptom, but likely not related in that case
<samueldr> (if it's what I think it is)
<samueldr> the "Starting kernel..." message is the last thing u-boot printed while it was in control
<samueldr> so it could be for _any_ reason where the kernel couldn't start using the display
<samueldr> (making this hard to properly diagnose!)
<koral> I can actually guess that NixOS booted correctly because if I blindly log as root and blindly type "reboot", it does reboot
<samueldr> that's one of the diagnostic steps :)
<koral> I don't have a serial connection available unfortunately
<samueldr> if it isn't a one-off that you're going to do SBC stuff, it's worth its weight in gold to have the serial cable
<samueldr> can be useful for non-SBC stuff too :)
<koral> it works :D
<koral> thank you sir !
<samueldr> I'll add a note to the wiki page about that current issue
<samueldr> it might be really hard to get right
<koral> oh, didn't know I was actually exploring unknown territory :D
<samueldr> since you either lose compatibility pre-change in u-boot, or lose compatibility post-change in u-boot
<samueldr> not exactly unknown, another user had a similar (the same I think) issue the other day
Thra11 has quit [Ping timeout: 244 seconds]
<koral> while I am at it, do you need specific information about the issue that may help you fix it ?
<samueldr> I don't think so, I'm not sure
<samueldr> it might not be exactly what I described (can't find proof from the sources)
<samueldr> it *might* be a bit more subtle
Thra11 has joined #nixos-aarch64
Thra11 has quit [Ping timeout: 245 seconds]
koral has quit [Quit: WeeChat 2.5]
orivej has quit [Ping timeout: 268 seconds]
orivej has joined #nixos-aarch64
Thra11 has joined #nixos-aarch64
v0|d has quit [Ping timeout: 248 seconds]
v0|d has joined #nixos-aarch64
v0|d has quit [Ping timeout: 258 seconds]
<Thra11> samueldr: Full boot log if you're interested: https://gist.github.com/39c4a1bae7a9f14d3c98d38e7b94e660
<Thra11> As far as I can see, "devtmpfs: error mounting -2" is the first thing which is definitely not ok
<samueldr> >> /nix/store/vwvbzqf3jd9y8pyvbwpyyffj014wm6fi-nixos-system-nixos-19.09.git.50b20e1/init: line 9: /proc/cmdline: No such file or directory
<samueldr> this sounds bad too
<Thra11> samueldr: Yes. Looks like it hasn't managed to mount any of its special filesystems
<samueldr> aaaah
<samueldr> you're skipping the initrd with root and rootfstype
<samueldr> [ 45.340408] rootfs image is not initramfs (invalid magic at start of compressed archive); looks like an initrd
<samueldr> so there's not even a stage-1 going on first
<samueldr> try*
<samueldr> not entirely sure how you should do that
<samueldr> but that's a u-boot option
<Thra11> Ah. ok
<samueldr> so, AFAIUI you're basically skipping the initrd entirely
<samueldr> (which didn't load for some reason)
<samueldr> the reason might be that it's being overwritten by the kernel
<samueldr> at least, I'm reasonably confident
<Thra11> Yes. I guess I've manually told it to skip it by fiddling with the command line, because when I didn't it just failed because the initrd was being overwritten
<Thra11> *flashbacks to a very similar situation with an imx6 at work several years ago*
<samueldr> plus ça change, plus c'est pareille ;)
<Thra11> samueldr: I'll try setting ramdisk_addr_r tomorrow evening. Thanks for your help!
<samueldr> you're welcome
<DigitalKiwi> cana kit has raspberry pi 4 4gb
<DigitalKiwi> err i think they have other ones too that's just the one i was on
<samueldr> I have received mine it's been a good while
<samueldr> though for a while it wasn't available standalone, only in "kits"
<DigitalKiwi> oh
<DigitalKiwi> soo nixos on it soon eh? :P
<samueldr> some users already have it booted using the usual raspberry pi foundation boot chain
<samueldr> u-boot is supposedly working
<samueldr> though last I tried (when the patch set was shared) I didn't get it to work
<samueldr> have been busy
<samueldr> I was hoping that the changes required to tianocore would land soon enough
<samueldr> I would personally prefer nudging towards EFI
<clever> *looks*
<clever> -rw-r--r-- 1 clever users 3.5M Jul 29 00:09 OVMF_CODE.fd
<clever> dang, no hope of fitting tianocore into 512kb
<samueldr> nah, pipe dream
<clever> we would have to replace the spi flash on the rpi
<samueldr> but still, having tianocore on SD card for Pi3 is just the best
<samueldr> even then, it would only work if the chipset knows how to address more
<samueldr> knows it can address more
<clever> samueldr: i think its more about the firmware you load from the initial segment
<samueldr> which is not necessarily a given
<samueldr> possible
<samueldr> not too hip on all that :)
<clever> the boot rom will just load a "bootcode.bin" style loader from spi flash to L2 cache, and execute it
<samueldr> though it's possible that it cannot load more than the...
<samueldr> yeah, size of the cache
<clever> currently, thats it, end of story
<clever> bootcode.bin then loads start.elf from the SD and everything happens as normal
<samueldr> because that blob will need to init the ram
<samueldr> (for rpi4)
<clever> and bootcode.bin is responsible for bringing the ram up, on every model
* samueldr schemes
<samueldr> I should dust off that PR to make edk2 "projects" build
<samueldr> and ship the raspberry pi firmware as UEFI :3
<clever> in theory, a custom bootcode.bin could do more
<clever> and once it has control of the cpu, it can deal with a larger spi flash
<clever> the problem, is that its executing on the VPU
<clever> and its responsible for loading start.elf
<samueldr> (and since u-boot can also boot uefi, it shouldn't be an issue to do so)
<clever> i just had a thought
<samueldr> it happens
<clever> what if we had a custom bootcode.bin, that would bring up the arm cpu, and basically just idle the VPU
<samueldr> what if?
<clever> and then linux can shove over start.elf, and boot up the gpu firmware
<clever> make it behave more like a typical gpu
<clever> and if you dont want unfree software, dont give it a start.elf
<samueldr> that would be neat
<clever> i think this can already do half of that
<clever> the only problem, is getting start.elf to run, after linux has already started
<clever> because start.elf typically brings up the arm, loads the linux kernel, and resets the arm cpu to phys addr 0
<clever> would need the foundation to help with that
orivej has quit [Ping timeout: 245 seconds]
ryantrinkle has quit [Ping timeout: 245 seconds]