ChanServ changed the topic of #robotnix to: Robotnix: https://github.com/danielfullmer/robotnix || Channel logs: https://logs.nix.samueldr.com/robotnix
danielrf[m] has quit [*.net *.split]
danielrf[m] has joined #robotnix
henri has joined #robotnix
<henri> I can't build the GrapheneOS release 2021.03.02.10 on sunfish. Somehow many of the Git hashes in repo-RQ2A.210305.006.2021.03.02.10.json are wrong but that seems to be a problem inherited from the upstream platform_manifest.
<henri> I'll write to Daniel Micay to see what's up.
henri has quit [Quit: ERC (IRC client for Emacs 27.1)]
henri has joined #robotnix
<henri> In the meantime I have a patch with fixed hashes here: https://git.henrimenke.de/henri/robotnix/commit/c7a54b5e2b338fe8a7b973593653fac555932e68
henri has quit [Client Quit]
<danielrf[m]> Yep, confirmed, looks like the tag changed
<danielrf[m]> I also independently updated and can confirm that I now get the same revisions hmenke did for that tag
<danielrf[m]> Here's the diff before/after for `device/google/sunfish` and `kernel/google/sunfish`
<danielrf[m]> Mostly just the change to fts_touch_s5, which I'll also switch in vanilla flavor
<ajs124> I managed to build the old tag, but apparently the touch screen is broken with that one.
<ajs124> Haven't tried it myself, but my flatmate updated his sunfish.
<danielrf[m]> I also have a sunfish running the vanilla flavor. I updated to the latest tag but my touchscreen still works
<danielrf[m]> even though the vanilla flavors for sunfish also uses the wrong touchscreen repo in the kernel
<danielrf[m]> strange
<samueldr> some phones have multiple variants of compatible parts depending on availability
<samueldr> so it might be that your sunfish has part A that works fine with the wrong repo, but phones with part B needs fixes as it's not actually compatible
<samueldr> (or maybe totally incompatible parts, but resolved at boot time by the firmware fixing up the FDT!)
<danielrf[m]> hmm, interesting
<danielrf[m]> I plan to update the vanilla flavor to use the new repo in either case
<danielrf[m]> Just need to verify that it doesn't break my device first
<ajs124> samueldr I've had that with raspberrypis. Same model, different rev and USB didn't come up with the exact same firmware and kernel.
<samueldr> btw, what I said was 100% conjecture, did not check anything
<samueldr> but I've seen it be a thing for things like even the display panel
<samueldr> where the firmware detects the panel, and forces values into the FDT
<ajs124> hm. my sunfish seems to bootloop after sideloading the update.
<danielrf[m]> ajs124: Is this with the latest commit on master?
<ajs124> c7a54b5e2b338fe8a7b973593653fac555932e68
<ajs124> although I had already flashed the one with the broken hashes beforehand through OTA, but the touchscreen stopped working as well
<ajs124> (couldn't stop it anymore after I found out…)
<danielrf[m]> And your earlier build was based on ae32ae5ec112bd6e2ead1978ffa1c5ba8ca850b7 ?
<danielrf[m]> device="sunfish" and flavor="grapheneos" ?
<ajs124> yes to both (or all three)
<ajs124> it's still unlocked, so I could just try and apply the update via fastboot, but idk if simply omitting -w in fastboot update leads it to not erase userdata
<danielrf[m]> you updated with `adb sideload` right?
<ajs124> the things is, it does start to boot and shows up over adb, but since grapheneos does not seem to persist adb keys, I can't view the logs
<ajs124> exactly, from the grapheneos recovery
<danielrf[m]> I would have hoped that with enough bootloops it would fallback to your earlier version
<danielrf[m]> the a/b update mechanism is supposed to do that IIRC
<danielrf[m]> also that I thought that issue with grapheneos not persisting adb keys was fixed earlier
<ajs124> maybe I haven't used adb since
<ajs124> I sort of assumed that was by design tbh
<danielrf[m]> too bad hmenke isn't on here since I wonder if he'd already tried the latest grapheneos on sunfish
<ajs124> I'm just #yolo downgrading with fastboot now. Maybe that will work 🤷‍♂️
<danielrf[m]> Once I've tested this issue with vanilla on my sunfish I will test grapheneos as well
<danielrf[m]> seemed to have worked for him
<danielrf[m]> I wonder what is different about your device
<ajs124> weird…