| minute | elb`: oh no :/ on battery or on ac? | 00:27 |
|---|---|---|
| elb` | battery | 00:48 |
| elb` | I didn't look at the battery remaining (I was lecturing!) but I'm guessing something lke 30-40% | 00:48 |
| minute | elb`: ahh ok... then it's probably the same issue i have when i have it play a youtube dj set in the bathroom for an hour | 00:52 |
| - jacqueline (QUIT: Ping timeout: 256 seconds) (~jacquelin@user/jacqueline) | 00:56 | |
| - Ar|stote|is (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~linx@149.210.0.217) | 01:00 | |
| + Ar|stote|is (~linx@149.210.0.217) | 01:00 | |
| elb` | I have two lectures back to back, I boot at the beginning of the first, fold it up and stick it in my bag for the walk from one to the next, and then continue in the second; total duration is about two hours, and I've usually used it some / carried it all day powered off before that | 01:04 |
| elb` | maybe one out of four days it does something hinky in one lecture or the other | 01:05 |
| + jacqueline (~jacquelin@user/jacqueline) | 01:17 | |
| - AnimaInvicta (PART: !!unknown attribute: msg!!) (~AnimaInvi@88-120-179-216.subs.proxad.net) | 01:17 | |
| minute | elb`: got it. as a workaround, to avoid this issue from happening until i find a definitive fix (probably hardware fix though, or cpu/gpu throttling at low voltages via software), you could use a usb powerbank to support it from like the second hour on | 01:24 |
| - mjw (QUIT: Ping timeout: 260 seconds) (~mjw@2a02-a472-ffa1-1-f2e6-4728-2974-38b8.fixed6.kpn.net) | 01:50 | |
| - paperManu (QUIT: Ping timeout: 264 seconds) (~paperManu@64.58.44.160) | 03:28 | |
| + paperManu (~paperManu@64.58.44.160) | 03:51 | |
| - kop316 (QUIT: Remote host closed the connection) (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 03:56 | |
| + kop316 (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 03:58 | |
| - kop316 (QUIT: Read error: Connection reset by peer) (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 04:06 | |
| + kop316 (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 04:07 | |
| - kop316 (QUIT: Remote host closed the connection) (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 04:07 | |
| + kop316 (m-6f6zq6@static.138.159.90.157.clients.your-server.de) | 04:07 | |
| + sigrid_ (~sigrid@ftrv.se) | 04:16 | |
| + plomlomp0m (~plom@oldplay.plomlompom.com) | 04:20 | |
| + mlarkin_ (~mlarkin@syn-076-081-194-027.biz.spectrum.com) | 04:20 | |
| elb` | does the battery board just need bost/buck circuitry so it can send a constant voltage to the motherboard? | 04:20 |
| + sknebel_ (~quassel@v22016013254630973.happysrv.de) | 04:21 | |
| + mesaoptimizer0 (~mesa@user/PapuaHardyNet) | 04:22 | |
| + trypokarydos (~linx@149.210.0.217) | 04:24 | |
| - Ar|stote|is (QUIT: *.net *.split) (~linx@149.210.0.217) | 04:25 | |
| - plomtest (QUIT: *.net *.split) (~plom@user/plomtest) | 04:25 | |
| - mesaoptimizer (QUIT: *.net *.split) (~mesa@user/PapuaHardyNet) | 04:25 | |
| - sigrid (QUIT: *.net *.split) (~sigrid@ftrv.se) | 04:25 | |
| - sknebel (QUIT: *.net *.split) (~quassel@v22016013254630973.happysrv.de) | 04:25 | |
| - mlarkin (QUIT: *.net *.split) (~mlarkin@syn-076-081-194-027.biz.spectrum.com) | 04:25 | |
| - plomlompom (QUIT: *.net *.split) (~plom@oldplay.plomlompom.com) | 04:25 | |
| - abortretryfail (QUIT: *.net *.split) (~arf@146.ip-149-56-132.net) | 04:25 | |
| - schneider (QUIT: *.net *.split) (~schneider@irc2.xtort.eu) | 04:25 | |
| - svp (QUIT: *.net *.split) (~svp@2002:4f07:f0bd:0:e488:5bf3:17dc:e1ea) | 04:25 | |
| + svp (~svp@2002:4f07:f0bd:0:e488:5bf3:17dc:e1ea) | 04:25 | |
| + abortretryfail (~arf@146.ip-149-56-132.net) | 04:32 | |
| + schneider (~schneider@irc2.xtort.eu) | 04:32 | |
| - paperManu (QUIT: Ping timeout: 248 seconds) (~paperManu@64.58.44.160) | 04:34 | |
| minute | elb`: it absolutely has a buck/boost converter ^^ | 05:03 |
| minute | elb`: it's the mp2650 charger chip itself. it's possible that it needs more output capacitance. or that the cable could to be thicker gauge. or that the 5v buck on the mainboard needs more capacitance. | 05:07 |
| - wose1 (QUIT: Ping timeout: 256 seconds) (wose@mail.zuendmasse.de) | 05:26 | |
| - S0rin (QUIT: Ping timeout: 256 seconds) (~S0rin@user/s0rin) | 05:26 | |
| + wose1 (wose@mail.zuendmasse.de) | 05:26 | |
| + S0rin (~S0rin@user/s0rin) | 05:27 | |
| + reformer (~reformer@softboy.mntmn.com) | 05:29 | |
| - BAndiT1983 (QUIT: Quit: No Ping reply in 180 seconds.) (~quassel@46.101.193.235) | 05:36 | |
| - S0rin (QUIT: Ping timeout: 260 seconds) (~S0rin@user/s0rin) | 05:36 | |
| + S0rin (~S0rin@user/s0rin) | 05:36 | |
| + BAndiT1983 (~quassel@46.101.193.235) | 05:37 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 06:07 | |
| - trypokarydos (QUIT: Quit: No Ping reply in 180 seconds.) (~linx@149.210.0.217) | 07:03 | |
| + Ar|stote|is (~linx@149.210.0.217) | 07:03 | |
| - rodion_goritskov (QUIT: Ping timeout: 256 seconds) (~rodion_go@mail.goritskov.com) | 07:17 | |
| - mrbcmorris (QUIT: Read error: Connection reset by peer) (~mrbcmorri@1513413-static.lxtnkya3.metronetinc.net) | 07:18 | |
| + mrbcmorris (~mrbcmorri@1513413-static.lxtnkya3.metronetinc.net) | 07:18 | |
| - wose1 (QUIT: Ping timeout: 256 seconds) (wose@mail.zuendmasse.de) | 07:18 | |
| - abortretryfail (QUIT: Ping timeout: 256 seconds) (~arf@146.ip-149-56-132.net) | 07:18 | |
| + abortretryfail (~arf@146.ip-149-56-132.net) | 07:18 | |
| + reformer (~reformer@softboy.mntmn.com) | 07:21 | |
| - natalie (QUIT: Ping timeout: 256 seconds) (~natalie@user/natalie) | 07:59 | |
| + natalie (~natalie@user/natalie) | 08:01 | |
| - Guest4250 (QUIT: Ping timeout: 246 seconds) (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c) | 08:20 | |
| + Guest4250 (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c) | 08:21 | |
| - wose1 (QUIT: Ping timeout: 256 seconds) (wose@mail.zuendmasse.de) | 08:28 | |
| + wose1 (wose@mail.zuendmasse.de) | 08:28 | |
| - arminweigl (QUIT: Ping timeout: 260 seconds) (~arminweig@sourcehut/user/arminweigl) | 08:44 | |
| - minute (QUIT: Ping timeout: 260 seconds) (~mntirc@softboy.mntmn.com) | 08:47 | |
| - laumann (QUIT: Ping timeout: 260 seconds) (31d44fbb95@user/laumann) | 08:59 | |
| + minute (~mntirc@softboy.mntmn.com) | 08:59 | |
| + arminweigl (~arminweig@sourcehut/user/arminweigl) | 09:00 | |
| + laumann (31d44fbb95@user/laumann) | 09:01 | |
| * mesaoptimizer0 -> mesaoptimizer | 09:17 | |
| + mrbcmorris_ (~mrbcmorri@1513413-static.lxtnkya3.metronetinc.net) | 09:20 | |
| - mrbcmorris (QUIT: Ping timeout: 256 seconds) (~mrbcmorri@1513413-static.lxtnkya3.metronetinc.net) | 09:20 | |
| - jahkosha (QUIT: Ping timeout: 244 seconds) (~jahkosha@user/jahkosha) | 09:25 | |
| + jahkosha (~jahkosha@user/jahkosha) | 09:25 | |
| - svp (QUIT: Quit: Gateway shutdown) (~svp@2002:4f07:f0bd:0:e488:5bf3:17dc:e1ea) | 09:47 | |
| - frickler (QUIT: Ping timeout: 265 seconds) (~jens@user/frickler) | 10:03 | |
| + frickler (~jens@user/frickler) | 10:03 | |
| [tj] | most of the time on my pocket when I issue "poweroff" the screen looses context immediately and I get some lines drawn as it powers down, sometimes it stays on until kernel shutdown and I get kernel messages (imx8mp) | 11:20 |
| [tj] | seems like a race condition in the shutdown logic | 11:20 |
| + josch-pocket (~josch-poc@5.42.207.8) | 11:21 | |
| josch-pocket | hi, i'd like some ideas on the following problem: on two motherboards, with different dsi cables, with both imx8mq as well as a311d and different system images i have no display output | 11:23 |
| josch-pocket | here are some dmesg outputs from a311d: https://mister-muffin.de/p/COM4.txt https://mister-muffin.de/p/UL-G.txt | 11:23 |
| josch-pocket | it's two different classic reform and i didn't touch the display cable so i say it's unlikely to be that | 11:24 |
| + svp (~svp@host-79-7-240-189.business.telecomitalia.it) | 11:24 | |
| josch-pocket | my hunch is the dsi cable but i tried 3 different dsi connectors already | 11:25 |
| josch-pocket | my second hunch is that i'm forgetting something very obvious | 11:25 |
| josch-pocket | the systems boot just fine otherwise (as you can see from the fact that i was able to grab dmesg output) | 11:25 |
| josch-pocket | i was about to investigate the 'no display' issue from the forum and now i have none either :D | 11:26 |
| josch-pocket | i suspected the kernel but the issue is the same with a system image from july | 11:26 |
| josch-pocket | ideas? :) | 11:27 |
| - jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu) | 11:29 | |
| + jogu (~jogu@user/jogu) | 11:35 | |
| - jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu) | 11:42 | |
| + jogu (~jogu@user/jogu) | 11:43 | |
| - jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu) | 11:43 | |
| + jogu (~jogu@user/jogu) | 11:44 | |
| - swivel (QUIT: Ping timeout: 248 seconds) (~swivel@shells.gnugeneration.com) | 11:46 | |
| + swivel (~swivel@shells.gnugeneration.com) | 11:47 | |
| - elektron (QUIT: Ping timeout: 252 seconds) (~elektron@apoc.halo.nu) | 11:47 | |
| + elektron (~elektron@apoc.halo.nu) | 11:47 | |
| + plomtest (~plom@user/plomtest) | 12:29 | |
| + mjw (~mjw@2a02:a472:ffa1:1:f2e6:4728:2974:38b8) | 12:35 | |
| - sigrid_ (QUIT: Quit: leaving) (~sigrid@ftrv.se) | 12:38 | |
| + sigrid (~sigrid@ftrv.se) | 12:38 | |
| + paperManu (~paperManu@64.58.44.160) | 12:39 | |
| + timide (~timide@user/timide) | 12:43 | |
| + gustav2 (~gustav@c-78-82-52-97.bbcust.telenor.se) | 13:02 | |
| - svp (QUIT: Quit: Gateway shutdown) (~svp@host-79-7-240-189.business.telecomitalia.it) | 13:17 | |
| + svp (~svp@host-79-7-240-189.business.telecomitalia.it) | 13:24 | |
| - paperManu (QUIT: Ping timeout: 246 seconds) (~paperManu@64.58.44.160) | 14:09 | |
| + SadOp (~SadOp@81.209.189.43) | 14:33 | |
| SadOp | Hey, I'm having a hard time entering the u-boot command prompt when booting an RK3588 Reform. I connected the serial console via USB-C but the autoboot into boot menu can't be interrupted. There seems to be no wait for any key presses to get the u-boot cmd prompt. | 14:36 |
| gordon1 | SadOp: yeah it is quite annoying, isn't it? try hitting a key a bit earlier than you anticipate the prompt would appear, it can be done with a bit of practice | 14:37 |
| gordon1 | it is possible to get to u-boot console, i can assure you | 14:38 |
| gordon1 | what do you need u-boot console for btw? | 14:39 |
| + paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca) | 14:40 | |
| SadOp | gordon1: it really is quite frustrating. | 14:41 |
| SadOp | I want to get OpenBSD, successfully installed it on the NVMe. But now I need to learn how that uboot system works. | 14:41 |
| SadOp | *Openbsd working | 14:42 |
| SadOp | on a RK3588 system | 14:42 |
| gordon1 | right, there is an issue with that, u-boot in reform has no env, so you cannot pretty much do any changes there from default state, so you either has to configure it to boot from extlinux config as it boots linux, or you have to compile your own | 14:43 |
| SadOp | I'm fine with building my own - the build almost finishes successfully on the Reform (some dtb issue iirc. but nothing that I can't figure out). Booting from the SD card and telling it to use the NVMe would also work for me, but the OpenBSD bootloader does not recognize the NVMe, albeit being able to install to it. | 14:45 |
| SadOp | I tried the easy ways, but I think the u-boot needs to get involved. | 14:45 |
| gordon1 | SadOp: are you using script to build u-boot from source.mnt.re? | 14:46 |
| SadOp | yes | 14:46 |
| gordon1 | good, i don't remember any issues with dtbs there tho | 14:46 |
| SadOp | let me double check real quick | 14:46 |
| SadOp | Starting the uboot console is impossible for me and I'm hammering the keys like a maniac over here | 14:47 |
| gordon1 | is it possible for openbsd to have boot partition on emmc? | 14:47 |
| - SadOp (QUIT: Quit: Client closed) (~SadOp@81.209.189.43) | 14:49 | |
| + SadOp (~SadOp@81.209.189.43) | 14:53 | |
| gordon1 | are you sure that your uart adapter can tx stuff (or if you are using usb-c thing, tx dip switch is fully engaged)? | 14:54 |
| SadOp | well it accepts all my other keystrokes as soon as I'm in the boot menu | 14:54 |
| gordon1 | ah, i see, means it works | 14:54 |
| [tj] | SadOp: you can reconfigure uboot from userspace on linux to have a longer delay | 14:55 |
| [tj] | https://community.nxp.com/t5/Layerscape/How-to-modify-U-boot-configuration-in-Linux-user-space/td-p/918642?profile.language=en | 14:56 |
| [tj] | something like that | 14:56 |
| gordon1 | no, you can't, there is no env | 14:56 |
| SadOp | yeah pretty sure I'd need to recompile u-boot to make that happen | 14:57 |
| [tj] | modify the flash in place? | 14:57 |
| gordon1 | SadOp: did you try ctrl-c the boot process? | 14:57 |
| [tj] | having done this I'd recommend not doing this | 14:57 |
| gordon1 | i'm pretty sure there is crc somewhere that will make u-boot borked | 14:57 |
| [tj] | yeah it breaks the env, but that then drops you to a prompt | 14:58 |
| [tj] | I have never claimed my ideas were good | 14:58 |
| SadOp | hehe, thanks for the idea though! | 14:59 |
| josch | SadOp: we purposefully minimized the delay for rk3588 to the absolute minimum (i think it was 0.1 seconds) because there is no boot menu other than over uart | 15:00 |
| [tj] | I have used this to get uboot control on an embeded tv box I was repurposing | 15:00 |
| josch | SadOp: the choice was made because those who have a uart adapter usually also know how to build their own u-boot with a longer custom delay | 15:00 |
| [tj] | afaict from uboot source you should always be able to break in if prompted, but in practice I find it really difficult | 15:00 |
| SadOp | josch: I'm building it right now, although getting an error like I said before: https://pastebin.com/ijv3d4c0 | 15:01 |
| SadOp | I hoped build.sh would do the job for me and I only need to touch the configs | 15:02 |
| josch | one sec, i have to get my browser back | 15:02 |
| josch | i was without display for 21 hours... | 15:02 |
| gordon1 | bootdelay is 0, i don't think you can have fractional values there | 15:03 |
| josch | no, bootdelay is 1 | 15:03 |
| josch | it was set to 0 in 14aed286a2c444331c42c2d2670209e5dc80b9a0 | 15:03 |
| gordon1 | but that shouldn't stop the abort from happening, there might be bootdelaykey or bootstopkey configured tho, i'm checking it out | 15:03 |
| gordon1 | well, it is 0 in my u-boot | 15:04 |
| josch | i reverted that in 3ec14e945457db8dc9c357925afcebc5a6c0085b | 15:04 |
| gordon1 | which i didn't change | 15:04 |
| gordon1 | yet | 15:04 |
| josch | gordon1: than your rk3588 u-boot is older than 2025-05-06 | 15:04 |
| josch | SadOp: i just looked at your error -- i've never seen that before, i have to investigate | 15:05 |
| - SadOp (QUIT: Quit: Client closed) (~SadOp@81.209.189.43) | 15:05 | |
| gordon1 | probably need to update it then (though i definitely received my reform way after may, more like in august) | 15:05 |
| gordon1 | weird, you don't need tpl for rk3588, do you? | 15:06 |
| + SadOp (~SadOp@81.209.189.43) | 15:07 | |
| SadOp | josch: thanks, I'm building this on the Reform itself if that makes a difference... | 15:07 |
| josch | SadOp: on the latest Debian system image? Then i should be able to reproduce it | 15:07 |
| SadOp | freshly updated today | 15:08 |
| SadOp | (with apt mind you) | 15:08 |
| josch | how else? :) | 15:09 |
| gordon1 | josch: you should really consider enabling u-boot env tho, it will solve all the problems | 15:09 |
| josch | gordon1: you mean that users can save their environment? | 15:10 |
| - SadOp (PART: !!unknown attribute: msg!!) (~SadOp@81.209.189.43) | 15:10 | |
| gordon1 | yes, like modifying the bootdelay for example | 15:10 |
| gordon1 | or changing the bootcmd, from linux using fw_setenv | 15:10 |
| josch | gordon1: it was explicitly disabled after users soft-bricked their device with bad variable values | 15:10 |
| + SadOp_ (~jan@81.209.189.43) | 15:10 | |
| gordon1 | right | 15:10 |
| gordon1 | oh well | 15:10 |
| * SadOp_ -> SadOp | 15:11 | |
| josch | it happened quite a few times during imx8mq days | 15:11 |
| gordon1 | maybe a factory reset pin that forces u-boot to reinit its env is worth considering | 15:11 |
| gordon1 | just a though | 15:11 |
| gordon1 | *thought | 15:11 |
| SadOp | I agree that bricking the device should be as hard as possible | 15:13 |
| SadOp | (or not as easy as possible at least) | 15:13 |
| gordon1 | josch: oh wait, is it because imx8 has u-boot console accessible on the display not uart? | 15:16 |
| josch | gordon1: because imx8mq has display for u-boot it was easier to modify the env, yes | 15:18 |
| gordon1 | right, i see now | 15:18 |
| + gachikuku_ (~gachikuku@kamil.gr) | 15:18 | |
| josch | but because the others do not and require a uart adapter you could even more argue: those who can interact with u-boot do have to have a uart adapter and are even more likely to be able to just build their own u-boot with whatever config they want | 15:19 |
| gordon1 | ffs u-boot refuses to compile at all now :( | 15:19 |
| josch | same error as SadOp? | 15:19 |
| gordon1 | no, something about python and swig | 15:19 |
| gordon1 | that's probably because gentoo is so blindingly bleeding edge | 15:20 |
| gordon1 | https://github.com/openwrt/openwrt/issues/17345 yep | 15:21 |
| josch | i suspect the problem is that build.sh for rk3588 blindly pulls in the latest git HEAD instead of pinning to a known tag | 15:21 |
| gordon1 | it does pin it, doesn't it? | 15:22 |
| gordon1 | yes, it does | 15:22 |
| SadOp | gordon1: you need to downgrade swig | 15:22 |
| gordon1 | that's a defeatist approach! | 15:22 |
| SadOp | I installed 4.2.1 from sources to get the compile step going | 15:22 |
| josch | gordon1: only for u-boot, not for the other repos | 15:22 |
| gordon1 | oh, i see now | 15:23 |
| gordon1 | ddr file name probably changed | 15:24 |
| josch | i have the known good hashes here, we can easily revert | 15:24 |
| gordon1 | SadOp: try to do git checkout 0f8ac860f0479da56a1decae207ddc99e289f2e2 in rkbin repo | 15:25 |
| gordon1 | that is a hash i used to build u-boot for my rock5b+ | 15:25 |
| gordon1 | (using reform build script) | 15:25 |
| SadOp | trying | 15:26 |
| gordon1 | SadOp: also if you are feeling adventurous you can change this line https://source.mnt.re/reform/reform-rk3588-uboot/-/blob/main/build.sh?ref_type=heads#L25 to v1.19 instead | 15:27 |
| SadOp | sure thing, the build.sh is already changed so lets have it | 15:29 |
| josch | i'm performing a couple of fixes in the rk3588-u-boot build.sh right now | 15:30 |
| josch | so if you want something changed, now is a good time to send me your wishlist :) | 15:30 |
| SadOp | I added a cleanup step that removes the cloned repos (cleaner way would be to check and update tho). | 15:31 |
| gordon1 | josch: that's the last commit with v1.18 272136301989d493000425935bd4ead0ef04d06b in rkbin | 15:31 |
| josch | minute: I think it would also be a good idea to have a mirror of collabora u-boot git, rkbin and atf on source.mnt.re in case collabora.com or github.com become unavailable for whatever reason | 15:31 |
| gordon1 | but i think it would be more beneficial to try to update ddr code to v1.19 | 15:31 |
| gordon1 | josch: i know it is too much to ask, but it would be real nice to have env back at least for rk3588 | 15:32 |
| gordon1 | at least before u-boot console is on the keyboard+display | 15:33 |
| josch | gordon1: but if you want it, why don't you just build your custom u-boot and call it a day? why must it be on for everybody? | 15:33 |
| gordon1 | yeah yeah, i could | 15:34 |
| gordon1 | just can't be bothered since everything works now... | 15:34 |
| gordon1 | i'll wait till something breaks | 15:34 |
| SadOp | oh nice this is not a arm64 binary: rkbin/tools/boot_merger | 15:35 |
| SadOp | this breaks now for building it on a reform | 15:35 |
| gordon1 | josch: i just think that _if_ uboot console accessible only through uart there is no harm of having the env | 15:35 |
| SadOp | gordon1: commit hash 272136301989d493000425935bd4ead0ef04d06b works, the other one didn't | 15:35 |
| gordon1 | and it helps someone who, like me, maybe want to tweak it a little but can't be bothered to reflash it completely | 15:36 |
| josch | thank you for confirming, lets pin it to 272136301989d493000425935bd4ead0ef04d06b for now | 15:36 |
| josch | gordon1: bring it up with minute -- it's not my call to make :) | 15:36 |
| gordon1 | sure, i mean tbf it's not something that i feel strongly about | 15:37 |
| gordon1 | https://github.com/openwrt/openwrt/pull/17352 FIY | 15:38 |
| gordon1 | not sure that rng seed patch is something to do with swig tho | 15:38 |
| josch | gordon1: you can read commit e391ee96a87a2b80c2165f65f80ec2581bc53500 from 2022 for some of the history | 15:39 |
| SadOp | gordon1: yeah, thats the reason I downgraded swig locally. I tried a more recent u-boot version but thats a whole other fight then | 15:40 |
| SadOp | I can't find the source for rkbin/tools/boot_merger, its just a binary. what does it even do? | 15:41 |
| gordon1 | SadOp: iirc it is not required to build the image, josch can you check since you are in build.sh anyways? | 15:42 |
| gordon1 | i remember there were some actions that weren't required to get final image | 15:42 |
| SadOp | I mean in general it would be superb to be able to build this on the Reform itself and not to rely on another amd64 machine | 15:43 |
| josch | please give me a minute, i'm fighting other things in the build process right now | 15:43 |
| josch | but yes, boot_merger has no source and we were wondering about that a few months ago already... | 15:43 |
| gordon1 | no rush | 15:44 |
| gordon1 | my rock5b+ build script does not use boot_merger | 15:45 |
| gordon1 | and it was based on reform script, so i guess 6 months ago i just nuked it from there w/o any adverse effects | 15:46 |
| gordon1 | applied patch from openwrt link, u-boot compiled successfully with 4.3.0 swig | 15:49 |
| SadOp | https://github.com/rockchip-toybrick/u-boot/blob/master/tools/rockchip/boot_merger.c found this for the boot_merger | 15:49 |
| SadOp | gordon1: nice | 15:50 |
| - mjw (QUIT: Ping timeout: 244 seconds) (~mjw@2a02:a472:ffa1:1:f2e6:4728:2974:38b8) | 15:52 | |
| - josch-pocket (QUIT: Quit: josch-pocket) (~josch-poc@5.42.207.8) | 16:03 | |
| SadOp | prefixing the boot_merger commands with qemu-amd64 works. so with a little bit of machhine type logic and the git hash change it now works. | 16:19 |
| josch | ah indeed, i have qemu enabled via linux binfmt | 16:21 |
| josch | so i'm not noticing that an amd64 binary is running | 16:21 |
| josch | and the MNT gitlab CI is cross-building this on amd64 | 16:21 |
| SadOp | I can link the diff of my build.sh in a bit. | 16:22 |
| josch | the MNT gitlab CI also didn't run into the swip issue, because it's building everything from Debian oldstable (bookworm) | 16:22 |
| SadOp | josch: to install the build u-boot I dd it to the boot partition? | 16:23 |
| josch | not for rk3588 -- you dd it at the right offset on the main disk, not to the boot0 partitions | 16:23 |
| josch | though maybe experiment with flashing it to sd-card first | 16:23 |
| josch | and flash to emmc once you know it works | 16:23 |
| josch | recovery could otherwise become tricky | 16:24 |
| josch | rk3588 will read u-boot from sd-card if it cannot find it on emmc, so you have to wipe the first 16MBit with zeroes | 16:24 |
| - SadOp (QUIT: Quit: Konversation terminated!) (~jan@81.209.189.43) | 16:38 | |
| josch | nooooooo we need the link to the build.sh! XD | 16:39 |
| + TheProtoHobo (~dfbarth@mobile-166-173-184-165.mycingular.net) | 16:40 | |
| - jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu) | 16:41 | |
| + jogu (~jogu@user/jogu) | 16:42 | |
| - gustav2 (QUIT: Quit: Quit) (~gustav@c-78-82-52-97.bbcust.telenor.se) | 16:53 | |
| + mjw (~mjw@178.230.76.61) | 17:06 | |
| minute | SadOp: normal if you press and hold ctrl-c a bit before the uboot output, it should interrupt | 17:15 |
| minute | (over serial) | 17:15 |
| minute | that's how i do it | 17:15 |
| minute | s/normal/normally | 17:15 |
| - mjw (QUIT: Read error: Connection reset by peer) (~mjw@178.230.76.61) | 17:16 | |
| + mjw (~mjw@178.230.76.61) | 17:16 | |
| - mjw (QUIT: Ping timeout: 245 seconds) (~mjw@178.230.76.61) | 17:22 | |
| + mjw (~mjw@178.230.76.61) | 17:22 | |
| - jogu (QUIT: Remote host closed the connection) (~jogu@user/jogu) | 17:24 | |
| + colinsane (~colinunin@97-113-89-123.tukw.qwest.net) | 17:40 | |
| - mjw (QUIT: Read error: Connection reset by peer) (~mjw@178.230.76.61) | 17:44 | |
| + SadOp (~jan@81.209.189.43) | 17:44 | |
| + mjw (~mjw@178.230.76.61) | 17:44 | |
| SadOp | minute: sorry I was in the kitchen and got disconnected. I'll try the ctrl+c method now | 17:45 |
| SadOp | josch: thanks for the additional info | 17:45 |
| SadOp | it's quite funny, when I hammer ctrl+c when booting the machine boots from the NVMe and fires up OpenBSD lol | 17:49 |
| SadOp | and I did not flash a new u-boot | 17:52 |
| - mjw (QUIT: Remote host closed the connection) (~mjw@178.230.76.61) | 17:54 | |
| + mjw (~mjw@178.230.76.61) | 17:54 | |
| josch | :D | 17:55 |
| josch | SadOp, gordon1: i'm looking for comments: https://source.mnt.re/reform/reform-rk3588-uboot/-/merge_requests/10 | 17:55 |
| josch | minute: did you ever compare battery runtimes of the different SoMs? When disconnecting screen and keyboard i'm getting insanely long runtimes when the system is idle. For example a311d manages more than 20 hours on the 2200 mAh cells in classic reform. | 17:56 |
| josch | i'll post my findings in the forum but was wondering if you maybe have an explanation | 17:56 |
| josch | are more than 12 hours of runtime just eaten up by the display backlight and is my reform then actually a rather intricate torch light? :D | 17:58 |
| josch | i should just connect keyboard+trackball and just display next to find out what difference that makes in runtime... | 17:58 |
| - mjw (QUIT: Ping timeout: 264 seconds) (~mjw@178.230.76.61) | 18:01 | |
| SadOp | josch: how about an if [[ `uname -m` == "aarch64" ]] for the boot_merger part? | 18:01 |
| SadOp | so that it can be build on the Reform with qemu | 18:02 |
| josch | SadOp: i'd rather fix this by building boot_merger from source if possible. I very much dislike running some proprietary blob... | 18:03 |
| josch | but i don't know if that's possible | 18:03 |
| josch | if not, then yes, we can special-case amd64 | 18:03 |
| josch | and use qemu for other architectures | 18:03 |
| SadOp | I looked for the source, but everything I found was an outdated on that does not understand the used config file fully | 18:03 |
| SadOp | there is even an open GitHub issue | 18:04 |
| josch | yes, i think that was the status also last time i checked :( | 18:04 |
| josch | SadOp: do you have the github issue at hand? | 18:04 |
| SadOp | https://github.com/rockchip-linux/u-boot/issues/90 | 18:05 |
| josch | thank you! | 18:05 |
| SadOp | The first comment translates (via google translate) to something like "i have the same issue: | 18:05 |
| SadOp | The first comment translates (via google translate) to something like "i have the same issue" | 18:05 |
| SadOp | oops, this is not Signal, sorry | 18:06 |
| josch | what would be different in signal? | 18:06 |
| SadOp | arrow up lets you edit the last message | 18:06 |
| josch | ah :) | 18:07 |
| josch | same with many (most?) xmpp and matrix clients | 18:07 |
| SadOp | I extracted the boot_merger source code and build it locally with the 2 needed files and a one line change. Too bad it doesn't work. | 18:07 |
| josch | SadOp: is it not this one? https://github.com/rockchip-linux/u-boot/blob/next-dev/tools/rockchip/boot_merger.c | 18:08 |
| SadOp | thats the one | 18:08 |
| josch | another item for the todo list... | 18:09 |
| SadOp | I only downloaded the c and h file used cc and copied it into the repo #include <version.h> should be #include <linux/version.h> and it builds | 18:09 |
| SadOp | errors out in the while loop starting on line 391 | 18:10 |
| SadOp | basically in the parser part of the tool | 18:10 |
| josch | you are further than we were the last time that boot_merger came up in this channel: https://mntre.com/reform-irc-logs/2025-02-07.log.html#t00:08:38 | 18:15 |
| SadOp | josch: running your build.sh locally now to see if everything works | 18:17 |
| josch | thank you! | 18:17 |
| SadOp | (with a little qemu-amd64 addition for the Reform) | 18:17 |
| SadOp | happy to help! | 18:17 |
| josch | SadOp: you could install qemu-user, then you don't have to patch build.sh | 18:18 |
| SadOp | wait, I have that installed | 18:18 |
| SadOp | does it figure it out by itself then? | 18:18 |
| josch | SadOp: sorry, it's called qemu-user-binfmt | 18:19 |
| josch | this will register qemu-user via the binfmt mechanism with the kernel | 18:19 |
| josch | and from that point on, foreign elf binaries will transparently be emulated with qemu | 18:20 |
| SadOp | I even have that. lol I did too much work :D | 18:20 |
| SadOp | it just works | 18:20 |
| josch | okay, nice :) | 18:20 |
| SadOp | thanks for the hint! | 18:20 |
| - ch (QUIT: Ping timeout: 260 seconds) (~ch@user/meow/ch) | 18:24 | |
| + ch (~ch@user/meow/ch) | 18:26 | |
| SadOp | josch: your patched build.sh works flawlessly on my Reform2 with qemu-user-binfmt installed | 18:28 |
| SadOp | thanks for that, I really appreciate! | 18:28 |
| josch | great to hear, thank you for testing | 18:29 |
| josch | i didn't test the resulting binary yet, though | 18:29 |
| SadOp | me neither, if I nuke the first 16MB (?) of the emmc and dd it to an sd card it should work, right? | 18:31 |
| josch | not necessarily but at least with that approach you don't soft-brick your rk3588 but can re-flash the sd-card with a working image :) | 18:32 |
| qbit | oooo nixos-hardware for the rk3588! | 18:33 |
| josch | so... i tried to figure out what boot_merger does but after commenting out the calls to boot_merger in boot.sh the resulting rk3588-mnt-reform2-flash.bin is bit-by-bit identical to before -- can this just be removed? o0 | 18:43 |
| - GNUmoon2 (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon) | 18:50 | |
| + GNUmoon2 (~GNUmoon@gateway/tor-sasl/gnumoon) | 18:51 | |
| - colinsane (QUIT: Quit: bye) (~colinunin@97-113-89-123.tukw.qwest.net) | 18:58 | |
| + colinsane (~colinunin@97-113-89-123.tukw.qwest.net) | 19:04 | |
| gordon1 | josch: i didn't check if reform's u-boot does boot if you remove boot_merger lines, but i can assure that it does _compile_ w/o boot_merger lines and also rock5b+ u-boot boots and works without them | 19:10 |
| gordon1 | josch: also s/SWIP/SWIG/ in the MR description | 19:10 |
| gordon1 | if it is bit-by-bit identical i don't see a reason why keep it tbf | 19:11 |
| SadOp | 100% agree. It's a useless binary blob then | 19:11 |
| + AnimaInvicta (~AnimaInvi@88-120-179-216.subs.proxad.net) | 19:12 | |
| elb` | minute: my mistake; for some reason I thought the raw combined battery voltage was what is transported to the mainboard, I must have misunderstood | 19:21 |
| - minute (QUIT: Ping timeout: 246 seconds) (~mntirc@softboy.mntmn.com) | 19:25 | |
| - jahkosha (QUIT: Ping timeout: 246 seconds) (~jahkosha@user/jahkosha) | 19:26 | |
| - n_to (QUIT: Ping timeout: 246 seconds) (~n_to@2a03:4000:6:3662:24b1:57ff:fec6:76c1) | 19:26 | |
| + n_to (~n_to@2a03:4000:6:3662:24b1:57ff:fec6:76c1) | 19:26 | |
| elb` | I'm handy with a soldering iron, if you're looking for people to try things that can be bodged, feel free to ask! | 19:27 |
| + jahkosha_ (~jahkosha@user/jahkosha) | 19:27 | |
| * jahkosha_ -> jahkosha | 19:27 | |
| + minute (~mntirc@softboy.mntmn.com) | 19:27 | |
| josch | gordon1, SadOp: thank you, i removed it in my MR | 19:39 |
| + BAndiT1983_ (~quassel@46.101.193.235) | 19:47 | |
| - BAndiT1983 (QUIT: *.net *.split) (~quassel@46.101.193.235) | 19:51 | |
| + reform15220 (~rene@2001:9e8:33b8:2900:7107:9943:42fe:8ef4) | 19:59 | |
| * reform15220 -> rwa_ | 19:59 | |
| - paperManu (QUIT: Ping timeout: 256 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca) | 20:20 | |
| - rwa_ (QUIT: Quit: Leaving) (~rene@2001:9e8:33b8:2900:7107:9943:42fe:8ef4) | 20:20 | |
| + paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca) | 20:22 | |
| - svp (QUIT: Quit: Gateway shutdown) (~svp@host-79-7-240-189.business.telecomitalia.it) | 20:23 | |
| + svp (~svp@host-79-7-240-189.business.telecomitalia.it) | 20:30 | |
| minute | josch: yeah the backlight always pulls a lot of power. that's why phones are so aggressively dimming on idle | 20:30 |
| SadOp | josch: new version of the script works. I'll now start testing the new u-boot | 20:30 |
| josch | thank you! | 20:32 |
| SadOp | josch: just to make sure, dd if=/dev/zero of=/dev/mmcblk0 bs=1M count=16 would be the correct thing to wipe the first 16mb of the emmc? | 20:37 |
| SadOp | well, only the target is of interest here, the rest should be fine :D | 20:38 |
| gordon1 | i don't think you have to do that | 20:38 |
| SadOp | ok, how do I test the u-boot on my sd-card then? | 20:38 |
| gordon1 | ah, you want to boot from sd card, i see, i thought you want to put new uboot on emmc | 20:39 |
| SadOp | after testing it on the sd-card I'd love to tho | 20:39 |
| gordon1 | tbh i don't see the benefits of testing it from sd card if you render emmc not bootable anyways... | 20:40 |
| gordon1 | SadOp: i would highly recommend you to get a contingency plan ready, in case it won't boot up - have separate machine with usb cable ready to program it using rkdevtool | 20:41 |
| gordon1 | *rkdeveloptool | 20:41 |
| gordon1 | but yeah if i remember correctly rk3588 searches for some magic on the device to try to boot from it so wiping should work | 20:42 |
| gordon1 | but i would wait for someone who actually knows what they're doing | 20:42 |
| SadOp | yes, I will! thank you tho! | 20:42 |
| SadOp | I should document my doings I think | 20:42 |
| gordon1 | i booted rk3588 successfully only from spi flash so far | 20:42 |
| SadOp | Since I haven't been doing anything with arm for a long time this is really new to me. But its fun so far. And OpenBSD boots when I hammer ctrl+c over serial. No idea why since its on the NVMe but it boots. | 20:47 |
| gordon1 | well, that's an achievement already! | 20:52 |
| SadOp | it really is. also the build.sh script is improved. thats nice | 20:54 |
| SadOp | Ok, holding ctrl+c down actually gets me into the uboot console | 20:56 |
| SadOp | another achievement :D | 20:56 |
| gordon1 | with new u-boot? | 20:57 |
| SadOp | no with the old | 20:57 |
| gordon1 | minute: how are you flashing u-boot on fresh factory fireflies btw? | 20:57 |
| gordon1 | if not a secret | 20:57 |
| - paperManu (QUIT: Ping timeout: 244 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca) | 21:10 | |
| kfx | if the som can't load u-boot from the emmc it will try the sd card | 21:10 |
| kfx | I've never had to mess with spi on mine | 21:10 |
| gordon1 | i would love to mess with spi but i don't think firefly's module has spi flash | 21:11 |
| josch | kfx: what i want to cautio about is: what if it *can* load your u-boot but you built a u-boot which is borked and does not boot you far enough such that you can repair the mess you made? | 21:11 |
| gordon1 | josch: dropped another comment to MR | 21:11 |
| + paperManu (~paperManu@modemcable141.205-200-24.mc.videotron.ca) | 21:12 | |
| kfx | josch: I've done that and sadly I didn't take good notes, now that I look. You can force maskrom mode by jumping GND and CLK on the SOM and then us rkdevtool to put another u-boot on emmc | 21:18 |
| kfx | usually I save the commands I use but I don't appear to have done that... guess I was in a hurry | 21:18 |
| gordon1 | finally managed to successfully print trackball keycaps, dunno why SLA gives me so much grief https://litter.catbox.moe/h3panh.JPG | 21:19 |
| gordon1 | i said "successfully" and then zoomed in, it is not particularly successful, is it? but at least much better than my other attempts | 21:20 |
| gordon1 | i have commands right here | 21:23 |
| gordon1 | rkdeveloptool db rk3588_spl_loader_v1.18.113.bin | 21:23 |
| gordon1 | rkdeveloptool wl 0 /tmp/spi.img | 21:23 |
| gordon1 | i think second command must be changed to wl 1 since build.sh generates image with 1 sector offset | 21:23 |
| gordon1 | and spl loader should be in rkbin | 21:24 |
| gordon1 | at least we can have it saved in irc logs | 21:24 |
| gordon1 | ah no i confused it with something else, there is a seek=16320 so wl 16320 i guess? | 21:26 |
| gordon1 | and that is the guide how to enter maskrom mode on icore https://wiki.t-firefly.com/en/iCore-3588Q/upgrade_maskrom_mode.html | 21:27 |
| kfx | gordon1: I'd be curious why seek=16320 since I thought u-boot went in 16384 ( from https://opensource.rock-chips.com/wiki_Partitions ) | 21:30 |
| gordon1 | kfx: ah, nevermind, i remember now, it prepares the image minus the mbr and partition table etc, so it puts idbloader first then after 16320 it puts u-boot, but expects the whole images to be written at 16384-16320 at 64 sectors, so i guess in this case the command will be rkdeveloptool wl 64 u-boot-flash.bin | 21:33 |
| gordon1 | it would be really nice if someone can verify the math ^ | 21:33 |
| gordon1 | and i think magic that rk3588 boot code searches for supposed to be at 32k (or 64 sectors) mark | 21:34 |
| + mjw (~mjw@gnu.wildebeest.org) | 22:44 | |
| kfx | gordon1: makes sense | 22:51 |
| gordon1 | i have magic RKNS located at 32k offset on my emmc, same as in the beginning of idbloader.bin | 22:53 |
| minute | gordon1: we have a microsd card that contains an uboot.scr that erases the first 16MB or so of emmc. after that, we boot a test and install microsd. | 22:59 |
| gordon1 | right, i see, so you don't use usbboot | 22:59 |
| minute | no | 23:00 |
| gordon1 | okay, that's something to discover on my own then | 23:01 |
| minute | gordon1: nice tb cap prints! | 23:01 |
| gordon1 | they have a long way to go, i still have those weird layer failure lines that i cannot understand | 23:01 |
| minute | gordon1: here's a far out idea for you: the system ctrl is connected to spi as a device. it could emulate a spi flash. | 23:02 |
| gordon1 | ooh, and extconn is connected to it | 23:02 |
| gordon1 | so it can potentially has more storage | 23:03 |
| gordon1 | thanks for the idea! | 23:03 |
| minute | right, there's also an i2c header on the board connected to sysctl | 23:03 |
| minute | also, it is itself running from spi flash and could pass parts of it through... | 23:04 |
| gordon1 | well, mine is running from onboard flash i would guess | 23:04 |
| gordon1 | but on rp2040-based ones, yes | 23:04 |
| minute | ah right sorry. classic reform | 23:04 |
| minute | but yeah, lpc11u24 could also be a usb host for the flashing port. | 23:05 |
| minute | (if you need storage i mean) | 23:05 |
| gordon1 | hmm | 23:05 |
| minute | or yeah, the "hack the planet" port is there of course :D | 23:05 |
| - paperManu (QUIT: Ping timeout: 256 seconds) (~paperManu@modemcable141.205-200-24.mc.videotron.ca) | 23:05 | |
| gordon1 | yeah that's probably path of least resistance, just plug a little board with extra spi flash or even microsd slot | 23:06 |
| - colinsane (QUIT: Quit: bye) (~colinunin@97-113-89-123.tukw.qwest.net) | 23:25 | |
| - AnimaInvicta (QUIT: Ping timeout: 256 seconds) (~AnimaInvi@88-120-179-216.subs.proxad.net) | 23:25 | |
| - frickler (QUIT: Ping timeout: 256 seconds) (~jens@user/frickler) | 23:25 | |
| + frickler (~jens@user/frickler) | 23:25 | |
| + AnimaInvicta (~AnimaInvi@88-120-179-216.subs.proxad.net) | 23:25 | |
| - xktr (QUIT: Ping timeout: 246 seconds) (~xktr@user/xktr) | 23:37 | |
| - Ar|stote|is (QUIT: Ping timeout: 246 seconds) (~linx@149.210.0.217) | 23:37 | |
| - gordon1 (QUIT: Ping timeout: 246 seconds) (~gordon1@citadel.niflheim.info) | 23:37 | |
| - blast007 (QUIT: Read error: Connection reset by peer) (~blast@user/blast007) | 23:37 | |
| - n_to (QUIT: Ping timeout: 246 seconds) (~n_to@2a03:4000:6:3662:24b1:57ff:fec6:76c1) | 23:37 | |
| - SadOp (QUIT: Ping timeout: 246 seconds) (~jan@81.209.189.43) | 23:37 | |
| + blast007` (~blast@user/blast007) | 23:39 | |
| + gordon2 (~gordon1@citadel.niflheim.info) | 23:39 | |
| + n_to_ (~n_to@2a03:4000:6:3662:24b1:57ff:fec6:76c1) | 23:39 | |
| + SadOp_ (~jan@81.209.189.43) | 23:39 | |
| + trypokarydos (~linx@149.210.0.217) | 23:39 | |
| + xktr_ (~xktr@user/xktr) | 23:39 | |
| - TheProtoHobo (QUIT: Quit: leaving) (~dfbarth@mobile-166-173-184-165.mycingular.net) | 23:48 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
