| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 00:57 | |
| josch | vagrantc: i have linux 6.12.32 boot successfully on rk3588 pocket | 01:04 |
|---|---|---|
| josch | vagrantc: i had to push another fix into ./linux/patches6.12 for that to work: https://source.mnt.re/reform/reform-debian-packages/-/commit/240a971740c9f9bb2d40ffe72a6d32f066d34bc8 | 01:06 |
| vagrantc | josch: will try it, thanks! :) | 01:18 |
| - mjw (QUIT: Ping timeout: 276 seconds) (~mjw@gnu.wildebeest.org) | 01:21 | |
| - robin (QUIT: Quit: Leaving) (~robin@user/terpri) | 01:42 | |
| - svp (QUIT: Quit: Gateway shutdown) (~svp@host-79-7-240-189.business.telecomitalia.it) | 01:54 | |
| + robin (~robin@user/terpri) | 01:55 | |
| + svp (~svp@host-79-7-240-189.business.telecomitalia.it) | 01:57 | |
| - AnimaInvicta (PART: !!unknown attribute: msg!!) (~AnimaInvi@88-120-179-216.subs.proxad.net) | 02:03 | |
| kfx | if I'm booting via extlinux.conf from u-boot, and I set fdtdir as normal | 02:25 |
| kfx | how does u-boot know whether to load rk3588-mnt-reform2 or rk3588-mnt-reform2-dsi? | 02:25 |
| vagrantc | for it to work correctly, your u-boot should know which platform you're using ... unless there is some auto-detection login inside u-boot, in which case it should set the appropriate environment variable (fdtfile?) | 02:32 |
| vagrantc | auto-detection logic ... | 02:33 |
| vagrantc | it is possible to run a preboot script or soemthing too ... not sure how it is supposed to work on mnt/reform at the moment | 02:33 |
| kfx | ah I see | 02:37 |
| kfx | there are two different u-boot builds | 02:37 |
| kfx | so, no autodetect, you have to switch builds. makes sense now, thanks | 02:37 |
| vagrantc | that's the simple old-school way | 02:39 |
| kfx | simple is good. it keeps me from having to sift through thousands of nearly-identical commits in some abandoned bsp tarball | 02:40 |
| vagrantc | there are a handful of platforms that are able to detect between variants if there is in fact some way to detect the difference, but most u-boot builds have a pretty 1-to-1 correspondence | 02:40 |
| vagrantc | josch: hrm. fails to build rk3588-mnt-reform2.dtb with an syntax error ... Label or path hdptxphy_hdmi0 not found | 03:08 |
| vagrantc | it seems to keep building despite the failure, though :/ | 03:08 |
| vagrantc | https://paste.debian.net/1381510/ | 03:10 |
| vagrantc | i guess with 8 cores and 9 build jobs ... "drivers" takes forever to finish | 03:21 |
| - paperManu (QUIT: Ping timeout: 268 seconds) (~paperManu@72.10.128.164) | 03:51 | |
| - nsc (QUIT: Ping timeout: 272 seconds) (~nicolas@204-96-142-46.pool.kielnet.net) | 03:53 | |
| + nsc (~nicolas@14-97-142-46.pool.kielnet.net) | 03:55 | |
| - op_4 (QUIT: Remote host closed the connection) (~tslil@user/op-4/x-9116473) | 04:05 | |
| + op_4 (~tslil@user/op-4/x-9116473) | 04:05 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 04:22 | |
| - Giraut (QUIT: Remote host closed the connection) (~Giraut@84-230-5-4.elisa-mobile.fi) | 06:44 | |
| + reform24306 (~jamie@s207-81-194-51.bc.hsia.telus.net) | 06:59 | |
| reform24306 | Just got my pocket reform today and I love the flexability! Keyboard a tad weird so I cannot type as fast yet. | 07:00 |
| - reform24306 (QUIT: Client Quit) (~jamie@s207-81-194-51.bc.hsia.telus.net) | 07:00 | |
| josch | minute: i recently reported that display backlight does not come up with linux 6.15 until gdm (or the setup wizard) starts. Turns out, this is not deterministic. Sometimes, display backlight come on already when just the tty is showing the boot messages. | 07:18 |
| josch | minute: but that behaviour apparently also existed before 6.15: https://community.mnt.re/t/screen-needs-several-attempts-to-turn-on-a311d/3081 | 07:18 |
| josch | I'm running 6.12 so i'm not experiencing these issues. It's likely, that when I tested the kernel versions after 6.12 I just got lucky and since I didn't test booting multiple times I failed to notice that even before there was a bug making the backlight not come on reproducibly? | 07:19 |
| josch | This is of course even more nasty for users of tuigreet because they don't get to see their login prompt. Switching to gdm would be a workaround for that. | 07:20 |
| + Giraut (~Giraut@84-230-5-4.elisa-mobile.fi) | 08:07 | |
| - Giraut (QUIT: Ping timeout: 248 seconds) (~Giraut@84-230-5-4.elisa-mobile.fi) | 09:08 | |
| + AnimaInvicta (~AnimaInvi@88-120-179-216.subs.proxad.net) | 09:46 | |
| + My-Bosnian-mom-M (~My-Bosnia@109.175.50.243) | 10:10 | |
| - My-Bosnian-mom-M (QUIT: Quit: Client closed) (~My-Bosnia@109.175.50.243) | 10:16 | |
| + My-Bosnian-mom-M (~My-Bosnia@109.175.50.243) | 10:18 | |
| - My-Bosnian-mom-M (QUIT: Quit: Client closed) (~My-Bosnia@109.175.50.243) | 10:23 | |
| josch | minute: gitlab outage? | 11:29 |
| josch | ah HTTP 502: Waiting for GitLab to boot | 11:29 |
| - antti (QUIT: Ping timeout: 244 seconds) (~antti@user/antti) | 11:48 | |
| + antti (~antti@user/antti) | 11:51 | |
| josch | debian-installer on the pocket reform: https://mister-muffin.de/p/PN4L.jpg | 12:00 |
| josch | but i don't understand the terminal font size and rotation despite having fbcon=rotate:3 fbcon=font:TER16x32 in the kernel cmdline: https://mister-muffin.de/p/8ghZ.txt | 12:03 |
| + mjw (~mjw@gnu.wildebeest.org) | 12:20 | |
| minute | josch: yes, outage apparently | 12:25 |
| minute | josch: backlight on classic? with a311d? rk3588? has "something is saving 0 backlight and system restores it" being ruled out? | 12:27 |
| minute | josch: a workaround would be adding something to reform-hw-setup that checks if brightness is 0 and sets to 50% if that is the case | 12:28 |
| bremner | am I right that UBSAN error in rc80211_minstrel_h is a known problem on the 3588 pocket reform? | 12:36 |
| bremner | unrelated, I wonder how my pocket reform managed to drain the battery while plugged in with the lid closed (and powered on) | 12:44 |
| bremner | it's charging again, but it's just odd. | 12:45 |
| josch | minute: backlight on a311d classic reform. This is with a vanilla system image on first boot, so nothing could've saved the backlight before. Also, this is not deterministic. I flash the same content to the sd-card and boot it and get different results. | 13:02 |
| minute | josch: btw gitlab outage was buckket installing security update | 13:12 |
| josch | ah well, that's a good thing to happen :) | 13:12 |
| minute | josch: aha, then i guess driver probing race | 13:47 |
| minute | josch: i.e. depends on which module gets loaded first | 13:47 |
| + chomwitt (~chomwitt@2a02:85f:9a3f:9300:1ac0:4dff:fedb:a3f1) | 14:00 | |
| + Guest51 (~Guest51@ip-062-143-164-159.um16.pools.vodafone-ip.de) | 14:07 | |
| - chomwitt (QUIT: Ping timeout: 272 seconds) (~chomwitt@2a02:85f:9a3f:9300:1ac0:4dff:fedb:a3f1) | 14:10 | |
| - L29Ah (QUIT: Ping timeout: 272 seconds) (~L29Ah@wikipedia/L29Ah) | 14:15 | |
| + enwu (~enwu@user/enwu) | 14:27 | |
| + paperManu (~paperManu@72.10.128.164) | 14:30 | |
| - murphnj (QUIT: Quit: Leaving) (~murph@user/murphnj) | 15:11 | |
| - Guest51 (QUIT: Ping timeout: 272 seconds) (~Guest51@ip-062-143-164-159.um16.pools.vodafone-ip.de) | 15:33 | |
| + chomwitt (~chomwitt@2a02:85f:9a3f:9300:1ac0:4dff:fedb:a3f1) | 15:39 | |
| - chomwitt (QUIT: Ping timeout: 260 seconds) (~chomwitt@2a02:85f:9a3f:9300:1ac0:4dff:fedb:a3f1) | 15:49 | |
| qbit | hola :D - has anyone reprogrammed the keyboard to a different layout? | 16:33 |
| kfx | qbit: look at matrix.h in the firmware source | 16:50 |
| bkeys | josch: How would I go about getting your linux source tarball? | 16:53 |
| kfx | bkeys: the build script pulls debian source and then applies a series of patches depending on the board type | 16:55 |
| kfx | https://source.mnt.re/reform/reform-debian-packages/-/blob/main/linux/build.sh?ref_type=heads | 16:56 |
| kfx | (the patches are also in that repo) | 16:56 |
| - exark (QUIT: Ping timeout: 252 seconds) (~exark@user/exark) | 16:56 | |
| bkeys | Thank you | 16:56 |
| bkeys | I'm hoping instead of me doing all this patching and stuff I can just use this script and feed the tarball he generates into the Fedora packaging | 16:56 |
| josch | bkeys: not quite | 16:57 |
| josch | because you have to actually *do* the packaging | 16:57 |
| josch | err... patching | 16:57 |
| josch | the patches in that repo are Debian packaging patches | 16:57 |
| kfx | that's what I did to make an alpine package | 16:57 |
| josch | so i throw them into the Debian source package and then during the build they get applied | 16:58 |
| kfx | downloaded vanilla kernel source, applied those patches, and pushed to a different repo | 16:58 |
| josch | at no point does a source package get created that has the patches applied | 16:58 |
| kfx | one of the patches doesn't apply to mainline kernel source because the debian kernel source is ... not mainline | 16:59 |
| bkeys | https://gitgud.io/bkeys/linux-reform-rk3588# | 17:00 |
| bkeys | Here is what I have, I applied the patches but for some reason I'm betting .dts files for the Reform in my dts folder instead of .dtb files | 17:01 |
| josch | kfx: oh? which one? | 17:02 |
| josch | kfx: the debian kernel source is from the stable linux branch, not from torvalds' branch | 17:03 |
| bkeys | Wait, nevermind | 17:03 |
| bkeys | I guess they are there now | 17:03 |
| bkeys | Maybe copr took a while to update the package | 17:03 |
| bkeys | I am building off of the Collabara kernel for rk3588 with patches applied | 17:04 |
| josch | minute: the backlight behaviour is also present on a311d classic reform with the latest system image V5 tag. Thus this problem is not new in 6.15. It can be observed in system images with 6.14 and it was observed by users in the forum months ago. | 17:05 |
| josch | minute: thus, do you want to test linux 6.15 before merging or should I hit the merge button right way as I tested it on a311d, imx8mq and rk3588 | 17:06 |
| kfx | josch: I can't get 0063 to cleanly apply on 6.14.11. I haven't brought it up because 6.14 is EOL anyway and I was just going to try the 6.15 stuff soon | 17:06 |
| - antti (QUIT: Quit: antti) (~antti@user/antti) | 17:16 | |
| + antti (~antti@user/antti) | 17:18 | |
| josch | kfx: ah the reason might be that the last release of 6.14 which I rebased the patches onto was 6.14.6 | 17:18 |
| josch | minute: here is an apt sources.list entry to try out linux 6.15: deb [trusted=yes arch=arm64] https://source.mnt.re/reform/reform-debian-packages/-/jobs/11327/artifacts/raw/repo reform main | 17:20 |
| josch | i have it running on my rk3588 pocket since last week | 17:20 |
| josch | on the other platforms i tested sound and wifi next to the obvious things like the display | 17:20 |
| + L29Ah (~L29Ah@wikipedia/L29Ah) | 17:33 | |
| - paperManu (QUIT: Ping timeout: 252 seconds) (~paperManu@72.10.128.164) | 17:47 | |
| + chomwitt (~chomwitt@2a02:85f:9a3f:9300:1ac0:4dff:fedb:a3f1) | 17:53 | |
| bkeys | josch, minute: On rk3588 which set of pins do I hook a serial adapter up to? The ones that point at my face or the ones pointing towards the front of the laptop? | 18:09 |
| bkeys | nvm I got it figured out | 18:15 |
| josch | bkeys: it's probably documented somewhere but i always pull up the kicad schematics these days... | 18:22 |
| bkeys | I found it | 18:23 |
| bkeys | The serial connection isn't showing anything I didn't already know | 18:23 |
| bkeys | kfx: Do you have a tarball of kernel source that's known to enable the rk3588 reform? | 18:24 |
| josch | bkeys: i can also tar the directory with patches applied | 18:26 |
| bkeys | That works, hopefully | 18:26 |
| bkeys | Although in theory there is nothing wrong with my kernel source, the patches are applied there on top of the 6.15 collabara kernel | 18:27 |
| + Giraut (~Giraut@84-230-5-4.elisa-mobile.fi) | 18:29 | |
| bkeys | josch: I'm investigating the MNT Reform u-boot, the maintainer for Fedora for ARM seems to think edk2 is complicating things (I honestly don't know how) and am seeing about uboot booting my Fedora install, but uboot doesn't seem to see either my nvme drive or my USB drive | 18:37 |
| bkeys | And running usb start on uboot runs but it seems to just freeze | 18:37 |
| - cow321 (QUIT: Remote host closed the connection) (~deflated8@user/meow/deflated8837) | 18:41 | |
| + cow321 (~deflated8@user/meow/deflated8837) | 18:42 | |
| qbit | kfx: thanks :D | 18:47 |
| - SavagePeanut (QUIT: Quit: Gateway shutdown) (59eaa45ac7@irc.cheogram.com) | 18:51 | |
| - BoostisBetter (QUIT: Quit: Gateway shutdown) (4a410829d7@irc.cheogram.com) | 18:51 | |
| minute | bkeys: serial pointing towards front (90 degrees), i.e. serial 1 | 18:54 |
| minute | josch: merge away! | 18:54 |
| bkeys | Yeah I got the serial part working | 18:54 |
| bkeys | minute: I can't remember, is setenv the evil command or saveenv? | 18:57 |
| josch | bkeys: maybe don't do either but just edit extlinux.conf? | 19:00 |
| josch | bkeys: (saveenv) | 19:01 |
| josch | setenv sets, saveenv saves your settings | 19:01 |
| bkeys | Well the reform uboot doesn't see my NVME drive or USB drives | 19:11 |
| bkeys | But I can try stuff from SD card, but I'll be afk so I will respond/mess with it further this evening | 19:11 |
| josch | bkeys: i don't think u-boot can do nvme (unless that was changed since the last time i checked) | 19:16 |
| josch | bkeys: we have a separate partition for /boot which is on eMMC or SD-card which contains linux and linux can then load the (potentially encrypted) rootfs from nvme | 19:16 |
| kfx | bkeys: is this the kernel I use https://git.sr.ht/~khm/linux-rcore/archive/6.14.11-2.tar.gz | 20:31 |
| kfx | josch: I have u-boot written to emmc, and it's loading extlinux.conf off my nvme /boot directory | 20:32 |
| kfx | this is reform2/rk3588... reform2/imx8mq can't do this | 20:32 |
| - zeha (QUIT: Quit: reboot) (~ch@user/meow/ch) | 20:39 | |
| josch | kfx: uuuuh sweet! | 20:53 |
| + ch (~ch@user/meow/ch) | 21:07 | |
| ch | josch: feel like trying the build result from here? https://github.com/grml/grml-live/pull/374#issuecomment-2994221534 | 21:07 |
| ch | josch: also - should that boot on a pocket too? | 21:07 |
| josch | ch: ** File not found /boot/grmlghaciarm64/dtbs/freescale/imx8mq-mnt-reform2.dtb ** | 21:22 |
| ch | hmm | 21:23 |
| - hairu (QUIT: Remote host closed the connection) (m-uotkmd@user/hairu) | 21:29 | |
| josch | ch: this probably does not help: mount: /mnt: wrong fs type, bad option, bad superblock on /dev/mmcblk0p1, missing codepage or helper program, or other error. | 21:30 |
| ch | so, issue 1 is that it copied only rk3588-mnt-reform2.dtb | 21:30 |
| + hairu (m-uotkmd@user/hairu) | 21:32 | |
| ch | the mount error seems weirder | 21:33 |
| josch | ch: issue 2 is your partition and filesystem size | 21:33 |
| josch | your filesystem is 260000*4096 = 1064960000 bytes long | 21:33 |
| josch | and your partition is 2048000 * 512 = 1048576000 bytes large | 21:34 |
| ch | right | 21:35 |
| ch | and probably start 512 | 21:35 |
| ch | not 32768 | 21:35 |
| josch | no 32768 is correct because 16MiB offset | 21:36 |
| josch | ch: in my parted call, just divide the numbers by 512 | 21:36 |
| ch | you're telling me the partition offsets are from _after_ the mbr length? | 21:37 |
| josch | wait, sfdisk takes offsets not sizes? | 21:37 |
| ch | iow isn't there now a 512byte hole somewhere? | 21:37 |
| josch | ah that you mean | 21:37 |
| josch | i'm unsure about the sfdisk details | 21:37 |
| josch | i'd hope that the sfdisk offset is from zero | 21:38 |
| ch | yeah i think it is from zero | 21:38 |
| josch | how did you end up with 2048000 anyway? | 21:38 |
| ch | unsure | 21:39 |
| ch | forcepushed, i'll look at it with hexdump later | 21:39 |
| ch | thanks | 21:40 |
| josch | ch: and it should also work on pocket reform but without X11 | 21:40 |
| josch | ch: also instead of going via a temporary $MBRTMP you could use tee | 21:41 |
| ch | wasnt sure about exit code propagation there | 21:42 |
| ch | i'm not good with shell | 21:42 |
| josch | then ignore what i said -- simple is also often better | 21:44 |
| + paperManu (~paperManu@72.10.128.164) | 22:59 | |
| - colinsane (QUIT: Remote host closed the connection) (~colinunin@97-126-5-139.tukw.qwest.net) | 23:31 | |
| + colinsane (~colinunin@97-126-5-139.tukw.qwest.net) | 23:34 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
