| AbortRetryFail | Who was it who had trouble getting FreeCAD working on the imx8 Reform? I got it working. | 01:16 |
|---|---|---|
| AbortRetryFail | Qt Wayland bugs I guess. QT_QPA_PLATFORM=xcb works | 01:18 |
| - jjbliss (QUIT: Ping timeout: 246 seconds) (~jjbliss@1464766-static.elnsmiaa.metronetinc.net) | 01:50 | |
| - robin (QUIT: Quit: Leaving) (~robin@user/terpri) | 01:55 | |
| - Ar|stote|is (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~linx@149.210.24.247) | 02:10 | |
| + Ar|stote|is (~linx@149.210.24.247) | 02:11 | |
| + jjbliss (~jjbliss@1464766-static.elnsmiaa.metronetinc.net) | 02:20 | |
| violet | minute: what'd you learn? anything i'd need to be thinking about? | 03:27 |
| noam | minute: I'll double check that port | 04:17 |
| noam | and if it doesn't work I'll try ptftpd, thanks :) | 04:17 |
| - Ar|stote|is (QUIT: Read error: Connection reset by peer) (~linx@149.210.24.247) | 04:31 | |
| + Ar|stote|is (~linx@149.210.25.116) | 04:38 | |
| - macewentoo (QUIT: Ping timeout: 246 seconds) (~macewento@gateway/vpn/pia/paperowl) | 05:00 | |
| - colinsane (QUIT: Ping timeout: 246 seconds) (~colinunin@97-113-136-129.tukw.qwest.net) | 05:48 | |
| - Ar|stote|is (QUIT: Ping timeout: 252 seconds) (~linx@149.210.25.116) | 06:08 | |
| + Ar|stote|is (~linx@149.210.24.197) | 06:13 | |
| + colinsane (~colinunin@97-113-136-129.tukw.qwest.net) | 06:23 | |
| - Ar|stote|is (QUIT: Ping timeout: 252 seconds) (~linx@149.210.24.197) | 06:25 | |
| + Ar|stote|is (~linx@149.210.26.156) | 06:30 | |
| - Ar|stote|is (QUIT: Ping timeout: 268 seconds) (~linx@149.210.26.156) | 06:58 | |
| + macewentoo (~macewento@gateway/vpn/pia/paperowl) | 07:00 | |
| + Ar|stote|is (~linx@149.210.17.185) | 07:17 | |
| - Ar|stote|is (QUIT: Ping timeout: 268 seconds) (~linx@149.210.17.185) | 07:31 | |
| + chomwitt (~chomwitt@2a02:587:7a1d:df00:1ac0:4dff:fedb:a3f1) | 07:50 | |
| josch | minute: maybe a topic worth your attention is the keyboard failing on ls1028a and a311d after the latest apt upgrades: https://community.mnt.re/t/no-keyboard-input-with-ls1028-board-and-latest-image/1994/11 | 07:56 |
| + jaume (~user@user/jaume) | 08:34 | |
| minute | josch: ugh, that's strange | 08:37 |
| - nybble (QUIT: Ping timeout: 256 seconds) (~sprang@user/csprng) | 09:01 | |
| noam | minute: the USB port by the HDMI works | 09:44 |
| noam | :) :) | 09:44 |
| minute | noam: ah very nice | 09:49 |
| noam | yeah, just gotta burn an image to that USB :P | 09:51 |
| + mjw (~mjw@gnu.wildebeest.org) | 09:53 | |
| minute | noam: it is possible that you'll have to (re)assign some load addresses for our boot.scr to work | 10:05 |
| noam | Oh! I can cheat :D | 10:07 |
| noam | I have a USB-A to USB-C adapter, and a hub from the pinebook pro that has a MMC slot :D | 10:07 |
| noam | > ## Transferring control to Plan 9 (at address 40100000) ... | 10:07 |
| noam | :D | 10:07 |
| noam | ...except it then proceeds to reboot almost immediately? :/ | 10:08 |
| noam | my custom kernel is definitely loading; some of the silly prints I added are showing up | 10:08 |
| noam | Ah, hm. > Failed to load '/plan9.ini' | 10:09 |
| noam | That's from the uboot script | 10:09 |
| noam | > ** Reading file would overwrite reserved memory ** | 10:10 |
| minute | noam: can you change the address? | 10:13 |
| noam | minute: in the u-boot script? | 10:13 |
| minute | yeah | 10:13 |
| noam | It's on a SD card, so probaly? :P | 10:14 |
| minute | another option would be to boot linux and flash our uboot first... or can you do that from 9front? | 10:14 |
| noam | I can do it from 9front | 10:16 |
| noam | ...if I can boot into it lmao | 10:16 |
| noam | minute: I haven't worked with u-boot extensively before; how do I look at the memory map and pick a new address? | 10:16 |
| noam | I'll probably have to update the kernel too since I assume it hardcodes the location where the config file is | 10:16 |
| noam | yeah okay, if I have a better address to put it in I can change it :) | 10:17 |
| noam | Of note is that plan9 kernel says there's only 1G available, but u-boot says 4G of DRAM (which is my expectation) | 10:19 |
| noam | so it looks like the u-boot config may be weird | 10:19 |
| noam | Actually, think I found the info under board info :) | 10:23 |
| noam | looks like DRAM is memory[1] [0x40000000-0x13fffffff], 0x100000000 bytes flags: 0, and reserved[0][0x40000000-0x4001ffff], 0x00020000 bytes flags: 0, so I can probably just do 40020000, I think | 10:24 |
| noam | oh! but I think the kernel hardcodes the usable memory regions? | 10:28 |
| noam | which might be why it's rebooting, might be trapping somewhere | 10:28 |
| sigrid | no it's not | 10:32 |
| noam | sigrid: conf.mem[2].base = 0x1000000000ULL; ? | 10:35 |
| noam | Or, sorry, which "it" am I wrong about? :P | 10:36 |
| sigrid | the kernel isn't in reserved area | 10:39 |
| sigrid | and this "reserved" area is most likely just uboot itself deciding it is | 10:40 |
| sigrid | memset seemed to work fine, anyway | 10:41 |
| sigrid | you can change the boot script to load elsewhere and then cp.b to the address 9front expects plan9.ini to be at, it would probably work | 10:41 |
| minute | josch: testing a311d reform2 now with https://source.mnt.re/reform/reform-system-image/-/jobs/4455/artifacts/file/pocket-reform-system-a311d.img.gz | 10:42 |
| noam | sigrid: yes, but we hardcode where in the map the DRAM is, no? | 10:43 |
| noam | VDRAM? | 10:44 |
| noam | I'll try just bumping config area outside of u-boot's reserved area and see what happens :P | 10:44 |
| josch | minute: thank you for testing -- i had some really weird lock-ups with kernel 6.9 and hope that it's my setup and not the kernel... | 10:44 |
| josch | wait... that link is for the pocket reform and not the reform2 :D | 10:45 |
| minute | josch: oh lol | 10:45 |
| minute | i didn't have my second coffee yet, thanks! | 10:45 |
| sigrid | noam: what? | 10:46 |
| sigrid | i do not understand. those addresses are defined for imx8mq | 10:46 |
| noam | sigrid: I'm almost certainly misunderstanding basically everything :P | 10:46 |
| sigrid | idk why their uboot decided to reserve that area | 10:47 |
| sigrid | this area is just ram | 10:47 |
| noam | What the heck? > fpu.c: sysreg.h does not exist | 10:50 |
| noam | I swear I've run into this before | 10:50 |
| noam | I think my tree is corrupt, but `git/revert`ing back to front doesn't fix it, what the heck | 10:50 |
| sigrid | why not try latest 9front instead of custom kernel | 10:50 |
| noam | sigrid: because that won't be able to mount my file system | 10:51 |
| sigrid | i am not even sure if yours is working | 10:51 |
| sigrid | yes, but the kernel would either boot or not | 10:51 |
| noam | Ah, as a test you mean? | 10:51 |
| sigrid | yes | 10:51 |
| noam | yeah, I'll try that | 10:52 |
| noam | it's entirely possible that my image is borked | 10:52 |
| noam | I _was_ self-hosting | 10:52 |
| noam | but I tried building this image from the thinkpad | 10:52 |
| noam | and I hadn't synced it in a long while, I think | 10:52 |
| sigrid | if there is no plan9.ini it should still be printing stuff on uart | 10:56 |
| noam | sigrid: it is! | 10:56 |
| noam | It prints the "Plan 9" line, and then nothing for a minute, and then u-boot starts printing again | 10:56 |
| sigrid | did uboot ever complain about mw.b? | 10:57 |
| noam | I don't believe so | 10:58 |
| noam | sigrid: https://pixelhero.dev/tmp/9.png | 10:59 |
| sigrid | wait. do you actually *have* plan9.ini? | 10:59 |
| noam | yes | 11:00 |
| noam | This is a fresh official 9front image | 11:00 |
| noam | it tries loading plan9.ini into memory u-boot _does actaully_ have listed as reserved | 11:00 |
| noam | https://pixelhero.dev/tmp/bdinfo.png | 11:00 |
| sigrid | i don't think it matters tbh | 11:02 |
| noam | not sure why it's rebooting :/ | 11:02 |
| sigrid | you can try loading plan9.ini manually to another region, then cp.b | 11:03 |
| sigrid | then dump a few lines from that reserved region | 11:03 |
| sigrid | see if uboot reboots itself | 11:03 |
| noam | confinit() happens before the `Plan 9` print, no? I don't think it should be rebooting even without the config file? | 11:04 |
| noam | bootargsinit(), rather | 11:05 |
| sigrid | i do not understand much from your screenshot | 11:07 |
| sigrid | was it your kernel booting there or 9front kernel? | 11:07 |
| noam | 9front | 11:08 |
| sigrid | maybe config memory range is filled with garbage | 11:08 |
| noam | Could be, given that it's reserved >_> | 11:08 |
| sigrid | that's why i'm asking to try my suggestion | 11:08 |
| noam | Yeah, I'm working on building a kernel for that rn | 11:08 |
| sigrid | it is reserved by uboot | 11:08 |
| noam | Right, that's what I mean - u-boot is likely using it | 11:09 |
| sigrid | which does not necessarily mean soc cares about it | 11:09 |
| noam | but, the 9front kernel is parsing that w/o issue, printing "Plan 9," and _then_ rebooting | 11:09 |
| sigrid | ok. add more prints around main() | 11:12 |
| sigrid | see where it reboots exactly | 11:12 |
| noam | In trapinit(), apparently | 11:15 |
| noam | going to do the same in there for more detail... | 11:16 |
| noam | Guessing the watchdog timer is rebooting | 11:16 |
| sigrid | maybe vendor uboot decided to do some extra stuff | 11:18 |
| sigrid | if we'd have git log we could check that too | 11:18 |
| noam | intrcupshutdown() is hanging... | 11:20 |
| noam | I... uh. Hm. | 11:24 |
| noam | trapinit() calls intrcpushutdown(). I put a print statement at the very beginning of that function, and it never prints :/ | 11:25 |
| noam | ah! no, wrong intrcpushutdown. There's one in arm64/ and one in imx8? | 11:26 |
| sigrid | yes | 11:27 |
| noam | k, I need to instrument the imx8 one then >_)> | 11:27 |
| noam | Okay, so syswr(ICC_IGRPEN0_EL1, 0) is hanging | 11:29 |
| - colinsane (QUIT: Quit: bye) (~colinunin@97-113-136-129.tukw.qwest.net) | 11:29 | |
| noam | I wonder if this is a permissions thing? | 11:31 |
| sigrid | it would trap not hang | 11:31 |
| noam | Sure, but it's possible that's happening? | 11:33 |
| noam | If this is getting trapped by u-boot / there's no handler? | 11:33 |
| noam | Given that this is happening at the very beginning of trapinit() so we're not handling traps yet | 11:33 |
| + colinsane (~colinunin@97-113-136-129.tukw.qwest.net) | 11:34 | |
| sigrid | you should probably iprint btw | 11:35 |
| sigrid | not print | 11:35 |
| noam | Non-interrupting, or? | 11:35 |
| sigrid | it for sure won't involve anything but uart | 11:36 |
| noam | gotcha | 11:36 |
| noam | I'm going to have to learn a great deal about ARM to fix this aren't I :) | 11:39 |
| sigrid | i don't think so | 11:44 |
| sigrid | still you haven't tried manual plan9.ini thing | 11:47 |
| sigrid | so i am not sure about anything | 11:47 |
| noam | sigrid: I can just bump CONFADDR and change the address in boot.scr? | 11:49 |
| sigrid | sure | 11:50 |
| - mjw (QUIT: Ping timeout: 268 seconds) (~mjw@gnu.wildebeest.org) | 11:54 | |
| noam | sigrid: same result :/ | 11:56 |
| noam | well, the plan9.ini file gets loaded now :P | 11:56 |
| noam | but it still hangs | 11:56 |
| sigrid | https://developer.arm.com/documentation/ddi0601/2022-06/AArch64-Registers/ICC-IGRPEN0-EL1--Interrupt-Controller-Interrupt-Group-0-Enable-register | 11:59 |
| sigrid | if you're interested | 11:59 |
| noam | already got that open in anopther tab :P | 12:00 |
| noam | > If the highest priority pending interrupt for that PE is a Group 0 interrupt using 1 of N model, then the interrupt will be targeted to another PE as a result of the Enable bit changing from 1 to 0. | 12:01 |
| noam | I wonder if there's an interrupt enabled with vendor u-boot that we don't expect to be, and that's causing Something to occur? | 12:01 |
| noam | </conjecture> | 12:01 |
| noam | the other possibility that comes to mind is the EL being 'wrong' from our perspective, but I'd think that'd break _everything_ (including LInux for instance) | 12:02 |
| sigrid | i don't think uboot uses interrupts | 12:02 |
| noam | ...I should try a linux image, and see if that works | 12:03 |
| sigrid | maybe syswr needs a mask, as not to rewrite the entire register | 12:03 |
| sigrid | strange. seemed to work fine anywhere else | 12:03 |
| noam | Why would it work on the older SoMs if that's an issue? Rather, why would that behavior change? | 12:05 |
| minute | somewhere NXP has a pdf note with the changes from AA to AB silicon | 12:08 |
| minute | but i think it's probably the different uboot version | 12:08 |
| noam | Going to burn the official reform image to a USB and try that | 12:09 |
| * Guest4962 -> mjw | 12:12 | |
| noam | Heh, apparently the boot.scr for the official image doesn't work on the new module :/ | 12:16 |
| noam | the load command gives out usage info | 12:17 |
| josch | noam: that might be because your vendor u-boot cannot handle our boot.scr which is written for MNT u-boot | 12:18 |
| noam | josch: ah. That's probably it :/ | 12:18 |
| noam | grrrr. Is there a way to flash eMMC from within uboot? | 12:19 |
| sigrid | i don't think boot.scr matters | 12:19 |
| sigrid | 9front ships with its own basic (very simple) boot.scr | 12:19 |
| noam | sigrid: I was testing the debian image | 12:20 |
| sigrid | i understand | 12:20 |
| noam | If uboot has file system drivers, it's got to be possible to use it to flash itself on the eMMC. Surely it's copying itself into RAM on startup anyways... | 12:21 |
| sigrid | emmc flashing would probably work but if you're writing the same region that current uboot is at, what happens then? | 12:21 |
| sigrid | maybe just try to boot linux manually | 12:22 |
| noam | woooo go u-boot! :P | 12:23 |
| noam | > BUG at drivers/usb/host/xhci-ring.c:530/abort_td()! | 12:23 |
| josch | noam: or re-write the system image boot.scr to be able to work with yout u-boot | 12:24 |
| noam | oh. oops. in its defense I _did_ remove the drive lol | 12:24 |
| josch | noam: or if the vendor u-boot is newer than 2014, maybe it has support for extlinux.conf which i personally find a bit easier to edit than boot.scr | 12:24 |
| minute | noam: this is what i meant earlier | 12:25 |
| minute | you need to (re)define some variables | 12:26 |
| minute | of the 3 load adresses | 12:26 |
| minute | fdt_addr_r | 12:28 |
| minute | kernel_addr_r | 12:28 |
| minute | ramdisk_addr_r | 12:28 |
| minute | noam: these three must be defined | 12:28 |
| minute | the vendor uboot has different names, you can see these by doing "printenv" | 12:28 |
| minute | for example, loadaddr or fdt_addr with a _r | 12:28 |
| minute | maybe: | 12:29 |
| minute | setenv fdt_addr_r 0x43000000 | 12:30 |
| minute | setenv kernel_addr_r 40800000 | 12:30 |
| minute | setenv ramdisk_addr_r 42800000 | 12:30 |
| minute | noam: if you set these, the boot script should work | 12:30 |
| minute | and yes, extlinux.conf is also an alternative good idea | 12:31 |
| noam | > Booting Debian 6.8.11-mnt-reform-arm64 from usb 0:1... | 12:37 |
| noam | > ERROR: Did not find a cmdline Flattened Device Tree | 12:37 |
| noam | :/ | 12:37 |
| josch | did you load the device tree dtb to fdt_addr_r? | 12:38 |
| noam | the boot script should have, I think | 12:39 |
| noam | double checking... | 12:39 |
| josch | the MNT boot.scr does, yes -- but maybe the variables are wrong? | 12:39 |
| noam | It definitely read it, but I suppose the addresses might have been wrong? | 12:40 |
| noam | ...oh wait, 0x is missing from two of the ones minute pasted, I think? | 12:40 |
| josch | hah indeed :) | 12:41 |
| noam | same issue though :/ | 12:41 |
| - colinsane (QUIT: Quit: bye) (~colinunin@97-113-136-129.tukw.qwest.net) | 12:43 | |
| + colinsane (~colinunin@97-113-136-129.tukw.qwest.net) | 12:47 | |
| noam | proooobably can't use vendor u-boot to load u-boot from another device, can I? :P | 12:54 |
| sigrid | that would probably end up wasting a lot of your time and without any guarantee it'd work | 12:56 |
| noam | yeaaaaaah | 12:57 |
| sigrid | do a printenv and paste it somewhere | 12:58 |
| sigrid | preferably not as a screenshot | 12:58 |
| noam | right >_> | 12:59 |
| noam | http://45.76.26.151/env | 13:00 |
| sigrid | so it's without _r | 13:00 |
| sigrid | and seems to do syslinux | 13:01 |
| sigrid | s/sys/ext/ | 13:01 |
| sigrid | try inserting a usb drive with linux mnt reform image and do usb_boot | 13:02 |
| noam | sigrid: that's what I did? | 13:03 |
| sigrid | ok. what were the errors? | 13:04 |
| sigrid | I didn't see the output | 13:04 |
| noam | Well first it just flat out doesn't work because the _r variables are missing | 13:04 |
| sigrid | ah ok | 13:04 |
| noam | I set those manually per minute's instructions above and it gave a "Did not find a cmdline Flattened Device Tree" error | 13:04 |
| sigrid | setenv fdt_addr_r ${fdt_addr} | 13:05 |
| sigrid | ah no wait | 13:05 |
| sigrid | it does set _r vars | 13:05 |
| noam | sigrid: that env print is from after I set _r vars manually | 13:06 |
| noam | just updated that linked file with the result if I didn't do that >_> | 13:06 |
| noam | ...oh. huh. Ezurio (the company that bought Boundary) has an upgrade.scr file | 13:08 |
| sigrid | slightly related: uboot upgrade seems to be described https://www.ezurio.com/resources/software-announcements/how-to-upgrade-u-boot | 13:08 |
| noam | Heh, yeah, just found that :D | 13:08 |
| sigrid | not that it helps | 13:08 |
| noam | Oh? | 13:09 |
| sigrid | I mean if it's uboot issue and somehow an upgrade is worth it, sure. but we haven't found out if it has anything to do with uboot | 13:10 |
| noam | ahhh, true | 13:11 |
| noam | and worst case I risk bricking it if there's a bug in their script >_> | 13:11 |
| sigrid | probably not but don't take my word for it | 13:12 |
| sigrid | I am no linux/uboot specialist | 13:12 |
| minute | noam: maybe for fdt you also need fdt_addr | 13:14 |
| noam | minute: _and_ fdt_addr_r? /confusion | 13:14 |
| minute | noam: yeah | 13:14 |
| josch | me neither. i just read the u-boot sources at boot/image-fdt.c which is where the error message comes from and it seems like u-boot is unable to find a valid device tree at the point where it's looking | 13:14 |
| sigrid | fdt_addr=0x43000000 | 13:14 |
| sigrid | fdt_addr_r=0x43000000 | 13:14 |
| noam | minute: it's set by default? > "put this on a storage media | 13:14 |
| noam | er | 13:14 |
| sigrid | that uboot sets both | 13:14 |
| noam | sigrid: no, that's my bad | 13:14 |
| noam | it sets fdt_addr=0x43000000 only | 13:14 |
| noam | I set fdt_addr_r before running `env print` | 13:15 |
| minute | you can also just do the load completely manually | 13:15 |
| sigrid | :/ | 13:15 |
| noam | I pushed a clean env file up >_> | 13:15 |
| sigrid | ok, then: setenv fdt_addr_r ${fdt_addr} | 13:15 |
| noam | minute: I tried :( | 13:15 |
| sigrid | and do the same for the other two | 13:15 |
| sigrid | then usb_boot | 13:15 |
| noam | $kernel_addr is unset | 13:15 |
| noam | ditto for $ramdisk_addr | 13:15 |
| sigrid | ${loadaddr} | 13:15 |
| noam | Ah, okay | 13:16 |
| noam | What's ramdisk_addr_r based off of? | 13:16 |
| sigrid | set it to 42800000 | 13:17 |
| sigrid | it's fileaddr | 13:17 |
| noam | gotcha | 13:17 |
| sigrid | though fileaddr doesn't seem to have 0x | 13:17 |
| noam | yeah, that looks wrrong? | 13:17 |
| noam | going to reset to double check it didn't get messed up | 13:17 |
| sigrid | idr uboot requiring 0x | 13:18 |
| noam | ah, fileaddr doesn't exist with a clean env | 13:18 |
| sigrid | it won't get messed up unless you do saveenv or something | 13:18 |
| sigrid | yeah, that is probably set by another var that you ran | 13:18 |
| sigrid | so just set it manually then, with 0x | 13:18 |
| noam | ...yeah, fdt_addr also doesn't exist on a clean env | 13:18 |
| sigrid | can you paste the clean one? | 13:19 |
| noam | er yes it does, my brain is just dumb | 13:19 |
| noam | yeah sure | 13:19 |
| sigrid | it's hard to understand what's going on | 13:19 |
| noam | https://45.76.26.151/cleanenv | 13:19 |
| noam | `setenv fdt_addr_r ${fdt_addr}; setenv kernel_addr_r ${loadaddr}; setenv ramdisk_addr_r 42800000; boot` on a clean boot? | 13:20 |
| noam | That still yields > ERROR: Did not find a cmdline Flattened Device Tree | 13:21 |
| sigrid | usb_boot, not boot | 13:22 |
| noam | > Unknown command 'usb_boot' | 13:22 |
| noam | `run ${usb_boot}`? | 13:22 |
| josch | distro_bootcmd should also work as that is running usb_boot on the usb0 entry in $boot_targets | 13:23 |
| josch | noam: run usb_boot | 13:23 |
| sigrid | "run usb_boot" | 13:23 |
| noam | ah, oops, thanks | 13:23 |
| - S0rin (QUIT: Quit: WeeChat 3.8) (~S0rin@user/s0rin) | 13:24 | |
| noam | => setenv fdt_addr_r ${fdt_addr}; setenv kernel_addr_r ${loadaddr}; setenv ramdisk_addr_r 42800000; run usb_boot | 13:24 |
| noam | same result | 13:24 |
| sigrid | do you have bootpart set? | 13:25 |
| noam | yep | 13:25 |
| noam | bootpart=1 | 13:25 |
| noam | [after running that line] | 13:25 |
| noam | [didn't check beforehand, can do if it's helpful] | 13:26 |
| noam | it's not set beforehand | 13:26 |
| sigrid | try manually | 13:26 |
| noam | Setting bootpart? to 1, or? | 13:26 |
| sigrid | something along the lines of "load ${devtype} ${devnum}:${bootpart} ${loadaddr} /boot/vmlinux..." | 13:27 |
| noam | those load commands work fine | 13:27 |
| sigrid | ok, have you tried the same for dtb? | 13:27 |
| noam | it's the booti command that gives an errorr | 13:27 |
| noam | the load command for the dtb works fine | 13:27 |
| sigrid | ok, what booti command looks like? | 13:27 |
| noam | booti ${kernel_addr_r} ${ramdisk_addr_r}:${filesize} ${fdt_addr_r} | 13:27 |
| sigrid | can you run "help booti"? | 13:28 |
| - jaume (QUIT: Ping timeout: 268 seconds) (~user@user/jaume) | 13:28 | |
| sigrid | also paste the exact command you ran to load dtb | 13:28 |
| noam | https://45.76.26.151/booti | 13:29 |
| noam | the dtb was loaded by the boot.scr | 13:29 |
| noam | 45.76.26.151/scr | 13:29 |
| noam | https://45.76.26.151/scr * | 13:30 |
| noam | > load ${devtype} ${devnum}:${partition} ${fdt_addr_r} ${pathprefix}${fdtpath} | 13:30 |
| sigrid | can you show ${pathprefix}${fdtpath} | 13:30 |
| noam | /dtb-6.8.11-mnt-reform-arm64 | 13:30 |
| minute | Pocket Reform handbook is out: https://mntre.com/documentation/pocket-reform-handbook.pdf | 13:31 |
| sigrid | can you heck if that file exists? | 13:31 |
| noam | that file exists, yes | 13:31 |
| - colinsane (QUIT: Ping timeout: 268 seconds) (~colinunin@97-113-136-129.tukw.qwest.net) | 13:31 | |
| noam | > System.map-6.8.11-mnt-reform-arm64 boot.scr config-6.8.11-mnt-reform-arm64 dtb dtb-6.8.11-mnt-reform-arm64 dtbs flash.bin initrd.img-6.8.11-mnt-reform-arm64 lost+found vmlinuz-6.8.11-mnt-reform-arm64 | 13:31 |
| sigrid | and it's not a directory? | 13:31 |
| noam | I think it may be a symlink, but it's shown as loaded, so I assume that's fine | 13:31 |
| noam | lrwxrwxrwx 1 root root 61 May 18 00:03 dtb-6.8.11-mnt-reform-arm64 -> dtbs/6.8.11-mnt-reform-arm64/freescale/imx8mq-mnt-reform2.dtb | 13:32 |
| noam | The file it points to is 41593 bytes | 13:33 |
| noam | > 41593 bytes read in 6 ms (6.6 MiB/s) | 13:33 |
| noam | so it's definitely being read in right | 13:33 |
| sigrid | ok. can you do "fdt addr ${fdt_addr_r}" | 13:33 |
| sigrid | then "fdt print" | 13:33 |
| noam | uh | 13:34 |
| noam | => fdt addr ${fdt_addr_r} | 13:34 |
| noam | libfdt fdt_check_header(): FDT_ERR_BADMAGIC | 13:34 |
| sigrid | ok | 13:35 |
| sigrid | ext2load ${devtype} ${devnum}:${partition} ${fdt_addr_r} ${pathprefix}${fdtpath} | 13:35 |
| + S0rin (~S0rin@user/s0rin) | 13:35 | |
| noam | then the fdt command again? | 13:35 |
| sigrid | yes | 13:35 |
| noam | yeah okay now that works? | 13:36 |
| sigrid | does it print the tree? | 13:36 |
| noam | yeah | 13:36 |
| sigrid | ok. try booting | 13:36 |
| sigrid | booti blah blah | 13:36 |
| noam | ....gahhhhhhh I need to redo it, I ran usb_boot and that refucked it up for some reason?? | 13:37 |
| sigrid | ofc it did | 13:37 |
| sigrid | one problem at a time | 13:37 |
| josch | noam: whatever you do -- please write up what worked in the end and post it in the forum! :D | 13:37 |
| sigrid | josch: of note: load vs ext2load | 13:37 |
| josch | yup, this is very interesting | 13:37 |
| sigrid | 41593 bytes seems like legit file size to me | 13:38 |
| josch | it's funny that the plain "load" didn't show an error | 13:38 |
| sigrid | I thought maybe it was loading symlink contents instead of actual file | 13:38 |
| sigrid | but I would never expect that to be the case | 13:38 |
| noam | current command is `setenv fdt_addr_r ${fdt_addr}; setenv kernel_addr_r ${loadaddr}; setenv ramdisk_addr_r 42800000; devtype=usb; ext2load ${devtype} ${devnum}:${partition} ${fdt_addr_r} ${pathprefix}${fdtpath}", need to add in the rest of the load commands and booti | 13:38 |
| noam | Yeah I'm wondering what it actually loaded >_> | 13:38 |
| noam | probably just the offset relative to disk instead of fs or some such? | 13:38 |
| sigrid | you can do a memory dump to see | 13:39 |
| noam | oh! I should use ext2load for the kernel and initrd too | 13:39 |
| sigrid | that is probably a good idea | 13:39 |
| sigrid | josch: what fs does the image ship with? | 13:40 |
| josch | maybe now that we know that, you an just s/load/ext2load/ in the boot.scr? | 13:40 |
| josch | sigrid: ext4 | 13:40 |
| noam | oh! no that's not doing it | 13:40 |
| noam | because loading the other files broke it :D | 13:40 |
| noam | ERROR: Did not find a cmdline Flattened Device Tree | 13:40 |
| noam | and fdt addr ${fdt_addr_r} complains | 13:40 |
| sigrid | perhaps using loadaddr is wrong? | 13:40 |
| noam | the initrd is at the wrong addr I think | 13:41 |
| noam | because loading it overwrites the fdt | 13:41 |
| sigrid | heh :( | 13:41 |
| noam | initrd at 42800000 | 13:41 |
| noam | fdt at 43000000 | 13:41 |
| noam | ramdisk is ~32MiB | 13:41 |
| sigrid | that would do it | 13:42 |
| noam | the distance between initrd start and fdt start is 2MiB | 13:42 |
| noam | yea | 13:42 |
| noam | I can totally just pick a different address for the initrd though | 13:42 |
| sigrid | set fdt addr to like 44000000 | 13:42 |
| sigrid | actually that isn't enough | 13:42 |
| noam | I was going to bump ramdisk addr instead :P | 13:42 |
| noam | and just set it to fdt+1MiB | 13:42 |
| sigrid | I think usually how it's done is the boot script loads one file, then set the address for the next one by adding the size of the last loaded file to initial address? | 13:43 |
| noam | > Starting kernel ... | 13:43 |
| noam | :O | 13:43 |
| noam | screen turned on!!! | 13:43 |
| noam | I see linux! | 13:43 |
| sigrid | idr, last time I did anything considerable with uboot was in 2006, I think | 13:43 |
| noam | woot woot! | 13:43 |
| sigrid | s/2006/2009/ | 13:44 |
| noam | Thanks!! :D <3 | 13:44 |
| noam | it failed to launch sway, though? | 13:44 |
| noam | (which I didn't run, the default boot script did?) | 13:44 |
| noam | but that's fine | 13:44 |
| noam | console should be plenty ehre | 13:44 |
| noam | here* | 13:44 |
| minute | noam: good job! | 13:45 |
| minute | noam: the default boot launches the setup wizard in sway nowadays, but that aborts if you have no input devices or no display | 13:45 |
| minute | i think we have a command to flash uboot | 13:46 |
| josch | noam: could you write down what whanges ultimately worked for you? | 13:46 |
| noam | fairly sure this would work: setenv fdt_addr_r ${fdt_addr}; setenv kernel_addr_r ${loadaddr}; setenv ramdisk_addr_r 43200000; boot | 13:46 |
| noam | load vs ext2load was a red herring :P | 13:47 |
| noam | the ramdisk and fdt were overlapping | 13:47 |
| josch | ah that makes sense | 13:47 |
| noam | minute: the failure to launch sway occurs on the Reform prroper | 13:48 |
| noam | where display is working | 13:48 |
| noam | ... | 13:48 |
| noam | keyboard isn't though | 13:48 |
| noam | >_> | 13:48 |
| minute | noam: yeah it's because uboot didn't reset the usb hub | 13:49 |
| minute | noam: so you can only use the usb port that is next to hdmi | 13:49 |
| noam | linux can't do it?? | 13:49 |
| noam | uboot was able to reset the usb hub a few times | 13:49 |
| noam | I saw the keyboard light up | 13:49 |
| noam | but only on some resets, not others | 13:49 |
| minute | noam: idk. it's not our uboot | 13:49 |
| minute | noam: you can do it in linux too with gpioset, but i don't remember the gpio number off the top of my head | 13:50 |
| noam | gotcha. | 13:50 |
| noam | Good news is, now I can boot linux, I can probably patch the USB drive to work automatically | 13:50 |
| minute | probably io 1-10 or sth | 13:50 |
| noam | and then I can at least put the hardwarre back together, and stop using the serial port | 13:50 |
| noam | and should be able to get 9 working using linux as the starting point | 13:50 |
| noam | hopefully | 13:50 |
| mjw | plomlompom, minute, Would it be possible to get an update on ticket #3488 ? I don't mind if it is just "all under control, but we are concentrating on getting all the pocket reforms out first". But I am getting a little worried that there has been no update for a month now. | 13:50 |
| minute | noam: maybe update to our uboot? | 13:50 |
| noam | minute: that's the plan :) | 13:51 |
| minute | mjw: hmm weird, i saw plomlompom replying to looots of tickets, let me see | 13:51 |
| noam | Worst case nothing changes | 13:51 |
| noam | best case that solves it | 13:51 |
| mjw | minute, If it is my spam filter eating them I apologize. | 13:51 |
| josch | noam: to update u-boot there is a script called reform-flash-uboot | 13:52 |
| minute | mjw: no, indeed i see we owe you a response there, i apologize. your package is here. i will take care of it this week, going in my list now | 13:52 |
| mjw | minute, phew. Thanks. | 13:53 |
| josch | noam: it will attempt downloading the latest u-boot but if you don't have interwobbles, you can disable that with --offline | 13:53 |
| noam | josch: that'll use a build on the image, I assume? | 13:53 |
| josch | noam: with --offline it will use /boot/flash.bin | 13:53 |
| noam | cool | 13:53 |
| noam | I need to find a screw for my NVMe drive so I can put the bottom on, I'll be back in a few days ;P | 13:54 |
| josch | it should be: sudo reform-flash-uboot --offline emmc | 13:54 |
| josch | minute: wow, the pocket reform handbook design, style and artwork is outstanding -- what an amazing piece of work! | 14:02 |
| noam | what kind of screw is needed for NVMe drives? I might have to buy new ones before I can rerassemble this >_> | 14:04 |
| noam | fuck it, going to glue gun the drive into place for now | 14:04 |
| josch | noam: i think it's just an M2 screw | 14:05 |
| noam | well, it's plastic now :P | 14:06 |
| minute | josch: thank you very much! | 14:17 |
| AbortRetryFail | noam: I have a strip of kapton tape holding the NVMe drive in mine. :D | 14:18 |
| noam | eughhhh need to clean the old thermal paste off the heat sink :/ | 14:20 |
| noam | i'll need to pick up alcohol today i suppose | 14:21 |
| sigrid | noam: when you've got time, comment out syswr(ICC_IGRPEN0_EL1, 0); | 14:32 |
| sigrid | for me it made no difference | 14:33 |
| noam | ...oops | 14:39 |
| noam | i just exploded a thing of loctite blue | 14:39 |
| noam | ...in the kitchen | 14:40 |
| noam | ...on cookware. :( | 14:40 |
| sigrid | uboot does not seem to do anything with that register and it's supposed to be reset to 0 on boot anyway | 14:41 |
| sigrid | so I am not sure why this syswr is needed at all | 14:42 |
| noam | josch: confirmed! | 14:50 |
| + mark_ (~mjw@gnu.wildebeest.org) | 14:51 | |
| noam | I put the three setenv commands in boot.scr, removed the header, regenerated it using mkimage, and the USB now boots automatically ;D | 14:51 |
| noam | there's some EGL_BAD_PARAMETER messages so sway doesn't work though | 14:51 |
| noam | and libinput fails, because the USB hub didn't reset :/ | 14:52 |
| noam | Fuuuuudge might have to take the bottom off again. Dunno why I expected the keyboard to magically work | 14:53 |
| noam | sigrid: nonzero odds hte write to IGRPEN1_EL1 hangs next :P | 14:57 |
| sigrid | comment that out as well | 14:58 |
| sigrid | and coherence() | 14:58 |
| minute | noam: strange @ usb hub. our uboot should've taken care of that | 15:17 |
| minute | noam: ah you still have a custom uboot.scr? | 15:17 |
| minute | boot.scr i mean | 15:17 |
| noam | minute: didn't flash the newer uboot yet >_> | 15:27 |
| minute | noam: then that's why the keyboard doesn't work | 16:24 |
| + Ar|stote|is (~linx@149.210.28.24) | 16:50 | |
| - Ar|stote|is (QUIT: Ping timeout: 255 seconds) (~linx@149.210.28.24) | 16:59 | |
| - iank_ (QUIT: Quit: ZNC 1.8.2+deb2+b1 - https://znc.in) (~iank@fsf/staff/iank) | 17:04 | |
| + iank (~iank@fsf/staff/iank) | 17:06 | |
| + Ar|stote|is (~linx@149.210.25.250) | 17:08 | |
| - Ar|stote|is (QUIT: Ping timeout: 268 seconds) (~linx@149.210.25.250) | 17:13 | |
| + colinsane (~colinunin@97-113-90-227.tukw.qwest.net) | 17:15 | |
| + Ar|stote|is (~linx@149.210.25.143) | 17:17 | |
| minute | ok, now testing a311d + ls1028a | 17:42 |
| minute | josch: usb hub, keyboard, mouse works fine for me on latest a311d image. i don't get an image on the display though. might be because this is not the usual display | 17:44 |
| - Ar|stote|is (QUIT: Ping timeout: 264 seconds) (~linx@149.210.25.143) | 17:57 | |
| + Ar|stote|is (~linx@149.210.0.91) | 18:11 | |
| minute | josch: indeed, no usb devices found @ ls1028a | 18:12 |
| minute | [ 46.526696] xhci-hcd xhci-hcd.0.auto: Error while assigning device slot ID: Command Aborted | 18:12 |
| minute | [ 46.526707] xhci-hcd xhci-hcd.1.auto: Error while assigning device slot ID: Command Aborted | 18:12 |
| minute | [ 46.526722] xhci-hcd xhci-hcd.0.auto: Max number of devices this xHCI host supports is 127. | 18:12 |
| minute | [ 46.526725] xhci-hcd xhci-hcd.1.auto: Max number of devices this xHCI host supports is 127. | 18:12 |
| minute | https://mntre.com/reform-irc-logs/2023-09-22.log.html#t16:06:28 | 18:19 |
| minute | rmmod & modprobe dwc3 used to be a workaround, but it doesn't work anymore | 18:19 |
| * mjw -> Guest4402 | 18:19 | |
| - Guest4402 (QUIT: Killed (silver.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 18:19 | |
| * mark_ -> mjw | 18:19 | |
| + Guest4402 (~mjw@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 18:20 | |
| minute | weird, it works with loglevel=7 | 18:21 |
| - Ar|stote|is (QUIT: Ping timeout: 252 seconds) (~linx@149.210.0.91) | 18:23 | |
| minute | it might be a timing issue where the logging accidentally adds a required delay somewhere | 18:29 |
| minute | i have turned off the workaround in /sbin/reform-hw-setup and added loglevel=7 to the bootargs and now it works every time (ok now on 3 consecutive power cycles) | 18:30 |
| + Ar|stote|is (~linx@149.210.8.125) | 18:33 | |
| minute | indeed | 18:36 |
| minute | even setting the loglevel dynamically to 3 again makes it fail again | 18:36 |
| minute | and setting it to 7 on console makes it work... w t f | 18:37 |
| + nybble (~sprang@user/csprng) | 18:37 | |
| - Ar|stote|is (QUIT: Ping timeout: 272 seconds) (~linx@149.210.8.125) | 18:40 | |
| + gustav28 (~gustav@c-0c35524e.019-141-67626730.bbcust.telenor.se) | 18:43 | |
| violet | that's terribly cursed | 18:45 |
| minute | violet: yes | 18:46 |
| minute | it could also be that the prints influence the cache behavior | 18:46 |
| minute | as those nxp patches sounded like there are coherency issues | 18:46 |
| violet | oh noooooooo | 18:46 |
| violet | i could see it | 18:47 |
| minute | it feels like stuff i encountered when messing with bare metal stuff on zynq etc | 18:47 |
| minute | josch: i guess a short-term workaround would be to revert loglevel to 7 on ls1028a in the uboot or boot.scr (not sure where it's ultimately set) | 18:48 |
| + Ar|stote|is (~linx@149.210.8.134) | 18:49 | |
| - Ar|stote|is (QUIT: Read error: Connection reset by peer) (~linx@149.210.8.134) | 18:58 | |
| - chomwitt (QUIT: Ping timeout: 256 seconds) (~chomwitt@2a02:587:7a1d:df00:1ac0:4dff:fedb:a3f1) | 19:04 | |
| + robin (~robin@user/terpri) | 19:42 | |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-0c35524e.019-141-67626730.bbcust.telenor.se) | 19:44 | |
| + chomwitt (~chomwitt@2a02:587:7a1b:2500:1ac0:4dff:fedb:a3f1) | 19:48 | |
| josch | minute: does a later loglevel setting in cmdline overwrite the former? Can you try editing /usr/share/flash-kernel/ubootenv.d/00reform2_ubootenv and add your loglevel setting at the end of this line: https://source.mnt.re/reform/reform-tools/-/blob/main/flash-kernel/ubootenv.d/00reform2_ubootenv?ref_type=heads#L22 | 19:59 |
| josch | and then run flash-kernel to re-generate your boot.scr | 19:59 |
| josch | if you want to make this solution more permanent, then you can do so by editing the default u-boot ${bootargs} here which (if later loglevel settings indeed overwrite former ones) will have the desired effect: https://source.mnt.re/reform/reform-ls1028a-uboot/-/blob/main/u-boot-config?ref_type=heads#L405 | 20:01 |
| + gustav28 (~gustav@c-0c35524e.019-141-67626730.bbcust.telenor.se) | 20:02 | |
| - cobra (QUIT: Ping timeout: 268 seconds) (~cobra@user/Cobra) | 21:05 | |
| + cobra (~cobra@user/Cobra) | 21:09 | |
| - robin (QUIT: Read error: Connection reset by peer) (~robin@user/terpri) | 21:26 | |
| + robin (~robin@user/terpri) | 21:27 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 21:30 | |
| minute | josch: ok i can try that tomorrow | 21:52 |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-0c35524e.019-141-67626730.bbcust.telenor.se) | 22:15 | |
| josch | i think i also like the solution of adding `dmesg -n 7` to /usr/sbin/reform-hw-setup | 23:00 |
| josch | then it does not mess with the bootargs and a comment next to it can elaborately explain what is going on | 23:00 |
| minute | josch: yeah, and can be removed once root caused | 23:55 |
| josch | i'll add that to https://source.mnt.re/reform/reform-tools/-/merge_requests/69 | 23:56 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
