| T42 | <HengYeDev> @elros34 Recreating the boot.img with hybris initramfs and kernel worked, now it stays on the logo screen for a minute or so before rebooting, which i'm guessing is the ofono issue? | 00:29 |
|---|---|---|
| T42 | <HengYeDev> However, there's nothing in dmesg or lsusb iserial about the mer debug telnet | 00:29 |
| T42 | <elros34> great, 1 minute and reboot is probably lack of rndis | 00:31 |
| T42 | <HengYeDev> @elros34 grep RNDIS in kernel config: | 00:37 |
| T42 | <HengYeDev> # CONFIG_USB_NET_RNDIS_HOST is not set | 00:37 |
| T42 | <HengYeDev> # CONFIG_USB_NET_RNDIS_WLAN is not set | 00:37 |
| T42 | <HengYeDev> # CONFIG_USB_RNDIS_MULTIPACKET_WITH_TIMER is not set | 00:37 |
| T42 | <HengYeDev> # CONFIG_USB_ANDROID_RNDIS_DWORD_ALIGNED is not set | 00:37 |
| T42 | <HengYeDev> CONFIG_USB_F_RNDIS=y | 00:37 |
| T42 | <HengYeDev> CONFIG_USB_CONFIGFS_RNDIS=y | 00:37 |
| T42 | <HengYeDev> CONFIG_USB_RNDIS_MULTIPACKET=y | 00:37 |
| T42 | <elros34> hmm, do you have /sys/class/android_usb/android0 in recovery? | 00:39 |
| T42 | <HengYeDev> @elros34 In lineage recovery? Yes | 00:39 |
| T42 | <elros34> try to change order here so gadget_dir directory is checked first: https://github.com/mer-hybris/hybris-boot/blob/master/init-script#L243-L247 | 00:41 |
| T42 | <HengYeDev> still nothing in dmesg | 00:44 |
| T42 | <elros34> did you correctly recreate initramfs and then hybris-boot.img? | 00:45 |
| T42 | <HengYeDev> i used AIK, went into ramdisk and edited init, then repackimg.sh (re @elros34: did you correctly re...) | 00:45 |
| T42 | <elros34> change in line 276 sleep value to something bigger to confirm this is really usb issue | 00:46 |
| T42 | <HengYeDev> @elros34 Yes, it definitely is, i changed it to 120 seconds and now it reboots after 2 mins | 00:50 |
| T42 | <elros34> ok so add "write $ANDROID_USB/enable 1" at the end of usb_setup_configs function, maybe also 0 at the beginning | 00:52 |
| T42 | <elros34> maybe is also needed to add "write $ANDROID_USB/functions $USB_FUNCTIONS" before writting to UDC line | 00:54 |
| T42 | <elros34> I am not sure this will help but this is how *usb.rc file for your device looks like | 00:55 |
| T42 | <dcmmc> mal: you are right, the generated .ks misses local repo | 04:51 |
| T42 | <dcmmc> after fix this, the generated rootfs.tar.bz2 failed to be extracted | 04:51 |
| T42 | <dcmmc> https://pastebin.com/hYvgMH8J | 04:51 |
| deathmist | soooo is OBS still shutting down for good? is it worth to even update my port there anymore? I've not updated it since 3.4 and I don't know if I should put in the effort at this point | 05:35 |
| T42 | <dcmmc> oh i can extract the tar.bz2 on linux but failed on my mac and TWRP | 07:08 |
| T42 | <dcmmc> TWRP failed with tar -xj -f $ARCHIVE -C $ROOT/ || abort 10 "Couldn't extract SFOS rootfs!" | 07:09 |
| rinigus | deathmist: it is not known what will happen with OBS after may 2022. I wouldn't call it "shutting down for good" | 07:14 |
| rinigus | I guess Jolla will have to make decision regarding it. | 07:15 |
| rinigus | just flypig pointed out in his post that original decision regarding obs refloating was done until may this year, nothing else. no official statement regarding it is issued | 07:16 |
| deathmist | @dcmmc find older TWRP, extract it from Android with root or copy the .tar.bz2 to /data and boot hybris-boot.img, then extract it via telnet to /data/.stowaways/sailfishos before doing "echo continue >/init-ctl/stdin" | 07:40 |
| deathmist | newer TWRP's don't even have support for unpacking bzip2 compressed archives anymore... | 07:40 |
| T42 | <dcmmc> thank you for the advice | 07:40 |
| T42 | <dcmmc> i found it’s twrp’s issue | 07:40 |
| deathmist | rinigus: I see, well I guess I may wait anyway, doesn't make sense to update it just for OBS to shut down in a few months (if thats' what ends up happening) | 07:43 |
| rinigus | deathmist: understandable. let's hope Jolla can make decision promptly and not drag it till the end as it looks to be impeding the use of OBS - as in your case | 07:45 |
| T42 | <elros34> @dcmmc but in your repos I saw hybris-installer which I assume contains workaround for twrp issue, what did you change so it doesn't work anymore? Missing local-repo is also because some changes you made to droid-config | 11:10 |
| deathmist | @elros34 hybris-installer is my project and I gave up on trying to support TWRP, the env just isn't worth supporting when they (or Android installed on /system) may remove any tools they want from busybox/toybox etc. at any point. still figuring out what I want to do but it seems I'll be going for a custom initramfs | 11:49 |
| T42 | <elros34> ok, he managed to flash image previously. sfos zip should contains all needed tools to unpack tarball (busybox-static) then | 11:55 |
| T42 | <dcmmc> @elros34 I used the out-of-dated version. After sync from upstrea | 15:45 |
| T42 | <dcmmc> updates, the issue has been solved. | 15:45 |
| T42 | <dcmmc> I found the maintainer of sfos-oneplus 6 has finished a preliminary port to sfos 4 in last year w/ hybris-17.1. He said it's easy to get UI by following the porter guide. I have no idea why I have trouble getting UI on hybris-16. Is hybris-17 more suitable for porting sfos 4 than hybris-16? | 15:52 |
| T42 | <calebccff> ah someone picking up the op6 port? that would be nice :> | 15:57 |
| T42 | <calebccff> i was considering having a go at a mainline port if i ever have a rainy day | 15:57 |
| T42 | <elros34> did you try what I asked about surfaceflinger? There is not connection between sfos 4 and hybris 16/17 base. Obviously there can be differences in android side across android versions | 15:58 |
| deathmist | @dcmmc check if you missed some manual steps from https://github.com/sailfish-oneplus6/sf-enchilada/blob/master/building.md | 16:00 |
| T42 | <dcmmc> @calebccff yep I’m trying to port the latest sf on op6 based on your GitHub repo. But I encountered some problems in getting UI. Lipstick crashed but I have no idea why. | 16:02 |
| T42 | <calebccff> blep, it's been actually almost exactly 2 years since i made the last SFOS release and i haven't really touched it since then | 16:02 |
| T42 | <dcmmc> @elros34 I’m now going to rebuild sf based on Caleb’s hybris-17 HAL | 16:03 |
| T42 | <elros34> I get it but have you tired surfaceflinger? | 16:03 |
| T42 | <dcmmc> deathmist: I have followed this manual carefully but no luck | 16:04 |
| T42 | <dcmmc> At present nope, wait me change to hubris-17 and rebuild sf | 16:05 |
| deathmist | @dcmmc doubt that'll help you with getting UI any faster | 16:07 |
| deathmist | I tried to move my OnePlus 5/5T port to 17.1 but didn't get UI, I probably missed something small but I've not tried it again in a while | 16:07 |
| T42 | <dcmmc> Yesterday I followed the hadk guide from scratch to build a lineage 17.1 based version but I just got reboot very quickly | 16:08 |
| deathmist | I assume you started with blank repositories for e.g. kernel and whatnot? | 16:08 |
| deathmist | (as in clones from LineageOS) | 16:08 |
| T42 | <dcmmc> Deathmist: thank you a lot | 16:08 |
| T42 | <dcmmc> Yesterday I try to start from blank and followed the official hadk guide step by step | 16:09 |
| T42 | <dcmmc> But it even failed to boot | 16:09 |
| deathmist | yeah soo you're missing all the important patches that make things work, look through the existing repositories for the device | 16:11 |
| deathmist | I know for a fact you need this to boot: https://github.com/sailfish-oneplus6/android_kernel_oneplus_sdm845-stable/commit/00e54e9 | 16:11 |
| T42 | <dcmmc> I’m going to follow Caleb’s code with hybris-17. Thanks him for providing the guide and code | 16:11 |
| T42 | <dcmmc> Oh I really miss this important patch! | 16:13 |
| T42 | <dcmmc> Thank you a ton | 16:13 |
| T42 | <dcmmc> Wait, I found I just forked Caleb’s kernel repo | 16:15 |
| T42 | <dcmmc> So it should have these patches | 16:16 |
| T42 | <dcmmc> @calebccff it’s amazing to use mainline. But I guess I will have more troubles to use mainline now. Looking forward to your wonderful progress on it😄 | 16:21 |
| T42 | <dcmmc> @elros34 after rebuilding I will try surfaceflinger as soon as possible if I still stuck in GUI | 16:22 |
| T42 | <calebccff> yeah getting hardware working on mainline will be a bit annoying, we'll need to run modemmanager and write some daemon to translate the userspace ofono integration with modemmanager backend | 16:25 |
| T42 | <calebccff> that said, almost everything works now, i did boot SFOS using the pinephone image and mainline kernel and got pretty good performance | 16:26 |
| T42 | <elros34> @dcmmc no need to, on 17 you will have different issues | 16:32 |
| T42 | <dcmmc> it's so great! could you please provide the source code so that I can follow you. I’m willing to have a try on the mainline port and helping for some simple test and minor fix | 16:33 |
| T42 | <dcmmc> @elros34 okay I will try it on this weekend | 16:34 |
| T42 | <calebccff> it should be fairly simple to follow the pinephone port, i only had to edit a few config files (some lipstick conf hardcodes the drm device for some reason). | 16:36 |
| T42 | <dcmmc> Alright, the workload sounds small | 16:42 |
| T42 | <Wardinbush1> Could I install any custom roms on M20? I tried some of them such as sailfish but I failed every single time. | 16:45 |
| deathmist | @Wardinbush1 this is no place to talk about ROMs, but only Sailfish OS. I'm assuming you mean the Samsung Galaxy M20; if it has an unlockable bootloader (and kernel sources) yes it's possible. any particular error message or logs you get? | 16:53 |
| T42 | <Wardinbush1> No, only black screen, the OS does not run for me (re @SailfishFreenodeIRCBridgeBot: <deathmist>@Wardinbu...) | 16:56 |
| T42 | <Wardinbush1> Yes Iam talking about the Samsung Galaxy M20 | 16:56 |
| T42 | <Verevka86> 4.3.0.15- https://www.toptal.com/developers/hastebin/inocatened.properties | 16:57 |
| T42 | <Verevka86> how to fix it? | 16:58 |
| deathmist | @Wardinbush1 how exactly are you even installing Sailfish OS on it? it doesn't look like the device has a port yet (unless you're porting it right now?) in which case link to your repositories. black screen is also potentially good | 17:02 |
| deathmist | are you reading https://sailfishos.org/develop/hadk and specifically section 9.2? | 17:02 |
| deathmist | @Verevka86 looks like Jolla repos might be down, try again later I suppose unless someone else knows how to help | 17:04 |
| T42 | <Wardinbush1> I tried installing it long ago, I will do it again and I will let you know (re @SailfishFreenodeIRCBridgeBot: <deathmist>@Verevka8...) | 17:04 |
| deathmist | @Verevka86 oh actually maybe you ran out of disk space? | 17:05 |
| T42 | <Verevka86> no 😊 (re @SailfishFreenodeIRCBridgeBot: <deathmist>@Verevka8...) | 17:05 |
| T42 | <elros34> have you tired refreshing repos and installing that package manually in target? Do you have all repos in target set to 4.3.0.15? | 17:06 |
| Thaodan | Why is there a need for modemmanager, has ofono no support for that modem? | 17:06 |
| deathmist | Thaodan: yes, ofono doesn't support QRTR properly which is uses to talk to the modem on modern mainline Qualcomm devices | 17:07 |
| deathmist | while ModemManager supports it quite well and even calls work | 17:07 |
| Thaodan | hm there's this https://github.com/msm8953-mainline/ofono/commits/master | 17:08 |
| deathmist | yeah I tried it, it didn't do anything interesting | 17:08 |
| deathmist | i.e. none of the test scripts worked | 17:08 |
| deathmist | @calebccff | 17:08 |
| T42 | <calebccff> Thaodan: yeah there's no way you're gonna get Ofono working on mainline qcom without a lot of effort, and there's also no point, modemmanager is just better | 17:09 |
| Thaodan | Talking about NIH. Idk I get why some pushed modemmanager since there's no networkmanager <-> ofono integration? But its lost mainpower in a way to have to services maintained. | 17:12 |
| T42 | <calebccff> not sure i understand what you're saying | 17:14 |
| Thaodan | My point is that now we have two services maintained that fill very similar roles | 17:19 |
| Thaodan | Work gets duplicated, where as before there was just mostly ofono now you have to choose where to contribute. | 17:21 |
| T42 | <calebccff> Thaodan: calling "ofono" maintained is a bit of a stretch from what i've heard | 17:21 |
| T42 | <calebccff> (ask why plasma mobile switched to modem manager) | 17:22 |
| Thaodan | Idk, I know that it wasn't that active in the first place if I look at the upstream commit history. | 17:28 |
| Thaodan | but why did they switch? | 17:28 |
| T42 | <calebccff> ofono is buggy as heck and ignoring patches / fixes | 17:32 |
| T42 | <calebccff> modemmanager has had a whole lotta work put into it to support the pinephone modem properly, likewise for sdm845 / qcom. so switching to modemmanager let them support loads more devices, and not have to carry a patched fork of ofono | 17:33 |
| Thaodan | oh that sounds bad. | 17:37 |
| T42 | <calebccff> yuh, that's my understanding at least | 17:38 |
| Thaodan | esp if patches are ignored but idk maybe its related to the missing manpower | 17:38 |
| deathmist | whatever it is it's certainly not helping make Ofono a good candidate for mainline devices if nothing is merged | 17:38 |
| T42 | <calebccff> now if only someone ports the hybris ofono plugin to modemmanager and ya'll switch over :D | 17:39 |
| deathmist | I was actually thinking of implementing one, maybe I'll look into binder magic one day, but I already have too many projects :/ | 17:39 |
| T42 | <calebccff> ahaha, mood | 17:40 |
| deathmist | slava: maybe you'd be interested;: how would switching from Ofono to ModemManager sound? | 17:40 |
| T42 | <calebccff> if only we could add support for MM in the SFOS ui ourselves 🤔 | 17:40 |
| Thaodan | could mm work with connman? | 17:41 |
| T42 | <adampigg> @calebccff thaodan: im always being told to make MM work with sfos just for the pinephone | 17:41 |
| deathmist | with a plugin why not (assuming it's not already merged), they'd probably be happy to merge it if one is implemented | 17:41 |
| T42 | <calebccff> @adampigg if you do it i'll bring up SFOS with mainline on the OnePlus 6 and SHIFT6mq | 17:42 |
| T42 | <slavamon> sounds pretty much impossible (re @SailfishFreenodeIRCBridgeBot: <deathmist>slava: ma...) | 17:42 |
| deathmist | nothing is impossible, how come? | 17:43 |
| T42 | <adampigg> well, today i got the PP modem FW guy using/debugging with ofono, so im hopeful i dont have to! | 17:43 |
| T42 | <adampigg> i did consider a MM/Ofonoe dbus bridge, so, you could run MM, but it would expose an ofono dbus api | 17:43 |
| T42 | <calebccff> b-but more mainline devices ! | 17:43 |
| T42 | <calebccff> that would be my solution | 17:43 |
| T42 | <calebccff> (given how i can't touch UI code :<) | 17:43 |
| T42 | <adampigg> you could send me ofono patches ;) | 17:44 |
| T42 | <slavamon> emulating ofono dbus api is probably easier than ripping off all the dependencies (re @adampigg: i did consider a MM/...) | 17:44 |
| T42 | <calebccff> I barely know how phones work, nevermind the modem :P | 17:44 |
| T42 | <calebccff> that would be significant work from what i know | 17:45 |
| T42 | <calebccff> (which is not very much XD) | 17:45 |
| mal | there is some much code in various packages using the ofono apis the work to change it to another api would be quite a lot of work | 17:45 |
| mal | *so much | 17:45 |
| mal | and of course rewriting the binder plugin etc | 17:46 |
| deathmist | so we should focus on implementing the inverse of https://github.com/droidian/oFono2MM, got it | 17:46 |
| deathmist | ideally the daemon for that would be able to do it both ways so there aren't 2 projects for the same thing | 17:47 |
| deathmist | and maybe not written in python :p | 17:47 |
| piggz | yes, ive seen that ... the reverse is what we would need | 17:47 |
| T42 | <eugenio_g7> Droidian would switch to MM directly at some point | 17:50 |
| deathmist | @eugenio_g7 with support for droid HALs et all? | 17:50 |
| deathmist | sounds nice | 17:50 |
| T42 | <calebccff> I'd settle for a UI that the community could hack on and add support themselves, that would probably deal with the "lot of work" part at least | 17:50 |
| T42 | <calebccff> ofono2mm is a big hack really | 17:51 |
| T42 | <eugenio_g7> deathmist: wanted to give that a shot at least | 17:51 |
| deathmist | @eugenio_g7 good luck! it sounds like a solid solution; if we get binder modem support in upstream that would be awesome | 17:51 |
| T42 | <eugenio_g7> thanks :D I don't think upstream would be interested though | 17:53 |
| mal | I assume modemmanager supports external plugins? | 17:53 |
| T42 | <eugenio_g7> yes (sorta) | 17:55 |
| T42 | <calebccff> mal: https://www.freedesktop.org/software/ModemManager/doc/latest/ModemManager/ref-overview-plugin-specific-modems.html | 17:56 |
| T42 | <calebccff> not sure if that's offers enough functionality. End goal should be to merge binder support upstream, they're quite open to that stuff afaict | 17:56 |
| T42 | <calebccff> e.g. support for QRTR/QMI modems requires 2 external libraries which are both there as well | 17:57 |
| mal | does it use both libqmi and libqrtr? | 17:57 |
| T42 | <calebccff> yeah | 17:57 |
| mal | I have used both when hacking qcom sensor support for mainline kernel | 17:58 |
| T42 | <calebccff> actually now that i think of it, i know some people have had success running Qcom downstream RILD on AOSP with mainline linux | 17:58 |
| T42 | <calebccff> so there's a decent chance that you could actually drop Android RILD entirely on qcom and just run modemmanager with native QMI/QRTR support | 17:58 |
| T42 | <AntonlX> Hi Thaodan, how is going xperia x sound fixing? | 18:42 |
| T42 | <HengYeDev> I added all of those, didn't work (re @elros34: ok so add "write $AN...) | 19:45 |
| T42 | <elros34> might little hard to figure out it, maybe you need to play with these settings in twrp. Are you sure configfs function is executed not android_usb? | 19:48 |
| T42 | <HengYeDev> how could i detect that? maybe adding different sleep periods? (re @elros34: might little hard to...) | 19:48 |
| T42 | <elros34> naah, if function is simple, just be sure (no idea how is your scripting knowledge that is why I ask) | 19:50 |
| T42 | <HengYeDev> @elros34 when usb_setup called setup_configfs should be called first if gadget_dir exists, which i've checked that it does, so i think it should be executed | 19:51 |
| T42 | <HengYeDev> should i just remove the if statement and just execute usb_setup_configfs | 19:51 |
| T42 | <elros34> no need to but it will not harm. Maybe you could check how it's done in android/twrp like: content of /config/usb_gadget/g1/functions/ and others. https://github.com/LineageOS/android_device_samsung_gta4xl-common/blob/lineage-17.1/config/init/init.exynos9611.usb.rc#L132 | 19:56 |
| T42 | <HengYeDev> @elros34 is it bad if the only thing in /config/usb_gadget/g1/functions is ffs.adb in recovery? | 19:58 |
| T42 | <elros34> or if you will remove usb setup in initramfs (DONE_SWITCH=no) then it should boot to real rootfs and try to setup it again with logging to /data/.stowaways/sailfishos/init.log | 19:58 |
| T42 | <elros34> ffs.adb probably expected becauuse only adb connection is active | 19:59 |
| T42 | <HengYeDev> @elros34 bootloop on comment out DONE_SWITCH | 20:02 |
| T42 | <elros34> no, you get it wrong | 20:02 |
| T42 | <elros34> the point is to not execute usb_setup at all in initramfs, same script is in sfos rootfs | 20:05 |
| T42 | <elros34> is DATA_PARTITION inside script in initramfs contains by-name path or path to device block? | 20:06 |
| T42 | <elros34> or maybe you have encrypted /data? There must be a reason why there is no switch to sfos rootfs but instead usb_setup is called | 20:12 |
| T42 | <HengYeDev> @elros34 data partition is sda23, data doesn't seem to be encrypted | 20:38 |
| T42 | <HengYeDev> `/dev/block/platform/13520000.ufs/by-name/userdata /data ext4 defaults 0 0`in fstab | 20:38 |
| T42 | <HengYeDev> what should i do instead of comment out DONE_SWITCH=no? | 20:39 |
| T42 | <elros34> in fstab but I ask how it looks inside initramfs | 20:47 |
| T42 | <HengYeDev> @elros34 not sure what you mean, i added https://gist.github.com/HengYeDev/52f0804124ad169e17e7b2de4f909103 to fixup-mountpoints, in initramfs DATA_PARTITION is set to DATA_PARTITION=/dev/sda32 | 20:52 |
| T42 | <HengYeDev> @elros34 not sure what you mean, i added https://gist.github.com/HengYeDev/52f0804124ad169e17e7b2de4f909103 to fixup-mountpoints, in initramfs DATA_PARTITION is set to DATA_PARTITION=/dev/sda32, which is the correct node (edited) | 20:52 |
| T42 | <elros34> ok, so if you do not have /data/.stowaways/sailfishos/init.log then either /data is ecrypted or check_kernel_config function fails | 20:56 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!