#asahi · 2022-02-18 06:57
marcanjmr2, David[m]123456789: I just updated the installer, please try again#asahi · 2022-02-18 05:38
marcanthat's new.#asahi · 2022-02-18 05:38
marcanwell then.#asahi · 2022-02-18 04:50
marcan-s#asahi · 2022-02-18 04:50
marcanthe "Volume Groups <foo>" should be the same as "Volume UUID: <foo>" for the data volume under it#asahi · 2022-02-18 04:49
marcanjmr2: diskutil apfs listVolumeGroups shows a different VGUUID from the data role volume?#asahi · 2022-02-17 13:20
marcanyesterday someone said B is a dissonant note. When pressed, they said "in the key of C" (as if that key is vastly more popular). I said B is literally in the C major scale and part of some of the most important chords in that scale. They insisted B and C together are dissonant. Guess they've never heard of CMaj7. I guess all jazz and most popular music is dissonant these days...#asahi · 2022-02-17 13:18
marcanas400[m]: being knowledgeable about a subject really makes it obvious how much nonsense gets said there...#asahi · 2022-02-16 18:34
marcanthere's always logs#asahi · 2022-02-16 18:34
marcanbut we first need to make sure this is really the way to go and macOS does the same#asahi · 2022-02-16 18:34
marcanso you never lose more than 1 second of data#asahi · 2022-02-16 18:34
marcanif this is really the only way, the only sane solution is to basically rate-limit/time flushes so you guarantee dirty data isn't cached for, say, more than 1 second#asahi · 2022-02-16 18:33
marcanwe were talking about that in -dev#asahi · 2022-02-16 18:33
marcanit's lying to the kernel about the drive having no cache or a nonvolatile cache#asahi · 2022-02-16 18:32
marcanwhich means no data is ever dirty so no flushes are needed#asahi · 2022-02-16 18:32
marcanit means read cache#asahi · 2022-02-16 18:30
marcanhence lost guarantees#asahi · 2022-02-16 18:30
marcanwhich it isn't#asahi · 2022-02-16 18:30
marcanwritethrough tells it the cache is writethrough#asahi · 2022-02-16 02:35
marcanI need to figure out if there's a way to get the true frequency#asahi · 2022-02-16 02:34
marcanlinux thinks it's 3.2 but it isn't#asahi · 2022-02-16 02:34
marcan3.2 is a lie though#asahi · 2022-02-16 02:34
marcanbut if it works now I assume you fixed that :p#asahi · 2022-02-16 02:34
marcanthis is my "it works on the m1pro, I'm going to sleep" tree from last night#asahi · 2022-02-16 02:33
marcanas I said the DTs need changing#asahi · 2022-02-16 02:33
marcankov: the DT for the mini does not have the cpufreq stuff yet#asahi · 2022-02-16 00:12
marcanso it's really <10% boost you're going to get for single-core perf, not a deal breaker for most users#asahi · 2022-02-16 00:12
marcanthat already works fine#asahi · 2022-02-16 00:12
marcankov: 3.0#asahi · 2022-02-16 00:12
marcanthat's where the PSCI story comes in#asahi · 2022-02-16 00:12
marcan(because for that the other cores need to be into a deeper sleep state than we currently support)#asahi · 2022-02-16 00:11
marcan(but not right now)#asahi · 2022-02-16 00:08
marcancurrent story is doing it via EFI#asahi · 2022-02-16 00:08
marcanso we need another mechanism#asahi · 2022-02-16 00:08
marcanso we can't just say m1n1 is the hypervisor for normal usage#asahi · 2022-02-16 00:08
marcanopticron: but the problem is linux needs to run at EL2 for KVM#asahi · 2022-02-16 00:08
marcanhttps://github.com/AsahiLinux/m1n1/commit/1a775715573ec8dff35e3486b82d59b3f498640d#asahi · 2022-02-16 00:07
marcanopticron: SMC trap to EL2 is an impdef bit#asahi · 2022-02-16 00:07
marcanone core per cluster can go into 3.2, or two into 3.1#asahi · 2022-02-15 07:09
marcangrab the dump while in recovery mode and while in macOS#asahi · 2022-02-15 07:08
marcanI *think* setting the main macOS partition explicitly as the boot drive should fix it#asahi · 2022-02-15 07:08
marcanjmr2: can you grab an ioreg -l -p IODeviceTree dump, and also installer output?#asahi · 2022-02-15 05:23
marcanstreaming in -stream in a bit, going to try wrap up SMC and CPUfreq#asahi · 2022-02-08 08:23
marcanI'll put together a bits branch soonish :)#asahi · 2022-02-08 08:19
marcanyou do need RTKit and all that... or do you mean something else?#asahi · 2022-02-08 08:04
marcanthey are different things#asahi · 2022-02-08 08:04
marcanbecause that's a property of the battery, not the AC adapter#asahi · 2022-02-08 08:03
marcanthere is no AC adapter device yet#asahi · 2022-02-08 04:46
marcan(it could be they remap within a cluster to always make the failed cores last, but that seems like too much effort)#asahi · 2022-02-08 04:45
marcanthat's what they'd do when they have too many good CPUs and need to just disable cores and sell them as lower-core ones