2025-06-28.log

- hramrach (QUIT: Ping timeout: 245 seconds) (~hramrach@193.86.92.181)01:17
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50)01:21
+ hramrach (~hramrach@193.86.92.181)01:31
- AnimaInvicta (PART: !!unknown attribute: msg!!) (~AnimaInvi@88-120-179-216.subs.proxad.net)02:06
thejevansminute: I wasn't able to add the products rss feed to my feed reader and it seems to be invalid according to the w3c checker https://validator.w3.org/feed/check.cgi?url=https%3A%2F%2Fshop.mntre.com%2Fproducts.rss02:09
thejevanskfx: were you able to get it to work?02:10
+ photomattmills (~photomatt@192-184-254-28.fiber.dynamic.sonic.net)02:42
- photomattmills (QUIT: Client Quit) (~photomatt@192-184-254-28.fiber.dynamic.sonic.net)02:44
chorcyes, miniflux also refused to recognize it02:44
- paperManu (QUIT: Ping timeout: 252 seconds) (~paperManu@72.10.128.164)03:04
minuteit's not a normal rss feed but google products thingie03:22
thejevansAh okay03:24
kfxthat feed checker is wrong.  it ignores the declared xml namespace and assumes a different one03:51
kfxanyway I parse it with awk so it works fine here!03:51
- mjw (QUIT: Ping timeout: 268 seconds) (~mjw@gnu.wildebeest.org)03:59
+ libercv (~libercv@user/libercv)08:52
+ chomwitt (~chomwitt@2a02:85f:9a3f:9300:1ac0:4dff:fedb:a3f1)09:32
+ gustav28 (~gustav@c-78-82-53-129.bbcust.telenor.se)10:02
+ mjw (~mjw@178.230.108.14)11:17
- mjw (QUIT: Remote host closed the connection) (~mjw@178.230.108.14)11:21
+ mjw (~mjw@178.230.108.14)11:21
- Rookie_one (QUIT: Remote host closed the connection) (~SJ@jegen.dev)11:32
- mjw (QUIT: Ping timeout: 276 seconds) (~mjw@178.230.108.14)11:54
- libercv (QUIT: Quit: Konversation terminated!) (~libercv@user/libercv)11:57
+ mjw (~mjw@178.230.108.14)12:23
- mjw (QUIT: Ping timeout: 276 seconds) (~mjw@178.230.108.14)12:39
+ mjw (~mjw@178.230.108.14)12:42
- mjw (QUIT: Ping timeout: 260 seconds) (~mjw@178.230.108.14)12:49
+ paperManu (~paperManu@72.10.128.164)13:05
- xha (QUIT: Quit: WeeChat 4.6.3) (~xha@user/xha)13:38
+ xha (~xha@user/xha)13:53
minutelpc poweroff is definitely broken. this will be fun to investigate oof14:35
minutecc josch, i saw you responded about this in community 14:36
minutemaybe kernel regression if the trace of O.o is true?14:36
minutealso, only slightly related https://docs.kernel.org/userspace-api/fwctl/fwctl.html14:40
minuteanyway i think this old method in our driver is wrong since the beginning15:42
minutei think we should use this api https://elixir.bootlin.com/linux/v6.15.3/source/include/linux/reboot.h#L6815:47
minuteotoh what ifff shutdown is broken and crashes before poweroff is reached on rk3588?17:22
joschminute: sorry :( and thank you!17:23
joschminute: about kernel regression: the latest system image with kernel 6.15 can do "systemctl poweroff" just fine on A311D classic reform with motherboard 2.517:48
josch(which doesn't mean that the kernel is not wrong, just a datapoint)17:48
+ mjw (~mjw@178.230.108.14)18:06
bkeysminute, josch: So is it a known issue that on KDE the max resolution an external display can be on the Reform is 1024x768?18:27
bkeysbrb18:28
- bkeys (QUIT: Quit: With every step we take, danger will follow closely) (~Thunderbi@h211.16.186.173.dynamic.ip.windstream.net)18:28
minutejosch: aha, that's very good to know. then there's probably a bug in one of the rk3588 drivers which crashes during shutdown18:31
minutejosch: btw i'm investigating audio dropout issues under load. i just discovered that pipewire/wireplumber expects rtkit. i don't have it installed. do you have that installed? just wondering if it is a default 18:32
minutei'm also doing an apt upgrade right now that gives me over 2000 upgraded packages :D18:32
minute(this is partly due to the fact that i run 4 architectures on my pocket)18:33
minutejosch: hmm, my user is not in the pipewire group which is mentioned in /etc/security/limits.d/25-pw-rlimits.conf18:39
minute> # It is up to the distribution/user to create the @pipewire group and to add the18:39
minute# relevant users to the group.18:39
joschminute: no, rtkit is not part of the default system image18:39
minute> # PipeWire will fall back to the RTKit DBus service when the user is not able to # acquire RT priorities with rlimits.18:40
minuteso pipewire can't set process niceness if one is not in the pipewire group nor rtkit is installed18:40
minutejosch: do you know what the debian policy is regarding this? does debian use the group or rtkit? 18:40
joschnot out of the top of my head -- i have to investigate this18:43
joschbut rtkit is also part of the Recommends package set18:43
joschthat hints that the maintainer expects it to be part of the default installation18:43
minuteah ok!18:43
minutethen i'll install it and check if this solves my dropout issues18:44
minutehdmi audio is also very sensitive to this18:44
minutebtw unrelated, there's a new panthor kernel driver written in rust by collabora/arm/google called "tyr" and reference platform is rk3588 currently18:47
+ bkeys (~Thunderbi@173.186.16.211)18:58
- mjw (QUIT: Ping timeout: 252 seconds) (~mjw@178.230.108.14)19:00
minuteafter installing rtkit, pipewire processes now have prio -1119:04
joschminute: pulseaudio and pipewire-bin both Recommend rtkit. I read in the gentoo and archlinux wiki about it and then read the package documentation in Debian as well. As a result I think that rtkit should indeed be part of the default installation: https://source.mnt.re/reform/reform-debian-packages/-/merge_requests/12219:31
minutejosch: great, thanks!19:35
bkeysjosch: Since nvme is giving me grief I want to write a Fedora raw image to the emmc, but the front of the emmc has edk on it. I tried to dd the raw image using dd seek=20 to skip over the UEFI but I don't see the partitions or anything in gparted, which I guess makes sense because the partition table is stored at the front of the device IIRC19:59
bkeysHow would you suggest I put edk2 and Fedora on the same emmc?19:59
joschbkeys: you could re-create the partition layout that i used on the efi debian image20:08
minutebkeys: maybe you could create the partition layout first, while leaving free space for edk2, write edk2 and install fedora to the premade partition from a usb stick and some fedora install tool?20:08
joschthe first sectors are for gpt, then there are 16mbit of free space and then the first partition starts20:08
minuteyeah20:08
bkeysYeah that's what I was thinking about doing was recreating it in gparted, then copying the files over20:08
minuteright, you could also make the partitions like josch says and just copy all the files from the mounted image using archive flag20:08
joschbkeys: also about kde, i could try out what rk3588 reform does for me when i install kde with plasma620:09
bkeysI just really hope edk2 doesn't get messed up, but I suppose if it does I can always load up from SD card20:09
joschbkeys: feel free to ping me in case i forget to try this out20:09
minutebkeys: before rebooting, just make sure to dd edk2 once again so you don't lose it20:09
bkeysjosch: Will do, but right now I really want Fedora installed on something internel, the emmc is 256GB which is plenty of space for me20:09
joschright, much success!20:09
minutepipeline for reform-debian-packages merge of hdmi audio is done20:10
+ mjw (~mjw@gnu.wildebeest.org)20:11
bkeysjosch: So in gparted it says the first 9 megabytes of the Fedora ARM image are unallocated, I think the maintainer does that so that uboot can be put there. (there is no 9 mb of unallocated space at the front of my x86 machine) I think I might be able to write Fedora and then edk20:16
bkeysI'll risk it and do it, worse case I just redo it on an SD card I'm not worried about bricking20:19
minutebkeys: as long as you write edk2 afterwards, you should be fine20:20
bkeysWill that mess up the partition table if I do that?20:20
joschbkeys: for the debian image i leave the first 32768 bytes untouched which seems to be sufficient for gpt and msdos both20:22
minutebkeys: no idea20:22
minuteok, hdmi audio seems stable with rtkit installed20:23
minutevery nice20:23
joschoh funny then it might've been "just" a scheduling issue20:26
bkeysYeah I did the write and it looks like the edk2 did get wiped out, I'll do like minute says and put it back20:26
minutejosch: yes20:26
minutejosch: maybe installing rtkit would also fix hdmi audio on a311d20:27
joschwait, a311d has hdmi audio??20:27
minuteyes, but it's possible i didn't ship that20:27
minutebecause of dropout issues20:27
joschXD20:27
minutei thought something was wrong with the hw20:27
bkeysI'm trying to find the exact command I told minute when I successfully flashed it the first time20:27
joschi'm currently feeding my TV with kodi via 3.5 mm audio line with A311D20:28
minutei wonder if i still have that patch20:28
joschbkeys: did you write it in irc? you have the logs, right?20:28
bkeys sudo dd if=aio-edk2-image.img of=/dev/mmcblk0 bs=4M conv=fsync20:28
bkeysYeah I found it in the logs20:28
bkeysOkay and sha1sum check passes20:30
bkeysminute: My Reform claims it has 256GB emmc and 16GB RAM, is that right?20:32
minutejosch: looks like we shipped hdmi audio https://github.com/torvalds/linux/blob/master/arch/arm64/boot/dts/amlogic/meson-g12b-bananapi-cm4-mnt-reform2.dts#L11820:32
bkeysI thought only the 32GB version used it20:32
minutebkeys: yes in the end i think we got all or most 16gb versions with 256gb emmc20:33
minutea gift from firefly i guess :D20:33
bkeysThat's great, 126 would have been a squeeze but if I can get 256 with a Fedora install on emmc I won't even worry about nvme20:33
bkeysminute, josch: It seems when I flasked edk2 back it wrecked the partition table20:34
minutejosch: i think you need to toggle some knobs in alsamixer to activate it20:34
bkeysI'll just have to create the partitions manually in gparted and cp the content over20:34
minutei think the problem is that edk2 image goes to the very beginning of the disk20:35
bkeysI mean if recreating the image is what I gotta do then I suppose I'll do it20:35
minutewhat's the bootloader position again on rockchip... maybe you can seek and skip the amount of bytes when writing edk2 to leave the part table intact20:36
bkeysI did that initially but it doesn't seem to pick up the partitions that were written by dd, in gparted it looks like nothing happened20:36
minutebkeys: https://github.com/edk2-porting/edk2-rk3588?tab=readme-ov-file#updating-the-firmware20:36
minutebkeys: write fedora to emmc and then use the dd command there20:37
minute> Here we skip the GPT and copy the firmware starting at offset 0x800020:37
bkeysAlright I will try that20:38
- xktr (QUIT: Ping timeout: 244 seconds) (~xktr@user/xktr)20:39
bkeysMan I have had such bad experiences with mpcie wifi on this thing, I think I'll just get a mpcie to USB adapter and use a USB wifi as internal20:39
+ xktr (~xktr@user/xktr)20:40
minutebkeys: what wifi card20:41
joschminute: thank you, i'll have a look in alsamixer20:41
joschbkeys: your dd call does not skip any sectors at the beginning of the disk, so yes it will overwrite the partition table20:41
bkeysI'd have to look at the specific model but lspci says "Qualcomm Atheros AR93xx Wireless Network Adapter (rev 01)20:41
minutejosch: i remember it was quite the puzzle, related to the TDM/DAI numbers you see in the device tree nodes in the link20:42
joschoh no XD20:42
minutebkeys: seems old20:42
minutebkeys: i can recommend intel ax210 ^^20:42
joschbkeys: funnily i had the opposite experience so far -- no usb wifi dongle i tried was any good20:42
joschthe last one i used on my trip to GPN literally got fried20:42
joschit stopped working and was too hot to touch :D20:43
joschso i bought a new dongle for a few EUR and that one would stop working the moment i download a few GB of data over wifi. I then have to wait a bit with wifi turned off and then i can continue downloading. Downloading for too long is a no-go.20:44
bkeysYeah I'm taking a train to Chicago on the 3rd so I'd love to have Fedora all finalized and ready to go and installing to internal memory is the only blocker left20:44
josch(why do i need a usb wifi dongle? because i have the quectel em06 lte card installed in my mpcie)20:44
joschand because the wifi dongle is usb A, i have it hanging from a usb-c hub like this: https://floss.social/@josch/11475777767211336320:46
minutei think the only good usb wifis are mediatek20:46
minutebut anyway limited to 480mbps usb2 afaik20:47
joschminute: i'll buy your pocket pcie upgrade before i buy another wifi dongle :D20:47
bkeysminute: Okay so I ran the command like it says and indeed my Fedora stuff doesn't appear to be at all overwritten20:47
minutebkeys: great20:48
bkeysSo in theory, I expand the partition for my root, and reboot and then hopefully I can boot to emmc from edk220:48
bkeysAnd if that works, I'd be really happy20:48
bkeysOh wait20:48
bkeysHmmmm20:48
bkeysI need to get the Reform kernel installed on the emmc20:49
minutepossible with chroot?20:49
bkeysIn theory20:49
bkeysI mean in theory edk2 is there, so as long as it's confirmed booting I can always boot back to this USB disk that my current install is on20:49
bkeysI'm just scared to reboot right now20:49
bkeysBut I mean, the things I'm discovering I think are valuable for others20:50
minutefor sure20:50
bkeysYeah I guess in theory even if I borked my emmc all up the worst thing I'd have to do is boot edk2 from SD card (easy), then boot back into this USB disk of mine and reflash edk2 to get it back20:52
bkeysminute: Wew, my emmc is not borked20:55
bkeysLets see if it sees my Fedora install20:56
bkeysIT DOES20:56
bkeysIt's almost over, I almost have Fedora on here working 20:56
minutebkeys: great20:59
minutecurrently building mesa from source with zink enabled21:00
+ reform8446 (~cqc@2601:249:9281:8440::92d7)21:04
bkeysIt looks like I'm gonna have to install the kernel before I can boot it, acpi doesn't work for some reason21:06
bkeysbrb gonna zip to the bike shop and get my USB wifi adapter21:08
- reform8446 (QUIT: Remote host closed the connection) (~cqc@2601:249:9281:8440::92d7)21:38
bkeysI guess I'm just gonna have to rebuild the kernel inside of the chroot21:52
bkeysUnless21:53
bkeysAlright the kernel is there, apparently the Linux makefile can create an rpm package so I want to test that out22:14
bkeysIt also created an srpm file so if this indeed works I'll upload the srpm to copr and I can have a repo others can just use and will simplify things for everyone while putting very little maintenance burden on me22:14
- gustav28 (QUIT: Quit: Quit) (~gustav@c-78-82-53-129.bbcust.telenor.se)22:15
bkeysThe kernel rpm installed fine, but for some reason now the UEFI isn't pulling up? Who knows, I g2g I'll make an SD card with edk2 on it and reflash the UEFI22:19
minuteas an experiment i just built a patched mutter/gnome-shell/gdm where i manually override the KMS overlay type for the last overlay to cursor... and this fixes all cursor stuttering issues, it's a night and day difference23:19
minutejosch: check out 0037-mfd-rk8xx-Fix-shutdown-handler.patch23:51
minutejosch: this patch wasn't there in my 6.14 stack23:51
joschooooh!23:52
joschi love the verbose explanation for a one-line change23:53
minuteso they were indeed messing with poweroff stuff23:54
joschwish i would learn to do write-ups like this in my git commits XD23:54
joschi could try and see whether power-off used to work in 6.14 -- do we know?23:54
minutejosch: it worked23:55

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