| violet | im working on a gentoo install on my pocket right now | 00:09 |
|---|---|---|
| dozens | neat! | 00:38 |
| - mtm (QUIT: Ping timeout: 252 seconds) (~textual@c-71-228-84-213.hsd1.fl.comcast.net) | 02:02 | |
| - mjw (QUIT: Ping timeout: 276 seconds) (~mjw@213-10-231-91.fixed.kpn.net) | 02:05 | |
| + mtm (~textual@c-71-228-84-213.hsd1.fl.comcast.net) | 02:05 | |
| - jacobk (QUIT: Ping timeout: 276 seconds) (~quassel@2601:380:837f:3520:b413:1241:5e3:eebd) | 04:28 | |
| + jacobk (~quassel@2601:380:837f:3520:1113:1c75:8d3b:2220) | 05:32 | |
| - jacobk (QUIT: Ping timeout: 260 seconds) (~quassel@2601:380:837f:3520:1113:1c75:8d3b:2220) | 06:18 | |
| - colinsane (QUIT: Quit: bye) (~colinunin@97-113-150-69.tukw.qwest.net) | 09:15 | |
| + colinsane (~colinunin@97-113-150-69.tukw.qwest.net) | 09:18 | |
| + mjw (~mjw@213-10-231-91.fixed.kpn.net) | 11:46 | |
| - nsc (QUIT: Ping timeout: 252 seconds) (~nicolas@i5C74DC43.versanet.de) | 11:54 | |
| minute | josch: sync was already there https://source.mnt.re/reform/pocket-reform/-/blob/display-panel2/scripts/update-sysctl-firmware.sh#L47 | 12:29 |
| josch | hrm... | 12:31 |
| josch | i wonder if this kind of stuff can be done with a mini-rescue usb-stick which loads a mini-system from an initramfs like d-i does it... | 12:31 |
| minute | josch: i mean, that is totally not a surefire solution, there can be any disk activity going on between that sync and the end of picotool | 12:31 |
| josch | i remember there used to be mini-iso images for usb-sticks or floppies back in the day for bios updates | 12:33 |
| josch | maybe the pocket needs something similar to avoid a hard-reset messing with mounted filesystems | 12:33 |
| josch | i'm not saying you should implement this | 12:33 |
| josch | i made a note about this in my todo list | 12:34 |
| minute | we do something similar here just with modified microsd card (images) | 12:34 |
| ch | would be an interesting quest to see if fwupd works on uboot platforms and if that can be used | 12:34 |
| minute | for factory install purposes | 12:34 |
| josch | ch: yes, i had a look at fwupd in the past already but it doesn't seem to be tailored at the kind of firmware the reform wants to flash | 12:35 |
| josch | ch: but yes, would be nice to use existing facilities to flash things like the keyboard, trackball etc | 12:35 |
| josch | ch: but even with fwupd, the hard-reset would be unavoidable, no? | 12:36 |
| josch | minute: right, if your sd-card disk image is disposable, then a hard-reset is irrelevant | 12:36 |
| ch | sure, but if fwupd already has infra to, say, do it from uboot, ... | 12:36 |
| josch | ch: i did not see any such thing | 12:37 |
| ch | :( | 12:37 |
| josch | if you know more about fwupd, please have a look -- maybe i was blind | 12:37 |
| ch | (i think on x86 it can use efi for that...) | 12:37 |
| josch | (happens too often) | 12:37 |
| josch | yes | 12:37 |
| josch | i found the efi stuff | 12:37 |
| ch | but yeah, if there's a way of knowing if the fw is already installed, putting a hook into initramfs would also work i think | 12:41 |
| ch | might be less fiddly than doing it from uboot | 12:41 |
| josch | ch: you mean the user would run a tool which would temporarily create an initramfs which does the right thing on the next reboot? | 12:42 |
| ch | josch: yeah. i guess it could also just be permanently in the initramfs, if it checks the version first | 12:42 |
| josch | hrm... i think that's too prone to accidental failure | 12:43 |
| ch | (having it temporarily gives better control i guess) | 12:43 |
| josch | i think it should be the user's explicit choice to upgrade their firmware | 12:43 |
| minute | googling around there's an extreme tool called > echo u > /proc/sysrq-trigger | 12:44 |
| ch | so, my wild idea: have the reform tools package install an initramfs hook which only gets activated if you have a flag file on disk. if so, it copies the fw files and picotool and so on and runs a script before mounting the rootfs. if versions dont match -> flash and reboot | 12:44 |
| josch | ch: that could be done, yes. We already install a few scripts into the initramfs | 12:45 |
| josch | (even some which are executed before mounting filesystems) | 12:45 |
| ch | something should delete the flag file afterwards, maybe a onetime systemd service could check if the firmware files now match and do that | 12:46 |
| ch | (and otherwise log an error or sth) | 12:46 |
| josch | minute: "echo u > /proc/sysrq-trigger" might be a suitable hot-fix for now | 12:46 |
| josch | i have a business trip this week and guests over for several days so my time is limited in the near future unfortunately XD | 12:46 |
| ch | if kexec works, that would also be an interesting option to explore | 12:47 |
| ch | and -maybe- systemd has a facility to run sth just before rebooting | 12:47 |
| ch | thinking of it, systemd has a new soft-reboot thing where it shuts down and then pivots_root into sth new, so there you also have a clean state (but dont need to do a reboot first) | 12:48 |
| - colinsane (QUIT: Quit: bye) (~colinunin@97-113-150-69.tukw.qwest.net) | 13:45 | |
| + colinsane (~colinunin@97-113-150-69.tukw.qwest.net) | 13:48 | |
| - mtm (QUIT: Ping timeout: 260 seconds) (~textual@c-71-228-84-213.hsd1.fl.comcast.net) | 14:04 | |
| + mtm (~textual@c-71-228-84-213.hsd1.fl.comcast.net) | 14:05 | |
| ch | josch: food for thought: https://gist.githubusercontent.com/zeha/f47ed474c34066090796e0d3bc218199/raw/869024ac3fb3cef2f879e7723173506c9c63e685/fwupdate.sh | 14:50 |
| - mjw (QUIT: Remote host closed the connection) (~mjw@213-10-231-91.fixed.kpn.net) | 16:05 | |
| + mjw (~mjw@213-10-231-91.fixed.kpn.net) | 16:05 | |
| + nsc (~nicolas@i5C74DC43.versanet.de) | 16:19 | |
| swivel | 12823 N Jul 22 eBay ( 50K) 🚚 EARLY DELIVERY UPDATE: Purism Librem 5 | 16:46 |
| swivel | 12824 N Jul 22 eBay ( 53K) Out for delivery: Purism Librem 5 | 16:46 |
| swivel | ooo, looks like another foss imx8 gadget should be showing up today | 16:46 |
| Twodisbetter | have fun with it. When it works it is ok, but mine has been boot looping for no reason for the last 3 weeks. It is now hanging out powered off and in a drawer. | 17:40 |
| swivel | doh | 17:54 |
| swivel | my friend's l5 also decided to stop working | 17:55 |
| + ZylonMaster (~hjcs@ool-ad02e01f.dyn.optonline.net) | 18:09 | |
| jfred | The Pocket Reform is such a cozy little device :D | 18:11 |
| jfred | It's nice to have a tiny laptop with an actually very good keyboard and trackball | 18:11 |
| minute | jfred: happy to hear :3 | 18:13 |
| + hjcfd (~hjcs@ool-ad02e01f.dyn.optonline.net) | 18:13 | |
| - hjcfd (QUIT: Client Quit) (~hjcs@ool-ad02e01f.dyn.optonline.net) | 18:14 | |
| - ZylonMaster (QUIT: Ping timeout: 260 seconds) (~hjcs@ool-ad02e01f.dyn.optonline.net) | 18:15 | |
| - Manis (QUIT: Ping timeout: 252 seconds) (01a66df340@84-73-19-122.dclient.hispeed.ch) | 18:31 | |
| + Jonas__ (~Jonas@45.134.79.119) | 18:43 | |
| josch | ch: wow, *interesting*! | 18:54 |
| josch | ch: please keep that gist online, i'm copying that link into my notes :) | 18:55 |
| ch | sure | 18:55 |
| josch | i think i want to try this out | 18:56 |
| josch | i'll have to figure out whether i like to tie the mechanism to systemd or to mkinitramfs better... | 18:56 |
| ch | yeah | 18:57 |
| ch | systemd has the advantage one doesn't need to fiddle with the boot setup | 18:57 |
| + Manis (01a66df340@84-73-19-122.dclient.hispeed.ch) | 19:18 | |
| josch | indeed | 19:50 |
| - Jonas__ (QUIT: Ping timeout: 255 seconds) (~Jonas@45.134.79.119) | 20:00 | |
| minute | minute! | 20:15 |
| minute | woops, wrong window | 20:16 |
| minute | :D | 20:16 |
| josch | ch: I've now read systemd.offline-updates(7) (thank you, that was interesting!) and i learned that the script will be run explicitly after filesystems have been mounted but that's exactly what we want to avoid because picotool will trigger a cold system reset | 20:36 |
| + Jonas (~Jonas@45.134.79.119) | 20:53 | |
| - Jonas (QUIT: Remote host closed the connection) (~Jonas@45.134.79.119) | 21:22 | |
| + Jonas (~Jonas@45.134.79.119) | 21:22 | |
| - Gooberpatrol66 (QUIT: Ping timeout: 260 seconds) (~Gooberpat@user/gooberpatrol66) | 21:29 | |
| josch | I'm looking at this change by mtm again: https://source.mnt.re/reform/reform-tools/-/merge_requests/75/diffs?commit_id=728698d83bb04429e6cd5ee8785c91ebc2ca3410 | 21:33 |
| josch | i don't like how this will override an explicit block by the user unconditionally | 21:34 |
| josch | does somebody have an idea of how we can do this better? | 21:34 |
| mtm | yeah, it was really just a quick hack to get bluetoothh running again | 21:34 |
| josch | that's fine -- it's a place from which we can start :) | 21:35 |
| mtm | I haven't tried booting without that with the newest kernel | 21:35 |
| josch | mtm: could you give that a try and see what happens? | 21:35 |
| mtm | sure thing | 21:35 |
| josch | thanks! | 21:35 |
| josch | also, 6.10 is sitting in experimental -- we could also try that one | 21:36 |
| mtm | well, good news: with 6.9.10 the hack is no longer needed, bluetooth fired right up | 21:39 |
| josch | hah, nice! | 21:40 |
| mtm | now I just need to spend a little time to figure out what's going on with the waybar battery meter bug | 21:41 |
| josch | ah yes i read about that in the forum | 21:41 |
| josch | was the waybar version updated or what changed? | 21:41 |
| mtm | I believe so. I haven't researched what exactly changed yet | 21:43 |
| mtm | what's the 'apt' command to see the various versions of a package? I forget | 21:50 |
| josch | apt-cache policy mypkg | 21:50 |
| mtm | thx | 21:51 |
| josch | minute: could you assign the intel gitlab CI runner to https://source.mnt.re/josch/reform-debian-packages -- thank you! | 21:54 |
| + Gooberpatrol66 (~Gooberpat@user/gooberpatrol66) | 21:55 | |
| - Jonas (QUIT: Remote host closed the connection) (~Jonas@45.134.79.119) | 22:18 | |
| ch | josch: re systemd.offline-updates after filesystems - yes, on a normal boot. the soft-reboot script avoids that | 22:33 |
| josch | ch: i read (what i think are the important parts of) systemd-soft-reboot.service(8) but i do not see how a soft reboot unmounts my file systems -- instead it says "File system mounts may remain mounted during the transition" | 22:39 |
| ch | josch: they remain -if- nextroot uses them. but nextroot is a tmpfs without any references to the old mount tree, and thus stuff gets unmounted | 22:39 |
| ch | (all of this can be done without the systemd.offline-updates machinery, it was just a bit nicer) | 22:40 |
| josch | (well, no need to duplicate the mechanism if it already exists) | 22:41 |
| ch | https://per.namespace.at/~ch/scr20240722224149.png screengrab from a vm with mount -l stuck into the .sh | 22:43 |
| ch | (i think the cp -av /dev is useless actually) | 22:44 |
| josch | (seems so, as there is an udev mounted in it) | 22:44 |
| ch | yeah so i'm sure this needs a bit of polishing, but i think it would be a nice setup | 22:45 |
| josch | what i find also tempting about the systemd approach is, that i can set up my own mini-chroot in /run/nextroot without the limitations of mkinitramfs | 22:46 |
| ch | indeed | 22:46 |
| josch | ch: so... suppose i'm using this... can you add a small copyright line at the top? thanks! | 22:46 |
| ch | note that os-release needs to exist for soft-reboot to consider nextroot to be valid | 22:46 |
| josch | weird but ok | 22:48 |
| ch | https://gist.githubusercontent.com/zeha/f47ed474c34066090796e0d3bc218199/raw/3e74ff72fe196bf7d132a381cfecb5d06e14fd74/fwupdate.sh | 22:49 |
| josch | ch: perfect, thank you! | 22:50 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
