| ex-parrot | dang the pocket reform looks so good | 00:00 |
|---|---|---|
| bluerise | ah, another designware mipi dsi . hmmm | 00:00 |
| minute | ex-parrot: thank you! it's not even the production version | 00:00 |
| ex-parrot | I' | 00:00 |
| ex-parrot | I've been very out of the loop recently, good to be catching up on all the amazing stuff happening | 00:00 |
| josch | minute: at some point you mentioned an upgraded speaker that can sit next to the trackball? | 00:01 |
| josch | is this something that will become reality? | 00:01 |
| josch | in https://files.mastodon.social/media_attachments/files/110/688/973/385/219/152/original/9b6f8c8f05d3bda0.png are you using networkmanager to connect to lte? | 00:02 |
| minute | josch: yes, networkmanager+modemmanager | 00:02 |
| minute | josch: sooo the speaker uprade is in my reform, it's just not super cleanly made and i basically forgot turning it into a product, thanks for bringing it up :3 | 00:03 |
| ex-parrot | it's very weird running gnome in the v3 system image for a change, I have almost exclusively been using LFS and then 9front up to now | 00:10 |
| ex-parrot | josch: switched my pristine v3 image to the reform.debian repo, all seems fine so far but I did get a config file change warning when installing apparmor | 00:20 |
| josch | ex-parrot: what warning was that? | 00:20 |
| ex-parrot | and a warning about not having zstd for the initrd | 00:20 |
| josch | the zstd warning can be safely ignored | 00:20 |
| josch | no idea why it's there and presented to every user not having zstd installed | 00:21 |
| minute | ex-parrot: how is gnome holding up in there? | 00:21 |
| minute | a zstd ad! | 00:22 |
| ex-parrot | I only barely know how to use gnome but it seems OK so far? the wifi password dialog didn't work | 00:23 |
| ex-parrot | josch: https://paste.hotplate.co.nz/PSix2KJ6ep#4D+oMbCDSs8HLXeJRyakMQ$0 here is the error I got | 00:23 |
| ex-parrot | minute: also the KDE packages are all installed as well and it makes the Gnome chooser experience a bit meh | 00:23 |
| ex-parrot | the battery amount is not working but I assume it's because I am on old LPC firmware | 00:23 |
| ex-parrot | going to upgrade that shortly | 00:24 |
| ex-parrot | the motd gets printed again when I run "gnome" which is kind of amusing but not harmful :P | 00:25 |
| josch | ex-parrot: ah you can indeed ignore that appormor bug (as you did) because it doesn't apply to you | 00:26 |
| ex-parrot | _b | 00:26 |
| josch | ex-parrot: thank you for trying out reform.debian.net! this is all highly experimental and i'm happy i didn't bork your system. XD | 00:27 |
| ex-parrot | it's fine, it's a fresh install of the latest v3 image that I downloaded while trying to troubleshoot my SD card situation anyway | 00:27 |
| ex-parrot | so nothing much to bork | 00:28 |
| ex-parrot | I have somehow never used the speakers before, good to know they work I suppose :P | 00:28 |
| ex-parrot | should the v3 reform images have the security and updates components enabled in sources.list? | 00:35 |
| ex-parrot | the one I grabbed from gitlab just had unstable/main | 00:35 |
| ex-parrot | I switched this to bookworm | 00:35 |
| josch | uuh good point | 00:35 |
| josch | let me update the instructions | 00:35 |
| ex-parrot | the updates and security components should prob be enabled in the reform image too? or is that on purpose | 00:37 |
| - mjw (QUIT: Ping timeout: 240 seconds) (~mjw@gnu.wildebeest.org) | 00:38 | |
| josch | ex-parrot: the image built by the mnt.re CI use unstable so there are no updates and security mirrors | 00:38 |
| josch | on reform.debian.net i will offer images for bookworm and those will include updates and security | 00:38 |
| ex-parrot | oh of course | 00:38 |
| ex-parrot | sorry, not thinking clearly this morning apparently :) | 00:39 |
| josch | ex-parrot: i added how your sources.list should look like now: https://reform.debian.net/ | 00:39 |
| ex-parrot | great | 00:39 |
| ex-parrot | I generally go for https for the mirror lines but I'm not sure this is very important | 00:39 |
| josch | you can go for https but it will not add much | 00:39 |
| ex-parrot | indeed | 00:40 |
| ex-parrot | ACTION <-- runs ftp.nz.debian.org :P | 00:40 |
| josch | uh nice! | 00:40 |
| ex-parrot | so I will use that mirror | 00:40 |
| ex-parrot | but yeah, this is looking great | 00:40 |
| ex-parrot | one thing on these instructions, the cat heredoc to update the sources list seems to assume a root shell where everything else here is prefixed with sudo | 00:41 |
| josch | good point -- one sec | 00:41 |
| josch | thanks, fixed! | 00:42 |
| ex-parrot | lovely, instructions worked | 00:43 |
| josch | sweet :) | 00:43 |
| ex-parrot | re-ran dist-upgrade and got another 16 packages to do | 00:43 |
| ex-parrot | josch: do you know if xorg is expected to work on the v3 sysimage w/ that 10-reform-etnaviv config file? | 00:55 |
| ex-parrot | I am getting an error at startup currently about not being able to load glx or modesetting | 00:55 |
| josch | ex-parrot: i never got it to work | 00:57 |
| ex-parrot | ah ok | 00:57 |
| ex-parrot | ACTION returns this project to the backlog | 00:58 |
| - S0rin (QUIT: Ping timeout: 250 seconds) (~S0rin@user/s0rin) | 01:06 | |
| + S0rin (~S0rin@user/s0rin) | 01:09 | |
| ex-parrot | do I want firmware-20_R3.bin for an early reform board? | 01:14 |
| ex-parrot | or firmware-25_R2.bin? | 01:14 |
| josch | firmware-20_R3.bin | 01:17 |
| josch | firmware-25_R2.bin is for mainboard 2.5 | 01:17 |
| ex-parrot | col | 01:17 |
| ex-parrot | er cool | 01:17 |
| ex-parrot | I wasn't sure if the RX or the 2X was the thing to be looking at | 01:18 |
| + bkeys (~Thunderbi@static-198-54-130-101.cust.tzulo.com) | 01:19 | |
| ex-parrot | it seems like there's an R4 floating around too? | 01:22 |
| minute | ex-parrot: what's gnome chooser? | 01:22 |
| ex-parrot | the application launcher thing you get when you hit the MNT button | 01:23 |
| minute | oh ah | 01:23 |
| minute | yeah i'm thinking to backpedal on kde (and maybe gnome) and go for a more lightweight image in the future incl instructions on how to install the DEs | 01:24 |
| ex-parrot | seems sensible to me | 01:24 |
| ex-parrot | ok LPC R3 is working | 01:24 |
| ex-parrot | amazing. battery stats in gnome | 01:25 |
| ex-parrot | ok reform-flash-uboot in the current v3 image complains about checksum mismatch for both --offline and download modes | 01:34 |
| josch | ex-parrot: what version of reform-tools are you on? | 01:37 |
| ex-parrot | 1.22 from the reform.debian repo | 01:37 |
| josch | hrm... | 01:37 |
| josch | oh shoot! | 01:39 |
| josch | i forgot to update that checksum | 01:40 |
| ex-parrot | another small bug: the keyboard flash script says to toggle the DIP switch to "ON" but "ON" is the non-programming state, it should say to toggle it to "PROG" | 01:40 |
| josch | minute: is that a mistake in the handbook? "Remove the keyboard’s frame and toggle the programming DIP switch SW84 on the keyboard to “ON”." https://mntre.com/reform2/handbook/parts.html#keyboard-firmware | 01:41 |
| ex-parrot | maybe mine has been in the wrong state all this time | 01:42 |
| josch | minute: now that you made kernel 6.4 official it's also really important that you merge this: https://source.mnt.re/reform/reform-tools/-/merge_requests/39 | 01:42 |
| ex-parrot | hmm ignore me, I think I am wrong about the switch thing | 01:42 |
| josch | minute: or otherwise building the lpc dkms module will fail | 01:43 |
| minute | josch: merged | 01:46 |
| minute | ex-parrot: ON is correct. the switch is called PROG | 01:46 |
| minute | need to sleep! n8y | 01:47 |
| ex-parrot | yeah sorry. I must have toggled it absent mindedly then toggled it again and gotten confused | 01:47 |
| ex-parrot | cya minute | 01:47 |
| josch | ex-parrot: could you try this patch with reform-flash-uboot: https://paste.debian.net/hidden/61634ee9/ | 01:48 |
| ex-parrot | I get "41: version: parameter not set" | 01:50 |
| ex-parrot | and it fails to run | 01:50 |
| josch | oh sec... | 01:50 |
| josch | ex-parrot: and this one on top of what you have: https://paste.debian.net/hidden/e591a3b4/ | 01:53 |
| josch | err... | 01:54 |
| ex-parrot | "1 computed checksum did NOT match" | 01:54 |
| josch | if [ -n "$version" ]; then | 01:54 |
| josch | should be | 01:54 |
| josch | if [ -z "$version" ]; then | 01:54 |
| ex-parrot | :D | 01:54 |
| ex-parrot | get a sha1sum warning but it proceeds anyway | 01:55 |
| ex-parrot | which seems like what would be expected | 01:55 |
| josch | what warning? | 01:55 |
| josch | it warns despite --quiet? | 01:55 |
| ex-parrot | "1 computed checksum did NOT match" | 01:55 |
| ex-parrot | "/boot/flash.bin: FAILED" | 01:56 |
| josch | aha | 01:57 |
| josch | got it | 01:57 |
| ex-parrot | semi related, I guess picotool isn't packaged in debian yet? | 01:58 |
| josch | i removed the --quiet and appended >/dev/null 2>&1 | 01:58 |
| ex-parrot | let me test that | 01:58 |
| josch | it is not | 01:59 |
| josch | okay, final diff: https://source.mnt.re/reform/reform-tools/-/merge_requests/41/diffs | 02:00 |
| josch | minute: please merge tomorrow :) | 02:00 |
| josch | thank you! | 02:00 |
| josch | now bedtime for me as well | 02:00 |
| ex-parrot | all looks good now here too | 02:01 |
| ex-parrot | cheers! | 02:01 |
| - mtm (QUIT: Ping timeout: 246 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 02:03 | |
| ex-parrot | josch: for tomorrow, the reform-flash-rescue script is also failing for me | 02:06 |
| ex-parrot | "FATAL: could not search modules" | 02:16 |
| ex-parrot | going to try shovelling the modules dir over manually in to the mmc... | 02:19 |
| - robin_ (QUIT: Quit: Leaving) (~robin@user/terpri) | 02:46 | |
| - chartreuse (QUIT: Read error: Connection reset by peer) (~chartreus@S0106f0f249dfd9c3.cg.shawcable.net) | 02:50 | |
| + chartreuse (~chartreus@S0106f0f249dfd9c3.cg.shawcable.net) | 03:14 | |
| - GNUmoon (QUIT: Ping timeout: 240 seconds) (~GNUmoon@gateway/tor-sasl/gnumoon) | 03:19 | |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 03:20 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 03:22 | |
| - S0rin (QUIT: Ping timeout: 245 seconds) (~S0rin@user/s0rin) | 03:45 | |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 04:09 | |
| violet | josch: im happy to test thekernel | 04:34 |
| violet | can you mention me in a link to the kernel i should be trying out? im having a hard time processing irclogs in my brain right now | 04:36 |
| + S0rin (~S0rin@user/s0rin) | 04:39 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 05:07 | |
| - jacobk (QUIT: Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) (~quassel@47-186-122-163.dlls.tx.frontiernet.net) | 06:10 | |
| - GNUmoon (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon) | 06:12 | |
| + GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon) | 06:12 | |
| - klardotsh (QUIT: Ping timeout: 245 seconds) (~klardotsh@98.97.35.74) | 06:21 | |
| - Boostisbetter (QUIT: Ping timeout: 240 seconds) (4a410829d7@irc.cheogram.com) | 07:16 | |
| josch | ex-parrot: do you have more bits of that log for me? | 07:56 |
| josch | violet: https://source.mnt.re/reform/reform-debian-packages/-/jobs/1559/artifacts/browse/repo/pool/main/l/linux/ | 07:57 |
| josch | minute: why did you cancel the second to last reform-debian-packages pipeline? | 07:58 |
| minute | josch: because i made a tag and it ran the whole thing again for the tag | 09:40 |
| josch | minute: yes, but I suppose you have your mirror set up so that it downloads the latest artifacts from the main branch, right? Those artifacts were not produced because the pipeline got cancelled and thus mntre.com still contains the old packages: https://mntre.com/reform-debian-repo/pool/main/l/linux/ | 09:44 |
| minute | josch: damn, sorry | 10:12 |
| - lifning (QUIT: Ping timeout: 245 seconds) (~lifning@user/lifning) | 10:15 | |
| minute | rerunning that pipeline now | 10:17 |
| + lifning (~lifning@user/lifning) | 10:18 | |
| josch | minute: the rack reform uses the same motherboard as the MNT Reform 2 and thus needs no changes in the software compared to the MNT Reform 2, right? | 10:19 |
| minute | josch: correct... except perhaps it will get a different lpc firmware, but that's not 100% clear yet | 10:22 |
| + mjw (~mjw@gnu.wildebeest.org) | 10:30 | |
| josch | early feedback welcome: https://reform.debian.net/ | 10:43 |
| + Boostisbetter (4a410829d7@irc.cheogram.com) | 12:03 | |
| amospalla[m] | josch: the links on that page are stubs right? | 12:16 |
| - mjw (QUIT: Ping timeout: 240 seconds) (~mjw@gnu.wildebeest.org) | 12:17 | |
| * mark_ -> mjw | 12:25 | |
| - Boostisbetter (QUIT: Ping timeout: 240 seconds) (4a410829d7@irc.cheogram.com) | 12:47 | |
| josch | amospalla[m]: some still are, yes | 12:51 |
| josch | amospalla[m]: you mean the ones on the bottom, right? | 12:52 |
| amospalla[m] | josch: I meant the repo and images, they did point to the page itself, but now they point to real data. | 12:53 |
| josch | yes, that's fixed now :) | 12:54 |
| amospalla[m] | I can't wait for my pocket :') | 12:55 |
| - Asmadeus (QUIT: Read error: Connection reset by peer) (~asmadeus@240b:13:8c80:d300::42:13) | 12:55 | |
| + Asmadeus (~asmadeus@240b:13:8c80:d300::42:13) | 13:00 | |
| - mtm (QUIT: Ping timeout: 245 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 14:03 | |
| + abortretryfail (~arf@146.ip-149-56-132.net) | 14:50 | |
| - abortretryfail (QUIT: Ping timeout: 260 seconds) (~arf@146.ip-149-56-132.net) | 14:57 | |
| + abortretryfail (~arf@146.ip-149-56-132.net) | 14:59 | |
| dozens | me either :) | 15:23 |
| + mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 16:09 | |
| + Boostisbetter (4a410829d7@irc.cheogram.com) | 16:38 | |
| minute | josch: i think the issue about the repo not updating was that the download url was still referencing a job called "build", but now it's called "reprepro" | 16:55 |
| minute | fixed that now, so i can now apt install linux-image-6.4.0-0-reform2-arm64 | 16:56 |
| sknebel | josch: minor wording thing: "To learn more about this" is a bit unclear what "this" means - I wouldve expected that to be specificially about the "debian stable images/repos" thing, not reform in general. so maybe "For more about MNT Reform computers, ..." ? | 16:57 |
| sevan | josch: keyboard firmware update with flash.sh | 17:03 |
| sevan | forgot to make clean; make before starting and initially flashed the old build artifact :) (I had updated the tree) | 17:04 |
| sevan | now reports keyboard r2 20230703 | 17:05 |
| josch | minute: ah that makes more sense, yes :) | 17:05 |
| minute | hmm, lpc dkms build fails on a311d | 17:07 |
| minute | /var/lib/dkms/reform2_lpc/1.21/build/reform2_lpc.c:430:27: error: implicit declaration of function ‘of_match_ptr’ [-Werror=implicit-function-declaration] | 17:07 |
| minute | 430 | .of_match_table = of_match_ptr(of_tis_spi_match), | 17:07 |
| minute | > cc1: some warnings being treated as errors | 17:07 |
| josch | sknebel: thank you, fixed! | 17:07 |
| josch | minute: do you have the latest reform-tools installed? | 17:07 |
| minute | lets see | 17:07 |
| josch | minute: this error message is fixed by the merge request you merged yesterday | 17:07 |
| minute | i did not! | 17:07 |
| minute | thanks | 17:07 |
| josch | on the topic of unmerged stuff, there is also https://source.mnt.re/reform/reform-tools/-/merge_requests/41 | 17:08 |
| josch | sevan: so, success, right? :) | 17:09 |
| sevan | indeed :) | 17:09 |
| minute | josch: woops, merged | 17:09 |
| josch | ex-parrot: problem solved! ^ | 17:10 |
| josch | minute: thank you! i'll trigger another run of the reform-debian-packages pipeline | 17:12 |
| minute | ok! | 17:12 |
| minute | kind of bizarre how i can install stuff on a311d without ethernet cable or a wifi antenna... | 17:14 |
| minute | (the rtw8822cs has enough wifi reception anyway) | 17:15 |
| minute | josch: hmm, i still get the same dkms build error. | 17:16 |
| minute | Package: reform-tools | 17:17 |
| minute | Version: 1.22 | 17:17 |
| minute | > reform-tools is already the newest version (1.22). | 17:18 |
| minute | but the good news is, a311d reform booted with that kernel. > Linux reform 6.4.0-0-reform2-arm64 #1 SMP Debian 6.4.1-1~exp1+reform20230710T195054Z1 (2023-07-10) aarch64 | 17:19 |
| josch | minute: you need 1.23 | 17:24 |
| josch | minute: | 17:25 |
| josch | https://source.mnt.re/reform/reform-tools/-/blob/main/debian/changelog | 17:25 |
| sevan | does the machine need to be offline in order to reflash the LPC or can I use it to flash itself? | 17:26 |
| minute | ok hmm | 17:26 |
| minute | sevan: it cannot flash itself | 17:26 |
| sevan | minute: ack | 17:26 |
| minute | josch: the latest reform-debian-packages that i just deployed only got repo/pool/main/r/reform-tools/reform-tools_1.22_all.deb | 17:27 |
| sevan | ACTION copies firmware.bin to another device before shutting down. | 17:28 |
| minute | https://source.mnt.re/reform/reform-debian-packages/-/jobs/1557#L1128 | 17:28 |
| josch | minute: you merged https://source.mnt.re/reform/reform-tools/-/merge_requests/39 tonight at 2:00 but the last reform-debian-packages run started at 00:39 | 17:29 |
| minute | oh :D | 17:29 |
| minute | that explains it, sorry | 17:29 |
| minute | that also explains why i'm so tired today lol | 17:30 |
| josch | minute: would you like for the reform-debian-packages pipeline to run more often than once a week? | 17:30 |
| minute | josch: no, it's fine! i just wanted to have the latest version on my a311d reform :3 | 17:30 |
| josch | okay :) | 17:30 |
| minute | i'm thinking to migrate my daily driver reform to it (at least for a while) to get a good feeling for the stability | 17:31 |
| josch | minute: how did you end up solving the ls1028a cpu situation? | 17:33 |
| minute | josch: the batch is split in 14 and 36 units | 17:33 |
| josch | oh that worked out -- nice | 17:34 |
| minute | josch: 14 processors i had in stock still. and i just acquired the memory chips via a broker | 17:34 |
| minute | josch: i found some alleged LS1028AXE7PQA stock in japan, but will see tomorrow or so if it is real. otherwise i will buy from a big vendor with 18 weeks lead time | 17:34 |
| josch | uff 18 weeks | 17:34 |
| minute | yeah :( better than 52 weeks though how it used to be... | 17:35 |
| josch | yeah that's nuts | 17:35 |
| minute | interesting, the part number for imx8mplus is not MIMX8MP, but MIX8ML | 17:38 |
| minute | MIMX8ML i mean | 17:38 |
| sevan | LPC flashed as well, now shows 20230703 date. | 17:44 |
| minute | nice | 17:46 |
| sevan | alrighty, now to switch to the system to Debian stable | 17:51 |
| sevan | josch: I'm going to work through the instructions on debian.net and switch. debian-keyring isn't preinstalled on the MNT unstable (< correct terminology for reference?) | 18:02 |
| sevan | josch: step 4 clobbers /etc/apt/sources.list, step 3 unnecessary? (http://reform.debian.net/repo/) | 18:08 |
| sevan | current status: running apt update | 18:11 |
| josch | sevan: yes, debian-keyring isn't preinstalled but you can install it if you wish. Though I only created the subkey a few days ago, so it is not yet part of the debian-keyring package. But you can still use my GPG key as shipped by the debian-keyring package to make sure that the signature for that repo is really mine. | 18:13 |
| josch | sevan: what do you mean by step 4 clobbering /etc/apt/sources.list? | 18:14 |
| josch | sevan: why is step 3 unnecessary? | 18:14 |
| sevan | josch: clobbering means that the previous contents was lost so the entry for the mnt repo at step 3 is potentially unnecessary (it's lost anyway, if that's the desired behaviour) | 18:16 |
| sevan | josch: re keyring, ack, I didn't quiet understand the output from gpg when it stated "nothing exported", assuming that it hadn't nothing to do, rather than implying key is not found? (mind melting :) ) | 18:19 |
| josch | sevan: step 3 clobbers, correct. But you said that step 4 does. | 18:20 |
| sevan | ugh, off by one. | 18:20 |
| josch | And why is step 3 unncessary? | 18:20 |
| sevan | steps 2 & 3 | 18:20 |
| josch | indeed, if one just runs the command in 3. then 2. is unncessary | 18:21 |
| sevan | Do i need to "comment out or remove https://mntre.com/reform-debian-repo from /etc/apt/sources.list" if in the next step it'll be overwritten | 18:21 |
| josch | let me make the "tee" into a "tee -a" | 18:21 |
| sevan | ok :) | 18:21 |
| josch | hm... | 18:21 |
| josch | no wait | 18:21 |
| josch | so, i don't think there is a way to express what should be done as a command that can be run | 18:22 |
| josch | i've written it as a HERE document, clobbering /etc/apt/sources.list in the hopes that people would know what it does and do the right thing | 18:22 |
| josch | sevan: i don't want to tell people "throw away everything and start over" | 18:24 |
| josch | sevan: do you have an idea how i can formulate this better? | 18:24 |
| Boostisbetter | Has anyone tested 6.4 kernel with suspend? Is it still the same thing since 5.12? (Ie: suspend is unstable?) | 18:24 |
| sevan | josch: how does dropping the commands to run at the shell sound and instead, suggesting that the the reader appends the following three deb repos to /etc/apt/sources.list ? | 18:25 |
| josch | sevan: well the command can remain if "tee" is replaced by "tee -a" | 18:27 |
| josch | sevan: but you also have to disable the "unstable" repo if it's in there | 18:27 |
| sevan | sevan: understood. | 18:28 |
| sevan | derp | 18:28 |
| sevan | josch: understood :) | 18:28 |
| josch | sevan: I change the wording a bit -- do you think it's better now? | 18:35 |
| sevan | josch: "3. switch out your Debian unstable entry to bookworm in your /etc/apt/sources.list" switch it out with what? | 18:37 |
| sevan | josch: ah, replace ustable with bookworm, example of what the resulting statement will look like (so someone can just copy it) | 18:39 |
| minute | Boostisbetter: not tested it yet, but should! | 18:39 |
| sevan | josch: it would help if unstable & bookworm were marked up with code tags for those unfamiliar with the sources.list contents (myself included) | 18:42 |
| josch | sevan: what are "code tags"? | 18:43 |
| josch | sevan: you mean html tags? | 18:43 |
| minute | i just tested suspend. the issue is that the sn65dsi86 fails DP locking. i guess it needs a delay | 18:44 |
| sevan | josch: yes, I wasn't sure if you're writing html, or using something else which generates the pages for you. | 18:45 |
| minute | (or retries or something) | 18:45 |
| jfred | working suspend would be amazing! I worry about draining my battery less now that I have the protected battery boards installed (they've already saved me once) but still being able to suspend the machine at night would be great | 18:59 |
| josch | sevan: i changed more stuff. Do you like it better? | 18:59 |
| josch | sevan: I fear the problem with adding more scripting solutions is, that people will just copypaste those without understanding them. | 18:59 |
| + mark_ (~mjw@gnu.wildebeest.org) | 19:01 | |
| - mjw (QUIT: Killed (NickServ (GHOST command used by mark_!~mjw@gnu.wildebeest.org))) (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 19:02 | |
| * mark_ -> mjw | 19:02 | |
| + mark_ (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 19:03 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 19:04 | |
| sevan | josch: it's not good, because you're making the reader roleplay a shell and parse the commands. There's an element of trust that in the worse case scenario I can just copy and paste the command to do "the right thing" | 19:10 |
| sevan | josch: I mean generally ^ | 19:11 |
| sevan | josch: not a comment on your most recent changes. :) | 19:11 |
| sevan | josch: on the latest changes, looks good. Can't try it since I didn't backup /etc/apt/sources.list before I started. short of downloading a new copy of the system image, is there a copy of sources.list somewhere? | 19:13 |
| josch | no, such a copy isn't downloadle anywhere i think | 19:22 |
| josch | i don't like the current version | 19:22 |
| josch | i have to sleep over this | 19:22 |
| sevan | no worries :) | 19:23 |
| + klardotsh (~klardotsh@98.97.115.75) | 19:31 | |
| violet | josch: tried the kernel you gave me. partial success! i now have /dev/dri/card2 and /sys/class/drm/card2-HDMI-A-1 . i dont know how to get sway to use the output though because `swaymsg -t get_outputs` still shows just eDP-1 | 19:31 |
| violet | but that is maybe a sway thing | 19:32 |
| violet | ok it just did not detect my HDMI to VGA adapter. i plugged in a normal HDMI cable and now it sees the display and gets EDID data i am getting an error | 19:40 |
| violet | https://dpaste.com/4NMGEYX33 | 19:41 |
| minute | oh, there is some bug where on a311d reform, update-initramfs -u changes the dtb to imx8mq-mnt-reform2-hdmi.dtb | 19:48 |
| minute | violet: can you try to force a resolution on output HDMI-A-1 via sway.config? | 19:50 |
| minute | also, update-initramfs didn't honor the new modules i added to /etc/initramfs-tools/modules, hmm | 19:54 |
| minute | josch: FYI there are two scripts that load the reform modules in the initrd now: /scripts/init-top/reform and /scripts/init-top/reform-modules-fix.sh | 20:01 |
| minute | josch: looks like some redundancy in there | 20:01 |
| josch | minute: where is reform-modules-fix.sh from? | 20:01 |
| violet | minute: tried that, https://dpaste.com/5LN9SGXPW it cannot modeset it apparently | 20:02 |
| josch | minute: what is in your /etc/flash-kernel/machine? | 20:02 |
| minute | josch: hmm, maybe it could be a leftover from an older version that is still around on my nvme | 20:02 |
| minute | and that got baked into the initramfs | 20:02 |
| minute | ah, root@reform:/etc/initramfs-tools/scripts/init-top# cat /etc/flash-kernel/machine | 20:03 |
| minute | MNT Reform 2 HDMI | 20:03 |
| minute | that's wrong : | 20:03 |
| josch | minute: the location of that script is definitely a leftover | 20:04 |
| minute | ah, but it makes sense | 20:04 |
| josch | minute: the canonical location is /usr/share/initramfs-tools/scripts/init-top/reform | 20:04 |
| minute | aha! | 20:04 |
| josch | if it lives in /etc we cannot upgrade it well | 20:04 |
| minute | thanks, deleted | 20:04 |
| minute | it's totally clear why /etc/flash-kernel/machine is wrong, because i took the drive out of the reform with imx8mq | 20:05 |
| violet | i also have this in my dmesg: [ 632.317143] imx-dcss 32e00000.display-controller: [drm] Cannot find any crtc or sizes | 20:06 |
| minute | violet: are you booting with hdmi connected? | 20:06 |
| minute | ah, nevermind | 20:06 |
| violet | cannot find any crtc seems relevant because that's like, the actual display clock right | 20:07 |
| minute | crtc means cathode ray tube connector | 20:07 |
| minute | :D | 20:07 |
| violet | they live on in our hearts | 20:07 |
| violet | and our kernel code | 20:07 |
| minute | it just means that it thinks nothing is connected / no mode set | 20:07 |
| violet | ah ok | 20:07 |
| minute | aha > Unsupported platform 'MNT Reform 2 with BPI-CM4 Module' | 20:08 |
| josch | minute: we should build patched flash-kernel in reform-debian-packages | 20:09 |
| minute | true | 20:09 |
| minute | what i need there is an entry with: | 20:10 |
| minute | Machine: MNT Reform 2 with BPI-CM4 Module | 20:10 |
| minute | DTB-Id: amlogic/meson-g12b-bananapi-cm4-mnt-reform2.dtb | 20:10 |
| violet | well the connection is definitely there because like i say i got monitor model-number data over the cable, and the monitor itself detects that a connection exists | 20:10 |
| josch | minute: before patching flash-kernel you can write an entry into your /etc/flash-kernel/db | 20:11 |
| violet | i have had troubles with my Rock64 where that device only supported 1080 display though, and could not handle 1280x1024. so i am going to go try it with that 1080p monitor and see if that works | 20:11 |
| minute | violet: what if you run wdisplays in sway? does it show HDMI-A-1? (btw it will get very slow, go to hamburger menu and disable the previews) | 20:11 |
| josch | minute: just copy and adjust the reform entry from /usr/share/flash-kernel/db/all.db | 20:11 |
| minute | ah yeah | 20:11 |
| josch | and later when we patch flash-kernel you can jus rm your custom /etc/flash-kernel/db | 20:12 |
| violet | displays wont run because "no provider of glFenceSync found. Requires one of: Desktop OpenGL 3.2, GL_ARB_sync, OpenGL ES 3.0, GL_APPLE_sync" lol | 20:12 |
| minute | gah | 20:12 |
| violet | but the display shows up in `swaymsg -t get_outputs` as Output HDMI-A-1 'Unknown GH18PS 0323ME0502' (inactive) | 20:12 |
| josch | i remember glFenceSync | 20:12 |
| violet | and GH18PS is the display model | 20:13 |
| violet | it doesnt have any modes available. that log i linked above was from me adding a custom mode using swaymsg 'output HDMI-A-1 mode --custom 1280x1024' | 20:13 |
| josch | re glFenceSync: https://gitlab.gnome.org/GNOME/gtk/-/issues/5749 | 20:14 |
| violet | oh gods you were not kidding about the lag | 20:15 |
| violet | anyways yes HDMI-A-1 shows up in wdisplays | 20:15 |
| violet | i used that GDK_GL=gl workaround | 20:15 |
| violet | but yeah as expected it cannot be enabled from wdisplays either | 20:16 |
| violet | ....... sigh | 20:17 |
| violet | yeah it works just fine with my 1920x1080 display | 20:17 |
| violet | plugged in and the display came right on and is working | 20:18 |
| violet | josch: well good news is your kernel build works for hdmi | 20:19 |
| violet | bad news is that this is now the second embedded graphics hardware ive used that apparently couldnt be bothered to test anything other than 1080p for their kernel driver hahahaha. oh well | 20:20 |
| violet | ill test 1440p later my friend is using my 1440p display right now | 20:20 |
| minute | i vaguely remember that for some clocks one has to add something to the hdmi driver table | 20:20 |
| violet | yeah i also remember seeing a table like that when i was hacking at the rock64 but i could never figure out how i was meant to modify it | 20:21 |
| minute | hmm, looks like my nvme died | 20:23 |
| violet | 1080p is the main thing that matters for me anyway though because i use a 1080p display when im in bed so its good enough! | 20:24 |
| minute | ah no, nvme still works on usb adapter | 20:25 |
| + jacobk (~quassel@47-186-122-163.dlls.tx.frontiernet.net) | 20:48 | |
| minute | josch: for initramfs-tools, we still use /etc/initramfs-tools/modules, right? | 21:06 |
| minute | lol, trying to assemble the right modules list for initrd is such a horror | 21:28 |
| vagrantc | and once you have it figured out ... you get to try again with debian-installer, which has an entirely different way of representing which modules to include! :) | 21:35 |
| - bkeys (QUIT: Remote host closed the connection) (~Thunderbi@static-198-54-130-101.cust.tzulo.com) | 21:37 | |
| minute | vagrantc: nice | 21:38 |
| minute | i'm one step before just dumping allll the loaded modules into that list | 21:39 |
| minute | aha. but it works now. the missing one was "display_connector", hadn't heard of that one before | 21:39 |
| vagrantc | totally valid tactic :) | 21:39 |
| josch | minute: yes, /etc/initramfs-tools/modules is the file filled by reform-system-image | 21:56 |
| josch | finding the list for imx8mq that was needed in /etc/initramfs-tools/modules was also lots of "fun" | 21:56 |
| minute | :D | 22:03 |
| vagrantc | i usually add hooks to /etc/initramfs-tools/hooks or scripts directories ... as you can drop in your own hooks without mangling initramfs-tools managed configuration files ... | 22:07 |
| vagrantc | i've even put hooks in there that just install all the modules :) | 22:08 |
| josch | vagrantc: but /etc/initramfs-tools/modules is not managed by /etc/initramfs-tools/modules, no? | 22:10 |
| vagrantc | ACTION wonders where it comes from | 22:11 |
| vagrantc | i still prefer things that can be isolated ... e.g. this specific part comes from this specific thing ... | 22:12 |
| josch | vagrantc: an you mean that by putting it into /etc/initramfs-tools/hooks/ and scripts you can give it an appropriate name instead of putting everything into the same file? | 22:13 |
| vagrantc | hooks with the various helper functions, e.g. manual_add_modules, add_modules_from_file, copy_modules_dir, etc. just seems a bit cleaner | 22:13 |
| vagrantc | yeah ... isolation of concerns | 22:13 |
| josch | actually that's a good idea | 22:14 |
| josch | this way, we could ship it under /usr with reform-tools | 22:14 |
| vagrantc | right | 22:14 |
| josch | vagrantc: do you have an example pair of hook and script doing the module loading? | 22:14 |
| vagrantc | and you could use add_modules_from_file if you really want a static list too | 22:14 |
| vagrantc | been a while since i've messed with it ... nothing handy at the moment | 22:15 |
| vagrantc | most of the time i didn't need to explicitly load it, it just needed to be present in the initramfs image | 22:15 |
| josch | add_modules_from_file() seems nice -- i'm reading it now | 22:15 |
| vagrantc | and by shipping the stuff in the packages rather than postinst scripts or whatever, you get no-nonsense upgrades :) | 22:16 |
| josch | this is a great idea :) | 22:17 |
| josch | it can also be introduced without touching the existing /etc/initramfs-tools/hooks | 22:17 |
| vagrantc | right | 22:17 |
| josch | and new systems can then have this file in its original state | 22:17 |
| vagrantc | there has been occasional talk of switching away from initramfs-tools ... but every other option always seems needlessly overly complicated and less flexible ... | 22:18 |
| josch | i've heard the talk | 22:18 |
| josch | like switching away from flash-kernel ;) | 22:19 |
| vagrantc | sometimes lack of active development on something is a good sign that it works | 22:19 |
| vagrantc | uh-oh, guilty as charged. | 22:19 |
| josch | i'm now using flash-kernel to install the dtb and u-boot-menu to choose my kernel | 22:20 |
| - lifning (QUIT: Ping timeout: 240 seconds) (~lifning@user/lifning) | 22:23 | |
| + lifning (~lifning@user/lifning) | 22:25 | |
| - jacobk (QUIT: Ping timeout: 252 seconds) (~quassel@47-186-122-163.dlls.tx.frontiernet.net) | 22:37 | |
| vagrantc | i swear recent u-boot-menu has an option to copy all the .dtb files | 22:39 |
| vagrantc | maybe it is just a merge request i never pulled in | 22:40 |
| vagrantc | josch: there's zz-sync-dtb in /usr/share/doc/u-boot-menu/examples | 22:42 |
| josch | vagrantc: that copies all dtbs but it doesn't create the symlink to the dtb we want | 22:43 |
| vagrantc | why do you need the symlink? broken versions of u-boot? | 22:47 |
| vagrantc | e.g. the name of which file you want should be available in u-boot's environment ... and u-boot-menu supports it with fdtdir | 22:48 |
| josch | vagrantc: then how do you switch the dtb if you want hdmi? | 22:48 |
| vagrantc | recompile and reinstall u-boot, of course :) | 22:48 |
| vagrantc | to have an image bootable by N possible incompatible boards, you're going to have to rely on the bootloader telling you the information you have ... | 22:49 |
| vagrantc | custom .dtb entries ... are much more tricky :) | 22:50 |
| vagrantc | josch: so you're essentially using u-boot-menu to select between hdmi and non-hdmi support? | 22:51 |
| josch | vagrantc: but the image contains the bootloader | 22:51 |
| vagrantc | if you get really lucky, you can install multiple bootloaders at compatible offsets ... but that falls down pretty fast :) | 22:51 |
| vagrantc | now i'm starting to talk like the tow-boot folks :) | 22:52 |
| vagrantc | you can have a single image and either boot from some other device, or install the bootloader to the image for your specific board | 22:53 |
| vagrantc | but not if you rely on symlinks ... then you'd have to adjust the symlinks as well | 22:53 |
| josch | vagrantc: are you just thinking out loud or do you have a proposal? | 22:54 |
| josch | so we soon have different boards -- some similar some very different | 22:54 |
| josch | i don't think we'll have a situation where the same image will boot on all of them | 22:55 |
| josch | right? | 22:55 |
| vagrantc | josch: somewhat thinking out loud, yes ... | 22:55 |
| vagrantc | but everything but the bootloader should work across the boards, no? | 22:55 |
| vagrantc | presuming you can pick the correct .dtb ? | 22:56 |
| josch | yes | 22:56 |
| josch | and assuming that the boards all work with the same u-boot | 22:56 |
| vagrantc | ok, i was assuming a different u-boot for each board, which is somehow otherwise installed (e.g. separate device, installed to the image before booting, etc.) | 22:57 |
| vagrantc | like debian-installer images which have the bootloader part, and the image | 22:57 |
| vagrantc | basically, to avoid shipping N different images for N boards, where only the bootloader/bootfirmware/u-boot differ | 22:58 |
| josch | vagrantc: my plan was to do that on the server to avoid endusers having to dd it together | 23:01 |
| - deflated8837 (QUIT: Read error: Connection reset by peer) (~deflated8@50.53.206.61) | 23:01 | |
| + bkeys (~Thunderbi@static-198-54-130-101.cust.tzulo.com) | 23:01 | |
| vagrantc | josch: ok, so you solve it a different way | 23:02 |
| josch | the operation is the same -- just the location where it's carried out will be different from what d-i does | 23:02 |
| vagrantc | sure | 23:02 |
| josch | i run mmdebstrap once to build / and then dd that onto N different filesystem images with the correct choice of u-boot | 23:03 |
| vagrantc | works fine for a handful of images | 23:03 |
| vagrantc | but keeping it simple for other folks is nice too | 23:04 |
| + deflated8837 (~deflated8@50.53.206.61) | 23:18 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
