| 4 Jun 2024 |
 erahhal | And was told that I should only need what's there. | 00:04:48 |
| erahhal | This is what it responds with:
ser@host:~/lineage-21$ cat .repo/local_manifests/muppets.xml
<?xml version="1.0" encoding="UTF-8"?>
<manifest>
<project name="TheMuppets/proprietary_vendor_google_cheetah" path="vendor/google/cheetah" remote="github" revision="lineage-21"/>
</manifest>
user@host:~/lineage-21$ rm -rf vendor/google/cheetah
user@host:~/lineage-21$ repo sync vendor/google/cheetah
| 00:05:24 |
|  raitobezarius changed their display name from raitobezarius (DECT: 7248) to raitobezarius. | 11:13:47 |
| erahhal | Ok, I just successfully built lineageos locally with a standard build environment. I can confirm that the source device dir contains pantah, pantah-kernel, gs101, gs201, and gs-common with no corresponding vendor dir and builds successfully. | 22:48:18 |
| erahhal | Does this indicate that the robotnix code needs to change to not expect vendor dirs for those devices? | 22:48:33 |
| erahhal | I'm going to try adding a "noVendor" filter to device-dirs.json and filter those out in lineageos/default.nix | 23:59:54 |
| 5 Jun 2024 |
| erahhal | Was able to successfully get the Robotnix source dir structure to match the structure in the standard build, but still getting the no kernel error Neither INSTALLED_KERNEL_TARGET nor INSTALLED_BOOTIMAGE_TARGET is defined when PRODUCT_OTA_ENFORCE_VINTF_KERNEL_REQUIREMENTS is true | 00:43:25 |
| erahhal | Realizing that maybe the original choosecombo failure when I first tried this might be the same root cause. Digging into the envseetup source | 03:01:58 |
| erahhal | Ok, I think I'm getting somewhere... Calling breakfast instead of choosecombo seems to do the trick. | 04:15:30 |
| erahhal | Sorry for spamming here. Please let me know if my level of messaging is inappropriate. | 04:15:46 |
| erahhal | This is what I'm currently working on:
robotnix-lineage_cheetah> FAILED:
robotnix-lineage_cheetah> In file included from tools/tradefederation/prebuilts/test_harness/Android.mk:55:
robotnix-lineage_cheetah> vendor/google/cheetah/Android.mk:9: error: vendor/google/cheetah/radio/abl.img SHA1 mismatch (8f468eae2412d469a892638518b83841fe1bf380 != 4e5ac9c69b8df933427d83c901bee75e7e777900).
robotnix-lineage_cheetah> 04:09:10 ckati failed with: exit status 1
error: builder for '/nix/store/2hkxwmn5yi63z33wkrs7z47l6bxvg4ak-robotnix-lineage_cheetah-2024060317.drv' failed with exit code 1;
last 10 log lines:
> [ 93% 296/315] including tools/platform-compat/Android.mk ...
> [ 94% 297/315] including tools/test/connectivity/Android.mk ...
> [ 94% 298/315] including tools/test/graphicsbenchmark/Android.mk ...
> [ 94% 299/315] including tools/tradefederation/core/Android.mk ...
> [ 95% 300/315] including tools/tradefederation/prebuilts/test_harness/Android.mk ...
> [ 95% 301/315] including vendor/google/cheetah/Android.mk ...
> FAILED:
> In file included from tools/tradefederation/prebuilts/test_harness/Android.mk:55:
> vendor/google/cheetah/Android.mk:9: error: vendor/google/cheetah/radio/abl.img SHA1 mismatch (8f468eae2412d469a892638518b83841fe1bf380 != 4e5ac9c69b8df933427d83c901bee75e7e777900).
> 04:09:10 ckati failed with: exit status 1
For full logs, run 'nix log /nix/store/2hkxwmn5yi63z33wkrs7z47l6bxvg4ak-robotnix-lineage_cheetah-2024060317.drv'.
error: 1 dependencies of derivation '/nix/store/zvkg6c7yhbsc2m65gk0z8sjymj35p7rf-cheetah-factory-2024060317.zip.drv' failed to build
| 04:16:30 |
| erahhal | Trying to figure out where the build system is getting that hash | 04:38:50 |
| erahhal | Ahhhh, the radio files are using git lfs | 05:25:57 |
| erahhal | Manually updated the vendor.json entry for cheetah setting fetchLFS to true to no effect. digging into why this is not working. | 05:50:46 |
 @samueldr:matrix.org | did you update the hash? | 05:52:16 |
| erahhal | No, trying that now. | 06:03:24 |
| @samueldr:matrix.org | overall, remember that Nix trusts you when you tell it what the hash of a thing is | 06:15:02 |
| erahhal | I pulled down the LFS repo, then realized I don't know what I should be hashing | 06:15:25 |
| erahhal | No idea how to get the hash for an LFS repo | 06:15:36 |
| erahhal | Better yet, how should the update_device_dirs.py script know that a repo has LFS files? I guess that would be specifically flagged similar to other devices with special cases? | 06:19:39 |
| erahhal | Did the old trick of corrupting the hash to see what Nix would expect, and it still wants to old hash despite fetchLFS being true. | 06:24:21 |
| @samueldr:matrix.org | the "old trick" is "TOFU"; trust on first use | 06:24:52 |
| @samueldr:matrix.org | and yeah | 06:24:56 |
| @samueldr:matrix.org | though I guess you need something else | 06:25:09 |
| @samueldr:matrix.org | (haven't touched robotnix in a while) | 06:25:14 |
| erahhal | In reply to @samueldr:matrix.org
the "old trick" is "TOFU"; trust on first use Learned a new term | 06:25:34 |
| @samueldr:matrix.org | it's the name of the pattern like with SSH where you are asked to "trust on first use" [though you should also check really] | 06:26:19 |
| @samueldr:matrix.org | T&COFU doesn't have a good sound to it | 06:26:32 |
| erahhal | So to be clear, even if the hash is changed, it will ignore the updated properties until it finds a new matching hash? | 06:28:18 |
| erahhal | otherwise erroring out and ignoring the props | 06:28:40 |
