| rinigus | @b100dian: thanks for dealing with the comm-adaptation. that should allow us to hook with ci at some point. | 06:46 |
|---|---|---|
| rinigus | hopefully you will sort out binfmt_misc issue. should be something with the host | 06:47 |
| T42 | <fredldotme> I am just now wondering whether libhybris with the wayland ws actually supports sub-surfaces at the moment. We seem to hit issues on Ubuntu Touch with firefox regarding that. | 09:43 |
| mal | @fredldotme what kind of things are sub-surfaces in wayland? | 11:46 |
| T42 | <fredldotme> for example the hamburger menu in gedit or gnome-text-editor spawns a subsurface as a child to the parent window surface. | 11:47 |
| mal | it's possible libhybris doesn't support those, not sure what kind of handling those need if any | 11:52 |
| mal | need to have a look into what could be needed | 12:25 |
| T42 | <fredldotme> I am unsure how the wl_surface <-> wl_buffer relationship is put together on hybris since the android-wayland.xml merely defines passing wl_buffers around. The relation from buffer to pertaining surface must happen somewhere else. In that case it could be it's working already, just not in an obvious fashion. | 12:27 |
| T42 | <fredldotme> Okay no, it makes sense that it does in fact not work: https://github.com/libhybris/libhybris/blob/9f61f26c44d9a3bf62efb67d4c32a7a0c89c21ca/hybris/egl/platforms/wayland/wayland_window.cpp#L112 | 12:35 |
| T42 | <fredldotme> unless there is one `WaylandNativeWindow` per subsurface also | 12:35 |
| rinigus | @b100dian: flashed and the first boot showed sfos logo. waited a bit and then rebooted | 18:23 |
| rinigus | right now have issues with getting back to fastboot in fastbootd mode. hang on with flashing... | 18:24 |
| rinigus | @b100dian: to get back to fastbootd after bad sfos boot, you have to flash aosp boot, vendor_boot and dtbo. then you can get back | 18:29 |
| rinigus | @b100dian: do you know what is the status for recovery? did you managed to get telnet working? | 18:31 |
| mal | I had issue of device rebooting quite soon also | 18:40 |
| mal | I found a way around it | 18:41 |
| mal | rinigus: did it reboot to what? for me it did this https://android.googlesource.com/platform/build/+/refs/heads/main/target/product/gsi/init.vndk-nodef.rc | 18:42 |
| T42 | <b100dian> rinigus: Telnet on boot (not recovery) was available on port 23 without rootfs. It should be on 2323 if you created mic fs | 18:42 |
| T42 | <b100dian> But you should be able to create some init_debug files to stop it more early right? Am AFK so don't know the exact file name | 18:44 |
| rinigus | mal: how would I know whether it rebooted on BOARD_VNDK_VERSION? | 18:44 |
| rinigus | now understood your message... | 18:44 |
| rinigus | device was connected to usb and rebooted back to sfos logo | 18:45 |
| rinigus | note that I flashed: boot dtbo vendor_boot and userdata | 18:45 |
| rinigus | possibly due to such combination, I could not get back to fastbootd | 18:46 |
| rinigus | for that had to flash dtbo boot vendor_boot from aosp | 18:46 |
| rinigus | @b100dian: hmm, I wonder how would it work. as it should be looking for root partition | 18:47 |
| rinigus | mal: on boot, after sony logo, I get sfos logo. then screen goes dark and, after some time, it reboots. while usb is connected, it does not boot into bootloader (blue light) but goes on to full boot. and the cycle repeats | 19:17 |
| rinigus | during this process, I have usb connected | 19:17 |
| rinigus | don't see any traces of telnet connection or something except bootloader messages in dmesg of my pc | 19:18 |
| rinigus | (wondering if something is needed on linux pc for it to work) | 19:19 |
| T42 | <elros34> aosp hadk build doesn't have telnet | 19:33 |
| rinigus | @b100dian: I was missing your last commit in hybris-boot. the one with message "got telnet" | 19:36 |
| rinigus | @elros34: that's strange, I think it used to have in recovery at least | 19:36 |
| T42 | <elros34> maybe in recovery (I don't know) but not in normal boot | 19:37 |
| T42 | <elros34> looks like yes https://github.com/mer-hybris/hybris-initrd/blob/master/recovery-init but you would probably need to add peripheral mode changes from @b100dian hybris-boot | 19:39 |
| mal | ah, yes, that should be fixed | 19:40 |
| rinigus | yes, recompiling now... | 19:44 |
| T42 | <b100dian> rinigus: recompiling boot image or recovery? | 19:45 |
| rinigus | right now was hybris-hal. now will have to look for correct image | 19:46 |
| rinigus | going through the maze of submodules :) | 19:46 |
| rinigus | ideally would like to get into recovery then it will be easier to debug | 19:47 |
| mal | I should also add parsing the udc device from commandline, both to hybris-boot and hybris-initrd | 19:49 |
| T42 | <b100dian> rinigus: but the init script that enables peripheral mode is in hybris-boot/recovery init script from ramdisk, not sure why you'd need whole hybris hal rebuilt | 20:02 |
| T42 | <b100dian> I'n creating a mic image now, since I got to my computer, but this time on my laptop directly instead of kvm host - didn't find anything ready for that binfmt_misc thingie | 20:03 |
| rinigus | from reading the repos, looks like recovery and boot are made using hybris-initrd as a submodule of droid-hal-img-boot-sony-nagara | 20:03 |
| rinigus | what I don't see is how fixup mountpoints and init script from hybris-boot gets integrated into it. | 20:04 |
| rinigus | @b100dian: great! hopefully that mic image will work :) | 20:04 |
| rinigus | @b100dian: re hybris-boot that you used - it was directly after building hal, generated by android. (my build was fast as I have most ready anyway) | 20:06 |
| rinigus | but question of fixup and whether your changes in init propagate to droid-hal-img-boot-sony-nagara is still there | 20:06 |
| T42 | <elros34> hybris-boot's init is not integrated in droid-hal-img-boot- which uses hybris-initrd | 20:13 |
| T42 | <b100dian> "losetup: root.img: failed to set up loop device: No such file or directory" seems that the spec/patterns somehow specify that there's a lvm build? | 20:14 |
| T42 | <b100dian> anyway, I do have the sfe-xqcq54-4.6.0.13.tar.gz so that'd be enough for me no | 20:17 |
| T42 | <b100dian> *now | 20:17 |
| rinigus | @elros34: thanks, that makes sense. but what about mountpoints? | 20:18 |
| rinigus | @b100dian: it could be the case that we have config already for lvm | 20:18 |
| rinigus | I will have to charge the phone and for that, due to our awesome aosp, I would have to go to stock. as otherwise I am loosing charge in aosp on charger | 20:19 |
| T42 | <b100dian> you mean it doesn't charge while off or in fastboot ;D? | 20:27 |
| rinigus | It doesn't stay off, but boots into android. As for fastboot, no idea | 20:29 |
| rinigus | Btw, if it's better for you, we can convert to regular non lvm image. Not sure how much it will help... Later we can always go back to lvm | 20:31 |
| T42 | <b100dian> no worries, I got the tar file - problems pushing that rn but I'll find a way | 20:35 |
| rinigus | @b100dian: adb push? not sure it will work as we have also system and vendor in that image, don't we? | 20:47 |
| rinigus | did you use KS that was in ssu rpm? if you look into it, it has few resize img commands and so on. that's probably not what we have in regular mic fs | 20:49 |
| T42 | <b100dian> understood, but I have just taken the tar file for now and gonna expand it in /data/.stowaways/sailfishos | 20:52 |
| T42 | <b100dian> like in https://github.com/sailfishos-on-sake#alternate-installation-instructions | 20:54 |
| T42 | <b100dian> It's so funny we come at different angles at this, based on our previous experience :) but that's a good thing | 20:54 |
| rinigus | I am just looking in sake config and trying to figure out what's different from nagara that makes it go lvm | 20:55 |
| rinigus | looks like just having https://github.com/sailfishos-sony-nagara/droid-config-sony-nagara/tree/main/kickstart | 20:57 |
| rinigus | @b100dian: so, to make regular images, we can probably just move kickstart into something else | 20:58 |
| rinigus | plus something has to be done with boot images | 21:01 |
| T42 | <b100dian> flashing hybris-recovery and booting with unpacked rootfs (which indeed includes system and vendor) does not get me very far, 38 seconds to reboot because of missing ipa_fws.mdt | 22:42 |
| T42 | <b100dian> https://pastebin.ubuntu.com/p/SF32h8KbVv/ and I need to audit=0 on the next boot image | 22:42 |
| T42 | <b100dian> that's a red herring, android recovery has those messages in a loop too | 23:05 |
| T42 | <b100dian> i've chrooted into /data/.stowaways/sailfishos and masked droid-hal-init but that is not the problem | 23:09 |
| T42 | <elros34> a lot of things happens in journal at 30s of boot time so check that, not only dmesg | 23:18 |
| T42 | <b100dian> yes, exactly my next thought: https://pastebin.ubuntu.com/p/6QWqJ2HBwJ/ | 23:22 |
| T42 | <b100dian> It might be related to "ohmd.service: Failed to set up mount namespacing: No such file or directory" | 23:22 |
| T42 | <b100dian> maybe the kernel is not having all the things required? https://github.com/sailfishos-sony-nagara/android_kernel_sony_msm/commit/HEAD | 23:23 |
| T42 | <elros34> it's common but usually ohmd works after it restarts once or more | 23:23 |
| T42 | <b100dian> hmm "Starting usb-moded USB gadget controller" maybe I should mask that too | 23:24 |
| T42 | <b100dian> on screen what's happening is that the SailfishOS logo appears (yay nice development!) and then after 30> secs it goes blank. I want that logo on all my ports by default | 23:25 |
| T42 | <b100dian> :D | 23:26 |
| T42 | <elros34> I would start with fixing or disabling all these timouting mount units | 23:26 |
| T42 | <b100dian> agreed, but I need stable telnet/usb access | 23:30 |
| T42 | <b100dian> which I now got after chrooting again to stowaways and masking usb-moded ;) | 23:31 |
| T42 | <elros34> do you have USER_NS? it may be needed | 23:31 |
| T42 | <b100dian> it doesn't appear in defconfig, let me check .config | 23:33 |
| T42 | <b100dian> wait, I may have re-built with default kernel 🤦 | 23:35 |
| T42 | <b100dian> this is actually possible because I cleaned up out/ and didnt make it again | 23:35 |
| T42 | <b100dian> and the default kernel has rndis flags correctly, just hybris-boot init peripheral missing... | 23:35 |
| T42 | <b100dian> at least this time it'l be audit=0. Interesting how you can make hybris-boot with a default kernel in sony's trees, if you're not paying attention | 23:40 |
| T42 | <b100dian> USER_NS is not defined | 23:50 |
| T42 | <b100dian> with the correct kernel 😊 logcat at least stays up complaining | 23:50 |
| T42 | <b100dian> > audioadsprpcd: audio adsp daemon error dlopen failed: library "libadsp_default_listener.so" not found | 23:50 |
| T42 | <b100dian> journal looks saner, will continue to investigate tomorrow though https://pastebin.ubuntu.com/p/PG3vzswpCS/ | 23:53 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!