| - jacobk (QUIT: Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) (~quassel@2600:1700:9e1e:7800:c40c:5464:17f4:498c) | 00:03 | |
| jfred | Just got myself a 100W portable solar panel, excited and nervous to try it out with my Reform haha | 00:53 |
|---|---|---|
| noam | got it! :DF | 00:54 |
| noam | :D :D :D | 00:54 |
| jfred | Looks like the stock barrel connector on the cable it comes with is the right one so don't even need any adapters | 00:55 |
| noam | My reform is booting p9 again :D | 00:55 |
| noam | I didn't even have to disassemble it | 00:55 |
| jfred | Woo! :D | 00:55 |
| noam | and it looks like the u-boot was the issue, since a stock kernel works now :P | 00:55 |
| noam | Used a USB port splitter, my PC's keyboard, and my modified Debian image; `reform-flash-uboot --offline emmc`, reboot, done | 00:57 |
| noam | Currently booting stock 9front off of SD (well, okay, slightly patched 9front since it's the image I was using for debugging :P), but that should be enough for me to mount the encrypted NVMe, and use that to bootstrap the eMMC further ahead :P | 00:58 |
| noam | lol no | 01:11 |
| - chomwitt (QUIT: Ping timeout: 255 seconds) (~chomwitt@2a02:587:7a10:cf00:1ac0:4dff:fedb:a3f1) | 01:16 | |
| + murphnj (~murph@user/murphnj) | 01:19 | |
| - mjw (QUIT: Ping timeout: 272 seconds) (~mjw@gnu.wildebeest.org) | 01:36 | |
| - digitalrayne (QUIT: Ping timeout: 240 seconds) (~digitalra@vps-446f4f39.vps.ovh.ca) | 04:22 | |
| + digitalrayne (~digitalra@vps-446f4f39.vps.ovh.ca) | 05:13 | |
| - hairu (QUIT: Remote host closed the connection) (m-uotkmd@user/hairu) | 06:43 | |
| + hairu (m-uotkmd@user/hairu) | 06:50 | |
| - hairu (QUIT: Remote host closed the connection) (m-uotkmd@user/hairu) | 07:26 | |
| + hairu (m-uotkmd@user/hairu) | 07:28 | |
| - GNUmoon2 (QUIT: Ping timeout: 260 seconds) (~GNUmoon@gateway/tor-sasl/gnumoon) | 07:48 | |
| + jaume (~user@user/jaume) | 07:56 | |
| + GNUmoon2 (~GNUmoon@gateway/tor-sasl/gnumoon) | 08:01 | |
| vkoskiv | jfred: Are you just hooking it up directly, or is there some hardware between the panel and the Reform? | 09:19 |
| vkoskiv | I've been curious about solar charging myself, too :D | 09:19 |
| vkoskiv | Not that I have much use for it, I live very much in the city, but it'd be a neat curiosity. | 09:19 |
| josch | vkoskiv: there are some in this channel who, in the past have reported that they have directly hooked it up without anything between the panel and the reform | 09:21 |
| noam | I wonder how hard it would be to design a custom bottom that gives easier access to the cells >_> | 09:21 |
| josch | noam: what do you want to have access to the cells for? | 09:21 |
| vkoskiv | Yeah, I recall something like that was discussed at some point. | 09:22 |
| josch | should be easy to find in the logs :) | 09:22 |
| vkoskiv | Yep! I have a local copy I can grep | 09:22 |
| josch | who doesn't :) | 09:22 |
| josch | wget -r FTW | 09:23 |
| vkoskiv | I've started building little datasets I can browse easily with fzf. I tried building a component database I could quickly look up part numbers in, to determine what that component does, for instance. | 09:23 |
| vkoskiv | So my dataset is just 'partnumber|short description|PDF link', and when I hit return in fzf after finding the thing, it brings that PDF right up. | 09:24 |
| vkoskiv | It's very nice, but I only managed to scrape 1.7 million part numbers for this list, so it's not very comprehensive. | 09:25 |
| vkoskiv | alldatasheets claims they have >50 million | 09:25 |
| vkoskiv | I don't want the PDFs archived, those I can get ondemand, I just want to remove the annoying "google for PN and click through to the thing to even find out what it is" part of the cycle | 09:26 |
| + Manis (01a66df340@77-56-188-94.dclient.hispeed.ch) | 10:08 | |
| - nybble (QUIT: Ping timeout: 240 seconds) (~sprang@user/csprng) | 10:20 | |
| + mjw (~mjw@gnu.wildebeest.org) | 11:42 | |
| minute | lol the copper heatsinks are _heavy_ https://mastodon.social/@mntmn/112591734408018823 | 11:59 |
| noam | josch: swapping them while it's live :P | 12:00 |
| josch | wow o0 | 12:01 |
| noam | minute: that means they're denser, in principle, and thus should have higher conductivity? | 12:01 |
| vkoskiv | That copper looks so nice | 12:01 |
| minute | noam: yeah makes sense :D | 12:01 |
| josch | i does! please leave it bare without coating? :) | 12:01 |
| noam | josch: in principle, it'd be cool to flip a switch so that it's running only from one set of four cells and interrupts the other one? | 12:01 |
| noam | and then swap the other set | 12:02 |
| minute | vkoskiv: yeah it has an interesting surface texture/color irl | 12:02 |
| noam | That'd probably need more modding :/ | 12:02 |
| minute | it's more pinkish and less orange than i expected | 12:02 |
| noam | Oh! Is the person who swapped sodium-ion cells in on IRC? | 12:02 |
| noam | I'm contemplating getting a set, though I'd want to patch the system controller so it's not permanently burning some of the capacity away on the first charge cycle :P | 12:03 |
| minute | josch: yeah might need some tape on the bottom though to not accidentally short any caps | 12:03 |
| josch | is shorting a potential problem with the existing heatsinks? | 12:03 |
| minute | josch: i don't think so, because of the geometry which has a protruding square on the bottom and hardcoat anodizing | 12:05 |
| minute | (well only V1 heatsink has the square, generic doesn't) | 12:06 |
| josch | ah okay, i wasn't aware that the coating made the surface non-conductive | 12:06 |
| noam | Also, for the record: https://hakadibattery.com/products/hakadi-18650-3-2v-1800mah-lifepo4-rechargeable-battery-cell-cycle-life-3000-for-diy-battery-pack-flashlight?variant=41712175841485 <- they look to actually have LFP 18650s in stock :D | 12:09 |
| + chomwitt (~chomwitt@2a02:587:7a10:cf00:1ac0:4dff:fedb:a3f1) | 12:09 | |
| noam | I don't need any right now, but I was struggling to find some last time I looked, so I figured I'd share the link in case anyone else needs :) | 12:09 |
| noam | and by "in stock" I mean "they have a 'buy 300' option" :P | 12:10 |
| - chomwitt (QUIT: Ping timeout: 268 seconds) (~chomwitt@2a02:587:7a10:cf00:1ac0:4dff:fedb:a3f1) | 12:30 | |
| josch | sir-photch (sitting behind me) just recommended this scrollable-tiling wayland compositor to me which looks quite lean and neat! https://github.com/YaLTeR/niri | 13:07 |
| - robin (QUIT: Ping timeout: 268 seconds) (~robin@user/terpri) | 13:23 | |
| jfred | josch: Ooh! I used PaperWM a while back but it always felt a bit hacky as a Gnome Shell extension - might have to give this a try | 14:06 |
| jfred | vkoskiv: Yeah the plan was to hook it up directly. The panel outputs 12-18V which *should* all be within spec for the Reform... I haven't actually tried it yet though 😅 | 14:10 |
| + robin (~robin@user/terpri) | 14:37 | |
| + andreas-e (~Andreas@2001:861:c4:f2f0::c64) | 14:45 | |
| AbortRetryFail | was there a guide for moving the Reform off of Debian sid somewhere? I'm about done with its antics | 15:02 |
| AbortRetryFail | https://community.mnt.re/t/strategy-for-apt-upgrade/2010/10 This thread seems to suggest downgrading isn't an option, dang. | 15:08 |
| Twodisbetter | Yes Josch and I are on stable although the system still reports SID. | 15:35 |
| Twodisbetter | It is not super recommended from what I recall. I have not, as of yet, had any issues though. | 15:36 |
| - mjw (QUIT: Killed (mercury.libera.chat (Nickname regained by services))) (~mjw@gnu.wildebeest.org) | 15:38 | |
| * Guest9859 -> mjw | 15:38 | |
| + Guest7546 (~mjw@gnu.wildebeest.org) | 15:38 | |
| AbortRetryFail | when was the last time you updated? mine is going berserk right now | 15:43 |
| AbortRetryFail | installed firefox-esr and its removing dependencies from underneath xwayland. | 15:43 |
| josch | Twodisbetter: where does your system report sid? | 16:13 |
| josch | AbortRetryFail: what are you attempting to do? | 16:13 |
| Twodisbetter | josch: neofetch | 16:16 |
| josch | Twodisbetter: i have no idea where neofetch gets that information from. What does your /etc/os-release say? | 16:17 |
| - ethulhu (QUIT: Quit: connection reset by purr) (ethulhu@nora.ethulhu.co.uk) | 17:03 | |
| minute | ok, now looking into the pocket reform migrate-to-ssd issue | 17:36 |
| josch | nice! | 17:42 |
| josch | maybe it helps to boot with a root shell in the initramfs? | 17:42 |
| josch | i.e. init=/bin/sh and see what the state of the system is | 17:43 |
| minute | well, first i have to move my system to ssd. | 17:45 |
| minute | installed the latest reform-tools for that | 17:47 |
| minute | update-initramfs is so slow aaaa | 17:48 |
| minute | ok finally | 17:48 |
| minute | > The encrypted NVMe is now set up. Do you want me to run reform-migrate now as well? | 17:51 |
| minute | y | 17:51 |
| minute | nice, 136MB/s | 17:51 |
| josch | initramfs devs were open to talks about improvements re progress indicator, so once me or somebody else finds the free time... | 17:51 |
| AbortRetryFail | josch: all i am attempting to do is stop apt from removing apps I use. | 17:52 |
| - robin (QUIT: Remote host closed the connection) (~robin@user/terpri) | 17:53 | |
| + robin (~robin@user/terpri) | 17:53 | |
| josch | AbortRetryFail: i wrote you on the forum how we can start debugging this | 17:54 |
| josch | AbortRetryFail: the important thing is, that you get into a stable state first | 17:54 |
| josch | that is a state where "apt install -f" exits successfully | 17:54 |
| AbortRetryFail | it's never exited unsuccessfully so far. If I tell it to install something, it does do that. | 17:55 |
| minute | haha the longest part is at 98%+ | 17:55 |
| josch | AbortRetryFail: just send me the output of the command i sent you in the forum then | 17:56 |
| AbortRetryFail | okay, waiting for that page to load rn | 17:56 |
| josch | minute: with rsync? | 17:56 |
| minute | josch: yeah... | 17:57 |
| minute | oooops thermal shutdown | 17:57 |
| minute | maybe not a good idea to do this without the heatsink | 17:57 |
| minute | but hey, at least thermal shutdown works cleanly | 17:58 |
| josch | maybe somebody knows an alternative to rsync (or a way to do this with rsync?) but for a robust progress indicator, rsync would first have to build the list of stuff to be copied so that it knows where 100% is | 17:58 |
| josch | there is a FIXME comment in reform-migrate which has the idea of maybe encoding the number of files and their overall size in the system image | 17:59 |
| josch | and put those values somewhere in /etc so that reform-migrate can tell the user: this is 100% | 17:59 |
| AbortRetryFail | josch: all 3k lines of it https://paste.debian.net/hidden/227d222f/ | 17:59 |
| josch | AbortRetryFail: can you also show me the output of "apt-cache policy"? | 18:00 |
| minute | josch: yeah it's not a biggie, it takes long for me because i already have 28GB of data on emmc | 18:01 |
| minute | i don't _actually_ want to migrate my system... i prefer the eMMC | 18:01 |
| minute | but i wanna solve the issue | 18:01 |
| AbortRetryFail | josch: apt-cache policy https://paste.debian.net/hidden/ea16544d/ | 18:01 |
| AbortRetryFail | I wish it would tell me why it thinks it needs to remove anything when all I have told it to do is update or install | 18:02 |
| josch | AbortRetryFail: we can also go the other way round and you send me an output of apt with debug flags on | 18:02 |
| josch | AbortRetryFail: can you run the apt install command that causes you problems with the following additional flags: | 18:04 |
| josch | AbortRetryFail: -oDebug::pkgProblemResolver=true -oDebug::pkgDepCache::Marker=1 -oDebug::pkgDepCache::AutoInstall=1 | 18:05 |
| AbortRetryFail | for clarity: apt install never fails - it always installs the thing I asked it to | 18:06 |
| josch | then i misunderstood the problem | 18:07 |
| josch | can you show me an example? | 18:07 |
| AbortRetryFail | sure, here's apt install -f https://paste.debian.net/hidden/423ef9fd/ | 18:08 |
| AbortRetryFail | all that stuff it's suggesting to autoremove? Deps for things I have installed. No idea why it thinks those aren't necessary anymore. | 18:09 |
| josch | oooooooooooooooh! :D | 18:09 |
| josch | because they are marked as automatic | 18:09 |
| josch | that's why they are listed | 18:09 |
| AbortRetryFail | Started when I was trying to do some stuff in FreeCAD, installed OpenSCAD, then it did crap like removed firefox, telegram, freecad itself, etc. | 18:09 |
| josch | apt is not removing things but it tells you that it would autoremove them | 18:09 |
| AbortRetryFail | I installed freecad again and apt wanted to remove kicad | 18:09 |
| josch | but kicad is not in your list | 18:10 |
| AbortRetryFail | Because I already reinstalled it manually | 18:10 |
| josch | tell me a package in the list you'd like to not autoremove because you need it | 18:10 |
| josch | AbortRetryFail: apt is not removing those packages -- it just tells you that it would remove them if you run "apt autoremove" | 18:11 |
| josch | see the last line in the output | 18:11 |
| josch | we can try figuring out what happened if you like | 18:11 |
| AbortRetryFail | linphone-common is an example. I had linphone-desktop installed and it removed it for whatever reason | 18:12 |
| AbortRetryFail | so now it thinks the linphone-common package is no longer needed. | 18:12 |
| josch | but it is no longer needed because you do not have linphone installed anymore | 18:12 |
| AbortRetryFail | Sure, I get that part. :) | 18:12 |
| josch | if linphone-common was only installed as a dependency of linphone, then if you remove linphone, it will have linphone-common marked for autoremoval | 18:13 |
| josch | so far, what you experience is what i consider normal | 18:13 |
| josch | if you want to mark packages in that list as "please keep this one" then you can run: | 18:13 |
| josch | sudo apt-mark manual mypackage | 18:14 |
| AbortRetryFail | Right, I know that too. Or even just `apt install mypackage` will mark it as manually installed. | 18:14 |
| josch | see the man page of "apt-mark" the entry for "manual": "manual is used to mark a package as being manually installed, which will prevent the package from being automatically removed if no other packages depend on it." | 18:14 |
| josch | yes | 18:14 |
| josch | then i still do not understand the problem :D | 18:14 |
| AbortRetryFail | maybe i need to paste all of /var/log/apt/history.log | 18:15 |
| josch | you could but that could contain very personal information | 18:15 |
| josch | depending on what you consider personal | 18:15 |
| josch | your autoremoval list looks like you at some point had something installed that needed a lot of Qt and opencv and then removed that application | 18:15 |
| AbortRetryFail | The problems basically all started yesterday evening when i did `apt install openscad -y` | 18:16 |
| AbortRetryFail | but something about the state of things made it remove stuff, which I didn't notice until this morning, when telegram didn't work. | 18:16 |
| josch | okay, you can find out what got removed by looking at /var/log/apt/history.log | 18:16 |
| AbortRetryFail | Right. The problem I am trying to solve is not what got removed, but *why*? | 18:17 |
| josch | you probably told it to install something which conflicted with something else | 18:18 |
| AbortRetryFail | If I tell it to install openscad, why does it remove kde-baseapps and a zillion other things? | 18:18 |
| josch | if this happens when you try installing openscad, show me the output with the debugging flags i pasted above | 18:19 |
| AbortRetryFail | openscad is already installed | 18:19 |
| AbortRetryFail | and the other things were already uninstalled. None of that *failed* | 18:19 |
| josch | i think i get confused by your use of tenses | 18:19 |
| AbortRetryFail | maybe. | 18:20 |
| josch | you said "if i tell it to install openscad" and i think what you meant is "yesterday when i told it to install openscad"? | 18:20 |
| AbortRetryFail | yeah that's right, sorry. | 18:20 |
| josch | maybe history.log can tell you more | 18:20 |
| AbortRetryFail | history.log tells me what was done, not why | 18:20 |
| josch | can you show me? | 18:20 |
| josch | please carefully review what apt is about to do before telling it to do it :) | 18:21 |
| AbortRetryFail | Right, the mistake was -y | 18:21 |
| AbortRetryFail | I get that | 18:21 |
| Twodisbetter | josch: from neofetch on my Reform: | 18:22 |
| Twodisbetter | _,met$$$$$gg. committed88@reform | 18:22 |
| Twodisbetter | ,g$$$$$$$$$$$$$$$P. ------------------ | 18:22 |
| Twodisbetter | ,g$$P" """Y$$.". OS: Debian GNU/Linux trixie/sid aarch64 | 18:22 |
| Twodisbetter | ,$$P' `$$$. Host: MNT Reform 2 | 18:22 |
| Twodisbetter | ',$$P ,ggs. `$$b: Kernel: 6.6.13+bpo-reform2-arm64 | 18:22 |
| Twodisbetter | `d$$' ,$P"' . $$$ Uptime: 2 days, 18 hours, 36 mins | 18:22 |
| Twodisbetter | $$P d$' , $$P Packages: 2245 (dpkg), 18 (flatpak) | 18:22 |
| Twodisbetter | $$: $$. - ,d$$' Shell: bash 5.2.15 | 18:22 |
| Twodisbetter | $$; Y$b._ _,d$P' Resolution: 1920x1080 | 18:22 |
| Twodisbetter | Y$$. `.`"Y$$$$P"' WM: sway | 18:22 |
| Twodisbetter | `$$b "-.__ Theme: Arc-Dark [GTK3] | 18:22 |
| Twodisbetter | `Y$$ Icons: Adwaita [GTK3] | 18:22 |
| AbortRetryFail | Even if I do review it. If what it says its going to do makes no sense I still don't know why it decides to do that or how to fix it. :) | 18:22 |
| josch | AbortRetryFail: you can get to know the "why" by returning to the state from yesterday (for example by using a backup) or by hand-installing some packages that got removed (according to history.log) and then trying to install openscad again but with the debugging flags | 18:23 |
| Twodisbetter | my os-release file says the exact same thing as neofetch reports. If I were to change that, would neofetch reflect it? | 18:24 |
| AbortRetryFail | Yeah, I'm just hand installing the packages at this point. I don't have backups of this machine since its not that important. | 18:24 |
| josch | Twodisbetter: please use a pastebin if you want to show more than 3 lines and what neofetch displays is much less important than what is in your /etc/os-release | 18:24 |
| josch | Twodisbetter: do not edit /etc/os-release | 18:24 |
| Twodisbetter | josch: roger dodger | 18:24 |
| Twodisbetter | will do | 18:24 |
| josch | Twodisbetter: this indicates that you have a system with mixed stable and unstable | 18:24 |
| Twodisbetter | josch: safe way to make it only stable? | 18:24 |
| Twodisbetter | or is that the dangerous thing that might break things, that are beyond what stable provides? | 18:25 |
| josch | AbortRetryFail: you can retrieve the *why* even after apt already finished by reading /var/log/apt/eipp.log.xz but that only stores the *last* apt run | 18:25 |
| josch | Twodisbetter: i'd only attempt a downgrade if you have backups | 18:26 |
| Twodisbetter | josch: but if I just leave it as is, am I fine? | 18:26 |
| Twodisbetter | does the kernel I'm using match what stable is using? | 18:26 |
| AbortRetryFail | hmm, okay | 18:27 |
| josch | Twodisbetter: your kernel seems to be from Debian backports, not stable | 18:27 |
| - khm (PART: .) (~kfx@216.126.196.60) | 18:28 | |
| AbortRetryFail | I did apt update again and now it seems to think I can upgrade 349 packages. I wonder if I had an unlucky roll and got some inconsistency from a mirror earlier. | 18:28 |
| josch | maybe -- you are using unstable after all | 18:29 |
| josch | this year we had a ton of churn in unstable | 18:29 |
| josch | because we are performing 64bit time_t and /usr-move at the same time | 18:29 |
| AbortRetryFail | Yeah, kind of why I'm wondering what the unstable -> stable migration path looks like. | 18:29 |
| josch | the migration path is: switch to testing first and then wait for the next stable release | 18:30 |
| AbortRetryFail | That seems reasonable. :) | 18:30 |
| josch | using testing instead of unstable will already make things quite a bit more sane, i think | 18:31 |
| josch | the only problem with using testing is, that occasionally, you'll be out-of-sync with the MNT mirrors for a while | 18:31 |
| josch | but these days, we only patch flash-kernel, so that should not be too bad | 18:31 |
| josch | AbortRetryFail: are things better now? | 18:35 |
| AbortRetryFail | I'm not sure. I removed the waydroid repo and I'm trying to do an upgrade and it has heaps of things "kept back" and doesn't say why. | 18:35 |
| AbortRetryFail | https://paste.debian.net/1319835/ | 18:36 |
| josch | an upgrade keeps things back that would result in additional packages getting installed | 18:36 |
| josch | some of those are just due to library soname bumps | 18:36 |
| josch | AbortRetryFail: run "apt dist-upgrade" to not have packages that are kept back | 18:37 |
| AbortRetryFail | oh, interesting | 18:37 |
| josch | AbortRetryFail: but the dist-upgrade will then likely also show you packages that get removed or installed -- review that list carefully | 18:37 |
| josch | AbortRetryFail: "apt-get upgrade" is meant for safe upgrades when you have stable installed | 18:38 |
| josch | AbortRetryFail: "apt-get dist-upgrade" is meant for... a dist upgrade :) | 18:38 |
| josch | since you are using unstable, you are going to be in the conditions for a dist upgrade all the time | 18:38 |
| AbortRetryFail | See, that's actually relevant information for the forum thread about "strategy for apt upgrade" | 18:38 |
| josch | AbortRetryFail: then maybe add it there :) | 18:39 |
| AbortRetryFail | I have no idea what I'm doing here, so telling other users what to do is off the table for me rn | 18:40 |
| AbortRetryFail | Okay, now dist-upgrade wants to remove a bunch more things I use... Still not sure why | 18:40 |
| josch | AbortRetryFail: then run with the debug flags i showed above | 18:40 |
| AbortRetryFail | trying that now. Where did you find those? | 18:42 |
| josch | AbortRetryFail: i'm a bad source for this kind of stuff as i'm a contributor to apt | 18:42 |
| josch | AbortRetryFail: so my personal answer would be: i read the source code | 18:43 |
| josch | it's probably documented somewhere | 18:43 |
| josch | generally, your questions are not about the reform but about debian | 18:43 |
| josch | so far, nothing in your questions was specific to the setup on the reform, i think | 18:44 |
| josch | so you would probably find the answers in a debian user forum | 18:44 |
| josch | minute: please stop me if i'm getting too off topic :) | 18:44 |
| AbortRetryFail | You're right. These definitely are debian sid problems | 18:44 |
| AbortRetryFail | I got the answer to my reform-specific one which is switch to testing and then stable once the next one is released. | 18:44 |
| josch | and even that is an answer you'd also get for a more vanilla debian :) | 18:45 |
| AbortRetryFail | sure, but I wouldn't install a PC with a SD card image of sid in the first place. | 18:46 |
| josch | good point! | 18:46 |
| josch | AbortRetryFail: for the faint of heart, reform.debian.net also has the usual debian-installer medium to get Debian stable on your reform :) | 18:46 |
| AbortRetryFail | I just wish it'd break when I was at home and another machine with a SD card reader is handy. Seems to only do this when I'm on the road. :) | 18:47 |
| minute | josch: all good | 18:48 |
| AbortRetryFail | this almost looks like the mirror I'm using is in some sort of flux where half the packages depend on some libraries that haven't updated and others dont | 18:49 |
| AbortRetryFail | I might just leave it alone and try again later/tomorrow. | 18:49 |
| + nybble (~sprang@user/csprng) | 18:57 | |
| minute | josch: migrating 28GB took 48 minutes so probably block copying would be more effective, but that only works if the NVMe is bigger than the eMMC or SD card of course | 19:00 |
| minute | josch: can we change this for imx8mp? > Your /boot partition will be on the SD-Card by default. Do you want it on eMMC instead? [y/N] | 19:01 |
| minute | > /dev/mmcblk2p1 is still mounted somewhere, which means that it is... | 19:01 |
| minute | > reform-boot-config FAILED to run | 19:01 |
| minute | ah, it doesn't offer to run /boot | 19:02 |
| minute | s/run/umount | 19:02 |
| minute | josch: at this point i'm helpless if i'm not a super expert. i spent 48 minutes migrating data but i don't know how to rerun reform-boot-config | 19:03 |
| minute | i.e. i don't know the right parameters for --emmc and encrypted nvme | 19:03 |
| minute | the help says > The disk has to be unlocked first with: cryptsetup luksOpen /dev/nvme0n1p1 crypt | 19:04 |
| minute | but it's actually /dev/nvme0n1 | 19:04 |
| minute | then it says to do reform-migrate /dev/mapper/crypt | 19:05 |
| minute | but that's also wrong | 19:05 |
| minute | as apparently reform-migrate made some kind of volume group | 19:05 |
| minute | i am trying > reform-boot-config --emmc /dev/mapper/reformvg-root | 19:06 |
| minute | ok that seemed to have been the right thing | 19:08 |
| minute | aha, local-block fails in initramfs, as expected | 19:08 |
| minute | so initramfs is buggy | 19:09 |
| minute | > ALERT! /dev/mapper/reformvg-root does not exist. Dropping to a shell! | 19:09 |
| minute | but there's no shell | 19:09 |
| minute | doing > setenv bootargs ro no_console_suspend pci=pcie_bus_perf nvme_core.default_ps_max_latency_us=0 fbcon=rotate:3 cma=256MB init=/bin/sh | 19:10 |
| minute | i miss reform-init... | 19:11 |
| minute | ok, i need to work around it wanting to use the non-initialized display as a console | 19:12 |
| minute | and also we really need to get display working in initramfs... this is unusable | 19:14 |
| minute | i wonder why initramfs-tools can't do the job right and include all the necessary things | 19:15 |
| minute | booting from microSD | 19:17 |
| minute | commenting out the line in boot.scr > setenv bootargs "${bootargs} console=tty1" | 19:19 |
| minute | and also setting loglevel to 7 | 19:19 |
| minute | > mkimage -c none -A arm -T script -d boot.scr.txt boot.scr | 19:21 |
| minute | and there's a console | 19:22 |
| minute | copying lsmod output to a file so i can compare vs a working system | 19:23 |
| - andreas-e (QUIT: Quit: Leaving) (~Andreas@2001:861:c4:f2f0::c64) | 19:23 | |
| minute | > diff --color pref-broken-boot-modules-sorted.txt working-boot-microsd-sorted.txt | 19:27 |
| minute | hmm tons of stuff missing | 19:27 |
| minute | huh, dwc3 is missing, weird | 19:28 |
| minute | imx_bus is missing | 19:29 |
| minute | that might be the culprit | 19:29 |
| minute | hmm didn't i bake that into the kernel recently? | 19:29 |
| minute | other candidates: imx_cpufreq_dt imx_mailbox imx_pcm_dma imx_sdma | 19:30 |
| minute | reset_imx7 | 19:30 |
| + chomwitt (~chomwitt@2a02:587:7a10:cf00:1ac0:4dff:fedb:a3f1) | 19:33 | |
| minute | imx_bus is CONFIG_ARM_IMX_BUS_DEVFREQ | 19:33 |
| minute | p sure imx_bus is the issue | 19:34 |
| minute | i'm in the initramfs again. trying to load that module from disk | 19:36 |
| minute | > mount /dev/mmcblk2p2 /mnt | 19:36 |
| minute | doesn't work because i have to (lol) > modprobe ext4 | 19:36 |
| minute | > insmod /mnt/lib/modules/6.8.11-mnt-reform-arm64/kernel/drivers/devfreq/imx-bus.ko.xz | 19:37 |
| minute | lmao > [ 169.585907] imx-bus-devfreq 32700000.interconnect: devfreq_add_device: Unable to find governor for the device | 19:38 |
| minute | > [ 169.603958] imx-bus-devfreq: probe of 32700000.interconnect failed with error -22 | 19:38 |
| minute | woooahhhh | 19:39 |
| minute | i did the following: | 19:39 |
| minute | insmod /mnt/lib/modules/6.8.11-mnt-reform-arm64/kernel/drivers/devfreq/governor_userspace.ko.xz | 19:39 |
| minute | insmod /mnt/lib/modules/6.8.11-mnt-reform-arm64/kernel/drivers/devfreq/governor_passive.ko.xz | 19:40 |
| minute | insmod mnt/lib/modules/6.8.11-mnt-reform-arm64/kernel/drivers/devfreq/governor_simpleondemand.ko.xz | 19:40 |
| minute | insmod /mnt/lib/modules/6.8.11-mnt-reform-arm64/kernel/drivers/devfreq/imx-bus.ko.xz | 19:41 |
| minute | then a lot of magic happened | 19:41 |
| minute | display went on! | 19:41 |
| minute | usb devices are proberd | 19:41 |
| minute | pcie is probed | 19:41 |
| minute | in the initramfs, most of these drivers are missing | 19:42 |
| minute | only governor_simpleondemand.ko is in /lib/modules/6.8.11-mnt-reform-arm64/kernel/drivers/devfreq/ | 19:42 |
| minute | josch: do you have any idea what could be buggy in initramfs-tools that would forget to include these drivers? i thought it looks at the list of loaded modules | 19:43 |
| minute | josch: both cryptroot/crypttab and etc/fstab in the initramfs are _empty_ | 19:44 |
| minute | gah | 19:44 |
| minute | so i'm doing > cryptsetup luksOpen /dev/nvme0 | 19:45 |
| - chomwitt (QUIT: Ping timeout: 272 seconds) (~chomwitt@2a02:587:7a10:cf00:1ac0:4dff:fedb:a3f1) | 19:45 | |
| minute | this works | 19:45 |
| minute | bizarrely after booting it wants to know the passphrase again | 19:45 |
| minute | ok, am at tuigreet | 19:45 |
| minute | system / is on encrypted nvme, everything works | 19:46 |
| minute | i'm trying to add imx_bus and cpufreq_dt to /etc/initramfs-tools/modules and rebuilding initramfs | 19:48 |
| minute | > cryptsetup: WARNING: target 'crypt' not found in /etc/crypttab | 19:48 |
| minute | ah, those were not enough, it still complains about a missing governor. so must be passive or userspace | 19:50 |
| minute | ok so the missing governor is: userspace | 19:58 |
| minute | hmm, crypttab has an entry reform_crypt | 20:00 |
| minute | shouldn't that be just "crypt"? | 20:02 |
| + gustav28 (~gustav@c-8b68524e.019-141-67626730.bbcust.telenor.se) | 20:02 | |
| minute | ah no it should be reform_crypt, but i didn't know the correct incantation for reform-boot-config earlier | 20:04 |
| minute | aha, magically we have display in initramfs now | 20:07 |
| minute | almost as good as having display in uboot! | 20:07 |
| * mjw -> Guest7569 | 20:10 | |
| - Guest7569 (QUIT: Killed (silver.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:10 | |
| * Guest7546 -> mjw | 20:10 | |
| + Guest7569 (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:11 | |
| minute | josch: can you release a new reform-tools with this? https://source.mnt.re/reform/reform-tools/-/merge_requests/71 | 20:13 |
| minute | now looking into suspend/resume | 20:16 |
| minute | it crashes when suspending with: > hdmi_pclk already disabled and > hdmi_pclk already unprepared | 20:16 |
| minute | so as expected, hdmi related... the patchstack we have for hdmi is a bit older | 20:17 |
| minute | https://lore.kernel.org/all/20240227220444.77566-1-aford173@gmail.com/ | 20:19 |
| minute | it looks like imx8mp hdmi is merged upstream | 20:19 |
| minute | yep https://github.com/torvalds/linux/commits/master/drivers/phy/freescale/phy-fsl-samsung-hdmi.c | 20:20 |
| minute | it appeared in 6.10-rc1 | 20:21 |
| - qbit (QUIT: Remote host closed the connection) (~qbit@mail.suah.dev) | 20:25 | |
| + qbit (~qbit@mail.suah.dev) | 20:26 | |
| josch | minute: sorry, i'll be back in around one hour and then read the backlog | 21:02 |
| minute | josch: no worries, i was expecting you to read it later when you have time ^^ | 21:03 |
| josch | if anything is time-critical of course feel free to git push, merge upload etc freely :) | 21:03 |
| + reform8246 (~bastien@2a01:e0a:b48:b300:b73f:8441:5693:ad4a) | 21:32 | |
| reform8246 | hello all | 21:33 |
| - reform8246 (QUIT: Quit: Leaving) (~bastien@2a01:e0a:b48:b300:b73f:8441:5693:ad4a) | 21:41 | |
| josch | oh, only 3 modules missing? that was a rather tame change | 21:44 |
| josch | if i didn't miss anything, the bigger problem is reform-setup-encrypted-nvme | 21:45 |
| josch | minute: first problem: "Your /boot partition will be on the SD-Card by default. Do you want it on eMMC instead?" -- do you mean to change the default? should the default be different for all platforms or just imx8mp? If it should be different, a new variable in machines/*.conf files is needed. | 21:48 |
| josch | wow, how did copying take 48 minutes? o0 | 21:49 |
| minute | josch: idk, used system with 28GB stuff | 21:52 |
| minute | many small files i guess | 21:52 |
| josch | okay, what do you want the default to be for reform-boot-config as called by reform-setup-encrypted-nvme? Should it put /boot on emmc or sd-card by default? | 21:53 |
| minute | josch: yeah, normally there is no SD card involved on imx8mp... only in debug cases | 21:53 |
| josch | does it make sense to change this for all platforms that are allowed to use the emmc? | 21:53 |
| minute | josch: i would say so yeah | 21:53 |
| - cobra (QUIT: Quit: ZNC 1.8.2 - https://znc.in) (~cobra@user/Cobra) | 21:54 | |
| josch | minute: okay, fixed. I'll start collecting the changes in this new MR: https://source.mnt.re/reform/reform-tools/-/merge_requests/72 | 21:55 |
| josch | next thing: "/dev/mmcblk2p1 is still mounted somewhere" | 21:55 |
| minute | josch: yeah. i would say if /boot is mounted before reform-boot-config, umount it | 21:56 |
| josch | i can see two ways to fix this: a) interactively ask the user to umount or b) check the mount early in reform-setup-encrypted-nvme | 21:56 |
| minute | josch: and remount it after reform-boot-config has run | 21:56 |
| josch | the reason i did it this way is, that i want to prevent users from nuking their /boot | 21:57 |
| minute | josch: checking early is also a good option, and giving the umount command | 21:57 |
| minute | josch: _and_ another convenience would be to output the reform-boot-config line before it is being run, so you have a chance to redo that | 21:57 |
| minute | if it fails | 21:57 |
| josch | we can do more than one of those things above | 21:57 |
| + cobra (~cobra@user/Cobra) | 21:58 | |
| minute | yeah! | 21:58 |
| josch | i think the important choice is (others depend on this): should reform-boot-config umount by itself and if yes, after a prompt or automatically if needed? | 21:58 |
| minute | josch: imho it's fine to do it automatically after a warning | 21:59 |
| josch | well, the warning is not of much help if it then just continues -- or do you mean a worning which allows to abort interactivey? | 21:59 |
| josch | haha worning | 21:59 |
| minute | yeah warning and the ability to continue y/n | 22:00 |
| josch | okay, one sec... | 22:00 |
| josch | minute: one topic ahead: even if reform-boot-config allows to interactively unmount, it can still fail | 22:04 |
| josch | minute: in that case, it is not so easy to recover because to do that, one first hast to unlock the luks, set up lvm etc | 22:04 |
| josch | so i wonder how that should best be done | 22:04 |
| josch | so essentially, the user would have to run something like: cryptsetup luksOpen ... && vgchange -ay reformvg && reform-boot-config /dev/reformvg/root && vgchange -an reformvg && cryptsetup luksClose reform_crypt | 22:07 |
| josch | that's a bit unweildy... | 22:07 |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-8b68524e.019-141-67626730.bbcust.telenor.se) | 22:15 | |
| minute | josch: hmm i didn't need to wrestle with vgchange etc myself for some reason | 22:17 |
| minute | josch: what's the failure mode? | 22:17 |
| josch | minute: imagine the user runs reform-setup-encrypted-nvme, everything gets copied, then reform-boot-config runs and asks the user to umount, the user chooses "no" | 22:19 |
| + ephase (~ephase@2a01:e0a:2a:5300:8af3:6216:8fce:7058) | 22:21 | |
| josch | minute: untested interactive umount: https://source.mnt.re/reform/reform-tools/-/merge_requests/72/diffs?commit_id=46c5b1a91992cb48f6a7c28590baca88c0cd7d92 | 22:22 |
| minute | josch: cool, thanks! | 22:27 |
| josch | while we are at it, this (untested) commit sets loglevel to 7 in the initramfs on ls1028a: https://source.mnt.re/reform/reform-tools/-/merge_requests/72/diffs?commit_id=d566dc61815809f85bb1ecfc9e170e8e2c5f6bde | 22:29 |
| josch | i do not know if it does that early enough for the luks passphrase though | 22:29 |
| josch | minute: and this one (untested) will print instructions what to do if reform-boot-config failed: https://source.mnt.re/reform/reform-tools/-/merge_requests/72/diffs?commit_id=9b2716eccd54c5540316cae51af6bd6b7062c51b | 22:40 |
| josch | minute: do you have capacity to test this before release? | 22:40 |
| minute | josch: not today unfortunately | 22:48 |
| minute | josch: only tomorrow | 22:48 |
| josch | minute: same here -- my ls1028a where i can run reform-setup-encrypted-nvme because nothing of value will get wiped is not with me at home | 22:49 |
| minute | josch: ok, it's fine that i'll test it tomorrow... | 22:49 |
| josch | would it make sense to send the link to the MR to the thread in the forum in case an affected user wants to test it before release? | 22:50 |
| minute | josch: most definitely | 22:52 |
| josch | okay, i'll post it unless you would like to do that | 22:56 |
| + gustav28 (~gustav@c-8b68524e.019-141-67626730.bbcust.telenor.se) | 22:56 | |
| josch | posted | 23:03 |
| + eibachd (~eibachd@p200300dcf74bee0070221fc8ae10b82c.dip0.t-ipconnect.de) | 23:19 | |
| + chomwitt (~chomwitt@2a02:587:7a10:cf00:1ac0:4dff:fedb:a3f1) | 23:37 | |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-8b68524e.019-141-67626730.bbcust.telenor.se) | 23:43 | |
| minute | nice, i finally have an acrylic bottom on my rk3588 reform again (and copper heatsink) | 23:48 |
| minute | finally decent wifi again... 200mbit/s on speedtest.net | 23:48 |
| minute | (with ye olde ath9k mpcie carfd) | 23:48 |
| minute | card | 23:49 |
| josch | finally acrylic bottom again? what was there instead? no bottom? :D | 23:49 |
| minute | josch: full aluminum bottom | 23:50 |
| minute | (there exists only one, it is quite heavy) | 23:50 |
| josch | the reform-next will not have an acrylic bottom, right? | 23:51 |
| minute | josch: correct, it is a very different design where the bottom is like an aluminum bathtub | 23:51 |
| josch | photo of the underside of your rk3588 reform on fedi? :) | 23:52 |
| minute | josch: ok one moment :D | 23:52 |
| josch | nice! i have the repost button ready! | 23:52 |
| minute | josch: there you go! https://mastodon.social/@mntmn/112594550083855290 | 23:55 |
| josch | the copper heatsink makes it look so alien! :D | 23:55 |
| josch | okay... i see the hdmi-dsi adapter, the wifi with the molex antenna instead of the laird, next to the trackball the speaker mod that i also have -- nice | 23:57 |
| minute | ~next gen~ | 23:57 |
| minute | :D | 23:57 |
| josch | i really love the look of copper | 23:57 |
| minute | hdmi-edp! | 23:57 |
| josch | errr, yes | 23:57 |
| minute | yeah the copper looks much better than i imagined somehow | 23:58 |
| minute | i wonder if it will turn green one day :D | 23:58 |
| josch | just leave it out in the rain often enough or something ;) | 23:58 |
| minute | hehe | 23:58 |
| minute | CPU temp currently at 41 degrees | 23:59 |
| minute | :3 | 23:59 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!