2025-09-01 19:31:30 <clandmeter> riscv containers have moved to subnet 5.x
2025-09-02 12:20:50 <ncopa> i have been able to boot the startpro64 machine (the alpha-one). It has been bumpy and is a bit challenging
2025-09-02 12:22:21 <ncopa> I bought the alpha-one due to the metal case for passive cooling. And I have had very positive experience with the hifive premier p550. starpro64 has same SoC, eswin eic7700
2025-09-02 12:23:45 <ncopa> the pre-loaded rockos kernel boots and works fine. but it gets hot. very hot
2025-09-02 12:25:02 <ncopa> problem 1: I need to open the case to be able to connect to the debug UART. So it does not get any proper cooling when working with debug uart
2025-09-02 12:26:01 <ncopa> problem 2: the adafruit TTL USB to serial does not work properly. there is random garbage in the serial console. Switching to another USB to serial cable worked better. but occationally there gets random characters on the serial tty
2025-09-02 12:26:21 <ncopa> this happens already from the OpenSBI
2025-09-02 12:27:12 <ncopa> I have been able to build a kernel that boots, but I had to specify the full FDT path. only setting fdtdir was not enough
2025-09-02 12:29:28 <ncopa> problem 3: the mmc SD card was not working properly. adding sdhci.debug_quirks2=0x4 appears to solve that.
2025-09-02 12:30:21 <ncopa> problem 4: network appears to be very unreliable. it may or may not get dhcp lease. order of eth0 and eth1 is random.
2025-09-02 12:30:38 <ncopa> once it gets dhcp lease, it appears to have 50% packet loss
2025-09-02 12:31:35 <ncopa> no idea what is going on with that. but it also means that I can only work with this thing with via debug UART (which gives heating problems)
2025-09-02 12:31:53 <ncopa> problem 5: only 24G mem was detected instead of 32G
2025-09-02 13:36:53 <ncopa> clandmeter: what happened with the riscv64 builder(s)? looks like we got new hw?
2025-09-02 15:21:16 <ikke> ncopa: What do you mean? It should still be the milkv pioneers?
2025-09-02 15:22:55 <ncopa> i was logged in to a ncopa-edge-riscv64 container earlier today, building kernel
2025-09-02 15:23:06 <ncopa> and it dissapeared
2025-09-02 15:23:21 <ncopa> then i saw carlos message above: iscv containers have moved to subnet 5.x
2025-09-02 15:23:31 <ikke> ah yes
2025-09-02 15:23:43 <ncopa> so I found a machine 172.16.5.6
2025-09-02 15:23:45 <ikke> Those are the dev containers
2025-09-02 15:23:49 <ikke> builders are on native hw
2025-09-02 15:23:51 <ncopa> which has 128 cores
2025-09-02 15:24:27 <ikke> It's one of the new servers we received
2025-09-02 15:24:52 <ncopa> it appears to be fast indeed
2025-09-02 15:25:24 <ncopa> but its qemu
2025-09-03 08:01:22 <mps> I see 'Sophgo Mango' in this new machine model. that means thead C920
2025-09-03 13:39:50 <ncopa> alright. I know why the network driver for the starpro64 is broken
2025-09-03 13:40:14 <ncopa> and I have an approx idea how it can be fixed. But its not trivial
2025-09-05 08:48:44 <mps> how we set max open files on alpine
2025-09-05 08:49:09 <mps>  /proc/sys/fs/file-max don't help
2025-09-05 10:21:03 <ikke> mps: ulimit -n <new-limit>
2025-09-05 10:21:12 <ikke> but you cannot go further than the hardlimit
2025-09-05 10:26:46 <mps> ikke: thanks
2025-09-05 10:27:42 <mps> trying to find why X start with udev and not with libudev-zero
2025-09-05 12:33:12 <mps> solved it by adding line 'ulimit -n 8192' at top of /etc/init.d/slim
2025-09-10 10:04:05 <mps> hm, first time this: `rm /home/mps/xdg/doc` => rm: can't stat '/home/mps/xdg/doc': Permission denied
2025-09-10 10:04:22 <mps> doing this as root
2025-09-11 06:45:15 <mps> acpid disappered from busybox 1.37.0-r23 ?
2025-09-11 07:43:17 <mps> ah `apk fix busybox` reinstalls it
2025-09-20 11:16:17 <mps> algitbot: retry master