#asahi · 2021-12-13 16:07
marcanNightRaven[m]: this week#asahi · 2021-12-12 00:42
marcancool!#asahi · 2021-12-12 00:05
marcanor rather they do, to the T2; then the Intel sees a weirdo virtual USB controller instead#asahi · 2021-12-12 00:05
marcanthose don't use SPI for the keyboard though#asahi · 2021-12-12 00:05
marcanI also in theory have access to an idle T2 mac (though I think it'll be put to use soonish); I want to use that to test the wifi patches and make them work on ACPI systems#asahi · 2021-12-12 00:04
marcanI'll ask them what year it was, it might be in range#asahi · 2021-12-12 00:03
marcan12 is apparently Early 2015?#asahi · 2021-12-12 00:03
marcansays the driver#asahi · 2021-12-12 00:02
marcan * All others need this driver. "#asahi · 2021-12-12 00:02
marcan * pins are only connected on the MacBookAir6 and 7 and the MacBookPro12.#asahi · 2021-12-12 00:02
marcan * MacBook8 and newer can be driven either by USB or SPI. However the USB#asahi · 2021-12-12 00:02
marcan" * The keyboard and touchpad controller on the MacBookAir6, MacBookPro12,#asahi · 2021-12-12 00:02
marcannot sure :-)#asahi · 2021-12-12 00:02
marcannot sure if it's the right gen though#asahi · 2021-12-12 00:01
marcanwould be nice to follow up with some patches to support that#asahi · 2021-12-12 00:01
marcanI wonder if I have access to an Intel MBP with that kind of keyboard... I might, I have a friend with an older (pre-T2) MBP they aren't using, maybe I can borrow that#asahi · 2021-12-11 23:58
marcan(to clarify: the wifi *patches* need a rewrite, the existing driver is fine; the keyboard *driver* needs a rewrite because the existing driver is doing entirely the wrong thing)#asahi · 2021-12-11 23:57
marcanideally that would also be ported to Intel macbooks and end up replacing the old driver altogether at some point in the future#asahi · 2021-12-11 23:57
marcansame with the keyboard driver btw; the patches to that aren't crazy, and it can be made to work (I spent some time the past few days poking around the corellium patches to that to validate the SPI driver I wrote), but it turns out the *original* driver in Linux there was written with the wrong understanding and we're going to be writing a new driver from scratch that does things properly#asahi · 2021-12-11 23:55
marcanthe way that driver does things is completely crazy and will never fly upstream; therefore trying to use it is largely a waste of time since it needs a complete rewrite#asahi · 2021-12-11 10:58
marcanI doubt anything recent in m1n1 caused it#asahi · 2021-12-11 10:56
marcanwe did have that breaking change about requiring a compatible DT, but that shouldn't be the problem here#asahi · 2021-12-11 10:56
marcan*appended#asahi · 2021-12-11 10:56
marcanyou can test it by chainloading the appending m1n1 instead of using linux.py#asahi · 2021-12-11 10:56
marcanlinux.py *should* be roughly equivalent to the cat method, but I can't say there aren't any bugs#asahi · 2021-12-11 10:32
marcanyeah, I gotcha :p#asahi · 2021-12-11 10:32
marcanthey are useful when booting macOS under the HV :)#asahi · 2021-12-11 10:31
marcanm1n1 does not attempt to pass through any of those to Linux#asahi · 2021-12-11 10:31
marcanthe commandline arguments specified in run_guest.py are *xnu bootargs*, not Linux args#asahi · 2021-12-11 10:31
marcanJamie[m]1: the hypervisor does not boot linux directly, it boots m1n1 which boots linux#asahi · 2021-12-10 06:16
marcanI think we have nobody else on tor, so let's do the practical thing#asahi · 2021-12-10 06:15
marcanlooks like they're connecting via tor#asahi · 2021-12-10 06:14
marcanjannau: they're already gone, these are bots that come and go#asahi · 2021-12-10 00:14
marcansimpledrm does in-kernel conversion instead#asahi · 2021-12-10 00:13
marcanwe won't be doing that#asahi · 2021-12-10 00:13
marcancorellium did a horrible hack to switch the hardware to 8-bit#asahi · 2021-12-10 00:13
marcanX does not like simplefb with 10-bit color#asahi · 2021-12-09 01:44
marcanwell that's a chromium bug if it can't handle 16K pages then#asahi · 2021-12-08 18:11
marcanI basically never rebooted during the SPI thing; once I started wedging the hardware I just started using the PMGR reset bit to reset it on every test run#asahi · 2021-12-08 18:10
marcanit's an experimentation platform; the whole point is to test stuff outside an OS interactively#asahi · 2021-12-08 18:08
marcanthe M1 speculates like crazy and is really good at bringing out memory ordering and concurrency bugs#asahi · 2021-12-08 18:08
marcanit's also possible it's literally a bug somewhere#asahi · 2021-12-08 18:05
marcanj_ey: even if you cause pin contention, modern silicon is quite resilient to shorts and such#asahi · 2021-12-08 18:04
marcan(it did crash USB-PD though)#asahi · 2021-12-08 18:04
marcanthat's a good sign for the reliability of these machines :p#asahi · 2021-12-08 18:04
marcanalso: during today's stream I managed to toggle all GPIO register bits by accident, and I didn't seem to cause any permanent damage#asahi · 2021-12-08 18:01
marcanfeel free to debug it :p#asahi · 2021-12-08 18:01
marcanit should be trying to use llvmpipe, but who knows what goes wrong#asahi · 2021-12-08 18:01
marcanI assume it is#asahi · 2021-12-08 18:00
marcanthat's my guess anyway, need to look into it