qyliss changed the topic of #spectrum to: A compartmentalized operating system | https://spectrum-os.org/ | Logs: https://logs.spectrum-os.org/spectrum/
<aaronjanse> Yay!
nicoo has joined #spectrum
cole-h has quit [Ping timeout: 240 seconds]
TheJollyRoger has joined #spectrum
jpds has joined #spectrum
<hyperfekt> sweet!
cole-h has joined #spectrum
<qyliss> hmm, I wonder why I can boot from a squashfs with crosvm but not QEMU
<MichaelRaskin> Please don't tell Qemu recognises its header as something it should handle as non-raw disk image
<qyliss> I was passing format=raw, so don't think so
<qyliss> it seems like I can't actually boot from ext4 either so something is just up with my setup I think
<qyliss> it's annoyingly hard to find info about using QEMU when not using an initramfs
<samueldr> wouldn't it be the kernel's job to deal with the block device contents?
<MichaelRaskin> I wonder if adding an initramfs just to see if you try to mount from shell would help
<qyliss> there's no reason not to just use an initramfs and no root filesystem, tbh, except that now I'm curious what's up
<samueldr> you're using the scheme where qemu deals with starting the kernel, right?
<qyliss> samueldr: yes, and the same kernel works in crosvm
<samueldr> loglevel to the max, anything suspicious in the kernel logs "around the same time"?
<MichaelRaskin> I wonder if Qemu passes zero-size initramfs or something, Although you have non-initramfs-supporting kernel, I guess
<qyliss> but yeah I'll figure out an initramfs so I can try mounting from userspace I think
<samueldr> could crosvm send more cmdline params through another means?
<qyliss> samueldr: it does add some default cmdline, but I matched them with qemu
<samueldr> e.g. on DT platforms (like ARM) cmdline params can be added to through the DT
<samueldr> not sure if when this happens this ends up shown in the logs
<qyliss> I checked /proc/cmdline
<samueldr> and there's *other* options not commandline options
<samueldr> like on ARM there's `/chosen/stdout` that may be added by the firmware
<samueldr> which will be used as the console
<qyliss> fascinating
<samueldr> that tripped me up with my work on $PROJECT recently
<samueldr> since the serial console was working, and I realized I never configured it
<qyliss> crosvm adds console=ttyS0 to the cmdline by default
<samueldr> but if the firmware loads another FDT to be used, e.g. the kernel's expected DT, those /chosen nodes end up not being present
<samueldr> so that's one way boot (on ARM) can be "broken" easily
<samueldr> but for $PROJECT that'll end up being quite useful
<qyliss> the actual error is VFS: Cannot open root device "vda" or unknown-block(0,0): error -6
<qyliss> which is EXNIO No such device or address
<samueldr> Please append a correct "root=" boot option I figure close by?
<MichaelRaskin> Hm, 0,0 ?
<qyliss> samueldr: yeah
<samueldr> "here are the available partitions:" showing nothing?
<qyliss> OH NO
<qyliss> THIS AGAIN
<samueldr> this?
<qyliss> I keep getting this wrong and then forgetting about it
<qyliss> root=/dev/vda, not root=vda
<samueldr> correct "root=" boot option :)
<samueldr> sorry
<qyliss> I swear this happened to me a week ago or so, and it took me ages to spot then, but then I immediately forgot about it until just now
<qyliss> :D
<qyliss> I'm going to write this on my whiteboard
<samueldr> talking it out helps
<MichaelRaskin> Anki but desktop backgrounds
<samueldr> hah
<qyliss> rude
<samueldr> I guess this falls into "technically it's no colours"
<samueldr> anki, but the deck is entirely composed of one card stating that root= parameter requires /dev/
<multi> qyliss: ah, you discovered chanserv's mlock? :)
<qyliss> My whiteboard now says root=/dev/vda in big writing :D
<MichaelRaskin> Shoud the topic of #spectrum also say that?
pastbytes has joined #spectrum
<qyliss> lol
<lukegb> ...oops
ehmry has joined #spectrum