| 5 Sep 2021 |
 kranzes | why are you suspecting that its probably the webview/chromium? | 19:33:04 |
 pie_ | hm ok;
$ cat expr.nix; nix-instantiate --eval expr.nix --argstr a 1 --argstr b 2
{a}: ({b ? 3}: b) {}
3
i guess that explains why this doesnt ever seem to break anything
"<cleverca22>the inner function already got ran, so --arg cant do anything do it"
| 19:34:10 |
 danielrf | I've had bootloops in the past caused by incorrect webview configuration--although it's been a while since that's happened for me. | 19:34:46 |
| pie_ | what does this mean? " The current code in robotnix only works with a single custom device type. E.g. the attestation service cannot handle robotnix-customized versions of both crosshatch and sunfish" | 19:50:16 |
| kranzes | 
Download image.png | 20:02:41 |
| kranzes | danielrf Now we wait | 20:02:43 |
| danielrf | pie_: We can patch the AttestationServer / Auditor packages to include supporting attestation for the custom builds that we make in robotnix. This involves adding certain AVB fingerprints that correspond to the keys you've generated for your device. | 20:11:47 |
| danielrf | Here i'm just saying that we only add support for at most one custom device. | 20:12:05 |
| danielrf | It's not hard to extend the patches to include multiple devices, but the tricky bit is just how to set up the robotnix configuration--how do we get the list of customized devices and AVB key fingerprints we need. | 20:13:07 |
| danielrf | Like, if you have multiple devices and all their keys all get included in the patchset, then suddenly the build for a single device depends on the configuration for other devices, because it needs key fingerprints to construct the patch. | 20:14:20 |
| danielrf | * Like, if you have multiple devices and all their keys get included in the patchset, then suddenly the build for a single device depends on the configuration for other devices, because it needs key fingerprints to construct the patch. | 20:15:05 |
| pie_ | i have acquired a BIG CHONK machine | 21:12:12 |
| pie_ | danielrf: how long does it take for the heavy cpu usage part of the build to kick in? | 21:12:31 |
| pie_ | it doesnt seem to be doing a lot yet but its done cloning repos | 21:12:41 |
| danielrf | If you're in the main derivation then usually ~1minute or so for it to bootstrap and configure its build system | 21:13:53 |
| danielrf | Once ninja starts running then it should be heavily parallel | 21:14:27 |
| pie_ | oh it just did some kernel building and its doing LTO on a single core rightnow | 21:17:46 |
| danielrf | Yeah looks like you're probably building the kernel and chromium first | 21:18:10 |
| danielrf | Those are available on a binary cache if you're rather not build them yourself: https://docs.robotnix.org/building.html?highlight=cachix#binary-cache | 21:18:39 |
| pie_ | hm yeah i suppose i should | 21:21:08 |
| pie_ | i wonder , theres probably some tool that can give you a 3d histogram of core usage | 21:22:03 |
| pie_ | i wonder if theres also something that can interactively tell you whats using the cores at each sample point | 21:22:26 |
| pie_ | yep, its definitely building now | 21:22:48 |
| danielrf | * Those are available on a binary cache if you'd rather not build them yourself: https://docs.robotnix.org/building.html?highlight=cachix#binary-cache | 21:23:14 |
| 6 Sep 2021 |
 dsx | In reply to @danielrf:matrix.org
under modules/pixel/default.nix, add "arch/arm64/boot/Image.lz4-dtb" to the kernel.buildProductFilenames toward the bottom of the file in the section where it conditions on config.deviceFamily == "redfin" That didn't help unfortunately, got infinite reboot. | 00:17:29 |
| pie_ | /nix/store/crdbdgc194p5zpqsn9v3zfgpmvps7n4z-crosshatch-img-2021080300.zip
real 132m27.225s
user 0m8.527s
sys 0m9.705s
| 05:17:06 |
| pie_ | danielrf: should that be 2x your reported 1 hour? | 05:17:27 |
| kranzes | danielrf Ok it isn't the webview/chromium | 09:53:45 |
| kranzes | it still does not work | 09:53:50 |
| kranzes | * it still does not boot | 09:53:58 |
