2026-03-10 16:13:05 <ikke> Someone bootstrapped GHC against musl https://wiki.gentoo.org/wiki/User:Mjo/GHC_binary_packages
2026-03-12 18:59:09 <ikke> Anyone happens to be available to debug this test panic only happening on riscv64: https://build.alpinelinux.org/buildlogs//build-edge-riscv64/community/opentofu/opentofu-1.11.5-r1.log?
2026-03-12 18:59:18 <ikke> --- FAIL: TestJSONView_PlannedChange (0.00s)
2026-03-12 18:59:20 <ikke> panic: runtime error: invalid memory address or nil pointer dereference [recovered, repanicked]
2026-03-12 19:32:35 <mps> i can on pioner lxc
2026-03-12 20:06:53 <mps> hm, it hangs
2026-03-12 20:08:19 <ikke> what?
2026-03-12 20:11:10 <mps> opentofu-1.11.5-r1 build on lxc hangs
2026-03-12 20:12:10 <ikke> It may take some time
2026-03-12 20:13:52 <mps> here https://tpaste.us/QKb0
2026-03-12 20:20:22 <mps> and i didn't yet finished to set development env on musebook, maybe on weekend
2026-03-12 20:21:30 <mps> (we are renewing and extending house so all is in the mess and chaos)
2026-03-13 06:02:09 <ikke> Fun, on my container: <autogenerated>:1: internal compiler error: '(*timeTimer).reset': unCached value v15 = ConstNil <func(any, uintptr, int64)> not found in cache, auxInt=0x0, adjusted aux=0xc759e285
2026-03-13 06:02:31 <ikke> Just on build
2026-03-13 11:53:39 <raspbeguy> Hello there, I have a build failure : "SIGILL: illegal instruction" https://gitlab.alpinelinux.org/raspbeguy/aports/-/jobs/2257679
2026-03-13 11:53:50 <raspbeguy> do you think there is a workaround?
2026-03-13 11:54:29 <ikke> raspbeguy: Not sure if it's the same issue, but Lotheac has been tracking down a similar issue
2026-03-13 11:54:52 <ikke> If it's the same issue, it would only happen on the nor-ci-1 runner
2026-03-13 11:55:34 <raspbeguy> Can I override the runner instance?
2026-03-13 11:56:39 <ikke> No
2026-03-13 11:57:27 <raspbeguy> how many riscv64 runner are there? Are the build jobs equally distributed among them?
2026-03-13 11:58:08 <ikke> 4, it's random
2026-03-13 11:58:49 <raspbeguy> ok so if I run the job again it will eventually pass (if that's the same issue than you're talking about)
2026-03-13 12:00:34 <raspbeguy> yep, it passes on fra-ci-1
2026-03-13 19:35:53 <mps> https://gdriscv.com/#/ and click "EN" if you don't understand chinese. interesting thing for devs
2026-03-14 12:08:20 <ikke> Something strange is going on with opentofu on riscv64
2026-03-14 12:08:35 <ikke> This is the lining causing the panic: return fmt.Sprintf("<invalid>.%s.%s", r.Type, r.Name)
2026-03-14 12:08:46 <ikke> r.Type and r.Name are both stringsd
2026-03-14 12:08:59 <ikke> but this is already part of a code path that "should not happen"
2026-03-14 13:51:17 <ikke> and *ofcourse* the test passes when running with a debugger
2026-03-15 04:36:15 <iggy>   I would say disable it and set a reminder for a couple months that hopefully it magically gets fixed in a later release
2026-03-15 04:37:26 <iggy> lame, but this seems like it's taking up an inordinate amount of time for a use case that probably isn't highly sought after
2026-03-15 05:23:48 <ikke> iggy: well, I was debugging it because it seems to go much deeper than just these test failures
2026-03-15 05:23:58 <ikke> https://github.com/golang/go/issues/78161
2026-03-15 05:24:47 <ikke> We have 2 work-arounds now
2026-03-29 07:10:04 <mps> algitbot: retry master
2026-03-29 20:18:26 <mps> algitbot: retry master
2026-03-30 16:13:47 <mps> trying to build qemu on riscv64 got this error '#error "Header 'rbd/librbd.h' could not be found"'
2026-03-30 16:14:14 <mps> what pkg have this file
2026-03-30 16:14:35 <achill> https://pkgs.alpinelinux.org/contents?file=librbd.h&path=&name=&branch=edge&repo=&arch= says ceph19-dev but im not sure if thats the correct package
2026-03-30 16:14:51 <achill> ah yeah rbd looks like a ceph thing
2026-03-30 16:17:33 <mps> log is here https://tpaste.us/KxXP
2026-03-30 16:17:48 <achill> o│>>> sssd: Signing the index...
2026-03-30 16:17:48 <achill> │>>> syncthing-gtk: Signing the index...
2026-03-30 16:18:04 <achill> oops wrong window
2026-03-30 16:19:04 <mps> ah, ceph-dev is not added to makepends on riscv64
2026-03-30 16:20:13 <mps> oh, it is, but mess
2026-03-30 16:24:42 <mps> ok, should we disable rbd in qemu on riscv64
2026-03-30 16:25:04 <achill> what why
2026-03-30 16:26:25 <mps> because it is not explicitly enabled in current APKBUILD
2026-03-30 16:26:51 <achill> looks like it is:
2026-03-30 16:26:56 <achill> case "$CARCH" in arm*|x86) ;; *) .....
2026-03-30 16:27:09 <achill> effectively means that its enabled for riscv64
2026-03-30 16:27:54 <mps> yes, but something doesn't work with this
2026-03-30 16:28:09 <mps> error i just posted
2026-03-30 16:28:18 <achill> right
2026-03-30 16:28:51 <achill> /usr/lib/gcc/riscv64-alpine-linux-musl/15.2.0/../../../../riscv64-alpine-linux-musl/bin/ld: /usr/lib/ceph/libceph-common.so.2: undefined reference to `udev_unref@LIBUDEV_183'
2026-03-30 16:29:07 <achill> these look like the actual errors tho
2026-03-30 16:29:28 <achill> i would just try rebuilding ceph
2026-03-30 16:30:11 <mps> it is slow on my local riscv