2025-10-23.log

- thelounge3120 (QUIT: Quit: The Lounge - https://thelounge.chat) (~thelounge@gidzit.org)00:04
- andreas-e (QUIT: Quit: Leaving) (~Andreas@2a02-8434-b6a3-e901-facc-8e87-8e54-890e.rev.sfr.net)00:38
+ wielaard (~mjw@gnu.wildebeest.org)00:47
- mjw (QUIT: Ping timeout: 240 seconds) (~mjw@gnu.wildebeest.org)00:49
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50)00:53
- chomwitt (QUIT: Ping timeout: 260 seconds) (~chomwitt@2a02:85f:9a5f:900:1ac0:4dff:fedb:a3f1)00:58
+ enwu (~enwu@user/enwu)01:11
- enwu6 (QUIT: Ping timeout: 246 seconds) (~enwu@user/enwu)01:13
- enwu (QUIT: Ping timeout: 248 seconds) (~enwu@user/enwu)01:23
+ enwu (~enwu@user/enwu)01:25
- enwu (QUIT: Ping timeout: 244 seconds) (~enwu@user/enwu)01:36
+ enwu1 (~enwu@user/enwu)01:36
- schalken (QUIT: Ping timeout: 244 seconds) (~schalken@117-118-178-69.gci.net)01:53
+ schalken (~schalken@117-118-178-69.gci.net)01:54
- n_to (QUIT: Quit: quitidiquit) (~n_to@2a03:4000:6:3662:24b1:57ff:fec6:76c1)02:00
+ n_to (~n_to@2a03:4000:6:3662:24b1:57ff:fec6:76c1)02:00
- svp (QUIT: Ping timeout: 248 seconds) (~svp@2002:4f07:f0bd:0:69b1:b463:d245:e861)02:06
- schalken (QUIT: Ping timeout: 264 seconds) (~schalken@117-118-178-69.gci.net)02:13
+ schalken (~schalken@117-118-178-69.gci.net)02:14
+ svp (~svp@2002:4f07:f0bd:0:69b1:b463:d245:e861)02:29
+ _justin_kelly71 (~justinkel@user/justin-kelly/x-6011154)02:40
- wielaard (QUIT: Ping timeout: 255 seconds) (~mjw@gnu.wildebeest.org)02:48
- schalken (QUIT: Ping timeout: 264 seconds) (~schalken@117-118-178-69.gci.net)03:06
+ schalken (~schalken@117-118-178-69.gci.net)03:10
- paperManu (QUIT: Ping timeout: 240 seconds) (~paperManu@198.58.139.163)03:30
- AbortRetryFail (QUIT: Ping timeout: 264 seconds) (~arf@146.ip-149-56-132.net)03:52
+ AbortRetryFail (~arf@146.ip-149-56-132.net)04:06
- paperManu_ (QUIT: Ping timeout: 240 seconds) (~paperManu@198.58.139.163)05:02
- GNUmoon2 (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon)06:02
+ GNUmoon2 (~GNUmoon@gateway/tor-sasl/gnumoon)06:03
- casparvitch (QUIT: Remote host closed the connection) (~casparvit@user/casparvitch)06:44
+ casparvitch (~casparvit@user/casparvitch)06:45
+ chomwitt (~chomwitt@2a02:85f:9a5f:900:1ac0:4dff:fedb:a3f1)07:25
- casparvitch (QUIT: Remote host closed the connection) (~casparvit@user/casparvitch)08:01
+ casparvitch (~casparvit@user/casparvitch)08:02
[tj]If anyone is looking for a 39c3 voucher, hit me up08:22
- casparvitch (QUIT: Remote host closed the connection) (~casparvit@user/casparvitch)08:26
+ casparvitch (~casparvit@user/casparvitch)08:27
dormitoThe handbook says that for the i.mx8m+ the uart labeled 'S1' @115200 baud is where u-boot and the kernel's consoles will be. I connected to that port (with 115200n8) on my 'pocket', and verified it worked in linux (/dev/ttymxc3), and then powered off, and back on. However I do not see any u-boot chatter at all. Is there something still missing?08:34
joschdormito: did you remember to swap RX and TX on your UART adapter?08:46
dormitojosch: yes. the uart adaptor works when linux is booted on the reform pocket (to be clear: I ran c-kermit on my other machine, and 'busybox microcom' on the pocket, using 11520n8, and I could type on one side, and see it on the other).09:20
dormito*115200n809:20
joschoh that is very weird, sorry i have to idea how your early boot messages would get lost but somehow serial just decides to work later o009:22
josch[tj]: i sent you a query, thank you! <309:22
dormitook, well thank you anyhow09:24
joschdormito: let me ping minute for you for more ideas :)09:33
+ thelounge2220 (~thelounge@148.168.138.88.rev.sfr.net)09:40
- ZetaR (QUIT: Ping timeout: 248 seconds) (~user@c-76-148-139-78.hsd1.fl.comcast.net)09:40
joschit seems i also need some assistance later09:50
joschi have a reform keyboard 20R01 here which seems dead. When I connect it to another PC via its usb-c port, nothing shows up. When I connect USB the top right JST connector, again nothing shows up on the other side. I tried pressing reset and flipped the prog switch to no avail. Ideas?09:51
+ ZetaR (~user@c-76-148-139-78.hsd1.fl.comcast.net)09:55
AbortRetryFailThose have an AVR, right?10:06
AbortRetryFailATMEGA32U4 from the schematic10:09
AbortRetryFailIf you have an AVRISP maybe it can tell if there are signs of life, but start with the basics: Does it have voltages? Is the clock running? etc.10:10
minutejosch: that one doesn't have a standalone switch iirc, so you need to provide 3v3 on the uart connector as well11:20
minutejosch: for example by using a reform11:21
- thelounge2220 (QUIT: Quit: The Lounge - https://thelounge.chat) (~thelounge@148.168.138.88.rev.sfr.net)11:30
- erle (QUIT: Quit: K-lined) (~erle@user/erle)11:31
- cow321 (QUIT: Read error: Connection reset by peer) (~deflated8@user/meow/deflated8837)11:59
+ cow321 (~deflated8@user/meow/deflated8837)11:59
- plomtest (QUIT: Remote host closed the connection) (~plom@user/plomtest)12:00
+ plomtest (~plom@user/plomtest)12:01
+ wielaard (~mjw@gnu.wildebeest.org)12:05
+ andreas-e (~Andreas@2a02-8434-b6a3-e901-facc-8e87-8e54-890e.rev.sfr.net)12:46
- plomtest (QUIT: Remote host closed the connection) (~plom@user/plomtest)12:46
+ plomtest (~plom@user/plomtest)12:47
+ paperManu (~paperManu@198.58.139.163)12:52
joschminute: oooooh i totally forgot about that, thank you!12:53
joschit seems i got too used to keyboard 3 and 4 already XD12:54
+ gustav25 (~gustav@c-78-82-53-228.bbcust.telenor.se)13:02
- plomtest (QUIT: Remote host closed the connection) (~plom@user/plomtest)13:02
+ plomtest (~plom@user/plomtest)13:03
+ ericsfraga (~user@2a00:23cc:b475:7c01:53b:6ddf:a36e:d38b)13:05
[tj]minute: josch I handle vouchers for Scotland (but the ccc are clear that the group is what you make of it). I get them for Congress and camp, for any future events I’m happy of you point mnt-reform related people to me for vouchers13:14
fricklerACTION now has a weird vision of the pink pocket chassis getting embroidered with a kilt pattern ;)13:21
- chomwitt (QUIT: Ping timeout: 244 seconds) (~chomwitt@2a02:85f:9a5f:900:1ac0:4dff:fedb:a3f1)13:22
- plomtest (QUIT: Remote host closed the connection) (~plom@user/plomtest)13:22
+ plomtest (~plom@user/plomtest)13:22
minutejosch: hmm, debian has a new multiarch bug maybe? it won't let me remove *:armhf >  E: Removing essential system-critical packages is not permitted.13:37
- Ar|stote|is (QUIT: Quit: No Ping reply in 180 seconds.) (~linx@149.210.6.130)13:37
minutebecause it thinks these are essential: > libgcc-s1:armhf libc6:armhf (due to libgcc-s1:armhf)13:37
minutetrying > sudo apt-get remove --allow-remove-essential libgcc-s1:armhf13:38
minutethat worked13:41
+ Ar|stote|is (~linx@149.210.6.130)13:42
- plomtest (QUIT: Remote host closed the connection) (~plom@user/plomtest)13:42
+ plomtest (~plom@user/plomtest)13:43
- paperManu (QUIT: Ping timeout: 265 seconds) (~paperManu@198.58.139.163)14:10
joschminute: yes, this is known. The problem is, that apt considers packages all packages in the essential set as essential (and thus warns on removal) irrespective of their architecture. There are arguments that apt should only consider packages of the native architecture as essential but there exist use-cases involving cross-grading from one architecture to the other where you actually need more than one 14:31
joscharchitecture in the essenital set. So this is something which pops up every once in a while but nobody is too motivated to work on changing the status quo. So your removal of libgcc-s1:armhf with --allow-remove-essential was exactly the right thing to do in your situation.14:32
minutejosch: ok, thanks for explaining! i just didn't encounter this in the past (longer time ago), probably because i didn't have that package installed14:33
minutejosch: btw maybe you also have a hint for how to remove a package with broken pre-removal script?14:34
minuteactually i can just google that, sorry14:35
- wielaard (QUIT: Ping timeout: 256 seconds) (~mjw@gnu.wildebeest.org)14:36
+ paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca)14:36
minuteok i deleted the prerm and postrm scripts14:38
minutejosch: btw we should change the default terminal to ptyxis, it appears to be the modern (and faster) replacement for gnome-terminal14:53
joschminute: broken removal scripts are *really* nasty. There is no silver bullet that i know of. We have a bunch of constraints in Debian of what packages never have to be removed because of removal scripts. And there are some nice multi-arch things we cannot have because they would create situations in which removal is no longer possible due to missing deps14:53
joschoh funky -- sure14:53
joschi never heard of it, i'm looking forward to trying it out14:54
minutefeels a lot smoother and looks a lot better than gnome-terminal14:54
joschwhat bugged me about gnome or gnome-terminal or the shell (no idea who the culprit is) is that when you want to launch another terminal it will not do that but instead highlight the terminal window you have already open -- is there an easy way to change this behaviour?14:55
minutejosch: well, gnome-terminal has to goooo14:56
minutejosch: how did you try to launch the terminal? click in the dock?14:56
joschsuper+enter14:56
joschor super and then i type "termi" and press enter when it highlights14:56
minuteohh ok i think maybe you were already using ptyxis because i experience this for the first time now :D14:57
joscho014:57
minutei guess the shortcut needs to be changed to --new-window14:57
joschno way :D14:57
minutei guess the shortcut needs to be changed to ptyxis --new-window14:57
joschi'll check later14:57
minutejosch: i only know ptyxis because yesterday we installed/set up gnome on holo_memory's pocket reform and there it was the default14:58
minutejosch: we did that by upgrading reform-desktop-full14:58
minuteok setting my super+return shortcut to "ptyxis --new-window" fixes that14:59
minutei mean, it's a matter of taste, one could also want to open new tabs14:59
minute(ptyxis --tab)14:59
joschfoot is also still among the dependencies of reform-desktop-full15:00
minutejosch: that can go then15:00
minutejosch: so far we had sensible-terminal i think in sway config and it starts gnome-terminal which is a bad experience on sway (and in general) because it's soooo slow for a terminal and also has a murky default theme15:01
- jacqueline (QUIT: Read error: Connection reset by peer) (~jacquelin@user/jacqueline)15:02
+ jacqueline (~jacquelin@user/jacqueline)15:02
joschgnome-terminal is not a direct dependency of reform-desktop-full but a dependency of gnome-core15:03
joschso after installing ptyxis, maybe reform-system-image has to manually change the default x-terminal-emulator -- we'll see15:03
josch(meaning, gnome-terminal will still be there even after adding ptyxis to the deps)15:03
joschthis might be doing what you had in mind: https://source.mnt.re/reform/reform-debian-packages/-/merge_requests/14115:07
minutejosch: nice, many thanks15:08
joschno worries, these are micro changes :)15:09
joschminute: of course also always feel free to edit that file yourself -- remember that i changed the reform-debian-packages build of reform-tools to work this way instead of patches because you wanted to have an easy way to do things like this yourself15:09
minutejosch: ok! just wanted to be verbose about it so you're in the loop 15:10
joschnot saying that you cannot/should not ping me about it, just saying that you should not feel like you had to or something :)15:10
joschthank you, sounds good! :)15:10
minuteok :)15:10
+ chomwitt (~chomwitt@2a02:85f:9a5f:900:1ac0:4dff:fedb:a3f1)15:11
- pomel0 (QUIT: Ping timeout: 252 seconds) (~pomel0@user/pomel0)15:21
+ pomel0 (~pomel0@user/pomel0)15:21
* Guest7939 -> mjw15:40
- pomel0 (QUIT: Remote host closed the connection) (~pomel0@user/pomel0)16:02
- digitalrane (QUIT: Ping timeout: 250 seconds) (~digitalra@user/digitalrane)16:09
- lidstah2 (QUIT: Remote host closed the connection) (~lidstah@gateway/tor-sasl/lidstah)16:20
+ lidstah (~lidstah@gateway/tor-sasl/lidstah)16:21
+ digitalrane (~digitalra@user/digitalrane)16:24
- paperManu (QUIT: Ping timeout: 252 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca)16:35
+ paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca)16:37
- paperManu (QUIT: Read error: Connection reset by peer) (~paperManu@modemcable141.205-200-24.mc.videotron.ca)16:54
+ paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca)16:58
elbminute: josch: I've been using foot under sway and it seems reasonably well-behaved17:48
- paperManu (QUIT: Ping timeout: 256 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca)17:54
+ paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca)17:56
minuteelb: foot under sway is good!17:58
minutegrimmware: elb: josch: i made a quick data logger tool to get more info about what happens before a crash reboot https://community.mnt.re/t/restarts-without-warning/3927/3517:59
minutethe idea is that you run this in the background and it will collect all kinds of hardware/sensor/cpu/gpu/memory/freq/irq data and hopefully there will be some pattern in the logs 18:00
minuteit collects the data every second18:00
minuteand syncs18:00
minuteso there's a chance that some data will be committed to disk a few seconds before the big bang18:00
joschminute: thank you! next time i encode a video i'll run this and remain connected via ssh to also get the last few log messages before the reboot18:04
minutejosch: awesome18:05
minuteimprovements to this script are welcome of course, but i had to start somewhere...18:05
grimmwarehttps://www.theregister.com/2025/10/22/linux_hibernation_patch/ interesting18:07
minutegrimmware: woah woah18:07
minutegrimmware: extra interesting that this is from collabora18:08
minutegrimmware: regarding the panel fixing stuff, i want to make a kernel-dev script that will help making a stripped down monokernel and document how to boot that over the network for faster iteration18:10
grimmwareminute: that would be amazing, although I would say that the biggest thing that slowed me down was having to edit a diff. A more generic workflow that doesn't require you to use emacs would be good.18:15
minutegrimmware: yeah totally. that would be included :D18:20
grimmwarefantastic :) more than happy to put in a few hours of development, review or testing on that if you get to a place where you can hand some of it off18:23
grimmwareminute: incidentally with the debugging for non-battery related reboot, I've got a fair bit of experience with ebpf so if you have some theories you want to prove or disprove we might be able to pivot those into some high performance non-polling tracing, just lmk18:34
minutegrimmware: ah very nice. i don't have any solid theories yet though18:34
minutegrimmware: so i wanna first check if there are any common "environmental" things among those crashes or no18:35
grimmwareminute: getting people to chuck you a `sudo journalctl -k -b -1` would probably be a good shout (dmesg from last boot)18:39
minutegrimmware: how often is that saved? or does it only contain dmesg up to login?18:44
grimmwareit's the entire dmesg for the last boot, so it's guarantees are essentially whatever journald's are for writing18:45
grimmwareI'd expect to see this just truncate for a power related reboot and have something terribly interesting for a kernel related one18:47
minuteok, good call18:50
grimmwareyeah, my last power related reboot happens to be my `-b -2` and it truncates after some wifi output whereas the last one has systemd handling the shutdown18:52
minutei'm writing the kernel dev script, and wanted to apply all patches with git-am, but there are some patches that it can't apply, "Patch format detection failed."18:53
minuteah, missing a subject18:53
grimmwareI'm excited to have a play with this18:54
+ wielaard (~mjw@gnu.wildebeest.org)18:58
minuteso far only 1 badly formatted patch...19:01
minuteinteresting, > error: Documentation/devicetree/bindings/display/rockchip/rockchip-vop2.yaml: patch does not apply19:04
minuteduring > Applying: ../reform-debian-packages/linux/patches6.16/rk3588-mnt-reform2/0058-dt-bindings-display-vop2-Add-VP-clock-resets.patch19:04
grimmwareI think holo_memory's current reboot issue is different from his old one, which is to say I think he has the one you're currently searching for19:05
grimmwareYou probably know that already but I think it bears saying out loud19:06
grimmwareNeonkAaa's looks unrelated19:08
grimmwaretorb's sounds like holo_memory's on the very small amount of information19:08
grimmwareR1ck's is the same19:09
minutegrimmware: i suspect so too @ holo_memory. but it's super mysterious what the reset is (it's also quite rare)19:10
grimmwareyeah, it could be so many different things. I like that you're going straight to getting them to just run diagnostics rather than getting people to debug19:11
rick_if i can test sth.. just ping me :D  an yeah.. the reboot problem is very.. strange and very hard to tackle.. cause nothing gets logged ^^v19:11
rick_at least i found nothing.. and i searched for some time19:12
grimmwarerick_: make sure to grab a `sudo journalctl -k -b -1` next time :)19:13
rick_ohh new stuff in the forum.. didn't saw taht19:13
rick_i did..  well without the k19:13
grimmwareoh, you still got it? can I have a look?19:14
grimmwareor just the end of it19:14
grimmwareminute: a `uname -a` and display version would probably be relevant data here too19:17
minutegrimmware: right, both should be added19:17
minute> error: arch/arm64/boot/dts/rockchip/rk3576-rock-4d.dts: patch does not apply19:18
minuteah, we have unrelated patches in the patchstack again19:18
rick_grimmware i think i can find a boot in my journel which had a brownout.. but i didn t change any settings which are mentioned in the forums.. ill try that now19:19
rick_was 'offline' yesterday19:19
+ synnfynn (~synnfynn@user/synnfynn)19:21
minutehm, hdmi driver recently gained CEC huh19:30
rick_kay, did lock the cpu/gpu at max and startet crashland in the background. lets see if we get sth ^^19:49
minuterick_: noice19:49
- mjw (QUIT: Killed (platinum.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c)19:54
* wielaard -> mjw19:54
+ Guest6281 (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c)19:54
minutegrimmware: first draft https://source.mnt.re/reform/reform-kernel-dev19:56
minutegrimmware: prepare.sh makes a local kernel git with the version that's currently running (this should get some option to make that customizable if you're cross building from another major version etc) and gets all reform specific patches and applies them as regular commits, so you have a git history of all the patches in the right order. then it also configures the kernel with my monolithic19:58
minutekernel config19:58
minuteok. now i need some uboot script to be able to boot this kernel over the network...19:59
minutehm, looks like we don't have tftpboot enabled in rk3588 uboot :#20:06
grimmwareoh lol sidequest20:09
+ LainIwakura (~LainIwaku@user/LainIwakura)20:19
+ vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50)20:19
- LainIwakura (QUIT: Ping timeout: 250 seconds) (~LainIwaku@user/LainIwakura)20:24
chwhi20:28
chhi20:28
minutegrimmware: yeah... https://source.mnt.re/reform/reform-rk3588-uboot/-/merge_requests/new?merge_request%5Bsource_branch%5D=networking20:30
minutech: hi!20:30
chhttps://source.mnt.re/reform/reform-rk3588-uboot/-/merge_requests/11 better link for us, i think20:30
minutegrimmware: sorry wrong link :D20:30
minutehaha20:30
minutecopied too early lol20:30
ch:)20:31
minutei wonder what the fanciest initramfses are out there20:32
chmh?20:34
chdropbear in initramfs is about the fanciest thing i know20:34
chdebian/grml's initramfs has a few nice things like fetch= etc to load the rootfs from elsewhere20:35
minutemhm20:36
grimmwarethis'll be so fuckin cool to boot ofter tftp20:40
grimmwareminute: the script apparently assumes a global git configuration http://paste.debian.net/1402335/20:58
chwell thats not unreasonable21:03
minutegrimmware: ah yeah that's just for git am21:03
minuteoh we got new ddr4 firmware (v1.19)21:06
grimmwarewhat's better about it?21:15
minuteno idea21:16
minutestrangely, my freshly baked kernel freezes during boot21:16
minute> Using asix_eth device21:25
minute> TFTP from server 192.168.1.1; our IP address is 192.168.1.19021:25
minute=> dns mntre.com21:29
minute91.250.115.1521:29
minuteheh, dns in uboot21:29
joschminute: i welcome your reform-kernel-dev project and i see reform-debian-packages!142 but that makes me fear that the changes you are adding will get outdated by linux6.17 changes21:30
minutejosch: doesn't matter i think... the question is how long should we follow the collabora stuff for rk3588 and when do we start preferring mainline and cherrypicking things...21:31
joschfor example, the "remove unneeded patches" commit removes 45 files and that work would need to be re-done for 6.1721:31
rick_oh nice, with 'export PAN_MESA_DEBUG=gl3' my slicer software finally runs without problems <3 saw a discusssion about that here a few days ago ^^21:32
joschaha yes, good21:32
minutejosch: it makes me uneasy that the rk3588 patchstack is growing and not shrinking21:32
joschminute: i'd like to drop the collabora patch stack rather sooner than later :)21:32
minutejosch: i think that's mainly because of rk3576 developments21:32
minuterick_: ah nice21:32
joschit is an oddity in the regular patch rebasing and causing extra work for little gain as many patches do not affect rk3588 anymore21:32
minutejosch: yes exactly. i'm just wondering why certain patches are not getting merged and are getting old now...21:33
+ op_4_ (~tslil@2a01:4f8:c0c:7952::1)21:33
joschi need to have a very close look at your MR142 and update my linux6.17 branch accordingly21:34
+ cararemixed_ (sid724089@id-724089.helmsley.irccloud.com)21:34
- synnfynn (QUIT: Quit: WeeChat 4.5.2) (~synnfynn@user/synnfynn)21:34
joschi am happy that you are fixing up the patch headers21:34
joschthat's something which in the past contributions by bremner and vagrantc did21:35
+ jahkosha_ (~jahkosha@user/jahkosha)21:35
+ jn_ (~quassel@2a0a-a54a-68ad-0-20d-b9ff-fe49-15fc.ipv6dyn.netcologne.de)21:35
- jn_ (QUIT: Changing host) (~quassel@2a0a-a54a-68ad-0-20d-b9ff-fe49-15fc.ipv6dyn.netcologne.de)21:35
+ jn_ (~quassel@user/jn/x-3390946)21:35
joschthere is also a more lengthy discussion of whether it makes sense to change the whole patching approach in https://source.mnt.re/reform/reform-debian-packages/-/merge_requests/139 in case you find yourself with too much free time one day :)21:35
- jahkosha (QUIT: Killed (zinc.libera.chat (Nickname regained by services))) (~jahkosha@user/jahkosha)21:36
* jahkosha_ -> jahkosha21:36
+ mlarkin_ (~mlarkin@syn-076-081-194-027.biz.spectrum.com)21:36
+ buckket1 (~buckket@vps.buckket.org)21:36
+ schalken1 (~schalken@117-118-178-69.gci.net)21:37
minutelol tftp boot over this 100mbit asix usb ethernet dongle is very slow > 678.7 KiB/s21:39
+ LainIwakura (~LainIwaku@user/LainIwakura)21:40
- Guest6281 (QUIT: *.net *.split) (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c)21:41
- schalken (QUIT: *.net *.split) (~schalken@117-118-178-69.gci.net)21:41
- colinsane (QUIT: *.net *.split) (~colinunin@97-113-71-58.tukw.qwest.net)21:41
- buckket (QUIT: *.net *.split) (~buckket@vps.buckket.org)21:41
- jn (QUIT: *.net *.split) (~quassel@user/jn/x-3390946)21:41
- op_4 (QUIT: *.net *.split) (~tslil@user/op-4/x-9116473)21:41
- cararemixed (QUIT: *.net *.split) (uid724089@id-724089.helmsley.irccloud.com)21:41
- mlarkin (QUIT: *.net *.split) (~mlarkin@syn-076-081-194-027.biz.spectrum.com)21:41
* op_4_ -> op_421:41
* cararemixed_ -> cararemixed21:41
- LainIwakura (QUIT: Client Quit) (~LainIwaku@user/LainIwakura)21:43
+ LainIwakura (~LainIwaku@user/LainIwakura)21:44
- schalken1 (QUIT: Ping timeout: 256 seconds) (~schalken@117-118-178-69.gci.net)21:46
+ schalken (~schalken@117-118-178-69.gci.net)21:46
+ Guest6281 (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c)21:47
+ colinsane (~colinunin@97-113-71-58.tukw.qwest.net)21:48
- LainIwakura (QUIT: Ping timeout: 250 seconds) (~LainIwaku@user/LainIwakura)21:52
+ LainIwakura (~LainIwaku@user/LainIwakura)21:54
minuteok nice, booted the first kernel over network21:55
minuteah, much faster with 1gbit asix21:58
- schalken (QUIT: Ping timeout: 244 seconds) (~schalken@117-118-178-69.gci.net)21:59
+ schalken (~schalken@117-118-178-69.gci.net)22:02
minutemhm mhm https://www.landley.net/toybox/status.html22:09
- gustav25 (QUIT: Quit: Quit) (~gustav@c-78-82-53-228.bbcust.telenor.se)22:15
- schalken (QUIT: Ping timeout: 260 seconds) (~schalken@117-118-178-69.gci.net)22:24
- elb (QUIT: Read error: Connection reset by peer) (~elb@68.133.31.194)22:24
+ elb (~elb@68.133.31.194)22:29
minuteok, i have my kernel booting over tftp and starting toybox as shell22:30
minutethis takes only a few seconds22:30
minuteoh, some kernel config is missing > [   12.256622] pcieport 0004:40:00.0: deferred probe timeout, ignoring dependency22:30
elbminute: the keyboard debounce logic looks like it appeared out of whole cloth when the initial keyboard firmware was committed; it's kind of complicated, but it appears that what it actually does is just checks to see that a key is pressed for two polling intervals in a row22:31
elbdoes that sound accurate?22:32
minuteelb: yes, that's accurate i think. in the beginning (very long time ago) i experimented with longer intervals etc22:32
elbI'm trying to decide if reducing the logic to exactly that check is valuable or not22:33
minuteelb: if it's still possible to change the number of polling intervals later, than yes22:34
minuteconsecutive ones, that is22:34
minutes/than/then22:34
elbok, then I won't change anything22:34
elbbeacuse the change i had in mind eliminated the debounce array entirely22:34
elb(basically debounce logic became state == 1 && pressed)22:35
minuteelb: ah no, that's too much22:35
elbit's pretty easy to use consecutive debounce intervals that are all 1 with just the state array22:35
elbbut if you want to be able to investigate history, might as well leave what's there now22:36
elb(basically that would be if state == k && pressed, it's pressed, otherwise if pressed == 1 then state++, otehrwise state = 0)22:36
elbbut not a big deal to just leave it as is22:37
+ nsc (~nicolas@i5C74DC34.versanet.de)22:38
minuteyeah, i still have to do some keyboard debugging on next, i would prefer to have that feature for a while still 22:38
elbare the next and pocket keyboard firmwares unified?22:39
- nsc (PART: !!unknown attribute: msg!!) (~nicolas@i5C74DC34.versanet.de)22:39
elb(the next is 2350 IIRC?)22:39
elbI'm not actually sure what I'm hunting at this point; all of this simplification is looking for reasons the keyboard might be drawing more power than it needs to, but I've dug at enough stuff that I think that's going to require measurement I'm not prepared to do at the moment22:45
- LainIwakura (QUIT: Quit: Client closed) (~LainIwaku@user/LainIwakura)22:45
elbalthough I'm also tracking a bug, my keyboard doesn't reliably light up the OLED and give the power on message when keys are pressed (it used to), and I think that's from making pressed_keys static (I need to verify this)22:50
elbI'm pretty sure there's a race between hid_task and main in accessing pressed_keys22:50
elb but I need to check the processor documentation to see if it's a race we should care about22:50
elb(the processor documentation will clarify both whether it's an actual race and whether if it is a race, we care about it; if it's not an actual race, we just make pressed_keys volatile and problem solved)22:53
minuteelb: pocket+next are not unified yet. classic kbd fw4, pocket and next all share code and could be unified potentially 22:55
elbnice, I like the sound of that22:55
minuteuhm classic kbd fw4+next are actually unified, sorry. but next is a branch there, not fully in main22:55
elbI'm willing to help with that if I can, although not having all three devices it may be that I can't22:56
minuteelb: that sounds verry interesting @ race22:56
minuteelb: we can provide you with extra keyboards etc if that helps22:56
minuteelb: in the next branch of reform2-keyboard4-fw i've implementing a diffing thing where it only sends usb reports when something has changed22:57
elbif it gets to the point that that's directly valuable to you we can talk, but my availibility for projects tends to be bursty, and I don't want to consume resources unless there's a goal I can commit to22:57
minuteelb: that greatly reduces usb traffic, but some keydown/up events are lost, and i haven't figured out why yet, and if you found a race, it might be an explanation.22:58
elbyeah I was just looking at that, actually, and wondering if that would save power22:58
minuteelb: i think it would, as usb is causing a lot of interrupts22:58
elbyeah but they're not _that_ fast22:58
minutei recently checked powertop and the input device is always high up there22:58
elb5ms22:58
minuteright...22:59
elbI need to learn more about tusb, too22:59
elbI wonder if it's trying to do USB things every 5 ms when powered off22:59
minuteit normally shouldn't, but who knows22:59
elbthat seems more like a power drain I'm concerned about, but I'm hoping tusb is shutting that down22:59
minutethere is an api for detecting if there's a host at least22:59
elbcertainly hid_task is trying22:59
elbyeah one of the things I want to look at is stopping hid_task when the device isn't enumerated and active23:00
minuteinteresting idea23:00
elbI think you could do a substantially slower poll of just the keys, instead23:00
minutecertainly has my blessing23:00
elbsubstantially like factor of 2-5, not 10023:00
minuteright23:01
elbbut again, a 5 ms wake cycle on a 2040 at 48 MHz shouldn't be enough drain to matter on those giant cells if it's not actually _doing_ anything23:01
elbthus needing to dig into tusb23:02
elbbut I dunno, mA add up23:02
minute> [  160.415616] PM: suspend exit23:05
joschwat23:05
minuteinteresting, suspend to ram test works on my monokernel23:05
elb!!23:05
joschrk3588?23:05
minutesure thing!23:05
minutei did this whole setup to be able to debug suspend and pocket panel foo23:05
- paperManu (QUIT: Ping timeout: 256 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca)23:05
minuteinterestingly there is an mem abort in _phy_state_machine+0x200/0x2f8 but it can continue past that23:06
minuteit's directly after gmac-dwmac so possibly related:23:07
minute> [  158.860198] rk_gmac-dwmac fe1b0000.ethernet: init for RGMII23:07
minute[  158.860337] Unable to handle kernel NULL pointer dereference at virtual address 00000000000001b823:07
joschwith or without pci/wifi/nvme?23:07
minutejosch: no pcie devices installed, pci root port is there in lspci though (in toybox)23:08
minuteaha, like i suspected, with ethernet cable plugged in, the mem abort isn't there23:09
minuteah wait. this is only the "devices" setting of pm_test23:09
minutei'll disable gwmac in the kernel and reboot + try again23:10
+ paperManu (~paperManu@198.58.139.163)23:17
- bkeys (QUIT: Remote host closed the connection) (~Thunderbi@98.19.131.193)23:19
vagrantcyay. updating the lpc firmware seems to have fixed my battery reporting issues from the kernel module... and serial console seems less noisy too ... a little bit of ghosting on the OLED display, though ... ghosted faded battery status information lingers when doing other things23:20
+ LainIwakura (~LainIwaku@user/LainIwakura)23:20
vagrantc(or at least, the first two boots seem to be reporting battery status correctly... have had the dumb luck "it worked!" before)23:20
elbwait I'm an idiot, we're only using one of the two cores, this absolutely _is not_ a bus-level race, and it's the kind of thing that can absolutely be fixed by volatile23:21
vagrantcentirely possible that the ground connection for the serial adapter was just loose...23:21
minuteelb: nice23:26
minutei wanted a text editor for my toybox linux and was happy that go stuff (like micro) comes as static binary, and it starts but complains that > open /dev/tty: no such device or address23:33
minutei should probably use "oneit"23:35
vagrantcok, pretty confident the serial problems are just a loose connector ... flipping it upside down torqued the serial cable and i can now see through the transparent bottom plate (yay!) that the ground pin has wiggled pretty loose23:35
joschbusybox would have vi as an editor23:37
minuteahaha i got it to work23:43
minutei renamed /dev/tty and ln -s /dev/console /dev/tty23:43
minute(and export TERM=xterm)23:43
minutejosch: i wasn't able to compile busybox, tried that first... it doesn't build on my debian for some reason23:43
minutejosch: doesn't find ncurses23:44
minutejosch: well, i didn't really build busybox itself, but tried make menuconfig23:44
- LainIwakura (QUIT: Ping timeout: 250 seconds) (~LainIwaku@user/LainIwakura)23:46
+ schalken (~schalken@117-118-178-69.gci.net)23:48
minuteok, so with gmac0 disabled, "devices" pm_test is clean23:49
joschminute: you can use the same tools used in reform-system-image to make a multi-gb system image to make tiny initramfs as well. For example, a "real" Debian (no busybox) fits into 24 MB gzipped. You'd trade the time you spend on fiddling with the limitations of busybox/toybox for a few more MB to transfer.23:50
minutejosch: well, i would have to learn how to build that 24mb debian system too23:50
minutejosch: also, the system is already fine now... i only wanted a text editor so i don't have to rebuild the system to make a change to some scripts23:51
joschokay!23:51
minuteok, devices test works23:51
minutedarn23:51
minutei meant, platform test works23:51
minuteah, i didn't realize that hdmi works all the time as well (i just saw my camlink usb stick blinking off and on during suspend test)23:51
minutepower goes down to 0.38A @ 12V during "platform" test23:52
minuteprocessors test works as well!23:53
- andreas-e (QUIT: Quit: Leaving) (~Andreas@2a02-8434-b6a3-e901-facc-8e87-8e54-890e.rev.sfr.net)23:54
minuteok so all those tests work, suspends and exits cleanly23:54
minutenow the question is, how to make it wake from an external input23:54
grimmware...23:55
grimmwarewow23:56

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