2025-10-26.log

- pomel0 (QUIT: Ping timeout: 256 seconds) (~pomel0@user/pomel0)00:06
+ pomel0 (~pomel0@user/pomel0)00:06
minutegordon2: maybe first try if it works on your machine too!00:11
+ aelius (~aelius@user/aelius)00:12
- wickedshell (QUIT: Ping timeout: 264 seconds) (~wickedshe@c-174-56-77-83.hsd1.nm.comcast.net)00:16
minuteimages built:00:19
minutebarebox-mnt-reform2-rk3588.img00:19
minute> Board: MNT Reform 2 with RCORE-DSI RK3588 Module00:20
minutelol > usb1-0: Bus 001 Device 002: ID 1235:8016 Scarlett 2i2 USB00:21
ch:>00:21
minuteit found something on one of the usb controllers...00:21
chis it correct? (:00:22
minuteyeah. it's just not the port of the usb hub :D00:22
minute(but maybe it's just missing usb hub reset)00:22
- jackhill (QUIT: Remote host closed the connection) (~jackhill@kalessin.dragonsnail.net)00:22
minutenow it found >   +-4 ID 0451:814000:27
minutethat's the USB hub00:27
minutebut it can't see the keyboard/trackball... maybe no OHCI support?00:28
chbtw, is this a new warning from the kernel during boot?00:30
chOkt 25 23:30:13 zeppelin kernel: Warning! ehci_hcd should always be loaded before uhci_hcd and ohci_hcd, not after00:31
gordon2is it kexec bootloader or just a bootloader that mimics linux kernel?00:31
gordon2looks like the latter, good, because last time i checked kexec was broken on rk358800:31
minuteah, the version of rk3588-base.dtsi is missing hdmi100:36
minutech: huh00:36
minutegordon2: it's an evolution of uboot that's more closely modeled on linux if i understand correctly00:36
gordon2cool00:37
minute(i was mistaken, hdmi1 is in rk3588-extra.dtsi)00:38
- natalie (QUIT: Read error: Connection reset by peer) (~natalie@user/natalie)00:50
- chomwitt (QUIT: Ping timeout: 244 seconds) (~chomwitt@2a02:85f:9a5f:900:1ac0:4dff:fedb:a3f1)00:50
minuteinteresting stuff about CPLL clock https://git.pengutronix.de/cgit/barebox/log/?h=next&showmsg=1&qt=grep&q=rk358800:50
+ natalie (~natalie@user/natalie)00:50
- Ar|stote|is (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~linx@149.210.6.130)01:00
+ Ar|stote|is (~linx@149.210.6.130)01:00
chhttps://lore.barebox.org/barebox/aN01SY3mFJVfdf2X@pengutronix.de/#t01:01
chwondering how linux does this bit01:07
- pomel0 (QUIT: Ping timeout: 256 seconds) (~pomel0@user/pomel0)01:16
+ pomel0 (~pomel0@user/pomel0)01:16
minutech: ugh interesting01:23
minutei've progressed a bit01:25
minutedwhdmiqp-rockchip fdea0000.hdmi@fdea0000.of: registered I2C bus driver01:25
minuteERROR: rockchip-vop2 fdd90000.vop@fdd90000.of: failed to get modes: Invalid argument01:25
minute:01:25
minute:D01:25
minutefbconsole: registered as fbconsole0-101:25
minuterockchip-vop2 fdd90000.vop@fdd90000.of: Registered fb0 on VP3, window Esmart1-win0, type primary01:25
minutehaha02:06
minute> rockchip-vop2 fdd90000.vop@fdd90000.of: dclk: 150015000000, pixclk_div: 1, dclk_div: 202:07
minutethat dclk is a bit too big02:07
+ wickedshell (~wickedshe@c-98-60-138-166.hsd1.nm.comcast.net)02:22
- RandyK (QUIT: Ping timeout: 272 seconds) (~RandyK@user/randyk)02:24
minutemeeeh > PANIC: unable to handle paging request at address 0x00000000fdea08e002:27
+ RandyK (~RandyK@user/randyk)02:31
minutehm, no more crash02:45
- paperManu_ (QUIT: Ping timeout: 256 seconds) (~paperManu@198.58.209.52)02:57
minuteok, enough for today. couldn't get a hdmi picture even if the code traces mostly correctly02:34
minuteanyway, barebox is pretty hackable02:34
- jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu)02:39
- pomel0 (QUIT: Ping timeout: 244 seconds) (~pomel0@user/pomel0)02:58
- op_4 (QUIT: Remote host closed the connection) (~tslil@2a01:4f8:c0c:7952::1)04:05
+ op_4 (~tslil@user/op-4/x-9116473)04:05
+ jackhill (~jackhill@kalessin.dragonsnail.net)04:09
- reform12606 (QUIT: Ping timeout: 244 seconds) (~heiko@2a11:fb80:3cc:f401:e981:d025:1a25:befe)04:54
- colinsane (QUIT: Ping timeout: 244 seconds) (~colinunin@97-113-71-58.tukw.qwest.net)05:35
- wickedshell (QUIT: Ping timeout: 240 seconds) (~wickedshe@c-98-60-138-166.hsd1.nm.comcast.net)06:28
+ reform12606 (~heiko@2a11:fb80:3fb:9001:e981:d025:1a25:befe)07:04
+ wickedshell (~wickedshe@c-98-60-138-166.hsd1.nm.comcast.net)07:21
reform12606  07:39
- shtrophic (QUIT: Ping timeout: 244 seconds) (~m-hrdsqi@user/shtrophic)07:44
+ chomwitt (~chomwitt@2a02:85f:9a5f:900:1ac0:4dff:fedb:a3f1)08:05
+ shtrophic (~m-hrdsqi@user/shtrophic)08:54
+ gustav25 (~gustav@c-78-82-53-228.bbcust.telenor.se)09:02
+ jogu (~jogu@user/jogu)10:32
- timide (QUIT: Remote host closed the connection) (~timide@user/timide)10:33
+ mjw (~mjw@89.205.206.232)10:44
- mjw (QUIT: Ping timeout: 244 seconds) (~mjw@89.205.206.232)10:57
+ digitalrane (~digitalra@user/digitalrane)11:12
+ andreas-e (~Andreas@2a02-8434-b6a3-e901-facc-8e87-8e54-890e.rev.sfr.net)11:30
+ mjw (~mjw@89.205.206.232)11:46
- iank (QUIT: Ping timeout: 256 seconds) (~iank@fsf/staff/iank)11:53
- mjw (QUIT: Ping timeout: 252 seconds) (~mjw@89.205.206.232)12:01
+ mjw (~mjw@89.205.206.232)12:29
+ iank (~iank@fsf/staff/iank)12:38
chnice12:39
- iank (QUIT: Quit: ZNC 1.8.2+deb2+deb11u1 - https://znc.in) (~iank@fsf/staff/iank)12:43
- mjw (QUIT: Ping timeout: 256 seconds) (~mjw@89.205.206.232)12:46
+ iank (~iank@fsf/staff/iank)12:47
+ paperManu (~paperManu@198.58.209.52)13:01
doki am really interested in getting barebox running on the mnt reform13:18
dokbut i am waiting on the next ;)13:18
dokif you have questions on barebox i might be able to answer them, but i am not as knowledgeable as a3f13:19
+ mjw (~mjw@gnu.wildebeest.org)13:27
minutedok: oh nice, good to know! here is my WIP hack branch https://source.mnt.re/reform/mnt-reform-barebox13:34
minutedok: i got VOP2 + hdptx to seemingly configure, without crashing, but i had to hardcode mode info and pixel clock in several places, so something is yet off, and i didn't get a signal on monitor 13:35
minutedok: but it was awesome how much progress i was able to do in one evening on an unfamiliar codebase (barebox), which speaks to its code quality 13:36
dokcool13:36
dokbut i don't know much about displays in barebox, nor about rockchip ^^'13:36
minutedok: haha no problem... do you know if barebox supports extlinux btw?13:36
doki dont know, i never heard of extlinux13:37
minuteand i wonder how hard it is to port things from uboot and/or linux (missing device drivers)13:37
minutedok: haha13:37
minutedok: ok then maybe you tell me what you like the most about barebox / what is your focus there :D13:37
dokyou can also ask in #barebox, a3f is very active13:37
minuteohhh nice13:37
doki worked with barebox at my previous job, porting drivers and such13:38
dokand i've ported barbox to Allwinner A6413:39
dokalthough very minimal support13:39
+ bkeys (~Thunderbi@h193.131.19.98.dynamic.ip.windstream.net)13:41
+ erle (~erle@user/erle)14:31
+ timide (~timide@user/timide)14:44
- xktr (QUIT: Ping timeout: 246 seconds) (~xktr@user/xktr)15:03
+ xktr (~xktr@user/xktr)15:05
joschgrimmware: for CONFIG_DEBUG_INFO_BTF=y, try removing pkg.linux.nokerneldbg pkg.linux.nokerneldbginfo from linux/build.sh15:09
elbman the documentation for tinyusb is _terrible_15:29
elb"uh yeah this library exists, here's some examples, laters"15:31
joschi'm maintaining tinyusb in debian -- if you happen to find some better documentation i'd be interested in including it in the package15:33
elbwill do15:39
elbalso, don't get me wrong, I'm thankful that it's written adn working, and I always appreciate open source and its contributors (been in this game a long time)15:39
elbbut I'm having to dig through an _awful_ lot of heavily #ifdef'd source to find answers to questions that I think should be simple15:40
elbstuff like tud_rhport_init (which is relatively straightforward once you puzzle through it) does not make me happy15:42
elbbut specifically I think there's a not-exactly-problem in pocket-hid dealing with the chaining of reports; we trigger a keyboard report on every hid_task interval, and each keyboard report completion triggers a report of the next type15:46
elbbut we don't find out about the report's completion until the next hid_task interval, because that's the only time we call tud_task()15:46
elbthis is currently kind of fine, because we have only two reports, keyboard and trackball, but I have some concern that attempting to modulate wakeups when the host is not listening, or when there is no data to send, is going to get complicated due to this architecture15:48
chyeah its all looking at the source with tinyusb15:58
chand reading the usb spec next to it15:58
joschch: I'm adding a --zero option to reform-flash-uboot to make it more straight forward what you wanted to do yesterday withdd16:05
- shtrophic (QUIT: Ping timeout: 240 seconds) (~m-hrdsqi@user/shtrophic)16:08
elbch: that's the part that's burning me, I'm really not very familiar with USB at the protocol level16:09
chits super painful, yes16:11
erlejosch minute note that since luanti 5.11 drastically worsed support for a bunch of hardware (in particular it won't work very well with etnaviv), it might be a good idea to version-pin it to 5.10 if you still wish to include it in the image. the difference is about a 1000 lines of code (they deleted one renderer).16:25
erlei am writing this because i saw https://salsa.debian.org/twrightsman/luanti/-/pipelines/96337916:25
+ shtrophic (~m-hrdsqi@user/shtrophic)16:45
elbI thought I saw, somewhere in the documentation, instructiosn for using the PROG and BOOT buttons on the keyboard PCB to put it into the bootloader for flashing if it's been bricked, does anyone know where that was?16:51
elbmaybe it was on the forum, I didn't look there16:51
elb(I'm going to link to tools for deeper troubleshooting from my HACKING document)16:58
joscherle: i'm now in front of a keyboard 2.0 (the one with two alt keys) and flashed the neo2 firmware and am trying to reproduce your bug17:00
erlejosch thanks! you have to press the key labeled hyper for fn and hyper + circle for fn lock.17:03
erlejosch so hyper + m3 should be enter/return17:03
erle*right* m3, that is17:03
erleshould be easy to reproduce, since there are only 3 keys affected (unless you found some more)17:05
joschah sorry, it was my problem17:05
joscherle: your patch works -- if you like, file a MR :)17:07
erlejosch nice. i'll assign you as a reviewer then ig?17:10
grimmwarejosch: nice, will do, thx17:12
- shtrophic (QUIT: Ping timeout: 240 seconds) (~m-hrdsqi@user/shtrophic)17:13
+ shtrophic (~m-hrdsqi@user/shtrophic)17:15
grimmwarejust set a build going, look forward to testing that out17:15
joscherle: yes, thank you!17:15
- shtrophic (QUIT: Ping timeout: 240 seconds) (~m-hrdsqi@user/shtrophic)17:28
+ shtrophic (~m-hrdsqi@user/shtrophic)17:35
- shtrophic (QUIT: Ping timeout: 246 seconds) (~m-hrdsqi@user/shtrophic)17:46
+ shtrophic (~m-hrdsqi@user/shtrophic)17:50
- shtrophic (QUIT: Ping timeout: 246 seconds) (~m-hrdsqi@user/shtrophic)17:58
cellmooseIf I want to fiddle around with the display cable (connecting/disconnecting different ends) is it recommended to disconnect the battery while doing it? I don't want to wear out the battery connetors, and it doesn't seem like it powers on from just external power being connected18:02
+ shtrophic (~m-hrdsqi@user/shtrophic)18:02
- erle (QUIT: Quit: K-lined) (~erle@user/erle)18:10
joschcellmoose: pocket reform?18:14
cellmoosejosch: yes18:16
joschin general, MNT recommends to disconnect power (including batteries) before performing changes to the electronics18:18
cellmoosejosch: I figured, that's why I have been disconnecting them so far. To give context, I'm not getting an image on the internal display (external works) so I'm trying to figure out if the display cable is connected properly18:29
- shtrophic (QUIT: Remote host closed the connection) (~m-hrdsqi@user/shtrophic)18:34
+ shtrophic (~m-hrdsqi@user/shtrophic)18:36
cellmooseFor anyone interested: It appears that I connected the ribbon cable upside down when I connected it to the display18:41
cellmooseThanks to everyone who answered questions!18:42
cellmooseAnd a special thanks to MNT that fixed the motherboard under warranty after I broke off the plastic tab on the display connector18:43
cellmooseI also have a suspiciously large amount of screws left for the number of steps left in the assembly guide18:52
elbI think I had almost no screws left until I got almost to the end, but one of the later packets has spares of almost every screw in the build18:58
elbat least two each of the M2 screws, and one M4 for the upper hinges, or so18:59
cellmooseelb: that sounds about the ones I have left19:13
+ FirefoxDeHuk (~FirefoxDe@109.108.69.106)19:27
- FirefoxDeHuk (QUIT: Quit: Client closed) (~FirefoxDe@109.108.69.106)19:31
+ AnimaInvicta (~AnimaInvi@88-120-179-216.subs.proxad.net)19:37
grimmwarejosch: you were on the money about those settings, I'm now able to introspect on my running kernel's data structures to write bpftrace scripts19:38
grimmwareI'm gonna write a blog post about how I set it up and how to use it so that minute can use it to debug the reboot issue;19:38
+ colinsane (~colinunin@97-113-137-58.tukw.qwest.net)20:07
- colinsane (QUIT: Ping timeout: 240 seconds) (~colinunin@97-113-137-58.tukw.qwest.net)20:12
+ vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50)20:14
grimmwareminute: you're gonna fuckin love this20:19
+ colinsane (~colinunin@97-113-142-187.tukw.qwest.net)20:27
+ pomel0 (~pomel0@user/pomel0)21:02
joschwow, the suspense! :D21:09
joschgrimmware: how much bigger is the debug kernel?21:09
vagrantcsoooo big.21:10
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50)21:12
vkoskivCan't bpftrace get the needed info from BTF?21:17
grimmwarehttps://community.mnt.re/t/kernel-tracing-on-the-reform-with-ebpf/396921:38
grimmwareminute: ^ take a look at this when you get a moment, should set you up to be able to do some tracing for the suspend stuff21:39
grimmwarevkoskiv: that's what I was attempting to enable, could be I've also ended up enabling stuff in the build that I didn't have to.21:39
joschgrimmware: are you sure that CONFIG_DEBUG_INFO_BTF=y is needed? That setting has been enabled for multiple years now and we don't disable it.21:53
grimmwareThe kernel didn’t have support built into it before and now it does21:54
joschgrimmware: yes, because you changed the build profile options, no?21:54
grimmwarejosch: I'll do a build tomorrow and then use bpftool to check if the bft data is in the resultant vmlinuz22:00
joschgrimmware: thank you! There is nothing in our kernel build scripts which disables CONFIG_DEBUG_INFO_BTF=y so if you indeed had to set CONFIG_DEBUG_INFO_BTF=y manually even after *not* building with the debug profiles enabled I'd be interested in hearing about it (and fix it).22:05
grimmwarejosch: yeah I'd also like to find the most lightweigh way of doing this because it would be absolutely gnarly to be able to provide people with tracing scripts they can run to debug hardware problems22:06
joschgrimmware: what are the sizes of the resulting packages? I guess there was a reason we disabled this was because of space (or build time?). This is in the scripts since their beginnings in 2022 without any rationale...22:10
- gustav25 (QUIT: Quit: Quit) (~gustav@c-78-82-53-228.bbcust.telenor.se)22:15
grimmwarejosch: http://paste.debian.net/1402861/22:25
grimmwareit's not totally implausible that there were other random packages in that directory22:26
grimmwareI use the filesystem like a complete shithead22:26
joschoh that explains :D22:32
joschi think we wanted to avoid shipping more than a gig of kernel artifacts22:32
grimmwareyeah, very legit22:34
grimmwarewell worst case scenario is that there's now a way to debug this stuff if you're willing to spare a gig22:34
- AnimaInvicta (PART: !!unknown attribute: msg!!) (~AnimaInvi@88-120-179-216.subs.proxad.net)22:57
+ dmisth (~dmisth@158.69.201.157)23:10
dmisthminute so the rk3588 reform won't have edk2? that's pretty disappointing imo23:14
joschdmisth: that's not what minute said23:15
minutedmisth: what's your main usecase for it?23:15
dmisththe arguments that person gave in the #linux-rockchip channel seem pretty non-sense. "edk2 codebase is extremely windows/dos like" what?23:15
dmisthwell for once the mainline uboot has no graphical output whatsoever23:15
dmisthso if anything goes wrong i really have to pull out a serial adapter..23:15
minutedmisth: no no23:15
minutedmisth: graphical ouput is exactly what i want23:16
minutedmisth: and EFI support23:16
minutedmisth: but at the moment it seems like this will be easier with barebox (or uboot)23:16
minutebecause edk2 is hard to work on23:16
dmisththe bsp uboot or?23:16
minutedmisth: no23:17
dmisthbecause mainline has had zero work for display output on the rk3588..23:17
minutewe don't use bsp uboot23:17
dmisthit looks like edk2 already has something in place....so how can that be harder?23:17
minutedmisth: i'm sorry but you come off as quite arrogant at the moment23:17
minuteminute: i'm one of two people who've spent a lot of time on making this work, not you... or am i wrong? why don't you just let me do my job?23:18
minuteminute: you'll get your graphical bootloader23:18
dmisthoh not my intention at all sorry i just read the chatlogs and was a bit disappointed seeing the arguments23:18
dmisthif we get graphical output with uboot then i'm happy :)23:19
minutedmisth: you'll get it23:19
minutei'm just trying to find a better and more stable way considering a lot of factors, that's all.23:19
dmisthso is the plan longterm to get this into upstream uboot?23:20
dmisthwould be pretty neat23:20
minutehttps://source.mnt.re/reform/mnt-reform-barebox/-/commits/reform2-rk3588?ref_type=HEADS#23:26
minutecurrently focusing on barebox, but i guess the code should be not too hard to port to uboot later23:26
- andreas-e (QUIT: Quit: Leaving) (~Andreas@2a02-8434-b6a3-e901-facc-8e87-8e54-890e.rev.sfr.net)23:26
minutegrimmware: awesome stuff btw!!!23:27
joschi remember how imx8mq u-boot display support was a community contribution by... nanocodebug? That was an epic commit23:28
joschi think mainly ported over from 9front23:28
- dmisth (QUIT: Quit: Ping timeout (120 seconds)) (~dmisth@158.69.201.157)23:34

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