2022-05-01 20:44:52 ikke: i enabled the new webhook container 2022-05-03 04:33:52 gitlab security notification 2022-05-05 07:45:25 re alpine-instealler-testsuite gitlab action. could we use a small equinix metal instance for it? ie. have a runner in a bare metal machine, so we dont need nested virtualizaion? 2022-05-05 07:45:53 or maybe we could even spawn it up and tear it down when running it 2022-05-05 07:46:46 there are no small equinix instances anymore 2022-05-05 07:47:00 the one we had is not available anymore 2022-05-05 07:47:58 ncopa: you can just use a shell runner 2022-05-05 07:47:58 s/small/smallest available/ 2022-05-05 07:47:58 ncopa meant to say: re alpine-instealler-testsuite gitlab action. could we use a smallest available equinix metal instance for it? ie. have a runner in a bare metal machine, so we dont need nested virtualizaion? 2022-05-05 07:48:01 its already available 2022-05-05 07:48:10 i use it to use qemu 2022-05-05 07:48:20 i set them up recently 2022-05-05 07:48:43 both for x86 and aarch64 2022-05-05 07:50:17 ncopa: https://gitlab.alpinelinux.org/clandmeter/alpine-disk-image/-/blob/master/.gitlab-ci.yml 2022-05-05 07:51:01 the idea is: when tagging a release, run a job that will: 1) spin up a bare metal machine (using terrafor), 2) install git qemu etc 3) checkout alpine-installer-testsuite 4) download the release artifacts to the baremetal machine, 5) run the testsuite 6) report back to gitlab runner 2022-05-05 07:52:40 the gitlab runner (shell) already has qemu and git installed 2022-05-05 07:53:24 oh, cool 2022-05-05 07:53:43 i think what i do is similar to what you do 2022-05-05 07:54:02 except i use packer and you use expect 2022-05-05 07:54:47 shell executor does not use any virtualization 2022-05-05 07:55:59 i do use docker, but its just to deploy things 2022-05-05 08:09:15 My idea is to use space we have on azure for that 2022-05-05 08:09:55 azure is baremetal? 2022-05-05 08:09:58 no 2022-05-05 08:10:02 that it's not 2022-05-05 08:10:10 hence the nested virtualization 2022-05-05 08:10:29 There are instance types which support it 2022-05-05 08:11:06 if we expose /dev/kvm in docker, what would be the downside to that? 2022-05-05 08:11:15 on real hw 2022-05-05 08:12:17 If we just run the suite on tags, which we control, I see little issue 2022-05-05 08:12:50 the disadvantage of using a shell executor is that its not a clean env on each run 2022-05-05 08:12:54 correct 2022-05-05 08:13:13 but if you do your magic in qemu, it does not really matter that much. 2022-05-05 08:13:23 also correct 2022-05-05 08:14:15 spinning a real machine on equinix on each run seems like lot of wasted time, i guess it will take some time to provision the server. 2022-05-05 08:19:05 We could check what we can do with spot instances? 2022-05-05 08:21:41 how long time does it takes to spin up an alpine instance on equinix? 2022-05-05 08:27:14 i have no clue, but only loading the efi firmware takes a lot of time 2022-05-05 08:27:32 is it 30 seconds or 5 mins? 2022-05-05 08:27:52 depends on hw i guess 2022-05-05 08:28:12 im using a server for work atm that takes minutes to boot 2022-05-05 08:29:45 i mean why would you want to spin a bare metal server to afterwards run qemu in it? 2022-05-05 08:30:17 to run the alpine-installer-testsuite 2022-05-05 08:30:21 what you are testing runs in qemu right? not on metal? 2022-05-05 08:30:28 yes 2022-05-05 08:30:57 so then its kind of useless to needing to spin a new server and wait for boot install and kexec 2022-05-05 08:31:02 i set up a matrix of different emulated hardware combinations (IDE disk, virtio disk, nvme) and run the installer on it 2022-05-05 08:31:25 i test with one or two disks 2022-05-05 08:31:33 test boot with bios or uefi 2022-05-05 08:31:45 how many arches are you testing? 2022-05-05 08:31:57 you want to run emulated or kvm? 2022-05-05 08:32:00 right now only x86 and x86_64 2022-05-05 08:32:08 preferrible not 2022-05-05 08:32:27 i'd also like to test on aarch64 hw 2022-05-05 08:32:31 some arches will be more difficult to test i guess? 2022-05-05 08:32:34 test armv7 and aarch64 2022-05-05 08:32:37 yup 2022-05-05 08:32:39 like s390x 2022-05-05 08:33:00 i think for ppc64le and s390x we might need to do some simple testing with emulated qemu 2022-05-05 08:33:11 running with kvm is always nice 2022-05-05 08:33:15 yup 2022-05-05 08:33:28 thats why i wonder why not just expose it in docker? 2022-05-05 08:33:49 you mean run qemu in docker? 2022-05-05 08:33:56 what would you miss out on? 2022-05-05 08:34:00 yeah 2022-05-05 08:34:06 i can do that 2022-05-05 08:34:14 just need to pass in /dev/kvm i guess 2022-05-05 08:34:18 i was thinking to do the same 2022-05-05 08:34:23 and move aways from shell executor 2022-05-05 08:34:51 but docker itself runs on bare-metal? 2022-05-05 08:34:56 we need to do that on the runner, but we can set up specific ones for just kvm usage 2022-05-05 08:35:39 docker runs on baremetal? 2022-05-05 08:35:46 dont understand that question :) 2022-05-05 08:36:24 i think "set up specific ones for just kvm usage" answered my question 2022-05-05 08:36:37 ok :) 2022-05-05 08:36:59 that way you dont need to wait for boot, installer, kexec 2022-05-05 08:37:08 just have a docker image fit for your needs 2022-05-05 08:37:51 right 2022-05-05 08:38:17 i was just thnking that we don't need have the bare-metal machine constantly running 2022-05-05 08:38:45 since those tests only need to run when we tag release candidates or releases 2022-05-05 08:39:33 well that could be possible, but with the cost mentioned above. 2022-05-05 08:39:40 and another thing 2022-05-05 08:39:45 we cant spin arm servers 2022-05-05 08:40:13 ok 2022-05-05 08:40:20 i still need to sent another email to arm ppl 2022-05-05 08:40:53 to show them how important alpine is to the arm community 2022-05-05 08:41:05 and 1 server is not enough for such tasks 2022-05-05 08:43:37 clandmeter: and riscv 2022-05-05 08:43:51 ? 2022-05-05 08:44:11 alpine is important for riscv, imo 2022-05-05 12:26:05 We do get hardware from riscvm, but here is no server-grade hw yet 2022-05-05 12:48:08 ikke: yes, i know, but I have rights to dream 2022-05-05 18:43:50 ikke: for some reason someone emailed me saying they can't even fork aports as public 2022-05-05 18:44:23 Well, public is disabled for users. They should be able to set it to internal 2022-05-05 18:44:44 yes, but you said it needs to be public for anything to work later 2022-05-05 18:45:18 and that things automatically go from internal to public 2022-05-05 18:45:23 yes 2022-05-05 18:45:24 so.. what is the point of disabling it 2022-05-05 18:45:30 We only do it for aports 2022-05-05 18:45:47 this is so frustrating that i wish we just allowed public-only and called it a day 2022-05-05 18:45:50 i'm so tired of this 2022-05-05 18:46:01 The reason we set it up like that is that we want to discourage users from using this instance for random projects 2022-05-05 18:46:51 The issue that prevents contributors from rebasing when the project is internal has been fixed, so just a little bit of patience left before that lands 2022-05-05 18:47:26 yes, but if they fork aports as private then the same issue remains 2022-05-05 18:47:43 nothing rewrites private into anything 2022-05-05 18:51:37 re random projects; is there no way to forbid new non-fork repo creation entirely in CE 2022-05-05 18:53:46 I suppose setting the projects limit to 0 2022-05-05 19:42:34 So it turned out I immediately upgraded to gitlab 14.10 instead of first to 14.9 2022-05-05 19:42:47 https://docs.gitlab.com/ee/update/#1490 2022-05-05 19:45:49 ah 2022-05-05 19:46:16 neat 2022-05-05 19:46:28 The mentioned background jobs are running 2022-05-05 19:47:41 they should perhaps complete.. 2022-05-05 19:48:00 yes, I see one already making progress 2022-05-05 19:48:06 MigratePersonalNamespaceProjectMaintainerToOwner: members 99.00% 2022-05-05 19:48:23 at last i can be a true owner /s 2022-05-05 19:48:30 :D 2022-05-05 19:52:55 How about NullifyOrphanRunnerIdOnCiBuilds: ci_builds 7.00% 2022-05-05 19:54:50 a whole null of my own, too 🥺 2022-05-05 20:56:41 rebasing !34009 also fails with the new hook 2022-05-05 21:05:26 I see what's going on, will look at it tomorrow 2022-05-06 09:51:18 we don't have rust in s390x and riscv-64 2022-05-06 09:53:01 new clamav needs rust 2022-05-06 12:13:57 psykose: one thing I did change is that by default projects are set to internal instead of private, maybe that will reduce the amount of forks that start as private 2022-05-06 12:15:52 hope so 2022-05-06 12:23:03 thanks :3 2022-05-06 16:24:01 14.10.2 does not solve the issue errors 2022-05-06 16:24:04 sadly 2022-05-06 17:54:50 Oof, the irony: a 500 response from gitlab when trying to report an issue :/ 2022-05-08 15:44:35 any plans to fix that Issues in Gitlab are no longer viewable if not logged in? Or is this a new Gitlab "feature" that cannot be reverted? 2022-05-08 15:46:42 https://gitlab.com/gitlab-org/gitlab/-/issues/361699 2022-05-08 15:46:54 Want to fix it, but trying to figure out how 2022-05-08 15:53:28 ok, thanks 2022-05-08 15:55:15 Something graphql related 2022-05-08 15:58:58 weird that the MR page works though when not logged in 2022-05-09 12:30:14 I received confirmation that the rv64 board is being shipped 2022-05-09 12:32:05 someone could've walked it across the atlantic in this time /s 2022-05-09 12:40:42 ikke: also 2022-05-09 12:40:48 few days ago 2022-05-09 12:41:36 I downloaded some docs about JH7100 and started to read when I have some time 2022-05-09 12:42:04 and I started to skim over ISA 2022-05-09 12:42:44 btw, last night I managed to destroy my macbook m1 2022-05-09 12:42:59 ouch, how did you manage to do that? 2022-05-09 12:43:46 simple, spilled water over it :\ 2022-05-09 12:44:59 bummer 2022-05-09 12:47:12 will try to call service if they can 'save' it 2022-05-09 13:35:45 "Your package is delayed due to local reasons." 2022-05-09 13:35:50 More delays 2022-05-09 13:57:39 giga delay 2022-05-09 14:03:16 ikke: can't you add another worker machine so the package will build quicker? ;-) 2022-05-09 14:16:54 Same here 2022-05-10 04:31:23 apparently we are not the only one having the problem with the error on the issue list when not being logged in 2022-05-10 18:00:32 https://git.pleroma.social/pleroma/pleroma/-/issues 2022-05-10 18:00:38 same issue 2022-05-10 18:00:57 minimal: ^ 2022-05-10 18:05:44 ikke: on Gitlab's own issues page you can't search for an issue (such as this problem) without being logged in....doh! 2022-05-10 18:06:52 Yeah, that's anoying 2022-05-10 18:07:04 But at least you can still see the issues 2022-05-10 18:24:31 yeah but short of trawling through all their recent issues there's no easy way to know (i.e. search) if its already on their radar 2022-05-10 18:24:48 I did search for it 2022-05-10 18:25:27 and someone else managed to find at least this one 2022-05-11 10:26:38 So apparently the error has to do with 'public' being restricted in our instance 2022-05-11 14:42:04 Someone from gitlab is working on a fix for the issue 2022-05-11 14:45:43 \o/ 2022-05-11 17:53:31 Ariadne told me she cannot push to main anymore 2022-05-11 17:54:02 can we just delete this stupid hook 2022-05-11 17:54:10 we either trust people to push to alpine or we don't 2022-05-11 17:54:38 i guess that 1. makes sense 2. would simplify things 2022-05-11 17:56:53 what is enforcing that limitation? 2022-05-11 18:00:29 I also have that. "You are not allowed to push to the main repository. Try again." 2022-05-11 18:01:33 but it seems like i can press there mer when pipeline finishes 2022-05-11 18:05:28 ikke: can we just delete this? i don't think it really gets us anything 2022-05-11 18:06:10 like i understand the theory behind it -- graduate people up to having main access -- but realistically, it just blocks people from getting work done 2022-05-11 18:06:39 and now it is blocking anyone from getting *anything* done 2022-05-11 18:06:51 does it? it seems to work for me now? 2022-05-11 18:06:54 not sure what is happening 2022-05-11 18:06:59 i saw the error message 2022-05-11 18:07:52 but after a refresh or something it disapeared? 2022-05-11 18:08:52 merge something, then :p 2022-05-11 18:09:10 i vote for !33751 2022-05-11 18:10:26 nope it does not work for me 2022-05-11 18:12:11 weird. I wonder how it is supposed to work. and what enforces it 2022-05-11 18:12:29 shell script with some greps and a magic file 2022-05-11 18:14:44 i think if we want to have this, we should do it as separate projects in gitlab 2022-05-11 18:14:55 instead of something ported from gitolite days 2022-05-11 18:15:06 but i am not convinced we really need it 2022-05-11 18:15:12 we should just get rid of it 2022-05-11 18:15:20 it's a point of failure that serves only political purpose 2022-05-11 18:15:47 psykose: do you know how/where they are executed? 2022-05-11 18:15:49 from where 2022-05-11 18:16:00 gitlab runs it as a pre-commit hook 2022-05-11 18:16:02 some hook container on something i don't have access to 2022-05-11 18:19:48 /srv/docker/repositories/alpine-aports/custom_hooks/update 2022-05-11 18:20:31 Sorry /srv/docker/gitlab/... 2022-05-11 18:21:54 can we just delete this damn thing 2022-05-11 18:22:51 it is political theatre and provides no actual security benefit to alpine -- if somebody wants to fuck up alpine with commit rights, they will just jump through the additional hoops to get main access anyway 2022-05-11 18:23:24 ikke: so it is not to be found from the webui? 2022-05-11 18:23:29 in fact, right now, it is causing security problems to alpine 2022-05-11 18:23:35 because i cannot mitigate CVEs 2022-05-11 18:25:35 ncopa: correct 2022-05-11 18:26:05 where does this machine run? 2022-05-11 18:26:19 deu2-dev1 2022-05-11 18:26:26 Linode 2022-05-11 18:29:51 ok. i dont have login there 2022-05-11 18:33:10 Oof 2022-05-11 21:15:00 ikke: was it the hook that failed? 2022-05-12 04:48:55 more like 100% 2022-05-12 04:50:05 100% is more than 95% :P 2022-05-12 04:51:27 indeed so 2022-05-12 04:51:28 :3 2022-05-12 04:51:49 afaik most of the space is trimmed at the end of the run once the old gopkgs will be trimmed 2022-05-12 04:51:58 just well, 6 worlds at once is a big growth 2022-05-12 04:52:35 Yeah, at leas in the current raid config, that box does not have enough space 2022-05-12 04:52:40 (lvm raid) 2022-05-12 04:57:55 also very fun, zabbix removed the unsupported tags from github, so 5.4 is gone for 3.15 2022-05-12 05:02:12 they have 'official' tarballs on the website, https://www.zabbix.com/download_sources#unsupported but they have dbschema missing or something 2022-05-12 05:04:15 they can be generated 2022-05-12 05:04:52 make dbschema 2022-05-12 05:06:32 that's what i mean :) it's undefined 2022-05-12 05:06:51 looking at the configure stuff, everything is gated behind #if DBSCHEMA 2022-05-12 05:06:58 and DBSCHEMA is set from a `test -d create` 2022-05-12 05:07:20 if you force make the directory, you get failures with ` No rule to make target '../../../create/src/schema.tmpl', needed by 'dbschema.c'. Stop. ` 2022-05-12 05:07:32 so, simply the create folder with the base things is not part of the tarball 2022-05-12 05:07:53 did you run autoreconf? 2022-05-12 05:08:00 https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/community/zabbix/APKBUILD#L98 2022-05-12 05:08:27 i mean.. the apkbuild is running it right there 2022-05-12 05:08:29 not doing this by hand 2022-05-12 05:08:33 ah ok 2022-05-12 05:09:23 idk why the 'official' tarballs are missing this 2022-05-12 05:10:25 oh, hm 2022-05-12 05:10:49 they are already built, that's why 2022-05-12 05:11:03 no need to run it at all, unless it's supposed to be updated 2022-05-12 05:12:36 Right, if you download it from github (source), you do need to do it, but hte ofificial tarballs already have it included 2022-05-12 05:13:00 mhm 2022-05-12 05:13:04 guess it can just be removed then 2022-05-12 05:13:07 But strange that it fails 2022-05-12 05:13:43 well, the rule is not generated without the base files being present 2022-05-12 05:13:51 they purposefully just remove them 2022-05-12 05:14:53 ah ok 2022-05-12 05:17:38 We do at least have the source we built with on distfiles 2022-05-12 05:18:55 are you doing a go upgrade? 2022-05-12 05:18:57 it's fine to swap i think 2022-05-12 05:19:00 yeah, this was the last thing 2022-05-12 05:19:13 just doublechecking a diff on the edge one 2022-05-12 05:19:23 5.4 is eol, will not receive any updates anymore 2022-05-12 05:19:24 removing make dbschema removes some files, now seeing if new source + remove is identical 2022-05-12 05:19:30 yeah, but it's a 3.15 rebuild 2022-05-12 05:19:36 ahuh 2022-05-12 05:19:58 I mean that there will be no other versions anymore anyway 2022-05-12 05:20:02 mhm 2022-05-12 05:22:29 it does have one more side effect 2022-05-12 05:22:51 some useless .gitignore folders are removed from usr/share/webapps/zabbix in zabbix-webif 2022-05-12 05:23:09 aha 2022-05-12 05:23:11 and all of usr/share/webapps/zabbix/tests/ is dropped, which leads to a 63->60 mb change 2022-05-12 05:23:21 you should change to the official tarball next upgrade 2022-05-12 05:23:42 The main reason I used source is that it's easier to test rc's 2022-05-12 05:23:48 yeah, it is 2022-05-12 05:24:03 in the past that was usefull specifically for go and 32-bits 2022-05-12 05:24:22 and with that, both go's are now sent 2022-05-12 05:24:26 rip to diskspace 2022-05-12 05:25:56 i think there's some $HOME/go / .cargo cache to trim by now 2022-05-12 05:26:31 yeah 2022-05-12 05:32:39 150G should be enough for a while 2022-05-12 05:32:58 yeah 2022-05-12 05:33:00 for builder in build-3-16-* build-3-15-* build-edge-*; do rm -rf ./$builder/rootfs/home/buildozer/.cargo ./$builder/rootfs/home/buildozer/.cache/go-build ./$builder/rootfs/home/buildozer/.cache/yarn ./$builder/roo 2022-05-12 05:33:02 tfs/home/buildozer/go; done 2022-05-12 05:33:28 it wasn't not enough currently, but there is that middle period where old packages aren't deleted yet, so unfortunate 2022-05-12 05:34:16 But we need to think about the next release 2022-05-12 05:34:32 i don't think it's possible without another machine 2022-05-12 05:34:49 At least this one still has unassigned space 2022-05-12 05:34:52 ah 2022-05-12 05:34:55 then sure 2022-05-12 05:35:03 there will be space freed up on proper release 2022-05-12 05:35:10 I'm just conservative in assigning it 2022-05-12 05:35:10 er, no 2022-05-12 05:35:43 The rc's can be removed 2022-05-12 05:36:09 those are pretty small anyway 2022-05-12 05:36:12 yup 2022-05-12 07:05:43 ikke: ? 2022-05-12 07:11:14 re update hook? 2022-05-12 07:14:14 yup 2022-05-12 07:24:17 the hook was working, the acl list outdated 2022-05-12 09:13:04 ah ok, thx for checking. 2022-05-12 09:16:44 rebooting deu1/deu7 for the usual kernel and things :) 2022-05-12 09:20:05 algitbot: ping 2022-05-12 09:21:19 ikke: we need to think about a plan to get those new t1 servers going 2022-05-12 09:21:29 or at least the new NL one 2022-05-12 09:31:47 You wanted to use zfs, right? 2022-05-12 09:47:01 we can, the disk setup is kind of optimal for zfs usage 2022-05-12 09:47:08 but its not a must for me 2022-05-12 09:50:34 I don't mind 2022-05-12 09:50:47 I have no experience with it myself though 2022-05-12 09:51:50 lets change that :) 2022-05-12 09:52:12 i have played with it, just not that much. 2022-05-12 09:52:41 do we want to run docker on it? 2022-05-12 09:57:07 good quesiton, not sure 2022-05-12 10:32:32 ugh, from 150G free to 36G free in a couple of hours 2022-05-12 10:34:31 i'm curious where the space went, considering no actual new packages were added 2022-05-12 10:35:34 ~/go apparently 2022-05-12 10:36:05 hah 2022-05-12 10:36:09 well, that makes sense 2022-05-12 11:42:49 i might need some handholding to set up a CI job for alpine-conf. I'd like to install kyua and run `make test` for every MR 2022-05-12 11:43:19 afaik that is editing the .gitlab-ci.yml in the alpine-conf repo 2022-05-12 11:43:54 +- perhaps hitting a toggle somewhere 2022-05-12 11:44:26 maybe I need to create a docker image as well? 2022-05-12 11:44:31 then the rest is the gitlab ci pipeline thing, for which the apk-tools ones might be of use https://gitlab.alpinelinux.org/alpine/apk-tools/-/blob/master/.gitlab-ci.yml 2022-05-12 11:44:36 ehh, i don't think so 2022-05-12 11:44:47 apk just uses standard alpine:latest 2022-05-12 11:45:04 oh... thats nice 2022-05-12 11:45:23 even if you want to 'add stuff' into the container, unless it's building custom stuff upfront, it's not slow to apk add / run a bit of shell 2022-05-12 11:45:39 if you want to build a whole thing then yeah, a custom image would be useful to make it faster 2022-05-12 11:46:42 but from the sound of it, this is just running make test on some shell scripts, so just alpine should be fine 2022-05-12 11:48:40 If it's just a few deps, it's maybe easier to do it directly in .gitlab-ci.yaml. If you need a more elaborate setup, we can make an image 2022-05-12 11:48:48 yeah 2022-05-12 11:48:58 i guess I'll create an MR with it 2022-05-12 11:49:55 Don't forget to add tags to the jobs to select the runners 2022-05-12 11:50:16 you can probably copy the apk one and leave only the test stage and go from there 2022-05-12 11:57:38 Something like this? 2022-05-12 11:57:38 https://tpaste.us/qaNa 2022-05-12 11:57:38 $ tpaste < .gitlab-ci.yaml 2022-05-12 11:58:33 Note that each jobs starts with a clean tree by default 2022-05-12 11:59:59 so if the goal is to just run tests (that depends on built files), it's probably enough to have a single job that does both 2022-05-12 12:00:06 yeah 2022-05-12 12:01:03 Multiple stages can make sense if you want to avoid expensive builds by checking things in advance 2022-05-12 12:04:50 do I need to add runners for it? https://gitlab.alpinelinux.org/alpine/alpine-conf/-/merge_requests/65 2022-05-12 12:05:14 ncopa: you need to enable shared runners here: https://gitlab.alpinelinux.org/alpine/alpine-conf/-/settings/ci_cd 2022-05-12 12:05:46 under the runners section 2022-05-12 12:06:16 And then restart the job 2022-05-12 12:06:54 yes! 2022-05-12 12:07:57 :3 2022-05-12 12:09:29 sweet! 2022-05-12 12:09:38 something is broken in the kyua setup though 2022-05-12 12:10:05 i assume the test-env it sources doesn't exist or something 2022-05-12 12:10:08 is something supposed to make it 2022-05-12 12:26:10 ah, the ole strace in the ci job :) 2022-05-12 12:26:47 :D 2022-05-12 13:01:37 i cannot figure out what it does and why it fails 2022-05-12 13:05:18 Can you reproduce it locally in a docker container? 2022-05-12 13:12:20 nope 2022-05-12 13:12:24 but I figured it out now 2022-05-12 13:12:49 i have a *.sh in .gitignore, so the tests/test_env.sh was never included in git 2022-05-12 13:13:04 aha 2022-05-12 13:13:06 hey i guessed that :) 2022-05-12 13:13:26 oh.. sorry i missed that while on lunch... 2022-05-12 13:13:33 hehe 2022-05-12 13:13:37 hope it was as tasty as mine 2022-05-12 13:13:47 psykose: i definitively need to better at listening 2022-05-12 13:19:24 \o/ its now in prod! 2022-05-12 13:19:48 \/o 2022-05-12 13:19:58 and with that i'm off to bed 2022-05-12 13:20:21 o/ 2022-05-12 13:34:24 good night! 2022-05-12 15:05:04 any idea why there are no runners for this MR? https://gitlab.alpinelinux.org/alpine/alpine-conf/-/merge_requests/64 2022-05-12 15:11:56 ncopa: did you enable shared runners after rebasing / starting the pipeline? 2022-05-12 15:12:12 no 2022-05-12 15:12:23 do i need enable shared runners for the users fork? 2022-05-12 15:12:36 Normally not, we don't need to do it for aports either 2022-05-12 15:12:45 And they are enabled 2022-05-12 16:35:22 clandmeter: What do you think about lifthing the restriction on public repos on our instance? More then once it has created issues 2022-05-12 16:35:52 i think we have to be careful with it 2022-05-12 16:36:51 what would be the reason to do it? 2022-05-12 16:37:29 1) get rid of issues caused by it (apparently gitlab is not testing with this enabled a lot) 2022-05-12 16:38:06 2) psykose (and I assume others as well) get frustrated because they cannot rebase some MRs in certain cases 2022-05-12 16:39:48 thats related to permsisions? 2022-05-12 16:39:49 it's also annoying because internal/private forks default to making MR in own repo (because alpine repos are public and gitlab is like: we need to protect you) 2022-05-12 16:40:59 clandmeter: Yes 2022-05-12 16:43:08 if psykose has issues with it, maybe she can help montior abuse if we switch? 2022-05-12 16:46:10 abuse is the main concern here right? 2022-05-12 16:47:17 The primary concert was users starting to host their own personal projects which is a liability for us 2022-05-12 16:47:45 We already have people agreeing to some terms regarding to that 2022-05-12 16:48:43 does it impact CI usage? 2022-05-12 16:49:08 No, in the sense that people can already abuse CI with private projects 2022-05-12 16:49:23 nod 2022-05-12 16:49:56 i guess in the end it does not really matter? 2022-05-12 16:50:06 Yeah, I was thinking the same 2022-05-12 16:52:46 So should I proceed and remove it? 2022-05-12 16:53:04 sounds good to me 2022-05-12 16:53:22 and would be good if we have somebody to keep an eye on it 2022-05-12 16:54:46 minimal: issues are visible again when not logged in 2022-05-12 16:55:04 clandmeter: I think you should be able to get ssh keys again now as well 2022-05-12 16:55:33 :) 2022-05-12 16:55:56 Pandoras box has opened 2022-05-12 17:40:29 ikke: good news :-) 2022-05-12 17:40:43 ike: was it a bug or a feature? 2022-05-12 17:41:00 bug 2022-05-12 17:41:22 but triggers by having public repos restricted 2022-05-12 17:41:27 triggered* 2022-05-12 19:34:45 ncopa: I think for existing forks, you'd need to enable shared runners, new forks will have it enabled 2022-05-12 19:35:03 https://gitlab.alpinelinux.org/dbradley/alpine-conf/-/jobs/719708 2022-05-12 23:52:31 sure, i can watch gitlab if you want me to, though i'm unsure of what lets one do that aside from some admin ui 2022-05-12 23:53:01 it's all fine by me though, i don't mind 2022-05-13 09:02:48 psykose: did you create MR for clamav 0.104.3 2022-05-13 09:03:47 no, how come 2022-05-13 09:03:47 I doubt that the 0.105.0 will be solved soon, it needs rust but we don't have rust on s390x and riscv64 2022-05-13 09:03:52 yep 2022-05-13 09:03:59 just thought i'd make it .3 since we had it in 3.15 already 2022-05-13 09:04:16 re: rust, imo it's fine to just disable it on those arches i guess 2022-05-13 09:04:25 psykose: ok, I will cherry pick from 3.15, I upgraded clamav there few days ago 2022-05-13 09:04:26 and perhaps in the next 6 months we will have rust there, finally :) 2022-05-13 09:04:40 mps: i mean that i already merged 0.104.3 into edge 2022-05-13 09:04:51 ah, good 2022-05-13 09:05:56 'perhaps' sounds like infinite ;) 2022-05-13 09:06:45 well, for s390x it's guaranteed 2022-05-13 09:06:52 or there will not be s390x anymore :p 2022-05-13 09:06:58 for riscv i guess it is more infinite.. 2022-05-13 09:07:11 but it already builds, just need to figure out some internal issues, hopefully upstream helps 2022-05-13 09:08:34 (I'm not sure we need s390x (and ppc64le) but ...) 2022-05-13 09:22:43 ppc64le at least usually works :) 2022-05-13 09:22:52 (until binutils or go or whatever else magically stops working..) 2022-05-13 09:54:28 anything that has ring as dependency: am I a joke to you? 2022-05-13 17:19:12 what do y'all use for arm builders, and where did you get them? 2022-05-13 17:29:45 zv: it is one neoverse-n1 machine and builders for armhf,armv7 and aarch64 are 3 lxc 2022-05-14 07:44:38 zv: arm sponsors us a machine 2022-05-14 07:45:19 netbox upgraded to 3.2.3 2022-05-14 14:34:20 ikke: do you have an account manager there / would you be in a position to share their contact info? 2022-05-16 06:44:25 zv: do you need arm64 resources for something? 2022-05-16 11:15:14 ikke: did you board get into the country? 2022-05-16 11:15:23 i guess its also stuck at customs? 2022-05-16 11:19:00 I didn't receive any updates yet 2022-05-16 11:19:24 mine is stuck at ehv airport 2022-05-16 11:19:37 mps: ? 2022-05-16 11:30:31 clandmeter: I've got mail this morning that it is 'on fedex' and that they will send updates as it goes 2022-05-16 11:31:04 'This shipment is scheduled to be sent on Mon 5/16/2022.' 2022-05-16 11:31:41 (what a stupid date format) 2022-05-16 11:32:00 I guess it is 2022-05-16 2022-05-16 12:46:34 clandmeter: forgot to add from mail: `Estimated delivery date Tuesday, 24/05/2022` 2022-05-16 14:00:25 anyone have idea how to run shell script on first boot with newly installed alpine 2022-05-16 14:00:56 /etc/local.d/ is ok but it is not 'interactive' 2022-05-16 14:02:25 /root/.profile? 2022-05-16 14:05:41 interactive means you need to login 2022-05-16 14:06:08 so that means profile or /etc/profile.d iirc 2022-05-16 14:07:57 yes, after root login 2022-05-16 14:55:37 hmm, just got mail that the board is delayed, starting to think it will not ever be here 2022-05-16 15:21:21 rails: yeah 2022-05-16 15:23:44 zv: It's the works for arm program 2022-05-16 15:24:21 ikke: we used to have that -- were kicked out of it *without notice* and our data deleted, including signing keys -- because we are not "server" enough 2022-05-16 15:24:38 ouch 2022-05-16 15:25:07 server enough in what sense? wonder what their reasoning was 2022-05-16 15:26:29 works on arm is eol iirc 2022-05-16 15:26:48 we'll write about it in the next blog post, but basically they went in circles with excueses 2022-05-16 15:26:51 "Further discussion involved bad-faith arguments and went in circles. It became clear that only corporate-benefiting projects “where there are clear signals of customer demand” would be supported. Other FOSS projects seem to have been treated similarly." 2022-05-16 15:26:53 at least the concept has changed 2022-05-16 15:27:25 zv: for which project are you asking? 2022-05-16 15:27:30 clandmeter: adelie 2022-05-16 15:27:35 ah ok 2022-05-16 15:27:52 i dont recall your nick 2022-05-16 15:27:59 Thalheim 2022-05-16 15:28:10 ah ok, that name does ring a bell 2022-05-16 15:28:43 maybe rails can help 2022-05-16 15:29:42 yeah, that looks like a me issue 2022-05-16 15:29:43 :> 2022-05-16 15:29:54 zv: hit up fossho.st 2022-05-16 15:30:11 I think we did, twice, to no avail. will check again. 2022-05-16 15:30:19 oh, really? 2022-05-16 15:30:30 get me ticket numbers and i can do some yelling 2022-05-16 15:30:33 ah, Thalheim is zv 2022-05-16 15:30:38 or vice versa 2022-05-16 15:32:03 rails: 491108 is one of them 2022-05-16 15:32:37 oh, sorry 2022-05-16 15:33:31 it looks like the application was migrated to a different system and we needed to resubmit it 2022-05-16 15:33:38 huh 2022-05-16 15:33:44 it's pre- osticket? 2022-05-16 15:34:20 pre- june 2021 2022-05-16 15:34:43 Ah. 2022-05-16 18:42:44 ncopa: will upgrade mutt and I think about using gnu sasl instead of cyrus, debian discussed this here https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=999672 2022-05-16 18:43:06 what is your stance 2022-05-16 18:43:49 look also here http://mutt.org/relnotes/2.2/ under 'GNU SASL support' section 2022-05-16 18:45:51 ok I guess. dont have energy to deal with it right now 2022-05-16 18:47:22 ok 2022-05-16 18:48:08 will make MR and first test it locally 2022-05-16 18:51:07 MR is !34497 2022-05-17 09:13:03 ncopa: !34510 2022-05-17 09:13:52 yes, it is ugly but I didn't got anything better, and would be nice to have latest iptables in 3.16 release 2022-05-17 09:13:58 mps: ping me when its no longer draft 2022-05-17 09:14:19 I can remove it right now 2022-05-17 09:14:53 done 2022-05-17 09:15:16 LGTM 2022-05-17 09:15:44 thanks 2022-05-17 09:16:07 I will wait one day to see if someone else have comments 2022-05-17 09:25:08 `It's a fairly small update. Maybe the patchs can get a small comment why they are necssary 2022-05-17 09:25:49 ikke: I wrote in commit msg 2022-05-17 09:26:03 and the source is quite clear 2022-05-17 09:26:32 u_int16_t is replaced with uint16_t 2022-05-17 09:27:37 do you want to see my samba upgrade :D 2022-05-17 09:28:02 it would hurt eyes 2022-05-17 09:46:23 this rv board shipment is delay on delay on delay 2022-05-17 09:57:16 :) 2022-05-17 10:19:52 they must go to Chinese secret service, and then to USA secret service, and then to Germany secret service and maybe some other. they have to put their spyware in boards 2022-05-17 13:26:48 it arrived :) 2022-05-17 13:27:25 but no time to play with it unfortunately 2022-05-17 13:30:35 mps, maybe it can replace your m1 ;-) 2022-05-17 13:31:23 clandmeter: how do you 'read' my dream 2022-05-17 13:31:58 clandmeter: didn't I told you that I have another M1? 2022-05-17 13:32:02 packaging is nice 2022-05-17 13:32:10 i had to unpack 5x :) 2022-05-17 13:32:44 I'm jealous 2022-05-17 13:33:03 not 5 boards 2022-05-17 13:34:19 5 pins on board for extensions ;) 2022-05-17 13:34:55 what model is it? 2022-05-17 13:36:38 clandmeter: my son bought new M1 pro macbook and gave me his old macbook m1 2022-05-17 13:36:54 yes you mentioned it 2022-05-17 13:36:59 you are spoiled 2022-05-17 13:37:06 and you spoiled 2022-05-17 13:37:09 aha, sorry, I forgot 2022-05-17 13:38:00 actually it was red wine, water was later for washing 2022-05-17 13:39:44 sure it wasnt something stronger? 2022-05-17 13:40:06 VisionFive v1 2022-05-17 13:40:45 it has a fan included, hope its of better quality as the previous one <- nmeum 2022-05-17 13:41:13 I also got an email from fedex 2022-05-17 13:41:22 but it's not here yet :( 2022-05-17 13:41:29 no, I don't drink unnatural drinks, wine, brandy only 2022-05-17 13:41:45 how natural is brandy? 2022-05-17 13:42:00 sorry for out of topic, lets move this :) 2022-05-17 13:42:05 hm, maybe I use wrong term 2022-05-17 13:42:44 if it have 8GB ram it could be quite usable as workstation 2022-05-17 13:43:17 small home server for sure 2022-05-17 13:44:29 but I think I will use it for connecting different electronic devices for experimenting, if it ever come here 2022-05-17 13:44:59 first it needs to boot alpine 2022-05-17 13:45:40 it will, trust me ;) 2022-05-17 13:45:54 (if I got it) 2022-05-17 15:19:45 It's radio silence for me 2022-05-17 17:34:53 minimal: what you think about build grub2 with `--enable-device-mapper` 2022-05-17 17:37:47 it is needed to run grub-install in chroot for loop mounted img files 2022-05-17 17:50:24 grub is fat anyways and install disk needs to ship lvm2 anyways so it sounds fine 2022-05-17 18:39:58 mps: I already run grub-install inside chroot on loop mounted files with no changes 2022-05-17 18:44:24 do I remember correctly that there was some json endpoint on a.o to grab list of releases? 2022-05-17 18:44:37 https://alpinelinux.org/releases.json 2022-05-17 18:44:44 cheers 2022-05-17 18:51:09 minimal: hm, on my all try it fails 2022-05-18 09:59:37 The vm crashed, restarted it 2022-05-18 14:32:59 ncopa: after reporting iptables bug to #netfilter someone told me that he is pushed few hours ago practically same patch as ones I created, https://git.netfilter.org/iptables/patch/?id=f319389525b066b7dc6d389c88f16a0df3b8f189 2022-05-18 14:33:34 so, should we pick this one and apply or keep iptables as it is now 2022-05-18 14:34:01 i think we should pick that one 2022-05-18 14:34:09 but its not critical 2022-05-18 14:34:24 ok 2022-05-18 14:34:42 there is also this one https://www.spinics.net/lists/netfilter-devel/msg76308.html 2022-05-18 14:35:05 but this is not merged 2022-05-18 14:37:00 not critical either. in general, I prefer use patches from upstream, but it is not a hard rule 2022-05-18 14:37:34 ok, I'll add one which is merged upstream 2022-05-18 21:08:54 FTR, the problem with the issue list not showing up when not being logged in while public visibility is restricted, has been resolved 2022-05-18 21:54:10 but for some reason the number of comments is not visible (when logged out) 2022-05-18 23:00:10 i think this is just a gitlab thing though 2022-05-19 17:09:58 my rv64 board arrived now as well 2022-05-19 17:15:09 nice 2022-05-19 17:15:33 my will arrive next year I guess 2022-05-19 17:16:37 but I'm getting mails that it will 'on the road' :) 2022-05-19 17:16:58 s/will/is/ 2022-05-19 17:16:58 mps meant to say: but I'm getting mails that it is 'on the road' :) 2022-05-19 17:42:58 nmeum: do you know if https://gitlab.alpinelinux.org/nmeum/alpine-unmatched also works with the VisionFive boards we received? 2022-05-19 17:48:10 Appears they forked u-boot: https://github.com/starfive-tech/u-boot 2022-05-19 18:05:12 ikke: as I read from docs looks like patched/forked u-boot will be needed 2022-05-19 20:19:52 Added deu1 to dmvpn (there was already a cert setup, but it was never installed) 2022-05-20 11:39:14 ncopa: I have local build for mutt where I 'fixed' making manual.txt by adding docbook-xml, docbook-xsl and elinks to makedepends. I think it would be good to have this in next release 2022-05-20 11:40:15 then push it 2022-05-20 11:40:33 ok, thanks 2022-05-20 14:25:39 https://i.imgur.com/mGnu6qB.png 2022-05-20 15:40:01 clandmeter: ^ traefik traffic monitoring you wondered about the other day 2022-05-21 08:00:12 ikke: I haven't looked at he VisionFive boot process yet as I have not received mine yet, but it will probably require a different image creation script 2022-05-21 08:00:21 s/he/the/ 2022-05-21 08:00:21 nmeum meant to say: ikke: I haven't looked at the VisionFive boot process yet as I have not received mine yet, but it will probably require a different image creation script 2022-05-21 08:00:43 I will look into it as soon as I receive mine in the mail :) 2022-05-21 08:14:39 When I have time I try to look into it as well. The instructions are here: https://rvspace.org/ 2022-05-21 08:48:15 I got mail with tracking fedex url but whenever I try I got `Unfortunately we are unable to retrieve your tracking results at this time. Please try again later.` 2022-05-21 08:49:01 only once saw status that it is on the Paris airport 2022-05-21 08:49:58 Yes, same for me 2022-05-21 08:50:10 And then suddenly it was just delivered 2022-05-21 08:50:45 ah, they want to make surprising gift to us ;) 2022-05-21 08:51:39 also I read docs on rvspace.org, doesn't look complicated to make it booting 2022-05-21 08:52:43 but u-boot patches aren't upstream yet 2022-05-21 08:59:07 Should we for the time being have a separate package, or try to patch the u-boot package? 2022-05-21 09:03:18 probably separate would be better 2022-05-21 09:03:50 there are not small number of patches 2022-05-21 09:04:44 I'm tempted to create separate u-boot aport for apple m1 2022-05-21 09:06:31 also, I found this https://www.youtube.com/watch?v=7jzgmz5qQss in chinese but have translation 2022-05-21 11:41:03 harfbuzz 4.3.0 is released, have (big) speed improvemnts https://github.com/harfbuzz/harfbuzz/releases/tag/4.3.0 2022-05-21 11:41:23 upgraded it locally, works fine 2022-05-21 11:41:46 not sure should we push it now for next stable 2022-05-21 12:34:29 does it break ABI? 2022-05-21 12:35:05 I think you can push it as long as it does not break ABI 2022-05-21 12:39:05 looks like it doesn't. running fine firefox, libreoffice, telegram, geeqie with just harfbuzz upgrade 2022-05-21 12:40:06 that is gui programs I have installed. and also awesome wm ofc 2022-05-21 12:41:58 oh, forgot I have kicad installed 2022-05-21 12:47:02 now I'm working on picolibc for rv64, think we will need it for visionfive board 2022-05-21 12:51:00 ncopa: also zathura pdf works. I guess it is safe to upgrade harfbuzz 2022-05-21 12:51:33 I will create MR for now to see if someone have objections 2022-05-21 12:52:02 sgtm 2022-05-21 12:52:58 do we have estimated release date for 3.16 2022-05-21 12:53:04 monday 2022-05-21 12:53:22 aha, ok 2022-05-21 12:53:56 I think I have to upgrade ncurses tomorrow, there is few fixes, not much important but anyway 2022-05-21 12:55:05 the only thing I request is no big builds on monday 2022-05-21 12:55:22 what gets in gets in 2022-05-21 12:55:23 ncurses is fast on builders 2022-05-21 12:55:32 no prooblemo 2022-05-21 12:55:46 as long as it does not break ABI... which I think it doesnt 2022-05-21 12:55:56 so I will keep gcc-cross-embed for rv64 after release 2022-05-21 12:56:13 no, ncurses have just few fixes 2022-05-21 12:57:38 I hope 'New API -> +hb_map_is_equal()' in harfbuzz doesn't count as risky 2022-05-21 12:57:57 https://github.com/harfbuzz/harfbuzz/releases/tag/4.3.0 2022-05-21 17:47:56 it's gonna fall over in a second, probably 2022-05-21 17:49:41 fail over? 2022-05-21 17:50:45 run out of space :) 2022-05-21 17:50:55 ~200G free still 2022-05-21 17:50:59 hm 2022-05-21 17:51:00 it's flapping around 95% 2022-05-21 17:51:18 ah, is it 4tb total 2022-05-21 17:51:46 oh, 85G, but still 2022-05-21 17:51:53 4 dotnet builds may or may not swipe that 2022-05-21 17:52:10 sucks when 3.16 and edge build at the same time 2022-05-21 17:52:15 aye 2022-05-21 17:52:32 /dev/mapper/vg0-mirror 1.9T 1.7T 85G 96% /srv/mirror 2022-05-21 17:53:00 oh, I'm looking at nld3 :P 2022-05-21 17:53:07 usa9 is obviously something else 2022-05-21 17:55:51 /dev/mapper/vg0-lv_root 1.1T 924G 96G 91% / 2022-05-22 07:09:34 would someone with some free time look at !34656 2022-05-22 07:36:10 psykose: could be !31848 skipped and simply upgrade mesa for 3.16-stable 2022-05-22 07:36:34 or it is too late 2022-05-22 10:46:49 yesterday around this time I started build of gcc-cross-embeded on rv64 lxc dev container and it is still working, 24 hours 2022-05-22 13:10:06 mps: not sure what you mean? 2022-05-22 13:12:00 if you mean skipping 22.x for 3.16, sure, that's fine 2022-05-22 13:14:06 psykose: I mean upgrade to 22.0.x for next stable 2022-05-22 13:14:18 ah 2022-05-22 13:14:48 tomorrow releasing will start 2022-05-22 13:14:53 i mean, feel free to do so, i wasn't going to merge it myself, and i don't really care about waiting 78 years until someone actually looks at it 2022-05-22 13:15:33 psykose: tbh, I didn't read MR comments carefully 2022-05-22 13:16:11 only what I'm ready to do is plain upgrade, if you agree 2022-05-22 13:17:38 though that will mean rebuild some other pkgs, I guess 2022-05-22 13:18:57 there is no i786 2022-05-22 13:19:23 i mean that you cannot just plain upgrade it 2022-05-22 13:20:22 never mind, i will put it on mr 2022-05-22 13:22:32 psykose: ok then 2022-05-22 13:22:52 (because all of -classic got removed) 2022-05-22 13:23:48 ok, maintainer should decide then 2022-05-22 14:31:07 what is with aarch64 and armv7 edge builders, they are not visible on build.a.o 2022-05-22 14:31:24 I've stopped them because there were too many havy jobs on a single machine 2022-05-22 14:31:42 Waiting for 3-16-armv7 to finish 2022-05-22 14:31:57 aha 2022-05-22 14:32:05 ok 2022-05-22 15:42:50 Finally finished 2022-05-22 15:46:17 Now qt5-qtwebengine 2022-05-22 19:04:43 ikke: safe to re-enable now 2022-05-22 19:08:54 I enabled aarch64 already 2022-05-22 19:09:15 which is building dotnet6 atm 2022-05-22 20:09:30 although it isn't a full bootstrap from source, I'm amazed someone actually got dotnet running 2022-05-22 22:38:38 technically microsoft got it working ages ago 2022-05-23 04:31:13 True, I meant in Alpine packaging of course which is a feat in itself 2022-05-23 07:40:40 37-38 hours to build gcc-cross-embeded on rv64 lxc 2022-05-23 16:37:57 dns from bra1-dev1 does not seem to work 2022-05-23 16:38:18 Could not resolve hostname build-3-16-ppc64le.bra1.alpin.pw 2022-05-23 16:41:53 looks like dmvpn is down 2022-05-23 16:42:07 it cannot route to 172.16.11.36 2022-05-23 16:44:20 hum.. maybe dmvpn was not even configured there 2022-05-23 16:47:29 i killed charon 2022-05-23 16:47:33 and restarted it 2022-05-23 17:12:46 ncopa: afaik it used to work 2022-05-23 17:13:48 it works again. charon was unhappy 2022-05-23 17:14:01 i did kill -9 2022-05-23 17:14:36 ok, thanks 2022-05-23 20:26:08 https://about.gitlab.com/releases/2022/05/22/gitlab-15-0-released/ 2022-05-23 21:36:14 ah, 3.16 is released 2022-05-24 00:25:55 did any policy change with regards to the EOL date? i'm picking up EOL date for 3.16 as 2024-05-23 instead of what I'd usually expect (2024-05-01) 2022-05-24 00:36:30 well, it just released, and 2 years from now is the former 2022-05-24 04:22:25 psykose: almost none of the past releases have been exactly on *-05-1 2022-05-24 04:22:38 ah 2022-05-24 04:37:15 nice, mirrors automatically picked up 3.16 now 2022-05-24 04:47:06 ikke: alpine-mksite PR waiting for you - i've confirmed locally that the cloud page is as expected 2022-05-24 04:47:59 tomalok: alright 2022-05-24 04:48:43 still rendering the old way for now -- cloud/variant filtering coming soonish 2022-05-24 04:51:20 tomalok: pushed 2022-05-24 04:54:17 thx! 2022-05-24 07:47:01 ikke: my VisionFive also arrived, did you find the time yet to boot alpine on it? 2022-05-24 07:49:49 nmeum: nice you got it 2022-05-24 07:50:29 did they called you by phone or it just 'appeared in inbox' 2022-05-24 07:51:20 I received an email from jeff with a fedex link and fedex send me an email today that it would arrive this morning and indeed it did 2022-05-24 07:52:05 btw: what do I need to do to get SMTP access for nmeum@alpinelinux.org? 2022-05-24 07:52:14 aha, also I got tracking url but it shows only error about tracking 2022-05-24 07:52:33 it took some time to arrive for me as well because it was stuck in customs for a while 2022-05-24 07:53:31 I'm preparing tools needed to build opensbi, u-boot and ddrinit on riscv 2022-05-24 07:54:02 I think we will need picolibc 2022-05-24 07:54:09 why? 2022-05-24 07:54:21 picolibc is just a tiny libc for embedded stuff 2022-05-24 07:54:23 to build u-boot and ddrinit 2022-05-24 07:55:14 I've built gcc-cross-embeded and newlib on rv64 lxc already in last few days 2022-05-24 07:56:06 as I read docs they use picolibc to make these basic things to boot to be small as possible 2022-05-24 07:57:02 but picolibc currently fails to build on alpine, something related to meson directive which I don't understand 2022-05-24 07:57:27 hm 2022-05-24 07:57:44 meson.build:737:10: ERROR: configuration_data.set got unknown keyword arguments "descrption" 2022-05-24 07:57:47 I will read up on the visionfive boot process after $dayjob 2022-05-24 07:58:34 if someone understand this would be nice to tell how to fix it 2022-05-24 07:59:03 looks like there is a typo somewhere? descrption → description 2022-05-24 07:59:47 if it's doing .set(desc=xyz) make it .set(desc, xyz) 2022-05-24 07:59:48 heh, https://github.com/picolibc/picolibc/issues/244 2022-05-24 07:59:54 found it 2022-05-24 08:00:02 ah, actual typo 2022-05-24 08:02:21 I wonder how it passed on builders 2022-05-24 08:03:26 nmeum: if all these tools pass today on my tests I'll inform you 2022-05-24 08:03:45 0.62 meson was recent 2022-05-24 08:04:03 the version is from december, back then it was.. 0.60 or 0.61 2022-05-24 08:04:13 or maybe even 0.59, i forget the dates 2022-05-24 08:05:35 mps: which documentation did you consult for the VisionFive? 2022-05-24 08:06:11 nmeum: I'm following this https://rvspace.org/en/Product/VisionFive/Technical_Documents/VisionFive_Software_Technical_Reference_Manual 2022-05-24 08:06:45 thanks! :) 2022-05-24 08:08:38 nmeum: haven't had time yet to try to boot it. 2022-05-24 08:11:46 ikke: I have also question as nmeum, how to access imap inbox on alpinelinux.org 2022-05-24 08:13:43 We don't have imap afaik 2022-05-24 08:13:45 for me, nmeum@ it just an alias which forwards to my private mail address which works well for me but would be nice to be able to send emails as nmeum@ as well (for which I would need smtp access) 2022-05-24 08:14:03 nmeum: I can look at that later 2022-05-24 08:16:53 thanks, not high priority at all :-) 2022-05-24 08:45:27 morning! 2022-05-24 08:45:40 what do I need to do to get v3.16 to show up on https://pkgs.alpinelinux.org/packages? 2022-05-24 08:47:24 check gbr2 tnux session 2022-05-24 08:47:27 Tmux 2022-05-24 08:47:44 If it's finished, the container needs to be restarted 2022-05-24 08:48:45 exit the container in the tnux session. Then docker compose restart 2022-05-24 09:13:00 Skipping: v3.3/testing/riscv64 2022-05-24 09:13:01 Update finished. 2022-05-24 09:14:33 done! and now v3.16 shows up. Thank you sir! 2022-05-24 09:19:24 👍 2022-05-24 16:53:02 ikke: nmeum: I built u-boot, openbios and ddrinit for visionfive board, only had problem with secondBoot 2022-05-24 16:53:41 (and hope that the board will arrive this week to test) 2022-05-24 16:56:57 my understanding is that this secondBoot, ddrinit, OpenSBI, as well as u-boot are all considered part of the firmware and already in the NOR flash 2022-05-24 16:57:56 so we don't need to package those for alpine but it's of cause still cool to be able to build them from source 2022-05-24 16:58:05 aha 2022-05-24 16:58:17 https://rvspace.org/technical_document/boot_flow_20211126.png 2022-05-24 16:58:19 nmeum: yes 2022-05-24 16:58:57 but would be nice to have all these in case of 'bricking' board 2022-05-24 16:59:09 right 2022-05-24 17:02:48 oh nice, just built secondBoot \o/ 2022-05-24 17:03:41 so everything could be built on alpine 2022-05-24 17:15:51 the fan included with the VisionFive is slightly less annoying than the one on the unmatched :D 2022-05-24 17:16:10 But still annoying 2022-05-24 17:18:36 and the kernel 5.18 is released yesterday, have some riscv improvements 2022-05-24 17:19:09 (lot of work for me these days ;) ) 2022-05-24 17:21:01 lets start, one by one 2022-05-24 17:21:25 we can push now big pkgs I think 2022-05-24 17:21:58 hmm, interesting. The sdcard was in the original packaging, but when I start it, it seems to boot into fedora 2022-05-24 17:22:43 I suppose they pre-flash them from the vendor 2022-05-24 17:22:50 as I read, they support fedora by default 2022-05-24 17:23:05 yes, that's on the rvspace.org 2022-05-24 17:23:37 yes I read there 2022-05-24 17:25:02 ikke: yes, I was also suprised that the packaged untouched sdcard contained a fedora image :D 2022-05-24 17:30:12 Sadly the fan noise means I most likely cannot leave the board running for a longer period 2022-05-24 17:30:53 ikke: put on the roof as clandmeter did :) 2022-05-24 19:43:07 nmeum: if even uboot is part of the firmware, do we need anything specific to boot alpine? 2022-05-24 19:43:37 This diagram suggest a dtb: https://rvspace.org/technical_document/boot_flow_20211126.png 2022-05-24 19:44:11 Seems like u-boot calls into grub? 2022-05-24 19:48:45 Seems like we do not have it yet? https://pkgs.alpinelinux.org/contents?file=*.dtb&path=&name=linux-edge&branch=edge&arch=riscv64 2022-05-24 19:53:34 https://github.com/starfive-tech/linux/tree/visionfive/arch/riscv/boot/dts/starfive 2022-05-24 19:54:12 Is not upstream yet apparently 2022-05-24 19:54:16 Is missing here: https://github.com/torvalds/linux/tree/master/arch/riscv/boot/dts/starfive 2022-05-24 20:00:30 ikke: JH7100 is in kernel 5.18 2022-05-24 20:00:53 though not sure is the DTS for visionfive 2022-05-24 20:01:16 it this board need separate DTS 2022-05-24 20:02:01 yes 2022-05-24 20:02:07 according to rvspace.org it needs it 2022-05-24 20:02:18 and that one is not upstream yet 2022-05-24 20:02:30 At least, as far as I can see 2022-05-24 20:05:47 visionfive_defconfig have comment 'not for upstream' 2022-05-24 20:07:27 and dts is quite small https://github.com/starfive-tech/linux/blob/visionfive/arch/riscv/boot/dts/starfive/jh7100-starfive-visionfive-v1.dts 2022-05-24 20:07:39 easy to add as patch 2022-05-24 20:08:56 ok 2022-05-24 20:11:39 I think I will tomorrow look at linux-edge 5.18 upgrade and will try to add these missing files for visionfive 2022-05-24 20:11:52 nice 2022-05-24 20:12:54 if it become to complicated maybe we can add separate linux-starfive kernel in testing till all is upstreamed 2022-05-24 20:13:03 too* 2022-05-24 20:13:28 yeah, that's what I was thinking, but if it's easy enough to add it to linux-edge, that's preferrable 2022-05-24 20:13:35 I already maintain few such specific kernels 2022-05-24 20:14:09 and have more in local branches 2022-05-24 20:48:11 ikke: their kernel patches, but other then that no 2022-05-24 20:49:13 I started working on a linux-starfive package it compiles atm so I haven't tested it yet, I don't think 3.18 supports everything from their tree yet 2022-05-24 20:49:30 aha ok 2022-05-24 20:50:58 https://github.com/starfive-tech/linux#status 2022-05-24 20:52:28 That's what they've upstreamed? 2022-05-24 20:53:28 not sure, I think this might just be what has been implemented in terms of peripheral support etc 2022-05-25 08:40:47 ikke: you could try to replace the fan with something like a noctua 2022-05-25 08:41:41 i hacked my aruba switch yesterday 2022-05-25 08:41:49 replaced the system fan with a noctua one. 2022-05-25 08:41:55 now i can sleep again 2022-05-25 08:51:17 clandmeter: what is noctua 2022-05-25 08:52:02 https://noctua.at/ 2022-05-25 08:52:04 fans that do not such (that much) 2022-05-25 08:52:11 suck* 2022-05-25 08:52:29 fans which doesn't works ;p 2022-05-25 08:53:11 I'm thinking about passive cooling 2022-05-25 09:11:11 im thinking of immersion cooling 2022-05-25 09:11:23 just put some oil in a bucket and sink it in 2022-05-25 09:14:11 Does it matter if it's sunflower oil or olive oil? :-P 2022-05-25 09:17:55 yes it matters a lot 2022-05-25 09:18:01 sunflower is much more expensive now ;-) 2022-05-25 09:18:54 but immersion oil is even more expensive, and very hard to get atm (fyi) 2022-05-25 10:27:15 Information on booting the fisionfive board https://rvspace.org/en/Product/JH7100/Technical_Documents/JH7100_Boot_User_Guide 2022-05-25 10:35:58 hmhm, read it few days ago, but board is not yet here to test /o\ 2022-05-25 10:37:33 ikke: what about separate channel for riscv? or we can use #riscv on libera for alpine discussion 2022-05-25 10:38:08 I have no idea how much time and effort is needed to keep irc channel running 2022-05-25 10:40:38 not much 2022-05-25 10:42:39 #alpine-riscv64 2022-05-25 10:45:28 OFTC? 2022-05-25 10:45:31 yes 2022-05-25 10:46:22 why 64 at the end? maybe we will have 32 one day 2022-05-25 10:46:34 Because right now we are targeting rv64 2022-05-25 10:47:41 ok, but soon 64 will have options to execute 32 code ;) 2022-05-25 11:00:01 i hope someone decides to remove that feature instead 2022-05-25 11:01:38 why? may I ask 2022-05-25 11:12:38 because aside from old devices and micro-level things 32-bit devices are not made anymore and are pointless, and riscv is a new architecture. so what is the point of adding the 32-bit emulation support? there are no 'old 32 bit' binaries to run with it, or similar things 2022-05-25 11:13:28 even on x86_64 where the x86 emulation exists, what is the point of it? musl doesn't even support multilib for instance, nobody is running around running software that hasn't been recompiled since 1990 for 32-bit only, etc.. 2022-05-25 11:14:01 it's merely more stuff to implement and a mess at the low-level for no gain that i can imagine, aside from maybe making it easier to develop things for riscv-32 proper 2022-05-25 11:16:16 we need 32bit emulation on aarch64 to run armv7 VM in qemu 2022-05-25 11:16:20 the recent apple m1 chips for instance don't have the arm-32 support at all, they are 64 bit only, and everything should follow that, because it doesn't really make sense to add these never-used modes and spend time implementing them and providing software support for them for nobody to use them 2022-05-25 11:16:35 sure, because we build things for armv7 2022-05-25 11:16:56 we could also make a cross-compiler to armv7 and build them that way, not that much different 2022-05-25 11:17:02 and we could build binaries for riscv32 2022-05-25 11:17:57 I also run 32bit armv7 userspace on arm64 machine 2022-05-25 11:18:45 and I also run armv7 VM in qemu on apple m1 2022-05-25 11:19:14 that one is emulated 2022-05-25 11:20:12 oh 2022-05-25 11:50:32 cross compiling is different 2022-05-25 11:50:50 but i agree riscv for 32bit is useless 2022-05-25 11:53:51 for you maybe, but for me is opposite ;p 2022-05-25 13:21:12 nmeum: we have #alpine-riscv64 channel, you can join if you want 2022-05-25 13:21:25 oh very nice 2022-05-25 14:08:08 riscv64 builder is stopped I think 2022-05-25 15:43:47 mps: rebooted it 2022-05-25 15:44:01 clandmeter: ^ :P 2022-05-25 20:38:45 ikke: seen e-mail? 2022-05-25 20:38:54 Just now 2022-05-25 20:39:08 Wonder what to reply 2022-05-25 20:39:50 3 months is over before you know it 2022-05-25 20:40:11 not sure what will be reevaluated 2022-05-25 20:40:27 right 2022-05-25 20:48:34 what's the evaluation about 2022-05-25 20:49:13 We were asking for a new arm server 2022-05-25 20:49:32 They say they can offer one for 3 months and then have to reevaluate 2022-05-25 21:08:25 ah 2022-05-26 16:30:38 hi, congrats for v3.16 release 2022-05-26 16:30:45 vkrishn: thanks 2022-05-26 16:31:21 is it ok that mirrors.alpinelinux.org does not show armv7 entry in status' ? 2022-05-26 16:32:11 Hmm, no, that's not normal