| 15 Oct 2025 |
 Atemu | With correct image and all | 16:08:31 |
 volth | not actually, but it is a good point to start:
INFO | Android emulator version 36.3.2.0 (build_id 14184492) (CL:N/A)
INFO | Graphics backend: gfxstream
INFO | Found systemPath /nix/store/nabl5nqq1hz6mns1cvmx8fss7v9c9yd7-fake-sdk/system-images/android/x86_64/
FATAL | QEMU2 emulator does not support arm64 CPU architecture
| 17:14:43 |
| Atemu | @volth:matrix.org yeah that means it's the wrong arch | 17:28:48 |
| volth | It runs! On the screen there is only sort of a screensaver with 3 circles (https://imgur.com/arsGi9L.png), while adb shell does work. And the guest works very slow as though it emulates a foreign architecture. probably -enable-kvm is required somewhere | 20:27:12 |
| Atemu | That's the LOS boot animation | 20:28:05 |
| Atemu | If you're emulating, it'll run veeeery slowly, so it getting stuck is unfortunately normal | 20:28:38 |
| Atemu | You can't do KVM with a foreign arch. Like, that conceptually does not work. You need to set the correct arch via the robotnix option then it works with KVM automagically | 20:30:05 |
| Atemu | At least it did half a year ago when I fixed the emulator | 20:30:22 |
| volth | It is not foreign, x86_64 everywhere | 20:31:40 |
| Atemu | Ah, I misread | 20:32:01 |
| Atemu | Hm, does KVM work otherwise on your system? | 20:32:25 |
| volth | yes. I tried BlissOS in normal qemu (not avd emulator) and it runs quite fast | 20:33:31 |
| Atemu | There was some way to make the emulator print debug logs | 20:33:57 |
| Atemu | And you could have a look at the cli args of the qemu process | 20:34:14 |
| volth | It is AOSP with virtio drivers | 20:34:15 |
| volth |
LOS boot animation
that boot animation runs for 25 minutes already....
| 20:35:59 |
| Atemu | Is it progressing or still stuck? | 20:36:22 |
| Atemu | Trying to figure out whether it did actually boot quickly and the graphics stack is simply stuck | 20:36:41 |
| Atemu | Check the logcat to figure out what state it's in | 20:36:56 |
| Atemu | And there was some way to check whether the system has booted | 20:37:11 |
| Atemu | Need to dig that out of my bachelor thesis lol | 20:37:22 |
| Atemu | getprop sys.boot_completed | 20:37:41 |
| Atemu | You could play with the different GPU options | 20:38:08 |
| Atemu | I did have to put some other version of mesa.drivers into the LD_LIBRARY_PATH when I last played around with that | 20:38:55 |
| Atemu | Probably the one from the nixpkgs input of the robotnix flake | 20:39:07 |
| volth | Probably this: Could not find 'aidl/activity' for ctl.interface_start from pid: 169 (/system/bin/servicemanager)
Likely I am the first one who run LOS 22.2 in the emulator, I had to start with something older and tested
| 20:50:35 |
| volth | logcat full of different errors, but missing aidl is most often and likely relevant | 20:51:26 |
| Atemu | Not sure; is there a stacktrace? Android logs tonnes of benign errors that sound critical | 20:51:42 |
| Atemu | Like, if the system can't boot for one reason or another, you'll usually have zygote error out with a huge-ass stack trace | 20:52:40 |
| Atemu | You could try 22.1 I suppose; that did work | 20:53:17 |
