2026-01-23.log

- voltaire28_ (QUIT: Ping timeout: 264 seconds) (~jlafon@28.162.2.93.rev.sfr.net)00:14
+ pomel0 (~pomel0@user/pomel0)00:32
gordon1til: if you run mpv while another mpv is playing another video (but not on screen), sometimes rk3588 hwdec bails out and just displays green screen only00:34
- kensanata (QUIT: Ping timeout: 245 seconds) (~alex@user/kensanata)01:04
- Ar|stote|is (QUIT: Quit: No Ping reply in 180 seconds.) (~linx@149.210.3.83)01:07
+ Ar|stote|is (~linx@149.210.3.83)01:13
- paperManu (QUIT: Ping timeout: 264 seconds) (~paperManu@146.71.9.156)02:57
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:20)03:40
- nsc (QUIT: Ping timeout: 246 seconds) (~nicolas@i5C74DD66.versanet.de)03:47
+ nsc (~nicolas@i5C74DDCC.versanet.de)03:49
+ voltaire28_ (~jlafon@28.162.2.93.rev.sfr.net)04:07
- voltaire28_ (QUIT: Ping timeout: 246 seconds) (~jlafon@28.162.2.93.rev.sfr.net)04:11
- mjw (QUIT: Ping timeout: 260 seconds) (~mjw@gnu.wildebeest.org)04:28
- paperManu_ (QUIT: Ping timeout: 264 seconds) (~paperManu@146.71.9.156)04:38
+ voltaire28_ (~jlafon@28.162.2.93.rev.sfr.net)05:06
- voltaire28_ (QUIT: Remote host closed the connection) (~jlafon@28.162.2.93.rev.sfr.net)05:08
- Ar|stote|is (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~linx@149.210.3.83)05:48
- lidstah (QUIT: Remote host closed the connection) (~lidstah@gateway/tor-sasl/lidstah)06:47
+ lidstah (~lidstah@gateway/tor-sasl/lidstah)06:50
+ Ar|stote|is (~linx@149.210.3.83)08:04
+ kensanata (~alex@user/kensanata)08:52
joschbkeys: hello! I have not forgotten you and I have tried out Fedora again recently.09:32
joschbkeys: Things got delayed because of the mess around gdm last weekend :(09:32
joschbkeys: But I need your help to get further. :)09:32
joschbkeys: Mainly I'd like to know how you got a kernel with the right patches and dtb into Fedora. Did you just hack it in or is there a better way?09:33
cararemixedHmm. Got the dreaded blank screen again on GDM. I think we've got a regression there.09:43
cararemixedI'm also failing to install the fw deps while I was going to go back and look at the keyboard firmware changes I have in mind. Anyone with better apt-fu than me have pointers?09:46
cararemixedhttps://www.irccloud.com/pastebin/CInWFY5y/09:46
cararemixedgcc-arm-none-eabi is at 15:14.2.rel1-1+b109:47
cararemixedNot sure how to convince it to try.09:47
joschcararemixed: you don't. It's just broken right now in unstable.09:49
cararemixedsigh09:49
cararemixedso no other trivial way to build fw on this device w/o cooking my own toolchain?09:50
joschcararemixed: you can build on this device but not with unstable right now.09:50
joschcararemixed: about gnome, do you have journalctl output to look at? Last time gdm was broken the error was hidden in journalctl09:51
cararemixedok, how do I accomplish this? a chroot with stable?09:51
joschyes, you can build inside a chroot09:51
joschcararemixed: the reason gcc-arm-none-eabi is broken right now is that Debian currently rebuilds nearly all arm64 Debian packages to enable GCS09:52
joschand since there are several tens of thousands of packages, build daemons take some time to crunch through them09:52
joschthe gdm issue also turned out to be a problem because of this kind of rebuild09:53
joschand that was more than a week ago...09:53
cararemixedSo this new GDM failure is a new package that is lagging I take it?09:54
cararemixedI went ahead and disabled it so the machine is still usable at least.09:55
joschcararemixed: i have no idea, i have not seen the issue yet09:57
joschwhat did you disable? gdm?09:57
cararemixedgdm yes10:01
joschokay, i'm trying to reproduce the gdm breakage10:02
joschcararemixed: cannot confirm. gdm comes up for me with all packages upgraded to latest in unstable10:05
cararemixedok odd. I failed an hour ago. I just ran update again and see a few new packages. I'll see if these fix it.10:09
cararemixedI had updated last night while hoping to get fw building10:10
joschif it still fails, maybe a look at journalctl output can find the culprit10:12
cararemixedYeah. Failed. Back in and checking the log10:16
cararemixedSome wireplumber and HDMI stuff on top of the usual crash I see in dmesg.10:40
cararemixedhttps://www.irccloud.com/pastebin/XSPWH6GQ/dmesg10:41
cararemixeddmesg output doesn't seem unusual.10:41
cararemixedGonna go make tea, Janine is sick so I need to attend to them for a bit. I'll  poke some more after10:42
joschcararemixed: the last time, the error appeared in the unit org.gnome.Shell@wayland.service10:42
josch(in case you don't want to read all of journalctl)10:42
cararemixedI will scroll through again. Didn't see many error or warn level lines but perhaps it's hidden with a different log level10:49
cararemixedsorry for the chopped lines https://www.irccloud.com/pastebin/FELxYNnY/11:06
cararemixedThere are these and then some other errors after I manually start sway. dbus and mesa related11:06
cararemixedIt's hard to tell exactly what's going on given how noisy the log is11:07
cararemixedI'm going to go find my known good SD card and check journalctl -b on that for a fresh boot and see what is and isn't there.11:08
cararemixedsame errors11:27
cararemixedback on the broken system11:27
cararemixedI can't start a gnome-session but sway works11:27
cararemixedNo idea what's broken as this point11:30
cararemixedIt was booting fine last night.11:31
- pasta161 (QUIT: Quit: WeeChat 4.8.1) (~pasta161@user/xha)11:35
+ pasta161 (~pasta161@user/xha)11:52
- schalken (QUIT: Ping timeout: 244 seconds) (~schalken@117-118-178-69.gci.net)12:11
- Ar|stote|is (QUIT: Ping timeout: 264 seconds) (~linx@149.210.3.83)12:15
+ schalken (~schalken@117-118-178-69.gci.net)12:18
+ Ar|stote|is (~linx@149.210.3.83)12:20
+ paperManu (~paperManu@146.71.9.156)12:28
- Chwoka (QUIT: Ping timeout: 260 seconds) (~Chwoka@216.73.127.21)12:49
+ mjw (~mjw@gnu.wildebeest.org)12:50
- pomel0 (QUIT: Ping timeout: 264 seconds) (~pomel0@user/pomel0)13:30
+ pomel0 (~pomel0@user/pomel0)13:37
- pomel0 (QUIT: Ping timeout: 240 seconds) (~pomel0@user/pomel0)13:49
+ pomel0 (~pomel0@user/pomel0)13:49
cararemixedI'm looking at the delta between my nvme system installation and the sd card which is fully upgraded and working14:00
- paperManu (QUIT: Ping timeout: 246 seconds) (~paperManu@146.71.9.156)14:05
cararemixedsigh. I have projects I want to focus on. I'm going to remigrate from the working image.14:17
cararemixedIf anyone else hits this on rk3588/pocket/rcorev1 lmk14:17
- pomel0 (QUIT: Ping timeout: 250 seconds) (~pomel0@user/pomel0)14:36
+ pomel0 (~pomel0@user/pomel0)14:37
+ paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca)14:43
- mjw (QUIT: Ping timeout: 250 seconds) (~mjw@gnu.wildebeest.org)15:35
* mark_ -> mjw15:43
gsoraminute: you mentioned being still on 6.18.3, is it still available in reform's repos?15:49
gsoraapt says it's referred by from another package 🤔15:52
+ bkeys1 (~Thunderbi@66.110.201.50)15:56
- bkeys1 (QUIT: Read error: Connection reset by peer) (~Thunderbi@66.110.201.50)15:56
- bkeys (QUIT: Ping timeout: 250 seconds) (~Thunderbi@98.19.128.69)15:56
+ bkeys (~Thunderbi@66.110.201.50)15:56
- bkeys (QUIT: Read error: Connection reset by peer) (~Thunderbi@66.110.201.50)15:59
+ bkeys (~Thunderbi@66.110.201.50)15:59
- bkeys (QUIT: Ping timeout: 260 seconds) (~Thunderbi@66.110.201.50)16:04
- lidstah (QUIT: Remote host closed the connection) (~lidstah@gateway/tor-sasl/lidstah)16:04
+ lidstah (~lidstah@gateway/tor-sasl/lidstah)16:05
- jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu)16:10
+ jogu (~jogu@user/jogu)16:26
- paperManu (QUIT: Ping timeout: 244 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca)16:31
minutei'm back at MNT16:32
minutei'm gonna try barebox and then latest kernel stuff on pocket16:32
minutebut i need to remember how to nuke uboot from emmc without killing the whole part table16:32
minuteah yeah, machine confs have the answer16:34
minute> # idbloader.img is expected at an offset of 32768 bytes and u-boot follows 8355840 bytes after16:35
minuteso lets kill some blocks at 3276816:35
gsoragood luck16:35
minute(64 sectors)16:35
+ paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca)16:36
vkoskivmeanwhile I've fallen in the obsidian rabbit hole. this program will surely fix all my problems, just so long as I figure out The Perfect Tagging System in one go!16:37
minuteok, erasing 2 blocks at seek=64 bs=512 successfully broke boot16:39
- paperManu (QUIT: Ping timeout: 264 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca)16:41
+ eery (~eery@192.150.253.237)16:41
^alexooh pocket work16:42
^alexwe need to replace our charger board when the new one comes in16:43
+ paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca)16:43
^alexwe don't feel like leaving it plugged in with the potentially-explodive board16:43
- jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu)16:49
eerysooo I updated my a311d reform, and nvme disappeared lol. Does downgrading to 6.17 fix it or will I need to fiddle with u-boot?16:49
- Batto_ (QUIT: Ping timeout: 265 seconds) (~joona@user/Batto)17:01
+ Batto (~joona@user/Batto)17:02
joschminute: reform-flash-uboot --zero emmc17:02
joscheery: what system is this? Pocket Reform?17:03
+ siviq (~siviq@user/siviq)17:05
eeryjosch: no, the big one :)17:06
joschokay, that is surpsing17:06
joschso far, the reports only came from pocket reform users17:06
joschi'd like to figure out what the problem is but it works fine for me: https://community.mnt.re/t/a311d-wont-boot-no-keyboard-input-after-kernel-6-18-3-update-temporary-fix/4151/1217:07
joscheery: are you thrown into an emergency shell?17:07
eeryyep, drops to the ramdisk shell, where I can confirm the nvme devices are missing. Other data point: my boot partition/boot blocks are on an SD card17:09
eeryI flashed the latest system image to an SD card and it boots fine, but still doesn't see nvme17:09
joscheery: maybe only some nvme drives have a problem with 6.18? Which one do you have? WD blue here17:12
eeryI'm seeing the same IRQ error in the kernel log as the poster though17:13
eeryuhhhh let me check17:13
eeryadvantages of transparent base :)17:13
joschWDC WDS100T2B0C-00PXH0 to be precise17:13
gsoraas a data point, i have a samsung 990 pro drive in my reform, which is also having pcie issues17:14
eeryThis one is a Crucial P317:14
- pomel0 (QUIT: Ping timeout: 240 seconds) (~pomel0@user/pomel0)17:24
minuteok barebox up on my rk3588 pocket17:28
minuteincl display17:28
minuteand usb and pcie too, but usb keyboard not found yet17:28
minutejosch: oh nice @ --zero17:28
gsoranice! does barebox display work on classic as well?17:29
+ pomel0 (~pomel0@user/pomel0)17:30
joschminute: you can also use --zero with sd-cards for foreign machines. Using the --machine option I can use --zero on an sd-card for RK3588 even though my system is running A311D.17:30
joschI destroyed my partition tables one time too often :D17:30
+ jogu (~jogu@user/jogu)17:33
minutejosch: very nice17:35
minutegsora: "only" HDMI, not DP yet or DSI->eDP bridge17:35
minuteneed to see how easy/difficult the DP stuff is to port17:36
joschgsora: but you had classic reform with rk3588 not a311d, right?17:36
gsorajosch: yes, my bad!17:37
joschgsora: can you remind me whether your ssd problem was with u-boot 2026-01-11 or with kernel 6.18?17:37
gsora6.18, but it's on the wifi/pcie side17:39
- siviq (QUIT: Ping timeout: 272 seconds) (~siviq@user/siviq)17:39
gsorasorry for the confusion, tried too hard to pattern match 17:39
gsorathis is how LLMs must feel17:39
minutehehe nice, there's a login option for barebox17:39
joschgsora: not with only 5 data points :D17:39
gsoraminute: better than nothing! hdmi is uge17:39
gsora*huge17:40
+ liberto (~liberto@user/libercv)17:47
minutenow trying > bootm -o /mnt/mmc0.0/dtb-6.17.13-mnt-reform-arm64 -r /mnt/mmc0.0/initrd.img-6.17.13-mnt-reform-arm64 /mnt/mmc0.0/vmlinuz-6.17.13-mnt-reform-arm6417:49
minutekernel is kerneling17:52
minuteserial console is behaving bizarrely around nvme unlock17:52
minutebut > Debian GNU/Linux forky/sid pocki3000 ttyS217:52
josch\o/17:53
minuteso it can boot! just display is empty :D17:53
minuteah > [    3.298960] dw-mipi-dsi2 fde30000.dsi: command interface is busy17:55
- marty (QUIT: Quit: WeeChat 4.6.3) (~marty@172.59.96.54)18:07
+ Chwoka (~Chwoka@216.73.127.21)18:07
minutehdmi still works and system works in general when booted from barebox...18:19
minutehdmi in barebox also works automatically when it's plugged in18:27
minuteinteresting https://www.barebox.org/doc/latest/user/booting-linux.html#boot-loader-specification18:32
minuteoh interesting > usb1-0: Bus 001 Device 002: ID 1209:6d07 Pocket Reform System Controller18:36
joschminute: that's the format of extlinux.conf18:36
minutejosch: oh it is?18:37
joschyes, minus the labels which are optional18:38
josch(if you don't have multiple entries)18:38
- liberto (QUIT: Quit: WeeChat 4.8.1) (~liberto@user/libercv)18:40
joschaha kernel-install is a tool to create /boot/loader/entries/*.conf files18:45
minuteinteresting18:55
joschbut we are back at the question: how does it auto-choose the device tree path? :)18:56
minutejosch: why would it auto choose?18:59
minutefor me the dtb path is in the extlinux.conf19:00
minutestill wondering why system controller is found, but keyboard isn't... 19:03
joschminute: i'm not able to find which key barebox expects in /boot/loader/entries/*.conf files to find the dtb and for kernel-install i read that if it indeed is following the uapi spec then it should be looking into the /dtb directory which already does exist (created by flash-kernel) but how does it select the right dtb from there?19:05
joschin extlinux.conf we do that by *not* manually selecting the dtb but instead we just set dtbdir option and then it's up to u-boot and its ${fdtfile} value to know which dtb to load from the fdtdir directory19:06
minutei thought /dtb is a symlink to a dtb19:08
minute> lrwxrwxrwx 1 root root       64 Jan 14 19:54 dtb -> dtbs/6.18.3-mnt-reform-arm64/rockchip/rk3588-mnt-reform-next.dtb19:08
minuteand for me the extlinux entries have > fdt /dtb-6.18.3-mnt-reform-arm6419:09
minutewhich is a symlink > lrwxrwxrwx 1 root root       64 Jan 14 19:54 dtb-6.18.3-mnt-reform-arm64 -> dtbs/6.18.3-mnt-reform-arm64/rockchip/rk3588-mnt-reform-next.dtb19:09
minutejosch: ok i opened common/blspec.c19:11
minutejosch: > devicetree = blspec_entry_var_get(entry, "devicetree");19:11
joschokay, nice :)19:11
minuteso it's a bit different than extlinux' fdtfile19:12
joschminute: having the fdtdir option in extlinux.conf is necessary to have the reform-system-any images19:12
joschminute: you have it selected explicitly because your system at some point ran u-boot-menu19:12
minuteah ok19:12
joschwithout u-boot/barebox auto-selecting the fdt, reform-system-any images cannot work19:13
joschthis was also the background of my efi question19:13
minutegot it19:13
joschyes, i know how to create my own uki blob with the dtb baked in19:13
joschbut that doesn't help us for the any images19:13
minuteright. so there's multiple ways to embed this information in the barebox build of course19:13
joschand since the fedora image ships hundreds of dtb files and uses efi booting i thought that it would somehow work19:14
minuteor do you want a bootloader that figures out itself what system it's running on?19:14
joschno, the bootloader is built for platform X, so it knows which dtb it needs19:14
joschthe idea of the any images is that the system already has u-boot/barebox on emmc19:14
joschand then you can pop in the same image, irrespective of the board19:14
minuteyes. and which piece of information is still missing?19:15
joschhow does barebox know to load /dtbs/${kernelver}/rockchip/rk3588-mnt-reform2.dtb from a reform-system-any image?19:15
joschbarebox built for rk3588-mnt-reform2 that is19:16
minutejosch: ah, so it knows "rk3588-mnt-reform2.dtb", but "/dtb/$kernelver/rockchip/" needs to come from the extlinux.conf, like fdtpath/fdtdir or what it was called?19:18
joschit knows rockchip/rk3588-mnt-reform2.dtb, that's what u-boot has stored in ${fdtfile}19:19
joschand it's given /dtbs via fdtdir19:19
minutejosch: but why only /dtbs, not /dtbs/kernelver?19:20
joschand ah sorry19:20
minutejosch: ok so then fdtdir support is missing in barebox right?19:20
minutejosch: there's multiple ways to add that in i guess...19:20
joschfdtdirs is /dtbs\\/'\"\$kvers\"'\\19:20
joschminus the quoting mess19:20
joschjust looked it up in mkimage.sh19:21
minuteah19:21
josch"then fdtdir support is missing in barebox" -- no idea, i don't know barebox19:21
minutejosch: we could add that functionality to common/blspec.c, or make a little binary for itthat loads extlinux.conf and boots that19:21
joschright, in the end everything can be hacked in19:22
joschlets see19:22
joschreform-system-any support also doesn't have to be in from day 1 :)19:22
+ siviq (~siviq@user/siviq)19:31
+ mark_ (~mjw@gnu.wildebeest.org)19:44
+ liberto (~liberto@user/libercv)19:47
* mjw -> Guest17720:07
- Guest177 (QUIT: Killed (calcium.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2486:a800:a09a:fc1c:5a8:e74d)20:07
* mark_ -> mjw20:07
eeryis there a way to get a copy of a system image build from a few months back?20:07
+ Guest177 (~mjw@2001:1c06:2486:a800:a09a:fc1c:5a8:e74d)20:07
- siviq (QUIT: Ping timeout: 272 seconds) (~siviq@user/siviq)20:10
joscheery: yes: https://source.mnt.re/reform/reform-system-image/#old-image-downloads20:27
- liberto (QUIT: Quit: WeeChat 4.8.1) (~liberto@user/libercv)20:28
+ marty (~marty@172.59.96.54)20:50
+ siviq (~siviq@user/siviq)21:13
minutenow writing a primitive HID version of the usb kbd driver that so far only does boot protocol21:15
minuteit already recognizes key reports!21:16
- pomel0 (QUIT: Ping timeout: 240 seconds) (~pomel0@user/pomel0)21:48
+ vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:20)21:49
+ pomel0 (~pomel0@user/pomel0)21:49
minutevideo of barebox with keyboard input on rk3588 pocket reform: https://mastodon.social/@mntmn/115946460182908608 cc josch22:12
+ Svp (~svp@2002:4f07:f0bd:0:95e7:dc62:c203:a24)22:16
- pomel0 (QUIT: Read error: Connection reset by peer) (~pomel0@user/pomel0)22:18
Svpman i forgot how much non-x86 annoys me when it comes to getting an OS to work, at this point i might just give up, put in a raspi on the rcm4 and have that do EFI so it'll be a least not AS an annoyance i guess22:18
minuteSvp: what's the problem22:19
Svpjust voicing some annoyances at getting netbsd and less mainstream distros to like the platform (both i.mx8mq and a311d, nothing too specific but it's way above my pay grade...), it's sadly not a new experience and i blame ARM(and riscv) designs for it22:21
minuteSvp: for imx8mq there's even 9front22:28
minuteand genode22:28
minutea311d is not that well supported on non-linux though i guess22:29
Svpyeah i'm aware, ran 9front for a while but i found it not particularly to my taste in design, huge props to the team though. upstream has an android port for the a311d but thats hardly of help and yeah outside of linux it's a wasteland for the most part. i recently obtained a loongson ls3a4000 from china (mips64!), and it managed to boot from an iso right there and then, but that's the privilege of having 22:35
Svpone upstream...22:35
Svp(idk if the whole message went through)22:35
f_Amlogic SoCs are not really well supported by anything non linux22:37
f_closest I saw was OpenBSD for Amlogic g12a (of which A311D is part of) and NetBSD for Libre Computer lepotato22:38
f_and 9front is out of the question22:38
+ pomel0 (~pomel0@user/pomel0)22:39
minuteSvp: in that case rpi cm4 would be def. more suitable for you, right22:42
minutei also wonder if cm5 works in rcm422:42
eeryjosch: thanks! can confirm that the nvme drive appears again with 6.1722:53
- siviq (QUIT: Quit: Client closed) (~siviq@user/siviq)22:59
- paperManu (QUIT: Ping timeout: 256 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca)23:02
minuteahaa cc josch > a3f[mx]>minute: barebox enumerates all blspec files, extracts the compatible of the referenced DT and then takes the one that matches23:04
minuteok, i have a working blspec file23:12
minuteso i can take this pocket reform home with barebox on it :D23:13
minute(lots of cleanup to do, but this is very promising)23:13
+ paperManu (~paperManu@146.71.9.156)23:23
+ paperManu_ (~paperManu@146.71.9.156)23:25
Svpminute: if raspi cm5: it *should* work as-is last i checked, if cm5 with three connectors probably not or not satisfactorily, the orange pi cm5 doesnt even have an ethernet phy so the pins are different23:27
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:20)23:29
minuteSvp: i meant rpi yeah23:30
Svpyeah if i remember well enough from the rpi docs it should be drop-in compatible23:30
- pomel0 (QUIT: Ping timeout: 246 seconds) (~pomel0@user/pomel0)23:31
Svpsome retooling might be required for the new features but it should work...?23:31
+ pomel0 (~pomel0@user/pomel0)23:33
- pomel0 (QUIT: Ping timeout: 260 seconds) (~pomel0@user/pomel0)23:38
+ pomel0 (~pomel0@user/pomel0)23:40
- pomel0 (QUIT: Ping timeout: 246 seconds) (~pomel0@user/pomel0)23:44
+ pomel0 (~pomel0@user/pomel0)23:45

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!