#asahi · 2022-03-29 13:33
marcan... chances of something else going wrong with the rest of the software which is alpha-quality anyway#asahi · 2022-03-29 13:33
marcanand I'm not, because I know the SSD controller is on-SoC, which means the chances of a bit flip there are ~the same as the chances of a bit flip in the main CPU, and NAND is error-corrected, and the PCIe links to the NAND chips are also error-checked, and really there is no reason to believe the SSD randomly corrupting stuff is a major concern on this particular platform, not any more than the ...#asahi · 2022-03-29 13:29
marcanit wasn't funny then#asahi · 2022-03-29 13:27
marcanor with being *told* to work on stuff just to fix problems you caused, like that 12.1 thing#asahi · 2022-03-29 13:26
marcanjust dealing with the extra workload of answering your questions that you wouldn't have to ask if you'd just waited for things to get merged into the relevant trees is tiring enough#asahi · 2022-03-29 13:25
marcanGlanzmann: I'm going to be honest, I'm scared of the time coming when you actually cause some damage and I need to deflect responsibility for the folks working on this properly#asahi · 2022-03-29 13:22
marcanand blindly merging from random testing branches where even the patch *author* hasn't said "this is mergeable" yet is crazy#asahi · 2022-03-29 13:21
marcanblindly merging that kind of thing is how people destroy their systems#asahi · 2022-03-29 13:21
marcanand I especially look into spots where I know there's danger, like that time I caught that the DTs povik sent me didn't have the speakers disabled in one model#asahi · 2022-03-29 13:20
marcanwhich is why I don't randomly merge things into asahi until folks (that I trust to begin with) tell me to and I take a look#asahi · 2022-03-29 13:20
marcanand knowing what people to trust with things and what kind of development/testing process works to get to a decent outcome#asahi · 2022-03-29 13:18
marcanso we *kind* of have a history of knowing what is safe to ship and what isn't#asahi · 2022-03-29 13:17
marcanand that was like a decade+ ago#asahi · 2022-03-29 13:17
marcanthe difference is sven and I have shipped software that 10m+ users have installed that makes core changes to the boot process of proprietary reverse engineered systems without bricking any, to our knowledge#asahi · 2022-03-29 13:08
marcanwhat you're doing here is pulling people's personal WIP branches randomly#asahi · 2022-03-29 13:08
marcan~everything in git head from torvalds has usually been reviewed by at least 3 people if not more#asahi · 2022-03-29 13:01
marcanthough I should probably also have a linux-next merge branch somewhere, because we might find things ahead of time there (*cough* that "fix a warning" patch that broke t6000 *cough*)#asahi · 2022-03-29 13:01
marcannot that much work and less likely to explode#asahi · 2022-03-29 13:01
marcannext asahi rebase will probably start tracking 5.18 RCs and I'll ditch the linux-next story, instead just merging individual upstreamed trees/branches as needed#asahi · 2022-03-29 13:00
marcanyeah#asahi · 2022-03-29 12:58
marcanthough I'll probably switch to -rcs soon anyway#asahi · 2022-03-29 12:58
marcanand here I thought I was playing with fire by shipping linux-next... but at least that stuff *has* been reviewed (mostly)#asahi · 2022-03-29 12:47
marcan(I'm not entirely sure you see, since you've been told this before and have kept on doing it for months)#asahi · 2022-03-29 12:40
marcanI just never got around to reviewing the linux driver to see if it wasn't broken too :(#asahi · 2022-03-29 12:40
marcanwith transaction ends#asahi · 2022-03-29 12:40
marcanyeah, and we already had this issue with i2c in m1n1...#asahi · 2022-03-29 04:45
marcanno, but don't run update-grub yet, I need to check that I didn't miss something there#asahi · 2022-03-28 17:17
marcanit's the same thing macOS does, except macOS does it by issuing a magic kind of panic (lol)#asahi · 2022-03-28 17:17
marcanthat allows the NVMe cache to flush#asahi · 2022-03-28 17:16
marcanwhen the shutdown button is held we trigger a hard shutdown before the hardware has a chance to#asahi · 2022-03-28 17:16
marcanit's the other way around#asahi · 2022-03-28 17:15
marcanyup#asahi · 2022-03-28 17:14
marcana few months ago maybe, I wrote the driver for that already#asahi · 2022-03-28 17:13
marcanno?#asahi · 2022-03-28 17:13
marcandeepsleep can mean several things too; CPU deep sleep, s2idle, or full s3 style sleep#asahi · 2022-03-28 17:12
marcanwhat do you mean by that?#asahi · 2022-03-28 17:12
marcanshutdown works fine?#asahi · 2022-03-28 17:11
marcan(and there are advantages to it)#asahi · 2022-03-28 17:11
marcanjannau: I want to keep the fb 10-bit, that falls into the "get people to fix their software" motivator list ;)#asahi · 2022-03-27 21:19
marcanit's not supposed to set >1920x1200 modes in auto mode#asahi · 2022-03-27 21:19
marcanthat's working as intended#asahi · 2022-03-27 10:51
marcanuntil either the next machines are released or we decide to upgrade for some important reason#asahi · 2022-03-27 10:51
marcanin fact the stub will remain at 12.3 for the foreseeable future#asahi · 2022-03-27 10:50
marcannot unless something goes wrong with the stub which would be very unlikely since it doesn't really try to do anything complicated#asahi · 2022-03-27 10:50
marcanno#asahi · 2022-03-27 10:50
marcanbut if something breaks it should fail gracefully, not stop everything from working#asahi · 2022-03-27 10:49
marcanthe system is designed to be backwards compatible, but I can't guarantee we won't run into a corner case because obviously we can't talk to apple directly about what the expectations are#asahi · 2022-03-27 10:49
marcanat least in principle#asahi · 2022-03-27 10:48
marcanyes#asahi · 2022-03-27 10:45
marcanyou have to upgrade to 12.3 to install in the first place