| ch | josch: i replied in the MR, but its only an "zwischen tuer und angel" reply | 00:24 |
|---|---|---|
| josch | ch: understood -- i still am pondering about this as i try to find the thing that makes this not work or at least not a good idea :D | 00:27 |
| josch | ch: thank you for your feedback! | 00:27 |
| ch | i think its a good idea, but -minimal could be a lot more -minimal | 00:28 |
| ch | (the eternal debian package set topic) | 00:28 |
| josch | i thin what you want as minimal is just installing reform-tools | 00:28 |
| josch | and a bit less minimal is reform-tools plus its recommends | 00:28 |
| ch | and a kernel and some firmwrae | 00:28 |
| ch | and rfkill, pcitools, i guess | 00:29 |
| josch | oh good point -- i must not have the kernel there | 00:29 |
| josch | we don't have linux-image-mnt-reform-arm64 in debian :D | 00:29 |
| ch | if we take a lesson from the debian discussions, trying to come up with a definition might tell you what should be -not- in there | 00:29 |
| josch | ch: also, x11-xserver-utils was in there for xhost which once upon a time was used by reform-synaptic... | 00:30 |
| ch | ah | 00:30 |
| josch | so out with it :) | 00:31 |
| ch | does xhost work on a wayland machine? (i really have no idea) | 00:31 |
| ch | the list seemed a bit 'old' in general | 00:31 |
| josch | ch: also, is the util-linux author trying to sneak util-linux-extra into the set?? ;) | 00:31 |
| josch | errr... s/author/maintainer/ | 00:31 |
| ch | let me check what is actually in -extra | 00:32 |
| ch | i thought about it because bsdextrautils is there | 00:32 |
| josch | oh the package has hwclock | 00:32 |
| ch | not sure if you want hwclock, hopefully(tm) not | 00:32 |
| josch | no idea -- it seems i have it installed on my box but... | 00:33 |
| josch | ch: the reason for avoiding Recommends is, that mmdebstrap does not install those by default, that's the whole reason | 00:33 |
| ch | IIRC there was some community discussion that hwclock is required, but maybe thats a symptom of another problem | 00:33 |
| ch | ah | 00:33 |
| ch | extremely OT: for grml-debootstrap we are pondering 'profiles', where the 'desktop profile' would probably want to install recommends | 00:34 |
| josch | makes sense | 00:35 |
| josch | why is gpgv in the Depends?? | 00:35 |
| ch | so it would also benefit from some mmdebstrap --with-recommends option | 00:35 |
| ch | ;) | 00:35 |
| ch | (there are great visions for the tool, but so little time) | 00:35 |
| josch | ch: --aptopt='Apt::Install-Recommends "true"' | 00:35 |
| ch | right | 00:36 |
| josch | (see man page) | 00:36 |
| ch | ok, if its in the man page i'll find it when the time comes | 00:36 |
| ch | thanks | 00:36 |
| josch | ch: we talked about that last week at the sprint in Würzburg -- the idea of mmdebstrap is for it to go away, so i'd like to avoid adding flags for things that are just apt options | 00:36 |
| ch | yeah makes sense | 00:37 |
| ch | g-d wants to kinda become the opposite: d-i but not d-i | 00:37 |
| ch | (just to be clear, we are millions miles away from that) | 00:38 |
| josch | minute: time for a MNT plymouth theme? https://community.mnt.re/t/installing-plymouth/3244/8 | 00:51 |
| ch | yes yes | 00:52 |
| ch | ;) | 00:52 |
| minute | oh yeahhh | 00:52 |
| minute | ch: random comment: yes, xhost works on wayland | 00:53 |
| minute | i do xhost + if i need to quick and dirty run some old gui tool as root | 00:54 |
| josch | plymouth seems to not work on imx8mq, imx8mplus nor a311d but other than some scary error messages, it seems to not mess anything up | 00:54 |
| minute | josch: interesting | 00:54 |
| minute | i'll definitely play with it | 00:54 |
| josch | minute: most worrysome is the a311d hdmi problem... i could bisect this but not sure when i find the time... | 00:55 |
| minute | josch: sorry, i probably missed out... what's wrong with a311d hdmi? just stopped working? | 00:55 |
| josch | minute: no, sway seems to think that there is a second monitor connected even though there is not | 00:56 |
| josch | so if you press super+2, you are on the non-existant second monitor | 00:56 |
| minute | josch: ohh ok... even if there's no dual display support | 00:57 |
| josch | https://community.mnt.re/t/sway-switch-workspace-erratic/3219/1 | 00:57 |
| josch | yes, this happens on unstable since recently | 00:57 |
| josch | i'm on bookworm with a311d classic reform and don't have the problem | 00:57 |
| josch | so likely a recent-ish regression somewhere | 00:57 |
| - sigrid (QUIT: Quit: leaving) (~sigrid@ftrv.se) | 01:06 | |
| - ZylonMaster (PART: Leaving) (~hjcs@ool-ad02e01f.dyn.optonline.net) | 01:14 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 01:20 | |
| - andypiper_ (QUIT: Ping timeout: 268 seconds) (~Srain@45.146.11.160) | 01:23 | |
| - Ar|stote|is (QUIT: Ping timeout: 268 seconds) (~linx@149.210.29.177) | 02:01 | |
| - paperManu (QUIT: Ping timeout: 260 seconds) (~paperManu@107.159.71.33) | 03:21 | |
| - nsc (QUIT: Ping timeout: 268 seconds) (~nicolas@181-98-142-46.pool.kielnet.net) | 03:50 | |
| + nsc (~nicolas@192-97-142-46.pool.kielnet.net) | 03:51 | |
| - mjw (QUIT: Remote host closed the connection) (~mjw@gnu.wildebeest.org) | 04:05 | |
| + Gooberpatrol66 (~Gooberpat@user/gooberpatrol66) | 04:31 | |
| + chomwitt (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 06:15 | |
| + sigrid (~sigrid@ftrv.se) | 08:56 | |
| + Ar|stote|is (~linx@149.210.8.52) | 09:14 | |
| + andypiper_ (~Srain@45.146.11.160) | 09:33 | |
| - andypiper_ (QUIT: Remote host closed the connection) (~Srain@45.146.11.160) | 09:57 | |
| + andypiper_ (~Srain@45.146.11.160) | 09:57 | |
| - Ar|stote|is (QUIT: Ping timeout: 244 seconds) (~linx@149.210.8.52) | 11:03 | |
| - jacobk (QUIT: Ping timeout: 248 seconds) (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 11:04 | |
| + jacobk (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 11:04 | |
| + L29Ah (~L29Ah@wikipedia/L29Ah) | 11:13 | |
| * Guest6513 -> mjw | 11:21 | |
| - chomwitt (QUIT: Ping timeout: 260 seconds) (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 11:25 | |
| + andreas-e (~Andreas@2001:861:c4:f2f0::c64) | 11:29 | |
| + Ar|stote|is (~linx@149.210.25.158) | 12:08 | |
| minute | josch: i wrote this in my "urgent" list now but it's not in the first place | 12:13 |
| minute | josch: my hunch is that something is wrong with the interpretation of the HPD signal | 12:13 |
| minute | josch: because a non plugged in hdmi display should not be available in sway | 12:13 |
| minute | the first thing i'm working on today is trying to get rcore rk3588 work in classic reform without the hdmi adapter | 12:14 |
| josch | minute: no worries, please do that instead -- i am investigating the a311d problem today | 12:16 |
| josch | (and rk3588 reform without needing a hdmi adapter would be super cool!) | 12:17 |
| - buckket (QUIT: Quit: buckket) (~buckket@vps.buckket.org) | 12:32 | |
| + buckket (~buckket@vps.buckket.org) | 12:33 | |
| josch | okay, it's not mesa. Same problem with mesa 24.3 | 12:44 |
| + paperManu (~paperManu@107.159.71.33) | 12:46 | |
| minute | josch: no, it has to be kernel related | 12:47 |
| minute | josch: mesa is only concerned with GPUs. the GPU doesn't know anything about displays. | 12:47 |
| + chomwitt (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 12:48 | |
| josch | right, i'm trying an older system image now | 12:48 |
| - andreas-e (QUIT: Quit: Leaving) (~Andreas@2001:861:c4:f2f0::c64) | 12:53 | |
| minute | josch: any changes in git history to bananapi cm4 dtsi? | 12:54 |
| minute | meson-g12b-bananapi-cm4.dtsi | 12:54 |
| josch | not on our side | 12:59 |
| josch | i'll look upstream | 12:59 |
| minute | https://github.com/torvalds/linux/blob/e48e99b6edf41c69c5528aa7ffb2daf3c59ee105/arch/arm64/boot/dts/amlogic/meson-g12-common.dtsi#L381 | 13:00 |
| minute | we should in general check if the bias here needs to be set for example. maybe it has always been wrong | 13:00 |
| minute | for example maybe we need bias-pull-up; in our .dts for this noe | 13:01 |
| minute | node | 13:01 |
| + gustav28 (~gustav@c-78-82-54-216.bbcust.telenor.se) | 13:02 | |
| minute | uff, EDP_RESETn is not connected on rcore ;/ | 13:06 |
| josch | bodging this with a small wire is maybe possible? | 13:09 |
| minute | possible yeah | 13:10 |
| minute | or maybe replacing the pulldown resistor on the motherboard with a capacitor | 13:11 |
| josch | hm ärgerlich :/ | 13:11 |
| josch | though probably heaper than shipping the hdmi adapters with each unit? | 13:13 |
| josch | *cheaper | 13:13 |
| josch | holy moly, the a311d hdmi option was even present in kernels we built last december... in particular, the last reform-system-image tag sysimage-v4-20250109 is also affacted | 13:30 |
| josch | but then i wonder: if i can reproduce this problem with 6.12.6-1+reform20241228T163100Z | 13:30 |
| josch | then why do i not see it on my a311d classic reform with Debian Bookworm with backported kernel 6.12.9-1~bpo12+1+reform20250123T060214Z | 13:31 |
| - chomwitt (QUIT: Ping timeout: 268 seconds) (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 13:51 | |
| josch | huh... | 14:08 |
| josch | the most recent system image from reform.debian.net has this kernel: | 14:08 |
| josch | Linux reform 6.12.9+bpo-mnt-reform-arm64 #1 SMP Debian 6.12.9-1~bpo12+1+reform20250123T060214Z (2025-01-2 aarch64 GNU/Linux | 14:08 |
| josch | and that version is *not* affected | 14:08 |
| josch | but the set of kernel patches is identical | 14:09 |
| josch | it's just built for bookworm instead of unstable | 14:09 |
| minute | hmm | 14:23 |
| minute | maybe it's sway after all? | 14:23 |
| minute | josch: can you find something about hpd in sysfs? | 14:24 |
| minute | or there's also some drm-related tool to see connector states | 14:24 |
| josch | sure, i have a directory /sys/firmware/devicetree/base/soc/bus@ff600000/bus@34400/pinctrl@40/hdmitx_hpd | 14:34 |
| josch | i'll try if upgrading sway from its version in bookworm to the version in unstable triggers the issue | 14:36 |
| - anzu (QUIT: Quit: WeeChat 3.5) (~anzu@melkki.cs.helsinki.fi) | 14:53 | |
| minute | josch: if you monitor this file, say with watch -n1 cat ... | 14:57 |
| minute | josch: does the content change when you plug/unplug a monitor? | 14:57 |
| minute | ah wait, that's just pinctrl | 14:57 |
| + chomwitt (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 15:08 | |
| josch | upgrading sway on bookworm to sway from unstable triggers the problem | 15:10 |
| minute | aha | 15:10 |
| minute | so probably wlroots changed their way of detecting if a display is connected? | 15:11 |
| josch | i'll now see how hard it is to compile and run sway from git on bookworm to make sure that it's really sway and not one of the libraries it depends on (which of course also got upgraded to their version in unstable) | 15:11 |
| minute | my hunch is this would be handled in wlroots, but i might be wrong | 15:11 |
| josch | yes, wlroots 0.15.1 got upgraded to 0.18.2 | 15:14 |
| josch | thank you for that hunch, i can try that next | 15:14 |
| minute | wow, quite a big version jump | 15:16 |
| josch | modern sway 1.10 needs wlroots 0.18 so to test that theory i'm building old sway 1.7 against new wlroots 0.18 and hope that there are no API breaks... | 15:24 |
| josch | nope, not possible | 15:28 |
| josch | meson.build requires a specific wlroots version | 15:29 |
| bluerise | https://github.com/torvalds/linux/blob/master/arch/arm64/boot/dts/rockchip/rk3588-mnt-reform2.dts | 15:30 |
| bluerise | \o/ | 15:30 |
| josch | nice!! this means that we can soon have https://source.mnt.re/reform/reform-debian-packages/-/merge_requests/89 | 15:32 |
| - gsora (QUIT: Ping timeout: 252 seconds) (~gsora@user/gsora) | 15:32 | |
| abortretryfail | Does anyone remember where the patches were to try and fix HDMI on the iMX8MQ Reform? Searching has got me nowhere. | 15:37 |
| josch | abortretryfail: sure! This restores HDMI for me on imx8mq classic reform: https://source.mnt.re/reform/reform-debian-packages/-/merge_requests/80 | 15:40 |
| + gsora (~gsora@user/gsora) | 15:42 | |
| abortretryfail | Thanks josch, I'll toss that on mine today. | 15:46 |
| josch | abortretryfail: thank you! If you like, please leave a comment on that MR with your results. | 15:50 |
| - chomwitt (QUIT: Ping timeout: 246 seconds) (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 16:16 | |
| + ericsfraga (~user@2a00:23cc:b452:2900::99e) | 16:54 | |
| minute | there's a bug in recent firmware builds for pre-2.5 motherboards: it never enables charging on those board versions | 17:02 |
| minute | (classic reform) | 17:02 |
| minute | josch: you recently changed something about the mnt logo font in waybar right? moved from css to font tag? it doesn't actually seem to work in the current system image | 17:23 |
| minute | (now the logo is gone) | 17:23 |
| - Ar|stote|is (QUIT: Ping timeout: 272 seconds) (~linx@149.210.25.158) | 17:30 | |
| - ericsfraga (QUIT: Quit: ERC 5.6.1-git (IRC client for GNU Emacs 31.0.50)) (~user@2a00:23cc:b452:2900::99e) | 17:33 | |
| abortretryfail | josch: I realize now I already tested that patch a while back. It was commented out in my apt sources. | 17:36 |
| + Ar|stote|is (~linx@149.210.24.31) | 17:39 | |
| - L29Ah (QUIT: Ping timeout: 252 seconds) (~L29Ah@wikipedia/L29Ah) | 17:42 | |
| + L29Ah (~L29Ah@wikipedia/L29Ah) | 17:43 | |
| josch | abortretryfail: lol nice :d | 17:49 |
| josch | minute: yes, i changed it but that change is only in git | 17:49 |
| abortretryfail | well, it would have been nice if I'd documented anything about it (or how to do it again) | 17:49 |
| josch | minute: if it shows up in the latest system image, let me quickly repro this | 17:50 |
| josch | (to confirm that my fix in git really fixes this) | 17:50 |
| josch | errr... | 17:50 |
| josch | i meant to say "if the bug shows up in..." not "if the logo shows up in..." -- the latter would not be the bug showing up as the bug showing up is the logo not showing up :D | 17:51 |
| minute | josch: :D | 17:52 |
| minute | josch: ah, you say "only in git" as in not merged? | 17:52 |
| minute | then the redirect of mnt.re/system-image still picks the wrong pipeline | 17:53 |
| josch | minute: no, we changed that | 17:53 |
| minute | what we did is download via that link | 17:53 |
| josch | minute: reform-tools is no longer tightly integrated | 17:53 |
| minute | ah | 17:53 |
| josch | it is now a normal upstream project that can be used by any distro | 17:53 |
| josch | this also means that the whole business with only being allowed to push to the staging branch is gone | 17:54 |
| minute | ok | 17:54 |
| josch | it's now normal MRs that get merged into main | 17:54 |
| josch | and then at some point i run "git tag" | 17:54 |
| josch | and then i update the debian packaging | 17:54 |
| josch | since the logo issue is quite bad, let me quickly confirm that reform-tools in git actually fixes the problem | 17:54 |
| josch | and if yes, i tag a new version | 17:54 |
| josch | and then i package that in debian unstable | 17:54 |
| minute | josch: btw the issue is that the logo is just completely invisible, not a white M this time | 17:55 |
| minute | josch: the font is installed correctly on the machine, i checked with font-viewer | 17:55 |
| josch | what a mess... i'll check whether the <span font=""> change adresses this | 17:56 |
| minute | the span font="" change was in there | 17:56 |
| minute | i checked .config/waybar/config | 17:56 |
| - jacobk (QUIT: Ping timeout: 248 seconds) (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 17:56 | |
| minute | do you have a link to your change commit? | 17:56 |
| josch | oh | 17:56 |
| josch | sorry | 17:56 |
| josch | reform-system-image commit 38a65433a06585ce9d4bec6cf2b5cb6ea6646643 | 17:57 |
| josch | sorry, i was already too far ahead -- i am integrating the /etc directory into reform-tools but that is not done yet | 17:57 |
| josch | so yes, this is not a reform-tools issue | 17:57 |
| josch | oh shoot i see the problem | 17:57 |
| minute | i see | 17:57 |
| minute | also, font-family: MNTIcons, Inter; was there before but now there's only MNTIcons in the font tag | 17:58 |
| josch | sorry for the mess, fixing it now... | 17:58 |
| minute | thanks! | 17:58 |
| josch | minute: i also wrote up my status re a311d hdmi bug in https://source.mnt.re/bugs/bugs/-/issues/19 | 17:58 |
| minute | ah yeah i see now | 17:58 |
| minute | span vs font :D | 17:58 |
| + chomwitt (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 17:58 | |
| josch | minute: it's tricky for me to continue debugging this because recent sway does not build on bookworm, so i'd have to do this from an installation on sd-card which let me not use my laptop for other things during that time | 17:59 |
| josch | minute: yes, exactly | 17:59 |
| josch | apparently i copypasted something that i failed to test properly | 17:59 |
| josch | at least you spotted it quickly, thanks | 17:59 |
| josch | the web says that it should be <span style="font-family: MNTIcons"> -- testing it now... | 18:00 |
| minute | josch: ah, but how would that be different from having it in the CSS? | 18:01 |
| josch | minute: the problem with the css is, that it relies on the style name never changing | 18:01 |
| minute | josch: i see | 18:01 |
| minute | ok | 18:01 |
| josch | by putting it in a <span> this becomes past and future proof (past proof is needed because i compile this for bookworm on reform.d.n) | 18:02 |
| minute | ah right, ok | 18:02 |
| josch | image is building... | 18:05 |
| - chomwitt (QUIT: Ping timeout: 245 seconds) (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 18:17 | |
| + chomwitt (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 18:26 | |
| - chomwitt (QUIT: Ping timeout: 268 seconds) (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 18:42 | |
| minute | josch: it seems like MMC debugging is on on our current a311d uboot | 18:50 |
| abortretryfail | josch: I'm not 100% sure I got it installed right, but the kernel from !80 detects HDMI monitors, but Sway can't use the output. | 18:52 |
| abortretryfail | I vaguely remember noticing this before, but the built-in panel is still running at 70hz, and I think that needs to be lower for HDMI to work. | 18:53 |
| minute | josch: this is slowing down uboot a bit on a311d | 18:57 |
| minute | josch: when running "reform-boot-config ssd" on a311d pocket, it errors with reform-boot-config: 132: EMMC: parameter not set | 19:00 |
| gustav28 | huh re rk3588-mnt-reform2.dts: does linux include files like these for *all* computers supporting linux officially? seems like there must be a better way than mainlining configs like this? | 19:01 |
| gustav28 | also ordered a mnt reform classic yesterday, excited :) | 19:01 |
| abortretryfail | gustav28: yeah, kind of. ARM is a bit of a mess. | 19:04 |
| abortretryfail | It's like this with SBCs and phones too | 19:05 |
| gustav28 | :0 | 19:05 |
| abortretryfail | Some of them use overlays though, and idk if any of the MNT stuff does. | 19:05 |
| minute | haha it's so hard to change from boot via emmc to boot via sd card -> ssd | 19:08 |
| minute | like, if you're on sd card, you can't use "reform-boot-config ssd" | 19:12 |
| minute | ah wait. you can | 19:12 |
| josch | minute: thank you, the reform-boot-config problem is a regression from 3a3305340a6a5dcf9d4d7fa56aa1e4aa3b82ede1 | 19:21 |
| josch | it also turns out that i'm stupid: pango is not html | 19:22 |
| josch | the correct markup is: "<span face='MNTIcons'>M</span>" | 19:22 |
| josch | i tested that with reform-branding not installed and then you just get a normal "M" | 19:23 |
| josch | which explains why i wrote the current version wrong: it's not <span font=... it's <span face=... | 19:24 |
| josch | the EMMC problem should be fixed in 55bf7b37dc350098f17464585ee1b547c028f8d2 | 19:27 |
| josch | minute: i never noticed anything "slow" about a311d u-boot but nice you found it! :) | 19:40 |
| josch | minute: the EMMC problem i think is severe enough such that we can cut a new reform-tools release today after all | 19:40 |
| josch | that will become 1.70 | 19:41 |
| josch | abortretryfail: did you try with another monitor? i found that hdmi is sometimes picky... | 19:41 |
| josch | minute: when you said "it's so hard to..." you figured out it was actually not -- maybe something in the documentation can be improved so that one does not get stuck like you just did? | 19:43 |
| abortretryfail | hmm, no. I can try with the TV and see if that works. | 19:46 |
| abortretryfail | josch: that was it. Reform's HDMI must not like that Dell monitor. It's OK with a Sony TV. | 19:50 |
| - andypiper_ (QUIT: Ping timeout: 260 seconds) (~Srain@45.146.11.160) | 19:50 | |
| minute | josch: ok the story was: wakest wanted to move their system to nvme on an a311d pocket that was running quite slowly from a 90% full microsd. so i said, use reform-setup-encrypted-disk. which worked well. but they said yes to the emmc boot option. the script then asked if they wanted to reboot into their system (which i didn't know was a feature). then it rebooted, and the microsd was still | 19:51 |
| minute | inserted. so it booted from the microsd. they thought it was the nvme system. they pulled out the microsd while the system was running. the system went into panic mode, icons stopped rendering, commands were not found anymore | 19:51 |
| minute | josch: then i said well, lets try to boot from emmc just as you selected. but it didn't work. connected debugger, ran into the a311d uboot bug about couldn't select voltage, -110. | 19:51 |
| minute | josch: so i was like ok, lets figure out how to boot via the microsd to your new encrypted nvme instead. | 19:52 |
| josch | oh wow, that's yet another issue in reform-tools we can fix then | 19:52 |
| minute | josch: so we ran reform-boot-config ssd and had to work around the EMMC not defined bug. then it said, can't find /dev/nvme0n1p1 | 19:53 |
| minute | josch: which is correct because that doesn't exist if you do reform-setup-encrypted-nvme, which sets up a VG on the whole disk, without a partition | 19:53 |
| minute | josch: so i tried to figure out how to mount this encrypted nvme in the correct way (had to google a bit) | 19:53 |
| minute | josch: so i did cryptsetup open /dev/nvme0n1 crypt, and then vgchange -ay, and then did reform-boot-config /dev/...something-vg-root (forgot), which worked. | 19:54 |
| minute | josch: wait. it warned me about /dev/mmcblk0p1 still being mounted somewhere. | 19:54 |
| josch | oh you tried to use reform-boot-config from a system running from sd-card *for* an encrypted installation on nvme? | 19:55 |
| minute | josch: so i was like, damn. it looks like we can't do that from the SD card then. because i lapsed and forgot that /dev/mmcblk0p1 is just /boot. there it would have been extra helpful if the script said "/boot" instead of "somewhere" | 19:55 |
| minute | josch: correct. it was the only option we had i think | 19:55 |
| josch | okay | 19:55 |
| minute | josch: anyway, after realizing that, i tried again and pointed it to the unlocked vg root partition, and it worked. but! | 19:56 |
| josch | i already have one other FIXME items from a few days ago for reform-boot-config which is to create the partition on emmc if it does not exist yet | 19:56 |
| minute | josch: i didn't know/remember that the expected name is "reform_crypt" | 19:56 |
| josch | but where should it say "/boot" instead of "somewhere"? | 19:56 |
| josch | yes, maybe reform-boot-config should support the lvm/luks layout created by reform-setup-encrypted-disk out-of-the-box | 19:57 |
| minute | josch: so there were some errors scrolling by about: warning, crypt is not in crypttab (i think). because i was afraid of messing it up, i did the whole process again with reform_crypt as the name for the initial cryptsetup open | 19:57 |
| minute | josch: and it worked. but it was quite challenging to get right | 19:57 |
| minute | josch: yeah, it would be very nice if those were compatible | 19:57 |
| minute | josch: > echo "/dev/$BOOTPART is still mounted somewhere, which means that it is" >&2 | 19:58 |
| minute | josch: the variable is named BOOTPART but when the script runs this information ("boot") is gone | 19:58 |
| josch | minute: okay, can you try and give me a list of actionables to make sure i don't miss things? | 19:59 |
| josch | minute: the "somewhere" message is from reform-migrate though, no? | 19:59 |
| minute | josch: no i grepped "somewhere" in /usr/bin/reform-boot-config | 19:59 |
| minute | josch: it's of course possible that it's also in reform-migrate | 20:00 |
| josch | ah i see it | 20:00 |
| josch | yes, needs to be fixed in both | 20:00 |
| minute | ok i can make a quick list in a new issue in bugs | 20:00 |
| josch | that'd be helpful thank you | 20:00 |
| minute | btw we realized that any member of source.mnt.re can create issues in source.mnt.re/bugs/bugs (mnt.re/bugs) | 20:00 |
| minute | (which is good) | 20:00 |
| josch | i think i like the common bug list | 20:01 |
| josch | i got reminded of this thread: https://community.mnt.re/t/collecting-long-standing-reform-issues/2125 | 20:01 |
| josch | and it was not clear back then where to file those | 20:01 |
| josch | so some are filed in /reform/reform, some in /reform/reform-debian-packages... | 20:01 |
| josch | thank you for creating a quick list -- we're eating dinner now, then bringing baby to bed etc and i fear i'll have forgotten half the things in a few hours. I'll re-read your last messages in the backlog if i need more details but those were quite a few items that can be improved | 20:04 |
| josch | (which is good!) | 20:04 |
| minute | josch: awesome. i also put the whole story in the issue | 20:05 |
| minute | ok, issue filed https://source.mnt.re/bugs/bugs/-/issues/24 | 20:08 |
| josch | minute: perfect! love the checklist :) | 20:28 |
| minute | nice! | 20:33 |
| minute | trying a monolithic kernel build for debugging r/n from reform-debian-packages linux and my special monolithic config | 20:44 |
| minute | > drivers/pci/controller/dwc/pcie-dw-rockchip.c:238:33: error: ‘PCIE_TYPE0_HDR_DBI2_OFFSET’ undeclared | 20:44 |
| minute | ah lol, that's my own experimental patch i guess | 20:46 |
| + chomwitt (~chomwitt@2a02:587:7a07:1100:1ac0:4dff:fedb:a3f1) | 21:35 | |
| minute | hmm this is going quite badly, rcu stalls | 21:49 |
| minute | > [ 58.511722] [ebf2bb3c66fc6b55] address between user and kernel address ranges | 22:04 |
| minute | i guess something in the dsi or sn65 driver badly corrupts memory | 22:06 |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-78-82-54-216.bbcust.telenor.se) | 22:15 | |
| + mark_ (~mjw@gnu.wildebeest.org) | 22:23 | |
| - mjw (QUIT: Killed (tantalum.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c) | 22:25 | |
| * mark_ -> mjw | 22:25 | |
| + Guest3974 (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c) | 22:25 | |
| + erle (~erle@user/erle) | 23:04 | |
| - erle (QUIT: Client Quit) (~erle@user/erle) | 23:07 | |
| josch | minute: i think i have implemented all the points except for one (but didn't test yet). I have a question about two of the points for which I cannot figure out the answer from what you wrote: | 23:52 |
| josch | a) "the script then asked if they wanted to reboot into their system (which i didn't know was a feature). then it rebooted" -- as far as I can see, the scripts say: "Restart MNT Reform (type: reboot) after saving your work to activate the changes." -- they do not reboot themselves but they do tell the user to reboot now. That message should be improved if the current system is on sd-card and the new | 23:54 |
| josch | /boot on emmc, yes. I just want to make sure that i didn't miss the place where one of our scripts *really* offered to reboot and then did it? | 23:54 |
| josch | b) reform-setup-encrypted-disk has a big warning if EMMC_USE is set to "warn" (which is the case for a311d): https://source.mnt.re/reform/reform-tools/-/blob/main/bin/reform-setup-encrypted-disk?ref_type=heads#L140 Did you find a code path where the warning is missing? | 23:56 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!