#asahi-dev · 2022-04-14 09:19
jannauyes, at least partially. Can't be removed the 'if (WARN(...'#asahi-dev · 2022-04-14 09:10
jannaufound how to clear surfaces with dcp_swap#asahi-dev · 2022-04-14 07:45
jannausven: nvme-v2 works with dcp in asahi, smc rtkit not adapted but disabled#asahi-dev · 2022-04-14 07:22
jannaumarcan: did submit the PR for the alarm rust PKGBUILD? update to 1.60 broke it#asahi-dev · 2022-04-13 16:16
jannaudcp works on the mac studio if I don't init the display in m1n1#asahi-dev · 2022-04-12 19:07
jannauyes, "mailbox: apple: Implement peek_data() operation" breaks it#asahi-dev · 2022-04-12 18:43
jannausven: nvme-v2 + dcp + nvme devicetree works so far#asahi-dev · 2022-04-12 18:38
jannauat least didn't intend to. I did at first but decided to avoid dealing with the conflicts#asahi-dev · 2022-04-12 18:35
jannauI didn't reapply the atomic patches#asahi-dev · 2022-04-12 18:33
jannaunvme and dcp are working#asahi-dev · 2022-04-12 18:33
jannausven: it works sometimes. second boot of nvme-v2 + dcp + nvme devicetree didn't schedule while atomic a single time#asahi-dev · 2022-04-12 18:15
jannausigh, I need to switch to the mac studio for compiling#asahi-dev · 2022-04-12 18:12
jannaucompiling now nvme-v2 + dcp, not really expecting other results but much clearer what's tested#asahi-dev · 2022-04-12 17:57
jannauI can test on the mac mini with just dcp on top of nvme-v2#asahi-dev · 2022-04-12 17:56
jannauI can push the tree but it's a mess#asahi-dev · 2022-04-12 17:55
jannaunvme-apple 393cc0000.nvme: ANS did not boot#asahi-dev · 2022-04-12 17:55
jannauI've reverted b0aec24ea7a8e79cfce311fc8a7bc6aa4bf14ccf and 16b8a4a8997d359639caf9b641cea8466e68c717 in asahi and I've disabled smc#asahi-dev · 2022-04-12 17:52
jannauhttps://paste.centos.org/view/a4cc4585#asahi-dev · 2022-04-12 17:48
jannausyslog still works#asahi-dev · 2022-04-12 17:48
jannausven: tons of "scheduling while atomic" BUGS#asahi-dev · 2022-04-12 17:27
jannauyes, would have been easier in order#asahi-dev · 2022-04-12 17:17
jannausven: I haven't yet because it's a little annoying. starting now, replaced all rtkit/nvme commits with the top two commits in nvme-v2#asahi-dev · 2022-04-11 10:59
jannau10% faster, 70s instead of 78s for make vmlinux with linux-5.15 arm64 defconfig. I guess I have to redo it on the m1 max as well#asahi-dev · 2022-04-11 10:15
jannaushould be 3, 4, 5 instead of 0, 1, 2#asahi-dev · 2022-04-11 10:15
jannaumissed to update the core's "apple,freq-domain = <&cpufreq_hw x>;" for the second die#asahi-dev · 2022-04-10 21:52
jannauhttps://github.com/jannau/m1n1/tree/t6002-bringup#asahi-dev · 2022-04-10 21:51
jannaumarcan: refreshed t6002-bringup branch pushed. not fully happy with the multi-die pmgr handling, macos guest handling still needs work#asahi-dev · 2022-04-10 21:08
jannauindication that BIT(1) of dart-t6000 is indeed uncached: iboot/macos set it for dcp/disp0 but not for dart-usb#asahi-dev · 2022-04-10 21:02
jannaupanic with initialized display is the AMCC UNEXP_RT_HIT_DIR with an address inside dart-disp0's first page table for dev 0#asahi-dev · 2022-04-10 20:27
jannaubut it is so slow that it panics on shutdown with "Halt/Restart Timed Out @IOPlatformExpert.cpp:883"#asahi-dev · 2022-04-10 20:22
jannausigh, HV + macos guest works on the m1 ultra if I don't initialize the display / map the framebuffer#asahi-dev · 2022-04-10 20:15
jannaulol, I can crash dcp in way which doesn't keeps the display scanout working. In that state I can turn the monitor off and on without breaking the display#asahi-dev · 2022-04-10 14:23
jannaufor pgmr we probably have to assume that always-on is correct and everything else can be disabled if there is no device using it#asahi-dev · 2022-04-10 14:23
jannauit's not a problem for now just annoying. it looks like it's the easiest way to hardcode it in m1n1 as well until a 4 die SoC arrives#asahi-dev · 2022-04-10 14:09
jannauI don't see anything in the pgmr ADT node to indicate whether devices are present on either die#asahi-dev · 2022-04-10 14:08
jannauI still don't see anything in the ADT which helps to deal with the magically present pgmr on the second die. we could infer the offset from ranges but it's a little annoying#asahi-dev · 2022-04-08 16:54
jannauno problem. what we have right now works and was good enough to get the first user going#asahi-dev · 2022-04-08 16:52
jannauI'll work on it over the weekend and post what I get done until sunday#asahi-dev · 2022-04-08 16:46
jannaumarcan: have you done any work on the m1 ultra? if not I can work on a m1n1 merge request and fix HV+macos#asahi-dev · 2022-04-08 06:25
jannauthe panics I looked at after inactivity were caused by suspend#asahi-dev · 2022-04-08 06:24
jannaurqou_: I think macos just auto-suspends/sleeps after a certain amount of inactivity. supporting that properly with the HV is probably not worth the effort. I haven't bothered to check if it goes away if macos is configured to stay on#asahi-dev · 2022-04-08 06:12
jannauthe panic reason ("AMCC2 PLANE3 UNEXP_RT_HIT_DIR error: ...") looks like it could be caused by something m1n1 did#asahi-dev · 2022-04-08 06:11
jannauJamie[m]1: I haven't on t6001, panics either on boot or on suspend but I'm also not using macos extensively under the HV. mostly short sessions to trace kbd or dcp#asahi-dev · 2022-04-07 11:02
jannauread write protection bits also differ between between t8110 and the defines for t8103 in the kernel#asahi-dev · 2022-04-07 11:00
jannauthe ARM_MALI_LPAE page table format seems to have also a different use for BIT1)#asahi-dev · 2022-04-07 10:55
jannauit appears t6000 and t8110 share the same PTE layout assuming the offset mask 37,10 in the kernel patch is correct and the 39,10 offset in m1n1 is an oversight#asahi-dev · 2022-04-07 10:49
jannauok, I'll ping it tonight#asahi-dev · 2022-04-07 10:47
jannausven: any progress on getting the t6000-dart merged in linux? would it help to ping the patch? we should try to get it into 5.19 together with basic t600x device trees#asahi-dev · 2022-04-07 05:45
jannaurqou_: it's entirely possible that it is the same race and broken dcp results in strange looking rtkit behavior for dcp but still unsure how that can break nvme#asahi-dev · 2022-04-07 05:42
jannaunot in a central place