#asahi-dev · 2022-08-29 07:27
svenand they didn't manage to fix it for M2 in time#asahi-dev · 2022-08-29 07:27
svenmaybe it's also just broken in some way#asahi-dev · 2022-08-29 07:26
sveni kinda hope this actually works. would be a fun tweet. "Here's my M1 macbook air with two external screen running Asahi Linux" :D#asahi-dev · 2022-08-29 07:23
svenhuh, also DCP itself needs to know to which atc its stream is routed? ("core%u -> atc%u,%u (supportsHPD=%u)")#asahi-dev · 2022-08-29 07:23
svenweird that apple doesn't make use of that then though#asahi-dev · 2022-08-29 07:22
sventhat would be fun#asahi-dev · 2022-08-29 07:22
svenlol :D#asahi-dev · 2022-08-29 07:19
sveninteresting #asahi-dev · 2022-08-29 07:06
sveni'll play around with it a bit once I actually get dp alt mode up in m1n1 :D#asahi-dev · 2022-08-29 07:05
svenand dpin0/1 is usb4 tunneling#asahi-dev · 2022-08-29 07:05
svenfirst %u in atc is the atc index second on selects between "dpphy", "dpin0" and "dpin1"#asahi-dev · 2022-08-29 07:05
svendispext%u,%u -> atc%u,%u (atc%u-%s)#asahi-dev · 2022-08-29 07:04
svenand XNU debug strings clearly call this dispextN,M#asahi-dev · 2022-08-29 07:04
svenat least the crossbar part I think#asahi-dev · 2022-08-29 07:04
svenor the atcphy0 crossbar to dispext1,0 and dispext2,0 if using USB4 tunneling#asahi-dev · 2022-08-29 07:03
sveni could also configure the atcphy0 crossbar to dispext0,0 and the atcphy1 crossbar to dispext0,1#asahi-dev · 2022-08-29 07:03
svenyup#asahi-dev · 2022-08-29 07:02
svenas I said, I don't think the crossbar even supports MST via DP alt-mode#asahi-dev · 2022-08-29 07:02
svenI can route two streams to DPIN0/1 and tunnel them both over USB4 then#asahi-dev · 2022-08-29 07:01
svenI don't think it supports MST fwiw. I can only route a single stream to DPPHY#asahi-dev · 2022-08-29 07:01
svenyeah, could be#asahi-dev · 2022-08-29 07:01
sven(in that crossbar mux)#asahi-dev · 2022-08-29 07:00
svendon't know yet, all i know is that 0 is dispext0,0 and 1 is dispext0,1#asahi-dev · 2022-08-28 18:52
svenand also a hack that only kept it in suspend for 2 seconds or so#asahi-dev · 2022-08-28 18:51
sveni had no smc driver so the power gpio stayed on#asahi-dev · 2022-08-28 18:50
svenyeah, that explains why it worked for bt#asahi-dev · 2022-08-28 12:15
svencause it’s so important that the keyboard backlight is properly calibrated!#asahi-dev · 2022-08-28 12:15
svenlol, nits to pwm :D#asahi-dev · 2022-08-28 11:46
svenwe’ll figure it out once we get to that point #asahi-dev · 2022-08-28 11:44
sventhat calibration sounds like something that could end up in the device tree as well#asahi-dev · 2022-08-28 11:43
svenwhy bother with things that may not even be firmware already?#asahi-dev · 2022-08-28 11:43
sventhe entire point of that script is to add new firmware once we discover it’s required #asahi-dev · 2022-08-28 11:42
svenit’s a good idea to do it every boot because then you can also be sure it works because it’s tested all the time#asahi-dev · 2022-08-28 11:13
svenyup, looks like it's just two streams per dispext#asahi-dev · 2022-08-28 10:51
svenxnu talks about dcpextN,M so maybe they want to support two streams inside each dcpext at some point? dunno#asahi-dev · 2022-08-28 10:50
svenhrm, interesting. looks like I’ll have to go and figure out how the dcpext maps to bits in that register then #asahi-dev · 2022-08-28 10:45
svenit’s using dcpext0 and whatever dcpext is mapped to 2#asahi-dev · 2022-08-28 10:45
svenoffset 0x60 is what controls the mux #asahi-dev · 2022-08-28 10:43
svennice, thanks. do you know which dcpext it ended up using?#asahi-dev · 2022-08-28 09:59
sven/arm-io/atcN-dpxbar actually#asahi-dev · 2022-08-28 09:58
svencan anyone with a m1 pro/max run trace_device("/arm-io/atxN-xbar") for all atcs and then plug two external screens into some usb-c ports using dp altmode?#asahi-dev · 2022-08-28 09:31
svenalso different incentives. Linux will probably support m1 drivers for 10+ years. Apple can just drop support or introduce breaking changes whenever they want to. #asahi-dev · 2022-08-28 09:26
svenso they went with the fake adt node #asahi-dev · 2022-08-28 09:26
svenmy best guess is that the gpu kext team didn’t want to or wasn’t allowed to talk to the iommu team to get some api to just register custom iommu ops#asahi-dev · 2022-08-28 09:25
svenso if the main kext allocates io memory it’ll call into their iommu api which calls the fake iommu kext which calls back into the main kext :D#asahi-dev · 2022-08-28 09:24
svenApple also has some horrible hacks in their kernel. I still enjoy the fake-gpu-iommu kext and adt node that’s just a wrapper that calls back into the main gpu kext#asahi-dev · 2022-08-28 09:17
svenLinux is a single binary that has to work everywhere #asahi-dev · 2022-08-28 09:17
sventhey also have a different kernel for each SoC which also helps things #asahi-dev · 2022-08-28 08:27
svenAs I said, almost everything is left #asahi-dev · 2022-08-28 08:23
sveni have a wip nhi driver but it’s not working yet