2026-05-15.log

Zabaaccording to https://www.kernel.org/doc/Documentation/devicetree/bindings/chosen.txt, the stdout-path property in the device tree can be used instead of the console= kernel parameter to set the boot console00:45
- mjw (QUIT: Ping timeout: 264 seconds) (~mjw@gnu.wildebeest.org)01:42
+ paperManu (~paperManu@204.244.197.237)01:43
+ paperManu_ (~paperManu@204.244.197.237)01:44
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50)02:22
- paperManu (QUIT: Ping timeout: 276 seconds) (~paperManu@204.244.197.237)03:06
- paperManu_ (QUIT: Ping timeout: 255 seconds) (~paperManu@204.244.197.237)03:25
+ paperManu (~paperManu@79.127.134.33)03:39
- paperManu (QUIT: Read error: Connection reset by peer) (~paperManu@79.127.134.33)04:15
- casparvitch (QUIT: Read error: Connection reset by peer) (~casparvit@user/casparvitch)04:33
- Chwoka (QUIT: Remote host closed the connection) (~Chwoka@216.73.127.21)04:41
+ Chwoka (~Chwoka@216.73.127.21)04:44
- sad_plan (QUIT: Quit: WeeChat 4.9.0) (sadplan@tilde.club)05:52
joschright, but the order of console= parameters on the cmdline matters and how would i express that via my dtb?06:44
joschefi frustrates me because while u-boot has its fundamental problems, we had all of this figured out already and now "the new guy" comes along and nothing works anymore06:45
joschi'm told that one of the advantages of booting via efi would be that just like with a "normal" x86 computer i can just ship a generic installation medium and it will boot on a variety of platforms -- that's why i don't understand why we have to bake in stuff like device tree or the cmdline into the uki and make the whole thing device-specific06:46
joschkcxt mentioned the "autodtb feature in UKI" the week before the last but i was unable to find more on that...06:47
kfxyou can shove multiple dtbs into one uki and there are a bunch of ways to make sure the right one is selected, depending whether you're using ukify or stubble or what06:55
kfxfedora uses stubble for this, not sure what debian uses.  ukify takes dtbs and a hwid file to do the mapping, but iirc it requires smbios to be configured in a particular way06:57
joschkfx: we could use stubble -- there is a Debian ITP for that. Right now we use systemd kernel-install as kcxt suggested which uses systemd-ukify.06:59
joschand with that, according to the docs, i can specifiy only a single dtb via /etc/kernel/devicetree07:00
kfxiirc the ukify thing uses systemd-stub .dtbauto sections07:01
kfxhttps://www.freedesktop.org/software/systemd/man/latest/systemd-stub.html#  described here07:02
joschokay, so maybe systemd-stub can do it and then the limiting point is /usr/lib/kernel/install.d/60-ukify.install07:04
joschbecause that one calls ukify only with --devicetree= and not with --devicetree-auto= which according to its man page populates the .dtbauto sections07:05
joschhrm... maybe the way to do is, is by filling /etc/kernel/uki.conf with DeviceTreeAuto= lines?07:15
joschsystem image is building, lets see what happens07:43
joschalso interesting in this context: https://github.com/systemd/systemd/issues/3683507:43
- lidstah (QUIT: Remote host closed the connection) (~lidstah@gateway/tor-sasl/lidstah)08:26
+ lidstah (~lidstah@gateway/tor-sasl/lidstah)08:31
joschhrm... nope, log says Using config file: /etc/kernel/uki.conf09:02
joschbut according to objdump the resulting efi has no .dtbauto sections09:03
joschhere is my /etc/kernel/uki.conf that i used for testing: https://mister-muffin.de/p/sOU-.txt09:04
joschbut there is a .dtb section with rk3588 (non-dsi) where does that come from??09:16
- wakest (QUIT: Remote host closed the connection) (m-a7d6fe@67.219.99.225)09:18
+ wakest (m-a7d6fe@67.219.99.225)09:18
+ casparvitch (~casparvit@user/casparvitch)10:30
+ andreas-e (~Andreas@2a02-8434-b6a3-e901-facc-8e87-8e54-890d.rev.sfr.net)10:43
+ sad_plan (sadplan@ssh.tilde.club)10:46
- andreas-e (QUIT: Quit: Leaving) (~Andreas@2a02-8434-b6a3-e901-facc-8e87-8e54-890d.rev.sfr.net)10:53
joschminute: when uploading pipeline artifacts i'm again gettings this failure: WARNING: Uploading artifacts as "archive" to coordinator... 504 Gateway Time-out  id=20777 responseStatus=504 Gateway Time-out status=504 token=glcbt-6411:21
joschminute: assuming that this is because the disk is full (i've been building system images) i'll just remove all artifacts from my "josch" namespace, okay? So the links i've sent you will now stop working. But I think that's okay because these were buggy anyways and the current version is fixing things. :)11:22
joschthat was 26.38 GiB of artifacts11:25
+ mjw (~mjw@gnu.wildebeest.org)11:38
joschokay, this is really silly...12:18
joschI now know how to create .dtbauto sections for all our dtbs in the uki but how to do it is very suboptimal. I'll probably have to contact systemd maintainers about this.12:19
minutejosch: hmmm i cleaned the disk not so long ago but i can check now12:24
minutejosch: i have a todo to fix up the backups which are endless and totally clog up everything there12:24
minutehmm12:25
minute> /dev/vda3                   1008G  528G  439G  55% /12:25
minute> 46.38.248.210:/voln779700a1 1000G  586G  415G  59% /nfs12:25
minutejosch: around 1TB free!12:25
minutejosch: from time to time i get a bunch of emails that source is not reachable by a cronjob on our website, mostly happens at night, i don't know yet what grinds it down... could be DoS by ai scrapers etc12:26
minutejosch: and sure, no problem @ your artifacts12:26
+ paperManu (~paperManu@204.244.197.237)12:30
+ talos1 (~talos@35.151.128.31)12:58
- talos (QUIT: Ping timeout: 246 seconds) (~talos@2600:6c5d:f0:1fd0:928a:37ab:7f64:2ac0)12:59
* talos1 -> talos12:59
- buckket (QUIT: Quit: buckket) (~buckket@vps.buckket.org)13:06
+ buckket (~buckket@vps.buckket.org)13:06
- lidstah (QUIT: Remote host closed the connection) (~lidstah@gateway/tor-sasl/lidstah)13:28
+ lidstah (~lidstah@gateway/tor-sasl/lidstah)13:28
joschokay, then the problem was not related to the size at all but maybe just a random hickup14:01
josch(but i think the disk is 1 TB and you have 415 G of that free?)14:01
- paperManu (QUIT: Ping timeout: 245 seconds) (~paperManu@204.244.197.237)14:08
+ Svp (~svp@2002:4f07:f0bd:0:a119:2b09:7ce5:44d7)14:32
+ paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca)14:33
- qbit (QUIT: Remote host closed the connection) (~qbit@user/qbit)15:43
+ qbit (~qbit@user/qbit)15:47
josch[    0.000000] efi: EFI v2.10 by Das U-Boot16:15
joschminute: i just booted via efi on rk3588-dsi classic reform16:15
joschhttps://mister-muffin.de/p/my50.txt16:15
joschdisplay remains black and at some point serial disconnects16:16
joschbut that uki includes all dtbs, each in one .dtbauto section16:17
joschminute: the image i used is from this job in case you want to experiment: https://source.mnt.re/josch/reform-system-image/-/jobs/2078316:17
- jah (QUIT: Ping timeout: 245 seconds) (~jahkosha@user/jahkosha)16:18
+ jahkosha (~jahkosha@user/jahkosha)16:22
* jahkosha -> jah16:22
- Svp (QUIT: Ping timeout: 252 seconds) (~svp@2002:4f07:f0bd:0:a119:2b09:7ce5:44d7)16:53
minutejosch: those are 2 disks16:59
minutejosch: one local one nfs16:59
joschminute: ah indeed my mistake :D17:15
+ Svp (~svp@host-79-7-240-189.business.telecomitalia.it)18:01
gordon1josch: forgot to ask, now you have the physical stick, does it work mechanically with the keyboard mod that i did (with 3d printed mockup)?18:43
joschgordon1: it did work with it back then, yes18:56
joschgordon1: the interesting question is: does the pcb somehow touch the stem in a way which influences the measurement18:56
joschthe second part is what i can try out now, yes18:56
joschand the problem is that it is very hard to secure the pcb in a way such that the stem does *not* touch the "keyboard" (the plastic mock-up)18:57
joschof course a bigger hole would help but we also know that there is not really space for a bigger hole18:57
joschi'm going to print a new mock-up soon which will come with some "camefering" (?) around the edges of the hole18:57
gordon1josch: it will also give you estimate on what sort of forces the thing will exert on mounting screws and the bracket18:58
joschyes, this is also suboptimal right now because the plastic bends :/18:58
josch(the next mockup will have a thick "frame" to reduce the bending)18:59
gordon1well if it will not bend at all that would mean that pcb wont either18:59
joschit's hard to describe :D i'll ping you once i have a better mock-up :)19:00
gordon1or you can print it out of something like cf-petg19:00
joschthe department next door also has an SLS printer -- maybe it will be more realistic when made out of resin19:02
joschi have some options and have to yet try them :)19:02
joschgordon1: i 100% need you once it comes to making/ordering the pcb, so i have to keep you into the loop :D19:06
gordon1sure19:07
gordon1josch: are you planning to order it with assembly or just the pcb and assemble it yourself?19:08
joschgordon1: i can solder switches but i cannot solder the chips19:09
gordon1tbh i never ordered stuff with assembly service so probably won't help you on that part, but certainly can review a pcb19:11
joschXD19:11
gordon1i usually assemble things  myself, and afaik if you need some less common chips you have to handle all the logistics yourself19:13
- paperManu (QUIT: Ping timeout: 255 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca)19:57
+ paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca)19:59
+ spew (~spew@user/spew)20:44
- spew (QUIT: Quit: nyaa~) (~spew@user/spew)20:51
joschgordon1: hey, it was you who said "do a keyboard as your first project" -- should be easy, right? XD20:55
gordon1yep20:59
gordon1cannot be easier than a keyboard! :D21:00
joschhaha :)21:00
joschwell, at least i'll learn a lot :D21:00
- xktr (QUIT: Ping timeout: 245 seconds) (~xktr@user/xktr)21:28
+ spew (~spew@user/spew)21:32
+ xktr (~xktr@user/xktr)21:35
- xktr (QUIT: Ping timeout: 272 seconds) (~xktr@user/xktr)22:08
+ xktr (~xktr@user/xktr)22:10
- paperManu (QUIT: Ping timeout: 246 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca)22:42
- xktr (QUIT: Ping timeout: 272 seconds) (~xktr@user/xktr)23:09
+ xktr (~xktr@user/xktr)23:16
+ paperManu (~paperManu@204.244.197.237)23:24
+ paperManu_ (~paperManu@204.244.197.237)23:25

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!