dgilmore changed the topic of #fedora-riscv to: Fedora on RISC-V https://fedoraproject.org/wiki/Architectures/RISC-V || Logs: https://libera.irclog.whitequark.org/fedora-riscv || Alt Arch discussions are welcome in #fedora-alt-arches
<davidlt[m]> the kernel compiled, but failed the packaging
<davidlt[m]> bpftool didn't work on the kernel
<davidlt[m]> seems this is new, as required kernel flag seems to be enabled, but maybe the tool doesn't support riscv64 or something is still missing in the kernel
<davidlt[m]> ok, new build is going :)
<davidlt[m]> probably for v5.14.17 I will do some backports
<davidlt[m]> the changes to the kernel also need to go into kernel-ark, not into kernel dist-git
jimwilson_ has quit [Remote host closed the connection]
cyberpear has quit [Quit: Connection closed for inactivity]
jimwilson has joined #fedora-riscv
bkircher has joined #fedora-riscv
jcajka has joined #fedora-riscv
bkircher has quit [Remote host closed the connection]
bkircher has joined #fedora-riscv
fede2cr has joined #fedora-riscv
<davidlt[m]> had another look into the spec, and I disabled a wrong line.
<davidlt[m]> fixed and restarted the build
<davidlt[m]> damn, AMD and their EPYC with 804MB cache per socket... that's insane!
<davidlt[m]> I also pinged Björn for BTF issue
fede2cr has quit [Ping timeout: 256 seconds]
bkircher has quit [Remote host closed the connection]
bkircher has joined #fedora-riscv
<rwmjones> that is a lot of cache
zsun has joined #fedora-riscv
<rwmjones> I remember getting a xeon with 20MB of L3 cache and thinking that was amazing
<davidlt[m]> Yeah, we are now almost at GiB for cache
<davidlt[m]> that's some expensive silicon
bkircher has quit [Ping timeout: 268 seconds]
bkircher has joined #fedora-riscv
cyberpear has joined #fedora-riscv
fede2cr has joined #fedora-riscv
<fede2cr> I'm testing Fedora no a nezhi board, but ssh kicks me out after a couple of minutes of running the board.
<davidlt[m]> See the logs?
<fede2cr> It seems to repond to ping well, it look like an ssh issue.
<davidlt[m]> I don't have idea why it would kick you out :)
<fede2cr> And for some reason I can't get the hdmi console active (I only have a vga monitor with an hdmi adapter)
<fede2cr> Let me remove the SD and check the logs.
<davidlt[m]> Is that working now?
<davidlt[m]> I don't think Fu Wei is connected right now
zsun has quit [Remote host closed the connection]
<fede2cr> What isn't working davidlt[m]? hdmi or hdmi over vga adapter?
<davidlt[m]> HDMI output
<fede2cr> Nothing useful on the logs. I'm trying with another SD and it keeps happening.
<fede2cr> The same board has worked fine with debian, but I need a more recent kernel for development.
<davidlt[m]> If you run ssh with -vvv what does it show?
<fede2cr> I'll run it with it and let you know when I get kicked out again. But the ssh terminal just freezes up, and keeps responding to ping.
<fede2cr> I already checked the entropy available and it seems fine.
<fede2cr> I'm also running dnf update under screen, but it kicks me out even before starting to upgrade the packages.
<fede2cr> Running ssh with -vvv, after a while I get
<fede2cr> client_loop: send disconnect: Broken pipe
<fede2cr> Running transaction test
<fede2cr> debug3: send packet: type 1
<fede2cr> client_loop: send disconnect: Broken pipe
<davidlt[m]> maybe try?
<davidlt[m]> -o TCPKeepAlive=yes -o ServerAliveCountMax=20 -o ServerAliveInterval=15
<davidlt[m]> or something similar?
<fede2cr> I got a segfault while doing dnf update.
<fede2cr> Even ping died on me this time.
<fede2cr> I'll start downloading an older image to see if it's something with the current kernel.
<fede2cr> The CPU temp is ok, it has a small heat sink on it.
<fede2cr> I'm now trying to upgrade the modules one by one, and I got:
<fede2cr>   Running scriptlet: python3-3.9.4-1.0.riscv64.fc33.riscv64 11/11
<fede2cr> debug3: send packet: type 1
<fede2cr> debug1: channel 0: free: client-session, nchannels 1
<fede2cr> debug3: channel 0: status: The following connections are open:
<fede2cr>   #0 client-session (t4 r0 i0/0 o0/0 e[write]/0 fd 4/5/6 sock -1 cc -1)
<fede2cr> debug3: fd 1 is not O_NONBLOCK
<fede2cr> Connection to 192.168.100.25 closed by remote host.
<fede2cr> Connection to 192.168.100.25 closed.
<fede2cr> Transferred: sent 5388, received 151308 bytes, in 460.5 seconds
<fede2cr> Bytes per second: sent 11.7, received 328.6
<fede2cr> debug1: Exit status -1
<BryceII> we hear you oh wise one!
<BryceII> erk wrong window focus 8/
bkircher has quit [Ping timeout: 250 seconds]
<fede2cr> I'm not testing the previous image on this Nezhi board, and I still get lockups. Now new ssh conections don't work, but the previous one is still up.
<fede2cr> But any command replies:
<fede2cr> [root@fedora-riscv ~]# dmesg
<fede2cr> Illegal instruction
<fede2cr> [root@fedora-riscv ~]#
<davidlt[m]> something is terrible wrong :/
<davidlt[m]> oh, the new kernel is almost done
<davidlt[m]> it finished building, now running various boring stuff
<fede2cr> Let me know if you need testing on a nezhi board. I'd be happy to.
fede2cr has quit [Ping timeout: 256 seconds]
fede2cr has joined #fedora-riscv
<davidlt[m]> Eh
<davidlt[m]> new kernel is still building, I guess, it will have to wait for tomorrow
sorear_ has joined #fedora-riscv
sorear has quit [*.net *.split]
djdelorie has quit [*.net *.split]
sorear_ is now known as sorear
djdelorie has joined #fedora-riscv
jcajka has quit [Quit: Leaving]
<djdelorie> do we have a current "getting started" guide for Fedora on unmatched?
<rwmjones> djdelorie: hey, fedora doesn't boot off unmatched
<rwmjones> have you assembled the hardware yet?
<djdelorie> yup
<djdelorie> I found a gpu and nvme for it too
<rwmjones> so I am running pssssht ubuntu kernel & bootloader + fedora userspace, assembled by hand
<rwmjones> what we need to do next is to work out the bootloader path for Fedora
<rwmjones> someone worked it out already, I'll find the link ...
<rwmjones> getting a Fedora image out with everything bootable on unmatched is job #1 at the moment imo
<rwmjones> davidlt[m]: ^^
<rwmjones> hopefully nirik will help here too
<rwmjones> btw which gpu are you using and how much did it cost? both mine are headless atmo because of the cost of getting a gpu plus I didn't really need one
<djdelorie> I'm guessing, ideally the microsd would boot the CPU, and hand off to the nvme or hd so those can look like "stock" fedora drives?
<djdelorie> I had an HD6870 from an old gaming rig, on the shelf. Same for power supply
<rwmjones> mine boots entirely off nvme I think (I would need to go to the office to check, but that was I think how it's set up)
<djdelorie> the unmatched manual implies that at least the first step has to be on microsd, but that just might be their default
<djdelorie> but a microsd "boostrap" image might simplify the installing-fedora step
<djdelorie> if it can make a riscv host look like an x86 pc, boot-wise
<rwmjones> I'm not 100% sure, would need to look at my boards in the office
<djdelorie> no rush ;-)
<rwmjones> going there tomorrow :-)
<rwmjones> I have time tomorrow afternoon (after 10.30am EST) when I can play with this and look at this together if you like
<djdelorie> I'll try to be in early but I usually fail at that (I work PST)
<rwmjones> oh no prob, later is fine too
<rwmjones> definitely don't get up early on my account :-)
<djdelorie> whew
<djdelorie> I have a work meeting 1-2pm EST but an otherwise free
fede2cr has quit [Ping timeout: 256 seconds]