| minute | i also got rid of "funky mode" in the process | 00:08 |
|---|---|---|
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 00:12 | |
| minute | oh we can bump i2c to 400khz | 00:17 |
| minute | WIP https://source.mnt.re/reform/pocket-reform/-/merge_requests/23 | 00:24 |
| minute | why didn't i do this much earlier lol... | 00:27 |
| ch | i might have seen that tinyusb could also run on the second core. but havent looked at tall | 00:32 |
| minute | yes, this was my first thought, but then i was like, lets try interrupts first... because working with core 1 is not 100% trivial (there are some side effects) | 00:35 |
| minute | there's also some way to get the picotool reset hook back in tinyusb hid afaik | 00:36 |
| josch | having the imx8mq reform from BoostisBetter taught me again how it is much more motiving to fix issues that you are personally affected by -- i knew the reports of new waybar reporting wrong battery information from the forum but didn't feel i needed to put this higher on my todo list because i'm running old waybar from bookworm... dogfooding is a thing | 00:37 |
| minute | it totally is! i have a free day and was trying to get steam, wasteland 2 and signalis to run and then i noticed that the builtin keyboard+trackball were not responding anymore in pocket | 00:38 |
| minute | and i also had that frustrated post by emery in the back of my head, and the reports by BoostisBetter here. | 00:38 |
| ch | yeah getting reset to work again is relatively easy | 00:39 |
| ch | i just always thought its deliberately not there | 00:39 |
| minute | ch: yeah maybe i was like, this could be unsafe and then didn't research more... | 00:40 |
| ch | if you dont need it now, i'll see about the reset interface tomorrow or so | 00:41 |
| ch | should share the files between sysctl and hid | 00:41 |
| minute | ch: cool! i don't need it now, would just be nice for later | 00:46 |
| minute | btw i just pushed a fix to that branch for... the shift key functionality that i messed up :D | 00:47 |
| - MyNetAz (QUIT: Remote host closed the connection) (~MyNetAz@user/MyNetAz) | 00:50 | |
| minute | hmm idle power consumption seems too low... between 0.2-0.3A? :D that would be like 1.8-2.4W | 00:51 |
| minute | what kind of power consumption are you seeing on rk3588 pocket reform in battery status page in the oled menu? | 00:52 |
| ch | that would be insane | 00:52 |
| minute | something can't be right about this number yeah... | 00:52 |
| ch | need to charge the pocket for a moment | 00:52 |
| ch | 0.65A idle | 00:57 |
| ch | no graphics | 00:57 |
| ch | + kbd leds off | 00:58 |
| minute | surely the keyboard can't be 50% of that? :D | 00:59 |
| minute | it would be fun though | 00:59 |
| minute | (it now has 5ms sleep in the main loop) | 00:59 |
| + MyNetAz (~MyNetAz@user/MyNetAz) | 01:01 | |
| ch | hmm | 01:02 |
| - mtm (QUIT: Ping timeout: 244 seconds) (~textual@47.202.75.129) | 01:03 | |
| + mtm (~textual@47.202.75.129) | 01:05 | |
| - cobra (QUIT: Ping timeout: 252 seconds) (~cobra@user/Cobra) | 01:08 | |
| ch | rp2040 datasheet claims ~35mA worst case for some intensive workload | 01:09 |
| - vagrantc (QUIT: Ping timeout: 246 seconds) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 01:12 | |
| ch | ok dunno, with keyboard flashed from main branch i get between 0.38 and 0.45A in idle now | 01:21 |
| ch | sometimes peaking up a bit | 01:21 |
| + colinsane (~colinunin@97-113-84-187.tukw.qwest.net) | 01:22 | |
| ch | (i knew the kbd leds take a ton of power) | 01:22 |
| ch | kbd fw from branch seems to settle on 0.40A with peaks upwards | 01:23 |
| ch | unscientifically i want to say the current stays a bit lower | 01:23 |
| ch | old thought of mine: limiting the maximum brightness of the kbd leds would go a long way | 01:26 |
| ch | but dunno what the design goal was :) | 01:26 |
| - colinsane (QUIT: Ping timeout: 265 seconds) (~colinunin@97-113-84-187.tukw.qwest.net) | 01:28 | |
| minute | well the default could be lower brightness for sure :D | 01:29 |
| ch | that too :) | 01:30 |
| minute | rk3588 has some funny features like hardware zlib etc decompressor | 01:38 |
| minute | also jpeg decoder :D | 01:39 |
| + colinsane (~colinunin@97-126-1-93.tukw.qwest.net) | 01:42 | |
| ch | i guess no support in zlib for that :/ | 01:44 |
| ch | Voice Activity Detect (VAD) | 01:46 |
| ch | fun | 01:46 |
| ch | EoTp is not supported in DataStream mode | 01:48 |
| minute | hmm | 01:51 |
| - wielaard (QUIT: Ping timeout: 252 seconds) (~mjw@gnu.wildebeest.org) | 03:17 | |
| - paperManu (QUIT: Ping timeout: 265 seconds) (~paperManu@107.159.243.8) | 03:50 | |
| + cobra (~cobra@user/Cobra) | 04:21 | |
| + erle (~erle@user/erle) | 04:26 | |
| - erle (QUIT: Quit: K-lined) (~erle@user/erle) | 04:49 | |
| - bpye (QUIT: Quit: The Lounge - https://thelounge.chat) (~bpye@user/bpye) | 05:02 | |
| + bpye (~bpye@user/bpye) | 05:09 | |
| + erle (~erle@user/erle) | 06:40 | |
| - erle (QUIT: Quit: K-lined) (~erle@user/erle) | 06:47 | |
| - bpye (QUIT: Ping timeout: 252 seconds) (~bpye@user/bpye) | 08:34 | |
| + nsc (~nicolas@i5C74DC72.versanet.de) | 09:05 | |
| + bpye (~bpye@user/bpye) | 10:01 | |
| - aloo_shu (QUIT: Ping timeout: 265 seconds) (~aloo_shu@90.166.193.242) | 10:03 | |
| + aloo_shu (~aloo_shu@85.51.18.56) | 10:06 | |
| + jacobk (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 10:33 | |
| BoostisBetter | minute: is using dpms to disable the monitor on the pocket ok? Or does that glitch and sometimes you can't get the monitor back? | 11:44 |
| grimmware | BoostisBetter: on pocket I use dpms off in swayidle, works really well but IMO if you’re fiddling with it it’s worth having a keybinding to manually turn it back on again. | 12:01 |
| BoostisBetter | grimmware: so you set a keybinding to exec dpms on? | 12:17 |
| + wielaard (~mjw@gnu.wildebeest.org) | 12:25 | |
| + gustav28 (~gustav@c-78-82-52-90.bbcust.telenor.se) | 13:03 | |
| - mtm (QUIT: Ping timeout: 252 seconds) (~textual@47.202.75.129) | 13:03 | |
| + mtm (~textual@47.202.75.129) | 13:05 | |
| grimmware | BoostisBetter: yeah, I was doing some stuff with automating turning the screen off and it just meant that I couldn’t cause myself headaches by writing shitty code | 13:24 |
| ryukazou | My pocket reform keyboard rarely will have some weird behaviour happen, it will suddenly stop working and I have to power down pocket reform and reset keyboard in order for it to work again. | 13:25 |
| ch | ryukazou: could try the firmware from here: https://source.mnt.re/reform/pocket-reform/-/merge_requests/23 | 13:27 |
| ryukazou | How could I flash this firmware? | 13:30 |
| ch | click Pipelines -> then the download icon on the newest one, then build:archive. that will give you a zip with the built firmware. then follow the normal instructions | 13:37 |
| ch | (be prepared to not have a working keyboard while you flash it) | 13:37 |
| minute | (i.e. best to have a external usb keyboard connected as well) | 13:39 |
| ryukazou | ch: Got it, just run the script inside artifacts like normal update, thanks. | 13:44 |
| ryukazou | minute: I see, I will update after I got home, thanks. | 13:44 |
| - glu_ (QUIT: Ping timeout: 244 seconds) (~glu@user/glu) | 13:45 | |
| - wielaard (QUIT: Ping timeout: 252 seconds) (~mjw@gnu.wildebeest.org) | 14:07 | |
| minute | ryukazou: i just pushed an update that will have a new convenience script for updating the keyboard. | 14:07 |
| * Guest5386 -> mjw | 14:12 | |
| + paperManu (~paperManu@107.159.243.8) | 14:18 | |
| ch | ah nice | 14:53 |
| ch | so many yaks to shave | 14:53 |
| minute | ch: yaks all the way down! but which one do you mean? :D | 15:07 |
| BoostisBetter | minute: is the link CH posted update you are talking about? | 15:09 |
| minute | BoostisBetter: yes | 15:10 |
| BoostisBetter | OK, when I get home I will try to pull it down an update. | 15:10 |
| BoostisBetter | This is just for keyboard or for the sysctrl? | 15:10 |
| ch | keyboard only | 15:13 |
| ch | minute: :D i have some extra yaks to shave here, but i think the immediate stack is: lets merge the fwupd branch for sysctl, and then stick the same bits into the hid firmware? | 15:14 |
| ch | (extra yaks: seems like my riscv board with the 3d printer software died, which i wanted to use as an airplay receiver too, etc) | 15:15 |
| minute | ch: yes @ fwupd bits | 15:16 |
| ch | minute: so maybe you can review/merge this easy MR: https://source.mnt.re/reform/pocket-reform/-/merge_requests/22 - removes pico-extras from sysctl | 15:18 |
| qbit | is there a guide anywhere for moving an install from the µsd to nvme? | 15:32 |
| ch | the reform-migrate tool should help with that | 15:33 |
| qbit | fancy, ty! | 15:33 |
| BoostisBetter | minute: your update should be run with the latest sysctrl firmware right? | 15:35 |
| - aperezdc (QUIT: Ping timeout: 248 seconds) (~aperezdc@2a03:6000:6e61:633::43) | 15:43 | |
| minute | BoostisBetter: it doesn't matter | 15:56 |
| ch | theres no way of turning on the kbd leds without also turning on the soc, right? | 15:58 |
| ch | (just wondering about the power draw) | 15:58 |
| minute | ch: correct, they need 5v | 15:58 |
| ch | ah, oh well | 15:58 |
| minute | and the 5v rail is leds+usb+soc | 15:58 |
| ryukazou | Keyboard update done, fast and easy. Thanks. | 16:19 |
| qbit | that worked wonderfully! nice work on the reform-* tools.. and everything in general :D | 16:21 |
| qbit | .. also my keymap is working in the console stuff now.. squeeee! | 16:21 |
| ch | at least i managed to close a few browser tabs m( | 16:22 |
| minute | ryukazou: cool, thanks for testing! | 16:31 |
| minute | btw the new fw doesn't hardcode hyper location now for kbd backlight setting via trackball | 16:32 |
| - jacobk (QUIT: Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 16:43 | |
| BoostisBetter | minute: how do we change the speed of the mouseball tracking. That seemed to reset as well. | 16:44 |
| BoostisBetter | minute: and that is awesome on the backlight memory | 16:44 |
| BoostisBetter | minute: also seems like the trackball keys were changed. Right click is now the outer right button | 16:47 |
| minute | BoostisBetter: ahh so you were on a really old fw before | 16:48 |
| minute | one of the first pockets we shipped, then | 16:48 |
| BoostisBetter | right on | 16:48 |
| minute | BoostisBetter: is the trackball now too fast or too slow for you? | 16:48 |
| BoostisBetter | too slow | 16:48 |
| BoostisBetter | amospalla: the new keyboard firmware breaks rmt tools | 16:49 |
| minute | i can tune it in the fw but the question is what would be the best default | 16:49 |
| BoostisBetter | Not sure, but I like it kind of fast | 16:49 |
| BoostisBetter | maybe double what it is now, or .75 of what it is now | 16:50 |
| minute | BoostisBetter: do you have any pointer acceleration configured in wayfire/sway or none? | 16:50 |
| BoostisBetter | minute: I never set up any, but let me double check real quick. That could be a quick fix, and then you could leave that alone. | 16:50 |
| minute | ok i can tweak the default speed later tonight | 16:50 |
| minute | BoostisBetter: just trying to get a baseline... if you have some multiplier already it might be different from what i see | 16:51 |
| minute | but very possible it got slower due to different scanning/reporting interval | 16:51 |
| BoostisBetter | no multipliers in there | 16:51 |
| [tj] | I stopped paying attention, does crowdsupply have pocket reforms to ship tomorrow? | 16:57 |
| BoostisBetter | It says stock is limited, but they do have stock. | 16:59 |
| BoostisBetter | That is crowdsupply saying that though. | 16:59 |
| [tj] | on the last update my shipping date was changed from end of november to jan 3rd | 17:00 |
| ch | i just needed three tries to turn on the soc | 17:00 |
| ch | dunno if thats the kbd fw or sth else | 17:00 |
| ch | bleh, why does fwupd not work now | 17:09 |
| BoostisBetter | minute: so the change in buttons for the trackball is just part of the update? I can retrain no worries, just making sure this is supposed to be like that. | 17:12 |
| minute | BoostisBetter: yes it was intentional for better ergonomics esp for scrolling with one hand | 17:13 |
| minute | [tj]: CS has some pockets in stock | 17:13 |
| BoostisBetter | so just to be sure the far left button is the middle mouse button? | 17:13 |
| [tj] | cool I'll be impatiently waiting for a shipped anouncement all weekend then :D | 17:14 |
| minute | [tj]: but maybe not your configuration. we will resupply starting next week. from now on incl a new display model | 17:14 |
| minute | BoostisBetter: correct | 17:14 |
| [tj] | I'm pretty sure plan9 for the imx8mp pocket reform would be about a days worth of very painful copying and pasting | 17:15 |
| minute | ch: this two/three tries stuff needs to be looked at yeah. it's some uart buffer messup between the two chips | 17:15 |
| minute | [tj]: very possible :D | 17:15 |
| [tj] | my confidence comes from having done it for freebsd | 17:16 |
| minute | [tj]: you can look at pocket reform stock numbers on mouser.com as a workaround | 17:16 |
| ch | looks like regression in fwupd 2.0.3, it still works with 2.0.2 | 17:27 |
| ch | got to be kidding me | 17:29 |
| ch | works now that i try to fill out the bug report | 17:30 |
| minute | uff | 17:32 |
| - yqshao (QUIT: Remote host closed the connection) (~yqshao@user/yqshao) | 17:33 | |
| + yqshao (~yqshao@user/yqshao) | 17:33 | |
| + aperezdc (~aperezdc@2a03:6000:6e61:633::43) | 17:40 | |
| ch | minute: do you want to keep tagging as 'YYYY-MM-DD' or is it ok to switch to 'YYYYMMDD'? | 17:42 |
| ch | the latter being somewhat easier | 17:42 |
| ch | problem at hand is that the build embeds an URL to the source, which f.e. now becomes this: | 17:42 |
| ch | Source: https://source.mnt.re/reform/pocket-reform/-/tags/20241224 | 17:42 |
| ch | but this is missing the dashes | 17:42 |
| minute | ch: latter is fine! | 17:43 |
| ch | cool | 17:43 |
| ch | wow, this is so weird | 17:44 |
| ch | when the sysctl on the pocket restarts, the mac thats attached to the same charger notices some hiccup on the powerdelivery | 17:44 |
| ch | minute: https://source.mnt.re/reform/pocket-reform/-/merge_requests/21 i think this is as ready as it gets | 17:46 |
| ch | YYYYMMDDz (f.e. 20241224b) wont work as its not a number. if that should be possible, we can strip off the character somewhere | 17:46 |
| minute | ch: ok, the question is how to do hotfix releases if the first release of that day has some error | 18:09 |
| minute | ch: should we do sth like 202412240 (0 for the first release on that day?). or increment the day past the current day... | 18:10 |
| C-Keen | josch: ah I see so, which branch and commit do I need to use now? | 18:13 |
| - chrcav (QUIT: Ping timeout: 264 seconds) (~chrcav@user/chrcav) | 18:16 | |
| ch | minute: i figured increasing the day would work | 18:17 |
| minute | ch: ok, lets do that then in emergencies | 18:17 |
| minute | releases from the future | 18:18 |
| ch | ok | 18:28 |
| ch | certainly the easiest thing to do | 18:28 |
| C-Keen | argh build things downloading stuff... | 18:28 |
| ch | minute: do you want me to do anything before merging 21? | 18:28 |
| - aloo_shu (QUIT: Ping timeout: 264 seconds) (~aloo_shu@85.51.18.56) | 18:48 | |
| + aloo_shu (~aloo_shu@85.51.17.59) | 18:50 | |
| minute | ch: ha, merged! | 19:01 |
| ch | \o/ | 19:07 |
| josch | C-Keen: context? | 19:13 |
| josch | ch: why does it need to be a number? | 19:20 |
| ch | josch: because we tell fwupd that its a number; which has defined lower/greater semantics | 19:20 |
| josch | ch: does fwupd grok floating point numbers? | 19:21 |
| josch | or only integers? | 19:21 |
| ch | it has a few formats it understands, i think <int>.<int>.<int> also works and some others | 19:22 |
| ch | probably <int>.<int> would also work | 19:22 |
| josch | 20241224.0 might be better than 202412240 because the latter is 10 times larger than the former and will always sort later than the next, no? | 19:23 |
| ch | whatever we do, we should make up our mind before shipping the first firmware in the update channel | 19:23 |
| josch | ah just use an epoch ;) | 19:23 |
| ch | i cant recall what the other involved components will do with 20241224.0 | 19:24 |
| ch | (hid, etc) | 19:24 |
| ch | i dont think fwupd does epochs :P | 19:25 |
| ch | i know it complains loudly when the old and the new firmware use different version formats | 19:25 |
| josch | of course :) | 19:26 |
| + wielaard (~mjw@gnu.wildebeest.org) | 19:34 | |
| josch | huh, i need help with this | 19:40 |
| josch | i successfully bisected a311d kernel using linus' git and the a311d patches | 19:41 |
| josch | i now started doing the same for imx8mq | 19:41 |
| josch | i'm on 6.12 right now | 19:41 |
| josch | everything builds fine but when booting, u-boot tells me: | 19:41 |
| josch | Bad Linux ARM64 Image magic! | 19:41 |
| josch | this is with the exact kernel config from the booted 6.12 Debian kernel | 19:42 |
| josch | yes, the kernel image is only 11M and not 29M | 19:43 |
| josch | but the kernel i successfully booted on a311d was also only 10.5M | 19:43 |
| josch | is this because imx8mq has old vendor u-boot that somehow does not understand this kernel format? | 19:44 |
| josch | but if so, then why does the Debian kernel with the same config boot? | 19:44 |
| minute | hmmmm | 19:44 |
| minute | another binutils bug or so? | 19:44 |
| josch | good hunch, let me check my binutils version | 19:45 |
| minute | my vmlinuz on this system is 29M | 19:45 |
| minute | must be some magical compression going on @ 11mb | 19:45 |
| minute | so i'm surprised it worked for your a311d | 19:45 |
| josch | maybe magical "head -c 11534336" compression :) | 19:46 |
| minute | hehe | 19:46 |
| minute | josch: what does "file" output on that 11MB vmlinuz? | 19:46 |
| josch | oh | 19:47 |
| josch | why did i not try this before? | 19:47 |
| josch | "gzip compressed data, max compression, from Unix, original size modulo 2^32 30315008" | 19:48 |
| minute | weird | 19:48 |
| minute | i get > vmlinuz-6.12.6-mnt-reform-arm64: Linux kernel ARM64 boot executable Image, little-endian, 4K pages | 19:48 |
| josch | which is not the same for the debian kernel which is Linux kernel ARM64 boot executable Image... | 19:48 |
| josch | yes :) | 19:48 |
| josch | so that's the problem i guess? | 19:48 |
| josch | let me try to not make it compressed | 19:48 |
| minute | let me gzip that... | 19:48 |
| minute | oh wow | 19:48 |
| minute | it gzips to 11mb | 19:48 |
| minute | so our kernels are uncompressed thus far | 19:49 |
| minute | bzip2 yields 10MB :D | 19:49 |
| op_4 | hey guys, so i finally have an sd card and am in the process of recovering my borked pocket reform os. I d/l'd latest image, booted no problem, but manual says to run `reform-migrate`, doesn't appear to exist by default on the image? | 19:51 |
| op_4 | (and reform-tools is at the newest version afaict) | 19:52 |
| op_4 | i mean i guess i can dd the sd card to the emmc? or is there something more complicated to installing to emmc? | 19:54 |
| josch | op_4: you probably tried running reform-migrate without sudo? | 19:54 |
| op_4 | josch: indeed, i suppose i should have tried that first. I also look at /usr/bin/reform-* and didn't find it, as well as at https://source.mnt.re/reform/reform-tools/-/tree/main/bin?ref_type=heads but i guess i was looking in all the wrong places. Thank you! | 19:56 |
| C-Keen | josch: newer kernel packages do not create uimg initrds anymore | 19:57 |
| op_4 | ACTION is a little embarrassed | 19:57 |
| C-Keen | I think the vendor uboot only supports the legacy format | 19:57 |
| C-Keen | so the culprit is more likely a bootscript not the kernel itself | 19:58 |
| C-Keen | ah nvm you already found the issue | 19:58 |
| josch | Linux debian 6.12.0-dirty #4 SMP ... | 20:01 |
| josch | yes, unzipping made it work | 20:01 |
| - yqshao (QUIT: Remote host closed the connection) (~yqshao@user/yqshao) | 20:11 | |
| + yqshao (~yqshao@user/yqshao) | 20:11 | |
| ch | https://9to5linux.com/fwupd-2-0-2-firmware-updater-adds-support-for-asus-rog-ally-raspberry-pi-pico | 20:12 |
| ch | til | 20:12 |
| ch | (^F pico ; ^F MNT) | 20:12 |
| * mjw -> Guest4113 | 20:13 | |
| * wielaard -> mjw | 20:13 | |
| minute | ch: oh nice | 20:17 |
| C-Keen | josch: yeah ok, giving the capacity file priority again solves the issue for me. but what remains is how it should be calculated correctly. my hunch is that waybar should also take the min current into account as the lpc code does | 20:37 |
| * jn_ -> jn | 21:21 | |
| ch | weird, the hid rp2040 behaves slightly differently on resetting it | 21:31 |
| ch | i think its basically the same behaviour as shown in https://github.com/fwupd/fwupd/pull/8100 | 21:32 |
| josch | C-Keen: for the pocket reform, a sysctl upgrade fixed things -- i wonder if this is not something that should be fixed in the reform lpc module? | 21:32 |
| ch | minute: https://source.mnt.re/reform/pocket-reform/-/merge_requests/24 adds reset + fwupd to pocket-hid | 21:35 |
| ch | reform-mcu-tool needs to learn about the pipe error, but that can happen separately/later | 21:37 |
| minute | ch: amazing, thanks a ton | 21:38 |
| ch | josch: https://source.mnt.re/reform/reform-tools/-/merge_requests/109 is the fix for the above thing ^ | 21:41 |
| josch | ch: thank you! set to auto-merge :) | 21:52 |
| + chrcav (~chrcav@user/chrcav) | 21:57 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 22:09 | |
| josch | vagrantc: do you have time to talk about u-boot-menu? :) | 22:15 |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-78-82-52-90.bbcust.telenor.se) | 22:15 | |
| vagrantc | josch: maybe? :) | 22:18 |
| vagrantc | ACTION feels lagging behind on all things u-boot* | 22:18 |
| josch | vagrantc: i'm on Debian trixie on the classic reform with imx8mq and installed u-boot-menu | 22:19 |
| josch | vagrantc: i think that's also your configuration? | 22:20 |
| vagrantc | josch: switched to rk3588 recently, but otherwise similar... | 22:20 |
| josch | by default, extlinux.conf does not have fdt nor fdtdir entries | 22:21 |
| josch | is that by design or a bug? | 22:21 |
| vagrantc | that sounds unusual | 22:21 |
| josch | this is with split /boot | 22:21 |
| vagrantc | should have fdtdir, unless there is something amiss | 22:21 |
| vagrantc | split /boot or no, that stuff should have been handled the last several releases | 22:22 |
| josch | i can make fdtdir entries appear by adding U_BOOT_FDT_DIR="/dtbs/" -- notice also the slashes which are required because u-boot-update concatenates variables without slashes between them? bug or feature? | 22:22 |
| josch | i have sh -x output which shows how it is failing | 22:22 |
| josch | and there are several oddities | 22:23 |
| josch | for example, there is this check: [ /usr/lib/linux-image- = "${U_BOOT_FDT_DIR}" ] | 22:24 |
| josch | but with split /boot that can never be true because read-config sets _FDT_DIR="/dtb-" | 22:24 |
| vagrantc | josch: can you try downgrading to u-boot-menu from bookworm, just to see? | 22:25 |
| vagrantc | there were some changes to /boot detection | 22:25 |
| josch | vagrantc: i bisected u-boot-menu already | 22:25 |
| vagrantc | oh, even better :) | 22:25 |
| josch | it broke with this commit: a10af24326bea12aea3f2a72e11d205b60ce1c4f | 22:25 |
| vagrantc | that was my hunch :) | 22:26 |
| josch | then lets continue via a bts bug | 22:26 |
| vagrantc | indeed! | 22:26 |
| josch | but now at least i know that it's a bug and not a feature :) | 22:26 |
| josch | vagrantc: i also just found your mail here: https://lists.debian.org/debian-arm/2017/02/msg00007.html | 22:26 |
| josch | vagrantc: just before you came online, i was fighting some "Bad Linux ARM64 Image magic!" messages of u-boot | 22:27 |
| josch | only to find out that just decompressing the image fixed the problem | 22:27 |
| josch | vagrantc: do you happen to know which knob i have to turn to let upstream linux' bindeb-pkg target create uncompressed images? | 22:28 |
| vagrantc | ugh. | 22:28 |
| vagrantc | not off the top of my head | 22:28 |
| josch | okay | 22:28 |
| ch | # CONFIG_COMPRESSED_INSTALL is not set | 22:29 |
| ch | is my tip | 22:29 |
| josch | ch: thank you! | 22:30 |
| ch | minute: feel like merging this too? :) https://source.mnt.re/reform/pocket-reform/-/merge_requests/22/diffs | 22:35 |
| minute | ch: alright! | 22:35 |
| ch | thx | 22:36 |
| josch | for the next person running into this, to build a debian package with an uncompressed kernel for our old boundary vendor u-boot (it's no issue with u-boot on a311d) you run: | 23:09 |
| josch | make KBUILD_IMAGE=arch/arm64/boot/Image bindeb-pkg | 23:09 |
| minute | josch: uff uff | 23:11 |
| minute | josch: thanks for figuring that out | 23:11 |
| josch | that's why copying the .config had no effect | 23:11 |
| josch | it's not something that's stored in the kernel config but passed as a parameter when building... | 23:11 |
| josch | minute: well, it's not relevant anymore for recent u-boot, so it's a non-issue for a311d, rk3588 and anything else in the future :) | 23:12 |
| vagrantc | josch: ah, i *think* i was running something based on mainline u-boot when i was running the imx8mq | 23:21 |
| + bkeys (~Thunderbi@173.16.175.75) | 23:23 | |
| josch | yes, i think bluerise put in lots of work except that their fork does not have display support (yet) which is why i'm personally still using the vendor fork... | 23:23 |
| vagrantc | ah. | 23:24 |
| vagrantc | so, now that i've moved on to rk3588 ... how many patches in the kernel, u-boot, etc. needing mainlining work? :) | 23:25 |
| vagrantc | josch: i have scoured that breaking u-boot-menu commit and wow it must be something subtle | 23:26 |
| vagrantc | it *looks* to just be moving some things from cut-and-pasted code into a function | 23:27 |
| - aperezdc (QUIT: Ping timeout: 265 seconds) (~aperezdc@2a03:6000:6e61:633::43) | 23:28 | |
| josch | vagrantc: are you using u-boot-menu right now? | 23:28 |
| josch | probably not because you don't have display support, right? | 23:28 |
| vagrantc | josch: right now i'm on an x86 machine :/ | 23:29 |
| josch | boo ;) | 23:30 |
| vagrantc | but definitely using it on other machines with split boot and so on, and even though mostly running bookworm, i had the newer u-boot-menu on at least one of them | 23:30 |
| vagrantc | so i am a little surprised ... but will have to dig into it later | 23:30 |
| josch | then let me make sure to send a mail before going to bed :) | 23:32 |
| vagrantc | feel free to cut-and-paste from irc :) | 23:36 |
| josch | vagrantc: thank you! you now have https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1091979 -- thanks in advance :) | 23:58 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!