| nightishaman | yep its stopping before the kernel modules | 00:05 |
|---|---|---|
| nightishaman | I did one after loadkernelmodules | 00:05 |
| mal | nightishaman: so reboot before load modules works but not after it? | 00:16 |
| nightishaman | i gotta try | 00:19 |
| nightishaman | it instantly reboots | 00:28 |
| mal | ok, so what is happening in module loading then | 00:31 |
| mal | why does it get stuck there | 00:31 |
| nightishaman | we could get funny creative, check if /dev/fb0 works, draw 50 different images and push a different based on how many modules loaded | 00:34 |
| mal | heh | 00:36 |
| mal | I don't think framebuffer usually works in android kernel | 00:38 |
| nightishaman | alternatively, does this device have ramps? | 00:38 |
| nightishaman | ramoops? | 00:38 |
| mal | usually pstore can be read | 00:40 |
| nightishaman | im booting into recovery and will check it out | 00:40 |
| mal | yeah, causing a kernel panic and then reading pstore in recovery it should contain kernel logs | 00:41 |
| nightishaman | when I tried to port the Sony Xperia 5 IV, I had used that to find a bug in a driver | 00:41 |
| nightishaman | to ubuntu touch | 00:41 |
| nightishaman | pstore might not be enabled | 00:42 |
| mal | but debugging with that is a bit difficult also since we can't know which module loading is causing the issue | 00:42 |
| nightishaman | what do you suggest? | 00:44 |
| mal | not sure really, other than very slow test of incrementally loading modules one by one and seeing when it hangs | 00:45 |
| nightishaman | oof | 00:45 |
| mal | so first loading the first module in modules.load list, then two and so on | 00:45 |
| nightishaman | better idea, binary searching | 00:45 |
| mal | sure | 00:45 |
| nightishaman | cut half the modules and try again | 00:45 |
| mal | faster that way probably | 00:46 |
| nightishaman | i just gotta figure out where the modules.load is defined | 00:46 |
| mal | you don't need to know that | 00:47 |
| mal | just edit the init-script load function to have a counter and set the limit number in each test | 00:47 |
| mal | like load 20 first modules and break the loop once counter is 20 | 00:48 |
| nightishaman | my bash skills arent that good, or lets say almost 0 | 00:49 |
| mal | btw how long does that make hybris-init_boot take for you? | 00:49 |
| nightishaman | approximately 20 seconds | 00:49 |
| mal | ok, then that is not a problem | 00:49 |
| nightishaman | the most part is android soong before starting ninja | 00:50 |
| mal | I just thought if the whole make is slow then there are other ways to edit the init_boot image | 00:50 |
| nightishaman | i got a ryzen 5800x3d and its on an nvme | 00:51 |
| nightishaman | running fedora linux | 00:52 |
| nightishaman | i got this now: https://pastebin.com/Za2hryk2 | 01:00 |
| nightishaman | the module seems to be in the first half | 01:05 |
| nightishaman | first quarter | 01:09 |
| nightishaman | the module is in the 2nd eigth | 01:13 |
| nightishaman | above 25, less than 51 | 01:13 |
| nightishaman | above 25, less than 37 | 01:15 |
| nightishaman | above 25, less than 31 | 01:19 |
| Solrac[m]1 | This is slightly off topic, but I wanted to ask, [@_oftc_mal:matrix.org](https://matrix.to/#/@_oftc_mal:matrix.org) do you have fp4 Hailium source? Do you think I can use it to test droidian? | 01:19 |
| nightishaman | Solrac[m]1 ask in the ubuntu touch telegram group, halium is their system | 01:21 |
| nightishaman | above 25, less than 28 | 01:21 |
| nightishaman | im gonna do -1 steps now | 01:21 |
| Solrac[m]1 | nightishaman: Their Hailium is based on a much older Hailium, for some, some networks do not work proper. I tired the official a11 and then flashed sfos, and had no connectivity,any version older than ofw was, didn't have this issue. I've yet to try e/OS for Hailium instead | 01:23 |
| nightishaman | sailfishos uses hybris | 01:24 |
| mal | Solrac[m]1: is a11 the latest official for fp4? | 01:25 |
| mal | Solrac[m]1: I didn't quite understand what the problem is | 01:25 |
| nightishaman | i think they want to use the boot system of halium for other os or somehow port it | 01:26 |
| nightishaman | mal: its module at index 26, max 26 it hangs, max 25 it doesnt | 01:26 |
| nightishaman | 0 indexed | 01:26 |
| mal | nightishaman: what is number 26? | 01:26 |
| nightishaman | mal: extra/private/google-modules/soc/gs/drivers/soc/google/debug/exynos-coresight-etm.ko | 01:29 |
| mal | nightishaman: ok, I suspected that when you said above 25, less than 28 | 01:29 |
| mal | so as a test add a hacky condition to the loop to skip that | 01:30 |
| Solrac[m]1 | nightishaman: That's pretty much it. Basically, I can't use Ubuntu Touch Helium because it requires a much older version of Android and 11 specifically, and the latest build had some issues with some carriers on mine included. So, I figured, instead of using that old version, maybe I could use the Helium version that's used for sfos and try porting Droidian. | 01:32 |
| Solrac[m]1 | Since Mobian and Postmarket OS are using a kernel that is not feature complete yet. | 01:32 |
| mal | Solrac[m]1: sailfish on fp4 doesn't use halium | 01:33 |
| nightishaman | while ubuntu touch uses the same base, their init system is different to Sailfish OS, you can't use that | 01:33 |
| mal | nightishaman: some sailfish ports use halium but not many | 01:33 |
| Solrac[m]1 | mal: Wait what, really? | 01:34 |
| mal | Solrac[m]1: I use the sailfish way of porting meaning building patched parts of android and so on | 01:34 |
| Solrac[m]1 | nightishaman: The problem with helium is that it's very dependent on the version of Android that you're using, and in my case, Android 11 gives sobre carrier issues | 01:34 |
| mal | Solrac[m]1: which android version is ok for you? | 01:35 |
| Solrac[m]1 | mal: Part in my ignorance since I haven't tried porting this OS to a device yet. Is this similar to what pmOS is doing? as in apply the relevant patches to a seemingly mainline kernel? | 01:36 |
| mal | Solrac[m]1: no, we use android hal for some things like graphics and other hw features | 01:36 |
| mal | Solrac[m]1: so you had issues when you tried sailfish os? you know before you install sailfish you can have android 14 or maybe even newer one on the device | 01:37 |
| mal | the fp4 sailfish port is actually using android 14 as base for the hardware parts | 01:39 |
| nightishaman | mal: I'll look into enabling pstore, because for some reason, it doesnt reboot at 25 anymore | 01:44 |
| Solrac[m]1 | mal: Do you might remember that I had some network issues, but that was when I flashed the first time. When I flashed it the next time I had no network issues at all. That was because I was using the wrong Android version the first time | 02:01 |
| Solrac[m]1 | s/Do/You/, s/you// | 02:02 |
| mal | yeah, to that second time you probably had some newer android on the device before sailfish | 02:02 |
| Solrac[m]1 | mal: Also, do part in my ignorance. I have only tried porting via Hailium once before. My understanding is that haleum uses whatever system and then bridges it to whatever kernel to boot. But how does hal work? | 02:03 |
| mal | sailfish on fp4 doesn't care that much about the android version since it only use some firmware etc from the android that was there before | 02:03 |
| Solrac[m]1 | mal: I had used official Android 11 and the second time I used eOS with Android 14. | 02:03 |
| Solrac[m]1 | mal: Ahh | 02:04 |
| nightishaman2 | mal: ok bad thing, my znc host is down | 02:04 |
| nightishaman2 | pstore is enabled and unfortunately emtpy | 02:04 |
| mal | Solrac[m]1: have you tried some halium installation but without reflashing everything from a11 for example, only flashing super partition content (system, vendor etc), I assume it probably needs those but the other parts probably can be from newer android version | 02:06 |
| nightishaman | mal: do you know which syntax the sh interpreter that is running the init script supports? | 02:09 |
| nightishaman | because maybe my counter is bugged or not interpreted correctly, although i find it weird that it rebooted one time | 02:09 |
| mal | not fully sure but maybe avoid bash stuff | 02:13 |
| nightishaman | mal: ok it fails even loading the first module | 02:35 |
| nightishaman | which is zsmalloc | 02:35 |
| Solrac[m]1 | mal: I used boot.img only, but the reason why I asked was to try porting Mobian, if it needs a custom setup, I imagine its better off starting from scratch | 02:37 |
| mal | nightishaman: very odd | 02:41 |
| nightishaman | should i try disabling module signature check? | 02:43 |
| mal | maybe | 02:44 |
| mal | but how would that cause modprobe to hang | 02:44 |
| mal | I need some sleep again | 02:45 |
| nightishaman | im gonna go soon too | 02:45 |
| nightishaman | claude says modules without signature cant access modules with signature | 02:45 |
| nightishaman | CONFIG_MODULE_SIG=y | 02:46 |
| nightishaman | CONFIG_MODULE_SIG_PROTECT=y | 02:46 |
| nightishaman | mal: im gonna test this hypothesis tomorrow, but it might be that when i flashed the kernel with additional flags, i didnt flash vendor_kernel_boot and the kernel is rejecting the modules. no idea why it hangs then tho | 03:43 |
| *** Nokius7 is now known as Nokius | 03:55 | |
| T42 | <elros34> you should try with clean kernel without any sfos related changes | 10:15 |
| T42 | <elros34> also same question did you check check_kernel_config function? Did you format /data and you are 100% sure there is no encyption there? | 10:27 |
| rinigus | mal and b100dian : I wrote down eSIM case as an issue: https://forum.sailfishos.org/t/enabling-esim-modem-prevents-lipstick-login/29652 . Sorry for long text, but I hope it has sufficient details to trace the origin. | 15:22 |
| rinigus | it seems to me that lipstick and PinQueryAgent interaction together with minimal ofono dbus properties for eSIM leads to such case | 15:23 |
| rinigus | I've included dbus data as well as results of tracing this bug in lipstick. unfortunately, I've hit closed-source wall and need someone behind it to check it out :) | 15:23 |
| mal | rinigus: thanks, I'll check that | 15:48 |
| rinigus | mal: thanks! | 15:48 |
| T42 | <b100dian> rinigus: wow! thanks for the thoroughness:) | 16:02 |
| mal | rinigus: in settings it shows the esim slot as if there is sim present all the time | 16:13 |
| rinigus | mal: is this a question or statement? in my case, I have to disable under multisim settings the second sim. then I can boot | 16:18 |
| mal | statement | 16:20 |
| mal | rinigus: that is observable when UI is up and then enabling the second slot | 16:21 |
| rinigus | mal: yes, if I enable in settings, SIM card 2 it is shown in as just enabled under Multisim, I can choose it for calls and messages. but not for mobile data | 16:24 |
| mal | rinigus: you have esim active on the device? | 16:25 |
| mal | I mean a functioning esim profile installed | 16:26 |
| rinigus | no, I don't have profile installed. testing without any profile as it is the first stage during install | 16:26 |
| rinigus | (have to leave for 20-30 mins, sorry) | 16:27 |
| mal | yeah, same here, no esim profile | 16:29 |
| T42 | <b100dian> I have an eSIM profile (or more) and that's why I didn't notice this bug (on the sister device) | 16:32 |
| mal | @b100dian so if there active esim profile then device boots fine? | 16:37 |
| T42 | <b100dian> Yes, like it would be a 2nd phisical sim | 16:37 |
| mal | ok, good to know | 16:38 |
| rinigus | would be good to get over this bump. one option would be to automatically disable sim2 on boot if it does not have a profile. I guess there is a way we can check for that while booting. but its more of a workaround than proper fix | 17:16 |
| mal | rinigus: I think I now what specific condition in PinQueryAgent is the problem | 17:19 |
| mal | not sure yet what is the best way to fix it | 17:20 |
| mal | I wonder how android shows esim slot where are no profiles | 17:21 |
| nightishaman | @elros34 data is formatted as ext4, i did that manually in recovery and pushed the sailfishos rootfs there, also we already tested and it hangs at the load_kernel_modules function, a reboot before the function happens, a reboot after it doesnt | 17:36 |
| nightishaman | i could flash the lineageos kernel but that would yield no results since it doesnt have rndis enabled | 17:38 |
| rinigus | mal: no access for me for such device. hmm, maybe there is some demo online. | 17:38 |
| nightishaman | rinigus what device do you need | 17:39 |
| rinigus | nightishaman: see question by mal above (I wonder how android shows esim slot where are no profiles) | 17:40 |
| nightishaman | ur working on the sony nagara port, right? | 17:40 |
| rinigus | yes. not working much - mainly using it :) | 17:41 |
| nightishaman | i have the sony xperia 5 iv lying around here because im currently on porting the pixel 7 to sfos, i used it to try and port ubuntu touch to it, but that one got stuck wlan firmware issues, so if you need anything, maybe i can flash lineageos back on | 17:43 |
| rinigus | nightishaman: good to know. we use LOS21 as a base on that. it hasn't been released via lineage, just available as a download on our SFOS repo. | 17:46 |
| rinigus | good luck with pixel 7 | 17:46 |
| nightishaman | i know, i tried flashing it but got stuck there too | 17:48 |
| nightishaman | sfos i mean, the port u did | 17:49 |
| mal | rinigus: as you suspected it's the Present value sim sim manager that is the problem, in esim it's false even if sim is claimed to be present | 17:51 |
| rinigus | mal: great! now we will have to wait for a fix... | 17:52 |
| mal | I'm flashing android to one device to see how it shows there | 17:52 |
| rinigus | ... meanwhile, we can test esim profiles and other aspects. | 17:53 |
| mal | does it show the second slot or not | 17:53 |
| rinigus | mal: one example - https://esim-on.com/installing-esim-on-android-devices-2/ | 17:57 |
| rinigus | but there are many android variants - some UI differences could be there | 17:57 |
| mal | rinigus: yeah and those can be from device with two physical slots and esim support | 17:58 |
| mal | but we'll see soon once the device I have is flashed | 17:58 |
| nightishaman | how good is the esim support in sfos, because i only use esims here in germany | 17:59 |
| rinigus | nightishaman: esim - baby steps right now. b100dian has been testing; see https://forum.sailfishos.org/t/esim-provider-and-hardware-support/29520 | 18:03 |
| rinigus | b100dian: do we have lpac at OBS? | 18:03 |
| rinigus | ... and SimPro? | 18:04 |
| *** Nokius5 is now known as Nokius | 18:04 | |
| nightishaman | im gonna try and flash both the kernel and kernel modules from the same build | 18:04 |
| mal | rinigus: sony has a bit different sim UI, without sim it directly asks if I want to add some esim | 18:09 |
| mal | in about phone page in android it does show two slots | 18:10 |
| nightishaman | mal: we got a reboot, which means the first kernel module loaded successfully | 18:23 |
| rinigus | mal: showing slots makes sense, even if one is empty. just status should probably reflect it, similar to empty sim slot in sfos | 18:25 |
| nightishaman | all kernel modules loaded successfully | 18:26 |
| mal | nightishaman: using which kernel build? | 18:33 |
| mal | hybris patched or lineage? | 18:33 |
| nightishaman | sfos kernel the same as yesterday just vendor_kernel_boot from the same build | 18:33 |
| nightishaman | we just got another reboot, indicating it got past the dbg stuff | 18:34 |
| mal | I did mention to you to flash always both kernel and that | 18:34 |
| mal | ok | 18:34 |
| nightishaman | so probably if i remove that we go into rootfs | 18:34 |
| mal | let's see if we can get usb up | 18:34 |
| mal | rinigus: which binder interface do you use on your device 1.5 or aidl? | 18:35 |
| mal | rinigus: or something older | 18:35 |
| nightishaman | should i force usb telnet debug or try and look if we can get to the boot screen? | 18:35 |
| mal | nightishaman: at which point is it now? | 18:35 |
| nightishaman | before kernel hotplug | 18:35 |
| rinigus | mal: for which service? for volte it was aidl | 18:36 |
| nightishaman | im gonna go into recovery and check out the data partition, it should've copied dmesg there | 18:36 |
| mal | rinigus: I mean ofono-binder-plugin | 18:36 |
| rinigus | radioInterface = 1.5 | 18:37 |
| mal | ok | 18:39 |
| mal | found a possible bug in 1.5, not sure if it affect much though | 18:39 |
| mal | on fairphone it doesn't seem to cause anything but theoretically could on others | 18:41 |
| nightishaman | ok without the reboots, we got the 60s again, but could already be in rootfs, im gonna force in initrd first | 18:45 |
| nightishaman | mal: we're stuck at usb not working in initrd again, although possibly it might have gotten into rootfs, i gotta check that | 18:53 |
| nightishaman | the only weird thing is that my cp commands are not copying onto the data partition | 18:59 |
| nightishaman | ok no, it is only stuck in initrd | 19:02 |
| mal | why does it end up trying to setup usb there, one of the conditions fail then | 19:07 |
| nightishaman | i wish i could do a beep or vibration depending on the state | 19:09 |
| mal | does the device have indicator led? | 19:11 |
| nightishaman | unfortunately not | 19:12 |
| T42 | <Mister_Magister> mal perhaps if vibrator is native you can use that as led? | 19:13 |
| T42 | <Mister_Magister> tho mind that i didn't read the context xd | 19:14 |
| T42 | <Mister_Magister> but if its debugging no display then mayhaps | 19:14 |
| nightishaman | im just really confused, i just checked mount_stowaways and added a reboot on failure, it didnt reboot immediately, so i guess mounting must have succeeded, but nonetheless no dmesg.log, no init.log or anything on the data partition | 19:16 |
| nightishaman | https://pastebin.com/A7PLXTPX | 19:40 |
| nightishaman | the startup file from google | 19:40 |
| nightishaman | mal: can it be that rndis is on another directory like here in the init.gs201.usb.rc | 19:56 |
| mal | not sure what you mean, we don't use those in init | 20:07 |
| nightishaman | well it writes to rndis.gs4 instead of rndis.usb0, so maybe we need to write there | 20:12 |
| mal | we can't use gsi in init afaik | 20:13 |
| nightishaman | i dont know what to do rn | 20:15 |
| mal | still odd that you can't get logs to userdata | 20:15 |
| nightishaman | i tried adding a reboot to the fail section of the mount, but it doesnt fail | 20:16 |
| nightishaman | I wonder whether I could write to pstore | 20:17 |
| mal | just to be sure add sync after writing, in case it doesn't sync the storage | 20:17 |
| nightishaman | still nothing | 20:26 |
| nightishaman | potential reason why could be | 20:32 |
| nightishaman | the init script resolves the data partition to /dev/sda31. At least in recovery mode, it is in /dev/block/sda31 | 20:33 |
| mal | but you said mounting seemed to work? can you show the init-script you use now | 20:45 |
| nightishaman | https://pastebin.com/0cc8zUiX | 20:46 |
| nightishaman | last change was hardcoding the partition | 20:46 |
| nightishaman | one thing i could try is writing [device] to /sys/class/typec/port0/data_role | 20:47 |
| T42 | <adampigg> Nightishaman: ensure your distro also doesnt block rndis on the host. Suse does. | 20:51 |
| nightishaman | i had rndis with the sony xperia 5 iv | 20:52 |
| nightishaman | so it doesnt | 20:52 |
| mal | nightishaman: do you have rootfs in place at the moment? | 20:57 |
| mal | nightishaman: can you show the .config from out/ that is used my kernel checker | 20:58 |
| mal | *used by | 20:58 |
| nightishaman | https://pastebin.com/4HirD6yW | 21:01 |
| nightishaman | yes rootfs is in place | 21:04 |
| mal | I meant the whole .config | 21:10 |
| nightishaman | oh, yeah wait | 21:13 |
| nightishaman | https://pastebin.com/CXpMCsZD | 21:15 |
| mal | nightishaman: did you compare the module.load and module.load.recovery lists I assume you can see the latter in android recovery? | 21:25 |
| mal | starting to think if the modules.load you have could be missing something needed | 21:25 |
| nightishaman | https://pastebin.com/mfuyhC6b | 21:28 |
| nightishaman | this is the one i use | 21:28 |
| mal | check the one in android recovery but in there modules.load.recovery if there is one | 21:29 |
| nightishaman | it also only has a modules.load | 21:30 |
| nightishaman | https://pastebin.com/uHpyeUZr | 21:30 |
| mal | ok, no relevant differences it seems | 21:34 |
| nightishaman | i could try reading the result of modprobe and reboot if it is non-zero | 21:41 |
| nightishaman | maybe modprobe doesnt like the relative file names in modules.load | 21:42 |
| mal | nightishaman: but you see that modprobe should be getting the basename | 21:48 |
| nightishaman | yeah but zsmalloc for example is at /lib/modules/6.1.145-android14-11-g7861b6103597/kernel/mm$ ls | 21:53 |
| nightishaman | zsmalloc.ko | 21:53 |
| nightishaman | does modprobe know where to look? | 21:53 |
| mal | good question | 21:53 |
| nightishaman | because if the modules just get loaded with module not found, the usb driver isnt setup, the ufs driver isnt too and we are getting nothing | 21:54 |
| mal | are all the modules in subfolders? | 21:55 |
| mal | seems they | 21:55 |
| mal | seems they are | 21:55 |
| mal | nightishaman: terribly ugly hack if it even works, using find command copy all .ko files from subfolders to the /lib/modules/6.1.145-android14-11-g7861b6103597/ or symlink them, which ever is easier for you to implement | 21:55 |
| mal | you can test the script function first locally to see that it works | 21:56 |
| mal | actually no need for find even | 21:57 |
| mal | since you have paths in modules.load already | 21:57 |
| nightishaman | we could give modprobe the full path | 21:57 |
| nightishaman | i just realized something | 21:57 |
| mal | modprobe doesn't like full paths | 21:58 |
| nightishaman | that im really stupid | 21:58 |
| Mister_Magister | i don't think modprobe can even use full path | 21:58 |
| Mister_Magister | insmod can | 21:58 |
| mal | but insmod doesn't handle dependencies | 21:58 |
| mal | nightishaman: in case you want to try symlink do something like ln -s "$1" "/lib/modules/$(uname -r)/"$(basename "$1") | 22:00 |
| mal | not sure if that works but test locally first | 22:00 |
| nightishaman | wait a minute, i fucked something up | 22:01 |
| mal | what? | 22:02 |
| nightishaman | i may have forgotten to change the path for the cfg file that i hardcoded after i flashed a new kernel this morning | 22:02 |
| mal | ok | 22:02 |
| mal | you should not hardcode the path | 22:02 |
| mal | since we have to assume the uname is correct | 22:02 |
| mal | if it's not then there is a mismatch of flashed partitions which could cause other issues | 22:03 |
| nightishaman | yeah im just confused because in recovery from lineageos it has an additional string panther:/ # uname -r | 22:05 |
| nightishaman | 6.1.145-android14-11-gec45f20f38ea-ab15260282 | 22:05 |
| nightishaman | where as mine only has Linux version 6.1.145-android14-11-g7861b6103597 | 22:06 |
| mal | well you should see the value your kernel has with the strings grep you tested before | 22:07 |
| mal | in which case it seemed ok | 22:08 |
| mal | so test with uname in the script and see what happens | 22:08 |
| nightishaman | so i checked that boot.img in out is the same as in device/google/pantah-kernels/6.1/boot.img | 22:10 |
| nightishaman | i flashed boot.img, then flashed vendor_kernel_boot.img from device/google/pantah-kernels/6.1 | 22:10 |
| nightishaman | then i corrected the cfg file path in init-script, build it and flashed init_boot | 22:10 |
| nightishaman | with the string i got from strings vmlinux | 22:10 |
| nightishaman | now the device is hanging again | 22:11 |
| mal | hmm | 22:11 |
| nightishaman | ill replace it with uname -r and see what happens | 22:12 |
| nightishaman | i feel very stupid tho that i didnt caught that | 22:14 |
| mal | so the progress wasn't really happening it seems | 22:15 |
| nightishaman | i might just remove all the debug kernel modules from it, i dont really use coresight since im not a kernel dev at google with the equipment | 22:15 |
| mal | but you could try the copy or symlink way to see if that helps | 22:15 |
| mal | but you said even first module hangs? | 22:15 |
| mal | maybe test the copying or symlinking of .ko files first | 22:16 |
| nightishaman | i have to check it again now that everything matches | 22:16 |
| nightishaman | https://wiki.postmarketos.org/wiki/Serial_debugging/Cable_schematics#Google_Pixel_.22Android_Debug_Cable.22 apparently u can enable uart, i wonder how that communicates | 22:53 |
| mal | nightishaman: try first the symlink idea in case module loading still fails | 22:54 |
| nightishaman | like this? ln -s "/lib/modules/$basename" "/lib/modules/$(uname -r)/$1" | 23:02 |
| nightishaman | basename=$(basename "$1" .ko) | 23:02 |
| mal | isn't that the wrong way | 23:06 |
| mal | something like this, but check the syntax that it works first: ln -s "$1" "/lib/modules/$(uname -r)/"$(basename "$1") | 23:07 |
| nightishaman | ive had problems in the past that ln required full paths | 23:13 |
| nightishaman | ah but the first argument is the target | 23:15 |
| nightishaman | i got this now: ln -s "/lib/modules/$(uname -r)/$1" "/lib/modules/$(uname -r)/"$(basename "$1") | 23:17 |
| mal | full path in both works also | 23:19 |
| mal | my code used relative path but doesn't matter | 23:19 |
| nightishaman | ok lets try it, if it doesnt work, i might have to buy a cable to read sbu1 and sbu2 the uart console | 23:23 |
| nightishaman | great success | 23:24 |
| nightishaman | it reboots after the first module, indicating it did load | 23:24 |
| nightishaman | ill let it load all now | 23:24 |
| nightishaman | ok it doesnt do anything when i let all load | 23:30 |
| nightishaman | lets try 25 | 23:30 |
| nightishaman | 25 loads | 23:31 |
| nightishaman | weird, the second time it rebooted with 25, it did reboot only after about a minute | 23:35 |
| nightishaman | im really confused, one reboot later it did again reboot after 6s, then again a minute | 23:38 |
| nightishaman | in both cases it reboots into the bootloader with the reason: BL1 requested | 23:38 |
| mal | I wonder if the kernel is crashing | 23:40 |
| nightishaman | if /sys/fs/pstore wasnt empty in recovery, we could check that | 23:41 |
| nightishaman | but the behavior is weird | 23:41 |
| nightishaman | claude suggests disabling this: CONFIG_PANIC_ON_OOPS=y | 23:43 |
| nightishaman | CONFIG_PANIC_TIMEOUT=-1 | 23:43 |
| nightishaman | https://pastebin.com/MGJiDFV7 | 23:48 |
| nightishaman | we got a problem | 23:48 |
| nightishaman | fastboot oem dmesg yielded this from the bootloader | 23:48 |
| nightishaman | with max=12 it doesnt hang and reboots normally, no bl1 requested | 23:56 |
| nightishaman | between 12 and 25 is a kernel module which hangs the entire system | 23:56 |
| nightishaman | the debug modules are the cause | 23:59 |
| nightishaman | max=17 reboots fine | 23:59 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!