2026-03-02 13:10:18 <raspbeguy> hello, I'm struggling with an error with a go build on loongarch64 with messages like "$WORK/b189/_pkg_.a(_x001.o): unknown relocation type 77; compiled without -fpic?" https://gitlab.alpinelinux.org/raspbeguy/aports/-/jobs/2241883
2026-03-03 01:29:17 <huajingyun> Hi  raspbeguy, thank you for your feedback, will look into it shortly
2026-03-03 07:34:53 <huajingyun> raspbeguy,I added comment in !98334, thanks for your feedback!
2026-03-05 11:58:58 <huajingyun> Does anyone have time to help merge !97222 ? 
2026-03-07 18:04:08 <ikke> https://build.alpinelinux.org/buildlogs/build-edge-loongarch64/community/grafana/grafana-12.2.1-r4.log: relocation truncated to fit: R_LARCH_B26 against symbol `malloc' defined in .text section
2026-03-07 18:04:12 <ikke> Someone able to check that?
2026-03-07 21:26:19 <aelin> ikke: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/98690 fixes it 
2026-03-07 21:27:03 <ikke> aelin: aha, thanks
2026-03-12 17:15:43 <jvvv> I'm having trouble with py3-lmdb build segfaulting in check() on the loongarch64 builder. It passed in CI and now has failed/segfaulted several times on the builder. https://build.alpinelinux.org/buildlogs/build-edge-loongarch64/community/py3-lmdb/py3-lmdb-1.7.5-r0.log
2026-03-12 17:17:15 <jvvv> Does anyone have any ideas? I don't have any hardware to do any local troubleshooting on.
2026-03-12 18:17:16 <aelin> jvvv: heya if you (or other Alpine devs) want hardware to troubleshoot on, send me a private message and I can give you SSH access to a 3B6000
2026-03-12 18:20:31 <ikke> We do have dev containers
2026-03-12 18:29:58 <jvvv> I'm working on setting up a qemu vm, as soon as I figure out how to boot it. Think I need to setup some efi images (code and vars).
2026-03-12 18:39:35 <aelin> there is edk2-loongarch64, that contains an EDK2 build
2026-03-12 18:40:22 <aelin> just run QEMU with -drive if=pflash,format=raw,readonly=on,file=/usr/share/edk2/loongarch64/QEMU_EFI.fd
2026-03-12 19:09:33 <jvvv> it may not have been efi... my qemu-system-loongarch64 version didn't match my qemu-ui-* packages. but my kernel just got updated so I have to reboot into the new kernel so my kernel modules aren't mismatched.
2026-03-12 19:10:11 <ikke> jvvv: I can give you a container
2026-03-12 19:10:31 <ikke> if you want
2026-03-12 19:10:35 <jvvv> ok, that sounds cool. brb, just rebooting.
2026-03-12 19:14:16 <ikke> Hmm, maybe not..
2026-03-12 19:14:53 <ikke> at least, it's not working yet
2026-03-12 19:17:38 <jvvv> No hurry. Having an impetus to setup another arch qemu config is worth while. If you can at some point, that would be cool, but don't put yourself out in any kind of hurry.
2026-03-12 19:19:12 <ikke> "illegal instruction" ouch
2026-03-12 19:20:42 <ikke> huajingyun: seems like one of our dev servers is now broken against edge
2026-03-12 19:22:58 <ikke> jvvv: so probably not going to be today
2026-03-12 19:25:11 <jvvv> ikke: No worries. I have a loongarch64 qemu vm rig setup now, so I will play with that. Thank you very much for trying!
2026-03-12 19:25:58 <jvvv> My qemu issue was indeed just a mismatch in the packages for the system emulator and the ui modules
2026-03-12 19:31:19 <mps> ikke: version of kernel on builder?
2026-03-12 19:32:18 <ikke> I believe 6.6.45
2026-03-12 19:32:33 <ikke> It's not the builder btw, it's a dedicated dev server with very few containers
2026-03-12 19:32:51 <mps> ah
2026-03-12 19:33:10 <mps> kernel is too old
2026-03-12 19:33:19 <ikke> yeah, I wanted to update it, but updating it broke things
2026-03-12 19:34:06 <mps> but my loongarch64 is not accessible to me RN
2026-03-12 19:34:41 <mps> maybe i'll manage to fix it tomorrow
2026-03-13 01:34:28 <huajingyun> Good morning
2026-03-13 01:36:22 <huajingyun> ikke: did you upgrade dev server kernel from 6.6.45 to the latest version in edge (currently linux-lts 6.18.16) and then encounter an "illegal instruction" error during the reboot process?
2026-03-13 01:36:55 <huajingyun> or did the dev server already reboot normally in the new kernel, and then an error occurred when executing a command? Could you describe the process in detail? 
2026-03-13 02:01:08 <huajingyun> Hi jvvv, did you continue checking afterward? I just built py3-lmdb on my machine and all tests passed (reverted: http://dup.pw/alpine/aports/546482578cd0)
2026-03-13 02:01:34 <huajingyun> I wanted to view the py3-lmdb test failure logs on the builder, but they have now been overwritten
2026-03-13 02:08:19 <jvvv> huajingyun:  Unfortunately I did not save one of the logs from build that failed that had failed. But I can scrape my browsers' cache, maybe I will find one.
2026-03-13 02:10:35 <huajingyun> jvvv: thanks
2026-03-13 02:15:24 <jvvv> I think the cache got cleared when rebooted earlier, which was after the last time I viewed one of the failure logs. Sorry.
2026-03-13 02:16:16 <jvvv> I have my browser settings to clear cache at exit
2026-03-13 02:28:47 <huajingyun> jvvv: thanks for taking the time to check! I haven't been able to reproduce it after multiple local builds
2026-03-13 02:29:25 <jvvv> I just tried to build it without skipping test in a qemu vm and it failed same as on the builder. just a moment, I can post a link to a paste so you can review it
2026-03-13 02:34:52 <jvvv> https://tpaste.us/KxQo
2026-03-13 02:35:39 <jvvv> It looks the same as the earlier failures
2026-03-13 02:43:25 <huajingyun> jvvv: Great! thank you
2026-03-13 02:44:10 <DWwanghao> I tested it on a physical machine with the 6.18.0-1-lts kernel and it worked fine. It seems we need to revert to an older kernel version for testing
2026-03-13 02:47:52 <jvvv> Welcome, happy to help
2026-03-13 03:16:24 <huajingyun> DWwanghao: fyi:I've tested it with both kernel 6.19.5-1-stable and 6.18.6-0-lts, and both py3-lmdb are fine
2026-03-13 03:22:24 <ikke> huajingyun: it already happened before rebooting 
2026-03-13 03:23:34 <ikke> It causes all apk triggers to fail including updating initfs 
2026-03-13 03:24:02 <ikke> I have not rebooted it yet
2026-03-13 03:31:03 <ikke> But yes, I've updated it from ~3.21 edge to current edge and it immediately started to fail with "illegal instruction" for all executables 
2026-03-13 03:31:36 <ikke> An killed by sig am
2026-03-13 03:31:43 <ikke> Signal 4
2026-03-13 04:54:54 <huajingyun> ikke: Based on this information, I have a guess that the issue might be caused by the environment still using kernel 6.6.45
2026-03-13 04:55:19 <huajingyun> This was the initial MR that enabled LSX and LASX for the kernel, at which time the linux-lts version was 6.6.46  https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/70603/diffs#7a18eb60fca8b7170f467f1227ae626970ae7dd9 
2026-03-13 04:55:38 <huajingyun> this indicates that the dev server kernel 6.6.45 you mentioned earlier does not yet support vectors
2026-03-13 04:56:14 <huajingyun> also, since you didn't reboot with the new kernel after upgrading, the "illegal instruction" error when executing commands is likely due to incompatibility with the old kernel
2026-03-13 04:59:36 <huajingyun> did you upgrade using "apk upgrade -aU"? If everything went smoothly during the system upgrade process, I guess it should work normally after rebooting
2026-03-13 05:18:12 <ikke> That's the problem, it didn't go smoothly 
2026-03-13 05:20:30 <ikke> And i haven't tried it yet, but i suspect even the reboot command no longer works 
2026-03-13 05:21:09 <ikke> In any case, i need to work to get OOB access 
2026-03-13 06:17:13 <huajingyun> Oh, yes, the machine is in EU
2026-03-13 06:17:40 <ikke> correct
2026-03-13 06:19:35 <huajingyun> ikke, could you also check the kernel versions of the other Builder servers? I remember the Builder kernel was upgraded earlier, but just in case, make sure it's at least a version higher than 6.6.46
2026-03-13 06:20:07 <ikke> 6.6.47-0-lts
2026-03-13 06:20:31 <ikke> CI: 6.6.54-0-lts
2026-03-13 06:21:58 <huajingyun> Ok,thanks
2026-03-13 06:23:27 <ikke> I wonder it it would be possible to restore busybox purely with a shell :D
2026-03-13 06:23:39 <ikke> printf into /bin/busybox or something like that
2026-03-13 06:28:46 <huajingyun> Let's give it a try
2026-03-13 06:30:08 <ikke> Would https://cdn.alpinelinux.org/v3.21/main/loongarch64/ still work with that kernel?
2026-03-13 06:32:54 <DWwanghao> Yes, I just tested that kernel version 3.21 also works, 6.12.76-0-lts
2026-03-13 06:33:27 <huajingyun> yes
2026-03-13 06:33:27 <ikke> I mean with the older kernel
2026-03-13 06:33:37 <ikke> 6.6.46
2026-03-13 06:33:44 <ikke> otherwise it doesn't matter
2026-03-13 06:42:10 <huajingyun> Is there a way to copy the extracted busybox file (https://cdn.alpinelinux.org/v3.21/main/loongarch64/busybox-1.37.0-r14.apk) to the dev server?
2026-03-13 06:42:54 <ikke> I can not create new ssh sessions, I only have a single existing session
2026-03-13 06:43:10 <ikke> So I gather the only way would be to echo it into a file using ash
2026-03-13 07:09:17 <huajingyun> another solution is to try rebooting via BMC, unsure if this server in the EU has BMC configured
2026-03-13 07:15:15 <huajingyun> Alternatively,  can try execute "ipmitool power reset" , but I suspect ipmitool might not be installed
2026-03-13 07:33:05 <ikke> Yeah, that's what I meant with OOB access
2026-03-13 07:47:29 <huajingyun> Understood
2026-03-13 07:55:19 <ikke> But it
2026-03-13 07:55:44 <ikke> But I see it also see it as a challenge to recover it without BMC
2026-03-13 09:06:47 <clandmeter> they have bmc iirc
2026-03-13 09:06:52 <huajingyun> Yes, I'm trying to find a kernel version 6.6.45, but I haven't found one yet, need to recompile
2026-03-13 09:07:01 <clandmeter> openbmc right
2026-03-13 09:42:55 <huajingyun> clandmeter: it shouled be
2026-03-13 09:55:25 <huajingyun> ikke: sorry,I'm afraid I can't reproduce your error scenario. After rebuilding and upgrading, the system failed to reboot (it encountered a command execution error, as expected), there are no available repo for the 6.6.45 kernel, that version is too old:-(
2026-03-13 09:55:51 <huajingyun> the kernel version released with Alpine v3.21 was already 6.12.1
2026-03-13 11:03:41 <aelin> the py3-lmdb tests all pass for me by the way
2026-03-13 11:03:52 <aelin> no segfaults
2026-03-13 11:50:03 <ikke> I've found the BMC ip and credentials
2026-03-13 16:01:11 <ikke> Booting a live iso, but taking ages over network
2026-03-13 16:23:38 <ikke> "OK: 172.1 MiB in 111 packages"
2026-03-13 16:23:44 <ikke> See if that was sufficient
2026-03-13 16:35:42 <ikke> It boots, but no network yet for some reason
2026-03-13 16:47:44 <ikke> ok, fixed, the server is back up \o/
2026-03-16 01:08:05 <huajingyun> ikke: thanks