#asahi · 2022-01-22 11:00
mazany HW virtualisation framework (HVF, KVM...) will give you close-to-native speed as long as you don't start oversubscribing the system.#asahi · 2022-01-20 11:06
mazthat's the core of the problem: instead of an abstraction layer, people want to use it as a quirk description mechanism.#asahi · 2022-01-20 11:01
mazlx2160a is the poster child of such madness.#asahi · 2022-01-20 11:00
mazexactly.#asahi · 2022-01-20 10:53
mazthe fun begins when people use the SR argument to justify adding terrible quirks to support ACPI on HW that really shouldn't use that.#asahi · 2022-01-20 10:49
maz(although their IORT table is... interesting...)#asahi · 2022-01-20 10:49
mazthey got an exemtion for it. I think Altra is sort of OK in that respect (I booted fairly old kernels on it without any extra patch).#asahi · 2022-01-20 10:47
mazto the point where ServerReady doesn't mean anything.#asahi · 2022-01-20 10:47
mazand ARM will hand out exemptions for each and every broken system anyway.#asahi · 2022-01-20 10:46
maza handful of machines, and mostly software models.#asahi · 2022-01-20 10:44
mazthere are. the Ampere stuff is reasonably correct, for example.#asahi · 2022-01-20 10:43
mazChaosPrincess: you can't fit AIC or dart in ACPI without a major revision of the spec.#asahi · 2022-01-20 10:43
mazbut I wouldn't put it past ARM to create a special ServerReady category just for the M1. that's what they do for every broken HW implementation out there.#asahi · 2022-01-20 10:42
mazwell, not if they only want to support standard HW.#asahi · 2022-01-20 10:40
mazit isn't even ARM architecture compliant, so forget about everything else.#asahi · 2022-01-20 10:39
mazschaha! wrong platform then.#asahi · 2022-01-20 10:28
mazthat's something they could work around (though shoehorning AIC and DART into ACPI could be... fun...)#asahi · 2022-01-20 10:25
maznico_32: there are no technical obstacle to it.#asahi · 2022-01-17 16:57
maz(not that I care about secure boot, but hey...)#asahi · 2022-01-17 16:57
mazyeah, and they deprecated it just before the stable release.#asahi · 2022-01-05 18:58
mazj`ey: I don't *only* communicate using FZ songs!#asahi · 2022-01-05 18:47
mazkettenis: that's courtesy of marcan (I really wouldn't have found that myself!).#asahi · 2022-01-05 18:37
maz(except for $work)#asahi · 2022-01-05 18:37
mazyeah, I don't do x86 either.#asahi · 2022-01-05 18:37
mazHiSi D05, 64xA72, 256GB of RAM. quite an odd/old machine.#asahi · 2022-01-05 18:34
maz(it also has only a 20th of the memory, but hey...)#asahi · 2022-01-05 18:34
mazbut given that my mini draws less than a 20th of my normal box for half the speed, I see an immediate benefit.#asahi · 2022-01-05 18:31
mazdunno, I don't do laptops.#asahi · 2022-01-05 18:31
mazhttps://paste.debian.net/1226030/#asahi · 2022-01-05 18:30
mazand with a tiny m1n1 hack, you get 3GHz on the p-cores.#asahi · 2022-01-05 17:04
mazwhat we don't (and won't) have is direct access to the counter from userspace (that'd be a new ABI, and I'm not going there), nor PMUs being exposed to KVM guests (for the same reasons).#asahi · 2022-01-05 17:03
maztpw_rules: yes. I have a driver on the list that does exactly that. the main issue is that the documentation for the various events is extremely sparse.#asahi · 2022-01-05 16:57
mazyes#asahi · 2022-01-05 16:56
maztpw_rules: M1 doesn't implement the ARMv8 PMU. instead, it has a home grown PMU that doesn't work like the standard one, and doesn't count the same things.#asahi · 2022-01-05 16:55
maztpw_rules: just not the architected version.#asahi · 2022-01-05 16:55
maztpw_rules: it does.#asahi · 2022-01-02 12:28
mazyes, absolutely. it may not be able to use the FS though.#asahi · 2022-01-02 12:26
mazthere are *two* grub config files in most installs: one that is colocated with bootaa64.efi (used to botstrap the environment), and another one with the full config. obviously, the first has to know how to find the second one.#asahi · 2022-01-02 12:23
mazit looks like grub doesn't find its configuration file, or that the fonfig file doesn't match the system that grub sees (UUIDs, partition layout and such)#asahi · 2022-01-02 12:07
mazAPFS != APFS container.#asahi · 2022-01-02 12:05
mazamw: the standard way to use u-boot from a distro these days is to use u-boot's EFI interface. it makes things a lot easier for Debian (and everyone else), as nobody has to care about the various versions of u-boot and their level of brokenness across 2000 different platforms.#asahi · 2022-01-02 11:59
mazand it's not like partitioning is wasting much space either...#asahi · 2022-01-02 11:58
mazlandscape15[m]: and you want to consider some swap as well.#asahi · 2022-01-02 11:57
mazlandscape15[m]: a lot? have you ever seen an Android install, or a ChromeOS install?#asahi · 2022-01-02 11:55
mazmps: well, everything else is a module that can live anywhere wher grub can load it. the core grub has to live in the EFI partition.#asahi · 2022-01-02 11:54
mazmps: that's grub.#asahi · 2022-01-02 11:53
mazmps: are you sure? you should have a file called bootaa64.efi in the EFI partition.#asahi · 2022-01-02 11:51
mazlandscape15[m]: grub needs to live somewhere where the EFI loader will find it, and that's in the EFI partition.#asahi · 2022-01-02 11:45
mazit could, if you were to port grub to m1n1. but the amount of work required for that is absolutely silly, and that's exactly why we have an EFI layer in u-boot...#asahi · 2021-12-06 10:38
mazthat being said, nobody needs to cater for my special needs. I'm happy to install m1n1/u-boot by hand.