#asahi · 2021-12-17 08:27
marcanheh#asahi · 2021-12-17 08:25
marcanhope your employer is okay with working on this :p#asahi · 2021-12-17 08:23
marcandoing this at work?#asahi · 2021-12-17 08:19
marcanotherwise, it's just a waste of time, I'd rather test it myself when I get there and fix it then#asahi · 2021-12-17 08:19
marcanthe thing is, debugging this kind of stuff is easy if you have a m1n1 setup and are familiar with booting that way and you can try kernels out#asahi · 2021-12-17 08:18
marcanj`ey: no, it has a new undocumented codec with no public equivalent#asahi · 2021-12-17 08:18
marcan(what did I say about targeting developers doing tethered boot... sigh.)#asahi · 2021-12-17 08:18
marcanno#asahi · 2021-12-17 08:17
marcaneither way, the only thing that might work on that model is the headphones out, the speakers definitely aren't instantiated yet#asahi · 2021-12-17 08:16
marcancan you run tools/dump_pmgr.py on m1n1 and paste the output?#asahi · 2021-12-17 08:15
marcanhaven't tested on that one yet, but that sounds like a pmgr tree issue#asahi · 2021-12-17 08:09
marcanmps: what machine?#asahi · 2021-12-17 07:37
marcanwe'll get to this soon#asahi · 2021-12-17 07:36
marcanthe fix is simple but if you insist on trying right now, you really shouldn't expect things to work#asahi · 2021-12-17 07:36
marcan(this is why I keep saying this isn't ready for end-users; there is literally no source tree right now that will give you a proper self-contained boot with u-boot+grub)#asahi · 2021-12-16 16:31
marcanI mean, sure, but it's all for reasonable reasons#asahi · 2021-12-16 16:30
marcan(and made it work seamlessly with macos, while the inner workings are less obvious)#asahi · 2021-12-16 16:29
marcanapple built the only decent consumer secureboot system#asahi · 2021-12-16 16:28
marcanideally we do a proper machine owner transfer on first Linux boot, once SEP support is done, and therefore create a proper separate user for linux#asahi · 2021-12-16 16:28
marcanI've considered doing something like copying the identity file to the iSCPreboot partition so it can outlive containers, but the whole thing is a bit dodgy anyway since that user is still tied to the macos user (e.g. if you delete it, kmutil stops working on the linux container)#asahi · 2021-12-16 16:27
marcanthat needs changes to support this flow#asahi · 2021-12-16 16:27
marcanyou won't be able to re-run the installer though, of course, without a primary macos#asahi · 2021-12-16 16:27
marcanuntil then I don't want to encourage that, though I won't stop you if you really want to nuke macos (you could just wipe the container after installing asahi, then using some partition mover tool to move asahi up into the free space)#asahi · 2021-12-16 16:24
marcanonce we know for sure we can live without macos, I might consider a "replace macos" install mode#asahi · 2021-12-16 16:23
marcanyou can install asahi after deleting macos *if* you copy the user identity file ahead of time; we don't currently have a flow for that and I'm not interested in adding it until we have very solid support in Linux for things like SEP user management / machine owners / firmware updates / etc#asahi · 2021-12-16 16:23
marcanthere is no way of avoiding doing the initial macos setup / user creation#asahi · 2021-12-16 16:17
marcanyes, keep the first one you had, since that one is guaranteed to be machine owner#asahi · 2021-12-16 16:07
marcanbecause then the machine will have an owner you no longer know the UUID for, and thus can't authenticate as#asahi · 2021-12-16 16:07
marcanthere is no way to install anything if you literally have zero containers; in fact, if you used to have one with an admin user and deleted it, I get the feeling you can't actually recover from that situation without a DFU flash#asahi · 2021-12-16 16:06
marcanas400: you need a macos partition initially, as you need to have an admin user on it#asahi · 2021-12-16 07:23
marcanI just force pushed my current branch, but I'm pretty sure those things have worked for a while now#asahi · 2021-12-16 07:20
marcanlandscape15[m]: those patches are in linux-asahi#asahi · 2021-12-16 07:05
marcanthis is what I do on my j314s#asahi · 2021-12-16 07:05
marcantpw_rules: hotplug also works fine with linux-asahi#asahi · 2021-12-16 07:04
marcantpw_rules: yes#asahi · 2021-12-15 16:52
marcan(the ~$30-40 ones)#asahi · 2021-12-15 16:52
marcanmini: I use it with a cheap usb3 one and it works#asahi · 2021-12-15 16:51
marcanit's not on-die, it's on package, fwiw#asahi · 2021-12-15 16:51
marcan(or, well, not cost effective anyway)#asahi · 2021-12-15 16:50
marcanI briefly thought it was but I think it isn't; in fact I think ECC is impossible with 16-bit LPDDR channels :(#asahi · 2021-12-15 16:28
marcanI've never really had it go wrong myself, but yeah...#asahi · 2021-12-15 16:27
marcangiven apple took it out entirely in 12.0.1 :/#asahi · 2021-12-15 16:27
marcansounds like another reason to put my own attempt at that in m1n1...#asahi · 2021-12-15 16:16
marcanbut it *should* be a superset ultimately#asahi · 2021-12-15 16:16
marcanwell, it's probably both right now due to missing stuff#asahi · 2021-12-15 16:15
marcanbut yes, I think right now you need a couple changes for u-boot on top of the kernel device tree#asahi · 2021-12-15 16:14
marcanISTR we need m1n1 to add some u-boot specific properties for the UART and its PMGR nodes#asahi · 2021-12-15 16:14
marcanthere are some differences that should eventually disappear#asahi · 2021-12-15 16:14
marcanah, well, that's a different thing, talk to kettenis#asahi · 2021-12-15 16:14
marcandevice trees are not (in principle) tied to kernels, they are just maintained alongside the kernel